1 2018-09-13 Eric Botcazou <ebotcazou@adacore.com>
4 * config/visium/visium.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define.
6 2018-09-13 Richard Biener <rguenther@suse.de>
8 PR tree-optimization/87263
9 * tree-ssa-sccvn.c (visit_phi): Revert some earlier changes.
10 (struct unwind_state): Add max_rpo field.
11 (do_rpo_vn): Allow up-to-date loop state to be used when not iterating.
12 Compute max_rpo, the max RPO number a block can be backwards reached
13 from. Re-write non-iterating mode to a RPO ordered worklist approach,
14 separating it from the iterating mode.
16 2018-09-13 Vlad Lazar <vlad.lazar@arm.com>
18 * haifa-sched.c (rank_for_schedule): Schedule by INSN_COST.
19 (rfs_decision): New scheduling decision.
21 2018-09-13 Richard Biener <rguenther@suse.de>
24 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Fix assert.
25 (vn_nary_op_insert_pieces_predicated): Do not write useless
26 valid_dominated_by_p entry outside of the allocated storage.
28 2018-09-13 Omar Sandoval <osandov@osandov.com>
29 Tom de Vries <tdevries@suse.de>
32 * dwarf2out.c (is_c): New function.
33 (add_subscript_info): Add DW_AT_count of 0 for C zero-length arrays.
35 2018-09-13 Sam Tebbs <sam.tebbs@arm.com>
38 * config/aarch64/aarch64.md (*aarch64_bfxil):
40 * config/aarch64/constraints.md (Ulc): Define.
41 * config/aarch64/aarch64-protos.h (aarch64_high_bits_all_ones_p):
43 * config/aarch64/aarch64.c (aarch64_high_bits_all_ones_p):
46 2018-09-13 Vlad Lazar <vlad.lazar@arm.com>
48 * config/aarch64/aarch64.h (TARGET_COMPUTE_FRAME_LAYOUT): Define.
49 * config/aarch64/aarch64.c (aarch64_expand_prologue): Remove
50 aarch64_layout_frame call.
51 (aarch64_expand_epilogue): Likewise.
52 (aarch64_initial_elimination_offset): Likewise.
53 (aarch64_get_separate_components): Likewise.
54 (aarch64_use_return_insn_p): Likewise.
55 (aarch64_layout_frame): Remove unneeded check.
57 2018-09-13 Jakub Jelinek <jakub@redhat.com>
59 * configure.ac: Only append
60 " : (reconfigured) $TOPLEVEL_CONFIGURE_ARGUMENTS" to
61 gcc_config_arguments if it was never reconfigured or last reconfigure
62 was with different arguments.
63 * configure: Regenerated.
65 2018-09-13 Jakub Jelinek <jakub@redhat.com>
66 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
69 * expr.c (maybe_optimize_pow2p_mod_cmp): New function.
70 (maybe_optimize_mod_cmp): Use it if integer_pow2p treeop1.
72 2018-09-13 Jakub Jelinek <jakub@redhat.com>
74 PR tree-optimization/87287
75 * fold-const.c (fold_binary_loc) <case EQ_EXPR>: Move signed modulo
76 X % C == 0 to X % (unsigned) C == 0 optimization to ...
77 * match.pd (X % C == 0): ... here. New optimization.
79 2018-09-12 Jakub Jelinek <jakub@redhat.com>
82 * expr.h (maybe_optimize_mod_cmp): Declare.
83 * expr.c (mod_inv): New function.
84 (maybe_optimize_mod_cmp): New function.
85 (do_store_flag): Use it.
86 * cfgexpand.c (expand_gimple_cond): Likewise.
88 2018-09-09 Cesar Philippidis <cesar@codesourcery.com>
89 Julian Brown <julian@codesourcery.com>
92 * gimplify.c (gimplify_scan_omp_clauses): Set
93 target_firstprivatize_array_bases in OpenACC parallel and kernels
94 region contexts. Remove GOMP_MAP_FIRSTPRIVATE_REFERENCE clauses from
97 2018-09-12 Uros Bizjak <ubizjak@gmail.com>
99 * config/i386/i386.md (sqrt_extend<mode>xf3_i387): Remove.
100 (sqrt<mode>2): Extend operand 1 to XFmode and generate sqrtxf3 insn.
102 2018-09-12 Richard Biener <rguenther@suse.de>
104 PR tree-optimization/87280
105 * tree-ssa-sccvn.c (process_bb): Handle the case of executable
106 edge but unreachable target.
107 (do_rpo_vn): For conservatively handling a PHI only mark
108 the backedge executable but not the block reachable.
110 2018-09-12 Richard Biener <rguenther@suse.de>
112 PR tree-optimization/87266
113 * tree-ssa-sccvn.c (do_rpo_vn): Always iterate to not yet
116 2018-09-12 Andreas Krebbel <krebbel@linux.ibm.com>
118 * config/s390/s390.md (PFPO_RND_MODE_DFP, PFPO_RND_MODE_BFP): New
120 ("trunc<BFP:mode><DFP_ALL:mode>2")
121 ("trunc<DFP_ALL:mode><BFP:mode>2")
122 ("extend<BFP:mode><DFP_ALL:mode>2")
123 ("extend<DFP_ALL:mode><BFP:mode>2"): Set proper rounding mode
124 according to the target operand type.
126 2018-09-12 Jakub Jelinek <jakub@redhat.com>
127 Andreas Krebbel <krebbel@linux.ibm.com>
129 PR tree-optimization/86844
130 * gimple-ssa-store-merging.c
131 (imm_store_chain_info::coalesce_immediate): For overlapping stores, if
132 there are any overlapping stores in between them, make sure they are
133 also coalesced or we give up completely.
135 2018-09-12 Jakub Jelinek <jakub@redhat.com>
138 * fold-const.c (fold_ternary_loc) <case COND_EXPR>: Verify also that
139 BIT_AND_EXPR's second operand is a power of two. Formatting fix.
141 2018-09-12 Tom de Vries <tdevries@suse.de>
143 * common.opt (gdescribe-dies): Add option.
144 * dwarf2out.c (add_name_and_src_coords_attributes): Add description
145 attribute for artifical and nameless decls.
146 (dwarf2out_register_external_die): Add description attribute to
147 external reference die.
148 (add_desc_attribute): New functions.
149 (gen_subprogram_die): Add description attribute to
150 DW_TAG_call_site_parameter.
151 * tree-pretty-print.c (print_generic_expr_to_str): New function.
152 * tree-pretty-print.h (print_generic_expr_to_str): Declare.
153 * doc/invoke.texi (@item Debugging Options): Add -gdescribe-dies and
155 (@item -gdescribe-dies): Add.
157 2018-09-12 Aldy Hernandez <aldyh@redhat.com>
159 * tree-vrp.c (vrp_shift_undefined_p): Remove.
160 (extract_range_from_binary_expr_1: Call
161 wide_int_range_shift_undefined_p instead of vrp_shift_undefined_p.
162 * wide-int-range.h (wide_int_range_shift_undefined_p): Do not
165 2018-09-12 Aldy Hernandez <aldyh@redhat.com>
167 * gimple-ssa-warn-alloca.c
168 (alloca_type_and_limit::alloca_type_and_limit): Initialize limit
169 field for ALLOCA_BOUND_*_LARGE.
171 2018-09-11 Nathan Sidwell <nathan@acm.org>
173 * gcc.c (load_specs, execute, run_attempt): Use %qs not '%s'.
175 2018-09-11 Uros Bizjak <ubizjak@gmail.com>
177 * reg-stack.c (subst_asm_stack_regs): Call replace_reg also
178 for clobbers. Remove obsolete comment.
180 2018-09-11 Uros Bizjak <ubizjak@gmail.com>
182 * config/i386/i386.md (define_attr "type"): Remove mpxmov, mpxmk,
183 mpxchk, mpxld and mpxst types.
184 (define_attr length_immediate): Remove all processing of mpx types.
185 (define_attr prefix_0f): Ditto.
186 (define_attr memory): Ditto.
188 2018-09-11 Uros Bizjak <ubizjak@gmail.com>
190 * config/i386/i386.md (fyl2x_extend<mode>xf3_i387): Remove.
191 (log<mode>2): Change operand 1 predicate to general_operand.
192 Extend operand 1 to XFmode and generate logxf3 insn.
193 (log10<mode>2): Change operand 1 predicate to general_operand.
194 Extend operand 1 to XFmode and generate log10xf3 insn.
195 (log2<mode>2): Change operand 1 predicate to general_operand.
196 Extend operand 1 to XFmode and generate log2xf3 insn.
197 (fyl2xp1_extend<mode>xf3_i387): Remove.
198 (log1p<mode>2): Change operand 1 predicate to general_operand.
199 Extend operand 1 to XFmode and generate log1pxf3 insn.
200 (fxtract_extend<mode>xf3_i387): Remove.
201 (logb<mode>2): Change operand 1 predicate to general_operand.
202 Extend operand 1 to XFmode and generate logbxf3 insn.
203 (ilogb<mode>2): Change operand 1 predicate to general_operand.
204 Extend operand 1 to XFmode and generate fxtractxf3_i387 insn.
205 (significand<mode>2): Change operand 1 predicate to general_operand.
206 Extend operand 1 to XFmode and generate significandxf3 insn.
208 2018-09-11 Nathan Sidwell <nathan@acm.org>
210 * gcc.c (perror_with_name, pfatal_with_name): Delete.
211 (load_specs): Use fatal_error.
212 (DELETE_IF_ORDINARY, process_command): Use error.
213 (execute, run_attempt): Use fatal_error.
215 2018-09-11 Andrew Stubbs <ams@codesourcery.com>
217 * diagnostic-core.h (sorry_at): New prototype.
218 * diagnostic.c (sorry_at): New function.
220 2018-09-11 Aldy Hernandez <aldyh@redhat.com>
222 * tree-vrp.c (extract_range_from_binary_expr_1): Treat all divisions
223 by zero as VR_UNDEFINED.
225 2018-09-10 Uros Bizjak <ubizjak@gmail.com>
227 * config/i386/i386.md (<sincos>xf2): Rename from *<sincos>xf2_i387.
228 (*<sincos>_extend<mode>xf2_i387): Remove insn pattern.
229 (<sincos>mode2): New expander.
230 (sincos_extend<mode>xf3_i387): Remove insn pattern.
231 (sincos -> sin, cos splitters): Remove splitter patterns.
232 (sincos<mode>3): Change operand 2 predicate to general_operand.
233 Extend operand 2 to XFmode and generate sincosxf3 insn.
234 (fptanxf4_i387): Change mode of operands 0 and 3 to SFmode.
235 Change operand 3 predicate to const1_operand.
236 (fptan_extend<mode>xf4_i387): Remove insn pattern.
237 (tanxf2): Update operands in the call to fptanxf4_i387.
238 (tan<mode>2): Change operand 1 predicate to general_operand.
239 Extend operand 1 to XFmode and generate tanxf3 insn.
240 (atan2xf3): Rename from *fpatanxf3_i387.
241 (fpatan_extend<mode>xf3_i387): Remove insn pattern.
242 (atan2xf3): Remove expander.
243 (atan2<mode<3): Change operand 1 and 2 predicates to general_operand.
244 Extend operands 1 and 2 to XFmode and generate atan2xf3 insn.
245 (atan<mode>2): Change operand 1 predicate to general_operand.
246 Extend operand 1 to XFmode and generate atanxf3 insn.
248 2018-09-10 Uros Bizjak <ubizjak@gmail.com>
250 * config/i386/i386.md (x87/SSE constant load splitter): Use
251 memory_operand instead of nonimmediate_operand for input operand
254 2018-09-09 Uros Bizjak <ubizjak@gmail.com>
256 * config/i386/i386.md (float partial SSE register stall splitter): Move
257 splitter near its instruction pattern.
258 (float_extend partial SSE register stall splitter): Ditto.
259 (float_truncate partial SSE register stall splitter): Ditto.
261 2018-09-09 Hans-Peter Nilsson <hp@bitrange.com>
264 * config/mmix/mmix.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Redefine
265 to speculation_safe_value_not_needed.
268 * config/mmix/mmix.c (mmix_assemble_integer): Handle byte-size
269 non-CONST_INT rtx:es using assemble_integer_with_op ".byte".
270 (MMIX_CFUN_NEEDS_SAVED_EH_RETURN_ADDRESS): Don't call
271 leaf_function_p, instead use has_hard_reg_initial_val.
273 2018-09-09 Nathan Sidwell <nathan@acm.org>
275 * gcc.h (pfatal_with_name): Don't declare here.
276 * gcc.c (pfatal_with_name): Make static.
278 2018-09-09 Xianmiao Qu <xianmiao_qu@c-sky.com>
280 * config/csky/csky.md (*cskyv2_adddi3, *cskyv2_subdi3): Add
283 2018-09-08 John David Anglin <danglin@gcc.gnu.org>
285 PR rtl-optimization/85458
286 * sel-sched.c (sel_target_adjust_priority): Allow backend adjust
287 priority hook to reduce the priority of EXPR.
289 2018-09-07 Uros Bizjak <ubizjak@gmail.com>
291 * config/i386/i386.md (float<SWI48x:mode><MODEF:mode>2) Enable
292 DImode for x87 on 32bit targets. Conditionally disable x87 modes
293 with X87_ENABLE_FLOAT. Remove preparation code.
294 (*float<SWI48:mode><MODEF:mode>2): Rename from
295 *float<SWI48:mode><MODEF:mode>2_mixed. Handle x87, SSE and mixed
296 math using "enabled" attribute.
297 (*floatdi<MODEF:mode>2_i387): Rename from
298 *float<SWI48x:mode><MODEF:mode>2_i387. Handle only DImode and
299 enable for 32bit targets only.
300 (floatdi<X87MODEF:mode>2_i387_with_xmm pre-reload splitter): New
302 (floatdi<X87MODEF:mode>2_i387_with_xmm): Use register_operand
303 as operand 1 predicate. Rewrite as define_insn_and_split.
304 (floatdi<X87MODEF:mode>2_i387_with_xmm memory input splitter): Remove.
306 2018-09-06 Uros Bizjak <ubizjak@gmail.com>
308 * reg-stack.c (get_true_reg) <case FLOAT_TRUNCATE>: Reorder
309 to fallthru to FLOAT case.
311 2018-09-06 Will Schmidt <will_schmidt@vnet.ibm.com>
314 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Update logic
315 around folding of vec_sl to handle out of range shift values.
317 2018-09-06 Uros Bizjak <ubizjak@gmail.com>
319 * config/i386/i386.md (fix_trunc<mode>_fisttp_i387_1): Remove.
320 Update callers to gen_fix_trunc<mode>_i387_fisttp
321 (fix_trunc<mode>_i387_fisttp): Change operand 0 predicate to
322 nonimmediate_operand.
323 (fix_trunc<mode>_i387_fisttp_with_temp): Remove insn pattern
324 and corresponding splitters.
325 (*fix_trunc<mode>_i387_1): Always expand with fix_trunc<mode>_i387.
326 (fix_truncdi_i387): Change operand 0 predicate to nonimmediate_operand.
327 (fix_truncdi_i387_with_temp): Remove insn pattern
328 and corresponding splitters.
329 (fix_trunc<mode>_i387): Change operand 0 predicate to
330 nonimmediate_operand.
331 (fix_trunc<mode>_i387_with_temp): Remove insn pattern
332 and corresponding splitters.
333 (*fistdi2_1): Remove.
335 (fistdi2_with_temp): Remove insn pattern and corresponding splitters.
336 (lrintxfdi2): Remove expander. Reimplement as define_insn.
337 (*fist<mode>2_1): Remove.
338 (fist<mode>2): Ditto.
339 (fist<mode>2_with_temp): Remove insn pattern and corresponding
341 (lrintxf<mode>2): Remove expander. Reimplement as define_insn.
342 (*fist<mode>2_<rounding>_1): Always expand with fist<mode>2_<rounding>.
343 (fistdi2_<rounding>): Change operand 0 predicate to
344 nonimmediate_operand.
345 (fistdi2_<rounding>_with_temp): Remove insn pattern
346 and corresponding splitters.
347 (fist<mode>2_<rounding>): Change operand 0 predicate to
348 nonimmediate_operand.
349 (fist<mode>2_<rounding>_with_temp): Remove insn pattern
350 and corresponding splitters.
352 (*fixuns_trunc<mode>si2_avx512f_zext): Depend on TARGET_SSE_MATH.
354 2018-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
356 * varasm.c (output_constructor_regular_field): Check TYPE_SIZE_UNIT of
359 2018-09-06 Will Schmidt <will_schmidt@vnet.ibm.com>
361 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
362 early gimple folding of vec_splat().
363 * tree-vect-generic.c: Remove static from tree_vec_extract() definition.
364 * gimple-fold.h: Add an extern define for tree_vec_extract().
366 2018-09-06 Will Schmidt <will_schmidt@vnet.ibm.com>
368 * config/rs6000/rs6000.c (fold_mergehl_helper): Add types_compatible_p
369 wrappers around TREE_TYPE comparisons.
371 2018-09-06 Ilya Leoshkevich <iii@linux.ibm.com>
374 * config/s390/predicates.md: Add nonsym_memory_operand.
375 * config/s390/s390.c (s390_legitimize_cs_operand): If operand
376 contains a SYMBOL_REF, load it into an intermediate pseudo.
377 (s390_emit_compare_and_swap): Legitimize operand.
378 * config/s390/s390.md: Use the new nonsym_memory_operand
379 with UNSPECV_CAS patterns.
381 2018-09-06 Ilya Leoshkevich <iii@linux.ibm.com>
384 * config/s390/s390-passes.def: New file.
385 * config/s390/s390-protos.h (class rtl_opt_pass): Add forward
387 (make_pass_s390_early_mach): Add declaration.
388 * config/s390/s390.c (make_pass_s390_early_mach):
389 (s390_option_override): Remove dynamic registration.
390 * config/s390/t-s390: Add s390-passes.def.
392 2018-09-06 Ilya Leoshkevich <iii@linux.ibm.com>
394 * config/s390/s390.c (s390_decompose_constant_pool_ref):
395 Remove UNSPEC_LTREL_BASE check.
396 (annotate_constant_pool_refs): Likewise.
397 (find_constant_pool_ref): Likewise.
398 (find_ltrel_base): Removed.
399 (replace_ltrel_base): Removed.
400 (s390_mainpool_finish): Remove replace_ltrel_base call.
401 (s390_chunkify_start): Remove pending LTREL_BASE logic.
402 (s390_chunkify_finish): Remove replace_ltrel_base call.
403 * config/s390/s390.md: Remove UNSPEC_LTREL_BASE.
405 2018-09-06 Hans-Peter Nilsson <hp@axis.com>
408 * config/cris/cris.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Redefine
409 to speculation_safe_value_not_needed.
411 2018-09-05 Cesar Philippidis <cesar@codesourcery.com>
412 Bernd Schmidt <bernds_cb1@t-online.de>
414 * config/nvptx/nvptx-opts.h: New file.
415 * config/nvptx/nvptx.c (nvptx_file_start): Print the correct .target.
416 * config/nvptx/nvptx.h: Include "nvptx-opts.h".
418 (TARGET_SM35): New macro.
419 * config/nvptx/nvptx.md (atomic_fetch_<logic><mode>): Enable with the
421 * config/nvptx/nvptx.opt (ptx_isa, sm_30, sm_35): New enum and its
424 * doc/invoke.texi (Nvidia PTX Options): Document -misa.
426 2018-09-05 Uros Bizjak <ubizjak@gmail.com>
428 * config/i386/i386.md (truncdfsf2): Remove expander.
429 (truncdfsf2_with_temp): Ditto.
430 (truncxf<mode>2): Ditto.
431 (*truncdfsf_fast_mixed): Remove insn pattern.
432 (*truncdfsf_fast_i387): Ditto.
433 (*truncdfsf_mixed): Ditto.
434 (*truncdfsf_i387): Ditto.
435 (*truncdfsf2_i387_1): Ditto.
436 (*truncxfsf2_mixed): Ditto.
437 (*truncxfdf2_mixed): Ditto.
438 (*truncxf<mode>2_i387_noop): Ditto. Update callers
439 to call gen_truncxf<mode>2 instead.
440 (*truncxf<mode>2_i387): Remove.
441 (reg->reg splitters): Remove splitter pattern.
442 (reg->mem splitters): Ditto.
444 (truncdfsf2): New insn pattern.
445 (truncxf<mode>2): Ditto.
447 2018-09-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
449 * tree-ssa-math-opts.c (is_mult_by): New function.
450 (is_square_of): Use the above.
451 (optimize_recip_sqrt): New function.
452 (pass_cse_reciprocals::execute): Use the above.
454 2018-09-05 Richard Biener <rguenther@suse.de>
457 * tree-ssa-sccvn.c (rpo_elim::eliminate_push_avail): Make sure
458 to zero-init the emplaced vec.
460 2018-09-05 Martin Liska <mliska@suse.cz>
462 PR tree-optimization/87205
463 * tree-switch-conversion.c (pass_lower_switch::execute):
464 Group cases for switch statements.
466 2018-09-05 Richard Biener <rguenther@suse.de>
468 PR tree-optimization/87217
469 * tree-ssa-sccvn.c (vuse_valueize): New.
470 (vn_reference_lookup_pieces): Use it.
471 (vn_reference_lookup): Likewise.
473 2018-09-05 Nathan Sidwell <nathan@acm.org>
476 * stor-layout.c (place_field): Scan forwards to check last
477 bitfield when ms_bitfield_placement is in effect.
479 2018-09-05 Richard Biener <rguenther@suse.de>
482 * tree-vect-stmts.c (vectorizable_simd_clone_call): Fix bogus
485 2018-09-05 Siddhesh Poyarekar <siddhesh@sourceware.org>
486 Kugan Vivekanandarajah <kugan.vivekanandarajah@linaro.org>
488 * config/aarch64/falkor-tag-collision-avoidance.c: New file.
489 * config.gcc (extra_objs): Build it.
490 * config/aarch64/t-aarch64 (falkor-tag-collision-avoidance.o):
492 * config/aarch64/aarch64-passes.def
493 (pass_tag_collision_avoidance): New pass.
494 * config/aarch64/aarch64.c (qdf24xx_tunings): Add
495 AARCH64_EXTRA_TUNE_RENAME_LOAD_REGS to tuning_flags.
496 (aarch64_classify_address): Remove static qualifier.
497 (aarch64_address_info, aarch64_address_type): Move to...
498 * config/aarch64/aarch64-protos.h: ... here.
499 (make_pass_tag_collision_avoidance): New function.
500 * config/aarch64/aarch64-tuning-flags.def (rename_load_regs):
503 2018-09-05 Martin Liska <mliska@suse.cz>
505 * doc/gcov.texi: Update documentation of humar
507 * gcov.c (format_count): Print one decimal place, it provides
508 more fine number of situations like '1G' vs. '1.4G'.
510 2018-09-05 Martin Liska <mliska@suse.cz>
513 * config/rs6000/rs6000.opt: Mark the option as Deprecated.
514 * optc-gen.awk: Allow 'Var' for Deprecated options in order
515 to generate a MASK value.
517 2018-09-04 H.J. Lu <hongjiu.lu@intel.com>
520 * dwarf2out.c (based_loc_descr): Allow hard frame pointer even
521 if frame pointer isn't used.
522 (compute_frame_pointer_to_fb_displacement): Likewise.
524 2018-09-04 Jakub Jelinek <jakub@redhat.com>
527 * common/config/i386/i386-common.c (OPTION_MASK_ISA_XSAVEOPT_SET,
528 OPTION_MASK_ISA_XSAVES_SET, OPTION_MASK_ISA_XSAVEC_SET): Use
529 OPTION_MASK_ISA_XSAVE_SET instead of OPTION_MASK_ISA_XSAVE.
530 (OPTION_MASK_ISA_XSAVE_UNSET): Add OPTION_MASK_ISA_XSAVES_UNSET
531 and OPTION_MASK_ISA_XSAVEC_UNSET.
533 2018-09-04 Max Filippov <jcmvbkbc@gmail.com>
535 * config/xtensa/xtensa.c (xtensa_expand_atomic): Reorder AND and
536 XOR operations in NAND case.
538 2018-09-04 Aldy Hernandez <aldyh@redhat.com>
540 * wide-int-range.cc (wide_int_range_convert): New.
541 * wide-int-range.h (wide_int_range_convert): New.
542 * tree-vrp.c (extract_range_from_unary_expr): Abstract wide int
543 code into wide_int_range_convert.
544 (extract_range_into_wide_ints): Do not munge anti range constants
545 into the entire domain. Just return the range back.
547 2018-09-04 Martin Liska <mliska@suse.cz>
549 * genmatch.c (output_line_directive): Add new argument
551 (dt_simplify::gen_1): Encapsulate dump within __builtin_expect.
553 2018-09-04 Jonathan Wakely <jwakely@redhat.com>
555 * doc/invoke.texi (Option Summary): Add whitespace.
557 * doc/invoke.texi (Option Summary): Add -Waligned-new.
559 2018-09-04 Richard Biener <rguenther@suse.de>
561 PR tree-optimization/87211
562 * tree-ssa-sccvn.c (visit_phi): When value-numbering to a
563 backedge value we're supposed to treat as VARYING also number
564 the PHI to VARYING in case it got a different value-number already.
566 2018-09-04 Aldy Hernandez <aldyh@redhat.com>
568 * tree-vrp.c (vrp_can_optimize_bit_op): Remove.
569 (extract_range_from_binary_expr_1): Do not call
570 vrp_can_optimize_bit_op.
571 * wide-int-range.cc (wide_int_range_can_optimize_bit_op): Make
573 (wide_int_range_get_mask_and_bounds): New.
574 (wide_int_range_optimize_bit_op): New.
575 (wide_int_range_bit_ior): Call wide_int_range_optimize_bit_op.
576 (wide_int_range_bit_and): Same.
577 * wide-int-range.h (wide_int_range_can_optimize_bit_op): Remove.
578 (wide_int_range_optimize_bit_op): New.
579 (wide_int_range_get_mask_and_bounds): New.
581 2018-09-04 Richard Biener <rguenther@suse.de>
583 PR tree-optimization/87176
584 * tree-ssa-sccvn.c (visit_phi): Remove redundant allsame
585 variable. When value-numbering a virtual PHI node make sure
586 to not value-number to the backedge value.
588 2018-09-04 Jonathan Wakely <jwakely@redhat.com>
590 * doc/extend.texi (Long Long, Hex Floats): Document support for
591 long long and hex floats in more recent versions of ISO C++.
593 2018-09-03 Richard Biener <rguenther@suse.de>
595 PR tree-optimization/87177
596 * tree-ssa-sccvn.c (vuse_ssa_val): Revert previous change, keep
599 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
601 * bb-reorder.c (edge_order): Convert to C-qsort-style
602 tri-state comparator.
603 (reorder_basic_blocks_simple): Change std::stable_sort to gcc_stablesort.
605 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
607 * tree-loop-distribution.c (offset_cmp): Convert to C-qsort-style
608 tri-state comparator.
609 (fuse_memset_builtins): Change std::stable_sort to gcc_stablesort.
611 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
613 * sort.cc (struct sort_ctx): New field 'nlim'. Use it...
614 (mergesort): ... here as maximum count for using netsort.
615 (gcc_qsort): Set nlim to 3 if stable sort is requested.
616 (gcc_stablesort): New.
617 * system.h (gcc_stablesort): Declare.
619 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
621 * sort.cc (gcc_qsort) [CHECKING_P]: Call qsort_chk.
622 * system.h (qsort): Always redirect to gcc_qsort. Update comment.
623 * vec.c (qsort_chk): Do not call gcc_qsort. Update comment.
625 2018-09-03 Segher Boessenkool <segher@kernel.crashing.org>
627 * config/rs6000/rs6000.md (*mov<mode>_hardfloat32): Remove %U from the
628 lxsdx and stxsdx alternatives.
629 (*mov<mode>_hardfloat64): Ditto.
630 * config/rs6000/vsx.md (*vsx_extract_<mode>_store): Ditto.
632 2018-09-03 Richard Biener <rguenther@suse.de>
634 PR tree-optimization/87200
635 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Valueize a
638 2018-09-03 Martin Liska <mliska@suse.cz>
640 PR tree-optimization/87201
641 * tree-switch-conversion.c (switch_decision_tree::balance_case_nodes):
642 Fix parenthesis in an expression.
644 2018-09-03 Richard Biener <rguenther@suse.de>
646 PR tree-optimization/87197
647 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Mark the new def
648 visited. CSE the VN_INFO hashtable lookup.
650 PR tree-optimization/87169
651 * tree-ssa-sccvn.c (do_rpo_vn): When marking loops for not
652 iterating make sure there's no extra backedges from irreducible
653 regions feeding the header. Mark the destination block
656 2018-09-03 Martin Liska <mliska@suse.cz>
659 * common/common-target.def: Add TARGET_GET_VALID_OPTION_VALUES.
660 * common/common-targhooks.c (default_get_valid_option_values):
662 * common/common-targhooks.h (default_get_valid_option_values):
664 * common/config/i386/i386-common.c: Move processor_target_table
666 (ix86_get_valid_option_values): New function.
667 (TARGET_GET_VALID_OPTION_VALUES): New macro.
668 * config/i386/i386.c (struct ptt): Move to i386-common.c.
669 (PTA_*): Move all defined masks into i386-common.c.
670 (ix86_function_specific_restore): Use new processor_cost_table.
671 * config/i386/i386.h (struct ptt): Moved from i386.c.
672 (struct pta): Likewise.
673 * doc/tm.texi: Document new TARGET_GET_VALID_OPTION_VALUES.
674 * doc/tm.texi.in: Likewise.
675 * opt-suggestions.c (option_proposer::suggest_option):
676 Pass prefix to build_option_suggestions.
677 (option_proposer::get_completions): Likewise.
678 (option_proposer::build_option_suggestions): Use the new target
680 * opts.c (struct option_help_tuple): New struct.
681 (print_filtered_help): Use the new target hook.
683 2018-09-03 Martin Liska <mliska@suse.cz>
686 * predict.c (set_even_probabilities): Add likely_edges
687 argument and handle cases where we have precisely one
689 (combine_predictions_for_bb): Catch also likely_edges.
690 (tree_predict_by_opcode): Handle gswitch statements.
691 * tree-cfg.h (find_case_label_for_value): New declaration.
692 (find_taken_edge_switch_expr): Likewise.
693 * tree-switch-conversion.c (switch_decision_tree::balance_case_nodes):
694 Find pivot in decision tree based on probabily, not by number of
697 2018-09-02 Gerald Pfeifer <gerald@pfeifer.com>
699 * doc/standards.texi (Standards): Update Objective-C reference.
701 2018-09-01 Gerald Pfeifer <gerald@pfeifer.com>
703 * doc/install.texi (Prerequisites): Update link for MPC.
705 2018-09-01 Michael Matz <matz@suse.de>
707 PR tree-optimization/87074
708 * gimple-loop-jam.c (unroll_jam_possible_p): Check loop exit
709 PHIs for outer-loop uses.
711 2018-09-01 Gerald Pfeifer <gerald@pfeifer.com>
713 * doc/generic.texi (OpenMP): Adjust link to openmp.org.
714 * doc/invoke.texi (C Dialect Options): Ditto.
716 2018-09-01 Gerald Pfeifer <gerald@pfeifer.com>
718 * doc/install.texi (Prerequisites): Adjust link mpfr.org.
720 2018-08-31 Richard Biener <rguenther@suse.de>
722 PR tree-optimization/87168
723 * tree-ssa-sccvn.c (SSA_VAL): Add visited output parameter.
724 (rpo_elim::eliminate_avail): When OP was not visited it must
727 2018-08-31 David Malcolm <dmalcolm@redhat.com>
729 * tree-vrp.c (copy_value_range): Convert param "from" from
730 "value_range *" to "const value_range *".
731 (range_is_null): Likewise for param "vr".
732 (range_int_cst_p): Likewise.
733 (range_int_cst_singleton_p): Likewise.
734 (symbolic_range_p): Likewise.
735 (value_ranges_intersect_p): Likewise for both params.
736 (value_range_nonnegative_p): Likewise for param "vr".
737 (value_range_constant_singleton): Likewise.
738 (vrp_set_zero_nonzero_bits): Likewise for param "ar".
739 (extract_range_into_wide_ints): Likewise for param "vr".
740 (extract_range_from_multiplicative_op): Likewise for params "vr0"
742 (vrp_can_optimize_bit_op): Likewise.
743 (extract_range_from_binary_expr_1): Likewise for params "vr0_" and
745 (extract_range_from_unary_expr): Likewise.
746 (debug_value_range): Likewise for param "vr".
747 (value_range::dump): Add "const" qualifier.
748 (vrp_prop::check_array_ref): Convert local "vr" from
749 "value_range *" to "const value_range *".
750 (vrp_prop::check_mem_ref): Likewise.
751 (vrp_prop::visit_stmt): Likewise for local "old_vr".
752 (vrp_intersect_ranges_1): Likewise for param "vr_1".
753 (vrp_intersect_ranges): Likewise.
754 (simplify_stmt_for_jump_threading): Likewise for local "vr".
755 (vrp_prop::vrp_finalize): Likewise.
756 * tree-vrp.h (value_range::dump): Add "const" qualifier.
757 (vrp_intersect_ranges): Add "const" qualifier to params as above.
758 (extract_range_from_unary_expr): Likewise.
759 (value_range_constant_singleton): Likewise.
760 (symbolic_range_p): Likewise.
761 (copy_value_range): Likewise.
762 (extract_range_from_binary_expr_1): Likewise.
763 (range_int_cst_p): Likewise.
764 (vrp_set_zero_nonzero_bits): Likewise.
765 (range_int_cst_singleton_p): Likewise.
767 2018-08-31 Vlad Lazar <vlad.lazar@arm.com>
769 * config/aarch64/arm_neon.h (vabsd_s64): New.
770 (vnegd_s64): Likewise.
772 2018-08-31 Martin Jambor <mjambor@suse.cz>
774 * ipa-cp.c (estimate_local_effects): Replace wrong MAX with MIN.
776 2018-08-31 Martin Liska <mliska@suse.cz>
778 * ipa-icf.c (sem_item::add_type): Use
779 sem_item::m_type_hash_cache.
780 * ipa-icf.h: Move the cache from sem_item_optimizer
783 2018-08-31 Nathan Sidwell <nathan@acm.org>
785 * doc/extend.texi (Backwards Compatibility): Remove implicit
786 extern C leeway of () being (...).
788 2018-08-31 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
790 * ipa-inline.c (can_inline_edge_by_limits_p): Fix typos in comment.
792 2018-08-31 Segher Boessenkool <segher@kernel.crashing.org>
796 * config/rs6000/rs6000.md (lround<mode>di2): Gate on TARGET_FPRND.
798 2018-08-31 Jakub Jelinek <jakub@redhat.com>
801 * expmed.c (expand_mult_const): Use immed_wide_int_const instead of
802 gen_int_mode. Formatting fixes.
804 2018-08-30 Sandra Loosemore <sandra@codesourcery.com>
806 * target.def (custom_function_descriptors): Improve documentation.
807 * doc/tm.texi.in (Trampolines): Expand discussion of function
808 descriptors and move TARGET_CUSTOM_FUNCTION_DESCRIPTORS to the
809 beginning of the section.
810 * doc/tm.texi: Regenerated.
812 2018-08-30 Jose E. Marchesi <jose.marchesi@oracle.com>
814 * cfg.h (class auto_edge_flag): Spell out the template-id of the
815 base class in the initializer list. This is a workaround for
816 building with older GCC.
817 (class auto_bb_flag): Likewise.
819 2018-08-30 Aaron Sawdey <acsawdey@linux.ibm.com>
821 * config/rs6000/altivec.md (altivec_eq<mode>): Remove star.
822 (altivec_vcmpequ<VI_char>_p): Remove star.
823 * config/rs6000/rs6000-string.c (do_load_for_compare): Support
825 (expand_strncmp_vec_sequence): New function.
826 (emit_final_str_compare_vec): New function.
827 (expand_strn_compare): Add support for vector strncmp.
828 * config/rs6000/rs6000.opt (-mstring-compare-inline-limit): Change
829 length specification to bytes.
830 * config/rs6000/vsx.md (vsx_ld_elemrev_v16qi_internal): Remove star.
831 (vcmpnezb_p): New pattern.
832 * doc/invoke.texi (RS/6000 and PowerPC Options): Update documentation
833 for option -mstring-compare-inline-limit.
835 2018-08-30 Thiago Macieira <thiago.macieira@intel.com>
837 * config/i386/i386.c (PTA_WESTMERE): Remove PTA_AES.
838 (PTA_SKYLAKE): Add PTA_AES.
839 (PTA_GOLDMONT): Likewise.
841 2018-08-29 Jan Hubicka <jh@suse.cz>
844 * lto-opts.c (lto_write_options): Always stream PIC/PIE mode.
845 * lto-wrapper.c (merge_and_complain): Fix merging of PIC/PIE.
847 2018-08-29 Jan Hubicka <jh@suse.cz>
849 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not follow
851 (hash_tree): Do not visit TYPE_STUB_DECL.
852 * tree-streamer-out.c (write_ts_type_common_tree_pointers): Do not
853 stream TYPE_STUB_DECL.
854 * tree-streamer-in.c (lto_input_ts_type_common_tree_pointers): Likewise.
855 * ipa-utils.h (type_with_linkage_p): Do not rely on TYPE_STUB_DECL
856 after free_lang_data.
857 (type_in_anonymous_namespace_p): Likewise.
859 2018-08-29 Jan Hubicka <jh@suse.cz>
861 * sreal.h (SREAL_PART_BITS): Change to 31; remove seemingly unnecessary
862 comment that it has to be even number.
863 (class sreal): Change m_sig type to int32_t.
864 * sreal.c (sreal::dump, sreal::to_int, opreator+, operator-): Use
865 int64_t for temporary calculations.
866 (sreal_verify_basics): Drop one bit from minimum and maximum.
868 2018-08-30 Richard Biener <rguenther@suse.de>
870 PR tree-optimization/87147
871 * tree-ssa-sccvn.c (SSA_VISITED): New function.
872 (visit_phi): When the degenerate result is from the backedge and
873 we didn't visit its definition yet drop to VARYING.
874 (do_rpo_vn): Properly mark blocks with incoming backedges as executable.
876 2018-08-29 Jan Hubicka <jh@suse.cz>
878 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not walk
880 (hash_tree): Likewise.
882 2018-08-29 Jan Hubicka <jh@suse.cz>
884 * tree.c (find_decls_types_r): Walk also TYPE_NEXT_PTR_TO
885 and TYPE_NEXT_REF_TO.
887 2018-08-29 Jan Hubicka <jh@suse.cz>
889 * sreal.h (SREAL_PART_BITS): Change to 31; remove seemingly unnecessary
890 comment that it has to be even number.
891 (class sreal): Change m_sig type to int32_t.
892 * sreal.c (sreal::dump, sreal::to_int, opreator+, operator-): Use
893 int64_t for temporary calculations.
894 (sreal_verify_basics): Drop one bit from minimum and maximum.
896 2018-08-30 Tamar Christina <tamar.christina@arm.com>
898 * config/aarch64/aarch64.c (aarch64_expand_movmem): Set TImode max.
900 2018-08-30 Vlad Lazar <vlad.lazar@arm.com>
903 * expmed.c (canonicalize_comparison): Use wi::sub instead of wi::add
904 if to_add is negative.
906 2018-08-29 Bernd Edlinger <bernd.edlinger@hotmail.de>
909 * builtins.c (c_strlen): Improve range checks.
911 2018-08-29 Martin Sebor <msebor@redhat.com>
912 Jeff Law <law@redhat.com>
914 PR tree-optimization/86714
915 PR tree-optimization/86711
916 * builtins.c (c_strlen): Add arguments to call to string_constant.
917 * expr.c (string_constant): Add argument. Detect missing nul
918 terminator and outermost declaration it's missing in.
919 * expr.h (string_constant): Add argument.
920 * fold-const.c (read_from_constant_string): Add arguments to call to
922 (c_getstr): Likewise.
923 * tree-ssa-forwprop.c (simplify_builtin_call): Likewise.
925 * tree-ssa-strlen.c (get_stridx): Likewise.
927 2018-08-29 Jan Hubicka <jh@suse.cz>
929 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers):
930 Do not stream DECL_VINDEX.
931 * tree-streamer-out.c (write_ts_function_decl_tree_pointers): Likewise.
932 * tree.c (free_lang_data_in_decl): Clear DECL_VINDEX.
933 (decl_function_context): Use DECL_VIRTUAL_P rather than DECL_VINDEX.
935 2018-08-29 Richard Biener <rguenther@suse.de>
937 * tree-ssa-sccvn.c (vuse_ssa_val): Return NULL for unvisited
938 virtual operands that are not default defs to honor region
940 (rpo_vn_valueize): Remove ineffective code here.
942 2018-08-29 Richard Biener <rguenther@suse.de>
944 PR tree-optimization/87132
945 * tree-ssa-alias.c (get_continuation_for_phi): Do not translate
946 when skipping defs reachable over backedges.
948 2018-08-29 Richard Biener <rguenther@suse.de>
950 * tree-core.h: Document use of deprecated_flag in SSA_NAME.
951 * tree.h (SSA_NAME_POINTS_TO_READONLY_MEMORY): Define.
952 * tree-into-ssa.c (pass_build_ssa::execute): Initialize
953 function parameters SSA_NAME_POINTS_TO_READONLY_MEMORY from fnspec.
954 * tree-ssa-sccvn.c (const_parms, init_const_parms): Remove.
955 (vn_reference_lookup_3): Remove use of const_parms.
956 (free_rpo_vn): Do not free const_parms.
957 (do_rpo_vn): Do not call init_const_parms.
958 * tree-ssa-alias.c (refs_may_alias_p_1): Honor
959 SSA_NAME_POINTS_TO_READONLY_MEMORY.
960 (call_may_clobber_ref_p_1): Likewise.
962 2018-08-29 Alexander Monakov <amonakov@ispras.ru>
965 * invoke.texi (Optimization Options): List -ftree-scev-cprop.
967 (-ftree-scev-cprop): Document.
969 2018-08-29 Jan Hubicka <jh@suse.cz>
971 * sreal.h (normalize, normalize_up, normalize_down): Add new_sig/new_exp
973 (sreal constructor): Update.
974 * sreal.c (sreal:operator+, sreal:operator-, sreal:operator*,
975 sreal:operator/): Update.
977 2018-08-29 Martin Liska <mliska@suse.cz>
979 * tree-switch-conversion.c (switch_conversion::expand):
980 Strenghten assumption about gswitch statements.
982 2018-08-29 Richard Biener <rguenther@suse.de>
984 PR tree-optimization/87117
985 * tree-ssa-sccvn.c (eliminate_dom_walker::eliminate_stmt): Only
986 re-value-number released SSA VDEFs.
988 2018-08-29 Richard Biener <rguenther@suse.de>
990 PR tree-optimization/87126
991 * tree-ssa-sccvn.c (vn_reference_insert): Remove assert.
993 2018-08-28 Jim Wilson <jimw@sifive.com>
995 * config/riscv/pic.md: Rewrite.
996 * config/riscv/riscv.c (riscv_address_insns): Return cost of 3 for
998 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Delete.
999 (SOFTF, default_load, softload, softstore): New.
1001 2018-08-28 Jeff Law <law@redhat.com>
1003 * fold-const.c (fold_binary_loc): Remove recently added assert.
1005 2018-08-28 Joern Rennecke <joern.rennecke@riscy-ip.com>
1007 * genpreds.c (write_predicate_subfunction): Also add ATTRIBUTE_UNUSED
1008 to OP parmeter of generated function.
1010 2018-08-28 MCC CS <deswurstes@users.noreply.github.com>
1012 PR tree-optimization/87009
1013 * match.pd: Add boolean optimizations.
1015 2018-08-28 Martin Sebor <msebor@redhat.com>
1018 * calls.c (alloc_max_size): Treat HOST_WIDE_INT special.
1019 * gimple-ssa-warn-alloca.c (adjusted_warn_limit): New function.
1020 (pass_walloca::gate): Use it.
1021 (alloca_call_type): Same.
1022 (pass_walloca::execute): Same.
1023 * stor-layout.c (layout_decl): Treat HOST_WIDE_INT special.
1025 2018-08-28 David Malcolm <dmalcolm@redhat.com>
1027 * dumpfile.h (ATTRIBUTE_GCC_DUMP_PRINTF): Change version check on
1028 GCC_VERSION for usage of "__gcc_dump_printf__" format from
1031 2018-08-28 Richard Biener <rguenther@suse.de>
1033 PR tree-optimization/87124
1034 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Guard against
1035 constants before looking up avail.
1037 2018-08-28 Jakub Jelinek <jakub@redhat.com>
1040 * calls.c (maybe_warn_nonstring_arg): Punt early if
1041 warn_stringop_overflow is zero. Don't call get_range_strlen
1042 on 3rd argument, keep iterating until lenrng[1] is INTEGER_CST.
1043 Swap comparison operands to have constants on rhs. Only use
1044 lenrng[1] if non-NULL and INTEGER_CST. Don't uselessly
1045 increment lenrng[0].
1047 2018-08-28 Richard Sandiford <richard.sandiford@arm.com>
1049 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Fix unguarded
1050 use of tree_to_shwi. Remove duplicated test for the size being
1051 a whole number of bytes.
1053 2018-08-28 Richard Biener <rguenther@suse.de>
1055 PR tree-optimization/87117
1056 * tree-ssa-sccvn.c (eliminate_dom_walker::eliminate_cleanup):
1057 Handle removed stmt without LHS (GIMPLE_NOP).
1059 2018-08-28 Richard Biener <rguenther@suse.de>
1061 PR tree-optimization/87117
1062 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Exclude
1063 void which is is_gimple_reg_type by checking for COMPLETE_TYPE_P.
1065 2018-08-28 Richard Biener <rguenther@suse.de>
1067 PR tree-optimization/87117
1068 * tree-ssa-pre.c (compute_avail): Do not make expressions
1069 with predicated values available.
1070 (get_expr_value_id): Assert we do not run into predicated value
1073 2018-08-28 Richard Biener <rguenther@suse.de>
1075 PR tree-optimization/87117
1076 * tree-ssa-operands.c (add_stmt_operand): STRING_CST may
1077 get virtual operands.
1078 (get_expr_operands): Handle STRING_CST like other decls.
1080 2018-08-28 Martin Liska <mliska@suse.cz>
1082 * tree.h: Update documentation of fndecl_built_in_p
1086 2018-08-27 Jeff Law <law@redhat.com>
1087 PR tree-optimization/87110
1088 * tree-ssa-dse.c (compute_trims): Handle non-constant
1091 2018-08-27 Martin Sebor <msebor@redhat.com>
1093 PR tree-optimization/86914
1094 * tree-ssa-strlen.c (maybe_set_strlen_range): Avoid MEM_REF.
1096 2018-08-27 Martin Sebor <msebor@redhat.com>
1098 PR tree-optimization/87112
1099 * builtins.c (expand_builtin_strnlen): Convert c_strlen result to
1100 the type of the bound argument.
1102 2018-08-27 Jeff Law <law@redhat.com>
1104 * tree-ssa-dse.c (compute_trims): Handle case where the reference's
1105 type does not have a TYPE_SIZE_UNIT.
1107 2018-08-27 Steve Ellcey <sellcey@cavium.com>
1109 * config/aarch64/aarch64-speculation.cc: Replace include of cfg.h
1110 with include of backend.h.
1112 2018-08-27 Richard Biener <rguenther@suse.de>
1114 PR tree-optimization/86927
1115 * tree-vect-loop.c (vect_create_epilog_for_reduction): Properly
1116 use const cond reduction code.
1118 2018-08-27 Alexander Monakov <amonakov@ispras.ru>
1120 PR tree-optimization/85758
1121 * match.pd ((X & Y) ^ Y): Add :s qualifier to inner expression.
1123 2018-08-27 David Malcolm <dmalcolm@redhat.com>
1126 * diagnostic-show-locus.c (class layout_range): Update for
1127 conversion of show_caret_p to a tri-state.
1128 (layout_range::layout_range): Likewise.
1129 (make_range): Likewise.
1130 (layout::maybe_add_location_range): Likewise.
1131 (layout::should_print_annotation_line_p): Don't show annotation
1132 lines for ranges that are SHOW_LINES_WITHOUT_RANGE.
1133 (layout::get_state_at_point): Update for conversion of
1134 show_caret_p to a tri-state. Bail out early for
1135 SHOW_LINES_WITHOUT_RANGE, so that such ranges don't affect
1136 underlining or source colorization.
1137 (gcc_rich_location::add_location_if_nearby): Update for conversion
1138 of show_caret_p to a tri-state.
1139 (selftest::test_one_liner_multiple_carets_and_ranges): Likewise.
1140 (selftest::test_one_liner_fixit_replace_equal_secondary_range):
1142 (selftest::test_one_liner_labels): Likewise.
1143 * gcc-rich-location.c (gcc_rich_location::add_expr): Update for
1144 conversion of show_caret_p to a tri-state.
1145 * pretty-print.c (text_info::set_location): Likewise.
1146 * pretty-print.h (text_info::set_location): Likewise.
1147 * substring-locations.c (format_warning_n_va): Likewise.
1148 * tree-diagnostic.c (default_tree_printer): Likewise.
1149 * tree-pretty-print.c (newline_and_indent): Likewise.
1151 2018-08-27 David Malcolm <dmalcolm@redhat.com>
1154 * diagnostic-show-locus.c (get_line_span_for_fixit_hint): Show the
1155 line above for line-insertion fix-it hints.
1156 (selftest::test_fixit_insert_containing_newline): Update the
1157 expected results, and add a test with line-numbering enabled.
1159 2018-08-27 Martin Liska <mliska@suse.cz>
1162 * sanopt.c (sanitize_rewrite_addressable_params): Ignore
1163 params with DECL_HAS_VALUE_EXPR_P.
1165 2018-08-27 Martin Liska <mliska@suse.cz>
1167 * config/i386/i386.c (ix86_expand_set_or_movmem): Dump
1168 selected expansion strategy.
1170 2018-08-27 Martin Liska <mliska@suse.cz>
1172 * builtins.h (is_builtin_fn): Remove and fndecl_built_in_p.
1173 * builtins.c (is_builtin_fn): Likewise.
1174 * attribs.c (diag_attr_exclusions): Use new function
1175 fndecl_built_in_p and remove check for FUNCTION_DECL if
1177 (builtin_mathfn_code): Likewise.
1178 (fold_builtin_expect): Likewise.
1179 (fold_call_expr): Likewise.
1180 (fold_builtin_call_array): Likewise.
1181 (fold_call_stmt): Likewise.
1182 (set_builtin_user_assembler_name): Likewise.
1183 (is_simple_builtin): Likewise.
1184 * calls.c (gimple_alloca_call_p): Likewise.
1185 (maybe_warn_nonstring_arg): Likewise.
1186 * cfgexpand.c (expand_call_stmt): Likewise.
1187 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Likewise.
1188 (cgraph_edge::verify_corresponds_to_fndecl): Likewise.
1189 (cgraph_node::verify_node): Likewise.
1190 * cgraphclones.c (build_function_decl_skip_args): Likewise.
1191 (cgraph_node::create_clone): Likewise.
1192 * config/arm/arm.c (arm_insert_attributes): Likewise.
1193 * config/i386/i386.c (ix86_gimple_fold_builtin): Likewise.
1194 * dse.c (scan_insn): Likewise.
1195 * expr.c (expand_expr_real_1): Likewise.
1196 * fold-const.c (operand_equal_p): Likewise.
1197 (fold_binary_loc): Likewise.
1198 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
1199 * gimple-low.c (lower_stmt): Likewise.
1200 * gimple-pretty-print.c (dump_gimple_call): Likewise.
1201 * gimple-ssa-warn-restrict.c (wrestrict_dom_walker::check_call): Likewise.
1202 * gimple.c (gimple_build_call_from_tree): Likewise.
1203 (gimple_call_builtin_p): Likewise.
1204 (gimple_call_combined_fn): Likewise.
1205 * gimplify.c (gimplify_call_expr): Likewise.
1206 (gimple_boolify): Likewise.
1207 (gimplify_modify_expr): Likewise.
1208 (gimplify_addr_expr): Likewise.
1209 * hsa-gen.c (gen_hsa_insns_for_call): Likewise.
1210 * ipa-cp.c (determine_versionability): Likewise.
1211 * ipa-fnsummary.c (compute_fn_summary): Likewise.
1212 * ipa-param-manipulation.c (ipa_modify_formal_parameters): Likewise.
1213 * ipa-split.c (visit_bb): Likewise.
1214 (split_function): Likewise.
1215 * ipa-visibility.c (cgraph_externally_visible_p): Likewise.
1216 * lto-cgraph.c (input_node): Likewise.
1217 * lto-streamer-out.c (write_symbol): Likewise.
1218 * omp-low.c (setjmp_or_longjmp_p): Likewise.
1219 (lower_omp_1): Likewise.
1220 * predict.c (strip_predict_hints): Likewise.
1221 * print-tree.c (print_node): Likewise.
1222 * symtab.c (symtab_node::output_to_lto_symbol_table_p): Likewise.
1223 * trans-mem.c (is_tm_irrevocable): Likewise.
1224 (is_tm_load): Likewise.
1225 (is_tm_simple_load): Likewise.
1226 (is_tm_store): Likewise.
1227 (is_tm_simple_store): Likewise.
1228 (is_tm_abort): Likewise.
1229 (tm_region_init_1): Likewise.
1230 * tree-call-cdce.c (gen_shrink_wrap_conditions): Likewise.
1231 * tree-cfg.c (verify_gimple_call): Likewise.
1232 (move_stmt_r): Likewise.
1233 (stmt_can_terminate_bb_p): Likewise.
1234 * tree-eh.c (lower_eh_constructs_2): Likewise.
1235 * tree-if-conv.c (if_convertible_stmt_p): Likewise.
1236 * tree-inline.c (remap_gimple_stmt): Likewise.
1237 (copy_bb): Likewise.
1238 (estimate_num_insns): Likewise.
1239 (fold_marked_statements): Likewise.
1240 * tree-sra.c (scan_function): Likewise.
1241 * tree-ssa-ccp.c (surely_varying_stmt_p): Likewise.
1242 (optimize_stack_restore): Likewise.
1243 (pass_fold_builtins::execute): Likewise.
1244 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
1245 (mark_all_reaching_defs_necessary_1): Likewise.
1246 * tree-ssa-dom.c (dom_opt_dom_walker::optimize_stmt): Likewise.
1247 * tree-ssa-forwprop.c (simplify_builtin_call): Likewise.
1248 (pass_forwprop::execute): Likewise.
1249 * tree-ssa-loop-im.c (stmt_cost): Likewise.
1250 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
1251 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Likewise.
1252 * tree-ssa-strlen.c (get_string_length): Likewise.
1253 * tree-ssa-structalias.c (handle_lhs_call): Likewise.
1254 (find_func_aliases_for_call): Likewise.
1255 * tree-ssa-ter.c (find_replaceable_in_bb): Likewise.
1256 * tree-stdarg.c (optimize_va_list_gpr_fpr_size): Likewise.
1257 * tree-tailcall.c (find_tail_calls): Likewise.
1258 * tree.c (need_assembler_name_p): Likewise.
1259 (free_lang_data_in_decl): Likewise.
1260 (get_call_combined_fn): Likewise.
1261 * ubsan.c (is_ubsan_builtin_p): Likewise.
1262 * varasm.c (incorporeal_function_p): Likewise.
1263 * tree.h (DECL_BUILT_IN): Remove and replace with
1265 (DECL_BUILT_IN_P): Transfort to fndecl_built_in_p.
1266 (fndecl_built_in_p): New.
1268 2018-08-27 Martin Liska <mliska@suse.cz>
1270 PR tree-optimization/86847
1271 * tree-switch-conversion.c (switch_decision_tree::dump_case_nodes):
1272 Dump also subtree probability.
1273 (switch_decision_tree::do_jump_if_equal): New function.
1274 (switch_decision_tree::emit_case_nodes): Handle special
1275 situations in balanced tree that can be emitted much simpler.
1276 Fix calculation of probabilities that happen in tree expansion.
1277 * tree-switch-conversion.h (struct cluster): Add
1279 (struct simple_cluster): Likewise.
1280 (struct case_tree_node): Add new function has_child.
1281 (do_jump_if_equal): New.
1283 2018-08-27 Martin Liska <mliska@suse.cz>
1285 * tree-switch-conversion.c (bit_test_cluster::find_bit_tests):
1286 Add new argument to bit_test_cluster constructor.
1287 (bit_test_cluster::emit): Set bits really number of values
1289 (bit_test_cluster::hoist_edge_and_branch_if_true): Add
1290 probability argument.
1291 * tree-switch-conversion.h (struct bit_test_cluster):
1292 Add m_handles_entire_switch.
1294 2018-08-27 Martin Liska <mliska@suse.cz>
1296 PR tree-optimization/86702
1297 * tree-switch-conversion.c (jump_table_cluster::emit):
1298 Make probabilities even for values in jump table
1299 according to number of cases handled.
1300 (switch_decision_tree::compute_cases_per_edge): Pass
1301 argument to reset_out_edges_aux function.
1302 (switch_decision_tree::analyze_switch_statement): Likewise.
1303 * tree-switch-conversion.h (switch_decision_tree::reset_out_edges_aux):
1306 2018-08-27 Martin Liska <mliska@suse.cz>
1308 * cfgexpand.c (expand_asm_stmt): Use label_to_block and pass
1309 cfun argument explicitly.
1310 * gimple-pretty-print.c (dump_gimple_switch): Likewise.
1311 * hsa-gen.c (gen_hsa_insns_for_switch_stmt): Use new
1312 function gimple_switch_default_bb.
1313 (convert_switch_statements):
1315 * ipa-fnsummary.c (set_switch_stmt_execution_predicate):
1316 * stmt.c (label_to_block_fn): Use label_to_block and pass
1317 cfun argument explicitly and use gimple_switch_label_bb.
1318 (expand_case): Likewise.
1319 * tree-cfg.c (lower_phi_internal_fn): Use label_to_block and pass
1320 cfun argument explicitly. Likewise.
1321 (make_edges_bb): Likewise.
1322 (make_cond_expr_edges): Likewise.
1323 (get_cases_for_edge): Likewise.
1324 (make_gimple_switch_edges): Likewise.
1325 (label_to_block_fn): Likewise.
1326 (label_to_block): Likewise.
1327 (make_goto_expr_edges): Likewise.
1328 (make_gimple_asm_edges): Likewise.
1329 (main_block_label): Likewise.
1330 (group_case_labels_stmt): Likewise.
1331 (find_taken_edge_computed_goto): Likewise.
1332 (find_taken_edge_switch_expr): Likewise.
1333 (gimple_verify_flow_info): Likewise.
1334 (gimple_redirect_edge_and_branch): Likewise.
1335 (gimple_switch_label_bb): New function.
1336 (gimple_switch_default_bb): Likewise.
1337 (gimple_switch_edge): Likewise.
1338 (gimple_switch_default_edge): Likewise.
1339 * tree-cfg.h (label_to_block_fn): Remove and replace ...
1340 (label_to_block): ... with this.
1341 (gimple_switch_label_bb): New.
1342 (gimple_switch_default_bb): Likewise.
1343 (gimple_switch_edge): Likewise.
1344 (gimple_switch_default_edge): Likewise.
1345 * tree-cfgcleanup.c (convert_single_case_switch): Use
1346 new gimple functions and pass new argument to label_to_block.
1347 (cleanup_control_flow_bb):
1348 * tree-eh.c (make_eh_dispatch_edges): Use label_to_block and pass
1349 cfun argument explicitly.
1350 (make_eh_edges): Likewise.
1351 (redirect_eh_dispatch_edge): Likewise.
1352 (lower_resx): Likewise.
1353 (lower_eh_dispatch): Likewise.
1354 (maybe_remove_unreachable_handlers): Likewise.
1355 (unsplit_eh): Likewise.
1356 (cleanup_empty_eh): Likewise.
1357 (verify_eh_edges): Likewise.
1358 (verify_eh_dispatch_edge): Likewise.
1359 * tree-ssa-dom.c (record_edge_info): Likewise.
1360 * tree-ssa-forwprop.c (simplify_gimple_switch_label_vec): Likewise.
1361 * tree-ssa-threadedge.c (thread_around_empty_blocks): Likewise.
1362 (thread_through_normal_block): Likewise.
1363 * tree-ssa-uncprop.c (associate_equivalences_with_edges): Likewise.
1364 * tree-ssa-uninit.c (convert_control_dep_chain_into_preds):
1365 * tree-switch-conversion.c (switch_conversion::collect): Use new
1367 (switch_conversion::check_final_bb): Likewise.
1368 (switch_conversion::gather_default_values): Pass new argument
1370 (switch_conversion::build_constructors): Likewise.
1371 (switch_decision_tree::compute_cases_per_edge): Use new
1372 gimple_switch_edge function.
1373 (switch_decision_tree::analyze_switch_statement): Pass new argument
1375 (switch_decision_tree::try_switch_expansion): Use
1376 gimple_switch_default_edge.
1377 * tree-vrp.c (find_switch_asserts): Pass new argument
1379 * vr-values.c (vr_values::vrp_visit_switch_stmt): Likewise.
1380 (vr_values::simplify_switch_using_ranges): Likewise.
1382 2018-08-27 Richard Biener <rguenther@suse.de>
1384 * cfganal.h (rev_post_order_and_mark_dfs_back_seme): Declare.
1385 * cfganal.c (rev_post_order_and_mark_dfs_back_seme): New function.
1387 * tree-ssa-sccvn.h (struct vn_pval): New structure.
1388 (struct vn_nary_op_s): Add unwind_to member. Add
1389 predicated_values flag and put result into a union together
1390 with a linked list of vn_pval.
1391 (struct vn_ssa_aux): Add name member to make maintaining
1392 a map of SSA name to vn_ssa_aux possible. Remove no longer
1393 needed info, dfsnum, low, visited, on_sccstack, use_processed
1394 and range_info_anti_range_p members.
1395 (run_scc_vn, vn_eliminate, free_scc_vn, vn_valueize): Remove.
1396 (do_rpo_vn, run_rpo_vn, eliminate_with_rpo_vn, free_rpo_vn):
1398 (vn_valueize): New global.
1399 (vn_context_bb): Likewise.
1400 (VN_INFO_RANGE_INFO, VN_INFO_ANTI_RANGE_P, VN_INFO_RANGE_TYPE,
1401 VN_INFO_PTR_INFO): Remove.
1402 * tree-ssa-sccvn.c: ... (rewrite)
1403 (pass_fre::execute): For -O2+ initialize loops and run
1404 RPO VN in optimistic mode (iterating). For -O1 and -Og
1405 run RPO VN in non-optimistic mode.
1406 * params.def (PARAM_SCCVN_MAX_SCC_SIZE): Remove.
1407 (PARAM_RPO_VN_MAX_LOOP_DEPTH): Add.
1408 * doc/invoke.texi (sccvn-max-scc-size): Remove.
1409 (rpo-vn-max-loop-depth): Document.
1410 * tree-ssa-alias.c (walk_non_aliased_vuses): Stop walking
1411 when valuezing the VUSE signals we walked out of the region.
1412 * tree-ssa-pre.c (phi_translate_1): Ignore predicated values.
1413 (phi_translate): Set VN context block to use for availability
1415 (compute_avail): Likewise.
1416 (pre_valueize): New function.
1417 (pass_pre::execute): Adjust to the RPO VN API.
1419 * tree-ssa-loop-ivcanon.c: Include tree-ssa-sccvn.h.
1420 (propagate_constants_for_unrolling): Remove.
1421 (tree_unroll_loops_completely): Perform value-numbering
1422 on the unrolled bodies loop parent.
1424 2018-08-27 Richard Biener <rguenther@suse.de>
1426 * tree-ssa-pre.c (compute_antic): Re-use inverted postorder
1427 for partial antic compute.
1429 2018-08-27 Jakub Jelinek <jakub@redhat.com>
1431 PR rtl-optimization/87065
1432 * combine.c (simplify_if_then_else): Formatting fix.
1433 (if_then_else_cond): Guard MULT optimization with SCALAR_INT_MODE_P
1435 (known_cond): Don't return const_true_rtx for vector modes. Use
1436 CONST0_RTX instead of const0_rtx. Formatting fixes.
1438 2018-08-27 Martin Liska <mliska@suse.cz>
1440 PR gcov-profile/87069
1441 * gcov.c (process_file): Record files already processed
1442 and warn about a file being processed multiple times.
1444 2018-08-27 Martin Liska <mliska@suse.cz>
1447 * config/aarch64/aarch64.c (aarch64_override_options_internal):
1448 Set default values for x_aarch64_*_string strings.
1449 * config/aarch64/aarch64.opt: Remove --{march,mcpu,mtune}==
1450 prefix. For -mabi do not print '=ABI' in help and use
1451 <option_value> format for -msve-vector-bits and -moverride
1454 2018-08-26 Jeff Law <law@redhat.com>
1456 * config/mips/frame-header-opt.c: Include "backend.h" rather than
1459 2018-08-26 Marek Polacek <polacek@redhat.com>
1461 PR c++/87029, Implement -Wredundant-move.
1462 * doc/invoke.texi: Document -Wredundant-move.
1464 2018-08-25 Martin Sebor <msebor@redhat.com>
1466 PR tree-optimization/87059
1467 * builtins.c (expand_builtin_strncmp): Convert MIN_EXPR operand
1468 to the same type as the other.
1469 * fold-const.c (fold_binary_loc): Assert expectation.
1471 2018-08-25 Iain Sandoe <iain@sandoe.co.uk>
1473 * config/darwin.c (machopic_legitimize_pic_address): Clean up
1474 extraneous parentheses, dead code section and formatting.
1476 2018-08-24 David Malcolm <dmalcolm@redhat.com>
1479 * diagnostic-show-locus.c (layout::layout): Ensure the margin is
1480 wide enough for jumps in the line-numbering to be visible.
1481 (layout::print_gap_in_line_numbering): New member function.
1482 (layout::calculate_line_spans): When using line numbering, merge
1483 line spans that are only 1 line apart.
1484 (diagnostic_show_locus): When printing line numbers, show gaps in
1485 line numbering directly, rather than printing headers.
1486 (selftest::test_diagnostic_show_locus_fixit_lines): Add test of
1487 line-numbering with multiple line spans.
1488 (selftest::test_fixit_insert_containing_newline_2): Add test of
1489 line-numbering, in which the spans are close enough to be merged.
1491 2018-08-24 Aldy Hernandez <aldyh@redhat.com>
1493 * gimple-ssa-evrp-analyze.c (set_ssa_range_info): Pass value_range
1494 to range_includes_zero_p. Do not special case VR_ANTI_RANGE.
1495 * tree-vrp.c (range_is_nonnull): Remove.
1496 (range_includes_zero_p): Accept value_range instead of min/max.
1497 (extract_range_from_binary_expr_1): Do not early bail on
1499 Use range_includes_zero_p instead of range_is_nonnull.
1500 (extract_range_from_unary_expr): Use range_includes_zero_p instead
1501 of range_is_nonnull.
1502 (vrp_meet_1): Pass value_range to range_includes_zero_p. Do not
1503 special case VR_ANTI_RANGE.
1504 (vrp_finalize): Same.
1505 * tree-vrp.h (range_includes_zero_p): Pass value_range as argument
1507 (range_is_nonnull): Remove.
1508 * vr-values.c (vrp_stmt_computes_nonzero): Use
1509 range_includes_zero_p instead of range_is_nonnull.
1510 (extract_range_basic): Pass value_range to range_includes_zero_p
1511 instead of range_is_nonnull.
1513 2018-08-24 Uros Bizjak <ubizjak@gmail.com>
1515 * emit-rtl.c (init_emit_once): Do not emit MODE_POINTER_BOUNDS RTXes.
1516 * emit-rtl.h (rtl_data): Remove return_bnd.
1517 * explow.c (trunc_int_for_mode): Do not handle POINTER_BOUNDS_MODE_P.
1518 * function.c (diddle_return_value): Do not handle crtl->return_bnd.
1519 * genmodes.c (complete_mode): Do not handle MODE_POINTER_BOUNDS.
1520 (POINTER_BOUNDS_MODE): Remove definition.
1521 (make_pointer_bounds_mode): Remove.
1522 (get_mode_class): Do not handle MODE_POINTER_BOUNDS.
1523 * machmode.h (POINTER_BOUNDS_MODE_P): Remove definition.
1524 (scalare_mode::includes_p): Do not handle MODE_POINTER_BOUNDS.
1525 * mode-classes.def: Do not define MODE_POINTER_BOUNDS.
1526 * stor-layout.c (int_mode_for_mode): Do not handle MODE_POINTER_BOUNDS.
1527 * tree-core.h (enum tree_index): Remove TI_POINTER_BOUNDS_TYPE.
1528 * varasm.c (output_constant_pool_2): Do not handle MODE_POINTER_BOUNDS.
1530 * config/i386/i386-modes.def (BND32, BND64): Remove.
1531 * config/i386/i386.c (dbx_register_map): Remove bound registers.
1532 (dbx64_register_map): Ditto.
1533 (svr4_dbx_register_map): Ditto.
1534 (indirect_thunk_bnd_needed): Remove.
1535 (indirect_thunks_bnd_used): Ditto.
1536 (indirect_return_bnd_needed): Ditto.
1537 (indirect_return_via_cx_bnd): Ditto.
1538 (enum indirect_thunk_prefix): Remove indirect_thunk_prefix_bnd.
1539 (indirect_thunk_name): Remove handling of indirect_thunk_prefix_bnd.
1540 (output_indirect_thunk): Ditto. Remove need_prefix argument.
1541 (output_indirect_thunk_function): Remove handling of
1542 indirect_return_bnd_needed, indirect_return_via_cx_bnd,
1543 indirect_thunk_bnd_needed and indirect_thunks_bnd_used variables.
1544 (ix86_save_reg): Remove handling of crtl->return_bnd.
1545 (ix86_legitimate_constant_p): Remove handling of POINTER_BOUNDS_MODE_P.
1546 (ix86_print_operand_address_as): Remove handling of UNSPEC_BNDMK_ADDR
1547 and UNSPEC_BNDLX_ADDR.
1548 (ix86_output_indirect_branch_via_reg): Remove handling of
1549 indirect_thunk_prefix_bnd.
1550 (ix86_output_indirect_branch_via_push): Ditto.
1551 (ix86_output_function_return): Ditto.
1552 (ix86_output_indirect_function_return): Ditto.
1553 (avoid_func_arg_motion): Do not handle UNSPEC_BNDSTX.
1554 * config/i386/i386.h (FIXED_REGISTERS): Remove bound registers.
1555 (CALL_USED_REGISTERS): Ditto.
1556 (REG_ALLOC_ORDER): Update for removal of bound registers.
1557 (HI_REGISTER_NAMES): Ditto.
1558 * config/i386/i386.md (UNSPEC_BNDMK, UNSPEC_BNDMK_ADDR, UNSPEC_BNDSTX)
1559 (UNSPEC_BNDLDX, UNSPEC_BNDLDX_ADDR, UNSPEC_BNDCL, UNSPEC_BNDCU)
1560 (UNSPEC_BNDCN, UNSPEC_MPX_FENCE): Remove.
1561 (BND0_REG, BND1_REG, BND2_REG, BND3_REG): Remove
1562 (FIRST_PSEUDO_REG): Update.
1563 (BND): Remove mode iterator.
1564 * config/i386/predicates.md (bnd_mem_operator): Remove.
1566 2018-08-24 Richard Sandiford <richard.sandiford@arm.com>
1568 * tree-vect-stmts.c (vectorizable_bswap): Handle variable-length
1571 2018-08-24 Richard Sandiford <richard.sandiford@arm.com>
1573 * tree-vect-slp.c (vect_transform_slp_perm_load): Separate out
1574 the case in which the permute needs only a single element and
1575 repeats for every vector of the result. Extend that case to
1576 handle variable-length vectors.
1577 * tree-vect-stmts.c (vectorizable_load): Update accordingly.
1579 2018-08-24 H.J. Lu <hongjiu.lu@intel.com>
1582 * dwarf2out.c (save_macinfo_strings): Call set_indirect_string
1583 on DW_MACINFO_start_file for -gsplit-dwarf -g3.
1585 2018-08-24 Richard Biener <rguenther@suse.de>
1587 * cfg.h (struct control_flow_graph): Add edge_flags_allocated and
1588 bb_flags_allocated members.
1589 (auto_flag): New RAII class for allocating flags.
1590 (auto_edge_flag): New RAII class for allocating edge flags.
1591 (auto_bb_flag): New RAII class for allocating bb flags.
1592 * cfgloop.c (verify_loop_structure): Allocate temporary edge
1594 * cfganal.c (dfs_enumerate_from): Remove use of visited sbitmap
1595 in favor of temporarily allocated BB flag.
1596 * hsa-brig.c: Re-order includes.
1597 * hsa-dump.c: Likewise.
1598 * hsa-regalloc.c: Likewise.
1599 * print-rtl.c: Likewise.
1600 * profile-count.c: Likewise.
1602 2018-08-24 Segher Boessenkool <segher@kernel.crashing.org>
1605 * config/rs6000/rs6000.c (toc_relative_expr_p): Check that the base is
1608 2018-08-24 Aldy Hernandez <aldyh@redhat.com>
1611 * wide-int-range.cc (wide_int_range_div): Do not ignore result
1612 from wide_int_range_multiplicative_op.
1614 2018-08-23 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
1616 * tree-vect-data-refs.c (vect_grouped_store_supported): Fix typo
1619 2018-08-23 Giuliano Belinassi <giuliano.belinassi@usp.br>
1621 * genmatch.c (parser::parse_operation): Fix typo 'exapnded'
1624 2018-08-23 Alexander Monakov <amonakov@ispras.ru>
1626 * tree-scalar-evolution.c (final_value_replacement_loop): Dump
1627 full GENERIC expression used for replacement.
1629 2018-08-23 Aldy Hernandez <aldyh@redhat.com>
1631 * tree-vrp.c (abs_extent_range): Remove.
1632 (extract_range_into_wide_ints): Pass wide ints by reference.
1633 (extract_range_from_binary_expr_1): Rewrite the *DIV_EXPR code.
1634 Pass wide ints by reference in all calls to
1635 extract_range_into_wide_ints.
1636 * wide-int-range.cc (wide_int_range_div): New.
1637 * wide-int-range.h (wide_int_range_div): New.
1638 (wide_int_range_includes_zero_p): New.
1639 (wide_int_range_zero_p): New.
1641 2018-08-23 Matthew Malcomson <matthew.malcomson@arm.com>
1643 * config/aarch64/aarch64.md (arches): New enum.
1644 (arch): New enum attr.
1645 (arch_enabled): New attr.
1646 (enabled): Now uses arch_enabled only.
1647 (simd, sve, fp16): Removed attribute.
1648 (fp): Attr now defined in terms of 'arch'.
1649 (*mov<mode>_aarch64, *movsi_aarch64, *movdi_aarch64, *movti_aarch64,
1650 *movhf_aarch64, <optab><fcvt_target><GPF:mode>2,
1651 <FCVT_F2FIXED:fcvt_fixed_insn><GPF:mode>3,
1652 <FCVT_FIXED2F:fcvt_fixed_insn><GPI:mode>3): Merge 'fp' and 'simd'
1653 attributes into 'arch'.
1654 (*movsf_aarch64, *movdf_aarch64, *movtf_aarch64, *add<mode>3_aarch64,
1655 subdi3, neg<mode>2, <optab><mode>3, one_cmpl<mode>2,
1656 *<NLOGICAL:optab>_one_cmpl<mode>3, *xor_one_cmpl<mode>3,
1657 *aarch64_ashl_sisd_or_int_<mode>3, *aarch64_lshr_sisd_or_int_<mode>3,
1658 *aarch64_ashr_sisd_or_int_<mode>3, *aarch64_sisd_ushl): Convert use of
1659 'simd' attribute into 'arch'.
1660 (load_pair_sw_<SX:mode><SX2:mode>, load_pair_dw_<DX:mode><DX2:mode>,
1661 store_pair_sw_<SX:mode><SX2:mode>, store_pair_dw_<DX:mode><DX2:mode>):
1662 Convert use of 'fp' attribute to 'arch'.
1663 * config/aarch64/aarch64-simd.md (move_lo_quad_internal_<mode>,
1664 move_lo_quad_internal_<mode>): (different modes) Merge 'fp' and 'simd'
1666 (move_lo_quad_internal_be_<mode>, move_lo_quad_internal_be_<mode>):
1667 (different modes) Merge 'fp' and 'simd' into 'arch'.
1668 (*aarch64_combinez<mode>, *aarch64_combinez_be<mode>): Merge 'fp' and
1671 2018-08-23 Segher Boessenkool <segher@kernel.crashing.org>
1673 PR rtl-optimization/87026
1674 * expmed.c (canonicalize_comparison): If we can no longer create
1675 pseudoregisters, don't.
1677 2018-08-23 Richard Earnshaw <rearnsha@arm.com>
1680 * config/arm/arm-protos.h (arm_emit_speculation_barrier): New
1682 * config/arm/arm.c (speculation_barrier_libfunc): New static
1684 (arm_init_libfuncs): Initialize it.
1685 (arm_emit_speculation_barrier): New function.
1686 * config/arm/arm.md (speculation_barrier): Call
1687 arm_emit_speculation_barrier for architectures that do not have
1689 (speculation_barrier_insn): Only match on Armv7 or later.
1691 2018-08-23 Richard Biener <rguenther@suse.de>
1694 * tree-inline.c (copy_bb): Drop unused __builtin_va_arg_pack_len
1697 2018-08-23 Richard Sandiford <richard.sandiford@arm.com>
1699 * config/aarch64/aarch64.c (aarch64_evpc_sve_tbl): Fix handling
1700 of single-vector TBLs.
1701 (aarch64_vectorize_vec_perm_const): Set one_vector_p when only
1704 2018-08-23 Richard Sandiford <richard.sandiford@arm.com>
1707 * config/aarch64/aarch64.c (aarch64_expand_vec_perm_const_1): Fix
1708 aarch64_evpc_tbl guard.
1710 2018-08-22 Bernd Edlinger <bernd.edlinger@hotmail.de>
1712 * tree-ssa-dse.c (compute_trims): Avoid folding away undefined
1715 2018-08-22 Martin Sebor <msebor@redhat.com>
1718 * tree-pretty-print.c (pretty_print_string): Add argument.
1719 (dump_generic_node): Call to pretty_print_string with string size.
1721 2018-08-22 Segher Boessenkool <segher@kernel.crashing.org>
1723 PR rtl-optimization/86771
1724 * combine.c (try_combine): Do not allow splitting a resulting PARALLEL
1725 of two SETs into those two SETs, one to be placed at i2, if that SETs
1726 destination is modified between i2 and i3.
1728 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
1730 PR tree-optimization/86725
1731 * tree-vect-loop.c (vect_inner_phi_in_double_reduction_p): New
1733 (vect_analyze_scalar_cycles_1): Check it.
1735 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
1737 PR tree-optimization/86725
1738 * tree-vect-loop.c (vect_is_simple_reduction): When treating
1739 an outer loop phi as a double reduction, make sure that the
1740 single user of the phi result is an inner loop phi.
1742 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
1744 * tree-vect-data-refs.c (vect_analyze_group_access_1): Convert
1745 grouped stores with gaps to a strided group.
1747 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
1749 * tree-vect-stmts.c (get_group_load_store_type)
1750 (get_load_store_type): Only test STMT_VINFO_STRIDED_P for the
1751 first statement in a group.
1753 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
1755 * config/darwin.h (LINK_COMMAND_SPEC_A): Sync LTO options with
1756 the sequence used in gcc/gcc.c.
1758 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
1761 * gcc-ar.c (main): Don’t try to invoke the plug-in if we’re not
1764 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
1766 * config/darwin10.h (LINK_GCC_C_SEQUENCE_SPEC): Adjust to use the
1767 Darwin10-specific unwinder-shim.
1768 * config/darwin12.h (LINK_GCC_C_SEQUENCE_SPEC): Remove.
1769 * config/rs6000/darwin.h (DARWIN_CRT1_SPEC, DARWIN_DYLIB1_SPEC):
1770 New to cater for Darwin10 Rosetta.
1772 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
1774 * config/i386/i386.c (ix86_output_addr_diff_elt): Move the MACH-O
1775 specific test before the one for HAVE_AS_GOTOFF_IN_DATA.
1777 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
1782 * gcc/dwarf2out.c (FUNC_SECOND_SECT_LABEL): New.
1783 (dwarf2out_switch_text_section): Generate a local label for the
1784 second function sub-section and apply it as the second FDE start
1786 * gcc/final.c (final_scan_insn_1): Emit second FDE label after the
1787 second sub-section start.
1789 2018-08-22 Richard Biener <rguenther@suse.de>
1791 PR tree-optimization/86988
1792 * tree-vrp.c (vrp_prop::check_mem_ref): Bail out on VLAs.
1794 2018-08-22 Richard Biener <rguenther@suse.de>
1796 PR tree-optimization/86945
1797 * tree-cfg.c (generate_range_test): Use unsigned arithmetic.
1799 2018-08-22 Alexandre Oliva <oliva@adacore.com>
1801 * config/rs6000/rs6000.c (SMALL_DATA_RELOC, SMALL_DATA_REG): Add
1802 a comment about how uses of r2 for .sdata2 come about.
1804 2018-08-22 Alexandre Oliva <aoliva@redhat.com>
1806 * tree-ssa-reassoc.c (is_reassociable_op): Fix cut&pasto.
1808 2018-08-21 Marek Polacek <polacek@redhat.com>
1810 PR c++/86981, Implement -Wpessimizing-move.
1811 * doc/invoke.texi: Document -Wpessimizing-move.
1813 2018-08-21 Jan Hubicka <jh@suse.cz>
1815 * tree.c (find_decls_types_r): Do not check for redundant typedefs.
1816 * tree.h (is_redundant_typedef): Remove.
1817 * dwarf2out.c (is_redundant_typedef): Turn into static function.
1819 2018-08-21 Jan Hubicka <jh@suse.cz>
1821 * tree.c (free_lang_data_in_decl): Remove types from DECL_CONTEXT
1824 2018-08-21 Tamar Christina <tamar.christina@arm.com>
1826 * expmed.c (extract_low_bits): Reject invalid subregs early.
1828 2018-08-21 Bernd Edlinger <bernd.edlinger@hotmail.de>
1831 * tree-ssa-strlen.c (adjust_last_stmt): Avoid folding away undefined
1834 2018-08-21 Rasmus Villemoes <rv@rasmusvillemoes.dk>
1836 * config/vxworks.h: Guard vxworks_asm_out_constructor and
1837 vxworks_asm_out_destructor by !HAVE_INITFINI_ARRAY_SUPPORT
1838 * config/vxworks.c: Likewise.
1840 2018-08-21 Rasmus Villemoes <rv@rasmusvillemoes.dk>
1842 * config/vxworks.c: Set targetm.have_ctors_dtors
1843 if HAVE_INITFINI_ARRAY_SUPPORT.
1844 * config/vxworks.h: Set SUPPORTS_INIT_PRIORITY
1845 if HAVE_INITFINI_ARRAY_SUPPORT.
1847 2018-08-21 Rasmus Villemoes <rv@rasmusvillemoes.dk>
1849 * config/vxworks.h: Add $(WIND_BASE)/target/h/wrn/coreip to
1850 default search path for VxWorks < 7.
1852 2018-08-21 Bernd Edlinger <bernd.edlinger@hotmail.de>
1854 * gimple-ssa-sprintf.c (decl_constant_value): Remove.
1855 (get_format_string): Refer to c_getstr.
1857 2018-08-21 Tom de Vries <tdevries@suse.de>
1859 * cgraph.h (debuginfo_early_init, debuginfo_init, debuginfo_fini)
1860 (debuginfo_start, debuginfo_stop, debuginfo_early_start)
1861 (debuginfo_early_stop): Declare.
1862 * cgraphunit.c (debuginfo_early_init, debuginfo_init, debuginfo_fini)
1863 (debuginfo_start, debuginfo_stop, debuginfo_early_start)
1864 (debuginfo_early_stop): New function.
1865 (symbol_table::finalize_compilation_unit): Call debuginfo_early_start
1866 and debuginfo_early_stop.
1867 * dwarf2out.c (dwarf2out_finish, dwarf2out_early_finish): Dump dwarf.
1868 * toplev.c (compile_file): Call debuginfo_start and debuginfo_stop.
1869 (general_init): Call debuginfo_early_init.
1870 (finalize): Call debuginfo_fini.
1871 (do_compile): Call debuginfo_init.
1872 * doc/invoke.texi (@gccoptlist): Add -fdump-debug and
1874 (@item -fdump-debug, @item -fdump-earlydebug): Add.
1876 2018-08-21 Tom de Vries <tdevries@suse.de>
1878 * dwarf2out.c (print_dw_val, print_loc_descr, print_die): Handle
1879 flag_dump_noaddr and flag_dump_unnumbered.
1881 2018-08-21 Aldy Hernandez <aldyh@redhat.com>
1883 * wide-int-range.cc (wide_int_range_abs): New.
1884 (wide_int_range_order_set): Rename from wide_int_range_min_max.
1885 * wide-int-range.h (wide_int_range_abs): New.
1886 (wide_int_range_min_max): New.
1887 * tree-vrp.c (extract_range_from_unary_expr): Rewrite ABS_EXPR
1888 case to call wide_int_range_abs.
1889 Rewrite MIN/MAX_EXPR to call wide_int_range_min_max.
1890 (extract_range_from_abs_expr): Delete.
1892 2018-08-20 Michael Meissner <meissner@linux.ibm.com>
1895 * config/rs6000/rs6000.md (extendsi<mode>2): Change constraints
1896 from 'Y' to 'YZ' to enable the LWAX instruction to be generated
1899 2018-08-20 Nathan Sidwell <nathan@acm.org>
1900 Jeff Law <law@redhat.com>
1902 * config/s390/s390-c (s390_macro_to_expand): Use cpp_macro_p.
1903 * config/spu/spu-c.c (spu_macro_to_expand): Likewise.
1905 2018-08-20 David Malcolm <dmalcolm@redhat.com>
1908 * attribs.c (diag_attr_exclusions): Add auto_diagnostic_group instance.
1909 (decl_attributes): Likewise.
1910 * calls.c (maybe_warn_nonstring_arg): Add auto_diagnostic_group
1912 * cgraphunit.c (maybe_diag_incompatible_alias): Likewise.
1913 * diagnostic-core.h (class auto_diagnostic_group): New class.
1914 * diagnostic.c (diagnostic_initialize): Initialize the new fields.
1915 (diagnostic_report_diagnostic): Handle the first diagnostics within
1917 (emit_diagnostic): Add auto_diagnostic_group instance.
1919 (inform_n): Likewise.
1920 (warning): Likewise.
1921 (warning_at): Likewise.
1922 (warning_n): Likewise.
1923 (pedwarn): Likewise.
1924 (permerror): Likewise.
1926 (error_n): Likewise.
1927 (error_at): Likewise.
1929 (fatal_error): Likewise.
1930 (internal_error): Likewise.
1931 (internal_error_no_backtrace): Likewise.
1932 (auto_diagnostic_group::auto_diagnostic_group): New ctor.
1933 (auto_diagnostic_group::~auto_diagnostic_group): New dtor.
1934 * diagnostic.h (struct diagnostic_context): Add fields
1935 "diagnostic_group_nesting_depth",
1936 "diagnostic_group_emission_count", "begin_group_cb",
1938 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior):
1939 Add auto_diagnostic_group instance(s).
1940 (find_explicit_erroneous_behavior): Likewise.
1941 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Likewise.
1942 * gimple-ssa-warn-restrict.c (maybe_diag_offset_bounds): Likewise.
1943 * gimplify.c (warn_implicit_fallthrough_r): Likewise.
1944 (gimplify_va_arg_expr): Likewise.
1945 * hsa-gen.c (HSA_SORRY_ATV): Likewise.
1946 (HSA_SORRY_AT): Likewise.
1947 * ipa-devirt.c (compare_virtual_tables): Likewise.
1948 (warn_odr): Likewise.
1949 * multiple_target.c (expand_target_clones): Likewise.
1950 * opts-common.c (cmdline_handle_error): Likewise.
1951 * reginfo.c (globalize_reg): Likewise.
1952 * substring-locations.c (format_warning_n_va): Likewise.
1953 * tree-inline.c (expand_call_inline): Likewise.
1954 * tree-ssa-ccp.c (pass_post_ipa_warn::execute): Likewise.
1955 * tree-ssa-loop-niter.c
1956 (do_warn_aggressive_loop_optimizations): Likewise.
1957 * tree-ssa-uninit.c (warn_uninit): Likewise.
1958 * tree.c (warn_deprecated_use): Likewise.
1960 2018-08-20 H.J. Lu <hongjiu.lu@intel.com>
1963 * config/i386/i386.md (eh_return): Always update EH return
1964 address in word_mode.
1966 2018-08-20 Chung-Lin Tang <cltang@codesourcery.com>
1968 * targhooks.c (std_gimplify_va_arg_expr): Properly handle case of when
1969 TARGET_SPLIT_COMPLEX_ARG is defined.
1971 2018-08-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
1973 * expr.c (store_field): Change gcc_assert to gcc_checking_assert.
1975 2018-08-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
1978 * expr.c (expand_assignment): Assert that bitpos is positive.
1979 (store_field): Likewise
1980 (expand_expr_real_1): Make sure that bitpos is positive.
1981 * config/alpha/alpha.h (CONSTANT_ADDRESS_P): Avoid signed
1984 2018-08-20 Nathan Sidwell <nathan@acm.org>
1986 * Makefile.in (CPP_ID_DATA_H): Delete.
1987 (CPP_INTERNAL_H): Don't add it.
1988 (GTFILES): Replace CPP_ID_DATA_H with CPPLIB_H.
1989 * gengtype.c (open_base_files): Replace cpp-id-data.h with cpplib.h
1991 2018-08-20 Richard Biener <rguenther@suse.de>
1993 PR tree-optimization/78655
1994 * tree-vrp.c (extract_range_from_binary_expr_1): Make
1995 pointer + offset nonnull if either operand is nonnull work.
1997 2018-08-20 Tom de Vries <tdevries@suse.de>
1999 * dwarf2out.c (add_scalar_info): Don't add reference to existing die
2000 unless the referenced die describes the added property using
2001 DW_AT_location or DW_AT_const_value. Fall back to exprloc case.
2002 Otherwise, add a DW_AT_location to the referenced die.
2004 2018-08-19 Uros Bizjak <ubizjak@gmail.com>
2007 * config/i386/i386.c (ix86_rtx_costs) [case SET]: Check source for
2008 register_operand when calling ix86_set_reg_reg_cost.
2009 [case CONST_INT, case CONST, case LABEL_REF, case SYMBOL_REF]:
2010 Set *total to 0 for operands that satisfy x86_64_immediate_operand
2011 predicate and to 1 otherwise.
2013 2018-08-18 Iain Sandoe <iain@sandoe.co.uk>
2015 * config/darwin.c (darwin_override_options): If -gsplit-dwarf is set,
2016 emit a diagnostic that it is not supported and reset the option.
2017 * config/darwin.h (DRIVER_SELF_SPECS): Note that gsplit-dwarf is not
2018 supported and consume the option. (ASM_FINAL_SPEC): New.
2020 2018-08-17 Segher Boessenkool <segher@kernel.crashing.org>
2022 * doc/md.texi (Patterns): Use @ref instead of @xref in the middle of
2025 2018-08-17 Sandra Loosemore <sandra@codesourcery.com>
2027 C-SKY port: Documentation
2029 * doc/extend.texi (C-SKY Function Attributes): New section.
2030 * doc/invoke.texi (Option Summary): Add C-SKY options.
2031 (C-SKY Options): New section.
2032 * doc/md.texi (Machine Constraints): Document C-SKY constraints.
2034 2018-08-17 Jojo <jijie_rong@c-sky.com>
2035 Huibin Wang <huibin_wang@c-sky.com>
2036 Sandra Loosemore <sandra@codesourcery.com>
2037 Chung-Lin Tang <cltang@codesourcery.com>
2039 C-SKY port: Backend implementation
2041 * config/csky/*: New.
2042 * common/config/csky/*: New.
2044 2018-08-17 Jojo <jijie_rong@c-sky.com>
2045 Huibin Wang <huibin_wang@c-sky.com>
2046 Sandra Loosemore <sandra@codesourcery.com>
2047 Chung-Lin Tang <cltang@codesourcery.com>
2048 Andrew Jenner <andrew@codesourcery.com>
2050 C-SKY port: Configury
2052 * config.gcc (csky-*-*): New.
2053 * configure.ac: Add csky to targets for dwarf2 debug_line support.
2054 * configure: Regenerated.
2056 2018-08-17 David Malcolm <dmalcolm@redhat.com>
2058 * dump-context.h: Include "dumpfile.h".
2059 (dump_context::dump_printf_va): Convert final param from va_list
2060 to va_list *. Convert from ATTRIBUTE_PRINTF to
2061 ATTRIBUTE_GCC_DUMP_PRINTF.
2062 (dump_context::dump_printf_loc_va): Likewise.
2063 * dumpfile.c: Include "stringpool.h".
2064 (make_item_for_dump_printf_va): Delete.
2065 (make_item_for_dump_printf): Delete.
2066 (class dump_pretty_printer): New class.
2067 (dump_pretty_printer::dump_pretty_printer): New ctor.
2068 (dump_pretty_printer::emit_items): New member function.
2069 (dump_pretty_printer::emit_any_pending_textual_chunks): New member
2071 (dump_pretty_printer::emit_item): New member function.
2072 (dump_pretty_printer::stash_item): New member function.
2073 (dump_pretty_printer::format_decoder_cb): New member function.
2074 (dump_pretty_printer::decode_format): New member function.
2075 (dump_context::dump_printf_va): Reimplement in terms of
2076 dump_pretty_printer.
2077 (dump_context::dump_printf_loc_va): Convert final param from va_list
2079 (dump_context::begin_scope): Reimplement call to
2080 make_item_for_dump_printf.
2081 (dump_printf): Update for change to dump_printf_va.
2082 (dump_printf_loc): Likewise.
2083 (selftest::test_capture_of_dump_calls): Convert "stmt" from
2084 greturn * to gimple *. Add a test_decl. Add tests of dump_printf
2085 with %T, %E, and %G.
2086 * dumpfile.h (ATTRIBUTE_GCC_DUMP_PRINTF): New macro.
2087 (dump_printf): Replace ATTRIBUTE_PRINTF_2 with
2088 ATTRIBUTE_GCC_DUMP_PRINTF (2, 3).
2089 (dump_printf_loc): Replace ATTRIBUTE_PRINTF_3 with
2090 ATTRIBUTE_GCC_DUMP_PRINTF (3, 0).
2091 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Convert
2092 use of HOST_WIDE_INT_PRINT_DEC on unsigned HOST_WIDE_INT "count"
2093 within a dump_printf_loc call to "%wu".
2094 (vector_alignment_reachable_p): Merge two dump_printf[_loc] calls,
2095 converting a use of HOST_WIDE_INT_PRINT_DEC to "%wd". Add a
2096 missing space after "=".
2097 * tree-vect-loop.c (vect_analyze_loop_2) Within a dump_printf
2098 call, convert use of HOST_WIDE_INT_PRINT_DEC to "%wd".
2099 * tree-vect-slp.c (vect_slp_bb): Within a dump_printf_loc call,
2100 convert use of HOST_WIDE_INT_PRINT_UNSIGNED to "%wu".
2101 * tree-vectorizer.c (try_vectorize_loop_1): Likewise. Remove
2102 duplicate "vectorized" from message.
2104 2018-08-17 Szabolcs Nagy <szabolcs.nagy@arm.com>
2106 * config/arm/arm-builtins.c (arm_init_simd_builtin_types): Clear
2107 polyNxK_t element's TYPE_STRING_FLAG.
2109 2018-08-17 Segher Boessenkool <segher@kernel.crashing.org>
2111 * config/rs6000/rs6000.md (*cbranch, *creturn): Name these patterns
2112 (they were unnamed before). Fix comments.
2114 2018-08-17 Nathan Sidwell <nathan@acm.org>
2116 * cppbuiltin.c: Include "cpplib.h", not "cpp-id-data.h".
2118 2018-08-17 Richard Biener <rguenther@suse.de>
2120 PR tree-optimization/86841
2121 * wide-int-range.cc (wide_int_range_lshift): Use to_uhwi.
2123 2018-08-17 Martin Liska <mliska@suse.cz>
2125 * common.opt: Remove Warn, Init and Report for options with
2126 Ignore/Deprecated flag. Warning is done automatically for
2128 * config/i386/i386.opt: Likewise.
2129 * config/ia64/ia64.opt: Likewise.
2130 * config/rs6000/rs6000.opt: Likewise.
2131 * cppbuiltin.c (define_builtin_macros_for_compilation_flags):
2132 Remove usage of flag_check_pointer_bounds.
2133 * lto-wrapper.c (merge_and_complain): Do not handle
2134 OPT_fcheck_pointer_bounds.
2135 (append_compiler_options): Likewise.
2136 * opt-functions.awk: Do not handle Deprecated.
2137 * optc-gen.awk: Check that Var, Report and Init are not
2138 used for an option with Ignore/Deprecated flag.
2139 * opts-common.c (decode_cmdline_option): Do not report
2141 (read_cmdline_option): Report warning for OPT_SPECIAL_deprecated
2143 * opts.h (struct cl_option): Remove cl_deprecated flag.
2144 (CL_ERR_DEPRECATED): Remove error enum value.
2146 2018-08-17 Richard Biener <rguenther@suse.de>
2149 * tree-inline.c (copy_bb): When inlining __builtin_va_arg_pack_len ()
2150 across a va-arg-pack using call adjust its return value accordingly.
2152 2018-08-16 Martin Sebor <msebor@redhat.com>
2154 PR tree-optimization/86853
2155 * gimple-ssa-sprintf.c (struct format_result): Rename member.
2156 (struct fmtresult): Add member and initialize it in ctors.
2157 (format_character): Handle %C. Extend range to NUL. Set MAYFAIL.
2158 (format_string): Handle %S the same as %ls. Set MAYFAIL.
2159 (format_directive): Set POSUNDER4K when MAYFAIL is set.
2160 (parse_directive): Handle %C same as %c.
2161 (sprintf_dom_walker::compute_format_length): Adjust.
2162 (is_call_safe): Adjust.
2164 2018-08-16 Bernd Edlinger <bernd.edlinger@hotmail.de>
2166 * builtins.c (c_strlen): Add new parameter eltsize. Use it
2167 for determining how to count the elements.
2168 * builtins.h (c_strlen): Adjust prototype.
2169 * expr.c (string_constant): Add new parameter mem_size.
2170 Set *mem_size appropriately.
2171 * expr.h (string_constant): Adjust protoype.
2172 * gimple-fold.c (get_range_strlen): Add new parameter eltsize.
2173 * gimple-fold.h (get_range_strlen): Adjust prototype.
2174 * gimple-ssa-sprintf.c (get_string_length): Add new parameter eltsize.
2175 (format_string): Call get_string_length with eltsize.
2177 2018-08-16 David Malcolm <dmalcolm@redhat.com>
2179 * diagnostic.c (default_diagnostic_start_span_fn): Call pp_string
2180 to emit the span, rather than setting it as the prefix.
2182 2018-08-16 David Malcolm <dmalcolm@redhat.com>
2184 * diagnostic-show-locus.c (layout::start_annotation_line): Add
2185 "margin_char" parameter, defaulting to space. Use it in place
2186 of pp_space for the initial part of the margin.
2187 (layout::print_leading_fixits): Use '+' when filling the margin
2188 of line-insertion fix-it hints.
2190 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
2192 * config/rs6000/rs6000.md (two unnamed define_insn and define_split):
2195 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
2197 * config/rs6000/altivec.md: Don't set length attribute to the default
2199 * config/rs6000/darwin.md: Ditto.
2200 * config/rs6000/dfp.md: Ditto.
2201 * config/rs6000/htm.md: Ditto.
2202 * config/rs6000/rs6000.md: Ditto.
2203 * config/rs6000/sync.md: Ditto.
2204 * config/rs6000/vsx.md: Ditto.
2206 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
2208 * config/rs6000/altivec.md: Don't set length attribute to the default
2209 value, for branch instructions.
2210 * config/rs6000/darwin.md: Ditto.
2211 * config/rs6000/rs6000.md: Ditto.
2213 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
2215 * config/rs6000/rs6000.md (length): Always define as const_int 4.
2216 (unnamed conditional branch define_insn): Set length to 4 or 8
2217 depending on offset.
2218 (<bd>_<mode>): Similar, for alternative 0.
2219 (<bd>tf_<mode>): Ditto.
2221 2018-08-16 Tamar Christina <tamar.christina@arm.com>
2223 * expr.c (copy_blkmode_to_reg): Perform larger copies when safe.
2225 2018-08-16 Matthew Malcomson <matthew.malcomson@arm.com>
2227 * doc/rtl.texi: Replace old RTX class names with new names.
2230 2018-08-16 Vlad Lazar <vlad.lazar@arm.com>
2232 * expmed.h (canonicalize_comparison): New declaration.
2233 * expmed.c (canonicalize_comparison, equivalent_cmp_code): New function.
2234 * expmed.c (emit_store_flag_1): Add call to canonicalize_comparison.
2235 * optabs.c (prepare_cmp_insn): Likewise.
2236 * rtl.h (unsigned_condition_p): New function which checks if a
2237 comparison operator is unsigned.
2239 2018-08-16 Nathan Sidwell <nathan@acm.org>
2241 * config/rs6000/rs6000-c.c (rs6000_macro_to_expend): Use cpp_macro_p.
2242 * config/powerpcspc/powerpcspe-c.c (rs6000_macro_to_expend): Likewise.
2244 2018-08-16 Tamar Christina <tamar.christina@arm.com>
2247 * config/arm/arm.c (arm_can_change_mode_class): Disallow subreg.
2248 * config/arm/neon.md (movv4hf, movv8hf): Refactored to..
2249 (mov<mov>): ..this and enable unconditionally.
2251 2018-08-16 Tamar Christina <tamar.christina@arm.com>
2253 * config/arm/neon.md (*neon_mov<mode>): Remove reg-to-reg alternative.
2255 2018-08-16 Sam Tebbs <sam.tebbs@arm.com>
2257 * config/aarch64/aarch64.opt (mlow-precision-recip-sqrt)
2258 (mlow-precision-sqrt, mlow-precision-div, mverbose-cost-dump): Replace
2259 "Common" with "Target".
2261 2018-08-15 Uros Bizjak <ubizjak@gmail.com>
2263 * config/i386/i386.opt (mmitigate-rop): Mark as deprecated.
2264 * doc/invoke.texi (mmitigate-rop): Remove.
2265 * config/i386/i386.c: Do not include "regrename.h".
2266 (ix86_rop_should_change_byte_p, reg_encoded_number)
2267 (ix86_get_modrm_for_rop, set_rop_modrm_reg_bits, ix86_mitigate_rop):
2269 (ix86_reorg): Remove call to ix86_mitigate_rop.
2270 * config/i386/i386.md (attr "modrm_class"): Remove.
2271 (cmp<mode>_ccno_1, mov<mode>_xor, movstrict<mode>_xor)
2272 (x86_mov<mode>cc_0_m1. x86_mov<mode>cc_0_m1_se)
2273 (x86_mov<mode>cc_0_m1_neg): Remove modrm_class attribute override.
2275 2018-08-15 Will Schmidt <will_schmidt@vnet.ibm.com>
2277 * config/rs6000/rs600.c (rs6000_gimple_fold_builtin): Add entries to
2278 allow folding of mergeh() and mergel() for the float and double types.
2279 (fold_mergehl_helper): Rework to handle building a permute tree
2282 2018-08-15 Uros Bizjak <ubizjak@gmail.com>
2284 * config/i386/i386.c (expand_vec_perm_movs): Enable V4SFmode
2287 2018-08-15 David Malcolm <dmalcolm@redhat.com>
2289 * common.opt (fdiagnostics-show-labels): New option.
2290 * diagnostic-show-locus.c (class layout_range): Add field
2292 (class layout): Add field "m_show_labels_p".
2293 (layout_range::layout_range): Add param "label" and use it to
2295 (make_range): Pass in NULL for new "label" param of layout_range's
2297 (layout::layout): Initialize m_show_labels_p.
2298 (layout::maybe_add_location_range): Pass in loc_range->m_label
2299 when constructing layout_range instances.
2300 (struct line_label): New struct.
2301 (layout::print_any_labels): New member function.
2302 (layout::print_line): Call it if label-printing is enabled.
2303 (selftest::test_one_liner_labels): New test.
2304 (selftest::test_diagnostic_show_locus_one_liner): Call it.
2305 * diagnostic.c (diagnostic_initialize): Initialize
2306 context->show_labels_p.
2307 * diagnostic.h (struct diagnostic_context): Add field
2309 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
2310 -fno-diagnostics-show-labels.
2311 * dwarf2out.c (gen_producer_string): Add
2312 OPT_fdiagnostics_show_labels to the ignored options.
2313 * gcc-rich-location.c (gcc_rich_location::add_expr): Add "label"
2315 (gcc_rich_location::maybe_add_expr): Likewise.
2316 * gcc-rich-location.h (gcc_rich_location::gcc_rich_location): Add
2317 label" param, defaulting to NULL.
2318 (gcc_rich_location::add_expr): Add "label" param.
2319 (gcc_rich_location::maybe_add_expr): Likewise.
2320 (class text_range_label): New class.
2321 (class range_label_for_type_mismatch): New class.
2322 * gimple-ssa-sprintf.c (fmtwarn): Pass NULL for new label params
2323 of format_warning_va.
2324 (fmtwarn_n): Likewise for new params of format_warning_n_va.
2325 * lto-wrapper.c (merge_and_complain): Add
2326 OPT_fdiagnostics_show_labels to the "pick one setting" options.
2327 (append_compiler_options): Likewise to the dropped options.
2328 (append_diag_options): Likewise to the passed-on options.
2329 * opts.c (common_handle_option): Handle the new option.
2330 * selftest-diagnostic.c
2331 (test_diagnostic_context::test_diagnostic_context): Enable
2333 * substring-locations.c: Include "gcc-rich-location.h".
2334 (format_warning_n_va): Add "fmt_label" and "param_label" params
2335 and use them as appropriate.
2336 (format_warning_va): Add "fmt_label" and "param_label" params,
2337 passing them on to format_warning_n_va.
2338 (format_warning_at_substring): Likewise.
2339 (format_warning_at_substring_n): Likewise.
2340 * substring-locations.h (format_warning_va): Add "fmt_label" and
2341 "param_label" params.
2342 (format_warning_n_va): Likewise.
2343 (format_warning_at_substring): Likewise.
2344 (format_warning_at_substring_n): Likewise.
2345 * toplev.c (general_init): Initialize global_dc->show_labels_p.
2347 2018-08-15 Qing Zhao <qing.zhao@oracle.com>
2350 * builtins.c (expand_builtin_memcmp): Do not expand the call
2351 when overflow is detected.
2353 2018-08-15 Martin Sebor <msebor@redhat.com>
2355 PR tree-optimization/71625
2356 * config/aarch64/aarch64-builtins.c
2357 (aarch64_init_simd_builtin_types): Clear Poly8_t's TYPE_STRING_FLAG.
2359 2018-08-15 Ilya Leoshkevich <iii@linux.ibm.com>
2361 * config/s390/s390.c (s390_reorg): Remove loop.
2363 2018-08-15 Iain Sandoe <iain@sandoe.co.uk>
2366 (darwin_function_switched_text_sections): Delete.
2367 * gcc/config/darwin.h
2368 (TARGET_ASM_FUNCTION_SWITCHED_TEXT_SECTIONS): Likewise.
2370 2018-08-15 Iain Sandoe <iain@sandoe.co.uk>
2373 * config/darwin.h: (DEBUG_STR_OFFSETS_SECTION, DEBUG_LOCLISTS_SECTION,
2374 DEBUG_RNGLISTS_SECTION) new macros. (DEBUG_PUBNAMES_SECTION,
2375 DEBUG_PUBTYPES_SECTION) update to include GNU variant.
2377 2018-08-15 Martin Liska <mliska@suse.cz>
2379 PR tree-optimization/86925
2380 * predict.c (expr_expected_value_1): When taking
2381 later predictor, assign also probability.
2382 Use fold_build2_initializer_loc in order to fold
2383 the expression in -frounding-math.
2385 2018-08-14 Allan Sandfeld Jensen <allan.jensen@qt.io>
2387 * config/i386/i386.c (expand_vec_perm_movs): New method matching movs
2389 (expand_vec_perm_1): Try the new method.
2391 2018-08-14 Ilya Leoshkevich <iii@linux.ibm.com>
2394 * lra-lives.c (remove_some_program_points_and_update_live_ranges):
2395 Check whether lra_live_max_point is 0 before dividing.
2397 2018-08-14 Martin Sebor <msebor@redhat.com>
2399 PR tree-optimization/86650
2400 * tree-vrp.c (vrp_prop::check_array_ref): Print an inform message.
2401 (vrp_prop::check_mem_ref): Same.
2403 2018-08-13 Liu Hao <lh_mouse@126.com>
2405 * pretty-print.c (eat_esc_sequence): Swap the foreground and
2406 background colors if the COMMON_LVB_REVERSE_VIDEO flag is set,
2407 and clear it thereafter, as it only works for DBCS.
2409 2018-08-13 Liu Hao <lh_mouse@126.com>
2411 * pretty-print.c (mingw_ansi_fputs): Do not call _close() on the
2412 handle returned by _get_osf_handle().
2414 2018-08-13 Will Schmidt <will_schmidt@vnet.ibm.com>
2416 * gcc/config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support
2417 for folding vec_perm.
2419 2018-08-13 Will Schmidt <will_schmidt@vnet.ibm.com>
2421 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin):
2422 Add support for gimple-folding of vec_pack() and vec_unpack()
2425 2018-08-13 Will Schmidt <will_schmidt@vnet.ibm.com>
2427 * config/rs6000/rs6000.c (rs6000_builtin_valid_without_lhs): Add
2428 vec_xst variants to the list.
2429 (rs6000_gimple_fold_builtin): Add support for folding unaligned
2430 vector loads and stores.
2432 2018-08-13 David Edelsohn <dje.gcc@gmail.com>
2434 * config.gcc (rs6000-ibm-aix4.x): Delete.
2435 (rs6000-ibm-aix5.1): Delete.
2436 (rs6000-ibm-aix5.2): Delete.
2437 (rs6000-ibm-aix5.3): Delete.
2438 * config/rs6000/aix43.h: Delete.
2439 * config/rs6000/aix51.h: Delete.
2440 * config/rs6000/aix52.h: Delete.
2441 * config/rs6000/t-aix43: Delete.
2443 2018-08-13 Ilya Leoshkevich <iii@linux.ibm.com>
2445 * config/s390/s390.c (s390_decompose_constant_pool_ref):
2447 (s390_decompose_address): Factor out constant pool ref
2450 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
2452 * config/nds32/nds32-predicates.c
2453 (nds32_can_use_bclr_p): Change return type as bool.
2454 (nds32_can_use_bset_p): Ditto.
2455 (nds32_can_use_btgl_p): Ditto.
2456 (nds32_can_use_bitci_p): Ditto.
2457 * config/nds32/nds32-protos.h
2458 (nds32_can_use_bclr_p): Change declaration.
2459 (nds32_can_use_bset_p): Ditto.
2460 (nds32_can_use_btgl_p): Ditto.
2461 (nds32_can_use_bitci_p): Ditto.
2463 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
2465 * config/nds32/nds32.c (nds32_expand_prologue, nds32_expand_epilogue):
2466 Support -msched-prolog-epilog option.
2467 * config/nds32/nds32.opt (msched-prolog-epilog): New option.
2469 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
2471 * common/config/nds32/nds32-common.c
2472 (nds32_option_optimization_table): Enalbe -malways-align.
2474 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
2476 * config.gcc (nds32*): Add nds32_isr.h and nds32_init.inc in
2478 * common/config/nds32/nds32-common.c (nds32_handle_option): Handle
2479 OPT_misr_secure_ case.
2480 * config/nds32/nds32-isr.c: Implementation of backward compatibility.
2481 * config/nds32/nds32-protos.h (nds32_isr_function_critical_p): New.
2482 * config/nds32/nds32.c (nds32_attribute_table): Add critical and
2484 * config/nds32/nds32.h (nds32_isr_nested_type): Add NDS32_CRITICAL.
2485 (nds32_isr_info): New field security_level.
2486 (TARGET_ISR_VECTOR_SIZE_4_BYTE): New macro.
2487 * config/nds32/nds32.md (return_internal): Consider critical attribute.
2488 * config/nds32/nds32.opt (misr-secure): New option.
2489 * config/nds32/nds32_init.inc: New file.
2490 * config/nds32/nds32_isr.h: New file.
2492 2018-08-11 John David Anglin <danglin@gcc.gnu.org>
2494 * config/pa/pa.md (UNSPEC_MEMORY_BARRIER): New unspec enum.
2495 Update comment for atomic instructions.
2496 (atomic_storeqi, atomic_storehi, atomic_storesi, atomic_storesf,
2497 atomic_loaddf, atomic_loaddf_1, atomic_storedf, atomic_storedf_1):
2499 (atomic_loaddi): Revise fence expansion to only emit fence prior to
2500 load for __ATOMIC_SEQ_CST model.
2501 (atomic_loaddi_1): Remove float register target.
2502 (atomic_storedi): Handle CONST_INT values.
2503 (atomic_storedi_1): Remove float register source. Add special case
2505 (memory_barrier): New expander and insn.
2507 2018-08-11 Jakub Jelinek <jakub@redhat.com>
2509 PR tree-optimization/86835
2510 * tree-ssa-math-opts.c (insert_reciprocals): Even when inserting
2511 new_stmt after def_gsi, make sure to insert new_square_stmt after
2512 that stmt, not 2 stmts before it.
2514 2018-08-10 Alexander Monakov <amonakov@ispras.ru>
2517 * config/i386/i386.md (<s>mul<mode>3_highpart): Use DWIH mode iterator
2520 2018-08-10 Martin Liska <mliska@suse.cz>
2523 * builtin-types.def (BT_FN_LONG_LONG_LONG_DOUBLE): Add new
2525 * builtins.c (expand_builtin_expect_with_probability):
2527 (expand_builtin_expect_with_probability): New function.
2528 (build_builtin_expect_predicate): Add new argumnet probability
2529 for BUILT_IN_EXPECT_WITH_PROBABILITY.
2530 (fold_builtin_expect):
2533 * builtins.def (BUILT_IN_EXPECT_WITH_PROBABILITY):
2534 * builtins.h (fold_builtin_expect): Set new argument.
2535 * doc/extend.texi: Document __builtin_expect_with_probability.
2536 * doc/invoke.texi: Likewise.
2537 * gimple-fold.c (gimple_fold_call): Pass new argument.
2538 * ipa-fnsummary.c (find_foldable_builtin_expect): Handle
2539 also BUILT_IN_EXPECT_WITH_PROBABILITY.
2540 * predict.c (get_predictor_value): New function.
2541 (expr_expected_value): Add new argument probability. Assume
2542 that predictor and probability are always non-null.
2543 (expr_expected_value_1): Likewise. For __builtin_expect and
2544 __builtin_expect_with_probability set probability. Handle
2545 combination in binary expressions.
2546 (tree_predict_by_opcode): Simplify code by simply calling
2547 get_predictor_value.
2548 (pass_strip_predict_hints::execute): Add handling of
2549 BUILT_IN_EXPECT_WITH_PROBABILITY.
2550 * predict.def (PRED_BUILTIN_EXPECT_WITH_PROBABILITY): Add
2552 * tree.h (DECL_BUILT_IN_P): New function.
2554 2018-08-10 Martin Liska <mliska@suse.cz>
2556 PR tree-optimization/85799
2557 * passes.def: Add argument for pass_strip_predict_hints.
2558 * predict.c (class pass_strip_predict_hints): Add new argument
2560 (strip_predictor_early): New function.
2561 (pass_strip_predict_hints::execute): Call the function to
2563 (strip_predict_hints): New function.
2564 * predict.def: Fix comment.
2566 2018-08-10 Thomas Preud'homme <thomas.preudhomme@linaro.org>
2568 * Makefile.in: Clarify which tm.texi to copy over to assert the
2569 right to grant a GFDL license for all.
2571 2018-08-09 Jeff Law <law@redhat.com>
2573 * config/m68k/m68k.c (m68k_adjust_decorated_operand): Remove
2576 2018-08-09 Andreas Schwab <schwab@linux-m68k.org>
2578 * config/m68k/m68k-protos.h (m68k_final_prescan_insn): Remove
2581 2018-08-09 Richard Sandiford <richard.sandiford@arm.com>
2583 * tree-vect-loop.c (vectorizable_reduction): Allow inner-loop
2584 reductions for variable-length vectors.
2586 2018-08-09 David Malcolm <dmalcolm@redhat.com>
2589 * common.opt (fdiagnostics-show-line-numbers): New option.
2590 * diagnostic-show-locus.c (class layout): Add fields
2591 "m_show_line_numbers_p" and "m_linenum_width";
2592 (num_digits): New function.
2593 (test_num_digits): New function.
2594 (layout::layout): Initialize new fields. Update m_x_offset
2595 logic to handle any left margin.
2596 (layout::print_source_line): Print line number when requested.
2597 (layout::start_annotation_line): New member function.
2598 (layout::print_annotation_line): Call it.
2599 (layout::print_leading_fixits): Likewise.
2600 (layout::print_trailing_fixits): Likewise. Update calls to
2601 move_to_column for new parameter.
2602 (layout::get_x_bound_for_row): Add "add_left_margin" param and use
2603 it to potentially call start_annotation_line.
2604 (layout::show_ruler): Call start_annotation_line.
2605 (selftest::test_line_numbers_multiline_range): New selftest.
2606 (selftest::diagnostic_show_locus_c_tests): Call test_num_digits
2607 and selftest::test_line_numbers_multiline_range.
2608 * diagnostic.c (diagnostic_initialize): Initialize
2609 show_line_numbers_p.
2610 * diagnostic.h (struct diagnostic_context): Add field
2611 "show_line_numbers_p".
2612 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
2613 -fno-diagnostics-show-line-numbers.
2614 * dwarf2out.c (gen_producer_string): Add
2615 OPT_fdiagnostics_show_line_numbers to the ignored options.
2616 * lto-wrapper.c (merge_and_complain): Likewise to the "pick
2617 one setting" options.
2618 (append_compiler_options): Likewise to the dropped options.
2619 (append_diag_options): Likewise to the passed-on options.
2620 * opts.c (common_handle_option): Handle the new option.
2621 * toplev.c (general_init): Set up global_dc->show_line_numbers_p.
2623 2018-08-09 Kelvin Nilsen <kelvin@gcc.gnu.org>
2625 * doc/extend.texi (PowerPC AltiVec Built-in Functions Available on
2626 ISA 2.07): Correct spelling of bcdsub to be __builtin_bcdsub. Add
2627 third argument of type "const signed char" to descriptions of
2628 __builtin_bcdadd, __builtin_bcdadd_lt, __builtin_bcdadd_eq,
2629 __builtin_bcdadd_gt, __builtin_bcdadd_ov, __builtin_bcdsub,
2630 __builtin_bcdsub_lt, __builtin_bcdsub_eq, __builtin_bcdsub_gt,
2631 __builtin_bcdsub_ov functions.
2633 2018-08-09 Richard Sandiford <richard.sandiford@arm.com>
2635 PR tree-optimization/86858
2636 * tree-vect-loop.c (vect_is_simple_reduction): Restore
2637 flow_bb_inside_loop_p calls.
2639 2018-08-09 Richard Sandiford <richard.sandiford@arm.com>
2641 PR tree-optimization/86871
2642 * tree-vect-stmts.c (vect_transform_stmt): Use gimple_get_lhs
2643 instead of gimple_assign_lhs.
2645 2018-08-09 Richard Earnshaw <rearnsha@arm.com>
2648 * config/aarch64/aarch64.md (add<mode>3_carryinC_zero): Add missing
2649 register constraint to operand 0.
2650 (add<mode>3_carryinC): Likewise.
2651 (add<mode>3_carryinV_zero, add<mode>3_carryinV): Likewise.
2653 2018-08-09 Martin Liska <mliska@suse.cz>
2656 * common.opt: Remove extra line.
2658 2018-08-09 Martin Liska <mliska@suse.cz>
2660 * params.def (PARAM_ALIGN_LOOP_ITERATIONS): Remove double dots
2661 at the end of a line, make first letter capital and end up
2662 a sentence with a dot.
2663 (PARAM_LOOP_INTERCHANGE_STRIDE_RATIO): Likewise.
2664 (PARAM_LOOP_BLOCK_TILE_SIZE): Likewise.
2665 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Likewise.
2666 (PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP): Likewise.
2667 (PARAM_MAX_ISL_OPERATIONS): Likewise.
2668 (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): Likewise.
2669 (PARAM_PROFILE_FUNC_INTERNAL_ID): Likewise.
2670 (PARAM_INDIR_CALL_TOPN_PROFILE): Likewise.
2671 (PARAM_SLP_MAX_INSNS_IN_BB): Likewise.
2672 (PARAM_IPA_CP_EVAL_THRESHOLD): Likewise.
2673 (PARAM_IPA_CP_RECURSION_PENALTY): Likewise.
2674 (PARAM_IPA_CP_SINGLE_CALL_PENALTY): Likewise.
2675 (PARAM_IPA_CP_LOOP_HINT_BONUS): Likewise.
2676 (PARAM_IPA_CP_ARRAY_INDEX_HINT_BONUS): Likewise.
2677 (PARAM_TREE_REASSOC_WIDTH): Likewise.
2678 (PARAM_HSA_GEN_DEBUG_STORES): Likewise.
2679 (PARAM_MAX_SPECULATIVE_DEVIRT_MAYDEFS): Likewise.
2680 (PARAM_MAX_VRP_SWITCH_ASSERTIONS): Likewise.
2682 2018-08-09 Andreas Krebbel <krebbel@linux.ibm.com>
2685 * config/s390/s390.c (s390_option_override_internal): Reduce the
2686 stack-clash-protection-probe-interval param if it would be too big
2689 2018-08-08 Andreas Schwab <schwab@linux-m68k.org>
2692 * config/m68k/m68k.h (FINAL_PRESCAN_INSN): Don't define.
2693 * config/m68k/m68k.c (handle_move_double): Don't call
2694 m68k_final_prescan_insn.
2695 (m68k_adjust_decorated_operand): Renamed from
2696 m68k_final_prescan_insn, remove first and third operand and
2698 (print_operand): Call it.
2699 (print_operand_address): Call it.
2701 2018-08-08 Nathan Sidwell <nathan@acm.org>
2703 * diagnostic.c (diagnostic_report_current_module): Use
2704 linemap_included_from & linemap_included_from_linemap.
2706 2018-08-08 Hongbo Zhang <hongbo.zhang@linaro.org>
2708 * config/aarch64/aarch64-cores.def: Add phecda core.
2709 * config/aarch64/aarch64-tune.md: Regenerate.
2710 * doc/invoke.texi: Add phecda core.
2712 2018-08-08 Andreas Krebbel <krebbel@linux.ibm.com>
2715 * config/s390/constraints.md ("NxHD0", "NxSD0"): New constraint
2717 * config/s390/s390.md ("movti"): Add more alternatives for
2718 constant to GPR copies.
2720 2018-08-08 Andreas Krebbel <krebbel@linux.ibm.com>
2722 * config/s390/s390.c: Fix whitespace damage throughout the file.
2723 * config/s390/s390.h: Likewise.
2724 * config/s390/tpf.h: Likewise.
2726 2018-08-08 Ilya Leoshkevich <iii@linux.ibm.com>
2728 * config/s390/s390.c (s390_loadrelative_operand_p):
2729 Remove TARGET_CPU_ZARCH usages.
2730 (s390_rtx_costs): Likewise.
2731 (s390_legitimate_constant_p): Likewise.
2732 (s390_cannot_force_const_mem): Likewise.
2733 (legitimate_reload_constant_p): Likewise.
2734 (s390_preferred_reload_class): Likewise.
2735 (legitimize_pic_address): Likewise.
2736 (legitimize_tls_address): Likewise.
2737 (s390_split_branches): Removed.
2738 (s390_add_execute): Removed.
2739 (s390_dump_pool): Remove TARGET_CPU_ZARCH usages.
2740 (s390_mainpool_start): Likewise.
2741 (s390_mainpool_finish): Likewise.
2742 (s390_mainpool_cancel): Removed.
2743 (s390_chunkify_start): Remove TARGET_CPU_ZARCH usages.
2744 (s390_chunkify_cancel): Likewise.
2745 (s390_return_addr_rtx): Likewise.
2746 (s390_register_info): Remove split_branches_pending_p uages.
2747 (s390_optimize_register_info): Likewise.
2748 (s390_init_frame_layout): Remove TARGET_CPU_ZARCH and
2749 split_branches_pending_p usages.
2750 (s390_can_eliminate): Remove TARGET_CPU_ZARCH usages.
2751 (s390_load_got): Likewise.
2752 (s390_expand_split_stack_prologue): Likewise.
2753 (output_asm_nops): Likewise.
2754 (s390_function_profiler): Likewise.
2755 (s390_emit_call): Likewise.
2756 (s390_conditional_register_usage): Likewise.
2757 (s390_optimize_prologue): Likewise.
2758 (s390_reorg): Remove TARGET_CPU_ZARCH and
2759 split_branches_pending_p usages.
2760 (s390_option_override_internal): Remove TARGET_CPU_ZARCH
2762 (s390_output_indirect_thunk_function): Likewise.
2763 * config/s390/s390.h (TARGET_CPU_ZARCH): Removed.
2764 (TARGET_CPU_ZARCH_P): Removed.
2765 (struct machine_function): Remove split_branches_pending_p.
2766 * config/s390/s390.md: Remove TARGET_CPU_ZARCH usages.
2768 2018-08-08 Ilya Leoshkevich <iii@linux.ibm.com>
2770 * common/config/s390/s390-common.c (processor_flags_table):
2772 * config.gcc: Remove with_arch/with_tune support.
2773 * config/s390/2064.md: Remove cpu attribute comparisons.
2774 * config/s390/driver-native.c (s390_host_detect_local_cpu):
2776 * config/s390/linux.h (ASM_SPEC):
2777 Remove -march support.
2778 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal):
2779 Use a table to get an arch level.
2780 * config/s390/s390-opts.h (enum processor_type):
2782 * config/s390/s390.c
2783 (processor_table): Remove entries, add arch_level values.
2784 (s390_issue_rate): Remove cases.
2785 (s390_option_override): Adjust
2786 s390_option_override_internal() call.
2787 (s390_option_override_internal): Remove deprecation warning.
2788 (s390_valid_target_attribute_tree): Adjust
2789 s390_option_override_internal() call.
2790 * config/s390/s390.h (struct s390_processor):
2791 Share with s390-c.c, add arch_level field.
2792 * config/s390/s390.md:
2793 Remove occurrences in cpu attribute.
2794 * config/s390/s390.opt: Remove -march/-mtune support.
2795 * config/s390/tpf.h (ASM_SPEC): Remove -march support.
2796 * doc/invoke.texi: Remove deprecation warning.
2798 2018-08-08 Luis Machado <luis.machado@linaro.org>
2800 * config/aarch64/aarch64.c (qdf24xx_vector_cost): New static
2802 (qdf24xx_tunings): Set vector cost structure to
2803 qdf24xx_vector_cost.
2805 * config/aarch64/aarch64.c (qdf24xx_addrcost_table)
2806 <register_sextend>: Set to 3.
2808 2018-08-07 Richard Sandiford <richard.sandiford@arm.com>
2811 * config/aarch64/iterators.md (FRECP, frecp_suffix): Delete.
2812 * config/aarch64/aarch64-simd.md
2813 (aarch64_frecp<FRECP:frecp_suffix><mode>): Fold FRECPE into...
2814 (@aarch64_frecpe<mode>): ...here and the move FRECPX to...
2815 (aarch64_frecpx<mode>): ...this new pattern.
2816 * config/aarch64/aarch64-simd-builtins.def: Remove comment
2817 about aarch64_frecp<FRECP:frecp_suffix><mode>.
2819 2018-08-07 Martin Liska <mliska@suse.cz>
2822 * predict.c (expr_expected_value_1): Handle DECL_IS_MALLOC,
2823 BUILT_IN_REALLOC and DECL_IS_OPERATOR_NEW.
2824 * predict.def (PRED_MALLOC_NONNULL): New predictor.
2825 * doc/extend.texi: Document that malloc attribute adds
2828 2018-08-06 John David Anglin <danglin@gcc.gnu.org>
2831 * config/pa/pa.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
2832 Define to speculation_safe_value_not_needed.
2834 2018-08-06 Jeff Law <law@redhat.com>
2836 * tree-ssa-dom.c (dom_opt_dom_walker::optimize_stmt): Pass down
2837 the vr_values instance to cprop_into_stmt.
2838 (cprop_into_stmt): Pass vr_values instance down to cprop_operand.
2839 (cprop_operand): Also query EVRP to determine if OP is a constant.
2841 2018-08-06 Nathan Sidwell <nathan@acm.org>
2843 * diagnostic.c (diagnostic_report_current_module): Reroll
2844 included-at loop. Translate text.
2846 2018-08-06 David Malcolm <dmalcolm@redhat.com>
2848 * function-tests.c (selftest::test_expansion_to_rtl): Call
2849 free_after_compilation.
2851 2018-08-06 Alan Hayward <alan.hayward@arm.com>
2853 * config/aarch64/aarch64.md: Add clobber highs to tls_desc.
2855 2018-08-06 Andreas Krebbel <krebbel@linux.ibm.com>
2857 * config/s390/s390.c (s390_loop_unroll_adjust): Prevent small
2858 loops with memory block operations from getting unrolled.
2860 2018-08-06 Ulrich Weigand <uweigand@de.ibm.com>
2863 * config/spu/spu.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
2864 Define to speculation_safe_value_not_needed.
2866 2018-08-06 Jeff Law <law@redhat.com>
2868 * reload1.c (forget_old_reloads_1): Adjust CLOBBER_HIGH
2871 2018-08-06 Jozef Lawrynowicz <jozef.l@mittosystems.com>
2874 * gcc/tree.c (build_common_tree_nodes): Initialize integer_types array
2875 with all enabled __intN types.
2877 * gcc/testsuite/gcc.target/msp430/pr86662.c: New test.
2879 2018-08-06 Alan Hayward <alan.hayward@arm.com>
2881 * alias.c (record_set): Check for clobber high.
2882 * cfgexpand.c (expand_gimple_stmt): Likewise.
2883 * combine-stack-adj.c (single_set_for_csa): Likewise.
2884 * combine.c (find_single_use_1): Likewise.
2885 (set_nonzero_bits_and_sign_copies): Likewise.
2886 (get_combine_src_dest): Likewise.
2887 (is_parallel_of_n_reg_sets): Likewise.
2888 (try_combine): Likewise.
2889 (record_dead_and_set_regs_1): Likewise.
2890 (reg_dead_at_p_1): Likewise.
2891 (reg_dead_at_p): Likewise.
2892 * dce.c (deletable_insn_p): Likewise.
2893 (mark_nonreg_stores_1): Likewise.
2894 (mark_nonreg_stores_2): Likewise.
2895 * df-scan.c (df_find_hard_reg_defs): Likewise.
2896 (df_uses_record): Likewise.
2897 (df_get_call_refs): Likewise.
2898 * dwarf2out.c (mem_loc_descriptor): Likewise.
2899 * haifa-sched.c (haifa_classify_rtx): Likewise.
2900 * ira-build.c (create_insn_allocnos): Likewise.
2901 * ira-costs.c (scan_one_insn): Likewise.
2902 * ira.c (equiv_init_movable_p): Likewise.
2903 (rtx_moveable_p): Likewise.
2904 (interesting_dest_for_shprep): Likewise.
2905 * jump.c (mark_jump_label_1): Likewise.
2906 * postreload-gcse.c (record_opr_changes): Likewise.
2907 * postreload.c (reload_cse_simplify): Likewise.
2908 (struct reg_use): Add source expr.
2909 (reload_combine): Check for clobber high.
2910 (reload_combine_note_use): Likewise.
2911 (reload_cse_move2add): Likewise.
2912 (move2add_note_store): Likewise.
2913 * print-rtl.c (print_pattern): Likewise.
2914 * recog.c (decode_asm_operands): Likewise.
2915 (store_data_bypass_p): Likewise.
2916 (if_test_bypass_p): Likewise.
2917 * regcprop.c (kill_clobbered_value): Likewise.
2918 (kill_set_value): Likewise.
2919 * reginfo.c (reg_scan_mark_refs): Likewise.
2920 * reload1.c (maybe_fix_stack_asms): Likewise.
2921 (eliminate_regs_1): Likewise.
2922 (elimination_effects): Likewise.
2923 (mark_not_eliminable): Likewise.
2924 (scan_paradoxical_subregs): Likewise.
2925 (forget_old_reloads_1): Likewise.
2926 * reorg.c (find_end_label): Likewise.
2927 (try_merge_delay_insns): Likewise.
2928 (redundant_insn): Likewise.
2929 (own_thread_p): Likewise.
2930 (fill_simple_delay_slots): Likewise.
2931 (fill_slots_from_thread): Likewise.
2932 (dbr_schedule): Likewise.
2933 * resource.c (update_live_status): Likewise.
2934 (mark_referenced_resources): Likewise.
2935 (mark_set_resources): Likewise.
2936 * rtl.c (copy_rtx): Likewise.
2937 * rtlanal.c (reg_referenced_p): Likewise.
2938 (single_set_2): Likewise.
2939 (noop_move_p): Likewise.
2940 (note_stores): Likewise.
2941 * sched-deps.c (sched_analyze_reg): Likewise.
2942 (sched_analyze_insn): Likewise.
2944 2018-08-06 Alan Hayward <alan.hayward@arm.com>
2946 * cse.c (invalidate_reg): New function extracted from...
2947 (invalidate): ...here.
2948 (canonicalize_insn): Check for clobber high.
2949 (invalidate_from_clobbers): invalidate clobber highs.
2950 (invalidate_from_sets_and_clobbers): Likewise.
2951 (count_reg_usage): Check for clobber high.
2952 (insn_live_p): Likewise.
2953 * cselib.c (cselib_expand_value_rtx_1):Likewise.
2954 (cselib_invalidate_regno): Check for clobber in setter.
2955 (cselib_invalidate_rtx): Pass through setter.
2956 (cselib_invalidate_rtx_note_stores):
2957 (cselib_process_insn): Check for clobber high.
2958 * cselib.h (cselib_invalidate_rtx): Add operand.
2960 2018-08-06 Alan Hayward <alan.hayward@arm.com>
2962 * lra-eliminations.c (lra_eliminate_regs_1): Check for clobber high.
2963 (mark_not_eliminable): Likewise.
2964 * lra-int.h (struct lra_insn_reg): Add clobber high marker.
2965 * lra-lives.c (process_bb_lives): Check for clobber high.
2966 * lra.c (new_insn_reg): Remember clobber highs.
2967 (collect_non_operand_hard_regs): Check for clobber high.
2968 (lra_set_insn_recog_data): Likewise.
2969 (add_regs_to_insn_regno_info): Likewise.
2970 (lra_update_insn_regno_info): Likewise.
2972 2018-08-06 Alan Hayward <alan.hayward@arm.com>
2974 * rtl.h (reg_is_clobbered_by_clobber_high): Add declarations.
2975 * rtlanal.c (reg_is_clobbered_by_clobber_high): Add function.
2977 2018-08-06 Alan Hayward <alan.hayward@arm.com>
2979 * emit-rtl.c (verify_rtx_sharing): Check for CLOBBER_HIGH.
2980 (copy_insn_1): Likewise.
2981 (gen_hard_reg_clobber_high): New gen function.
2982 * genconfig.c (walk_insn_part): Check for CLOBBER_HIGH.
2983 * genemit.c (gen_exp): Likewise.
2984 (gen_emit_seq): Pass through info.
2985 (gen_insn): Check for CLOBBER_HIGH.
2986 (gen_expand): Pass through info.
2987 (gen_split): Likewise.
2988 (output_add_clobbers): Likewise.
2989 * genrecog.c (validate_pattern): Check for CLOBBER_HIGH.
2990 (remove_clobbers): Likewise.
2991 * rtl.h (gen_hard_reg_clobber_high): New declaration.
2993 2018-08-06 Alan Hayward <alan.hayward@arm.com>
2995 * doc/rtl.texi (clobber_high): Add.
2996 (parallel): Add in clobber high
2997 * rtl.c (rtl_check_failed_code3): Add function.
2998 * rtl.def (CLOBBER_HIGH): Add expression.
2999 * rtl.h (RTL_CHECKC3): Add macro.
3000 (rtl_check_failed_code3): Add declaration.
3001 (XC3EXP): Add macro.
3003 2018-08-05 H.J. Lu <hongjiu.lu@intel.com>
3006 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Set
3007 cfun->machine->max_used_stack_alignment if needed.
3009 2018-08-04 Martin Sebor <msebor@redhat.com>
3011 PR tree-optimization/86571
3012 * gimple-ssa-sprintf.c (format_floating): Extend upper bound of
3015 2018-08-03 Sandra Loosemore <sandra@codesourcery.com>
3018 * config/nios2/nios2.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
3021 2018-08-03 Jeff Law <law@redhat.com>
3024 * config/mn10300/mn10300.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
3025 Define to speculation_safe_value_not_needed.
3027 2018-08-03 David Malcolm <dmalcolm@redhat.com>
3029 * doc/gcov.texi (-x): Remove duplicate "to".
3030 * doc/invoke.texi (-Wnoexcept-type): Remove duplicate "calls".
3031 (-Wif-not-aligned): Remove duplicate "is".
3032 (-flto): Remove duplicate "the".
3033 (MicroBlaze Options): In examples of "-mcpu=cpu-type", remove
3034 duplicate "v5.00.b".
3035 (MSP430 Options): Remove duplicate "and" from the description
3036 of "-mgprel-sec=regexp".
3037 (x86 Options): Remove duplicate copies of "vmldLog102" and
3038 vmlsLog104 from description of "-mveclibabi=type".
3040 2018-08-03 Richard Sandiford <richard.sandiford@arm.com>
3042 * internal-fn.h (first_commutative_argument): Declare.
3043 * internal-fn.c (first_commutative_argument): New function.
3044 * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove extra
3045 restrictions for pattern statements. Use first_commutative_argument
3046 to look for commutative operands in calls to internal functions.
3048 2018-08-03 Aldy Hernandez <aldyh@redhat.com>
3050 * Makefile.in (wide-int-range.o): New.
3051 * tree-vrp.c: Move all the wide_int_* functions to...
3052 * wide-int-range.cc: ...here.
3053 * tree-vrp.h: Move all the wide_int_* prototypes to...
3054 * wide-int-range.h: ...here.
3056 2018-08-03 Tom de Vries <tdevries@suse.de>
3058 * common/config/nvptx/nvptx-common.c (nvptx_except_unwind_info): Return
3060 * config/nvptx/nvptx.c (TARGET_ASM_BYTE_OP): Remove define.
3061 * except.c (output_function_exception_table): Do early exit if
3062 targetm_common.except_unwind_info (&global_options) == UI_NONE.
3064 2018-08-03 Martin Liska <mliska@suse.cz>
3066 * predict.c (dump_prediction): Change to 2 digits
3069 2018-08-03 Siddhesh Poyarekar <siddhesh@sourceware.org>
3071 * config/aarch64/falkor.md (falkor_am_1_vxvy_vxvy): Move
3073 (falkor_am_1_gtov_gtov): ... a new insn reservation.
3075 2018-07-19 Ilya Leoshkevich <iii@linux.ibm.com>
3077 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Replace > with >=.
3078 * df-problems.c (df_remove_dead_eq_notes): Replace > with >=.
3079 * dwarf2out.c (mem_loc_descriptor): Replace > with >=.
3080 * lra-constraints.c (spill_hard_reg_in_range): Replace <= with <.
3081 * lra-remat.c (call_used_input_regno_present_p): Replace <= with <.
3083 2018-08-02 David Malcolm <dmalcolm@redhat.com>
3085 * diagnostic-show-locus.c (diagnostic_show_locus): Use
3086 pp_take_prefix when saving the existing prefix.
3087 * diagnostic.c (diagnostic_append_note): Likewise.
3088 * langhooks.c (lhd_print_error_function): Likewise.
3089 * pretty-print.c (pp_set_prefix): Drop the "const" from "prefix"
3090 param's type. Free the existing prefix.
3091 (pp_take_prefix): New function.
3092 (pretty_printer::pretty_printer): Drop the prefix parameter.
3093 Rename the length parameter to match the comment.
3094 (pretty_printer::~pretty_printer): Free the prefix.
3095 * pretty-print.h (pretty_printer::pretty_printer): Drop the prefix
3097 (struct pretty_printer): Drop the "const" from "prefix" field's
3098 type and clarify memory management.
3099 (pp_set_prefix): Drop the "const" from the 2nd param.
3100 (pp_take_prefix): New decl.
3102 2018-08-02 Aaron Sawdey <acsawdey@linux.ibm.com>
3104 * config/rs6000/rs6000-string.c (select_block_compare_mode): Move test
3105 for word_mode_ok here instead of passing as argument.
3106 (expand_block_compare): Change select_block_compare_mode() call.
3107 (expand_strncmp_gpr_sequence): New function.
3108 (expand_strn_compare): Make use of expand_strncmp_gpr_sequence.
3110 2018-08-02 Jeff Law <law@redhat.com>
3113 * config/m68k/m68k.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
3114 Define to speculation_safe_value_not_needed.
3117 * config/h8300/h8300.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
3118 Define to speculation_safe_value_not_needed.
3120 2018-08-02 Tom de Vries <tdevries@suse.de>
3123 * common/config/nvptx/nvptx-common.c (nvptx_except_unwind_info): New
3124 function. Return UI_TARGET unconditionally.
3125 (TARGET_EXCEPT_UNWIND_INFO): Redefine to nvptx_except_unwind_info.
3126 * config/nvptx/nvptx.c (TARGET_ASM_BYTE_OP): Emit commented out '.byte'.
3128 2018-08-02 Richard Sandiford <richard.sandiford@arm.com>
3130 * genemit.c (print_overload_test): Fix typo.
3132 2018-08-02 Richard Biener <rguenther@suse.de>
3134 PR tree-optimization/86816
3135 * tree-ssa-tail-merge.c (tail_merge_valueize): New function
3136 which checks for value availability before querying it.
3137 (gvn_uses_equal): Use it.
3138 (same_succ_hash): Likewise.
3139 (gimple_equal_p): Likewise.
3141 2018-08-02 Nick Clifton <nickc@redhat.com>
3144 * config/stormy16/stormy16.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
3145 Define to speculation_safe_value_not_needed.
3148 * config/v850/v850.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
3149 Define to speculation_safe_value_not_needed.
3152 * config/v850/v850.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
3153 Define to speculation_safe_value_not_needed.
3156 * config/rx/rx.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
3157 Define to speculation_safe_value_not_needed.
3160 * config/msp430/msp430.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
3161 Define to speculation_safe_value_not_needed.
3164 * config/mcore/mcore.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
3165 Define to speculation_safe_value_not_needed.
3168 * config/m32r/m32r.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
3169 Define to speculation_safe_value_not_needed.
3172 * config/iq2000/iq2000.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
3173 Define to speculation_safe_value_not_needed.
3176 * config/frv/frv.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define to
3177 speculation_safe_value_not_needed.
3180 * config/fr30/fr30.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define
3181 to speculation_safe_value_not_needed.
3183 2018-08-02 Richard Sandiford <richard.sandiford@arm.com>
3185 * doc/md.texi: Expand the documentation of instruction names
3186 to mention port-local uses. Document '@' in pattern names.
3187 * read-md.h (overloaded_instance, overloaded_name): New structs.
3189 (md_reader::handle_overloaded_name): New member function.
3190 (md_reader::get_overloads): Likewise.
3191 (md_reader::m_first_overload): New member variable.
3192 (md_reader::m_next_overload_ptr): Likewise.
3193 (md_reader::m_overloads_htab): Likewise.
3194 * read-md.c (md_reader::md_reader): Initialize m_first_overload,
3195 m_next_overload_ptr and m_overloads_htab.
3196 * read-rtl.c (iterator_group): Add "type" and "get_c_token" fields.
3197 (get_mode_token, get_code_token, get_int_token): New functions.
3198 (map_attr_string): Add an optional argument that passes back
3199 the associated iterator.
3200 (overloaded_name_hash, overloaded_name_eq_p, named_rtx_p):
3201 (md_reader::handle_overloaded_name, add_overload_instance): New
3203 (apply_iterators): Handle '@' names. Report an error if '@'
3204 is used without iterators.
3205 (initialize_iterators): Initialize the new iterator_group fields.
3206 * genopinit.c (handle_overloaded_code_for)
3207 (handle_overloaded_gen): New functions.
3208 (main): Use them to print declarations of maybe_code_for_* and
3209 maybe_gen_* functions, and inline definitions of code_for_* and gen_*.
3210 * genemit.c (print_overload_arguments, print_overload_test)
3211 (handle_overloaded_code_for, handle_overloaded_gen): New functions.
3212 (main): Use it to print definitions of maybe_code_for_* and
3213 maybe_gen_* functions.
3214 * config/aarch64/aarch64.c (aarch64_split_128bit_move): Use
3215 gen_aarch64_mov{low,high}_di and gen_aarch64_movdi_{low,high}
3216 instead of explicit mode checks.
3217 (aarch64_split_simd_combine): Likewise gen_aarch64_simd_combine.
3218 (aarch64_split_simd_move): Likewise gen_aarch64_split_simd_mov.
3219 (aarch64_emit_load_exclusive): Likewise gen_aarch64_load_exclusive.
3220 (aarch64_emit_store_exclusive): Likewise gen_aarch64_store_exclusive.
3221 (aarch64_expand_compare_and_swap): Likewise
3222 gen_aarch64_compare_and_swap and gen_aarch64_compare_and_swap_lse
3223 (aarch64_gen_atomic_cas): Likewise gen_aarch64_atomic_cas.
3224 (aarch64_emit_atomic_swap): Likewise gen_aarch64_atomic_swp.
3225 (aarch64_constant_pool_reload_icode): Delete.
3226 (aarch64_secondary_reload): Use code_for_aarch64_reload_movcp
3227 instead of aarch64_constant_pool_reload_icode. Use
3228 code_for_aarch64_reload_mov instead of explicit mode checks.
3229 (rsqrte_type, get_rsqrte_type, rsqrts_type, get_rsqrts_type): Delete.
3230 (aarch64_emit_approx_sqrt): Use gen_aarch64_rsqrte instead of
3231 get_rsqrte_type and gen_aarch64_rsqrts instead of gen_rqrts_type.
3232 (recpe_type, get_recpe_type, recps_type, get_recps_type): Delete.
3233 (aarch64_emit_approx_div): Use gen_aarch64_frecpe instead of
3234 get_recpe_type and gen_aarch64_frecps instead of get_recps_type.
3235 (aarch64_atomic_load_op_code): Delete.
3236 (aarch64_emit_atomic_load_op): Likewise.
3237 (aarch64_gen_atomic_ldop): Use UNSPECV_ATOMIC_* instead of
3238 aarch64_atomic_load_op_code. Use gen_aarch64_atomic_load
3239 instead of aarch64_emit_atomic_load_op.
3240 * config/aarch64/aarch64.md (aarch64_reload_movcp<GPF_TF:mode><P:mode>)
3241 (aarch64_reload_movcp<VALL:mode><P:mode>, aarch64_reload_mov<mode>)
3242 (aarch64_movdi_<mode>low, aarch64_movdi_<mode>high)
3243 (aarch64_mov<mode>high_di, aarch64_mov<mode>low_di): Add a '@'
3244 character before the pattern name.
3245 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
3246 (aarch64_rsqrte<mode>, aarch64_rsqrts<mode>)
3247 (aarch64_simd_combine<mode>, aarch64_frecpe<mode>)
3248 (aarch64_frecps<mode>): Likewise.
3249 * config/aarch64/atomics.md (atomic_compare_and_swap<mode>)
3250 (aarch64_compare_and_swap<mode>, aarch64_compare_and_swap<mode>_lse)
3251 (aarch64_load_exclusive<mode>, aarch64_store_exclusive<mode>)
3252 (aarch64_atomic_swp<mode>, aarch64_atomic_cas<mode>)
3253 (aarch64_atomic_load<atomic_ldop><mode>): Likewise.
3255 2018-08-02 Richard Sandiford <richard.sandiford@arm.com>
3257 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
3258 Allow HFmode constants if TARGET_FP_F16INST.
3260 2018-08-02 Jackson Woodruff <jackson.woodruff@arm.com>
3263 * config/aarch64/aarch64.c (aarch64_operands_adjust_ok_for_ldpstp):
3264 No longer check last store for clobber of address register.
3266 2018-08-02 Martin Liska <mliska@suse.cz>
3268 PR gcov-profile/86817
3269 * gcov.c (process_all_functions): New function.
3271 (process_file): Move functions processing to
3272 process_all_functions.
3274 2018-08-02 David Malcolm <dmalcolm@redhat.com>
3276 * dumpfile.c (dump_user_location_t::dump_user_location_t): Add
3277 "const" to the "gimple *" and "rtx_insn *" parameters.
3278 * dumpfile.h (dump_user_location_t::dump_user_location_t):
3280 (dump_location_t::dump_location_t): Likewise.
3282 2018-08-01 Martin Sebor <msebor@redhat.com>
3284 PR tree-optimization/86650
3285 * gimple-pretty-print.c (percent_G_format): Accept a "gimple *"
3286 rather than a "gcall *". Directly pass the data of interest
3287 to percent_K_format, rather than building a temporary CALL_EXPR
3289 * gimple-fold.c (gimple_fold_builtin_strncpy): Adjust.
3290 (gimple_fold_builtin_strncat): Adjust.
3291 * gimple-ssa-warn-restrict.h (check_bounds_or_overlap): Replace
3292 gcall* argument with gimple*.
3293 * gimple-ssa-warn-restrict.c (check_call): Same.
3294 (wrestrict_dom_walker::before_dom_children): Same.
3295 (builtin_access::builtin_access): Same.
3296 (check_bounds_or_overlap): Same
3297 (maybe_diag_overlap): Same.
3298 (maybe_diag_offset_bounds): Same.
3299 * tree-diagnostic.c (default_tree_printer): Move usage of
3300 EXPR_LOCATION (t) and TREE_BLOCK (t) from within percent_K_format
3302 * tree-pretty-print.c (percent_K_format): Add argument.
3303 * tree-pretty-print.h: Add argument.
3304 * tree-ssa-ccp.c (pass_post_ipa_warn::execute): Adjust.
3305 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Adjust.
3306 (maybe_diag_stxncpy_trunc): Same.
3307 (handle_builtin_stxncpy): Same.
3308 (handle_builtin_strcat): Same.
3310 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
3312 * match.pd: Optimise pointer range checks.
3314 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
3316 PR tree-optimization/86758
3317 * tree-vect-stmts.c (vectorizable_simd_clone_call): Don't try
3318 to remove pattern statements.
3320 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
3322 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Use the
3323 result of dfs_enumerate_from when constructing stmt_vec_infos,
3324 instead of additionally calling get_loop_body.
3326 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
3328 * tree-vectorizer.h (vect_create_data_ref_ptr): Remove inv_p
3330 * tree-vect-data-refs.c (vect_create_data_ref_ptr): Likewise.
3331 When creating an iv, assert that the step is not known to be zero.
3332 (vect_setup_realignment): Update call accordingly.
3333 * tree-vect-stmts.c (vectorizable_store): Likewise.
3334 (vectorizable_load): Likewise. Handle VMAT_INVARIANT separately.
3336 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
3338 * tree-vectorizer.h (vect_stmt_to_vectorize): New function.
3339 * tree-vect-loop.c (vect_update_vf_for_slp): Use it.
3340 (vectorizable_reduction): Likewise.
3341 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
3342 (vect_detect_hybrid_slp_stmts): Likewise.
3343 * tree-vect-stmts.c (vect_is_simple_use): Likewise.
3345 2018-08-01 Aldy Hernandez <aldyh@redhat.com>
3347 * tree-vrp (zero_nonzero_bits_from_bounds): Rename to...
3348 (wide_int_set_zero_nonzero_bits): ...this.
3349 (zero_nonzero_bits_from_vr): Rename to...
3350 (vrp_set_zero_nonzero_bits): ...this.
3351 (extract_range_from_multiplicative_op_1): Abstract wide int
3353 (wide_int_range_multiplicative_op): ...here.
3354 (extract_range_from_binary_expr_1): Extract wide int binary
3355 operations into their own functions.
3356 (wide_int_range_lshift): New.
3357 (wide_int_range_can_optimize_bit_op): New.
3358 (wide_int_range_shift_undefined_p): New.
3359 (wide_int_range_bit_xor): New.
3360 (wide_int_range_bit_ior): New.
3361 (wide_int_range_bit_and): New.
3362 (wide_int_range_trunc_mod): New.
3363 (extract_range_into_wide_ints): New.
3364 (vrp_shift_undefined_p): New.
3365 (extract_range_from_multiplicative_op): New.
3366 (vrp_can_optimize_bit_op): New.
3367 * tree-vrp.h (value_range::dump): New.
3368 (wide_int_range_multiplicative_op): New.
3369 (wide_int_range_lshift):New.
3370 (wide_int_range_shift_undefined_p): New.
3371 (wide_int_range_bit_xor): New.
3372 (wide_int_range_bit_ior): New.
3373 (wide_int_range_bit_and): New.
3374 (wide_int_range_trunc_mod): New.
3375 (zero_nonzero_bits_from_bounds): Rename to...
3376 (wide_int_set_zero_nonzero_bits): ...this.
3377 (zero_nonzero_bits_from_vr): Rename to...
3378 (vrp_set_zero_nonzero_bits): ...this.
3379 (range_easy_mask_min_max): Rename to...
3380 (wide_int_range_can_optimize_bit_op): this.
3381 * vr-values.c (simplify_bit_ops_using_ranges): Rename
3382 zero_nonzero_bits_from_vr into vrp_set_zero_nonzero_bits.
3384 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
3386 * tree-vectorizer.h (vect_orig_stmt): New function.
3387 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Use it.
3388 * tree-vect-loop.c (vect_model_reduction_cost): Likewise.
3389 (vect_create_epilog_for_reduction): Likewise.
3390 (vectorizable_live_operation): Likewise.
3391 * tree-vect-slp.c (vect_find_last_scalar_stmt_in_slp): Likewise.
3392 (vect_detect_hybrid_slp_stmts, vect_schedule_slp): Likewise.
3393 * tree-vect-stmts.c (vectorizable_call): Likewise.
3394 (vectorizable_simd_clone_call, vect_remove_stores): Likewise.
3396 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
3398 * tree-vectorizer.h (vect_transform_stmt): Remove grouped_store
3400 * tree-vect-stmts.c (vect_transform_stmt): Likewise.
3401 * tree-vect-loop.c (vect_transform_loop_stmt): Update call accordingly.
3402 (vect_transform_loop): Likewise.
3403 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
3405 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
3407 * tree-vectorizer.h (vect_schedule_slp): Return void.
3408 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
3409 (vect_schedule_slp): Likewise.
3411 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
3413 * tree-vect-loop.c (vect_transform_loop_stmt): Remove slp_scheduled
3415 (vect_transform_loop): Update calls accordingly. Schedule SLP
3416 instances before the main loop, if any exist.
3418 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
3420 PR tree-optimization/86749
3421 * tree-vect-patterns.c (vect_determine_min_output_precision_1):
3422 If the lhs is used in a COND_EXPR, check that it is being used
3423 as the "then" or "else" value.
3425 2018-08-01 Tom de Vries <tdevries@suse.de>
3428 * config/nvptx/nvptx.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define to
3429 speculation_safe_value_not_needed.
3431 2018-08-01 Richard Biener <rguenther@suse.de>
3433 * tree-ssa-sccvn.c (visit_phi): Compare invariant addresses
3436 2018-08-01 Martin Liska <mliska@suse.cz>
3438 * value-prof.c (gimple_divmod_fixed_value_transform): Unify
3439 format how successful transformation is dumped.
3440 (gimple_mod_pow2_value_transform): Likewise.
3441 (gimple_mod_subtract_transform): Likewise.
3442 (gimple_stringops_transform): Likewise.
3444 2018-08-01 Martin Liska <mliska@suse.cz>
3447 * value-prof.c (interesting_stringop_to_profile_p):
3448 Simplify the code and add BUILT_IN_MEMMOVE.
3449 (gimple_stringops_transform): Likewise.
3451 2018-08-01 Sam Tebbs <sam.tebbs@arm.com>
3453 * config/aarch64/aarch64-simd.md
3454 (*aarch64_get_lane_zero_extendsi<mode>): Rename to...
3455 (*aarch64_get_lane_zero_extend<GPI:mode><VDQQH:mode>): ... This and
3456 use GPI iterator instead of SI mode.
3458 2018-08-01 Richard Earnshaw <rearnsha@arm.com>
3460 * config/rs6000/rs6000.md (speculation_barrier): Renamed from
3461 rs6000_speculation_barrier.
3462 * config/rs6000/rs6000.c (rs6000_expand_builtin): Adjust for
3463 new barrier pattern name.
3465 2018-08-01 Richard Earnshaw <rearnsha@arm.com>
3467 * config/i386/i386.md (unspecv): Add UNSPECV_SPECULATION_BARRIER.
3468 (speculation_barrier): New insn.
3470 2018-08-01 Richard Biener <rguenther@suse.de>
3473 * graphite.h: Include isl/id.h and isl/space.h to allow build
3476 2018-08-01 Jan Willem Jagersma <jwjagersma@gmail.com>
3479 * dwarf2out.c (dwarf2out_early_finish): Do not generate assembly in LTO
3480 mode for COFF targets.
3481 * defaults.h (TARGET_COFF): Define.
3482 * config/i386/djgpp.h (TARGET_ASM_LTO_START, TARGET_ASM_LTO_END,
3483 TARGET_COFF): Define.
3484 (i386_djgpp_asm_lto_start, i386_djgpp_asm_lto_end): Declare.
3485 * config/i386/djgpp.c (saved_debug_info_level): New static variable.
3486 (i386_djgpp_asm_lto_start, i386_djgpp_asm_lto_end): New functions.
3488 2018-07-31 Alexandre Oliva <oliva@adacore.com>
3490 * gimple-streamer-in.c (input_bb): Restore BB discriminator.
3491 * gimple-streamer-out.c (output_bb): Save it.
3492 * lto-streamer-in.c (input_struct_function_base): Restore
3493 instance discriminator if available. Create map on demand.
3494 * lto-streamer-out.c (output_struct_function_base): Save it if
3496 * final.c (decl_to_instance_map): Document LTO strategy.
3498 2018-07-31 Alexandre Oliva <oliva@adacore.com>
3499 Olivier Hainque <hainque@adacore.com>
3501 * debug.h (decl_to_instance_map_t): New type.
3502 (decl_to_instance_map): Declare.
3503 (maybe_create_decl_to_instance_map): New inline function.
3504 * final.c (bb_discriminator, last_bb_discriminator): New statics,
3505 to track basic block discriminators.
3506 (final_start_function_1): Initialize them.
3507 (final_scan_insn_1): On NOTE_INSN_BASIC_BLOCK, track
3509 (decl_to_instance_map): New variable.
3510 (map_decl_to_instance, maybe_set_discriminator): New functions.
3511 (notice_source_line): Set discriminator.
3513 2018-07-31 Ian Lance Taylor <iant@golang.org>
3515 * targhooks.c (default_have_speculation_safe_value): Add
3518 2018-07-31 David Malcolm <dmalcolm@redhat.com>
3520 * dump-context.h: Include "pretty-print.h".
3521 (dump_context::refresh_dumps_are_enabled): New decl.
3522 (dump_context::emit_item): New decl.
3523 (class dump_context): Add fields "m_test_pp" and
3525 (temp_dump_context::temp_dump_context): Add param "test_pp_flags".
3526 (temp_dump_context::get_dumped_text): New decl.
3527 (class temp_dump_context): Add field "m_pp".
3528 * dumpfile.c (refresh_dumps_are_enabled): Convert to...
3529 (dump_context::refresh_dumps_are_enabled): ...and add a test for
3531 (set_dump_file): Update for above change.
3532 (set_alt_dump_file): Likewise.
3533 (dump_loc): New overload, taking a pretty_printer *.
3534 (dump_context::dump_loc): Call end_any_optinfo. Dump the location
3535 to any test pretty-printer.
3536 (make_item_for_dump_gimple_stmt): New function, adapted from
3537 optinfo::add_gimple_stmt.
3538 (dump_context::dump_gimple_stmt): Call it, and use the result,
3539 eliminating the direct usage of dump_file and alt_dump_file in
3540 favor of indirectly using them via emit_item.
3541 (make_item_for_dump_gimple_expr): New function, adapted from
3542 optinfo::add_gimple_expr.
3543 (dump_context::dump_gimple_expr): Call it, and use the result,
3544 eliminating the direct usage of dump_file and alt_dump_file in
3545 favor of indirectly using them via emit_item.
3546 (make_item_for_dump_generic_expr): New function, adapted from
3548 (dump_context::dump_generic_expr): Call it, and use the result,
3549 eliminating the direct usage of dump_file and alt_dump_file in
3550 favor of indirectly using them via emit_item.
3551 (make_item_for_dump_printf_va): New function, adapted from
3552 optinfo::add_printf_va.
3553 (make_item_for_dump_printf): New function.
3554 (dump_context::dump_printf_va): Call make_item_for_dump_printf_va,
3555 and use the result, eliminating the direct usage of dump_file and
3556 alt_dump_file in favor of indirectly using them via emit_item.
3557 (make_item_for_dump_dec): New function.
3558 (dump_context::dump_dec): Call it, and use the result,
3559 eliminating the direct usage of dump_file and alt_dump_file in
3560 favor of indirectly using them via emit_item.
3561 (make_item_for_dump_symtab_node): New function, adapted from
3562 optinfo::add_symtab_node.
3563 (dump_context::dump_symtab_node): Call it, and use the result,
3564 eliminating the direct usage of dump_file and alt_dump_file in
3565 favor of indirectly using them via emit_item.
3566 (dump_context::begin_scope): Reimplement, avoiding direct usage
3567 of dump_file and alt_dump_file in favor of indirectly using them
3569 (dump_context::emit_item): New member function.
3570 (temp_dump_context::temp_dump_context): Add param "test_pp_flags".
3571 Set up test pretty-printer on the underlying context. Call
3572 refresh_dumps_are_enabled.
3573 (temp_dump_context::~temp_dump_context): Call
3574 refresh_dumps_are_enabled.
3575 (temp_dump_context::get_dumped_text): New member function.
3576 (selftest::verify_dumped_text): New function.
3577 (ASSERT_DUMPED_TEXT_EQ): New macro.
3578 (selftest::test_capture_of_dump_calls): Run all tests twice, with
3579 and then without optinfo enabled. Add uses of
3580 ASSERT_DUMPED_TEXT_EQ to all tests. Add test of nested scopes.
3581 * dumpfile.h: Update comment for the dump_* API.
3582 * optinfo-emit-json.cc
3583 (selftest::test_building_json_from_dump_calls): Update for new
3584 param for temp_dump_context ctor.
3585 * optinfo.cc (optinfo_item::optinfo_item): Remove "owned" param
3586 and "m_owned" field.
3587 (optinfo_item::~optinfo_item): Likewise.
3588 (optinfo::add_item): New member function.
3589 (optinfo::emit): Update comment.
3590 (optinfo::add_string): Delete.
3591 (optinfo::add_printf): Delete.
3592 (optinfo::add_printf_va): Delete.
3593 (optinfo::add_gimple_stmt): Delete.
3594 (optinfo::add_gimple_expr): Delete.
3595 (optinfo::add_tree): Delete.
3596 (optinfo::add_symtab_node): Delete.
3597 (optinfo::add_dec): Delete.
3598 * optinfo.h (class dump_context): New forward decl.
3599 (optinfo::add_item): New decl.
3600 (optinfo::add_string): Delete.
3601 (optinfo::add_printf): Delete.
3602 (optinfo::add_printf_va): Delete.
3603 (optinfo::add_gimple_stmt): Delete.
3604 (optinfo::add_gimple_expr): Delete.
3605 (optinfo::add_tree): Delete.
3606 (optinfo::add_symtab_node): Delete.
3607 (optinfo::add_dec): Delete.
3608 (optinfo::add_poly_int): Delete.
3609 (optinfo_item::optinfo_item): Remove "owned" param.
3610 (class optinfo_item): Remove field "m_owned".
3612 2018-07-31 Jozef Lawrynowicz <jozef.l@mittosystems.com>
3615 * gcc/cfgexpand.c (set_parm_rtl): Use the alignment of Pmode when
3616 MAX_SUPPORTED_STACK_ALIGNMENT would otherwise be exceeded by the
3617 requested variable alignment.
3618 (expand_one_ssa_partition): Likewise.
3619 (expand_one_var): Likewise.
3621 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
3623 * config/pdp11/pdp11.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Redefine
3624 to speculation_safe_value_not_needed.
3626 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
3628 * targhooks.h (speculation_safe_value_not_needed): New prototype.
3629 * targhooks.c (speculation_safe_value_not_needed): New function.
3630 * target.def (have_speculation_safe_value): Update documentation.
3631 * doc/tm.texi: Regenerated.
3633 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
3635 * config/aarch64/iterators.md (ALLI_TI): New iterator.
3636 * config/aarch64/aarch64.md (despeculate_copy<ALLI_TI:mode>): New
3638 (despeculate_copy<ALLI:mode>_insn): New insn.
3639 (despeculate_copyti_insn): New insn.
3640 (despeculate_simple<ALLI:mode>): New insn
3641 (despeculate_simpleti): New insn.
3642 * config/aarch64/aarch64.c (aarch64_speculation_safe_value): New
3644 (TARGET_SPECULATION_SAFE_VALUE): Redefine to
3645 aarch64_speculation_safe_value.
3646 (aarch64_print_operand): Handle const0_rtx in modifier 'H'.
3648 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
3650 * config/aarch64/aarch64-speculation.cc: New file.
3651 * config/aarch64/aarch64-passes.def (pass_track_speculation): Add
3652 before pass_reorder_blocks.
3653 * config/aarch64/aarch64-protos.h (make_pass_track_speculation): Add
3655 * config/aarch64/aarch64.c (aarch64_conditional_register_usage): Fix
3656 X14 and X15 when tracking speculation.
3657 * config/aarch64/aarch64.md (register name constants): Add
3658 SPECULATION_TRACKER_REGNUM and SPECULATION_SCRATCH_REGNUM.
3659 (unspec): Add UNSPEC_SPECULATION_TRACKER.
3660 (speculation_barrier): New insn attribute.
3661 (cmp<mode>): Allow SP in comparisons.
3662 (speculation_tracker): New insn.
3663 (speculation_barrier): Add speculation_barrier attribute.
3664 * config/aarch64/t-aarch64: Add make rule for aarch64-speculation.o.
3665 * config.gcc (aarch64*-*-*): Add aarch64-speculation.o to extra_objs.
3666 * doc/invoke.texi (AArch64 Options): Document -mtrack-speculation.
3668 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
3670 * config/aarch64/aarch64.md (cb<optab><mode>1): Disable when
3671 aarch64_track_speculation is true.
3672 (tb<optab><mode>1): Likewise.
3673 * config/aarch64/aarch64.c (aarch64_split_compare_regs): Do not
3674 generate CB[N]Z when tracking speculation.
3675 (aarch64_split_compare_and_swap): Likewise.
3676 (aarch64_split_atomic_op): Likewise.
3678 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
3680 * config/aarch64/aarch64.opt (mtrack-speculation): New target option.
3682 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
3684 * config/aarch64.md (unspecv): Add UNSPECV_SPECULAION_BARRIER.
3685 (speculation_barrier): New insn.
3687 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
3689 * config/arm/unspecs.md (unspecv): Add VUNSPEC_SPECULATION_BARRIER.
3690 * config/arm/arm.md (speculation_barrier): New expand.
3691 (speculation_barrier_insn): New pattern.
3693 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
3695 * builtin-types.def (BT_FN_PTR_PTR_VAR): New function type.
3696 (BT_FN_I1_I1_VAR, BT_FN_I2_I2_VAR, BT_FN_I4_I4_VAR): Likewise.
3697 (BT_FN_I8_I8_VAR, BT_FN_I16_I16_VAR): Likewise.
3698 * builtin-attrs.def (ATTR_NOVOPS_NOTHROW_LEAF_LIST): New attribute
3700 * builtins.def (BUILT_IN_SPECULATION_SAFE_VALUE_N): New builtin.
3701 (BUILT_IN_SPECULATION_SAFE_VALUE_PTR): New internal builtin.
3702 (BUILT_IN_SPECULATION_SAFE_VALUE_1): Likewise.
3703 (BUILT_IN_SPECULATION_SAFE_VALUE_2): Likewise.
3704 (BUILT_IN_SPECULATION_SAFE_VALUE_4): Likewise.
3705 (BUILT_IN_SPECULATION_SAFE_VALUE_8): Likewise.
3706 (BUILT_IN_SPECULATION_SAFE_VALUE_16): Likewise.
3707 * builtins.c (expand_speculation_safe_value): New function.
3708 (expand_builtin): Call it.
3709 * doc/cpp.texi: Document predefine __HAVE_SPECULATION_SAFE_VALUE.
3710 * doc/extend.texi: Document __builtin_speculation_safe_value.
3711 * doc/md.texi: Document "speculation_barrier" pattern.
3712 * doc/tm.texi.in: Pull in TARGET_SPECULATION_SAFE_VALUE and
3713 TARGET_HAVE_SPECULATION_SAFE_VALUE.
3714 * doc/tm.texi: Regenerated.
3715 * target.def (have_speculation_safe_value, speculation_safe_value): New
3717 * targhooks.c (default_have_speculation_safe_value): New function.
3718 (default_speculation_safe_value): New function.
3719 * targhooks.h (default_have_speculation_safe_value): Add prototype.
3720 (default_speculation_safe_value): Add prototype.
3722 2018-07-31 David Malcolm <dmalcolm@redhat.com>
3724 * dump-context.h (dump_context::dump_loc): New decl.
3725 * dumpfile.c (dump_context::dump_loc): New member function.
3726 (dump_context::dump_gimple_stmt_loc): Reimplement using dump_loc
3727 and dump_gimple_stmt.
3728 (dump_context::dump_gimple_expr_loc): Likewise, using
3730 (dump_context::dump_generic_expr_loc): Likewise, using
3732 (dump_context::dump_printf_loc_va): Likewise, using
3734 (dump_context::begin_scope): Explicitly using the global function
3735 "dump_loc", rather than the member function.
3737 2018-07-31 Martin Sebor <msebor@redhat.com>
3739 PR tree-optimization/86741
3740 * tree-vrp.c (vrp_prop::check_mem_ref): Avoid incomplete types.
3742 2018-07-31 Andreas Krebbel <krebbel@linux.ibm.com>
3744 * config/s390/s390.c (s390_expand_setmem): Make the unrolling to
3745 depend on whether prefetch instructions will be emitted or not.
3746 Use TARGET_SETMEM_PFD for checking whether prefetch instructions
3747 will be emitted or not.
3748 * config/s390/s390.h (TARGET_SETMEM_PREFETCH_DISTANCE)
3749 (TARGET_SETMEM_PFD): New macros.
3751 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
3753 * tree-vectorizer.h (stmt_vec_info): Turn back into a typedef.
3754 (NULL_STMT_VEC_INFO): Delete.
3755 (stmt_vec_info::operator*): Likewise.
3756 (stmt_vec_info::operator gimple *): Likewise.
3757 * tree-vect-loop.c (vectorizable_reduction): Use NULL instead
3758 of NULL_STMT_VEC_INFO.
3759 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
3760 (vect_reassociating_reduction_p): Likewise.
3761 * tree-vect-stmts.c (vect_build_gather_load_calls): Likewise.
3762 (vectorizable_store): Likewise.
3763 * tree-vectorizer.c (vec_info::set_vinfo_for_stmt): Likewise.
3764 (vec_info::free_stmt_vec_infos): Likewise.
3766 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
3768 * tree-vectorizer.h (vect_stmt_in_region_p): Delete.
3769 * tree-vectorizer.c (vect_stmt_in_region_p): Likewise.
3771 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
3773 * tree-vectorizer.h (vec_info::new_vinfo_for_stmt)
3774 (vec_info::set_vinfo_for_stmt, vec_info::free_stmt_vec_infos)
3775 (vec_info::free_stmt_vec_info): New private member functions.
3776 (set_stmt_vec_info_vec, free_stmt_vec_infos, vinfo_for_stmt)
3777 (set_vinfo_for_stmt, new_stmt_vec_info, free_stmt_vec_info): Delete.
3778 * tree-parloops.c (gather_scalar_reductions): Remove calls to
3779 set_stmt_vec_info_vec and free_stmt_vec_infos.
3780 * tree-vect-loop.c (_loop_vec_info): Remove call to
3781 set_stmt_vec_info_vec.
3782 * tree-vect-stmts.c (new_stmt_vec_info, set_stmt_vec_info_vec)
3783 (free_stmt_vec_infos, free_stmt_vec_info): Delete in favor of...
3784 * tree-vectorizer.c (vec_info::new_stmt_vec_info)
3785 (vec_info::set_vinfo_for_stmt, vec_info::free_stmt_vec_infos)
3786 (vec_info::free_stmt_vec_info): ...these new functions. Remove
3787 assignments in {vec_info::,}new_stmt_vec_info that are redundant
3788 with the clearing in the xcalloc.
3789 (stmt_vec_info_vec): Delete.
3790 (vec_info::vec_info): Don't call set_stmt_vec_info_vec.
3791 (vectorize_loops): Likewise.
3792 (vec_info::~vec_info): Remove argument from call to
3793 free_stmt_vec_infos.
3794 (vec_info::add_stmt): Remove vinfo argument from call to
3797 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
3799 * tree-vectorizer.h (free_stmt_vec_info): Take a stmt_vec_info
3800 rather than a gimple stmt.
3801 * tree-vect-stmts.c (free_stmt_vec_info): Likewise. Don't free
3802 information for pattern statements when passed the original
3803 statement; instead wait to be passed the pattern statement itself.
3804 Don't call set_vinfo_for_stmt here.
3805 (free_stmt_vec_infos): Update call to free_stmt_vec_info.
3806 * tree-vect-loop.c (_loop_vec_info::~loop_vec_info): Don't free
3807 stmt_vec_infos here.
3808 * tree-vect-slp.c (_bb_vec_info::~bb_vec_info): Likewise.
3809 * tree-vectorizer.c (vec_info::remove_stmt): Nullify the statement's
3810 stmt_vec_infos entry.
3812 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
3814 * tree-vectorizer.h (vec_info::replace_stmt): Declare.
3815 * tree-vectorizer.c (vec_info::replace_stmt): New function.
3816 * tree-vect-slp.c (vect_remove_slp_scalar_calls): Use it.
3817 * tree-vect-stmts.c (vectorizable_call): Likewise.
3818 (vectorizable_simd_clone_call): Likewise.
3820 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
3822 * tree-vectorizer.h (vec_info::remove_stmt): Declare.
3823 * tree-vectorizer.c (vec_info::remove_stmt): New function.
3824 * tree-vect-loop-manip.c (vect_set_loop_condition): Use it.
3825 * tree-vect-loop.c (vect_transform_loop): Likewise.
3826 * tree-vect-slp.c (vect_schedule_slp): Likewise.
3827 * tree-vect-stmts.c (vect_remove_stores): Likewise.
3829 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
3831 * tree-vectorizer.h (vec_info::lookup_dr): New member function.
3832 (vect_dr_stmt): Delete.
3833 * tree-vectorizer.c (vec_info::lookup_dr): New function.
3834 * tree-vect-loop-manip.c (vect_update_inits_of_drs): Use it instead
3836 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr)
3837 (vect_analyze_data_ref_dependence, vect_record_base_alignments)
3838 (vect_verify_datarefs_alignment, vect_peeling_supportable)
3839 (vect_analyze_data_ref_accesses, vect_prune_runtime_alias_test_list)
3840 (vect_analyze_data_refs): Likewise.
3841 (vect_slp_analyze_data_ref_dependence): Likewise. Take a vec_info
3843 (vect_find_same_alignment_drs): Likewise.
3844 (vect_slp_analyze_node_dependences): Update calls accordingly.
3845 (vect_analyze_data_refs_alignment): Likewise. Use vec_info::lookup_dr
3846 instead of DR_VECT_AUX.
3847 (vect_get_peeling_costs_all_drs): Take a loop_vec_info instead
3848 of a vector data references. Use vec_info::lookup_dr instead of
3850 (vect_peeling_hash_get_lowest_cost): Update calls accordingly.
3851 (vect_enhance_data_refs_alignment): Likewise. Use vec_info::lookup_dr
3852 instead of DR_VECT_AUX.
3854 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
3856 * tree-vectorizer.h (_loop_vec_info::unaligned_dr): Change to
3858 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Update
3860 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
3861 * tree-vect-loop-manip.c (get_misalign_in_elems): Likewise.
3862 (vect_gen_prolog_loop_niters): Likewise.
3864 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
3866 * tree-vectorizer.h (set_dr_misalignment, dr_misalignment)
3867 (DR_TARGET_ALIGNMENT, aligned_access_p, known_alignment_for_access_p)
3868 (vect_known_alignment_in_bytes, vect_dr_behavior)
3869 (vect_get_scalar_dr_size): Take references as dr_vec_infos
3870 instead of data_references. Update calls to other routines for
3871 which the same change has been made.
3872 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Take
3873 dr_vec_infos instead of stmt_vec_infos.
3874 (vect_analyze_data_ref_dependence): Update call accordingly.
3875 (vect_slp_analyze_data_ref_dependence)
3876 (vect_record_base_alignments): Use DR_VECT_AUX.
3877 (vect_calculate_target_alignment, vect_compute_data_ref_alignment)
3878 (vect_update_misalignment_for_peel, verify_data_ref_alignment)
3879 (vector_alignment_reachable_p, vect_get_data_access_cost)
3880 (vect_peeling_supportable, vect_analyze_group_access_1)
3881 (vect_analyze_group_access, vect_analyze_data_ref_access)
3882 (vect_vfa_segment_size, vect_vfa_access_size, vect_vfa_align)
3883 (vect_compile_time_alias, vect_small_gap_p)
3884 (vectorizable_with_step_bound_p, vect_duplicate_ssa_name_ptr_info):
3885 (vect_supportable_dr_alignment): Take references as dr_vec_infos
3886 instead of data_references. Update calls to other routines for
3887 which the same change has been made.
3888 (vect_verify_datarefs_alignment, vect_get_peeling_costs_all_drs)
3889 (vect_find_same_alignment_drs, vect_analyze_data_refs_alignment)
3890 (vect_slp_analyze_and_verify_node_alignment)
3891 (vect_analyze_data_ref_accesses, vect_prune_runtime_alias_test_list)
3892 (vect_create_addr_base_for_vector_ref, vect_create_data_ref_ptr)
3893 (vect_setup_realignment): Use dr_vec_infos. Update calls after
3895 (_vect_peel_info::dr): Replace with...
3896 (_vect_peel_info::dr_info): ...this new field.
3897 (vect_peeling_hash_get_most_frequent)
3898 (vect_peeling_hash_choose_best_peeling): Update accordingly.
3899 (vect_peeling_hash_get_lowest_cost):
3900 (vect_enhance_data_refs_alignment): Likewise. Update calls to other
3901 routines for which the same change has been made.
3902 (vect_peeling_hash_insert): Likewise. Take a dr_vec_info instead of a
3904 * tree-vect-loop-manip.c (get_misalign_in_elems)
3905 (vect_gen_prolog_loop_niters): Use dr_vec_infos. Update calls after
3907 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
3908 * tree-vect-stmts.c (vect_get_store_cost, vect_get_load_cost)
3909 (vect_truncate_gather_scatter_offset, compare_step_with_zero)
3910 (get_group_load_store_type, get_negative_load_store_type)
3911 (vect_get_data_ptr_increment, vectorizable_store)
3912 (vectorizable_load): Likewise.
3913 (ensure_base_align): Take a dr_vec_info instead of a data_reference.
3914 Update calls to other routines for which the same change has been made.
3916 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
3918 * tree-vectorizer.h (vec_info::move_dr): New member function.
3919 (dataref_aux): Rename to...
3920 (dr_vec_info): ...this and add "dr" and "stmt" fields.
3921 (_stmt_vec_info::dr_aux): Update accordingly.
3922 (_stmt_vec_info::data_ref_info): Delete.
3923 (STMT_VINFO_GROUPED_ACCESS, DR_GROUP_FIRST_ELEMENT)
3924 (DR_GROUP_NEXT_ELEMENT, DR_GROUP_SIZE, DR_GROUP_STORE_COUNT)
3925 (DR_GROUP_GAP, DR_GROUP_SAME_DR_STMT, REDUC_GROUP_FIRST_ELEMENT):
3926 (REDUC_GROUP_NEXT_ELEMENT, REDUC_GROUP_SIZE): Use dr_aux.dr instead
3928 (STMT_VINFO_DATA_REF): Likewise. Turn into an lvalue.
3929 (STMT_VINFO_DR_INFO): New macro.
3930 (DR_VECT_AUX): Use STMT_VINFO_DR_INKFO and vect_dr_stmt.
3931 (set_dr_misalignment): Update after rename of dataref_aux.
3932 (vect_dr_stmt): Move earlier in file. Return dr_aux.stmt.
3933 * tree-vect-stmts.c (new_stmt_vec_info): Remove redundant
3934 initialization of STMT_VINFO_DATA_REF.
3935 * tree-vectorizer.c (vec_info::move_dr): New function.
3936 * tree-vect-patterns.c (vect_recog_bool_pattern)
3937 (vect_recog_mask_conversion_pattern)
3938 (vect_recog_gather_scatter_pattern): Use it.
3939 * tree-vect-data-refs.c (vect_analyze_data_refs): Initialize
3940 the "dr" and "stmt" fields of dr_vec_info instead of
3941 STMT_VINFO_DATA_REF.
3943 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
3945 * tree-vectorizer.h (_stmt_vec_info::pattern_stmt_p): New field.
3946 (is_pattern_stmt_p): Use it.
3947 * tree-vect-patterns.c (vect_init_pattern_stmt): Set pattern_stmt_p
3948 on pattern statements.
3950 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
3952 * tree-vect-patterns.c (vect_mark_pattern_stmts): Take the
3953 original stmt as a stmt_vec_info rather than a gimple stmt.
3954 (vect_pattern_recog_1): Take the statement directly as a
3955 stmt_vec_info, rather than via a gimple_stmt_iterator.
3956 Update call to vect_mark_pattern_stmts.
3957 (vect_pattern_recog): Update calls accordingly.
3959 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
3961 * tree-vectorizer.h (vect_get_vec_defs_for_stmt_copy)
3962 (vect_get_vec_def_for_stmt_copy): Take a vec_info rather than
3963 a vect_def_type for the first argument.
3964 * tree-vect-stmts.c (vect_get_vec_defs_for_stmt_copy): Likewise.
3965 (vect_get_vec_def_for_stmt_copy): Likewise. Return the original
3966 operand if it isn't defined by a vectorized statement.
3967 (vect_build_gather_load_calls): Remove the mask_dt argument and
3968 update calls to vect_get_vec_def_for_stmt_copy.
3969 (vectorizable_bswap): Likewise the dt argument.
3970 (vectorizable_call): Update calls to vectorizable_bswap and
3971 vect_get_vec_def_for_stmt_copy.
3972 (vectorizable_simd_clone_call, vectorizable_assignment)
3973 (vectorizable_shift, vectorizable_operation, vectorizable_condition)
3974 (vectorizable_comparison): Update calls to
3975 vect_get_vec_def_for_stmt_copy.
3976 (vectorizable_store): Likewise. Remove now-unnecessary calls to
3978 (vect_get_loop_based_defs): Remove dt argument and update call
3979 to vect_get_vec_def_for_stmt_copy.
3980 (vectorizable_conversion): Update calls to vect_get_loop_based_defs
3981 and vect_get_vec_def_for_stmt_copy.
3982 (vectorizable_load): Update calls to vect_build_gather_load_calls
3983 and vect_get_vec_def_for_stmt_copy.
3984 * tree-vect-loop.c (vect_create_epilog_for_reduction)
3985 (vectorizable_reduction, vectorizable_live_operation): Update calls
3986 to vect_get_vec_def_for_stmt_copy.
3988 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
3990 * tree-vect-data-refs.c (vect_record_base_alignment): Replace vec_info
3991 and gimple stmt arguments with a stmt_vec_info.
3992 (vect_record_base_alignments): Update calls accordingly.
3993 * tree-vect-slp.c (vect_record_max_nunits): Replace vec_info
3994 and gimple stmt arguments with a stmt_vec_info.
3995 (vect_build_slp_tree_1): Remove vinfo argument and update call
3996 to vect_record_max_nunits.
3997 (vect_build_slp_tree_2): Update calls to vect_build_slp_tree_1
3998 and vect_record_max_nunits.
4000 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4002 * tree-vectorizer.h (nested_in_vect_loop_p): Move further down
4003 file and take a stmt_vec_info instead of a gimple stmt.
4004 (supportable_widening_operation, vect_finish_replace_stmt)
4005 (vect_finish_stmt_generation, vect_get_store_rhs)
4006 (vect_get_vec_def_for_operand_1, vect_get_vec_def_for_operand)
4007 (vect_get_vec_defs, vect_init_vector, vect_transform_stmt)
4008 (vect_remove_stores, vect_analyze_stmt, vectorizable_condition)
4009 (vect_get_smallest_scalar_type, vect_check_gather_scatter)
4010 (vect_create_data_ref_ptr, bump_vector_ptr)
4011 (vect_permute_store_chain, vect_setup_realignment)
4012 (vect_transform_grouped_load, vect_record_grouped_load_vectors)
4013 (vect_create_addr_base_for_vector_ref, vectorizable_live_operation)
4014 (vectorizable_reduction, vectorizable_induction)
4015 (get_initial_def_for_reduction, is_simple_and_all_uses_invariant)
4016 (vect_get_place_in_interleaving_chain): Take stmt_vec_infos rather
4017 than gimple stmts as arguments.
4018 * tree-vect-data-refs.c (vect_get_smallest_scalar_type)
4019 (vect_preserves_scalar_order_p, vect_slp_analyze_node_dependences)
4020 (can_group_stmts_p, vect_check_gather_scatter)
4021 (vect_create_addr_base_for_vector_ref, vect_create_data_ref_ptr)
4022 (bump_vector_ptr, vect_permute_store_chain, vect_setup_realignment)
4023 (vect_permute_load_chain, vect_shift_permute_load_chain)
4024 (vect_transform_grouped_load)
4025 (vect_record_grouped_load_vectors): Likewise.
4026 * tree-vect-loop.c (vect_fixup_reduc_chain)
4027 (get_initial_def_for_reduction, vect_create_epilog_for_reduction)
4028 (vectorize_fold_left_reduction, is_nonwrapping_integer_induction)
4029 (vectorizable_reduction, vectorizable_induction)
4030 (vectorizable_live_operation, vect_loop_kill_debug_uses): Likewise.
4031 * tree-vect-patterns.c (type_conversion_p, adjust_bool_stmts)
4032 (vect_get_load_store_mask): Likewise.
4033 * tree-vect-slp.c (vect_get_place_in_interleaving_chain)
4034 (vect_analyze_slp_instance, vect_mask_constant_operand_p): Likewise.
4035 * tree-vect-stmts.c (vect_mark_relevant)
4036 (is_simple_and_all_uses_invariant)
4037 (exist_non_indexing_operands_for_use_p, process_use)
4038 (vect_init_vector_1, vect_init_vector, vect_get_vec_def_for_operand_1)
4039 (vect_get_vec_def_for_operand, vect_get_vec_defs)
4040 (vect_finish_stmt_generation_1, vect_finish_replace_stmt)
4041 (vect_finish_stmt_generation, vect_truncate_gather_scatter_offset)
4042 (compare_step_with_zero, vect_get_store_rhs, get_group_load_store_type)
4043 (get_negative_load_store_type, get_load_store_type)
4044 (vect_check_load_store_mask, vect_check_store_rhs)
4045 (vect_build_gather_load_calls, vect_get_strided_load_store_ops)
4046 (vectorizable_bswap, vectorizable_call, vectorizable_simd_clone_call)
4047 (vect_create_vectorized_demotion_stmts, vectorizable_conversion)
4048 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
4049 (get_group_alias_ptr_type, vectorizable_store, hoist_defs_of_uses)
4050 (vectorizable_load, vectorizable_condition, vectorizable_comparison)
4051 (vect_analyze_stmt, vect_transform_stmt, vect_remove_stores)
4052 (supportable_widening_operation): Likewise.
4054 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4056 * tree-vect-data-refs.c (vect_describe_gather_scatter_call): Take
4057 a stmt_vec_info instead of a gcall.
4058 (vect_check_gather_scatter): Update call accordingly.
4059 * tree-vect-loop-manip.c (iv_phi_p): Take a stmt_vec_info instead
4061 (vect_can_advance_ivs_p, vect_update_ivs_after_vectorizer)
4062 (slpeel_update_phi_nodes_for_loops):): Update calls accordingly.
4063 * tree-vect-loop.c (vect_transform_loop_stmt): Take a stmt_vec_info
4064 instead of a gimple stmt.
4065 (vect_transform_loop): Update calls accordingly.
4066 * tree-vect-slp.c (vect_split_slp_store_group): Take and return
4067 stmt_vec_infos instead of gimple stmts.
4068 (vect_analyze_slp_instance): Update use accordingly.
4069 * tree-vect-stmts.c (read_vector_array, write_vector_array)
4070 (vect_clobber_variable, vect_stmt_relevant_p, permute_vec_elements)
4071 (vect_use_strided_gather_scatters_p, vect_build_all_ones_mask)
4072 (vect_build_zero_merge_argument, vect_get_gather_scatter_ops)
4073 (vect_gen_widened_results_half, vect_get_loop_based_defs)
4074 (vect_create_vectorized_promotion_stmts, can_vectorize_live_stmts):
4075 Take a stmt_vec_info instead of a gimple stmt and pass stmt_vec_infos
4076 down to subroutines.
4078 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4080 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Change the type
4081 of the worklist from a vector of gimple stmts to a vector of
4083 * tree-vect-stmts.c (vect_mark_relevant, process_use)
4084 (vect_mark_stmts_to_be_vectorized): Likewise
4086 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4088 * tree-vect-loop.c (vect_analyze_loop_operations): Look up the
4089 statement before passing it to vect_analyze_stmt.
4090 (vect_create_epilog_for_reduction): Use a stmt_vec_info to walk
4091 the chain of phi vector definitions. Track the exit phi via its
4093 (vectorizable_reduction): Set cond_stmt_vinfo directly from the
4094 STMT_VINFO_REDUC_DEF.
4095 * tree-vect-slp.c (vect_get_place_in_interleaving_chain): Use
4096 stmt_vec_infos to handle the statement chains.
4097 (vect_get_slp_defs): Record the first statement in the node
4098 using a stmt_vec_info.
4099 * tree-vect-stmts.c (vect_mark_stmts_to_be_vectorized): Look up
4100 statements here and pass their stmt_vec_info down to subroutines.
4101 (vect_init_vector_1): Hoist call to vinfo_for_stmt and pass it
4102 down to vect_finish_stmt_generation.
4103 (vect_init_vector, vect_get_vec_defs, vect_finish_replace_stmt)
4104 (vect_finish_stmt_generation): Call vinfo_for_stmt and pass
4105 stmt_vec_infos to subroutines.
4106 (vect_remove_stores): Use stmt_vec_infos to handle the statement
4109 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4111 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences):
4112 (vect_check_gather_scatter, vect_create_data_ref_ptr, bump_vector_ptr)
4113 (vect_permute_store_chain, vect_setup_realignment)
4114 (vect_permute_load_chain, vect_shift_permute_load_chain)
4115 (vect_transform_grouped_load): Use stmt_vec_info rather than gimple
4116 stmts internally, and when passing values to other vectorizer routines.
4117 * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Likewise.
4118 * tree-vect-loop.c (vect_analyze_scalar_cycles_1)
4119 (vect_analyze_loop_operations, get_initial_def_for_reduction)
4120 (vect_create_epilog_for_reduction, vectorize_fold_left_reduction)
4121 (vectorizable_reduction, vectorizable_induction)
4122 (vectorizable_live_operation, vect_transform_loop_stmt)
4123 (vect_transform_loop): Likewise.
4124 * tree-vect-patterns.c (vect_reassociating_reduction_p)
4125 (vect_recog_widen_op_pattern, vect_recog_mixed_size_cond_pattern)
4126 (vect_recog_bool_pattern, vect_recog_gather_scatter_pattern): Likewise.
4127 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
4128 (vect_slp_analyze_node_operations_1): Likewise.
4129 * tree-vect-stmts.c (vect_mark_relevant, process_use)
4130 (exist_non_indexing_operands_for_use_p, vect_init_vector_1)
4131 (vect_mark_stmts_to_be_vectorized, vect_get_vec_def_for_operand)
4132 (vect_finish_stmt_generation_1, get_group_load_store_type)
4133 (get_load_store_type, vect_build_gather_load_calls)
4134 (vectorizable_bswap, vectorizable_call, vectorizable_simd_clone_call)
4135 (vect_create_vectorized_demotion_stmts, vectorizable_conversion)
4136 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
4137 (vectorizable_store, vectorizable_load, vectorizable_condition)
4138 (vectorizable_comparison, vect_analyze_stmt, vect_transform_stmt)
4139 (supportable_widening_operation): Likewise.
4140 (vect_get_vector_types_for_stmt): Likewise.
4141 * tree-vectorizer.h (vect_dr_behavior): Likewise.
4143 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4145 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence)
4146 (vect_slp_analyze_node_dependences, vect_analyze_data_ref_accesses)
4147 (vect_permute_store_chain, vect_permute_load_chain)
4148 (vect_shift_permute_load_chain, vect_transform_grouped_load): Avoid
4149 repeated stmt_vec_info lookups.
4150 * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Likewise.
4151 (vect_update_ivs_after_vectorizer): Likewise.
4152 * tree-vect-loop.c (vect_is_simple_reduction): Likewise.
4153 (vect_create_epilog_for_reduction, vectorizable_reduction): Likewise.
4154 * tree-vect-patterns.c (adjust_bool_stmts): Likewise.
4155 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
4156 (vect_bb_slp_scalar_cost): Likewise.
4157 * tree-vect-stmts.c (get_group_alias_ptr_type): Likewise.
4159 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4161 * tree-vect-data-refs.c (vect_check_gather_scatter): Pass the
4162 gcall rather than the generic gimple stmt to gimple_call_internal_fn.
4163 (vect_get_smallest_scalar_type, can_group_stmts_p): Use dyn_cast
4164 to get gassigns and gcalls, rather than operating on generc gimple
4166 * tree-vect-stmts.c (exist_non_indexing_operands_for_use_p)
4167 (vect_mark_stmts_to_be_vectorized, vectorizable_store)
4168 (vectorizable_load, vect_analyze_stmt): Likewise.
4169 * tree-vect-loop.c (vectorizable_reduction): Likewise gphi.
4171 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4173 * tree-vectorizer.h (get_earlier_stmt, get_later_stmt): Take and
4174 return stmt_vec_infos rather than gimple stmts. Do not accept
4176 (vect_find_last_scalar_stmt_in_slp): Return a stmt_vec_info instead
4178 * tree-vect-slp.c (vect_find_last_scalar_stmt_in_slp): Likewise.
4179 Update use of get_later_stmt.
4180 (vect_get_constant_vectors): Update call accordingly.
4181 (vect_schedule_slp_instance): Likewise
4182 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences): Likewise.
4183 (vect_slp_analyze_instance_dependence): Likewise.
4184 (vect_preserves_scalar_order_p): Update use of get_earlier_stmt.
4186 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4188 * tree-vectorizer.h (stmt_info_for_cost::stmt): Replace with...
4189 (stmt_info_for_cost::stmt_info): ...this new field.
4190 (add_stmt_costs): Update accordingly.
4191 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost)
4192 (vect_get_known_peeling_cost): Likewise.
4193 (vect_estimate_min_profitable_iters): Likewise.
4194 * tree-vect-stmts.c (record_stmt_cost): Likewise.
4196 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4198 * tree-vectorizer.h (_loop_vec_info::may_misalign_stmts): Change
4199 from an auto_vec<gimple *> to an auto_vec<stmt_vec_info>.
4200 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Update
4202 * tree-vect-loop-manip.c (vect_create_cond_for_align_checks): Likewise.
4204 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4206 * tree-vectorizer.h (_stmt_vec_info::same_dr_stmt): Change from
4207 a gimple stmt to a stmt_vec_info.
4208 * tree-vect-stmts.c (vectorizable_load): Update accordingly.
4210 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4212 * tree-vectorizer.h (vec_info::grouped_stores): Change from
4213 an auto_vec<gimple *> to an auto_vec<stmt_vec_info>.
4214 (_loop_vec_info::reduction_chains): Likewise.
4215 * tree-vect-loop.c (vect_fixup_scalar_cycles_with_patterns): Update
4217 * tree-vect-slp.c (vect_analyze_slp): Likewise.
4219 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4221 * tree-vectorizer.h (_stmt_vec_info::first_element): Change from
4222 a gimple stmt to a stmt_vec_info.
4223 (_stmt_vec_info::next_element): Likewise.
4224 * tree-vect-data-refs.c (vect_update_misalignment_for_peel)
4225 (vect_slp_analyze_and_verify_node_alignment)
4226 (vect_analyze_group_access_1, vect_analyze_group_access)
4227 (vect_small_gap_p, vect_prune_runtime_alias_test_list)
4228 (vect_create_data_ref_ptr, vect_record_grouped_load_vectors)
4229 (vect_supportable_dr_alignment): Update accordingly.
4230 * tree-vect-loop.c (vect_fixup_reduc_chain): Likewise.
4231 (vect_fixup_scalar_cycles_with_patterns, vect_is_slp_reduction)
4232 (vect_is_simple_reduction, vectorizable_reduction): Likewise.
4233 * tree-vect-patterns.c (vect_reassociating_reduction_p): Likewise.
4234 * tree-vect-slp.c (vect_build_slp_tree_1)
4235 (vect_attempt_slp_rearrange_stmts, vect_supported_load_permutation_p)
4236 (vect_split_slp_store_group, vect_analyze_slp_instance)
4237 (vect_analyze_slp, vect_transform_slp_perm_load): Likewise.
4238 * tree-vect-stmts.c (vect_model_store_cost, vect_model_load_cost)
4239 (get_group_load_store_type, get_load_store_type)
4240 (get_group_alias_ptr_type, vectorizable_store, vectorizable_load)
4241 (vect_transform_stmt, vect_remove_stores): Likewise.
4243 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4245 * tree-vectorizer.h (vect_dr_stmt): Return a stmt_vec_info rather
4247 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence)
4248 (vect_slp_analyze_data_ref_dependence, vect_record_base_alignments)
4249 (vect_calculate_target_alignmentm, vect_compute_data_ref_alignment)
4250 (vect_update_misalignment_for_peel, vect_verify_datarefs_alignment)
4251 (vector_alignment_reachable_p, vect_get_data_access_cost)
4252 (vect_get_peeling_costs_all_drs, vect_peeling_hash_get_lowest_cost)
4253 (vect_peeling_supportable, vect_enhance_data_refs_alignment)
4254 (vect_find_same_alignment_drs, vect_analyze_data_refs_alignment)
4255 (vect_analyze_group_access_1, vect_analyze_group_access)
4256 (vect_analyze_data_ref_access, vect_analyze_data_ref_accesses)
4257 (vect_vfa_access_size, vect_small_gap_p, vect_analyze_data_refs)
4258 (vect_supportable_dr_alignment): Remove vinfo_for_stmt from the
4259 result of vect_dr_stmt and use the stmt_vec_info instead of
4260 the associated gimple stmt.
4261 * tree-vect-loop-manip.c (get_misalign_in_elems): Likewise.
4262 (vect_gen_prolog_loop_niters): Likewise.
4263 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
4265 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4267 * tree-vectorizer.h (_slp_tree::stmts): Change from a vec<gimple *>
4268 to a vec<stmt_vec_info>.
4269 * tree-vect-slp.c (vect_free_slp_tree): Update accordingly.
4270 (vect_create_new_slp_node): Take a vec<gimple *> instead of a
4272 (_slp_oprnd_info::def_stmts): Change from a vec<gimple *>
4273 to a vec<stmt_vec_info>.
4274 (bst_traits::value_type, bst_traits::value_type): Likewise.
4275 (bst_traits::hash): Update accordingly.
4276 (vect_get_and_check_slp_defs): Change the stmts parameter from
4277 a vec<gimple *> to a vec<stmt_vec_info>.
4278 (vect_two_operations_perm_ok_p, vect_build_slp_tree_1): Likewise.
4279 (vect_build_slp_tree): Likewise.
4280 (vect_build_slp_tree_2): Likewise. Update uses of
4281 SLP_TREE_SCALAR_STMTS.
4282 (vect_print_slp_tree): Update uses of SLP_TREE_SCALAR_STMTS.
4283 (vect_mark_slp_stmts, vect_mark_slp_stmts_relevant)
4284 (vect_slp_rearrange_stmts, vect_attempt_slp_rearrange_stmts)
4285 (vect_supported_load_permutation_p, vect_find_last_scalar_stmt_in_slp)
4286 (vect_detect_hybrid_slp_stmts, vect_slp_analyze_node_operations_1)
4287 (vect_slp_analyze_node_operations, vect_slp_analyze_operations)
4288 (vect_bb_slp_scalar_cost, vect_slp_analyze_bb_1)
4289 (vect_get_constant_vectors, vect_get_slp_defs)
4290 (vect_transform_slp_perm_load, vect_schedule_slp_instance)
4291 (vect_remove_slp_scalar_calls, vect_schedule_slp): Likewise.
4292 (vect_analyze_slp_instance): Build up a vec of stmt_vec_infos
4293 instead of gimple stmts.
4294 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences): Change
4295 the stores parameter for a vec<gimple *> to a vec<stmt_vec_info>.
4296 (vect_slp_analyze_instance_dependence): Update uses of
4297 SLP_TREE_SCALAR_STMTS.
4298 (vect_slp_analyze_and_verify_node_alignment): Likewise.
4299 (vect_slp_analyze_and_verify_instance_alignment): Likewise.
4300 * tree-vect-loop.c (neutral_op_for_slp_reduction): Likewise.
4301 (get_initial_defs_for_reduction): Likewise.
4302 (vect_create_epilog_for_reduction): Likewise.
4303 (vectorize_fold_left_reduction): Likewise.
4304 * tree-vect-stmts.c (vect_prologue_cost_for_slp_op): Likewise.
4305 (vect_model_simple_cost, vectorizable_shift, vectorizable_load)
4306 (can_vectorize_live_stmts): Likewise.
4308 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4310 * tree-vectorizer.h (_loop_vec_info::reductions): Change from an
4311 auto_vec<gimple *> to an auto_vec<stmt_vec_info>.
4312 (vect_force_simple_reduction): Take and return stmt_vec_infos rather
4314 * tree-parloops.c (valid_reduction_p): Take a stmt_vec_info instead
4316 (gather_scalar_reductions): Update after above interface changes.
4317 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Likewise.
4318 (vect_is_simple_reduction): Take and return stmt_vec_infos rather
4320 (vect_force_simple_reduction): Likewise.
4321 * tree-vect-patterns.c (vect_pattern_recog_1): Update use of
4322 LOOP_VINFO_REDUCTIONS.
4323 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
4325 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4327 * tree-vectorizer.h (_stmt_vec_info::reduc_def): Change from
4328 a gimple stmt to a stmt_vec_info.
4329 * tree-vect-loop.c (vect_active_double_reduction_p)
4330 (vect_force_simple_reduction, vectorizable_reduction): Update
4333 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4335 * tree-vectorizer.h (_slp_tree::vec_stmts): Change from a
4336 vec<gimple *> to a vec<stmt_vec_info>.
4337 * tree-vect-loop.c (vect_create_epilog_for_reduction): Change
4338 the reduction_phis argument from a vec<gimple *> to a
4340 (vectorizable_reduction): Likewise the phis local variable that
4341 is passed to vect_create_epilog_for_reduction. Update for new type
4342 of SLP_TREE_VEC_STMTS.
4343 (vectorizable_induction): Update for new type of SLP_TREE_VEC_STMTS.
4344 (vectorizable_live_operation): Likewise.
4345 * tree-vect-slp.c (vect_get_slp_vect_defs): Likewise.
4346 (vect_transform_slp_perm_load, vect_schedule_slp_instance): Likewise.
4348 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4350 * tree-vectorizer.h (_stmt_vec_info::vectorized_stmt): Change from
4351 a gimple stmt to a stmt_vec_info.
4352 (vectorizable_condition, vectorizable_live_operation)
4353 (vectorizable_reduction, vectorizable_induction): Pass back the
4354 vectorized statement as a stmt_vec_info.
4355 * tree-vect-data-refs.c (vect_record_grouped_load_vectors): Update
4356 use of STMT_VINFO_VEC_STMT.
4357 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise,
4358 accumulating the inner phis that feed the STMT_VINFO_VEC_STMT
4359 as stmt_vec_infos rather than gimple stmts.
4360 (vectorize_fold_left_reduction): Change vec_stmt from a gimple stmt
4362 (vectorizable_live_operation): Likewise.
4363 (vectorizable_reduction, vectorizable_induction): Likewise,
4364 updating use of STMT_VINFO_VEC_STMT.
4365 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Update use
4366 of STMT_VINFO_VEC_STMT.
4367 (vect_build_gather_load_calls, vectorizable_bswap, vectorizable_call)
4368 (vectorizable_simd_clone_call, vectorizable_conversion)
4369 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
4370 (vectorizable_store, vectorizable_load, vectorizable_condition)
4371 (vectorizable_comparison, can_vectorize_live_stmts): Change vec_stmt
4372 from a gimple stmt to a stmt_vec_info.
4373 (vect_transform_stmt): Update use of STMT_VINFO_VEC_STMT. Pass a
4374 pointer to a stmt_vec_info to the vectorizable_* routines.
4376 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4378 * tree-vectorizer.h (_stmt_vec_info::related_stmt): Change from
4379 a gimple stmt to a stmt_vec_info.
4380 (is_pattern_stmt_p): Update accordingly.
4381 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Likewise.
4382 (vect_record_grouped_load_vectors): Likewise.
4383 * tree-vect-loop.c (vect_determine_vf_for_stmt): Likewise.
4384 (vect_fixup_reduc_chain, vect_update_vf_for_slp): Likewise.
4385 (vect_model_reduction_cost): Likewise.
4386 (vect_create_epilog_for_reduction): Likewise.
4387 (vectorizable_reduction, vectorizable_induction): Likewise.
4388 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
4389 Return the stmt_vec_info for the pattern statement.
4390 (vect_set_pattern_stmt): Update use of STMT_VINFO_RELATED_STMT.
4391 (vect_split_statement, vect_mark_pattern_stmts): Likewise.
4392 * tree-vect-slp.c (vect_detect_hybrid_slp_stmts): Likewise.
4393 (vect_detect_hybrid_slp, vect_get_slp_defs): Likewise.
4394 * tree-vect-stmts.c (vect_mark_relevant): Likewise.
4395 (vect_get_vec_def_for_operand_1, vectorizable_call): Likewise.
4396 (vectorizable_simd_clone_call, vect_analyze_stmt, new_stmt_vec_info)
4397 (free_stmt_vec_info, vect_is_simple_use): Likewise.
4399 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4401 * tree-vectorizer.h (vect_finish_replace_stmt): Return a stmt_vec_info
4402 (vect_finish_stmt_generation): Likewise.
4403 * tree-vect-stmts.c (vect_finish_stmt_generation_1): Likewise.
4404 (vect_finish_replace_stmt, vect_finish_stmt_generation): Likewise.
4405 (vect_build_gather_load_calls): Use the return value of the above
4406 functions instead of a separate call to vinfo_for_stmt. Use narrow
4407 scopes for the input gimple stmt and wider scopes for the associated
4408 stmt_vec_info. Use vec_info::lookup_def when setting these
4409 stmt_vec_infos from an SSA_NAME definition.
4410 (vectorizable_bswap, vectorizable_call, vectorizable_simd_clone_call)
4411 (vect_create_vectorized_demotion_stmts, vectorizable_conversion)
4412 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
4413 (vectorizable_store, vectorizable_load, vectorizable_condition)
4414 (vectorizable_comparison): Likewise.
4415 * tree-vect-loop.c (vectorize_fold_left_reduction): Likewise.
4416 (vectorizable_reduction): Likewise.
4418 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4420 * tree-vectorizer.h (vect_is_simple_use): Add an optional
4421 stmt_vec_info * parameter before the optional gimple **.
4422 * tree-vect-stmts.c (vect_is_simple_use): Likewise.
4423 (process_use, vect_get_vec_def_for_operand_1): Update callers.
4424 (vect_get_vec_def_for_operand, vectorizable_shift): Likewise.
4425 * tree-vect-loop.c (vectorizable_reduction): Likewise.
4426 (vectorizable_live_operation): Likewise.
4427 * tree-vect-patterns.c (type_conversion_p): Likewise.
4428 (vect_look_through_possible_promotion): Likewise.
4429 (vect_recog_rotate_pattern): Likewise.
4430 * tree-vect-slp.c (vect_get_and_check_slp_defs): Likewise.
4432 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4434 * tree-vectorizer.h (stmt_vec_info): Temporarily change from
4435 a typedef to a wrapper class.
4436 (NULL_STMT_VEC_INFO): New macro.
4437 (vec_info::stmt_infos): Change to vec<stmt_vec_info>.
4438 (stmt_vec_info::operator*): New function.
4439 (stmt_vec_info::operator gimple *): Likewise.
4440 (set_vinfo_for_stmt): Use NULL_STMT_VEC_INFO.
4441 (add_stmt_costs): Likewise.
4442 * tree-vect-loop-manip.c (iv_phi_p): Likewise.
4443 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost)
4444 (vect_get_known_peeling_cost): Likewise.
4445 (vect_estimate_min_profitable_iters): Likewise.
4446 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
4447 * tree-vect-slp.c (vect_remove_slp_scalar_calls): Likewise.
4448 * tree-vect-stmts.c (vect_build_gather_load_calls): Likewise.
4449 (vectorizable_store, free_stmt_vec_infos): Likewise.
4450 (new_stmt_vec_info): Change return type of xcalloc to
4453 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4455 * tree-vectorizer.h (vec_info::lookup_single_use): Declare.
4456 * tree-vectorizer.c (vec_info::lookup_single_use): New function.
4457 * tree-vect-loop.c (vectorizable_reduction): Use it instead of
4458 a single_imm_use-based sequence.
4459 * tree-vect-stmts.c (supportable_widening_operation): Likewise.
4461 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4463 * tree-vectorizer.h (vec_info::lookup_def): Declare.
4464 * tree-vectorizer.c (vec_info::lookup_def): New function.
4465 * tree-vect-patterns.c (vect_get_internal_def): Use it.
4466 (vect_widened_op_tree): Likewise.
4467 * tree-vect-stmts.c (vect_is_simple_use): Likewise.
4468 * tree-vect-loop.c (vect_analyze_loop_operations): Likewise.
4469 (vectorizable_reduction): Likewise.
4470 (vect_valid_reduction_input_p): Take a stmt_vec_info instead
4472 (vect_is_slp_reduction): Update calls accordingly. Use
4473 vec_info::lookup_def.
4474 (vect_is_simple_reduction): Likewise
4475 * tree-vect-slp.c (vect_detect_hybrid_slp_1): Use vec_info::lookup_def.
4477 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4479 * tree-vectorizer.h (vec_info::lookup_stmt): Declare.
4480 * tree-vectorizer.c (vec_info::lookup_stmt): New function.
4481 * tree-vect-loop.c (vect_determine_vf_for_stmt): Use it instead
4483 (vect_determine_vectorization_factor, vect_analyze_scalar_cycles_1)
4484 (vect_compute_single_scalar_iteration_cost, vect_analyze_loop_form)
4485 (vect_update_vf_for_slp, vect_analyze_loop_operations)
4486 (vect_is_slp_reduction, vectorizable_induction)
4487 (vect_transform_loop_stmt, vect_transform_loop): Likewise.
4488 * tree-vect-patterns.c (vect_init_pattern_stmt):
4489 (vect_determine_min_output_precision_1, vect_determine_precisions)
4490 (vect_pattern_recog): Likewise.
4491 * tree-vect-stmts.c (vect_analyze_stmt, vect_transform_stmt): Likewise.
4492 * config/powerpcspe/powerpcspe.c (rs6000_density_test): Likewise.
4493 * config/rs6000/rs6000.c (rs6000_density_test): Likewise.
4494 * tree-vect-slp.c (vect_detect_hybrid_slp_stmts): Likewise.
4495 (vect_detect_hybrid_slp_1, vect_detect_hybrid_slp_2)
4496 (vect_detect_hybrid_slp): Likewise. Change the walk_stmt_info
4497 info field from a loop to a loop_vec_info.
4499 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4501 * tree-vectorizer.h (stmt_vec_info): Move typedef earlier in file.
4502 (vec_info::add_stmt): Declare.
4503 * tree-vectorizer.c (vec_info::add_stmt): New function.
4504 * tree-vect-data-refs.c (vect_create_data_ref_ptr): Use it.
4505 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Likewise.
4506 (vect_create_epilog_for_reduction, vectorizable_reduction): Likewise.
4507 (vectorizable_induction): Likewise.
4508 * tree-vect-slp.c (_bb_vec_info::_bb_vec_info): Likewise.
4509 * tree-vect-stmts.c (vect_finish_stmt_generation_1): Likewise.
4510 (vectorizable_simd_clone_call, vectorizable_store): Likewise.
4511 (vectorizable_load): Likewise.
4512 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
4513 (vect_recog_bool_pattern, vect_recog_mask_conversion_pattern)
4514 (vect_recog_gather_scatter_pattern): Likewise.
4515 (append_pattern_def_seq): Likewise. Remove a check that is
4516 performed by add_stmt itself.
4518 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4520 * tree-vect-loop.c (vectorizable_reduction): Fix an instance in
4521 which make_ssa_name was called with new_stmt before new_stmt
4524 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4526 * tree-vect-loop.c (vect_valid_reduction_input_p): New function,
4528 (vect_is_slp_reduction): ...here...
4529 (vect_is_simple_reduction): ...and here. Remove repetition of tests
4530 that are already known to be false.
4532 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4534 * tree-vectorizer.h (vect_free_slp_instance): Add a final_p parameter.
4535 * tree-vect-slp.c (vect_free_slp_tree): Likewise. Don't update
4536 STMT_VINFO_NUM_SLP_USES when it's true.
4537 (vect_free_slp_instance): Add a final_p parameter and pass it to
4539 (vect_build_slp_tree_2): Update call to vect_free_slp_instance.
4540 (vect_analyze_slp_instance): Likewise.
4541 (vect_slp_analyze_operations): Likewise.
4542 (vect_slp_analyze_bb_1): Likewise.
4543 * tree-vectorizer.c (vec_info): Likewise.
4544 * tree-vect-loop.c (vect_transform_loop): Likewise.
4546 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4548 * tree-vect-loop.c (vectorizable_reduction): Assert that the
4549 function is not called for second and subsequent members of
4552 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4554 * tree-vect-loop.c (get_initial_def_for_reduction): Move special
4555 cases for nested loops from here to ...
4556 (vect_create_epilog_for_reduction): ...here. Only call
4557 vect_is_simple_use for inner-loop reductions.
4559 2018-07-31 Martin Liska <mliska@suse.cz>
4561 PR gcov-profile/85338
4562 PR gcov-profile/85350
4563 PR gcov-profile/85372
4564 * profile.c (struct location_triplet): New.
4565 (struct location_triplet_hash): Likewise.
4566 (output_location): Do not output a BB that
4567 is already recorded for a line.
4568 (branch_prob): Use streamed_locations.
4570 2018-07-31 Martin Liska <mliska@suse.cz>
4572 PR gcov-profile/85370
4573 * coverage.c (coverage_begin_function): Do not mark target
4574 clones as artificial functions.
4576 2018-07-31 Martin Liska <mliska@suse.cz>
4578 PR gcov-profile/83813
4579 PR gcov-profile/84758
4580 PR gcov-profile/85217
4581 PR gcov-profile/85332
4582 * profile.c (branch_prob): Do not record GOTO expressions
4583 for GIMPLE statements which locations are already streamed.
4585 2018-07-31 Olivier Hainque <hainque@adacore.com>
4587 * gcc.c (handle_spec_function): Accept a soft_matched_part
4588 argument, as do_spec_1. Pass it down to ...
4589 (eval_spec_function): Accept a soft_matched_part argument,
4590 and pass it down to ...
4591 (do_spec_2): Accept a soft_matched_part argument, and pass
4592 it down to do_spec_1.
4593 (do_spec_1): Pass soft_matched_part to handle_spec_function.
4594 (handle_braces): Update call to handle_spec_function.
4595 (driver::set_up_specs): Update calls to do_spec_2.
4596 (compare_debug_dump_opt_spec_function): Likewise.
4597 (compare_debug_self_opt_spec_function): Likewise.
4599 2018-07-31 Olivier Hainque <hainque@adacore.com>
4601 * common.opt (nolibc): New option.
4602 * doc/invoke.texi (Link Options): Document it.
4603 * gcc.c (LINK_GCC_C_SEQUENCE_SPEC): Honor nolibc.
4604 * config/alpha/linux.h: Likewise.
4605 * config/arc/elf.h: Likewise.
4606 * config/arm/uclinux-elf.h: Likewise.
4607 * config/arm/unknown-elf.h: Likewise.
4608 * config/avr/avrlibc.h: Likewise.
4609 * config/bfin/bfin.h: Likewise.
4610 * config/bfin/linux.h: Likewise.
4611 * config/bfin/uclinux.h: Likewise.
4612 * config/darwin.h: Likewise.
4613 * config/darwin10.h: Likewise.
4614 * config/darwin12.h: Likewise.
4615 * config/gnu-user.h: Likewise.
4616 * config/lm32/uclinux-elf.h: Likewise.
4617 * config/pa/pa-hpux11.h: Likewise.
4618 * config/pa/pa64-hpux.h: Likewise.
4619 * config/sparc/sparc.h: Likewise.
4621 2018-07-31 Olivier Hainque <hainque@adacore.com>
4623 * gcc.c (getenv_spec_function): Prepend '/' to value for allowed
4624 undefined variables.
4626 2018-07-30 Segher Boessenkool <segher@kernel.crashing.org>
4629 * config/arm/arm.c (arm_block_set_aligned_vect): Use gen_int_mode
4632 2018-07-30 Bernd Edlinger <bernd.edlinger@hotmail.de>
4634 * tree-ssa-forwprop.c (simplify_builtin_call): Don't create a not NUL
4635 terminated string literal.
4637 2018-07-30 Segher Boessenkool <segher@kernel.crashing.org>
4639 PR rtl-optimization/85160
4640 * combine.c (is_just_move): New function.
4641 (try_combine): Allow combining two instructions into two if neither of
4642 the original instructions was a move.
4644 2018-07-30 Alexander Monakov <amonakov@ispras.ru>
4647 * doc/extend.texi (Global Register Variables): Discourage use of type
4649 (Local Register Variables): Likewise.
4651 2018-07-30 Richard Sandiford <richard.sandiford@arm.com>
4653 PR tree-optimization/86506
4654 * hwint.h (ceil_log2): Resync with hwint.c implementation.
4656 2018-07-30 Ilya Leoshkevich <iii@linux.ibm.com>
4659 * lra-constraints.c (spill_hard_reg_in_range): When selecting the
4660 hard_regno, make sure no insn between `from` and `to` clobbers it.
4662 2018-07-30 Cesar Philippidis <cesar@codesourcery.com>
4663 Tom de Vries <tdevries@suse.de>
4665 * config/nvptx/nvptx.c (PTX_GANG_DEFAULT): Rename to ...
4666 (PTX_DEFAULT_RUNTIME_DIM): ... this.
4667 (nvptx_goacc_validate_dims): Set default worker and gang dims to
4668 PTX_DEFAULT_RUNTIME_DIM.
4669 (nvptx_dim_limit): Ignore GOMP_DIM_WORKER.
4671 2018-07-29 John David Anglin <danglin@gcc.gnu.org>
4673 * config/pa/pa.c (pa_output_addr_vec): Align address table.
4674 * config/pa/pa.h (JUMP_TABLES_IN_TEXT_SECTION): Revise comment.
4675 * config/pa/pa32-linux.h (JUMP_TABLES_IN_TEXT_SECTION): Define.
4677 2018-07-27 Michael Meissner <meissner@linux.ibm.com>
4679 * config/rs6000/constraints.md (wG constraint): Delete, no longer
4681 * config/rs6000/predicates.md (p9_fusion_reg_operand): Rename
4682 predicate to reflect toc fusion has been deleted.
4683 (toc_fusion_mem_raw): Delete, no longer used.
4684 (toc_fusion_mem_wrapped): Likewise.
4685 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete toc
4687 * config/rs6000/rs6000-protos.h (fusion_wrap_memory_address):
4688 Delete, no longer used.
4689 * config/rs6000/rs6000.c (struct rs6000_reg_addr): Delete fields
4690 meant to be used for toc fusion.
4691 (rs6000_debug_print_mode): Delete toc fusion debugging.
4692 (rs6000_debug_reg_global): Likewise.
4693 (rs6000_init_hard_regno_mode_ok): Delete setting up fields for toc
4694 fusion and secondary reload support that were never used.
4695 (rs6000_option_override_internal): Delete TOC fusion, that was only
4696 partially defined, and it did not work unless you also used the
4698 (rs6000_legitimate_address_p): Delete TOC fusion support.
4699 (rs6000_opt_masks): Likewise.
4700 (fusion_wrap_memory_address): Delete function, no longer used.
4701 (fusion_split_address); Delete TOC fusion support.
4702 * config/rs6000/rs6000.h (TARGET_TOC_FUSION_INT): Delete, no
4703 longer used with toc fusion being deleted.
4704 (TARGET_TOC_FUSION_FP): Likewise.
4705 * config/rs6000/rs6000.md (UNSPEC_FUSION_ADDIS): Delete TOC fusion
4707 (toc fusion spliter): Delete TOC fusion support.
4708 (toc_fusionload_<mode>): Likewise.
4709 (toc_fusionload_di): Likewise.
4710 (fusion_gpr_load_<mode>): Delete generator function, this insn no
4711 longer needs to be named. Rename predicate to delete TOC fusion.
4712 (fusion_gpr_<P:mode>_<GPR_FUSION:mode>_load): Likewise.
4713 (fusion_gpr_<P:mode>_<GPR_FUSION:mode>_store): Likewise.
4714 (fusion_vsx_<P:mode>_<GPR_FUSION:mode>_load): Likewise.
4715 (fusion_vsx_<P:mode>_<GPR_FUSION:mode>_store): Likewise.
4716 (p9 fusion peephole2s): Rename predicate to delete TOC fusion.
4718 2018-07-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
4720 * doc/extend.texi (Basic PowerPC Built-in Functions Available on
4721 ISA 2.05): Replace __uint128_t with __uint128 and __int128_t with
4722 __int128 in built-in function prototypes.
4723 (PowerPC AltiVec Built-in Functions on ISA 2.07): Likewise.
4724 (PowerPC AltiVec Built-in Functions on ISA 3.0): Likewise.
4726 2018-07-27 Martin Sebor <msebor@redhat.com>
4728 PR tree-optimization/86696
4729 * tree-ssa-strlen.c (get_min_string_length): Handle all integer
4730 types, including enums.
4731 (handle_char_store): Be prepared for the above function to fail.
4733 2018-07-26 Qing Zhao <qing.zhao@oracle.com>
4735 * builtins.c (inline_expand_builtin_string_cmp): Disable inlining
4736 when optimization level is lower than 2 or optimize for size.
4738 2018-07-26 Martin Sebor <msebor@redhat.com>
4740 PR tree-optimization/86043
4741 PR tree-optimization/86042
4742 * tree-ssa-strlen.c (handle_builtin_memcpy): Handle strict overlaps.
4743 (get_string_cst_length): Rename...
4744 (get_min_string_length): ...to this. Add argument.
4745 (handle_char_store): Extend to handle multi-character stores by
4747 * tree.c (initializer_zerop): Use new argument. Handle MEM_REF.
4748 * tree.h (initializer_zerop): Add argument.
4750 2018-07-26 Jakub Jelinek <jakub@redhat.com>
4753 * omp-low.c (scan_sharing_clauses): Don't ignore map clauses for
4754 declare target to variables if they have always,{to,from,tofrom} map
4757 2018-07-26 Martin Liska <mliska@suse.cz>
4760 * lto-wrapper.c: Add linker_output as prefix
4761 for ltrans_output_file.
4763 2018-07-26 Segher Boessenkool <segher@kernel.crashing.org>
4765 PR rtl-optimization/85805
4766 * combine.c (reg_nonzero_bits_for_combine): Only use the last set
4767 value for hard registers if that was written in the same mode.
4769 2018-07-26 Martin Liska <mliska@suse.cz>
4771 PR gcov-profile/86536
4772 * gcov.c (format_gcov): Use printf format %.*f directly
4773 and do not handle special values.
4775 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
4777 * common/config/arc/arc-common.c (arc_option_optimization_table):
4778 Update default optimizations for size.
4780 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
4782 * config/arc/arc.md (movsf_insn): Add short instruction selection.
4783 * config/arc/constraints.md (CfZ): New constraint.
4784 * config/arc/fpu.md (addssf3_fpu): Use CfZ constraint.
4785 (subsf3_fpu): Likewise.
4786 (cmpsf_fpu): Likewise.
4787 (cmpsf_fpu_uneq): Likewise.
4789 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
4791 * config/arc/arc.c (compact_memory_operand_p): Check for uncached
4793 (arc_is_uncached_mem_p): uncached applies to both the variable and
4796 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
4798 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Add additional
4801 2018-07-25 David Malcolm <dmalcolm@redhat.com>
4803 * optinfo-emit-json.cc (class optrecord_json_writer): Convert
4804 field "m_scopes" from vec to auto_vec.
4806 2018-07-25 Martin Liska <mliska@suse.cz>
4808 * config/powerpcspe/powerpcspe-protos.h (rs6000_loop_align): Fix
4811 2018-07-25 Richard Biener <rguenther@suse.de>
4814 * dwarf2out.c (dwarf2out_decl): Do not handle nested functions
4815 special wrt context_die late.
4816 (gen_subprogram_die): Re-use DIEs in local scope.
4818 2018-07-25 Richard Sandiford <richard.sandiford@arm.com>
4820 PR tree-optimization/86644
4821 * hwint.c (ceil_log2): Fix comment. Return 0 for 0.
4823 2018-07-25 Martin Liska <mliska@suse.cz>
4826 * dumpfile.c: And excluded values with TDF_ALL_VALUES.
4827 * dumpfile.h (enum dump_flag): Defince TDF_ALL_VALUES.
4829 2018-07-25 Martin Liska <mliska@suse.cz>
4832 * params.def: Explain ASan abbreviation and provide
4833 a documentation link.
4835 2018-07-24 Martin Sebor <msebor@redhat.com>
4837 PR tree-optimization/86622
4838 PR tree-optimization/86532
4839 * builtins.h (string_length): Declare.
4840 * builtins.c (c_strlen): Correct handling of non-constant offsets.
4841 (check_access): Be prepared for non-constant length ranges.
4842 (string_length): Make extern.
4843 * expr.c (string_constant): Only handle the minor non-constant
4844 array index. Use string_constant to compute the length of
4845 a generic string constant.
4847 2018-07-24 Richard Sandiford <richard.sandiford@arm.com>
4849 PR tree-optimization/86618
4850 * tree-vect-stmts.c (vectorizable_call): Don't take the address
4851 of LOOP_VINFO_MASKS (loop_vinfo) when loop_vinfo is null.
4853 2018-07-24 David Malcolm <dmalcolm@redhat.com>
4855 PR tree-optimization/86636
4856 * json.cc (json::object::set): Fix comment. Add assertions.
4857 (json::array::append): Move here from json.h. Add comment and an
4859 (json::string::string): Likewise.
4860 * json.h (json::array::append): Move to json.cc.
4861 (json::string::string): Likewise.
4862 * optinfo-emit-json.cc
4863 (optrecord_json_writer::impl_location_to_json): Assert that we
4864 aren't attempting to write out UNKNOWN_LOCATION, or an ad-hoc
4865 wrapper around it. Expand the location once, rather than three
4867 (optrecord_json_writer::inlining_chain_to_json): Fix the check for
4868 UNKNOWN_LOCATION, to use LOCATION_LOCUS to look through ad-hoc
4870 (optrecord_json_writer::optinfo_to_json): Likewise, in four
4871 places. Fix some overlong lines.
4873 2018-07-24 Matthew Malcomson <matthew.malcomson@arm.com>
4875 * config/aarch64/aarch64-simd.md
4876 (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w<mode>): Split into...
4877 (aarch64_<ANY_EXTEND:su>subw<mode>): ... This...
4878 (aarch64_<ANY_EXTEND:su>addw<mode>): ... And this.
4879 (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w<mode>_internal): Split into...
4880 (aarch64_<ANY_EXTEND:su>subw<mode>_internal): ... This...
4881 (aarch64_<ANY_EXTEND:su>addw<mode>_internal): ... And this.
4882 (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w2<mode>_internal): Split into...
4883 (aarch64_<ANY_EXTEND:su>subw2<mode>_internal): ... This...
4884 (aarch64_<ANY_EXTEND:su>addw2<mode>_internal): ... And this.
4886 2018-07-24 Jakub Jelinek <jakub@redhat.com>
4889 * expmed.c (expand_divmod): Punt if d == HOST_WIDE_INT_MIN
4890 and size > HOST_BITS_PER_WIDE_INT. For size > HOST_BITS_PER_WIDE_INT
4891 and abs_d == d, do the power of two handling if profitable.
4893 2018-07-24 Richard Biener <rguenther@suse.de>
4895 * match.pd: Add BIT_FIELD_REF canonicalizations.
4897 2018-07-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
4900 * genmatch.c (dt_operand::gen_match_op): Avoid folding volatile values.
4902 2018-07-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
4904 * gimple-fold.c (gimple_fold_builtin_printf): Don't create a not NUL
4905 terminated STRING_CST object.
4907 2018-07-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
4909 hsa-dump.c (dump_hsa_symbol): Avoid out of scope access to buf.
4911 2018-07-23 Segher Boessenkool <segher@kernel.crashing.org>
4913 * config/rs6000/rs6000-p8swap.c (rtx_is_swappable_p): Adjust.
4914 * config/rs6000/rs6000-protos.h (rs6000_split_v4si_init): Delete.
4915 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Always force
4916 the elements into a register.
4917 (rs6000_split_v4si_init_di_reg): Delete.
4918 (rs6000_split_v4si_init): Delete.
4919 * config/rs6000/vsx.md (unspec): Delete UNSPEC_VSX_VEC_INIT.
4920 (vsx_init_v4si): Rewrite as a define_expand.
4922 2018-07-23 Segher Boessenkool <segher@kernel.crashing.org>
4924 * config/rs6000/rs6000.md (splitters for rldimi and rlwimi with the
4925 zero_extend argument from memory): New.
4927 2018-07-22 Martin Sebor <msebor@redhat.com>
4930 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Avoid
4931 diagnosing calls with unknown arguments unless -Walloca-larger-than
4932 is restricted to less than PTRDIFF_MAX bytes.
4934 2018-07-22 Gerald Pfeifer <gerald@pfeifer.com>
4936 * doc/gcov.texi (Invoking Gcov): Editorial changes.
4938 2018-07-20 David Malcolm <dmalcolm@redhat.com>
4940 * pretty-print.c (text_info::set_location): Remove redundant
4941 "line_table" parameter from call to rich_location::set_range.
4943 2018-07-20 Martin Sebor <msebor@redhat.com>
4946 * builtins.c (expand_builtin_alloca): Adjust.
4947 * calls.c (alloc_max_size): Simplify.
4948 * cgraphunit.c (cgraph_node::expand): Adjust.
4949 * common.opt (larger_than_size, warn_frame_larger_than): Remove
4951 (frame_larger_than_size): Same.
4952 (-Wframe-larger-than, -Wlarger-than, -Wstack-usage): Change options
4953 to take a HOST_WIDE_INT argument and accept a byte-size suffix.
4955 * doc/invoke.texi (GCC Command Options): Document option arguments.
4956 Explain byte-size arguments and suffixes.
4957 (-Wvla-larger-than, -Wno-alloc-size-larger-than): Update.
4958 (-Wno-alloca-larger-than, -Wno-vla-larger-than): Same.
4959 (-Wframe-larger-than, -Wlarger-than, -Wstack-usage): Same.
4960 * doc/options.texi (UInteger): Expand.
4961 (Host_Wide_Int, ByteSize): Document new properties.
4962 * final.c (final_start_function_1): Include sizes in an error message.
4963 * function.c (frame_offset_overflow): Same.
4964 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Adjust.
4965 (alloca_call_type_by_arg): Change function argument to HOST_WIDE_INT.
4966 Diagnose unbounded alloca calls only for limits of less than
4968 (alloca_call_type): Adjust. Diagnose possibly out-of-bounds alloca
4969 calls and VLA size only for limits of less than PTRDIFF_MAX. Same
4971 (pass_walloca::execute): Adjust. Diagnose alloca calls in loops
4972 only for limits of less than PTRDIFF_MAX.
4973 * langhooks-def.h (lhd_handle_option): Change function argument
4975 * langhooks.c (lhd_handle_option): Same.
4976 * langhooks.h (handle_option): Same.
4977 * opt-functions.awk (switch_bit_fields): Handle Host_Wide_Int and
4979 (var_type, var_type_struct): Same.
4980 (var_set): Handle ByteSize flag.
4981 * optc-gen.awk: Add comments to output to ease debugging. Make
4982 use of HOST_WIDE_INT where appropriate.
4983 * opts-gen-save.awk: Use %lx to format unsigned long.
4984 * opth-gen.awk: Change function argument to HOST_WIDE_INT.
4985 * opts-common.c (integral_argument): Return HOST_WIDE_INT and add
4986 arguments. Parse bytes-size suffixes.
4987 (enum_arg_to_value): Change function argument to HOST_WIDE_INT.
4988 (enum_value_to_arg): Same.
4989 (decode_cmdline_option): Handle cl_host_wide_int. Adjust.
4990 (handle_option): Adjust.
4991 (generate_option): Change function argument to HOST_WIDE_INT.
4992 (cmdline_handle_error): Adjust.
4993 (read_cmdline_option): Change function argument to HOST_WIDE_INT.
4994 (set_option): Change function argument to HOST_WIDE_INT.
4995 (option_enabled): Handle cl_host_wide_int.
4996 (get_option_state): Handle CLVC_SIZE.
4997 (control_warning_option): Same.
4998 * opts.c (common_handle_option): Change function argument to
4999 HOST_WIDE_INT. Remove handling of OPT_Walloca_larger_than_ and
5000 OPT_Wvla_larger_than_.
5001 * opts.h (enum cl_var_type): Add an enumerator.
5002 * stor-layout.c (layout_decl): Print a more meaningful warning.
5003 * toplev.c (output_stack_usage): Adjust.
5005 2018-07-20 Qing Zhao <qing.zhao@oracle.com>
5007 * builtins.c (expand_builtin_memcmp): Delete the last parameter for
5008 call to inline_expand_builtin_string_cmp.
5009 (expand_builtin_strcmp): Likewise.
5010 (expand_builtin_strncmp): Likewise.
5011 (inline_string_cmp): Delete the last parameter, change char_type_node
5012 to unsigned_char_type_node for strcmp/strncmp, add conversions to the
5014 (inline_expand_builtin_string_cmp): Delete the last parameter, give up
5015 the inlining expansion on target where the type of the call has same or
5016 narrower precision than unsigned char.
5018 2018-07-20 David Malcolm <dmalcolm@redhat.com>
5020 * Makefile.in (OBJS): Add json.o and optinfo-emit-json.o.
5021 (CFLAGS-optinfo-emit-json.o): Define TARGET_NAME.
5022 * common.opt (fsave-optimization-record): New option.
5023 * coretypes.h (struct kv_pair): Move here from dumpfile.c.
5024 * doc/invoke.texi (-fsave-optimization-record): New option.
5025 * dumpfile.c: Include "optinfo-emit-json.h".
5026 (struct kv_pair): Move to coretypes.h.
5027 (optgroup_options): Make non-static.
5028 (dump_context::end_scope): Call
5029 optimization_records_maybe_pop_dump_scope.
5030 * dumpfile.h (optgroup_options): New decl.
5031 * json.cc: New file.
5033 * optinfo-emit-json.cc: New file.
5034 * optinfo-emit-json.h: New file.
5035 * optinfo.cc: Include "optinfo-emit-json.h".
5036 (optinfo::emit): Call optimization_records_maybe_record_optinfo.
5037 (optinfo_enabled_p): Check optimization_records_enabled_p.
5038 (optinfo_wants_inlining_info_p): Likewise.
5039 * optinfo.h: Update comment.
5040 * profile-count.c (profile_quality_as_string): New function.
5041 * profile-count.h (profile_quality_as_string): New decl.
5042 (profile_count::quality): New accessor.
5043 * selftest-run-tests.c (selftest::run_tests): Call json_cc_tests
5044 and optinfo_emit_json_cc_tests.
5045 * selftest.h (selftest::json_cc_tests): New decl.
5046 (selftest::optinfo_emit_json_cc_tests): New decl.
5047 * toplev.c: Include "optinfo-emit-json.h".
5048 (compile_file): Call optimization_records_finish.
5049 (do_compile): Call optimization_records_start.
5050 * tree-ssa-live.c: Include optinfo.h.
5051 (remove_unused_scope_block_p): Retain inlining information if
5052 optinfo_wants_inlining_info_p returns true.
5054 2018-07-20 Richard Biener <rguenther@suse.de>
5057 * dwarf2out.c (dwarf2out_die_ref_for_decl): Test in_lto_p
5058 to cover -flto-partition=none.
5060 2018-07-20 Martin Liska <mliska@suse.cz>
5062 * tree.h (DECL_LOCATION_RANGE): Remove unused macro.
5063 (get_decl_source_range): Remove unused function.
5065 2018-07-20 Richard Biener <rguenther@suse.de>
5067 * tree-ssa-sccvn.h (struct vn_nary_op_s): Add next member.
5068 (struct vn_phi_s): Likewise.
5069 (struct vn_reference_s): Likewise.
5070 * tree-ssa-sccvn.c (vn_nary_op_hasher::equal): Add shortcut
5071 for searching the slot of an entry known to be in the hash itself.
5072 (vn_phi_hasher::equal): Likewise.
5073 (vn_reference_hasher::equal): Likewise.
5074 (last_inserted_ref, last_inserted_phi, last_inserted_nary): New
5076 (optimistic_info, current_info): Remove, keeping only valid_info.
5077 (vn_reference_lookup_1): Remove fallback lookup.
5078 (vn_reference_lookup_2): Likewise.
5079 (vn_nary_op_lookup_1): Likewise.
5080 (vn_phi_lookup): Likewise.
5081 (vn_nary_build_or_lookup_1): Make sure to not chain the built
5083 (vn_reference_insert): Adjust, chain the inserted hash element
5084 at last_inserted_ref.
5085 (vn_reference_insert_pieces): Likewise.
5086 (visit_reference_op_call): Likewise.
5087 (vn_nary_op_insert_into): Chain the inserted hash element at
5089 (vn_nary_op_insert_pieces): Adjust.
5090 (vn_nary_op_insert): Likewise.
5091 (vn_nary_op_insert_stmt): Likewise.
5092 (vn_phi_insert): Adjust, chain the inserted hash element at
5094 (process_scc): Remove clearing and copying the optimistic
5095 table. Instead remove elements inserted during an optimistic
5096 iteration from the single table we maintain.
5097 (init_scc_vn): Adjust.
5098 (free_scc_vn): Likewise.
5099 (sccvn_dom_walker::record_cond): Likewise.
5100 (sccvn_dom_walker::after_dom_children): Likewise.
5102 2018-07-19 Martin Sebor <msebor@redhat.com>
5104 PR tree-optimization/84047
5105 PR tree-optimization/83776
5106 * tree-vrp.c (vrp_prop::check_mem_ref): New function.
5107 (check_array_bounds): Call it.
5109 2018-07-19 Martin Sebor <msebor@redhat.com>
5111 * align.h (align_flags): Use member initialization.
5113 2018-07-19 David Malcolm <dmalcolm@redhat.com>
5115 * Makefile.in (OBJS): Add optinfo.o.
5116 * coretypes.h (class symtab_node): New forward decl.
5117 (struct cgraph_node): New forward decl.
5118 (class varpool_node): New forward decl.
5119 * dump-context.h: New file.
5120 * dumpfile.c: Include "optinfo.h", "dump-context.h", "cgraph.h",
5122 (refresh_dumps_are_enabled): Use optinfo_enabled_p.
5123 (set_dump_file): Call dumpfile_ensure_any_optinfo_are_flushed.
5124 (set_alt_dump_file): Likewise.
5125 (dump_context::~dump_context): New dtor.
5126 (dump_gimple_stmt): Move implementation to...
5127 (dump_context::dump_gimple_stmt): ...this new member function.
5128 Add the stmt to any pending optinfo, creating one if need be.
5129 (dump_gimple_stmt_loc): Move implementation to...
5130 (dump_context::dump_gimple_stmt_loc): ...this new member function.
5131 Start a new optinfo and add the stmt to it.
5132 (dump_gimple_expr): Move implementation to...
5133 (dump_context::dump_gimple_expr): ...this new member function.
5134 Add the stmt to any pending optinfo, creating one if need be.
5135 (dump_gimple_expr_loc): Move implementation to...
5136 (dump_context::dump_gimple_expr_loc): ...this new member function.
5137 Start a new optinfo and add the stmt to it.
5138 (dump_generic_expr): Move implementation to...
5139 (dump_context::dump_generic_expr): ...this new member function.
5140 Add the tree to any pending optinfo, creating one if need be.
5141 (dump_generic_expr_loc): Move implementation to...
5142 (dump_context::dump_generic_expr_loc): ...this new member
5143 function. Add the tree to any pending optinfo, creating one if
5145 (dump_printf): Move implementation to...
5146 (dump_context::dump_printf_va): ...this new member function. Add
5147 the text to any pending optinfo, creating one if need be.
5148 (dump_printf_loc): Move implementation to...
5149 (dump_context::dump_printf_loc_va): ...this new member function.
5150 Start a new optinfo and add the stmt to it.
5151 (dump_dec): Move implementation to...
5152 (dump_context::dump_dec): ...this new member function. Add the
5153 value to any pending optinfo, creating one if need be.
5154 (dump_context::dump_symtab_node): New member function.
5155 (dump_context::get_scope_depth): New member function.
5156 (dump_context::begin_scope): New member function.
5157 (dump_context::end_scope): New member function.
5158 (dump_context::ensure_pending_optinfo): New member function.
5159 (dump_context::begin_next_optinfo): New member function.
5160 (dump_context::end_any_optinfo): New member function.
5161 (dump_context::s_current): New global.
5162 (dump_context::s_default): New global.
5163 (dump_scope_depth): Delete global.
5164 (dumpfile_ensure_any_optinfo_are_flushed): New function.
5165 (dump_symtab_node): New function.
5166 (get_dump_scope_depth): Reimplement in terms of dump_context.
5167 (dump_begin_scope): Likewise.
5168 (dump_end_scope): Likewise.
5169 (selftest::temp_dump_context::temp_dump_context): New ctor.
5170 (selftest::temp_dump_context::~temp_dump_context): New dtor.
5171 (selftest::verify_item): New function.
5172 (ASSERT_IS_TEXT): New macro.
5173 (ASSERT_IS_TREE): New macro.
5174 (ASSERT_IS_GIMPLE): New macro.
5175 (selftest::test_capture_of_dump_calls): New test.
5176 (selftest::dumpfile_c_tests): Call it.
5177 * dumpfile.h (dump_printf, dump_printf_loc, dump_basic_block)
5178 (dump_generic_expr_loc, dump_generic_expr, dump_gimple_stmt_loc)
5179 (dump_gimple_stmt, dump_dec): Gather these related decls and add a
5180 descriptive comment.
5181 (dump_function, print_combine_total_stats, enable_rtl_dump_file)
5182 (dump_node, dump_bb): Move these unrelated decls.
5183 (class dump_manager): Add leading comment.
5184 * optinfo.cc: New file.
5185 * optinfo.h: New file.
5187 2018-07-19 Michael Collison <michael.collison@arm.com>
5188 Richard Henderson <rth@redhat.com>
5190 * config/aarch64/aarch64.md (subv<GPI>4, usubv<GPI>4): New patterns.
5191 (subti): Handle op1 zero.
5192 (subvti4, usub4ti4): New.
5193 (*sub<GPI>3_compare1_imm): New.
5194 (sub<GPI>3_carryinCV): New.
5195 (*sub<GPI>3_carryinCV_z1_z2, *sub<GPI>3_carryinCV_z1): New.
5196 (*sub<GPI>3_carryinCV_z2, *sub<GPI>3_carryinCV): New.
5198 2018-07-19 Michael Collison <michael.collison@arm.com>
5199 Richard Henderson <rth@redhat.com>
5201 * config/aarch64/aarch64.md: (addv<GPI>4, uaddv<GPI>4): New.
5202 (addti3): Create simpler code if low part is already known to be 0.
5203 (addvti4, uaddvti4): New.
5204 (*add<GPI>3_compareC_cconly_imm): New.
5205 (*add<GPI>3_compareC_cconly): New.
5206 (*add<GPI>3_compareC_imm): New.
5207 (*add<GPI>3_compareC): Rename from add<GPI>3_compare1; do not
5208 handle constants within this pattern..
5209 (*add<GPI>3_compareV_cconly_imm): New.
5210 (*add<GPI>3_compareV_cconly): New.
5211 (*add<GPI>3_compareV_imm): New.
5212 (add<GPI>3_compareV): New.
5213 (add<GPI>3_carryinC, add<GPI>3_carryinV): New.
5214 (*add<GPI>3_carryinC_zero, *add<GPI>3_carryinV_zero): New.
5215 (*add<GPI>3_carryinC, *add<GPI>3_carryinV): New.
5216 ((*add<GPI>3_compareC_cconly_imm): Replace 'ne' operator
5217 with 'comparison' operator.
5218 (*add<GPI>3_compareV_cconly_imm): Ditto.
5219 (*add<GPI>3_compareV_cconly): Ditto.
5220 (*add<GPI>3_compareV_imm): Ditto.
5221 (add<GPI>3_compareV): Ditto.
5222 (add<mode>3_carryinC): Ditto.
5223 (*add<mode>3_carryinC_zero): Ditto.
5224 (*add<mode>3_carryinC): Ditto.
5225 (add<mode>3_carryinV): Ditto.
5226 (*add<mode>3_carryinV_zero): Ditto.
5227 (*add<mode>3_carryinV): Ditto.
5229 2018-07-19 Michael Collison <michael.collison@arm.com>
5230 Richard Henderson <rth@redhat.com>
5232 * config/aarch64/aarch64-modes.def (CC_V): New.
5233 * config/aarch64/aarch64-protos.h
5234 (aarch64_addti_scratch_regs): Declare
5235 (aarch64_subvti_scratch_regs): Declare.
5236 (aarch64_expand_subvti): Declare.
5237 (aarch64_gen_unlikely_cbranch): Declare
5238 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Test
5239 for signed overflow using CC_Vmode.
5240 (aarch64_get_condition_code_1): Handle CC_Vmode.
5241 (aarch64_gen_unlikely_cbranch): New function.
5242 (aarch64_addti_scratch_regs): New function.
5243 (aarch64_subvti_scratch_regs): New function.
5244 (aarch64_expand_subvti): New function.
5246 2018-07-19 Andre Vieira <andre.simoesdiasvieira@arm.com>
5248 * config/aarch64/aarch64-option-extensions.def: New entry for profile
5250 * config/aarch64/aarch64.h (AARCH64_FL_PROFILE): New.
5251 * doc/invoke.texi (aarch64-feature-modifiers): New entry for profile
5254 2018-07-19 Andre Vieira <andre.simoesdiasvieira@arm.com>
5257 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Make
5258 address check not strict.
5260 2018-07-19 Andre Vieira <andre.simoesdiasvieira@arm.com>
5262 * config/aarch64/aarch64-simd.md (aarch64_simd_mov<VQ:mode>): Replace
5264 (store_pair_lanes<mode>): Likewise.
5265 * config/aarch64/aarch64-protos.h (aarch64_addr_query_type): Add new
5266 enum value 'ADDR_QUERY_LDP_STP_N'.
5267 * config/aarch64/aarch64.c (aarch64_addr_query_type): Likewise.
5268 (aarch64_print_address_internal): Add declaration.
5269 (aarch64_print_ldpstp_address): Remove.
5270 (aarch64_classify_address): Adapt mode for 'ADDR_QUERY_LDP_STP_N'.
5271 (aarch64_print_operand): Change printing of 'y'.
5272 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Use
5273 new enum value 'ADDR_QUERY_LDP_STP_N', don't hardcode mode and use
5274 'true' rather than '1'.
5275 * config/aarch64/constraints.md (Uml): Likewise.
5276 (Uml): Rename to Umn.
5279 2018-07-19 Richard Biener <rguenther@suse.de>
5281 * tree-ssa-sccvn.h (struct vn_phi_s): Make phiargs member
5283 * tree-ssa-sccvn.c: Remove alloc-pool.h use.
5284 (vn_phi_hasher): Derive from nofree_ptr_hash and remove remove method.
5285 (vn_reference_hasher): Likewise.
5286 (struct vn_tables_s): Remove obstack and alloc-pool members.
5287 (vn_tables_obstack, vn_tables_insert_obstack): New global obstacks.
5288 (vn_nary_build_or_lookup_1): Manually build in vn_tables_insert_obstack.
5289 (vn_reference_insert): Allocate from obstack instead of from alloc-pool.
5290 (vn_reference_insert_pieces): Likewise.
5291 (alloc_vn_nary_op_noinit): Adjust.
5292 (vn_nary_op_insert_stmt): Allocate phiargs in-place.
5293 (vn_phi_eq): Adjust.
5294 (shared_lookup_phiargs): Remove.
5295 (vn_phi_lookup): Allocate temporary vn_phi_s on the stack.
5296 (vn_phi_insert): Allocate from obstack instead of from alloc-pool.
5297 (visit_reference_op_call): Likewise.
5298 (copy_nary, copy_phi, copy_reference): Remove.
5299 (process_scc): Rewind the obstack when iterating. Do not
5300 copy the elements to valid_info but just move them from one
5301 hashtable to the other.
5302 (allocate_vn_table): Adjust.
5303 (free_vn_table): Likewise.
5304 (init_scc_vn): Likewise.
5305 (free_scc_vn): Likewise.
5307 2018-07-19 H.J. Lu <hongjiu.lu@intel.com>
5310 * config/i386/i386.c (rest_of_insert_endbranch): Lookup
5311 indirect_return as function type attribute.
5312 (ix86_attribute_table): Change indirect_return to function
5314 * doc/extend.texi: Update indirect_return attribute.
5316 2018-07-19 Aldy Hernandez <aldyh@redhat.com>
5318 * wide-int.h (widest2_int): New.
5319 * gimple-fold.c (arith_overflowed_p): Use it.
5320 * tree.h (widest2_int_cst): New.
5321 * tree-vrp.c (wide_int_binop_overflow): Rename from
5322 vrp_int_const_binop.
5323 Rewrite to work on trees.
5324 (extract_range_from_multiplicative_op_1): Abstract code to...
5325 (wide_int_range_min_max): ...here.
5326 (wide_int_range_cross_product): ...and here.
5327 (extract_range_from_binary_expr_1): Abstract overflow code to...
5328 (wide_int_range_mult_wrapping): ...here.
5329 * tree-vrp.h (wide_int_range_cross_product): New.
5330 (wide_int_range_mult_wrapping): New.
5332 2018-07-19 Andrew Senkevich <andrew.senkevich@intel.com>
5333 Julia Koval <julia.koval@intel.com>
5335 * config/i386/x86-tune-costs.h (skylake_memcpy,
5336 skylake_memset): Replace rep_prefix with unrolling for size 512.
5338 2018-07-18 Kugan Vivekanandarajah <kuganv@linaro.org>
5341 * tree-ssa-phiopt.c (cond_removal_in_popcount_pattern): Handle
5342 comparision with EQ_EXPR in last stmt.
5344 2018-07-18 Kelvin Nilsen <kelvin@gcc.gnu.org>
5346 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Rename
5347 this subsection to "PowerPC AltiVec/VSX Built-in Functions".
5348 (PowerPC AltiVec/VSX Built-in Functions): New name for subsection
5349 previously known as "PowerPC AltiVec Built-in Functions". Move
5350 some material to new subsubsections "PowerPC AltiVec Built-in
5351 Functions on ISA 2.06" and "PowerPC AltiVec Built-in Functions on
5353 (PowerPC Altivec Built-in Functions on ISA 2.05): New subsubsection.
5354 (PowerPC Altivec Built-in Functions on ISA 2.06): Likewise.
5355 (PowerPC Altivec Built-in Functions on ISA 2.07): Likewise.
5356 (PowerPC Altivec Built-in Functions on ISA 3.0): Likewise.
5358 2018-07-18 Richard Biener <rguenther@suse.de>
5360 PR tree-optimization/86557
5361 * tree-vect-patterns.c (vect_recog_divmod_pattern): Also handle
5364 2018-07-18 Ilya Leoshkevich <iii@linux.ibm.com>
5366 * config/s390/s390.c (s390_function_profiler): Generate CFI.
5368 2018-07-17 Jeff Law <law@redhat.com>
5370 * config/arm/arm.c (get_label_padding): Update for recent
5371 changes to label_to_alignment.
5373 PR tree-optimization/86010
5374 * tree-ssa-dse.c (compute_trims): Fix typo/thinko.
5376 * config/mips/mips.c (vr4130_align_insns): Update for recent
5377 changes to label_to_alignment.
5379 * config/frv/frv.c (frv_label_align): Update for recent changes
5380 to label_to_alignment.
5382 * config/nios2/nios2.c (nios2_label_align): Update for recent
5383 changes which dropped ALIGN_LABELS_LOG.
5385 2018-07-17 Andreas Schwab <schwab@linux-m68k.org>
5387 * config/m68k/m68k.md (umulsi3_highpart+1, const_umulsi3_highpart)
5388 (smulsi3_highpart+1, const_smulsi3_highpart): Add CC_STATUS_INIT.
5390 2018-07-17 Claudiu Zissulescu <claziss@synopsys.com>
5392 * config/arc/arc.c (arc_label_align): Use align_labels instead of
5393 deprecated align_labels_log.
5395 2018-07-17 Richard Biener <rguenther@suse.de>
5398 * dwarf2out.c (init_sections_and_labels): Always generate
5399 a debug_line_str_section for early LTO debug.
5400 (dwarf2out_finish): Reset debug_line_str_hash output early.
5401 Bump counter for extra dwarf5 .debug_loc labels to not conflict
5403 (dwarf2out_early_finish): Output debug_line_str.
5405 2018-07-17 Robin Dapp <rdapp@linux.ibm.com>
5407 * config/s390/s390.c (preferred_la_operand_p): Do not use LA with
5408 index register on z196 or later.
5410 2018-07-17 Robin Dapp <rdapp@linux.ibm.com>
5412 * config/s390/s390.c (s390_default_align): Set default function
5414 (s390_override_options_after_change): Call s390_default align.
5415 (s390_option_override_internal): Call s390_default_align.
5416 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define.
5418 2018-07-17 Jakub Jelinek <jakub@redhat.com>
5421 * omp-low.c (create_task_copyfn): Copy over also fields corresponding
5422 to _looptemp_ clauses, other than the first two.
5424 2018-07-17 Martin Liska <mliska@suse.cz>
5426 * opts.c: Do not enable OPT_falign_* for -Os.
5428 2018-07-17 Martin Liska <mliska@suse.cz>
5430 * align.h (MAX_CODE_ALIGN): New.
5431 (MAX_CODE_ALIGN_VALUE): New.
5432 * common/config/i386/i386-common.c (ix86_handle_option):
5433 (MAX_CODE_ALIGN): Moved to align.h.
5434 * final.c (MAX_CODE_ALIGN): Likewise.
5435 * opts.c (parse_and_check_align_values):
5436 (MAX_CODE_ALIGN): Likewise.
5437 (MAX_CODE_ALIGN_VALUE): Likewise.
5439 2018-07-17 Martin Liska <mliska@suse.cz>
5441 * config/i386/att.h (ASM_OUTPUT_ALIGN): Fix spacing
5442 in order to fulfil coding style.
5443 * config/i386/cygming.h (ASM_OUTPUT_ALIGN): Likewise.
5444 * config/i386/gas.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
5445 * config/i386/x86-64.h (ASM_OUTPUT_MAX_SKIP_PAD): Likewise.
5446 * config/iq2000/iq2000.h (ASM_OUTPUT_ALIGN): Likewise.
5447 * config/pa/pa.h (ASM_OUTPUT_ALIGN): Likewise.
5448 * config/sparc/sol2.h (ASM_OUTPUT_ALIGN_WITH_NOP): Likewise.
5449 * config/sparc/sparc.h (ASM_OUTPUT_ALIGN): Likewise.
5450 * config/visium/visium.h (ASM_OUTPUT_ALIGN): Likewise.
5451 (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
5453 2018-07-17 Martin Liska <mliska@suse.cz>
5455 * align.h: New file.
5456 * config/alpha/alpha.c (alpha_align_insns_1): Use align_functions
5458 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Use new return type
5459 align_flags of label_to_alignment.
5460 * config/m32r/m32r.h (LOOP_ALIGN): Wrap returned values into
5462 * config/m68k/m68k.c: Do not use removed align_labels_value and
5464 * config/nds32/nds32.h (JUMP_ALIGN): Wrap result into align_flags class.
5465 (LOOP_ALIGN): Likewise.
5466 (LABEL_ALIGN): Likewise.
5467 * config/powerpcspe/powerpcspe.c (TARGET_ASM_LOOP_ALIGN_MAX_SKIP):
5468 Remove not used macro.
5469 (rs6000_loop_align): Change return type to align_flags.
5470 (rs6000_loop_align_max_skip): Remove.
5471 * config/rs6000/rs6000-protos.h (rs6000_loop_align):
5472 Change return type to align_flags.
5473 * config/rs6000/rs6000.c (TARGET_ASM_LOOP_ALIGN_MAX_SKIP):
5474 Remove not used macro.
5475 (rs6000_loop_align): Change return type to align_flags.
5476 (rs6000_loop_align_max_skip): Remove.
5477 * config/rx/rx.h (JUMP_ALIGN): Wrap integer values
5478 * config/rx/rx-protos.h (rx_align_for_label): Make it
5480 * config/rx/rx.c (rx_align_for_label): Change return type
5482 (rx_max_skip_for_label): Remove TARGET_ASM_*_ALIGN_MAX_SKIP
5484 into align_flags class.
5485 (LABEL_ALIGN): Likewise.
5486 (LOOP_ALIGN): Likewise.
5487 * config/s390/s390.c (s390_label_align): Use align_flags
5489 (s390_asm_output_function_label): Likewise.
5490 * config/sh/sh.c (sh_override_options_after_change):
5491 Use align_flags class directly without macros.
5492 (find_barrier): Likewise.
5493 (barrier_align): Likewise.
5494 (sh_loop_align): Likewise.
5495 * config/spu/spu.c (spu_option_override):
5496 Use align_flags_tuple::get_value instead of removed macros.
5497 (spu_sched_init): Likewise.
5498 * config/spu/spu.h (GTY): Likewise.
5499 * config/visium/visium.c (visium_option_override):
5500 Set "8" as default secondary alignment.
5501 * config/visium/visium.h (SUBALIGN_LOG): Define to 3
5502 in order to guarantee secondary alignment of 8.
5503 * coretypes.h: Include align.h header file.
5504 * doc/tm.texi: Remove TARGET_ASM_JUMP_ALIGN_MAX_SKIP,
5505 TARGET_ASM_LOOP_ALIGN_MAX_SKIP, TARGET_ASM_LABEL_ALIGN_MAX_SKIP
5506 and TARGET_ASM_LABEL_ALIGN_AFTER_BARRIER_MAX_SKIP macros.
5507 * doc/tm.texi.in: Likewise.
5508 * final.c (struct label_alignment): Remove not used structure.
5509 (LABEL_ALIGN): Change type to align_flags.
5510 (LOOP_ALIGN): Likewise.
5511 (JUMP_ALIGN): Likewise.
5512 (default_loop_align_max_skip): Remove.
5513 (default_label_align_max_skip): Likewise.
5514 (default_jump_align_max_skip): Likewise.
5515 (default_label_align_after_barrier_max_skip):
5516 (LABEL_TO_ALIGNMENT): Change to access label_align vector.
5517 (LABEL_TO_MAX_SKIP): Remove.
5518 (label_to_alignment): Return align_flags type instead of integer.
5519 (label_to_max_skip): Remove.
5520 (align_fuzz): Use align_flags type.
5521 (compute_alignments): Use align_flags type and use align_flags::max
5522 to combine multiple alignments.
5523 (grow_label_align): Grow vec instead of C array.
5524 (update_alignments): Assign just LABEL_TO_ALIGNMENT.
5525 (shorten_branches): Use align_flags type and use align_flags::max
5526 to combine multiple alignments.
5527 (final_scan_insn_1): Remove usage of secondary alignment that comes
5528 from label alignment, but instead use proper secondary alignment
5529 which is computed in grow_label_align.
5530 * flags.h (struct align_flags_tuple): Move to align.h.
5531 (struct align_flags): Likewise.
5532 (state_align_loops): Rename to align_loops.
5533 (state_align_jumps): Rename to align_jumps.
5534 (state_align_labels): Rename to align_labels.
5535 (state_align_functions): Rename to align_functions.
5536 (align_loops_log): Remove.
5537 (align_jumps_log): Remove.
5538 (align_labels_log): Remove.
5539 (align_functions_log): Remove.
5540 (align_loops_max_skip): Remove.
5541 (align_jumps_max_skip): Remove.
5542 (align_labels_max_skip): Remove.
5543 (align_functions_max_skip): Remove.
5544 (align_loops_value): Remove.
5545 (align_jumps_value): Remove.
5546 (align_labels_value): Remove.
5547 (align_functions_value): Remove.
5548 * output.h (label_to_alignment): Change return type to align_flags.
5549 (label_to_max_skip): Remove.
5550 * target.def: Remove loop_align_max_skip, label_align_max_skip,
5551 jump_align_max_skip macros.
5552 * targhooks.h (default_loop_align_max_skip): Remove.
5553 (default_label_align_max_skip): Likewise.
5554 (default_jump_align_max_skip): Likewise.
5555 (default_label_align_after_barrier_max_skip): Remove.
5556 * toplev.c (read_log_maxskip): Use ::normalize function.
5557 (parse_N_M): Remove not used argument and also call ::normalize.
5558 (parse_alignment_opts): Do not pass unused arguments.
5559 * varasm.c (assemble_start_function): Use directly align_functions
5560 instead of removed macros.
5561 * system.h: Do not poison removed macros.
5563 2018-07-17 Jakub Jelinek <jakub@redhat.com>
5566 * gimplify.c (gimplify_omp_for): Ensure taskloop firstprivatized init
5567 and cond temporaries don't have reference type if iterator has
5568 pointer type. For init use &for_pre_body instead of pre_p if
5569 for_pre_body is non-empty.
5571 2018-07-16 Segher Boessenkool <segher@kernel.crashing.org>
5573 * config/rs6000/rs6000.md (trunc<mode>sf2): Expand truncates of
5574 double-double modes to SFmode directly directly.
5575 (trunc<mode>sf2_fprs): Delete.
5577 2018-07-16 Segher Boessenkool <segher@kernel.crashing.org>
5579 * config/rs6000/rs6000.c (init_float128_ibm): Use the correct names
5580 for conversions between IFmode and the decimal floating point modes.
5581 (init_float128_ieee): Use the correct names for conversions between
5582 KFmode and the decimal floating point modes.
5584 2018-07-16 Segher Boessenkool <segher@kernel.crashing.org>
5586 * config/rs6000/rs6000.c (init_float128_ibm): Use more correct names
5587 for the conversions between TDmode and IFmode.
5588 (init_float128_ieee): Use more correct names for the conversions
5589 between TDmode and KFmode.
5591 2018-07-16 Jakub Jelinek <jakub@redhat.com>
5593 PR tree-optimization/86526
5594 * builtins.c (expand_builtin_memcmp): Formatting fixes.
5595 (inline_expand_builtin_string_cmp): Likewise.
5596 (inline_string_cmp): Likewise. Use c_readstr instead of
5597 builtin_memcpy_read_str. Add unit_mode temporary.
5599 2018-07-16 Bernd Edlinger <bernd.edlinger@hotmail.de>
5602 * builtins.c (check_access): Bail out if range[0] is no INTEGER_CST.
5603 * expr.c (string_constant): Fix the element size of ARRAY_TYPE.
5605 2018-07-16 Kelvin Nilsen <kelvin@gcc.gnu.org>
5607 * doc/extend.texi (PowerPC AltiVec Built-in Functions):
5608 Alphabetize prototypes of built-in functions, separating out
5609 built-in functions that are listed in this section but should be
5610 described elsewhere.
5612 2018-07-16 Uros Bizjak <ubizjak@gmail.com>
5615 * expmed.c (emit_store_flag): Do not emit setcc followed by a
5616 conditional move when trapping comparison was split to a
5617 non-trapping one (and vice versa).
5619 2018-07-16 Ilya Leoshkevich <iii@linux.ibm.com>
5621 * config/s390/s390.c (s390_function_profiler): Generate nops
5622 instead of profiler call sequences.
5623 * config/s390/s390.opt: Add the new option.
5625 2018-07-16 Ilya Leoshkevich <iii@linux.ibm.com>
5627 * config/s390/s390.c (s390_function_profiler): Generate
5628 __mcount_loc section.
5629 * config/s390/s390.opt: Add the new option.
5631 2018-07-16 Ilya Leoshkevich <iii@linux.ibm.com>
5633 * common.opt: Add the new warning.
5634 * config/s390/s390.c (s390_function_profiler): Emit "brasl
5635 %r0,__fentry__" when -mfentry is specified.
5636 (s390_option_override_internal): Disallow -mfentry for 31-bit
5638 * config/s390/s390.opt: Add the new option.
5640 2018-07-16 Richard Biener <rguenther@suse.de>
5643 * dwarf2out.c (dwarf2out_register_external_die): Assign DIE parents
5644 for function-local FUNCTION_DECL and RESULT_DECL immediately.
5646 2018-07-16 Martin Liska <mliska@suse.cz>
5649 * ipa-pure-const.c (malloc_candidate_p): Revert ::get
5652 2017-07-16 Claudiu Zissulescu <claziss@synopsys.com>
5654 * config/arc/arcHS.md: Update ARCHS scheduling rules.
5656 2017-07-16 Claudiu Zissulescu <claziss@synopsys.com>
5658 * config/arc/arc-arch.h (arc_tune_attr): Add new tune parameters
5660 * config/arc/arc-cpus.def (hs4x): New cpu.
5662 * config/arc/arc-tables.opt: Regenerate.
5663 * config/arc/arc.c (arc_sched_issue_rate): New function.
5664 (TARGET_SCHED_ISSUE_RATE): Define.
5665 (TARGET_SCHED_EXPOSED_PIPELINE): Likewise.
5666 * config/arc/arc.md (attr type): Add fpu_fuse, fpu_sdiv, fpu_ddiv,
5668 (attr tune): Add ARCHS4x tune values.
5669 (attr tune_dspmpy): Define.
5670 (*tst): Correct instruction type.
5671 * config/arc/arcHS.md: Don't use this automaton for ARCHS4x cpus.
5672 * config/arc/arcHS4x.md: New file.
5673 * config/arc/fpu.md: Update instruction type attributes.
5674 * config/arc/t-multilib: Regenerate.
5676 2018-07-16 Tom de Vries <tdevries@suse.de>
5679 * var-tracking.c (vt_initialize): Fix pre_dec handling.
5681 2018-07-16 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
5683 * config/aarch64/atomics.md (aarch64_store_execlusive<mode>): Add
5686 2018-07-16 Eric Botcazou <ebotcazou@adacore.com>
5688 PR tree-optimization/86514
5689 * tree-ssa-reassoc.c (init_range_entry) <CASE_CONVERT>: Return for a
5690 conversion to a boolean type from a type with greater precision.
5692 2018-07-16 Tom de Vries <tdevries@suse.de>
5694 * var-tracking.c (vt_initialize): Print adjusted insn slim if
5695 dump_flags request TDF_SLIM.
5697 2018-07-16 Aldy Hernandez <aldyh@redhat.com>
5699 * fold-const.c (int_const_binop_1): Abstract...
5700 (wide_int_binop): ...wide int code here.
5701 (poly_int_binop): ...poly int code here.
5702 Abstract the rest of int_const_binop_1 into int_const_binop.
5703 * fold-const.h (wide_int_binop): New.
5704 * tree-vrp.c (vrp_int_const_binop): Call wide_int_binop.
5705 Remove useless PLUS/MINUS_EXPR case.
5706 (zero_nonzero_bits_from_vr): Move wide int code...
5707 (zero_nonzero_bits_from_bounds): ...here.
5708 (extract_range_from_binary_expr_1): Move mask optimization code...
5709 (range_easy_mask_min_max): ...here.
5710 * tree-vrp.h (zero_nonzero_bits_from_bounds): New.
5711 (range_easy_mask_min_max): New.
5713 2018-07-15 Jeff Law <law@redhat.com>
5716 * config/sh/sh.c (output_mi_thunk): Remove dead conditional
5719 2018-07-14 Jim Wilson <jimw@sifive.com>
5721 * config/riscv/linux.h (TARGET_ASM_FILE_END): New.
5723 2018-07-14 Paul Koning <ni1d@arrl.net>
5725 * config/pdp11/pdp11.c (pdp11_rtx_costs): Bugfixes.
5727 2018-07-13 Jan Hubicka <hubicka@ucw.cz>
5729 * lto-streamer-out.c (copy_function_or_variable): Dump info about
5732 2018-07-13 Bill Schmidt <wschmidt@linux.ibm.com>
5733 Steve Munroe <munroesj52@gmail.com>
5735 * config/rs6000/emmintrin.h (_mm_and_si128): New function.
5736 (_mm_andnot_si128): Likewise.
5737 (_mm_or_si128): Likewise.
5738 (_mm_xor_si128): Likewise.
5740 2018-07-13 Qing Zhao <qing.zhao@oracle.com>
5743 * builtins.c (expand_builtin_memcmp): Inline the calls first
5744 when result_eq is false.
5745 (expand_builtin_strcmp): Inline the calls first.
5746 (expand_builtin_strncmp): Likewise.
5747 (inline_string_cmp): New routine. Expand a string compare
5748 call by using a sequence of char comparison.
5749 (inline_expand_builtin_string_cmp): New routine. Inline expansion
5750 a call to str(n)cmp/memcmp.
5751 * doc/invoke.texi (--param builtin-string-cmp-inline-length):
5753 * params.def (BUILTIN_STRING_CMP_INLINE_LENGTH): New.
5755 2018-07-13 Richard Earnshaw <rearnsha@arm.com>
5757 * config/arm/driver-arm.c: Include arm-native.h.
5758 (host_detect_local_cpu): Use auto-generated data tables.
5759 (vendors, arm_cpu_table): Delete. Move part information to ...
5760 * config/arm/arm-cpus.in: ... here.
5761 * config/arm/parsecpu.awk (gen_native): New function.
5762 (vendor, part): New CPU fields.
5763 (END): Add support for building the native CPU detection tables.
5764 * config/arm/t-arm (arm-native.h): Add build rule.
5765 (driver-arm.o): Add dependency on arm-native.h.
5767 2018-07-13 Richard Biener <rguenther@suse.de>
5770 * match.pd (addr1 - addr2): Allow either of the operand to
5773 2018-07-13 Tom de Vries <tdevries@suse.de>
5775 * tree-inline.c (remap_ssa_name): Save and reuse debug exprs generated
5778 2018-07-13 Jackson Woodruff <jackson.woodruff@arm.com>
5780 * config/aarch64/aarch64.c (aarch64_operands_adjust_ok_for_ldpstp): Use
5781 arrays instead of numbered variables.
5783 2018-07-13 Eric Botcazou <ebotcazou@adacore.com>
5785 * config/sparc/sparc-protos.h (sparc_compute_frame_size): Delete.
5786 * config/sparc/sparc.c (sparc_compute_frame_size): Make static.
5788 2018-07-13 Richard Biener <rguenther@suse.de>
5791 * dwarf2out.c (gen_type_die_with_usage): Use scope_die_for
5792 instead of get_context_die.
5794 2018-07-13 Kugan Vivekanandarajah <kuganv@linaro.org>
5795 Richard Biener <rguenther@suse.de>
5798 * tree-ssa-loop-niter.c (number_of_iterations_popcount): Check
5799 that the loop latch destination where phi is defined.
5801 2018-07-12 Kito Cheng <kito.cheng@gmail.com>
5803 * config/riscv/riscv.c (enum riscv_privilege_levels): Add UNKNOWN_MODE.
5804 (riscv_expand_epilogue): Add assertion to check interrupt mode.
5805 (riscv_set_current_function): Extract getting interrupt type to new
5807 (riscv_get_interrupt_type): New function.
5808 (riscv_merge_decl_attributes): New function, checking interrupt type is
5810 (TARGET_MERGE_DECL_ATTRIBUTES): Define.
5812 2018-07-12 Paul Koning <ni1d@arrl.net>
5814 * config/pdp11/pdp11.c (pdp11_output_def): Fix typo in .set
5817 2018-07-12 Paul Koning <ni1d@arrl.net>
5819 * doc/rtl.texi (REG_NONNEG): Remove decrement and branch until
5820 zero reference, add doloop_end instead.
5821 * doc/md.texi (decrement_and_branch_until_zero): Remove.
5822 (Looping patterns): Remove decrement_and_branch_until_zero. Add
5823 detail for doloop_end.
5825 2018-07-12 Martin Sebor <msebor@redhat.com>
5828 * attribs.c (decl_attributes): Reject conflicting attributes before
5829 calling attribute handlers.
5831 2018-07-12 Jan Hubicka <hubicka@ucw.cz>
5833 * dumpfile.c (gcc::dump_manager::get_dump_file_name): Add PART
5835 (gcc::dump_manager::get_dump_file_name): likewise.
5836 (dump_begin): Likewise.
5837 * dumpfile.h (dump_begin): Update prototype.
5838 (gcc::dump_manager::get_dump_file_name,
5839 gcc::dump_manager::get_dump_file_name): Update prototype.
5841 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
5843 * internal-fn.h (vectorizable_internal_fn_p): New function.
5844 * tree-vect-slp.c (compatible_calls_p): Likewise.
5845 (vect_build_slp_tree_1): Remove nops argument. Handle calls
5846 to internal functions.
5847 (vect_build_slp_tree_2): Update call to vect_build_slp_tree_1.
5849 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
5851 * fold-const.h (inverse_conditions_p): Declare.
5852 * fold-const.c (inverse_conditions_p): New function.
5853 * match.pd: Use inverse_conditions_p. Add folds of view_converts
5854 that test the inverse condition of a conditional internal function.
5855 * internal-fn.h (vectorized_internal_fn_supported_p): Declare.
5856 * internal-fn.c (internal_fn_mask_index): Handle conditional
5858 (vectorized_internal_fn_supported_p): New function.
5859 * tree-if-conv.c: Include internal-fn.h and fold-const.h.
5860 (any_pred_load_store): Replace with...
5861 (need_to_predicate): ...this new variable.
5862 (redundant_ssa_names): New variable.
5863 (ifcvt_can_use_mask_load_store): Move initial checks to...
5864 (ifcvt_can_predicate): ...this new function. Handle tree codes
5865 for which a conditional internal function exists.
5866 (if_convertible_gimple_assign_stmt_p): Use ifcvt_can_predicate
5867 instead of ifcvt_can_use_mask_load_store. Update after variable
5869 (predicate_load_or_store): New function, split out from
5870 predicate_mem_writes.
5871 (check_redundant_cond_expr): New function.
5872 (value_available_p): Likewise.
5873 (predicate_rhs_code): Likewise.
5874 (predicate_mem_writes): Rename to...
5875 (predicate_statements): ...this. Use predicate_load_or_store
5876 and predicate_rhs_code.
5877 (combine_blocks, tree_if_conversion): Update after above name changes.
5878 (ifcvt_local_dce): Handle redundant_ssa_names.
5879 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern): Handle
5880 general conditional functions.
5881 * tree-vect-stmts.c (vectorizable_call): Likewise.
5883 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
5884 Alan Hayward <alan.hayward@arm.com>
5885 David Sherwood <david.sherwood@arm.com>
5887 * internal-fn.h (can_interpret_as_conditional_op_p): Declare.
5888 * internal-fn.c (can_interpret_as_conditional_op_p): New function.
5889 * tree-ssa-math-opts.c (convert_mult_to_fma_1): Handle conditional
5890 plus and minus and convert them into IFN_COND_FMA-based sequences.
5891 (convert_mult_to_fma): Handle conditional plus and minus.
5893 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
5895 * doc/md.texi (cond_fma, cond_fms, cond_fnma, cond_fnms): Document.
5896 * optabs.def (cond_fma_optab, cond_fms_optab, cond_fnma_optab)
5897 (cond_fnms_optab): New optabs.
5898 * internal-fn.def (COND_FMA, COND_FMS, COND_FNMA, COND_FNMS): New
5900 (FMA): Use DEF_INTERNAL_FLT_FN rather than DEF_INTERNAL_FLT_FLOATN_FN.
5901 * internal-fn.h (get_conditional_internal_fn): Declare.
5902 (get_unconditional_internal_fn): Likewise.
5903 * internal-fn.c (cond_ternary_direct): New macro.
5904 (expand_cond_ternary_optab_fn): Likewise.
5905 (direct_cond_ternary_optab_supported_p): Likewise.
5906 (FOR_EACH_COND_FN_PAIR): Likewise.
5907 (get_conditional_internal_fn): New function.
5908 (get_unconditional_internal_fn): Likewise.
5909 * gimple-match.h (gimple_match_op::MAX_NUM_OPS): Bump to 5.
5910 (gimple_match_op::gimple_match_op): Add a new overload for 5
5912 (gimple_match_op::set_op): Likewise.
5913 (gimple_resimplify5): Declare.
5914 * genmatch.c (decision_tree::gen): Generate simplifications for
5916 * gimple-match-head.c (gimple_simplify): Define an overload for
5917 5 operands. Handle calls with 5 arguments in the top-level overload.
5918 (convert_conditional_op): Handle conversions from unconditional
5919 internal functions to conditional ones.
5920 (gimple_resimplify5): New function.
5921 (build_call_internal): Pass a fifth operand.
5922 (maybe_push_res_to_seq): Likewise.
5923 (try_conditional_simplification): Try converting conditional
5924 internal functions to unconditional internal functions.
5925 Handle 3-operand unconditional forms.
5926 * match.pd (UNCOND_TERNARY, COND_TERNARY): Operator lists.
5927 Define ternary equivalents of the current rules for binary conditional
5929 * config/aarch64/aarch64.c (aarch64_preferred_else_value): Handle
5931 * config/aarch64/iterators.md (UNSPEC_COND_FMLA, UNSPEC_COND_FMLS)
5932 (UNSPEC_COND_FNMLA, UNSPEC_COND_FNMLS): New unspecs.
5933 (optab): Handle them.
5934 (SVE_COND_FP_TERNARY): New int iterator.
5935 (sve_fmla_op, sve_fmad_op): New int attributes.
5936 * config/aarch64/aarch64-sve.md (cond_<optab><mode>)
5937 (*cond_<optab><mode>_2, *cond_<optab><mode_4)
5938 (*cond_<optab><mode>_any): New SVE_COND_FP_TERNARY patterns.
5940 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
5942 * target.def (preferred_else_value): New target hook.
5943 * doc/tm.texi.in (TARGET_PREFERRED_ELSE_VALUE): New hook.
5944 * doc/tm.texi: Regenerate.
5945 * targhooks.h (default_preferred_else_value): Declare.
5946 * targhooks.c (default_preferred_else_value): New function.
5947 * internal-fn.h (conditional_internal_fn_code): Declare.
5948 * internal-fn.c (FOR_EACH_CODE_MAPPING): New macro.
5949 (get_conditional_internal_fn): Use it.
5950 (conditional_internal_fn_code): New function.
5951 * gimple-match.h (gimple_match_cond): New struct.
5952 (gimple_match_op): Add a cond member function.
5953 (gimple_match_op::gimple_match_op): Update all forms to take a
5955 * genmatch.c (expr::gen_transform): Use the same condition as res_op
5956 for the suboperation, but don't specify a particular else_value.
5957 * tree-ssa-sccvn.c (vn_nary_simplify, vn_reference_lookup_3)
5958 (visit_nary_op, visit_reference_op_load): Pass
5959 gimple_match_cond::UNCOND to the gimple_match_op constructor.
5960 * gimple-match-head.c: Include tree-eh.h
5961 (convert_conditional_op): New function.
5962 (maybe_resimplify_conditional_op): Likewise.
5963 (gimple_resimplify1): Call maybe_resimplify_conditional_op.
5964 (gimple_resimplify2): Likewise.
5965 (gimple_resimplify3): Likewise.
5966 (gimple_resimplify4): Likewise.
5967 (maybe_push_res_to_seq): Return null for conditional operations.
5968 (try_conditional_simplification): New function.
5969 (gimple_simplify): Call it. Pass conditions to the gimple_match_op
5971 * match.pd: Fold VEC_COND_EXPRs of an IFN_COND_* call to a new
5973 * config/aarch64/aarch64.c (aarch64_preferred_else_value): New
5975 (TARGET_PREFERRED_ELSE_VALUE): Redefine.
5977 2018-07-12 Jan Hubicka <hubicka@ucw.cz>
5979 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not stream
5981 (hash_tree): Do not hash DECL_FCONTEXT
5982 * tree-streamer-in.c (lto_input_ts_field_decl_tree_pointers):
5983 Do not stream DECL_FCONTEXT.
5984 * tree-streamer-out.c (write_ts_field_decl_tree_pointers): Likewise.
5985 * tree.c (free_lang_data_in_decl): Free DECL_FCONTEXT.
5987 2018-07-12 Richard Biener <rguenther@suse.de>
5990 * dwarf2out.c (gen_block_die): Only output blocks when they have
5991 at least one !DECL_IGNORED_P variable.
5993 2018-07-12 Richard Biener <rguenther@suse.de>
5996 * config/gnu-user.h (GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC):
5997 Remove -mieee-fp handling.
5999 2018-07-12 Richard Biener <rguenther@suse.de>
6001 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Remove bogus
6002 left-over from last patch.
6004 2018-07-12 Jakub Jelinek <jakub@redhat.com>
6006 PR tree-optimization/86492
6007 * gimple-ssa-store-merging.c
6008 (imm_store_chain_info::coalesce_immediate_stores): Call
6009 check_no_overlap even for the merge_overlapping case. Formatting fix.
6011 2018-07-12 Richard Biener <rguenther@suse.de>
6014 * fold-const.c (fold_binary_op_with_conditional_arg): Do not
6015 move possibly trapping operations into the conditional.
6017 2018-07-12 Richard Biener <rguenther@suse.de>
6019 * tree-ssa-sccvn.c (mprts_hook_cnt): Remove.
6020 (vn_lookup_simplify_result): Remove recursion limit applied
6022 (vn_nary_build_or_lookup_1): Adjust.
6023 (try_to_simplify): Likewise.
6024 * gimple-match-head.c (gimple_resimplify1): Instead apply one
6026 (gimple_resimplify2): Likewise.
6027 (gimple_resimplify3): Likewise.
6028 (gimple_resimplify4): Likewise.
6030 2018-07-11 Jakub Jelinek <jakub@redhat.com>
6032 * config/i386/avx512bitalgintrin.h (_mm512_mask_bitshuffle_epi64_mask):
6033 Use __mmask64 type instead of __mmask8 for __M argument.
6034 * config/i386/avx512fintrin.h (_mm512_mask_xor_epi64,
6035 _mm512_maskz_xor_epi64): Use __mmask8 type instead of __mmask16 for
6037 (_mm512_mask_cmpneq_epi64_mask): Use __mmask8 type instead of
6038 __mmask16 for __M argument.
6039 (_mm512_maskz_insertf32x4, _mm512_maskz_inserti32x4,
6040 _mm512_mask_insertf32x4, _mm512_mask_inserti32x4): Cast last argument
6041 to __mmask16 instead of __mmask8.
6042 * config/i386/avx512vlintrin.h (_mm_mask_add_ps, _mm_maskz_add_ps,
6043 _mm256_mask_add_ps, _mm256_maskz_add_ps, _mm_mask_sub_ps,
6044 _mm_maskz_sub_ps, _mm256_mask_sub_ps, _mm256_maskz_sub_ps,
6045 _mm256_maskz_cvtepi32_ps, _mm_maskz_cvtepi32_ps): Use __mmask8 type
6046 instead of __mmask16 for __U argument.
6047 * config/i386/avx512vlbwintrin.h (_mm_mask_cmp_epi8_mask): Use
6048 __mmask16 instead of __mmask8 for __U argument.
6049 (_mm256_mask_cmp_epi8_mask): Use __mmask32 instead of __mmask16 for
6051 (_mm256_cmp_epi8_mask): Use __mmask32 return type instead of
6053 (_mm_mask_cmp_epu8_mask): Use __mmask16 instead of __mmask8 for __U
6055 (_mm256_mask_cmp_epu8_mask): Use __mmask32 instead of __mmask16 for
6057 (_mm256_cmp_epu8_mask): Use __mmask32 return type instead of
6059 (_mm_mask_cmp_epi16_mask): Cast last argument to __mmask8 instead
6061 (_mm256_mask_cvtepi8_epi16): Use __mmask16 instead of __mmask32 for
6063 (_mm_mask_cvtepi8_epi16): Use __mmask8 instead of __mmask32 for
6065 (_mm256_mask_cvtepu8_epi16): Use __mmask16 instead of __mmask32 for
6067 (_mm_mask_cvtepu8_epi16): Use __mmask8 instead of __mmask32 for
6069 (_mm256_mask_cmpneq_epu8_mask, _mm256_mask_cmplt_epu8_mask,
6070 _mm256_mask_cmpge_epu8_mask, _mm256_mask_cmple_epu8_mask): Change
6071 return type as well as __M argument type and all casts from __mmask8
6073 (_mm256_mask_cmpneq_epu16_mask, _mm256_mask_cmplt_epu16_mask,
6074 _mm256_mask_cmpge_epu16_mask, _mm256_mask_cmple_epu16_mask): Change
6075 return type as well as __M argument type and all casts from __mmask8
6077 (_mm256_mask_cmpneq_epi8_mask, _mm256_mask_cmplt_epi8_mask,
6078 _mm256_mask_cmpge_epi8_mask, _mm256_mask_cmple_epi8_mask): Change
6079 return type as well as __M argument type and all casts from __mmask8
6081 (_mm256_mask_cmpneq_epi16_mask, _mm256_mask_cmplt_epi16_mask,
6082 _mm256_mask_cmpge_epi16_mask, _mm256_mask_cmple_epi16_mask): Change
6083 return type as well as __M argument type and all casts from __mmask8
6085 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_shrdi_epi32,
6086 _mm_mask_shldi_epi32): Cast last argument to __mmask8 instead of
6089 2018-07-11 Grazvydas Ignotas <notasas@gmail.com>
6091 * config/i386/avx512bwintrin.h: (_mm512_mask_cmp_epi8_mask,
6092 _mm512_mask_cmp_epu8_mask): Use __mmask64 type instead of __mmask32
6095 2018-07-11 Paul Koning <ni1d@arrl.net>
6097 * doc/md.texi (define_subst): Document how multiple occurrences of
6098 the same argument in the replacement pattern are handled.
6100 2018-07-11 Paul Koning <ni1d@arrl.net>
6102 * doc/extend.texi (Common Variable Attributes): Move "mode" into
6104 (Common Type Attributes): Add "mode" attribute.
6106 2018-07-11 Jan Hubicka <hubicka@ucw.cz>
6108 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not
6109 stream DECL_ORIGINAL_TYPE.
6110 (DFS::DFS_write_tree_body): Drop hack handling local external decls.
6111 (hash_tree): Do not walk DECL_ORIGINAL_TYPE.
6112 * tree-streamer-in.c (lto_input_ts_decl_non_common_tree_pointers):
6113 Do not walk original type.
6114 * tree-streamer-out.c (streamer_write_chain): Drop hack handling
6116 (write_ts_decl_non_common_tree_pointers): Do not stream
6118 * tree.c (free_lang_data_in_decl): Clear DECL_ORIGINAL_TYPE.
6119 (find_decls_types_r): Do not walk DEC_ORIGINAL_TYPE.
6121 2018-07-11 Aldy Hernandez <aldyh@redhat.com>
6123 * tree-ssa-threadupdate.c (thread_through_all_blocks): Do not jump
6124 thread twice from the same starting edge.
6126 2018-07-11 Aldy Hernandez <aldyh@redhat.com>
6128 * vr-values.c (gimple_stmt_nonzero_p): Abstract common code to...
6129 * gimple.c (gimple_call_nonnull_result_p): ...here...
6130 (gimple_call_nonnull_arg): ...and here.
6131 * gimple.h (gimple_call_nonnull_result_p): New.
6132 (gimple_call_nonnull_arg): New.
6134 2018-07-11 Richard Earnshaw <rearnsha@arm.com>
6136 * config/arm/arm-cpus.in: Move information from fpu field of each
6137 cpu definition to the isa field.
6138 * config/arm/parsecpu.awk (fpu): Delete match rule.
6139 (gen_comm_data): Don't add bits from the CPU's FPU entry.
6141 2018-07-11 Richard Biener <rguenther@suse.de>
6144 * dwarf2out.c (init_sections_and_labels): Use
6145 output_asm_line_debug_info consistently.
6146 (dwarf2out_early_finish): Likewise.
6147 (dwarf2out_finish): Remove DW_AT_stmt_list from early generated
6150 2018-07-11 Richard Biener <rguenther@suse.de>
6152 * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1):
6153 Rework father_bb setting in a way to avoid propagating constants
6154 multiple times on a loop body.
6156 2018-07-10 Mark Wielaard <mark@klomp.org>
6159 * dwarf2out.c (output_macinfo_op): Fix dwarf_FORM typo in gcc_assert.
6161 2018-07-10 Richard Biener <rguenther@suse.de>
6163 * hash-map.h (hash_map::iterator::operator*): Return
6164 references to key and value.
6166 2018-07-10 Jakub Jelinek <jakub@redhat.com>
6169 * gimplify.c (find_combined_omp_for): Add DATA argument, in addition
6170 to finding the inner OMP_FOR/OMP_SIMD stmt find non-trivial wrappers,
6171 BLOCKs with BLOCK_VARs, OMP_PARALLEL in between, OMP_FOR in between.
6172 (gimplify_omp_for): For composite loops, move outer
6173 OMP_{DISTRIBUTE,TASKLOOP,FOR,PARALLEL} right around innermost
6174 OMP_FOR/OMP_SIMD if there are any non-trivial wrappers. For class
6175 iterators add any needed clauses. Allow OMP_FOR_ORIG_DECLS to contain
6176 TREE_LIST for both the original class iterator and the "last" helper
6177 var. Gimplify OMP_FOR_PRE_BODY before the outermost composite
6178 loop, remember has_decl_expr from outer composite loops for the
6179 innermost OMP_SIMD in TREE_PRIVATE bit on OMP_FOR_INIT.
6181 2018-07-09 Martin Sebor <msebor@redhat.com>
6185 * builtins.c (c_strlen): Avoid out-of-bounds warnings when
6186 accessing implicitly initialized array elements.
6187 * expr.c (string_constant): Handle string initializers of
6188 character arrays within aggregates.
6189 * gimple-fold.c (fold_array_ctor_reference): Add argument.
6190 Store element offset. As a special case, handle zero size.
6191 (fold_nonarray_ctor_reference): Same.
6192 (fold_ctor_reference): Add argument. Store subobject offset.
6193 * gimple-fold.h (fold_ctor_reference): Add argument.
6195 2018-07-09 Paul Koning <ni1d@arrl.net>
6197 * config/pdp11/pdp11.c (pdp11_addr_cost): New function.
6198 (pdp11_insn_cost): New function.
6199 (pdp11_md_asm_adjust): New function.
6200 (TARGET_INVALID_WITHIN_DOLOOP): Define.
6201 (pdp11_rtx_costs): Update to match machine better.
6202 (output_addr_const_pdp11): Correct format mismatch warnings.
6203 * config/pdp11/pdp11.h (SLOW_BYTE_ACCESS): Correct definition.
6204 * config/pdp11/pdp11.md: General change to add base_cost and/or
6205 length attributes for use by new pdp11_insn_cost function.
6206 (MIN_BRANCH): Correct definition.
6208 (doloop_end): Use standard pattern name for looping pattern.
6209 (doloop_end_nocc): New.
6210 (movsf): Add another constraint alternative.
6211 (zero_extendqihi2): Add constraint alternatives for not in place
6213 (zero_extendhisi2): Remove.
6214 (shift patterns): Add CC handling variants.
6218 (define_peephole2): New peephole to recognize mov that sets CC for
6221 2018-07-09 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
6223 * config/sparc/sparc.c (sparc_fold_builtin) <SPARC_BUILTIN_PDIST,
6224 SPARC_BUILTIN_PDISTN>: Adapt for signature change of wi::neg,
6227 2018-07-09 Jakub Jelinek <jakub@redhat.com>
6230 * real.c (real_nextafter): Return true if result is denormal.
6232 2018-07-09 Martin Liska <mliska@suse.cz>
6234 * common.opt: Add back wrongly removed attribute.
6236 2018-07-09 Richard Biener <rguenther@suse.de>
6239 * dwarf2out.c (gen_block_die): For an early generated DIE
6240 always output high/low PC attributes.
6242 2018-07-09 Tom de Vries <tdevries@suse.de>
6244 * cfgexpand.c (expand_debug_source_expr): Handle VAR_DECL.
6245 * tree-inline.c (remap_ssa_name): Handle default def ssa_name mapping
6246 onto VAR_DECL with abstract origin.
6248 2018-07-07 Jim Wilson <jimw@sifive.com>
6250 * config/riscv/riscv.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): New.
6252 2018-07-07 Tom de Vries <tdevries@suse.de>
6254 * tree-dfa.c (dump_enumerated_decls): Handle cfun->cfg == NULL.
6256 2018-07-07 Aldy Hernandez <aldyh@redhat.com>
6258 * tree-vrp.c (vrp_int_const_binop): Change overflow type to
6260 (combine_bound): Use wide-int overflow calculation instead of
6262 * calls.c (maybe_warn_alloc_args_overflow): Change overflow type to
6264 * fold-const.c (int_const_binop_2): Same.
6265 (extract_muldiv_1): Same.
6266 (fold_div_compare): Same.
6267 (fold_abs_const): Same.
6269 * poly-int.h (add): Same.
6273 * predict.c (predict_iv_comparison): Same.
6274 * profile-count.c (slow_safe_scale_64bit): Same.
6275 * simplify-rtx.c (simplify_const_binary_operation): Same.
6276 * tree-chrec.c (tree_fold_binomial): Same.
6277 * tree-data-ref.c (split_constant_offset_1): Same.
6278 * tree-if-conv.c (idx_within_array_bound): Same.
6279 * tree-scalar-evolution.c (iv_can_overflow_p): Same.
6280 * tree-ssa-phiopt.c (minmax_replacement): Same.
6281 * tree-vect-loop.c (is_nonwrapping_integer_induction): Same.
6282 * tree-vect-stmts.c (vect_truncate_gather_scatter_offset): Same.
6283 * vr-values.c (vr_values::adjust_range_with_scev): Same.
6284 * wide-int.cc (wi::add_large): Same.
6285 (wi::mul_internal): Same.
6286 (wi::sub_large): Same.
6287 (wi::divmod_internal): Same.
6288 * wide-int.h: Change overflow type to overflow_type for neg, add,
6289 mul, smul, umul, div_trunc, div_floor, div_ceil, div_round,
6290 mod_trunc, mod_ceil, mod_round, add_large, sub_large,
6291 mul_internal, divmod_internal.
6292 (overflow_type): New enum.
6293 (accumulate_overflow): New.
6295 2018-07-06 Kugan Vivekanandarajah <kugan.vivekanandarajah@linaro.org>
6297 * tree-ssa-phiopt.c (cond_removal_in_popcount_pattern): New.
6298 (tree_ssa_phiopt_worker): Call cond_removal_in_popcount_pattern.
6300 2018-07-06 Kugan Vivekanandarajah <kuganv@linaro.org>
6302 * tree-ssa-loop-niter.c (number_of_iterations_popcount): If popcount
6303 argument is checked for zero before entering loop, avoid checking again.
6305 2018-07-06 Kugan Vivekanandarajah <kuganv@linaro.org>
6307 * gimplify.h (generic_expr_could_trap_p): Set as global function.
6308 * gimplify.h (generic_expr_could_trap_p): Likwise.
6309 * tree-scalar-evolution.c (expression_expensive_p): Handle COND_EXPR.
6311 2018-07-06 Jakub Jelinek <jakub@redhat.com>
6313 PR tree-optimization/86401
6314 * fold-const.c (fold_binary_loc) <case BIT_AND_EXPR>: Move the
6315 ((A & N) + B) & M -> (A + B) & M etc. optimization into ...
6316 (fold_bit_and_mask): ... here. New helper function for match.pd.
6317 * fold-const.h (fold_bit_and_mask): Declare.
6318 * match.pd (((A & N) + B) & M -> (A + B) & M): New optimization.
6320 2018-07-06 Peter Bergner <bergner@linux.ibm.com>
6323 * target.def (translate_mode_attribute): New hook.
6324 * targhooks.h (default_translate_mode_attribute): Declare.
6325 * targhooks.c (default_translate_mode_attribute): New function.
6326 * doc/tm.texi.in (TARGET_TRANSLATE_MODE_ATTRIBUTE): New hook.
6327 * doc/tm.texi: Regenerate.
6328 * config/rs6000/rs6000.c (TARGET_TRANSLATE_MODE_ATTRIBUTE): Define.
6329 (rs6000_translate_mode_attribute): New function.
6331 2018-07-06 Paul Koning <ni1d@arrl.net>
6333 * doc/md.texi (define_split): Document DONE and FAIL.
6334 (define_peephole2): Ditto.
6336 2018-07-05 Jeff Law <law@redhat.com>
6338 PR tree-optimization/86010
6339 * tree-ssa-dse.c (compute_trims): More aggressively trim at
6340 both the head and tail of mem* and str* calls.
6342 2018-07-05 Jim Wilson <jimw@sifive.com>
6344 * config.gcc (riscv*-*-*): When setting xlen, handle riscv-*.
6346 2018-07-05 Indu Bhagat <indu.bhagat@oracle.com>
6348 * config/aarch64/aarch64-simd.md: correct flags text for
6349 MIN_EXPR replacement.
6351 2018-07-05 James Clarke <jrtc27@jrtc27.com>
6353 * configure: Regenerated.
6355 2018-07-05 Carl Love <cel@us.ibm.com>
6357 * config/rs6000/rs6000-c.c: Map ALTIVEC_BUILTIN_VEC_UNPACKH for
6358 float argument to VSX_BUILTIN_DOUBLEH_V4SF.
6359 Map ALTIVEC_BUILTIN_VEC_UNPACKL for float argument to
6360 VSX_BUILTIN_DOUBLEL_V4SF.
6362 2018-07-05 Martin Sebor <msebor@redhat.com>
6365 * tree-ssa-strlen.c (maybe_set_strlen_range): Use type size rather
6366 than its domain to compute its the upper bound of a char array.
6368 2018-07-05 Nathan Sidwell <nathan@acm.org>
6370 Replace NO_IMPLICIT_EXTERN_C with SYSTEM_IMPLICIT_EXTERN_C.
6371 * doc/cpp.texi: Update comment.
6372 * doc/tm.texi: Rebuilt.
6373 * doc/tm.texi.in (NO_IMPLICIT_EXTERN_C): Replace with ...
6374 (SYSTEM_IMPLICIT_EXTERN_C): ... this, opposite sense.
6375 * doc/extend.texi (Backwards Compatibility): Clarify it is system
6376 headers affected by extern "C".
6377 * system.h: Poison NO_IMPLICIT_EXTERN_C.
6378 * config/alpha/alpha.h, config/arm/uclinux-elf.h,
6379 config/bfin/elf.h, config/cris/cris.h, config/darwin.h,
6380 config/dragonfly.h, config/freebsd.h, config/gnu-user.h,
6381 config/i386/cygming.h, config/i386/djgpp.h, config/i386/nto.h,
6382 config/ia64/hpux.h, config/lm32/lm32.h, config/lm32/uclinux-elf.h,
6383 config/lynx.h, config/mips/elf.h, config/mmix/mmix.h,
6384 config/netbsd.h, config/pa/pa-hpux.h, config/powerpcspe/sysv4.h,
6385 config/riscv/elf.h, config/rs6000/sysv4.h, config/rtems.h,
6386 config/s390/tpf.h, config/sh/newlib.h, config/sol2.h,
6387 config/sparc/openbsd64.h, config/sparc/sp-elf.h,
6388 config/sparc/sp64-elf.h, config/spu/spu.h,
6389 config/stormy16/stormy16.h, config/v850/v850.h,
6390 config/visium/visium.h, config/vx-common.h, config/xtensa/elf.h: Don't
6391 define NO_IMPLICIT_EXTERN_C.
6392 * config/rs6000/aix.h: Set SYSTEM_IMPLICIT_EXTERN_C.
6394 2018-07-05 Tamar Christina <tamar.christina@arm.com>
6397 * config/arm/arm.c (arm_can_change_mode_class): Use GET_MODE_UNIT_SIZE
6398 instead of GET_MODE_SIZE when comparing Units.
6400 2018-07-05 Tamar Christina <tamar.christina@arm.com>
6403 * rtlanal.c (set_noop_p): Constrain on mode change,
6404 include hard-reg-set.h
6406 2018-07-05 Tamar Christina <tamar.christina@arm.com>
6408 * config/aarch64/aarch64.c (aarch64_expand_movmem): Fix mode size.
6410 2018-07-05 Jakub Jelinek <jakub@redhat.com>
6413 2018-07-04 Maxim Ostapenko <m.ostapenko@samsung.com>
6416 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Pass -lstdc++ for static
6418 * gcc.c: Do not pass LIBUBSAN_SPEC if ASan is enabled with UBSan.
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 Eric Botcazou <ebotcazou@adacore.com>
6430 * expmed.c (choose_multiplier): Fix incorrect comparison with mask.
6432 2018-07-04 Aldy Hernandez <aldyh@redhat.com>
6434 * tree-vrp.c (extract_range_from_binary_expr_1): Initialize
6435 neg_*_op* variables.
6437 2018-07-04 Martin Liska <mliska@suse.cz>
6439 * tree-switch-conversion.c: Define
6440 max_ratio_for_speed and max_ratio_for_size constants.
6442 2018-07-04 Denys Vlasenko <dvlasenk@redhat.com>
6443 Martin Liska <mliska@suse.cz>
6448 * common.opt: Rename align options with 'str_' prefix.
6449 * common/config/i386/i386-common.c (set_malign_value): New
6451 (ix86_handle_option): Use it to set -falign-* options/
6452 * config/aarch64/aarch64-protos.h (struct tune_params): Change
6453 type from int to string.
6454 * config/aarch64/aarch64.c: Update default values from int
6456 * config/alpha/alpha.c (alpha_override_options_after_change):
6458 * config/arm/arm.c (arm_override_options_after_change_1): Likewise.
6459 * config/i386/dragonfly.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
6460 max skip conditionally.
6461 * config/i386/freebsd.h (SUBALIGN_LOG): New.
6462 (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
6463 max skip conditionally.
6464 * config/i386/gas.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
6465 max skip conditionally.
6466 * config/i386/gnu-user.h (SUBALIGN_LOG): New.
6467 (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
6468 max skip conditionally.
6469 * config/i386/i386.c (struct ptt): Change type from int to
6471 (ix86_default_align): Set default values.
6472 * config/i386/i386.h (ASM_OUTPUT_MAX_SKIP_PAD): Print
6473 max skip conditionally.
6474 * config/i386/iamcu.h (SUBALIGN_LOG): New.
6475 (ASM_OUTPUT_MAX_SKIP_ALIGN):
6476 * config/i386/lynx.h (ASM_OUTPUT_MAX_SKIP_ALIGN):
6477 * config/i386/netbsd-elf.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
6478 max skip conditionally.
6479 * config/i386/openbsdelf.h (SUBALIGN_LOG): New.
6480 (ASM_OUTPUT_MAX_SKIP_ALIGN) Print max skip conditionally.:
6481 * config/i386/x86-64.h (SUBALIGN_LOG): New.
6482 (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
6483 max skip conditionally.
6484 (ASM_OUTPUT_MAX_SKIP_PAD): Likewise.
6485 * config/ia64/ia64.c (ia64_option_override): Set default values
6486 for alignment options.
6487 * config/m68k/m68k.c: Handle new str_align_* options.
6488 * config/mips/mips.c (mips_set_compression_mode): Change
6490 (mips_option_override): Set default values for options.
6491 * config/powerpcspe/powerpcspe.c (rs6000_option_override_internal):
6493 * config/rs6000/rs6000.c (rs6000_option_override_internal):
6495 * config/rx/rx.c (rx_option_override): Likewise.
6496 * config/rx/rx.h (JUMP_ALIGN): Use align_jumps_log.
6497 (LABEL_ALIGN): Use align_labels_log.
6498 (LOOP_ALIGN): Use align_loops_align.
6499 * config/s390/s390.c (s390_asm_output_function_label): Use new
6501 * config/sh/sh.c (sh_override_options_after_change):
6502 Change type of constants.
6503 * config/spu/spu.c (spu_sched_init): Likewise.
6504 * config/sparc/sparc.c (sparc_option_override): Set default
6506 * config/visium/visium.c (visium_option_override): Likewise.
6507 * config/visium/visium.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Do not
6508 emit p2align format with last argument if it's not needed.
6509 * doc/invoke.texi: Document extended format of -falign-*.
6510 * final.c: Use align_labels alignment.
6511 * flags.h (struct target_flag_state): Change type to use
6513 (struct align_flags_tuple): New.
6514 (struct align_flags): Likewise.
6515 (align_loops_log): Redefine macro to use new types.
6516 (align_loops_max_skip): Redefine macro to use new types.
6517 (align_jumps_log): Redefine macro to use new types.
6518 (align_jumps_max_skip): Redefine macro to use new types.
6519 (align_labels_log): Redefine macro to use new types.
6520 (align_labels_max_skip): Redefine macro to use new types.
6521 (align_functions_log): Redefine macro to use new types.
6522 (align_loops): Redefine macro to use new types.
6523 (align_jumps): Redefine macro to use new types.
6524 (align_labels): Redefine macro to use new types.
6525 (align_functions): Redefine macro to use new types.
6526 (align_functions_max_skip): Redefine macro to use new types.
6527 (align_loops_value): New macro.
6528 (align_jumps_value): New macro.
6529 (align_labels_value): New macro.
6530 (align_functions_value): New macro.
6531 * function.c (invoke_set_current_function_hook): Propagate
6532 alignment values from flags to global variables default in
6534 * ipa-icf.c (sem_function::equals_wpa): Use
6535 cl_optimization_option_eq instead of memcmp.
6536 * lto-streamer.h (cl_optimization_stream_out): Support streaming
6538 (cl_optimization_stream_in): Likewise.
6539 * optc-save-gen.awk: Support strings in cl_optimization.
6540 * opth-gen.awk: Likewise.
6541 * opts.c (finish_options): Remove error checking of invalid
6543 (MAX_CODE_ALIGN): Remove.
6544 (MAX_CODE_ALIGN_VALUE): Likewise.
6545 (parse_and_check_align_values): New function.
6546 (check_alignment_argument): Likewise.
6547 (common_handle_option): Use check_alignment_argument.
6548 * opts.h (parse_and_check_align_values): Declare.
6549 * toplev.c (init_alignments): Remove.
6550 (read_log_maxskip): New.
6551 (parse_N_M): Likewise.
6552 (parse_alignment_opts): Likewise.
6553 (backend_init_target): Remove usage of init_alignments.
6554 * toplev.h (parse_alignment_opts): Declare.
6555 * tree-streamer-in.c (streamer_read_tree_bitfields): Add new
6557 * tree-streamer-out.c (streamer_write_tree_bitfields): Likewise.
6558 * tree.c (cl_option_hasher::equal): New.
6559 * varasm.c: Use new global macros.
6561 2018-07-04 Denys Vlasenko <dvlasenk@redhat.com>
6563 * config/i386/dragonfly.h: (ASM_OUTPUT_MAX_SKIP_ALIGN):
6564 Use a simpler align directive also if MAXSKIP = ALIGN-1.
6565 * config/i386/gas.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
6566 * config/i386/lynx.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
6567 * config/i386/netbsd-elf.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
6568 * config/i386/i386.h (ASM_OUTPUT_MAX_SKIP_PAD): Likewise.
6569 * config/i386/freebsd.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Remove "If N
6570 is large, do at least 8 byte alignment" code. Add SUBALIGN_LOG
6571 define. Use a simpler align directive also if MAXSKIP = ALIGN-1.
6572 * config/i386/gnu-user.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
6573 * config/i386/iamcu.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
6574 * config/i386/openbsdelf.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
6575 * config/i386/x86-64.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
6577 2018-07-04 Martin Liska <mliska@suse.cz>
6578 Jonathan Wakely <jwakely@redhat.com>
6580 * coverage.c: Use correct type.
6581 * doc/invoke.texi: Language correction.
6583 2018-07-03 H.J. Lu <hongjiu.lu@intel.com>
6586 * config/i386/i386.c (rest_of_insert_endbranch): Also generate
6587 ENDBRANCH for non-tail call which may return via indirect branch.
6588 * doc/extend.texi: Document indirect_return attribute.
6590 2018-07-03 Martin Sebor <msebor@redhat.com>
6592 PR tree-optimization/86274
6593 * gimple-ssa-sprintf.c (fmtresult::type_max_digits): Verify
6595 (format_floating): Correct handling of infinities and NaNs.
6597 2018-07-03 Martin Sebor <msebor@redhat.com>
6599 * print-tree.c (print_real_cst): New function.
6600 (print_node_brief): Call it.
6601 (print_node): Ditto.
6603 2018-07-03 Jeff Law <law@redhat.com>
6605 * config/h8300/h8300.md (logical<mode>3_sn, logical<mode>3): Merge
6606 into a single pattern.
6608 * config/h8300/h8300.md (ors code_iterator): New.
6609 (bsetqi_msx, bnotqi_msx patterns and splitters): Consolidate into
6610 a single pattern and single splitter.
6611 (bsethi_msx, bnothi_msx patterns): Consolidate into a single pattern.
6612 (iorqi3_1, xorqi3_1): Likewise.
6613 (iorqi3, xorqi3 expanders): Similarly.
6615 * config/h8300/h8300.md (movmd_internal_normal): Consolidated with
6616 (movmd_internal) into a single pattern using the P mode iterator.
6617 (movmd splitters): Similarly.
6618 (stpcpy_internal_normal, stpcpy_internal): Similarly for thes patterns.
6619 (movsd splitters): Similarly.
6621 * config/h8300/h8300.c (h8300_insn_length_from_table): Consolidate
6622 ADDB, ADDW and ADDL into a single ADD attribute which selects the
6623 right table based on the size of the operand.
6624 * config/h8300/h8300.md (length_table): Corresponding changes. All
6625 references to "addb", "addw" and "addl" changed to "add".
6626 (btst patterns): Merge two variants into a single pattern.
6627 (tstqi, tsthi): Likewise.
6628 (addhi3_incdec, addsi3_incdec): Likewise.
6629 (subhi3_h8300hs, subsi3_h8300hs): Likewise.
6630 (mulhi3, mulsi3): Likewise.
6631 (udivhi3, udivsi3): Likewise.
6632 (divhi3, divsi3): Likewise.
6633 (andorqi3, andorhi3, andorsi3): Likewise.
6635 2018-07-03 Uros Bizjak <ubizjak@gmail.com>
6638 * config/i386/sse.md (uavg<mode>3_ceil): New expander.
6639 (<sse2_avx2>_uavg<mode>3<mask_name>): Simplify expander.
6641 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
6643 PR tree-optimization/85694
6644 * config/aarch64/iterators.md (HADD, RHADD): New int iterators.
6645 (u): Handle UNSPEC_SHADD, UNSPEC_UHADD, UNSPEC_SRHADD and
6647 * config/aarch64/aarch64-simd.md (<u>avg<mode>3_floor)
6648 (<u>avg<mode>3_ceil): New patterns.
6650 2018-07-03 David Malcolm <dmalcolm@redhat.com>
6652 * gcc.dg/vect/slp-perm-1.c: Remove "note: " prefix from
6653 scan-tree-dump directive.
6654 * gcc.dg/vect/slp-perm-2.c: Likewise.
6655 * gcc.dg/vect/slp-perm-3.c: Likewise.
6656 * gcc.dg/vect/slp-perm-5.c: Likewise.
6657 * gcc.dg/vect/slp-perm-6.c: Likewise.
6658 * gcc.dg/vect/slp-perm-7.c: Likewise.
6659 * gcc.dg/vect/slp-perm-8.c: Likewise.
6661 2018-07-03 Marek Polacek <polacek@redhat.com>
6664 * gimple-fold.c (size_must_be_zero_p): Check the type of the size.
6666 2018-07-03 Richard Biener <rguenther@suse.de>
6669 * tree-ssa-structalias.c (find_func_clobbers): Properly
6670 handle indirect calls.
6672 2018-07-03 Jeff Law <law@redhat.com>
6674 * config/h8300/h8300.md (HSI, QHSI, QHSIF): New mode iterators.
6675 (shifts): New code iterator.
6676 (movqi, movhi, movsi, movsf expanders): Consolidate into a single
6677 expander. Fix HImode handling on H8/SX.
6678 (addqi3, addhi3, addsi3 expanders): Consolidate into a single expander.
6679 (subqi3, subhi3, subsi3 expanders): Likewise.
6680 (andqi3, andhi3, andsi3 expanders): Likewise.
6681 (iorqi3, iorhi3, iorsi3 expanders): Likewise.
6682 (xorqi3, xorhi3, xorsi3 expanders): Likewise.
6683 (negqi2, neghi2, negsi2, negsf2 expanders): Likewise.
6684 (one_cmplqi2, one_cmplhi2, one_cmplsi2): Likewise.
6685 (zero_extendqihi2, zero_extendqisi2): Likewise.
6686 (extendqihi2, extendqisi2): Likewise.
6687 (rotlqi3, rotlhi3, rotlsi3): Likewise.
6688 (neghi2_h8300, negsi2_h8300): Likewise for these patterns.
6689 (rotlqi3_1, rotlhi3_1): Likewise.
6690 (logicalhi3_sn, logicalsi3_sn): Likewise.
6691 (logicalhi3, logicalsi3): Likewise.
6693 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
6695 * tree-vect-patterns.c (vect_recog_rotate_pattern)
6696 (vect_recog_vector_vector_shift_pattern, vect_recog_divmod_pattern)
6697 (vect_recog_mixed_size_cond_pattern, adjust_bool_pattern_cast)
6698 (adjust_bool_pattern, vect_recog_bool_pattern): Pass the vector
6699 type to append_pattern_def_seq instead of creating a stmt_vec_info
6701 (build_mask_conversion): Likewise. Remove vinfo argument.
6702 (vect_add_conversion_to_patterm): Likewise, renaming to...
6703 (vect_add_conversion_to_pattern): ...this.
6704 (vect_recog_mask_conversion_pattern): Update call to
6705 build_mask_conversion. Pass the vector type to
6706 append_pattern_def_seq here too.
6707 (vect_recog_gather_scatter_pattern): Update call to
6708 vect_add_conversion_to_pattern.
6710 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
6712 * tree-vect-patterns.c (new_pattern_def_seq): Delete.
6713 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern)
6714 (vect_recog_widen_op_pattern, vect_recog_over_widening_pattern)
6715 (vect_recog_rotate_pattern, vect_synth_mult_by_constant): Don't set
6716 STMT_VINFO_PATTERN_DEF_SEQ to null here.
6717 (vect_recog_pow_pattern, vect_recog_vector_vector_shift_pattern)
6718 (vect_recog_mixed_size_cond_pattern, vect_recog_bool_pattern): Use
6719 append_pattern_def_seq instead of new_pattern_def_seq.
6720 (vect_recog_divmod_pattern): Do both of the above.
6721 (vect_pattern_recog_1): Assert that STMT_VINO_PATTERN_DEF_SEQ
6724 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
6726 * tree-vect-patterns.c (vect_recog_dot_prod_pattern):
6727 (vect_recog_sad_pattern, vect_recog_widen_op_pattern)
6728 (vect_recog_widen_mult_pattern, vect_recog_pow_pattern):
6729 (vect_recog_widen_sum_pattern, vect_recog_over_widening_pattern)
6730 (vect_recog_average_pattern, vect_recog_cast_forwprop_pattern)
6731 (vect_recog_widen_shift_pattern, vect_recog_rotate_pattern)
6732 (vect_recog_vector_vector_shift_pattern, vect_synth_mult_by_constant)
6733 (vect_recog_mult_pattern, vect_recog_divmod_pattern)
6734 (vect_recog_mixed_size_cond_pattern, vect_recog_bool_pattern)
6735 (vect_recog_mask_conversion_pattern): Replace vec<gimple *>
6736 parameter with a single stmt_vec_info.
6737 (vect_recog_func_ptr): Likewise.
6738 (vect_recog_gather_scatter_pattern): Likewise, folding in...
6739 (vect_try_gather_scatter_pattern): ...this.
6740 (vect_pattern_recog_1): Remove stmts_to_replace and just pass
6741 the stmt_vec_info of the statement to be matched. Don't clear
6742 STMT_VINFO_RELATED_STMT.
6743 (vect_pattern_recog): Update call accordingly.
6745 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
6747 PR tree-optimization/85694
6748 * doc/md.texi (avgM3_floor, uavgM3_floor, avgM3_ceil)
6749 (uavgM3_ceil): Document new optabs.
6750 * doc/sourcebuild.texi (vect_avg_qi): Document new target selector.
6751 * internal-fn.def (IFN_AVG_FLOOR, IFN_AVG_CEIL): New internal
6753 * optabs.def (savg_floor_optab, uavg_floor_optab, savg_ceil_optab)
6754 (savg_ceil_optab): New optabs.
6755 * tree-vect-patterns.c (vect_recog_average_pattern): New function.
6756 (vect_vect_recog_func_ptrs): Add it.
6757 * tree-vect-stmts.c (vectorizable_call): Get the type of the zero
6758 constant directly from the associated lhs.
6760 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
6762 * tree-vect-patterns.c (vect_split_statement): New function.
6763 (vect_convert_input): Use it to try to split an existing cast.
6765 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
6767 * poly-int.h (print_hex): New function.
6768 * dumpfile.h (dump_dec, dump_hex): Declare.
6769 * dumpfile.c (dump_dec, dump_hex): New poly_wide_int functions.
6770 * tree-vectorizer.h (_stmt_vec_info): Add min_output_precision,
6771 min_input_precision, operation_precision and operation_sign.
6772 * tree-vect-patterns.c (vect_get_range_info): New function.
6773 (vect_same_loop_or_bb_p, vect_single_imm_use)
6774 (vect_operation_fits_smaller_type): Delete.
6775 (vect_look_through_possible_promotion): Add an optional
6776 single_use_p parameter.
6777 (vect_recog_over_widening_pattern): Rewrite to use new
6778 stmt_vec_info infomration. Handle one operation at a time.
6779 (vect_recog_cast_forwprop_pattern, vect_narrowable_type_p)
6780 (vect_truncatable_operation_p, vect_set_operation_type)
6781 (vect_set_min_input_precision): New functions.
6782 (vect_determine_min_output_precision_1): Likewise.
6783 (vect_determine_min_output_precision): Likewise.
6784 (vect_determine_precisions_from_range): Likewise.
6785 (vect_determine_precisions_from_users): Likewise.
6786 (vect_determine_stmt_precisions, vect_determine_precisions): Likewise.
6787 (vect_vect_recog_func_ptrs): Put over_widening first.
6789 (vect_pattern_recog): Call vect_determine_precisions.
6791 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
6793 * tree-vect-patterns.c (vect_mark_pattern_stmts): Remove pattern
6794 statements that have been replaced by further pattern statements.
6795 (vect_pattern_recog_1): Clear STMT_VINFO_PATTERN_DEF_SEQ on failure.
6797 2018-07-03 Richard Biener <rguenther@suse.de>
6799 * tree-vect-stmts.c (vect_is_simple_use): Consolidate dumping,
6800 always set *dt. Dump vectype in vectype overload.
6801 * dumpfile.h (dump_gimple_expr): New function.
6802 (dump_gimple_expr_loc): Likewise.
6803 * dumpfile.c (dump_gimple_expr): New function.
6804 (dump_gimple_expr_loc): Likewise.
6806 2018-07-02 Jeff Law <law@redhat.com>
6808 * config/h8300/h8300.md (movqi_h8300, movqi_h8300hs): Consolidate
6809 the H8/300, H8/300H and H8/S variants into a single pattern.
6810 (movhi_h8300, movqi_h8300hs): Similarly.
6811 (pushqi_h8300hs, pushhi_h8300hs): Consolidate into a single pattern.
6812 (QHI mode iterator): New.
6814 * config/h8300/h8300.md: Remove trailing whitespace.
6816 2018-07-02 Jim Wilson <jimw@sifive.com>
6818 * config/riscv/riscv.c (riscv_expand_epilogue): Use emit_jump_insn
6819 instead of emit_insn for interrupt returns.
6820 * config/riscv/riscv.md (riscv_met): Add (return) to rtl.
6821 (riscv_sret, riscv_uret): Likewise.
6823 2018-07-02 David Malcolm <dmalcolm@redhat.com>
6825 * pretty-print.c (selftest::test_pp_format): Move save and restore
6826 of quotes to class auto_fix_quotes, and add an instance.
6827 * selftest.c: Include "intl.h".
6828 (selftest::auto_fix_quotes::auto_fix_quotes): New ctor.
6829 (selftest::auto_fix_quotes::~auto_fix_quotes): New dtor.
6830 * selftest.h (selftest::auto_fix_quotes): New class.
6832 2018-07-02 Richard Henderson <richard.henderson@linaro.org>
6834 * config/aarch64/aarch64-protos.h, config/aarch64/aarch64.c
6835 (aarch64_sve_prepare_conditional_op): Remove.
6836 * config/aarch64/aarch64-sve.md (cond_<SVE_INT_BINARY><SVE_I>):
6837 Allow aarch64_simd_reg_or_zero as select operand; remove
6838 the aarch64_sve_prepare_conditional_op call.
6839 (cond_<SVE_INT_BINARY_SD><SVE_SDI>): Likewise.
6840 (cond_<SVE_COND_FP_BINARY><SVE_F>): Likewise.
6841 (*cond_<SVE_INT_BINARY><SVE_I>_z): New pattern.
6842 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_z): New pattern.
6843 (*cond_<SVE_COND_FP_BINARY><SVE_F>_z): New pattern.
6844 (*cond_<SVE_INT_BINARY><SVE_I>_any): New pattern.
6845 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_any): New pattern.
6846 (*cond_<SVE_COND_FP_BINARY><SVE_F>_any): New pattern
6847 and a splitters to match all of the *_any patterns.
6848 * config/aarch64/predicates.md (aarch64_sve_any_binary_operator): New.
6850 * config/aarch64/iterators.md (SVE_INT_BINARY_REV): Remove.
6851 (SVE_COND_FP_BINARY_REV): Remove.
6852 (sve_int_op_rev, sve_fp_op_rev): New.
6853 * config/aarch64/aarch64-sve.md (*cond_<SVE_INT_BINARY><SVE_I>_0): New.
6854 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_0): New.
6855 (*cond_<SVE_COND_FP_BINARY><SVE_F>_0): New.
6856 (*cond_<SVE_INT_BINARY><SVE_I>_2): Rename, add movprfx alternative.
6857 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_2): Similarly.
6858 (*cond_<SVE_COND_FP_BINARY><SVE_F>_2): Similarly.
6859 (*cond_<SVE_INT_BINARY><SVE_I>_3): Similarly; use sve_int_op_rev.
6860 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_3): Similarly.
6861 (*cond_<SVE_COND_FP_BINARY><SVE_F>_3): Similarly; use sve_fp_op_rev.
6863 * config/aarch64/aarch64-sve.md (cond_<SVE_COND_FP_BINARY><SVE_F>):
6864 Remove match_dup 1 from the inner unspec.
6865 (*cond_<SVE_COND_FP_BINARY><SVE_F>): Likewise.
6867 * config/aarch64/aarch64.md (movprfx): New attr.
6868 (length): Default movprfx to 8.
6869 * config/aarch64/aarch64-sve.md (*mul<SVE_I>3): Add movprfx alt.
6870 (*madd<SVE_I>, *msub<SVE_I): Likewise.
6871 (*<su>mul<SVE_I>3_highpart): Likewise.
6872 (*<SVE_INT_BINARY_SD><SVE_SDI>3): Likewise.
6873 (*v<ASHIFT><SVE_I>3): Likewise.
6874 (*<su><MAXMIN><SVE_I>3): Likewise.
6875 (*<su><MAXMIN><SVE_F>3): Likewise.
6876 (*fma<SVE_F>4, *fnma<SVE_F>4): Likewise.
6877 (*fms<SVE_F>4, *fnms<SVE_F>4): Likewise.
6878 (*div<SVE_F>4): Likewise.
6880 2018-07-02 Richard Sandiford <richard.sandiford@arm.com>
6882 * tree-vect-patterns.c (vect_recog_widen_shift_pattern): Fix typo
6885 2018-07-02 Richard Biener <rguenther@suse.de>
6887 PR tree-optimization/86363
6888 * tree-ssa-sccvn.c (vn_reference_lookup_3): Check the
6889 memset argument refers to a non-variable address.
6891 2018-07-02 Aldy Hernandez <aldyh@redhat.com>
6893 * tree-vrp.c (extract_range_from_binary_expr_1): Abstract a lot of the
6894 {PLUS,MINUS}_EXPR code to...
6895 (adjust_symbolic_bound): ...here,
6896 (combine_bound): ...here,
6897 (set_value_range_with_overflow): ...and here.
6899 2018-07-02 Aldy Hernandez <aldyh@redhat.com>
6901 * tree-vrp.c (extract_range_from_unary_expr): Abstract ABS_EXPR
6903 (extract_range_from_abs_expr): ...here.
6905 2018-07-02 Eric Botcazou <ebotcazou@adacore.com>
6907 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Do not overrule
6908 -fno-omit-frame-pointer when not optimizing.
6910 2018-07-02 Martin Liska <mliska@suse.cz>
6913 * ipa-pure-const.c (malloc_candidate_p): Revert usage of ::get.
6914 (propagate_nothrow): Likewise.
6916 2018-07-02 Martin Liska <mliska@suse.cz>
6919 * ipa-inline.c (early_inliner): Revert wrongly added ::get call.
6921 2018-07-02 David Malcolm <dmalcolm@redhat.com>
6923 * dumpfile.c (dump_generic_expr_loc): Undo removal of this
6924 function in r262149, changing "loc" param from source_location to
6925 const dump_location_t &.
6926 * dumpfile.h (dump_generic_expr_loc): Undo removal of this
6927 declaration, as above.
6929 2018-07-01 Paul Koning <ni1d@arrl.net>
6931 * common/config/pdp11/pdp11-common.c (pdp11_handle_option): Handle
6932 -munit-asm, -mgnu-asm, -mdec-asm.
6933 * config/pdp11/pdp11-protos.h (pdp11_gen_int_label): New.
6934 (pdp11_output_labelref): New.
6935 (pdp11_output_def): New.
6936 (pdp11_output_addr_vec_elt): New.
6937 * config/pdp11/pdp11.c: Use tab between opcode and operands. Use
6938 %# and %@ format codes.
6939 (pdp11_option_override): New.
6940 (TARGET_ASM_FILE_START_FILE_DIRECTIVE): Define.
6941 (pdp11_output_ident): New.
6942 (pdp11_asm_named_section): New.
6943 (pdp11_asm_init_sections): New.
6944 (pdp11_file_start): New.
6945 (pdp11_file_end): New.
6946 (output_ascii): Use .ascii/.asciz for -mdec-asm.
6947 (pdp11_asm_print_operand): Update %# and %$ for -mdec-asm. Add
6948 %o, like %c but octal.
6949 (pdp11_option_override): New.
6950 * config/pdp11/pdp11.h (TEXT_SECTION_ASM_OP): Update for
6952 (DATA_SECTION_ASM_OP): Ditto.
6953 (READONLY_DATA_SECTION_ASM_OP): New.
6954 (IS_ASM_LOGICAL_LINE_SEPARATOR): New.
6955 (ASM_GENERATE_INTERNAL_LABEL): Use new function.
6956 (ASM_OUTPUT_LABELREF): Ditto.
6957 (ASM_OUTPUT_DEF): Ditto.
6958 (ASM_OUTPUT_EXTERNAL): New.
6959 (ASM_OUTPUT_SOURCE_FILENAME): New.
6960 (ASM_OUTPUT_ADDR_VEC_ELT): Use new function.
6961 (ASM_OUTPUT_SKIP): Update for -mdec-asm.
6962 * config/pdp11/pdp11.md: Use tab between opcode and operands. Use
6963 %# and %@ format codes.
6964 * config/pdp11/pdp11.opt (mgnu-asm): New.
6965 (mdec-asm): Conflicts with -mgnu-asm and -munix-asm.
6966 (munix-asm): Conflicts with -mdec-asm and -mgnu-asm.
6967 * doc/invoke.txt (PDP-11 Options): Add -mgnu-asm.
6969 2018-07-01 Aldy Hernandez <aldyh@redhat.com>
6971 * tree-ssa-threadupdate.c (mark_threaded_blocks): Avoid
6972 dereferencing path[] beyond its length.
6974 (debug_all_paths): New.
6975 (rewire_first_differing_edge): New.
6976 (adjust_paths_after_duplication): New.
6977 (duplicate_thread_path): Call adjust_paths_after_duplication.
6979 (thread_through_all_blocks): Add new argument to
6980 duplicate_thread_path.
6982 2018-06-30 Jim Wilson <jimw@sifive.com>
6984 * config/riscv/predicates.md (p2m1_shift_operand): New.
6985 (high_mask_shift_operand): New.
6986 * config/riscv/riscv.md (lshrsi3_zero_extend_3+1): New combiner
6987 pattern using p2m1_shift_operand.
6988 (lshsi3_zero_extend_3+2): New combiner pattern using
6989 high_mask_shift_operand.
6991 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
6993 * tree-vect-patterns.c (vect_get_external_def_edge): New function,
6995 (vect_recog_rotate_pattern): ...here.
6996 (vect_convert_input): Try to insert casts of invariants in the
6998 * tree-vect-loop-manip.c (vect_loop_versioning): Don't require the
6999 preheader to be empty.
7001 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
7003 * tree-vect-patterns.c (append_pattern_def_seq): Take an optional
7004 vector type. If given, install it in the new statement's
7006 (vect_element_precision): New function.
7007 (vect_unpromoted_value): New struct.
7008 (vect_unpromoted_value::vect_unpromoted_value): New function.
7009 (vect_unpromoted_value::set_op): Likewise.
7010 (vect_look_through_possible_promotion): Likewise.
7011 (vect_joust_widened_integer, vect_joust_widened_type): Likewise.
7012 (vect_widened_op_tree, vect_convert_input): Likewise.
7013 (vect_convert_inputs, vect_convert_output): Likewise.
7014 (vect_recog_dot_prod_pattern): Use vect_look_through_possible_promotion
7015 to handle the optional cast of the multiplication result and
7016 vect_widened_op_tree to detect the widened multiplication itself.
7017 Do not require the input and output of promotion casts to have
7018 the same sign, but base the signedness of the operation on the
7019 input rather than the result. If the pattern includes two
7020 promotions, check that those promotions have the same sign.
7021 Do not restrict the MULT_EXPR handling to a double-width result;
7022 handle quadruple-width results and wider. Use vect_convert_inputs
7023 to convert the inputs to the common type.
7024 (vect_recog_sad_pattern): Use vect_look_through_possible_promotion
7025 to handle the optional cast of the ABS result. Also allow a sign
7026 change or a sign extension between the ABS and MINUS.
7027 Use vect_widened_op_tree to detect the widened subtraction and use
7028 vect_convert_inputs to convert the inputs to the common type.
7029 (vect_handle_widen_op_by_const): Delete.
7030 (vect_recog_widen_op_pattern): New function.
7031 (vect_recog_widen_mult_pattern): Use it.
7032 (vect_recog_widen_shift_pattern): Likewise.
7033 (vect_recog_widen_sum_pattern): Use
7034 vect_look_through_possible_promotion to handle the promoted
7037 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
7039 * gimple-iterator.c (gsi_for_stmt): Add a new overload that takes
7040 the containing gimple_seq *.
7041 * gimple-iterator.h (gsi_for_stmt): Declare it.
7042 * tree-vect-patterns.c (vect_recog_dot_prod_pattern)
7043 (vect_recog_sad_pattern, vect_recog_widen_sum_pattern)
7044 (vect_recog_widen_shift_pattern, vect_recog_rotate_pattern)
7045 (vect_recog_vector_vector_shift_pattern, vect_recog_divmod_pattern)
7046 (vect_recog_mask_conversion_pattern): Remove STMT_VINFO_IN_PATTERN_P
7048 (vect_init_pattern_stmt, vect_set_pattern_stmt): New functions,
7050 (vect_mark_pattern_stmts): ...here. Handle cases in which the
7051 statement being replaced is part of an existing pattern
7052 definition sequence, inserting the new pattern statements before
7054 (vect_pattern_recog_1): Don't return a bool. If the statement
7055 is already part of a pattern, instead apply pattern matching
7056 to the pattern definition statements. Don't clear the
7057 STMT_VINFO_RELATED_STMT if is_pattern_stmt_p.
7058 (vect_pattern_recog): Don't break after the first match;
7059 continue processing the pattern definition statements instead.
7060 Don't bail out for STMT_VINFO_IN_PATTERN_P here.
7062 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
7064 * tree-vect-patterns.c (vect_reassociating_reduction_p): New function.
7065 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern)
7066 (vect_recog_widen_sum_pattern): Use it.
7068 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
7070 * tree-vect-loop.c (vectorizable_reduction): Assert that the
7071 phi is not a pattern statement and has not been replaced by
7072 a pattern statement.
7073 * tree-vect-patterns.c (type_conversion_p): Don't check
7074 STMT_VINFO_IN_PATTERN_P.
7075 (vect_recog_vector_vector_shift_pattern): Likewise.
7076 (vect_recog_dot_prod_pattern): Expect vect_is_simple_use to return
7077 the pattern statement rather than the original statement; check
7078 directly for a WIDEN_MULT_EXPR here.
7079 * tree-vect-slp.c (vect_get_and_check_slp_defs): Expect
7080 vect_is_simple_use to return the pattern statement rather
7081 than the original statement; use is_pattern_stmt_p to check
7082 for such a pattern statement.
7083 * tree-vect-stmts.c (process_use): Expect vect_is_simple_use
7084 to return the pattern statement rather than the original statement;
7085 don't do the same transformation here.
7086 (vect_is_simple_use): If the defining statement has been replaced
7087 by a pattern statement, return the pattern statement instead.
7088 Remove the corresponding (local) transformation from the vectype
7091 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
7093 * tree-vectorizer.h (vect_is_simple_use): Move the gimple ** to the
7094 end and default to null.
7095 * tree-vect-loop.c (vect_create_epilog_for_reduction)
7096 (vectorizable_reduction): Update calls accordingly, dropping the
7097 gimple ** argument if the passed-back statement isn't needed.
7098 * tree-vect-patterns.c (vect_get_internal_def, type_conversion_p)
7099 (vect_recog_rotate_pattern): Likewise.
7100 (vect_recog_mask_conversion_pattern): Likewise.
7101 * tree-vect-slp.c (vect_get_and_check_slp_defs): Likewise.
7102 (vect_mask_constant_operand_p): Likewise.
7103 * tree-vect-stmts.c (is_simple_and_all_uses_invariant, process_use):
7104 (vect_model_simple_cost, vect_get_vec_def_for_operand): Likewise.
7105 (get_group_load_store_type, get_load_store_type): Likewise.
7106 (vect_check_load_store_mask, vect_check_store_rhs): Likewise.
7107 (vectorizable_call, vectorizable_simd_clone_call): Likewise.
7108 (vectorizable_conversion, vectorizable_assignment): Likewise.
7109 (vectorizable_shift, vectorizable_operation): Likewise.
7110 (vectorizable_store, vect_is_simple_cond): Likewise.
7111 (vectorizable_condition, vectorizable_comparison): Likewise.
7112 (get_same_sized_vectype, vect_get_mask_type_for_stmt): Likewise.
7113 (vect_is_simple_use): Rename the def_stmt argument to def_stmt_out
7114 and move it to the end. Cope with null def_stmt_outs.
7116 2018-06-30 Bernd Edlinger <bernd.edlinger@hotmail.de>
7118 * Makefile.in (FLAGS_TO_PASS): Add RANLIB_FOR_TARGET.
7120 2018-06-29 Jeff Law <law@redhat.com>
7122 * config/v850/v850.c (v850_legitimate_address_p): Handle large
7123 displacements for TARGET_V850E2V3 and newer.
7124 (TARGET_LRA_P): Remove. Defaults to LRA now.
7125 * config/v850/v850.md (sign23byte_load): Remove.
7126 (unsign23byte_load, sign23hword_load, unsign23hword_load): Likewise.
7127 (23word_load, 23byte_store, 23hword_store, 23word_store): Likewise.
7129 2018-06-29 Martin Liska <mliska@suse.cz>
7132 * coverage.c (coverage_init): Mangle full path name.
7133 * doc/invoke.texi: Document the change.
7134 * gcov-io.c (mangle_path): New.
7135 * gcov-io.h (mangle_path): Likewise.
7136 * gcov.c (mangle_name): Use mangle_path for path mangling.
7138 2018-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7140 * config/arm/arm.c (output_move_double): Don't allow STRD instructions
7141 if starting source register is not even.
7143 2018-06-29 Martin Liska <mliska@suse.cz>
7145 PR tree-optimization/86263
7146 * tree-switch-conversion.c (switch_decision_tree::try_switch_expansion):
7147 Make edge redirection.
7149 2018-06-29 David Malcolm <dmalcolm@redhat.com>
7151 * dumpfile.c (dump_loc): Add indentation based on scope depth.
7152 (dump_scope_depth): New variable.
7153 (get_dump_scope_depth): New function.
7154 (dump_begin_scope): New function.
7155 (dump_end_scope): New function.
7156 * dumpfile.h (get_dump_scope_depth): New declaration.
7157 (dump_begin_scope): New declaration.
7158 (dump_end_scope): New declaration.
7159 (class auto_dump_scope): New class.
7160 (AUTO_DUMP_SCOPE): New macro.
7161 * tree-vectorizer.h (DUMP_VECT_SCOPE): Reimplement in terms of
7164 2018-06-29 Richard Biener <rguenther@suse.de>
7166 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Assert
7167 compute_all_dependences succeeds.
7168 * tree-vect-loop.c (vect_get_datarefs_in_loop): Fail early if we
7169 exceed --param loop-max-datarefs-for-datadeps.
7171 2018-06-29 Jakub Jelinek <jakub@redhat.com>
7173 * config/rs6000/t-rs6000: Append rs6000-modes.h to TM_H.
7175 2018-06-28 Uros Bizjak <ubizjak@gmail.com>
7178 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): Use
7179 alternative 0 in preferred_for_speed attribute.
7181 2018-06-28 Paul Koning <ni1d@arrl.net>
7183 * config/pdp11/pdp11-protos.h (pdp11_shift_length): New function.
7184 * config/pdp11/pdp11.c (pdp11_shift_length): New function.
7185 * config/pdp11/pdp11.h (ADJUST_INSN_LENGTH): Remove.
7186 * config/pdp11/pdp11.md: Correct "length" attribute calculation
7187 for shift insn patterns.
7189 2018-06-28 David Malcolm <dmalcolm@redhat.com>
7191 * cgraph.c (cgraph_node::get_body): Replace assignments to
7192 "dump_file" with calls to set_dump_file.
7193 * dumpfile.c (alt_dump_file): Make static, and group with...
7194 (alt_flags): ...this definition.
7195 (dumps_are_enabled): New variable.
7196 (refresh_dumps_are_enabled): New function.
7197 (set_dump_file): New function.
7198 (set_alt_dump_file): New function.
7199 (gcc::dump_manager::dump_start): Replace assignments to
7200 "dump_file" and "alt_dump_file" with calls to set_dump_file and
7202 (gcc::dump_manager::dump_finish): Likewise.
7203 * dumpfile.h (alt_dump_file): Delete decl.
7204 (dumps_are_enabled): New variable decl.
7205 (set_dump_file): New function decl.
7206 (dump_enabled_p): Rewrite in terms of new "dumps_are_enabled"
7208 * tree-nested.c (lower_nested_functions): Replace assignments to
7209 "dump_file" with calls to set_dump_file.
7211 2018-06-28 Eric Botcazou <ebotcazou@adacore.com>
7213 * tree-cfg.c (verify_gimple_in_cfg): Call verify_location on the
7214 goto_locus of each outgoing edge of each basic block.
7216 2018-06-28 Richard Biener <rguenther@suse.de>
7218 * dwarf2out.c (decl_scope_table): Remove.
7219 (push_decl_scope): Likewise.
7220 (pop_decl_scope): Likewise.
7221 (gen_type_die_for_member): Do not call push/pop_decl_scope.
7222 (gen_struct_or_union_type_die): Likewise.
7223 (gen_tagged_type_die): Likewise.
7224 (dwarf2out_init): Do not initialize decl_scope_table.
7225 (dwarf2out_c_finalize): Do not free it.
7227 2018-06-28 Richard Biener <rguenther@suse.de>
7229 * dwarf2out.c (gen_subprogram_die): Use is_unit_die when
7230 deciding whether to not re-use a DIE.
7232 2018-06-28 Richard Biener <rguenther@suse.de>
7234 * dwarf2out.c (gen_subprogram_die): Always re-use DIEs with an
7235 DW_AT_abstract_origin attribute.
7237 2018-06-28 Martin Liska <mliska@suse.cz>
7239 * tree-switch-conversion.c (jump_table_cluster::can_be_handled):
7240 Use newly introduced constants.
7241 * tree-switch-conversion.h (struct jump_table_cluster):
7242 Define max_ratio_for_size and max_ratio_for_speed.
7244 2018-06-28 Martin Liska <mliska@suse.cz>
7246 * tree-switch-conversion.c (jump_table_cluster::find_jump_tables):
7247 Add new checking assert to catch invalid state.
7248 (jump_table_cluster::can_be_handled): Handle single case
7250 (jump_table_cluster::is_beneficial): Bail out for such case.
7251 (bit_test_cluster::find_bit_tests):
7252 Add new checking assert to catch invalid state.
7253 (bit_test_cluster::can_be_handled): Handle single case
7255 (bit_test_cluster::is_beneficial): Bail out for such case.
7256 (switch_decision_tree::analyze_switch_statement):
7259 2018-06-28 Martin Liska <mliska@suse.cz>
7261 * common.opt: Introduce -completion option.
7262 * gcc.c (driver_handle_option): Handle it.
7263 (driver::main): Print completions if completion
7265 * opt-suggestions.c (option_proposer::get_completions):
7267 (option_proposer::suggest_completion): Likewise.
7268 (option_proposer::find_param_completions): Likewise.
7269 (verify_autocompletions): Likewise.
7270 (test_completion_valid_options): Likewise.
7271 (test_completion_valid_params): Likewise.
7272 (in_completion_p): Likewise.
7273 (empty_completion_p): Likewise.
7274 (test_completion_partial_match): Likewise.
7275 (test_completion_garbage): Likewise.
7276 (opt_proposer_c_tests): Likewise.
7277 * opt-suggestions.h: Declare new functions.
7278 * opts.c (common_handle_option): Handle OPT__completion_.
7279 * selftest-run-tests.c (selftest::run_tests): Add
7280 opt_proposer_c_tests.
7281 * selftest.c (assert_str_startswith): New.
7282 * selftest.h (assert_str_startswith): Likewise.
7283 (opt_proposer_c_tests): New.
7284 (ASSERT_STR_STARTSWITH): Likewise.
7286 2018-06-28 Martin Liska <mliska@suse.cz>
7288 * Makefile.in: Add opt-suggestions.o.
7289 * gcc-main.c: Include opt-suggestions.h.
7290 * gcc.c (driver::driver): Likewise.
7291 (driver::~driver): Remove m_option_suggestions.
7292 (driver::build_option_suggestions): Moved to option_proposer.
7293 (driver::suggest_option): Likewise.
7294 (driver::handle_unrecognized_options): Use option_proposer.
7295 * gcc.h (class driver): Add new memver m_option_proposer.
7296 * opt-suggestions.c: New file.
7297 * opt-suggestions.h: New file.
7299 2018-06-28 Martin Liska <mliska@suse.cz>
7301 * vec.h (class auto_string_vec): New (moved from auto_argvec).
7302 (auto_string_vec::~auto_string_vec): Likewise.
7304 2018-06-28 Eric Botcazou <ebotcazou@adacore.com>
7306 * tree-inline.h (struct copy_body_data): Move remapping_type_depth and
7307 prevent_decl_creation_for_types fields up and add reset_location field.
7308 * tree-inline.c (remap_gimple_stmt): Force input_location on the new
7309 statement if id->reset_location is true.
7310 (copy_edges_for_bb): Do not set goto_locus on the new edges if
7311 id->reset_location is true.
7312 (copy_phis_for_bb): Force input_location on the arguments if
7313 id->reset_location is true.
7314 (expand_call_inline): Set id->reset_location if DECL_IGNORED_P
7315 is set on the function to be inlined.
7317 2018-06-27 Stephan Bergmann <sbergman@redhat.com>
7319 * doc/invoke.texi (Debugging Options): Add -gsplit-dwarf.
7321 2018-06-27 Dimitar Dimitrov <dimitar@dinux.eu>
7323 * lra-eliminations.c (update_reg_eliminate): Mark all spanning hard
7324 registers for Pmode.
7325 * lra-lives.c (check_pseudos_live_through_calls): Mark all spanning
7326 hard registers for the clobbered pseudo.
7328 2018-06-27 Paul Koning <ni1d@arrl.net>
7330 * common/config/pdp11/pdp11-common.c (pdp11_handle_option): Handle
7331 mutually exclusive options.
7332 * config/pdp11/constraints.md (h): New constraint.
7333 (O): Update definition to match shift code generation.
7334 (D): New constraint.
7335 * config/pdp11/pdp11-modes.def (CCNZ): Define mode.
7337 * config/pdp11/pdp11-protos.h (int_no_side_effect_operand): New
7339 (output_jump): Change arguments.
7340 (pdp11_fixed_cc_regs): New function.
7341 (pdp11_cc_mode): Ditto.
7342 (pdp11_expand_shift): Ditto.
7343 (pdp11_assemble_shift): Ditto.
7344 (pdp11_small_shift): Ditto.
7345 (pdp11_branch_cost): Remove.
7346 * config/pdp11/pdp11.c (pdp11_assemble_integer): Remove comments
7348 (pdp11_register_move_cost): Update for CC registers.
7349 (pdp11_rtx_costs): Add case for LSHIFTRT.
7350 (pdp11_output_jump): Add CCNZ mode conditional branches.
7351 (notice_update_cc_on_set): Remove.
7352 (pdp11_cc_mode): New function.
7353 (simple_memory_operand): Correct pre/post decrement case.
7354 (no_side_effect_operand): New function.
7355 (pdp11_regno_reg_class): Add CC_REGS class.
7356 (pdp11_fixed_cc_regs): New function.
7357 (pdp11_small_shift): New function.
7358 (pdp11_expand_shift): New function to expand shift insns.
7359 (pdp11_assemble_shift): New function to output shifts.
7360 (pdp11_branch_cost): Remove.
7361 (pdp11_modes_tieable_p): Make QI/HI modes tieable.
7362 * config/pdp11/pdp11.h (SIZE_TYPE): Ensure 16-bit type.
7363 (WCHAR_TYPE): Ditto.
7364 (PTRDIFF_TYPE): Ditto.
7365 (ADJUST_INSN_LENGTH): New macro.
7366 (FIXED_REGISTERS): Add CC registers.
7367 (CALL_USED_REGISTERS): Ditto.
7369 (REG_CLASS_NAMES): Ditto.
7370 (REG_CLASS_CONTENTS): Ditto.
7371 (SELECT_CC_MODE): Use new function.
7372 (TARGET_FLAGS_REGNUM): New macro.
7373 (TARGET_FIXED_CONDITION_CODE_REGS): Ditto.
7374 (cc0_reg_rtx): Remove.
7375 (CC_STATUS_MDEP): Remove.
7376 (CC_STATUS_MDEFP_INIT): Remove.
7377 (CC_IN_FPU): Remove.
7378 (NOTICE_UPDATE_CC): Remove.
7379 (REGISTER_NAMES): Add CC registers.
7380 (BRANCH_COST): Change to constant 1.
7381 * config/pdp11/pdp11.md: Rewrite for CCmode condition code
7383 * config/pdp11/pdp11.opt (mbcopy): Remove.
7384 (mbcopy-builtin): Remove.
7385 (mbranch-cheap): Remove.
7386 (mbranch-expensive): Remove.
7387 * config/pdp11/predicates.md (expand_shift_operand): Update to
7388 match shift code generation.
7389 (ccnz_operator): New predicate.
7390 * doc/invoke.texi (PDP-11 Options): Remove deleted options
7391 -mbcopy, -mbcopy-builtin, -mbranch-cheap, -mbranch-expensive.
7392 Remove non-existent option -mabshi, -mno-abshi. Document mutually
7394 * doc/md.texi (PDP-11): Document new D and h constraints. Update
7395 description of O constraint.
7397 2018-06-27 Jeff Law <law@redhat.com>
7398 Austin Law <austinklaw@gmail.com>
7400 * config/v850/v850.md (addsi3_set_flags): New pattern.
7401 (subsi3_set_flags, negsi2_set_flags, andsi3_set_flags): Likewise.
7402 (iorsi3_set_flags, xorsi3_set_flags, one_cmplsi2_set_flags): Likewise.
7403 (zero_extendhisi2_v850_set_flags): Likewise.
7404 (zero_extendqisi2_v850_set_flags): Likewise.
7405 (ashlsi3_set_flags, ashlsi3_v850e2_set_flags): Likewise.
7406 (lshrsi3_set_flags, lshrsi3_v850e2_set_flags): Likewise.
7407 (ashrsi3_set_flags, ashrsi3_v850e2_set_flags): Likewise.
7409 * config/v850/v850-protos.h (notice_update_cc): Remove.
7410 * config/v850/v850.c (v850_compare_op0, v850_compare_op1): Remove.
7411 (v850_print_operand): Handle 'D' and "d".
7412 (v850_select_cc_mode): Remove ATTRIBUTE_UNUSED for last argument.
7413 Add handling of arithmetic/logical operations compared against zero.
7414 (v850_gen_float_compare): Remove ATTRIBUTE_UNUSED for last argument.
7415 Do not look at v850_compare_op, instead get mode from last argument.
7416 (v850_gen_compare): Remove
7417 (increment_stack): Use addsi3_clobber_flags to avoid splitting failure
7418 after reload for prologue insns.
7419 (expand_prologue): Account for CLOBBER of CC_REGNUM in various
7421 (construct_save_jarl): Likewise.
7422 (TARGET_FLAGS_REGNUM): Define.
7423 * config/v850/v850.h (v850_compare_op0, v850_compare_op1): Remove.
7424 (NOTICE_UPDATE_CC): Remove.
7425 * config/v850/v850.md (v850_tst1): Use (reg:CCZ CC_REGNUM) rather
7426 than cc0. Conditionalize on reload_completed.
7427 (cmpsi_insn, setfcc_insn): Likewise.
7428 (tst1 splitter): Turn into define_and_split which sets the flags
7430 (cstoresi4, cbranchsf4, cbranchdf4, cbranchsi4_insn): Likewise.
7431 (cbranchsi4, branch_normal, branch_invert): Do not expose cc0 here.
7432 (cstoresf4, cstoredf4): Clobber the flags.
7433 (cmpsi, cmpsf, cmpdf): Remove expanders.
7434 (setf_insn): Remove pattern.
7435 (addsi3): Turn into define_and_split which clobbers the flags after
7436 reload and a suitable pattern (addsi3_clobber_flags) for use after
7438 (subsi3, negsi2, andsi3, iorsi3, xorsi3, one_cmplsi2) Likewise.
7439 (ashlsi3, ashlsi3_v850e2, lshrsi3, lsh4si3_v850e2): Likewise.
7440 (ashrsi3, ashrsi3_v850e2): Likewise.
7441 (bins): Clobber the flags.
7442 (movsicc_normal_cc, movsicc_normal, movsicc_tst1): Likewise.
7443 (movsicc_tst1_revesed, sasf, swap and rotate patterns): Likewise.
7444 (fix_loop_counter, call_internal_short, call_internal_long): Likewise.
7445 (call_value_internal_short, call_value_internal_long): Likewise.
7446 (callt_save_interrupt, callt_return_interrupt): Likewise.
7447 (save_interrupt, return_interrupt): Likewise.
7448 (callt_save_all_interrupt, save_all_interrupt): Likewise.
7449 (_save_all_interrupt, callt_restore_all_interrupt): Likewise.
7450 (restore_all_interrupt, _restore_all_interrupt): Likewise.
7451 (All FP comparisons): Only allow after reload has completed.
7453 (divh, divhu): Tweak output template.
7454 (branch_z_normal, branch_z_invert): Remove
7455 (branch_nz_normal, branch_nz_invert): Likewise.
7456 (extendhisi_insn, extendqisi_insn): Do not clobber flags.
7458 * config/v850/v850-modes.def (CCZ, CCNZ): Add new modes.
7459 * config/v850/v850.c (notice_update_cc): Remove.
7460 * config/v850/v850.h (CC_OVERFLOW_UNUSABLE): Remove
7461 (CC_NO_CARRY): Likewise.
7462 (NOTICE_UPDATE_CC): Define to nothing.
7463 * config/v850/v850.md: Remove block comment on cc0 handling
7464 Remove "cc" attribute from all patterns. Remove cc_status handling
7465 from all patterns. Minor formatting fixes.
7467 2018-06-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7469 * config/aarch64/aarch64-cores.def (cortex-a76): New entry.
7470 (cortex-a76.cortex-a55): Likewise.
7471 * config/aarch64/aarch64-tune.md: Regenerate.
7472 * doc/invoke.texi (AArch64 Options): Document cortex-a76 and
7473 cortex-a76.cortex-a55.
7475 2018-06-27 Jeff Law <law@redhat.com>
7477 * config/v850/t-v850 (MULTILIB_OPTIONS): Remove 8byte-align.
7478 (MULTILIB_DIRNAMES): Similarly.
7480 2018-06-27 Eric Botcazou <ebotcazou@adacore.com>
7482 * gimple.h (gimple_return_retbnd): Delete.
7483 (gimple_return_set_retbnd): Likewise.
7484 * cgraphunit.c (cgraph_node::expand_thunk): Remove call to
7485 gimple_return_set_retbnd.
7486 * gimple-pretty-print.c (dump_gimple_return): Remove call to
7487 gimple_return_retbnd and adjust.
7488 * tree-inline.h (struct copy_body_data): Remove retbnd field.
7489 * tree-inline.c (remap_gimple_stmt): Remove handling of retbnd.
7490 Explicitly return NULL in a couple more cases. Move assertion
7491 on debug statements and remove unreachable code.
7492 (reset_debug_binding): Do not test id->retbnd.
7493 (expand_call_inline): Do not set it.
7495 2018-06-27 Rasmus Villemoes <rv@rasmusvillemoes.dk>
7497 * configure.ac: Add --disable-gcov option.
7498 * configure: Regenerate.
7499 * Makefile.in: Honour @enable_gcov@.
7500 * doc/install.texi: Document --disable-gcov.
7502 2018-06-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7504 * config/arm/arm-cpus.in (cortex-a76): New entry.
7505 (cortex-a76.cortex-a55): Likewise.
7506 * config/arm/arm-tables.opt: Regenerate.
7507 * config/arm/arm-tune.md: Likewise.
7508 * config/arm/driver-arm.c (arm_cpu_table): Add Cortex-A76 entry.
7509 * doc/invoke.texi (ARM Options): Document cortex-a76 and
7510 cortex-a76.cortex-a55.
7512 2018-06-27 Tamar Christina <tamar.christina@arm.com>
7515 * config/aarch64/aarch64.md (*movhf_aarch64): Add dup v0.4h pattern.
7517 2018-06-27 Siddhesh Poyarekar <siddhesh@sourceware.org>
7519 * config/aarch64/aarch64.h (CALL_USE_REGISTERS): Fix obsolete
7521 (EPILOGUE_USES): Likewise.
7523 2018-06-26 Eric Botcazou <ebotcazou@adacore.com>
7525 * tree-inline.c (remap_location): New function extracted from...
7526 (copy_edges_for_bb): Add ID parameter. Remap goto_locus.
7527 (copy_phis_for_bb): ...here. Call remap_location.
7528 (copy_cfg_body): Adjust call to copy_edges_for_bb.
7530 2018-06-26 Aaron Sawdey <acsawdey@linux.ibm.com>
7532 * config/rs6000/rs6000-string.c (expand_block_clear): Don't use
7533 unaligned vsx for 16B memset.
7535 2018-06-26 Segher Boessenkool <segher@kernel.crashing.org>
7538 * config/rs6000/rs6000.c (rs6000_init_builtins): Do not set
7539 ieee128_float_type_node to long_double_type_node unless
7540 TARGET_LONG_DOUBLE_128 is set.
7542 2018-06-26 David Malcolm <dmalcolm@redhat.com>
7544 * cfgloop.c (get_loop_location): Convert return type from
7545 location_t to dump_user_location_t, replacing INSN_LOCATION lookups
7546 by implicit construction from rtx_insn *, and using
7547 dump_user_location_t::from_function_decl for the fallback case.
7548 * cfgloop.h (get_loop_location): Convert return type from
7549 location_t to dump_user_location_t.
7550 * cgraphunit.c (walk_polymorphic_call_targets): Update call to
7551 dump_printf_loc to pass in a dump_location_t rather than a
7552 location_t, via the gimple stmt.
7553 * coverage.c (get_coverage_counts): Update calls to
7554 dump_printf_loc to pass in dump_location_t rather than a
7556 * doc/optinfo.texi (Dump types): Convert example of
7557 dump_printf_loc from taking "locus" to taking "insn". Update
7558 description of the "_loc" calls to cover dump_location_t.
7559 * dumpfile.c: Include "backend.h", "gimple.h", "rtl.h", and
7561 (dump_user_location_t::dump_user_location_t): New constructors,
7562 from gimple *stmt and rtx_insn *.
7563 (dump_user_location_t::from_function_decl): New function.
7564 (dump_loc): Make static.
7565 (dump_gimple_stmt_loc): Convert param "loc" from location_t to
7566 const dump_location_t &.
7567 (dump_generic_expr_loc): Delete.
7568 (dump_printf_loc): Convert param "loc" from location_t to
7569 const dump_location_t &.
7570 (selftest::test_impl_location): New function.
7571 (selftest::dumpfile_c_tests): New function.
7572 * dumpfile.h: Include "profile-count.h".
7573 (class dump_user_location_t): New class.
7574 (struct dump_impl_location_t): New struct.
7575 (class dump_location_t): New class.
7576 (dump_printf_loc): Convert 2nd param from source_location to
7577 const dump_location_t &.
7578 (dump_generic_expr_loc): Delete.
7579 (dump_gimple_stmt_loc): Convert 2nd param from source_location to
7580 const dump_location_t &.
7581 * gimple-fold.c (fold_gimple_assign): Update call to
7582 dump_printf_loc to pass in a dump_location_t rather than a
7583 location_t, via the gimple stmt.
7584 (gimple_fold_call): Likewise.
7585 * gimple-loop-interchange.cc
7586 (loop_cand::analyze_iloop_reduction_var): Update for change to
7587 check_reduction_path.
7588 (tree_loop_interchange::interchange): Update for change to
7590 * graphite-isl-ast-to-gimple.c (scop_to_isl_ast): Update for
7591 change in return-type of find_loop_location.
7592 (graphite_regenerate_ast_isl): Likewise.
7593 * graphite-optimize-isl.c (optimize_isl): Likewise.
7594 * graphite.c (graphite_transform_loops): Likewise.
7595 * ipa-devirt.c (ipa_devirt): Update call to dump_printf_loc to
7596 pass in a dump_location_t rather than a location_t, via the
7598 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
7599 * ipa.c (walk_polymorphic_call_targets): Likewise.
7600 * loop-unroll.c (report_unroll): Convert "locus" param from
7601 location_t to dump_location_t.
7602 (decide_unrolling): Update for change to get_loop_location's
7604 * omp-grid.c (struct grid_prop): Convert field "target_loc" from
7605 location_t to dump_user_location_t.
7606 (grid_find_single_omp_among_assignments_1): Updates calls to
7607 dump_printf_loc to pass in a dump_location_t rather than a
7608 location_t, via the gimple stmt.
7609 (grid_parallel_clauses_gridifiable): Convert "tloc" from
7610 location_t to dump_location_t. Updates calls to dump_printf_loc
7611 to pass in a dump_location_t rather than a location_t, via the
7613 (grid_inner_loop_gridifiable_p): Likewise.
7614 (grid_dist_follows_simple_pattern): Likewise.
7615 (grid_gfor_follows_tiling_pattern): Likewise.
7616 (grid_target_follows_gridifiable_pattern): Likewise.
7617 (grid_attempt_target_gridification): Convert initialization
7618 of local "grid" from memset to zero-initialization; FIXME: does
7619 this require C++11? Update call to dump_printf_loc to pass in a
7620 optinfo_location rather than a location_t, via the gimple stmt.
7621 * profile.c (read_profile_edge_counts): Updates call to
7622 dump_printf_loc to pass in a dump_location_t rather than a
7624 (compute_branch_probabilities): Likewise.
7625 * selftest-run-tests.c (selftest::run_tests): Call
7627 * selftest.h (dumpfile_c_tests): New decl.
7628 * tree-loop-distribution.c (pass_loop_distribution::execute):
7629 Update for change in return type of find_loop_location.
7630 * tree-parloops.c (parallelize_loops): Likewise.
7631 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Convert
7632 "locus" from location_t to dump_user_location_t.
7633 (canonicalize_loop_induction_variables): Likewise.
7634 * tree-ssa-loop-ivopts.c (tree_ssa_iv_optimize_loop): Update
7635 for change in return type of find_loop_location.
7636 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update call
7637 to dump_printf_loc to pass in a dump_location_t rather than a
7638 location_t, via the stmt.
7639 * tree-ssa-sccvn.c (eliminate_dom_walker::before_dom_children):
7641 * tree-vect-loop-manip.c (find_loop_location): Convert return
7642 type from source_location to dump_user_location_t.
7643 (vect_do_peeling): Update for above change.
7644 (vect_loop_versioning): Update for change in type of
7646 * tree-vect-loop.c (check_reduction_path): Convert "loc" param
7647 from location_t to dump_user_location_t.
7648 (vect_estimate_min_profitable_iters): Update for change in type
7650 * tree-vect-slp.c (vect_print_slp_tree): Convert param "loc" from
7651 location_t to dump_location_t.
7652 (vect_slp_bb): Update for change in type of vect_location.
7653 * tree-vectorizer.c (vect_location): Convert from source_location
7654 to dump_user_location_t.
7655 (try_vectorize_loop_1): Update for change in vect_location's type.
7656 (vectorize_loops): Likewise.
7657 (increase_alignment): Likewise.
7658 * tree-vectorizer.h (vect_location): Convert from source_location
7659 to dump_user_location_t.
7660 (find_loop_location): Convert return type from source_location to
7661 dump_user_location_t.
7662 (check_reduction_path): Convert 1st param from location_t to
7663 dump_user_location_t.
7664 * value-prof.c (check_counter): Update call to dump_printf_loc to
7665 pass in a dump_user_location_t rather than a location_t; update
7666 call to error_at for change in type of "locus".
7667 (check_ic_target): Update call to dump_printf_loc to
7668 pass in a dump_user_location_t rather than a location_t, via the
7671 2018-06-26 Robin Dapp <rdapp@linux.vnet.ibm.com>
7673 * config/s390/s390.h (enum processor_flags): Do not use
7674 default tune parameter when -march was specified.
7676 2018-06-26 Jakub Jelinek <jakub@redhat.com>
7679 * config/i386/i386.md (setcc + movzbl to xor + setcc peephole2s):
7680 Check reg_overlap_mentioned_p in addition to reg_set_p with the same
7683 2018-06-26 Richard Biener <rguenther@suse.de>
7685 PR tree-optimization/86287
7687 * tree-vect-loop.c (vect_transform_loop_stmt): Fix read-after-free.
7688 (vect_analyze_loop): Initialize n_stmts.
7690 2018-06-26 Richard Biener <rguenther@suse.de>
7693 * fold-const.c (fold_convertible_p): Pointer extension
7696 2018-06-26 Alexandre Oliva <aoliva@redhat.com>
7699 * dwarf2out.c (loc_list_has_views): Adjust comments.
7700 (dw_loc_list): Split single cross-partition range with
7703 2018-06-25 Jeff Law <law@redhat.com>
7705 * common/config/v850/v850-common.c (TARGET_DEFAULT_TARGET_FLAGS): Turn
7706 on -mbig-switch by default.
7708 * config/v850/predicates.md (const_float_1_operand): Fix match_code
7710 (const_float_0_operand): Remove unused predicate.
7711 * config/v850/v850.md (define_constants): Remove UNSPEC_LOOP.
7712 (define_c_enum unspec): Add LOOP, RCP and RSQRT constants.
7713 (recipsf2): New expander. Original pattern now called
7715 (recipdf2, recipdf2_insn): Similarly.
7716 (rsqrtsf2, rsqrtsf2_insn): Similarly
7717 (rsqrtdf2, rsqrtdf2_insn): Similarly
7719 2018-06-26 Gerald Pfeifer <gerald@pfeifer.com>
7721 * ginclude/stddef.h: Remove an obsolete comment on FreeBSD 5.
7722 Simplify logic for FreeBSD (twice).
7724 2018-06-25 Martin Sebor <msebor@redhat.com>
7726 PR tree-optimization/86204
7727 * tree-ssa-strlen.c (handle_builtin_strlen): Avoid storing
7728 a strnlen result if it's less than the length of the string.
7730 2018-06-25 Martin Sebor <msebor@redhat.com>
7732 PR tree-optimization/85700
7733 * gimple-fold.c (gimple_fold_builtin_strncat): Adjust comment.
7734 * tree-ssa-strlen.c (is_strlen_related_p): Handle integer subtraction.
7735 (maybe_diag_stxncpy_trunc): Distinguish strncat from strncpy.
7737 2018-06-25 Martin Sebor <msebor@redhat.com>
7739 * doc/extend.texi (Zero-length arrays): Update and clarify.
7741 2018-06-25 Michael Meissner <meissner@linux.ibm.com>
7743 * config.gcc (powerpc64le*): Revert January 16th, 2018 patch that
7744 added IEEE/IBM long double multilib support on PowerPC little
7745 endian Linux systems.
7746 * config/rs6000/linux64.h (MULTILIB_DEFAULTS_IEEE): Likewise.
7747 (MULTILIB_DEFAULTS): Likewise.
7748 * config/rs6000/rs6000.c (rs6000_option_override_internal):
7750 * config/rs6000/rs6000.h (TARGET_IEEEQUAD_MULTILIB): Likewise.
7751 * config/rs6000/t-ldouble-linux64le-ibm: Delete, no longer used.
7752 * config/rs6000/t-ldouble-linux64le-ieee: Delete, no longer used.
7754 2018-06-25 Alexander Monakov <amonakov@ispras.ru>
7757 * sort.cc (REORDER_23): Avoid memcpy with same destination and source.
7758 (REORDER_45): Likewise.
7760 2018-06-25 Jeff Law <law@redhat.com>
7762 * config/v850/v850.md (divmodhi4): Make sure to sign extend the
7763 dividend to 32 bits. Adjust length.
7764 (udivmodhi4): Cleanup output template. Fix length.
7766 2018-06-25 Carl Love <cel@us.ibm.com>
7768 * config/rs6000/vsx.md: Change word selector to prefered location.
7770 2018-06-25 Richard Biener <rguenther@suse.de>
7772 PR tree-optimization/86304
7773 * tree-vectorizer.c (vectorize_loops): Walk over new possibly
7774 epilogue-if-converted loops as well.
7776 2018-06-25 Jan Hubicka <hubicka@ucw.cz>
7778 * lto-section-out.c (lto_begin_section): Do not print section
7779 name for noaddr and unnumbered dumps.
7781 2018-06-25 Richard Biener <rguenther@suse.de>
7783 * tree-vectorizer.h (struct vec_info_shared): New structure
7784 with parts split out from struct vec_info and loop_nest from
7785 struct _loop_vec_info.
7786 (struct vec_info): Adjust accordingly.
7787 (struct _loop_vec_info): Likewise.
7788 (LOOP_VINFO_LOOP_NEST): Adjust.
7789 (LOOP_VINFO_DATAREFS): Likewise.
7790 (LOOP_VINFO_DDRS): Likewise.
7791 (struct _bb_vec_info): Likewise.
7792 (BB_VINFO_DATAREFS): Likewise.
7793 (BB_VINFO_DDRS): Likewise.
7794 (struct _stmt_vec_info): Add dr_aux member.
7795 (DR_VECT_AUX): Adjust to refer to member of DR_STMTs vinfo.
7796 (DR_MISALIGNMENT_UNINITIALIZED): New.
7797 (set_dr_misalignment): Adjust.
7798 (dr_misalignment): Assert misalign isn't DR_MISALIGNMENT_UNINITIALIZED.
7799 (vect_analyze_loop): Adjust prototype.
7800 (vect_analyze_loop_form): Likewise.
7801 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
7802 Compute dependences lazily.
7803 (vect_record_base_alignments): Use shared datarefs/ddrs.
7804 (vect_verify_datarefs_alignment): Likewise.
7805 (vect_analyze_data_refs_alignment): Likewise.
7806 (vect_analyze_data_ref_accesses): Likewise.
7807 (vect_analyze_data_refs): Likewise.
7808 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Add
7809 constructor parameter for shared part.
7810 (vect_analyze_loop_form): Pass in shared part and adjust.
7811 (vect_analyze_loop_2): Pass in storage for the number of
7812 stmts. Move loop nest finding to the caller. Compute
7814 (vect_analyze_loop): Pass in shared part.
7815 (vect_transform_loop): Verify shared datarefs are unchanged.
7816 * tree-vect-slp.c (_bb_vec_info::_bb_vec_info): Add
7817 constructor parameter for shared part.
7818 (vect_slp_analyze_bb_1): Pass in shared part and adjust.
7819 (vect_slp_bb): Verify shared datarefs are unchanged before
7821 * tree-vect-stmts.c (ensure_base_align): Adjust for DR_AUX
7823 (new_stmt_vec_info): Initialize DR_AUX misalignment to
7824 DR_MISALIGNMENT_UNINITIALIZED.
7825 * tree-vectorizer.c (vec_info::vec_info): Add constructor
7826 parameter for shared part.
7827 (vec_info::~vec_info): Adjust.
7828 (vec_info_shared::vec_info_shared): New.
7829 (vec_info_shared::~vec_info_shared): Likewise.
7830 (vec_info_shared::save_datarefs): Likewise.
7831 (vec_info_shared::check_datarefs): Likewise.
7832 (try_vectorize_loop_1): Construct shared part live for analyses
7833 of a single loop for multiple vector sizes.
7834 * tree-parloops.c (gather_scalar_reductions): Adjust.
7836 2018-06-25 Richard Biener <rguenther@suse.de>
7838 * tree-vect-data-refs.c (vect_find_stmt_data_reference): Modify
7839 DR for SIMD lane accesses here and mark DR with (void *)-1 aux.
7840 (vect_analyze_data_refs): Remove similar code from here and
7841 simplify accordingly.
7843 2018-06-25 Richard Biener <rguenther@suse.de>
7845 * tree-vect-data-refs.c (vect_check_gather_scatter): Fail
7846 for reverse storage order accesses rather than asserting
7847 they cannot happen here.
7849 2018-06-25 Tom de Vries <tdevries@suse.de>
7852 * config/i386/i386.md (define_insn "*tls_global_dynamic_64_<mode>"):
7853 Use data16 instead of .byte for insn prefix.
7855 2018-06-25 Andreas Krebbel <krebbel@linux.ibm.com>
7858 * parser.c (make_char_string_pack): Pass this literal chars
7859 through cpp_interpret_string.
7860 (cp_parser_userdef_numeric_literal): Check the result of
7861 make_char_string_pack.
7863 2018-06-24 Maya Rashish <coypu@sdf.org>
7865 * ginclude/stddef.h: Simplify conditions around avoiding
7866 re-definition of __size_t.
7868 2018-06-22 Jan Hubicka <hubicka@ucw.cz>
7870 * lto-streamer-out.c (tree_is_indexable): Make LABEL_DECL nonindexable
7871 unless it is forced or nonlocal; assert that we stream no IMPORTED_DECL.
7873 2018-06-22 Maya Rashish <coypu@sdf.org>
7875 * doc/invoke.texi (mno-fancy-math-387): Update for changes
7876 made to OpenBSD and NetBSD through the years.
7878 2018-06-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
7880 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Change
7881 behavior of vec_pack (vector double, vector double) to match
7882 behavior of vec_float2 (vector double, vector double).
7884 2018-06-22 Olivier Hainque <hainque@adacore.com>
7886 * gimplify.c (gimplify_function_tree): Prevent creation
7887 of a trampoline for the address of the current function
7888 passed to entry/exit instrumentation hooks.
7890 2018-06-22 Aaron Sawdey <acsawdey@linux.ibm.com>
7893 * config/rs6000/rs6000-string.c (expand_strn_compare): Handle -m32
7896 2018-06-22 Martin Liska <mliska@suse.cz>
7898 PR tree-optimization/86263
7899 * tree-switch-conversion.c (jump_table_cluster::find_jump_tables):
7900 Bail out if is_enabled is false.
7901 * tree-switch-conversion.h (jump_table_cluster::is_enabled):
7903 (jump_table_cluster::is_enabled): New function.
7905 2018-06-22 Jan Hubicka <hubicka@ucw.cz>
7907 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not stream
7908 BINFO_BASE_ACCESSES and BINFO_VPTR_FIELD.
7909 * tree-streamer-in.c (streamer_read_tree_bitfields): Likewise.
7910 (lto_input_ts_binfo_tree_pointers): Likewise.
7911 * tree-streamer-out.c (streamer_write_tree_bitfields,
7912 write_ts_binfo_tree_pointers): Likewise.
7913 * tree.c (free_lang_data_in_binfo): Clear BINFO_VPTR_FIELD.
7915 2018-06-22 Jan Hubicka <hubicka@ucw.cz>
7917 * tree.c (free_lang_data_in_type): Free all TYPE_VFIELDs.
7919 2018-06-22 Martin Liska <mliska@suse.cz>
7921 * symbol-summary.h (get): Make it pure and inline move
7922 functionality from ::get function.
7923 (get): Remove and inline into ::get and ::get_create.
7924 (get_create): Move code from ::get function.
7926 2018-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
7929 * config/i386/sol2.h (CPP_SPEC): Don't pass -P for
7930 -x assembler-with-cpp.
7932 2018-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
7934 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define
7935 _FILE_OFFSET_BITS=64 for C++.
7937 2018-06-21 Michael Meissner <meissner@linux.ibm.com>
7939 * config/rs6000/rs6000.md (extendtfif2): Add missing 128-bit
7940 conversion insn that shows up when pr85657-3.c is compiled using
7941 IEEE 128-bit long double.
7942 (neg<mode>2_internal): Use the correct mode to check whether the
7943 mode is IBM extended.
7944 * config/rs6000/rs6000.c (init_float128_ieee): Prevent complex
7945 multiply and divide external functions from being created more
7948 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
7950 * cfgrtl.c (fixup_reorder_chain): Do not emit NOPs in DECL_IGNORED_P
7952 (rtl_merge_blocks): Likewise. Do not emit a NOP if the location of
7953 the edge can be forwarded.
7954 (cfg_layout_merge_blocks): Likewise.
7956 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
7958 * except.c (finish_eh_generation): Commit edge insertions only after
7959 the EH edges have been redirected from post-landing to landing pads.
7961 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
7963 * tree-nested.c (get_frame_type): Use create_tmp_var_raw instead of
7964 create_tmp_var_for to create the FRAME decl.
7965 (finalize_nesting_tree_1): Do not unchain the FRAME decl.
7967 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
7969 * tree-inline.c (copy_edges_for_bb): Minor tweak.
7970 (maybe_move_debug_stmts_to_successors): Also reset the locus of the
7971 debug statement when resetting its value.
7972 (expand_call_inline): Copy the locus of the call onto the assignment
7973 of the return value, if any. Use local variable in more cases.
7975 2018-06-21 Martin Liska <mliska@suse.cz>
7977 * ipa-pure-const.c (propagate_nothrow): Use
7978 funct_state_summaries->get.
7979 (dump_malloc_lattice): Likewise.
7980 (propagate_malloc): Likewise.
7982 2018-06-21 Richard Biener <rguenther@suse.de>
7984 * lto-streamer-out.c (DFS::DFS_write_tree_body): Update outdated
7985 comment. Follow BLOCK_ABSTRACT_ORIGIN unconditionally.
7986 * tree-streamer-in.c (lto_input_ts_block_tree_pointers): Update
7988 * tree-streamer-out.c (write_ts_block_tree_pointers): Stream
7989 BLOCK_ABSTRACT_ORIGIN unconditionally.
7991 2018-06-21 David Malcolm <dmalcolm@redhat.com>
7993 * ipa-cp.c (ipcp_driver): Set edge_clone_summaries to NULL after
7995 * ipa-reference.c (ipa_reference_c_finalize): Delete
7996 ipa_ref_opt_sum_summaries and set it to NULL.
7998 2018-06-21 Tom de Vries <tdevries@suse.de>
8000 PR tree-optimization/85859
8001 * tree-ssa-tail-merge.c (stmt_local_def): Copy gimple_is_call
8002 test with comment from bb_no_side_effects_p.
8004 2018-06-21 Richard Biener <rguenther@suse.de>
8006 PR tree-optimization/86232
8007 * tree-ssa-loop-niter.c (number_of_iterations_popcount): Adjust
8008 max for constant niter.
8010 2018-06-21 Andre Vieira <andre.simoesdiasvieira@arm.com>
8012 * config/aarch64/aarch64-simd.md
8013 (*aarch64_crypto_aes<aes_op>v16qi_xor_combine): New.
8015 2018-06-21 Andre Vieira <andre.simoesdiasvieira@arm.com>
8017 * config/aarch64/aarch64-simd.md (aarch64_crypto_aes<aes_op>v16qi):
8018 Make opernads of the unspec commutative.
8020 2018-06-21 Richard Biener <rguenther@suse.de>
8022 * tree-data-ref.c (dr_step_indicator): Handle NULL DR_STEP.
8023 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
8024 Avoid calling vect_mark_for_runtime_alias_test with gathers or scatters.
8025 (vect_analyze_data_ref_dependence): Re-order checks to deal with
8027 (vect_record_base_alignments): Do not record base alignment
8028 for gathers or scatters.
8029 (vect_compute_data_ref_alignment): Drop return value that is always
8030 true. Bail out early for gathers or scatters.
8031 (vect_enhance_data_refs_alignment): Bail out early for gathers
8033 (vect_find_same_alignment_drs): Likewise.
8034 (vect_analyze_data_refs_alignment): Remove dead code.
8035 (vect_slp_analyze_and_verify_node_alignment): Likewise.
8036 (vect_analyze_data_refs): For possible gathers or scatters do
8037 not create an alternate DR, just check their possible validity
8038 and mark them. Adjust DECL_NONALIASED handling to not rely
8040 * tree-vect-loop-manip.c (vect_update_inits_of_drs): Do not
8041 update inits of gathers or scatters.
8042 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern):
8043 Also copy gather/scatter flag to pattern vinfo.
8045 2018-06-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
8047 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Change
8048 behavior of vec_packsu (vector unsigned long long, vector unsigned
8049 long long) to match behavior of vec_packs with same signature.
8051 2018-06-20 Chung-Lin Tang <cltang@codesourcery.com>
8052 Thomas Schwinge <thomas@codesourcery.com>
8053 Cesar Philippidis <cesar@codesourcery.com>
8055 * gimplify.c (gimplify_scan_omp_clauses): Add support for
8056 OMP_CLAUSE_{IF_PRESENT,FINALIZE}.
8057 (gimplify_adjust_omp_clauses): Likewise.
8058 (gimplify_oacc_declare_1): Add support for GOMP_MAP_RELEASE, remove
8059 support for GOMP_MAP_FORCE_{ALLOC,TO,FROM,TOFROM}.
8060 (gimplify_omp_target_update): Update handling of acc update and
8062 * omp-low.c (install_var_field): Remove unused parameter
8063 base_pointers_restrict.
8064 (scan_sharing_clauses): Remove base_pointers_restrict parameter.
8065 Update call to install_var_field. Handle OMP_CLAUSE_{IF_PRESENT,
8067 (omp_target_base_pointers_restrict_p): Delete.
8068 (scan_omp_target): Update call to scan_sharing_clauses.
8069 * tree-core.h (enum omp_clause_code): Add OMP_CLAUSE_{IF_PRESENT,
8071 * tree-nested.c (convert_nonlocal_omp_clauses): Handle
8072 OMP_CLAUSE_{IF_PRESENT,FINALIZE}.
8073 (convert_local_omp_clauses): Likewise.
8074 * tree-pretty-print.c (dump_omp_clause): Likewise.
8075 * tree.c (omp_clause_num_ops): Add entries for OMP_CLAUSE_{IF_PRESENT,
8077 (omp_clause_code_name): Likewise.
8079 2018-06-20 Jakub Jelinek <jakub@redhat.com>
8082 * var-tracking.c (use_narrower_mode_test): Check if shift amount can
8085 PR tree-optimization/86231
8086 * tree-vrp.c (union_ranges): For ( [ ) ] or ( )[ ] range and
8087 anti-range don't overwrite *vr0min before using it to compute *vr0max.
8089 2018-06-20 Tom de Vries <tdevries@suse.de>
8091 PR tree-optimization/86097
8092 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Also convert *nit to
8093 iv type if signedness of iv type is not the same as that of *nit.
8095 2018-06-20 Jakub Jelinek <jakub@redhat.com>
8097 * cfgrtl.c (rtl_verify_edges): Formatting fix. If bb->preds has any
8098 EDGE_EH edges, verify they are all EDGE_EH.
8100 2018-06-20 Maya Rashish <coypu@sdf.org>
8102 * ginclude/stddef.h: Limit #include <machine/ansi.h> to NetBSD.
8104 2018-06-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8106 * config/aarch64/aarch64-tuning-flags.def (no_ldp_stp_qregs): New.
8107 * config/aarch64/aarch64.c (xgene1_tunings): Add
8108 AARCH64_EXTRA_TUNE_NO_LDP_STP_QREGS to tune_flags.
8109 (aarch64_mode_valid_for_sched_fusion_p):
8110 Allow 16-byte modes.
8111 (aarch64_classify_address): Allow 16-byte modes for load_store_pair_p.
8112 * config/aarch64/aarch64-ldpstp.md: Add peepholes for LDP STP of
8114 * config/aarch64/aarch64-simd.md (load_pair<VQ:mode><VQ2:mode>):
8116 (vec_store_pair<VQ:mode><VQ2:mode>): Likewise.
8117 * config/aarch64/iterators.md (VQ2): New mode iterator.
8119 2018-06-20 Martin Liska <mliska@suse.cz>
8121 * tree-switch-conversion.c (jump_table_cluster::can_be_handled):
8122 Change default ratio from 10 to 8.
8124 2018-06-20 Martin Liska <mliska@suse.cz>
8126 * tree-switch-conversion.c (jump_table_cluster::find_jump_tables):
8128 (bit_test_cluster::find_bit_tests): Likewise.
8129 (switch_decision_tree::analyze_switch_statement): Find clusters.
8130 * tree-switch-conversion.h (struct jump_table_cluster): Document
8133 2018-06-20 Martin Liska <mliska@suse.cz>
8135 * tree-switch-conversion.c (switch_conversion::collect):
8136 Record m_uniq property.
8137 (switch_conversion::expand): Bail out for special conditions.
8138 (group_cluster::~group_cluster): New.
8139 (group_cluster::group_cluster): Likewise.
8140 (group_cluster::dump): Likewise.
8141 (jump_table_cluster::emit): New.
8142 (switch_decision_tree::fix_phi_operands_for_edges): New.
8143 (struct case_node): Remove struct.
8144 (jump_table_cluster::can_be_handled): New.
8145 (case_values_threshold): Moved to header.
8146 (reset_out_edges_aux): Likewise.
8147 (jump_table_cluster::is_beneficial): New.
8148 (bit_test_cluster::can_be_handled): Likewise.
8149 (add_case_node): Remove.
8150 (bit_test_cluster::is_beneficial): New.
8151 (case_bit_test::cmp): New.
8152 (bit_test_cluster::emit): New.
8153 (expand_switch_as_decision_tree_p): Remove.
8154 (bit_test_cluster::hoist_edge_and_branch_if_true): New.
8155 (fix_phi_operands_for_edge): Likewise.
8156 (switch_decision_tree::analyze_switch_statement): New.
8157 (compute_cases_per_edge): Move ...
8158 (switch_decision_tree::compute_cases_per_edge): ... here.
8159 (try_switch_expansion): Likewise.
8160 (switch_decision_tree::try_switch_expansion): Likewise.
8161 (record_phi_operand_mapping): Likewise.
8162 (switch_decision_tree::record_phi_operand_mapping): Likewise.
8163 (emit_case_decision_tree): Likewise.
8164 (switch_decision_tree::emit): Likewise.
8165 (balance_case_nodes): Likewise.
8166 (switch_decision_tree::balance_case_nodes): Likewise.
8167 (dump_case_nodes): Likewise.
8168 (switch_decision_tree::dump_case_nodes): Likewise.
8169 (emit_jump): Likewise.
8170 (switch_decision_tree::emit_jump): Likewise.
8171 (emit_cmp_and_jump_insns): Likewise.
8172 (switch_decision_tree::emit_cmp_and_jump_insns): Likewise.
8173 (emit_case_nodes): Likewise.
8174 (switch_decision_tree::emit_case_nodes): Likewise.
8175 (conditional_probability): Remove.
8176 * tree-switch-conversion.h (enum cluster_type): New.
8178 (struct cluster): Likewise.
8179 (cluster::cluster): Likewise.
8180 (struct simple_cluster): Likewise.
8181 (simple_cluster::simple_cluster): Likewise.
8182 (struct group_cluster): Likewise.
8183 (struct jump_table_cluster): Likewise.
8184 (struct bit_test_cluster): Likewise.
8185 (struct min_cluster_item): Likewise.
8186 (struct case_tree_node): Likewise.
8187 (case_tree_node::case_tree_node): Likewise.
8188 (jump_table_cluster::case_values_threshold): Likewise.
8189 (struct case_bit_test): Likewise.
8190 (struct switch_decision_tree): Likewise.
8191 (struct switch_conversion): Likewise.
8192 (switch_decision_tree::reset_out_edges_aux): Likewise.
8194 2018-06-20 Martin Liska <mliska@suse.cz>
8196 * tree-switch-conversion.c (MAX_CASE_BIT_TESTS): Remove.
8197 (hoist_edge_and_branch_if_true): Likewise.
8198 (expand_switch_using_bit_tests_p): Likewise.
8199 (struct case_bit_test): Likewise.
8200 (case_bit_test_cmp): Likewise.
8201 (emit_case_bit_tests): Likewise.
8202 (switch_conversion::switch_conversion): New class.
8203 (struct switch_conv_info): Remove old struct.
8204 (collect_switch_conv_info): More to ...
8205 (switch_conversion::collect): ... this.
8206 (check_range): Likewise.
8207 (switch_conversion::check_range): Likewise.
8208 (check_all_empty_except_final): Likewise.
8209 (switch_conversion::check_all_empty_except_final): Likewise.
8210 (check_final_bb): Likewise.
8211 (switch_conversion::check_final_bb): Likewise.
8212 (create_temp_arrays): Likewise.
8213 (switch_conversion::create_temp_arrays): Likewise.
8214 (free_temp_arrays): Likewise.
8215 (gather_default_values): Likewise.
8216 (switch_conversion::gather_default_values): Likewise.
8217 (build_constructors): Likewise.
8218 (switch_conversion::build_constructors): Likewise.
8219 (constructor_contains_same_values_p): Likewise.
8220 (switch_conversion::contains_same_values_p): Likewise.
8221 (array_value_type): Likewise.
8222 (switch_conversion::array_value_type): Likewise.
8223 (build_one_array): Likewise.
8224 (switch_conversion::build_one_array): Likewise.
8225 (build_arrays): Likewise.
8226 (switch_conversion::build_arrays): Likewise.
8227 (gen_def_assigns): Likewise.
8228 (switch_conversion::gen_def_assigns): Likewise.
8229 (prune_bbs): Likewise.
8230 (switch_conversion::prune_bbs): Likewise.
8231 (fix_phi_nodes): Likewise.
8232 (switch_conversion::fix_phi_nodes): Likewise.
8233 (gen_inbound_check): Likewise.
8234 (switch_conversion::gen_inbound_check): Likewise.
8235 (process_switch): Use the newly created class.
8236 (switch_conversion::expand): New.
8237 (switch_conversion::~switch_conversion): New.
8238 * tree-switch-conversion.h: New file.
8240 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
8242 * tree-vectorizer.h (NUM_PATTERNS, vect_recog_func_ptr): Move to
8243 tree-vect-patterns.c.
8244 * tree-vect-patterns.c (vect_supportable_direct_optab_p): New function.
8245 (vect_recog_dot_prod_pattern): Use it. Remove the type_in argument.
8246 (vect_recog_sad_pattern): Likewise.
8247 (vect_recog_widen_sum_pattern): Likewise.
8248 (vect_recog_pow_pattern): Likewise. Check for a null vectype.
8249 (vect_recog_widen_shift_pattern): Remove the type_in argument.
8250 (vect_recog_rotate_pattern): Likewise.
8251 (vect_recog_mult_pattern): Likewise.
8252 (vect_recog_vector_vector_shift_pattern): Likewise.
8253 (vect_recog_divmod_pattern): Likewise.
8254 (vect_recog_mixed_size_cond_pattern): Likewise.
8255 (vect_recog_bool_pattern): Likewise.
8256 (vect_recog_mask_conversion_pattern): Likewise.
8257 (vect_try_gather_scatter_pattern): Likewise.
8258 (vect_recog_widen_mult_pattern): Likewise. Check for a null vectype.
8259 (vect_recog_over_widening_pattern): Likewise.
8260 (vect_recog_gather_scatter_pattern): Likewise.
8261 (vect_recog_func_ptr): Move from tree-vectorizer.h
8262 (vect_vect_recog_func_ptrs): Move further down the file.
8263 (vect_recog_func): Likewise. Remove the third argument.
8264 (NUM_PATTERNS): Define based on vect_vect_recog_func_ptrs.
8265 (vect_pattern_recog_1): Expect the pattern function to do any
8266 necessary target tests. Also expect it to provide a vector type.
8267 Remove the type_in handling.
8269 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
8271 * tree-vect-patterns.c (vect_pattern_detected): New function.
8272 (vect_recog_dot_prod_patternm, vect_recog_sad_pattern)
8273 (vect_recog_widen_mult_pattern, vect_recog_widen_sum_pattern)
8274 (vect_recog_over_widening_pattern, vect_recog_widen_shift_pattern
8275 (vect_recog_rotate_pattern, vect_recog_vector_vector_shift_pattern)
8276 (vect_recog_mult_pattern, vect_recog_divmod_pattern)
8277 (vect_recog_mixed_size_cond_pattern, vect_recog_bool_pattern)
8278 (vect_recog_mask_conversion_pattern)
8279 (vect_try_gather_scatter_pattern): Likewise.
8281 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
8283 * tree-vect-patterns.c (vect_get_internal_def): New function.
8284 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern)
8285 (vect_recog_vector_vector_shift_pattern, check_bool_pattern)
8286 (search_type_for_mask_1): Use it.
8288 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
8290 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Remove
8291 redundant WIDEN_SUM_EXPR handling.
8292 (vect_recog_sad_pattern): Likewise.
8294 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
8296 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Remove
8297 redundant check that the types of a PLUS_EXPR or MULT_EXPR agree.
8298 (vect_recog_sad_pattern): Likewise PLUS_EXPR, ABS_EXPR and MINUS_EXPR.
8299 (vect_recog_widen_mult_pattern): Likewise MULT_EXPR.
8300 (vect_recog_widen_sum_pattern): Likewise PLUS_EXPR.
8302 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
8304 * tree-vect-stmts.c (vectorizable_call): Make sure that we
8305 use the stmt_vec_info of the original bb statement for the
8306 new zero assignment, even if the call is part of a pattern.
8308 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
8310 * tree-vectorizer.h (_stmt_vec_info): Note above pattern_def_seq
8311 that the sequence is attached to the original statement rather
8312 than the pattern statement.
8313 * tree-vect-loop.c (vect_determine_vf_for_stmt): Take the
8314 PATTERN_DEF_SEQ from the original statement rather than
8315 the main pattern statement.
8316 * tree-vect-stmts.c (free_stmt_vec_info): Likewise.
8317 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Likewise.
8318 (vect_mark_pattern_stmts): Don't copy the PATTERN_DEF_SEQ.
8320 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
8322 * tree-vect-stmts.c (vect_analyze_stmt): Move the handling of pattern
8323 definition statements before the early exit for statements that aren't
8325 * tree-vect-loop.c (vect_transform_loop_stmt): New function,
8327 (vect_transform_loop): ...here. Process pattern definition
8328 statements without first checking whether the main pattern
8329 statement is live or relevant.
8331 2018-06-19 Eric Botcazou <ebotcazou@adacore.com>
8333 * tree-cfgcleanup.c (tree_forwarder_block_p): Do not return false at
8334 -O0 if the locus represent UNKNOWN_LOCATION but have different values.
8336 2018-06-19 Aaron Sawdey <acsawdey@linux.ibm.com>
8338 * config/rs6000/rs6000-string.c (select_block_compare_mode): Check
8339 TARGET_EFFICIENT_OVERLAPPING_UNALIGNED here instead of in caller.
8340 (do_and3, do_and3_mask, do_compb3, do_rotl3): New functions.
8341 (expand_block_compare): Change select_block_compare_mode call.
8342 (expand_strncmp_align_check): Use new functions, fix comment.
8343 (emit_final_str_compare_gpr): New function.
8344 (expand_strn_compare): Refactor and clean up code.
8345 * config/rs6000/vsx.md (vsx_mov<mode>_64bit): Remove *.
8347 2018-06-19 Tony Reix <tony.reix@atos.com>
8348 Damien Bergamini <damien.bergamini@atos.com>
8349 David Edelsohn <dje.gcc@gmail.com>
8351 * collect2.c (static_obj): New variable.
8352 (static_libs): New variable.
8353 (is_in_list): Uncomment declaration.
8354 (main): Track AIX libraries linked statically.
8355 (is_in_list): Uncomment definition.
8356 (scan_prog_file): Don't add AIX shared libraries initializer
8357 to constructor list if linking statically.
8359 2018-06-19 Max Filippov <jcmvbkbc@gmail.com>
8361 * config/xtensa/xtensa.md (UNSPEC_FRAME_BLOCKAGE): New unspec
8363 (allocate_stack, frame_blockage, *frame_blockage): New patterns.
8365 2018-06-19 Jan Hubicka <hubicka@ucw.cz>
8367 * tree.c (find_decls_types_r): Remove all non-VAR_DECLs from
8370 2018-06-19 Martin Liska <mliska@suse.cz>
8372 * config/i386/i386.c (ix86_can_inline_p): Do not use
8373 ipa_fn_summaries::get_create.
8374 * ipa-cp.c (ipcp_cloning_candidate_p): Replace get_create with
8376 (devirtualization_time_bonus): Likewise.
8377 (ipcp_propagate_stage): Likewise.
8378 * ipa-fnsummary.c (redirect_to_unreachable): Likewise.
8379 (edge_set_predicate): Likewise.
8380 (evaluate_conditions_for_known_args): Likewise.
8381 (evaluate_properties_for_edge): Likewise.
8382 (ipa_call_summary::reset): Tranform to ...
8383 (ipa_call_summary::~ipa_call_summary): ... this.
8384 (ipa_fn_summary::reset): Transform to ...
8385 (ipa_fn_summary::~ipa_fn_summary): ... this.
8386 (ipa_fn_summary_t::remove): Rename to ...
8387 (ipa_fn_summary_t::remove_callees): ... this.
8388 (ipa_fn_summary_t::duplicate): Use placement new
8389 instead of memory copy.
8390 (ipa_call_summary_t::duplicate): Likewise.
8391 (ipa_call_summary_t::remove): Remove.
8392 (dump_ipa_call_summary): Change get_create to get.
8393 (ipa_dump_fn_summary): Dump only when summary exists.
8394 (analyze_function_body): Use symbol_summary::get instead
8396 (compute_fn_summary): Likewise.
8397 (estimate_edge_devirt_benefit): Likewise.
8398 (estimate_edge_size_and_time): Likewise.
8399 (inline_update_callee_summaries): Likewise.
8400 (remap_edge_change_prob): Likewise.
8401 (remap_edge_summaries): Likewise.
8402 (ipa_merge_fn_summary_after_inlining): Likewise.
8403 (write_ipa_call_summary): Likewise.
8404 (ipa_fn_summary_write): Likewise.
8405 (ipa_free_fn_summary): Likewise.
8406 * ipa-fnsummary.h (struct GTY): Add new ctor and copy ctor.
8407 (struct ipa_call_summary): Likewise.
8408 * ipa-icf.c (sem_function::merge): Use symbol_summary::get instead
8410 * ipa-inline-analysis.c (do_estimate_edge_time): Likewise.
8411 (estimate_size_after_inlining): Likewise.
8412 (estimate_growth): Likewise.
8413 (growth_likely_positive): Likewise.
8414 * ipa-inline-transform.c (clone_inlined_nodes): Likewise.
8415 (inline_call): Likewise.
8416 * ipa-inline.c (caller_growth_limits): Likewise.
8417 (can_inline_edge_p): Likewise.
8418 (can_inline_edge_by_limits_p): Likewise.
8419 (compute_uninlined_call_time): Likewise.
8420 (compute_inlined_call_time): Likewise.
8421 (want_inline_small_function_p): Likewise.
8422 (edge_badness): Likewise.
8423 (update_caller_keys): Likewise.
8424 (update_callee_keys): Likewise.
8425 (inline_small_functions): Likewise.
8426 (inline_to_all_callers_1): Likewise.
8427 (dump_overall_stats): Likewise.
8428 (early_inline_small_functions): Likewise.
8429 (early_inliner): Likewise.
8430 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
8431 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
8432 * ipa-pure-const.c (malloc_candidate_p): Likewise.
8433 * ipa-split.c (execute_split_functions): Likewise.
8434 * symbol-summary.h: Likewise.
8435 * tree-sra.c (ipa_sra_preliminary_function_checks): Likewise.
8437 2018-06-19 Richard Biener <rguenther@suse.de>
8439 * tree-vectorizer.c (try_vectorize_loop_1): Split out of ...
8440 (vectorize_loops): ... here. Fix dbgcnt handling.
8441 (try_vectorize_loop): Wrap try_vectorize_loop_1.
8443 2018-06-19 Segher Boessenkool <segher@kernel.crashing.org>
8446 * config/rs6000/rs6000.md (rs6000_discover_homogeneous_aggregate): An
8447 ieee128 argument takes up only one (vector) register, not two (floating
8450 2018-06-19 Eric Botcazou <ebotcazou@adacore.com>
8452 * gimplify.c (gimplify_init_constructor): Really never clear for an
8453 incomplete constructor if CONSTRUCTOR_NO_CLEARING is set.
8455 2018-06-19 Richard Biener <rguenther@suse.de>
8457 PR tree-optimization/86179
8458 * tree-vect-patterns.c (vect_pattern_recog_1): Clean up
8459 after failed recognition.
8461 2018-06-18 Martin Sebor <msebor@redhat.com>
8464 * calls.c (maybe_warn_nonstring_arg): Handle strncat.
8465 * tree-ssa-strlen.c (is_strlen_related_p): Make extern.
8466 Handle integer subtraction.
8467 (maybe_diag_stxncpy_trunc): Handle nonstring source arguments.
8468 * tree-ssa-strlen.h (is_strlen_related_p): Declare.
8470 2018-06-18 David Malcolm <dmalcolm@redhat.com>
8472 * config/frv/frv-protos.h (frv_ifcvt_modify_insn): Strengthen 3rd
8473 param from rtx to rtx_insn *.
8474 * config/frv/frv.c (frv_ifcvt_add_insn): Likewise for "insn"
8476 (frv_ifcvt_modify_insn): Likwise.
8477 (frv_ifcvt_modify_final): Likwise for local "existing_insn",
8478 adding an as_a <rtx_insn *> cast. Likewise for local "insn".
8479 * config/mips/mips.c (r10k_insert_cache_barriers): Add an
8480 as_a <rtx_insn *> cast to local "unprotected_region" once
8481 it's been established that it's not NULL or pc_rtx.
8482 * config/nds32/nds32-relax-opt.c (nds32_group_insns): Strengthen
8483 param "sethi" from rtx to rtx_insn *.
8484 (nds32_group_float_insns): Likewise for param "insn".
8485 * config/vax/vax-protos.h (vax_output_int_add): Likewise for 1st
8487 (vax_output_int_subtract): Likewise.
8488 * config/vax/vax.c (vax_output_int_add): Likewise for param
8490 (vax_output_int_subtract): Likewise.
8491 * emit-rtl.c (set_insn_deleted): Likewise, removing cast.
8492 (emit_pattern_after): Likewise for param "after".
8493 (emit_insn_after): Likewise.
8494 (emit_jump_insn_after): Likewise.
8495 (emit_call_insn_after): Likewise.
8496 (emit_debug_insn_after): Likewise.
8497 (emit_pattern_before): Likewise for param "before".
8498 (emit_insn_before): Likewise.
8499 (emit_jump_insn_before): Likewise.
8500 * final.c (get_insn_template): Likewise for param "insn", removing
8502 * output.h (get_insn_template): Likewise for 2nd param.
8503 * rtl.h (emit_insn_before): Likewise.
8504 (emit_jump_insn_before): Likewise.
8505 (emit_debug_insn_before_noloc): Likewise.
8506 (emit_insn_after): Likewise.
8507 (emit_jump_insn_after): Likewise.
8508 (emit_call_insn_after): Likewise.
8509 (emit_debug_insn_after): Likewise.
8510 (set_insn_deleted): Likewise for param.
8512 2018-06-18 Michael Meissner <meissner@linux.ibm.com>
8515 * config/rs6000/rs6000-modes.def (toplevel): Rework the 128-bit
8516 floating point modes, so that IFmode is numerically greater than
8517 TFmode, which is greater than KFmode using FRACTIONAL_FLOAT_MODE
8518 to declare the ordering. This prevents IFmode from being
8519 converted to TFmode when long double is IEEE 128-bit on an ISA 3.0
8520 machine. Include rs6000-modes.h to share the fractional values
8521 between genmodes* and the rest of the compiler.
8525 * config/rs6000/rs6000-modes.h: New file.
8526 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Change the
8527 meaning of rs6000_long_double_size so that 126..128 selects an
8528 appropriate 128-bit floating point type.
8529 (rs6000_option_override_internal): Likewise.
8530 * config/rs6000/rs6000.h (toplevel): Include rs6000-modes.h.
8531 (TARGET_LONG_DOUBLE_128): Change the meaning of
8532 rs6000_long_double_size so that 126..128 selects an appropriate
8533 128-bit floating point type.
8534 (LONG_DOUBLE_TYPE_SIZE): Update comment.
8535 * config/rs6000/rs6000.md (trunciftf2): Correct the modes of the
8536 source and destination to match the standard usage.
8537 (truncifkf2): Likewise.
8538 (copysign<mode>3, IEEE iterator): Rework copysign of float128 on
8539 ISA 2.07 to use an explicit clobber, instead of passing in a
8541 (copysign<mode>3_soft): Likewise.
8543 2018-06-18 David Malcolm <dmalcolm@redhat.com>
8545 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
8546 Replace dump_printf_loc call with DUMP_VECT_SCOPE.
8547 (vect_slp_analyze_instance_dependence): Likewise.
8548 (vect_enhance_data_refs_alignment): Likewise.
8549 (vect_analyze_data_refs_alignment): Likewise.
8550 (vect_slp_analyze_and_verify_instance_alignment
8551 (vect_analyze_data_ref_accesses): Likewise.
8552 (vect_prune_runtime_alias_test_list): Likewise.
8553 (vect_analyze_data_refs): Likewise.
8554 * tree-vect-loop-manip.c (vect_update_inits_of_drs): Likewise.
8555 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
8556 (vect_analyze_scalar_cycles_1): Likewise.
8557 (vect_get_loop_niters): Likewise.
8558 (vect_analyze_loop_form_1): Likewise.
8559 (vect_update_vf_for_slp): Likewise.
8560 (vect_analyze_loop_operations): Likewise.
8561 (vect_analyze_loop): Likewise.
8562 (vectorizable_induction): Likewise.
8563 (vect_transform_loop): Likewise.
8564 * tree-vect-patterns.c (vect_pattern_recog): Likewise.
8565 * tree-vect-slp.c (vect_analyze_slp): Likewise.
8566 (vect_make_slp_decision): Likewise.
8567 (vect_detect_hybrid_slp): Likewise.
8568 (vect_slp_analyze_operations): Likewise.
8569 (vect_slp_bb): Likewise.
8570 * tree-vect-stmts.c (vect_mark_stmts_to_be_vectorized): Likewise.
8571 (vectorizable_bswap): Likewise.
8572 (vectorizable_call): Likewise.
8573 (vectorizable_simd_clone_call): Likewise.
8574 (vectorizable_conversion): Likewise.
8575 (vectorizable_assignment): Likewise.
8576 (vectorizable_shift): Likewise.
8577 (vectorizable_operation): Likewise.
8578 * tree-vectorizer.h (DUMP_VECT_SCOPE): New macro.
8580 2018-06-18 Martin Sebor <msebor@redhat.com>
8582 PR tree-optimization/81384
8583 * builtin-types.def (BT_FN_SIZE_CONST_STRING_SIZE): New.
8584 * builtins.c (expand_builtin_strnlen): New function.
8585 (expand_builtin): Call it.
8586 (fold_builtin_n): Avoid setting TREE_NO_WARNING.
8587 * builtins.def (BUILT_IN_STRNLEN): New.
8588 * calls.c (maybe_warn_nonstring_arg): Handle BUILT_IN_STRNLEN.
8589 Warn for bounds in excess of maximum object size.
8590 * tree-ssa-strlen.c (maybe_set_strlen_range): Return tree representing
8591 single-value ranges. Handle strnlen.
8592 (handle_builtin_strlen): Handle strnlen.
8593 (strlen_check_and_optimize_stmt): Same.
8594 * doc/extend.texi (Other Builtins): Document strnlen.
8596 2018-06-18 Maya Rashish <coypu@sdf.org>
8598 * config/alpha/openbsd.h (TARGET_DEFAULT): Define.
8599 (LINK_SPEC, STARTFILE_SPEC, ENDFILE_SPEC): Likewise.
8600 (INTMAX_TYPE, UINTMAX_TYPE, WINT_TYPE): Likewise.
8602 * config/alpha/elf.h (STARTFILE_SPEC, ENDFILE_SPEC): Move from
8604 * config/alpha/linux.h (STARTFILE_SPEC, ENDFILE_SPEC): Here.
8606 2018-06-18 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
8608 * tree.c (escaped_string::escape): Replace cast to char * by
8609 const_cast<char *> (unescaped).
8611 2018-06-18 Nick Clifton <nickc@redhat.com>
8614 * tree.c (escaped_string): New class. Converts an unescaped
8615 string into its escaped equivalent.
8616 (warn_deprecated_use): Use the new class to convert the
8617 deprecation message, if present.
8618 (test_escaped_strings): New self test.
8619 (test_c_tests): Add test_escaped_strings.
8620 * doc/extend.texi (deprecated): Add a note that the
8621 deprecation message is affected by the -fmessage-length
8622 option, and that control characters will be escaped.
8623 (#pragma GCC error): Document this pragma.
8624 (#pragma GCC warning): Likewise.
8625 * doc/invoke.texi (-fmessage-length): Document this option's
8626 effect on the #warning and #error preprocessor directives and
8627 the deprecated attribute.
8629 2018-06-18 Eric Botcazou <ebotcazou@adacore.com>
8631 * tree.c (decl_value_expr_lookup): Revert latest change.
8632 (decl_value_expr_insert): Likewise.
8634 2018-06-17 Eric Botcazou <ebotcazou@adacore.com>
8636 * gimplify.c (nonlocal_vlas): Delete.
8637 (nonlocal_vla_vars): Likewise.
8638 (gimplify_var_or_parm_decl): Do not add debug VAR_DECLs for non-local
8640 (gimplify_body): Do not create and destroy nonlocal_vlas.
8641 * tree-nested.c: Include diagnostic.h.
8642 (use_pointer_in_frame): Tweak.
8643 (lookup_field_for_decl): Add assertion and declare the transformation.
8644 (convert_nonlocal_reference_op) <PARM_DECL>: Rework and issue an
8645 internal error when the reference is in a wrong context. Do not
8646 create a debug decl by default.
8647 (note_nonlocal_block_vlas): Delete.
8648 (convert_nonlocal_reference_stmt) <GIMPLE_BIND>: Do not call it.
8649 (convert_local_reference_op) <PARM_DECL>: Skip the frame decl. Do not
8650 create a debug decl by default.
8651 (convert_gimple_call) <GIMPLE_CALL>: Issue an internal error when the
8652 call is in a wrong context.
8653 (fixup_vla_decls): New function.
8654 (finalize_nesting_tree_1): Adjust comment. Call fixup_vla_decls if no
8655 debug variables were created.
8656 * tree.c (decl_value_expr_lookup): Add checking assertion.
8657 (decl_value_expr_insert): Likewise.
8659 2018-06-16 Kugan Vivekanandarajah <kuganv@linaro.org>
8662 * ipa-fnsummary.c (will_be_nonconstant_expr_predicate): Handle CALL_EXPR.
8663 * tree-scalar-evolution.c (interpret_expr): Likewise.
8664 (expression_expensive_p): Likewise.
8665 * tree-ssa-loop-ivopts.c (contains_abnormal_ssa_name_p): Likewise.
8666 * tree-ssa-loop-niter.c (number_of_iterations_popcount): New.
8667 (number_of_iterations_exit_assumptions): Use number_of_iterations_popcount.
8668 (ssa_defined_by_minus_one_stmt_p): New.
8670 2018-06-16 Kugan Vivekanandarajah <kuganv@linaro.org>
8673 * cfgexpand.c (expand_debug_expr): Hande ABSU_EXPR.
8674 * config/i386/i386.c (ix86_add_stmt_cost): Likewise.
8675 * dojump.c (do_jump): Likewise.
8676 * expr.c (expand_expr_real_2): Check operand type's sign.
8677 * fold-const.c (const_unop): Handle ABSU_EXPR.
8678 (fold_abs_const): Likewise.
8679 * gimple-pretty-print.c (dump_unary_rhs): Likewise.
8680 * gimple-ssa-backprop.c (backprop::process_assign_use): Likesie.
8681 (strip_sign_op_1): Likesise.
8682 * match.pd: Add new pattern to generate ABSU_EXPR.
8683 * optabs-tree.c (optab_for_tree_code): Handle ABSU_EXPR.
8684 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
8685 * tree-eh.c (operation_could_trap_helper_p): Likewise.
8686 * tree-inline.c (estimate_operator_cost): Likewise.
8687 * tree-pretty-print.c (dump_generic_node): Likewise.
8688 * tree-vect-patterns.c (vect_recog_sad_pattern): Likewise.
8689 * tree.def (ABSU_EXPR): New.
8691 2018-06-16 Jakub Jelinek <jakub@redhat.com>
8694 * common.opt (Wunsafe-loop-optimizations): Add Ignore, remove Var,
8695 documented as preserved for backward compatibility only.
8696 * doc/invoke.texi: Remove -Wunsafe-loop-optimizations documentation.
8698 PR rtl-optimization/86108
8699 * bb-reorder.c (create_forwarder_block): Renamed to ...
8700 (create_eh_forwarder_block): ... this. Split OLD_BB after labels and
8701 jump from new landing pad to the second part.
8702 (sjlj_fix_up_crossing_landing_pad, dw2_fix_up_crossing_landing_pad):
8705 2018-06-15 Jakub Jelinek <jakub@redhat.com>
8708 * expr.c (expand_assignment): Remove now redundant COMPLEX_MODE_P
8709 check from first store_expr, use to_mode instead of GET_MODE (to_rtx).
8710 Only call store_expr for halves if the mode is the same.
8713 * match.pd ((X / Y) == 0 -> X < Y): Don't transform complex divisions.
8714 Fix up comment formatting.
8716 2018-06-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
8718 * typed-splay-tree.h (typed_splay_tree::remove): New function.
8719 (typed_splay_tree::closure,
8720 typed_splay_tree::inner_foreach_fn, typed_splay_tree::m_inner): Deleted.
8721 (typed_splay_tree::typed_splay_tree,
8722 typed_splay_tree::operator =): Declared private.
8723 (typed_splay_tree::splay_tree_key, typed_splay_tree::splay_tree_value,
8724 typed_splay_tree::splay_tree_node_s, typed_splay_tree::KDEL,
8725 typed_splay_tree::VDEL, typed_splay_tree::splay_tree_delete_helper,
8726 typed_splay_tree::rotate_left, typed_splay_tree::rotate_right,
8727 typed_splay_tree::splay_tree_splay,
8728 typed_splay_tree::splay_tree_foreach_helper,
8729 typed_splay_tree::splay_tree_insert,
8730 typed_splay_tree::splay_tree_remove,
8731 typed_splay_tree::splay_tree_lookup,
8732 typed_splay_tree::splay_tree_predecessor,
8733 typed_splay_tree::splay_tree_successor,
8734 typed_splay_tree::splay_tree_min,
8735 typed_splay_tree::splay_tree_max): Took over from splay-tree.c/.h.
8736 (typed_splay_tree::root, typed_splay_tree::comp,
8737 typed_splay_tree::delete_key,
8738 typed_splay_tree::delete_value): New data members.
8739 * typed-splay-tree.c (selftest::test_str_to_int): Add a test for
8740 typed_splay_tree::remove.
8742 2018-06-15 Matthew Fortune <matthew.fortune@mips.com>
8744 * config/mips/mips.h (ASM_SPEC): Pass through -mcrc, -mno-crc,
8745 -mginv and -mno-ginv to the assembler.
8746 * config/mips/mips.opt (-mcrc): New option.
8748 * doc/invoke.text (-mcrc): Document.
8751 2018-06-15 Nick Clifton <nickc@redhat.com>
8754 * tree.c (escaped_string): New class. Converts an unescaped
8755 string into its escaped equivalent.
8756 (warn_deprecated_use): Use the new class to convert the
8757 deprecation message, if present.
8758 (test_escaped_strings): New self test.
8759 (test_c_tests): Add test_escaped_strings.
8760 * doc/extend.texi (deprecated): Add a note that the
8761 deprecation message is affected by the -fmessage-length
8762 option, and that control characters will be escaped.
8763 (#pragma GCC error): Document this pragma.
8764 (#pragma GCC warning): Likewise.
8765 * doc/invoke.texi (-fmessage-length): Document this option's
8766 effect on the #warning and #error preprocessor directives and
8767 the deprecated attribute.
8769 2018-06-15 Richard Biener <rguenther@suse.de>
8771 * tree-vect-slp.c (vect_slp_bb): Dump MSG_OPTIMIZED_LOCATIONS
8772 here, also noting vector size used.
8773 * tree-vectorizer.c (vectorize_loops): Adjust. Note vector
8774 size used in MSG_OPTIMIZED_LOCATIONS dump.
8775 (pass_slp_vectorize::execute): Adjust.
8777 2018-06-15 Claudiu Zissulescu <claziss@synopsys.com>
8780 * config/arc/arc.c (arc_return_address_register): Fix
8783 2018-06-15 Richard Biener <rguenther@suse.de>
8786 * tree-cfg.c (gimplify_build3): Do not strip sign conversions,
8787 leave useless conversion stripping to force_gimple_operand_gsi.
8788 (gimplify_build2): Likewise.
8789 (gimplify_build1): Likewise.
8791 2018-06-15 Richard Biener <rguenther@suse.de>
8794 * tree-cfg.c (move_stmt_op): unshare invariant addresses
8795 before adjusting their block.
8797 2018-06-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
8799 * config.gcc (riscv*-*-elf* | riscv*-*-rtems*): Use custom
8800 multilibs for *-*-rtems*.
8801 * config/riscv/t-rtems: New file.
8803 2018-06-14 Jakub Jelinek <jakub@redhat.com>
8806 * match.pd ((A +- CST1) +- CST2): Punt if last resort
8807 unsigned_type_for returns NULL.
8810 * lower-subreg.c (find_decomposable_subregs): Don't decompose float
8811 subregs of multi-word pseudos unless the float mode has word size.
8813 2018-06-14 Richard Biener <rguenther@suse.de>
8816 * tree-vect-generic.c (build_word_mode_vector_type): Remove
8817 duplicate and harmful type_hash_canon.
8818 * tree.c (type_hash_canon): Assert we didn't find ourselves.
8820 2018-06-14 Richard Biener <rguenther@suse.de>
8823 * tree-ssa-struct-alias.c (create_variable_info_for): Handle
8826 2018-06-14 Sebastian Huber <sebastian.huber@embedded-brains.de>
8828 * config/rtems.h (STDINT_LONG32): Define.
8830 2018-06-13 Matthew Fortune <matthew.fortune@mips.com>
8831 Prachi Godbole <prachi.godbole@imgtec.com>
8833 * config/mips/mips-cpus.def: Define P6600.
8834 * config/mips/mips-tables.opt: Regenerate.
8835 * config/mips/mips.c (mips_ucbranch_type): New enum.
8836 (mips_rtx_cost_data): Add support for P6600.
8837 (mips_issue_rate): Likewise.
8838 (mips_multipass_dfa_lookahead): Likewise.
8839 (mips_avoid_hazard): Likewise.
8840 (mips_reorg_process_insns): Likewise.
8841 (mips_classify_branch_p6600): New function.
8842 * config/mips/mips.h (TUNE_P6600): New define.
8843 (MIPS_ISA_LEVEL_SPEC): Infer mips64r6 from p6600.
8844 (ENABLE_LD_ST_PAIRS): Enable load/store bonding for p6600.
8845 * config/mips/mips.md: Include p6600.md.
8846 (processor): Add p6600.
8847 * config/mips/p6600.md: New file.
8848 * doc/invoke.texi: Add p6600 to supported architectures.
8850 2018-06-13 Martin Sebor <msebor@redhat.com>
8852 PR tree-optimization/86114
8853 * gimple-fold.c (gimple_fold_builtin_strlen): Only handle LHS
8855 * tree-ssa-strlen.c (maybe_set_strlen_range): Same.
8857 2018-06-13 Richard Biener <rguenther@suse.de>
8859 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern):
8860 Properly set vector type of the intermediate stmt.
8861 * tree-vect-stmts.c (vectorizable_operation): The destination
8862 var always has vectype_out type.
8864 2018-06-13 Jeff Law <law@redhat.com>
8866 * config/rl78/rl78.c (move_elim_pass): Use TDF_NONE rather than
8867 integer 0 for argument to print_rtl_with_bb.
8868 (rl78_reorg): Likewise.
8870 2018-06-13 David Malcolm <dmalcolm@redhat.com>
8872 * config/arc/arc.c (hwloop_optimize): Strengthen local "end_label"
8873 from rtx to rtx_insn *.
8874 * config/bfin/bfin.c (hwloop_optimize): Likewise for local
8876 (add_sched_insns_for_speculation): Likewise for local "target",
8877 converting usage of JUMP_LABEL to JUMP_LABEL_AS_INSN.
8878 * config/c6x/c6x.c (reorg_split_calls): Strengthen param "call_labels"
8879 from rtx_insn ** to rtx_code_label **.
8880 (reorg_emit_nops): Likewise.
8881 (c6x_reorg): Likewise for local "call_labels".
8882 * config/sh/sh-protos.h (get_dest_uid): Strengthen 1st param from
8884 * config/sh/sh.c (dump_table): Strengthen local "lab" from rtx to
8885 rtx_code_label *, adding safe_as_a <rtx_code_label *> casts to
8886 the loops over LABEL_REFS.
8887 (fixup_addr_diff_vecs): Add as_a <rtx_insn *> to usage of
8889 (barrier_align): Convert usage of JUMP_LABEL to JUMP_LABEL_AS_INSN.
8890 (get_dest_uid): Strengthen param "label" from rtx to rtx_insn *.
8891 (split_branches): Strengthen local "olabel" from rtx to
8892 rtx_insn *, adding a safe_as_a cast.
8893 * emit-rtl.c (next_real_insn): Strengthen param from "rtx"
8895 (add_insn_after): Likewise for first two params.
8896 (add_insn_before): Likewise.
8897 (remove_insn): Likewise for param.
8898 (emit_pattern_before_noloc): Likewise for second and third params.
8899 (emit_jump_insn_before_noloc): Convert NULL_RTX to NULL.
8900 (emit_call_insn_before_noloc): Likewise.
8901 (emit_debug_insn_before_noloc): Strengthen "before" param from "rtx"
8903 (emit_barrier_before): Likewise.
8904 (emit_label_before): Strengthen "label" param from "rtx" to
8905 "rtx_code_label *". Strengthen "before" param from "rtx" to
8907 (emit_insn_after_1): Strengthen "after" param from "rtx" to
8909 (emit_pattern_after_noloc): Likewise.
8910 (emit_insn_after_noloc): Likewise.
8911 (emit_jump_insn_after_noloc): Likewise.
8912 (emit_call_insn_after_noloc): Likewise.
8913 (emit_debug_insn_after_noloc): Likewise.
8914 (emit_barrier_after): Likewise.
8915 (emit_label_after): Likewise for both params.
8916 (emit_pattern_after_setloc): Likewise for "after" param. Convert
8917 "loc" param from "int" to "location_t".
8918 (emit_insn_after_setloc): Likewise.
8919 (emit_jump_insn_after_setloc): Likewise.
8920 (emit_call_insn_after_setloc): Likewise.
8921 (emit_debug_insn_after_setloc): Likewise.
8922 (emit_pattern_before_setloc): Likewise for "before" param. Convert
8923 "loc" param from "int" to "location_t".
8924 (emit_pattern_before): Convert NULL_RTX to NULL.
8925 (emit_insn_before_setloc): Convert "loc" param from "int" to
8927 (emit_jump_insn_before_setloc): Likewise.
8928 (emit_call_insn_before_setloc): Likewise.
8929 (emit_debug_insn_before_setloc): Strengthen "before" param from rtx to
8930 rtx_insn *. Convert "loc" param from "int" to "location_t".
8931 * rtl.h (emit_insn_before_setloc, emit_jump_insn_before_setloc,
8932 emit_call_insn_before_setloc, emit_debug_insn_before_setloc):
8933 Convert 3rd param from "int" to "location_t".
8934 (emit_barrier_before, emit_barrier_after, next_real_insn):
8935 Strengthen param from rtx to rtx_insn *.
8936 (emit_label_before): Strengthen 1st param from "rtx" to
8937 "rtx_code_label *". Strengthen 2nd param from "rtx" to
8939 (emit_insn_after_noloc, emit_jump_insn_after_noloc,
8940 emit_call_insn_after_noloc, emit_debug_insn_after_noloc):
8941 Strengthen 2nd param from "rtx" to "rtx_insn *".
8942 (emit_insn_after_setloc, emit_jump_insn_after_setloc)
8943 emit_call_insn_after_setloc, emit_debug_insn_after_setloc):
8944 Likewise. Convert 3rd param from "int" to "location_t".
8945 (emit_label_after): Strengthen 1st param from "rtx" to
8947 (next_real_insn, remove_insn): Strengthen param from "rtx" to
8949 (add_insn_before, add_insn_after): Strengthen 1st and 2nd params
8950 from "rtx" to "rtx_insn *".
8952 2018-06-13 Jan Hubicka <hubicka@gcc.gnu.org>
8954 * cgraph.c (cgraph_node::get_untransformed_body): Dump function
8955 bodies streamed in with -Q.
8956 * dumpfile.c (dump_files): Add lto-stream-out dump file.
8957 * dumpfile.h (tree_dump_index): Add lto_stream_out.
8958 * gimple-streamer-out.c: Include gimple-pretty-print.h
8959 (output_bb): Dump stmts streamed.
8960 * lto-section-out.c: Include print-tree.h
8961 (lto_begin_section): Dump sections created.
8962 (lto_output_decl_index): Dump decl encoded.
8963 * lto-streamer-out.c: Include print-tree.h
8964 (create_output_block): Dump output block created.
8965 (DFS::DFS_write_tree_body): Dump DFS SCCs streamed.
8966 (output_function): Dump function output.
8967 (output_constructor): Dump constructor streamed.
8968 (write_global_stream): Output indexes encoded.
8969 (produce_asm_for_decls): Dump streams encoded.
8970 * lto-streamer.c (streamer_dump_file): New global var.
8971 * lto-streamer.h (streamer_dump_file): Declare.
8972 * passes.c (ipa_write_summaries): Initialize streamer dump.
8973 * varpool.c (varpool_node::get_constructor): Dump constructors streamed
8976 2018-06-13 Eric Botcazou <ebotcazou@adacore.com>
8979 * config/i386/winnt.c (i386_pe_seh_cold_init): Do not emit negative
8980 offsets for register save directives. Emit a second batch of save
8981 directives, if need be, when the function accesses prior frames.
8983 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
8985 * config/arc/fpu.md (fmasf4): Force operand to register.
8986 (fnmasf4): Likewise.
8988 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
8990 * config/arc/arc-protos.h (arc_pad_return): Remove.
8991 * config/arc/arc.c (machine_function): Remove force_short_suffix
8993 (arc_print_operand): Adjust printing of '&'.
8994 (arc_verify_short): Remove conditional printing of short suffix.
8995 (arc_final_prescan_insn): Remove reference to size_reason.
8996 (pad_return): New function.
8997 (arc_reorg): Call pad_return.
8998 (arc_pad_return): Remove.
8999 (arc_init_machine_status): Remove reference to force_short_suffix.
9000 * config/arc/arc.md (vunspec): Add VUNSPEC_ARC_BLOCKAGE.
9001 (attr length): When attribute iscompact is true force to 2
9002 regardless; in the case of maybe check if we want to force the
9003 instruction to have 4 bytes length.
9004 (nopv): Change it to generate 4 byte long nop as well.
9005 (blockage): New pattern.
9006 (simple_return): Remove call to arc_pad_return.
9007 (p_return_i): Likewise.
9009 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
9011 * config/arc/elf.h (LINK_GCC_C_SEQUENCE_SPEC): Define.
9013 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
9015 * config/arc/builtins.def (SYNC): SYNC instruction is valid on all
9018 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
9020 * config/arc/arc.c (atomic_exchangesi): EX instruction is default
9021 for ARC700 and ARCv2.
9023 2018-06-13 Chenghua Xu <paul.hua.gm@gmail.com>
9026 * config/mips/loongson.md (vec_setv4hi): Gen_lowpart for
9027 operands[2] instead of operands[1].
9030 2018-06-12 Richard Sandiford <richard.sandiford@linaro.org>
9032 * lra-constraints.c (simplify_operand_subreg): In the paradoxical
9033 case, check whether the outer register overlaps an unallocatable
9034 register, not just whether it fits the required class.
9036 2018-06-12 Richard Sandiford <richard.sandiford@linaro.org>
9038 * poly-int.h (can_div_trunc_p): Add new overload in which all values
9040 * alias.c (get_addr): Extend CONST_INT handling to poly_int_rtx_p.
9041 (memrefs_conflict_p): Likewise.
9042 (init_alias_analysis): Likewise.
9043 * cfgexpand.c (expand_debug_expr): Likewise.
9044 * combine.c (combine_simplify_rtx, force_int_to_mode): Likewise.
9045 * cse.c (fold_rtx): Likewise.
9046 * explow.c (adjust_stack, anti_adjust_stack): Likewise.
9047 * expr.c (emit_block_move_hints): Likewise.
9048 (clear_storage_hints, push_block, emit_push_insn): Likewise.
9049 (store_expr_with_bounds, reduce_to_bit_field_precision): Likewise.
9050 (emit_group_load_1): Use rtx_to_poly_int64 for group offsets.
9051 (emit_group_store): Likewise.
9052 (find_args_size_adjust): Use strip_offset. Use rtx_to_poly_int64
9053 to read the PRE/POST_MODIFY increment.
9054 * calls.c (store_one_arg): Use strip_offset.
9055 * rtlanal.c (rtx_addr_can_trap_p_1): Extend CONST_INT handling to
9057 (set_noop_p): Use rtx_to_poly_int64 for the elements selected
9059 * simplify-rtx.c (avoid_constant_pool_reference): Use strip_offset.
9060 (simplify_binary_operation_1): Extend CONST_INT handling to
9062 * var-tracking.c (compute_cfa_pointer): Take a poly_int64 rather
9063 than a HOST_WIDE_INT.
9064 (hard_frame_pointer_adjustment): Change from HOST_WIDE_INT to
9066 (adjust_mems, add_stores): Update accodingly.
9067 (vt_canonicalize_addr): Track polynomial offsets.
9068 (emit_note_insn_var_location): Likewise.
9069 (vt_add_function_parameter): Likewise.
9070 (vt_initialize): Likewise.
9072 2018-06-12 Jeff Law <law@redhat.com>
9074 * config.gcc (alpha*-*-freebsd*): Remove.
9075 * config/alpha/freebsd.h: Remove.
9077 2018-06-12 David Malcolm <dmalcolm@redhat.com>
9080 * spellcheck-tree.c (levenshtein_distance): Rename to...
9081 (get_edit_distance): ...this, and update for underlying renaming.
9082 * spellcheck-tree.h (levenshtein_distance): Rename to...
9083 (get_edit_distance): ...this.
9084 * spellcheck.c (levenshtein_distance): Rename to...
9085 (get_edit_distance): ...this. Convert from Levenshtein distance
9086 to Damerau-Levenshtein distance by supporting transpositions of
9087 adjacent characters. Rename "v1" to "v_next" and "v0" to
9089 (selftest::levenshtein_distance_unit_test_oneway): Rename to...
9090 (selftest::test_edit_distance_unit_test_oneway): ...this, and
9091 update for underlying renaming.
9092 (selftest::levenshtein_distance_unit_test): Rename to...
9093 (selftest::test_get_edit_distance_unit): ...this, and update for
9094 underlying renaming.
9095 (selftest::test_find_closest_string): Add example from PR 69968
9096 where transposition helps
9097 (selftest::test_metric_conditions): Update for renaming.
9098 (selftest::test_metric_conditions): Likewise.
9099 (selftest::spellcheck_c_tests): Likewise.
9100 * spellcheck.h (levenshtein_distance): Rename both overloads to...
9101 (get_edit_distance): ...this.
9102 (best_match::consider): Update for renaming.
9104 2018-06-12 Martin Sebor <msebor@redhat.com>
9106 PR tree-optimization/85259
9107 * builtins.c (compute_objsize): Handle constant offsets.
9108 * gimple-ssa-warn-restrict.c (maybe_diag_offset_bounds): Return
9109 true iff a warning has been issued.
9110 * gimple.h (gimple_nonartificial_location): New function.
9111 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Call
9112 gimple_nonartificial_location and handle -Wno-system-headers.
9113 (handle_builtin_stxncpy): Same.
9115 2018-06-12 Martin Sebor <msebor@redhat.com>
9118 * fold-const.c (operand_equal_p): Handle SAVE_EXPR.
9120 2018-06-12 Will Schmidt <will_schmidt@vnet.ibm.com>
9122 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
9123 BUILTIN_VEC_XST entries for pointer to double and long long.
9125 2018-06-12 H.J. Lu <hongjiu.lu@intel.com>
9128 * config/i386/gnu-user.h (TARGET_THREAD_SPLIT_STACK_OFFSET):
9130 * config/i386/gnu-user64.h (TARGET_THREAD_SPLIT_STACK_OFFSET):
9133 2018-06-12 Martin Liska <mliska@suse.cz>
9135 * doc/options.texi: Document IntegerRange.
9137 2018-06-12 Martin Liska <mliska@suse.cz>
9139 * config/i386/i386.opt: Make MPX-related options as Deprecated.
9140 * opt-functions.awk: Handle Deprecated flag.
9141 * opts-common.c (decode_cmdline_option): Handle cl_deprecated
9143 (read_cmdline_option): Report warning for a deprecated option.
9144 * opts.h (struct cl_option): Add new field cl_deprecated.
9145 (CL_ERR_DEPRECATED): New.
9147 2018-06-12 Martin Liska <mliska@suse.cz>
9149 * doc/options.texi: Document Deprecated option flag.
9151 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
9153 * config/arc/arc-arch.h (arc_extras): New enum.
9154 (arc_cpu_t):Add field extra.
9155 (arc_cpu_types): Consider the extras.
9156 * config/arc/arc-cpus.def: Add extras info.
9157 * config/arc/arc-opts.h (processor_type): Consider extra field.
9158 * config/arc/arc.c (arc_override_options): Handle extra field.
9160 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
9162 * config/arc/arc-arch.h: Update ARC_OPTX macro.
9163 * config/arc/arc-options.def (ARC_OPTX): Introduce a new doc
9165 * config/arc/arc.c (arc_init): Update pic warning.
9166 (irq_range): Update irq range parsing warnings.
9167 (arc_override_options): Update various warning messages.
9168 (arc_handle_aux_attribute): Likewise.
9170 2018-06-12 Robert Suchanek <robert.suchanek@mips.com>
9172 * config/mips/i6400.md (i6400_fpu_fadd): Remove frint.
9174 2018-06-12 Jozef Lawrynowicz <jozef.l@mittosystems.com>
9176 * doc/sourcebuild.texi: Document usage of line number 0 in verify
9177 compiler messages directives.
9179 2018-06-12 Matthew Fortune <mfortune@gmail.com>
9181 * config/mips/mips-cpus.def: New MIPS_CPU for i6500.
9182 * config/mips/mips-tables.opt: Regenerate.
9183 * config/mips/mips.h (MIPS_ISA_LEVEL_SPEC): Mark i6500 as
9185 * doc/invoke.texi: Document -march=i6500.
9187 2018-06-12 Prachi Godbole <prachi.godbole@imgtec.com>
9189 * config/mips/i6400.md (i6400_gpmuldiv): Remove cpu_unit.
9190 (i6400_gpmul): Add cpu_unit.
9191 (i6400_gpdiv): Likewise.
9192 (i6400_msa_add_d): Update reservations.
9193 (i6400_msa_int_add) Likewise.
9194 (i6400_msa_short_logic3) Likewise.
9195 (i6400_msa_short_logic2) Likewise.
9196 (i6400_msa_short_logic) Likewise.
9197 (i6400_msa_move) Likewise.
9198 (i6400_msa_cmp) Likewise.
9199 (i6400_msa_short_float2) Likewise.
9200 (i6400_msa_div_d) Likewise.
9201 (i6400_msa_long_logic1) Likewise.
9202 (i6400_msa_long_logic2) Likewise.
9203 (i6400_msa_mult) Likewise.
9204 (i6400_msa_long_float2) Likewise.
9205 (i6400_msa_long_float4) Likewise.
9206 (i6400_msa_long_float5) Likewise.
9207 (i6400_msa_long_float8) Likewise.
9208 (i6400_fpu_fadd): Include frint type.
9209 (i6400_fpu_store): New define_insn_reservation.
9210 (i6400_fpu_load): Likewise.
9211 (i6400_fpu_move): Likewise.
9212 (i6400_fpu_fcmp): Likewise.
9213 (i6400_fpu_fmadd): Likewise.
9214 (i6400_int_mult): Include imul3nc type and update reservation.
9215 (i6400_int_div): Include idiv3 type and update reservation.
9216 (i6400_int_load): Update to check type not move_type.
9217 (i6400_int_store): Likewise.
9218 (i6400_int_prefetch): Set zero latency.
9220 2018-06-12 Eric Botcazou <ebotcazou@adacore.com>
9222 * gcc.c: Document new %@{...} sequence.
9223 (LINK_COMMAND_SPEC): Use it for the -L switches.
9224 (cpp_unique_options): Use it for the -I switches.
9225 (at_file_argbuf): New global variable.
9226 (in_at_file): Likewise.
9227 (alloc_args): Create at_file_argbuf.
9228 (clear_args): Truncate at_file_argbuf.
9229 (store_arg): If in_at_file, push the argument onto at_file_argbuf.
9230 (open_at_file): New function.
9231 (close_at_file): Likewise.
9232 (create_at_file): Delete.
9233 (do_spec_1) <'i'>: Use open_at_file/close_at_file.
9236 (validate_switches_from_spec): Deal with %@{...} sequence.
9237 (validate_switches): Likewise.
9238 (driver::finalize): Call clear_args.
9240 2018-06-11 Rasmus Villemoes <rasmus.villemoes@prevas.dk>
9242 * config/vx-common.h (USE_TM_CLONE_REGISTRY): #define to 0.
9244 2018-06-11 Martin Sebor <msebor@redhat.com>
9246 * doc/invoke.texi (-Wall): List -Wc++17-compat.
9247 (Wno-class-memaccess): Add @opindex.
9248 (Wno-templates, Wno-multiple-inheritance): Same.
9249 (Wno-virtual-inheritance, Wno-namespaces): Same.
9250 (Wno-pedantic, Wno-chkp, Wcoverage-mismatch): Same.
9251 (Wno-format-overflow, Wno-format-truncation): Same.
9252 (Wno-unused-local-typedefs, Walloc-size-larger-than): Same.
9253 (Wno-alloc-size-larger-than, Wframe-larger-than): Same
9254 (Wno-stack-usage, Wno-c++-compat, Wno-c++11-compat): Same.
9255 (Wno-c++14-compat, Wno-c++17-compat, Wno-openmp-simd): Same.
9256 (Wno-unsuffixed-float-constants, Wno-addr-space-convert): Same.
9257 (Wno-misspelled-isr): Same.
9259 2018-06-11 Martin Sebor <msebor@redhat.com>
9261 * PR tree-optimization/86083
9262 * tree-ssa-strlen.c (handle_char_store): Use tree_expr_nonzero_p.
9264 2018-06-11 Zhouyi Zhou <zhouzhouyi@gmail.com>
9266 * tree-eh.c (lower_eh_constructs_2): Add a comma to comment.
9268 2018-06-11 Segher Boessenkool <segher@kernel.crashing.org>
9271 * config/rs6000/rs6000.md (*movdi_internal32): Put constraint modifiers
9272 on the correct operand.
9273 (*movdi_internal64): Ditto.
9275 2018-06-11 Martin Liska <mliska@suse.cz>
9277 PR tree-optimization/86089
9278 * tree-ssa-strlen.c (get_string_length): Move back removed hunk.
9280 2018-06-11 Julia Koval <julia.koval@intel.com>
9282 * config/i386/vaesintrin.h (_mm_aesdec_epi128, _mm_aesdeclast_epi128)
9283 _mm_aesenc_epi128, _mm_aesenclast_epi128: Remove.
9284 * config/i386/vpclmulqdqintrin.h (_mm_clmulepi64_epi128): Remove.
9286 2018-06-11 Olivier Hainque <hainque@adacore.com>
9288 * dwarf2out.c (gen_compile_unit_die): Fallout to DW_LANG_Ada83
9289 for Ada with strict dwarf2.
9291 2018-06-08 Peter Bergner <bergner@vnet.ibm.com>
9294 * config/rs6000/rs6000.c (mem_operand_gpr): Enable PRE_INC and PRE_DEC
9297 2018-06-08 Jan Hubicka <hubicka@ucw.cz>
9299 * dumpfile.c (FIRST_ME_AUTO_NUMBERED_DUMP): Bump to 4.
9301 2018-06-08 David Edelsohn <dje.gcc@gmail.com>
9303 * config/rs6000/rs6000.c (rs6000_passes_ieee128): Protect with #if
9306 2018-06-08 Martin Liska <mliska@suse.cz>
9308 * tree-cfg.h (debug_function): Fix argument type to match
9311 2018-06-08 Martin Liska <mliska@suse.cz>
9313 * config/powerpcspe/powerpcspe.c (rs6000_xcoff_visibility):
9314 Remove usage of MPX-related (and removed) fields.
9315 * config/rs6000/rs6000.c (rs6000_xcoff_visibility): Likewise.
9317 2018-06-08 David Malcolm <dmalcolm@redhat.com>
9319 * cfg.c (debug): Use TDF_NONE rather than 0.
9320 * cfghooks.c (debug): Likewise.
9321 * dumpfile.c (DUMP_FILE_INFO): Likewise; also for OPTGROUP.
9322 (struct dump_option_value_info): Convert to...
9323 (struct kv_pair): ...this template type.
9324 (dump_options): Convert to kv_pair<dump_flags_t>; use TDF_NONE
9326 (optinfo_verbosity_options): Likewise.
9327 (optgroup_options): Convert to kv_pair<optgroup_flags_t>; use
9329 (gcc::dump_manager::dump_register): Use optgroup_flags_t rather
9330 than int for "optgroup_flags" param.
9331 (dump_generic_expr_loc): Use dump_flags_t rather than int for
9333 (dump_dec): Likewise.
9334 (dump_finish): Use TDF_NONE rather than 0.
9335 (gcc::dump_manager::opt_info_enable_passes): Use optgroup_flags_t
9336 rather than int for "optgroup_flags" param. Use TDF_NONE rather
9337 than 0. Update for change to option_ptr.
9338 (opt_info_switch_p_1): Convert "optgroup_flags" param from int *
9339 to optgroup_flags_t *. Use TDF_NONE and OPTGROUP_NONE rather than
9340 0. Update for changes to optinfo_verbosity_options and
9342 (opt_info_switch_p): Convert optgroup_flags from int to
9344 (dump_basic_block): Use dump_flags_t rather than int
9345 for "dump_kind" param.
9346 * dumpfile.h (TDF_ADDRESS, TDF_SLIM, TDF_RAW, TDF_DETAILS,
9347 TDF_STATS, TDF_BLOCKS, TDF_VOPS, TDF_LINENO, TDF_UID)
9348 TDF_STMTADDR, TDF_GRAPH, TDF_MEMSYMS, TDF_RHS_ONLY, TDF_ASMNAME,
9349 TDF_EH, TDF_NOUID, TDF_ALIAS, TDF_ENUMERATE_LOCALS, TDF_CSELIB,
9350 TDF_SCEV, TDF_GIMPLE, TDF_FOLDING, MSG_OPTIMIZED_LOCATIONS,
9351 MSG_MISSED_OPTIMIZATION, MSG_NOTE, MSG_ALL, TDF_COMPARE_DEBUG,
9352 TDF_NONE): Convert from macros to...
9353 (enum dump_flag): ...this new enum.
9354 (dump_flags_t): Update to use enum.
9355 (operator|, operator&, operator~, operator|=, operator&=):
9356 Implement for dump_flags_t.
9357 (OPTGROUP_NONE, OPTGROUP_IPA, OPTGROUP_LOOP, OPTGROUP_INLINE,
9358 OPTGROUP_OMP, OPTGROUP_VEC, OPTGROUP_OTHER, OPTGROUP_ALL):
9359 Convert from macros to...
9360 (enum optgroup_flag): ...this new enum.
9361 (optgroup_flags_t): New typedef.
9362 (operator|, operator|=): Implement for optgroup_flags_t.
9363 (struct dump_file_info): Convert field "alt_flags" to
9364 dump_flags_t. Convert field "optgroup_flags" to
9366 (dump_basic_block): Use dump_flags_t rather than int for param.
9367 (dump_generic_expr_loc): Likewise.
9368 (dump_dec): Likewise.
9369 (dump_register): Convert param "optgroup_flags" to
9371 (opt_info_enable_passes): Likewise.
9372 * early-remat.c (early_remat::dump_edge_list): Use TDF_NONE rather
9374 * gimple-pretty-print.c (debug): Likewise.
9375 * gimple-ssa-store-merging.c (bswap_replace): Likewise.
9376 (merged_store_group::apply_stores): Likewise.
9377 * gimple-ssa-strength-reduction.c (insert_initializers): Likewise.
9378 * gimple.c (verify_gimple_pp): Likewise.
9379 * graphite-poly.c (print_pbb_body): Likewise.
9380 * passes.c (pass_manager::register_one_dump_file): Convert
9381 local "optgroup_flags" to optgroup_flags_t.
9382 * print-tree.c (print_node): Use TDF_NONE rather than 0.
9384 (debug_body): Likewise.
9385 * tree-pass.h (struct pass_data): Convert field "optgroup_flags"
9386 to optgroup_flags_t.
9387 * tree-pretty-print.c (print_struct_decl): Use TDF_NONE rather
9389 * tree-ssa-math-opts.c (convert_mult_to_fma_1): Likewise.
9390 (convert_mult_to_fma): Likewise.
9391 * tree-ssa-reassoc.c (undistribute_ops_list): Likewise.
9392 * tree-ssa-sccvn.c (vn_eliminate): Likewise.
9393 * tree-vect-data-refs.c (dump_lower_bound): Convert param
9394 "dump_kind" to dump_flags_t.
9396 2018-06-08 Segher Boessenkool <segher@kernel.crashing.org>
9398 * config/rs6000/rs6000.c (min, max): Delete.
9400 2018-06-08 Segher Boessenkool <segher@kernel.crashing.org>
9402 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete mention of
9403 -mabi=spe and -mabi=no-spe.
9405 2018-06-08 Martin Liska <mliska@suse.cz>
9407 * ipa-pure-const.c (propagate_pure_const): Use ::get at places
9408 where we expect an existing summary.
9410 2018-06-08 Martin Liska <mliska@suse.cz>
9412 * ipa-inline-analysis.c (simple_edge_hints): Use ::get method.
9413 * ipa-inline.h (estimate_edge_growth): Likewise.
9415 2018-06-08 Martin Liska <mliska@suse.cz>
9417 * cgraph.c (function_version_hasher::hash): Use
9418 cgraph_node::get_uid ().
9419 (function_version_hasher::equal):
9420 * cgraph.h (cgraph_node::get_uid): New method.
9421 * ipa-inline.c (update_caller_keys): Use
9422 cgraph_node::get_uid ().
9423 (update_callee_keys): Likewise.
9424 * ipa-utils.c (searchc): Likewise.
9425 (ipa_reduced_postorder): Likewise.
9426 * lto-cgraph.c (input_node): Likewise.
9427 * passes.c (is_pass_explicitly_enabled_or_disabled): Likewise.
9428 * symbol-summary.h (symtab_insertion): Likewise.
9429 (symtab_removal): Likewise.
9430 (symtab_duplication): Likewise.
9431 * tree-pretty-print.c (dump_function_header): Likewise.
9432 * tree-sra.c (convert_callers_for_node): Likewise.
9434 2018-06-08 Martin Liska <mliska@suse.cz>
9436 * cgraph.c (symbol_table::create_edge): Always assign a new
9438 (symbol_table::free_edge): Do not recycle numbers.
9439 * cgraph.h (cgraph_edge::get): New method.
9440 * symbol-summary.h (symtab_removal): Use it.
9441 (symtab_duplication): Likewise.
9442 (call_summary::hashable_uid): Remove.
9444 2018-06-08 Martin Liska <mliska@suse.cz>
9446 * ipa-inline-analysis.c (inline_edge_removal_hook): Remove.
9447 (initialize_growth_caches): Remove.
9448 (free_growth_caches): Likewise.
9449 (do_estimate_edge_time): Use edge_growth_cache.
9450 (do_estimate_edge_size): Likewise.
9451 (do_estimate_edge_hints): Likewise.
9452 * ipa-inline.c (reset_edge_caches): Likewise.
9453 (recursive_inlining): Likewise.
9454 (inline_small_functions): Likewise.
9455 * ipa-inline.h (initialize_growth_caches): Remove.
9456 (estimate_edge_size): Likewise.
9457 (estimate_edge_time): Likewise.
9458 (estimate_edge_hints): Likewise.
9459 (reset_edge_growth_cache): Likewise.
9460 * symbol-summary.h (call_summary::remove): New method.
9462 2018-06-08 Martin Liska <mliska@suse.cz>
9464 * ipa-cp.c (class edge_clone_summary): New summary.
9465 (grow_edge_clone_vectors): Remove.
9466 (ipcp_edge_duplication_hook): Remove.
9467 (class edge_clone_summary_t): New call_summary class.
9468 (ipcp_edge_removal_hook): Remove.
9469 (edge_clone_summary_t::duplicate): New function.
9470 (get_next_cgraph_edge_clone): Use edge_clone_summaries.
9471 (create_specialized_node): Likewise.
9472 (ipcp_driver): Initialize edge_clone_summaries and do not
9475 2018-06-08 Martin Liska <mliska@suse.cz>
9477 * symbol-summary.h (get): New function.
9478 (call_summary::m_initialize_when_cloning): New class member.
9480 2018-06-08 Martin Liska <mliska@suse.cz>
9482 * cgraph.c (cgraph_node::remove): Do not recycle uid.
9483 * cgraph.h (symbol_table::release_symbol): Do not pass uid.
9484 (symbol_table::allocate_cgraph_symbol): Do not set uid.
9485 * passes.c (uid_hash_t): Record removed_nodes by their uids.
9486 (remove_cgraph_node_from_order): Use the removed_nodes set.
9487 (do_per_function_toporder): Likwise.
9488 * symbol-summary.h (symtab_insertion): Use cgraph_node::uid
9489 instead of summary_uid.
9490 (symtab_removal): Likewise.
9491 (symtab_duplication): Likewise.
9493 2018-06-08 Martin Liska <mliska@suse.cz>
9495 * ipa-cp.c (ipcp_store_bits_results): Use
9496 ipcp_transformation_sum.
9497 (ipcp_store_vr_results): Likewise.
9498 * ipa-prop.c (ipcp_grow_transformations_if_necessary): Renamed
9500 (ipcp_transformation_initialize): ... this.
9501 (ipa_set_node_agg_value_chain):
9502 (ipa_node_params_t::duplicate): Use ipcp_transformation_sum.
9503 (write_ipcp_transformation_info): Likewise.
9504 (read_ipcp_transformation_info): Likewise.
9505 (ipcp_update_bits): Likewise.
9506 (ipcp_update_vr): Likewise.
9507 (ipcp_transform_function): Likewise.
9508 * ipa-prop.h: Rename ipcp_transformation_summary to
9509 ipcp_transformation.
9510 (class ipcp_transformation_t): New function summary.
9511 (ipcp_get_transformation_summary): Use ipcp_transformation_sum.
9512 (ipa_get_agg_replacements_for_node): Likewise.
9514 2018-06-08 Martin Liska <mliska@suse.cz>
9516 * ipa-pure-const.c (struct funct_state_d): Do it class instead
9518 (class funct_state_summary_t): New function_summary class.
9519 (has_function_state): Remove.
9520 (get_function_state): Likewise.
9521 (set_function_state): Likewise.
9522 (add_new_function): Likewise.
9523 (funct_state_summary_t::insert): New function.
9524 (duplicate_node_data): Remove.
9525 (remove_node_data): Remove.
9526 (funct_state_summary_t::duplicate): New function.
9527 (register_hooks): Create new funct_state_summaries.
9528 (pure_const_generate_summary): Use it.
9529 (pure_const_write_summary): Likewise.
9530 (pure_const_read_summary): Likewise.
9531 (propagate_pure_const): Likewise.
9532 (propagate_nothrow): Likewise.
9533 (dump_malloc_lattice): Likewise.
9534 (propagate_malloc): Likewise.
9535 (execute): Do not register hooks, just remove summary
9537 (pass_ipa_pure_const::pass_ipa_pure_const): Simplify
9540 2018-06-08 Martin Liska <mliska@suse.cz>
9542 * ipa-reference.c (remove_node_data): Remove.
9543 (duplicate_node_data): Likewise.
9544 (class ipa_ref_var_info_summary_t): New class.
9545 (class ipa_ref_opt_summary_t): Likewise.
9546 (get_reference_vars_info): Use ipa_ref_var_info_summaries.
9547 (get_reference_optimization_summary): Use
9548 ipa_ref_opt_sum_summaries.
9549 (set_reference_vars_info): Remove.
9550 (set_reference_optimization_summary): Likewise.
9551 (ipa_init): Create summaries.
9552 (init_function_info): Use function summary.
9553 (ipa_ref_opt_summary_t::duplicate): New function.
9554 (ipa_ref_opt_summary_t::remove): New function.
9555 (get_read_write_all_from_node): Fix GNU coding style.
9556 (propagate): Use function summary.
9557 (write_node_summary_p): Fix GNU coding style.
9558 (stream_out_bitmap): Likewise.
9559 (ipa_reference_read_optimization_summary): Use function summary.
9560 (ipa_reference_c_finalize): Do not release hooks.
9562 2018-06-08 Martin Liska <mliska@suse.cz>
9564 * ipa-fnsummary.c (dump_ipa_call_summary): Use ::get method.
9565 (analyze_function_body): Extract multiple calls of get_create.
9566 * ipa-inline-analysis.c (simple_edge_hints): Likewise.
9567 * ipa-inline.c (recursive_inlining): Use ::get method.
9568 * ipa-inline.h (estimate_edge_growth): Likewise.
9570 2018-06-08 Martin Liska <mliska@suse.cz>
9572 * hsa-common.h (enum hsa_function_kind): Rename HSA_NONE to
9574 (hsa_function_summary::hsa_function_summary): Use the new enum
9576 (hsa_gpu_implementation_p): Use hsa_summaries::get.
9577 * hsa-gen.c (hsa_get_host_function): Likewise.
9578 (get_brig_function_name): Likewise.
9579 * ipa-hsa.c (process_hsa_functions): Likewise.
9580 (ipa_hsa_write_summary): Likewise.
9581 * symbol-summary.h (symtab_duplication): Use ::get function/
9582 (get): New function.
9584 2018-06-08 Martin Liska <mliska@suse.cz>
9586 * config/i386/i386.c (ix86_can_inline_p): Use get_create instead
9588 * hsa-common.c (hsa_summary_t::link_functions): Likewise.
9589 (hsa_register_kernel): Likewise.
9590 * hsa-common.h (hsa_gpu_implementation_p): Likewise.
9591 * hsa-gen.c (hsa_get_host_function): Likewise.
9592 (get_brig_function_name): Likewise.
9593 (generate_hsa): Likewise.
9594 (pass_gen_hsail::execute): Likewise.
9595 * ipa-cp.c (ipcp_cloning_candidate_p): Likewise.
9596 (devirtualization_time_bonus): Likewise.
9597 (ipcp_propagate_stage): Likewise.
9598 * ipa-fnsummary.c (redirect_to_unreachable): Likewise.
9599 (edge_set_predicate): Likewise.
9600 (evaluate_conditions_for_known_args): Likewise.
9601 (evaluate_properties_for_edge): Likewise.
9602 (ipa_fn_summary::reset): Likewise.
9603 (ipa_fn_summary_t::duplicate): Likewise.
9604 (dump_ipa_call_summary): Likewise.
9605 (ipa_dump_fn_summary): Likewise.
9606 (analyze_function_body): Likewise.
9607 (compute_fn_summary): Likewise.
9608 (estimate_edge_devirt_benefit): Likewise.
9609 (estimate_edge_size_and_time): Likewise.
9610 (estimate_calls_size_and_time): Likewise.
9611 (estimate_node_size_and_time): Likewise.
9612 (inline_update_callee_summaries): Likewise.
9613 (remap_edge_change_prob): Likewise.
9614 (remap_edge_summaries): Likewise.
9615 (ipa_merge_fn_summary_after_inlining): Likewise.
9616 (ipa_update_overall_fn_summary): Likewise.
9617 (read_ipa_call_summary): Likewise.
9618 (inline_read_section): Likewise.
9619 (write_ipa_call_summary): Likewise.
9620 (ipa_fn_summary_write): Likewise.
9621 (ipa_free_fn_summary): Likewise.
9622 * ipa-hsa.c (process_hsa_functions): Likewise.
9623 (ipa_hsa_write_summary): Likewise.
9624 (ipa_hsa_read_section): Likewise.
9625 * ipa-icf.c (sem_function::merge): Likewise.
9626 * ipa-inline-analysis.c (simple_edge_hints): Likewise.
9627 (do_estimate_edge_time): Likewise.
9628 (estimate_size_after_inlining): Likewise.
9629 (estimate_growth): Likewise.
9630 (growth_likely_positive): Likewise.
9631 * ipa-inline-transform.c (clone_inlined_nodes): Likewise.
9632 (inline_call): Likewise.
9633 * ipa-inline.c (caller_growth_limits): Likewise.
9634 (can_inline_edge_p): Likewise.
9635 (can_inline_edge_by_limits_p): Likewise.
9636 (compute_uninlined_call_time): Likewise.
9637 (compute_inlined_call_time): Likewise.
9638 (want_inline_small_function_p): Likewise.
9639 (edge_badness): Likewise.
9640 (update_caller_keys): Likewise.
9641 (update_callee_keys): Likewise.
9642 (recursive_inlining): Likewise.
9643 (inline_small_functions): Likewise.
9644 (inline_to_all_callers_1): Likewise.
9645 (dump_overall_stats): Likewise.
9646 (early_inline_small_functions): Likewise.
9647 (early_inliner): Likewise.
9648 * ipa-inline.h (estimate_edge_growth): Likewise.
9649 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
9650 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
9651 * ipa-prop.h (IPA_NODE_REF): Likewise.
9652 (IPA_EDGE_REF): Likewise.
9653 * ipa-pure-const.c (malloc_candidate_p): Likewise.
9654 (propagate_malloc): Likewise.
9655 * ipa-split.c (execute_split_functions): Likewise.
9656 * symbol-summary.h: Rename get to get_create.
9658 (get_create): Likewise.
9659 * tree-sra.c (ipa_sra_preliminary_function_checks): Likewise.
9661 2018-06-08 Martin Liska <mliska@suse.cz>
9663 * symbol-summary.h (release): Move definition out of class
9665 (symtab_removal): Likewise.
9666 (symtab_duplication): Likewise.
9668 2018-06-08 Martin Liska <mliska@suse.cz>
9670 * symbol-summary.h (function_summary): Move constructor
9671 implementation out of class declaration.
9672 (release): Likewise.
9673 (symtab_insertion): Likewise.
9674 (symtab_removal): Likewise.
9675 (symtab_duplication): Likewise.
9678 2018-06-08 Martin Liska <mliska@suse.cz>
9680 * Makefile.in: Remove support for MPX (macros, related functions,
9681 fields in cgraph_node, ...).
9682 * builtin-types.def (BT_BND): Likewise.
9683 (BT_FN_BND_CONST_PTR): Likewise.
9684 (BT_FN_CONST_PTR_BND): Likewise.
9685 (BT_FN_VOID_PTR_BND): Likewise.
9686 (BT_FN_BND_CONST_PTR_SIZE): Likewise.
9687 (BT_FN_VOID_CONST_PTR_BND_CONST_PTR): Likewise.
9688 * builtins.c (expand_builtin_memcpy_with_bounds): Likewise.
9689 (expand_builtin_mempcpy_with_bounds): Likewise.
9690 (expand_builtin_memset_with_bounds): Likewise.
9691 (expand_builtin_memset_args): Likewise.
9692 (std_expand_builtin_va_start): Likewise.
9693 (expand_builtin): Likewise.
9694 (expand_builtin_with_bounds): Likewise.
9695 * builtins.def (DEF_BUILTIN_CHKP): Likewise.
9696 (DEF_LIB_BUILTIN_CHKP): Likewise.
9697 (DEF_EXT_LIB_BUILTIN_CHKP): Likewise.
9698 (DEF_CHKP_BUILTIN): Likewise.
9699 (BUILT_IN_MEMCPY): Likewise.
9700 (BUILT_IN_MEMMOVE): Likewise.
9701 (BUILT_IN_MEMPCPY): Likewise.
9702 (BUILT_IN_MEMSET): Likewise.
9703 (BUILT_IN_STPCPY): Likewise.
9704 (BUILT_IN_STRCAT): Likewise.
9705 (BUILT_IN_STRCHR): Likewise.
9706 (BUILT_IN_STRCPY): Likewise.
9707 (BUILT_IN_STRLEN): Likewise.
9708 (BUILT_IN_MEMCPY_CHK): Likewise.
9709 (BUILT_IN_MEMMOVE_CHK): Likewise.
9710 (BUILT_IN_MEMPCPY_CHK): Likewise.
9711 (BUILT_IN_MEMSET_CHK): Likewise.
9712 (BUILT_IN_STPCPY_CHK): Likewise.
9713 (BUILT_IN_STRCAT_CHK): Likewise.
9714 (BUILT_IN_STRCPY_CHK): Likewise.
9715 * calls.c (store_bounds): Likewise.
9716 (emit_call_1): Likewise.
9717 (special_function_p): Likewise.
9718 (maybe_warn_nonstring_arg): Likewise.
9719 (initialize_argument_information): Likewise.
9720 (finalize_must_preallocate): Likewise.
9721 (compute_argument_addresses): Likewise.
9722 (expand_call): Likewise.
9723 * cfgexpand.c (expand_call_stmt): Likewise.
9724 (expand_return): Likewise.
9725 (expand_gimple_stmt_1): Likewise.
9726 (pass_expand::execute): Likewise.
9727 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
9728 (cgraph_node::remove): Likewise.
9729 (cgraph_node::dump): Likewise.
9730 (cgraph_node::verify_node): Likewise.
9731 * cgraph.h (chkp_function_instrumented_p): Likewise.
9732 (symtab_node::get_alias_target): Likewise.
9733 (cgraph_node::can_remove_if_no_direct_calls_and_refs_p): Likewise.
9734 (cgraph_local_p): Likewise.
9735 * cgraphbuild.c (cgraph_edge::rebuild_edges): Likewise.
9736 (cgraph_edge::rebuild_references): Likewise.
9737 * cgraphunit.c (varpool_node::finalize_decl): Likewise.
9738 (walk_polymorphic_call_targets): Likewise.
9739 (cgraph_node::expand_thunk): Likewise.
9740 (symbol_table::output_weakrefs): Likewise.
9741 * common/config/i386/i386-common.c (OPTION_MASK_ISA2_GENERAL_REGS_ONLY_UNSET): Likewise.
9742 (ix86_handle_option): Likewise.
9743 * config/i386/constraints.md: Likewise.
9744 * config/i386/i386-builtin-types.def (BND): Likewise.
9748 * config/i386/i386-builtin.def (BDESC_END): Likewise.
9749 (BDESC_FIRST): Likewise.
9751 * config/i386/i386-c.c (ix86_target_macros_internal): Likewise.
9752 * config/i386/i386-protos.h (ix86_bnd_prefixed_insn_p): Likewise.
9753 * config/i386/i386.c (enum reg_class): Likewise.
9754 (ix86_target_string): Likewise.
9755 (ix86_option_override_internal): Likewise.
9756 (ix86_conditional_register_usage): Likewise.
9757 (ix86_valid_target_attribute_inner_p): Likewise.
9758 (ix86_set_indirect_branch_type): Likewise.
9759 (ix86_set_current_function): Likewise.
9760 (ix86_function_arg_regno_p): Likewise.
9761 (init_cumulative_args): Likewise.
9762 (ix86_function_arg_advance): Likewise.
9763 (ix86_function_arg): Likewise.
9764 (ix86_pass_by_reference): Likewise.
9765 (ix86_function_value_regno_p): Likewise.
9766 (ix86_function_value_1): Likewise.
9767 (ix86_function_value_bounds): Likewise.
9768 (ix86_return_in_memory): Likewise.
9769 (ix86_setup_incoming_vararg_bounds): Likewise.
9770 (ix86_va_start): Likewise.
9771 (indirect_thunk_need_prefix): Likewise.
9772 (print_reg): Likewise.
9773 (ix86_print_operand): Likewise.
9774 (ix86_expand_call): Likewise.
9775 (ix86_output_function_return): Likewise.
9776 (reg_encoded_number): Likewise.
9777 (BDESC_VERIFYS): Likewise.
9778 (ix86_init_mpx_builtins): Likewise.
9779 (ix86_init_builtins): Likewise.
9780 (ix86_emit_cmove): Likewise.
9781 (ix86_emit_move_max): Likewise.
9782 (ix86_expand_builtin): Likewise.
9783 (ix86_builtin_mpx_function): Likewise.
9784 (ix86_get_arg_address_for_bt): Likewise.
9785 (ix86_load_bounds): Likewise.
9786 (ix86_store_bounds): Likewise.
9787 (ix86_load_returned_bounds): Likewise.
9788 (ix86_store_returned_bounds): Likewise.
9789 (ix86_class_likely_spilled_p): Likewise.
9790 (ix86_hard_regno_mode_ok): Likewise.
9791 (x86_order_regs_for_local_alloc): Likewise.
9792 (ix86_mitigate_rop): Likewise.
9793 (ix86_bnd_prefixed_insn_p): Likewise.
9794 (ix86_mpx_bound_mode): Likewise.
9795 (ix86_make_bounds_constant): Likewise.
9796 (ix86_initialize_bounds): Likewise.
9797 (TARGET_LOAD_BOUNDS_FOR_ARG): Likewise.
9798 (TARGET_STORE_BOUNDS_FOR_ARG): Likewise.
9799 (TARGET_LOAD_RETURNED_BOUNDS): Likewise.
9800 (TARGET_STORE_RETURNED_BOUNDS): Likewise.
9801 (TARGET_CHKP_BOUND_MODE): Likewise.
9802 (TARGET_BUILTIN_CHKP_FUNCTION): Likewise.
9803 (TARGET_CHKP_FUNCTION_VALUE_BOUNDS): Likewise.
9804 (TARGET_CHKP_MAKE_BOUNDS_CONSTANT): Likewise.
9805 (TARGET_CHKP_INITIALIZE_BOUNDS): Likewise.
9806 * config/i386/i386.h (TARGET_MPX): Likewise.
9807 (TARGET_MPX_P): Likewise.
9808 (VALID_BND_REG_MODE): Likewise.
9809 (FIRST_BND_REG): Likewise.
9810 (LAST_BND_REG): Likewise.
9811 (enum reg_class): Likewise.
9812 (BND_REG_P): Likewise.
9813 (BND_REGNO_P): Likewise.
9814 (BNDmode): Likewise.
9815 (ADJUST_INSN_LENGTH): Likewise.
9816 * config/i386/i386.md: Likewise.
9817 * config/i386/i386.opt: Likewise.
9818 * config/i386/linux-common.h (LIBMPX_LIBS): Likewise.
9819 (defined): Likewise.
9820 (LINK_MPX): Likewise.
9821 (MPX_SPEC): Likewise.
9822 (LIBMPX_SPEC): Likewise.
9823 (LIBMPXWRAPPERS_SPEC): Likewise.
9824 (CHKP_SPEC): Likewise.
9825 * config/i386/predicates.md: Likewise.
9826 * dbxout.c (dbxout_type): Likewise.
9827 * doc/extend.texi: Likewise.
9828 * doc/invoke.texi: Likewise.
9829 * doc/md.texi: Likewise.
9830 * doc/tm.texi: Likewise.
9831 * doc/tm.texi.in: Likewise.
9832 * dwarf2out.c (is_base_type): Likewise.
9833 (gen_formal_types_die): Likewise.
9834 (gen_subprogram_die): Likewise.
9835 (gen_type_die_with_usage): Likewise.
9836 (gen_decl_die): Likewise.
9837 (dwarf2out_late_global_decl): Likewise.
9838 * expr.c (expand_assignment): Likewise.
9839 (emit_storent_insn): Likewise.
9840 (store_expr_with_bounds): Likewise.
9841 (store_expr): Likewise.
9842 (expand_expr_real_1): Likewise.
9843 * expr.h (store_expr_with_bounds): Likewise.
9844 * function.c (use_register_for_decl): Likewise.
9845 (struct bounds_parm_data): Likewise.
9846 (assign_parms_augmented_arg_list): Likewise.
9847 (assign_parm_find_entry_rtl): Likewise.
9848 (assign_parm_is_stack_parm): Likewise.
9849 (assign_parm_load_bounds): Likewise.
9850 (assign_bounds): Likewise.
9851 (assign_parms): Likewise.
9852 (expand_function_start): Likewise.
9853 * gcc.c (CHKP_SPEC): Likewise.
9854 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
9855 * gimple-ssa-warn-restrict.c (builtin_access::builtin_access): Likewise.
9856 (wrestrict_dom_walker::check_call): Likewise.
9857 * gimple.c (gimple_build_call_from_tree): Likewise.
9858 * gimple.h (enum gf_mask): Likewise.
9859 (gimple_call_with_bounds_p): Likewise.
9860 (gimple_call_set_with_bounds): Likewise.
9861 * gimplify.c (gimplify_init_constructor): Likewise.
9862 * ipa-cp.c (initialize_node_lattices): Likewise.
9863 (propagate_constants_across_call): Likewise.
9864 (find_more_scalar_values_for_callers_subset): Likewise.
9865 * ipa-hsa.c (process_hsa_functions): Likewise.
9866 * ipa-icf-gimple.c (func_checker::compare_gimple_call): Likewise.
9867 * ipa-icf.c (sem_function::merge): Likewise.
9868 * ipa-inline.c (early_inliner): Likewise.
9869 * ipa-pure-const.c (warn_function_noreturn): Likewise.
9870 (warn_function_cold): Likewise.
9871 (propagate_pure_const): Likewise.
9872 * ipa-ref.h (enum GTY): Likewise.
9873 * ipa-split.c (find_retbnd): Likewise.
9874 (consider_split): Likewise.
9875 (split_function): Likewise.
9876 * ipa-visibility.c (cgraph_externally_visible_p): Likewise.
9877 * ipa.c (walk_polymorphic_call_targets): Likewise.
9878 (symbol_table::remove_unreachable_nodes): Likewise.
9879 (process_references): Likewise.
9880 (cgraph_build_static_cdtor_1): Likewise.
9881 * lto-cgraph.c (lto_output_node): Likewise.
9882 (output_refs): Likewise.
9883 (compute_ltrans_boundary): Likewise.
9884 (input_overwrite_node): Likewise.
9885 (input_node): Likewise.
9886 (input_cgraph_1): Likewise.
9887 * params.def (PARAM_CHKP_MAX_CTOR_SIZE): Likewise.
9888 * passes.c (pass_manager::execute_early_local_passes): Likewise.
9889 (class pass_chkp_instrumentation_passes): Likewise.
9890 (make_pass_chkp_instrumentation_passes): Likewise.
9891 * passes.def: Likewise.
9892 * rtl.h (struct GTY): Likewise.
9893 (CALL_EXPR_WITH_BOUNDS_P): Likewise.
9894 * stor-layout.c (layout_type): Likewise.
9895 * symtab.c: Likewise.
9896 * target.def: Likewise.
9897 * targhooks.c (default_chkp_bound_type): Likewise.
9898 (default_chkp_bound_mode): Likewise.
9899 (default_builtin_chkp_function): Likewise.
9900 (default_chkp_function_value_bounds): Likewise.
9901 (default_chkp_make_bounds_constant): Likewise.
9902 (default_chkp_initialize_bounds): Likewise.
9903 * targhooks.h (default_chkp_bound_type): Likewise.
9904 (default_chkp_bound_mode): Likewise.
9905 (default_builtin_chkp_function): Likewise.
9906 (default_chkp_function_value_bounds): Likewise.
9907 (default_chkp_make_bounds_constant): Likewise.
9908 (default_chkp_initialize_bounds): Likewise.
9909 * toplev.c (compile_file): Likewise.
9910 (process_options): Likewise.
9911 * tree-core.h (DEF_BUILTIN): Likewise.
9912 (DEF_BUILTIN_CHKP): Likewise.
9913 * tree-inline.c (declare_return_variable): Likewise.
9914 (remap_gimple_stmt): Likewise.
9915 (copy_bb): Likewise.
9916 (initialize_inlined_parameters): Likewise.
9917 (expand_call_inline): Likewise.
9918 * tree-pass.h (make_pass_ipa_chkp_versioning): Likewise.
9919 (make_pass_ipa_chkp_early_produce_thunks): Likewise.
9920 (make_pass_ipa_chkp_produce_thunks): Likewise.
9921 (make_pass_chkp): Likewise.
9922 (make_pass_chkp_opt): Likewise.
9923 (make_pass_chkp_instrumentation_passes): Likewise.
9924 * tree-pretty-print.c (dump_generic_node): Likewise.
9925 * tree-ssa-ccp.c (insert_clobber_before_stack_restore): Likewise.
9926 * tree-ssa-dce.c (propagate_necessity): Likewise.
9927 (eliminate_unnecessary_stmts): Likewise.
9928 * tree-ssa-pre.c (create_expression_by_pieces): Likewise.
9929 * tree-ssa-sccvn.c (copy_reference_ops_from_call): Likewise.
9930 * tree-ssa-sccvn.h: Likewise.
9931 * tree-ssa-strlen.c (get_string_length): Likewise.
9932 (valid_builtin_call): Likewise.
9933 (adjust_last_stmt): Likewise.
9934 (handle_builtin_strchr): Likewise.
9935 (handle_builtin_strcpy): Likewise.
9936 (handle_builtin_stxncpy): Likewise.
9937 (handle_builtin_memcpy): Likewise.
9938 (handle_builtin_strcat): Likewise.
9939 (strlen_check_and_optimize_stmt): Likewise.
9940 * tree-stdarg.c (expand_ifn_va_arg_1): Likewise.
9941 * tree-streamer-in.c: Likewise.
9942 * tree-streamer.c (record_common_node): Likewise.
9943 * tree.c (tree_code_size): Likewise.
9944 (wide_int_to_tree_1): Likewise.
9945 (type_contains_placeholder_1): Likewise.
9946 (build_common_tree_nodes): Likewise.
9947 * tree.def (POINTER_BOUNDS_TYPE): Likewise.
9948 * tree.h (POINTER_BOUNDS_TYPE_P): Likewise.
9949 (POINTER_BOUNDS_P): Likewise.
9950 (BOUNDED_TYPE_P): Likewise.
9951 (BOUNDED_P): Likewise.
9952 (CALL_WITH_BOUNDS_P): Likewise.
9953 (pointer_bounds_type_node): Likewise.
9954 * value-prof.c (gimple_ic): Likewise.
9955 * var-tracking.c (vt_add_function_parameters): Likewise.
9956 * varasm.c (make_decl_rtl): Likewise.
9957 (assemble_start_function): Likewise.
9958 (output_constant): Likewise.
9959 (maybe_assemble_visibility): Likewise.
9960 * varpool.c (ctor_for_folding): Likewise.
9961 * chkp-builtins.def: Remove.
9962 * ipa-chkp.c: Remove.
9963 * ipa-chkp.h: Remove.
9964 * rtl-chkp.c: Remove.
9965 * rtl-chkp.h: Remove.
9966 * tree-chkp-opt.c: Remove.
9967 * tree-chkp.c: Remove.
9968 * tree-chkp.h: Remove.
9970 2018-06-07 Carl Love <cel@us.ibm.com>
9972 * config/rs6000/vsx.md (vextract_fp_from_shorth,
9973 vextract_fp_from_shortl): Add BE support.
9975 2018-06-07 Paul Koning <ni1d@arrl.net>
9977 * compare-elim.c (try_merge_compare): Don't merge compare if
9978 address contains a side effect.
9979 (try_eliminate_compare): Likewise.
9981 2018-06-07 Olga Makhotina <olga.makhotina@intel.com>
9983 * config.gcc: Support "tremont".
9984 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "tremont".
9985 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
9987 * config/i386/i386.c (m_TREMONT): Define.
9988 (processor_target_table): Add "tremont".
9989 (PTA_TREMONT): Define.
9990 (ix86_lea_outperforms): Add TARGET_TREMONT.
9991 (get_builtin_code_for_version): Handle PROCESSOR_TREMONT.
9992 (fold_builtin_cpu): Add M_INTEL_TREMONT, replace M_INTEL_GOLDMONT
9993 and M_INTEL_GOLDMONT_PLUS.
9994 (fold_builtin_cpu): Add "tremont".
9995 (ix86_add_stmt_cost): Add TARGET_TREMONT.
9996 (ix86_option_override_internal): Add "tremont".
9997 * config/i386/i386.h (processor_costs): Define TARGET_TREMONT.
9998 (processor_type): Add PROCESSOR_TREMONT.
9999 * config/i386/x86-tune.def: Add m_TREMONT.
10000 * doc/invoke.texi: Add tremont as x86 -march=/-mtune= CPU type.
10002 2018-06-07 Jozef Lawrynowicz <jozef.l@mittosystems.com>
10004 * config/msp430/msp430.c (msp430_mcu_name): Set the "i" in the
10005 symbol defined for msp430i* devices to be lower case.
10007 2018-06-07 Richard Biener <rguenther@suse.de>
10009 * graphite-sese-to-poly.c (extract_affine): Avoid unneded
10010 wrapping. Properly wrap the result of a BIT_NOT_EXPR.
10011 Properly wrap signed arithmetic if overflow wraps.
10013 2018-06-07 Jakub Jelinek <jakub@redhat.com>
10015 PR tree-optimization/69615
10016 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If rhs2 is lhs
10017 of a cast from a same precision integral SSA_NAME in a bb dominated
10018 by first_bb, retry with rhs2 set to the rhs1 of the cast. Don't emit
10019 cast to utype if rhs2 has already a compatible type.
10021 2018-06-07 Richard Biener <rguenther@suse.de>
10023 PR tree-optimization/85935
10024 * graphite-scop-detection.c (find_params_in_bb): Analyze
10025 condition operands with respect to the correct loop. Assert
10026 the analysis doesn't fail.
10028 2018-06-04 Carl Love <cel@us.ibm.com>
10030 * config/rs6000/vsx.md (first_match_index_<mode>): Calculate index
10031 using natural element order. Use gen_lshrsi3 instead of gen_ashrsi3
10032 as it is slightly cheaper.
10033 (first_match_or_eos_index_<mode>):
10034 Calculate index using natural element order.
10035 (first_match_index_<mode>):
10036 Calculate index using natural element order.
10037 (first_match_or_eos_index_<mode>):
10038 Calculate index using natural order.
10039 (define_insn vclzlsbb): Change to define_insn vclzlsbb_<mode>.
10040 for BE and LE modes.
10041 * config/rs6000/rs6000-c.c: Rename P9V_BUILTIN_VCLZLSBB,
10042 P9V_BUILTIN_VCLZLSBB_V16QI.
10043 * config/rs6000/rs6000-builtin.def: Make VCLZLSBB mode
10046 2018-06-06 Kelvin Nilsen <kelvin@gcc.gnu.org>
10048 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
10049 indentation and line wrap for many prototypes. Add missing
10050 @smallexample directives around block of prototypes for vec_xl and
10053 2018-06-05 Michael Meissner <meissner@linux.ibm.com>
10055 * config/rs6000/rs6000.c (rs6000_passes_ieee128): New boolean to
10056 track if we pass or return IEEE 128-bit floating point.
10057 (ieee128_mangling_gcc_8_1): New boolean to say whether to generate
10058 C++ mangling that is compatible with GCC 8.1.
10059 (TARGET_ASM_GLOBALIZE_DECL_NAME): Override target hook.
10060 (init_cumulative_args): Note if we pass or return IEEE 128-bit
10061 floating point types.
10062 (rs6000_function_arg_advance_1): Likewise.
10063 (rs6000_mangle_type): Optionally generate mangled names that match
10064 what GCC 8.1 generated for IEEE 128-bit floating point types.
10065 (rs6000_globalize_decl_name): If we have an external function that
10066 passes or returns IEEE 128-bit types, generate a weak reference
10067 from the mangled name used in GCC 8.1 to the current mangled
10069 (rs6000_init_builtins): Make __ibm128 use the long double type if
10070 long double is IBM extended double. Make __float128 use the long
10071 double type if long double is IEEE 128-bit.
10074 * config/rs6000/rs6000-builtin.def (BU_IBM128_2): New helper
10075 macro for __ibm128 built-in functions.
10076 (PACK_IF): Add __ibm128 pack/unpack functions.
10077 (UNPACK_IF): Likewise.
10078 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Do not
10079 enable long double built-in functions if long double is IEEE
10080 128-bit floating point.
10081 (rs6000_invalid_builtin): Update long double built-in function
10083 (rs6000_expand_builtin): For PACK_IF and UNPACK_IF built-in
10084 functions, adjust the built-in function to use the long double
10085 built-in function if __ibm128 and long double are the same type.
10086 * doc/extend.texi (PowerPC builtins): Update documention for
10087 __builtin_{,un}pack_longdouble. Add documentation for
10088 __builtin_{,un}pack_ibm128.
10090 2018-06-06 Jim Wilson <jimw@sifive.com>
10092 * config/riscv/riscv.c (enum riscv_privilege_levels): New.
10093 (struct machine_function): New field interrupt_mode.
10094 (riscv_handle_type_attribute): New function. Add forward declaration.
10095 (riscv_attribute_table) <interrupt>: Use riscv_handle_type_attribute.
10096 (riscv_expand_epilogue): Check interrupt_mode field.
10097 (riscv_set_current_function): Check interrupt attribute args and
10098 set interrupt_mode field.
10099 * config/riscv/riscv.md (UNSPECV_SRET, UNSPECV_URET): New.
10100 (riscv_sret, riscv_uret): New.
10101 * doc/extend.texi (RISC-V Function Attributes) <interrupt>: Document
10102 new arguments to interrupt attribute.
10104 2018-06-06 Peter Bergner <bergner@vnet.ibm.com>
10107 * /config/rs6000/rs6000.h (ASM_CPU_SPEC): Add support for -mpower9.
10108 Don't handle -mcpu=power8 if -mpower9-vector is also used.
10110 2018-06-06 Kelvin Nilsen <kelvin@gcc.gnu.org>
10112 * config/rs6000/rs6000-builtin.def (VSX_BUILTIN_VEC_LD,
10113 VSX_BUILTIN_VEC_ST): Add comment to explain non-traditional uses.
10114 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
10115 several redundant entries.
10117 2018-06-06 David Malcolm <dmalcolm@redhat.com>
10119 * config/i386/i386-protos.h (ix86_expand_call): Strengthen return
10120 type from "rtx" to "rtx_insn *".
10121 * config/i386/i386.c (ix86_expand_split_stack_prologue): Likewise
10122 for local "call_insn", removing cast.
10123 (ix86_expand_call): Likewise, introducing a "call_insn" local.
10125 2018-06-06 Eric Botcazou <ebotcazou@adacore.com>
10127 PR tree-optimization/86066
10128 * gimple-ssa-store-merging.c (process_store): Do not bypass BIT_NOT_EXPR
10129 for BIT_INSERT_EXPR stores.
10131 2018-06-06 Richard Biener <rguenther@suse.de>
10133 PR tree-optimization/86062
10134 * tree-ssa-sccvn.c (vn_reference_lookup_3): Handle arbitrary
10135 component refs ontop
10136 of to be offsetted base.
10138 2018-06-06 Jozef Lawrynowicz <jozef.l@mittosystems.com>
10140 * config/msp430/msp430.c (msp430_attr): Allow interrupt handlers
10141 to be static and remove check on interrupt attribute name.
10143 2018-06-05 Kelvin Nilsen <kelvin@gcc.gnu.org>
10145 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Remove
10146 volatile qualifier from vec_lvsl and vec_lvsr argument prototypes.
10148 2018-06-05 Steve Ellcey <sellcey@cavium.com>
10151 * config/aarch64/aarch64-protos.h (aarch64_err_no_fpadvsimd): Remove
10153 * config/aarch64/aarch64-protos..c (aarch64_err_no_fpadvsimd):
10154 Remove second argument, change how error is called.
10155 (aarch64_layout_arg): Remove second argument from
10156 aarch64_err_no_fpadvsimd call.
10157 (aarch64_init_cumulative_args): Ditto.
10158 (aarch64_gimplify_va_arg_expr): Ditto.
10159 * config/aarch64/aarch64.md (mov<mode>): Ditto.
10161 2018-06-05 Uros Bizjak <ubizjak@gmail.com>
10163 * config/i386/i386.md (simple_return_indirect_internal): New expander.
10164 (*simple_return_indirect_internal<mode>): Rename from
10165 simple_return_indirect_internal. Use W mode iterator.
10166 (rstorssp): New expander.
10167 (*rstorssp<mode>): Rename from rstorssp. Use P mode iterator.
10168 (clrssbsy): New expander.
10169 (*clrssbsy<mode>): Rename from clrssbsy. Use P mode iterator.
10171 2018-06-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
10173 * config/arm/arm_cmse.h (cmse_nsfptr_create): Change typeof to
10175 (cmse_check_pointed_object): Likewise.
10177 2018-06-05 Martin Liska <mliska@suse.cz>
10179 PR gcov-profile/47618
10180 * doc/invoke.texi: Document how -fprofile-dir format
10183 2018-06-05 Richard Biener <rguenther@suse.de>
10185 * tree-cfgcleanup.c (cleanup_control_flow_pre): For edge
10186 removal pretend DOM info isn't available so we do not update
10187 it and only remove edges, not dominated blocks. Actually free
10188 DOM info in case we removed something. Remove unreachable blocks.
10189 (mfb_keep_latches): Work with either DOM info or marked backedges.
10190 (cleanup_tree_cfg_noloop): Do not remove unreachable blocks
10191 first. Mark backedges if DOM info isn't available.
10192 (Re-)compute DOM info after cleanup_control_flow_pre.
10194 2018-06-05 Richard Biener <rguenther@suse.de>
10196 * tree-cfg.c (struct locus_discrim_map): Store line, not location.
10197 (locus_discrim_hasher::hash): Adjust.
10198 (locus_discrim_hasher::equal): Likewise.
10199 (next_discriminator_for_locus): Work on line directly.
10200 (same_line_p): Pass in expanded locus1 as well.
10201 (assign_discriminators): Avoid redundant location expansions.
10203 2018-06-05 Richard Biener <rguenther@suse.de>
10205 PR tree-optimization/86046
10206 * tree-ssa.c (maybe_optimize_var): Clear DECL_GIMPLE_REG_P
10207 if required after clearing TREE_ADDRESSABLE.
10209 2018-06-05 Richard Biener <rguenther@suse.de>
10211 PR tree-optimization/86047
10212 * tree-ssa-loop.c (for_each_index): Glob handling of all
10213 decls and constants and really handle all of them.
10215 2018-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10218 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
10219 qualifier_void_pointer and qualifier_const_void_pointer.
10220 (arm_ldc_qualifiers, arm_stc_qualifiers): Use the above.
10221 (arm_init_builtins): Handle the above.
10222 * config/arm/arm_acle.h (__arm_cdp, __arm_ldc, __arm_ldcl, __arm_stc,
10223 __arm_stcl, __arm_mcr, __arm_cdp2, __arm_ldc2, __arm_ldcl2, __arm_stc2,
10224 __arm_stcl2,__arm_mcr2, __arm_mcrr, __arm_mcrr2): Remove return for
10227 2018-06-05 Martin Liska <mliska@suse.cz>
10229 * auto-profile.c (read_autofdo_file): Do not use
10230 gcov_ctr_summary struct.
10231 (afdo_callsite_hot_enough_for_early_inline): Likewise.
10232 * coverage.c (struct counts_entry): Likewise.
10233 (read_counts_file): Read just single summary entry.
10234 (get_coverage_counts): Use gcov_summary struct.
10235 * coverage.h (get_coverage_counts): Likewise.
10236 * gcov-dump.c (dump_working_sets): Likewise.
10237 (tag_summary): Dump just single summary.
10238 * gcov-io.c (gcov_write_summary): Write just histogram
10240 (gcov_read_summary): Read just single summary.
10241 (compute_working_sets): Use gcov_summary struct.
10242 * gcov-io.h (GCOV_TAG_SUMMARY_LENGTH): Remove usage
10243 of GCOV_COUNTERS_SUMMABLE.
10244 (GCOV_COUNTERS_SUMMABLE): Remove.
10245 (GCOV_FIRST_VALUE_COUNTER): Replace with
10246 GCOV_COUNTER_V_INTERVAL.
10247 (struct gcov_ctr_summary): Remove.
10248 (struct gcov_summary): Directly use fields of former
10250 (compute_working_sets): Use gcov_summary struct.
10251 * gcov.c (read_count_file): Do not use ctrs fields.
10252 * lto-cgraph.c (merge_profile_summaries): Use gcov_summary
10254 * lto-streamer.h (struct GTY): Make profile_info gcov_summary
10256 * profile.c: Likewise.
10257 * profile.h: Likewise.
10259 2018-06-05 Martin Liska <mliska@suse.cz>
10261 PR gcov-profile/84846
10262 * gcov.c (output_lines): Print working directory only
10263 in intermediate format.
10265 2018-06-05 Andreas Krebbel <krebbel@linux.ibm.com>
10267 * config/s390/s390-builtin-types.def: Add void function type.
10268 * config/s390/s390-builtins.def: Use the function type for the
10271 2018-06-04 Jim Wilson <jimw@sifive.com>
10273 * config/riscv/riscv-protos.h (riscv_expand_epilogue): Change bool arg
10275 * config/riscv/riscv.c (riscv_for_each_saved_reg): New args epilogue
10276 and maybe_eh_return. Change regno to unsigned int. Use new args to
10277 handle EH_RETURN_DATA_REGNO registers properly.
10278 (riscv_expand_prologue): Pass new args to riscv_for_each_saved_reg.
10279 (riscv_expand_epilogue): Update comment. Change argument name and
10280 type. Update code to use new name and type. Pass new args to
10281 riscv_for_each_saved_reg. Only use EH_RETURN_STACKADJ_RTX when
10283 * config/riscv/riscv.md (NORMAL_RETURN): New.
10284 (SIBCALL_RETURN, EXCEPTION_RETURN): New.
10285 (epilogue, sibcall_epilogue): Update riscv_expand_epilogue arg.
10286 (eh_return): Call gen_eh_return_internal and emit barrier.
10287 (eh_return_internal): Call riscv_expand_epilogue.
10289 2018-06-04 Eric Botcazou <ebotcazou@adacore.com>
10291 * gimple-ssa-store-merging.c (struct merged_store_group): Move up
10292 bit_insertion field and declare can_be_merged_into method.
10293 (merged_store_group::can_be_merged_into): New method.
10294 (imm_store_chain_info::coalesce_immediate): Call it to decide whether
10295 consecutive non-overlapping stores can be merged. Turn MEM_REF stores
10296 into BIT_INSERT_EXPR stores if the group contains a non-MEM_REF store.
10298 2018-06-04 Richard Biener <rguenther@suse.de>
10300 PR tree-optimization/85955
10301 * builtins.c (fold_builtin_sincos): Convert pointers to
10302 destination to appropriate type before dereferencing.
10304 2018-06-04 Segher Boessenkool <segher@kernel.crashing.org>
10306 * config/rs6000/rs6000.md (abs<mode>2 for FLOAT128): Handle IFmode.
10308 2018-06-04 Richard Sandiford <richard.sandiford@linaro.org>
10310 * expr.c (expand_expr_real_1): Force the operand into memory if
10311 its TYPE_MODE is BLKmode and if there is no integer mode for
10312 the number of bits being extracted.
10314 2018-06-04 Jakub Jelinek <jakub@redhat.com>
10318 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>_1):
10319 Use vptestnm rather than vptestm in (=Yc,v,C) variant.
10321 2018-06-04 Richard Biener <rguenther@suse.de>
10323 * tree-cfgcleanup.c (cleanup_tree_cfg_1): Fold into...
10324 (cleanup_tree_cfg_noloop): ... single caller. Do
10325 start_recording_case_labels later.
10327 2018-06-04 Sebastian Peryt <sebastian.peryt@intel.com>
10329 * config/i386/cldemoteintrin.h: Change define from _X86INTRIN_H_INCLUDED
10330 to _IMMINTRIN_H_INCLUDED.
10331 * config/i386/pconfigintrin.h: Ditto.
10332 * config/i386/waitpkgintrin.h: Ditto.
10333 * config/i386/immintrin.h: Add includes for sgxintrin.h,
10334 pconfigintrin.h, waitpkgintrin.h and cldemoteintrin.h.
10335 * config/i386/x86intrin.h: Remove includes for mintrin.h, xmmintrin.h,
10336 emmintrin.h, pmmintrin.h, tmmintrin.h, smmintrin.h, wmmintrin.h,
10337 bmiintrin.h, bmi2intrin.h, lzcntintrin.h, sgxintrin.h, pconfigintrin.h,
10338 waitpkgintrin.h and cldemoteintrin.h.
10340 2018-06-04 Richard Biener <rguenther@suse.de>
10342 PR tree-optimization/86038
10343 * tracer.c (find_best_successor): Check probability for
10344 being initialized, bail out if not.
10346 2018-06-04 Richard Earnshaw <rearnsha@arm.com>
10349 * config/arm/arm-cpus.in (ALL_QUIRKS): Add xscale feature to the list
10350 of bits to ignore when comparing architectures.
10352 2018-06-04 Jakub Jelinek <jakub@redhat.com>
10354 PR tree-optimization/69615
10355 * fold-const.c (merge_ranges): If range1 is - [x, x] and x is the
10356 maximum or minimum of the type, try to merge it also as if
10357 range1 is + [-, x - 1] or + [x + 1, -].
10360 * tree.c (inchash::add_expr): Handle IDENTIFIER_NODE.
10362 2018-06-03 Eric Botcazou <ebotcazou@adacore.com>
10364 PR tree-optimization/86034
10365 * gimple-ssa-store-merging.c (output_merged_store): Convert the RHS to
10366 the unsigned bitfield type in a bit insertion sequence if it does not
10367 have a larger precision than the bitfield size.
10368 (process_store): Also bypass widening conversions for BIT_INSERT_EXPR.
10370 2018-06-03 Kito Cheng <kito.cheng@gmail.com>
10372 * config/nds32/nds32-peephole2.md: Add new patterns for code size.
10374 2018-06-03 Chung-Ju Wu <jasonwucj@gmail.com>
10376 * config/nds32/nds32-opts.h (nds32_arch_type): Add ARCH_V3J.
10377 * config/nds32/nds32.c (nds32_option_override): Consider ARCH_V3J.
10378 * config/nds32/nds32.h (TARGET_ISA_V3): Add ARCH_V3J checking.
10379 * config/nds32/nds32.opt (march): Add enum value Value(ARCH_V3J).
10381 2018-06-03 Chung-Ju Wu <jasonwucj@gmail.com>
10383 * common/config/nds32/nds32-common.c (nds32_option_optimization_table):
10384 Disable -fdelete-null-pointer-checks for ELF toolchain.
10386 2018-06-02 Chung-Ju Wu <jasonwucj@gmail.com>
10387 Kito Cheng <kito.cheng@gmail.com>
10389 * config.gcc (nds32*): Use nds32-linux.opt and nds32-elf.opt.
10390 (nds32le-*-*, nds32be-*-*): Integrate checking process.
10391 (nds32*-*-*): Add glibc and uclibc conditions.
10392 * common/config/nds32/nds32-common.c (nds32_except_unwind_info): New.
10393 (TARGET_EXCEPT_UNWIND_INFO): Define.
10394 * config/nds32/elf.h: New file.
10395 * config/nds32/linux.h: New file.
10396 * config/nds32/nds32-elf.opt: New file.
10397 * config/nds32/nds32-linux.opt: New file.
10398 * config/nds32/nds32-fp-as-gp.c
10399 (pass_nds32_fp_as_gp::gate): Consider TARGET_LINUX_ABI.
10400 * config/nds32/nds32.c (nds32_conditional_register_usage): Consider
10402 (nds32_asm_file_end): Ditto.
10403 (nds32_print_operand): Ditto.
10404 (nds32_insert_attributes): Ditto.
10405 (nds32_init_libfuncs): New function.
10406 (TARGET_HAVE_TLS): Define.
10407 (TARGET_INIT_LIBFUNCS): Define.
10408 * config/nds32/nds32.h (TARGET_DEFAULT_RELAX): Apply different relax
10410 (TARGET_ELF): Apply different mcmodel setting.
10411 (LINK_SPEC, LIB_SPEC, STARTFILE_SPEC, ENDFILE_SPEC): The content has
10412 been migrated into elf.h and linux.h files.
10413 * config/nds32/nds32.md (add_pc): Consider TARGET_LINUX_ABI.
10414 * config/nds32/nds32.opt (mvh): Consider TARGET_LINUX_ABI.
10415 (mcmodel): The content has been migrated into nds32-elf.opt and
10416 nds32-linux.opt files.
10417 * config/nds32/t-elf: New file.
10418 * config/nds32/t-linux: New file.
10420 2018-06-02 Chung-Ju Wu <jasonwucj@gmail.com>
10421 Shiva Chen <shiva0217@gmail.com>
10423 * config/nds32/constants.md (unspec_volatile_element): Add
10424 UNSPEC_VOLATILE_OMIT_FP_BEGIN and UNSPEC_VOLATILE_OMIT_FP_END.
10425 * config/nds32/nds32-fp-as-gp.c: New implementation of fp_as_gp
10427 * config/nds32/nds32-protos.h (nds32_naked_function_p): Declare.
10428 (make_pass_nds32_fp_as_gp): Declare.
10429 * config/nds32/nds32.c (nds32_register_passes): Add fp_as_gp as one
10431 (nds32_asm_function_end_prologue): Remove unused asm output.
10432 (nds32_asm_function_begin_epilogue): Remove unused asm output.
10433 (nds32_asm_file_start): Output necessary fp_as_gp information.
10434 (nds32_option_override): Adjust register usage.
10435 (nds32_expand_prologue): Consider fp_as_gp situation.
10436 (nds32_expand_prologue_v3push): Consider fp_as_gp situation.
10437 * config/nds32/nds32.md (prologue): Check fp_as_gp_p and naked_p.
10440 (simple_return): Ditto.
10441 (omit_fp_begin): Output special directive for fp_as_gp.
10442 (omit_fp_end): Output special directive for fp_as_gp.
10443 * config/nds32/nds32.opt (mfp-as-gp, mno-fp-as-gp, mforce-fp-as-gp,
10444 mforbid-fp-as-gp): New options.
10446 2018-06-01 Mark Wielaard <mark@klomp.org>
10448 * dwarf2out.c (dwarf2out_finish): Remove generation of
10449 DW_AT_loclists_base.
10451 2018-06-01 Eric Botcazou <ebotcazou@adacore.com>
10453 * gimple-ssa-store-merging.c: Include gimple-fold.h.
10454 (struct store_immediate_info): Document BIT_INSERT_EXPR stores.
10455 (struct merged_store_group): Add bit_insertion field.
10456 (dump_char_array): Use standard hexadecimal format.
10457 (merged_store_group::merged_store_group): Set bit_insertion to false.
10458 (merged_store_group::apply_stores): Use optimal buffer size. Deal
10459 with BIT_INSERT_EXPR stores. Move up code updating the mask and
10460 also print the mask in the dump file.
10461 (pass_store_merging::gate): Minor tweak.
10462 (imm_store_chain_info::coalesce_immediate): Fix wrong association
10463 of stores with groups in dump. Allow coalescing of BIT_INSERT_EXPR
10464 stores with INTEGER_CST stores.
10465 (count_multiple_uses) <BIT_INSERT_EXPR>: New case.
10466 (imm_store_chain_info::output_merged_store): Add try_bitpos variable
10467 and use it throughout. Generate bit insertion sequences if need be.
10468 (pass_store_merging::process_store): Remove redundant condition.
10469 Record stores from a SSA name to a bit-field with BIT_INSERT_EXPR.
10471 2018-06-01 Segher Boessenkool <segher@kernel.crashing.org>
10473 * config/rs6000/rs6000.c (rs6000_mangle_type): Change the mangling of
10474 the 128-bit floating point types. Fix function comment.
10476 2018-06-01 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10478 * config/aarch64/aarch64-simd.md
10479 (aarch64_simd_vec_unpack<su>_lo_<mode>): Use UXTL and SXTL assembler
10481 (aarch64_simd_vec_unpack<su>_hi_<mode>): Use UXTL2 and SXTL2 assembler
10484 2018-06-01 Richard Sandiford <richard.sandiford@linaro.org>
10486 PR tree-optimization/85989
10487 * gimple-ssa-backprop.c (backprop::m_visited_phis): New member
10489 (backprop::intersect_uses): Check it when deciding whether this
10490 is a backedge reference.
10491 (backprop::process_block): Add each phi to m_visited_phis
10492 after visiting it, then clear it at the end.
10494 2018-06-01 Richard Biener <rguenther@suse.de>
10496 * tree-vectorizer.h (vect_dr_stmt): New function.
10497 (vect_get_load_cost): Adjust.
10498 (vect_get_store_cost): Likewise.
10499 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence):
10500 Use vect_dr_stmt instead of DR_SMTT.
10501 (vect_record_base_alignments): Likewise.
10502 (vect_calculate_target_alignment): Likewise.
10503 (vect_compute_data_ref_alignment): Likewise and make static.
10504 (vect_update_misalignment_for_peel): Likewise.
10505 (vect_verify_datarefs_alignment): Likewise.
10506 (vector_alignment_reachable_p): Likewise.
10507 (vect_get_data_access_cost): Likewise. Pass down
10508 vinfo to vect_get_load_cost/vect_get_store_cost instead of DR.
10509 (vect_get_peeling_costs_all_drs): Likewise.
10510 (vect_peeling_hash_get_lowest_cost): Likewise.
10511 (vect_enhance_data_refs_alignment): Likewise.
10512 (vect_find_same_alignment_drs): Likewise.
10513 (vect_analyze_data_refs_alignment): Likewise.
10514 (vect_analyze_group_access_1): Likewise.
10515 (vect_analyze_group_access): Likewise.
10516 (vect_analyze_data_ref_access): Likewise.
10517 (vect_analyze_data_ref_accesses): Likewise.
10518 (vect_vfa_segment_size): Likewise.
10519 (vect_small_gap_p): Likewise.
10520 (vectorizable_with_step_bound_p): Likewise.
10521 (vect_prune_runtime_alias_test_list): Likewise.
10522 (vect_analyze_data_refs): Likewise.
10523 (vect_supportable_dr_alignment): Likewise.
10524 * tree-vect-loop-manip.c (get_misalign_in_elems): Likewise.
10525 (vect_gen_prolog_loop_niters): Likewise.
10526 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
10527 * tree-vect-patterns.c (vect_recog_bool_pattern): Do not
10529 (vect_recog_mask_conversion_pattern): Likewise.
10530 (vect_try_gather_scatter_pattern): Likewise.
10531 * tree-vect-stmts.c (vect_model_store_cost): Pass stmt_info
10532 to vect_get_store_cost.
10533 (vect_get_store_cost): Get stmt_info instead of DR.
10534 (vect_model_load_cost): Pass stmt_info to vect_get_load_cost.
10535 (vect_get_load_cost): Get stmt_info instead of DR.
10537 2018-06-01 Richard Biener <rguenther@suse.de>
10539 PR middle-end/86017
10540 * gimple-fold.c (var_decl_component_p): Also allow offsetted
10541 vars wrapped in MEM_REFs.
10543 2018-06-01 Richard Sandiford <richard.sandiford@linaro.org>
10545 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
10546 Fix subreg tests so that we only return a choice between
10547 GENERAL_REGS and FP_REGS if the original classes included both.
10549 2018-06-01 Richard Biener <rguenther@suse.de>
10552 * tree-ssa-structalias.c (get_function_part_constraint):
10553 Handle NULL fi->decl.
10554 (find_func_aliases_for_call): Properly handle indirect
10555 fi from direct call.
10556 (find_func_clobbers): Likewise.
10557 (ipa_pta_execute): Likewise.
10558 (create_variable_info_for): For functions that are ifunc_resolver
10559 resolve to a varinfo that contains the result of the resolver call.
10560 (associate_varinfo_to_alias): Do not treat ifunc resolvers as
10563 2018-05-31 Michael Collison <michael.collison@arm.com>
10565 * config/aarch64/aarch64.md:
10566 (*fix_to_zero_extenddfdi2): New pattern.
10567 * gcc.target/aarch64/fix_extend1.c: New testcase.
10569 2018-05-31 Qing Zhao <qing.zhao@oracle.com>
10571 PR middle-end/78809
10572 PR middle-end/83026
10573 * builtins.c (expand_builtin): Add the handling of BUILT_IN_STRCMP_EQ
10574 and BUILT_IN_STRNCMP_EQ.
10575 * builtins.def: Add new builtins BUILT_IN_STRCMP_EQ and
10576 BUILT_IN_STRNCMP_EQ.
10577 * gimple-fold.c (gimple_fold_builtin_string_compare): Add the
10578 handling of BUILTIN_IN_STRCMP_EQ and BUILT_IN_STRNCMP_EQ.
10579 (gimple_fold_builtin): Likewise.
10580 * tree-ssa-strlen.c (compute_string_length): New function.
10581 (determine_min_obsize): New function.
10582 (handle_builtin_string_cmp): New function to handle calls to
10583 string compare functions.
10584 (strlen_optimize_stmt): Add handling to builtin string compare
10586 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
10587 Add the handling of BUILT_IN_STRCMP_EQ and BUILT_IN_STRNCMP_EQ.
10588 * tree.c (build_common_builtin_nodes): Add new defines of
10589 BUILT_IN_STRNCMP_EQ and BUILT_IN_STRCMP_EQ.
10591 2018-05-31 Jakub Jelinek <jakub@redhat.com>
10594 * bb-reorder.c (pass_partition_blocks::gate): Return false for
10595 functions with naked attribute.
10597 2018-05-31 Uros Bizjak <ubizjak@gmail.com>
10599 * config/i386/sse.md (avx_vec_concat<mode>):
10600 Substitute concat_tg_mode mode attribute with xtg_mode.
10601 (<mask_codefor>avx512dq_broadcast<mode><mask_name>_1): Ditto.
10602 (concat_tg_mode): Remove mode attribute.
10604 2018-05-31 Martin Sebor <msebor@redhat.com>
10607 * calls.c (alloc_max_size): Correct a logic error/typo.
10608 Treat excessive arguments as infinite. Warn for invalid arguments.
10609 * doc/invoke.texi (-Walloc-size-larger-than): Update.
10611 2018-05-31 H.J. Lu <hongjiu.lu@intel.com>
10614 * config/i386/x86-tune.def: Re-enable partial_reg_dependency
10615 and movx for Haswell.
10617 2018-05-31 Chung-Lin Tang <cltang@codesourcery.com>
10618 Cesar Philippidis <cesar@codesourcery.com>
10620 PR middle-end/85879
10621 * gimplify.c (gimplify_adjust_omp_clauses): Add 'remove = true'
10622 when emitting error on private/firstprivate reductions.
10623 * omp-low.c (lower_omp_target): Avoid reference-type processing
10624 on pointers for firstprivate clause.
10626 2018-05-31 Sameera Deshpande <sameera.deshpande@linaro.org>
10628 * config/aarch64/aarch64-simd-builtins.def (ld1x3): New.
10631 * config/aarch64/aarch64-simd.md
10632 (aarch64_ld1x3<VALLDIF:mode>): New pattern.
10633 (aarch64_ld1_x3_<mode>): Likewise
10634 (aarch64_st1x2<VALLDIF:mode>): Likewise
10635 (aarch64_st1_x2_<mode>): Likewise
10636 (aarch64_st1x3<VALLDIF:mode>): Likewise
10637 (aarch64_st1_x3_<mode>): Likewise
10638 * config/aarch64/arm_neon.h (vld1_u8_x3): New function.
10639 (vld1_s8_x3): Likewise.
10640 (vld1_u16_x3): Likewise.
10641 (vld1_s16_x3): Likewise.
10642 (vld1_u32_x3): Likewise.
10643 (vld1_s32_x3): Likewise.
10644 (vld1_u64_x3): Likewise.
10645 (vld1_s64_x3): Likewise.
10646 (vld1_f16_x3): Likewise.
10647 (vld1_f32_x3): Likewise.
10648 (vld1_f64_x3): Likewise.
10649 (vld1_p8_x3): Likewise.
10650 (vld1_p16_x3): Likewise.
10651 (vld1_p64_x3): Likewise.
10652 (vld1q_u8_x3): Likewise.
10653 (vld1q_s8_x3): Likewise.
10654 (vld1q_u16_x3): Likewise.
10655 (vld1q_s16_x3): Likewise.
10656 (vld1q_u32_x3): Likewise.
10657 (vld1q_s32_x3): Likewise.
10658 (vld1q_u64_x3): Likewise.
10659 (vld1q_s64_x3): Likewise.
10660 (vld1q_f16_x3): Likewise.
10661 (vld1q_f32_x3): Likewise.
10662 (vld1q_f64_x3): Likewise.
10663 (vld1q_p8_x3): Likewise.
10664 (vld1q_p16_x3): Likewise.
10665 (vld1q_p64_x3): Likewise.
10666 (vst1_s64_x2): Likewise.
10667 (vst1_u64_x2): Likewise.
10668 (vst1_f64_x2): Likewise.
10669 (vst1_s8_x2): Likewise.
10670 (vst1_p8_x2): Likewise.
10671 (vst1_s16_x2): Likewise.
10672 (vst1_p16_x2): Likewise.
10673 (vst1_s32_x2): Likewise.
10674 (vst1_u8_x2): Likewise.
10675 (vst1_u16_x2): Likewise.
10676 (vst1_u32_x2): Likewise.
10677 (vst1_f16_x2): Likewise.
10678 (vst1_f32_x2): Likewise.
10679 (vst1_p64_x2): Likewise.
10680 (vst1q_s8_x2): Likewise.
10681 (vst1q_p8_x2): Likewise.
10682 (vst1q_s16_x2): Likewise.
10683 (vst1q_p16_x2): Likewise.
10684 (vst1q_s32_x2): Likewise.
10685 (vst1q_s64_x2): Likewise.
10686 (vst1q_u8_x2): Likewise.
10687 (vst1q_u16_x2): Likewise.
10688 (vst1q_u32_x2): Likewise.
10689 (vst1q_u64_x2): Likewise.
10690 (vst1q_f16_x2): Likewise.
10691 (vst1q_f32_x2): Likewise.
10692 (vst1q_f64_x2): Likewise.
10693 (vst1q_p64_x2): Likewise.
10694 (vst1_s64_x3): Likewise.
10695 (vst1_u64_x3): Likewise.
10696 (vst1_f64_x3): Likewise.
10697 (vst1_s8_x3): Likewise.
10698 (vst1_p8_x3): Likewise.
10699 (vst1_s16_x3): Likewise.
10700 (vst1_p16_x3): Likewise.
10701 (vst1_s32_x3): Likewise.
10702 (vst1_u8_x3): Likewise.
10703 (vst1_u16_x3): Likewise.
10704 (vst1_u32_x3): Likewise.
10705 (vst1_f16_x3): Likewise.
10706 (vst1_f32_x3): Likewise.
10707 (vst1_p64_x3): Likewise.
10708 (vst1q_s8_x3): Likewise.
10709 (vst1q_p8_x3): Likewise.
10710 (vst1q_s16_x3): Likewise.
10711 (vst1q_p16_x3): Likewise.
10712 (vst1q_s32_x3): Likewise.
10713 (vst1q_s64_x3): Likewise.
10714 (vst1q_u8_x3): Likewise.
10715 (vst1q_u16_x3): Likewise.
10716 (vst1q_u32_x3): Likewise.
10717 (vst1q_u64_x3): Likewise.
10718 (vst1q_f16_x3): Likewise.
10719 (vst1q_f32_x3): Likewise.
10720 (vst1q_f64_x3): Likewise.
10721 (vst1q_p64_x3): Likewise.
10723 2018-05-30 Jozef Lawrynowicz <jozef.l@mittosystems.com>
10725 * config/msp430/msp430.c (msp430_output_labelref): Prepend
10726 user_label_prefix to name.
10728 * tree-core.h: Update comment about the format of NAME string
10729 passed to handler in attribute_spec.
10731 * config/msp430/msp430.md: Remove erroneous subreg expression from
10732 zero_extendqisi2 insn pattern. Remove msp430x ISA restriction on
10733 zero_extend{q,h}isi2.
10735 2018-05-30 Borislav Petkov <bp@suse.de>
10737 * doc/extend.texi: Document some architecture specific
10738 constraints and sort entries.
10740 2018-05-30 Martin Sebor <msebor@redhat.com>
10742 PR middle-end/85369
10743 * builtins.c (expand_builtin_stpcpy_1): New function.
10744 (expand_builtin_stpcpy): Call it, and call maybe_warn_nonstring_arg
10745 only if the former succeeds.
10747 2018-05-31 Sameera Deshpande <sameera.deshpande@linaro.org>
10749 * config/aarch64/aarch64-cores.def (saphira) : Add support for ARMv8.4
10752 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
10754 * doc/invoke.texi (-flinker-output): Document
10756 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
10758 * passes.c (ipa_write_summaries): Only modify statements if body
10760 * cgraphunit.c (ipa_passes): Also produce intermeidate code when
10761 incrementally linking.
10762 (ipa_passes): Likewise.
10763 * lto-cgraph.c (lto_output_node): When incrementally linking do not
10764 pass down resolution info.
10765 * common.opt (flag_incremental_link): Update info.
10766 * gcc.c (plugin specs): Turn flinker-output=* to
10767 -plugin-opt=-linker-output-known
10768 * toplev.c (compile_file): Also cut compilation when doing incremental
10770 * flag-types. (enum lto_partition_model): Add
10771 LTO_LINKER_OUTPUT_NOLTOREL.
10772 (invoke.texi): Add -flinker-output docs.
10773 * ipa.c (symbol_table::remove_unreachable_nodes): Handle LTO incremental
10774 link same way as WPA; do not stream in dead initializers.
10776 * dwarf2out.c (dwarf2out_die_ref_for_decl,
10777 darf2out_register_external_decl): Support incremental link.
10779 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
10781 * lto-opts.c (lto_write_options): Skip OPT_dumpdir, OPT_fresolution_.
10783 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
10785 * lto-wrapper.c (debug_objcopy): Add rename parameter; pass
10786 it down to simple_object_copy_lto_debug_sections.
10787 (run_gcc): Determine incremental LTO link time and configure
10788 lto1 into non-wpa mode, disable renaming of debug sections.
10790 2018-05-30 Kelvin Nilsen <kelvin@gcc.gnu.org>
10792 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Remove
10793 descriptions of various incorrectly documented functions.
10795 2018-05-30 Andre Vieira <andre.simoesdiasvieira@arm.com>
10798 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Make
10799 address check not strict.
10801 2018-05-30 Richard Biener <rguenther@suse.de>
10803 PR tree-optimization/85964
10804 * tracer.c (better_p): Drop initialized count check, we only
10805 call the function with initialized counts now.
10806 (find_best_successor): Do find a best edge if one
10807 has uninitialized count.
10808 (find_best_predecessor): Likewise. Do BB frequency check only
10809 if count is initialized.
10811 2017-05-30 Jackson Woodruff <jackson.woodruff@arm.com>
10813 * config/aarch64/aarch64.c (aarch64_host_wide_int_compare): New.
10814 (aarch64_ldrstr_offset_compare): New.
10815 (aarch64_operands_adjust_ok_for_ldpstp): Update to consider all
10816 load/store orderings.
10817 (aarch64_gen_adjusted_ldpstp): Likewise.
10819 2018-05-30 Wilco Dijkstra <wdijkstr@arm.com>
10821 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
10822 Check for subset of GENERAL_REGS and FP_REGS.
10823 * config/aarch64/aarch64-simd.md (aarch64_get_lane): Increase cost of
10826 2018-05-30 Richard Sandiford <richard.sandiford@linaro.org>
10828 * alias.c (adjust_offset_for_component_ref): Use poly_int_tree_p
10829 and wi::to_poly_offset. Add the current offset and then check
10830 whether the sum fits, rather than using an unchecked addition of
10831 a checked term. Check for a shwi rather than a uhwi.
10832 * expr.c (get_bit_range): Use tree_to_poly_uint64.
10833 (store_constructor): Use poly_int_tree_p.
10834 (expand_expr_real_1): Likewise.
10835 * function.c (assign_temp): Likewise.
10836 * fold-const.c (const_binop): Use poly_int_tree_p and
10837 wi::to_poly_offset.
10838 (fold_indirect_ref_1): Likewise. Use multiple_p to attempt an exact
10840 * ipa-icf-gimple.c (func_checker::compare_operand): Use
10841 to_poly_offset for MEM offsets.
10842 * ipa-icf.c (sem_variable::equals): Likewise.
10843 * stor-layout.c (compute_record_mode): Use poly_int_tree_p.
10844 * tree-ssa-sccvn.c (ao_ref_init_from_vn_reference): Use
10845 wi::to_poly_offset for BIT_FIELD_REF offsets.
10846 (vn_reference_maybe_forwprop_address): Use poly_int_tree_p and
10847 wi::to_poly_offset.
10848 * var-tracking.c (emit_note_insn_var_location): Use
10849 tree_to_poly_uint64.
10851 2018-05-29 Jim Wilson <jimw@sifive.com>
10853 * config/riscv/riscv.c (riscv_interrupt_type): Fix comment typo.
10855 2018-05-29 Uros Bizjak <ubizjak@gmail.com>
10858 * config/i386/i386.md (l<rounding_insn><MODEF:mode><SWI48:mode>2):
10859 Enable for TARGET_SSE4_1 and generate rounds{s,d} and cvtts{s,d}2si{,q}
10861 (sse4_1_round<mode>2): Use nonimmediate_operand
10862 for operand 1 predicate.
10864 2018-05-29 Martin Sebor <msebor@redhat.com>
10865 Richard Biener <rguenther@suse.de>
10868 * calls.c (get_size_range): Call determine_value_range instead
10869 of get_value_range..
10870 * tree-vrp.h (determine_value_range): Declared new function.
10871 * tree-vrp.c (determine_value_range_1, determine_value_range): New.
10873 2018-05-29 Richard Biener <rguenther@suse.de>
10875 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Make
10876 sure to use non-pattern stmts for get_earlier_stmt arguments.
10877 * tree-vectorizer.h (get_earlier_stmt): Assert we do not get
10878 called on pattern stmts.
10879 (get_later_stmt): Likewise.
10881 2018-05-29 Martin Liska <mliska@suse.cz>
10883 PR gcov-profile/85759
10884 * doc/gcov.texi: Document GCOV_ERROR_FILE and GCOV_EXIT_AT_ERROR
10887 2018-05-29 Jakub Jelinek <jakub@redhat.com>
10889 * tree-cfg.c (verify_gimple_assign_unary): Add checking for
10891 (verify_gimple_assign_binary): Check TYPE_VECTOR_SUBPARTS for
10895 * tree.def (VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR,
10896 VEC_PACK_FLOAT_EXPR): New tree codes.
10897 * tree-pretty-print.c (op_code_prio): Handle
10898 VEC_UNPACK_FIX_TRUNC_HI_EXPR and VEC_UNPACK_FIX_TRUNC_LO_EXPR.
10899 (dump_generic_node): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR,
10900 VEC_UNPACK_FIX_TRUNC_LO_EXPR and VEC_PACK_FLOAT_EXPR.
10901 * tree-inline.c (estimate_operator_cost): Likewise.
10902 * gimple-pretty-print.c (dump_binary_rhs): Handle VEC_PACK_FLOAT_EXPR.
10903 * fold-const.c (const_binop): Likewise.
10904 (const_unop): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR and
10905 VEC_UNPACK_FIX_TRUNC_LO_EXPR.
10906 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
10907 (verify_gimple_assign_binary): Handle VEC_PACK_FLOAT_EXPR.
10908 * cfgexpand.c (expand_debug_expr): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR,
10909 VEC_UNPACK_FIX_TRUNC_LO_EXPR and VEC_PACK_FLOAT_EXPR.
10910 * expr.c (expand_expr_real_2): Likewise.
10911 * optabs.def (vec_packs_float_optab, vec_packu_float_optab,
10912 vec_unpack_sfix_trunc_hi_optab, vec_unpack_sfix_trunc_lo_optab,
10913 vec_unpack_ufix_trunc_hi_optab, vec_unpack_ufix_trunc_lo_optab): New
10915 * optabs.c (expand_widen_pattern_expr): For
10916 VEC_UNPACK_FIX_TRUNC_HI_EXPR and VEC_UNPACK_FIX_TRUNC_LO_EXPR use
10917 sign from result type rather than operand's type.
10918 (expand_binop_directly): For vec_packu_float_optab and
10919 vec_packs_float_optab allow result type to be different from operand's
10921 * optabs-tree.c (optab_for_tree_code): Handle
10922 VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR and
10923 VEC_PACK_FLOAT_EXPR. Formatting fixes.
10924 * tree-vect-generic.c (expand_vector_operations_1): Handle
10925 VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR and
10926 VEC_PACK_FLOAT_EXPR.
10927 * tree-vect-stmts.c (supportable_widening_operation): Handle
10929 (supportable_narrowing_operation): Handle FLOAT_EXPR.
10930 * config/i386/i386.md (fixprefix, floatprefix): New code attributes.
10931 * config/i386/sse.md (*float<floatunssuffix>v2div2sf2): Rename to ...
10932 (float<floatunssuffix>v2div2sf2): ... this. Formatting fix.
10933 (vpckfloat_concat_mode, vpckfloat_temp_mode, vpckfloat_op_mode): New
10935 (vec_pack<floatprefix>_float_<mode>): New expander.
10936 (vunpckfixt_mode, vunpckfixt_model, vunpckfixt_extract_mode): New mode
10938 (vec_unpack_<fixprefix>fix_trunc_lo_<mode>,
10939 vec_unpack_<fixprefix>fix_trunc_hi_<mode>): New expanders.
10940 * doc/md.texi (vec_packs_float_@var{m}, vec_packu_float_@var{m},
10941 vec_unpack_sfix_trunc_hi_@var{m}, vec_unpack_sfix_trunc_lo_@var{m},
10942 vec_unpack_ufix_trunc_hi_@var{m}, vec_unpack_ufix_trunc_lo_@var{m}):
10944 * doc/generic.texi (VEC_UNPACK_FLOAT_HI_EXPR,
10945 VEC_UNPACK_FLOAT_LO_EXPR): Fix pasto in description.
10946 (VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR,
10947 VEC_PACK_FLOAT_EXPR): Document.
10949 2018-05-29 Richard Biener <rguenther@suse.de>
10951 * tree-vectorizer.h (struct vec_info): Add stmt_vec_infos
10953 (stmt_vec_info_vec): Make pointer.
10954 (init_stmt_vec_info_vec): Remove.
10955 (free_stmt_vec_info_vec): Likewise.
10956 (set_stmt_vec_info_vec): New function.
10957 (free_stmt_vec_infos): Likewise.
10958 (vinfo_for_stmt): Adjust for stmt_vec_info_vec indirection.
10959 (set_vinfo_for_stmt): Likewise.
10960 (get_earlier_stmt): Likewise.
10961 (get_later_stmt): Likewise.
10962 * tree-vectorizer.c (stmt_vec_info_vec): Make pointer.
10963 (vec_info::vec_info): Allocate stmt_vec_infos and set the global.
10964 (vec_info::~vec_info): Free stmt_vec_infos.
10965 (vectorize_loops): Set the global stmt_vec_info_vec to NULL.
10966 Remove old init_stmt_vec_info_vec/free_stmt_vec_info_vec calls.
10967 (pass_slp_vectorize::execute): Likewise.
10968 * tree-vect-stmts.c (init_stmt_vec_info_vec): Remove.
10969 (free_stmt_vec_info_vec): Likewise.
10970 (set_stmt_vec_info_vec): New function.
10971 (free_stmt_vec_infos): Likewise.
10972 * tree-vect-loop.c (_loop_vec_info::~_loop_vec_info): Set
10973 the global stmt_vec_info_vec.
10974 * tree-parloops.c (gather_scalar_reductions): Use
10975 set_stmt_vec_info_vec/free_stmt_vec_infos and maintain a local
10978 2018-05-29 Richard Biener <rguenther@suse.de>
10980 * dominance.c (iterate_fix_dominators): Push/pop TV_DOMINANCE.
10982 2018-05-29 Martin Liska <mliska@suse.cz>
10983 David Malcolm <dmalcolm@redhat.com>
10985 * vec.c (test_reverse): New.
10986 (vec_c_tests): Add new test.
10987 * vec.h (vl_ptr>::reverse): New function.
10989 2018-05-29 Gerald Pfeifer <gerald@pfeifer.com>
10991 * config.gcc: Identify FreeBSD 3.x and 4.x as unsupported.
10993 * config/freebsd-spec.h (FBSD_LIB_SPEC): Only consider FreeBSD 5
10996 2018-05-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
10998 * tree-dump.c (dump_node): Use splay_tree_delete_pointers.
11000 2018-05-28 Richard Biener <rguenther@suse.de>
11002 PR tree-optimization/85933
11003 * tree-vect-data-refs.c (vect_record_base_alignments): Only
11004 look at stmts marked as vectorizable.
11006 2018-05-28 Richard Biener <rguenther@suse.de>
11008 PR tree-optimization/85934
11009 * tree-vect-generic.c (expand_vector_operations_1): Hoist
11010 vector boolean check before scalar optimization.
11012 2018-05-28 Jakub Jelinek <jakub@redhat.com>
11014 * doc/invoke.texi (ARM Options): Use @item instead of @itemx
11017 2018-05-28 Mark Wielaard <mark@klomp.org>
11019 * dwarf2asm.c (dw2_asm_output_delta_uleb128): Add brackets around lab2
11020 if it is an expression containing a minus sign.
11022 2018-05-27 John David Anglin <danglin@gcc.gnu.org>
11024 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Define to 0.
11026 2018-05-27 Paul Koning <ni1d@arrl.net>
11028 * config/pdp11/pdp11.md (truncsihi2): Remove.
11030 2018-05-27 Monk Chiang <sh.chiang04@gmail.com>
11031 Chung-Ju Wu <jasonwucj@gmail.com>
11033 * config/nds32/nds32-intrinsic.md (unaligned_storedi): Modify patterns
11035 (unaligned_store_dw): Ditto.
11036 * config/nds32/nds32-memory-manipulation.c
11037 (nds32_expand_movmemsi_loop_known_size): Refactoring implementation.
11038 (nds32_gen_dup_4_byte_to_word_value): Rename to ...
11039 (nds32_gen_dup_4_byte_to_word_value_aux): ... this.
11040 (emit_setmem_word_loop): Rename to ...
11041 (emit_setmem_doubleword_loop): ... this.
11042 (nds32_gen_dup_4_byte_to_word_value): New function.
11043 (nds32_gen_dup_8_byte_to_double_word_value): New function.
11044 (nds32_expand_setmem_loop): Refine implementation.
11045 (nds32_expand_setmem_loop_v3m): Ditto.
11046 * config/nds32/nds32-multiple.md (unaligned_store_update_base_dw): New
11049 2018-05-27 Chung-Ju Wu <jasonwucj@gmail.com>
11051 * config/nds32/nds32.md (bswapsi2, bswaphi2): New patterns.
11053 2018-05-27 Chung-Ju Wu <jasonwucj@gmail.com>
11055 * config/nds32/nds32.c (nds32_attribute_table): Add "no_prologue".
11056 (nds32_init_machine_status): Initialize machine->attr_naked_p and
11057 machine->attr_no_prologue_p.
11058 (nds32_compute_stack_frame): Check "naked" and "no_prologue" attributes.
11059 (nds32_naked_function_p): Handle "naked" and "no_prologue" attributes.
11060 (nds32_expand_epilogue): Consider attr_naked_p.
11061 (nds32_expand_epilogue_v3pop): Likewise.
11062 (nds32_can_use_return_insn): Likewise.
11063 * config/nds32/nds32.h (machine_function): Add attr_naked_p and
11064 attr_no_prologue_p fields.
11065 * config/nds32/nds32.opt (mret-in-naked-func): New option.
11067 2018-05-27 Jakub Jelinek <jakub@redhat.com>
11070 * config/i386/i386.md (fixunssuffix, floatunssuffix): New code
11072 * config/i386/sse.md
11073 (<floatsuffix>float<sseintvecmodelower><mode>2<mask_name><round_name>):
11075 (float<floatunssuffix><sseintvecmodelower><mode>2<mask_name><round_name>):
11077 (<floatsuffix>float<sselongvecmodelower><mode>2<mask_name><round_name>):
11079 (float<floatunssuffix><sselongvecmodelower><mode>2<mask_name><round_name>):
11081 (*<floatsuffix>floatv2div2sf2): Rename to ...
11082 (*float<floatunssuffix>v2div2sf2): ... this.
11083 (<floatsuffix>floatv2div2sf2_mask): Rename to ...
11084 (float<floatunssuffix>v2div2sf2_mask): ... this.
11085 (*<floatsuffix>floatv2div2sf2_mask_1): Rename to ...
11086 (*float<floatunssuffix>v2div2sf2_mask_1): ... this.
11087 (<fixsuffix>fix_truncv8dfv8si2<mask_name><round_saeonly_name>): Rename
11089 (fix<fixunssuffix>_truncv8dfv8si2<mask_name><round_saeonly_name>):
11091 (<fixsuffix>fix_trunc<mode><sseintvecmodelower>2<mask_name><round_saeonly_name>):
11093 (fix<fixunssuffix>_trunc<mode><sseintvecmodelower>2<mask_name><round_saeonly_name>):
11095 (<fixsuffix>fix_trunc<mode><sselongvecmodelower>2<mask_name><round_saeonly_name>):
11097 (fix<fixunssuffix>_trunc<mode><sselongvecmodelower>2<mask_name><round_saeonly_name>):
11099 (<fixsuffix>fix_truncv2sfv2di2<mask_name>): Rename to ...
11100 (fix<fixunssuffix>_truncv2sfv2di2<mask_name>): ... this.
11101 (vec_pack_ufix_trunc_<mode>): Use gen_fixuns_truncv8dfv8si2 instead of
11102 gen_ufix_truncv8dfv8si2.
11103 * config/i386/i386-builtin.def (__builtin_ia32_cvttpd2uqq256_mask,
11104 __builtin_ia32_cvttpd2uqq128_mask, __builtin_ia32_cvttps2uqq256_mask,
11105 __builtin_ia32_cvttps2uqq128_mask, __builtin_ia32_cvtuqq2ps256_mask,
11106 __builtin_ia32_cvtuqq2ps128_mask, __builtin_ia32_cvtuqq2pd256_mask,
11107 __builtin_ia32_cvtuqq2pd128_mask, __builtin_ia32_cvttpd2udq512_mask,
11108 __builtin_ia32_cvtuqq2ps512_mask, __builtin_ia32_cvtuqq2pd512_mask,
11109 __builtin_ia32_cvttps2uqq512_mask, __builtin_ia32_cvttpd2uqq512_mask):
11110 Use fixuns instead ufix or floatuns instead ufloat in CODE_FOR_ names.
11112 2018-05-24 H.J. Lu <hongjiu.lu@intel.com>
11116 * varasm.c (assemble_alias): Lookup ifunc attribute on error.
11118 2018-05-25 Jim Wilson <jimw@sifive.com>
11120 * config/riscv/riscv-protos.h (riscv_epilogue_uses): New.
11121 * config/riscv/riscv.c (struct machine_function): Add
11122 interrupt_handler_p and attribute_checked_p fields.
11123 (riscv_attribute_table): Add interrupt.
11124 (riscv_interrupt_type_p): New.
11125 (riscv_save_reg_p): Save extra regs for interrupt handler.
11126 (riscv_use_save_libcall): Return false for interrupt handler.
11127 (riscv_first_stack_step): Add forward declaration.
11128 (riscv_compute_frame_info): New local interrupt_save_t1. Set it
11129 for interrupt handler with large frame. Use it for saved reg list.
11130 (riscv_expand_prologue): Move flag_stack_usage_info support to
11131 eliminate duplication.
11132 (riscv_expand_epilogue): Generate mret for interrupt handler.
11133 (riscv_epilogue_uses): New.
11134 (riscv_can_use_return_insn): Return false for interrupt handler.
11135 (riscv_function_ok_for_sibcall): Likewise.
11136 (riscv_set_current_function): Add interrupt handler support.
11137 * config/riscv/riscv.h (EPILOGUE_USES): Call riscv_epilogue_uses.
11138 * config/riscv/riscv.md (UNSPECV_MRET): New.
11140 (riscv_frflags, riscv_fsflags): Use tab after opcode.
11142 * doc/extend.texi (RISC-V Function Attributes) <interrupt>: New.
11144 2018-05-25 Bill Schmidt <wschmidt@linux.ibm.com>
11146 PR tree-optimization/85712
11147 * gimple-ssa-strength-reduction.c (replace_one_candidate): Skip if
11148 this candidate has already been replaced in-situ by a copy.
11150 2018-05-25 Jason Merrill <jason@redhat.com>
11152 PR c++/80485 - inline function non-zero address.
11153 * symtab.c (nonzero_address): Check DECL_COMDAT.
11155 2018-05-25 Uros Bizjak <ubizjak@gmail.com>
11158 * config/alpha/alpha.md (ashlsi3): New insn pattern.
11159 (*ashlsi_se): Rename from *ashldi_se. Define as sign
11160 extension of SImode operation. Use const123_operand predicate.
11161 (*saddsi_1): Remove.
11162 (*saddl_se_1): Ditto.
11163 (*ssubsi_1): Ditto.
11164 (*ssubl_se_1): Ditto.
11165 * config/alpha/predicates.md (const123_operand): New predicate.
11166 * config/alpha/constraints.md (P): Use IN_RANGE.
11168 2018-05-25 Richard Biener <rguenther@suse.de>
11170 * tree-ssa-alias.h (refs_may_alias_p): Add tbaa_p bool parameter,
11172 (ref_maybe_used_by_stmt_p): Likewise.
11173 (stmt_may_clobber_ref_p): Likewise.
11174 (stmt_may_clobber_ref_p_1): Likewise.
11175 * tree-ssa-alias.c (refs_may_alias_p): Add tbaa_p bool parameter
11177 (ref_maybe_used_by_stmt_p): Likewise.
11178 (stmt_may_clobber_ref_p): Likewise.
11179 (stmt_may_clobber_ref_p_1): Likewise.
11180 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences): Use
11181 the alias oracle to disambiguate DRs with stmts DR analysis
11183 (vect_analyze_data_refs): Do not give up on not analyzable
11184 DRs for BB vectorization. Remove code truncating the dataref
11187 2018-05-25 Jakub Jelinek <jakub@redhat.com>
11190 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>_1):
11191 Add (=Yk,v,C) variant using vptestm insn. Use TARGET_AVX512BW
11192 in test instead of TARGET_AVX512F for VI12_AVX512VL iterator.
11194 2018-05-25 Richard Biener <rguenther@suse.de>
11196 * tree-vect-data-refs.c (vect_find_stmt_data_reference): New
11197 function, combining stmt data ref gathering and fatal analysis
11199 (vect_analyze_data_refs): Remove now redudnant code and simplify.
11200 * tree-vect-loop.c (vect_get_datarefs_in_loop): Factor out from
11201 vect_analyze_loop_2 and use vect_find_stmt_data_reference.
11202 * tree-vect-slp.c (vect_slp_bb): Use vect_find_stmt_data_reference.
11203 * tree-vectorizer.h (vect_find_stmt_data_reference): Declare.
11205 2018-05-25 Bin Cheng <bin.cheng@arm.com>
11207 PR tree-optimization/85720
11208 * tree-loop-distribution.c (break_alias_scc_partitions): Don't merge
11209 SCC if all partitions are builtins.
11210 (version_loop_by_alias_check): New parameter. Generate cancelable
11211 runtime alias check if all partitions are builtins.
11212 (distribute_loop): Update call to above function.
11214 2018-05-25 Bin Cheng <bin.cheng@arm.com>
11216 * tree-outof-ssa.c (tree-ssa.h, tree-dfa.h): Include header files.
11217 (create_default_def, for_all_parms): Moved from tree-ssa-coalesce.c.
11218 (parm_default_def_partition_arg): Ditto.
11219 (set_parm_default_def_partition): Ditto.
11220 (get_parm_default_def_partitions): Ditto and make it static.
11221 (get_undefined_value_partitions): Ditto and make it static.
11222 (remove_ssa_form): Refactor call to init_var_map here.
11223 * tree-ssa-coalesce.c (build_ssa_conflict_graph): Support live range
11224 computation for loop region.
11225 (coalesce_partitions, compute_optimized_partition_bases): Ditto.
11226 (register_default_def): Delete.
11227 (for_all_parms, create_default_def): Move to tree-outof-ssa.c.
11228 (parm_default_def_partition_arg): Ditto.
11229 (set_parm_default_def_partition): Ditto.
11230 (get_parm_default_def_partitions): Ditto and make it static.
11231 (get_undefined_value_partitions): Ditto and make it static.
11232 (coalesce_with_default, coalesce_with_default): Update comment.
11233 (create_coalesce_list_for_region): New func factored out from
11234 create_outofssa_var_map.
11235 (populate_coalesce_list_for_outofssa): New func factored out from
11236 create_outofssa_var_map and coalesce_ssa_name.
11237 (create_outofssa_var_map): Delete.
11238 (coalesce_ssa_name): Refactor to support live range computation.
11239 * tree-ssa-coalesce.h (coalesce_ssa_name): Change decl.
11240 (get_parm_default_def_partitions): Delete.
11241 (get_undefined_value_partitions): Ditto.
11242 * tree-ssa-live.c (init_var_map, delete_var_map): Support live range
11243 computation for loop region.
11244 (new_tree_live_info, loe_visit_block): Ditto.
11245 (live_worklist, set_var_live_on_entry): Ditto.
11246 (calculate_live_on_exit, verify_live_on_entry): Ditto.
11247 * tree-ssa-live.h (struct _var_map): New fields.
11248 (init_var_map): Change decl.
11249 (region_contains_p): New.
11251 2018-05-25 Bin Cheng <bin.cheng@arm.com>
11253 * tree-ssa-live.h (live_merge_and_clear): Delete.
11255 2018-05-25 Richard Biener <rguenther@suse.de>
11258 * tree-dump.c (dequeue_and_dump): Remove access to removed
11259 operand 2 of a SWITCH_EXPR.
11261 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
11263 * doc/sourcebuild.texi (vect_double_cond_arith): Include
11264 multiplication and division.
11265 * doc/md.texi (cond_mul@var{m}, cond_div@var{m}, cond_mod@var{m})
11266 (cond_udiv@var{m}, cond_umod@var{m}): Document.
11267 * optabs.def (cond_smul_optab, cond_sdiv_optab, cond_smod_optab)
11268 (cond_udiv_optab, cond_umod_optab): New optabs.
11269 * internal-fn.def (IFN_COND_MUL, IFN_COND_DIV, IFN_COND_MOD)
11270 (IFN_COND_RDIV): New internal functions.
11271 * internal-fn.c (get_conditional_internal_fn): Handle TRUNC_DIV_EXPR,
11272 TRUNC_MOD_EXPR and RDIV_EXPR.
11273 * match.pd (UNCOND_BINARY, COND_BINARY): Handle them.
11274 * config/aarch64/iterators.md (UNSPEC_COND_MUL, UNSPEC_COND_DIV):
11276 (SVE_INT_BINARY): Include mult.
11277 (SVE_COND_FP_BINARY): Include UNSPEC_MUL and UNSPEC_DIV.
11278 (optab, sve_int_op): Handle mult.
11279 (optab, sve_fp_op, commutative): Handle UNSPEC_COND_MUL and
11281 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): New pattern
11282 for SVE_INT_BINARY_SD.
11284 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
11286 * config/aarch64/iterators.md (SVE_INT_BINARY_SD): New code iterator.
11287 (optab, sve_int_op): Handle div and udiv.
11288 * config/aarch64/aarch64-sve.md (<optab><mode>3): New expander
11289 for SVE_INT_BINARY_SD.
11290 (*<optab><mode>3): New insn for the same.
11292 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
11294 * tree-vect-patterns.c: Include predict.h.
11295 (vect_recog_divmod_pattern): Restrict check for division support
11296 to when optimizing for size.
11298 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
11300 * doc/sourcebuild.texi (vect_double_cond_arith: Document.
11301 * gimple-match.h (gimple_match_op::MAX_NUM_OPS): Bump to 4.
11302 (gimple_match_op::gimple_match_op): Add an overload for 4 operands.
11303 (gimple_match_op::set_op): Likewise.
11304 (gimple_resimplify4): Declare.
11305 * genmatch.c (get_operand_type): Handle CFN_COND_* functions.
11306 (expr::gen_transform): Likewise.
11307 (decision_tree::gen): Generate a simplification routine for 4 operands.
11308 * gimple-match-head.c (gimple_simplify): Add an overload for
11309 4 operands. In the top-level function, handle up to 4 call
11310 arguments and call gimple_resimplify4.
11311 (gimple_resimplify4): New function.
11312 (build_call_internal): Pass a fourth operand.
11313 (maybe_push_to_seq): Likewise.
11314 * match.pd (UNCOND_BINARY, COND_BINARY): New operator lists.
11315 Fold VEC_COND_EXPRs of an operation and a default value into
11316 an IFN_COND_* function if possible.
11317 * config/aarch64/iterators.md (UNSPEC_COND_MAX, UNSPEC_COND_MIN):
11319 (SVE_COND_FP_BINARY): Include them.
11320 (optab, sve_fp_op): Handle them.
11321 (SVE_INT_BINARY_REV): New code iterator.
11322 (SVE_COND_FP_BINARY_REV): New int iterator.
11323 (commutative): New int attribute.
11324 * config/aarch64/aarch64-protos.h (aarch64_sve_prepare_conditional_op):
11326 * config/aarch64/aarch64.c (aarch64_sve_prepare_conditional_op): New
11328 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): Use it.
11329 (*cond_<optab><mode>): New patterns for reversed operands.
11331 2018-05-25 Richard Biener <rguenther@suse.de>
11333 * tree-vectorizer.h (STMT_VINFO_GROUP_*, GROUP_*): Remove.
11334 (DR_GROUP_*): New, assert we have non-NULL ->data_ref_info.
11335 (REDUC_GROUP_*): New, assert we have NULL ->data_ref_info.
11336 (STMT_VINFO_GROUPED_ACCESS): Adjust.
11337 * tree-vect-data-refs.c (everywhere): Adjust users.
11338 * tree-vect-loop.c (everywhere): Likewise.
11339 * tree-vect-slp.c (everywhere): Likewise.
11340 * tree-vect-stmts.c (everywhere): Likewise.
11341 * tree-vect-patterns.c (vect_reassociating_reduction_p): Likewise.
11343 2018-05-25 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
11345 * configure.ac (gcc_cv_as_section_has_e): Move to common section.
11347 (gcc_cv_as_section_exclude): ... this.
11348 Try Solaris as #exclude syntax.
11349 * configure: Regenerate.
11350 * config.in: Regenerate.
11351 * config/i386/i386.c (i386_solaris_elf_named_section): Handle
11353 * config/sparc/sparc.c (sparc_solaris_elf_asm_named_section)
11354 [HAVE_GAS_SECTION_EXCLUDE]: Handle SECTION_EXCLUDE.
11356 * varasm.c (default_elf_asm_named_section): Don't check if
11357 HAVE_GAS_SECTION_EXCLUDE is defined.
11359 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
11361 * doc/md.texi: Update the documentation of the cond_* optabs
11362 to mention the new final operand. Fix GET_MODE_NUNITS call.
11363 Describe the scalar case too.
11364 * internal-fn.def (IFN_EXTRACT_LAST): Change type to fold_left.
11365 * internal-fn.c (expand_cond_unary_optab_fn): Expect 3 operands
11367 (expand_cond_binary_optab_fn): Expect 4 operands instead of 3.
11368 (get_conditional_internal_fn): Update comment.
11369 * tree-vect-loop.c (vectorizable_reduction): Pass the original
11370 accumulator value as a final argument to conditional functions.
11371 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): Turn into
11372 a define_expand and add an "else" operand. Assert for now that
11373 the else operand is equal to operand 2. Use SVE_INT_BINARY and
11374 SVE_COND_FP_BINARY instead of SVE_COND_INT_OP and SVE_COND_FP_OP.
11375 (*cond_<optab><mode>): New patterns.
11376 * config/aarch64/iterators.md (UNSPEC_COND_SMAX, UNSPEC_COND_UMAX)
11377 (UNSPEC_COND_SMIN, UNSPEC_COND_UMIN, UNSPEC_COND_AND, UNSPEC_COND_ORR)
11378 (UNSPEC_COND_EOR): Delete.
11379 (optab): Remove associated mappings.
11380 (SVE_INT_BINARY): New code iterator.
11381 (sve_int_op): Remove int attribute and add "minus" to the code
11383 (SVE_COND_INT_OP): Delete.
11384 (SVE_COND_FP_OP): Rename to...
11385 (SVE_COND_FP_BINARY): ...this.
11387 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
11389 * optabs.c (can_reuse_operands_p): New function.
11390 (maybe_legitimize_operands): Try to reuse the results for
11393 2018-05-24 Uros Bizjak <ubizjak@gmail.com>
11395 * config/i386/sse.md (cvtusi2<ssescalarmodesuffix>64<round_name>):
11396 Add {q} suffix to insn mnemonic.
11398 2018-05-23 Jozef Lawrynowicz <jozef.l@mittosystems.com>
11400 * config/msp430/msp430.c (TARGET_WARN_FUNC_RETURN): Define.
11401 (msp430_warn_func_return): New.
11403 2018-05-24 Roger Sayle <roger@nextmovesoftware.com>
11405 * fold-const.c (tree_nonzero_bits): New function.
11406 * fold-const.h (tree_nonzero_bits): Likewise.
11407 * match.pd (POPCOUNT): New patterns to fold BUILTIN_POPCOUNT and
11408 friends. POPCOUNT(x&1) => x&1, POPCOUNT(x)==0 => x==0, etc.
11410 2018-05-24 H.J. Lu <hongjiu.lu@intel.com>
11414 * varasm.c (assemble_alias): Check ifunc_resolver only on
11417 2018-05-24 Uros Bizjak <ubizjak@gmail.com>
11420 * config/i386/sse.md (movdi_to_sse): Do not generate pseudo
11421 when memory input operand is handled.
11423 2018-05-24 Luis Machado <luis.machado@linaro.org>
11425 * config/aarch64/aarch64.c (qdf24xx_addrcost_table): New static
11427 (qdf24xx_tunings) <addr_costs>: Set to qdf24xx_addrcost_table.
11429 2018-05-24 Richard Sandiford <richard.sandiford@linaro.org>
11431 * match.pd: Delay FMA folds until after vectorization.
11433 2018-05-24 Andre Vieira <andre.simoesdiasvieira@arm.com>
11436 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Make
11437 address check not strict.
11439 2018-05-24 Richard Sandiford <richard.sandiford@linaro.org>
11441 * gimple-match.h (gimple_match_op): New class.
11442 (mprts_hook): Replace parameters with a gimple_match_op *.
11443 (maybe_build_generic_op): Likewise.
11444 (gimple_simplified_result_is_gimple_val): Replace parameters with
11445 a const gimple_match_op *.
11446 (gimple_simplify): Replace code_helper * and tree * parameters with
11447 a gimple_match_op * parameter.
11448 (gimple_resimplify1): Replace code_helper *, tree and tree *
11449 parameters with a gimple_match_op * parameter.
11450 (gimple_resimplify2): Likewise.
11451 (gimple_resimplify3): Likewise.
11452 (maybe_push_res_to_seq): Replace code_helper, tree and tree *
11453 parameters with a gimple_match_op * parameter.
11454 * gimple-match-head.c (gimple_simplify): Change prototypes of
11455 auto-generated functions to take a gimple_match_op * instead of
11456 separate code_helper * and tree * parameters. Make the same
11457 change in the top-level overload and update calls to the
11458 gimple_resimplify routines. Update calls to the auto-generated
11459 functions and to maybe_push_res_to_seq in the publicly-facing
11460 operation-specific gimple_simplify overloads.
11461 (gimple_match_op::MAX_NUM_OPS): Define.
11462 (gimple_resimplify1): Replace rcode and ops with a single res_op
11463 parameter. Update call to gimple_simplify.
11464 (gimple_resimplify2): Likewise.
11465 (gimple_resimplify3): Likewise.
11466 (mprts_hook): Replace parameters with a gimple_match_op *.
11467 (maybe_build_generic_op): Likewise.
11468 (build_call_internal): Replace type, nargs and ops with
11469 a gimple_match_op *.
11470 (maybe_push_res_to_seq): Replace res_code, type and ops parameters
11471 with a single gimple_match_op *. Update calls to mprts_hook,
11472 build_call_internal and gimple_simplified_result_is_gimple_val.
11473 Factor out code that is common to the tree_code and combined_fn cases.
11474 * genmatch.c (expr::gen_transform): Replace tem_code and
11475 tem_ops with a gimple_match_op called tem_op. Update calls
11476 to the gimple_resimplify functions and maybe_push_res_to_seq.
11477 (dt_simplify::gen_1): Manipulate res_op instead of res_code and
11478 res_ops. Update call to the gimple_resimplify functions.
11479 (dt_simplify::gen): Pass res_op instead of res_code and res_ops.
11480 (decision_tree::gen): Make the functions take a gimple_match_op *
11481 called res_op instead of separate res_code and res_ops parameters.
11482 Update call accordingly.
11483 * gimple-fold.c (replace_stmt_with_simplification): Replace rcode
11484 and ops with a single res_op parameter. Update calls to
11485 maybe_build_generic_op and maybe_push_res_to_seq.
11486 (fold_stmt_1): Update calls to gimple_simplify and
11487 replace_stmt_with_simplification.
11488 (gimple_fold_stmt_to_constant_1): Update calls to gimple_simplify
11489 and gimple_simplified_result_is_gimple_val.
11490 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update call to
11492 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Replace parameters
11493 with a gimple_match_op *.
11494 (vn_nary_build_or_lookup): Likewise. Update call to
11495 vn_nary_build_or_lookup_1.
11496 (vn_nary_build_or_lookup_1): Replace rcode, type and ops with a
11497 gimple_match_op *. Update calls to the gimple_resimplify routines
11498 and to gimple_simplified_result_is_gimple_val.
11499 (vn_nary_simplify): Update call to vn_nary_build_or_lookup_1.
11500 Use gimple_match_op::MAX_NUM_OPS instead of a hard-coded 3.
11501 (vn_reference_lookup_3): Update call to vn_nary_build_or_lookup.
11502 (visit_nary_op): Likewise.
11503 (visit_reference_op_load): Likewise.
11505 2018-05-23 Luis Machado <luis.machado@linaro.org>
11507 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Use correct type
11508 modifier for printing the step amount.
11510 2018-05-23 Jozef Lawrynowicz <jozef.l@somniumtech.com>
11513 * tree.c (build_common_tree_nodes): Dont set TYPE_SIZE for __intN
11516 2018-05-23 Segher Boessenkool <segher@kernel.crashing.org>
11518 * doc/sourcebuild.texi (Endianness): New subsubsection.
11520 2018-05-23 Luis Machado <luis.machado@linaro.org>
11522 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
11523 <prefetch_dynamic_strides>: New const bool field.
11524 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
11525 prefetch_dynamic_strides.
11526 (exynosm1_prefetch_tune): Likewise.
11527 (thunderxt88_prefetch_tune): Likewise.
11528 (thunderx_prefetch_tune): Likewise.
11529 (thunderx2t99_prefetch_tune): Likewise.
11530 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides to
11532 (aarch64_override_options_internal): Update to set
11533 PARAM_PREFETCH_DYNAMIC_STRIDES.
11534 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
11535 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
11536 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
11537 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
11538 prefetch-dynamic-strides setting.
11540 2018-05-23 Luis Machado <luis.machado@linaro.org>
11542 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
11543 <minimum_stride>: New const int field.
11544 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
11545 minimum_stride field defaulting to -1.
11546 (exynosm1_prefetch_tune): Likewise.
11547 (thunderxt88_prefetch_tune): Likewise.
11548 (thunderx_prefetch_tune): Likewise.
11549 (thunderx2t99_prefetch_tune): Likewise.
11550 (qdf24xx_prefetch_tune) <minimum_stride>: Set to 2048.
11551 <default_opt_level>: Set to 3.
11552 (aarch64_override_options_internal): Update to set
11553 PARAM_PREFETCH_MINIMUM_STRIDE.
11554 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
11555 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
11556 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
11557 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
11558 stride is constant and is below the minimum stride threshold.
11560 2018-05-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11562 * config/arm/arm-cpus.in (mode26): Delete.
11563 (armv4): Delete mode26 reference.
11564 * config/arm/arm.c (arm_configure_build_target): Delete use of
11567 2018-05-23 Uros Bizjak <ubizjak@gmail.com>
11569 * config/i386/i386.md (*floatuns<SWI48:mode><MODEF:mode>2_avx512):
11571 (floatunssi<mode>2): Also enable for AVX512F and TARGET_SSE_MATH.
11572 Rewrite expander pattern. Emit gen_floatunssi<mode>2_i387_with_xmm
11574 (floatunsdisf2): Rewrite expander pattern. Hanlde TARGET_AVX512F.
11575 (floatunsdidf2): Ditto.
11577 2018-05-23 Uros Bizjak <ubizjak@gmail.com>
11579 * config/i386/i386.md (fixuns_trunc<mode>di2): New insn pattern.
11580 (fixuns_trunc<mode>si2_avx512f): Ditto.
11581 (*fixuns_trunc<mode>si2_avx512f_zext): Ditto.
11582 (fixuns_trunc<mode>si2): Also enable for AVX512F and TARGET_SSE_MATH.
11583 Emit fixuns_trunc<mode>si2_avx512f for AVX512F targets.
11585 2018-05-23 Alexander Monakov <amonakov@ispras.ru>
11587 PR rtl-optimization/79985
11588 * df-scan.c (df_insn_refs_collect): Remove special case for
11589 global registers and asm statements.
11591 2018-05-23 Alexander Monakov <amonakov@ispras.ru>
11593 * extend.texi (Global Register Variables): Rewrite the bullet list.
11594 Note that the register is available for allocation. Note that access
11595 via inline asm must use constraints. Add note about async-signal
11596 handlers. Remove paragraph about automagic register selection.
11598 2018-05-23 Richard Biener <rguenther@suse.de>
11600 * tree-ssa-sccvn.c (vn_reference_lookup_3): Remove restriction
11601 of fixed offset from memset VN.
11603 2018-05-23 Bill Schmidt <wschmidt@linux.ibm.com>
11605 * gimple-ssa-strength-reduction.c (struct slsr_cand_d): Add
11606 first_interp field.
11607 (alloc_cand_and_find_basis): Initialize first_interp field.
11608 (slsr_process_mul): Modify first_interp field.
11609 (slsr_process_add): Likewise.
11610 (slsr_process_cast): Modify first_interp field for each new
11612 (slsr_process_copy): Likewise.
11613 (dump_candidate): Dump first_interp field.
11614 (replace_mult_candidate): Process all interpretations, not just
11616 (replace_rhs_if_not_dup): Likewise.
11617 (replace_one_candidate): Likewise.
11619 2018-05-23 Wilco Dijkstra <wdijkstr@arm.com>
11621 * config/aarch64/aarch64.c (aarch64_use_frame_pointer):
11623 (aarch64_needs_frame_chain): New function.
11624 (aarch64_parse_override_string): Set aarch64_use_frame_pointer.
11626 2018-05-23 Sudakshina Das <sudi.das@arm.com>
11629 * common/config/aarch64/aarch64-common.c (aarch64_handle_option):
11630 Check val before adding MASK_STRICT_ALIGN to opts->x_target_flags.
11631 * config/aarch64/aarch64.opt (mstrict-align): Remove RejectNegative.
11632 * config/aarch64/aarch64.c (aarch64_attributes): Mark allow_neg
11633 as true for strict-align.
11634 (aarch64_can_inline_p): Perform checks even when callee has no
11635 attributes to check for strict alignment.
11636 * doc/extend.texi (AArch64 Function Attributes): Document
11638 * doc/invoke.texi: (AArch64 Options): Likewise.
11640 2018-05-23 Richard Sandiford <richard.sandiford@linaro.org>
11642 PR tree-optimization/85853
11643 * tree-vect-slp.c (vect_slp_analyze_node_operations): Split out
11644 the handling of the root of the node to...
11645 (vect_slp_analyze_node_operations_1): ...this new function,
11646 and run the whole thing with the child nodes' def types
11647 set according to their SLP node's def type.
11649 2018-05-23 Richard Biener <rguenther@suse.de>
11651 PR middle-end/85874
11652 * tree-data-ref.c (create_runtime_alias_checks): Defer
11653 and ignore overflow warnings.
11655 2018-05-23 Yury Gribov <tetra2005@gmail.com>
11657 PR tree-optimization/85822
11658 * tree-vrp.c (is_masked_range_test): Fix handling of negative
11661 2018-05-23 Richard Biener <rguenther@suse.de>
11663 * tree-ssa-sccvn.c (vn_reference_lookup_3): Handle arbitrary
11664 memset constants via native_interpret_expr.
11666 2018-05-22 H.J. Lu <hongjiu.lu@intel.com>
11669 * cgraph.h (cgraph_node::create): Set ifunc_resolver for ifunc
11671 (cgraph_node::create_alias): Likewise.
11672 (cgraph_node::get_availability): Check ifunc_resolver instead
11673 of looking up ifunc attribute.
11674 * cgraphunit.c (maybe_diag_incompatible_alias): Likewise.
11675 * varasm.c (do_assemble_alias): Likewise.
11676 (assemble_alias): Likewise.
11677 (default_binds_local_p_3): Likewise.
11678 * cgraph.h (cgraph_node): Add ifunc_resolver.
11679 (cgraph_node::only_called_directly_or_aliased_p): Return false
11680 for IFUNC resolver.
11681 * lto-cgraph.c (input_node): Set ifunc_resolver for ifunc
11683 * symtab.c (symtab_node::verify_base): Verify that ifunc_resolver
11684 is equivalent to lookup_attribute ("ifunc", DECL_ATTRIBUTES (decl)).
11685 (symtab_node::binds_to_current_def_p): Check ifunc_resolver
11686 instead of looking up ifunc attribute.
11688 2018-05-22 Luis Machado <luis.machado@linaro.org>
11690 * config/aarch64/aarch64.md (*ashift<mode>_extv_bfiz): New pattern.
11692 2018-05-22 Martin Sebor <msebor@redhat.com>
11694 PR middle-end/85359
11695 * builtins.c (expand_builtin_strcpy): Call maybe_warn_nonstring_arg
11696 only when expasion succeeds.
11697 (expand_builtin_strcmp): Same.
11698 (expand_builtin_strncmp): Same.
11700 2018-05-22 Martin Sebor <msebor@redhat.com>
11702 * calls.c (maybe_warn_nonstring_arg): Fix a typo in a comment.
11704 2018-05-22 Jackson Woodruff <jackson.woodruff@arm.com>
11705 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11707 * config/aarch64/aarch64-ldpstp.md: Replace uses of
11708 aarch64_mem_pair_operand with memory_operand and delete operand swapping
11710 * config/aarch64/aarch64.c (aarch64_operands_ok_for_ldpstp):
11711 Add check for legitimate_address.
11712 (aarch64_gen_adjusted_ldpstp): Swap operands where appropriate.
11713 (aarch64_swap_ldrstr_operands): New.
11714 * config/aarch64/aarch64-protos.h (aarch64_swap_ldrstr_operands):
11717 2018-05-22 Jackson Woodruff <jackson.woodruff@arm.com>
11718 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11720 * config/aarch64/aarch64.md: New patterns to generate stp
11722 (store_pair_sw, store_pair_dw): New patterns to generate stp for
11723 single words and double words.
11724 (load_pair_sw, load_pair_dw): Likewise.
11725 (store_pair_sf, store_pair_df, store_pair_si, store_pair_di):
11727 (load_pair_sf, load_pair_df, load_pair_si, load_pair_di):
11729 * config/aarch64/aarch64-ldpstp.md: Modify peephole
11730 for different mode ldpstp and add peephole for merged zero stores.
11731 Likewise for loads.
11732 * config/aarch64/aarch64.c (aarch64_operands_ok_for_ldpstp):
11734 (aarch64_gen_store_pair): Rename calls to match new patterns.
11735 (aarch64_gen_load_pair): Rename calls to match new patterns.
11736 * config/aarch64/aarch64-simd.md (load_pair<mode>): Rename to...
11737 (load_pair<DREG:mode><DREG2:mode>): ... This.
11738 (store_pair<mode>): Rename to...
11739 (vec_store_pair<DREG:mode><DREG2:mode>): ... This.
11740 * config/aarch64/iterators.md (DREG, DREG2, DX2, SX, SX2, DSX):
11741 New mode iterators.
11742 (V_INT_EQUIV): Handle SImode.
11743 * config/aarch64/predicates.md (aarch64_reg_zero_or_fp_zero):
11746 2018-05-22 Martin Sebor <msebor@redhat.com>
11749 * calls.c (maybe_warn_nonstring_arg): Use string length to set
11750 or ajust the presumed bound on an operation to avoid unnecessary
11753 2018-05-22 Martin Sebor <msebor@redhat.com>
11755 PR tree-optimization/85826
11756 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Avoid
11757 assuming that a DECL necesarily has a constant size.
11759 2018-05-22 Richard Sandiford <richard.sandiford@linaro.org>
11761 PR middle-end/85862
11762 * internal-fn.c (expand_direct_optab_fn): Cope with a null lhs.
11764 2018-05-22 Richard Biener <rguenther@suse.de>
11766 PR tree-optimization/85834
11767 * tree-ssa-sccvn.c (vn_reference_lookup_3): Properly handle
11768 non-constant and non-zero memset arguments.
11770 2018-05-22 Martin Liska <mliska@suse.cz>
11773 * ipa-icf.c (sem_item::add_type): Do not ICE for incomplete types.
11775 2018-05-22 Richard Biener <rguenther@suse.de>
11777 PR tree-optimization/85863
11778 * tree-vect-stmts.c (vect_is_simple_cond): Only widen invariant
11779 comparisons when vectype is specified.
11780 (vectorizable_condition): Do not specify vectype for
11781 vect_is_simple_cond when SLP vectorizing.
11783 2018-05-21 Michael Meissner <meissner@linux.ibm.com>
11786 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Do not
11787 define __ibm128 as long double.
11788 * config/rs6000/rs6000.c (rs6000_init_builtins): Create __ibm128
11789 as a distinct type when IEEE 128-bit support is enabled.
11790 (init_float128_ieee): Fix up conversions between IFmode and IEEE
11791 128-bit types to use the correct functions.
11792 (rs6000_expand_float128_convert): Use explicit FLOAT_EXTEND to
11793 convert between 128-bit floating point types that have different
11794 modes but the same representation, instead of using gen_lowpart to
11796 * config/rs6000/rs6000.md (IFKF): New iterator for IFmode and
11798 (IFKF_reg): New attributes to give the register constraints for
11800 (extend<mode>tf2_internal): New insns to mark an explicit
11801 conversion between 128-bit floating point types that have a
11802 different mode but share the same representation.
11804 2018-05-21 Richard Sandiford <richard.sandiford@linaro.org>
11806 PR tree-optimization/85814
11807 * tree-ssa-strlen.c (get_stridx_plus_constant): Cope with
11808 a null return from get_strinfo when unsharing the next
11809 strinfo in the chain.
11811 2018-05-21 Vladimir Mezentsev <vladimir.mezentsev@oracle.com>
11814 * varasm.c (weak_finish): Clean up weak_decls.
11816 2018-05-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11818 * config/aarch64/aarch64.md ("unspec"): Define UNSPEC_SABAL,
11819 UNSPEC_SABDL2, UNSPEC_SADALP, UNSPEC_UABAL, UNSPEC_UABDL2,
11820 UNSPEC_UADALP values.
11821 * config/aarch64/iterators.md (ABAL): New int iterator.
11824 (sur): Add mappings for the above.
11825 * config/aarch64/aarch64-simd.md (aarch64_<sur>abdl2<mode>_3):
11827 (aarch64_<sur>abal<mode>_4): Likewise.
11828 (aarch64_<sur>adalp<mode>_3): Likewise.
11829 (<sur>sadv16qi): New define_expand.
11831 2018-05-21 Alexander Nesterovskiy <alexander.nesterovskiy@intel.com>
11833 * config/i386/i386.md (*movsf_internal): AVX falsedep fix.
11834 (*movdf_internal): Ditto.
11835 (*rcpsf2_sse): Ditto.
11836 (*rsqrtsf2_sse): Ditto.
11837 (*sqrt<mode>2_sse): Ditto.
11839 2018-05-21 Tamar Christina <tamar.christina@arm.com>
11841 * config/aarch64/aarch64-simd.md (aarch64_eor3qv8hi): Change to
11843 (aarch64_bcaxqv8hi): Change to bcaxq<mode>4.
11844 * config/aarch64/aarch64-simd-builtins.def (veor3q_u8, veor3q_u32,
11845 veor3q_u64, veor3q_s8, veor3q_s16, veor3q_s32, veor3q_s64, vbcaxq_u8,
11846 vbcaxq_u32, vbcaxq_u64, vbcaxq_s8, vbcaxq_s16, vbcaxq_s32,
11848 * config/aarch64/arm_neon.h: Likewise.
11849 * config/aarch64/iterators.md (VQ_I): New.
11851 2018-05-21 Alexey Brodkin <abrodkin@synopsys.com>
11853 * config.gcc: Add arc/t-multilib-linux to tmake_file for
11855 * config/arc/t-multilib-linux: Specify MULTILIB_OPTIONS and
11858 2018-05-20 Chung-Ju Wu <jasonwucj@gmail.com>
11860 * config/nds32/constraints.md (S): New constraint.
11861 * config/nds32/nds32.md (call_internal): Use constraint S.
11862 (call_value_internal): Likewise.
11863 (sibcall_internal): Likewise.
11864 (sibcall_value_internal): Likewise.
11866 2018-05-20 Kito Cheng <kito.cheng@gmail.com>
11867 Chung-Ju Wu <jasonwucj@gmail.com>
11869 * config/nds32/nds32.c (nds32_register_move_cost): Take garywolf cpu
11870 into consideration.
11872 2018-05-20 Kito Cheng <kito.cheng@gmail.com>
11873 Chung-Ju Wu <jasonwucj@gmail.com>
11875 * config/nds32/nds32-cost.c (rtx_cost_model_t): New structure.
11876 (insn_size_16bit, insn_size_32bit): New variables for cost evaluation.
11877 (nds32_rtx_costs_impl): Simplify.
11878 (nds32_address_cost_impl): Simplify.
11879 (nds32_init_rtx_costs): New function.
11880 (nds32_rtx_costs_speed_prefer): Likewise.
11881 (nds32_rtx_costs_size_prefer): Likewise.
11882 (nds32_address_cost_speed_prefer): Likewise.
11883 (nds32_address_cost_speed_fwprop): Likewise.
11884 (nds32_address_cost_size_prefer): Likewise.
11885 * config/nds32/nds32-protos.h (nds32_init_rtx_costs): Declare.
11886 * config/nds32/nds32.c (nds32_option_override): Use
11887 nds32_init_rtx_costs function.
11889 2018-05-20 Chung-Ju Wu <jasonwucj@gmail.com>
11891 * config/nds32/nds32.c (nds32_asm_file_start): Output pipeline model.
11892 * config/nds32/nds32.h (TARGET_PIPELINE_N7): Define.
11893 (TARGET_PIPELINE_N8): Likewise.
11894 (TARGET_PIPELINE_N10): Likewise.
11895 (TARGET_PIPELINE_N13): Likewise.
11896 (TARGET_PIPELINE_GRAYWOLF): Likewise.
11898 2018-05-19 Monk Chiang <sh.chiang04@gmail.com>
11900 * config/nds32/nds32-fpu.md: Update copyright year.
11902 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
11904 * config/nds32/nds32.h (ASM_SPEC): Adjust spec rule.
11906 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
11908 * config/nds32/nds32.c
11909 (nds32_md_asm_adjust): Consider flag_inline_asm_r15 variable.
11910 * config/nds32/nds32.opt (minline-asm-r15): New option.
11912 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
11914 * common/config/nds32/nds32-common.c (TARGET_DEFAULT_TARGET_FLAGS): Add
11916 * config/nds32/nds32.md (abssi2): New pattern.
11918 2018-05-19 Uros Bizjak <ubizjak@gmail.com>
11920 * config/i386/i386.md (rex64namesuffix): New mode attribute.
11921 * config/i386/sse.md (sse_cvtsi2ss<rex64namesuffix><round_name>):
11922 Merge insn pattern from sse_cvtsi2ss<round_name> and
11923 sse_cvtsi2ssq<round_name> using SWI48 mode iterator.
11924 (sse_cvtss2si<rex64namesuffix><round_name>): Merge insn pattern
11925 from sse_cvtss2si<round_name> and sse_cvtss2siq<round_name>
11926 using SWI48 mode iterator.
11927 (sse_cvtss2si<rex64namesuffix>_2): Merge insn pattern from
11928 sse_cvtss2si_2 and sse_cvtss2siq_2 using SWI48 mode iterator.
11929 (sse_cvttss2si<rex64namesuffix><round_saeonly_name>): Merge insn
11930 pattern from sse_cvttss2si<round_saeonly_name>
11931 and sse_cvttss2siq<round_saeonly_name> using SWI48 mode iterator.
11932 (avx512f_vcvtss2usi<rex64namesuffix><round_name>): Merge insn pattern
11933 from avx512f_vcvtss2usi<round_name> and avx512f_vcvtss2usiq<round_name>
11934 using SWI48 mode iterator.
11935 (avx512f_vcvttss2usi<rex64namesuffix><round_saeonly_name>): Merge
11936 insn pattern from avx512f_vcvttss2usi<round_saeonly_name> and
11937 avx512f_vcvttss2usiq<round_saeonly_name> using SWI48 mode iterator.
11938 (avx512f_vcvtsd2usi<rex64namesuffix><round_name>): Merge insn pattern
11939 from avx512f_vcvtsd2usi<round_name> and avx512f_vcvtsd2usiq<round_name>
11940 using SWI48 mode iterator.
11941 (avx512f_vcvttsd2usi<rex64namesuffix><round_saeonly_name>): Merge
11942 insn pattern from avx512f_vcvttsd2usi<round_saeonly_name> and
11943 avx512f_vcvttsd2usiq<round_saeonly_name> using SWI48 mode iterator.
11944 (sse2_cvtsd2si<rex64namesuffix><round_name>): Merge insn pattern from
11945 sse2_cvtsd2si<round_name> and sse2_cvtsd2siq<round_name> using
11946 SWI48 mode iterator.
11947 (sse2_cvtsd2si<rex64namesuffix>_2): Merge insn pattern from
11948 sse2_cvtsd2si_2 and sse2_cvtsd2siq_2 using SWI48 mode iterator.
11949 (sse_cvttsd2si<rex64namesuffix><round_saeonly_name>): Merge insn
11950 pattern from sse_cvttsd2si<round_saeonly_name>
11951 and sse_cvttsd2siq<round_saeonly_name> using SWI48 mode iterator.
11953 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
11955 * config/nds32/nds32-md-auxiliary.c
11956 (nds32_valid_smw_lwm_base_p): Refine.
11957 (nds32_output_smw_single_word): Refine.
11958 (nds32_output_smw_double_word): New.
11959 * config/nds32/nds32-protos.h (nds32_output_smw_double_word): New.
11961 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
11963 * config/nds32/nds32-md-auxiliary.c (nds32_output_stack_push): Refine.
11964 (nds32_output_stack_pop): Refine.
11965 (nds32_expand_unaligned_load): Refine.
11966 (nds32_expand_unaligned_store): Refine.
11968 2018-05-19 Kuan-Lin Chen <kuanlinchentw@gmail.com>
11969 Chung-Ju Wu <jasonwucj@gmail.com>
11971 * config/nds32/constants.md: Add TP_REGNUM constant.
11972 (unspec_element): Add UNSPEC_GOTINIT, UNSPEC_GOT, UNSPEC_GOTOFF,
11973 UNSPEC_PLT, UNSPEC_TLSGD, UNSPEC_TLSLD, UNSPEC_TLSIE, UNSPEC_TLSLE and
11975 * config/nds32/nds32-doubleword.md: Consider flag_pic.
11976 * config/nds32/nds32-dspext.md (mov<mode>): Expand TLS and PIC cases.
11977 * config/nds32/nds32-predicates.c (nds32_const_unspec_p): New.
11978 * config/nds32/nds32-md-auxiliary.c: Implementation that support TLS
11979 and PIC code generation.
11980 * config/nds32/nds32-protos.h: Declarations that support TLS and PIC
11982 * config/nds32/nds32-relax-opt.c: Consider TLS and PIC for relax
11984 * config/nds32/nds32.md: Support TLS and PIC.
11985 * config/nds32/nds32.c: Support TLS and PIC.
11986 * config/nds32/nds32.h (nds32_relax_insn_type): New enum type.
11987 * config/nds32/predicates.md (nds32_nonunspec_symbolic_operand): New
11990 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
11992 * config/nds32/nds32-predicates.c (const_vector_to_hwint): Use machine
11993 mode with E_ prefix.
11995 2018-05-19 Kuan-Lin Chen <kuanlinchentw@gmail.com>
11996 Chung-Ju Wu <jasonwucj@gmail.com>
11998 * config/nds32/constants.md (unspec_element): Add UNSPEC_ICT.
11999 * config/nds32/nds32-md-auxiliary.c
12000 (symbolic_reference_mentioned_p): New.
12001 (nds32_legitimize_ict_address): New.
12002 (nds32_expand_ict_move): New.
12003 (nds32_indirect_call_referenced_p): New.
12004 (nds32_symbol_binds_local_p): Delete.
12005 (nds32_long_call_p): Modify.
12006 * config/nds32/nds32-opts.h (nds32_ict_model_type): New enum type.
12007 * config/nds32/nds32-protos.h
12008 (symbolic_reference_mentioned_p): Declare.
12009 (nds32_legitimize_ict_address): Declare.
12010 (nds32_expand_ict_move): Declare.
12011 (nds32_indirect_call_referenced_p): Declare.
12012 * config/nds32/nds32-relax-opt.c (nds32_ict_const_p): New.
12013 (nds32_relax_group): Use nds32_ict_const_p as condition.
12014 * config/nds32/nds32.c (nds32_attribute_table): Add "indirect_call".
12015 (nds32_asm_file_start): Output ict_model directive in asm code.
12016 (nds32_legitimate_address_p): Consider indirect call.
12017 (nds32_print_operand): Consider indirect call.
12018 (nds32_print_operand_address): Consider indirect call.
12019 (nds32_insert_attributes): Handle "indirect_call" attribute.
12020 (TARGET_LEGITIMATE_ADDRESS_P): Define.
12021 (TARGET_LEGITIMATE_CONSTANT_P): Define.
12022 (TARGET_CANNOT_FORCE_CONST_MEM): Define.
12023 (TARGET_DELEGITIMIZE_ADDRESS): Define.
12024 (TARGET_ASM_OUTPUT_ADDR_CONST_EXTRA): Define.
12025 * config/nds32/nds32.h (SYMBOLIC_CONST_P): Define.
12026 (TARGET_ICT_MODEL_SMALL): Define.
12027 (TARGET_ICT_MODEL_LARGE): Define.
12028 * config/nds32/nds32.md (movsi): Consider ict model.
12029 (call, call_value): Consider ict model.
12030 (sibcall, sibcall_value): Consider ict model.
12031 * config/nds32/nds32.opt (mict-model): New option.
12032 * config/nds32/predicates.md (nds32_symbolic_operand): Consider ict
12035 2018-05-18 Kito Cheng <kito.cheng@gmail.com>
12036 Monk Chiang <sh.chiang04@gmail.com>
12037 Jim Wilson <jimw@sifive.com>
12039 * common/config/riscv/riscv-common.c (riscv_parse_arch_string):
12040 Add support to parse rv32e*. Clear MASK_RVE for rv32i and rv64i.
12041 * config.gcc (riscv*-*-*): Add support for rv32e* and ilp32e.
12042 * config/riscv/riscv-c.c (riscv_cpu_cpp_builtins): Define
12043 __riscv_32e when TARGET_RVE. Handle ABI_ILP32E as soft-float ABI.
12044 * config/riscv/riscv-opts.h (riscv_abi_type): Add ABI_ILP32E.
12045 * config/riscv/riscv.c (riscv_compute_frame_info): When TARGET_RVE,
12046 compute save_libcall_adjustment properly.
12047 (riscv_option_override): Call error if TARGET_RVE and not ABI_ILP32E.
12048 (riscv_conditional_register_usage): Handle TARGET_RVE and ABI_ILP32E.
12049 * config/riscv/riscv.h (UNITS_PER_FP_ARG): Handle ABI_ILP32E.
12050 (STACK_BOUNDARY, ABI_STACK_BOUNDARY): Handle TARGET_RVE.
12051 (GP_REG_LAST, MAX_ARGS_IN_REGISTERS): Likewise.
12052 (ABI_SPEC): Handle mabi=ilp32e.
12053 * config/riscv/riscv.opt (abi_type): Add ABI_ILP32E.
12054 (RVE): Add RVE mask.
12055 * doc/invoke.texi (RISC-V options) <-mabi>: Add ilp32e info.
12056 <-march>: Add rv32e as an example.
12058 2018-05-18 Marc Glisse <marc.glisse@inria.fr>
12061 * tree-ssa-structalias.c (create_variable_info_for_1): Extra argument.
12062 (intra_create_variable_infos): Handle C++ constructors.
12064 2018-05-18 Martin Liska <mliska@suse.cz>
12066 * passes.def: Remove a redundant pass.
12068 2018-05-18 Eric Botcazou <ebotcazou@adacore.com>
12071 * config/sparc/sparc.c (sparc_expand_builtin): Always initialize op[0].
12073 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12075 * config/arm/arm-cpus.in (armv3m, mode32): Delete features.
12077 (ARMv2, ARMv3, ARMv3m): Delete fgroups.
12079 (armv2, armv2a, armv3, armv3m): Delete architectures.
12080 (arm2, arm250, arm3, arm6, arm60, arm600, arm610, arm620,
12081 arm7, arm7d, arm7di, arm70, arm700, arm700i, arm710, arm720,
12082 arm710c, arm7100, arm7500, arm7500fe, arm7m, arm7dm, arm7dmi):
12084 * config/arm/arm.md (maddsidi4): Remove check for arm_arch3m.
12085 (*mulsidi3adddi): Likewise.
12086 (mulsidi3): Likewise.
12087 (*mulsidi3_nov6): Likewise.
12088 (umulsidi3): Likewise.
12089 (umulsidi3_nov6): Likewise.
12090 (umaddsidi4): Likewise.
12091 (*umulsidi3adddi): Likewise.
12092 (smulsi3_highpart): Likewise.
12093 (*smulsi3_highpart_nov6): Likewise.
12094 (umulsi3_highpart): Likewise.
12095 (*umulsi3_highpart_nov6): Likewise.
12096 * config/arm/arm.h (arm_arch3m): Delete.
12097 * config/arm/arm.c (arm_arch3m): Delete.
12098 (arm_option_override_internal): Update armv3-related comment.
12099 (arm_configure_build_target): Delete use of isa_bit_mode32.
12100 (arm_option_reconfigure_globals): Delete set of arm_ach3m.
12101 (arm_rtx_costs_internal): Delete check of arm_arch3m.
12102 * config/arm/arm-fixed.md (mulsq3): Delete check for arm_arch3m.
12103 (mulsa3): Likewise.
12104 (mulusa3): Likewise.
12105 * config/arm/arm-protos.h (arm_arch3m): Delete.
12106 * config/arm/arm-tables.opt: Regenerate.
12107 * config/arm/arm-tune.md: Likewise.
12108 * config/arm/t-arm-elf (all_early_nofp): Delete mentions of
12109 deleted architectures.
12111 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12113 * config/arm/arm-cpus.in (armv5, armv5e): Delete features.
12114 (armv5t, armv5te): New features.
12115 (ARMv5, ARMv5e): Delete fgroups.
12116 (ARMv5t, ARMv5te): Adjust for above changes.
12117 (ARMv6m): Likewise.
12118 (armv5, armv5e): Delete arches.
12119 * config/arm/arm.md (*call_reg_armv5): Use arm_arch5t instead of
12121 (*call_reg_arm): Likewise.
12122 (*call_value_reg_armv5): Likewise.
12123 (*call_value_reg_arm): Likewise.
12124 (*call_symbol): Likewise.
12125 (*call_value_symbol): Likewise.
12126 (*sibcall_insn): Likewise.
12127 (*sibcall_value_insn): Likewise.
12128 (clzsi2): Likewise.
12129 (prefetch): Likewise.
12130 (define_split and define_peephole2 dependent on arm_arch5):
12132 * config/arm/arm.h (TARGET_LDRD): Use arm_arch5te instead of
12134 (TARGET_ARM_QBIT): Likewise.
12135 (TARGET_DSP_MULTIPLY): Likewise.
12136 (enum base_architecture): Delete BASE_ARCH_5, BASE_ARCH_5E.
12137 (arm_arch5, arm_arch5e): Delete.
12138 (arm_arch5t, arm_arch5te): Declare.
12139 * config/arm/arm.c (arm_arch5, arm_arch5e): Delete.
12140 (arm_arch5t): Declare.
12141 (arm_option_reconfigure_globals): Update for the above.
12142 (arm_options_perform_arch_sanity_checks): Update comment, replace
12143 use of arm_arch5 with arm_arch5t.
12144 (use_return_insn): Likewise.
12145 (arm_emit_call_insn): Likewise.
12146 (output_return_instruction): Likewise.
12147 (arm_final_prescan_insn): Likewise.
12148 (arm_coproc_builtin_available): Likewise.
12149 * config/arm/arm-c.c (arm_cpu_builtins): Replace arm_arch5 and
12150 arm_arch5e with arm_arch5t and arm_arch5te.
12151 * config/arm/arm-protos.h (arm_arch5, arm_arch5e): Delete.
12152 (arm_arch5t, arm_arch5te): Declare.
12153 * config/arm/arm-tables.opt: Regenerate.
12154 * config/arm/t-arm-elf: Remove references to armv5, armv5e.
12155 * config/arm/t-multilib: Likewise.
12156 * config/arm/thumb1.md (*call_reg_thumb1_v5): Check arm_arch5t
12157 instead of arm_arch5.
12158 (*call_reg_thumb1): Likewise.
12159 (*call_value_reg_thumb1_v5): Likewise.
12160 (*call_value_reg_thumb1): Likewise.
12161 * config/arm/vxworks.h (TARGET_OS_CPP_BUILTINS): Remove now
12163 * doc/invoke.texi (ARM Options): Remove references to armv5, armv5e.
12165 2018-05-18 Martin Liska <mliska@suse.cz>
12167 PR gcov-profile/84846
12168 * doc/gcov.texi: Document -t option of gcov tool.
12170 2018-05-18 Martin Liska <mliska@suse.cz>
12172 PR gcov-profile/84846
12173 * gcov.c (print_usage): Add new -t option.
12174 (process_args): Handle the option.
12175 (generate_results): Use stdout as output when requested by
12178 2018-05-18 Martin Liska <mliska@suse.cz>
12180 PR gcov-profile/84846
12181 * coverage.c (coverage_init): Write PWD to .gcno file.
12182 * doc/gcov.texi: Document how working directory is printed.
12183 * gcov-dump.c (dump_gcov_file): Print PWD.
12184 * gcov.c (output_intermediate_file): Likewise.
12185 (read_graph_file): Read PWD string.
12186 (output_lines): Print PWD.
12188 2018-05-18 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
12190 PR middle-end/85817
12191 * ipa-pure-const.c (malloc_candidate_p): Remove the check integer_zerop
12192 for retval and return false if all args to phi are zero.
12194 2018-05-18 Richard Biener <rguenther@suse.de>
12196 * gimple-ssa-evrp.c (class evrp_folder): Add simplify_stmt_using_ranges
12198 (evrp_dom_walker::before_dom_children): Call it.
12200 2018-05-18 Richard Biener <rguenther@suse.de>
12202 * tree-dfa.c (get_ref_base_and_extent): Use range-info to refine
12203 results when processing array refs with variable index.
12205 2018-05-18 Toon Moene <toon@moene.org>
12207 * doc/invoke.texi: Move -floop-unroll-and-jam documentation
12208 directly after that of -floop-interchange. Indicate that both
12209 options are enabled by default when specifying -O3.
12211 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12213 * config/aarch64/aarch64-simd.md (vec_set<mode>): Use VALL_F16 mode
12214 iterator. Delete separate integer-mode vec_set<mode> expander.
12215 (aarch64_simd_vec_setv2di): Delete.
12216 (vec_setv2di): Delete.
12217 (aarch64_simd_vec_set<mode>): Delete all other patterns with that name.
12218 Use VALL_F16 mode iterator. Add LD1 alternative and use vwcore for
12219 the "w, r" alternative.
12221 2018-05-18 Martin Liska <mliska@suse.cz>
12223 * passes.def: Add pass_lower_switch and pass_lower_switch_O0.
12224 * tree-pass.h (make_pass_lower_switch_O0): New function.
12225 * tree-switch-conversion.c (node_has_low_bound): Remove.
12226 (node_has_high_bound): Likewise.
12227 (node_is_bounded): Likewise.
12228 (class pass_lower_switch): Make it a template type and create
12230 (pass_lower_switch::execute): Add template argument.
12231 (make_pass_lower_switch): New function.
12232 (make_pass_lower_switch_O0): New function.
12233 (do_jump_if_equal): Remove.
12234 (emit_case_nodes): Simplify to just handle all 3 cases and leave
12235 all the hard work to tree optimization passes.
12237 2018-05-18 Martin Liska <mliska@suse.cz>
12239 * dbgcnt.c (limit_low): Renamed from limit.
12240 (limit_high): New variable.
12241 (dbg_cnt_is_enabled): Check for upper limit.
12242 (dbg_cnt): Adjust dumping.
12243 (dbg_cnt_set_limit_by_index): Add new argument for high
12245 (dbg_cnt_set_limit_by_name): Likewise.
12246 (dbg_cnt_process_single_pair): Parse new format.
12247 (dbg_cnt_process_opt): Use strtok.
12248 (dbg_cnt_list_all_counters): Remove 'value' and add
12250 * doc/invoke.texi: Document changes.
12252 2018-05-18 Richard Sandiford <richard.sandiford@linaro.org>
12254 * doc/sourcebuild.texi (scalar_all_fma): Document.
12255 * tree.def (FMA_EXPR): Delete.
12256 * internal-fn.def (FMA, FMS, FNMA, FNMS): New internal functions.
12257 * internal-fn.c (ternary_direct): New macro.
12258 (expand_ternary_optab_fn): Likewise.
12259 (direct_ternary_optab_supported_p): Likewise.
12260 * Makefile.in (build/genmatch.o): Depend on case-fn-macros.h.
12261 * builtins.c (fold_builtin_fma): Delete.
12262 (fold_builtin_3): Don't call it.
12263 * cfgexpand.c (expand_debug_expr): Remove FMA_EXPR handling.
12264 * expr.c (expand_expr_real_2): Likewise.
12265 * fold-const.c (operand_equal_p): Likewise.
12266 (fold_ternary_loc): Likewise.
12267 * gimple-pretty-print.c (dump_ternary_rhs): Likewise.
12268 * gimple.c (DEFTREECODE): Likewise.
12269 * gimplify.c (gimplify_expr): Likewise.
12270 * optabs-tree.c (optab_for_tree_code): Likewise.
12271 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
12272 * tree-eh.c (operation_could_trap_p): Likewise.
12273 (stmt_could_throw_1_p): Likewise.
12274 * tree-inline.c (estimate_operator_cost): Likewise.
12275 * tree-pretty-print.c (dump_generic_node): Likewise.
12276 (op_code_prio): Likewise.
12277 * tree-ssa-loop-im.c (stmt_cost): Likewise.
12278 * tree-ssa-operands.c (get_expr_operands): Likewise.
12279 * tree.c (commutative_ternary_tree_code, add_expr): Likewise.
12280 * fold-const-call.h (fold_fma): Delete.
12281 * fold-const-call.c (fold_const_call_ssss): Handle CFN_FMS,
12282 CFN_FNMA and CFN_FNMS.
12283 (fold_fma): Delete.
12284 * genmatch.c (combined_fn): New enum.
12285 (commutative_ternary_tree_code): Remove FMA_EXPR handling.
12286 (commutative_op): New function.
12287 (commutate): Use it. Handle more than 2 operands.
12288 (dt_operand::gen_gimple_expr): Use commutative_op.
12289 (parser::parse_expr): Allow :c to be used with non-binary
12290 operators if the commutative operand is known.
12291 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Handle
12292 CFN_FMS, CFN_FNMA and CFN_FNMS.
12293 (backprop::process_assign_use): Remove FMA_EXPR handling.
12294 * hsa-gen.c (gen_hsa_insns_for_operation_assignment): Likewise.
12295 (gen_hsa_fma): New function.
12296 (gen_hsa_insn_for_internal_fn_call): Use it for IFN_FMA, IFN_FMS,
12297 IFN_FNMA and IFN_FNMS.
12298 * match.pd: Add folds for IFN_FMS, IFN_FNMA and IFN_FNMS.
12299 * gimple-fold.h (follow_all_ssa_edges): Declare.
12300 * gimple-fold.c (follow_all_ssa_edges): New function.
12301 * tree-ssa-math-opts.c (convert_mult_to_fma_1): Use the
12302 gimple_build interface and use follow_all_ssa_edges to fold the result.
12303 (convert_mult_to_fma): Use direct_internal_fn_suppoerted_p
12304 instead of checking for optabs directly.
12305 * config/i386/i386.c (ix86_add_stmt_cost): Recognize FMAs as calls
12306 rather than FMA_EXPRs.
12307 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Create a
12308 call to IFN_FMA instead of an FMA_EXPR.
12310 2018-05-17 Jim Wilson <jimw@sifive.com>
12312 * expr.c (do_tablejump): When converting index to Pmode, if we have a
12313 sign extended promoted subreg, and the range does not have the sign bit
12314 set, then do a sign extend.
12316 * config/riscv/riscv.c (riscv_extend_comparands): In unsigned QImode
12317 test, check for sign extended subreg and/or constant operands, and
12318 do a sign extend in that case.
12320 2018-05-17 Steve Ellcey <sellcey@cavium.com>
12322 * config/aarch64/thunderx2t99.md (thunderx2t99_ls_both): Delete.
12323 (thunderx2t99_multiple): Delete psuedo-units from used cpus.
12325 (thunderx2t99_alu_shift): Remove alu_shift_reg, alus_shift_reg.
12326 Change logics_shift_reg to logics_shift_imm.
12327 (thunderx2t99_fp_loadpair_basic): Delete.
12328 (thunderx2t99_fp_storepair_basic): Delete.
12329 (thunderx2t99_asimd_int): Add neon_sub and neon_sub_q types.
12330 (thunderx2t99_asimd_polynomial): Delete.
12331 (thunderx2t99_asimd_fp_simple): Add neon_fp_mul_s_scalar_q
12332 and neon_fp_mul_d_scalar_q.
12333 (thunderx2t99_asimd_fp_conv): Add *int_to_fp* types.
12334 (thunderx2t99_asimd_misc): Delete neon_dup and neon_dup_q.
12335 (thunderx2t99_asimd_recip_step): Add missing *sqrt* types.
12336 (thunderx2t99_asimd_lut): Add missing tbl types.
12337 (thunderx2t99_asimd_ext): Delete.
12338 (thunderx2t99_asimd_load1_1_mult): Delete.
12339 (thunderx2t99_asimd_load1_2_mult): Delete.
12340 (thunderx2t99_asimd_load1_ldp): New.
12341 (thunderx2t99_asimd_load1): New.
12342 (thunderx2t99_asimd_load2): Add missing *load2* types.
12343 (thunderx2t99_asimd_load3): New.
12344 (thunderx2t99_asimd_load4): New.
12345 (thunderx2t99_asimd_store1_1_mult): Delete.
12346 (thunderx2t99_asimd_store1_2_mult): Delete.
12347 (thunderx2t99_asimd_store2_mult): Delete.
12348 (thunderx2t99_asimd_store2_onelane): Delete.
12349 (thunderx2t99_asimd_store_stp): New.
12350 (thunderx2t99_asimd_store1): New.
12351 (thunderx2t99_asimd_store2): New.
12352 (thunderx2t99_asimd_store3): New.
12353 (thunderx2t99_asimd_store4): New.
12355 2018-05-17 Jerome Lambourg <lambourg@adacore.com>
12357 * config/arm/arm_cmse.h (cmse_nsfptr_create, cmse_is_nsfptr): Remove
12358 #include <stdint.h>. Replace intptr_t with __INTPTR_TYPE__.
12360 2018-05-17 Pat Haugen <pthaugen@us.ibm.com>
12361 Segher Boessenkool <segher@kernel.crashing.org>
12364 * config/rs6000/rs6000.c (rs6000_output_move_128bit): Check dest
12367 2018-05-17 Richard Biener <rguenther@suse.de>
12369 * tree-ssa-dse.c (dse_classify_store): Fix iterator increment
12370 for pruning loop and prune defs feeding only already visited PHIs.
12372 2018-05-17 Richard Biener <rguenther@suse.de>
12374 * tree-ssa-sccvn.c (vn_reference_lookup_3): Improve memset handling.
12376 2018-05-17 Bin Cheng <bin.cheng@arm.com>
12377 Richard Biener <rguenther@suse.de>
12379 PR tree-optimization/85793
12380 * tree-vect-stmts.c (vectorizable_load): Handle 1 element-wise load
12381 for VMAT_ELEMENTWISE.
12383 2018-05-17 Richard Sandiford <richard.sandiford@linaro.org>
12385 * internal-fn.h (lookup_internal_fn): Declare
12386 * internal-fn.c (lookup_internal_fn): New function.
12387 * gimple.c (gimple_build_call_from_tree): Handle calls to
12388 internal functions.
12389 * gimple-pretty-print.c (dump_gimple_call): Print "." before
12390 internal function names.
12391 * tree-pretty-print.c (dump_generic_node): Likewise.
12392 * tree-ssa-scopedtables.c (expr_hash_elt::print): Likewise.
12394 2018-05-17 Richard Sandiford <richard.sandiford@linaro.org>
12396 * gimple-fold.h (gimple_build): Make the function forms take
12397 combined_fn rather than built_in_function.
12398 (gimple_simplify): Likewise.
12399 * gimple-match-head.c (gimple_simplify): Likewise.
12400 * gimple-fold.c (gimple_build): Likewise.
12401 * tree-vect-loop.c (get_initial_def_for_reduction): Use gimple_build
12402 rather than gimple_build_call_internal.
12403 (get_initial_defs_for_reduction): Likewise.
12404 (vect_create_epilog_for_reduction): Likewise.
12405 (vectorizable_live_operation): Likewise.
12407 2018-05-17 Martin Liska <mliska@suse.cz>
12409 * gimple-ssa-sprintf.c (format_directive): Do not use
12410 space in between 'G_' and '('.
12412 2018-05-17 Jakub Jelinek <jakub@redhat.com>
12415 * config/i386/i386.c (ix86_fold_builtin): Handle masked shifts
12416 even if the mask is not all ones.
12419 * config/i386/i386.c (ix86_fold_builtin): Fold shift builtins by
12421 (ix86_gimple_fold_builtin): Likewise.
12424 * config/i386/i386.c: Include tree-vector-builder.h.
12425 (ix86_vector_shift_count): New function.
12426 (ix86_fold_builtin): Fold shift builtins by scalar count.
12427 (ix86_gimple_fold_builtin): Likewise.
12429 * config/i386/avx512fintrin.h (_mm512_set_epi16, _mm512_set_epi8,
12430 _mm512_setzero): New intrinsics.
12432 2018-05-17 James Greenhalgh <james.greenhalgh@arm.com>
12433 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12435 * config/aarch64/aarch64.c (aarch64_expand_vector_init): Modify
12436 code generation for cases where splatting a value is not useful.
12437 * simplify-rtx.c (simplify_ternary_operation): Simplify
12438 vec_merge across a vec_duplicate and a paradoxical subreg forming
12439 a vector mode to a vec_concat.
12441 2018-05-17 Olga Makhotina <olga.makhotina@intel.com>
12443 * config.gcc: Support "goldmont-plus".
12444 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
12446 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
12447 PROCESSOR_GOLDMONT_PLUS.
12448 * config/i386/i386.c (m_GOLDMONT_PLUS): Define.
12449 (processor_target_table): Add "goldmont-plus".
12450 (PTA_GOLDMONT_PLUS): Define.
12451 (ix86_lea_outperforms): Add TARGET_GOLDMONT_PLUS.
12452 (get_builtin_code_for_version): Handle PROCESSOR_GOLDMONT_PLUS.
12453 (fold_builtin_cpu): Add M_INTEL_GOLDMONT_PLUS.
12454 (fold_builtin_cpu): Add "goldmont-plus".
12455 (ix86_add_stmt_cost): Add TARGET_GOLDMONT_PLUS.
12456 (ix86_option_override_internal): Add "goldmont-plus".
12457 * config/i386/i386.h (processor_costs): Define TARGET_GOLDMONT_PLUS.
12458 (processor_type): Add PROCESSOR_GOLDMONT_PLUS.
12459 * config/i386/x86-tune.def: Add m_GOLDMONT_PLUS.
12460 * doc/invoke.texi: Add goldmont-plus as x86 -march=/-mtune= CPU type.
12462 2018-05-17 Richard Biener <rguenther@suse.de>
12464 PR tree-optimization/85757
12465 * tree-ssa-dse.c (dse_classify_store): Record a PHI def and
12466 remove defs that only feed that PHI from further processing.
12468 2018-05-16 Jim Wilson <jimw@sifive.com>
12470 * config/riscv/riscv.md (<optab>si3_mask, <optab>si3_mask_1): Prepend
12472 (<optab>di3_mask, <optab>di3_mask_1): Likewise.
12474 2018-05-16 Mark Wielaard <mark@klomp.org>
12476 * dwarf2out.c (count_index_strings): New function.
12477 (output_indirect_strings): Call count_index_strings and generate
12478 header for dwarf_version >= 5.
12480 2018-05-16 Mark Wielaard <mark@klomp.org>
12482 * dwarf2out.c (dwarf_FORM): New function.
12483 (set_indirect_string): Use dwarf_FORM.
12484 (reset_indirect_string): Likewise.
12485 (size_of_die): Likewise.
12486 (value_format): Likewise.
12487 (output_die): Likewise.
12488 (add_skeleton_AT_string): Likewise.
12489 (output_macinfo_op): Likewise.
12490 (index_string): Likewise.
12491 (output_index_string_offset): Likewise.
12492 (output_index_string): Likewise.
12493 (count_index_strings): Likewise.
12495 2018-05-16 Carl Love <cel@us.ibm.com>
12497 * config/rs6000/rs6000.md (prefetch): Generate ISA 2.06 instructions
12498 dcbt and dcbtstt with TH=16 if operands[2] is 0 and Power 8 or newer.
12500 2018-05-16 Martin Jambor <mjambor@suse.cz>
12502 * ipa-prop.c (ipa_free_all_edge_args): Remove.
12503 * ipa-prop.h (ipa_free_all_edge_args): Likewise.
12505 2018-05-16 Wilco Dijkstra <wdijkstr@arm.com>
12507 * config/aarch64/aarch64.md (fma<mode>4): Change into expand pattern.
12508 (fnma<mode>4): Likewise.
12509 (fms<mode>4): Likewise.
12510 (fnms<mode>4): Likewise.
12511 (aarch64_fma<mode>4): Rename insn, reorder accumulator operand.
12512 (aarch64_fnma<mode>4): Likewise.
12513 (aarch64_fms<mode>4): Likewise.
12514 (aarch64_fnms<mode>4): Likewise.
12515 (aarch64_fnmadd<mode>4): Likewise.
12517 2018-05-16 Jason Merrill <jason@redhat.com>
12519 * tree.c (warn_deprecated_use): Return bool. Simplify logic.
12521 2018-05-16 Richard Biener <rguenther@suse.de>
12523 * tree-vectorizer.h (struct stmt_info_for_cost): Add where member.
12524 (dump_stmt_cost): Declare.
12525 (add_stmt_cost): Dump cost we add.
12526 (add_stmt_costs): New function.
12527 (vect_model_simple_cost, vect_model_store_cost, vect_model_load_cost):
12528 No longer exported.
12529 (vect_analyze_stmt): Adjust prototype.
12530 (vectorizable_condition): Likewise.
12531 (vectorizable_live_operation): Likewise.
12532 (vectorizable_reduction): Likewise.
12533 (vectorizable_induction): Likewise.
12534 * tree-vect-loop.c (vect_analyze_loop_operations): Create local
12535 cost vector to pass to vectorizable_ and record afterwards.
12536 (vect_model_reduction_cost): Take cost vector argument and adjust.
12537 (vect_model_induction_cost): Likewise.
12538 (vectorizable_reduction): Likewise.
12539 (vectorizable_induction): Likewise.
12540 (vectorizable_live_operation): Likewise.
12541 * tree-vect-slp.c (vect_create_new_slp_node): Initialize
12542 SLP_TREE_NUMBER_OF_VEC_STMTS.
12543 (vect_analyze_slp_cost_1): Remove.
12544 (vect_analyze_slp_cost): Likewise.
12545 (vect_slp_analyze_node_operations): Take visited args and
12546 a target cost vector. Avoid processing already visited stmt sets.
12547 (vect_slp_analyze_operations): Use a local cost vector to gather
12548 costs and register those of non-discarded instances.
12549 (vect_bb_vectorization_profitable_p): Use add_stmt_costs.
12550 (vect_schedule_slp_instance): Remove copying of
12551 SLP_TREE_NUMBER_OF_VEC_STMTS. Instead assert that it is not
12553 * tree-vect-stmts.c (record_stmt_cost): Remove path directly
12554 adding cost. Record cost entry location.
12555 (vect_prologue_cost_for_slp_op): Function to compute cost of
12556 a constant or invariant generated for SLP vect in the prologue,
12557 split out from vect_analyze_slp_cost_1.
12558 (vect_model_simple_cost): Make static. Adjust for SLP costing.
12559 (vect_model_promotion_demotion_cost): Likewise.
12560 (vect_model_store_cost): Likewise, make static.
12561 (vect_model_load_cost): Likewise.
12562 (vectorizable_bswap): Add cost vector arg and adjust.
12563 (vectorizable_call): Likewise.
12564 (vectorizable_simd_clone_call): Likewise.
12565 (vectorizable_conversion): Likewise.
12566 (vectorizable_assignment): Likewise.
12567 (vectorizable_shift): Likewise.
12568 (vectorizable_operation): Likewise.
12569 (vectorizable_store): Likewise.
12570 (vectorizable_load): Likewise.
12571 (vectorizable_condition): Likewise.
12572 (vectorizable_comparison): Likewise.
12573 (can_vectorize_live_stmts): Likewise.
12574 (vect_analyze_stmt): Likewise.
12575 (vect_transform_stmt): Adjust calls to vectorizable_*.
12576 * tree-vectorizer.c: Include gimple-pretty-print.h.
12577 (dump_stmt_cost): New function.
12579 2018-05-16 Richard Biener <rguenther@suse.de>
12581 * params.def (PARAM_DSE_MAX_ALIAS_QUERIES_PER_STORE): New param.
12582 * doc/invoke.texi (dse-max-alias-queries-per-store): Document.
12583 * tree-ssa-dse.c: Include tree-ssa-loop.h.
12584 (check_name): New callback.
12585 (dse_classify_store): Track cycles via a visited bitmap of PHI
12586 defs and simplify handling of in-loop and across loop dead stores
12587 and properly fail for loop-variant refs. Handle byte-tracking with
12588 multiple defs. Use PARAM_DSE_MAX_ALIAS_QUERIES_PER_STORE for
12591 2018-05-16 Richard Sandiford <richard.sandiford@linaro.org>
12593 * tree-vectorizer.h (vect_get_vector_types_for_stmt): Declare.
12594 (vect_get_mask_type_for_stmt): Likewise.
12595 * tree-vect-slp.c (vect_two_operations_perm_ok_p): New function,
12597 (vect_build_slp_tree_1): ...here. Use vect_get_vector_types_for_stmt
12598 to determine the statement's vector type and the vector type that
12599 should be used for calculating nunits. Deal with cases in which
12600 the type has to be deferred.
12601 (vect_slp_analyze_node_operations): Use vect_get_vector_types_for_stmt
12602 and vect_get_mask_type_for_stmt to calculate STMT_VINFO_VECTYPE.
12603 * tree-vect-loop.c (vect_determine_vf_for_stmt_1)
12604 (vect_determine_vf_for_stmt): New functions, split out from...
12605 (vect_determine_vectorization_factor): ...here.
12606 * tree-vect-stmts.c (vect_get_vector_types_for_stmt)
12607 (vect_get_mask_type_for_stmt): New functions, split out from
12608 vect_determine_vectorization_factor.
12610 2018-05-16 Richard Biener <rguenther@suse.de>
12612 * tree-cfg.c (verify_gimple_assign_ternary): Properly
12613 verify the [VEC_]COND_EXPR embedded comparison.
12615 2018-05-15 Martin Sebor <msebor@redhat.com>
12617 PR tree-optimization/85753
12618 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Handle
12619 RECORD_TYPE in addition to ARRAY_TYPE.
12621 2018-05-15 Martin Sebor <msebor@redhat.com>
12623 PR middle-end/85643
12624 * calls.c (get_attr_nonstring_decl): Handle MEM_REF.
12626 2018-05-15 Richard Biener <rguenther@suse.de>
12628 * tree-ssa-dse.c (dse_classify_store): Remove use_stmt parameter,
12629 add by_clobber_p one. Change algorithm to collect all defs
12630 representing uses we need to walk and try reducing them to
12631 a single one before failing.
12632 (dse_dom_walker::dse_optimize_stmt): Adjust.
12634 2018-05-13 Mark Wielaard <mark@klomp.org>
12636 * dwarf2out.c (dwarf_OP): Handle DW_OP_addrx and DW_OP_constx.
12637 (size_of_loc_descr): Likewise.
12638 (output_loc_operands): Likewise.
12639 (output_loc_operands_raw): Likewise.
12640 (dw_addr_op): Use dwarf_OP () for DW_OP_constx and DW_OP_addrx.
12641 (resolve_addr_in_expr): Handle DW_OP_addrx and DW_OP_constx.
12642 (hash_loc_operands): Likewise.
12643 (compare_loc_operands): Likewise.
12645 2018-05-14 Mark Wielaard <mark@klomp.org>
12647 * dwarf2out.c (count_index_addrs): New function.
12648 (dwarf2out_finish): Use count_index_addrs to calculate addrs_length.
12650 2018-05-15 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
12652 PR tree-optimization/83648
12653 * ipa-pure-const.c (malloc_candidate_p): Allow function with NULL
12654 return value as malloc candidate.
12656 2018-05-15 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
12659 * ipa-pure-const.c (warn_function_malloc): Pass value of known_finite
12660 param as true in call to suggest_attribute.
12662 2018-05-14 Segher Boessenkool <segher@kernel.crashing.org>
12664 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete @itemx for
12665 -mreadonly-in-sdata.
12667 2018-05-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12669 * config/aarch64/aarch64-simd.md (*aarch64_crypto_aese_fused):
12671 (aarch64_crypto_aesd_fused): Likewise.
12673 2018-05-14 Wilco Dijkstra <wdijkstr@arm.com>
12675 * config/aarch64/aarch64.md (mov<mode>): Remove '*' in alternatives.
12676 (movsi_aarch64): Likewise.
12677 (load_pairsi): Likewise.
12678 (load_pairdi): Likewise.
12679 (store_pairsi): Likewise.
12680 (store_pairdi): Likewise.
12681 (load_pairsf): Likewise.
12682 (load_pairdf): Likewise.
12683 (store_pairsf): Likewise.
12684 (store_pairdf): Likewise.
12685 (zero_extend): Likewise.
12686 (trunc): Swap alternatives.
12687 (fcvt_target): Add '?' to prefer w over r.
12689 2018-05-14 Jakub Jelinek <jakub@redhat.com>
12692 * config/i386/i386.md: Disallow non-commutative arithmetics in
12693 last twpeephole for mem {+,-,&,|,^}= x; mem != 0 after cmpelim
12694 optimization. Use COMMUTATIVE_ARITH_P test rather than != MINUS
12695 in the peephole2 before it.
12697 2018-05-14 Sebastian Peryt <sebastian.peryt@intel.com>
12699 * common/config/i386/i386-common.c (OPTION_MASK_ISA_CLDEMOTE_SET,
12700 OPTION_MASK_ISA_CLDEMOTE_UNSET): New defines.
12701 (ix86_handle_option): Handle -mcldemote.
12702 * config.gcc: New header.
12703 * config/i386/cldemoteintrin.h: New file.
12704 * config/i386/cpuid.h (bit_CLDEMOTE): New bit.
12705 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
12707 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
12708 OPTION_MASK_ISA_CLDEMOTE.
12709 * config/i386/i386.c (ix86_target_string): Add -mcldemote.
12710 (ix86_valid_target_attribute_inner_p): Ditto.
12711 (enum ix86_builtins): Add IX86_BUILTIN_CLDEMOTE.
12712 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_cldemote.
12713 (ix86_expand_builtin): Expand IX86_BUILTIN_CLDEMOTE.
12714 * config/i386/i386.h (TARGET_CLDEMOTE, TARGET_CLDEMOTE_P): New.
12715 * config/i386/i386.md (UNSPECV_CLDEMOTE): New.
12717 * config/i386/i386.opt: Add -mcldemote.
12718 * config/i386/x86intrin.h: New header.
12719 * doc/invoke.texi: Add -mcldemote.
12721 2018-05-14 Richard Biener <rguenther@suse.de>
12723 * doc/match-and-simplify.texi: Adjust :s documentation.
12725 2018-05-14 Alexander Monakov <amonakov@ispras.ru>
12727 * sort.cc (REORDER_23): Pass the type for the temporaries instead of
12728 intended memcpy size.
12729 (REORDER_45): Likewise.
12731 2018-05-13 Alexander Monakov <amonakov@ispras.ru>
12733 * sort.cc: New file.
12734 * system.h [!CHECKING_P] (qsort): Redirect to gcc_qsort.
12735 * vec.c (qsort_chk): Use gcc_qsort.
12736 * Makefile.in (OBJS-libcommon): Add sort.o.
12737 (build/sort.o): New target. Use it...
12738 (BUILD_RTL): ... here, and...
12739 (build/gencfn-macros): ... here, and...
12740 (build/genmatch): ... here.
12742 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
12743 Chung-Ju Wu <jasonwucj@gmail.com>
12745 * config.gcc (nds32*-*-*): Check that n15 is valid to --with-cpu.
12746 * config/nds32/nds32-graywolf.md: New file.
12747 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_GRAYWOLF.
12748 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n15
12750 * config/nds32/nds32-protos.h: More declarations for n15 pipeline.
12751 * config/nds32/nds32-utils.c: More implementations for n15 pipeline.
12752 * config/nds32/nds32.md (pipeline_model): Add graywolf.
12753 * config/nds32/nds32.opt (mcpu): Support n15 pipeline cpus.
12754 * config/nds32/pipelines.md: Include n15 settings.
12756 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
12757 Chung-Ju Wu <jasonwucj@gmail.com>
12759 * config.gcc (nds32*-*-*): Check that n12/n13 are valid to --with-cpu.
12760 * config/nds32/nds32-n13.md: New file.
12761 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N12 and CPU_N13.
12762 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n12/n13
12764 * config/nds32/nds32-protos.h: More declarations for n12/n13 pipeline.
12765 * config/nds32/nds32.md (pipeline_model): Add n13.
12766 * config/nds32/nds32.opt (mcpu): Support n12/n13 pipeline cpus.
12767 * config/nds32/pipelines.md: Include n13 settings.
12769 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
12770 Chung-Ju Wu <jasonwucj@gmail.com>
12772 * config.gcc (nds32*-*-*): Check that n10/d10 are valid to --with-cpu.
12773 * config/nds32/nds32-n10.md: New file.
12774 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N10.
12775 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n10
12777 * config/nds32/nds32-protos.h: More declarations for n10 pipeline.
12778 * config/nds32/nds32-utils.c: More implementations for n10 pipeline.
12779 * config/nds32/nds32.md (pipeline_model): Add n10.
12780 * config/nds32/nds32.opt (mcpu): Support n10 pipeline cpus.
12781 * config/nds32/pipelines.md: Include n10 settings.
12783 2018-05-13 Monk Chiang <sh.chiang04@gmail.com>
12784 Kito Cheng <kito.cheng@gmail.com>
12785 Chung-Ju Wu <jasonwucj@gmail.com>
12787 * config.gcc (nds32be-*-*): Handle --with-ext-dsp.
12788 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
12789 Add enum values for DSP extension instructions.
12790 * config/nds32/constraints.md (Iu06, IU06, CVp5, CVs5, CVs2, CVhi):
12792 * config/nds32/iterators.md (shifts, shiftrt, sat_plus, all_plus,
12793 sat_minus, all_minus, plus_minus, extend, sumax, sumin, sumin_max):
12794 New code iterators.
12795 (su, zs, uk, opcode, add_rsub, add_sub): New code attributes.
12796 * config/nds32/nds32-dspext.md: New file for DSP implementation.
12797 * config/nds32/nds32-intrinsic.c: Implementation of DSP extension.
12798 * config/nds32/nds32-intrinsic.md: Likewise.
12799 * config/nds32/nds32_intrinsic.h: Likewise.
12800 * config/nds32/nds32-md-auxiliary.c: Likewise.
12801 * config/nds32/nds32-memory-manipulation.c: Consider DSP extension.
12802 * config/nds32/nds32-predicates.c (const_vector_to_hwint): New.
12803 (nds32_valid_CVp5_p, nds32_valid_CVs5_p): New.
12804 (nds32_valid_CVs2_p, nds32_valid_CVhi_p): New.
12805 * config/nds32/nds32-protos.h: New declarations for DSP extension.
12806 * config/nds32/nds32-utils.c (extract_mac_non_acc_rtx): New case
12807 TYPE_DMAC in switch statement.
12808 * config/nds32/nds32.c: New checking and implementation for DSP
12809 extension instructions.
12810 * config/nds32/nds32.h: Likewise.
12811 * config/nds32/nds32.md: Likewise.
12812 * config/nds32/nds32.opt (mhw-abs, mext-dsp): New options.
12813 * config/nds32/predicates.md: Implement new predicates for DSP
12816 2018-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
12818 * config/rs6000/rs6000.md (mov<mode>_softfloat, FMOVE32):
12819 Reformat alternatives and attributes so it is easier to identify
12820 which constraints/attributes go with which instruction.
12821 (mov<mode>_hardfloat32, FMOVE64): Likewise.
12822 (mov<mode>_softfloat32, FMOVE64): Likewise.
12823 (mov<mode>_hardfloat64, FMOVE64): Likewise.
12824 (mov<mode>_softfloat64, FMOVE64): Likewise.
12826 2018-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
12828 * doc/extend.texi (PowerPC Built-in Functions): Rename this
12830 (Basic PowerPC Built-in Functions): The new name of the
12831 subsection previously known as "PowerPC Built-in Functions".
12832 (Basic PowerPC Built-in Functions Available on all Configurations):
12834 (Basic PowerPC Built-in Functions Available on ISA 2.05): Likewise.
12835 (Basic PowerPC Built-in Functions Available on ISA 2.06): Likewise.
12836 (Basic PowerPC Built-in Functions Available on ISA 2.07): Likewise.
12837 (Basic PowerPC Built-in Functions Available on ISA 3.0): Likewise.
12839 2018-05-11 Martin Jambor <mjambor@suse.cz>
12842 * ipa-cp.c (intersect_with_plats): Check that the lattice contains
12845 2018-05-11 Richard Earnshaw <rearnsha@arm.com>
12848 * config/arm/arm-cpus.in (fgroup ARMv6m): Add be8 feature.
12850 2018-05-11 Sebastian Peryt <sebastian.peryt@intel.com>
12852 * common/config/i386/i386-common.c (OPTION_MASK_ISA_WAITPKG_SET,
12853 OPTION_MASK_ISA_WAITPKG_UNSET): New defines.
12854 (ix86_handle_option): Handle -mwaitpkg.
12855 * config.gcc: New header.
12856 * config/i386/cpuid.h (bit_WAITPKG): New bit.
12857 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mwaitpkg.
12858 * config/i386/i386-builtin-types.def ((UINT8, UNSIGNED, UINT64)): New
12860 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
12861 OPTION_MASK_ISA_WAITPKG.
12862 * config/i386/i386.c (ix86_target_string): Add -mwaitpkg.
12863 (ix86_option_override_internal): Add PTA_WAITPKG.
12864 (ix86_valid_target_attribute_inner_p): Add -mwaitpkg.
12865 (enum ix86_builtins): Add IX86_BUILTIN_UMONITOR, IX86_BUILTIN_UMWAIT,
12866 IX86_BUILTIN_TPAUSE.
12867 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_umonitor,
12868 __builtin_ia32_umwait and __builtin_ia32_tpause.
12869 (ix86_expand_builtin): Expand IX86_BUILTIN_UMONITOR,
12870 IX86_BUILTIN_UMWAIT, IX86_BUILTIN_TPAUSE.
12871 * config/i386/i386.h (TARGET_WAITPKG, TARGET_WAITPKG_P): New.
12872 * config/i386/i386.md (UNSPECV_UMWAIT, UNSPECV_UMONITOR,
12873 UNSPECV_TPAUSE): New.
12874 (umwait, umwait_rex64, umonitor_<mode>, tpause, tpause_rex64): New.
12875 * config/i386/i386.opt: Add -mwaitpkg.
12876 * config/i386/waitpkgintrin.h: New file.
12877 * config/i386/x86intrin.h: New header.
12878 * doc/invoke.texi: Add -mwaitpkg.
12880 2018-05-11 Richard Earnshaw <rearnsha@arm.com>
12883 * config/arm/arm-cpus.in: Add comment that ARMv6-m and ARMv6S-m are now
12885 (cortex-m0): Use armv6s-m isa.
12886 (cortex-m0plus): Likewise.
12887 (cortex-m1): Likewise.
12888 (cortex-m0.small-multiply): Likewise.
12889 (cortex-m0plus.small-multiply): Likewise.
12890 (cortex-m1.small-multiply): Likewise.
12892 2018-05-11 Allan Sandfeld Jensen <allan.jensen@qt.io>
12893 Jakub Jelinek <jakub@redhat.com>
12895 PR tree-optimization/85692
12896 * tree-ssa-forwprop.c (simplify_vector_constructor): Try two
12897 source permute as well.
12899 2018-05-11 Martin Liska <mliska@suse.cz>
12902 * doc/extend.texi: Document LLVM style format for no_sanitize
12905 2018-05-10 Michael Meissner <meissner@linux.vnet.ibm.com>
12907 * config/rs6000/rs6000.c (mode_supports_dq_form): Rename
12908 mode_supports_vsx_dform_quad to mode_supports_dq_form.
12909 (mode_supports_vsx_dform_quad): Likewise.
12910 (mode_supports_vmx_dform): Move these functions to be next to the
12911 other mode_supports functions.
12912 (mode_supports_dq_form): Likewise.
12913 (quad_address_p): Change calls of mode_supports_vsx_dform_quad to
12914 mode_supports_dq_form.
12915 (reg_offset_addressing_ok_p): Likewise.
12916 (offsettable_ok_by_alignment): Likewise.
12917 (rs6000_legitimate_offset_address_p): Likewise.
12918 (legitimate_lo_sum_address_p): Likewise.
12919 (rs6000_legitimize_address): Likewise.
12920 (rs6000_legitimize_reload_address): Likewise.
12921 (rs6000_secondary_reload_inner): Likewise.
12922 (rs6000_preferred_reload_class): Likewise.
12923 (rs6000_output_move_128bit): Likewise.
12925 2018-05-10 Uros Bizjak <ubizjak@gmail.com>
12927 * config/i386/i386.c (ix86_expand_builtin) <case IX86_BUILTIN_RDPID>:
12928 Generate SImode target register for null target.
12929 <case IX86_BUILTIN_XGETBV>: Ditto.
12930 <case IX86_BUILTIN_XSETBV>: Optimize LSHIFTRT generation.
12931 * config/i386/xsaveintrin.h (_xgetbv): Add missing return.
12933 2018-05-10 Carl Love <cel@us.ibm.com>
12935 * config/rs6000/rs6000.md (prefetch): Generate ISA 2.06 instructions
12936 dcbtt and dcbtstt if operands[2] is 0.
12938 2018-05-10 Uros Bizjak <ubizjak@gmail.com>
12941 * config/i386/sse.md (usadv64qi): New expander.
12943 2018-05-10 Segher Boessenkool <segher@kernel.crashing.org>
12945 * config/rs6000/altivec.md (altivec_vmrghb, altivec_vmrghh,
12946 altivec_vmrghw, altivec_vmrglb, altivec_vmrglh, altivec_vmrglw): Remove
12947 -maltivec=be support.
12948 (vec_widen_umult_even_v16qi, vec_widen_smult_even_v16qi,
12949 vec_widen_umult_even_v8hi, vec_widen_smult_even_v8hi,
12950 vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
12951 vec_widen_umult_odd_v16qi, vec_widen_smult_odd_v16qi,
12952 vec_widen_umult_odd_v8hi, vec_widen_smult_odd_v8hi,
12953 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si, altivec_vpkpx,
12954 altivec_vpks<VI_char>ss, altivec_vpks<VI_char>us,
12955 altivec_vpku<VI_char>us, altivec_vpku<VI_char>um, altivec_vsum2sws,
12956 altivec_vsumsws): Adjust.
12957 (altivec_vspltb *altivec_vspltb_internal, altivec_vsplth,
12958 *altivec_vsplth_internal, altivec_vspltw, *altivec_vspltw_internal,
12959 altivec_vspltsf, *altivec_vspltsf_internal): Remove -maltivec=be
12961 (altivec_vperm_<mode>, altivec_vperm_<mode>_uns,
12962 altivec_vupkhs<VU_char>, altivec_vupkls<VU_char>, altivec_vupkhpx,
12963 altivec_vupklpx, altivec_lvsl, altivec_lvsr): Adjust.
12964 (altivec_lve<VI_char>x): Delete expand.
12965 (*altivec_lve<VI_char>x_internal): Rename to...
12966 (altivec_lve<VI_char>x): ... this.
12967 (altivec_lvxl_<mode>): Delete expand.
12968 (*altivec_lvxl_<mode>_internal): Rename to ...
12969 (altivec_lvxl_<mode>): ... this.
12970 (altivec_stvxl_<mode>): Delete expand.
12971 (*altivec_stvxl_<mode>_internal): Rename to ...
12972 (altivec_stvxl_<mode>): ... this.
12973 (altivec_stve<VI_char>x): Delete expand.
12974 (*altivec_stve<VI_char>x_internal): Rename to ...
12975 (altivec_stve<VI_char>x): ... this.
12976 (doublee<mode>2, unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2,
12977 doubleh<mode>2, unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2,
12978 reduc_plus_scal_<mode>): Adjust.
12979 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Adjust
12981 (rs6000_cpu_cpp_builtins): Adjust.
12982 (altivec_resolve_overloaded_builtin): Remove -maltivec=be support.
12983 * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be,
12984 altivec_expand_stvx_be, altivec_expand_stvex_be): Delete.
12985 * config/rs6000/rs6000.c (rs6000_option_override_internal): Remove
12986 -maltivec=be support.
12987 (rs6000_split_vec_extract_var): Adjust.
12988 (rs6000_split_v4si_init): Adjust.
12989 (swap_selector_for_mode): Delete.
12990 (altivec_expand_lvx_be, altivec_expand_stvx_be,
12991 altivec_expand_stvex_be): Delete.
12992 (altivec_expand_lv_builtin, altivec_expand_stv_builtin): Remove
12993 -maltivec=be support.
12994 (rs6000_gimple_fold_builtin): Ditto.
12995 (rs6000_generate_float2_double_code, rs6000_generate_float2_code):
12997 * config/rs6000/rs6000.h (VECTOR_ELT_ORDER_BIG): Delete.
12998 (TARGET_DIRECT_MOVE_64BIT): Adjust.
12999 * config/rs6000/rs6000.md (split for extendsidi2 for vectors): Adjust.
13000 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Delete.
13001 * config/rs6000/vsx.md (floate<mode>, unsfloatev2di, floato<mode>,
13002 unsfloatov2di, vsignedo_v2df, vsignede_v2df, vunsignedo_v2df,
13003 vunsignede_v2df, vsx_extract_<mode>_p9, *vsx_extract_si,
13004 *vsx_extract_<mode>_p8, *vsx_extract_si_<uns>float_df,
13005 *vsx_extract_si_<uns>float_<mode>, vsx_set_<mode>_p9, vsx_set_v4sf_p9,
13006 *vsx_insert_extract_v4sf_p9, *vsx_insert_extract_v4sf_p9_2, and an
13007 anonymous split): Adjust.
13008 (vsx_mergel_<mode>, vsx_mergeh_<mode>): Remove -maltivec=be support.
13009 (vsx_xxspltd_<mode>, extract4b, insert4b): Adjust.
13011 2018-05-10 Eric Botcazou <ebotcazou@adacore.com>
13013 * configure.ac (gcc_gxx_include_dir_add_sysroot): Set it to 1 only
13014 when --with-gxx-include-dir is also specified.
13015 * configure: Regenerate.
13017 2018-05-09 Jim Wilson <jimw@sifive.com>
13020 * config.gcc (riscv*-*-*): Handle --with-multilib-list.
13021 * config/riscv/t-withmultilib: New.
13022 * config/riscv/withmultilib.h: New.
13023 * doc/install.texi: Document RISC-V --with-multilib-list support.
13025 2018-05-09 Richard Biener <rguenther@suse.de>
13027 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Fill a cost
13029 (vect_bb_vectorization_profitable_p): Adjust. Compute
13030 actual scalar cost using the cost vector and the add_stmt_cost
13033 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
13035 PR rtl-optimization/85645
13036 * config/rs6000/rs6000.c (rs6000_emit_prologue_components): Put a SET
13037 in the REG_CFA_REGISTER note for LR, don't leave it empty.
13039 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
13041 PR rtl-optimization/85645
13042 * shrink-wrap.c (spread_components): Return a boolean saying if
13043 anything was changed.
13044 (try_shrink_wrapping_separate): Iterate spread_components until
13045 nothing changes anymore.
13047 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
13049 PR rtl-optimization/85645
13050 * regrename.c (build_def_use): Also kill the chains that include the
13051 destination of a REG_CFA_REGISTER note.
13053 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
13055 PR rtl-optimization/85645
13056 * regcprop.c (copyprop_hardreg_forward_1): Don't propagate into an
13057 insn that has a REG_CFA_REGISTER note.
13059 2018-05-09 Richard Sandiford <richard.sandiford@linaro.org>
13061 * cfgexpand.c (expand_clobber): New function.
13062 (expand_gimple_stmt_1): Use it.
13063 * tree-vect-stmts.c (vect_clobber_variable): New function,
13065 (vectorizable_simd_clone_call): ...here.
13066 (vectorizable_store): Emit a clobber either side of an
13067 IFN_STORE_LANES sequence.
13068 (vectorizable_load): Emit a clobber after an IFN_LOAD_LANES sequence.
13070 2018-05-09 Tom de Vries <tom@codesourcery.com>
13073 * config/nvptx/nvptx.md (define_insn "trap", define_insn "trap_if_true")
13074 (define_insn "trap_if_false"): Add exit after trap.
13076 2018-05-09 Eric Botcazou <ebotcazou@adacore.com>
13078 PR rtl-optimization/85638
13079 * bb-reorder.c: Include common/common-target.h.
13080 (create_forwarder_block): New function extracted from...
13081 (fix_up_crossing_landing_pad): ...here. Rename into...
13082 (dw2_fix_up_crossing_landing_pad): ...this.
13083 (sjlj_fix_up_crossing_landing_pad): New function.
13084 (find_rarely_executed_basic_blocks_and_crossing_edges): In SJLJ mode,
13085 call sjlj_fix_up_crossing_landing_pad if there are incoming EH edges
13086 from both partitions and exit the loop after one iteration.
13088 2018-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
13091 * doc/extend.texi (PowerPC Built-in Functions): Rename this
13093 (Basic PowerPC Built-in Functions): The new name of the
13094 subsection previously known as "PowerPC Built-in Functions".
13095 (Basic PowerPC Built-in Functions Available on all Configurations):
13097 (Basic PowerPC Built-in Functions Available on ISA 2.05): New
13099 (Basic PowerPC Built-in Functions Available on ISA 2.06): New
13101 (Basic PowerPC Built-in Functions Available on ISA 2.07): New
13103 (Basic PowerPC Built-in Functions Available on ISA 3.0): New
13106 2018-05-08 Jim Wilson <jimw@sifive.com>
13108 * config/riscv/linux.h (MUSL_ABI_SUFFIX): Delete unnecessary backslash.
13109 (LD_EMUL_SUFFIX): New.
13110 (LINK_SPEC): Use it.
13112 2018-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
13114 * doc/extend.texi (PowerPC Built-in Functions): Rename this
13116 (Basic PowerPC Built-in Functions): The new name of the
13117 subsection previously known as "PowerPC Built-in Functions".
13118 (Basic PowerPC Built-in Functions Available on all Configurations):
13120 (Basic PowerPC Built-in Functions Available on ISA 2.05): New
13122 (Basic PowerPC Built-in Functions Available on ISA 2.06): New
13124 (Basic PowerPC Built-in Functions Available on ISA 2.07): New
13126 (Basic PowerPC Built-in Functions Available on ISA 3.0): New
13129 2018-05-08 Jakub Jelinek <jakub@redhat.com>
13132 * config/i386/i386.md: Add peepholes for mem {+,-,&,|,^}= x; mem != 0
13133 after cmpelim optimization.
13135 2018-05-08 Olga Makhotina <olga.makhotina@intel.com>
13137 * config.gcc: Support "goldmont".
13138 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "goldmont".
13139 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
13140 PROCESSOR_GOLDMONT.
13141 * config/i386/i386.c (m_GOLDMONT): Define.
13142 (processor_target_table): Add "goldmont".
13143 (PTA_GOLDMONT): Define.
13144 (ix86_lea_outperforms): Add TARGET_GOLDMONT.
13145 (get_builtin_code_for_version): Handle PROCESSOR_GOLDMONT.
13146 (fold_builtin_cpu): Add M_INTEL_GOLDMONT.
13147 (fold_builtin_cpu): Add "goldmont".
13148 (ix86_add_stmt_cost): Add TARGET_GOLDMONT.
13149 (ix86_option_override_internal): Add "goldmont".
13150 * config/i386/i386.h (processor_costs): Define TARGET_GOLDMONT.
13151 (processor_type): Add PROCESSOR_GOLDMONT.
13152 * config/i386/i386.md: Add CPU "glm".
13153 * config/i386/glm.md: New file.
13154 * config/i386/x86-tune.def: Add m_GOLDMONT.
13155 * doc/invoke.texi: Add goldmont as x86 -march=/-mtune= CPU type.
13157 2018-05-08 Jakub Jelinek <jakub@redhat.com>
13160 * config/i386/i386.c (ix86_expand_sse2_abs): Handle E_V2DImode and
13162 * config/i386/sse.md (abs<mode>2): Use VI_AVX2 iterator instead of
13163 VI1248_AVX512VL_AVX512BW. Handle V2DImode and V4DImode if not
13164 TARGET_AVX512VL using ix86_expand_sse2_abs. Formatting fixes.
13167 * config/i386/i386.c (ix86_fold_builtin): Handle
13168 IX86_BUILTIN_{,P}MOVMSK{PS,PD,B}{,128,256}.
13171 * config/i386/sse.md (ssequaterinsnmode): New mode attribute.
13172 (*<extract_type>_vinsert<shuffletype><extract_suf>_0): New pattern.
13174 2018-05-08 Richard Earnshaw <rearnsha@arm.com>
13177 * config/arm/parsecpu.awk (check_cpu): Fix operator precedence.
13178 (check_arch): Likewise.
13179 (check_fpu): Return the result rather than printing it.
13180 (end arch): Fix operator precedence.
13181 (end cpu): Likewise.
13182 (END): Print the result from check_fpu.
13184 2018-05-08 Richard Sandiford <richard.sandiford@linaro.org>
13185 Alan Hayward <alan.hayward@arm.com>
13186 David Sherwood <david.sherwood@arm.com>
13188 * config/aarch64/aarch64-sve.md (*pred_cmp<cmp_op><mode>_combine)
13189 (*pred_cmp<cmp_op><mode>, *fcm<cmp_op><mode>_and_combine)
13190 (*fcmuo<mode>_and_combine, *fcm<cmp_op><mode>_and)
13191 (*fcmuo<mode>_and): New patterns.
13193 2018-05-08 Richard Sandiford <richard.sandiford@linaro.org>
13195 * config/aarch64/iterators.md (UNSPEC_COND_LO, UNSPEC_COND_LS)
13196 (UNSPEC_COND_HI, UNSPEC_COND_HS, UNSPEC_COND_UO): Delete.
13197 (SVE_INT_CMP, SVE_FP_CMP): New code iterators.
13198 (cmp_op, sve_imm_con): New code attributes.
13199 (SVE_COND_INT_CMP, imm_con): Delete.
13200 (cmp_op): Remove above unspecs from int attribute.
13201 * config/aarch64/aarch64-sve.md (*vec_cmp<cmp_op>_<mode>): Rename
13203 (*cmp<cmp_op><mode>): ...this. Use UNSPEC_MERGE_PTRUE instead of
13204 comparison-specific unspecs.
13205 (*vec_cmp<cmp_op>_<mode>_ptest): Rename to...
13206 (*cmp<cmp_op><mode>_ptest): ...this and adjust likewise.
13207 (*vec_cmp<cmp_op>_<mode>_cc): Rename to...
13208 (*cmp<cmp_op><mode>_cc): ...this and adjust likewise.
13209 (*vec_fcm<cmp_op><mode>): Rename to...
13210 (*fcm<cmp_op><mode>): ...this and adjust likewise.
13211 (*vec_fcmuo<mode>): Rename to...
13212 (*fcmuo<mode>): ...this and adjust likewise.
13213 (*pred_fcm<cmp_op><mode>): New pattern.
13214 * config/aarch64/aarch64.c (aarch64_emit_unop, aarch64_emit_binop)
13215 (aarch64_emit_sve_ptrue_op, aarch64_emit_sve_ptrue_op_cc): New
13217 (aarch64_unspec_cond_code): Remove handling of LTU, GTU, LEU, GEU
13219 (aarch64_gen_unspec_cond, aarch64_emit_unspec_cond): Delete.
13220 (aarch64_emit_sve_predicated_cond): New function.
13221 (aarch64_expand_sve_vec_cmp_int): Use aarch64_emit_sve_ptrue_op_cc.
13222 (aarch64_emit_unspec_cond_or): Replace with...
13223 (aarch64_emit_sve_or_conds): ...this new function. Use
13224 aarch64_emit_sve_ptrue_op for the individual comparisons and
13225 aarch64_emit_binop to OR them together.
13226 (aarch64_emit_inverted_unspec_cond): Replace with...
13227 (aarch64_emit_sve_inverted_cond): ...this new function. Use
13228 aarch64_emit_sve_ptrue_op for the comparison and
13229 aarch64_emit_unop to invert the result.
13230 (aarch64_expand_sve_vec_cmp_float): Update after the above
13231 changes. Use aarch64_emit_sve_ptrue_op for native comparisons.
13233 2018-05-07 Nathan Sidwell <nathan@acm.org>
13235 * doc/invoke.texi (C++ Dialect Options): Remove -ffor-scope.
13236 * doc/extend.texi (Deprecated Features): Remove -fno-for-scope
13237 (Backwards Compatibility): Likewise.
13239 2018-05-07 Luis Machado <luis.machado@linaro.org>
13243 2018-05-07 Luis Machado <luis.machado@linaro.org>
13245 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
13246 <prefetch_dynamic_strides>: New const bool field.
13247 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
13248 prefetch_dynamic_strides.
13249 (exynosm1_prefetch_tune): Likewise.
13250 (thunderxt88_prefetch_tune): Likewise.
13251 (thunderx_prefetch_tune): Likewise.
13252 (thunderx2t99_prefetch_tune): Likewise.
13253 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides
13255 (aarch64_override_options_internal): Update to set
13256 PARAM_PREFETCH_DYNAMIC_STRIDES.
13257 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
13258 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
13259 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
13260 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
13261 prefetch-dynamic-strides setting.
13263 2018-05-07 Luis Machado <luis.machado@linaro.org>
13265 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
13266 <minimum_stride>: New const int field.
13267 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
13268 minimum_stride field.
13269 (exynosm1_prefetch_tune): Likewise.
13270 (thunderxt88_prefetch_tune): Likewise.
13271 (thunderx_prefetch_tune): Likewise.
13272 (thunderx2t99_prefetch_tune): Likewise.
13273 (qdf24xx_prefetch_tune): Likewise. Set minimum_stride to 2048.
13274 (aarch64_override_options_internal): Update to set
13275 PARAM_PREFETCH_MINIMUM_STRIDE.
13276 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
13277 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
13278 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
13279 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
13280 stride is constant and is below the minimum stride threshold.
13282 2018-05-07 Luis Machado <luis.machado@linaro.org>
13284 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune) <l2_cache_size>: Set
13287 2018-05-07 Luis Machado <luis.machado@linaro.org>
13289 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
13290 <prefetch_dynamic_strides>: New const bool field.
13291 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
13292 prefetch_dynamic_strides.
13293 (exynosm1_prefetch_tune): Likewise.
13294 (thunderxt88_prefetch_tune): Likewise.
13295 (thunderx_prefetch_tune): Likewise.
13296 (thunderx2t99_prefetch_tune): Likewise.
13297 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides
13299 (aarch64_override_options_internal): Update to set
13300 PARAM_PREFETCH_DYNAMIC_STRIDES.
13301 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
13302 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
13303 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
13304 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
13305 prefetch-dynamic-strides setting.
13307 2018-05-07 Luis Machado <luis.machado@linaro.org>
13309 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
13310 <minimum_stride>: New const int field.
13311 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
13312 minimum_stride field.
13313 (exynosm1_prefetch_tune): Likewise.
13314 (thunderxt88_prefetch_tune): Likewise.
13315 (thunderx_prefetch_tune): Likewise.
13316 (thunderx2t99_prefetch_tune): Likewise.
13317 (qdf24xx_prefetch_tune): Likewise. Set minimum_stride to 2048.
13318 (aarch64_override_options_internal): Update to set
13319 PARAM_PREFETCH_MINIMUM_STRIDE.
13320 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
13321 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
13322 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
13323 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
13324 stride is constant and is below the minimum stride threshold.
13326 2018-05-06 Jakub Jelinek <jakub@redhat.com>
13329 * cfgexpand.c (expand_asm_stmt): Don't create a temporary if
13330 the type is addressable. Don't force op into register if it has
13333 2018-05-05 Roland McGrath <mcgrathr@google.com>
13336 * varasm.c (default_section_type_flags): Set SECTION_NOTYPE for
13337 any section for which we don't know a specific type it should have,
13338 regardless of name. Previously this was done only for the exact
13339 names ".init_array", ".fini_array", and ".preinit_array".
13340 (default_elf_asm_named_section): Add comment about
13341 relationship with default_section_type_flags and SECTION_NOTYPE.
13342 (get_section): Don't consider it a type conflict if one side has
13343 SECTION_NOTYPE and the other doesn't, as long as neither has the
13344 SECTION_BSS et al used in the default_section_type_flags logic.
13346 2018-05-05 Tom de Vries <tom@codesourcery.com>
13349 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG_3): Define.
13350 (workaround_barsyncs): New function.
13351 (nvptx_reorg): Use workaround_barsyncs.
13352 * config/nvptx/nvptx.md (define_c_enum "unspecv"): Add UNSPECV_MEMBAR.
13353 (define_expand "nvptx_membar_cta"): New define_expand.
13354 (define_insn "*nvptx_membar_cta"): New insn.
13356 2018-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
13358 * brig-builtins.def: Add consts to ptrs etc. in BRIG builtin defs.
13359 To improve optimization opportunities.
13360 * builtin-types.def: The new needed builtin types for the above.
13362 2018-05-04 Richard Biener <rguenther@suse.de>
13364 * bb-reorder.c (sanitize_hot_paths): Release hot_bbs_to_check.
13365 * gimple-ssa-store-merging.c
13366 (imm_store_chain_info::output_merged_store): Remove redundant create,
13367 release split_store vector contents on failure.
13368 * tree-vect-slp.c (vect_schedule_slp_instance): Avoid leaking
13369 scalar stmt vector on cache hit.
13371 2018-05-04 Segher Boessenkool <segher@kernel.crashing.org>
13373 * common/config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
13375 * config.gcc (powerpc-xilinx-eabi*): Remove.
13376 * config/rs6000/predicates.md (easy_fp_constant): Remove Xilinx FP
13378 (fusion_addis_mem_combo_load): Ditto.
13379 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove Xilinx
13381 (rs6000_cpu_cpp_builtins): Ditto.
13382 * config/rs6000/rs6000-linux.c
13383 (rs6000_linux_float_exceptions_rounding_supported_p): Ditto.
13384 * config/rs6000/rs6000-opts.h (enum fpu_type_t): Delete.
13385 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove Xilinx FP
13387 (rs6000_setup_reg_addr_masks): Ditto.
13388 (rs6000_init_hard_regno_mode_ok): Ditto.
13389 (rs6000_option_override_internal): Ditto.
13390 (legitimate_lo_sum_address_p): Ditto.
13391 (rs6000_legitimize_address): Ditto.
13392 (rs6000_legitimize_reload_address): Ditto.
13393 (rs6000_legitimate_address_p): Ditto.
13394 (abi_v4_pass_in_fpr): Ditto.
13395 (setup_incoming_varargs): Ditto.
13396 (rs6000_gimplify_va_arg): Ditto.
13397 (rs6000_split_multireg_move): Ditto.
13398 (rs6000_savres_strategy): Ditto.
13399 (rs6000_emit_prologue_components): Ditto.
13400 (rs6000_emit_epilogue_components): Ditto.
13401 (rs6000_emit_prologue): Ditto.
13402 (rs6000_emit_epilogue): Ditto.
13403 (rs6000_elf_file_end): Ditto.
13404 (rs6000_function_value): Ditto.
13405 (rs6000_libcall_value): Ditto.
13406 * config/rs6000/rs6000.h: Ditto.
13407 (TARGET_MINMAX_SF, TARGET_MINMAX_DF): Delete, merge to ...
13408 (TARGET_MINMAX): ... this. New.
13409 (TARGET_SF_FPR, TARGET_DF_FPR, TARGET_SF_INSN, TARGET_DF_INSN): Delete.
13410 * config/rs6000/rs6000.md: Remove Xilinx FP support.
13411 (*movsi_internal1_single): Delete.
13412 * config/rs6000/rs6000.opt (msingle-float, mdouble-float, msimple-fpu,
13413 mfpu=, mxilinx-fpu): Delete.
13414 * config/rs6000/singlefp.h: Delete.
13415 * config/rs6000/sysv4.h: Remove Xilinx FP support.
13416 * config/rs6000/t-rs6000: Ditto.
13417 * config/rs6000/t-xilinx: Delete.
13418 * config/rs6000/titan.md: Adjust for fp_type removal.
13419 * config/rs6000/vsx.md: Remove Xilinx FP support.
13420 (VStype_simple): Delete.
13421 (VSfptype_simple, VSfptype_mul, VSfptype_div, VSfptype_sqrt): Delete.
13422 * config/rs6000/xfpu.h: Delete.
13423 * config/rs6000/xfpu.md: Delete.
13424 * config/rs6000/xilinx.h: Delete.
13425 * config/rs6000/xilinx.opt: Delete.
13426 * doc/invoke.texi (RS/6000 and PowerPC Options): Remove
13427 -msingle-float, -mdouble-float, -msimple-fpu, -mfpu=, and -mxilinx-fpu.
13429 2018-05-04 Tom de Vries <tom@codesourcery.com>
13432 * builtins.c (expand_builtin_goacc_parlevel_id_size): Handle null target
13435 2018-05-04 Richard Biener <rguenther@suse.de>
13437 PR middle-end/85627
13438 * tree-complex.c (update_complex_assignment): We are always in SSA form.
13439 (expand_complex_div_wide): Likewise.
13440 (expand_complex_operations_1): Likewise.
13441 (expand_complex_libcall): Preserve EH info of the original stmt.
13442 (tree_lower_complex): Handle removed blocks.
13443 * tree.c (build_common_builtin_nodes): Do not set ECF_NOTRHOW
13444 on complex multiplication and division libcall builtins.
13446 2018-05-04 Richard Biener <rguenther@suse.de>
13448 PR middle-end/85574
13449 * fold-const.c (negate_expr_p): Restrict negation of operand
13450 zero of a division to when we know that can happen without
13452 (fold_negate_expr_1): Likewise.
13454 2018-05-04 Jakub Jelinek <jakub@redhat.com>
13457 * real.h (real_nextafter): Declare.
13458 * real.c (real_nextafter): New function.
13459 * fold-const-call.c (fold_const_nextafter): New function.
13460 (fold_const_call_sss): Call it for CASE_CFN_NEXTAFTER and
13461 CASE_CFN_NEXTTOWARD.
13462 (fold_const_call_1): For CASE_CFN_NEXTTOWARD call fold_const_call_sss
13463 even when arg1_mode is different from arg0_mode.
13465 2018-05-03 Nathan Sidwell <nathan@acm.org>
13467 * doc/extend.texi (Deprecated Features): Remove
13468 -ffriend-injection.
13469 (Backwards Compatibility): Likewise.
13470 * doc/invoke.texi (C++ Language Options): Likewise.
13471 (C++ Dialect Options): Likewise.
13473 2018-05-03 Jakub Jelinek <jakub@redhat.com>
13476 * config/i386/avx512fintrin.h (_mm512_mullox_epi64,
13477 _mm512_mask_mullox_epi64): New intrinsics.
13479 2018-05-03 Tom de Vries <tom@codesourcery.com>
13482 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
13483 dump files): Add offload-tree.
13485 2018-05-03 Richard Biener <rguenther@suse.de>
13487 PR tree-optimization/85615
13488 * tree-ssa-threadupdate.c (thread_block_1): Only allow exits
13489 to loops not nested in BBs loop father to avoid creating multi-entry
13492 2018-05-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13494 PR tree-optimization/70291
13495 * tree-complex.c (expand_complex_libcall): Add type, inplace_p
13496 arguments. Change return type to tree. Emit libcall as a new
13497 statement rather than replacing existing one when inplace_p is true.
13498 (expand_complex_multiplication_components): New function.
13499 (expand_complex_multiplication): Expand floating-point complex
13500 multiplication using the above.
13501 (expand_complex_division): Rename inner_type parameter to type.
13502 Update expand_complex_libcall call-site.
13503 (expand_complex_operations_1): Update expand_complex_multiplication
13504 and expand_complex_division call-sites.
13506 2018-05-02 Jakub Jelinek <jakub@redhat.com>
13509 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask,
13510 *ashl<dwi>3_doubleword_mask_1, *<shift_insn><dwi>3_doubleword_mask,
13511 *<shift_insn><dwi>3_doubleword_mask_1): In condition require that
13512 the highest significant bit of the shift count mask is clear. In
13513 check whether and[sq]i3 is needed verify that all significant bits
13514 of the shift count other than the highest are set.
13516 2018-05-02 Tom de Vries <tom@codesourcery.com>
13519 * builtins.def (DEF_GOACC_BUILTIN_ONLY): Define.
13520 * omp-builtins.def (BUILT_IN_GOACC_PARLEVEL_ID)
13521 (BUILT_IN_GOACC_PARLEVEL_SIZE): New builtin.
13522 * builtins.c (expand_builtin_goacc_parlevel_id_size): New function.
13523 (expand_builtin): Call expand_builtin_goacc_parlevel_id_size.
13524 * doc/extend.texi (Other Builtins): Add __builtin_goacc_parlevel_id and
13525 __builtin_goacc_parlevel_size.
13527 2018-05-02 Richard Biener <rguenther@suse.de>
13529 PR tree-optimization/85597
13530 * tree-vect-stmts.c (vectorizable_operation): For ternary SLP
13531 do not use split vect_get_vec_defs call but call vect_get_slp_defs
13534 2018-05-02 Tom de Vries <tom@codesourcery.com>
13537 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
13538 dump files): Add ltrans-tree.
13540 2018-05-02 Tom de Vries <tom@codesourcery.com>
13543 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
13544 dump files): Add wpa-ipa.
13546 2018-05-02 Segher Boessenkool <segher@kernel.crashing.org>
13548 * config.gcc (powerpc*-*-*): Remove paired.h. Unsupport the
13549 powerpc*-*-linux*paired* target.
13550 * config/rs6000/750cl.h: Delete.
13551 * config/rs6000/paired.h: Delete.
13552 * config/rs6000/paired.md: Delete.
13553 * config/rs6000/predicates.md (easy_vector_constant): Remove paired
13555 * config/rs6000/rs6000-builtin.def: Remove paired float support.
13556 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
13557 comment. Remove paired float support.
13558 * config/rs6000/rs6000-modes.def: Remove V2SF and V2SI.
13559 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete
13561 * config/rs6000/rs6000-protos.h (paired_expand_vector_init,
13562 paired_emit_vector_cond_expr, paired_expand_vector_move): Delete
13564 * config/rs6000/rs6000.c: Remove paired float support.
13565 (paired_expand_vector_init, paired_expand_vector_move,
13566 paired_emit_vector_compare, paired_emit_vector_cond_expr,
13567 (paired_expand_lv_builtin, paired_expand_stv_builtin,
13568 paired_expand_builtin, paired_expand_predicate_builtin,
13569 paired_init_builtins): Delete.
13570 * config/rs6000/rs6000.h: Remove paired float support.
13571 * config/rs6000/rs6000.md: Remove paired float support.
13572 (move_from_CR_ov_bit): Delete.
13573 * config/rs6000/rs6000.opt (mpaired): Delete.
13574 * config/rs6000/t-rs6000: Remove paired.md from MD_INCLUDES.
13575 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mpaired.
13577 2018-05-02 Richard Biener <rguenther@suse.de>
13579 PR middle-end/85567
13580 * gimplify.c (gimplify_save_expr): When in SSA form allow
13581 SAVE_EXPRs to compute to SSA vars.
13583 2018-05-02 Jakub Jelinek <jakub@redhat.com>
13586 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask,
13587 *ashl<dwi>3_doubleword_mask_1, *<shift_insn><dwi>3_doubleword_mask,
13588 *<shift_insn><dwi>3_doubleword_mask_1): If and[sq]i3 is needed, don't
13589 clobber operands[2], instead use a new pseudo. Formatting fixes.
13591 2018-05-02 Richard Sandiford <richard.sandiford@linaro.org>
13593 PR tree-optimization/85586
13594 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Only
13595 exit early for statements in the same group if the accesses are
13598 2018-05-02 Tom de Vries <tom@codesourcery.com>
13601 * lto-wrapper.c (compile_offload_image): Add "could not find mkoffload"
13604 2018-05-01 Marc Glisse <marc.glisse@inria.fr>
13606 PR tree-optimization/85143
13607 * match.pd (A<B&A<C): Extend to BIT_IOR_EXPR.
13609 2018-05-01 Tom de Vries <tom@codesourcery.com>
13612 * config/nvptx/mkoffload.c (main): Suggest using -B in "offload compiler
13613 not found" error message.
13615 2018-05-01 Tom de Vries <tom@codesourcery.com>
13618 * vec.h (VEC_ORDERED_REMOVE_IF, VEC_ORDERED_REMOVE_IF_FROM_TO): Define.
13619 * vec.c (test_ordered_remove_if): New function.
13620 (vec_c_tests): Call test_ordered_remove_if.
13621 * dwarf2cfi.c (connect_traces): Use VEC_ORDERED_REMOVE_IF_FROM_TO.
13622 * lto-streamer-out.c (prune_offload_funcs): Use VEC_ORDERED_REMOVE_IF.
13623 * tree-vect-patterns.c (vect_pattern_recog_1): Use
13624 VEC_ORDERED_REMOVE_IF.
13626 2018-05-01 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
13628 PR tree-optimization/82665
13629 * vr-values.c (vr_values::extract_range_from_binary_expr): Handle
13630 pointer subtraction where arguments come from a memchr call.
13632 2018-05-01 Jakub Jelinek <jakub@redhat.com>
13634 * configure.ac (LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION): Use
13635 --push-state --as-needed and --pop-state instead of --as-needed and
13636 --no-as-needed if ld supports it.
13637 * configure: Regenerated.
13640 * doc/install.texi2html: Replace _002d with - and _002a with * in
13641 generated html files using sed.
13643 2018-04-30 David Malcolm <dmalcolm@redhat.com>
13646 * gcc-rich-location.c (blank_line_before_p): New function.
13647 (use_new_line): New function.
13648 (gcc_rich_location::add_fixit_insert_formatted): New function.
13649 * gcc-rich-location.h
13650 (gcc_rich_location::add_fixit_insert_formatted): New function.
13652 2018-04-30 David Malcolm <dmalcolm@redhat.com>
13654 * selftest.c (assert_streq): Rename "expected" and "actual" to
13655 "val1" and "val2". Extend NULL-handling to cover both inputs
13656 symmetrically, while still requiring both to be non-NULL for a pass.
13657 * selftest.h (assert_streq): Rename "expected" and "actual" to
13659 (ASSERT_EQ): Likewise.
13660 (ASSERT_EQ_AT): Likewise.
13661 (ASSERT_KNOWN_EQ): Likewise.
13662 (ASSERT_KNOWN_EQ_AT): Likewise.
13663 (ASSERT_NE): Likewise.
13664 (ASSERT_MAYBE_NE): Likewise.
13665 (ASSERT_MAYBE_NE_AT): Likewise.
13666 (ASSERT_STREQ): Likewise. Clarify that both must be non-NULL for
13667 the assertion to pass.
13668 (ASSERT_STREQ_AT): Likewise.
13670 2018-04-30 Jonathan Wakely <jwakely@redhat.com>
13672 * doc/invoke.texi (-fpie, -fPIE): Fix grammar and clarify
13673 interaction with -pie.
13675 2018-04-30 David Malcolm <dmalcolm@redhat.com>
13677 * selftest.h: Fix alphabetization of per-source-file selftest
13680 2018-04-30 Jason Merrill <jason@redhat.com>
13682 PR c++/61982 - dead stores to destroyed objects.
13683 * gimplify.c (gimplify_modify_expr): Simplify complex lvalue on LHS
13686 2018-04-30 Jason Merrill <jason@redhat.com>
13688 * tree.c (build_clobber): New.
13689 * tree.h: Declare it.
13690 * gimplify.c (gimplify_bind_expr, gimplify_target_expr): Use it.
13692 2018-04-30 David Malcolm <dmalcolm@redhat.com>
13694 * diagnostic-show-locus.c (layout::layout): Update for
13695 location_get_source_line returning a char_span.
13696 (struct char_span): Move to input.h.
13697 (struct correction): Update for fields in char_span becoming
13699 (struct source_line): Update for location_get_source_line
13700 returning a char_span.
13701 (layout::print_line): Likewise.
13702 * edit-context.c (edited_file::print_content): Likewise.
13703 (edited_file::print_diff_hunk): Likewise.
13704 (edited_file::print_run_of_changed_lines): Likewise.
13705 (edited_file::get_num_lines): Likewise.
13706 (edited_line::edited_line): Likewise.
13707 * final.c (asm_show_source): Likewise.
13708 * input.c (location_get_source_line): Convert return type
13709 from const char * to char_span, losing the final "line_len"
13711 (dump_location_info): Update for the above.
13712 (get_substring_ranges_for_loc): Likewise. Use a char_span
13713 when handling the literal within the line.
13714 (test_reading_source_line): Update for location_get_source_line
13715 returning a char_span.
13716 * input.h (class char_span): Move here from
13717 diagnostic-show-locus.c, converting from a struct to a class.
13718 Make data members private.
13719 (char_span::operator bool): New.
13720 (char_span::length): New.
13721 (char_span::get_buffer): New.
13722 (char_span::operator[]): New.
13723 (char_span::subspan): Make const.
13724 (char_span::xstrdup): New.
13725 (location_get_source_line): Convert return type from const char *
13726 to char_span, losing the final "line_size" param.
13728 2018-04-30 Jan Hubicka <jh@suse.cz>
13730 * lto-wrapper.c (ltrans_priorities): New static var.
13731 (cmp_priority): New.
13732 (run_gcc): Read priorities and if doing parallel build order
13733 the Makefile by them.
13735 2018-04-30 David Malcolm <dmalcolm@redhat.com>
13737 * input.h (builtins_location_check): Convert to a STATIC_ASSERT.
13739 2018-04-30 Richard Biener <rguenther@suse.de>
13741 * tree-cfg.c (verify_address): Remove base argument, add
13742 flag whether to check TREE_ADDRESSABLE and do that.
13743 (verify_expr): Remove.
13744 (verify_types_in_gimple_reference): Add pieces from verify_expr.
13745 (verify_gimple_assign_single): Likewise.
13746 (verify_gimple_switch): Likewise.
13747 (verify_expr_location_1): Dereference tp once. Add (disabled)
13748 piece from verify_expr.
13749 (verify_gimple_in_cfg): Do not call verify_expr on all ops.
13751 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
13753 * config/arc/linux.h (CLEAR_INSN_CACHE): Define.
13755 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
13757 * config/arc/arc-protos.h (prepare_extend_operands): Remove.
13758 (small_data_pattern): Likewise.
13759 (arc_rewrite_small_data): Likewise.
13760 * config/arc/arc.c (LEGITIMATE_SMALL_DATA_OFFSET_P): Remove.
13761 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Likewise.
13762 (get_symbol_alignment): New function.
13763 (legitimate_small_data_address_p): Likewise.
13764 (legitimate_scaled_address): Update, call
13765 legitimate_small_data_address_p.
13766 (output_sdata): New static variable.
13767 (arc_print_operand): Update how we handle small data operands.
13768 (arc_print_operand_address): Likewise.
13769 (arc_legitimate_address_p): Update, use
13770 legitimate_small_data_address_p.
13771 (arc_rewrite_small_data_p): Remove.
13772 (arc_rewrite_small_data_1): Likewise.
13773 (arc_rewrite_small_data): Likewise.
13774 (small_data_pattern): Likewise.
13775 (compact_sda_memory_operand): Update to use
13776 legitimate_small_data_address_p and get_symbol_alignment.
13777 (prepare_move_operands): Don't rewite sdata pattern.
13778 (prepare_extend_operands): Remove.
13779 * config/arc/arc.md (zero_extendqihi2): Don't rewrite sdata
13781 (zero_extendqisi2): Likewise.
13782 (zero_extendhisi2): Likewise.
13783 (extendqihi2): Likewise.
13784 (extendqisi2): Likewise.
13785 (extendhisi2): Likewise.
13786 (addsi3): Likewise.
13787 (subsi3): Likewise.
13788 (andsi3): Likewise.
13789 * config/arc/constraints.md (Usd): Change it to memory constraint.
13791 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
13793 * config/arc/arc.c (arc_split_move): Allow signed 6-bit constants
13794 as source of std instructions.
13795 * config/arc/arc.md (movsi_insn): Update pattern predicate to
13796 allow 6-bit constants as source for store instructions.
13797 (movdi_insn): Update instruction pattern to allow 6-bit constants
13798 as source for store instructions.
13800 2018-04-30 Jonathan Wakely <jwakely@redhat.com>
13802 * doc/invoke.texi (-fdebug-types-section): Fix grammar.
13804 2018-04-30 Nathan Sidwell <nathan@acm.org>
13805 Sandra Loosemore <sandra@codesourcery.com>
13807 * dumpfile.c (dump_open): Allow '-' for stdout.
13808 * doc/invoke.texi (Developer Options): Document dump filename
13809 determination early. Document stdin/stdout selection.
13811 2018-04-30 Andrew Sadek <andrew.sadek.se@gmail.com>
13813 Microblaze Target: PIC data text relative
13815 * config/microblaze/microblaze.opt: add new option -mpic-data-text-rel.
13816 * config/microblaze/microblaze-protos.h (microblaze_constant_address_p):
13818 * config/microblaze/microblaze.h (microblaze_constant_address_p):
13819 CONSTANT_ADDRESS_P definition to microblaze_constant_address_p.
13820 * config/microblaze/microblaze.c (TARGET_PIC_DATA_TEXT_REL):
13821 New addressing mode for data-text relative position indepenedent code.
13822 (microblaze_classify_unspec): add 'UNSPEC_TEXT' case ->
13823 'ADDRESS_SYMBOLIC_TXT_REL'.
13824 (microblaze_classify_address): Add handling for UNSPEC + CONST_INT.
13825 (microblaze_legitimate_pic_operand): Exclude function calls from
13826 pic operands in case of TARGET_PIC_DATA_TEXT_REL option.
13827 (microblaze_legitimize_address): Generate 'UNSPEC_TEXT' for all possible
13829 (microblaze_address_insns): Add 'ADDRESS_SYMBOLIC_TXT_REL' case.
13830 (print_operand): Add 'ADDRESS_SYMBOLIC_TXT_REL' case.
13831 (print_operand_address): Add 'ADDRESS_SYMBOLIC_TXT_REL' case + handling
13832 for 'address + offset'.
13833 (microblaze_expand_prologue): Add new function prologue call for
13835 (microblaze_asm_generate_pic_addr_dif_vec): Override new target hook
13836 'TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC' to disable address diff vector
13837 table in case of TARGET_PIC_DATA_TEXT_REL.
13838 (expand_pic_symbol_ref): Add handling for 'UNSPEC_TEXT'.
13839 * config/microblaze/microblaze.md (TARGET_PIC_DATA_TEXT_REL):
13840 Add new macros 'UNSPEC_TEXT',
13841 'UNSPEC_SET_TEXT' + add rule for setting r20 in function prologue
13842 + exclude function calls from 'UNSPEC_PLT' in case of data text
13844 * doc/tm.texi.in (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Add
13845 new target hook for generating address diff vector tables in case of
13847 * doc/tm.texi : Regenerate.
13848 * stmt.c (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Append new condition
13849 'targetm.asm_out.generate_pic_addr_diff_vec' to flag_pic in case
13850 of addr diff vector generation.
13851 * target.def (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Add
13852 target hook definition.
13853 * targhooks.h, targhooks.c (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC):
13854 Add default function for generate_pic_addr_diff_vec -> flag_pic.
13855 * doc/invoke.texi (Add new pic option): Add new microblaze pic
13856 option for data text relative.
13858 2018-04-30 Richard Biener <rguenther@suse.de>
13860 * tree-chrec.h (evolution_function_is_constant_p): Remove
13862 * tree-cfg.c (tree_node_can_be_shared): Re-order checks.
13864 2018-04-30 Richard Biener <rguenther@suse.de>
13867 * dwarf2out.c (gen_producer_string): Ignore -fchecking[=].
13869 2018-04-30 Richard Biener <rguenther@suse.de>
13871 PR tree-optimization/28364
13872 PR tree-optimization/85275
13873 * tree-ssa-loop-ch.c (ch_base::copy_headers): Stop after
13874 copying first exit test.
13876 2018-04-28 Mark Wielaard <mark@klomp.org>
13878 * dwarf2out.c (dwarf2out_finish): Add .debug_addr table header for
13879 dwarf_version >= 5.
13880 (dwarf_AT): Handle DW_AT_addr_base.
13881 (add_top_level_skeleton_die_attrs): Use dwarf_AT for DW_AT_addr_base.
13883 2018-04-28 Uros Bizjak <ubizjak@gmail.com>
13886 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask): New pattern.
13887 (*ashl<dwi>3_doubleword_mask_1): Ditto.
13888 (*<shift_insn><dwi>3_doubleword_mask): Ditto.
13889 (*<shift_insn><dwi>3_doubleword_mask_1): Ditto.
13891 2018-04-28 Richard Biener <rguenther@suse.de>
13893 * tree-cfg.c (verify_gimple_phi): Take a gphi * argument.
13894 (verify_gimple_in_cfg): Rename visited_stmts to visited_throwing_stmts
13895 to reflect use. Only add interesting stmts.
13897 2018-04-27 Martin Jambor <mjambor@suse.cz>
13900 * ipa-cp.c (find_aggregate_values_for_callers_subset): Make sure
13901 the jump function allows for passing through aggregate values.
13903 2018-04-27 David Malcolm <dmalcolm@redhat.com>
13905 * input.h (in_system_header_at): Convert from macro to inline
13907 (from_macro_expansion_at): Likewise.
13908 (from_macro_definition_at): Likewise.
13910 2018-04-27 Jeff Law <law@redhat.com>
13912 * config.gcc: Mark tile* targets as deprecated/obsolete.
13914 2018-04-27 Richard Biener <rguenther@suse.de>
13916 * config/aarch64/aarch64.c: Simplify ap.__stack advance and
13919 2018-04-27 Richard Biener <rguenther@suse.de>
13921 * tree-cfg.c (verify_expr): Make dead code hit gcc_unreachable.
13923 2018-04-27 Uros Bizjak <ubizjak@gmail.com>
13925 * config/i386/i386.md (*movti_internal): Substitute Ye constraint
13926 with Yd constraint. Set "preferred_for_speed" attribute from
13927 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC for alternatives
13928 with Yd constraint.
13929 (*movdi_internal): Ditto.
13930 (movti_interunit splitters): Remove
13931 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC from insn condition.
13932 (movdi_interunit splitters): Ditto.
13933 * config/i386/constraints.md (Ye): Remove.
13934 (Yd): Do not depend on TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC.
13936 2018-04-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13939 * config/aarch64/constraints.md (Usg): Limit to 31.
13940 (Usj): Limit to 63.
13942 2018-04-27 Jakub Jelinek <jakub@redhat.com>
13944 PR tree-optimization/85529
13945 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Add FIRST_BB
13946 argument. Don't call get_nonzero_bits if opcode is ERROR_MARK_NODE,
13947 rhs2 def stmt's bb is dominated by first_bb and it isn't an obvious
13948 zero extension or masking of the MSB bit.
13949 (optimize_range_tests): Add FIRST_BB argument, pass it through
13950 to optimize_range_tests_var_bound.
13951 (maybe_optimize_range_tests, reassociate_bb): Adjust
13952 optimize_range_tests callers.
13954 2018-04-26 Richard Biener <rguenther@suse.de>
13955 Jakub Jelinek <jakub@redhat.com>
13957 * cgraph.h (symbol_table): Just declare debug method here.
13958 * symtab.c (symbol_table::debug): Define.
13960 2018-04-26 Eric Botcazou <ebotcazou@adacore.com>
13962 * loop-invariant.c (may_assign_reg_p): Return false for frame pointer.
13964 2018-04-26 Uros Bizjak <ubizjak@gmail.com>
13966 * config/i386/i386.md ("isa" attribute): Add x64_sse2.
13967 ("enabled" attribute): Handle x64_sse2 "isa" attribute.
13968 (*movdi_internal): Substitute Yi and Yj constraint with x
13969 and Ym and Yn constraint with y constraint. Update "isa"
13970 attribute and set "preferred_for_speed" attribute from
13971 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC for updated alternatives.
13972 (*movsi_internal): Ditto.
13973 (*movdf_internal): Ditto.
13974 (*movsf_internal): Ditto.
13975 (*zero_extendsidi2): Ditto.
13976 * config/i386/sse.md (vec_set<mode>_0): Ditto.
13977 (sse2_loadld): Ditto.
13978 (*vec_extract<ssevecmodelower>_0): Ditto.
13979 (*vec_extractv4si_0_zext_sse4): Ditto.
13980 (vec_concatv2di): Ditto.
13981 (*vec_dup<mode>): Ditto.
13982 * config/i386/mmx.md (*mov<mode>_internal): Ditto.
13983 * config/i386/constraints.md (Yi): Remove.
13988 2018-04-26 Nathan Sidwell <nathan@acm.org>
13990 * dumpfile.c (dump_open): New.
13991 (dump_open_alternate_stream, dump_start, dump_begin): Call it.
13992 (dump_finish): Detect stdio/stderr by value not name.
13994 2018-04-26 Jonathan Wakely <jwakely@redhat.com>
13996 * doc/invoke.texi (-Wreturn-type): Document default status for C++.
13998 2018-04-26 Tom de Vries <tom@codesourcery.com>
14001 * config/nvptx/nvptx.c (verify_neutering_jumps)
14002 (verify_neutering_labels): New function
14003 (nvptx_single): Use verify_neutering_jumps and verify_neutering_labels.
14005 2018-04-26 Tom de Vries <tom@codesourcery.com>
14008 * config/nvptx/nvptx.c (needs_neutering_p): New function.
14009 (nvptx_single): Use needs_neutering_p to skip over insns that do not
14012 2018-04-26 Richard Biener <rguenther@suse.de>
14013 Tom de Vries <tom@codesourcery.com>
14016 * lto-streamer-out.c (output_function): Fixup loops if required to match
14017 discovery done in the reader.
14019 2018-04-26 Richard Biener <rguenther@suse.de>
14021 PR tree-optimization/85116
14022 * tree-ssa-loop-ch.c (do_while_loop_p): A do-while loop should
14023 have a loop exit from the single latch predecessor. Remove
14024 case of header with just condition.
14025 (ch_base::copy_headers): Exclude infinite loops from any
14027 (pass_ch::execute): Record exits.
14029 2018-04-26 Richard Biener <rguenther@suse.de>
14031 * tree-vect-data-refs.c (vect_get_data_access_cost): Get
14032 prologue cost vector and pass it to vect_get_load_cost.
14033 (vect_get_peeling_costs_all_drs): Likewise.
14034 (vect_peeling_hash_get_lowest_cost): Likewise.
14035 (vect_enhance_data_refs_alignment): Likewise.
14037 2018-04-26 Richard Biener <rguenther@suse.de>
14039 PR middle-end/85450
14040 * tree-cfg.c (verify_gimple_assign_unary): Restore proper
14041 checking of integer<->pointer conversions.
14042 * omp-expand.c (expand_omp_for_static_nochunk): Avoid
14043 sign-/zero-extending pointer types.
14044 (expand_omp_for_static_chunk): Likewise.
14046 2018-03-22 Hans-Peter Nilsson <hp@axis.com>
14047 Jean Lee <xiaoyur347@gmail.com>
14049 * config/mips/mips.c (mips_asan_shadow_offset): New function.
14050 (TARGET_ASAN_SHADOW_OFFSET): Define.
14051 * config/mips/mips.h (FRAME_GROWS_DOWNWARD): Augment to also be
14052 true for -fsanitize=address.
14054 2018-04-25 Mark Wielaard <mark@klomp.org>
14056 * dwarf2out.c (file_info_cmp): Sort longer dir prefixes before
14059 2018-04-25 Jakub Jelinek <jakub@redhat.com>
14061 * config/i386/i386.md (*x86_mov<mode>cc_0_m1): Use type "alu1" rather
14062 than "alu", remove explicit "memory" and "imm_disp" attributes.
14063 (*x86_mov<mode>cc_0_m1_se, *x86_mov<mode>cc_0_m1_neg): Likewise.
14065 PR middle-end/85414
14066 * simplify-rtx.c (simplify_unary_operation_1) <case SIGN_EXTEND,
14067 case ZERO_EXTEND>: Pass SUBREG_REG (op) rather than op to
14068 gen_lowpart_no_emit.
14070 2018-04-25 Sebastian Peryt <sebastian.peryt@intel.com>
14073 * config/i386/i386.c (ix86_expand_builtin): Change memory
14074 operand to XI, extend p0 to Pmode.
14075 * config/i386/i386.md: Change unspec volatile and operand
14076 1 mode to XI, change operand 0 mode to P.
14078 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
14080 * config/nds32/nds32-predicates.c (nds32_can_use_bclr_p): Mask with
14081 GET_MODE_MASK before any checking.
14082 (nds32_can_use_bset_p): Likewise.
14083 (nds32_can_use_btgl_p): Likewise.
14085 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
14087 * config/nds32/nds32-doubleword.md: New define_split pattern for
14088 illegal register number.
14090 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
14092 * config/nds32/nds32.c (nds32_print_operand): Set op_value ealier.
14094 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
14096 * config/nds32/nds32.h (ASM_APP_ON): Add missing newline character.
14098 2018-04-25 Richard Biener <rguenther@suse.de>
14100 * lto-streamer.h (LTO_major_version): Bump to 8.
14102 2018-04-25 Jakub Jelinek <jakub@redhat.com>
14104 * BASE-VER: Set to 9.0.0.
14106 2018-04-24 Segher Boessenkool <segher@kernel.crashing.org>
14108 * config/rs6000/rs6000.c (init_float128_ieee): Fix spelling mistakes
14109 in __abskf2 and __powikf2.
14111 2018-04-24 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14114 * config/aarch64/constraints.md (Usg, Usj): New constraints.
14115 * config/aarch64/iterators.md (cmode_simd): New mode attribute.
14116 * config/aarch64/aarch64.md (*aarch64_ashr_sisd_or_int_<mode>3):
14117 Use the above on operand 2. Reindent.
14118 (*aarch64_lshr_sisd_or_int_<mode>3): Likewise.
14120 2018-04-24 H.J. Lu <hongjiu.lu@intel.com>
14123 * common/config/i386/i386-common.c (ix86_handle_option): Don't
14125 * config/i386/i386.opt (mcet): Removed.
14126 * doc/install.texi: Remove -mcet documentation.
14127 * doc/invoke.texi: Likewise.
14129 2018-04-24 H.J. Lu <hongjiu.lu@intel.com>
14132 * doc/install.texi: Remove -mcet from bootstrap-cet.
14134 2018-04-24 Jakub Jelinek <jakub@redhat.com>
14137 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Don't define
14138 __builtin_ia32_readeflags_u32 and __builtin_ia32_writeeflags_u32
14142 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Punt if
14143 const_vector is not CONST_VECTOR or SYMBOL_REF for a constant pool
14144 containing a CONST_VECTOR.
14146 2018-04-24 Cesar Philippidis <cesar@codesourcery.com>
14148 * doc/install.texi: Update newlib dependency for nvptx.
14150 2018-04-24 Jakub Jelinek <jakub@redhat.com>
14153 * config/i386/i386.c (ix86_expand_vector_init_one_var): Use UINTVAL
14154 instead of INTVAL when shifting x left.
14156 2018-04-24 Andreas Krebbel <krebbel@linux.ibm.com>
14158 PR tree-optimization/85478
14159 * tree-vect-loop.c (vect_analyze_loop_2): Do not call
14160 vect_grouped_store_supported for single element vectors.
14162 2018-04-24 Richard Biener <rguenther@suse.de>
14165 * config/i386/i386.c (ix86_add_stmt_cost): Restrict strided
14166 load cost increase to the case of non-constant step.
14168 2018-04-24 Jakub Jelinek <jakub@redhat.com>
14171 * reg-stack.c (move_for_stack_reg): Don't fail assertion about dead
14172 destination if any_malformed_asm.
14174 2018-04-23 Eric Botcazou <ebotcazou@adacore.com>
14176 PR middle-end/85496
14177 * expr.c (store_field): In the bitfield case, if the value comes from
14178 a function call and is returned in registers by means of a PARALLEL,
14179 do not change the mode of the temporary unless BLKmode and VOIDmode.
14181 2018-04-23 Andrey Belevantsev <abel@ispras.ru>
14183 PR rtl-optimization/85423
14184 * sel-sched-ir.c (has_dependence_note_mem_dep): Only discard
14185 dependencies to debug insns when the previous insn is non-debug.
14187 2018-04-23 Claudiu Zissulescu <claziss@synopsys.com>
14189 * config/arc/arc.md ("vunspec"): Delete it, unify all the unspec
14190 enums into a single definition.
14191 (fls): Fix predicates and printing.
14194 2018-04-23 Claudiu Zissulescu <claziss@synopsys.com>
14196 * config/arc/arc-protos.h (check_if_valid_sleep_operand): Remove.
14197 * config/arc/arc.c (arc_expand_builtin): Sleep accepts registers
14198 and short u6 immediate.
14199 (check_if_valid_sleep_operand): Remove.
14200 * config/arc/arc.md (Sleep): Accepts registers and u6 immediates.
14202 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
14204 * config/nds32/nds32.c (nds32_compute_stack_frame): Consider
14205 flag_always_save_lp condition.
14206 * config/nds32/nds32.opt (malways-save-lp): New option.
14208 2018-04-22 Shiva Chen <shiva0217@gmail.com>
14210 * config/nds32/nds32-protos.h (nds32_use_load_post_increment): Declare.
14211 * config/nds32/nds32.c (nds32_use_load_post_increment): New.
14212 * config/nds32/nds32.h
14213 (USE_LOAD_POST_INCREMENT, USE_LOAD_POST_DECREMENT): Define.
14214 (USE_STORE_POST_INCREMENT, USE_STORE_POST_DECREMENT): Define.
14216 2018-04-22 Shiva Chen <shiva0217@gmail.com>
14218 * config/nds32/nds32-protos.h (nds32_ls_333_p): Remove.
14219 * config/nds32/nds32.c (nds32_ls_333_p): Remove.
14221 2018-04-22 Shiva Chen <shiva0217@gmail.com>
14222 Chung-Ju Wu <jasonwucj@gmail.com>
14224 * config/nds32/nds32-protos.h (nds32_case_vector_shorten_mode):
14226 * config/nds32/nds32.c (nds32_case_vector_shorten_mode): New function.
14227 * config/nds32/nds32.h (CASE_VECTOR_SHORTEN_MODE): Modify.
14229 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
14231 * config/nds32/nds32.c (nds32_compute_stack_frame): Fix wrong value.
14233 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
14235 * config/nds32/nds32-protos.h (nds32_data_alignment,
14236 nds32_local_alignment): Declare.
14237 * config/nds32/nds32.c (nds32_data_alignment, nds32_constant_alignment,
14238 nds32_local_alignment): New functions.
14239 (TARGET_CONSTANT_ALIGNMENT): Define.
14240 * config/nds32/nds32.h (DATA_ALIGNMENT, LOCAL_ALIGNMENT): Define.
14242 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
14244 * config/nds32/nds32.c
14245 (TARGET_HARD_REGNO_MODE_OK): Move to the bottom of file.
14246 (TARGET_MODES_TIEABLE_P): Likewise.
14248 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
14250 * config/nds32/nds32.c (nds32_asm_file_start): Display optimization
14251 level Ofast and Og.
14253 2018-04-22 Monk Chiang <sh.chiang04@gmail.com>
14254 Chung-Ju Wu <jasonwucj@gmail.com>
14256 * config/nds32/constants.md (unspec_volatile_element): Add enum values
14257 for unaligned access.
14258 * config/nds32/nds32-intrinsic.c: Implementation of expanding
14260 * config/nds32/nds32-intrinsic.md: Likewise.
14261 * config/nds32/nds32_intrinsic.h: Likewise.
14262 * config/nds32/nds32.h (nds32_builtins): Likewise.
14263 * config/nds32/nds32.opt (munaligned-access): New option.
14264 * config/nds32/nds32.c (nds32_asm_file_start): Display
14265 flag_unaligned_access status.
14267 2018-04-20 Kito Cheng <kito.cheng@gmail.com>
14269 * config/riscv/elf.h (LINK_SPEC): Pass --no-relax if
14270 -mno-relax is present.
14271 * config/riscv/linux.h (LINK_SPEC): Ditto.
14273 2018-04-20 Martin Sebor <msebor@redhat.com>
14276 * gimple-fold.c (gimple_fold_builtin_strcpy): Suppress -Wrestrict
14278 (gimple_fold_builtin_stxcpy_chk): Same.
14279 * gimple-ssa-warn-restrict.c (check_bounds_or_overlap): Same.
14281 2018-04-20 Michael Meissner <meissner@linux.ibm.com>
14284 * config/rs6000/rs6000.c (init_float128_ieee): Add support to call
14285 __powikf2 when long double is IEEE 128-bit.
14287 2018-04-20 Kito Cheng <kito.cheng@gmail.com>
14289 * config/riscv/riscv.c (riscv_first_stack_step): Round up min
14290 step to make sure stack always aligned.
14292 2018-04-20 Carl Love <cel@us.ibm.com>
14295 * config/rs6000/rs6000-c.c (rs6000_gimple_fold_builtin): Add
14296 size check for arg0.
14298 2018-04-20 Nathan Sidwell <nathan@codesourcery.com>
14299 Tom de Vries <tom@codesourcery.com>
14302 * config/nvptx/nvptx.c (nvptx_emit_forking, nvptx_emit_joining):
14303 Emit insns for calls too.
14304 (nvptx_find_par): Always look for worker-level predecessor insn.
14305 (nvptx_propagate): Add is_call parm, return bool. Copy frame for
14307 (nvptx_vpropagate, nvptx_wpropagate): Adjust.
14308 (nvptx_process_pars): Propagate frames for calls.
14310 2018-04-20 H.J. Lu <hongjiu.lu@intel.com>
14313 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET):
14315 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
14316 (ix86_handle_option): Don't handle OPT_mibt.
14317 * config/i386/cet.h: Check __CET__ instead of __IBT__ and
14319 * config/i386/driver-i386.c (host_detect_local_cpu): Remove
14321 * config/i386/i386-c.c (ix86_target_macros_internal): Don't
14322 check OPTION_MASK_ISA_IBT nor flag_cf_protection.
14323 (ix86_target_macros): Define __CET__ with flag_cf_protection
14324 for -fcf-protection.
14325 * config/i386/i386.c (isa2_opts): Remove -mibt.
14326 * config/i386/i386.h (TARGET_IBT): Removed.
14327 (TARGET_IBT_P): Likewise.
14328 (ix86_valid_target_attribute_inner_p): Don't check OPT_mibt.
14329 * config/i386/i386.md (nop_endbr): Don't check TARGET_IBT.
14330 * config/i386/i386.opt (mcet): Update help message.
14331 (mshstk): Likewise.
14333 * doc/invoke.texi: Remove -mibt. Document __CET__. Document
14334 -mcet as an alias for -mshstk.
14336 2018-04-20 Richard Biener <rguenther@suse.de>
14338 PR middle-end/85475
14339 * match.pd ((X * CST) * Y -> (X * Y) * CST): Avoid exponential
14340 complexity by forcing a single use of the multiply operand.
14342 2018-04-20 Martin Jambor <mjambor@suse.cz>
14345 * ipa-cp.c (cgraph_edge_brings_value_p): Move check for self-feeding
14346 recursion dependency to only apply to non-clones.
14348 2018-04-20 Martin Jambor <mjambor@suse.cz>
14351 * ipa-cp.c (create_specialized_node): Check that clones of
14352 self-recursive edges exist during IPA-CP.
14354 2018-04-19 Toon Moene <toon@moene.org>
14356 * doc/invoke.texi: Add -floop-unroll-and-jam to options enabled
14359 2018-04-19 Jakub Jelinek <jakub@redhat.com>
14361 PR tree-optimization/85467
14362 * fold-const.c (fold_ternary_loc) <case BIT_FIELD_REF>: Use
14363 VECTOR_TYPE_P macro. If type is vector type, VIEW_CONVERT_EXPR the
14364 VECTOR_CST element to type.
14366 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
14369 * config/i386/i386.h (STACK_SAVEAREA_MODE): New.
14370 * config/i386/i386.md (builtin_setjmp_setup): Removed.
14371 (builtin_longjmp): Likewise.
14372 (save_stack_nonlocal): New pattern.
14373 (restore_stack_nonlocal): Likewise.
14375 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
14378 * config/i386/cet.c (file_end_indicate_exec_stack_and_cet):
14379 Replace ASM_OUTPUT_LABEL with fprintf.
14381 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
14384 * config/i386/cet.c (file_end_indicate_exec_stack_and_cet):
14385 Check flag_cf_protection instead of TARGET_IBT and TARGET_SHSTK.
14386 * config/i386/i386-c.c (ix86_target_macros_internal): Also
14387 define __IBT__ and __SHSTK__ for -fcf-protection.
14388 * config/i386/i386.c (pass_insert_endbranch::gate): Don't check
14390 (ix86_trampoline_init): Likewise.
14391 (x86_output_mi_thunk): Likewise.
14392 (ix86_notrack_prefixed_insn_p): Likewise.
14393 (ix86_option_override_internal): Don't disallow -fcf-protection.
14394 * config/i386/i386.md (rdssp<mode>): Also enable for
14396 (incssp<mode>): Likewise.
14397 (nop_endbr): Likewise.
14398 * config/i386/i386.opt (mcet): Change help message to built-in
14401 (mshstk): Likewise.
14402 * doc/invoke.texi: Remove -mcet, -mibt and -mshstk condition
14403 on -fcf-protection. Change -mcet, -mibt and -mshstk to only
14404 enable CET built-in functions.
14406 2018-04-19 Sebastian Peryt <sebastian.peryt@intel.com>
14408 * common/config/i386/i386-common.c
14409 (OPTION_MASK_ISA_MOVDIRI_SET, OPTION_MASK_ISA_MOVDIR64B_SET,
14410 OPTION_MASK_ISA_MOVDIRI_UNSET,
14411 OPTION_MASK_ISA_MOVDIR64B_UNSET): New defines.
14412 (ix86_handle_option): Handle -mmovdiri and -mmovdir64b.
14413 * config.gcc (movdirintrin.h): New header.
14414 * config/i386/cpuid.h (bit_MOVDIRI,
14415 bit_MOVDIR64B): New bits.
14416 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mmovdiri
14418 * config/i386/i386-builtin-types.def ((VOID, PUNSIGNED, UNSIGNED),
14419 (VOID, PVOID, PCVOID)): New function types.
14420 * config/i386/i386-builtin.def (__builtin_ia32_directstoreu_u32,
14421 __builtin_ia32_directstoreu_u64,
14422 __builtin_ia32_movdir64b): New builtins.
14423 * config/i386/i386-c.c (__MOVDIRI__, __MOVDIR64B__): New.
14424 * config/i386/i386.c (ix86_target_string): Added -mmovdir64b
14426 (ix86_valid_target_attribute_inner_p): Ditto.
14427 (ix86_expand_special_args_builtin): Added VOID_FTYPE_PUNSIGNED_UNSIGNED
14428 and VOID_FTYPE_PUNSIGNED_UNSIGNED.
14429 (ix86_expand_builtin): Expand IX86_BUILTIN_MOVDIR64B.
14430 * config/i386/i386.h (TARGET_MOVDIRI, TARGET_MOVDIRI_P,
14431 TARGET_MOVDIR64B, TARGET_MOVDIR64B_P): New.
14432 * config/i386/i386.md (UNSPECV_MOVDIRI, UNSPECV_MOVDIR64B): New.
14433 (movdiri<mode>, movdir64b_<mode>): New.
14434 * config/i386/i386.opt: Add -mmovdiri and -mmovdir64b.
14435 * config/i386/immintrin.h: Include movdirintrin.h.
14436 * config/i386/movdirintrin.h: New file.
14437 * doc/invoke.texi: Added -mmovdiri and -mmovdir64b.
14439 2018-04-19 Richard Biener <rguenther@suse.de>
14441 PR middle-end/85455
14442 * cfg.c (clear_bb_flags): When loop state says we have
14443 marked irreducible regions also preserve BB_IRREDUCIBLE_LOOP.
14445 2018-04-19 Richard Biener <rguenther@suse.de>
14447 PR tree-optimization/84737
14448 * tree-vect-data-refs.c (vect_copy_ref_info): New function
14449 copying restrict info.
14450 (vect_setup_realignment): Use it.
14451 * tree-vectorizer.h (vect_copy_ref_info): Declare.
14452 * tree-vect-stmts.c (vectorizable_store): Copy ref info from
14453 the first DR to all generated stores.
14454 (vectorizable_load): Likewise for loads.
14456 2018-04-19 Jakub Jelinek <jakub@redhat.com>
14458 PR tree-optimization/85446
14459 * match.pd ((intptr_t) x eq/ne CST to x eq/ne (typeof x) cst): Require
14460 the integral and pointer types to have the same precision.
14462 * doc/install.texi: Document --disable-cet being the default and
14465 2018-04-18 Martin Liska <mliska@suse.cz>
14467 * ipa-devirt.c (odr_subtypes_equivalent_p): Fix GNU coding
14470 2018-04-18 Martin Liska <mliska@suse.cz>
14473 2018-03-02 Eric Botcazou <ebotcazou@adacore.com>
14476 * ipa-devirt.c (odr_subtypes_equivalent_p): Get the ODR type of both
14477 arguments if they are comparable.
14479 2018-04-18 Martin Liska <mliska@suse.cz>
14482 2018-03-13 Eric Botcazou <ebotcazou@adacore.com>
14485 * ipa-devirt.c (odr_subtypes_equivalent_p): Do not get the ODR type of
14488 2018-04-18 H.J. Lu <hongjiu.lu@intel.com>
14491 * config/i386/i386.c (ix86_expand_split_stack_prologue): Insert
14492 ENDBR after calling __morestack.
14494 2018-04-18 David Malcolm <dmalcolm@redhat.com>
14497 * configure.ac (gcc-driver-name.h): Honor --with-gcc-major-version
14498 by using gcc_base_ver to generate a gcc_driver_version, and use
14499 it when generating GCC_DRIVER_NAME.
14500 * configure: Regenerate.
14502 2018-04-18 Jakub Jelinek <jakub@redhat.com>
14505 * config.gcc: Obsolete powerpc*-*-*spe*.
14507 2018-04-17 Jakub Jelinek <jakub@redhat.com>
14510 * dbxout.c (dbxout_int): Perform negation in unsigned int type.
14511 (stabstr_D): Change type of unum from unsigned int to
14512 unsigned HOST_WIDE_INT. Perform negation in unsigned HOST_WIDE_INT
14515 2018-04-17 Jim Wilson <jimw@sifive.com>
14518 * config/riscv/riscv.c (riscv_compute_frame_info): Add calls to
14519 RISCV_STACK_ALIGN when using outgoing_args_size and pretend_args_size.
14520 Set arg_pointer_offset after using pretend_args_size.
14522 2018-04-17 Jakub Jelinek <jakub@redhat.com>
14524 PR rtl-optimization/85431
14525 * dse.c (record_store): Ignore zero width stores.
14528 * asan.c (handle_builtin_stack_restore): Adjust comment. Emit
14529 __asan_allocas_unpoison call and last_alloca_addr = new_sp before
14530 __builtin_stack_restore rather than after it.
14531 * builtins.c (expand_asan_emit_allocas_unpoison): Pass
14532 arg1 + (virtual_dynamic_stack_rtx - stack_pointer_rtx) as second
14533 argument instead of virtual_dynamic_stack_rtx.
14535 2018-04-17 Kelvin Nilsen <kelvin@gcc.gnu.org>
14537 * config/rs6000/rs6000-protos.h (rs6000_builtin_is_supported_p):
14539 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
14540 Add note to error message to explain internal mapping of overloaded
14541 built-in function name to non-overloaded built-in function name.
14542 * config/rs6000/rs6000.c (rs6000_builtin_is_supported_p): New
14545 2018-04-17 Michael Meissner <meissner@linux.vnet.ibm.com>
14548 * config/rs6000/rs6000.md (pack<mode>): Do not try handle a pack
14549 where the inputs overlap with the output.
14551 2018-04-17 Jakub Jelinek <jakub@redhat.com>
14553 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): Add
14554 (=v, v) alternative and explicit "memory" attribute.
14555 (vec_extract_lo_<mode><mask_name>): Likewise. Also add
14556 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
14558 (vec_extract_lo_<mode><mask_name>): Add (=v, v) alternative and use
14559 "sselog1" type instead of "sselog".
14560 (vec_extract_hi_<mode><mask_name>): Use "sselog1" type instead of
14561 "sselog". Remove explicit "memory" attribute.
14562 (vec_extract_lo_v32hi): Add (=v, v) alternative and explicit "memory",
14563 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
14565 (vec_extract_hi_v32hi): Merge all alternatives into one, use
14566 "sselog1" type instead of "sselog". Remove explicit "memory"
14568 (vec_extract_hi_v16hi): Merge each pair of alternatives into one,
14569 use "sselog1" type instead of "sselog". Remove explicit "memory"
14571 (vec_extract_lo_v64qi): Add (=v, v) alternative and explicit "memory",
14572 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
14574 (vec_extract_hi_v64qi): Merge all alternatives into one, use
14575 "sselog1" type instead of "sselog". Remove explicit "memory"
14577 (vec_extract_hi_v32qi): Merge each pair of alternatives into one,
14578 use "sselog1" type instead of "sselog". Remove explicit "memory"
14582 * config/i386/i386.md (*ashlqi3_1_slp): Use alu1 type instead of alu.
14584 PR middle-end/85414
14585 * rtlhooks.c (gen_lowpart_if_possible): Don't call gen_lowpart_SUBREG
14588 2018-04-17 Martin Jambor <mjambor@suse.cz>
14591 * ipa-cp.c (create_specialized_node): Call
14592 expand_all_artificial_thunks if necessary.
14594 2018-04-17 Martin Liska <mliska@suse.cz>
14597 * ipa-devirt.c (odr_types_equivalent_p): Remove trailing
14598 in message, remote space in between '_G' and '('.
14600 2018-04-17 Jakub Jelinek <jakub@redhat.com>
14603 * config/i386/sse.md (reduces<mode><mask_scalar_name>,
14604 avx512f_vmcmp<mode>3<round_saeonly_name>,
14605 avx512f_vmcmp<mode>3_mask<round_saeonly_name>,
14606 avx512f_sgetexp<mode><mask_scalar_name><round_saeonly_scalar_name>,
14607 avx512f_rndscale<mode><round_saeonly_name>,
14608 avx512dq_ranges<mode><mask_scalar_name><round_saeonly_scalar_name>,
14609 avx512f_vgetmant<mode><mask_scalar_name><round_saeonly_scalar_name>):
14610 Use %<iptr>2 instead of %2 for -masm=intel.
14611 (avx512f_vcvtss2usi<round_name>, avx512f_vcvtss2usiq<round_name>,
14612 avx512f_vcvttss2usi<round_saeonly_name>,
14613 avx512f_vcvttss2usiq<round_saeonly_name>): Use %k1 instead of %1 for
14615 (avx512f_vcvtsd2usi<round_name>, avx512f_vcvtsd2usiq<round_name>,
14616 avx512f_vcvttsd2usi<round_saeonly_name>,
14617 avx512f_vcvttsd2usiq<round_saeonly_name>, ufloatv2siv2df2<mask_name>):
14618 Use %q1 instead of %1 for -masm=intel.
14619 (avx512f_sfixupimm<mode><sd_maskz_name><round_saeonly_name>,
14620 avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Use %<iptr>3 instead
14621 of %3 for -masm=intel.
14622 (sse2_shufpd_v2df_mask): Fix a typo, change %{6%} to %{%6%} for
14624 (*avx512vl_<code>v2div2qi2_store): Use %w0 instead of %0 for
14626 (*avx512vl_<code><mode>v4qi2_store): Use %k0 instead of %0 for
14628 (avx512vl_<code><mode>v4qi2_mask_store): Use a single pattern with
14629 %k0 and %1 for -masm=intel rather than two patterns, one with %0 and
14631 (*avx512vl_<code><mode>v8qi2_store): Use %q0 instead of %0 for
14633 (avx512vl_<code><mode>v8qi2_mask_store): Use a single pattern with
14634 %q0 and %1 for -masm=intel rather than two patterns, one with %0 and
14635 %g1 and one with %0 and %1.
14636 (avx512er_vmrcp28<mode><round_saeonly_name>,
14637 avx512er_vmrsqrt28<mode><round_saeonly_name>): Use %<iptr>1 instead of
14638 %1 for -masm=intel.
14639 (avx5124fmaddps_4fmaddps_mask, avx5124fmaddps_4fmaddss_mask,
14640 avx5124fmaddps_4fnmaddps_mask, avx5124fmaddps_4fnmaddss_mask,
14641 avx5124vnniw_vp4dpwssd_mask, avx5124vnniw_vp4dpwssds_mask): Swap order
14642 of %0 and %{%4%} for -masm=intel.
14643 (avx5124fmaddps_4fmaddps_maskz, avx5124fmaddps_4fmaddss_maskz,
14644 avx5124fmaddps_4fnmaddps_maskz, avx5124fmaddps_4fnmaddss_maskz,
14645 avx5124vnniw_vp4dpwssd_maskz, avx5124vnniw_vp4dpwssds_maskz): Swap
14646 order of %0 and %{%5%}%{z%} for -masm=intel.
14648 2018-04-17 Jan Hubicka <jh@suse.cz>
14651 * ipa-devirt.c (odr_types_equivalent_p): Handle bit fields.
14653 2018-04-17 Martin Liska <mliska@suse.cz>
14656 * multiple_target.c (create_dispatcher_calls): Set apostrophes
14657 for target_clone error message. Make default implementation
14658 clone to be a local declaration.
14659 (separate_attrs): Add new argument and check for an empty
14661 (expand_target_clones): Handle it.
14662 (ipa_target_clone): Make redirection just for target_clones
14665 2018-04-16 Cesar Philippidis <cesar@codesourcery.com>
14666 Tom de Vries <tom@codesourcery.com>
14668 PR middle-end/84955
14669 * omp-expand.c (expand_oacc_for): Add dummy false branch for
14670 tiled basic blocks without omp continue statements.
14672 2018-04-16 Aaron Sawdey <acsawdey@linux.ibm.com>
14675 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Mark
14676 vec_extract expression as having side effects to make sure it gets
14679 2018-04-16 H.J. Lu <hongjiu.lu@intel.com>
14682 * config/i386/i386.c (get_builtin_code_for_version): Check
14685 2018-04-16 Olga Makhotina <olga.makhotina@intel.com>
14688 * config.gcc: Support "skylake".
14689 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
14691 * config/i386/i386.c (m_SKYLAKE): Define.
14692 (processor_target_table): Add "skylake".
14693 (ix86_option_override_internal): Add "skylake".
14694 (get_builtin_code_for_version): Handle PROCESSOR_SKYLAKE,
14695 PROCESSOR_CANNONLAKE.
14696 (get_builtin_code_for_version): Fix priority for
14697 PROCESSOR_ICELAKE_CLIENT, PROCESSOR_ICELAKE_SERVER,
14698 PROCESSOR_SKYLAKE-AVX512.
14699 * config/i386/i386.h (processor_costs): Define TARGET_SKYLAKE.
14700 (processor_type): Add PROCESSOR_SKYLAKE.
14702 2018-04-16 Paolo Carlini <paolo.carlini@oracle.com>
14703 Jason Merrill <jason@redhat.com>
14706 * convert.c (convert_to_integer_1): Use direct recursion for
14707 enumeral types and types with a precision less than the number
14708 of bits in their mode.
14710 2018-04-16 Julia Koval <julia.koval@intel.com>
14713 * config/i386/x86-tune.def (X86_TUNE_SSE_UNALIGNED_LOAD_OPTIMAL,
14714 X86_TUNE_SSE_UNALIGNED_STORE_OPTIMAL): Add m_SKYLAKE_AVX512
14716 2018-04-14 Segher Boessenkool <segher@kernel.crashing.org>
14719 * config/rs6000/rs6000.opt (mdirect-move): Make deprecated.
14720 * doc/invoke.texi (RS/6000 and PowerPC Options): Remove -mdirect-move
14721 and -mno-direct-move.
14723 2018-04-13 Paul A. Clarke <pc@us.ibm.com>
14726 * config/rs6000/emmintrin.h (_mm_slli_epi{16,32,64}):
14727 Ensure that vec_splat_s32 is only called with 0 <= shift < 16.
14728 Ensure negative shifts result in {0}.
14730 2018-04-13 Vladimir Makarov <vmakarov@redhat.com>
14732 PR rtl-optimization/79916
14733 * config/rs6000/rs6000.c (rs6000_emit_move): Use assigned hard
14734 regs (if any) to define how to gnerate SD moves when LRA is in
14737 2018-04-13 Jakub Jelinek <jakub@redhat.com>
14739 PR rtl-optimization/85393
14740 * except.h (expand_dw2_landing_pad_for_region): Remove declaration.
14741 * except.c (expand_dw2_landing_pad_for_region): Make static.
14742 * bb-reorder.c (fix_up_crossing_landing_pad): In new_bb emit just
14743 a label and unconditional jump to old_bb, rather than
14744 expand_dw2_landing_pad_for_region insn(s) and jump to single_succ
14747 PR rtl-optimization/85376
14748 * simplify-rtx.c (simplify_const_unary_operation): For CLZ and CTZ and
14749 zero op0, if C?Z_DEFINED_VALUE_AT_ZERO is false, return NULL_RTX
14750 instead of a specific value.
14752 2018-04-13 Jan Hubicka <hubicka@ucw.cz>
14753 Bin Cheng <bin.cheng@arm.com>
14755 PR tree-optimization/82965
14756 PR tree-optimization/83991
14757 * cfgloopanal.c (expected_loop_iterations_unbounded): Add
14758 by_profile_only parameter.
14759 * cfgloopmanip.c (scale_loop_profile): Further scale loop's profile
14760 information if the loop was predicted to iterate too many times.
14761 * cfgloop.h (expected_loop_iterations_unbounded): Update prototype
14763 2018-04-13 Jan Hubicka <hubicka@ucw.cz>
14766 * config/i386/i386.c (ix86_can_inline_p): Allow safe transitions for
14769 2018-04-13 Martin Liska <mliska@suse.cz>
14770 Jakub Jelinek <jakub@redhat.com>
14772 PR middle-end/81657
14773 * expr.h (enum block_op_methods): Add BLOCK_OP_NO_LIBCALL_RET.
14774 * expr.c (emit_block_move_hints): Handle BLOCK_OP_NO_LIBCALL_RET.
14775 * builtins.c (expand_builtin_memory_copy_args): Use
14776 BLOCK_OP_NO_LIBCALL_RET method for mempcpy with non-ignored target,
14777 handle dest_addr == pc_rtx.
14779 2018-04-12 Segher Boessenkool <segher@kernel.crashing.org>
14782 * config/rs6000/rs6000.md (fix_trunc<mode>si2): Use legacy code if
14783 asked to not generate direct moves.
14784 (fix_trunc<mode>si2_stfiwx): Similar.
14785 (fix_trunc<mode>si2_internal): Similar.
14787 2018-04-12 Jakub Jelinek <jakub@redhat.com>
14790 * var-tracking.c (add_stores): Handle STRICT_LOW_PART SET_DEST.
14791 * cselib.c (cselib_record_sets): For STRICT_LOW_PART dest,
14792 lookup if dest in some wider mode is known to be const0_rtx and
14793 if so, record permanent equivalence for it to be ZERO_EXTEND of
14794 the narrower mode destination.
14796 2018-04-12 Cesar Philippidis <cesar@codesourcery.com>
14798 * lto-streamer-out.c (output_function): Revert 259346.
14799 * omp-expand.c (expand_oacc_for): Likewise.
14801 2018-04-12 Alexander Monakov <amonakov@ispras.ru>
14803 PR rtl-optimization/85354
14804 * sel-sched-ir.c (sel_init_pipelining): Move cfg_cleanup call...
14805 * sel-sched.c (sel_global_init): ... here.
14807 2018-04-12 Eric Botcazou <ebotcazou@adacore.com>
14810 * lto-wrapper.c (debug_objcopy): Open the files in binary mode.
14811 * dwarf2out.c (dwarf2out_early_finish): Do not generate assembly in LTO
14812 mode for PE-COFF targets.
14813 * config/i386/i386-protos.h (i386_pe_asm_lto_start): Declare.
14814 (i386_pe_asm_lto_end): Likewise.
14815 * config/i386/cygming.h (TARGET_ASM_LTO_START): Define.
14816 (TARGET_ASM_LTO_END): Likewise.
14817 * config/i386/winnt.c (saved_debug_info_level): New static variable.
14818 (i386_pe_asm_lto_start): New function.
14819 (i386_pe_asm_lto_end): Likewise.
14821 2018-04-12 Cesar Philippidis <cesar@codesourcery.com>
14822 Richard Biener <rguenther@suse.de>
14824 PR middle-end/84955
14825 * lto-streamer-out.c (output_function): Fix CFG loop state before
14827 * omp-expand.c (expand_oacc_for): Handle calls to internal
14828 functions like regular functions.
14830 2018-04-12 Richard Biener <rguenther@suse.de>
14833 * dwarf2out.c (init_sections_and_labels): Use debug_line_section[_label]
14834 for the early LTO debug to properly generate references to it
14835 during DIE emission. Do not re-use that for the skeleton for
14837 (dwarf2out_early_finish): Likewise.
14839 2018-04-12 Jakub Jelinek <jakub@redhat.com>
14842 * config/i386/sse.md
14843 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name> split,
14844 <mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name> split,
14845 vec_extract_lo_<mode><mask_name> split, vec_extract_lo_v32hi,
14846 vec_extract_lo_v64qi): For non-AVX512VL if input is xmm16+ reg
14847 and output is a reg, avoid creating invalid lowpart subreg, but
14848 instead split into a 512-bit move. Don't split if not AVX512VL,
14849 input is xmm16+ reg and output is a mem.
14850 (vec_extract_lo_<mode><mask_name>, vec_extract_lo_v32hi,
14851 vec_extract_lo_v64qi): Don't require split if not AVX512VL, input is
14852 xmm16+ reg and output is a mem.
14854 2018-04-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14856 * config/s390/s390.c (s390_output_indirect_thunk_function): Check
14857 also for flag_dwarf2_cfi_asm.
14859 2018-04-12 Jakub Jelinek <jakub@redhat.com>
14861 PR rtl-optimization/85342
14862 * regcprop.c (copyprop_hardreg_forward_1): Remove replaced array, use
14863 a bool scalar var inside of the loop instead. Don't try to update
14864 recog_data.operand after failed apply_change_group.
14866 2018-04-12 Tom de Vries <tom@codesourcery.com>
14869 * config/nvptx/nvptx.c (flexible_array_member_type_p): New function.
14870 (nvptx_assemble_decl_begin): Add undefined param. Declare undefined
14871 array with flexible array member as array without given dimension.
14872 (nvptx_assemble_undefined_decl): Set nvptx_assemble_decl_begin call
14873 argument for undefined param to true.
14875 2018-04-11 Aaron Sawdey <acsawdey@linux.ibm.com>
14878 * doc/invoke.texi (RS/6000 and PowerPC Options): Document options
14879 -mcall- and -mtraceback=. Remove options -mabi=spe and -mabi=no-spe
14880 from PowerPC section.
14881 * config/rs6000/sysv4.opt (mcall-): Improve help text.
14882 * config/rs6000/rs6000.opt (mblock-compare-inline-limit=): Trim
14883 help text that is too long.
14884 * config/rs6000/rs6000.opt (mblock-compare-inline-loop-limit=): Trim
14885 help text that is too long.
14886 * config/rs6000/rs6000.opt (mstring-compare-inline-limit=): Trim
14887 help text that is too long.
14889 2018-04-11 Uros Bizjak <ubizjak@gmail.com>
14891 * config/alpha/alpha.md (stack_probe_internal): Rename
14892 from "probe_stack". Update all callers.
14894 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
14896 PR rtl-optimization/84566
14897 * sched-deps.c (sched_analyze_insn): Check deps->readonly when invoking
14898 sched_macro_fuse_insns.
14900 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
14903 * sched-rgn.c (add_branch_dependences): Move sel_sched_p check here...
14904 (compute_block_dependences): ... from here.
14906 2018-04-11 Jakub Jelinek <jakub@redhat.com>
14908 PR tree-optimization/85331
14909 * vec-perm-indices.h (vec_perm_indices::clamp): Change input type
14910 from int to HOST_WIDE_INT.
14912 2018-04-11 Martin Jambor <mjambor@suse.cz>
14915 * ipa-cp.c (propagate_vals_across_pass_through): Expand comment.
14916 (cgraph_edge_brings_value_p): New parameter dest_val, check if it is
14917 not the same as the source val.
14918 (cgraph_edge_brings_value_p): New parameter.
14919 (gather_edges_for_value): Pass destination value to
14920 cgraph_edge_brings_value_p.
14921 (perhaps_add_new_callers): Likewise.
14922 (get_info_about_necessary_edges): Likewise and exclude values brought
14923 only by self-recursive edges.
14924 (create_specialized_node): Redirect only clones of self-calling edges.
14925 (+self_recursive_pass_through_p): New function.
14926 (find_more_scalar_values_for_callers_subset): Use it.
14927 (find_aggregate_values_for_callers_subset): Likewise.
14928 (known_aggs_to_agg_replacement_list): Removed.
14929 (decide_whether_version_node): Re-calculate known constants for all
14930 remaining context clones.
14932 2018-04-11 Richard Biener <rguenther@suse.de>
14935 * dwarf2out.c (dwarf2out_finish): Remove DW_AT_stmt_list attribute
14936 from early DWARF output.
14937 (dwarf2out_early_finish): Output line info unconditionally into
14938 early DWARF and add reference to it.
14940 2018-04-11 Jakub Jelinek <jakub@redhat.com>
14943 * config/i386/sse.md (iptr): Add V16SFmode and V8DFmode cases.
14944 (<avx512>_vec_dup<mode><mask_name>): Use a single pattern for modes
14945 other than V2DFmode using iptr mode attribute.
14946 (<avx512>_vec_dup<mode><mask_name>): Use iptr mode attribute.
14948 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
14950 PR rtl-optimization/84659
14951 * sel-sched-ir.c (sel_init_pipelining): Invoke cleanup_cfg.
14953 2018-04-11 Jakub Jelinek <jakub@redhat.com>
14956 * dwarf2out.c (skip_loc_list_entry): Don't call size_of_locs if
14958 (output_loc_list): Pass address of a dummy size variable even in the
14959 locview handling loop.
14960 (index_location_lists): Add comment on why skip_loc_list_entry can't
14963 2018-04-11 Thomas Preud'homme <thomas.preudhomme@arm.com>
14966 * config/arm/arm-builtins.c (arm_expand_builtin): Force input operand
14969 2018-04-10 Aaron Sawdey <acsawdey@linux.ibm.com>
14972 * doc/invoke.texi (RS/6000 and PowerPC Options): Document options
14973 -mblock-compare-inline-limit, -mblock-compare-inline-loop-limit,
14974 and -mstring-compare-inline-limit.
14976 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
14979 * config/rs6000/rs6000.md (allocate_stack): Put the residual size
14980 for stack clash protection in a register whenever we need it to be in
14983 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
14985 * common/config/rs6000/rs6000-common.c (rs6000_option_init_struct):
14986 Enable -fasynchronous-unwind-tables by default if OBJECT_FORMAT_ELF.
14988 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
14991 * config/rs6000/rs6000.opt (mtraceback=): Show the allowed values in
14993 (mlong-double-): Ditto.
14994 * config/rs6000/sysv4.opt (msdata=): Ditto.
14995 (mtls-size=): Ditto.
14997 2018-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
14999 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
15000 erroneous entries for
15001 "vector int vec_ldl (int, long int *)", and
15002 "vector unsigned int vec_ldl (int, unsigned long int *)".
15003 Add comments and entries for
15004 "vector bool char vec_ldl (int, bool char *)",
15005 "vector bool short vec_ldl (int, bool short *)",
15006 "vector bool int vec_ldl (int, bool int *)",
15007 "vector bool long long vec_ldl (int, bool long long *)",
15008 "vector pixel vec_ldl (int, pixel *)",
15009 "vector long long vec_ldl (int, long long *)",
15010 "vector unsigned long long vec_ldl (int, unsigned long long *)".
15011 * config/rs6000/rs6000.c (rs6000_init_builtins): Initialize new
15012 type tree bool_long_long_type_node and correct definition of
15013 bool_V2DI_type_node to make reference to this new type tree.
15014 (rs6000_mangle_type): Replace erroneous reference to
15015 bool_long_type_node with bool_long_long_type_node.
15016 * config/rs6000/rs6000.h (enum rs6000_builtin_type_index): Add
15017 comments to emphasize sign distinctions for char and int types and
15018 replace RS6000_BTI_bool_long constant with
15019 RS6000_BTI_bool_long_long constant. Also add comment to restrict
15020 use of RS6000_BTI_pixel.
15021 (bool_long_type_node): Remove this macro definition.
15022 (bool_long_long_type_node): New macro definition
15024 2018-04-10 Jakub Jelinek <jakub@redhat.com>
15026 PR rtl-optimization/85300
15027 * combine.c (subst): Handle subst of CONST_SCALAR_INT_P new_rtx also
15028 into FLOAT and UNSIGNED_FLOAT like ZERO_EXTEND, return a CLOBBER if
15029 simplify_unary_operation fails.
15031 2018-04-10 Martin Liska <mliska@suse.cz>
15033 * gdbhooks.py: Add pretty-printers for varpool_node, symtab_node,
15034 cgraph_edge and ipa_ref.
15036 2018-04-10 Jakub Jelinek <jakub@redhat.com>
15040 * config/i386/sse.md
15041 (<extract_type>_vinsert<shuffletype><extract_suf>_mask): Fix
15042 computation of the VEC_MERGE selector from mask.
15043 (<extract_type>_vinsert<shuffletype><extract_suf>_1<mask_name>):
15044 Fix decoding of the VEC_MERGE selector into mask.
15046 2018-04-10 Richard Sandiford <richard.sandiford@linaro.org>
15048 PR tree-optimization/85286
15049 * tree-vect-data-refs.c (vect_get_smallest_scalar_type):
15051 2018-04-10 Richard Sandiford <richard.sandiford@linaro.org>
15053 * final.c (final_1): Set insn_last_address as well as
15054 insn_current_address.
15056 2018-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15059 * explow.c (emit_stack_probe): Call validize_mem on memory location
15060 before passing it to gen_probe_stack. Create address operand and
15061 legitimize it for the probe_stack_address case.
15063 2018-04-09 Jan Hubicka <jh@suse.cz>
15066 * ipa-devirt.c (rebuild_type_inheritance-hash): New.
15067 * ipa-utils.h (rebuild_type_inheritance-hash): Declare.
15068 * tree.c (free_lang_data_in_type): Fix handling of binfos;
15070 (free_lang_data): Rebuild type inheritance graph.
15072 2018-04-09 Martin Sebor <msebor@redhat.com>
15074 * invoke.texi (-finline-small-functions): Mention other optimization
15076 (-findirect-inlining, -fpartial-inlining): Same.
15077 (-finline-functions-called-once): Same.
15078 (-freorder-blocks-and-partition): Same.
15080 2018-04-09 Jan Hubicka <jh@suse.cz>
15083 * cfgcleanup.c (try_forward_edges): Do not give up on crossing
15084 jumps; choose last target that matches the criteria (i.e.
15085 no partition changes for non-crossing jumps).
15086 * cfgrtl.c (cfg_layout_redirect_edge_and_branch): Add basic
15087 support for redirecting crossing jumps to non-crossing.
15089 2018-04-09 Alexey Brodkin <abrodkin@synopsys.com>
15091 * config/arc/arc.c (arc_expand_prologue): Set stack usage info
15092 also for naked functions.
15094 2018-04-09 Claudiu Zissulescu <claziss@synopsys.com>
15096 * config/arc/arc.md (add_shift): New pattern.
15097 (add_shift2): Likewise.
15098 (sub_shift): Likewise.
15099 (sub_shift_cmp0_noout): Likewise.
15100 (compare_si_ashiftsi): Likewise.
15101 (xbfu_cmp0_noout): New combine pattern.
15102 (xbfu_cmp0"): Likewise.
15103 (movsi_set_cc_insn): Place the predicable variant first.
15104 (commutative_binary_cmp0_noout): Remove clobber.
15105 (commutative_binary_cmp0): New pattern.
15106 (noncommutative_binary_cmp0): Likewise.
15107 (noncommutative_binary_cmp0_noout): Likewise.
15108 (noncommutative_binary_comparison_result_used): Removed.
15109 (rsub_cmp0): New pattern.
15110 (rsub_cmp0_noout): Likewise.
15111 (extzvsi): Changed, keep only meaningful variants.
15112 (SQH, SEZ): New iterators.
15113 (SQH_postfix): New mode attribute.
15114 (SEZ_prefix): New code attribute.
15115 (<SEZ_prefix>xt<SQH_postfix>_cmp0_noout): New instruction pattern.
15116 (<SEZ_prefix>xt<SQH_postfix>_cmp0): Likewise.
15117 * config/arc/predicates.md (cc_set_register): Use CC_REG instead
15118 of numerical value.
15119 (noncommutative_operator): Check the availability of barrel
15122 2018-04-09 Richard Biener <rguenther@suse.de>
15124 PR tree-optimization/85284
15125 * tree-ssa-loop-niter.c (number_of_iterations_exit_assumptions):
15126 Only use the niter constraining form of simple_iv when the exit
15127 is always executed.
15129 2018-04-09 Tom de Vries <tom@codesourcery.com>
15132 * config/nvptx/nvptx.md (define_c_enum "unspecv"): Add UNSPECV_MEMBAR.
15133 (define_expand "*memory_barrier"): New define_expand.
15134 (define_insn "memory_barrier"): New insn.
15136 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
15138 PR rtl-optimization/80463
15139 PR rtl-optimization/83972
15140 PR rtl-optimization/83480
15142 * sel-sched-ir.c (has_dependence_note_mem_dep): Take into account the
15143 correct producer for the insn.
15144 (tidy_control_flow): Fixup seqnos in case of debug insns.
15146 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
15148 PR rtl-optimization/83913
15150 * sel-sched-ir.c (merge_expr_data): Choose the middle between two
15151 different sched-times when merging exprs.
15153 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
15155 PR rtl-optimization/83962
15157 * sel-sched-ir.c (tidy_control_flow): Correct the order in which we call
15158 tidy_fallthru_edge and tidy_control_flow.
15160 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
15162 PR rtl-optimization/83530
15164 * sel-sched.c (force_next_insn): New global variable.
15165 (remove_insn_for_debug): When force_next_insn is true, also leave only
15166 next insn in the ready list.
15167 (sel_sched_region): When the region wasn't scheduled, make another pass
15168 over it with force_next_insn set to 1.
15170 2018-04-08 Monk Chiang <sh.chiang04@gmail.com>
15172 * config.gcc (nds32le-*-*, nds32be-*-*): Add nds32/nds32_intrinsic.h
15174 * config/nds32/constants.md (unspec_volatile_element): Add enum values
15175 for interrupt control.
15176 * config/nds32/nds32-intrinsic.c: Implementation of intrinsic
15177 functions for interrupt control.
15178 * config/nds32/nds32-intrinsic.md: Likewise.
15179 * config/nds32/nds32_intrinsic.h: Likewise.
15180 * config/nds32/nds32.h (nds32_builtins): Likewise.
15182 2018-04-08 Chung-Ju Wu <jasonwucj@gmail.com>
15184 * config/nds32/nds32.c (nds32_init_machine_status,
15185 nds32_legitimate_index_p, nds32_legitimate_address_p): Consider
15186 strict_aligned_p field.
15187 (nds32_expand_to_rtl_hook): New function.
15188 (TARGET_EXPAND_TO_RTL_HOOK): Define.
15189 * config/nds32/nds32.h (machine_function): Add strict_aligned_p field.
15191 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
15192 Chung-Ju Wu <jasonwucj@gmail.com>
15194 * config.gcc (nds32*-*-*): Check that n7 is valid to --with-cpu.
15195 * config/nds32/nds32-n7.md: New file.
15196 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N7.
15197 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n7
15199 * config/nds32/nds32-protos.h: More declarations for n7 pipeline.
15200 * config/nds32/nds32.md (pipeline_model): Add n7.
15201 * config/nds32/nds32.opt (mcpu): Support n7 pipeline cpus.
15202 * config/nds32/pipelines.md: Include n7 settings.
15204 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
15205 Chung-Ju Wu <jasonwucj@gmail.com>
15207 * config.gcc (nds32*-*-*): Check that e8 is valid to --with-cpu.
15208 * config/nds32/nds32-e8.md: New file.
15209 * config/nds32/nds32-opts.h (nds32-cpu_type): Add CPU_E8.
15210 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for e8
15212 * config/nds32/nds32-protos.h: More declarations for e8 pipeline.
15213 * config/nds32/nds32.md (pipeline_model): Add e8.
15214 * config/nds32/nds32.opt (mcpu): Support e8 pipeline cpus.
15215 * config/nds32/pipelines.md: Include e8 settings.
15217 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
15218 Chung-Ju Wu <jasonwucj@gmail.com>
15220 * config.gcc (nds32*-*-*): Check that n6/n8/s8 are valid to --with-cpu.
15221 * config/nds32/nds32-n8.md: New file.
15222 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N6 and CPU_N8.
15223 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n8
15225 * config/nds32/nds32-protos.h: More declarations for n8 pipeline.
15226 * config/nds32/nds32-utils.c: More implementations for n8 pipeline.
15227 * config/nds32/nds32.md (pipeline_model): Add n8.
15228 * config/nds32/nds32.opt (mcpu): Support n8 pipeline cpus.
15229 * config/nds32/pipelines.md: Include n8 settings.
15231 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
15232 Chung-Ju Wu <jasonwucj@gmail.com>
15234 * config.gcc (nds32*): Add nds32-utils.o into extra_objs.
15235 * config/nds32/nds32-n9-2r1w.md: New file.
15236 * config/nds32/nds32-n9-3r2w.md: New file.
15237 * config/nds32/nds32-opts.h (nds32_cpu_type, nds32_mul_type,
15238 nds32_register_ports): New or modify for cpu n9.
15239 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n9
15241 * config/nds32/nds32-protos.h: More declarations for n9 pipeline.
15242 * config/nds32/nds32-utils.c: New file.
15243 * config/nds32/nds32.h (TARGET_PIPELINE_N9, TARGET_PIPELINE_SIMPLE,
15244 TARGET_MUL_SLOW): Define.
15245 * config/nds32/nds32.md (pipeline_model): New attribute.
15246 * config/nds32/nds32.opt (mcpu, mconfig-mul, mconfig-register-ports):
15247 New options that support cpu n9.
15248 * config/nds32/pipelines.md: Include n9 settings.
15249 * config/nds32/t-nds32 (nds32-utils.o): Add dependency.
15251 2018-04-08 Chung-Ju Wu <jasonwucj@gmail.com>
15253 * config/nds32/nds32-md-auxiliary.c (output_cond_branch): Output align
15254 information if necessary.
15255 (output_cond_branch_compare_zero): Likewise.
15256 * config/nds32/nds32.c (nds32_adjust_insn_length): Consider align case.
15257 (nds32_target_alignment): Refine for alignment.
15258 * config/nds32/nds32.h (NDS32_ALIGN_P): Define.
15259 (FUNCTION_BOUNDARY): Modify.
15260 * config/nds32/nds32.md (call_internal, call_value_internal): Consider
15262 * config/nds32/nds32.opt (malways-align, malign-functions): New.
15264 2018-04-08 Monk Chiang <sh.chiang04@gmail.com>
15266 * config/nds32/constants.md (unspec_volatile_element): Add values for
15267 TLB operation and data prefetch.
15268 * config/nds32/nds32-intrinsic.c: Implementation of intrinsic
15269 functions for TLB operation and data prefetch.
15270 * config/nds32/nds32-intrinsic.md: Likewise.
15271 * config/nds32/nds32_intrinsic.h: Likewise.
15272 * config/nds32/nds32.c (nds32_dpref_names): Likewise.
15273 (nds32_print_operand): Likewise.
15274 * config/nds32/nds32.h (nds32_builtins): Likewise.
15276 2018-04-07 Thomas Koenig <tkoenig@gcc.gnu.org>
15277 Andrew Pinski <pinsika@gcc.gnu.org>
15279 PR middle-end/82976
15280 * match.pd: Use constant_boolean_node of correct type instead of
15281 boolean_true_node or boolean_false_node for simplifying
15282 pointer comparisons to zero.
15284 2018-04-07 Jakub Jelinek <jakub@redhat.com>
15286 PR tree-optimization/80021
15287 * tree.c (verify_type_variant): Make error call in verify_variant_match
15288 translatable and remove final full stop.
15290 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
15292 * config/nds32/constants.md (unspec_volatile_element): Add
15293 UNSPEC_VOLATILE_EH_RETURN.
15294 * config/nds32/nds32-md-auxiliary.c (nds32_output_stack_push,
15295 nds32_output_stack_pop): Support dwarf exception handling process.
15296 * config/nds32/nds32-protos.h (nds32_dynamic_chain_address): Declare.
15297 * config/nds32/nds32.c (nds32_init_machine_status): Support dwarf
15298 exception handling process.
15299 (nds32_compute_stack_frame): Likewise.
15300 (nds32_return_addr_rtx): Likewise.
15301 (nds32_initial_elimination_offset): Likewise.
15302 (nds32_expand_prologue): Likewise.
15303 (nds32_expand_epilogue): Likewise.
15304 (nds32_dynamic_chain_address): New function.
15305 * config/nds32/nds32.h (machine_function): Add fields for dwarf
15306 exception handling.
15307 (DYNAMIC_CHAIN_ADDRESS): Define.
15308 (EH_RETURN_DATA_REGNO): Define.
15309 (EH_RETURN_STACKADJ_RTX): Define.
15310 * config/nds32/nds32.md (eh_return, nds32_eh_return): Implement
15311 patterns for dwarf exception handling.
15313 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
15315 * config/nds32/nds32.h: Clean up obsolete macros.
15317 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
15319 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
15320 Add enum values for particular instructions.
15321 * config/nds32/nds32-intrinsic.c: Implementation of expanding
15322 particular intrinsic functions.
15323 * config/nds32/nds32-intrinsic.md: Likewise.
15324 * config/nds32/nds32_intrinsic.h: Likewise.
15325 * config/nds32/nds32.h (nds32_builtins): Likewise.
15326 * config/nds32/nds32.md (type): Add pbsad and pbsada.
15327 (btst, ave): New patterns for particular instructions.
15329 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
15331 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
15332 Add enum values for atomic load/store and memory sync.
15333 * config/nds32/nds32-intrinsic.c: Implementation for atomic load/store
15335 * config/nds32/nds32-intrinsic.md: Likewise.
15336 * config/nds32/nds32_intrinsic.h: Likewise.
15337 * config/nds32/nds32.h (nds32_builtins): Likewise.
15339 2018-04-07 Jakub Jelinek <jakub@redhat.com>
15341 PR tree-optimization/85257
15342 * fold-const.c (native_encode_vector): If not all elts could fit
15343 and off is -1, return 0 rather than offset.
15344 * tree-ssa-sccvn.c (vn_reference_lookup_3): Pass
15345 (offseti - offset2) / BITS_PER_UNIT as 4th argument to
15346 native_encode_expr. Verify len * BITS_PER_UNIT >= maxsizei. Don't
15347 adjust buffer in native_interpret_expr call.
15349 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
15351 * config/nds32/constants.md (unspec_volatile_element): Add cache
15352 control enum values.
15353 * config/nds32/nds32-intrinsic.c: Add cache control expand functions.
15354 * config/nds32/nds32-intrinsic.md: Add cache control patterns.
15355 * config/nds32/nds32.c (nds32_cctl_names): New.
15356 (nds32_print_operand): Handle cache control register names.
15357 * config/nds32/nds32.h (nds32_builtins): New enum values.
15358 * config/nds32/nds32_intrinsic.h: Add cache control enum types and
15360 * config/nds32/nds32.md (type): Add mmu.
15361 * config/nds32/pipelines.md (simple_insn): Add mmu.
15363 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
15365 * config/nds32/nds32.md (type): Remove call.
15366 * config/nds32/pipelines.md (simple_insn): Likewise.
15368 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
15370 * config/nds32/constants.md (unspec_volatile_element): Add
15371 UNSPEC_VOLATILE_FMFCSR, UNSPEC_VOLATILE_FMTCSR and
15372 UNSPEC_VOLATILE_FMFCFG.
15373 * config/nds32/nds32-intrinsic.c (bdesc_noarg): New builtin
15374 description for fmfcfg and fmfcsr.
15375 (bdesc_1arg): Add fmtcsr.
15376 (bdesc_2arg): Add fcpynss, fcpyss, fcpynsd and fcpysd.
15377 (nds32_expand_builtin_impl): Deal with FPU intrinsic functions.
15378 * config/nds32/nds32-intrinsic.md (unspec_fcpynsd, unspec_fcpysd,
15379 unspec_fcpynss, unspec_fcpysd, unspec_fcpyss, unspec_fmfcsr,
15380 unspec_fmfcfg): New patterns.
15381 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_FMFCFG,
15382 NDS32_BUILTIN_FMFCSR, NDS32_BUILTIN_FMTCSR, NDS32_BUILTIN_FCPYNSS,
15383 NDS32_BUILTIN_FCPYSS,NDS32_BUILTIN_FCPYNSD and NDS32_BUILTIN_FCPYSD.
15384 * config/nds32/nds32_intrinsic.h (__nds32__fcpynsd, __nds32__fcpynss,
15385 __nds32__fcpysd, __nds32__fcpyss, __nds32__fmfcsr, __nds32__fmtcsr,
15386 __nds32__fmfcfg): Define.
15388 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
15390 * config/nds32/nds32.c (nds32_intrinsic_register_names): Add more
15391 intrinsic register names.
15392 * config/nds32/nds32_intrinsic.h (nds32_intrinsic_registers): Add more
15393 intrinsic register enum values and macros.
15395 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
15397 * config/nds32/nds32.c (nds32_legitimate_index_p): Modify condition
15398 for load/store addressing form.
15399 (nds32_print_operand_address): Likewise.
15401 2018-04-06 Eric Botcazou <ebotcazou@adacore.com>
15404 * config/sparc/sparc.c (sparc_expand_move): Deal with symbolic operands
15405 based on LABEL_REF. Remove useless assertion.
15406 (pic_address_needs_scratch): Fix formatting.
15407 (sparc_legitimize_pic_address): Minor tweaks.
15408 (sparc_delegitimize_address): Adjust assertion accordingly.
15409 * config/sparc/sparc.md (movsi_pic_label_ref): Change label_ref_operand
15410 into symbolic_operand.
15411 (movsi_high_pic_label_ref): Likewise.
15412 (movsi_lo_sum_pic_label_ref): Likewise.
15413 (movdi_pic_label_ref): Likewise.
15414 (movdi_high_pic_label_ref): Likewise.
15415 (movdi_lo_sum_pic_label_ref): Likewise.
15417 2018-04-06 Amaan Cheval <amaan.cheval@gmail.com>
15419 * config.gcc (x86_64-*-rtems*): Add rtems.h to tm_file for
15420 custom LIB_SPEC setup.
15422 2018-04-06 Ruslan Bukin <br@bsdpad.com>
15423 Kito Cheng <kito.cheng@gmail.com>
15425 * config.gcc (riscv*-*-freebsd*): Add RISC-V FreeBSD support.
15426 * config/riscv/freebsd.h: New.
15428 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
15430 * config/nds32/nds32.c (nds32_adjust_insn_length): Refine.
15431 * config/nds32/nds32.h (ADJUST_INSN_LENGTH): Change the location in
15434 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
15435 Kito Cheng <kito.cheng@gmail.com>
15437 * config/nds32/nds32-md-auxiliary.c (nds32_output_return,
15438 nds32_output_call, nds32_symbol_binds_local_p): New functions.
15439 * config/nds32/nds32-protos.h (nds32_output_call,
15440 nds32_output_return): Declare.
15441 * config/nds32/nds32.md: Refine all the call and return patterns.
15443 2018-04-06 Jakub Jelinek <jakub@redhat.com>
15446 * dwarf2out.c (rtl_for_decl_init): For STRING_CST initializer only
15447 build CONST_STRING if TYPE_MAX_VALUE is non-NULL and is INTEGER_CST.
15449 PR rtl-optimization/84872
15450 * cfgloopmanip.c (create_preheader): Use make_forwarder_block even if
15451 nentry == 1 when CP_FALLTHRU_PREHEADERS and single_entry is
15452 EDGE_CROSSING edge.
15454 2018-04-06 Tamar Christina <tamar.christina@arm.com>
15456 * expr.c (copy_blkmode_to_reg): Revert 254862.
15457 * doc/sourcebuild.texi (word_mode_no_slow_unalign): Likewise.
15459 2018-04-06 Richard Biener <rguenther@suse.de>
15461 PR middle-end/85244
15462 * tree-dfa.c (get_ref_base_and_extent): Reset seen_variable_array_ref
15463 after seeing a component reference with an adjacent field. Treat
15464 refs to arrays at struct end of external decls similar to
15465 refs to unconstrained commons.
15467 2018-04-06 Jakub Jelinek <jakub@redhat.com>
15470 * fold-const.c (twoval_comparison_p): Remove SAVE_P argument and don't
15471 look through SAVE_EXPRs with non-side-effects argument. Adjust
15473 (fold_comparison): Adjust twoval_comparison_p caller, don't handle
15476 2018-04-06 Richard Biener <rguenther@suse.de>
15478 PR middle-end/85180
15479 * alias.c (find_base_term): New wrapper around find_base_term
15480 unwinding CSELIB_VAL_PTR changes.
15481 (find_base_term): Do not restore CSELIB_VAL_PTR during the
15484 2018-04-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15486 * config/s390/s390.c (s390_z10_optimize_cmp): Expand dedicated NOP
15488 * config/s390/s390.md (UNSPECV_NOP_LR_0, UNSPECV_NOP_LR_1): New
15489 constant definitions.
15490 ("nop"): lr 0,0 -> nopr r0
15491 ("nop_lr0", "nop_lr1"): New insn definitions.
15493 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
15495 * config/nds32/nds32.md (*stack_push, *stack_pop): Use
15496 NDS32_V3PUSH_AVAILABLE_P macro.
15498 2018-04-06 Monk Chiang <sh.chiang04@gmail.com>
15499 Chung-Ju Wu <jasonwucj@gmail.com>
15501 * config.gcc (nds32*-*-*): Add v2j v3f v3s checking.
15502 (nds32*-*-*): Add float and fpu_config into supported_defaults.
15503 * common/config/nds32/nds32-common.c (TARGET_DEFAULT_TARGET_FLAGS):
15504 Include TARGET_DEFAULT_FPU_ISA and TARGET_DEFAULT_FPU_FMA.
15505 * config/nds32/constants.md (unspec_element): Add UNSPEC_COPYSIGN,
15506 UNSPEC_FCPYNSD, UNSPEC_FCPYNSS, UNSPEC_FCPYSD and UNSPEC_FCPYSS.
15507 * config/nds32/constraints.md: New constraints and checking for hard
15508 float configuration.
15509 * config/nds32/iterators.md: New mode iterator and attribute for hard
15510 float configuration.
15511 * config/nds32/nds32-doubleword.md: Use hard float alternatives and
15513 * config/nds32/nds32-fpu.md: New file.
15514 * config/nds32/nds32-md-auxiliary.c: New functions and checkings to
15515 deal with hard float code generation.
15516 * config/nds32/nds32-opts.h (nds32_arch_type): Add ARCH_V3F and
15518 (abi_type, float_reg_number): New enum type.
15519 * config/nds32/nds32-predicates.c: New predicates for hard float.
15520 * config/nds32/nds32-protos.h: Declare functions for hard float.
15521 * config/nds32/nds32.c: Implementation for hard float configuration.
15522 * config/nds32/nds32.h: Definitions for hard float configuration.
15523 * config/nds32/nds32.md: Include hard float machine description and
15524 modify patterns for hard float configuration.
15525 * config/nds32/nds32.opt: New options for hard float configuration.
15526 * config/nds32/predicates.md: New predicates for hard float
15529 2018-04-06 Kuan-Lin Chen <kuanlinchentw@gmail.com>
15531 * common/config/nds32/nds32-common.c
15532 (nds32_option_optimization_table): Enable -mreleax-hint by default.
15534 2018-04-05 Jakub Jelinek <jakub@redhat.com>
15536 PR middle-end/85195
15537 * match.pd (BIT_FIELD_REF CONSTRUCTOR@0 @1 @2): Use view_convert around
15538 CONSTRUCTOR_ELT (ctor, ...)->value.
15540 2018-04-05 Uros Bizjak <ubizjak@gmail.com>
15543 * config/i386/i386.md (define_attr "memory"): Handle rotate1 type.
15545 2018-04-05 Tom de Vries <tom@codesourcery.com>
15548 * config/nvptx/nvptx.c (nvptx_single): Fix neutering of bb with only
15551 2018-04-05 Shiva Chen <shiva0217@gmail.com>
15552 Kito Cheng <kito.cheng@gmail.com>
15554 * config/nds32/constraints.md (U33): Fine-tune checking condition.
15555 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Ditto.
15556 * config/nds32/nds32.h (nds32_16bit_address_type): Add
15557 ADDRESS_POST_MODIFY_LO_REG_IMM3U.
15559 2018-04-05 Shiva Chen <shiva0217@gmail.com>
15560 Kito Cheng <kito.cheng@gmail.com>
15562 * config/nds32/constraints.md (Ufe): New memory constraint.
15563 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format,
15564 nds32_output_16bit_load): Consider r8 register for lwi45.fe format.
15565 * config/nds32/nds32.c (nds32_print_operand): Output lwi45.fe
15567 * config/nds32/nds32.h (nds32_16bit_address_type): Add ADDRESS_R8_IMM7U.
15568 * config/nds32/nds32.md (*mov<mode>): Adjust pattern.
15570 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
15572 * config/nds32/nds32.md: Use optimize_size in the condition for
15573 alu-shift instructions.
15575 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
15577 * config/nds32/nds32.md (divsi4, udivsi4): New patterns.
15579 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
15581 * config/nds32/nds32.md (negsi2): Refine pattern.
15583 2018-04-05 Kito Cheng <kito.cheng@gmail.com>
15584 Chung-Ju Wu <jasonwucj@gmail.com>
15586 * config/nds32/iterators.md (shift_rotate): New code iterator.
15587 (shift): New code attribute.
15588 * config/nds32/nds32-md-auxiliary.c (nds32_expand_constant): New.
15589 * config/nds32/nds32-protos.h (nds32_expand_constant): Declare.
15590 * config/nds32/nds32.c (nds32_print_operand): Deal with more cases.
15591 * config/nds32/nds32.md (addsi3, *add_srli): Refine implementation for
15592 bit-wise operations.
15593 (andsi3, *andsi3): Ditto.
15594 (iorsi3, *iorsi3, *or_slli, *or_srli): Ditto.
15595 (xorsi3, *xorsi3, *xor_slli, *xor_srli): Ditto.
15596 (<shift>si3, *ashlsi3, *ashrsi3, *lshrsi3, *rotrsi3): Ditto.
15597 * config/nds32/predicates.md (nds32_rimm5u_operand, nds32_and_operand,
15598 nds32_ior_operand, nds32_xor_operand): New predicates.
15600 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
15602 * config/nds32/nds32.md (add<mode>3, sub<mode>3): Rename to ...
15603 (addsi3, subsi3): ... this.
15605 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
15607 * config/nds32/nds32.md (*sub_srli, *and_slli): Fine-tune predicator.
15609 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
15611 * config/nds32/nds32.md: Adjust indention.
15613 2018-04-05 Kito Cheng <kito.cheng@gmail.com>
15615 * config/nds32/nds32.md (feature): New attribute.
15617 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
15619 * config/nds32/nds32.md (subtype): New attribute.
15621 2018-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
15624 * config/arm/arm-builtins.c (arm_expand_builtin): Change
15625 expansion to perform a bitwise AND of the argument followed by a
15626 boolean negation of the result.
15628 2018-04-04 Peter Bergner <bergner@vnet.ibm.com>
15630 PR rtl-optimization/84878
15631 * ddg.c (add_cross_iteration_register_deps): Use DF_REF_BB to determine
15632 the basic block. Assert the use reference is not artificial and that
15633 it has an associated insn.
15635 2018-04-04 Michael Matz <matz@suse.de>
15637 * builtins.c (compute_objsize): Pass correct operand
15638 to array_at_struct_end_p.
15640 2018-04-04 Richard Biener <rguenther@suse.de>
15643 * dwarf2out.c (dwarf2out_register_external_die): Peel namespaces
15644 from contexts for DINFO_LEVEL_TERSE and below.
15646 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
15648 * config/nds32/nds32-doubleword.md (move_<mode>): Require
15649 resiter_operand condition.
15650 * config/nds32/nds32.md (*move<mode>): Ditto.
15652 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
15653 Monk Chiang <sh.chiang04@gmail.com>
15655 * config/nds32/nds32.md (movmisalign<mode>): New pattern.
15657 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
15659 * config/nds32/nds32.md (movqi, movhi): Merge into mov<mode>.
15661 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
15662 Kito Cheng <kito.cheng@gmail.com>
15664 * config/nds32/nds32-md-auxiliary.c (nds32_inverse_cond_code,
15665 nds32_cond_code_str, output_cond_branch,
15666 output_cond_branch_compare_zero, nds32_expand_cbranch,
15667 nds32_expand_cstore, nds32_expand_movcc,
15668 nds32_output_cbranchsi4_equality_zero,
15669 nds32_output_cbranchsi4_equality_reg,
15670 nds32_output_cbranchsi4_equality_reg_or_const_int,
15671 nds32_output_cbranchsi4_greater_less_zero: New functions.
15672 * config/nds32/nds32-protos.h (nds32_expand_cbranch,
15673 nds32_expand_cstore, nds32_expand_movcc,
15674 nds32_output_cbranchsi4_equality_zero,
15675 nds32_output_cbranchsi4_equality_reg,
15676 nds32_output_cbranchsi4_equality_reg_or_const_int,
15677 nds32_output_cbranchsi4_greater_less_zero): Declare.
15678 * config/nds32/predicates.md (nds32_movecc_comparison_operator,
15679 nds32_rimm11s_operand): New predicates.
15680 * config/nds32/nds32.h (nds32_expand_result_type): New enum type.
15681 * config/nds32/nds32.md: Rewrite all the branch and conditional move
15684 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
15686 * config/nds32/nds32-doubleword.md: Refine all the instruction type.
15687 * config/nds32/nds32.md: Ditto.
15688 * config/nds32/pipelines.md: Ditto.
15690 2018-04-04 Richard Biener <rguenther@suse.de>
15692 PR tree-optimization/85168
15693 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address): Avoid
15694 propagating abnormals.
15696 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
15698 * config/nds32/nds32.md (enabled): Use yes/no for this attribute.
15700 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
15701 Kito Cheng <kito.cheng@gmail.com>
15703 * config/nds32/nds32-md-auxiliary.c (nds32_long_call_p): New function.
15704 * config/nds32/nds32-protos.h (nds32_long_call_p): Declare.
15705 * config/nds32/nds32.c (nds32_function_ok_for_sibcall): New function.
15706 (TARGET_FUNCTION_OK_FOR_SIBCALL): Define.
15707 * config/nds32/nds32.md (sibcall_internal): New.
15708 (sibcall_register): Remove.
15709 (sibcall_immediate): Remove.
15710 (sibcall_value_internal): New.
15711 (sibcall_value_register): Remove.
15712 (sibcall_value_immediate): Remove.
15713 * config/nds32/predicates.md (nds32_general_register_operand): New.
15714 (nds32_call_address_operand): New.
15716 2018-04-03 Jakub Jelinek <jakub@redhat.com>
15718 PR rtl-optimization/85167
15719 * shrink-wrap.c (move_insn_for_shrink_wrap): Don't set bb_uses and
15720 bb_defs if *split_p, instead preinitialize it to NULL.
15722 PR tree-optimization/85156
15723 * builtins.c (fold_builtin_expect): Use save_expr on arg1 to avoid
15724 evaluating the argument multiple times.
15726 2018-04-03 Bill Schmidt <wschmidt@linux.ibm.com>
15728 * config/rs6000/emmintrin.h (_mm_cvtpd_epi32): Use __vector rather
15730 (_mm_cvtpd_ps): Likewise.
15731 (_mm_cvttpd_epi32): Likewise.
15732 * config/rs6000/mmintrin.h (_mm_unpacklo_pi8): Likewise.
15733 * config/rs6000/xmmintrin.h: For strict-ANSI C++ or C11, undefine
15734 vector, pixel, and bool following altivec.h include.
15736 2018-04-03 Martin Sebor <msebor@redhat.com>
15738 * doc/extend.texi (Common Function Attributes): Clarify.
15739 (const attribute): Likewise.
15740 (pure attribute): Likewise.
15742 2018-04-03 Jakub Jelinek <jakub@redhat.com>
15745 * config/i386/i386.c (ix86_expand_vector_set): Use
15746 HOST_WIDE_INT_1U << elt instead of 1 << elt. Formatting fix.
15748 2018-04-03 Uros Bizjak <ubizjak@gmail.com>
15750 * config/i386/i386.c (emit_i387_cw_initialization): Always use logic
15751 instructions when changing rounding bits to preserve precision bits
15752 in the x87 control word.
15754 2018-04-03 Martin Liska <mliska@suse.cz>
15756 PR tree-optimization/82491
15757 * rtl.h (strip_offset_and_add): Replace += suboffset with
15758 poly_uint64 () + suboffset.
15760 2018-03-29 Martin Liska <mliska@suse.cz>
15761 Martin Jambor <mjambor@suse.cz>
15764 * ipa-cp.c (propagate_bits_across_jump_function): Bail out if
15765 param_type is not an integral or pointer type.
15767 2018-04-03 Richard Biener <rguenther@suse.de>
15769 * sese.h (recompute_all_dominators): Remove.
15771 2018-04-02 Martin Sebor <msebor@redhat.com>
15773 * doc/invoke.texi (-Wrestrict): Fix typos.
15775 2018-04-02 Jim Wilson <jimw@sifive.com>
15777 * config/riscv/riscv.h (SHIFT_COUNT_TRUNCATED): Set to zero.
15778 * config/riscv/riscv.md (<optab>si3): Use QImode shift count.
15779 (<optab>di3, <optab>si3_extend): Likewise.
15780 (<optab>si3_mask, <optab>si3_mask_1): New.
15781 (<optab>di3_mask, <optab>di3_mask_1): New.
15782 (<optab>si3_extend_mask, <optab>si3_extend_mask_1): New.
15783 (lshrsi3_zero_extend_1): Use VOIDmode shift count.
15784 * config/riscv/sync.md (atomic_test_and_set): Emit QImode shift count.
15786 2018-04-02 Gerald Pfeifer <gerald@pfeifer.com>
15788 * doc/cpp.texi (Variadic Macros): Fix line continuation in an
15791 2018-04-02 Chung-Ju Wu <jasonwucj@gmail.com>
15793 * config/nds32/nds32.c (TARGET_CANONICALIZE_COMPARISON): Define.
15794 (nds32_canonicalize_comparison): New function.
15796 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
15797 Kito Cheng <kito.cheng@gmail.com>
15798 Kuan-Lin Chen <kuanlinchentw@gmail.com>
15800 * config.gcc (nds32): Add nds32-relax-opt.o into extra_objs.
15801 * config/nds32/constants.md (unspec_volatile_element): Add
15802 UNSPEC_VOLATILE_RELAX_GROUP.
15803 * config/nds32/nds32-relax-opt.c: New file.
15804 * config/nds32/nds32-predicates.c
15805 (nds32_symbol_load_store_p): New function.
15806 * config/nds32/nds32-protos.h
15807 (nds32_symbol_load_store_p): Declare function.
15808 (make_pass_nds32_relax_opt): Declare new rtl pass function.
15809 * config/nds32/nds32.c
15810 (nds32_register_pass): New function to register pass.
15811 (nds32_register_passes): New function to register passes.
15812 * config/nds32/nds32.md (relax_group): New pattern.
15813 * config/nds32/nds32.opt (mrelax-hint): New option.
15814 * config/nds32/t-nds32 (nds32-relax-opt.o): New dependency.
15816 2018-04-01 Kito Cheng <kito.cheng@gmail.com>
15818 * config/nds32/t-nds32: Modify files dependency.
15820 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
15822 * config/nds32/nds32.h (FUNCTION_PROFILER): Output newline character.
15823 (PROFILE_HOOK): Define its implementation.
15825 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
15827 * config/nds32/nds32.h (WCHAR_TYPE, WCHAR_TYPE_SIZE): Use unsigned int
15828 type and 32-bit size.
15830 2018-04-01 Jakub Jelinek <jakub@redhat.com>
15832 PR middle-end/85090
15833 * config/i386/sse.md (V): Add V64QI and V32HI for TARGET_AVX512F.
15834 (V_128_256): New mode iterator.
15835 (*avx512dq_vextract<shuffletype>64x2_1 splitter): New define_split.
15836 (*avx512f_vextract<shuffletype>32x4_1 splitter): Likewise.
15837 (xop_pcmov_<mode><avxsizesuffix>): Use V_128_256 mode iterator instead
15839 * config/i386/i386.c (ix86_expand_vector_set): Improve V32HImode and
15840 V64QImode expansion for !TARGET_AVX512BW && TARGET_AVX512F.
15842 2018-03-31 Segher Boessenkool <segher@kernel.crashing.org>
15845 * config/rs6000/xmmintrin.h (_mm_set_ps, _mm_max_ps): Handle (quiet)
15846 NaN inputs correctly.
15848 2018-03-30 Peter Bergner <bergner@vnet.ibm.com>
15851 * config/rs6000/vsx.md (??r): New mode attribute.
15852 (*vsx_mov<mode>_64bit): Use it.
15853 (*vsx_mov<mode>_32bit): Likewise.
15855 2018-03-30 Martin Sebor <msebor@redhat.com>
15857 PR tree-optimization/84818
15858 * builtins.c (check_access): Use warning_n.
15860 2018-03-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
15863 * config/rs6000/rs6000-string.c (expand_compare_loop): Fix redundant
15865 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Fix redundant
15868 2018-03-30 Julia Koval <julia.koval@intel.com>
15871 * x86-tune.def (movx, partial_reg_dependency): Enable for
15874 2018-03-29 Vladimir Makarov <vmakarov@redhat.com>
15876 PR inline-asm/84985
15877 * lra-constraints.c (process_alt_operands): Move setting
15878 this_alternative_matches below.
15880 2018-03-29 Martin Liska <mliska@suse.cz>
15883 * doc/invoke.texi: Document how LTO works with debug info.
15884 Describe auto-load support of binutils. Mention 'x86-64'
15885 as valid option value of -march option.
15887 2018-03-29 Jakub Jelinek <jakub@redhat.com>
15889 * config/i386/sse.md (<avx512>_blendm<mode>): Use <sseintprefix>.
15892 * fold-const.c (operand_equal_p): Handle DEBUG_BEGIN_STMT.
15893 For STATEMENT_LIST, pass down OEP_LEXICOGRAPHIC and maybe
15894 OEP_NO_HASH_CHECK for recursive call, to avoid exponential
15897 2018-03-28 Peter Bergner <bergner@vnet.ibm.com>
15900 * config/rs6000/rs6000.h (RS6000_BTM_POWERPC64): New define.
15901 (RS6000_BTM_COMMON): Add RS6000_BTM_POWERPC64.
15902 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add support
15903 for RS6000_BTM_POWERPC64.
15904 (rs6000_invalid_builtin): Add handling for RS6000_BTM_POWERPC64
15905 (rs6000_builtin_mask_names): Add RS6000_BTM_POWERPC64.
15906 * config/rs6000/rs6000-builtin.def (BU_P7_POWERPC64_MISC_2): New macro
15909 (DIVDEU): Likewise.
15911 2018-03-28 Carl Love <cel@us.ibm.com>
15914 2017-09-27 Carl Love <cel@us.ibm.com>
15916 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
15917 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
15918 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
15921 2018-03-28 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15923 * config/rs6000/xmmintrin.h (_mm_max_pi16): Use __vector __bool
15924 instead of __vector bool.
15925 (_mm_max_pu8): Likewise.
15926 (_mm_min_pi16): Likewise.
15928 2018-03-28 Peter Bergner <bergner@vnet.ibm.com>
15931 * config/rs6000/rs6000-builtin.def (DIVWEO): Delete macro expansion.
15932 (DIVWEUO): Likewise.
15933 (DIVDEO): Likewise.
15934 (DIVDEUO): Likewise.
15935 * config/rs6000/rs6000.c (builtin_function_type): Remove support for
15936 DIVWEUO and DIVDEUO.
15937 * config/rs6000/rs6000.md (UNSPEC_DIVEO, UNSPEC_DIVEUO): Delete unspecs.
15938 (UNSPEC_DIV_EXTEND): Remove deleted unspecs.
15939 (div_extend): Likewise.
15940 * doc/extend.texi (__builtin_divweo): Remove documentation for deleted
15942 (__builtin_divweuo): Likewise.
15943 (__builtin_divdeo): Likewise.
15944 (__builtin_divdeuo): Likewise.
15946 2018-03-28 Jakub Jelinek <jakub@redhat.com>
15949 * config/i386/i386.md (*add<mode>3_carry_0, *addsi3_carry_zext_0,
15950 *sub<mode>3_carry_0, *subsi3_carry_zext_0): New patterns.
15952 PR tree-optimization/82004
15953 * gimple-match-head.c (optimize_pow_to_exp): New function.
15954 * match.pd (pow(C,x) -> exp(log(C)*x)): Wrap with #if GIMPLE.
15955 Don't fold to exp if optimize_pow_to_exp is false.
15957 2018-03-28 Martin Liska <mliska@suse.cz>
15960 * calls.c (initialize_argument_information): Fix trailing space.
15961 * common.opt: Fix typo and provide better explanation for
15962 -fsanitize-coverage option.
15963 * config/i386/i386.opt: Fix typo.
15965 2018-03-28 Jakub Jelinek <jakub@redhat.com>
15966 Martin Liska <mliska@suse.cz>
15969 * gimplify.c (asan_poison_variable): Don't do the check for
15970 gimplify_omp_ctxp here.
15971 (gimplify_decl_expr): Do it here.
15972 (gimplify_target_expr): Likewise.
15974 2018-03-28 Martin Liska <mliska@suse.cz>
15977 * config/i386/i386.c (ix86_function_arg_advance): Do not call
15978 chkp_type_bounds_count if MPX is not enabled.
15980 2018-03-27 Chung-Ju Wu <jasonwucj@gmail.com>
15982 * config/nds32/nds32.h (BRANCH_COST): Adjust cost.
15984 2018-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
15987 * config/rs6000/rs6000.c (create_complex_muldiv): New helper
15988 function to create the function decl for complex long double
15989 multiply and divide for -mabi=ieeelongdouble.
15990 (init_float128_ieee): Call it.
15992 2018-03-27 H.J. Lu <hongjiu.lu@intel.com>
15995 * config/i386/i386.c (ix86_trampoline_init): Insert ENDBR for
15996 -fcf-protection=branch -mibt.
15997 * config/i386/i386.h (TRAMPOLINE_SIZE): Increased by 4 bytes.
15999 2018-03-27 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
16002 * config/arm/arm.c (arm_valid_symbolic_address): Handle
16003 arm_word_relocations.
16005 2018-03-27 Cesar Philippidis <cesar@codesourcery.com>
16008 * config/nvptx/nvptx.c (nvptx_assemble_decl_begin): Add '[]' to
16009 extern array declarations.
16011 2018-03-27 Richard Biener <rguenther@suse.de>
16013 PR middle-end/84067
16014 * match.pd ((A * C) +- (B * C) -> (A+-B) * C): Guard with
16015 explicit single_use checks.
16017 2018-03-27 Richard Biener <rguenther@suse.de>
16019 PR tree-optimization/85082
16020 * tree-ssa-sccvn.c (vn_reference_lookup_or_insert_for_pieces):
16023 2018-03-27 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
16025 * config.gcc (aarch64*-*-linux*): New TARGET_DEFAULT_ASYNC_UNWIND_TABLES
16026 * common/config/aarch64/aarch64-common.c (aarch64_optimization_table[]):
16027 Turn on fasynchronous-unwind-tables and funwind-tables.
16029 2018-03-26 Uros Bizjak <ubizjak@gmail.com>
16032 * config/i386/i386.md (*bmi_blsr_<mode>_cmp): New insn pattern.
16033 (*bmi_blsr_<mode>_ccz): Ditto.
16035 2018-03-26 Tom de Vries <tom@codesourcery.com>
16037 PR tree-optimization/85063
16038 * omp-general.c (offloading_function_p): New function. Factor out
16040 * omp-offload.c (pass_omp_target_link::gate): ... here.
16041 * omp-general.h (offloading_function_p): Declare.
16042 * tree-switch-conversion.c (build_one_array): Mark CSWTCH.x variable
16043 with attribute omp declare target for offloading functions.
16045 2018-03-24 Richard Sandiford <richard.sandiford@linaro.org>
16047 PR tree-optimization/84005
16048 * tree-data-ref.h (get_base_for_alignment): Declare.
16049 * tree-data-ref.c (get_base_for_alignment_1): New function.
16050 (get_base_for_alignment): Likewise.
16051 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
16052 get_base_for_alignment to find a suitable base object, instead
16053 of always using drb->base_address.
16055 2018-03-23 Jakub Jelinek <jakub@redhat.com>
16057 PR inline-asm/85022
16058 * emit-rtl.c (init_emit_regs): Indicate that VOIDmode MEMs don't have
16059 known size by default.
16061 2018-03-23 Vladimir Makarov <vmakarov@redhat.com>
16063 PR inline-asm/85030
16064 * lra-constraints.c (process_alt_operands): Don't match BLKmode
16065 and non BLKmode operands.
16067 2018-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16070 * config/arm/arm.md (unaligned_loadhis): Remove first alternative.
16071 Clean up attributes.
16073 2018-03-23 Richard Biener <rguenther@suse.de>
16076 * dwarf2out.c (rtl_for_decl_location): Do not generate RTL early when
16077 we are going to emit early debug for LTO.
16079 2018-03-23 Jakub Jelinek <jakub@redhat.com>
16081 PR inline-asm/85034
16082 * function.c (match_asm_constraints_1): Don't optimize if input
16083 doesn't satisfy general_operand predicate for output's mode.
16085 PR inline-asm/85022
16086 * alias.c (write_dependence_p): Don't require for x_canonicalized
16087 non-VOIDmode if x has VOIDmode.
16090 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): If DECL_REGISTER (base),
16091 just don't try to optimize it rather than assert it never happens.
16093 2018-03-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
16095 * config/rs6000/rs6000-builtin.def: Remove various BU_ALTIVEC_X
16096 macro expansions for definition of ST_INTERNAL_<mode> and
16097 LD_INTERNAL_<mode> builtins.
16098 * config/rs6000/rs6000-protos.h (rs6000_address_for_altivec):
16100 * config/rs6000/rs6000.c (altivec_expand_ld_builtin): Delete this
16102 (altivec_expand_st_builtin): Likewise.
16103 (altivec_expand_builtin): Remove calls to deleted functions.
16104 (rs6000_address_for_altivec): Delete this function.
16105 * config/rs6000/vector.md: Remove expands for
16106 vector_altivec_load_<mode> and vector_altivec_store_<mode>.
16108 2018-03-22 Sudakshina Das <sudi.das@arm.com>
16111 * config/arm/arm.h (machine_function): Add static_chain_stack_bytes.
16112 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Avoid
16113 re-computing once computed.
16114 (arm_expand_prologue): Compute machine->static_chain_stack_bytes.
16115 (arm_init_machine_status): Initialize
16116 machine->static_chain_stack_bytes.
16118 2018-03-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
16121 * doc/extend.texi: Add four new prototypes for vec_ld.
16122 * config/rs6000/rs6000-builtin.def (LVX_V1TI): Reorder symbol
16123 definitions for more logical presentation.
16124 * config/rs6000/rs6000-c.c: (altivec_overloaded_builtins): Add
16125 entries for V1TI variants of __builtin_altivec_ld builtin.
16126 * config/rs6000/rs6000.c: (altivec_expand_lv_builtin): Add test and
16127 handling of V1TI variant of LVX icode pattern.
16128 (altivec_expand_builtin): Add case for ALTIVEC_BUILTIN_LVX_V1TI.
16129 (rs6000_gimple_fold_builtin): Likewise.
16130 (altivec_init_builtins): Add code to define
16131 __builtin_altivec_lvx_v1ti function.
16133 2018-03-22 Jakub Jelinek <jakub@redhat.com>
16135 PR inline-asm/84941
16136 * function.c (match_asm_constraints_1): Don't do the optimization
16137 if input isn't a REG, SUBREG, MEM or constant.
16139 2018-03-22 Tom de Vries <tom@codesourcery.com>
16141 PR tree-optimization/84956
16142 * tree-ssa-tail-merge.c (find_clusters_1): Skip bbs with
16143 bb_has_abnormal_pred.
16145 2018-03-22 Jakub Jelinek <jakub@redhat.com>
16148 * dwarf2asm.c (dw2_output_indirect_constant_1): Set
16149 DECL_INITIAL (decl) to decl at the end.
16150 * varasm.c (use_blocks_for_decl_p): Revert the 2018-03-20 change,
16151 adjust the comment.
16153 2018-03-21 Joseph Myers <joseph@codesourcery.com>
16155 * doc/extend.texi (__builtin_tgmath): Document when complex
16156 integer types are treated as _Complex _Float64.
16158 2018-03-21 Tom de Vries <tom@codesourcery.com>
16160 * doc/extend.texi (__builtin_extend_pointer): Remove pasto.
16162 2018-03-21 Jakub Jelinek <jakub@redhat.com>
16164 PR tree-optimization/84960
16165 * tree-cfg.c (remove_bb): Don't move forced labels into bb->prev_bb
16166 if it is ENTRY block, move them into single succ of ENTRY in that case.
16168 2018-03-21 Richard Sandiford <richard.sandiford@linaro.org>
16170 PR tree-optimization/84811
16171 * poly-int.h (poly_span_traits): Remove the T3 parameter and
16172 promote HOST_WIDE_INT T2 - T1 results to unsigned HOST_WIDE_INT.
16173 (maybe_in_range_p, known_in_range_p, ranges_known_overlap_p):
16174 (known_subrange_p): Update accordingly. Cast each value involved
16175 in the size comparison, rather than casting the result of the
16178 2018-03-21 Jakub Jelinek <jakub@redhat.com>
16180 PR tree-optimization/84982
16181 * gimple-ssa-store-merging.c (invert_op): Handle boolean inversion
16182 by flipping the least significant bit rather than all bits from
16183 bitpos to bitpos + bitsize - 1.
16185 2018-03-21 Nathan Sidwell <nathan@acm.org>
16187 * doc/extend.texi (Deprecated Features): Remove mention of
16188 long-deleted deprecations.
16190 2018-03-21 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
16193 * configure.ac (gcc_cv_ld_soname) <*-*-solaris2*>: Set.
16194 * configure: Regenerate.
16196 2018-03-21 Tom de Vries <tom@codesourcery.com>
16198 PR tree-optimization/83126
16199 * tree-parloops.c (num_phis): New function.
16200 (gen_parallel_loop): Detect and handle canonicalize_loop_ivs failure.
16202 2018-03-21 Nathan Sidwell <nathan@acm.org>
16204 * doc/extend.texi (Deprecated Features): Update deprecated flags,
16205 mention anon-struct/union members and trailing attributes.
16207 2018-03-21 Bin Cheng <bin.cheng@arm.com>
16209 PR tree-optimization/84969
16210 * tree-loop-distribution.c (fuse_memset_builtins): Don't reorder
16211 builtin memset partitions if they set different rhs values.
16213 2018-03-21 Jakub Jelinek <jakub@redhat.com>
16215 PR rtl-optimization/84989
16216 * simplify-rtx.c (simplify_unary_operation_1): Don't try to simplify
16217 VEC_DUPLICATE with scalar result mode.
16219 2018-03-21 Martin Liska <mliska@suse.cz>
16222 * ipa-icf.c (sem_item_optimizer::fixup_points_to_sets): Remove
16223 not intended return statement.
16225 2018-03-21 Martin Liska <mliska@suse.cz>
16228 * tree-chkp.c (CHKP_ARRAY_MAX_CHECK_STEPS): Define a new macro.
16229 (chkp_find_bound_slots_1): Limit number of iterations.
16231 2018-03-20 David H. Gutteridge <dhgutteridge@sympatico.ca>
16234 * Minor grammar fixes for x86 options.
16236 2018-03-20 Jakub Jelinek <jakub@redhat.com>
16239 * dce.c (delete_unmarked_insns): Don't remove frame related noop moves
16240 holding REG_CFA_RESTORE notes, instead turn them into a USE.
16242 2018-03-20 Peter Bergner <bergner@vnet.ibm.com>
16245 * config/rs6000/altivec.md (altivec_lvx_<mode>_2op): Delete define_insn.
16246 (altivec_lvx_<mode>_1op): Likewise.
16247 (altivec_stvx_<mode>_2op): Likewise.
16248 (altivec_stvx_<mode>_1op): Likewise.
16249 (altivec_lvx_<VM2:mode>): New define_expand.
16250 (altivec_stvx_<VM2:mode>): Likewise.
16251 (altivec_lvx_<VM2:mode>_2op_<P:mptrsize>): New define_insn.
16252 (altivec_lvx_<VM2:mode>_1op_<P:mptrsize>): Likewise.
16253 (altivec_stvx_<VM2:mode>_2op_<P:mptrsize>): Likewise.
16254 (altivec_stvx_<VM2:mode>_1op_<P:mptrsize>): Likewise.
16255 * config/rs6000/rs6000-p8swap.c (rs6000_gen_stvx): Use new expanders.
16256 (rs6000_gen_lvx): Likewise.
16257 * config/rs6000/rs6000.c (altivec_expand_lv_builtin): Likewise.
16258 (altivec_expand_stv_builtin): Likewise.
16259 (altivec_expand_builtin): Likewise.
16260 * config/rs6000/vector.md: Likewise.
16262 2018-03-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16265 * config/arm/arm.c (arm_array_mode_supported_p): Return false for
16268 2018-03-20 Richard Biener <rguenther@suse.de>
16271 * config/i386/i386.c (ix86_add_stmt_cost): Only cost
16272 sign-conversions as zero, fall back to standard scalar_stmt
16275 2018-03-20 Martin Liska <mliska@suse.cz>
16278 * predict.c (rebuild_frequencies): Handle case when we have
16279 PROFILE_ABSENT, but flag_guess_branch_prob is false.
16281 2018-03-20 Jakub Jelinek <jakub@redhat.com>
16284 * dwarf2asm.c (dw2_output_indirect_constant_1): Temporarily turn off
16285 flag_section_anchors.
16286 * varasm.c (use_blocks_for_decl_p): Remove hack for
16287 dw2_force_const_mem.
16290 * config/aarch64/aarch64.md (*aarch64_reg_<mode>3_neg_mask2): Rename
16292 (*aarch64_<optab>_reg_<mode>3_neg_mask2): ... this. If pseudos can't
16293 be created, use lowpart_subreg of operands[0] rather than operands[0]
16295 (*aarch64_reg_<mode>3_minus_mask): Rename to ...
16296 (*aarch64_ashl_reg_<mode>3_minus_mask): ... this.
16297 (*aarch64_<optab>_reg_di3_mask2): Use const_int_operand predicate
16298 and n constraint instead of aarch64_shift_imm_di and Usd.
16299 (*aarch64_reg_<optab>_minus<mode>3): Rename to ...
16300 (*aarch64_<optab>_reg_minus<mode>3): ... this.
16302 2018-03-20 Sudakshina Das <sudi.das@arm.com>
16305 * config/arm/neon.md (ashldi3_neon): Update ?s for constraints
16306 to favor GPR over NEON registers.
16307 (<shift>di3_neon): Likewise.
16309 2018-03-20 Tom de Vries <tom@codesourcery.com>
16312 * config/nvptx/nvptx.c (nvptx_single): Don't neuter bar.sync.
16313 (nvptx_process_pars): Emit bar.sync asap and alap.
16315 2018-03-20 Tom de Vries <tom@codesourcery.com>
16318 * config/nvptx/nvptx.c (prevent_branch_around_nothing): Also update
16319 seen_label if seen_label is already set.
16321 2018-03-20 Jakub Jelinek <jakub@redhat.com>
16324 * config/i386/i386.c (fold_builtin_cpu): For features above 31
16325 use __cpu_features2 variable instead of __cpu_model.__cpu_features[0].
16326 Use 1U instead of 1. Formatting fixes.
16329 * builtins.c (fold_builtin_strpbrk): For strpbrk(x, "") use type
16330 instead of TREE_TYPE (s1) for the return value.
16332 2018-03-19 Jakub Jelinek <jakub@redhat.com>
16334 PR tree-optimization/84946
16335 * gimple-ssa-store-merging.c (mem_valid_for_store_merging): Compute
16336 bitsize + bitsize in poly_uint64 rather than poly_int64.
16339 * dwarf2asm.c: Include fold-const.c.
16340 (dw2_output_indirect_constant_1): Set DECL_INITIAL (decl) to ADDR_EXPR
16341 of decl rather than decl itself.
16343 PR rtl-optimization/84643
16344 * memmodel.h (enum memmodel): Add MEMMODEL_MAX enumerator.
16346 2018-03-19 Maxim Ostapenko <m.ostapenko@samsung.com>
16349 * dwarf2asm.c (dw2_output_indirect_constant_1): Disable ASan before
16350 calling assemble_variable.
16352 2018-03-19 Sudakshina Das <sudi.das@arm.com>
16355 * config/aarch64/aarch64-simd.md (vec_cmp<mode><v_int_equiv>): Modify
16356 instructions for UNLT, UNLE, UNGT, UNGE, UNEQ, UNORDERED and ORDERED.
16358 2018-03-19 Jim Wilson <jimw@sifive.com>
16361 * config/riscv/riscv.c (riscv_function_arg_boundary): Use
16362 PREFERRED_STACK_BOUNDARY instead of STACK_BOUNDARY.
16363 (riscv_first_stack_step): Likewise.
16364 (riscv_option_override): Use STACK_BOUNDARY instead of
16365 MIN_STACK_BOUNDARY.
16366 * config/riscv/riscv.h (STACK_BOUNDARY): Renamed from
16367 MIN_STACK_BOUNDARY.
16368 (BIGGEST_ALIGNMENT): Set to 128.
16369 (PREFERRED_STACK_BOUNDARY): Renamed from STACK_BOUNDARY.
16370 (RISCV_STACK_ALIGN): Use PREFERRED_STACK_BOUNDARY instead of
16373 2018-03-19 Richard Biener <rguenther@suse.de>
16375 PR tree-optimization/84933
16376 * tree-vrp.c (set_and_canonicalize_value_range): Treat out-of-bound
16377 values as -INF/INF when canonicalizing an ANTI_RANGE to a RANGE.
16379 2018-03-19 Richard Biener <rguenther@suse.de>
16381 PR tree-optimization/84859
16382 * tree-ssa-phiopt.c (single_trailing_store_in_bb): New function.
16383 (cond_if_else_store_replacement): Perform sinking operation on
16384 single-store BBs regardless of MAX_STORES_TO_SINK setting.
16385 Generalize what a BB with a single eligible store is.
16387 2018-03-19 Richard Biener <rguenther@suse.de>
16389 PR tree-optimization/84929
16390 * tree-data-ref.c (analyze_siv_subscript_cst_affine): Guard
16391 chrec_is_positive against non-chrec arg.
16393 2018-03-19 Tamar Christina <tamar.christina@arm.com>
16396 * config/arm/arm.c (arm_can_change_mode_class): revert r258554.
16398 2018-03-18 Martin Liska <mliska@suse.cz>
16400 PR rtl-optimization/84635
16401 * regrename.c (build_def_use): Use matches_mode only when
16404 2018-03-18 Richard Sandiford <richard.sandiford@linaro.org>
16406 PR tree-optimization/84913
16407 * tree-vect-loop.c (vectorizable_reduction): Don't try to
16408 vectorize chains of COND_EXPRs.
16410 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
16412 * config/nds32/nds32.h (MAX_REGS_PER_ADDRESS): Fix the value.
16414 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
16416 * config/nds32/nds32.h (LOGICAL_OP_NON_SHORT_CIRCUIT): Define.
16418 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
16420 * config/nds32/nds32.h (CLZ_DEFINED_VALUE_AT_ZERO): Define.
16422 2018-03-17 Chung-Ju Wu <jasonwucj@gmail.com>
16423 Kito Cheng <kito.cheng@gmail.com>
16425 * config/nds32/nds32-protos.h (nds32_adjust_reg_alloc_order): Declare.
16426 * config/nds32/nds32.c (nds32_reg_alloc_order_for_speed): New array.
16427 (nds32_adjust_reg_alloc_order): New function.
16428 * config/nds32/nds32.h (ADJUST_REG_ALLOC_ORDER): Define.
16430 2018-03-17 Kito Cheng <kito.cheng@gmail.com>
16432 * config/nds32/nds32.c (nds32_asm_output_mi_thunk,
16433 nds32_print_operand, nds32_print_operand_address): Use
16434 HOST_WIDE_INT_PRINT_DEC instead.
16436 2018-03-17 Chung-Ju Wu <jasonwucj@gmail.com>
16438 * config/nds32/nds32.c (nds32_register_priority): Modify cost.
16440 2018-03-17 Jakub Jelinek <jakub@redhat.com>
16443 * config/i386/i386.c (initial_ix86_tune_features,
16444 initial_ix86_arch_features): Use unsigned HOST_WIDE_INT rather than
16445 unsigned long long.
16446 (set_ix86_tune_features): Change ix86_tune_mask from unsigned int
16447 to unsigned HOST_WIDE_INT, initialize to HOST_WIDE_INT_1U << ix86_tune
16448 rather than 1u << ix86_tune. Formatting fix.
16449 (ix86_option_override_internal): Change ix86_arch_mask from
16450 unsigned int to unsigned HOST_WIDE_INT, initialize to
16451 HOST_WIDE_INT_1U << ix86_arch rather than 1u << ix86_arch.
16452 (ix86_function_specific_restore): Likewise.
16454 2018-03-16 Jakub Jelinek <jakub@redhat.com>
16457 * postreload.c (reload_combine_recognize_pattern): Perform
16458 INTVAL addition in unsigned HOST_WIDE_INT type to avoid UB and
16459 truncate_int_for_mode the result for the destination's mode.
16462 * hsa-gen.c (mem_type_for_type): Fix comment typo.
16463 * tree-vect-loop-manip.c (vect_create_cond_for_niters_checks):
16465 * gimple-ssa-warn-restrict.c (builtin_memref::set_base_and_offset):
16468 2018-03-16 Vladimir Makarov <vmakarov@redhat.com>
16471 * lra-assigns.c (lra_split_hard_reg_for): Don't use
16472 regno_allocno_class_array and sorted_pseudos.
16473 * lra-constraints.c (spill_hard_reg_in_range): Ignore hard regs in
16474 insns where regno is used.
16476 2018-03-16 Martin Liska <mliska@suse.cz>
16479 * multiple_target.c (create_dispatcher_calls): Redirect
16480 reference in the symbol table.
16482 2018-03-16 Martin Liska <mliska@suse.cz>
16485 * multiple_target.c (create_dispatcher_calls): Redirect also
16488 2018-03-16 Jakub Jelinek <jakub@redhat.com>
16492 * tree.h (TARGET_EXPR_NO_ELIDE): Define.
16493 * gimplify.c (gimplify_modify_expr_rhs): Don't elide TARGET_EXPRs with
16494 TARGET_EXPR_NO_ELIDE flag set unless *expr_p is INIT_EXPR.
16496 2018-03-16 Julia Koval <julia.koval@intel.com>
16498 * doc/invoke.texi (Skylake Server): Add CLWB.
16499 Cannonlake): Remove CLWB.
16501 2018-03-16 Jakub Jelinek <jakub@redhat.com>
16503 PR tree-optimization/84841
16504 * tree-ssa-reassoc.c (INTEGER_CONST_TYPE): Change to 1 << 4 from
16506 (FLOAT_ONE_CONST_TYPE): Define.
16507 (constant_type): Return FLOAT_ONE_CONST_TYPE for -1.0 and 1.0.
16508 (sort_by_operand_rank): Put entries with higher constant_type last
16509 rather than first to match comments.
16511 2018-03-15 Sandra Loosemore <sandra@codesourcery.com>
16513 * config/nios2/nios2.md (movsi_internal): Fix thinko in
16516 2018-03-15 Jakub Jelinek <jakub@redhat.com>
16519 * calls.c (expand_call): For TREE_ADDRESSABLE rettype ignore alignment
16520 check and use address of target always.
16522 2018-03-15 H.J. Lu <hongjiu.lu@intel.com>
16525 * config/i386/i386.c (indirect_thunk_needed): Update comments.
16526 (indirect_thunk_bnd_needed): Likewise.
16527 (indirect_thunks_used): Likewise.
16528 (indirect_thunks_bnd_used): Likewise.
16529 (indirect_return_needed): New.
16530 (indirect_return_bnd_needed): Likewise.
16531 (output_indirect_thunk_function): Add a bool argument for
16533 (output_indirect_thunk_function): Don't generate alias for
16534 function return thunk.
16535 (ix86_code_end): Call output_indirect_thunk_function to generate
16536 function return thunks.
16537 (ix86_output_function_return): Set indirect_return_bnd_needed
16538 and indirect_return_needed instead of indirect_thunk_bnd_needed
16539 and indirect_thunk_needed.
16541 2018-03-15 Olga Makhotina <olga.makhotina@intel.com>
16543 * config/i386/sgxintrin.h (_enclv_u32): New intrinsic.
16544 (__enclv_bc, __enclv_cd, __enclv_generic): New definitions.
16545 (ERDINFO, ETRACKC, ELDBC, ELDUC): New leaves.
16547 2018-03-15 David Malcolm <dmalcolm@redhat.com>
16548 Paul Hua <paul.hua.gm@gmail.com>
16551 * gcc.dg/fixits-pr84852-1.c: Fix filename in dg-regexp.
16553 2018-03-15 Segher Boessenkool <segher@kernel.crashing.org>
16555 * config/rs6000/rs6000.c (abi_v4_pass_in_fpr): Add back the
16556 TARGET_DOUBLE_FLOAT and TARGET_SINGLE_FLOAT conditions on the DFmode
16557 resp. SFmode cases.
16559 2018-03-15 Tamar Christina <tamar.christina@arm.com>
16562 * config/arm/arm.c (arm_can_change_mode_class): Use GET_MODE_UNIT_SIZE
16563 instead of GET_MODE_SIZE when comparing Units.
16565 2018-03-15 Vladimir Mezentsev <vladimir.mezentsev@oracle.com>
16568 * varasm.c (hash_section): Return an unchangeble hash value
16569 * config/aarch64/aarch64.c (aarch64_use_blocks_for_constant_p):
16570 Return !aarch64_can_use_per_function_literal_pools_p ().
16572 2018-03-15 Jakub Jelinek <jakub@redhat.com>
16575 * optabs.c (emit_conditional_move): Pass address of cmode's copy
16576 rather than address of cmode as last argument to prepare_cmp_insn.
16578 2018-03-15 Julia Koval <julia.koval@intel.com>
16580 * config/i386/i386.c (F_AVX512VBMI2, F_GFNI, F_VPCLMULQDQ,
16581 F_AVX512VNNI, F_AVX512BITALG): New.
16583 2018-03-14 John David Anglin <danglin@gcc.gnu.org>
16586 * config/pa/pa.c (pa_emit_move_sequence): Always emit secondary reload
16587 insn for floating-point loads and stores.
16589 2018-03-14 Carl Love <cel@us.ibm.com>
16591 * config/rs6000/rs6000-c.c: Add macro definitions for
16592 ALTIVEC_BUILTIN_VEC_PERMXOR.
16593 * config/rs6000/rs6000.h: Add #define for vec_permxor builtin.
16594 * config/rs6000/rs6000-builtin.def: Add macro expansions for VPERMXOR.
16595 * config/rs6000/altivec.md (altivec_vpermxor): New define expand.
16596 * config/rs6000/rs6000-p8swap.c (rtx_is_swappable_p): Add case
16598 * config/doc/extend.texi: Add prototypes for vec_permxor.
16600 2018-03-14 David Malcolm <dmalcolm@redhat.com>
16603 * diagnostic-show-locus.c (class layout_point): Convert m_line
16604 from int to linenum_type.
16605 (line_span::comparator): Use linenum "compare" function when
16606 comparing line numbers.
16607 (test_line_span): New function.
16608 (layout_range::contains_point): Convert param "row" from int to
16610 (layout_range::intersects_line_p): Likewise.
16611 (layout::will_show_line_p): Likewise.
16612 (layout::print_source_line): Likewise.
16613 (layout::should_print_annotation_line_p): Likewise.
16614 (layout::print_annotation_line): Likewise.
16615 (layout::print_leading_fixits): Likewise.
16616 (layout::annotation_line_showed_range_p): Likewise.
16617 (struct line_corrections): Likewise for field m_row.
16618 (line_corrections::line_corrections): Likewise for param "row".
16619 (layout::print_trailing_fixits): Likewise.
16620 (layout::get_state_at_point): Likewise.
16621 (layout::get_x_bound_for_row): Likewise.
16622 (layout::print_line): Likewise.
16623 (diagnostic_show_locus): Likewise for locals "last_line" and "row".
16624 (selftest::diagnostic_show_locus_c_tests): Call test_line_span.
16625 * input.c (selftest::test_linenum_comparisons): New function.
16626 (selftest::input_c_tests): Call it.
16627 * selftest.c (selftest::test_assertions): Test ASSERT_GT,
16628 ASSERT_GT_AT, ASSERT_LT, and ASSERT_LT_AT.
16629 * selftest.h (ASSERT_GT): New macro.
16630 (ASSERT_GT_AT): New macro.
16631 (ASSERT_LT): New macro.
16632 (ASSERT_LT_AT): New macro.
16634 2018-03-14 Segher Boessenkool <segher@kernel.crashing.org>
16636 PR rtl-optimization/84780
16637 * combine.c (distribute_links): Don't make a link based on pc_rtx.
16639 2018-03-14 Martin Liska <mliska@suse.cz>
16641 * tree.c (record_node_allocation_statistics): Use
16642 get_stats_node_kind.
16643 (get_stats_node_kind): New function extracted from
16644 record_node_allocation_statistics.
16645 (free_node): Use get_stats_node_kind.
16647 2018-03-14 Richard Biener <rguenther@suse.de>
16649 * tree-ssa-pre.c (compute_antic_aux): Remove code that asserts
16650 that the value-set of ANTIC_IN doesn't grow.
16653 * tree-ssa-pre.c (struct bb_bitmap_sets): Add visited_with_visited_succs
16655 (BB_VISITED_WITH_VISITED_SUCCS): New define.
16656 (compute_antic): Initialize BB_VISITED_WITH_VISITED_SUCCS.
16658 2018-03-14 Julia Koval <julia.koval@intel.com>
16660 * config.gcc (icelake-client, icelake-server): New.
16662 * config/i386/i386.c (initial_ix86_tune_features): Extend to 64 bit.
16663 (initial_ix86_arch_features): Ditto.
16664 (PTA_SKYLAKE): Add SGX.
16665 (PTA_ICELAKE): Remove.
16666 (PTA_ICELAKE_CLIENT): New.
16667 (PTA_ICELAKE_SERVER): New.
16668 (ix86_option_override_internal): Split up icelake on icelake client and
16670 (get_builtin_code_for_version): Ditto.
16671 (fold_builtin_cpu): Ditto.
16672 * config/i386/driver-i386.c (config/i386/driver-i386.c): Ditto.
16673 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto
16674 * config/i386/i386.h (processor_type): Ditto.
16675 * doc/invoke.texi: Ditto.
16677 2018-03-14 Jakub Jelinek <jakub@redhat.com>
16680 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): Handle also
16681 INTEGER_CST offset, add it together with bitpos / 8 and
16682 sign extend based on POINTER_SIZE.
16686 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
16689 * config/i386/constraints.md (Yc): New register constraint.
16690 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
16691 Use Yc constraint for alternative 2 of operand 0. Remove
16692 preferred_for_speed attribute.
16694 2018-03-14 Richard Biener <rguenther@suse.de>
16696 PR tree-optimization/84830
16697 * tree-ssa-pre.c (compute_antic_aux): Intersect the new ANTIC_IN
16698 with the old one to avoid oscillations.
16700 2018-03-13 Vladimir Makarov <vmakarov@redhat.com>
16703 * lra-assigns.c (find_all_spills_for): Ignore uninteresting
16705 (assign_by_spills): Return a flag of reload assignment failure.
16706 Do not process the reload assignment failures. Do not spill other
16707 reload pseudos if they has the same reg class. Update n if
16709 (lra_assign): Add a return arg. Set up from the result of
16710 assign_by_spills call.
16711 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
16712 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
16713 usage_insns if it is not NULL.
16714 (spill_hard_reg_in_range): New function.
16715 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
16716 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
16717 function prototypes.
16718 (lra_assign): Change prototype.
16719 * lra.c (lra): Add code to deal with fails by splitting hard reg
16722 2018-03-01 Palmer Dabbelt <palmer@sifive.com>
16724 * config/riscv/riscv.opt (mrelax): New option.
16725 * config/riscv/riscv.c (riscv_file_start): Emit ".option
16726 "norelax" when riscv_mrelax is disabled.
16727 * doc/invoke.texi (RISC-V): Document "-mrelax" and "-mno-relax".
16729 2018-03-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
16732 * config/rs6000/rs6000.c (rs6000_reassociation_width): Disable parallel
16733 reassociation for int modes.
16735 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
16737 * tree-vect-loop-manip.c (vect_maybe_permute_loop_masks):
16738 Reverse the choice between VEC_UNPACK_LO_EXPR and VEC_UNPACK_HI_EXPR
16740 * config/aarch64/iterators.md (hi_lanes_optab): New int attribute.
16741 * config/aarch64/aarch64-sve.md
16742 (*aarch64_sve_<perm_insn><perm_hilo><mode>): Rename to...
16743 (aarch64_sve_<perm_insn><perm_hilo><mode>): ...this.
16744 (*extend<mode><Vwide>2): Rename to...
16745 (aarch64_sve_extend<mode><Vwide>2): ...this.
16746 (vec_unpack<su>_<perm_hilo>_<mode>): Turn into a define_expand,
16747 renaming the old pattern to...
16748 (aarch64_sve_punpk<perm_hilo>_<mode>): ...this. Only define
16750 (vec_unpack<su>_<perm_hilo>_<SVE_BHSI:mode>): Turn into a
16751 define_expand, renaming the old pattern to...
16752 (aarch64_sve_<su>unpk<perm_hilo>_<SVE_BHSI:mode>): ...this.
16753 (*vec_unpacku_<perm_hilo>_<mode>_no_convert): Delete.
16754 (vec_unpacks_<perm_hilo>_<mode>): Take BYTES_BIG_ENDIAN into
16755 account when deciding which SVE instruction the optab should use.
16756 (vec_unpack<su_optab>_float_<perm_hilo>_vnx4si): Likewise.
16758 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
16760 * config/aarch64/aarch64.md (V4_REGNUM, V8_REGNUM, V12_REGNUM)
16761 (V20_REGNUM, V24_REGNUM, V28_REGNUM, P1_REGNUM, P2_REGNUM, P3_REGNUM)
16762 (P4_REGNUM, P5_REGNUM, P6_REGNUM, P8_REGNUM, P9_REGNUM, P10_REGNUM)
16763 (P11_REGNUM, P12_REGNUM, P13_REGNUM, P14_REGNUM): New define_constants.
16764 (tlsdesc_small_<mode>): Turn a define_expand and use
16765 tlsdesc_small_sve_<mode> for SVE. Rename original define_insn to...
16766 (tlsdesc_small_advsimd_<mode>): ...this.
16767 (tlsdesc_small_sve_<mode>): New pattern.
16769 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
16771 * config/aarch64/iterators.md (UNSPEC_SMUL_HIGHPART)
16772 (UNSPEC_UMUL_HIGHPART): New constants.
16773 (MUL_HIGHPART): New int iteraor.
16774 (su): Handle UNSPEC_SMUL_HIGHPART and UNSPEC_UMUL_HIGHPART.
16775 * config/aarch64/aarch64-sve.md (<su>mul<mode>3_highpart): New
16777 (*<su>mul<mode>3_highpart): New define_insn.
16779 2018-03-13 Eric Botcazou <ebotcazou@adacore.com>
16782 * ipa-devirt.c (odr_subtypes_equivalent_p): Do not get the ODR type of
16785 2018-03-13 Martin Liska <mliska@suse.cz>
16788 * (sem_item_optimizer::sem_item_optimizer): Initialize new
16790 (sem_item_optimizer::~sem_item_optimizer): Release it.
16791 (sem_item_optimizer::merge_classes): Register variable aliases.
16792 (sem_item_optimizer::fixup_pt_set): New function.
16793 (sem_item_optimizer::fixup_points_to_sets): Likewise.
16794 * ipa-icf.h: Declare new variables and functions.
16796 2018-03-13 Jakub Jelinek <jakub@redhat.com>
16798 PR middle-end/84834
16799 * match.pd ((A & C) != 0 ? D : 0): Use INTEGER_CST@2 instead of
16800 integer_pow2p@2 and test integer_pow2p in condition.
16801 (A < 0 ? C : 0): Similarly for @1.
16803 PR middle-end/84831
16804 * stmt.c (parse_output_constraint): If the CONSTRAINT_LEN (*p, p)
16805 characters starting at p contain '\0' character, don't look beyond
16809 * config/i386/i386.md (round<mode>2): For 387 fancy math, disable
16810 pattern if -ftrapping-math -fno-fp-int-builtin-inexact.
16813 * reg-stack.c (change_stack): Change update_end var from int to
16814 rtx_insn *, if non-NULL don't update just BB_END (current_block), but
16815 also call set_block_for_insn on the newly added insns and rescan.
16818 * config/i386/sse.md (sse2_loadhpd): Use Yv constraint rather than v
16819 on the last operand.
16822 * tree.c (stabilize_reference_1): Return save_expr (e) for
16823 STATEMENT_LIST even if it doesn't have side-effects.
16825 2018-03-12 Jonathan Wakely <jwakely@redhat.com>
16827 * doc/invoke.texi (-mclflushopt): Fix spelling of option.
16829 2018-03-12 Renlin Li <renlin.li@arm.com>
16831 * config/aarch64/aarch64.md (movhf_aarch64): Fix mode argument to
16832 aarch64_output_scalar_simd_mov_immediate.
16834 2018-03-12 Martin Sebor <msebor@redhat.com>
16836 PR tree-optimization/83456
16837 * gimple-fold.c (gimple_fold_builtin_memory_op): Avoid warning
16838 for perfectly overlapping calls to memcpy.
16839 (gimple_fold_builtin_memory_chk): Same.
16840 (gimple_fold_builtin_strcpy): Handle no-warning.
16841 (gimple_fold_builtin_stxcpy_chk): Same.
16842 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Handle no-warning.
16844 2018-03-12 Segher Boessenkool <segher@kernel.crashing.org>
16846 * config/rs6000/rs6000.c (abi_v4_pass_in_fpr): Add bool "named"
16847 parameter. Use it for SFmode.
16848 (rs6000_function_arg_advance_1): Adjust.
16849 (rs6000_function_arg): Adjust.
16850 (rs6000_gimplify_va_arg): Pass false for that new parameter.
16852 2018-03-12 Segher Boessenkool <segher@kernel.crashing.org>
16854 PR rtl-optimization/84169
16855 PR rtl-optimization/84780
16856 * combine.c (can_combine_p): Check for a 2-insn combination whether
16857 the destination register is used between the two insns, too.
16859 2018-03-12 Richard Biener <rguenther@suse.de>
16861 PR tree-optimization/84803
16862 * tree-if-conv.c (ifcvt_memrefs_wont_trap): Don't do anything
16863 for refs DR analysis didn't process.
16865 2018-03-12 Richard Biener <rguenther@suse.de>
16867 PR tree-optimization/84777
16868 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): For
16869 force-vectorize loops ignore whether we are optimizing for size.
16871 2018-03-12 Chung-Ju Wu <jasonwucj@gmail.com>
16873 * config/nds32/nds32.c (nds32_md_asm_adjust): New function.
16874 (TARGET_MD_ASM_ADJUST): Define.
16876 2018-03-12 Monk Chiang <sh.chiang04@gmail.com>
16877 Kito Cheng <kito.cheng@gmail.com>
16878 Chung-Ju Wu <jasonwucj@gmail.com>
16880 * config/nds32/nds32.c (nds32_compute_stack_frame,
16881 nds32_emit_stack_push_multiple, nds32_emit_stack_pop_multiple,
16882 nds32_emit_stack_v3push, nds32_emit_stack_v3pop,
16883 nds32_emit_adjust_frame, nds32_expand_prologue, nds32_expand_epilogue,
16884 nds32_expand_prologue_v3push, nds32_expand_epilogue_v3pop): Refine.
16885 * config/nds32/nds32.h (NDS32_FIRST_CALLEE_SAVE_GPR_REGNUM,
16886 NDS32_LAST_CALLEE_SAVE_GPR_REGNUM, NDS32_V3PUSH_AVAILABLE_P): New.
16887 * config/nds32/nds32.md (prologue, epilogue): Use macro
16888 NDS32_V3PUSH_AVAILABLE_P to do checking.
16890 2018-03-11 Jakub Jelinek <jakub@redhat.com>
16893 * dwarf2out.c (gen_enumeration_type_die): Don't guard adding
16894 DW_AT_declaration for ENUM_IS_OPAQUE on -gdwarf-4 or -gno-strict-dwarf,
16895 but on TYPE_SIZE. Don't do anything for ENUM_IS_OPAQUE if not creating
16896 a new die. Don't set TREE_ASM_WRITTEN if ENUM_IS_OPAQUE. Guard
16897 addition of most attributes on !orig_type_die or the attribute not
16898 being present already. Assert TYPE_VALUES is NULL for ENUM_IS_OPAQUE.
16900 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
16901 Chung-Ju Wu <jasonwucj@gmail.com>
16903 * config/nds32/nds32.c (nds32_cpu_cpp_builtins): Modify to define
16904 __NDS32_VH__ macro.
16905 * config/nds32/nds32.opt (mvh): New option.
16907 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
16908 Chung-Ju Wu <jasonwucj@gmail.com>
16910 * config/nds32/nds32-protos.h (nds32_cpu_cpp_builtins): Declare
16912 * config/nds32/nds32.c (nds32_cpu_cpp_builtins): New function.
16913 * config/nds32/nds32.h (TARGET_CPU_CPP_BUILTINS): Modify its
16916 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
16917 Chung-Ju Wu <jasonwucj@gmail.com>
16919 * config/nds32/nds32-memory-manipulation.c (nds32_expand_strlen): New
16921 * config/nds32/nds32-multiple.md (strlensi): New pattern.
16922 * config/nds32/nds32-protos.h (nds32_expand_strlen): Declare function.
16924 2018-03-11 Monk Chiang <sh.chiang04@gmail.com>
16925 Kito Cheng <kito.cheng@gmail.com>
16926 Chung-Ju Wu <jasonwucj@gmail.com>
16928 * config/nds32/constants.md (unspec_element): Add UNSPEC_FFB,
16929 UNSPEC_FFMISM and UNSPEC_FLMISM.
16930 * config/nds32/nds32-intrinsic.c (bdesc_2arg): Add builtin description
16931 for ffb, ffmism and flmism.
16932 * config/nds32/nds32-intrinsic.md (unspec_ffb): Define new pattern.
16933 (unspec_ffmism): Ditto.
16934 (unspec_flmism): Ditto.
16935 (nds32_expand_builtin_impl): Check if string extension is available.
16936 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_FFB,
16937 NDS32_BUILTIN_FFMISM and NDS32_BUILTIN_FLMISM.
16939 2018-03-10 Vladimir Makarov <vmakarov@redhat.com>
16942 2018-03-09 Vladimir Makarov <vmakarov@redhat.com>
16945 * lra-assigns.c (assign_by_spills): Return a flag of reload
16946 assignment failure. Do not process the reload assignment
16947 failures. Do not spill other reload pseudos if they has the same
16949 (lra_assign): Add a return arg. Set up from the result of
16950 assign_by_spills call.
16951 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
16952 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
16953 usage_insns if it is not NULL.
16954 (spill_hard_reg_in_range): New function.
16955 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
16956 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
16957 function prototypes.
16958 (lra_assign): Change prototype.
16959 * lra.c (lra): Add code to deal with fails by splitting hard reg
16962 2018-03-10 H.J. Lu <hongjiu.lu@intel.com>
16965 * config/i386/i386.opt: Replace Enforcment with Enforcement.
16967 2018-03-10 Alexandre Oliva <aoliva@redhat.com>
16970 * dwarf2out.h (dw_val_class): Add dw_val_class_symview.
16971 (dw_val_node): Add val_symbolic_view.
16972 * dwarf2out.c (dw_line_info_table): Add symviews_since_reset.
16973 (symview_upper_bound): New.
16974 (new_line_info_table): Initialize symviews_since_reset.
16975 (dwarf2out_source_line): Count symviews_since_reset and set
16976 symview_upper_bound.
16977 (dw_val_equal_p): Handle symview.
16978 (add_AT_symview): New.
16979 (print_dw_val): Handle symview.
16980 (attr_checksum, attr_checksum_ordered): Likewise.
16981 (same_dw_val_p, size_of_die): Likewise.
16982 (value_format, output_die): Likewise.
16983 (add_high_low_attributes): Use add_AT_symview for entry_view.
16984 (dwarf2out_finish): Reset symview_upper_bound, clear
16987 2018-03-09 Peter Bergner <bergner@vnet.ibm.com>
16990 * config/rs6000/rs6000.c (rs6000_offsettable_memref_p): New prototype.
16991 Add strict argument and use it.
16992 (rs6000_split_multireg_move): Update for new strict argument.
16993 (mem_operand_gpr): Disallow all non-offsettable addresses.
16994 * config/rs6000/rs6000.md (*movdi_internal64): Use YZ constraint.
16996 2018-03-09 Jakub Jelinek <jakub@redhat.com>
16999 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Mark va_arg_tmp
17000 temporary TREE_ADDRESSABLE before gimplification of BUILT_IN_MEMCPY.
17001 * config/powerpcspe/powerpcspe.c (rs6000_gimplify_va_arg): Likewise.
17004 * tree-inline.c (copy_tree_body_r): For INDIRECT_REF of a remapped
17005 decl, use remap_type if we want to use the type.
17007 2018-03-09 Martin Sebor <msebor@redhat.com>
17009 PR tree-optimization/84526
17010 * gimple-ssa-warn-restrict.c (builtin_memref::set_base_and_offset):
17012 (builtin_access::generic_overlap): Be prepared to handle non-array
17015 2018-03-09 Alexandre Oliva <aoliva@redhat.com>
17017 PR rtl-optimization/84682
17018 * lra-constraints.c (process_address_1): Check is_address flag
17019 for address constraints.
17020 (process_alt_operands): Likewise.
17021 * lra.c (lra_set_insn_recog_data): Pass asm operand locs to
17022 preprocess_constraints.
17023 * recog.h (preprocess_constraints): Add oploc parameter.
17025 * recog.c (preprocess_constraints): Test address_operand for
17026 CT_ADDRESS constraints.
17028 2018-03-09 Vladimir Makarov <vmakarov@redhat.com>
17031 * lra-assigns.c (assign_by_spills): Return a flag of reload
17032 assignment failure. Do not process the reload assignment
17033 failures. Do not spill other reload pseudos if they has the same
17035 (lra_assign): Add a return arg. Set up from the result of
17036 assign_by_spills call.
17037 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
17038 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
17039 usage_insns if it is not NULL.
17040 (spill_hard_reg_in_range): New function.
17041 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
17042 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
17043 function prototypes.
17044 (lra_assign): Change prototype.
17045 * lra.c (lra): Add code to deal with fails by splitting hard reg
17048 2018-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17051 * common/config/arm/arm-common.c (arm_parse_arch_option_name):
17052 Accept complain bool parameter. Only emit errors if it is true.
17053 (arm_parse_cpu_option_name): Likewise.
17054 (arm_target_thumb_only): Adjust callers of the above.
17055 * config/arm/arm-protos.h (arm_parse_cpu_option_name): Adjust
17056 prototype to take a default true bool parameter.
17057 (arm_parse_arch_option_name): Likewise.
17059 2018-03-09 David Malcolm <dmalcolm@redhat.com>
17060 Francois-Xavier Coudert <fxcoudert@gcc.gnu.org>
17064 * Makefile.in (LD_VERSION_SCRIPT_OPTION, LD_SONAME_OPTION): New.
17065 * configure: Regenerate.
17066 * configure.ac ("linker --version-script option"): New.
17067 ("linker soname option"): New.
17069 2018-03-09 Richard Biener <rguenther@suse.de>
17071 PR tree-optimization/84775
17072 * tree-if-conv.c (add_bb_predicate_gimplified_stmts): Delink
17073 immediate uses of predicate stmts and mark them modified.
17076 PR tree-optimization/84178
17077 * tree-if-conv.c (combine_blocks): Move insert_gimplified_predicates
17079 (version_loop_for_if_conversion): Delay update_ssa call.
17080 (tree_if_conversion): Delay update_ssa until after predicate
17083 2018-03-09 Eric Botcazou <ebotcazou@adacore.com>
17086 * config/i386/winnt.c (i386_pe_seh_cold_init): Use small pre-allocation
17087 when the function accesses prior frames.
17089 2018-03-08 Jakub Jelinek <jakub@redhat.com>
17092 * dwarf2out.c (dw_loc_list): If list && loc_list->first->next, call
17093 gen_llsym, otherwise call maybe_gen_llsym.
17095 PR inline-asm/84742
17096 * recog.c (asm_operand_ok): Return 0 if multi-character constraint
17097 has ',' character inside of it.
17099 2018-03-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17102 * config/aarch64/aarch64.md (*compare_cstore<mode>_insn): Mark pattern
17103 as clobbering CC_REGNUM.
17105 2018-03-08 Richard Biener <rguenther@suse.de>
17107 PR middle-end/84552
17108 * tree-scalar-evolution.c: Include tree-into-ssa.h.
17109 (follow_copies_to_constant): Do not follow SSA names registered
17112 2018-03-08 Richard Biener <rguenther@suse.de>
17114 PR tree-optimization/84178
17115 * tree-if-conv.c (combine_blocks): Move insert_gimplified_predicates
17117 (version_loop_for_if_conversion): Delay update_ssa call.
17118 (tree_if_conversion): Delay update_ssa until after predicate
17121 2018-03-08 David Malcolm <dmalcolm@redhat.com>
17123 PR tree-optimization/84178
17124 * tree-if-conv.c (release_bb_predicate): Remove the
17125 the assertion that the stmts have NULL use_ops.
17126 Discard the statements, asserting that they haven't
17127 yet been added to a BB.
17129 2018-03-08 Richard Biener <rguenther@suse.de>
17131 PR tree-optimization/84746
17132 * tree-ssa-pre.c (find_leader_in_sets): Deal with SET1 being NULL.
17133 (phi_translate): Pass in destination ANTIC_OUT set.
17134 (phi_translate_1): Likewise. For a simplified result lookup
17135 a leader in ANTIC_OUT and AVAIL_OUT, not the ANTIC_IN sets.
17136 (phi_translate_set): Adjust.
17137 (do_pre_regular_insertion): Likewise.
17138 (do_pre_partial_partial_insertion): Likewise.
17140 2018-03-08 Martin Liska <mliska@suse.cz>
17142 PR gcov-profile/84735
17143 * doc/gcov.texi: Document usage of profile files.
17144 * gcov-io.h: Document changes in the format.
17146 2018-03-08 Alexandre Oliva <aoliva@redhat.com>
17150 * dwarf2out.c (struct dw_line_info_table): Update comments for
17152 (FORCE_RESET_NEXT_VIEW): New.
17153 (FORCE_RESETTING_VIEW_P): New.
17154 (RESETTING_VIEW_P): Check for -1 too.
17155 (ZERO_VIEW_P): Likewise.
17156 (new_line_info_table): Force-reset next view.
17157 (dwarf2out_begin_function): Likewise.
17158 (dwarf2out_source_line): Simplify zero_view_p initialization.
17159 Test FORCE_RESETTING_VIEW_P and RESETTING_VIEW_P instead of
17160 view directly. Omit view when omitting .loc at line 0.
17162 2018-03-08 Jakub Jelinek <jakub@redhat.com>
17164 PR tree-optimization/84740
17165 * tree-switch-conversion.c (process_switch): Call build_constructors
17166 only if info.phi_count is non-zero.
17168 PR tree-optimization/84739
17169 * tree-tailcall.c (find_tail_calls): Check call arguments against
17170 DECL_ARGUMENTS (current_function_decl) rather than
17171 DECL_ARGUMENTS (func) when checking for tail recursion.
17173 2018-03-07 Jakub Jelinek <jakub@redhat.com>
17175 * doc/contrib.texi: Add entries for Martin Liska, David Malcolm,
17176 Marek Polacek, extend Vladimir Makarov's, Jonathan Wakely's and
17177 Volker Reichelt's entry and add entries for people that perform
17178 GCC fuzzy testing and report numerous bugs.
17180 2018-03-07 Segher Boessenkool <segher@kernel.crashing.org>
17183 * config/rs6000/rs6000.c (rs6000_elf_in_small_data_p): Don't put
17184 readonly data in sdata, if that is disabled.
17185 * config/rs6000/sysv4.opt (mreadonly-in-sdata): New option.
17186 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
17187 -mreadonly-in-sdata option.
17189 2018-03-07 Martin Sebor <msebor@redhat.com>
17191 PR tree-optimization/84468
17192 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Consider successor
17193 basic block when looking for nul assignment.
17195 2018-03-07 Eric Botcazou <ebotcazou@adacore.com>
17198 * except.h (output_function_exception_table): Adjust prototype.
17199 * except.c (output_function_exception_table): Remove FNNAME parameter
17200 and add SECTION parameter. Ouput one part of the table at a time.
17201 * final.c (final_scan_insn_1) <NOTE_INSN_SWITCH_TEXT_SECTIONS>: Output
17202 the first part of the exception table and emit unwind directives.
17203 * config/i386/i386-protos.h (i386_pe_end_cold_function): Declare.
17204 (i386_pe_seh_cold_init): Likewise.
17205 * config/i386/cygming.h (ASM_DECLARE_COLD_FUNCTION_NAME): New macro.
17206 (ASM_DECLARE_COLD_FUNCTION_SIZE): Likewise.
17207 * config/i386/i386.c (x86_expand_epilogue): Fix wording in comment.
17208 (ix86_output_call_insn): Emit a nop in one more case for SEH.
17209 * config/i386/winnt.c: Include except.h.
17210 (struct seh_frame_state): Add reg_offset, after_prologue and
17211 in_cold_section fields.
17212 (i386_pe_seh_end_prologue): Set seh->after_prologue.
17213 (i386_pe_seh_cold_init): New function.
17214 (i386_pe_seh_fini): Add COLD parameter and bail out if it is not equal
17215 to seh->in_cold_section.
17216 (seh_emit_push): Record the offset of the push.
17217 (seh_emit_save): Record the offet of the save.
17218 (i386_pe_seh_unwind_emit): Deal with NOTE_INSN_SWITCH_TEXT_SECTIONS.
17219 Test seh->after_prologue to disregard the epilogue.
17220 (i386_pe_end_function): Pass FALSE to i386_pe_seh_fini.
17221 (i386_pe_end_cold_function): New function.
17223 2018-03-07 Jakub Jelinek <jakub@redhat.com>
17226 * config/aarch64/predicates.md (aarch64_simd_reg_or_zero): Use
17227 aarch64_simd_or_scalar_imm_zero rather than aarch64_simd_imm_zero.
17230 * gimple-expr.c (create_tmp_var_raw): Set DECL_NAMELESS flag
17232 * tree-pretty-print.c (dump_decl_name): For TDF_COMPARE_DEBUG,
17233 don't print names of DECL_NAMELESS DECL_IGNORED_P decls.
17235 PR middle-end/84723
17236 * multiple_target.c: Include tree-inline.h and intl.h.
17237 (expand_target_clones): Diagnose and fail if node->definition and
17238 !tree_versionable_function_p (node->decl).
17240 2018-03-06 John David Anglin <danglin@gcc.gnu.org>
17242 * config/pa/pa.h (ASM_GENERATE_INTERNAL_LABEL): Revise to use
17244 (ASM_OUTPUT_ADDR_VEC_ELT): Revise for above change.
17245 (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
17246 * config/pa/pa64-hpux.h (ASM_GENERATE_INTERNAL_LABEL): Revise as above.
17248 2018-03-06 Jakub Jelinek <jakub@redhat.com>
17251 * combine.c (try_combine): Use reg_or_subregno instead of handling
17252 just paradoxical SUBREGs and REGs.
17254 2018-03-06 Claudiu Zissulescu <claziss@synopsys.com>
17256 * config/arc/arc.c (arc_finalize_pic): Remove function.
17257 (arc_must_save_register): We use single base PIC register, remove
17258 checks to save/restore the PIC register.
17259 (arc_expand_prologue): Likewise.
17260 * config/arc/arc-protos.h (arc_set_default_type_attributes):
17262 (arc_verify_short): Likewise.
17263 (arc_attr_type): Likewise.
17264 * config/arc/arc.c (arc_set_default_type_attributes): Remove.
17265 (walk_stores): Likewise.
17266 (arc_address_cost): Make it static.
17267 (arc_verify_short): Likewise.
17268 (branch_dest): Likewise.
17269 (arc_attr_type): Likewise.
17270 * config/arc/arc.c (TARGET_ADJUST_INSN_LENGTH): Remove.
17271 (TARGET_INSN_LENGTH_PARAMETERS): Likewise.
17272 (arc_final_prescan_insn): Remove inserting the nops due to
17273 hardware hazards. It is done in reorg step.
17274 (insn_length_variant_t): Remove.
17275 (insn_length_parameters_t): Likewise.
17276 (arc_insn_length_parameters): Likewise.
17277 (arc_get_insn_variants): Likewise.
17278 * config/arc/arc.h (TARGET_UPSIZE_DBR): Remove.
17280 2018-03-06 Jakub Jelinek <jakub@redhat.com>
17282 PR inline-asm/84683
17283 * reg-stack.c (move_for_stack_reg): If any_malformed_asm, avoid
17286 PR tree-optimization/84687
17287 * omp-simd-clone.c (simd_clone_create): Clear DECL_BUILT_IN_CLASS
17289 * match.pd (pow(C,x)*expN(y) -> expN(logN(C)*x+y)): New optimization.
17291 2018-03-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17293 * config/rs6000/rs6000-builtin.def (rs6000_speculation_barrier):
17294 Rename to ppc_speculation_barrier.
17295 * config/rs6000/rs6000.c (rs6000_init_builtins): Rename builtin to
17296 __builtin_ppc_speculation_barrier.
17298 2018-03-05 Jakub Jelinek <jakub@redhat.com>
17301 * combine.c (combine_simplify_rtx): Don't try to simplify if
17302 if_then_else_cond returned non-NULL, but either true_rtx or false_rtx
17305 2018-03-05 Segher Boessenkool <segher@kernel.crashing.org>
17307 * config/rs6000/rs6000.c (rs6000_loop_align): Don't align tiny loops
17308 to 32 bytes when compiling for POWER9.
17310 2018-03-05 Jakub Jelinek <jakub@redhat.com>
17313 * config/i386/i386.c (ix86_function_ok_for_sibcall): Check for
17314 regparm >= 3 with no arg reg available also for calls with
17315 flag_force_indirect_call. Pass decl to ix86_function_regparm.
17318 * config/i386/sse.md (*<code><mode>3): Replace <mask_prefix3> with
17320 (*<plusminus_insn><mode>3): Likewise. Remove <mask_operand3> uses.
17322 2018-03-05 Peter Bergner <bergner@vnet.ibm.com>
17325 * config/rs6000/vector.md (mov<mode>): Disallow altivec memory operands.
17327 2018-03-05 Richard Biener <rguenther@suse.de>
17329 PR tree-optimization/84486
17330 * tree-ssa-pre.c (create_expression_by_pieces): Remove dead code.
17331 When inserting a __builtin_assume_aligned call set the LHS
17332 SSA name alignment info accordingly.
17334 2018-03-05 Wilco Dijkstra <wdijkstr@arm.com>
17336 PR tree-optimization/84114
17337 * config/aarch64/aarch64.c (aarch64_reassociation_width)
17338 Avoid reassociation of FLOAT_MODE addition.
17340 2018-03-05 Olga Makhotina <olga.makhotina@intel.com>
17342 * common/config/i386/i386-common.c (OPTION_MASK_ISA_PCONFIG_SET,
17343 OPTION_MASK_ISA_PCONFIG_UNSET, OPTION_MASK_ISA_WBNOINVD_SET,
17344 OPTION_MASK_ISA_WBNOINVD_UNSET): New definitions.
17345 (ix86_handle_option): Handle -mpconfig and -mwbnoinvd.
17346 * config.gcc (pconfigintrin.h, wbnoinvdintrin.h) : Add headers.
17347 * config/i386/cpuid.h (bit_PCONFIG, bit_WBNOINVD): New.
17348 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mpconfig
17350 * config/i386/i386-builtin.def (__builtin_ia32_wbnoinvd,
17351 __builtin_ia32_wbinvd): New builtins.
17352 (SPECIAL_ARGS2): New.
17353 * config/i386/i386-c.c (__WBNOINVD__, __PCONFIG__): New.
17354 (SPECIAL_ARGS2): New.
17355 * config/i386/i386.c (ix86_target_string): Add -mpconfig and -mwbnoinvd.
17356 (ix86_valid_target_attribute_inner_p): Ditto.
17357 (ix86_init_mmx_sse_builtins): Add special_args2.
17358 * config/i386/i386.h (TARGET_PCONFIG, TARGET_PCONFIG_P, TARGET_WBNOINVD,
17359 TARGET_WBNOINVD_P): New.
17360 * config/i386/i386.md (UNSPECV_WBINVD, UNSPECV_WBNOINVD): New.
17361 (define_insn "wbinvd", define_insn "wbnoinvd"): New.
17362 * config/i386/i386.opt: Add -mpconfig and -mwbnoinvd.
17363 * config/i386/immintrin.h (_wbinvd): New intrinsic.
17364 * config/i386/pconfigintrin.h: New file.
17365 * config/i386/wbnoinvdintrin.h: Ditto.
17366 * config/i386/x86intrin.h: Add headers pconfigintrin.h and
17368 * doc/invoke.texi (-mpconfig, -mwbnoinvd): New.
17370 2018-03-05 Richard Biener <rguenther@suse.de>
17372 PR tree-optimization/84670
17373 * tree-ssa-pre.c (struct bb_bitmap_sets): Add visited_with_visited_succs
17375 (BB_VISITED_WITH_VISITED_SUCCS): New define.
17376 (compute_antic): Initialize BB_VISITED_WITH_VISITED_SUCCS.
17377 (compute_antic_aux): Only assert the number of values in ANTIC_IN
17378 doesn't grow if all successors (recursively) were visited at least
17381 2018-03-05 Richard Biener <rguenther@suse.de>
17383 PR tree-optimization/84650
17384 * tree-ssa-loop-im.c (pass_lim::execute): Reset the SCEV cache
17385 if executed in the loop pipeline.
17387 2018-03-05 Sandra Loosemore <sandra@codesourcery.com>
17389 * doc/configfiles.texi (Configuration Files): Move info about
17390 conditionalizing $target-protos.h to...
17391 * doc/sourcebuild.texi (Back End): Here. Explain how $target.h
17392 differs from $target-protos.h.
17394 2018-03-05 Kito Cheng <kito.cheng@gmail.com>
17395 Chung-Ju Wu <jasonwucj@gmail.com>
17397 * config/nds32/nds32-protos.h (nds32_expand_setmem): Declare.
17398 * config/nds32/nds32-multiple.md (setmemsi): Define.
17399 * config/nds32/nds32-memory-manipulation.c
17400 (nds32_gen_dup_4_byte_to_word_value): New.
17401 (emit_setmem_word_loop): New.
17402 (emit_setmem_byte_loop): New.
17403 (nds32_expand_setmem_loop): New.
17404 (nds32_expand_setmem_loop_v3m): New.
17405 (nds32_expand_setmem_unroll): New.
17406 (nds32_expand_setmem): New.
17408 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
17409 Chung-Ju Wu <jasonwucj@gmail.com>
17411 * config/nds32/nds32-memory-manipulation.c
17412 (nds32_emit_load_store): New.
17413 (nds32_emit_post_inc_load_store): New.
17414 (nds32_emit_mem_move): New.
17415 (nds32_emit_mem_move_block): New.
17416 (nds32_expand_movmemsi_loop_unknown_size): New.
17417 (nds32_expand_movmemsi_loop_known_size): New.
17418 (nds32_expand_movmemsi_loop): New.
17419 (nds32_expand_movmemsi_unroll): New.
17420 (nds32_expand_movmemqi): Rename ...
17421 (nds32_expand_movmemsi): ... to this.
17422 * config/nds32/nds32-multiple.md (movmemqi): Rename ...
17423 (movmemsi): ... to this.
17424 * config/nds32/nds32-protos.h (nds32_expand_movmemqi): Rename ...
17425 (nds32_expand_movmemsi): ... to this.
17427 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
17428 Monk Chiang <sh.chiang04@gmail.com>
17429 Chung-Ju Wu <jasonwucj@gmail.com>
17431 * config/nds32/nds32-protos.h
17432 (nds32_expand_load_multiple): New arguments.
17433 (nds32_expand_store_multiple): Ditto.
17434 (nds32_valid_multiple_load_store): Rename ...
17435 (nds32_valid_multiple_load_store_p): ... to this.
17436 * config/nds32/nds32-memory-manipulation.c
17437 (nds32_expand_load_multiple): Refine implementation.
17438 (nds32_expand_store_multiple): Ditto.
17439 * config/nds32/nds32-multiple.md
17440 (load_multiple): Update nds32_expand_load_multiple interface.
17441 (store_multiple): Update nds32_expand_store_multiple interface.
17442 * config/nds32/nds32-predicates.c
17443 (nds32_valid_multiple_load_store): Rename ...
17444 (nds32_valid_multiple_load_store_p): ... to this and refine
17446 * config/nds32/predicates.md
17447 (nds32_load_multiple_and_update_address_operation): New predicate.
17448 (nds32_store_multiple_and_update_address_operation): New predicate.
17450 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
17451 Chung-Ju Wu <jasonwucj@gmail.com>
17453 * config/nds32/nds32.md (type): Add load_multiple and store_multiple.
17454 (combo): New attribute.
17455 * config/nds32/nds32-multiple.md: Refine patterns with new attributes.
17457 2018-03-03 Chung-Ju Wu <jasonwucj@gmail.com>
17459 * config/nds32/nds32.opt: Change -mcmodel= default value.
17461 2018-03-03 Kito Cheng <kito.cheng@gmail.com>
17462 Monk Chiang <sh.chiang04@gmail.com>
17463 Chung-Ju Wu <jasonwucj@gmail.com>
17465 * config/nds32/constants.md (unspec_element): New enum.
17466 * config/nds32/constraints.md (Umw): New constraint.
17467 * config/nds32/nds32-intrinsic.c: Add more builtin functions.
17468 * config/nds32/nds32-intrinsic.md: Likewise.
17469 * config/nds32/nds32-md-auxiliary.c (nds32_regno_to_enable4): New.
17470 (nds32_valid_smw_lwm_base_p): New.
17471 (nds32_output_smw_single_word): New.
17472 (nds32_output_lmw_single_word): New.
17473 (nds32_expand_unaligned_load): New.
17474 (nds32_expand_unaligned_store): New.
17475 * config/nds32/nds32-protos.h (nds32_valid_smw_lwm_base_p): Declare.
17476 (nds32_output_smw_single_word): Declare.
17477 (nds32_output_lmw_single_word): Declare.
17478 (nds32_expand_unaligned_load): Declare.
17479 (nds32_expand_unaligned_store): Declare.
17480 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_UALOAD_HW,
17481 NDS32_BUILTIN_UALOAD_W, NDS32_BUILTIN_UALOAD_DW,
17482 NDS32_BUILTIN_UASTORE_HW, NDS32_BUILTIN_UASTORE_W,
17483 NDS32_BUILTIN_UASTORE_DW.
17484 * config/nds32/predicates.md (nds32_lmw_smw_base_operand): New
17487 2018-03-03 Monk Chiang <sh.chiang04@gmail.com>
17488 Kito Cheng <kito.cheng@gmail.com>
17489 Chung-Ju Wu <jasonwucj@gmail.com>
17491 * config/nds32/nds32-intrinsic.c
17492 (nds32_expand_builtin_null_ftype_reg): Delete.
17493 (nds32_expand_builtin_reg_ftype_imm): Ditto.
17494 (nds32_expand_builtin_null_ftype_reg_imm): Ditto.
17495 (nds32_read_argument): New.
17496 (nds32_legitimize_target): Ditto.
17497 (nds32_legitimize_argument): Ditto.
17498 (nds32_check_constant_argument): Ditto.
17499 (nds32_expand_unop_builtin): Ditto.
17500 (nds32_expand_unopimm_builtin): Ditto.
17501 (nds32_expand_binop_builtin): Ditto.
17502 (nds32_builtin_decl_impl): Ditto.
17503 (builtin_description): Ditto.
17504 (nds32_expand_builtin_impl): Rewrite with new infrastructure.
17505 (nds32_init_builtins_impl): Ditto.
17506 * config/nds32/nds32.c (TARGET_BUILTIN_DECL): Define.
17507 (nds32_builtin_decl): New.
17508 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_COUNT.
17509 * config/nds32/nds32-protos.h (nds32_builtin_decl_impl): Declare.
17511 2018-03-02 Jeff Law <law@redhat.com>
17513 * reorg.c (stop_search_p): Handle DEBUG_INSN.
17514 (redundant_insn, fill_simple_delay_slots): Likewise.
17515 (fill_slots_from_thread): Likewise.
17516 * resource.c (mark_referenced_resources): Likewise.
17517 (mark_set_resources, find_dead_or_set_registers): Likewise.
17519 2018-03-02 Jakub Jelinek <jakub@redhat.com>
17521 * substring-locations.h (format_warning_va): Formatting fix for
17522 ATTRIBUTE_GCC_DIAG.
17523 (format_warning_at_substring): Fix up ATTRIBUTE_GCC_DIAG second
17525 (format_warning_n_va, format_warning_at_substring_n): New prototypes.
17526 * substring-locations.c: Include intl.h.
17527 (format_warning_va): Turned into small wrapper around
17528 format_warning_n_va, renamed to ...
17529 (format_warning_n_va): ... this, add N and PLURAL_GMSGID arguments,
17530 rename GMSGID to SINGULAR_GMSGID, if SINGULAR_GMSGID != PLURAL_GMSGID,
17532 (format_warning_at_substring_n): New function.
17533 * gimple-ssa-sprintf.c: Remove GCC diagnostic ignored pragma.
17534 (fmtwarn): Add ATTRIBUTE_GCC_DIAG. Turn into a copy of
17535 format_warning_at_substring with just a shorter name instead of
17536 const function pointer.
17537 (fmtwarn_n): New function.
17538 (maybe_warn, format_directive, parse_directive): Use fmtwarn_n where
17539 appropriate, get rid of all the fmtstr temporaries, move conditionals
17540 with G_() wrapped string literals directly into fmtwarn arguments,
17541 cast dir.len to (int), formatting fixes.
17543 2018-03-02 Thomas Schwinge <thomas@codesourcery.com>
17545 * doc/invoke.texi: Remove "Cilk Plus" references.
17547 2018-03-02 Jakub Jelinek <jakub@redhat.com>
17548 Richard Biener <rguenther@suse.de>
17551 * expr.c (expand_expr_real_1) <case CALL_EXPR>: Don't emit diagnostics
17552 for error or warning attributes if CALL_FROM_THUNK_P is set.
17555 2018-03-02 Jakub Jelinek <jakub@redhat.com>
17558 * config/pa/pa.h (TARGET_CPU_CPP_BUILTINS): Predefine
17559 __SIZEOF_128__ macro if HPUX_LONG_DOUBLE_LIBRARY.
17562 * config/ia64/ia64.h (TARGET_CPU_CPP_BUILTINS): Predefine
17563 __SIZEOF_{FPREG,FLOAT{80,128}}__ macros.
17565 * predict.c (test_prediction_value_range): Use PROB_UNINITIALIZED
17566 instead of -1U in last predictors element's probability member.
17568 2018-03-02 Eric Botcazou <ebotcazou@adacore.com>
17571 * ipa-devirt.c (odr_subtypes_equivalent_p): Get the ODR type of both
17572 arguments if they are comparable.
17574 2018-03-02 Richard Sandiford <richard.sandiford@linaro.org>
17576 PR tree-optimization/84634
17577 * tree-vect-stmts.c (vectorizable_store, vectorizable_load): Replace
17578 masks and masked_loop_p with a single loop_masks, making sure it's
17579 null for bb vectorization.
17581 2018-03-02 Richard Sandiford <richard.sandiford@linaro.org>
17583 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence)
17584 (vect_analyze_data_ref_access): Use loop->safe_len rather than
17585 loop->force_vectorize to check whether there is no alias.
17587 2018-03-02 Jakub Jelinek <jakub@redhat.com>
17590 * rtl.h (prev_real_nondebug_insn, next_real_nondebug_insn): New
17592 * emit-rtl.c (next_real_insn, prev_real_insn): Fix up function
17594 (next_real_nondebug_insn, prev_real_nondebug_insn): New functions.
17595 * cfgcleanup.c (try_head_merge_bb): Use prev_real_nondebug_insn
17596 instead of a loop around prev_real_insn.
17597 * combine.c (move_deaths): Use prev_real_nondebug_insn instead of
17600 PR inline-asm/84625
17601 * config/i386/i386.c (ix86_print_operand): Use conditional
17602 output_operand_lossage instead of gcc_assert if CONST_VECTOR is not
17605 2018-03-02 Richard Biener <rguenther@suse.de>
17607 PR tree-optimization/84427
17608 * tree-ssa-pre.c (bitmap_remove_expr_from_set): Remove.
17609 (bitmap_set_subtract_values): Rewrite to handle multiple
17612 (prune_clobbered_mems): Likewise.
17613 (phi_translate): Take edge instead of pred/phiblock.
17614 (phi_translate_1): Likewise.
17615 (phi_translate_set): Likewise. Insert all translated
17616 exprs for a value into the set, keeping possibly multiple
17617 expressions per value.
17618 (compute_antic_aux): Adjust for phi_translate changes.
17619 When intersecting union the expressions and prune those
17620 not in the final value set, keeping possibly multiple
17621 expressions per value. Do not use value-insertion
17622 for unioning ANTIC_OUT U EXP_GEN - TMP_GEN but merge
17623 all expressions. Add verification that the value-sets
17624 only shrink during iteration.
17625 (compute_partial_antic_aux): Adjust for the phi_translate changes.
17626 (do_pre_regular_insertion): Likewise.
17627 (do_pre_partial_partial_insertion): Likewise.
17629 2018-03-02 Richard Biener <rguenther@suse.de>
17632 * config/darwin.c (saved_debug_info_level): New static global.
17633 (darwin_asm_lto_start): Disable debug info generation for LTO out.
17634 (darwin_asm_lto_end): Restore debug info generation settings.
17636 2018-03-01 Martin Liska <mliska@suse.cz>
17639 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
17640 volatile arguments.
17642 2018-03-01 Richard Biener <rguenther@suse.de>
17645 * dwarf2out.c (gen_variable_die): Properly handle late VLA
17646 type annotation with LTO when debug was disabled at compile-time.
17648 2018-03-01 Matthew Fortune <mfortune@gmail.com>
17650 * config/mips/mips.c (mips_final_prescan_insn): Fix incorrect
17652 (mips_final_postscan_insn): Likewise.
17654 2018-03-01 Richard Sandiford <richard.sandiford@linaro.org>
17656 PR rtl-optimization/84528
17657 * alias.c (init_alias_target): Add commentary.
17658 (init_alias_analysis): Only give HARD_FRAME_POINTER_REGNUM
17659 a unique base value if the frame pointer is not eliminated
17660 to the stack pointer.
17662 2018-03-01 Tom de Vries <tom@codesourcery.com>
17664 PR rtl-optimization/83327
17665 * lra-int.h (hard_regs_spilled_into): Declare.
17666 * lra.c (hard_regs_spilled_into): Define.
17667 (init_reg_info): Init hard_regs_spilled_into.
17668 * lra-spills.c (assign_spill_hard_regs): Update hard_regs_spilled_into.
17669 * lra-lives.c (make_hard_regno_born, make_hard_regno_dead)
17670 (process_bb_lives): Handle hard_regs_spilled_into.
17671 (lra_create_live_ranges_1): Before doing liveness propagation, clear
17672 regs in all_hard_regs_bitmap if set in hard_regs_spilled_into.
17674 2018-02-28 David Edelsohn <dje.gcc@gmail.com>
17676 * config.gcc (powerpc-ibm-aix7.1.*): New stanza.
17677 (powerpc-ibm-aix[789]*): Default to AIX 7.2.
17678 * config/rs6000/aix71.h (TARGET_DEFAULT): Revert to Power4 ISA.
17679 * config/rs6000/aix72.h: New file.
17681 2018-02-28 Jakub Jelinek <jakub@redhat.com>
17683 * gimple-ssa-warn-restrict.c (maybe_diag_overlap): Use warning_n
17684 instead of warning_at with conditional singular and plural messages
17688 * stor-layout.c (update_alignment_for_field): For
17689 targetm.ms_bitfield_layout_p (rli->t), if !is_bitfield
17690 && !DECL_PACKED (field), do the alignment update, just use
17691 only desired_align instead of MAX (type_align, desired_align)
17693 (place_field): Don't do known_align < desired_align handling
17694 early if targetm.ms_bitfield_layout_p (rli->t) and rli->prev_field
17695 is non-NULL, instead do it after rli->prev_field handling and
17696 only if not within a bitfield word. For DECL_PACKED (field)
17697 use type_align of BITS_PER_UNIT.
17699 2018-02-28 Eric Botcazou <ebotcazou@adacore.com>
17701 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Remove
17702 superfluous parentheses and trailing spaces.
17704 2018-02-28 Richard Biener <rguenther@suse.de>
17706 PR tree-optimization/84584
17707 * graphite-scop-detection.c (scop_detection::add_scop): Discard
17708 SCoPs with fake exit edge.
17710 2018-02-28 Martin Liska <mliska@suse.cz>
17713 * timevar.c (timer::print): Fix format to properly print 100%
17716 2018-02-28 Richard Biener <rguenther@suse.de>
17718 PR middle-end/84607
17719 * genmatch.c (capture_info::walk_match): Do not mark
17720 captured expressions without operands as expr_p given
17721 they act more like predicates and should be subject to
17722 "lost tail" side-effect preserving.
17724 2018-02-28 Alexandre Oliva <aoliva@redhat.com>
17726 PR rtl-optimization/81611
17727 * auto-inc-dec.c (attempt_change): Move dead note from
17728 mem_insn if it's the next use of regno
17729 (find_address): Take address use of reg holding
17730 non-incremented value. Add parm to limit search to the named
17732 (merge_in_block): Attempt to use a mem insn that is the next
17733 use of the original regno.
17735 2018-02-27 Martin Sebor <msebor@redhat.com>
17738 * doc/invoke.texi (-Wmissing-attributes): New option.
17739 * print-tree.c (print_node): Handle DECL_UNINLINABLE.
17741 2018-02-27 Martin Sebor <msebor@redhat.com>
17743 PR translation/84207
17744 * diagnostic-core.h (warning_n, error_n, inform_n): Change
17745 n argument to unsigned HOST_WIDE_INT.
17746 * diagnostic.c (warning_n, error_n, inform_n): Ditto.
17747 (diagnostic_n_impl): Ditto. Handle arguments in excess of LONG_MAX.
17748 * gimple-ssa-sprintf.c (format_directive): Simplify inform_n call.
17749 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Use warning_n.
17751 2018-02-27 Richard Biener <rguenther@suse.de>
17753 PR tree-optimization/84512
17754 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost):
17755 Do not use the estimate returned from record_stmt_cost for
17756 the scalar iteration cost but sum properly using add_stmt_cost.
17758 2018-02-27 Richard Biener <rguenther@suse.de>
17760 PR tree-optimization/84466
17761 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
17762 Adjust last change to less strictly validate use operands.
17764 2018-02-27 Martin Liska <mliska@suse.cz>
17766 PR gcov-profile/84548
17767 * gcov.c (process_file): Allow partial overlap and consider it
17768 also as group functions.
17769 (output_lines): Properly calculate range of lines for a group.
17771 2018-02-27 Martin Liska <mliska@suse.cz>
17773 * timevar.c (timer::print_row): Remove 'usr', 'sys', 'wall' and
17774 'ggc' suffixes. Change first column width.
17775 (timer::print): Fix formatting of the column.
17777 2018-02-27 Alexandre Oliva <aoliva@redhat.com>
17779 * tree-ssa-live.c (remove_unused_scope_block_p): Do not
17780 preserve inline entry blocks for the sake of debug inline
17781 entry point markers alone.
17782 (remove_unused_locals): Suggest in comments a better place to
17783 force the preservation of inline entry blocks that are
17784 otherwise unused, but do not preserve them.
17786 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
17788 * config/i386/i386.c (ix86_output_indirect_jmp): Update comments.
17790 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
17793 * config/i386/constraints.md (Bs): Replace
17794 ix86_indirect_branch_register with
17795 TARGET_INDIRECT_BRANCH_REGISTER.
17797 * config/i386/i386.md (indirect_jump): Likewise.
17798 (tablejump): Likewise.
17799 (*sibcall_memory): Likewise.
17800 (*sibcall_value_memory): Likewise.
17801 Peepholes of indirect call and jump via memory: Likewise.
17802 (*sibcall_GOT_32): Disallowed for TARGET_INDIRECT_BRANCH_REGISTER.
17803 (*sibcall_value_GOT_32): Likewise.
17804 * config/i386/predicates.md (indirect_branch_operand): Likewise.
17805 (GOT_memory_operand): Likewise.
17806 (call_insn_operand): Likewise.
17807 (sibcall_insn_operand): Likewise.
17808 (GOT32_symbol_operand): Likewise.
17809 * config/i386/i386.h (TARGET_INDIRECT_BRANCH_REGISTER): New.
17811 2018-02-26 Eric Botcazou <ebotcazou@adacore.com>
17813 PR rtl-optimization/83496
17814 * reorg.c (steal_delay_list_from_target): Change REDUNDANT array from
17815 booleans to RTXes. Call fix_reg_dead_note on every non-null element.
17816 (steal_delay_list_from_fallthrough): Call fix_reg_dead_note on a
17817 redundant insn, if any.
17818 (relax_delay_slots): Likewise.
17819 (update_reg_unused_notes): Rename REDUNDANT_INSN to OTHER_INSN.
17821 2018-02-26 Richard Sandiford <richard.sandiford@linaro.org>
17823 PR tree-optimization/83965
17824 * tree-vect-patterns.c (vect_reassociating_reduction_p): Assume
17825 that grouped statements are part of a reduction chain. Return
17826 true if the statement is not marked as a reduction itself but
17827 is part of a group.
17828 (vect_recog_dot_prod_pattern): Don't check whether the statement
17829 is part of a group here.
17830 (vect_recog_sad_pattern): Likewise.
17831 (vect_recog_widen_sum_pattern): Likewise.
17833 2018-02-26 Eric Botcazou <ebotcazou@adacore.com>
17836 * final.c (rest_of_clean_state): Also look for calls inside sequences.
17838 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
17841 * config/i386/i386-protos.h (ix86_output_indirect_jmp): Remove
17843 (ix86_output_indirect_function_return): New prototype.
17844 (ix86_split_simple_return_pop_internal): Likewise.
17845 * config/i386/i386.c (indirect_return_via_cx): New.
17846 (indirect_return_via_cx_bnd): Likewise.
17847 (indirect_thunk_name): Handle return va CX_REG.
17848 (output_indirect_thunk_function): Create alias for
17849 __x86_return_thunk_[re]cx and __x86_return_thunk_[re]cx_bnd.
17850 (ix86_output_indirect_jmp): Remove the bool argument.
17851 (ix86_output_indirect_function_return): New function.
17852 (ix86_split_simple_return_pop_internal): Likewise.
17853 * config/i386/i386.md (*indirect_jump): Don't pass false
17854 to ix86_output_indirect_jmp.
17855 (*tablejump_1): Likewise.
17856 (simple_return_pop_internal): Change it to define_insn_and_split.
17857 Call ix86_split_simple_return_pop_internal to split it for
17858 -mfunction-return=.
17859 (simple_return_indirect_internal): Call
17860 ix86_output_indirect_function_return instead of
17861 ix86_output_indirect_jmp.
17863 2018-02-26 Jakub Jelinek <jakub@redhat.com>
17866 * vec.h (vec_default_construct): For BROKEN_VALUE_INITIALIZATION use
17867 memset and value initialization afterwards.
17869 2018-02-26 Christophe Lyon <christophe.lyon@linaro.org>
17871 * Makefile.in (lto-wrapper): Use ALL_LINKERFLAGS.
17873 2018-02-26 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
17876 * common/config/aarch64/aarch64-common.c
17877 (aarch_option_optimization_table[]): Switch
17878 off fomit-frame-pointer
17880 2018-02-26 Kito Cheng <kito.cheng@gmail.com>
17881 Chung-Ju Wu <jasonwucj@gmail.com>
17883 * config/nds32/nds32-multiple.md (load_multiple): Disallow
17885 (store_multiple): Ditto.
17887 2018-02-26 Kito Cheng <kito.cheng@gmail.com>
17889 * config.gcc: Add --with-cpu support for nds32 target.
17890 * config/nds32/nds32-opts.h (nds32_cpu_type): New.
17891 * config/nds32/nds32.opt: Add -mcpu= option.
17893 2018-02-25 Segher Boessenkool <segher@kernel.crashing.org>
17895 * config/rs6000/rs6000.opt (mvrsave=no, mvrsave=yes, isel=no,
17896 isel=yes): Warn for these deprecated options.
17898 2018-02-23 David Edelsohn <dje.gcc@gmail.com>
17900 * config/rs6000/aix71.h (TARGET_DEFAULT): Change to
17901 ISA_2_5_MASKS_EMBEDDED.
17903 2018-02-23 Jakub Jelinek <jakub@redhat.com>
17905 * ipa-prop.c (ipa_vr_ggc_hash_traits::hash): Hash p->min and
17906 p->max as pointers rather than using iterative_hash_expr.
17908 2018-02-23 Carl Love <cel@us.ibm.com>
17910 * config/rs6000/rs6000-builtin.def: Change VSIGNED2 and VUNSIGNED2
17911 macro expansions from BU_VSX_2 to BU_P8V_VSX_2 and BU_VSX_OVERLOAD_2 to
17913 * config/rs6000/rs6000-c.c: Change VSX_BUILTIN_VEC_VSIGNED2 to
17914 P8V_BUILTIN_VEC_VSIGNED2. Change VSX_BUILTIN_VEC_VUNSIGNED2 to
17915 P8V_BUILTIN_VEC_VUNSIGNED2.
17917 2018-02-22 Vladimir Makarov <vmakarov@redhat.com>
17920 * lra-int.h (LRA_UNKNOWN_ALT, LRA_NON_CLOBBERED_ALT): New macros.
17921 * lra.c (lra_set_insn_recog_data, lra_update_insn_recog_data): Use
17923 * lra-constraints.c (curr_insn_transform): Set up
17924 LRA_NON_CLOBBERED_ALT for moves processed on the fast path. Use
17926 (remove_inheritance_pseudos): Use LRA_UNKNOWN_ALT.
17927 * lra-eliminations.c (spill_pseudos): Ditto.
17928 (process_insn_for_elimination): Ditto.
17929 * lra-lives.c (reg_early_clobber_p): Use the new macros.
17930 * lra-spills.c (spill_pseudos): Use LRA_UNKNOWN_ALT and
17931 LRA_NON_CLOBBERED_ALT.
17933 2018-02-22 Martin Sebor <msebor@redhat.com>
17935 PR tree-optimization/84480
17936 * gimple-fold.c (gimple_fold_builtin_strcpy): Move warnings
17937 to maybe_diag_stxncpy_trunc. Call it.
17938 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Integrate warnings
17939 from gimple_fold_builtin_strcpy. Print inlining stack.
17940 (handle_builtin_stxncpy): Print inlining stack.
17941 * tree-ssa-strlen.h (maybe_diag_stxncpy_trunc): Declare.
17943 2018-02-22 H.J. Lu <hongjiu.lu@intel.com>
17946 * config/i386/i386.c (ix86_set_indirect_branch_type): Issue an
17947 error when -mindirect-branch=thunk-extern, -fcf-protection=branch
17948 and -fcheck-pointer-bounds are used together.
17949 (indirect_thunk_prefix): New enum.
17950 (indirect_thunk_need_prefix): New function.
17951 (indirect_thunk_name): Replace need_bnd_p with need_prefix. Use
17952 "_nt" instead of "_bnd" for NOTRACK prefix.
17953 (output_indirect_thunk): Replace need_bnd_p with need_prefix.
17954 (output_indirect_thunk_function): Likewise.
17956 (ix86_code_end): Update output_indirect_thunk_function calls.
17957 (ix86_output_indirect_branch_via_reg): Replace
17958 ix86_bnd_prefixed_insn_p with indirect_thunk_need_prefix.
17959 (ix86_output_indirect_branch_via_push): Likewise.
17960 (ix86_output_function_return): Likewise.
17961 * doc/invoke.texi: Document -mindirect-branch=thunk-extern is
17962 incompatible with -fcf-protection=branch and
17963 -fcheck-pointer-bounds.
17965 2018-02-22 Steve Ellcey <sellcey@cavium.com>
17968 * config/aarch64/aarch64.c (aarch64_print_address_internal):
17969 Change gcc_assert call to output_operand_lossage.
17971 2018-02-22 Steve Ellcey <sellcey@cavium.com>
17973 * doc/extend.texi (__builtin_extend_pointer): Document builtin.
17975 2018-02-22 DJ Delorie <dj@redhat.com>
17976 Sebastian Perta <sebastian.perta@renesas.com>
17977 Oleg Endo <olegendo@gcc.gnu.org>
17979 * config/rx/rx.c (rx_rtx_costs): New function.
17980 (TARGET_RTX_COSTS): Override to use rx_rtx_costs.
17982 2018-02-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
17984 * config/arm/t-multilib: Map Armv8-R to Armv7 multilibs.
17986 2018-02-22 Martin Liska <mliska@suse.cz>
17989 * common/config/arm/arm-common.c (arm_print_hint_for_cpu_option):
17990 Add "native" as a possible value.
17992 2018-02-22 Martin Liska <mliska@suse.cz>
17995 * config/i386/i386.c (ix86_option_override_internal):
17996 Add "native" as a possible value for -march and -mtune.
17998 2018-02-22 Jakub Jelinek <jakub@redhat.com>
18001 * stor-layout.c (finalize_type_size): Propagate TYPE_EMPTY_P flag
18002 to all type variants.
18004 PR tree-optimization/84503
18005 * gimple-ssa-store-merging.c (merged_store_group::merge_into): Compute
18006 width as info->bitpos + info->bitsize - start.
18007 (merged_store_group::merge_overlapping): Simplify width computation.
18008 (check_no_overlap): New function.
18009 (imm_store_chain_info::try_coalesce_bswap): Compute expected
18010 start + width and last_order of the group, fail if check_no_overlap
18012 (imm_store_chain_info::coalesce_immediate_stores): Don't merge info
18013 to group if check_no_overlap fails.
18015 2018-02-21 Segher Boessenkool <segher@kernel.crashing.org>
18017 * config/rs6000/altivec.md: Delete contraint arguments to
18018 define_expand, define_split, and define_peephole2, and in
18019 define_insn_and_split if always unused.
18020 * config/rs6000/darwin.md: Ditto.
18021 * config/rs6000/dfp.md: Ditto.
18022 * config/rs6000/rs6000.md: Ditto.
18023 * config/rs6000/sync.md: Ditto.
18024 * config/rs6000/vector.md: Ditto.
18025 * config/rs6000/vsx.md: Ditto.
18027 2018-02-21 Segher Boessenkool <segher@kernel.crashing.org>
18029 * config/rs6000/altivec.md: Write output control strings as braced
18030 blocks instead of double-quoted strings.
18031 * config/rs6000/darwin.md: Ditto.
18032 * config/rs6000/rs6000.md: Ditto.
18033 * config/rs6000/vector.md: Ditto.
18034 * config/rs6000/vsx.md: Ditto.
18036 2018-02-21 Jason Merrill <jason@redhat.com>
18038 PR c++/84314 - ICE with templates and fastcall attribute.
18039 * attribs.c (build_type_attribute_qual_variant): Remove assert.
18041 2018-02-21 Jan Hubicka <hubicka@ucw.cz>
18043 * ipa-cp.c (determine_versionability): Fix comment typos.
18045 2018-02-21 Jan Hubicka <hubicka@ucw.cz>
18048 * ipa-cp.c (determine_versionability): Do not version functions caling
18051 2018-02-21 Martin Liska <mliska@suse.cz>
18054 * config/aarch64/aarch64.c (aarch64_print_hint_for_core_or_arch):
18055 Add "native" as a possible value.
18056 * config/aarch64/aarch64.h (HAVE_LOCAL_CPU_DETECT): Define
18057 the macro when native cpu detection is available.
18059 2018-02-21 Martin Liska <mliska@suse.cz>
18062 * common/config/arm/arm-common.c (arm_print_hint_for_arch_option):
18063 Add "native" as a possible value.
18064 * config/arm/arm.h (HAVE_LOCAL_CPU_DETECT): Define the macro
18065 when native cpu detection is available.
18067 2018-02-21 Jakub Jelinek <jakub@redhat.com>
18068 Martin Sebor <msebor@redhat.com>
18070 PR tree-optimization/84478
18071 * gimple-fold.h (get_range_strlen): Add a bool argument defaulted to
18073 * gimple-fold.c (get_range_strlen): Make minlen const and assume it
18074 can't be NULL. Change FUZZY from bool to int, for 1 add PHI/COND_EXPR
18075 support which is conservatively correct, for 2 only stay conservative
18076 for maxlen. Formatting and comment capitalization fixes. Add STRICT
18077 argument to the 2 argument get_range_strlen, adjust 6 arg
18078 get_range_strlen caller and clear minmaxlen[0] and [1] if it returned
18080 (get_maxval_strlen): Adjust 6 arg get_range_strlen caller.
18081 (gimple_fold_builtin_strlen): Pass true as last argument to
18084 2018-02-20 Martin Sebor <msebor@redhat.com>
18086 PR middle-end/84095
18087 * gimple-ssa-warn-restrict.c (builtin_memref::extend_offset_range): New.
18088 (builtin_memref::set_base_and_offset): Same. Handle inner references.
18089 (builtin_memref::builtin_memref): Factor out parts into
18090 set_base_and_offset and call it.
18092 2018-02-20 Richard Sandiford <richard.sandiford@linaro.org>
18094 PR middle-end/84406
18095 * optabs-query.c (find_widening_optab_handler_and_mode): If from_mode
18096 is a scalar_int_mode, assert that to_mode is a scalar_int_mode with
18097 greater precision. If to_mode is a MODE_PARTIAL_INT, stop the
18098 search at the associated MODE_INT.
18100 2018-02-20 Jeff Law <law@redhat.com>
18102 PR middle-end/82123
18103 PR tree-optimization/81592
18104 PR middle-end/79257
18105 * gimple-ssa-sprintf.c (format_integer): Query EVRP range analyzer
18106 for range data rather than using global data.
18107 * gimple-ssa-sprintf.c (get_int_range): Query EVRP range analyzer for
18108 range data rather than using global data.
18109 * gimple-ssa-sprintf.c (get_int_range): Accept vr_values parameter
18110 pass it to children as needed.
18111 (struct directive::fmtresult): Similarly.
18112 (struct directive::set_width): Similarly.
18113 (struct directive::set_precision): Similarly.
18114 (format_integer, format_directive, parse_directive): Similarly.
18115 (format_none): Accept unnamed vr_values parameter.
18116 (format_percent, format_floating, format_character): Similarly.
18117 (format_string, format_plain): Similarly.
18118 * gimple-ssa-sprintf.c (sprintf_dom_walker::handle_gimple_call): Query
18119 the EVRP range analyzer for range data rather than using global data.
18120 * gimple-ssa-sprintf.c: Include alloc-pool.h, vr-values.h and
18121 gimple-ssa-evrp-analyze.h
18122 (class sprintf_dom_walker): Add after_dom_children member function.
18123 Add evrp_range_analyzer member.
18124 (sprintf_dom_walker::before_dom_children): Call into the EVRP
18125 range analyzer as needed.
18126 (sprintf_dom_walker::after_dom_children): New member function.
18127 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::enter): Do nothing
18129 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
18130 (evrp_range_analyzer::pop_to_marker): Likewise.
18132 2018-02-20 Richard Sandiford <richard.sandiford@linaro.org>
18134 PR tree-optimization/84419
18135 * internal-fn.c (expand_call_mem_ref): Create a TARGET_MEM_REF
18136 with the required type if its current type is compatible but
18139 2018-02-20 Jakub Jelinek <jakub@redhat.com>
18141 PR middle-end/82004
18142 * match.pd (pow(C,x) -> exp(log(C)*x)): Delay all folding until
18143 after vectorization.
18145 2018-02-20 Martin Liska <mliska@suse.cz>
18148 * config/aarch64/aarch64.c (aarch64_print_hint_for_core_or_arch): Print
18149 possible values if we don't have a hint.
18151 2018-02-20 Martin Liska <mliska@suse.cz>
18155 * final.c (shorten_branches): Build align_tab array with one
18157 * opts.c (finish_options): Add alignment option limit check.
18158 (MAX_CODE_ALIGN): Likewise.
18159 (MAX_CODE_ALIGN_VALUE): Likewise.
18160 * doc/invoke.texi: Document maximum allowed option value for
18161 all -falign-* options.
18163 2018-02-19 Jakub Jelinek <jakub@redhat.com>
18166 * reg-notes.def (REG_CALL_ARG_LOCATION): New reg note.
18167 * insn-notes.def (NOTE_INSN_CALL_ARG_LOCATION): Remove.
18168 * var-tracking.c (emit_note_insn_var_location): Remove all references
18169 to NOTE_INSN_CALL_ARG_LOCATION.
18170 (emit_notes_in_bb): Emit arguments as REG_CALL_ARG_LOCATION note on
18171 the CALL_INSN rather than separate NOTE_INSN_CALL_ARG_LOCATION note.
18172 Use copy_rtx_if_shared.
18173 * dwarf2out.c (gen_subprogram_die): Use XEXP with 0 instead of
18174 NOTE_VAR_LOCATION on ca_loc->call_arg_loc_note.
18175 (dwarf2out_var_location): Remove handling of
18176 NOTE_INSN_CALL_ARG_LOCATION, instead handle REG_CALL_ARG_LOCATION note
18178 * final.c (final_scan_insn): Remove all references to
18179 NOTE_INSN_CALL_ARG_LOCATION.
18180 (rest_of_clean_state): Likewise. Remove REG_CALL_ARG_LOCATION notes
18181 before dumping final insns.
18182 * except.c (emit_note_eh_region_end): Remove all references to
18183 NOTE_INSN_CALL_ARG_LOCATION.
18184 * config/alpha/alpha.c (alpha_pad_function_end): Likewise.
18185 * config/c6x/c6x.c (c6x_gen_bundles): Likewise.
18186 * config/arc/arc.c (hwloop_optimize): Likewise.
18187 * config/arm/arm.c (create_fix_barrier): Likewise.
18188 * config/s390/s390.c (s390_chunkify_start): Likewise.
18189 * config/sh/sh.c (find_barrier): Likewise.
18190 * config/i386/i386.c (rest_of_insert_endbranch,
18191 ix86_seh_fixup_eh_fallthru): Likewise.
18192 * config/xtensa/xtensa.c (hwloop_optimize): Likewise.
18193 * config/iq2000/iq2000.c (final_prescan_insn): Likewise.
18194 * config/frv/frv.c (frv_function_prologue): Likewise.
18195 * emit-rtl.c (try_split): Likewise. Copy over REG_CALL_ARG_LOCATION
18197 (note_outside_basic_block_p): Remove all references to
18198 NOTE_INSN_CALL_ARG_LOCATION.
18199 * gengtype.c (adjust_field_rtx_def): Likewise.
18200 * print-rtl.c (rtx_writer::print_rtx_operand_code_0, print_insn):
18202 * jump.c (cleanup_barriers, delete_related_insns): Likewise.
18203 * cfgrtl.c (force_nonfallthru_and_redirect): Likewise.
18206 * builtins.c (builtin_mathfn_code): Don't check if CALL_EXPR_FN (t)
18209 PR tree-optimization/84452
18210 * tree-vect-patterns.c (vect_recog_pow_pattern): Don't call
18211 expand_simd_clones if targetm.simd_clone.compute_vecsize_and_simdlen
18214 2018-02-19 Martin Liska <mliska@suse.cz>
18217 * passes.def: Put pass_sancov_O0 before pass_lower_switch with -O0.
18219 2018-02-19 Martin Liska <mliska@suse.cz>
18220 Richard Sandiford <richard.sandiford@linaro.org>
18222 PR tree-optimization/82491
18223 * gimple-fold.c (get_base_constructor): Make earlier bail out
18226 2018-02-19 Carl Love <cel@us.ibm.com>
18228 * config/rs6000/rs6000-builtin.def: Change NEG macro expansions from
18229 BU_ALTIVEC_A to BU_P8V_AV_1 and BU_ALTIVEC_OVERLOAD_1 to
18231 * config/rs6000/rs6000-c.c: Change ALTIVEC_BUILTIN_VEC_NEG to
18232 P8V_BUILTIN_VEC_NEG.
18234 2018-02-19 Sebastian Perta <sebastian.perta@renesas.com>
18236 * config/rl78/rl78.md (movdf): New define expand.
18238 2018-02-19 Martin Liska <mliska@suse.cz>
18241 * doc/invoke.texi: Fix typo.
18242 * params.def (PARAM_MAX_LOOP_HEADER_INSNS): Likewise.
18244 2018-02-18 Segher Boessenkool <segher@kernel.crashing.org>
18246 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
18247 handle rs6000_single_float and rs6000_double_float specially for
18250 2018-02-16 Jeff Law <law@redhat.com>
18252 * config/rx/rx.c (add_pop_cfi_notes): New function.;
18253 (pop_regs): Use it.
18255 2018-02-16 Jakub Jelinek <jakub@redhat.com>
18258 * ipa-inline.c (inline_small_functions): Fix a typo.
18260 2018-02-16 Nathan Sidwell <nathan@acm.org>
18262 * doc/extend.texi (Backwards Compatibility): Americanize 'behaviour'.
18264 2018-02-16 Carl Love <cel@us.ibm.com>
18266 * config/rs6000/rs6000-builtin.def: Add BU_P8V_VSX_2 macro definition.
18267 Change FLOAT2 expansions from BU_VSX_2 to BU_P8V_VSX_2 and
18268 from BU_VSX_OVERLOAD_2 to BU_P8V_OVERLOAD_2.
18269 * config/rs6000/rs6000-c.c: Changed macro VSX_BUILTIN_VEC_FLOAT2
18270 expansion to P8V_BUILTIN_VEC_FLOAT2.
18272 2018-02-16 Vladimir Makarov <vmakarov@redhat.com>
18274 PR rtl-optimization/70023
18275 * lra-constraints.c (inherit_in_ebb): Take hard reg mode of
18276 src_regno into account.
18278 2018-02-16 Carl Love <cel@us.ibm.com>
18280 * config/rs6000/altivec.h: Remove vec_vextract4b and vec_vinsert4b.
18281 * config/rs6000/rs6000-builtin.def: Remove macro expansion for
18282 VEXTRACT4B, VINSERT4B, VINSERT4B_DI and VEXTRACT4B.
18283 * config/rs6000/rs6000.c: Remove case statements for
18284 P9V_BUILTIN_VEXTRACT4B, P9V_BUILTIN_VEC_VEXTRACT4B,
18285 P9V_BUILTIN_VINSERT4B, P9V_BUILTIN_VINSERT4B_DI,
18286 and P9V_BUILTIN_VEC_VINSERT4B.
18287 * config/rs6000/rs6000-c.c (altivec_expand_builtin): Remove entries for
18288 P9V_BUILTIN_VEC_VEXTRACT4B and P9V_BUILTIN_VEC_VINSERT4B.
18289 * config/rs6000/vsx.md:
18290 * doc/extend.texi: Remove vec_vextract4b, non ABI definitions for
18293 2018-02-16 Carl Love <cel@us.ibm.com>
18295 * config/rs6000/altivec.h: Add builtin names vec_extract4b
18297 * config/rs6000/rs6000-builtin.def: Add INSERT4B and EXTRACT4B
18299 * config/rs6000/rs6000-c.c: Add the definitions for
18300 P9V_BUILTIN_VEC_EXTRACT4B and P9V_BUILTIN_VEC_INSERT4B.
18301 * config/rs6000/rs6000.c (altivec_expand_builtin): Add
18302 P9V_BUILTIN_EXTRACT4B and P9V_BUILTIN_INSERT4B case statements.
18303 * config/rs6000/vsx.md: Add define_insn extract4b. Add define_expand
18304 definition for insert4b and define insn *insert3b_internal.
18305 * doc/extend.texi: Add documentation for vec_extract4b.
18307 2018-02-16 Nathan Sidwell <nathan@acm.org>
18309 * doc/extend.texi (Backwards Compatibility): Mention friend
18310 injection. Note for-scope is deprecated.
18311 * doc/invoke.texi (-ffriend-injection): Deprecate.
18313 2018-02-16 Segher Boessenkool <segher@kernel.crashing.org>
18315 * combine.c (try_combine): When adjusting LOG_LINKS for the destination
18316 that moved to I2, also allow destinations that are a paradoxical
18317 subreg (instead of a normal reg).
18319 2018-02-16 Oleg Endo <olegendo@gcc.gnu.org>
18322 * config/rx/rx.c (rx_fuse_in_memory_bitop): Convert shift operand
18325 2018-02-16 Richard Biener <rguenther@suse.de>
18327 PR tree-optimization/84037
18328 PR tree-optimization/84016
18330 * config/i386/i386.c (ix86_builtin_vectorization_cost):
18331 Adjust vec_construct for the fact we need additional higher latency
18332 128bit inserts for AVX256 and AVX512 vector builds.
18333 (ix86_add_stmt_cost): Scale vector construction cost for
18336 2018-02-16 Richard Biener <rguenther@suse.de>
18338 PR tree-optimization/84417
18339 * tree-ssa.c (non_rewritable_mem_ref_base): Properly constrain
18340 the MEM_REF offset when conversion to BIT_FIELD_REF is desired.
18341 (non_rewritable_lvalue_p): Likewise, use poly-ints.
18343 2018-02-16 Martin Liska <mliska@suse.cz>
18346 * internal-fn.def (ASAN_CHECK): Set proper flags.
18347 (ASAN_MARK): Likewise.
18349 2018-02-16 Julia Koval <julia.koval@intel.com>
18351 * config/i386/i386.c (ix86_option_override_internal): Remove PTA_CLWB
18352 from PTA_CANNONLAKE.
18354 2018-02-16 Jakub Jelinek <jakub@redhat.com>
18357 * config/aarch64/cortex-a57-fma-steering.c (fma_forest::merge_forest):
18358 Use ++iter rather than iter++ for std::list iterators.
18359 (func_fma_steering::dfs): Likewise. Don't delete nodes right away,
18360 defer deleting them until all nodes in the forest are processed. Do
18361 free even leaf nodes. Change to_process into auto_vec.
18364 * system.h (BROKEN_VALUE_INITIALIZATION): Define for GCC < 4.3.
18365 * vec.h (vec_default_construct): Use memset instead of placement new
18366 if BROKEN_VALUE_INITIALIZATION is defined.
18367 * hash-table.h (hash_table<Descriptor, Allocator>::empty_slow): Use
18368 memset instead of value initialization if BROKEN_VALUE_INITIALIZATION
18371 PR rtl-optimization/83723
18372 * lra-int.h (lra_substitute_pseudo): Add DEBUG_P argument.
18373 * lra.c (lra_substitute_pseudo): Likewise. If true, use
18374 gen_rtx_raw_SUBREG instead of gen_rtx_SUBREG. Pass DEBUG_P to
18376 (lra_substitute_pseudo_within_insn): Adjust lra_substitute_pseudo
18378 * lra-constraints.c (inherit_reload_reg, split_reg): Likewise.
18380 2018-02-16 Eric Botcazou <ebotcazou@adacore.com>
18382 PR rtl-optimization/81443
18383 * rtlanal.c (num_sign_bit_copies1) <SUBREG>: Do not propagate results
18384 from inner REGs to paradoxical SUBREGs.
18386 2018-02-16 Richard Biener <rguenther@suse.de>
18388 PR tree-optimization/84399
18389 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
18390 For operands we can analyze at their definition make sure we can
18391 analyze them at each use as well.
18393 2018-02-16 Richard Biener <rguenther@suse.de>
18395 PR tree-optimization/84190
18396 * tree-ssa.c (non_rewritable_mem_ref_base): Do not touch
18397 volatile accesses if the decl isn't volatile.
18399 2018-02-15 Jason Merrill <jason@redhat.com>
18401 PR c++/84314 - ICE with templates and fastcall attribute.
18402 * attribs.c (build_type_attribute_qual_variant): Don't clobber
18403 TYPE_CANONICAL on an existing type.
18405 2018-02-15 Jakub Jelinek <jakub@redhat.com>
18407 PR tree-optimization/84383
18408 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Don't look at
18409 dstoff nor call operand_equal_p if dstbase is NULL.
18411 PR tree-optimization/84334
18412 * match.pd ((A +- CST1) +- CST2 -> A + CST3): If A is
18413 also a CONSTANT_CLASS_P, punt.
18415 2018-02-14 Jim Wilson <jimw@sifive.com>
18417 * config/riscv/riscv.c (riscv_first_stack_step): Move locals after
18418 first SMALL_OPERAND check. New local min_second_step. Move assert
18419 to where locals are set. Add TARGET_RVC support.
18420 * config/riscv/riscv.h (C_SxSP_BITS, SWSP_REACH, SDSP_REACH): New.
18422 2018-02-14 Indu Bhagat <indu.bhagat@oracle.com>
18424 * doc/invoke.texi: Correct -Wformat-overflow code sample.
18426 2018-02-14 Martin Sebor <msebor@redhat.com>
18428 PR tree-optimization/83698
18429 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): For
18430 arrays constrain the offset range to their bounds.
18431 (builtin_access::strcat_overlap): Adjust the bounds of overlap offset.
18432 (builtin_access::overlap): Avoid setting the size of overlap if it's
18434 (maybe_diag_overlap): Also consider arrays when deciding what values
18435 of offsets to include in diagnostics.
18437 2018-02-14 Martin Sebor <msebor@redhat.com>
18440 * attribs.c (diag_attr_exclusions): Consider the exclusion(s)
18441 that correspond to the kind of a declaration.
18443 2018-02-14 John David Anglin <danglin@gcc.gnu.org>
18446 * config/pa/pa.md: Load address of PIC label using the linkage table
18447 if the label is nonlocal.
18449 2018-02-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
18451 * config/rs6000/rs6000.c (rs6000_option_override_internal): Issue
18452 warning message if user requests -maltivec=be.
18453 * doc/invoke.texi: Document deprecation of -maltivec=be.
18455 2018-02-14 Will Schmidt <will_schmidt@vnet.ibm.com>
18458 * config/rs6000/rs6000-c.c: Update definitions for
18459 ALTIVEC_BUILTIN_VEC_SLD, ALTIVEC_BUILTIN_VEC_SLDW,
18460 VEC_XXSLDWI and ALTIVEC_BUILTIN_VEC_XXPERMDI builtins.
18462 2018-02-14 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
18465 * config/i386/cetintrin.h: Remove _rdssp[d|q] and
18466 add _get_ssp intrinsics. Remove argument from
18467 __builtin_ia32_rdssp[d|q].
18468 * config/i386/i386-builtin-types.def: Add UINT_FTYPE_VOID.
18469 * config/i386/i386-builtin.def: Remove argument from
18470 __builtin_ia32_rdssp[d|q].
18471 * config/i386/i386.c: Use UINT_FTYPE_VOID. Use
18472 ix86_expand_special_args_builtin for _rdssp[d|q].
18473 * config/i386/i386.md: Remove argument from rdssp[si|di] insn.
18474 Clear register before usage.
18475 * doc/extend.texi: Remove argument from __builtin_ia32_rdssp[d|q].
18476 Add documentation for new _get_ssp and _inc_ssp intrinsics.
18478 2018-02-14 Richard Sandiford <richard.sandiford@linaro.org>
18480 PR tree-optimization/84357
18481 * tree-data-ref.c (object_address_invariant_in_loop_p): Check
18482 operand 1 of an ARRAY_REF too.
18484 2018-02-14 Oleg Endo <olegendo@gcc.gnu.org>
18487 * config/rx/rx-protos.h (rx_reg_dead_or_unused_after_insn,
18488 rx_copy_reg_dead_or_unused_notes, rx_fuse_in_memory_bitop): New
18490 (set_of_reg): New struct.
18491 (rx_find_set_of_reg, rx_find_use_of_reg): New functions.
18492 * config/rx/rx.c (rx_reg_dead_or_unused_after_insn,
18493 rx_copy_reg_dead_or_unused_notes, rx_fuse_in_memory_bitop): New
18495 * config/rx/rx.md (andsi3, iorsi3, xorsi3): Convert to insn_and_split.
18496 Split into bitclr, bitset, bitinvert patterns if appropriate.
18497 (*bitset, *bitinvert, *bitclr): Convert to named insn_and_split and
18498 use rx_fuse_in_memory_bitop.
18499 (*bitset_in_memory, *bitinvert_in_memory, *bitclr_in_memory): Convert
18500 to named insn, correct maximum insn length.
18502 2018-02-14 Jozef Lawrynowicz <jozefl.gcc@gmail.com>
18505 * machmode.def: Define a complex mode for PARTIAL_INT.
18506 * genmodes.c (complex_class): Return MODE_COMPLEX_INT for
18508 * doc/rtl.texi: Document CSPImode.
18509 * config/msp430/msp430.c (msp430_hard_regno_nregs): Add CPSImode
18511 (msp430_hard_regno_nregs_with_padding): Likewise.
18513 2018-02-13 Peter Bergner <bergner@vnet.ibm.com>
18516 * config/rs6000/rs6000.c (mem_operand_gpr): Disallow altivec addresses.
18518 2018-02-13 Segher Boessenkool <segher@kernel.crashing.org>
18520 PR rtl-optimization/84169
18521 * combine.c (try_combine): New variable split_i2i3. Set it to true if
18522 we generated a parallel as new i3 and we split that to new i2 and i3
18523 instructions. Handle split_i2i3 similar to swap_i2i3: scan the
18524 LOG_LINKs of i3 to see which of those need to link to i2 now. Link
18525 those to i2, not i1. Partially rewrite this scan code.
18527 2018-02-13 Jakub Jelinek <jakub@redhat.com>
18530 * stor-layout.c (place_field): For variable length fields, adjust
18531 offset_align afterwards not just based on the field's alignment,
18532 but also on the size.
18534 PR middle-end/84309
18535 * match.pd (pow(C,x) -> exp(log(C)*x)): Use exp2s and log2s instead
18536 of exps and logs in the use_exp2 case.
18538 2018-02-13 Jeff Law <law@redhat.com>
18540 * config/rl/rl78.c (rl78_attribute_table): Fix terminator and
18541 entry for "vector".
18543 * config/rl78/rl78.c (rl78_handle_func_attribute): Mark
18546 2018-02-13 Alexandre Oliva <aoliva@redhat.com>
18550 * common.opt (gas-loc-support, gas-locview-support): New.
18551 (ginline-points, ginternal-reset-location-views): New.
18552 * doc/invoke.texi: Document them. Use @itemx where intended.
18553 (gvariable-location-views): Adjust.
18554 * target.def (reset_location_view): New.
18555 * doc/tm.texi.in (DWARF2_ASM_VIEW_DEBUG_INFO): New.
18556 (TARGET_RESET_LOCATION_VIEW): New.
18557 * doc/tm.texi: Rebuilt.
18558 * dwarf2out.c (dwarf2out_default_as_loc_support): New.
18559 (dwarf2out_default_as_locview_support): New.
18560 (output_asm_line_debug_info): Use option variables.
18561 (dwarf2out_maybe_output_loclist_view_pair): Likewise.
18562 (output_loc_list): Likewise.
18563 (add_high_low_attributes): Check option variables.
18564 Don't output entry view attribute in strict mode.
18565 (gen_inlined_subroutine_die): Check option variables.
18566 (dwarf2out_inline_entry): Likewise.
18567 (init_sections_and_labels): Likewise.
18568 (dwarf2out_early_finish): Likewise.
18569 (maybe_reset_location_view): New, from...
18570 (dwarf2out_var_location): ... here. Call it.
18571 * debug.h (dwarf2out_default_as_loc_support): Declare.
18572 (dwarf2out_default_as_locview_support): Declare.
18573 * hooks.c (hook_int_rtx_insn_0): New.
18574 * hooks.h (hook_int_rtx_insn_0): Declare.
18575 * toplev.c (process_options): Take -gas-loc-support and
18576 -gas-locview-support from dwarf2out. Enable
18577 -gvariable-location-views by default only with locview
18578 assembler support. Enable -ginternal-reset-location-views by
18579 default only if the target defines the corresponding hook.
18580 Enable -ginline-points by default if location views are
18581 enabled; force it disabled if statement frontiers are
18583 * tree-inline.c (expand_call_inline): Check option variables.
18584 * tree-ssa-live.c (remove_unused_scope_block_p): Likewise.
18586 2018-02-13 Richard Sandiford <richard.sandiford@linaro.org>
18588 PR tree-optimization/84321
18589 * tree-vrp.c (intersect_range_with_nonzero_bits): Fix VR_ANTI_RANGE
18590 handling. Also check whether the anti-range contains any values
18591 that satisfy the mask; switch to a VR_RANGE if not.
18593 2018-02-13 Paolo Bonzini <bonzini@gnu.org>
18596 * internal-fn.def (ASAN_CHECK, ASAN_MARK): Revert changes to fnspec.
18598 2018-02-13 Martin Jambor <mjambor@suse.cz>
18601 * ipa-param-manipulation.c (ipa_modify_call_arguments): Use location
18602 of call statements, also set location of a load to a temporary.
18604 2018-02-13 Sebastian Perta <sebastian.perta@renesas.com>
18606 * config/rl78/rl78.c (add_vector_labels): New function.
18607 * config/rl78/rl78.c (rl78_handle_vector_attribute): New function.
18608 * config/rl78/rl78.c (rl78_start_function): Call add_vector_labels.
18609 * config/rl78/rl78.c (rl78_handle_func_attribute): Removed the assert
18610 which checks that no arguments are passed.
18611 * config/rl78/rl78.c (rl78_attribute_table): Add "vector" attribute.
18612 * doc/extend.texi: Documentation for the new attribute.
18614 2018-02-13 Andreas Schwab <schwab@suse.de>
18616 * config/riscv/linux.h (CPP_SPEC): Define.
18618 2018-02-13 Jakub Jelinek <jakub@redhat.com>
18621 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Pass
18622 OPTION_MASK_ISA_AES | OPTION_MASK_ISA_SSE2 instead of
18623 OPTION_MASK_ISA_AES as first argument to def_builtin_const
18624 for AES builtins. Pass OPTION_MASK_ISA_PCLMUL | OPTION_MASK_ISA_SSE2
18625 instead of OPTION_MASK_ISA_PCLMUL as first argument to
18626 def_builtin_const for __builtin_ia32_pclmulqdq128 builtin.
18627 * config/i386/wmmintrin.h: If __SSE2__ is not defined, enable it
18628 temporarily for AES and PCLMUL builtins.
18630 PR tree-optimization/84339
18631 * gimple-fold.c (get_range_strlen): Set *FLEXP to true when handling
18632 ARRAY_REF where first operand is array_at_struct_end_p COMPONENT_REF.
18635 PR middle-end/84309
18636 * match.pd (pow(C,x) -> exp(log(C)*x)): Optimize instead into
18637 exp2(log2(C)*x) if C is a power of 2 and c99 runtime is available.
18638 * generic-match-head.c (canonicalize_math_after_vectorization_p): New
18640 * gimple-match-head.c (canonicalize_math_after_vectorization_p): New
18642 * omp-simd-clone.h: New file.
18643 * omp-simd-clone.c: Include omp-simd-clone.h.
18644 (expand_simd_clones): No longer static.
18645 * tree-vect-patterns.c: Include fold-const-call.h, attribs.h,
18646 cgraph.h and omp-simd-clone.h.
18647 (vect_recog_pow_pattern): Optimize pow(C,x) to exp(log(C)*x).
18648 (vect_recog_widen_shift_pattern): Formatting fix.
18649 (vect_pattern_recog_1): Don't check optab for calls.
18652 * config/i386/sse.md (<avx512>_vpermi2var<mode>3_mask): Force
18653 operands[2] into a REG before using gen_lowpart on it.
18655 2018-02-12 Jeff Law <law@redhat.com>
18658 * config/sh/sh.c (find_barrier): Consider a sibling call
18661 * cse.c (try_back_substitute_reg): Move any REG_ARGS_SIZE note when
18662 successfully back substituting a reg.
18664 2018-02-12 Richard Biener <rguenther@suse.de>
18666 PR tree-optimization/84037
18667 * tree-vect-slp.c (vect_analyze_slp_cost): Add visited
18668 parameter, move visited init to caller.
18669 (vect_slp_analyze_operations): Separate cost from validity
18670 check, initialize visited once for all instances.
18671 (vect_schedule_slp): Analyze map to CSE vectorized nodes once
18673 * tree-vect-stmts.c (vect_model_simple_cost): Make early
18675 (vect_model_promotion_demotion_cost): Likewise.
18676 (vectorizable_bswap): Guard cost modeling with !slp_node
18677 instead of !PURE_SLP_STMT to avoid double-counting on hybrid
18679 (vectorizable_call): Likewise.
18680 (vectorizable_conversion): Likewise.
18681 (vectorizable_assignment): Likewise.
18682 (vectorizable_shift): Likewise.
18683 (vectorizable_operation): Likewise.
18684 (vectorizable_store): Likewise.
18685 (vectorizable_load): Likewise.
18686 (vectorizable_condition): Likewise.
18687 (vectorizable_comparison): Likewise.
18689 2018-02-12 Paolo Bonzini <bonzini@gnu.org>
18692 * internal-fn.def (ASAN_CHECK): Fix fnspec to account for return value.
18693 (ASAN_MARK): Fix fnspec to account for return value, change pointer
18694 argument from 'R' to 'W' so that the pointed-to datum is clobbered.
18696 2018-02-08 Jan Hubicka <hubicka@ucw.cz>
18698 PR middle-end/83665
18699 * params.def (inline-min-speedup): Increase from 8 to 15.
18700 (max-inline-insns-auto): Decrease from 40 to 30.
18701 * ipa-split.c (consider_split): Add some buffer for function to
18702 be considered inlining candidate.
18703 * invoke.texi (max-inline-insns-auto, inline-min-speedup): UPdate
18706 2018-02-12 Richard Biener <rguenther@suse.de>
18708 PR tree-optimization/84037
18709 * tree-vect-slp.c (vect_build_slp_tree_2): Try swapping the
18710 matched stmts if we cannot swap the non-matched ones.
18712 2018-02-12 Olga Makhotina <olga.makhotina@intel.com>
18714 * config/i386/avx512fintrin.h (_mm_mask_scalef_round_sd,
18715 _mm_maskz_scalef_round_sd, _mm_mask_scalef_round_ss,
18716 _mm_maskz_scalef_round_ss): New intrinsics.
18717 (__builtin_ia32_scalefsd_round, __builtin_ia32_scalefss_round): Fix.
18718 * config/i386/i386-builtin.def (__builtin_ia32_scalefsd_round,
18719 __builtin_ia32_scalefss_round): Remove.
18720 (__builtin_ia32_scalefsd_mask_round,
18721 __builtin_ia32_scalefss_mask_round): New intrinsics.
18722 * config/i386/sse.md (vmscalef<mode><round_name>): Renamed to ...
18723 (vmscalef<mode><mask_scalar_name><round_scalar_name>): ... this.
18724 ((match_operand:VF_128 2 "<round_nimm_predicate>"
18725 "<round_constraint>")): Changed to ...
18726 ((match_operand:VF_128 2 "<round_scalar_nimm_predicate>"
18727 "<round_scalar_constraint>")): ... this.
18728 ("vscalef<ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|
18729 %0, %1, %2<round_op3>}"): Changed to ...
18730 ("vscalef<ssescalarmodesuffix>\t{<round_scalar_mask_op3>%2, %1,
18731 %0<mask_scalar_operand3>|%0<mask_scalar_operand3>, %1,
18732 %2<round_scalar_mask_op3>}"): ... this.
18733 * config/i386/subst.md (round_scalar_nimm_predicate): New.
18735 2018-02-12 Olga Makhotina <olga.makhotina@intel.com>
18737 * config/i386/avx512fintrin.h (_mm_mask_sqrt_round_sd)
18738 (_mm_maskz_sqrt_round_sd, _mm_mask_sqrt_round_ss)
18739 (_mm_maskz_sqrt_round_ss): New intrinsics.
18740 (__builtin_ia32_sqrtsd_round, __builtin_ia32_sqrtss_round): Remove.
18741 (__builtin_ia32_sqrtsd_mask_round)
18742 (__builtin_ia32_sqrtss_mask_round): New builtins.
18743 * config/i386/i386-builtin.def (__builtin_ia32_sqrtsd_round)
18744 (__builtin_ia32_sqrtss_round): Remove.
18745 (__builtin_ia32_sqrtsd_mask_round)
18746 (__builtin_ia32_sqrtss_mask_round): New builtins.
18747 * config/i386/sse.md (vmsqrt<mode>2<round_name>): Renamed to ...
18748 (vmsqrt<mode>2<mask_scalar_name><round_scalar_name>): ... this.
18749 ((match_operand:VF_128 1 "vector_operand"
18750 "xBm,<round_constraint>")): Changed to ...
18751 ((match_operand:VF_128 1 "vector_operand"
18752 "xBm,<round_scalar_constraint>")): ... this.
18753 (vsqrt<ssescalarmodesuffix>\t{<round_op3>%1, %2, %0|
18754 %0, %2, %<iptr>1<round_op3>}): Changed to ...
18755 (vsqrt<ssescalarmodesuffix>\t{<round_scalar_mask_op3>%1, %2,
18756 %0<mask_scalar_operand3>|%0<mask_scalar_operand3>, %2,
18757 %<iptr>1<round_scalar_mask_op3>}): ... this.
18758 ((set_attr "prefix" "<round_prefix>")): Changed to ...
18759 ((set_attr "prefix" "<round_scalar_prefix>")): ... this.
18761 2018-02-11 Steven Munroe <munroesj@gcc.gnu.org>
18764 * config/rs6000/mmintrin.h (_mm_cmpeq_pi32 [_ARCH_PWR9]):
18765 Cast vec_cmpeq result to correct type.
18766 * config/rs6000/mmintrin.h (_mm_cmpgt_pi32 [_ARCH_PWR9]):
18767 Cast vec_cmpgt result to correct type.
18769 2018-02-11 Alexandre Oliva <aoliva@redhat.com>
18771 * final.c (final_scan_insn_1): Renamed from...
18772 (final_scan_insn): ... this. New wrapper, to recover
18773 seen from the outermost call in recursive ones.
18774 * config/sparc/sparc.c (output_return): Drop seen from call.
18775 (output_sibcall): Likewise.
18776 * config/visium/visium.c (output_branch): Likewise.
18778 2018-02-10 John David Anglin <danglin@gcc.gnu.org>
18780 * config/pa/pa.c (hppa_profile_hook): Mark SYMBOL_REF for _mcount as
18783 2018-02-10 Alan Modra <amodra@gmail.com>
18786 * config/rs6000/rs6000.md (split_stack_return): Remove (use ..).
18787 Specify LR as an input.
18789 2018-02-10 Jakub Jelinek <jakub@redhat.com>
18792 * omp-low.c (maybe_remove_omp_member_access_dummy_vars,
18793 remove_member_access_dummy_vars): New functions.
18794 (lower_omp_for, lower_omp_taskreg, lower_omp_target,
18795 lower_omp_1, execute_lower_omp): Use them.
18797 PR rtl-optimization/84308
18798 * shrink-wrap.c (spread_components): Release todo vector.
18800 2018-02-09 Vladimir Makarov <vmakarov@redhat.com>
18802 PR rtl-optimization/57193
18803 * ira-color.c (struct allocno_color_data): Add member
18804 conflict_allocno_hard_prefs.
18805 (update_conflict_allocno_hard_prefs): New.
18806 (bucket_allocno_compare_func): Add a preference based on
18807 conflict_allocno_hard_prefs.
18808 (push_allocno_to_stack): Update conflict_allocno_hard_prefs.
18809 (color_allocnos): Remove a dead code. Initiate
18810 conflict_allocno_hard_prefs. Call update_costs_from_prefs.
18812 2018-02-09 Jakub Jelinek <jakub@redhat.com>
18815 * config/rs6000/vsx.md (p9_xxbrq_v16qi): Change input operand
18816 constraint from =wa to wa. Avoid a subreg on the output operand,
18817 instead use a pseudo and subreg it in a move.
18818 (p9_xxbrd_<mode>): Changed to ...
18819 (p9_xxbrd_v2di): ... this insn, without VSX_D iterator.
18820 (p9_xxbrd_v2df): New expander.
18821 (p9_xxbrw_<mode>): Changed to ...
18822 (p9_xxbrw_v4si): ... this insn, without VSX_W iterator.
18823 (p9_xxbrw_v4sf): New expander.
18825 2018-02-09 Sebastian Perta <sebastian.perta@renesas.com>
18827 * config/rx/rx.md (movsicc): Update expander to be matched by GCC.
18829 2018-02-09 Peter Bergner <bergner@vnet.ibm.com>
18832 * config/rs6000/vsx.md (vsx_mul_v2di): Handle generating a 64-bit
18833 multiply in 32-bit mode.
18834 (vsx_div_v2di): Handle generating a 64-bit signed divide in 32-bit mode.
18835 (vsx_udiv_v2di): Handle generating a 64-bit unsigned divide in 32-bit
18838 2018-02-09 Sebastian Perta <sebastian.perta@renesas.com>
18840 * config/rx/constraints.md (CALL_OP_SYMBOL_REF): Added new constraint
18841 to allow or block "symbol_ref" depending on the value of TARGET_JSR.
18842 * config/rx/rx.md (call_internal): Use CALL_OP_SYMBOL_REF.
18843 * config/rx/rx.md (call_value_internal): Use CALL_OP_SYMBOL_REF.
18845 2018-02-09 Pierre-Marie de Rodat <derodat@adacore.com>
18848 * dwarf2out.c (is_trivial_indirect_ref): New function.
18849 (dwarf2out_late_global_decl): Do not generate a location
18850 attribute for variables that have a non-trivial DECL_VALUE_EXPR
18851 and that are not defined in the current unit.
18853 2018-02-09 Eric Botcazou <ebotcazou@adacore.com>
18855 * optabs.c (prepare_cmp_insn): Try harder to emit a direct comparison
18856 instead of a libcall for UNORDERED.
18858 2018-02-09 Tamar Christina <tamar.christina@arm.com>
18861 * config/arm/arm-c.c (arm_cpu_builtins): Un-define __ARM_FEATURE_LDREX,
18862 __ARM_ARCH_PROFILE, __ARM_ARCH_ISA_THUMB, __ARM_FP and __ARM_NEON_FP.
18864 2018-02-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18867 * config/s390/s390.c (s390_set_current_function): Invoke
18868 s390_indirect_branch_settings also if fndecl didn't change.
18870 2018-02-09 Alexandre Oliva <aoliva@redhat.com>
18872 * config/rs6000/rs6000.md (blockage): Set length to zero.
18874 2018-02-09 Eric Botcazou <ebotcazou@adacore.com>
18876 * expr.c (optimize_bitfield_assignment_op): Remove obsolete assertion.
18878 2018-02-09 Jakub Jelinek <jakub@redhat.com>
18881 * gcc.c (STATIC_LIBASAN_LIBS, STATIC_LIBTSAN_LIBS,
18882 STATIC_LIBLSAN_LIBS, STATIC_LIBUBSAN_LIBS): Handle -static like
18886 * var-tracking.c (vt_add_function_parameter): Punt for non-onepart
18887 PARALLEL incoming that failed vt_get_decl_and_offset check.
18889 PR middle-end/84237
18890 * output.h (bss_initializer_p): Add NAMED argument, defaulted to false.
18891 * varasm.c (bss_initializer_p): Add NAMED argument, if true, ignore
18893 (get_variable_section): For decls in named .bss* sections pass true as
18894 second argument to bss_initializer_p.
18896 2018-02-09 Marek Polacek <polacek@redhat.com>
18897 Jakub Jelinek <jakub@redhat.com>
18900 * fold-const.c (fold_indirect_ref_1): Use VECTOR_TYPE_P macro.
18901 Formatting fixes. Verify first that tree_fits_poly_int64_p (op01).
18902 Sync some changes from cxx_fold_indirect_ref.
18904 2018-02-09 Alexandre Oliva <aoliva@redhat.com>
18906 * cfgexpand.c (expand_gimple_basic_block): Handle inline entry
18908 * dwarf2out.c (dwarf2_debug_hooks): Enable inline_entry hook.
18909 (BLOCK_INLINE_ENTRY_LABEL): New.
18910 (dwarf2out_var_location): Disregard inline entry markers.
18911 (inline_entry_data): New struct.
18912 (inline_entry_data_hasher): New hashtable type.
18913 (inline_entry_data_hasher::hash): New.
18914 (inline_entry_data_hasher::equal): New.
18915 (inline_entry_data_table): New variable.
18916 (add_high_low_attributes): Add DW_AT_entry_pc and
18917 DW_AT_GNU_entry_view attributes if a pending entry is found
18918 in inline_entry_data_table. Add old entry_pc attribute only
18919 if debug nonbinding markers are disabled.
18920 (gen_inlined_subroutine_die): Set BLOCK_DIE if nonbinding
18921 markers are enabled.
18922 (block_within_block_p, dwarf2out_inline_entry): New.
18923 (dwarf2out_finish): Check that no entries remained in
18924 inline_entry_data_table.
18925 * final.c (reemit_insn_block_notes): Handle inline entry notes.
18926 (final_scan_insn, notice_source_line): Likewise.
18927 (rest_of_clean_state): Skip inline entry markers.
18928 * gimple-pretty-print.c (dump_gimple_debug): Handle inline entry
18930 * gimple.c (gimple_build_debug_inline_entry): New.
18931 * gimple.h (enum gimple_debug_subcode): Add
18932 GIMPLE_DEBUG_INLINE_ENTRY.
18933 (gimple_build_debug_inline_entry): Declare.
18934 (gimple_debug_inline_entry_p): New.
18935 (gimple_debug_nonbind_marker_p): Adjust.
18936 * insn-notes.def (INLINE_ENTRY): New.
18937 * print-rtl.c (rtx_writer::print_rtx_operand_code_0): Handle
18938 inline entry marker notes.
18939 (print_insn): Likewise.
18940 * rtl.h (NOTE_MARKER_P): Add INLINE_ENTRY support.
18941 (INSN_DEBUG_MARKER_KIND): Likewise.
18942 (GEN_RTX_DEBUG_MARKER_INLINE_ENTRY_PAT): New.
18943 * tree-inline.c (expand_call_inline): Build and insert
18944 debug_inline_entry stmt.
18945 * tree-ssa-live.c (remove_unused_scope_block_p): Preserve
18946 inline entry blocks early, if nonbind markers are enabled.
18947 (dump_scope_block): Dump fragment info.
18948 * var-tracking.c (reemit_marker_as_note): Handle inline entry note.
18949 * doc/gimple.texi (gimple_debug_inline_entry_p): New.
18950 (gimple_build_debug_inline_entry): New.
18951 * doc/invoke.texi (gstatement-frontiers, gno-statement-frontiers):
18952 Enable/disable inline entry points too.
18953 * doc/rtl.texi (NOTE_INSN_INLINE_ENTRY): New.
18954 (DEBUG_INSN): Describe inline entry markers.
18956 * common.opt (gvariable-location-views): New.
18957 (gvariable-location-views=incompat5): New.
18958 * config.in: Rebuilt.
18959 * configure: Rebuilt.
18960 * configure.ac: Test assembler for view support.
18961 * dwarf2asm.c (dw2_asm_output_symname_uleb128): New.
18962 * dwarf2asm.h (dw2_asm_output_symname_uleb128): Declare.
18963 * dwarf2out.c (var_loc_view): New typedef.
18964 (struct dw_loc_list_struct): Add vl_symbol, vbegin, vend.
18965 (dwarf2out_locviews_in_attribute): New.
18966 (dwarf2out_locviews_in_loclist): New.
18967 (dw_val_equal_p): Compare val_view_list of dw_val_class_view_lists.
18968 (enum dw_line_info_opcode): Add LI_adv_address.
18969 (struct dw_line_info_table): Add view.
18970 (RESET_NEXT_VIEW, RESETTING_VIEW_P): New macros.
18971 (DWARF2_ASM_VIEW_DEBUG_INFO): Define default.
18972 (zero_view_p): New variable.
18973 (ZERO_VIEW_P): New macro.
18974 (output_asm_line_debug_info): New.
18975 (struct var_loc_node): Add view.
18976 (add_AT_view_list, AT_loc_list): New.
18977 (add_var_loc_to_decl): Add view param. Test it against last.
18978 (new_loc_list): Add view params. Record them.
18979 (AT_loc_list_ptr): Handle loc and view lists.
18980 (view_list_to_loc_list_val_node): New.
18981 (print_dw_val): Handle dw_val_class_view_list.
18982 (size_of_die): Likewise.
18983 (value_format): Likewise.
18984 (loc_list_has_views): New.
18985 (gen_llsym): Set vl_symbol too.
18986 (maybe_gen_llsym, skip_loc_list_entry): New.
18987 (dwarf2out_maybe_output_loclist_view_pair): New.
18988 (output_loc_list): Output view list or entries too.
18989 (output_view_list_offset): New.
18990 (output_die): Handle dw_val_class_view_list.
18991 (output_dwarf_version): New.
18992 (output_compilation_unit_header): Use it.
18993 (output_skeleton_debug_sections): Likewise.
18994 (output_rnglists, output_line_info): Likewise.
18995 (output_pubnames, output_aranges): Update version comments.
18996 (output_one_line_info_table): Output view numbers in asm comments.
18997 (dw_loc_list): Determine current endview, pass it to new_loc_list.
18998 Call maybe_gen_llsym.
18999 (loc_list_from_tree_1): Adjust.
19000 (add_AT_location_description): Create view list attribute if
19001 needed, check it's absent otherwise.
19002 (convert_cfa_to_fb_loc_list): Adjust.
19003 (maybe_emit_file): Call output_asm_line_debug_info for test.
19004 (dwarf2out_var_location): Reset views as needed. Precompute
19005 add_var_loc_to_decl args. Call get_attr_min_length only if we have the
19006 attribute. Set view.
19007 (new_line_info_table): Reset next view.
19008 (set_cur_line_info_table): Call output_asm_line_debug_info for test.
19009 (dwarf2out_source_line): Likewise. Output view resets and labels to
19010 the assembler, or select appropriate line info opcodes.
19011 (prune_unused_types_walk_attribs): Handle dw_val_class_view_list.
19012 (optimize_string_length): Catch it. Adjust.
19013 (resolve_addr): Copy vl_symbol along with ll_symbol. Handle
19014 dw_val_class_view_list, and remove it if no longer needed.
19015 (hash_loc_list): Hash view numbers.
19016 (loc_list_hasher::equal): Compare them.
19017 (optimize_location_lists): Check whether a view list symbol is
19018 needed, and whether the locview attribute is present, and
19019 whether they match. Remove the locview attribute if no longer
19021 (index_location_lists): Call skip_loc_list_entry for test.
19022 (dwarf2out_finish): Call output_asm_line_debug_info for test.
19023 Use output_dwarf_version.
19024 * dwarf2out.h (enum dw_val_class): Add dw_val_class_view_list.
19025 (struct dw_val_node): Add val_view_list.
19026 * final.c (SEEN_NEXT_VIEW): New.
19027 (set_next_view_needed): New.
19028 (clear_next_view_needed): New.
19029 (maybe_output_next_view): New.
19030 (final_start_function): Rename to...
19031 (final_start_function_1): ... this. Take pointer to FIRST,
19032 add SEEN parameter. Emit param bindings in the initial view.
19033 (final_start_function): Reintroduce SEEN-less interface.
19034 (final): Rename to...
19035 (final_1): ... this. Take SEEN parameter. Output final pending
19036 next view at the end.
19037 (final): Reintroduce seen-less interface.
19038 (final_scan_insn): Output pending next view before switching
19039 sections or ending a block. Mark the next view as needed when
19040 outputting variable locations. Notify debug backend of section
19041 changes, and of location view changes.
19042 (rest_of_handle_final): Adjust.
19043 * toplev.c (process_options): Autodetect value for debug variable
19044 location views option. Warn on incompat5 without -gdwarf-5.
19045 * doc/invoke.texi (gvariable-location-views): New.
19046 (gvariable-location-views=incompat5): New.
19047 (gno-variable-location-views): New.
19049 2018-02-08 David Malcolm <dmalcolm@redhat.com>
19051 PR tree-optimization/84136
19052 * tree-cfg.c (find_taken_edge_computed_goto): Remove assertion
19053 that the result of find_edge is non-NULL.
19055 2018-02-08 Sergey Shalnov <sergey.shalnov@intel.com>
19058 * config/i386/x86-tune-costs.h (skylake_cost): Fix cost of
19059 storing integer register in SImode. Fix cost of 256 and 512
19060 byte aligned SSE register store.
19062 2018-02-08 Sergey Shalnov <sergey.shalnov@intel.com>
19064 * config/i386/i386.c (ix86_multiplication_cost): Fix
19065 multiplication cost for TARGET_AVX512DQ.
19067 2018-02-08 Marek Polacek <polacek@redhat.com>
19069 PR tree-optimization/84238
19070 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Verify the result of
19073 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
19075 PR tree-optimization/84265
19076 * tree-vect-stmts.c (vectorizable_store): Don't treat
19077 VMAT_CONTIGUOUS accesses as grouped.
19078 (vectorizable_load): Likewise.
19080 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
19082 PR tree-optimization/81635
19083 * wide-int.h (wi::round_down_for_mask, wi::round_up_for_mask): Declare.
19084 * wide-int.cc (wi::round_down_for_mask, wi::round_up_for_mask)
19085 (test_round_for_mask): New functions.
19086 (wide_int_cc_tests): Call test_round_for_mask.
19087 * tree-vrp.h (intersect_range_with_nonzero_bits): Declare.
19088 * tree-vrp.c (intersect_range_with_nonzero_bits): New function.
19089 * tree-data-ref.c (split_constant_offset_1): Use it to refine the
19090 range returned by get_range_info.
19092 2018-02-08 Jan Hubicka <hubicka@ucw.cz>
19095 * cgraph.h (symtab_node::output_to_lto_symbol_table_p): Declare
19096 * symtab.c: Include builtins.h
19097 (symtab_node::output_to_lto_symbol_table_p): Move here
19098 from lto-streamer-out.c:output_symbol_p.
19099 * lto-streamer-out.c (write_symbol): Turn early exit to assert.
19100 (output_symbol_p): Move all logic to symtab.c
19101 (produce_symtab): Update.
19103 2018-02-08 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19105 * config/s390/s390-opts.h (enum indirect_branch): Define.
19106 * config/s390/s390-protos.h (s390_return_addr_from_memory)
19107 (s390_indirect_branch_via_thunk)
19108 (s390_indirect_branch_via_inline_thunk): Add function prototypes.
19109 (enum s390_indirect_branch_type): Define.
19110 * config/s390/s390.c (struct s390_frame_layout, struct
19111 machine_function): Remove.
19112 (indirect_branch_prez10thunk_mask, indirect_branch_z10thunk_mask)
19113 (indirect_branch_table_label_no, indirect_branch_table_name):
19115 (INDIRECT_BRANCH_NUM_OPTIONS): Define macro.
19116 (enum s390_indirect_branch_option): Define.
19117 (s390_return_addr_from_memory): New function.
19118 (s390_handle_string_attribute): New function.
19119 (s390_attribute_table): Add new attribute handler.
19120 (s390_execute_label): Handle UNSPEC_EXECUTE_JUMP patterns.
19121 (s390_indirect_branch_via_thunk): New function.
19122 (s390_indirect_branch_via_inline_thunk): New function.
19123 (s390_function_ok_for_sibcall): When jumping via thunk disallow
19124 sibling call optimization for non z10 compiles.
19125 (s390_emit_call): Force indirect branch target to be a single
19126 register. Add r1 clobber for non-z10 compiles.
19127 (s390_emit_epilogue): Emit return jump via return_use expander.
19128 (s390_reorg): Handle JUMP_INSNs as execute targets.
19129 (s390_option_override_internal): Perform validity checks for the
19130 new command line options.
19131 (s390_indirect_branch_attrvalue): New function.
19132 (s390_indirect_branch_settings): New function.
19133 (s390_set_current_function): Invoke s390_indirect_branch_settings.
19134 (s390_output_indirect_thunk_function): New function.
19135 (s390_code_end): Implement target hook.
19136 (s390_case_values_threshold): Implement target hook.
19137 (TARGET_ASM_CODE_END, TARGET_CASE_VALUES_THRESHOLD): Define target
19139 * config/s390/s390.h (struct s390_frame_layout)
19140 (struct machine_function): Move here from s390.c.
19141 (TARGET_INDIRECT_BRANCH_NOBP_RET)
19142 (TARGET_INDIRECT_BRANCH_NOBP_JUMP)
19143 (TARGET_INDIRECT_BRANCH_NOBP_JUMP_THUNK)
19144 (TARGET_INDIRECT_BRANCH_NOBP_JUMP_INLINE_THUNK)
19145 (TARGET_INDIRECT_BRANCH_NOBP_CALL)
19146 (TARGET_DEFAULT_INDIRECT_BRANCH_TABLE)
19147 (TARGET_INDIRECT_BRANCH_THUNK_NAME_EXRL)
19148 (TARGET_INDIRECT_BRANCH_THUNK_NAME_EX)
19149 (TARGET_INDIRECT_BRANCH_TABLE): Define macros.
19150 * config/s390/s390.md (UNSPEC_EXECUTE_JUMP)
19151 (INDIRECT_BRANCH_THUNK_REGNUM): Define constants.
19152 (mnemonic attribute): Add values which aren't recognized
19154 ("*cjump_long", "*icjump_long", "*basr", "*basr_r"): Disable
19155 pattern for branch conversion. Fix mnemonic attribute.
19156 ("*c<code>", "*sibcall_br", "*sibcall_value_br", "*return"): Emit
19157 indirect branch via thunk if requested.
19158 ("indirect_jump", "<code>"): Expand patterns for branch conversion.
19159 ("*indirect_jump"): Disable for branch conversion using out of
19161 ("indirect_jump_via_thunk<mode>_z10")
19162 ("indirect_jump_via_thunk<mode>")
19163 ("indirect_jump_via_inlinethunk<mode>_z10")
19164 ("indirect_jump_via_inlinethunk<mode>", "*casesi_jump")
19165 ("casesi_jump_via_thunk<mode>_z10", "casesi_jump_via_thunk<mode>")
19166 ("casesi_jump_via_inlinethunk<mode>_z10")
19167 ("casesi_jump_via_inlinethunk<mode>", "*basr_via_thunk<mode>_z10")
19168 ("*basr_via_thunk<mode>", "*basr_r_via_thunk_z10")
19169 ("*basr_r_via_thunk", "return<mode>_prez10"): New pattern.
19170 ("*indirect2_jump"): Disable for branch conversion.
19171 ("casesi_jump"): Turn into expander and expand patterns for branch
19173 ("return_use"): New expander.
19174 ("*return"): Emit return via thunk and rename it to ...
19175 ("*return<mode>"): ... this one.
19176 * config/s390/s390.opt: Add new options and and enum for the
19179 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
19181 * lra-constraints.c (match_reload): Unconditionally use
19182 gen_lowpart_SUBREG, rather than selecting between that
19183 and equivalent gen_rtx_SUBREG code.
19185 2018-02-08 Richard Biener <rguenther@suse.de>
19187 PR tree-optimization/84233
19188 * tree-ssa-phiprop.c (propagate_with_phi): Use separate
19189 changed flag instead of boguously re-using phi_inserted.
19191 2018-02-08 Martin Jambor <mjambor@suse.cz>
19193 * hsa-gen.c (get_symbol_for_decl): Set program allocation for
19194 static local variables.
19196 2018-02-08 Richard Biener <rguenther@suse.de>
19198 PR tree-optimization/84278
19199 * tree-vect-stmts.c (vectorizable_store): When looking for
19200 smaller vector types to perform grouped strided loads/stores
19201 make sure the mode is supported by the target.
19202 (vectorizable_load): Likewise.
19204 2018-02-08 Wilco Dijkstra <wdijkstr@arm.com>
19206 * config/aarch64/aarch64.c (aarch64_components_for_bb):
19207 Increase LDP/STP opportunities by adding adjacent callee-saves.
19209 2018-02-08 Wilco Dijkstra <wdijkstr@arm.com>
19211 PR rtl-optimization/84068
19212 PR rtl-optimization/83459
19213 * haifa-sched.c (rank_for_schedule): Fix SCHED_PRESSURE_MODEL sorting.
19215 2018-02-08 Aldy Hernandez <aldyh@redhat.com>
19217 PR tree-optimization/84224
19218 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Remove assert.
19219 * calls.c (gimple_alloca_call_p): Only return TRUE when we have
19220 non-zero arguments.
19222 2018-02-07 Iain Sandoe <iain@codesourcery.com>
19225 * config/rs6000/altivec.md (*restore_world): Remove LR use.
19226 * config/rs6000/predicates.md (restore_world_operation): Adjust op
19227 count, remove one USE.
19229 2018-02-07 Michael Meissner <meissner@linux.vnet.ibm.com>
19231 * doc/install.texi (Configuration): Document the
19232 --with-long-double-format={ibm,ieee} PowerPC configuration
19236 * config/rs6000/rs6000.md (fix_trunc<SFDF:mode><QHI:mode>2):
19237 Convert from define_expand to be define_insn_and_split. Rework
19238 float/double/_Float128 conversions to QI/HI/SImode to work with
19239 both ISA 2.07 (power8) or ISA 3.0 (power9). Fix regression where
19240 conversions to QI/HImode types did a store and then a load to
19241 truncate the value. For conversions to VSX registers, don't split
19242 the insn, instead emit the code directly. Use the code iterator
19243 any_fix to combine signed and unsigned conversions.
19244 (fix<uns>_trunc<SFDF:mode>si2_p8): Likewise.
19245 (fixuns_trunc<SFDF:mode><QHI:mode>2): Likewise.
19246 (fix_trunc<IEEE128:mode><QHI:mode>2): Likewise.
19247 (fix<uns>_trunc<SFDF:mode><QHI:mode>2): Likewise.
19248 (fix_<mode>di2_hw): Likewise.
19249 (fixuns_<mode>di2_hw): Likewise.
19250 (fix_<mode>si2_hw): Likewise.
19251 (fixuns_<mode>si2_hw): Likewise.
19252 (fix<uns>_<IEEE128:mode><SDI:mode>2_hw): Likewise.
19253 (fix<uns>_trunc<IEEE128:mode><QHI:mode>2): Likewise.
19254 (fctiw<u>z_<mode>_smallint): Rename fctiw<u>z_<mode>_smallint to
19255 fix<uns>_trunc<SFDF:mode>si2_p8.
19256 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Delete, no longer used.
19257 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
19258 (fix<uns>_<mode>_mem): Likewise.
19259 (fctiw<u>z_<mode>_mem): Likewise.
19260 (fix<uns>_<mode>_mem): Likewise.
19261 (fix<uns>_trunc<SFDF:mode><QHSI:mode>2_mem): On ISA 3.0, prevent
19262 the register allocator from doing a direct move to the GPRs to do
19263 a store, and instead use the ISA 3.0 store byte/half-word from
19264 vector register instruction. For IEEE 128-bit floating point,
19265 also optimize stores of 32-bit ints.
19266 (fix<uns>_trunc<IEEE128:mode><QHSI:mode>2_mem): Likewise.
19268 2018-02-07 Alan Hayward <alan.hayward@arm.com>
19270 * genextract.c (push_pathstr_operand): New function to support
19272 (walk_rtx): Call push_pathstr_operand.
19273 (print_path): Support [a-zA-Z].
19275 2018-02-07 Richard Biener <rguenther@suse.de>
19277 PR tree-optimization/84037
19278 * tree-vectorizer.h (struct _loop_vec_info): Add ivexpr_map member.
19279 (cse_and_gimplify_to_preheader): Declare.
19280 (vect_get_place_in_interleaving_chain): Likewise.
19281 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
19283 (_loop_vec_info::~_loop_vec_info): Delete it.
19284 (cse_and_gimplify_to_preheader): New function.
19285 * tree-vect-slp.c (vect_get_place_in_interleaving_chain): Export.
19286 * tree-vect-stmts.c (vectorizable_store): CSE base and steps.
19287 (vectorizable_load): Likewise. For grouped stores always base
19288 the IV on the first element.
19289 * tree-vect-loop-manip.c (vect_loop_versioning): Unshare versioning
19290 condition before gimplifying.
19292 2018-02-07 Jakub Jelinek <jakub@redhat.com>
19294 * tree-eh.c (operation_could_trap_helper_p): Ignore honor_trapv for
19295 *DIV_EXPR and *MOD_EXPR.
19297 2018-02-07 H.J. Lu <hongjiu.lu@intel.com>
19300 * config/i386/i386.c (ix86_option_override_internal): Mask out
19301 the CF_SET bit when checking -fcf-protection.
19303 2018-02-07 Tom de Vries <tom@codesourcery.com>
19306 * omp-expand.c (expand_oacc_collapse_init): Ensure diff_type is large
19309 2018-02-07 Richard Biener <rguenther@suse.de>
19311 PR tree-optimization/84204
19312 * tree-chrec.c (chrec_fold_plus_1): Remove size limiting in
19315 PR tree-optimization/84205
19316 * graphite-isl-ast-to-gimple.c (binary_op_to_tree): Also
19317 special-case isl_ast_op_zdiv_r.
19319 PR tree-optimization/84223
19320 * graphite-scop-detection.c (gather_bbs::before_dom_children):
19321 Only add conditions from within the region.
19322 (gather_bbs::after_dom_children): Adjust.
19324 2018-02-07 Georg-Johann Lay <avr@gjlay.de>
19327 * config/avr/avr.h (GENERAL_REGNO_P, GENERAL_REG_P): New macros.
19328 * config/avr/avr.md: Only post-reload split REG-REG moves if
19329 either register is GENERAL_REG_P.
19331 2018-02-07 Jakub Jelinek <jakub@redhat.com>
19333 PR tree-optimization/84235
19334 * tree-ssa-scopedtables.c
19335 (avail_exprs_stack::simplify_binary_operation): Fir MINUS_EXPR, punt
19336 if the subtraction is performed in floating point type where NaNs are
19337 honored. For *DIV_EXPR, punt for ALL_FRACT_MODE_Ps where we can't
19338 build 1. Formatting fix.
19340 2018-02-06 Jakub Jelinek <jakub@redhat.com>
19343 * config/i386/i386.c (rest_of_insert_endbranch): Only skip
19344 NOTE_INSN_CALL_ARG_LOCATION after a call, not anything else,
19345 and skip it regardless of bb boundaries. Use CALL_P macro,
19346 don't test INSN_P (insn) together with CALL_P or JUMP_P check
19347 unnecessarily, formatting fix.
19349 2018-02-06 Michael Collison <michael.collison@arm.com>
19351 * config/arm/thumb2.md:
19352 (*thumb2_mov_negscc): Split only if TARGET_THUMB2 && !arm_restrict_it.
19353 (*thumb_mov_notscc): Ditto.
19355 2018-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
19358 * config/rs6000/rs6000.md (su code attribute): Use "u" for
19359 unsigned_fix, not "s".
19361 2018-02-06 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
19363 * configure.ac (gcc_fn_eh_frame_ro): New function.
19364 (gcc_cv_as_cfi_directive): Check both 32 and 64-bit assembler for
19365 correct .eh_frame permissions.
19366 * configure: Regenerate.
19368 2018-02-06 Andrew Jenner <andrew@codeourcery.com>
19370 * doc/invoke.texi: Add section for the PowerPC SPE backend. Remove
19371 irrelevant options.
19373 2018-02-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19375 * config/rs6000/rs6000.c (rs6000_option_override_internal):
19376 Display warning message for -mno-speculate-indirect-jumps.
19378 2018-02-06 Andrew Jenner <andrew@codesourcery.com>
19380 * config/powerpcspe/powerpcspe.opt: (msimple-fpu, mfpu) Add
19382 * config/powerpcspe/sysv4.opt (mbit-align): Likewise.
19384 2018-02-06 Aldy Hernandez <aldyh@redhat.com>
19386 PR tree-optimization/84225
19387 * tree-eh.c (find_trapping_overflow): Only call
19388 operation_no_trapping_overflow when ANY_INTEGRAL_TYPE_P.
19390 2018-02-06 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
19393 * config/i386/i386.c: Reimplement the check of possible options
19394 -mibt/-mshstk conbination. Change error messages.
19395 * doc/invoke.texi: Fix a typo: remove extra '='.
19397 2018-02-06 Marek Polacek <polacek@redhat.com>
19399 PR tree-optimization/84228
19400 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Skip debug statements.
19402 2018-02-06 Tamar Christina <tamar.christina@arm.com>
19405 * config/arm/arm.c (arm_print_asm_arch_directives): Record already
19406 emitted arch directives.
19407 * config/arm/arm-c.c (arm_cpu_builtins): Undefine __ARM_ARCH and
19408 __ARM_FEATURE_COPROC before changing architectures.
19410 2018-02-06 Richard Biener <rguenther@suse.de>
19412 * config/i386/i386.c (print_reg): Fix typo.
19413 (ix86_loop_unroll_adjust): Do not unroll beyond the original nunroll.
19415 2018-02-06 Eric Botcazou <ebotcazou@adacore.com>
19417 * configure: Regenerate.
19419 2018-02-05 Martin Sebor <msebor@redhat.com>
19421 PR tree-optimization/83369
19422 * tree-ssa-ccp.c (pass_post_ipa_warn::execute): Use %G to print
19425 2018-02-05 Martin Liska <mliska@suse.cz>
19427 * doc/invoke.texi: Cherry-pick upstream r323995.
19429 2018-02-05 Richard Sandiford <richard.sandiford@linaro.org>
19431 * ira.c (ira_init_register_move_cost): Adjust comment.
19433 2018-02-05 Martin Liska <mliska@suse.cz>
19435 PR gcov-profile/84137
19436 * doc/gcov.texi: Fix typo in documentation.
19438 2018-02-05 Martin Liska <mliska@suse.cz>
19440 PR gcov-profile/83879
19441 * doc/gcov.texi: Document necessity of --dynamic-list-data when
19442 using dlopen functionality.
19444 2018-02-05 Olga Makhotina <olga.makhotina@intel.com>
19446 * config/i386/avx512dqintrin.h (_mm_mask_range_sd, _mm_maskz_range_sd,
19447 _mm_mask_range_round_sd, _mm_maskz_range_round_sd, _mm_mask_range_ss,
19448 _mm_maskz_range_ss, _mm_mask_range_round_ss,
19449 _mm_maskz_range_round_ss): New intrinsics.
19450 (__builtin_ia32_rangesd128_round)
19451 (__builtin_ia32_rangess128_round): Remove.
19452 (__builtin_ia32_rangesd128_mask_round,
19453 __builtin_ia32_rangess128_mask_round): New builtins.
19454 * config/i386/i386-builtin.def (__builtin_ia32_rangesd128_round,
19455 __builtin_ia32_rangess128_round): Remove.
19456 (__builtin_ia32_rangesd128_mask_round,
19457 __builtin_ia32_rangess128_mask_round): New builtins.
19458 * config/i386/sse.md (ranges<mode><round_saeonly_name>): Renamed to ...
19459 (ranges<mode><mask_scalar_name><round_saeonly_scalar_name>): ... this.
19460 ((match_operand:VF_128 2 "<round_saeonly_nimm_predicate>"
19461 "<round_saeonly_constraint>")): Changed to ...
19462 ((match_operand:VF_128 2 "<round_saeonly_scalar_nimm_predicate>"
19463 "<round_saeonly_scalar_constraint>")): ... this.
19464 ("vrange<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
19465 %0, %1, %2<round_saeonly_op4>, %3}"): Changed to ...
19466 ("vrange<ssescalarmodesuffix>\t{%3, <round_saeonly_scalar_mask_op4>%2,
19467 %1, %0<mask_scalar_operand4>|%0<mask_scalar_operand4>, %1,
19468 %2<round_saeonly_scalar_mask_op4>, %3}"): ... this.
19470 2018-02-02 Andrew Jenner <andrew@codesourcery.com>
19472 * config/powerpcspe/powerpcspe.opt: Add Undocumented to irrelevant
19474 * config/powerpcspe/powerpcspe-tables.opt (rs6000_cpu_opt_value):
19475 Remove all values except native, 8540 and 8548.
19477 2018-02-02 H.J. Lu <hongjiu.lu@intel.com>
19479 * config/i386/i386.c (ix86_output_function_return): Pass
19480 INVALID_REGNUM, instead of -1, as invalid register number to
19481 indirect_thunk_name and output_indirect_thunk.
19483 2018-02-02 Julia Koval <julia.koval@intel.com>
19485 * config.gcc: Add -march=icelake.
19486 * config/i386/driver-i386.c (host_detect_local_cpu): Detect icelake.
19487 * config/i386/i386-c.c (ix86_target_macros_internal): Handle icelake.
19488 * config/i386/i386.c (processor_costs): Add m_ICELAKE.
19489 (PTA_ICELAKE, PTA_AVX512VNNI, PTA_GFNI, PTA_VAES, PTA_AVX512VBMI2,
19490 PTA_VPCLMULQDQ, PTA_RDPID, PTA_AVX512BITALG): New.
19491 (processor_target_table): Add icelake.
19492 (ix86_option_override_internal): Handle new PTAs.
19493 (get_builtin_code_for_version): Handle icelake.
19494 (M_INTEL_COREI7_ICELAKE): New.
19495 (fold_builtin_cpu): Handle icelake.
19496 * config/i386/i386.h (TARGET_ICELAKE, PROCESSOR_ICELAKE): New.
19497 * doc/invoke.texi: Add -march=icelake.
19499 2018-02-02 Julia Koval <julia.koval@intel.com>
19501 * config/i386/i386.c (ix86_option_override_internal): Change flags type
19502 to wide_int_bitmask.
19503 * wide-int-bitmask.h: New.
19505 2018-02-02 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
19508 * config/i386/i386.md: Replace Pmode with word_mode in
19509 builtin_setjmp_setup and builtin_longjmp to support x32.
19511 2018-02-01 Peter Bergner <bergner@vnet.ibm.com>
19515 * config/rs6000/driver-rs6000.c: #include "diagnostic.h".
19517 (rs6000_supported_cpu_names): New static variable.
19518 (linux_cpu_translation_table): Likewise.
19519 (elf_platform) <cpu>: Define new static variable and use it.
19520 Translate kernel AT_PLATFORM name to canonical name if needed.
19521 Error if platform name is unknown.
19523 2018-02-01 Aldy Hernandez <aldyh@redhat.com>
19526 * config/pa/predicates.md (base14_operand): Handle E_VOIDmode.
19528 2018-02-01 Jeff Law <law@redhat.com>
19531 * config/i386/i386.c (release_scratch_register_on_entry): Add new
19532 OFFSET and RELEASE_VIA_POP arguments. Use SP+OFFSET to restore
19533 the scratch if RELEASE_VIA_POP is false.
19534 (ix86_adjust_stack_and_probe_stack_clash): Un-constify SIZE.
19535 If we have to save a temporary register, decrement SIZE appropriately.
19536 Pass new arguments to release_scratch_register_on_entry.
19537 (ix86_adjust_stack_and_probe): Likewise.
19538 (ix86_emit_probe_stack_range): Pass new arguments to
19539 release_scratch_register_on_entry.
19541 2018-02-01 Uros Bizjak <ubizjak@gmail.com>
19543 PR rtl-optimization/84157
19544 * combine.c (change_zero_ext): Use REG_P predicate in
19545 front of HARD_REGISTER_P predicate.
19547 2018-02-01 Georg-Johann Lay <avr@gjlay.de>
19549 * config/avr/avr.c (avr_option_override): Move disabling of
19550 -fdelete-null-pointer-checks to...
19551 * common/config/avr/avr-common.c (avr_option_optimization_table):
19554 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
19556 PR tree-optimization/81635
19557 * tree-data-ref.c (split_constant_offset_1): For types that
19558 wrap on overflow, try to use range info to prove that wrapping
19561 2018-02-01 Renlin Li <renlin.li@arm.com>
19564 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
19565 TAILCALL_ADDR_REGS.
19566 (aarch64_register_move_cost): Likewise.
19567 * config/aarch64/aarch64.h (reg_class): Rename CALLER_SAVE_REGS to
19568 TAILCALL_ADDR_REGS.
19569 (REG_CLASS_NAMES): Likewise.
19570 (REG_CLASS_CONTENTS): Rename CALLER_SAVE_REGS to
19571 TAILCALL_ADDR_REGS. Remove IP registers.
19572 * config/aarch64/aarch64.md (Ucs): Update register constraint.
19574 2018-02-01 Richard Biener <rguenther@suse.de>
19576 * domwalk.h (dom_walker::dom_walker): Add additional constructor
19577 for specifying RPO order and allow NULL for that.
19578 * domwalk.c (dom_walker::dom_walker): Likewise.
19579 (dom_walker::walk): Handle NULL RPO order.
19580 * tree-into-ssa.c (rewrite_dom_walker): Do not walk dom children
19582 (rewrite_update_dom_walker): Likewise.
19583 (mark_def_dom_walker): Likewise.
19585 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
19587 * config/aarch64/aarch64-protos.h (aarch64_split_sve_subreg_move)
19588 (aarch64_maybe_expand_sve_subreg_move): Declare.
19589 * config/aarch64/aarch64.md (UNSPEC_REV_SUBREG): New unspec.
19590 * config/aarch64/predicates.md (aarch64_any_register_operand): New
19592 * config/aarch64/aarch64-sve.md (mov<mode>): Optimize subreg moves
19593 that are semantically a reverse operation.
19594 (*aarch64_sve_mov<mode>_subreg_be): New pattern.
19595 * config/aarch64/aarch64.c (aarch64_maybe_expand_sve_subreg_move):
19596 (aarch64_replace_reg_mode, aarch64_split_sve_subreg_move): New
19598 (aarch64_can_change_mode_class): For big-endian, forbid changes
19599 between two SVE modes if they have different element sizes.
19601 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
19603 * config/aarch64/aarch64.c (aarch64_expand_sve_const_vector): Prefer
19604 the TImode handling for big-endian targets.
19606 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
19608 * config/aarch64/aarch64-sve.md (sve_ld1rq): Replace with...
19609 (*sve_ld1rq<Vesize>): ... this new pattern. Handle all element sizes,
19611 * config/aarch64/aarch64.c (aarch64_expand_sve_widened_duplicate):
19612 Remove BSWAP handing for big-endian targets and use the form of
19613 LD1RQ appropariate for the mode.
19615 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
19617 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Handle
19618 all CONST_VECTOR_DUPLICATE_P vectors, not just those with a single
19619 duplicated element.
19621 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
19624 * config/aarch64/aarch64.c (aarch64_secondary_reload): Tighten
19625 check for operands that need to go through aarch64_sve_reload_be.
19627 2018-02-01 Jakub Jelinek <jakub@redhat.com>
19629 PR tree-optimization/81661
19630 PR tree-optimization/84117
19631 * tree-eh.h (rewrite_to_non_trapping_overflow): Declare.
19632 * tree-eh.c: Include gimplify.h.
19633 (find_trapping_overflow, replace_trapping_overflow,
19634 rewrite_to_non_trapping_overflow): New functions.
19635 * tree-vect-loop.c: Include tree-eh.h.
19636 (vect_get_loop_niters): Use rewrite_to_non_trapping_overflow.
19637 * tree-data-ref.c: Include tree-eh.h.
19638 (get_segment_min_max): Use rewrite_to_non_trapping_overflow.
19640 2018-01-31 Uros Bizjak <ubizjak@gmail.com>
19642 PR rtl-optimization/84123
19643 * combine.c (change_zero_ext): Check if hard register satisfies
19644 can_change_dest_mode before calling gen_lowpart_SUBREG.
19646 2018-01-31 Vladimir Makarov <vmakarov@redhat.com>
19649 * ira.c (ira_init_register_move_cost): Remove assert.
19651 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
19653 PR rtl-optimization/84071
19654 * doc/tm.texi.in (WORD_REGISTER_OPERATIONS): Add explicit case.
19655 * doc/tm.texi: Regenerate.
19657 2018-01-31 Richard Biener <rguenther@suse.de>
19659 PR tree-optimization/84132
19660 * tree-data-ref.c (analyze_miv_subscript): Properly
19661 check whether evolution_function_is_affine_multivariate_p
19662 before calling gcd_of_steps_may_divide_p.
19664 2018-01-31 Julia Koval <julia.koval@intel.com>
19667 * config/i386/i386.c (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
19668 * config/i386/i386.md (rdpid_rex64) New.
19669 (rdpid): Make 32bit only.
19671 2018-01-29 Aldy Hernandez <aldyh@redhat.com>
19674 * tree-pretty-print.c (dump_generic_node): Handle a TYPE_NAME with
19675 an IDENTIFIER_NODE for FUNCTION_TYPE's.
19677 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
19680 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
19682 * config/sparc/sparc.md (vxworks_load_got): Set the GOT register.
19684 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
19686 PR rtl-optimization/84071
19687 * combine.c (record_dead_and_set_regs_1): Record the source unmodified
19688 for a paradoxical SUBREG on a WORD_REGISTER_OPERATIONS target.
19690 2018-01-31 Claudiu Zissulescu <claziss@synopsys.com>
19692 * config/arc/arc.c (arc_handle_aux_attribute): New function.
19693 (arc_attribute_table): Add 'aux' attribute.
19694 (arc_in_small_data_p): Consider aux like variables.
19695 (arc_is_aux_reg_p): New function.
19696 (arc_asm_output_aligned_decl_local): Ignore 'aux' like variables.
19697 (arc_get_aux_arg): New function.
19698 (prepare_move_operands): Handle aux-register access.
19699 (arc_handle_aux_attribute): New function.
19700 * doc/extend.texi (ARC Variable attributes): Add subsection.
19702 2018-01-31 Claudiu Zissulescu <claziss@synopsys.com>
19704 * config/arc/arc-protos.h (arc_is_uncached_mem_p): Function proto.
19705 * config/arc/arc.c (arc_handle_uncached_attribute): New function.
19706 (arc_attribute_table): Add 'uncached' attribute.
19707 (arc_print_operand): Print '.di' flag for uncached memory
19709 (arc_in_small_data_p): Do not consider for small data the uncached
19711 (arc_is_uncached_mem_p): New function.
19712 * config/arc/predicates.md (compact_store_memory_operand): Check
19713 for uncached memory accesses.
19714 (nonvol_nonimm_operand): Likewise.
19715 * doc/extend.texi (ARC Type Attribute): New subsection.
19717 2018-01-31 Jakub Jelinek <jakub@redhat.com>
19720 * common.opt (falign-functions=, falign-jumps=, falign-labels=,
19721 falign-loops=): Add Optimization flag.
19723 2018-01-30 Jeff Law <law@redhat.com>
19726 * i386.c (ix86_adjust_stack_and_probe_stack_clash): New argument
19727 INT_REGISTERS_SAVED. Check it prior to calling
19728 get_scratch_register_on_entry.
19729 (ix86_adjust_stack_and_probe): Similarly.
19730 (ix86_emit_probe_stack_range): Similarly.
19731 (ix86_expand_prologue): Corresponding changes.
19733 2018-01-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
19736 * config/sol2.h (STARTFILE_ARCH_SPEC): Use -std=c*,
19737 -std=iso9899:199409 instead of -pedantic to select values-Xc.o.
19739 2018-01-30 Vladimir Makarov <vmakarov@redhat.com>
19742 * lra-constraints.c (curr_insn_transform): Process AND in the
19745 2018-01-30 Jakub Jelinek <jakub@redhat.com>
19747 PR rtl-optimization/83986
19748 * sched-deps.c (sched_analyze_insn): For frame related insns, add anti
19749 dependence against last_pending_memory_flush in addition to
19750 pending_jump_insns.
19752 2018-01-30 Alexandre Oliva <aoliva@redhat.com>
19754 PR tree-optimization/81611
19755 * tree-ssa-dom.c (simple_iv_increment_p): Skip intervening
19758 2018-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19761 * config/rs6000/rs6000.c (rs6000_internal_arg_pointer): Only return
19764 2018-01-30 Richard Biener <rguenther@suse.de>
19765 Jakub Jelinek <jakub@redhat.com>
19767 PR tree-optimization/84111
19768 * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1): Skip
19769 inner loops added during recursion, as they don't have up-to-date
19772 2018-01-30 Jan Hubicka <hubicka@ucw.cz>
19775 * ipa-inline.c (can_inline_edge_p): Break out late tests to...
19776 (can_inline_edge_by_limits_p): ... here.
19777 (can_early_inline_edge_p, check_callers,
19778 update_caller_keys, update_callee_keys, recursive_inlining,
19779 add_new_edges_to_heap, speculation_useful_p,
19780 inline_small_functions,
19781 inline_small_functions, flatten_function,
19782 inline_to_all_callers_1): Update.
19784 2018-01-30 Jan Hubicka <hubicka@ucw.cz>
19786 * profile-count.c (profile_count::combine_with_ipa_count): Handle
19789 2018-01-30 Richard Biener <rguenther@suse.de>
19791 PR tree-optimization/83008
19792 * tree-vect-slp.c (vect_analyze_slp_cost_1): Properly cost
19793 invariant and constant vector uses in stmts when they need
19794 more than one stmt.
19796 2018-01-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
19799 * configure.ac (gcc_cv_as_shf_merge): Disable on Solaris 10/x86.
19800 * configure: Regenerate.
19802 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
19804 * config/aarch64/aarch64-sve.md (*vec_extract<mode><Vel>_0): New
19806 (*vec_extract<mode><Vel>_v128): Require a nonzero lane number.
19807 Use gen_rtx_REG rather than gen_lowpart.
19809 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
19811 * lra-constraints.c (match_reload): Use subreg_lowpart_offset
19812 rather than 0 when creating partial subregs.
19814 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
19816 * vec-perm-indices.c (vec_perm_indices::series_p): Give examples
19819 2018-01-29 Michael Meissner <meissner@linux.vnet.ibm.com>
19822 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): If DFmode
19823 and SFmode can go in Altivec registers (-mcpu=power7 for DFmode,
19824 -mcpu=power8 for SFmode) don't set the PRE_INCDEC or PRE_MODIFY
19825 flags. This restores the settings used before the 2017-07-24.
19826 Turning off pre increment/decrement/modify allows IVOPTS to
19827 optimize DF/SF loops where the index is an int.
19829 2018-01-29 Richard Biener <rguenther@suse.de>
19830 Kelvin Nilsen <kelvin@gcc.gnu.org>
19833 * tree-vect-stmts.c (vectorizable_call): Don't call
19834 targetm.vectorize_builtin_md_vectorized_function if callee is
19837 2018-01-22 Carl Love <cel@us.ibm.com>
19839 * doc/extend.tex: Fix typo in second arg in
19840 __builtin_bcdadd_{lt|eq|gt|ov} and __builtin_bcdsub_{lt|eq|gt|ov}.
19842 2018-01-29 Richard Biener <rguenther@suse.de>
19844 PR tree-optimization/84086
19845 * tree-ssanames.c: Include cfgloop.h and tree-scalar-evolution.h.
19846 (flush_ssaname_freelist): When SSA names were released reset
19847 the SCEV hash table.
19849 2018-01-29 Richard Biener <rguenther@suse.de>
19851 PR tree-optimization/84057
19852 * tree-ssa-loop-ivcanon.c (unloop_loops): Deal with already
19853 removed paths when removing edges.
19855 2018-01-27 H.J. Lu <hongjiu.lu@intel.com>
19857 * doc/invoke.texi: Replace -mfunction-return==@var{choice} with
19858 -mfunction-return=@var{choice}.
19860 2018-01-27 Bernd Edlinger <bernd.edlinger@hotmail.de>
19862 PR diagnostic/84034
19863 * diagnostic-show-locus.c (get_line_width_without_trailing_whitespace):
19864 Handle CR like TAB.
19865 (layout::print_source_line): Likewise.
19866 (test_get_line_width_without_trailing_whitespace): Add test cases.
19868 2018-01-27 Jakub Jelinek <jakub@redhat.com>
19870 PR middle-end/84040
19871 * sched-deps.c (sched_macro_fuse_insns): Return immediately for
19874 2018-01-26 Jim Wilson <jimw@sifive.com>
19876 * config/riscv/riscv.h (MAX_FIXED_MODE_SIZE): New.
19878 * config/riscv/elf.h (LIB_SPEC): Don't include -lgloss when nosys.specs
19881 2018-01-26 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19883 * config/aarch64/aarch64.md: Add peepholes for CMP + SUB -> SUBS
19884 and CMP + SUB-immediate -> SUBS.
19886 2018-01-26 Martin Sebor <msebor@redhat.com>
19888 PR tree-optimization/83896
19889 * tree-ssa-strlen.c (get_string_len): Rename...
19890 (get_string_cst_length): ...to this. Return HOST_WIDE_INT.
19891 Avoid assuming length is constant.
19892 (handle_char_store): Use HOST_WIDE_INT for string length.
19894 2018-01-26 Uros Bizjak <ubizjak@gmail.com>
19897 * config/i386/i386.md (*andndi3_doubleword): Add earlyclobber
19898 to (=&r,r,rm) alternative. Add (=r,0,rm) and (=r,r,0) alternatives.
19900 2018-01-26 Richard Biener <rguenther@suse.de>
19902 PR rtl-optimization/84003
19903 * dse.c (record_store): Only record redundant stores when
19904 the earlier store aliases at least all accesses the later one does.
19906 2018-01-26 Jakub Jelinek <jakub@redhat.com>
19908 PR rtl-optimization/83985
19909 * dce.c (deletable_insn_p): Return false for separate shrink wrapping
19910 REG_CFA_RESTORE insns.
19911 (delete_unmarked_insns): Don't ignore separate shrink wrapping
19912 REG_CFA_RESTORE insns here.
19915 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Don't
19916 use SSA_NAME_VAR as base for SSA_NAMEs with non-NULL SSA_NAME_VAR.
19918 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
19920 * config/arc/arc-arch.h (arc_tune_attr): Add ARC_TUNE_CORE_3.
19921 * config/arc/arc.c (arc_sched_issue_rate): Use ARC_TUNE_... .
19922 (arc_init): Likewise.
19923 (arc_override_options): Likewise.
19924 (arc_file_start): Choose Tag_ARC_CPU_variation based on arc_tune
19926 (hwloop_fail): Use TARGET_DBNZ when we want to check for dbnz insn
19928 * config/arc/arc.h (TARGET_DBNZ): Define.
19929 * config/arc/arc.md (attr tune): Add core_3, use ARC_TUNE_... to
19930 properly set the tune attribute.
19931 (dbnz): Use TARGET_DBNZ guard.
19932 * config/arc/arc.opt (mtune): Add core3 option.
19934 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
19936 * config/arc/arc.c (arc_delegitimize_address_0): Refactored to
19937 recognize new pic like addresses.
19938 (arc_delegitimize_address): Clean up.
19940 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
19942 * config/arc/arc-arches.def: Option mrf16 valid for all
19944 * config/arc/arc-c.def (__ARC_RF16__): New predefined macro.
19945 * config/arc/arc-cpus.def (em_mini): New cpu with rf16 on.
19946 * config/arc/arc-options.def (FL_RF16): Add mrf16 option.
19947 * config/arc/arc-tables.opt: Regenerate.
19948 * config/arc/arc.c (arc_conditional_register_usage): Handle
19949 reduced register file case.
19950 (arc_file_start): Set must have build attributes.
19951 * config/arc/arc.h (MAX_ARC_PARM_REGS): Conditional define using
19952 mrf16 option value.
19953 * config/arc/arc.opt (mrf16): Add new option.
19954 * config/arc/elf.h (ATTRIBUTE_PCS): Define.
19955 * config/arc/genmultilib.awk: Handle new mrf16 option.
19956 * config/arc/linux.h (ATTRIBUTE_PCS): Define.
19957 * config/arc/t-multilib: Regenerate.
19958 * doc/invoke.texi (ARC Options): Document mrf16 option.
19960 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
19962 * config/arc/arc-protos.h: Add arc_is_secure_call_p proto.
19963 * config/arc/arc.c (arc_handle_secure_attribute): New function.
19964 (arc_attribute_table): Add 'secure_call' attribute.
19965 (arc_print_operand): Print secure call operand.
19966 (arc_function_ok_for_sibcall): Don't optimize tail calls when
19968 (arc_is_secure_call_p): New function. * config/arc/arc.md
19969 (call_i): Add support for sjli instruction.
19970 (call_value_i): Likewise.
19971 * config/arc/constraints.md (Csc): New constraint.
19973 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
19974 John Eric Martin <John.Martin@emmicro-us.com>
19976 * config/arc/arc-protos.h: Add arc_is_jli_call_p proto.
19977 * config/arc/arc.c (_arc_jli_section): New struct.
19978 (arc_jli_section): New type.
19979 (rc_jli_sections): New static variable.
19980 (arc_handle_jli_attribute): New function.
19981 (arc_attribute_table): Add jli_always and jli_fixed attribute.
19982 (arc_file_end): New function.
19983 (TARGET_ASM_FILE_END): Define.
19984 (arc_print_operand): Reuse 'S' letter for JLI output instruction.
19985 (arc_add_jli_section): New function.
19986 (jli_call_scan): Likewise.
19987 (arc_reorg): Call jli_call_scan.
19988 (arc_output_addsi): Remove 'S' from printing asm operand.
19989 (arc_is_jli_call_p): New function.
19990 * config/arc/arc.md (movqi_insn): Remove 'S' from printing asm
19992 (movhi_insn): Likewise.
19993 (movsi_insn): Likewise.
19994 (movsi_set_cc_insn): Likewise.
19995 (loadqi_update): Likewise.
19996 (load_zeroextendqisi_update): Likewise.
19997 (load_signextendqisi_update): Likewise.
19998 (loadhi_update): Likewise.
19999 (load_zeroextendhisi_update): Likewise.
20000 (load_signextendhisi_update): Likewise.
20001 (loadsi_update): Likewise.
20002 (loadsf_update): Likewise.
20003 (movsicc_insn): Likewise.
20004 (bset_insn): Likewise.
20005 (bxor_insn): Likewise.
20006 (bclr_insn): Likewise.
20007 (bmsk_insn): Likewise.
20008 (bicsi3_insn): Likewise.
20009 (cmpsi_cc_c_insn): Likewise.
20010 (movsi_ne): Likewise.
20011 (movsi_cond_exec): Likewise.
20012 (clrsbsi2): Likewise.
20013 (norm_f): Likewise.
20023 (call_i): Remove 'S' asm letter, add jli instruction.
20024 (call_value_i): Likewise.
20025 * config/arc/arc.op (mjli-always): New option.
20026 * config/arc/constraints.md (Cji): New constraint.
20027 * config/arc/fpx.md (addsf3_fpx): Remove 'S' from printing asm
20029 (subsf3_fpx): Likewise.
20030 (mulsf3_fpx): Likewise.
20031 * config/arc/simdext.md (vendrec_insn): Remove 'S' from printing
20033 * doc/extend.texi (ARC): Document 'jli-always' and 'jli-fixed'
20034 function attrbutes.
20035 * doc/invoke.texi (ARC): Document mjli-always option.
20037 2018-01-26 Sebastian Perta <sebastian.perta@renesas.com>
20039 * config/rl78/rl78.c (rl78_addsi3_internal): If operand 2 is const
20040 avoid addition with 0 and use incw and decw where possible.
20042 2018-01-26 Richard Biener <rguenther@suse.de>
20044 PR tree-optimization/81082
20045 * fold-const.c (fold_plusminus_mult_expr): Do not perform the
20046 association if it requires casting to unsigned.
20047 * match.pd ((A * C) +- (B * C) -> (A+-B)): New patterns derived
20048 from fold_plusminus_mult_expr to catch important cases late when
20049 range info is available.
20051 2018-01-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
20053 * config/i386/sol2.h (USE_HIDDEN_LINKONCE): Remove.
20054 * configure.ac (hidden_linkonce): New test.
20055 * configure: Regenerate.
20056 * config.in: Regenerate.
20058 2018-01-26 Julia Koval <julia.koval@intel.com>
20060 * config/i386/avx512bitalgintrin.h (_mm512_bitshuffle_epi64_mask,
20061 _mm512_mask_bitshuffle_epi64_mask, _mm256_bitshuffle_epi64_mask,
20062 _mm256_mask_bitshuffle_epi64_mask, _mm_bitshuffle_epi64_mask,
20063 _mm_mask_bitshuffle_epi64_mask): Fix type.
20064 * config/i386/i386-builtin-types.def (UHI_FTYPE_V2DI_V2DI_UHI,
20065 USI_FTYPE_V4DI_V4DI_USI): Remove.
20066 * config/i386/i386-builtin.def (__builtin_ia32_vpshufbitqmb512_mask,
20067 __builtin_ia32_vpshufbitqmb256_mask,
20068 __builtin_ia32_vpshufbitqmb128_mask): Fix types.
20069 * config/i386/i386.c (ix86_expand_args_builtin): Remove old types.
20070 * config/i386/sse.md (VI1_AVX512VLBW): Change types.
20072 2018-01-26 Alan Modra <amodra@gmail.com>
20075 * config/rs6000/rs6000-p8swap.c (rtx_is_swappable_p): Exclude
20076 UNSPEC_VBPERMQ. Sort other unspecs.
20078 2018-01-25 David Edelsohn <dje.gcc@gmail.com>
20080 * doc/invoke.texi (PowerPC Options): Document 'native' cpu type.
20082 2018-01-25 Jan Hubicka <hubicka@ucw.cz>
20084 PR middle-end/83055
20085 * predict.c (drop_profile): Do not push/pop cfun; update also
20087 (handle_missing_profiles): Fix logic looking for zero profiles.
20089 2018-01-25 Jakub Jelinek <jakub@redhat.com>
20091 PR middle-end/83977
20092 * ipa-fnsummary.c (compute_fn_summary): Clear can_change_signature
20093 on functions with #pragma omp declare simd or functions with simd
20095 * omp-simd-clone.c (expand_simd_clones): Revert 2018-01-24 change.
20096 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
20097 Remove trailing \n from warning_at calls.
20099 2018-01-25 Tom de Vries <tom@codesourcery.com>
20102 * config/nvptx/nvptx.c (nvptx_single): Add exit insn after noreturn call
20103 for neutered workers.
20105 2018-01-24 Joseph Myers <joseph@codesourcery.com>
20108 * config/m68k/m68k.c (m68k_promote_function_mode): New function.
20109 (TARGET_PROMOTE_FUNCTION_MODE): New macro.
20111 2018-01-24 Jeff Law <law@redhat.com>
20114 * i386.c (get_probe_interval): Move to earlier point.
20115 (ix86_compute_frame_layout): If -fstack-clash-protection and
20116 the frame is larger than the probe interval, then use pushes
20117 to save registers rather than reg->mem moves.
20118 (ix86_expand_prologue): Remove conditional for int_registers_saved
20121 2018-01-24 Vladimir Makarov <vmakarov@redhat.com>
20124 * ira-build.c (setup_min_max_allocno_live_range_point): Set up
20125 min/max for never referenced object.
20127 2018-01-24 Jakub Jelinek <jakub@redhat.com>
20129 PR middle-end/83977
20130 * tree.c (free_lang_data_in_decl): Don't clear DECL_ABSTRACT_ORIGIN
20132 * omp-low.c (create_omp_child_function): Remove "omp declare simd"
20133 attributes from DECL_ATTRIBUTES (decl) without affecting
20134 DECL_ATTRIBUTES (current_function_decl).
20135 * omp-simd-clone.c (expand_simd_clones): Ignore DECL_ARTIFICIAL
20136 functions with non-NULL DECL_ABSTRACT_ORIGIN.
20138 2018-01-24 Richard Sandiford <richard.sandiford@linaro.org>
20140 PR tree-optimization/83979
20141 * fold-const.c (fold_comparison): Use constant_boolean_node
20142 instead of boolean_{true,false}_node.
20144 2018-01-24 Jan Hubicka <hubicka@ucw.cz>
20146 * ipa-profile.c (ipa_propagate_frequency_1): Fix logic skipping calls
20149 2018-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20151 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
20152 Simplify the clause that sets the length attribute.
20153 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
20154 (*sibcall_nonlocal_sysv<mode>): Clean up code block; simplify the
20155 clause that sets the length attribute.
20156 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
20158 2018-01-24 Tom de Vries <tom@codesourcery.com>
20161 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG_2): Define to 1.
20162 (nvptx_pc_set, nvptx_condjump_label): New function. Copy from jump.c.
20163 Add strict parameter.
20164 (prevent_branch_around_nothing): Insert dummy insn between branch to
20165 label and label with no ptx insn inbetween.
20166 * config/nvptx/nvptx.md (define_insn "fake_nop"): New insn.
20168 2018-01-24 Tom de Vries <tom@codesourcery.com>
20171 * config/nvptx/nvptx.c (nvptx_single): Add exit insn after noreturn call
20172 for neutered threads in warp.
20173 * config/nvptx/nvptx.md (define_insn "exit"): New insn.
20175 2018-01-24 Richard Biener <rguenther@suse.de>
20177 PR tree-optimization/83176
20178 * tree-chrec.c (chrec_fold_plus_1): Handle (signed T){(T) .. }
20181 2018-01-24 Richard Biener <rguenther@suse.de>
20183 PR tree-optimization/82819
20184 * graphite-isl-ast-to-gimple.c (binary_op_to_tree): Avoid
20185 code generating pluses that are no-ops in the target precision.
20187 2018-01-24 Richard Biener <rguenther@suse.de>
20189 PR middle-end/84000
20190 * tree-cfg.c (replace_loop_annotate): Handle annot_expr_parallel_kind.
20192 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
20194 * cfgcleanup.c (try_crossjump_to_edge): Use combine_with_count
20195 to merge probabilities.
20196 * predict.c (probably_never_executed): Also mark as cold functions
20197 with global 0 profile and guessed local profile.
20198 * profile-count.c (profile_probability::combine_with_count): New
20200 * profile-count.h (profile_probability::operator*,
20201 profile_probability::operator*=, profile_probability::operator/,
20202 profile_probability::operator/=): Reduce precision to adjusted
20203 and set value to guessed on contradictory divisions.
20204 (profile_probability::combine_with_freq): Remove.
20205 (profile_probability::combine_wiht_count): Declare.
20206 (profile_count::force_nonzero):: Set to adjusted.
20207 (profile_count::probability_in):: Set quality to adjusted.
20208 * tree-ssa-tail-merge.c (replace_block_by): Use
20209 combine_with_count.
20211 2018-01-23 Andrew Waterman <andrew@sifive.com>
20212 Jim Wilson <jimw@sifive.com>
20214 * config/riscv/riscv.c (riscv_stack_boundary): New.
20215 (riscv_option_override): Set riscv_stack_boundary. Handle
20216 riscv_preferred_stack_boundary_arg.
20217 * config/riscv/riscv.h (MIN_STACK_BOUNDARY, ABI_STACK_BOUNDARY): New.
20218 (BIGGEST_ALIGNMENT): Set to STACK_BOUNDARY.
20219 (STACK_BOUNDARY): Set to riscv_stack_boundary.
20220 (RISCV_STACK_ALIGN): Use STACK_BOUNDARY.
20221 * config/riscv/riscv.opt (mpreferred-stack-boundary): New.
20222 * doc/invoke.tex (RISC-V Options): Add -mpreferred-stack-boundary.
20224 2018-01-23 H.J. Lu <hongjiu.lu@intel.com>
20227 * config/i386/i386.c (ix86_expand_prologue): Use cost reference
20228 of struct ix86_frame.
20229 (ix86_expand_epilogue): Likewise. Add a local variable for
20230 the reg_save_offset field in struct ix86_frame.
20232 2018-01-23 Bin Cheng <bin.cheng@arm.com>
20234 PR tree-optimization/82604
20235 * tree-loop-distribution.c (enum partition_kind): New enum item
20236 PKIND_PARTIAL_MEMSET.
20237 (partition_builtin_p): Support above new enum item.
20238 (generate_code_for_partition): Ditto.
20239 (compute_access_range): Differentiate cases that equality can be
20240 proven at all loops, the innermost loops or no loops.
20241 (classify_builtin_st, classify_builtin_ldst): Adjust call to above
20242 function. Set PKIND_PARTIAL_MEMSET for partition appropriately.
20243 (finalize_partitions, distribute_loop): Don't fuse partition of
20244 PKIND_PARTIAL_MEMSET kind when distributing 3-level loop nest.
20245 (prepare_perfect_loop_nest): Distribute 3-level loop nest only if
20246 parloop is enabled.
20248 2018-01-23 Martin Liska <mliska@suse.cz>
20250 * predict.def (PRED_INDIR_CALL): Set probability to PROB_EVEN in
20251 order to ignore the predictor.
20252 (PRED_POLYMORPHIC_CALL): Likewise.
20253 (PRED_RECURSIVE_CALL): Likewise.
20255 2018-01-23 Martin Liska <mliska@suse.cz>
20257 * tree-profile.c (tree_profiling): Print function header to
20258 aware reader which function we are working on.
20259 * value-prof.c (gimple_find_values_to_profile): Do not print
20260 not interesting value histograms.
20262 2018-01-23 Martin Liska <mliska@suse.cz>
20264 * profile-count.h (enum profile_quality): Add
20265 profile_uninitialized as the first value. Do not number values
20266 as they are zero based.
20267 (profile_count::verify): Update sanity check.
20268 (profile_probability::verify): Likewise.
20270 2018-01-23 Nathan Sidwell <nathan@acm.org>
20272 * doc/invoke.texi (ffor-scope): Deprecate.
20274 2018-01-23 David Malcolm <dmalcolm@redhat.com>
20276 PR tree-optimization/83510
20277 * domwalk.c (set_all_edges_as_executable): New function.
20278 (dom_walker::dom_walker): Convert bool param
20279 "skip_unreachable_blocks" to enum reachability. Move setup of
20280 edge flags to set_all_edges_as_executable and only do it when
20281 reachability is REACHABLE_BLOCKS.
20282 * domwalk.h (enum dom_walker::reachability): New enum.
20283 (dom_walker::dom_walker): Convert bool param
20284 "skip_unreachable_blocks" to enum reachability.
20285 (set_all_edges_as_executable): New decl.
20286 * graphite-scop-detection.c (gather_bbs::gather_bbs): Convert
20287 from false for "skip_unreachable_blocks" to ALL_BLOCKS for
20289 * tree-ssa-dom.c (dom_opt_dom_walker::dom_opt_dom_walker): Likewise,
20290 but converting true to REACHABLE_BLOCKS.
20291 * tree-ssa-sccvn.c (sccvn_dom_walker::sccvn_dom_walker): Likewise.
20293 (check_array_bounds_dom_walker::check_array_bounds_dom_walker):
20294 Likewise, but converting it to REACHABLE_BLOCKS_PRESERVING_FLAGS.
20295 (vrp_dom_walker::vrp_dom_walker): Likewise, but converting it to
20297 (vrp_prop::vrp_finalize): Call set_all_edges_as_executable
20298 if check_all_array_refs will be called.
20300 2018-01-23 David Malcolm <dmalcolm@redhat.com>
20302 * tree.c (selftest::test_location_wrappers): Add more test
20305 2018-01-23 David Malcolm <dmalcolm@redhat.com>
20307 * sbitmap.c (selftest::test_set_range): Fix memory leaks.
20308 (selftest::test_bit_in_range): Likewise.
20310 2018-01-23 Richard Sandiford <richard.sandiford@linaro.org>
20313 * doc/sourcebuild.texi (vect_float): Say that the selector
20314 only describes the situation when -funsafe-math-optimizations is on.
20315 (vect_float_strict): Document.
20317 2018-01-23 Richard Sandiford <richard.sandiford@linaro.org>
20319 PR tree-optimization/83965
20320 * tree-vect-patterns.c (vect_reassociating_reduction_p): New function.
20321 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern): Use it
20322 instead of checking only for a reduction.
20323 (vect_recog_widen_sum_pattern): Likewise.
20325 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
20327 * predict.c (probably_never_executed): Only use precise profile info.
20328 (compute_function_frequency): Skip after inlining hack since we now
20329 have quality checking.
20331 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
20333 * profile-count.h (profile_probability::very_unlikely,
20334 profile_probability::unlikely, profile_probability::even): Set
20335 precision to guessed.
20337 2018-01-23 Richard Biener <rguenther@suse.de>
20339 PR tree-optimization/83963
20340 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
20341 Properly terminate dominator walk when crossing the exit edge not
20342 when visiting its source block.
20344 2018-01-23 Jakub Jelinek <jakub@redhat.com>
20347 * tree.c (maybe_wrap_with_location): Use NON_LVALUE_EXPR rather than
20348 VIEW_CONVERT_EXPR to wrap CONST_DECLs.
20350 2018-01-22 Jakub Jelinek <jakub@redhat.com>
20352 PR tree-optimization/83957
20353 * omp-expand.c (expand_omp_for_generic): Ignore virtual PHIs. Remove
20354 semicolon after for body surrounded by braces.
20356 PR tree-optimization/83081
20357 * profile-count.h (profile_probability::split): New method.
20358 * dojump.c (do_jump_1) <case TRUTH_ANDIF_EXPR, case TRUTH_ORIF_EXPR>:
20359 Use profile_probability::split.
20360 (do_compare_rtx_and_jump): Fix adjustment of probabilities
20361 when splitting a single conditional jump into 2.
20363 2018-01-22 David Malcolm <dmalcolm@redhat.com>
20365 PR tree-optimization/69452
20366 * tree-ssa-loop-im.c (class move_computations_dom_walker): Remove
20369 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
20371 * config/rl78/rl78-expand.md (bswaphi2): New define_expand.
20372 * config/rl78/rl78-virt.md (*bswaphi2_virt): New define_insn.
20373 * config/rl78/rl78-real.md (*bswaphi2_real): New define_insn.
20375 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
20377 * config/rl78/rl78-protos.h (rl78_split_movdi): New function
20379 * config/rl78/rl78.md (movdi): New define_expand.
20380 * config/rl78/rl78.c (rl78_split_movdi): New function.
20382 2018-01-22 Michael Meissner <meissner@linux.vnet.ibm.com>
20385 * config/rs6000/rs6000-protos.h (rs6000_split_signbit): Delete,
20387 * config/rs6000/rs6000.c (rs6000_split_signbit): Likewise.
20388 * config/rs6000/rs6000.md (signbit<mode>2): Change code for IEEE
20389 128-bit to produce an UNSPEC move to get the double word with the
20390 signbit and then a shift directly to do signbit.
20391 (signbit<mode>2_dm): Replace old IEEE 128-bit signbit
20392 implementation with a new version that just does either a direct
20393 move or a regular move. Move memory interface to separate insns.
20394 Move insns so they are next to the expander.
20395 (signbit<mode>2_dm_mem_be): New combiner insns to combine load
20396 with signbit move. Split big and little endian case.
20397 (signbit<mode>2_dm_mem_le): Likewise.
20398 (signbit<mode>2_dm_<su>ext): Delete, no longer used.
20399 (signbit<mode>2_dm2): Likewise.
20401 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
20403 * config/rl78/rl78.md (anddi3): New define_expand.
20405 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
20407 * config/rl78/rl78.md (umindi3): New define_expand.
20409 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
20411 * config/rl78/rl78.md (smindi3): New define_expand.
20413 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
20415 * config/rl78/rl78.md (smaxdi3): New define_expand.
20417 2018-01-22 Carl Love <cel@us.ibm.com>
20419 * config/rs6000/rs6000-builtin.def (ST_ELEMREV_V1TI, LD_ELEMREV_V1TI,
20420 LVX_V1TI): Add macro expansion.
20421 * config/rs6000/rs6000-c.c (altivec_builtin_types): Add argument
20422 definitions for VSX_BUILTIN_VEC_XST_BE, VSX_BUILTIN_VEC_ST,
20423 VSX_BUILTIN_VEC_XL, LD_ELEMREV_V1TI builtins.
20424 * config/rs6000/rs6000-p8swap.c (insn_is_swappable_p);
20425 Change check to determine if the instruction is a byte reversing
20426 entry. Fix typo in comment.
20427 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case entry
20428 for VSX_BUILTIN_ST_ELEMREV_V1TI and VSX_BUILTIN_LD_ELEMREV_V1TI.
20429 Add def_builtin calls for new builtins.
20430 * config/rs6000/vsx.md (vsx_st_elemrev_v1ti, vsx_ld_elemrev_v1ti):
20431 Add define_insn expansion.
20433 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
20435 * config/rl78/rl78.md (umaxdi3): New define_expand.
20437 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
20439 * config/rl78/rl78.c (rl78_note_reg_set): Fixed dead reg check
20440 for non-QImode registers.
20442 2018-01-22 Richard Biener <rguenther@suse.de>
20444 PR tree-optimization/83963
20445 * graphite-scop-detection.c (scop_detection::get_sese): Delay
20446 including the loop exit block.
20447 (scop_detection::merge_sese): Likewise.
20448 (scop_detection::add_scop): Do it here instead.
20450 2018-01-22 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20452 * doc/sourcebuild.texi (arm_softfloat): Document.
20454 2018-01-21 John David Anglin <danglin@gcc.gnu.org>
20457 * config/pa/pa.c (pa_function_ok_for_sibcall): Use
20458 targetm.binds_local_p instead of TREE_PUBLIC to check local binding.
20459 Move TARGET_PORTABLE_RUNTIME check after TARGET_64BIT check.
20461 2018-01-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20462 David Edelsohn <dje.gcc@gmail.com>
20465 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
20466 Change "crset eq" to "crset 2".
20467 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
20468 (*call_indirect_aix<mode>_nospec): Likewise.
20469 (*call_value_indirect_aix<mode>_nospec): Likewise.
20470 (*call_indirect_elfv2<mode>_nospec): Likewise.
20471 (*call_value_indirect_elfv2<mode>_nospec): Likewise.
20472 (*sibcall_nonlocal_sysv<mode>): Change "crset eq" to "crset 2";
20473 change assembly output from . to $.
20474 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
20475 (indirect_jump<mode>_nospec): Change assembly output from . to $.
20476 (*tablejump<mode>_internal1_nospec): Likewise.
20478 2018-01-21 Oleg Endo <olegendo@gcc.gnu.org>
20481 * config/sh/sh_optimize_sett_clrt.cc:
20482 Use INCLUDE_ALGORITHM and INCLUDE_VECTOR instead of direct includes.
20484 2018-01-20 Richard Sandiford <richard.sandiford@linaro.org>
20486 PR tree-optimization/83940
20487 * tree-vect-stmts.c (vect_truncate_gather_scatter_offset): Set
20488 offset_dt to vect_constant_def rather than vect_unknown_def_type.
20489 (vect_check_load_store_mask): Add a mask_dt_out parameter and
20490 use it to pass back the definition type.
20491 (vect_check_store_rhs): Likewise rhs_dt_out.
20492 (vect_build_gather_load_calls): Add a mask_dt argument and use
20493 it instead of a call to vect_is_simple_use.
20494 (vectorizable_store): Update calls to vect_check_load_store_mask
20495 and vect_check_store_rhs. Use the dt returned by the latter instead
20496 of scatter_src_dt. Use the cached mask_dt and gs_info.offset_dt
20497 instead of calls to vect_is_simple_use. Pass the scalar rather
20498 than the vector operand to vect_is_simple_use when handling
20499 second and subsequent copies of an rhs value.
20500 (vectorizable_load): Update calls to vect_check_load_store_mask
20501 and vect_build_gather_load_calls. Use the cached mask_dt and
20502 gs_info.offset_dt instead of calls to vect_is_simple_use.
20504 2018-01-20 Jakub Jelinek <jakub@redhat.com>
20506 PR middle-end/83945
20507 * tree-emutls.c: Include gimplify.h.
20508 (lower_emutls_2): New function.
20509 (lower_emutls_1): If ADDR_EXPR is a gimple invariant and walk_tree
20510 with lower_emutls_2 callback finds some TLS decl in it, unshare_expr
20511 it before further processing.
20514 * simplify-rtx.c (simplify_binary_operation_1) <case UMOD>: Use
20515 UINTVAL (trueop1) instead of INTVAL (op1).
20517 2018-01-19 Jakub Jelinek <jakub@redhat.com>
20521 * dwarf2cfi.c (DEFAULT_INCOMING_FRAME_SP_OFFSET): Define to
20522 INCOMING_FRAME_SP_OFFSET if not defined.
20523 (scan_trace): Add ENTRY argument. If true and
20524 DEFAULT_INCOMING_FRAME_SP_OFFSET != INCOMING_FRAME_SP_OFFSET,
20525 emit a note to adjust the CFA offset.
20526 (create_cfi_notes): Adjust scan_trace callers.
20527 (create_cie_data): Use DEFAULT_INCOMING_FRAME_SP_OFFSET rather than
20528 INCOMING_FRAME_SP_OFFSET in the CIE.
20529 * config/i386/i386.h (DEFAULT_INCOMING_FRAME_SP_OFFSET): Define.
20530 * config/stormy16/stormy16.h (DEFAULT_INCOMING_FRAME_SP_OFFSET):
20532 * doc/tm.texi.in (DEFAULT_INCOMING_FRAME_SP_OFFSET): Document.
20533 * doc/tm.texi: Regenerated.
20535 2018-01-19 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20537 PR rtl-optimization/83147
20538 * lra-constraints.c (remove_inheritance_pseudos): Use
20539 lra_substitute_pseudo_within_insn.
20541 2018-01-19 Tom de Vries <tom@codesourcery.com>
20542 Cesar Philippidis <cesar@codesourcery.com>
20545 * config/nvptx/nvptx.c (nvptx_single): Fix jit workaround.
20547 2018-01-19 Cesar Philippidis <cesar@codesourcery.com>
20550 * config/nvptx/nvptx.c (output_init_frag): Don't use generic address
20551 spaces for function labels.
20553 2018-01-19 Martin Liska <mliska@suse.cz>
20555 * predict.def (PRED_LOOP_EXIT): Change from 85 to 89.
20556 (PRED_LOOP_EXIT_WITH_RECURSION): Change from 72 to 78.
20557 (PRED_LOOP_EXTRA_EXIT): Change from 83 to 67.
20558 (PRED_OPCODE_POSITIVE): Change from 64 to 59.
20559 (PRED_TREE_OPCODE_POSITIVE): Change from 64 to 59.
20560 (PRED_CONST_RETURN): Change from 69 to 65.
20561 (PRED_NULL_RETURN): Change from 91 to 71.
20562 (PRED_LOOP_IV_COMPARE_GUESS): Change from 98 to 64.
20563 (PRED_LOOP_GUARD): Change from 66 to 73.
20565 2018-01-19 Martin Liska <mliska@suse.cz>
20567 * predict.c (predict_insn_def): Add new assert.
20568 (struct branch_predictor): Change type to signed integer.
20569 (test_prediction_value_range): Amend test to cover
20570 PROB_UNINITIALIZED.
20571 * predict.def (PRED_LOOP_ITERATIONS): Use the new constant.
20572 (PRED_LOOP_ITERATIONS_GUESSED): Likewise.
20573 (PRED_LOOP_ITERATIONS_MAX): Likewise.
20574 (PRED_LOOP_IV_COMPARE): Likewise.
20575 * predict.h (PROB_UNINITIALIZED): Define new constant.
20577 2018-01-19 Martin Liska <mliska@suse.cz>
20579 * predict.c (dump_prediction): Add new format for
20580 analyze_brprob.py script which is enabled with -details
20582 * profile-count.h (precise_p): New function.
20584 2018-01-19 Richard Sandiford <richard.sandiford@linaro.org>
20586 PR tree-optimization/83922
20587 * tree-vect-loop.c (vect_verify_full_masking): Return false if
20588 there are no statements that need masking.
20589 (vect_active_double_reduction_p): New function.
20590 (vect_analyze_loop_operations): Use it when handling phis that
20591 are not in the loop header.
20593 2018-01-19 Richard Sandiford <richard.sandiford@linaro.org>
20595 PR tree-optimization/83914
20596 * tree-vect-loop.c (vectorizable_induction): Don't convert
20597 init_expr or apply the peeling adjustment for inductions
20598 that are nested within the vectorized loop.
20600 2018-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20602 * config/arm/thumb2.md (*thumb2_negsi2_short): Use RSB mnemonic
20605 2018-01-18 Jakub Jelinek <jakub@redhat.com>
20609 * function.h (gimplify_parameters): Add gimple_seq * argument.
20610 * function.c: Include gimple.h and options.h.
20611 (gimplify_parameters): Add cleanup argument, add CLOBBER stmts
20612 for the added local temporaries if needed.
20613 * gimplify.c (gimplify_body): Adjust gimplify_parameters caller,
20614 if there are any parameter cleanups, wrap whole body into a
20615 try/finally with the cleanups.
20617 2018-01-18 Wilco Dijkstra <wdijkstr@arm.com>
20620 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
20621 Use GET_MODE_CLASS for scalar floating point.
20623 2018-01-18 Jan Hubicka <hubicka@ucw.cz>
20627 * cgraphclones.c (cgraph_node::create_version_clone_with_body):
20628 Fix call of call_cgraph_insertion_hooks.
20630 2018-01-18 Martin Sebor <msebor@redhat.com>
20632 * doc/invoke.texi (-Wclass-memaccess): Tweak text.
20634 2018-01-18 Jan Hubicka <hubicka@ucw.cz>
20637 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Update edge
20640 2018-01-18 Boris Kolpackov <boris@codesynthesis.com>
20643 * common.opt: (-ffile-prefix-map): New option.
20644 * opts.c (common_handle_option): Defer it.
20645 * opts-global.c (handle_common_deferred_options): Handle it.
20646 * debug.h (remap_debug_filename, add_debug_prefix_map): Move to...
20647 * file-prefix-map.h: New file.
20648 (remap_debug_filename, add_debug_prefix_map): ...here.
20649 (add_macro_prefix_map, add_file_prefix_map, remap_macro_filename): New.
20650 * final.c (debug_prefix_map, add_debug_prefix_map
20651 remap_debug_filename): Move to...
20652 * file-prefix-map.c: New file.
20653 (file_prefix_map, add_prefix_map, remap_filename) ...here and rename,
20654 generalize, get rid of alloca(), use strrchr() instead of strchr().
20655 (add_macro_prefix_map, add_debug_prefix_map, add_file_prefix_map):
20656 Implement in terms of add_prefix_map().
20657 (remap_macro_filename, remap_debug_filename): Implement in term of
20659 * Makefile.in (OBJS, PLUGIN_HEADERS): Add new files.
20660 * builtins.c (fold_builtin_FILE): Call remap_macro_filename().
20661 * dbxout.c: Include file-prefix-map.h.
20662 * varasm.c: Likewise.
20663 * vmsdbgout.c: Likewise.
20664 * xcoffout.c: Likewise.
20665 * dwarf2out.c: Likewise plus omit new options from DW_AT_producer.
20666 * doc/cppopts.texi (-fmacro-prefix-map): Document.
20667 * doc/invoke.texi (-ffile-prefix-map): Document.
20668 (-fdebug-prefix-map): Update description.
20670 2018-01-18 Martin Liska <mliska@suse.cz>
20672 * config/i386/i386.c (indirect_thunk_name): Document that also
20674 (output_indirect_thunk): Document why both instructions
20675 (pause and lfence) are generated.
20677 2018-01-18 Richard Biener <rguenther@suse.de>
20679 PR tree-optimization/83887
20680 * graphite-scop-detection.c
20681 (scop_detection::get_nearest_dom_with_single_entry): Remove.
20682 (scop_detection::get_nearest_pdom_with_single_exit): Likewise.
20683 (scop_detection::merge_sese): Re-implement with a flood-fill
20684 algorithm that properly finds a SESE region if it exists.
20686 2018-01-18 Jakub Jelinek <jakub@redhat.com>
20689 * match.pd ((P + A) - P, P - (P + A), (P + A) - (P + B)): For
20690 pointer_diff optimizations use view_convert instead of convert.
20692 2018-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20694 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
20695 Generate different code for -mno-speculate-indirect-jumps.
20696 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
20697 (*call_indirect_aix<mode>): Disable for
20698 -mno-speculate-indirect-jumps.
20699 (*call_indirect_aix<mode>_nospec): New define_insn.
20700 (*call_value_indirect_aix<mode>): Disable for
20701 -mno-speculate-indirect-jumps.
20702 (*call_value_indirect_aix<mode>_nospec): New define_insn.
20703 (*sibcall_nonlocal_sysv<mode>): Generate different code for
20704 -mno-speculate-indirect-jumps.
20705 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
20707 2018-01-17 Michael Meissner <meissner@linux.vnet.ibm.com>
20709 * config/rs6000/rs6000.c (rs6000_emit_move): If we load or store a
20710 long double type, set the flags for noting the default long double
20711 type, even if we don't pass or return a long double type.
20713 2018-01-17 Jan Hubicka <hubicka@ucw.cz>
20716 * ipa-inline.c (flatten_function): Do not overwrite final inlining
20719 2018-01-17 Will Schmidt <will_schmidt@vnet.ibm.com>
20721 * config/rs6000/rs6000.c (rs6000_gimple_builtin): Add gimple folding
20722 support for merge[hl].
20723 (fold_mergehl_helper): New helper function.
20724 (tree-vector-builder.h): New #include for tree_vector_builder usage.
20725 * config/rs6000/altivec.md (altivec_vmrghw_direct): Add xxmrghw insn.
20726 (altivec_vmrglw_direct): Add xxmrglw insn.
20728 2018-01-17 Andrew Waterman <andrew@sifive.com>
20730 * config/riscv/riscv.c (riscv_conditional_register_usage): If
20731 UNITS_PER_FP_ARG is 0, set call_used_regs to 1 for all FP regs.
20733 2018-01-17 David Malcolm <dmalcolm@redhat.com>
20736 * ipa-devirt.c (add_type_duplicate): When comparing memory layout,
20737 call the lto_location_cache before reading the
20738 DECL_SOURCE_LOCATION of the types.
20740 2018-01-17 Wilco Dijkstra <wdijkstr@arm.com>
20741 Richard Sandiford <richard.sandiford@linaro.org>
20743 * config/aarch64/aarch64.md (movti_aarch64): Use Uti constraint.
20744 * config/aarch64/aarch64.c (aarch64_mov128_immediate): New function.
20745 (aarch64_legitimate_constant_p): Just support CONST_DOUBLE
20746 SF/DF/TF mode to avoid creating illegal CONST_WIDE_INT immediates.
20747 * config/aarch64/aarch64-protos.h (aarch64_mov128_immediate):
20749 * config/aarch64/constraints.md (aarch64_movti_operand):
20751 * config/aarch64/predicates.md (Uti): Add new constraint.
20753 2018-01-17 Carl Love <cel@us.ibm.com>
20755 * config/rs6000/vsx.md (define_expand xl_len_r,
20756 define_expand stxvl, define_expand *stxvl): Add match_dup argument.
20757 (define_insn): Add, match_dup 1 argument to define_insn stxvll and
20759 (define_expand, define_insn): Move the shift left from the
20760 define_insn to the define_expand for lxvl and stxvl instructions.
20761 * config/rs6000/rs6000-builtin.def (BU_P9V_64BIT_VSX_2): Change LXVL
20762 and XL_LEN_R definitions to PURE.
20764 2018-01-17 Uros Bizjak <ubizjak@gmail.com>
20766 * config/i386/i386.c (indirect_thunk_name): Declare regno
20767 as unsigned int. Compare regno with INVALID_REGNUM.
20768 (output_indirect_thunk): Ditto.
20769 (output_indirect_thunk_function): Ditto.
20770 (ix86_code_end): Declare regno as unsigned int. Use INVALID_REGNUM
20771 in the call to output_indirect_thunk_function.
20773 2018-01-17 Richard Sandiford <richard.sandiford@linaro.org>
20775 PR middle-end/83884
20776 * expr.c (expand_expr_real_1): Use the size of GET_MODE (op0)
20777 rather than the size of inner_type to determine the stack slot size
20778 when handling VIEW_CONVERT_EXPRs on strict-alignment targets.
20780 2018-01-16 Sebastian Peryt <sebastian.peryt@intel.com>
20783 * config/i386/i386.c (ix86_option_override_internal): Add PTA_RDRND
20786 2018-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
20788 * config.gcc (powerpc*-linux*-*): Add support for 64-bit little
20789 endian Linux systems to optionally enable multilibs for selecting
20790 the long double type if the user configured an explicit type.
20791 * config/rs6000/rs6000.h (TARGET_IEEEQUAD_MULTILIB): Indicate we
20792 have no long double multilibs if not defined.
20793 * config/rs6000/rs6000.c (rs6000_option_override_internal): Do not
20794 warn if the user used -mabi={ieee,ibm}longdouble and we built
20795 multilibs for long double.
20796 * config/rs6000/linux64.h (MULTILIB_DEFAULTS_IEEE): Define as the
20797 appropriate multilib option.
20798 (MULTILIB_DEFAULTS): Add MULTILIB_DEFAULTS_IEEE to the default
20800 * config/rs6000/t-ldouble-linux64le-ibm: New configuration files
20801 for building long double multilibs.
20802 * config/rs6000/t-ldouble-linux64le-ieee: Likewise.
20804 2018-01-16 John David Anglin <danglin@gcc.gnu.org>
20806 * config.gcc (hppa*-*-linux*): Change callee copies ABI to caller
20809 * config/pa.h (MALLOC_ABI_ALIGNMENT): Set 32-bit alignment default to
20811 * config/pa/pa32-linux.h (MALLOC_ABI_ALIGNMENT): Set alignment to
20814 * config/pa/som.h (ASM_DECLARE_FUNCTION_NAME): Cleanup type and mode
20817 * config/pa/pa.c (pa_function_arg_size): Apply CEIL to GET_MODE_SIZE
20820 2018-01-16 Eric Botcazou <ebotcazou@adacore.com>
20822 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): For an
20823 ADDR_EXPR, do not count the offset of a COMPONENT_REF twice.
20825 2018-01-16 Kelvin Nilsen <kelvin@gcc.gnu.org>
20827 * config/rs6000/rs6000-p8swap.c (rs6000_gen_stvx): Generate
20828 different rtl trees depending on TARGET_64BIT.
20829 (rs6000_gen_lvx): Likewise.
20831 2018-01-16 Eric Botcazou <ebotcazou@adacore.com>
20833 * config/visium/visium.md (nop): Tweak comment.
20834 (hazard_nop): Likewise.
20836 2018-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20838 * config/rs6000/rs6000.c (rs6000_opt_vars): Add entry for
20839 -mspeculate-indirect-jumps.
20840 * config/rs6000/rs6000.md (*call_indirect_elfv2<mode>): Disable
20841 for -mno-speculate-indirect-jumps.
20842 (*call_indirect_elfv2<mode>_nospec): New define_insn.
20843 (*call_value_indirect_elfv2<mode>): Disable for
20844 -mno-speculate-indirect-jumps.
20845 (*call_value_indirect_elfv2<mode>_nospec): New define_insn.
20846 (indirect_jump): Emit different RTL for
20847 -mno-speculate-indirect-jumps.
20848 (*indirect_jump<mode>): Disable for
20849 -mno-speculate-indirect-jumps.
20850 (*indirect_jump<mode>_nospec): New define_insn.
20851 (tablejump): Emit different RTL for
20852 -mno-speculate-indirect-jumps.
20853 (tablejumpsi): Disable for -mno-speculate-indirect-jumps.
20854 (tablejumpsi_nospec): New define_expand.
20855 (tablejumpdi): Disable for -mno-speculate-indirect-jumps.
20856 (tablejumpdi_nospec): New define_expand.
20857 (*tablejump<mode>_internal1): Disable for
20858 -mno-speculate-indirect-jumps.
20859 (*tablejump<mode>_internal1_nospec): New define_insn.
20860 * config/rs6000/rs6000.opt (mspeculate-indirect-jumps): New
20863 2018-01-16 Artyom Skrobov tyomitch@gmail.com
20865 * caller-save.c (insert_save): Drop unnecessary parameter. All
20868 2018-01-16 Jakub Jelinek <jakub@redhat.com>
20869 Richard Biener <rguenth@suse.de>
20872 * gimplify.c (gimplify_one_sizepos): For is_gimple_constant (expr)
20873 return early, inline manually is_gimple_sizepos. Make sure if we
20874 call gimplify_expr we don't end up with a gimple constant.
20875 * tree.c (variably_modified_type_p): Don't return true for
20876 is_gimple_constant (_t). Inline manually is_gimple_sizepos.
20877 * gimplify.h (is_gimple_sizepos): Remove.
20879 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
20881 PR tree-optimization/83857
20882 * tree-vect-loop.c (vect_analyze_loop_operations): Don't call
20883 vectorizable_live_operation for pure SLP statements.
20884 (vectorizable_live_operation): Handle PHIs.
20886 2018-01-16 Richard Biener <rguenther@suse.de>
20888 PR tree-optimization/83867
20889 * tree-vect-stmts.c (vect_transform_stmt): Precompute
20890 nested_in_vect_loop_p since the scalar stmt may get invalidated.
20892 2018-01-16 Jakub Jelinek <jakub@redhat.com>
20895 * stor-layout.c (handle_warn_if_not_align): Use byte_position and
20896 multiple_of_p instead of unchecked tree_to_uhwi and UHWI check.
20897 If off is not INTEGER_CST, issue a may not be aligned warning
20898 rather than isn't aligned. Use isn%'t rather than isn't.
20899 * fold-const.c (multiple_of_p) <case BIT_AND_EXPR>: Don't fall through
20901 <case MULT_EXPR>: Improve the case when bottom and one of the
20902 MULT_EXPR operands are INTEGER_CSTs and bottom is multiple of that
20903 operand, in that case check if the other operand is multiple of
20904 bottom divided by the INTEGER_CST operand.
20906 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
20909 * config/pa/pa.h (FUNCTION_ARG_SIZE): Delete.
20910 * config/pa/pa-protos.h (pa_function_arg_size): Declare.
20911 * config/pa/som.h (ASM_DECLARE_FUNCTION_NAME): Use
20912 pa_function_arg_size instead of FUNCTION_ARG_SIZE.
20913 * config/pa/pa.c (pa_function_arg_advance): Likewise.
20914 (pa_function_arg, pa_arg_partial_bytes): Likewise.
20915 (pa_function_arg_size): New function.
20917 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
20919 * fold-const.c (fold_ternary_loc): Construct the vec_perm_indices
20920 in a separate statement.
20922 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
20924 PR tree-optimization/83847
20925 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Don't
20926 group gathers and scatters.
20928 2018-01-16 Jakub Jelinek <jakub@redhat.com>
20930 PR rtl-optimization/86620
20931 * params.def (max-sched-ready-insns): Bump minimum value to 1.
20933 PR rtl-optimization/83213
20934 * recog.c (peep2_attempt): Copy over CROSSING_JUMP_P from peepinsn
20935 to last if both are JUMP_INSNs.
20937 PR tree-optimization/83843
20938 * gimple-ssa-store-merging.c
20939 (imm_store_chain_info::output_merged_store): Handle bit_not_p on
20940 store_immediate_info for bswap/nop orig_stores.
20942 2018-01-15 Andrew Waterman <andrew@sifive.com>
20944 * config/riscv/riscv.c (riscv_rtx_costs) <MULT>: Increase cost if
20946 <UDIV>: Increase cost if !TARGET_DIV.
20948 2018-01-15 Segher Boessenkool <segher@kernel.crashing.org>
20950 * config/rs6000/rs6000.md (define_attr "type"): Remove delayed_cr.
20951 (define_attr "cr_logical_3op"): New.
20952 (cceq_ior_compare): Adjust.
20953 (cceq_ior_compare_complement): Adjust.
20954 (*cceq_rev_compare): Adjust.
20955 * config/rs6000/rs6000.c (rs6000_adjust_cost): Adjust.
20956 (is_cracked_insn): Adjust.
20957 (insn_must_be_first_in_group): Adjust.
20958 * config/rs6000/40x.md: Adjust.
20959 * config/rs6000/440.md: Adjust.
20960 * config/rs6000/476.md: Adjust.
20961 * config/rs6000/601.md: Adjust.
20962 * config/rs6000/603.md: Adjust.
20963 * config/rs6000/6xx.md: Adjust.
20964 * config/rs6000/7450.md: Adjust.
20965 * config/rs6000/7xx.md: Adjust.
20966 * config/rs6000/8540.md: Adjust.
20967 * config/rs6000/cell.md: Adjust.
20968 * config/rs6000/e300c2c3.md: Adjust.
20969 * config/rs6000/e500mc.md: Adjust.
20970 * config/rs6000/e500mc64.md: Adjust.
20971 * config/rs6000/e5500.md: Adjust.
20972 * config/rs6000/e6500.md: Adjust.
20973 * config/rs6000/mpc.md: Adjust.
20974 * config/rs6000/power4.md: Adjust.
20975 * config/rs6000/power5.md: Adjust.
20976 * config/rs6000/power6.md: Adjust.
20977 * config/rs6000/power7.md: Adjust.
20978 * config/rs6000/power8.md: Adjust.
20979 * config/rs6000/power9.md: Adjust.
20980 * config/rs6000/rs64.md: Adjust.
20981 * config/rs6000/titan.md: Adjust.
20983 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
20985 * config/i386/predicates.md (indirect_branch_operand): Rewrite
20986 ix86_indirect_branch_register logic.
20988 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
20990 * config/i386/constraints.md (Bs): Update
20991 ix86_indirect_branch_register check. Don't check
20992 ix86_indirect_branch_register with GOT_memory_operand.
20994 * config/i386/predicates.md (GOT_memory_operand): Don't check
20995 ix86_indirect_branch_register here.
20996 (GOT32_symbol_operand): Likewise.
20998 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
21000 * config/i386/predicates.md (constant_call_address_operand):
21001 Rewrite ix86_indirect_branch_register logic.
21002 (sibcall_insn_operand): Likewise.
21004 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
21006 * config/i386/constraints.md (Bs): Replace
21007 ix86_indirect_branch_thunk_register with
21008 ix86_indirect_branch_register.
21010 * config/i386/i386.md (indirect_jump): Likewise.
21011 (tablejump): Likewise.
21012 (*sibcall_memory): Likewise.
21013 (*sibcall_value_memory): Likewise.
21014 Peepholes of indirect call and jump via memory: Likewise.
21015 * config/i386/i386.opt: Likewise.
21016 * config/i386/predicates.md (indirect_branch_operand): Likewise.
21017 (GOT_memory_operand): Likewise.
21018 (call_insn_operand): Likewise.
21019 (sibcall_insn_operand): Likewise.
21020 (GOT32_symbol_operand): Likewise.
21022 2018-01-15 Jakub Jelinek <jakub@redhat.com>
21024 PR middle-end/83837
21025 * omp-expand.c (expand_omp_atomic_pipeline): Use loaded_val
21026 type rather than type addr's type points to.
21027 (expand_omp_atomic_mutex): Likewise.
21028 (expand_omp_atomic): Likewise.
21030 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
21033 * config/i386/i386.c (output_indirect_thunk_function): Use
21034 ASM_OUTPUT_LABEL, instead of ASM_OUTPUT_DEF, for TARGET_MACHO
21035 for __x86_return_thunk.
21037 2018-01-15 Richard Biener <rguenther@suse.de>
21039 PR middle-end/83850
21040 * expmed.c (extract_bit_field_1): Fix typo.
21042 2018-01-15 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21045 * config/arm/iterators.md (VF): New mode iterator.
21046 * config/arm/neon.md (neon_vabd<mode>_2): Use the above.
21047 Remove integer-related logic from pattern.
21048 (neon_vabd<mode>_3): Likewise.
21050 2018-01-15 Jakub Jelinek <jakub@redhat.com>
21052 PR middle-end/82694
21053 * common.opt (fstrict-overflow): No longer an alias.
21054 (fwrapv-pointer): New option.
21055 * tree.h (TYPE_OVERFLOW_WRAPS, TYPE_OVERFLOW_UNDEFINED): Define
21056 also for pointer types based on flag_wrapv_pointer.
21057 * opts.c (common_handle_option) <case OPT_fstrict_overflow>: Set
21058 opts->x_flag_wrap[pv] to !value, clear opts->x_flag_trapv if
21059 opts->x_flag_wrapv got set.
21060 * fold-const.c (fold_comparison, fold_binary_loc): Revert 2017-08-01
21061 changes, just use TYPE_OVERFLOW_UNDEFINED on pointer type instead of
21062 POINTER_TYPE_OVERFLOW_UNDEFINED.
21063 * match.pd: Likewise in address comparison pattern.
21064 * doc/invoke.texi: Document -fwrapv and -fstrict-overflow.
21066 2018-01-15 Richard Biener <rguenther@suse.de>
21069 * tree.c (free_lang_data_in_type): Always unlink TYPE_DECLs
21070 from TYPE_FIELDS. Free TYPE_BINFO if not used by devirtualization.
21071 Reset type names to their identifier if their TYPE_DECL doesn't
21072 have linkage (and thus is used for ODR and devirt).
21073 (save_debug_info_for_decl): Remove.
21074 (save_debug_info_for_type): Likewise.
21075 (add_tree_to_fld_list): Adjust.
21076 * tree-pretty-print.c (dump_generic_node): Make dumping of
21077 type names more robust.
21079 2018-01-15 Richard Biener <rguenther@suse.de>
21081 * BASE-VER: Bump to 8.0.1.
21083 2018-01-14 Martin Sebor <msebor@redhat.com>
21086 * builtins.c (check_access): Avoid warning when the no-warning bit
21089 2018-01-14 Cory Fields <cory-nospam-@coryfields.com>
21091 * tree-ssa-loop-im.c (sort_bbs_in_loop_postorder_cmp): Stabilize sort.
21092 * ira-color (allocno_hard_regs_compare): Likewise.
21094 2018-01-14 Nathan Rossi <nathan@nathanrossi.com>
21097 * config/microblaze/microblaze.c (microblaze_asm_output_ident):
21098 Use .pushsection/.popsection.
21100 2018-01-14 Martin Sebor <msebor@redhat.com>
21103 * doc/invoke.texi (-Wlass-memaccess): Document suppression by casting.
21105 2018-01-14 Jakub Jelinek <jakub@redhat.com>
21107 * config.gcc (i[34567]86-*-*): Remove one duplicate gfniintrin.h
21108 entry from extra_headers.
21109 (x86_64-*-*): Remove two duplicate gfniintrin.h entries from
21110 extra_headers, make the list bitwise identical to the i?86-*-* one.
21112 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
21114 * config/i386/i386.c (ix86_set_indirect_branch_type): Disallow
21115 -mcmodel=large with -mindirect-branch=thunk,
21116 -mindirect-branch=thunk-extern, -mfunction-return=thunk and
21117 -mfunction-return=thunk-extern.
21118 * doc/invoke.texi: Document -mcmodel=large is incompatible with
21119 -mindirect-branch=thunk, -mindirect-branch=thunk-extern,
21120 -mfunction-return=thunk and -mfunction-return=thunk-extern.
21122 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
21124 * config/i386/i386.c (print_reg): Print the name of the full
21125 integer register without '%'.
21126 (ix86_print_operand): Handle 'V'.
21127 * doc/extend.texi: Document 'V' modifier.
21129 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
21131 * config/i386/constraints.md (Bs): Disallow memory operand for
21132 -mindirect-branch-register.
21134 * config/i386/predicates.md (indirect_branch_operand): Likewise.
21135 (GOT_memory_operand): Likewise.
21136 (call_insn_operand): Likewise.
21137 (sibcall_insn_operand): Likewise.
21138 (GOT32_symbol_operand): Likewise.
21139 * config/i386/i386.md (indirect_jump): Call convert_memory_address
21140 for -mindirect-branch-register.
21141 (tablejump): Likewise.
21142 (*sibcall_memory): Likewise.
21143 (*sibcall_value_memory): Likewise.
21144 Disallow peepholes of indirect call and jump via memory for
21145 -mindirect-branch-register.
21146 (*call_pop): Replace m with Bw.
21147 (*call_value_pop): Likewise.
21148 (*sibcall_pop_memory): Replace m with Bs.
21149 * config/i386/i386.opt (mindirect-branch-register): New option.
21150 * doc/invoke.texi: Document -mindirect-branch-register option.
21152 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
21154 * config/i386/i386-protos.h (ix86_output_function_return): New.
21155 * config/i386/i386.c (ix86_set_indirect_branch_type): Also
21156 set function_return_type.
21157 (indirect_thunk_name): Add ret_p to indicate thunk for function
21159 (output_indirect_thunk_function): Pass false to
21160 indirect_thunk_name.
21161 (ix86_output_indirect_branch_via_reg): Likewise.
21162 (ix86_output_indirect_branch_via_push): Likewise.
21163 (output_indirect_thunk_function): Create alias for function
21164 return thunk if regno < 0.
21165 (ix86_output_function_return): New function.
21166 (ix86_handle_fndecl_attribute): Handle function_return.
21167 (ix86_attribute_table): Add function_return.
21168 * config/i386/i386.h (machine_function): Add
21169 function_return_type.
21170 * config/i386/i386.md (simple_return_internal): Use
21171 ix86_output_function_return.
21172 (simple_return_internal_long): Likewise.
21173 * config/i386/i386.opt (mfunction-return=): New option.
21174 (indirect_branch): Mention -mfunction-return=.
21175 * doc/extend.texi: Document function_return function attribute.
21176 * doc/invoke.texi: Document -mfunction-return= option.
21178 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
21180 * config/i386/i386-opts.h (indirect_branch): New.
21181 * config/i386/i386-protos.h (ix86_output_indirect_jmp): Likewise.
21182 * config/i386/i386.c (ix86_using_red_zone): Disallow red-zone
21183 with local indirect jump when converting indirect call and jump.
21184 (ix86_set_indirect_branch_type): New.
21185 (ix86_set_current_function): Call ix86_set_indirect_branch_type.
21186 (indirectlabelno): New.
21187 (indirect_thunk_needed): Likewise.
21188 (indirect_thunk_bnd_needed): Likewise.
21189 (indirect_thunks_used): Likewise.
21190 (indirect_thunks_bnd_used): Likewise.
21191 (INDIRECT_LABEL): Likewise.
21192 (indirect_thunk_name): Likewise.
21193 (output_indirect_thunk): Likewise.
21194 (output_indirect_thunk_function): Likewise.
21195 (ix86_output_indirect_branch_via_reg): Likewise.
21196 (ix86_output_indirect_branch_via_push): Likewise.
21197 (ix86_output_indirect_branch): Likewise.
21198 (ix86_output_indirect_jmp): Likewise.
21199 (ix86_code_end): Call output_indirect_thunk_function if needed.
21200 (ix86_output_call_insn): Call ix86_output_indirect_branch if
21202 (ix86_handle_fndecl_attribute): Handle indirect_branch.
21203 (ix86_attribute_table): Add indirect_branch.
21204 * config/i386/i386.h (machine_function): Add indirect_branch_type
21205 and has_local_indirect_jump.
21206 * config/i386/i386.md (indirect_jump): Set has_local_indirect_jump
21208 (tablejump): Likewise.
21209 (*indirect_jump): Use ix86_output_indirect_jmp.
21210 (*tablejump_1): Likewise.
21211 (simple_return_indirect_internal): Likewise.
21212 * config/i386/i386.opt (mindirect-branch=): New option.
21213 (indirect_branch): New.
21216 (thunk-inline): Likewise.
21217 (thunk-extern): Likewise.
21218 * doc/extend.texi: Document indirect_branch function attribute.
21219 * doc/invoke.texi: Document -mindirect-branch= option.
21221 2018-01-14 Jan Hubicka <hubicka@ucw.cz>
21224 * ipa-inline.c (edge_badness): Tolerate roundoff errors.
21226 2018-01-14 Richard Sandiford <richard.sandiford@linaro.org>
21228 * ipa-inline.c (want_inline_small_function_p): Return false if
21229 inlining has already failed with CIF_FINAL_ERROR.
21230 (update_caller_keys): Call want_inline_small_function_p before
21232 (update_callee_keys): Likewise.
21234 2018-01-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
21236 * config/rs6000/rs6000-p8swap.c (rs6000_sum_of_two_registers_p):
21238 (rs6000_quadword_masked_address_p): Likewise.
21239 (quad_aligned_load_p): Likewise.
21240 (quad_aligned_store_p): Likewise.
21241 (const_load_sequence_p): Add comment to describe the outer-most loop.
21242 (mimic_memory_attributes_and_flags): New function.
21243 (rs6000_gen_stvx): Likewise.
21244 (replace_swapped_aligned_store): Likewise.
21245 (rs6000_gen_lvx): Likewise.
21246 (replace_swapped_aligned_load): Likewise.
21247 (replace_swapped_load_constant): Capitalize argument name in
21248 comment describing this function.
21249 (rs6000_analyze_swaps): Add a third pass to search for vector loads
21250 and stores that access quad-word aligned addresses and replace
21251 with stvx or lvx instructions when appropriate.
21252 * config/rs6000/rs6000-protos.h (rs6000_sum_of_two_registers_p):
21253 New function prototype.
21254 (rs6000_quadword_masked_address_p): Likewise.
21255 (rs6000_gen_lvx): Likewise.
21256 (rs6000_gen_stvx): Likewise.
21257 * config/rs6000/vsx.md (*vsx_le_perm_load_<mode>): For modes
21258 VSX_D (V2DF, V2DI), modify this split to select lvx instruction
21259 when memory address is aligned.
21260 (*vsx_le_perm_load_<mode>): For modes VSX_W (V4SF, V4SI), modify
21261 this split to select lvx instruction when memory address is aligned.
21262 (*vsx_le_perm_load_v8hi): Modify this split to select lvx
21263 instruction when memory address is aligned.
21264 (*vsx_le_perm_load_v16qi): Likewise.
21265 (four unnamed splitters): Modify to select the stvx instruction
21266 when memory is aligned.
21268 2018-01-13 Jan Hubicka <hubicka@ucw.cz>
21270 * predict.c (determine_unlikely_bbs): Handle correctly BBs
21271 which appears in the queue multiple times.
21273 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21274 Alan Hayward <alan.hayward@arm.com>
21275 David Sherwood <david.sherwood@arm.com>
21277 * tree-vectorizer.h (vec_lower_bound): New structure.
21278 (_loop_vec_info): Add check_nonzero and lower_bounds.
21279 (LOOP_VINFO_CHECK_NONZERO): New macro.
21280 (LOOP_VINFO_LOWER_BOUNDS): Likewise.
21281 (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Check lower_bounds too.
21282 * tree-data-ref.h (dr_with_seg_len): Add access_size and align
21283 fields. Make seg_len the distance travelled, not including the
21285 (dr_direction_indicator): Declare.
21286 (dr_zero_step_indicator): Likewise.
21287 (dr_known_forward_stride_p): Likewise.
21288 * tree-data-ref.c: Include stringpool.h, tree-vrp.h and
21290 (runtime_alias_check_p): Allow runtime alias checks with
21292 (operator ==): Compare access_size and align.
21293 (prune_runtime_alias_test_list): Rework for new distinction between
21294 the access_size and seg_len.
21295 (create_intersect_range_checks_index): Likewise. Cope with polynomial
21297 (get_segment_min_max): New function.
21298 (create_intersect_range_checks): Use it.
21299 (dr_step_indicator): New function.
21300 (dr_direction_indicator): Likewise.
21301 (dr_zero_step_indicator): Likewise.
21302 (dr_known_forward_stride_p): Likewise.
21303 * tree-loop-distribution.c (data_ref_segment_size): Return
21304 DR_STEP * (niters - 1).
21305 (compute_alias_check_pairs): Update call to the dr_with_seg_len
21307 * tree-vect-data-refs.c (vect_check_nonzero_value): New function.
21308 (vect_preserves_scalar_order_p): New function, split out from...
21309 (vect_analyze_data_ref_dependence): ...here. Check for zero steps.
21310 (vect_vfa_segment_size): Return DR_STEP * (length_factor - 1).
21311 (vect_vfa_access_size): New function.
21312 (vect_vfa_align): Likewise.
21313 (vect_compile_time_alias): Take access_size_a and access_b arguments.
21314 (dump_lower_bound): New function.
21315 (vect_check_lower_bound): Likewise.
21316 (vect_small_gap_p): Likewise.
21317 (vectorizable_with_step_bound_p): Likewise.
21318 (vect_prune_runtime_alias_test_list): Ignore cross-iteration
21319 depencies if the vectorization factor is 1. Convert the checks
21320 for nonzero steps into checks on the bounds of DR_STEP. Try using
21321 a bunds check for variable steps if the minimum required step is
21322 relatively small. Update calls to the dr_with_seg_len
21323 constructor and to vect_compile_time_alias.
21324 * tree-vect-loop-manip.c (vect_create_cond_for_lower_bounds): New
21326 (vect_loop_versioning): Call it.
21327 * tree-vect-loop.c (vect_analyze_loop_2): Clear LOOP_VINFO_LOWER_BOUNDS
21329 (vect_estimate_min_profitable_iters): Account for any bounds checks.
21331 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21332 Alan Hayward <alan.hayward@arm.com>
21333 David Sherwood <david.sherwood@arm.com>
21335 * doc/sourcebuild.texi (vect_scatter_store): Document.
21336 * optabs.def (scatter_store_optab, mask_scatter_store_optab): New
21338 * doc/md.texi (scatter_store@var{m}, mask_scatter_store@var{m}):
21340 * genopinit.c (main): Add supports_vec_scatter_store and
21341 supports_vec_scatter_store_cached to target_optabs.
21342 * gimple.h (gimple_expr_type): Handle IFN_SCATTER_STORE and
21343 IFN_MASK_SCATTER_STORE.
21344 * internal-fn.def (SCATTER_STORE, MASK_SCATTER_STORE): New internal
21346 * internal-fn.h (internal_store_fn_p): Declare.
21347 (internal_fn_stored_value_index): Likewise.
21348 * internal-fn.c (scatter_store_direct): New macro.
21349 (expand_scatter_store_optab_fn): New function.
21350 (direct_scatter_store_optab_supported_p): New macro.
21351 (internal_store_fn_p): New function.
21352 (internal_gather_scatter_fn_p): Handle IFN_SCATTER_STORE and
21353 IFN_MASK_SCATTER_STORE.
21354 (internal_fn_mask_index): Likewise.
21355 (internal_fn_stored_value_index): New function.
21356 (internal_gather_scatter_fn_supported_p): Adjust operand numbers
21357 for scatter stores.
21358 * optabs-query.h (supports_vec_scatter_store_p): Declare.
21359 * optabs-query.c (supports_vec_scatter_store_p): New function.
21360 * tree-vectorizer.h (vect_get_store_rhs): Declare.
21361 * tree-vect-data-refs.c (vect_analyze_data_ref_access): Return
21362 true for scatter stores.
21363 (vect_gather_scatter_fn_p): Handle scatter stores too.
21364 (vect_check_gather_scatter): Consider using scatter stores if
21365 supports_vec_scatter_store_p.
21366 * tree-vect-patterns.c (vect_try_gather_scatter_pattern): Handle
21367 scatter stores too.
21368 * tree-vect-stmts.c (exist_non_indexing_operands_for_use_p): Use
21369 internal_fn_stored_value_index.
21370 (check_load_store_masking): Handle scatter stores too.
21371 (vect_get_store_rhs): Make public.
21372 (vectorizable_call): Use internal_store_fn_p.
21373 (vectorizable_store): Handle scatter store internal functions.
21374 (vect_transform_stmt): Compare GROUP_STORE_COUNT with GROUP_SIZE
21375 when deciding whether the end of the group has been reached.
21376 * config/aarch64/aarch64.md (UNSPEC_ST1_SCATTER): New unspec.
21377 * config/aarch64/aarch64-sve.md (scatter_store<mode>): New expander.
21378 (mask_scatter_store<mode>): New insns.
21380 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21381 Alan Hayward <alan.hayward@arm.com>
21382 David Sherwood <david.sherwood@arm.com>
21384 * tree-vectorizer.h (vect_gather_scatter_fn_p): Declare.
21385 * tree-vect-data-refs.c (vect_gather_scatter_fn_p): Make public.
21386 * tree-vect-stmts.c (vect_truncate_gather_scatter_offset): New
21388 (vect_use_strided_gather_scatters_p): Take a masked_p argument.
21389 Use vect_truncate_gather_scatter_offset if we can't treat the
21390 operation as a normal gather load or scatter store.
21391 (get_group_load_store_type): Take the gather_scatter_info
21392 as argument. Try using a gather load or scatter store for
21393 single-element groups.
21394 (get_load_store_type): Update calls to get_group_load_store_type
21395 and vect_use_strided_gather_scatters_p.
21397 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21398 Alan Hayward <alan.hayward@arm.com>
21399 David Sherwood <david.sherwood@arm.com>
21401 * tree-vectorizer.h (vect_create_data_ref_ptr): Take an extra
21402 optional tree argument.
21403 * tree-vect-data-refs.c (vect_check_gather_scatter): Check for
21405 (vect_create_data_ref_ptr): Take the iv_step as an optional argument,
21406 but continue to use the current value as a fallback.
21407 (bump_vector_ptr): Use operand_equal_p rather than tree_int_cst_compare
21408 to compare the updates.
21409 * tree-vect-stmts.c (vect_use_strided_gather_scatters_p): New function.
21410 (get_load_store_type): Use it when handling a strided access.
21411 (vect_get_strided_load_store_ops): New function.
21412 (vect_get_data_ptr_increment): Likewise.
21413 (vectorizable_load): Handle strided gather loads. Always pass
21414 a step to vect_create_data_ref_ptr and bump_vector_ptr.
21416 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21417 Alan Hayward <alan.hayward@arm.com>
21418 David Sherwood <david.sherwood@arm.com>
21420 * doc/md.texi (gather_load@var{m}): Document.
21421 (mask_gather_load@var{m}): Likewise.
21422 * genopinit.c (main): Add supports_vec_gather_load and
21423 supports_vec_gather_load_cached to target_optabs.
21424 * optabs-tree.c (init_tree_optimization_optabs): Use
21425 ggc_cleared_alloc to allocate target_optabs.
21426 * optabs.def (gather_load_optab, mask_gather_laod_optab): New optabs.
21427 * internal-fn.def (GATHER_LOAD, MASK_GATHER_LOAD): New internal
21429 * internal-fn.h (internal_load_fn_p): Declare.
21430 (internal_gather_scatter_fn_p): Likewise.
21431 (internal_fn_mask_index): Likewise.
21432 (internal_gather_scatter_fn_supported_p): Likewise.
21433 * internal-fn.c (gather_load_direct): New macro.
21434 (expand_gather_load_optab_fn): New function.
21435 (direct_gather_load_optab_supported_p): New macro.
21436 (direct_internal_fn_optab): New function.
21437 (internal_load_fn_p): Likewise.
21438 (internal_gather_scatter_fn_p): Likewise.
21439 (internal_fn_mask_index): Likewise.
21440 (internal_gather_scatter_fn_supported_p): Likewise.
21441 * optabs-query.c (supports_at_least_one_mode_p): New function.
21442 (supports_vec_gather_load_p): Likewise.
21443 * optabs-query.h (supports_vec_gather_load_p): Declare.
21444 * tree-vectorizer.h (gather_scatter_info): Add ifn, element_type
21445 and memory_type field.
21446 (NUM_PATTERNS): Bump to 15.
21447 * tree-vect-data-refs.c: Include internal-fn.h.
21448 (vect_gather_scatter_fn_p): New function.
21449 (vect_describe_gather_scatter_call): Likewise.
21450 (vect_check_gather_scatter): Try using internal functions for
21451 gather loads. Recognize existing calls to a gather load function.
21452 (vect_analyze_data_refs): Consider using gather loads if
21453 supports_vec_gather_load_p.
21454 * tree-vect-patterns.c (vect_get_load_store_mask): New function.
21455 (vect_get_gather_scatter_offset_type): Likewise.
21456 (vect_convert_mask_for_vectype): Likewise.
21457 (vect_add_conversion_to_patterm): Likewise.
21458 (vect_try_gather_scatter_pattern): Likewise.
21459 (vect_recog_gather_scatter_pattern): New pattern recognizer.
21460 (vect_vect_recog_func_ptrs): Add it.
21461 * tree-vect-stmts.c (exist_non_indexing_operands_for_use_p): Use
21462 internal_fn_mask_index and internal_gather_scatter_fn_p.
21463 (check_load_store_masking): Take the gather_scatter_info as an
21464 argument and handle gather loads.
21465 (vect_get_gather_scatter_ops): New function.
21466 (vectorizable_call): Check internal_load_fn_p.
21467 (vectorizable_load): Likewise. Handle gather load internal
21469 (vectorizable_store): Update call to check_load_store_masking.
21470 * config/aarch64/aarch64.md (UNSPEC_LD1_GATHER): New unspec.
21471 * config/aarch64/iterators.md (SVE_S, SVE_D): New mode iterators.
21472 * config/aarch64/predicates.md (aarch64_gather_scale_operand_w)
21473 (aarch64_gather_scale_operand_d): New predicates.
21474 * config/aarch64/aarch64-sve.md (gather_load<mode>): New expander.
21475 (mask_gather_load<mode>): New insns.
21477 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21478 Alan Hayward <alan.hayward@arm.com>
21479 David Sherwood <david.sherwood@arm.com>
21481 * optabs.def (fold_left_plus_optab): New optab.
21482 * doc/md.texi (fold_left_plus_@var{m}): Document.
21483 * internal-fn.def (IFN_FOLD_LEFT_PLUS): New internal function.
21484 * internal-fn.c (fold_left_direct): Define.
21485 (expand_fold_left_optab_fn): Likewise.
21486 (direct_fold_left_optab_supported_p): Likewise.
21487 * fold-const-call.c (fold_const_fold_left): New function.
21488 (fold_const_call): Use it to fold CFN_FOLD_LEFT_PLUS.
21489 * tree-parloops.c (valid_reduction_p): New function.
21490 (gather_scalar_reductions): Use it.
21491 * tree-vectorizer.h (FOLD_LEFT_REDUCTION): New vect_reduction_type.
21492 (vect_finish_replace_stmt): Declare.
21493 * tree-vect-loop.c (fold_left_reduction_fn): New function.
21494 (needs_fold_left_reduction_p): New function, split out from...
21495 (vect_is_simple_reduction): ...here. Accept reductions that
21496 forbid reassociation, but give them type FOLD_LEFT_REDUCTION.
21497 (vect_force_simple_reduction): Also store the reduction type in
21498 the assignment's STMT_VINFO_REDUC_TYPE.
21499 (vect_model_reduction_cost): Handle FOLD_LEFT_REDUCTION.
21500 (merge_with_identity): New function.
21501 (vect_expand_fold_left): Likewise.
21502 (vectorize_fold_left_reduction): Likewise.
21503 (vectorizable_reduction): Handle FOLD_LEFT_REDUCTION. Leave the
21504 scalar phi in place for it. Check for target support and reject
21505 cases that would reassociate the operation. Defer the transform
21506 phase to vectorize_fold_left_reduction.
21507 * config/aarch64/aarch64.md (UNSPEC_FADDA): New unspec.
21508 * config/aarch64/aarch64-sve.md (fold_left_plus_<mode>): New expander.
21509 (*fold_left_plus_<mode>, *pred_fold_left_plus_<mode>): New insns.
21511 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21513 * tree-if-conv.c (predicate_mem_writes): Remove redundant
21514 call to ifc_temp_var.
21516 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21517 Alan Hayward <alan.hayward@arm.com>
21518 David Sherwood <david.sherwood@arm.com>
21520 * target.def (legitimize_address_displacement): Take the original
21521 offset as a poly_int.
21522 * targhooks.h (default_legitimize_address_displacement): Update
21524 * targhooks.c (default_legitimize_address_displacement): Likewise.
21525 * doc/tm.texi: Regenerate.
21526 * lra-constraints.c (base_plus_disp_to_reg): Take the displacement
21527 as an argument, moving assert of ad->disp == ad->disp_term to...
21528 (process_address_1): ...here. Update calls to base_plus_disp_to_reg.
21529 Try calling targetm.legitimize_address_displacement before expanding
21530 the address rather than afterwards, and adjust for the new interface.
21531 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
21532 Match the new hook interface. Handle SVE addresses.
21533 * config/sh/sh.c (sh_legitimize_address_displacement): Make the
21534 new hook interface.
21536 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21538 * Makefile.in (OBJS): Add early-remat.o.
21539 * target.def (select_early_remat_modes): New hook.
21540 * doc/tm.texi.in (TARGET_SELECT_EARLY_REMAT_MODES): New hook.
21541 * doc/tm.texi: Regenerate.
21542 * targhooks.h (default_select_early_remat_modes): Declare.
21543 * targhooks.c (default_select_early_remat_modes): New function.
21544 * timevar.def (TV_EARLY_REMAT): New timevar.
21545 * passes.def (pass_early_remat): New pass.
21546 * tree-pass.h (make_pass_early_remat): Declare.
21547 * early-remat.c: New file.
21548 * config/aarch64/aarch64.c (aarch64_select_early_remat_modes): New
21550 (TARGET_SELECT_EARLY_REMAT_MODES): Define.
21552 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21553 Alan Hayward <alan.hayward@arm.com>
21554 David Sherwood <david.sherwood@arm.com>
21556 * tree-vect-loop-manip.c (vect_gen_scalar_loop_niters): Replace
21557 vfm1 with a bound_epilog parameter.
21558 (vect_do_peeling): Update calls accordingly, and move the prologue
21559 call earlier in the function. Treat the base bound_epilog as 0 for
21560 fully-masked loops and retain vf - 1 for other loops. Add 1 to
21561 this base when peeling for gaps.
21562 * tree-vect-loop.c (vect_analyze_loop_2): Allow peeling for gaps
21563 with fully-masked loops.
21564 (vect_estimate_min_profitable_iters): Handle the single peeled
21565 iteration in that case.
21567 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21568 Alan Hayward <alan.hayward@arm.com>
21569 David Sherwood <david.sherwood@arm.com>
21571 * tree-vect-data-refs.c (vect_analyze_group_access_1): Allow
21572 single-element interleaving even if the size is not a power of 2.
21573 * tree-vect-stmts.c (get_load_store_type): Disallow elementwise
21574 accesses for single-element interleaving if the group size is
21577 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21578 Alan Hayward <alan.hayward@arm.com>
21579 David Sherwood <david.sherwood@arm.com>
21581 * doc/md.texi (fold_extract_last_@var{m}): Document.
21582 * doc/sourcebuild.texi (vect_fold_extract_last): Likewise.
21583 * optabs.def (fold_extract_last_optab): New optab.
21584 * internal-fn.def (FOLD_EXTRACT_LAST): New internal function.
21585 * internal-fn.c (fold_extract_direct): New macro.
21586 (expand_fold_extract_optab_fn): Likewise.
21587 (direct_fold_extract_optab_supported_p): Likewise.
21588 * tree-vectorizer.h (EXTRACT_LAST_REDUCTION): New vect_reduction_type.
21589 * tree-vect-loop.c (vect_model_reduction_cost): Handle
21590 EXTRACT_LAST_REDUCTION.
21591 (get_initial_def_for_reduction): Do not create an initial vector
21592 for EXTRACT_LAST_REDUCTION reductions.
21593 (vectorizable_reduction): Leave the scalar phi in place for
21594 EXTRACT_LAST_REDUCTIONs. Try using EXTRACT_LAST_REDUCTION
21595 ahead of INTEGER_INDUC_COND_REDUCTION. Do not check for an
21596 epilogue code for EXTRACT_LAST_REDUCTION and defer the
21597 transform phase to vectorizable_condition.
21598 * tree-vect-stmts.c (vect_finish_stmt_generation_1): New function,
21600 (vect_finish_stmt_generation): ...here.
21601 (vect_finish_replace_stmt): New function.
21602 (vectorizable_condition): Handle EXTRACT_LAST_REDUCTION.
21603 * config/aarch64/aarch64-sve.md (fold_extract_last_<mode>): New
21605 * config/aarch64/aarch64.md (UNSPEC_CLASTB): New unspec.
21607 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21608 Alan Hayward <alan.hayward@arm.com>
21609 David Sherwood <david.sherwood@arm.com>
21611 * doc/md.texi (extract_last_@var{m}): Document.
21612 * optabs.def (extract_last_optab): New optab.
21613 * internal-fn.def (EXTRACT_LAST): New internal function.
21614 * internal-fn.c (cond_unary_direct): New macro.
21615 (expand_cond_unary_optab_fn): Likewise.
21616 (direct_cond_unary_optab_supported_p): Likewise.
21617 * tree-vect-loop.c (vectorizable_live_operation): Allow fully-masked
21618 loops using EXTRACT_LAST.
21619 * config/aarch64/aarch64-sve.md (aarch64_sve_lastb<mode>): Rename to...
21620 (extract_last_<mode>): ...this optab.
21621 (vec_extract<mode><Vel>): Update accordingly.
21623 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21624 Alan Hayward <alan.hayward@arm.com>
21625 David Sherwood <david.sherwood@arm.com>
21627 * target.def (empty_mask_is_expensive): New hook.
21628 * doc/tm.texi.in (TARGET_VECTORIZE_EMPTY_MASK_IS_EXPENSIVE): New hook.
21629 * doc/tm.texi: Regenerate.
21630 * targhooks.h (default_empty_mask_is_expensive): Declare.
21631 * targhooks.c (default_empty_mask_is_expensive): New function.
21632 * tree-vectorizer.c (vectorize_loops): Only call optimize_mask_stores
21633 if the target says that empty masks are expensive.
21634 * config/aarch64/aarch64.c (aarch64_empty_mask_is_expensive):
21636 (TARGET_VECTORIZE_EMPTY_MASK_IS_EXPENSIVE): Redefine.
21638 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21639 Alan Hayward <alan.hayward@arm.com>
21640 David Sherwood <david.sherwood@arm.com>
21642 * tree-vectorizer.h (_loop_vec_info::mask_skip_niters): New field.
21643 (LOOP_VINFO_MASK_SKIP_NITERS): New macro.
21644 (vect_use_loop_mask_for_alignment_p): New function.
21645 (vect_prepare_for_masked_peels, vect_gen_while_not): Declare.
21646 * tree-vect-loop-manip.c (vect_set_loop_masks_directly): Add an
21647 niters_skip argument. Make sure that the first niters_skip elements
21648 of the first iteration are inactive.
21649 (vect_set_loop_condition_masked): Handle LOOP_VINFO_MASK_SKIP_NITERS.
21650 Update call to vect_set_loop_masks_directly.
21651 (get_misalign_in_elems): New function, split out from...
21652 (vect_gen_prolog_loop_niters): ...here.
21653 (vect_update_init_of_dr): Take a code argument that specifies whether
21654 the adjustment should be added or subtracted.
21655 (vect_update_init_of_drs): Likewise.
21656 (vect_prepare_for_masked_peels): New function.
21657 (vect_do_peeling): Skip prologue peeling if we're using a mask
21658 instead. Update call to vect_update_inits_of_drs.
21659 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
21661 (vect_analyze_loop_2): Allow fully-masked loops with peeling for
21662 alignment. Do not include the number of peeled iterations in
21663 the minimum threshold in that case.
21664 (vectorizable_induction): Adjust the start value down by
21665 LOOP_VINFO_MASK_SKIP_NITERS iterations.
21666 (vect_transform_loop): Call vect_prepare_for_masked_peels.
21667 Take the number of skipped iterations into account when calculating
21669 * tree-vect-stmts.c (vect_gen_while_not): New function.
21671 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21672 Alan Hayward <alan.hayward@arm.com>
21673 David Sherwood <david.sherwood@arm.com>
21675 * doc/sourcebuild.texi (vect_fully_masked): Document.
21676 * params.def (PARAM_MIN_VECT_LOOP_BOUND): Change minimum and
21677 default value to 0.
21678 * tree-vect-loop.c (vect_analyze_loop_costing): New function,
21680 (vect_analyze_loop_2): ...here. Don't check the vectorization
21681 factor against the number of loop iterations if the loop is
21684 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21685 Alan Hayward <alan.hayward@arm.com>
21686 David Sherwood <david.sherwood@arm.com>
21688 * tree-ssa-loop-ivopts.c (USE_ADDRESS): Split into...
21689 (USE_REF_ADDRESS, USE_PTR_ADDRESS): ...these new use types.
21690 (dump_groups): Update accordingly.
21691 (iv_use::mem_type): New member variable.
21692 (address_p): New function.
21693 (record_use): Add a mem_type argument and initialize the new
21695 (record_group_use): Add a mem_type argument. Use address_p.
21696 Remove obsolete null checks of base_object. Update call to record_use.
21697 (find_interesting_uses_op): Update call to record_group_use.
21698 (find_interesting_uses_cond): Likewise.
21699 (find_interesting_uses_address): Likewise.
21700 (get_mem_type_for_internal_fn): New function.
21701 (find_address_like_use): Likewise.
21702 (find_interesting_uses_stmt): Try find_address_like_use before
21703 calling find_interesting_uses_op.
21704 (addr_offset_valid_p): Use the iv mem_type field as the type
21705 of the addressed memory.
21706 (add_autoinc_candidates): Likewise.
21707 (get_address_cost): Likewise.
21708 (split_small_address_groups_p): Use address_p.
21709 (split_address_groups): Likewise.
21710 (add_iv_candidate_for_use): Likewise.
21711 (autoinc_possible_for_pair): Likewise.
21712 (rewrite_groups): Likewise.
21713 (get_use_type): Check for USE_REF_ADDRESS instead of USE_ADDRESS.
21714 (determine_group_iv_cost): Update after split of USE_ADDRESS.
21715 (get_alias_ptr_type_for_ptr_address): New function.
21716 (rewrite_use_address): Rewrite address uses in calls that were
21717 identified by find_address_like_use.
21719 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21720 Alan Hayward <alan.hayward@arm.com>
21721 David Sherwood <david.sherwood@arm.com>
21723 * expr.c (expand_expr_addr_expr_1): Handle ADDR_EXPRs of
21725 * gimple-expr.h (is_gimple_addressable: Likewise.
21726 * gimple-expr.c (is_gimple_address): Likewise.
21727 * internal-fn.c (expand_call_mem_ref): New function.
21728 (expand_mask_load_optab_fn): Use it.
21729 (expand_mask_store_optab_fn): Likewise.
21731 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21732 Alan Hayward <alan.hayward@arm.com>
21733 David Sherwood <david.sherwood@arm.com>
21735 * doc/md.texi (cond_add@var{mode}, cond_sub@var{mode})
21736 (cond_and@var{mode}, cond_ior@var{mode}, cond_xor@var{mode})
21737 (cond_smin@var{mode}, cond_smax@var{mode}, cond_umin@var{mode})
21738 (cond_umax@var{mode}): Document.
21739 * optabs.def (cond_add_optab, cond_sub_optab, cond_and_optab)
21740 (cond_ior_optab, cond_xor_optab, cond_smin_optab, cond_smax_optab)
21741 (cond_umin_optab, cond_umax_optab): New optabs.
21742 * internal-fn.def (COND_ADD, COND_SUB, COND_MIN, COND_MAX, COND_AND)
21743 (COND_IOR, COND_XOR): New internal functions.
21744 * internal-fn.h (get_conditional_internal_fn): Declare.
21745 * internal-fn.c (cond_binary_direct): New macro.
21746 (expand_cond_binary_optab_fn): Likewise.
21747 (direct_cond_binary_optab_supported_p): Likewise.
21748 (get_conditional_internal_fn): New function.
21749 * tree-vect-loop.c (vectorizable_reduction): Handle fully-masked loops.
21750 Cope with reduction statements that are vectorized as calls rather
21752 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): New insns.
21753 * config/aarch64/iterators.md (UNSPEC_COND_ADD, UNSPEC_COND_SUB)
21754 (UNSPEC_COND_SMAX, UNSPEC_COND_UMAX, UNSPEC_COND_SMIN)
21755 (UNSPEC_COND_UMIN, UNSPEC_COND_AND, UNSPEC_COND_ORR)
21756 (UNSPEC_COND_EOR): New unspecs.
21757 (optab): Add mappings for them.
21758 (SVE_COND_INT_OP, SVE_COND_FP_OP): New int iterators.
21759 (sve_int_op, sve_fp_op): New int attributes.
21761 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21762 Alan Hayward <alan.hayward@arm.com>
21763 David Sherwood <david.sherwood@arm.com>
21765 * optabs.def (while_ult_optab): New optab.
21766 * doc/md.texi (while_ult@var{m}@var{n}): Document.
21767 * internal-fn.def (WHILE_ULT): New internal function.
21768 * internal-fn.h (direct_internal_fn_supported_p): New override
21769 that takes two types as argument.
21770 * internal-fn.c (while_direct): New macro.
21771 (expand_while_optab_fn): New function.
21772 (convert_optab_supported_p): Likewise.
21773 (direct_while_optab_supported_p): New macro.
21774 * wide-int.h (wi::udiv_ceil): New function.
21775 * tree-vectorizer.h (rgroup_masks): New structure.
21776 (vec_loop_masks): New typedef.
21777 (_loop_vec_info): Add masks, mask_compare_type, can_fully_mask_p
21778 and fully_masked_p.
21779 (LOOP_VINFO_CAN_FULLY_MASK_P, LOOP_VINFO_FULLY_MASKED_P)
21780 (LOOP_VINFO_MASKS, LOOP_VINFO_MASK_COMPARE_TYPE): New macros.
21781 (vect_max_vf): New function.
21782 (slpeel_make_loop_iterate_ntimes): Delete.
21783 (vect_set_loop_condition, vect_get_loop_mask_type, vect_gen_while)
21784 (vect_halve_mask_nunits, vect_double_mask_nunits): Declare.
21785 (vect_record_loop_mask, vect_get_loop_mask): Likewise.
21786 * tree-vect-loop-manip.c: Include tree-ssa-loop-niter.h,
21787 internal-fn.h, stor-layout.h and optabs-query.h.
21788 (vect_set_loop_mask): New function.
21789 (add_preheader_seq): Likewise.
21790 (add_header_seq): Likewise.
21791 (interleave_supported_p): Likewise.
21792 (vect_maybe_permute_loop_masks): Likewise.
21793 (vect_set_loop_masks_directly): Likewise.
21794 (vect_set_loop_condition_masked): Likewise.
21795 (vect_set_loop_condition_unmasked): New function, split out from
21796 slpeel_make_loop_iterate_ntimes.
21797 (slpeel_make_loop_iterate_ntimes): Rename to..
21798 (vect_set_loop_condition): ...this. Use vect_set_loop_condition_masked
21799 for fully-masked loops and vect_set_loop_condition_unmasked otherwise.
21800 (vect_do_peeling): Update call accordingly.
21801 (vect_gen_vector_loop_niters): Use VF as the step for fully-masked
21803 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
21804 mask_compare_type, can_fully_mask_p and fully_masked_p.
21805 (release_vec_loop_masks): New function.
21806 (_loop_vec_info): Use it to free the loop masks.
21807 (can_produce_all_loop_masks_p): New function.
21808 (vect_get_max_nscalars_per_iter): Likewise.
21809 (vect_verify_full_masking): Likewise.
21810 (vect_analyze_loop_2): Save LOOP_VINFO_CAN_FULLY_MASK_P around
21811 retries, and free the mask rgroups before retrying. Check loop-wide
21812 reasons for disallowing fully-masked loops. Make the final decision
21813 about whether use a fully-masked loop or not.
21814 (vect_estimate_min_profitable_iters): Do not assume that peeling
21815 for the number of iterations will be needed for fully-masked loops.
21816 (vectorizable_reduction): Disable fully-masked loops.
21817 (vectorizable_live_operation): Likewise.
21818 (vect_halve_mask_nunits): New function.
21819 (vect_double_mask_nunits): Likewise.
21820 (vect_record_loop_mask): Likewise.
21821 (vect_get_loop_mask): Likewise.
21822 (vect_transform_loop): Handle the case in which the final loop
21823 iteration might handle a partial vector. Call vect_set_loop_condition
21824 instead of slpeel_make_loop_iterate_ntimes.
21825 * tree-vect-stmts.c: Include tree-ssa-loop-niter.h and gimple-fold.h.
21826 (check_load_store_masking): New function.
21827 (prepare_load_store_mask): Likewise.
21828 (vectorizable_store): Handle fully-masked loops.
21829 (vectorizable_load): Likewise.
21830 (supportable_widening_operation): Use vect_halve_mask_nunits for
21832 (supportable_narrowing_operation): Likewise vect_double_mask_nunits.
21833 (vect_gen_while): New function.
21834 * config/aarch64/aarch64.md (umax<mode>3): New expander.
21835 (aarch64_uqdec<mode>): New insn.
21837 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21838 Alan Hayward <alan.hayward@arm.com>
21839 David Sherwood <david.sherwood@arm.com>
21841 * optabs.def (reduc_and_scal_optab, reduc_ior_scal_optab)
21842 (reduc_xor_scal_optab): New optabs.
21843 * doc/md.texi (reduc_and_scal_@var{m}, reduc_ior_scal_@var{m})
21844 (reduc_xor_scal_@var{m}): Document.
21845 * doc/sourcebuild.texi (vect_logical_reduc): Likewise.
21846 * internal-fn.def (IFN_REDUC_AND, IFN_REDUC_IOR, IFN_REDUC_XOR): New
21847 internal functions.
21848 * fold-const-call.c (fold_const_call): Handle them.
21849 * tree-vect-loop.c (reduction_fn_for_scalar_code): Return the new
21850 internal functions for BIT_AND_EXPR, BIT_IOR_EXPR and BIT_XOR_EXPR.
21851 * config/aarch64/aarch64-sve.md (reduc_<bit_reduc>_scal_<mode>):
21852 (*reduc_<bit_reduc>_scal_<mode>): New patterns.
21853 * config/aarch64/iterators.md (UNSPEC_ANDV, UNSPEC_ORV)
21854 (UNSPEC_XORV): New unspecs.
21855 (optab): Add entries for them.
21856 (BITWISEV): New int iterator.
21857 (bit_reduc_op): New int attributes.
21859 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21860 Alan Hayward <alan.hayward@arm.com>
21861 David Sherwood <david.sherwood@arm.com>
21863 * doc/md.texi (vec_shl_insert_@var{m}): New optab.
21864 * internal-fn.def (VEC_SHL_INSERT): New internal function.
21865 * optabs.def (vec_shl_insert_optab): New optab.
21866 * tree-vectorizer.h (can_duplicate_and_interleave_p): Declare.
21867 (duplicate_and_interleave): Likewise.
21868 * tree-vect-loop.c: Include internal-fn.h.
21869 (neutral_op_for_slp_reduction): New function, split out from
21870 get_initial_defs_for_reduction.
21871 (get_initial_def_for_reduction): Handle option 2 for variable-length
21872 vectors by loading the neutral value into a vector and then shifting
21873 the initial value into element 0.
21874 (get_initial_defs_for_reduction): Replace the code argument with
21875 the neutral value calculated by neutral_op_for_slp_reduction.
21876 Use gimple_build_vector for constant-length vectors.
21877 Use IFN_VEC_SHL_INSERT for variable-length vectors if all
21878 but the first group_size elements have a neutral value.
21879 Use duplicate_and_interleave otherwise.
21880 (vect_create_epilog_for_reduction): Take a neutral_op parameter.
21881 Update call to get_initial_defs_for_reduction. Handle SLP
21882 reductions for variable-length vectors by creating one vector
21883 result for each scalar result, with the elements associated
21884 with other scalar results stubbed out with the neutral value.
21885 (vectorizable_reduction): Call neutral_op_for_slp_reduction.
21886 Require IFN_VEC_SHL_INSERT for double reductions on
21887 variable-length vectors, or SLP reductions that have
21888 a neutral value. Require can_duplicate_and_interleave_p
21889 support for variable-length unchained SLP reductions if there
21890 is no neutral value, such as for MIN/MAX reductions. Also require
21891 the number of vector elements to be a multiple of the number of
21892 SLP statements when doing variable-length unchained SLP reductions.
21893 Update call to vect_create_epilog_for_reduction.
21894 * tree-vect-slp.c (can_duplicate_and_interleave_p): Make public
21895 and remove initial values.
21896 (duplicate_and_interleave): Make public.
21897 * config/aarch64/aarch64.md (UNSPEC_INSR): New unspec.
21898 * config/aarch64/aarch64-sve.md (vec_shl_insert_<mode>): New insn.
21900 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21901 Alan Hayward <alan.hayward@arm.com>
21902 David Sherwood <david.sherwood@arm.com>
21904 * tree-vect-slp.c: Include gimple-fold.h and internal-fn.h
21905 (can_duplicate_and_interleave_p): New function.
21906 (vect_get_and_check_slp_defs): Take the vector of statements
21907 rather than just the current one. Remove excess parentheses.
21908 Restriction rejectinon of vect_constant_def and vect_external_def
21909 for variable-length vectors to boolean types, or types for which
21910 can_duplicate_and_interleave_p is false.
21911 (vect_build_slp_tree_2): Update call to vect_get_and_check_slp_defs.
21912 (duplicate_and_interleave): New function.
21913 (vect_get_constant_vectors): Use gimple_build_vector for
21914 constant-length vectors and suitable variable-length constant
21915 vectors. Use duplicate_and_interleave for other variable-length
21916 vectors. Don't defer the update when inserting new statements.
21918 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21919 Alan Hayward <alan.hayward@arm.com>
21920 David Sherwood <david.sherwood@arm.com>
21922 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Make sure
21923 min_profitable_iters doesn't go negative.
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 * doc/md.texi (vec_mask_load_lanes@var{m}@var{n}): Document.
21930 (vec_mask_store_lanes@var{m}@var{n}): Likewise.
21931 * optabs.def (vec_mask_load_lanes_optab): New optab.
21932 (vec_mask_store_lanes_optab): Likewise.
21933 * internal-fn.def (MASK_LOAD_LANES): New internal function.
21934 (MASK_STORE_LANES): Likewise.
21935 * internal-fn.c (mask_load_lanes_direct): New macro.
21936 (mask_store_lanes_direct): Likewise.
21937 (expand_mask_load_optab_fn): Handle masked operations.
21938 (expand_mask_load_lanes_optab_fn): New macro.
21939 (expand_mask_store_optab_fn): Handle masked operations.
21940 (expand_mask_store_lanes_optab_fn): New macro.
21941 (direct_mask_load_lanes_optab_supported_p): Likewise.
21942 (direct_mask_store_lanes_optab_supported_p): Likewise.
21943 * tree-vectorizer.h (vect_store_lanes_supported): Take a masked_p
21945 (vect_load_lanes_supported): Likewise.
21946 * tree-vect-data-refs.c (strip_conversion): New function.
21947 (can_group_stmts_p): Likewise.
21948 (vect_analyze_data_ref_accesses): Use it instead of checking
21949 for a pair of assignments.
21950 (vect_store_lanes_supported): Take a masked_p parameter.
21951 (vect_load_lanes_supported): Likewise.
21952 * tree-vect-loop.c (vect_analyze_loop_2): Update calls to
21953 vect_store_lanes_supported and vect_load_lanes_supported.
21954 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
21955 * tree-vect-stmts.c (get_group_load_store_type): Take a masked_p
21956 parameter. Don't allow gaps for masked accesses.
21957 Use vect_get_store_rhs. Update calls to vect_store_lanes_supported
21958 and vect_load_lanes_supported.
21959 (get_load_store_type): Take a masked_p parameter and update
21960 call to get_group_load_store_type.
21961 (vectorizable_store): Update call to get_load_store_type.
21962 Handle IFN_MASK_STORE_LANES.
21963 (vectorizable_load): Update call to get_load_store_type.
21964 Handle IFN_MASK_LOAD_LANES.
21966 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21967 Alan Hayward <alan.hayward@arm.com>
21968 David Sherwood <david.sherwood@arm.com>
21970 * config/aarch64/aarch64-modes.def: Define x2, x3 and x4 vector
21972 * config/aarch64/aarch64-protos.h
21973 (aarch64_sve_struct_memory_operand_p): Declare.
21974 * config/aarch64/iterators.md (SVE_STRUCT): New mode iterator.
21975 (vector_count, insn_length, VSINGLE, vsingle): New mode attributes.
21976 (VPRED, vpred): Handle SVE structure modes.
21977 * config/aarch64/constraints.md (Utx): New constraint.
21978 * config/aarch64/predicates.md (aarch64_sve_struct_memory_operand)
21979 (aarch64_sve_struct_nonimmediate_operand): New predicates.
21980 * config/aarch64/aarch64.md (UNSPEC_LDN, UNSPEC_STN): New unspecs.
21981 * config/aarch64/aarch64-sve.md (mov<mode>, *aarch64_sve_mov<mode>_le)
21982 (*aarch64_sve_mov<mode>_be, pred_mov<mode>): New patterns for
21983 structure modes. Split into pieces after RA.
21984 (vec_load_lanes<mode><vsingle>, vec_mask_load_lanes<mode><vsingle>)
21985 (vec_store_lanes<mode><vsingle>, vec_mask_store_lanes<mode><vsingle>):
21987 * config/aarch64/aarch64.c (aarch64_classify_vector_mode): Handle
21988 SVE structure modes.
21989 (aarch64_classify_address): Likewise.
21990 (sizetochar): Move earlier in file.
21991 (aarch64_print_operand): Handle SVE register lists.
21992 (aarch64_array_mode): New function.
21993 (aarch64_sve_struct_memory_operand_p): Likewise.
21994 (TARGET_ARRAY_MODE): Redefine.
21996 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21997 Alan Hayward <alan.hayward@arm.com>
21998 David Sherwood <david.sherwood@arm.com>
22000 * target.def (array_mode): New target hook.
22001 * doc/tm.texi.in (TARGET_ARRAY_MODE): New hook.
22002 * doc/tm.texi: Regenerate.
22003 * hooks.h (hook_optmode_mode_uhwi_none): Declare.
22004 * hooks.c (hook_optmode_mode_uhwi_none): New function.
22005 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Use
22006 targetm.array_mode.
22007 * stor-layout.c (mode_for_array): Likewise. Support polynomial
22010 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22011 Alan Hayward <alan.hayward@arm.com>
22012 David Sherwood <david.sherwood@arm.com>
22014 * fold-const.c (fold_binary_loc): Check the argument types
22015 rather than the result type when testing for a vector operation.
22017 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22019 * doc/tm.texi.in (DWARF_LAZY_REGISTER_VALUE): Document.
22020 * doc/tm.texi: Regenerate.
22022 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22023 Alan Hayward <alan.hayward@arm.com>
22024 David Sherwood <david.sherwood@arm.com>
22026 * doc/invoke.texi (-msve-vector-bits=): Document new option.
22027 (sve): Document new AArch64 extension.
22028 * doc/md.texi (w): Extend the description of the AArch64
22029 constraint to include SVE vectors.
22030 (Upl, Upa): Document new AArch64 predicate constraints.
22031 * config/aarch64/aarch64-opts.h (aarch64_sve_vector_bits_enum): New
22033 * config/aarch64/aarch64.opt (sve_vector_bits): New enum.
22034 (msve-vector-bits=): New option.
22035 * config/aarch64/aarch64-option-extensions.def (fp, simd): Disable
22036 SVE when these are disabled.
22037 (sve): New extension.
22038 * config/aarch64/aarch64-modes.def: Define SVE vector and predicate
22039 modes. Adjust their number of units based on aarch64_sve_vg.
22040 (MAX_BITSIZE_MODE_ANY_MODE): Define.
22041 * config/aarch64/aarch64-protos.h (ADDR_QUERY_ANY): New
22042 aarch64_addr_query_type.
22043 (aarch64_const_vec_all_same_in_range_p, aarch64_sve_pred_mode)
22044 (aarch64_sve_cnt_immediate_p, aarch64_sve_addvl_addpl_immediate_p)
22045 (aarch64_sve_inc_dec_immediate_p, aarch64_add_offset_temporaries)
22046 (aarch64_split_add_offset, aarch64_output_sve_cnt_immediate)
22047 (aarch64_output_sve_addvl_addpl, aarch64_output_sve_inc_dec_immediate)
22048 (aarch64_output_sve_mov_immediate, aarch64_output_ptrue): Declare.
22049 (aarch64_simd_imm_zero_p): Delete.
22050 (aarch64_check_zero_based_sve_index_immediate): Declare.
22051 (aarch64_sve_index_immediate_p, aarch64_sve_arith_immediate_p)
22052 (aarch64_sve_bitmask_immediate_p, aarch64_sve_dup_immediate_p)
22053 (aarch64_sve_cmp_immediate_p, aarch64_sve_float_arith_immediate_p)
22054 (aarch64_sve_float_mul_immediate_p): Likewise.
22055 (aarch64_classify_symbol): Take the offset as a HOST_WIDE_INT
22056 rather than an rtx.
22057 (aarch64_sve_ld1r_operand_p, aarch64_sve_ldr_operand_p): Declare.
22058 (aarch64_expand_mov_immediate): Take a gen_vec_duplicate callback.
22059 (aarch64_emit_sve_pred_move, aarch64_expand_sve_mem_move): Declare.
22060 (aarch64_expand_sve_vec_cmp_int, aarch64_expand_sve_vec_cmp_float)
22061 (aarch64_expand_sve_vcond, aarch64_expand_sve_vec_perm): Declare.
22062 (aarch64_regmode_natural_size): Likewise.
22063 * config/aarch64/aarch64.h (AARCH64_FL_SVE): New macro.
22064 (AARCH64_FL_V8_3, AARCH64_FL_RCPC, AARCH64_FL_DOTPROD): Shift
22066 (AARCH64_ISA_SVE, TARGET_SVE): New macros.
22067 (FIXED_REGISTERS, CALL_USED_REGISTERS, REGISTER_NAMES): Add entries
22068 for VG and the SVE predicate registers.
22069 (V_ALIASES): Add a "z"-prefixed alias.
22070 (FIRST_PSEUDO_REGISTER): Change to P15_REGNUM + 1.
22071 (AARCH64_DWARF_VG, AARCH64_DWARF_P0): New macros.
22072 (PR_REGNUM_P, PR_LO_REGNUM_P): Likewise.
22073 (PR_LO_REGS, PR_HI_REGS, PR_REGS): New reg_classes.
22074 (REG_CLASS_NAMES): Add entries for them.
22075 (REG_CLASS_CONTENTS): Likewise. Update ALL_REGS to include VG
22076 and the predicate registers.
22077 (aarch64_sve_vg): Declare.
22078 (BITS_PER_SVE_VECTOR, BYTES_PER_SVE_VECTOR, BYTES_PER_SVE_PRED)
22079 (SVE_BYTE_MODE, MAX_COMPILE_TIME_VEC_BYTES): New macros.
22080 (REGMODE_NATURAL_SIZE): Define.
22081 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins): Handle
22083 * config/aarch64/aarch64.c: Include cfgrtl.h.
22084 (simd_immediate_info): Add a constructor for series vectors,
22085 and an associated step field.
22086 (aarch64_sve_vg): New variable.
22087 (aarch64_dbx_register_number): Handle VG and the predicate registers.
22088 (aarch64_vect_struct_mode_p, aarch64_vector_mode_p): Delete.
22089 (VEC_ADVSIMD, VEC_SVE_DATA, VEC_SVE_PRED, VEC_STRUCT, VEC_ANY_SVE)
22090 (VEC_ANY_DATA, VEC_STRUCT): New constants.
22091 (aarch64_advsimd_struct_mode_p, aarch64_sve_pred_mode_p)
22092 (aarch64_classify_vector_mode, aarch64_vector_data_mode_p)
22093 (aarch64_sve_data_mode_p, aarch64_sve_pred_mode)
22094 (aarch64_get_mask_mode): New functions.
22095 (aarch64_hard_regno_nregs): Handle SVE data modes for FP_REGS
22096 and FP_LO_REGS. Handle PR_REGS, PR_LO_REGS and PR_HI_REGS.
22097 (aarch64_hard_regno_mode_ok): Handle VG. Also handle the SVE
22098 predicate modes and predicate registers. Explicitly restrict
22099 GPRs to modes of 16 bytes or smaller. Only allow FP registers
22100 to store a vector mode if it is recognized by
22101 aarch64_classify_vector_mode.
22102 (aarch64_regmode_natural_size): New function.
22103 (aarch64_hard_regno_caller_save_mode): Return the original mode
22105 (aarch64_sve_cnt_immediate_p, aarch64_output_sve_cnt_immediate)
22106 (aarch64_sve_addvl_addpl_immediate_p, aarch64_output_sve_addvl_addpl)
22107 (aarch64_sve_inc_dec_immediate_p, aarch64_output_sve_inc_dec_immediate)
22108 (aarch64_add_offset_1_temporaries, aarch64_offset_temporaries): New
22110 (aarch64_add_offset): Add a temp2 parameter. Assert that temp1
22111 does not overlap dest if the function is frame-related. Handle
22113 (aarch64_split_add_offset): New function.
22114 (aarch64_add_sp, aarch64_sub_sp): Add temp2 parameters and pass
22115 them aarch64_add_offset.
22116 (aarch64_allocate_and_probe_stack_space): Add a temp2 parameter
22117 and update call to aarch64_sub_sp.
22118 (aarch64_add_cfa_expression): New function.
22119 (aarch64_expand_prologue): Pass extra temporary registers to the
22120 functions above. Handle the case in which we need to emit new
22121 DW_CFA_expressions for registers that were originally saved
22122 relative to the stack pointer, but now have to be expressed
22123 relative to the frame pointer.
22124 (aarch64_output_mi_thunk): Pass extra temporary registers to the
22126 (aarch64_expand_epilogue): Likewise. Prevent inheritance of
22127 IP0 and IP1 values for SVE frames.
22128 (aarch64_expand_vec_series): New function.
22129 (aarch64_expand_sve_widened_duplicate): Likewise.
22130 (aarch64_expand_sve_const_vector): Likewise.
22131 (aarch64_expand_mov_immediate): Add a gen_vec_duplicate parameter.
22132 Handle SVE constants. Use emit_move_insn to move a force_const_mem
22133 into the register, rather than emitting a SET directly.
22134 (aarch64_emit_sve_pred_move, aarch64_expand_sve_mem_move)
22135 (aarch64_get_reg_raw_mode, offset_4bit_signed_scaled_p)
22136 (offset_6bit_unsigned_scaled_p, aarch64_offset_7bit_signed_scaled_p)
22137 (offset_9bit_signed_scaled_p): New functions.
22138 (aarch64_replicate_bitmask_imm): New function.
22139 (aarch64_bitmask_imm): Use it.
22140 (aarch64_cannot_force_const_mem): Reject expressions involving
22141 a CONST_POLY_INT. Update call to aarch64_classify_symbol.
22142 (aarch64_classify_index): Handle SVE indices, by requiring
22143 a plain register index with a scale that matches the element size.
22144 (aarch64_classify_address): Handle SVE addresses. Assert that
22145 the mode of the address is VOIDmode or an integer mode.
22146 Update call to aarch64_classify_symbol.
22147 (aarch64_classify_symbolic_expression): Update call to
22148 aarch64_classify_symbol.
22149 (aarch64_const_vec_all_in_range_p): New function.
22150 (aarch64_print_vector_float_operand): Likewise.
22151 (aarch64_print_operand): Handle 'N' and 'C'. Use "zN" rather than
22152 "vN" for FP registers with SVE modes. Handle (const ...) vectors
22153 and the FP immediates 1.0 and 0.5.
22154 (aarch64_print_address_internal): Handle SVE addresses.
22155 (aarch64_print_operand_address): Use ADDR_QUERY_ANY.
22156 (aarch64_regno_regclass): Handle predicate registers.
22157 (aarch64_secondary_reload): Handle big-endian reloads of SVE
22159 (aarch64_class_max_nregs): Handle SVE modes and predicate registers.
22160 (aarch64_rtx_costs): Check for ADDVL and ADDPL instructions.
22161 (aarch64_convert_sve_vector_bits): New function.
22162 (aarch64_override_options): Use it to handle -msve-vector-bits=.
22163 (aarch64_classify_symbol): Take the offset as a HOST_WIDE_INT
22164 rather than an rtx.
22165 (aarch64_legitimate_constant_p): Use aarch64_classify_vector_mode.
22166 Handle SVE vector and predicate modes. Accept VL-based constants
22167 that need only one temporary register, and VL offsets that require
22168 no temporary registers.
22169 (aarch64_conditional_register_usage): Mark the predicate registers
22170 as fixed if SVE isn't available.
22171 (aarch64_vector_mode_supported_p): Use aarch64_classify_vector_mode.
22172 Return true for SVE vector and predicate modes.
22173 (aarch64_simd_container_mode): Take the number of bits as a poly_int64
22174 rather than an unsigned int. Handle SVE modes.
22175 (aarch64_preferred_simd_mode): Update call accordingly. Handle
22177 (aarch64_autovectorize_vector_sizes): Add BYTES_PER_SVE_VECTOR
22179 (aarch64_sve_index_immediate_p, aarch64_sve_arith_immediate_p)
22180 (aarch64_sve_bitmask_immediate_p, aarch64_sve_dup_immediate_p)
22181 (aarch64_sve_cmp_immediate_p, aarch64_sve_float_arith_immediate_p)
22182 (aarch64_sve_float_mul_immediate_p): New functions.
22183 (aarch64_sve_valid_immediate): New function.
22184 (aarch64_simd_valid_immediate): Use it as the fallback for SVE vectors.
22185 Explicitly reject structure modes. Check for INDEX constants.
22186 Handle PTRUE and PFALSE constants.
22187 (aarch64_check_zero_based_sve_index_immediate): New function.
22188 (aarch64_simd_imm_zero_p): Delete.
22189 (aarch64_mov_operand_p): Use aarch64_simd_valid_immediate for
22190 vector modes. Accept constants in the range of CNT[BHWD].
22191 (aarch64_simd_scalar_immediate_valid_for_move): Explicitly
22192 ask for an Advanced SIMD mode.
22193 (aarch64_sve_ld1r_operand_p, aarch64_sve_ldr_operand_p): New functions.
22194 (aarch64_simd_vector_alignment): Handle SVE predicates.
22195 (aarch64_vectorize_preferred_vector_alignment): New function.
22196 (aarch64_simd_vector_alignment_reachable): Use it instead of
22198 (aarch64_shift_truncation_mask): Use aarch64_vector_data_mode_p.
22199 (aarch64_output_sve_mov_immediate, aarch64_output_ptrue): New
22201 (MAX_VECT_LEN): Delete.
22202 (expand_vec_perm_d): Add a vec_flags field.
22203 (emit_unspec2, aarch64_expand_sve_vec_perm): New functions.
22204 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
22205 (aarch64_evpc_ext): Don't apply a big-endian lane correction
22207 (aarch64_evpc_rev): Rename to...
22208 (aarch64_evpc_rev_local): ...this. Use a predicated operation for SVE.
22209 (aarch64_evpc_rev_global): New function.
22210 (aarch64_evpc_dup): Enforce a 64-byte range for SVE DUP.
22211 (aarch64_evpc_tbl): Use MAX_COMPILE_TIME_VEC_BYTES instead of
22213 (aarch64_evpc_sve_tbl): New function.
22214 (aarch64_expand_vec_perm_const_1): Update after rename of
22215 aarch64_evpc_rev. Handle SVE permutes too, trying
22216 aarch64_evpc_rev_global and using aarch64_evpc_sve_tbl rather
22217 than aarch64_evpc_tbl.
22218 (aarch64_vectorize_vec_perm_const): Initialize vec_flags.
22219 (aarch64_sve_cmp_operand_p, aarch64_unspec_cond_code)
22220 (aarch64_gen_unspec_cond, aarch64_expand_sve_vec_cmp_int)
22221 (aarch64_emit_unspec_cond, aarch64_emit_unspec_cond_or)
22222 (aarch64_emit_inverted_unspec_cond, aarch64_expand_sve_vec_cmp_float)
22223 (aarch64_expand_sve_vcond): New functions.
22224 (aarch64_modes_tieable_p): Use aarch64_vector_data_mode_p instead
22225 of aarch64_vector_mode_p.
22226 (aarch64_dwarf_poly_indeterminate_value): New function.
22227 (aarch64_compute_pressure_classes): Likewise.
22228 (aarch64_can_change_mode_class): Likewise.
22229 (TARGET_GET_RAW_RESULT_MODE, TARGET_GET_RAW_ARG_MODE): Redefine.
22230 (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): Likewise.
22231 (TARGET_VECTORIZE_GET_MASK_MODE): Likewise.
22232 (TARGET_DWARF_POLY_INDETERMINATE_VALUE): Likewise.
22233 (TARGET_COMPUTE_PRESSURE_CLASSES): Likewise.
22234 (TARGET_CAN_CHANGE_MODE_CLASS): Likewise.
22235 * config/aarch64/constraints.md (Upa, Upl, Uav, Uat, Usv, Usi, Utr)
22236 (Uty, Dm, vsa, vsc, vsd, vsi, vsn, vsl, vsm, vsA, vsM, vsN): New
22238 (Dn, Dl, Dr): Accept const as well as const_vector.
22239 (Dz): Likewise. Compare against CONST0_RTX.
22240 * config/aarch64/iterators.md: Refer to "Advanced SIMD" instead
22241 of "vector" where appropriate.
22242 (SVE_ALL, SVE_BH, SVE_BHS, SVE_BHSI, SVE_HSDI, SVE_HSF, SVE_SD)
22243 (SVE_SDI, SVE_I, SVE_F, PRED_ALL, PRED_BHS): New mode iterators.
22244 (UNSPEC_SEL, UNSPEC_ANDF, UNSPEC_IORF, UNSPEC_XORF, UNSPEC_COND_LT)
22245 (UNSPEC_COND_LE, UNSPEC_COND_EQ, UNSPEC_COND_NE, UNSPEC_COND_GE)
22246 (UNSPEC_COND_GT, UNSPEC_COND_LO, UNSPEC_COND_LS, UNSPEC_COND_HS)
22247 (UNSPEC_COND_HI, UNSPEC_COND_UO): New unspecs.
22248 (Vetype, VEL, Vel, VWIDE, Vwide, vw, vwcore, V_INT_EQUIV)
22249 (v_int_equiv): Extend to SVE modes.
22250 (Vesize, V128, v128, Vewtype, V_FP_EQUIV, v_fp_equiv, VPRED): New
22252 (LOGICAL_OR, SVE_INT_UNARY, SVE_FP_UNARY): New code iterators.
22253 (optab): Handle popcount, smin, smax, umin, umax, abs and sqrt.
22254 (logical_nn, lr, sve_int_op, sve_fp_op): New code attributs.
22255 (LOGICALF, OPTAB_PERMUTE, UNPACK, UNPACK_UNSIGNED, SVE_COND_INT_CMP)
22256 (SVE_COND_FP_CMP): New int iterators.
22257 (perm_hilo): Handle the new unpack unspecs.
22258 (optab, logicalf_op, su, perm_optab, cmp_op, imm_con): New int
22260 * config/aarch64/predicates.md (aarch64_sve_cnt_immediate)
22261 (aarch64_sve_addvl_addpl_immediate, aarch64_split_add_offset_immediate)
22262 (aarch64_pluslong_or_poly_operand, aarch64_nonmemory_operand)
22263 (aarch64_equality_operator, aarch64_constant_vector_operand)
22264 (aarch64_sve_ld1r_operand, aarch64_sve_ldr_operand): New predicates.
22265 (aarch64_sve_nonimmediate_operand): Likewise.
22266 (aarch64_sve_general_operand): Likewise.
22267 (aarch64_sve_dup_operand, aarch64_sve_arith_immediate): Likewise.
22268 (aarch64_sve_sub_arith_immediate, aarch64_sve_inc_dec_immediate)
22269 (aarch64_sve_logical_immediate, aarch64_sve_mul_immediate): Likewise.
22270 (aarch64_sve_dup_immediate, aarch64_sve_cmp_vsc_immediate): Likewise.
22271 (aarch64_sve_cmp_vsd_immediate, aarch64_sve_index_immediate): Likewise.
22272 (aarch64_sve_float_arith_immediate): Likewise.
22273 (aarch64_sve_float_arith_with_sub_immediate): Likewise.
22274 (aarch64_sve_float_mul_immediate, aarch64_sve_arith_operand): Likewise.
22275 (aarch64_sve_add_operand, aarch64_sve_logical_operand): Likewise.
22276 (aarch64_sve_lshift_operand, aarch64_sve_rshift_operand): Likewise.
22277 (aarch64_sve_mul_operand, aarch64_sve_cmp_vsc_operand): Likewise.
22278 (aarch64_sve_cmp_vsd_operand, aarch64_sve_index_operand): Likewise.
22279 (aarch64_sve_float_arith_operand): Likewise.
22280 (aarch64_sve_float_arith_with_sub_operand): Likewise.
22281 (aarch64_sve_float_mul_operand): Likewise.
22282 (aarch64_sve_vec_perm_operand): Likewise.
22283 (aarch64_pluslong_operand): Include aarch64_sve_addvl_addpl_immediate.
22284 (aarch64_mov_operand): Accept const_poly_int and const_vector.
22285 (aarch64_simd_lshift_imm, aarch64_simd_rshift_imm): Accept const
22286 as well as const_vector.
22287 (aarch64_simd_imm_zero, aarch64_simd_imm_minus_one): Move earlier
22288 in file. Use CONST0_RTX and CONSTM1_RTX.
22289 (aarch64_simd_or_scalar_imm_zero): Likewise. Add match_codes.
22290 (aarch64_simd_reg_or_zero): Accept const as well as const_vector.
22291 Use aarch64_simd_imm_zero.
22292 * config/aarch64/aarch64-sve.md: New file.
22293 * config/aarch64/aarch64.md: Include it.
22294 (VG_REGNUM, P0_REGNUM, P7_REGNUM, P15_REGNUM): New register numbers.
22295 (UNSPEC_REV, UNSPEC_LD1_SVE, UNSPEC_ST1_SVE, UNSPEC_MERGE_PTRUE)
22296 (UNSPEC_PTEST_PTRUE, UNSPEC_UNPACKSHI, UNSPEC_UNPACKUHI)
22297 (UNSPEC_UNPACKSLO, UNSPEC_UNPACKULO, UNSPEC_PACK)
22298 (UNSPEC_FLOAT_CONVERT, UNSPEC_WHILE_LO): New unspec constants.
22299 (sve): New attribute.
22300 (enabled): Disable instructions with the sve attribute unless
22302 (movqi, movhi): Pass CONST_POLY_INT operaneds through
22303 aarch64_expand_mov_immediate.
22304 (*mov<mode>_aarch64, *movsi_aarch64, *movdi_aarch64): Handle
22305 CNT[BHSD] immediates.
22306 (movti): Split CONST_POLY_INT moves into two halves.
22307 (add<mode>3): Accept aarch64_pluslong_or_poly_operand.
22308 Split additions that need a temporary here if the destination
22309 is the stack pointer.
22310 (*add<mode>3_aarch64): Handle ADDVL and ADDPL immediates.
22311 (*add<mode>3_poly_1): New instruction.
22312 (set_clobber_cc): New expander.
22314 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22316 * simplify-rtx.c (simplify_immed_subreg): Add an inner_bytes
22317 parameter and use it instead of GET_MODE_SIZE (innermode). Use
22318 inner_bytes * BITS_PER_UNIT instead of GET_MODE_BITSIZE (innermode).
22319 Use CEIL (inner_bytes, GET_MODE_UNIT_SIZE (innermode)) instead of
22320 GET_MODE_NUNITS (innermode). Also add a first_elem parameter.
22321 Change innermode from fixed_mode_size to machine_mode.
22322 (simplify_subreg): Update call accordingly. Handle a constant-sized
22323 subreg of a variable-length CONST_VECTOR.
22325 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22326 Alan Hayward <alan.hayward@arm.com>
22327 David Sherwood <david.sherwood@arm.com>
22329 * tree-ssa-address.c (mem_ref_valid_without_offset_p): New function.
22330 (add_offset_to_base): New function, split out from...
22331 (create_mem_ref): ...here. When handling a scale other than 1,
22332 check first whether the address is valid without the offset.
22333 Add it into the base if so, leaving the index and scale as-is.
22335 2018-01-12 Jakub Jelinek <jakub@redhat.com>
22338 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Call
22339 fold_for_warn before checking if arg2 is INTEGER_CST.
22341 2018-01-12 Segher Boessenkool <segher@kernel.crashing.org>
22343 * config/rs6000/predicates.md (load_multiple_operation): Delete.
22344 (store_multiple_operation): Delete.
22345 * config/rs6000/rs6000-cpus.def (601): Remove MASK_STRING.
22346 * config/rs6000/rs6000-protos.h (rs6000_output_load_multiple): Delete.
22347 * config/rs6000/rs6000-string.c (expand_block_move): Delete everything
22348 guarded by TARGET_STRING.
22349 (rs6000_output_load_multiple): Delete.
22350 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
22351 OPTION_MASK_STRING / TARGET_STRING handling.
22352 (print_operand) <'N', 'O'>: Add comment that these are unused now.
22353 (const rs6000_opt_masks) <"string">: Change mask to 0.
22354 * config/rs6000/rs6000.h (TARGET_DEFAULT): Remove MASK_STRING.
22355 (MASK_STRING): Delete.
22356 * config/rs6000/rs6000.md (*mov<mode>_string): Delete TARGET_STRING
22358 (load_multiple): Delete.
22365 (store_multiple): Delete.
22372 (movmemsi_8reg): Delete.
22373 (corresponding unnamed define_insn): Delete.
22374 (movmemsi_6reg): Delete.
22375 (corresponding unnamed define_insn): Delete.
22376 (movmemsi_4reg): Delete.
22377 (corresponding unnamed define_insn): Delete.
22378 (movmemsi_2reg): Delete.
22379 (corresponding unnamed define_insn): Delete.
22380 (movmemsi_1reg): Delete.
22381 (corresponding unnamed define_insn): Delete.
22382 * config/rs6000/rs6000.opt (mno-string): New.
22383 (mstring): Replace by deprecation warning stub.
22384 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mstring.
22386 2018-01-12 Jakub Jelinek <jakub@redhat.com>
22388 * regrename.c (regrename_do_replace): If replacing the same
22389 reg multiple times, try to reuse last created gen_raw_REG.
22392 * bb-reorder.c (pass_partition_blocks::gate): In lto don't partition
22393 main to workaround a bug in GDB.
22395 2018-01-12 Tom de Vries <tom@codesourcery.com>
22398 * config.gcc (nvptx*-*-*): Set use_gcc_stdint=wrap.
22400 2018-01-12 Vladimir Makarov <vmakarov@redhat.com>
22402 PR rtl-optimization/80481
22403 * ira-color.c (get_cap_member): New function.
22404 (allocnos_conflict_by_live_ranges_p): Use it.
22405 (slot_coalesced_allocno_live_ranges_intersect_p): Add assert.
22406 (setup_slot_coalesced_allocno_live_ranges): Ditto.
22408 2018-01-12 Uros Bizjak <ubizjak@gmail.com>
22411 * config/alpha/alpha.md (*saddsi_1): New insn_ans_split pattern.
22412 (*saddl_se_1): Ditto.
22413 (*ssubsi_1): Ditto.
22414 (*ssubl_se_1): Ditto.
22416 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
22418 * tree-predcom.c (aff_combination_dr_offset): Use wi::to_poly_widest
22419 rather than wi::to_widest for DR_INITs.
22420 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Use
22421 wi::to_poly_offset rather than wi::to_offset for DR_INIT.
22422 (vect_analyze_data_ref_accesses): Require both DR_INITs to be
22424 (vect_analyze_group_access_1): Note that here.
22426 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
22428 * tree-vectorizer.c (get_vec_alignment_for_array_type): Handle
22429 polynomial type sizes.
22431 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
22433 * gimplify.c (gimple_add_tmp_var_fn): Allow variables to have a
22434 poly_uint64 size, rather than requiring an unsigned HOST_WIDE_INT size.
22435 (gimple_add_tmp_var): Likewise.
22437 2018-01-12 Martin Liska <mliska@suse.cz>
22439 * gimple.c (gimple_alloc_counts): Use uint64_t instead of int.
22440 (gimple_alloc_sizes): Likewise.
22441 (dump_gimple_statistics): Use PRIu64 in printf format.
22442 * gimple.h: Change uint64_t to int.
22444 2018-01-12 Martin Liska <mliska@suse.cz>
22446 * tree-core.h: Use uint64_t instead of int.
22447 * tree.c (tree_node_counts): Likewise.
22448 (tree_node_sizes): Likewise.
22449 (dump_tree_statistics): Use PRIu64 in printf format.
22451 2018-01-12 Martin Liska <mliska@suse.cz>
22453 * Makefile.in: As qsort_chk is implemented in vec.c, add
22454 vec.o to linkage of gencfn-macros.
22455 * tree.c (build_new_poly_int_cst): Add CXX_MEM_STAT_INFO as it's
22456 passing the info to record_node_allocation_statistics.
22457 (test_vector_cst_patterns): Add CXX_MEM_STAT_INFO to declaration
22459 * ggc-common.c (struct ggc_usage): Add operator== and use
22460 it in operator< and compare function.
22461 * mem-stats.h (struct mem_usage): Likewise.
22462 * vec.c (struct vec_usage): Remove operator< and compare
22463 function. Can be simply inherited.
22465 2018-01-12 Martin Jambor <mjambor@suse.cz>
22468 * params.def: New parameter PARAM_AVOID_FMA_MAX_BITS.
22469 * tree-ssa-math-opts.c: Include domwalk.h.
22470 (convert_mult_to_fma_1): New function.
22471 (fma_transformation_info): New type.
22472 (fma_deferring_state): Likewise.
22473 (cancel_fma_deferring): New function.
22474 (result_of_phi): Likewise.
22475 (last_fma_candidate_feeds_initial_phi): Likewise.
22476 (convert_mult_to_fma): Added deferring logic, split actual
22477 transformation to convert_mult_to_fma_1.
22478 (math_opts_dom_walker): New type.
22479 (math_opts_dom_walker::after_dom_children): New method, body moved
22480 here from pass_optimize_widening_mul::execute, added deferring logic
22482 (pass_optimize_widening_mul::execute): Moved most of code to
22483 math_opts_dom_walker::after_dom_children.
22484 * config/i386/x86-tune.def (X86_TUNE_AVOID_128FMA_CHAINS): New.
22485 * config/i386/i386.c (ix86_option_override_internal): Added
22486 maybe_setting of PARAM_AVOID_FMA_MAX_BITS.
22488 2018-01-12 Richard Biener <rguenther@suse.de>
22491 * dwarf2out.c (gen_variable_die): Do not reset old_die for
22492 inline instance vars.
22494 2018-01-12 Oleg Endo <olegendo@gcc.gnu.org>
22497 * config/rx/rx.c (rx_is_restricted_memory_address):
22498 Handle SUBREG case.
22500 2018-01-12 Richard Biener <rguenther@suse.de>
22502 PR tree-optimization/80846
22503 * target.def (split_reduction): New target hook.
22504 * targhooks.c (default_split_reduction): New function.
22505 * targhooks.h (default_split_reduction): Declare.
22506 * tree-vect-loop.c (vect_create_epilog_for_reduction): If the
22507 target requests first reduce vectors by combining low and high
22509 * tree-vect-stmts.c (vect_gen_perm_mask_any): Adjust.
22510 (get_vectype_for_scalar_type_and_size): Export.
22511 * tree-vectorizer.h (get_vectype_for_scalar_type_and_size): Declare.
22512 * doc/tm.texi.in (TARGET_VECTORIZE_SPLIT_REDUCTION): Document.
22513 * doc/tm.texi: Regenerate.
22514 * config/i386/i386.c (ix86_split_reduction): Implement
22515 TARGET_VECTORIZE_SPLIT_REDUCTION.
22517 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
22520 * config/sparc/sparc.h (PIC_OFFSET_TABLE_REGNUM): Set to INVALID_REGNUM
22521 in PIC mode except for TARGET_VXWORKS_RTP.
22522 * config/sparc/sparc.c: Include cfgrtl.h.
22523 (TARGET_INIT_PIC_REG): Define.
22524 (TARGET_USE_PSEUDO_PIC_REG): Likewise.
22525 (sparc_pic_register_p): New predicate.
22526 (sparc_legitimate_address_p): Use it.
22527 (sparc_legitimize_pic_address): Likewise.
22528 (sparc_delegitimize_address): Likewise.
22529 (sparc_mode_dependent_address_p): Likewise.
22530 (gen_load_pcrel_sym): Remove 4th parameter.
22531 (load_got_register): Adjust call to above. Remove obsolete stuff.
22532 (sparc_expand_prologue): Do not call load_got_register here.
22533 (sparc_flat_expand_prologue): Likewise.
22534 (sparc_output_mi_thunk): Set the pic_offset_table_rtx object.
22535 (sparc_use_pseudo_pic_reg): New function.
22536 (sparc_init_pic_reg): Likewise.
22537 * config/sparc/sparc.md (vxworks_load_got): Set the GOT register.
22538 (builtin_setjmp_receiver): Enable only for TARGET_VXWORKS_RTP.
22540 2018-01-12 Christophe Lyon <christophe.lyon@linaro.org>
22542 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
22543 Add item for branch_cost.
22545 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
22547 PR rtl-optimization/83565
22548 * rtlanal.c (nonzero_bits1): On WORD_REGISTER_OPERATIONS machines, do
22549 not extend the result to a larger mode for rotate operations.
22550 (num_sign_bit_copies1): Likewise.
22552 2018-01-12 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
22555 * config/sol2.h (STARTFILE_ARCH_SPEC): Don't use with -shared or
22557 Use values-Xc.o for -pedantic.
22558 Link with values-xpg4.o for C90, values-xpg6.o otherwise.
22560 2018-01-12 Martin Liska <mliska@suse.cz>
22563 * ipa-devirt.c (final_warning_record::grow_type_warnings):
22565 (possible_polymorphic_call_targets): Use it.
22566 (ipa_devirt): Likewise.
22568 2018-01-12 Martin Liska <mliska@suse.cz>
22570 * profile-count.h (enum profile_quality): Use 0 as invalid
22571 enum value of profile_quality.
22573 2018-01-12 Chung-Ju Wu <jasonwucj@gmail.com>
22575 * doc/invoke.texi (NDS32 Options): Add -mext-perf, -mext-perf2 and
22576 -mext-string options.
22578 2018-01-12 Richard Biener <rguenther@suse.de>
22580 * lto-streamer-out.c (DFS::DFS_write_tree_body): Process
22581 DECL_DEBUG_EXPR conditional on DECL_HAS_DEBUG_EXPR_P.
22582 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
22584 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
22586 2018-01-11 Michael Meissner <meissner@linux.vnet.ibm.com>
22588 * configure.ac (--with-long-double-format): Add support for the
22589 configuration option to change the default long double format on
22591 * config.gcc (powerpc*-linux*-*): Likewise.
22592 * configure: Regenerate.
22593 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If long
22594 double is IEEE, define __KC__ and __KF__ to allow floatn.h to be
22595 used without modification.
22597 2018-01-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22599 * config/rs6000/rs6000-builtin.def (BU_P7_MISC_X): New #define.
22600 (SPEC_BARRIER): New instantiation of BU_P7_MISC_X.
22601 * config/rs6000/rs6000.c (rs6000_expand_builtin): Handle
22602 MISC_BUILTIN_SPEC_BARRIER.
22603 (rs6000_init_builtins): Likewise.
22604 * config/rs6000/rs6000.md (UNSPECV_SPEC_BARRIER): New UNSPECV
22606 (speculation_barrier): New define_insn.
22607 * doc/extend.texi: Document __builtin_speculation_barrier.
22609 2018-01-11 Jakub Jelinek <jakub@redhat.com>
22612 * config/i386/i386.c (ix86_expand_vector_init_one_nonzero): If one_var
22613 is 0, for V{8,16}S[IF] and V[48]D[IF]mode use gen_vec_set<mode>_0.
22614 * config/i386/sse.md (VI8_AVX_AVX512F, VI4F_256_512): New mode
22616 (ssescalarmodesuffix): Add 512-bit vectors. Use "d" or "q" for
22617 integral modes instead of "ss" and "sd".
22618 (vec_set<mode>_0): New define_insns for 256-bit and 512-bit
22619 vectors with 32-bit and 64-bit elements.
22620 (vecdupssescalarmodesuffix): New mode attribute.
22621 (vec_dup<mode>): Use it.
22623 2018-01-11 H.J. Lu <hongjiu.lu@intel.com>
22626 * config/i386/i386.c (ix86_compute_frame_layout): Align stack
22627 frame if argument is passed on stack.
22629 2018-01-11 Jakub Jelinek <jakub@redhat.com>
22632 * ree.c (combine_reaching_defs): Optimize also
22633 reg2=exp; reg1=reg2; reg2=any_extend(reg1); into
22634 reg2=any_extend(exp); reg1=reg2;, formatting fix.
22636 2018-01-11 Jan Hubicka <hubicka@ucw.cz>
22638 PR middle-end/83189
22639 * gimple-ssa-isolate-paths.c (isolate_path): Fix profile update.
22641 2018-01-11 Jan Hubicka <hubicka@ucw.cz>
22643 PR middle-end/83718
22644 * tree-inline.c (copy_cfg_body): Adjust num&den for scaling
22645 after they are computed.
22647 2018-01-11 Bin Cheng <bin.cheng@arm.com>
22649 PR tree-optimization/83695
22650 * gimple-loop-linterchange.cc
22651 (tree_loop_interchange::interchange_loops): Call scev_reset_htab to
22652 reset cached scev information after interchange.
22653 (pass_linterchange::execute): Remove call to scev_reset_htab.
22655 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22657 * config/arm/arm_neon.h (vfmlal_lane_low_u32, vfmlal_lane_high_u32,
22658 vfmlalq_laneq_low_u32, vfmlalq_lane_low_u32, vfmlal_laneq_low_u32,
22659 vfmlalq_laneq_high_u32, vfmlalq_lane_high_u32, vfmlal_laneq_high_u32,
22660 vfmlsl_lane_low_u32, vfmlsl_lane_high_u32, vfmlslq_laneq_low_u32,
22661 vfmlslq_lane_low_u32, vfmlsl_laneq_low_u32, vfmlslq_laneq_high_u32,
22662 vfmlslq_lane_high_u32, vfmlsl_laneq_high_u32): Define.
22663 * config/arm/arm_neon_builtins.def (vfmal_lane_low,
22664 vfmal_lane_lowv4hf, vfmal_lane_lowv8hf, vfmal_lane_high,
22665 vfmal_lane_highv4hf, vfmal_lane_highv8hf, vfmsl_lane_low,
22666 vfmsl_lane_lowv4hf, vfmsl_lane_lowv8hf, vfmsl_lane_high,
22667 vfmsl_lane_highv4hf, vfmsl_lane_highv8hf): New sets of builtins.
22668 * config/arm/iterators.md (VFMLSEL2, vfmlsel2): New mode attributes.
22669 (V_lane_reg): Likewise.
22670 * config/arm/neon.md (neon_vfm<vfml_op>l_lane_<vfml_half><VCVTF:mode>):
22672 (neon_vfm<vfml_op>l_lane_<vfml_half><vfmlsel2><mode>): Likewise.
22673 (vfmal_lane_low<mode>_intrinsic,
22674 vfmal_lane_low<vfmlsel2><mode>_intrinsic,
22675 vfmal_lane_high<vfmlsel2><mode>_intrinsic,
22676 vfmal_lane_high<mode>_intrinsic, vfmsl_lane_low<mode>_intrinsic,
22677 vfmsl_lane_low<vfmlsel2><mode>_intrinsic,
22678 vfmsl_lane_high<vfmlsel2><mode>_intrinsic,
22679 vfmsl_lane_high<mode>_intrinsic): New define_insns.
22681 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22683 * config/arm/arm-cpus.in (fp16fml): New feature.
22684 (ALL_SIMD): Add fp16fml.
22685 (armv8.2-a): Add fp16fml as an option.
22686 (armv8.3-a): Likewise.
22687 (armv8.4-a): Add fp16fml as part of fp16.
22688 * config/arm/arm.h (TARGET_FP16FML): Define.
22689 * config/arm/arm-c.c (arm_cpu_builtins): Define __ARM_FEATURE_FP16_FML
22691 * config/arm/arm-modes.def (V2HF): Define.
22692 * config/arm/arm_neon.h (vfmlal_low_u32, vfmlsl_low_u32,
22693 vfmlal_high_u32, vfmlsl_high_u32, vfmlalq_low_u32,
22694 vfmlslq_low_u32, vfmlalq_high_u32, vfmlslq_high_u32): Define.
22695 * config/arm/arm_neon_builtins.def (vfmal_low, vfmal_high,
22696 vfmsl_low, vfmsl_high): New set of builtins.
22697 * config/arm/iterators.md (PLUSMINUS): New code iterator.
22698 (vfml_op): New code attribute.
22699 (VFMLHALVES): New int iterator.
22700 (VFML, VFMLSEL): New mode attributes.
22701 (V_reg): Define mapping for V2HF.
22702 (V_hi, V_lo): New mode attributes.
22703 (VF_constraint): Likewise.
22704 (vfml_half, vfml_half_selector): New int attributes.
22705 * config/arm/neon.md (neon_vfm<vfml_op>l_<vfml_half><mode>): New
22707 (vfmal_low<mode>_intrinsic, vfmsl_high<mode>_intrinsic,
22708 vfmal_high<mode>_intrinsic, vfmsl_low<mode>_intrinsic):
22710 * config/arm/t-arm-elf (v8_fps): Add fp16fml.
22711 * config/arm/t-multilib (v8_2_a_simd_variants): Add fp16fml.
22712 * config/arm/unspecs.md (UNSPEC_VFML_LO, UNSPEC_VFML_HI): New unspecs.
22713 * doc/invoke.texi (ARM Options): Document fp16fml. Update armv8.4-a
22715 * doc/sourcebuild.texi (arm_fp16fml_neon_ok, arm_fp16fml_neon):
22716 Document new effective target and option set.
22718 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22720 * config/arm/arm-cpus.in (armv8_4): New feature.
22721 (ARMv8_4a): New fgroup.
22722 (armv8.4-a): New arch.
22723 * config/arm/arm-tables.opt: Regenerate.
22724 * config/arm/t-aprofile: Add matching rules for -march=armv8.4-a.
22725 * config/arm/t-arm-elf (all_v8_archs): Add armv8.4-a.
22726 * config/arm/t-multilib (v8_4_a_simd_variants): New variable.
22727 Add matching rules for -march=armv8.4-a and extensions.
22728 * doc/invoke.texi (ARM Options): Document -march=armv8.4-a.
22730 2018-01-11 Oleg Endo <olegendo@gcc.gnu.org>
22733 * config/rx/rx.md (BW): New mode attribute.
22734 (sync_lock_test_and_setsi): Add mode suffix to insn output.
22736 2018-01-11 Richard Biener <rguenther@suse.de>
22738 PR tree-optimization/83435
22739 * graphite.c (canonicalize_loop_form): Ignore fake loop exit edges.
22740 * graphite-scop-detection.c (scop_detection::get_sese): Likewise.
22741 * tree-vrp.c (add_assert_info): Drop TREE_OVERFLOW if they appear.
22743 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
22744 Alan Hayward <alan.hayward@arm.com>
22745 David Sherwood <david.sherwood@arm.com>
22747 * config/aarch64/aarch64.c (aarch64_address_info): Add a const_offset
22749 (aarch64_classify_address): Initialize it. Track polynomial offsets.
22750 (aarch64_print_address_internal): Use it to check for a zero offset.
22752 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
22753 Alan Hayward <alan.hayward@arm.com>
22754 David Sherwood <david.sherwood@arm.com>
22756 * config/aarch64/aarch64-modes.def (NUM_POLY_INT_COEFFS): Set to 2.
22757 * config/aarch64/aarch64-protos.h (aarch64_initial_elimination_offset):
22758 Return a poly_int64 rather than a HOST_WIDE_INT.
22759 (aarch64_offset_7bit_signed_scaled_p): Take the offset as a poly_int64
22760 rather than a HOST_WIDE_INT.
22761 * config/aarch64/aarch64.h (aarch64_frame): Protect with
22762 HAVE_POLY_INT_H rather than HOST_WIDE_INT. Change locals_offset,
22763 hard_fp_offset, frame_size, initial_adjust, callee_offset and
22764 final_offset from HOST_WIDE_INT to poly_int64.
22765 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args): Use
22766 to_constant when getting the number of units in an Advanced SIMD
22768 (aarch64_builtin_vectorized_function): Check for a constant number
22770 * config/aarch64/aarch64-simd.md (mov<mode>): Handle polynomial
22772 (aarch64_ld<VSTRUCT:nregs>_lane<VALLDIF:mode>): Use the nunits
22773 attribute instead of GET_MODE_NUNITS.
22774 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs)
22775 (aarch64_class_max_nregs): Use the constant_lowest_bound of the
22776 GET_MODE_SIZE for fixed-size registers.
22777 (aarch64_const_vec_all_same_in_range_p): Use const_vec_duplicate_p.
22778 (aarch64_hard_regno_call_part_clobbered, aarch64_classify_index)
22779 (aarch64_mode_valid_for_sched_fusion_p, aarch64_classify_address)
22780 (aarch64_legitimize_address_displacement, aarch64_secondary_reload)
22781 (aarch64_print_operand, aarch64_print_address_internal)
22782 (aarch64_address_cost, aarch64_rtx_costs, aarch64_register_move_cost)
22783 (aarch64_short_vector_p, aapcs_vfp_sub_candidate)
22784 (aarch64_simd_attr_length_rglist, aarch64_operands_ok_for_ldpstp):
22785 Handle polynomial GET_MODE_SIZE.
22786 (aarch64_hard_regno_caller_save_mode): Likewise. Return modes
22787 wider than SImode without modification.
22788 (tls_symbolic_operand_type): Use strip_offset instead of split_const.
22789 (aarch64_pass_by_reference, aarch64_layout_arg, aarch64_pad_reg_upward)
22790 (aarch64_gimplify_va_arg_expr): Assert that we don't yet handle
22791 passing and returning SVE modes.
22792 (aarch64_function_value, aarch64_layout_arg): Use gen_int_mode
22793 rather than GEN_INT.
22794 (aarch64_emit_probe_stack_range): Take the size as a poly_int64
22795 rather than a HOST_WIDE_INT, but call sorry if it isn't constant.
22796 (aarch64_allocate_and_probe_stack_space): Likewise.
22797 (aarch64_layout_frame): Cope with polynomial offsets.
22798 (aarch64_save_callee_saves, aarch64_restore_callee_saves): Take the
22799 start_offset as a poly_int64 rather than a HOST_WIDE_INT. Track
22800 polynomial offsets.
22801 (offset_9bit_signed_unscaled_p, offset_12bit_unsigned_scaled_p)
22802 (aarch64_offset_7bit_signed_scaled_p): Take the offset as a
22803 poly_int64 rather than a HOST_WIDE_INT.
22804 (aarch64_get_separate_components, aarch64_process_components)
22805 (aarch64_expand_prologue, aarch64_expand_epilogue)
22806 (aarch64_use_return_insn_p): Handle polynomial frame offsets.
22807 (aarch64_anchor_offset): New function, split out from...
22808 (aarch64_legitimize_address): ...here.
22809 (aarch64_builtin_vectorization_cost): Handle polynomial
22810 TYPE_VECTOR_SUBPARTS.
22811 (aarch64_simd_check_vect_par_cnst_half): Handle polynomial
22813 (aarch64_simd_make_constant, aarch64_expand_vector_init): Get the
22814 number of elements from the PARALLEL rather than the mode.
22815 (aarch64_shift_truncation_mask): Use GET_MODE_UNIT_BITSIZE
22816 rather than GET_MODE_BITSIZE.
22817 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_ext)
22818 (aarch64_evpc_rev, aarch64_evpc_dup, aarch64_evpc_zip)
22819 (aarch64_expand_vec_perm_const_1): Handle polynomial
22820 d->perm.length () and d->perm elements.
22821 (aarch64_evpc_tbl): Likewise. Use nelt rather than GET_MODE_NUNITS.
22822 Apply to_constant to d->perm elements.
22823 (aarch64_simd_valid_immediate, aarch64_vec_fpconst_pow_of_2): Handle
22824 polynomial CONST_VECTOR_NUNITS.
22825 (aarch64_move_pointer): Take amount as a poly_int64 rather
22827 (aarch64_progress_pointer): Avoid temporary variable.
22828 * config/aarch64/aarch64.md (aarch64_<crc_variant>): Use
22829 the mode attribute instead of GET_MODE.
22831 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
22832 Alan Hayward <alan.hayward@arm.com>
22833 David Sherwood <david.sherwood@arm.com>
22835 * config/aarch64/aarch64.c (aarch64_force_temporary): Assert that
22836 x exists before using it.
22837 (aarch64_add_constant_internal): Rename to...
22838 (aarch64_add_offset_1): ...this. Replace regnum with separate
22839 src and dest rtxes. Handle the case in which they're different,
22840 including when the offset is zero. Replace scratchreg with an rtx.
22841 Use 2 additions if there is no spare register into which we can
22842 move a 16-bit constant.
22843 (aarch64_add_constant): Delete.
22844 (aarch64_add_offset): Replace reg with separate src and dest
22845 rtxes. Take a poly_int64 offset instead of a HOST_WIDE_INT.
22846 Use aarch64_add_offset_1.
22847 (aarch64_add_sp, aarch64_sub_sp): Take the scratch register as
22848 an rtx rather than an int. Take the delta as a poly_int64
22849 rather than a HOST_WIDE_INT. Use aarch64_add_offset.
22850 (aarch64_expand_mov_immediate): Update uses of aarch64_add_offset.
22851 (aarch64_expand_prologue): Update calls to aarch64_sub_sp,
22852 aarch64_allocate_and_probe_stack_space and aarch64_add_offset.
22853 (aarch64_expand_epilogue): Update calls to aarch64_add_offset
22854 and aarch64_add_sp.
22855 (aarch64_output_mi_thunk): Use aarch64_add_offset rather than
22856 aarch64_add_constant.
22858 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
22860 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
22861 Use scalar_float_mode.
22863 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
22865 * config/aarch64/aarch64-simd.md
22866 (aarch64_fml<f16mac1>l<f16quad>_low<mode>): Avoid GET_MODE_NUNITS.
22867 (aarch64_fml<f16mac1>l<f16quad>_high<mode>): Likewise.
22868 (aarch64_fml<f16mac1>l_lane_lowv2sf): Likewise.
22869 (aarch64_fml<f16mac1>l_lane_highv2sf): Likewise.
22870 (aarch64_fml<f16mac1>lq_laneq_lowv4sf): Likewise.
22871 (aarch64_fml<f16mac1>lq_laneq_highv4sf): Likewise.
22872 (aarch64_fml<f16mac1>l_laneq_lowv2sf): Likewise.
22873 (aarch64_fml<f16mac1>l_laneq_highv2sf): Likewise.
22874 (aarch64_fml<f16mac1>lq_lane_lowv4sf): Likewise.
22875 (aarch64_fml<f16mac1>lq_lane_highv4sf): Likewise.
22877 2018-01-11 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
22880 * config/arm/arm.c (arm_declare_function_name): Set arch_to_print if
22881 targ_options->x_arm_arch_string is non NULL.
22883 2018-01-11 Tamar Christina <tamar.christina@arm.com>
22885 * config/aarch64/aarch64.h
22886 (AARCH64_FL_FOR_ARCH8_4): Add AARCH64_FL_DOTPROD.
22888 2018-01-11 Sudakshina Das <sudi.das@arm.com>
22891 * expmed.c (emit_store_flag_force): Swap if const op0
22892 and change VOIDmode to mode of op0.
22894 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
22896 PR rtl-optimization/83761
22897 * caller-save.c (replace_reg_with_saved_mem): Pass bits rather
22898 than bytes to mode_for_size.
22900 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
22902 PR middle-end/83189
22903 * gfortran.fortran-torture/compile/pr83189.f90: New testcase.
22904 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Handle zero
22907 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
22909 PR middle-end/83575
22910 * cfgrtl.c (rtl_verify_edges): Only verify fixability of partition
22911 when in layout mode.
22912 (cfg_layout_finalize): Do not verify cfg before we are out of layout.
22913 * cfgcleanup.c (try_optimize_cfg): Only verify flow info when doing
22916 2018-01-10 Michael Collison <michael.collison@arm.com>
22918 * config/aarch64/aarch64-modes.def (V2HF): New VECTOR_MODE.
22919 * config/aarch64/aarch64-option-extension.def: Add
22920 AARCH64_OPT_EXTENSION of 'fp16fml'.
22921 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
22922 (__ARM_FEATURE_FP16_FML): Define if TARGET_F16FML is true.
22923 * config/aarch64/predicates.md (aarch64_lane_imm3): New predicate.
22924 * config/aarch64/constraints.md (Ui7): New constraint.
22925 * config/aarch64/iterators.md (VFMLA_W): New mode iterator.
22926 (VFMLA_SEL_W): Ditto.
22929 (VFMLA16_LOW): New int iterator.
22930 (VFMLA16_HIGH): Ditto.
22931 (UNSPEC_FMLAL): New unspec.
22932 (UNSPEC_FMLSL): Ditto.
22933 (UNSPEC_FMLAL2): Ditto.
22934 (UNSPEC_FMLSL2): Ditto.
22935 (f16mac): New code attribute.
22936 * config/aarch64/aarch64-simd-builtins.def
22937 (aarch64_fmlal_lowv2sf): Ditto.
22938 (aarch64_fmlsl_lowv2sf): Ditto.
22939 (aarch64_fmlalq_lowv4sf): Ditto.
22940 (aarch64_fmlslq_lowv4sf): Ditto.
22941 (aarch64_fmlal_highv2sf): Ditto.
22942 (aarch64_fmlsl_highv2sf): Ditto.
22943 (aarch64_fmlalq_highv4sf): Ditto.
22944 (aarch64_fmlslq_highv4sf): Ditto.
22945 (aarch64_fmlal_lane_lowv2sf): Ditto.
22946 (aarch64_fmlsl_lane_lowv2sf): Ditto.
22947 (aarch64_fmlal_laneq_lowv2sf): Ditto.
22948 (aarch64_fmlsl_laneq_lowv2sf): Ditto.
22949 (aarch64_fmlalq_lane_lowv4sf): Ditto.
22950 (aarch64_fmlsl_lane_lowv4sf): Ditto.
22951 (aarch64_fmlalq_laneq_lowv4sf): Ditto.
22952 (aarch64_fmlsl_laneq_lowv4sf): Ditto.
22953 (aarch64_fmlal_lane_highv2sf): Ditto.
22954 (aarch64_fmlsl_lane_highv2sf): Ditto.
22955 (aarch64_fmlal_laneq_highv2sf): Ditto.
22956 (aarch64_fmlsl_laneq_highv2sf): Ditto.
22957 (aarch64_fmlalq_lane_highv4sf): Ditto.
22958 (aarch64_fmlsl_lane_highv4sf): Ditto.
22959 (aarch64_fmlalq_laneq_highv4sf): Ditto.
22960 (aarch64_fmlsl_laneq_highv4sf): Ditto.
22961 * config/aarch64/aarch64-simd.md:
22962 (aarch64_fml<f16mac1>l<f16quad>_low<mode>): New pattern.
22963 (aarch64_fml<f16mac1>l<f16quad>_high<mode>): Ditto.
22964 (aarch64_simd_fml<f16mac1>l<f16quad>_low<mode>): Ditto.
22965 (aarch64_simd_fml<f16mac1>l<f16quad>_high<mode>): Ditto.
22966 (aarch64_fml<f16mac1>l_lane_lowv2sf): Ditto.
22967 (aarch64_fml<f16mac1>l_lane_highv2sf): Ditto.
22968 (aarch64_simd_fml<f16mac>l_lane_lowv2sf): Ditto.
22969 (aarch64_simd_fml<f16mac>l_lane_highv2sf): Ditto.
22970 (aarch64_fml<f16mac1>lq_laneq_lowv4sf): Ditto.
22971 (aarch64_fml<f16mac1>lq_laneq_highv4sf): Ditto.
22972 (aarch64_simd_fml<f16mac>lq_laneq_lowv4sf): Ditto.
22973 (aarch64_simd_fml<f16mac>lq_laneq_highv4sf): Ditto.
22974 (aarch64_fml<f16mac1>l_laneq_lowv2sf): Ditto.
22975 (aarch64_fml<f16mac1>l_laneq_highv2sf): Ditto.
22976 (aarch64_simd_fml<f16mac>l_laneq_lowv2sf): Ditto.
22977 (aarch64_simd_fml<f16mac>l_laneq_highv2sf): Ditto.
22978 (aarch64_fml<f16mac1>lq_lane_lowv4sf): Ditto.
22979 (aarch64_fml<f16mac1>lq_lane_highv4sf): Ditto.
22980 (aarch64_simd_fml<f16mac>lq_lane_lowv4sf): Ditto.
22981 (aarch64_simd_fml<f16mac>lq_lane_highv4sf): Ditto.
22982 * config/aarch64/arm_neon.h (vfmlal_low_u32): New intrinsic.
22983 (vfmlsl_low_u32): Ditto.
22984 (vfmlalq_low_u32): Ditto.
22985 (vfmlslq_low_u32): Ditto.
22986 (vfmlal_high_u32): Ditto.
22987 (vfmlsl_high_u32): Ditto.
22988 (vfmlalq_high_u32): Ditto.
22989 (vfmlslq_high_u32): Ditto.
22990 (vfmlal_lane_low_u32): Ditto.
22991 (vfmlsl_lane_low_u32): Ditto.
22992 (vfmlal_laneq_low_u32): Ditto.
22993 (vfmlsl_laneq_low_u32): Ditto.
22994 (vfmlalq_lane_low_u32): Ditto.
22995 (vfmlslq_lane_low_u32): Ditto.
22996 (vfmlalq_laneq_low_u32): Ditto.
22997 (vfmlslq_laneq_low_u32): Ditto.
22998 (vfmlal_lane_high_u32): Ditto.
22999 (vfmlsl_lane_high_u32): Ditto.
23000 (vfmlal_laneq_high_u32): Ditto.
23001 (vfmlsl_laneq_high_u32): Ditto.
23002 (vfmlalq_lane_high_u32): Ditto.
23003 (vfmlslq_lane_high_u32): Ditto.
23004 (vfmlalq_laneq_high_u32): Ditto.
23005 (vfmlslq_laneq_high_u32): Ditto.
23006 * config/aarch64/aarch64.h (AARCH64_FL_F16SML): New flag.
23007 (AARCH64_FL_FOR_ARCH8_4): New.
23008 (AARCH64_ISA_F16FML): New ISA flag.
23009 (TARGET_F16FML): New feature flag for fp16fml.
23010 (doc/invoke.texi): Document new fp16fml option.
23012 2018-01-10 Michael Collison <michael.collison@arm.com>
23014 * config/aarch64/aarch64-builtins.c:
23015 (aarch64_types_ternopu_imm_qualifiers, TYPES_TERNOPUI): New.
23016 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
23017 (__ARM_FEATURE_SHA3): Define if TARGET_SHA3 is true.
23018 * config/aarch64/aarch64.h (AARCH64_FL_SHA3): New flags.
23019 (AARCH64_ISA_SHA3): New ISA flag.
23020 (TARGET_SHA3): New feature flag for sha3.
23021 * config/aarch64/iterators.md (sha512_op): New int attribute.
23022 (CRYPTO_SHA512): New int iterator.
23023 (UNSPEC_SHA512H): New unspec.
23024 (UNSPEC_SHA512H2): Ditto.
23025 (UNSPEC_SHA512SU0): Ditto.
23026 (UNSPEC_SHA512SU1): Ditto.
23027 * config/aarch64/aarch64-simd-builtins.def
23028 (aarch64_crypto_sha512hqv2di): New builtin.
23029 (aarch64_crypto_sha512h2qv2di): Ditto.
23030 (aarch64_crypto_sha512su0qv2di): Ditto.
23031 (aarch64_crypto_sha512su1qv2di): Ditto.
23032 (aarch64_eor3qv8hi): Ditto.
23033 (aarch64_rax1qv2di): Ditto.
23034 (aarch64_xarqv2di): Ditto.
23035 (aarch64_bcaxqv8hi): Ditto.
23036 * config/aarch64/aarch64-simd.md:
23037 (aarch64_crypto_sha512h<sha512_op>qv2di): New pattern.
23038 (aarch64_crypto_sha512su0qv2di): Ditto.
23039 (aarch64_crypto_sha512su1qv2di): Ditto.
23040 (aarch64_eor3qv8hi): Ditto.
23041 (aarch64_rax1qv2di): Ditto.
23042 (aarch64_xarqv2di): Ditto.
23043 (aarch64_bcaxqv8hi): Ditto.
23044 * config/aarch64/arm_neon.h (vsha512hq_u64): New intrinsic.
23045 (vsha512h2q_u64): Ditto.
23046 (vsha512su0q_u64): Ditto.
23047 (vsha512su1q_u64): Ditto.
23048 (veor3q_u16): Ditto.
23049 (vrax1q_u64): Ditto.
23050 (vxarq_u64): Ditto.
23051 (vbcaxq_u16): Ditto.
23052 * config/arm/types.md (crypto_sha512): New type attribute.
23053 (crypto_sha3): Ditto.
23054 (doc/invoke.texi): Document new sha3 option.
23056 2018-01-10 Michael Collison <michael.collison@arm.com>
23058 * config/aarch64/aarch64-builtins.c:
23059 (aarch64_types_quadopu_imm_qualifiers, TYPES_QUADOPUI): New.
23060 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
23061 (__ARM_FEATURE_SM3): Define if TARGET_SM4 is true.
23062 (__ARM_FEATURE_SM4): Define if TARGET_SM4 is true.
23063 * config/aarch64/aarch64.h (AARCH64_FL_SM4): New flags.
23064 (AARCH64_ISA_SM4): New ISA flag.
23065 (TARGET_SM4): New feature flag for sm4.
23066 * config/aarch64/aarch64-simd-builtins.def
23067 (aarch64_sm3ss1qv4si): Ditto.
23068 (aarch64_sm3tt1aq4si): Ditto.
23069 (aarch64_sm3tt1bq4si): Ditto.
23070 (aarch64_sm3tt2aq4si): Ditto.
23071 (aarch64_sm3tt2bq4si): Ditto.
23072 (aarch64_sm3partw1qv4si): Ditto.
23073 (aarch64_sm3partw2qv4si): Ditto.
23074 (aarch64_sm4eqv4si): Ditto.
23075 (aarch64_sm4ekeyqv4si): Ditto.
23076 * config/aarch64/aarch64-simd.md:
23077 (aarch64_sm3ss1qv4si): Ditto.
23078 (aarch64_sm3tt<sm3tt_op>qv4si): Ditto.
23079 (aarch64_sm3partw<sm3part_op>qv4si): Ditto.
23080 (aarch64_sm4eqv4si): Ditto.
23081 (aarch64_sm4ekeyqv4si): Ditto.
23082 * config/aarch64/iterators.md (sm3tt_op): New int iterator.
23083 (sm3part_op): Ditto.
23084 (CRYPTO_SM3TT): Ditto.
23085 (CRYPTO_SM3PART): Ditto.
23086 (UNSPEC_SM3SS1): New unspec.
23087 (UNSPEC_SM3TT1A): Ditto.
23088 (UNSPEC_SM3TT1B): Ditto.
23089 (UNSPEC_SM3TT2A): Ditto.
23090 (UNSPEC_SM3TT2B): Ditto.
23091 (UNSPEC_SM3PARTW1): Ditto.
23092 (UNSPEC_SM3PARTW2): Ditto.
23093 (UNSPEC_SM4E): Ditto.
23094 (UNSPEC_SM4EKEY): Ditto.
23095 * config/aarch64/constraints.md (Ui2): New constraint.
23096 * config/aarch64/predicates.md (aarch64_imm2): New predicate.
23097 * config/arm/types.md (crypto_sm3): New type attribute.
23098 (crypto_sm4): Ditto.
23099 * config/aarch64/arm_neon.h (vsm3ss1q_u32): New intrinsic.
23100 (vsm3tt1aq_u32): Ditto.
23101 (vsm3tt1bq_u32): Ditto.
23102 (vsm3tt2aq_u32): Ditto.
23103 (vsm3tt2bq_u32): Ditto.
23104 (vsm3partw1q_u32): Ditto.
23105 (vsm3partw2q_u32): Ditto.
23106 (vsm4eq_u32): Ditto.
23107 (vsm4ekeyq_u32): Ditto.
23108 (doc/invoke.texi): Document new sm4 option.
23110 2018-01-10 Michael Collison <michael.collison@arm.com>
23112 * config/aarch64/aarch64-arches.def (armv8.4-a): New architecture.
23113 * config/aarch64/aarch64.h (AARCH64_ISA_V8_4): New ISA flag.
23114 (AARCH64_FL_FOR_ARCH8_4): New.
23115 (AARCH64_FL_V8_4): New flag.
23116 (doc/invoke.texi): Document new armv8.4-a option.
23118 2018-01-10 Michael Collison <michael.collison@arm.com>
23120 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
23121 (__ARM_FEATURE_AES): Define if TARGET_AES is true.
23122 (__ARM_FEATURE_SHA2): Define if TARGET_SHA2 is true.
23123 * config/aarch64/aarch64-option-extension.def: Add
23124 AARCH64_OPT_EXTENSION of 'sha2'.
23125 (aes): Add AARCH64_OPT_EXTENSION of 'aes'.
23126 (crypto): Disable sha2 and aes if crypto disabled.
23127 (crypto): Enable aes and sha2 if enabled.
23128 (simd): Disable sha2 and aes if simd disabled.
23129 * config/aarch64/aarch64.h (AARCH64_FL_AES, AARCH64_FL_SHA2):
23131 (AARCH64_ISA_AES, AARCH64_ISA_SHA2): New ISA flags.
23132 (TARGET_SHA2): New feature flag for sha2.
23133 (TARGET_AES): New feature flag for aes.
23134 * config/aarch64/aarch64-simd.md:
23135 (aarch64_crypto_aes<aes_op>v16qi): Make pattern
23136 conditional on TARGET_AES.
23137 (aarch64_crypto_aes<aesmc_op>v16qi): Ditto.
23138 (aarch64_crypto_sha1hsi): Make pattern conditional
23140 (aarch64_crypto_sha1hv4si): Ditto.
23141 (aarch64_be_crypto_sha1hv4si): Ditto.
23142 (aarch64_crypto_sha1su1v4si): Ditto.
23143 (aarch64_crypto_sha1<sha1_op>v4si): Ditto.
23144 (aarch64_crypto_sha1su0v4si): Ditto.
23145 (aarch64_crypto_sha256h<sha256_op>v4si): Ditto.
23146 (aarch64_crypto_sha256su0v4si): Ditto.
23147 (aarch64_crypto_sha256su1v4si): Ditto.
23148 (doc/invoke.texi): Document new aes and sha2 options.
23150 2018-01-10 Martin Sebor <msebor@redhat.com>
23152 PR tree-optimization/83781
23153 * gimple-fold.c (get_range_strlen): Avoid treating arrays of pointers
23156 2018-01-11 Martin Sebor <msebor@gmail.com>
23157 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
23159 PR tree-optimization/83501
23160 PR tree-optimization/81703
23162 * tree-ssa-strlen.c (get_string_cst): Rename...
23163 (get_string_len): ...to this. Handle global constants.
23164 (handle_char_store): Adjust.
23166 2018-01-10 Kito Cheng <kito.cheng@gmail.com>
23167 Jim Wilson <jimw@sifive.com>
23169 * config/riscv/riscv-protos.h (riscv_output_return): New.
23170 * config/riscv/riscv.c (struct machine_function): New naked_p field.
23171 (riscv_attribute_table, riscv_output_return),
23172 (riscv_handle_fndecl_attribute, riscv_naked_function_p),
23173 (riscv_allocate_stack_slots_for_args, riscv_warn_func_return): New.
23174 (riscv_compute_frame_info): Only compute frame->mask if not a naked
23176 (riscv_expand_prologue): Add early return for naked function.
23177 (riscv_expand_epilogue): Likewise.
23178 (riscv_function_ok_for_sibcall): Return false for naked function.
23179 (riscv_set_current_function): New.
23180 (TARGET_SET_CURRENT_FUNCTION, TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS),
23181 (TARGET_ATTRIBUTE_TABLE, TARGET_WARN_FUNC_RETURN): New.
23182 * config/riscv/riscv.md (simple_return): Call riscv_output_return.
23183 * doc/extend.texi (RISC-V Function Attributes): New.
23185 2018-01-10 Michael Meissner <meissner@linux.vnet.ibm.com>
23187 * config/rs6000/rs6000.c (is_complex_IBM_long_double): Explicitly
23188 check for 128-bit long double before checking TCmode.
23189 * config/rs6000/rs6000.h (FLOAT128_IEEE_P): Explicitly check for
23190 128-bit long doubles before checking TFmode or TCmode.
23191 (FLOAT128_IBM_P): Likewise.
23193 2018-01-10 Martin Sebor <msebor@redhat.com>
23195 PR tree-optimization/83671
23196 * builtins.c (c_strlen): Unconditionally return zero for the empty
23198 Use -Warray-bounds for warnings.
23199 * gimple-fold.c (get_range_strlen): Handle non-constant lengths
23200 for non-constant array indices with COMPONENT_REF, arrays of
23201 arrays, and pointers to arrays.
23202 (gimple_fold_builtin_strlen): Determine and set length range for
23203 non-constant character arrays.
23205 2018-01-10 Aldy Hernandez <aldyh@redhat.com>
23207 PR middle-end/81897
23208 * tree-ssa-uninit.c (convert_control_dep_chain_into_preds): Skip
23211 2018-01-10 Eric Botcazou <ebotcazou@adacore.com>
23213 * dwarf2out.c (dwarf2out_var_location): Do not pass NULL to fprintf.
23215 2018-01-10 Peter Bergner <bergner@vnet.ibm.com>
23218 * config/rs6000/rs6000.c (print_operand) <'y'>: Use
23219 VECTOR_MEM_ALTIVEC_OR_VSX_P.
23220 * config/rs6000/vsx.md (*vsx_le_perm_load_<mode> for VSX_D): Use
23221 indexed_or_indirect_operand predicate.
23222 (*vsx_le_perm_load_<mode> for VSX_W): Likewise.
23223 (*vsx_le_perm_load_v8hi): Likewise.
23224 (*vsx_le_perm_load_v16qi): Likewise.
23225 (*vsx_le_perm_store_<mode> for VSX_D): Likewise.
23226 (*vsx_le_perm_store_<mode> for VSX_W): Likewise.
23227 (*vsx_le_perm_store_v8hi): Likewise.
23228 (*vsx_le_perm_store_v16qi): Likewise.
23229 (eight unnamed splitters): Likewise.
23231 2018-01-10 Peter Bergner <bergner@vnet.ibm.com>
23233 * config/rs6000/x86intrin.h: Change #warning to #error. Update message.
23234 * config/rs6000/emmintrin.h: Likewise.
23235 * config/rs6000/mmintrin.h: Likewise.
23236 * config/rs6000/xmmintrin.h: Likewise.
23238 2018-01-10 David Malcolm <dmalcolm@redhat.com>
23241 * tree-core.h: Document EXPR_LOCATION_WRAPPER_P's usage of
23243 * tree.c (tree_nop_conversion): Return true for location wrapper
23245 (maybe_wrap_with_location): New function.
23246 (selftest::check_strip_nops): New function.
23247 (selftest::test_location_wrappers): New function.
23248 (selftest::tree_c_tests): Call it.
23249 * tree.h (STRIP_ANY_LOCATION_WRAPPER): New macro.
23250 (maybe_wrap_with_location): New decl.
23251 (EXPR_LOCATION_WRAPPER_P): New macro.
23252 (location_wrapper_p): New inline function.
23253 (tree_strip_any_location_wrapper): New inline function.
23255 2018-01-10 H.J. Lu <hongjiu.lu@intel.com>
23258 * config/i386/i386.c (ix86_compute_frame_layout): Always adjust
23259 stack_realign_offset for the largest alignment of stack slot
23261 (ix86_find_max_used_stack_alignment): New function.
23262 (ix86_finalize_stack_frame_flags): Use it. Set
23263 max_used_stack_alignment if we don't realign stack.
23264 * config/i386/i386.h (machine_function): Add
23265 max_used_stack_alignment.
23267 2018-01-10 Christophe Lyon <christophe.lyon@linaro.org>
23269 * config/arm/arm.opt (-mbranch-cost): New option.
23270 * config/arm/arm.h (BRANCH_COST): Take arm_branch_cost into
23273 2018-01-10 Segher Boessenkool <segher@kernel.crashing.org>
23276 * config/rs6000/rs6000.md (load_toc_v4_PIC_2, load_toc_v4_PIC_3b,
23277 load_toc_v4_PIC_3c): Wrap const term in CONST RTL.
23279 2018-01-10 Richard Biener <rguenther@suse.de>
23282 * dwarf2out.c (gen_subprogram_die): Hoist old_die && declaration
23283 early out so it also covers the case where we have a non-NULL
23286 2018-01-10 Richard Sandiford <richard.sandiford@linaro.org>
23288 PR tree-optimization/83753
23289 * tree-vect-stmts.c (get_group_load_store_type): Use VMAT_CONTIGUOUS
23290 for non-strided grouped accesses if the number of elements is 1.
23292 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
23295 * i386.c (ix86_vectorize_builtin_gather): Check TARGET_USE_GATHER.
23296 * i386.h (TARGET_USE_GATHER): Define.
23297 * x86-tune.def (X86_TUNE_USE_GATHER): New.
23299 2018-01-10 Martin Liska <mliska@suse.cz>
23302 * basic-block.h (CLEANUP_NO_PARTITIONING): New define.
23303 * bb-reorder.c (pass_reorder_blocks::execute): Do not clean up
23305 * cfgcleanup.c (try_optimize_cfg): Fix up partitioning if
23306 CLEANUP_NO_PARTITIONING is not set.
23308 2018-01-10 Richard Sandiford <richard.sandiford@linaro.org>
23310 * doc/rtl.texi: Remove documentation of (const ...) wrappers
23311 for vectors, as a partial revert of r254296.
23312 * rtl.h (const_vec_p): Delete.
23313 (const_vec_duplicate_p): Don't test for vector CONSTs.
23314 (unwrap_const_vec_duplicate, const_vec_series_p): Likewise.
23315 * expmed.c (make_tree): Likewise.
23318 * common.md (E, F): Use CONSTANT_P instead of checking for
23320 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
23321 checking for CONST_VECTOR.
23323 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
23325 PR middle-end/83575
23326 * predict.c (force_edge_cold): Handle in more sane way edges
23327 with no prediction.
23329 2018-01-09 Carl Love <cel@us.ibm.com>
23331 * config/rs6002/altivec.md (p8_vmrgow): Add support for V2DI, V2DF,
23333 (p8_vmrgew): Add support for V2DI, V2DF, V4SF types.
23334 * config/rs6000/rs6000-builtin.def: Add definitions for FLOAT2_V2DF,
23335 VMRGEW_V2DI, VMRGEW_V2DF, VMRGEW_V4SF, VMRGOW_V4SI, VMRGOW_V4SF,
23336 VMRGOW_V2DI, VMRGOW_V2DF. Remove definition for VMRGOW.
23337 * config/rs6000/rs6000-c.c (VSX_BUILTIN_VEC_FLOAT2,
23338 P8V_BUILTIN_VEC_VMRGEW, P8V_BUILTIN_VEC_VMRGOW): Add definitions.
23339 * config/rs6000/rs6000-protos.h: Add extern defition for
23340 rs6000_generate_float2_double_code.
23341 * config/rs6000/rs6000.c (rs6000_generate_float2_double_code): Add
23343 * config/rs6000/vsx.md (vsx_xvcdpsp): Add define_insn.
23344 (float2_v2df): Add define_expand.
23346 2018-01-09 Uros Bizjak <ubizjak@gmail.com>
23349 * combine.c (force_int_to_mode) <case ASHIFT>: Use mode instead of
23350 op_mode in the force_to_mode call.
23352 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
23354 * config/aarch64/aarch64.c (aarch64_evpc_trn): Use d.perm.series_p
23355 instead of checking each element individually.
23356 (aarch64_evpc_uzp): Likewise.
23357 (aarch64_evpc_zip): Likewise.
23358 (aarch64_evpc_ext): Likewise.
23359 (aarch64_evpc_rev): Likewise.
23360 (aarch64_evpc_dup): Test the encoding for a single duplicated element,
23361 instead of checking each element individually. Return true without
23363 (aarch64_vectorize_vec_perm_const): Use all_from_input_p to test
23364 whether all selected elements come from the same input, instead of
23365 checking each element individually. Remove calls to gen_rtx_REG,
23366 start_sequence and end_sequence and instead assert that no rtl is
23369 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
23371 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p): Fix
23372 order of HIGH and CONST checks.
23374 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
23376 * tree-vect-stmts.c (permute_vec_elements): Create a fresh variable
23377 if the destination isn't an SSA_NAME.
23379 2018-01-09 Richard Biener <rguenther@suse.de>
23381 PR tree-optimization/83668
23382 * graphite.c (canonicalize_loop_closed_ssa): Add edge argument,
23384 (canonicalize_loop_form): ... here, renamed from ...
23385 (canonicalize_loop_closed_ssa_form): ... this and amended to
23386 swap successor edges for loop exit blocks to make us use
23387 the RPO order we need for initial schedule generation.
23389 2018-01-09 Joseph Myers <joseph@codesourcery.com>
23391 PR tree-optimization/64811
23392 * match.pd: When optimizing comparisons with Inf, avoid
23393 introducing or losing exceptions from comparisons with NaN.
23395 2018-01-09 Martin Liska <mliska@suse.cz>
23398 * asan.c (shadow_mem_size): Add gcc_assert.
23400 2018-01-09 Georg-Johann Lay <avr@gjlay.de>
23402 Don't save registers in main().
23405 * doc/invoke.texi (AVR Options) [-mmain-is-OS_task]: Document it.
23406 * config/avr/avr.opt (-mmain-is-OS_task): New target option.
23407 * config/avr/avr.c (avr_set_current_function): Don't error if
23408 naked, OS_task or OS_main are specified at the same time.
23409 (avr_function_ok_for_sibcall): Don't disable sibcalls for OS_task,
23411 (avr_insert_attributes) [-mmain-is-OS_task] <main>: Add OS_task
23413 * common/config/avr/avr-common.c (avr_option_optimization_table):
23414 Switch on -mmain-is-OS_task for optimizing compilations.
23416 2018-01-09 Richard Biener <rguenther@suse.de>
23418 PR tree-optimization/83572
23419 * graphite.c: Include cfganal.h.
23420 (graphite_transform_loops): Connect infinite loops to exit
23421 and remove fake edges at the end.
23423 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
23425 * ipa-inline.c (edge_badness): Revert accidental checkin.
23427 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
23430 * ipa-comdats.c (set_comdat_group): Only set comdat group of real
23431 symbols; not inline clones.
23433 2018-01-09 Jakub Jelinek <jakub@redhat.com>
23436 * modulo-sched.c (schedule_reg_moves): Punt if we'd need to move
23437 hard registers. Formatting fixes.
23439 PR preprocessor/83722
23440 * gcc.c (try_generate_repro): Pass
23441 &temp_stderr_files[RETRY_ICE_ATTEMPTS - 1] rather than
23442 &temp_stdout_files[RETRY_ICE_ATTEMPTS - 1] as last argument to
23445 2018-01-08 Monk Chiang <sh.chiang04@gmail.com>
23446 Kito Cheng <kito.cheng@gmail.com>
23448 * config/riscv/riscv.c (machine_function::is_leaf): Remove field.
23449 (riscv_leaf_function_p): Delete.
23450 (riscv_function_ok_for_sibcall): Return false when TARGET_SAVE_RESTORE.
23452 2018-01-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23454 * config/rs6000/rs6000-string.c (do_load_for_compare_from_addr): New
23456 (do_ifelse): New function.
23457 (do_isel): New function.
23458 (do_sub3): New function.
23459 (do_add3): New function.
23460 (do_load_mask_compare): New function.
23461 (do_overlap_load_compare): New function.
23462 (expand_compare_loop): New function.
23463 (expand_block_compare): Call expand_compare_loop() when appropriate.
23464 * config/rs6000/rs6000.opt (-mblock-compare-inline-limit): Change
23465 option description.
23466 (-mblock-compare-inline-loop-limit): New option.
23468 2018-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23471 * config/rs6000/altivec.md (*altivec_vpermr_<mode>_internal):
23472 Reverse order of second and third operands in first alternative.
23473 * config/rs6000/rs6000.c (rs6000_expand_vector_set): Reverse order
23474 of first and second elements in UNSPEC_VPERMR vector.
23475 (altivec_expand_vec_perm_le): Likewise.
23477 2018-01-08 Jeff Law <law@redhat.com>
23479 PR rtl-optimizatin/81308
23480 * tree-switch-conversion.c (cfg_altered): New file scoped static.
23481 (process_switch): If group_case_labels makes a change, then set
23483 (pass_convert_switch::execute): If a switch is converted, then
23484 set cfg_altered. Return TODO_cfg_cleanup if cfg_altered is true.
23486 PR rtl-optimization/81308
23487 * recog.c (split_all_insns): Conditionally cleanup the CFG after
23490 2018-01-08 Vidya Praveen <vidyapraveen@arm.com>
23492 PR target/83663 - Revert r255946
23493 * config/aarch64/aarch64.c (aarch64_expand_vector_init): Modify code
23494 generation for cases where splatting a value is not useful.
23495 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
23496 across a vec_duplicate and a paradoxical subreg forming a vector
23497 mode to a vec_concat.
23499 2018-01-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23501 * config/arm/t-aprofile (MULTILIB_MATCHES): Add mapping rules for
23502 -march=armv8.3-a variants.
23503 * config/arm/t-multilib: Likewise.
23504 * config/arm/t-arm-elf: Likewise. Handle dotprod extension.
23506 2018-01-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23508 * config/rs6000/rs6000.md (cceq_ior_compare): Remove * so I can use it
23510 (cceq_ior_compare_complement): Give it a name so I can use it, and
23511 change boolean_or_operator predicate to boolean_operator so it can
23512 be used to generate a crand.
23513 (eqne): New code iterator.
23514 (bd/bd_neg): New code_attrs.
23515 (<bd>_<mode>): New name for ctr<mode>_internal[12] now combined into
23516 a single define_insn.
23517 (<bd>tf_<mode>): A new insn pattern for the conditional form branch
23518 decrement (bdnzt/bdnzf/bdzt/bdzf).
23519 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Updated
23520 with the new names of the branch decrement patterns, and added the
23521 names of the branch decrement conditional patterns.
23523 2018-01-08 Richard Biener <rguenther@suse.de>
23525 PR tree-optimization/83563
23526 * graphite.c (canonicalize_loop_closed_ssa_form): Reset the SCEV
23529 2018-01-08 Richard Biener <rguenther@suse.de>
23531 PR middle-end/83713
23532 * convert.c (do_narrow): Properly guard TYPE_OVERFLOW_WRAPS checks.
23534 2018-01-08 Richard Biener <rguenther@suse.de>
23536 PR tree-optimization/83685
23537 * tree-ssa-pre.c (create_expression_by_pieces): Do not insert
23538 references to abnormals.
23540 2018-01-08 Richard Biener <rguenther@suse.de>
23543 * dwarf2out.c (output_indirect_strings): Handle empty
23544 skeleton_debug_str_hash.
23545 (dwarf2out_early_finish): Index strings for -gsplit-dwarf.
23547 2018-01-08 Claudiu Zissulescu <claziss@synopsys.com>
23549 * config/arc/arc.c (TARGET_TRAMPOLINE_ADJUST_ADDRESS): Delete.
23550 (emit_store_direct): Likewise.
23551 (arc_trampoline_adjust_address): Likewise.
23552 (arc_asm_trampoline_template): New function.
23553 (arc_initialize_trampoline): Use asm_trampoline_template.
23554 (TARGET_ASM_TRAMPOLINE_TEMPLATE): Define.
23555 * config/arc/arc.h (TRAMPOLINE_SIZE): Adjust to 16.
23556 * config/arc/arc.md (flush_icache): Delete pattern.
23558 2018-01-08 Claudiu Zissulescu <claziss@synopsys.com>
23560 * config/arc/arc-c.def (__ARC_UNALIGNED__): New define.
23561 * config/arc/arc.h (STRICT_ALIGNMENT): Control this macro using
23564 2018-01-08 Sebastian Huber <sebastian.huber@embedded-brains.de>
23567 * config/epiphany/epiphany.h (make_pass_mode_switch_use): Guard
23568 by not USED_FOR_TARGET.
23569 (make_pass_resolve_sw_modes): Likewise.
23571 2018-01-08 Sebastian Huber <sebastian.huber@embedded-brains.de>
23573 * config/nios2/nios2.h (nios2_section_threshold): Guard by not
23576 2018-01-08 Richard Biener <rguenther@suse.de>
23578 PR middle-end/83580
23579 * tree-data-ref.c (split_constant_offset): Remove STRIP_NOPS.
23581 2018-01-08 Richard Biener <rguenther@suse.de>
23583 PR middle-end/83517
23584 * match.pd ((t * 2) / 2) -> t): Add missing :c.
23586 2018-01-06 Aldy Hernandez <aldyh@redhat.com>
23588 PR middle-end/81897
23589 * tree-ssa-uninit.c (compute_control_dep_chain): Do not bail on
23590 basic blocks with a small number of successors.
23591 (convert_control_dep_chain_into_preds): Improve handling of
23593 (dump_predicates): Split apart into...
23594 (dump_pred_chain): ...here...
23595 (dump_pred_info): ...and here.
23596 (can_one_predicate_be_invalidated_p): Add debugging printfs.
23597 (can_chain_union_be_invalidated_p): Improve check for invalidation
23599 (uninit_uses_cannot_happen): Avoid unnecessary if
23600 convert_control_dep_chain_into_preds yielded nothing.
23602 2018-01-06 Martin Sebor <msebor@redhat.com>
23604 PR tree-optimization/83640
23605 * gimple-ssa-warn-restrict.c (builtin_access::builtin_access): Avoid
23606 subtracting negative offset from size.
23607 (builtin_access::overlap): Adjust offset bounds of the access to fall
23608 within the size of the object if possible.
23610 2018-01-06 Richard Sandiford <richard.sandiford@linaro.org>
23612 PR rtl-optimization/83699
23613 * expmed.c (extract_bit_field_1): Restrict the vector usage of
23614 extract_bit_field_as_subreg to cases in which the extracted
23615 value is also a vector.
23617 * lra-constraints.c (process_alt_operands): Test for the equivalence
23618 substitutions when detecting a possible reload cycle.
23620 2018-01-06 Jakub Jelinek <jakub@redhat.com>
23623 * toplev.c (process_options): Don't enable debug_nonbind_markers_p
23624 by default if flag_selective_schedling{,2}. Formatting fixes.
23626 PR rtl-optimization/83682
23627 * rtl.h (const_vec_duplicate_p): Only return true for VEC_DUPLICATE
23628 if it has non-VECTOR_MODE element mode.
23629 (vec_duplicate_p): Likewise.
23631 PR middle-end/83694
23632 * cfgexpand.c (expand_debug_expr): Punt if mode1 is VOIDmode
23633 and bitsize might be greater than MAX_BITSIZE_MODE_ANY_INT.
23635 2018-01-05 Jakub Jelinek <jakub@redhat.com>
23638 * config/i386/i386-builtin.def
23639 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
23640 __builtin_ia32_vgf2p8affineqb_v64qi, __builtin_ia32_vgf2p8mulb_v64qi):
23641 Require also OPTION_MASK_ISA_AVX512F in addition to
23642 OPTION_MASK_ISA_GFNI.
23643 (__builtin_ia32_vgf2p8affineinvqb_v16qi_mask,
23644 __builtin_ia32_vgf2p8affineqb_v16qi_mask): Require
23645 OPTION_MASK_ISA_AVX512VL instead of OPTION_MASK_ISA_SSE in addition
23646 to OPTION_MASK_ISA_GFNI.
23647 (__builtin_ia32_vgf2p8mulb_v32qi_mask): Require
23648 OPTION_MASK_ISA_AVX512VL in addition to OPTION_MASK_ISA_GFNI and
23649 OPTION_MASK_ISA_AVX512BW.
23650 (__builtin_ia32_vgf2p8mulb_v16qi_mask): Require
23651 OPTION_MASK_ISA_AVX512VL instead of OPTION_MASK_ISA_AVX512BW in
23652 addition to OPTION_MASK_ISA_GFNI.
23653 (__builtin_ia32_vgf2p8affineinvqb_v16qi,
23654 __builtin_ia32_vgf2p8affineqb_v16qi, __builtin_ia32_vgf2p8mulb_v16qi):
23655 Require OPTION_MASK_ISA_SSE2 instead of OPTION_MASK_ISA_SSE in addition
23656 to OPTION_MASK_ISA_GFNI.
23657 * config/i386/i386.c (def_builtin): Change to builtin isa/isa2 being
23658 a requirement for all ISAs rather than any of them with a few
23660 (ix86_add_new_builtins): Clear OPTION_MASK_ISA_64BIT from isa before
23662 (ix86_expand_builtin): Require all ISAs from builtin's isa and isa2
23663 bitmasks to be enabled with 3 exceptions, instead of requiring any
23664 enabled ISA with lots of exceptions.
23665 * config/i386/sse.md (vgf2p8affineinvqb_<mode><mask_name>,
23666 vgf2p8affineqb_<mode><mask_name>, vgf2p8mulb_<mode><mask_name>):
23667 Change avx512bw in isa attribute to avx512f.
23668 * config/i386/sgxintrin.h: Add license boilerplate.
23669 * config/i386/vaesintrin.h: Likewise. Fix macro spelling __AVX512F
23670 to __AVX512F__ and __AVX512VL to __AVX512VL__.
23671 (_mm256_aesdec_epi128, _mm256_aesdeclast_epi128, _mm256_aesenc_epi128,
23672 _mm256_aesenclast_epi128): Enable temporarily avx if __AVX__ is not
23674 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8,
23675 _mm_gf2p8affineinv_epi64_epi8, _mm_gf2p8affine_epi64_epi8): Enable
23676 temporarily sse2 rather than sse if not enabled already.
23679 * config/i386/sse.md (VI248_VLBW): Rename to ...
23680 (VI248_AVX512VL): ... this. Don't guard V32HI with TARGET_AVX512BW.
23681 (vpshrd_<mode><mask_name>, vpshld_<mode><mask_name>,
23682 vpshrdv_<mode>, vpshrdv_<mode>_mask, vpshrdv_<mode>_maskz,
23683 vpshrdv_<mode>_maskz_1, vpshldv_<mode>, vpshldv_<mode>_mask,
23684 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): Use VI248_AVX512VL
23685 mode iterator instead of VI248_VLBW.
23687 2018-01-05 Jan Hubicka <hubicka@ucw.cz>
23689 * ipa-fnsummary.c (record_modified_bb_info): Add OP.
23690 (record_modified): Skip clobbers; add debug output.
23691 (param_change_prob): Use sreal frequencies.
23693 2018-01-05 Richard Sandiford <richard.sandiford@linaro.org>
23695 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
23696 punt for user-aligned variables.
23698 2018-01-05 Richard Sandiford <richard.sandiford@linaro.org>
23700 * tree-chrec.c (chrec_contains_symbols): Return true for
23703 2018-01-05 Sudakshina Das <sudi.das@arm.com>
23706 * simplify-rtx.c (simplify_relational_operation_1): Add simplifications
23707 of (x|y) == x for BICS pattern.
23709 2018-01-05 Jakub Jelinek <jakub@redhat.com>
23711 PR tree-optimization/83605
23712 * gimple-ssa-strength-reduction.c: Include tree-eh.h.
23713 (find_candidates_dom_walker::before_dom_children): Ignore stmts that
23716 2018-01-05 Sebastian Huber <sebastian.huber@embedded-brains.de>
23718 * config.gcc (epiphany-*-elf*): Add (epiphany-*-rtems*) configuration.
23719 * config/epiphany/rtems.h: New file.
23721 2018-01-04 Jakub Jelinek <jakub@redhat.com>
23722 Uros Bizjak <ubizjak@gmail.com>
23725 * config/i386/i386.md (*<rotate_insn>hi3_1 splitter): Use
23726 QIreg_operand instead of register_operand predicate.
23727 * config/i386/i386.c (ix86_rop_should_change_byte_p,
23728 set_rop_modrm_reg_bits, ix86_mitigate_rop): Use -mmitigate-rop in
23729 comments instead of -fmitigate[-_]rop.
23731 2018-01-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
23734 * cgraphunit.c (symbol_table::compile): Switch to text_section
23735 before calling assembly_start debug hook.
23736 * run-rtl-passes.c (run_rtl_passes): Likewise.
23739 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
23741 * tree-vrp.c (extract_range_from_binary_expr_1): Check
23742 range_int_cst_p rather than !symbolic_range_p before calling
23743 extract_range_from_multiplicative_op_1.
23745 2018-01-04 Jeff Law <law@redhat.com>
23747 * tree-ssa-math-opts.c (execute_cse_reciprocals_1): Remove
23748 redundant test in assertion.
23750 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
23752 * doc/rtl.texi: Document machine_mode wrapper classes.
23754 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
23756 * fold-const.c (fold_ternary_loc): Check tree_fits_uhwi_p before
23757 using tree_to_uhwi.
23759 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
23761 * tree-ssa-forwprop.c (is_combined_permutation_identity): Allow
23762 the VEC_PERM_EXPR fold to fail.
23764 2018-01-04 Jakub Jelinek <jakub@redhat.com>
23767 * bb-reorder.c (insert_section_boundary_note): Set has_bb_partition
23768 to switched_sections.
23770 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
23773 * config/arm/arm.c (arm_vectorize_vec_perm_const): Fix inverted
23774 test for d.testing.
23776 2018-01-04 Peter Bergner <bergner@vnet.ibm.com>
23779 * config/rs6000/rs6000.c (rs6000_discover_homogeneous_aggregate): Do not
23780 allow arguments in FP registers if TARGET_HARD_FLOAT is false.
23782 2018-01-04 Jakub Jelinek <jakub@redhat.com>
23785 * cfgexpand.c (expand_debug_expr) <case BIT_FIELD_REF>: Punt if mode
23786 is BLKmode and bitpos not zero or mode change is needed.
23788 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
23791 * config/sparc/sparc.c (sparc_vectorize_vec_perm_const): Require
23794 2018-01-04 Uros Bizjak <ubizjak@gmail.com>
23797 * config/alpha/alpha.md (*sadd<modesuffix>): Use ASHIFT
23798 instead of MULT rtx. Update all corresponding splitters.
23799 (*saddl_se): Ditto.
23800 (*ssub<modesuffix>): Ditto.
23801 (*ssubl_se): Ditto.
23802 (*cmp_sadd_di): Update split patterns.
23803 (*cmp_sadd_si): Ditto.
23804 (*cmp_sadd_sidi): Ditto.
23805 (*cmp_ssub_di): Ditto.
23806 (*cmp_ssub_si): Ditto.
23807 (*cmp_ssub_sidi): Ditto.
23808 * config/alpha/predicates.md (const23_operand): New predicate.
23809 * config/alpha/alpha.c (alpha_rtx_costs) [PLUS, MINUS]:
23810 Look for ASHIFT, not MULT inner operand.
23811 (alpha_split_conditional_move): Update for *sadd<modesuffix> change.
23813 2018-01-04 Martin Liska <mliska@suse.cz>
23815 PR gcov-profile/83669
23816 * gcov.c (output_intermediate_file): Add version to intermediate
23818 * doc/gcov.texi: Document new field 'version' in intermediate
23819 file format. Fix location of '-k' option of gcov command.
23821 2018-01-04 Martin Liska <mliska@suse.cz>
23824 * ipa-icf.c (sem_function::merge): Do not cross comdat boundary.
23826 2018-01-04 Jakub Jelinek <jakub@redhat.com>
23828 * gimple-ssa-sprintf.c (parse_directive): Cast second dir.len to uhwi.
23830 2018-01-03 Martin Sebor <msebor@redhat.com>
23832 PR tree-optimization/83655
23833 * gimple-ssa-warn-restrict.c (wrestrict_dom_walker::check_call): Avoid
23834 checking calls with invalid arguments.
23836 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23838 * tree-vect-stmts.c (vect_get_store_rhs): New function.
23839 (vectorizable_mask_load_store): Delete.
23840 (vectorizable_call): Return false for masked loads and stores.
23841 (vectorizable_store): Handle IFN_MASK_STORE. Use vect_get_store_rhs
23842 instead of gimple_assign_rhs1.
23843 (vectorizable_load): Handle IFN_MASK_LOAD.
23844 (vect_transform_stmt): Don't set is_store for call_vec_info_type.
23846 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23848 * tree-vect-stmts.c (vect_build_gather_load_calls): New function,
23850 (vectorizable_mask_load_store): ...here.
23851 (vectorizable_load): ...and here.
23853 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23855 * tree-vect-stmts.c (vect_build_all_ones_mask)
23856 (vect_build_zero_merge_argument): New functions, split out from...
23857 (vectorizable_load): ...here.
23859 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23861 * tree-vect-stmts.c (vect_check_store_rhs): New function,
23863 (vectorizable_mask_load_store): ...here.
23864 (vectorizable_store): ...and here.
23866 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23868 * tree-vect-stmts.c (vect_check_load_store_mask): New function,
23870 (vectorizable_mask_load_store): ...here.
23872 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23874 * tree-vectorizer.h (vec_load_store_type): Moved from tree-vec-stmts.c
23875 (vect_model_store_cost): Take a vec_load_store_type instead of a
23877 * tree-vect-stmts.c (vec_load_store_type): Move to tree-vectorizer.h.
23878 (vect_model_store_cost): Take a vec_load_store_type instead of a
23880 (vectorizable_mask_load_store): Update accordingly.
23881 (vectorizable_store): Likewise.
23882 * tree-vect-slp.c (vect_analyze_slp_cost_1): Update accordingly.
23884 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23886 * tree-vect-loop.c (vect_transform_loop): Stub out scalar
23887 IFN_MASK_LOAD calls here rather than...
23888 * tree-vect-stmts.c (vectorizable_mask_load_store): ...here.
23890 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23891 Alan Hayward <alan.hayward@arm.com>
23892 David Sherwood <david.sherwood@arm.com>
23894 * expmed.c (extract_bit_field_1): For vector extracts,
23895 fall back to extract_bit_field_as_subreg if vec_extract
23898 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23899 Alan Hayward <alan.hayward@arm.com>
23900 David Sherwood <david.sherwood@arm.com>
23902 * lra-spills.c (pseudo_reg_slot_compare): Sort slots by whether
23903 they are variable or constant sized.
23904 (assign_stack_slot_num_and_sort_pseudos): Don't reuse variable-sized
23905 slots for constant-sized data.
23907 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23908 Alan Hayward <alan.hayward@arm.com>
23909 David Sherwood <david.sherwood@arm.com>
23911 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern): When
23912 handling COND_EXPRs with boolean comparisons, try to find a better
23913 basis for the mask type than the boolean itself.
23915 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23917 * doc/rtl.texi (MAX_BITSIZE_MODE_ANY_MODE): Describe how the default
23918 is calculated and how it can be overridden.
23919 * genmodes.c (max_bitsize_mode_any_mode): New variable.
23920 (create_modes): Initialize it from MAX_BITSIZE_MODE_ANY_MODE,
23922 (emit_max_int): Use it to set the output MAX_BITSIZE_MODE_ANY_MODE,
23925 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23926 Alan Hayward <alan.hayward@arm.com>
23927 David Sherwood <david.sherwood@arm.com>
23929 * config/aarch64/aarch64-protos.h (aarch64_output_simd_mov_immediate):
23930 Remove the mode argument.
23931 (aarch64_simd_valid_immediate): Remove the mode and inverse
23933 * config/aarch64/iterators.md (bitsize): New iterator.
23934 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>, and<mode>3)
23935 (ior<mode>3): Update calls to aarch64_output_simd_mov_immediate.
23936 * config/aarch64/constraints.md (Do, Db, Dn): Update calls to
23937 aarch64_simd_valid_immediate.
23938 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Likewise.
23939 (aarch64_reg_or_bic_imm): Likewise.
23940 * config/aarch64/aarch64.c (simd_immediate_info): Replace mvn
23941 with an insn_type enum and msl with a modifier_type enum.
23942 Replace element_width with a scalar_mode. Change the shift
23943 to unsigned int. Add constructors for scalar_float_mode and
23944 scalar_int_mode elements.
23945 (aarch64_vect_float_const_representable_p): Delete.
23946 (aarch64_can_const_movi_rtx_p)
23947 (aarch64_simd_scalar_immediate_valid_for_move)
23948 (aarch64_simd_make_constant): Update call to
23949 aarch64_simd_valid_immediate.
23950 (aarch64_advsimd_valid_immediate_hs): New function.
23951 (aarch64_advsimd_valid_immediate): Likewise.
23952 (aarch64_simd_valid_immediate): Remove mode and inverse
23953 arguments. Rewrite to use the above. Use const_vec_duplicate_p
23954 to detect duplicated constants and use aarch64_float_const_zero_rtx_p
23955 and aarch64_float_const_representable_p on the result.
23956 (aarch64_output_simd_mov_immediate): Remove mode argument.
23957 Update call to aarch64_simd_valid_immediate and use of
23958 simd_immediate_info.
23959 (aarch64_output_scalar_simd_mov_immediate): Update call
23962 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23963 Alan Hayward <alan.hayward@arm.com>
23964 David Sherwood <david.sherwood@arm.com>
23966 * machmode.h (mode_precision): Prefix with CONST_MODE_PRECISION.
23967 (mode_nunits): Likewise CONST_MODE_NUNITS.
23968 * machmode.def (ADJUST_NUNITS): Document.
23969 * genmodes.c (mode_data::need_nunits_adj): New field.
23970 (blank_mode): Update accordingly.
23971 (adj_nunits): New variable.
23972 (print_maybe_const_decl): Replace CATEGORY with a NEEDS_ADJ
23974 (emit_mode_size_inline): Set need_bytesize_adj for all modes
23975 listed in adj_nunits.
23976 (emit_mode_nunits_inline): Set need_nunits_adj for all modes
23977 listed in adj_nunits. Don't emit case statements for such modes.
23978 (emit_insn_modes_h): Emit definitions of CONST_MODE_NUNITS
23979 and CONST_MODE_PRECISION. Make CONST_MODE_SIZE expand to
23980 nothing if adj_nunits is nonnull.
23981 (emit_mode_precision, emit_mode_nunits): Use print_maybe_const_decl.
23982 (emit_mode_unit_size, emit_mode_base_align, emit_mode_ibit)
23983 (emit_mode_fbit): Update use of print_maybe_const_decl.
23984 (emit_move_size): Likewise. Treat the array as non-const
23986 (emit_mode_adjustments): Handle adj_nunits.
23988 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23990 * machmode.def (VECTOR_MODES_WITH_PREFIX): Document.
23991 * genmodes.c (VECTOR_MODES_WITH_PREFIX): New macro.
23992 (VECTOR_MODES): Use it.
23993 (make_vector_modes): Take the prefix as an argument.
23995 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
23996 Alan Hayward <alan.hayward@arm.com>
23997 David Sherwood <david.sherwood@arm.com>
23999 * mode-classes.def (MODE_VECTOR_BOOL): New mode class.
24000 * machmode.h (INTEGRAL_MODE_P, VECTOR_MODE_P): Return true
24001 for MODE_VECTOR_BOOL.
24002 * machmode.def (VECTOR_BOOL_MODE): Document.
24003 * genmodes.c (VECTOR_BOOL_MODE): New macro.
24004 (make_vector_bool_mode): New function.
24005 (complete_mode, emit_mode_wider, emit_mode_adjustments): Handle
24007 * lto-streamer-in.c (lto_input_mode_table): Likewise.
24008 * rtx-vector-builder.c (rtx_vector_builder::find_cached_value):
24010 * stor-layout.c (int_mode_for_mode): Likewise.
24011 * tree.c (build_vector_type_for_mode): Likewise.
24012 * varasm.c (output_constant_pool_2): Likewise.
24013 * emit-rtl.c (init_emit_once): Make sure that CONST1_RTX (BImode) and
24014 CONSTM1_RTX (BImode) are the same thing. Initialize const_tiny_rtx
24015 for MODE_VECTOR_BOOL.
24016 * expr.c (expand_expr_real_1): Use VECTOR_MODE_P instead of a list
24017 of mode class checks.
24018 * tree-vect-generic.c (expand_vector_operation): Use VECTOR_MODE_P
24019 instead of a list of mode class checks.
24020 (expand_vector_scalar_condition): Likewise.
24021 (type_for_widest_vector_mode): Handle BImode as an inner mode.
24023 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24024 Alan Hayward <alan.hayward@arm.com>
24025 David Sherwood <david.sherwood@arm.com>
24027 * machmode.h (mode_size): Change from unsigned short to
24029 (mode_to_bytes): Return a poly_uint16 rather than an unsigned short.
24030 (GET_MODE_SIZE): Return a constant if ONLY_FIXED_SIZE_MODES,
24031 or if measurement_type is not polynomial.
24032 (fixed_size_mode::includes_p): Check for constant-sized modes.
24033 * genmodes.c (emit_mode_size_inline): Make mode_size_inline
24034 return a poly_uint16 rather than an unsigned short.
24035 (emit_mode_size): Change the type of mode_size from unsigned short
24036 to poly_uint16_pod. Use ZERO_COEFFS for the initializer.
24037 (emit_mode_adjustments): Cope with polynomial vector sizes.
24038 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
24040 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
24042 * auto-inc-dec.c (try_merge): Treat GET_MODE_SIZE as polynomial.
24043 * builtins.c (expand_ifn_atomic_compare_exchange_into_call): Likewise.
24044 * caller-save.c (setup_save_areas): Likewise.
24045 (replace_reg_with_saved_mem): Likewise.
24046 * calls.c (emit_library_call_value_1): Likewise.
24047 * combine-stack-adj.c (combine_stack_adjustments_for_block): Likewise.
24048 * combine.c (simplify_set, make_extraction, simplify_shift_const_1)
24049 (gen_lowpart_for_combine): Likewise.
24050 * convert.c (convert_to_integer_1): Likewise.
24051 * cse.c (equiv_constant, cse_insn): Likewise.
24052 * cselib.c (autoinc_split, cselib_hash_rtx): Likewise.
24053 (cselib_subst_to_values): Likewise.
24054 * dce.c (word_dce_process_block): Likewise.
24055 * df-problems.c (df_word_lr_mark_ref): Likewise.
24056 * dwarf2cfi.c (init_one_dwarf_reg_size): Likewise.
24057 * dwarf2out.c (multiple_reg_loc_descriptor, mem_loc_descriptor)
24058 (concat_loc_descriptor, concatn_loc_descriptor, loc_descriptor)
24059 (rtl_for_decl_location): Likewise.
24060 * emit-rtl.c (gen_highpart, widen_memory_access): Likewise.
24061 * expmed.c (extract_bit_field_1, extract_integral_bit_field): Likewise.
24062 * expr.c (emit_group_load_1, clear_storage_hints): Likewise.
24063 (emit_move_complex, emit_move_multi_word, emit_push_insn): Likewise.
24064 (expand_expr_real_1): Likewise.
24065 * function.c (assign_parm_setup_block_p, assign_parm_setup_block)
24066 (pad_below): Likewise.
24067 * gimple-fold.c (optimize_atomic_compare_exchange_p): Likewise.
24068 * gimple-ssa-store-merging.c (rhs_valid_for_store_merging_p): Likewise.
24069 * ira.c (get_subreg_tracking_sizes): Likewise.
24070 * ira-build.c (ira_create_allocno_objects): Likewise.
24071 * ira-color.c (coalesced_pseudo_reg_slot_compare): Likewise.
24072 (ira_sort_regnos_for_alter_reg): Likewise.
24073 * ira-costs.c (record_operand_costs): Likewise.
24074 * lower-subreg.c (interesting_mode_p, simplify_gen_subreg_concatn)
24075 (resolve_simple_move): Likewise.
24076 * lra-constraints.c (get_reload_reg, operands_match_p): Likewise.
24077 (process_addr_reg, simplify_operand_subreg, curr_insn_transform)
24078 (lra_constraints): Likewise.
24079 (CONST_POOL_OK_P): Reject variable-sized modes.
24080 * lra-spills.c (slot, assign_mem_slot, pseudo_reg_slot_compare)
24081 (add_pseudo_to_slot, lra_spill): Likewise.
24082 * omp-low.c (omp_clause_aligned_alignment): Likewise.
24083 * optabs-query.c (get_best_extraction_insn): Likewise.
24084 * optabs-tree.c (expand_vec_cond_expr_p): Likewise.
24085 * optabs.c (expand_vec_perm_var, expand_vec_cond_expr): Likewise.
24086 (expand_mult_highpart, valid_multiword_target_p): Likewise.
24087 * recog.c (offsettable_address_addr_space_p): Likewise.
24088 * regcprop.c (maybe_mode_change): Likewise.
24089 * reginfo.c (choose_hard_reg_mode, record_subregs_of_mode): Likewise.
24090 * regrename.c (build_def_use): Likewise.
24091 * regstat.c (dump_reg_info): Likewise.
24092 * reload.c (complex_word_subreg_p, push_reload, find_dummy_reload)
24093 (find_reloads, find_reloads_subreg_address): Likewise.
24094 * reload1.c (eliminate_regs_1): Likewise.
24095 * rtlanal.c (for_each_inc_dec_find_inc_dec, rtx_cost): Likewise.
24096 * simplify-rtx.c (avoid_constant_pool_reference): Likewise.
24097 (simplify_binary_operation_1, simplify_subreg): Likewise.
24098 * targhooks.c (default_function_arg_padding): Likewise.
24099 (default_hard_regno_nregs, default_class_max_nregs): Likewise.
24100 * tree-cfg.c (verify_gimple_assign_binary): Likewise.
24101 (verify_gimple_assign_ternary): Likewise.
24102 * tree-inline.c (estimate_move_cost): Likewise.
24103 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
24104 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Likewise.
24105 (get_address_cost_ainc): Likewise.
24106 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
24107 (vect_supportable_dr_alignment): Likewise.
24108 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
24109 (vectorizable_reduction): Likewise.
24110 * tree-vect-stmts.c (vectorizable_assignment, vectorizable_shift)
24111 (vectorizable_operation, vectorizable_load): Likewise.
24112 * tree.c (build_same_sized_truth_vector_type): Likewise.
24113 * valtrack.c (cleanup_auto_inc_dec): Likewise.
24114 * var-tracking.c (emit_note_insn_var_location): Likewise.
24115 * config/arc/arc.h (ASM_OUTPUT_CASE_END): Use as_a <scalar_int_mode>.
24116 (ADDR_VEC_ALIGN): Likewise.
24118 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24119 Alan Hayward <alan.hayward@arm.com>
24120 David Sherwood <david.sherwood@arm.com>
24122 * machmode.h (mode_to_bits): Return a poly_uint16 rather than an
24124 (GET_MODE_BITSIZE): Return a constant if ONLY_FIXED_SIZE_MODES,
24125 or if measurement_type is polynomial.
24126 * calls.c (shift_return_value): Treat GET_MODE_BITSIZE as polynomial.
24127 * combine.c (make_extraction): Likewise.
24128 * dse.c (find_shift_sequence): Likewise.
24129 * dwarf2out.c (mem_loc_descriptor): Likewise.
24130 * expmed.c (store_integral_bit_field, extract_bit_field_1): Likewise.
24131 (extract_bit_field, extract_low_bits): Likewise.
24132 * expr.c (convert_move, convert_modes, emit_move_insn_1): Likewise.
24133 (optimize_bitfield_assignment_op, expand_assignment): Likewise.
24134 (store_expr_with_bounds, store_field, expand_expr_real_1): Likewise.
24135 * fold-const.c (optimize_bit_field_compare, merge_ranges): Likewise.
24136 * gimple-fold.c (optimize_atomic_compare_exchange_p): Likewise.
24137 * reload.c (find_reloads): Likewise.
24138 * reload1.c (alter_reg): Likewise.
24139 * stor-layout.c (bitwise_mode_for_mode, compute_record_mode): Likewise.
24140 * targhooks.c (default_secondary_memory_needed_mode): Likewise.
24141 * tree-if-conv.c (predicate_mem_writes): Likewise.
24142 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
24143 * tree-vect-patterns.c (adjust_bool_pattern): Likewise.
24144 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
24145 * valtrack.c (dead_debug_insert_temp): Likewise.
24146 * varasm.c (mergeable_constant_section): Likewise.
24147 * config/sh/sh.h (LOCAL_ALIGNMENT): Use as_a <fixed_size_mode>.
24149 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24150 Alan Hayward <alan.hayward@arm.com>
24151 David Sherwood <david.sherwood@arm.com>
24153 * expr.c (expand_assignment): Cope with polynomial mode sizes
24154 when assigning to a CONCAT.
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 * machmode.h (mode_precision): Change from unsigned short to
24162 (mode_to_precision): Return a poly_uint16 rather than an unsigned
24164 (GET_MODE_PRECISION): Return a constant if ONLY_FIXED_SIZE_MODES,
24165 or if measurement_type is not polynomial.
24166 (HWI_COMPUTABLE_MODE_P): Turn into a function. Optimize the case
24167 in which the mode is already known to be a scalar_int_mode.
24168 * genmodes.c (emit_mode_precision): Change the type of mode_precision
24169 from unsigned short to poly_uint16_pod. Use ZERO_COEFFS for the
24171 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
24172 for GET_MODE_PRECISION.
24173 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
24174 for GET_MODE_PRECISION.
24175 * combine.c (update_rsp_from_reg_equal): Treat GET_MODE_PRECISION
24177 (try_combine, find_split_point, combine_simplify_rtx): Likewise.
24178 (expand_field_assignment, make_extraction): Likewise.
24179 (make_compound_operation_int, record_dead_and_set_regs_1): Likewise.
24180 (get_last_value): Likewise.
24181 * convert.c (convert_to_integer_1): Likewise.
24182 * cse.c (cse_insn): Likewise.
24183 * expr.c (expand_expr_real_1): Likewise.
24184 * lra-constraints.c (simplify_operand_subreg): Likewise.
24185 * optabs-query.c (can_atomic_load_p): Likewise.
24186 * optabs.c (expand_atomic_load): Likewise.
24187 (expand_atomic_store): Likewise.
24188 * ree.c (combine_reaching_defs): Likewise.
24189 * rtl.h (partial_subreg_p, paradoxical_subreg_p): Likewise.
24190 * rtlanal.c (nonzero_bits1, lsb_bitfield_op_p): Likewise.
24191 * tree.h (type_has_mode_precision_p): Likewise.
24192 * ubsan.c (instrument_si_overflow): Likewise.
24194 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24195 Alan Hayward <alan.hayward@arm.com>
24196 David Sherwood <david.sherwood@arm.com>
24198 * tree.h (TYPE_VECTOR_SUBPARTS): Turn into a function and handle
24199 polynomial numbers of units.
24200 (SET_TYPE_VECTOR_SUBPARTS): Likewise.
24201 (valid_vector_subparts_p): New function.
24202 (build_vector_type): Remove temporary shim and take the number
24203 of units as a poly_uint64 rather than an int.
24204 (build_opaque_vector_type): Take the number of units as a
24205 poly_uint64 rather than an int.
24206 * tree.c (build_vector_from_ctor): Handle polynomial
24207 TYPE_VECTOR_SUBPARTS.
24208 (type_hash_canon_hash, type_cache_hasher::equal): Likewise.
24209 (uniform_vector_p, vector_type_mode, build_vector): Likewise.
24210 (build_vector_from_val): If the number of units is variable,
24211 use build_vec_duplicate_cst for constant operands and
24212 VEC_DUPLICATE_EXPR otherwise.
24213 (make_vector_type): Remove temporary is_constant ().
24214 (build_vector_type, build_opaque_vector_type): Take the number of
24215 units as a poly_uint64 rather than an int.
24216 (check_vector_cst): Handle polynomial TYPE_VECTOR_SUBPARTS and
24218 * cfgexpand.c (expand_debug_expr): Likewise.
24219 * expr.c (count_type_elements, categorize_ctor_elements_1): Likewise.
24220 (store_constructor, expand_expr_real_1): Likewise.
24221 (const_scalar_mask_from_tree): Likewise.
24222 * fold-const-call.c (fold_const_reduction): Likewise.
24223 * fold-const.c (const_binop, const_unop, fold_convert_const): Likewise.
24224 (operand_equal_p, fold_vec_perm, fold_ternary_loc): Likewise.
24225 (native_encode_vector, vec_cst_ctor_to_array): Likewise.
24226 (fold_relational_const): Likewise.
24227 (native_interpret_vector): Likewise. Change the size from an
24228 int to an unsigned int.
24229 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Handle polynomial
24230 TYPE_VECTOR_SUBPARTS.
24231 (gimple_fold_indirect_ref, gimple_build_vector): Likewise.
24232 (gimple_build_vector_from_val): Use VEC_DUPLICATE_EXPR when
24233 duplicating a non-constant operand into a variable-length vector.
24234 * hsa-brig.c (hsa_op_immed::emit_to_buffer): Handle polynomial
24235 TYPE_VECTOR_SUBPARTS and VECTOR_CST_NELTS.
24236 * ipa-icf.c (sem_variable::equals): Likewise.
24237 * match.pd: Likewise.
24238 * omp-simd-clone.c (simd_clone_subparts): Likewise.
24239 * print-tree.c (print_node): Likewise.
24240 * stor-layout.c (layout_type): Likewise.
24241 * targhooks.c (default_builtin_vectorization_cost): Likewise.
24242 * tree-cfg.c (verify_gimple_comparison): Likewise.
24243 (verify_gimple_assign_binary): Likewise.
24244 (verify_gimple_assign_ternary): Likewise.
24245 (verify_gimple_assign_single): Likewise.
24246 * tree-pretty-print.c (dump_generic_node): Likewise.
24247 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
24248 (simplify_bitfield_ref, is_combined_permutation_identity): Likewise.
24249 * tree-vect-data-refs.c (vect_permute_store_chain): Likewise.
24250 (vect_grouped_load_supported, vect_permute_load_chain): Likewise.
24251 (vect_shift_permute_load_chain): Likewise.
24252 * tree-vect-generic.c (nunits_for_known_piecewise_op): Likewise.
24253 (expand_vector_condition, optimize_vector_constructor): Likewise.
24254 (lower_vec_perm, get_compute_type): Likewise.
24255 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
24256 (get_initial_defs_for_reduction, vect_transform_loop): Likewise.
24257 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
24258 (vect_recog_mask_conversion_pattern): Likewise.
24259 * tree-vect-slp.c (vect_supported_load_permutation_p): Likewise.
24260 (vect_get_constant_vectors, vect_transform_slp_perm_load): Likewise.
24261 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
24262 (get_group_load_store_type, vectorizable_mask_load_store): Likewise.
24263 (vectorizable_bswap, simd_clone_subparts, vectorizable_assignment)
24264 (vectorizable_shift, vectorizable_operation, vectorizable_store)
24265 (vectorizable_load, vect_is_simple_cond, vectorizable_comparison)
24266 (supportable_widening_operation): Likewise.
24267 (supportable_narrowing_operation): Likewise.
24268 * tree-vector-builder.c (tree_vector_builder::binary_encoded_nelts):
24270 * varasm.c (output_constant): Likewise.
24272 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24273 Alan Hayward <alan.hayward@arm.com>
24274 David Sherwood <david.sherwood@arm.com>
24276 * tree-vect-data-refs.c (vect_permute_store_chain): Reorganize
24277 so that both the length == 3 and length != 3 cases set up their
24278 own permute vectors. Add comments explaining why we know the
24279 number of elements is constant.
24280 (vect_permute_load_chain): Likewise.
24282 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24283 Alan Hayward <alan.hayward@arm.com>
24284 David Sherwood <david.sherwood@arm.com>
24286 * machmode.h (mode_nunits): Change from unsigned char to
24288 (ONLY_FIXED_SIZE_MODES): New macro.
24289 (pod_mode::measurement_type, scalar_int_mode::measurement_type)
24290 (scalar_float_mode::measurement_type, scalar_mode::measurement_type)
24291 (complex_mode::measurement_type, fixed_size_mode::measurement_type):
24293 (mode_to_nunits): Return a poly_uint16 rather than an unsigned short.
24294 (GET_MODE_NUNITS): Return a constant if ONLY_FIXED_SIZE_MODES,
24295 or if measurement_type is not polynomial.
24296 * genmodes.c (ZERO_COEFFS): New macro.
24297 (emit_mode_nunits_inline): Make mode_nunits_inline return a
24299 (emit_mode_nunits): Change the type of mode_nunits to poly_uint16_pod.
24300 Use ZERO_COEFFS when emitting initializers.
24301 * data-streamer.h (bp_pack_poly_value): New function.
24302 (bp_unpack_poly_value): Likewise.
24303 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
24304 for GET_MODE_NUNITS.
24305 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
24306 for GET_MODE_NUNITS.
24307 * tree.c (make_vector_type): Remove temporary shim and make
24308 the real function take the number of units as a poly_uint64
24309 rather than an int.
24310 (build_vector_type_for_mode): Handle polynomial nunits.
24311 * dwarf2out.c (loc_descriptor, add_const_value_attribute): Likewise.
24312 * emit-rtl.c (const_vec_series_p_1): Likewise.
24313 (gen_rtx_CONST_VECTOR): Likewise.
24314 * fold-const.c (test_vec_duplicate_folding): Likewise.
24315 * genrecog.c (validate_pattern): Likewise.
24316 * optabs-query.c (can_vec_perm_var_p, can_mult_highpart_p): Likewise.
24317 * optabs-tree.c (expand_vec_cond_expr_p): Likewise.
24318 * optabs.c (expand_vector_broadcast, expand_binop_directly): Likewise.
24319 (shift_amt_for_vec_perm_mask, expand_vec_perm_var): Likewise.
24320 (expand_vec_cond_expr, expand_mult_highpart): Likewise.
24321 * rtlanal.c (subreg_get_info): Likewise.
24322 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
24323 (vect_grouped_load_supported): Likewise.
24324 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
24325 * tree-vect-loop.c (have_whole_vector_shift): Likewise.
24326 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
24327 (simplify_const_unary_operation, simplify_binary_operation_1)
24328 (simplify_const_binary_operation, simplify_ternary_operation)
24329 (test_vector_ops_duplicate, test_vector_ops): Likewise.
24330 (simplify_immed_subreg): Use GET_MODE_NUNITS on a fixed_size_mode
24331 instead of CONST_VECTOR_NUNITS.
24332 * varasm.c (output_constant_pool_2): Likewise.
24333 * rtx-vector-builder.c (rtx_vector_builder::build): Only include the
24334 explicit-encoded elements in the XVEC for variable-length vectors.
24336 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24338 * lra-constraints.c (curr_insn_transform): Use partial_subreg_p.
24340 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24341 Alan Hayward <alan.hayward@arm.com>
24342 David Sherwood <david.sherwood@arm.com>
24344 * coretypes.h (fixed_size_mode): Declare.
24345 (fixed_size_mode_pod): New typedef.
24346 * builtins.h (target_builtins::x_apply_args_mode)
24347 (target_builtins::x_apply_result_mode): Change type to
24348 fixed_size_mode_pod.
24349 * builtins.c (apply_args_size, apply_result_size, result_vector)
24350 (expand_builtin_apply_args_1, expand_builtin_apply)
24351 (expand_builtin_return): Update accordingly.
24353 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24355 * cse.c (hash_rtx_cb): Hash only the encoded elements.
24356 * cselib.c (cselib_hash_rtx): Likewise.
24357 * expmed.c (make_tree): Build VECTOR_CSTs directly from the
24358 CONST_VECTOR encoding.
24360 2018-01-03 Jakub Jelinek <jakub@redhat.com>
24361 Jeff Law <law@redhat.com>
24364 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): For
24365 noreturn probe, use gen_pop instead of ix86_emit_restore_reg_using_pop,
24366 only set RTX_FRAME_RELATED_P on both the push and pop if cfa_reg is sp
24367 and add REG_CFA_ADJUST_CFA notes in that case to both insns.
24370 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Do not
24371 explicitly probe *sp in a noreturn function if there were any callee
24372 register saves or frame pointer is needed.
24374 2018-01-03 Jakub Jelinek <jakub@redhat.com>
24377 * cfgexpand.c (expand_debug_expr): Return NULL if mode is
24378 BLKmode for ternary, binary or unary expressions.
24381 * var-tracking.c (delete_vta_debug_insn): New inline function.
24382 (delete_vta_debug_insns): Add USE_CFG argument, if true, walk just
24383 insns from get_insns () to NULL instead of each bb separately.
24384 Use delete_vta_debug_insn. No longer static.
24385 (vt_debug_insns_local, variable_tracking_main_1): Adjust
24386 delete_vta_debug_insns callers.
24387 * rtl.h (delete_vta_debug_insns): Declare.
24388 * final.c (rest_of_handle_final): Call delete_vta_debug_insns
24389 instead of variable_tracking_main.
24391 2018-01-03 Martin Sebor <msebor@redhat.com>
24393 PR tree-optimization/83603
24394 * calls.c (maybe_warn_nonstring_arg): Avoid accessing function
24395 arguments past the endof the argument list in functions declared
24396 without a prototype.
24397 * gimple-ssa-warn-restrict.c (wrestrict_dom_walker::check_call):
24398 Avoid checking when arguments are null.
24400 2018-01-03 Martin Sebor <msebor@redhat.com>
24403 * doc/extend.texi (attribute const): Fix a typo.
24404 * ipa-pure-const.c ((warn_function_const, warn_function_pure): Avoid
24405 issuing -Wsuggest-attribute for void functions.
24407 2018-01-03 Martin Sebor <msebor@redhat.com>
24409 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Use
24410 offset_int::from instead of wide_int::to_shwi.
24411 (maybe_diag_overlap): Remove assertion.
24412 Use HOST_WIDE_INT_PRINT_DEC instead of %lli.
24413 * gimple-ssa-sprintf.c (format_directive): Same.
24414 (parse_directive): Same.
24415 (sprintf_dom_walker::compute_format_length): Same.
24416 (try_substitute_return_value): Same.
24418 2018-01-03 Jeff Law <law@redhat.com>
24420 PR middle-end/83654
24421 * explow.c (anti_adjust_stack_and_probe_stack_clash): Test a
24422 non-constant residual for zero at runtime and avoid probing in
24423 that case. Reorganize code for trailing problem to mirror handling
24426 2018-01-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
24428 PR tree-optimization/83501
24429 * tree-ssa-strlen.c (get_string_cst): New.
24430 (handle_char_store): Call get_string_cst.
24432 2018-01-03 Martin Liska <mliska@suse.cz>
24434 PR tree-optimization/83593
24435 * tree-ssa-strlen.c: Include tree-cfg.h.
24436 (strlen_check_and_optimize_stmt): Add new argument cleanup_eh.
24437 (strlen_dom_walker): Add new member variable m_cleanup_cfg.
24438 (strlen_dom_walker::strlen_dom_walker): Initialize m_cleanup_cfg
24440 (strlen_dom_walker::before_dom_children): Call
24441 gimple_purge_dead_eh_edges. Dump tranformation with details
24443 (strlen_dom_walker::before_dom_children): Update call by adding
24444 new argument cleanup_eh.
24445 (pass_strlen::execute): Return TODO_cleanup_cfg if needed.
24447 2018-01-03 Martin Liska <mliska@suse.cz>
24450 * cif-code.def (VARIADIC_THUNK): New enum value.
24451 * ipa-fnsummary.c (compute_fn_summary): Do not inline variadic
24454 2018-01-03 Jan Beulich <jbeulich@suse.com>
24456 * sse.md (mov<mode>_internal): Tighten condition for when to use
24457 vmovdqu<ssescalarsize> for TI and OI modes.
24459 2018-01-03 Jakub Jelinek <jakub@redhat.com>
24461 Update copyright years.
24463 2018-01-03 Martin Liska <mliska@suse.cz>
24466 * ipa-visibility.c (function_and_variable_visibility): Skip
24467 functions with noipa attribure.
24469 2018-01-03 Jakub Jelinek <jakub@redhat.com>
24471 * gcc.c (process_command): Update copyright notice dates.
24472 * gcov-dump.c (print_version): Ditto.
24473 * gcov.c (print_version): Ditto.
24474 * gcov-tool.c (print_version): Ditto.
24475 * gengtype.c (create_file): Ditto.
24476 * doc/cpp.texi: Bump @copying's copyright year.
24477 * doc/cppinternals.texi: Ditto.
24478 * doc/gcc.texi: Ditto.
24479 * doc/gccint.texi: Ditto.
24480 * doc/gcov.texi: Ditto.
24481 * doc/install.texi: Ditto.
24482 * doc/invoke.texi: Ditto.
24484 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24486 * vector-builder.h (vector_builder::m_full_nelts): Change from
24487 unsigned int to poly_uint64.
24488 (vector_builder::full_nelts): Update prototype accordingly.
24489 (vector_builder::new_vector): Likewise.
24490 (vector_builder::encoded_full_vector_p): Handle polynomial full_nelts.
24491 (vector_builder::operator ==): Likewise.
24492 (vector_builder::finalize): Likewise.
24493 * int-vector-builder.h (int_vector_builder::int_vector_builder):
24494 Take the number of elements as a poly_uint64 rather than an
24496 * vec-perm-indices.h (vec_perm_indices::m_nelts_per_input): Change
24497 from unsigned int to poly_uint64.
24498 (vec_perm_indices::vec_perm_indices): Update prototype accordingly.
24499 (vec_perm_indices::new_vector): Likewise.
24500 (vec_perm_indices::length): Likewise.
24501 (vec_perm_indices::nelts_per_input): Likewise.
24502 (vec_perm_indices::input_nelts): Likewise.
24503 * vec-perm-indices.c (vec_perm_indices::new_vector): Take the
24504 number of elements per input as a poly_uint64 rather than an
24505 unsigned int. Use the original encoding for variable-length
24506 vectors, rather than clamping each individual element.
24507 For the second and subsequent elements in each pattern,
24508 clamp the step and base before clamping their sum.
24509 (vec_perm_indices::series_p): Handle polynomial element counts.
24510 (vec_perm_indices::all_in_range_p): Likewise.
24511 (vec_perm_indices_to_tree): Likewise.
24512 (vec_perm_indices_to_rtx): Likewise.
24513 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
24514 * tree-vector-builder.c (tree_vector_builder::new_unary_operation)
24515 (tree_vector_builder::new_binary_operation): Handle polynomial
24516 element counts. Return false if we need to know the number
24517 of elements at compile time.
24518 * fold-const.c (fold_vec_perm): Punt if the number of elements
24519 isn't known at compile time.
24521 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24523 * vec-perm-indices.h (vec_perm_builder): Change element type
24524 from HOST_WIDE_INT to poly_int64.
24525 (vec_perm_indices::element_type): Update accordingly.
24526 (vec_perm_indices::clamp): Handle polynomial element_types.
24527 * vec-perm-indices.c (vec_perm_indices::series_p): Likewise.
24528 (vec_perm_indices::all_in_range_p): Likewise.
24529 (tree_to_vec_perm_builder): Check for poly_int64 trees rather
24531 * vector-builder.h (vector_builder::stepped_sequence_p): Handle
24532 polynomial vec_perm_indices element types.
24533 * int-vector-builder.h (int_vector_builder::equal_p): Likewise.
24534 * fold-const.c (fold_vec_perm): Likewise.
24535 * optabs.c (shift_amt_for_vec_perm_mask): Likewise.
24536 * tree-vect-generic.c (lower_vec_perm): Likewise.
24537 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
24538 * config/aarch64/aarch64.c (aarch64_evpc_tbl): Cast d->perm
24539 element type to HOST_WIDE_INT.
24541 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24542 Alan Hayward <alan.hayward@arm.com>
24543 David Sherwood <david.sherwood@arm.com>
24545 * alias.c (addr_side_effect_eval): Take the size as a poly_int64
24546 rather than an int. Use plus_constant.
24547 (memrefs_conflict_p): Take the sizes as poly_int64s rather than ints.
24548 Take the offset "c" as a poly_int64 rather than a HOST_WIDE_INT.
24550 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24551 Alan Hayward <alan.hayward@arm.com>
24552 David Sherwood <david.sherwood@arm.com>
24554 * calls.c (emit_call_1, expand_call): Change struct_value_size from
24555 a HOST_WIDE_INT to a poly_int64.
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 (load_register_parameters): Cope with polynomial
24562 mode sizes. Require a constant size for BLKmode parameters
24563 that aren't described by a PARALLEL. If BLOCK_REG_PADDING
24564 forces a parameter to be padded at the lsb end in order to
24565 fill a complete number of words, require the parameter size
24566 to be ordered wrt UNITS_PER_WORD.
24568 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24569 Alan Hayward <alan.hayward@arm.com>
24570 David Sherwood <david.sherwood@arm.com>
24572 * reload1.c (spill_stack_slot_width): Change element type
24573 from unsigned int to poly_uint64_pod.
24574 (alter_reg): Treat mode sizes as polynomial.
24576 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24577 Alan Hayward <alan.hayward@arm.com>
24578 David Sherwood <david.sherwood@arm.com>
24580 * reload.c (complex_word_subreg_p): New function.
24581 (reload_inner_reg_of_subreg, push_reload): Use it.
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 * lra-constraints.c (process_alt_operands): Reject matched
24588 operands whose sizes aren't ordered.
24589 (match_reload): Refer to this check here.
24591 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24592 Alan Hayward <alan.hayward@arm.com>
24593 David Sherwood <david.sherwood@arm.com>
24595 * builtins.c (expand_ifn_atomic_compare_exchange_into_call): Assert
24596 that the mode size is in the set {1, 2, 4, 8, 16}.
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 * var-tracking.c (adjust_mems): Treat mode sizes as polynomial.
24603 Use plus_constant instead of gen_rtx_PLUS.
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 * config/cr16/cr16-protos.h (cr16_push_rounding): Declare.
24610 * config/cr16/cr16.h (PUSH_ROUNDING): Move implementation to...
24611 * config/cr16/cr16.c (cr16_push_rounding): ...this new function.
24612 * config/h8300/h8300-protos.h (h8300_push_rounding): Declare.
24613 * config/h8300/h8300.h (PUSH_ROUNDING): Move implementation to...
24614 * config/h8300/h8300.c (h8300_push_rounding): ...this new function.
24615 * config/i386/i386-protos.h (ix86_push_rounding): Declare.
24616 * config/i386/i386.h (PUSH_ROUNDING): Move implementation to...
24617 * config/i386/i386.c (ix86_push_rounding): ...this new function.
24618 * config/m32c/m32c-protos.h (m32c_push_rounding): Take and return
24620 * config/m32c/m32c.c (m32c_push_rounding): Likewise.
24621 * config/m68k/m68k-protos.h (m68k_push_rounding): Declare.
24622 * config/m68k/m68k.h (PUSH_ROUNDING): Move implementation to...
24623 * config/m68k/m68k.c (m68k_push_rounding): ...this new function.
24624 * config/pdp11/pdp11-protos.h (pdp11_push_rounding): Declare.
24625 * config/pdp11/pdp11.h (PUSH_ROUNDING): Move implementation to...
24626 * config/pdp11/pdp11.c (pdp11_push_rounding): ...this new function.
24627 * config/stormy16/stormy16-protos.h (xstormy16_push_rounding): Declare.
24628 * config/stormy16/stormy16.h (PUSH_ROUNDING): Move implementation to...
24629 * config/stormy16/stormy16.c (xstormy16_push_rounding): ...this new
24631 * expr.c (emit_move_resolve_push): Treat the input and result
24632 of PUSH_ROUNDING as a poly_int64.
24633 (emit_move_complex_push, emit_single_push_insn_1): Likewise.
24634 (emit_push_insn): Likewise.
24635 * lra-eliminations.c (mark_not_eliminable): Likewise.
24636 * recog.c (push_operand): Likewise.
24637 * reload1.c (elimination_effects): Likewise.
24638 * rtlanal.c (nonzero_bits1): Likewise.
24639 * calls.c (store_one_arg): Likewise. Require the padding to be
24640 known at compile time.
24642 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24643 Alan Hayward <alan.hayward@arm.com>
24644 David Sherwood <david.sherwood@arm.com>
24646 * expr.c (emit_single_push_insn_1): Treat mode sizes as polynomial.
24647 Use plus_constant instead of gen_rtx_PLUS.
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 * auto-inc-dec.c (set_inc_state): Take the mode size as a poly_int64
24654 rather than an int.
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 * expr.c (expand_expr_real_1): Use tree_to_poly_uint64
24661 instead of int_size_in_bytes when handling VIEW_CONVERT_EXPRs
24662 via stack temporaries. Treat the mode size as polynomial too.
24664 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24665 Alan Hayward <alan.hayward@arm.com>
24666 David Sherwood <david.sherwood@arm.com>
24668 * expr.c (expand_expr_real_2): When handling conversions involving
24669 unions, apply tree_to_poly_uint64 to the TYPE_SIZE rather than
24670 multiplying int_size_in_bytes by BITS_PER_UNIT. Treat GET_MODE_BISIZE
24671 as a poly_uint64 too.
24673 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24674 Alan Hayward <alan.hayward@arm.com>
24675 David Sherwood <david.sherwood@arm.com>
24677 * rtlanal.c (subreg_get_info): Handle polynomial mode sizes.
24679 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24680 Alan Hayward <alan.hayward@arm.com>
24681 David Sherwood <david.sherwood@arm.com>
24683 * combine.c (can_change_dest_mode): Handle polynomial
24684 REGMODE_NATURAL_SIZE.
24685 * expmed.c (store_bit_field_1): Likewise.
24686 * expr.c (store_constructor): Likewise.
24687 * emit-rtl.c (validate_subreg): Operate on polynomial mode sizes
24688 and polynomial REGMODE_NATURAL_SIZE.
24689 (gen_lowpart_common): Likewise.
24690 * reginfo.c (record_subregs_of_mode): Likewise.
24691 * rtlanal.c (read_modify_subreg_p): Likewise.
24693 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24694 Alan Hayward <alan.hayward@arm.com>
24695 David Sherwood <david.sherwood@arm.com>
24697 * internal-fn.c (expand_vector_ubsan_overflow): Handle polynomial
24698 numbers of elements.
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 * match.pd: Cope with polynomial numbers of vector elements.
24706 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24707 Alan Hayward <alan.hayward@arm.com>
24708 David Sherwood <david.sherwood@arm.com>
24710 * fold-const.c (fold_indirect_ref_1): Handle polynomial offsets
24711 in a POINTER_PLUS_EXPR.
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 * omp-simd-clone.c (simd_clone_subparts): New function.
24718 (simd_clone_init_simd_arrays): Use it instead of TYPE_VECTOR_SUBPARTS.
24719 (ipa_simd_modify_function_body): Likewise.
24721 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24722 Alan Hayward <alan.hayward@arm.com>
24723 David Sherwood <david.sherwood@arm.com>
24725 * tree-vect-generic.c (nunits_for_known_piecewise_op): New function.
24726 (expand_vector_piecewise): Use it instead of TYPE_VECTOR_SUBPARTS.
24727 (expand_vector_addition, add_rshift, expand_vector_divmod): Likewise.
24728 (expand_vector_condition, vector_element): Likewise.
24729 (subparts_gt): New function.
24730 (get_compute_type): Use subparts_gt.
24731 (count_type_subparts): Delete.
24732 (expand_vector_operations_1): Use subparts_gt instead of
24733 count_type_subparts.
24735 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24736 Alan Hayward <alan.hayward@arm.com>
24737 David Sherwood <david.sherwood@arm.com>
24739 * tree-vect-data-refs.c (vect_no_alias_p): Replace with...
24740 (vect_compile_time_alias): ...this new function. Do the calculation
24741 on poly_ints rather than trees.
24742 (vect_prune_runtime_alias_test_list): Update call accordingly.
24744 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24745 Alan Hayward <alan.hayward@arm.com>
24746 David Sherwood <david.sherwood@arm.com>
24748 * tree-vect-slp.c (vect_build_slp_tree_1): Handle polynomial
24750 (vect_schedule_slp_instance): Likewise.
24752 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24753 Alan Hayward <alan.hayward@arm.com>
24754 David Sherwood <david.sherwood@arm.com>
24756 * tree-vect-slp.c (vect_get_and_check_slp_defs): Reject
24757 constant and extern definitions for variable-length vectors.
24758 (vect_get_constant_vectors): Note that the number of units
24759 is known to be constant.
24761 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24762 Alan Hayward <alan.hayward@arm.com>
24763 David Sherwood <david.sherwood@arm.com>
24765 * tree-vect-stmts.c (vectorizable_conversion): Treat the number
24766 of units as polynomial. Choose between WIDE and NARROW based
24769 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24770 Alan Hayward <alan.hayward@arm.com>
24771 David Sherwood <david.sherwood@arm.com>
24773 * tree-vect-stmts.c (simd_clone_subparts): New function.
24774 (vectorizable_simd_clone_call): Use it instead of TYPE_VECTOR_SUBPARTS.
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 (vectorizable_call): Treat the number of
24781 vectors as polynomial. Use build_index_vector for
24782 IFN_GOMP_SIMD_LANE.
24784 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24785 Alan Hayward <alan.hayward@arm.com>
24786 David Sherwood <david.sherwood@arm.com>
24788 * tree-vect-stmts.c (get_load_store_type): Treat the number of
24789 units as polynomial. Reject VMAT_ELEMENTWISE and VMAT_STRIDED_SLP
24790 for variable-length vectors.
24791 (vectorizable_mask_load_store): Treat the number of units as
24792 polynomial, asserting that it is constant if the condition has
24793 already been enforced.
24794 (vectorizable_store, vectorizable_load): Likewise.
24796 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24797 Alan Hayward <alan.hayward@arm.com>
24798 David Sherwood <david.sherwood@arm.com>
24800 * tree-vect-loop.c (vectorizable_live_operation): Treat the number
24801 of units as polynomial. Punt if we can't tell at compile time
24802 which vector contains the final result.
24804 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24805 Alan Hayward <alan.hayward@arm.com>
24806 David Sherwood <david.sherwood@arm.com>
24808 * tree-vect-loop.c (vectorizable_induction): Treat the number
24809 of units as polynomial. Punt on SLP inductions. Use an integer
24810 VEC_SERIES_EXPR for variable-length integer reductions. Use a
24811 cast of such a series for variable-length floating-point
24814 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24815 Alan Hayward <alan.hayward@arm.com>
24816 David Sherwood <david.sherwood@arm.com>
24818 * tree.h (build_index_vector): Declare.
24819 * tree.c (build_index_vector): New function.
24820 * tree-vect-loop.c (get_initial_defs_for_reduction): Treat the number
24821 of units as polynomial, forcibly converting it to a constant if
24822 vectorizable_reduction has already enforced the condition.
24823 (vect_create_epilog_for_reduction): Likewise. Use build_index_vector
24824 to create a {1,2,3,...} vector.
24825 (vectorizable_reduction): Treat the number of units as polynomial.
24826 Choose vectype_in based on the largest scalar element size rather
24827 than the smallest number of units. Enforce the restrictions
24830 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24831 Alan Hayward <alan.hayward@arm.com>
24832 David Sherwood <david.sherwood@arm.com>
24834 * tree-vect-data-refs.c (vector_alignment_reachable_p): Treat the
24835 number of units as polynomial.
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 * target.h (vector_sizes, auto_vector_sizes): New typedefs.
24842 * target.def (autovectorize_vector_sizes): Return the vector sizes
24843 by pointer, using vector_sizes rather than a bitmask.
24844 * targhooks.h (default_autovectorize_vector_sizes): Update accordingly.
24845 * targhooks.c (default_autovectorize_vector_sizes): Likewise.
24846 * config/aarch64/aarch64.c (aarch64_autovectorize_vector_sizes):
24848 * config/arc/arc.c (arc_autovectorize_vector_sizes): Likewise.
24849 * config/arm/arm.c (arm_autovectorize_vector_sizes): Likewise.
24850 * config/i386/i386.c (ix86_autovectorize_vector_sizes): Likewise.
24851 * config/mips/mips.c (mips_autovectorize_vector_sizes): Likewise.
24852 * omp-general.c (omp_max_vf): Likewise.
24853 * omp-low.c (omp_clause_aligned_alignment): Likewise.
24854 * optabs-query.c (can_vec_mask_load_store_p): Likewise.
24855 * tree-vect-loop.c (vect_analyze_loop): Likewise.
24856 * tree-vect-slp.c (vect_slp_bb): Likewise.
24857 * doc/tm.texi: Regenerate.
24858 * tree-vectorizer.h (current_vector_size): Change from an unsigned int
24860 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Take
24861 the vector size as a poly_uint64 rather than an unsigned int.
24862 (current_vector_size): Change from an unsigned int to a poly_uint64.
24863 (get_vectype_for_scalar_type): Update accordingly.
24864 * tree.h (build_truth_vector_type): Take the size and number of
24865 units as a poly_uint64 rather than an unsigned int.
24866 (build_vector_type): Add a temporary overload that takes
24867 the number of units as a poly_uint64 rather than an unsigned int.
24868 * tree.c (make_vector_type): Likewise.
24869 (build_truth_vector_type): Take the number of units as a poly_uint64
24870 rather than an unsigned int.
24872 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24873 Alan Hayward <alan.hayward@arm.com>
24874 David Sherwood <david.sherwood@arm.com>
24876 * target.def (get_mask_mode): Take the number of units and length
24877 as poly_uint64s rather than unsigned ints.
24878 * targhooks.h (default_get_mask_mode): Update accordingly.
24879 * targhooks.c (default_get_mask_mode): Likewise.
24880 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
24881 * doc/tm.texi: Regenerate.
24883 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24884 Alan Hayward <alan.hayward@arm.com>
24885 David Sherwood <david.sherwood@arm.com>
24887 * omp-general.h (omp_max_vf): Return a poly_uint64 instead of an int.
24888 * omp-general.c (omp_max_vf): Likewise.
24889 * omp-expand.c (omp_adjust_chunk_size): Update call to omp_max_vf.
24890 (expand_omp_simd): Handle polynomial safelen.
24891 * omp-low.c (omplow_simd_context): Add a default constructor.
24892 (omplow_simd_context::max_vf): Change from int to poly_uint64.
24893 (lower_rec_simd_input_clauses): Update accordingly.
24894 (lower_rec_input_clauses): Likewise.
24896 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24897 Alan Hayward <alan.hayward@arm.com>
24898 David Sherwood <david.sherwood@arm.com>
24900 * tree-vectorizer.h (vect_nunits_for_cost): New function.
24901 * tree-vect-loop.c (vect_model_reduction_cost): Use it.
24902 * tree-vect-slp.c (vect_analyze_slp_cost_1): Likewise.
24903 (vect_analyze_slp_cost): Likewise.
24904 * tree-vect-stmts.c (vect_model_store_cost): Likewise.
24905 (vect_model_load_cost): Likewise.
24907 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24908 Alan Hayward <alan.hayward@arm.com>
24909 David Sherwood <david.sherwood@arm.com>
24911 * tree-vect-slp.c (vect_record_max_nunits, vect_build_slp_tree_1)
24912 (vect_build_slp_tree_2, vect_build_slp_tree): Change max_nunits
24913 from an unsigned int * to a poly_uint64_pod *.
24914 (calculate_unrolling_factor): New function.
24915 (vect_analyze_slp_instance): Use it. Track polynomial max_nunits.
24917 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24918 Alan Hayward <alan.hayward@arm.com>
24919 David Sherwood <david.sherwood@arm.com>
24921 * tree-vectorizer.h (_slp_instance::unrolling_factor): Change
24922 from an unsigned int to a poly_uint64.
24923 (_loop_vec_info::slp_unrolling_factor): Likewise.
24924 (_loop_vec_info::vectorization_factor): Change from an int
24926 (MAX_VECTORIZATION_FACTOR): Bump from 64 to INT_MAX.
24927 (vect_get_num_vectors): New function.
24928 (vect_update_max_nunits, vect_vf_for_cost): Likewise.
24929 (vect_get_num_copies): Use vect_get_num_vectors.
24930 (vect_analyze_data_ref_dependences): Change max_vf from an int *
24931 to an unsigned int *.
24932 (vect_analyze_data_refs): Change min_vf from an int * to a
24934 (vect_transform_slp_perm_load): Take the vf as a poly_uint64 rather
24935 than an unsigned HOST_WIDE_INT.
24936 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr)
24937 (vect_analyze_data_ref_dependence): Change max_vf from an int *
24938 to an unsigned int *.
24939 (vect_analyze_data_ref_dependences): Likewise.
24940 (vect_compute_data_ref_alignment): Handle polynomial vf.
24941 (vect_enhance_data_refs_alignment): Likewise.
24942 (vect_prune_runtime_alias_test_list): Likewise.
24943 (vect_shift_permute_load_chain): Likewise.
24944 (vect_supportable_dr_alignment): Likewise.
24945 (dependence_distance_ge_vf): Take the vectorization factor as a
24946 poly_uint64 rather than an unsigned HOST_WIDE_INT.
24947 (vect_analyze_data_refs): Change min_vf from an int * to a
24949 * tree-vect-loop-manip.c (vect_gen_scalar_loop_niters): Take
24950 vfm1 as a poly_uint64 rather than an int. Make the same change
24951 for the returned bound_scalar.
24952 (vect_gen_vector_loop_niters): Handle polynomial vf.
24953 (vect_do_peeling): Likewise. Update call to
24954 vect_gen_scalar_loop_niters and handle polynomial bound_scalars.
24955 (vect_gen_vector_loop_niters_mult_vf): Assert that the vf must
24957 * tree-vect-loop.c (vect_determine_vectorization_factor)
24958 (vect_update_vf_for_slp, vect_analyze_loop_2): Handle polynomial vf.
24959 (vect_get_known_peeling_cost): Likewise.
24960 (vect_estimate_min_profitable_iters, vectorizable_reduction): Likewise.
24961 (vect_worthwhile_without_simd_p, vectorizable_induction): Likewise.
24962 (vect_transform_loop): Likewise. Use the lowest possible VF when
24963 updating the upper bounds of the loop.
24964 (vect_min_worthwhile_factor): Make static. Return an unsigned int
24965 rather than an int.
24966 * tree-vect-slp.c (vect_attempt_slp_rearrange_stmts): Cope with
24967 polynomial unroll factors.
24968 (vect_analyze_slp_cost_1, vect_analyze_slp_instance): Likewise.
24969 (vect_make_slp_decision): Likewise.
24970 (vect_supported_load_permutation_p): Likewise, and polynomial
24972 (vect_analyze_slp_cost): Handle polynomial vf.
24973 (vect_slp_analyze_node_operations): Likewise.
24974 (vect_slp_analyze_bb_1): Likewise.
24975 (vect_transform_slp_perm_load): Take the vf as a poly_uint64 rather
24976 than an unsigned HOST_WIDE_INT.
24977 * tree-vect-stmts.c (vectorizable_simd_clone_call, vectorizable_store)
24978 (vectorizable_load): Handle polynomial vf.
24979 * tree-vectorizer.c (simduid_to_vf::vf): Change from an int to
24981 (adjust_simduid_builtins, shrink_simd_arrays): Update accordingly.
24983 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24984 Alan Hayward <alan.hayward@arm.com>
24985 David Sherwood <david.sherwood@arm.com>
24987 * match.pd: Handle bit operations involving three constants
24988 and try to fold one pair.
24990 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24992 * tree-vect-loop-manip.c: Include gimple-fold.h.
24993 (slpeel_make_loop_iterate_ntimes): Add step, final_iv and
24994 niters_maybe_zero parameters. Handle other cases besides a step of 1.
24995 (vect_gen_vector_loop_niters): Add a step_vector_ptr parameter.
24996 Add a path that uses a step of VF instead of 1, but disable it
24998 (vect_do_peeling): Add step_vector, niters_vector_mult_vf_var
24999 and niters_no_overflow parameters. Update calls to
25000 slpeel_make_loop_iterate_ntimes and vect_gen_vector_loop_niters.
25001 Create a new SSA name if the latter choses to use a ste other
25002 than zero, and return it via niters_vector_mult_vf_var.
25003 * tree-vect-loop.c (vect_transform_loop): Update calls to
25004 vect_do_peeling, vect_gen_vector_loop_niters and
25005 slpeel_make_loop_iterate_ntimes.
25006 * tree-vectorizer.h (slpeel_make_loop_iterate_ntimes, vect_do_peeling)
25007 (vect_gen_vector_loop_niters): Update declarations after above changes.
25009 2018-01-02 Michael Meissner <meissner@linux.vnet.ibm.com>
25011 * config/rs6000/rs6000.md (floor<mode>2): Add support for IEEE
25012 128-bit round to integer instructions.
25013 (ceil<mode>2): Likewise.
25014 (btrunc<mode>2): Likewise.
25015 (round<mode>2): Likewise.
25017 2018-01-02 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25019 * config/rs6000/rs6000-string.c (expand_block_move): Allow the use of
25020 unaligned VSX load/store on P8/P9.
25021 (expand_block_clear): Allow the use of unaligned VSX
25022 load/store on P8/P9.
25024 2018-01-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25026 * config/rs6000/rs6000-p8swap.c (swap_feeds_both_load_and_store):
25028 (rs6000_analyze_swaps): Mark a web unoptimizable if it contains a
25029 swap associated with both a load and a store.
25031 2018-01-02 Andrew Waterman <andrew@sifive.com>
25033 * config/riscv/linux.h (ICACHE_FLUSH_FUNC): New.
25034 * config/riscv/riscv.md (clear_cache): Use it.
25036 2018-01-02 Artyom Skrobov <tyomitch@gmail.com>
25038 * web.c: Remove out-of-date comment.
25040 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25042 * expr.c (fixup_args_size_notes): Check that any existing
25043 REG_ARGS_SIZE notes are correct, and don't try to re-add them.
25044 (emit_single_push_insn_1): Move stack_pointer_delta adjustment to...
25045 (emit_single_push_insn): ...here.
25047 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25049 * rtl.h (CONST_VECTOR_ELT): Redefine to const_vector_elt.
25050 (const_vector_encoded_nelts): New function.
25051 (CONST_VECTOR_NUNITS): Redefine to use GET_MODE_NUNITS.
25052 (const_vector_int_elt, const_vector_elt): Declare.
25053 * emit-rtl.c (const_vector_int_elt_1): New function.
25054 (const_vector_elt): Likewise.
25055 * simplify-rtx.c (simplify_immed_subreg): Avoid taking the address
25056 of CONST_VECTOR_ELT.
25058 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25060 * expr.c: Include rtx-vector-builder.h.
25061 (const_vector_mask_from_tree): Use rtx_vector_builder and operate
25062 directly on the tree encoding.
25063 (const_vector_from_tree): Likewise.
25064 * optabs.c: Include rtx-vector-builder.h.
25065 (expand_vec_perm_var): Use rtx_vector_builder and create a repeating
25066 sequence of "u" values.
25067 * vec-perm-indices.c: Include rtx-vector-builder.h.
25068 (vec_perm_indices_to_rtx): Use rtx_vector_builder and operate
25069 directly on the vec_perm_indices encoding.
25071 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25073 * doc/rtl.texi (const_vector): Describe new encoding scheme.
25074 * Makefile.in (OBJS): Add rtx-vector-builder.o.
25075 * rtx-vector-builder.h: New file.
25076 * rtx-vector-builder.c: Likewise.
25077 * rtl.h (rtx_def::u2): Add a const_vector field.
25078 (CONST_VECTOR_NPATTERNS): New macro.
25079 (CONST_VECTOR_NELTS_PER_PATTERN): Likewise.
25080 (CONST_VECTOR_DUPLICATE_P): Likewise.
25081 (CONST_VECTOR_STEPPED_P): Likewise.
25082 (CONST_VECTOR_ENCODED_ELT): Likewise.
25083 (const_vec_duplicate_p): Check for a duplicated vector encoding.
25084 (unwrap_const_vec_duplicate): Likewise.
25085 (const_vec_series_p): Check for a non-duplicated vector encoding.
25086 Say that the function only returns true for integer vectors.
25087 * emit-rtl.c: Include rtx-vector-builder.h.
25088 (gen_const_vec_duplicate_1): Delete.
25089 (gen_const_vector): Call gen_const_vec_duplicate instead of
25090 gen_const_vec_duplicate_1.
25091 (const_vec_series_p_1): Operate directly on the CONST_VECTOR encoding.
25092 (gen_const_vec_duplicate): Use rtx_vector_builder.
25093 (gen_const_vec_series): Likewise.
25094 (gen_rtx_CONST_VECTOR): Likewise.
25095 * config/powerpcspe/powerpcspe.c: Include rtx-vector-builder.h.
25096 (swap_const_vector_halves): Take an rtx pointer rather than rtx.
25097 Build a new vector rather than modifying a CONST_VECTOR in-place.
25098 (handle_special_swappables): Update call accordingly.
25099 * config/rs6000/rs6000-p8swap.c: Include rtx-vector-builder.h.
25100 (swap_const_vector_halves): Take an rtx pointer rather than rtx.
25101 Build a new vector rather than modifying a CONST_VECTOR in-place.
25102 (handle_special_swappables): Update call accordingly.
25104 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25106 * simplify-rtx.c (simplify_const_binary_operation): Use
25107 CONST_VECTOR_ELT instead of XVECEXP.
25109 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25111 * tree-cfg.c (verify_gimple_assign_ternary): Allow the size of
25112 the selector elements to be different from the data elements
25113 if the selector is a VECTOR_CST.
25114 * tree-vect-stmts.c (vect_gen_perm_mask_any): Use a vector of
25115 ssizetype for the selector.
25117 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25119 * optabs.c (shift_amt_for_vec_perm_mask): Try using series_p
25120 before testing each element individually.
25121 * tree-vect-generic.c (lower_vec_perm): Likewise.
25123 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25125 * selftest.h (selftest::vec_perm_indices_c_tests): Declare.
25126 * selftest-run-tests.c (selftest::run_tests): Call it.
25127 * vector-builder.h (vector_builder::operator ==): New function.
25128 (vector_builder::operator !=): Likewise.
25129 * vec-perm-indices.h (vec_perm_indices::series_p): Declare.
25130 (vec_perm_indices::all_from_input_p): New function.
25131 * vec-perm-indices.c (vec_perm_indices::series_p): Likewise.
25132 (test_vec_perm_12, selftest::vec_perm_indices_c_tests): Likewise.
25133 * fold-const.c (fold_ternary_loc): Use tree_to_vec_perm_builder
25134 instead of reading the VECTOR_CST directly. Detect whether both
25135 vector inputs are the same before constructing the vec_perm_indices,
25136 and update the number of inputs argument accordingly. Use the
25137 utility functions added above. Only construct sel2 if we need to.
25139 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25141 * optabs.c (expand_vec_perm_var): Use an explicit encoding for
25142 the broadcast of the low byte.
25143 (expand_mult_highpart): Use an explicit encoding for the permutes.
25144 * optabs-query.c (can_mult_highpart_p): Likewise.
25145 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
25146 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
25147 (vectorizable_bswap): Likewise.
25148 * tree-vect-data-refs.c (vect_grouped_store_supported): Use an
25149 explicit encoding for the power-of-2 permutes.
25150 (vect_permute_store_chain): Likewise.
25151 (vect_grouped_load_supported): Likewise.
25152 (vect_permute_load_chain): Likewise.
25154 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25156 * vec-perm-indices.h (vec_perm_indices_to_tree): Declare.
25157 * vec-perm-indices.c (vec_perm_indices_to_tree): New function.
25158 * tree-ssa-forwprop.c (simplify_vector_constructor): Use it.
25159 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
25160 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
25161 (vect_gen_perm_mask_any): Likewise.
25163 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25165 * int-vector-builder.h: New file.
25166 * vec-perm-indices.h: Include int-vector-builder.h.
25167 (vec_perm_indices): Redefine as an int_vector_builder.
25168 (auto_vec_perm_indices): Delete.
25169 (vec_perm_builder): Redefine as a stand-alone class.
25170 (vec_perm_indices::vec_perm_indices): New function.
25171 (vec_perm_indices::clamp): Likewise.
25172 * vec-perm-indices.c: Include fold-const.h and tree-vector-builder.h.
25173 (vec_perm_indices::new_vector): New function.
25174 (vec_perm_indices::new_expanded_vector): Update for new
25175 vec_perm_indices class.
25176 (vec_perm_indices::rotate_inputs): New function.
25177 (vec_perm_indices::all_in_range_p): Operate directly on the
25178 encoded form, without computing elided elements.
25179 (tree_to_vec_perm_builder): Operate directly on the VECTOR_CST
25180 encoding. Update for new vec_perm_indices class.
25181 * optabs.c (expand_vec_perm_const): Create a vec_perm_indices for
25182 the given vec_perm_builder.
25183 (expand_vec_perm_var): Update vec_perm_builder constructor.
25184 (expand_mult_highpart): Use vec_perm_builder instead of
25185 auto_vec_perm_indices.
25186 * optabs-query.c (can_mult_highpart_p): Use vec_perm_builder and
25187 vec_perm_indices instead of auto_vec_perm_indices. Use a single
25188 or double series encoding as appropriate.
25189 * fold-const.c (fold_ternary_loc): Use vec_perm_builder and
25190 vec_perm_indices instead of auto_vec_perm_indices.
25191 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
25192 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
25193 (vect_permute_store_chain): Likewise.
25194 (vect_grouped_load_supported): Likewise.
25195 (vect_permute_load_chain): Likewise.
25196 (vect_shift_permute_load_chain): Likewise.
25197 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
25198 (vect_transform_slp_perm_load): Likewise.
25199 (vect_schedule_slp_instance): Likewise.
25200 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
25201 (vectorizable_mask_load_store): Likewise.
25202 (vectorizable_bswap): Likewise.
25203 (vectorizable_store): Likewise.
25204 (vectorizable_load): Likewise.
25205 * tree-vect-generic.c (lower_vec_perm): Use vec_perm_builder and
25206 vec_perm_indices instead of auto_vec_perm_indices. Use
25207 tree_to_vec_perm_builder to read the vector from a tree.
25208 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Take a
25209 vec_perm_builder instead of a vec_perm_indices.
25210 (have_whole_vector_shift): Use vec_perm_builder and
25211 vec_perm_indices instead of auto_vec_perm_indices. Leave the
25212 truncation to calc_vec_perm_mask_for_shift.
25213 (vect_create_epilog_for_reduction): Likewise.
25214 * config/aarch64/aarch64.c (expand_vec_perm_d::perm): Change
25215 from auto_vec_perm_indices to vec_perm_indices.
25216 (aarch64_expand_vec_perm_const_1): Use rotate_inputs on d.perm
25217 instead of changing individual elements.
25218 (aarch64_vectorize_vec_perm_const): Use new_vector to install
25219 the vector in d.perm.
25220 * config/arm/arm.c (expand_vec_perm_d::perm): Change
25221 from auto_vec_perm_indices to vec_perm_indices.
25222 (arm_expand_vec_perm_const_1): Use rotate_inputs on d.perm
25223 instead of changing individual elements.
25224 (arm_vectorize_vec_perm_const): Use new_vector to install
25225 the vector in d.perm.
25226 * config/powerpcspe/powerpcspe.c (rs6000_expand_extract_even):
25227 Update vec_perm_builder constructor.
25228 (rs6000_expand_interleave): Likewise.
25229 * config/rs6000/rs6000.c (rs6000_expand_extract_even): Likewise.
25230 (rs6000_expand_interleave): Likewise.
25232 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25234 * optabs-query.c (can_vec_perm_var_p): Check whether lowering
25235 to qimode could truncate the indices.
25236 * optabs.c (expand_vec_perm_var): Likewise.
25238 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25240 * Makefile.in (OBJS): Add vec-perm-indices.o.
25241 * vec-perm-indices.h: New file.
25242 * vec-perm-indices.c: Likewise.
25243 * target.h (vec_perm_indices): Replace with a forward class
25245 (auto_vec_perm_indices): Move to vec-perm-indices.h.
25246 * optabs.h: Include vec-perm-indices.h.
25247 (expand_vec_perm): Delete.
25248 (selector_fits_mode_p, expand_vec_perm_var): Declare.
25249 (expand_vec_perm_const): Declare.
25250 * target.def (vec_perm_const_ok): Replace with...
25251 (vec_perm_const): ...this new hook.
25252 * doc/tm.texi.in (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Replace with...
25253 (TARGET_VECTORIZE_VEC_PERM_CONST): ...this new hook.
25254 * doc/tm.texi: Regenerate.
25255 * optabs.def (vec_perm_const): Delete.
25256 * doc/md.texi (vec_perm_const): Likewise.
25257 (vec_perm): Refer to TARGET_VECTORIZE_VEC_PERM_CONST.
25258 * expr.c (expand_expr_real_2): Use expand_vec_perm_const rather than
25259 expand_vec_perm for constant permutation vectors. Assert that
25260 the mode of variable permutation vectors is the integer equivalent
25261 of the mode that is being permuted.
25262 * optabs-query.h (selector_fits_mode_p): Declare.
25263 * optabs-query.c: Include vec-perm-indices.h.
25264 (selector_fits_mode_p): New function.
25265 (can_vec_perm_const_p): Check whether targetm.vectorize.vec_perm_const
25266 is defined, instead of checking whether the vec_perm_const_optab
25267 exists. Use targetm.vectorize.vec_perm_const instead of
25268 targetm.vectorize.vec_perm_const_ok. Check whether the indices
25269 fit in the vector mode before using a variable permute.
25270 * optabs.c (shift_amt_for_vec_perm_mask): Take a mode and a
25271 vec_perm_indices instead of an rtx.
25272 (expand_vec_perm): Replace with...
25273 (expand_vec_perm_const): ...this new function. Take the selector
25274 as a vec_perm_indices rather than an rtx. Also take the mode of
25275 the selector. Update call to shift_amt_for_vec_perm_mask.
25276 Use targetm.vectorize.vec_perm_const instead of vec_perm_const_optab.
25277 Use vec_perm_indices::new_expanded_vector to expand the original
25278 selector into bytes. Check whether the indices fit in the vector
25279 mode before using a variable permute.
25280 (expand_vec_perm_var): Make global.
25281 (expand_mult_highpart): Use expand_vec_perm_const.
25282 * fold-const.c: Includes vec-perm-indices.h.
25283 * tree-ssa-forwprop.c: Likewise.
25284 * tree-vect-data-refs.c: Likewise.
25285 * tree-vect-generic.c: Likewise.
25286 * tree-vect-loop.c: Likewise.
25287 * tree-vect-slp.c: Likewise.
25288 * tree-vect-stmts.c: Likewise.
25289 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm_const):
25291 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>): Delete.
25292 * config/aarch64/aarch64.c (aarch64_expand_vec_perm_const)
25293 (aarch64_vectorize_vec_perm_const_ok): Fuse into...
25294 (aarch64_vectorize_vec_perm_const): ...this new function.
25295 (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
25296 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
25297 * config/arm/arm-protos.h (arm_expand_vec_perm_const): Delete.
25298 * config/arm/vec-common.md (vec_perm_const<mode>): Delete.
25299 * config/arm/arm.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
25300 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
25301 (arm_expand_vec_perm_const, arm_vectorize_vec_perm_const_ok): Merge
25303 (arm_vectorize_vec_perm_const): ...this new function. Explicitly
25304 check for NEON modes.
25305 * config/i386/i386-protos.h (ix86_expand_vec_perm_const): Delete.
25306 * config/i386/sse.md (VEC_PERM_CONST, vec_perm_const<mode>): Delete.
25307 * config/i386/i386.c (ix86_expand_vec_perm_const_1): Update comment.
25308 (ix86_expand_vec_perm_const, ix86_vectorize_vec_perm_const_ok): Merge
25310 (ix86_vectorize_vec_perm_const): ...this new function. Incorporate
25311 the old VEC_PERM_CONST conditions.
25312 * config/ia64/ia64-protos.h (ia64_expand_vec_perm_const): Delete.
25313 * config/ia64/vect.md (vec_perm_const<mode>): Delete.
25314 * config/ia64/ia64.c (ia64_expand_vec_perm_const)
25315 (ia64_vectorize_vec_perm_const_ok): Merge into...
25316 (ia64_vectorize_vec_perm_const): ...this new function.
25317 * config/mips/loongson.md (vec_perm_const<mode>): Delete.
25318 * config/mips/mips-msa.md (vec_perm_const<mode>): Delete.
25319 * config/mips/mips-ps-3d.md (vec_perm_constv2sf): Delete.
25320 * config/mips/mips-protos.h (mips_expand_vec_perm_const): Delete.
25321 * config/mips/mips.c (mips_expand_vec_perm_const)
25322 (mips_vectorize_vec_perm_const_ok): Merge into...
25323 (mips_vectorize_vec_perm_const): ...this new function.
25324 * config/powerpcspe/altivec.md (vec_perm_constv16qi): Delete.
25325 * config/powerpcspe/paired.md (vec_perm_constv2sf): Delete.
25326 * config/powerpcspe/spe.md (vec_perm_constv2si): Delete.
25327 * config/powerpcspe/vsx.md (vec_perm_const<mode>): Delete.
25328 * config/powerpcspe/powerpcspe-protos.h (altivec_expand_vec_perm_const)
25329 (rs6000_expand_vec_perm_const): Delete.
25330 * config/powerpcspe/powerpcspe.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK):
25332 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
25333 (altivec_expand_vec_perm_const_le): Take each operand individually.
25334 Operate on constant selectors rather than rtxes.
25335 (altivec_expand_vec_perm_const): Likewise. Update call to
25336 altivec_expand_vec_perm_const_le.
25337 (rs6000_expand_vec_perm_const): Delete.
25338 (rs6000_vectorize_vec_perm_const_ok): Delete.
25339 (rs6000_vectorize_vec_perm_const): New function.
25340 (rs6000_do_expand_vec_perm): Take a vec_perm_builder instead of
25341 an element count and rtx array.
25342 (rs6000_expand_extract_even): Update call accordingly.
25343 (rs6000_expand_interleave): Likewise.
25344 * config/rs6000/altivec.md (vec_perm_constv16qi): Delete.
25345 * config/rs6000/paired.md (vec_perm_constv2sf): Delete.
25346 * config/rs6000/vsx.md (vec_perm_const<mode>): Delete.
25347 * config/rs6000/rs6000-protos.h (altivec_expand_vec_perm_const)
25348 (rs6000_expand_vec_perm_const): Delete.
25349 * config/rs6000/rs6000.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
25350 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
25351 (altivec_expand_vec_perm_const_le): Take each operand individually.
25352 Operate on constant selectors rather than rtxes.
25353 (altivec_expand_vec_perm_const): Likewise. Update call to
25354 altivec_expand_vec_perm_const_le.
25355 (rs6000_expand_vec_perm_const): Delete.
25356 (rs6000_vectorize_vec_perm_const_ok): Delete.
25357 (rs6000_vectorize_vec_perm_const): New function. Remove stray
25358 reference to the SPE evmerge intructions.
25359 (rs6000_do_expand_vec_perm): Take a vec_perm_builder instead of
25360 an element count and rtx array.
25361 (rs6000_expand_extract_even): Update call accordingly.
25362 (rs6000_expand_interleave): Likewise.
25363 * config/sparc/sparc.md (vec_perm_constv8qi): Delete in favor of...
25364 * config/sparc/sparc.c (sparc_vectorize_vec_perm_const): ...this
25366 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
25368 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25370 * optabs.c (expand_vec_perm_1): Assert that SEL has an integer
25371 vector mode and that that mode matches the mode of the data
25373 (expand_vec_perm): Split handling of non-CONST_VECTOR selectors
25374 out into expand_vec_perm_var. Do all CONST_VECTOR handling here,
25375 directly using expand_vec_perm_1 when forcing selectors into
25377 (expand_vec_perm_var): New function, split out from expand_vec_perm.
25379 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25381 * optabs-query.h (can_vec_perm_p): Delete.
25382 (can_vec_perm_var_p, can_vec_perm_const_p): Declare.
25383 * optabs-query.c (can_vec_perm_p): Split into...
25384 (can_vec_perm_var_p, can_vec_perm_const_p): ...these two functions.
25385 (can_mult_highpart_p): Use can_vec_perm_const_p to test whether a
25386 particular selector is valid.
25387 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
25388 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
25389 (vect_grouped_load_supported): Likewise.
25390 (vect_shift_permute_load_chain): Likewise.
25391 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
25392 (vect_transform_slp_perm_load): Likewise.
25393 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
25394 (vectorizable_bswap): Likewise.
25395 (vect_gen_perm_mask_checked): Likewise.
25396 * fold-const.c (fold_ternary_loc): Likewise. Don't take
25397 implementations of variable permutation vectors into account
25398 when deciding which selector to use.
25399 * tree-vect-loop.c (have_whole_vector_shift): Don't check whether
25400 vec_perm_const_optab is supported; instead use can_vec_perm_const_p
25401 with a false third argument.
25402 * tree-vect-generic.c (lower_vec_perm): Use can_vec_perm_const_p
25403 to test whether the constant selector is valid and can_vec_perm_var_p
25404 to test whether a variable selector is valid.
25406 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25408 * optabs-query.h (can_vec_perm_p): Take a const vec_perm_indices *.
25409 * optabs-query.c (can_vec_perm_p): Likewise.
25410 * fold-const.c (fold_vec_perm): Take a const vec_perm_indices &
25411 instead of vec_perm_indices.
25412 * tree-vectorizer.h (vect_gen_perm_mask_any): Likewise,
25413 (vect_gen_perm_mask_checked): Likewise,
25414 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise,
25415 (vect_gen_perm_mask_checked): Likewise,
25417 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25419 * optabs-query.h (qimode_for_vec_perm): Declare.
25420 * optabs-query.c (can_vec_perm_p): Split out qimode search to...
25421 (qimode_for_vec_perm): ...this new function.
25422 * optabs.c (expand_vec_perm): Use qimode_for_vec_perm.
25424 2018-01-02 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25426 * rtlanal.c (canonicalize_condition): Return 0 if final rtx
25427 does not have a conditional at the top.
25429 2018-01-02 Richard Biener <rguenther@suse.de>
25431 * ipa-inline.c (big_speedup_p): Fix expression.
25433 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
25436 * config/i386/x86-tune-costs.h: Increase cost of integer load costs
25439 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
25443 * x86-tune-costs.h (generic_cost): Reduce cost of FDIV 20->17,
25444 cost of sqrt 20->14, DIVSS 18->13, DIVSD 32->17, SQRtSS 30->14
25445 and SQRTsD 58->18, cond_not_taken_branch_cost. 2->1. Increase
25446 cond_taken_branch_cost 3->4.
25448 2018-01-01 Jakub Jelinek <jakub@redhat.com>
25450 PR tree-optimization/83581
25451 * tree-loop-distribution.c (pass_loop_distribution::execute): Return
25452 TODO_cleanup_cfg if any changes have been made.
25454 PR middle-end/83608
25455 * expr.c (store_expr_with_bounds): Use simplify_gen_subreg instead of
25456 convert_modes if target mode has the right side, but different mode
25459 PR middle-end/83609
25460 * expr.c (expand_assignment): Fix up a typo in simplify_gen_subreg
25461 last argument when extracting from CONCAT. If either from_real or
25462 from_imag is NULL, use expansion through memory. If result is not
25463 a CONCAT and simplify_gen_subreg fails, try to simplify_gen_subreg
25464 the parts directly to inner mode, if even that fails, use expansion
25467 PR middle-end/83623
25468 * expmed.c (expand_shift_1): For 2-byte rotates by BITS_PER_UNIT,
25469 check for bswap in mode rather than HImode and use that in expand_unop
25472 Copyright (C) 2018 Free Software Foundation, Inc.
25474 Copying and distribution of this file, with or without modification,
25475 are permitted in any medium without royalty provided the copyright
25476 notice and this notice are preserved.