1 2018-11-12 Richard Biener <rguenther@suse.de>
3 * tree-vrp.h (value_range_base::symbolic_p,
4 value_range_base::constant_p, value_range_base::zero_p,
5 value_range_base::singleton_p): Move from value_range.
6 (value_range::dump): Add.
7 * gimple-ssa-evrp-analyze.c
8 (evrp_range_analyzer::record_ranges_from_phis): Use set_varying.
9 * ipa-cp.c (ipcp_vr_lattice::print): Use dump_value_range.
10 * tree-ssa-threadedge.c (record_temporary_equivalences_from_phis):
12 * tree-vrp.c (value_range::symbolic_p): Move to value_range_base.
13 (value_range::constant_p): Likewise.
14 (value_range::singleton_p): Likewise.
15 (value_range_base::dump): Add.
16 (set_value_range_to_undefined): Remove.
17 (set_value_range_to_varying): Likewise.
18 (range_int_cst_p): Take value_range_base argument.
19 (range_int_cst_singleton_p): Likewise.
20 (value_range_constant_singleton): Likewise.
21 (vrp_set_zero_nonzero_bits): Likewise.
22 (extract_range_from_multiplicative_op): Use set_varying.
23 (extract_range_from_binary_expr_1): Likewise. Use set_undefined.
24 (extract_range_from_unary_expr): Likewise.
25 (dump_value_range_base): Change to overload of dump_value_range.
26 (vrp_prop::vrp_initialize): Use set_varying and set_undefined.
27 (vrp_prop::visit_stmt): Likewise.
28 (value_range::intersect_helper): Likewise.
29 (value_range::union_helper): Likewise.
30 (determine_value_range_1): Likewise.
32 2018-11-12 Richard Biener <rguenther@suse.de>
34 * tree-vrp.c (set_value_range_to_nonnull): Clear equiv.
35 (set_value_range_to_null): Likewise.
36 * vr-values.c (vr_values::extract_range_from_comparison):
37 Clear equiv for constant singleton ranges.
39 2018-11-12 Wei Xiao <wei3.xiao@intel.com>
41 * config/i386/sse.md: Combine VFIXUPIMM* patterns
42 (<avx512>_fixupimm<mode>_maskz<round_saeonly_expand_name>): Update.
43 (<avx512>_fixupimm<mode><sd_maskz_name><round_saeonly_name>): Update.
44 (<avx512>_fixupimm<mode>_mask<round_saeonly_name>): Remove.
45 (avx512f_sfixupimm<mode>_maskz<round_saeonly_expand_name>): Update.
46 (avx512f_sfixupimm<mode><sd_maskz_name><round_saeonly_name>): Update.
47 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Remove.
49 2018-11-11 Sandra Loosemore <sandra@codesourcery.com>
52 * doc/extend.texi (Common Type Attributes): For the align type
53 attribute, copy language about decreasing alignment from the
54 corresponding variable attribute.
56 2018-11-11 Xianmiao Qu <xianmiao_qu@c-sky.com>
58 * config/csky/csky.md (*fpuv2_nmulsf3_1, *fpuv2_nmuldf3_1): Handle
61 2018-11-11 Sandra Loosemore <sandra@codesourcery.com>
64 * doc/invoke.texi (C++ Dialect Options): Add warning about mixing
65 -frtti and -fno-rtti code.
67 2018-11-11 Sandra Loosemore <sandra@codesourcery.com>
70 * doc/extend.texi (Other Builtins): Document probability associated
71 with __builtin_expect.
73 2018-11-11 Uros Bizjak <ubizjak@gmail.com>
76 * config/i386/i386.h (STACK_BOUNDARY): Use TARGET_64BIT_MS_ABI
77 instead of (TARGET_64BIT && ix86_abi == MS_ABI).
78 * config/i386/darwin.h (STACK_BOUNDARY): Ditto.
79 * config/i386/cygming.h (STACK_BOUNDARY): Remove.
81 2018-11-11 Xianmiao Qu <xianmiao_qu@c-sky.com>
83 * config/csky/csky-linux-elf.h (CC1_SPEC): Support -profile.
85 2018-11-11 Xianmiao Qu <xianmiao_qu@c-sky.com>
87 * config/csky/csky.h (ASM_PREFERRED_EH_DATA_FORMAT): Define.
89 2018-11-11 Richard Biener <rguenther@suse.de>
91 * tree-vrp.h (class value_range_base): New base class for
92 value_range containing all but the m_equiv member.
93 (dump_value_range_base): Add.
94 (range_includes_zero_p): Work on value_range_base.
95 * tree-vrp.c (value_range_base::set): Split out base handling
97 (value_range::set): this.
98 (value_range::set_equiv): New.
99 (value_range_base::value_range_base): New constructors.
100 (value_range_base::check): Split out base handling from...
101 (value_range::check): this.
102 (value_range::equal_p): Refactor in terms of
103 ignore_equivs_equal_p which is now member of the base.
104 (value_range_base::set_undefined): New.
105 (value_range_base::set_varying): Likewise.
106 (value_range_base::dump):Split out base handling from...
107 (value_range::dump): this.
108 (value_range_base::set_and_canonicalize): Split out base handling
110 (value_range::set_and_canonicalize): this.
111 (value_range_base::union_): New.
112 * ipa-prop.h (struct ipa_jump_func): Use value_range_base *
114 * ipa-cp.c (class ipcp_vr_lattice): Use value_range_base
115 instead of value_range everywhere.
116 (ipcp_vr_lattice::print): Use dump_value_range_base.
117 (ipcp_vr_lattice::meet_with): Adjust.
118 (ipcp_vr_lattice::meet_with_1): Likewise.
119 (ipa_vr_operation_and_type_effects): Likewise.
120 (propagate_vr_across_jump_function): Likewise.
121 * ipa-prop.c (struct ipa_vr_ggc_hash_traits): Likewise.
122 (ipa_get_value_range): Likewise.
123 (ipa_set_jfunc_vr): Likewise.
124 (ipa_compute_jump_functions_for_edge): Likewise.
126 2018-11-10 Sandra Loosemore <sandra@codesourcery.com>
129 * doc/invoke.texi (Optimize Options): Add @opindex entries
130 for the positive forms of -fno-xxx and -mno-xxx options
131 that were lacking them.
133 2018-11-10 Segher Boessenkool <segher@kernel.crashing.org>
135 * combine.c (make_more_copies): Only make an intermediate copy if the
136 dest of a move is a pseudo.
138 2018-11-09 Maya Rashish <coypu@sdf.org>
141 * config/netbsd-elf.h (NETBSD_STARTFILE_SPEC): Use crtbeginS.o for PIE.
142 (NETBSD_ENDFILE_SPEC): Use crtendS.o for PIE.
144 2018-11-09 Sandra Loosemore <sandra@codesourcery.com>
148 * doc/invoke.texi (Optimize Options): Clarify default behavior
149 for -fno-toplevel-reorder, -fno-defer-pop, and -fno-branch-count-reg.
151 2018-11-09 Bernd Edlinger <bernd.edlinger@hotmail.de>
153 PR tree-optimization/87940
154 * expr.c (string_constant): Don't strip NOPS in subexpressions.
155 Fold PLUS_EXPR correctly.
157 2018-11-09 Ilya Leoshkevich <iii@linux.ibm.com>
160 * config/s390/s390.c (s390_safe_relative_long_p): New function.
161 (annotate_constant_pool_refs): Skip insns which support
163 (annotate_constant_pool_refs_1): New helper function.
164 (find_constant_pool_ref): Skip insns which support relative
166 (find_constant_pool_ref_1): New helper function.
167 (replace_constant_pool_ref): Skip insns which support
169 (replace_constant_pool_ref_1): New helper function.
170 (s390_mainpool_start): Adapt to the new signature.
171 (s390_mainpool_finish): Likewise.
172 (s390_chunkify_start): Likewise.
173 (s390_chunkify_finish): Likewise.
174 (pass_s390_early_mach::execute): Likewise.
175 (s390_prologue_plus_offset): Likewise.
176 (s390_emit_prologue): Likewise.
177 (s390_emit_epilogue): Likewise.
179 2018-11-09 Jakub Jelinek <jakub@redhat.com>
181 * gimplify.c (gimplify_scan_omp_clauses): Call sorry_at for valid
182 but unsupported lastprivate with conditional modifier.
184 2018-11-09 Jeff Law <law@redhat.com>
186 * config/mips/mips.c (mips_loongson_ext2_prefetch_cookie): Handle
187 unused argument better. Add gcc_unreachable to silence warning.
189 2018-11-09 Martin Sebor <msebor@redhat.com>
192 * attribs.c (has_attribute): New helper function.
193 (decls_mismatched_attributes, maybe_diag_alias_attributes): Same.
194 * attribs.h (decls_mismatched_attributes): Declare.
195 * cgraphunit.c (handle_alias_pairs): Call maybe_diag_alias_attributes.
196 (maybe_diag_incompatible_alias): Use OPT_Wattribute_alias_.
197 * common.opt (-Wattribute-alias): Take an argument.
198 (-Wno-attribute-alias): New option.
199 * doc/extend.texi (Common Function Attributes): Document copy.
200 (Common Variable Attributes): Same.
201 * doc/invoke.texi (-Wmissing-attributes): Document enhancement.
202 (-Wattribute-alias): Document new option argument.
204 2018-11-09 Richard Earnshaw <rearnsha@arm.com>
206 * config/arm/parsecpu.awk (/alias/): Tighten invisible alias
207 matching criteria. Remove unused array initializer.
209 2018-11-09 Bill Schmidt <wschmidt@linux.ibm.com>
210 Jinsong Ji <jji@us.ibm.com>
212 * config/rs6000/xmmintrin.h (_mm_cvtss_si32): Fix incorrect
213 constraints by introducing a new temporary.
214 (_mm_cvtss_si64): Likewise.
216 2018-11-09 Martin Liska <mliska@suse.cz>
218 * common.opt: Add -fipa-stack-alignment flag.
219 * doc/invoke.texi: Document it.
220 * final.c (rest_of_clean_state): Guard stack
223 2018-11-09 Martin Liska <mliska@suse.cz>
225 * cgraph.h (ipa_discover_readonly_nonaddressable_vars): Rename
227 (ipa_discover_variable_flags): ... this.
228 * common.opt: Come up with new flag -fipa-reference-addressable.
229 * doc/invoke.texi: Document it.
230 * ipa-reference.c (propagate): Call the renamed fn.
231 * ipa-visibility.c (whole_program_function_and_variable_visibility):
233 * ipa.c (ipa_discover_readonly_nonaddressable_vars): Renamed to
235 (ipa_discover_variable_flags): ... this. Discover
236 non-addressable variables only with the newly added flag.
237 * opts.c: Enable the newly added flag with -O1 and higher
240 2018-11-09 David Malcolm <dmalcolm@redhat.com>
242 * json.cc (selftest::test_writing_literals): Fix comment.
244 2018-11-09 Martin Liska <mliska@suse.cz>
246 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not copy
247 string to a stack buffer.
248 (aarch64_parse_cpu): Likewise.
249 (aarch64_parse_tune): Likewise.
251 2018-11-09 Richard Biener <rguenther@suse.de>
253 PR tree-optimization/87953
254 * tree-vect-loop.c (vectorizable_reduction): For analysis
255 always pass ops[0] to vectorizable_condition.
257 2018-11-09 Stafford Horne <shorne@gmail.com>
258 Richard Henderson <rth@twiddle.net>
259 Joel Sherrill <joel@rtems.org>
261 * common/config/or1k/or1k-common.c: New file.
262 * config/or1k/*: New.
263 * config.gcc (or1k*-*-*): New.
264 * configure.ac (or1k*-*-*): New test for openrisc tls.
265 * configure: Regenerated.
266 * doc/install.texi: Document OpenRISC triplets.
267 * doc/invoke.texi: Document OpenRISC arguments.
268 * doc/md.texi: Document OpenRISC.
270 2018-11-09 Richard Earnshaw <rearnsha@arm.com>
272 * config/arm/arm-cpus.in (arm7tdmi): Add an alias for arm7tdmi-s.
273 (arm7tdmi-s): Delete CPU.
274 (arm710t): Add aliases for arm720t and arm740t.
275 (arm720t, arm740t): Delete CPUs.
276 (arm920t): Add aliases for arm920, arm922t and arm940t.
277 (arm920, arm922t, arm940t): Delete CPUs.
278 (arm10tdmi): Add alias for arm1020t.
279 (arm1020t): Delete CPU.
280 (arm9e): Add aliases for arm946e-s, arm966e-s and arm968e-s.
281 (arm946e-s, arm966e-s, arm968e-s): Delete CPUs.
282 (arm10e): Add aliases for arm1020e and arm1022e.
283 (arm1020e, arm1022e): Delete CPU.
284 * config/arm/arm.md (generic_sched): Remove entries that are now
286 (generic_vfp): Likewise.
287 * config/arm/arm1020e.md: Simplify tuning selection based on alias
289 * config/arm/arm-tune.md: Regenerated.
290 * config/arm/arm-tables.opt: Regenerated.
292 2018-11-09 Richard Biener <rguenther@suse.de>
294 PR tree-optimization/87621
295 * tree-vect-loop.c (vectorizable_reduction): Handle reduction
296 op with only phi inputs.
297 * tree-ssa-loop-ch.c: Include tree-ssa-sccvn.h.
298 (ch_base::copy_headers): Run CSE on copied loop headers.
299 (pass_ch_vect::process_loop_p): Simplify.
301 2018-11-09 Alexandre Oliva <oliva@adacore.com>
303 * config/i386/mingw32.h (LINK_SPEC_LARGE_ADDR_AWARE): Adjust
304 for non-w64 x86_64 biarch.
306 2018-11-09 Alexandre Oliva <aoliva@redhat.com>
308 PR rtl-optimization/86438
309 * compare-elim.c (try_eliminate_compare): Use SET_SRC instead
310 of in_b for the compare if in_b is SET_DEST.
313 * config/i386/i386.c (ix86_const_not_ok_for_debug_p): Reject
316 2018-11-08 Aldy Hernandez <aldyh@redhat.com>
318 * tree-vrp.c (value_range::check): Do not access internals
320 (value_range::singleton_p): Same.
321 (value_range::type): Same.
322 (vrp_finalize): Use value_range API.
324 2018-11-08 Aldy Hernandez <aldyh@redhat.com>
326 * tree-vrp.c (may_contain_p): Do not access m_min/m_max directly.
328 2018-11-08 Aldy Hernandez <aldyh@redhat.com>
330 * vr-values.c (vr_values::get_value_range): Use value_range API
331 instead of piecing together ranges.
332 (vr_values::update_value_range): Same.
334 2018-11-08 Aldy Hernandez <aldyh@redhat.com>
336 * gimple-fold.c (size_must_be_zero_p): Use value_range API instead
337 of performing ad-hoc calculations.
338 * tree-ssanames.c (set_range_info): New overloaded function
339 accepting value_range &.
340 (get_range_info): Same.
341 * tree-ssanames.h (set_range_info_raw): Remove.
342 (set_range_info): New prototype.
343 (get_range_info): Same.
344 * tree-vrp.h (value_range::null_p): Rename to zero_p.
345 * tree-vrp.c (value_range::null_p): Same.
347 2018-11-09 Jan Hubicka <jh@suse.cz>
349 * tree.c (fld_type_variant_equal_p): Test user align flag.
350 (flt_type_variant): Copy user align flag.
351 (fld_incomplete_type_of): Clear it.
353 2018-11-09 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
355 * config/arm/neon.md (div<mode>3): New pattern.
357 2018-11-08 Andi Kleen <ak@linux.intel.com>
359 * common/config/i386/i386-common.c (OPTION_MASK_ISA_PTWRITE_SET): New.
360 (OPTION_MASK_ISA_PTWRITE_UNSET): New.
361 (ix86_handle_option): Handle OPT_mptwrite.
362 * config/i386/cpuid.h (bit_PTWRITE): Add.
363 * config/i386/driver-i386.c (host_detect_local_cpu): Detect ptwrite.
364 * config/i386/i386-builtin.def (BDESC): Add ptwrite32/64.
365 * config/i386/i386-c.c (ix86_target_macros_internal): Define __PTWRITE__.
366 * config/i386/i386.c (ix86_target_string): Handle ptwrite.
367 (ix86_option_override_internal): Handle PTA_PTWRITE.
368 (ix86_valid_target_attribute_inner_p): Define ptwrite.
369 (def_builtin2): Force UINT64 to be 64bit only.
370 * config/i386/i386.h (TARGET_PTWRITE): Add.
371 (TARGET_PTWRITE_P): Add.
373 * config/i386/i386.md: Define ptwrite.
374 * config/i386/i386.opt: Add -mptwrite.
375 * config/i386/immintrin.h (_ptwrite64): Add.
377 * doc/extend.texi: Document __builtin_ia32_ptwrite*.
378 * doc/invoke.texi: Document -mptwrite.
380 2018-11-08 Peter Bergner <bergner@linux.ibm.com>
382 PR rtl-optimization/87600
383 * cfgexpand.c (expand_asm_stmt): Catch illegal asm constraint usage.
384 * lra-constraints.c (process_alt_operands): Skip illegal hard
385 register usage. Prefer reloading non hard register operands.
387 2018-11-08 Sandra Loosemore <sandra@codesourcery.com>
390 * doc/invoke.texi (Optimize Options): Clarify default behavior
391 for -fno-sched-interblock and -fno-sched-spec.
393 2018-11-08 Roman Geissler <roman.geissler@amadeus.com>
395 * collect2.c (linker_select): Add USE_LLD_LD.
396 (ld_suffixes): Add ld.lld.
397 (main): Handle -fuse-ld=lld.
398 * common.opt (-fuse-ld=lld): New option.
399 * doc/invoke.texi (-fuse-ld=lld): Document.
400 * opts.c (common_handle_option): Handle OPT_fuse_ld_lld.
402 2018-11-08 Paul Koning <ni1d@arrl.net>
404 * config/pdp11/constraints.md: Add "Z" series constraints for use
405 with pre-dec and post-inc addressing.
406 * config/pdp11/pdp11-protos.m (expand_block_move): Delete.
407 (pdp11_expand_operands): Add int argument (word count).
408 (pdp11_sp_frame_offset): Delete.
409 (pdp11_cmp_length): New function.
410 (pushpop_regeq): New function.
411 * config/pdp11/pdp11.c (TARGET_STACK_PROTECT_RUNTIME_ENABLED_P):
413 (pdp11_expand_prologue, pdp11_expand_epilogue): Rewrite for new
415 (pdp11_initial_elimination_offset): Ditto.
416 (pdp11_expand_operands): Add word count argument. Bugfixes.
417 (output_move_multiple): Change how pointer adjustment is done.
418 (pdp11_gen_int_label): Correct format.
419 (output_ascii): Ditto.
420 (pdp11_asm_output_var): Add code for DEC assembler case.
421 (pdp11_asm_print_operand): Bugfix for CONST_DOUBLE holding integer
423 (legitimate_const_double_p): Ditto.
424 (pdp11_register_move_cost): Adjust for new register classes.
425 (pdp11_regno_reg_class): Ditto.
426 (expand_block_move): Delete.
427 (pushpop_regeq): New function.
428 (pdp11_legitimate_address_p): Bugfix in check for constant
430 (pdp11_sp_frame_offset): Delete.
431 (pdp11_reg_save_size): New helper function for new frame layout.
432 (output_addr_const_pdp11): Remove CONST_DOUBLE case.
433 (pdp11_expand_shift): Bugfix in check for constant shift count.
434 (pdp11_shift_length): Ditto.
435 (pdp11_assemble_shift): Copy input to pdp11_expand_operands.
436 (pdp11_cmp_length): New function.
437 * config/pdp11/pdp11.h (TARGET_CPU_CPP_BUILTINS): Add macros for
438 some compile options.
439 (FIXED_REGISTERS): Remove HARD_FRAME_POINTER_REGNUM.
440 (CALL_USED_REGISTERS): Ditto.
441 (ELIMINABLE_REGS): Ditto.
442 (REGISTER_NAMES): Ditto.
443 (reg_class): Add classes NOTR0_REG through NOTSP_REG for use by Z
445 (REG_CLASS_NAMES): Ditto.
446 (REG_CLASS_CONTENTS): Ditto. Also remove
447 HARD_FRAME_POINTER_REGNUM.
448 (CPU_REG_CLASS): New macro.
449 (CLASS_MAX_NREGS): Adjust for new register classes.
450 (FUNCTION_PROFILER): Make no-op.
451 (may_call_alloca): Remove unused declaration.
452 (ASM_OUTPUT_ALIGN): Add workaround for PR87795.
453 (ASM_OUTPUT_SKIP): Fix format.
454 * config/pdp11/pdp11.md (unspecv): Add UNSPECV_MOVMEM.
455 (HARD_FRAME_POINTER_REGNUM): Remove.
457 (*rts): Rename. Remove epilogue related checks.
458 (cmpsi, cmpdi): New insn.
459 (cbranch<mode>4): Change to apply to SI and DI modes as well.
460 (mov<mode>): Change constraints to enforce that push/pop
461 destination cannot use the same register as source.
462 (*mov<mode><cc_cc>): Ditto.
463 (movmemhi, movmemhi1, movmemhi_nocc): Change to expand block move
464 at assembly output rather than as RTL expander.
465 (zero_extendqihi2): Bugfix in check for same registers.
466 (adddi3_nocc): Bugfix in check for constant operand.
467 (addsi3_nocc): Ditto.
468 (subdi3_nocc): Ditto.
469 (subsi3_nocc): Ditto.
470 (negdi2_nocc): Copy input to pdp11_expand_operands.
471 (negsi2_nocc): Ditto.
472 (bswap2_nocc): Ditto.
473 * config/pdp11/pdp11.opt (mlra): Fix documentation.
474 * config/pdp11/t-pdp11: Use -Os.
476 2018-11-08 Richard Earnshaw <rearnsha@arm.com>
478 * config/arm/parsecpu.awk (/alias/): New parsing rule.
479 (/begin cpu/): Check that the cpu name hasn't been previously defined.
480 (gen_comm_data): Print out CPU alias tables.
481 (check_cpu): Match aliases when checking the CPU name.
482 * config/arm/arm-protos.h (cpu_alias): New structure.
483 (cpu_option): Add entry for aliases.
484 * config/arm/arm-cpus.in (strongarm): Add aliases for strongarm110
485 strongarm1100 and strongarm1110.
486 (strongarm110, strongarm1100, strongarm1110): Delete CPU entries.
487 (config/arm/arm-generic.md): Remove redundant references to
488 strongarm110, strongarm1100 and strongarm1110.
489 * common/config/arm/arm-common.c (arm_print_hint_for_cpu_option):
490 Scan aliases for additional hints.
491 (arm_parse_cpu_option_name): Also match a cpu name against the list
493 * config/arm/arm-tables.opt: Regenerated.
494 * config/arm/arm-tune.md: Regenerated.
496 2018-11-08 Jakub Jelinek <jakub@redhat.com>
498 * builtin-types.def (BT_FN_VOID_BOOL, BT_FN_VOID_SIZE_SIZE_PTR,
499 BT_FN_UINT_UINT_PTR_PTR, BT_FN_UINT_OMPFN_PTR_UINT_UINT,
500 BT_FN_BOOL_UINT_LONGPTR_LONG_LONG_LONGPTR_LONGPTR_PTR_PTR,
501 BT_FN_BOOL_UINT_ULLPTR_LONG_ULL_ULLPTR_ULLPTR_PTR_PTR,
502 BT_FN_BOOL_LONG_LONG_LONG_LONG_LONG_LONGPTR_LONGPTR_PTR_PTR,
503 BT_FN_BOOL_BOOL_ULL_ULL_ULL_LONG_ULL_ULLPTR_ULLPTR_PTR_PTR): New.
504 * gengtype.c (open_base_files): Add omp-general.h.
505 * gimple.c (gimple_build_omp_critical):
506 (gimple_build_omp_taskgroup): Add CLAUSES argument. Call
507 gimple_omp_taskgroup_set_clauses.
508 (gimple_build_omp_atomic_load): Add mo argument, call
509 gimple_omp_atomic_set_memory_order.
510 (gimple_build_omp_atomic_store): Likewise.
511 (gimple_copy): Adjust handling of GIMPLE_OMP_TASKGROUP.
512 * gimple.def (GIMPLE_OMP_TASKGROUP): Use GSS_OMP_SINGLE_LAYOUT
514 (GIMPLE_OMP_TEAMS): Use GSS_OMP_PARALLEL_LAYOUT instead
515 of GSS_OMP_SINGLE_LAYOUT, adjust comments.
516 * gimple.h (enum gf_mask): Add GF_OMP_TEAMS_HOST, GF_OMP_TASK_TASKWAIT
517 and GF_OMP_ATOMIC_MEMORY_ORDER. Remove GF_OMP_ATOMIC_SEQ_CST, use
518 different value for GF_OMP_ATOMIC_NEED_VALUE.
519 (struct gimple_statement_omp_taskreg): Add GIMPLE_OMP_TEAMS to
521 (struct gimple_statement_omp_single_layout): And remove here.
522 (struct gomp_teams): Inherit from gimple_statement_omp_taskreg rather
523 than gimple_statement_omp_single_layout.
524 (is_a_helper <gimple_statement_omp_taskreg *>::test): Allow
526 (is_a_helper <const gimple_statement_omp_taskreg *>::test): Likewise.
527 (gimple_omp_subcode): Formatting fix.
528 (gimple_omp_teams_child_fn, gimple_omp_teams_child_fn_ptr,
529 gimple_omp_teams_set_child_fn, gimple_omp_teams_data_arg,
530 gimple_omp_teams_data_arg_ptr, gimple_omp_teams_set_data_arg,
531 gimple_omp_teams_host, gimple_omp_teams_set_host,
532 gimple_omp_task_taskwait_p, gimple_omp_task_set_taskwait_p,
533 gimple_omp_taskgroup_clauses, gimple_omp_taskgroup_clauses_ptr,
534 gimple_omp_taskgroup_set_clauses): New inline functions.
535 (gimple_build_omp_atomic_load): Add enum omp_memory_order argument.
536 (gimple_build_omp_atomic_store): Likewise.
537 (gimple_omp_atomic_seq_cst_p): Remove.
538 (gimple_omp_atomic_memory_order): New function.
539 (gimple_omp_atomic_set_seq_cst): Remove.
540 (gimple_omp_atomic_set_memory_order): New function.
541 (gimple_build_omp_taskgroup): Add clauses argument.
542 * gimple-pretty-print.c (dump_gimple_omp_taskgroup): New function.
543 (dump_gimple_omp_task): Print taskwait with depend clauses.
544 (dump_gimple_omp_atomic_load, dump_gimple_omp_atomic_store): Use
545 dump_omp_atomic_memory_order.
546 (pp_gimple_stmt_1): Handle GIMPLE_OMP_TASKGROUP.
547 * gimplify.c (enum gimplify_omp_var_data): Add GOVD_MAP_ALLOC_ONLY,
548 GOVD_MAP_FROM_ONLY and GOVD_NONTEMPORAL.
549 (enum omp_region_type): Reserve bits 1 and 2 for auxiliary flags,
550 renumber values of most of ORT_* enumerators, add ORT_HOST_TEAMS,
551 ORT_COMBINED_HOST_TEAMS, ORT_TASKGROUP, ORT_TASKLOOP and
552 ORT_UNTIED_TASKLOOP enumerators.
553 (enum gimplify_defaultmap_kind): New.
554 (struct gimplify_omp_ctx): Remove target_map_scalars_firstprivate and
555 target_map_pointers_as_0len_arrays members, add defaultmap.
556 (new_omp_context): Initialize defaultmap member.
557 (gimple_add_tmp_var): Handle ORT_TASKGROUP like ORT_WORKSHARE.
558 (maybe_fold_stmt): Don't fold even in host teams regions.
559 (omp_firstprivatize_variable): Handle ORT_TASKGROUP like
560 ORT_WORKSHARE. Test ctx->defaultmap[GDMK_SCALAR] instead of
561 ctx->omp_firstprivatize_variable.
562 (omp_add_variable): Don't add private/firstprivate for VLAs in
564 (omp_default_clause): Print "taskloop" rather than "task" if
566 (omp_notice_variable): Handle ORT_TASKGROUP like ORT_WORKSHARE.
567 Handle new defaultmap clause kinds.
568 (omp_is_private): Handle ORT_TASKGROUP like ORT_WORKSHARE. Allow simd
569 iterator to be lastprivate or private. Fix up diagnostics if linear
570 is used on collapse>1 simd iterator.
571 (omp_check_private): Handle ORT_TASKGROUP like ORT_WORKSHARE.
572 (gimplify_omp_depend): New function.
573 (gimplify_scan_omp_clauses): Add shared clause on parallel for
574 combined parallel master taskloop{, simd} if taskloop has
575 firstprivate, lastprivate or reduction clause. Handle
576 OMP_CLAUSE_REDUCTION_TASK diagnostics. Adjust tests for
577 ORT_COMBINED_TEAMS. Gimplify depend clauses with iterators. Handle
578 cancel and simd OMP_CLAUSE_IF_MODIFIERs. Handle
579 OMP_CLAUSE_NONTEMPORAL. Handle new defaultmap clause kinds. Handle
580 OMP_CLAUSE_{TASK,IN}_REDUCTION. Diagnose invalid conditional
582 (gimplify_adjust_omp_clauses_1): Ignore GOVD_NONTEMPORAL. Handle
583 GOVD_MAP_ALLOC_ONLY and GOVD_MAP_FROM_ONLY.
584 (gimplify_adjust_omp_clauses): Handle OMP_CLAUSE_NONTEMPORAL. Handle
585 OMP_CLAUSE_{TASK,IN}_REDUCTION.
586 (gimplify_omp_task): Handle taskwait with depend clauses.
587 (gimplify_omp_for): Add shared clause on parallel for combined
588 parallel master taskloop{, simd} if taskloop has firstprivate,
589 lastprivate or reduction clause. Use ORT_TASKLOOP or
590 ORT_UNTIED_TASKLOOP instead of ORT_TASK or ORT_UNTIED_TASK. Adjust
591 tests for ORT_COMBINED_TEAMS. Handle C++ range for loops with
592 NULL TREE_PURPOSE in OMP_FOR_ORIG_DECLS. Firstprivatize
593 __for_end and __for_range temporaries on OMP_PARALLEL for
594 distribute parallel for{, simd}. Move OMP_CLAUSE_REDUCTION
595 and OMP_CLAUSE_IN_REDUCTION from taskloop to the task construct
596 sandwiched in between two taskloops.
597 (computable_teams_clause): Test ctx->defaultmap[GDMK_SCALAR]
598 instead of ctx->omp_firstprivatize_variable.
599 (gimplify_omp_workshare): Set ort to ORT_HOST_TEAMS or
600 ORT_COMBINED_HOST_TEAMS if not inside of target construct. If
601 host teams, use gimplify_and_return_first etc. for body like
602 for target or target data constructs, and at the end call
603 gimple_omp_teams_set_host on the GIMPLE_OMP_TEAMS object.
604 (gimplify_omp_atomic): Use OMP_ATOMIC_MEMORY_ORDER instead
605 of OMP_ATOMIC_SEQ_CST, pass it as new argument to
606 gimple_build_omp_atomic_load and gimple_build_omp_atomic_store, remove
607 gimple_omp_atomic_set_seq_cst calls.
608 (gimplify_expr) <case OMP_TASKGROUP>: Move handling into a separate
609 case, handle taskgroup clauses.
610 * lto-streamer-out.c (hash_tree): Handle
611 OMP_CLAUSE_{TASK,IN}_REDUCTION.
612 * Makefile.in (GTFILES): Add omp-general.h.
613 * omp-builtins.def (BUILT_IN_GOMP_TASKWAIT_DEPEND,
614 BUILT_IN_GOMP_LOOP_NONMONOTONIC_RUNTIME_START,
615 BUILT_IN_GOMP_LOOP_MAYBE_NONMONOTONIC_RUNTIME_START,
616 BUILT_IN_GOMP_LOOP_START, BUILT_IN_GOMP_LOOP_ORDERED_START,
617 BUILT_IN_GOMP_LOOP_DOACROSS_START,
618 BUILT_IN_GOMP_LOOP_NONMONOTONIC_RUNTIME_NEXT,
619 BUILT_IN_GOMP_LOOP_MAYBE_NONMONOTONIC_RUNTIME_NEXT,
620 BUILT_IN_GOMP_LOOP_ULL_NONMONOTONIC_RUNTIME_START,
621 BUILT_IN_GOMP_LOOP_ULL_MAYBE_NONMONOTONIC_RUNTIME_START,
622 BUILT_IN_GOMP_LOOP_ULL_START, BUILT_IN_GOMP_LOOP_ULL_ORDERED_START,
623 BUILT_IN_GOMP_LOOP_ULL_DOACROSS_START,
624 BUILT_IN_GOMP_LOOP_ULL_NONMONOTONIC_RUNTIME_NEXT,
625 BUILT_IN_GOMP_LOOP_ULL_MAYBE_NONMONOTONIC_RUNTIME_NEXT,
626 BUILT_IN_GOMP_PARALLEL_LOOP_NONMONOTONIC_RUNTIME,
627 BUILT_IN_GOMP_PARALLEL_LOOP_MAYBE_NONMONOTONIC_RUNTIME,
628 BUILT_IN_GOMP_PARALLEL_REDUCTIONS, BUILT_IN_GOMP_SECTIONS2_START,
629 BUILT_IN_GOMP_TEAMS_REG, BUILT_IN_GOMP_TASKGROUP_REDUCTION_REGISTER,
630 BUILT_IN_GOMP_TASKGROUP_REDUCTION_UNREGISTER,
631 BUILT_IN_GOMP_TASK_REDUCTION_REMAP,
632 BUILT_IN_GOMP_WORKSHARE_TASK_REDUCTION_UNREGISTER): New builtins.
633 * omp-expand.c (workshare_safe_to_combine_p): Return false for
634 non-worksharing loops.
635 (omp_adjust_chunk_size): Don't adjust anything if chunk_size is zero.
636 (determine_parallel_type): Don't combine parallel with worksharing
637 which has _reductemp_ clause.
638 (expand_parallel_call): Emit the GOMP_*nonmonotonic_runtime* or
639 GOMP_*maybe_nonmonotonic_runtime* builtins instead of GOMP_*runtime*
640 if there is nonmonotonic modifier or if there is no modifier and no
641 ordered clause. For dynamic and guided schedule without monotonic
642 and nonmonotonic modifier, default to nonmonotonic.
643 (expand_omp_for): Likewise. Adjust expand_omp_for_generic caller, use
644 GOMP_loop{,_ull}{,_ordered,_doacross}_start builtins if there are
646 (expand_task_call): Add GOMP_TASK_FLAG_REDUCTION flag to flags if
647 there are any reduction clauses.
648 (expand_taskwait_call): New function.
649 (expand_teams_call): New function.
650 (expand_omp_taskreg): Allow GIMPLE_OMP_TEAMS and call
651 expand_teams_call for it. Formatting fix. Handle taskwait with
653 (expand_omp_for_generic): Add SCHED_ARG argument. Handle expansion
654 of worksharing loops with task reductions.
655 (expand_omp_for_static_nochunk, expand_omp_for_static_chunk): Handle
656 expansion of worksharing loops with task reductions.
657 (expand_omp_sections): Handle expansion of sections with task
659 (expand_omp_synch): For host teams call expand_omp_taskreg.
660 (omp_memory_order_to_memmodel): New function.
661 (expand_omp_atomic_load, expand_omp_atomic_store,
662 expand_omp_atomic_fetch_op): Use it and gimple_omp_atomic_memory_order
663 instead of gimple_omp_atomic_seq_cst_p.
664 (build_omp_regions_1, omp_make_gimple_edges): Treat taskwait with
665 depend clauses as a standalone directive.
666 * omp-general.c (enum omp_requires): New variable.
667 (omp_extract_for_data): Initialize have_reductemp member. Allow
668 NE_EXPR even in OpenMP loops, transform them into LT_EXPR or
669 GT_EXPR loops depending on incr sign. Formatting fixes.
670 * omp-general.h (struct omp_for_data): Add have_reductemp member.
671 (enum omp_requires): New enum.
672 (omp_requires_mask): Declare.
673 * omp-grid.c (grid_eliminate_combined_simd_part): Formatting fix.
675 * omp-low.c (struct omp_context): Add task_reductions and
676 task_reduction_map fields.
677 (is_host_teams_ctx): New function.
678 (is_taskreg_ctx): Return true also if is_host_teams_ctx.
679 (use_pointer_for_field): Use is_global_var instead of
680 TREE_STATIC || DECL_EXTERNAL, and apply only if not privatized
682 (build_outer_var_ref): Ignore taskgroup outer contexts.
683 (delete_omp_context): Release task_reductions and task_reduction_map.
684 (scan_sharing_clauses): Don't add any fields for reduction clause on
685 taskloop. Handle OMP_CLAUSE__REDUCTEMP_. Handle
686 OMP_CLAUSE_{IN,TASK}_REDUCTION and OMP_CLAUSE_REDUCTION with task
687 modifier. Don't ignore shared clauses in is_host_teams_ctx contexts.
688 Handle OMP_CLAUSE_NONTEMPORAL.
689 (add_taskreg_looptemp_clauses): Add OMP_CLAUSE__REDUCTEMP_ clause if
691 (scan_omp_parallel): Add _reductemp_ clause if there are any reduction
692 clauses with task modifier.
693 (scan_omp_task): Handle taskwait with depend clauses.
694 (finish_taskreg_scan): Move field corresponding to _reductemp_ clause
695 first. Move also OMP_CLAUSE__REDUCTEMP_ clause in front if present.
696 Handle GIMPLE_OMP_TEAMS like GIMPLE_OMP_PARALLEL.
697 (scan_omp_for): Fix comment formatting.
698 (scan_omp_teams): Handle host teams constructs.
699 (check_omp_nesting_restrictions): Allow teams with no outer
700 OpenMP context. Adjust diagnostics for teams strictly nested into
701 some explicit OpenMP construct other than target. Allow OpenMP atomics
702 inside of simd regions.
703 (scan_omp_1_stmt): Call scan_sharing_clauses for taskgroups.
704 (scan_omp_1_stmt) <case GIMPLE_OMP_TEAMS>: Temporarily bump
705 taskreg_nesting_level while scanning host teams construct.
706 (task_reduction_read): New function.
707 (lower_rec_input_clauses): Handle OMP_CLAUSE_REDUCTION on taskloop
708 construct. Handle OMP_CLAUSE_IN_REDUCTION and OMP_CLAUSE__REDUCTEMP_
709 clauses. Handle OMP_CLAUSE_REDUCTION with task modifier. Remove
710 second argument create_tmp_var if it is NULL. Don't ignore shared
711 clauses in is_host_teams_ctx contexts. Handle
712 OMP_CLAUSE_FIRSTPRIVATE_NO_REFERENCE on OMP_CLAUSE_FIRSTPRIVATE
714 (lower_reduction_clauses): Ignore reduction clauses with task
715 modifier. Remove second argument create_tmp_var if it is NULL.
716 Initialize OMP_ATOMIC_MEMORY_ORDER to relaxed.
717 (lower_send_clauses): Ignore reduction clauses with task modifier.
718 Handle OMP_CLAUSE__REDUCTEMP_. Don't send anything for
719 OMP_CLAUSE_REDUCTION on taskloop. Handle OMP_CLAUSE_IN_REDUCTION.
720 (maybe_add_implicit_barrier_cancel): Add OMP_RETURN argument, don't
721 rely that it is the last stmt in body so far. Ignore outer taskgroup
723 (omp_task_reductions_find_first, omp_task_reduction_iterate,
724 lower_omp_task_reductions): New functions.
725 (lower_omp_sections): Handle reduction clauses with taskgroup
726 modifiers. Adjust maybe_add_implicit_barrier_cancel caller.
727 (lower_omp_single): Adjust maybe_add_implicit_barrier_cancel caller.
728 (lower_omp_for): Likewise. Handle reduction clauses with taskgroup
730 (lower_omp_taskgroup): Handle taskgroup reductions.
731 (create_task_copyfn): Copy over OMP_CLAUSE__REDUCTEMP_ pointer.
732 Handle OMP_CLAUSE_IN_REDUCTION and OMP_CLAUSE_REDUCTION clauses.
733 (lower_depend_clauses): If there are any
734 OMP_CLAUSE_DEPEND_DEPOBJ or OMP_CLAUSE_DEPEND_MUTEXINOUTSET
735 depend clauses, use a new array format. If OMP_CLAUSE_DEPEND_LAST is
736 seen, assume lowering is done already and return early. Set kind
737 on artificial depend clause to OMP_CLAUSE_DEPEND_LAST.
738 (lower_omp_taskreg): Handle reduction clauses with task modifier on
739 parallel construct. Handle reduction clause on taskloop construct.
740 Handle taskwait with depend clauses.
741 (lower_omp_1): Use lower_omp_taskreg instead of lower_omp_teams
742 for host teams constructs.
743 * tree.c (omp_clause_num_ops): Add in_reduction, task_reduction,
744 nontemporal and _reductemp_ clause entries.
745 (omp_clause_code_name): Likewise.
746 (walk_tree_1): Handle OMP_CLAUSE_{IN,TASK}_REDUCTION,
747 OMP_CLAUSE_NONTEMPORAL and OMP_CLAUSE__REDUCTEMP_.
748 * tree-core.h (enum omp_clause_code): Add
749 OMP_CLAUSE_{{IN,TASK}_REDUCTION,NONTEMPORAL,_REDUCTEMP_}.
750 (enum omp_clause_defaultmap_kind, enum omp_memory_order): New.
751 (struct tree_base): Add omp_atomic_memory_order field into union.
752 Remove OMP_ATOMIC_SEQ_CST comment.
753 (enum omp_clause_depend_kind): Add OMP_CLAUSE_DEPEND_MUTEXINOUTSET
754 and OMP_CLAUSE_DEPEND_DEPOBJ.
755 (struct tree_omp_clause): Add subcode.defaultmap_kind.
756 * tree.def (OMP_TASKGROUP): Add another operand, move next to other
757 OpenMP constructs with body and clauses operands.
758 * tree.h (OMP_BODY): Use OMP_MASTER instead of OMP_TASKGROUP.
759 (OMP_CLAUSES): Use OMP_TASKGROUP instead of OMP_SINGLE.
760 (OMP_TASKGROUP_CLAUSES): Define.
761 (OMP_CLAUSE_DECL): Use OMP_CLAUSE__REDUCTEMP_ instead of
762 OMP_CLAUSE__LOOPTEMP_.
763 (OMP_ATOMIC_SEQ_CST): Remove.
764 (OMP_ATOMIC_MEMORY_ORDER, OMP_CLAUSE_FIRSTPRIVATE_NO_REFERENCE,
765 OMP_CLAUSE_LASTPRIVATE_CONDITIONAL): Define.
766 (OMP_CLAUSE_REDUCTION_CODE, OMP_CLAUSE_REDUCTION_INIT,
767 OMP_CLAUSE_REDUCTION_MERGE, OMP_CLAUSE_REDUCTION_PLACEHOLDER,
768 OMP_CLAUSE_REDUCTION_DECL_PLACEHOLDER,
769 OMP_CLAUSE_REDUCTION_OMP_ORIG_REF): Handle
770 OMP_CLAUSE_{,IN_,TASK_}REDUCTION.
771 (OMP_CLAUSE_REDUCTION_TASK, OMP_CLAUSE_REDUCTION_INSCAN,
772 OMP_CLAUSE_DEFAULTMAP_KIND, OMP_CLAUSE_DEFAULTMAP_CATEGORY,
773 OMP_CLAUSE_DEFAULTMAP_BEHAVIOR, OMP_CLAUSE_DEFAULTMAP_SET_KIND):
775 * tree-inline.c (remap_gimple_stmt): Remap taskgroup clauses.
776 * tree-nested.c (convert_nonlocal_omp_clauses): Handle
777 OMP_CLAUSE__REDUCTEMP_, OMP_CLAUSE_NONTEMPORAL.
778 (convert_local_omp_clauses): Likewise. Remove useless test.
779 * tree-parloops.c (create_call_for_reduction_1): Pass
780 OMP_MEMORY_ORDER_RELAXED as new argument to
781 dump_gimple_omp_atomic_load and dump_gimple_omp_atomic_store.
782 * tree-pretty-print.c (dump_omp_iterators): New function.
783 (dump_omp_clause): Handle OMP_CLAUSE__REDUCTEMP_,
784 OMP_CLAUSE_NONTEMPORAL, OMP_CLAUSE_{TASK,IN}_REDUCTION. Print
785 reduction modifiers. Handle OMP_CLAUSE_DEPEND_DEPOBJ and
786 OMP_CLAUSE_DEPEND_MUTEXINOUTSET. Print iterators in depend clauses.
787 Print __internal__ for OMP_CLAUSE_DEPEND_LAST. Handle cancel and
788 simd OMP_CLAUSE_IF_MODIFIERs. Handle new kinds of
789 OMP_CLAUSE_DEFAULTMAP. Print conditional: for
790 OMP_CLAUSE_LASTPRIVATE_CONDITIONAL.
791 (dump_omp_atomic_memory_order): New function.
792 (dump_generic_node): Use it. Print taskgroup clauses. Print
793 taskwait with depend clauses.
794 * tree-pretty-print.h (dump_omp_atomic_memory_order): Declare.
795 * tree-streamer-in.c (unpack_ts_omp_clause_value_fields):
796 Handle OMP_CLAUSE_{TASK,IN}_REDUCTION.
797 * tree-streamer-out.c (pack_ts_omp_clause_value_fields,
798 write_ts_omp_clause_tree_pointers): Likewise.
800 2018-11-08 David Malcolm <dmalcolm@redhat.com>
803 * doc/invoke.texi (-fdump-ipa-): Document the "-optimized",
804 "-missed", "-note", and "-all" sub-options.
805 * ipa-inline.c (caller_growth_limits): Port from fprintf to dump
807 (can_early_inline_edge_p): Likewise.
808 (want_early_inline_function_p): Likewise.
809 (want_inline_self_recursive_call_p): Likewise.
810 (recursive_inlining): Likewise.
811 (inline_small_functions): Likewise.
812 (flatten_function): Likewise.
813 (ipa_inline): Likewise.
814 (inline_always_inline_functions): Likewise.
815 (early_inline_small_functions): Likewise.
816 (early_inliner): Likewise.
817 * tree-inline.c (expand_call_inline): Likewise.
819 2018-11-08 David Malcolm <dmalcolm@redhat.com>
821 * pretty-print.c (pp_format): Handle %f.
822 (selftest::test_pp_format): Add test of %f.
823 * pretty-print.h (pp_double): New macro.
825 2018-11-08 David Malcolm <dmalcolm@redhat.com>
827 * dump-context.h (ASSERT_IS_CGRAPH_NODE): New macro.
828 * dumpfile.c (make_item_for_dump_cgraph_node): Move to before...
829 (dump_pretty_printer::decode_format): Implement "%C" for
831 (selftest::test_capture_of_dump_calls): Rename "where" to
832 "stmt_loc". Convert test_decl to a function decl and set its
833 location. Add a symbol_table_test RAII instance and a
834 cgraph_node, using it to test "%C" and dump_symtab_node.
836 2018-11-08 Eric Botcazou <ebotcazou@adacore.com>
839 * cgraphclones.c (duplicate_thunk_for_node): Also set DECL_IGNORED_P.
841 2018-11-08 David Malcolm <dmalcolm@redhat.com>
843 * cgraph.c: Include "selftest.h".
844 (saved_symtab): New variable.
845 (selftest::symbol_table_test::symbol_table_test): New ctor.
846 (selftest::symbol_table_test::~symbol_table_test): New dtor.
847 (selftest::test_symbol_table_test): New test.
848 (selftest::cgraph_c_tests): New.
849 * cgraph.h (saved_symtab): New decl.
850 (selftest::symbol_table_test): New class.
851 * selftest-run-tests.c (selftest::run_tests): Call
852 selftest::cgraph_c_tests.
853 * selftest.h (selftest::cgraph_c_tests): New decl.
855 2018-11-08 Richard Biener <rguenther@suse.de>
857 * tree-data-ref.h (lambda_int): New typedef.
858 (lambda_vector_gcd): Adjust.
859 (lambda_vector_new): Likewise.
860 (lambda_matrix_new): Likewise.
861 * tree-data-ref.c (print_lambda_vector): Adjust.
863 2018-11-08 Richard Biener <rguenther@suse.de>
865 PR tree-optimization/87929
866 * tree-complex.c (expand_complex_comparison): Clean EH.
868 2018-11-08 Martin Liska <mliska@suse.cz>
870 * doc/extend.texi: Reword.
871 * predict.c (expr_expected_value_1): Likewise.
873 2018-11-08 Richard Biener <rguenther@suse.de>
875 PR tree-optimization/87913
876 * tree-ssa-phiopt.c (minmax_replacement): Turn EQ/NE compares
877 of extreme values to ordered comparisons.
879 2018-11-07 Sandra Loosemore <sandra@codesourcery.com>
882 * doc/invoke.texi (Code Gen Options): Clarify -fno-common behavior.
884 2018-11-07 Sandra Loosemore <sandra@codesourcery.com>
886 * doc/invoke.texi: Remove leading dash from @opindex entries
889 2018-11-07 Sandra Loosemore <sandra@codesourcery.com>
892 * doc/invoke.texi (Option Summary): Add -e and --entry.
893 (Link Options): Likewise.
895 2018-11-07 Nathan Sidwell <nathan@acm.org>
898 * Makefile.in (bitmap.o-warn): Add -Wno-error to unbreak
899 --disable-checking bootstrap.
901 2018-11-07 Hafiz Abid Qadeer <abidh@codesourcery.com>
903 * configure: Regenerated.
905 2018-11-07 Jozef Lawrynowicz <jozef.l@mittosystems.com>
908 * stor-layout.c (compute_record_mode): Set TYPE_MODE of UNION_TYPE
909 to the mode of the widest field iff the widest field has mode class
910 MODE_INT, or MODE_PARTIAL_INT and the union would be passed by
913 2018-11-07 Nikolai Merinov <n.merinov@inango-systems.com>
915 * common.opt: Add -Wattribute-warning.
916 * doc/invoke.texi: Add documentation for -Wno-attribute-warning.
917 * expr.c (expand_expr_real_1): Add new attribute to warning_at
918 call to allow user configure behavior of "warning" attribute.
920 2018-11-07 Segher Boessenkool <segher@kernel.crashing.org>
922 * target.def: Put @: after every vs., e.g., and i.e. where it is
923 followed by whitespace.
924 * doc/extend.texi: Ditto.
925 * doc/fragments.texi: Ditto.
926 * doc/gimple.texi: Ditto.
927 * doc/implement-c.texi: Ditto.
928 * doc/install.texi: Ditto.
929 * doc/invoke.texi: Ditto.
930 * doc/md.texi: Ditto.
931 * doc/plugins.texi: Ditto.
932 * doc/rtl.texi: Ditto.
933 * doc/sourcebuild.texi: Ditto.
934 * doc/tm.texi.in: Ditto.
935 * doc/ux.texi: Ditto.
936 * doc/tm.texi: Regenerate.
938 2018-11-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
940 * config/arm/arm-cpus.in (ares): New entry.
941 * config/arm/arm-tables.opt: Regenerate.
942 * config/arm/arm-tune.md: Likewise.
943 * doc/invoke.texi (ARM Options): Document ares.
945 2018-11-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
947 * config/aarch64/aarch64-cores.def (ares): Define.
948 * config/aarch64/aarch64-tune.md: Regenerate.
949 * doc/invoke.texi (AArch64 Options): Document ares value for mtune.
951 2018-11-07 Jan Hubicka <jh@suse.cz>
953 * tree.c (fld_incomplete_type_of): Clear TREE_ADDRESSABLE.
954 (free_lang_data_in_decl): Set TREE_ADDRESSABLE for public vars and
955 functions; clear TYPE_DECL_SUPPRESS_DEBUG and DECL_MODE for
958 2018-11-07 Richard Biener <rguenther@suse.de>
960 PR tree-optimization/87914
961 * tree-vect-loop.c (vect_is_simple_reduction): Improve detection
963 (vectorizable_reduction): Handle shifts and rotates by dispatching
964 to vectorizable_shift.
965 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Handle
966 in-loop uses of vect_nested_cycle defs. Merge cycle and internal
968 (vectorizable_shift): Export and handle being called as
970 (vect_analyze_stmt): Call vectorizable_shift after
971 vectorizable_reduction.
972 * tree-vectorizer.h (vectorizable_shift): Declare.
974 2018-11-07 Jan Hubicka <jh@suse.cz>
976 * ipa-devirt.c (odr_types_equivalent_p): Expect constants
977 than const decls in TREE_VALUE of enum.
978 (dump_type_inheritance_graph): Improve duplicate dumping.
979 (free_enum_values): New.
980 (build_type_inheritance_graph): Use it.
981 * tree.c (free_lang_data_in_type): Free TYPE_VALUES of enums
982 which are not main variants or not ODR types.
983 (verify_type_variant): Expect variants to have no TYPE_VALUES.
985 2018-11-07 Richard Biener <rguenther@suse.de>
987 * ipa-inline.c (want_inline_small_function_p): Compute
988 big_speedup_p lazily and last.
990 2018-11-07 Jan Hubicka <jh@suse.cz>
992 * tree.c (fld_type_variant_equal_p): Skip TYPE_ALIGN check when
993 building incomplete variant of complete type.
994 (fld_type_variant): Do not copy TYPE_ALIGN when building incomplete
995 variant of complete type.
997 2018-11-07 Chenghua Xu <paul.hua.gm@gmail.com>
999 * config/mips/mips.c: Fix typo in documentation of
1000 mips_loongson_ext2_prefetch_cookie.
1001 (mips_option_override): fix brain twister logical.
1002 * config/mips/mips.h: Fix typo in documentation of
1003 ISA_HAS_CTZ_CTO and define pattern.
1004 * config/mips/mips.md (prefetch): Hoist EXT2 above
1006 (prefetch_indexed): Hoist EXT2 above the EXT block.
1008 2018-11-07 Jan Hubicka <jh@suse.cz>
1010 * tree.c (free_lang_data_in_type): Add fld parameter; simplify
1011 return and parameter types of function and method types.
1012 (free_lang_data_in_cgraph): Update.
1014 2018-11-07 Martin Liska <mliska@suse.cz>
1016 PR rtl-optimization/87868
1017 * postreload-gcse.c (eliminate_partially_redundant_load): Set
1018 threshold to max_count if we would overflow.
1019 * profile-count.h: Make max_count a public constant.
1021 2018-11-07 Martin Liska <mliska@suse.cz>
1023 * mem-stats.h: Fix GNU coding style.
1025 2018-11-07 Chenghua Xu <paul.hua.gm@gmail.com>
1027 * config/mips/gs264e.md: New.
1028 * config/mips/mips-cpus.def: Define gs264e.
1029 * config/mips/mips-tables.opt: Regenerate.
1030 * config/mips/mips.c (mips_rtx_cost_data): Add DEFAULT_COSTS for
1032 (mips_issue_rate): Add support for gs264e.
1033 (mips_multipass_dfa_lookahead): Likewise.
1034 * config/mips/mips.h: Define TARGET_GS264E and TUNE_GS264E.
1035 (MIPS_ISA_LEVEL_SPEC): Infer mips64r2 from gs264e.
1036 (MIPS_ASE_MSA_SPEC): New.
1037 (BASE_DRIVER_SELF_SPECS): march=gs264e implies -mmsa.
1038 (ISA_HAS_FUSED_MADD4): Enable for TARGET_GS264E.
1039 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_GS264E.
1040 * config/mips/mips.md: Include gs264e.md.
1041 (processor): Add gs264e.
1042 * config/mips/mips.opt (MSA): Use Mask instead of Var.
1043 * doc/invoke.texi: Add gs264e to supported architectures.
1045 2018-11-07 Chenghua Xu <paul.hua.gm@gmail.com>
1047 * config/mips/gs464e.md: New.
1048 * config/mips/mips-cpus.def: Define gs464e.
1049 * config/mips/mips-tables.opt: Regenerate.
1050 * config/mips/mips.c (mips_rtx_cost_data): Add DEFAULT_COSTS for
1052 (mips_issue_rate): Add support for gs464e.
1053 (mips_multipass_dfa_lookahead): Likewise.
1054 (mips_option_override): Enable MMI, EXT and EXT2 for gs464e.
1055 * config/mips/mips.h: Define TARGET_GS464E and TUNE_GS464E.
1056 (MIPS_ISA_LEVEL_SPEC): Infer mips64r2 from gs464e.
1057 (ISA_HAS_FUSED_MADD4): Enable for TARGET_GS464E.
1058 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_GS464E.
1059 * config/mips/mips.md: Include gs464e.md.
1060 (processor): Add gs464e.
1061 * doc/invoke.texi: Add gs464e to supported architectures.
1063 2018-11-07 Chenghua Xu <paul.hua.gm@gmail.com>
1065 * config/mips/loongson3a.md: Rename to ...
1066 * config/mips/gs464.md: ... here.
1067 * config/mips/mips-cpus.def: Define gs464; Add loongson3a
1068 as an alias of gs464 processor.
1069 * config/mips/mips-tables.opt: Regenerate.
1070 * config/mips/mips.c (mips_issue_rate): Use PROCESSOR_GS464
1071 instead of PROCESSOR_LOONGSON_3A.
1072 (mips_multipass_dfa_lookahead): Use TUNE_GS464 instead of
1074 (mips_option_override): Enable MMI and EXT for gs464.
1075 * config/mips/mips.h: Rename TARGET_LOONGSON_3A to TARGET_GS464;
1076 Rename TUNE_LOONGSON_3A to TUNE_GS464.
1077 (MIPS_ISA_LEVEL_SPEC): Infer mips64r2 from gs464.
1078 (ISA_HAS_ODD_SPREG, ISA_AVOID_DIV_HILO, ISA_HAS_FUSED_MADD4,
1079 ISA_HAS_UNFUSED_MADD4): Use TARGET_GS464 instead of
1081 * config/mips/mips.md: Include gs464.md instead of loongson3a.md.
1082 (processor): Add gs464;
1083 * doc/invoke.texi: Add gs464 to supported architectures.
1085 2018-11-07 Chenghua Xu <paul.hua.gm@gmail.com>
1087 * config/mips/mips-protos.h
1088 (mips_loongson_ext2_prefetch_cookie): New prototype.
1089 * config/mips/mips.c (mips_loongson_ext2_prefetch_cookie): New.
1090 (mips_option_override): Enable TARGET_LOONGSON_EXT when
1091 TARGET_LOONGSON_EXT2 is true.
1092 * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Define
1093 __mips_loongson_ext2, __mips_loongson_ext_rev=2.
1094 (ISA_HAS_CTZ_CTO): New, true if TARGET_LOONGSON_EXT2.
1095 (ISA_HAS_PREFETCH): Include TARGET_LOONGSON_EXT and
1096 TARGET_LOONGSON_EXT2.
1097 (ASM_SPEC): Add mloongson-ext2 and mno-loongson-ext2.
1098 (define_insn "ctz<mode>2"): New insn pattern.
1099 (define_insn "prefetch"): Include TARGET_LOONGSON_EXT2.
1100 (define_insn "prefetch_indexed_<mode>"): Include
1101 TARGET_LOONGSON_EXT and TARGET_LOONGSON_EXT2.
1102 * config/mips/mips.opt (-mloongson-ext2): Add option.
1103 * gcc/doc/invoke.texi (-mloongson-ext2): Document.
1105 2018-11-07 Chenghua Xu <paul.hua.gm@gmail.com>
1107 * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Add
1108 __mips_loongson_ext.
1109 (MIPS_ASE_LOONGSON_EXT_SPEC): New.
1110 (BASE_DRIVER_SELF_SPECS): march=loongson3a implies
1112 (ASM_SPEC): Add mloongson-ext and mno-loongson-ext.
1113 * config/mips/mips.md (mul<mode>3, mul<mode>3_mul3_nohilo,
1114 <u>div<mode>3, <u>mod<mode>3, prefetch): Use TARGET_LOONGSON_EXT
1115 instead of TARGET_LOONGSON_3A.
1116 * config/mips/mips.opt (-mloongson-ext): Add option.
1117 * gcc/doc/invoke.texi (-mloongson-ext): Document.
1119 2018-11-07 Chenghua Xu <paul.hua.gm@gmail.com>
1121 * config.gcc (extra_headers): Add loongson-mmiintrin.h.
1122 * config/mips/loongson.md: Move to ...
1123 * config/mips/loongson-mmi.md: here; Adjustment.
1124 * config/mips/loongson.h: Move to ...
1125 State as deprecated. Include loongson-mmiintrin.h for back
1126 compatibility and warning.
1127 * config/mips/loongson-mmiintrin.h: ... here.
1128 * config/mips/mips.c (mips_hard_regno_mode_ok_uncached,
1129 mips_vector_mode_supported_p, AVAIL_NON_MIPS16): Use
1130 TARGET_LOONGSON_MMI instead of TARGET_LOONGSON_VECTORS.
1131 (mips_option_override): Make sure MMI use hard float;
1132 (mips_shift_truncation_mask, mips_expand_vpc_loongson_even_odd,
1133 mips_expand_vpc_loongson_pshufh, mips_expand_vpc_loongson_bcast,
1134 mips_expand_vector_init): Use TARGET_LOONGSON_MMI instead of
1135 TARGET_LOONGSON_VECTORS.
1136 * gcc/config/mips/mips.h (TARGET_LOONGSON_VECTORS): Delete.
1137 (TARGET_CPU_CPP_BUILTINS): Add __mips_loongson_mmi.
1138 (MIPS_ASE_DSP_SPEC, MIPS_ASE_LOONGSON_MMI_SPEC): New.
1139 (BASE_DRIVER_SELF_SPECS): march=loongson2e/2f/3a implies
1141 (SHIFT_COUNT_TRUNCATED): Use TARGET_LOONGSON_MMI instead of
1142 TARGET_LOONGSON_VECTORS.
1143 * gcc/config/mips/mips.md (MOVE64, MOVE128): Use
1144 TARGET_LOONGSON_MMI instead of TARGET_LOONGSON_VECTORS.
1145 (Loongson MMI patterns): Include loongson-mmi.md instead of
1147 * gcc/config/mips/mips.opt (-mloongson-mmi): New option.
1148 * gcc/doc/invoke.texi (-mloongson-mmi): Document.
1150 2018-11-07 Richard Biener <rguenther@suse.de>
1153 * tree-streamer-in.c (lto_input_ts_block_tree_pointers): Fixup
1154 BLOCK_ABSTRACT_ORIGIN to be the ultimate origin.
1156 2018-11-07 Alexandre Oliva <aoliva@redhat.com>
1158 PR rtl-optimization/87874
1159 * lra.c (lra_substitute_pseudo): Do not create a subreg for
1162 2018-11-06 Aaron Sawdey <acsawdey@linux.ibm.com>
1164 * config/rs6000/rs6000.md (bswap<mode>2): Force address into register
1165 if not in indexed or indirect form.
1166 (bswap<mode>2_load): Change predicate to indexed_or_indirect_operand.
1167 (bswap<mode>2_store): Ditto.
1169 2018-11-06 Richard Earnshaw <rearnsha@arm.com>
1171 * config/aarch64/aarch64.md (speculation_tracker): Set the mode for
1174 2018-11-06 Richard Biener <rguenther@suse.de>
1176 PR tree-optimization/86850
1177 * vec.h (vec<T, va_heap, vl_ptr>::splice): Check src.length ()
1178 instead of src.m_vec.
1180 2018-11-06 Jan Hubicka <jh@suse.cz>
1182 * tree.c (fld_simplified_type_name): Break out form ...
1183 (free_lang_data_in_type): ... here.
1184 (fld_type_variant_equal_p): Use it.
1186 2018-11-06 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1188 * config/default-d.c: Include memmodel.h.
1190 * config/sol2-d.c: New file.
1191 * config/t-sol2 (sol2-d.o): New rule.
1192 * config.gcc <*-*-solaris2*>: Set d_target_objs,
1193 target_has_targetdm.
1195 2018-11-06 Jan Hubicka <jh@suse.cz>
1197 * tree.c (fld_type_variant): Also copy alignment; be sure that
1198 new variant is equal.
1200 2018-11-06 Ilya Leoshkevich <iii@linux.ibm.com>
1203 * config/s390/s390.md: Add relative_long attribute.
1205 2018-11-06 Jan Hubicka <jh@suse.cz>
1207 * ipa-pure-const.c (check_decl): Do not test TYPE_NEEDS_CONSTRUCTING.
1208 * lto-streamer-out.c (hash_tree): Do not hash TYPE_NEEDS_CONSTRUCTING.
1209 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Do not
1210 stream TYPE_NEEDS_CONSTRUCTING.
1211 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
1212 * tree.c (free_lang_data_in_type): Clear TYPE_NEEDS_CONSTRUCTING.
1214 2018-11-06 Richard Biener <rguenther@suse.de>
1216 * tree-vect-slp.c (vect_slp_bb): Move opening of vect_slp_analyze_bb
1218 (vect_slp_analyze_bb_1): ... here to avoid hiding optimized locations.
1220 2018-11-06 Jozef Lawrynowicz <jozef.l@mittosystems.com>
1222 * gcc/config/msp430/msp430.h (REG_CLASS_CONTENTS): Add R0 to
1223 REG_CLASS_CONTENTS[GEN_REGS].
1224 (REGNO_REG_CLASS): Return NO_REGS for R2 and R3.
1226 2018-11-06 Jan Hubicka <jh@suse.cz>
1228 * tree.c (fld_simplified_type_of): Clear TYPELESS_STORAGE flag.
1230 2018-11-06 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1233 * config/sol2.h (ASAN_CC1_SPEC): Define.
1234 (LD_WHOLE_ARCHIVE_OPTION): Define.
1235 (LD_NO_WHOLE_ARCHIVE_OPTION): Define.
1236 (ASAN_REJECT_SPEC): Provide default.
1237 (LIBASAN_EARLY_SPEC): Define.
1238 (LIBTSAN_EARLY_SPEC): Define.
1239 (LIBLSAN_EARLY_SPEC): Define.
1240 * config/i386/sol2.h (CC1_SPEC): Redefine.
1241 (ASAN_REJECT_SPEC): Define.
1243 * config/sparc/sparc.c (sparc_asan_shadow_offset): Declare.
1244 (TARGET_ASAN_SHADOW_OFFSET): Define.
1245 (sparc_asan_shadow_offset): New function.
1246 * config/sparc/sol2.h (CC1_SPEC): Append ASAN_CC1_SPEC.
1247 (ASAN_REJECT_SPEC): Define.
1249 2018-11-06 Jan Hubicka <jh@suse.cz>
1251 * tree.c (fld_type_variant): Copy canonical type.
1252 (fld_incomplete_type_of): Check that canonical types looks sane;
1253 copy canonical type.
1254 (verify_type): Accept when incomplete type has complete canonical type.
1256 2018-11-06 Jan Hubicka <jh@suse.cz>
1258 * tree.c (free_lang_data): Reset overwite_assembler_name,
1259 print_xnode, print_decl, print_type and print_identifier of
1262 2018-11-06 Richard Biener <rguenther@suse.de>
1264 PR tree-optimization/87889
1265 * tree-vect-loop-manip.c (slpeel_duplicate_current_defs_from_edges):
1266 Do nothing if old and new arg are the same
1268 2018-11-06 Andreas Krebbel <krebbel@linux.ibm.com>
1271 * config/s390/s390.md ("*r<noxa>sbg_di_rotl"): Remove mode
1272 attributes for operands 3 and 4.
1274 2018-11-06 Richard Biener <rguenther@suse.de>
1277 * simplify-rtx.c (simplify_binary_operation_1): Add pattern
1278 matching bitfield insertion.
1280 2018-11-06 Alexandre Oliva <aoliva@redhat.com>
1282 * auto-inc-dec.c: Include valtrack.h. Improve comments.
1283 (reg_next_debug_use): New.
1284 (attempt_change): Propagate adjusted expression into affected
1286 (merge_in_block): Track uses in debug insns.
1287 (pass_inc_dec::execute): Allocate and release
1290 2018-11-06 Wei Xiao <wei3.xiao@intel.com>
1292 * config/i386/avx512fintrin.h: Update VFIXUPIMM* intrinsics.
1293 (_mm512_fixupimm_round_pd): Update parameters and builtin.
1294 (_mm512_maskz_fixupimm_round_pd): Ditto.
1295 (_mm512_fixupimm_round_ps): Ditto.
1296 (_mm512_maskz_fixupimm_round_ps): Ditto.
1297 (_mm_fixupimm_round_sd): Ditto.
1298 (_mm_maskz_fixupimm_round_sd): Ditto.
1299 (_mm_fixupimm_round_ss): Ditto.
1300 (_mm_maskz_fixupimm_round_ss): Ditto.
1301 (_mm512_fixupimm_pd): Ditto.
1302 (_mm512_maskz_fixupimm_pd): Ditto.
1303 (_mm512_fixupimm_ps): Ditto.
1304 (_mm512_maskz_fixupimm_ps): Ditto.
1305 (_mm_fixupimm_sd): Ditto.
1306 (_mm_maskz_fixupimm_sd): Ditto.
1307 (_mm_fixupimm_ss): Ditto.
1308 (_mm_maskz_fixupimm_ss): Ditto.
1309 (_mm512_mask_fixupimm_round_pd): Update builtin.
1310 (_mm512_mask_fixupimm_round_ps): Ditto.
1311 (_mm_mask_fixupimm_round_sd): Ditto.
1312 (_mm_mask_fixupimm_round_ss): Ditto.
1313 (_mm512_mask_fixupimm_pd): Ditto.
1314 (_mm512_mask_fixupimm_ps): Ditto.
1315 (_mm_mask_fixupimm_sd): Ditto.
1316 (_mm_mask_fixupimm_ss): Ditto.
1317 * config/i386/avx512vlintrin.h:
1318 (_mm256_fixupimm_pd): Update parameters and builtin.
1319 (_mm256_maskz_fixupimm_pd): Ditto.
1320 (_mm256_fixupimm_ps): Ditto.
1321 (_mm256_maskz_fixupimm_ps): Ditto.
1322 (_mm_fixupimm_pd): Ditto.
1323 (_mm_maskz_fixupimm_pd): Ditto.
1324 (_mm_fixupimm_ps): Ditto.
1325 (_mm_maskz_fixupimm_ps): Ditto.
1326 (_mm256_mask_fixupimm_pd): Update builtin.
1327 (_mm256_mask_fixupimm_ps): Ditto.
1328 (_mm_mask_fixupimm_pd): Ditto.
1329 (_mm_mask_fixupimm_ps): Ditto.
1330 * config/i386/i386-builtin-types.def: Add new types and remove useless ones.
1331 * config/i386/i386-builtin.def: Update builtin definitions.
1332 * config/i386/i386.c: Handle new builtin types and remove useless ones.
1333 * config/i386/sse.md: Update VFIXUPIMM* patterns.
1334 (<avx512>_fixupimm<mode>_maskz<round_saeonly_expand_name>): Update.
1335 (<avx512>_fixupimm<mode><sd_maskz_name><round_saeonly_name>): Update.
1336 (<avx512>_fixupimm<mode>_mask<round_saeonly_name>): Update.
1337 (avx512f_sfixupimm<mode>_maskz<round_saeonly_expand_name>): Update.
1338 (avx512f_sfixupimm<mode><sd_maskz_name><round_saeonly_name>): Update.
1339 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Update.
1340 * config/i386/subst.md:
1341 (round_saeonly_sd_mask_operand4): Add new subst_attr.
1342 (round_saeonly_sd_mask_op4): Ditto.
1343 (round_saeonly_expand_operand5): Ditto.
1344 (round_saeonly_expand): Update.
1346 2018-11-05 Max Filippov <jcmvbkbc@gmail.com>
1348 * config/xtensa/uclinux.h (XTENSA_ALWAYS_PIC): Change to 0.
1350 2018-11-05 Segher Boessenkool <segher@kernel.crashing.org>
1352 PR rtl-optimization/87871
1353 * combine.c (make_more_copies): Skip if dest is frame_pointer_rtx.
1355 2018-11-05 Paul Koning <ni1d@arrl.net>
1357 * doc/sourcebuild.texi (target attributes): Document new "inf"
1358 effective target keyword.
1360 2018-11-05 Robin Dapp <rdapp@linux.ibm.com>
1362 * config/s390/s390.c (s390_register_move_cost): Increase costs for
1363 moves involving the CC reg.
1365 2018-11-05 Richard Biener <rguenther@suse.de>
1367 PR tree-optimization/87873
1368 * tree-ssa-loop-manip.h (split_loop_exit_edge): Add copy_constants_p
1370 * tree-ssa-loop-manip.c (split_loop_exit_edge): Likewise.
1371 * tree-vect-loop.c (vect_transform_loop): When splitting the
1372 loop exit also create forwarder PHIs for constants.
1373 * tree-vect-loop-manip.c (slpeel_duplicate_current_defs_from_edges):
1374 Handle constant to_arg, add extra checking we match up the correct
1377 2018-11-05 Robin Dapp <rdapp@linux.ibm.com>
1379 * config/s390/s390.md: QImode and HImode for load on condition.
1381 2018-11-05 Robin Dapp <rdapp@linux.ibm.com>
1383 * config/s390/predicates.md: Fix typo.
1384 * config/s390/s390.md: Allow immediates for load on condition.
1386 2018-11-05 Martin Liska <mliska@suse.cz>
1388 * alloc-pool.h (struct pool_usage): Use SIZE_AMOUNT.
1389 * bitmap.h (struct bitmap_usage): Likewise.
1390 * ggc-common.c (SCALE): Remove.
1392 (struct ggc_usage): Use SIZE_AMOUNT. And update
1394 * ggc-page.c (SCALE): Remove.
1395 (STAT_LABEL): Remove.
1396 (ggc_print_statistics): Use SIZE_AMOUNT.
1397 * gimple.h (SCALE): Remove.
1399 * input.c (ONE_K): Remove.
1402 (STAT_LABEL): Likewise.
1403 (FORMAT_AMOUNT): Likewise.
1404 (dump_line_table_statistics): Use SIZE_AMOUNT.
1405 * mem-stats.h (struct mem_usage): Likewise.
1406 * rtl.c (dump_rtx_statistics): Likewise.
1407 (rtx_alloc_counts): Change type to size_t.
1408 (rtx_alloc_sizes): Likewise.
1409 (rtx_count_cmp): New.
1410 (dump_rtx_statistics): Sort first based on counts.
1411 * tree.c (tree_nodes_cmp): New.
1412 (tree_codes_cmp): New.
1413 (dump_tree_statistics): Sort first based on counts.
1414 * system.h (ONE_K): New.
1416 (SIZE_SCALE): Likewise.
1417 (SIZE_LABEL): Likewise.
1418 (SIZE_AMOUNT): Likewise.
1419 * tree-cfg.c (dump_cfg_stats): Use SIZE_AMOUNT.
1420 * tree-dfa.c (dump_dfa_stats): Likewise.
1421 * tree-phinodes.c (phinodes_print_statistics): Likewise.
1422 * tree-ssanames.c (ssanames_print_statistics): Likewise.
1423 * tree.c (dump_tree_statistics): Likewise.
1424 * vec.c (struct vec_usage): Likewise.
1425 * trans-mem.c (tm_mangle): Enlarge buffer in order to not
1426 trigger a -Werror=format-overflow with
1427 --enable-gather-detailed-stats.
1429 2018-11-05 Martin Liska <mliska@suse.cz>
1431 * mem-stats.h (mem_alloc_description::release_instance_overhead):
1433 * vec.c (struct vec_usage): Register m_element_size.
1434 (vec_prefix::register_overhead): New arguments: elements and
1436 (vec_prefix::release_overhead): Subtract elements.
1437 * vec.h (struct vec_prefix): Change signature.
1438 (va_heap::reserve): Pass proper arguments.
1439 (va_heap::release): Likewise.
1441 2018-11-05 Martin Liska <mliska@suse.cz>
1443 * mem-stats.h (mem_alloc_description::get_list): Fix GNU coding
1447 2018-11-05 Richard Biener <rguenther@suse.de>
1449 * tree-scalar-evolution.h (final_value_replacement_loop): Update
1451 * tree-scalar-evolution.c (final_value_replacement_loop): Return
1452 whether anything was done.
1453 (scev_const_prop): Remove constant propagation part, fold
1455 * tree-ssa-loop.c (pass_scev_cprop::execute): ... here.
1456 (pass_data_scev_cprop): TODO_cleanup_cfg is now done
1459 2018-11-05 Jakub Jelinek <jakub@redhat.com>
1461 PR tree-optimization/87859
1462 * gimple-ssa-store-merging.c (struct merged_store_group): Add
1463 only_constants and first_nonmergeable_order members.
1464 (merged_store_group::merged_store_group): Initialize them.
1465 (merged_store_group::do_merge): Clear only_constants member if
1466 adding something other than INTEGER_CST store.
1467 (imm_store_chain_info::coalesce_immediate_stores): Don't merge
1468 stores with order >= first_nonmergeable_order. Use
1469 merged_store->only_constants instead of always recomputing it.
1470 Set merged_store->first_nonmergeable_order if we've skipped any
1471 stores. Attempt to merge overlapping INTEGER_CST stores that
1472 we would otherwise skip.
1475 * match.pd (X + Y < X): Don't optimize if TYPE_OVERFLOW_SANITIZED.
1477 2018-11-05 Xuepeng Guo <xuepeng.guo@intel.com>
1480 * config/i386/emmintrin.h (__v16qs): New to cope with option
1482 (_mm_cmpeq_epi8): Replace __v16qi with __v16qs.
1483 (_mm_cmplt_epi8): Likewise.
1484 (_mm_cmpgt_epi8): Likewise.
1486 2018-11-05 Richard Biener <rguenther@suse.de>
1488 PR rtl-optimization/87852
1489 * fwprop.c (use_killed_between): Only consider single-defs of the
1490 use whose definition statement dominates the use.
1492 2018-11-05 Martin Liska <mliska@suse.cz>
1495 * doc/invoke.texi: Remove options that are
1496 not disabled with -Os.
1498 2018-11-05 Martin Liska <mliska@suse.cz>
1501 * doc/extend.texi: Update constrain about the last argument
1502 of __builtin_expect_with_probability.
1504 2018-11-05 Martin Liska <mliska@suse.cz>
1507 * predict.c (expr_expected_value_1): Verify
1508 that last argument is a real constants and emit
1511 2018-11-05 Martin Liska <mliska@suse.cz>
1513 PR gcov-profile/77698
1514 * ipa-profile.c (ipa_profile): Adjust hotness threshold
1517 2018-11-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
1519 PR tree-optimization/86572
1520 * builtins.c (c_strlen): Handle negative offsets in a safe way.
1522 2018-11-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
1524 PR tree-optimization/87672
1525 * gimple-fold.c (gimple_fold_builtin_stxcpy_chk): Gimplify.
1526 * tree-ssa-strlen.c (handle_builtin_strcat): Adjust object size.
1528 2018-11-04 Uros Bizjak <ubizjak@gmail.com>
1531 * cfgexpand.c (pass_expand::execute): Move the call to
1532 finish_eh_generation in front of the call to expand_stack_alignment.
1534 2018-11-04 Venkataramanan Kumar <venkataramanan.kumar@amd.com>
1536 * common/config/i386/i386-common.c (processor_alias_table): Add
1538 * config.gcc (i[34567]86-*-linux* | ...): Add znver2.
1539 (case ${target}): Add znver2.
1540 * config/i386/driver-i386.c: (host_detect_local_cpu): Let
1541 -march=native recognize znver2 processors.
1542 * config/i386/i386-c.c (ix86_target_macros_internal): Add znver2.
1543 * config/i386/i386.c (m_znver2): New definition.
1544 (m_ZNVER): New definition.
1545 (m_AMD_MULTIPLE): Includes m_znver2.
1546 (processor_cost_table): Add znver2 entry.
1547 (processor_target_table): Add znver2 entry.
1548 (get_builtin_code_for_version): Set priority for
1550 (processor_model): Add M_AMDFAM17H_ZNVER2.
1551 (arch_names_table): Ditto.
1552 (ix86_reassociation_width): Include znver2.
1553 * config/i386/i386.h (TARGET_znver2): New definition.
1554 (struct ix86_size_cost): Add TARGET_ZNVER2.
1555 (enum processor_type): Add PROCESSOR_ZNVER2.
1556 * config/i386/i386.md (define_attr "cpu"): Add znver2.
1557 * config/i386/x86-tune-costs.h: (processor_costs) Add znver2 costs.
1558 * config/i386/x86-tune-sched.c: (ix86_issue_rate): Add znver2.
1559 (ix86_adjust_cost): Add znver2.
1560 * config/i386/x86-tune.def: Replace m_ZNVER1 by m_ZNVER.
1561 * gcc/doc/extend.texi: Add details about znver2.
1562 * gcc/doc/invoke.texi: Add details about znver2.
1564 2018-11-03 Sandra Loosemore <sandra@codesourcery.com>
1568 * config/nios2/nios2.c (nios2_rtx_costs): Recognize <mul>sidi3
1571 2018-11-02 Aaron Sawdey <acsawdey@linux.ibm.com>
1573 * config/rs6000/rs6000-string.c (expand_strncmp_gpr_sequence): Pay
1574 attention to TARGET_AVOID_XFORM and BYTES_BIG_ENDIAN.
1576 2018-11-02 Richard Earnshaw <rearnsha@arm.com>
1578 * config/aarch64/aarch64.c ((aarch64_override_options): Disable
1579 shrink-wrapping when -mtrack-speculation.
1581 2018-11-02 Richard Biener <rguenther@suse.de>
1583 * tree-ssa-coalesce.c (struct coalesce_list): Add obstack member.
1584 (pop_cost_one_pair): Do not free pair.
1585 (pop_best_coalesce): Likewise.
1586 (create_coalesce_list): Initialize obstack.
1587 (delete_coalesce_list): Free obstack.
1588 (find_coalesce_pair): Obstack-allocate coalesce pairs.
1589 (add_cost_one_coalesce): Likewise.
1590 (struct live_track): Remove bitmap pointer indirections.
1591 (new_live_track): Adjust.
1592 (delete_live_track): Likewise.
1593 (live_track_remove_partition): Likewise.
1594 (live_track_add_partition): Likewise.
1595 (live_track_live_p): Likewise.
1596 (live_track_process_def): Likewise.
1597 (live_track_clear_base_vars): Likewise.
1599 2018-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1601 * configure.ac (gcc_cv_as_sparc_register_op): Remove.
1602 * configure: Regenerate.
1603 * config.in: Regenerate.
1604 * config/sparc/sparc.c (sparc_output_scratch_registers): Remove
1605 HAVE_AS_REGISTER_PSEUDO_OP guard.
1606 * config/sparc/sparc.h (ASM_ARCH64_SPEC): Remove
1607 !HAVE_AS_REGISTER_PSEUDO_OP support.
1608 (ASM_DECLARE_REGISTER_GLOBAL): Remove HAVE_AS_REGISTER_PSEUDO_OP
1611 2018-11-02 Richard Biener <rguenther@suse.de>
1613 * dwarf2out.c (build_abbrev_table): Guard lookup_external_ref call
1616 2018-11-02 Richard Biener <rguenther@suse.de>
1618 PR tree-optimization/87776
1619 * tree-ssa-sccvn.c (do_rpo_vn): Do not mark backedges
1620 executable when iterating but running into rpo-vn-max-loop-depth
1621 and not eliding the iteration.
1623 2018-11-30 Jan Hubicka <jh@suse.cz>
1625 * tree.c (free_lang_data_in_decl): Free TREE_TYPE and alignment of
1628 2018-11-01 Aaron Sawdey <acsawdey@linux.ibm.com>
1630 * config/rs6000/rs6000-protos.h (rs6000_address_for_fpconvert): Remove
1632 * config/rs6000/rs6000.c (rs6000_force_indexed_or_indirect_mem):
1633 Combine with rs6000_address_for_fpconvert.
1634 (rs6000_address_for_fpconvert): Combine with
1635 rs6000_force_indexed_or_indirect_mem.
1636 (rs6000_expand_vector_init): Change function call from
1637 rs6000_address_for_fpconvert to rs6000_force_indexed_or_indirect_mem.
1638 * config/rs6000/rs6000.md (floatsi<mode>2_lfiwax): Change call from
1639 rs6000_address_for_fpconvert to rs6000_force_indexed_or_indirect_mem.
1640 (floatsi<mode>2_lfiwax_mem): Ditto.
1641 (floatunssi<mode>2_lfiwzx): Ditto.
1642 (floatunssi<mode>2_lfiwzx_mem): Ditto.
1643 (float<QHI:mode><FP_ISA3:mode>2): Ditto.
1644 (floatuns<QHI:mode><FP_ISA3:mode>2): Ditto.
1645 (fix_trunc<mode>si2_stfiwx): Ditto.
1646 (fixuns_trunc<mode>si2_stfiwx): Ditto.
1647 (float_<mode>si2_hw): Ditto.
1648 (floatuns_<mode>si2_hw): Ditto.
1649 * config/rs6000/vsx.md (*vsx_extract_si): Ditto.
1650 (vsx_splat_<mode>): Ditto.
1652 2018-11-01 Joseph Myers <joseph@codesourcery.com>
1654 * configure.ac (gcc_cv_as_ia64_ltoffx_ldxmov_relocs): Add newline
1655 at end of assembler input text.
1656 * configure: Regenerate.
1658 2018-11-01 Jakub Jelinek <jakub@redhat.com>
1660 PR tree-optimization/87826
1661 * gimple-ssa-store-merging.c (do_shift_rotate): Punt if count is
1662 negative or larger or equal to type's precision.
1664 2018-10-31 Alexandre Oliva <aoliva@redhat.com>
1666 * opts.c (default_options_table): Do not enable
1667 OPT_fdelayed_branch at -Og.
1668 * doc/invoke.texi (-fdelayed-branch): Document it.
1670 2018-10-31 Richard Henderson <richard.henderson@linaro.org>
1672 * optabs-libfuncs.c (build_libfunc_function_visibility):
1673 New, split out from...
1674 (build_libfunc_function): ... here.
1675 (init_one_libfunc_visibility): New, split out from ...
1676 (init_one_libfunc): ... here.
1678 * config/aarch64/atomics.md (aarch64_atomic_<ATOMIC_LDOP><ALLI>_lse):
1679 scratch register need not be early-clobber. Document the reason
1680 why we cannot use ST<OP>.
1682 2018-10-31 Joseph Myers <joseph@codesourcery.com>
1685 * configure.ac: Remove AC_PREREQ. Use AC_LANG_SOURCE. Use single
1686 line for second argument of AC_DEFINE_UNQUOTED.
1687 * doc/install.texi (Tools/packages necessary for modifying GCC):
1688 Update to autoconf 2.69 and automake 1.15.1.
1689 * aclocal.m4, config.in, configure: Regenerate.
1691 2018-10-31 Pat Haugen <pthaugen@us.ibm.com>
1693 * tree-ssa-loop-ivcanon.c (try_peel_loop): Fix typo and remove dead
1696 2018-10-31 Martin Liska <mliska@suse.cz>
1699 * common/config/aarch64/aarch64-common.c (aarch64_parse_extension):
1700 Add new argument invalid_extension.
1701 (aarch64_get_all_extension_candidates): New function.
1702 (aarch64_rewrite_selected_cpu): Add NULL to function call.
1703 * config/aarch64/aarch64-protos.h (aarch64_parse_extension): Add
1705 (aarch64_get_all_extension_candidates): New function.
1706 * config/aarch64/aarch64.c (aarch64_parse_arch): Add new
1707 argument invalid_extension.
1708 (aarch64_parse_cpu): Likewise.
1709 (aarch64_print_hint_for_extensions): New function.
1710 (aarch64_validate_mcpu): Provide hint about invalid extension.
1711 (aarch64_validate_march): Likewise.
1712 (aarch64_handle_attr_arch): Pass new argument.
1713 (aarch64_handle_attr_cpu): Provide hint about invalid extension.
1714 (aarch64_handle_attr_isa_flags): Likewise.
1716 2018-10-31 Richard Biener <rguenther@suse.de>
1720 * tree-outof-ssa.c (insert_backedge_copies): Restrict
1721 copy generation to useful cases. Place the copy before
1722 the definition of the backedge value when possible.
1724 2018-10-31 Claudiu Zissulescu <claziss@synopsys.com>
1726 * config/arc/arc-arch.h (ARC_TUNE_ARC7XX): New tune value.
1727 * config/arc/arc.c (arc_active_insn): New function.
1728 (check_store_cacheline_hazard): Likewise.
1729 (workaround_arc_anomaly): Use check_store_cacheline_hazard.
1730 (arc_override_options): Disable delay slot scheduler for older
1732 (arc_store_addr_hazard_p): New implementation, old one renamed to
1734 (arc_store_addr_hazard_internal_p): Renamed.
1735 (arc_reorg): Don't combine into brcc instructions which are part
1736 of hardware hazard solution.
1737 * config/arc/arc.md (attr tune): Consider new arc7xx tune value.
1738 (tune_arc700): Likewise.
1739 * config/arc/arc.opt (arc7xx): New tune value.
1740 * config/arc/arc700.md: Improve A7 scheduler.
1742 2018-10-31 Claudiu Zissulescu <claziss@synopsys.com>
1744 * config/arc/arc.c (arc_override_options): Remove
1745 TARGET_COMPACT_CASESI.
1746 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Update.
1747 (CASE_VECTOR_MODE): Likewise.
1748 (CASE_VECTOR_PC_RELATIVE): Likewise.
1749 (CASE_VECTOR_SHORTEN_MODE): Likewise.
1750 (CASE_VECTOR_SHORTEN_MODE1): Delete.
1751 (ADDR_VEC_ALIGN): Update.
1752 (ASM_OUTPUT_CASE_LABEL): Undefine.
1753 (ASM_OUTPUT_BEFORE_CASE_LABEL): Undefine.
1754 (TARGET_BI_BIH): Define.
1755 (DEFAULT_BRANCH_INDEX): Likewise.
1756 * config/arc/arc.md (casesi): Rework to accept BI/BIH
1757 instructions, remove compact_casesi use case.
1758 (casesi_compact_jump): Remove.
1759 (casesi_dispatch): New pattern.
1760 * config/arc/arc.opt: Add mbranch-index option. Deprecate
1761 compact_casesi option.
1762 * doc/invoke.texi: Document mbranch-index option.
1764 2018-10-31 Claudiu Zissulescu <claziss@synopsys.com>
1766 * config/arc/arc.c (arc_get_tp): Remove function.
1767 (arc_emit_call_tls_get_addr): Likewise.
1768 (arc_call_tls_get_addr): New function.
1769 (arc_legitimize_tls_address): Make use of arc_call_tls_get_addr.
1770 * config/arc/arc.md (tls_load_tp_soft): Remove.
1771 (tls_gd_get_addr): Likewise.
1773 2018-10-31 Claudiu Zissulescu <claziss@synopsys.com>
1775 * config/arc/arc.md (mulsi3): Remove call to mulsi_600_lib.
1776 (mulsi3_600_lib): Remove pattern.
1777 (umulsi3_highpart_600_lib_le): Likewise.
1778 (umulsi3_highpart): Remove call to umulsi3_highpart_600_lib_le.
1779 (umulsidi3): Remove call to umulsidi3_600_lib.
1780 (umulsidi3_600_lib): Remove pattern.
1781 (peephole2): Remove peephole using the above deprecated patterns.
1783 2018-10-31 Thomas Preud'homme <thomas.preudhomme@linaro.org>
1786 * config/arm/arm.c (arm_option_check_internal): Disable the combined
1787 use of -mslow-flash-data and -mword-relocations.
1788 (arm_option_override): Enable -mword-relocations if -fpic or -fPIC.
1789 * config/arm/arm.md (SYMBOL_REF MOVT splitter): Stop checking for
1791 * doc/invoke.texi (-mword-relocations): Mention conflict with
1793 (-mslow-flash-data): Reciprocally.
1795 2018-10-31 Richard Henderson <richard.henderson@linaro.org>
1797 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Force
1798 16-byte modes held in GP registers to use an even regno.
1800 * config/aarch64/aarch64.c (aarch64_emit_bic): Remove.
1801 (aarch64_atomic_ldop_supported_p): Remove.
1802 (aarch64_gen_atomic_ldop): Remove.
1803 * config/aarch64/atomic.md (atomic_<atomic_optab><ALLI>):
1804 Fully expand LSE operations here.
1805 (atomic_fetch_<atomic_optab><ALLI>): Likewise.
1806 (atomic_<atomic_optab>_fetch<ALLI>): Likewise.
1807 (aarch64_atomic_<ATOMIC_LDOP><ALLI>_lse): Drop atomic_op iterator
1808 and use ATOMIC_LDOP instead; use register_operand for the input;
1809 drop the split and emit insns directly.
1810 (aarch64_atomic_fetch_<ATOMIC_LDOP><ALLI>_lse): Likewise.
1811 (aarch64_atomic_<atomic_op>_fetch<ALLI>_lse): Remove.
1812 (@aarch64_atomic_load<ATOMIC_LDOP><ALLI>): Remove.
1814 * config/aarch64/aarch64.c (aarch64_emit_atomic_swap): Remove.
1815 (aarch64_gen_atomic_ldop): Don't call it.
1816 * config/aarch64/atomics.md (atomic_exchange<ALLI>):
1817 Use aarch64_reg_or_zero.
1818 (aarch64_atomic_exchange<ALLI>): Likewise.
1819 (aarch64_atomic_exchange<ALLI>_lse): Remove split; remove & from
1820 operand 0; use aarch64_reg_or_zero for input; merge ...
1821 (@aarch64_atomic_swp<ALLI>): ... this and remove.
1823 * config/aarch64/aarch64.c (aarch64_gen_compare_reg_maybe_ze): New.
1824 (aarch64_split_compare_and_swap): Use it.
1825 (aarch64_expand_compare_and_swap): Likewise. Remove convert_modes;
1826 test oldval against the proper predicate.
1827 * config/aarch64/atomics.md (@atomic_compare_and_swap<ALLI>):
1828 Use nonmemory_operand for expected.
1829 (cas_short_expected_pred): New.
1830 (@aarch64_compare_and_swap<SHORT>): Use it; use "rn" not "rI" to match.
1831 (@aarch64_compare_and_swap<GPI>): Use "rn" not "rI" for expected.
1832 * config/aarch64/predicates.md (aarch64_plushi_immediate): New.
1833 (aarch64_plushi_operand): New.
1835 * config/aarch64/aarch64.c (aarch64_expand_compare_and_swap):
1836 Force oldval into the rval register for TARGET_LSE; emit the compare
1837 during initial expansion so that it may be deleted if unused.
1838 (aarch64_gen_atomic_cas): Remove.
1839 * config/aarch64/atomics.md (@aarch64_compare_and_swap<SHORT>_lse):
1840 Change =&r to +r for operand 0; use match_dup for operand 2;
1841 remove is_weak and mod_f operands as unused. Drop the split
1843 (@aarch64_atomic_cas<SHORT>): ... this pattern's output; remove.
1844 (@aarch64_compare_and_swap<GPI>_lse): Similarly.
1845 (@aarch64_atomic_cas<GPI>): Similarly.
1847 2018-10-31 Richard Biener <rguenther@suse.de>
1849 * tree-eh.c (replace_trapping_overflow): Simplify ABS_EXPR case
1852 2018-10-31 Sameera Deshpande <sameera.deshpande@linaro.org>
1854 * config/aarch64/aarch64-cores.def (saphira): Use saphira pipeline.
1855 * config/aarch64/aarch64.md: Include saphira.md
1856 * config/aarch64/saphira.md: New file for pipeline description.
1858 2018-10-30 Martin Sebor <msebor@redhat.com>
1861 * gimple-ssa-sprintf.c (format_directive): Use %G to include
1863 (sprintf_dom_walker::compute_format_length):
1864 Avoid setting POSUNDER4K here.
1865 (get_destination_size): Handle null argument values.
1866 (get_user_idx_format): New function.
1867 (sprintf_dom_walker::handle_gimple_call): Handle all printf-like
1868 functions, including user-defined with attribute format printf.
1869 Use %G to include inlining context.
1870 Set POSUNDER4K here.
1872 2018-10-30 Jan Hubicka <jh@suse.cz>
1874 * params.def (lto-partitions): Bump from 32 to 128.
1876 2018-10-30 Jan Hubicka <jh@suse.cz>
1879 (free_lang_data_d, add_tree_to_fld_list, fld_worklist_push): Move
1881 (free_lang_data_in_type): Forward declare.
1882 (fld_type_variant_equal_p): New function.
1883 (fld_type_variant): New function
1884 (fld_incomplete_types): New hash.
1885 (fld_incomplete_type_of): New function
1886 (fld_simplfied-type): New function.
1887 (free_lang_data_in_decl): Add fld parameter; simplify type of FIELD_DECL
1888 (free_lang_data): Allocate and free fld_incomplete_type; update call
1889 of free_lang_data_in_decl.
1891 2018-10-30 Eric Botcazou <ebotcazou@adacore.com>
1893 * gcov.c (output_lines): Remove duplicate line.
1895 2018-10-30 Aaron Sawdey <acsawdey@linux.ibm.com>
1897 * config/rs6000/rs6000.md (bswapdi2): Force address into register
1898 if not in indexed or indirect form.
1899 (bswapdi2_load): Change predicate to indexed_or_indirect_operand.
1900 (bswapdi2_store): Ditto.
1901 * config/rs6000/rs6000.c (rs6000_force_indexed_or_indirect_mem): New
1903 * config/rs6000/rs6000-protos.h (rs6000_force_indexed_or_indirect_mem):
1904 Prototype for helper function.
1906 2018-10-30 Martin Sebor <msebor@redhat.com>
1908 * doc/extend.texi (optimize): Clarify/expand attribute documentation.
1909 (target, pragma GCC optimize, pragma GCC target): Ditto.
1911 2018-10-30 Jonathan Wakely <jwakely@redhat.com>
1913 * doc/extend.texi: Fix prototype and description of
1914 __builtin_expect_with_probability.
1916 2018-10-30 Michael Ploujnikov <michael.ploujnikov@oracle.com>
1918 * cgraph.h (clone_function_name_1): Replaced by new
1919 clone_function_name_numbered that takes name as string; for
1920 privatize_symbol_name_1 use only.
1921 (clone_function_name): Renamed to
1922 clone_function_name_numbered to be explicit about numbering.
1923 (clone_function_name): New two-argument function that does
1924 not number its output.
1925 (clone_function_name): New three-argument function that
1926 takes a number to append to its output.
1927 * cgraphclones.c (duplicate_thunk_for_node):
1928 (clone_function_name_1): Renamed.
1929 (clone_function_name_numbered): Two new functions.
1930 (clone_function_name): Improved documentation.
1931 (cgraph_node::create_virtual_clone): Use clone_function_name_numbered.
1932 * config/rs6000/rs6000.c (make_resolver_func): Ditto.
1933 * final.c (final_scan_insn_1): Use the new clone_function_name
1935 * multiple_target.c (create_dispatcher_calls): Ditto.
1936 (create_target_clone): Ditto.
1937 * omp-expand.c (grid_expand_target_grid_body): Ditto.
1938 * omp-low.c (create_omp_child_function_name): Ditto.
1939 * omp-simd-clone.c (simd_clone_create): Ditto.
1940 * symtab.c (simd_symtab_node::noninterposable_alias): Use the
1941 new clone_function_name without numbering.
1943 2018-10-30 Richard Earnshaw <rearnsha@arm.com>
1945 * alloc-pool.h (base_pool_allocator <TBlockAllocator>::initialize):
1946 Assert that the allocation size is not zero.
1948 2018-10-30 Richard Biener <rguenther@suse.de>
1950 PR tree-optimization/87800
1951 * tree-vect-slp.c (vect_build_slp_tree_2): Reject any
1952 non-induction or reduction PHIs.
1954 2018-10-30 Sameera Deshpande <sameera.deshpande@linaro.org
1956 * config/aarch64/falkor-tag-collision-avoidance.c
1957 (execute_tag_collision_avoidance): Call df_note_add_problem.
1959 2018-10-30 Martin Liska <mliska@suse.cz>
1961 * doc/extend.texi: Fix typo in documentation
1962 of __builtin_expect_with_probability.
1964 2018-10-29 David Malcolm <dmalcolm@redhat.com>
1967 * input.c (get_substring_ranges_for_loc): Detect if
1968 linemap_resolve_location gives us a NULL map, and reject
1971 2018-10-29 Iain Buclaw <ibuclaw@gdcproject.org>
1973 * config.gcc (xstormy16-*-elf): Set tm_d_file.
1975 2018-10-29 David Malcolm <dmalcolm@redhat.com>
1976 Martin Sebor <msebor@redhat.com>
1977 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
1979 * doc/ux.texi (Quoting): New subsection, adapted from material at
1980 https://gcc.gnu.org/wiki/DiagnosticsGuidelines written by
1981 MartinSebor and ManuelLopezIbanez.
1982 (Fix-it hints): Note that fix-it hints shouldn't be marked for
1985 2018-10-29 Kugan Vivekanandarajah <kuganv@linaro.org>
1988 * tree-ssa-loop-niter.c (number_of_iterations_popcount): Fix niter
1991 2018-10-29 Paul A. Clarke <pc@us.ibm.com>
1993 * gcc.target/powerpc/ssse3-check.h: Remove inaccurate comment.
1995 2018-10-29 Paul Koning <ni1d@arrl.net>
1997 * doc/rtl.texi (CONST_WIDE_INT_ELT): Give correct macro name.
1999 2018-10-29 Paul A. Clarke <pc@us.ibm.com>
2001 * gcc/config/rs6000/mmintrin.h (_mm_packs_pi16, _mm_packs_pi32,
2002 _mm_packs_pu16, _mm_unpackhi_pi8, _mm_unpacklo_pi8, _mm_add_pi8,
2003 _mm_add_pi16, _mm_add_pi32, _mm_sub_pi8, _mm_sub_pi16, _mm_sub_pi32,
2004 _mm_cmpgt_pi8, _mm_cmpeq_pi16, _mm_cmpgt_pi16, _mm_cmpeq_pi32,
2005 _mm_cmpgt_pi32, _mm_adds_pi8, _mm_adds_pi16, _mm_adds_pu8,
2006 _mm_adds_pu16, _mm_subs_pi8, _mm_subs_pi16, _mm_subs_pu8,
2007 _mm_subs_pu16, _mm_madd_pi16, _mm_mulhi_pi16, _mm_mullo_pi16,
2008 _mm_sll_pi16, _mm_sra_pi16, _mm_srl_pi16, _mm_set1_pi16, _mm_set1_pi8):
2009 Change 'vector' to '__vector'.
2010 * gcc/config/rs6000/xmmintrin.h (_mm_cvtps_pi32, _mm_cvttps_pi32,
2011 _mm_cvtps_pi16, _mm_cvtps_pi8, _mm_max_pi16, _mm_max_pu8, _mm_min_pi16,
2012 _mm_min_pu8, _mm_mulhi_pu16, _mm_shuffle_pi16, _mm_avg_pu8,
2013 _mm_avg_pu16): Likewise. And, whitespace corrections.
2015 2018-10-29 Richard Biener <rguenther@suse.de>
2017 PR tree-optimization/87785
2018 * tree-vect-slp.c (vect_gather_slp_loads): Only gather
2021 2018-10-29 Olivier Hainque <hainque@adacore.com>
2023 * VXWORKS_NET_LIBS_RTP: New macro, network part of VXWORKS_LIBS_RTP.
2025 2018-10-29 Olivier Hainque <hainque@adacore.com>
2027 * config/vxworks.h (VXWORKS_PERSONALITY): New VxWorks
2028 ports configuration macro, defaults to "gnu".
2029 (VXWORKS_OS_CPP_BUILTINS): Use it. Feed TOOL and
2030 TOOL_FAMILY instead of _VX_TOOL/_VXTOOL_FAMILY.
2032 2018-10-29 Olivier Hainque <hainque@adacore.com>
2034 * config/vxworks.h (CLEAR_INSN_CACHE): #undef before #define.
2036 2018-10-29 Richard Biener <rguenther@suse.de>
2038 PR tree-optimization/87790
2039 * tree-vect-slp.c (vect_mark_slp_stmts): Simplify.
2040 (vect_make_slp_decision): Adjust.
2041 (vect_slp_analyze_bb_1): Likewise.
2042 (vect_detect_hybrid_slp_stmts): Properly union SLP type over
2045 2018-10-29 Richard Biener <rguenther@suse.de>
2047 PR tree-optimization/87785
2048 * tree-vect-slp.c (vect_build_slp_tree_2): Remove loads argument
2050 (vect_build_slp_tree): Likewise.
2051 (vect_gather_slp_loads): New function.
2052 (vect_analyze_slp_instance): Gather loads separately from the
2055 2018-10-29 Martin Liska <mliska@suse.cz>
2057 * Makefile.in: Make dependency to json.o.
2058 * doc/gcov.texi: Document new JSON format, remove
2059 old intermediate format documentation.
2060 * gcov.c (struct function_info): Come up with m_name and
2062 (function_info::function_info): Initialize it.
2063 (function_info::~function_info): Release it.
2064 (main): Rename flag_intermediate_format to flag_json_format.
2065 (print_usage): Describe --json-format.
2066 (process_args): Set flag_json_format.
2067 (output_intermediate_line): Remove.
2068 (output_intermediate_json_line): Likewise.
2069 (get_gcov_intermediate_filename): Return new extension
2071 (output_intermediate_file): Implement JSON emission.
2072 (output_json_intermediate_file): Implement JSON emission.
2073 (generate_results): Use ::get_name for function name.
2074 Handle JSON output file.
2075 (read_graph_file): Use ::get_name instead of cplus_demangle.
2076 (read_count_file): Likewise.
2077 (solve_flow_graph): Likewise.
2078 (add_line_counts): Likewise.
2079 (accumulate_line_counts): Use new flag_json_format.
2080 (output_function_details): Use ::get_name instead of cplus_demangle.
2081 (output_lines): Likewise.
2082 * json.cc (test_writing_literals): Add new tests.
2083 * json.h (class literal): Add new boolean constructor.
2085 2018-10-29 Segher Boessenkool <segher@kernel.crashing.org>
2087 PR rtl-optimization/87701
2088 PR rtl-optimization/87780
2089 * combine.c (make_more_copies): Rewrite.
2091 2018-10-28 Kugan Vivekanandarajah <kuganv@linaro.org>
2093 * doc/generic.texi (ABSU_EXPR): Document.
2094 * match.pd (absu(x)*absu(x) -> x*x): Handle.
2095 (absu(absu(X)) -> absu(X)): Likewise.
2096 (absu(-X) -> absu(X)): Likewise.
2097 (absu(X) where X is nonnegative -> X): Likewise.
2099 2018-10-28 Iain Buclaw <ibuclaw@gdcproject.org>
2101 * Makefile.in (tm_d_file_list, tm_d_include_list): New variables.
2102 (TM_D_H, D_TARGET_DEF, D_TARGET_H, D_TARGET_OBJS): New variables.
2103 (tm_d.h, cs-tm_d.h, default-d.o): New rules.
2104 (d/d-target-hooks-def.h, s-d-target-hooks-def-h): New rules.
2105 (s-tm-texi): Also check timestamp on d-target.def.
2106 (generated_files): Add TM_D_H and d-target-hooks-def.h.
2107 (build/genhooks.o): Also depend on D_TARGET_DEF.
2108 * config.gcc (tm_d_file, d_target_objs, target_has_targetdm): New
2110 * config/aarch64/aarch64-d.c: New file.
2111 * config/aarch64/aarch64-linux.h (GNU_USER_TARGET_D_CRITSEC_SIZE):
2113 * config/aarch64/aarch64-protos.h (aarch64_d_target_versions): New
2115 * config/aarch64/aarch64.h (TARGET_D_CPU_VERSIONS): Define.
2116 * config/aarch64/t-aarch64 (aarch64-d.o): New rule.
2117 * config/arm/arm-d.c: New file.
2118 * config/arm/arm-protos.h (arm_d_target_versions): New prototype.
2119 * config/arm/arm.h (TARGET_D_CPU_VERSIONS): Define.
2120 * config/arm/linux-eabi.h (EXTRA_TARGET_D_OS_VERSIONS): Define.
2121 * config/arm/t-arm (arm-d.o): New rule.
2122 * config/default-d.c: New file.
2123 * config/glibc-d.c: New file.
2124 * config/gnu.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
2125 * config/i386/i386-d.c: New file.
2126 * config/i386/i386-protos.h (ix86_d_target_versions): New prototype.
2127 * config/i386/i386.h (TARGET_D_CPU_VERSIONS): Define.
2128 * config/i386/linux-common.h (EXTRA_TARGET_D_OS_VERSIONS): Define.
2129 (GNU_USER_TARGET_D_CRITSEC_SIZE): Define.
2130 * config/i386/t-i386 (i386-d.o): New rule.
2131 * config/kfreebsd-gnu.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
2132 * config/kopensolaris-gnu.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
2133 * config/linux-android.h (ANDROID_TARGET_D_OS_VERSIONS): Define.
2134 * config/linux.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
2135 * config/mips/linux-common.h (EXTRA_TARGET_D_OS_VERSIONS): Define.
2136 * config/mips/mips-d.c: New file.
2137 * config/mips/mips-protos.h (mips_d_target_versions): New prototype.
2138 * config/mips/mips.h (TARGET_D_CPU_VERSIONS): Define.
2139 * config/mips/t-mips (mips-d.o): New rule.
2140 * config/powerpcspe/linux.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
2141 * config/powerpcspe/linux64.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
2142 * config/powerpcspe/powerpcspe-d.c: New file.
2143 * config/powerpcspe/powerpcspe-protos.h (rs6000_d_target_versions):
2145 * config/powerpcspe/powerpcspe.c (rs6000_output_function_epilogue):
2146 Support GNU D by using 0 as the language type.
2147 * config/powerpcspe/powerpcspe.h (TARGET_D_CPU_VERSIONS): Define.
2148 * config/powerpcspe/t-powerpcspe (powerpcspe-d.o): New rule.
2149 * config/riscv/riscv-d.c: New file.
2150 * config/riscv/riscv-protos.h (riscv_d_target_versions): New
2152 * config/riscv/riscv.h (TARGET_D_CPU_VERSIONS): Define.
2153 * config/riscv/t-riscv (riscv-d.o): New rule.
2154 * config/rs6000/linux.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
2155 * config/rs6000/linux64.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
2156 * config/rs6000/rs6000-d.c: New file.
2157 * config/rs6000/rs6000-protos.h (rs6000_d_target_versions): New
2159 * config/rs6000/rs6000.c (rs6000_output_function_epilogue):
2160 Support GNU D by using 0 as the language type.
2161 * config/rs6000/rs6000.h (TARGET_D_CPU_VERSIONS): Define.
2162 * config/rs6000/t-rs6000 (rs6000-d.o): New rule.
2163 * config/s390/s390-d.c: New file.
2164 * config/s390/s390-protos.h (s390_d_target_versions): New prototype.
2165 * config/s390/s390.h (TARGET_D_CPU_VERSIONS): Define.
2166 * config/s390/t-s390 (s390-d.o): New rule.
2167 * config/sparc/sparc-d.c: New file.
2168 * config/sparc/sparc-protos.h (sparc_d_target_versions): New
2170 * config/sparc/sparc.h (TARGET_D_CPU_VERSIONS): Define.
2171 * config/sparc/t-sparc (sparc-d.o): New rule.
2172 * config/t-glibc (glibc-d.o): New rule.
2173 * configure: Regenerated.
2174 * configure.ac (tm_d_file): New variable.
2175 (tm_d_file_list, tm_d_include_list, d_target_objs): Add substitutes.
2176 * doc/contrib.texi (Contributors): Add self for the D frontend.
2177 * doc/frontends.texi (G++ and GCC): Mention D as a supported language.
2178 * doc/install.texi (Configuration): Mention libphobos as an option for
2179 --enable-shared. Mention d as an option for --enable-languages.
2180 (Testing): Mention check-d as a target.
2181 * doc/invoke.texi (Overall Options): Mention .d, .dd, and .di as file
2182 name suffixes. Mention d as a -x option.
2183 * doc/sourcebuild.texi (Top Level): Mention libphobos.
2184 * doc/standards.texi (Standards): Add section on D language.
2185 * doc/tm.texi: Regenerated.
2186 * doc/tm.texi.in: Add @node for D language and ABI, and @hook for
2187 TARGET_CPU_VERSIONS, TARGET_D_OS_VERSIONS, and TARGET_D_CRITSEC_SIZE.
2188 * dwarf2out.c (is_dlang): New function.
2189 (gen_compile_unit_die): Use DW_LANG_D for D.
2190 (declare_in_namespace): Return module die for D, instead of adding
2191 extra declarations into the namespace.
2192 (gen_namespace_die): Generate DW_TAG_module for D.
2193 (gen_decl_die): Handle CONST_DECLSs for D.
2194 (dwarf2out_decl): Likewise.
2195 (prune_unused_types_walk_local_classes): Handle DW_tag_interface_type.
2196 (prune_unused_types_walk): Handle DW_tag_interface_type same as other
2197 kinds of aggregates.
2198 * gcc.c (default_compilers): Add entries for .d, .dd and .di.
2199 * genhooks.c: Include d/d-target.def.
2201 2018-10-28 Iain Sandoe <iain@sandoe.co.uk>
2204 * config/rs6000/darwin.h (STACK_BOUNDARY): New.
2205 (RS6000_STARTING_FRAME_OFFSET): Adjust to preserve 16byte alignment.
2206 (STACK_DYNAMIC_OFFSET): Likewise.
2208 2018-10-27 Sandra Loosemore <sandra@codesourcery.com>
2211 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Fix
2214 2018-10-26 Tulio Magno Quites Machado Filho <tuliom@linux.ibm.com>
2216 * doc/extend.texi (PowerPC builtins): Fix __builtin_unpack_ibm128
2217 return type and other typos.
2219 2018-10-26 Aaron Sawdey <acsawdey@linux.ibm.com>
2221 * config/rs6000/rs6000-string.c (expand_strncmp_gpr_sequence): Change to
2222 a shorter sequence with fewer branches.
2223 (emit_final_str_compare_gpr): Ditto.
2225 2018-10-26 Paul A. Clarke <pc@us.ibm.com>
2227 * config/rs6000/tmmintrin.h: New file.
2228 * config.gcc (powerpc*-*-*): Add tmmintrin.h to extra_headers.
2230 2018-10-26 Paul A. Clarke <pc@us.ibm.com>
2232 * config/rs6000/mmintrin.h: Enable 32bit compilation.
2233 * config/rs6000/xmmintrin.h: Likewise.
2235 2018-10-26 Paul A. Clarke <pc@us.ibm.com>
2237 * config/rs6000/xmmintrin.h (_mm_extract_pi16): Fix for big-endian.
2239 2018-10-26 Richard Biener <rguenther@suse.de>
2241 * tree-vect-slp.c (vect_mark_slp_stmts): Add visited hash_set
2243 (vect_mark_slp_stmts_relevant): Likewise.
2244 (vect_detect_hybrid_slp_stmts): Likewise.
2245 (vect_bb_slp_scalar_cost): Likewise.
2246 (vect_remove_slp_scalar_calls): Likewise.
2248 2018-10-26 Jan Hubicka <jh@suse.cz>
2250 * ipa-devirt.c (odr_subtypes_equivalent_p): Fix recursion.
2251 (warn_types_mismatch): Fix walk of DECL_NAME.
2252 (odr_types_equivalent_p): Fix overactive assert.
2254 2018-10-26 Richard Biener <rguenther@suse.de>
2256 PR tree-optimization/87105
2257 * tree-vectorizer.h (_slp_tree::refcnt): New member.
2258 * tree-vect-slp.c (vect_free_slp_tree): Decrement and honor
2260 (vect_create_new_slp_node): Initialize refcnt to one.
2262 (scalar_stmts_set_t, bst_fail): Remove.
2263 (vect_build_slp_tree_2): Add bst_map argument and adjust calls.
2264 (vect_build_slp_tree): Add bst_map argument and lookup
2265 already created SLP nodes.
2266 (vect_print_slp_tree): Handle a SLP graph, print SLP node
2268 (vect_slp_rearrange_stmts): Handle a SLP graph.
2269 (vect_analyze_slp_instance): Adjust and free SLP nodes from
2270 the CSE map. Fix indenting.
2271 (vect_schedule_slp_instance): Add short-cut.
2273 2018-10-26 Martin Liska <mliska@suse.cz>
2276 * ipa-prop.c (struct ipa_vr_ggc_hash_traits): Hash with
2277 addr_expr and not with pointers.
2279 2018-10-26 Jan Hubicka <jh@suse.cz>
2281 * tree.c (free_lang_data_in_type): Only check main variants.
2282 * ipa-devirt.c (warn_odr): Make static.
2283 (types_same_for_odr): Drop strict variant.
2284 (types_odr_comparable): Likewise.
2285 (odr_or_derived_type_p): Look for main variants.
2286 (odr_name_hasher::equal): Cleanup comment.
2287 (odr_subtypes_equivalent): Add warn and warned arguments; check main
2289 (type_variants_equivalent_p): break out from ...
2290 (odr_types_equivalent): ... here; go for main variants where needed.
2291 (warn_odr): ... here; turn static.
2292 (warn_types_mismatch): Compare mangled names of main variants.
2293 * ipa-utils.h (types_odr_comparable): Drop strict parameter.
2294 (type_with_linkage_p): Sanity check that we look at main variant.
2295 * lto.c (lto_read_decls): Only consider main variant to be ODR type.
2296 * tree.h (types_same_for_odr): Drop strict argument.
2298 2018-10-26 Richard Biener <rguenther@suse.de>
2300 PR tree-optimization/87746
2301 * tree-vect-data-refs.c (vect_update_misalignment_for_peel):
2302 Simplify and fix WRT strided store groups with size not
2303 equal to step in element count.
2304 (vect_analyze_group_access_1): Dump the whole group.
2306 2018-10-25 Carl Love <cel@us.ibm.com>
2308 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_VSCEDPGT,
2309 P9V_BUILTIN_VEC_VSCEDPLT, P9V_BUILTIN_VEC_VSCEDPEQ,
2310 P9V_BUILTIN_VEC_VSCEDPUO): Rename base overloaded name. Add quad
2311 precicion entry for each overloaded builtin.
2312 * config/rs6000/rs6000-builtin.def (VSCEDPGT, VSCEDPLT, VSCEDPEQ,
2313 VSCEDPUO): Rename overloaded name.
2314 (VSCEDPGT, VSCEQPGT, VSCEDPLT, VSCEQPLT, VSCEDPEQ, VSCEQPEQ,
2315 VSCEDPUO, VSCEQPUO): Add defitions for overloaded builtins.
2316 * config/rs6000/vsx.md (xscmpexpqp_<code>_<mode>): Add
2317 define_expand for xscmpexqp instruction.
2318 (*xscmpexpqp): Add define_insn for the xscmpexqp instruction.
2320 2018-10-25 Bill Schmidt <wschmidt@linux.ibm.com>
2321 Jinsong Ji <jji@us.ibm.com>
2323 * config/rs6000/emmintrin.h (_mm_slli_epi16): Replace deprecated
2324 function with vec_sl.
2325 (_mm_slli_epi32): Likewise.
2326 (_mm_slli_epi64): Likewise.
2327 (_mm_srai_epi16): Replace deprecated function with vec_sra.
2328 (_mm_srai_epi32): Likewise.
2329 (_mm_srli_epi16): Replace deprecated function with vec_sr.
2330 (_mm_srli_epi32): Likewise.
2331 (_mm_srli_epi64): Likewise.
2332 (_mm_sll_epi16): Replace deprecated function with vec_sl.
2333 (_mm_sll_epi32): Likewise.
2334 (_mm_sll_epi64): Likewise.
2335 (_mm_sra_epi16): Replace deprecated function with vec_sra.
2336 (_mm_sra_epi32): Likewise.
2337 (_mm_srl_epi16): Replace deprecated function with vec_sr.
2338 (_mm_srl_epi32): Likewise.
2339 (_mm_srl_epi64): Likewise.
2341 2018-10-25 Bill Schmidt <wschmidt@linux.ibm.com>
2342 Jinsong Ji <jji@us.ibm.com>
2344 * gcc/config/rs6000/emmintrin.h (_mm_sll_epi16): Replace
2345 comparison operators with vec_cmp* for compatibility due to
2346 unfortunate history; clean up formatting and use types more
2348 (_mm_sll_epi32): Likewise.
2349 (_mm_sll_epi64): Likewise.
2350 (_mm_srl_epi16): Likewise.
2351 (_mm_srl_epi32): Likewise.
2352 (_mm_srl_epi64): Likewise.
2354 2018-10-25 Bill Schmidt <wschmidt@linux.ibm.com>
2355 Jinsong Ji <jji@us.ibm.com>
2357 * config/rs6000/emmintrin.h (_mm_sll_epi64): Remove wrong cast.
2358 * config/rs6000/xmmintrin.h (_mm_min_ps): Change m's type to
2359 __vector __bool int. Use vec_cmpgt in preference to deprecated
2360 function vec_vcmpgtfp.
2361 (_mm_max_ps): Likewise.
2363 2018-10-25 Jeff Law <law@redhat.com>
2365 * config/rl78/rl78.c (insn_ok_now): Always re-recognize the insn
2368 2018-10-25 Martin Sebor <msebor@redhat.com>
2370 * doc/extend.texi (aligned): Expand attribute description.
2371 (Alignment): Rename section. Discuss function arguments.
2373 2018-10-25 Jan Hubicka <jh@suse.cz>
2375 * ipa-devirt.c (main_odr_variant): Remove.
2376 (hash_odr_name, types_same_for_odr, types_odr_comparable,
2377 odr_name_hasher::equal, odr_subtypes_equivalent_p):
2378 Drop use of main_odr_variant.
2379 (add_type_duplicate): Silence confused warnings on integer types.
2380 (get_odr_type): Always look for main variant.
2381 (register_odr_type): Simplify.
2383 2018-10-25 Richard Biener <rguenther@suse.de>
2385 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses):
2386 Initialize ng to silence error with release checking bootstrap.
2388 2018-10-25 Richard Biener <rguenther@suse.de>
2390 * tree-if-conv.c: Include tree-ssa-sccvn.h.
2391 (tree_if_conversion): Run CSE on the if-converted loop body.
2393 2018-10-25 Ilya Leoshkevich <iii@linux.ibm.com>
2395 * config/s390/constraints.md (ZL): New constraint.
2396 * config/s390/s390.c (legitimate_pic_operand_p): Accept LARL
2398 * config/s390/s390.md (movdi_larl): Remove.
2399 (movdi_64): Add the LARL alternative.
2401 2018-10-25 Ilya Leoshkevich <iii@linux.ibm.com>
2404 * rtl.c (RTX_CODE_HWINT_P_1): New helper macro.
2405 (RTX_CODE_HWINT_P): New macro.
2406 (rtx_code_size): Use RTX_CODE_HWINT_P ().
2408 2018-10-25 Jan Hubicka <jh@suse.cz>
2410 * ipa-devirt.c (odr_types_equivalent_p): Do not ICE if one of types
2413 2018-10-25 Richard Biener <rguenther@suse.de>
2415 PR tree-optimization/87665
2416 PR tree-optimization/87745
2417 * tree-vectorizer.h (get_earlier_stmt): Remove.
2418 (get_later_stmt): Pick up UID from the original non-pattern stmt.
2420 2018-10-25 Sam Tebbs <sam.tebbs@arm.com>
2422 * options.texi (Deprecated): Move list to Var section.
2424 2018-10-24 Bill Schmidt <wschmidt@linux.ibm.com>
2425 Jinsong Ji <jji@us.ibm.com>
2427 * config/rs6000/emmintrin.h (_mm_cvtpd_epi32): Change deprecated
2428 __vector long to __vector long long.
2429 (_mm_cvtpd_ps): Likewise.
2430 (_mm_cvttpd_epi32): Likewise.
2431 (_mm_cvtpi32_pd): Likewise.
2432 (_mm_unpackhi_epi64): Likewise.
2433 (_mm_unpacklo_epi64): Likewise.
2435 2018-10-24 Segher Boessenkool <segher@kernel.crashing.org>
2437 PR rtl-optimization/87720
2438 * combine.c (make_more_copies): Skip if the dest is pc_rtx.
2440 2018-10-24 Alexandre Oliva <aoliva@redhat.com>
2442 * gimple-ssa-isolate-paths.c
2443 (find_implicit_erroneous_behavior): Do not change code if the
2444 pass is running for warnings only.
2445 (find_explicit_erroneous_behavior): Likewise.
2447 2018-10-24 Michael Meissner <meissner@linux.ibm.com>
2449 * config/rs6000/rs6000.c (TARGET_MANGLE_DECL_ASSEMBLER_NAME):
2450 Define as rs6000_mangle_decl_assembler_name.
2451 (rs6000_mangle_decl_assembler_name): If the user switched from IBM
2452 long double to IEEE long double, switch the names of the long
2453 double built-in functions to be <func>f128 instead of <func>l.
2455 2018-10-24 Martin Sebor <msebor@redhat.com>
2457 * doc/extend.texi (nonnull): List no-argument form. Reference
2458 -fno-delete-null-pointer-checks and -fisolate-erroneous-paths-attribute.
2460 2018-10-24 Richard Biener <rguenther@suse.de>
2462 * tree-ssa-sccvn.c (do_rpo_vn): Free rpo_state.
2464 2018-10-24 Martin Liska <mliska@suse.cz>
2466 PR tree-optimization/84436
2467 * tree-switch-conversion.c (switch_conversion::contains_same_values_p):
2469 (switch_conversion::contains_linear_function_p): New.
2470 (switch_conversion::build_one_array): Support linear
2471 transformation on input.
2472 * tree-switch-conversion.h (struct switch_conversion): Add
2473 contains_linear_function_p declaration.
2475 2018-10-24 Richard Biener <rguenther@suse.de>
2477 * varasm.c (const_hash_1): Return hash of ADDR_EXPR
2478 if its argument is CONSTANT_CLASS_P.
2480 2018-10-11 Jan Hubicka <hubicka@ucw.cz>
2482 * ipa-utils.h (type_with_linkage_p): No longer check for TYPE_STUB_DECL;
2483 it is wrong for forward declarations.
2485 2018-10-24 Ilya Leoshkevich <iii@linux.ibm.com>
2487 * config/s390/s390.c (s390_check_qrst_address): Add the missing
2488 SYMBOL_REF_P () check.
2490 2018-10-24 Richard Biener <rguenther@suse.de>
2492 PR tree-optimization/87105
2493 * tree-vect-data-refs.c (vect_analyze_group_access_1): Adjust
2494 dump classification.
2495 (vect_analyze_data_ref_accesses): Handle duplicate loads and
2496 stores by splitting the affected group after the fact.
2497 * tree-vect-slp.c (vect_build_slp_tree_2): Dump when we
2498 fail the SLP build because of size constraints.
2500 2018-10-24 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2502 * configure.ac (gcc_cv_ld_aligned_shf_merge): New test.
2503 * configure: Regenerate.
2504 * config.in: Regenerate.
2505 * varasm.c (mergeable_string_section): Use readonly_data_section
2506 if linker doesn't support SHF_MERGE with alignment > 8.
2507 (mergeable_constant_section): Likewise.
2509 2018-10-24 Richard Biener <rguenther@suse.de>
2511 PR tree-optimization/84013
2512 * tree-ssa-structalias.c (struct msdi_data): New struct for
2513 marshalling data to walk_stmt_load_store_ops.
2514 (maybe_set_dependence_info): Refactor as callback for
2515 walk_stmt_load_store_ops.
2516 (compute_dependence_clique): Set restrict info on all stmt kinds.
2518 2018-10-24 Martin Liska <mliska@suse.cz>
2520 * cgraph.c (cgraph_node::dump):
2521 Remove reduntant dumps and make tp_first_run dump more compact.
2523 2018-10-24 Richard Biener <rguenther@suse.de>
2525 PR tree-optimization/87665
2526 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Adjust
2529 2018-10-12 Jeff Law <law@redhat.com>
2531 * config/h8300/h8300.c (h8300_expand_prologue): Fix stm generation
2534 2018-10-23 Richard Biener <rguenther@suse.de>
2536 * tree-vrp.c (add_assert_info): Guard dump_printf with
2538 * gimple-ssa-evrp-analyze.c
2539 (evrp_range_analyzer::record_ranges_from_incoming_edge):
2540 Use value_range::ignore_equivs_equal_p.
2542 2018-10-23 Richard Biener <rguenther@suse.de>
2544 PR tree-optimization/87105
2545 PR tree-optimization/87608
2546 * passes.def (pass_all_early_optimizations): Add early phi-opt
2548 * tree-ssa-phiopt.c (value_replacement): Ignore NOPs and predicts in
2549 addition to debug stmts.
2550 (tree_ssa_phiopt_worker): Add early_p argument, do only min/max
2551 and abs replacement early.
2552 * tree-cfg.c (gimple_empty_block_p): Likewise.
2554 2018-10-23 Richard Earnshaw <rearnsha@arm.com>
2557 * config.gcc (arm*-*-netbsdelf*): Default to StrongARM if no CPU
2558 specified to configure.
2559 (arm*-*-*): Use ARM7TDMI as the target CPU if no default provided.
2561 2018-10-23 Richard Biener <rguenther@suse.de>
2563 PR tree-optimization/87700
2564 * tree-ssa-copy.c (set_copy_of_val): Fix change detection logic.
2566 2018-10-23 Jakub Jelinek <jakub@redhat.com>
2569 * config/i386/avx512vlintrin.h (_mm_mask_mullo_epi32): Change type of
2570 second argument from __mmask16 to __mmask8.
2571 * config/i386/avx512vlbwintrin.h (_mm_mask_packus_epi32,
2572 _mm_mask_packs_epi32): Likewise.
2573 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i64scatter_ps):
2575 (_mm512_mask_prefetch_i64scatter_pd): Likewise. Formatting fix.
2577 2018-10-23 Richard Biener <rguenther@suse.de>
2579 * tree-vect-stmts.c (vect_analyze_stmt): Fix typo in comment.
2581 2018-10-23 Richard Biener <rguenther@suse.de>
2583 PR tree-optimization/86144
2584 * tree-vect-stmts.c (vect_analyze_stmt): Prefer -mveclibabi
2585 over simd attribute.
2587 2018-10-23 Richard Biener <rguenther@suse.de>
2589 PR tree-optimization/87693
2590 * tree-ssa-threadedge.c (thread_around_empty_blocks): Handle
2591 the case we do not find the taken edge.
2593 2018-10-22 Bill Schmidt <wschmidt@linux.ibm.com>
2594 Jinsong Ji <jji@us.ibm.com>
2596 * config/rs6000/emmintrin.h (_MM_SHUFFLE2): Comment cleanup.
2597 (_mm_store_pd): Use unaligned vector type for pointer cast.
2598 (_mm_maskmoveu_si128): Likewise.
2599 * config/rs6000/xmmintrin.h (__m128_u): New typedef.
2600 (_mm_store_ps): Use unaligned vector type for pointer cast.
2602 2018-10-22 Paul Koning <ni1d@arrl.net>
2604 * symtab.c (symtab_node::increase_alignment): Correct max
2607 2018-10-22 Yury Gribov <tetra2005@gmail.com>
2609 PR tree-optimization/87633
2610 * match.pd: Do not generate unordered integer comparisons.
2612 2018-10-22 Segher Boessenkool <segher@kernel.crashing.org>
2614 PR rtl-optimization/87600
2615 * combine.c: Add include of expr.h.
2616 (cant_combine_insn_p): Do not combine moves from any hard non-fixed
2617 register to a pseudo.
2618 (make_more_copies): New function, add a copy to a new pseudo after
2619 the moves from hard registers into pseudos.
2620 (rest_of_handle_combine): Declare rebuild_jump_labels_after_combine
2621 later. Call make_more_copies.
2623 2018-10-22 Andrew Stubbs <ams@codesourcery.com>
2625 * lra-constraints.c (process_alt_operands): New local array,
2626 matching_early_clobber. Check matching_early_clobber before
2627 decrementing reject, and set matching_early_clobber after.
2629 2018-10-22 Segher Boessenkool <segher@kernel.crashing.org>
2632 * config/rs6000/rs6000.c (print_operand_address): For unexpected RTL
2633 call output_addr_const and hope for the best.
2635 2018-10-22 Richard Biener <rguenther@suse.de>
2637 * gimple-ssa-evrp-analyze.c
2638 (evrp_range_analyzer::record_ranges_from_incoming_edge): Be
2639 smarter about what ranges to use.
2640 * tree-vrp.c (add_assert_info): Dump here.
2641 (register_edge_assert_for_2): Instead of here at multiple but
2644 * gcc.dg/tree-ssa/evrp12.c: New testcase.
2645 * gcc.dg/predict-6.c: Adjust.
2646 * gcc.dg/tree-ssa/vrp33.c: Disable EVRP.
2647 * gcc.dg/tree-ssa/vrp02.c: Likewise.
2648 * gcc.dg/tree-ssa/cunroll-9.c: Likewise.
2650 2018-10-22 Steven Bosscher <steven@gcc.gnu.org>
2651 Richard Biener <rguenther@suse.de>
2653 * bitmap.h: Update data structure documentation, including a
2654 description of bitmap views as either linked-lists or splay trees.
2655 (struct bitmap_element_def): Update comments for splay tree bitmaps.
2656 (struct bitmap_head_def): Likewise.
2657 (bitmap_list_view, bitmap_tree_view): New prototypes.
2658 (bitmap_initialize_stat): Initialize a bitmap_head's indx and
2660 (bmp_iter_set_init): Assert the iterated bitmaps are in list form.
2661 (bmp_iter_and_init, bmp_iter_and_compl_init): Likewise.
2662 * bitmap.c (bitmap_elem_to_freelist): Unregister overhead of a
2663 released bitmap element here.
2664 (bitmap_element_free): Remove.
2665 (bitmap_elt_clear_from): Work on splay tree bitmaps.
2666 (bitmap_list_link_element): Renamed from bitmap_element_link. Move
2667 this function similar ones such that linked-list bitmap implementation
2668 functions are grouped.
2669 (bitmap_list_unlink_element): Renamed from bitmap_element_unlink,
2670 and moved for grouping.
2671 (bitmap_list_insert_element_after): Renamed from
2672 bitmap_elt_insert_after, and moved for grouping.
2673 (bitmap_list_find_element): New function spliced from bitmap_find_bit.
2674 (bitmap_tree_link_left, bitmap_tree_link_right,
2675 bitmap_tree_rotate_left, bitmap_tree_rotate_right, bitmap_tree_splay,
2676 bitmap_tree_link_element, bitmap_tree_unlink_element,
2677 bitmap_tree_find_element): New functions for splay-tree bitmap
2679 (bitmap_element_link, bitmap_element_unlink, bitmap_elt_insert_after):
2680 Renamed and moved, see above entries.
2681 (bitmap_tree_listify_from): New function to convert part of a splay
2682 tree bitmap to a linked-list bitmap.
2683 (bitmap_list_view): Convert a splay tree bitmap to linked-list form.
2684 (bitmap_tree_view): Convert a linked-list bitmap to splay tree form.
2685 (bitmap_find_bit): Remove.
2686 (bitmap_clear, bitmap_clear_bit, bitmap_set_bit,
2687 bitmap_single_bit_set_p, bitmap_first_set_bit, bitmap_last_set_bit):
2688 Handle splay tree bitmaps.
2689 (bitmap_copy, bitmap_count_bits, bitmap_and, bitmap_and_into,
2690 bitmap_elt_copy, bitmap_and_compl, bitmap_and_compl_into,
2691 bitmap_compl_and_into, bitmap_elt_ior, bitmap_ior, bitmap_ior_into,
2692 bitmap_xor, bitmap_xor_into, bitmap_equal_p, bitmap_intersect_p,
2693 bitmap_intersect_compl_p, bitmap_ior_and_compl,
2694 bitmap_ior_and_compl_into, bitmap_set_range, bitmap_clear_range,
2695 bitmap_hash): Reject trying to act on splay tree bitmaps. Make
2696 corresponding changes to use linked-list specific bitmap_element
2697 manipulation functions as applicable for efficiency.
2698 (bitmap_tree_to_vec): New function.
2699 (debug_bitmap_elt_file): New function split out from ...
2700 (debug_bitmap_file): ... here. Handle splay tree bitmaps.
2701 (bitmap_print): Likewise.
2703 PR tree-optimization/63155
2704 * tree-ssa-propagate.c (ssa_prop_init): Use tree-view for the
2706 * tree-ssa-coalesce.c (coalesce_ssa_name): Populate used_in_copies
2709 2018-10-22 Martin Liska <mliska@suse.cz>
2711 PR tree-optimization/87686
2713 2018-08-29 Martin Liska <mliska@suse.cz>
2715 * tree-switch-conversion.c (switch_conversion::expand):
2716 Strenghten assumption about gswitch statements.
2718 2018-10-22 Martin Liska <mliska@suse.cz>
2720 * ipa-icf.c (sem_item::compare_attributes): Remove.
2721 (sem_item::compare_referenced_symbol_properties): Use
2722 attribute_list_equal instead.
2723 (sem_function::equals_wpa): Likewise.
2724 * ipa-icf.h: Remove compare_attributes.
2726 2018-10-22 Richard Biener <rguenther@suse.de>
2729 * mem-stats.h (mem_usage::operator==): Fix pasto.
2731 2018-10-22 Richard Biener <rguenther@suse.de>
2733 PR tree-optimization/87640
2734 * tree-vrp.c (set_value_range_with_overflow): Decompose
2736 (extract_range_from_binary_expr_1): Adjust.
2738 2018-10-22 Martin Jambor <mjambor@suse.cz>
2740 * tree-eh.h (stmt_could_throw_p): Add function parameter.
2741 (stmt_can_throw_external): Likewise.
2742 (stmt_can_throw_internal): Likewise.
2743 * tree-eh.c (lower_eh_constructs_2): Pass cfun to stmt_could_throw_p.
2744 (lower_eh_constructs_2): Likewise.
2745 (stmt_could_throw_p): Add fun parameter, use it instead of cfun.
2746 (stmt_can_throw_external): Likewise.
2747 (stmt_can_throw_internal): Likewise.
2748 (maybe_clean_eh_stmt_fn): Pass cfun to stmt_could_throw_p.
2749 (maybe_clean_or_replace_eh_stmt): Pass cfun to stmt_could_throw_p.
2750 (maybe_duplicate_eh_stmt_fn): Pass new_fun to stmt_could_throw_p.
2751 (maybe_duplicate_eh_stmt): Pass cfun to stmt_could_throw_p.
2752 (pass_lower_eh_dispatch::execute): Pass cfun to
2753 stmt_can_throw_external.
2754 (cleanup_empty_eh): Likewise.
2755 (verify_eh_edges): Pass cfun to stmt_could_throw_p.
2756 * cgraph.c (cgraph_edge::set_call_stmt): Pass a function to
2757 stmt_can_throw_external instead of pushing it to cfun.
2758 (symbol_table::create_edge): Likewise.
2759 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Pass cfun to
2760 stmt_can_throw_internal.
2761 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Pass cfun
2762 to stmt_could_throw_p.
2763 * gimple-ssa-store-merging.c (handled_load): Pass cfun to
2764 stmt_can_throw_internal.
2765 (pass_store_merging::execute): Likewise.
2766 * gimple-ssa-strength-reduction.c
2767 (find_candidates_dom_walker::before_dom_children): Pass cfun to
2769 * gimplify-me.c (gimple_regimplify_operands): Pass cfun to
2770 stmt_can_throw_internal.
2771 * ipa-pure-const.c (check_call): Pass cfun to stmt_could_throw_p and
2772 to stmt_can_throw_external.
2773 (check_stmt): Pass cfun to stmt_could_throw_p.
2774 (check_stmt): Pass cfun to stmt_can_throw_external.
2775 (pass_nothrow::execute): Likewise.
2776 * trans-mem.c (expand_call_tm): Pass cfun to stmt_can_throw_internal.
2777 * tree-cfg.c (is_ctrl_altering_stmt): Pass cfun to
2778 stmt_can_throw_internal.
2779 (verify_gimple_in_cfg): Pass cfun to stmt_could_throw_p.
2780 (stmt_can_terminate_bb_p): Pass cfun to stmt_can_throw_external.
2781 (gimple_purge_dead_eh_edges): Pass cfun to stmt_can_throw_internal.
2782 * tree-complex.c (expand_complex_libcall): Pass cfun to
2783 stmt_could_throw_p and to stmt_can_throw_internal.
2784 (expand_complex_multiplication): Pass cfun to stmt_can_throw_internal.
2785 * tree-inline.c (copy_edges_for_bb): Likewise.
2786 (maybe_move_debug_stmts_to_successors): Likewise.
2787 * tree-outof-ssa.c (ssa_is_replaceable_p): Pass cfun to
2789 * tree-parloops.c (oacc_entry_exit_ok_1): Likewise.
2790 * tree-sra.c (scan_function): Pass cfun to stmt_can_throw_external.
2791 * tree-ssa-alias.c (stmt_kills_ref_p): Pass cfun to
2792 stmt_can_throw_internal.
2793 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Likewise.
2794 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Pass cfun to
2796 (mark_aliased_reaching_defs_necessary_1): Pass cfun to
2797 stmt_can_throw_internal.
2798 * tree-ssa-forwprop.c (pass_forwprop::execute): Likewise.
2799 * tree-ssa-loop-im.c (movement_possibility): Pass cfun to
2801 * tree-ssa-loop-ivopts.c (find_givs_in_stmt_scev): Likewise.
2802 (add_autoinc_candidates): Pass cfun to stmt_can_throw_internal.
2803 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
2804 (convert_mult_to_fma_1): Likewise.
2805 (convert_to_divmod): Likewise.
2806 * tree-ssa-phiprop.c (propagate_with_phi): Likewise.
2807 * tree-ssa-pre.c (compute_avail): Pass cfun to stmt_could_throw_p.
2808 * tree-ssa-propagate.c
2809 (substitute_and_fold_dom_walker::before_dom_children): Likewise.
2810 * tree-ssa-reassoc.c (suitable_cond_bb): Likewise.
2811 (maybe_optimize_range_tests): Likewise.
2812 (linearize_expr_tree): Likewise.
2813 (reassociate_bb): Likewise.
2814 * tree-ssa-sccvn.c (copy_reference_ops_from_call): Likewise.
2815 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Likewise.
2816 * tree-ssa-strlen.c (adjust_last_stmt): Likewise.
2817 (handle_char_store): Likewise.
2818 * tree-vect-data-refs.c (vect_find_stmt_data_reference): Pass cfun to
2819 stmt_can_throw_internal.
2820 * tree-vect-patterns.c (check_bool_pattern): Pass cfun to
2822 * tree-vect-stmts.c (vect_finish_stmt_generation_1): Likewise.
2823 (vectorizable_call): Pass cfun to stmt_can_throw_internal.
2824 (vectorizable_simd_clone_call): Likewise.
2825 * value-prof.c (gimple_ic): Pass cfun to stmt_could_throw_p.
2826 (gimple_stringop_fixed_value): Likewise.
2828 2018-10-22 Ilya Leoshkevich <iii@linux.ibm.com>
2830 * config/s390/s390.c (s390_loadrelative_operand_p): Accept
2831 literal pool references.
2832 (s390_check_qrst_address): Adapt to the new behavior of
2833 s390_loadrelative_operand_p ().
2835 2018-10-22 H.J. Lu <hongjiu.lu@intel.com>
2838 * config/i386/sse.md (*andnot<mode>3_bcst): New.
2840 2018-10-22 H.J. Lu <hongjiu.lu@intel.com>
2843 * config/i386/sse.md (*<code><mode>3_bcst): New.
2845 2018-10-22 H.J. Lu <hongjiu.lu@intel.com>
2848 * config/i386/sse.md (avx512bcst): Updated for V4SI, V2DI, V8SI,
2849 V4DI, V16SI and V8DI.
2850 (*sub<mode>3<mask_name>_bcst): New.
2851 (*add<mode>3<mask_name>_bcst): Likewise.
2853 2018-10-21 Bill Schmidt <wschmidt@linux.ibm.com>
2854 Jinsong Ji <jji@us.ibm.com>
2856 * config/rs6000/emmintrin.h (_mm_movemask_pd): Replace __vector
2857 __m64 with __vector unsigned long long for compatibility.
2858 (_mm_movemask_epi8): Likewise.
2859 * config/rs6000/xmmintrin.h (_mm_cvtps_pi32): Likewise.
2860 (_mm_cvttps_pi32): Likewise.
2861 (_mm_cvtpi32_ps): Likewise.
2862 (_mm_cvtps_pi16): Likewise.
2863 (_mm_loadh_pi): Likewise.
2864 (_mm_storeh_pi): Likewise.
2865 (_mm_movehl_ps): Likewise.
2866 (_mm_movelh_ps): Likewise.
2867 (_mm_loadl_pi): Likewise.
2868 (_mm_storel_pi): Likewise.
2869 (_mm_movemask_ps): Likewise.
2870 (_mm_shuffle_pi16): Likewise.
2872 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
2875 * config/i386/avx512fintrin.h (_mm512_fnmsub_round_pd): Use
2876 __builtin_ia32_vfnmsubpd512_mask.
2877 (_mm512_mask_fnmsub_round_pd): Likewise.
2878 (_mm512_fnmsub_pd): Likewise.
2879 (_mm512_mask_fnmsub_pd): Likewise.
2880 (_mm512_maskz_fnmsub_round_pd): Use
2881 __builtin_ia32_vfnmsubpd512_maskz.
2882 (_mm512_maskz_fnmsub_pd): Likewise.
2883 (_mm512_fnmsub_round_ps): Use __builtin_ia32_vfnmsubps512_mask.
2884 (_mm512_mask_fnmsub_round_ps): Likewise.
2885 (_mm512_fnmsub_ps): Likewise.
2886 (_mm512_mask_fnmsub_ps): Likewise.
2887 (_mm512_maskz_fnmsub_round_ps): Use
2888 __builtin_ia32_vfnmsubps512_maskz.
2889 (_mm512_maskz_fnmsub_ps): Likewise.
2890 * config/i386/avx512vlintrin.h (_mm256_mask_fnmsub_pd): Use
2891 __builtin_ia32_vfnmsubpd256_mask.
2892 (_mm256_maskz_fnmsub_pd): Use __builtin_ia32_vfnmsubpd256_maskz.
2893 (_mm_mask_fnmsub_pd): Use __builtin_ia32_vfmaddpd128_mask
2894 (_mm_maskz_fnmsub_pd): Use __builtin_ia32_vfnmsubpd128_maskz.
2895 (_mm256_mask_fnmsub_ps): Use __builtin_ia32_vfnmsubps256_mask.
2896 (_mm256_mask_fnmsub_ps): Use __builtin_ia32_vfnmsubps256_mask.
2897 (_mm256_maskz_fnmsub_ps): Use __builtin_ia32_vfnmsubps256_maskz.
2898 (_mm_mask_fnmsub_ps): Use __builtin_ia32_vfnmsubps128_mask.
2899 (_mm_maskz_fnmsub_ps): Use __builtin_ia32_vfnmsubps128_maskz.
2900 * config/i386/fmaintrin.h (_mm_fnmsub_pd): Use
2901 __builtin_ia32_vfnmsubpd.
2902 (_mm256_fnmsub_pd): Use __builtin_ia32_vfnmsubpd256.
2903 (_mm_fnmsub_ps): Use __builtin_ia32_vfnmsubps.
2904 (_mm256_fnmsub_ps): Use __builtin_ia32_vfnmsubps256.
2905 (_mm_fnmsub_sd): Use __builtin_ia32_vfnmsubsd3.
2906 (_mm_fnmsub_ss): Use __builtin_ia32_vfnmsubss3.
2907 * config/i386/i386-builtin.def: Add
2908 __builtin_ia32_vfnmsubpd256_mask,
2909 __builtin_ia32_vfnmsubpd256_maskz,
2910 __builtin_ia32_vfnmsubpd128_mask,
2911 __builtin_ia32_vfnmsubpd128_maskz,
2912 __builtin_ia32_vfnmsubps256_mask,
2913 __builtin_ia32_vfnmsubps256_maskz,
2914 __builtin_ia32_vfnmsubps128_mask,
2915 __builtin_ia32_vfnmsubps128_maskz,
2916 __builtin_ia32_vfnmsubpd512_mask,
2917 __builtin_ia32_vfnmsubpd512_maskz,
2918 __builtin_ia32_vfnmsubps512_mask,
2919 __builtin_ia32_vfnmsubps512_maskz, __builtin_ia32_vfnmsubss3,
2920 __builtin_ia32_vfnmsubsd3, __builtin_ia32_vfnmsubps,
2921 __builtin_ia32_vfnmsubpd, __builtin_ia32_vfnmsubps256 and.
2922 __builtin_ia32_vfnmsubpd256.
2923 * config/i386/sse.md (fma4i_fnmsub_<mode>): New.
2924 (<avx512>_fnmsub_<mode>_maskz<round_expand_name>): Likewise.
2925 (*<sd_mask_codefor>fma_fnmsub_<mode><sd_maskz_name>_bcst_1):
2927 (*<sd_mask_codefor>fma_fnmsub_<mode><sd_maskz_name>_bcst_2):
2929 (*<sd_mask_codefor>fma_fnmsub_<mode><sd_maskz_name>_bcst_3):
2931 (fmai_vmfnmsub_<mode><round_name>): Likewise.
2933 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
2936 * config/i386/avx512fintrin.h (_mm512_fnmadd_round_pd): Use
2937 __builtin_ia32_vfnmaddpd512_mask.
2938 (_mm512_mask_fnmadd_round_pd): Likewise.
2939 (_mm512_fnmadd_pd): Likewise.
2940 (_mm512_mask_fnmadd_pd): Likewise.
2941 (_mm512_maskz_fnmadd_round_pd): Use
2942 __builtin_ia32_vfnmaddpd512_maskz.
2943 (_mm512_maskz_fnmadd_pd): Likewise.
2944 (_mm512_fnmadd_round_ps): Use __builtin_ia32_vfnmaddps512_mask.
2945 (_mm512_mask_fnmadd_round_ps): Likewise.
2946 (_mm512_fnmadd_ps): Likewise.
2947 (_mm512_mask_fnmadd_ps): Likewise.
2948 (_mm512_maskz_fnmadd_round_ps): Use
2949 __builtin_ia32_vfnmaddps512_maskz.
2950 (_mm512_maskz_fnmadd_ps): Likewise.
2951 * config/i386/avx512vlintrin.h (_mm256_mask_fnmadd_pd): Use
2952 __builtin_ia32_vfnmaddpd256_mask.
2953 (_mm256_maskz_fnmadd_pd): Use __builtin_ia32_vfnmaddpd256_maskz.
2954 (_mm_mask_fnmadd_pd): Use __builtin_ia32_vfmaddpd128_mask
2955 (_mm_maskz_fnmadd_pd): Use __builtin_ia32_vfnmaddpd128_maskz.
2956 (_mm256_mask_fnmadd_ps): Use __builtin_ia32_vfnmaddps256_mask.
2957 (_mm256_mask_fnmadd_ps): Use __builtin_ia32_vfnmaddps256_mask.
2958 (_mm256_maskz_fnmadd_ps): Use __builtin_ia32_vfnmaddps256_maskz.
2959 (_mm_mask_fnmadd_ps): Use __builtin_ia32_vfnmaddps128_mask.
2960 (_mm_maskz_fnmadd_ps): Use __builtin_ia32_vfnmaddps128_maskz.
2961 * config/i386/fmaintrin.h (_mm_fnmadd_pd): Use
2962 __builtin_ia32_vfnmaddpd.
2963 (_mm256_fnmadd_pd): Use __builtin_ia32_vfnmaddpd256.
2964 (_mm_fnmadd_ps): Use __builtin_ia32_vfnmaddps.
2965 (_mm256_fnmadd_ps): Use __builtin_ia32_vfnmaddps256.
2966 (_mm_fnmadd_sd): Use __builtin_ia32_vfnmaddsd3.
2967 (_mm_fnmadd_ss): Use __builtin_ia32_vfnmaddss3.
2968 * config/i386/i386-builtin.def: Add
2969 __builtin_ia32_vfnmaddpd256_mask,
2970 __builtin_ia32_vfnmaddpd256_maskz,
2971 __builtin_ia32_vfnmaddpd128_mask,
2972 __builtin_ia32_vfnmaddpd128_maskz,
2973 __builtin_ia32_vfnmaddps256_mask,
2974 __builtin_ia32_vfnmaddps256_maskz,
2975 __builtin_ia32_vfnmaddps128_mask,
2976 __builtin_ia32_vfnmaddps128_maskz,
2977 __builtin_ia32_vfnmaddpd512_mask,
2978 __builtin_ia32_vfnmaddpd512_maskz,
2979 __builtin_ia32_vfnmaddps512_mask,
2980 __builtin_ia32_vfnmaddps512_maskz, __builtin_ia32_vfnmaddss3,
2981 __builtin_ia32_vfnmaddsd3, __builtin_ia32_vfnmaddps,
2982 __builtin_ia32_vfnmaddpd, __builtin_ia32_vfnmaddps256 and.
2983 __builtin_ia32_vfnmaddpd256.
2984 * config/i386/sse.md (fma4i_fnmadd_<mode>): New.
2985 (<avx512>_fnmadd_<mode>_maskz<round_expand_name>): Likewise.
2986 (*<sd_mask_codefor>fma_fnmadd_<mode><sd_maskz_name>_bcst_1):
2988 (*<sd_mask_codefor>fma_fnmadd_<mode><sd_maskz_name>_bcst_2):
2990 (*<sd_mask_codefor>fma_fnmadd_<mode><sd_maskz_name>_bcst_3):
2992 (fmai_vmfnmadd_<mode><round_name>): Likewise.
2994 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
2997 * config/i386/avx512fintrin.h (_mm512_fmsub_round_pd): Use
2998 __builtin_ia32_vfmsubpd512_mask.
2999 (_mm512_mask_fmsub_round_pd): Likewise.
3000 (_mm512_fmsub_pd): Likewise.
3001 (_mm512_mask_fmsub_pd): Likewise.
3002 (_mm512_maskz_fmsub_round_pd): Use
3003 __builtin_ia32_vfmsubpd512_maskz.
3004 (_mm512_maskz_fmsub_pd): Likewise.
3005 (_mm512_fmsub_round_ps): Use __builtin_ia32_vfmsubps512_mask.
3006 (_mm512_mask_fmsub_round_ps): Likewise.
3007 (_mm512_fmsub_ps): Likewise.
3008 (_mm512_mask_fmsub_ps): Likewise.
3009 (_mm512_maskz_fmsub_round_ps): Use
3010 __builtin_ia32_vfmsubps512_maskz.
3011 (_mm512_maskz_fmsub_ps): Likewise.
3012 * config/i386/avx512vlintrin.h (_mm256_mask_fmsub_pd): Use
3013 __builtin_ia32_vfmsubpd256_mask.
3014 (_mm256_maskz_fmsub_pd): Use __builtin_ia32_vfmsubpd256_maskz.
3015 (_mm_mask_fmsub_pd): Use __builtin_ia32_vfmaddpd128_mask
3016 (_mm_maskz_fmsub_pd): Use __builtin_ia32_vfmsubpd128_maskz.
3017 (_mm256_mask_fmsub_ps): Use __builtin_ia32_vfmsubps256_mask.
3018 (_mm256_mask_fmsub_ps): Use __builtin_ia32_vfmsubps256_mask.
3019 (_mm256_maskz_fmsub_ps): Use __builtin_ia32_vfmsubps256_maskz.
3020 (_mm_mask_fmsub_ps): Use __builtin_ia32_vfmsubps128_mask.
3021 (_mm_maskz_fmsub_ps): Use __builtin_ia32_vfmsubps128_maskz.
3022 * config/i386/fmaintrin.h (_mm_fmsub_pd): Use
3023 __builtin_ia32_vfmsubpd.
3024 (_mm256_fmsub_pd): Use __builtin_ia32_vfmsubpd256.
3025 (_mm_fmsub_ps): Use __builtin_ia32_vfmsubps.
3026 (_mm256_fmsub_ps): Use __builtin_ia32_vfmsubps256.
3027 (_mm_fmsub_sd): Use __builtin_ia32_vfmsubsd3.
3028 (_mm_fmsub_ss): Use __builtin_ia32_vfmsubss3.
3029 * config/i386/i386-builtin.def: Add
3030 __builtin_ia32_vfmsubpd256_mask,
3031 __builtin_ia32_vfmsubpd256_maskz,
3032 __builtin_ia32_vfmsubpd128_mask,
3033 __builtin_ia32_vfmsubpd128_maskz,
3034 __builtin_ia32_vfmsubps256_mask,
3035 __builtin_ia32_vfmsubps256_maskz,
3036 __builtin_ia32_vfmsubps128_mask,
3037 __builtin_ia32_vfmsubps128_maskz,
3038 __builtin_ia32_vfmsubpd512_mask,
3039 __builtin_ia32_vfmsubpd512_maskz,
3040 __builtin_ia32_vfmsubps512_mask,
3041 __builtin_ia32_vfmsubps512_maskz, __builtin_ia32_vfmsubss3,
3042 __builtin_ia32_vfmsubsd3, __builtin_ia32_vfmsubps,
3043 __builtin_ia32_vfmsubpd, __builtin_ia32_vfmsubps256 and.
3044 __builtin_ia32_vfmsubpd256.
3045 * config/i386/sse.md (fma4i_fmsub_<mode>): New.
3046 (<avx512>_fmsub_<mode>_maskz<round_expand_name>): Likewise.
3047 (*<sd_mask_codefor>fma_fmsub_<mode><sd_maskz_name>_bcst_1):
3049 (*<sd_mask_codefor>fma_fmsub_<mode><sd_maskz_name>_bcst_2):
3051 (*<sd_mask_codefor>fma_fmsub_<mode><sd_maskz_name>_bcst_3):
3053 (fmai_vmfmsub_<mode><round_name>): Likewise.
3055 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
3057 * config/i386/sse.md (*<plusminus_insn><mode>3<mask_name>_bcst_1):
3058 Remove plus. Renamed to ...
3059 (*sub<mode>3<mask_name>_bcst): This.
3060 (*add<mode>3<mask_name>_bcst_2): Renamede to ...
3061 (*add<mode>3<mask_name>_bcst): This.
3063 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
3066 * config/i386/sse.md (*mul<mode>3<mask_name>_bcst): New.
3068 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
3071 * i386/avx512vlintrin.h (_mm256_or_epi32): New.
3072 (_mm_or_epi32): Likewise.
3073 (_mm256_xor_epi32): Likewise.
3074 (_mm_xor_epi32): Likewise.
3075 (_mm256_or_epi64): Likewise.
3076 (_mm_or_epi64): Likewise.
3077 (_mm256_xor_epi64): Likewise.
3078 (_mm_xor_epi64): Likewise.
3080 2018-10-20 H.J. Lu <hongjiu.lu@intel.com>
3083 * config/i386/sse.md (*<avx512>_div<mode>3<mask_name>_bcst): New.
3085 2018-10-20 Jakub Jelinek <jakub@redhat.com>
3088 * varasm.c (decode_addr_const): Handle COMPOUND_LITERAL_EXPR.
3090 2018-10-20 Andreas Schwab <schwab@linux-m68k.org>
3092 * doc/ux.texi: Move @section directly after @node.
3094 2018-10-19 Jakub Jelinek <jakub@redhat.com>
3098 * omp-low.c (check_omp_nesting_restrictions): Diagnose ordered without
3099 depend closely nested inside of loop with ordered clause with
3102 2018-10-19 David Malcolm <dmalcolm@redhat.com>
3104 * Makefile.in (TEXI_GCCINT_FILES): Add ux.texi.
3105 * doc/gccint.texi: Include ux.texi and use it in top-level menu.
3106 * doc/ux.texi: New file.
3108 2018-10-19 Segher Boessenkool <segher@kernel.crashing.org>
3110 * config/rs6000/rs6000.h (REG_ALLOC_ORDER): Move 68 (that is, CR0) to
3111 be the first CR field allocated.
3113 2018-10-19 Richard Biener <rguenther@suse.de>
3116 * config/i386/i386.c (ix86_builtin_vectorization_cost): Use
3117 TYPE_VECTOR_SUBPARTS and avoid relying on vector mode.
3119 2018-10-19 H.J. Lu <hongjiu.lu@intel.com>
3122 * config/i386/sse.md
3123 (*<plusminus_insn><mode>3<mask_name>_bcst_1): New.
3124 (*add<mode>3<mask_name>_bcst_2): Likewise.
3126 2018-10-19 H.J. Lu <hongjiu.lu@intel.com>
3128 * config/i386/sse.md
3129 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_1):
3130 Replace nonimmediate_operand with register_operand.
3131 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_2):
3133 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_3):
3136 2018-10-19 Ilya Leoshkevich <iii@linux.ibm.com>
3138 PR rtl-optimization/87596
3139 * lra-constraints.c (spill_hard_reg_in_range): Use INSN_P () +
3140 lra_get_insn_recog_data () instead of lra_insn_recog_data[]
3141 for instructions in FROM..TO range.
3143 2018-10-19 Eric Botcazou <ebotcazou@adacore.com>
3145 * cfgexpand.c (expand_one_var): Use specific wording in error message
3146 for non-local frame variables.
3147 * stor-layout.c (layout_decl): Do not issue a warning for them.
3149 2018-10-19 Robin Dapp <rdapp@linux.ibm.com>
3151 * haifa-sched.c (priority): Add force_recompute parameter.
3152 (apply_replacement): Call priority () with force_recompute = true.
3153 (restore_pattern): Likewise.
3155 2018-10-18 H.J. Lu <hongjiu.lu@intel.com>
3157 * simplify-rtx.c (simplify_subreg): Limit mask of vec_merge to
3158 HOST_BITS_PER_WIDE_INT.
3159 (test_vector_ops_duplicate): Likewise.
3161 2018-10-18 H.J. Lu <hongjiu.lu@intel.com>
3164 * config/i386/sse.md (VF_AVX512): New.
3165 (avx512bcst): Likewise.
3166 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_1):
3168 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_2):
3170 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_3):
3173 2018-10-18 Jonathan Wakely <jwakely@redhat.com>
3175 * doc/invoke.texi (-dumpversion): Improve grammar.
3176 (-dumpfullversion): Make more consistent with -dumpversion.
3178 2018-10-18 Uros Bizjak <ubizjak@gmail.com>
3180 * config/i386/i386.c (ix86_emit_fp_unordered_jump):
3181 Set JUMP_LABEL to the jump insn.
3182 (ix86_emit_i387_log1p): Use ix86_expand_branch to expand branch.
3183 Predict emitted jump and add label to jump insn.
3185 2018-10-18 David Malcolm <dmalcolm@redhat.com>
3187 PR tree-optimization/87562
3188 * input.c (get_substring_ranges_for_loc): Use
3189 LRK_SPELLING_LOCATION rather than LRK_MACRO_EXPANSION_POINT when
3190 getting the linemap for the endpoint. Verify that it's either
3191 in the same linemap as the start point's spelling location, or
3192 at least in the same file.
3194 2018-10-18 Richard Biener <rguenther@suse.de>
3196 * config/i386/i386.c (ix86_builtin_vectorization_cost): Do not
3197 feed width-specific load/store costs through ix86_vec_cost.
3198 * config/i386/x86-tune-costs.h (athlon_cost): Adjust.
3199 (k8_cost): Likewise.
3200 (bdver_cost): Likewise.
3201 (znver1_cost): Likewise.
3202 (btver1_cost): Likewise.
3203 (btver2_cost): Likewise.
3205 2018-10-18 H.J. Lu <hongjiu.lu@intel.com>
3207 * simplify-rtx.c (simplify_subreg): Call simplify_gen_subreg
3208 to simplify subreg of vec_merge.
3210 2018-10-18 Richard Biener <rguenther@suse.de>
3212 * config/i386/i386.c: Fix costing of vector FMA.
3214 2018-10-18 Richard Biener <rguenther@suse.de>
3216 * config/i386/i386.c (ix86_vec_cost): Remove !parallel path
3218 (ix86_builtin_vectorization_cost): For vec_construct properly
3219 cost insertion into SSE regs.
3220 (...): Adjust calls to ix86_vec_cost.
3222 2018-10-18 Richard Biener <rguenther@suse.de>
3226 2018-02-07 Richard Biener <rguenther@suse.de>
3228 PR tree-optimization/84204
3229 * tree-chrec.c (chrec_fold_plus_1): Remove size limiting in
3232 2018-10-18 H.J. Lu <hongjiu.lu@intel.com>
3235 * simplify-rtx.c (simplify_subreg): Simplify subreg of vec_merge
3237 (test_vector_ops_duplicate): Add test for a scalar subreg of a
3238 VEC_MERGE of a VEC_DUPLICATE.
3240 2018-10-17 Joseph Myers <joseph@codesourcery.com>
3242 * doc/cpp.texi (__STDC_VERSION__): Document C2X handling.
3243 * doc/invoke.texi (-std=c2x, -std=gnu2x): Document new options.
3244 * doc/standards.texi (C Language): Document C2X.
3245 * dwarf2out.c (highest_c_language), config/rl78/rl78.c
3246 (rl78_option_override): Handle "GNU C2X" language name.
3248 2018-10-17 Joseph Myers <joseph@codesourcery.com>
3250 * doc/invoke.texi (-std=c17), doc/standards.texi (C Language):
3251 Document C17 as published in 2018.
3253 2018-10-17 Eric Botcazou <ebotcazou@adacore.com>
3256 * fold-const.c (fold_truth_andor_1): If the right side is not constant,
3257 bail out if both sides do not have the same storage order.
3259 2018-10-17 Aldy Hernandez <aldyh@redhat.com>
3261 * bitmap.c (bitmap_head::dump): New.
3262 * bitmap.h (bitmap_head): Add dump().
3263 * gimple-ssa-evrp-analyze.c
3264 (evrp_range_analyzer::try_find_new_range): Adjust for value_range API.
3265 (evrp_range_analyzer::set_ssa_range_info): Same.
3266 (evrp_range_analyzer::record_ranges_from_phis): Same.
3267 (evrp_range_analyzer::record_ranges_from_stmt): Same.
3268 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Same.
3269 * gimple-ssa-sprintf.c (get_int_range): Same.
3270 (format_integer): Same.
3271 (sprintf_dom_walker::handle_gimple_call): Same.
3272 * ipa-cp.c (ipcp_vr_lattice::meet_with_1): Same.
3273 (ipcp_vr_lattice::top_p): Same.
3274 (ipcp_vr_lattice::bottom_p): Same.
3275 (ipcp_vr_lattice::set_to_bottom): Same.
3276 (ipa_vr_operation_and_type_effects): Same.
3277 (propagate_vr_across_jump_function): Same.
3278 (ipcp_store_vr_results): Same.
3279 * ipa-prop.c (struct ipa_vr_ggc_hash_traits): Same.
3280 (ipa_print_node_jump_functions_for_edge): Same.
3281 (ipa_get_value_range): Same.
3282 (ipa_compute_jump_functions_for_edge): Same.
3283 (ipa_write_jump_function): Same.
3284 * tree-ssa-dom.c (simplify_stmt_for_jump_threading): Same.
3285 * tree-ssa-threadedge.c (record_temporary_equivalences_from_phis):
3287 * vr-values.c (set_value_range_to_nonnegative): Same.
3288 (set_value_range_to_truthvalue): Same.
3289 (vr_values::get_value_range): Same.
3290 (vr_values::set_defs_to_varying): Same.
3291 (vr_values::update_value_range): Same.
3292 (symbolic_range_based_on_p): Same.
3293 (vr_values::op_with_boolean_value_range_p): Same.
3294 (vr_values::extract_range_for_var_from_comparison_expr): Same.
3295 (vr_values::extract_range_from_ssa_name): Same.
3296 (vr_values::extract_range_from_binary_expr): Same.
3297 (vr_values::extract_range_from_unary_expr): Same.
3298 (vr_values::extract_range_from_cond_expr): Same.
3299 (vr_values::extract_range_from_comparison): Same.
3300 (vr_values::check_for_binary_op_overflow): Same.
3301 (vr_values::extract_range_basic): Same.
3302 (vr_values::extract_range_from_assignment): Same.
3303 (compare_ranges): Same.
3304 (compare_range_with_value): Same.
3305 (vr_values::adjust_range_with_scev): Same.
3306 (vrp_valueize): Same.
3307 (vrp_valueize_1): Same.
3308 (vr_values::get_vr_for_comparison): Same.
3309 (vr_values::compare_name_with_value): Same.
3310 (vr_values::compare_names): Same.
3311 (vr_values::vrp_evaluate_conditional): Same.
3312 (find_case_label_ranges): Same.
3313 (vr_values::vrp_visit_switch_stmt): Same.
3314 (vr_values::extract_range_from_phi_node): Same.
3315 (vr_values::simplify_div_or_mod_using_ranges): Same.
3316 (vr_values::simplify_bit_ops_using_ranges): Same.
3317 (test_for_singularity): Same.
3318 (range_fits_type_p): Same.
3319 (vr_values::simplify_cond_using_ranges_1): Same.
3320 (vr_values::simplify_switch_using_ranges): Same.
3321 (vr_values::simplify_float_conversion_using_ranges): Same.
3322 (vr_values::two_valued_val_range_p): Same.
3323 (vr_values::add_equivalence): Move to value_range::equiv_add.
3324 * vr-values.h (vr_values::add_equivalence): Remove.
3325 (VR_INITIALIZER): Remove.
3326 * tree-vrp.c (value_range::set): New.
3327 (value_range::equiv_add): New.
3328 (value_range::value_range): New.
3329 (value_range::deep_copy): New.
3330 (value_range::check): New.
3331 (value_range::equal_p): New.
3332 (value_range::ignore_equivs_equal_p): New.
3333 (value_range::operator==): New.
3334 (value_range::operator!=): New.
3335 (value_range::symbolic_p): New.
3336 (value_range::numeric_p): New.
3337 (value_range::set_undefined): New.
3338 (value_range::set_varying): New.
3339 (value_range::may_contain_p): New.
3340 (value_range::equiv_clear): New.
3341 (value_range::singleton_p): New.
3342 (value_range::intersect): New.
3343 (value_range::dump): New.
3344 (value_range::set_and_canonicalize): New.
3345 (set_value_range): Adjust for value_range API.
3346 (set_value_range_to_undefined): Same.
3347 (set_value_range_to_varying): Same.
3348 (set_and_canonicalize_value_range): Same.
3349 (set_value_range_to_nonnull): Same.
3350 (set_value_range_to_null): Same.
3351 (range_is_null): Same.
3352 (range_is_nonnull): Same.
3353 (range_int_cst_p): Same.
3354 (range_int_cst_singleton_p): Same.
3355 (symbolic_range_p): Same.
3356 (range_includes_zero_p): Same.
3357 (value_range_constant_singleton): Same.
3358 (vrp_set_zero_nonzero_bits): Same.
3359 (ranges_from_anti_range): Same.
3360 (extract_range_into_wide_ints): Same.
3361 (extract_range_from_multiplicative_op): Same.
3362 (set_value_range_with_overflow): Same.
3363 (extract_range_from_binary_expr_1): Same.
3364 (extract_range_from_unary_expr): Same.
3365 (dump_value_range): Same.
3366 (debug_value_range): Same.
3367 (vrp_prop::check_array_ref): Same.
3368 (vrp_prop::check_mem_ref): Same.
3369 (vrp_prop::vrp_initialize): Same.
3370 (vrp_prop::visit_stmt): Same.
3371 (intersect_ranges): Same.
3372 (vrp_prop::visit_phi): Same.
3373 (vrp_prop::vrp_finalize): Same.
3374 (determine_value_range_1): Same.
3375 (determine_value_range): Same.
3376 (vrp_intersect_ranges_1): Rename to...
3377 (vrp_intersect_1): this.
3378 (vrp_intersect_ranges): Rename to...
3379 (value_range::intersect_helper): ...this.
3380 (vrp_meet_1): Rename to...
3381 (value_range::union_helper): ...this.
3382 (vrp_meet): Rename to...
3383 (value_range::union_): ...this.
3384 (copy_value_range): Remove.
3385 * tree-vrp.h (struct value_range): Rewrite into a proper class.
3386 (value_range::vrtype): New.
3387 (value_range::type): New.
3388 (value_range::equiv): New.
3389 (value_range::min): New.
3390 (value_range::max): New.
3391 (value_range::varying_p): New.
3392 (value_range::undefined_p): New.
3393 (value_range::null_p): New.
3394 (value_range::equiv_add): New.
3395 (copy_value_range): Remove.
3397 2018-10-17 David Malcolm <dmalcolm@redhat.com>
3399 * Makefile.in (SELFTEST_TARGETS): New.
3400 (selftest) Change from s-selftest-c to $(SELFTEST_TARGETS).
3401 (C_SELFTEST_FLAGS, C_SELFTEST_DEPS, s-selftest-c, selftest-c-gdb)
3402 (selftest-gdb, selftest-c-valgrind, selftest-valgrind): Move to
3404 (CPP_SELFTEST_FLAGS, CPP_SELFTEST_DEPS, s-selftest-c++)
3405 (selftest-c++-gdb, selftest-c++-valgrind): Move to
3407 * configure: Regenerate.
3408 * configure.ac (selftest_languages): New.
3410 2018-10-17 Aldy Hernandez <aldyh@redhat.com>
3412 * tree-vrp.c (extract_range_from_multiplicative_op): Remove
3413 overflow wraps argument.
3414 (extract_range_from_binary_expr_1): Do not pass overflow wraps to
3415 wide_int_range_multiplicative_op.
3416 * wide-int-range.cc (wide_int_range_mult_wrapping): Remove
3417 overflow wraps argument.
3418 (wide_int_range_multiplicative_op): Same.
3419 (wide_int_range_lshift): Same.
3420 (wide_int_range_div): Same.
3421 * wide-int-range.h (wide_int_range_multiplicative_op): Same.
3422 (wide_int_range_lshift): Same.
3423 (wide_int_range_div): Same.
3425 2018-10-17 Aldy Hernandez <aldyh@redhat.com>
3427 * wide-int-range.h (wide_int_range_shift_undefined_p): Adjust to
3428 use sign as argument.
3429 * tree-vrp.c (extract_range_from_binary_expr_1): Pass sign to
3430 wide_int_range_shift_undefined_p.
3432 2018-10-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3434 * config/aarch64/aarch64.md (despeculate_copy<ALLI_TI:mode>):
3436 (@despeculate_copy<ALLI_TI:mode>): ... This.
3437 * config/aarch64/aarch64.c (aarch64_speculation_safe_value): Remove
3440 2018-10-16 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3442 * config.gcc: Obsolete *-*-solaris2.10*.
3443 * doc/install.texi (Specific, *-*-solaris2*): Document it.
3445 2018-10-12 Jeff Law <law@redhat.com>
3447 * config/ft32/ft32.md (ft32_general_movsrc_operand): Disable
3448 reg + sym +- const_int addressing modes.
3450 2018-10-15 David Malcolm <dmalcolm@redhat.com>
3452 * common.opt (fdiagnostics-minimum-margin-width=): New option.
3453 * diagnostic-show-locus.c (layout::layout): Apply the minimum
3455 (layout::start_annotation_line): Only print up to 3 of the
3456 margin character, to avoid touching the left-hand side.
3457 (selftest::test_diagnostic_show_locus_fixit_lines): Update for
3458 minimum margin width, as set by test_diagnostic_context's ctor.
3459 (selftest::test_fixit_insert_containing_newline): Likewise.
3460 (selftest::test_fixit_insert_containing_newline_2): Likewise.
3461 (selftest::test_line_numbers_multiline_range): Clear
3462 dc.min_margin_width.
3463 * diagnostic.c (diagnostic_initialize): Initialize
3465 * diagnostic.h (struct diagnostic_context): Add field
3467 * doc/invoke.texi: Add -fdiagnostics-minimum-margin-width=.
3468 * opts.c (common_handle_option): Handle
3469 OPT_fdiagnostics_minimum_margin_width_.
3470 * selftest-diagnostic.c
3471 (selftest::test_diagnostic_context::test_diagnostic_context):
3472 Initialize min_margin_width to 6.
3473 * toplev.c (general_init): Initialize global_dc->min_margin_width.
3475 2018-10-15 David Malcolm <dmalcolm@redhat.com>
3477 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
3478 Fix usage of "error_at_rich_loc" in the comment.
3480 2018-10-15 Renlin Li <renlin.li@arm.com>
3483 * tree-vectorizer.c (try_vectorize_loop_1): Don't use
3484 if-conversioned loop when it contains ifn with types not
3485 supported by backend.
3486 * internal-fn.c (expand_direct_optab_fn): Add an assert.
3487 (direct_internal_fn_supported_p): New helper function.
3488 * internal-fn.h (direct_internal_fn_supported_p): Declare.
3490 2018-10-15 Jakub Jelinek <jakub@redhat.com>
3493 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512F_UNSET):
3494 Remove OPTION_MASK_ISA_AVX5124FMAPS_UNSET and
3495 OPTION_MASK_ISA_AVX5124VNNIW_UNSET.
3497 2018-10-15 Bin Cheng <bin.cheng@linux.alibaba.com>
3499 PR tree-optimization/87022
3500 * tree-loop-distribution.c (pg_add_dependence_edges): Check all
3501 bits in dist vector rather than the first one.
3503 2018-10-15 Richard Biener <rguenther@suse.de>
3506 * tree-ssa-structalias.c (struct vls_data): Add escaped_p member.
3507 (visit_loadstore): When a used restrict tag escaped verify that
3508 the points-to solution of "other" pointers do not include
3510 (compute_dependence_clique): If a used restrict tag escaped
3511 communicated that down to visit_loadstore.
3513 2018-10-15 Andreas Krebbel <krebbel@linux.ibm.com>
3515 * config/s390/s390.c (s390_expand_vec_init): Force vector element
3516 into reg if it isn't a general operand.
3518 2018-10-14 H.J. Lu <hongjiu.lu@intel.com>
3521 * config/i386/sse.md (*vec_dupv2di): Add register source to
3524 2018-10-14 H.J. Lu <hongjiu.lu@intel.com>
3527 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512F_UNSET):
3528 Add OPTION_MASK_ISA_AVX512IFMA_UNSET,
3529 OPTION_MASK_ISA_AVX5124FMAPS_UNSET and
3530 OPTION_MASK_ISA_AVX5124VNNIW_UNSET.
3532 2018-10-13 Eric Botcazou <ebotcazou@adacore.com>
3534 * dwarf2cfi.c (struct dw_trace_info): Add args_size_defined_for_eh.
3535 (notice_args_size): Set it in the current trace if no insn that can
3536 throw internally has been seen yet.
3537 (connect_traces): When connecting args_size between traces, allow the
3538 incoming values not to match if there is an insn setting it before the
3539 first insn that can throw internally; in that case, force the creation
3540 of a CFI note on this latter insn.
3542 2018-10-13 Jonathan Wakely <jwakely@redhat.com>
3544 * opt-problem.h (opt_wrapper): Use template-argument-list when naming
3545 the base class, because using the injected-class-name was not clearly
3546 specified until DR 176.
3548 2018-10-12 Paul Koning <ni1d@arrl.net>
3550 * config/pdp11/pdp11.md (doloop_end): New expander.
3551 (doloop_end_insn): renamed from "doloop_end".
3552 (addqi3): New pattern.
3553 (subqi3): New pattern.
3554 * config/pdp11/predicates.md (incdec_operand): New predicate.
3556 2018-10-12 Yury Gribov <tetra2005@gmail.com>
3559 * real.c (format_helper::can_represent_integral_type_p): New function
3560 * real.h (format_helper::can_represent_integral_type_p): Ditto.
3561 * match.pd: New pattern.
3563 2018-10-12 Alexandre Oliva <oliva@adacore.com>
3565 * configure.ac: Introduce --enable-large-address-aware
3566 to define MINGW_DEFAULT_LARGE_ADDR_AWARE.
3567 * doc/install.texi: Document it.
3568 * configure, config.in: Rebuilt.
3569 * config/i386/mingw32.h (LINK_SPEC_LARGE_ADDR_AWARE): Define,
3570 based on MINGW_DEFAULT_LARGE_ADDR_AWARE.
3571 (LINK_SPEC): Insert it.
3572 * config/i386/mingw-w64.h: Likewise.
3574 * cygming.h (HAVE_GAS_ALIGNED_COMM): Fallback-define.
3576 2018-10-12 Peter Bergner <bergner@linux.ibm.com>
3578 PR rtl-optimization/87600
3579 * ira-lives (non_conflicting_reg_copy_p): Disable for non LRA targets.
3581 2018-10-12 Paul Koning <ni1d@arrl.net>
3583 * doc/md.texi (doloop_end): Document that the pattern code may
3584 need to check operand mode.
3586 2018-10-12 Wilco Dijkstra <wdijkstr@arm.com>
3588 * config/aarch64/aarch64.md (zero_extendsidi2_aarch64): Add alternatives
3589 to zero-extend between int and floating-point registers.
3590 (load_pair_zero_extendsidi2_aarch64): Add alternative for zero-extended
3591 ldp into floating-point registers. Add type and arch attributes.
3592 (zero_extend<SHORT:mode><GPI:mode>2_aarch64): Add arch attribute.
3593 Use f_loads for type attribute.
3595 2018-10-11 Martin Sebor <msebor@redhat.com>
3597 * doc/extend.texi (attribute packed): Correct typos.
3599 2018-10-11 Martin Sebor <msebor@redhat.com>
3601 * doc/extend.texi (attribute flatten): Mention interaction with
3604 2018-10-11 Jan Hubicka <hubicka@ucw.cz>
3607 * cgraphclones.c (cgraph_node::create_version_clone_with_body):
3608 Set new_decl virtual flag to zero.
3610 2018-10-11 Martin Sebor <msebor@redhat.com>
3613 * doc/extend.texi (attribute format_arg): Discuss using multiple
3614 attributes on a single function.
3616 2018-10-11 Giuliano Belinassi <giuliano.belinassi@usp.br>
3618 PR tree-optimization/86829
3619 * match.pd (sin (atan (x))): New simplification rules.
3620 (cos (atan (x))): Likewise.
3621 * real.c (build_sinatan_real): New function.
3622 * real.h (build_sinatan_real): Prototype.
3624 2018-10-11 Will Schmidt <will_schmidt@vnet.ibm.com>
3626 * config/rs6000/rs6000.c (map_to_integral_tree_type): New helper
3628 (fold_mergeeo_helper): New helper function.
3629 (rs6000_gimple_fold_builtin): Add hooks for vec_mergee and vec_mergeo
3630 intrinsics. Correct some whitespace indentation issues.
3632 2018-10-11 Wilco Dijkstra <wdijkstr@arm.com>
3635 * config/aarch64/aarch64.c (aarch64_mask_and_shift_for_ubfiz_p):
3636 Use HOST_WIDE_INT_1U for shift.
3638 2018-10-11 Doug Rupp <rupp@adacore.com>
3639 Olivier Hainque <hainque@adacore.com>
3641 * config/rs6000/vxworks.h (VXWORKS_RELAX_LINK_SPEC): New macro.
3642 Pass --relax to the linker for RTPs.
3643 (LINK_SPEC): Append VXWORKS_RELAX_LINK_SPEC.
3645 2018-10-11 Andrew Stubbs <ams@codesourcery.com>
3646 Jan Hubicka <jh@suse.cz>
3647 Martin Jambor <mjambor@suse.cz>
3649 * print-rtl.c (print_rtx_operand_codes_E_and_V): Print how many times
3650 the same elements are repeated rather than printing all of them.
3651 * read-rtl.c (rtx_reader::read_rtx_operand): Recognize and expand
3652 "repeated" elements.
3653 * read-rtl-function.c (test_loading_repeat): New function.
3654 (read_rtl_function_c_tests): Call test_loading_repeat.
3655 * rtl-tests.c (test_dumping_repeat): New function.
3656 (rtl_tests_c_tests): Call test_dumping_repeat.
3658 2018-10-11 Richard Biener <rguenther@suse.de>
3660 * config/i386/x86-tune-costs.h (bdver?_memcpy, bdver?_memset,
3661 bdver?_cost): Unify to ...
3662 (bdver_memcpy, bdver_memset, bdver_cost): ... this.
3663 * config/i386/i386.c (processor_cost_table): Adjust.
3665 2018-10-10 Eric Botcazou <ebotcazou@adacore.com>
3668 * cgraphunit.c (cgraph_node::expand_thunk): Force DECL_IGNORED_P on
3669 the thunk when expanding to GIMPLE.
3671 2018-10-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
3673 * varasm.c (mergeable_string_section): Don't try to move zero-length
3674 strings to the merge section.
3676 2018-10-10 Uros Bizjak <ubizjak@gmail.com>
3679 * config/i386/mmx.md (const_vector 0 -> mem splitter): New splitter.
3681 2018-10-10 Jakub Jelinek <jakub@redhat.com>
3684 * config/i386/i386-builtin.def (IX86_BUILTIN_RDPMC): Move from args set
3685 to special_args set.
3687 2018-10-10 Richard Biener <rguenther@suse.de>
3689 * config/i386/sse.md (reduc_plus_scal_v8df, reduc_plus_scal_v4df,
3690 reduc_plus_scal_v2df, reduc_plus_scal_v16sf, reduc_plus_scal_v8sf,
3691 reduc_plus_scal_v4sf): Merge into pattern reducing to half width
3692 and recursing and pattern terminating the recursion on SSE
3693 vector width using ix86_expand_reduc.
3694 (reduc_sminmax_scal_<mode>): Split into part reducing to half
3695 width and recursing and SSE2 vector variant doing the final
3696 reduction with ix86_expand_reduc.
3697 (reduc_uminmax_scal_<mode>): Likewise for the AVX512 variants
3698 with terminating the recursion at AVX level, splitting that
3701 2018-10-09 David Malcolm <dmalcolm@redhat.com>
3703 * genmatch.c (error_cb): Rename to...
3704 (diagnostic_cb): ...this, converting int params to enums.
3705 (fatal_at): Update for renaming.
3706 (warning_at): Likewise.
3708 * input.c (selftest::ebcdic_execution_charset::apply):
3709 Update for renaming of...
3710 (selftest::ebcdic_execution_charset::on_error): ...this, renaming
3712 (selftest::ebcdic_execution_charset::on_diagnostic): ...this,
3713 converting level and reason to enums.
3714 (class selftest::lexer_error_sink): Rename to...
3715 (class selftest::lexer_test_options): ...this, renaming field
3716 "m_errors" to "m_diagnostics".
3717 (selftest::lexer_test_options::apply): Update for renaming of...
3718 (selftest::lexer_test_options::on_error): ...this, renaming to...
3719 (selftest::lexer_test_options::on_diagnostic): ...this
3720 converting level and reason to enums.
3721 (selftest::test_lexer_string_locations_raw_string_unterminated):
3722 Update for renamings.
3723 * opth-gen.awk (struct cpp_reason_option_codes_t): Use enum for
3726 2018-10-09 Paul A. Clarke <pc@us.ibm.com>
3728 * config.gcc (powerpc*-*-*): Add pmmintrin.h to extra_headers.
3729 * config/rs6000/pmmintrin.h: New file.
3731 2018-10-09 Eric Botcazou <ebotcazou@adacore.com>
3733 PR tree-optimization/86659
3734 * gimple-match.h (gimple_match_op constructors): Initialize reverse.
3736 2018-10-09 Richard Biener <rguenther@suse.de>
3738 PR tree-optimization/63155
3739 * tree-ssa-structalias.c: Include tree-ssa.h.
3740 (get_constraint_for_ssa_var): For undefs return nothing_id.
3741 (find_func_aliases): Cleanup PHI handling.
3743 2018-10-09 Richard Biener <rguenther@suse.de>
3745 * tree-vectorizer.c (dump_stmt_cost): Fix cut&paste missing
3748 2018-10-09 Martin Liska <mliska@suse.cz>
3750 * asan.c (asan_emit_stack_protection): If a stack variable
3751 is located in a same file as current function, then emit
3752 line info into variable definition string.
3754 2018-10-08 Eric Botcazou <ebotcazou@adacore.com>
3756 * print-rtl.c (rtx_writer::print_rtx_operand_code_i): Print column
3759 2018-10-08 Eric Botcazou <ebotcazou@adacore.com>
3761 * cgraphunit.c (cgraph_node::expand_thunk): Do not force DECL_IGNORED_P
3764 2018-10-08 Eric Botcazou <ebotcazou@adacore.com>
3766 PR tree-optimization/86659
3767 * gimple-match.h (struct gimple_match_op): Add reverse field.
3768 (gimple_match_op::set_op): New overloaded method.
3769 * gimple-match-head.c (maybe_build_generic_op) <BIT_FIELD_REF>: Set
3770 the REF_REVERSE_STORAGE_ORDER flag on the value.
3771 (gimple_simplify) <GIMPLE_ASSIGN>: For BIT_FIELD_REF, propagate the
3772 REF_REVERSE_STORAGE_ORDER flag and avoid simplifying if it is set.
3774 2018-10-08 Richard Sandiford <richard.sandiford@arm.com>
3777 * gimple-ssa-backprop.c (backprop::intersect_uses): Use
3778 FOR_EACH_IMM_USE_FAST instead of FOR_EACH_IMM_USE_STMT.
3780 2018-10-08 H.J. Lu <hongjiu.lu@intel.com>
3783 * config/i386/avx512fintrin.h (_mm512_mask_fmaddsub_round_pd):
3784 Defined with __builtin_ia32_vfmaddsubpd512_mask.
3786 2018-10-08 Richard Biener <rguenther@suse.de>
3788 * config/i386/x86-tune-costs.h (znver1_cost): Make AVX256 vector loads
3789 cost the same as AVX128 ones.
3791 2018-10-08 Paul Koning <ni1d@arrl.net>
3793 * config/pdp11/pdp11-protos.h (output_block_move): Remove.
3794 (expand_block_move): New function.
3795 * config/pdp11/pdp11.c (output_block_move): Remove.
3796 (expand_block_move): New function.
3797 * config/pdp11/pdp11.h (MOVE_RATIO): New definition.
3798 * config/pdp11/pdp11.md (movmemhi): Use expand_block_move.
3799 (*movmemhi1): Remove.
3801 2018-10-08 Robin Dapp <rdapp@linux.ibm.com>
3803 * config/s390/2827.md: Increase latencies for some FP instructions.
3805 2018-10-08 Richard Biener <rguenther@suse.de>
3807 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost):
3809 * tree-vectorizer.c (dump_stmt_cost): Add cost param and dump it.
3810 * tree-vectorizer.h (dump_stmt_cost): Adjust.
3811 (add_stmt_cost): Dump return value of the hook.
3813 2018-10-08 Richard Biener <rguenther@suse.de>
3815 PR tree-optimization/63155
3816 * tree-ssa-propagate.c (add_ssa_edge): Do cheap check first.
3817 (ssa_propagation_engine::ssa_propagate): Remove redundant
3818 bitmap bit clearing.
3820 2018-10-05 Peter Bergner <bergner@linux.ibm.com>
3822 PR rtl-optimization/86939
3823 PR rtl-optimization/87479
3824 * ira.h (non_conflicting_reg_copy_p): New prototype.
3825 * ira-lives.c (ignore_reg_for_conflicts): New static variable.
3826 (make_hard_regno_dead): Don't add conflicts for register
3827 ignore_reg_for_conflicts.
3828 (make_object_dead): Likewise.
3829 (non_conflicting_reg_copy_p): New function.
3830 (process_bb_node_lives): Set ignore_reg_for_conflicts for copies.
3831 Remove special conflict handling of REAL_PIC_OFFSET_TABLE_REGNUM.
3832 * lra-lives.c (ignore_reg_for_conflicts): New static variable.
3833 (make_hard_regno_dead): Don't add conflicts for register
3834 ignore_reg_for_conflicts. Remove special conflict handling of
3835 REAL_PIC_OFFSET_TABLE_REGNUM. Remove now unused argument
3836 check_pic_pseudo_p and update callers.
3837 (mark_pseudo_dead): Don't add conflicts for register
3838 ignore_reg_for_conflicts.
3839 (process_bb_lives): Set ignore_reg_for_conflicts for copies.
3841 2018-10-05 Andrew Waterman <andrew@sifive.com>
3842 Jim Wilson <jimw@sifive.com>
3844 * config/riscv/riscv.md (f<quiet_pattern>_quiet<ANYF:mode><X:mode>4):
3845 Add define_expand. Add ! HONOR_SNANS check to current pattern. Add
3846 new pattern using HONOR_SNANS that emits one extra instruction.
3848 2018-10-05 Segher Boessenkool <segher@kernel.crashing.org>
3850 * config/rs6000/rs6000.md (unnamed mfcr scc_comparison_operator
3851 patterns): Merge SI and DI patterns to a GPR pattern.
3852 (unnamed define_insn and define_split for record form of that): Merge
3853 to a single define_insn_and_split pattern.
3855 2018-10-05 David Malcolm <dmalcolm@redhat.com>
3858 * input.c (expand_location_to_spelling_point): Add param "aspect"
3859 and use rather than hardcoding LOCATION_ASPECT_CARET.
3860 (get_substring_ranges_for_loc): Handle the case of a single token
3861 within a macro expansion.
3862 * input.h (expand_location_to_spelling_point): Add "aspect" param,
3863 defaulting to LOCATION_ASPECT_CARET.
3865 2018-10-05 Paul Koning <ni1d@arrl.net>
3867 * config/pdp11/pdp11.c (TARGET_CXX_GUARD_TYPE): Define.
3868 (TARGET_CXX_CLASS_DATA_ALWAYS_COMDAT): Define.
3869 (TARGET_CXX_LIBRARY_RTTI_COMDAT): Define.
3870 (pdp11_guard_type): New function.
3872 2018-10-05 Paul Koning <ni1d@arrl.net>
3874 * config/pdp11/pdp11.h (FLOAT_TYPE_SIZE): Always 32.
3875 * config/pdp11/pdp11.opt (mfloat32): Remove.
3877 * doc/invoke.texi (pdp11 -mfloat32): Remove:
3878 (pdp11 -mfloat64): Remove.
3880 2018-10-05 Uros Bizjak <ubizjak@gmail.com>
3882 * config/i386/i386.md (*cmpxf_cc_i387): Remove pattern.
3883 (*cmp<mode>_cc_i387): Ditto.
3884 (*cmpu<mode>_cc_i387): Ditto.
3885 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
3886 * config/i386/i386.c (ix86_expand_fp_compare): Remove
3888 <case IX86_FPCMP_SAHF>: Do not generate pattern with HImode clobber.
3889 Emit x86_sahf_1 pattern.
3890 (ix86_expand_compare): Update call to ix86_expand_fp_compare.
3891 (ix86_expand_carry_flag_compare): Ditto.
3893 2018-10-05 Uros Bizjak <ubizjak@gmail.com>
3895 * config/i386/i386.md (*cmpxf_i387): Change operand 2 predicate
3896 to reg_or_0_operand. Add "C" constraint.
3897 (*cmpxf_cc_i387): Ditto.
3898 (*cmp<mode>_i387): Change operand 2 predicate
3899 to nonimm_or_0_operand. Add "C" constraint.
3900 (*cmp<mode>_cc_i387): Ditto.
3901 (*cmp<mode>_0_i387): Remove insn pattern.
3902 (*cmp<mode>_0_cc_i387): Ditto.
3904 2018-10-05 Uros Bizjak <ubizjak@gmail.com>
3906 * config/i386/constraints.md ("C"): Do not depend on TARGET_SSE.
3907 * config/i386/predicates.md (nonimm_or_0_operand): Rename
3908 from vector_move_operand. Update all uses.
3910 2018-10-05 Martin Sebor <msebor@redhat.com>
3912 PR tree-optimization/87490
3913 * builtins.c (expand_builtin_strnlen): Handle a null data.decl
3916 2018-10-05 Richard Biener <rguenther@suse.de>
3918 PR tree-optimization/63155
3919 * tree-ssa-ccp.c (ccp_propagate::visit_phi): Avoid excess
3920 vertical space in dumpfiles.
3921 * tree-ssa-propagate.h
3922 (ssa_propagation_engine::process_ssa_edge_worklist): Remove.
3923 * tree-ssa-propagate.c (cfg_blocks_back): New global.
3924 (ssa_edge_worklist_back): Likewise.
3925 (curr_order): Likewise.
3926 (cfg_blocks_get): Remove abstraction.
3927 (cfg_blocks_add): Likewise.
3928 (cfg_blocks_empty_p): Likewise.
3929 (add_ssa_edge): Add to current or next worklist based on
3931 (add_control_edge): Likewise.
3932 (ssa_propagation_engine::process_ssa_edge_worklist): Fold
3934 (ssa_propagation_engine::ssa_propagate): ... here. Unify
3935 iteration from CFG and SSA edge worklist so we process
3936 everything in RPO order, prioritizing forward progress
3938 (ssa_prop_init): Allocate new worklists, do not dump
3940 (ssa_prop_fini): Free new worklists.
3942 2018-10-05 Richard Biener <rguenther@suse.de>
3944 * tree-core.h (tree_block::abstract_flag): Remove.
3945 (tree_block::block_num): Make full 32bits.
3946 * tree.def (BLOCK): Remove docs about BLOCK_ABSTRACT.
3947 * tree.h (BLOCK_ABSTRACT): Remove.
3948 * dwarf2out.c (gen_lexical_block_die): Remove dead code
3949 resulting from BLOCK_ABSTRACT being always false.
3950 (gen_inlined_subroutine_die): Likewise.
3951 (gen_block_die): Likewise.
3952 * tree.c (block_ultimate_origin): Likewise.
3953 * tree-pretty-print.c (dump_block_node): Remove code dealing
3954 with BLOCK_ABSTRACT.
3955 * tree-ssa-live.c (dump_scope_block): Likewise.
3956 * tree-streamer-in.c (unpack_ts_block_value_fields): Likewise.
3957 * tree-streamer-out.c (pack_ts_block_value_fields): Likewise.
3959 2018-10-05 Richard Biener <rguenther@suse.de>
3961 * config/i386/i386.c (ix86_add_stmt_cost): When scalar cost
3962 is asked for initialize mode to the component mode of the
3965 2018-10-05 H.J. Lu <hongjiu.lu@intel.com>
3968 * config/i386/gnu-user.h (ASM_SPEC): Don't pass -msse2avx to
3969 assembler for -mavx.
3970 * config/i386/gnu-user64.h (ASM_SPEC): Likewise.
3972 2018-10-05 Segher Boessenkool <segher@kernel.crashing.org>
3975 * config/rs6000/rs6000-builtin.def (RS6000_BUILTIN_SET_FPSCR_DRN): Use
3977 * config/rs6000/rs6000.md (rs6000_set_fpscr_rn): Require the operand
3978 to be DImode. When using mffscrn, force the operand to a register.
3980 2018-10-04 Uros Bizjak <ubizjak@gmail.com>
3982 * config/i386/i386.md (*fop_<X87MODEF:mode>_2_i387): Macroize insn
3983 from *fop_<MODEF:mode>_2_i387 and *fop_xf_2_i387 using
3984 X87MODEF mode iterator.
3985 (*fop_<X87MODEF:mode>_3_i387): Macroize insn from
3986 *fop_<MODEF:mode>_3_i387 and *fop_xf_3_i387 using
3987 X87MODEF mode iterator.
3989 2018-10-04 Vinay Kumar <vinay.kumar@blackfigtech.com>
3991 * doc/invoke.texi (-Wno-prio-ctor-dtor): Document new warning
3992 -Wno-prio-ctor-dtor.
3994 2018-10-04 David Malcolm <dmalcolm@redhat.com>
3996 * Makefile.in (OBJS): Add opt-problem.o.
3997 * dump-context.h: Include "selftest.h.
3998 (selftest::temp_dump_context): New forward decl.
3999 (class dump_context): Make friend of class
4000 selftest::temp_dump_context.
4001 (dump_context::dump_loc_immediate): New decl.
4002 (class dump_pretty_printer): Move here from dumpfile.c.
4003 (class temp_dump_context): Move to namespace selftest.
4004 (temp_dump_context::temp_dump_context): Add param
4005 "forcibly_enable_dumping".
4006 (selftest::verify_dumped_text):
4007 (ASSERT_DUMPED_TEXT_EQ): Move here from dumpfile.c.
4008 (selftest::verify_item):
4009 (ASSERT_IS_TEXT): Move here from dumpfile.c.
4010 (ASSERT_IS_TREE): Likewise.
4011 (ASSERT_IS_GIMPLE): Likewise.
4012 * dumpfile.c (dump_context::dump_loc): Move immediate dumping
4014 (dump_context::dump_loc_immediate): ...this new function.
4015 (class dump_pretty_printer): Move to dump-context.h.
4016 (dump_switch_p_1): Don't enable MSG_PRIORITY_REEMITTED.
4017 (opt_info_switch_p_1): Enable MSG_PRIORITY_REEMITTED.
4018 (temp_dump_context::temp_dump_context): Move to "selftest"
4019 namespace. Add param "forcibly_enable_dumping", and use it to
4020 conditionalize the use of m_pp;
4021 (selftest::verify_dumped_text): Make non-static.
4022 (ASSERT_DUMPED_TEXT_EQ): Move to dump-context.h.
4023 (selftest::verify_item): Make non-static.
4024 (ASSERT_IS_TEXT): Move to dump-context.h.
4025 (ASSERT_IS_TREE): Likewise.
4026 (ASSERT_IS_GIMPLE): Likewise.
4027 (selftest::test_capture_of_dump_calls): Pass "true" for new
4028 param of temp_dump_context.
4029 * dumpfile.h (enum dump_flag): Add MSG_PRIORITY_REEMITTED, adding
4030 it to MSG_ALL_PRIORITIES. Update values of TDF_COMPARE_DEBUG and
4032 * opt-problem.cc: New file.
4033 * opt-problem.h: New file.
4034 * optinfo-emit-json.cc
4035 (selftest::test_building_json_from_dump_calls): Pass "true" for
4036 new param of temp_dump_context.
4037 * optinfo.cc (optinfo_kind_to_dump_flag): New function.
4038 (optinfo::emit_for_opt_problem): New function.
4039 (optinfo::emit): Clarity which emit_item is used.
4040 * optinfo.h (optinfo::get_dump_location): New accessor.
4041 (optinfo::emit_for_opt_problem): New decl.
4042 (optinfo::emit): Make const.
4043 * selftest-run-tests.c (selftest::run_tests): Call
4044 selftest::opt_problem_cc_tests.
4045 * selftest.h (selftest::opt_problem_cc_tests): New decl.
4046 * tree-data-ref.c (dr_analyze_innermost): Convert return type from
4047 bool to opt_result, converting fprintf messages to
4048 opt_result::failure_at calls. Add "stmt" param for use by the
4050 (create_data_ref): Pass "stmt" to the dr_analyze_innermost call.
4051 (runtime_alias_check_p): Convert return type from bool to
4052 opt_result, converting dump_printf calls to
4053 opt_result::failure_at, using the statement DDR_A for their
4055 (find_data_references_in_stmt): Convert return type from bool to
4056 opt_result, converting "return false" to opt_result::failure_at
4058 * tree-data-ref.h: Include "opt-problem.h".
4059 (dr_analyze_innermost): Convert return type from bool to opt_result,
4060 and add a const gimple * param.
4061 (find_data_references_in_stmt): Convert return type from bool to
4063 (runtime_alias_check_p): Likewise.
4064 * tree-predcom.c (find_looparound_phi): Pass "init_stmt" to
4065 dr_analyze_innermost.
4066 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test):
4067 Convert return type from bool to opt_result, adding a message for
4068 the PARAM_VECT_MAX_VERSION_FOR_ALIAS_CHECKS zero case.
4069 (vect_analyze_data_ref_dependence): Convert return type from bool
4070 to opt_result. Change sense of return type from "false"
4071 effectively meaning "no problems" to "false" meaning a problem,
4072 so that "return false" becomes "return opt_result::success".
4073 Convert "return true" calls to opt_result::failure_at, using
4074 the location of statement A rather than vect_location.
4075 (vect_analyze_data_ref_dependences): Convert return type from bool
4077 (verify_data_ref_alignment): Likewise, converting dump_printf_loc
4078 calls to opt_result::failure_at, using the stmt location rather
4080 (vect_verify_datarefs_alignment): Convert return type from bool
4082 (vect_enhance_data_refs_alignment): Likewise. Split local "stat"
4083 into multiple more-tightly-scoped copies.
4084 (vect_analyze_data_refs_alignment): Convert return type from bool
4086 (vect_analyze_data_ref_accesses): Likewise, converting a
4087 "return false" to a "return opt_result::failure_at", adding a
4089 (vect_prune_runtime_alias_test_list): Convert return type from
4090 bool to opt_result, converting dump_printf_loc to
4091 opt_result::failure_at. Add a %G to show the pertinent statement,
4092 and use the stmt's location rather than vect_location.
4093 (vect_find_stmt_data_reference): Convert return type from
4094 bool to opt_result, converting dump_printf_loc to
4095 opt_result::failure_at, using stmt's location.
4096 (vect_analyze_data_refs): Convert return type from bool to
4097 opt_result. Convert "return false" to "return
4098 opt_result::failure_at", adding messages as needed.
4099 * tree-vect-loop.c (vect_determine_vf_for_stmt_1): Convert return
4100 type from bool to opt_result.
4101 (vect_determine_vf_for_stmt): Likewise.
4102 (vect_determine_vectorization_factor): Likewise, converting
4103 dump_printf_loc to opt_result::failure_at, using location of phi
4104 rather than vect_location.
4105 (vect_analyze_loop_form_1): Convert return type from bool to
4106 opt_result, converting dump_printf_loc calls, retaining the use of
4108 (vect_analyze_loop_form): Convert return type from loop_vec_info
4109 to opt_loop_vec_info.
4110 (vect_analyze_loop_operations): Convert return type from bool to
4111 opt_result, converting dump_printf_loc calls, using the location
4112 of phi/stmt rather than vect_location where available. Convert
4113 various "return false" to "return opt_result::failure_at" with
4114 "unsupported phi" messages.
4115 (vect_get_datarefs_in_loop): Convert return type from bool to
4116 opt_result. Add a message for the
4117 PARAM_LOOP_MAX_DATAREFS_FOR_DATADEPS failure.
4118 (vect_analyze_loop_2): Convert return type from bool to
4119 opt_result. Ensure "ok" is set to a opt_result::failure_at before
4120 each "goto again;", adding new messages where needed.
4121 Add "unsupported grouped {store|load}" messages.
4122 (vect_analyze_loop): Convert return type from loop_vec_info to
4124 * tree-vect-slp.c (vect_analyze_slp): Convert return type from
4126 * tree-vect-stmts.c (process_use): Likewise, converting
4127 dump_printf_loc call and using stmt location, rather than
4129 (vect_mark_stmts_to_be_vectorized): Likeise.
4130 (vect_analyze_stmt): Likewise, adding a %G.
4131 (vect_get_vector_types_for_stmt): Convert return type from bool to
4132 opt_result, converting dump_printf_loc calls and using stmt
4133 location, rather than vect_location.
4134 (vect_get_mask_type_for_stmt): Convert return type from tree to
4135 opt_tree, converting dump_printf_loc calls and using stmt location.
4136 * tree-vectorizer.c: Include "opt-problem.h.
4137 (try_vectorize_loop_1): Flag "Analyzing loop at" dump message as
4138 MSG_PRIORITY_INTERNALS. Convert local "loop_vinfo" from
4139 loop_vec_info to opt_loop_vec_info. If if fails, and dumping is
4140 enabled, use it to report at the top level "couldn't vectorize
4141 loop" followed by the problem.
4142 * tree-vectorizer.h (opt_loop_vec_info): New typedef.
4143 (vect_mark_stmts_to_be_vectorized): Convert return type from bool
4145 (vect_analyze_stmt): Likewise.
4146 (vect_get_vector_types_for_stmt): Likewise.
4147 (tree vect_get_mask_type_for_stmt): Likewise.
4148 (vect_analyze_data_ref_dependences): Likewise.
4149 (vect_enhance_data_refs_alignment): Likewise.
4150 (vect_analyze_data_refs_alignment): Likewise.
4151 (vect_verify_datarefs_alignment): Likewise.
4152 (vect_analyze_data_ref_accesses): Likewise.
4153 (vect_prune_runtime_alias_test_list): Likewise.
4154 (vect_find_stmt_data_reference): Likewise.
4155 (vect_analyze_data_refs): Likewise.
4156 (vect_analyze_loop): Convert return type from loop_vec_info to
4158 (vect_analyze_loop_form): Likewise.
4159 (vect_analyze_slp): Convert return type from bool to opt_result.
4161 2018-10-04 David Malcolm <dmalcolm@redhat.com>
4163 * doc/invoke.texi (-fopt-info): Document new "internals"
4165 * dump-context.h (dump_context::apply_dump_filter_p): New decl.
4166 * dumpfile.c (dump_options): Update for renaming of MSG_ALL to
4168 (optinfo_verbosity_options): Add "internals".
4169 (kind_as_string): Update for renaming of MSG_ALL to MSG_ALL_KINDS.
4170 (dump_context::apply_dump_filter_p): New member function.
4171 (dump_context::dump_loc): Use apply_dump_filter_p rather than
4172 explicitly masking the dump_kind.
4173 (dump_context::begin_scope): Increment the scope depth first. Use
4174 apply_dump_filter_p rather than explicitly masking the dump_kind.
4175 (dump_context::emit_item): Use apply_dump_filter_p rather than
4176 explicitly masking the dump_kind.
4177 (dump_dec): Likewise.
4178 (dump_hex): Likewise.
4179 (dump_switch_p_1): Default to MSG_ALL_PRIORITIES.
4180 (opt_info_switch_p_1): Default to MSG_PRIORITY_USER_FACING.
4181 (opt_info_switch_p): Update handling of default
4182 MSG_OPTIMIZED_LOCATIONS to cope with default of
4183 MSG_PRIORITY_USER_FACING.
4184 (dump_basic_block): Use apply_dump_filter_p rather than explicitly
4185 masking the dump_kind.
4186 (selftest::test_capture_of_dump_calls): Update test_dump_context
4187 instances to use MSG_ALL_KINDS | MSG_PRIORITY_USER_FACING rather
4188 than MSG_ALL. Generalize scope test to be run at all four
4189 combinations of with/without MSG_PRIORITY_USER_FACING and
4190 MSG_PRIORITY_INTERNALS, adding examples of explicit priority
4191 for each of the two values.
4192 * dumpfile.h (enum dump_flag): Add comment about the MSG_* flags.
4193 Rename MSG_ALL to MSG_ALL_KINDS. Add MSG_PRIORITY_USER_FACING,
4194 MSG_PRIORITY_INTERNALS, and MSG_ALL_PRIORITIES, updating the
4195 values for TDF_COMPARE_DEBUG and TDF_ALL_VALUES.
4196 (AUTO_DUMP_SCOPE): Add a note to the comment about the interaction
4197 with MSG_PRIORITY_*.
4198 * tree-vect-loop-manip.c (vect_loop_versioning): Mark versioning
4199 dump messages as MSG_PRIORITY_USER_FACING.
4200 * tree-vectorizer.h (DUMP_VECT_SCOPE): Add a note to the comment
4201 about the interaction with MSG_PRIORITY_*.
4203 2018-10-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
4205 * varasm.c (output_constant): Add new parameter merge_strings.
4206 Make strings properly zero terminated in merge string sections.
4207 (mergeable_string_section): Don't fail if the last char is non-zero.
4208 (assemble_variable_contents): Handle merge string sections.
4209 (assemble_variable): Likewise.
4210 (assemble_constant_contents): Likewise.
4211 (output_constant_def_contents): Likewise.
4212 (output_constructor_array_range,
4213 output_constructor_regular_field): Adjust call to output_constant.
4214 (output_object_block): Adjust call to assemble_constant_contents
4215 and assemble_variable_contents.
4217 2018-10-04 Martin Liska <mliska@suse.cz>
4220 * cgraphunit.c (process_function_and_variable_attributes):
4221 Warn about a function with alias attribute and a body.
4223 2018-10-04 Martin Liska <mliska@suse.cz>
4226 * multiple_target.c (redirect_to_specific_clone): New function.
4227 (ipa_target_clone): Use it.
4228 * tree-inline.c: Fix comment.
4230 2018-10-04 David Malcolm <dmalcolm@redhat.com>
4232 * dumpfile.c (gcc::dump_manager::dump_manager): Initialize new
4234 (gcc::dump_manager::~dump_manager): Free m_optinfo_filename.
4235 (gcc::dump_manager::register_pass): New member function, adapted
4236 from loop body in gcc::pass_manager::register_pass, adding a
4237 call to update_dfi_for_opt_info.
4238 (gcc::dump_manager::opt_info_enable_passes): Store the
4239 -fopt-info options into the new fields. Move the loop
4241 (gcc::dump_manager::update_dfi_for_opt_info): ...this new member
4243 * dumpfile.h (struct opt_pass): New forward decl.
4244 (gcc::dump_manager::register_pass): New decl.
4245 (gcc::dump_manager::update_dfi_for_opt_info): New decl.
4246 (class gcc::dump_manager): Add fields "m_optgroup_flags",
4247 "m_optinfo_flags", and "m_optinfo_filename".
4248 * passes.c (gcc::pass_manager::register_pass): Move all of the
4249 dump-handling code to gcc::dump_manager::register_pass.
4251 2018-10-04 Peter Bergner <bergner@linux.ibm.com>
4253 PR rtl-optimization/87466
4254 * target.def (setjmp_preserves_nonvolatile_regs_p): New target hook.
4255 * doc/tm.texi.in (TARGET_SETJMP_PRESERVES_NONVOLATILE_REGS_P): New hook.
4256 * doc/tm.texi: Regenerate.
4257 * ira-lives.c (process_bb_node_lives): Use the new target hook.
4258 * lra-lives.c (process_bb_lives): Likewise.
4259 * config/rs6000/rs6000.c (TARGET_SETJMP_PRESERVES_NONVOLATILE_REGS_P):
4262 2018-10-04 Tamar Christina <tamar.christina@arm.com>
4264 * params.c (add_params): Fix initialization.
4266 2018-10-04 Martin Liska <mliska@suse.cz>
4268 PR gcov-profile/84107
4269 * tree-profile.c (init_ic_make_global_vars):
4270 Remove ic_void_ptr_var and ic_gcov_type_ptr_var.
4271 Come up with new ic_tuple* variables. Emit
4272 __gcov_indirect_call{,_topn} variables.
4273 (gimple_gen_ic_profiler): Access the variable
4275 (gimple_gen_ic_func_profiler): Access
4276 __gcov_indirect_call.callee field.
4277 (gimple_init_gcov_profiler): Use ptr_type_node.
4278 * value-prof.c (gimple_ic): Use ptr_type_node.
4280 2018-10-04 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
4282 PR tree-optimization/85787
4283 * ipa-pure-const.c (malloc_candidate_p_1): Move most of malloc_candidate_p
4284 into this function and add support for detecting multiple phis.
4285 (DUMP_AND_RETURN): Move from malloc_candidate_p into top-level macro.
4287 2018-10-04 Martin Liska <mliska@suse.cz>
4290 * ipa-inline.c (inline_to_all_callers_1):
4291 Call ultimate_alias_target for node being inlined.
4293 2018-10-03 Jeff Law <law@redhat.com>
4295 * gimple-ssa-sprintf.c (format_string): Do not hardcode size of
4297 * tree.c (get_typenode_from_name): Moved from fortran/trans-types.c.
4298 * tree.h (get_typenode_from_name): Prototype.
4300 2018-10-03 Uros Bizjak <ubizjak@gmail.com>
4302 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
4303 Change operand 2 predicate to nonimmediate_operand.
4304 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
4306 2018-10-03 Martin Sebor <msebor@redhat.com>
4307 Jeff Law <law@redhat.com>
4309 * gimple-ssa-sprintf.c (struct fmtresult): Add new member and
4311 (get_string_length): Detect unterminated arrays.
4312 (format_string): Same.
4313 (format_directive): Warn about unterminated arrays.
4314 (handle_gimple_call): Mark statements with no_warning as needed.
4316 2018-10-03 Jim Wilson <jimw@sifive.com>
4318 * config/riscv/riscv-c.c (riscv_cpu_cpp_builtins): For ABI_ILP32E,
4319 also define __riscv_abi_rve. Delete trailing white space.
4321 2018-10-03 Paul Koning <ni1d@arrl.net>
4323 Enable LRA register allocator for PDP11.
4324 * config/pdp11/constraints.md (Q): Use define_memory_constraint.
4327 * config/pdp11/pdp11.c (pdp11_lra_p): New function.
4328 * config/pdp11/pdp11.opt (-mlra): New option.
4329 * doc/invoke.texi (PDP-11 Options): Document -mlra.
4331 2018-10-03 Uros Bizjak <ubizjak@gmail.com>
4333 * config/i386/i386.md (*<absneg:code>extendsfdf2): Remove.
4334 (*<absneg:code>extend<mode>xf2): Ditto.
4336 2018-10-03 Aldy Hernandez <aldyh@redhat.com>
4338 PR tree-optimization/87415
4339 * tree-vrp.c (set_value_range_with_overflow): Special case one bit
4342 2018-10-02 Jeff Law <law@redhat.com>
4344 * gimple-fold.c (get_range_strlen): Only set *nonstr when
4345 an unterminated string is discovered. Bubble up range
4346 even for unterminated strings.
4347 (gimple_fold_builtin_strlen): Do not fold if get_range_strlen
4348 indicates the string was not terminated via NONSTR.
4350 2018-10-03 Aldy Hernandez <aldyh@redhat.com>
4352 * tree-vrp.c (extract_range_from_unary_expr): Special case all
4353 pointer conversions.
4354 Do not do anything special for anti-ranges.
4356 2018-10-03 Jérôme Lambourg <lambourg@adacore.com>
4358 * config/arm/vxworks.h (ARM_TARGET2_DWARF_FORMAT): Adjust to
4359 DW_EH_PE_pcrel | DW_EH_PE_indirect for RTPs.
4361 2018-10-03 Martin Liska <mliska@suse.cz>
4363 PR gcov-profile/86109
4364 * coverage.c (coverage_begin_function): Do not
4365 mark lambdas as artificial.
4366 * tree-core.h (struct GTY): Remove tm_clone_flag
4367 and introduce new lambda_function.
4368 * tree.h (DECL_LAMBDA_FUNCTION): New macro.
4370 2018-10-02 Aaron Sawdey <acsawdey@linux.ibm.com>
4373 * config/rs6000/rs6000-string.c (expand_strn_compare): Check that both
4374 P8_VECTOR and VSX are enabled.
4376 2018-10-02 Andreas Krebbel <krebbel@linux.ibm.com>
4378 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
4379 0x3907 as CPU model number.
4381 2018-10-02 Andreas Krebbel <krebbel@linux.ibm.com>
4383 * common/config/s390/s390-common.c: Rename PF_ARCH12 to PF_Z14.
4384 * config/s390/s390.h (enum processor_flags): Rename PF_ARCH12 to
4385 PF_Z14. Rename TARGET_CPU_ARCH12 to TARGET_CPU_Z14,
4386 TARGET_CPU_ARCH12_P to TARGET_CPU_Z14_P, TARGET_ARCH12 to
4387 TARGET_Z14, and TARGET_ARCH12_P to TARGET_Z14_P.
4388 * config/s390/s390.md: Likewise. Rename also the cpu attribute
4389 value from arch12 to z14.
4391 2018-10-02 Uros Bizjak <ubizjak@gmail.com>
4393 * config/i386/i386.md (fxam<mode>2_i387_with_temp): Remove.
4395 (isinf<mode>2): Ditto.
4397 2018-10-02 Uros Bizjak <ubizjak@gmail.com>
4399 * config/i386/i386.c (ix86_emit_i387_round): Extend op1 to XFmode
4400 before emitting fxam. Perform calculations in XFmode.
4402 2018-10-02 Marc Glisse <marc.glisse@inria.fr>
4404 * match.pd (((X /[ex] A) +- B) * A): New transformation.
4406 2018-10-02 Marc Glisse <marc.glisse@inria.fr>
4409 * fold-const.c (fold_plusminus_mult_expr): Handle complex and vectors.
4410 * tree.c (signed_or_unsigned_type_for): Handle complex.
4412 2018-10-02 Jeff Law <law@redhat.com>
4414 * gimple-fold.c (get_range_strlen): Remove dead code.
4416 2018-10-02 Martin Sebor <msebor@redhat.com>
4417 Jeff Law <law@redhat.com>
4419 * builtins.c (unterminated_array): Add new arguments.
4420 If argument is not terminated, bubble up size and exact
4422 (expand_builtin_strnlen): Detect, avoid expanding
4423 and diagnose unterminated arrays.
4424 (c_strlen): Fill in offset of start of unterminated strings.
4425 * builtins.h (unterminated_array): Update prototype.
4427 2018-10-02 Richard Biener <rguenther@suse.de>
4429 * config/i386/sse.md (reduc_plus_scal_v4df): Avoid the use
4430 of haddv4df, first reduce to SSE width and exploit the fact
4431 that we only need element zero with the reduction result.
4432 (reduc_plus_scal_v2df): Likewise.
4434 2018-10-02 Eric Botcazou <ebotcazou@adacore.com>
4436 * dojump.h (do_jump): Delete.
4437 (do_jump_1): Likewise.
4438 (split_comparison): Move around.
4439 * dojump.c (do_jump): Make static.
4440 (do_jump_1): Likewise.
4441 (jumpifnot): Move around.
4442 (jumpifnot_1): Likewise.
4444 (jumpif_1): Likewise.
4445 * expr.c (expand_expr_real_1): Call jumpif[not] instead of do_jump.
4447 2018-10-02 Eric Botcazou <ebotcazou@adacore.com>
4449 * reorg.c (make_return_insns): Use emit_copy_of_insn_after for the
4450 insns in the delay slot and add_insn_after for the jump insn.
4452 2018-10-02 Richard Biener <rguenther@suse.de>
4454 * tree-inline.c (expand_call_inline): Use the location of
4455 the callee declaration for the inline-entry marker.
4456 * final.c (notice_source_line): Remove special-casing of
4457 NOTE_INSN_INLINE_ENTRY.
4459 2018-10-01 Carl Love <cel@us.ibm.com>
4462 * config/rs6000/rs6000-builtin.def (__builtin_mffsl): New.
4463 (__builtin_mtfsb0): New.
4464 (__builtin_mtfsb1): New.
4465 ( __builtin_set_fpscr_rn): New.
4466 (__builtin_set_fpscr_drn): New.
4467 * config/rs6000/rs6000.c (rs6000_expand_mtfsb_builtin): Add.
4468 (rs6000_expand_set_fpscr_rn_builtin): Add.
4469 (rs6000_expand_set_fpscr_drn_builtin): Add.
4470 (rs6000_expand_builtin): Add case statement entries for
4471 RS6000_BUILTIN_MTFSB0, RS6000_BUILTIN_MTFSB1,
4472 RS6000_BUILTIN_SET_FPSCR_RN, RS6000_BUILTIN_SET_FPSCR_DRN,
4473 RS6000_BUILTIN_MFFSL.
4474 (rs6000_init_builtins): Add ftype initialization and def_builtin
4475 calls for __builtin_mffsl, __builtin_mtfsb0, __builtin_mtfsb1,
4476 __builtin_set_fpscr_rn, __builtin_set_fpscr_drn.
4477 * config/rs6000.md (rs6000_mtfsb0, rs6000_mtfsb1, rs6000_mffscrn,
4478 rs6000_mffscdrn): Add define_insn.
4479 (rs6000_set_fpscr_rn, rs6000_set_fpscr_drn): Add define_expand.
4480 * doc/extend.texi: Add documentation for the builtins.
4482 2018-10-01 Richard Biener <rguenther@suse.de>
4484 PR tree-optimization/87465
4485 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Fix typo
4486 causing branch miscounts.
4488 2018-10-01 Tamar Christina <tamar.christina@arm.com>
4490 * common/config/aarch64/aarch64-common.c (TARGET_OPTION_DEFAULT_PARAM,
4491 aarch64_option_default_param): New.
4492 (params.h): Include.
4493 (TARGET_OPTION_VALIDATE_PARAM, aarch64_option_validate_param): New.
4494 * config/aarch64/aarch64.c (aarch64_override_options_internal): Simplify
4495 stack-clash protection validation code.
4497 2018-10-01 Tamar Christina <tamar.christina@arm.com>
4499 * params.c (validate_param): New.
4500 (add_params): Use it.
4501 (set_param_value): Refactor param validation into validate_param.
4502 (diagnostic.h): Include.
4503 * diagnostic.h (diagnostic_ready_p): New.
4505 2018-10-01 Tamar Christina <tamar.christina@arm.com>
4507 * params.c (set_param_value):
4508 Add index of parameter being validated.
4509 * common/common-target.def (option_validate_param): New.
4510 * common/common-targhooks.h (default_option_validate_param): New.
4511 * common/common-targhooks.c (default_option_validate_param): New.
4512 * doc/tm.texi.in (TARGET_OPTION_VALIDATE_PARAM): New.
4513 * doc/tm.texi: Regenerate.
4515 2018-10-01 Tamar Christina <tamar.christina@arm.com>
4518 * config/aarch64/aarch64.c (aarch64_override_options_internal):
4519 Add validation for stack-clash parameters and set defaults.
4521 2018-10-01 Tamar Christina <tamar.christina@arm.com>
4524 * configure.ac: Add stack-clash-protection-guard-size.
4525 * doc/install.texi: Document it.
4526 * config.in (DEFAULT_STK_CLASH_GUARD_SIZE): New.
4527 * params.def: Update comment for guard-size.
4528 (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE,
4529 PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Update description.
4530 * configure: Regenerate.
4532 2018-10-01 Tamar Christina <tamar.christina@arm.com>
4535 * config/aarch64/aarch64.h (STACK_CLASH_MIN_BYTES_OUTGOING_ARGS,
4536 STACK_DYNAMIC_OFFSET): New.
4537 * config/aarch64/aarch64.c (aarch64_layout_frame):
4538 Update outgoing args size.
4539 (aarch64_stack_clash_protection_alloca_probe_range,
4540 TARGET_STACK_CLASH_PROTECTION_ALLOCA_PROBE_RANGE): New.
4542 2018-10-01 Tamar Christina <tamar.christina@arm.com>
4545 * explow.c (anti_adjust_stack_and_probe_stack_clash): Support custom
4547 * target.def (stack_clash_protection_alloca_probe_range): New.
4548 (stack_clash_protection_final_dynamic_probe): Remove.
4549 * targhooks.h (default_stack_clash_protection_alloca_probe_range) New.
4550 (default_stack_clash_protection_final_dynamic_probe): Remove.
4551 * targhooks.c: Likewise.
4552 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_ALLOCA_PROBE_RANGE): New.
4553 (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE): Remove.
4554 * doc/tm.texi: Regenerate.
4556 2018-10-01 Tamar Christina <tamar.christina@arm.com>
4559 * config/aarch64/aarch64-protos.h (aarch64_output_probe_sve_stack_clash): New.
4560 * config/aarch64/aarch64.c (aarch64_output_probe_sve_stack_clash,
4561 aarch64_clamp_to_uimm12_shift): New.
4562 (aarch64_allocate_and_probe_stack_space): Add SVE specific section.
4563 * config/aarch64/aarch64.md (probe_sve_stack_clash): New.
4565 2018-10-01 Tamar Christina <tamar.christina@arm.com>
4568 * config/aarch64/aarch64.c (aarch64_layout_frame): Add assert.
4570 2018-10-01 Jeff Law <law@redhat.com>
4571 Richard Sandiford <richard.sandiford@linaro.org>
4572 Tamar Christina <tamar.christina@arm.com>
4575 * config/aarch64/aarch64.md
4576 (probe_stack_range): Add k (SP) constraint.
4577 * config/aarch64/aarch64.h (STACK_CLASH_CALLER_GUARD,
4578 STACK_CLASH_MAX_UNROLL_PAGES): New.
4579 * config/aarch64/aarch64.c (aarch64_output_probe_stack_range): Emit
4580 stack probes for stack clash.
4581 (aarch64_allocate_and_probe_stack_space): New.
4582 (aarch64_expand_prologue): Use it.
4583 (aarch64_expand_epilogue): Likewise and update IP regs re-use criteria.
4584 (aarch64_sub_sp): Add emit_move_imm optional param.
4586 2018-10-01 MCC CS <deswurstes@users.noreply.github.com>
4588 PR tree-optimization/87261
4589 * match.pd: Remove trailing whitespace.
4590 Add (x & y) | ~(x | y) -> ~(x ^ y),
4591 (~x | y) ^ (x ^ y) -> x | ~y and (x ^ y) | ~(x | y) -> ~(x & y)
4593 2018-10-01 Claudiu Zissulescu <claziss@synopsys.com>
4595 * config/arc/arc.md (*add_n): Clean up pattern, update instruction
4597 (ashlsi3_insn): Update instruction constraints.
4598 (ashrsi3_insn): Likewise.
4599 (rotrsi3): Likewise.
4600 (add_shift): Likewise.
4601 * config/arc/constraints.md (Csz): New 32 bit constraint. It
4602 avoids placing in the limm field small constants which, otherwise,
4603 could end into a small instruction.
4605 2018-10-01 Claudiu Zissulescu <claziss@synopsys.com>
4607 * config/arc/arc.md (maddsidi4_split): Don't use dmac if the
4608 destination register is not odd-even.
4609 (umaddsidi4_split): Likewise.
4611 2018-10-01 Richard Biener <rguenther@suse.de>
4613 * tree-inline.c (expand_call_inline): Store origin of fn
4614 in BLOCK_ABSTRACT_ORIGIN for the inline BLOCK.
4615 * tree.c (block_ultimate_origin): Simplify and do some
4618 2018-09-30 Uros Bizjak <ubizjak@gmail.com>
4620 * config/i386/mmx.md (EMMS): New int iterator.
4621 (emms): New int attribute.
4622 (mmx_<emms>): Macroize insn from *mmx_emms and *mmx_femms using
4623 EMMS int iterator. Explicitly declare clobbers.
4624 (mmx_emms): Remove expander.
4626 * config/i386/predicates.md (emms_operation): Remove predicate.
4627 (vzeroall_pattern): New predicate.
4628 (vzeroupper_pattern): Rename from vzeroupper_operation.
4629 * config/i386/i386.c (ix86_avx_u128_mode_after): Use
4630 vzeroupper_pattern and vzeroall_pattern predicates.
4632 2018-09-30 Peter Bergner <bergner@linux.ibm.com>
4634 PR rtl-optimization/86939
4635 * ira-lives.c (make_hard_regno_born): Rename from this...
4636 (make_hard_regno_live): ... to this. Remove update to conflict
4637 information. Update function comment.
4638 (make_hard_regno_dead): Add conflict information update. Update
4640 (make_object_born): Rename from this...
4641 (make_object_live): ... to this. Remove update to conflict information.
4642 Update function comment.
4643 (make_object_dead): Add conflict information update. Update function
4645 (mark_pseudo_regno_live): Call make_object_live.
4646 (mark_pseudo_regno_subword_live): Likewise.
4647 (mark_hard_reg_dead): Update function comment.
4648 (mark_hard_reg_live): Call make_hard_regno_live.
4649 (process_bb_node_lives): Likewise.
4650 * lra-lives.c (make_hard_regno_born): Rename from this...
4651 (make_hard_regno_live): ... to this. Remove update to conflict
4652 information. Remove now uneeded check_pic_pseudo_p argument.
4653 Update function comment.
4654 (make_hard_regno_dead): Add check_pic_pseudo_p argument and add update
4655 to conflict information. Update function comment.
4656 (mark_pseudo_live): Remove update to conflict information. Update
4658 (mark_pseudo_dead): Add conflict information update.
4659 (mark_regno_live): Call make_hard_regno_live.
4660 (mark_regno_dead): Call make_hard_regno_dead with new arguement.
4661 (process_bb_lives): Call make_hard_regno_live and make_hard_regno_dead.
4663 2018-09-29 H.J. Lu <hongjiu.lu@intel.com>
4666 * config/i386/i386.c (construct_container): Use TImode for
4667 BLKmode values in 2 integer registers.
4669 2018-09-29 Jeff Law <law@redhat.com>
4671 * builtins.c (unterminated_array): Pass in c_strlen_data * to
4672 c_strlen rather than just a tree *.
4673 (c_strlen): Change NONSTR argument to a c_strlen_data pointer.
4674 Update recursive calls appropriately. If caller did not provide a
4675 suitable data pointer, create a local one. When a non-terminated
4676 string is discovered, bubble up information about the string via the
4677 c_strlen_data object.
4678 * builtins.h (c_strlen): Update prototype.
4679 (c_strlen_data): New structure.
4680 * gimple-fold.c (get_range_strlen): Update calls to c_strlen.
4681 For a type 2 call, if c_strlen indicates a non-terminated string
4682 use the length of the non-terminated string.
4683 (gimple_fold_builtin_stpcpy): Update calls to c_strlen.
4685 2018-09-29 Jakub Jelinek <jakub@redhat.com>
4688 * config/i386/avx512fintrin.h (_mm512_abs_pd, _mm512_mask_abs_pd): Use
4689 __m512d type for __A argument rather than __m512.
4691 2018-09-28 John David Anglin <danglin@gcc.gnu.org>
4693 * match.pd (simple_comparison): Don't optimize if either operand is
4694 a function pointer when target needs function pointer canonicalization.
4696 2018-09-28 Segher Boessenkool <segher@kernel.crashing.org>
4698 * config/rs6000/driver-rs6000.c (asm_names): Adjust the entries for
4699 power5 .. power9 to remove indirection.
4700 * config/rs6000/rs6000.h (ASM_CPU_POWER5_SPEC, ASM_CPU_POWER6_SPEC,
4701 ASM_CPU_POWER7_SPEC, ASM_CPU_POWER8_SPEC, ASM_CPU_POWER9_SPEC,
4702 ASM_CPU_476_SPEC): Delete.
4703 (ASM_CPU_SPEC): Adjust.
4704 (EXTRA_SPECS): Delete asm_cpu_power5, asm_cpu_power6, asm_cpu_power7,
4705 asm_cpu_power8, asm_cpu_power9, asm_cpu_476.
4707 2018-09-28 Segher Boessenkool <segher@kernel.crashing.org>
4709 * config.in: Delete HAVE_AS_DCI.
4710 * config/powerpcspe/powerpcspe.h: Treat HAVE_AS_DCI as always true.
4711 * config/rs6000/rs6000.h: Ditto.
4712 * configure.ac: Delete HAVE_AS_DCI.
4713 * configure: Regenerate.
4715 2018-09-28 Segher Boessenkool <segher@kernel.crashing.org>
4717 * config.in (HAVE_AS_LWSYNC): Delete.
4718 * config/powerpcspe/powerpcspe.h (TARGET_LWSYNC_INSTRUCTION): Delete.
4719 * config/powerpcspe/sync.md (*lwsync): Always generate lwsync, never
4721 * config/rs6000/rs6000.h (TARGET_LWSYNC_INSTRUCTION): Delete.
4722 * config/rs6000/sync.md (*lwsync): Always generate lwsync, never do it
4724 * configure.ac: Delete HAVE_AS_LWSYNC.
4725 * configure: Regenerate.
4727 2018-09-28 Eric Botcazou <ebotcazou@adacore.com>
4728 Pierre-Marie de Rodat <derodat@adacore.com>
4730 * calls.c (expand_call): Try to do a tail call for thunks at -O0 too.
4731 * cgraph.h (struct cgraph_thunk_info): Add indirect_offset.
4732 (cgraph_node::create_thunk): Add indirect_offset parameter.
4733 (thunk_adjust): Likewise.
4734 * cgraph.c (cgraph_node::create_thunk): Add indirect_offset parameter
4735 and initialize the corresponding field with it.
4736 (cgraph_node::dump): Dump indirect_offset field.
4737 * cgraphclones.c (duplicate_thunk_for_node): Deal with indirect_offset.
4738 * cgraphunit.c (cgraph_node::analyze): Be prepared for external thunks.
4739 (thunk_adjust): Add indirect_offset parameter and deal with it.
4740 (cgraph_node::expand_thunk): Deal with the indirect_offset field and
4741 pass it to thunk_adjust. Do not call the target hook if it's non-zero
4742 or if the thunk is external or local. Fix formatting. Do not chain
4743 the RESULT_DECL to BLOCK_VARS. Pass the static chain to the target,
4744 if any, in the GIMPLE representation.
4745 * ipa-icf.c (sem_function::equals_wpa): Deal with indirect_offset.
4746 * lto-cgraph.c (lto_output_node): Write indirect_offset field.
4747 (input_node): Read indirect_offset field.
4748 * tree-inline.c (expand_call_inline): Pass indirect_offset field in the
4749 call to thunk_adjust.
4750 * tree-nested.c (struct nesting_info): Add thunk_p field.
4751 (create_nesting_tree): Set it.
4752 (convert_all_function_calls): Copy static chain from targets to thunks.
4753 (finalize_nesting_tree_1): Return early for thunks.
4754 (unnest_nesting_tree_1): Do not finalize thunks.
4755 (gimplify_all_functions): Do not gimplify thunks.
4757 2018-09-28 David Malcolm <dmalcolm@redhat.com>
4759 * opt-suggestions.c (option_proposer::build_option_suggestions):
4760 Release "option_values".
4762 2018-09-28 David Malcolm <dmalcolm@redhat.com>
4764 * coverage.c (get_coverage_counts): Convert problem-reporting dump
4765 messages from MSG_OPTIMIZED_LOCATIONS to MSG_MISSED_OPTIMIZATION.
4766 * dumpfile.c (kind_as_string): New function.
4767 (dump_loc): Rather than a hardcoded prefix of "note: ", use
4768 kind_as_string to vary the prefix based on dump_kind.
4769 (selftest::test_capture_of_dump_calls): Update for above.
4771 2018-09-28 Uros Bizjak <ubizjak@gmail.com>
4773 * config/i386/i386.h (SSE_REGNO): Fix check for FIRST_REX_SSE_REG.
4774 (GET_SSE_REGNO): Rename from SSE_REGNO. Update all uses for rename.
4776 2018-09-28 Uros Bizjak <ubizjak@gmail.com>
4778 * config/i386/i386.h (CC_REGNO): Remove FPSR_REG.
4779 * config/i386/i386.c (ix86_fixed_condition_code_regs): Use
4780 INVALID_REGNUM instead of FPSR_REG.
4781 (ix86_md_asm_adjust): Do not clobber FPSR_REG.
4782 * config/i386/i386.md: Update comment of FP compares.
4783 (fldenv): Do not clobber FPSR_REG.
4785 2018-09-28 Richard Biener <rguenther@suse.de>
4787 * tree.h (BLOCK_ORIGIN): New.
4788 * omp-expand.c (grid_expand_target_grid_body): Assign
4789 BLOCK_ORIGIN to BLOCK_ABSTRACT_ORIGIN.
4790 * tree-inline.c (remap_block): Likewise.
4791 * auto-profile.c (get_function_decl_from_block): Simplify
4792 by eliding the BLOCK_ABSTRACT_ORIGIN chasing.
4793 * langhooks.c (lhd_print_error_function): Likewise.
4794 * optinfo-emit-json.cc (optrecord_json_writer::inlining_chain_to):
4796 * tree-ssa-live.c (remove_unused_scope_block_p): Likewise.
4797 * tree.c (block_nonartificial_location): Likewise.
4798 (block_ultimate_origin): Likewise.
4799 * tree-pretty-print.c (percent_K_format): Likewise. Remove
4800 no longer needed LTO case.
4802 2018-09-28 Andrew Stubbs <ams@codesourcery.com>
4803 Jan Hubicka <jh@suse.cz>
4804 Martin Jambor <mjambor@suse.cz>
4806 * simplify-rtx.c (simplify_merge_mask): New function.
4807 (simplify_ternary_operation): Use it, also see if VEC_MERGEs with the
4808 same masks are used in op1 or op2.
4809 (test_vec_merge): New function.
4810 (test_vector_ops): Call test_vec_merge.
4812 2018-09-28 Eric Botcazou <ebotcazou@adacore.com>
4814 * config/sparc/sparc-protos.h (sparc_branch_cost): Declare.
4815 * config/sparc/sparc.h (BRANCH_COST): Call sparc_branch_cost.
4816 * config/sparc/sparc.c (struct processor_costs): Add branch_cost field.
4817 (cypress_costs): Set it.
4818 (supersparc_costs): Likewise.
4819 (hypersparc_costs): Likewise.
4820 (leon_cost): Likewise.
4821 (leon3_costs): Likewise.
4822 (sparclet_costs): Likewise.
4823 (ultrasparc_costs): Likewise.
4824 (ultrasparc_costs): Likewise.
4825 (niagara_costs): Likewise.
4826 (niagara2_costs): Likewise.
4827 (niagara3_costs): Likewise.
4828 (niagara4_costs): Likewise.
4829 (niagara7_costs): Likewise.
4830 (m8_costs): Likewise.
4831 (TARGET_CAN_FOLLOW_JUMP): Define.
4832 (pass_work_around_errata::gate): Minor tweak.
4833 (sparc_option_override): Remove MASK_FSMULD mask for V7 processors.
4834 Do not set both MASK_VIS4 and MASK_VIS4B for M8 processor.
4835 Automaitcally clear MASK_FSMULD mask for V7 processors.
4836 (sparc_can_follow_jump): New static function.
4837 (output_ubranch): Deal with CROSSING_JUMP_P.
4838 (sparc_use_sched_lookahead): Rewrite using switch statement.
4839 (sparc_issue_rate): Reorder.
4840 (sparc_branch_cost): New function.
4842 2018-09-27 Martin Sebor <msebor@redhat.com>
4844 * tree.h (tree_to_shwi): Add attribute nonnull and pure.
4845 (tree_to_poly_int64, tree_to_uhwi, tree_to_poly_uint64): Same.
4846 (int_fits_type_p): Same.
4848 2018-09-27 Uros Bizjak <ubizjak@gmail.com>
4850 * config/i386/i386.md (FPCR_REG): Remove.
4851 (UNSPEC_FLDCW): Remove.
4852 (x86_fnstcw_1): Use (const_int 0) instead of FPCR_REG.
4853 (x86_fldcw_1): Remove insn pattern.
4854 (fnstenv): Do not clobber FPCR_REG.
4856 * config/i386/i386.h (FIXED_REGISTERS) Remove fpsr register.
4857 (CALL_USED_REGISTERS): Ditto.
4858 (REG_ALLOC_ORDER): Ditto.
4859 (REG_CLASS_CONTENTS): Ditto.
4860 (HI_REGISTER_NAMES): Ditto.
4861 (ADDITIONAL_REGISTER_NAMES): Use defines instead
4862 of numerical constants.
4863 * config/i386/i386.c (regclass_map): Remove fpsr register.
4864 (dbx_register_map): Ditto.
4865 (dbx64_register_map): Ditto.
4866 (svr4_dbx_register_map): Ditto.
4867 (print_reg): Do not handle FPCR_REG.
4869 2018-09-27 Segher Boessenkool <segher@kernel.crashing.org>
4872 * config.in (HAVE_AS_CMPB, HAVE_AS_DFP, HAVE_AS_FPRND, HAVE_AS_MFPGPR,
4873 HAVE_AS_POPCNTB, HAVE_AS_POPCNTD, HAVE_AS_POWER8, HAVE_AS_POWER9):
4874 Delete, always treat as true.
4875 * config/powerpcspe/powerpcspe.c (rs6000_option_override_internal):
4876 Ditto. Simplify remaining code.
4877 * config/powerpcspe/powerpcspe.h: Ditto.
4878 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ditto.
4879 Simplify remaining code.
4880 (rs6000_expand_builtin): Ditto.
4881 * config/rs6000/rs6000.h: Ditto.
4882 * configure.ac: Ditto.
4883 * configure: Regenerate.
4885 2018-09-27 Martin Liska <mliska@suse.cz>
4887 * coverage.c (get_coverage_counts): Revert the formatting
4888 of missing profile opt info.
4890 2018-09-27 Richard Biener <rguenther@suse.de>
4894 * dwarf2out.c (set_block_origin_self): Do not mark outermost
4895 block as we do not output that.
4896 (gen_inlined_subroutine_die): Elide the originally outermost
4897 block, matching what we do for concrete instances.
4898 (decls_for_scope): Add parameter specifying whether to recurse
4901 2018-09-27 Andrew Stubbs <ams@codesourcery.com>
4902 Tom de Vries <tom@codesourcery.com>
4906 * expmed.c (emit_cstore): Fix handling of result_mode == BImode and
4907 STORE_FLAG_VALUE == 1.
4909 2018-09-27 Andreas Krebbel <krebbel@linux.ibm.com>
4911 * config/s390/s390.md (PPA_TX_ABORT, PPA_OOO_BARRIER): New
4912 constant definitions.
4913 ("tx_assist"): Replace magic number with PPA_TX_ABORT.
4914 ("*ppa"): Enable pattern also for -march=zEC12 -mno-htm.
4915 ("speculation_barrier"): New expander definition.
4917 2018-09-26 Indu Bhagat <indu.bhagat@oracle.com>
4919 PR gcov-profile/86957
4920 * common.opt: New warning option -Wmissing-profile.
4921 * coverage.c (get_coverage_counts): Add warning for missing .gcda file.
4922 * doc/invoke.texi: Document -Wmissing-profile.
4924 2018-09-26 Jim Wilson <jimw@sifive.com>
4926 * config/riscv/riscv.md (subsi3_extended2): Add J constraint.
4927 (negdi2, negsi2, negsi2_extended, negsi2_extended2): New.
4929 2018-09-26 Martin Sebor <msebor@redhat.com>
4931 * tree.c (zerop): Change return type to bool.
4932 (integer_zerop, integer_onep, integer_each_onep): Same.
4933 (integer_all_onesp, integer_minus_onep, integer_pow2p): Same.
4934 (integer_nonzerop, integer_truep, tree_ctz, real_zerop): Same.
4935 (real_onep, real_minus_onep, chain_index): Same.
4936 (print_type_hash_statistics, type_list_equal): Same.
4937 * tree.h (zerop): Same.
4938 (zerop, integer_zerop, integer_onep, integer_each_onep): Same.
4939 (integer_all_onesp, integer_minus_onep, integer_pow2p): Same.
4940 (integer_nonzerop, integer_truep, tree_ctz, real_zerop): Same.
4941 (real_onep, real_minus_onep, chain_index): Same.
4942 (print_type_hash_statistics, type_list_equal): Same.
4944 2018-09-26 Jim Wilson <jimw@sifive.com>
4946 * config/riscv/riscv.h (FUNCTION_ARG_REGNO_P): Fix comment.
4948 2018-09-26 Jakub Jelinek <jakub@redhat.com>
4951 * config/i386/i386.c: Include debug.h and dwarf2out.h.
4952 (output_indirect_thunk): Emit DW_CFA_def_cfa_offset after the
4955 2018-09-25 Andrew Stubbs <ams@codesourcery.com>
4957 * builtins.c (get_builtin_sync_mem): Force address mode conversion.
4959 2018-09-26 Uros Bizjak <ubizjak@gmail.com>
4961 * config/i386/i386.h (enum reg_class): Remove FP_TOP_SSE_REGS
4962 and FP_SECOND_SSE_REGS.
4963 (REG_CLASS_NAMES): Ditto.
4964 (REG_CLASS_CONTENTS): Ditto.
4965 * config/i386/i386.c (ix86_preferred_reload_class) Do not handle
4966 FP_TOP_SSE_REGS and FP_SECOND_SSE_REGS classes.
4967 (ix86_preferred_output_reload_class): Ditto.
4968 * config/i386/i386.md (fix_trunc<mode>_i387_fisttp): Change "=&1f"
4969 clobber constraint to "=&f".
4970 (fix_truncdi_i387): Ditto.
4971 (lrintxfdi2): Ditto.
4972 (fistdi2_<rounding>): Ditto.
4973 (fpremxf4_i387): Change "=u" constraint to "=f".
4974 (fprem1xf4_i387): Ditto.
4976 (fptanxf4_i387): Ditto.
4977 (fxtractxf3_i387): Ditto.
4978 (fscalexf4_i387): Ditto.
4979 (atan2xf3): Change "u" constraint to "f".
4980 (fyl2xxf3_i387): Ditto.
4981 (fyl2xp1xf3_i387): Ditto.
4983 2018-09-26 Uros Bizjak <ubizjak@gmail.com>
4986 * config/i386/i386.h (NUM_MODES_FOR_MODE_SWITCHING): Update
4987 for removed I387_MASK_PM entity.
4990 2018-09-26 Jeff Law <law@redhat.com>
4992 2018-09-26 Alexey Neyman <stilor@att.net>
4994 * graphite.h: Include <isl/id.h> and <isl/space.h>; these
4995 headers are no longer pulled in by <isl/val.h>.
4997 2018-09-26 Richard Biener <rguenther@suse.de>
5000 * dwarf2out.c (gen_lexical_block_die): Do not equate inline
5001 or concrete instance DIE to the tree. Create abstract origin
5002 attributes also for concrete instances.
5004 2018-09-26 Alexey Neyman <stilor@att.net>
5006 * graphite.h: Include <isl/id.h> and <isl/space.h>; these
5007 headers are no longer pulled in by <isl/val.h>.
5009 2018-09-26 Matthew Malcomson <matthew.malcomson@arm.com>
5011 * config/arm/arm.c (arm_split_compare_and_swap, arm_split_atomic_op):
5012 Use new helper functions.
5013 * config/arm/sync.md (atomic_load<mode>, atomic_store<mode>):
5014 Use new helper functions.
5015 * config/arm/aarch-common-protos.h (aarch_mm_needs_acquire,
5016 aarch_mm_needs_release): New declarations.
5017 * config/arm/aarch-common.c (aarch_mm_needs_acquire,
5018 aarch_mm_needs_release): New.
5020 2018-09-26 Eric Botcazou <ebotcazou@adacore.com>
5022 * config/arm/arm.c (arm_reorg): Skip Thumb reorg pass for thunks.
5023 (arm32_output_mi_thunk): Deal with long calls.
5025 2018-09-26 Richard Biener <rguenther@suse.de>
5029 * tree-inline.c (expand_call_inline): When the location
5030 of the call is UNKNOWN_LOCATION use DECL_SOURCE_LOCATION
5031 or BUILTINS_LOCATION for the BLOCK_SOURCE_LOCATION of
5032 the inserted BLOCK to make inlined_function_outer_scope_p
5034 * dwarf2out.c (add_call_src_coords_attributes): Do not add
5035 coords for reserved locations.
5037 2018-09-25 Segher Boessenkool <segher@kernel.crashing.org>
5039 * config/rs6000/rs6000.md (*movcc_internal1): Use set_attr_alternative.
5040 (*call_indirect_nonlocal_sysv<mode>): Ditto.
5041 (*call_value_indirect_nonlocal_sysv<mode>): Ditto.
5042 (*sibcall_nonlocal_sysv<mode>): Ditto.
5043 (*sibcall_value_nonlocal_sysv<mode>): Ditto.
5044 (<bd>_<mode>): Ditto.
5045 (<bd>tf_<mode>): Ditto.
5047 2018-09-25 Segher Boessenkool <segher@kernel.crashing.org>
5049 * config/rs6000/altivec.md (*altivec_mov<mode>): Write the output
5050 control string as a list of templates instead of as C code.
5051 (*altivec_movti): Ditto.
5052 * config/rs6000/darwin.md (movdf_low_di): Ditto.
5054 2018-09-25 Jim Wilson <jimw@sifive.com>
5056 * config/riscv/riscv.c (riscv_split_symbol): Mark auipc label as weak
5057 when target symbol is weak.
5059 2018-09-25 Bernd Edlinger <bernd.edlinger@hotmail.de>
5062 * builtins.c (unterminated_array): Simplify.
5063 * expr.c (string_constant): Handle SSA_NAME. Add more exceptions
5064 where pointer arithmetic is safe.
5066 2018-09-25 Segher Boessenkool <segher@kernel.crashing.org>
5069 * config/rs6000/altivec.md (altivec_vspltb): Use
5070 const_0_to_15_operand instead of u5bit_cint_operand.
5071 (*altivec_vspltb_internal): Ditto.
5072 (altivec_vspltb_direct): Ditto.
5073 (altivec_vsplth): Use const_0_to_7_operand instead of
5075 (*altivec_vsplth_internal): Ditto.
5076 (altivec_vsplth_direct): Ditto.
5077 (altivec_vspltw): Use const_0_to_3_operand instead of
5079 (*altivec_vspltw_internal): Ditto.
5080 (altivec_vspltw_direct): Ditto.
5081 (altivec_vspltsf): Ditto.
5082 (*altivec_vspltsf_internal): Ditto.
5083 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Handle the
5084 various splats with the proper size immediate. Reorder the various
5085 cases by ascending size of immediate, and put all such together.
5087 2018-09-25 Richard Biener <rguenther@suse.de>
5090 * dwarf2out.c (add_AT_external_die_ref): Remove now redundant
5092 (maybe_create_die_with_external_ref): Do not create
5093 DW_TAG_imported_unit here.
5094 (add_abstract_origin_attribute): Handle external BLOCK refs.
5095 (dwarf2out_abstract_function): Simplify LTO case.
5096 (dwarf2out_early_finish): Create DW_TAG_imported_unit explicitely
5097 rather than using maybe_create_die_with_external_ref.
5099 2018-09-25 Uros Bizjak <ubizjak@gmail.com>
5102 * config/i386/i386.md (frndintxf2_mask_pm): Remove.
5103 (frndintxf2_mask_pm_i387): Ditto.
5104 (nearbyintxf2): Rewrite expander pattern to match rintxf2.
5105 Enable for !flag_trapping_math.
5106 (nearbyint<mode>2): Enable x87 modes for !flag_trapping_math.
5107 Enable SSE modes for TARGET_SSE4_1 and expand them with round insn.
5108 Change operand 1 predicate to nonimmediate_operand.
5109 (attr "i387_cw"): Remove mask_pm.
5110 * config/i386/i386.h (enum ix86_stack_slot): Remove SLOT_CW_MASK_PM.
5111 (enum ix86_entity): Remove I387_MASK_PM.
5112 * config/i386/i386.c (ix86_i387_mode_needed): Do not
5113 handle I387_MASK_PM.
5114 (ix86_mode_needed): Ditto.
5115 (ix86_mode_after): Ditto.
5116 (ix86_mode_entry): Ditto.
5117 (ix86_mode_exit): Ditto.
5118 (emit_i387_cw_initialization): Do not handle I387_CW_MASK_PM.
5120 2018-09-25 Jakub Jelinek <jakub@redhat.com>
5122 * vr-values.c (vr_values::vr_values): Initialize to_remove_edges and
5123 to_update_switch_stmts to vNULL instead of calling create on them
5126 2018-09-25 Richard Biener <rguenther@suse.de>
5128 PR tree-optimization/87402
5129 * tree-ssa-sccvn.c (SSA_VISITED): Remove unused function.
5130 (visit_phi): Re-instantiate handling of supposed to be VARYING
5131 but non-VARYING backedge value.
5133 2018-09-25 Richard Biener <rguenther@suse.de>
5136 * dwarf2out.c (struct sym_off_pair): New.
5137 (external_die_map): New global.
5138 (lookup_decl_die): When in LTO create DIEs lazily from the
5140 (lookup_block_die): New function, create DIEs lazily in LTO.
5141 (equate_block_to_die): New function.
5142 (dwarf2out_die_ref_for_decl): During WPA get the association
5143 from the external DIE map.
5144 (dwarf2out_register_external_die): Record mapping into the
5146 (maybe_create_die_with_external_ref): New function split out from
5147 DIE generation part of old dwarf2out_register_external_die.
5148 (add_abstract_origin_attribute): Do not return the DIE. When
5149 in LTO reference externals directly.
5150 (dwarf2out_abstract_function): When in LTO ignore calls for
5151 decls with external DIEs (already present abstract instances).
5152 (gen_call_site_die): Adjust.
5153 (add_high_low_attributes): Likewise.
5154 (gen_lexical_block_die): Likewise.
5155 (gen_inlined_subroutine_die): Likewie.
5156 (gen_block_die): Likewise.
5157 (dwarf2out_inline_entry): Likewise.
5158 (dwarf2out_early_finish): In LTRANS phase create DW_TAG_imported_unit
5161 2018-09-25 Martin Liska <mliska@suse.cz>
5163 * ipa-fnsummary.c (estimate_node_size_and_time): Scale by two
5164 integers and not by a float value.
5166 2018-09-25 Martin Liska <mliska@suse.cz>
5169 * dbgcnt.c (dbg_cnt_process_single_pair): Return false
5171 * dumpfile.c (dump_enable_all): Remove extra parenthesis.
5172 * gcov-tool.c: Declare the function with ATTRIBUTE_NORETURN.
5173 * godump.c (go_format_type): Remove extra parenthesis.
5175 2018-09-25 Martin Liska <mliska@suse.cz>
5177 * alias.c (set_dest_equal_p): Remove unused function.
5178 * config/i386/i386.c (def_builtin_pure2): Likewise.
5179 * diagnostic-show-locus.c (class layout): Remove
5181 (layout::layout): Likewise here.
5182 * dump-context.h (class temp_dump_context): Likewise.
5183 * dwarf2out.c (add_AT_fde_ref): Remove unused function.
5184 (add_AT_loclistsptr): Likewise.
5185 (add_AT_offset): Likewise.
5186 (get_AT_hi_pc): Likewise.
5187 (is_comdat_die): Likewise.
5188 (type_is_enum): Likewise.
5189 (ceiling): Likewise.
5190 (add_AT_vms_delta): Likewise.
5191 (is_class_die): Likewise.
5192 * edit-context.c (class line_event): Remove unused field.
5193 * graphite-sese-to-poly.c (tree_int_to_gmp): Remove
5195 * ipa-cp.c (ipa_get_vr_lat): Likewise.
5196 * lra-constraints.c (ok_for_index_p_nonstrict): Likewise.
5197 (ok_for_base_p_nonstrict): Likewise.
5198 * tree-chrec.c (is_not_constant_evolution): Likewise.
5199 (chrec_fold_poly_cst): Likewise.
5200 * tree-if-conv.c (has_pred_critical_p): Likewise.
5201 * tree-ssa-coalesce.c (print_exprs): Likewise.
5202 * tree-ssa-pre.c (bitmap_set_contains_expr): Likewise.
5203 * tree-ssa-uninit.c (is_and_or_or_p): Likewise.
5204 * tree-vrp.c (value_ranges_intersect_p): Likewise.
5205 (value_range_nonnegative_p): Likewise.
5207 2018-09-25 Martin Liska <mliska@suse.cz>
5209 * config/powerpcspe/powerpcspe.c (rs6000_output_function_epilogue):
5210 Do not handle "GNU Pascal".
5211 * config/rs6000/rs6000.c (rs6000_output_function_epilogue):
5213 * config/sparc/sparc.c (sparc_pass_by_reference): Remove Pascal
5214 from documentation. Likewise.
5215 * dbxout.c (dbxout_range_type): Likewise.
5216 * doc/cpp.texi: Likewise.
5217 * doc/extend.texi: Likewise.
5218 * doc/frontends.texi: Likewise.
5219 * doc/invoke.texi: Remove Pascal entry.
5220 * tree.def (CLEANUP_POINT_EXPR): Likewise.
5221 * doc/rtl.texi (MODE_FUNCTION): Remove not used entry.
5223 2018-09-25 Martin Liska <mliska@suse.cz>
5226 * doc/invoke.texi: Document all parameters and remove default
5229 2018-09-25 Ilya Leoshkevich <iii@linux.ibm.com>
5232 * rtl.c (rtx_code_size): Take into account that EQ_ATTR_ALT
5233 contains HOST_WIDE_INTs when computing its size.
5235 2018-09-24 Jim Wilson <jimw@sifive.com>
5238 * config/riscv/riscv.h (STACK_BOUNDARY): Test riscv_abi == ABI_ILP32E
5240 (ABI_STACK_BOUNDARY, MAX_ARGS_IN_REGISTERS): Likewise.
5242 2018-09-24 Andrew Pinski <apinski@marvell.com>
5244 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Don't
5245 access prev before checking it for NULLness in the
5246 AARCH64_FUSE_CMP_BRANCH case.
5248 2018-09-24 H.J. Lu <hongjiu.lu@intel.com>
5251 * config/i386/i386.c (rest_of_insert_endbranch): Set
5252 endbr_queued_at_entrance to true and don't insert ENDBR if
5253 x86_function_profiler will be called.
5254 (x86_function_profiler): Insert ENDBR if endbr_queued_at_entrance
5256 * config/i386/i386.h (machine_function): Add
5257 endbr_queued_at_entrance.
5259 2018-09-24 Ilya Leoshkevich <iii@linux.ibm.com>
5261 * genattrtab.c (mk_attr_alt): Use alternative_mask.
5262 (attr_rtx_1): Adjust caching to match the new EQ_ATTR_ALT field
5264 (check_attr_test): Use alternative_mask.
5265 (get_attr_value): Likewise.
5266 (compute_alternative_mask): Use alternative_mask and XWINT.
5267 (make_alternative_compare): Use alternative_mask.
5268 (attr_alt_subset_p): Use XWINT.
5269 (attr_alt_subset_of_compl_p): Likewise.
5270 (attr_alt_intersection): Use alternative_mask and XWINT.
5271 (attr_alt_union): Likewise.
5272 (attr_alt_complement): Use HOST_WIDE_INT and XWINT.
5273 (mk_attr_alt): Use alternative_mask and HOST_WIDE_INT.
5274 (simplify_test_exp): Use alternative_mask and XWINT.
5275 (write_test_expr): Use alternative_mask and XWINT, adjust bit
5276 number calculation to support 64 bits. Generate code that
5277 checks 64-bit masks.
5278 (main): Use alternative_mask.
5279 * rtl.def (EQ_ATTR_ALT): Change field types from ii to ww.
5281 2018-09-24 Ilya Leoshkevich <iii@linux.ibm.com>
5284 * config/s390/s390.c (s390_emit_epilogue): Do not use PARALLEL
5285 RETURN+USE when returning via %r14.
5287 2018-09-24 Martin Liska <mliska@suse.cz>
5289 * gcov.c (output_lines): Print colorization legend
5290 for both flag_use_colors and flag_use_hotness_colors.
5293 2018-09-24 Martin Liska <mliska@suse.cz>
5295 * coverage.c (get_coverage_counts): Use warning_at
5296 with current_function_decl location. Use %qD in warning
5299 2018-09-24 Martin Liska <mliska@suse.cz>
5301 * memory-block.h (memory_block_pool::release): Annotate with
5302 valgrind that the memory is not accessible.
5304 2018-09-24 Martin Liska <mliska@suse.cz>
5307 * asan.c: Make asan_handled_variables extern.
5309 * cfgexpand.c (expand_stack_vars): Make sure
5310 a representative is unpoison if another
5311 variable in the partition is handled by
5312 use-after-scope sanitization.
5314 2018-09-24 Richard Biener <rguenther@suse.de>
5316 PR tree-optimization/63155
5317 * tree-ssa-propagate.c (add_ssa_edge): Avoid adding PHIs to
5318 the worklist when the edge of the respective argument isn't
5321 2018-09-23 Uros Bizjak <ubizjak@gmail.com>
5323 * config/i386/i386.h (enum reg_class): Rename MASK_REGS to
5324 ALL_MASK_REGS and MASK_EVEX_REGS to MASK_REGS.
5325 (MASK_CLASS_P): Update for rename.
5326 (MAYBE_MASK_CLASS_P): Ditto.
5327 (REG_CLASS_NAMES): Update.
5328 (REG_CLASS_CONTENT): Update.
5329 * config/i386/i386.c (regclass_map): Update for MASK_REG
5330 and ALL_MASK_REGS rename.
5331 * config/i386/constraints.md (Yk): Update for rename.
5334 2018-09-23 Uros Bizjak <ubizjak@gmail.com>
5336 * config/i386/i386.h (enum reg_class): Remove
5337 EVEX_SSE_REGS and MOD4_SSE_REGS.
5338 (REG_CLASS_NAMES): Update.
5339 (REG_CLASS_CONTENT): Update.
5340 * config/i386/i386.c (regclass_map): Declare AVX-512 SSE
5341 registers as ALL_SSE_REGS.
5342 (ix86_additional_allocno_class_p): Remove.
5343 (TARGET_ADDITIONAL_ALLOCNO_CLASS_P): Remove.
5344 (ix86_register_priority): Lower priority of EVEX SSE registers.
5345 Use IN_RANGE macro where appropriate.
5346 (ix86_hard_regno_mode_ok): Merge AVX-5124FMAPS and
5347 AVX-5124VNNIW checks.
5348 (ix86_modes_tieable_p): Tie 512-bit SSE modes.
5349 * config/i386/sse.md (avx5124fmaddps_4fmaddps)
5350 (avx5124fmaddps_4fmaddps_mask, avx5124fmaddps_4fmaddps_maskz)
5351 (avx5124fmaddps_4fmaddss, avx5124fmaddps_4fmaddss_mask)
5352 (avx5124fmaddps_4fmaddss_maskz, avx5124fmaddps_4fnmaddps)
5353 (avx5124fmaddps_4fnmaddps_mask, avx5124fmaddps_4fnmaddps_maskz)
5354 (avx5124fmaddps_4fnmaddss, avx5124fmaddps_4fnmaddss_mask)
5355 (avx5124fmaddps_4fnmaddss_maskz, avx5124vnniw_vp4dpwssd)
5356 (avx5124vnniw_vp4dpwssd_mask, avx5124vnniw_vp4dpwssd_maskz)
5357 (avx5124vnniw_vp4dpwssds, avx5124vnniw_vp4dpwssds_mask)
5358 (avx5124vnniw_vp4dpwssds_maskz): Use "v" instead of "Yh" constraint.
5359 * config/i386/constraints.md (Yh): Remove.
5361 2018-09-23 Uros Bizjak <ubizjak@gmail.com>
5363 * config/i386/i386.c (regclass_map): Declare integer REX registers
5366 2018-09-23 Gerald Pfeifer <gerald@pfeifer.com>
5368 * doc/service.texi (Service): Switch the fsf.org link to https.
5370 2018-09-22 Chung-Ju Wu <jasonwucj@gmail.com>
5373 * config/nds32/nds32.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
5374 Define to speculation_safe_value_not_needed.
5376 2018-09-21 Florian Weimer <fweimer@redhat.com>
5379 * doc/extend.texi (Common Function Attributes): Mention that
5380 noreturn suppresses tail call optimization.
5382 2018-09-21 Jeff Law <law@redhat.com>
5384 * gimple-ssa-evrp.c (evrp_dom_walker::cleanup): Call
5385 vr_values::cleanup_edges_and_switches.
5386 * tree-vrp.c (to_remove_edges, to_update_switch_stmts): Moved into
5388 (identify_jump_threads): Remove EDGE_IGNORE handling.
5389 (execute_vrp): Move handling of to_remove_edges and
5390 to_update_switch_stmts into vr_values class member functions.
5391 * tree-vrp.h (switch_update, to_remove_edges): Remove declarations.
5392 (to_update_switch_stmts): Likewise.
5393 * vr-values.c: Include cfghooks.h.
5394 (vr_values::vr_values): Initialize to_remove_edges and
5395 to_update_switch_stmts.
5396 (vr_values::~vr_values): Verify to_remove_edges and
5397 to_update_switch_stmts are empty.
5398 (vr_values::simplify_switch_using_ranges): Set EDGE_IGNORE as needed.
5399 (vr_values::cleanup_edges_and_switches): New member function.
5400 * vr-values.h (vr_values): Add cleanup_edges_and_switches member
5401 function. Add new data members.
5403 2018-09-21 David Malcolm <dmalcolm@redhat.com>
5405 PR tree-optimization/87309
5406 * dumpfile.c (dump_context::begin_scope): Filter the dump_loc
5407 calls with pflags and alt_flags.
5408 (selftest::test_capture_of_dump_calls): Add test of interaction of
5409 MSG_OPTIMIZED_LOCATIONS with AUTO_DUMP_SCOPE.
5411 2018-09-21 Olivier Hainque <hainque@adacore.com>
5413 * config.gcc: Factorize and comment inclusion of vxworks-dummy.h.
5415 2018-09-21 Olivier Hainque <hainque@adacore.com>
5417 * config/vxworks.h (CLEAR_INSN_CACHE): #define to 1.
5419 2018-09-21 Olivier Hainque <hainque@adacore.com>
5421 * config/vxworks.h (VXWORKS_LIBGCC_SPEC): Remove -lc_internal.
5422 Merge block comment with the one ahead of VXWORKS_LIBS_RTP. Then:
5423 (VXWORKS_LIBS_RTP): Minor reordering.
5425 2018-09-21 Olivier Hainque <hainque@adacore.com>
5427 * config/vxworks.h (STARTFILE_PREFIX_SPEC): Define.
5428 (VXWORKS_LIBS_DIR_RTP): Remove definition and use.
5430 2018-09-21 Olivier Hainque <hainque@adacore.com>
5432 * config/vxworks.h (SIZE_TYPE): Account for TARGET_VXWORKS64.
5433 (PTRDIFF_TYPE): Likewise.
5435 2018-09-21 Olivier Hainque <hainque@adacore.com>
5437 * config.gcc: Enforce def of TARGET_VXWORKS64 to 1 from
5438 triplet, similar to support for VxWorks7.
5439 * config/vxworks-dummy.h: Provide a default definition
5440 of TARGET_VXWORKS64 to 0.
5442 2018-09-21 Olivier Hainque <hainque@adacore.com>
5444 * config/vxworks.h (TARGET_VXWORKS7): Move default definition ...
5445 * config/vxworks-dummy.h: here.
5447 2018-09-21 Olivier Hainque <hainque@adacore.com>
5449 * config.gcc: Prepend vxworks-dummy.h to tm_file for powerpc*
5451 2018-09-21 Shaokun Zhang <zhangshaokun@hisilicon.com>
5452 Bo Zhou <zbo.zhou@hisilicon.com>
5454 * config/aarch64/aarch64-cores.def (tsv110): New CPU.
5455 * config/aarch64/aarch64-tune.md: Regenerated.
5456 * doc/invoke.texi (AArch64 Options/-mtune): Add "tsv110".
5457 * config/aarch64/aarch64.c (tsv110_tunings): New tuning table.
5458 * config/aarch64/aarch64-cost-tables.h: Add "tsv110" extra costs.
5460 2018-09-21 Andrew Stubbs <ams@codesourcery.com>
5461 Julian Brown <julian@codesourcery.com>
5463 * builtins.c (get_builtin_sync_mem): Handle address spaces.
5465 2018-09-21 Eric Botcazou <ebotcazou@adacore.com>
5467 * config/rs6000/rs6000.c (rs6000_function_ok_for_sibcall): Return false
5468 if the call takes a static chain.
5470 2018-09-21 Martin Liska <mliska@suse.cz>
5472 * auto-profile.c (autofdo_source_profile::read): Do not
5474 (read_profile): Do not add working sets.
5475 (read_autofdo_file): Remove sum_all.
5476 (afdo_callsite_hot_enough_for_early_inline): Remove const
5478 * coverage.c (struct counts_entry): Remove gcov_summary.
5479 (read_counts_file): Read new GCOV_TAG_OBJECT_SUMMARY,
5480 do not support GCOV_TAG_PROGRAM_SUMMARY.
5481 (get_coverage_counts): Remove summary and expected
5483 * coverage.h (get_coverage_counts): Likewise.
5484 * doc/gcov-dump.texi: Remove -w option.
5485 * gcov-dump.c (dump_working_sets): Remove.
5486 (main): Do not support '-w' option.
5487 (print_usage): Likewise.
5488 (tag_summary): Likewise.
5489 * gcov-io.c (gcov_write_summary): Do not dump
5491 (gcov_read_summary): Likewise.
5492 (gcov_histo_index): Remove.
5493 (gcov_histogram_merge): Likewise.
5494 (compute_working_sets): Likewise.
5495 * gcov-io.h (GCOV_TAG_OBJECT_SUMMARY): Mark
5497 (GCOV_TAG_PROGRAM_SUMMARY): Mark it obsolete.
5498 (GCOV_TAG_SUMMARY_LENGTH): Adjust.
5499 (GCOV_HISTOGRAM_SIZE): Remove.
5500 (GCOV_HISTOGRAM_BITVECTOR_SIZE): Likewise.
5501 (struct gcov_summary): Simplify rapidly just
5502 to runs and sum_max fields.
5503 (gcov_histo_index): Remove.
5504 (NUM_GCOV_WORKING_SETS): Likewise.
5505 (compute_working_sets): Likewise.
5506 * gcov-tool.c (print_overlap_usage_message): Remove
5507 trailing empty line.
5508 * gcov.c (read_count_file): Read GCOV_TAG_OBJECT_SUMMARY.
5509 (output_lines): Remove program related line.
5510 * ipa-profile.c (ipa_profile): Do not consider GCOV histogram.
5511 * lto-cgraph.c (output_profile_summary): Do not stream GCOV
5513 (input_profile_summary): Do not read it.
5514 (merge_profile_summaries): And do not merge it.
5515 (input_symtab): Do not call removed function.
5516 * modulo-sched.c (sms_schedule): Do not print sum_max.
5517 * params.def (HOT_BB_COUNT_FRACTION): Reincarnate param that was
5518 removed when histogram method was invented.
5519 (HOT_BB_COUNT_WS_PERMILLE): Mention that it's used only in LTO
5521 * postreload-gcse.c (eliminate_partially_redundant_load): Fix
5523 * predict.c (get_hot_bb_threshold): Use HOT_BB_COUNT_FRACTION
5524 and dump selected value.
5525 * profile.c (add_working_set): Remove.
5526 (get_working_sets): Likewise.
5527 (find_working_set): Likewise.
5528 (get_exec_counts): Do not work with working sets.
5529 (read_profile_edge_counts): Do not inform as sum_max is removed.
5530 (compute_branch_probabilities): Likewise.
5531 (compute_value_histograms): Remove argument for call of
5532 get_coverage_counts.
5533 * profile.h: Do not make gcov_summary const.
5535 2018-09-21 Monk Chiang <sh.chiang04@gmail.com>
5537 * config.gcc (nds32*-*-*): Set TARGET_DEFAULT_TLSDESC_TRAMPOLINE=0.
5539 2018-09-21 Eric Botcazou <ebotcazou@adacore.com>
5541 PR tree-optimization/86990
5542 * gimple-ssa-store-merging.c (imm_store_chain_info:coalesce_immediate):
5543 Check that the entire merged store group is made of constants only for
5546 2018-09-20 Allan Sandfeld Jensen <allan.jensen@qt.io>
5548 * gcc.c (LINK_COMMAND_SPEC): Handle -r like -nostdlib.
5549 (VTABLE_VERIFICATION_SPEC): Likewise.
5550 (SANITIZER_EARLY_SPEC): Likewise.
5551 (SANITIZER_SPEC): Likewise.
5552 * config/darwin.h (LINK_COMMAND_SPEC): Likewise.
5553 * doc/invoke.texi (Link Options): Document -r.
5555 2018-09-20 Richard Biener <rguenther@suse.de>
5558 * gimplify.c (gimplify_expr): Retain alignment of
5559 addressable lvalue in dereference.
5561 2018-09-20 Alexandre Oliva <aoliva@redhat.com>
5564 * configure.ac: Check for .loc is_stmt support.
5565 * configure, config.in: Rebuilt.
5566 * dwarf2out.c (dwarf2out_source_line): Skip is_stmt
5569 2018-09-20 Segher Boessenkool <segher@kernel.crashing.org>
5571 * config/rs6000/rs6000.opt (misel=no, misel=yes): Delete.
5572 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -misel=yes and
5575 2018-09-20 Segher Boessenkool <segher@kernel.crashing.org>
5577 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete
5579 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete
5582 2018-09-20 Marek Polacek <polacek@redhat.com>
5584 * doc/invoke.texi: Add -Wno-init-list-lifetime to C++ Language Options.
5586 2018-09-20 Richard Sandiford <richard.sandiford@arm.com>
5588 PR tree-optimization/87288
5589 * tree-vect-loop.c (vect_analyze_loop_2): Take PEELING_FOR_GAPS
5590 into account when determining PEELING_FOR_NITERS.
5592 2018-09-20 Richard Sandiford <richard.sandiford@arm.com>
5594 PR tree-optimization/86877
5595 * tree-vect-loop.c (vect_analyze_loop_2): Call
5596 vect_verify_datarefs_alignment.
5598 2018-09-19 Marek Polacek <polacek@redhat.com>
5600 * doc/invoke.texi: Document -Wclass-conversion.
5602 2018-09-19 John David Anglin <danglin@gcc.gnu.org>
5604 * config/pa/pa.c (pa_adjust_priority): Delete.
5605 (TARGET_SCHED_ADJUST_PRIORITY): Delete define.
5607 * config/pa/pa.md (atomic_storeqi): Restore deleted expander.
5608 (atomic_storehi): Likewise.
5609 (atomic_storesi): Likewise.
5610 (atomic_loaddi): Restore compare and swap exchange loop code.
5612 2018-09-19 Segher Boessenkool <segher@kernel.crashing.org>
5614 PR rtl-optimization/86902
5615 * combine.c (try_combine): When changing the CC mode used, don't change
5616 an unrelated mode in other_insn to that new CC mode.
5618 2018-09-19 David Malcolm <dmalcolm@redhat.com>
5620 * tree-data-ref.c (runtime_alias_check_p): Use formatted printing
5621 with %T in place of calls to dump_generic_expr.
5622 (prune_runtime_alias_test_list): Likewise.
5623 (create_runtime_alias_checks): Likewise.
5624 * tree-vect-data-refs.c (vect_check_nonzero_value): Likewise.
5625 (vect_analyze_data_ref_dependence): Likewise.
5626 (vect_slp_analyze_data_ref_dependence): Likewise.
5627 (vect_record_base_alignment): Likewise. Use %G in place of call
5628 to dump_gimple_stmt.
5629 (vect_compute_data_ref_alignment): Likewise.
5630 (verify_data_ref_alignment): Likewise.
5631 (vect_find_same_alignment_drs): Likewise.
5632 (vect_analyze_group_access_1): Likewise.
5633 (vect_analyze_data_ref_accesses): Likewise.
5634 (dependence_distance_ge_vf): Likewise.
5635 (dump_lower_bound): Likewise.
5636 (vect_prune_runtime_alias_test_list): Likewise.
5637 (vect_find_stmt_data_reference): Likewise.
5638 (vect_analyze_data_refs): Likewise.
5639 (vect_create_addr_base_for_vector_ref): Likewise.
5640 (vect_create_data_ref_ptr): Likewise.
5641 * tree-vect-loop-manip.c (vect_set_loop_condition): Likewise.
5642 (vect_can_advance_ivs_p): Likewise.
5643 (vect_update_ivs_after_vectorizer): Likewise.
5644 (vect_gen_prolog_loop_niters): Likewise.
5645 (vect_prepare_for_masked_peels): Likewise.
5646 * tree-vect-loop.c (vect_determine_vf_for_stmt): Likewise.
5647 (vect_determine_vectorization_factor): Likewise.
5648 (vect_is_simple_iv_evolution): Likewise.
5649 (vect_analyze_scalar_cycles_1): Likewise.
5650 (vect_analyze_loop_operations): Likewise.
5651 (report_vect_op): Likewise.
5652 (vect_is_slp_reduction): Likewise.
5653 (check_reduction_path): Likewise.
5654 (vect_is_simple_reduction): Likewise.
5655 (vect_create_epilog_for_reduction): Likewise.
5656 (vect_finalize_reduction:): Likewise.
5657 (vectorizable_induction): Likewise.
5658 (vect_transform_loop_stmt): Likewise.
5659 (vect_transform_loop): Likewise.
5660 (optimize_mask_stores): Likewise.
5661 * tree-vect-patterns.c (vect_pattern_detected): Likewise.
5662 (vect_split_statement): Likewise.
5663 (vect_recog_over_widening_pattern): Likewise.
5664 (vect_recog_average_pattern): Likewise.
5665 (vect_determine_min_output_precision_1): Likewise.
5666 (vect_determine_precisions_from_range): Likewise.
5667 (vect_determine_precisions_from_users): Likewise.
5668 (vect_mark_pattern_stmts): Likewise.
5669 (vect_pattern_recog_1): Likewise.
5670 * tree-vect-slp.c (vect_get_and_check_slp_defs): Likewise.
5671 (vect_record_max_nunits): Likewise.
5672 (vect_build_slp_tree_1): Likewise.
5673 (vect_build_slp_tree_2): Likewise.
5674 (vect_print_slp_tree): Likewise.
5675 (vect_analyze_slp_instance): Likewise.
5676 (vect_detect_hybrid_slp_stmts): Likewise.
5677 (vect_detect_hybrid_slp_1): Likewise.
5678 (vect_slp_analyze_operations): Likewise.
5679 (vect_slp_analyze_bb_1): Likewise.
5680 (vect_transform_slp_perm_load): Likewise.
5681 (vect_schedule_slp_instance): Likewise.
5682 * tree-vect-stmts.c (vect_mark_relevant): Likewise.
5683 (vect_mark_stmts_to_be_vectorized): Likewise.
5684 (vect_init_vector_1): Likewise.
5685 (vect_get_vec_def_for_operand): Likewise.
5686 (vect_finish_stmt_generation_1): Likewise.
5687 (vect_check_load_store_mask): Likewise.
5688 (vectorizable_call): Likewise.
5689 (vectorizable_conversion): Likewise.
5690 (vectorizable_operation): Likewise.
5691 (vectorizable_load): Likewise.
5692 (vect_analyze_stmt): Likewise.
5693 (vect_is_simple_use): Likewise.
5694 (vect_get_vector_types_for_stmt): Likewise.
5695 (vect_get_mask_type_for_stmt): Likewise.
5696 * tree-vectorizer.c (increase_alignment): Likewise.
5698 2018-09-19 Andrew Stubbs <ams@codesourcery.com>
5700 * doc/rtl.texi: Adjust vec_select description.
5701 * simplify-rtx.c (simplify_binary_operation_1): Allow VEC_SELECT to use
5702 non-constant selectors.
5704 2018-09-19 Matthew Malcomson <matthew.malcomson@arm.com>
5706 * config/aarch64/aarch64-protos.h
5707 (aarch64_offset_9bit_signed_unscaled_p): New declaration.
5708 * config/aarch64/aarch64.md (arches): New "rcpc8_4" attribute value.
5709 (arch_enabled): Add check for "rcpc8_4" attribute value of "arch".
5710 * config/aarch64/aarch64.h (AARCH64_FL_RCPC8_4): New bitfield.
5711 (AARCH64_FL_FOR_ARCH8_4): Include AARCH64_FL_RCPC8_4.
5712 (AARCH64_FL_PROFILE): Move index so flags are ordered.
5713 (AARCH64_ISA_RCPC8_4): New flag.
5714 * config/aarch64/aarch64.c (offset_9bit_signed_unscaled_p): Renamed
5715 to aarch64_offset_9bit_signed_unscaled_p.
5716 * config/aarch64/atomics.md (atomic_store<mode>): Allow offset
5718 * config/aarch64/constraints.md (Ust): New constraint.
5719 * config/aarch64/predicates.md.
5720 (aarch64_9bit_offset_memory_operand): New predicate.
5721 (aarch64_rcpc_memory_operand): New predicate.
5723 2018-09-19 Eric Botcazou <ebotcazou@adacore.com>
5725 PR rtl-optimization/87361
5726 * rtlanal.c (nonzero_bits1): Revert accidental change.
5728 2018-09-19 Richard Biener <rguenther@suse.de>
5730 PR tree-optimization/87349
5731 PR tree-optimization/87342
5732 * tree-ssa-sccvn.c (do_rpo_vn): Iterate max_rpo computation.
5734 2018-09-18 Marek Polacek <polacek@redhat.com>
5736 P1064R0 - Allowing Virtual Function Calls in Constant Expressions
5737 * gimple-fold.c (gimple_get_virt_method_for_vtable): Adjust assert.
5739 2018-09-18 Segher Boessenkool <segher@kernel.crashing.org>
5741 * config/rs6000/rs6000.md: Remove old "Cygnus sibcall" comment.
5743 2018-09-18 Segher Boessenkool <segher@kernel.crashing.org>
5745 PR rtl-optimization/86882
5746 * rtlanal.c (reg_overlap_mentioned_p): Handle CLOBBER.
5748 2018-09-18 Uros Bizjak <ubizjak@gmail.com>
5750 * config/i386/i386.md (*<code>extend<mode>xf2): Macroize insn from
5751 *<code>extendsfxf2 and *<code>extenddfxf2 using MODEF mode iterator.
5753 2018-09-18 Jonathan Wakely <jwakely@redhat.com>
5756 * doc/invoke.texi (Link Options): Fix formatting and grammar.
5758 2018-09-18 Richard Biener <rguenther@suse.de>
5761 * tree-ssa-coalesce.c (tree_int_map_hasher): Remove.
5762 (compute_samebase_partition_bases): Likewise.
5763 (coalesce_ssa_name): Always use compute_optimized_partition_bases.
5764 (gimple_can_coalesce_p): Simplify.
5766 2018-09-18 Hans-Peter Nilsson <hp@bitrange.com>
5768 Handle a library implementation of ffs calling __builtin_ffs.
5769 * config/mmix/mmix.c (TARGET_INIT_LIBFUNCS): Override with...
5770 (mmix_init_libfuncs): New function: make __builtin_ffs expand
5773 2018-09-17 David Malcolm <dmalcolm@redhat.com>
5775 * diagnostic-show-locus.c (class layout_range): Add field
5777 (layout_range::layout_range): Add "original_idx" param and use it
5778 to initialize new field.
5779 (make_range): Use 0 for original_idx.
5780 (layout::layout): Pass in index to calls to
5781 maybe_add_location_range.
5782 (layout::maybe_add_location_range): Add param "original_idx" and
5783 pass it on to layout_range.
5784 (layout::print_any_labels): Pass on range->m_original_idx to
5786 (gcc_rich_location::add_location_if_nearby): Use 0 for
5788 * gcc-rich-location.h (text_range_label::get_text): Update for new
5790 (range_label_for_type_mismatch::get_text): Likewise.
5792 2018-09-17 Uros Bizjak <ubizjak@gmail.com>
5794 * config/i386/i386.c (ix86_emit_i387_log1p): Emit fldln2 earlier.
5796 2018-09-17 David Malcolm <dmalcolm@redhat.com>
5798 * gimple-ssa-sprintf.c (fmtwarn): Update for introduction of
5799 format_string_diagnostic_t.
5800 (fmtwarn_n): Likewise.
5801 * substring-locations.c
5802 (format_string_diagnostic_t::format_string_diagnostic_t) New ctor.
5803 (format_warning_n_va): Convert to...
5804 (format_string_diagnostic_t::emit_warning_n_va): ...this.
5805 (format_warning_va): Convert to...
5806 (format_string_diagnostic_t::emit_warning_va): ...this.
5807 (format_warning_at_substring): Convert to...
5808 (format_string_diagnostic_t::emit_warning): ...this.
5809 (format_warning_at_substring_n): Convert to...
5810 (format_string_diagnostic_t::emit_warning_n): ...this.
5811 * substring-locations.h (class format_string_diagnostic_t): New
5813 (format_warning_va): Convert to
5814 format_string_diagnostic_t::emit_warning_va.
5815 (format_warning_n_va): Convert to
5816 format_string_diagnostic_t::emit_warning_n_va.
5817 (format_warning_at_substring): Convert to
5818 format_string_diagnostic_t::emit_warning.
5819 (format_warning_at_substring_n): Convert to
5820 format_string_diagnostic_t::emit_warning_n.
5822 2018-09-17 Cesar Philippidis <cesar@codesourcery.com>
5823 Bernd Schmidt <bernds_cb1@t-online.de>
5825 * config/nvptx/nvptx.md (atomic_fetch_<logic><mode>): Enable with
5828 2018-09-17 Uros Bizjak <ubizjak@gmail.com>
5830 * config/i386/i386.md (truncxf<mode>2_i387_noop_unspec): Change
5831 operand 0 predicate to nonimmediate operand.
5832 (rint<mode>2_frndint): Remove insn pattern.
5833 (rint<mode>2): Change operand 1 predicate to general_operand.
5834 Extend operand 1 to XFmode and generate rintxf2 insn.
5835 (frndintxf2_<rounding>): Rename from frndint<mode>2_<rounding>.
5836 Do not use X87MODEF mode macro.
5837 (frndintxf2_<rounding>_i387): Rename from
5838 frndint<mode>2_<rounding>_i387. Do not use X87MODEF mode macro.
5839 (<rounding_insn><mode>2): For non-SSE modes, extend operand 1
5840 to XFmode and generate significandxf3 insn.
5842 2018-09-17 Richard Biener <rguenther@suse.de>
5844 PR tree-optimization/87328
5845 * tree-ssa-sccvn.c (process_bb): Remove assertion about not
5846 visiting unexecutable backedges when not iterating.
5847 (do_rpo_vn): Mark all edges not executable even when not
5850 2018-09-17 Martin Jambor <mjambor@suse.cz>
5853 * doc/invoke.texi (Warning Options): Likewise.
5855 2018-09-17 Richard Biener <rguenther@suse.de>
5857 PR tree-optimization/87301
5858 * tree-ssa-sccvn.c (eliminate_dom_walker::eliminate_cleanup): Properly
5859 clean EH info from leftover copy assignments.
5861 2018-09-17 Martin Liska <mliska@suse.cz>
5863 PR gcov-profile/85871
5864 * gcov.c (output_intermediate_file): Fix out of bounds
5867 2018-09-17 Vineet Gupta <vgupta@synopsys.com>
5869 * config/arc/arc.c: Object attributes for core4 not reflected
5871 * config/arc/arc.h: Don't restrict DBNZ to core3 (core4 includes
5874 2018-09-17 Alexey Brodkin <abrodkin@synopsys.com>
5876 * config/arc/linux.h (LINK_EH_SPEC): Add missing space.
5878 2018-09-17 Martin Liska <mliska@suse.cz>
5880 * doc/gcov.texi: Document new option --use-hotness-colors.
5881 * gcov.c (struct source_info): Declare new field.
5882 (source_info::source_info): Set default for maximum_count.
5883 (print_usage): Add new -q option.
5884 (process_args): Process it.
5885 (accumulate_line_info): Save src->maximum_count.
5886 (output_line_beginning): Make color line number if
5887 flag_use_hotness_colors is set.
5888 (output_line_details): Pass default argument value.
5889 (output_lines): Pass src->maximum_count.
5891 2018-09-17 Martin Liska <mliska@suse.cz>
5893 * common/config/i386/i386-common.c (ix86_get_valid_option_values):
5894 Use processor_names table.
5895 * config/i386/i386.c (ix86_default_align): Use
5896 processor_cost_table for alignment values.
5897 (ix86_option_override_internal): Use processor_names.
5898 (ix86_function_specific_print): Likewise.
5899 * config/i386/i386.h (struct processor_costs):
5900 Add alignment values.
5901 (struct ptt): Remove and replace with const char *.
5902 * config/i386/x86-tune-costs.h (struct processor_costs):
5903 Declare default alignments for all costs.
5905 2018-09-17 Aldy Hernandez <aldyh@redhat.com>
5907 * tree-vrp.c (extract_range_from_unary_expr): Do not special case
5908 symbolics or VR_VARYING ranges for ABS_EXPR.
5909 * wide-int-range.cc (wide_int_range_abs): Return positive numbers
5910 when range will wrap.
5912 2018-09-15 Eric Botcazou <ebotcazou@adacore.com>
5915 * cfgexpand.c (expand_gimple_basic_block): Be prepared for a BARRIER
5916 before and after a JUMP_TABLE_DATA.
5918 2018-09-14 John David Anglin <danglin@gcc.gnu.org>
5921 * dojump.c (do_compare_and_jump): Canonicalize function pointers
5922 when one operand is a function pointer. Use POINTER_TYPE_P and
5923 FUNC_OR_METHOD_TYPE_P.
5924 * expr.c (do_store_flag): Use POINTER_TYPE_P and FUNC_OR_METHOD_TYPE_P.
5925 * fold-const.c (build_range_check): Likewise.
5926 * match.pd (simple_comparison): Likewise.
5928 2018-09-14 David Malcolm <dmalcolm@redhat.com>
5931 * spellcheck.c (get_edit_distance_cutoff): New function.
5932 (selftest::test_edit_distance_unit_test_oneway): Rename to...
5933 (selftest::test_get_edit_distance_one_way): ...this.
5934 (selftest::test_get_edit_distance_unit): Rename to...
5935 (selftest::test_get_edit_distance_both_ways): ...this.
5936 (selftest::test_edit_distances): Move tests to this new function,
5937 and test some more pairs of strings. Update for above renaming.
5938 (selftest::get_old_cutoff): New function.
5939 (selftest::test_get_edit_distance_cutoff): New function.
5940 (selftest::assert_suggested_for): New function.
5941 (ASSERT_SUGGESTED_FOR): New macro.
5942 (selftest::assert_not_suggested_for): New function.
5943 (ASSERT_NOT_SUGGESTED_FOR): New macro.
5944 (selftest::test_suggestions): New function.
5945 (selftest::spellcheck_c_tests): Move test_get_edit_distance_unit
5946 tests to selftest::test_edit_distances and call it. Add calls to
5947 selftest::test_get_edit_distance_cutoff and
5948 selftest::test_suggestions.
5949 * spellcheck.h (get_edit_distance_cutoff): New function declaration.
5950 (best_match::consider): Replace hard-coded cutoff calculation with
5952 (best_match::get_cutoff): New declaration.
5953 (best_match::get_best_meaningful_candidate): Likewise.
5955 2018-09-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
5957 * builtins.c (fold_builtin_strlen): Remove TODO comment.
5959 2018-09-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
5962 2018-07-30 Bernd Edlinger <bernd.edlinger@hotmail.de>
5964 * tree-ssa-forwprop.c (simplify_builtin_call): Don't create a not NUL
5965 terminated string literal.
5967 2018-09-14 Martin Sebor <msebor@redhat.com>
5969 * builtins.c (unterminated_array): Handle ARRAY_REF.
5970 (expand_builtin_stpcpy_1): Detect unterminated char arrays.
5971 * builtins.h (unterminated_array): Declare extern.
5972 * gimple-fold.c (gimple_fold_builtin_stpcpy): Detect unterminated
5974 (gimple_fold_builtin_sprintf): Propagate NO_WARNING to transformed
5977 2018-09-14 Martin Sebor <msebor@redhat.com>
5978 Jeff Law <law@redhat.com>
5980 * builtins.c (unterminated_array): New.
5981 (expand_builtin_strcpy): Adjust.
5982 (expand_builtin_strcpy_args): Detect unterminated arrays.
5983 * gimple-fold.c (get_maxval_strlen): Add argument. Detect
5984 unterminated arrays.
5985 * gimple-fold.h (get_maxval_strlen): Add argument.
5986 (gimple_fold_builtin_strcpy): Detec unterminated arrays.
5988 * gimple-fold.c (get_range_strlen): Add argument.
5989 (get_maxval_strlen): Adjust.
5990 * gimple-fold.h (get_range_strlen): Add argument.
5992 2018-09-14 Wei Xiao <wei3.xiao@intel.com>
5994 * config/i386/movdirintrin.h: Fix copyright year.
5996 2018-09-14 Uros Bizjak <ubizjak@gmail.com>
5998 * reg-stack.c: Include regs.h.
5999 (replace_reg): Assert that mode is MODE_FLOAT or MODE_COMPLEX_FLOAT.
6000 (emit_pop_insn): Default pop insn mode to the reg_raw_mode of
6001 FIRST_STACK_REG, not DFmode.
6002 (emit_swap_insn): Default swap insn mode to the reg_raw_mode of
6003 FIRST_STACK_REG, not XFmode. Explicitly construct swap RTX.
6004 (change stack): Default register mode to the reg_raw_mode of
6005 FIRST_STACK_REG, not DFmode.
6006 * config/i386/i386.md (*swap<mode>): Remove insn pattern.
6007 (*swapxf): Rename from swapxf.
6009 2018-09-14 Carl Love <cel@us.ibm.com>
6011 * config/rs6000/emmintrin.h: Add _MM_SHUFFLE2.
6012 * config/rs6000/xmmintrin.h: Add _MM_SHUFFLE.
6014 2018-09-14 Segher Boessenkool <segher@kernel.crashing.org>
6017 * config/rs6000/rs6000.md (*mov<mode>_hardfloat64): Add Z to the Y
6020 2018-09-14 Sam Tebbs <sam.tebbs@arm.com>
6023 * config/aarch64/aarch64.md (*aarch64_bfxilsi_uxtw): Define.
6025 2018-09-14 Jason Merrill <jason@redhat.com>
6027 Fix --enable-gather-detailed-mem-stats.
6028 * hash-table.c (hash_table_usage): Change from variable to function.
6029 * hash-table.h: Adjust.
6030 * Makefile.in: Add missing dependencies on hash-table.h.
6032 2018-09-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6034 PR tree-optimization/87259
6036 (pass_cse_reciprocals::execute): Run optimize_recip_sqrt after
6037 execute_cse_reciprocals_1 has tried transforming.
6039 2018-09-14 Aldy Hernandez <aldyh@redhat.com>
6041 * tree-vrp.c (extract_range_from_binary_expr_1): Normalize
6042 VR_VARYING for PLUS/MINUS_EXPR.
6044 2018-09-14 Ilya Leoshkevich <iii@linux.ibm.com>
6046 * config/s390/s390-passes.def (INSERT_PASS_BEFORE): Improve
6049 2018-09-14 Richard Biener <rguenther@suse.de>
6052 * tree-ssa-coalesce.c (ssa_conflicts_merge): Remove conflict
6053 bits for the merged partition.
6055 2018-09-13 Martin Sebor <msebor@redhat.com>
6056 Bernd Edlinger <bernd.edlinger@hotmail.de>
6058 * builtins.h (c_srlen): Add argument.
6059 * builtins.c (warn_string_no_nul): New function.
6060 (c_strlen): Add argument and use it. Update recursive calls.
6061 Pass DECL argument to string_constant to get info on non
6062 terminated strings. Update *NONSTR as needed.
6063 (fold_builtin_strlen): Add argument to calls to c_strlen.
6064 Warn for unterminated arrays.
6065 (warn_string_no_null): Add prototype.
6066 * expr.c (string_constant): Update arguments. Update recursive
6067 calls appropriately. Detect missing NUL terminator and outermost
6068 declaration its missing in.
6069 Improve checks for arrays with nonzero lower bound or elements
6070 that are not a single byte. Simplify offset computation.
6071 Simplify checks for non-NUL terminated strings.
6072 * gimple-fold.c (get_range_strlen): Add argument to c_strlen call.
6073 * gimple-ssa-sprintf.c (get_string_length): Remove unnecessary code.
6075 2018-09-13 Bernd Edlinger <bernd.edlinger@hotmail.de>
6077 * builtins.c (c_strlen): Handle not zero terminated STRING_CSTs
6079 * fold-const.c (c_getstr): Fix function comment. Remove unused third
6080 argument. Fix range checks.
6081 * fold-const.h (c_getstr): Adjust protoype.
6082 * gimple-fold.c (gimple_fold_builtin_memory_op): Avoid folding when
6083 string is constant but contains no NUL byte.
6085 * expr.c (string_constant): Adjust function comment.
6086 Remove bogus check for zero termination.
6088 * fold-const.c (c_getstr): Clamp STRING_LENGTH to STRING_SIZE.
6090 * varasm.c (compare_constant): Compare type size of STRING_CSTs.
6091 (get_constant_size): Don't make STRING_CSTs larger than they are.
6092 (check_string_literal): New check function for STRING_CSTs.
6093 (output_constant): Use it.
6095 2018-09-13 Eric Botcazou <ebotcazou@adacore.com>
6098 * config/visium/visium.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define.
6100 2018-09-13 Richard Biener <rguenther@suse.de>
6102 PR tree-optimization/87263
6103 * tree-ssa-sccvn.c (visit_phi): Revert some earlier changes.
6104 (struct unwind_state): Add max_rpo field.
6105 (do_rpo_vn): Allow up-to-date loop state to be used when not iterating.
6106 Compute max_rpo, the max RPO number a block can be backwards reached
6107 from. Re-write non-iterating mode to a RPO ordered worklist approach,
6108 separating it from the iterating mode.
6110 2018-09-13 Vlad Lazar <vlad.lazar@arm.com>
6112 * haifa-sched.c (rank_for_schedule): Schedule by INSN_COST.
6113 (rfs_decision): New scheduling decision.
6115 2018-09-13 Richard Biener <rguenther@suse.de>
6118 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Fix assert.
6119 (vn_nary_op_insert_pieces_predicated): Do not write useless
6120 valid_dominated_by_p entry outside of the allocated storage.
6122 2018-09-13 Omar Sandoval <osandov@osandov.com>
6123 Tom de Vries <tdevries@suse.de>
6126 * dwarf2out.c (is_c): New function.
6127 (add_subscript_info): Add DW_AT_count of 0 for C zero-length arrays.
6129 2018-09-13 Sam Tebbs <sam.tebbs@arm.com>
6132 * config/aarch64/aarch64.md (*aarch64_bfxil):
6134 * config/aarch64/constraints.md (Ulc): Define.
6135 * config/aarch64/aarch64-protos.h (aarch64_high_bits_all_ones_p):
6137 * config/aarch64/aarch64.c (aarch64_high_bits_all_ones_p):
6140 2018-09-13 Vlad Lazar <vlad.lazar@arm.com>
6142 * config/aarch64/aarch64.h (TARGET_COMPUTE_FRAME_LAYOUT): Define.
6143 * config/aarch64/aarch64.c (aarch64_expand_prologue): Remove
6144 aarch64_layout_frame call.
6145 (aarch64_expand_epilogue): Likewise.
6146 (aarch64_initial_elimination_offset): Likewise.
6147 (aarch64_get_separate_components): Likewise.
6148 (aarch64_use_return_insn_p): Likewise.
6149 (aarch64_layout_frame): Remove unneeded check.
6151 2018-09-13 Jakub Jelinek <jakub@redhat.com>
6153 * configure.ac: Only append
6154 " : (reconfigured) $TOPLEVEL_CONFIGURE_ARGUMENTS" to
6155 gcc_config_arguments if it was never reconfigured or last reconfigure
6156 was with different arguments.
6157 * configure: Regenerated.
6159 2018-09-13 Jakub Jelinek <jakub@redhat.com>
6160 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6163 * expr.c (maybe_optimize_pow2p_mod_cmp): New function.
6164 (maybe_optimize_mod_cmp): Use it if integer_pow2p treeop1.
6166 2018-09-13 Jakub Jelinek <jakub@redhat.com>
6168 PR tree-optimization/87287
6169 * fold-const.c (fold_binary_loc) <case EQ_EXPR>: Move signed modulo
6170 X % C == 0 to X % (unsigned) C == 0 optimization to ...
6171 * match.pd (X % C == 0): ... here. New optimization.
6173 2018-09-12 Jakub Jelinek <jakub@redhat.com>
6176 * expr.h (maybe_optimize_mod_cmp): Declare.
6177 * expr.c (mod_inv): New function.
6178 (maybe_optimize_mod_cmp): New function.
6179 (do_store_flag): Use it.
6180 * cfgexpand.c (expand_gimple_cond): Likewise.
6182 2018-09-09 Cesar Philippidis <cesar@codesourcery.com>
6183 Julian Brown <julian@codesourcery.com>
6186 * gimplify.c (gimplify_scan_omp_clauses): Set
6187 target_firstprivatize_array_bases in OpenACC parallel and kernels
6188 region contexts. Remove GOMP_MAP_FIRSTPRIVATE_REFERENCE clauses from
6189 OpenACC data regions.
6191 2018-09-12 Uros Bizjak <ubizjak@gmail.com>
6193 * config/i386/i386.md (sqrt_extend<mode>xf3_i387): Remove.
6194 (sqrt<mode>2): Extend operand 1 to XFmode and generate sqrtxf3 insn.
6196 2018-09-12 Richard Biener <rguenther@suse.de>
6198 PR tree-optimization/87280
6199 * tree-ssa-sccvn.c (process_bb): Handle the case of executable
6200 edge but unreachable target.
6201 (do_rpo_vn): For conservatively handling a PHI only mark
6202 the backedge executable but not the block reachable.
6204 2018-09-12 Richard Biener <rguenther@suse.de>
6206 PR tree-optimization/87266
6207 * tree-ssa-sccvn.c (do_rpo_vn): Always iterate to not yet
6210 2018-09-12 Andreas Krebbel <krebbel@linux.ibm.com>
6212 * config/s390/s390.md (PFPO_RND_MODE_DFP, PFPO_RND_MODE_BFP): New
6214 ("trunc<BFP:mode><DFP_ALL:mode>2")
6215 ("trunc<DFP_ALL:mode><BFP:mode>2")
6216 ("extend<BFP:mode><DFP_ALL:mode>2")
6217 ("extend<DFP_ALL:mode><BFP:mode>2"): Set proper rounding mode
6218 according to the target operand type.
6220 2018-09-12 Jakub Jelinek <jakub@redhat.com>
6221 Andreas Krebbel <krebbel@linux.ibm.com>
6223 PR tree-optimization/86844
6224 * gimple-ssa-store-merging.c
6225 (imm_store_chain_info::coalesce_immediate): For overlapping stores, if
6226 there are any overlapping stores in between them, make sure they are
6227 also coalesced or we give up completely.
6229 2018-09-12 Jakub Jelinek <jakub@redhat.com>
6232 * fold-const.c (fold_ternary_loc) <case COND_EXPR>: Verify also that
6233 BIT_AND_EXPR's second operand is a power of two. Formatting fix.
6235 2018-09-12 Tom de Vries <tdevries@suse.de>
6237 * common.opt (gdescribe-dies): Add option.
6238 * dwarf2out.c (add_name_and_src_coords_attributes): Add description
6239 attribute for artifical and nameless decls.
6240 (dwarf2out_register_external_die): Add description attribute to
6241 external reference die.
6242 (add_desc_attribute): New functions.
6243 (gen_subprogram_die): Add description attribute to
6244 DW_TAG_call_site_parameter.
6245 * tree-pretty-print.c (print_generic_expr_to_str): New function.
6246 * tree-pretty-print.h (print_generic_expr_to_str): Declare.
6247 * doc/invoke.texi (@item Debugging Options): Add -gdescribe-dies and
6249 (@item -gdescribe-dies): Add.
6251 2018-09-12 Aldy Hernandez <aldyh@redhat.com>
6253 * tree-vrp.c (vrp_shift_undefined_p): Remove.
6254 (extract_range_from_binary_expr_1: Call
6255 wide_int_range_shift_undefined_p instead of vrp_shift_undefined_p.
6256 * wide-int-range.h (wide_int_range_shift_undefined_p): Do not
6259 2018-09-12 Aldy Hernandez <aldyh@redhat.com>
6261 * gimple-ssa-warn-alloca.c
6262 (alloca_type_and_limit::alloca_type_and_limit): Initialize limit
6263 field for ALLOCA_BOUND_*_LARGE.
6265 2018-09-11 Nathan Sidwell <nathan@acm.org>
6267 * gcc.c (load_specs, execute, run_attempt): Use %qs not '%s'.
6269 2018-09-11 Uros Bizjak <ubizjak@gmail.com>
6271 * reg-stack.c (subst_asm_stack_regs): Call replace_reg also
6272 for clobbers. Remove obsolete comment.
6274 2018-09-11 Uros Bizjak <ubizjak@gmail.com>
6276 * config/i386/i386.md (define_attr "type"): Remove mpxmov, mpxmk,
6277 mpxchk, mpxld and mpxst types.
6278 (define_attr length_immediate): Remove all processing of mpx types.
6279 (define_attr prefix_0f): Ditto.
6280 (define_attr memory): Ditto.
6282 2018-09-11 Uros Bizjak <ubizjak@gmail.com>
6284 * config/i386/i386.md (fyl2x_extend<mode>xf3_i387): Remove.
6285 (log<mode>2): Change operand 1 predicate to general_operand.
6286 Extend operand 1 to XFmode and generate logxf3 insn.
6287 (log10<mode>2): Change operand 1 predicate to general_operand.
6288 Extend operand 1 to XFmode and generate log10xf3 insn.
6289 (log2<mode>2): Change operand 1 predicate to general_operand.
6290 Extend operand 1 to XFmode and generate log2xf3 insn.
6291 (fyl2xp1_extend<mode>xf3_i387): Remove.
6292 (log1p<mode>2): Change operand 1 predicate to general_operand.
6293 Extend operand 1 to XFmode and generate log1pxf3 insn.
6294 (fxtract_extend<mode>xf3_i387): Remove.
6295 (logb<mode>2): Change operand 1 predicate to general_operand.
6296 Extend operand 1 to XFmode and generate logbxf3 insn.
6297 (ilogb<mode>2): Change operand 1 predicate to general_operand.
6298 Extend operand 1 to XFmode and generate fxtractxf3_i387 insn.
6299 (significand<mode>2): Change operand 1 predicate to general_operand.
6300 Extend operand 1 to XFmode and generate significandxf3 insn.
6302 2018-09-11 Nathan Sidwell <nathan@acm.org>
6304 * gcc.c (perror_with_name, pfatal_with_name): Delete.
6305 (load_specs): Use fatal_error.
6306 (DELETE_IF_ORDINARY, process_command): Use error.
6307 (execute, run_attempt): Use fatal_error.
6309 2018-09-11 Andrew Stubbs <ams@codesourcery.com>
6311 * diagnostic-core.h (sorry_at): New prototype.
6312 * diagnostic.c (sorry_at): New function.
6314 2018-09-11 Aldy Hernandez <aldyh@redhat.com>
6316 * tree-vrp.c (extract_range_from_binary_expr_1): Treat all divisions
6317 by zero as VR_UNDEFINED.
6319 2018-09-10 Uros Bizjak <ubizjak@gmail.com>
6321 * config/i386/i386.md (<sincos>xf2): Rename from *<sincos>xf2_i387.
6322 (*<sincos>_extend<mode>xf2_i387): Remove insn pattern.
6323 (<sincos>mode2): New expander.
6324 (sincos_extend<mode>xf3_i387): Remove insn pattern.
6325 (sincos -> sin, cos splitters): Remove splitter patterns.
6326 (sincos<mode>3): Change operand 2 predicate to general_operand.
6327 Extend operand 2 to XFmode and generate sincosxf3 insn.
6328 (fptanxf4_i387): Change mode of operands 0 and 3 to SFmode.
6329 Change operand 3 predicate to const1_operand.
6330 (fptan_extend<mode>xf4_i387): Remove insn pattern.
6331 (tanxf2): Update operands in the call to fptanxf4_i387.
6332 (tan<mode>2): Change operand 1 predicate to general_operand.
6333 Extend operand 1 to XFmode and generate tanxf3 insn.
6334 (atan2xf3): Rename from *fpatanxf3_i387.
6335 (fpatan_extend<mode>xf3_i387): Remove insn pattern.
6336 (atan2xf3): Remove expander.
6337 (atan2<mode<3): Change operand 1 and 2 predicates to general_operand.
6338 Extend operands 1 and 2 to XFmode and generate atan2xf3 insn.
6339 (atan<mode>2): Change operand 1 predicate to general_operand.
6340 Extend operand 1 to XFmode and generate atanxf3 insn.
6342 2018-09-10 Uros Bizjak <ubizjak@gmail.com>
6344 * config/i386/i386.md (x87/SSE constant load splitter): Use
6345 memory_operand instead of nonimmediate_operand for input operand
6348 2018-09-09 Uros Bizjak <ubizjak@gmail.com>
6350 * config/i386/i386.md (float partial SSE register stall splitter): Move
6351 splitter near its instruction pattern.
6352 (float_extend partial SSE register stall splitter): Ditto.
6353 (float_truncate partial SSE register stall splitter): Ditto.
6355 2018-09-09 Hans-Peter Nilsson <hp@bitrange.com>
6358 * config/mmix/mmix.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Redefine
6359 to speculation_safe_value_not_needed.
6362 * config/mmix/mmix.c (mmix_assemble_integer): Handle byte-size
6363 non-CONST_INT rtx:es using assemble_integer_with_op ".byte".
6364 (MMIX_CFUN_NEEDS_SAVED_EH_RETURN_ADDRESS): Don't call
6365 leaf_function_p, instead use has_hard_reg_initial_val.
6367 2018-09-09 Nathan Sidwell <nathan@acm.org>
6369 * gcc.h (pfatal_with_name): Don't declare here.
6370 * gcc.c (pfatal_with_name): Make static.
6372 2018-09-09 Xianmiao Qu <xianmiao_qu@c-sky.com>
6374 * config/csky/csky.md (*cskyv2_adddi3, *cskyv2_subdi3): Add
6377 2018-09-08 John David Anglin <danglin@gcc.gnu.org>
6379 PR rtl-optimization/85458
6380 * sel-sched.c (sel_target_adjust_priority): Allow backend adjust
6381 priority hook to reduce the priority of EXPR.
6383 2018-09-07 Uros Bizjak <ubizjak@gmail.com>
6385 * config/i386/i386.md (float<SWI48x:mode><MODEF:mode>2) Enable
6386 DImode for x87 on 32bit targets. Conditionally disable x87 modes
6387 with X87_ENABLE_FLOAT. Remove preparation code.
6388 (*float<SWI48:mode><MODEF:mode>2): Rename from
6389 *float<SWI48:mode><MODEF:mode>2_mixed. Handle x87, SSE and mixed
6390 math using "enabled" attribute.
6391 (*floatdi<MODEF:mode>2_i387): Rename from
6392 *float<SWI48x:mode><MODEF:mode>2_i387. Handle only DImode and
6393 enable for 32bit targets only.
6394 (floatdi<X87MODEF:mode>2_i387_with_xmm pre-reload splitter): New
6396 (floatdi<X87MODEF:mode>2_i387_with_xmm): Use register_operand
6397 as operand 1 predicate. Rewrite as define_insn_and_split.
6398 (floatdi<X87MODEF:mode>2_i387_with_xmm memory input splitter): Remove.
6400 2018-09-06 Uros Bizjak <ubizjak@gmail.com>
6402 * reg-stack.c (get_true_reg) <case FLOAT_TRUNCATE>: Reorder
6403 to fallthru to FLOAT case.
6405 2018-09-06 Will Schmidt <will_schmidt@vnet.ibm.com>
6408 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Update logic
6409 around folding of vec_sl to handle out of range shift values.
6411 2018-09-06 Uros Bizjak <ubizjak@gmail.com>
6413 * config/i386/i386.md (fix_trunc<mode>_fisttp_i387_1): Remove.
6414 Update callers to gen_fix_trunc<mode>_i387_fisttp
6415 (fix_trunc<mode>_i387_fisttp): Change operand 0 predicate to
6416 nonimmediate_operand.
6417 (fix_trunc<mode>_i387_fisttp_with_temp): Remove insn pattern
6418 and corresponding splitters.
6419 (*fix_trunc<mode>_i387_1): Always expand with fix_trunc<mode>_i387.
6420 (fix_truncdi_i387): Change operand 0 predicate to nonimmediate_operand.
6421 (fix_truncdi_i387_with_temp): Remove insn pattern
6422 and corresponding splitters.
6423 (fix_trunc<mode>_i387): Change operand 0 predicate to
6424 nonimmediate_operand.
6425 (fix_trunc<mode>_i387_with_temp): Remove insn pattern
6426 and corresponding splitters.
6427 (*fistdi2_1): Remove.
6429 (fistdi2_with_temp): Remove insn pattern and corresponding splitters.
6430 (lrintxfdi2): Remove expander. Reimplement as define_insn.
6431 (*fist<mode>2_1): Remove.
6432 (fist<mode>2): Ditto.
6433 (fist<mode>2_with_temp): Remove insn pattern and corresponding
6435 (lrintxf<mode>2): Remove expander. Reimplement as define_insn.
6436 (*fist<mode>2_<rounding>_1): Always expand with fist<mode>2_<rounding>.
6437 (fistdi2_<rounding>): Change operand 0 predicate to
6438 nonimmediate_operand.
6439 (fistdi2_<rounding>_with_temp): Remove insn pattern
6440 and corresponding splitters.
6441 (fist<mode>2_<rounding>): Change operand 0 predicate to
6442 nonimmediate_operand.
6443 (fist<mode>2_<rounding>_with_temp): Remove insn pattern
6444 and corresponding splitters.
6446 (*fixuns_trunc<mode>si2_avx512f_zext): Depend on TARGET_SSE_MATH.
6448 2018-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
6450 * varasm.c (output_constructor_regular_field): Check TYPE_SIZE_UNIT of
6453 2018-09-06 Will Schmidt <will_schmidt@vnet.ibm.com>
6455 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
6456 early gimple folding of vec_splat().
6457 * tree-vect-generic.c: Remove static from tree_vec_extract() definition.
6458 * gimple-fold.h: Add an extern define for tree_vec_extract().
6460 2018-09-06 Will Schmidt <will_schmidt@vnet.ibm.com>
6462 * config/rs6000/rs6000.c (fold_mergehl_helper): Add types_compatible_p
6463 wrappers around TREE_TYPE comparisons.
6465 2018-09-06 Ilya Leoshkevich <iii@linux.ibm.com>
6468 * config/s390/predicates.md: Add nonsym_memory_operand.
6469 * config/s390/s390.c (s390_legitimize_cs_operand): If operand
6470 contains a SYMBOL_REF, load it into an intermediate pseudo.
6471 (s390_emit_compare_and_swap): Legitimize operand.
6472 * config/s390/s390.md: Use the new nonsym_memory_operand
6473 with UNSPECV_CAS patterns.
6475 2018-09-06 Ilya Leoshkevich <iii@linux.ibm.com>
6478 * config/s390/s390-passes.def: New file.
6479 * config/s390/s390-protos.h (class rtl_opt_pass): Add forward
6481 (make_pass_s390_early_mach): Add declaration.
6482 * config/s390/s390.c (make_pass_s390_early_mach):
6483 (s390_option_override): Remove dynamic registration.
6484 * config/s390/t-s390: Add s390-passes.def.
6486 2018-09-06 Ilya Leoshkevich <iii@linux.ibm.com>
6488 * config/s390/s390.c (s390_decompose_constant_pool_ref):
6489 Remove UNSPEC_LTREL_BASE check.
6490 (annotate_constant_pool_refs): Likewise.
6491 (find_constant_pool_ref): Likewise.
6492 (find_ltrel_base): Removed.
6493 (replace_ltrel_base): Removed.
6494 (s390_mainpool_finish): Remove replace_ltrel_base call.
6495 (s390_chunkify_start): Remove pending LTREL_BASE logic.
6496 (s390_chunkify_finish): Remove replace_ltrel_base call.
6497 * config/s390/s390.md: Remove UNSPEC_LTREL_BASE.
6499 2018-09-06 Hans-Peter Nilsson <hp@axis.com>
6502 * config/cris/cris.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Redefine
6503 to speculation_safe_value_not_needed.
6505 2018-09-05 Cesar Philippidis <cesar@codesourcery.com>
6506 Bernd Schmidt <bernds_cb1@t-online.de>
6508 * config/nvptx/nvptx-opts.h: New file.
6509 * config/nvptx/nvptx.c (nvptx_file_start): Print the correct .target.
6510 * config/nvptx/nvptx.h: Include "nvptx-opts.h".
6512 (TARGET_SM35): New macro.
6513 * config/nvptx/nvptx.md (atomic_fetch_<logic><mode>): Enable with the
6515 * config/nvptx/nvptx.opt (ptx_isa, sm_30, sm_35): New enum and its
6517 (misa=): New option.
6518 * doc/invoke.texi (Nvidia PTX Options): Document -misa.
6520 2018-09-05 Uros Bizjak <ubizjak@gmail.com>
6522 * config/i386/i386.md (truncdfsf2): Remove expander.
6523 (truncdfsf2_with_temp): Ditto.
6524 (truncxf<mode>2): Ditto.
6525 (*truncdfsf_fast_mixed): Remove insn pattern.
6526 (*truncdfsf_fast_i387): Ditto.
6527 (*truncdfsf_mixed): Ditto.
6528 (*truncdfsf_i387): Ditto.
6529 (*truncdfsf2_i387_1): Ditto.
6530 (*truncxfsf2_mixed): Ditto.
6531 (*truncxfdf2_mixed): Ditto.
6532 (*truncxf<mode>2_i387_noop): Ditto. Update callers
6533 to call gen_truncxf<mode>2 instead.
6534 (*truncxf<mode>2_i387): Remove.
6535 (reg->reg splitters): Remove splitter pattern.
6536 (reg->mem splitters): Ditto.
6538 (truncdfsf2): New insn pattern.
6539 (truncxf<mode>2): Ditto.
6541 2018-09-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6543 * tree-ssa-math-opts.c (is_mult_by): New function.
6544 (is_square_of): Use the above.
6545 (optimize_recip_sqrt): New function.
6546 (pass_cse_reciprocals::execute): Use the above.
6548 2018-09-05 Richard Biener <rguenther@suse.de>
6551 * tree-ssa-sccvn.c (rpo_elim::eliminate_push_avail): Make sure
6552 to zero-init the emplaced vec.
6554 2018-09-05 Martin Liska <mliska@suse.cz>
6556 PR tree-optimization/87205
6557 * tree-switch-conversion.c (pass_lower_switch::execute):
6558 Group cases for switch statements.
6560 2018-09-05 Richard Biener <rguenther@suse.de>
6562 PR tree-optimization/87217
6563 * tree-ssa-sccvn.c (vuse_valueize): New.
6564 (vn_reference_lookup_pieces): Use it.
6565 (vn_reference_lookup): Likewise.
6567 2018-09-05 Nathan Sidwell <nathan@acm.org>
6570 * stor-layout.c (place_field): Scan forwards to check last
6571 bitfield when ms_bitfield_placement is in effect.
6573 2018-09-05 Richard Biener <rguenther@suse.de>
6576 * tree-vect-stmts.c (vectorizable_simd_clone_call): Fix bogus
6579 2018-09-05 Siddhesh Poyarekar <siddhesh@sourceware.org>
6580 Kugan Vivekanandarajah <kugan.vivekanandarajah@linaro.org>
6582 * config/aarch64/falkor-tag-collision-avoidance.c: New file.
6583 * config.gcc (extra_objs): Build it.
6584 * config/aarch64/t-aarch64 (falkor-tag-collision-avoidance.o):
6586 * config/aarch64/aarch64-passes.def
6587 (pass_tag_collision_avoidance): New pass.
6588 * config/aarch64/aarch64.c (qdf24xx_tunings): Add
6589 AARCH64_EXTRA_TUNE_RENAME_LOAD_REGS to tuning_flags.
6590 (aarch64_classify_address): Remove static qualifier.
6591 (aarch64_address_info, aarch64_address_type): Move to...
6592 * config/aarch64/aarch64-protos.h: ... here.
6593 (make_pass_tag_collision_avoidance): New function.
6594 * config/aarch64/aarch64-tuning-flags.def (rename_load_regs):
6597 2018-09-05 Martin Liska <mliska@suse.cz>
6599 * doc/gcov.texi: Update documentation of humar
6601 * gcov.c (format_count): Print one decimal place, it provides
6602 more fine number of situations like '1G' vs. '1.4G'.
6604 2018-09-05 Martin Liska <mliska@suse.cz>
6607 * config/rs6000/rs6000.opt: Mark the option as Deprecated.
6608 * optc-gen.awk: Allow 'Var' for Deprecated options in order
6609 to generate a MASK value.
6611 2018-09-04 H.J. Lu <hongjiu.lu@intel.com>
6614 * dwarf2out.c (based_loc_descr): Allow hard frame pointer even
6615 if frame pointer isn't used.
6616 (compute_frame_pointer_to_fb_displacement): Likewise.
6618 2018-09-04 Jakub Jelinek <jakub@redhat.com>
6621 * common/config/i386/i386-common.c (OPTION_MASK_ISA_XSAVEOPT_SET,
6622 OPTION_MASK_ISA_XSAVES_SET, OPTION_MASK_ISA_XSAVEC_SET): Use
6623 OPTION_MASK_ISA_XSAVE_SET instead of OPTION_MASK_ISA_XSAVE.
6624 (OPTION_MASK_ISA_XSAVE_UNSET): Add OPTION_MASK_ISA_XSAVES_UNSET
6625 and OPTION_MASK_ISA_XSAVEC_UNSET.
6627 2018-09-04 Max Filippov <jcmvbkbc@gmail.com>
6629 * config/xtensa/xtensa.c (xtensa_expand_atomic): Reorder AND and
6630 XOR operations in NAND case.
6632 2018-09-04 Aldy Hernandez <aldyh@redhat.com>
6634 * wide-int-range.cc (wide_int_range_convert): New.
6635 * wide-int-range.h (wide_int_range_convert): New.
6636 * tree-vrp.c (extract_range_from_unary_expr): Abstract wide int
6637 code into wide_int_range_convert.
6638 (extract_range_into_wide_ints): Do not munge anti range constants
6639 into the entire domain. Just return the range back.
6641 2018-09-04 Martin Liska <mliska@suse.cz>
6643 * genmatch.c (output_line_directive): Add new argument
6645 (dt_simplify::gen_1): Encapsulate dump within __builtin_expect.
6647 2018-09-04 Jonathan Wakely <jwakely@redhat.com>
6649 * doc/invoke.texi (Option Summary): Add whitespace.
6651 * doc/invoke.texi (Option Summary): Add -Waligned-new.
6653 2018-09-04 Richard Biener <rguenther@suse.de>
6655 PR tree-optimization/87211
6656 * tree-ssa-sccvn.c (visit_phi): When value-numbering to a
6657 backedge value we're supposed to treat as VARYING also number
6658 the PHI to VARYING in case it got a different value-number already.
6660 2018-09-04 Aldy Hernandez <aldyh@redhat.com>
6662 * tree-vrp.c (vrp_can_optimize_bit_op): Remove.
6663 (extract_range_from_binary_expr_1): Do not call
6664 vrp_can_optimize_bit_op.
6665 * wide-int-range.cc (wide_int_range_can_optimize_bit_op): Make
6667 (wide_int_range_get_mask_and_bounds): New.
6668 (wide_int_range_optimize_bit_op): New.
6669 (wide_int_range_bit_ior): Call wide_int_range_optimize_bit_op.
6670 (wide_int_range_bit_and): Same.
6671 * wide-int-range.h (wide_int_range_can_optimize_bit_op): Remove.
6672 (wide_int_range_optimize_bit_op): New.
6673 (wide_int_range_get_mask_and_bounds): New.
6675 2018-09-04 Richard Biener <rguenther@suse.de>
6677 PR tree-optimization/87176
6678 * tree-ssa-sccvn.c (visit_phi): Remove redundant allsame
6679 variable. When value-numbering a virtual PHI node make sure
6680 to not value-number to the backedge value.
6682 2018-09-04 Jonathan Wakely <jwakely@redhat.com>
6684 * doc/extend.texi (Long Long, Hex Floats): Document support for
6685 long long and hex floats in more recent versions of ISO C++.
6687 2018-09-03 Richard Biener <rguenther@suse.de>
6689 PR tree-optimization/87177
6690 * tree-ssa-sccvn.c (vuse_ssa_val): Revert previous change, keep
6693 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
6695 * bb-reorder.c (edge_order): Convert to C-qsort-style
6696 tri-state comparator.
6697 (reorder_basic_blocks_simple): Change std::stable_sort to
6700 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
6702 * tree-loop-distribution.c (offset_cmp): Convert to C-qsort-style
6703 tri-state comparator.
6704 (fuse_memset_builtins): Change std::stable_sort to gcc_stablesort.
6706 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
6708 * sort.cc (struct sort_ctx): New field 'nlim'. Use it...
6709 (mergesort): ... here as maximum count for using netsort.
6710 (gcc_qsort): Set nlim to 3 if stable sort is requested.
6711 (gcc_stablesort): New.
6712 * system.h (gcc_stablesort): Declare.
6714 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
6716 * sort.cc (gcc_qsort) [CHECKING_P]: Call qsort_chk.
6717 * system.h (qsort): Always redirect to gcc_qsort. Update comment.
6718 * vec.c (qsort_chk): Do not call gcc_qsort. Update comment.
6720 2018-09-03 Segher Boessenkool <segher@kernel.crashing.org>
6722 * config/rs6000/rs6000.md (*mov<mode>_hardfloat32): Remove %U from the
6723 lxsdx and stxsdx alternatives.
6724 (*mov<mode>_hardfloat64): Ditto.
6725 * config/rs6000/vsx.md (*vsx_extract_<mode>_store): Ditto.
6727 2018-09-03 Richard Biener <rguenther@suse.de>
6729 PR tree-optimization/87200
6730 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Valueize a
6733 2018-09-03 Martin Liska <mliska@suse.cz>
6735 PR tree-optimization/87201
6736 * tree-switch-conversion.c (switch_decision_tree::balance_case_nodes):
6737 Fix parenthesis in an expression.
6739 2018-09-03 Richard Biener <rguenther@suse.de>
6741 PR tree-optimization/87197
6742 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Mark the new def
6743 visited. CSE the VN_INFO hashtable lookup.
6745 PR tree-optimization/87169
6746 * tree-ssa-sccvn.c (do_rpo_vn): When marking loops for not
6747 iterating make sure there's no extra backedges from irreducible
6748 regions feeding the header. Mark the destination block
6751 2018-09-03 Martin Liska <mliska@suse.cz>
6754 * common/common-target.def: Add TARGET_GET_VALID_OPTION_VALUES.
6755 * common/common-targhooks.c (default_get_valid_option_values):
6757 * common/common-targhooks.h (default_get_valid_option_values):
6759 * common/config/i386/i386-common.c: Move processor_target_table
6761 (ix86_get_valid_option_values): New function.
6762 (TARGET_GET_VALID_OPTION_VALUES): New macro.
6763 * config/i386/i386.c (struct ptt): Move to i386-common.c.
6764 (PTA_*): Move all defined masks into i386-common.c.
6765 (ix86_function_specific_restore): Use new processor_cost_table.
6766 * config/i386/i386.h (struct ptt): Moved from i386.c.
6767 (struct pta): Likewise.
6768 * doc/tm.texi: Document new TARGET_GET_VALID_OPTION_VALUES.
6769 * doc/tm.texi.in: Likewise.
6770 * opt-suggestions.c (option_proposer::suggest_option):
6771 Pass prefix to build_option_suggestions.
6772 (option_proposer::get_completions): Likewise.
6773 (option_proposer::build_option_suggestions): Use the new target
6775 * opts.c (struct option_help_tuple): New struct.
6776 (print_filtered_help): Use the new target hook.
6778 2018-09-03 Martin Liska <mliska@suse.cz>
6781 * predict.c (set_even_probabilities): Add likely_edges
6782 argument and handle cases where we have precisely one
6784 (combine_predictions_for_bb): Catch also likely_edges.
6785 (tree_predict_by_opcode): Handle gswitch statements.
6786 * tree-cfg.h (find_case_label_for_value): New declaration.
6787 (find_taken_edge_switch_expr): Likewise.
6788 * tree-switch-conversion.c (switch_decision_tree::balance_case_nodes):
6789 Find pivot in decision tree based on probabily, not by number of
6792 2018-09-02 Gerald Pfeifer <gerald@pfeifer.com>
6794 * doc/standards.texi (Standards): Update Objective-C reference.
6796 2018-09-01 Gerald Pfeifer <gerald@pfeifer.com>
6798 * doc/install.texi (Prerequisites): Update link for MPC.
6800 2018-09-01 Michael Matz <matz@suse.de>
6802 PR tree-optimization/87074
6803 * gimple-loop-jam.c (unroll_jam_possible_p): Check loop exit
6804 PHIs for outer-loop uses.
6806 2018-09-01 Gerald Pfeifer <gerald@pfeifer.com>
6808 * doc/generic.texi (OpenMP): Adjust link to openmp.org.
6809 * doc/invoke.texi (C Dialect Options): Ditto.
6811 2018-09-01 Gerald Pfeifer <gerald@pfeifer.com>
6813 * doc/install.texi (Prerequisites): Adjust link mpfr.org.
6815 2018-08-31 Richard Biener <rguenther@suse.de>
6817 PR tree-optimization/87168
6818 * tree-ssa-sccvn.c (SSA_VAL): Add visited output parameter.
6819 (rpo_elim::eliminate_avail): When OP was not visited it must
6822 2018-08-31 David Malcolm <dmalcolm@redhat.com>
6824 * tree-vrp.c (copy_value_range): Convert param "from" from
6825 "value_range *" to "const value_range *".
6826 (range_is_null): Likewise for param "vr".
6827 (range_int_cst_p): Likewise.
6828 (range_int_cst_singleton_p): Likewise.
6829 (symbolic_range_p): Likewise.
6830 (value_ranges_intersect_p): Likewise for both params.
6831 (value_range_nonnegative_p): Likewise for param "vr".
6832 (value_range_constant_singleton): Likewise.
6833 (vrp_set_zero_nonzero_bits): Likewise for param "ar".
6834 (extract_range_into_wide_ints): Likewise for param "vr".
6835 (extract_range_from_multiplicative_op): Likewise for params "vr0"
6837 (vrp_can_optimize_bit_op): Likewise.
6838 (extract_range_from_binary_expr_1): Likewise for params "vr0_" and
6840 (extract_range_from_unary_expr): Likewise.
6841 (debug_value_range): Likewise for param "vr".
6842 (value_range::dump): Add "const" qualifier.
6843 (vrp_prop::check_array_ref): Convert local "vr" from
6844 "value_range *" to "const value_range *".
6845 (vrp_prop::check_mem_ref): Likewise.
6846 (vrp_prop::visit_stmt): Likewise for local "old_vr".
6847 (vrp_intersect_ranges_1): Likewise for param "vr_1".
6848 (vrp_intersect_ranges): Likewise.
6849 (simplify_stmt_for_jump_threading): Likewise for local "vr".
6850 (vrp_prop::vrp_finalize): Likewise.
6851 * tree-vrp.h (value_range::dump): Add "const" qualifier.
6852 (vrp_intersect_ranges): Add "const" qualifier to params as above.
6853 (extract_range_from_unary_expr): Likewise.
6854 (value_range_constant_singleton): Likewise.
6855 (symbolic_range_p): Likewise.
6856 (copy_value_range): Likewise.
6857 (extract_range_from_binary_expr_1): Likewise.
6858 (range_int_cst_p): Likewise.
6859 (vrp_set_zero_nonzero_bits): Likewise.
6860 (range_int_cst_singleton_p): Likewise.
6862 2018-08-31 Vlad Lazar <vlad.lazar@arm.com>
6864 * config/aarch64/arm_neon.h (vabsd_s64): New.
6865 (vnegd_s64): Likewise.
6867 2018-08-31 Martin Jambor <mjambor@suse.cz>
6869 * ipa-cp.c (estimate_local_effects): Replace wrong MAX with MIN.
6871 2018-08-31 Martin Liska <mliska@suse.cz>
6873 * ipa-icf.c (sem_item::add_type): Use
6874 sem_item::m_type_hash_cache.
6875 * ipa-icf.h: Move the cache from sem_item_optimizer
6878 2018-08-31 Nathan Sidwell <nathan@acm.org>
6880 * doc/extend.texi (Backwards Compatibility): Remove implicit
6881 extern C leeway of () being (...).
6883 2018-08-31 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6885 * ipa-inline.c (can_inline_edge_by_limits_p): Fix typos in comment.
6887 2018-08-31 Segher Boessenkool <segher@kernel.crashing.org>
6891 * config/rs6000/rs6000.md (lround<mode>di2): Gate on TARGET_FPRND.
6893 2018-08-31 Jakub Jelinek <jakub@redhat.com>
6896 * expmed.c (expand_mult_const): Use immed_wide_int_const instead of
6897 gen_int_mode. Formatting fixes.
6899 2018-08-30 Sandra Loosemore <sandra@codesourcery.com>
6901 * target.def (custom_function_descriptors): Improve documentation.
6902 * doc/tm.texi.in (Trampolines): Expand discussion of function
6903 descriptors and move TARGET_CUSTOM_FUNCTION_DESCRIPTORS to the
6904 beginning of the section.
6905 * doc/tm.texi: Regenerated.
6907 2018-08-30 Jose E. Marchesi <jose.marchesi@oracle.com>
6909 * cfg.h (class auto_edge_flag): Spell out the template-id of the
6910 base class in the initializer list. This is a workaround for
6911 building with older GCC.
6912 (class auto_bb_flag): Likewise.
6914 2018-08-30 Aaron Sawdey <acsawdey@linux.ibm.com>
6916 * config/rs6000/altivec.md (altivec_eq<mode>): Remove star.
6917 (altivec_vcmpequ<VI_char>_p): Remove star.
6918 * config/rs6000/rs6000-string.c (do_load_for_compare): Support
6920 (expand_strncmp_vec_sequence): New function.
6921 (emit_final_str_compare_vec): New function.
6922 (expand_strn_compare): Add support for vector strncmp.
6923 * config/rs6000/rs6000.opt (-mstring-compare-inline-limit): Change
6924 length specification to bytes.
6925 * config/rs6000/vsx.md (vsx_ld_elemrev_v16qi_internal): Remove star.
6926 (vcmpnezb_p): New pattern.
6927 * doc/invoke.texi (RS/6000 and PowerPC Options): Update documentation
6928 for option -mstring-compare-inline-limit.
6930 2018-08-30 Thiago Macieira <thiago.macieira@intel.com>
6932 * config/i386/i386.c (PTA_WESTMERE): Remove PTA_AES.
6933 (PTA_SKYLAKE): Add PTA_AES.
6934 (PTA_GOLDMONT): Likewise.
6936 2018-08-29 Jan Hubicka <jh@suse.cz>
6939 * lto-opts.c (lto_write_options): Always stream PIC/PIE mode.
6940 * lto-wrapper.c (merge_and_complain): Fix merging of PIC/PIE.
6942 2018-08-29 Jan Hubicka <jh@suse.cz>
6944 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not follow
6946 (hash_tree): Do not visit TYPE_STUB_DECL.
6947 * tree-streamer-out.c (write_ts_type_common_tree_pointers): Do not
6948 stream TYPE_STUB_DECL.
6949 * tree-streamer-in.c (lto_input_ts_type_common_tree_pointers): Likewise.
6950 * ipa-utils.h (type_with_linkage_p): Do not rely on TYPE_STUB_DECL
6951 after free_lang_data.
6952 (type_in_anonymous_namespace_p): Likewise.
6954 2018-08-29 Jan Hubicka <jh@suse.cz>
6956 * sreal.h (SREAL_PART_BITS): Change to 31; remove seemingly unnecessary
6957 comment that it has to be even number.
6958 (class sreal): Change m_sig type to int32_t.
6959 * sreal.c (sreal::dump, sreal::to_int, opreator+, operator-): Use
6960 int64_t for temporary calculations.
6961 (sreal_verify_basics): Drop one bit from minimum and maximum.
6963 2018-08-30 Richard Biener <rguenther@suse.de>
6965 PR tree-optimization/87147
6966 * tree-ssa-sccvn.c (SSA_VISITED): New function.
6967 (visit_phi): When the degenerate result is from the backedge and
6968 we didn't visit its definition yet drop to VARYING.
6969 (do_rpo_vn): Properly mark blocks with incoming backedges as executable.
6971 2018-08-29 Jan Hubicka <jh@suse.cz>
6973 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not walk
6975 (hash_tree): Likewise.
6977 2018-08-29 Jan Hubicka <jh@suse.cz>
6979 * tree.c (find_decls_types_r): Walk also TYPE_NEXT_PTR_TO
6980 and TYPE_NEXT_REF_TO.
6982 2018-08-29 Jan Hubicka <jh@suse.cz>
6984 * sreal.h (SREAL_PART_BITS): Change to 31; remove seemingly unnecessary
6985 comment that it has to be even number.
6986 (class sreal): Change m_sig type to int32_t.
6987 * sreal.c (sreal::dump, sreal::to_int, opreator+, operator-): Use
6988 int64_t for temporary calculations.
6989 (sreal_verify_basics): Drop one bit from minimum and maximum.
6991 2018-08-30 Tamar Christina <tamar.christina@arm.com>
6993 * config/aarch64/aarch64.c (aarch64_expand_movmem): Set TImode max.
6995 2018-08-30 Vlad Lazar <vlad.lazar@arm.com>
6998 * expmed.c (canonicalize_comparison): Use wi::sub instead of wi::add
6999 if to_add is negative.
7001 2018-08-29 Bernd Edlinger <bernd.edlinger@hotmail.de>
7004 * builtins.c (c_strlen): Improve range checks.
7006 2018-08-29 Martin Sebor <msebor@redhat.com>
7007 Jeff Law <law@redhat.com>
7009 PR tree-optimization/86714
7010 PR tree-optimization/86711
7011 * builtins.c (c_strlen): Add arguments to call to string_constant.
7012 * expr.c (string_constant): Add argument. Detect missing nul
7013 terminator and outermost declaration it's missing in.
7014 * expr.h (string_constant): Add argument.
7015 * fold-const.c (read_from_constant_string): Add arguments to call to
7017 (c_getstr): Likewise.
7018 * tree-ssa-forwprop.c (simplify_builtin_call): Likewise.
7020 * tree-ssa-strlen.c (get_stridx): Likewise.
7022 2018-08-29 Jan Hubicka <jh@suse.cz>
7024 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers):
7025 Do not stream DECL_VINDEX.
7026 * tree-streamer-out.c (write_ts_function_decl_tree_pointers): Likewise.
7027 * tree.c (free_lang_data_in_decl): Clear DECL_VINDEX.
7028 (decl_function_context): Use DECL_VIRTUAL_P rather than DECL_VINDEX.
7030 2018-08-29 Richard Biener <rguenther@suse.de>
7032 * tree-ssa-sccvn.c (vuse_ssa_val): Return NULL for unvisited
7033 virtual operands that are not default defs to honor region
7035 (rpo_vn_valueize): Remove ineffective code here.
7037 2018-08-29 Richard Biener <rguenther@suse.de>
7039 PR tree-optimization/87132
7040 * tree-ssa-alias.c (get_continuation_for_phi): Do not translate
7041 when skipping defs reachable over backedges.
7043 2018-08-29 Richard Biener <rguenther@suse.de>
7045 * tree-core.h: Document use of deprecated_flag in SSA_NAME.
7046 * tree.h (SSA_NAME_POINTS_TO_READONLY_MEMORY): Define.
7047 * tree-into-ssa.c (pass_build_ssa::execute): Initialize
7048 function parameters SSA_NAME_POINTS_TO_READONLY_MEMORY from fnspec.
7049 * tree-ssa-sccvn.c (const_parms, init_const_parms): Remove.
7050 (vn_reference_lookup_3): Remove use of const_parms.
7051 (free_rpo_vn): Do not free const_parms.
7052 (do_rpo_vn): Do not call init_const_parms.
7053 * tree-ssa-alias.c (refs_may_alias_p_1): Honor
7054 SSA_NAME_POINTS_TO_READONLY_MEMORY.
7055 (call_may_clobber_ref_p_1): Likewise.
7057 2018-08-29 Alexander Monakov <amonakov@ispras.ru>
7060 * invoke.texi (Optimization Options): List -ftree-scev-cprop.
7062 (-ftree-scev-cprop): Document.
7064 2018-08-29 Jan Hubicka <jh@suse.cz>
7066 * sreal.h (normalize, normalize_up, normalize_down): Add new_sig/new_exp
7068 (sreal constructor): Update.
7069 * sreal.c (sreal:operator+, sreal:operator-, sreal:operator*,
7070 sreal:operator/): Update.
7072 2018-08-29 Martin Liska <mliska@suse.cz>
7074 * tree-switch-conversion.c (switch_conversion::expand):
7075 Strenghten assumption about gswitch statements.
7077 2018-08-29 Richard Biener <rguenther@suse.de>
7079 PR tree-optimization/87117
7080 * tree-ssa-sccvn.c (eliminate_dom_walker::eliminate_stmt): Only
7081 re-value-number released SSA VDEFs.
7083 2018-08-29 Richard Biener <rguenther@suse.de>
7085 PR tree-optimization/87126
7086 * tree-ssa-sccvn.c (vn_reference_insert): Remove assert.
7088 2018-08-28 Jim Wilson <jimw@sifive.com>
7090 * config/riscv/pic.md: Rewrite.
7091 * config/riscv/riscv.c (riscv_address_insns): Return cost of 3 for
7093 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Delete.
7094 (SOFTF, default_load, softload, softstore): New.
7096 2018-08-28 Jeff Law <law@redhat.com>
7098 * fold-const.c (fold_binary_loc): Remove recently added assert.
7100 2018-08-28 Joern Rennecke <joern.rennecke@riscy-ip.com>
7102 * genpreds.c (write_predicate_subfunction): Also add ATTRIBUTE_UNUSED
7103 to OP parmeter of generated function.
7105 2018-08-28 MCC CS <deswurstes@users.noreply.github.com>
7107 PR tree-optimization/87009
7108 * match.pd: Add boolean optimizations.
7110 2018-08-28 Martin Sebor <msebor@redhat.com>
7113 * calls.c (alloc_max_size): Treat HOST_WIDE_INT special.
7114 * gimple-ssa-warn-alloca.c (adjusted_warn_limit): New function.
7115 (pass_walloca::gate): Use it.
7116 (alloca_call_type): Same.
7117 (pass_walloca::execute): Same.
7118 * stor-layout.c (layout_decl): Treat HOST_WIDE_INT special.
7120 2018-08-28 David Malcolm <dmalcolm@redhat.com>
7122 * dumpfile.h (ATTRIBUTE_GCC_DUMP_PRINTF): Change version check on
7123 GCC_VERSION for usage of "__gcc_dump_printf__" format from
7126 2018-08-28 Richard Biener <rguenther@suse.de>
7128 PR tree-optimization/87124
7129 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Guard against
7130 constants before looking up avail.
7132 2018-08-28 Jakub Jelinek <jakub@redhat.com>
7135 * calls.c (maybe_warn_nonstring_arg): Punt early if
7136 warn_stringop_overflow is zero. Don't call get_range_strlen
7137 on 3rd argument, keep iterating until lenrng[1] is INTEGER_CST.
7138 Swap comparison operands to have constants on rhs. Only use
7139 lenrng[1] if non-NULL and INTEGER_CST. Don't uselessly
7140 increment lenrng[0].
7142 2018-08-28 Richard Sandiford <richard.sandiford@arm.com>
7144 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Fix unguarded
7145 use of tree_to_shwi. Remove duplicated test for the size being
7146 a whole number of bytes.
7148 2018-08-28 Richard Biener <rguenther@suse.de>
7150 PR tree-optimization/87117
7151 * tree-ssa-sccvn.c (eliminate_dom_walker::eliminate_cleanup):
7152 Handle removed stmt without LHS (GIMPLE_NOP).
7154 2018-08-28 Richard Biener <rguenther@suse.de>
7156 PR tree-optimization/87117
7157 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Exclude
7158 void which is is_gimple_reg_type by checking for COMPLETE_TYPE_P.
7160 2018-08-28 Richard Biener <rguenther@suse.de>
7162 PR tree-optimization/87117
7163 * tree-ssa-pre.c (compute_avail): Do not make expressions
7164 with predicated values available.
7165 (get_expr_value_id): Assert we do not run into predicated value
7168 2018-08-28 Richard Biener <rguenther@suse.de>
7170 PR tree-optimization/87117
7171 * tree-ssa-operands.c (add_stmt_operand): STRING_CST may
7172 get virtual operands.
7173 (get_expr_operands): Handle STRING_CST like other decls.
7175 2018-08-28 Martin Liska <mliska@suse.cz>
7177 * tree.h: Update documentation of fndecl_built_in_p
7181 2018-08-27 Jeff Law <law@redhat.com>
7182 PR tree-optimization/87110
7183 * tree-ssa-dse.c (compute_trims): Handle non-constant
7186 2018-08-27 Martin Sebor <msebor@redhat.com>
7188 PR tree-optimization/86914
7189 * tree-ssa-strlen.c (maybe_set_strlen_range): Avoid MEM_REF.
7191 2018-08-27 Martin Sebor <msebor@redhat.com>
7193 PR tree-optimization/87112
7194 * builtins.c (expand_builtin_strnlen): Convert c_strlen result to
7195 the type of the bound argument.
7197 2018-08-27 Jeff Law <law@redhat.com>
7199 * tree-ssa-dse.c (compute_trims): Handle case where the reference's
7200 type does not have a TYPE_SIZE_UNIT.
7202 2018-08-27 Steve Ellcey <sellcey@cavium.com>
7204 * config/aarch64/aarch64-speculation.cc: Replace include of cfg.h
7205 with include of backend.h.
7207 2018-08-27 Richard Biener <rguenther@suse.de>
7209 PR tree-optimization/86927
7210 * tree-vect-loop.c (vect_create_epilog_for_reduction): Properly
7211 use const cond reduction code.
7213 2018-08-27 Alexander Monakov <amonakov@ispras.ru>
7215 PR tree-optimization/85758
7216 * match.pd ((X & Y) ^ Y): Add :s qualifier to inner expression.
7218 2018-08-27 David Malcolm <dmalcolm@redhat.com>
7221 * diagnostic-show-locus.c (class layout_range): Update for
7222 conversion of show_caret_p to a tri-state.
7223 (layout_range::layout_range): Likewise.
7224 (make_range): Likewise.
7225 (layout::maybe_add_location_range): Likewise.
7226 (layout::should_print_annotation_line_p): Don't show annotation
7227 lines for ranges that are SHOW_LINES_WITHOUT_RANGE.
7228 (layout::get_state_at_point): Update for conversion of
7229 show_caret_p to a tri-state. Bail out early for
7230 SHOW_LINES_WITHOUT_RANGE, so that such ranges don't affect
7231 underlining or source colorization.
7232 (gcc_rich_location::add_location_if_nearby): Update for conversion
7233 of show_caret_p to a tri-state.
7234 (selftest::test_one_liner_multiple_carets_and_ranges): Likewise.
7235 (selftest::test_one_liner_fixit_replace_equal_secondary_range):
7237 (selftest::test_one_liner_labels): Likewise.
7238 * gcc-rich-location.c (gcc_rich_location::add_expr): Update for
7239 conversion of show_caret_p to a tri-state.
7240 * pretty-print.c (text_info::set_location): Likewise.
7241 * pretty-print.h (text_info::set_location): Likewise.
7242 * substring-locations.c (format_warning_n_va): Likewise.
7243 * tree-diagnostic.c (default_tree_printer): Likewise.
7244 * tree-pretty-print.c (newline_and_indent): Likewise.
7246 2018-08-27 David Malcolm <dmalcolm@redhat.com>
7249 * diagnostic-show-locus.c (get_line_span_for_fixit_hint): Show the
7250 line above for line-insertion fix-it hints.
7251 (selftest::test_fixit_insert_containing_newline): Update the
7252 expected results, and add a test with line-numbering enabled.
7254 2018-08-27 Martin Liska <mliska@suse.cz>
7257 * sanopt.c (sanitize_rewrite_addressable_params): Ignore
7258 params with DECL_HAS_VALUE_EXPR_P.
7260 2018-08-27 Martin Liska <mliska@suse.cz>
7262 * config/i386/i386.c (ix86_expand_set_or_movmem): Dump
7263 selected expansion strategy.
7265 2018-08-27 Martin Liska <mliska@suse.cz>
7267 * builtins.h (is_builtin_fn): Remove and fndecl_built_in_p.
7268 * builtins.c (is_builtin_fn): Likewise.
7269 * attribs.c (diag_attr_exclusions): Use new function
7270 fndecl_built_in_p and remove check for FUNCTION_DECL if
7272 (builtin_mathfn_code): Likewise.
7273 (fold_builtin_expect): Likewise.
7274 (fold_call_expr): Likewise.
7275 (fold_builtin_call_array): Likewise.
7276 (fold_call_stmt): Likewise.
7277 (set_builtin_user_assembler_name): Likewise.
7278 (is_simple_builtin): Likewise.
7279 * calls.c (gimple_alloca_call_p): Likewise.
7280 (maybe_warn_nonstring_arg): Likewise.
7281 * cfgexpand.c (expand_call_stmt): Likewise.
7282 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Likewise.
7283 (cgraph_edge::verify_corresponds_to_fndecl): Likewise.
7284 (cgraph_node::verify_node): Likewise.
7285 * cgraphclones.c (build_function_decl_skip_args): Likewise.
7286 (cgraph_node::create_clone): Likewise.
7287 * config/arm/arm.c (arm_insert_attributes): Likewise.
7288 * config/i386/i386.c (ix86_gimple_fold_builtin): Likewise.
7289 * dse.c (scan_insn): Likewise.
7290 * expr.c (expand_expr_real_1): Likewise.
7291 * fold-const.c (operand_equal_p): Likewise.
7292 (fold_binary_loc): Likewise.
7293 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
7294 * gimple-low.c (lower_stmt): Likewise.
7295 * gimple-pretty-print.c (dump_gimple_call): Likewise.
7296 * gimple-ssa-warn-restrict.c (wrestrict_dom_walker::check_call):
7298 * gimple.c (gimple_build_call_from_tree): Likewise.
7299 (gimple_call_builtin_p): Likewise.
7300 (gimple_call_combined_fn): Likewise.
7301 * gimplify.c (gimplify_call_expr): Likewise.
7302 (gimple_boolify): Likewise.
7303 (gimplify_modify_expr): Likewise.
7304 (gimplify_addr_expr): Likewise.
7305 * hsa-gen.c (gen_hsa_insns_for_call): Likewise.
7306 * ipa-cp.c (determine_versionability): Likewise.
7307 * ipa-fnsummary.c (compute_fn_summary): Likewise.
7308 * ipa-param-manipulation.c (ipa_modify_formal_parameters): Likewise.
7309 * ipa-split.c (visit_bb): Likewise.
7310 (split_function): Likewise.
7311 * ipa-visibility.c (cgraph_externally_visible_p): Likewise.
7312 * lto-cgraph.c (input_node): Likewise.
7313 * lto-streamer-out.c (write_symbol): Likewise.
7314 * omp-low.c (setjmp_or_longjmp_p): Likewise.
7315 (lower_omp_1): Likewise.
7316 * predict.c (strip_predict_hints): Likewise.
7317 * print-tree.c (print_node): Likewise.
7318 * symtab.c (symtab_node::output_to_lto_symbol_table_p): Likewise.
7319 * trans-mem.c (is_tm_irrevocable): Likewise.
7320 (is_tm_load): Likewise.
7321 (is_tm_simple_load): Likewise.
7322 (is_tm_store): Likewise.
7323 (is_tm_simple_store): Likewise.
7324 (is_tm_abort): Likewise.
7325 (tm_region_init_1): Likewise.
7326 * tree-call-cdce.c (gen_shrink_wrap_conditions): Likewise.
7327 * tree-cfg.c (verify_gimple_call): Likewise.
7328 (move_stmt_r): Likewise.
7329 (stmt_can_terminate_bb_p): Likewise.
7330 * tree-eh.c (lower_eh_constructs_2): Likewise.
7331 * tree-if-conv.c (if_convertible_stmt_p): Likewise.
7332 * tree-inline.c (remap_gimple_stmt): Likewise.
7333 (copy_bb): Likewise.
7334 (estimate_num_insns): Likewise.
7335 (fold_marked_statements): Likewise.
7336 * tree-sra.c (scan_function): Likewise.
7337 * tree-ssa-ccp.c (surely_varying_stmt_p): Likewise.
7338 (optimize_stack_restore): Likewise.
7339 (pass_fold_builtins::execute): Likewise.
7340 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
7341 (mark_all_reaching_defs_necessary_1): Likewise.
7342 * tree-ssa-dom.c (dom_opt_dom_walker::optimize_stmt): Likewise.
7343 * tree-ssa-forwprop.c (simplify_builtin_call): Likewise.
7344 (pass_forwprop::execute): Likewise.
7345 * tree-ssa-loop-im.c (stmt_cost): Likewise.
7346 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
7347 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Likewise.
7348 * tree-ssa-strlen.c (get_string_length): Likewise.
7349 * tree-ssa-structalias.c (handle_lhs_call): Likewise.
7350 (find_func_aliases_for_call): Likewise.
7351 * tree-ssa-ter.c (find_replaceable_in_bb): Likewise.
7352 * tree-stdarg.c (optimize_va_list_gpr_fpr_size): Likewise.
7353 * tree-tailcall.c (find_tail_calls): Likewise.
7354 * tree.c (need_assembler_name_p): Likewise.
7355 (free_lang_data_in_decl): Likewise.
7356 (get_call_combined_fn): Likewise.
7357 * ubsan.c (is_ubsan_builtin_p): Likewise.
7358 * varasm.c (incorporeal_function_p): Likewise.
7359 * tree.h (DECL_BUILT_IN): Remove and replace with
7361 (DECL_BUILT_IN_P): Transfort to fndecl_built_in_p.
7362 (fndecl_built_in_p): New.
7364 2018-08-27 Martin Liska <mliska@suse.cz>
7366 PR tree-optimization/86847
7367 * tree-switch-conversion.c (switch_decision_tree::dump_case_nodes):
7368 Dump also subtree probability.
7369 (switch_decision_tree::do_jump_if_equal): New function.
7370 (switch_decision_tree::emit_case_nodes): Handle special
7371 situations in balanced tree that can be emitted much simpler.
7372 Fix calculation of probabilities that happen in tree expansion.
7373 * tree-switch-conversion.h (struct cluster): Add
7375 (struct simple_cluster): Likewise.
7376 (struct case_tree_node): Add new function has_child.
7377 (do_jump_if_equal): New.
7379 2018-08-27 Martin Liska <mliska@suse.cz>
7381 * tree-switch-conversion.c (bit_test_cluster::find_bit_tests):
7382 Add new argument to bit_test_cluster constructor.
7383 (bit_test_cluster::emit): Set bits really number of values
7385 (bit_test_cluster::hoist_edge_and_branch_if_true): Add
7386 probability argument.
7387 * tree-switch-conversion.h (struct bit_test_cluster):
7388 Add m_handles_entire_switch.
7390 2018-08-27 Martin Liska <mliska@suse.cz>
7392 PR tree-optimization/86702
7393 * tree-switch-conversion.c (jump_table_cluster::emit):
7394 Make probabilities even for values in jump table
7395 according to number of cases handled.
7396 (switch_decision_tree::compute_cases_per_edge): Pass
7397 argument to reset_out_edges_aux function.
7398 (switch_decision_tree::analyze_switch_statement): Likewise.
7399 * tree-switch-conversion.h (switch_decision_tree::reset_out_edges_aux):
7402 2018-08-27 Martin Liska <mliska@suse.cz>
7404 * cfgexpand.c (expand_asm_stmt): Use label_to_block and pass
7405 cfun argument explicitly.
7406 * gimple-pretty-print.c (dump_gimple_switch): Likewise.
7407 * hsa-gen.c (gen_hsa_insns_for_switch_stmt): Use new
7408 function gimple_switch_default_bb.
7409 (convert_switch_statements):
7411 * ipa-fnsummary.c (set_switch_stmt_execution_predicate):
7412 * stmt.c (label_to_block_fn): Use label_to_block and pass
7413 cfun argument explicitly and use gimple_switch_label_bb.
7414 (expand_case): Likewise.
7415 * tree-cfg.c (lower_phi_internal_fn): Use label_to_block and pass
7416 cfun argument explicitly. Likewise.
7417 (make_edges_bb): Likewise.
7418 (make_cond_expr_edges): Likewise.
7419 (get_cases_for_edge): Likewise.
7420 (make_gimple_switch_edges): Likewise.
7421 (label_to_block_fn): Likewise.
7422 (label_to_block): Likewise.
7423 (make_goto_expr_edges): Likewise.
7424 (make_gimple_asm_edges): Likewise.
7425 (main_block_label): Likewise.
7426 (group_case_labels_stmt): Likewise.
7427 (find_taken_edge_computed_goto): Likewise.
7428 (find_taken_edge_switch_expr): Likewise.
7429 (gimple_verify_flow_info): Likewise.
7430 (gimple_redirect_edge_and_branch): Likewise.
7431 (gimple_switch_label_bb): New function.
7432 (gimple_switch_default_bb): Likewise.
7433 (gimple_switch_edge): Likewise.
7434 (gimple_switch_default_edge): Likewise.
7435 * tree-cfg.h (label_to_block_fn): Remove and replace ...
7436 (label_to_block): ... with this.
7437 (gimple_switch_label_bb): New.
7438 (gimple_switch_default_bb): Likewise.
7439 (gimple_switch_edge): Likewise.
7440 (gimple_switch_default_edge): Likewise.
7441 * tree-cfgcleanup.c (convert_single_case_switch): Use
7442 new gimple functions and pass new argument to label_to_block.
7443 (cleanup_control_flow_bb):
7444 * tree-eh.c (make_eh_dispatch_edges): Use label_to_block and pass
7445 cfun argument explicitly.
7446 (make_eh_edges): Likewise.
7447 (redirect_eh_dispatch_edge): Likewise.
7448 (lower_resx): Likewise.
7449 (lower_eh_dispatch): Likewise.
7450 (maybe_remove_unreachable_handlers): Likewise.
7451 (unsplit_eh): Likewise.
7452 (cleanup_empty_eh): Likewise.
7453 (verify_eh_edges): Likewise.
7454 (verify_eh_dispatch_edge): Likewise.
7455 * tree-ssa-dom.c (record_edge_info): Likewise.
7456 * tree-ssa-forwprop.c (simplify_gimple_switch_label_vec): Likewise.
7457 * tree-ssa-threadedge.c (thread_around_empty_blocks): Likewise.
7458 (thread_through_normal_block): Likewise.
7459 * tree-ssa-uncprop.c (associate_equivalences_with_edges): Likewise.
7460 * tree-ssa-uninit.c (convert_control_dep_chain_into_preds):
7461 * tree-switch-conversion.c (switch_conversion::collect): Use new
7463 (switch_conversion::check_final_bb): Likewise.
7464 (switch_conversion::gather_default_values): Pass new argument
7466 (switch_conversion::build_constructors): Likewise.
7467 (switch_decision_tree::compute_cases_per_edge): Use new
7468 gimple_switch_edge function.
7469 (switch_decision_tree::analyze_switch_statement): Pass new argument
7471 (switch_decision_tree::try_switch_expansion): Use
7472 gimple_switch_default_edge.
7473 * tree-vrp.c (find_switch_asserts): Pass new argument
7475 * vr-values.c (vr_values::vrp_visit_switch_stmt): Likewise.
7476 (vr_values::simplify_switch_using_ranges): Likewise.
7478 2018-08-27 Richard Biener <rguenther@suse.de>
7480 * cfganal.h (rev_post_order_and_mark_dfs_back_seme): Declare.
7481 * cfganal.c (rev_post_order_and_mark_dfs_back_seme): New function.
7483 * tree-ssa-sccvn.h (struct vn_pval): New structure.
7484 (struct vn_nary_op_s): Add unwind_to member. Add
7485 predicated_values flag and put result into a union together
7486 with a linked list of vn_pval.
7487 (struct vn_ssa_aux): Add name member to make maintaining
7488 a map of SSA name to vn_ssa_aux possible. Remove no longer
7489 needed info, dfsnum, low, visited, on_sccstack, use_processed
7490 and range_info_anti_range_p members.
7491 (run_scc_vn, vn_eliminate, free_scc_vn, vn_valueize): Remove.
7492 (do_rpo_vn, run_rpo_vn, eliminate_with_rpo_vn, free_rpo_vn):
7494 (vn_valueize): New global.
7495 (vn_context_bb): Likewise.
7496 (VN_INFO_RANGE_INFO, VN_INFO_ANTI_RANGE_P, VN_INFO_RANGE_TYPE,
7497 VN_INFO_PTR_INFO): Remove.
7498 * tree-ssa-sccvn.c: ... (rewrite)
7499 (pass_fre::execute): For -O2+ initialize loops and run
7500 RPO VN in optimistic mode (iterating). For -O1 and -Og
7501 run RPO VN in non-optimistic mode.
7502 * params.def (PARAM_SCCVN_MAX_SCC_SIZE): Remove.
7503 (PARAM_RPO_VN_MAX_LOOP_DEPTH): Add.
7504 * doc/invoke.texi (sccvn-max-scc-size): Remove.
7505 (rpo-vn-max-loop-depth): Document.
7506 * tree-ssa-alias.c (walk_non_aliased_vuses): Stop walking
7507 when valuezing the VUSE signals we walked out of the region.
7508 * tree-ssa-pre.c (phi_translate_1): Ignore predicated values.
7509 (phi_translate): Set VN context block to use for availability
7511 (compute_avail): Likewise.
7512 (pre_valueize): New function.
7513 (pass_pre::execute): Adjust to the RPO VN API.
7515 * tree-ssa-loop-ivcanon.c: Include tree-ssa-sccvn.h.
7516 (propagate_constants_for_unrolling): Remove.
7517 (tree_unroll_loops_completely): Perform value-numbering
7518 on the unrolled bodies loop parent.
7520 2018-08-27 Richard Biener <rguenther@suse.de>
7522 * tree-ssa-pre.c (compute_antic): Re-use inverted postorder
7523 for partial antic compute.
7525 2018-08-27 Jakub Jelinek <jakub@redhat.com>
7527 PR rtl-optimization/87065
7528 * combine.c (simplify_if_then_else): Formatting fix.
7529 (if_then_else_cond): Guard MULT optimization with SCALAR_INT_MODE_P
7531 (known_cond): Don't return const_true_rtx for vector modes. Use
7532 CONST0_RTX instead of const0_rtx. Formatting fixes.
7534 2018-08-27 Martin Liska <mliska@suse.cz>
7536 PR gcov-profile/87069
7537 * gcov.c (process_file): Record files already processed
7538 and warn about a file being processed multiple times.
7540 2018-08-27 Martin Liska <mliska@suse.cz>
7543 * config/aarch64/aarch64.c (aarch64_override_options_internal):
7544 Set default values for x_aarch64_*_string strings.
7545 * config/aarch64/aarch64.opt: Remove --{march,mcpu,mtune}==
7546 prefix. For -mabi do not print '=ABI' in help and use
7547 <option_value> format for -msve-vector-bits and -moverride
7550 2018-08-26 Jeff Law <law@redhat.com>
7552 * config/mips/frame-header-opt.c: Include "backend.h" rather than
7555 2018-08-26 Marek Polacek <polacek@redhat.com>
7557 PR c++/87029, Implement -Wredundant-move.
7558 * doc/invoke.texi: Document -Wredundant-move.
7560 2018-08-25 Martin Sebor <msebor@redhat.com>
7562 PR tree-optimization/87059
7563 * builtins.c (expand_builtin_strncmp): Convert MIN_EXPR operand
7564 to the same type as the other.
7565 * fold-const.c (fold_binary_loc): Assert expectation.
7567 2018-08-25 Iain Sandoe <iain@sandoe.co.uk>
7569 * config/darwin.c (machopic_legitimize_pic_address): Clean up
7570 extraneous parentheses, dead code section and formatting.
7572 2018-08-24 David Malcolm <dmalcolm@redhat.com>
7575 * diagnostic-show-locus.c (layout::layout): Ensure the margin is
7576 wide enough for jumps in the line-numbering to be visible.
7577 (layout::print_gap_in_line_numbering): New member function.
7578 (layout::calculate_line_spans): When using line numbering, merge
7579 line spans that are only 1 line apart.
7580 (diagnostic_show_locus): When printing line numbers, show gaps in
7581 line numbering directly, rather than printing headers.
7582 (selftest::test_diagnostic_show_locus_fixit_lines): Add test of
7583 line-numbering with multiple line spans.
7584 (selftest::test_fixit_insert_containing_newline_2): Add test of
7585 line-numbering, in which the spans are close enough to be merged.
7587 2018-08-24 Aldy Hernandez <aldyh@redhat.com>
7589 * gimple-ssa-evrp-analyze.c (set_ssa_range_info): Pass value_range
7590 to range_includes_zero_p. Do not special case VR_ANTI_RANGE.
7591 * tree-vrp.c (range_is_nonnull): Remove.
7592 (range_includes_zero_p): Accept value_range instead of min/max.
7593 (extract_range_from_binary_expr_1): Do not early bail on
7595 Use range_includes_zero_p instead of range_is_nonnull.
7596 (extract_range_from_unary_expr): Use range_includes_zero_p instead
7597 of range_is_nonnull.
7598 (vrp_meet_1): Pass value_range to range_includes_zero_p. Do not
7599 special case VR_ANTI_RANGE.
7600 (vrp_finalize): Same.
7601 * tree-vrp.h (range_includes_zero_p): Pass value_range as argument
7603 (range_is_nonnull): Remove.
7604 * vr-values.c (vrp_stmt_computes_nonzero): Use
7605 range_includes_zero_p instead of range_is_nonnull.
7606 (extract_range_basic): Pass value_range to range_includes_zero_p
7607 instead of range_is_nonnull.
7609 2018-08-24 Uros Bizjak <ubizjak@gmail.com>
7611 * emit-rtl.c (init_emit_once): Do not emit MODE_POINTER_BOUNDS RTXes.
7612 * emit-rtl.h (rtl_data): Remove return_bnd.
7613 * explow.c (trunc_int_for_mode): Do not handle POINTER_BOUNDS_MODE_P.
7614 * function.c (diddle_return_value): Do not handle crtl->return_bnd.
7615 * genmodes.c (complete_mode): Do not handle MODE_POINTER_BOUNDS.
7616 (POINTER_BOUNDS_MODE): Remove definition.
7617 (make_pointer_bounds_mode): Remove.
7618 (get_mode_class): Do not handle MODE_POINTER_BOUNDS.
7619 * machmode.h (POINTER_BOUNDS_MODE_P): Remove definition.
7620 (scalare_mode::includes_p): Do not handle MODE_POINTER_BOUNDS.
7621 * mode-classes.def: Do not define MODE_POINTER_BOUNDS.
7622 * stor-layout.c (int_mode_for_mode): Do not handle MODE_POINTER_BOUNDS.
7623 * tree-core.h (enum tree_index): Remove TI_POINTER_BOUNDS_TYPE.
7624 * varasm.c (output_constant_pool_2): Do not handle MODE_POINTER_BOUNDS.
7626 * config/i386/i386-modes.def (BND32, BND64): Remove.
7627 * config/i386/i386.c (dbx_register_map): Remove bound registers.
7628 (dbx64_register_map): Ditto.
7629 (svr4_dbx_register_map): Ditto.
7630 (indirect_thunk_bnd_needed): Remove.
7631 (indirect_thunks_bnd_used): Ditto.
7632 (indirect_return_bnd_needed): Ditto.
7633 (indirect_return_via_cx_bnd): Ditto.
7634 (enum indirect_thunk_prefix): Remove indirect_thunk_prefix_bnd.
7635 (indirect_thunk_name): Remove handling of indirect_thunk_prefix_bnd.
7636 (output_indirect_thunk): Ditto. Remove need_prefix argument.
7637 (output_indirect_thunk_function): Remove handling of
7638 indirect_return_bnd_needed, indirect_return_via_cx_bnd,
7639 indirect_thunk_bnd_needed and indirect_thunks_bnd_used variables.
7640 (ix86_save_reg): Remove handling of crtl->return_bnd.
7641 (ix86_legitimate_constant_p): Remove handling of POINTER_BOUNDS_MODE_P.
7642 (ix86_print_operand_address_as): Remove handling of UNSPEC_BNDMK_ADDR
7643 and UNSPEC_BNDLX_ADDR.
7644 (ix86_output_indirect_branch_via_reg): Remove handling of
7645 indirect_thunk_prefix_bnd.
7646 (ix86_output_indirect_branch_via_push): Ditto.
7647 (ix86_output_function_return): Ditto.
7648 (ix86_output_indirect_function_return): Ditto.
7649 (avoid_func_arg_motion): Do not handle UNSPEC_BNDSTX.
7650 * config/i386/i386.h (FIXED_REGISTERS): Remove bound registers.
7651 (CALL_USED_REGISTERS): Ditto.
7652 (REG_ALLOC_ORDER): Update for removal of bound registers.
7653 (HI_REGISTER_NAMES): Ditto.
7654 * config/i386/i386.md (UNSPEC_BNDMK, UNSPEC_BNDMK_ADDR, UNSPEC_BNDSTX)
7655 (UNSPEC_BNDLDX, UNSPEC_BNDLDX_ADDR, UNSPEC_BNDCL, UNSPEC_BNDCU)
7656 (UNSPEC_BNDCN, UNSPEC_MPX_FENCE): Remove.
7657 (BND0_REG, BND1_REG, BND2_REG, BND3_REG): Remove
7658 (FIRST_PSEUDO_REG): Update.
7659 (BND): Remove mode iterator.
7660 * config/i386/predicates.md (bnd_mem_operator): Remove.
7662 2018-08-24 Richard Sandiford <richard.sandiford@arm.com>
7664 * tree-vect-stmts.c (vectorizable_bswap): Handle variable-length
7667 2018-08-24 Richard Sandiford <richard.sandiford@arm.com>
7669 * tree-vect-slp.c (vect_transform_slp_perm_load): Separate out
7670 the case in which the permute needs only a single element and
7671 repeats for every vector of the result. Extend that case to
7672 handle variable-length vectors.
7673 * tree-vect-stmts.c (vectorizable_load): Update accordingly.
7675 2018-08-24 H.J. Lu <hongjiu.lu@intel.com>
7678 * dwarf2out.c (save_macinfo_strings): Call set_indirect_string
7679 on DW_MACINFO_start_file for -gsplit-dwarf -g3.
7681 2018-08-24 Richard Biener <rguenther@suse.de>
7683 * cfg.h (struct control_flow_graph): Add edge_flags_allocated and
7684 bb_flags_allocated members.
7685 (auto_flag): New RAII class for allocating flags.
7686 (auto_edge_flag): New RAII class for allocating edge flags.
7687 (auto_bb_flag): New RAII class for allocating bb flags.
7688 * cfgloop.c (verify_loop_structure): Allocate temporary edge
7690 * cfganal.c (dfs_enumerate_from): Remove use of visited sbitmap
7691 in favor of temporarily allocated BB flag.
7692 * hsa-brig.c: Re-order includes.
7693 * hsa-dump.c: Likewise.
7694 * hsa-regalloc.c: Likewise.
7695 * print-rtl.c: Likewise.
7696 * profile-count.c: Likewise.
7698 2018-08-24 Segher Boessenkool <segher@kernel.crashing.org>
7701 * config/rs6000/rs6000.c (toc_relative_expr_p): Check that the base is
7704 2018-08-24 Aldy Hernandez <aldyh@redhat.com>
7707 * wide-int-range.cc (wide_int_range_div): Do not ignore result
7708 from wide_int_range_multiplicative_op.
7710 2018-08-23 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
7712 * tree-vect-data-refs.c (vect_grouped_store_supported): Fix typo
7715 2018-08-23 Giuliano Belinassi <giuliano.belinassi@usp.br>
7717 * genmatch.c (parser::parse_operation): Fix typo 'exapnded'
7720 2018-08-23 Alexander Monakov <amonakov@ispras.ru>
7722 * tree-scalar-evolution.c (final_value_replacement_loop): Dump
7723 full GENERIC expression used for replacement.
7725 2018-08-23 Aldy Hernandez <aldyh@redhat.com>
7727 * tree-vrp.c (abs_extent_range): Remove.
7728 (extract_range_into_wide_ints): Pass wide ints by reference.
7729 (extract_range_from_binary_expr_1): Rewrite the *DIV_EXPR code.
7730 Pass wide ints by reference in all calls to
7731 extract_range_into_wide_ints.
7732 * wide-int-range.cc (wide_int_range_div): New.
7733 * wide-int-range.h (wide_int_range_div): New.
7734 (wide_int_range_includes_zero_p): New.
7735 (wide_int_range_zero_p): New.
7737 2018-08-23 Matthew Malcomson <matthew.malcomson@arm.com>
7739 * config/aarch64/aarch64.md (arches): New enum.
7740 (arch): New enum attr.
7741 (arch_enabled): New attr.
7742 (enabled): Now uses arch_enabled only.
7743 (simd, sve, fp16): Removed attribute.
7744 (fp): Attr now defined in terms of 'arch'.
7745 (*mov<mode>_aarch64, *movsi_aarch64, *movdi_aarch64, *movti_aarch64,
7746 *movhf_aarch64, <optab><fcvt_target><GPF:mode>2,
7747 <FCVT_F2FIXED:fcvt_fixed_insn><GPF:mode>3,
7748 <FCVT_FIXED2F:fcvt_fixed_insn><GPI:mode>3): Merge 'fp' and 'simd'
7749 attributes into 'arch'.
7750 (*movsf_aarch64, *movdf_aarch64, *movtf_aarch64, *add<mode>3_aarch64,
7751 subdi3, neg<mode>2, <optab><mode>3, one_cmpl<mode>2,
7752 *<NLOGICAL:optab>_one_cmpl<mode>3, *xor_one_cmpl<mode>3,
7753 *aarch64_ashl_sisd_or_int_<mode>3, *aarch64_lshr_sisd_or_int_<mode>3,
7754 *aarch64_ashr_sisd_or_int_<mode>3, *aarch64_sisd_ushl): Convert use of
7755 'simd' attribute into 'arch'.
7756 (load_pair_sw_<SX:mode><SX2:mode>, load_pair_dw_<DX:mode><DX2:mode>,
7757 store_pair_sw_<SX:mode><SX2:mode>, store_pair_dw_<DX:mode><DX2:mode>):
7758 Convert use of 'fp' attribute to 'arch'.
7759 * config/aarch64/aarch64-simd.md (move_lo_quad_internal_<mode>,
7760 move_lo_quad_internal_<mode>): (different modes) Merge 'fp' and 'simd'
7762 (move_lo_quad_internal_be_<mode>, move_lo_quad_internal_be_<mode>):
7763 (different modes) Merge 'fp' and 'simd' into 'arch'.
7764 (*aarch64_combinez<mode>, *aarch64_combinez_be<mode>): Merge 'fp' and
7767 2018-08-23 Segher Boessenkool <segher@kernel.crashing.org>
7769 PR rtl-optimization/87026
7770 * expmed.c (canonicalize_comparison): If we can no longer create
7771 pseudoregisters, don't.
7773 2018-08-23 Richard Earnshaw <rearnsha@arm.com>
7776 * config/arm/arm-protos.h (arm_emit_speculation_barrier): New
7778 * config/arm/arm.c (speculation_barrier_libfunc): New static
7780 (arm_init_libfuncs): Initialize it.
7781 (arm_emit_speculation_barrier): New function.
7782 * config/arm/arm.md (speculation_barrier): Call
7783 arm_emit_speculation_barrier for architectures that do not have
7785 (speculation_barrier_insn): Only match on Armv7 or later.
7787 2018-08-23 Richard Biener <rguenther@suse.de>
7790 * tree-inline.c (copy_bb): Drop unused __builtin_va_arg_pack_len
7793 2018-08-23 Richard Sandiford <richard.sandiford@arm.com>
7795 * config/aarch64/aarch64.c (aarch64_evpc_sve_tbl): Fix handling
7796 of single-vector TBLs.
7797 (aarch64_vectorize_vec_perm_const): Set one_vector_p when only
7800 2018-08-23 Richard Sandiford <richard.sandiford@arm.com>
7803 * config/aarch64/aarch64.c (aarch64_expand_vec_perm_const_1): Fix
7804 aarch64_evpc_tbl guard.
7806 2018-08-22 Bernd Edlinger <bernd.edlinger@hotmail.de>
7808 * tree-ssa-dse.c (compute_trims): Avoid folding away undefined
7811 2018-08-22 Martin Sebor <msebor@redhat.com>
7814 * tree-pretty-print.c (pretty_print_string): Add argument.
7815 (dump_generic_node): Call to pretty_print_string with string size.
7817 2018-08-22 Segher Boessenkool <segher@kernel.crashing.org>
7819 PR rtl-optimization/86771
7820 * combine.c (try_combine): Do not allow splitting a resulting PARALLEL
7821 of two SETs into those two SETs, one to be placed at i2, if that SETs
7822 destination is modified between i2 and i3.
7824 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
7826 PR tree-optimization/86725
7827 * tree-vect-loop.c (vect_inner_phi_in_double_reduction_p): New
7829 (vect_analyze_scalar_cycles_1): Check it.
7831 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
7833 PR tree-optimization/86725
7834 * tree-vect-loop.c (vect_is_simple_reduction): When treating
7835 an outer loop phi as a double reduction, make sure that the
7836 single user of the phi result is an inner loop phi.
7838 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
7840 * tree-vect-data-refs.c (vect_analyze_group_access_1): Convert
7841 grouped stores with gaps to a strided group.
7843 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
7845 * tree-vect-stmts.c (get_group_load_store_type)
7846 (get_load_store_type): Only test STMT_VINFO_STRIDED_P for the
7847 first statement in a group.
7849 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
7851 * config/darwin.h (LINK_COMMAND_SPEC_A): Sync LTO options with
7852 the sequence used in gcc/gcc.c.
7854 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
7857 * gcc-ar.c (main): Don’t try to invoke the plug-in if we’re not
7860 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
7862 * config/darwin10.h (LINK_GCC_C_SEQUENCE_SPEC): Adjust to use the
7863 Darwin10-specific unwinder-shim.
7864 * config/darwin12.h (LINK_GCC_C_SEQUENCE_SPEC): Remove.
7865 * config/rs6000/darwin.h (DARWIN_CRT1_SPEC, DARWIN_DYLIB1_SPEC):
7866 New to cater for Darwin10 Rosetta.
7868 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
7870 * config/i386/i386.c (ix86_output_addr_diff_elt): Move the MACH-O
7871 specific test before the one for HAVE_AS_GOTOFF_IN_DATA.
7873 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
7878 * gcc/dwarf2out.c (FUNC_SECOND_SECT_LABEL): New.
7879 (dwarf2out_switch_text_section): Generate a local label for the
7880 second function sub-section and apply it as the second FDE start
7882 * gcc/final.c (final_scan_insn_1): Emit second FDE label after the
7883 second sub-section start.
7885 2018-08-22 Richard Biener <rguenther@suse.de>
7887 PR tree-optimization/86988
7888 * tree-vrp.c (vrp_prop::check_mem_ref): Bail out on VLAs.
7890 2018-08-22 Richard Biener <rguenther@suse.de>
7892 PR tree-optimization/86945
7893 * tree-cfg.c (generate_range_test): Use unsigned arithmetic.
7895 2018-08-22 Alexandre Oliva <oliva@adacore.com>
7897 * config/rs6000/rs6000.c (SMALL_DATA_RELOC, SMALL_DATA_REG): Add
7898 a comment about how uses of r2 for .sdata2 come about.
7900 2018-08-22 Alexandre Oliva <aoliva@redhat.com>
7902 * tree-ssa-reassoc.c (is_reassociable_op): Fix cut&pasto.
7904 2018-08-21 Marek Polacek <polacek@redhat.com>
7906 PR c++/86981, Implement -Wpessimizing-move.
7907 * doc/invoke.texi: Document -Wpessimizing-move.
7909 2018-08-21 Jan Hubicka <jh@suse.cz>
7911 * tree.c (find_decls_types_r): Do not check for redundant typedefs.
7912 * tree.h (is_redundant_typedef): Remove.
7913 * dwarf2out.c (is_redundant_typedef): Turn into static function.
7915 2018-08-21 Jan Hubicka <jh@suse.cz>
7917 * tree.c (free_lang_data_in_decl): Remove types from DECL_CONTEXT
7920 2018-08-21 Tamar Christina <tamar.christina@arm.com>
7922 * expmed.c (extract_low_bits): Reject invalid subregs early.
7924 2018-08-21 Bernd Edlinger <bernd.edlinger@hotmail.de>
7927 * tree-ssa-strlen.c (adjust_last_stmt): Avoid folding away undefined
7930 2018-08-21 Rasmus Villemoes <rv@rasmusvillemoes.dk>
7932 * config/vxworks.h: Guard vxworks_asm_out_constructor and
7933 vxworks_asm_out_destructor by !HAVE_INITFINI_ARRAY_SUPPORT
7934 * config/vxworks.c: Likewise.
7936 2018-08-21 Rasmus Villemoes <rv@rasmusvillemoes.dk>
7938 * config/vxworks.c: Set targetm.have_ctors_dtors
7939 if HAVE_INITFINI_ARRAY_SUPPORT.
7940 * config/vxworks.h: Set SUPPORTS_INIT_PRIORITY
7941 if HAVE_INITFINI_ARRAY_SUPPORT.
7943 2018-08-21 Rasmus Villemoes <rv@rasmusvillemoes.dk>
7945 * config/vxworks.h: Add $(WIND_BASE)/target/h/wrn/coreip to
7946 default search path for VxWorks < 7.
7948 2018-08-21 Bernd Edlinger <bernd.edlinger@hotmail.de>
7950 * gimple-ssa-sprintf.c (decl_constant_value): Remove.
7951 (get_format_string): Refer to c_getstr.
7953 2018-08-21 Tom de Vries <tdevries@suse.de>
7955 * cgraph.h (debuginfo_early_init, debuginfo_init, debuginfo_fini)
7956 (debuginfo_start, debuginfo_stop, debuginfo_early_start)
7957 (debuginfo_early_stop): Declare.
7958 * cgraphunit.c (debuginfo_early_init, debuginfo_init, debuginfo_fini)
7959 (debuginfo_start, debuginfo_stop, debuginfo_early_start)
7960 (debuginfo_early_stop): New function.
7961 (symbol_table::finalize_compilation_unit): Call debuginfo_early_start
7962 and debuginfo_early_stop.
7963 * dwarf2out.c (dwarf2out_finish, dwarf2out_early_finish): Dump dwarf.
7964 * toplev.c (compile_file): Call debuginfo_start and debuginfo_stop.
7965 (general_init): Call debuginfo_early_init.
7966 (finalize): Call debuginfo_fini.
7967 (do_compile): Call debuginfo_init.
7968 * doc/invoke.texi (@gccoptlist): Add -fdump-debug and
7970 (@item -fdump-debug, @item -fdump-earlydebug): Add.
7972 2018-08-21 Tom de Vries <tdevries@suse.de>
7974 * dwarf2out.c (print_dw_val, print_loc_descr, print_die): Handle
7975 flag_dump_noaddr and flag_dump_unnumbered.
7977 2018-08-21 Aldy Hernandez <aldyh@redhat.com>
7979 * wide-int-range.cc (wide_int_range_abs): New.
7980 (wide_int_range_order_set): Rename from wide_int_range_min_max.
7981 * wide-int-range.h (wide_int_range_abs): New.
7982 (wide_int_range_min_max): New.
7983 * tree-vrp.c (extract_range_from_unary_expr): Rewrite ABS_EXPR
7984 case to call wide_int_range_abs.
7985 Rewrite MIN/MAX_EXPR to call wide_int_range_min_max.
7986 (extract_range_from_abs_expr): Delete.
7988 2018-08-20 Michael Meissner <meissner@linux.ibm.com>
7991 * config/rs6000/rs6000.md (extendsi<mode>2): Change constraints
7992 from 'Y' to 'YZ' to enable the LWAX instruction to be generated
7995 2018-08-20 Nathan Sidwell <nathan@acm.org>
7996 Jeff Law <law@redhat.com>
7998 * config/s390/s390-c (s390_macro_to_expand): Use cpp_macro_p.
7999 * config/spu/spu-c.c (spu_macro_to_expand): Likewise.
8001 2018-08-20 David Malcolm <dmalcolm@redhat.com>
8004 * attribs.c (diag_attr_exclusions): Add auto_diagnostic_group instance.
8005 (decl_attributes): Likewise.
8006 * calls.c (maybe_warn_nonstring_arg): Add auto_diagnostic_group
8008 * cgraphunit.c (maybe_diag_incompatible_alias): Likewise.
8009 * diagnostic-core.h (class auto_diagnostic_group): New class.
8010 * diagnostic.c (diagnostic_initialize): Initialize the new fields.
8011 (diagnostic_report_diagnostic): Handle the first diagnostics within
8013 (emit_diagnostic): Add auto_diagnostic_group instance.
8015 (inform_n): Likewise.
8016 (warning): Likewise.
8017 (warning_at): Likewise.
8018 (warning_n): Likewise.
8019 (pedwarn): Likewise.
8020 (permerror): Likewise.
8022 (error_n): Likewise.
8023 (error_at): Likewise.
8025 (fatal_error): Likewise.
8026 (internal_error): Likewise.
8027 (internal_error_no_backtrace): Likewise.
8028 (auto_diagnostic_group::auto_diagnostic_group): New ctor.
8029 (auto_diagnostic_group::~auto_diagnostic_group): New dtor.
8030 * diagnostic.h (struct diagnostic_context): Add fields
8031 "diagnostic_group_nesting_depth",
8032 "diagnostic_group_emission_count", "begin_group_cb",
8034 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior):
8035 Add auto_diagnostic_group instance(s).
8036 (find_explicit_erroneous_behavior): Likewise.
8037 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Likewise.
8038 * gimple-ssa-warn-restrict.c (maybe_diag_offset_bounds): Likewise.
8039 * gimplify.c (warn_implicit_fallthrough_r): Likewise.
8040 (gimplify_va_arg_expr): Likewise.
8041 * hsa-gen.c (HSA_SORRY_ATV): Likewise.
8042 (HSA_SORRY_AT): Likewise.
8043 * ipa-devirt.c (compare_virtual_tables): Likewise.
8044 (warn_odr): Likewise.
8045 * multiple_target.c (expand_target_clones): Likewise.
8046 * opts-common.c (cmdline_handle_error): Likewise.
8047 * reginfo.c (globalize_reg): Likewise.
8048 * substring-locations.c (format_warning_n_va): Likewise.
8049 * tree-inline.c (expand_call_inline): Likewise.
8050 * tree-ssa-ccp.c (pass_post_ipa_warn::execute): Likewise.
8051 * tree-ssa-loop-niter.c
8052 (do_warn_aggressive_loop_optimizations): Likewise.
8053 * tree-ssa-uninit.c (warn_uninit): Likewise.
8054 * tree.c (warn_deprecated_use): Likewise.
8056 2018-08-20 H.J. Lu <hongjiu.lu@intel.com>
8059 * config/i386/i386.md (eh_return): Always update EH return
8060 address in word_mode.
8062 2018-08-20 Chung-Lin Tang <cltang@codesourcery.com>
8064 * targhooks.c (std_gimplify_va_arg_expr): Properly handle case of when
8065 TARGET_SPLIT_COMPLEX_ARG is defined.
8067 2018-08-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
8069 * expr.c (store_field): Change gcc_assert to gcc_checking_assert.
8071 2018-08-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
8074 * expr.c (expand_assignment): Assert that bitpos is positive.
8075 (store_field): Likewise
8076 (expand_expr_real_1): Make sure that bitpos is positive.
8077 * config/alpha/alpha.h (CONSTANT_ADDRESS_P): Avoid signed
8080 2018-08-20 Nathan Sidwell <nathan@acm.org>
8082 * Makefile.in (CPP_ID_DATA_H): Delete.
8083 (CPP_INTERNAL_H): Don't add it.
8084 (GTFILES): Replace CPP_ID_DATA_H with CPPLIB_H.
8085 * gengtype.c (open_base_files): Replace cpp-id-data.h with cpplib.h
8087 2018-08-20 Richard Biener <rguenther@suse.de>
8089 PR tree-optimization/78655
8090 * tree-vrp.c (extract_range_from_binary_expr_1): Make
8091 pointer + offset nonnull if either operand is nonnull work.
8093 2018-08-20 Tom de Vries <tdevries@suse.de>
8095 * dwarf2out.c (add_scalar_info): Don't add reference to existing die
8096 unless the referenced die describes the added property using
8097 DW_AT_location or DW_AT_const_value. Fall back to exprloc case.
8098 Otherwise, add a DW_AT_location to the referenced die.
8100 2018-08-19 Uros Bizjak <ubizjak@gmail.com>
8103 * config/i386/i386.c (ix86_rtx_costs) [case SET]: Check source for
8104 register_operand when calling ix86_set_reg_reg_cost.
8105 [case CONST_INT, case CONST, case LABEL_REF, case SYMBOL_REF]:
8106 Set *total to 0 for operands that satisfy x86_64_immediate_operand
8107 predicate and to 1 otherwise.
8109 2018-08-18 Iain Sandoe <iain@sandoe.co.uk>
8111 * config/darwin.c (darwin_override_options): If -gsplit-dwarf is set,
8112 emit a diagnostic that it is not supported and reset the option.
8113 * config/darwin.h (DRIVER_SELF_SPECS): Note that gsplit-dwarf is not
8114 supported and consume the option. (ASM_FINAL_SPEC): New.
8116 2018-08-17 Segher Boessenkool <segher@kernel.crashing.org>
8118 * doc/md.texi (Patterns): Use @ref instead of @xref in the middle of
8121 2018-08-17 Sandra Loosemore <sandra@codesourcery.com>
8123 C-SKY port: Documentation
8125 * doc/extend.texi (C-SKY Function Attributes): New section.
8126 * doc/invoke.texi (Option Summary): Add C-SKY options.
8127 (C-SKY Options): New section.
8128 * doc/md.texi (Machine Constraints): Document C-SKY constraints.
8130 2018-08-17 Jojo <jijie_rong@c-sky.com>
8131 Huibin Wang <huibin_wang@c-sky.com>
8132 Sandra Loosemore <sandra@codesourcery.com>
8133 Chung-Lin Tang <cltang@codesourcery.com>
8135 C-SKY port: Backend implementation
8137 * config/csky/*: New.
8138 * common/config/csky/*: New.
8140 2018-08-17 Jojo <jijie_rong@c-sky.com>
8141 Huibin Wang <huibin_wang@c-sky.com>
8142 Sandra Loosemore <sandra@codesourcery.com>
8143 Chung-Lin Tang <cltang@codesourcery.com>
8144 Andrew Jenner <andrew@codesourcery.com>
8146 C-SKY port: Configury
8148 * config.gcc (csky-*-*): New.
8149 * configure.ac: Add csky to targets for dwarf2 debug_line support.
8150 * configure: Regenerated.
8152 2018-08-17 David Malcolm <dmalcolm@redhat.com>
8154 * dump-context.h: Include "dumpfile.h".
8155 (dump_context::dump_printf_va): Convert final param from va_list
8156 to va_list *. Convert from ATTRIBUTE_PRINTF to
8157 ATTRIBUTE_GCC_DUMP_PRINTF.
8158 (dump_context::dump_printf_loc_va): Likewise.
8159 * dumpfile.c: Include "stringpool.h".
8160 (make_item_for_dump_printf_va): Delete.
8161 (make_item_for_dump_printf): Delete.
8162 (class dump_pretty_printer): New class.
8163 (dump_pretty_printer::dump_pretty_printer): New ctor.
8164 (dump_pretty_printer::emit_items): New member function.
8165 (dump_pretty_printer::emit_any_pending_textual_chunks): New member
8167 (dump_pretty_printer::emit_item): New member function.
8168 (dump_pretty_printer::stash_item): New member function.
8169 (dump_pretty_printer::format_decoder_cb): New member function.
8170 (dump_pretty_printer::decode_format): New member function.
8171 (dump_context::dump_printf_va): Reimplement in terms of
8172 dump_pretty_printer.
8173 (dump_context::dump_printf_loc_va): Convert final param from va_list
8175 (dump_context::begin_scope): Reimplement call to
8176 make_item_for_dump_printf.
8177 (dump_printf): Update for change to dump_printf_va.
8178 (dump_printf_loc): Likewise.
8179 (selftest::test_capture_of_dump_calls): Convert "stmt" from
8180 greturn * to gimple *. Add a test_decl. Add tests of dump_printf
8181 with %T, %E, and %G.
8182 * dumpfile.h (ATTRIBUTE_GCC_DUMP_PRINTF): New macro.
8183 (dump_printf): Replace ATTRIBUTE_PRINTF_2 with
8184 ATTRIBUTE_GCC_DUMP_PRINTF (2, 3).
8185 (dump_printf_loc): Replace ATTRIBUTE_PRINTF_3 with
8186 ATTRIBUTE_GCC_DUMP_PRINTF (3, 0).
8187 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Convert
8188 use of HOST_WIDE_INT_PRINT_DEC on unsigned HOST_WIDE_INT "count"
8189 within a dump_printf_loc call to "%wu".
8190 (vector_alignment_reachable_p): Merge two dump_printf[_loc] calls,
8191 converting a use of HOST_WIDE_INT_PRINT_DEC to "%wd". Add a
8192 missing space after "=".
8193 * tree-vect-loop.c (vect_analyze_loop_2) Within a dump_printf
8194 call, convert use of HOST_WIDE_INT_PRINT_DEC to "%wd".
8195 * tree-vect-slp.c (vect_slp_bb): Within a dump_printf_loc call,
8196 convert use of HOST_WIDE_INT_PRINT_UNSIGNED to "%wu".
8197 * tree-vectorizer.c (try_vectorize_loop_1): Likewise. Remove
8198 duplicate "vectorized" from message.
8200 2018-08-17 Szabolcs Nagy <szabolcs.nagy@arm.com>
8202 * config/arm/arm-builtins.c (arm_init_simd_builtin_types): Clear
8203 polyNxK_t element's TYPE_STRING_FLAG.
8205 2018-08-17 Segher Boessenkool <segher@kernel.crashing.org>
8207 * config/rs6000/rs6000.md (*cbranch, *creturn): Name these patterns
8208 (they were unnamed before). Fix comments.
8210 2018-08-17 Nathan Sidwell <nathan@acm.org>
8212 * cppbuiltin.c: Include "cpplib.h", not "cpp-id-data.h".
8214 2018-08-17 Richard Biener <rguenther@suse.de>
8216 PR tree-optimization/86841
8217 * wide-int-range.cc (wide_int_range_lshift): Use to_uhwi.
8219 2018-08-17 Martin Liska <mliska@suse.cz>
8221 * common.opt: Remove Warn, Init and Report for options with
8222 Ignore/Deprecated flag. Warning is done automatically for
8224 * config/i386/i386.opt: Likewise.
8225 * config/ia64/ia64.opt: Likewise.
8226 * config/rs6000/rs6000.opt: Likewise.
8227 * cppbuiltin.c (define_builtin_macros_for_compilation_flags):
8228 Remove usage of flag_check_pointer_bounds.
8229 * lto-wrapper.c (merge_and_complain): Do not handle
8230 OPT_fcheck_pointer_bounds.
8231 (append_compiler_options): Likewise.
8232 * opt-functions.awk: Do not handle Deprecated.
8233 * optc-gen.awk: Check that Var, Report and Init are not
8234 used for an option with Ignore/Deprecated flag.
8235 * opts-common.c (decode_cmdline_option): Do not report
8237 (read_cmdline_option): Report warning for OPT_SPECIAL_deprecated
8239 * opts.h (struct cl_option): Remove cl_deprecated flag.
8240 (CL_ERR_DEPRECATED): Remove error enum value.
8242 2018-08-17 Richard Biener <rguenther@suse.de>
8245 * tree-inline.c (copy_bb): When inlining __builtin_va_arg_pack_len ()
8246 across a va-arg-pack using call adjust its return value accordingly.
8248 2018-08-16 Martin Sebor <msebor@redhat.com>
8250 PR tree-optimization/86853
8251 * gimple-ssa-sprintf.c (struct format_result): Rename member.
8252 (struct fmtresult): Add member and initialize it in ctors.
8253 (format_character): Handle %C. Extend range to NUL. Set MAYFAIL.
8254 (format_string): Handle %S the same as %ls. Set MAYFAIL.
8255 (format_directive): Set POSUNDER4K when MAYFAIL is set.
8256 (parse_directive): Handle %C same as %c.
8257 (sprintf_dom_walker::compute_format_length): Adjust.
8258 (is_call_safe): Adjust.
8260 2018-08-16 Bernd Edlinger <bernd.edlinger@hotmail.de>
8262 * builtins.c (c_strlen): Add new parameter eltsize. Use it
8263 for determining how to count the elements.
8264 * builtins.h (c_strlen): Adjust prototype.
8265 * expr.c (string_constant): Add new parameter mem_size.
8266 Set *mem_size appropriately.
8267 * expr.h (string_constant): Adjust protoype.
8268 * gimple-fold.c (get_range_strlen): Add new parameter eltsize.
8269 * gimple-fold.h (get_range_strlen): Adjust prototype.
8270 * gimple-ssa-sprintf.c (get_string_length): Add new parameter eltsize.
8271 (format_string): Call get_string_length with eltsize.
8273 2018-08-16 David Malcolm <dmalcolm@redhat.com>
8275 * diagnostic.c (default_diagnostic_start_span_fn): Call pp_string
8276 to emit the span, rather than setting it as the prefix.
8278 2018-08-16 David Malcolm <dmalcolm@redhat.com>
8280 * diagnostic-show-locus.c (layout::start_annotation_line): Add
8281 "margin_char" parameter, defaulting to space. Use it in place
8282 of pp_space for the initial part of the margin.
8283 (layout::print_leading_fixits): Use '+' when filling the margin
8284 of line-insertion fix-it hints.
8286 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
8288 * config/rs6000/rs6000.md (two unnamed define_insn and define_split):
8291 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
8293 * config/rs6000/altivec.md: Don't set length attribute to the default
8295 * config/rs6000/darwin.md: Ditto.
8296 * config/rs6000/dfp.md: Ditto.
8297 * config/rs6000/htm.md: Ditto.
8298 * config/rs6000/rs6000.md: Ditto.
8299 * config/rs6000/sync.md: Ditto.
8300 * config/rs6000/vsx.md: Ditto.
8302 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
8304 * config/rs6000/altivec.md: Don't set length attribute to the default
8305 value, for branch instructions.
8306 * config/rs6000/darwin.md: Ditto.
8307 * config/rs6000/rs6000.md: Ditto.
8309 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
8311 * config/rs6000/rs6000.md (length): Always define as const_int 4.
8312 (unnamed conditional branch define_insn): Set length to 4 or 8
8313 depending on offset.
8314 (<bd>_<mode>): Similar, for alternative 0.
8315 (<bd>tf_<mode>): Ditto.
8317 2018-08-16 Tamar Christina <tamar.christina@arm.com>
8319 * expr.c (copy_blkmode_to_reg): Perform larger copies when safe.
8321 2018-08-16 Matthew Malcomson <matthew.malcomson@arm.com>
8323 * doc/rtl.texi: Replace old RTX class names with new names.
8326 2018-08-16 Vlad Lazar <vlad.lazar@arm.com>
8328 * expmed.h (canonicalize_comparison): New declaration.
8329 * expmed.c (canonicalize_comparison, equivalent_cmp_code): New function.
8330 * expmed.c (emit_store_flag_1): Add call to canonicalize_comparison.
8331 * optabs.c (prepare_cmp_insn): Likewise.
8332 * rtl.h (unsigned_condition_p): New function which checks if a
8333 comparison operator is unsigned.
8335 2018-08-16 Nathan Sidwell <nathan@acm.org>
8337 * config/rs6000/rs6000-c.c (rs6000_macro_to_expend): Use cpp_macro_p.
8338 * config/powerpcspc/powerpcspe-c.c (rs6000_macro_to_expend): Likewise.
8340 2018-08-16 Tamar Christina <tamar.christina@arm.com>
8343 * config/arm/arm.c (arm_can_change_mode_class): Disallow subreg.
8344 * config/arm/neon.md (movv4hf, movv8hf): Refactored to..
8345 (mov<mov>): ..this and enable unconditionally.
8347 2018-08-16 Tamar Christina <tamar.christina@arm.com>
8349 * config/arm/neon.md (*neon_mov<mode>): Remove reg-to-reg alternative.
8351 2018-08-16 Sam Tebbs <sam.tebbs@arm.com>
8353 * config/aarch64/aarch64.opt (mlow-precision-recip-sqrt)
8354 (mlow-precision-sqrt, mlow-precision-div, mverbose-cost-dump): Replace
8355 "Common" with "Target".
8357 2018-08-15 Uros Bizjak <ubizjak@gmail.com>
8359 * config/i386/i386.opt (mmitigate-rop): Mark as deprecated.
8360 * doc/invoke.texi (mmitigate-rop): Remove.
8361 * config/i386/i386.c: Do not include "regrename.h".
8362 (ix86_rop_should_change_byte_p, reg_encoded_number)
8363 (ix86_get_modrm_for_rop, set_rop_modrm_reg_bits, ix86_mitigate_rop):
8365 (ix86_reorg): Remove call to ix86_mitigate_rop.
8366 * config/i386/i386.md (attr "modrm_class"): Remove.
8367 (cmp<mode>_ccno_1, mov<mode>_xor, movstrict<mode>_xor)
8368 (x86_mov<mode>cc_0_m1. x86_mov<mode>cc_0_m1_se)
8369 (x86_mov<mode>cc_0_m1_neg): Remove modrm_class attribute override.
8371 2018-08-15 Will Schmidt <will_schmidt@vnet.ibm.com>
8373 * config/rs6000/rs600.c (rs6000_gimple_fold_builtin): Add entries to
8374 allow folding of mergeh() and mergel() for the float and double types.
8375 (fold_mergehl_helper): Rework to handle building a permute tree
8378 2018-08-15 Uros Bizjak <ubizjak@gmail.com>
8380 * config/i386/i386.c (expand_vec_perm_movs): Enable V4SFmode
8383 2018-08-15 David Malcolm <dmalcolm@redhat.com>
8385 * common.opt (fdiagnostics-show-labels): New option.
8386 * diagnostic-show-locus.c (class layout_range): Add field
8388 (class layout): Add field "m_show_labels_p".
8389 (layout_range::layout_range): Add param "label" and use it to
8391 (make_range): Pass in NULL for new "label" param of layout_range's
8393 (layout::layout): Initialize m_show_labels_p.
8394 (layout::maybe_add_location_range): Pass in loc_range->m_label
8395 when constructing layout_range instances.
8396 (struct line_label): New struct.
8397 (layout::print_any_labels): New member function.
8398 (layout::print_line): Call it if label-printing is enabled.
8399 (selftest::test_one_liner_labels): New test.
8400 (selftest::test_diagnostic_show_locus_one_liner): Call it.
8401 * diagnostic.c (diagnostic_initialize): Initialize
8402 context->show_labels_p.
8403 * diagnostic.h (struct diagnostic_context): Add field
8405 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
8406 -fno-diagnostics-show-labels.
8407 * dwarf2out.c (gen_producer_string): Add
8408 OPT_fdiagnostics_show_labels to the ignored options.
8409 * gcc-rich-location.c (gcc_rich_location::add_expr): Add "label"
8411 (gcc_rich_location::maybe_add_expr): Likewise.
8412 * gcc-rich-location.h (gcc_rich_location::gcc_rich_location): Add
8413 label" param, defaulting to NULL.
8414 (gcc_rich_location::add_expr): Add "label" param.
8415 (gcc_rich_location::maybe_add_expr): Likewise.
8416 (class text_range_label): New class.
8417 (class range_label_for_type_mismatch): New class.
8418 * gimple-ssa-sprintf.c (fmtwarn): Pass NULL for new label params
8419 of format_warning_va.
8420 (fmtwarn_n): Likewise for new params of format_warning_n_va.
8421 * lto-wrapper.c (merge_and_complain): Add
8422 OPT_fdiagnostics_show_labels to the "pick one setting" options.
8423 (append_compiler_options): Likewise to the dropped options.
8424 (append_diag_options): Likewise to the passed-on options.
8425 * opts.c (common_handle_option): Handle the new option.
8426 * selftest-diagnostic.c
8427 (test_diagnostic_context::test_diagnostic_context): Enable
8429 * substring-locations.c: Include "gcc-rich-location.h".
8430 (format_warning_n_va): Add "fmt_label" and "param_label" params
8431 and use them as appropriate.
8432 (format_warning_va): Add "fmt_label" and "param_label" params,
8433 passing them on to format_warning_n_va.
8434 (format_warning_at_substring): Likewise.
8435 (format_warning_at_substring_n): Likewise.
8436 * substring-locations.h (format_warning_va): Add "fmt_label" and
8437 "param_label" params.
8438 (format_warning_n_va): Likewise.
8439 (format_warning_at_substring): Likewise.
8440 (format_warning_at_substring_n): Likewise.
8441 * toplev.c (general_init): Initialize global_dc->show_labels_p.
8443 2018-08-15 Qing Zhao <qing.zhao@oracle.com>
8446 * builtins.c (expand_builtin_memcmp): Do not expand the call
8447 when overflow is detected.
8449 2018-08-15 Martin Sebor <msebor@redhat.com>
8451 PR tree-optimization/71625
8452 * config/aarch64/aarch64-builtins.c
8453 (aarch64_init_simd_builtin_types): Clear Poly8_t's TYPE_STRING_FLAG.
8455 2018-08-15 Ilya Leoshkevich <iii@linux.ibm.com>
8457 * config/s390/s390.c (s390_reorg): Remove loop.
8459 2018-08-15 Iain Sandoe <iain@sandoe.co.uk>
8462 (darwin_function_switched_text_sections): Delete.
8463 * gcc/config/darwin.h
8464 (TARGET_ASM_FUNCTION_SWITCHED_TEXT_SECTIONS): Likewise.
8466 2018-08-15 Iain Sandoe <iain@sandoe.co.uk>
8469 * config/darwin.h: (DEBUG_STR_OFFSETS_SECTION, DEBUG_LOCLISTS_SECTION,
8470 DEBUG_RNGLISTS_SECTION) new macros. (DEBUG_PUBNAMES_SECTION,
8471 DEBUG_PUBTYPES_SECTION) update to include GNU variant.
8473 2018-08-15 Martin Liska <mliska@suse.cz>
8475 PR tree-optimization/86925
8476 * predict.c (expr_expected_value_1): When taking
8477 later predictor, assign also probability.
8478 Use fold_build2_initializer_loc in order to fold
8479 the expression in -frounding-math.
8481 2018-08-14 Allan Sandfeld Jensen <allan.jensen@qt.io>
8483 * config/i386/i386.c (expand_vec_perm_movs): New method matching movs
8485 (expand_vec_perm_1): Try the new method.
8487 2018-08-14 Ilya Leoshkevich <iii@linux.ibm.com>
8490 * lra-lives.c (remove_some_program_points_and_update_live_ranges):
8491 Check whether lra_live_max_point is 0 before dividing.
8493 2018-08-14 Martin Sebor <msebor@redhat.com>
8495 PR tree-optimization/86650
8496 * tree-vrp.c (vrp_prop::check_array_ref): Print an inform message.
8497 (vrp_prop::check_mem_ref): Same.
8499 2018-08-13 Liu Hao <lh_mouse@126.com>
8501 * pretty-print.c (eat_esc_sequence): Swap the foreground and
8502 background colors if the COMMON_LVB_REVERSE_VIDEO flag is set,
8503 and clear it thereafter, as it only works for DBCS.
8505 2018-08-13 Liu Hao <lh_mouse@126.com>
8507 * pretty-print.c (mingw_ansi_fputs): Do not call _close() on the
8508 handle returned by _get_osf_handle().
8510 2018-08-13 Will Schmidt <will_schmidt@vnet.ibm.com>
8512 * gcc/config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support
8513 for folding vec_perm.
8515 2018-08-13 Will Schmidt <will_schmidt@vnet.ibm.com>
8517 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin):
8518 Add support for gimple-folding of vec_pack() and vec_unpack()
8521 2018-08-13 Will Schmidt <will_schmidt@vnet.ibm.com>
8523 * config/rs6000/rs6000.c (rs6000_builtin_valid_without_lhs): Add
8524 vec_xst variants to the list.
8525 (rs6000_gimple_fold_builtin): Add support for folding unaligned
8526 vector loads and stores.
8528 2018-08-13 David Edelsohn <dje.gcc@gmail.com>
8530 * config.gcc (rs6000-ibm-aix4.x): Delete.
8531 (rs6000-ibm-aix5.1): Delete.
8532 (rs6000-ibm-aix5.2): Delete.
8533 (rs6000-ibm-aix5.3): Delete.
8534 * config/rs6000/aix43.h: Delete.
8535 * config/rs6000/aix51.h: Delete.
8536 * config/rs6000/aix52.h: Delete.
8537 * config/rs6000/t-aix43: Delete.
8539 2018-08-13 Ilya Leoshkevich <iii@linux.ibm.com>
8541 * config/s390/s390.c (s390_decompose_constant_pool_ref):
8543 (s390_decompose_address): Factor out constant pool ref
8546 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
8548 * config/nds32/nds32-predicates.c
8549 (nds32_can_use_bclr_p): Change return type as bool.
8550 (nds32_can_use_bset_p): Ditto.
8551 (nds32_can_use_btgl_p): Ditto.
8552 (nds32_can_use_bitci_p): Ditto.
8553 * config/nds32/nds32-protos.h
8554 (nds32_can_use_bclr_p): Change declaration.
8555 (nds32_can_use_bset_p): Ditto.
8556 (nds32_can_use_btgl_p): Ditto.
8557 (nds32_can_use_bitci_p): Ditto.
8559 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
8561 * config/nds32/nds32.c (nds32_expand_prologue, nds32_expand_epilogue):
8562 Support -msched-prolog-epilog option.
8563 * config/nds32/nds32.opt (msched-prolog-epilog): New option.
8565 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
8567 * common/config/nds32/nds32-common.c
8568 (nds32_option_optimization_table): Enalbe -malways-align.
8570 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
8572 * config.gcc (nds32*): Add nds32_isr.h and nds32_init.inc in
8574 * common/config/nds32/nds32-common.c (nds32_handle_option): Handle
8575 OPT_misr_secure_ case.
8576 * config/nds32/nds32-isr.c: Implementation of backward compatibility.
8577 * config/nds32/nds32-protos.h (nds32_isr_function_critical_p): New.
8578 * config/nds32/nds32.c (nds32_attribute_table): Add critical and
8580 * config/nds32/nds32.h (nds32_isr_nested_type): Add NDS32_CRITICAL.
8581 (nds32_isr_info): New field security_level.
8582 (TARGET_ISR_VECTOR_SIZE_4_BYTE): New macro.
8583 * config/nds32/nds32.md (return_internal): Consider critical attribute.
8584 * config/nds32/nds32.opt (misr-secure): New option.
8585 * config/nds32/nds32_init.inc: New file.
8586 * config/nds32/nds32_isr.h: New file.
8588 2018-08-11 John David Anglin <danglin@gcc.gnu.org>
8590 * config/pa/pa.md (UNSPEC_MEMORY_BARRIER): New unspec enum.
8591 Update comment for atomic instructions.
8592 (atomic_storeqi, atomic_storehi, atomic_storesi, atomic_storesf,
8593 atomic_loaddf, atomic_loaddf_1, atomic_storedf, atomic_storedf_1):
8595 (atomic_loaddi): Revise fence expansion to only emit fence prior to
8596 load for __ATOMIC_SEQ_CST model.
8597 (atomic_loaddi_1): Remove float register target.
8598 (atomic_storedi): Handle CONST_INT values.
8599 (atomic_storedi_1): Remove float register source. Add special case
8601 (memory_barrier): New expander and insn.
8603 2018-08-11 Jakub Jelinek <jakub@redhat.com>
8605 PR tree-optimization/86835
8606 * tree-ssa-math-opts.c (insert_reciprocals): Even when inserting
8607 new_stmt after def_gsi, make sure to insert new_square_stmt after
8608 that stmt, not 2 stmts before it.
8610 2018-08-10 Alexander Monakov <amonakov@ispras.ru>
8613 * config/i386/i386.md (<s>mul<mode>3_highpart): Use DWIH mode iterator
8616 2018-08-10 Martin Liska <mliska@suse.cz>
8619 * builtin-types.def (BT_FN_LONG_LONG_LONG_DOUBLE): Add new
8621 * builtins.c (expand_builtin_expect_with_probability):
8623 (expand_builtin_expect_with_probability): New function.
8624 (build_builtin_expect_predicate): Add new argumnet probability
8625 for BUILT_IN_EXPECT_WITH_PROBABILITY.
8626 (fold_builtin_expect):
8629 * builtins.def (BUILT_IN_EXPECT_WITH_PROBABILITY):
8630 * builtins.h (fold_builtin_expect): Set new argument.
8631 * doc/extend.texi: Document __builtin_expect_with_probability.
8632 * doc/invoke.texi: Likewise.
8633 * gimple-fold.c (gimple_fold_call): Pass new argument.
8634 * ipa-fnsummary.c (find_foldable_builtin_expect): Handle
8635 also BUILT_IN_EXPECT_WITH_PROBABILITY.
8636 * predict.c (get_predictor_value): New function.
8637 (expr_expected_value): Add new argument probability. Assume
8638 that predictor and probability are always non-null.
8639 (expr_expected_value_1): Likewise. For __builtin_expect and
8640 __builtin_expect_with_probability set probability. Handle
8641 combination in binary expressions.
8642 (tree_predict_by_opcode): Simplify code by simply calling
8643 get_predictor_value.
8644 (pass_strip_predict_hints::execute): Add handling of
8645 BUILT_IN_EXPECT_WITH_PROBABILITY.
8646 * predict.def (PRED_BUILTIN_EXPECT_WITH_PROBABILITY): Add
8648 * tree.h (DECL_BUILT_IN_P): New function.
8650 2018-08-10 Martin Liska <mliska@suse.cz>
8652 PR tree-optimization/85799
8653 * passes.def: Add argument for pass_strip_predict_hints.
8654 * predict.c (class pass_strip_predict_hints): Add new argument
8656 (strip_predictor_early): New function.
8657 (pass_strip_predict_hints::execute): Call the function to
8659 (strip_predict_hints): New function.
8660 * predict.def: Fix comment.
8662 2018-08-10 Thomas Preud'homme <thomas.preudhomme@linaro.org>
8664 * Makefile.in: Clarify which tm.texi to copy over to assert the
8665 right to grant a GFDL license for all.
8667 2018-08-09 Jeff Law <law@redhat.com>
8669 * config/m68k/m68k.c (m68k_adjust_decorated_operand): Remove
8672 2018-08-09 Andreas Schwab <schwab@linux-m68k.org>
8674 * config/m68k/m68k-protos.h (m68k_final_prescan_insn): Remove
8677 2018-08-09 Richard Sandiford <richard.sandiford@arm.com>
8679 * tree-vect-loop.c (vectorizable_reduction): Allow inner-loop
8680 reductions for variable-length vectors.
8682 2018-08-09 David Malcolm <dmalcolm@redhat.com>
8685 * common.opt (fdiagnostics-show-line-numbers): New option.
8686 * diagnostic-show-locus.c (class layout): Add fields
8687 "m_show_line_numbers_p" and "m_linenum_width";
8688 (num_digits): New function.
8689 (test_num_digits): New function.
8690 (layout::layout): Initialize new fields. Update m_x_offset
8691 logic to handle any left margin.
8692 (layout::print_source_line): Print line number when requested.
8693 (layout::start_annotation_line): New member function.
8694 (layout::print_annotation_line): Call it.
8695 (layout::print_leading_fixits): Likewise.
8696 (layout::print_trailing_fixits): Likewise. Update calls to
8697 move_to_column for new parameter.
8698 (layout::get_x_bound_for_row): Add "add_left_margin" param and use
8699 it to potentially call start_annotation_line.
8700 (layout::show_ruler): Call start_annotation_line.
8701 (selftest::test_line_numbers_multiline_range): New selftest.
8702 (selftest::diagnostic_show_locus_c_tests): Call test_num_digits
8703 and selftest::test_line_numbers_multiline_range.
8704 * diagnostic.c (diagnostic_initialize): Initialize
8705 show_line_numbers_p.
8706 * diagnostic.h (struct diagnostic_context): Add field
8707 "show_line_numbers_p".
8708 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
8709 -fno-diagnostics-show-line-numbers.
8710 * dwarf2out.c (gen_producer_string): Add
8711 OPT_fdiagnostics_show_line_numbers to the ignored options.
8712 * lto-wrapper.c (merge_and_complain): Likewise to the "pick
8713 one setting" options.
8714 (append_compiler_options): Likewise to the dropped options.
8715 (append_diag_options): Likewise to the passed-on options.
8716 * opts.c (common_handle_option): Handle the new option.
8717 * toplev.c (general_init): Set up global_dc->show_line_numbers_p.
8719 2018-08-09 Kelvin Nilsen <kelvin@gcc.gnu.org>
8721 * doc/extend.texi (PowerPC AltiVec Built-in Functions Available on
8722 ISA 2.07): Correct spelling of bcdsub to be __builtin_bcdsub. Add
8723 third argument of type "const signed char" to descriptions of
8724 __builtin_bcdadd, __builtin_bcdadd_lt, __builtin_bcdadd_eq,
8725 __builtin_bcdadd_gt, __builtin_bcdadd_ov, __builtin_bcdsub,
8726 __builtin_bcdsub_lt, __builtin_bcdsub_eq, __builtin_bcdsub_gt,
8727 __builtin_bcdsub_ov functions.
8729 2018-08-09 Richard Sandiford <richard.sandiford@arm.com>
8731 PR tree-optimization/86858
8732 * tree-vect-loop.c (vect_is_simple_reduction): Restore
8733 flow_bb_inside_loop_p calls.
8735 2018-08-09 Richard Sandiford <richard.sandiford@arm.com>
8737 PR tree-optimization/86871
8738 * tree-vect-stmts.c (vect_transform_stmt): Use gimple_get_lhs
8739 instead of gimple_assign_lhs.
8741 2018-08-09 Richard Earnshaw <rearnsha@arm.com>
8744 * config/aarch64/aarch64.md (add<mode>3_carryinC_zero): Add missing
8745 register constraint to operand 0.
8746 (add<mode>3_carryinC): Likewise.
8747 (add<mode>3_carryinV_zero, add<mode>3_carryinV): Likewise.
8749 2018-08-09 Martin Liska <mliska@suse.cz>
8752 * common.opt: Remove extra line.
8754 2018-08-09 Martin Liska <mliska@suse.cz>
8756 * params.def (PARAM_ALIGN_LOOP_ITERATIONS): Remove double dots
8757 at the end of a line, make first letter capital and end up
8758 a sentence with a dot.
8759 (PARAM_LOOP_INTERCHANGE_STRIDE_RATIO): Likewise.
8760 (PARAM_LOOP_BLOCK_TILE_SIZE): Likewise.
8761 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Likewise.
8762 (PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP): Likewise.
8763 (PARAM_MAX_ISL_OPERATIONS): Likewise.
8764 (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): Likewise.
8765 (PARAM_PROFILE_FUNC_INTERNAL_ID): Likewise.
8766 (PARAM_INDIR_CALL_TOPN_PROFILE): Likewise.
8767 (PARAM_SLP_MAX_INSNS_IN_BB): Likewise.
8768 (PARAM_IPA_CP_EVAL_THRESHOLD): Likewise.
8769 (PARAM_IPA_CP_RECURSION_PENALTY): Likewise.
8770 (PARAM_IPA_CP_SINGLE_CALL_PENALTY): Likewise.
8771 (PARAM_IPA_CP_LOOP_HINT_BONUS): Likewise.
8772 (PARAM_IPA_CP_ARRAY_INDEX_HINT_BONUS): Likewise.
8773 (PARAM_TREE_REASSOC_WIDTH): Likewise.
8774 (PARAM_HSA_GEN_DEBUG_STORES): Likewise.
8775 (PARAM_MAX_SPECULATIVE_DEVIRT_MAYDEFS): Likewise.
8776 (PARAM_MAX_VRP_SWITCH_ASSERTIONS): Likewise.
8778 2018-08-09 Andreas Krebbel <krebbel@linux.ibm.com>
8781 * config/s390/s390.c (s390_option_override_internal): Reduce the
8782 stack-clash-protection-probe-interval param if it would be too big
8785 2018-08-08 Andreas Schwab <schwab@linux-m68k.org>
8788 * config/m68k/m68k.h (FINAL_PRESCAN_INSN): Don't define.
8789 * config/m68k/m68k.c (handle_move_double): Don't call
8790 m68k_final_prescan_insn.
8791 (m68k_adjust_decorated_operand): Renamed from
8792 m68k_final_prescan_insn, remove first and third operand and
8794 (print_operand): Call it.
8795 (print_operand_address): Call it.
8797 2018-08-08 Nathan Sidwell <nathan@acm.org>
8799 * diagnostic.c (diagnostic_report_current_module): Use
8800 linemap_included_from & linemap_included_from_linemap.
8802 2018-08-08 Hongbo Zhang <hongbo.zhang@linaro.org>
8804 * config/aarch64/aarch64-cores.def: Add phecda core.
8805 * config/aarch64/aarch64-tune.md: Regenerate.
8806 * doc/invoke.texi: Add phecda core.
8808 2018-08-08 Andreas Krebbel <krebbel@linux.ibm.com>
8811 * config/s390/constraints.md ("NxHD0", "NxSD0"): New constraint
8813 * config/s390/s390.md ("movti"): Add more alternatives for
8814 constant to GPR copies.
8816 2018-08-08 Andreas Krebbel <krebbel@linux.ibm.com>
8818 * config/s390/s390.c: Fix whitespace damage throughout the file.
8819 * config/s390/s390.h: Likewise.
8820 * config/s390/tpf.h: Likewise.
8822 2018-08-08 Ilya Leoshkevich <iii@linux.ibm.com>
8824 * config/s390/s390.c (s390_loadrelative_operand_p):
8825 Remove TARGET_CPU_ZARCH usages.
8826 (s390_rtx_costs): Likewise.
8827 (s390_legitimate_constant_p): Likewise.
8828 (s390_cannot_force_const_mem): Likewise.
8829 (legitimate_reload_constant_p): Likewise.
8830 (s390_preferred_reload_class): Likewise.
8831 (legitimize_pic_address): Likewise.
8832 (legitimize_tls_address): Likewise.
8833 (s390_split_branches): Removed.
8834 (s390_add_execute): Removed.
8835 (s390_dump_pool): Remove TARGET_CPU_ZARCH usages.
8836 (s390_mainpool_start): Likewise.
8837 (s390_mainpool_finish): Likewise.
8838 (s390_mainpool_cancel): Removed.
8839 (s390_chunkify_start): Remove TARGET_CPU_ZARCH usages.
8840 (s390_chunkify_cancel): Likewise.
8841 (s390_return_addr_rtx): Likewise.
8842 (s390_register_info): Remove split_branches_pending_p uages.
8843 (s390_optimize_register_info): Likewise.
8844 (s390_init_frame_layout): Remove TARGET_CPU_ZARCH and
8845 split_branches_pending_p usages.
8846 (s390_can_eliminate): Remove TARGET_CPU_ZARCH usages.
8847 (s390_load_got): Likewise.
8848 (s390_expand_split_stack_prologue): Likewise.
8849 (output_asm_nops): Likewise.
8850 (s390_function_profiler): Likewise.
8851 (s390_emit_call): Likewise.
8852 (s390_conditional_register_usage): Likewise.
8853 (s390_optimize_prologue): Likewise.
8854 (s390_reorg): Remove TARGET_CPU_ZARCH and
8855 split_branches_pending_p usages.
8856 (s390_option_override_internal): Remove TARGET_CPU_ZARCH
8858 (s390_output_indirect_thunk_function): Likewise.
8859 * config/s390/s390.h (TARGET_CPU_ZARCH): Removed.
8860 (TARGET_CPU_ZARCH_P): Removed.
8861 (struct machine_function): Remove split_branches_pending_p.
8862 * config/s390/s390.md: Remove TARGET_CPU_ZARCH usages.
8864 2018-08-08 Ilya Leoshkevich <iii@linux.ibm.com>
8866 * common/config/s390/s390-common.c (processor_flags_table):
8868 * config.gcc: Remove with_arch/with_tune support.
8869 * config/s390/2064.md: Remove cpu attribute comparisons.
8870 * config/s390/driver-native.c (s390_host_detect_local_cpu):
8872 * config/s390/linux.h (ASM_SPEC):
8873 Remove -march support.
8874 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal):
8875 Use a table to get an arch level.
8876 * config/s390/s390-opts.h (enum processor_type):
8878 * config/s390/s390.c
8879 (processor_table): Remove entries, add arch_level values.
8880 (s390_issue_rate): Remove cases.
8881 (s390_option_override): Adjust
8882 s390_option_override_internal() call.
8883 (s390_option_override_internal): Remove deprecation warning.
8884 (s390_valid_target_attribute_tree): Adjust
8885 s390_option_override_internal() call.
8886 * config/s390/s390.h (struct s390_processor):
8887 Share with s390-c.c, add arch_level field.
8888 * config/s390/s390.md:
8889 Remove occurrences in cpu attribute.
8890 * config/s390/s390.opt: Remove -march/-mtune support.
8891 * config/s390/tpf.h (ASM_SPEC): Remove -march support.
8892 * doc/invoke.texi: Remove deprecation warning.
8894 2018-08-08 Luis Machado <luis.machado@linaro.org>
8896 * config/aarch64/aarch64.c (qdf24xx_vector_cost): New static
8898 (qdf24xx_tunings): Set vector cost structure to
8899 qdf24xx_vector_cost.
8901 * config/aarch64/aarch64.c (qdf24xx_addrcost_table)
8902 <register_sextend>: Set to 3.
8904 2018-08-07 Richard Sandiford <richard.sandiford@arm.com>
8907 * config/aarch64/iterators.md (FRECP, frecp_suffix): Delete.
8908 * config/aarch64/aarch64-simd.md
8909 (aarch64_frecp<FRECP:frecp_suffix><mode>): Fold FRECPE into...
8910 (@aarch64_frecpe<mode>): ...here and the move FRECPX to...
8911 (aarch64_frecpx<mode>): ...this new pattern.
8912 * config/aarch64/aarch64-simd-builtins.def: Remove comment
8913 about aarch64_frecp<FRECP:frecp_suffix><mode>.
8915 2018-08-07 Martin Liska <mliska@suse.cz>
8918 * predict.c (expr_expected_value_1): Handle DECL_IS_MALLOC,
8919 BUILT_IN_REALLOC and DECL_IS_OPERATOR_NEW.
8920 * predict.def (PRED_MALLOC_NONNULL): New predictor.
8921 * doc/extend.texi: Document that malloc attribute adds
8924 2018-08-06 John David Anglin <danglin@gcc.gnu.org>
8927 * config/pa/pa.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
8928 Define to speculation_safe_value_not_needed.
8930 2018-08-06 Jeff Law <law@redhat.com>
8932 * tree-ssa-dom.c (dom_opt_dom_walker::optimize_stmt): Pass down
8933 the vr_values instance to cprop_into_stmt.
8934 (cprop_into_stmt): Pass vr_values instance down to cprop_operand.
8935 (cprop_operand): Also query EVRP to determine if OP is a constant.
8937 2018-08-06 Nathan Sidwell <nathan@acm.org>
8939 * diagnostic.c (diagnostic_report_current_module): Reroll
8940 included-at loop. Translate text.
8942 2018-08-06 David Malcolm <dmalcolm@redhat.com>
8944 * function-tests.c (selftest::test_expansion_to_rtl): Call
8945 free_after_compilation.
8947 2018-08-06 Alan Hayward <alan.hayward@arm.com>
8949 * config/aarch64/aarch64.md: Add clobber highs to tls_desc.
8951 2018-08-06 Andreas Krebbel <krebbel@linux.ibm.com>
8953 * config/s390/s390.c (s390_loop_unroll_adjust): Prevent small
8954 loops with memory block operations from getting unrolled.
8956 2018-08-06 Ulrich Weigand <uweigand@de.ibm.com>
8959 * config/spu/spu.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
8960 Define to speculation_safe_value_not_needed.
8962 2018-08-06 Jeff Law <law@redhat.com>
8964 * reload1.c (forget_old_reloads_1): Adjust CLOBBER_HIGH
8967 2018-08-06 Jozef Lawrynowicz <jozef.l@mittosystems.com>
8970 * gcc/tree.c (build_common_tree_nodes): Initialize integer_types array
8971 with all enabled __intN types.
8973 * gcc/testsuite/gcc.target/msp430/pr86662.c: New test.
8975 2018-08-06 Alan Hayward <alan.hayward@arm.com>
8977 * alias.c (record_set): Check for clobber high.
8978 * cfgexpand.c (expand_gimple_stmt): Likewise.
8979 * combine-stack-adj.c (single_set_for_csa): Likewise.
8980 * combine.c (find_single_use_1): Likewise.
8981 (set_nonzero_bits_and_sign_copies): Likewise.
8982 (get_combine_src_dest): Likewise.
8983 (is_parallel_of_n_reg_sets): Likewise.
8984 (try_combine): Likewise.
8985 (record_dead_and_set_regs_1): Likewise.
8986 (reg_dead_at_p_1): Likewise.
8987 (reg_dead_at_p): Likewise.
8988 * dce.c (deletable_insn_p): Likewise.
8989 (mark_nonreg_stores_1): Likewise.
8990 (mark_nonreg_stores_2): Likewise.
8991 * df-scan.c (df_find_hard_reg_defs): Likewise.
8992 (df_uses_record): Likewise.
8993 (df_get_call_refs): Likewise.
8994 * dwarf2out.c (mem_loc_descriptor): Likewise.
8995 * haifa-sched.c (haifa_classify_rtx): Likewise.
8996 * ira-build.c (create_insn_allocnos): Likewise.
8997 * ira-costs.c (scan_one_insn): Likewise.
8998 * ira.c (equiv_init_movable_p): Likewise.
8999 (rtx_moveable_p): Likewise.
9000 (interesting_dest_for_shprep): Likewise.
9001 * jump.c (mark_jump_label_1): Likewise.
9002 * postreload-gcse.c (record_opr_changes): Likewise.
9003 * postreload.c (reload_cse_simplify): Likewise.
9004 (struct reg_use): Add source expr.
9005 (reload_combine): Check for clobber high.
9006 (reload_combine_note_use): Likewise.
9007 (reload_cse_move2add): Likewise.
9008 (move2add_note_store): Likewise.
9009 * print-rtl.c (print_pattern): Likewise.
9010 * recog.c (decode_asm_operands): Likewise.
9011 (store_data_bypass_p): Likewise.
9012 (if_test_bypass_p): Likewise.
9013 * regcprop.c (kill_clobbered_value): Likewise.
9014 (kill_set_value): Likewise.
9015 * reginfo.c (reg_scan_mark_refs): Likewise.
9016 * reload1.c (maybe_fix_stack_asms): Likewise.
9017 (eliminate_regs_1): Likewise.
9018 (elimination_effects): Likewise.
9019 (mark_not_eliminable): Likewise.
9020 (scan_paradoxical_subregs): Likewise.
9021 (forget_old_reloads_1): Likewise.
9022 * reorg.c (find_end_label): Likewise.
9023 (try_merge_delay_insns): Likewise.
9024 (redundant_insn): Likewise.
9025 (own_thread_p): Likewise.
9026 (fill_simple_delay_slots): Likewise.
9027 (fill_slots_from_thread): Likewise.
9028 (dbr_schedule): Likewise.
9029 * resource.c (update_live_status): Likewise.
9030 (mark_referenced_resources): Likewise.
9031 (mark_set_resources): Likewise.
9032 * rtl.c (copy_rtx): Likewise.
9033 * rtlanal.c (reg_referenced_p): Likewise.
9034 (single_set_2): Likewise.
9035 (noop_move_p): Likewise.
9036 (note_stores): Likewise.
9037 * sched-deps.c (sched_analyze_reg): Likewise.
9038 (sched_analyze_insn): Likewise.
9040 2018-08-06 Alan Hayward <alan.hayward@arm.com>
9042 * cse.c (invalidate_reg): New function extracted from...
9043 (invalidate): ...here.
9044 (canonicalize_insn): Check for clobber high.
9045 (invalidate_from_clobbers): invalidate clobber highs.
9046 (invalidate_from_sets_and_clobbers): Likewise.
9047 (count_reg_usage): Check for clobber high.
9048 (insn_live_p): Likewise.
9049 * cselib.c (cselib_expand_value_rtx_1):Likewise.
9050 (cselib_invalidate_regno): Check for clobber in setter.
9051 (cselib_invalidate_rtx): Pass through setter.
9052 (cselib_invalidate_rtx_note_stores):
9053 (cselib_process_insn): Check for clobber high.
9054 * cselib.h (cselib_invalidate_rtx): Add operand.
9056 2018-08-06 Alan Hayward <alan.hayward@arm.com>
9058 * lra-eliminations.c (lra_eliminate_regs_1): Check for clobber high.
9059 (mark_not_eliminable): Likewise.
9060 * lra-int.h (struct lra_insn_reg): Add clobber high marker.
9061 * lra-lives.c (process_bb_lives): Check for clobber high.
9062 * lra.c (new_insn_reg): Remember clobber highs.
9063 (collect_non_operand_hard_regs): Check for clobber high.
9064 (lra_set_insn_recog_data): Likewise.
9065 (add_regs_to_insn_regno_info): Likewise.
9066 (lra_update_insn_regno_info): Likewise.
9068 2018-08-06 Alan Hayward <alan.hayward@arm.com>
9070 * rtl.h (reg_is_clobbered_by_clobber_high): Add declarations.
9071 * rtlanal.c (reg_is_clobbered_by_clobber_high): Add function.
9073 2018-08-06 Alan Hayward <alan.hayward@arm.com>
9075 * emit-rtl.c (verify_rtx_sharing): Check for CLOBBER_HIGH.
9076 (copy_insn_1): Likewise.
9077 (gen_hard_reg_clobber_high): New gen function.
9078 * genconfig.c (walk_insn_part): Check for CLOBBER_HIGH.
9079 * genemit.c (gen_exp): Likewise.
9080 (gen_emit_seq): Pass through info.
9081 (gen_insn): Check for CLOBBER_HIGH.
9082 (gen_expand): Pass through info.
9083 (gen_split): Likewise.
9084 (output_add_clobbers): Likewise.
9085 * genrecog.c (validate_pattern): Check for CLOBBER_HIGH.
9086 (remove_clobbers): Likewise.
9087 * rtl.h (gen_hard_reg_clobber_high): New declaration.
9089 2018-08-06 Alan Hayward <alan.hayward@arm.com>
9091 * doc/rtl.texi (clobber_high): Add.
9092 (parallel): Add in clobber high
9093 * rtl.c (rtl_check_failed_code3): Add function.
9094 * rtl.def (CLOBBER_HIGH): Add expression.
9095 * rtl.h (RTL_CHECKC3): Add macro.
9096 (rtl_check_failed_code3): Add declaration.
9097 (XC3EXP): Add macro.
9099 2018-08-05 H.J. Lu <hongjiu.lu@intel.com>
9102 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Set
9103 cfun->machine->max_used_stack_alignment if needed.
9105 2018-08-04 Martin Sebor <msebor@redhat.com>
9107 PR tree-optimization/86571
9108 * gimple-ssa-sprintf.c (format_floating): Extend upper bound of
9111 2018-08-03 Sandra Loosemore <sandra@codesourcery.com>
9114 * config/nios2/nios2.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
9117 2018-08-03 Jeff Law <law@redhat.com>
9120 * config/mn10300/mn10300.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
9121 Define to speculation_safe_value_not_needed.
9123 2018-08-03 David Malcolm <dmalcolm@redhat.com>
9125 * doc/gcov.texi (-x): Remove duplicate "to".
9126 * doc/invoke.texi (-Wnoexcept-type): Remove duplicate "calls".
9127 (-Wif-not-aligned): Remove duplicate "is".
9128 (-flto): Remove duplicate "the".
9129 (MicroBlaze Options): In examples of "-mcpu=cpu-type", remove
9130 duplicate "v5.00.b".
9131 (MSP430 Options): Remove duplicate "and" from the description
9132 of "-mgprel-sec=regexp".
9133 (x86 Options): Remove duplicate copies of "vmldLog102" and
9134 vmlsLog104 from description of "-mveclibabi=type".
9136 2018-08-03 Richard Sandiford <richard.sandiford@arm.com>
9138 * internal-fn.h (first_commutative_argument): Declare.
9139 * internal-fn.c (first_commutative_argument): New function.
9140 * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove extra
9141 restrictions for pattern statements. Use first_commutative_argument
9142 to look for commutative operands in calls to internal functions.
9144 2018-08-03 Aldy Hernandez <aldyh@redhat.com>
9146 * Makefile.in (wide-int-range.o): New.
9147 * tree-vrp.c: Move all the wide_int_* functions to...
9148 * wide-int-range.cc: ...here.
9149 * tree-vrp.h: Move all the wide_int_* prototypes to...
9150 * wide-int-range.h: ...here.
9152 2018-08-03 Tom de Vries <tdevries@suse.de>
9154 * common/config/nvptx/nvptx-common.c (nvptx_except_unwind_info): Return
9156 * config/nvptx/nvptx.c (TARGET_ASM_BYTE_OP): Remove define.
9157 * except.c (output_function_exception_table): Do early exit if
9158 targetm_common.except_unwind_info (&global_options) == UI_NONE.
9160 2018-08-03 Martin Liska <mliska@suse.cz>
9162 * predict.c (dump_prediction): Change to 2 digits
9165 2018-08-03 Siddhesh Poyarekar <siddhesh@sourceware.org>
9167 * config/aarch64/falkor.md (falkor_am_1_vxvy_vxvy): Move
9169 (falkor_am_1_gtov_gtov): ... a new insn reservation.
9171 2018-07-19 Ilya Leoshkevich <iii@linux.ibm.com>
9173 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Replace > with >=.
9174 * df-problems.c (df_remove_dead_eq_notes): Replace > with >=.
9175 * dwarf2out.c (mem_loc_descriptor): Replace > with >=.
9176 * lra-constraints.c (spill_hard_reg_in_range): Replace <= with <.
9177 * lra-remat.c (call_used_input_regno_present_p): Replace <= with <.
9179 2018-08-02 David Malcolm <dmalcolm@redhat.com>
9181 * diagnostic-show-locus.c (diagnostic_show_locus): Use
9182 pp_take_prefix when saving the existing prefix.
9183 * diagnostic.c (diagnostic_append_note): Likewise.
9184 * langhooks.c (lhd_print_error_function): Likewise.
9185 * pretty-print.c (pp_set_prefix): Drop the "const" from "prefix"
9186 param's type. Free the existing prefix.
9187 (pp_take_prefix): New function.
9188 (pretty_printer::pretty_printer): Drop the prefix parameter.
9189 Rename the length parameter to match the comment.
9190 (pretty_printer::~pretty_printer): Free the prefix.
9191 * pretty-print.h (pretty_printer::pretty_printer): Drop the prefix
9193 (struct pretty_printer): Drop the "const" from "prefix" field's
9194 type and clarify memory management.
9195 (pp_set_prefix): Drop the "const" from the 2nd param.
9196 (pp_take_prefix): New decl.
9198 2018-08-02 Aaron Sawdey <acsawdey@linux.ibm.com>
9200 * config/rs6000/rs6000-string.c (select_block_compare_mode): Move test
9201 for word_mode_ok here instead of passing as argument.
9202 (expand_block_compare): Change select_block_compare_mode() call.
9203 (expand_strncmp_gpr_sequence): New function.
9204 (expand_strn_compare): Make use of expand_strncmp_gpr_sequence.
9206 2018-08-02 Jeff Law <law@redhat.com>
9209 * config/m68k/m68k.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
9210 Define to speculation_safe_value_not_needed.
9213 * config/h8300/h8300.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
9214 Define to speculation_safe_value_not_needed.
9216 2018-08-02 Tom de Vries <tdevries@suse.de>
9219 * common/config/nvptx/nvptx-common.c (nvptx_except_unwind_info): New
9220 function. Return UI_TARGET unconditionally.
9221 (TARGET_EXCEPT_UNWIND_INFO): Redefine to nvptx_except_unwind_info.
9222 * config/nvptx/nvptx.c (TARGET_ASM_BYTE_OP): Emit commented out '.byte'.
9224 2018-08-02 Richard Sandiford <richard.sandiford@arm.com>
9226 * genemit.c (print_overload_test): Fix typo.
9228 2018-08-02 Richard Biener <rguenther@suse.de>
9230 PR tree-optimization/86816
9231 * tree-ssa-tail-merge.c (tail_merge_valueize): New function
9232 which checks for value availability before querying it.
9233 (gvn_uses_equal): Use it.
9234 (same_succ_hash): Likewise.
9235 (gimple_equal_p): Likewise.
9237 2018-08-02 Nick Clifton <nickc@redhat.com>
9240 * config/stormy16/stormy16.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
9241 Define to speculation_safe_value_not_needed.
9244 * config/v850/v850.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
9245 Define to speculation_safe_value_not_needed.
9248 * config/v850/v850.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
9249 Define to speculation_safe_value_not_needed.
9252 * config/rx/rx.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
9253 Define to speculation_safe_value_not_needed.
9256 * config/msp430/msp430.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
9257 Define to speculation_safe_value_not_needed.
9260 * config/mcore/mcore.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
9261 Define to speculation_safe_value_not_needed.
9264 * config/m32r/m32r.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
9265 Define to speculation_safe_value_not_needed.
9268 * config/iq2000/iq2000.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
9269 Define to speculation_safe_value_not_needed.
9272 * config/frv/frv.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define to
9273 speculation_safe_value_not_needed.
9276 * config/fr30/fr30.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define
9277 to speculation_safe_value_not_needed.
9279 2018-08-02 Richard Sandiford <richard.sandiford@arm.com>
9281 * doc/md.texi: Expand the documentation of instruction names
9282 to mention port-local uses. Document '@' in pattern names.
9283 * read-md.h (overloaded_instance, overloaded_name): New structs.
9285 (md_reader::handle_overloaded_name): New member function.
9286 (md_reader::get_overloads): Likewise.
9287 (md_reader::m_first_overload): New member variable.
9288 (md_reader::m_next_overload_ptr): Likewise.
9289 (md_reader::m_overloads_htab): Likewise.
9290 * read-md.c (md_reader::md_reader): Initialize m_first_overload,
9291 m_next_overload_ptr and m_overloads_htab.
9292 * read-rtl.c (iterator_group): Add "type" and "get_c_token" fields.
9293 (get_mode_token, get_code_token, get_int_token): New functions.
9294 (map_attr_string): Add an optional argument that passes back
9295 the associated iterator.
9296 (overloaded_name_hash, overloaded_name_eq_p, named_rtx_p):
9297 (md_reader::handle_overloaded_name, add_overload_instance): New
9299 (apply_iterators): Handle '@' names. Report an error if '@'
9300 is used without iterators.
9301 (initialize_iterators): Initialize the new iterator_group fields.
9302 * genopinit.c (handle_overloaded_code_for)
9303 (handle_overloaded_gen): New functions.
9304 (main): Use them to print declarations of maybe_code_for_* and
9305 maybe_gen_* functions, and inline definitions of code_for_* and gen_*.
9306 * genemit.c (print_overload_arguments, print_overload_test)
9307 (handle_overloaded_code_for, handle_overloaded_gen): New functions.
9308 (main): Use it to print definitions of maybe_code_for_* and
9309 maybe_gen_* functions.
9310 * config/aarch64/aarch64.c (aarch64_split_128bit_move): Use
9311 gen_aarch64_mov{low,high}_di and gen_aarch64_movdi_{low,high}
9312 instead of explicit mode checks.
9313 (aarch64_split_simd_combine): Likewise gen_aarch64_simd_combine.
9314 (aarch64_split_simd_move): Likewise gen_aarch64_split_simd_mov.
9315 (aarch64_emit_load_exclusive): Likewise gen_aarch64_load_exclusive.
9316 (aarch64_emit_store_exclusive): Likewise gen_aarch64_store_exclusive.
9317 (aarch64_expand_compare_and_swap): Likewise
9318 gen_aarch64_compare_and_swap and gen_aarch64_compare_and_swap_lse
9319 (aarch64_gen_atomic_cas): Likewise gen_aarch64_atomic_cas.
9320 (aarch64_emit_atomic_swap): Likewise gen_aarch64_atomic_swp.
9321 (aarch64_constant_pool_reload_icode): Delete.
9322 (aarch64_secondary_reload): Use code_for_aarch64_reload_movcp
9323 instead of aarch64_constant_pool_reload_icode. Use
9324 code_for_aarch64_reload_mov instead of explicit mode checks.
9325 (rsqrte_type, get_rsqrte_type, rsqrts_type, get_rsqrts_type): Delete.
9326 (aarch64_emit_approx_sqrt): Use gen_aarch64_rsqrte instead of
9327 get_rsqrte_type and gen_aarch64_rsqrts instead of gen_rqrts_type.
9328 (recpe_type, get_recpe_type, recps_type, get_recps_type): Delete.
9329 (aarch64_emit_approx_div): Use gen_aarch64_frecpe instead of
9330 get_recpe_type and gen_aarch64_frecps instead of get_recps_type.
9331 (aarch64_atomic_load_op_code): Delete.
9332 (aarch64_emit_atomic_load_op): Likewise.
9333 (aarch64_gen_atomic_ldop): Use UNSPECV_ATOMIC_* instead of
9334 aarch64_atomic_load_op_code. Use gen_aarch64_atomic_load
9335 instead of aarch64_emit_atomic_load_op.
9336 * config/aarch64/aarch64.md (aarch64_reload_movcp<GPF_TF:mode><P:mode>)
9337 (aarch64_reload_movcp<VALL:mode><P:mode>, aarch64_reload_mov<mode>)
9338 (aarch64_movdi_<mode>low, aarch64_movdi_<mode>high)
9339 (aarch64_mov<mode>high_di, aarch64_mov<mode>low_di): Add a '@'
9340 character before the pattern name.
9341 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
9342 (aarch64_rsqrte<mode>, aarch64_rsqrts<mode>)
9343 (aarch64_simd_combine<mode>, aarch64_frecpe<mode>)
9344 (aarch64_frecps<mode>): Likewise.
9345 * config/aarch64/atomics.md (atomic_compare_and_swap<mode>)
9346 (aarch64_compare_and_swap<mode>, aarch64_compare_and_swap<mode>_lse)
9347 (aarch64_load_exclusive<mode>, aarch64_store_exclusive<mode>)
9348 (aarch64_atomic_swp<mode>, aarch64_atomic_cas<mode>)
9349 (aarch64_atomic_load<atomic_ldop><mode>): Likewise.
9351 2018-08-02 Richard Sandiford <richard.sandiford@arm.com>
9353 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
9354 Allow HFmode constants if TARGET_FP_F16INST.
9356 2018-08-02 Jackson Woodruff <jackson.woodruff@arm.com>
9359 * config/aarch64/aarch64.c (aarch64_operands_adjust_ok_for_ldpstp):
9360 No longer check last store for clobber of address register.
9362 2018-08-02 Martin Liska <mliska@suse.cz>
9364 PR gcov-profile/86817
9365 * gcov.c (process_all_functions): New function.
9367 (process_file): Move functions processing to
9368 process_all_functions.
9370 2018-08-02 David Malcolm <dmalcolm@redhat.com>
9372 * dumpfile.c (dump_user_location_t::dump_user_location_t): Add
9373 "const" to the "gimple *" and "rtx_insn *" parameters.
9374 * dumpfile.h (dump_user_location_t::dump_user_location_t):
9376 (dump_location_t::dump_location_t): Likewise.
9378 2018-08-01 Martin Sebor <msebor@redhat.com>
9380 PR tree-optimization/86650
9381 * gimple-pretty-print.c (percent_G_format): Accept a "gimple *"
9382 rather than a "gcall *". Directly pass the data of interest
9383 to percent_K_format, rather than building a temporary CALL_EXPR
9385 * gimple-fold.c (gimple_fold_builtin_strncpy): Adjust.
9386 (gimple_fold_builtin_strncat): Adjust.
9387 * gimple-ssa-warn-restrict.h (check_bounds_or_overlap): Replace
9388 gcall* argument with gimple*.
9389 * gimple-ssa-warn-restrict.c (check_call): Same.
9390 (wrestrict_dom_walker::before_dom_children): Same.
9391 (builtin_access::builtin_access): Same.
9392 (check_bounds_or_overlap): Same
9393 (maybe_diag_overlap): Same.
9394 (maybe_diag_offset_bounds): Same.
9395 * tree-diagnostic.c (default_tree_printer): Move usage of
9396 EXPR_LOCATION (t) and TREE_BLOCK (t) from within percent_K_format
9398 * tree-pretty-print.c (percent_K_format): Add argument.
9399 * tree-pretty-print.h: Add argument.
9400 * tree-ssa-ccp.c (pass_post_ipa_warn::execute): Adjust.
9401 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Adjust.
9402 (maybe_diag_stxncpy_trunc): Same.
9403 (handle_builtin_stxncpy): Same.
9404 (handle_builtin_strcat): Same.
9406 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
9408 * match.pd: Optimise pointer range checks.
9410 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
9412 PR tree-optimization/86758
9413 * tree-vect-stmts.c (vectorizable_simd_clone_call): Don't try
9414 to remove pattern statements.
9416 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
9418 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Use the
9419 result of dfs_enumerate_from when constructing stmt_vec_infos,
9420 instead of additionally calling get_loop_body.
9422 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
9424 * tree-vectorizer.h (vect_create_data_ref_ptr): Remove inv_p
9426 * tree-vect-data-refs.c (vect_create_data_ref_ptr): Likewise.
9427 When creating an iv, assert that the step is not known to be zero.
9428 (vect_setup_realignment): Update call accordingly.
9429 * tree-vect-stmts.c (vectorizable_store): Likewise.
9430 (vectorizable_load): Likewise. Handle VMAT_INVARIANT separately.
9432 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
9434 * tree-vectorizer.h (vect_stmt_to_vectorize): New function.
9435 * tree-vect-loop.c (vect_update_vf_for_slp): Use it.
9436 (vectorizable_reduction): Likewise.
9437 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
9438 (vect_detect_hybrid_slp_stmts): Likewise.
9439 * tree-vect-stmts.c (vect_is_simple_use): Likewise.
9441 2018-08-01 Aldy Hernandez <aldyh@redhat.com>
9443 * tree-vrp (zero_nonzero_bits_from_bounds): Rename to...
9444 (wide_int_set_zero_nonzero_bits): ...this.
9445 (zero_nonzero_bits_from_vr): Rename to...
9446 (vrp_set_zero_nonzero_bits): ...this.
9447 (extract_range_from_multiplicative_op_1): Abstract wide int
9449 (wide_int_range_multiplicative_op): ...here.
9450 (extract_range_from_binary_expr_1): Extract wide int binary
9451 operations into their own functions.
9452 (wide_int_range_lshift): New.
9453 (wide_int_range_can_optimize_bit_op): New.
9454 (wide_int_range_shift_undefined_p): New.
9455 (wide_int_range_bit_xor): New.
9456 (wide_int_range_bit_ior): New.
9457 (wide_int_range_bit_and): New.
9458 (wide_int_range_trunc_mod): New.
9459 (extract_range_into_wide_ints): New.
9460 (vrp_shift_undefined_p): New.
9461 (extract_range_from_multiplicative_op): New.
9462 (vrp_can_optimize_bit_op): New.
9463 * tree-vrp.h (value_range::dump): New.
9464 (wide_int_range_multiplicative_op): New.
9465 (wide_int_range_lshift):New.
9466 (wide_int_range_shift_undefined_p): New.
9467 (wide_int_range_bit_xor): New.
9468 (wide_int_range_bit_ior): New.
9469 (wide_int_range_bit_and): New.
9470 (wide_int_range_trunc_mod): New.
9471 (zero_nonzero_bits_from_bounds): Rename to...
9472 (wide_int_set_zero_nonzero_bits): ...this.
9473 (zero_nonzero_bits_from_vr): Rename to...
9474 (vrp_set_zero_nonzero_bits): ...this.
9475 (range_easy_mask_min_max): Rename to...
9476 (wide_int_range_can_optimize_bit_op): this.
9477 * vr-values.c (simplify_bit_ops_using_ranges): Rename
9478 zero_nonzero_bits_from_vr into vrp_set_zero_nonzero_bits.
9480 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
9482 * tree-vectorizer.h (vect_orig_stmt): New function.
9483 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Use it.
9484 * tree-vect-loop.c (vect_model_reduction_cost): Likewise.
9485 (vect_create_epilog_for_reduction): Likewise.
9486 (vectorizable_live_operation): Likewise.
9487 * tree-vect-slp.c (vect_find_last_scalar_stmt_in_slp): Likewise.
9488 (vect_detect_hybrid_slp_stmts, vect_schedule_slp): Likewise.
9489 * tree-vect-stmts.c (vectorizable_call): Likewise.
9490 (vectorizable_simd_clone_call, vect_remove_stores): Likewise.
9492 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
9494 * tree-vectorizer.h (vect_transform_stmt): Remove grouped_store
9496 * tree-vect-stmts.c (vect_transform_stmt): Likewise.
9497 * tree-vect-loop.c (vect_transform_loop_stmt): Update call accordingly.
9498 (vect_transform_loop): Likewise.
9499 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
9501 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
9503 * tree-vectorizer.h (vect_schedule_slp): Return void.
9504 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
9505 (vect_schedule_slp): Likewise.
9507 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
9509 * tree-vect-loop.c (vect_transform_loop_stmt): Remove slp_scheduled
9511 (vect_transform_loop): Update calls accordingly. Schedule SLP
9512 instances before the main loop, if any exist.
9514 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
9516 PR tree-optimization/86749
9517 * tree-vect-patterns.c (vect_determine_min_output_precision_1):
9518 If the lhs is used in a COND_EXPR, check that it is being used
9519 as the "then" or "else" value.
9521 2018-08-01 Tom de Vries <tdevries@suse.de>
9524 * config/nvptx/nvptx.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define to
9525 speculation_safe_value_not_needed.
9527 2018-08-01 Richard Biener <rguenther@suse.de>
9529 * tree-ssa-sccvn.c (visit_phi): Compare invariant addresses
9532 2018-08-01 Martin Liska <mliska@suse.cz>
9534 * value-prof.c (gimple_divmod_fixed_value_transform): Unify
9535 format how successful transformation is dumped.
9536 (gimple_mod_pow2_value_transform): Likewise.
9537 (gimple_mod_subtract_transform): Likewise.
9538 (gimple_stringops_transform): Likewise.
9540 2018-08-01 Martin Liska <mliska@suse.cz>
9543 * value-prof.c (interesting_stringop_to_profile_p):
9544 Simplify the code and add BUILT_IN_MEMMOVE.
9545 (gimple_stringops_transform): Likewise.
9547 2018-08-01 Sam Tebbs <sam.tebbs@arm.com>
9549 * config/aarch64/aarch64-simd.md
9550 (*aarch64_get_lane_zero_extendsi<mode>): Rename to...
9551 (*aarch64_get_lane_zero_extend<GPI:mode><VDQQH:mode>): ... This and
9552 use GPI iterator instead of SI mode.
9554 2018-08-01 Richard Earnshaw <rearnsha@arm.com>
9556 * config/rs6000/rs6000.md (speculation_barrier): Renamed from
9557 rs6000_speculation_barrier.
9558 * config/rs6000/rs6000.c (rs6000_expand_builtin): Adjust for
9559 new barrier pattern name.
9561 2018-08-01 Richard Earnshaw <rearnsha@arm.com>
9563 * config/i386/i386.md (unspecv): Add UNSPECV_SPECULATION_BARRIER.
9564 (speculation_barrier): New insn.
9566 2018-08-01 Richard Biener <rguenther@suse.de>
9569 * graphite.h: Include isl/id.h and isl/space.h to allow build
9572 2018-08-01 Jan Willem Jagersma <jwjagersma@gmail.com>
9575 * dwarf2out.c (dwarf2out_early_finish): Do not generate assembly in LTO
9576 mode for COFF targets.
9577 * defaults.h (TARGET_COFF): Define.
9578 * config/i386/djgpp.h (TARGET_ASM_LTO_START, TARGET_ASM_LTO_END,
9579 TARGET_COFF): Define.
9580 (i386_djgpp_asm_lto_start, i386_djgpp_asm_lto_end): Declare.
9581 * config/i386/djgpp.c (saved_debug_info_level): New static variable.
9582 (i386_djgpp_asm_lto_start, i386_djgpp_asm_lto_end): New functions.
9584 2018-07-31 Alexandre Oliva <oliva@adacore.com>
9586 * gimple-streamer-in.c (input_bb): Restore BB discriminator.
9587 * gimple-streamer-out.c (output_bb): Save it.
9588 * lto-streamer-in.c (input_struct_function_base): Restore
9589 instance discriminator if available. Create map on demand.
9590 * lto-streamer-out.c (output_struct_function_base): Save it if
9592 * final.c (decl_to_instance_map): Document LTO strategy.
9594 2018-07-31 Alexandre Oliva <oliva@adacore.com>
9595 Olivier Hainque <hainque@adacore.com>
9597 * debug.h (decl_to_instance_map_t): New type.
9598 (decl_to_instance_map): Declare.
9599 (maybe_create_decl_to_instance_map): New inline function.
9600 * final.c (bb_discriminator, last_bb_discriminator): New statics,
9601 to track basic block discriminators.
9602 (final_start_function_1): Initialize them.
9603 (final_scan_insn_1): On NOTE_INSN_BASIC_BLOCK, track
9605 (decl_to_instance_map): New variable.
9606 (map_decl_to_instance, maybe_set_discriminator): New functions.
9607 (notice_source_line): Set discriminator.
9609 2018-07-31 Ian Lance Taylor <iant@golang.org>
9611 * targhooks.c (default_have_speculation_safe_value): Add
9614 2018-07-31 David Malcolm <dmalcolm@redhat.com>
9616 * dump-context.h: Include "pretty-print.h".
9617 (dump_context::refresh_dumps_are_enabled): New decl.
9618 (dump_context::emit_item): New decl.
9619 (class dump_context): Add fields "m_test_pp" and
9621 (temp_dump_context::temp_dump_context): Add param "test_pp_flags".
9622 (temp_dump_context::get_dumped_text): New decl.
9623 (class temp_dump_context): Add field "m_pp".
9624 * dumpfile.c (refresh_dumps_are_enabled): Convert to...
9625 (dump_context::refresh_dumps_are_enabled): ...and add a test for
9627 (set_dump_file): Update for above change.
9628 (set_alt_dump_file): Likewise.
9629 (dump_loc): New overload, taking a pretty_printer *.
9630 (dump_context::dump_loc): Call end_any_optinfo. Dump the location
9631 to any test pretty-printer.
9632 (make_item_for_dump_gimple_stmt): New function, adapted from
9633 optinfo::add_gimple_stmt.
9634 (dump_context::dump_gimple_stmt): Call it, and use the result,
9635 eliminating the direct usage of dump_file and alt_dump_file in
9636 favor of indirectly using them via emit_item.
9637 (make_item_for_dump_gimple_expr): New function, adapted from
9638 optinfo::add_gimple_expr.
9639 (dump_context::dump_gimple_expr): Call it, and use the result,
9640 eliminating the direct usage of dump_file and alt_dump_file in
9641 favor of indirectly using them via emit_item.
9642 (make_item_for_dump_generic_expr): New function, adapted from
9644 (dump_context::dump_generic_expr): Call it, and use the result,
9645 eliminating the direct usage of dump_file and alt_dump_file in
9646 favor of indirectly using them via emit_item.
9647 (make_item_for_dump_printf_va): New function, adapted from
9648 optinfo::add_printf_va.
9649 (make_item_for_dump_printf): New function.
9650 (dump_context::dump_printf_va): Call make_item_for_dump_printf_va,
9651 and use the result, eliminating the direct usage of dump_file and
9652 alt_dump_file in favor of indirectly using them via emit_item.
9653 (make_item_for_dump_dec): New function.
9654 (dump_context::dump_dec): Call it, and use the result,
9655 eliminating the direct usage of dump_file and alt_dump_file in
9656 favor of indirectly using them via emit_item.
9657 (make_item_for_dump_symtab_node): New function, adapted from
9658 optinfo::add_symtab_node.
9659 (dump_context::dump_symtab_node): Call it, and use the result,
9660 eliminating the direct usage of dump_file and alt_dump_file in
9661 favor of indirectly using them via emit_item.
9662 (dump_context::begin_scope): Reimplement, avoiding direct usage
9663 of dump_file and alt_dump_file in favor of indirectly using them
9665 (dump_context::emit_item): New member function.
9666 (temp_dump_context::temp_dump_context): Add param "test_pp_flags".
9667 Set up test pretty-printer on the underlying context. Call
9668 refresh_dumps_are_enabled.
9669 (temp_dump_context::~temp_dump_context): Call
9670 refresh_dumps_are_enabled.
9671 (temp_dump_context::get_dumped_text): New member function.
9672 (selftest::verify_dumped_text): New function.
9673 (ASSERT_DUMPED_TEXT_EQ): New macro.
9674 (selftest::test_capture_of_dump_calls): Run all tests twice, with
9675 and then without optinfo enabled. Add uses of
9676 ASSERT_DUMPED_TEXT_EQ to all tests. Add test of nested scopes.
9677 * dumpfile.h: Update comment for the dump_* API.
9678 * optinfo-emit-json.cc
9679 (selftest::test_building_json_from_dump_calls): Update for new
9680 param for temp_dump_context ctor.
9681 * optinfo.cc (optinfo_item::optinfo_item): Remove "owned" param
9682 and "m_owned" field.
9683 (optinfo_item::~optinfo_item): Likewise.
9684 (optinfo::add_item): New member function.
9685 (optinfo::emit): Update comment.
9686 (optinfo::add_string): Delete.
9687 (optinfo::add_printf): Delete.
9688 (optinfo::add_printf_va): Delete.
9689 (optinfo::add_gimple_stmt): Delete.
9690 (optinfo::add_gimple_expr): Delete.
9691 (optinfo::add_tree): Delete.
9692 (optinfo::add_symtab_node): Delete.
9693 (optinfo::add_dec): Delete.
9694 * optinfo.h (class dump_context): New forward decl.
9695 (optinfo::add_item): New decl.
9696 (optinfo::add_string): Delete.
9697 (optinfo::add_printf): Delete.
9698 (optinfo::add_printf_va): Delete.
9699 (optinfo::add_gimple_stmt): Delete.
9700 (optinfo::add_gimple_expr): Delete.
9701 (optinfo::add_tree): Delete.
9702 (optinfo::add_symtab_node): Delete.
9703 (optinfo::add_dec): Delete.
9704 (optinfo::add_poly_int): Delete.
9705 (optinfo_item::optinfo_item): Remove "owned" param.
9706 (class optinfo_item): Remove field "m_owned".
9708 2018-07-31 Jozef Lawrynowicz <jozef.l@mittosystems.com>
9711 * gcc/cfgexpand.c (set_parm_rtl): Use the alignment of Pmode when
9712 MAX_SUPPORTED_STACK_ALIGNMENT would otherwise be exceeded by the
9713 requested variable alignment.
9714 (expand_one_ssa_partition): Likewise.
9715 (expand_one_var): Likewise.
9717 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
9719 * config/pdp11/pdp11.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Redefine
9720 to speculation_safe_value_not_needed.
9722 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
9724 * targhooks.h (speculation_safe_value_not_needed): New prototype.
9725 * targhooks.c (speculation_safe_value_not_needed): New function.
9726 * target.def (have_speculation_safe_value): Update documentation.
9727 * doc/tm.texi: Regenerated.
9729 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
9731 * config/aarch64/iterators.md (ALLI_TI): New iterator.
9732 * config/aarch64/aarch64.md (despeculate_copy<ALLI_TI:mode>): New
9734 (despeculate_copy<ALLI:mode>_insn): New insn.
9735 (despeculate_copyti_insn): New insn.
9736 (despeculate_simple<ALLI:mode>): New insn
9737 (despeculate_simpleti): New insn.
9738 * config/aarch64/aarch64.c (aarch64_speculation_safe_value): New
9740 (TARGET_SPECULATION_SAFE_VALUE): Redefine to
9741 aarch64_speculation_safe_value.
9742 (aarch64_print_operand): Handle const0_rtx in modifier 'H'.
9744 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
9746 * config/aarch64/aarch64-speculation.cc: New file.
9747 * config/aarch64/aarch64-passes.def (pass_track_speculation): Add
9748 before pass_reorder_blocks.
9749 * config/aarch64/aarch64-protos.h (make_pass_track_speculation): Add
9751 * config/aarch64/aarch64.c (aarch64_conditional_register_usage): Fix
9752 X14 and X15 when tracking speculation.
9753 * config/aarch64/aarch64.md (register name constants): Add
9754 SPECULATION_TRACKER_REGNUM and SPECULATION_SCRATCH_REGNUM.
9755 (unspec): Add UNSPEC_SPECULATION_TRACKER.
9756 (speculation_barrier): New insn attribute.
9757 (cmp<mode>): Allow SP in comparisons.
9758 (speculation_tracker): New insn.
9759 (speculation_barrier): Add speculation_barrier attribute.
9760 * config/aarch64/t-aarch64: Add make rule for aarch64-speculation.o.
9761 * config.gcc (aarch64*-*-*): Add aarch64-speculation.o to extra_objs.
9762 * doc/invoke.texi (AArch64 Options): Document -mtrack-speculation.
9764 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
9766 * config/aarch64/aarch64.md (cb<optab><mode>1): Disable when
9767 aarch64_track_speculation is true.
9768 (tb<optab><mode>1): Likewise.
9769 * config/aarch64/aarch64.c (aarch64_split_compare_regs): Do not
9770 generate CB[N]Z when tracking speculation.
9771 (aarch64_split_compare_and_swap): Likewise.
9772 (aarch64_split_atomic_op): Likewise.
9774 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
9776 * config/aarch64/aarch64.opt (mtrack-speculation): New target option.
9778 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
9780 * config/aarch64.md (unspecv): Add UNSPECV_SPECULAION_BARRIER.
9781 (speculation_barrier): New insn.
9783 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
9785 * config/arm/unspecs.md (unspecv): Add VUNSPEC_SPECULATION_BARRIER.
9786 * config/arm/arm.md (speculation_barrier): New expand.
9787 (speculation_barrier_insn): New pattern.
9789 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
9791 * builtin-types.def (BT_FN_PTR_PTR_VAR): New function type.
9792 (BT_FN_I1_I1_VAR, BT_FN_I2_I2_VAR, BT_FN_I4_I4_VAR): Likewise.
9793 (BT_FN_I8_I8_VAR, BT_FN_I16_I16_VAR): Likewise.
9794 * builtin-attrs.def (ATTR_NOVOPS_NOTHROW_LEAF_LIST): New attribute
9796 * builtins.def (BUILT_IN_SPECULATION_SAFE_VALUE_N): New builtin.
9797 (BUILT_IN_SPECULATION_SAFE_VALUE_PTR): New internal builtin.
9798 (BUILT_IN_SPECULATION_SAFE_VALUE_1): Likewise.
9799 (BUILT_IN_SPECULATION_SAFE_VALUE_2): Likewise.
9800 (BUILT_IN_SPECULATION_SAFE_VALUE_4): Likewise.
9801 (BUILT_IN_SPECULATION_SAFE_VALUE_8): Likewise.
9802 (BUILT_IN_SPECULATION_SAFE_VALUE_16): Likewise.
9803 * builtins.c (expand_speculation_safe_value): New function.
9804 (expand_builtin): Call it.
9805 * doc/cpp.texi: Document predefine __HAVE_SPECULATION_SAFE_VALUE.
9806 * doc/extend.texi: Document __builtin_speculation_safe_value.
9807 * doc/md.texi: Document "speculation_barrier" pattern.
9808 * doc/tm.texi.in: Pull in TARGET_SPECULATION_SAFE_VALUE and
9809 TARGET_HAVE_SPECULATION_SAFE_VALUE.
9810 * doc/tm.texi: Regenerated.
9811 * target.def (have_speculation_safe_value, speculation_safe_value): New
9813 * targhooks.c (default_have_speculation_safe_value): New function.
9814 (default_speculation_safe_value): New function.
9815 * targhooks.h (default_have_speculation_safe_value): Add prototype.
9816 (default_speculation_safe_value): Add prototype.
9818 2018-07-31 David Malcolm <dmalcolm@redhat.com>
9820 * dump-context.h (dump_context::dump_loc): New decl.
9821 * dumpfile.c (dump_context::dump_loc): New member function.
9822 (dump_context::dump_gimple_stmt_loc): Reimplement using dump_loc
9823 and dump_gimple_stmt.
9824 (dump_context::dump_gimple_expr_loc): Likewise, using
9826 (dump_context::dump_generic_expr_loc): Likewise, using
9828 (dump_context::dump_printf_loc_va): Likewise, using
9830 (dump_context::begin_scope): Explicitly using the global function
9831 "dump_loc", rather than the member function.
9833 2018-07-31 Martin Sebor <msebor@redhat.com>
9835 PR tree-optimization/86741
9836 * tree-vrp.c (vrp_prop::check_mem_ref): Avoid incomplete types.
9838 2018-07-31 Andreas Krebbel <krebbel@linux.ibm.com>
9840 * config/s390/s390.c (s390_expand_setmem): Make the unrolling to
9841 depend on whether prefetch instructions will be emitted or not.
9842 Use TARGET_SETMEM_PFD for checking whether prefetch instructions
9843 will be emitted or not.
9844 * config/s390/s390.h (TARGET_SETMEM_PREFETCH_DISTANCE)
9845 (TARGET_SETMEM_PFD): New macros.
9847 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
9849 * tree-vectorizer.h (stmt_vec_info): Turn back into a typedef.
9850 (NULL_STMT_VEC_INFO): Delete.
9851 (stmt_vec_info::operator*): Likewise.
9852 (stmt_vec_info::operator gimple *): Likewise.
9853 * tree-vect-loop.c (vectorizable_reduction): Use NULL instead
9854 of NULL_STMT_VEC_INFO.
9855 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
9856 (vect_reassociating_reduction_p): Likewise.
9857 * tree-vect-stmts.c (vect_build_gather_load_calls): Likewise.
9858 (vectorizable_store): Likewise.
9859 * tree-vectorizer.c (vec_info::set_vinfo_for_stmt): Likewise.
9860 (vec_info::free_stmt_vec_infos): Likewise.
9862 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
9864 * tree-vectorizer.h (vect_stmt_in_region_p): Delete.
9865 * tree-vectorizer.c (vect_stmt_in_region_p): Likewise.
9867 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
9869 * tree-vectorizer.h (vec_info::new_vinfo_for_stmt)
9870 (vec_info::set_vinfo_for_stmt, vec_info::free_stmt_vec_infos)
9871 (vec_info::free_stmt_vec_info): New private member functions.
9872 (set_stmt_vec_info_vec, free_stmt_vec_infos, vinfo_for_stmt)
9873 (set_vinfo_for_stmt, new_stmt_vec_info, free_stmt_vec_info): Delete.
9874 * tree-parloops.c (gather_scalar_reductions): Remove calls to
9875 set_stmt_vec_info_vec and free_stmt_vec_infos.
9876 * tree-vect-loop.c (_loop_vec_info): Remove call to
9877 set_stmt_vec_info_vec.
9878 * tree-vect-stmts.c (new_stmt_vec_info, set_stmt_vec_info_vec)
9879 (free_stmt_vec_infos, free_stmt_vec_info): Delete in favor of...
9880 * tree-vectorizer.c (vec_info::new_stmt_vec_info)
9881 (vec_info::set_vinfo_for_stmt, vec_info::free_stmt_vec_infos)
9882 (vec_info::free_stmt_vec_info): ...these new functions. Remove
9883 assignments in {vec_info::,}new_stmt_vec_info that are redundant
9884 with the clearing in the xcalloc.
9885 (stmt_vec_info_vec): Delete.
9886 (vec_info::vec_info): Don't call set_stmt_vec_info_vec.
9887 (vectorize_loops): Likewise.
9888 (vec_info::~vec_info): Remove argument from call to
9889 free_stmt_vec_infos.
9890 (vec_info::add_stmt): Remove vinfo argument from call to
9893 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
9895 * tree-vectorizer.h (free_stmt_vec_info): Take a stmt_vec_info
9896 rather than a gimple stmt.
9897 * tree-vect-stmts.c (free_stmt_vec_info): Likewise. Don't free
9898 information for pattern statements when passed the original
9899 statement; instead wait to be passed the pattern statement itself.
9900 Don't call set_vinfo_for_stmt here.
9901 (free_stmt_vec_infos): Update call to free_stmt_vec_info.
9902 * tree-vect-loop.c (_loop_vec_info::~loop_vec_info): Don't free
9903 stmt_vec_infos here.
9904 * tree-vect-slp.c (_bb_vec_info::~bb_vec_info): Likewise.
9905 * tree-vectorizer.c (vec_info::remove_stmt): Nullify the statement's
9906 stmt_vec_infos entry.
9908 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
9910 * tree-vectorizer.h (vec_info::replace_stmt): Declare.
9911 * tree-vectorizer.c (vec_info::replace_stmt): New function.
9912 * tree-vect-slp.c (vect_remove_slp_scalar_calls): Use it.
9913 * tree-vect-stmts.c (vectorizable_call): Likewise.
9914 (vectorizable_simd_clone_call): Likewise.
9916 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
9918 * tree-vectorizer.h (vec_info::remove_stmt): Declare.
9919 * tree-vectorizer.c (vec_info::remove_stmt): New function.
9920 * tree-vect-loop-manip.c (vect_set_loop_condition): Use it.
9921 * tree-vect-loop.c (vect_transform_loop): Likewise.
9922 * tree-vect-slp.c (vect_schedule_slp): Likewise.
9923 * tree-vect-stmts.c (vect_remove_stores): Likewise.
9925 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
9927 * tree-vectorizer.h (vec_info::lookup_dr): New member function.
9928 (vect_dr_stmt): Delete.
9929 * tree-vectorizer.c (vec_info::lookup_dr): New function.
9930 * tree-vect-loop-manip.c (vect_update_inits_of_drs): Use it instead
9932 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr)
9933 (vect_analyze_data_ref_dependence, vect_record_base_alignments)
9934 (vect_verify_datarefs_alignment, vect_peeling_supportable)
9935 (vect_analyze_data_ref_accesses, vect_prune_runtime_alias_test_list)
9936 (vect_analyze_data_refs): Likewise.
9937 (vect_slp_analyze_data_ref_dependence): Likewise. Take a vec_info
9939 (vect_find_same_alignment_drs): Likewise.
9940 (vect_slp_analyze_node_dependences): Update calls accordingly.
9941 (vect_analyze_data_refs_alignment): Likewise. Use vec_info::lookup_dr
9942 instead of DR_VECT_AUX.
9943 (vect_get_peeling_costs_all_drs): Take a loop_vec_info instead
9944 of a vector data references. Use vec_info::lookup_dr instead of
9946 (vect_peeling_hash_get_lowest_cost): Update calls accordingly.
9947 (vect_enhance_data_refs_alignment): Likewise. Use vec_info::lookup_dr
9948 instead of DR_VECT_AUX.
9950 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
9952 * tree-vectorizer.h (_loop_vec_info::unaligned_dr): Change to
9954 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Update
9956 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
9957 * tree-vect-loop-manip.c (get_misalign_in_elems): Likewise.
9958 (vect_gen_prolog_loop_niters): Likewise.
9960 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
9962 * tree-vectorizer.h (set_dr_misalignment, dr_misalignment)
9963 (DR_TARGET_ALIGNMENT, aligned_access_p, known_alignment_for_access_p)
9964 (vect_known_alignment_in_bytes, vect_dr_behavior)
9965 (vect_get_scalar_dr_size): Take references as dr_vec_infos
9966 instead of data_references. Update calls to other routines for
9967 which the same change has been made.
9968 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Take
9969 dr_vec_infos instead of stmt_vec_infos.
9970 (vect_analyze_data_ref_dependence): Update call accordingly.
9971 (vect_slp_analyze_data_ref_dependence)
9972 (vect_record_base_alignments): Use DR_VECT_AUX.
9973 (vect_calculate_target_alignment, vect_compute_data_ref_alignment)
9974 (vect_update_misalignment_for_peel, verify_data_ref_alignment)
9975 (vector_alignment_reachable_p, vect_get_data_access_cost)
9976 (vect_peeling_supportable, vect_analyze_group_access_1)
9977 (vect_analyze_group_access, vect_analyze_data_ref_access)
9978 (vect_vfa_segment_size, vect_vfa_access_size, vect_vfa_align)
9979 (vect_compile_time_alias, vect_small_gap_p)
9980 (vectorizable_with_step_bound_p, vect_duplicate_ssa_name_ptr_info):
9981 (vect_supportable_dr_alignment): Take references as dr_vec_infos
9982 instead of data_references. Update calls to other routines for
9983 which the same change has been made.
9984 (vect_verify_datarefs_alignment, vect_get_peeling_costs_all_drs)
9985 (vect_find_same_alignment_drs, vect_analyze_data_refs_alignment)
9986 (vect_slp_analyze_and_verify_node_alignment)
9987 (vect_analyze_data_ref_accesses, vect_prune_runtime_alias_test_list)
9988 (vect_create_addr_base_for_vector_ref, vect_create_data_ref_ptr)
9989 (vect_setup_realignment): Use dr_vec_infos. Update calls after
9991 (_vect_peel_info::dr): Replace with...
9992 (_vect_peel_info::dr_info): ...this new field.
9993 (vect_peeling_hash_get_most_frequent)
9994 (vect_peeling_hash_choose_best_peeling): Update accordingly.
9995 (vect_peeling_hash_get_lowest_cost):
9996 (vect_enhance_data_refs_alignment): Likewise. Update calls to other
9997 routines for which the same change has been made.
9998 (vect_peeling_hash_insert): Likewise. Take a dr_vec_info instead of a
10000 * tree-vect-loop-manip.c (get_misalign_in_elems)
10001 (vect_gen_prolog_loop_niters): Use dr_vec_infos. Update calls after
10003 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
10004 * tree-vect-stmts.c (vect_get_store_cost, vect_get_load_cost)
10005 (vect_truncate_gather_scatter_offset, compare_step_with_zero)
10006 (get_group_load_store_type, get_negative_load_store_type)
10007 (vect_get_data_ptr_increment, vectorizable_store)
10008 (vectorizable_load): Likewise.
10009 (ensure_base_align): Take a dr_vec_info instead of a data_reference.
10010 Update calls to other routines for which the same change has been made.
10012 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
10014 * tree-vectorizer.h (vec_info::move_dr): New member function.
10015 (dataref_aux): Rename to...
10016 (dr_vec_info): ...this and add "dr" and "stmt" fields.
10017 (_stmt_vec_info::dr_aux): Update accordingly.
10018 (_stmt_vec_info::data_ref_info): Delete.
10019 (STMT_VINFO_GROUPED_ACCESS, DR_GROUP_FIRST_ELEMENT)
10020 (DR_GROUP_NEXT_ELEMENT, DR_GROUP_SIZE, DR_GROUP_STORE_COUNT)
10021 (DR_GROUP_GAP, DR_GROUP_SAME_DR_STMT, REDUC_GROUP_FIRST_ELEMENT):
10022 (REDUC_GROUP_NEXT_ELEMENT, REDUC_GROUP_SIZE): Use dr_aux.dr instead
10024 (STMT_VINFO_DATA_REF): Likewise. Turn into an lvalue.
10025 (STMT_VINFO_DR_INFO): New macro.
10026 (DR_VECT_AUX): Use STMT_VINFO_DR_INKFO and vect_dr_stmt.
10027 (set_dr_misalignment): Update after rename of dataref_aux.
10028 (vect_dr_stmt): Move earlier in file. Return dr_aux.stmt.
10029 * tree-vect-stmts.c (new_stmt_vec_info): Remove redundant
10030 initialization of STMT_VINFO_DATA_REF.
10031 * tree-vectorizer.c (vec_info::move_dr): New function.
10032 * tree-vect-patterns.c (vect_recog_bool_pattern)
10033 (vect_recog_mask_conversion_pattern)
10034 (vect_recog_gather_scatter_pattern): Use it.
10035 * tree-vect-data-refs.c (vect_analyze_data_refs): Initialize
10036 the "dr" and "stmt" fields of dr_vec_info instead of
10037 STMT_VINFO_DATA_REF.
10039 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
10041 * tree-vectorizer.h (_stmt_vec_info::pattern_stmt_p): New field.
10042 (is_pattern_stmt_p): Use it.
10043 * tree-vect-patterns.c (vect_init_pattern_stmt): Set pattern_stmt_p
10044 on pattern statements.
10046 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
10048 * tree-vect-patterns.c (vect_mark_pattern_stmts): Take the
10049 original stmt as a stmt_vec_info rather than a gimple stmt.
10050 (vect_pattern_recog_1): Take the statement directly as a
10051 stmt_vec_info, rather than via a gimple_stmt_iterator.
10052 Update call to vect_mark_pattern_stmts.
10053 (vect_pattern_recog): Update calls accordingly.
10055 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
10057 * tree-vectorizer.h (vect_get_vec_defs_for_stmt_copy)
10058 (vect_get_vec_def_for_stmt_copy): Take a vec_info rather than
10059 a vect_def_type for the first argument.
10060 * tree-vect-stmts.c (vect_get_vec_defs_for_stmt_copy): Likewise.
10061 (vect_get_vec_def_for_stmt_copy): Likewise. Return the original
10062 operand if it isn't defined by a vectorized statement.
10063 (vect_build_gather_load_calls): Remove the mask_dt argument and
10064 update calls to vect_get_vec_def_for_stmt_copy.
10065 (vectorizable_bswap): Likewise the dt argument.
10066 (vectorizable_call): Update calls to vectorizable_bswap and
10067 vect_get_vec_def_for_stmt_copy.
10068 (vectorizable_simd_clone_call, vectorizable_assignment)
10069 (vectorizable_shift, vectorizable_operation, vectorizable_condition)
10070 (vectorizable_comparison): Update calls to
10071 vect_get_vec_def_for_stmt_copy.
10072 (vectorizable_store): Likewise. Remove now-unnecessary calls to
10073 vect_is_simple_use.
10074 (vect_get_loop_based_defs): Remove dt argument and update call
10075 to vect_get_vec_def_for_stmt_copy.
10076 (vectorizable_conversion): Update calls to vect_get_loop_based_defs
10077 and vect_get_vec_def_for_stmt_copy.
10078 (vectorizable_load): Update calls to vect_build_gather_load_calls
10079 and vect_get_vec_def_for_stmt_copy.
10080 * tree-vect-loop.c (vect_create_epilog_for_reduction)
10081 (vectorizable_reduction, vectorizable_live_operation): Update calls
10082 to vect_get_vec_def_for_stmt_copy.
10084 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
10086 * tree-vect-data-refs.c (vect_record_base_alignment): Replace vec_info
10087 and gimple stmt arguments with a stmt_vec_info.
10088 (vect_record_base_alignments): Update calls accordingly.
10089 * tree-vect-slp.c (vect_record_max_nunits): Replace vec_info
10090 and gimple stmt arguments with a stmt_vec_info.
10091 (vect_build_slp_tree_1): Remove vinfo argument and update call
10092 to vect_record_max_nunits.
10093 (vect_build_slp_tree_2): Update calls to vect_build_slp_tree_1
10094 and vect_record_max_nunits.
10096 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
10098 * tree-vectorizer.h (nested_in_vect_loop_p): Move further down
10099 file and take a stmt_vec_info instead of a gimple stmt.
10100 (supportable_widening_operation, vect_finish_replace_stmt)
10101 (vect_finish_stmt_generation, vect_get_store_rhs)
10102 (vect_get_vec_def_for_operand_1, vect_get_vec_def_for_operand)
10103 (vect_get_vec_defs, vect_init_vector, vect_transform_stmt)
10104 (vect_remove_stores, vect_analyze_stmt, vectorizable_condition)
10105 (vect_get_smallest_scalar_type, vect_check_gather_scatter)
10106 (vect_create_data_ref_ptr, bump_vector_ptr)
10107 (vect_permute_store_chain, vect_setup_realignment)
10108 (vect_transform_grouped_load, vect_record_grouped_load_vectors)
10109 (vect_create_addr_base_for_vector_ref, vectorizable_live_operation)
10110 (vectorizable_reduction, vectorizable_induction)
10111 (get_initial_def_for_reduction, is_simple_and_all_uses_invariant)
10112 (vect_get_place_in_interleaving_chain): Take stmt_vec_infos rather
10113 than gimple stmts as arguments.
10114 * tree-vect-data-refs.c (vect_get_smallest_scalar_type)
10115 (vect_preserves_scalar_order_p, vect_slp_analyze_node_dependences)
10116 (can_group_stmts_p, vect_check_gather_scatter)
10117 (vect_create_addr_base_for_vector_ref, vect_create_data_ref_ptr)
10118 (bump_vector_ptr, vect_permute_store_chain, vect_setup_realignment)
10119 (vect_permute_load_chain, vect_shift_permute_load_chain)
10120 (vect_transform_grouped_load)
10121 (vect_record_grouped_load_vectors): Likewise.
10122 * tree-vect-loop.c (vect_fixup_reduc_chain)
10123 (get_initial_def_for_reduction, vect_create_epilog_for_reduction)
10124 (vectorize_fold_left_reduction, is_nonwrapping_integer_induction)
10125 (vectorizable_reduction, vectorizable_induction)
10126 (vectorizable_live_operation, vect_loop_kill_debug_uses): Likewise.
10127 * tree-vect-patterns.c (type_conversion_p, adjust_bool_stmts)
10128 (vect_get_load_store_mask): Likewise.
10129 * tree-vect-slp.c (vect_get_place_in_interleaving_chain)
10130 (vect_analyze_slp_instance, vect_mask_constant_operand_p): Likewise.
10131 * tree-vect-stmts.c (vect_mark_relevant)
10132 (is_simple_and_all_uses_invariant)
10133 (exist_non_indexing_operands_for_use_p, process_use)
10134 (vect_init_vector_1, vect_init_vector, vect_get_vec_def_for_operand_1)
10135 (vect_get_vec_def_for_operand, vect_get_vec_defs)
10136 (vect_finish_stmt_generation_1, vect_finish_replace_stmt)
10137 (vect_finish_stmt_generation, vect_truncate_gather_scatter_offset)
10138 (compare_step_with_zero, vect_get_store_rhs, get_group_load_store_type)
10139 (get_negative_load_store_type, get_load_store_type)
10140 (vect_check_load_store_mask, vect_check_store_rhs)
10141 (vect_build_gather_load_calls, vect_get_strided_load_store_ops)
10142 (vectorizable_bswap, vectorizable_call, vectorizable_simd_clone_call)
10143 (vect_create_vectorized_demotion_stmts, vectorizable_conversion)
10144 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
10145 (get_group_alias_ptr_type, vectorizable_store, hoist_defs_of_uses)
10146 (vectorizable_load, vectorizable_condition, vectorizable_comparison)
10147 (vect_analyze_stmt, vect_transform_stmt, vect_remove_stores)
10148 (supportable_widening_operation): Likewise.
10150 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
10152 * tree-vect-data-refs.c (vect_describe_gather_scatter_call): Take
10153 a stmt_vec_info instead of a gcall.
10154 (vect_check_gather_scatter): Update call accordingly.
10155 * tree-vect-loop-manip.c (iv_phi_p): Take a stmt_vec_info instead
10157 (vect_can_advance_ivs_p, vect_update_ivs_after_vectorizer)
10158 (slpeel_update_phi_nodes_for_loops):): Update calls accordingly.
10159 * tree-vect-loop.c (vect_transform_loop_stmt): Take a stmt_vec_info
10160 instead of a gimple stmt.
10161 (vect_transform_loop): Update calls accordingly.
10162 * tree-vect-slp.c (vect_split_slp_store_group): Take and return
10163 stmt_vec_infos instead of gimple stmts.
10164 (vect_analyze_slp_instance): Update use accordingly.
10165 * tree-vect-stmts.c (read_vector_array, write_vector_array)
10166 (vect_clobber_variable, vect_stmt_relevant_p, permute_vec_elements)
10167 (vect_use_strided_gather_scatters_p, vect_build_all_ones_mask)
10168 (vect_build_zero_merge_argument, vect_get_gather_scatter_ops)
10169 (vect_gen_widened_results_half, vect_get_loop_based_defs)
10170 (vect_create_vectorized_promotion_stmts, can_vectorize_live_stmts):
10171 Take a stmt_vec_info instead of a gimple stmt and pass stmt_vec_infos
10172 down to subroutines.
10174 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
10176 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Change the type
10177 of the worklist from a vector of gimple stmts to a vector of
10179 * tree-vect-stmts.c (vect_mark_relevant, process_use)
10180 (vect_mark_stmts_to_be_vectorized): Likewise
10182 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
10184 * tree-vect-loop.c (vect_analyze_loop_operations): Look up the
10185 statement before passing it to vect_analyze_stmt.
10186 (vect_create_epilog_for_reduction): Use a stmt_vec_info to walk
10187 the chain of phi vector definitions. Track the exit phi via its
10189 (vectorizable_reduction): Set cond_stmt_vinfo directly from the
10190 STMT_VINFO_REDUC_DEF.
10191 * tree-vect-slp.c (vect_get_place_in_interleaving_chain): Use
10192 stmt_vec_infos to handle the statement chains.
10193 (vect_get_slp_defs): Record the first statement in the node
10194 using a stmt_vec_info.
10195 * tree-vect-stmts.c (vect_mark_stmts_to_be_vectorized): Look up
10196 statements here and pass their stmt_vec_info down to subroutines.
10197 (vect_init_vector_1): Hoist call to vinfo_for_stmt and pass it
10198 down to vect_finish_stmt_generation.
10199 (vect_init_vector, vect_get_vec_defs, vect_finish_replace_stmt)
10200 (vect_finish_stmt_generation): Call vinfo_for_stmt and pass
10201 stmt_vec_infos to subroutines.
10202 (vect_remove_stores): Use stmt_vec_infos to handle the statement
10205 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
10207 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences):
10208 (vect_check_gather_scatter, vect_create_data_ref_ptr, bump_vector_ptr)
10209 (vect_permute_store_chain, vect_setup_realignment)
10210 (vect_permute_load_chain, vect_shift_permute_load_chain)
10211 (vect_transform_grouped_load): Use stmt_vec_info rather than gimple
10212 stmts internally, and when passing values to other vectorizer routines.
10213 * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Likewise.
10214 * tree-vect-loop.c (vect_analyze_scalar_cycles_1)
10215 (vect_analyze_loop_operations, get_initial_def_for_reduction)
10216 (vect_create_epilog_for_reduction, vectorize_fold_left_reduction)
10217 (vectorizable_reduction, vectorizable_induction)
10218 (vectorizable_live_operation, vect_transform_loop_stmt)
10219 (vect_transform_loop): Likewise.
10220 * tree-vect-patterns.c (vect_reassociating_reduction_p)
10221 (vect_recog_widen_op_pattern, vect_recog_mixed_size_cond_pattern)
10222 (vect_recog_bool_pattern, vect_recog_gather_scatter_pattern): Likewise.
10223 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
10224 (vect_slp_analyze_node_operations_1): Likewise.
10225 * tree-vect-stmts.c (vect_mark_relevant, process_use)
10226 (exist_non_indexing_operands_for_use_p, vect_init_vector_1)
10227 (vect_mark_stmts_to_be_vectorized, vect_get_vec_def_for_operand)
10228 (vect_finish_stmt_generation_1, get_group_load_store_type)
10229 (get_load_store_type, vect_build_gather_load_calls)
10230 (vectorizable_bswap, vectorizable_call, vectorizable_simd_clone_call)
10231 (vect_create_vectorized_demotion_stmts, vectorizable_conversion)
10232 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
10233 (vectorizable_store, vectorizable_load, vectorizable_condition)
10234 (vectorizable_comparison, vect_analyze_stmt, vect_transform_stmt)
10235 (supportable_widening_operation): Likewise.
10236 (vect_get_vector_types_for_stmt): Likewise.
10237 * tree-vectorizer.h (vect_dr_behavior): Likewise.
10239 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
10241 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence)
10242 (vect_slp_analyze_node_dependences, vect_analyze_data_ref_accesses)
10243 (vect_permute_store_chain, vect_permute_load_chain)
10244 (vect_shift_permute_load_chain, vect_transform_grouped_load): Avoid
10245 repeated stmt_vec_info lookups.
10246 * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Likewise.
10247 (vect_update_ivs_after_vectorizer): Likewise.
10248 * tree-vect-loop.c (vect_is_simple_reduction): Likewise.
10249 (vect_create_epilog_for_reduction, vectorizable_reduction): Likewise.
10250 * tree-vect-patterns.c (adjust_bool_stmts): Likewise.
10251 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
10252 (vect_bb_slp_scalar_cost): Likewise.
10253 * tree-vect-stmts.c (get_group_alias_ptr_type): Likewise.
10255 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
10257 * tree-vect-data-refs.c (vect_check_gather_scatter): Pass the
10258 gcall rather than the generic gimple stmt to gimple_call_internal_fn.
10259 (vect_get_smallest_scalar_type, can_group_stmts_p): Use dyn_cast
10260 to get gassigns and gcalls, rather than operating on generc gimple
10262 * tree-vect-stmts.c (exist_non_indexing_operands_for_use_p)
10263 (vect_mark_stmts_to_be_vectorized, vectorizable_store)
10264 (vectorizable_load, vect_analyze_stmt): Likewise.
10265 * tree-vect-loop.c (vectorizable_reduction): Likewise gphi.
10267 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
10269 * tree-vectorizer.h (get_earlier_stmt, get_later_stmt): Take and
10270 return stmt_vec_infos rather than gimple stmts. Do not accept
10272 (vect_find_last_scalar_stmt_in_slp): Return a stmt_vec_info instead
10274 * tree-vect-slp.c (vect_find_last_scalar_stmt_in_slp): Likewise.
10275 Update use of get_later_stmt.
10276 (vect_get_constant_vectors): Update call accordingly.
10277 (vect_schedule_slp_instance): Likewise
10278 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences): Likewise.
10279 (vect_slp_analyze_instance_dependence): Likewise.
10280 (vect_preserves_scalar_order_p): Update use of get_earlier_stmt.
10282 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
10284 * tree-vectorizer.h (stmt_info_for_cost::stmt): Replace with...
10285 (stmt_info_for_cost::stmt_info): ...this new field.
10286 (add_stmt_costs): Update accordingly.
10287 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost)
10288 (vect_get_known_peeling_cost): Likewise.
10289 (vect_estimate_min_profitable_iters): Likewise.
10290 * tree-vect-stmts.c (record_stmt_cost): Likewise.
10292 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
10294 * tree-vectorizer.h (_loop_vec_info::may_misalign_stmts): Change
10295 from an auto_vec<gimple *> to an auto_vec<stmt_vec_info>.
10296 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Update
10298 * tree-vect-loop-manip.c (vect_create_cond_for_align_checks): Likewise.
10300 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
10302 * tree-vectorizer.h (_stmt_vec_info::same_dr_stmt): Change from
10303 a gimple stmt to a stmt_vec_info.
10304 * tree-vect-stmts.c (vectorizable_load): Update accordingly.
10306 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
10308 * tree-vectorizer.h (vec_info::grouped_stores): Change from
10309 an auto_vec<gimple *> to an auto_vec<stmt_vec_info>.
10310 (_loop_vec_info::reduction_chains): Likewise.
10311 * tree-vect-loop.c (vect_fixup_scalar_cycles_with_patterns): Update
10313 * tree-vect-slp.c (vect_analyze_slp): Likewise.
10315 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
10317 * tree-vectorizer.h (_stmt_vec_info::first_element): Change from
10318 a gimple stmt to a stmt_vec_info.
10319 (_stmt_vec_info::next_element): Likewise.
10320 * tree-vect-data-refs.c (vect_update_misalignment_for_peel)
10321 (vect_slp_analyze_and_verify_node_alignment)
10322 (vect_analyze_group_access_1, vect_analyze_group_access)
10323 (vect_small_gap_p, vect_prune_runtime_alias_test_list)
10324 (vect_create_data_ref_ptr, vect_record_grouped_load_vectors)
10325 (vect_supportable_dr_alignment): Update accordingly.
10326 * tree-vect-loop.c (vect_fixup_reduc_chain): Likewise.
10327 (vect_fixup_scalar_cycles_with_patterns, vect_is_slp_reduction)
10328 (vect_is_simple_reduction, vectorizable_reduction): Likewise.
10329 * tree-vect-patterns.c (vect_reassociating_reduction_p): Likewise.
10330 * tree-vect-slp.c (vect_build_slp_tree_1)
10331 (vect_attempt_slp_rearrange_stmts, vect_supported_load_permutation_p)
10332 (vect_split_slp_store_group, vect_analyze_slp_instance)
10333 (vect_analyze_slp, vect_transform_slp_perm_load): Likewise.
10334 * tree-vect-stmts.c (vect_model_store_cost, vect_model_load_cost)
10335 (get_group_load_store_type, get_load_store_type)
10336 (get_group_alias_ptr_type, vectorizable_store, vectorizable_load)
10337 (vect_transform_stmt, vect_remove_stores): Likewise.
10339 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
10341 * tree-vectorizer.h (vect_dr_stmt): Return a stmt_vec_info rather
10342 than a gimple stmt.
10343 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence)
10344 (vect_slp_analyze_data_ref_dependence, vect_record_base_alignments)
10345 (vect_calculate_target_alignmentm, vect_compute_data_ref_alignment)
10346 (vect_update_misalignment_for_peel, vect_verify_datarefs_alignment)
10347 (vector_alignment_reachable_p, vect_get_data_access_cost)
10348 (vect_get_peeling_costs_all_drs, vect_peeling_hash_get_lowest_cost)
10349 (vect_peeling_supportable, vect_enhance_data_refs_alignment)
10350 (vect_find_same_alignment_drs, vect_analyze_data_refs_alignment)
10351 (vect_analyze_group_access_1, vect_analyze_group_access)
10352 (vect_analyze_data_ref_access, vect_analyze_data_ref_accesses)
10353 (vect_vfa_access_size, vect_small_gap_p, vect_analyze_data_refs)
10354 (vect_supportable_dr_alignment): Remove vinfo_for_stmt from the
10355 result of vect_dr_stmt and use the stmt_vec_info instead of
10356 the associated gimple stmt.
10357 * tree-vect-loop-manip.c (get_misalign_in_elems): Likewise.
10358 (vect_gen_prolog_loop_niters): Likewise.
10359 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
10361 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
10363 * tree-vectorizer.h (_slp_tree::stmts): Change from a vec<gimple *>
10364 to a vec<stmt_vec_info>.
10365 * tree-vect-slp.c (vect_free_slp_tree): Update accordingly.
10366 (vect_create_new_slp_node): Take a vec<gimple *> instead of a
10367 vec<stmt_vec_info>.
10368 (_slp_oprnd_info::def_stmts): Change from a vec<gimple *>
10369 to a vec<stmt_vec_info>.
10370 (bst_traits::value_type, bst_traits::value_type): Likewise.
10371 (bst_traits::hash): Update accordingly.
10372 (vect_get_and_check_slp_defs): Change the stmts parameter from
10373 a vec<gimple *> to a vec<stmt_vec_info>.
10374 (vect_two_operations_perm_ok_p, vect_build_slp_tree_1): Likewise.
10375 (vect_build_slp_tree): Likewise.
10376 (vect_build_slp_tree_2): Likewise. Update uses of
10377 SLP_TREE_SCALAR_STMTS.
10378 (vect_print_slp_tree): Update uses of SLP_TREE_SCALAR_STMTS.
10379 (vect_mark_slp_stmts, vect_mark_slp_stmts_relevant)
10380 (vect_slp_rearrange_stmts, vect_attempt_slp_rearrange_stmts)
10381 (vect_supported_load_permutation_p, vect_find_last_scalar_stmt_in_slp)
10382 (vect_detect_hybrid_slp_stmts, vect_slp_analyze_node_operations_1)
10383 (vect_slp_analyze_node_operations, vect_slp_analyze_operations)
10384 (vect_bb_slp_scalar_cost, vect_slp_analyze_bb_1)
10385 (vect_get_constant_vectors, vect_get_slp_defs)
10386 (vect_transform_slp_perm_load, vect_schedule_slp_instance)
10387 (vect_remove_slp_scalar_calls, vect_schedule_slp): Likewise.
10388 (vect_analyze_slp_instance): Build up a vec of stmt_vec_infos
10389 instead of gimple stmts.
10390 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences): Change
10391 the stores parameter for a vec<gimple *> to a vec<stmt_vec_info>.
10392 (vect_slp_analyze_instance_dependence): Update uses of
10393 SLP_TREE_SCALAR_STMTS.
10394 (vect_slp_analyze_and_verify_node_alignment): Likewise.
10395 (vect_slp_analyze_and_verify_instance_alignment): Likewise.
10396 * tree-vect-loop.c (neutral_op_for_slp_reduction): Likewise.
10397 (get_initial_defs_for_reduction): Likewise.
10398 (vect_create_epilog_for_reduction): Likewise.
10399 (vectorize_fold_left_reduction): Likewise.
10400 * tree-vect-stmts.c (vect_prologue_cost_for_slp_op): Likewise.
10401 (vect_model_simple_cost, vectorizable_shift, vectorizable_load)
10402 (can_vectorize_live_stmts): Likewise.
10404 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
10406 * tree-vectorizer.h (_loop_vec_info::reductions): Change from an
10407 auto_vec<gimple *> to an auto_vec<stmt_vec_info>.
10408 (vect_force_simple_reduction): Take and return stmt_vec_infos rather
10410 * tree-parloops.c (valid_reduction_p): Take a stmt_vec_info instead
10412 (gather_scalar_reductions): Update after above interface changes.
10413 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Likewise.
10414 (vect_is_simple_reduction): Take and return stmt_vec_infos rather
10416 (vect_force_simple_reduction): Likewise.
10417 * tree-vect-patterns.c (vect_pattern_recog_1): Update use of
10418 LOOP_VINFO_REDUCTIONS.
10419 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
10421 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
10423 * tree-vectorizer.h (_stmt_vec_info::reduc_def): Change from
10424 a gimple stmt to a stmt_vec_info.
10425 * tree-vect-loop.c (vect_active_double_reduction_p)
10426 (vect_force_simple_reduction, vectorizable_reduction): Update
10429 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
10431 * tree-vectorizer.h (_slp_tree::vec_stmts): Change from a
10432 vec<gimple *> to a vec<stmt_vec_info>.
10433 * tree-vect-loop.c (vect_create_epilog_for_reduction): Change
10434 the reduction_phis argument from a vec<gimple *> to a
10435 vec<stmt_vec_info>.
10436 (vectorizable_reduction): Likewise the phis local variable that
10437 is passed to vect_create_epilog_for_reduction. Update for new type
10438 of SLP_TREE_VEC_STMTS.
10439 (vectorizable_induction): Update for new type of SLP_TREE_VEC_STMTS.
10440 (vectorizable_live_operation): Likewise.
10441 * tree-vect-slp.c (vect_get_slp_vect_defs): Likewise.
10442 (vect_transform_slp_perm_load, vect_schedule_slp_instance): Likewise.
10444 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
10446 * tree-vectorizer.h (_stmt_vec_info::vectorized_stmt): Change from
10447 a gimple stmt to a stmt_vec_info.
10448 (vectorizable_condition, vectorizable_live_operation)
10449 (vectorizable_reduction, vectorizable_induction): Pass back the
10450 vectorized statement as a stmt_vec_info.
10451 * tree-vect-data-refs.c (vect_record_grouped_load_vectors): Update
10452 use of STMT_VINFO_VEC_STMT.
10453 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise,
10454 accumulating the inner phis that feed the STMT_VINFO_VEC_STMT
10455 as stmt_vec_infos rather than gimple stmts.
10456 (vectorize_fold_left_reduction): Change vec_stmt from a gimple stmt
10457 to a stmt_vec_info.
10458 (vectorizable_live_operation): Likewise.
10459 (vectorizable_reduction, vectorizable_induction): Likewise,
10460 updating use of STMT_VINFO_VEC_STMT.
10461 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Update use
10462 of STMT_VINFO_VEC_STMT.
10463 (vect_build_gather_load_calls, vectorizable_bswap, vectorizable_call)
10464 (vectorizable_simd_clone_call, vectorizable_conversion)
10465 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
10466 (vectorizable_store, vectorizable_load, vectorizable_condition)
10467 (vectorizable_comparison, can_vectorize_live_stmts): Change vec_stmt
10468 from a gimple stmt to a stmt_vec_info.
10469 (vect_transform_stmt): Update use of STMT_VINFO_VEC_STMT. Pass a
10470 pointer to a stmt_vec_info to the vectorizable_* routines.
10472 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
10474 * tree-vectorizer.h (_stmt_vec_info::related_stmt): Change from
10475 a gimple stmt to a stmt_vec_info.
10476 (is_pattern_stmt_p): Update accordingly.
10477 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Likewise.
10478 (vect_record_grouped_load_vectors): Likewise.
10479 * tree-vect-loop.c (vect_determine_vf_for_stmt): Likewise.
10480 (vect_fixup_reduc_chain, vect_update_vf_for_slp): Likewise.
10481 (vect_model_reduction_cost): Likewise.
10482 (vect_create_epilog_for_reduction): Likewise.
10483 (vectorizable_reduction, vectorizable_induction): Likewise.
10484 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
10485 Return the stmt_vec_info for the pattern statement.
10486 (vect_set_pattern_stmt): Update use of STMT_VINFO_RELATED_STMT.
10487 (vect_split_statement, vect_mark_pattern_stmts): Likewise.
10488 * tree-vect-slp.c (vect_detect_hybrid_slp_stmts): Likewise.
10489 (vect_detect_hybrid_slp, vect_get_slp_defs): Likewise.
10490 * tree-vect-stmts.c (vect_mark_relevant): Likewise.
10491 (vect_get_vec_def_for_operand_1, vectorizable_call): Likewise.
10492 (vectorizable_simd_clone_call, vect_analyze_stmt, new_stmt_vec_info)
10493 (free_stmt_vec_info, vect_is_simple_use): Likewise.
10495 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
10497 * tree-vectorizer.h (vect_finish_replace_stmt): Return a stmt_vec_info
10498 (vect_finish_stmt_generation): Likewise.
10499 * tree-vect-stmts.c (vect_finish_stmt_generation_1): Likewise.
10500 (vect_finish_replace_stmt, vect_finish_stmt_generation): Likewise.
10501 (vect_build_gather_load_calls): Use the return value of the above
10502 functions instead of a separate call to vinfo_for_stmt. Use narrow
10503 scopes for the input gimple stmt and wider scopes for the associated
10504 stmt_vec_info. Use vec_info::lookup_def when setting these
10505 stmt_vec_infos from an SSA_NAME definition.
10506 (vectorizable_bswap, vectorizable_call, vectorizable_simd_clone_call)
10507 (vect_create_vectorized_demotion_stmts, vectorizable_conversion)
10508 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
10509 (vectorizable_store, vectorizable_load, vectorizable_condition)
10510 (vectorizable_comparison): Likewise.
10511 * tree-vect-loop.c (vectorize_fold_left_reduction): Likewise.
10512 (vectorizable_reduction): Likewise.
10514 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
10516 * tree-vectorizer.h (vect_is_simple_use): Add an optional
10517 stmt_vec_info * parameter before the optional gimple **.
10518 * tree-vect-stmts.c (vect_is_simple_use): Likewise.
10519 (process_use, vect_get_vec_def_for_operand_1): Update callers.
10520 (vect_get_vec_def_for_operand, vectorizable_shift): Likewise.
10521 * tree-vect-loop.c (vectorizable_reduction): Likewise.
10522 (vectorizable_live_operation): Likewise.
10523 * tree-vect-patterns.c (type_conversion_p): Likewise.
10524 (vect_look_through_possible_promotion): Likewise.
10525 (vect_recog_rotate_pattern): Likewise.
10526 * tree-vect-slp.c (vect_get_and_check_slp_defs): Likewise.
10528 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
10530 * tree-vectorizer.h (stmt_vec_info): Temporarily change from
10531 a typedef to a wrapper class.
10532 (NULL_STMT_VEC_INFO): New macro.
10533 (vec_info::stmt_infos): Change to vec<stmt_vec_info>.
10534 (stmt_vec_info::operator*): New function.
10535 (stmt_vec_info::operator gimple *): Likewise.
10536 (set_vinfo_for_stmt): Use NULL_STMT_VEC_INFO.
10537 (add_stmt_costs): Likewise.
10538 * tree-vect-loop-manip.c (iv_phi_p): Likewise.
10539 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost)
10540 (vect_get_known_peeling_cost): Likewise.
10541 (vect_estimate_min_profitable_iters): Likewise.
10542 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
10543 * tree-vect-slp.c (vect_remove_slp_scalar_calls): Likewise.
10544 * tree-vect-stmts.c (vect_build_gather_load_calls): Likewise.
10545 (vectorizable_store, free_stmt_vec_infos): Likewise.
10546 (new_stmt_vec_info): Change return type of xcalloc to
10549 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
10551 * tree-vectorizer.h (vec_info::lookup_single_use): Declare.
10552 * tree-vectorizer.c (vec_info::lookup_single_use): New function.
10553 * tree-vect-loop.c (vectorizable_reduction): Use it instead of
10554 a single_imm_use-based sequence.
10555 * tree-vect-stmts.c (supportable_widening_operation): Likewise.
10557 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
10559 * tree-vectorizer.h (vec_info::lookup_def): Declare.
10560 * tree-vectorizer.c (vec_info::lookup_def): New function.
10561 * tree-vect-patterns.c (vect_get_internal_def): Use it.
10562 (vect_widened_op_tree): Likewise.
10563 * tree-vect-stmts.c (vect_is_simple_use): Likewise.
10564 * tree-vect-loop.c (vect_analyze_loop_operations): Likewise.
10565 (vectorizable_reduction): Likewise.
10566 (vect_valid_reduction_input_p): Take a stmt_vec_info instead
10568 (vect_is_slp_reduction): Update calls accordingly. Use
10569 vec_info::lookup_def.
10570 (vect_is_simple_reduction): Likewise
10571 * tree-vect-slp.c (vect_detect_hybrid_slp_1): Use vec_info::lookup_def.
10573 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
10575 * tree-vectorizer.h (vec_info::lookup_stmt): Declare.
10576 * tree-vectorizer.c (vec_info::lookup_stmt): New function.
10577 * tree-vect-loop.c (vect_determine_vf_for_stmt): Use it instead
10579 (vect_determine_vectorization_factor, vect_analyze_scalar_cycles_1)
10580 (vect_compute_single_scalar_iteration_cost, vect_analyze_loop_form)
10581 (vect_update_vf_for_slp, vect_analyze_loop_operations)
10582 (vect_is_slp_reduction, vectorizable_induction)
10583 (vect_transform_loop_stmt, vect_transform_loop): Likewise.
10584 * tree-vect-patterns.c (vect_init_pattern_stmt):
10585 (vect_determine_min_output_precision_1, vect_determine_precisions)
10586 (vect_pattern_recog): Likewise.
10587 * tree-vect-stmts.c (vect_analyze_stmt, vect_transform_stmt): Likewise.
10588 * config/powerpcspe/powerpcspe.c (rs6000_density_test): Likewise.
10589 * config/rs6000/rs6000.c (rs6000_density_test): Likewise.
10590 * tree-vect-slp.c (vect_detect_hybrid_slp_stmts): Likewise.
10591 (vect_detect_hybrid_slp_1, vect_detect_hybrid_slp_2)
10592 (vect_detect_hybrid_slp): Likewise. Change the walk_stmt_info
10593 info field from a loop to a loop_vec_info.
10595 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
10597 * tree-vectorizer.h (stmt_vec_info): Move typedef earlier in file.
10598 (vec_info::add_stmt): Declare.
10599 * tree-vectorizer.c (vec_info::add_stmt): New function.
10600 * tree-vect-data-refs.c (vect_create_data_ref_ptr): Use it.
10601 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Likewise.
10602 (vect_create_epilog_for_reduction, vectorizable_reduction): Likewise.
10603 (vectorizable_induction): Likewise.
10604 * tree-vect-slp.c (_bb_vec_info::_bb_vec_info): Likewise.
10605 * tree-vect-stmts.c (vect_finish_stmt_generation_1): Likewise.
10606 (vectorizable_simd_clone_call, vectorizable_store): Likewise.
10607 (vectorizable_load): Likewise.
10608 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
10609 (vect_recog_bool_pattern, vect_recog_mask_conversion_pattern)
10610 (vect_recog_gather_scatter_pattern): Likewise.
10611 (append_pattern_def_seq): Likewise. Remove a check that is
10612 performed by add_stmt itself.
10614 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
10616 * tree-vect-loop.c (vectorizable_reduction): Fix an instance in
10617 which make_ssa_name was called with new_stmt before new_stmt
10620 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
10622 * tree-vect-loop.c (vect_valid_reduction_input_p): New function,
10624 (vect_is_slp_reduction): ...here...
10625 (vect_is_simple_reduction): ...and here. Remove repetition of tests
10626 that are already known to be false.
10628 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
10630 * tree-vectorizer.h (vect_free_slp_instance): Add a final_p parameter.
10631 * tree-vect-slp.c (vect_free_slp_tree): Likewise. Don't update
10632 STMT_VINFO_NUM_SLP_USES when it's true.
10633 (vect_free_slp_instance): Add a final_p parameter and pass it to
10634 vect_free_slp_tree.
10635 (vect_build_slp_tree_2): Update call to vect_free_slp_instance.
10636 (vect_analyze_slp_instance): Likewise.
10637 (vect_slp_analyze_operations): Likewise.
10638 (vect_slp_analyze_bb_1): Likewise.
10639 * tree-vectorizer.c (vec_info): Likewise.
10640 * tree-vect-loop.c (vect_transform_loop): Likewise.
10642 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
10644 * tree-vect-loop.c (vectorizable_reduction): Assert that the
10645 function is not called for second and subsequent members of
10648 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
10650 * tree-vect-loop.c (get_initial_def_for_reduction): Move special
10651 cases for nested loops from here to ...
10652 (vect_create_epilog_for_reduction): ...here. Only call
10653 vect_is_simple_use for inner-loop reductions.
10655 2018-07-31 Martin Liska <mliska@suse.cz>
10657 PR gcov-profile/85338
10658 PR gcov-profile/85350
10659 PR gcov-profile/85372
10660 * profile.c (struct location_triplet): New.
10661 (struct location_triplet_hash): Likewise.
10662 (output_location): Do not output a BB that
10663 is already recorded for a line.
10664 (branch_prob): Use streamed_locations.
10666 2018-07-31 Martin Liska <mliska@suse.cz>
10668 PR gcov-profile/85370
10669 * coverage.c (coverage_begin_function): Do not mark target
10670 clones as artificial functions.
10672 2018-07-31 Martin Liska <mliska@suse.cz>
10674 PR gcov-profile/83813
10675 PR gcov-profile/84758
10676 PR gcov-profile/85217
10677 PR gcov-profile/85332
10678 * profile.c (branch_prob): Do not record GOTO expressions
10679 for GIMPLE statements which locations are already streamed.
10681 2018-07-31 Olivier Hainque <hainque@adacore.com>
10683 * gcc.c (handle_spec_function): Accept a soft_matched_part
10684 argument, as do_spec_1. Pass it down to ...
10685 (eval_spec_function): Accept a soft_matched_part argument,
10686 and pass it down to ...
10687 (do_spec_2): Accept a soft_matched_part argument, and pass
10688 it down to do_spec_1.
10689 (do_spec_1): Pass soft_matched_part to handle_spec_function.
10690 (handle_braces): Update call to handle_spec_function.
10691 (driver::set_up_specs): Update calls to do_spec_2.
10692 (compare_debug_dump_opt_spec_function): Likewise.
10693 (compare_debug_self_opt_spec_function): Likewise.
10695 2018-07-31 Olivier Hainque <hainque@adacore.com>
10697 * common.opt (nolibc): New option.
10698 * doc/invoke.texi (Link Options): Document it.
10699 * gcc.c (LINK_GCC_C_SEQUENCE_SPEC): Honor nolibc.
10700 * config/alpha/linux.h: Likewise.
10701 * config/arc/elf.h: Likewise.
10702 * config/arm/uclinux-elf.h: Likewise.
10703 * config/arm/unknown-elf.h: Likewise.
10704 * config/avr/avrlibc.h: Likewise.
10705 * config/bfin/bfin.h: Likewise.
10706 * config/bfin/linux.h: Likewise.
10707 * config/bfin/uclinux.h: Likewise.
10708 * config/darwin.h: Likewise.
10709 * config/darwin10.h: Likewise.
10710 * config/darwin12.h: Likewise.
10711 * config/gnu-user.h: Likewise.
10712 * config/lm32/uclinux-elf.h: Likewise.
10713 * config/pa/pa-hpux11.h: Likewise.
10714 * config/pa/pa64-hpux.h: Likewise.
10715 * config/sparc/sparc.h: Likewise.
10717 2018-07-31 Olivier Hainque <hainque@adacore.com>
10719 * gcc.c (getenv_spec_function): Prepend '/' to value for allowed
10720 undefined variables.
10722 2018-07-30 Segher Boessenkool <segher@kernel.crashing.org>
10725 * config/arm/arm.c (arm_block_set_aligned_vect): Use gen_int_mode
10726 instead of GEN_INT.
10728 2018-07-30 Bernd Edlinger <bernd.edlinger@hotmail.de>
10730 * tree-ssa-forwprop.c (simplify_builtin_call): Don't create a not NUL
10731 terminated string literal.
10733 2018-07-30 Segher Boessenkool <segher@kernel.crashing.org>
10735 PR rtl-optimization/85160
10736 * combine.c (is_just_move): New function.
10737 (try_combine): Allow combining two instructions into two if neither of
10738 the original instructions was a move.
10740 2018-07-30 Alexander Monakov <amonakov@ispras.ru>
10743 * doc/extend.texi (Global Register Variables): Discourage use of type
10745 (Local Register Variables): Likewise.
10747 2018-07-30 Richard Sandiford <richard.sandiford@arm.com>
10749 PR tree-optimization/86506
10750 * hwint.h (ceil_log2): Resync with hwint.c implementation.
10752 2018-07-30 Ilya Leoshkevich <iii@linux.ibm.com>
10755 * lra-constraints.c (spill_hard_reg_in_range): When selecting the
10756 hard_regno, make sure no insn between `from` and `to` clobbers it.
10758 2018-07-30 Cesar Philippidis <cesar@codesourcery.com>
10759 Tom de Vries <tdevries@suse.de>
10761 * config/nvptx/nvptx.c (PTX_GANG_DEFAULT): Rename to ...
10762 (PTX_DEFAULT_RUNTIME_DIM): ... this.
10763 (nvptx_goacc_validate_dims): Set default worker and gang dims to
10764 PTX_DEFAULT_RUNTIME_DIM.
10765 (nvptx_dim_limit): Ignore GOMP_DIM_WORKER.
10767 2018-07-29 John David Anglin <danglin@gcc.gnu.org>
10769 * config/pa/pa.c (pa_output_addr_vec): Align address table.
10770 * config/pa/pa.h (JUMP_TABLES_IN_TEXT_SECTION): Revise comment.
10771 * config/pa/pa32-linux.h (JUMP_TABLES_IN_TEXT_SECTION): Define.
10773 2018-07-27 Michael Meissner <meissner@linux.ibm.com>
10775 * config/rs6000/constraints.md (wG constraint): Delete, no longer
10777 * config/rs6000/predicates.md (p9_fusion_reg_operand): Rename
10778 predicate to reflect toc fusion has been deleted.
10779 (toc_fusion_mem_raw): Delete, no longer used.
10780 (toc_fusion_mem_wrapped): Likewise.
10781 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete toc
10783 * config/rs6000/rs6000-protos.h (fusion_wrap_memory_address):
10784 Delete, no longer used.
10785 * config/rs6000/rs6000.c (struct rs6000_reg_addr): Delete fields
10786 meant to be used for toc fusion.
10787 (rs6000_debug_print_mode): Delete toc fusion debugging.
10788 (rs6000_debug_reg_global): Likewise.
10789 (rs6000_init_hard_regno_mode_ok): Delete setting up fields for toc
10790 fusion and secondary reload support that were never used.
10791 (rs6000_option_override_internal): Delete TOC fusion, that was only
10792 partially defined, and it did not work unless you also used the
10794 (rs6000_legitimate_address_p): Delete TOC fusion support.
10795 (rs6000_opt_masks): Likewise.
10796 (fusion_wrap_memory_address): Delete function, no longer used.
10797 (fusion_split_address); Delete TOC fusion support.
10798 * config/rs6000/rs6000.h (TARGET_TOC_FUSION_INT): Delete, no
10799 longer used with toc fusion being deleted.
10800 (TARGET_TOC_FUSION_FP): Likewise.
10801 * config/rs6000/rs6000.md (UNSPEC_FUSION_ADDIS): Delete TOC fusion
10803 (toc fusion spliter): Delete TOC fusion support.
10804 (toc_fusionload_<mode>): Likewise.
10805 (toc_fusionload_di): Likewise.
10806 (fusion_gpr_load_<mode>): Delete generator function, this insn no
10807 longer needs to be named. Rename predicate to delete TOC fusion.
10808 (fusion_gpr_<P:mode>_<GPR_FUSION:mode>_load): Likewise.
10809 (fusion_gpr_<P:mode>_<GPR_FUSION:mode>_store): Likewise.
10810 (fusion_vsx_<P:mode>_<GPR_FUSION:mode>_load): Likewise.
10811 (fusion_vsx_<P:mode>_<GPR_FUSION:mode>_store): Likewise.
10812 (p9 fusion peephole2s): Rename predicate to delete TOC fusion.
10814 2018-07-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
10816 * doc/extend.texi (Basic PowerPC Built-in Functions Available on
10817 ISA 2.05): Replace __uint128_t with __uint128 and __int128_t with
10818 __int128 in built-in function prototypes.
10819 (PowerPC AltiVec Built-in Functions on ISA 2.07): Likewise.
10820 (PowerPC AltiVec Built-in Functions on ISA 3.0): Likewise.
10822 2018-07-27 Martin Sebor <msebor@redhat.com>
10824 PR tree-optimization/86696
10825 * tree-ssa-strlen.c (get_min_string_length): Handle all integer
10826 types, including enums.
10827 (handle_char_store): Be prepared for the above function to fail.
10829 2018-07-26 Qing Zhao <qing.zhao@oracle.com>
10831 * builtins.c (inline_expand_builtin_string_cmp): Disable inlining
10832 when optimization level is lower than 2 or optimize for size.
10834 2018-07-26 Martin Sebor <msebor@redhat.com>
10836 PR tree-optimization/86043
10837 PR tree-optimization/86042
10838 * tree-ssa-strlen.c (handle_builtin_memcpy): Handle strict overlaps.
10839 (get_string_cst_length): Rename...
10840 (get_min_string_length): ...to this. Add argument.
10841 (handle_char_store): Extend to handle multi-character stores by
10843 * tree.c (initializer_zerop): Use new argument. Handle MEM_REF.
10844 * tree.h (initializer_zerop): Add argument.
10846 2018-07-26 Jakub Jelinek <jakub@redhat.com>
10848 PR middle-end/86660
10849 * omp-low.c (scan_sharing_clauses): Don't ignore map clauses for
10850 declare target to variables if they have always,{to,from,tofrom} map
10853 2018-07-26 Martin Liska <mliska@suse.cz>
10856 * lto-wrapper.c: Add linker_output as prefix
10857 for ltrans_output_file.
10859 2018-07-26 Segher Boessenkool <segher@kernel.crashing.org>
10861 PR rtl-optimization/85805
10862 * combine.c (reg_nonzero_bits_for_combine): Only use the last set
10863 value for hard registers if that was written in the same mode.
10865 2018-07-26 Martin Liska <mliska@suse.cz>
10867 PR gcov-profile/86536
10868 * gcov.c (format_gcov): Use printf format %.*f directly
10869 and do not handle special values.
10871 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
10873 * common/config/arc/arc-common.c (arc_option_optimization_table):
10874 Update default optimizations for size.
10876 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
10878 * config/arc/arc.md (movsf_insn): Add short instruction selection.
10879 * config/arc/constraints.md (CfZ): New constraint.
10880 * config/arc/fpu.md (addssf3_fpu): Use CfZ constraint.
10881 (subsf3_fpu): Likewise.
10882 (cmpsf_fpu): Likewise.
10883 (cmpsf_fpu_uneq): Likewise.
10885 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
10887 * config/arc/arc.c (compact_memory_operand_p): Check for uncached
10889 (arc_is_uncached_mem_p): uncached applies to both the variable and
10892 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
10894 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Add additional
10897 2018-07-25 David Malcolm <dmalcolm@redhat.com>
10899 * optinfo-emit-json.cc (class optrecord_json_writer): Convert
10900 field "m_scopes" from vec to auto_vec.
10902 2018-07-25 Martin Liska <mliska@suse.cz>
10904 * config/powerpcspe/powerpcspe-protos.h (rs6000_loop_align): Fix
10907 2018-07-25 Richard Biener <rguenther@suse.de>
10910 * dwarf2out.c (dwarf2out_decl): Do not handle nested functions
10911 special wrt context_die late.
10912 (gen_subprogram_die): Re-use DIEs in local scope.
10914 2018-07-25 Richard Sandiford <richard.sandiford@arm.com>
10916 PR tree-optimization/86644
10917 * hwint.c (ceil_log2): Fix comment. Return 0 for 0.
10919 2018-07-25 Martin Liska <mliska@suse.cz>
10921 PR middle-end/86645
10922 * dumpfile.c: And excluded values with TDF_ALL_VALUES.
10923 * dumpfile.h (enum dump_flag): Defince TDF_ALL_VALUES.
10925 2018-07-25 Martin Liska <mliska@suse.cz>
10928 * params.def: Explain ASan abbreviation and provide
10929 a documentation link.
10931 2018-07-24 Martin Sebor <msebor@redhat.com>
10933 PR tree-optimization/86622
10934 PR tree-optimization/86532
10935 * builtins.h (string_length): Declare.
10936 * builtins.c (c_strlen): Correct handling of non-constant offsets.
10937 (check_access): Be prepared for non-constant length ranges.
10938 (string_length): Make extern.
10939 * expr.c (string_constant): Only handle the minor non-constant
10940 array index. Use string_constant to compute the length of
10941 a generic string constant.
10943 2018-07-24 Richard Sandiford <richard.sandiford@arm.com>
10945 PR tree-optimization/86618
10946 * tree-vect-stmts.c (vectorizable_call): Don't take the address
10947 of LOOP_VINFO_MASKS (loop_vinfo) when loop_vinfo is null.
10949 2018-07-24 David Malcolm <dmalcolm@redhat.com>
10951 PR tree-optimization/86636
10952 * json.cc (json::object::set): Fix comment. Add assertions.
10953 (json::array::append): Move here from json.h. Add comment and an
10955 (json::string::string): Likewise.
10956 * json.h (json::array::append): Move to json.cc.
10957 (json::string::string): Likewise.
10958 * optinfo-emit-json.cc
10959 (optrecord_json_writer::impl_location_to_json): Assert that we
10960 aren't attempting to write out UNKNOWN_LOCATION, or an ad-hoc
10961 wrapper around it. Expand the location once, rather than three
10963 (optrecord_json_writer::inlining_chain_to_json): Fix the check for
10964 UNKNOWN_LOCATION, to use LOCATION_LOCUS to look through ad-hoc
10966 (optrecord_json_writer::optinfo_to_json): Likewise, in four
10967 places. Fix some overlong lines.
10969 2018-07-24 Matthew Malcomson <matthew.malcomson@arm.com>
10971 * config/aarch64/aarch64-simd.md
10972 (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w<mode>): Split into...
10973 (aarch64_<ANY_EXTEND:su>subw<mode>): ... This...
10974 (aarch64_<ANY_EXTEND:su>addw<mode>): ... And this.
10975 (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w<mode>_internal): Split into...
10976 (aarch64_<ANY_EXTEND:su>subw<mode>_internal): ... This...
10977 (aarch64_<ANY_EXTEND:su>addw<mode>_internal): ... And this.
10978 (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w2<mode>_internal): Split into...
10979 (aarch64_<ANY_EXTEND:su>subw2<mode>_internal): ... This...
10980 (aarch64_<ANY_EXTEND:su>addw2<mode>_internal): ... And this.
10982 2018-07-24 Jakub Jelinek <jakub@redhat.com>
10984 PR middle-end/86627
10985 * expmed.c (expand_divmod): Punt if d == HOST_WIDE_INT_MIN
10986 and size > HOST_BITS_PER_WIDE_INT. For size > HOST_BITS_PER_WIDE_INT
10987 and abs_d == d, do the power of two handling if profitable.
10989 2018-07-24 Richard Biener <rguenther@suse.de>
10991 * match.pd: Add BIT_FIELD_REF canonicalizations.
10993 2018-07-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
10996 * genmatch.c (dt_operand::gen_match_op): Avoid folding volatile values.
10998 2018-07-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
11000 * gimple-fold.c (gimple_fold_builtin_printf): Don't create a not NUL
11001 terminated STRING_CST object.
11003 2018-07-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
11005 hsa-dump.c (dump_hsa_symbol): Avoid out of scope access to buf.
11007 2018-07-23 Segher Boessenkool <segher@kernel.crashing.org>
11009 * config/rs6000/rs6000-p8swap.c (rtx_is_swappable_p): Adjust.
11010 * config/rs6000/rs6000-protos.h (rs6000_split_v4si_init): Delete.
11011 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Always force
11012 the elements into a register.
11013 (rs6000_split_v4si_init_di_reg): Delete.
11014 (rs6000_split_v4si_init): Delete.
11015 * config/rs6000/vsx.md (unspec): Delete UNSPEC_VSX_VEC_INIT.
11016 (vsx_init_v4si): Rewrite as a define_expand.
11018 2018-07-23 Segher Boessenkool <segher@kernel.crashing.org>
11020 * config/rs6000/rs6000.md (splitters for rldimi and rlwimi with the
11021 zero_extend argument from memory): New.
11023 2018-07-22 Martin Sebor <msebor@redhat.com>
11026 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Avoid
11027 diagnosing calls with unknown arguments unless -Walloca-larger-than
11028 is restricted to less than PTRDIFF_MAX bytes.
11030 2018-07-22 Gerald Pfeifer <gerald@pfeifer.com>
11032 * doc/gcov.texi (Invoking Gcov): Editorial changes.
11034 2018-07-20 David Malcolm <dmalcolm@redhat.com>
11036 * pretty-print.c (text_info::set_location): Remove redundant
11037 "line_table" parameter from call to rich_location::set_range.
11039 2018-07-20 Martin Sebor <msebor@redhat.com>
11041 PR middle-end/82063
11042 * builtins.c (expand_builtin_alloca): Adjust.
11043 * calls.c (alloc_max_size): Simplify.
11044 * cgraphunit.c (cgraph_node::expand): Adjust.
11045 * common.opt (larger_than_size, warn_frame_larger_than): Remove
11047 (frame_larger_than_size): Same.
11048 (-Wframe-larger-than, -Wlarger-than, -Wstack-usage): Change options
11049 to take a HOST_WIDE_INT argument and accept a byte-size suffix.
11051 * doc/invoke.texi (GCC Command Options): Document option arguments.
11052 Explain byte-size arguments and suffixes.
11053 (-Wvla-larger-than, -Wno-alloc-size-larger-than): Update.
11054 (-Wno-alloca-larger-than, -Wno-vla-larger-than): Same.
11055 (-Wframe-larger-than, -Wlarger-than, -Wstack-usage): Same.
11056 * doc/options.texi (UInteger): Expand.
11057 (Host_Wide_Int, ByteSize): Document new properties.
11058 * final.c (final_start_function_1): Include sizes in an error message.
11059 * function.c (frame_offset_overflow): Same.
11060 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Adjust.
11061 (alloca_call_type_by_arg): Change function argument to HOST_WIDE_INT.
11062 Diagnose unbounded alloca calls only for limits of less than
11064 (alloca_call_type): Adjust. Diagnose possibly out-of-bounds alloca
11065 calls and VLA size only for limits of less than PTRDIFF_MAX. Same
11067 (pass_walloca::execute): Adjust. Diagnose alloca calls in loops
11068 only for limits of less than PTRDIFF_MAX.
11069 * langhooks-def.h (lhd_handle_option): Change function argument
11071 * langhooks.c (lhd_handle_option): Same.
11072 * langhooks.h (handle_option): Same.
11073 * opt-functions.awk (switch_bit_fields): Handle Host_Wide_Int and
11075 (var_type, var_type_struct): Same.
11076 (var_set): Handle ByteSize flag.
11077 * optc-gen.awk: Add comments to output to ease debugging. Make
11078 use of HOST_WIDE_INT where appropriate.
11079 * opts-gen-save.awk: Use %lx to format unsigned long.
11080 * opth-gen.awk: Change function argument to HOST_WIDE_INT.
11081 * opts-common.c (integral_argument): Return HOST_WIDE_INT and add
11082 arguments. Parse bytes-size suffixes.
11083 (enum_arg_to_value): Change function argument to HOST_WIDE_INT.
11084 (enum_value_to_arg): Same.
11085 (decode_cmdline_option): Handle cl_host_wide_int. Adjust.
11086 (handle_option): Adjust.
11087 (generate_option): Change function argument to HOST_WIDE_INT.
11088 (cmdline_handle_error): Adjust.
11089 (read_cmdline_option): Change function argument to HOST_WIDE_INT.
11090 (set_option): Change function argument to HOST_WIDE_INT.
11091 (option_enabled): Handle cl_host_wide_int.
11092 (get_option_state): Handle CLVC_SIZE.
11093 (control_warning_option): Same.
11094 * opts.c (common_handle_option): Change function argument to
11095 HOST_WIDE_INT. Remove handling of OPT_Walloca_larger_than_ and
11096 OPT_Wvla_larger_than_.
11097 * opts.h (enum cl_var_type): Add an enumerator.
11098 * stor-layout.c (layout_decl): Print a more meaningful warning.
11099 * toplev.c (output_stack_usage): Adjust.
11101 2018-07-20 Qing Zhao <qing.zhao@oracle.com>
11103 * builtins.c (expand_builtin_memcmp): Delete the last parameter for
11104 call to inline_expand_builtin_string_cmp.
11105 (expand_builtin_strcmp): Likewise.
11106 (expand_builtin_strncmp): Likewise.
11107 (inline_string_cmp): Delete the last parameter, change char_type_node
11108 to unsigned_char_type_node for strcmp/strncmp, add conversions to the
11110 (inline_expand_builtin_string_cmp): Delete the last parameter, give up
11111 the inlining expansion on target where the type of the call has same or
11112 narrower precision than unsigned char.
11114 2018-07-20 David Malcolm <dmalcolm@redhat.com>
11116 * Makefile.in (OBJS): Add json.o and optinfo-emit-json.o.
11117 (CFLAGS-optinfo-emit-json.o): Define TARGET_NAME.
11118 * common.opt (fsave-optimization-record): New option.
11119 * coretypes.h (struct kv_pair): Move here from dumpfile.c.
11120 * doc/invoke.texi (-fsave-optimization-record): New option.
11121 * dumpfile.c: Include "optinfo-emit-json.h".
11122 (struct kv_pair): Move to coretypes.h.
11123 (optgroup_options): Make non-static.
11124 (dump_context::end_scope): Call
11125 optimization_records_maybe_pop_dump_scope.
11126 * dumpfile.h (optgroup_options): New decl.
11127 * json.cc: New file.
11128 * json.h: New file.
11129 * optinfo-emit-json.cc: New file.
11130 * optinfo-emit-json.h: New file.
11131 * optinfo.cc: Include "optinfo-emit-json.h".
11132 (optinfo::emit): Call optimization_records_maybe_record_optinfo.
11133 (optinfo_enabled_p): Check optimization_records_enabled_p.
11134 (optinfo_wants_inlining_info_p): Likewise.
11135 * optinfo.h: Update comment.
11136 * profile-count.c (profile_quality_as_string): New function.
11137 * profile-count.h (profile_quality_as_string): New decl.
11138 (profile_count::quality): New accessor.
11139 * selftest-run-tests.c (selftest::run_tests): Call json_cc_tests
11140 and optinfo_emit_json_cc_tests.
11141 * selftest.h (selftest::json_cc_tests): New decl.
11142 (selftest::optinfo_emit_json_cc_tests): New decl.
11143 * toplev.c: Include "optinfo-emit-json.h".
11144 (compile_file): Call optimization_records_finish.
11145 (do_compile): Call optimization_records_start.
11146 * tree-ssa-live.c: Include optinfo.h.
11147 (remove_unused_scope_block_p): Retain inlining information if
11148 optinfo_wants_inlining_info_p returns true.
11150 2018-07-20 Richard Biener <rguenther@suse.de>
11153 * dwarf2out.c (dwarf2out_die_ref_for_decl): Test in_lto_p
11154 to cover -flto-partition=none.
11156 2018-07-20 Martin Liska <mliska@suse.cz>
11158 * tree.h (DECL_LOCATION_RANGE): Remove unused macro.
11159 (get_decl_source_range): Remove unused function.
11161 2018-07-20 Richard Biener <rguenther@suse.de>
11163 * tree-ssa-sccvn.h (struct vn_nary_op_s): Add next member.
11164 (struct vn_phi_s): Likewise.
11165 (struct vn_reference_s): Likewise.
11166 * tree-ssa-sccvn.c (vn_nary_op_hasher::equal): Add shortcut
11167 for searching the slot of an entry known to be in the hash itself.
11168 (vn_phi_hasher::equal): Likewise.
11169 (vn_reference_hasher::equal): Likewise.
11170 (last_inserted_ref, last_inserted_phi, last_inserted_nary): New
11172 (optimistic_info, current_info): Remove, keeping only valid_info.
11173 (vn_reference_lookup_1): Remove fallback lookup.
11174 (vn_reference_lookup_2): Likewise.
11175 (vn_nary_op_lookup_1): Likewise.
11176 (vn_phi_lookup): Likewise.
11177 (vn_nary_build_or_lookup_1): Make sure to not chain the built
11179 (vn_reference_insert): Adjust, chain the inserted hash element
11180 at last_inserted_ref.
11181 (vn_reference_insert_pieces): Likewise.
11182 (visit_reference_op_call): Likewise.
11183 (vn_nary_op_insert_into): Chain the inserted hash element at
11184 last_inserted_nary.
11185 (vn_nary_op_insert_pieces): Adjust.
11186 (vn_nary_op_insert): Likewise.
11187 (vn_nary_op_insert_stmt): Likewise.
11188 (vn_phi_insert): Adjust, chain the inserted hash element at
11190 (process_scc): Remove clearing and copying the optimistic
11191 table. Instead remove elements inserted during an optimistic
11192 iteration from the single table we maintain.
11193 (init_scc_vn): Adjust.
11194 (free_scc_vn): Likewise.
11195 (sccvn_dom_walker::record_cond): Likewise.
11196 (sccvn_dom_walker::after_dom_children): Likewise.
11198 2018-07-19 Martin Sebor <msebor@redhat.com>
11200 PR tree-optimization/84047
11201 PR tree-optimization/83776
11202 * tree-vrp.c (vrp_prop::check_mem_ref): New function.
11203 (check_array_bounds): Call it.
11205 2018-07-19 Martin Sebor <msebor@redhat.com>
11207 * align.h (align_flags): Use member initialization.
11209 2018-07-19 David Malcolm <dmalcolm@redhat.com>
11211 * Makefile.in (OBJS): Add optinfo.o.
11212 * coretypes.h (class symtab_node): New forward decl.
11213 (struct cgraph_node): New forward decl.
11214 (class varpool_node): New forward decl.
11215 * dump-context.h: New file.
11216 * dumpfile.c: Include "optinfo.h", "dump-context.h", "cgraph.h",
11218 (refresh_dumps_are_enabled): Use optinfo_enabled_p.
11219 (set_dump_file): Call dumpfile_ensure_any_optinfo_are_flushed.
11220 (set_alt_dump_file): Likewise.
11221 (dump_context::~dump_context): New dtor.
11222 (dump_gimple_stmt): Move implementation to...
11223 (dump_context::dump_gimple_stmt): ...this new member function.
11224 Add the stmt to any pending optinfo, creating one if need be.
11225 (dump_gimple_stmt_loc): Move implementation to...
11226 (dump_context::dump_gimple_stmt_loc): ...this new member function.
11227 Start a new optinfo and add the stmt to it.
11228 (dump_gimple_expr): Move implementation to...
11229 (dump_context::dump_gimple_expr): ...this new member function.
11230 Add the stmt to any pending optinfo, creating one if need be.
11231 (dump_gimple_expr_loc): Move implementation to...
11232 (dump_context::dump_gimple_expr_loc): ...this new member function.
11233 Start a new optinfo and add the stmt to it.
11234 (dump_generic_expr): Move implementation to...
11235 (dump_context::dump_generic_expr): ...this new member function.
11236 Add the tree to any pending optinfo, creating one if need be.
11237 (dump_generic_expr_loc): Move implementation to...
11238 (dump_context::dump_generic_expr_loc): ...this new member
11239 function. Add the tree to any pending optinfo, creating one if
11241 (dump_printf): Move implementation to...
11242 (dump_context::dump_printf_va): ...this new member function. Add
11243 the text to any pending optinfo, creating one if need be.
11244 (dump_printf_loc): Move implementation to...
11245 (dump_context::dump_printf_loc_va): ...this new member function.
11246 Start a new optinfo and add the stmt to it.
11247 (dump_dec): Move implementation to...
11248 (dump_context::dump_dec): ...this new member function. Add the
11249 value to any pending optinfo, creating one if need be.
11250 (dump_context::dump_symtab_node): New member function.
11251 (dump_context::get_scope_depth): New member function.
11252 (dump_context::begin_scope): New member function.
11253 (dump_context::end_scope): New member function.
11254 (dump_context::ensure_pending_optinfo): New member function.
11255 (dump_context::begin_next_optinfo): New member function.
11256 (dump_context::end_any_optinfo): New member function.
11257 (dump_context::s_current): New global.
11258 (dump_context::s_default): New global.
11259 (dump_scope_depth): Delete global.
11260 (dumpfile_ensure_any_optinfo_are_flushed): New function.
11261 (dump_symtab_node): New function.
11262 (get_dump_scope_depth): Reimplement in terms of dump_context.
11263 (dump_begin_scope): Likewise.
11264 (dump_end_scope): Likewise.
11265 (selftest::temp_dump_context::temp_dump_context): New ctor.
11266 (selftest::temp_dump_context::~temp_dump_context): New dtor.
11267 (selftest::verify_item): New function.
11268 (ASSERT_IS_TEXT): New macro.
11269 (ASSERT_IS_TREE): New macro.
11270 (ASSERT_IS_GIMPLE): New macro.
11271 (selftest::test_capture_of_dump_calls): New test.
11272 (selftest::dumpfile_c_tests): Call it.
11273 * dumpfile.h (dump_printf, dump_printf_loc, dump_basic_block)
11274 (dump_generic_expr_loc, dump_generic_expr, dump_gimple_stmt_loc)
11275 (dump_gimple_stmt, dump_dec): Gather these related decls and add a
11276 descriptive comment.
11277 (dump_function, print_combine_total_stats, enable_rtl_dump_file)
11278 (dump_node, dump_bb): Move these unrelated decls.
11279 (class dump_manager): Add leading comment.
11280 * optinfo.cc: New file.
11281 * optinfo.h: New file.
11283 2018-07-19 Michael Collison <michael.collison@arm.com>
11284 Richard Henderson <rth@redhat.com>
11286 * config/aarch64/aarch64.md (subv<GPI>4, usubv<GPI>4): New patterns.
11287 (subti): Handle op1 zero.
11288 (subvti4, usub4ti4): New.
11289 (*sub<GPI>3_compare1_imm): New.
11290 (sub<GPI>3_carryinCV): New.
11291 (*sub<GPI>3_carryinCV_z1_z2, *sub<GPI>3_carryinCV_z1): New.
11292 (*sub<GPI>3_carryinCV_z2, *sub<GPI>3_carryinCV): New.
11294 2018-07-19 Michael Collison <michael.collison@arm.com>
11295 Richard Henderson <rth@redhat.com>
11297 * config/aarch64/aarch64.md: (addv<GPI>4, uaddv<GPI>4): New.
11298 (addti3): Create simpler code if low part is already known to be 0.
11299 (addvti4, uaddvti4): New.
11300 (*add<GPI>3_compareC_cconly_imm): New.
11301 (*add<GPI>3_compareC_cconly): New.
11302 (*add<GPI>3_compareC_imm): New.
11303 (*add<GPI>3_compareC): Rename from add<GPI>3_compare1; do not
11304 handle constants within this pattern..
11305 (*add<GPI>3_compareV_cconly_imm): New.
11306 (*add<GPI>3_compareV_cconly): New.
11307 (*add<GPI>3_compareV_imm): New.
11308 (add<GPI>3_compareV): New.
11309 (add<GPI>3_carryinC, add<GPI>3_carryinV): New.
11310 (*add<GPI>3_carryinC_zero, *add<GPI>3_carryinV_zero): New.
11311 (*add<GPI>3_carryinC, *add<GPI>3_carryinV): New.
11312 ((*add<GPI>3_compareC_cconly_imm): Replace 'ne' operator
11313 with 'comparison' operator.
11314 (*add<GPI>3_compareV_cconly_imm): Ditto.
11315 (*add<GPI>3_compareV_cconly): Ditto.
11316 (*add<GPI>3_compareV_imm): Ditto.
11317 (add<GPI>3_compareV): Ditto.
11318 (add<mode>3_carryinC): Ditto.
11319 (*add<mode>3_carryinC_zero): Ditto.
11320 (*add<mode>3_carryinC): Ditto.
11321 (add<mode>3_carryinV): Ditto.
11322 (*add<mode>3_carryinV_zero): Ditto.
11323 (*add<mode>3_carryinV): Ditto.
11325 2018-07-19 Michael Collison <michael.collison@arm.com>
11326 Richard Henderson <rth@redhat.com>
11328 * config/aarch64/aarch64-modes.def (CC_V): New.
11329 * config/aarch64/aarch64-protos.h
11330 (aarch64_addti_scratch_regs): Declare
11331 (aarch64_subvti_scratch_regs): Declare.
11332 (aarch64_expand_subvti): Declare.
11333 (aarch64_gen_unlikely_cbranch): Declare
11334 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Test
11335 for signed overflow using CC_Vmode.
11336 (aarch64_get_condition_code_1): Handle CC_Vmode.
11337 (aarch64_gen_unlikely_cbranch): New function.
11338 (aarch64_addti_scratch_regs): New function.
11339 (aarch64_subvti_scratch_regs): New function.
11340 (aarch64_expand_subvti): New function.
11342 2018-07-19 Andre Vieira <andre.simoesdiasvieira@arm.com>
11344 * config/aarch64/aarch64-option-extensions.def: New entry for profile
11346 * config/aarch64/aarch64.h (AARCH64_FL_PROFILE): New.
11347 * doc/invoke.texi (aarch64-feature-modifiers): New entry for profile
11350 2018-07-19 Andre Vieira <andre.simoesdiasvieira@arm.com>
11353 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Make
11354 address check not strict.
11356 2018-07-19 Andre Vieira <andre.simoesdiasvieira@arm.com>
11358 * config/aarch64/aarch64-simd.md (aarch64_simd_mov<VQ:mode>): Replace
11360 (store_pair_lanes<mode>): Likewise.
11361 * config/aarch64/aarch64-protos.h (aarch64_addr_query_type): Add new
11362 enum value 'ADDR_QUERY_LDP_STP_N'.
11363 * config/aarch64/aarch64.c (aarch64_addr_query_type): Likewise.
11364 (aarch64_print_address_internal): Add declaration.
11365 (aarch64_print_ldpstp_address): Remove.
11366 (aarch64_classify_address): Adapt mode for 'ADDR_QUERY_LDP_STP_N'.
11367 (aarch64_print_operand): Change printing of 'y'.
11368 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Use
11369 new enum value 'ADDR_QUERY_LDP_STP_N', don't hardcode mode and use
11370 'true' rather than '1'.
11371 * config/aarch64/constraints.md (Uml): Likewise.
11372 (Uml): Rename to Umn.
11375 2018-07-19 Richard Biener <rguenther@suse.de>
11377 * tree-ssa-sccvn.h (struct vn_phi_s): Make phiargs member
11379 * tree-ssa-sccvn.c: Remove alloc-pool.h use.
11380 (vn_phi_hasher): Derive from nofree_ptr_hash and remove remove method.
11381 (vn_reference_hasher): Likewise.
11382 (struct vn_tables_s): Remove obstack and alloc-pool members.
11383 (vn_tables_obstack, vn_tables_insert_obstack): New global obstacks.
11384 (vn_nary_build_or_lookup_1): Manually build in vn_tables_insert_obstack.
11385 (vn_reference_insert): Allocate from obstack instead of from alloc-pool.
11386 (vn_reference_insert_pieces): Likewise.
11387 (alloc_vn_nary_op_noinit): Adjust.
11388 (vn_nary_op_insert_stmt): Allocate phiargs in-place.
11389 (vn_phi_eq): Adjust.
11390 (shared_lookup_phiargs): Remove.
11391 (vn_phi_lookup): Allocate temporary vn_phi_s on the stack.
11392 (vn_phi_insert): Allocate from obstack instead of from alloc-pool.
11393 (visit_reference_op_call): Likewise.
11394 (copy_nary, copy_phi, copy_reference): Remove.
11395 (process_scc): Rewind the obstack when iterating. Do not
11396 copy the elements to valid_info but just move them from one
11397 hashtable to the other.
11398 (allocate_vn_table): Adjust.
11399 (free_vn_table): Likewise.
11400 (init_scc_vn): Likewise.
11401 (free_scc_vn): Likewise.
11403 2018-07-19 H.J. Lu <hongjiu.lu@intel.com>
11406 * config/i386/i386.c (rest_of_insert_endbranch): Lookup
11407 indirect_return as function type attribute.
11408 (ix86_attribute_table): Change indirect_return to function
11410 * doc/extend.texi: Update indirect_return attribute.
11412 2018-07-19 Aldy Hernandez <aldyh@redhat.com>
11414 * wide-int.h (widest2_int): New.
11415 * gimple-fold.c (arith_overflowed_p): Use it.
11416 * tree.h (widest2_int_cst): New.
11417 * tree-vrp.c (wide_int_binop_overflow): Rename from
11418 vrp_int_const_binop.
11419 Rewrite to work on trees.
11420 (extract_range_from_multiplicative_op_1): Abstract code to...
11421 (wide_int_range_min_max): ...here.
11422 (wide_int_range_cross_product): ...and here.
11423 (extract_range_from_binary_expr_1): Abstract overflow code to...
11424 (wide_int_range_mult_wrapping): ...here.
11425 * tree-vrp.h (wide_int_range_cross_product): New.
11426 (wide_int_range_mult_wrapping): New.
11428 2018-07-19 Andrew Senkevich <andrew.senkevich@intel.com>
11429 Julia Koval <julia.koval@intel.com>
11431 * config/i386/x86-tune-costs.h (skylake_memcpy,
11432 skylake_memset): Replace rep_prefix with unrolling for size 512.
11434 2018-07-18 Kugan Vivekanandarajah <kuganv@linaro.org>
11436 PR middle-end/86544
11437 * tree-ssa-phiopt.c (cond_removal_in_popcount_pattern): Handle
11438 comparision with EQ_EXPR in last stmt.
11440 2018-07-18 Kelvin Nilsen <kelvin@gcc.gnu.org>
11442 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Rename
11443 this subsection to "PowerPC AltiVec/VSX Built-in Functions".
11444 (PowerPC AltiVec/VSX Built-in Functions): New name for subsection
11445 previously known as "PowerPC AltiVec Built-in Functions". Move
11446 some material to new subsubsections "PowerPC AltiVec Built-in
11447 Functions on ISA 2.06" and "PowerPC AltiVec Built-in Functions on
11449 (PowerPC Altivec Built-in Functions on ISA 2.05): New subsubsection.
11450 (PowerPC Altivec Built-in Functions on ISA 2.06): Likewise.
11451 (PowerPC Altivec Built-in Functions on ISA 2.07): Likewise.
11452 (PowerPC Altivec Built-in Functions on ISA 3.0): Likewise.
11454 2018-07-18 Richard Biener <rguenther@suse.de>
11456 PR tree-optimization/86557
11457 * tree-vect-patterns.c (vect_recog_divmod_pattern): Also handle
11460 2018-07-18 Ilya Leoshkevich <iii@linux.ibm.com>
11462 * config/s390/s390.c (s390_function_profiler): Generate CFI.
11464 2018-07-17 Jeff Law <law@redhat.com>
11466 * config/arm/arm.c (get_label_padding): Update for recent
11467 changes to label_to_alignment.
11469 PR tree-optimization/86010
11470 * tree-ssa-dse.c (compute_trims): Fix typo/thinko.
11472 * config/mips/mips.c (vr4130_align_insns): Update for recent
11473 changes to label_to_alignment.
11475 * config/frv/frv.c (frv_label_align): Update for recent changes
11476 to label_to_alignment.
11478 * config/nios2/nios2.c (nios2_label_align): Update for recent
11479 changes which dropped ALIGN_LABELS_LOG.
11481 2018-07-17 Andreas Schwab <schwab@linux-m68k.org>
11483 * config/m68k/m68k.md (umulsi3_highpart+1, const_umulsi3_highpart)
11484 (smulsi3_highpart+1, const_smulsi3_highpart): Add CC_STATUS_INIT.
11486 2018-07-17 Claudiu Zissulescu <claziss@synopsys.com>
11488 * config/arc/arc.c (arc_label_align): Use align_labels instead of
11489 deprecated align_labels_log.
11491 2018-07-17 Richard Biener <rguenther@suse.de>
11494 * dwarf2out.c (init_sections_and_labels): Always generate
11495 a debug_line_str_section for early LTO debug.
11496 (dwarf2out_finish): Reset debug_line_str_hash output early.
11497 Bump counter for extra dwarf5 .debug_loc labels to not conflict
11499 (dwarf2out_early_finish): Output debug_line_str.
11501 2018-07-17 Robin Dapp <rdapp@linux.ibm.com>
11503 * config/s390/s390.c (preferred_la_operand_p): Do not use LA with
11504 index register on z196 or later.
11506 2018-07-17 Robin Dapp <rdapp@linux.ibm.com>
11508 * config/s390/s390.c (s390_default_align): Set default function
11510 (s390_override_options_after_change): Call s390_default align.
11511 (s390_option_override_internal): Call s390_default_align.
11512 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define.
11514 2018-07-17 Jakub Jelinek <jakub@redhat.com>
11516 PR middle-end/86542
11517 * omp-low.c (create_task_copyfn): Copy over also fields corresponding
11518 to _looptemp_ clauses, other than the first two.
11520 2018-07-17 Martin Liska <mliska@suse.cz>
11522 * opts.c: Do not enable OPT_falign_* for -Os.
11524 2018-07-17 Martin Liska <mliska@suse.cz>
11526 * align.h (MAX_CODE_ALIGN): New.
11527 (MAX_CODE_ALIGN_VALUE): New.
11528 * common/config/i386/i386-common.c (ix86_handle_option):
11529 (MAX_CODE_ALIGN): Moved to align.h.
11530 * final.c (MAX_CODE_ALIGN): Likewise.
11531 * opts.c (parse_and_check_align_values):
11532 (MAX_CODE_ALIGN): Likewise.
11533 (MAX_CODE_ALIGN_VALUE): Likewise.
11535 2018-07-17 Martin Liska <mliska@suse.cz>
11537 * config/i386/att.h (ASM_OUTPUT_ALIGN): Fix spacing
11538 in order to fulfil coding style.
11539 * config/i386/cygming.h (ASM_OUTPUT_ALIGN): Likewise.
11540 * config/i386/gas.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
11541 * config/i386/x86-64.h (ASM_OUTPUT_MAX_SKIP_PAD): Likewise.
11542 * config/iq2000/iq2000.h (ASM_OUTPUT_ALIGN): Likewise.
11543 * config/pa/pa.h (ASM_OUTPUT_ALIGN): Likewise.
11544 * config/sparc/sol2.h (ASM_OUTPUT_ALIGN_WITH_NOP): Likewise.
11545 * config/sparc/sparc.h (ASM_OUTPUT_ALIGN): Likewise.
11546 * config/visium/visium.h (ASM_OUTPUT_ALIGN): Likewise.
11547 (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
11549 2018-07-17 Martin Liska <mliska@suse.cz>
11551 * align.h: New file.
11552 * config/alpha/alpha.c (alpha_align_insns_1): Use align_functions
11554 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Use new return type
11555 align_flags of label_to_alignment.
11556 * config/m32r/m32r.h (LOOP_ALIGN): Wrap returned values into
11558 * config/m68k/m68k.c: Do not use removed align_labels_value and
11560 * config/nds32/nds32.h (JUMP_ALIGN): Wrap result into align_flags class.
11561 (LOOP_ALIGN): Likewise.
11562 (LABEL_ALIGN): Likewise.
11563 * config/powerpcspe/powerpcspe.c (TARGET_ASM_LOOP_ALIGN_MAX_SKIP):
11564 Remove not used macro.
11565 (rs6000_loop_align): Change return type to align_flags.
11566 (rs6000_loop_align_max_skip): Remove.
11567 * config/rs6000/rs6000-protos.h (rs6000_loop_align):
11568 Change return type to align_flags.
11569 * config/rs6000/rs6000.c (TARGET_ASM_LOOP_ALIGN_MAX_SKIP):
11570 Remove not used macro.
11571 (rs6000_loop_align): Change return type to align_flags.
11572 (rs6000_loop_align_max_skip): Remove.
11573 * config/rx/rx.h (JUMP_ALIGN): Wrap integer values
11574 * config/rx/rx-protos.h (rx_align_for_label): Make it
11576 * config/rx/rx.c (rx_align_for_label): Change return type
11578 (rx_max_skip_for_label): Remove TARGET_ASM_*_ALIGN_MAX_SKIP
11580 into align_flags class.
11581 (LABEL_ALIGN): Likewise.
11582 (LOOP_ALIGN): Likewise.
11583 * config/s390/s390.c (s390_label_align): Use align_flags
11585 (s390_asm_output_function_label): Likewise.
11586 * config/sh/sh.c (sh_override_options_after_change):
11587 Use align_flags class directly without macros.
11588 (find_barrier): Likewise.
11589 (barrier_align): Likewise.
11590 (sh_loop_align): Likewise.
11591 * config/spu/spu.c (spu_option_override):
11592 Use align_flags_tuple::get_value instead of removed macros.
11593 (spu_sched_init): Likewise.
11594 * config/spu/spu.h (GTY): Likewise.
11595 * config/visium/visium.c (visium_option_override):
11596 Set "8" as default secondary alignment.
11597 * config/visium/visium.h (SUBALIGN_LOG): Define to 3
11598 in order to guarantee secondary alignment of 8.
11599 * coretypes.h: Include align.h header file.
11600 * doc/tm.texi: Remove TARGET_ASM_JUMP_ALIGN_MAX_SKIP,
11601 TARGET_ASM_LOOP_ALIGN_MAX_SKIP, TARGET_ASM_LABEL_ALIGN_MAX_SKIP
11602 and TARGET_ASM_LABEL_ALIGN_AFTER_BARRIER_MAX_SKIP macros.
11603 * doc/tm.texi.in: Likewise.
11604 * final.c (struct label_alignment): Remove not used structure.
11605 (LABEL_ALIGN): Change type to align_flags.
11606 (LOOP_ALIGN): Likewise.
11607 (JUMP_ALIGN): Likewise.
11608 (default_loop_align_max_skip): Remove.
11609 (default_label_align_max_skip): Likewise.
11610 (default_jump_align_max_skip): Likewise.
11611 (default_label_align_after_barrier_max_skip):
11612 (LABEL_TO_ALIGNMENT): Change to access label_align vector.
11613 (LABEL_TO_MAX_SKIP): Remove.
11614 (label_to_alignment): Return align_flags type instead of integer.
11615 (label_to_max_skip): Remove.
11616 (align_fuzz): Use align_flags type.
11617 (compute_alignments): Use align_flags type and use align_flags::max
11618 to combine multiple alignments.
11619 (grow_label_align): Grow vec instead of C array.
11620 (update_alignments): Assign just LABEL_TO_ALIGNMENT.
11621 (shorten_branches): Use align_flags type and use align_flags::max
11622 to combine multiple alignments.
11623 (final_scan_insn_1): Remove usage of secondary alignment that comes
11624 from label alignment, but instead use proper secondary alignment
11625 which is computed in grow_label_align.
11626 * flags.h (struct align_flags_tuple): Move to align.h.
11627 (struct align_flags): Likewise.
11628 (state_align_loops): Rename to align_loops.
11629 (state_align_jumps): Rename to align_jumps.
11630 (state_align_labels): Rename to align_labels.
11631 (state_align_functions): Rename to align_functions.
11632 (align_loops_log): Remove.
11633 (align_jumps_log): Remove.
11634 (align_labels_log): Remove.
11635 (align_functions_log): Remove.
11636 (align_loops_max_skip): Remove.
11637 (align_jumps_max_skip): Remove.
11638 (align_labels_max_skip): Remove.
11639 (align_functions_max_skip): Remove.
11640 (align_loops_value): Remove.
11641 (align_jumps_value): Remove.
11642 (align_labels_value): Remove.
11643 (align_functions_value): Remove.
11644 * output.h (label_to_alignment): Change return type to align_flags.
11645 (label_to_max_skip): Remove.
11646 * target.def: Remove loop_align_max_skip, label_align_max_skip,
11647 jump_align_max_skip macros.
11648 * targhooks.h (default_loop_align_max_skip): Remove.
11649 (default_label_align_max_skip): Likewise.
11650 (default_jump_align_max_skip): Likewise.
11651 (default_label_align_after_barrier_max_skip): Remove.
11652 * toplev.c (read_log_maxskip): Use ::normalize function.
11653 (parse_N_M): Remove not used argument and also call ::normalize.
11654 (parse_alignment_opts): Do not pass unused arguments.
11655 * varasm.c (assemble_start_function): Use directly align_functions
11656 instead of removed macros.
11657 * system.h: Do not poison removed macros.
11659 2018-07-17 Jakub Jelinek <jakub@redhat.com>
11661 PR middle-end/86539
11662 * gimplify.c (gimplify_omp_for): Ensure taskloop firstprivatized init
11663 and cond temporaries don't have reference type if iterator has
11664 pointer type. For init use &for_pre_body instead of pre_p if
11665 for_pre_body is non-empty.
11667 2018-07-16 Segher Boessenkool <segher@kernel.crashing.org>
11669 * config/rs6000/rs6000.md (trunc<mode>sf2): Expand truncates of
11670 double-double modes to SFmode directly directly.
11671 (trunc<mode>sf2_fprs): Delete.
11673 2018-07-16 Segher Boessenkool <segher@kernel.crashing.org>
11675 * config/rs6000/rs6000.c (init_float128_ibm): Use the correct names
11676 for conversions between IFmode and the decimal floating point modes.
11677 (init_float128_ieee): Use the correct names for conversions between
11678 KFmode and the decimal floating point modes.
11680 2018-07-16 Segher Boessenkool <segher@kernel.crashing.org>
11682 * config/rs6000/rs6000.c (init_float128_ibm): Use more correct names
11683 for the conversions between TDmode and IFmode.
11684 (init_float128_ieee): Use more correct names for the conversions
11685 between TDmode and KFmode.
11687 2018-07-16 Jakub Jelinek <jakub@redhat.com>
11689 PR tree-optimization/86526
11690 * builtins.c (expand_builtin_memcmp): Formatting fixes.
11691 (inline_expand_builtin_string_cmp): Likewise.
11692 (inline_string_cmp): Likewise. Use c_readstr instead of
11693 builtin_memcpy_read_str. Add unit_mode temporary.
11695 2018-07-16 Bernd Edlinger <bernd.edlinger@hotmail.de>
11697 PR middle-end/86528
11698 * builtins.c (check_access): Bail out if range[0] is no INTEGER_CST.
11699 * expr.c (string_constant): Fix the element size of ARRAY_TYPE.
11701 2018-07-16 Kelvin Nilsen <kelvin@gcc.gnu.org>
11703 * doc/extend.texi (PowerPC AltiVec Built-in Functions):
11704 Alphabetize prototypes of built-in functions, separating out
11705 built-in functions that are listed in this section but should be
11706 described elsewhere.
11708 2018-07-16 Uros Bizjak <ubizjak@gmail.com>
11711 * expmed.c (emit_store_flag): Do not emit setcc followed by a
11712 conditional move when trapping comparison was split to a
11713 non-trapping one (and vice versa).
11715 2018-07-16 Ilya Leoshkevich <iii@linux.ibm.com>
11717 * config/s390/s390.c (s390_function_profiler): Generate nops
11718 instead of profiler call sequences.
11719 * config/s390/s390.opt: Add the new option.
11721 2018-07-16 Ilya Leoshkevich <iii@linux.ibm.com>
11723 * config/s390/s390.c (s390_function_profiler): Generate
11724 __mcount_loc section.
11725 * config/s390/s390.opt: Add the new option.
11727 2018-07-16 Ilya Leoshkevich <iii@linux.ibm.com>
11729 * common.opt: Add the new warning.
11730 * config/s390/s390.c (s390_function_profiler): Emit "brasl
11731 %r0,__fentry__" when -mfentry is specified.
11732 (s390_option_override_internal): Disallow -mfentry for 31-bit
11734 * config/s390/s390.opt: Add the new option.
11736 2018-07-16 Richard Biener <rguenther@suse.de>
11739 * dwarf2out.c (dwarf2out_register_external_die): Assign DIE parents
11740 for function-local FUNCTION_DECL and RESULT_DECL immediately.
11742 2018-07-16 Martin Liska <mliska@suse.cz>
11745 * ipa-pure-const.c (malloc_candidate_p): Revert ::get
11748 2017-07-16 Claudiu Zissulescu <claziss@synopsys.com>
11750 * config/arc/arcHS.md: Update ARCHS scheduling rules.
11752 2017-07-16 Claudiu Zissulescu <claziss@synopsys.com>
11754 * config/arc/arc-arch.h (arc_tune_attr): Add new tune parameters
11756 * config/arc/arc-cpus.def (hs4x): New cpu.
11758 * config/arc/arc-tables.opt: Regenerate.
11759 * config/arc/arc.c (arc_sched_issue_rate): New function.
11760 (TARGET_SCHED_ISSUE_RATE): Define.
11761 (TARGET_SCHED_EXPOSED_PIPELINE): Likewise.
11762 * config/arc/arc.md (attr type): Add fpu_fuse, fpu_sdiv, fpu_ddiv,
11764 (attr tune): Add ARCHS4x tune values.
11765 (attr tune_dspmpy): Define.
11766 (*tst): Correct instruction type.
11767 * config/arc/arcHS.md: Don't use this automaton for ARCHS4x cpus.
11768 * config/arc/arcHS4x.md: New file.
11769 * config/arc/fpu.md: Update instruction type attributes.
11770 * config/arc/t-multilib: Regenerate.
11772 2018-07-16 Tom de Vries <tdevries@suse.de>
11775 * var-tracking.c (vt_initialize): Fix pre_dec handling.
11777 2018-07-16 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
11779 * config/aarch64/atomics.md (aarch64_store_execlusive<mode>): Add
11782 2018-07-16 Eric Botcazou <ebotcazou@adacore.com>
11784 PR tree-optimization/86514
11785 * tree-ssa-reassoc.c (init_range_entry) <CASE_CONVERT>: Return for a
11786 conversion to a boolean type from a type with greater precision.
11788 2018-07-16 Tom de Vries <tdevries@suse.de>
11790 * var-tracking.c (vt_initialize): Print adjusted insn slim if
11791 dump_flags request TDF_SLIM.
11793 2018-07-16 Aldy Hernandez <aldyh@redhat.com>
11795 * fold-const.c (int_const_binop_1): Abstract...
11796 (wide_int_binop): ...wide int code here.
11797 (poly_int_binop): ...poly int code here.
11798 Abstract the rest of int_const_binop_1 into int_const_binop.
11799 * fold-const.h (wide_int_binop): New.
11800 * tree-vrp.c (vrp_int_const_binop): Call wide_int_binop.
11801 Remove useless PLUS/MINUS_EXPR case.
11802 (zero_nonzero_bits_from_vr): Move wide int code...
11803 (zero_nonzero_bits_from_bounds): ...here.
11804 (extract_range_from_binary_expr_1): Move mask optimization code...
11805 (range_easy_mask_min_max): ...here.
11806 * tree-vrp.h (zero_nonzero_bits_from_bounds): New.
11807 (range_easy_mask_min_max): New.
11809 2018-07-15 Jeff Law <law@redhat.com>
11812 * config/sh/sh.c (output_mi_thunk): Remove dead conditional
11815 2018-07-14 Jim Wilson <jimw@sifive.com>
11817 * config/riscv/linux.h (TARGET_ASM_FILE_END): New.
11819 2018-07-14 Paul Koning <ni1d@arrl.net>
11821 * config/pdp11/pdp11.c (pdp11_rtx_costs): Bugfixes.
11823 2018-07-13 Jan Hubicka <hubicka@ucw.cz>
11825 * lto-streamer-out.c (copy_function_or_variable): Dump info about
11828 2018-07-13 Bill Schmidt <wschmidt@linux.ibm.com>
11829 Steve Munroe <munroesj52@gmail.com>
11831 * config/rs6000/emmintrin.h (_mm_and_si128): New function.
11832 (_mm_andnot_si128): Likewise.
11833 (_mm_or_si128): Likewise.
11834 (_mm_xor_si128): Likewise.
11836 2018-07-13 Qing Zhao <qing.zhao@oracle.com>
11838 PR middle-end/78809
11839 * builtins.c (expand_builtin_memcmp): Inline the calls first
11840 when result_eq is false.
11841 (expand_builtin_strcmp): Inline the calls first.
11842 (expand_builtin_strncmp): Likewise.
11843 (inline_string_cmp): New routine. Expand a string compare
11844 call by using a sequence of char comparison.
11845 (inline_expand_builtin_string_cmp): New routine. Inline expansion
11846 a call to str(n)cmp/memcmp.
11847 * doc/invoke.texi (--param builtin-string-cmp-inline-length):
11849 * params.def (BUILTIN_STRING_CMP_INLINE_LENGTH): New.
11851 2018-07-13 Richard Earnshaw <rearnsha@arm.com>
11853 * config/arm/driver-arm.c: Include arm-native.h.
11854 (host_detect_local_cpu): Use auto-generated data tables.
11855 (vendors, arm_cpu_table): Delete. Move part information to ...
11856 * config/arm/arm-cpus.in: ... here.
11857 * config/arm/parsecpu.awk (gen_native): New function.
11858 (vendor, part): New CPU fields.
11859 (END): Add support for building the native CPU detection tables.
11860 * config/arm/t-arm (arm-native.h): Add build rule.
11861 (driver-arm.o): Add dependency on arm-native.h.
11863 2018-07-13 Richard Biener <rguenther@suse.de>
11865 PR middle-end/85974
11866 * match.pd (addr1 - addr2): Allow either of the operand to
11869 2018-07-13 Tom de Vries <tdevries@suse.de>
11871 * tree-inline.c (remap_ssa_name): Save and reuse debug exprs generated
11874 2018-07-13 Jackson Woodruff <jackson.woodruff@arm.com>
11876 * config/aarch64/aarch64.c (aarch64_operands_adjust_ok_for_ldpstp): Use
11877 arrays instead of numbered variables.
11879 2018-07-13 Eric Botcazou <ebotcazou@adacore.com>
11881 * config/sparc/sparc-protos.h (sparc_compute_frame_size): Delete.
11882 * config/sparc/sparc.c (sparc_compute_frame_size): Make static.
11884 2018-07-13 Richard Biener <rguenther@suse.de>
11887 * dwarf2out.c (gen_type_die_with_usage): Use scope_die_for
11888 instead of get_context_die.
11890 2018-07-13 Kugan Vivekanandarajah <kuganv@linaro.org>
11891 Richard Biener <rguenther@suse.de>
11893 PR middle-end/86489
11894 * tree-ssa-loop-niter.c (number_of_iterations_popcount): Check
11895 that the loop latch destination where phi is defined.
11897 2018-07-12 Kito Cheng <kito.cheng@gmail.com>
11899 * config/riscv/riscv.c (enum riscv_privilege_levels): Add UNKNOWN_MODE.
11900 (riscv_expand_epilogue): Add assertion to check interrupt mode.
11901 (riscv_set_current_function): Extract getting interrupt type to new
11903 (riscv_get_interrupt_type): New function.
11904 (riscv_merge_decl_attributes): New function, checking interrupt type is
11906 (TARGET_MERGE_DECL_ATTRIBUTES): Define.
11908 2018-07-12 Paul Koning <ni1d@arrl.net>
11910 * config/pdp11/pdp11.c (pdp11_output_def): Fix typo in .set
11913 2018-07-12 Paul Koning <ni1d@arrl.net>
11915 * doc/rtl.texi (REG_NONNEG): Remove decrement and branch until
11916 zero reference, add doloop_end instead.
11917 * doc/md.texi (decrement_and_branch_until_zero): Remove.
11918 (Looping patterns): Remove decrement_and_branch_until_zero. Add
11919 detail for doloop_end.
11921 2018-07-12 Martin Sebor <msebor@redhat.com>
11924 * attribs.c (decl_attributes): Reject conflicting attributes before
11925 calling attribute handlers.
11927 2018-07-12 Jan Hubicka <hubicka@ucw.cz>
11929 * dumpfile.c (gcc::dump_manager::get_dump_file_name): Add PART
11931 (gcc::dump_manager::get_dump_file_name): likewise.
11932 (dump_begin): Likewise.
11933 * dumpfile.h (dump_begin): Update prototype.
11934 (gcc::dump_manager::get_dump_file_name,
11935 gcc::dump_manager::get_dump_file_name): Update prototype.
11937 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
11939 * internal-fn.h (vectorizable_internal_fn_p): New function.
11940 * tree-vect-slp.c (compatible_calls_p): Likewise.
11941 (vect_build_slp_tree_1): Remove nops argument. Handle calls
11942 to internal functions.
11943 (vect_build_slp_tree_2): Update call to vect_build_slp_tree_1.
11945 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
11947 * fold-const.h (inverse_conditions_p): Declare.
11948 * fold-const.c (inverse_conditions_p): New function.
11949 * match.pd: Use inverse_conditions_p. Add folds of view_converts
11950 that test the inverse condition of a conditional internal function.
11951 * internal-fn.h (vectorized_internal_fn_supported_p): Declare.
11952 * internal-fn.c (internal_fn_mask_index): Handle conditional
11953 internal functions.
11954 (vectorized_internal_fn_supported_p): New function.
11955 * tree-if-conv.c: Include internal-fn.h and fold-const.h.
11956 (any_pred_load_store): Replace with...
11957 (need_to_predicate): ...this new variable.
11958 (redundant_ssa_names): New variable.
11959 (ifcvt_can_use_mask_load_store): Move initial checks to...
11960 (ifcvt_can_predicate): ...this new function. Handle tree codes
11961 for which a conditional internal function exists.
11962 (if_convertible_gimple_assign_stmt_p): Use ifcvt_can_predicate
11963 instead of ifcvt_can_use_mask_load_store. Update after variable
11965 (predicate_load_or_store): New function, split out from
11966 predicate_mem_writes.
11967 (check_redundant_cond_expr): New function.
11968 (value_available_p): Likewise.
11969 (predicate_rhs_code): Likewise.
11970 (predicate_mem_writes): Rename to...
11971 (predicate_statements): ...this. Use predicate_load_or_store
11972 and predicate_rhs_code.
11973 (combine_blocks, tree_if_conversion): Update after above name changes.
11974 (ifcvt_local_dce): Handle redundant_ssa_names.
11975 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern): Handle
11976 general conditional functions.
11977 * tree-vect-stmts.c (vectorizable_call): Likewise.
11979 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
11980 Alan Hayward <alan.hayward@arm.com>
11981 David Sherwood <david.sherwood@arm.com>
11983 * internal-fn.h (can_interpret_as_conditional_op_p): Declare.
11984 * internal-fn.c (can_interpret_as_conditional_op_p): New function.
11985 * tree-ssa-math-opts.c (convert_mult_to_fma_1): Handle conditional
11986 plus and minus and convert them into IFN_COND_FMA-based sequences.
11987 (convert_mult_to_fma): Handle conditional plus and minus.
11989 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
11991 * doc/md.texi (cond_fma, cond_fms, cond_fnma, cond_fnms): Document.
11992 * optabs.def (cond_fma_optab, cond_fms_optab, cond_fnma_optab)
11993 (cond_fnms_optab): New optabs.
11994 * internal-fn.def (COND_FMA, COND_FMS, COND_FNMA, COND_FNMS): New
11995 internal functions.
11996 (FMA): Use DEF_INTERNAL_FLT_FN rather than DEF_INTERNAL_FLT_FLOATN_FN.
11997 * internal-fn.h (get_conditional_internal_fn): Declare.
11998 (get_unconditional_internal_fn): Likewise.
11999 * internal-fn.c (cond_ternary_direct): New macro.
12000 (expand_cond_ternary_optab_fn): Likewise.
12001 (direct_cond_ternary_optab_supported_p): Likewise.
12002 (FOR_EACH_COND_FN_PAIR): Likewise.
12003 (get_conditional_internal_fn): New function.
12004 (get_unconditional_internal_fn): Likewise.
12005 * gimple-match.h (gimple_match_op::MAX_NUM_OPS): Bump to 5.
12006 (gimple_match_op::gimple_match_op): Add a new overload for 5
12008 (gimple_match_op::set_op): Likewise.
12009 (gimple_resimplify5): Declare.
12010 * genmatch.c (decision_tree::gen): Generate simplifications for
12012 * gimple-match-head.c (gimple_simplify): Define an overload for
12013 5 operands. Handle calls with 5 arguments in the top-level overload.
12014 (convert_conditional_op): Handle conversions from unconditional
12015 internal functions to conditional ones.
12016 (gimple_resimplify5): New function.
12017 (build_call_internal): Pass a fifth operand.
12018 (maybe_push_res_to_seq): Likewise.
12019 (try_conditional_simplification): Try converting conditional
12020 internal functions to unconditional internal functions.
12021 Handle 3-operand unconditional forms.
12022 * match.pd (UNCOND_TERNARY, COND_TERNARY): Operator lists.
12023 Define ternary equivalents of the current rules for binary conditional
12024 internal functions.
12025 * config/aarch64/aarch64.c (aarch64_preferred_else_value): Handle
12026 ternary operations.
12027 * config/aarch64/iterators.md (UNSPEC_COND_FMLA, UNSPEC_COND_FMLS)
12028 (UNSPEC_COND_FNMLA, UNSPEC_COND_FNMLS): New unspecs.
12029 (optab): Handle them.
12030 (SVE_COND_FP_TERNARY): New int iterator.
12031 (sve_fmla_op, sve_fmad_op): New int attributes.
12032 * config/aarch64/aarch64-sve.md (cond_<optab><mode>)
12033 (*cond_<optab><mode>_2, *cond_<optab><mode_4)
12034 (*cond_<optab><mode>_any): New SVE_COND_FP_TERNARY patterns.
12036 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
12038 * target.def (preferred_else_value): New target hook.
12039 * doc/tm.texi.in (TARGET_PREFERRED_ELSE_VALUE): New hook.
12040 * doc/tm.texi: Regenerate.
12041 * targhooks.h (default_preferred_else_value): Declare.
12042 * targhooks.c (default_preferred_else_value): New function.
12043 * internal-fn.h (conditional_internal_fn_code): Declare.
12044 * internal-fn.c (FOR_EACH_CODE_MAPPING): New macro.
12045 (get_conditional_internal_fn): Use it.
12046 (conditional_internal_fn_code): New function.
12047 * gimple-match.h (gimple_match_cond): New struct.
12048 (gimple_match_op): Add a cond member function.
12049 (gimple_match_op::gimple_match_op): Update all forms to take a
12051 * genmatch.c (expr::gen_transform): Use the same condition as res_op
12052 for the suboperation, but don't specify a particular else_value.
12053 * tree-ssa-sccvn.c (vn_nary_simplify, vn_reference_lookup_3)
12054 (visit_nary_op, visit_reference_op_load): Pass
12055 gimple_match_cond::UNCOND to the gimple_match_op constructor.
12056 * gimple-match-head.c: Include tree-eh.h
12057 (convert_conditional_op): New function.
12058 (maybe_resimplify_conditional_op): Likewise.
12059 (gimple_resimplify1): Call maybe_resimplify_conditional_op.
12060 (gimple_resimplify2): Likewise.
12061 (gimple_resimplify3): Likewise.
12062 (gimple_resimplify4): Likewise.
12063 (maybe_push_res_to_seq): Return null for conditional operations.
12064 (try_conditional_simplification): New function.
12065 (gimple_simplify): Call it. Pass conditions to the gimple_match_op
12067 * match.pd: Fold VEC_COND_EXPRs of an IFN_COND_* call to a new
12069 * config/aarch64/aarch64.c (aarch64_preferred_else_value): New
12071 (TARGET_PREFERRED_ELSE_VALUE): Redefine.
12073 2018-07-12 Jan Hubicka <hubicka@ucw.cz>
12075 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not stream
12077 (hash_tree): Do not hash DECL_FCONTEXT
12078 * tree-streamer-in.c (lto_input_ts_field_decl_tree_pointers):
12079 Do not stream DECL_FCONTEXT.
12080 * tree-streamer-out.c (write_ts_field_decl_tree_pointers): Likewise.
12081 * tree.c (free_lang_data_in_decl): Free DECL_FCONTEXT.
12083 2018-07-12 Richard Biener <rguenther@suse.de>
12086 * dwarf2out.c (gen_block_die): Only output blocks when they have
12087 at least one !DECL_IGNORED_P variable.
12089 2018-07-12 Richard Biener <rguenther@suse.de>
12092 * config/gnu-user.h (GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC):
12093 Remove -mieee-fp handling.
12095 2018-07-12 Richard Biener <rguenther@suse.de>
12097 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Remove bogus
12098 left-over from last patch.
12100 2018-07-12 Jakub Jelinek <jakub@redhat.com>
12102 PR tree-optimization/86492
12103 * gimple-ssa-store-merging.c
12104 (imm_store_chain_info::coalesce_immediate_stores): Call
12105 check_no_overlap even for the merge_overlapping case. Formatting fix.
12107 2018-07-12 Richard Biener <rguenther@suse.de>
12109 PR middle-end/86479
12110 * fold-const.c (fold_binary_op_with_conditional_arg): Do not
12111 move possibly trapping operations into the conditional.
12113 2018-07-12 Richard Biener <rguenther@suse.de>
12115 * tree-ssa-sccvn.c (mprts_hook_cnt): Remove.
12116 (vn_lookup_simplify_result): Remove recursion limit applied
12118 (vn_nary_build_or_lookup_1): Adjust.
12119 (try_to_simplify): Likewise.
12120 * gimple-match-head.c (gimple_resimplify1): Instead apply one
12122 (gimple_resimplify2): Likewise.
12123 (gimple_resimplify3): Likewise.
12124 (gimple_resimplify4): Likewise.
12126 2018-07-11 Jakub Jelinek <jakub@redhat.com>
12128 * config/i386/avx512bitalgintrin.h (_mm512_mask_bitshuffle_epi64_mask):
12129 Use __mmask64 type instead of __mmask8 for __M argument.
12130 * config/i386/avx512fintrin.h (_mm512_mask_xor_epi64,
12131 _mm512_maskz_xor_epi64): Use __mmask8 type instead of __mmask16 for
12133 (_mm512_mask_cmpneq_epi64_mask): Use __mmask8 type instead of
12134 __mmask16 for __M argument.
12135 (_mm512_maskz_insertf32x4, _mm512_maskz_inserti32x4,
12136 _mm512_mask_insertf32x4, _mm512_mask_inserti32x4): Cast last argument
12137 to __mmask16 instead of __mmask8.
12138 * config/i386/avx512vlintrin.h (_mm_mask_add_ps, _mm_maskz_add_ps,
12139 _mm256_mask_add_ps, _mm256_maskz_add_ps, _mm_mask_sub_ps,
12140 _mm_maskz_sub_ps, _mm256_mask_sub_ps, _mm256_maskz_sub_ps,
12141 _mm256_maskz_cvtepi32_ps, _mm_maskz_cvtepi32_ps): Use __mmask8 type
12142 instead of __mmask16 for __U argument.
12143 * config/i386/avx512vlbwintrin.h (_mm_mask_cmp_epi8_mask): Use
12144 __mmask16 instead of __mmask8 for __U argument.
12145 (_mm256_mask_cmp_epi8_mask): Use __mmask32 instead of __mmask16 for
12147 (_mm256_cmp_epi8_mask): Use __mmask32 return type instead of
12149 (_mm_mask_cmp_epu8_mask): Use __mmask16 instead of __mmask8 for __U
12151 (_mm256_mask_cmp_epu8_mask): Use __mmask32 instead of __mmask16 for
12153 (_mm256_cmp_epu8_mask): Use __mmask32 return type instead of
12155 (_mm_mask_cmp_epi16_mask): Cast last argument to __mmask8 instead
12157 (_mm256_mask_cvtepi8_epi16): Use __mmask16 instead of __mmask32 for
12159 (_mm_mask_cvtepi8_epi16): Use __mmask8 instead of __mmask32 for
12161 (_mm256_mask_cvtepu8_epi16): Use __mmask16 instead of __mmask32 for
12163 (_mm_mask_cvtepu8_epi16): Use __mmask8 instead of __mmask32 for
12165 (_mm256_mask_cmpneq_epu8_mask, _mm256_mask_cmplt_epu8_mask,
12166 _mm256_mask_cmpge_epu8_mask, _mm256_mask_cmple_epu8_mask): Change
12167 return type as well as __M argument type and all casts from __mmask8
12169 (_mm256_mask_cmpneq_epu16_mask, _mm256_mask_cmplt_epu16_mask,
12170 _mm256_mask_cmpge_epu16_mask, _mm256_mask_cmple_epu16_mask): Change
12171 return type as well as __M argument type and all casts from __mmask8
12173 (_mm256_mask_cmpneq_epi8_mask, _mm256_mask_cmplt_epi8_mask,
12174 _mm256_mask_cmpge_epi8_mask, _mm256_mask_cmple_epi8_mask): Change
12175 return type as well as __M argument type and all casts from __mmask8
12177 (_mm256_mask_cmpneq_epi16_mask, _mm256_mask_cmplt_epi16_mask,
12178 _mm256_mask_cmpge_epi16_mask, _mm256_mask_cmple_epi16_mask): Change
12179 return type as well as __M argument type and all casts from __mmask8
12181 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_shrdi_epi32,
12182 _mm_mask_shldi_epi32): Cast last argument to __mmask8 instead of
12185 2018-07-11 Grazvydas Ignotas <notasas@gmail.com>
12187 * config/i386/avx512bwintrin.h: (_mm512_mask_cmp_epi8_mask,
12188 _mm512_mask_cmp_epu8_mask): Use __mmask64 type instead of __mmask32
12191 2018-07-11 Paul Koning <ni1d@arrl.net>
12193 * doc/md.texi (define_subst): Document how multiple occurrences of
12194 the same argument in the replacement pattern are handled.
12196 2018-07-11 Paul Koning <ni1d@arrl.net>
12198 * doc/extend.texi (Common Variable Attributes): Move "mode" into
12199 alphabetical order.
12200 (Common Type Attributes): Add "mode" attribute.
12202 2018-07-11 Jan Hubicka <hubicka@ucw.cz>
12204 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not
12205 stream DECL_ORIGINAL_TYPE.
12206 (DFS::DFS_write_tree_body): Drop hack handling local external decls.
12207 (hash_tree): Do not walk DECL_ORIGINAL_TYPE.
12208 * tree-streamer-in.c (lto_input_ts_decl_non_common_tree_pointers):
12209 Do not walk original type.
12210 * tree-streamer-out.c (streamer_write_chain): Drop hack handling
12212 (write_ts_decl_non_common_tree_pointers): Do not stream
12214 * tree.c (free_lang_data_in_decl): Clear DECL_ORIGINAL_TYPE.
12215 (find_decls_types_r): Do not walk DEC_ORIGINAL_TYPE.
12217 2018-07-11 Aldy Hernandez <aldyh@redhat.com>
12219 * tree-ssa-threadupdate.c (thread_through_all_blocks): Do not jump
12220 thread twice from the same starting edge.
12222 2018-07-11 Aldy Hernandez <aldyh@redhat.com>
12224 * vr-values.c (gimple_stmt_nonzero_p): Abstract common code to...
12225 * gimple.c (gimple_call_nonnull_result_p): ...here...
12226 (gimple_call_nonnull_arg): ...and here.
12227 * gimple.h (gimple_call_nonnull_result_p): New.
12228 (gimple_call_nonnull_arg): New.
12230 2018-07-11 Richard Earnshaw <rearnsha@arm.com>
12232 * config/arm/arm-cpus.in: Move information from fpu field of each
12233 cpu definition to the isa field.
12234 * config/arm/parsecpu.awk (fpu): Delete match rule.
12235 (gen_comm_data): Don't add bits from the CPU's FPU entry.
12237 2018-07-11 Richard Biener <rguenther@suse.de>
12240 * dwarf2out.c (init_sections_and_labels): Use
12241 output_asm_line_debug_info consistently.
12242 (dwarf2out_early_finish): Likewise.
12243 (dwarf2out_finish): Remove DW_AT_stmt_list from early generated
12246 2018-07-11 Richard Biener <rguenther@suse.de>
12248 * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1):
12249 Rework father_bb setting in a way to avoid propagating constants
12250 multiple times on a loop body.
12252 2018-07-10 Mark Wielaard <mark@klomp.org>
12255 * dwarf2out.c (output_macinfo_op): Fix dwarf_FORM typo in gcc_assert.
12257 2018-07-10 Richard Biener <rguenther@suse.de>
12259 * hash-map.h (hash_map::iterator::operator*): Return
12260 references to key and value.
12262 2018-07-10 Jakub Jelinek <jakub@redhat.com>
12265 * gimplify.c (find_combined_omp_for): Add DATA argument, in addition
12266 to finding the inner OMP_FOR/OMP_SIMD stmt find non-trivial wrappers,
12267 BLOCKs with BLOCK_VARs, OMP_PARALLEL in between, OMP_FOR in between.
12268 (gimplify_omp_for): For composite loops, move outer
12269 OMP_{DISTRIBUTE,TASKLOOP,FOR,PARALLEL} right around innermost
12270 OMP_FOR/OMP_SIMD if there are any non-trivial wrappers. For class
12271 iterators add any needed clauses. Allow OMP_FOR_ORIG_DECLS to contain
12272 TREE_LIST for both the original class iterator and the "last" helper
12273 var. Gimplify OMP_FOR_PRE_BODY before the outermost composite
12274 loop, remember has_decl_expr from outer composite loops for the
12275 innermost OMP_SIMD in TREE_PRIVATE bit on OMP_FOR_INIT.
12277 2018-07-09 Martin Sebor <msebor@redhat.com>
12279 PR middle-end/77357
12280 PR middle-end/86428
12281 * builtins.c (c_strlen): Avoid out-of-bounds warnings when
12282 accessing implicitly initialized array elements.
12283 * expr.c (string_constant): Handle string initializers of
12284 character arrays within aggregates.
12285 * gimple-fold.c (fold_array_ctor_reference): Add argument.
12286 Store element offset. As a special case, handle zero size.
12287 (fold_nonarray_ctor_reference): Same.
12288 (fold_ctor_reference): Add argument. Store subobject offset.
12289 * gimple-fold.h (fold_ctor_reference): Add argument.
12291 2018-07-09 Paul Koning <ni1d@arrl.net>
12293 * config/pdp11/pdp11.c (pdp11_addr_cost): New function.
12294 (pdp11_insn_cost): New function.
12295 (pdp11_md_asm_adjust): New function.
12296 (TARGET_INVALID_WITHIN_DOLOOP): Define.
12297 (pdp11_rtx_costs): Update to match machine better.
12298 (output_addr_const_pdp11): Correct format mismatch warnings.
12299 * config/pdp11/pdp11.h (SLOW_BYTE_ACCESS): Correct definition.
12300 * config/pdp11/pdp11.md: General change to add base_cost and/or
12301 length attributes for use by new pdp11_insn_cost function.
12302 (MIN_BRANCH): Correct definition.
12304 (doloop_end): Use standard pattern name for looping pattern.
12305 (doloop_end_nocc): New.
12306 (movsf): Add another constraint alternative.
12307 (zero_extendqihi2): Add constraint alternatives for not in place
12309 (zero_extendhisi2): Remove.
12310 (shift patterns): Add CC handling variants.
12314 (define_peephole2): New peephole to recognize mov that sets CC for
12317 2018-07-09 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
12319 * config/sparc/sparc.c (sparc_fold_builtin) <SPARC_BUILTIN_PDIST,
12320 SPARC_BUILTIN_PDISTN>: Adapt for signature change of wi::neg,
12323 2018-07-09 Jakub Jelinek <jakub@redhat.com>
12326 * real.c (real_nextafter): Return true if result is denormal.
12328 2018-07-09 Martin Liska <mliska@suse.cz>
12330 * common.opt: Add back wrongly removed attribute.
12332 2018-07-09 Richard Biener <rguenther@suse.de>
12335 * dwarf2out.c (gen_block_die): For an early generated DIE
12336 always output high/low PC attributes.
12338 2018-07-09 Tom de Vries <tdevries@suse.de>
12340 * cfgexpand.c (expand_debug_source_expr): Handle VAR_DECL.
12341 * tree-inline.c (remap_ssa_name): Handle default def ssa_name mapping
12342 onto VAR_DECL with abstract origin.
12344 2018-07-07 Jim Wilson <jimw@sifive.com>
12346 * config/riscv/riscv.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): New.
12348 2018-07-07 Tom de Vries <tdevries@suse.de>
12350 * tree-dfa.c (dump_enumerated_decls): Handle cfun->cfg == NULL.
12352 2018-07-07 Aldy Hernandez <aldyh@redhat.com>
12354 * tree-vrp.c (vrp_int_const_binop): Change overflow type to
12356 (combine_bound): Use wide-int overflow calculation instead of
12358 * calls.c (maybe_warn_alloc_args_overflow): Change overflow type to
12360 * fold-const.c (int_const_binop_2): Same.
12361 (extract_muldiv_1): Same.
12362 (fold_div_compare): Same.
12363 (fold_abs_const): Same.
12365 * poly-int.h (add): Same.
12369 * predict.c (predict_iv_comparison): Same.
12370 * profile-count.c (slow_safe_scale_64bit): Same.
12371 * simplify-rtx.c (simplify_const_binary_operation): Same.
12372 * tree-chrec.c (tree_fold_binomial): Same.
12373 * tree-data-ref.c (split_constant_offset_1): Same.
12374 * tree-if-conv.c (idx_within_array_bound): Same.
12375 * tree-scalar-evolution.c (iv_can_overflow_p): Same.
12376 * tree-ssa-phiopt.c (minmax_replacement): Same.
12377 * tree-vect-loop.c (is_nonwrapping_integer_induction): Same.
12378 * tree-vect-stmts.c (vect_truncate_gather_scatter_offset): Same.
12379 * vr-values.c (vr_values::adjust_range_with_scev): Same.
12380 * wide-int.cc (wi::add_large): Same.
12381 (wi::mul_internal): Same.
12382 (wi::sub_large): Same.
12383 (wi::divmod_internal): Same.
12384 * wide-int.h: Change overflow type to overflow_type for neg, add,
12385 mul, smul, umul, div_trunc, div_floor, div_ceil, div_round,
12386 mod_trunc, mod_ceil, mod_round, add_large, sub_large,
12387 mul_internal, divmod_internal.
12388 (overflow_type): New enum.
12389 (accumulate_overflow): New.
12391 2018-07-06 Kugan Vivekanandarajah <kugan.vivekanandarajah@linaro.org>
12393 * tree-ssa-phiopt.c (cond_removal_in_popcount_pattern): New.
12394 (tree_ssa_phiopt_worker): Call cond_removal_in_popcount_pattern.
12396 2018-07-06 Kugan Vivekanandarajah <kuganv@linaro.org>
12398 * tree-ssa-loop-niter.c (number_of_iterations_popcount): If popcount
12399 argument is checked for zero before entering loop, avoid checking again.
12401 2018-07-06 Kugan Vivekanandarajah <kuganv@linaro.org>
12403 * gimplify.h (generic_expr_could_trap_p): Set as global function.
12404 * gimplify.h (generic_expr_could_trap_p): Likwise.
12405 * tree-scalar-evolution.c (expression_expensive_p): Handle COND_EXPR.
12407 2018-07-06 Jakub Jelinek <jakub@redhat.com>
12409 PR tree-optimization/86401
12410 * fold-const.c (fold_binary_loc) <case BIT_AND_EXPR>: Move the
12411 ((A & N) + B) & M -> (A + B) & M etc. optimization into ...
12412 (fold_bit_and_mask): ... here. New helper function for match.pd.
12413 * fold-const.h (fold_bit_and_mask): Declare.
12414 * match.pd (((A & N) + B) & M -> (A + B) & M): New optimization.
12416 2018-07-06 Peter Bergner <bergner@linux.ibm.com>
12419 * target.def (translate_mode_attribute): New hook.
12420 * targhooks.h (default_translate_mode_attribute): Declare.
12421 * targhooks.c (default_translate_mode_attribute): New function.
12422 * doc/tm.texi.in (TARGET_TRANSLATE_MODE_ATTRIBUTE): New hook.
12423 * doc/tm.texi: Regenerate.
12424 * config/rs6000/rs6000.c (TARGET_TRANSLATE_MODE_ATTRIBUTE): Define.
12425 (rs6000_translate_mode_attribute): New function.
12427 2018-07-06 Paul Koning <ni1d@arrl.net>
12429 * doc/md.texi (define_split): Document DONE and FAIL.
12430 (define_peephole2): Ditto.
12432 2018-07-05 Jeff Law <law@redhat.com>
12434 PR tree-optimization/86010
12435 * tree-ssa-dse.c (compute_trims): More aggressively trim at
12436 both the head and tail of mem* and str* calls.
12438 2018-07-05 Jim Wilson <jimw@sifive.com>
12440 * config.gcc (riscv*-*-*): When setting xlen, handle riscv-*.
12442 2018-07-05 Indu Bhagat <indu.bhagat@oracle.com>
12444 * config/aarch64/aarch64-simd.md: correct flags text for
12445 MIN_EXPR replacement.
12447 2018-07-05 James Clarke <jrtc27@jrtc27.com>
12449 * configure: Regenerated.
12451 2018-07-05 Carl Love <cel@us.ibm.com>
12453 * config/rs6000/rs6000-c.c: Map ALTIVEC_BUILTIN_VEC_UNPACKH for
12454 float argument to VSX_BUILTIN_DOUBLEH_V4SF.
12455 Map ALTIVEC_BUILTIN_VEC_UNPACKL for float argument to
12456 VSX_BUILTIN_DOUBLEL_V4SF.
12458 2018-07-05 Martin Sebor <msebor@redhat.com>
12461 * tree-ssa-strlen.c (maybe_set_strlen_range): Use type size rather
12462 than its domain to compute its the upper bound of a char array.
12464 2018-07-05 Nathan Sidwell <nathan@acm.org>
12466 Replace NO_IMPLICIT_EXTERN_C with SYSTEM_IMPLICIT_EXTERN_C.
12467 * doc/cpp.texi: Update comment.
12468 * doc/tm.texi: Rebuilt.
12469 * doc/tm.texi.in (NO_IMPLICIT_EXTERN_C): Replace with ...
12470 (SYSTEM_IMPLICIT_EXTERN_C): ... this, opposite sense.
12471 * doc/extend.texi (Backwards Compatibility): Clarify it is system
12472 headers affected by extern "C".
12473 * system.h: Poison NO_IMPLICIT_EXTERN_C.
12474 * config/alpha/alpha.h, config/arm/uclinux-elf.h,
12475 config/bfin/elf.h, config/cris/cris.h, config/darwin.h,
12476 config/dragonfly.h, config/freebsd.h, config/gnu-user.h,
12477 config/i386/cygming.h, config/i386/djgpp.h, config/i386/nto.h,
12478 config/ia64/hpux.h, config/lm32/lm32.h, config/lm32/uclinux-elf.h,
12479 config/lynx.h, config/mips/elf.h, config/mmix/mmix.h,
12480 config/netbsd.h, config/pa/pa-hpux.h, config/powerpcspe/sysv4.h,
12481 config/riscv/elf.h, config/rs6000/sysv4.h, config/rtems.h,
12482 config/s390/tpf.h, config/sh/newlib.h, config/sol2.h,
12483 config/sparc/openbsd64.h, config/sparc/sp-elf.h,
12484 config/sparc/sp64-elf.h, config/spu/spu.h,
12485 config/stormy16/stormy16.h, config/v850/v850.h,
12486 config/visium/visium.h, config/vx-common.h, config/xtensa/elf.h: Don't
12487 define NO_IMPLICIT_EXTERN_C.
12488 * config/rs6000/aix.h: Set SYSTEM_IMPLICIT_EXTERN_C.
12490 2018-07-05 Tamar Christina <tamar.christina@arm.com>
12493 * config/arm/arm.c (arm_can_change_mode_class): Use GET_MODE_UNIT_SIZE
12494 instead of GET_MODE_SIZE when comparing Units.
12496 2018-07-05 Tamar Christina <tamar.christina@arm.com>
12499 * rtlanal.c (set_noop_p): Constrain on mode change,
12500 include hard-reg-set.h
12502 2018-07-05 Tamar Christina <tamar.christina@arm.com>
12504 * config/aarch64/aarch64.c (aarch64_expand_movmem): Fix mode size.
12506 2018-07-05 Jakub Jelinek <jakub@redhat.com>
12509 2018-07-04 Maxim Ostapenko <m.ostapenko@samsung.com>
12512 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Pass -lstdc++ for static
12514 * gcc.c: Do not pass LIBUBSAN_SPEC if ASan is enabled with UBSan.
12516 2018-07-04 Maxim Ostapenko <m.ostapenko@samsung.com>
12519 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Pass -lstdc++ for static
12521 * gcc.c: Do not pass LIBUBSAN_SPEC if ASan is enabled with UBSan.
12523 2018-07-04 Eric Botcazou <ebotcazou@adacore.com>
12525 PR middle-end/86380
12526 * expmed.c (choose_multiplier): Fix incorrect comparison with mask.
12528 2018-07-04 Aldy Hernandez <aldyh@redhat.com>
12530 * tree-vrp.c (extract_range_from_binary_expr_1): Initialize
12531 neg_*_op* variables.
12533 2018-07-04 Martin Liska <mliska@suse.cz>
12535 * tree-switch-conversion.c: Define
12536 max_ratio_for_speed and max_ratio_for_size constants.
12538 2018-07-04 Denys Vlasenko <dvlasenk@redhat.com>
12539 Martin Liska <mliska@suse.cz>
12541 PR middle-end/66240
12544 * common.opt: Rename align options with 'str_' prefix.
12545 * common/config/i386/i386-common.c (set_malign_value): New
12547 (ix86_handle_option): Use it to set -falign-* options/
12548 * config/aarch64/aarch64-protos.h (struct tune_params): Change
12549 type from int to string.
12550 * config/aarch64/aarch64.c: Update default values from int
12552 * config/alpha/alpha.c (alpha_override_options_after_change):
12554 * config/arm/arm.c (arm_override_options_after_change_1): Likewise.
12555 * config/i386/dragonfly.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
12556 max skip conditionally.
12557 * config/i386/freebsd.h (SUBALIGN_LOG): New.
12558 (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
12559 max skip conditionally.
12560 * config/i386/gas.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
12561 max skip conditionally.
12562 * config/i386/gnu-user.h (SUBALIGN_LOG): New.
12563 (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
12564 max skip conditionally.
12565 * config/i386/i386.c (struct ptt): Change type from int to
12567 (ix86_default_align): Set default values.
12568 * config/i386/i386.h (ASM_OUTPUT_MAX_SKIP_PAD): Print
12569 max skip conditionally.
12570 * config/i386/iamcu.h (SUBALIGN_LOG): New.
12571 (ASM_OUTPUT_MAX_SKIP_ALIGN):
12572 * config/i386/lynx.h (ASM_OUTPUT_MAX_SKIP_ALIGN):
12573 * config/i386/netbsd-elf.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
12574 max skip conditionally.
12575 * config/i386/openbsdelf.h (SUBALIGN_LOG): New.
12576 (ASM_OUTPUT_MAX_SKIP_ALIGN) Print max skip conditionally.:
12577 * config/i386/x86-64.h (SUBALIGN_LOG): New.
12578 (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
12579 max skip conditionally.
12580 (ASM_OUTPUT_MAX_SKIP_PAD): Likewise.
12581 * config/ia64/ia64.c (ia64_option_override): Set default values
12582 for alignment options.
12583 * config/m68k/m68k.c: Handle new str_align_* options.
12584 * config/mips/mips.c (mips_set_compression_mode): Change
12586 (mips_option_override): Set default values for options.
12587 * config/powerpcspe/powerpcspe.c (rs6000_option_override_internal):
12589 * config/rs6000/rs6000.c (rs6000_option_override_internal):
12591 * config/rx/rx.c (rx_option_override): Likewise.
12592 * config/rx/rx.h (JUMP_ALIGN): Use align_jumps_log.
12593 (LABEL_ALIGN): Use align_labels_log.
12594 (LOOP_ALIGN): Use align_loops_align.
12595 * config/s390/s390.c (s390_asm_output_function_label): Use new
12597 * config/sh/sh.c (sh_override_options_after_change):
12598 Change type of constants.
12599 * config/spu/spu.c (spu_sched_init): Likewise.
12600 * config/sparc/sparc.c (sparc_option_override): Set default
12601 values for options.
12602 * config/visium/visium.c (visium_option_override): Likewise.
12603 * config/visium/visium.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Do not
12604 emit p2align format with last argument if it's not needed.
12605 * doc/invoke.texi: Document extended format of -falign-*.
12606 * final.c: Use align_labels alignment.
12607 * flags.h (struct target_flag_state): Change type to use
12609 (struct align_flags_tuple): New.
12610 (struct align_flags): Likewise.
12611 (align_loops_log): Redefine macro to use new types.
12612 (align_loops_max_skip): Redefine macro to use new types.
12613 (align_jumps_log): Redefine macro to use new types.
12614 (align_jumps_max_skip): Redefine macro to use new types.
12615 (align_labels_log): Redefine macro to use new types.
12616 (align_labels_max_skip): Redefine macro to use new types.
12617 (align_functions_log): Redefine macro to use new types.
12618 (align_loops): Redefine macro to use new types.
12619 (align_jumps): Redefine macro to use new types.
12620 (align_labels): Redefine macro to use new types.
12621 (align_functions): Redefine macro to use new types.
12622 (align_functions_max_skip): Redefine macro to use new types.
12623 (align_loops_value): New macro.
12624 (align_jumps_value): New macro.
12625 (align_labels_value): New macro.
12626 (align_functions_value): New macro.
12627 * function.c (invoke_set_current_function_hook): Propagate
12628 alignment values from flags to global variables default in
12630 * ipa-icf.c (sem_function::equals_wpa): Use
12631 cl_optimization_option_eq instead of memcmp.
12632 * lto-streamer.h (cl_optimization_stream_out): Support streaming
12634 (cl_optimization_stream_in): Likewise.
12635 * optc-save-gen.awk: Support strings in cl_optimization.
12636 * opth-gen.awk: Likewise.
12637 * opts.c (finish_options): Remove error checking of invalid
12639 (MAX_CODE_ALIGN): Remove.
12640 (MAX_CODE_ALIGN_VALUE): Likewise.
12641 (parse_and_check_align_values): New function.
12642 (check_alignment_argument): Likewise.
12643 (common_handle_option): Use check_alignment_argument.
12644 * opts.h (parse_and_check_align_values): Declare.
12645 * toplev.c (init_alignments): Remove.
12646 (read_log_maxskip): New.
12647 (parse_N_M): Likewise.
12648 (parse_alignment_opts): Likewise.
12649 (backend_init_target): Remove usage of init_alignments.
12650 * toplev.h (parse_alignment_opts): Declare.
12651 * tree-streamer-in.c (streamer_read_tree_bitfields): Add new
12653 * tree-streamer-out.c (streamer_write_tree_bitfields): Likewise.
12654 * tree.c (cl_option_hasher::equal): New.
12655 * varasm.c: Use new global macros.
12657 2018-07-04 Denys Vlasenko <dvlasenk@redhat.com>
12659 * config/i386/dragonfly.h: (ASM_OUTPUT_MAX_SKIP_ALIGN):
12660 Use a simpler align directive also if MAXSKIP = ALIGN-1.
12661 * config/i386/gas.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
12662 * config/i386/lynx.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
12663 * config/i386/netbsd-elf.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
12664 * config/i386/i386.h (ASM_OUTPUT_MAX_SKIP_PAD): Likewise.
12665 * config/i386/freebsd.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Remove "If N
12666 is large, do at least 8 byte alignment" code. Add SUBALIGN_LOG
12667 define. Use a simpler align directive also if MAXSKIP = ALIGN-1.
12668 * config/i386/gnu-user.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
12669 * config/i386/iamcu.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
12670 * config/i386/openbsdelf.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
12671 * config/i386/x86-64.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
12673 2018-07-04 Martin Liska <mliska@suse.cz>
12674 Jonathan Wakely <jwakely@redhat.com>
12676 * coverage.c: Use correct type.
12677 * doc/invoke.texi: Language correction.
12679 2018-07-03 H.J. Lu <hongjiu.lu@intel.com>
12682 * config/i386/i386.c (rest_of_insert_endbranch): Also generate
12683 ENDBRANCH for non-tail call which may return via indirect branch.
12684 * doc/extend.texi: Document indirect_return attribute.
12686 2018-07-03 Martin Sebor <msebor@redhat.com>
12688 PR tree-optimization/86274
12689 * gimple-ssa-sprintf.c (fmtresult::type_max_digits): Verify
12691 (format_floating): Correct handling of infinities and NaNs.
12693 2018-07-03 Martin Sebor <msebor@redhat.com>
12695 * print-tree.c (print_real_cst): New function.
12696 (print_node_brief): Call it.
12697 (print_node): Ditto.
12699 2018-07-03 Jeff Law <law@redhat.com>
12701 * config/h8300/h8300.md (logical<mode>3_sn, logical<mode>3): Merge
12702 into a single pattern.
12704 * config/h8300/h8300.md (ors code_iterator): New.
12705 (bsetqi_msx, bnotqi_msx patterns and splitters): Consolidate into
12706 a single pattern and single splitter.
12707 (bsethi_msx, bnothi_msx patterns): Consolidate into a single pattern.
12708 (iorqi3_1, xorqi3_1): Likewise.
12709 (iorqi3, xorqi3 expanders): Similarly.
12711 * config/h8300/h8300.md (movmd_internal_normal): Consolidated with
12712 (movmd_internal) into a single pattern using the P mode iterator.
12713 (movmd splitters): Similarly.
12714 (stpcpy_internal_normal, stpcpy_internal): Similarly for thes patterns.
12715 (movsd splitters): Similarly.
12717 * config/h8300/h8300.c (h8300_insn_length_from_table): Consolidate
12718 ADDB, ADDW and ADDL into a single ADD attribute which selects the
12719 right table based on the size of the operand.
12720 * config/h8300/h8300.md (length_table): Corresponding changes. All
12721 references to "addb", "addw" and "addl" changed to "add".
12722 (btst patterns): Merge two variants into a single pattern.
12723 (tstqi, tsthi): Likewise.
12724 (addhi3_incdec, addsi3_incdec): Likewise.
12725 (subhi3_h8300hs, subsi3_h8300hs): Likewise.
12726 (mulhi3, mulsi3): Likewise.
12727 (udivhi3, udivsi3): Likewise.
12728 (divhi3, divsi3): Likewise.
12729 (andorqi3, andorhi3, andorsi3): Likewise.
12731 2018-07-03 Uros Bizjak <ubizjak@gmail.com>
12734 * config/i386/sse.md (uavg<mode>3_ceil): New expander.
12735 (<sse2_avx2>_uavg<mode>3<mask_name>): Simplify expander.
12737 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
12739 PR tree-optimization/85694
12740 * config/aarch64/iterators.md (HADD, RHADD): New int iterators.
12741 (u): Handle UNSPEC_SHADD, UNSPEC_UHADD, UNSPEC_SRHADD and
12743 * config/aarch64/aarch64-simd.md (<u>avg<mode>3_floor)
12744 (<u>avg<mode>3_ceil): New patterns.
12746 2018-07-03 David Malcolm <dmalcolm@redhat.com>
12748 * gcc.dg/vect/slp-perm-1.c: Remove "note: " prefix from
12749 scan-tree-dump directive.
12750 * gcc.dg/vect/slp-perm-2.c: Likewise.
12751 * gcc.dg/vect/slp-perm-3.c: Likewise.
12752 * gcc.dg/vect/slp-perm-5.c: Likewise.
12753 * gcc.dg/vect/slp-perm-6.c: Likewise.
12754 * gcc.dg/vect/slp-perm-7.c: Likewise.
12755 * gcc.dg/vect/slp-perm-8.c: Likewise.
12757 2018-07-03 Marek Polacek <polacek@redhat.com>
12759 PR middle-end/86202
12760 * gimple-fold.c (size_must_be_zero_p): Check the type of the size.
12762 2018-07-03 Richard Biener <rguenther@suse.de>
12765 * tree-ssa-structalias.c (find_func_clobbers): Properly
12766 handle indirect calls.
12768 2018-07-03 Jeff Law <law@redhat.com>
12770 * config/h8300/h8300.md (HSI, QHSI, QHSIF): New mode iterators.
12771 (shifts): New code iterator.
12772 (movqi, movhi, movsi, movsf expanders): Consolidate into a single
12773 expander. Fix HImode handling on H8/SX.
12774 (addqi3, addhi3, addsi3 expanders): Consolidate into a single expander.
12775 (subqi3, subhi3, subsi3 expanders): Likewise.
12776 (andqi3, andhi3, andsi3 expanders): Likewise.
12777 (iorqi3, iorhi3, iorsi3 expanders): Likewise.
12778 (xorqi3, xorhi3, xorsi3 expanders): Likewise.
12779 (negqi2, neghi2, negsi2, negsf2 expanders): Likewise.
12780 (one_cmplqi2, one_cmplhi2, one_cmplsi2): Likewise.
12781 (zero_extendqihi2, zero_extendqisi2): Likewise.
12782 (extendqihi2, extendqisi2): Likewise.
12783 (rotlqi3, rotlhi3, rotlsi3): Likewise.
12784 (neghi2_h8300, negsi2_h8300): Likewise for these patterns.
12785 (rotlqi3_1, rotlhi3_1): Likewise.
12786 (logicalhi3_sn, logicalsi3_sn): Likewise.
12787 (logicalhi3, logicalsi3): Likewise.
12789 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
12791 * tree-vect-patterns.c (vect_recog_rotate_pattern)
12792 (vect_recog_vector_vector_shift_pattern, vect_recog_divmod_pattern)
12793 (vect_recog_mixed_size_cond_pattern, adjust_bool_pattern_cast)
12794 (adjust_bool_pattern, vect_recog_bool_pattern): Pass the vector
12795 type to append_pattern_def_seq instead of creating a stmt_vec_info
12797 (build_mask_conversion): Likewise. Remove vinfo argument.
12798 (vect_add_conversion_to_patterm): Likewise, renaming to...
12799 (vect_add_conversion_to_pattern): ...this.
12800 (vect_recog_mask_conversion_pattern): Update call to
12801 build_mask_conversion. Pass the vector type to
12802 append_pattern_def_seq here too.
12803 (vect_recog_gather_scatter_pattern): Update call to
12804 vect_add_conversion_to_pattern.
12806 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
12808 * tree-vect-patterns.c (new_pattern_def_seq): Delete.
12809 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern)
12810 (vect_recog_widen_op_pattern, vect_recog_over_widening_pattern)
12811 (vect_recog_rotate_pattern, vect_synth_mult_by_constant): Don't set
12812 STMT_VINFO_PATTERN_DEF_SEQ to null here.
12813 (vect_recog_pow_pattern, vect_recog_vector_vector_shift_pattern)
12814 (vect_recog_mixed_size_cond_pattern, vect_recog_bool_pattern): Use
12815 append_pattern_def_seq instead of new_pattern_def_seq.
12816 (vect_recog_divmod_pattern): Do both of the above.
12817 (vect_pattern_recog_1): Assert that STMT_VINO_PATTERN_DEF_SEQ
12820 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
12822 * tree-vect-patterns.c (vect_recog_dot_prod_pattern):
12823 (vect_recog_sad_pattern, vect_recog_widen_op_pattern)
12824 (vect_recog_widen_mult_pattern, vect_recog_pow_pattern):
12825 (vect_recog_widen_sum_pattern, vect_recog_over_widening_pattern)
12826 (vect_recog_average_pattern, vect_recog_cast_forwprop_pattern)
12827 (vect_recog_widen_shift_pattern, vect_recog_rotate_pattern)
12828 (vect_recog_vector_vector_shift_pattern, vect_synth_mult_by_constant)
12829 (vect_recog_mult_pattern, vect_recog_divmod_pattern)
12830 (vect_recog_mixed_size_cond_pattern, vect_recog_bool_pattern)
12831 (vect_recog_mask_conversion_pattern): Replace vec<gimple *>
12832 parameter with a single stmt_vec_info.
12833 (vect_recog_func_ptr): Likewise.
12834 (vect_recog_gather_scatter_pattern): Likewise, folding in...
12835 (vect_try_gather_scatter_pattern): ...this.
12836 (vect_pattern_recog_1): Remove stmts_to_replace and just pass
12837 the stmt_vec_info of the statement to be matched. Don't clear
12838 STMT_VINFO_RELATED_STMT.
12839 (vect_pattern_recog): Update call accordingly.
12841 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
12843 PR tree-optimization/85694
12844 * doc/md.texi (avgM3_floor, uavgM3_floor, avgM3_ceil)
12845 (uavgM3_ceil): Document new optabs.
12846 * doc/sourcebuild.texi (vect_avg_qi): Document new target selector.
12847 * internal-fn.def (IFN_AVG_FLOOR, IFN_AVG_CEIL): New internal
12849 * optabs.def (savg_floor_optab, uavg_floor_optab, savg_ceil_optab)
12850 (savg_ceil_optab): New optabs.
12851 * tree-vect-patterns.c (vect_recog_average_pattern): New function.
12852 (vect_vect_recog_func_ptrs): Add it.
12853 * tree-vect-stmts.c (vectorizable_call): Get the type of the zero
12854 constant directly from the associated lhs.
12856 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
12858 * tree-vect-patterns.c (vect_split_statement): New function.
12859 (vect_convert_input): Use it to try to split an existing cast.
12861 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
12863 * poly-int.h (print_hex): New function.
12864 * dumpfile.h (dump_dec, dump_hex): Declare.
12865 * dumpfile.c (dump_dec, dump_hex): New poly_wide_int functions.
12866 * tree-vectorizer.h (_stmt_vec_info): Add min_output_precision,
12867 min_input_precision, operation_precision and operation_sign.
12868 * tree-vect-patterns.c (vect_get_range_info): New function.
12869 (vect_same_loop_or_bb_p, vect_single_imm_use)
12870 (vect_operation_fits_smaller_type): Delete.
12871 (vect_look_through_possible_promotion): Add an optional
12872 single_use_p parameter.
12873 (vect_recog_over_widening_pattern): Rewrite to use new
12874 stmt_vec_info infomration. Handle one operation at a time.
12875 (vect_recog_cast_forwprop_pattern, vect_narrowable_type_p)
12876 (vect_truncatable_operation_p, vect_set_operation_type)
12877 (vect_set_min_input_precision): New functions.
12878 (vect_determine_min_output_precision_1): Likewise.
12879 (vect_determine_min_output_precision): Likewise.
12880 (vect_determine_precisions_from_range): Likewise.
12881 (vect_determine_precisions_from_users): Likewise.
12882 (vect_determine_stmt_precisions, vect_determine_precisions): Likewise.
12883 (vect_vect_recog_func_ptrs): Put over_widening first.
12885 (vect_pattern_recog): Call vect_determine_precisions.
12887 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
12889 * tree-vect-patterns.c (vect_mark_pattern_stmts): Remove pattern
12890 statements that have been replaced by further pattern statements.
12891 (vect_pattern_recog_1): Clear STMT_VINFO_PATTERN_DEF_SEQ on failure.
12893 2018-07-03 Richard Biener <rguenther@suse.de>
12895 * tree-vect-stmts.c (vect_is_simple_use): Consolidate dumping,
12896 always set *dt. Dump vectype in vectype overload.
12897 * dumpfile.h (dump_gimple_expr): New function.
12898 (dump_gimple_expr_loc): Likewise.
12899 * dumpfile.c (dump_gimple_expr): New function.
12900 (dump_gimple_expr_loc): Likewise.
12902 2018-07-02 Jeff Law <law@redhat.com>
12904 * config/h8300/h8300.md (movqi_h8300, movqi_h8300hs): Consolidate
12905 the H8/300, H8/300H and H8/S variants into a single pattern.
12906 (movhi_h8300, movqi_h8300hs): Similarly.
12907 (pushqi_h8300hs, pushhi_h8300hs): Consolidate into a single pattern.
12908 (QHI mode iterator): New.
12910 * config/h8300/h8300.md: Remove trailing whitespace.
12912 2018-07-02 Jim Wilson <jimw@sifive.com>
12914 * config/riscv/riscv.c (riscv_expand_epilogue): Use emit_jump_insn
12915 instead of emit_insn for interrupt returns.
12916 * config/riscv/riscv.md (riscv_met): Add (return) to rtl.
12917 (riscv_sret, riscv_uret): Likewise.
12919 2018-07-02 David Malcolm <dmalcolm@redhat.com>
12921 * pretty-print.c (selftest::test_pp_format): Move save and restore
12922 of quotes to class auto_fix_quotes, and add an instance.
12923 * selftest.c: Include "intl.h".
12924 (selftest::auto_fix_quotes::auto_fix_quotes): New ctor.
12925 (selftest::auto_fix_quotes::~auto_fix_quotes): New dtor.
12926 * selftest.h (selftest::auto_fix_quotes): New class.
12928 2018-07-02 Richard Henderson <richard.henderson@linaro.org>
12930 * config/aarch64/aarch64-protos.h, config/aarch64/aarch64.c
12931 (aarch64_sve_prepare_conditional_op): Remove.
12932 * config/aarch64/aarch64-sve.md (cond_<SVE_INT_BINARY><SVE_I>):
12933 Allow aarch64_simd_reg_or_zero as select operand; remove
12934 the aarch64_sve_prepare_conditional_op call.
12935 (cond_<SVE_INT_BINARY_SD><SVE_SDI>): Likewise.
12936 (cond_<SVE_COND_FP_BINARY><SVE_F>): Likewise.
12937 (*cond_<SVE_INT_BINARY><SVE_I>_z): New pattern.
12938 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_z): New pattern.
12939 (*cond_<SVE_COND_FP_BINARY><SVE_F>_z): New pattern.
12940 (*cond_<SVE_INT_BINARY><SVE_I>_any): New pattern.
12941 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_any): New pattern.
12942 (*cond_<SVE_COND_FP_BINARY><SVE_F>_any): New pattern
12943 and a splitters to match all of the *_any patterns.
12944 * config/aarch64/predicates.md (aarch64_sve_any_binary_operator): New.
12946 * config/aarch64/iterators.md (SVE_INT_BINARY_REV): Remove.
12947 (SVE_COND_FP_BINARY_REV): Remove.
12948 (sve_int_op_rev, sve_fp_op_rev): New.
12949 * config/aarch64/aarch64-sve.md (*cond_<SVE_INT_BINARY><SVE_I>_0): New.
12950 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_0): New.
12951 (*cond_<SVE_COND_FP_BINARY><SVE_F>_0): New.
12952 (*cond_<SVE_INT_BINARY><SVE_I>_2): Rename, add movprfx alternative.
12953 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_2): Similarly.
12954 (*cond_<SVE_COND_FP_BINARY><SVE_F>_2): Similarly.
12955 (*cond_<SVE_INT_BINARY><SVE_I>_3): Similarly; use sve_int_op_rev.
12956 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_3): Similarly.
12957 (*cond_<SVE_COND_FP_BINARY><SVE_F>_3): Similarly; use sve_fp_op_rev.
12959 * config/aarch64/aarch64-sve.md (cond_<SVE_COND_FP_BINARY><SVE_F>):
12960 Remove match_dup 1 from the inner unspec.
12961 (*cond_<SVE_COND_FP_BINARY><SVE_F>): Likewise.
12963 * config/aarch64/aarch64.md (movprfx): New attr.
12964 (length): Default movprfx to 8.
12965 * config/aarch64/aarch64-sve.md (*mul<SVE_I>3): Add movprfx alt.
12966 (*madd<SVE_I>, *msub<SVE_I): Likewise.
12967 (*<su>mul<SVE_I>3_highpart): Likewise.
12968 (*<SVE_INT_BINARY_SD><SVE_SDI>3): Likewise.
12969 (*v<ASHIFT><SVE_I>3): Likewise.
12970 (*<su><MAXMIN><SVE_I>3): Likewise.
12971 (*<su><MAXMIN><SVE_F>3): Likewise.
12972 (*fma<SVE_F>4, *fnma<SVE_F>4): Likewise.
12973 (*fms<SVE_F>4, *fnms<SVE_F>4): Likewise.
12974 (*div<SVE_F>4): Likewise.
12976 2018-07-02 Richard Sandiford <richard.sandiford@arm.com>
12978 * tree-vect-patterns.c (vect_recog_widen_shift_pattern): Fix typo
12981 2018-07-02 Richard Biener <rguenther@suse.de>
12983 PR tree-optimization/86363
12984 * tree-ssa-sccvn.c (vn_reference_lookup_3): Check the
12985 memset argument refers to a non-variable address.
12987 2018-07-02 Aldy Hernandez <aldyh@redhat.com>
12989 * tree-vrp.c (extract_range_from_binary_expr_1): Abstract a lot of the
12990 {PLUS,MINUS}_EXPR code to...
12991 (adjust_symbolic_bound): ...here,
12992 (combine_bound): ...here,
12993 (set_value_range_with_overflow): ...and here.
12995 2018-07-02 Aldy Hernandez <aldyh@redhat.com>
12997 * tree-vrp.c (extract_range_from_unary_expr): Abstract ABS_EXPR
12999 (extract_range_from_abs_expr): ...here.
13001 2018-07-02 Eric Botcazou <ebotcazou@adacore.com>
13003 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Do not overrule
13004 -fno-omit-frame-pointer when not optimizing.
13006 2018-07-02 Martin Liska <mliska@suse.cz>
13009 * ipa-pure-const.c (malloc_candidate_p): Revert usage of ::get.
13010 (propagate_nothrow): Likewise.
13012 2018-07-02 Martin Liska <mliska@suse.cz>
13015 * ipa-inline.c (early_inliner): Revert wrongly added ::get call.
13017 2018-07-02 David Malcolm <dmalcolm@redhat.com>
13019 * dumpfile.c (dump_generic_expr_loc): Undo removal of this
13020 function in r262149, changing "loc" param from source_location to
13021 const dump_location_t &.
13022 * dumpfile.h (dump_generic_expr_loc): Undo removal of this
13023 declaration, as above.
13025 2018-07-01 Paul Koning <ni1d@arrl.net>
13027 * common/config/pdp11/pdp11-common.c (pdp11_handle_option): Handle
13028 -munit-asm, -mgnu-asm, -mdec-asm.
13029 * config/pdp11/pdp11-protos.h (pdp11_gen_int_label): New.
13030 (pdp11_output_labelref): New.
13031 (pdp11_output_def): New.
13032 (pdp11_output_addr_vec_elt): New.
13033 * config/pdp11/pdp11.c: Use tab between opcode and operands. Use
13034 %# and %@ format codes.
13035 (pdp11_option_override): New.
13036 (TARGET_ASM_FILE_START_FILE_DIRECTIVE): Define.
13037 (pdp11_output_ident): New.
13038 (pdp11_asm_named_section): New.
13039 (pdp11_asm_init_sections): New.
13040 (pdp11_file_start): New.
13041 (pdp11_file_end): New.
13042 (output_ascii): Use .ascii/.asciz for -mdec-asm.
13043 (pdp11_asm_print_operand): Update %# and %$ for -mdec-asm. Add
13044 %o, like %c but octal.
13045 (pdp11_option_override): New.
13046 * config/pdp11/pdp11.h (TEXT_SECTION_ASM_OP): Update for
13048 (DATA_SECTION_ASM_OP): Ditto.
13049 (READONLY_DATA_SECTION_ASM_OP): New.
13050 (IS_ASM_LOGICAL_LINE_SEPARATOR): New.
13051 (ASM_GENERATE_INTERNAL_LABEL): Use new function.
13052 (ASM_OUTPUT_LABELREF): Ditto.
13053 (ASM_OUTPUT_DEF): Ditto.
13054 (ASM_OUTPUT_EXTERNAL): New.
13055 (ASM_OUTPUT_SOURCE_FILENAME): New.
13056 (ASM_OUTPUT_ADDR_VEC_ELT): Use new function.
13057 (ASM_OUTPUT_SKIP): Update for -mdec-asm.
13058 * config/pdp11/pdp11.md: Use tab between opcode and operands. Use
13059 %# and %@ format codes.
13060 * config/pdp11/pdp11.opt (mgnu-asm): New.
13061 (mdec-asm): Conflicts with -mgnu-asm and -munix-asm.
13062 (munix-asm): Conflicts with -mdec-asm and -mgnu-asm.
13063 * doc/invoke.txt (PDP-11 Options): Add -mgnu-asm.
13065 2018-07-01 Aldy Hernandez <aldyh@redhat.com>
13067 * tree-ssa-threadupdate.c (mark_threaded_blocks): Avoid
13068 dereferencing path[] beyond its length.
13070 (debug_all_paths): New.
13071 (rewire_first_differing_edge): New.
13072 (adjust_paths_after_duplication): New.
13073 (duplicate_thread_path): Call adjust_paths_after_duplication.
13075 (thread_through_all_blocks): Add new argument to
13076 duplicate_thread_path.
13078 2018-06-30 Jim Wilson <jimw@sifive.com>
13080 * config/riscv/predicates.md (p2m1_shift_operand): New.
13081 (high_mask_shift_operand): New.
13082 * config/riscv/riscv.md (lshrsi3_zero_extend_3+1): New combiner
13083 pattern using p2m1_shift_operand.
13084 (lshsi3_zero_extend_3+2): New combiner pattern using
13085 high_mask_shift_operand.
13087 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
13089 * tree-vect-patterns.c (vect_get_external_def_edge): New function,
13091 (vect_recog_rotate_pattern): ...here.
13092 (vect_convert_input): Try to insert casts of invariants in the
13094 * tree-vect-loop-manip.c (vect_loop_versioning): Don't require the
13095 preheader to be empty.
13097 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
13099 * tree-vect-patterns.c (append_pattern_def_seq): Take an optional
13100 vector type. If given, install it in the new statement's
13101 STMT_VINFO_VECTYPE.
13102 (vect_element_precision): New function.
13103 (vect_unpromoted_value): New struct.
13104 (vect_unpromoted_value::vect_unpromoted_value): New function.
13105 (vect_unpromoted_value::set_op): Likewise.
13106 (vect_look_through_possible_promotion): Likewise.
13107 (vect_joust_widened_integer, vect_joust_widened_type): Likewise.
13108 (vect_widened_op_tree, vect_convert_input): Likewise.
13109 (vect_convert_inputs, vect_convert_output): Likewise.
13110 (vect_recog_dot_prod_pattern): Use vect_look_through_possible_promotion
13111 to handle the optional cast of the multiplication result and
13112 vect_widened_op_tree to detect the widened multiplication itself.
13113 Do not require the input and output of promotion casts to have
13114 the same sign, but base the signedness of the operation on the
13115 input rather than the result. If the pattern includes two
13116 promotions, check that those promotions have the same sign.
13117 Do not restrict the MULT_EXPR handling to a double-width result;
13118 handle quadruple-width results and wider. Use vect_convert_inputs
13119 to convert the inputs to the common type.
13120 (vect_recog_sad_pattern): Use vect_look_through_possible_promotion
13121 to handle the optional cast of the ABS result. Also allow a sign
13122 change or a sign extension between the ABS and MINUS.
13123 Use vect_widened_op_tree to detect the widened subtraction and use
13124 vect_convert_inputs to convert the inputs to the common type.
13125 (vect_handle_widen_op_by_const): Delete.
13126 (vect_recog_widen_op_pattern): New function.
13127 (vect_recog_widen_mult_pattern): Use it.
13128 (vect_recog_widen_shift_pattern): Likewise.
13129 (vect_recog_widen_sum_pattern): Use
13130 vect_look_through_possible_promotion to handle the promoted
13133 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
13135 * gimple-iterator.c (gsi_for_stmt): Add a new overload that takes
13136 the containing gimple_seq *.
13137 * gimple-iterator.h (gsi_for_stmt): Declare it.
13138 * tree-vect-patterns.c (vect_recog_dot_prod_pattern)
13139 (vect_recog_sad_pattern, vect_recog_widen_sum_pattern)
13140 (vect_recog_widen_shift_pattern, vect_recog_rotate_pattern)
13141 (vect_recog_vector_vector_shift_pattern, vect_recog_divmod_pattern)
13142 (vect_recog_mask_conversion_pattern): Remove STMT_VINFO_IN_PATTERN_P
13144 (vect_init_pattern_stmt, vect_set_pattern_stmt): New functions,
13146 (vect_mark_pattern_stmts): ...here. Handle cases in which the
13147 statement being replaced is part of an existing pattern
13148 definition sequence, inserting the new pattern statements before
13150 (vect_pattern_recog_1): Don't return a bool. If the statement
13151 is already part of a pattern, instead apply pattern matching
13152 to the pattern definition statements. Don't clear the
13153 STMT_VINFO_RELATED_STMT if is_pattern_stmt_p.
13154 (vect_pattern_recog): Don't break after the first match;
13155 continue processing the pattern definition statements instead.
13156 Don't bail out for STMT_VINFO_IN_PATTERN_P here.
13158 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
13160 * tree-vect-patterns.c (vect_reassociating_reduction_p): New function.
13161 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern)
13162 (vect_recog_widen_sum_pattern): Use it.
13164 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
13166 * tree-vect-loop.c (vectorizable_reduction): Assert that the
13167 phi is not a pattern statement and has not been replaced by
13168 a pattern statement.
13169 * tree-vect-patterns.c (type_conversion_p): Don't check
13170 STMT_VINFO_IN_PATTERN_P.
13171 (vect_recog_vector_vector_shift_pattern): Likewise.
13172 (vect_recog_dot_prod_pattern): Expect vect_is_simple_use to return
13173 the pattern statement rather than the original statement; check
13174 directly for a WIDEN_MULT_EXPR here.
13175 * tree-vect-slp.c (vect_get_and_check_slp_defs): Expect
13176 vect_is_simple_use to return the pattern statement rather
13177 than the original statement; use is_pattern_stmt_p to check
13178 for such a pattern statement.
13179 * tree-vect-stmts.c (process_use): Expect vect_is_simple_use
13180 to return the pattern statement rather than the original statement;
13181 don't do the same transformation here.
13182 (vect_is_simple_use): If the defining statement has been replaced
13183 by a pattern statement, return the pattern statement instead.
13184 Remove the corresponding (local) transformation from the vectype
13187 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
13189 * tree-vectorizer.h (vect_is_simple_use): Move the gimple ** to the
13190 end and default to null.
13191 * tree-vect-loop.c (vect_create_epilog_for_reduction)
13192 (vectorizable_reduction): Update calls accordingly, dropping the
13193 gimple ** argument if the passed-back statement isn't needed.
13194 * tree-vect-patterns.c (vect_get_internal_def, type_conversion_p)
13195 (vect_recog_rotate_pattern): Likewise.
13196 (vect_recog_mask_conversion_pattern): Likewise.
13197 * tree-vect-slp.c (vect_get_and_check_slp_defs): Likewise.
13198 (vect_mask_constant_operand_p): Likewise.
13199 * tree-vect-stmts.c (is_simple_and_all_uses_invariant, process_use):
13200 (vect_model_simple_cost, vect_get_vec_def_for_operand): Likewise.
13201 (get_group_load_store_type, get_load_store_type): Likewise.
13202 (vect_check_load_store_mask, vect_check_store_rhs): Likewise.
13203 (vectorizable_call, vectorizable_simd_clone_call): Likewise.
13204 (vectorizable_conversion, vectorizable_assignment): Likewise.
13205 (vectorizable_shift, vectorizable_operation): Likewise.
13206 (vectorizable_store, vect_is_simple_cond): Likewise.
13207 (vectorizable_condition, vectorizable_comparison): Likewise.
13208 (get_same_sized_vectype, vect_get_mask_type_for_stmt): Likewise.
13209 (vect_is_simple_use): Rename the def_stmt argument to def_stmt_out
13210 and move it to the end. Cope with null def_stmt_outs.
13212 2018-06-30 Bernd Edlinger <bernd.edlinger@hotmail.de>
13214 * Makefile.in (FLAGS_TO_PASS): Add RANLIB_FOR_TARGET.
13216 2018-06-29 Jeff Law <law@redhat.com>
13218 * config/v850/v850.c (v850_legitimate_address_p): Handle large
13219 displacements for TARGET_V850E2V3 and newer.
13220 (TARGET_LRA_P): Remove. Defaults to LRA now.
13221 * config/v850/v850.md (sign23byte_load): Remove.
13222 (unsign23byte_load, sign23hword_load, unsign23hword_load): Likewise.
13223 (23word_load, 23byte_store, 23hword_store, 23word_store): Likewise.
13225 2018-06-29 Martin Liska <mliska@suse.cz>
13228 * coverage.c (coverage_init): Mangle full path name.
13229 * doc/invoke.texi: Document the change.
13230 * gcov-io.c (mangle_path): New.
13231 * gcov-io.h (mangle_path): Likewise.
13232 * gcov.c (mangle_name): Use mangle_path for path mangling.
13234 2018-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13236 * config/arm/arm.c (output_move_double): Don't allow STRD instructions
13237 if starting source register is not even.
13239 2018-06-29 Martin Liska <mliska@suse.cz>
13241 PR tree-optimization/86263
13242 * tree-switch-conversion.c (switch_decision_tree::try_switch_expansion):
13243 Make edge redirection.
13245 2018-06-29 David Malcolm <dmalcolm@redhat.com>
13247 * dumpfile.c (dump_loc): Add indentation based on scope depth.
13248 (dump_scope_depth): New variable.
13249 (get_dump_scope_depth): New function.
13250 (dump_begin_scope): New function.
13251 (dump_end_scope): New function.
13252 * dumpfile.h (get_dump_scope_depth): New declaration.
13253 (dump_begin_scope): New declaration.
13254 (dump_end_scope): New declaration.
13255 (class auto_dump_scope): New class.
13256 (AUTO_DUMP_SCOPE): New macro.
13257 * tree-vectorizer.h (DUMP_VECT_SCOPE): Reimplement in terms of
13260 2018-06-29 Richard Biener <rguenther@suse.de>
13262 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Assert
13263 compute_all_dependences succeeds.
13264 * tree-vect-loop.c (vect_get_datarefs_in_loop): Fail early if we
13265 exceed --param loop-max-datarefs-for-datadeps.
13267 2018-06-29 Jakub Jelinek <jakub@redhat.com>
13269 * config/rs6000/t-rs6000: Append rs6000-modes.h to TM_H.
13271 2018-06-28 Uros Bizjak <ubizjak@gmail.com>
13274 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): Use
13275 alternative 0 in preferred_for_speed attribute.
13277 2018-06-28 Paul Koning <ni1d@arrl.net>
13279 * config/pdp11/pdp11-protos.h (pdp11_shift_length): New function.
13280 * config/pdp11/pdp11.c (pdp11_shift_length): New function.
13281 * config/pdp11/pdp11.h (ADJUST_INSN_LENGTH): Remove.
13282 * config/pdp11/pdp11.md: Correct "length" attribute calculation
13283 for shift insn patterns.
13285 2018-06-28 David Malcolm <dmalcolm@redhat.com>
13287 * cgraph.c (cgraph_node::get_body): Replace assignments to
13288 "dump_file" with calls to set_dump_file.
13289 * dumpfile.c (alt_dump_file): Make static, and group with...
13290 (alt_flags): ...this definition.
13291 (dumps_are_enabled): New variable.
13292 (refresh_dumps_are_enabled): New function.
13293 (set_dump_file): New function.
13294 (set_alt_dump_file): New function.
13295 (gcc::dump_manager::dump_start): Replace assignments to
13296 "dump_file" and "alt_dump_file" with calls to set_dump_file and
13298 (gcc::dump_manager::dump_finish): Likewise.
13299 * dumpfile.h (alt_dump_file): Delete decl.
13300 (dumps_are_enabled): New variable decl.
13301 (set_dump_file): New function decl.
13302 (dump_enabled_p): Rewrite in terms of new "dumps_are_enabled"
13304 * tree-nested.c (lower_nested_functions): Replace assignments to
13305 "dump_file" with calls to set_dump_file.
13307 2018-06-28 Eric Botcazou <ebotcazou@adacore.com>
13309 * tree-cfg.c (verify_gimple_in_cfg): Call verify_location on the
13310 goto_locus of each outgoing edge of each basic block.
13312 2018-06-28 Richard Biener <rguenther@suse.de>
13314 * dwarf2out.c (decl_scope_table): Remove.
13315 (push_decl_scope): Likewise.
13316 (pop_decl_scope): Likewise.
13317 (gen_type_die_for_member): Do not call push/pop_decl_scope.
13318 (gen_struct_or_union_type_die): Likewise.
13319 (gen_tagged_type_die): Likewise.
13320 (dwarf2out_init): Do not initialize decl_scope_table.
13321 (dwarf2out_c_finalize): Do not free it.
13323 2018-06-28 Richard Biener <rguenther@suse.de>
13325 * dwarf2out.c (gen_subprogram_die): Use is_unit_die when
13326 deciding whether to not re-use a DIE.
13328 2018-06-28 Richard Biener <rguenther@suse.de>
13330 * dwarf2out.c (gen_subprogram_die): Always re-use DIEs with an
13331 DW_AT_abstract_origin attribute.
13333 2018-06-28 Martin Liska <mliska@suse.cz>
13335 * tree-switch-conversion.c (jump_table_cluster::can_be_handled):
13336 Use newly introduced constants.
13337 * tree-switch-conversion.h (struct jump_table_cluster):
13338 Define max_ratio_for_size and max_ratio_for_speed.
13340 2018-06-28 Martin Liska <mliska@suse.cz>
13342 * tree-switch-conversion.c (jump_table_cluster::find_jump_tables):
13343 Add new checking assert to catch invalid state.
13344 (jump_table_cluster::can_be_handled): Handle single case
13346 (jump_table_cluster::is_beneficial): Bail out for such case.
13347 (bit_test_cluster::find_bit_tests):
13348 Add new checking assert to catch invalid state.
13349 (bit_test_cluster::can_be_handled): Handle single case
13351 (bit_test_cluster::is_beneficial): Bail out for such case.
13352 (switch_decision_tree::analyze_switch_statement):
13355 2018-06-28 Martin Liska <mliska@suse.cz>
13357 * common.opt: Introduce -completion option.
13358 * gcc.c (driver_handle_option): Handle it.
13359 (driver::main): Print completions if completion
13361 * opt-suggestions.c (option_proposer::get_completions):
13363 (option_proposer::suggest_completion): Likewise.
13364 (option_proposer::find_param_completions): Likewise.
13365 (verify_autocompletions): Likewise.
13366 (test_completion_valid_options): Likewise.
13367 (test_completion_valid_params): Likewise.
13368 (in_completion_p): Likewise.
13369 (empty_completion_p): Likewise.
13370 (test_completion_partial_match): Likewise.
13371 (test_completion_garbage): Likewise.
13372 (opt_proposer_c_tests): Likewise.
13373 * opt-suggestions.h: Declare new functions.
13374 * opts.c (common_handle_option): Handle OPT__completion_.
13375 * selftest-run-tests.c (selftest::run_tests): Add
13376 opt_proposer_c_tests.
13377 * selftest.c (assert_str_startswith): New.
13378 * selftest.h (assert_str_startswith): Likewise.
13379 (opt_proposer_c_tests): New.
13380 (ASSERT_STR_STARTSWITH): Likewise.
13382 2018-06-28 Martin Liska <mliska@suse.cz>
13384 * Makefile.in: Add opt-suggestions.o.
13385 * gcc-main.c: Include opt-suggestions.h.
13386 * gcc.c (driver::driver): Likewise.
13387 (driver::~driver): Remove m_option_suggestions.
13388 (driver::build_option_suggestions): Moved to option_proposer.
13389 (driver::suggest_option): Likewise.
13390 (driver::handle_unrecognized_options): Use option_proposer.
13391 * gcc.h (class driver): Add new memver m_option_proposer.
13392 * opt-suggestions.c: New file.
13393 * opt-suggestions.h: New file.
13395 2018-06-28 Martin Liska <mliska@suse.cz>
13397 * vec.h (class auto_string_vec): New (moved from auto_argvec).
13398 (auto_string_vec::~auto_string_vec): Likewise.
13400 2018-06-28 Eric Botcazou <ebotcazou@adacore.com>
13402 * tree-inline.h (struct copy_body_data): Move remapping_type_depth and
13403 prevent_decl_creation_for_types fields up and add reset_location field.
13404 * tree-inline.c (remap_gimple_stmt): Force input_location on the new
13405 statement if id->reset_location is true.
13406 (copy_edges_for_bb): Do not set goto_locus on the new edges if
13407 id->reset_location is true.
13408 (copy_phis_for_bb): Force input_location on the arguments if
13409 id->reset_location is true.
13410 (expand_call_inline): Set id->reset_location if DECL_IGNORED_P
13411 is set on the function to be inlined.
13413 2018-06-27 Stephan Bergmann <sbergman@redhat.com>
13415 * doc/invoke.texi (Debugging Options): Add -gsplit-dwarf.
13417 2018-06-27 Dimitar Dimitrov <dimitar@dinux.eu>
13419 * lra-eliminations.c (update_reg_eliminate): Mark all spanning hard
13420 registers for Pmode.
13421 * lra-lives.c (check_pseudos_live_through_calls): Mark all spanning
13422 hard registers for the clobbered pseudo.
13424 2018-06-27 Paul Koning <ni1d@arrl.net>
13426 * common/config/pdp11/pdp11-common.c (pdp11_handle_option): Handle
13427 mutually exclusive options.
13428 * config/pdp11/constraints.md (h): New constraint.
13429 (O): Update definition to match shift code generation.
13430 (D): New constraint.
13431 * config/pdp11/pdp11-modes.def (CCNZ): Define mode.
13433 * config/pdp11/pdp11-protos.h (int_no_side_effect_operand): New
13435 (output_jump): Change arguments.
13436 (pdp11_fixed_cc_regs): New function.
13437 (pdp11_cc_mode): Ditto.
13438 (pdp11_expand_shift): Ditto.
13439 (pdp11_assemble_shift): Ditto.
13440 (pdp11_small_shift): Ditto.
13441 (pdp11_branch_cost): Remove.
13442 * config/pdp11/pdp11.c (pdp11_assemble_integer): Remove comments
13444 (pdp11_register_move_cost): Update for CC registers.
13445 (pdp11_rtx_costs): Add case for LSHIFTRT.
13446 (pdp11_output_jump): Add CCNZ mode conditional branches.
13447 (notice_update_cc_on_set): Remove.
13448 (pdp11_cc_mode): New function.
13449 (simple_memory_operand): Correct pre/post decrement case.
13450 (no_side_effect_operand): New function.
13451 (pdp11_regno_reg_class): Add CC_REGS class.
13452 (pdp11_fixed_cc_regs): New function.
13453 (pdp11_small_shift): New function.
13454 (pdp11_expand_shift): New function to expand shift insns.
13455 (pdp11_assemble_shift): New function to output shifts.
13456 (pdp11_branch_cost): Remove.
13457 (pdp11_modes_tieable_p): Make QI/HI modes tieable.
13458 * config/pdp11/pdp11.h (SIZE_TYPE): Ensure 16-bit type.
13459 (WCHAR_TYPE): Ditto.
13460 (PTRDIFF_TYPE): Ditto.
13461 (ADJUST_INSN_LENGTH): New macro.
13462 (FIXED_REGISTERS): Add CC registers.
13463 (CALL_USED_REGISTERS): Ditto.
13464 (reg_class): Ditto.
13465 (REG_CLASS_NAMES): Ditto.
13466 (REG_CLASS_CONTENTS): Ditto.
13467 (SELECT_CC_MODE): Use new function.
13468 (TARGET_FLAGS_REGNUM): New macro.
13469 (TARGET_FIXED_CONDITION_CODE_REGS): Ditto.
13470 (cc0_reg_rtx): Remove.
13471 (CC_STATUS_MDEP): Remove.
13472 (CC_STATUS_MDEFP_INIT): Remove.
13473 (CC_IN_FPU): Remove.
13474 (NOTICE_UPDATE_CC): Remove.
13475 (REGISTER_NAMES): Add CC registers.
13476 (BRANCH_COST): Change to constant 1.
13477 * config/pdp11/pdp11.md: Rewrite for CCmode condition code
13479 * config/pdp11/pdp11.opt (mbcopy): Remove.
13480 (mbcopy-builtin): Remove.
13481 (mbranch-cheap): Remove.
13482 (mbranch-expensive): Remove.
13483 * config/pdp11/predicates.md (expand_shift_operand): Update to
13484 match shift code generation.
13485 (ccnz_operator): New predicate.
13486 * doc/invoke.texi (PDP-11 Options): Remove deleted options
13487 -mbcopy, -mbcopy-builtin, -mbranch-cheap, -mbranch-expensive.
13488 Remove non-existent option -mabshi, -mno-abshi. Document mutually
13490 * doc/md.texi (PDP-11): Document new D and h constraints. Update
13491 description of O constraint.
13493 2018-06-27 Jeff Law <law@redhat.com>
13494 Austin Law <austinklaw@gmail.com>
13496 * config/v850/v850.md (addsi3_set_flags): New pattern.
13497 (subsi3_set_flags, negsi2_set_flags, andsi3_set_flags): Likewise.
13498 (iorsi3_set_flags, xorsi3_set_flags, one_cmplsi2_set_flags): Likewise.
13499 (zero_extendhisi2_v850_set_flags): Likewise.
13500 (zero_extendqisi2_v850_set_flags): Likewise.
13501 (ashlsi3_set_flags, ashlsi3_v850e2_set_flags): Likewise.
13502 (lshrsi3_set_flags, lshrsi3_v850e2_set_flags): Likewise.
13503 (ashrsi3_set_flags, ashrsi3_v850e2_set_flags): Likewise.
13505 * config/v850/v850-protos.h (notice_update_cc): Remove.
13506 * config/v850/v850.c (v850_compare_op0, v850_compare_op1): Remove.
13507 (v850_print_operand): Handle 'D' and "d".
13508 (v850_select_cc_mode): Remove ATTRIBUTE_UNUSED for last argument.
13509 Add handling of arithmetic/logical operations compared against zero.
13510 (v850_gen_float_compare): Remove ATTRIBUTE_UNUSED for last argument.
13511 Do not look at v850_compare_op, instead get mode from last argument.
13512 (v850_gen_compare): Remove
13513 (increment_stack): Use addsi3_clobber_flags to avoid splitting failure
13514 after reload for prologue insns.
13515 (expand_prologue): Account for CLOBBER of CC_REGNUM in various
13517 (construct_save_jarl): Likewise.
13518 (TARGET_FLAGS_REGNUM): Define.
13519 * config/v850/v850.h (v850_compare_op0, v850_compare_op1): Remove.
13520 (NOTICE_UPDATE_CC): Remove.
13521 * config/v850/v850.md (v850_tst1): Use (reg:CCZ CC_REGNUM) rather
13522 than cc0. Conditionalize on reload_completed.
13523 (cmpsi_insn, setfcc_insn): Likewise.
13524 (tst1 splitter): Turn into define_and_split which sets the flags
13526 (cstoresi4, cbranchsf4, cbranchdf4, cbranchsi4_insn): Likewise.
13527 (cbranchsi4, branch_normal, branch_invert): Do not expose cc0 here.
13528 (cstoresf4, cstoredf4): Clobber the flags.
13529 (cmpsi, cmpsf, cmpdf): Remove expanders.
13530 (setf_insn): Remove pattern.
13531 (addsi3): Turn into define_and_split which clobbers the flags after
13532 reload and a suitable pattern (addsi3_clobber_flags) for use after
13534 (subsi3, negsi2, andsi3, iorsi3, xorsi3, one_cmplsi2) Likewise.
13535 (ashlsi3, ashlsi3_v850e2, lshrsi3, lsh4si3_v850e2): Likewise.
13536 (ashrsi3, ashrsi3_v850e2): Likewise.
13537 (bins): Clobber the flags.
13538 (movsicc_normal_cc, movsicc_normal, movsicc_tst1): Likewise.
13539 (movsicc_tst1_revesed, sasf, swap and rotate patterns): Likewise.
13540 (fix_loop_counter, call_internal_short, call_internal_long): Likewise.
13541 (call_value_internal_short, call_value_internal_long): Likewise.
13542 (callt_save_interrupt, callt_return_interrupt): Likewise.
13543 (save_interrupt, return_interrupt): Likewise.
13544 (callt_save_all_interrupt, save_all_interrupt): Likewise.
13545 (_save_all_interrupt, callt_restore_all_interrupt): Likewise.
13546 (restore_all_interrupt, _restore_all_interrupt): Likewise.
13547 (All FP comparisons): Only allow after reload has completed.
13549 (divh, divhu): Tweak output template.
13550 (branch_z_normal, branch_z_invert): Remove
13551 (branch_nz_normal, branch_nz_invert): Likewise.
13552 (extendhisi_insn, extendqisi_insn): Do not clobber flags.
13554 * config/v850/v850-modes.def (CCZ, CCNZ): Add new modes.
13555 * config/v850/v850.c (notice_update_cc): Remove.
13556 * config/v850/v850.h (CC_OVERFLOW_UNUSABLE): Remove
13557 (CC_NO_CARRY): Likewise.
13558 (NOTICE_UPDATE_CC): Define to nothing.
13559 * config/v850/v850.md: Remove block comment on cc0 handling
13560 Remove "cc" attribute from all patterns. Remove cc_status handling
13561 from all patterns. Minor formatting fixes.
13563 2018-06-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13565 * config/aarch64/aarch64-cores.def (cortex-a76): New entry.
13566 (cortex-a76.cortex-a55): Likewise.
13567 * config/aarch64/aarch64-tune.md: Regenerate.
13568 * doc/invoke.texi (AArch64 Options): Document cortex-a76 and
13569 cortex-a76.cortex-a55.
13571 2018-06-27 Jeff Law <law@redhat.com>
13573 * config/v850/t-v850 (MULTILIB_OPTIONS): Remove 8byte-align.
13574 (MULTILIB_DIRNAMES): Similarly.
13576 2018-06-27 Eric Botcazou <ebotcazou@adacore.com>
13578 * gimple.h (gimple_return_retbnd): Delete.
13579 (gimple_return_set_retbnd): Likewise.
13580 * cgraphunit.c (cgraph_node::expand_thunk): Remove call to
13581 gimple_return_set_retbnd.
13582 * gimple-pretty-print.c (dump_gimple_return): Remove call to
13583 gimple_return_retbnd and adjust.
13584 * tree-inline.h (struct copy_body_data): Remove retbnd field.
13585 * tree-inline.c (remap_gimple_stmt): Remove handling of retbnd.
13586 Explicitly return NULL in a couple more cases. Move assertion
13587 on debug statements and remove unreachable code.
13588 (reset_debug_binding): Do not test id->retbnd.
13589 (expand_call_inline): Do not set it.
13591 2018-06-27 Rasmus Villemoes <rv@rasmusvillemoes.dk>
13593 * configure.ac: Add --disable-gcov option.
13594 * configure: Regenerate.
13595 * Makefile.in: Honour @enable_gcov@.
13596 * doc/install.texi: Document --disable-gcov.
13598 2018-06-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13600 * config/arm/arm-cpus.in (cortex-a76): New entry.
13601 (cortex-a76.cortex-a55): Likewise.
13602 * config/arm/arm-tables.opt: Regenerate.
13603 * config/arm/arm-tune.md: Likewise.
13604 * config/arm/driver-arm.c (arm_cpu_table): Add Cortex-A76 entry.
13605 * doc/invoke.texi (ARM Options): Document cortex-a76 and
13606 cortex-a76.cortex-a55.
13608 2018-06-27 Tamar Christina <tamar.christina@arm.com>
13611 * config/aarch64/aarch64.md (*movhf_aarch64): Add dup v0.4h pattern.
13613 2018-06-27 Siddhesh Poyarekar <siddhesh@sourceware.org>
13615 * config/aarch64/aarch64.h (CALL_USE_REGISTERS): Fix obsolete
13617 (EPILOGUE_USES): Likewise.
13619 2018-06-26 Eric Botcazou <ebotcazou@adacore.com>
13621 * tree-inline.c (remap_location): New function extracted from...
13622 (copy_edges_for_bb): Add ID parameter. Remap goto_locus.
13623 (copy_phis_for_bb): ...here. Call remap_location.
13624 (copy_cfg_body): Adjust call to copy_edges_for_bb.
13626 2018-06-26 Aaron Sawdey <acsawdey@linux.ibm.com>
13628 * config/rs6000/rs6000-string.c (expand_block_clear): Don't use
13629 unaligned vsx for 16B memset.
13631 2018-06-26 Segher Boessenkool <segher@kernel.crashing.org>
13634 * config/rs6000/rs6000.c (rs6000_init_builtins): Do not set
13635 ieee128_float_type_node to long_double_type_node unless
13636 TARGET_LONG_DOUBLE_128 is set.
13638 2018-06-26 David Malcolm <dmalcolm@redhat.com>
13640 * cfgloop.c (get_loop_location): Convert return type from
13641 location_t to dump_user_location_t, replacing INSN_LOCATION lookups
13642 by implicit construction from rtx_insn *, and using
13643 dump_user_location_t::from_function_decl for the fallback case.
13644 * cfgloop.h (get_loop_location): Convert return type from
13645 location_t to dump_user_location_t.
13646 * cgraphunit.c (walk_polymorphic_call_targets): Update call to
13647 dump_printf_loc to pass in a dump_location_t rather than a
13648 location_t, via the gimple stmt.
13649 * coverage.c (get_coverage_counts): Update calls to
13650 dump_printf_loc to pass in dump_location_t rather than a
13652 * doc/optinfo.texi (Dump types): Convert example of
13653 dump_printf_loc from taking "locus" to taking "insn". Update
13654 description of the "_loc" calls to cover dump_location_t.
13655 * dumpfile.c: Include "backend.h", "gimple.h", "rtl.h", and
13657 (dump_user_location_t::dump_user_location_t): New constructors,
13658 from gimple *stmt and rtx_insn *.
13659 (dump_user_location_t::from_function_decl): New function.
13660 (dump_loc): Make static.
13661 (dump_gimple_stmt_loc): Convert param "loc" from location_t to
13662 const dump_location_t &.
13663 (dump_generic_expr_loc): Delete.
13664 (dump_printf_loc): Convert param "loc" from location_t to
13665 const dump_location_t &.
13666 (selftest::test_impl_location): New function.
13667 (selftest::dumpfile_c_tests): New function.
13668 * dumpfile.h: Include "profile-count.h".
13669 (class dump_user_location_t): New class.
13670 (struct dump_impl_location_t): New struct.
13671 (class dump_location_t): New class.
13672 (dump_printf_loc): Convert 2nd param from source_location to
13673 const dump_location_t &.
13674 (dump_generic_expr_loc): Delete.
13675 (dump_gimple_stmt_loc): Convert 2nd param from source_location to
13676 const dump_location_t &.
13677 * gimple-fold.c (fold_gimple_assign): Update call to
13678 dump_printf_loc to pass in a dump_location_t rather than a
13679 location_t, via the gimple stmt.
13680 (gimple_fold_call): Likewise.
13681 * gimple-loop-interchange.cc
13682 (loop_cand::analyze_iloop_reduction_var): Update for change to
13683 check_reduction_path.
13684 (tree_loop_interchange::interchange): Update for change to
13685 find_loop_location.
13686 * graphite-isl-ast-to-gimple.c (scop_to_isl_ast): Update for
13687 change in return-type of find_loop_location.
13688 (graphite_regenerate_ast_isl): Likewise.
13689 * graphite-optimize-isl.c (optimize_isl): Likewise.
13690 * graphite.c (graphite_transform_loops): Likewise.
13691 * ipa-devirt.c (ipa_devirt): Update call to dump_printf_loc to
13692 pass in a dump_location_t rather than a location_t, via the
13694 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
13695 * ipa.c (walk_polymorphic_call_targets): Likewise.
13696 * loop-unroll.c (report_unroll): Convert "locus" param from
13697 location_t to dump_location_t.
13698 (decide_unrolling): Update for change to get_loop_location's
13700 * omp-grid.c (struct grid_prop): Convert field "target_loc" from
13701 location_t to dump_user_location_t.
13702 (grid_find_single_omp_among_assignments_1): Updates calls to
13703 dump_printf_loc to pass in a dump_location_t rather than a
13704 location_t, via the gimple stmt.
13705 (grid_parallel_clauses_gridifiable): Convert "tloc" from
13706 location_t to dump_location_t. Updates calls to dump_printf_loc
13707 to pass in a dump_location_t rather than a location_t, via the
13709 (grid_inner_loop_gridifiable_p): Likewise.
13710 (grid_dist_follows_simple_pattern): Likewise.
13711 (grid_gfor_follows_tiling_pattern): Likewise.
13712 (grid_target_follows_gridifiable_pattern): Likewise.
13713 (grid_attempt_target_gridification): Convert initialization
13714 of local "grid" from memset to zero-initialization; FIXME: does
13715 this require C++11? Update call to dump_printf_loc to pass in a
13716 optinfo_location rather than a location_t, via the gimple stmt.
13717 * profile.c (read_profile_edge_counts): Updates call to
13718 dump_printf_loc to pass in a dump_location_t rather than a
13720 (compute_branch_probabilities): Likewise.
13721 * selftest-run-tests.c (selftest::run_tests): Call
13723 * selftest.h (dumpfile_c_tests): New decl.
13724 * tree-loop-distribution.c (pass_loop_distribution::execute):
13725 Update for change in return type of find_loop_location.
13726 * tree-parloops.c (parallelize_loops): Likewise.
13727 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Convert
13728 "locus" from location_t to dump_user_location_t.
13729 (canonicalize_loop_induction_variables): Likewise.
13730 * tree-ssa-loop-ivopts.c (tree_ssa_iv_optimize_loop): Update
13731 for change in return type of find_loop_location.
13732 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update call
13733 to dump_printf_loc to pass in a dump_location_t rather than a
13734 location_t, via the stmt.
13735 * tree-ssa-sccvn.c (eliminate_dom_walker::before_dom_children):
13737 * tree-vect-loop-manip.c (find_loop_location): Convert return
13738 type from source_location to dump_user_location_t.
13739 (vect_do_peeling): Update for above change.
13740 (vect_loop_versioning): Update for change in type of
13742 * tree-vect-loop.c (check_reduction_path): Convert "loc" param
13743 from location_t to dump_user_location_t.
13744 (vect_estimate_min_profitable_iters): Update for change in type
13746 * tree-vect-slp.c (vect_print_slp_tree): Convert param "loc" from
13747 location_t to dump_location_t.
13748 (vect_slp_bb): Update for change in type of vect_location.
13749 * tree-vectorizer.c (vect_location): Convert from source_location
13750 to dump_user_location_t.
13751 (try_vectorize_loop_1): Update for change in vect_location's type.
13752 (vectorize_loops): Likewise.
13753 (increase_alignment): Likewise.
13754 * tree-vectorizer.h (vect_location): Convert from source_location
13755 to dump_user_location_t.
13756 (find_loop_location): Convert return type from source_location to
13757 dump_user_location_t.
13758 (check_reduction_path): Convert 1st param from location_t to
13759 dump_user_location_t.
13760 * value-prof.c (check_counter): Update call to dump_printf_loc to
13761 pass in a dump_user_location_t rather than a location_t; update
13762 call to error_at for change in type of "locus".
13763 (check_ic_target): Update call to dump_printf_loc to
13764 pass in a dump_user_location_t rather than a location_t, via the
13767 2018-06-26 Robin Dapp <rdapp@linux.vnet.ibm.com>
13769 * config/s390/s390.h (enum processor_flags): Do not use
13770 default tune parameter when -march was specified.
13772 2018-06-26 Jakub Jelinek <jakub@redhat.com>
13775 * config/i386/i386.md (setcc + movzbl to xor + setcc peephole2s):
13776 Check reg_overlap_mentioned_p in addition to reg_set_p with the same
13779 2018-06-26 Richard Biener <rguenther@suse.de>
13781 PR tree-optimization/86287
13783 * tree-vect-loop.c (vect_transform_loop_stmt): Fix read-after-free.
13784 (vect_analyze_loop): Initialize n_stmts.
13786 2018-06-26 Richard Biener <rguenther@suse.de>
13788 PR middle-end/86271
13789 * fold-const.c (fold_convertible_p): Pointer extension
13792 2018-06-26 Alexandre Oliva <aoliva@redhat.com>
13795 * dwarf2out.c (loc_list_has_views): Adjust comments.
13796 (dw_loc_list): Split single cross-partition range with
13799 2018-06-25 Jeff Law <law@redhat.com>
13801 * common/config/v850/v850-common.c (TARGET_DEFAULT_TARGET_FLAGS): Turn
13802 on -mbig-switch by default.
13804 * config/v850/predicates.md (const_float_1_operand): Fix match_code
13806 (const_float_0_operand): Remove unused predicate.
13807 * config/v850/v850.md (define_constants): Remove UNSPEC_LOOP.
13808 (define_c_enum unspec): Add LOOP, RCP and RSQRT constants.
13809 (recipsf2): New expander. Original pattern now called
13811 (recipdf2, recipdf2_insn): Similarly.
13812 (rsqrtsf2, rsqrtsf2_insn): Similarly
13813 (rsqrtdf2, rsqrtdf2_insn): Similarly
13815 2018-06-26 Gerald Pfeifer <gerald@pfeifer.com>
13817 * ginclude/stddef.h: Remove an obsolete comment on FreeBSD 5.
13818 Simplify logic for FreeBSD (twice).
13820 2018-06-25 Martin Sebor <msebor@redhat.com>
13822 PR tree-optimization/86204
13823 * tree-ssa-strlen.c (handle_builtin_strlen): Avoid storing
13824 a strnlen result if it's less than the length of the string.
13826 2018-06-25 Martin Sebor <msebor@redhat.com>
13828 PR tree-optimization/85700
13829 * gimple-fold.c (gimple_fold_builtin_strncat): Adjust comment.
13830 * tree-ssa-strlen.c (is_strlen_related_p): Handle integer subtraction.
13831 (maybe_diag_stxncpy_trunc): Distinguish strncat from strncpy.
13833 2018-06-25 Martin Sebor <msebor@redhat.com>
13835 * doc/extend.texi (Zero-length arrays): Update and clarify.
13837 2018-06-25 Michael Meissner <meissner@linux.ibm.com>
13839 * config.gcc (powerpc64le*): Revert January 16th, 2018 patch that
13840 added IEEE/IBM long double multilib support on PowerPC little
13841 endian Linux systems.
13842 * config/rs6000/linux64.h (MULTILIB_DEFAULTS_IEEE): Likewise.
13843 (MULTILIB_DEFAULTS): Likewise.
13844 * config/rs6000/rs6000.c (rs6000_option_override_internal):
13846 * config/rs6000/rs6000.h (TARGET_IEEEQUAD_MULTILIB): Likewise.
13847 * config/rs6000/t-ldouble-linux64le-ibm: Delete, no longer used.
13848 * config/rs6000/t-ldouble-linux64le-ieee: Delete, no longer used.
13850 2018-06-25 Alexander Monakov <amonakov@ispras.ru>
13852 PR middle-end/86311
13853 * sort.cc (REORDER_23): Avoid memcpy with same destination and source.
13854 (REORDER_45): Likewise.
13856 2018-06-25 Jeff Law <law@redhat.com>
13858 * config/v850/v850.md (divmodhi4): Make sure to sign extend the
13859 dividend to 32 bits. Adjust length.
13860 (udivmodhi4): Cleanup output template. Fix length.
13862 2018-06-25 Carl Love <cel@us.ibm.com>
13864 * config/rs6000/vsx.md: Change word selector to prefered location.
13866 2018-06-25 Richard Biener <rguenther@suse.de>
13868 PR tree-optimization/86304
13869 * tree-vectorizer.c (vectorize_loops): Walk over new possibly
13870 epilogue-if-converted loops as well.
13872 2018-06-25 Jan Hubicka <hubicka@ucw.cz>
13874 * lto-section-out.c (lto_begin_section): Do not print section
13875 name for noaddr and unnumbered dumps.
13877 2018-06-25 Richard Biener <rguenther@suse.de>
13879 * tree-vectorizer.h (struct vec_info_shared): New structure
13880 with parts split out from struct vec_info and loop_nest from
13881 struct _loop_vec_info.
13882 (struct vec_info): Adjust accordingly.
13883 (struct _loop_vec_info): Likewise.
13884 (LOOP_VINFO_LOOP_NEST): Adjust.
13885 (LOOP_VINFO_DATAREFS): Likewise.
13886 (LOOP_VINFO_DDRS): Likewise.
13887 (struct _bb_vec_info): Likewise.
13888 (BB_VINFO_DATAREFS): Likewise.
13889 (BB_VINFO_DDRS): Likewise.
13890 (struct _stmt_vec_info): Add dr_aux member.
13891 (DR_VECT_AUX): Adjust to refer to member of DR_STMTs vinfo.
13892 (DR_MISALIGNMENT_UNINITIALIZED): New.
13893 (set_dr_misalignment): Adjust.
13894 (dr_misalignment): Assert misalign isn't DR_MISALIGNMENT_UNINITIALIZED.
13895 (vect_analyze_loop): Adjust prototype.
13896 (vect_analyze_loop_form): Likewise.
13897 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
13898 Compute dependences lazily.
13899 (vect_record_base_alignments): Use shared datarefs/ddrs.
13900 (vect_verify_datarefs_alignment): Likewise.
13901 (vect_analyze_data_refs_alignment): Likewise.
13902 (vect_analyze_data_ref_accesses): Likewise.
13903 (vect_analyze_data_refs): Likewise.
13904 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Add
13905 constructor parameter for shared part.
13906 (vect_analyze_loop_form): Pass in shared part and adjust.
13907 (vect_analyze_loop_2): Pass in storage for the number of
13908 stmts. Move loop nest finding to the caller. Compute
13910 (vect_analyze_loop): Pass in shared part.
13911 (vect_transform_loop): Verify shared datarefs are unchanged.
13912 * tree-vect-slp.c (_bb_vec_info::_bb_vec_info): Add
13913 constructor parameter for shared part.
13914 (vect_slp_analyze_bb_1): Pass in shared part and adjust.
13915 (vect_slp_bb): Verify shared datarefs are unchanged before
13917 * tree-vect-stmts.c (ensure_base_align): Adjust for DR_AUX
13919 (new_stmt_vec_info): Initialize DR_AUX misalignment to
13920 DR_MISALIGNMENT_UNINITIALIZED.
13921 * tree-vectorizer.c (vec_info::vec_info): Add constructor
13922 parameter for shared part.
13923 (vec_info::~vec_info): Adjust.
13924 (vec_info_shared::vec_info_shared): New.
13925 (vec_info_shared::~vec_info_shared): Likewise.
13926 (vec_info_shared::save_datarefs): Likewise.
13927 (vec_info_shared::check_datarefs): Likewise.
13928 (try_vectorize_loop_1): Construct shared part live for analyses
13929 of a single loop for multiple vector sizes.
13930 * tree-parloops.c (gather_scalar_reductions): Adjust.
13932 2018-06-25 Richard Biener <rguenther@suse.de>
13934 * tree-vect-data-refs.c (vect_find_stmt_data_reference): Modify
13935 DR for SIMD lane accesses here and mark DR with (void *)-1 aux.
13936 (vect_analyze_data_refs): Remove similar code from here and
13937 simplify accordingly.
13939 2018-06-25 Richard Biener <rguenther@suse.de>
13941 * tree-vect-data-refs.c (vect_check_gather_scatter): Fail
13942 for reverse storage order accesses rather than asserting
13943 they cannot happen here.
13945 2018-06-25 Tom de Vries <tdevries@suse.de>
13948 * config/i386/i386.md (define_insn "*tls_global_dynamic_64_<mode>"):
13949 Use data16 instead of .byte for insn prefix.
13951 2018-06-25 Andreas Krebbel <krebbel@linux.ibm.com>
13954 * parser.c (make_char_string_pack): Pass this literal chars
13955 through cpp_interpret_string.
13956 (cp_parser_userdef_numeric_literal): Check the result of
13957 make_char_string_pack.
13959 2018-06-24 Maya Rashish <coypu@sdf.org>
13961 * ginclude/stddef.h: Simplify conditions around avoiding
13962 re-definition of __size_t.
13964 2018-06-22 Jan Hubicka <hubicka@ucw.cz>
13966 * lto-streamer-out.c (tree_is_indexable): Make LABEL_DECL nonindexable
13967 unless it is forced or nonlocal; assert that we stream no IMPORTED_DECL.
13969 2018-06-22 Maya Rashish <coypu@sdf.org>
13971 * doc/invoke.texi (mno-fancy-math-387): Update for changes
13972 made to OpenBSD and NetBSD through the years.
13974 2018-06-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
13976 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Change
13977 behavior of vec_pack (vector double, vector double) to match
13978 behavior of vec_float2 (vector double, vector double).
13980 2018-06-22 Olivier Hainque <hainque@adacore.com>
13982 * gimplify.c (gimplify_function_tree): Prevent creation
13983 of a trampoline for the address of the current function
13984 passed to entry/exit instrumentation hooks.
13986 2018-06-22 Aaron Sawdey <acsawdey@linux.ibm.com>
13989 * config/rs6000/rs6000-string.c (expand_strn_compare): Handle -m32
13992 2018-06-22 Martin Liska <mliska@suse.cz>
13994 PR tree-optimization/86263
13995 * tree-switch-conversion.c (jump_table_cluster::find_jump_tables):
13996 Bail out if is_enabled is false.
13997 * tree-switch-conversion.h (jump_table_cluster::is_enabled):
13999 (jump_table_cluster::is_enabled): New function.
14001 2018-06-22 Jan Hubicka <hubicka@ucw.cz>
14003 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not stream
14004 BINFO_BASE_ACCESSES and BINFO_VPTR_FIELD.
14005 * tree-streamer-in.c (streamer_read_tree_bitfields): Likewise.
14006 (lto_input_ts_binfo_tree_pointers): Likewise.
14007 * tree-streamer-out.c (streamer_write_tree_bitfields,
14008 write_ts_binfo_tree_pointers): Likewise.
14009 * tree.c (free_lang_data_in_binfo): Clear BINFO_VPTR_FIELD.
14011 2018-06-22 Jan Hubicka <hubicka@ucw.cz>
14013 * tree.c (free_lang_data_in_type): Free all TYPE_VFIELDs.
14015 2018-06-22 Martin Liska <mliska@suse.cz>
14017 * symbol-summary.h (get): Make it pure and inline move
14018 functionality from ::get function.
14019 (get): Remove and inline into ::get and ::get_create.
14020 (get_create): Move code from ::get function.
14022 2018-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
14025 * config/i386/sol2.h (CPP_SPEC): Don't pass -P for
14026 -x assembler-with-cpp.
14028 2018-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
14030 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define
14031 _FILE_OFFSET_BITS=64 for C++.
14033 2018-06-21 Michael Meissner <meissner@linux.ibm.com>
14035 * config/rs6000/rs6000.md (extendtfif2): Add missing 128-bit
14036 conversion insn that shows up when pr85657-3.c is compiled using
14037 IEEE 128-bit long double.
14038 (neg<mode>2_internal): Use the correct mode to check whether the
14039 mode is IBM extended.
14040 * config/rs6000/rs6000.c (init_float128_ieee): Prevent complex
14041 multiply and divide external functions from being created more
14044 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
14046 * cfgrtl.c (fixup_reorder_chain): Do not emit NOPs in DECL_IGNORED_P
14048 (rtl_merge_blocks): Likewise. Do not emit a NOP if the location of
14049 the edge can be forwarded.
14050 (cfg_layout_merge_blocks): Likewise.
14052 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
14054 * except.c (finish_eh_generation): Commit edge insertions only after
14055 the EH edges have been redirected from post-landing to landing pads.
14057 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
14059 * tree-nested.c (get_frame_type): Use create_tmp_var_raw instead of
14060 create_tmp_var_for to create the FRAME decl.
14061 (finalize_nesting_tree_1): Do not unchain the FRAME decl.
14063 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
14065 * tree-inline.c (copy_edges_for_bb): Minor tweak.
14066 (maybe_move_debug_stmts_to_successors): Also reset the locus of the
14067 debug statement when resetting its value.
14068 (expand_call_inline): Copy the locus of the call onto the assignment
14069 of the return value, if any. Use local variable in more cases.
14071 2018-06-21 Martin Liska <mliska@suse.cz>
14073 * ipa-pure-const.c (propagate_nothrow): Use
14074 funct_state_summaries->get.
14075 (dump_malloc_lattice): Likewise.
14076 (propagate_malloc): Likewise.
14078 2018-06-21 Richard Biener <rguenther@suse.de>
14080 * lto-streamer-out.c (DFS::DFS_write_tree_body): Update outdated
14081 comment. Follow BLOCK_ABSTRACT_ORIGIN unconditionally.
14082 * tree-streamer-in.c (lto_input_ts_block_tree_pointers): Update
14084 * tree-streamer-out.c (write_ts_block_tree_pointers): Stream
14085 BLOCK_ABSTRACT_ORIGIN unconditionally.
14087 2018-06-21 David Malcolm <dmalcolm@redhat.com>
14089 * ipa-cp.c (ipcp_driver): Set edge_clone_summaries to NULL after
14091 * ipa-reference.c (ipa_reference_c_finalize): Delete
14092 ipa_ref_opt_sum_summaries and set it to NULL.
14094 2018-06-21 Tom de Vries <tdevries@suse.de>
14096 PR tree-optimization/85859
14097 * tree-ssa-tail-merge.c (stmt_local_def): Copy gimple_is_call
14098 test with comment from bb_no_side_effects_p.
14100 2018-06-21 Richard Biener <rguenther@suse.de>
14102 PR tree-optimization/86232
14103 * tree-ssa-loop-niter.c (number_of_iterations_popcount): Adjust
14104 max for constant niter.
14106 2018-06-21 Andre Vieira <andre.simoesdiasvieira@arm.com>
14108 * config/aarch64/aarch64-simd.md
14109 (*aarch64_crypto_aes<aes_op>v16qi_xor_combine): New.
14111 2018-06-21 Andre Vieira <andre.simoesdiasvieira@arm.com>
14113 * config/aarch64/aarch64-simd.md (aarch64_crypto_aes<aes_op>v16qi):
14114 Make opernads of the unspec commutative.
14116 2018-06-21 Richard Biener <rguenther@suse.de>
14118 * tree-data-ref.c (dr_step_indicator): Handle NULL DR_STEP.
14119 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
14120 Avoid calling vect_mark_for_runtime_alias_test with gathers or scatters.
14121 (vect_analyze_data_ref_dependence): Re-order checks to deal with
14123 (vect_record_base_alignments): Do not record base alignment
14124 for gathers or scatters.
14125 (vect_compute_data_ref_alignment): Drop return value that is always
14126 true. Bail out early for gathers or scatters.
14127 (vect_enhance_data_refs_alignment): Bail out early for gathers
14129 (vect_find_same_alignment_drs): Likewise.
14130 (vect_analyze_data_refs_alignment): Remove dead code.
14131 (vect_slp_analyze_and_verify_node_alignment): Likewise.
14132 (vect_analyze_data_refs): For possible gathers or scatters do
14133 not create an alternate DR, just check their possible validity
14134 and mark them. Adjust DECL_NONALIASED handling to not rely
14135 on DR_BASE_ADDRESS.
14136 * tree-vect-loop-manip.c (vect_update_inits_of_drs): Do not
14137 update inits of gathers or scatters.
14138 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern):
14139 Also copy gather/scatter flag to pattern vinfo.
14141 2018-06-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
14143 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Change
14144 behavior of vec_packsu (vector unsigned long long, vector unsigned
14145 long long) to match behavior of vec_packs with same signature.
14147 2018-06-20 Chung-Lin Tang <cltang@codesourcery.com>
14148 Thomas Schwinge <thomas@codesourcery.com>
14149 Cesar Philippidis <cesar@codesourcery.com>
14151 * gimplify.c (gimplify_scan_omp_clauses): Add support for
14152 OMP_CLAUSE_{IF_PRESENT,FINALIZE}.
14153 (gimplify_adjust_omp_clauses): Likewise.
14154 (gimplify_oacc_declare_1): Add support for GOMP_MAP_RELEASE, remove
14155 support for GOMP_MAP_FORCE_{ALLOC,TO,FROM,TOFROM}.
14156 (gimplify_omp_target_update): Update handling of acc update and
14158 * omp-low.c (install_var_field): Remove unused parameter
14159 base_pointers_restrict.
14160 (scan_sharing_clauses): Remove base_pointers_restrict parameter.
14161 Update call to install_var_field. Handle OMP_CLAUSE_{IF_PRESENT,
14163 (omp_target_base_pointers_restrict_p): Delete.
14164 (scan_omp_target): Update call to scan_sharing_clauses.
14165 * tree-core.h (enum omp_clause_code): Add OMP_CLAUSE_{IF_PRESENT,
14167 * tree-nested.c (convert_nonlocal_omp_clauses): Handle
14168 OMP_CLAUSE_{IF_PRESENT,FINALIZE}.
14169 (convert_local_omp_clauses): Likewise.
14170 * tree-pretty-print.c (dump_omp_clause): Likewise.
14171 * tree.c (omp_clause_num_ops): Add entries for OMP_CLAUSE_{IF_PRESENT,
14173 (omp_clause_code_name): Likewise.
14175 2018-06-20 Jakub Jelinek <jakub@redhat.com>
14178 * var-tracking.c (use_narrower_mode_test): Check if shift amount can
14181 PR tree-optimization/86231
14182 * tree-vrp.c (union_ranges): For ( [ ) ] or ( )[ ] range and
14183 anti-range don't overwrite *vr0min before using it to compute *vr0max.
14185 2018-06-20 Tom de Vries <tdevries@suse.de>
14187 PR tree-optimization/86097
14188 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Also convert *nit to
14189 iv type if signedness of iv type is not the same as that of *nit.
14191 2018-06-20 Jakub Jelinek <jakub@redhat.com>
14193 * cfgrtl.c (rtl_verify_edges): Formatting fix. If bb->preds has any
14194 EDGE_EH edges, verify they are all EDGE_EH.
14196 2018-06-20 Maya Rashish <coypu@sdf.org>
14198 * ginclude/stddef.h: Limit #include <machine/ansi.h> to NetBSD.
14200 2018-06-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14202 * config/aarch64/aarch64-tuning-flags.def (no_ldp_stp_qregs): New.
14203 * config/aarch64/aarch64.c (xgene1_tunings): Add
14204 AARCH64_EXTRA_TUNE_NO_LDP_STP_QREGS to tune_flags.
14205 (aarch64_mode_valid_for_sched_fusion_p):
14206 Allow 16-byte modes.
14207 (aarch64_classify_address): Allow 16-byte modes for load_store_pair_p.
14208 * config/aarch64/aarch64-ldpstp.md: Add peepholes for LDP STP of
14210 * config/aarch64/aarch64-simd.md (load_pair<VQ:mode><VQ2:mode>):
14212 (vec_store_pair<VQ:mode><VQ2:mode>): Likewise.
14213 * config/aarch64/iterators.md (VQ2): New mode iterator.
14215 2018-06-20 Martin Liska <mliska@suse.cz>
14217 * tree-switch-conversion.c (jump_table_cluster::can_be_handled):
14218 Change default ratio from 10 to 8.
14220 2018-06-20 Martin Liska <mliska@suse.cz>
14222 * tree-switch-conversion.c (jump_table_cluster::find_jump_tables):
14224 (bit_test_cluster::find_bit_tests): Likewise.
14225 (switch_decision_tree::analyze_switch_statement): Find clusters.
14226 * tree-switch-conversion.h (struct jump_table_cluster): Document
14229 2018-06-20 Martin Liska <mliska@suse.cz>
14231 * tree-switch-conversion.c (switch_conversion::collect):
14232 Record m_uniq property.
14233 (switch_conversion::expand): Bail out for special conditions.
14234 (group_cluster::~group_cluster): New.
14235 (group_cluster::group_cluster): Likewise.
14236 (group_cluster::dump): Likewise.
14237 (jump_table_cluster::emit): New.
14238 (switch_decision_tree::fix_phi_operands_for_edges): New.
14239 (struct case_node): Remove struct.
14240 (jump_table_cluster::can_be_handled): New.
14241 (case_values_threshold): Moved to header.
14242 (reset_out_edges_aux): Likewise.
14243 (jump_table_cluster::is_beneficial): New.
14244 (bit_test_cluster::can_be_handled): Likewise.
14245 (add_case_node): Remove.
14246 (bit_test_cluster::is_beneficial): New.
14247 (case_bit_test::cmp): New.
14248 (bit_test_cluster::emit): New.
14249 (expand_switch_as_decision_tree_p): Remove.
14250 (bit_test_cluster::hoist_edge_and_branch_if_true): New.
14251 (fix_phi_operands_for_edge): Likewise.
14252 (switch_decision_tree::analyze_switch_statement): New.
14253 (compute_cases_per_edge): Move ...
14254 (switch_decision_tree::compute_cases_per_edge): ... here.
14255 (try_switch_expansion): Likewise.
14256 (switch_decision_tree::try_switch_expansion): Likewise.
14257 (record_phi_operand_mapping): Likewise.
14258 (switch_decision_tree::record_phi_operand_mapping): Likewise.
14259 (emit_case_decision_tree): Likewise.
14260 (switch_decision_tree::emit): Likewise.
14261 (balance_case_nodes): Likewise.
14262 (switch_decision_tree::balance_case_nodes): Likewise.
14263 (dump_case_nodes): Likewise.
14264 (switch_decision_tree::dump_case_nodes): Likewise.
14265 (emit_jump): Likewise.
14266 (switch_decision_tree::emit_jump): Likewise.
14267 (emit_cmp_and_jump_insns): Likewise.
14268 (switch_decision_tree::emit_cmp_and_jump_insns): Likewise.
14269 (emit_case_nodes): Likewise.
14270 (switch_decision_tree::emit_case_nodes): Likewise.
14271 (conditional_probability): Remove.
14272 * tree-switch-conversion.h (enum cluster_type): New.
14274 (struct cluster): Likewise.
14275 (cluster::cluster): Likewise.
14276 (struct simple_cluster): Likewise.
14277 (simple_cluster::simple_cluster): Likewise.
14278 (struct group_cluster): Likewise.
14279 (struct jump_table_cluster): Likewise.
14280 (struct bit_test_cluster): Likewise.
14281 (struct min_cluster_item): Likewise.
14282 (struct case_tree_node): Likewise.
14283 (case_tree_node::case_tree_node): Likewise.
14284 (jump_table_cluster::case_values_threshold): Likewise.
14285 (struct case_bit_test): Likewise.
14286 (struct switch_decision_tree): Likewise.
14287 (struct switch_conversion): Likewise.
14288 (switch_decision_tree::reset_out_edges_aux): Likewise.
14290 2018-06-20 Martin Liska <mliska@suse.cz>
14292 * tree-switch-conversion.c (MAX_CASE_BIT_TESTS): Remove.
14293 (hoist_edge_and_branch_if_true): Likewise.
14294 (expand_switch_using_bit_tests_p): Likewise.
14295 (struct case_bit_test): Likewise.
14296 (case_bit_test_cmp): Likewise.
14297 (emit_case_bit_tests): Likewise.
14298 (switch_conversion::switch_conversion): New class.
14299 (struct switch_conv_info): Remove old struct.
14300 (collect_switch_conv_info): More to ...
14301 (switch_conversion::collect): ... this.
14302 (check_range): Likewise.
14303 (switch_conversion::check_range): Likewise.
14304 (check_all_empty_except_final): Likewise.
14305 (switch_conversion::check_all_empty_except_final): Likewise.
14306 (check_final_bb): Likewise.
14307 (switch_conversion::check_final_bb): Likewise.
14308 (create_temp_arrays): Likewise.
14309 (switch_conversion::create_temp_arrays): Likewise.
14310 (free_temp_arrays): Likewise.
14311 (gather_default_values): Likewise.
14312 (switch_conversion::gather_default_values): Likewise.
14313 (build_constructors): Likewise.
14314 (switch_conversion::build_constructors): Likewise.
14315 (constructor_contains_same_values_p): Likewise.
14316 (switch_conversion::contains_same_values_p): Likewise.
14317 (array_value_type): Likewise.
14318 (switch_conversion::array_value_type): Likewise.
14319 (build_one_array): Likewise.
14320 (switch_conversion::build_one_array): Likewise.
14321 (build_arrays): Likewise.
14322 (switch_conversion::build_arrays): Likewise.
14323 (gen_def_assigns): Likewise.
14324 (switch_conversion::gen_def_assigns): Likewise.
14325 (prune_bbs): Likewise.
14326 (switch_conversion::prune_bbs): Likewise.
14327 (fix_phi_nodes): Likewise.
14328 (switch_conversion::fix_phi_nodes): Likewise.
14329 (gen_inbound_check): Likewise.
14330 (switch_conversion::gen_inbound_check): Likewise.
14331 (process_switch): Use the newly created class.
14332 (switch_conversion::expand): New.
14333 (switch_conversion::~switch_conversion): New.
14334 * tree-switch-conversion.h: New file.
14336 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
14338 * tree-vectorizer.h (NUM_PATTERNS, vect_recog_func_ptr): Move to
14339 tree-vect-patterns.c.
14340 * tree-vect-patterns.c (vect_supportable_direct_optab_p): New function.
14341 (vect_recog_dot_prod_pattern): Use it. Remove the type_in argument.
14342 (vect_recog_sad_pattern): Likewise.
14343 (vect_recog_widen_sum_pattern): Likewise.
14344 (vect_recog_pow_pattern): Likewise. Check for a null vectype.
14345 (vect_recog_widen_shift_pattern): Remove the type_in argument.
14346 (vect_recog_rotate_pattern): Likewise.
14347 (vect_recog_mult_pattern): Likewise.
14348 (vect_recog_vector_vector_shift_pattern): Likewise.
14349 (vect_recog_divmod_pattern): Likewise.
14350 (vect_recog_mixed_size_cond_pattern): Likewise.
14351 (vect_recog_bool_pattern): Likewise.
14352 (vect_recog_mask_conversion_pattern): Likewise.
14353 (vect_try_gather_scatter_pattern): Likewise.
14354 (vect_recog_widen_mult_pattern): Likewise. Check for a null vectype.
14355 (vect_recog_over_widening_pattern): Likewise.
14356 (vect_recog_gather_scatter_pattern): Likewise.
14357 (vect_recog_func_ptr): Move from tree-vectorizer.h
14358 (vect_vect_recog_func_ptrs): Move further down the file.
14359 (vect_recog_func): Likewise. Remove the third argument.
14360 (NUM_PATTERNS): Define based on vect_vect_recog_func_ptrs.
14361 (vect_pattern_recog_1): Expect the pattern function to do any
14362 necessary target tests. Also expect it to provide a vector type.
14363 Remove the type_in handling.
14365 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
14367 * tree-vect-patterns.c (vect_pattern_detected): New function.
14368 (vect_recog_dot_prod_patternm, vect_recog_sad_pattern)
14369 (vect_recog_widen_mult_pattern, vect_recog_widen_sum_pattern)
14370 (vect_recog_over_widening_pattern, vect_recog_widen_shift_pattern
14371 (vect_recog_rotate_pattern, vect_recog_vector_vector_shift_pattern)
14372 (vect_recog_mult_pattern, vect_recog_divmod_pattern)
14373 (vect_recog_mixed_size_cond_pattern, vect_recog_bool_pattern)
14374 (vect_recog_mask_conversion_pattern)
14375 (vect_try_gather_scatter_pattern): Likewise.
14377 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
14379 * tree-vect-patterns.c (vect_get_internal_def): New function.
14380 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern)
14381 (vect_recog_vector_vector_shift_pattern, check_bool_pattern)
14382 (search_type_for_mask_1): Use it.
14384 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
14386 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Remove
14387 redundant WIDEN_SUM_EXPR handling.
14388 (vect_recog_sad_pattern): Likewise.
14390 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
14392 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Remove
14393 redundant check that the types of a PLUS_EXPR or MULT_EXPR agree.
14394 (vect_recog_sad_pattern): Likewise PLUS_EXPR, ABS_EXPR and MINUS_EXPR.
14395 (vect_recog_widen_mult_pattern): Likewise MULT_EXPR.
14396 (vect_recog_widen_sum_pattern): Likewise PLUS_EXPR.
14398 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
14400 * tree-vect-stmts.c (vectorizable_call): Make sure that we
14401 use the stmt_vec_info of the original bb statement for the
14402 new zero assignment, even if the call is part of a pattern.
14404 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
14406 * tree-vectorizer.h (_stmt_vec_info): Note above pattern_def_seq
14407 that the sequence is attached to the original statement rather
14408 than the pattern statement.
14409 * tree-vect-loop.c (vect_determine_vf_for_stmt): Take the
14410 PATTERN_DEF_SEQ from the original statement rather than
14411 the main pattern statement.
14412 * tree-vect-stmts.c (free_stmt_vec_info): Likewise.
14413 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Likewise.
14414 (vect_mark_pattern_stmts): Don't copy the PATTERN_DEF_SEQ.
14416 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
14418 * tree-vect-stmts.c (vect_analyze_stmt): Move the handling of pattern
14419 definition statements before the early exit for statements that aren't
14421 * tree-vect-loop.c (vect_transform_loop_stmt): New function,
14423 (vect_transform_loop): ...here. Process pattern definition
14424 statements without first checking whether the main pattern
14425 statement is live or relevant.
14427 2018-06-19 Eric Botcazou <ebotcazou@adacore.com>
14429 * tree-cfgcleanup.c (tree_forwarder_block_p): Do not return false at
14430 -O0 if the locus represent UNKNOWN_LOCATION but have different values.
14432 2018-06-19 Aaron Sawdey <acsawdey@linux.ibm.com>
14434 * config/rs6000/rs6000-string.c (select_block_compare_mode): Check
14435 TARGET_EFFICIENT_OVERLAPPING_UNALIGNED here instead of in caller.
14436 (do_and3, do_and3_mask, do_compb3, do_rotl3): New functions.
14437 (expand_block_compare): Change select_block_compare_mode call.
14438 (expand_strncmp_align_check): Use new functions, fix comment.
14439 (emit_final_str_compare_gpr): New function.
14440 (expand_strn_compare): Refactor and clean up code.
14441 * config/rs6000/vsx.md (vsx_mov<mode>_64bit): Remove *.
14443 2018-06-19 Tony Reix <tony.reix@atos.com>
14444 Damien Bergamini <damien.bergamini@atos.com>
14445 David Edelsohn <dje.gcc@gmail.com>
14447 * collect2.c (static_obj): New variable.
14448 (static_libs): New variable.
14449 (is_in_list): Uncomment declaration.
14450 (main): Track AIX libraries linked statically.
14451 (is_in_list): Uncomment definition.
14452 (scan_prog_file): Don't add AIX shared libraries initializer
14453 to constructor list if linking statically.
14455 2018-06-19 Max Filippov <jcmvbkbc@gmail.com>
14457 * config/xtensa/xtensa.md (UNSPEC_FRAME_BLOCKAGE): New unspec
14459 (allocate_stack, frame_blockage, *frame_blockage): New patterns.
14461 2018-06-19 Jan Hubicka <hubicka@ucw.cz>
14463 * tree.c (find_decls_types_r): Remove all non-VAR_DECLs from
14466 2018-06-19 Martin Liska <mliska@suse.cz>
14468 * config/i386/i386.c (ix86_can_inline_p): Do not use
14469 ipa_fn_summaries::get_create.
14470 * ipa-cp.c (ipcp_cloning_candidate_p): Replace get_create with
14472 (devirtualization_time_bonus): Likewise.
14473 (ipcp_propagate_stage): Likewise.
14474 * ipa-fnsummary.c (redirect_to_unreachable): Likewise.
14475 (edge_set_predicate): Likewise.
14476 (evaluate_conditions_for_known_args): Likewise.
14477 (evaluate_properties_for_edge): Likewise.
14478 (ipa_call_summary::reset): Tranform to ...
14479 (ipa_call_summary::~ipa_call_summary): ... this.
14480 (ipa_fn_summary::reset): Transform to ...
14481 (ipa_fn_summary::~ipa_fn_summary): ... this.
14482 (ipa_fn_summary_t::remove): Rename to ...
14483 (ipa_fn_summary_t::remove_callees): ... this.
14484 (ipa_fn_summary_t::duplicate): Use placement new
14485 instead of memory copy.
14486 (ipa_call_summary_t::duplicate): Likewise.
14487 (ipa_call_summary_t::remove): Remove.
14488 (dump_ipa_call_summary): Change get_create to get.
14489 (ipa_dump_fn_summary): Dump only when summary exists.
14490 (analyze_function_body): Use symbol_summary::get instead
14492 (compute_fn_summary): Likewise.
14493 (estimate_edge_devirt_benefit): Likewise.
14494 (estimate_edge_size_and_time): Likewise.
14495 (inline_update_callee_summaries): Likewise.
14496 (remap_edge_change_prob): Likewise.
14497 (remap_edge_summaries): Likewise.
14498 (ipa_merge_fn_summary_after_inlining): Likewise.
14499 (write_ipa_call_summary): Likewise.
14500 (ipa_fn_summary_write): Likewise.
14501 (ipa_free_fn_summary): Likewise.
14502 * ipa-fnsummary.h (struct GTY): Add new ctor and copy ctor.
14503 (struct ipa_call_summary): Likewise.
14504 * ipa-icf.c (sem_function::merge): Use symbol_summary::get instead
14506 * ipa-inline-analysis.c (do_estimate_edge_time): Likewise.
14507 (estimate_size_after_inlining): Likewise.
14508 (estimate_growth): Likewise.
14509 (growth_likely_positive): Likewise.
14510 * ipa-inline-transform.c (clone_inlined_nodes): Likewise.
14511 (inline_call): Likewise.
14512 * ipa-inline.c (caller_growth_limits): Likewise.
14513 (can_inline_edge_p): Likewise.
14514 (can_inline_edge_by_limits_p): Likewise.
14515 (compute_uninlined_call_time): Likewise.
14516 (compute_inlined_call_time): Likewise.
14517 (want_inline_small_function_p): Likewise.
14518 (edge_badness): Likewise.
14519 (update_caller_keys): Likewise.
14520 (update_callee_keys): Likewise.
14521 (inline_small_functions): Likewise.
14522 (inline_to_all_callers_1): Likewise.
14523 (dump_overall_stats): Likewise.
14524 (early_inline_small_functions): Likewise.
14525 (early_inliner): Likewise.
14526 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
14527 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
14528 * ipa-pure-const.c (malloc_candidate_p): Likewise.
14529 * ipa-split.c (execute_split_functions): Likewise.
14530 * symbol-summary.h: Likewise.
14531 * tree-sra.c (ipa_sra_preliminary_function_checks): Likewise.
14533 2018-06-19 Richard Biener <rguenther@suse.de>
14535 * tree-vectorizer.c (try_vectorize_loop_1): Split out of ...
14536 (vectorize_loops): ... here. Fix dbgcnt handling.
14537 (try_vectorize_loop): Wrap try_vectorize_loop_1.
14539 2018-06-19 Segher Boessenkool <segher@kernel.crashing.org>
14542 * config/rs6000/rs6000.md (rs6000_discover_homogeneous_aggregate): An
14543 ieee128 argument takes up only one (vector) register, not two (floating
14546 2018-06-19 Eric Botcazou <ebotcazou@adacore.com>
14548 * gimplify.c (gimplify_init_constructor): Really never clear for an
14549 incomplete constructor if CONSTRUCTOR_NO_CLEARING is set.
14551 2018-06-19 Richard Biener <rguenther@suse.de>
14553 PR tree-optimization/86179
14554 * tree-vect-patterns.c (vect_pattern_recog_1): Clean up
14555 after failed recognition.
14557 2018-06-18 Martin Sebor <msebor@redhat.com>
14559 PR middle-end/85602
14560 * calls.c (maybe_warn_nonstring_arg): Handle strncat.
14561 * tree-ssa-strlen.c (is_strlen_related_p): Make extern.
14562 Handle integer subtraction.
14563 (maybe_diag_stxncpy_trunc): Handle nonstring source arguments.
14564 * tree-ssa-strlen.h (is_strlen_related_p): Declare.
14566 2018-06-18 David Malcolm <dmalcolm@redhat.com>
14568 * config/frv/frv-protos.h (frv_ifcvt_modify_insn): Strengthen 3rd
14569 param from rtx to rtx_insn *.
14570 * config/frv/frv.c (frv_ifcvt_add_insn): Likewise for "insn"
14572 (frv_ifcvt_modify_insn): Likwise.
14573 (frv_ifcvt_modify_final): Likwise for local "existing_insn",
14574 adding an as_a <rtx_insn *> cast. Likewise for local "insn".
14575 * config/mips/mips.c (r10k_insert_cache_barriers): Add an
14576 as_a <rtx_insn *> cast to local "unprotected_region" once
14577 it's been established that it's not NULL or pc_rtx.
14578 * config/nds32/nds32-relax-opt.c (nds32_group_insns): Strengthen
14579 param "sethi" from rtx to rtx_insn *.
14580 (nds32_group_float_insns): Likewise for param "insn".
14581 * config/vax/vax-protos.h (vax_output_int_add): Likewise for 1st
14583 (vax_output_int_subtract): Likewise.
14584 * config/vax/vax.c (vax_output_int_add): Likewise for param
14586 (vax_output_int_subtract): Likewise.
14587 * emit-rtl.c (set_insn_deleted): Likewise, removing cast.
14588 (emit_pattern_after): Likewise for param "after".
14589 (emit_insn_after): Likewise.
14590 (emit_jump_insn_after): Likewise.
14591 (emit_call_insn_after): Likewise.
14592 (emit_debug_insn_after): Likewise.
14593 (emit_pattern_before): Likewise for param "before".
14594 (emit_insn_before): Likewise.
14595 (emit_jump_insn_before): Likewise.
14596 * final.c (get_insn_template): Likewise for param "insn", removing
14598 * output.h (get_insn_template): Likewise for 2nd param.
14599 * rtl.h (emit_insn_before): Likewise.
14600 (emit_jump_insn_before): Likewise.
14601 (emit_debug_insn_before_noloc): Likewise.
14602 (emit_insn_after): Likewise.
14603 (emit_jump_insn_after): Likewise.
14604 (emit_call_insn_after): Likewise.
14605 (emit_debug_insn_after): Likewise.
14606 (set_insn_deleted): Likewise for param.
14608 2018-06-18 Michael Meissner <meissner@linux.ibm.com>
14611 * config/rs6000/rs6000-modes.def (toplevel): Rework the 128-bit
14612 floating point modes, so that IFmode is numerically greater than
14613 TFmode, which is greater than KFmode using FRACTIONAL_FLOAT_MODE
14614 to declare the ordering. This prevents IFmode from being
14615 converted to TFmode when long double is IEEE 128-bit on an ISA 3.0
14616 machine. Include rs6000-modes.h to share the fractional values
14617 between genmodes* and the rest of the compiler.
14618 (IFmode): Likewise.
14619 (KFmode): Likewise.
14620 (TFmode): Likewise.
14621 * config/rs6000/rs6000-modes.h: New file.
14622 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Change the
14623 meaning of rs6000_long_double_size so that 126..128 selects an
14624 appropriate 128-bit floating point type.
14625 (rs6000_option_override_internal): Likewise.
14626 * config/rs6000/rs6000.h (toplevel): Include rs6000-modes.h.
14627 (TARGET_LONG_DOUBLE_128): Change the meaning of
14628 rs6000_long_double_size so that 126..128 selects an appropriate
14629 128-bit floating point type.
14630 (LONG_DOUBLE_TYPE_SIZE): Update comment.
14631 * config/rs6000/rs6000.md (trunciftf2): Correct the modes of the
14632 source and destination to match the standard usage.
14633 (truncifkf2): Likewise.
14634 (copysign<mode>3, IEEE iterator): Rework copysign of float128 on
14635 ISA 2.07 to use an explicit clobber, instead of passing in a
14637 (copysign<mode>3_soft): Likewise.
14639 2018-06-18 David Malcolm <dmalcolm@redhat.com>
14641 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
14642 Replace dump_printf_loc call with DUMP_VECT_SCOPE.
14643 (vect_slp_analyze_instance_dependence): Likewise.
14644 (vect_enhance_data_refs_alignment): Likewise.
14645 (vect_analyze_data_refs_alignment): Likewise.
14646 (vect_slp_analyze_and_verify_instance_alignment
14647 (vect_analyze_data_ref_accesses): Likewise.
14648 (vect_prune_runtime_alias_test_list): Likewise.
14649 (vect_analyze_data_refs): Likewise.
14650 * tree-vect-loop-manip.c (vect_update_inits_of_drs): Likewise.
14651 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
14652 (vect_analyze_scalar_cycles_1): Likewise.
14653 (vect_get_loop_niters): Likewise.
14654 (vect_analyze_loop_form_1): Likewise.
14655 (vect_update_vf_for_slp): Likewise.
14656 (vect_analyze_loop_operations): Likewise.
14657 (vect_analyze_loop): Likewise.
14658 (vectorizable_induction): Likewise.
14659 (vect_transform_loop): Likewise.
14660 * tree-vect-patterns.c (vect_pattern_recog): Likewise.
14661 * tree-vect-slp.c (vect_analyze_slp): Likewise.
14662 (vect_make_slp_decision): Likewise.
14663 (vect_detect_hybrid_slp): Likewise.
14664 (vect_slp_analyze_operations): Likewise.
14665 (vect_slp_bb): Likewise.
14666 * tree-vect-stmts.c (vect_mark_stmts_to_be_vectorized): Likewise.
14667 (vectorizable_bswap): Likewise.
14668 (vectorizable_call): Likewise.
14669 (vectorizable_simd_clone_call): Likewise.
14670 (vectorizable_conversion): Likewise.
14671 (vectorizable_assignment): Likewise.
14672 (vectorizable_shift): Likewise.
14673 (vectorizable_operation): Likewise.
14674 * tree-vectorizer.h (DUMP_VECT_SCOPE): New macro.
14676 2018-06-18 Martin Sebor <msebor@redhat.com>
14678 PR tree-optimization/81384
14679 * builtin-types.def (BT_FN_SIZE_CONST_STRING_SIZE): New.
14680 * builtins.c (expand_builtin_strnlen): New function.
14681 (expand_builtin): Call it.
14682 (fold_builtin_n): Avoid setting TREE_NO_WARNING.
14683 * builtins.def (BUILT_IN_STRNLEN): New.
14684 * calls.c (maybe_warn_nonstring_arg): Handle BUILT_IN_STRNLEN.
14685 Warn for bounds in excess of maximum object size.
14686 * tree-ssa-strlen.c (maybe_set_strlen_range): Return tree representing
14687 single-value ranges. Handle strnlen.
14688 (handle_builtin_strlen): Handle strnlen.
14689 (strlen_check_and_optimize_stmt): Same.
14690 * doc/extend.texi (Other Builtins): Document strnlen.
14692 2018-06-18 Maya Rashish <coypu@sdf.org>
14694 * config/alpha/openbsd.h (TARGET_DEFAULT): Define.
14695 (LINK_SPEC, STARTFILE_SPEC, ENDFILE_SPEC): Likewise.
14696 (INTMAX_TYPE, UINTMAX_TYPE, WINT_TYPE): Likewise.
14698 * config/alpha/elf.h (STARTFILE_SPEC, ENDFILE_SPEC): Move from
14700 * config/alpha/linux.h (STARTFILE_SPEC, ENDFILE_SPEC): Here.
14702 2018-06-18 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
14704 * tree.c (escaped_string::escape): Replace cast to char * by
14705 const_cast<char *> (unescaped).
14707 2018-06-18 Nick Clifton <nickc@redhat.com>
14710 * tree.c (escaped_string): New class. Converts an unescaped
14711 string into its escaped equivalent.
14712 (warn_deprecated_use): Use the new class to convert the
14713 deprecation message, if present.
14714 (test_escaped_strings): New self test.
14715 (test_c_tests): Add test_escaped_strings.
14716 * doc/extend.texi (deprecated): Add a note that the
14717 deprecation message is affected by the -fmessage-length
14718 option, and that control characters will be escaped.
14719 (#pragma GCC error): Document this pragma.
14720 (#pragma GCC warning): Likewise.
14721 * doc/invoke.texi (-fmessage-length): Document this option's
14722 effect on the #warning and #error preprocessor directives and
14723 the deprecated attribute.
14725 2018-06-18 Eric Botcazou <ebotcazou@adacore.com>
14727 * tree.c (decl_value_expr_lookup): Revert latest change.
14728 (decl_value_expr_insert): Likewise.
14730 2018-06-17 Eric Botcazou <ebotcazou@adacore.com>
14732 * gimplify.c (nonlocal_vlas): Delete.
14733 (nonlocal_vla_vars): Likewise.
14734 (gimplify_var_or_parm_decl): Do not add debug VAR_DECLs for non-local
14736 (gimplify_body): Do not create and destroy nonlocal_vlas.
14737 * tree-nested.c: Include diagnostic.h.
14738 (use_pointer_in_frame): Tweak.
14739 (lookup_field_for_decl): Add assertion and declare the transformation.
14740 (convert_nonlocal_reference_op) <PARM_DECL>: Rework and issue an
14741 internal error when the reference is in a wrong context. Do not
14742 create a debug decl by default.
14743 (note_nonlocal_block_vlas): Delete.
14744 (convert_nonlocal_reference_stmt) <GIMPLE_BIND>: Do not call it.
14745 (convert_local_reference_op) <PARM_DECL>: Skip the frame decl. Do not
14746 create a debug decl by default.
14747 (convert_gimple_call) <GIMPLE_CALL>: Issue an internal error when the
14748 call is in a wrong context.
14749 (fixup_vla_decls): New function.
14750 (finalize_nesting_tree_1): Adjust comment. Call fixup_vla_decls if no
14751 debug variables were created.
14752 * tree.c (decl_value_expr_lookup): Add checking assertion.
14753 (decl_value_expr_insert): Likewise.
14755 2018-06-16 Kugan Vivekanandarajah <kuganv@linaro.org>
14757 PR middle-end/82479
14758 * ipa-fnsummary.c (will_be_nonconstant_expr_predicate): Handle CALL_EXPR.
14759 * tree-scalar-evolution.c (interpret_expr): Likewise.
14760 (expression_expensive_p): Likewise.
14761 * tree-ssa-loop-ivopts.c (contains_abnormal_ssa_name_p): Likewise.
14762 * tree-ssa-loop-niter.c (number_of_iterations_popcount): New.
14763 (number_of_iterations_exit_assumptions): Use number_of_iterations_popcount.
14764 (ssa_defined_by_minus_one_stmt_p): New.
14766 2018-06-16 Kugan Vivekanandarajah <kuganv@linaro.org>
14768 PR middle-end/64946
14769 * cfgexpand.c (expand_debug_expr): Hande ABSU_EXPR.
14770 * config/i386/i386.c (ix86_add_stmt_cost): Likewise.
14771 * dojump.c (do_jump): Likewise.
14772 * expr.c (expand_expr_real_2): Check operand type's sign.
14773 * fold-const.c (const_unop): Handle ABSU_EXPR.
14774 (fold_abs_const): Likewise.
14775 * gimple-pretty-print.c (dump_unary_rhs): Likewise.
14776 * gimple-ssa-backprop.c (backprop::process_assign_use): Likesie.
14777 (strip_sign_op_1): Likesise.
14778 * match.pd: Add new pattern to generate ABSU_EXPR.
14779 * optabs-tree.c (optab_for_tree_code): Handle ABSU_EXPR.
14780 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
14781 * tree-eh.c (operation_could_trap_helper_p): Likewise.
14782 * tree-inline.c (estimate_operator_cost): Likewise.
14783 * tree-pretty-print.c (dump_generic_node): Likewise.
14784 * tree-vect-patterns.c (vect_recog_sad_pattern): Likewise.
14785 * tree.def (ABSU_EXPR): New.
14787 2018-06-16 Jakub Jelinek <jakub@redhat.com>
14789 PR middle-end/86095
14790 * common.opt (Wunsafe-loop-optimizations): Add Ignore, remove Var,
14791 documented as preserved for backward compatibility only.
14792 * doc/invoke.texi: Remove -Wunsafe-loop-optimizations documentation.
14794 PR rtl-optimization/86108
14795 * bb-reorder.c (create_forwarder_block): Renamed to ...
14796 (create_eh_forwarder_block): ... this. Split OLD_BB after labels and
14797 jump from new landing pad to the second part.
14798 (sjlj_fix_up_crossing_landing_pad, dw2_fix_up_crossing_landing_pad):
14801 2018-06-15 Jakub Jelinek <jakub@redhat.com>
14803 PR middle-end/85878
14804 * expr.c (expand_assignment): Remove now redundant COMPLEX_MODE_P
14805 check from first store_expr, use to_mode instead of GET_MODE (to_rtx).
14806 Only call store_expr for halves if the mode is the same.
14808 PR middle-end/86123
14809 * match.pd ((X / Y) == 0 -> X < Y): Don't transform complex divisions.
14810 Fix up comment formatting.
14812 2018-06-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
14814 * typed-splay-tree.h (typed_splay_tree::remove): New function.
14815 (typed_splay_tree::closure,
14816 typed_splay_tree::inner_foreach_fn, typed_splay_tree::m_inner): Deleted.
14817 (typed_splay_tree::typed_splay_tree,
14818 typed_splay_tree::operator =): Declared private.
14819 (typed_splay_tree::splay_tree_key, typed_splay_tree::splay_tree_value,
14820 typed_splay_tree::splay_tree_node_s, typed_splay_tree::KDEL,
14821 typed_splay_tree::VDEL, typed_splay_tree::splay_tree_delete_helper,
14822 typed_splay_tree::rotate_left, typed_splay_tree::rotate_right,
14823 typed_splay_tree::splay_tree_splay,
14824 typed_splay_tree::splay_tree_foreach_helper,
14825 typed_splay_tree::splay_tree_insert,
14826 typed_splay_tree::splay_tree_remove,
14827 typed_splay_tree::splay_tree_lookup,
14828 typed_splay_tree::splay_tree_predecessor,
14829 typed_splay_tree::splay_tree_successor,
14830 typed_splay_tree::splay_tree_min,
14831 typed_splay_tree::splay_tree_max): Took over from splay-tree.c/.h.
14832 (typed_splay_tree::root, typed_splay_tree::comp,
14833 typed_splay_tree::delete_key,
14834 typed_splay_tree::delete_value): New data members.
14835 * typed-splay-tree.c (selftest::test_str_to_int): Add a test for
14836 typed_splay_tree::remove.
14838 2018-06-15 Matthew Fortune <matthew.fortune@mips.com>
14840 * config/mips/mips.h (ASM_SPEC): Pass through -mcrc, -mno-crc,
14841 -mginv and -mno-ginv to the assembler.
14842 * config/mips/mips.opt (-mcrc): New option.
14843 (-mginv): Likewise.
14844 * doc/invoke.text (-mcrc): Document.
14845 (-mginv): Likewise.
14847 2018-06-15 Nick Clifton <nickc@redhat.com>
14850 * tree.c (escaped_string): New class. Converts an unescaped
14851 string into its escaped equivalent.
14852 (warn_deprecated_use): Use the new class to convert the
14853 deprecation message, if present.
14854 (test_escaped_strings): New self test.
14855 (test_c_tests): Add test_escaped_strings.
14856 * doc/extend.texi (deprecated): Add a note that the
14857 deprecation message is affected by the -fmessage-length
14858 option, and that control characters will be escaped.
14859 (#pragma GCC error): Document this pragma.
14860 (#pragma GCC warning): Likewise.
14861 * doc/invoke.texi (-fmessage-length): Document this option's
14862 effect on the #warning and #error preprocessor directives and
14863 the deprecated attribute.
14865 2018-06-15 Richard Biener <rguenther@suse.de>
14867 * tree-vect-slp.c (vect_slp_bb): Dump MSG_OPTIMIZED_LOCATIONS
14868 here, also noting vector size used.
14869 * tree-vectorizer.c (vectorize_loops): Adjust. Note vector
14870 size used in MSG_OPTIMIZED_LOCATIONS dump.
14871 (pass_slp_vectorize::execute): Adjust.
14873 2018-06-15 Claudiu Zissulescu <claziss@synopsys.com>
14876 * config/arc/arc.c (arc_return_address_register): Fix
14879 2018-06-15 Richard Biener <rguenther@suse.de>
14881 PR middle-end/86159
14882 * tree-cfg.c (gimplify_build3): Do not strip sign conversions,
14883 leave useless conversion stripping to force_gimple_operand_gsi.
14884 (gimplify_build2): Likewise.
14885 (gimplify_build1): Likewise.
14887 2018-06-15 Richard Biener <rguenther@suse.de>
14889 PR middle-end/86076
14890 * tree-cfg.c (move_stmt_op): unshare invariant addresses
14891 before adjusting their block.
14893 2018-06-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
14895 * config.gcc (riscv*-*-elf* | riscv*-*-rtems*): Use custom
14896 multilibs for *-*-rtems*.
14897 * config/riscv/t-rtems: New file.
14899 2018-06-14 Jakub Jelinek <jakub@redhat.com>
14901 PR middle-end/86122
14902 * match.pd ((A +- CST1) +- CST2): Punt if last resort
14903 unsigned_type_for returns NULL.
14906 * lower-subreg.c (find_decomposable_subregs): Don't decompose float
14907 subregs of multi-word pseudos unless the float mode has word size.
14909 2018-06-14 Richard Biener <rguenther@suse.de>
14911 PR middle-end/86139
14912 * tree-vect-generic.c (build_word_mode_vector_type): Remove
14913 duplicate and harmful type_hash_canon.
14914 * tree.c (type_hash_canon): Assert we didn't find ourselves.
14916 2018-06-14 Richard Biener <rguenther@suse.de>
14919 * tree-ssa-struct-alias.c (create_variable_info_for): Handle
14922 2018-06-14 Sebastian Huber <sebastian.huber@embedded-brains.de>
14924 * config/rtems.h (STDINT_LONG32): Define.
14926 2018-06-13 Matthew Fortune <matthew.fortune@mips.com>
14927 Prachi Godbole <prachi.godbole@imgtec.com>
14929 * config/mips/mips-cpus.def: Define P6600.
14930 * config/mips/mips-tables.opt: Regenerate.
14931 * config/mips/mips.c (mips_ucbranch_type): New enum.
14932 (mips_rtx_cost_data): Add support for P6600.
14933 (mips_issue_rate): Likewise.
14934 (mips_multipass_dfa_lookahead): Likewise.
14935 (mips_avoid_hazard): Likewise.
14936 (mips_reorg_process_insns): Likewise.
14937 (mips_classify_branch_p6600): New function.
14938 * config/mips/mips.h (TUNE_P6600): New define.
14939 (MIPS_ISA_LEVEL_SPEC): Infer mips64r6 from p6600.
14940 (ENABLE_LD_ST_PAIRS): Enable load/store bonding for p6600.
14941 * config/mips/mips.md: Include p6600.md.
14942 (processor): Add p6600.
14943 * config/mips/p6600.md: New file.
14944 * doc/invoke.texi: Add p6600 to supported architectures.
14946 2018-06-13 Martin Sebor <msebor@redhat.com>
14948 PR tree-optimization/86114
14949 * gimple-fold.c (gimple_fold_builtin_strlen): Only handle LHS
14951 * tree-ssa-strlen.c (maybe_set_strlen_range): Same.
14953 2018-06-13 Richard Biener <rguenther@suse.de>
14955 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern):
14956 Properly set vector type of the intermediate stmt.
14957 * tree-vect-stmts.c (vectorizable_operation): The destination
14958 var always has vectype_out type.
14960 2018-06-13 Jeff Law <law@redhat.com>
14962 * config/rl78/rl78.c (move_elim_pass): Use TDF_NONE rather than
14963 integer 0 for argument to print_rtl_with_bb.
14964 (rl78_reorg): Likewise.
14966 2018-06-13 David Malcolm <dmalcolm@redhat.com>
14968 * config/arc/arc.c (hwloop_optimize): Strengthen local "end_label"
14969 from rtx to rtx_insn *.
14970 * config/bfin/bfin.c (hwloop_optimize): Likewise for local
14972 (add_sched_insns_for_speculation): Likewise for local "target",
14973 converting usage of JUMP_LABEL to JUMP_LABEL_AS_INSN.
14974 * config/c6x/c6x.c (reorg_split_calls): Strengthen param "call_labels"
14975 from rtx_insn ** to rtx_code_label **.
14976 (reorg_emit_nops): Likewise.
14977 (c6x_reorg): Likewise for local "call_labels".
14978 * config/sh/sh-protos.h (get_dest_uid): Strengthen 1st param from
14980 * config/sh/sh.c (dump_table): Strengthen local "lab" from rtx to
14981 rtx_code_label *, adding safe_as_a <rtx_code_label *> casts to
14982 the loops over LABEL_REFS.
14983 (fixup_addr_diff_vecs): Add as_a <rtx_insn *> to usage of
14985 (barrier_align): Convert usage of JUMP_LABEL to JUMP_LABEL_AS_INSN.
14986 (get_dest_uid): Strengthen param "label" from rtx to rtx_insn *.
14987 (split_branches): Strengthen local "olabel" from rtx to
14988 rtx_insn *, adding a safe_as_a cast.
14989 * emit-rtl.c (next_real_insn): Strengthen param from "rtx"
14991 (add_insn_after): Likewise for first two params.
14992 (add_insn_before): Likewise.
14993 (remove_insn): Likewise for param.
14994 (emit_pattern_before_noloc): Likewise for second and third params.
14995 (emit_jump_insn_before_noloc): Convert NULL_RTX to NULL.
14996 (emit_call_insn_before_noloc): Likewise.
14997 (emit_debug_insn_before_noloc): Strengthen "before" param from "rtx"
14999 (emit_barrier_before): Likewise.
15000 (emit_label_before): Strengthen "label" param from "rtx" to
15001 "rtx_code_label *". Strengthen "before" param from "rtx" to
15003 (emit_insn_after_1): Strengthen "after" param from "rtx" to
15005 (emit_pattern_after_noloc): Likewise.
15006 (emit_insn_after_noloc): Likewise.
15007 (emit_jump_insn_after_noloc): Likewise.
15008 (emit_call_insn_after_noloc): Likewise.
15009 (emit_debug_insn_after_noloc): Likewise.
15010 (emit_barrier_after): Likewise.
15011 (emit_label_after): Likewise for both params.
15012 (emit_pattern_after_setloc): Likewise for "after" param. Convert
15013 "loc" param from "int" to "location_t".
15014 (emit_insn_after_setloc): Likewise.
15015 (emit_jump_insn_after_setloc): Likewise.
15016 (emit_call_insn_after_setloc): Likewise.
15017 (emit_debug_insn_after_setloc): Likewise.
15018 (emit_pattern_before_setloc): Likewise for "before" param. Convert
15019 "loc" param from "int" to "location_t".
15020 (emit_pattern_before): Convert NULL_RTX to NULL.
15021 (emit_insn_before_setloc): Convert "loc" param from "int" to
15023 (emit_jump_insn_before_setloc): Likewise.
15024 (emit_call_insn_before_setloc): Likewise.
15025 (emit_debug_insn_before_setloc): Strengthen "before" param from rtx to
15026 rtx_insn *. Convert "loc" param from "int" to "location_t".
15027 * rtl.h (emit_insn_before_setloc, emit_jump_insn_before_setloc,
15028 emit_call_insn_before_setloc, emit_debug_insn_before_setloc):
15029 Convert 3rd param from "int" to "location_t".
15030 (emit_barrier_before, emit_barrier_after, next_real_insn):
15031 Strengthen param from rtx to rtx_insn *.
15032 (emit_label_before): Strengthen 1st param from "rtx" to
15033 "rtx_code_label *". Strengthen 2nd param from "rtx" to
15035 (emit_insn_after_noloc, emit_jump_insn_after_noloc,
15036 emit_call_insn_after_noloc, emit_debug_insn_after_noloc):
15037 Strengthen 2nd param from "rtx" to "rtx_insn *".
15038 (emit_insn_after_setloc, emit_jump_insn_after_setloc)
15039 emit_call_insn_after_setloc, emit_debug_insn_after_setloc):
15040 Likewise. Convert 3rd param from "int" to "location_t".
15041 (emit_label_after): Strengthen 1st param from "rtx" to
15042 "rtx_code_label *".
15043 (next_real_insn, remove_insn): Strengthen param from "rtx" to
15045 (add_insn_before, add_insn_after): Strengthen 1st and 2nd params
15046 from "rtx" to "rtx_insn *".
15048 2018-06-13 Jan Hubicka <hubicka@gcc.gnu.org>
15050 * cgraph.c (cgraph_node::get_untransformed_body): Dump function
15051 bodies streamed in with -Q.
15052 * dumpfile.c (dump_files): Add lto-stream-out dump file.
15053 * dumpfile.h (tree_dump_index): Add lto_stream_out.
15054 * gimple-streamer-out.c: Include gimple-pretty-print.h
15055 (output_bb): Dump stmts streamed.
15056 * lto-section-out.c: Include print-tree.h
15057 (lto_begin_section): Dump sections created.
15058 (lto_output_decl_index): Dump decl encoded.
15059 * lto-streamer-out.c: Include print-tree.h
15060 (create_output_block): Dump output block created.
15061 (DFS::DFS_write_tree_body): Dump DFS SCCs streamed.
15062 (output_function): Dump function output.
15063 (output_constructor): Dump constructor streamed.
15064 (write_global_stream): Output indexes encoded.
15065 (produce_asm_for_decls): Dump streams encoded.
15066 * lto-streamer.c (streamer_dump_file): New global var.
15067 * lto-streamer.h (streamer_dump_file): Declare.
15068 * passes.c (ipa_write_summaries): Initialize streamer dump.
15069 * varpool.c (varpool_node::get_constructor): Dump constructors streamed
15072 2018-06-13 Eric Botcazou <ebotcazou@adacore.com>
15075 * config/i386/winnt.c (i386_pe_seh_cold_init): Do not emit negative
15076 offsets for register save directives. Emit a second batch of save
15077 directives, if need be, when the function accesses prior frames.
15079 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
15081 * config/arc/fpu.md (fmasf4): Force operand to register.
15082 (fnmasf4): Likewise.
15084 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
15086 * config/arc/arc-protos.h (arc_pad_return): Remove.
15087 * config/arc/arc.c (machine_function): Remove force_short_suffix
15089 (arc_print_operand): Adjust printing of '&'.
15090 (arc_verify_short): Remove conditional printing of short suffix.
15091 (arc_final_prescan_insn): Remove reference to size_reason.
15092 (pad_return): New function.
15093 (arc_reorg): Call pad_return.
15094 (arc_pad_return): Remove.
15095 (arc_init_machine_status): Remove reference to force_short_suffix.
15096 * config/arc/arc.md (vunspec): Add VUNSPEC_ARC_BLOCKAGE.
15097 (attr length): When attribute iscompact is true force to 2
15098 regardless; in the case of maybe check if we want to force the
15099 instruction to have 4 bytes length.
15100 (nopv): Change it to generate 4 byte long nop as well.
15101 (blockage): New pattern.
15102 (simple_return): Remove call to arc_pad_return.
15103 (p_return_i): Likewise.
15105 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
15107 * config/arc/elf.h (LINK_GCC_C_SEQUENCE_SPEC): Define.
15109 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
15111 * config/arc/builtins.def (SYNC): SYNC instruction is valid on all
15114 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
15116 * config/arc/arc.c (atomic_exchangesi): EX instruction is default
15117 for ARC700 and ARCv2.
15119 2018-06-13 Chenghua Xu <paul.hua.gm@gmail.com>
15122 * config/mips/loongson.md (vec_setv4hi): Gen_lowpart for
15123 operands[2] instead of operands[1].
15126 2018-06-12 Richard Sandiford <richard.sandiford@linaro.org>
15128 * lra-constraints.c (simplify_operand_subreg): In the paradoxical
15129 case, check whether the outer register overlaps an unallocatable
15130 register, not just whether it fits the required class.
15132 2018-06-12 Richard Sandiford <richard.sandiford@linaro.org>
15134 * poly-int.h (can_div_trunc_p): Add new overload in which all values
15136 * alias.c (get_addr): Extend CONST_INT handling to poly_int_rtx_p.
15137 (memrefs_conflict_p): Likewise.
15138 (init_alias_analysis): Likewise.
15139 * cfgexpand.c (expand_debug_expr): Likewise.
15140 * combine.c (combine_simplify_rtx, force_int_to_mode): Likewise.
15141 * cse.c (fold_rtx): Likewise.
15142 * explow.c (adjust_stack, anti_adjust_stack): Likewise.
15143 * expr.c (emit_block_move_hints): Likewise.
15144 (clear_storage_hints, push_block, emit_push_insn): Likewise.
15145 (store_expr_with_bounds, reduce_to_bit_field_precision): Likewise.
15146 (emit_group_load_1): Use rtx_to_poly_int64 for group offsets.
15147 (emit_group_store): Likewise.
15148 (find_args_size_adjust): Use strip_offset. Use rtx_to_poly_int64
15149 to read the PRE/POST_MODIFY increment.
15150 * calls.c (store_one_arg): Use strip_offset.
15151 * rtlanal.c (rtx_addr_can_trap_p_1): Extend CONST_INT handling to
15153 (set_noop_p): Use rtx_to_poly_int64 for the elements selected
15155 * simplify-rtx.c (avoid_constant_pool_reference): Use strip_offset.
15156 (simplify_binary_operation_1): Extend CONST_INT handling to
15158 * var-tracking.c (compute_cfa_pointer): Take a poly_int64 rather
15159 than a HOST_WIDE_INT.
15160 (hard_frame_pointer_adjustment): Change from HOST_WIDE_INT to
15162 (adjust_mems, add_stores): Update accodingly.
15163 (vt_canonicalize_addr): Track polynomial offsets.
15164 (emit_note_insn_var_location): Likewise.
15165 (vt_add_function_parameter): Likewise.
15166 (vt_initialize): Likewise.
15168 2018-06-12 Jeff Law <law@redhat.com>
15170 * config.gcc (alpha*-*-freebsd*): Remove.
15171 * config/alpha/freebsd.h: Remove.
15173 2018-06-12 David Malcolm <dmalcolm@redhat.com>
15176 * spellcheck-tree.c (levenshtein_distance): Rename to...
15177 (get_edit_distance): ...this, and update for underlying renaming.
15178 * spellcheck-tree.h (levenshtein_distance): Rename to...
15179 (get_edit_distance): ...this.
15180 * spellcheck.c (levenshtein_distance): Rename to...
15181 (get_edit_distance): ...this. Convert from Levenshtein distance
15182 to Damerau-Levenshtein distance by supporting transpositions of
15183 adjacent characters. Rename "v1" to "v_next" and "v0" to
15185 (selftest::levenshtein_distance_unit_test_oneway): Rename to...
15186 (selftest::test_edit_distance_unit_test_oneway): ...this, and
15187 update for underlying renaming.
15188 (selftest::levenshtein_distance_unit_test): Rename to...
15189 (selftest::test_get_edit_distance_unit): ...this, and update for
15190 underlying renaming.
15191 (selftest::test_find_closest_string): Add example from PR 69968
15192 where transposition helps
15193 (selftest::test_metric_conditions): Update for renaming.
15194 (selftest::test_metric_conditions): Likewise.
15195 (selftest::spellcheck_c_tests): Likewise.
15196 * spellcheck.h (levenshtein_distance): Rename both overloads to...
15197 (get_edit_distance): ...this.
15198 (best_match::consider): Update for renaming.
15200 2018-06-12 Martin Sebor <msebor@redhat.com>
15202 PR tree-optimization/85259
15203 * builtins.c (compute_objsize): Handle constant offsets.
15204 * gimple-ssa-warn-restrict.c (maybe_diag_offset_bounds): Return
15205 true iff a warning has been issued.
15206 * gimple.h (gimple_nonartificial_location): New function.
15207 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Call
15208 gimple_nonartificial_location and handle -Wno-system-headers.
15209 (handle_builtin_stxncpy): Same.
15211 2018-06-12 Martin Sebor <msebor@redhat.com>
15214 * fold-const.c (operand_equal_p): Handle SAVE_EXPR.
15216 2018-06-12 Will Schmidt <will_schmidt@vnet.ibm.com>
15218 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15219 BUILTIN_VEC_XST entries for pointer to double and long long.
15221 2018-06-12 H.J. Lu <hongjiu.lu@intel.com>
15224 * config/i386/gnu-user.h (TARGET_THREAD_SPLIT_STACK_OFFSET):
15226 * config/i386/gnu-user64.h (TARGET_THREAD_SPLIT_STACK_OFFSET):
15229 2018-06-12 Martin Liska <mliska@suse.cz>
15231 * doc/options.texi: Document IntegerRange.
15233 2018-06-12 Martin Liska <mliska@suse.cz>
15235 * config/i386/i386.opt: Make MPX-related options as Deprecated.
15236 * opt-functions.awk: Handle Deprecated flag.
15237 * opts-common.c (decode_cmdline_option): Handle cl_deprecated
15239 (read_cmdline_option): Report warning for a deprecated option.
15240 * opts.h (struct cl_option): Add new field cl_deprecated.
15241 (CL_ERR_DEPRECATED): New.
15243 2018-06-12 Martin Liska <mliska@suse.cz>
15245 * doc/options.texi: Document Deprecated option flag.
15247 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
15249 * config/arc/arc-arch.h (arc_extras): New enum.
15250 (arc_cpu_t):Add field extra.
15251 (arc_cpu_types): Consider the extras.
15252 * config/arc/arc-cpus.def: Add extras info.
15253 * config/arc/arc-opts.h (processor_type): Consider extra field.
15254 * config/arc/arc.c (arc_override_options): Handle extra field.
15256 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
15258 * config/arc/arc-arch.h: Update ARC_OPTX macro.
15259 * config/arc/arc-options.def (ARC_OPTX): Introduce a new doc
15261 * config/arc/arc.c (arc_init): Update pic warning.
15262 (irq_range): Update irq range parsing warnings.
15263 (arc_override_options): Update various warning messages.
15264 (arc_handle_aux_attribute): Likewise.
15266 2018-06-12 Robert Suchanek <robert.suchanek@mips.com>
15268 * config/mips/i6400.md (i6400_fpu_fadd): Remove frint.
15270 2018-06-12 Jozef Lawrynowicz <jozef.l@mittosystems.com>
15272 * doc/sourcebuild.texi: Document usage of line number 0 in verify
15273 compiler messages directives.
15275 2018-06-12 Matthew Fortune <mfortune@gmail.com>
15277 * config/mips/mips-cpus.def: New MIPS_CPU for i6500.
15278 * config/mips/mips-tables.opt: Regenerate.
15279 * config/mips/mips.h (MIPS_ISA_LEVEL_SPEC): Mark i6500 as
15281 * doc/invoke.texi: Document -march=i6500.
15283 2018-06-12 Prachi Godbole <prachi.godbole@imgtec.com>
15285 * config/mips/i6400.md (i6400_gpmuldiv): Remove cpu_unit.
15286 (i6400_gpmul): Add cpu_unit.
15287 (i6400_gpdiv): Likewise.
15288 (i6400_msa_add_d): Update reservations.
15289 (i6400_msa_int_add) Likewise.
15290 (i6400_msa_short_logic3) Likewise.
15291 (i6400_msa_short_logic2) Likewise.
15292 (i6400_msa_short_logic) Likewise.
15293 (i6400_msa_move) Likewise.
15294 (i6400_msa_cmp) Likewise.
15295 (i6400_msa_short_float2) Likewise.
15296 (i6400_msa_div_d) Likewise.
15297 (i6400_msa_long_logic1) Likewise.
15298 (i6400_msa_long_logic2) Likewise.
15299 (i6400_msa_mult) Likewise.
15300 (i6400_msa_long_float2) Likewise.
15301 (i6400_msa_long_float4) Likewise.
15302 (i6400_msa_long_float5) Likewise.
15303 (i6400_msa_long_float8) Likewise.
15304 (i6400_fpu_fadd): Include frint type.
15305 (i6400_fpu_store): New define_insn_reservation.
15306 (i6400_fpu_load): Likewise.
15307 (i6400_fpu_move): Likewise.
15308 (i6400_fpu_fcmp): Likewise.
15309 (i6400_fpu_fmadd): Likewise.
15310 (i6400_int_mult): Include imul3nc type and update reservation.
15311 (i6400_int_div): Include idiv3 type and update reservation.
15312 (i6400_int_load): Update to check type not move_type.
15313 (i6400_int_store): Likewise.
15314 (i6400_int_prefetch): Set zero latency.
15316 2018-06-12 Eric Botcazou <ebotcazou@adacore.com>
15318 * gcc.c: Document new %@{...} sequence.
15319 (LINK_COMMAND_SPEC): Use it for the -L switches.
15320 (cpp_unique_options): Use it for the -I switches.
15321 (at_file_argbuf): New global variable.
15322 (in_at_file): Likewise.
15323 (alloc_args): Create at_file_argbuf.
15324 (clear_args): Truncate at_file_argbuf.
15325 (store_arg): If in_at_file, push the argument onto at_file_argbuf.
15326 (open_at_file): New function.
15327 (close_at_file): Likewise.
15328 (create_at_file): Delete.
15329 (do_spec_1) <'i'>: Use open_at_file/close_at_file.
15332 (validate_switches_from_spec): Deal with %@{...} sequence.
15333 (validate_switches): Likewise.
15334 (driver::finalize): Call clear_args.
15336 2018-06-11 Rasmus Villemoes <rasmus.villemoes@prevas.dk>
15338 * config/vx-common.h (USE_TM_CLONE_REGISTRY): #define to 0.
15340 2018-06-11 Martin Sebor <msebor@redhat.com>
15342 * doc/invoke.texi (-Wall): List -Wc++17-compat.
15343 (Wno-class-memaccess): Add @opindex.
15344 (Wno-templates, Wno-multiple-inheritance): Same.
15345 (Wno-virtual-inheritance, Wno-namespaces): Same.
15346 (Wno-pedantic, Wno-chkp, Wcoverage-mismatch): Same.
15347 (Wno-format-overflow, Wno-format-truncation): Same.
15348 (Wno-unused-local-typedefs, Walloc-size-larger-than): Same.
15349 (Wno-alloc-size-larger-than, Wframe-larger-than): Same
15350 (Wno-stack-usage, Wno-c++-compat, Wno-c++11-compat): Same.
15351 (Wno-c++14-compat, Wno-c++17-compat, Wno-openmp-simd): Same.
15352 (Wno-unsuffixed-float-constants, Wno-addr-space-convert): Same.
15353 (Wno-misspelled-isr): Same.
15355 2018-06-11 Martin Sebor <msebor@redhat.com>
15357 * PR tree-optimization/86083
15358 * tree-ssa-strlen.c (handle_char_store): Use tree_expr_nonzero_p.
15360 2018-06-11 Zhouyi Zhou <zhouzhouyi@gmail.com>
15362 * tree-eh.c (lower_eh_constructs_2): Add a comma to comment.
15364 2018-06-11 Segher Boessenkool <segher@kernel.crashing.org>
15367 * config/rs6000/rs6000.md (*movdi_internal32): Put constraint modifiers
15368 on the correct operand.
15369 (*movdi_internal64): Ditto.
15371 2018-06-11 Martin Liska <mliska@suse.cz>
15373 PR tree-optimization/86089
15374 * tree-ssa-strlen.c (get_string_length): Move back removed hunk.
15376 2018-06-11 Julia Koval <julia.koval@intel.com>
15378 * config/i386/vaesintrin.h (_mm_aesdec_epi128, _mm_aesdeclast_epi128)
15379 _mm_aesenc_epi128, _mm_aesenclast_epi128: Remove.
15380 * config/i386/vpclmulqdqintrin.h (_mm_clmulepi64_epi128): Remove.
15382 2018-06-11 Olivier Hainque <hainque@adacore.com>
15384 * dwarf2out.c (gen_compile_unit_die): Fallout to DW_LANG_Ada83
15385 for Ada with strict dwarf2.
15387 2018-06-08 Peter Bergner <bergner@vnet.ibm.com>
15390 * config/rs6000/rs6000.c (mem_operand_gpr): Enable PRE_INC and PRE_DEC
15393 2018-06-08 Jan Hubicka <hubicka@ucw.cz>
15395 * dumpfile.c (FIRST_ME_AUTO_NUMBERED_DUMP): Bump to 4.
15397 2018-06-08 David Edelsohn <dje.gcc@gmail.com>
15399 * config/rs6000/rs6000.c (rs6000_passes_ieee128): Protect with #if
15402 2018-06-08 Martin Liska <mliska@suse.cz>
15404 * tree-cfg.h (debug_function): Fix argument type to match
15407 2018-06-08 Martin Liska <mliska@suse.cz>
15409 * config/powerpcspe/powerpcspe.c (rs6000_xcoff_visibility):
15410 Remove usage of MPX-related (and removed) fields.
15411 * config/rs6000/rs6000.c (rs6000_xcoff_visibility): Likewise.
15413 2018-06-08 David Malcolm <dmalcolm@redhat.com>
15415 * cfg.c (debug): Use TDF_NONE rather than 0.
15416 * cfghooks.c (debug): Likewise.
15417 * dumpfile.c (DUMP_FILE_INFO): Likewise; also for OPTGROUP.
15418 (struct dump_option_value_info): Convert to...
15419 (struct kv_pair): ...this template type.
15420 (dump_options): Convert to kv_pair<dump_flags_t>; use TDF_NONE
15422 (optinfo_verbosity_options): Likewise.
15423 (optgroup_options): Convert to kv_pair<optgroup_flags_t>; use
15425 (gcc::dump_manager::dump_register): Use optgroup_flags_t rather
15426 than int for "optgroup_flags" param.
15427 (dump_generic_expr_loc): Use dump_flags_t rather than int for
15429 (dump_dec): Likewise.
15430 (dump_finish): Use TDF_NONE rather than 0.
15431 (gcc::dump_manager::opt_info_enable_passes): Use optgroup_flags_t
15432 rather than int for "optgroup_flags" param. Use TDF_NONE rather
15433 than 0. Update for change to option_ptr.
15434 (opt_info_switch_p_1): Convert "optgroup_flags" param from int *
15435 to optgroup_flags_t *. Use TDF_NONE and OPTGROUP_NONE rather than
15436 0. Update for changes to optinfo_verbosity_options and
15438 (opt_info_switch_p): Convert optgroup_flags from int to
15440 (dump_basic_block): Use dump_flags_t rather than int
15441 for "dump_kind" param.
15442 * dumpfile.h (TDF_ADDRESS, TDF_SLIM, TDF_RAW, TDF_DETAILS,
15443 TDF_STATS, TDF_BLOCKS, TDF_VOPS, TDF_LINENO, TDF_UID)
15444 TDF_STMTADDR, TDF_GRAPH, TDF_MEMSYMS, TDF_RHS_ONLY, TDF_ASMNAME,
15445 TDF_EH, TDF_NOUID, TDF_ALIAS, TDF_ENUMERATE_LOCALS, TDF_CSELIB,
15446 TDF_SCEV, TDF_GIMPLE, TDF_FOLDING, MSG_OPTIMIZED_LOCATIONS,
15447 MSG_MISSED_OPTIMIZATION, MSG_NOTE, MSG_ALL, TDF_COMPARE_DEBUG,
15448 TDF_NONE): Convert from macros to...
15449 (enum dump_flag): ...this new enum.
15450 (dump_flags_t): Update to use enum.
15451 (operator|, operator&, operator~, operator|=, operator&=):
15452 Implement for dump_flags_t.
15453 (OPTGROUP_NONE, OPTGROUP_IPA, OPTGROUP_LOOP, OPTGROUP_INLINE,
15454 OPTGROUP_OMP, OPTGROUP_VEC, OPTGROUP_OTHER, OPTGROUP_ALL):
15455 Convert from macros to...
15456 (enum optgroup_flag): ...this new enum.
15457 (optgroup_flags_t): New typedef.
15458 (operator|, operator|=): Implement for optgroup_flags_t.
15459 (struct dump_file_info): Convert field "alt_flags" to
15460 dump_flags_t. Convert field "optgroup_flags" to
15462 (dump_basic_block): Use dump_flags_t rather than int for param.
15463 (dump_generic_expr_loc): Likewise.
15464 (dump_dec): Likewise.
15465 (dump_register): Convert param "optgroup_flags" to
15467 (opt_info_enable_passes): Likewise.
15468 * early-remat.c (early_remat::dump_edge_list): Use TDF_NONE rather
15470 * gimple-pretty-print.c (debug): Likewise.
15471 * gimple-ssa-store-merging.c (bswap_replace): Likewise.
15472 (merged_store_group::apply_stores): Likewise.
15473 * gimple-ssa-strength-reduction.c (insert_initializers): Likewise.
15474 * gimple.c (verify_gimple_pp): Likewise.
15475 * graphite-poly.c (print_pbb_body): Likewise.
15476 * passes.c (pass_manager::register_one_dump_file): Convert
15477 local "optgroup_flags" to optgroup_flags_t.
15478 * print-tree.c (print_node): Use TDF_NONE rather than 0.
15480 (debug_body): Likewise.
15481 * tree-pass.h (struct pass_data): Convert field "optgroup_flags"
15482 to optgroup_flags_t.
15483 * tree-pretty-print.c (print_struct_decl): Use TDF_NONE rather
15485 * tree-ssa-math-opts.c (convert_mult_to_fma_1): Likewise.
15486 (convert_mult_to_fma): Likewise.
15487 * tree-ssa-reassoc.c (undistribute_ops_list): Likewise.
15488 * tree-ssa-sccvn.c (vn_eliminate): Likewise.
15489 * tree-vect-data-refs.c (dump_lower_bound): Convert param
15490 "dump_kind" to dump_flags_t.
15492 2018-06-08 Segher Boessenkool <segher@kernel.crashing.org>
15494 * config/rs6000/rs6000.c (min, max): Delete.
15496 2018-06-08 Segher Boessenkool <segher@kernel.crashing.org>
15498 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete mention of
15499 -mabi=spe and -mabi=no-spe.
15501 2018-06-08 Martin Liska <mliska@suse.cz>
15503 * ipa-pure-const.c (propagate_pure_const): Use ::get at places
15504 where we expect an existing summary.
15506 2018-06-08 Martin Liska <mliska@suse.cz>
15508 * ipa-inline-analysis.c (simple_edge_hints): Use ::get method.
15509 * ipa-inline.h (estimate_edge_growth): Likewise.
15511 2018-06-08 Martin Liska <mliska@suse.cz>
15513 * cgraph.c (function_version_hasher::hash): Use
15514 cgraph_node::get_uid ().
15515 (function_version_hasher::equal):
15516 * cgraph.h (cgraph_node::get_uid): New method.
15517 * ipa-inline.c (update_caller_keys): Use
15518 cgraph_node::get_uid ().
15519 (update_callee_keys): Likewise.
15520 * ipa-utils.c (searchc): Likewise.
15521 (ipa_reduced_postorder): Likewise.
15522 * lto-cgraph.c (input_node): Likewise.
15523 * passes.c (is_pass_explicitly_enabled_or_disabled): Likewise.
15524 * symbol-summary.h (symtab_insertion): Likewise.
15525 (symtab_removal): Likewise.
15526 (symtab_duplication): Likewise.
15527 * tree-pretty-print.c (dump_function_header): Likewise.
15528 * tree-sra.c (convert_callers_for_node): Likewise.
15530 2018-06-08 Martin Liska <mliska@suse.cz>
15532 * cgraph.c (symbol_table::create_edge): Always assign a new
15534 (symbol_table::free_edge): Do not recycle numbers.
15535 * cgraph.h (cgraph_edge::get): New method.
15536 * symbol-summary.h (symtab_removal): Use it.
15537 (symtab_duplication): Likewise.
15538 (call_summary::hashable_uid): Remove.
15540 2018-06-08 Martin Liska <mliska@suse.cz>
15542 * ipa-inline-analysis.c (inline_edge_removal_hook): Remove.
15543 (initialize_growth_caches): Remove.
15544 (free_growth_caches): Likewise.
15545 (do_estimate_edge_time): Use edge_growth_cache.
15546 (do_estimate_edge_size): Likewise.
15547 (do_estimate_edge_hints): Likewise.
15548 * ipa-inline.c (reset_edge_caches): Likewise.
15549 (recursive_inlining): Likewise.
15550 (inline_small_functions): Likewise.
15551 * ipa-inline.h (initialize_growth_caches): Remove.
15552 (estimate_edge_size): Likewise.
15553 (estimate_edge_time): Likewise.
15554 (estimate_edge_hints): Likewise.
15555 (reset_edge_growth_cache): Likewise.
15556 * symbol-summary.h (call_summary::remove): New method.
15558 2018-06-08 Martin Liska <mliska@suse.cz>
15560 * ipa-cp.c (class edge_clone_summary): New summary.
15561 (grow_edge_clone_vectors): Remove.
15562 (ipcp_edge_duplication_hook): Remove.
15563 (class edge_clone_summary_t): New call_summary class.
15564 (ipcp_edge_removal_hook): Remove.
15565 (edge_clone_summary_t::duplicate): New function.
15566 (get_next_cgraph_edge_clone): Use edge_clone_summaries.
15567 (create_specialized_node): Likewise.
15568 (ipcp_driver): Initialize edge_clone_summaries and do not
15571 2018-06-08 Martin Liska <mliska@suse.cz>
15573 * symbol-summary.h (get): New function.
15574 (call_summary::m_initialize_when_cloning): New class member.
15576 2018-06-08 Martin Liska <mliska@suse.cz>
15578 * cgraph.c (cgraph_node::remove): Do not recycle uid.
15579 * cgraph.h (symbol_table::release_symbol): Do not pass uid.
15580 (symbol_table::allocate_cgraph_symbol): Do not set uid.
15581 * passes.c (uid_hash_t): Record removed_nodes by their uids.
15582 (remove_cgraph_node_from_order): Use the removed_nodes set.
15583 (do_per_function_toporder): Likwise.
15584 * symbol-summary.h (symtab_insertion): Use cgraph_node::uid
15585 instead of summary_uid.
15586 (symtab_removal): Likewise.
15587 (symtab_duplication): Likewise.
15589 2018-06-08 Martin Liska <mliska@suse.cz>
15591 * ipa-cp.c (ipcp_store_bits_results): Use
15592 ipcp_transformation_sum.
15593 (ipcp_store_vr_results): Likewise.
15594 * ipa-prop.c (ipcp_grow_transformations_if_necessary): Renamed
15596 (ipcp_transformation_initialize): ... this.
15597 (ipa_set_node_agg_value_chain):
15598 (ipa_node_params_t::duplicate): Use ipcp_transformation_sum.
15599 (write_ipcp_transformation_info): Likewise.
15600 (read_ipcp_transformation_info): Likewise.
15601 (ipcp_update_bits): Likewise.
15602 (ipcp_update_vr): Likewise.
15603 (ipcp_transform_function): Likewise.
15604 * ipa-prop.h: Rename ipcp_transformation_summary to
15605 ipcp_transformation.
15606 (class ipcp_transformation_t): New function summary.
15607 (ipcp_get_transformation_summary): Use ipcp_transformation_sum.
15608 (ipa_get_agg_replacements_for_node): Likewise.
15610 2018-06-08 Martin Liska <mliska@suse.cz>
15612 * ipa-pure-const.c (struct funct_state_d): Do it class instead
15614 (class funct_state_summary_t): New function_summary class.
15615 (has_function_state): Remove.
15616 (get_function_state): Likewise.
15617 (set_function_state): Likewise.
15618 (add_new_function): Likewise.
15619 (funct_state_summary_t::insert): New function.
15620 (duplicate_node_data): Remove.
15621 (remove_node_data): Remove.
15622 (funct_state_summary_t::duplicate): New function.
15623 (register_hooks): Create new funct_state_summaries.
15624 (pure_const_generate_summary): Use it.
15625 (pure_const_write_summary): Likewise.
15626 (pure_const_read_summary): Likewise.
15627 (propagate_pure_const): Likewise.
15628 (propagate_nothrow): Likewise.
15629 (dump_malloc_lattice): Likewise.
15630 (propagate_malloc): Likewise.
15631 (execute): Do not register hooks, just remove summary
15633 (pass_ipa_pure_const::pass_ipa_pure_const): Simplify
15636 2018-06-08 Martin Liska <mliska@suse.cz>
15638 * ipa-reference.c (remove_node_data): Remove.
15639 (duplicate_node_data): Likewise.
15640 (class ipa_ref_var_info_summary_t): New class.
15641 (class ipa_ref_opt_summary_t): Likewise.
15642 (get_reference_vars_info): Use ipa_ref_var_info_summaries.
15643 (get_reference_optimization_summary): Use
15644 ipa_ref_opt_sum_summaries.
15645 (set_reference_vars_info): Remove.
15646 (set_reference_optimization_summary): Likewise.
15647 (ipa_init): Create summaries.
15648 (init_function_info): Use function summary.
15649 (ipa_ref_opt_summary_t::duplicate): New function.
15650 (ipa_ref_opt_summary_t::remove): New function.
15651 (get_read_write_all_from_node): Fix GNU coding style.
15652 (propagate): Use function summary.
15653 (write_node_summary_p): Fix GNU coding style.
15654 (stream_out_bitmap): Likewise.
15655 (ipa_reference_read_optimization_summary): Use function summary.
15656 (ipa_reference_c_finalize): Do not release hooks.
15658 2018-06-08 Martin Liska <mliska@suse.cz>
15660 * ipa-fnsummary.c (dump_ipa_call_summary): Use ::get method.
15661 (analyze_function_body): Extract multiple calls of get_create.
15662 * ipa-inline-analysis.c (simple_edge_hints): Likewise.
15663 * ipa-inline.c (recursive_inlining): Use ::get method.
15664 * ipa-inline.h (estimate_edge_growth): Likewise.
15666 2018-06-08 Martin Liska <mliska@suse.cz>
15668 * hsa-common.h (enum hsa_function_kind): Rename HSA_NONE to
15670 (hsa_function_summary::hsa_function_summary): Use the new enum
15672 (hsa_gpu_implementation_p): Use hsa_summaries::get.
15673 * hsa-gen.c (hsa_get_host_function): Likewise.
15674 (get_brig_function_name): Likewise.
15675 * ipa-hsa.c (process_hsa_functions): Likewise.
15676 (ipa_hsa_write_summary): Likewise.
15677 * symbol-summary.h (symtab_duplication): Use ::get function/
15678 (get): New function.
15680 2018-06-08 Martin Liska <mliska@suse.cz>
15682 * config/i386/i386.c (ix86_can_inline_p): Use get_create instead
15684 * hsa-common.c (hsa_summary_t::link_functions): Likewise.
15685 (hsa_register_kernel): Likewise.
15686 * hsa-common.h (hsa_gpu_implementation_p): Likewise.
15687 * hsa-gen.c (hsa_get_host_function): Likewise.
15688 (get_brig_function_name): Likewise.
15689 (generate_hsa): Likewise.
15690 (pass_gen_hsail::execute): Likewise.
15691 * ipa-cp.c (ipcp_cloning_candidate_p): Likewise.
15692 (devirtualization_time_bonus): Likewise.
15693 (ipcp_propagate_stage): Likewise.
15694 * ipa-fnsummary.c (redirect_to_unreachable): Likewise.
15695 (edge_set_predicate): Likewise.
15696 (evaluate_conditions_for_known_args): Likewise.
15697 (evaluate_properties_for_edge): Likewise.
15698 (ipa_fn_summary::reset): Likewise.
15699 (ipa_fn_summary_t::duplicate): Likewise.
15700 (dump_ipa_call_summary): Likewise.
15701 (ipa_dump_fn_summary): Likewise.
15702 (analyze_function_body): Likewise.
15703 (compute_fn_summary): Likewise.
15704 (estimate_edge_devirt_benefit): Likewise.
15705 (estimate_edge_size_and_time): Likewise.
15706 (estimate_calls_size_and_time): Likewise.
15707 (estimate_node_size_and_time): Likewise.
15708 (inline_update_callee_summaries): Likewise.
15709 (remap_edge_change_prob): Likewise.
15710 (remap_edge_summaries): Likewise.
15711 (ipa_merge_fn_summary_after_inlining): Likewise.
15712 (ipa_update_overall_fn_summary): Likewise.
15713 (read_ipa_call_summary): Likewise.
15714 (inline_read_section): Likewise.
15715 (write_ipa_call_summary): Likewise.
15716 (ipa_fn_summary_write): Likewise.
15717 (ipa_free_fn_summary): Likewise.
15718 * ipa-hsa.c (process_hsa_functions): Likewise.
15719 (ipa_hsa_write_summary): Likewise.
15720 (ipa_hsa_read_section): Likewise.
15721 * ipa-icf.c (sem_function::merge): Likewise.
15722 * ipa-inline-analysis.c (simple_edge_hints): Likewise.
15723 (do_estimate_edge_time): Likewise.
15724 (estimate_size_after_inlining): Likewise.
15725 (estimate_growth): Likewise.
15726 (growth_likely_positive): Likewise.
15727 * ipa-inline-transform.c (clone_inlined_nodes): Likewise.
15728 (inline_call): Likewise.
15729 * ipa-inline.c (caller_growth_limits): Likewise.
15730 (can_inline_edge_p): Likewise.
15731 (can_inline_edge_by_limits_p): Likewise.
15732 (compute_uninlined_call_time): Likewise.
15733 (compute_inlined_call_time): Likewise.
15734 (want_inline_small_function_p): Likewise.
15735 (edge_badness): Likewise.
15736 (update_caller_keys): Likewise.
15737 (update_callee_keys): Likewise.
15738 (recursive_inlining): Likewise.
15739 (inline_small_functions): Likewise.
15740 (inline_to_all_callers_1): Likewise.
15741 (dump_overall_stats): Likewise.
15742 (early_inline_small_functions): Likewise.
15743 (early_inliner): Likewise.
15744 * ipa-inline.h (estimate_edge_growth): Likewise.
15745 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
15746 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
15747 * ipa-prop.h (IPA_NODE_REF): Likewise.
15748 (IPA_EDGE_REF): Likewise.
15749 * ipa-pure-const.c (malloc_candidate_p): Likewise.
15750 (propagate_malloc): Likewise.
15751 * ipa-split.c (execute_split_functions): Likewise.
15752 * symbol-summary.h: Rename get to get_create.
15754 (get_create): Likewise.
15755 * tree-sra.c (ipa_sra_preliminary_function_checks): Likewise.
15757 2018-06-08 Martin Liska <mliska@suse.cz>
15759 * symbol-summary.h (release): Move definition out of class
15761 (symtab_removal): Likewise.
15762 (symtab_duplication): Likewise.
15764 2018-06-08 Martin Liska <mliska@suse.cz>
15766 * symbol-summary.h (function_summary): Move constructor
15767 implementation out of class declaration.
15768 (release): Likewise.
15769 (symtab_insertion): Likewise.
15770 (symtab_removal): Likewise.
15771 (symtab_duplication): Likewise.
15774 2018-06-08 Martin Liska <mliska@suse.cz>
15776 * Makefile.in: Remove support for MPX (macros, related functions,
15777 fields in cgraph_node, ...).
15778 * builtin-types.def (BT_BND): Likewise.
15779 (BT_FN_BND_CONST_PTR): Likewise.
15780 (BT_FN_CONST_PTR_BND): Likewise.
15781 (BT_FN_VOID_PTR_BND): Likewise.
15782 (BT_FN_BND_CONST_PTR_SIZE): Likewise.
15783 (BT_FN_VOID_CONST_PTR_BND_CONST_PTR): Likewise.
15784 * builtins.c (expand_builtin_memcpy_with_bounds): Likewise.
15785 (expand_builtin_mempcpy_with_bounds): Likewise.
15786 (expand_builtin_memset_with_bounds): Likewise.
15787 (expand_builtin_memset_args): Likewise.
15788 (std_expand_builtin_va_start): Likewise.
15789 (expand_builtin): Likewise.
15790 (expand_builtin_with_bounds): Likewise.
15791 * builtins.def (DEF_BUILTIN_CHKP): Likewise.
15792 (DEF_LIB_BUILTIN_CHKP): Likewise.
15793 (DEF_EXT_LIB_BUILTIN_CHKP): Likewise.
15794 (DEF_CHKP_BUILTIN): Likewise.
15795 (BUILT_IN_MEMCPY): Likewise.
15796 (BUILT_IN_MEMMOVE): Likewise.
15797 (BUILT_IN_MEMPCPY): Likewise.
15798 (BUILT_IN_MEMSET): Likewise.
15799 (BUILT_IN_STPCPY): Likewise.
15800 (BUILT_IN_STRCAT): Likewise.
15801 (BUILT_IN_STRCHR): Likewise.
15802 (BUILT_IN_STRCPY): Likewise.
15803 (BUILT_IN_STRLEN): Likewise.
15804 (BUILT_IN_MEMCPY_CHK): Likewise.
15805 (BUILT_IN_MEMMOVE_CHK): Likewise.
15806 (BUILT_IN_MEMPCPY_CHK): Likewise.
15807 (BUILT_IN_MEMSET_CHK): Likewise.
15808 (BUILT_IN_STPCPY_CHK): Likewise.
15809 (BUILT_IN_STRCAT_CHK): Likewise.
15810 (BUILT_IN_STRCPY_CHK): Likewise.
15811 * calls.c (store_bounds): Likewise.
15812 (emit_call_1): Likewise.
15813 (special_function_p): Likewise.
15814 (maybe_warn_nonstring_arg): Likewise.
15815 (initialize_argument_information): Likewise.
15816 (finalize_must_preallocate): Likewise.
15817 (compute_argument_addresses): Likewise.
15818 (expand_call): Likewise.
15819 * cfgexpand.c (expand_call_stmt): Likewise.
15820 (expand_return): Likewise.
15821 (expand_gimple_stmt_1): Likewise.
15822 (pass_expand::execute): Likewise.
15823 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
15824 (cgraph_node::remove): Likewise.
15825 (cgraph_node::dump): Likewise.
15826 (cgraph_node::verify_node): Likewise.
15827 * cgraph.h (chkp_function_instrumented_p): Likewise.
15828 (symtab_node::get_alias_target): Likewise.
15829 (cgraph_node::can_remove_if_no_direct_calls_and_refs_p): Likewise.
15830 (cgraph_local_p): Likewise.
15831 * cgraphbuild.c (cgraph_edge::rebuild_edges): Likewise.
15832 (cgraph_edge::rebuild_references): Likewise.
15833 * cgraphunit.c (varpool_node::finalize_decl): Likewise.
15834 (walk_polymorphic_call_targets): Likewise.
15835 (cgraph_node::expand_thunk): Likewise.
15836 (symbol_table::output_weakrefs): Likewise.
15837 * common/config/i386/i386-common.c (OPTION_MASK_ISA2_GENERAL_REGS_ONLY_UNSET): Likewise.
15838 (ix86_handle_option): Likewise.
15839 * config/i386/constraints.md: Likewise.
15840 * config/i386/i386-builtin-types.def (BND): Likewise.
15844 * config/i386/i386-builtin.def (BDESC_END): Likewise.
15845 (BDESC_FIRST): Likewise.
15847 * config/i386/i386-c.c (ix86_target_macros_internal): Likewise.
15848 * config/i386/i386-protos.h (ix86_bnd_prefixed_insn_p): Likewise.
15849 * config/i386/i386.c (enum reg_class): Likewise.
15850 (ix86_target_string): Likewise.
15851 (ix86_option_override_internal): Likewise.
15852 (ix86_conditional_register_usage): Likewise.
15853 (ix86_valid_target_attribute_inner_p): Likewise.
15854 (ix86_set_indirect_branch_type): Likewise.
15855 (ix86_set_current_function): Likewise.
15856 (ix86_function_arg_regno_p): Likewise.
15857 (init_cumulative_args): Likewise.
15858 (ix86_function_arg_advance): Likewise.
15859 (ix86_function_arg): Likewise.
15860 (ix86_pass_by_reference): Likewise.
15861 (ix86_function_value_regno_p): Likewise.
15862 (ix86_function_value_1): Likewise.
15863 (ix86_function_value_bounds): Likewise.
15864 (ix86_return_in_memory): Likewise.
15865 (ix86_setup_incoming_vararg_bounds): Likewise.
15866 (ix86_va_start): Likewise.
15867 (indirect_thunk_need_prefix): Likewise.
15868 (print_reg): Likewise.
15869 (ix86_print_operand): Likewise.
15870 (ix86_expand_call): Likewise.
15871 (ix86_output_function_return): Likewise.
15872 (reg_encoded_number): Likewise.
15873 (BDESC_VERIFYS): Likewise.
15874 (ix86_init_mpx_builtins): Likewise.
15875 (ix86_init_builtins): Likewise.
15876 (ix86_emit_cmove): Likewise.
15877 (ix86_emit_move_max): Likewise.
15878 (ix86_expand_builtin): Likewise.
15879 (ix86_builtin_mpx_function): Likewise.
15880 (ix86_get_arg_address_for_bt): Likewise.
15881 (ix86_load_bounds): Likewise.
15882 (ix86_store_bounds): Likewise.
15883 (ix86_load_returned_bounds): Likewise.
15884 (ix86_store_returned_bounds): Likewise.
15885 (ix86_class_likely_spilled_p): Likewise.
15886 (ix86_hard_regno_mode_ok): Likewise.
15887 (x86_order_regs_for_local_alloc): Likewise.
15888 (ix86_mitigate_rop): Likewise.
15889 (ix86_bnd_prefixed_insn_p): Likewise.
15890 (ix86_mpx_bound_mode): Likewise.
15891 (ix86_make_bounds_constant): Likewise.
15892 (ix86_initialize_bounds): Likewise.
15893 (TARGET_LOAD_BOUNDS_FOR_ARG): Likewise.
15894 (TARGET_STORE_BOUNDS_FOR_ARG): Likewise.
15895 (TARGET_LOAD_RETURNED_BOUNDS): Likewise.
15896 (TARGET_STORE_RETURNED_BOUNDS): Likewise.
15897 (TARGET_CHKP_BOUND_MODE): Likewise.
15898 (TARGET_BUILTIN_CHKP_FUNCTION): Likewise.
15899 (TARGET_CHKP_FUNCTION_VALUE_BOUNDS): Likewise.
15900 (TARGET_CHKP_MAKE_BOUNDS_CONSTANT): Likewise.
15901 (TARGET_CHKP_INITIALIZE_BOUNDS): Likewise.
15902 * config/i386/i386.h (TARGET_MPX): Likewise.
15903 (TARGET_MPX_P): Likewise.
15904 (VALID_BND_REG_MODE): Likewise.
15905 (FIRST_BND_REG): Likewise.
15906 (LAST_BND_REG): Likewise.
15907 (enum reg_class): Likewise.
15908 (BND_REG_P): Likewise.
15909 (BND_REGNO_P): Likewise.
15910 (BNDmode): Likewise.
15911 (ADJUST_INSN_LENGTH): Likewise.
15912 * config/i386/i386.md: Likewise.
15913 * config/i386/i386.opt: Likewise.
15914 * config/i386/linux-common.h (LIBMPX_LIBS): Likewise.
15915 (defined): Likewise.
15916 (LINK_MPX): Likewise.
15917 (MPX_SPEC): Likewise.
15918 (LIBMPX_SPEC): Likewise.
15919 (LIBMPXWRAPPERS_SPEC): Likewise.
15920 (CHKP_SPEC): Likewise.
15921 * config/i386/predicates.md: Likewise.
15922 * dbxout.c (dbxout_type): Likewise.
15923 * doc/extend.texi: Likewise.
15924 * doc/invoke.texi: Likewise.
15925 * doc/md.texi: Likewise.
15926 * doc/tm.texi: Likewise.
15927 * doc/tm.texi.in: Likewise.
15928 * dwarf2out.c (is_base_type): Likewise.
15929 (gen_formal_types_die): Likewise.
15930 (gen_subprogram_die): Likewise.
15931 (gen_type_die_with_usage): Likewise.
15932 (gen_decl_die): Likewise.
15933 (dwarf2out_late_global_decl): Likewise.
15934 * expr.c (expand_assignment): Likewise.
15935 (emit_storent_insn): Likewise.
15936 (store_expr_with_bounds): Likewise.
15937 (store_expr): Likewise.
15938 (expand_expr_real_1): Likewise.
15939 * expr.h (store_expr_with_bounds): Likewise.
15940 * function.c (use_register_for_decl): Likewise.
15941 (struct bounds_parm_data): Likewise.
15942 (assign_parms_augmented_arg_list): Likewise.
15943 (assign_parm_find_entry_rtl): Likewise.
15944 (assign_parm_is_stack_parm): Likewise.
15945 (assign_parm_load_bounds): Likewise.
15946 (assign_bounds): Likewise.
15947 (assign_parms): Likewise.
15948 (expand_function_start): Likewise.
15949 * gcc.c (CHKP_SPEC): Likewise.
15950 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
15951 * gimple-ssa-warn-restrict.c (builtin_access::builtin_access): Likewise.
15952 (wrestrict_dom_walker::check_call): Likewise.
15953 * gimple.c (gimple_build_call_from_tree): Likewise.
15954 * gimple.h (enum gf_mask): Likewise.
15955 (gimple_call_with_bounds_p): Likewise.
15956 (gimple_call_set_with_bounds): Likewise.
15957 * gimplify.c (gimplify_init_constructor): Likewise.
15958 * ipa-cp.c (initialize_node_lattices): Likewise.
15959 (propagate_constants_across_call): Likewise.
15960 (find_more_scalar_values_for_callers_subset): Likewise.
15961 * ipa-hsa.c (process_hsa_functions): Likewise.
15962 * ipa-icf-gimple.c (func_checker::compare_gimple_call): Likewise.
15963 * ipa-icf.c (sem_function::merge): Likewise.
15964 * ipa-inline.c (early_inliner): Likewise.
15965 * ipa-pure-const.c (warn_function_noreturn): Likewise.
15966 (warn_function_cold): Likewise.
15967 (propagate_pure_const): Likewise.
15968 * ipa-ref.h (enum GTY): Likewise.
15969 * ipa-split.c (find_retbnd): Likewise.
15970 (consider_split): Likewise.
15971 (split_function): Likewise.
15972 * ipa-visibility.c (cgraph_externally_visible_p): Likewise.
15973 * ipa.c (walk_polymorphic_call_targets): Likewise.
15974 (symbol_table::remove_unreachable_nodes): Likewise.
15975 (process_references): Likewise.
15976 (cgraph_build_static_cdtor_1): Likewise.
15977 * lto-cgraph.c (lto_output_node): Likewise.
15978 (output_refs): Likewise.
15979 (compute_ltrans_boundary): Likewise.
15980 (input_overwrite_node): Likewise.
15981 (input_node): Likewise.
15982 (input_cgraph_1): Likewise.
15983 * params.def (PARAM_CHKP_MAX_CTOR_SIZE): Likewise.
15984 * passes.c (pass_manager::execute_early_local_passes): Likewise.
15985 (class pass_chkp_instrumentation_passes): Likewise.
15986 (make_pass_chkp_instrumentation_passes): Likewise.
15987 * passes.def: Likewise.
15988 * rtl.h (struct GTY): Likewise.
15989 (CALL_EXPR_WITH_BOUNDS_P): Likewise.
15990 * stor-layout.c (layout_type): Likewise.
15991 * symtab.c: Likewise.
15992 * target.def: Likewise.
15993 * targhooks.c (default_chkp_bound_type): Likewise.
15994 (default_chkp_bound_mode): Likewise.
15995 (default_builtin_chkp_function): Likewise.
15996 (default_chkp_function_value_bounds): Likewise.
15997 (default_chkp_make_bounds_constant): Likewise.
15998 (default_chkp_initialize_bounds): Likewise.
15999 * targhooks.h (default_chkp_bound_type): Likewise.
16000 (default_chkp_bound_mode): Likewise.
16001 (default_builtin_chkp_function): Likewise.
16002 (default_chkp_function_value_bounds): Likewise.
16003 (default_chkp_make_bounds_constant): Likewise.
16004 (default_chkp_initialize_bounds): Likewise.
16005 * toplev.c (compile_file): Likewise.
16006 (process_options): Likewise.
16007 * tree-core.h (DEF_BUILTIN): Likewise.
16008 (DEF_BUILTIN_CHKP): Likewise.
16009 * tree-inline.c (declare_return_variable): Likewise.
16010 (remap_gimple_stmt): Likewise.
16011 (copy_bb): Likewise.
16012 (initialize_inlined_parameters): Likewise.
16013 (expand_call_inline): Likewise.
16014 * tree-pass.h (make_pass_ipa_chkp_versioning): Likewise.
16015 (make_pass_ipa_chkp_early_produce_thunks): Likewise.
16016 (make_pass_ipa_chkp_produce_thunks): Likewise.
16017 (make_pass_chkp): Likewise.
16018 (make_pass_chkp_opt): Likewise.
16019 (make_pass_chkp_instrumentation_passes): Likewise.
16020 * tree-pretty-print.c (dump_generic_node): Likewise.
16021 * tree-ssa-ccp.c (insert_clobber_before_stack_restore): Likewise.
16022 * tree-ssa-dce.c (propagate_necessity): Likewise.
16023 (eliminate_unnecessary_stmts): Likewise.
16024 * tree-ssa-pre.c (create_expression_by_pieces): Likewise.
16025 * tree-ssa-sccvn.c (copy_reference_ops_from_call): Likewise.
16026 * tree-ssa-sccvn.h: Likewise.
16027 * tree-ssa-strlen.c (get_string_length): Likewise.
16028 (valid_builtin_call): Likewise.
16029 (adjust_last_stmt): Likewise.
16030 (handle_builtin_strchr): Likewise.
16031 (handle_builtin_strcpy): Likewise.
16032 (handle_builtin_stxncpy): Likewise.
16033 (handle_builtin_memcpy): Likewise.
16034 (handle_builtin_strcat): Likewise.
16035 (strlen_check_and_optimize_stmt): Likewise.
16036 * tree-stdarg.c (expand_ifn_va_arg_1): Likewise.
16037 * tree-streamer-in.c: Likewise.
16038 * tree-streamer.c (record_common_node): Likewise.
16039 * tree.c (tree_code_size): Likewise.
16040 (wide_int_to_tree_1): Likewise.
16041 (type_contains_placeholder_1): Likewise.
16042 (build_common_tree_nodes): Likewise.
16043 * tree.def (POINTER_BOUNDS_TYPE): Likewise.
16044 * tree.h (POINTER_BOUNDS_TYPE_P): Likewise.
16045 (POINTER_BOUNDS_P): Likewise.
16046 (BOUNDED_TYPE_P): Likewise.
16047 (BOUNDED_P): Likewise.
16048 (CALL_WITH_BOUNDS_P): Likewise.
16049 (pointer_bounds_type_node): Likewise.
16050 * value-prof.c (gimple_ic): Likewise.
16051 * var-tracking.c (vt_add_function_parameters): Likewise.
16052 * varasm.c (make_decl_rtl): Likewise.
16053 (assemble_start_function): Likewise.
16054 (output_constant): Likewise.
16055 (maybe_assemble_visibility): Likewise.
16056 * varpool.c (ctor_for_folding): Likewise.
16057 * chkp-builtins.def: Remove.
16058 * ipa-chkp.c: Remove.
16059 * ipa-chkp.h: Remove.
16060 * rtl-chkp.c: Remove.
16061 * rtl-chkp.h: Remove.
16062 * tree-chkp-opt.c: Remove.
16063 * tree-chkp.c: Remove.
16064 * tree-chkp.h: Remove.
16066 2018-06-07 Carl Love <cel@us.ibm.com>
16068 * config/rs6000/vsx.md (vextract_fp_from_shorth,
16069 vextract_fp_from_shortl): Add BE support.
16071 2018-06-07 Paul Koning <ni1d@arrl.net>
16073 * compare-elim.c (try_merge_compare): Don't merge compare if
16074 address contains a side effect.
16075 (try_eliminate_compare): Likewise.
16077 2018-06-07 Olga Makhotina <olga.makhotina@intel.com>
16079 * config.gcc: Support "tremont".
16080 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "tremont".
16081 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
16083 * config/i386/i386.c (m_TREMONT): Define.
16084 (processor_target_table): Add "tremont".
16085 (PTA_TREMONT): Define.
16086 (ix86_lea_outperforms): Add TARGET_TREMONT.
16087 (get_builtin_code_for_version): Handle PROCESSOR_TREMONT.
16088 (fold_builtin_cpu): Add M_INTEL_TREMONT, replace M_INTEL_GOLDMONT
16089 and M_INTEL_GOLDMONT_PLUS.
16090 (fold_builtin_cpu): Add "tremont".
16091 (ix86_add_stmt_cost): Add TARGET_TREMONT.
16092 (ix86_option_override_internal): Add "tremont".
16093 * config/i386/i386.h (processor_costs): Define TARGET_TREMONT.
16094 (processor_type): Add PROCESSOR_TREMONT.
16095 * config/i386/x86-tune.def: Add m_TREMONT.
16096 * doc/invoke.texi: Add tremont as x86 -march=/-mtune= CPU type.
16098 2018-06-07 Jozef Lawrynowicz <jozef.l@mittosystems.com>
16100 * config/msp430/msp430.c (msp430_mcu_name): Set the "i" in the
16101 symbol defined for msp430i* devices to be lower case.
16103 2018-06-07 Richard Biener <rguenther@suse.de>
16105 * graphite-sese-to-poly.c (extract_affine): Avoid unneded
16106 wrapping. Properly wrap the result of a BIT_NOT_EXPR.
16107 Properly wrap signed arithmetic if overflow wraps.
16109 2018-06-07 Jakub Jelinek <jakub@redhat.com>
16111 PR tree-optimization/69615
16112 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If rhs2 is lhs
16113 of a cast from a same precision integral SSA_NAME in a bb dominated
16114 by first_bb, retry with rhs2 set to the rhs1 of the cast. Don't emit
16115 cast to utype if rhs2 has already a compatible type.
16117 2018-06-07 Richard Biener <rguenther@suse.de>
16119 PR tree-optimization/85935
16120 * graphite-scop-detection.c (find_params_in_bb): Analyze
16121 condition operands with respect to the correct loop. Assert
16122 the analysis doesn't fail.
16124 2018-06-04 Carl Love <cel@us.ibm.com>
16126 * config/rs6000/vsx.md (first_match_index_<mode>): Calculate index
16127 using natural element order. Use gen_lshrsi3 instead of gen_ashrsi3
16128 as it is slightly cheaper.
16129 (first_match_or_eos_index_<mode>):
16130 Calculate index using natural element order.
16131 (first_match_index_<mode>):
16132 Calculate index using natural element order.
16133 (first_match_or_eos_index_<mode>):
16134 Calculate index using natural order.
16135 (define_insn vclzlsbb): Change to define_insn vclzlsbb_<mode>.
16136 for BE and LE modes.
16137 * config/rs6000/rs6000-c.c: Rename P9V_BUILTIN_VCLZLSBB,
16138 P9V_BUILTIN_VCLZLSBB_V16QI.
16139 * config/rs6000/rs6000-builtin.def: Make VCLZLSBB mode
16142 2018-06-06 Kelvin Nilsen <kelvin@gcc.gnu.org>
16144 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
16145 indentation and line wrap for many prototypes. Add missing
16146 @smallexample directives around block of prototypes for vec_xl and
16149 2018-06-05 Michael Meissner <meissner@linux.ibm.com>
16151 * config/rs6000/rs6000.c (rs6000_passes_ieee128): New boolean to
16152 track if we pass or return IEEE 128-bit floating point.
16153 (ieee128_mangling_gcc_8_1): New boolean to say whether to generate
16154 C++ mangling that is compatible with GCC 8.1.
16155 (TARGET_ASM_GLOBALIZE_DECL_NAME): Override target hook.
16156 (init_cumulative_args): Note if we pass or return IEEE 128-bit
16157 floating point types.
16158 (rs6000_function_arg_advance_1): Likewise.
16159 (rs6000_mangle_type): Optionally generate mangled names that match
16160 what GCC 8.1 generated for IEEE 128-bit floating point types.
16161 (rs6000_globalize_decl_name): If we have an external function that
16162 passes or returns IEEE 128-bit types, generate a weak reference
16163 from the mangled name used in GCC 8.1 to the current mangled
16165 (rs6000_init_builtins): Make __ibm128 use the long double type if
16166 long double is IBM extended double. Make __float128 use the long
16167 double type if long double is IEEE 128-bit.
16170 * config/rs6000/rs6000-builtin.def (BU_IBM128_2): New helper
16171 macro for __ibm128 built-in functions.
16172 (PACK_IF): Add __ibm128 pack/unpack functions.
16173 (UNPACK_IF): Likewise.
16174 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Do not
16175 enable long double built-in functions if long double is IEEE
16176 128-bit floating point.
16177 (rs6000_invalid_builtin): Update long double built-in function
16179 (rs6000_expand_builtin): For PACK_IF and UNPACK_IF built-in
16180 functions, adjust the built-in function to use the long double
16181 built-in function if __ibm128 and long double are the same type.
16182 * doc/extend.texi (PowerPC builtins): Update documention for
16183 __builtin_{,un}pack_longdouble. Add documentation for
16184 __builtin_{,un}pack_ibm128.
16186 2018-06-06 Jim Wilson <jimw@sifive.com>
16188 * config/riscv/riscv.c (enum riscv_privilege_levels): New.
16189 (struct machine_function): New field interrupt_mode.
16190 (riscv_handle_type_attribute): New function. Add forward declaration.
16191 (riscv_attribute_table) <interrupt>: Use riscv_handle_type_attribute.
16192 (riscv_expand_epilogue): Check interrupt_mode field.
16193 (riscv_set_current_function): Check interrupt attribute args and
16194 set interrupt_mode field.
16195 * config/riscv/riscv.md (UNSPECV_SRET, UNSPECV_URET): New.
16196 (riscv_sret, riscv_uret): New.
16197 * doc/extend.texi (RISC-V Function Attributes) <interrupt>: Document
16198 new arguments to interrupt attribute.
16200 2018-06-06 Peter Bergner <bergner@vnet.ibm.com>
16203 * /config/rs6000/rs6000.h (ASM_CPU_SPEC): Add support for -mpower9.
16204 Don't handle -mcpu=power8 if -mpower9-vector is also used.
16206 2018-06-06 Kelvin Nilsen <kelvin@gcc.gnu.org>
16208 * config/rs6000/rs6000-builtin.def (VSX_BUILTIN_VEC_LD,
16209 VSX_BUILTIN_VEC_ST): Add comment to explain non-traditional uses.
16210 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
16211 several redundant entries.
16213 2018-06-06 David Malcolm <dmalcolm@redhat.com>
16215 * config/i386/i386-protos.h (ix86_expand_call): Strengthen return
16216 type from "rtx" to "rtx_insn *".
16217 * config/i386/i386.c (ix86_expand_split_stack_prologue): Likewise
16218 for local "call_insn", removing cast.
16219 (ix86_expand_call): Likewise, introducing a "call_insn" local.
16221 2018-06-06 Eric Botcazou <ebotcazou@adacore.com>
16223 PR tree-optimization/86066
16224 * gimple-ssa-store-merging.c (process_store): Do not bypass BIT_NOT_EXPR
16225 for BIT_INSERT_EXPR stores.
16227 2018-06-06 Richard Biener <rguenther@suse.de>
16229 PR tree-optimization/86062
16230 * tree-ssa-sccvn.c (vn_reference_lookup_3): Handle arbitrary
16231 component refs ontop
16232 of to be offsetted base.
16234 2018-06-06 Jozef Lawrynowicz <jozef.l@mittosystems.com>
16236 * config/msp430/msp430.c (msp430_attr): Allow interrupt handlers
16237 to be static and remove check on interrupt attribute name.
16239 2018-06-05 Kelvin Nilsen <kelvin@gcc.gnu.org>
16241 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Remove
16242 volatile qualifier from vec_lvsl and vec_lvsr argument prototypes.
16244 2018-06-05 Steve Ellcey <sellcey@cavium.com>
16247 * config/aarch64/aarch64-protos.h (aarch64_err_no_fpadvsimd): Remove
16249 * config/aarch64/aarch64-protos..c (aarch64_err_no_fpadvsimd):
16250 Remove second argument, change how error is called.
16251 (aarch64_layout_arg): Remove second argument from
16252 aarch64_err_no_fpadvsimd call.
16253 (aarch64_init_cumulative_args): Ditto.
16254 (aarch64_gimplify_va_arg_expr): Ditto.
16255 * config/aarch64/aarch64.md (mov<mode>): Ditto.
16257 2018-06-05 Uros Bizjak <ubizjak@gmail.com>
16259 * config/i386/i386.md (simple_return_indirect_internal): New expander.
16260 (*simple_return_indirect_internal<mode>): Rename from
16261 simple_return_indirect_internal. Use W mode iterator.
16262 (rstorssp): New expander.
16263 (*rstorssp<mode>): Rename from rstorssp. Use P mode iterator.
16264 (clrssbsy): New expander.
16265 (*clrssbsy<mode>): Rename from clrssbsy. Use P mode iterator.
16267 2018-06-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
16269 * config/arm/arm_cmse.h (cmse_nsfptr_create): Change typeof to
16271 (cmse_check_pointed_object): Likewise.
16273 2018-06-05 Martin Liska <mliska@suse.cz>
16275 PR gcov-profile/47618
16276 * doc/invoke.texi: Document how -fprofile-dir format
16279 2018-06-05 Richard Biener <rguenther@suse.de>
16281 * tree-cfgcleanup.c (cleanup_control_flow_pre): For edge
16282 removal pretend DOM info isn't available so we do not update
16283 it and only remove edges, not dominated blocks. Actually free
16284 DOM info in case we removed something. Remove unreachable blocks.
16285 (mfb_keep_latches): Work with either DOM info or marked backedges.
16286 (cleanup_tree_cfg_noloop): Do not remove unreachable blocks
16287 first. Mark backedges if DOM info isn't available.
16288 (Re-)compute DOM info after cleanup_control_flow_pre.
16290 2018-06-05 Richard Biener <rguenther@suse.de>
16292 * tree-cfg.c (struct locus_discrim_map): Store line, not location.
16293 (locus_discrim_hasher::hash): Adjust.
16294 (locus_discrim_hasher::equal): Likewise.
16295 (next_discriminator_for_locus): Work on line directly.
16296 (same_line_p): Pass in expanded locus1 as well.
16297 (assign_discriminators): Avoid redundant location expansions.
16299 2018-06-05 Richard Biener <rguenther@suse.de>
16301 PR tree-optimization/86046
16302 * tree-ssa.c (maybe_optimize_var): Clear DECL_GIMPLE_REG_P
16303 if required after clearing TREE_ADDRESSABLE.
16305 2018-06-05 Richard Biener <rguenther@suse.de>
16307 PR tree-optimization/86047
16308 * tree-ssa-loop.c (for_each_index): Glob handling of all
16309 decls and constants and really handle all of them.
16311 2018-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16314 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
16315 qualifier_void_pointer and qualifier_const_void_pointer.
16316 (arm_ldc_qualifiers, arm_stc_qualifiers): Use the above.
16317 (arm_init_builtins): Handle the above.
16318 * config/arm/arm_acle.h (__arm_cdp, __arm_ldc, __arm_ldcl, __arm_stc,
16319 __arm_stcl, __arm_mcr, __arm_cdp2, __arm_ldc2, __arm_ldcl2, __arm_stc2,
16320 __arm_stcl2,__arm_mcr2, __arm_mcrr, __arm_mcrr2): Remove return for
16323 2018-06-05 Martin Liska <mliska@suse.cz>
16325 * auto-profile.c (read_autofdo_file): Do not use
16326 gcov_ctr_summary struct.
16327 (afdo_callsite_hot_enough_for_early_inline): Likewise.
16328 * coverage.c (struct counts_entry): Likewise.
16329 (read_counts_file): Read just single summary entry.
16330 (get_coverage_counts): Use gcov_summary struct.
16331 * coverage.h (get_coverage_counts): Likewise.
16332 * gcov-dump.c (dump_working_sets): Likewise.
16333 (tag_summary): Dump just single summary.
16334 * gcov-io.c (gcov_write_summary): Write just histogram
16336 (gcov_read_summary): Read just single summary.
16337 (compute_working_sets): Use gcov_summary struct.
16338 * gcov-io.h (GCOV_TAG_SUMMARY_LENGTH): Remove usage
16339 of GCOV_COUNTERS_SUMMABLE.
16340 (GCOV_COUNTERS_SUMMABLE): Remove.
16341 (GCOV_FIRST_VALUE_COUNTER): Replace with
16342 GCOV_COUNTER_V_INTERVAL.
16343 (struct gcov_ctr_summary): Remove.
16344 (struct gcov_summary): Directly use fields of former
16346 (compute_working_sets): Use gcov_summary struct.
16347 * gcov.c (read_count_file): Do not use ctrs fields.
16348 * lto-cgraph.c (merge_profile_summaries): Use gcov_summary
16350 * lto-streamer.h (struct GTY): Make profile_info gcov_summary
16352 * profile.c: Likewise.
16353 * profile.h: Likewise.
16355 2018-06-05 Martin Liska <mliska@suse.cz>
16357 PR gcov-profile/84846
16358 * gcov.c (output_lines): Print working directory only
16359 in intermediate format.
16361 2018-06-05 Andreas Krebbel <krebbel@linux.ibm.com>
16363 * config/s390/s390-builtin-types.def: Add void function type.
16364 * config/s390/s390-builtins.def: Use the function type for the
16367 2018-06-04 Jim Wilson <jimw@sifive.com>
16369 * config/riscv/riscv-protos.h (riscv_expand_epilogue): Change bool arg
16371 * config/riscv/riscv.c (riscv_for_each_saved_reg): New args epilogue
16372 and maybe_eh_return. Change regno to unsigned int. Use new args to
16373 handle EH_RETURN_DATA_REGNO registers properly.
16374 (riscv_expand_prologue): Pass new args to riscv_for_each_saved_reg.
16375 (riscv_expand_epilogue): Update comment. Change argument name and
16376 type. Update code to use new name and type. Pass new args to
16377 riscv_for_each_saved_reg. Only use EH_RETURN_STACKADJ_RTX when
16379 * config/riscv/riscv.md (NORMAL_RETURN): New.
16380 (SIBCALL_RETURN, EXCEPTION_RETURN): New.
16381 (epilogue, sibcall_epilogue): Update riscv_expand_epilogue arg.
16382 (eh_return): Call gen_eh_return_internal and emit barrier.
16383 (eh_return_internal): Call riscv_expand_epilogue.
16385 2018-06-04 Eric Botcazou <ebotcazou@adacore.com>
16387 * gimple-ssa-store-merging.c (struct merged_store_group): Move up
16388 bit_insertion field and declare can_be_merged_into method.
16389 (merged_store_group::can_be_merged_into): New method.
16390 (imm_store_chain_info::coalesce_immediate): Call it to decide whether
16391 consecutive non-overlapping stores can be merged. Turn MEM_REF stores
16392 into BIT_INSERT_EXPR stores if the group contains a non-MEM_REF store.
16394 2018-06-04 Richard Biener <rguenther@suse.de>
16396 PR tree-optimization/85955
16397 * builtins.c (fold_builtin_sincos): Convert pointers to
16398 destination to appropriate type before dereferencing.
16400 2018-06-04 Segher Boessenkool <segher@kernel.crashing.org>
16402 * config/rs6000/rs6000.md (abs<mode>2 for FLOAT128): Handle IFmode.
16404 2018-06-04 Richard Sandiford <richard.sandiford@linaro.org>
16406 * expr.c (expand_expr_real_1): Force the operand into memory if
16407 its TYPE_MODE is BLKmode and if there is no integer mode for
16408 the number of bits being extracted.
16410 2018-06-04 Jakub Jelinek <jakub@redhat.com>
16414 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>_1):
16415 Use vptestnm rather than vptestm in (=Yc,v,C) variant.
16417 2018-06-04 Richard Biener <rguenther@suse.de>
16419 * tree-cfgcleanup.c (cleanup_tree_cfg_1): Fold into...
16420 (cleanup_tree_cfg_noloop): ... single caller. Do
16421 start_recording_case_labels later.
16423 2018-06-04 Sebastian Peryt <sebastian.peryt@intel.com>
16425 * config/i386/cldemoteintrin.h: Change define from _X86INTRIN_H_INCLUDED
16426 to _IMMINTRIN_H_INCLUDED.
16427 * config/i386/pconfigintrin.h: Ditto.
16428 * config/i386/waitpkgintrin.h: Ditto.
16429 * config/i386/immintrin.h: Add includes for sgxintrin.h,
16430 pconfigintrin.h, waitpkgintrin.h and cldemoteintrin.h.
16431 * config/i386/x86intrin.h: Remove includes for mintrin.h, xmmintrin.h,
16432 emmintrin.h, pmmintrin.h, tmmintrin.h, smmintrin.h, wmmintrin.h,
16433 bmiintrin.h, bmi2intrin.h, lzcntintrin.h, sgxintrin.h, pconfigintrin.h,
16434 waitpkgintrin.h and cldemoteintrin.h.
16436 2018-06-04 Richard Biener <rguenther@suse.de>
16438 PR tree-optimization/86038
16439 * tracer.c (find_best_successor): Check probability for
16440 being initialized, bail out if not.
16442 2018-06-04 Richard Earnshaw <rearnsha@arm.com>
16445 * config/arm/arm-cpus.in (ALL_QUIRKS): Add xscale feature to the list
16446 of bits to ignore when comparing architectures.
16448 2018-06-04 Jakub Jelinek <jakub@redhat.com>
16450 PR tree-optimization/69615
16451 * fold-const.c (merge_ranges): If range1 is - [x, x] and x is the
16452 maximum or minimum of the type, try to merge it also as if
16453 range1 is + [-, x - 1] or + [x + 1, -].
16456 * tree.c (inchash::add_expr): Handle IDENTIFIER_NODE.
16458 2018-06-03 Eric Botcazou <ebotcazou@adacore.com>
16460 PR tree-optimization/86034
16461 * gimple-ssa-store-merging.c (output_merged_store): Convert the RHS to
16462 the unsigned bitfield type in a bit insertion sequence if it does not
16463 have a larger precision than the bitfield size.
16464 (process_store): Also bypass widening conversions for BIT_INSERT_EXPR.
16466 2018-06-03 Kito Cheng <kito.cheng@gmail.com>
16468 * config/nds32/nds32-peephole2.md: Add new patterns for code size.
16470 2018-06-03 Chung-Ju Wu <jasonwucj@gmail.com>
16472 * config/nds32/nds32-opts.h (nds32_arch_type): Add ARCH_V3J.
16473 * config/nds32/nds32.c (nds32_option_override): Consider ARCH_V3J.
16474 * config/nds32/nds32.h (TARGET_ISA_V3): Add ARCH_V3J checking.
16475 * config/nds32/nds32.opt (march): Add enum value Value(ARCH_V3J).
16477 2018-06-03 Chung-Ju Wu <jasonwucj@gmail.com>
16479 * common/config/nds32/nds32-common.c (nds32_option_optimization_table):
16480 Disable -fdelete-null-pointer-checks for ELF toolchain.
16482 2018-06-02 Chung-Ju Wu <jasonwucj@gmail.com>
16483 Kito Cheng <kito.cheng@gmail.com>
16485 * config.gcc (nds32*): Use nds32-linux.opt and nds32-elf.opt.
16486 (nds32le-*-*, nds32be-*-*): Integrate checking process.
16487 (nds32*-*-*): Add glibc and uclibc conditions.
16488 * common/config/nds32/nds32-common.c (nds32_except_unwind_info): New.
16489 (TARGET_EXCEPT_UNWIND_INFO): Define.
16490 * config/nds32/elf.h: New file.
16491 * config/nds32/linux.h: New file.
16492 * config/nds32/nds32-elf.opt: New file.
16493 * config/nds32/nds32-linux.opt: New file.
16494 * config/nds32/nds32-fp-as-gp.c
16495 (pass_nds32_fp_as_gp::gate): Consider TARGET_LINUX_ABI.
16496 * config/nds32/nds32.c (nds32_conditional_register_usage): Consider
16498 (nds32_asm_file_end): Ditto.
16499 (nds32_print_operand): Ditto.
16500 (nds32_insert_attributes): Ditto.
16501 (nds32_init_libfuncs): New function.
16502 (TARGET_HAVE_TLS): Define.
16503 (TARGET_INIT_LIBFUNCS): Define.
16504 * config/nds32/nds32.h (TARGET_DEFAULT_RELAX): Apply different relax
16506 (TARGET_ELF): Apply different mcmodel setting.
16507 (LINK_SPEC, LIB_SPEC, STARTFILE_SPEC, ENDFILE_SPEC): The content has
16508 been migrated into elf.h and linux.h files.
16509 * config/nds32/nds32.md (add_pc): Consider TARGET_LINUX_ABI.
16510 * config/nds32/nds32.opt (mvh): Consider TARGET_LINUX_ABI.
16511 (mcmodel): The content has been migrated into nds32-elf.opt and
16512 nds32-linux.opt files.
16513 * config/nds32/t-elf: New file.
16514 * config/nds32/t-linux: New file.
16516 2018-06-02 Chung-Ju Wu <jasonwucj@gmail.com>
16517 Shiva Chen <shiva0217@gmail.com>
16519 * config/nds32/constants.md (unspec_volatile_element): Add
16520 UNSPEC_VOLATILE_OMIT_FP_BEGIN and UNSPEC_VOLATILE_OMIT_FP_END.
16521 * config/nds32/nds32-fp-as-gp.c: New implementation of fp_as_gp
16523 * config/nds32/nds32-protos.h (nds32_naked_function_p): Declare.
16524 (make_pass_nds32_fp_as_gp): Declare.
16525 * config/nds32/nds32.c (nds32_register_passes): Add fp_as_gp as one
16527 (nds32_asm_function_end_prologue): Remove unused asm output.
16528 (nds32_asm_function_begin_epilogue): Remove unused asm output.
16529 (nds32_asm_file_start): Output necessary fp_as_gp information.
16530 (nds32_option_override): Adjust register usage.
16531 (nds32_expand_prologue): Consider fp_as_gp situation.
16532 (nds32_expand_prologue_v3push): Consider fp_as_gp situation.
16533 * config/nds32/nds32.md (prologue): Check fp_as_gp_p and naked_p.
16536 (simple_return): Ditto.
16537 (omit_fp_begin): Output special directive for fp_as_gp.
16538 (omit_fp_end): Output special directive for fp_as_gp.
16539 * config/nds32/nds32.opt (mfp-as-gp, mno-fp-as-gp, mforce-fp-as-gp,
16540 mforbid-fp-as-gp): New options.
16542 2018-06-01 Mark Wielaard <mark@klomp.org>
16544 * dwarf2out.c (dwarf2out_finish): Remove generation of
16545 DW_AT_loclists_base.
16547 2018-06-01 Eric Botcazou <ebotcazou@adacore.com>
16549 * gimple-ssa-store-merging.c: Include gimple-fold.h.
16550 (struct store_immediate_info): Document BIT_INSERT_EXPR stores.
16551 (struct merged_store_group): Add bit_insertion field.
16552 (dump_char_array): Use standard hexadecimal format.
16553 (merged_store_group::merged_store_group): Set bit_insertion to false.
16554 (merged_store_group::apply_stores): Use optimal buffer size. Deal
16555 with BIT_INSERT_EXPR stores. Move up code updating the mask and
16556 also print the mask in the dump file.
16557 (pass_store_merging::gate): Minor tweak.
16558 (imm_store_chain_info::coalesce_immediate): Fix wrong association
16559 of stores with groups in dump. Allow coalescing of BIT_INSERT_EXPR
16560 stores with INTEGER_CST stores.
16561 (count_multiple_uses) <BIT_INSERT_EXPR>: New case.
16562 (imm_store_chain_info::output_merged_store): Add try_bitpos variable
16563 and use it throughout. Generate bit insertion sequences if need be.
16564 (pass_store_merging::process_store): Remove redundant condition.
16565 Record stores from a SSA name to a bit-field with BIT_INSERT_EXPR.
16567 2018-06-01 Segher Boessenkool <segher@kernel.crashing.org>
16569 * config/rs6000/rs6000.c (rs6000_mangle_type): Change the mangling of
16570 the 128-bit floating point types. Fix function comment.
16572 2018-06-01 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16574 * config/aarch64/aarch64-simd.md
16575 (aarch64_simd_vec_unpack<su>_lo_<mode>): Use UXTL and SXTL assembler
16577 (aarch64_simd_vec_unpack<su>_hi_<mode>): Use UXTL2 and SXTL2 assembler
16580 2018-06-01 Richard Sandiford <richard.sandiford@linaro.org>
16582 PR tree-optimization/85989
16583 * gimple-ssa-backprop.c (backprop::m_visited_phis): New member
16585 (backprop::intersect_uses): Check it when deciding whether this
16586 is a backedge reference.
16587 (backprop::process_block): Add each phi to m_visited_phis
16588 after visiting it, then clear it at the end.
16590 2018-06-01 Richard Biener <rguenther@suse.de>
16592 * tree-vectorizer.h (vect_dr_stmt): New function.
16593 (vect_get_load_cost): Adjust.
16594 (vect_get_store_cost): Likewise.
16595 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence):
16596 Use vect_dr_stmt instead of DR_SMTT.
16597 (vect_record_base_alignments): Likewise.
16598 (vect_calculate_target_alignment): Likewise.
16599 (vect_compute_data_ref_alignment): Likewise and make static.
16600 (vect_update_misalignment_for_peel): Likewise.
16601 (vect_verify_datarefs_alignment): Likewise.
16602 (vector_alignment_reachable_p): Likewise.
16603 (vect_get_data_access_cost): Likewise. Pass down
16604 vinfo to vect_get_load_cost/vect_get_store_cost instead of DR.
16605 (vect_get_peeling_costs_all_drs): Likewise.
16606 (vect_peeling_hash_get_lowest_cost): Likewise.
16607 (vect_enhance_data_refs_alignment): Likewise.
16608 (vect_find_same_alignment_drs): Likewise.
16609 (vect_analyze_data_refs_alignment): Likewise.
16610 (vect_analyze_group_access_1): Likewise.
16611 (vect_analyze_group_access): Likewise.
16612 (vect_analyze_data_ref_access): Likewise.
16613 (vect_analyze_data_ref_accesses): Likewise.
16614 (vect_vfa_segment_size): Likewise.
16615 (vect_small_gap_p): Likewise.
16616 (vectorizable_with_step_bound_p): Likewise.
16617 (vect_prune_runtime_alias_test_list): Likewise.
16618 (vect_analyze_data_refs): Likewise.
16619 (vect_supportable_dr_alignment): Likewise.
16620 * tree-vect-loop-manip.c (get_misalign_in_elems): Likewise.
16621 (vect_gen_prolog_loop_niters): Likewise.
16622 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
16623 * tree-vect-patterns.c (vect_recog_bool_pattern): Do not
16625 (vect_recog_mask_conversion_pattern): Likewise.
16626 (vect_try_gather_scatter_pattern): Likewise.
16627 * tree-vect-stmts.c (vect_model_store_cost): Pass stmt_info
16628 to vect_get_store_cost.
16629 (vect_get_store_cost): Get stmt_info instead of DR.
16630 (vect_model_load_cost): Pass stmt_info to vect_get_load_cost.
16631 (vect_get_load_cost): Get stmt_info instead of DR.
16633 2018-06-01 Richard Biener <rguenther@suse.de>
16635 PR middle-end/86017
16636 * gimple-fold.c (var_decl_component_p): Also allow offsetted
16637 vars wrapped in MEM_REFs.
16639 2018-06-01 Richard Sandiford <richard.sandiford@linaro.org>
16641 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
16642 Fix subreg tests so that we only return a choice between
16643 GENERAL_REGS and FP_REGS if the original classes included both.
16645 2018-06-01 Richard Biener <rguenther@suse.de>
16648 * tree-ssa-structalias.c (get_function_part_constraint):
16649 Handle NULL fi->decl.
16650 (find_func_aliases_for_call): Properly handle indirect
16651 fi from direct call.
16652 (find_func_clobbers): Likewise.
16653 (ipa_pta_execute): Likewise.
16654 (create_variable_info_for): For functions that are ifunc_resolver
16655 resolve to a varinfo that contains the result of the resolver call.
16656 (associate_varinfo_to_alias): Do not treat ifunc resolvers as
16659 2018-05-31 Michael Collison <michael.collison@arm.com>
16661 * config/aarch64/aarch64.md:
16662 (*fix_to_zero_extenddfdi2): New pattern.
16663 * gcc.target/aarch64/fix_extend1.c: New testcase.
16665 2018-05-31 Qing Zhao <qing.zhao@oracle.com>
16667 PR middle-end/78809
16668 PR middle-end/83026
16669 * builtins.c (expand_builtin): Add the handling of BUILT_IN_STRCMP_EQ
16670 and BUILT_IN_STRNCMP_EQ.
16671 * builtins.def: Add new builtins BUILT_IN_STRCMP_EQ and
16672 BUILT_IN_STRNCMP_EQ.
16673 * gimple-fold.c (gimple_fold_builtin_string_compare): Add the
16674 handling of BUILTIN_IN_STRCMP_EQ and BUILT_IN_STRNCMP_EQ.
16675 (gimple_fold_builtin): Likewise.
16676 * tree-ssa-strlen.c (compute_string_length): New function.
16677 (determine_min_obsize): New function.
16678 (handle_builtin_string_cmp): New function to handle calls to
16679 string compare functions.
16680 (strlen_optimize_stmt): Add handling to builtin string compare
16682 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
16683 Add the handling of BUILT_IN_STRCMP_EQ and BUILT_IN_STRNCMP_EQ.
16684 * tree.c (build_common_builtin_nodes): Add new defines of
16685 BUILT_IN_STRNCMP_EQ and BUILT_IN_STRCMP_EQ.
16687 2018-05-31 Jakub Jelinek <jakub@redhat.com>
16690 * bb-reorder.c (pass_partition_blocks::gate): Return false for
16691 functions with naked attribute.
16693 2018-05-31 Uros Bizjak <ubizjak@gmail.com>
16695 * config/i386/sse.md (avx_vec_concat<mode>):
16696 Substitute concat_tg_mode mode attribute with xtg_mode.
16697 (<mask_codefor>avx512dq_broadcast<mode><mask_name>_1): Ditto.
16698 (concat_tg_mode): Remove mode attribute.
16700 2018-05-31 Martin Sebor <msebor@redhat.com>
16703 * calls.c (alloc_max_size): Correct a logic error/typo.
16704 Treat excessive arguments as infinite. Warn for invalid arguments.
16705 * doc/invoke.texi (-Walloc-size-larger-than): Update.
16707 2018-05-31 H.J. Lu <hongjiu.lu@intel.com>
16710 * config/i386/x86-tune.def: Re-enable partial_reg_dependency
16711 and movx for Haswell.
16713 2018-05-31 Chung-Lin Tang <cltang@codesourcery.com>
16714 Cesar Philippidis <cesar@codesourcery.com>
16716 PR middle-end/85879
16717 * gimplify.c (gimplify_adjust_omp_clauses): Add 'remove = true'
16718 when emitting error on private/firstprivate reductions.
16719 * omp-low.c (lower_omp_target): Avoid reference-type processing
16720 on pointers for firstprivate clause.
16722 2018-05-31 Sameera Deshpande <sameera.deshpande@linaro.org>
16724 * config/aarch64/aarch64-simd-builtins.def (ld1x3): New.
16727 * config/aarch64/aarch64-simd.md
16728 (aarch64_ld1x3<VALLDIF:mode>): New pattern.
16729 (aarch64_ld1_x3_<mode>): Likewise
16730 (aarch64_st1x2<VALLDIF:mode>): Likewise
16731 (aarch64_st1_x2_<mode>): Likewise
16732 (aarch64_st1x3<VALLDIF:mode>): Likewise
16733 (aarch64_st1_x3_<mode>): Likewise
16734 * config/aarch64/arm_neon.h (vld1_u8_x3): New function.
16735 (vld1_s8_x3): Likewise.
16736 (vld1_u16_x3): Likewise.
16737 (vld1_s16_x3): Likewise.
16738 (vld1_u32_x3): Likewise.
16739 (vld1_s32_x3): Likewise.
16740 (vld1_u64_x3): Likewise.
16741 (vld1_s64_x3): Likewise.
16742 (vld1_f16_x3): Likewise.
16743 (vld1_f32_x3): Likewise.
16744 (vld1_f64_x3): Likewise.
16745 (vld1_p8_x3): Likewise.
16746 (vld1_p16_x3): Likewise.
16747 (vld1_p64_x3): Likewise.
16748 (vld1q_u8_x3): Likewise.
16749 (vld1q_s8_x3): Likewise.
16750 (vld1q_u16_x3): Likewise.
16751 (vld1q_s16_x3): Likewise.
16752 (vld1q_u32_x3): Likewise.
16753 (vld1q_s32_x3): Likewise.
16754 (vld1q_u64_x3): Likewise.
16755 (vld1q_s64_x3): Likewise.
16756 (vld1q_f16_x3): Likewise.
16757 (vld1q_f32_x3): Likewise.
16758 (vld1q_f64_x3): Likewise.
16759 (vld1q_p8_x3): Likewise.
16760 (vld1q_p16_x3): Likewise.
16761 (vld1q_p64_x3): Likewise.
16762 (vst1_s64_x2): Likewise.
16763 (vst1_u64_x2): Likewise.
16764 (vst1_f64_x2): Likewise.
16765 (vst1_s8_x2): Likewise.
16766 (vst1_p8_x2): Likewise.
16767 (vst1_s16_x2): Likewise.
16768 (vst1_p16_x2): Likewise.
16769 (vst1_s32_x2): Likewise.
16770 (vst1_u8_x2): Likewise.
16771 (vst1_u16_x2): Likewise.
16772 (vst1_u32_x2): Likewise.
16773 (vst1_f16_x2): Likewise.
16774 (vst1_f32_x2): Likewise.
16775 (vst1_p64_x2): Likewise.
16776 (vst1q_s8_x2): Likewise.
16777 (vst1q_p8_x2): Likewise.
16778 (vst1q_s16_x2): Likewise.
16779 (vst1q_p16_x2): Likewise.
16780 (vst1q_s32_x2): Likewise.
16781 (vst1q_s64_x2): Likewise.
16782 (vst1q_u8_x2): Likewise.
16783 (vst1q_u16_x2): Likewise.
16784 (vst1q_u32_x2): Likewise.
16785 (vst1q_u64_x2): Likewise.
16786 (vst1q_f16_x2): Likewise.
16787 (vst1q_f32_x2): Likewise.
16788 (vst1q_f64_x2): Likewise.
16789 (vst1q_p64_x2): Likewise.
16790 (vst1_s64_x3): Likewise.
16791 (vst1_u64_x3): Likewise.
16792 (vst1_f64_x3): Likewise.
16793 (vst1_s8_x3): Likewise.
16794 (vst1_p8_x3): Likewise.
16795 (vst1_s16_x3): Likewise.
16796 (vst1_p16_x3): Likewise.
16797 (vst1_s32_x3): Likewise.
16798 (vst1_u8_x3): Likewise.
16799 (vst1_u16_x3): Likewise.
16800 (vst1_u32_x3): Likewise.
16801 (vst1_f16_x3): Likewise.
16802 (vst1_f32_x3): Likewise.
16803 (vst1_p64_x3): Likewise.
16804 (vst1q_s8_x3): Likewise.
16805 (vst1q_p8_x3): Likewise.
16806 (vst1q_s16_x3): Likewise.
16807 (vst1q_p16_x3): Likewise.
16808 (vst1q_s32_x3): Likewise.
16809 (vst1q_s64_x3): Likewise.
16810 (vst1q_u8_x3): Likewise.
16811 (vst1q_u16_x3): Likewise.
16812 (vst1q_u32_x3): Likewise.
16813 (vst1q_u64_x3): Likewise.
16814 (vst1q_f16_x3): Likewise.
16815 (vst1q_f32_x3): Likewise.
16816 (vst1q_f64_x3): Likewise.
16817 (vst1q_p64_x3): Likewise.
16819 2018-05-30 Jozef Lawrynowicz <jozef.l@mittosystems.com>
16821 * config/msp430/msp430.c (msp430_output_labelref): Prepend
16822 user_label_prefix to name.
16824 * tree-core.h: Update comment about the format of NAME string
16825 passed to handler in attribute_spec.
16827 * config/msp430/msp430.md: Remove erroneous subreg expression from
16828 zero_extendqisi2 insn pattern. Remove msp430x ISA restriction on
16829 zero_extend{q,h}isi2.
16831 2018-05-30 Borislav Petkov <bp@suse.de>
16833 * doc/extend.texi: Document some architecture specific
16834 constraints and sort entries.
16836 2018-05-30 Martin Sebor <msebor@redhat.com>
16838 PR middle-end/85369
16839 * builtins.c (expand_builtin_stpcpy_1): New function.
16840 (expand_builtin_stpcpy): Call it, and call maybe_warn_nonstring_arg
16841 only if the former succeeds.
16843 2018-05-31 Sameera Deshpande <sameera.deshpande@linaro.org>
16845 * config/aarch64/aarch64-cores.def (saphira) : Add support for ARMv8.4
16848 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
16850 * doc/invoke.texi (-flinker-output): Document
16852 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
16854 * passes.c (ipa_write_summaries): Only modify statements if body
16856 * cgraphunit.c (ipa_passes): Also produce intermeidate code when
16857 incrementally linking.
16858 (ipa_passes): Likewise.
16859 * lto-cgraph.c (lto_output_node): When incrementally linking do not
16860 pass down resolution info.
16861 * common.opt (flag_incremental_link): Update info.
16862 * gcc.c (plugin specs): Turn flinker-output=* to
16863 -plugin-opt=-linker-output-known
16864 * toplev.c (compile_file): Also cut compilation when doing incremental
16866 * flag-types. (enum lto_partition_model): Add
16867 LTO_LINKER_OUTPUT_NOLTOREL.
16868 (invoke.texi): Add -flinker-output docs.
16869 * ipa.c (symbol_table::remove_unreachable_nodes): Handle LTO incremental
16870 link same way as WPA; do not stream in dead initializers.
16872 * dwarf2out.c (dwarf2out_die_ref_for_decl,
16873 darf2out_register_external_decl): Support incremental link.
16875 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
16877 * lto-opts.c (lto_write_options): Skip OPT_dumpdir, OPT_fresolution_.
16879 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
16881 * lto-wrapper.c (debug_objcopy): Add rename parameter; pass
16882 it down to simple_object_copy_lto_debug_sections.
16883 (run_gcc): Determine incremental LTO link time and configure
16884 lto1 into non-wpa mode, disable renaming of debug sections.
16886 2018-05-30 Kelvin Nilsen <kelvin@gcc.gnu.org>
16888 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Remove
16889 descriptions of various incorrectly documented functions.
16891 2018-05-30 Andre Vieira <andre.simoesdiasvieira@arm.com>
16894 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Make
16895 address check not strict.
16897 2018-05-30 Richard Biener <rguenther@suse.de>
16899 PR tree-optimization/85964
16900 * tracer.c (better_p): Drop initialized count check, we only
16901 call the function with initialized counts now.
16902 (find_best_successor): Do find a best edge if one
16903 has uninitialized count.
16904 (find_best_predecessor): Likewise. Do BB frequency check only
16905 if count is initialized.
16907 2017-05-30 Jackson Woodruff <jackson.woodruff@arm.com>
16909 * config/aarch64/aarch64.c (aarch64_host_wide_int_compare): New.
16910 (aarch64_ldrstr_offset_compare): New.
16911 (aarch64_operands_adjust_ok_for_ldpstp): Update to consider all
16912 load/store orderings.
16913 (aarch64_gen_adjusted_ldpstp): Likewise.
16915 2018-05-30 Wilco Dijkstra <wdijkstr@arm.com>
16917 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
16918 Check for subset of GENERAL_REGS and FP_REGS.
16919 * config/aarch64/aarch64-simd.md (aarch64_get_lane): Increase cost of
16922 2018-05-30 Richard Sandiford <richard.sandiford@linaro.org>
16924 * alias.c (adjust_offset_for_component_ref): Use poly_int_tree_p
16925 and wi::to_poly_offset. Add the current offset and then check
16926 whether the sum fits, rather than using an unchecked addition of
16927 a checked term. Check for a shwi rather than a uhwi.
16928 * expr.c (get_bit_range): Use tree_to_poly_uint64.
16929 (store_constructor): Use poly_int_tree_p.
16930 (expand_expr_real_1): Likewise.
16931 * function.c (assign_temp): Likewise.
16932 * fold-const.c (const_binop): Use poly_int_tree_p and
16933 wi::to_poly_offset.
16934 (fold_indirect_ref_1): Likewise. Use multiple_p to attempt an exact
16936 * ipa-icf-gimple.c (func_checker::compare_operand): Use
16937 to_poly_offset for MEM offsets.
16938 * ipa-icf.c (sem_variable::equals): Likewise.
16939 * stor-layout.c (compute_record_mode): Use poly_int_tree_p.
16940 * tree-ssa-sccvn.c (ao_ref_init_from_vn_reference): Use
16941 wi::to_poly_offset for BIT_FIELD_REF offsets.
16942 (vn_reference_maybe_forwprop_address): Use poly_int_tree_p and
16943 wi::to_poly_offset.
16944 * var-tracking.c (emit_note_insn_var_location): Use
16945 tree_to_poly_uint64.
16947 2018-05-29 Jim Wilson <jimw@sifive.com>
16949 * config/riscv/riscv.c (riscv_interrupt_type): Fix comment typo.
16951 2018-05-29 Uros Bizjak <ubizjak@gmail.com>
16954 * config/i386/i386.md (l<rounding_insn><MODEF:mode><SWI48:mode>2):
16955 Enable for TARGET_SSE4_1 and generate rounds{s,d} and cvtts{s,d}2si{,q}
16957 (sse4_1_round<mode>2): Use nonimmediate_operand
16958 for operand 1 predicate.
16960 2018-05-29 Martin Sebor <msebor@redhat.com>
16961 Richard Biener <rguenther@suse.de>
16964 * calls.c (get_size_range): Call determine_value_range instead
16965 of get_value_range..
16966 * tree-vrp.h (determine_value_range): Declared new function.
16967 * tree-vrp.c (determine_value_range_1, determine_value_range): New.
16969 2018-05-29 Richard Biener <rguenther@suse.de>
16971 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Make
16972 sure to use non-pattern stmts for get_earlier_stmt arguments.
16973 * tree-vectorizer.h (get_earlier_stmt): Assert we do not get
16974 called on pattern stmts.
16975 (get_later_stmt): Likewise.
16977 2018-05-29 Martin Liska <mliska@suse.cz>
16979 PR gcov-profile/85759
16980 * doc/gcov.texi: Document GCOV_ERROR_FILE and GCOV_EXIT_AT_ERROR
16983 2018-05-29 Jakub Jelinek <jakub@redhat.com>
16985 * tree-cfg.c (verify_gimple_assign_unary): Add checking for
16987 (verify_gimple_assign_binary): Check TYPE_VECTOR_SUBPARTS for
16991 * tree.def (VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR,
16992 VEC_PACK_FLOAT_EXPR): New tree codes.
16993 * tree-pretty-print.c (op_code_prio): Handle
16994 VEC_UNPACK_FIX_TRUNC_HI_EXPR and VEC_UNPACK_FIX_TRUNC_LO_EXPR.
16995 (dump_generic_node): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR,
16996 VEC_UNPACK_FIX_TRUNC_LO_EXPR and VEC_PACK_FLOAT_EXPR.
16997 * tree-inline.c (estimate_operator_cost): Likewise.
16998 * gimple-pretty-print.c (dump_binary_rhs): Handle VEC_PACK_FLOAT_EXPR.
16999 * fold-const.c (const_binop): Likewise.
17000 (const_unop): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR and
17001 VEC_UNPACK_FIX_TRUNC_LO_EXPR.
17002 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
17003 (verify_gimple_assign_binary): Handle VEC_PACK_FLOAT_EXPR.
17004 * cfgexpand.c (expand_debug_expr): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR,
17005 VEC_UNPACK_FIX_TRUNC_LO_EXPR and VEC_PACK_FLOAT_EXPR.
17006 * expr.c (expand_expr_real_2): Likewise.
17007 * optabs.def (vec_packs_float_optab, vec_packu_float_optab,
17008 vec_unpack_sfix_trunc_hi_optab, vec_unpack_sfix_trunc_lo_optab,
17009 vec_unpack_ufix_trunc_hi_optab, vec_unpack_ufix_trunc_lo_optab): New
17011 * optabs.c (expand_widen_pattern_expr): For
17012 VEC_UNPACK_FIX_TRUNC_HI_EXPR and VEC_UNPACK_FIX_TRUNC_LO_EXPR use
17013 sign from result type rather than operand's type.
17014 (expand_binop_directly): For vec_packu_float_optab and
17015 vec_packs_float_optab allow result type to be different from operand's
17017 * optabs-tree.c (optab_for_tree_code): Handle
17018 VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR and
17019 VEC_PACK_FLOAT_EXPR. Formatting fixes.
17020 * tree-vect-generic.c (expand_vector_operations_1): Handle
17021 VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR and
17022 VEC_PACK_FLOAT_EXPR.
17023 * tree-vect-stmts.c (supportable_widening_operation): Handle
17025 (supportable_narrowing_operation): Handle FLOAT_EXPR.
17026 * config/i386/i386.md (fixprefix, floatprefix): New code attributes.
17027 * config/i386/sse.md (*float<floatunssuffix>v2div2sf2): Rename to ...
17028 (float<floatunssuffix>v2div2sf2): ... this. Formatting fix.
17029 (vpckfloat_concat_mode, vpckfloat_temp_mode, vpckfloat_op_mode): New
17031 (vec_pack<floatprefix>_float_<mode>): New expander.
17032 (vunpckfixt_mode, vunpckfixt_model, vunpckfixt_extract_mode): New mode
17034 (vec_unpack_<fixprefix>fix_trunc_lo_<mode>,
17035 vec_unpack_<fixprefix>fix_trunc_hi_<mode>): New expanders.
17036 * doc/md.texi (vec_packs_float_@var{m}, vec_packu_float_@var{m},
17037 vec_unpack_sfix_trunc_hi_@var{m}, vec_unpack_sfix_trunc_lo_@var{m},
17038 vec_unpack_ufix_trunc_hi_@var{m}, vec_unpack_ufix_trunc_lo_@var{m}):
17040 * doc/generic.texi (VEC_UNPACK_FLOAT_HI_EXPR,
17041 VEC_UNPACK_FLOAT_LO_EXPR): Fix pasto in description.
17042 (VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR,
17043 VEC_PACK_FLOAT_EXPR): Document.
17045 2018-05-29 Richard Biener <rguenther@suse.de>
17047 * tree-vectorizer.h (struct vec_info): Add stmt_vec_infos
17049 (stmt_vec_info_vec): Make pointer.
17050 (init_stmt_vec_info_vec): Remove.
17051 (free_stmt_vec_info_vec): Likewise.
17052 (set_stmt_vec_info_vec): New function.
17053 (free_stmt_vec_infos): Likewise.
17054 (vinfo_for_stmt): Adjust for stmt_vec_info_vec indirection.
17055 (set_vinfo_for_stmt): Likewise.
17056 (get_earlier_stmt): Likewise.
17057 (get_later_stmt): Likewise.
17058 * tree-vectorizer.c (stmt_vec_info_vec): Make pointer.
17059 (vec_info::vec_info): Allocate stmt_vec_infos and set the global.
17060 (vec_info::~vec_info): Free stmt_vec_infos.
17061 (vectorize_loops): Set the global stmt_vec_info_vec to NULL.
17062 Remove old init_stmt_vec_info_vec/free_stmt_vec_info_vec calls.
17063 (pass_slp_vectorize::execute): Likewise.
17064 * tree-vect-stmts.c (init_stmt_vec_info_vec): Remove.
17065 (free_stmt_vec_info_vec): Likewise.
17066 (set_stmt_vec_info_vec): New function.
17067 (free_stmt_vec_infos): Likewise.
17068 * tree-vect-loop.c (_loop_vec_info::~_loop_vec_info): Set
17069 the global stmt_vec_info_vec.
17070 * tree-parloops.c (gather_scalar_reductions): Use
17071 set_stmt_vec_info_vec/free_stmt_vec_infos and maintain a local
17074 2018-05-29 Richard Biener <rguenther@suse.de>
17076 * dominance.c (iterate_fix_dominators): Push/pop TV_DOMINANCE.
17078 2018-05-29 Martin Liska <mliska@suse.cz>
17079 David Malcolm <dmalcolm@redhat.com>
17081 * vec.c (test_reverse): New.
17082 (vec_c_tests): Add new test.
17083 * vec.h (vl_ptr>::reverse): New function.
17085 2018-05-29 Gerald Pfeifer <gerald@pfeifer.com>
17087 * config.gcc: Identify FreeBSD 3.x and 4.x as unsupported.
17089 * config/freebsd-spec.h (FBSD_LIB_SPEC): Only consider FreeBSD 5
17092 2018-05-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
17094 * tree-dump.c (dump_node): Use splay_tree_delete_pointers.
17096 2018-05-28 Richard Biener <rguenther@suse.de>
17098 PR tree-optimization/85933
17099 * tree-vect-data-refs.c (vect_record_base_alignments): Only
17100 look at stmts marked as vectorizable.
17102 2018-05-28 Richard Biener <rguenther@suse.de>
17104 PR tree-optimization/85934
17105 * tree-vect-generic.c (expand_vector_operations_1): Hoist
17106 vector boolean check before scalar optimization.
17108 2018-05-28 Jakub Jelinek <jakub@redhat.com>
17110 * doc/invoke.texi (ARM Options): Use @item instead of @itemx
17113 2018-05-28 Mark Wielaard <mark@klomp.org>
17115 * dwarf2asm.c (dw2_asm_output_delta_uleb128): Add brackets around lab2
17116 if it is an expression containing a minus sign.
17118 2018-05-27 John David Anglin <danglin@gcc.gnu.org>
17120 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Define to 0.
17122 2018-05-27 Paul Koning <ni1d@arrl.net>
17124 * config/pdp11/pdp11.md (truncsihi2): Remove.
17126 2018-05-27 Monk Chiang <sh.chiang04@gmail.com>
17127 Chung-Ju Wu <jasonwucj@gmail.com>
17129 * config/nds32/nds32-intrinsic.md (unaligned_storedi): Modify patterns
17131 (unaligned_store_dw): Ditto.
17132 * config/nds32/nds32-memory-manipulation.c
17133 (nds32_expand_movmemsi_loop_known_size): Refactoring implementation.
17134 (nds32_gen_dup_4_byte_to_word_value): Rename to ...
17135 (nds32_gen_dup_4_byte_to_word_value_aux): ... this.
17136 (emit_setmem_word_loop): Rename to ...
17137 (emit_setmem_doubleword_loop): ... this.
17138 (nds32_gen_dup_4_byte_to_word_value): New function.
17139 (nds32_gen_dup_8_byte_to_double_word_value): New function.
17140 (nds32_expand_setmem_loop): Refine implementation.
17141 (nds32_expand_setmem_loop_v3m): Ditto.
17142 * config/nds32/nds32-multiple.md (unaligned_store_update_base_dw): New
17145 2018-05-27 Chung-Ju Wu <jasonwucj@gmail.com>
17147 * config/nds32/nds32.md (bswapsi2, bswaphi2): New patterns.
17149 2018-05-27 Chung-Ju Wu <jasonwucj@gmail.com>
17151 * config/nds32/nds32.c (nds32_attribute_table): Add "no_prologue".
17152 (nds32_init_machine_status): Initialize machine->attr_naked_p and
17153 machine->attr_no_prologue_p.
17154 (nds32_compute_stack_frame): Check "naked" and "no_prologue" attributes.
17155 (nds32_naked_function_p): Handle "naked" and "no_prologue" attributes.
17156 (nds32_expand_epilogue): Consider attr_naked_p.
17157 (nds32_expand_epilogue_v3pop): Likewise.
17158 (nds32_can_use_return_insn): Likewise.
17159 * config/nds32/nds32.h (machine_function): Add attr_naked_p and
17160 attr_no_prologue_p fields.
17161 * config/nds32/nds32.opt (mret-in-naked-func): New option.
17163 2018-05-27 Jakub Jelinek <jakub@redhat.com>
17166 * config/i386/i386.md (fixunssuffix, floatunssuffix): New code
17168 * config/i386/sse.md
17169 (<floatsuffix>float<sseintvecmodelower><mode>2<mask_name><round_name>):
17171 (float<floatunssuffix><sseintvecmodelower><mode>2<mask_name><round_name>):
17173 (<floatsuffix>float<sselongvecmodelower><mode>2<mask_name><round_name>):
17175 (float<floatunssuffix><sselongvecmodelower><mode>2<mask_name><round_name>):
17177 (*<floatsuffix>floatv2div2sf2): Rename to ...
17178 (*float<floatunssuffix>v2div2sf2): ... this.
17179 (<floatsuffix>floatv2div2sf2_mask): Rename to ...
17180 (float<floatunssuffix>v2div2sf2_mask): ... this.
17181 (*<floatsuffix>floatv2div2sf2_mask_1): Rename to ...
17182 (*float<floatunssuffix>v2div2sf2_mask_1): ... this.
17183 (<fixsuffix>fix_truncv8dfv8si2<mask_name><round_saeonly_name>): Rename
17185 (fix<fixunssuffix>_truncv8dfv8si2<mask_name><round_saeonly_name>):
17187 (<fixsuffix>fix_trunc<mode><sseintvecmodelower>2<mask_name><round_saeonly_name>):
17189 (fix<fixunssuffix>_trunc<mode><sseintvecmodelower>2<mask_name><round_saeonly_name>):
17191 (<fixsuffix>fix_trunc<mode><sselongvecmodelower>2<mask_name><round_saeonly_name>):
17193 (fix<fixunssuffix>_trunc<mode><sselongvecmodelower>2<mask_name><round_saeonly_name>):
17195 (<fixsuffix>fix_truncv2sfv2di2<mask_name>): Rename to ...
17196 (fix<fixunssuffix>_truncv2sfv2di2<mask_name>): ... this.
17197 (vec_pack_ufix_trunc_<mode>): Use gen_fixuns_truncv8dfv8si2 instead of
17198 gen_ufix_truncv8dfv8si2.
17199 * config/i386/i386-builtin.def (__builtin_ia32_cvttpd2uqq256_mask,
17200 __builtin_ia32_cvttpd2uqq128_mask, __builtin_ia32_cvttps2uqq256_mask,
17201 __builtin_ia32_cvttps2uqq128_mask, __builtin_ia32_cvtuqq2ps256_mask,
17202 __builtin_ia32_cvtuqq2ps128_mask, __builtin_ia32_cvtuqq2pd256_mask,
17203 __builtin_ia32_cvtuqq2pd128_mask, __builtin_ia32_cvttpd2udq512_mask,
17204 __builtin_ia32_cvtuqq2ps512_mask, __builtin_ia32_cvtuqq2pd512_mask,
17205 __builtin_ia32_cvttps2uqq512_mask, __builtin_ia32_cvttpd2uqq512_mask):
17206 Use fixuns instead ufix or floatuns instead ufloat in CODE_FOR_ names.
17208 2018-05-24 H.J. Lu <hongjiu.lu@intel.com>
17212 * varasm.c (assemble_alias): Lookup ifunc attribute on error.
17214 2018-05-25 Jim Wilson <jimw@sifive.com>
17216 * config/riscv/riscv-protos.h (riscv_epilogue_uses): New.
17217 * config/riscv/riscv.c (struct machine_function): Add
17218 interrupt_handler_p and attribute_checked_p fields.
17219 (riscv_attribute_table): Add interrupt.
17220 (riscv_interrupt_type_p): New.
17221 (riscv_save_reg_p): Save extra regs for interrupt handler.
17222 (riscv_use_save_libcall): Return false for interrupt handler.
17223 (riscv_first_stack_step): Add forward declaration.
17224 (riscv_compute_frame_info): New local interrupt_save_t1. Set it
17225 for interrupt handler with large frame. Use it for saved reg list.
17226 (riscv_expand_prologue): Move flag_stack_usage_info support to
17227 eliminate duplication.
17228 (riscv_expand_epilogue): Generate mret for interrupt handler.
17229 (riscv_epilogue_uses): New.
17230 (riscv_can_use_return_insn): Return false for interrupt handler.
17231 (riscv_function_ok_for_sibcall): Likewise.
17232 (riscv_set_current_function): Add interrupt handler support.
17233 * config/riscv/riscv.h (EPILOGUE_USES): Call riscv_epilogue_uses.
17234 * config/riscv/riscv.md (UNSPECV_MRET): New.
17236 (riscv_frflags, riscv_fsflags): Use tab after opcode.
17238 * doc/extend.texi (RISC-V Function Attributes) <interrupt>: New.
17240 2018-05-25 Bill Schmidt <wschmidt@linux.ibm.com>
17242 PR tree-optimization/85712
17243 * gimple-ssa-strength-reduction.c (replace_one_candidate): Skip if
17244 this candidate has already been replaced in-situ by a copy.
17246 2018-05-25 Jason Merrill <jason@redhat.com>
17248 PR c++/80485 - inline function non-zero address.
17249 * symtab.c (nonzero_address): Check DECL_COMDAT.
17251 2018-05-25 Uros Bizjak <ubizjak@gmail.com>
17254 * config/alpha/alpha.md (ashlsi3): New insn pattern.
17255 (*ashlsi_se): Rename from *ashldi_se. Define as sign
17256 extension of SImode operation. Use const123_operand predicate.
17257 (*saddsi_1): Remove.
17258 (*saddl_se_1): Ditto.
17259 (*ssubsi_1): Ditto.
17260 (*ssubl_se_1): Ditto.
17261 * config/alpha/predicates.md (const123_operand): New predicate.
17262 * config/alpha/constraints.md (P): Use IN_RANGE.
17264 2018-05-25 Richard Biener <rguenther@suse.de>
17266 * tree-ssa-alias.h (refs_may_alias_p): Add tbaa_p bool parameter,
17268 (ref_maybe_used_by_stmt_p): Likewise.
17269 (stmt_may_clobber_ref_p): Likewise.
17270 (stmt_may_clobber_ref_p_1): Likewise.
17271 * tree-ssa-alias.c (refs_may_alias_p): Add tbaa_p bool parameter
17273 (ref_maybe_used_by_stmt_p): Likewise.
17274 (stmt_may_clobber_ref_p): Likewise.
17275 (stmt_may_clobber_ref_p_1): Likewise.
17276 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences): Use
17277 the alias oracle to disambiguate DRs with stmts DR analysis
17279 (vect_analyze_data_refs): Do not give up on not analyzable
17280 DRs for BB vectorization. Remove code truncating the dataref
17283 2018-05-25 Jakub Jelinek <jakub@redhat.com>
17286 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>_1):
17287 Add (=Yk,v,C) variant using vptestm insn. Use TARGET_AVX512BW
17288 in test instead of TARGET_AVX512F for VI12_AVX512VL iterator.
17290 2018-05-25 Richard Biener <rguenther@suse.de>
17292 * tree-vect-data-refs.c (vect_find_stmt_data_reference): New
17293 function, combining stmt data ref gathering and fatal analysis
17295 (vect_analyze_data_refs): Remove now redudnant code and simplify.
17296 * tree-vect-loop.c (vect_get_datarefs_in_loop): Factor out from
17297 vect_analyze_loop_2 and use vect_find_stmt_data_reference.
17298 * tree-vect-slp.c (vect_slp_bb): Use vect_find_stmt_data_reference.
17299 * tree-vectorizer.h (vect_find_stmt_data_reference): Declare.
17301 2018-05-25 Bin Cheng <bin.cheng@arm.com>
17303 PR tree-optimization/85720
17304 * tree-loop-distribution.c (break_alias_scc_partitions): Don't merge
17305 SCC if all partitions are builtins.
17306 (version_loop_by_alias_check): New parameter. Generate cancelable
17307 runtime alias check if all partitions are builtins.
17308 (distribute_loop): Update call to above function.
17310 2018-05-25 Bin Cheng <bin.cheng@arm.com>
17312 * tree-outof-ssa.c (tree-ssa.h, tree-dfa.h): Include header files.
17313 (create_default_def, for_all_parms): Moved from tree-ssa-coalesce.c.
17314 (parm_default_def_partition_arg): Ditto.
17315 (set_parm_default_def_partition): Ditto.
17316 (get_parm_default_def_partitions): Ditto and make it static.
17317 (get_undefined_value_partitions): Ditto and make it static.
17318 (remove_ssa_form): Refactor call to init_var_map here.
17319 * tree-ssa-coalesce.c (build_ssa_conflict_graph): Support live range
17320 computation for loop region.
17321 (coalesce_partitions, compute_optimized_partition_bases): Ditto.
17322 (register_default_def): Delete.
17323 (for_all_parms, create_default_def): Move to tree-outof-ssa.c.
17324 (parm_default_def_partition_arg): Ditto.
17325 (set_parm_default_def_partition): Ditto.
17326 (get_parm_default_def_partitions): Ditto and make it static.
17327 (get_undefined_value_partitions): Ditto and make it static.
17328 (coalesce_with_default, coalesce_with_default): Update comment.
17329 (create_coalesce_list_for_region): New func factored out from
17330 create_outofssa_var_map.
17331 (populate_coalesce_list_for_outofssa): New func factored out from
17332 create_outofssa_var_map and coalesce_ssa_name.
17333 (create_outofssa_var_map): Delete.
17334 (coalesce_ssa_name): Refactor to support live range computation.
17335 * tree-ssa-coalesce.h (coalesce_ssa_name): Change decl.
17336 (get_parm_default_def_partitions): Delete.
17337 (get_undefined_value_partitions): Ditto.
17338 * tree-ssa-live.c (init_var_map, delete_var_map): Support live range
17339 computation for loop region.
17340 (new_tree_live_info, loe_visit_block): Ditto.
17341 (live_worklist, set_var_live_on_entry): Ditto.
17342 (calculate_live_on_exit, verify_live_on_entry): Ditto.
17343 * tree-ssa-live.h (struct _var_map): New fields.
17344 (init_var_map): Change decl.
17345 (region_contains_p): New.
17347 2018-05-25 Bin Cheng <bin.cheng@arm.com>
17349 * tree-ssa-live.h (live_merge_and_clear): Delete.
17351 2018-05-25 Richard Biener <rguenther@suse.de>
17354 * tree-dump.c (dequeue_and_dump): Remove access to removed
17355 operand 2 of a SWITCH_EXPR.
17357 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
17359 * doc/sourcebuild.texi (vect_double_cond_arith): Include
17360 multiplication and division.
17361 * doc/md.texi (cond_mul@var{m}, cond_div@var{m}, cond_mod@var{m})
17362 (cond_udiv@var{m}, cond_umod@var{m}): Document.
17363 * optabs.def (cond_smul_optab, cond_sdiv_optab, cond_smod_optab)
17364 (cond_udiv_optab, cond_umod_optab): New optabs.
17365 * internal-fn.def (IFN_COND_MUL, IFN_COND_DIV, IFN_COND_MOD)
17366 (IFN_COND_RDIV): New internal functions.
17367 * internal-fn.c (get_conditional_internal_fn): Handle TRUNC_DIV_EXPR,
17368 TRUNC_MOD_EXPR and RDIV_EXPR.
17369 * match.pd (UNCOND_BINARY, COND_BINARY): Handle them.
17370 * config/aarch64/iterators.md (UNSPEC_COND_MUL, UNSPEC_COND_DIV):
17372 (SVE_INT_BINARY): Include mult.
17373 (SVE_COND_FP_BINARY): Include UNSPEC_MUL and UNSPEC_DIV.
17374 (optab, sve_int_op): Handle mult.
17375 (optab, sve_fp_op, commutative): Handle UNSPEC_COND_MUL and
17377 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): New pattern
17378 for SVE_INT_BINARY_SD.
17380 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
17382 * config/aarch64/iterators.md (SVE_INT_BINARY_SD): New code iterator.
17383 (optab, sve_int_op): Handle div and udiv.
17384 * config/aarch64/aarch64-sve.md (<optab><mode>3): New expander
17385 for SVE_INT_BINARY_SD.
17386 (*<optab><mode>3): New insn for the same.
17388 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
17390 * tree-vect-patterns.c: Include predict.h.
17391 (vect_recog_divmod_pattern): Restrict check for division support
17392 to when optimizing for size.
17394 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
17396 * doc/sourcebuild.texi (vect_double_cond_arith: Document.
17397 * gimple-match.h (gimple_match_op::MAX_NUM_OPS): Bump to 4.
17398 (gimple_match_op::gimple_match_op): Add an overload for 4 operands.
17399 (gimple_match_op::set_op): Likewise.
17400 (gimple_resimplify4): Declare.
17401 * genmatch.c (get_operand_type): Handle CFN_COND_* functions.
17402 (expr::gen_transform): Likewise.
17403 (decision_tree::gen): Generate a simplification routine for 4 operands.
17404 * gimple-match-head.c (gimple_simplify): Add an overload for
17405 4 operands. In the top-level function, handle up to 4 call
17406 arguments and call gimple_resimplify4.
17407 (gimple_resimplify4): New function.
17408 (build_call_internal): Pass a fourth operand.
17409 (maybe_push_to_seq): Likewise.
17410 * match.pd (UNCOND_BINARY, COND_BINARY): New operator lists.
17411 Fold VEC_COND_EXPRs of an operation and a default value into
17412 an IFN_COND_* function if possible.
17413 * config/aarch64/iterators.md (UNSPEC_COND_MAX, UNSPEC_COND_MIN):
17415 (SVE_COND_FP_BINARY): Include them.
17416 (optab, sve_fp_op): Handle them.
17417 (SVE_INT_BINARY_REV): New code iterator.
17418 (SVE_COND_FP_BINARY_REV): New int iterator.
17419 (commutative): New int attribute.
17420 * config/aarch64/aarch64-protos.h (aarch64_sve_prepare_conditional_op):
17422 * config/aarch64/aarch64.c (aarch64_sve_prepare_conditional_op): New
17424 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): Use it.
17425 (*cond_<optab><mode>): New patterns for reversed operands.
17427 2018-05-25 Richard Biener <rguenther@suse.de>
17429 * tree-vectorizer.h (STMT_VINFO_GROUP_*, GROUP_*): Remove.
17430 (DR_GROUP_*): New, assert we have non-NULL ->data_ref_info.
17431 (REDUC_GROUP_*): New, assert we have NULL ->data_ref_info.
17432 (STMT_VINFO_GROUPED_ACCESS): Adjust.
17433 * tree-vect-data-refs.c (everywhere): Adjust users.
17434 * tree-vect-loop.c (everywhere): Likewise.
17435 * tree-vect-slp.c (everywhere): Likewise.
17436 * tree-vect-stmts.c (everywhere): Likewise.
17437 * tree-vect-patterns.c (vect_reassociating_reduction_p): Likewise.
17439 2018-05-25 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
17441 * configure.ac (gcc_cv_as_section_has_e): Move to common section.
17443 (gcc_cv_as_section_exclude): ... this.
17444 Try Solaris as #exclude syntax.
17445 * configure: Regenerate.
17446 * config.in: Regenerate.
17447 * config/i386/i386.c (i386_solaris_elf_named_section): Handle
17449 * config/sparc/sparc.c (sparc_solaris_elf_asm_named_section)
17450 [HAVE_GAS_SECTION_EXCLUDE]: Handle SECTION_EXCLUDE.
17452 * varasm.c (default_elf_asm_named_section): Don't check if
17453 HAVE_GAS_SECTION_EXCLUDE is defined.
17455 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
17457 * doc/md.texi: Update the documentation of the cond_* optabs
17458 to mention the new final operand. Fix GET_MODE_NUNITS call.
17459 Describe the scalar case too.
17460 * internal-fn.def (IFN_EXTRACT_LAST): Change type to fold_left.
17461 * internal-fn.c (expand_cond_unary_optab_fn): Expect 3 operands
17463 (expand_cond_binary_optab_fn): Expect 4 operands instead of 3.
17464 (get_conditional_internal_fn): Update comment.
17465 * tree-vect-loop.c (vectorizable_reduction): Pass the original
17466 accumulator value as a final argument to conditional functions.
17467 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): Turn into
17468 a define_expand and add an "else" operand. Assert for now that
17469 the else operand is equal to operand 2. Use SVE_INT_BINARY and
17470 SVE_COND_FP_BINARY instead of SVE_COND_INT_OP and SVE_COND_FP_OP.
17471 (*cond_<optab><mode>): New patterns.
17472 * config/aarch64/iterators.md (UNSPEC_COND_SMAX, UNSPEC_COND_UMAX)
17473 (UNSPEC_COND_SMIN, UNSPEC_COND_UMIN, UNSPEC_COND_AND, UNSPEC_COND_ORR)
17474 (UNSPEC_COND_EOR): Delete.
17475 (optab): Remove associated mappings.
17476 (SVE_INT_BINARY): New code iterator.
17477 (sve_int_op): Remove int attribute and add "minus" to the code
17479 (SVE_COND_INT_OP): Delete.
17480 (SVE_COND_FP_OP): Rename to...
17481 (SVE_COND_FP_BINARY): ...this.
17483 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
17485 * optabs.c (can_reuse_operands_p): New function.
17486 (maybe_legitimize_operands): Try to reuse the results for
17489 2018-05-24 Uros Bizjak <ubizjak@gmail.com>
17491 * config/i386/sse.md (cvtusi2<ssescalarmodesuffix>64<round_name>):
17492 Add {q} suffix to insn mnemonic.
17494 2018-05-23 Jozef Lawrynowicz <jozef.l@mittosystems.com>
17496 * config/msp430/msp430.c (TARGET_WARN_FUNC_RETURN): Define.
17497 (msp430_warn_func_return): New.
17499 2018-05-24 Roger Sayle <roger@nextmovesoftware.com>
17501 * fold-const.c (tree_nonzero_bits): New function.
17502 * fold-const.h (tree_nonzero_bits): Likewise.
17503 * match.pd (POPCOUNT): New patterns to fold BUILTIN_POPCOUNT and
17504 friends. POPCOUNT(x&1) => x&1, POPCOUNT(x)==0 => x==0, etc.
17506 2018-05-24 H.J. Lu <hongjiu.lu@intel.com>
17510 * varasm.c (assemble_alias): Check ifunc_resolver only on
17513 2018-05-24 Uros Bizjak <ubizjak@gmail.com>
17516 * config/i386/sse.md (movdi_to_sse): Do not generate pseudo
17517 when memory input operand is handled.
17519 2018-05-24 Luis Machado <luis.machado@linaro.org>
17521 * config/aarch64/aarch64.c (qdf24xx_addrcost_table): New static
17523 (qdf24xx_tunings) <addr_costs>: Set to qdf24xx_addrcost_table.
17525 2018-05-24 Richard Sandiford <richard.sandiford@linaro.org>
17527 * match.pd: Delay FMA folds until after vectorization.
17529 2018-05-24 Andre Vieira <andre.simoesdiasvieira@arm.com>
17532 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Make
17533 address check not strict.
17535 2018-05-24 Richard Sandiford <richard.sandiford@linaro.org>
17537 * gimple-match.h (gimple_match_op): New class.
17538 (mprts_hook): Replace parameters with a gimple_match_op *.
17539 (maybe_build_generic_op): Likewise.
17540 (gimple_simplified_result_is_gimple_val): Replace parameters with
17541 a const gimple_match_op *.
17542 (gimple_simplify): Replace code_helper * and tree * parameters with
17543 a gimple_match_op * parameter.
17544 (gimple_resimplify1): Replace code_helper *, tree and tree *
17545 parameters with a gimple_match_op * parameter.
17546 (gimple_resimplify2): Likewise.
17547 (gimple_resimplify3): Likewise.
17548 (maybe_push_res_to_seq): Replace code_helper, tree and tree *
17549 parameters with a gimple_match_op * parameter.
17550 * gimple-match-head.c (gimple_simplify): Change prototypes of
17551 auto-generated functions to take a gimple_match_op * instead of
17552 separate code_helper * and tree * parameters. Make the same
17553 change in the top-level overload and update calls to the
17554 gimple_resimplify routines. Update calls to the auto-generated
17555 functions and to maybe_push_res_to_seq in the publicly-facing
17556 operation-specific gimple_simplify overloads.
17557 (gimple_match_op::MAX_NUM_OPS): Define.
17558 (gimple_resimplify1): Replace rcode and ops with a single res_op
17559 parameter. Update call to gimple_simplify.
17560 (gimple_resimplify2): Likewise.
17561 (gimple_resimplify3): Likewise.
17562 (mprts_hook): Replace parameters with a gimple_match_op *.
17563 (maybe_build_generic_op): Likewise.
17564 (build_call_internal): Replace type, nargs and ops with
17565 a gimple_match_op *.
17566 (maybe_push_res_to_seq): Replace res_code, type and ops parameters
17567 with a single gimple_match_op *. Update calls to mprts_hook,
17568 build_call_internal and gimple_simplified_result_is_gimple_val.
17569 Factor out code that is common to the tree_code and combined_fn cases.
17570 * genmatch.c (expr::gen_transform): Replace tem_code and
17571 tem_ops with a gimple_match_op called tem_op. Update calls
17572 to the gimple_resimplify functions and maybe_push_res_to_seq.
17573 (dt_simplify::gen_1): Manipulate res_op instead of res_code and
17574 res_ops. Update call to the gimple_resimplify functions.
17575 (dt_simplify::gen): Pass res_op instead of res_code and res_ops.
17576 (decision_tree::gen): Make the functions take a gimple_match_op *
17577 called res_op instead of separate res_code and res_ops parameters.
17578 Update call accordingly.
17579 * gimple-fold.c (replace_stmt_with_simplification): Replace rcode
17580 and ops with a single res_op parameter. Update calls to
17581 maybe_build_generic_op and maybe_push_res_to_seq.
17582 (fold_stmt_1): Update calls to gimple_simplify and
17583 replace_stmt_with_simplification.
17584 (gimple_fold_stmt_to_constant_1): Update calls to gimple_simplify
17585 and gimple_simplified_result_is_gimple_val.
17586 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update call to
17588 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Replace parameters
17589 with a gimple_match_op *.
17590 (vn_nary_build_or_lookup): Likewise. Update call to
17591 vn_nary_build_or_lookup_1.
17592 (vn_nary_build_or_lookup_1): Replace rcode, type and ops with a
17593 gimple_match_op *. Update calls to the gimple_resimplify routines
17594 and to gimple_simplified_result_is_gimple_val.
17595 (vn_nary_simplify): Update call to vn_nary_build_or_lookup_1.
17596 Use gimple_match_op::MAX_NUM_OPS instead of a hard-coded 3.
17597 (vn_reference_lookup_3): Update call to vn_nary_build_or_lookup.
17598 (visit_nary_op): Likewise.
17599 (visit_reference_op_load): Likewise.
17601 2018-05-23 Luis Machado <luis.machado@linaro.org>
17603 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Use correct type
17604 modifier for printing the step amount.
17606 2018-05-23 Jozef Lawrynowicz <jozef.l@somniumtech.com>
17609 * tree.c (build_common_tree_nodes): Dont set TYPE_SIZE for __intN
17612 2018-05-23 Segher Boessenkool <segher@kernel.crashing.org>
17614 * doc/sourcebuild.texi (Endianness): New subsubsection.
17616 2018-05-23 Luis Machado <luis.machado@linaro.org>
17618 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
17619 <prefetch_dynamic_strides>: New const bool field.
17620 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
17621 prefetch_dynamic_strides.
17622 (exynosm1_prefetch_tune): Likewise.
17623 (thunderxt88_prefetch_tune): Likewise.
17624 (thunderx_prefetch_tune): Likewise.
17625 (thunderx2t99_prefetch_tune): Likewise.
17626 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides to
17628 (aarch64_override_options_internal): Update to set
17629 PARAM_PREFETCH_DYNAMIC_STRIDES.
17630 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
17631 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
17632 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
17633 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
17634 prefetch-dynamic-strides setting.
17636 2018-05-23 Luis Machado <luis.machado@linaro.org>
17638 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
17639 <minimum_stride>: New const int field.
17640 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
17641 minimum_stride field defaulting to -1.
17642 (exynosm1_prefetch_tune): Likewise.
17643 (thunderxt88_prefetch_tune): Likewise.
17644 (thunderx_prefetch_tune): Likewise.
17645 (thunderx2t99_prefetch_tune): Likewise.
17646 (qdf24xx_prefetch_tune) <minimum_stride>: Set to 2048.
17647 <default_opt_level>: Set to 3.
17648 (aarch64_override_options_internal): Update to set
17649 PARAM_PREFETCH_MINIMUM_STRIDE.
17650 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
17651 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
17652 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
17653 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
17654 stride is constant and is below the minimum stride threshold.
17656 2018-05-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17658 * config/arm/arm-cpus.in (mode26): Delete.
17659 (armv4): Delete mode26 reference.
17660 * config/arm/arm.c (arm_configure_build_target): Delete use of
17663 2018-05-23 Uros Bizjak <ubizjak@gmail.com>
17665 * config/i386/i386.md (*floatuns<SWI48:mode><MODEF:mode>2_avx512):
17667 (floatunssi<mode>2): Also enable for AVX512F and TARGET_SSE_MATH.
17668 Rewrite expander pattern. Emit gen_floatunssi<mode>2_i387_with_xmm
17670 (floatunsdisf2): Rewrite expander pattern. Hanlde TARGET_AVX512F.
17671 (floatunsdidf2): Ditto.
17673 2018-05-23 Uros Bizjak <ubizjak@gmail.com>
17675 * config/i386/i386.md (fixuns_trunc<mode>di2): New insn pattern.
17676 (fixuns_trunc<mode>si2_avx512f): Ditto.
17677 (*fixuns_trunc<mode>si2_avx512f_zext): Ditto.
17678 (fixuns_trunc<mode>si2): Also enable for AVX512F and TARGET_SSE_MATH.
17679 Emit fixuns_trunc<mode>si2_avx512f for AVX512F targets.
17681 2018-05-23 Alexander Monakov <amonakov@ispras.ru>
17683 PR rtl-optimization/79985
17684 * df-scan.c (df_insn_refs_collect): Remove special case for
17685 global registers and asm statements.
17687 2018-05-23 Alexander Monakov <amonakov@ispras.ru>
17689 * extend.texi (Global Register Variables): Rewrite the bullet list.
17690 Note that the register is available for allocation. Note that access
17691 via inline asm must use constraints. Add note about async-signal
17692 handlers. Remove paragraph about automagic register selection.
17694 2018-05-23 Richard Biener <rguenther@suse.de>
17696 * tree-ssa-sccvn.c (vn_reference_lookup_3): Remove restriction
17697 of fixed offset from memset VN.
17699 2018-05-23 Bill Schmidt <wschmidt@linux.ibm.com>
17701 * gimple-ssa-strength-reduction.c (struct slsr_cand_d): Add
17702 first_interp field.
17703 (alloc_cand_and_find_basis): Initialize first_interp field.
17704 (slsr_process_mul): Modify first_interp field.
17705 (slsr_process_add): Likewise.
17706 (slsr_process_cast): Modify first_interp field for each new
17708 (slsr_process_copy): Likewise.
17709 (dump_candidate): Dump first_interp field.
17710 (replace_mult_candidate): Process all interpretations, not just
17712 (replace_rhs_if_not_dup): Likewise.
17713 (replace_one_candidate): Likewise.
17715 2018-05-23 Wilco Dijkstra <wdijkstr@arm.com>
17717 * config/aarch64/aarch64.c (aarch64_use_frame_pointer):
17719 (aarch64_needs_frame_chain): New function.
17720 (aarch64_parse_override_string): Set aarch64_use_frame_pointer.
17722 2018-05-23 Sudakshina Das <sudi.das@arm.com>
17725 * common/config/aarch64/aarch64-common.c (aarch64_handle_option):
17726 Check val before adding MASK_STRICT_ALIGN to opts->x_target_flags.
17727 * config/aarch64/aarch64.opt (mstrict-align): Remove RejectNegative.
17728 * config/aarch64/aarch64.c (aarch64_attributes): Mark allow_neg
17729 as true for strict-align.
17730 (aarch64_can_inline_p): Perform checks even when callee has no
17731 attributes to check for strict alignment.
17732 * doc/extend.texi (AArch64 Function Attributes): Document
17734 * doc/invoke.texi: (AArch64 Options): Likewise.
17736 2018-05-23 Richard Sandiford <richard.sandiford@linaro.org>
17738 PR tree-optimization/85853
17739 * tree-vect-slp.c (vect_slp_analyze_node_operations): Split out
17740 the handling of the root of the node to...
17741 (vect_slp_analyze_node_operations_1): ...this new function,
17742 and run the whole thing with the child nodes' def types
17743 set according to their SLP node's def type.
17745 2018-05-23 Richard Biener <rguenther@suse.de>
17747 PR middle-end/85874
17748 * tree-data-ref.c (create_runtime_alias_checks): Defer
17749 and ignore overflow warnings.
17751 2018-05-23 Yury Gribov <tetra2005@gmail.com>
17753 PR tree-optimization/85822
17754 * tree-vrp.c (is_masked_range_test): Fix handling of negative
17757 2018-05-23 Richard Biener <rguenther@suse.de>
17759 * tree-ssa-sccvn.c (vn_reference_lookup_3): Handle arbitrary
17760 memset constants via native_interpret_expr.
17762 2018-05-22 H.J. Lu <hongjiu.lu@intel.com>
17765 * cgraph.h (cgraph_node::create): Set ifunc_resolver for ifunc
17767 (cgraph_node::create_alias): Likewise.
17768 (cgraph_node::get_availability): Check ifunc_resolver instead
17769 of looking up ifunc attribute.
17770 * cgraphunit.c (maybe_diag_incompatible_alias): Likewise.
17771 * varasm.c (do_assemble_alias): Likewise.
17772 (assemble_alias): Likewise.
17773 (default_binds_local_p_3): Likewise.
17774 * cgraph.h (cgraph_node): Add ifunc_resolver.
17775 (cgraph_node::only_called_directly_or_aliased_p): Return false
17776 for IFUNC resolver.
17777 * lto-cgraph.c (input_node): Set ifunc_resolver for ifunc
17779 * symtab.c (symtab_node::verify_base): Verify that ifunc_resolver
17780 is equivalent to lookup_attribute ("ifunc", DECL_ATTRIBUTES (decl)).
17781 (symtab_node::binds_to_current_def_p): Check ifunc_resolver
17782 instead of looking up ifunc attribute.
17784 2018-05-22 Luis Machado <luis.machado@linaro.org>
17786 * config/aarch64/aarch64.md (*ashift<mode>_extv_bfiz): New pattern.
17788 2018-05-22 Martin Sebor <msebor@redhat.com>
17790 PR middle-end/85359
17791 * builtins.c (expand_builtin_strcpy): Call maybe_warn_nonstring_arg
17792 only when expasion succeeds.
17793 (expand_builtin_strcmp): Same.
17794 (expand_builtin_strncmp): Same.
17796 2018-05-22 Martin Sebor <msebor@redhat.com>
17798 * calls.c (maybe_warn_nonstring_arg): Fix a typo in a comment.
17800 2018-05-22 Jackson Woodruff <jackson.woodruff@arm.com>
17801 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17803 * config/aarch64/aarch64-ldpstp.md: Replace uses of
17804 aarch64_mem_pair_operand with memory_operand and delete operand swapping
17806 * config/aarch64/aarch64.c (aarch64_operands_ok_for_ldpstp):
17807 Add check for legitimate_address.
17808 (aarch64_gen_adjusted_ldpstp): Swap operands where appropriate.
17809 (aarch64_swap_ldrstr_operands): New.
17810 * config/aarch64/aarch64-protos.h (aarch64_swap_ldrstr_operands):
17813 2018-05-22 Jackson Woodruff <jackson.woodruff@arm.com>
17814 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17816 * config/aarch64/aarch64.md: New patterns to generate stp
17818 (store_pair_sw, store_pair_dw): New patterns to generate stp for
17819 single words and double words.
17820 (load_pair_sw, load_pair_dw): Likewise.
17821 (store_pair_sf, store_pair_df, store_pair_si, store_pair_di):
17823 (load_pair_sf, load_pair_df, load_pair_si, load_pair_di):
17825 * config/aarch64/aarch64-ldpstp.md: Modify peephole
17826 for different mode ldpstp and add peephole for merged zero stores.
17827 Likewise for loads.
17828 * config/aarch64/aarch64.c (aarch64_operands_ok_for_ldpstp):
17830 (aarch64_gen_store_pair): Rename calls to match new patterns.
17831 (aarch64_gen_load_pair): Rename calls to match new patterns.
17832 * config/aarch64/aarch64-simd.md (load_pair<mode>): Rename to...
17833 (load_pair<DREG:mode><DREG2:mode>): ... This.
17834 (store_pair<mode>): Rename to...
17835 (vec_store_pair<DREG:mode><DREG2:mode>): ... This.
17836 * config/aarch64/iterators.md (DREG, DREG2, DX2, SX, SX2, DSX):
17837 New mode iterators.
17838 (V_INT_EQUIV): Handle SImode.
17839 * config/aarch64/predicates.md (aarch64_reg_zero_or_fp_zero):
17842 2018-05-22 Martin Sebor <msebor@redhat.com>
17845 * calls.c (maybe_warn_nonstring_arg): Use string length to set
17846 or ajust the presumed bound on an operation to avoid unnecessary
17849 2018-05-22 Martin Sebor <msebor@redhat.com>
17851 PR tree-optimization/85826
17852 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Avoid
17853 assuming that a DECL necesarily has a constant size.
17855 2018-05-22 Richard Sandiford <richard.sandiford@linaro.org>
17857 PR middle-end/85862
17858 * internal-fn.c (expand_direct_optab_fn): Cope with a null lhs.
17860 2018-05-22 Richard Biener <rguenther@suse.de>
17862 PR tree-optimization/85834
17863 * tree-ssa-sccvn.c (vn_reference_lookup_3): Properly handle
17864 non-constant and non-zero memset arguments.
17866 2018-05-22 Martin Liska <mliska@suse.cz>
17869 * ipa-icf.c (sem_item::add_type): Do not ICE for incomplete types.
17871 2018-05-22 Richard Biener <rguenther@suse.de>
17873 PR tree-optimization/85863
17874 * tree-vect-stmts.c (vect_is_simple_cond): Only widen invariant
17875 comparisons when vectype is specified.
17876 (vectorizable_condition): Do not specify vectype for
17877 vect_is_simple_cond when SLP vectorizing.
17879 2018-05-21 Michael Meissner <meissner@linux.ibm.com>
17882 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Do not
17883 define __ibm128 as long double.
17884 * config/rs6000/rs6000.c (rs6000_init_builtins): Create __ibm128
17885 as a distinct type when IEEE 128-bit support is enabled.
17886 (init_float128_ieee): Fix up conversions between IFmode and IEEE
17887 128-bit types to use the correct functions.
17888 (rs6000_expand_float128_convert): Use explicit FLOAT_EXTEND to
17889 convert between 128-bit floating point types that have different
17890 modes but the same representation, instead of using gen_lowpart to
17892 * config/rs6000/rs6000.md (IFKF): New iterator for IFmode and
17894 (IFKF_reg): New attributes to give the register constraints for
17896 (extend<mode>tf2_internal): New insns to mark an explicit
17897 conversion between 128-bit floating point types that have a
17898 different mode but share the same representation.
17900 2018-05-21 Richard Sandiford <richard.sandiford@linaro.org>
17902 PR tree-optimization/85814
17903 * tree-ssa-strlen.c (get_stridx_plus_constant): Cope with
17904 a null return from get_strinfo when unsharing the next
17905 strinfo in the chain.
17907 2018-05-21 Vladimir Mezentsev <vladimir.mezentsev@oracle.com>
17910 * varasm.c (weak_finish): Clean up weak_decls.
17912 2018-05-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17914 * config/aarch64/aarch64.md ("unspec"): Define UNSPEC_SABAL,
17915 UNSPEC_SABDL2, UNSPEC_SADALP, UNSPEC_UABAL, UNSPEC_UABDL2,
17916 UNSPEC_UADALP values.
17917 * config/aarch64/iterators.md (ABAL): New int iterator.
17920 (sur): Add mappings for the above.
17921 * config/aarch64/aarch64-simd.md (aarch64_<sur>abdl2<mode>_3):
17923 (aarch64_<sur>abal<mode>_4): Likewise.
17924 (aarch64_<sur>adalp<mode>_3): Likewise.
17925 (<sur>sadv16qi): New define_expand.
17927 2018-05-21 Alexander Nesterovskiy <alexander.nesterovskiy@intel.com>
17929 * config/i386/i386.md (*movsf_internal): AVX falsedep fix.
17930 (*movdf_internal): Ditto.
17931 (*rcpsf2_sse): Ditto.
17932 (*rsqrtsf2_sse): Ditto.
17933 (*sqrt<mode>2_sse): Ditto.
17935 2018-05-21 Tamar Christina <tamar.christina@arm.com>
17937 * config/aarch64/aarch64-simd.md (aarch64_eor3qv8hi): Change to
17939 (aarch64_bcaxqv8hi): Change to bcaxq<mode>4.
17940 * config/aarch64/aarch64-simd-builtins.def (veor3q_u8, veor3q_u32,
17941 veor3q_u64, veor3q_s8, veor3q_s16, veor3q_s32, veor3q_s64, vbcaxq_u8,
17942 vbcaxq_u32, vbcaxq_u64, vbcaxq_s8, vbcaxq_s16, vbcaxq_s32,
17944 * config/aarch64/arm_neon.h: Likewise.
17945 * config/aarch64/iterators.md (VQ_I): New.
17947 2018-05-21 Alexey Brodkin <abrodkin@synopsys.com>
17949 * config.gcc: Add arc/t-multilib-linux to tmake_file for
17951 * config/arc/t-multilib-linux: Specify MULTILIB_OPTIONS and
17954 2018-05-20 Chung-Ju Wu <jasonwucj@gmail.com>
17956 * config/nds32/constraints.md (S): New constraint.
17957 * config/nds32/nds32.md (call_internal): Use constraint S.
17958 (call_value_internal): Likewise.
17959 (sibcall_internal): Likewise.
17960 (sibcall_value_internal): Likewise.
17962 2018-05-20 Kito Cheng <kito.cheng@gmail.com>
17963 Chung-Ju Wu <jasonwucj@gmail.com>
17965 * config/nds32/nds32.c (nds32_register_move_cost): Take garywolf cpu
17966 into consideration.
17968 2018-05-20 Kito Cheng <kito.cheng@gmail.com>
17969 Chung-Ju Wu <jasonwucj@gmail.com>
17971 * config/nds32/nds32-cost.c (rtx_cost_model_t): New structure.
17972 (insn_size_16bit, insn_size_32bit): New variables for cost evaluation.
17973 (nds32_rtx_costs_impl): Simplify.
17974 (nds32_address_cost_impl): Simplify.
17975 (nds32_init_rtx_costs): New function.
17976 (nds32_rtx_costs_speed_prefer): Likewise.
17977 (nds32_rtx_costs_size_prefer): Likewise.
17978 (nds32_address_cost_speed_prefer): Likewise.
17979 (nds32_address_cost_speed_fwprop): Likewise.
17980 (nds32_address_cost_size_prefer): Likewise.
17981 * config/nds32/nds32-protos.h (nds32_init_rtx_costs): Declare.
17982 * config/nds32/nds32.c (nds32_option_override): Use
17983 nds32_init_rtx_costs function.
17985 2018-05-20 Chung-Ju Wu <jasonwucj@gmail.com>
17987 * config/nds32/nds32.c (nds32_asm_file_start): Output pipeline model.
17988 * config/nds32/nds32.h (TARGET_PIPELINE_N7): Define.
17989 (TARGET_PIPELINE_N8): Likewise.
17990 (TARGET_PIPELINE_N10): Likewise.
17991 (TARGET_PIPELINE_N13): Likewise.
17992 (TARGET_PIPELINE_GRAYWOLF): Likewise.
17994 2018-05-19 Monk Chiang <sh.chiang04@gmail.com>
17996 * config/nds32/nds32-fpu.md: Update copyright year.
17998 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
18000 * config/nds32/nds32.h (ASM_SPEC): Adjust spec rule.
18002 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
18004 * config/nds32/nds32.c
18005 (nds32_md_asm_adjust): Consider flag_inline_asm_r15 variable.
18006 * config/nds32/nds32.opt (minline-asm-r15): New option.
18008 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
18010 * common/config/nds32/nds32-common.c (TARGET_DEFAULT_TARGET_FLAGS): Add
18012 * config/nds32/nds32.md (abssi2): New pattern.
18014 2018-05-19 Uros Bizjak <ubizjak@gmail.com>
18016 * config/i386/i386.md (rex64namesuffix): New mode attribute.
18017 * config/i386/sse.md (sse_cvtsi2ss<rex64namesuffix><round_name>):
18018 Merge insn pattern from sse_cvtsi2ss<round_name> and
18019 sse_cvtsi2ssq<round_name> using SWI48 mode iterator.
18020 (sse_cvtss2si<rex64namesuffix><round_name>): Merge insn pattern
18021 from sse_cvtss2si<round_name> and sse_cvtss2siq<round_name>
18022 using SWI48 mode iterator.
18023 (sse_cvtss2si<rex64namesuffix>_2): Merge insn pattern from
18024 sse_cvtss2si_2 and sse_cvtss2siq_2 using SWI48 mode iterator.
18025 (sse_cvttss2si<rex64namesuffix><round_saeonly_name>): Merge insn
18026 pattern from sse_cvttss2si<round_saeonly_name>
18027 and sse_cvttss2siq<round_saeonly_name> using SWI48 mode iterator.
18028 (avx512f_vcvtss2usi<rex64namesuffix><round_name>): Merge insn pattern
18029 from avx512f_vcvtss2usi<round_name> and avx512f_vcvtss2usiq<round_name>
18030 using SWI48 mode iterator.
18031 (avx512f_vcvttss2usi<rex64namesuffix><round_saeonly_name>): Merge
18032 insn pattern from avx512f_vcvttss2usi<round_saeonly_name> and
18033 avx512f_vcvttss2usiq<round_saeonly_name> using SWI48 mode iterator.
18034 (avx512f_vcvtsd2usi<rex64namesuffix><round_name>): Merge insn pattern
18035 from avx512f_vcvtsd2usi<round_name> and avx512f_vcvtsd2usiq<round_name>
18036 using SWI48 mode iterator.
18037 (avx512f_vcvttsd2usi<rex64namesuffix><round_saeonly_name>): Merge
18038 insn pattern from avx512f_vcvttsd2usi<round_saeonly_name> and
18039 avx512f_vcvttsd2usiq<round_saeonly_name> using SWI48 mode iterator.
18040 (sse2_cvtsd2si<rex64namesuffix><round_name>): Merge insn pattern from
18041 sse2_cvtsd2si<round_name> and sse2_cvtsd2siq<round_name> using
18042 SWI48 mode iterator.
18043 (sse2_cvtsd2si<rex64namesuffix>_2): Merge insn pattern from
18044 sse2_cvtsd2si_2 and sse2_cvtsd2siq_2 using SWI48 mode iterator.
18045 (sse_cvttsd2si<rex64namesuffix><round_saeonly_name>): Merge insn
18046 pattern from sse_cvttsd2si<round_saeonly_name>
18047 and sse_cvttsd2siq<round_saeonly_name> using SWI48 mode iterator.
18049 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
18051 * config/nds32/nds32-md-auxiliary.c
18052 (nds32_valid_smw_lwm_base_p): Refine.
18053 (nds32_output_smw_single_word): Refine.
18054 (nds32_output_smw_double_word): New.
18055 * config/nds32/nds32-protos.h (nds32_output_smw_double_word): New.
18057 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
18059 * config/nds32/nds32-md-auxiliary.c (nds32_output_stack_push): Refine.
18060 (nds32_output_stack_pop): Refine.
18061 (nds32_expand_unaligned_load): Refine.
18062 (nds32_expand_unaligned_store): Refine.
18064 2018-05-19 Kuan-Lin Chen <kuanlinchentw@gmail.com>
18065 Chung-Ju Wu <jasonwucj@gmail.com>
18067 * config/nds32/constants.md: Add TP_REGNUM constant.
18068 (unspec_element): Add UNSPEC_GOTINIT, UNSPEC_GOT, UNSPEC_GOTOFF,
18069 UNSPEC_PLT, UNSPEC_TLSGD, UNSPEC_TLSLD, UNSPEC_TLSIE, UNSPEC_TLSLE and
18071 * config/nds32/nds32-doubleword.md: Consider flag_pic.
18072 * config/nds32/nds32-dspext.md (mov<mode>): Expand TLS and PIC cases.
18073 * config/nds32/nds32-predicates.c (nds32_const_unspec_p): New.
18074 * config/nds32/nds32-md-auxiliary.c: Implementation that support TLS
18075 and PIC code generation.
18076 * config/nds32/nds32-protos.h: Declarations that support TLS and PIC
18078 * config/nds32/nds32-relax-opt.c: Consider TLS and PIC for relax
18080 * config/nds32/nds32.md: Support TLS and PIC.
18081 * config/nds32/nds32.c: Support TLS and PIC.
18082 * config/nds32/nds32.h (nds32_relax_insn_type): New enum type.
18083 * config/nds32/predicates.md (nds32_nonunspec_symbolic_operand): New
18086 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
18088 * config/nds32/nds32-predicates.c (const_vector_to_hwint): Use machine
18089 mode with E_ prefix.
18091 2018-05-19 Kuan-Lin Chen <kuanlinchentw@gmail.com>
18092 Chung-Ju Wu <jasonwucj@gmail.com>
18094 * config/nds32/constants.md (unspec_element): Add UNSPEC_ICT.
18095 * config/nds32/nds32-md-auxiliary.c
18096 (symbolic_reference_mentioned_p): New.
18097 (nds32_legitimize_ict_address): New.
18098 (nds32_expand_ict_move): New.
18099 (nds32_indirect_call_referenced_p): New.
18100 (nds32_symbol_binds_local_p): Delete.
18101 (nds32_long_call_p): Modify.
18102 * config/nds32/nds32-opts.h (nds32_ict_model_type): New enum type.
18103 * config/nds32/nds32-protos.h
18104 (symbolic_reference_mentioned_p): Declare.
18105 (nds32_legitimize_ict_address): Declare.
18106 (nds32_expand_ict_move): Declare.
18107 (nds32_indirect_call_referenced_p): Declare.
18108 * config/nds32/nds32-relax-opt.c (nds32_ict_const_p): New.
18109 (nds32_relax_group): Use nds32_ict_const_p as condition.
18110 * config/nds32/nds32.c (nds32_attribute_table): Add "indirect_call".
18111 (nds32_asm_file_start): Output ict_model directive in asm code.
18112 (nds32_legitimate_address_p): Consider indirect call.
18113 (nds32_print_operand): Consider indirect call.
18114 (nds32_print_operand_address): Consider indirect call.
18115 (nds32_insert_attributes): Handle "indirect_call" attribute.
18116 (TARGET_LEGITIMATE_ADDRESS_P): Define.
18117 (TARGET_LEGITIMATE_CONSTANT_P): Define.
18118 (TARGET_CANNOT_FORCE_CONST_MEM): Define.
18119 (TARGET_DELEGITIMIZE_ADDRESS): Define.
18120 (TARGET_ASM_OUTPUT_ADDR_CONST_EXTRA): Define.
18121 * config/nds32/nds32.h (SYMBOLIC_CONST_P): Define.
18122 (TARGET_ICT_MODEL_SMALL): Define.
18123 (TARGET_ICT_MODEL_LARGE): Define.
18124 * config/nds32/nds32.md (movsi): Consider ict model.
18125 (call, call_value): Consider ict model.
18126 (sibcall, sibcall_value): Consider ict model.
18127 * config/nds32/nds32.opt (mict-model): New option.
18128 * config/nds32/predicates.md (nds32_symbolic_operand): Consider ict
18131 2018-05-18 Kito Cheng <kito.cheng@gmail.com>
18132 Monk Chiang <sh.chiang04@gmail.com>
18133 Jim Wilson <jimw@sifive.com>
18135 * common/config/riscv/riscv-common.c (riscv_parse_arch_string):
18136 Add support to parse rv32e*. Clear MASK_RVE for rv32i and rv64i.
18137 * config.gcc (riscv*-*-*): Add support for rv32e* and ilp32e.
18138 * config/riscv/riscv-c.c (riscv_cpu_cpp_builtins): Define
18139 __riscv_32e when TARGET_RVE. Handle ABI_ILP32E as soft-float ABI.
18140 * config/riscv/riscv-opts.h (riscv_abi_type): Add ABI_ILP32E.
18141 * config/riscv/riscv.c (riscv_compute_frame_info): When TARGET_RVE,
18142 compute save_libcall_adjustment properly.
18143 (riscv_option_override): Call error if TARGET_RVE and not ABI_ILP32E.
18144 (riscv_conditional_register_usage): Handle TARGET_RVE and ABI_ILP32E.
18145 * config/riscv/riscv.h (UNITS_PER_FP_ARG): Handle ABI_ILP32E.
18146 (STACK_BOUNDARY, ABI_STACK_BOUNDARY): Handle TARGET_RVE.
18147 (GP_REG_LAST, MAX_ARGS_IN_REGISTERS): Likewise.
18148 (ABI_SPEC): Handle mabi=ilp32e.
18149 * config/riscv/riscv.opt (abi_type): Add ABI_ILP32E.
18150 (RVE): Add RVE mask.
18151 * doc/invoke.texi (RISC-V options) <-mabi>: Add ilp32e info.
18152 <-march>: Add rv32e as an example.
18154 2018-05-18 Marc Glisse <marc.glisse@inria.fr>
18157 * tree-ssa-structalias.c (create_variable_info_for_1): Extra argument.
18158 (intra_create_variable_infos): Handle C++ constructors.
18160 2018-05-18 Martin Liska <mliska@suse.cz>
18162 * passes.def: Remove a redundant pass.
18164 2018-05-18 Eric Botcazou <ebotcazou@adacore.com>
18167 * config/sparc/sparc.c (sparc_expand_builtin): Always initialize op[0].
18169 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18171 * config/arm/arm-cpus.in (armv3m, mode32): Delete features.
18173 (ARMv2, ARMv3, ARMv3m): Delete fgroups.
18175 (armv2, armv2a, armv3, armv3m): Delete architectures.
18176 (arm2, arm250, arm3, arm6, arm60, arm600, arm610, arm620,
18177 arm7, arm7d, arm7di, arm70, arm700, arm700i, arm710, arm720,
18178 arm710c, arm7100, arm7500, arm7500fe, arm7m, arm7dm, arm7dmi):
18180 * config/arm/arm.md (maddsidi4): Remove check for arm_arch3m.
18181 (*mulsidi3adddi): Likewise.
18182 (mulsidi3): Likewise.
18183 (*mulsidi3_nov6): Likewise.
18184 (umulsidi3): Likewise.
18185 (umulsidi3_nov6): Likewise.
18186 (umaddsidi4): Likewise.
18187 (*umulsidi3adddi): Likewise.
18188 (smulsi3_highpart): Likewise.
18189 (*smulsi3_highpart_nov6): Likewise.
18190 (umulsi3_highpart): Likewise.
18191 (*umulsi3_highpart_nov6): Likewise.
18192 * config/arm/arm.h (arm_arch3m): Delete.
18193 * config/arm/arm.c (arm_arch3m): Delete.
18194 (arm_option_override_internal): Update armv3-related comment.
18195 (arm_configure_build_target): Delete use of isa_bit_mode32.
18196 (arm_option_reconfigure_globals): Delete set of arm_ach3m.
18197 (arm_rtx_costs_internal): Delete check of arm_arch3m.
18198 * config/arm/arm-fixed.md (mulsq3): Delete check for arm_arch3m.
18199 (mulsa3): Likewise.
18200 (mulusa3): Likewise.
18201 * config/arm/arm-protos.h (arm_arch3m): Delete.
18202 * config/arm/arm-tables.opt: Regenerate.
18203 * config/arm/arm-tune.md: Likewise.
18204 * config/arm/t-arm-elf (all_early_nofp): Delete mentions of
18205 deleted architectures.
18207 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18209 * config/arm/arm-cpus.in (armv5, armv5e): Delete features.
18210 (armv5t, armv5te): New features.
18211 (ARMv5, ARMv5e): Delete fgroups.
18212 (ARMv5t, ARMv5te): Adjust for above changes.
18213 (ARMv6m): Likewise.
18214 (armv5, armv5e): Delete arches.
18215 * config/arm/arm.md (*call_reg_armv5): Use arm_arch5t instead of
18217 (*call_reg_arm): Likewise.
18218 (*call_value_reg_armv5): Likewise.
18219 (*call_value_reg_arm): Likewise.
18220 (*call_symbol): Likewise.
18221 (*call_value_symbol): Likewise.
18222 (*sibcall_insn): Likewise.
18223 (*sibcall_value_insn): Likewise.
18224 (clzsi2): Likewise.
18225 (prefetch): Likewise.
18226 (define_split and define_peephole2 dependent on arm_arch5):
18228 * config/arm/arm.h (TARGET_LDRD): Use arm_arch5te instead of
18230 (TARGET_ARM_QBIT): Likewise.
18231 (TARGET_DSP_MULTIPLY): Likewise.
18232 (enum base_architecture): Delete BASE_ARCH_5, BASE_ARCH_5E.
18233 (arm_arch5, arm_arch5e): Delete.
18234 (arm_arch5t, arm_arch5te): Declare.
18235 * config/arm/arm.c (arm_arch5, arm_arch5e): Delete.
18236 (arm_arch5t): Declare.
18237 (arm_option_reconfigure_globals): Update for the above.
18238 (arm_options_perform_arch_sanity_checks): Update comment, replace
18239 use of arm_arch5 with arm_arch5t.
18240 (use_return_insn): Likewise.
18241 (arm_emit_call_insn): Likewise.
18242 (output_return_instruction): Likewise.
18243 (arm_final_prescan_insn): Likewise.
18244 (arm_coproc_builtin_available): Likewise.
18245 * config/arm/arm-c.c (arm_cpu_builtins): Replace arm_arch5 and
18246 arm_arch5e with arm_arch5t and arm_arch5te.
18247 * config/arm/arm-protos.h (arm_arch5, arm_arch5e): Delete.
18248 (arm_arch5t, arm_arch5te): Declare.
18249 * config/arm/arm-tables.opt: Regenerate.
18250 * config/arm/t-arm-elf: Remove references to armv5, armv5e.
18251 * config/arm/t-multilib: Likewise.
18252 * config/arm/thumb1.md (*call_reg_thumb1_v5): Check arm_arch5t
18253 instead of arm_arch5.
18254 (*call_reg_thumb1): Likewise.
18255 (*call_value_reg_thumb1_v5): Likewise.
18256 (*call_value_reg_thumb1): Likewise.
18257 * config/arm/vxworks.h (TARGET_OS_CPP_BUILTINS): Remove now
18259 * doc/invoke.texi (ARM Options): Remove references to armv5, armv5e.
18261 2018-05-18 Martin Liska <mliska@suse.cz>
18263 PR gcov-profile/84846
18264 * doc/gcov.texi: Document -t option of gcov tool.
18266 2018-05-18 Martin Liska <mliska@suse.cz>
18268 PR gcov-profile/84846
18269 * gcov.c (print_usage): Add new -t option.
18270 (process_args): Handle the option.
18271 (generate_results): Use stdout as output when requested by
18274 2018-05-18 Martin Liska <mliska@suse.cz>
18276 PR gcov-profile/84846
18277 * coverage.c (coverage_init): Write PWD to .gcno file.
18278 * doc/gcov.texi: Document how working directory is printed.
18279 * gcov-dump.c (dump_gcov_file): Print PWD.
18280 * gcov.c (output_intermediate_file): Likewise.
18281 (read_graph_file): Read PWD string.
18282 (output_lines): Print PWD.
18284 2018-05-18 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
18286 PR middle-end/85817
18287 * ipa-pure-const.c (malloc_candidate_p): Remove the check integer_zerop
18288 for retval and return false if all args to phi are zero.
18290 2018-05-18 Richard Biener <rguenther@suse.de>
18292 * gimple-ssa-evrp.c (class evrp_folder): Add simplify_stmt_using_ranges
18294 (evrp_dom_walker::before_dom_children): Call it.
18296 2018-05-18 Richard Biener <rguenther@suse.de>
18298 * tree-dfa.c (get_ref_base_and_extent): Use range-info to refine
18299 results when processing array refs with variable index.
18301 2018-05-18 Toon Moene <toon@moene.org>
18303 * doc/invoke.texi: Move -floop-unroll-and-jam documentation
18304 directly after that of -floop-interchange. Indicate that both
18305 options are enabled by default when specifying -O3.
18307 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18309 * config/aarch64/aarch64-simd.md (vec_set<mode>): Use VALL_F16 mode
18310 iterator. Delete separate integer-mode vec_set<mode> expander.
18311 (aarch64_simd_vec_setv2di): Delete.
18312 (vec_setv2di): Delete.
18313 (aarch64_simd_vec_set<mode>): Delete all other patterns with that name.
18314 Use VALL_F16 mode iterator. Add LD1 alternative and use vwcore for
18315 the "w, r" alternative.
18317 2018-05-18 Martin Liska <mliska@suse.cz>
18319 * passes.def: Add pass_lower_switch and pass_lower_switch_O0.
18320 * tree-pass.h (make_pass_lower_switch_O0): New function.
18321 * tree-switch-conversion.c (node_has_low_bound): Remove.
18322 (node_has_high_bound): Likewise.
18323 (node_is_bounded): Likewise.
18324 (class pass_lower_switch): Make it a template type and create
18326 (pass_lower_switch::execute): Add template argument.
18327 (make_pass_lower_switch): New function.
18328 (make_pass_lower_switch_O0): New function.
18329 (do_jump_if_equal): Remove.
18330 (emit_case_nodes): Simplify to just handle all 3 cases and leave
18331 all the hard work to tree optimization passes.
18333 2018-05-18 Martin Liska <mliska@suse.cz>
18335 * dbgcnt.c (limit_low): Renamed from limit.
18336 (limit_high): New variable.
18337 (dbg_cnt_is_enabled): Check for upper limit.
18338 (dbg_cnt): Adjust dumping.
18339 (dbg_cnt_set_limit_by_index): Add new argument for high
18341 (dbg_cnt_set_limit_by_name): Likewise.
18342 (dbg_cnt_process_single_pair): Parse new format.
18343 (dbg_cnt_process_opt): Use strtok.
18344 (dbg_cnt_list_all_counters): Remove 'value' and add
18346 * doc/invoke.texi: Document changes.
18348 2018-05-18 Richard Sandiford <richard.sandiford@linaro.org>
18350 * doc/sourcebuild.texi (scalar_all_fma): Document.
18351 * tree.def (FMA_EXPR): Delete.
18352 * internal-fn.def (FMA, FMS, FNMA, FNMS): New internal functions.
18353 * internal-fn.c (ternary_direct): New macro.
18354 (expand_ternary_optab_fn): Likewise.
18355 (direct_ternary_optab_supported_p): Likewise.
18356 * Makefile.in (build/genmatch.o): Depend on case-fn-macros.h.
18357 * builtins.c (fold_builtin_fma): Delete.
18358 (fold_builtin_3): Don't call it.
18359 * cfgexpand.c (expand_debug_expr): Remove FMA_EXPR handling.
18360 * expr.c (expand_expr_real_2): Likewise.
18361 * fold-const.c (operand_equal_p): Likewise.
18362 (fold_ternary_loc): Likewise.
18363 * gimple-pretty-print.c (dump_ternary_rhs): Likewise.
18364 * gimple.c (DEFTREECODE): Likewise.
18365 * gimplify.c (gimplify_expr): Likewise.
18366 * optabs-tree.c (optab_for_tree_code): Likewise.
18367 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
18368 * tree-eh.c (operation_could_trap_p): Likewise.
18369 (stmt_could_throw_1_p): Likewise.
18370 * tree-inline.c (estimate_operator_cost): Likewise.
18371 * tree-pretty-print.c (dump_generic_node): Likewise.
18372 (op_code_prio): Likewise.
18373 * tree-ssa-loop-im.c (stmt_cost): Likewise.
18374 * tree-ssa-operands.c (get_expr_operands): Likewise.
18375 * tree.c (commutative_ternary_tree_code, add_expr): Likewise.
18376 * fold-const-call.h (fold_fma): Delete.
18377 * fold-const-call.c (fold_const_call_ssss): Handle CFN_FMS,
18378 CFN_FNMA and CFN_FNMS.
18379 (fold_fma): Delete.
18380 * genmatch.c (combined_fn): New enum.
18381 (commutative_ternary_tree_code): Remove FMA_EXPR handling.
18382 (commutative_op): New function.
18383 (commutate): Use it. Handle more than 2 operands.
18384 (dt_operand::gen_gimple_expr): Use commutative_op.
18385 (parser::parse_expr): Allow :c to be used with non-binary
18386 operators if the commutative operand is known.
18387 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Handle
18388 CFN_FMS, CFN_FNMA and CFN_FNMS.
18389 (backprop::process_assign_use): Remove FMA_EXPR handling.
18390 * hsa-gen.c (gen_hsa_insns_for_operation_assignment): Likewise.
18391 (gen_hsa_fma): New function.
18392 (gen_hsa_insn_for_internal_fn_call): Use it for IFN_FMA, IFN_FMS,
18393 IFN_FNMA and IFN_FNMS.
18394 * match.pd: Add folds for IFN_FMS, IFN_FNMA and IFN_FNMS.
18395 * gimple-fold.h (follow_all_ssa_edges): Declare.
18396 * gimple-fold.c (follow_all_ssa_edges): New function.
18397 * tree-ssa-math-opts.c (convert_mult_to_fma_1): Use the
18398 gimple_build interface and use follow_all_ssa_edges to fold the result.
18399 (convert_mult_to_fma): Use direct_internal_fn_suppoerted_p
18400 instead of checking for optabs directly.
18401 * config/i386/i386.c (ix86_add_stmt_cost): Recognize FMAs as calls
18402 rather than FMA_EXPRs.
18403 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Create a
18404 call to IFN_FMA instead of an FMA_EXPR.
18406 2018-05-17 Jim Wilson <jimw@sifive.com>
18408 * expr.c (do_tablejump): When converting index to Pmode, if we have a
18409 sign extended promoted subreg, and the range does not have the sign bit
18410 set, then do a sign extend.
18412 * config/riscv/riscv.c (riscv_extend_comparands): In unsigned QImode
18413 test, check for sign extended subreg and/or constant operands, and
18414 do a sign extend in that case.
18416 2018-05-17 Steve Ellcey <sellcey@cavium.com>
18418 * config/aarch64/thunderx2t99.md (thunderx2t99_ls_both): Delete.
18419 (thunderx2t99_multiple): Delete psuedo-units from used cpus.
18421 (thunderx2t99_alu_shift): Remove alu_shift_reg, alus_shift_reg.
18422 Change logics_shift_reg to logics_shift_imm.
18423 (thunderx2t99_fp_loadpair_basic): Delete.
18424 (thunderx2t99_fp_storepair_basic): Delete.
18425 (thunderx2t99_asimd_int): Add neon_sub and neon_sub_q types.
18426 (thunderx2t99_asimd_polynomial): Delete.
18427 (thunderx2t99_asimd_fp_simple): Add neon_fp_mul_s_scalar_q
18428 and neon_fp_mul_d_scalar_q.
18429 (thunderx2t99_asimd_fp_conv): Add *int_to_fp* types.
18430 (thunderx2t99_asimd_misc): Delete neon_dup and neon_dup_q.
18431 (thunderx2t99_asimd_recip_step): Add missing *sqrt* types.
18432 (thunderx2t99_asimd_lut): Add missing tbl types.
18433 (thunderx2t99_asimd_ext): Delete.
18434 (thunderx2t99_asimd_load1_1_mult): Delete.
18435 (thunderx2t99_asimd_load1_2_mult): Delete.
18436 (thunderx2t99_asimd_load1_ldp): New.
18437 (thunderx2t99_asimd_load1): New.
18438 (thunderx2t99_asimd_load2): Add missing *load2* types.
18439 (thunderx2t99_asimd_load3): New.
18440 (thunderx2t99_asimd_load4): New.
18441 (thunderx2t99_asimd_store1_1_mult): Delete.
18442 (thunderx2t99_asimd_store1_2_mult): Delete.
18443 (thunderx2t99_asimd_store2_mult): Delete.
18444 (thunderx2t99_asimd_store2_onelane): Delete.
18445 (thunderx2t99_asimd_store_stp): New.
18446 (thunderx2t99_asimd_store1): New.
18447 (thunderx2t99_asimd_store2): New.
18448 (thunderx2t99_asimd_store3): New.
18449 (thunderx2t99_asimd_store4): New.
18451 2018-05-17 Jerome Lambourg <lambourg@adacore.com>
18453 * config/arm/arm_cmse.h (cmse_nsfptr_create, cmse_is_nsfptr): Remove
18454 #include <stdint.h>. Replace intptr_t with __INTPTR_TYPE__.
18456 2018-05-17 Pat Haugen <pthaugen@us.ibm.com>
18457 Segher Boessenkool <segher@kernel.crashing.org>
18460 * config/rs6000/rs6000.c (rs6000_output_move_128bit): Check dest
18463 2018-05-17 Richard Biener <rguenther@suse.de>
18465 * tree-ssa-dse.c (dse_classify_store): Fix iterator increment
18466 for pruning loop and prune defs feeding only already visited PHIs.
18468 2018-05-17 Richard Biener <rguenther@suse.de>
18470 * tree-ssa-sccvn.c (vn_reference_lookup_3): Improve memset handling.
18472 2018-05-17 Bin Cheng <bin.cheng@arm.com>
18473 Richard Biener <rguenther@suse.de>
18475 PR tree-optimization/85793
18476 * tree-vect-stmts.c (vectorizable_load): Handle 1 element-wise load
18477 for VMAT_ELEMENTWISE.
18479 2018-05-17 Richard Sandiford <richard.sandiford@linaro.org>
18481 * internal-fn.h (lookup_internal_fn): Declare
18482 * internal-fn.c (lookup_internal_fn): New function.
18483 * gimple.c (gimple_build_call_from_tree): Handle calls to
18484 internal functions.
18485 * gimple-pretty-print.c (dump_gimple_call): Print "." before
18486 internal function names.
18487 * tree-pretty-print.c (dump_generic_node): Likewise.
18488 * tree-ssa-scopedtables.c (expr_hash_elt::print): Likewise.
18490 2018-05-17 Richard Sandiford <richard.sandiford@linaro.org>
18492 * gimple-fold.h (gimple_build): Make the function forms take
18493 combined_fn rather than built_in_function.
18494 (gimple_simplify): Likewise.
18495 * gimple-match-head.c (gimple_simplify): Likewise.
18496 * gimple-fold.c (gimple_build): Likewise.
18497 * tree-vect-loop.c (get_initial_def_for_reduction): Use gimple_build
18498 rather than gimple_build_call_internal.
18499 (get_initial_defs_for_reduction): Likewise.
18500 (vect_create_epilog_for_reduction): Likewise.
18501 (vectorizable_live_operation): Likewise.
18503 2018-05-17 Martin Liska <mliska@suse.cz>
18505 * gimple-ssa-sprintf.c (format_directive): Do not use
18506 space in between 'G_' and '('.
18508 2018-05-17 Jakub Jelinek <jakub@redhat.com>
18511 * config/i386/i386.c (ix86_fold_builtin): Handle masked shifts
18512 even if the mask is not all ones.
18515 * config/i386/i386.c (ix86_fold_builtin): Fold shift builtins by
18517 (ix86_gimple_fold_builtin): Likewise.
18520 * config/i386/i386.c: Include tree-vector-builder.h.
18521 (ix86_vector_shift_count): New function.
18522 (ix86_fold_builtin): Fold shift builtins by scalar count.
18523 (ix86_gimple_fold_builtin): Likewise.
18525 * config/i386/avx512fintrin.h (_mm512_set_epi16, _mm512_set_epi8,
18526 _mm512_setzero): New intrinsics.
18528 2018-05-17 James Greenhalgh <james.greenhalgh@arm.com>
18529 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18531 * config/aarch64/aarch64.c (aarch64_expand_vector_init): Modify
18532 code generation for cases where splatting a value is not useful.
18533 * simplify-rtx.c (simplify_ternary_operation): Simplify
18534 vec_merge across a vec_duplicate and a paradoxical subreg forming
18535 a vector mode to a vec_concat.
18537 2018-05-17 Olga Makhotina <olga.makhotina@intel.com>
18539 * config.gcc: Support "goldmont-plus".
18540 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
18542 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
18543 PROCESSOR_GOLDMONT_PLUS.
18544 * config/i386/i386.c (m_GOLDMONT_PLUS): Define.
18545 (processor_target_table): Add "goldmont-plus".
18546 (PTA_GOLDMONT_PLUS): Define.
18547 (ix86_lea_outperforms): Add TARGET_GOLDMONT_PLUS.
18548 (get_builtin_code_for_version): Handle PROCESSOR_GOLDMONT_PLUS.
18549 (fold_builtin_cpu): Add M_INTEL_GOLDMONT_PLUS.
18550 (fold_builtin_cpu): Add "goldmont-plus".
18551 (ix86_add_stmt_cost): Add TARGET_GOLDMONT_PLUS.
18552 (ix86_option_override_internal): Add "goldmont-plus".
18553 * config/i386/i386.h (processor_costs): Define TARGET_GOLDMONT_PLUS.
18554 (processor_type): Add PROCESSOR_GOLDMONT_PLUS.
18555 * config/i386/x86-tune.def: Add m_GOLDMONT_PLUS.
18556 * doc/invoke.texi: Add goldmont-plus as x86 -march=/-mtune= CPU type.
18558 2018-05-17 Richard Biener <rguenther@suse.de>
18560 PR tree-optimization/85757
18561 * tree-ssa-dse.c (dse_classify_store): Record a PHI def and
18562 remove defs that only feed that PHI from further processing.
18564 2018-05-16 Jim Wilson <jimw@sifive.com>
18566 * config/riscv/riscv.md (<optab>si3_mask, <optab>si3_mask_1): Prepend
18568 (<optab>di3_mask, <optab>di3_mask_1): Likewise.
18570 2018-05-16 Mark Wielaard <mark@klomp.org>
18572 * dwarf2out.c (count_index_strings): New function.
18573 (output_indirect_strings): Call count_index_strings and generate
18574 header for dwarf_version >= 5.
18576 2018-05-16 Mark Wielaard <mark@klomp.org>
18578 * dwarf2out.c (dwarf_FORM): New function.
18579 (set_indirect_string): Use dwarf_FORM.
18580 (reset_indirect_string): Likewise.
18581 (size_of_die): Likewise.
18582 (value_format): Likewise.
18583 (output_die): Likewise.
18584 (add_skeleton_AT_string): Likewise.
18585 (output_macinfo_op): Likewise.
18586 (index_string): Likewise.
18587 (output_index_string_offset): Likewise.
18588 (output_index_string): Likewise.
18589 (count_index_strings): Likewise.
18591 2018-05-16 Carl Love <cel@us.ibm.com>
18593 * config/rs6000/rs6000.md (prefetch): Generate ISA 2.06 instructions
18594 dcbt and dcbtstt with TH=16 if operands[2] is 0 and Power 8 or newer.
18596 2018-05-16 Martin Jambor <mjambor@suse.cz>
18598 * ipa-prop.c (ipa_free_all_edge_args): Remove.
18599 * ipa-prop.h (ipa_free_all_edge_args): Likewise.
18601 2018-05-16 Wilco Dijkstra <wdijkstr@arm.com>
18603 * config/aarch64/aarch64.md (fma<mode>4): Change into expand pattern.
18604 (fnma<mode>4): Likewise.
18605 (fms<mode>4): Likewise.
18606 (fnms<mode>4): Likewise.
18607 (aarch64_fma<mode>4): Rename insn, reorder accumulator operand.
18608 (aarch64_fnma<mode>4): Likewise.
18609 (aarch64_fms<mode>4): Likewise.
18610 (aarch64_fnms<mode>4): Likewise.
18611 (aarch64_fnmadd<mode>4): Likewise.
18613 2018-05-16 Jason Merrill <jason@redhat.com>
18615 * tree.c (warn_deprecated_use): Return bool. Simplify logic.
18617 2018-05-16 Richard Biener <rguenther@suse.de>
18619 * tree-vectorizer.h (struct stmt_info_for_cost): Add where member.
18620 (dump_stmt_cost): Declare.
18621 (add_stmt_cost): Dump cost we add.
18622 (add_stmt_costs): New function.
18623 (vect_model_simple_cost, vect_model_store_cost, vect_model_load_cost):
18624 No longer exported.
18625 (vect_analyze_stmt): Adjust prototype.
18626 (vectorizable_condition): Likewise.
18627 (vectorizable_live_operation): Likewise.
18628 (vectorizable_reduction): Likewise.
18629 (vectorizable_induction): Likewise.
18630 * tree-vect-loop.c (vect_analyze_loop_operations): Create local
18631 cost vector to pass to vectorizable_ and record afterwards.
18632 (vect_model_reduction_cost): Take cost vector argument and adjust.
18633 (vect_model_induction_cost): Likewise.
18634 (vectorizable_reduction): Likewise.
18635 (vectorizable_induction): Likewise.
18636 (vectorizable_live_operation): Likewise.
18637 * tree-vect-slp.c (vect_create_new_slp_node): Initialize
18638 SLP_TREE_NUMBER_OF_VEC_STMTS.
18639 (vect_analyze_slp_cost_1): Remove.
18640 (vect_analyze_slp_cost): Likewise.
18641 (vect_slp_analyze_node_operations): Take visited args and
18642 a target cost vector. Avoid processing already visited stmt sets.
18643 (vect_slp_analyze_operations): Use a local cost vector to gather
18644 costs and register those of non-discarded instances.
18645 (vect_bb_vectorization_profitable_p): Use add_stmt_costs.
18646 (vect_schedule_slp_instance): Remove copying of
18647 SLP_TREE_NUMBER_OF_VEC_STMTS. Instead assert that it is not
18649 * tree-vect-stmts.c (record_stmt_cost): Remove path directly
18650 adding cost. Record cost entry location.
18651 (vect_prologue_cost_for_slp_op): Function to compute cost of
18652 a constant or invariant generated for SLP vect in the prologue,
18653 split out from vect_analyze_slp_cost_1.
18654 (vect_model_simple_cost): Make static. Adjust for SLP costing.
18655 (vect_model_promotion_demotion_cost): Likewise.
18656 (vect_model_store_cost): Likewise, make static.
18657 (vect_model_load_cost): Likewise.
18658 (vectorizable_bswap): Add cost vector arg and adjust.
18659 (vectorizable_call): Likewise.
18660 (vectorizable_simd_clone_call): Likewise.
18661 (vectorizable_conversion): Likewise.
18662 (vectorizable_assignment): Likewise.
18663 (vectorizable_shift): Likewise.
18664 (vectorizable_operation): Likewise.
18665 (vectorizable_store): Likewise.
18666 (vectorizable_load): Likewise.
18667 (vectorizable_condition): Likewise.
18668 (vectorizable_comparison): Likewise.
18669 (can_vectorize_live_stmts): Likewise.
18670 (vect_analyze_stmt): Likewise.
18671 (vect_transform_stmt): Adjust calls to vectorizable_*.
18672 * tree-vectorizer.c: Include gimple-pretty-print.h.
18673 (dump_stmt_cost): New function.
18675 2018-05-16 Richard Biener <rguenther@suse.de>
18677 * params.def (PARAM_DSE_MAX_ALIAS_QUERIES_PER_STORE): New param.
18678 * doc/invoke.texi (dse-max-alias-queries-per-store): Document.
18679 * tree-ssa-dse.c: Include tree-ssa-loop.h.
18680 (check_name): New callback.
18681 (dse_classify_store): Track cycles via a visited bitmap of PHI
18682 defs and simplify handling of in-loop and across loop dead stores
18683 and properly fail for loop-variant refs. Handle byte-tracking with
18684 multiple defs. Use PARAM_DSE_MAX_ALIAS_QUERIES_PER_STORE for
18687 2018-05-16 Richard Sandiford <richard.sandiford@linaro.org>
18689 * tree-vectorizer.h (vect_get_vector_types_for_stmt): Declare.
18690 (vect_get_mask_type_for_stmt): Likewise.
18691 * tree-vect-slp.c (vect_two_operations_perm_ok_p): New function,
18693 (vect_build_slp_tree_1): ...here. Use vect_get_vector_types_for_stmt
18694 to determine the statement's vector type and the vector type that
18695 should be used for calculating nunits. Deal with cases in which
18696 the type has to be deferred.
18697 (vect_slp_analyze_node_operations): Use vect_get_vector_types_for_stmt
18698 and vect_get_mask_type_for_stmt to calculate STMT_VINFO_VECTYPE.
18699 * tree-vect-loop.c (vect_determine_vf_for_stmt_1)
18700 (vect_determine_vf_for_stmt): New functions, split out from...
18701 (vect_determine_vectorization_factor): ...here.
18702 * tree-vect-stmts.c (vect_get_vector_types_for_stmt)
18703 (vect_get_mask_type_for_stmt): New functions, split out from
18704 vect_determine_vectorization_factor.
18706 2018-05-16 Richard Biener <rguenther@suse.de>
18708 * tree-cfg.c (verify_gimple_assign_ternary): Properly
18709 verify the [VEC_]COND_EXPR embedded comparison.
18711 2018-05-15 Martin Sebor <msebor@redhat.com>
18713 PR tree-optimization/85753
18714 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Handle
18715 RECORD_TYPE in addition to ARRAY_TYPE.
18717 2018-05-15 Martin Sebor <msebor@redhat.com>
18719 PR middle-end/85643
18720 * calls.c (get_attr_nonstring_decl): Handle MEM_REF.
18722 2018-05-15 Richard Biener <rguenther@suse.de>
18724 * tree-ssa-dse.c (dse_classify_store): Remove use_stmt parameter,
18725 add by_clobber_p one. Change algorithm to collect all defs
18726 representing uses we need to walk and try reducing them to
18727 a single one before failing.
18728 (dse_dom_walker::dse_optimize_stmt): Adjust.
18730 2018-05-13 Mark Wielaard <mark@klomp.org>
18732 * dwarf2out.c (dwarf_OP): Handle DW_OP_addrx and DW_OP_constx.
18733 (size_of_loc_descr): Likewise.
18734 (output_loc_operands): Likewise.
18735 (output_loc_operands_raw): Likewise.
18736 (dw_addr_op): Use dwarf_OP () for DW_OP_constx and DW_OP_addrx.
18737 (resolve_addr_in_expr): Handle DW_OP_addrx and DW_OP_constx.
18738 (hash_loc_operands): Likewise.
18739 (compare_loc_operands): Likewise.
18741 2018-05-14 Mark Wielaard <mark@klomp.org>
18743 * dwarf2out.c (count_index_addrs): New function.
18744 (dwarf2out_finish): Use count_index_addrs to calculate addrs_length.
18746 2018-05-15 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
18748 PR tree-optimization/83648
18749 * ipa-pure-const.c (malloc_candidate_p): Allow function with NULL
18750 return value as malloc candidate.
18752 2018-05-15 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
18755 * ipa-pure-const.c (warn_function_malloc): Pass value of known_finite
18756 param as true in call to suggest_attribute.
18758 2018-05-14 Segher Boessenkool <segher@kernel.crashing.org>
18760 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete @itemx for
18761 -mreadonly-in-sdata.
18763 2018-05-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18765 * config/aarch64/aarch64-simd.md (*aarch64_crypto_aese_fused):
18767 (aarch64_crypto_aesd_fused): Likewise.
18769 2018-05-14 Wilco Dijkstra <wdijkstr@arm.com>
18771 * config/aarch64/aarch64.md (mov<mode>): Remove '*' in alternatives.
18772 (movsi_aarch64): Likewise.
18773 (load_pairsi): Likewise.
18774 (load_pairdi): Likewise.
18775 (store_pairsi): Likewise.
18776 (store_pairdi): Likewise.
18777 (load_pairsf): Likewise.
18778 (load_pairdf): Likewise.
18779 (store_pairsf): Likewise.
18780 (store_pairdf): Likewise.
18781 (zero_extend): Likewise.
18782 (trunc): Swap alternatives.
18783 (fcvt_target): Add '?' to prefer w over r.
18785 2018-05-14 Jakub Jelinek <jakub@redhat.com>
18788 * config/i386/i386.md: Disallow non-commutative arithmetics in
18789 last twpeephole for mem {+,-,&,|,^}= x; mem != 0 after cmpelim
18790 optimization. Use COMMUTATIVE_ARITH_P test rather than != MINUS
18791 in the peephole2 before it.
18793 2018-05-14 Sebastian Peryt <sebastian.peryt@intel.com>
18795 * common/config/i386/i386-common.c (OPTION_MASK_ISA_CLDEMOTE_SET,
18796 OPTION_MASK_ISA_CLDEMOTE_UNSET): New defines.
18797 (ix86_handle_option): Handle -mcldemote.
18798 * config.gcc: New header.
18799 * config/i386/cldemoteintrin.h: New file.
18800 * config/i386/cpuid.h (bit_CLDEMOTE): New bit.
18801 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
18803 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
18804 OPTION_MASK_ISA_CLDEMOTE.
18805 * config/i386/i386.c (ix86_target_string): Add -mcldemote.
18806 (ix86_valid_target_attribute_inner_p): Ditto.
18807 (enum ix86_builtins): Add IX86_BUILTIN_CLDEMOTE.
18808 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_cldemote.
18809 (ix86_expand_builtin): Expand IX86_BUILTIN_CLDEMOTE.
18810 * config/i386/i386.h (TARGET_CLDEMOTE, TARGET_CLDEMOTE_P): New.
18811 * config/i386/i386.md (UNSPECV_CLDEMOTE): New.
18813 * config/i386/i386.opt: Add -mcldemote.
18814 * config/i386/x86intrin.h: New header.
18815 * doc/invoke.texi: Add -mcldemote.
18817 2018-05-14 Richard Biener <rguenther@suse.de>
18819 * doc/match-and-simplify.texi: Adjust :s documentation.
18821 2018-05-14 Alexander Monakov <amonakov@ispras.ru>
18823 * sort.cc (REORDER_23): Pass the type for the temporaries instead of
18824 intended memcpy size.
18825 (REORDER_45): Likewise.
18827 2018-05-13 Alexander Monakov <amonakov@ispras.ru>
18829 * sort.cc: New file.
18830 * system.h [!CHECKING_P] (qsort): Redirect to gcc_qsort.
18831 * vec.c (qsort_chk): Use gcc_qsort.
18832 * Makefile.in (OBJS-libcommon): Add sort.o.
18833 (build/sort.o): New target. Use it...
18834 (BUILD_RTL): ... here, and...
18835 (build/gencfn-macros): ... here, and...
18836 (build/genmatch): ... here.
18838 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
18839 Chung-Ju Wu <jasonwucj@gmail.com>
18841 * config.gcc (nds32*-*-*): Check that n15 is valid to --with-cpu.
18842 * config/nds32/nds32-graywolf.md: New file.
18843 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_GRAYWOLF.
18844 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n15
18846 * config/nds32/nds32-protos.h: More declarations for n15 pipeline.
18847 * config/nds32/nds32-utils.c: More implementations for n15 pipeline.
18848 * config/nds32/nds32.md (pipeline_model): Add graywolf.
18849 * config/nds32/nds32.opt (mcpu): Support n15 pipeline cpus.
18850 * config/nds32/pipelines.md: Include n15 settings.
18852 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
18853 Chung-Ju Wu <jasonwucj@gmail.com>
18855 * config.gcc (nds32*-*-*): Check that n12/n13 are valid to --with-cpu.
18856 * config/nds32/nds32-n13.md: New file.
18857 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N12 and CPU_N13.
18858 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n12/n13
18860 * config/nds32/nds32-protos.h: More declarations for n12/n13 pipeline.
18861 * config/nds32/nds32.md (pipeline_model): Add n13.
18862 * config/nds32/nds32.opt (mcpu): Support n12/n13 pipeline cpus.
18863 * config/nds32/pipelines.md: Include n13 settings.
18865 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
18866 Chung-Ju Wu <jasonwucj@gmail.com>
18868 * config.gcc (nds32*-*-*): Check that n10/d10 are valid to --with-cpu.
18869 * config/nds32/nds32-n10.md: New file.
18870 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N10.
18871 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n10
18873 * config/nds32/nds32-protos.h: More declarations for n10 pipeline.
18874 * config/nds32/nds32-utils.c: More implementations for n10 pipeline.
18875 * config/nds32/nds32.md (pipeline_model): Add n10.
18876 * config/nds32/nds32.opt (mcpu): Support n10 pipeline cpus.
18877 * config/nds32/pipelines.md: Include n10 settings.
18879 2018-05-13 Monk Chiang <sh.chiang04@gmail.com>
18880 Kito Cheng <kito.cheng@gmail.com>
18881 Chung-Ju Wu <jasonwucj@gmail.com>
18883 * config.gcc (nds32be-*-*): Handle --with-ext-dsp.
18884 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
18885 Add enum values for DSP extension instructions.
18886 * config/nds32/constraints.md (Iu06, IU06, CVp5, CVs5, CVs2, CVhi):
18888 * config/nds32/iterators.md (shifts, shiftrt, sat_plus, all_plus,
18889 sat_minus, all_minus, plus_minus, extend, sumax, sumin, sumin_max):
18890 New code iterators.
18891 (su, zs, uk, opcode, add_rsub, add_sub): New code attributes.
18892 * config/nds32/nds32-dspext.md: New file for DSP implementation.
18893 * config/nds32/nds32-intrinsic.c: Implementation of DSP extension.
18894 * config/nds32/nds32-intrinsic.md: Likewise.
18895 * config/nds32/nds32_intrinsic.h: Likewise.
18896 * config/nds32/nds32-md-auxiliary.c: Likewise.
18897 * config/nds32/nds32-memory-manipulation.c: Consider DSP extension.
18898 * config/nds32/nds32-predicates.c (const_vector_to_hwint): New.
18899 (nds32_valid_CVp5_p, nds32_valid_CVs5_p): New.
18900 (nds32_valid_CVs2_p, nds32_valid_CVhi_p): New.
18901 * config/nds32/nds32-protos.h: New declarations for DSP extension.
18902 * config/nds32/nds32-utils.c (extract_mac_non_acc_rtx): New case
18903 TYPE_DMAC in switch statement.
18904 * config/nds32/nds32.c: New checking and implementation for DSP
18905 extension instructions.
18906 * config/nds32/nds32.h: Likewise.
18907 * config/nds32/nds32.md: Likewise.
18908 * config/nds32/nds32.opt (mhw-abs, mext-dsp): New options.
18909 * config/nds32/predicates.md: Implement new predicates for DSP
18912 2018-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
18914 * config/rs6000/rs6000.md (mov<mode>_softfloat, FMOVE32):
18915 Reformat alternatives and attributes so it is easier to identify
18916 which constraints/attributes go with which instruction.
18917 (mov<mode>_hardfloat32, FMOVE64): Likewise.
18918 (mov<mode>_softfloat32, FMOVE64): Likewise.
18919 (mov<mode>_hardfloat64, FMOVE64): Likewise.
18920 (mov<mode>_softfloat64, FMOVE64): Likewise.
18922 2018-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
18924 * doc/extend.texi (PowerPC Built-in Functions): Rename this
18926 (Basic PowerPC Built-in Functions): The new name of the
18927 subsection previously known as "PowerPC Built-in Functions".
18928 (Basic PowerPC Built-in Functions Available on all Configurations):
18930 (Basic PowerPC Built-in Functions Available on ISA 2.05): Likewise.
18931 (Basic PowerPC Built-in Functions Available on ISA 2.06): Likewise.
18932 (Basic PowerPC Built-in Functions Available on ISA 2.07): Likewise.
18933 (Basic PowerPC Built-in Functions Available on ISA 3.0): Likewise.
18935 2018-05-11 Martin Jambor <mjambor@suse.cz>
18938 * ipa-cp.c (intersect_with_plats): Check that the lattice contains
18941 2018-05-11 Richard Earnshaw <rearnsha@arm.com>
18944 * config/arm/arm-cpus.in (fgroup ARMv6m): Add be8 feature.
18946 2018-05-11 Sebastian Peryt <sebastian.peryt@intel.com>
18948 * common/config/i386/i386-common.c (OPTION_MASK_ISA_WAITPKG_SET,
18949 OPTION_MASK_ISA_WAITPKG_UNSET): New defines.
18950 (ix86_handle_option): Handle -mwaitpkg.
18951 * config.gcc: New header.
18952 * config/i386/cpuid.h (bit_WAITPKG): New bit.
18953 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mwaitpkg.
18954 * config/i386/i386-builtin-types.def ((UINT8, UNSIGNED, UINT64)): New
18956 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
18957 OPTION_MASK_ISA_WAITPKG.
18958 * config/i386/i386.c (ix86_target_string): Add -mwaitpkg.
18959 (ix86_option_override_internal): Add PTA_WAITPKG.
18960 (ix86_valid_target_attribute_inner_p): Add -mwaitpkg.
18961 (enum ix86_builtins): Add IX86_BUILTIN_UMONITOR, IX86_BUILTIN_UMWAIT,
18962 IX86_BUILTIN_TPAUSE.
18963 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_umonitor,
18964 __builtin_ia32_umwait and __builtin_ia32_tpause.
18965 (ix86_expand_builtin): Expand IX86_BUILTIN_UMONITOR,
18966 IX86_BUILTIN_UMWAIT, IX86_BUILTIN_TPAUSE.
18967 * config/i386/i386.h (TARGET_WAITPKG, TARGET_WAITPKG_P): New.
18968 * config/i386/i386.md (UNSPECV_UMWAIT, UNSPECV_UMONITOR,
18969 UNSPECV_TPAUSE): New.
18970 (umwait, umwait_rex64, umonitor_<mode>, tpause, tpause_rex64): New.
18971 * config/i386/i386.opt: Add -mwaitpkg.
18972 * config/i386/waitpkgintrin.h: New file.
18973 * config/i386/x86intrin.h: New header.
18974 * doc/invoke.texi: Add -mwaitpkg.
18976 2018-05-11 Richard Earnshaw <rearnsha@arm.com>
18979 * config/arm/arm-cpus.in: Add comment that ARMv6-m and ARMv6S-m are now
18981 (cortex-m0): Use armv6s-m isa.
18982 (cortex-m0plus): Likewise.
18983 (cortex-m1): Likewise.
18984 (cortex-m0.small-multiply): Likewise.
18985 (cortex-m0plus.small-multiply): Likewise.
18986 (cortex-m1.small-multiply): Likewise.
18988 2018-05-11 Allan Sandfeld Jensen <allan.jensen@qt.io>
18989 Jakub Jelinek <jakub@redhat.com>
18991 PR tree-optimization/85692
18992 * tree-ssa-forwprop.c (simplify_vector_constructor): Try two
18993 source permute as well.
18995 2018-05-11 Martin Liska <mliska@suse.cz>
18998 * doc/extend.texi: Document LLVM style format for no_sanitize
19001 2018-05-10 Michael Meissner <meissner@linux.vnet.ibm.com>
19003 * config/rs6000/rs6000.c (mode_supports_dq_form): Rename
19004 mode_supports_vsx_dform_quad to mode_supports_dq_form.
19005 (mode_supports_vsx_dform_quad): Likewise.
19006 (mode_supports_vmx_dform): Move these functions to be next to the
19007 other mode_supports functions.
19008 (mode_supports_dq_form): Likewise.
19009 (quad_address_p): Change calls of mode_supports_vsx_dform_quad to
19010 mode_supports_dq_form.
19011 (reg_offset_addressing_ok_p): Likewise.
19012 (offsettable_ok_by_alignment): Likewise.
19013 (rs6000_legitimate_offset_address_p): Likewise.
19014 (legitimate_lo_sum_address_p): Likewise.
19015 (rs6000_legitimize_address): Likewise.
19016 (rs6000_legitimize_reload_address): Likewise.
19017 (rs6000_secondary_reload_inner): Likewise.
19018 (rs6000_preferred_reload_class): Likewise.
19019 (rs6000_output_move_128bit): Likewise.
19021 2018-05-10 Uros Bizjak <ubizjak@gmail.com>
19023 * config/i386/i386.c (ix86_expand_builtin) <case IX86_BUILTIN_RDPID>:
19024 Generate SImode target register for null target.
19025 <case IX86_BUILTIN_XGETBV>: Ditto.
19026 <case IX86_BUILTIN_XSETBV>: Optimize LSHIFTRT generation.
19027 * config/i386/xsaveintrin.h (_xgetbv): Add missing return.
19029 2018-05-10 Carl Love <cel@us.ibm.com>
19031 * config/rs6000/rs6000.md (prefetch): Generate ISA 2.06 instructions
19032 dcbtt and dcbtstt if operands[2] is 0.
19034 2018-05-10 Uros Bizjak <ubizjak@gmail.com>
19037 * config/i386/sse.md (usadv64qi): New expander.
19039 2018-05-10 Segher Boessenkool <segher@kernel.crashing.org>
19041 * config/rs6000/altivec.md (altivec_vmrghb, altivec_vmrghh,
19042 altivec_vmrghw, altivec_vmrglb, altivec_vmrglh, altivec_vmrglw): Remove
19043 -maltivec=be support.
19044 (vec_widen_umult_even_v16qi, vec_widen_smult_even_v16qi,
19045 vec_widen_umult_even_v8hi, vec_widen_smult_even_v8hi,
19046 vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
19047 vec_widen_umult_odd_v16qi, vec_widen_smult_odd_v16qi,
19048 vec_widen_umult_odd_v8hi, vec_widen_smult_odd_v8hi,
19049 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si, altivec_vpkpx,
19050 altivec_vpks<VI_char>ss, altivec_vpks<VI_char>us,
19051 altivec_vpku<VI_char>us, altivec_vpku<VI_char>um, altivec_vsum2sws,
19052 altivec_vsumsws): Adjust.
19053 (altivec_vspltb *altivec_vspltb_internal, altivec_vsplth,
19054 *altivec_vsplth_internal, altivec_vspltw, *altivec_vspltw_internal,
19055 altivec_vspltsf, *altivec_vspltsf_internal): Remove -maltivec=be
19057 (altivec_vperm_<mode>, altivec_vperm_<mode>_uns,
19058 altivec_vupkhs<VU_char>, altivec_vupkls<VU_char>, altivec_vupkhpx,
19059 altivec_vupklpx, altivec_lvsl, altivec_lvsr): Adjust.
19060 (altivec_lve<VI_char>x): Delete expand.
19061 (*altivec_lve<VI_char>x_internal): Rename to...
19062 (altivec_lve<VI_char>x): ... this.
19063 (altivec_lvxl_<mode>): Delete expand.
19064 (*altivec_lvxl_<mode>_internal): Rename to ...
19065 (altivec_lvxl_<mode>): ... this.
19066 (altivec_stvxl_<mode>): Delete expand.
19067 (*altivec_stvxl_<mode>_internal): Rename to ...
19068 (altivec_stvxl_<mode>): ... this.
19069 (altivec_stve<VI_char>x): Delete expand.
19070 (*altivec_stve<VI_char>x_internal): Rename to ...
19071 (altivec_stve<VI_char>x): ... this.
19072 (doublee<mode>2, unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2,
19073 doubleh<mode>2, unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2,
19074 reduc_plus_scal_<mode>): Adjust.
19075 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Adjust
19077 (rs6000_cpu_cpp_builtins): Adjust.
19078 (altivec_resolve_overloaded_builtin): Remove -maltivec=be support.
19079 * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be,
19080 altivec_expand_stvx_be, altivec_expand_stvex_be): Delete.
19081 * config/rs6000/rs6000.c (rs6000_option_override_internal): Remove
19082 -maltivec=be support.
19083 (rs6000_split_vec_extract_var): Adjust.
19084 (rs6000_split_v4si_init): Adjust.
19085 (swap_selector_for_mode): Delete.
19086 (altivec_expand_lvx_be, altivec_expand_stvx_be,
19087 altivec_expand_stvex_be): Delete.
19088 (altivec_expand_lv_builtin, altivec_expand_stv_builtin): Remove
19089 -maltivec=be support.
19090 (rs6000_gimple_fold_builtin): Ditto.
19091 (rs6000_generate_float2_double_code, rs6000_generate_float2_code):
19093 * config/rs6000/rs6000.h (VECTOR_ELT_ORDER_BIG): Delete.
19094 (TARGET_DIRECT_MOVE_64BIT): Adjust.
19095 * config/rs6000/rs6000.md (split for extendsidi2 for vectors): Adjust.
19096 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Delete.
19097 * config/rs6000/vsx.md (floate<mode>, unsfloatev2di, floato<mode>,
19098 unsfloatov2di, vsignedo_v2df, vsignede_v2df, vunsignedo_v2df,
19099 vunsignede_v2df, vsx_extract_<mode>_p9, *vsx_extract_si,
19100 *vsx_extract_<mode>_p8, *vsx_extract_si_<uns>float_df,
19101 *vsx_extract_si_<uns>float_<mode>, vsx_set_<mode>_p9, vsx_set_v4sf_p9,
19102 *vsx_insert_extract_v4sf_p9, *vsx_insert_extract_v4sf_p9_2, and an
19103 anonymous split): Adjust.
19104 (vsx_mergel_<mode>, vsx_mergeh_<mode>): Remove -maltivec=be support.
19105 (vsx_xxspltd_<mode>, extract4b, insert4b): Adjust.
19107 2018-05-10 Eric Botcazou <ebotcazou@adacore.com>
19109 * configure.ac (gcc_gxx_include_dir_add_sysroot): Set it to 1 only
19110 when --with-gxx-include-dir is also specified.
19111 * configure: Regenerate.
19113 2018-05-09 Jim Wilson <jimw@sifive.com>
19116 * config.gcc (riscv*-*-*): Handle --with-multilib-list.
19117 * config/riscv/t-withmultilib: New.
19118 * config/riscv/withmultilib.h: New.
19119 * doc/install.texi: Document RISC-V --with-multilib-list support.
19121 2018-05-09 Richard Biener <rguenther@suse.de>
19123 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Fill a cost
19125 (vect_bb_vectorization_profitable_p): Adjust. Compute
19126 actual scalar cost using the cost vector and the add_stmt_cost
19129 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
19131 PR rtl-optimization/85645
19132 * config/rs6000/rs6000.c (rs6000_emit_prologue_components): Put a SET
19133 in the REG_CFA_REGISTER note for LR, don't leave it empty.
19135 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
19137 PR rtl-optimization/85645
19138 * shrink-wrap.c (spread_components): Return a boolean saying if
19139 anything was changed.
19140 (try_shrink_wrapping_separate): Iterate spread_components until
19141 nothing changes anymore.
19143 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
19145 PR rtl-optimization/85645
19146 * regrename.c (build_def_use): Also kill the chains that include the
19147 destination of a REG_CFA_REGISTER note.
19149 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
19151 PR rtl-optimization/85645
19152 * regcprop.c (copyprop_hardreg_forward_1): Don't propagate into an
19153 insn that has a REG_CFA_REGISTER note.
19155 2018-05-09 Richard Sandiford <richard.sandiford@linaro.org>
19157 * cfgexpand.c (expand_clobber): New function.
19158 (expand_gimple_stmt_1): Use it.
19159 * tree-vect-stmts.c (vect_clobber_variable): New function,
19161 (vectorizable_simd_clone_call): ...here.
19162 (vectorizable_store): Emit a clobber either side of an
19163 IFN_STORE_LANES sequence.
19164 (vectorizable_load): Emit a clobber after an IFN_LOAD_LANES sequence.
19166 2018-05-09 Tom de Vries <tom@codesourcery.com>
19169 * config/nvptx/nvptx.md (define_insn "trap", define_insn "trap_if_true")
19170 (define_insn "trap_if_false"): Add exit after trap.
19172 2018-05-09 Eric Botcazou <ebotcazou@adacore.com>
19174 PR rtl-optimization/85638
19175 * bb-reorder.c: Include common/common-target.h.
19176 (create_forwarder_block): New function extracted from...
19177 (fix_up_crossing_landing_pad): ...here. Rename into...
19178 (dw2_fix_up_crossing_landing_pad): ...this.
19179 (sjlj_fix_up_crossing_landing_pad): New function.
19180 (find_rarely_executed_basic_blocks_and_crossing_edges): In SJLJ mode,
19181 call sjlj_fix_up_crossing_landing_pad if there are incoming EH edges
19182 from both partitions and exit the loop after one iteration.
19184 2018-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
19187 * doc/extend.texi (PowerPC Built-in Functions): Rename this
19189 (Basic PowerPC Built-in Functions): The new name of the
19190 subsection previously known as "PowerPC Built-in Functions".
19191 (Basic PowerPC Built-in Functions Available on all Configurations):
19193 (Basic PowerPC Built-in Functions Available on ISA 2.05): New
19195 (Basic PowerPC Built-in Functions Available on ISA 2.06): New
19197 (Basic PowerPC Built-in Functions Available on ISA 2.07): New
19199 (Basic PowerPC Built-in Functions Available on ISA 3.0): New
19202 2018-05-08 Jim Wilson <jimw@sifive.com>
19204 * config/riscv/linux.h (MUSL_ABI_SUFFIX): Delete unnecessary backslash.
19205 (LD_EMUL_SUFFIX): New.
19206 (LINK_SPEC): Use it.
19208 2018-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
19210 * doc/extend.texi (PowerPC Built-in Functions): Rename this
19212 (Basic PowerPC Built-in Functions): The new name of the
19213 subsection previously known as "PowerPC Built-in Functions".
19214 (Basic PowerPC Built-in Functions Available on all Configurations):
19216 (Basic PowerPC Built-in Functions Available on ISA 2.05): New
19218 (Basic PowerPC Built-in Functions Available on ISA 2.06): New
19220 (Basic PowerPC Built-in Functions Available on ISA 2.07): New
19222 (Basic PowerPC Built-in Functions Available on ISA 3.0): New
19225 2018-05-08 Jakub Jelinek <jakub@redhat.com>
19228 * config/i386/i386.md: Add peepholes for mem {+,-,&,|,^}= x; mem != 0
19229 after cmpelim optimization.
19231 2018-05-08 Olga Makhotina <olga.makhotina@intel.com>
19233 * config.gcc: Support "goldmont".
19234 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "goldmont".
19235 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
19236 PROCESSOR_GOLDMONT.
19237 * config/i386/i386.c (m_GOLDMONT): Define.
19238 (processor_target_table): Add "goldmont".
19239 (PTA_GOLDMONT): Define.
19240 (ix86_lea_outperforms): Add TARGET_GOLDMONT.
19241 (get_builtin_code_for_version): Handle PROCESSOR_GOLDMONT.
19242 (fold_builtin_cpu): Add M_INTEL_GOLDMONT.
19243 (fold_builtin_cpu): Add "goldmont".
19244 (ix86_add_stmt_cost): Add TARGET_GOLDMONT.
19245 (ix86_option_override_internal): Add "goldmont".
19246 * config/i386/i386.h (processor_costs): Define TARGET_GOLDMONT.
19247 (processor_type): Add PROCESSOR_GOLDMONT.
19248 * config/i386/i386.md: Add CPU "glm".
19249 * config/i386/glm.md: New file.
19250 * config/i386/x86-tune.def: Add m_GOLDMONT.
19251 * doc/invoke.texi: Add goldmont as x86 -march=/-mtune= CPU type.
19253 2018-05-08 Jakub Jelinek <jakub@redhat.com>
19256 * config/i386/i386.c (ix86_expand_sse2_abs): Handle E_V2DImode and
19258 * config/i386/sse.md (abs<mode>2): Use VI_AVX2 iterator instead of
19259 VI1248_AVX512VL_AVX512BW. Handle V2DImode and V4DImode if not
19260 TARGET_AVX512VL using ix86_expand_sse2_abs. Formatting fixes.
19263 * config/i386/i386.c (ix86_fold_builtin): Handle
19264 IX86_BUILTIN_{,P}MOVMSK{PS,PD,B}{,128,256}.
19267 * config/i386/sse.md (ssequaterinsnmode): New mode attribute.
19268 (*<extract_type>_vinsert<shuffletype><extract_suf>_0): New pattern.
19270 2018-05-08 Richard Earnshaw <rearnsha@arm.com>
19273 * config/arm/parsecpu.awk (check_cpu): Fix operator precedence.
19274 (check_arch): Likewise.
19275 (check_fpu): Return the result rather than printing it.
19276 (end arch): Fix operator precedence.
19277 (end cpu): Likewise.
19278 (END): Print the result from check_fpu.
19280 2018-05-08 Richard Sandiford <richard.sandiford@linaro.org>
19281 Alan Hayward <alan.hayward@arm.com>
19282 David Sherwood <david.sherwood@arm.com>
19284 * config/aarch64/aarch64-sve.md (*pred_cmp<cmp_op><mode>_combine)
19285 (*pred_cmp<cmp_op><mode>, *fcm<cmp_op><mode>_and_combine)
19286 (*fcmuo<mode>_and_combine, *fcm<cmp_op><mode>_and)
19287 (*fcmuo<mode>_and): New patterns.
19289 2018-05-08 Richard Sandiford <richard.sandiford@linaro.org>
19291 * config/aarch64/iterators.md (UNSPEC_COND_LO, UNSPEC_COND_LS)
19292 (UNSPEC_COND_HI, UNSPEC_COND_HS, UNSPEC_COND_UO): Delete.
19293 (SVE_INT_CMP, SVE_FP_CMP): New code iterators.
19294 (cmp_op, sve_imm_con): New code attributes.
19295 (SVE_COND_INT_CMP, imm_con): Delete.
19296 (cmp_op): Remove above unspecs from int attribute.
19297 * config/aarch64/aarch64-sve.md (*vec_cmp<cmp_op>_<mode>): Rename
19299 (*cmp<cmp_op><mode>): ...this. Use UNSPEC_MERGE_PTRUE instead of
19300 comparison-specific unspecs.
19301 (*vec_cmp<cmp_op>_<mode>_ptest): Rename to...
19302 (*cmp<cmp_op><mode>_ptest): ...this and adjust likewise.
19303 (*vec_cmp<cmp_op>_<mode>_cc): Rename to...
19304 (*cmp<cmp_op><mode>_cc): ...this and adjust likewise.
19305 (*vec_fcm<cmp_op><mode>): Rename to...
19306 (*fcm<cmp_op><mode>): ...this and adjust likewise.
19307 (*vec_fcmuo<mode>): Rename to...
19308 (*fcmuo<mode>): ...this and adjust likewise.
19309 (*pred_fcm<cmp_op><mode>): New pattern.
19310 * config/aarch64/aarch64.c (aarch64_emit_unop, aarch64_emit_binop)
19311 (aarch64_emit_sve_ptrue_op, aarch64_emit_sve_ptrue_op_cc): New
19313 (aarch64_unspec_cond_code): Remove handling of LTU, GTU, LEU, GEU
19315 (aarch64_gen_unspec_cond, aarch64_emit_unspec_cond): Delete.
19316 (aarch64_emit_sve_predicated_cond): New function.
19317 (aarch64_expand_sve_vec_cmp_int): Use aarch64_emit_sve_ptrue_op_cc.
19318 (aarch64_emit_unspec_cond_or): Replace with...
19319 (aarch64_emit_sve_or_conds): ...this new function. Use
19320 aarch64_emit_sve_ptrue_op for the individual comparisons and
19321 aarch64_emit_binop to OR them together.
19322 (aarch64_emit_inverted_unspec_cond): Replace with...
19323 (aarch64_emit_sve_inverted_cond): ...this new function. Use
19324 aarch64_emit_sve_ptrue_op for the comparison and
19325 aarch64_emit_unop to invert the result.
19326 (aarch64_expand_sve_vec_cmp_float): Update after the above
19327 changes. Use aarch64_emit_sve_ptrue_op for native comparisons.
19329 2018-05-07 Nathan Sidwell <nathan@acm.org>
19331 * doc/invoke.texi (C++ Dialect Options): Remove -ffor-scope.
19332 * doc/extend.texi (Deprecated Features): Remove -fno-for-scope
19333 (Backwards Compatibility): Likewise.
19335 2018-05-07 Luis Machado <luis.machado@linaro.org>
19339 2018-05-07 Luis Machado <luis.machado@linaro.org>
19341 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
19342 <prefetch_dynamic_strides>: New const bool field.
19343 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
19344 prefetch_dynamic_strides.
19345 (exynosm1_prefetch_tune): Likewise.
19346 (thunderxt88_prefetch_tune): Likewise.
19347 (thunderx_prefetch_tune): Likewise.
19348 (thunderx2t99_prefetch_tune): Likewise.
19349 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides
19351 (aarch64_override_options_internal): Update to set
19352 PARAM_PREFETCH_DYNAMIC_STRIDES.
19353 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
19354 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
19355 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
19356 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
19357 prefetch-dynamic-strides setting.
19359 2018-05-07 Luis Machado <luis.machado@linaro.org>
19361 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
19362 <minimum_stride>: New const int field.
19363 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
19364 minimum_stride field.
19365 (exynosm1_prefetch_tune): Likewise.
19366 (thunderxt88_prefetch_tune): Likewise.
19367 (thunderx_prefetch_tune): Likewise.
19368 (thunderx2t99_prefetch_tune): Likewise.
19369 (qdf24xx_prefetch_tune): Likewise. Set minimum_stride to 2048.
19370 (aarch64_override_options_internal): Update to set
19371 PARAM_PREFETCH_MINIMUM_STRIDE.
19372 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
19373 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
19374 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
19375 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
19376 stride is constant and is below the minimum stride threshold.
19378 2018-05-07 Luis Machado <luis.machado@linaro.org>
19380 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune) <l2_cache_size>: Set
19383 2018-05-07 Luis Machado <luis.machado@linaro.org>
19385 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
19386 <prefetch_dynamic_strides>: New const bool field.
19387 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
19388 prefetch_dynamic_strides.
19389 (exynosm1_prefetch_tune): Likewise.
19390 (thunderxt88_prefetch_tune): Likewise.
19391 (thunderx_prefetch_tune): Likewise.
19392 (thunderx2t99_prefetch_tune): Likewise.
19393 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides
19395 (aarch64_override_options_internal): Update to set
19396 PARAM_PREFETCH_DYNAMIC_STRIDES.
19397 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
19398 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
19399 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
19400 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
19401 prefetch-dynamic-strides setting.
19403 2018-05-07 Luis Machado <luis.machado@linaro.org>
19405 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
19406 <minimum_stride>: New const int field.
19407 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
19408 minimum_stride field.
19409 (exynosm1_prefetch_tune): Likewise.
19410 (thunderxt88_prefetch_tune): Likewise.
19411 (thunderx_prefetch_tune): Likewise.
19412 (thunderx2t99_prefetch_tune): Likewise.
19413 (qdf24xx_prefetch_tune): Likewise. Set minimum_stride to 2048.
19414 (aarch64_override_options_internal): Update to set
19415 PARAM_PREFETCH_MINIMUM_STRIDE.
19416 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
19417 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
19418 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
19419 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
19420 stride is constant and is below the minimum stride threshold.
19422 2018-05-06 Jakub Jelinek <jakub@redhat.com>
19425 * cfgexpand.c (expand_asm_stmt): Don't create a temporary if
19426 the type is addressable. Don't force op into register if it has
19429 2018-05-05 Roland McGrath <mcgrathr@google.com>
19432 * varasm.c (default_section_type_flags): Set SECTION_NOTYPE for
19433 any section for which we don't know a specific type it should have,
19434 regardless of name. Previously this was done only for the exact
19435 names ".init_array", ".fini_array", and ".preinit_array".
19436 (default_elf_asm_named_section): Add comment about
19437 relationship with default_section_type_flags and SECTION_NOTYPE.
19438 (get_section): Don't consider it a type conflict if one side has
19439 SECTION_NOTYPE and the other doesn't, as long as neither has the
19440 SECTION_BSS et al used in the default_section_type_flags logic.
19442 2018-05-05 Tom de Vries <tom@codesourcery.com>
19445 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG_3): Define.
19446 (workaround_barsyncs): New function.
19447 (nvptx_reorg): Use workaround_barsyncs.
19448 * config/nvptx/nvptx.md (define_c_enum "unspecv"): Add UNSPECV_MEMBAR.
19449 (define_expand "nvptx_membar_cta"): New define_expand.
19450 (define_insn "*nvptx_membar_cta"): New insn.
19452 2018-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
19454 * brig-builtins.def: Add consts to ptrs etc. in BRIG builtin defs.
19455 To improve optimization opportunities.
19456 * builtin-types.def: The new needed builtin types for the above.
19458 2018-05-04 Richard Biener <rguenther@suse.de>
19460 * bb-reorder.c (sanitize_hot_paths): Release hot_bbs_to_check.
19461 * gimple-ssa-store-merging.c
19462 (imm_store_chain_info::output_merged_store): Remove redundant create,
19463 release split_store vector contents on failure.
19464 * tree-vect-slp.c (vect_schedule_slp_instance): Avoid leaking
19465 scalar stmt vector on cache hit.
19467 2018-05-04 Segher Boessenkool <segher@kernel.crashing.org>
19469 * common/config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
19471 * config.gcc (powerpc-xilinx-eabi*): Remove.
19472 * config/rs6000/predicates.md (easy_fp_constant): Remove Xilinx FP
19474 (fusion_addis_mem_combo_load): Ditto.
19475 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove Xilinx
19477 (rs6000_cpu_cpp_builtins): Ditto.
19478 * config/rs6000/rs6000-linux.c
19479 (rs6000_linux_float_exceptions_rounding_supported_p): Ditto.
19480 * config/rs6000/rs6000-opts.h (enum fpu_type_t): Delete.
19481 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove Xilinx FP
19483 (rs6000_setup_reg_addr_masks): Ditto.
19484 (rs6000_init_hard_regno_mode_ok): Ditto.
19485 (rs6000_option_override_internal): Ditto.
19486 (legitimate_lo_sum_address_p): Ditto.
19487 (rs6000_legitimize_address): Ditto.
19488 (rs6000_legitimize_reload_address): Ditto.
19489 (rs6000_legitimate_address_p): Ditto.
19490 (abi_v4_pass_in_fpr): Ditto.
19491 (setup_incoming_varargs): Ditto.
19492 (rs6000_gimplify_va_arg): Ditto.
19493 (rs6000_split_multireg_move): Ditto.
19494 (rs6000_savres_strategy): Ditto.
19495 (rs6000_emit_prologue_components): Ditto.
19496 (rs6000_emit_epilogue_components): Ditto.
19497 (rs6000_emit_prologue): Ditto.
19498 (rs6000_emit_epilogue): Ditto.
19499 (rs6000_elf_file_end): Ditto.
19500 (rs6000_function_value): Ditto.
19501 (rs6000_libcall_value): Ditto.
19502 * config/rs6000/rs6000.h: Ditto.
19503 (TARGET_MINMAX_SF, TARGET_MINMAX_DF): Delete, merge to ...
19504 (TARGET_MINMAX): ... this. New.
19505 (TARGET_SF_FPR, TARGET_DF_FPR, TARGET_SF_INSN, TARGET_DF_INSN): Delete.
19506 * config/rs6000/rs6000.md: Remove Xilinx FP support.
19507 (*movsi_internal1_single): Delete.
19508 * config/rs6000/rs6000.opt (msingle-float, mdouble-float, msimple-fpu,
19509 mfpu=, mxilinx-fpu): Delete.
19510 * config/rs6000/singlefp.h: Delete.
19511 * config/rs6000/sysv4.h: Remove Xilinx FP support.
19512 * config/rs6000/t-rs6000: Ditto.
19513 * config/rs6000/t-xilinx: Delete.
19514 * config/rs6000/titan.md: Adjust for fp_type removal.
19515 * config/rs6000/vsx.md: Remove Xilinx FP support.
19516 (VStype_simple): Delete.
19517 (VSfptype_simple, VSfptype_mul, VSfptype_div, VSfptype_sqrt): Delete.
19518 * config/rs6000/xfpu.h: Delete.
19519 * config/rs6000/xfpu.md: Delete.
19520 * config/rs6000/xilinx.h: Delete.
19521 * config/rs6000/xilinx.opt: Delete.
19522 * doc/invoke.texi (RS/6000 and PowerPC Options): Remove
19523 -msingle-float, -mdouble-float, -msimple-fpu, -mfpu=, and -mxilinx-fpu.
19525 2018-05-04 Tom de Vries <tom@codesourcery.com>
19528 * builtins.c (expand_builtin_goacc_parlevel_id_size): Handle null target
19531 2018-05-04 Richard Biener <rguenther@suse.de>
19533 PR middle-end/85627
19534 * tree-complex.c (update_complex_assignment): We are always in SSA form.
19535 (expand_complex_div_wide): Likewise.
19536 (expand_complex_operations_1): Likewise.
19537 (expand_complex_libcall): Preserve EH info of the original stmt.
19538 (tree_lower_complex): Handle removed blocks.
19539 * tree.c (build_common_builtin_nodes): Do not set ECF_NOTRHOW
19540 on complex multiplication and division libcall builtins.
19542 2018-05-04 Richard Biener <rguenther@suse.de>
19544 PR middle-end/85574
19545 * fold-const.c (negate_expr_p): Restrict negation of operand
19546 zero of a division to when we know that can happen without
19548 (fold_negate_expr_1): Likewise.
19550 2018-05-04 Jakub Jelinek <jakub@redhat.com>
19553 * real.h (real_nextafter): Declare.
19554 * real.c (real_nextafter): New function.
19555 * fold-const-call.c (fold_const_nextafter): New function.
19556 (fold_const_call_sss): Call it for CASE_CFN_NEXTAFTER and
19557 CASE_CFN_NEXTTOWARD.
19558 (fold_const_call_1): For CASE_CFN_NEXTTOWARD call fold_const_call_sss
19559 even when arg1_mode is different from arg0_mode.
19561 2018-05-03 Nathan Sidwell <nathan@acm.org>
19563 * doc/extend.texi (Deprecated Features): Remove
19564 -ffriend-injection.
19565 (Backwards Compatibility): Likewise.
19566 * doc/invoke.texi (C++ Language Options): Likewise.
19567 (C++ Dialect Options): Likewise.
19569 2018-05-03 Jakub Jelinek <jakub@redhat.com>
19572 * config/i386/avx512fintrin.h (_mm512_mullox_epi64,
19573 _mm512_mask_mullox_epi64): New intrinsics.
19575 2018-05-03 Tom de Vries <tom@codesourcery.com>
19578 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
19579 dump files): Add offload-tree.
19581 2018-05-03 Richard Biener <rguenther@suse.de>
19583 PR tree-optimization/85615
19584 * tree-ssa-threadupdate.c (thread_block_1): Only allow exits
19585 to loops not nested in BBs loop father to avoid creating multi-entry
19588 2018-05-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19590 PR tree-optimization/70291
19591 * tree-complex.c (expand_complex_libcall): Add type, inplace_p
19592 arguments. Change return type to tree. Emit libcall as a new
19593 statement rather than replacing existing one when inplace_p is true.
19594 (expand_complex_multiplication_components): New function.
19595 (expand_complex_multiplication): Expand floating-point complex
19596 multiplication using the above.
19597 (expand_complex_division): Rename inner_type parameter to type.
19598 Update expand_complex_libcall call-site.
19599 (expand_complex_operations_1): Update expand_complex_multiplication
19600 and expand_complex_division call-sites.
19602 2018-05-02 Jakub Jelinek <jakub@redhat.com>
19605 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask,
19606 *ashl<dwi>3_doubleword_mask_1, *<shift_insn><dwi>3_doubleword_mask,
19607 *<shift_insn><dwi>3_doubleword_mask_1): In condition require that
19608 the highest significant bit of the shift count mask is clear. In
19609 check whether and[sq]i3 is needed verify that all significant bits
19610 of the shift count other than the highest are set.
19612 2018-05-02 Tom de Vries <tom@codesourcery.com>
19615 * builtins.def (DEF_GOACC_BUILTIN_ONLY): Define.
19616 * omp-builtins.def (BUILT_IN_GOACC_PARLEVEL_ID)
19617 (BUILT_IN_GOACC_PARLEVEL_SIZE): New builtin.
19618 * builtins.c (expand_builtin_goacc_parlevel_id_size): New function.
19619 (expand_builtin): Call expand_builtin_goacc_parlevel_id_size.
19620 * doc/extend.texi (Other Builtins): Add __builtin_goacc_parlevel_id and
19621 __builtin_goacc_parlevel_size.
19623 2018-05-02 Richard Biener <rguenther@suse.de>
19625 PR tree-optimization/85597
19626 * tree-vect-stmts.c (vectorizable_operation): For ternary SLP
19627 do not use split vect_get_vec_defs call but call vect_get_slp_defs
19630 2018-05-02 Tom de Vries <tom@codesourcery.com>
19633 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
19634 dump files): Add ltrans-tree.
19636 2018-05-02 Tom de Vries <tom@codesourcery.com>
19639 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
19640 dump files): Add wpa-ipa.
19642 2018-05-02 Segher Boessenkool <segher@kernel.crashing.org>
19644 * config.gcc (powerpc*-*-*): Remove paired.h. Unsupport the
19645 powerpc*-*-linux*paired* target.
19646 * config/rs6000/750cl.h: Delete.
19647 * config/rs6000/paired.h: Delete.
19648 * config/rs6000/paired.md: Delete.
19649 * config/rs6000/predicates.md (easy_vector_constant): Remove paired
19651 * config/rs6000/rs6000-builtin.def: Remove paired float support.
19652 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
19653 comment. Remove paired float support.
19654 * config/rs6000/rs6000-modes.def: Remove V2SF and V2SI.
19655 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete
19657 * config/rs6000/rs6000-protos.h (paired_expand_vector_init,
19658 paired_emit_vector_cond_expr, paired_expand_vector_move): Delete
19660 * config/rs6000/rs6000.c: Remove paired float support.
19661 (paired_expand_vector_init, paired_expand_vector_move,
19662 paired_emit_vector_compare, paired_emit_vector_cond_expr,
19663 (paired_expand_lv_builtin, paired_expand_stv_builtin,
19664 paired_expand_builtin, paired_expand_predicate_builtin,
19665 paired_init_builtins): Delete.
19666 * config/rs6000/rs6000.h: Remove paired float support.
19667 * config/rs6000/rs6000.md: Remove paired float support.
19668 (move_from_CR_ov_bit): Delete.
19669 * config/rs6000/rs6000.opt (mpaired): Delete.
19670 * config/rs6000/t-rs6000: Remove paired.md from MD_INCLUDES.
19671 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mpaired.
19673 2018-05-02 Richard Biener <rguenther@suse.de>
19675 PR middle-end/85567
19676 * gimplify.c (gimplify_save_expr): When in SSA form allow
19677 SAVE_EXPRs to compute to SSA vars.
19679 2018-05-02 Jakub Jelinek <jakub@redhat.com>
19682 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask,
19683 *ashl<dwi>3_doubleword_mask_1, *<shift_insn><dwi>3_doubleword_mask,
19684 *<shift_insn><dwi>3_doubleword_mask_1): If and[sq]i3 is needed, don't
19685 clobber operands[2], instead use a new pseudo. Formatting fixes.
19687 2018-05-02 Richard Sandiford <richard.sandiford@linaro.org>
19689 PR tree-optimization/85586
19690 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Only
19691 exit early for statements in the same group if the accesses are
19694 2018-05-02 Tom de Vries <tom@codesourcery.com>
19697 * lto-wrapper.c (compile_offload_image): Add "could not find mkoffload"
19700 2018-05-01 Marc Glisse <marc.glisse@inria.fr>
19702 PR tree-optimization/85143
19703 * match.pd (A<B&A<C): Extend to BIT_IOR_EXPR.
19705 2018-05-01 Tom de Vries <tom@codesourcery.com>
19708 * config/nvptx/mkoffload.c (main): Suggest using -B in "offload compiler
19709 not found" error message.
19711 2018-05-01 Tom de Vries <tom@codesourcery.com>
19714 * vec.h (VEC_ORDERED_REMOVE_IF, VEC_ORDERED_REMOVE_IF_FROM_TO): Define.
19715 * vec.c (test_ordered_remove_if): New function.
19716 (vec_c_tests): Call test_ordered_remove_if.
19717 * dwarf2cfi.c (connect_traces): Use VEC_ORDERED_REMOVE_IF_FROM_TO.
19718 * lto-streamer-out.c (prune_offload_funcs): Use VEC_ORDERED_REMOVE_IF.
19719 * tree-vect-patterns.c (vect_pattern_recog_1): Use
19720 VEC_ORDERED_REMOVE_IF.
19722 2018-05-01 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
19724 PR tree-optimization/82665
19725 * vr-values.c (vr_values::extract_range_from_binary_expr): Handle
19726 pointer subtraction where arguments come from a memchr call.
19728 2018-05-01 Jakub Jelinek <jakub@redhat.com>
19730 * configure.ac (LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION): Use
19731 --push-state --as-needed and --pop-state instead of --as-needed and
19732 --no-as-needed if ld supports it.
19733 * configure: Regenerated.
19736 * doc/install.texi2html: Replace _002d with - and _002a with * in
19737 generated html files using sed.
19739 2018-04-30 David Malcolm <dmalcolm@redhat.com>
19742 * gcc-rich-location.c (blank_line_before_p): New function.
19743 (use_new_line): New function.
19744 (gcc_rich_location::add_fixit_insert_formatted): New function.
19745 * gcc-rich-location.h
19746 (gcc_rich_location::add_fixit_insert_formatted): New function.
19748 2018-04-30 David Malcolm <dmalcolm@redhat.com>
19750 * selftest.c (assert_streq): Rename "expected" and "actual" to
19751 "val1" and "val2". Extend NULL-handling to cover both inputs
19752 symmetrically, while still requiring both to be non-NULL for a pass.
19753 * selftest.h (assert_streq): Rename "expected" and "actual" to
19755 (ASSERT_EQ): Likewise.
19756 (ASSERT_EQ_AT): Likewise.
19757 (ASSERT_KNOWN_EQ): Likewise.
19758 (ASSERT_KNOWN_EQ_AT): Likewise.
19759 (ASSERT_NE): Likewise.
19760 (ASSERT_MAYBE_NE): Likewise.
19761 (ASSERT_MAYBE_NE_AT): Likewise.
19762 (ASSERT_STREQ): Likewise. Clarify that both must be non-NULL for
19763 the assertion to pass.
19764 (ASSERT_STREQ_AT): Likewise.
19766 2018-04-30 Jonathan Wakely <jwakely@redhat.com>
19768 * doc/invoke.texi (-fpie, -fPIE): Fix grammar and clarify
19769 interaction with -pie.
19771 2018-04-30 David Malcolm <dmalcolm@redhat.com>
19773 * selftest.h: Fix alphabetization of per-source-file selftest
19776 2018-04-30 Jason Merrill <jason@redhat.com>
19778 PR c++/61982 - dead stores to destroyed objects.
19779 * gimplify.c (gimplify_modify_expr): Simplify complex lvalue on LHS
19782 2018-04-30 Jason Merrill <jason@redhat.com>
19784 * tree.c (build_clobber): New.
19785 * tree.h: Declare it.
19786 * gimplify.c (gimplify_bind_expr, gimplify_target_expr): Use it.
19788 2018-04-30 David Malcolm <dmalcolm@redhat.com>
19790 * diagnostic-show-locus.c (layout::layout): Update for
19791 location_get_source_line returning a char_span.
19792 (struct char_span): Move to input.h.
19793 (struct correction): Update for fields in char_span becoming
19795 (struct source_line): Update for location_get_source_line
19796 returning a char_span.
19797 (layout::print_line): Likewise.
19798 * edit-context.c (edited_file::print_content): Likewise.
19799 (edited_file::print_diff_hunk): Likewise.
19800 (edited_file::print_run_of_changed_lines): Likewise.
19801 (edited_file::get_num_lines): Likewise.
19802 (edited_line::edited_line): Likewise.
19803 * final.c (asm_show_source): Likewise.
19804 * input.c (location_get_source_line): Convert return type
19805 from const char * to char_span, losing the final "line_len"
19807 (dump_location_info): Update for the above.
19808 (get_substring_ranges_for_loc): Likewise. Use a char_span
19809 when handling the literal within the line.
19810 (test_reading_source_line): Update for location_get_source_line
19811 returning a char_span.
19812 * input.h (class char_span): Move here from
19813 diagnostic-show-locus.c, converting from a struct to a class.
19814 Make data members private.
19815 (char_span::operator bool): New.
19816 (char_span::length): New.
19817 (char_span::get_buffer): New.
19818 (char_span::operator[]): New.
19819 (char_span::subspan): Make const.
19820 (char_span::xstrdup): New.
19821 (location_get_source_line): Convert return type from const char *
19822 to char_span, losing the final "line_size" param.
19824 2018-04-30 Jan Hubicka <jh@suse.cz>
19826 * lto-wrapper.c (ltrans_priorities): New static var.
19827 (cmp_priority): New.
19828 (run_gcc): Read priorities and if doing parallel build order
19829 the Makefile by them.
19831 2018-04-30 David Malcolm <dmalcolm@redhat.com>
19833 * input.h (builtins_location_check): Convert to a STATIC_ASSERT.
19835 2018-04-30 Richard Biener <rguenther@suse.de>
19837 * tree-cfg.c (verify_address): Remove base argument, add
19838 flag whether to check TREE_ADDRESSABLE and do that.
19839 (verify_expr): Remove.
19840 (verify_types_in_gimple_reference): Add pieces from verify_expr.
19841 (verify_gimple_assign_single): Likewise.
19842 (verify_gimple_switch): Likewise.
19843 (verify_expr_location_1): Dereference tp once. Add (disabled)
19844 piece from verify_expr.
19845 (verify_gimple_in_cfg): Do not call verify_expr on all ops.
19847 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
19849 * config/arc/linux.h (CLEAR_INSN_CACHE): Define.
19851 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
19853 * config/arc/arc-protos.h (prepare_extend_operands): Remove.
19854 (small_data_pattern): Likewise.
19855 (arc_rewrite_small_data): Likewise.
19856 * config/arc/arc.c (LEGITIMATE_SMALL_DATA_OFFSET_P): Remove.
19857 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Likewise.
19858 (get_symbol_alignment): New function.
19859 (legitimate_small_data_address_p): Likewise.
19860 (legitimate_scaled_address): Update, call
19861 legitimate_small_data_address_p.
19862 (output_sdata): New static variable.
19863 (arc_print_operand): Update how we handle small data operands.
19864 (arc_print_operand_address): Likewise.
19865 (arc_legitimate_address_p): Update, use
19866 legitimate_small_data_address_p.
19867 (arc_rewrite_small_data_p): Remove.
19868 (arc_rewrite_small_data_1): Likewise.
19869 (arc_rewrite_small_data): Likewise.
19870 (small_data_pattern): Likewise.
19871 (compact_sda_memory_operand): Update to use
19872 legitimate_small_data_address_p and get_symbol_alignment.
19873 (prepare_move_operands): Don't rewite sdata pattern.
19874 (prepare_extend_operands): Remove.
19875 * config/arc/arc.md (zero_extendqihi2): Don't rewrite sdata
19877 (zero_extendqisi2): Likewise.
19878 (zero_extendhisi2): Likewise.
19879 (extendqihi2): Likewise.
19880 (extendqisi2): Likewise.
19881 (extendhisi2): Likewise.
19882 (addsi3): Likewise.
19883 (subsi3): Likewise.
19884 (andsi3): Likewise.
19885 * config/arc/constraints.md (Usd): Change it to memory constraint.
19887 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
19889 * config/arc/arc.c (arc_split_move): Allow signed 6-bit constants
19890 as source of std instructions.
19891 * config/arc/arc.md (movsi_insn): Update pattern predicate to
19892 allow 6-bit constants as source for store instructions.
19893 (movdi_insn): Update instruction pattern to allow 6-bit constants
19894 as source for store instructions.
19896 2018-04-30 Jonathan Wakely <jwakely@redhat.com>
19898 * doc/invoke.texi (-fdebug-types-section): Fix grammar.
19900 2018-04-30 Nathan Sidwell <nathan@acm.org>
19901 Sandra Loosemore <sandra@codesourcery.com>
19903 * dumpfile.c (dump_open): Allow '-' for stdout.
19904 * doc/invoke.texi (Developer Options): Document dump filename
19905 determination early. Document stdin/stdout selection.
19907 2018-04-30 Andrew Sadek <andrew.sadek.se@gmail.com>
19909 Microblaze Target: PIC data text relative
19911 * config/microblaze/microblaze.opt: add new option -mpic-data-text-rel.
19912 * config/microblaze/microblaze-protos.h (microblaze_constant_address_p):
19914 * config/microblaze/microblaze.h (microblaze_constant_address_p):
19915 CONSTANT_ADDRESS_P definition to microblaze_constant_address_p.
19916 * config/microblaze/microblaze.c (TARGET_PIC_DATA_TEXT_REL):
19917 New addressing mode for data-text relative position indepenedent code.
19918 (microblaze_classify_unspec): add 'UNSPEC_TEXT' case ->
19919 'ADDRESS_SYMBOLIC_TXT_REL'.
19920 (microblaze_classify_address): Add handling for UNSPEC + CONST_INT.
19921 (microblaze_legitimate_pic_operand): Exclude function calls from
19922 pic operands in case of TARGET_PIC_DATA_TEXT_REL option.
19923 (microblaze_legitimize_address): Generate 'UNSPEC_TEXT' for all possible
19925 (microblaze_address_insns): Add 'ADDRESS_SYMBOLIC_TXT_REL' case.
19926 (print_operand): Add 'ADDRESS_SYMBOLIC_TXT_REL' case.
19927 (print_operand_address): Add 'ADDRESS_SYMBOLIC_TXT_REL' case + handling
19928 for 'address + offset'.
19929 (microblaze_expand_prologue): Add new function prologue call for
19931 (microblaze_asm_generate_pic_addr_dif_vec): Override new target hook
19932 'TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC' to disable address diff vector
19933 table in case of TARGET_PIC_DATA_TEXT_REL.
19934 (expand_pic_symbol_ref): Add handling for 'UNSPEC_TEXT'.
19935 * config/microblaze/microblaze.md (TARGET_PIC_DATA_TEXT_REL):
19936 Add new macros 'UNSPEC_TEXT',
19937 'UNSPEC_SET_TEXT' + add rule for setting r20 in function prologue
19938 + exclude function calls from 'UNSPEC_PLT' in case of data text
19940 * doc/tm.texi.in (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Add
19941 new target hook for generating address diff vector tables in case of
19943 * doc/tm.texi : Regenerate.
19944 * stmt.c (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Append new condition
19945 'targetm.asm_out.generate_pic_addr_diff_vec' to flag_pic in case
19946 of addr diff vector generation.
19947 * target.def (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Add
19948 target hook definition.
19949 * targhooks.h, targhooks.c (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC):
19950 Add default function for generate_pic_addr_diff_vec -> flag_pic.
19951 * doc/invoke.texi (Add new pic option): Add new microblaze pic
19952 option for data text relative.
19954 2018-04-30 Richard Biener <rguenther@suse.de>
19956 * tree-chrec.h (evolution_function_is_constant_p): Remove
19958 * tree-cfg.c (tree_node_can_be_shared): Re-order checks.
19960 2018-04-30 Richard Biener <rguenther@suse.de>
19963 * dwarf2out.c (gen_producer_string): Ignore -fchecking[=].
19965 2018-04-30 Richard Biener <rguenther@suse.de>
19967 PR tree-optimization/28364
19968 PR tree-optimization/85275
19969 * tree-ssa-loop-ch.c (ch_base::copy_headers): Stop after
19970 copying first exit test.
19972 2018-04-28 Mark Wielaard <mark@klomp.org>
19974 * dwarf2out.c (dwarf2out_finish): Add .debug_addr table header for
19975 dwarf_version >= 5.
19976 (dwarf_AT): Handle DW_AT_addr_base.
19977 (add_top_level_skeleton_die_attrs): Use dwarf_AT for DW_AT_addr_base.
19979 2018-04-28 Uros Bizjak <ubizjak@gmail.com>
19982 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask): New pattern.
19983 (*ashl<dwi>3_doubleword_mask_1): Ditto.
19984 (*<shift_insn><dwi>3_doubleword_mask): Ditto.
19985 (*<shift_insn><dwi>3_doubleword_mask_1): Ditto.
19987 2018-04-28 Richard Biener <rguenther@suse.de>
19989 * tree-cfg.c (verify_gimple_phi): Take a gphi * argument.
19990 (verify_gimple_in_cfg): Rename visited_stmts to visited_throwing_stmts
19991 to reflect use. Only add interesting stmts.
19993 2018-04-27 Martin Jambor <mjambor@suse.cz>
19996 * ipa-cp.c (find_aggregate_values_for_callers_subset): Make sure
19997 the jump function allows for passing through aggregate values.
19999 2018-04-27 David Malcolm <dmalcolm@redhat.com>
20001 * input.h (in_system_header_at): Convert from macro to inline
20003 (from_macro_expansion_at): Likewise.
20004 (from_macro_definition_at): Likewise.
20006 2018-04-27 Jeff Law <law@redhat.com>
20008 * config.gcc: Mark tile* targets as deprecated/obsolete.
20010 2018-04-27 Richard Biener <rguenther@suse.de>
20012 * config/aarch64/aarch64.c: Simplify ap.__stack advance and
20015 2018-04-27 Richard Biener <rguenther@suse.de>
20017 * tree-cfg.c (verify_expr): Make dead code hit gcc_unreachable.
20019 2018-04-27 Uros Bizjak <ubizjak@gmail.com>
20021 * config/i386/i386.md (*movti_internal): Substitute Ye constraint
20022 with Yd constraint. Set "preferred_for_speed" attribute from
20023 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC for alternatives
20024 with Yd constraint.
20025 (*movdi_internal): Ditto.
20026 (movti_interunit splitters): Remove
20027 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC from insn condition.
20028 (movdi_interunit splitters): Ditto.
20029 * config/i386/constraints.md (Ye): Remove.
20030 (Yd): Do not depend on TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC.
20032 2018-04-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20035 * config/aarch64/constraints.md (Usg): Limit to 31.
20036 (Usj): Limit to 63.
20038 2018-04-27 Jakub Jelinek <jakub@redhat.com>
20040 PR tree-optimization/85529
20041 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Add FIRST_BB
20042 argument. Don't call get_nonzero_bits if opcode is ERROR_MARK_NODE,
20043 rhs2 def stmt's bb is dominated by first_bb and it isn't an obvious
20044 zero extension or masking of the MSB bit.
20045 (optimize_range_tests): Add FIRST_BB argument, pass it through
20046 to optimize_range_tests_var_bound.
20047 (maybe_optimize_range_tests, reassociate_bb): Adjust
20048 optimize_range_tests callers.
20050 2018-04-26 Richard Biener <rguenther@suse.de>
20051 Jakub Jelinek <jakub@redhat.com>
20053 * cgraph.h (symbol_table): Just declare debug method here.
20054 * symtab.c (symbol_table::debug): Define.
20056 2018-04-26 Eric Botcazou <ebotcazou@adacore.com>
20058 * loop-invariant.c (may_assign_reg_p): Return false for frame pointer.
20060 2018-04-26 Uros Bizjak <ubizjak@gmail.com>
20062 * config/i386/i386.md ("isa" attribute): Add x64_sse2.
20063 ("enabled" attribute): Handle x64_sse2 "isa" attribute.
20064 (*movdi_internal): Substitute Yi and Yj constraint with x
20065 and Ym and Yn constraint with y constraint. Update "isa"
20066 attribute and set "preferred_for_speed" attribute from
20067 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC for updated alternatives.
20068 (*movsi_internal): Ditto.
20069 (*movdf_internal): Ditto.
20070 (*movsf_internal): Ditto.
20071 (*zero_extendsidi2): Ditto.
20072 * config/i386/sse.md (vec_set<mode>_0): Ditto.
20073 (sse2_loadld): Ditto.
20074 (*vec_extract<ssevecmodelower>_0): Ditto.
20075 (*vec_extractv4si_0_zext_sse4): Ditto.
20076 (vec_concatv2di): Ditto.
20077 (*vec_dup<mode>): Ditto.
20078 * config/i386/mmx.md (*mov<mode>_internal): Ditto.
20079 * config/i386/constraints.md (Yi): Remove.
20084 2018-04-26 Nathan Sidwell <nathan@acm.org>
20086 * dumpfile.c (dump_open): New.
20087 (dump_open_alternate_stream, dump_start, dump_begin): Call it.
20088 (dump_finish): Detect stdio/stderr by value not name.
20090 2018-04-26 Jonathan Wakely <jwakely@redhat.com>
20092 * doc/invoke.texi (-Wreturn-type): Document default status for C++.
20094 2018-04-26 Tom de Vries <tom@codesourcery.com>
20097 * config/nvptx/nvptx.c (verify_neutering_jumps)
20098 (verify_neutering_labels): New function
20099 (nvptx_single): Use verify_neutering_jumps and verify_neutering_labels.
20101 2018-04-26 Tom de Vries <tom@codesourcery.com>
20104 * config/nvptx/nvptx.c (needs_neutering_p): New function.
20105 (nvptx_single): Use needs_neutering_p to skip over insns that do not
20108 2018-04-26 Richard Biener <rguenther@suse.de>
20109 Tom de Vries <tom@codesourcery.com>
20112 * lto-streamer-out.c (output_function): Fixup loops if required to match
20113 discovery done in the reader.
20115 2018-04-26 Richard Biener <rguenther@suse.de>
20117 PR tree-optimization/85116
20118 * tree-ssa-loop-ch.c (do_while_loop_p): A do-while loop should
20119 have a loop exit from the single latch predecessor. Remove
20120 case of header with just condition.
20121 (ch_base::copy_headers): Exclude infinite loops from any
20123 (pass_ch::execute): Record exits.
20125 2018-04-26 Richard Biener <rguenther@suse.de>
20127 * tree-vect-data-refs.c (vect_get_data_access_cost): Get
20128 prologue cost vector and pass it to vect_get_load_cost.
20129 (vect_get_peeling_costs_all_drs): Likewise.
20130 (vect_peeling_hash_get_lowest_cost): Likewise.
20131 (vect_enhance_data_refs_alignment): Likewise.
20133 2018-04-26 Richard Biener <rguenther@suse.de>
20135 PR middle-end/85450
20136 * tree-cfg.c (verify_gimple_assign_unary): Restore proper
20137 checking of integer<->pointer conversions.
20138 * omp-expand.c (expand_omp_for_static_nochunk): Avoid
20139 sign-/zero-extending pointer types.
20140 (expand_omp_for_static_chunk): Likewise.
20142 2018-03-22 Hans-Peter Nilsson <hp@axis.com>
20143 Jean Lee <xiaoyur347@gmail.com>
20145 * config/mips/mips.c (mips_asan_shadow_offset): New function.
20146 (TARGET_ASAN_SHADOW_OFFSET): Define.
20147 * config/mips/mips.h (FRAME_GROWS_DOWNWARD): Augment to also be
20148 true for -fsanitize=address.
20150 2018-04-25 Mark Wielaard <mark@klomp.org>
20152 * dwarf2out.c (file_info_cmp): Sort longer dir prefixes before
20155 2018-04-25 Jakub Jelinek <jakub@redhat.com>
20157 * config/i386/i386.md (*x86_mov<mode>cc_0_m1): Use type "alu1" rather
20158 than "alu", remove explicit "memory" and "imm_disp" attributes.
20159 (*x86_mov<mode>cc_0_m1_se, *x86_mov<mode>cc_0_m1_neg): Likewise.
20161 PR middle-end/85414
20162 * simplify-rtx.c (simplify_unary_operation_1) <case SIGN_EXTEND,
20163 case ZERO_EXTEND>: Pass SUBREG_REG (op) rather than op to
20164 gen_lowpart_no_emit.
20166 2018-04-25 Sebastian Peryt <sebastian.peryt@intel.com>
20169 * config/i386/i386.c (ix86_expand_builtin): Change memory
20170 operand to XI, extend p0 to Pmode.
20171 * config/i386/i386.md: Change unspec volatile and operand
20172 1 mode to XI, change operand 0 mode to P.
20174 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
20176 * config/nds32/nds32-predicates.c (nds32_can_use_bclr_p): Mask with
20177 GET_MODE_MASK before any checking.
20178 (nds32_can_use_bset_p): Likewise.
20179 (nds32_can_use_btgl_p): Likewise.
20181 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
20183 * config/nds32/nds32-doubleword.md: New define_split pattern for
20184 illegal register number.
20186 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
20188 * config/nds32/nds32.c (nds32_print_operand): Set op_value ealier.
20190 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
20192 * config/nds32/nds32.h (ASM_APP_ON): Add missing newline character.
20194 2018-04-25 Richard Biener <rguenther@suse.de>
20196 * lto-streamer.h (LTO_major_version): Bump to 8.
20198 2018-04-25 Jakub Jelinek <jakub@redhat.com>
20200 * BASE-VER: Set to 9.0.0.
20202 2018-04-24 Segher Boessenkool <segher@kernel.crashing.org>
20204 * config/rs6000/rs6000.c (init_float128_ieee): Fix spelling mistakes
20205 in __abskf2 and __powikf2.
20207 2018-04-24 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20210 * config/aarch64/constraints.md (Usg, Usj): New constraints.
20211 * config/aarch64/iterators.md (cmode_simd): New mode attribute.
20212 * config/aarch64/aarch64.md (*aarch64_ashr_sisd_or_int_<mode>3):
20213 Use the above on operand 2. Reindent.
20214 (*aarch64_lshr_sisd_or_int_<mode>3): Likewise.
20216 2018-04-24 H.J. Lu <hongjiu.lu@intel.com>
20219 * common/config/i386/i386-common.c (ix86_handle_option): Don't
20221 * config/i386/i386.opt (mcet): Removed.
20222 * doc/install.texi: Remove -mcet documentation.
20223 * doc/invoke.texi: Likewise.
20225 2018-04-24 H.J. Lu <hongjiu.lu@intel.com>
20228 * doc/install.texi: Remove -mcet from bootstrap-cet.
20230 2018-04-24 Jakub Jelinek <jakub@redhat.com>
20233 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Don't define
20234 __builtin_ia32_readeflags_u32 and __builtin_ia32_writeeflags_u32
20238 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Punt if
20239 const_vector is not CONST_VECTOR or SYMBOL_REF for a constant pool
20240 containing a CONST_VECTOR.
20242 2018-04-24 Cesar Philippidis <cesar@codesourcery.com>
20244 * doc/install.texi: Update newlib dependency for nvptx.
20246 2018-04-24 Jakub Jelinek <jakub@redhat.com>
20249 * config/i386/i386.c (ix86_expand_vector_init_one_var): Use UINTVAL
20250 instead of INTVAL when shifting x left.
20252 2018-04-24 Andreas Krebbel <krebbel@linux.ibm.com>
20254 PR tree-optimization/85478
20255 * tree-vect-loop.c (vect_analyze_loop_2): Do not call
20256 vect_grouped_store_supported for single element vectors.
20258 2018-04-24 Richard Biener <rguenther@suse.de>
20261 * config/i386/i386.c (ix86_add_stmt_cost): Restrict strided
20262 load cost increase to the case of non-constant step.
20264 2018-04-24 Jakub Jelinek <jakub@redhat.com>
20267 * reg-stack.c (move_for_stack_reg): Don't fail assertion about dead
20268 destination if any_malformed_asm.
20270 2018-04-23 Eric Botcazou <ebotcazou@adacore.com>
20272 PR middle-end/85496
20273 * expr.c (store_field): In the bitfield case, if the value comes from
20274 a function call and is returned in registers by means of a PARALLEL,
20275 do not change the mode of the temporary unless BLKmode and VOIDmode.
20277 2018-04-23 Andrey Belevantsev <abel@ispras.ru>
20279 PR rtl-optimization/85423
20280 * sel-sched-ir.c (has_dependence_note_mem_dep): Only discard
20281 dependencies to debug insns when the previous insn is non-debug.
20283 2018-04-23 Claudiu Zissulescu <claziss@synopsys.com>
20285 * config/arc/arc.md ("vunspec"): Delete it, unify all the unspec
20286 enums into a single definition.
20287 (fls): Fix predicates and printing.
20290 2018-04-23 Claudiu Zissulescu <claziss@synopsys.com>
20292 * config/arc/arc-protos.h (check_if_valid_sleep_operand): Remove.
20293 * config/arc/arc.c (arc_expand_builtin): Sleep accepts registers
20294 and short u6 immediate.
20295 (check_if_valid_sleep_operand): Remove.
20296 * config/arc/arc.md (Sleep): Accepts registers and u6 immediates.
20298 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
20300 * config/nds32/nds32.c (nds32_compute_stack_frame): Consider
20301 flag_always_save_lp condition.
20302 * config/nds32/nds32.opt (malways-save-lp): New option.
20304 2018-04-22 Shiva Chen <shiva0217@gmail.com>
20306 * config/nds32/nds32-protos.h (nds32_use_load_post_increment): Declare.
20307 * config/nds32/nds32.c (nds32_use_load_post_increment): New.
20308 * config/nds32/nds32.h
20309 (USE_LOAD_POST_INCREMENT, USE_LOAD_POST_DECREMENT): Define.
20310 (USE_STORE_POST_INCREMENT, USE_STORE_POST_DECREMENT): Define.
20312 2018-04-22 Shiva Chen <shiva0217@gmail.com>
20314 * config/nds32/nds32-protos.h (nds32_ls_333_p): Remove.
20315 * config/nds32/nds32.c (nds32_ls_333_p): Remove.
20317 2018-04-22 Shiva Chen <shiva0217@gmail.com>
20318 Chung-Ju Wu <jasonwucj@gmail.com>
20320 * config/nds32/nds32-protos.h (nds32_case_vector_shorten_mode):
20322 * config/nds32/nds32.c (nds32_case_vector_shorten_mode): New function.
20323 * config/nds32/nds32.h (CASE_VECTOR_SHORTEN_MODE): Modify.
20325 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
20327 * config/nds32/nds32.c (nds32_compute_stack_frame): Fix wrong value.
20329 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
20331 * config/nds32/nds32-protos.h (nds32_data_alignment,
20332 nds32_local_alignment): Declare.
20333 * config/nds32/nds32.c (nds32_data_alignment, nds32_constant_alignment,
20334 nds32_local_alignment): New functions.
20335 (TARGET_CONSTANT_ALIGNMENT): Define.
20336 * config/nds32/nds32.h (DATA_ALIGNMENT, LOCAL_ALIGNMENT): Define.
20338 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
20340 * config/nds32/nds32.c
20341 (TARGET_HARD_REGNO_MODE_OK): Move to the bottom of file.
20342 (TARGET_MODES_TIEABLE_P): Likewise.
20344 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
20346 * config/nds32/nds32.c (nds32_asm_file_start): Display optimization
20347 level Ofast and Og.
20349 2018-04-22 Monk Chiang <sh.chiang04@gmail.com>
20350 Chung-Ju Wu <jasonwucj@gmail.com>
20352 * config/nds32/constants.md (unspec_volatile_element): Add enum values
20353 for unaligned access.
20354 * config/nds32/nds32-intrinsic.c: Implementation of expanding
20356 * config/nds32/nds32-intrinsic.md: Likewise.
20357 * config/nds32/nds32_intrinsic.h: Likewise.
20358 * config/nds32/nds32.h (nds32_builtins): Likewise.
20359 * config/nds32/nds32.opt (munaligned-access): New option.
20360 * config/nds32/nds32.c (nds32_asm_file_start): Display
20361 flag_unaligned_access status.
20363 2018-04-20 Kito Cheng <kito.cheng@gmail.com>
20365 * config/riscv/elf.h (LINK_SPEC): Pass --no-relax if
20366 -mno-relax is present.
20367 * config/riscv/linux.h (LINK_SPEC): Ditto.
20369 2018-04-20 Martin Sebor <msebor@redhat.com>
20372 * gimple-fold.c (gimple_fold_builtin_strcpy): Suppress -Wrestrict
20374 (gimple_fold_builtin_stxcpy_chk): Same.
20375 * gimple-ssa-warn-restrict.c (check_bounds_or_overlap): Same.
20377 2018-04-20 Michael Meissner <meissner@linux.ibm.com>
20380 * config/rs6000/rs6000.c (init_float128_ieee): Add support to call
20381 __powikf2 when long double is IEEE 128-bit.
20383 2018-04-20 Kito Cheng <kito.cheng@gmail.com>
20385 * config/riscv/riscv.c (riscv_first_stack_step): Round up min
20386 step to make sure stack always aligned.
20388 2018-04-20 Carl Love <cel@us.ibm.com>
20391 * config/rs6000/rs6000-c.c (rs6000_gimple_fold_builtin): Add
20392 size check for arg0.
20394 2018-04-20 Nathan Sidwell <nathan@codesourcery.com>
20395 Tom de Vries <tom@codesourcery.com>
20398 * config/nvptx/nvptx.c (nvptx_emit_forking, nvptx_emit_joining):
20399 Emit insns for calls too.
20400 (nvptx_find_par): Always look for worker-level predecessor insn.
20401 (nvptx_propagate): Add is_call parm, return bool. Copy frame for
20403 (nvptx_vpropagate, nvptx_wpropagate): Adjust.
20404 (nvptx_process_pars): Propagate frames for calls.
20406 2018-04-20 H.J. Lu <hongjiu.lu@intel.com>
20409 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET):
20411 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
20412 (ix86_handle_option): Don't handle OPT_mibt.
20413 * config/i386/cet.h: Check __CET__ instead of __IBT__ and
20415 * config/i386/driver-i386.c (host_detect_local_cpu): Remove
20417 * config/i386/i386-c.c (ix86_target_macros_internal): Don't
20418 check OPTION_MASK_ISA_IBT nor flag_cf_protection.
20419 (ix86_target_macros): Define __CET__ with flag_cf_protection
20420 for -fcf-protection.
20421 * config/i386/i386.c (isa2_opts): Remove -mibt.
20422 * config/i386/i386.h (TARGET_IBT): Removed.
20423 (TARGET_IBT_P): Likewise.
20424 (ix86_valid_target_attribute_inner_p): Don't check OPT_mibt.
20425 * config/i386/i386.md (nop_endbr): Don't check TARGET_IBT.
20426 * config/i386/i386.opt (mcet): Update help message.
20427 (mshstk): Likewise.
20429 * doc/invoke.texi: Remove -mibt. Document __CET__. Document
20430 -mcet as an alias for -mshstk.
20432 2018-04-20 Richard Biener <rguenther@suse.de>
20434 PR middle-end/85475
20435 * match.pd ((X * CST) * Y -> (X * Y) * CST): Avoid exponential
20436 complexity by forcing a single use of the multiply operand.
20438 2018-04-20 Martin Jambor <mjambor@suse.cz>
20441 * ipa-cp.c (cgraph_edge_brings_value_p): Move check for self-feeding
20442 recursion dependency to only apply to non-clones.
20444 2018-04-20 Martin Jambor <mjambor@suse.cz>
20447 * ipa-cp.c (create_specialized_node): Check that clones of
20448 self-recursive edges exist during IPA-CP.
20450 2018-04-19 Toon Moene <toon@moene.org>
20452 * doc/invoke.texi: Add -floop-unroll-and-jam to options enabled
20455 2018-04-19 Jakub Jelinek <jakub@redhat.com>
20457 PR tree-optimization/85467
20458 * fold-const.c (fold_ternary_loc) <case BIT_FIELD_REF>: Use
20459 VECTOR_TYPE_P macro. If type is vector type, VIEW_CONVERT_EXPR the
20460 VECTOR_CST element to type.
20462 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
20465 * config/i386/i386.h (STACK_SAVEAREA_MODE): New.
20466 * config/i386/i386.md (builtin_setjmp_setup): Removed.
20467 (builtin_longjmp): Likewise.
20468 (save_stack_nonlocal): New pattern.
20469 (restore_stack_nonlocal): Likewise.
20471 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
20474 * config/i386/cet.c (file_end_indicate_exec_stack_and_cet):
20475 Replace ASM_OUTPUT_LABEL with fprintf.
20477 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
20480 * config/i386/cet.c (file_end_indicate_exec_stack_and_cet):
20481 Check flag_cf_protection instead of TARGET_IBT and TARGET_SHSTK.
20482 * config/i386/i386-c.c (ix86_target_macros_internal): Also
20483 define __IBT__ and __SHSTK__ for -fcf-protection.
20484 * config/i386/i386.c (pass_insert_endbranch::gate): Don't check
20486 (ix86_trampoline_init): Likewise.
20487 (x86_output_mi_thunk): Likewise.
20488 (ix86_notrack_prefixed_insn_p): Likewise.
20489 (ix86_option_override_internal): Don't disallow -fcf-protection.
20490 * config/i386/i386.md (rdssp<mode>): Also enable for
20492 (incssp<mode>): Likewise.
20493 (nop_endbr): Likewise.
20494 * config/i386/i386.opt (mcet): Change help message to built-in
20497 (mshstk): Likewise.
20498 * doc/invoke.texi: Remove -mcet, -mibt and -mshstk condition
20499 on -fcf-protection. Change -mcet, -mibt and -mshstk to only
20500 enable CET built-in functions.
20502 2018-04-19 Sebastian Peryt <sebastian.peryt@intel.com>
20504 * common/config/i386/i386-common.c
20505 (OPTION_MASK_ISA_MOVDIRI_SET, OPTION_MASK_ISA_MOVDIR64B_SET,
20506 OPTION_MASK_ISA_MOVDIRI_UNSET,
20507 OPTION_MASK_ISA_MOVDIR64B_UNSET): New defines.
20508 (ix86_handle_option): Handle -mmovdiri and -mmovdir64b.
20509 * config.gcc (movdirintrin.h): New header.
20510 * config/i386/cpuid.h (bit_MOVDIRI,
20511 bit_MOVDIR64B): New bits.
20512 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mmovdiri
20514 * config/i386/i386-builtin-types.def ((VOID, PUNSIGNED, UNSIGNED),
20515 (VOID, PVOID, PCVOID)): New function types.
20516 * config/i386/i386-builtin.def (__builtin_ia32_directstoreu_u32,
20517 __builtin_ia32_directstoreu_u64,
20518 __builtin_ia32_movdir64b): New builtins.
20519 * config/i386/i386-c.c (__MOVDIRI__, __MOVDIR64B__): New.
20520 * config/i386/i386.c (ix86_target_string): Added -mmovdir64b
20522 (ix86_valid_target_attribute_inner_p): Ditto.
20523 (ix86_expand_special_args_builtin): Added VOID_FTYPE_PUNSIGNED_UNSIGNED
20524 and VOID_FTYPE_PUNSIGNED_UNSIGNED.
20525 (ix86_expand_builtin): Expand IX86_BUILTIN_MOVDIR64B.
20526 * config/i386/i386.h (TARGET_MOVDIRI, TARGET_MOVDIRI_P,
20527 TARGET_MOVDIR64B, TARGET_MOVDIR64B_P): New.
20528 * config/i386/i386.md (UNSPECV_MOVDIRI, UNSPECV_MOVDIR64B): New.
20529 (movdiri<mode>, movdir64b_<mode>): New.
20530 * config/i386/i386.opt: Add -mmovdiri and -mmovdir64b.
20531 * config/i386/immintrin.h: Include movdirintrin.h.
20532 * config/i386/movdirintrin.h: New file.
20533 * doc/invoke.texi: Added -mmovdiri and -mmovdir64b.
20535 2018-04-19 Richard Biener <rguenther@suse.de>
20537 PR middle-end/85455
20538 * cfg.c (clear_bb_flags): When loop state says we have
20539 marked irreducible regions also preserve BB_IRREDUCIBLE_LOOP.
20541 2018-04-19 Richard Biener <rguenther@suse.de>
20543 PR tree-optimization/84737
20544 * tree-vect-data-refs.c (vect_copy_ref_info): New function
20545 copying restrict info.
20546 (vect_setup_realignment): Use it.
20547 * tree-vectorizer.h (vect_copy_ref_info): Declare.
20548 * tree-vect-stmts.c (vectorizable_store): Copy ref info from
20549 the first DR to all generated stores.
20550 (vectorizable_load): Likewise for loads.
20552 2018-04-19 Jakub Jelinek <jakub@redhat.com>
20554 PR tree-optimization/85446
20555 * match.pd ((intptr_t) x eq/ne CST to x eq/ne (typeof x) cst): Require
20556 the integral and pointer types to have the same precision.
20558 * doc/install.texi: Document --disable-cet being the default and
20561 2018-04-18 Martin Liska <mliska@suse.cz>
20563 * ipa-devirt.c (odr_subtypes_equivalent_p): Fix GNU coding
20566 2018-04-18 Martin Liska <mliska@suse.cz>
20569 2018-03-02 Eric Botcazou <ebotcazou@adacore.com>
20572 * ipa-devirt.c (odr_subtypes_equivalent_p): Get the ODR type of both
20573 arguments if they are comparable.
20575 2018-04-18 Martin Liska <mliska@suse.cz>
20578 2018-03-13 Eric Botcazou <ebotcazou@adacore.com>
20581 * ipa-devirt.c (odr_subtypes_equivalent_p): Do not get the ODR type of
20584 2018-04-18 H.J. Lu <hongjiu.lu@intel.com>
20587 * config/i386/i386.c (ix86_expand_split_stack_prologue): Insert
20588 ENDBR after calling __morestack.
20590 2018-04-18 David Malcolm <dmalcolm@redhat.com>
20593 * configure.ac (gcc-driver-name.h): Honor --with-gcc-major-version
20594 by using gcc_base_ver to generate a gcc_driver_version, and use
20595 it when generating GCC_DRIVER_NAME.
20596 * configure: Regenerate.
20598 2018-04-18 Jakub Jelinek <jakub@redhat.com>
20601 * config.gcc: Obsolete powerpc*-*-*spe*.
20603 2018-04-17 Jakub Jelinek <jakub@redhat.com>
20606 * dbxout.c (dbxout_int): Perform negation in unsigned int type.
20607 (stabstr_D): Change type of unum from unsigned int to
20608 unsigned HOST_WIDE_INT. Perform negation in unsigned HOST_WIDE_INT
20611 2018-04-17 Jim Wilson <jimw@sifive.com>
20614 * config/riscv/riscv.c (riscv_compute_frame_info): Add calls to
20615 RISCV_STACK_ALIGN when using outgoing_args_size and pretend_args_size.
20616 Set arg_pointer_offset after using pretend_args_size.
20618 2018-04-17 Jakub Jelinek <jakub@redhat.com>
20620 PR rtl-optimization/85431
20621 * dse.c (record_store): Ignore zero width stores.
20624 * asan.c (handle_builtin_stack_restore): Adjust comment. Emit
20625 __asan_allocas_unpoison call and last_alloca_addr = new_sp before
20626 __builtin_stack_restore rather than after it.
20627 * builtins.c (expand_asan_emit_allocas_unpoison): Pass
20628 arg1 + (virtual_dynamic_stack_rtx - stack_pointer_rtx) as second
20629 argument instead of virtual_dynamic_stack_rtx.
20631 2018-04-17 Kelvin Nilsen <kelvin@gcc.gnu.org>
20633 * config/rs6000/rs6000-protos.h (rs6000_builtin_is_supported_p):
20635 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
20636 Add note to error message to explain internal mapping of overloaded
20637 built-in function name to non-overloaded built-in function name.
20638 * config/rs6000/rs6000.c (rs6000_builtin_is_supported_p): New
20641 2018-04-17 Michael Meissner <meissner@linux.vnet.ibm.com>
20644 * config/rs6000/rs6000.md (pack<mode>): Do not try handle a pack
20645 where the inputs overlap with the output.
20647 2018-04-17 Jakub Jelinek <jakub@redhat.com>
20649 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): Add
20650 (=v, v) alternative and explicit "memory" attribute.
20651 (vec_extract_lo_<mode><mask_name>): Likewise. Also add
20652 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
20654 (vec_extract_lo_<mode><mask_name>): Add (=v, v) alternative and use
20655 "sselog1" type instead of "sselog".
20656 (vec_extract_hi_<mode><mask_name>): Use "sselog1" type instead of
20657 "sselog". Remove explicit "memory" attribute.
20658 (vec_extract_lo_v32hi): Add (=v, v) alternative and explicit "memory",
20659 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
20661 (vec_extract_hi_v32hi): Merge all alternatives into one, use
20662 "sselog1" type instead of "sselog". Remove explicit "memory"
20664 (vec_extract_hi_v16hi): Merge each pair of alternatives into one,
20665 use "sselog1" type instead of "sselog". Remove explicit "memory"
20667 (vec_extract_lo_v64qi): Add (=v, v) alternative and explicit "memory",
20668 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
20670 (vec_extract_hi_v64qi): Merge all alternatives into one, use
20671 "sselog1" type instead of "sselog". Remove explicit "memory"
20673 (vec_extract_hi_v32qi): Merge each pair of alternatives into one,
20674 use "sselog1" type instead of "sselog". Remove explicit "memory"
20678 * config/i386/i386.md (*ashlqi3_1_slp): Use alu1 type instead of alu.
20680 PR middle-end/85414
20681 * rtlhooks.c (gen_lowpart_if_possible): Don't call gen_lowpart_SUBREG
20684 2018-04-17 Martin Jambor <mjambor@suse.cz>
20687 * ipa-cp.c (create_specialized_node): Call
20688 expand_all_artificial_thunks if necessary.
20690 2018-04-17 Martin Liska <mliska@suse.cz>
20693 * ipa-devirt.c (odr_types_equivalent_p): Remove trailing
20694 in message, remote space in between '_G' and '('.
20696 2018-04-17 Jakub Jelinek <jakub@redhat.com>
20699 * config/i386/sse.md (reduces<mode><mask_scalar_name>,
20700 avx512f_vmcmp<mode>3<round_saeonly_name>,
20701 avx512f_vmcmp<mode>3_mask<round_saeonly_name>,
20702 avx512f_sgetexp<mode><mask_scalar_name><round_saeonly_scalar_name>,
20703 avx512f_rndscale<mode><round_saeonly_name>,
20704 avx512dq_ranges<mode><mask_scalar_name><round_saeonly_scalar_name>,
20705 avx512f_vgetmant<mode><mask_scalar_name><round_saeonly_scalar_name>):
20706 Use %<iptr>2 instead of %2 for -masm=intel.
20707 (avx512f_vcvtss2usi<round_name>, avx512f_vcvtss2usiq<round_name>,
20708 avx512f_vcvttss2usi<round_saeonly_name>,
20709 avx512f_vcvttss2usiq<round_saeonly_name>): Use %k1 instead of %1 for
20711 (avx512f_vcvtsd2usi<round_name>, avx512f_vcvtsd2usiq<round_name>,
20712 avx512f_vcvttsd2usi<round_saeonly_name>,
20713 avx512f_vcvttsd2usiq<round_saeonly_name>, ufloatv2siv2df2<mask_name>):
20714 Use %q1 instead of %1 for -masm=intel.
20715 (avx512f_sfixupimm<mode><sd_maskz_name><round_saeonly_name>,
20716 avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Use %<iptr>3 instead
20717 of %3 for -masm=intel.
20718 (sse2_shufpd_v2df_mask): Fix a typo, change %{6%} to %{%6%} for
20720 (*avx512vl_<code>v2div2qi2_store): Use %w0 instead of %0 for
20722 (*avx512vl_<code><mode>v4qi2_store): Use %k0 instead of %0 for
20724 (avx512vl_<code><mode>v4qi2_mask_store): Use a single pattern with
20725 %k0 and %1 for -masm=intel rather than two patterns, one with %0 and
20727 (*avx512vl_<code><mode>v8qi2_store): Use %q0 instead of %0 for
20729 (avx512vl_<code><mode>v8qi2_mask_store): Use a single pattern with
20730 %q0 and %1 for -masm=intel rather than two patterns, one with %0 and
20731 %g1 and one with %0 and %1.
20732 (avx512er_vmrcp28<mode><round_saeonly_name>,
20733 avx512er_vmrsqrt28<mode><round_saeonly_name>): Use %<iptr>1 instead of
20734 %1 for -masm=intel.
20735 (avx5124fmaddps_4fmaddps_mask, avx5124fmaddps_4fmaddss_mask,
20736 avx5124fmaddps_4fnmaddps_mask, avx5124fmaddps_4fnmaddss_mask,
20737 avx5124vnniw_vp4dpwssd_mask, avx5124vnniw_vp4dpwssds_mask): Swap order
20738 of %0 and %{%4%} for -masm=intel.
20739 (avx5124fmaddps_4fmaddps_maskz, avx5124fmaddps_4fmaddss_maskz,
20740 avx5124fmaddps_4fnmaddps_maskz, avx5124fmaddps_4fnmaddss_maskz,
20741 avx5124vnniw_vp4dpwssd_maskz, avx5124vnniw_vp4dpwssds_maskz): Swap
20742 order of %0 and %{%5%}%{z%} for -masm=intel.
20744 2018-04-17 Jan Hubicka <jh@suse.cz>
20747 * ipa-devirt.c (odr_types_equivalent_p): Handle bit fields.
20749 2018-04-17 Martin Liska <mliska@suse.cz>
20752 * multiple_target.c (create_dispatcher_calls): Set apostrophes
20753 for target_clone error message. Make default implementation
20754 clone to be a local declaration.
20755 (separate_attrs): Add new argument and check for an empty
20757 (expand_target_clones): Handle it.
20758 (ipa_target_clone): Make redirection just for target_clones
20761 2018-04-16 Cesar Philippidis <cesar@codesourcery.com>
20762 Tom de Vries <tom@codesourcery.com>
20764 PR middle-end/84955
20765 * omp-expand.c (expand_oacc_for): Add dummy false branch for
20766 tiled basic blocks without omp continue statements.
20768 2018-04-16 Aaron Sawdey <acsawdey@linux.ibm.com>
20771 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Mark
20772 vec_extract expression as having side effects to make sure it gets
20775 2018-04-16 H.J. Lu <hongjiu.lu@intel.com>
20778 * config/i386/i386.c (get_builtin_code_for_version): Check
20781 2018-04-16 Olga Makhotina <olga.makhotina@intel.com>
20784 * config.gcc: Support "skylake".
20785 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
20787 * config/i386/i386.c (m_SKYLAKE): Define.
20788 (processor_target_table): Add "skylake".
20789 (ix86_option_override_internal): Add "skylake".
20790 (get_builtin_code_for_version): Handle PROCESSOR_SKYLAKE,
20791 PROCESSOR_CANNONLAKE.
20792 (get_builtin_code_for_version): Fix priority for
20793 PROCESSOR_ICELAKE_CLIENT, PROCESSOR_ICELAKE_SERVER,
20794 PROCESSOR_SKYLAKE-AVX512.
20795 * config/i386/i386.h (processor_costs): Define TARGET_SKYLAKE.
20796 (processor_type): Add PROCESSOR_SKYLAKE.
20798 2018-04-16 Paolo Carlini <paolo.carlini@oracle.com>
20799 Jason Merrill <jason@redhat.com>
20802 * convert.c (convert_to_integer_1): Use direct recursion for
20803 enumeral types and types with a precision less than the number
20804 of bits in their mode.
20806 2018-04-16 Julia Koval <julia.koval@intel.com>
20809 * config/i386/x86-tune.def (X86_TUNE_SSE_UNALIGNED_LOAD_OPTIMAL,
20810 X86_TUNE_SSE_UNALIGNED_STORE_OPTIMAL): Add m_SKYLAKE_AVX512
20812 2018-04-14 Segher Boessenkool <segher@kernel.crashing.org>
20815 * config/rs6000/rs6000.opt (mdirect-move): Make deprecated.
20816 * doc/invoke.texi (RS/6000 and PowerPC Options): Remove -mdirect-move
20817 and -mno-direct-move.
20819 2018-04-13 Paul A. Clarke <pc@us.ibm.com>
20822 * config/rs6000/emmintrin.h (_mm_slli_epi{16,32,64}):
20823 Ensure that vec_splat_s32 is only called with 0 <= shift < 16.
20824 Ensure negative shifts result in {0}.
20826 2018-04-13 Vladimir Makarov <vmakarov@redhat.com>
20828 PR rtl-optimization/79916
20829 * config/rs6000/rs6000.c (rs6000_emit_move): Use assigned hard
20830 regs (if any) to define how to gnerate SD moves when LRA is in
20833 2018-04-13 Jakub Jelinek <jakub@redhat.com>
20835 PR rtl-optimization/85393
20836 * except.h (expand_dw2_landing_pad_for_region): Remove declaration.
20837 * except.c (expand_dw2_landing_pad_for_region): Make static.
20838 * bb-reorder.c (fix_up_crossing_landing_pad): In new_bb emit just
20839 a label and unconditional jump to old_bb, rather than
20840 expand_dw2_landing_pad_for_region insn(s) and jump to single_succ
20843 PR rtl-optimization/85376
20844 * simplify-rtx.c (simplify_const_unary_operation): For CLZ and CTZ and
20845 zero op0, if C?Z_DEFINED_VALUE_AT_ZERO is false, return NULL_RTX
20846 instead of a specific value.
20848 2018-04-13 Jan Hubicka <hubicka@ucw.cz>
20849 Bin Cheng <bin.cheng@arm.com>
20851 PR tree-optimization/82965
20852 PR tree-optimization/83991
20853 * cfgloopanal.c (expected_loop_iterations_unbounded): Add
20854 by_profile_only parameter.
20855 * cfgloopmanip.c (scale_loop_profile): Further scale loop's profile
20856 information if the loop was predicted to iterate too many times.
20857 * cfgloop.h (expected_loop_iterations_unbounded): Update prototype
20859 2018-04-13 Jan Hubicka <hubicka@ucw.cz>
20862 * config/i386/i386.c (ix86_can_inline_p): Allow safe transitions for
20865 2018-04-13 Martin Liska <mliska@suse.cz>
20866 Jakub Jelinek <jakub@redhat.com>
20868 PR middle-end/81657
20869 * expr.h (enum block_op_methods): Add BLOCK_OP_NO_LIBCALL_RET.
20870 * expr.c (emit_block_move_hints): Handle BLOCK_OP_NO_LIBCALL_RET.
20871 * builtins.c (expand_builtin_memory_copy_args): Use
20872 BLOCK_OP_NO_LIBCALL_RET method for mempcpy with non-ignored target,
20873 handle dest_addr == pc_rtx.
20875 2018-04-12 Segher Boessenkool <segher@kernel.crashing.org>
20878 * config/rs6000/rs6000.md (fix_trunc<mode>si2): Use legacy code if
20879 asked to not generate direct moves.
20880 (fix_trunc<mode>si2_stfiwx): Similar.
20881 (fix_trunc<mode>si2_internal): Similar.
20883 2018-04-12 Jakub Jelinek <jakub@redhat.com>
20886 * var-tracking.c (add_stores): Handle STRICT_LOW_PART SET_DEST.
20887 * cselib.c (cselib_record_sets): For STRICT_LOW_PART dest,
20888 lookup if dest in some wider mode is known to be const0_rtx and
20889 if so, record permanent equivalence for it to be ZERO_EXTEND of
20890 the narrower mode destination.
20892 2018-04-12 Cesar Philippidis <cesar@codesourcery.com>
20894 * lto-streamer-out.c (output_function): Revert 259346.
20895 * omp-expand.c (expand_oacc_for): Likewise.
20897 2018-04-12 Alexander Monakov <amonakov@ispras.ru>
20899 PR rtl-optimization/85354
20900 * sel-sched-ir.c (sel_init_pipelining): Move cfg_cleanup call...
20901 * sel-sched.c (sel_global_init): ... here.
20903 2018-04-12 Eric Botcazou <ebotcazou@adacore.com>
20906 * lto-wrapper.c (debug_objcopy): Open the files in binary mode.
20907 * dwarf2out.c (dwarf2out_early_finish): Do not generate assembly in LTO
20908 mode for PE-COFF targets.
20909 * config/i386/i386-protos.h (i386_pe_asm_lto_start): Declare.
20910 (i386_pe_asm_lto_end): Likewise.
20911 * config/i386/cygming.h (TARGET_ASM_LTO_START): Define.
20912 (TARGET_ASM_LTO_END): Likewise.
20913 * config/i386/winnt.c (saved_debug_info_level): New static variable.
20914 (i386_pe_asm_lto_start): New function.
20915 (i386_pe_asm_lto_end): Likewise.
20917 2018-04-12 Cesar Philippidis <cesar@codesourcery.com>
20918 Richard Biener <rguenther@suse.de>
20920 PR middle-end/84955
20921 * lto-streamer-out.c (output_function): Fix CFG loop state before
20923 * omp-expand.c (expand_oacc_for): Handle calls to internal
20924 functions like regular functions.
20926 2018-04-12 Richard Biener <rguenther@suse.de>
20929 * dwarf2out.c (init_sections_and_labels): Use debug_line_section[_label]
20930 for the early LTO debug to properly generate references to it
20931 during DIE emission. Do not re-use that for the skeleton for
20933 (dwarf2out_early_finish): Likewise.
20935 2018-04-12 Jakub Jelinek <jakub@redhat.com>
20938 * config/i386/sse.md
20939 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name> split,
20940 <mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name> split,
20941 vec_extract_lo_<mode><mask_name> split, vec_extract_lo_v32hi,
20942 vec_extract_lo_v64qi): For non-AVX512VL if input is xmm16+ reg
20943 and output is a reg, avoid creating invalid lowpart subreg, but
20944 instead split into a 512-bit move. Don't split if not AVX512VL,
20945 input is xmm16+ reg and output is a mem.
20946 (vec_extract_lo_<mode><mask_name>, vec_extract_lo_v32hi,
20947 vec_extract_lo_v64qi): Don't require split if not AVX512VL, input is
20948 xmm16+ reg and output is a mem.
20950 2018-04-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20952 * config/s390/s390.c (s390_output_indirect_thunk_function): Check
20953 also for flag_dwarf2_cfi_asm.
20955 2018-04-12 Jakub Jelinek <jakub@redhat.com>
20957 PR rtl-optimization/85342
20958 * regcprop.c (copyprop_hardreg_forward_1): Remove replaced array, use
20959 a bool scalar var inside of the loop instead. Don't try to update
20960 recog_data.operand after failed apply_change_group.
20962 2018-04-12 Tom de Vries <tom@codesourcery.com>
20965 * config/nvptx/nvptx.c (flexible_array_member_type_p): New function.
20966 (nvptx_assemble_decl_begin): Add undefined param. Declare undefined
20967 array with flexible array member as array without given dimension.
20968 (nvptx_assemble_undefined_decl): Set nvptx_assemble_decl_begin call
20969 argument for undefined param to true.
20971 2018-04-11 Aaron Sawdey <acsawdey@linux.ibm.com>
20974 * doc/invoke.texi (RS/6000 and PowerPC Options): Document options
20975 -mcall- and -mtraceback=. Remove options -mabi=spe and -mabi=no-spe
20976 from PowerPC section.
20977 * config/rs6000/sysv4.opt (mcall-): Improve help text.
20978 * config/rs6000/rs6000.opt (mblock-compare-inline-limit=): Trim
20979 help text that is too long.
20980 * config/rs6000/rs6000.opt (mblock-compare-inline-loop-limit=): Trim
20981 help text that is too long.
20982 * config/rs6000/rs6000.opt (mstring-compare-inline-limit=): Trim
20983 help text that is too long.
20985 2018-04-11 Uros Bizjak <ubizjak@gmail.com>
20987 * config/alpha/alpha.md (stack_probe_internal): Rename
20988 from "probe_stack". Update all callers.
20990 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
20992 PR rtl-optimization/84566
20993 * sched-deps.c (sched_analyze_insn): Check deps->readonly when invoking
20994 sched_macro_fuse_insns.
20996 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
20999 * sched-rgn.c (add_branch_dependences): Move sel_sched_p check here...
21000 (compute_block_dependences): ... from here.
21002 2018-04-11 Jakub Jelinek <jakub@redhat.com>
21004 PR tree-optimization/85331
21005 * vec-perm-indices.h (vec_perm_indices::clamp): Change input type
21006 from int to HOST_WIDE_INT.
21008 2018-04-11 Martin Jambor <mjambor@suse.cz>
21011 * ipa-cp.c (propagate_vals_across_pass_through): Expand comment.
21012 (cgraph_edge_brings_value_p): New parameter dest_val, check if it is
21013 not the same as the source val.
21014 (cgraph_edge_brings_value_p): New parameter.
21015 (gather_edges_for_value): Pass destination value to
21016 cgraph_edge_brings_value_p.
21017 (perhaps_add_new_callers): Likewise.
21018 (get_info_about_necessary_edges): Likewise and exclude values brought
21019 only by self-recursive edges.
21020 (create_specialized_node): Redirect only clones of self-calling edges.
21021 (+self_recursive_pass_through_p): New function.
21022 (find_more_scalar_values_for_callers_subset): Use it.
21023 (find_aggregate_values_for_callers_subset): Likewise.
21024 (known_aggs_to_agg_replacement_list): Removed.
21025 (decide_whether_version_node): Re-calculate known constants for all
21026 remaining context clones.
21028 2018-04-11 Richard Biener <rguenther@suse.de>
21031 * dwarf2out.c (dwarf2out_finish): Remove DW_AT_stmt_list attribute
21032 from early DWARF output.
21033 (dwarf2out_early_finish): Output line info unconditionally into
21034 early DWARF and add reference to it.
21036 2018-04-11 Jakub Jelinek <jakub@redhat.com>
21039 * config/i386/sse.md (iptr): Add V16SFmode and V8DFmode cases.
21040 (<avx512>_vec_dup<mode><mask_name>): Use a single pattern for modes
21041 other than V2DFmode using iptr mode attribute.
21042 (<avx512>_vec_dup<mode><mask_name>): Use iptr mode attribute.
21044 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
21046 PR rtl-optimization/84659
21047 * sel-sched-ir.c (sel_init_pipelining): Invoke cleanup_cfg.
21049 2018-04-11 Jakub Jelinek <jakub@redhat.com>
21052 * dwarf2out.c (skip_loc_list_entry): Don't call size_of_locs if
21054 (output_loc_list): Pass address of a dummy size variable even in the
21055 locview handling loop.
21056 (index_location_lists): Add comment on why skip_loc_list_entry can't
21059 2018-04-11 Thomas Preud'homme <thomas.preudhomme@arm.com>
21062 * config/arm/arm-builtins.c (arm_expand_builtin): Force input operand
21065 2018-04-10 Aaron Sawdey <acsawdey@linux.ibm.com>
21068 * doc/invoke.texi (RS/6000 and PowerPC Options): Document options
21069 -mblock-compare-inline-limit, -mblock-compare-inline-loop-limit,
21070 and -mstring-compare-inline-limit.
21072 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
21075 * config/rs6000/rs6000.md (allocate_stack): Put the residual size
21076 for stack clash protection in a register whenever we need it to be in
21079 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
21081 * common/config/rs6000/rs6000-common.c (rs6000_option_init_struct):
21082 Enable -fasynchronous-unwind-tables by default if OBJECT_FORMAT_ELF.
21084 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
21087 * config/rs6000/rs6000.opt (mtraceback=): Show the allowed values in
21089 (mlong-double-): Ditto.
21090 * config/rs6000/sysv4.opt (msdata=): Ditto.
21091 (mtls-size=): Ditto.
21093 2018-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
21095 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
21096 erroneous entries for
21097 "vector int vec_ldl (int, long int *)", and
21098 "vector unsigned int vec_ldl (int, unsigned long int *)".
21099 Add comments and entries for
21100 "vector bool char vec_ldl (int, bool char *)",
21101 "vector bool short vec_ldl (int, bool short *)",
21102 "vector bool int vec_ldl (int, bool int *)",
21103 "vector bool long long vec_ldl (int, bool long long *)",
21104 "vector pixel vec_ldl (int, pixel *)",
21105 "vector long long vec_ldl (int, long long *)",
21106 "vector unsigned long long vec_ldl (int, unsigned long long *)".
21107 * config/rs6000/rs6000.c (rs6000_init_builtins): Initialize new
21108 type tree bool_long_long_type_node and correct definition of
21109 bool_V2DI_type_node to make reference to this new type tree.
21110 (rs6000_mangle_type): Replace erroneous reference to
21111 bool_long_type_node with bool_long_long_type_node.
21112 * config/rs6000/rs6000.h (enum rs6000_builtin_type_index): Add
21113 comments to emphasize sign distinctions for char and int types and
21114 replace RS6000_BTI_bool_long constant with
21115 RS6000_BTI_bool_long_long constant. Also add comment to restrict
21116 use of RS6000_BTI_pixel.
21117 (bool_long_type_node): Remove this macro definition.
21118 (bool_long_long_type_node): New macro definition
21120 2018-04-10 Jakub Jelinek <jakub@redhat.com>
21122 PR rtl-optimization/85300
21123 * combine.c (subst): Handle subst of CONST_SCALAR_INT_P new_rtx also
21124 into FLOAT and UNSIGNED_FLOAT like ZERO_EXTEND, return a CLOBBER if
21125 simplify_unary_operation fails.
21127 2018-04-10 Martin Liska <mliska@suse.cz>
21129 * gdbhooks.py: Add pretty-printers for varpool_node, symtab_node,
21130 cgraph_edge and ipa_ref.
21132 2018-04-10 Jakub Jelinek <jakub@redhat.com>
21136 * config/i386/sse.md
21137 (<extract_type>_vinsert<shuffletype><extract_suf>_mask): Fix
21138 computation of the VEC_MERGE selector from mask.
21139 (<extract_type>_vinsert<shuffletype><extract_suf>_1<mask_name>):
21140 Fix decoding of the VEC_MERGE selector into mask.
21142 2018-04-10 Richard Sandiford <richard.sandiford@linaro.org>
21144 PR tree-optimization/85286
21145 * tree-vect-data-refs.c (vect_get_smallest_scalar_type):
21147 2018-04-10 Richard Sandiford <richard.sandiford@linaro.org>
21149 * final.c (final_1): Set insn_last_address as well as
21150 insn_current_address.
21152 2018-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21155 * explow.c (emit_stack_probe): Call validize_mem on memory location
21156 before passing it to gen_probe_stack. Create address operand and
21157 legitimize it for the probe_stack_address case.
21159 2018-04-09 Jan Hubicka <jh@suse.cz>
21162 * ipa-devirt.c (rebuild_type_inheritance-hash): New.
21163 * ipa-utils.h (rebuild_type_inheritance-hash): Declare.
21164 * tree.c (free_lang_data_in_type): Fix handling of binfos;
21166 (free_lang_data): Rebuild type inheritance graph.
21168 2018-04-09 Martin Sebor <msebor@redhat.com>
21170 * invoke.texi (-finline-small-functions): Mention other optimization
21172 (-findirect-inlining, -fpartial-inlining): Same.
21173 (-finline-functions-called-once): Same.
21174 (-freorder-blocks-and-partition): Same.
21176 2018-04-09 Jan Hubicka <jh@suse.cz>
21179 * cfgcleanup.c (try_forward_edges): Do not give up on crossing
21180 jumps; choose last target that matches the criteria (i.e.
21181 no partition changes for non-crossing jumps).
21182 * cfgrtl.c (cfg_layout_redirect_edge_and_branch): Add basic
21183 support for redirecting crossing jumps to non-crossing.
21185 2018-04-09 Alexey Brodkin <abrodkin@synopsys.com>
21187 * config/arc/arc.c (arc_expand_prologue): Set stack usage info
21188 also for naked functions.
21190 2018-04-09 Claudiu Zissulescu <claziss@synopsys.com>
21192 * config/arc/arc.md (add_shift): New pattern.
21193 (add_shift2): Likewise.
21194 (sub_shift): Likewise.
21195 (sub_shift_cmp0_noout): Likewise.
21196 (compare_si_ashiftsi): Likewise.
21197 (xbfu_cmp0_noout): New combine pattern.
21198 (xbfu_cmp0"): Likewise.
21199 (movsi_set_cc_insn): Place the predicable variant first.
21200 (commutative_binary_cmp0_noout): Remove clobber.
21201 (commutative_binary_cmp0): New pattern.
21202 (noncommutative_binary_cmp0): Likewise.
21203 (noncommutative_binary_cmp0_noout): Likewise.
21204 (noncommutative_binary_comparison_result_used): Removed.
21205 (rsub_cmp0): New pattern.
21206 (rsub_cmp0_noout): Likewise.
21207 (extzvsi): Changed, keep only meaningful variants.
21208 (SQH, SEZ): New iterators.
21209 (SQH_postfix): New mode attribute.
21210 (SEZ_prefix): New code attribute.
21211 (<SEZ_prefix>xt<SQH_postfix>_cmp0_noout): New instruction pattern.
21212 (<SEZ_prefix>xt<SQH_postfix>_cmp0): Likewise.
21213 * config/arc/predicates.md (cc_set_register): Use CC_REG instead
21214 of numerical value.
21215 (noncommutative_operator): Check the availability of barrel
21218 2018-04-09 Richard Biener <rguenther@suse.de>
21220 PR tree-optimization/85284
21221 * tree-ssa-loop-niter.c (number_of_iterations_exit_assumptions):
21222 Only use the niter constraining form of simple_iv when the exit
21223 is always executed.
21225 2018-04-09 Tom de Vries <tom@codesourcery.com>
21228 * config/nvptx/nvptx.md (define_c_enum "unspecv"): Add UNSPECV_MEMBAR.
21229 (define_expand "*memory_barrier"): New define_expand.
21230 (define_insn "memory_barrier"): New insn.
21232 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
21234 PR rtl-optimization/80463
21235 PR rtl-optimization/83972
21236 PR rtl-optimization/83480
21238 * sel-sched-ir.c (has_dependence_note_mem_dep): Take into account the
21239 correct producer for the insn.
21240 (tidy_control_flow): Fixup seqnos in case of debug insns.
21242 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
21244 PR rtl-optimization/83913
21246 * sel-sched-ir.c (merge_expr_data): Choose the middle between two
21247 different sched-times when merging exprs.
21249 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
21251 PR rtl-optimization/83962
21253 * sel-sched-ir.c (tidy_control_flow): Correct the order in which we call
21254 tidy_fallthru_edge and tidy_control_flow.
21256 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
21258 PR rtl-optimization/83530
21260 * sel-sched.c (force_next_insn): New global variable.
21261 (remove_insn_for_debug): When force_next_insn is true, also leave only
21262 next insn in the ready list.
21263 (sel_sched_region): When the region wasn't scheduled, make another pass
21264 over it with force_next_insn set to 1.
21266 2018-04-08 Monk Chiang <sh.chiang04@gmail.com>
21268 * config.gcc (nds32le-*-*, nds32be-*-*): Add nds32/nds32_intrinsic.h
21270 * config/nds32/constants.md (unspec_volatile_element): Add enum values
21271 for interrupt control.
21272 * config/nds32/nds32-intrinsic.c: Implementation of intrinsic
21273 functions for interrupt control.
21274 * config/nds32/nds32-intrinsic.md: Likewise.
21275 * config/nds32/nds32_intrinsic.h: Likewise.
21276 * config/nds32/nds32.h (nds32_builtins): Likewise.
21278 2018-04-08 Chung-Ju Wu <jasonwucj@gmail.com>
21280 * config/nds32/nds32.c (nds32_init_machine_status,
21281 nds32_legitimate_index_p, nds32_legitimate_address_p): Consider
21282 strict_aligned_p field.
21283 (nds32_expand_to_rtl_hook): New function.
21284 (TARGET_EXPAND_TO_RTL_HOOK): Define.
21285 * config/nds32/nds32.h (machine_function): Add strict_aligned_p field.
21287 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
21288 Chung-Ju Wu <jasonwucj@gmail.com>
21290 * config.gcc (nds32*-*-*): Check that n7 is valid to --with-cpu.
21291 * config/nds32/nds32-n7.md: New file.
21292 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N7.
21293 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n7
21295 * config/nds32/nds32-protos.h: More declarations for n7 pipeline.
21296 * config/nds32/nds32.md (pipeline_model): Add n7.
21297 * config/nds32/nds32.opt (mcpu): Support n7 pipeline cpus.
21298 * config/nds32/pipelines.md: Include n7 settings.
21300 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
21301 Chung-Ju Wu <jasonwucj@gmail.com>
21303 * config.gcc (nds32*-*-*): Check that e8 is valid to --with-cpu.
21304 * config/nds32/nds32-e8.md: New file.
21305 * config/nds32/nds32-opts.h (nds32-cpu_type): Add CPU_E8.
21306 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for e8
21308 * config/nds32/nds32-protos.h: More declarations for e8 pipeline.
21309 * config/nds32/nds32.md (pipeline_model): Add e8.
21310 * config/nds32/nds32.opt (mcpu): Support e8 pipeline cpus.
21311 * config/nds32/pipelines.md: Include e8 settings.
21313 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
21314 Chung-Ju Wu <jasonwucj@gmail.com>
21316 * config.gcc (nds32*-*-*): Check that n6/n8/s8 are valid to --with-cpu.
21317 * config/nds32/nds32-n8.md: New file.
21318 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N6 and CPU_N8.
21319 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n8
21321 * config/nds32/nds32-protos.h: More declarations for n8 pipeline.
21322 * config/nds32/nds32-utils.c: More implementations for n8 pipeline.
21323 * config/nds32/nds32.md (pipeline_model): Add n8.
21324 * config/nds32/nds32.opt (mcpu): Support n8 pipeline cpus.
21325 * config/nds32/pipelines.md: Include n8 settings.
21327 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
21328 Chung-Ju Wu <jasonwucj@gmail.com>
21330 * config.gcc (nds32*): Add nds32-utils.o into extra_objs.
21331 * config/nds32/nds32-n9-2r1w.md: New file.
21332 * config/nds32/nds32-n9-3r2w.md: New file.
21333 * config/nds32/nds32-opts.h (nds32_cpu_type, nds32_mul_type,
21334 nds32_register_ports): New or modify for cpu n9.
21335 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n9
21337 * config/nds32/nds32-protos.h: More declarations for n9 pipeline.
21338 * config/nds32/nds32-utils.c: New file.
21339 * config/nds32/nds32.h (TARGET_PIPELINE_N9, TARGET_PIPELINE_SIMPLE,
21340 TARGET_MUL_SLOW): Define.
21341 * config/nds32/nds32.md (pipeline_model): New attribute.
21342 * config/nds32/nds32.opt (mcpu, mconfig-mul, mconfig-register-ports):
21343 New options that support cpu n9.
21344 * config/nds32/pipelines.md: Include n9 settings.
21345 * config/nds32/t-nds32 (nds32-utils.o): Add dependency.
21347 2018-04-08 Chung-Ju Wu <jasonwucj@gmail.com>
21349 * config/nds32/nds32-md-auxiliary.c (output_cond_branch): Output align
21350 information if necessary.
21351 (output_cond_branch_compare_zero): Likewise.
21352 * config/nds32/nds32.c (nds32_adjust_insn_length): Consider align case.
21353 (nds32_target_alignment): Refine for alignment.
21354 * config/nds32/nds32.h (NDS32_ALIGN_P): Define.
21355 (FUNCTION_BOUNDARY): Modify.
21356 * config/nds32/nds32.md (call_internal, call_value_internal): Consider
21358 * config/nds32/nds32.opt (malways-align, malign-functions): New.
21360 2018-04-08 Monk Chiang <sh.chiang04@gmail.com>
21362 * config/nds32/constants.md (unspec_volatile_element): Add values for
21363 TLB operation and data prefetch.
21364 * config/nds32/nds32-intrinsic.c: Implementation of intrinsic
21365 functions for TLB operation and data prefetch.
21366 * config/nds32/nds32-intrinsic.md: Likewise.
21367 * config/nds32/nds32_intrinsic.h: Likewise.
21368 * config/nds32/nds32.c (nds32_dpref_names): Likewise.
21369 (nds32_print_operand): Likewise.
21370 * config/nds32/nds32.h (nds32_builtins): Likewise.
21372 2018-04-07 Thomas Koenig <tkoenig@gcc.gnu.org>
21373 Andrew Pinski <pinsika@gcc.gnu.org>
21375 PR middle-end/82976
21376 * match.pd: Use constant_boolean_node of correct type instead of
21377 boolean_true_node or boolean_false_node for simplifying
21378 pointer comparisons to zero.
21380 2018-04-07 Jakub Jelinek <jakub@redhat.com>
21382 PR tree-optimization/80021
21383 * tree.c (verify_type_variant): Make error call in verify_variant_match
21384 translatable and remove final full stop.
21386 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
21388 * config/nds32/constants.md (unspec_volatile_element): Add
21389 UNSPEC_VOLATILE_EH_RETURN.
21390 * config/nds32/nds32-md-auxiliary.c (nds32_output_stack_push,
21391 nds32_output_stack_pop): Support dwarf exception handling process.
21392 * config/nds32/nds32-protos.h (nds32_dynamic_chain_address): Declare.
21393 * config/nds32/nds32.c (nds32_init_machine_status): Support dwarf
21394 exception handling process.
21395 (nds32_compute_stack_frame): Likewise.
21396 (nds32_return_addr_rtx): Likewise.
21397 (nds32_initial_elimination_offset): Likewise.
21398 (nds32_expand_prologue): Likewise.
21399 (nds32_expand_epilogue): Likewise.
21400 (nds32_dynamic_chain_address): New function.
21401 * config/nds32/nds32.h (machine_function): Add fields for dwarf
21402 exception handling.
21403 (DYNAMIC_CHAIN_ADDRESS): Define.
21404 (EH_RETURN_DATA_REGNO): Define.
21405 (EH_RETURN_STACKADJ_RTX): Define.
21406 * config/nds32/nds32.md (eh_return, nds32_eh_return): Implement
21407 patterns for dwarf exception handling.
21409 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
21411 * config/nds32/nds32.h: Clean up obsolete macros.
21413 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
21415 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
21416 Add enum values for particular instructions.
21417 * config/nds32/nds32-intrinsic.c: Implementation of expanding
21418 particular intrinsic functions.
21419 * config/nds32/nds32-intrinsic.md: Likewise.
21420 * config/nds32/nds32_intrinsic.h: Likewise.
21421 * config/nds32/nds32.h (nds32_builtins): Likewise.
21422 * config/nds32/nds32.md (type): Add pbsad and pbsada.
21423 (btst, ave): New patterns for particular instructions.
21425 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
21427 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
21428 Add enum values for atomic load/store and memory sync.
21429 * config/nds32/nds32-intrinsic.c: Implementation for atomic load/store
21431 * config/nds32/nds32-intrinsic.md: Likewise.
21432 * config/nds32/nds32_intrinsic.h: Likewise.
21433 * config/nds32/nds32.h (nds32_builtins): Likewise.
21435 2018-04-07 Jakub Jelinek <jakub@redhat.com>
21437 PR tree-optimization/85257
21438 * fold-const.c (native_encode_vector): If not all elts could fit
21439 and off is -1, return 0 rather than offset.
21440 * tree-ssa-sccvn.c (vn_reference_lookup_3): Pass
21441 (offseti - offset2) / BITS_PER_UNIT as 4th argument to
21442 native_encode_expr. Verify len * BITS_PER_UNIT >= maxsizei. Don't
21443 adjust buffer in native_interpret_expr call.
21445 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
21447 * config/nds32/constants.md (unspec_volatile_element): Add cache
21448 control enum values.
21449 * config/nds32/nds32-intrinsic.c: Add cache control expand functions.
21450 * config/nds32/nds32-intrinsic.md: Add cache control patterns.
21451 * config/nds32/nds32.c (nds32_cctl_names): New.
21452 (nds32_print_operand): Handle cache control register names.
21453 * config/nds32/nds32.h (nds32_builtins): New enum values.
21454 * config/nds32/nds32_intrinsic.h: Add cache control enum types and
21456 * config/nds32/nds32.md (type): Add mmu.
21457 * config/nds32/pipelines.md (simple_insn): Add mmu.
21459 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
21461 * config/nds32/nds32.md (type): Remove call.
21462 * config/nds32/pipelines.md (simple_insn): Likewise.
21464 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
21466 * config/nds32/constants.md (unspec_volatile_element): Add
21467 UNSPEC_VOLATILE_FMFCSR, UNSPEC_VOLATILE_FMTCSR and
21468 UNSPEC_VOLATILE_FMFCFG.
21469 * config/nds32/nds32-intrinsic.c (bdesc_noarg): New builtin
21470 description for fmfcfg and fmfcsr.
21471 (bdesc_1arg): Add fmtcsr.
21472 (bdesc_2arg): Add fcpynss, fcpyss, fcpynsd and fcpysd.
21473 (nds32_expand_builtin_impl): Deal with FPU intrinsic functions.
21474 * config/nds32/nds32-intrinsic.md (unspec_fcpynsd, unspec_fcpysd,
21475 unspec_fcpynss, unspec_fcpysd, unspec_fcpyss, unspec_fmfcsr,
21476 unspec_fmfcfg): New patterns.
21477 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_FMFCFG,
21478 NDS32_BUILTIN_FMFCSR, NDS32_BUILTIN_FMTCSR, NDS32_BUILTIN_FCPYNSS,
21479 NDS32_BUILTIN_FCPYSS,NDS32_BUILTIN_FCPYNSD and NDS32_BUILTIN_FCPYSD.
21480 * config/nds32/nds32_intrinsic.h (__nds32__fcpynsd, __nds32__fcpynss,
21481 __nds32__fcpysd, __nds32__fcpyss, __nds32__fmfcsr, __nds32__fmtcsr,
21482 __nds32__fmfcfg): Define.
21484 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
21486 * config/nds32/nds32.c (nds32_intrinsic_register_names): Add more
21487 intrinsic register names.
21488 * config/nds32/nds32_intrinsic.h (nds32_intrinsic_registers): Add more
21489 intrinsic register enum values and macros.
21491 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
21493 * config/nds32/nds32.c (nds32_legitimate_index_p): Modify condition
21494 for load/store addressing form.
21495 (nds32_print_operand_address): Likewise.
21497 2018-04-06 Eric Botcazou <ebotcazou@adacore.com>
21500 * config/sparc/sparc.c (sparc_expand_move): Deal with symbolic operands
21501 based on LABEL_REF. Remove useless assertion.
21502 (pic_address_needs_scratch): Fix formatting.
21503 (sparc_legitimize_pic_address): Minor tweaks.
21504 (sparc_delegitimize_address): Adjust assertion accordingly.
21505 * config/sparc/sparc.md (movsi_pic_label_ref): Change label_ref_operand
21506 into symbolic_operand.
21507 (movsi_high_pic_label_ref): Likewise.
21508 (movsi_lo_sum_pic_label_ref): Likewise.
21509 (movdi_pic_label_ref): Likewise.
21510 (movdi_high_pic_label_ref): Likewise.
21511 (movdi_lo_sum_pic_label_ref): Likewise.
21513 2018-04-06 Amaan Cheval <amaan.cheval@gmail.com>
21515 * config.gcc (x86_64-*-rtems*): Add rtems.h to tm_file for
21516 custom LIB_SPEC setup.
21518 2018-04-06 Ruslan Bukin <br@bsdpad.com>
21519 Kito Cheng <kito.cheng@gmail.com>
21521 * config.gcc (riscv*-*-freebsd*): Add RISC-V FreeBSD support.
21522 * config/riscv/freebsd.h: New.
21524 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
21526 * config/nds32/nds32.c (nds32_adjust_insn_length): Refine.
21527 * config/nds32/nds32.h (ADJUST_INSN_LENGTH): Change the location in
21530 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
21531 Kito Cheng <kito.cheng@gmail.com>
21533 * config/nds32/nds32-md-auxiliary.c (nds32_output_return,
21534 nds32_output_call, nds32_symbol_binds_local_p): New functions.
21535 * config/nds32/nds32-protos.h (nds32_output_call,
21536 nds32_output_return): Declare.
21537 * config/nds32/nds32.md: Refine all the call and return patterns.
21539 2018-04-06 Jakub Jelinek <jakub@redhat.com>
21542 * dwarf2out.c (rtl_for_decl_init): For STRING_CST initializer only
21543 build CONST_STRING if TYPE_MAX_VALUE is non-NULL and is INTEGER_CST.
21545 PR rtl-optimization/84872
21546 * cfgloopmanip.c (create_preheader): Use make_forwarder_block even if
21547 nentry == 1 when CP_FALLTHRU_PREHEADERS and single_entry is
21548 EDGE_CROSSING edge.
21550 2018-04-06 Tamar Christina <tamar.christina@arm.com>
21552 * expr.c (copy_blkmode_to_reg): Revert 254862.
21553 * doc/sourcebuild.texi (word_mode_no_slow_unalign): Likewise.
21555 2018-04-06 Richard Biener <rguenther@suse.de>
21557 PR middle-end/85244
21558 * tree-dfa.c (get_ref_base_and_extent): Reset seen_variable_array_ref
21559 after seeing a component reference with an adjacent field. Treat
21560 refs to arrays at struct end of external decls similar to
21561 refs to unconstrained commons.
21563 2018-04-06 Jakub Jelinek <jakub@redhat.com>
21566 * fold-const.c (twoval_comparison_p): Remove SAVE_P argument and don't
21567 look through SAVE_EXPRs with non-side-effects argument. Adjust
21569 (fold_comparison): Adjust twoval_comparison_p caller, don't handle
21572 2018-04-06 Richard Biener <rguenther@suse.de>
21574 PR middle-end/85180
21575 * alias.c (find_base_term): New wrapper around find_base_term
21576 unwinding CSELIB_VAL_PTR changes.
21577 (find_base_term): Do not restore CSELIB_VAL_PTR during the
21580 2018-04-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21582 * config/s390/s390.c (s390_z10_optimize_cmp): Expand dedicated NOP
21584 * config/s390/s390.md (UNSPECV_NOP_LR_0, UNSPECV_NOP_LR_1): New
21585 constant definitions.
21586 ("nop"): lr 0,0 -> nopr r0
21587 ("nop_lr0", "nop_lr1"): New insn definitions.
21589 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
21591 * config/nds32/nds32.md (*stack_push, *stack_pop): Use
21592 NDS32_V3PUSH_AVAILABLE_P macro.
21594 2018-04-06 Monk Chiang <sh.chiang04@gmail.com>
21595 Chung-Ju Wu <jasonwucj@gmail.com>
21597 * config.gcc (nds32*-*-*): Add v2j v3f v3s checking.
21598 (nds32*-*-*): Add float and fpu_config into supported_defaults.
21599 * common/config/nds32/nds32-common.c (TARGET_DEFAULT_TARGET_FLAGS):
21600 Include TARGET_DEFAULT_FPU_ISA and TARGET_DEFAULT_FPU_FMA.
21601 * config/nds32/constants.md (unspec_element): Add UNSPEC_COPYSIGN,
21602 UNSPEC_FCPYNSD, UNSPEC_FCPYNSS, UNSPEC_FCPYSD and UNSPEC_FCPYSS.
21603 * config/nds32/constraints.md: New constraints and checking for hard
21604 float configuration.
21605 * config/nds32/iterators.md: New mode iterator and attribute for hard
21606 float configuration.
21607 * config/nds32/nds32-doubleword.md: Use hard float alternatives and
21609 * config/nds32/nds32-fpu.md: New file.
21610 * config/nds32/nds32-md-auxiliary.c: New functions and checkings to
21611 deal with hard float code generation.
21612 * config/nds32/nds32-opts.h (nds32_arch_type): Add ARCH_V3F and
21614 (abi_type, float_reg_number): New enum type.
21615 * config/nds32/nds32-predicates.c: New predicates for hard float.
21616 * config/nds32/nds32-protos.h: Declare functions for hard float.
21617 * config/nds32/nds32.c: Implementation for hard float configuration.
21618 * config/nds32/nds32.h: Definitions for hard float configuration.
21619 * config/nds32/nds32.md: Include hard float machine description and
21620 modify patterns for hard float configuration.
21621 * config/nds32/nds32.opt: New options for hard float configuration.
21622 * config/nds32/predicates.md: New predicates for hard float
21625 2018-04-06 Kuan-Lin Chen <kuanlinchentw@gmail.com>
21627 * common/config/nds32/nds32-common.c
21628 (nds32_option_optimization_table): Enable -mreleax-hint by default.
21630 2018-04-05 Jakub Jelinek <jakub@redhat.com>
21632 PR middle-end/85195
21633 * match.pd (BIT_FIELD_REF CONSTRUCTOR@0 @1 @2): Use view_convert around
21634 CONSTRUCTOR_ELT (ctor, ...)->value.
21636 2018-04-05 Uros Bizjak <ubizjak@gmail.com>
21639 * config/i386/i386.md (define_attr "memory"): Handle rotate1 type.
21641 2018-04-05 Tom de Vries <tom@codesourcery.com>
21644 * config/nvptx/nvptx.c (nvptx_single): Fix neutering of bb with only
21647 2018-04-05 Shiva Chen <shiva0217@gmail.com>
21648 Kito Cheng <kito.cheng@gmail.com>
21650 * config/nds32/constraints.md (U33): Fine-tune checking condition.
21651 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Ditto.
21652 * config/nds32/nds32.h (nds32_16bit_address_type): Add
21653 ADDRESS_POST_MODIFY_LO_REG_IMM3U.
21655 2018-04-05 Shiva Chen <shiva0217@gmail.com>
21656 Kito Cheng <kito.cheng@gmail.com>
21658 * config/nds32/constraints.md (Ufe): New memory constraint.
21659 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format,
21660 nds32_output_16bit_load): Consider r8 register for lwi45.fe format.
21661 * config/nds32/nds32.c (nds32_print_operand): Output lwi45.fe
21663 * config/nds32/nds32.h (nds32_16bit_address_type): Add ADDRESS_R8_IMM7U.
21664 * config/nds32/nds32.md (*mov<mode>): Adjust pattern.
21666 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
21668 * config/nds32/nds32.md: Use optimize_size in the condition for
21669 alu-shift instructions.
21671 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
21673 * config/nds32/nds32.md (divsi4, udivsi4): New patterns.
21675 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
21677 * config/nds32/nds32.md (negsi2): Refine pattern.
21679 2018-04-05 Kito Cheng <kito.cheng@gmail.com>
21680 Chung-Ju Wu <jasonwucj@gmail.com>
21682 * config/nds32/iterators.md (shift_rotate): New code iterator.
21683 (shift): New code attribute.
21684 * config/nds32/nds32-md-auxiliary.c (nds32_expand_constant): New.
21685 * config/nds32/nds32-protos.h (nds32_expand_constant): Declare.
21686 * config/nds32/nds32.c (nds32_print_operand): Deal with more cases.
21687 * config/nds32/nds32.md (addsi3, *add_srli): Refine implementation for
21688 bit-wise operations.
21689 (andsi3, *andsi3): Ditto.
21690 (iorsi3, *iorsi3, *or_slli, *or_srli): Ditto.
21691 (xorsi3, *xorsi3, *xor_slli, *xor_srli): Ditto.
21692 (<shift>si3, *ashlsi3, *ashrsi3, *lshrsi3, *rotrsi3): Ditto.
21693 * config/nds32/predicates.md (nds32_rimm5u_operand, nds32_and_operand,
21694 nds32_ior_operand, nds32_xor_operand): New predicates.
21696 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
21698 * config/nds32/nds32.md (add<mode>3, sub<mode>3): Rename to ...
21699 (addsi3, subsi3): ... this.
21701 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
21703 * config/nds32/nds32.md (*sub_srli, *and_slli): Fine-tune predicator.
21705 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
21707 * config/nds32/nds32.md: Adjust indention.
21709 2018-04-05 Kito Cheng <kito.cheng@gmail.com>
21711 * config/nds32/nds32.md (feature): New attribute.
21713 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
21715 * config/nds32/nds32.md (subtype): New attribute.
21717 2018-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
21720 * config/arm/arm-builtins.c (arm_expand_builtin): Change
21721 expansion to perform a bitwise AND of the argument followed by a
21722 boolean negation of the result.
21724 2018-04-04 Peter Bergner <bergner@vnet.ibm.com>
21726 PR rtl-optimization/84878
21727 * ddg.c (add_cross_iteration_register_deps): Use DF_REF_BB to determine
21728 the basic block. Assert the use reference is not artificial and that
21729 it has an associated insn.
21731 2018-04-04 Michael Matz <matz@suse.de>
21733 * builtins.c (compute_objsize): Pass correct operand
21734 to array_at_struct_end_p.
21736 2018-04-04 Richard Biener <rguenther@suse.de>
21739 * dwarf2out.c (dwarf2out_register_external_die): Peel namespaces
21740 from contexts for DINFO_LEVEL_TERSE and below.
21742 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
21744 * config/nds32/nds32-doubleword.md (move_<mode>): Require
21745 resiter_operand condition.
21746 * config/nds32/nds32.md (*move<mode>): Ditto.
21748 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
21749 Monk Chiang <sh.chiang04@gmail.com>
21751 * config/nds32/nds32.md (movmisalign<mode>): New pattern.
21753 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
21755 * config/nds32/nds32.md (movqi, movhi): Merge into mov<mode>.
21757 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
21758 Kito Cheng <kito.cheng@gmail.com>
21760 * config/nds32/nds32-md-auxiliary.c (nds32_inverse_cond_code,
21761 nds32_cond_code_str, output_cond_branch,
21762 output_cond_branch_compare_zero, nds32_expand_cbranch,
21763 nds32_expand_cstore, nds32_expand_movcc,
21764 nds32_output_cbranchsi4_equality_zero,
21765 nds32_output_cbranchsi4_equality_reg,
21766 nds32_output_cbranchsi4_equality_reg_or_const_int,
21767 nds32_output_cbranchsi4_greater_less_zero: New functions.
21768 * config/nds32/nds32-protos.h (nds32_expand_cbranch,
21769 nds32_expand_cstore, nds32_expand_movcc,
21770 nds32_output_cbranchsi4_equality_zero,
21771 nds32_output_cbranchsi4_equality_reg,
21772 nds32_output_cbranchsi4_equality_reg_or_const_int,
21773 nds32_output_cbranchsi4_greater_less_zero): Declare.
21774 * config/nds32/predicates.md (nds32_movecc_comparison_operator,
21775 nds32_rimm11s_operand): New predicates.
21776 * config/nds32/nds32.h (nds32_expand_result_type): New enum type.
21777 * config/nds32/nds32.md: Rewrite all the branch and conditional move
21780 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
21782 * config/nds32/nds32-doubleword.md: Refine all the instruction type.
21783 * config/nds32/nds32.md: Ditto.
21784 * config/nds32/pipelines.md: Ditto.
21786 2018-04-04 Richard Biener <rguenther@suse.de>
21788 PR tree-optimization/85168
21789 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address): Avoid
21790 propagating abnormals.
21792 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
21794 * config/nds32/nds32.md (enabled): Use yes/no for this attribute.
21796 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
21797 Kito Cheng <kito.cheng@gmail.com>
21799 * config/nds32/nds32-md-auxiliary.c (nds32_long_call_p): New function.
21800 * config/nds32/nds32-protos.h (nds32_long_call_p): Declare.
21801 * config/nds32/nds32.c (nds32_function_ok_for_sibcall): New function.
21802 (TARGET_FUNCTION_OK_FOR_SIBCALL): Define.
21803 * config/nds32/nds32.md (sibcall_internal): New.
21804 (sibcall_register): Remove.
21805 (sibcall_immediate): Remove.
21806 (sibcall_value_internal): New.
21807 (sibcall_value_register): Remove.
21808 (sibcall_value_immediate): Remove.
21809 * config/nds32/predicates.md (nds32_general_register_operand): New.
21810 (nds32_call_address_operand): New.
21812 2018-04-03 Jakub Jelinek <jakub@redhat.com>
21814 PR rtl-optimization/85167
21815 * shrink-wrap.c (move_insn_for_shrink_wrap): Don't set bb_uses and
21816 bb_defs if *split_p, instead preinitialize it to NULL.
21818 PR tree-optimization/85156
21819 * builtins.c (fold_builtin_expect): Use save_expr on arg1 to avoid
21820 evaluating the argument multiple times.
21822 2018-04-03 Bill Schmidt <wschmidt@linux.ibm.com>
21824 * config/rs6000/emmintrin.h (_mm_cvtpd_epi32): Use __vector rather
21826 (_mm_cvtpd_ps): Likewise.
21827 (_mm_cvttpd_epi32): Likewise.
21828 * config/rs6000/mmintrin.h (_mm_unpacklo_pi8): Likewise.
21829 * config/rs6000/xmmintrin.h: For strict-ANSI C++ or C11, undefine
21830 vector, pixel, and bool following altivec.h include.
21832 2018-04-03 Martin Sebor <msebor@redhat.com>
21834 * doc/extend.texi (Common Function Attributes): Clarify.
21835 (const attribute): Likewise.
21836 (pure attribute): Likewise.
21838 2018-04-03 Jakub Jelinek <jakub@redhat.com>
21841 * config/i386/i386.c (ix86_expand_vector_set): Use
21842 HOST_WIDE_INT_1U << elt instead of 1 << elt. Formatting fix.
21844 2018-04-03 Uros Bizjak <ubizjak@gmail.com>
21846 * config/i386/i386.c (emit_i387_cw_initialization): Always use logic
21847 instructions when changing rounding bits to preserve precision bits
21848 in the x87 control word.
21850 2018-04-03 Martin Liska <mliska@suse.cz>
21852 PR tree-optimization/82491
21853 * rtl.h (strip_offset_and_add): Replace += suboffset with
21854 poly_uint64 () + suboffset.
21856 2018-03-29 Martin Liska <mliska@suse.cz>
21857 Martin Jambor <mjambor@suse.cz>
21860 * ipa-cp.c (propagate_bits_across_jump_function): Bail out if
21861 param_type is not an integral or pointer type.
21863 2018-04-03 Richard Biener <rguenther@suse.de>
21865 * sese.h (recompute_all_dominators): Remove.
21867 2018-04-02 Martin Sebor <msebor@redhat.com>
21869 * doc/invoke.texi (-Wrestrict): Fix typos.
21871 2018-04-02 Jim Wilson <jimw@sifive.com>
21873 * config/riscv/riscv.h (SHIFT_COUNT_TRUNCATED): Set to zero.
21874 * config/riscv/riscv.md (<optab>si3): Use QImode shift count.
21875 (<optab>di3, <optab>si3_extend): Likewise.
21876 (<optab>si3_mask, <optab>si3_mask_1): New.
21877 (<optab>di3_mask, <optab>di3_mask_1): New.
21878 (<optab>si3_extend_mask, <optab>si3_extend_mask_1): New.
21879 (lshrsi3_zero_extend_1): Use VOIDmode shift count.
21880 * config/riscv/sync.md (atomic_test_and_set): Emit QImode shift count.
21882 2018-04-02 Gerald Pfeifer <gerald@pfeifer.com>
21884 * doc/cpp.texi (Variadic Macros): Fix line continuation in an
21887 2018-04-02 Chung-Ju Wu <jasonwucj@gmail.com>
21889 * config/nds32/nds32.c (TARGET_CANONICALIZE_COMPARISON): Define.
21890 (nds32_canonicalize_comparison): New function.
21892 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
21893 Kito Cheng <kito.cheng@gmail.com>
21894 Kuan-Lin Chen <kuanlinchentw@gmail.com>
21896 * config.gcc (nds32): Add nds32-relax-opt.o into extra_objs.
21897 * config/nds32/constants.md (unspec_volatile_element): Add
21898 UNSPEC_VOLATILE_RELAX_GROUP.
21899 * config/nds32/nds32-relax-opt.c: New file.
21900 * config/nds32/nds32-predicates.c
21901 (nds32_symbol_load_store_p): New function.
21902 * config/nds32/nds32-protos.h
21903 (nds32_symbol_load_store_p): Declare function.
21904 (make_pass_nds32_relax_opt): Declare new rtl pass function.
21905 * config/nds32/nds32.c
21906 (nds32_register_pass): New function to register pass.
21907 (nds32_register_passes): New function to register passes.
21908 * config/nds32/nds32.md (relax_group): New pattern.
21909 * config/nds32/nds32.opt (mrelax-hint): New option.
21910 * config/nds32/t-nds32 (nds32-relax-opt.o): New dependency.
21912 2018-04-01 Kito Cheng <kito.cheng@gmail.com>
21914 * config/nds32/t-nds32: Modify files dependency.
21916 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
21918 * config/nds32/nds32.h (FUNCTION_PROFILER): Output newline character.
21919 (PROFILE_HOOK): Define its implementation.
21921 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
21923 * config/nds32/nds32.h (WCHAR_TYPE, WCHAR_TYPE_SIZE): Use unsigned int
21924 type and 32-bit size.
21926 2018-04-01 Jakub Jelinek <jakub@redhat.com>
21928 PR middle-end/85090
21929 * config/i386/sse.md (V): Add V64QI and V32HI for TARGET_AVX512F.
21930 (V_128_256): New mode iterator.
21931 (*avx512dq_vextract<shuffletype>64x2_1 splitter): New define_split.
21932 (*avx512f_vextract<shuffletype>32x4_1 splitter): Likewise.
21933 (xop_pcmov_<mode><avxsizesuffix>): Use V_128_256 mode iterator instead
21935 * config/i386/i386.c (ix86_expand_vector_set): Improve V32HImode and
21936 V64QImode expansion for !TARGET_AVX512BW && TARGET_AVX512F.
21938 2018-03-31 Segher Boessenkool <segher@kernel.crashing.org>
21941 * config/rs6000/xmmintrin.h (_mm_set_ps, _mm_max_ps): Handle (quiet)
21942 NaN inputs correctly.
21944 2018-03-30 Peter Bergner <bergner@vnet.ibm.com>
21947 * config/rs6000/vsx.md (??r): New mode attribute.
21948 (*vsx_mov<mode>_64bit): Use it.
21949 (*vsx_mov<mode>_32bit): Likewise.
21951 2018-03-30 Martin Sebor <msebor@redhat.com>
21953 PR tree-optimization/84818
21954 * builtins.c (check_access): Use warning_n.
21956 2018-03-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21959 * config/rs6000/rs6000-string.c (expand_compare_loop): Fix redundant
21961 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Fix redundant
21964 2018-03-30 Julia Koval <julia.koval@intel.com>
21967 * x86-tune.def (movx, partial_reg_dependency): Enable for
21970 2018-03-29 Vladimir Makarov <vmakarov@redhat.com>
21972 PR inline-asm/84985
21973 * lra-constraints.c (process_alt_operands): Move setting
21974 this_alternative_matches below.
21976 2018-03-29 Martin Liska <mliska@suse.cz>
21979 * doc/invoke.texi: Document how LTO works with debug info.
21980 Describe auto-load support of binutils. Mention 'x86-64'
21981 as valid option value of -march option.
21983 2018-03-29 Jakub Jelinek <jakub@redhat.com>
21985 * config/i386/sse.md (<avx512>_blendm<mode>): Use <sseintprefix>.
21988 * fold-const.c (operand_equal_p): Handle DEBUG_BEGIN_STMT.
21989 For STATEMENT_LIST, pass down OEP_LEXICOGRAPHIC and maybe
21990 OEP_NO_HASH_CHECK for recursive call, to avoid exponential
21993 2018-03-28 Peter Bergner <bergner@vnet.ibm.com>
21996 * config/rs6000/rs6000.h (RS6000_BTM_POWERPC64): New define.
21997 (RS6000_BTM_COMMON): Add RS6000_BTM_POWERPC64.
21998 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add support
21999 for RS6000_BTM_POWERPC64.
22000 (rs6000_invalid_builtin): Add handling for RS6000_BTM_POWERPC64
22001 (rs6000_builtin_mask_names): Add RS6000_BTM_POWERPC64.
22002 * config/rs6000/rs6000-builtin.def (BU_P7_POWERPC64_MISC_2): New macro
22005 (DIVDEU): Likewise.
22007 2018-03-28 Carl Love <cel@us.ibm.com>
22010 2017-09-27 Carl Love <cel@us.ibm.com>
22012 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
22013 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
22014 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
22017 2018-03-28 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22019 * config/rs6000/xmmintrin.h (_mm_max_pi16): Use __vector __bool
22020 instead of __vector bool.
22021 (_mm_max_pu8): Likewise.
22022 (_mm_min_pi16): Likewise.
22024 2018-03-28 Peter Bergner <bergner@vnet.ibm.com>
22027 * config/rs6000/rs6000-builtin.def (DIVWEO): Delete macro expansion.
22028 (DIVWEUO): Likewise.
22029 (DIVDEO): Likewise.
22030 (DIVDEUO): Likewise.
22031 * config/rs6000/rs6000.c (builtin_function_type): Remove support for
22032 DIVWEUO and DIVDEUO.
22033 * config/rs6000/rs6000.md (UNSPEC_DIVEO, UNSPEC_DIVEUO): Delete unspecs.
22034 (UNSPEC_DIV_EXTEND): Remove deleted unspecs.
22035 (div_extend): Likewise.
22036 * doc/extend.texi (__builtin_divweo): Remove documentation for deleted
22038 (__builtin_divweuo): Likewise.
22039 (__builtin_divdeo): Likewise.
22040 (__builtin_divdeuo): Likewise.
22042 2018-03-28 Jakub Jelinek <jakub@redhat.com>
22045 * config/i386/i386.md (*add<mode>3_carry_0, *addsi3_carry_zext_0,
22046 *sub<mode>3_carry_0, *subsi3_carry_zext_0): New patterns.
22048 PR tree-optimization/82004
22049 * gimple-match-head.c (optimize_pow_to_exp): New function.
22050 * match.pd (pow(C,x) -> exp(log(C)*x)): Wrap with #if GIMPLE.
22051 Don't fold to exp if optimize_pow_to_exp is false.
22053 2018-03-28 Martin Liska <mliska@suse.cz>
22056 * calls.c (initialize_argument_information): Fix trailing space.
22057 * common.opt: Fix typo and provide better explanation for
22058 -fsanitize-coverage option.
22059 * config/i386/i386.opt: Fix typo.
22061 2018-03-28 Jakub Jelinek <jakub@redhat.com>
22062 Martin Liska <mliska@suse.cz>
22065 * gimplify.c (asan_poison_variable): Don't do the check for
22066 gimplify_omp_ctxp here.
22067 (gimplify_decl_expr): Do it here.
22068 (gimplify_target_expr): Likewise.
22070 2018-03-28 Martin Liska <mliska@suse.cz>
22073 * config/i386/i386.c (ix86_function_arg_advance): Do not call
22074 chkp_type_bounds_count if MPX is not enabled.
22076 2018-03-27 Chung-Ju Wu <jasonwucj@gmail.com>
22078 * config/nds32/nds32.h (BRANCH_COST): Adjust cost.
22080 2018-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
22083 * config/rs6000/rs6000.c (create_complex_muldiv): New helper
22084 function to create the function decl for complex long double
22085 multiply and divide for -mabi=ieeelongdouble.
22086 (init_float128_ieee): Call it.
22088 2018-03-27 H.J. Lu <hongjiu.lu@intel.com>
22091 * config/i386/i386.c (ix86_trampoline_init): Insert ENDBR for
22092 -fcf-protection=branch -mibt.
22093 * config/i386/i386.h (TRAMPOLINE_SIZE): Increased by 4 bytes.
22095 2018-03-27 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
22098 * config/arm/arm.c (arm_valid_symbolic_address): Handle
22099 arm_word_relocations.
22101 2018-03-27 Cesar Philippidis <cesar@codesourcery.com>
22104 * config/nvptx/nvptx.c (nvptx_assemble_decl_begin): Add '[]' to
22105 extern array declarations.
22107 2018-03-27 Richard Biener <rguenther@suse.de>
22109 PR middle-end/84067
22110 * match.pd ((A * C) +- (B * C) -> (A+-B) * C): Guard with
22111 explicit single_use checks.
22113 2018-03-27 Richard Biener <rguenther@suse.de>
22115 PR tree-optimization/85082
22116 * tree-ssa-sccvn.c (vn_reference_lookup_or_insert_for_pieces):
22119 2018-03-27 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
22121 * config.gcc (aarch64*-*-linux*): New TARGET_DEFAULT_ASYNC_UNWIND_TABLES
22122 * common/config/aarch64/aarch64-common.c (aarch64_optimization_table[]):
22123 Turn on fasynchronous-unwind-tables and funwind-tables.
22125 2018-03-26 Uros Bizjak <ubizjak@gmail.com>
22128 * config/i386/i386.md (*bmi_blsr_<mode>_cmp): New insn pattern.
22129 (*bmi_blsr_<mode>_ccz): Ditto.
22131 2018-03-26 Tom de Vries <tom@codesourcery.com>
22133 PR tree-optimization/85063
22134 * omp-general.c (offloading_function_p): New function. Factor out
22136 * omp-offload.c (pass_omp_target_link::gate): ... here.
22137 * omp-general.h (offloading_function_p): Declare.
22138 * tree-switch-conversion.c (build_one_array): Mark CSWTCH.x variable
22139 with attribute omp declare target for offloading functions.
22141 2018-03-24 Richard Sandiford <richard.sandiford@linaro.org>
22143 PR tree-optimization/84005
22144 * tree-data-ref.h (get_base_for_alignment): Declare.
22145 * tree-data-ref.c (get_base_for_alignment_1): New function.
22146 (get_base_for_alignment): Likewise.
22147 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
22148 get_base_for_alignment to find a suitable base object, instead
22149 of always using drb->base_address.
22151 2018-03-23 Jakub Jelinek <jakub@redhat.com>
22153 PR inline-asm/85022
22154 * emit-rtl.c (init_emit_regs): Indicate that VOIDmode MEMs don't have
22155 known size by default.
22157 2018-03-23 Vladimir Makarov <vmakarov@redhat.com>
22159 PR inline-asm/85030
22160 * lra-constraints.c (process_alt_operands): Don't match BLKmode
22161 and non BLKmode operands.
22163 2018-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22166 * config/arm/arm.md (unaligned_loadhis): Remove first alternative.
22167 Clean up attributes.
22169 2018-03-23 Richard Biener <rguenther@suse.de>
22172 * dwarf2out.c (rtl_for_decl_location): Do not generate RTL early when
22173 we are going to emit early debug for LTO.
22175 2018-03-23 Jakub Jelinek <jakub@redhat.com>
22177 PR inline-asm/85034
22178 * function.c (match_asm_constraints_1): Don't optimize if input
22179 doesn't satisfy general_operand predicate for output's mode.
22181 PR inline-asm/85022
22182 * alias.c (write_dependence_p): Don't require for x_canonicalized
22183 non-VOIDmode if x has VOIDmode.
22186 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): If DECL_REGISTER (base),
22187 just don't try to optimize it rather than assert it never happens.
22189 2018-03-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
22191 * config/rs6000/rs6000-builtin.def: Remove various BU_ALTIVEC_X
22192 macro expansions for definition of ST_INTERNAL_<mode> and
22193 LD_INTERNAL_<mode> builtins.
22194 * config/rs6000/rs6000-protos.h (rs6000_address_for_altivec):
22196 * config/rs6000/rs6000.c (altivec_expand_ld_builtin): Delete this
22198 (altivec_expand_st_builtin): Likewise.
22199 (altivec_expand_builtin): Remove calls to deleted functions.
22200 (rs6000_address_for_altivec): Delete this function.
22201 * config/rs6000/vector.md: Remove expands for
22202 vector_altivec_load_<mode> and vector_altivec_store_<mode>.
22204 2018-03-22 Sudakshina Das <sudi.das@arm.com>
22207 * config/arm/arm.h (machine_function): Add static_chain_stack_bytes.
22208 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Avoid
22209 re-computing once computed.
22210 (arm_expand_prologue): Compute machine->static_chain_stack_bytes.
22211 (arm_init_machine_status): Initialize
22212 machine->static_chain_stack_bytes.
22214 2018-03-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
22217 * doc/extend.texi: Add four new prototypes for vec_ld.
22218 * config/rs6000/rs6000-builtin.def (LVX_V1TI): Reorder symbol
22219 definitions for more logical presentation.
22220 * config/rs6000/rs6000-c.c: (altivec_overloaded_builtins): Add
22221 entries for V1TI variants of __builtin_altivec_ld builtin.
22222 * config/rs6000/rs6000.c: (altivec_expand_lv_builtin): Add test and
22223 handling of V1TI variant of LVX icode pattern.
22224 (altivec_expand_builtin): Add case for ALTIVEC_BUILTIN_LVX_V1TI.
22225 (rs6000_gimple_fold_builtin): Likewise.
22226 (altivec_init_builtins): Add code to define
22227 __builtin_altivec_lvx_v1ti function.
22229 2018-03-22 Jakub Jelinek <jakub@redhat.com>
22231 PR inline-asm/84941
22232 * function.c (match_asm_constraints_1): Don't do the optimization
22233 if input isn't a REG, SUBREG, MEM or constant.
22235 2018-03-22 Tom de Vries <tom@codesourcery.com>
22237 PR tree-optimization/84956
22238 * tree-ssa-tail-merge.c (find_clusters_1): Skip bbs with
22239 bb_has_abnormal_pred.
22241 2018-03-22 Jakub Jelinek <jakub@redhat.com>
22244 * dwarf2asm.c (dw2_output_indirect_constant_1): Set
22245 DECL_INITIAL (decl) to decl at the end.
22246 * varasm.c (use_blocks_for_decl_p): Revert the 2018-03-20 change,
22247 adjust the comment.
22249 2018-03-21 Joseph Myers <joseph@codesourcery.com>
22251 * doc/extend.texi (__builtin_tgmath): Document when complex
22252 integer types are treated as _Complex _Float64.
22254 2018-03-21 Tom de Vries <tom@codesourcery.com>
22256 * doc/extend.texi (__builtin_extend_pointer): Remove pasto.
22258 2018-03-21 Jakub Jelinek <jakub@redhat.com>
22260 PR tree-optimization/84960
22261 * tree-cfg.c (remove_bb): Don't move forced labels into bb->prev_bb
22262 if it is ENTRY block, move them into single succ of ENTRY in that case.
22264 2018-03-21 Richard Sandiford <richard.sandiford@linaro.org>
22266 PR tree-optimization/84811
22267 * poly-int.h (poly_span_traits): Remove the T3 parameter and
22268 promote HOST_WIDE_INT T2 - T1 results to unsigned HOST_WIDE_INT.
22269 (maybe_in_range_p, known_in_range_p, ranges_known_overlap_p):
22270 (known_subrange_p): Update accordingly. Cast each value involved
22271 in the size comparison, rather than casting the result of the
22274 2018-03-21 Jakub Jelinek <jakub@redhat.com>
22276 PR tree-optimization/84982
22277 * gimple-ssa-store-merging.c (invert_op): Handle boolean inversion
22278 by flipping the least significant bit rather than all bits from
22279 bitpos to bitpos + bitsize - 1.
22281 2018-03-21 Nathan Sidwell <nathan@acm.org>
22283 * doc/extend.texi (Deprecated Features): Remove mention of
22284 long-deleted deprecations.
22286 2018-03-21 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
22289 * configure.ac (gcc_cv_ld_soname) <*-*-solaris2*>: Set.
22290 * configure: Regenerate.
22292 2018-03-21 Tom de Vries <tom@codesourcery.com>
22294 PR tree-optimization/83126
22295 * tree-parloops.c (num_phis): New function.
22296 (gen_parallel_loop): Detect and handle canonicalize_loop_ivs failure.
22298 2018-03-21 Nathan Sidwell <nathan@acm.org>
22300 * doc/extend.texi (Deprecated Features): Update deprecated flags,
22301 mention anon-struct/union members and trailing attributes.
22303 2018-03-21 Bin Cheng <bin.cheng@arm.com>
22305 PR tree-optimization/84969
22306 * tree-loop-distribution.c (fuse_memset_builtins): Don't reorder
22307 builtin memset partitions if they set different rhs values.
22309 2018-03-21 Jakub Jelinek <jakub@redhat.com>
22311 PR rtl-optimization/84989
22312 * simplify-rtx.c (simplify_unary_operation_1): Don't try to simplify
22313 VEC_DUPLICATE with scalar result mode.
22315 2018-03-21 Martin Liska <mliska@suse.cz>
22318 * ipa-icf.c (sem_item_optimizer::fixup_points_to_sets): Remove
22319 not intended return statement.
22321 2018-03-21 Martin Liska <mliska@suse.cz>
22324 * tree-chkp.c (CHKP_ARRAY_MAX_CHECK_STEPS): Define a new macro.
22325 (chkp_find_bound_slots_1): Limit number of iterations.
22327 2018-03-20 David H. Gutteridge <dhgutteridge@sympatico.ca>
22330 * Minor grammar fixes for x86 options.
22332 2018-03-20 Jakub Jelinek <jakub@redhat.com>
22335 * dce.c (delete_unmarked_insns): Don't remove frame related noop moves
22336 holding REG_CFA_RESTORE notes, instead turn them into a USE.
22338 2018-03-20 Peter Bergner <bergner@vnet.ibm.com>
22341 * config/rs6000/altivec.md (altivec_lvx_<mode>_2op): Delete define_insn.
22342 (altivec_lvx_<mode>_1op): Likewise.
22343 (altivec_stvx_<mode>_2op): Likewise.
22344 (altivec_stvx_<mode>_1op): Likewise.
22345 (altivec_lvx_<VM2:mode>): New define_expand.
22346 (altivec_stvx_<VM2:mode>): Likewise.
22347 (altivec_lvx_<VM2:mode>_2op_<P:mptrsize>): New define_insn.
22348 (altivec_lvx_<VM2:mode>_1op_<P:mptrsize>): Likewise.
22349 (altivec_stvx_<VM2:mode>_2op_<P:mptrsize>): Likewise.
22350 (altivec_stvx_<VM2:mode>_1op_<P:mptrsize>): Likewise.
22351 * config/rs6000/rs6000-p8swap.c (rs6000_gen_stvx): Use new expanders.
22352 (rs6000_gen_lvx): Likewise.
22353 * config/rs6000/rs6000.c (altivec_expand_lv_builtin): Likewise.
22354 (altivec_expand_stv_builtin): Likewise.
22355 (altivec_expand_builtin): Likewise.
22356 * config/rs6000/vector.md: Likewise.
22358 2018-03-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22361 * config/arm/arm.c (arm_array_mode_supported_p): Return false for
22364 2018-03-20 Richard Biener <rguenther@suse.de>
22367 * config/i386/i386.c (ix86_add_stmt_cost): Only cost
22368 sign-conversions as zero, fall back to standard scalar_stmt
22371 2018-03-20 Martin Liska <mliska@suse.cz>
22374 * predict.c (rebuild_frequencies): Handle case when we have
22375 PROFILE_ABSENT, but flag_guess_branch_prob is false.
22377 2018-03-20 Jakub Jelinek <jakub@redhat.com>
22380 * dwarf2asm.c (dw2_output_indirect_constant_1): Temporarily turn off
22381 flag_section_anchors.
22382 * varasm.c (use_blocks_for_decl_p): Remove hack for
22383 dw2_force_const_mem.
22386 * config/aarch64/aarch64.md (*aarch64_reg_<mode>3_neg_mask2): Rename
22388 (*aarch64_<optab>_reg_<mode>3_neg_mask2): ... this. If pseudos can't
22389 be created, use lowpart_subreg of operands[0] rather than operands[0]
22391 (*aarch64_reg_<mode>3_minus_mask): Rename to ...
22392 (*aarch64_ashl_reg_<mode>3_minus_mask): ... this.
22393 (*aarch64_<optab>_reg_di3_mask2): Use const_int_operand predicate
22394 and n constraint instead of aarch64_shift_imm_di and Usd.
22395 (*aarch64_reg_<optab>_minus<mode>3): Rename to ...
22396 (*aarch64_<optab>_reg_minus<mode>3): ... this.
22398 2018-03-20 Sudakshina Das <sudi.das@arm.com>
22401 * config/arm/neon.md (ashldi3_neon): Update ?s for constraints
22402 to favor GPR over NEON registers.
22403 (<shift>di3_neon): Likewise.
22405 2018-03-20 Tom de Vries <tom@codesourcery.com>
22408 * config/nvptx/nvptx.c (nvptx_single): Don't neuter bar.sync.
22409 (nvptx_process_pars): Emit bar.sync asap and alap.
22411 2018-03-20 Tom de Vries <tom@codesourcery.com>
22414 * config/nvptx/nvptx.c (prevent_branch_around_nothing): Also update
22415 seen_label if seen_label is already set.
22417 2018-03-20 Jakub Jelinek <jakub@redhat.com>
22420 * config/i386/i386.c (fold_builtin_cpu): For features above 31
22421 use __cpu_features2 variable instead of __cpu_model.__cpu_features[0].
22422 Use 1U instead of 1. Formatting fixes.
22425 * builtins.c (fold_builtin_strpbrk): For strpbrk(x, "") use type
22426 instead of TREE_TYPE (s1) for the return value.
22428 2018-03-19 Jakub Jelinek <jakub@redhat.com>
22430 PR tree-optimization/84946
22431 * gimple-ssa-store-merging.c (mem_valid_for_store_merging): Compute
22432 bitsize + bitsize in poly_uint64 rather than poly_int64.
22435 * dwarf2asm.c: Include fold-const.c.
22436 (dw2_output_indirect_constant_1): Set DECL_INITIAL (decl) to ADDR_EXPR
22437 of decl rather than decl itself.
22439 PR rtl-optimization/84643
22440 * memmodel.h (enum memmodel): Add MEMMODEL_MAX enumerator.
22442 2018-03-19 Maxim Ostapenko <m.ostapenko@samsung.com>
22445 * dwarf2asm.c (dw2_output_indirect_constant_1): Disable ASan before
22446 calling assemble_variable.
22448 2018-03-19 Sudakshina Das <sudi.das@arm.com>
22451 * config/aarch64/aarch64-simd.md (vec_cmp<mode><v_int_equiv>): Modify
22452 instructions for UNLT, UNLE, UNGT, UNGE, UNEQ, UNORDERED and ORDERED.
22454 2018-03-19 Jim Wilson <jimw@sifive.com>
22457 * config/riscv/riscv.c (riscv_function_arg_boundary): Use
22458 PREFERRED_STACK_BOUNDARY instead of STACK_BOUNDARY.
22459 (riscv_first_stack_step): Likewise.
22460 (riscv_option_override): Use STACK_BOUNDARY instead of
22461 MIN_STACK_BOUNDARY.
22462 * config/riscv/riscv.h (STACK_BOUNDARY): Renamed from
22463 MIN_STACK_BOUNDARY.
22464 (BIGGEST_ALIGNMENT): Set to 128.
22465 (PREFERRED_STACK_BOUNDARY): Renamed from STACK_BOUNDARY.
22466 (RISCV_STACK_ALIGN): Use PREFERRED_STACK_BOUNDARY instead of
22469 2018-03-19 Richard Biener <rguenther@suse.de>
22471 PR tree-optimization/84933
22472 * tree-vrp.c (set_and_canonicalize_value_range): Treat out-of-bound
22473 values as -INF/INF when canonicalizing an ANTI_RANGE to a RANGE.
22475 2018-03-19 Richard Biener <rguenther@suse.de>
22477 PR tree-optimization/84859
22478 * tree-ssa-phiopt.c (single_trailing_store_in_bb): New function.
22479 (cond_if_else_store_replacement): Perform sinking operation on
22480 single-store BBs regardless of MAX_STORES_TO_SINK setting.
22481 Generalize what a BB with a single eligible store is.
22483 2018-03-19 Richard Biener <rguenther@suse.de>
22485 PR tree-optimization/84929
22486 * tree-data-ref.c (analyze_siv_subscript_cst_affine): Guard
22487 chrec_is_positive against non-chrec arg.
22489 2018-03-19 Tamar Christina <tamar.christina@arm.com>
22492 * config/arm/arm.c (arm_can_change_mode_class): revert r258554.
22494 2018-03-18 Martin Liska <mliska@suse.cz>
22496 PR rtl-optimization/84635
22497 * regrename.c (build_def_use): Use matches_mode only when
22500 2018-03-18 Richard Sandiford <richard.sandiford@linaro.org>
22502 PR tree-optimization/84913
22503 * tree-vect-loop.c (vectorizable_reduction): Don't try to
22504 vectorize chains of COND_EXPRs.
22506 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
22508 * config/nds32/nds32.h (MAX_REGS_PER_ADDRESS): Fix the value.
22510 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
22512 * config/nds32/nds32.h (LOGICAL_OP_NON_SHORT_CIRCUIT): Define.
22514 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
22516 * config/nds32/nds32.h (CLZ_DEFINED_VALUE_AT_ZERO): Define.
22518 2018-03-17 Chung-Ju Wu <jasonwucj@gmail.com>
22519 Kito Cheng <kito.cheng@gmail.com>
22521 * config/nds32/nds32-protos.h (nds32_adjust_reg_alloc_order): Declare.
22522 * config/nds32/nds32.c (nds32_reg_alloc_order_for_speed): New array.
22523 (nds32_adjust_reg_alloc_order): New function.
22524 * config/nds32/nds32.h (ADJUST_REG_ALLOC_ORDER): Define.
22526 2018-03-17 Kito Cheng <kito.cheng@gmail.com>
22528 * config/nds32/nds32.c (nds32_asm_output_mi_thunk,
22529 nds32_print_operand, nds32_print_operand_address): Use
22530 HOST_WIDE_INT_PRINT_DEC instead.
22532 2018-03-17 Chung-Ju Wu <jasonwucj@gmail.com>
22534 * config/nds32/nds32.c (nds32_register_priority): Modify cost.
22536 2018-03-17 Jakub Jelinek <jakub@redhat.com>
22539 * config/i386/i386.c (initial_ix86_tune_features,
22540 initial_ix86_arch_features): Use unsigned HOST_WIDE_INT rather than
22541 unsigned long long.
22542 (set_ix86_tune_features): Change ix86_tune_mask from unsigned int
22543 to unsigned HOST_WIDE_INT, initialize to HOST_WIDE_INT_1U << ix86_tune
22544 rather than 1u << ix86_tune. Formatting fix.
22545 (ix86_option_override_internal): Change ix86_arch_mask from
22546 unsigned int to unsigned HOST_WIDE_INT, initialize to
22547 HOST_WIDE_INT_1U << ix86_arch rather than 1u << ix86_arch.
22548 (ix86_function_specific_restore): Likewise.
22550 2018-03-16 Jakub Jelinek <jakub@redhat.com>
22553 * postreload.c (reload_combine_recognize_pattern): Perform
22554 INTVAL addition in unsigned HOST_WIDE_INT type to avoid UB and
22555 truncate_int_for_mode the result for the destination's mode.
22558 * hsa-gen.c (mem_type_for_type): Fix comment typo.
22559 * tree-vect-loop-manip.c (vect_create_cond_for_niters_checks):
22561 * gimple-ssa-warn-restrict.c (builtin_memref::set_base_and_offset):
22564 2018-03-16 Vladimir Makarov <vmakarov@redhat.com>
22567 * lra-assigns.c (lra_split_hard_reg_for): Don't use
22568 regno_allocno_class_array and sorted_pseudos.
22569 * lra-constraints.c (spill_hard_reg_in_range): Ignore hard regs in
22570 insns where regno is used.
22572 2018-03-16 Martin Liska <mliska@suse.cz>
22575 * multiple_target.c (create_dispatcher_calls): Redirect
22576 reference in the symbol table.
22578 2018-03-16 Martin Liska <mliska@suse.cz>
22581 * multiple_target.c (create_dispatcher_calls): Redirect also
22584 2018-03-16 Jakub Jelinek <jakub@redhat.com>
22588 * tree.h (TARGET_EXPR_NO_ELIDE): Define.
22589 * gimplify.c (gimplify_modify_expr_rhs): Don't elide TARGET_EXPRs with
22590 TARGET_EXPR_NO_ELIDE flag set unless *expr_p is INIT_EXPR.
22592 2018-03-16 Julia Koval <julia.koval@intel.com>
22594 * doc/invoke.texi (Skylake Server): Add CLWB.
22595 Cannonlake): Remove CLWB.
22597 2018-03-16 Jakub Jelinek <jakub@redhat.com>
22599 PR tree-optimization/84841
22600 * tree-ssa-reassoc.c (INTEGER_CONST_TYPE): Change to 1 << 4 from
22602 (FLOAT_ONE_CONST_TYPE): Define.
22603 (constant_type): Return FLOAT_ONE_CONST_TYPE for -1.0 and 1.0.
22604 (sort_by_operand_rank): Put entries with higher constant_type last
22605 rather than first to match comments.
22607 2018-03-15 Sandra Loosemore <sandra@codesourcery.com>
22609 * config/nios2/nios2.md (movsi_internal): Fix thinko in
22612 2018-03-15 Jakub Jelinek <jakub@redhat.com>
22615 * calls.c (expand_call): For TREE_ADDRESSABLE rettype ignore alignment
22616 check and use address of target always.
22618 2018-03-15 H.J. Lu <hongjiu.lu@intel.com>
22621 * config/i386/i386.c (indirect_thunk_needed): Update comments.
22622 (indirect_thunk_bnd_needed): Likewise.
22623 (indirect_thunks_used): Likewise.
22624 (indirect_thunks_bnd_used): Likewise.
22625 (indirect_return_needed): New.
22626 (indirect_return_bnd_needed): Likewise.
22627 (output_indirect_thunk_function): Add a bool argument for
22629 (output_indirect_thunk_function): Don't generate alias for
22630 function return thunk.
22631 (ix86_code_end): Call output_indirect_thunk_function to generate
22632 function return thunks.
22633 (ix86_output_function_return): Set indirect_return_bnd_needed
22634 and indirect_return_needed instead of indirect_thunk_bnd_needed
22635 and indirect_thunk_needed.
22637 2018-03-15 Olga Makhotina <olga.makhotina@intel.com>
22639 * config/i386/sgxintrin.h (_enclv_u32): New intrinsic.
22640 (__enclv_bc, __enclv_cd, __enclv_generic): New definitions.
22641 (ERDINFO, ETRACKC, ELDBC, ELDUC): New leaves.
22643 2018-03-15 David Malcolm <dmalcolm@redhat.com>
22644 Paul Hua <paul.hua.gm@gmail.com>
22647 * gcc.dg/fixits-pr84852-1.c: Fix filename in dg-regexp.
22649 2018-03-15 Segher Boessenkool <segher@kernel.crashing.org>
22651 * config/rs6000/rs6000.c (abi_v4_pass_in_fpr): Add back the
22652 TARGET_DOUBLE_FLOAT and TARGET_SINGLE_FLOAT conditions on the DFmode
22653 resp. SFmode cases.
22655 2018-03-15 Tamar Christina <tamar.christina@arm.com>
22658 * config/arm/arm.c (arm_can_change_mode_class): Use GET_MODE_UNIT_SIZE
22659 instead of GET_MODE_SIZE when comparing Units.
22661 2018-03-15 Vladimir Mezentsev <vladimir.mezentsev@oracle.com>
22664 * varasm.c (hash_section): Return an unchangeble hash value
22665 * config/aarch64/aarch64.c (aarch64_use_blocks_for_constant_p):
22666 Return !aarch64_can_use_per_function_literal_pools_p ().
22668 2018-03-15 Jakub Jelinek <jakub@redhat.com>
22671 * optabs.c (emit_conditional_move): Pass address of cmode's copy
22672 rather than address of cmode as last argument to prepare_cmp_insn.
22674 2018-03-15 Julia Koval <julia.koval@intel.com>
22676 * config/i386/i386.c (F_AVX512VBMI2, F_GFNI, F_VPCLMULQDQ,
22677 F_AVX512VNNI, F_AVX512BITALG): New.
22679 2018-03-14 John David Anglin <danglin@gcc.gnu.org>
22682 * config/pa/pa.c (pa_emit_move_sequence): Always emit secondary reload
22683 insn for floating-point loads and stores.
22685 2018-03-14 Carl Love <cel@us.ibm.com>
22687 * config/rs6000/rs6000-c.c: Add macro definitions for
22688 ALTIVEC_BUILTIN_VEC_PERMXOR.
22689 * config/rs6000/rs6000.h: Add #define for vec_permxor builtin.
22690 * config/rs6000/rs6000-builtin.def: Add macro expansions for VPERMXOR.
22691 * config/rs6000/altivec.md (altivec_vpermxor): New define expand.
22692 * config/rs6000/rs6000-p8swap.c (rtx_is_swappable_p): Add case
22694 * config/doc/extend.texi: Add prototypes for vec_permxor.
22696 2018-03-14 David Malcolm <dmalcolm@redhat.com>
22699 * diagnostic-show-locus.c (class layout_point): Convert m_line
22700 from int to linenum_type.
22701 (line_span::comparator): Use linenum "compare" function when
22702 comparing line numbers.
22703 (test_line_span): New function.
22704 (layout_range::contains_point): Convert param "row" from int to
22706 (layout_range::intersects_line_p): Likewise.
22707 (layout::will_show_line_p): Likewise.
22708 (layout::print_source_line): Likewise.
22709 (layout::should_print_annotation_line_p): Likewise.
22710 (layout::print_annotation_line): Likewise.
22711 (layout::print_leading_fixits): Likewise.
22712 (layout::annotation_line_showed_range_p): Likewise.
22713 (struct line_corrections): Likewise for field m_row.
22714 (line_corrections::line_corrections): Likewise for param "row".
22715 (layout::print_trailing_fixits): Likewise.
22716 (layout::get_state_at_point): Likewise.
22717 (layout::get_x_bound_for_row): Likewise.
22718 (layout::print_line): Likewise.
22719 (diagnostic_show_locus): Likewise for locals "last_line" and "row".
22720 (selftest::diagnostic_show_locus_c_tests): Call test_line_span.
22721 * input.c (selftest::test_linenum_comparisons): New function.
22722 (selftest::input_c_tests): Call it.
22723 * selftest.c (selftest::test_assertions): Test ASSERT_GT,
22724 ASSERT_GT_AT, ASSERT_LT, and ASSERT_LT_AT.
22725 * selftest.h (ASSERT_GT): New macro.
22726 (ASSERT_GT_AT): New macro.
22727 (ASSERT_LT): New macro.
22728 (ASSERT_LT_AT): New macro.
22730 2018-03-14 Segher Boessenkool <segher@kernel.crashing.org>
22732 PR rtl-optimization/84780
22733 * combine.c (distribute_links): Don't make a link based on pc_rtx.
22735 2018-03-14 Martin Liska <mliska@suse.cz>
22737 * tree.c (record_node_allocation_statistics): Use
22738 get_stats_node_kind.
22739 (get_stats_node_kind): New function extracted from
22740 record_node_allocation_statistics.
22741 (free_node): Use get_stats_node_kind.
22743 2018-03-14 Richard Biener <rguenther@suse.de>
22745 * tree-ssa-pre.c (compute_antic_aux): Remove code that asserts
22746 that the value-set of ANTIC_IN doesn't grow.
22749 * tree-ssa-pre.c (struct bb_bitmap_sets): Add visited_with_visited_succs
22751 (BB_VISITED_WITH_VISITED_SUCCS): New define.
22752 (compute_antic): Initialize BB_VISITED_WITH_VISITED_SUCCS.
22754 2018-03-14 Julia Koval <julia.koval@intel.com>
22756 * config.gcc (icelake-client, icelake-server): New.
22758 * config/i386/i386.c (initial_ix86_tune_features): Extend to 64 bit.
22759 (initial_ix86_arch_features): Ditto.
22760 (PTA_SKYLAKE): Add SGX.
22761 (PTA_ICELAKE): Remove.
22762 (PTA_ICELAKE_CLIENT): New.
22763 (PTA_ICELAKE_SERVER): New.
22764 (ix86_option_override_internal): Split up icelake on icelake client and
22766 (get_builtin_code_for_version): Ditto.
22767 (fold_builtin_cpu): Ditto.
22768 * config/i386/driver-i386.c (config/i386/driver-i386.c): Ditto.
22769 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto
22770 * config/i386/i386.h (processor_type): Ditto.
22771 * doc/invoke.texi: Ditto.
22773 2018-03-14 Jakub Jelinek <jakub@redhat.com>
22776 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): Handle also
22777 INTEGER_CST offset, add it together with bitpos / 8 and
22778 sign extend based on POINTER_SIZE.
22782 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
22785 * config/i386/constraints.md (Yc): New register constraint.
22786 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
22787 Use Yc constraint for alternative 2 of operand 0. Remove
22788 preferred_for_speed attribute.
22790 2018-03-14 Richard Biener <rguenther@suse.de>
22792 PR tree-optimization/84830
22793 * tree-ssa-pre.c (compute_antic_aux): Intersect the new ANTIC_IN
22794 with the old one to avoid oscillations.
22796 2018-03-13 Vladimir Makarov <vmakarov@redhat.com>
22799 * lra-assigns.c (find_all_spills_for): Ignore uninteresting
22801 (assign_by_spills): Return a flag of reload assignment failure.
22802 Do not process the reload assignment failures. Do not spill other
22803 reload pseudos if they has the same reg class. Update n if
22805 (lra_assign): Add a return arg. Set up from the result of
22806 assign_by_spills call.
22807 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
22808 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
22809 usage_insns if it is not NULL.
22810 (spill_hard_reg_in_range): New function.
22811 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
22812 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
22813 function prototypes.
22814 (lra_assign): Change prototype.
22815 * lra.c (lra): Add code to deal with fails by splitting hard reg
22818 2018-03-01 Palmer Dabbelt <palmer@sifive.com>
22820 * config/riscv/riscv.opt (mrelax): New option.
22821 * config/riscv/riscv.c (riscv_file_start): Emit ".option
22822 "norelax" when riscv_mrelax is disabled.
22823 * doc/invoke.texi (RISC-V): Document "-mrelax" and "-mno-relax".
22825 2018-03-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22828 * config/rs6000/rs6000.c (rs6000_reassociation_width): Disable parallel
22829 reassociation for int modes.
22831 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
22833 * tree-vect-loop-manip.c (vect_maybe_permute_loop_masks):
22834 Reverse the choice between VEC_UNPACK_LO_EXPR and VEC_UNPACK_HI_EXPR
22836 * config/aarch64/iterators.md (hi_lanes_optab): New int attribute.
22837 * config/aarch64/aarch64-sve.md
22838 (*aarch64_sve_<perm_insn><perm_hilo><mode>): Rename to...
22839 (aarch64_sve_<perm_insn><perm_hilo><mode>): ...this.
22840 (*extend<mode><Vwide>2): Rename to...
22841 (aarch64_sve_extend<mode><Vwide>2): ...this.
22842 (vec_unpack<su>_<perm_hilo>_<mode>): Turn into a define_expand,
22843 renaming the old pattern to...
22844 (aarch64_sve_punpk<perm_hilo>_<mode>): ...this. Only define
22846 (vec_unpack<su>_<perm_hilo>_<SVE_BHSI:mode>): Turn into a
22847 define_expand, renaming the old pattern to...
22848 (aarch64_sve_<su>unpk<perm_hilo>_<SVE_BHSI:mode>): ...this.
22849 (*vec_unpacku_<perm_hilo>_<mode>_no_convert): Delete.
22850 (vec_unpacks_<perm_hilo>_<mode>): Take BYTES_BIG_ENDIAN into
22851 account when deciding which SVE instruction the optab should use.
22852 (vec_unpack<su_optab>_float_<perm_hilo>_vnx4si): Likewise.
22854 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
22856 * config/aarch64/aarch64.md (V4_REGNUM, V8_REGNUM, V12_REGNUM)
22857 (V20_REGNUM, V24_REGNUM, V28_REGNUM, P1_REGNUM, P2_REGNUM, P3_REGNUM)
22858 (P4_REGNUM, P5_REGNUM, P6_REGNUM, P8_REGNUM, P9_REGNUM, P10_REGNUM)
22859 (P11_REGNUM, P12_REGNUM, P13_REGNUM, P14_REGNUM): New define_constants.
22860 (tlsdesc_small_<mode>): Turn a define_expand and use
22861 tlsdesc_small_sve_<mode> for SVE. Rename original define_insn to...
22862 (tlsdesc_small_advsimd_<mode>): ...this.
22863 (tlsdesc_small_sve_<mode>): New pattern.
22865 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
22867 * config/aarch64/iterators.md (UNSPEC_SMUL_HIGHPART)
22868 (UNSPEC_UMUL_HIGHPART): New constants.
22869 (MUL_HIGHPART): New int iteraor.
22870 (su): Handle UNSPEC_SMUL_HIGHPART and UNSPEC_UMUL_HIGHPART.
22871 * config/aarch64/aarch64-sve.md (<su>mul<mode>3_highpart): New
22873 (*<su>mul<mode>3_highpart): New define_insn.
22875 2018-03-13 Eric Botcazou <ebotcazou@adacore.com>
22878 * ipa-devirt.c (odr_subtypes_equivalent_p): Do not get the ODR type of
22881 2018-03-13 Martin Liska <mliska@suse.cz>
22884 * (sem_item_optimizer::sem_item_optimizer): Initialize new
22886 (sem_item_optimizer::~sem_item_optimizer): Release it.
22887 (sem_item_optimizer::merge_classes): Register variable aliases.
22888 (sem_item_optimizer::fixup_pt_set): New function.
22889 (sem_item_optimizer::fixup_points_to_sets): Likewise.
22890 * ipa-icf.h: Declare new variables and functions.
22892 2018-03-13 Jakub Jelinek <jakub@redhat.com>
22894 PR middle-end/84834
22895 * match.pd ((A & C) != 0 ? D : 0): Use INTEGER_CST@2 instead of
22896 integer_pow2p@2 and test integer_pow2p in condition.
22897 (A < 0 ? C : 0): Similarly for @1.
22899 PR middle-end/84831
22900 * stmt.c (parse_output_constraint): If the CONSTRAINT_LEN (*p, p)
22901 characters starting at p contain '\0' character, don't look beyond
22905 * config/i386/i386.md (round<mode>2): For 387 fancy math, disable
22906 pattern if -ftrapping-math -fno-fp-int-builtin-inexact.
22909 * reg-stack.c (change_stack): Change update_end var from int to
22910 rtx_insn *, if non-NULL don't update just BB_END (current_block), but
22911 also call set_block_for_insn on the newly added insns and rescan.
22914 * config/i386/sse.md (sse2_loadhpd): Use Yv constraint rather than v
22915 on the last operand.
22918 * tree.c (stabilize_reference_1): Return save_expr (e) for
22919 STATEMENT_LIST even if it doesn't have side-effects.
22921 2018-03-12 Jonathan Wakely <jwakely@redhat.com>
22923 * doc/invoke.texi (-mclflushopt): Fix spelling of option.
22925 2018-03-12 Renlin Li <renlin.li@arm.com>
22927 * config/aarch64/aarch64.md (movhf_aarch64): Fix mode argument to
22928 aarch64_output_scalar_simd_mov_immediate.
22930 2018-03-12 Martin Sebor <msebor@redhat.com>
22932 PR tree-optimization/83456
22933 * gimple-fold.c (gimple_fold_builtin_memory_op): Avoid warning
22934 for perfectly overlapping calls to memcpy.
22935 (gimple_fold_builtin_memory_chk): Same.
22936 (gimple_fold_builtin_strcpy): Handle no-warning.
22937 (gimple_fold_builtin_stxcpy_chk): Same.
22938 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Handle no-warning.
22940 2018-03-12 Segher Boessenkool <segher@kernel.crashing.org>
22942 * config/rs6000/rs6000.c (abi_v4_pass_in_fpr): Add bool "named"
22943 parameter. Use it for SFmode.
22944 (rs6000_function_arg_advance_1): Adjust.
22945 (rs6000_function_arg): Adjust.
22946 (rs6000_gimplify_va_arg): Pass false for that new parameter.
22948 2018-03-12 Segher Boessenkool <segher@kernel.crashing.org>
22950 PR rtl-optimization/84169
22951 PR rtl-optimization/84780
22952 * combine.c (can_combine_p): Check for a 2-insn combination whether
22953 the destination register is used between the two insns, too.
22955 2018-03-12 Richard Biener <rguenther@suse.de>
22957 PR tree-optimization/84803
22958 * tree-if-conv.c (ifcvt_memrefs_wont_trap): Don't do anything
22959 for refs DR analysis didn't process.
22961 2018-03-12 Richard Biener <rguenther@suse.de>
22963 PR tree-optimization/84777
22964 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): For
22965 force-vectorize loops ignore whether we are optimizing for size.
22967 2018-03-12 Chung-Ju Wu <jasonwucj@gmail.com>
22969 * config/nds32/nds32.c (nds32_md_asm_adjust): New function.
22970 (TARGET_MD_ASM_ADJUST): Define.
22972 2018-03-12 Monk Chiang <sh.chiang04@gmail.com>
22973 Kito Cheng <kito.cheng@gmail.com>
22974 Chung-Ju Wu <jasonwucj@gmail.com>
22976 * config/nds32/nds32.c (nds32_compute_stack_frame,
22977 nds32_emit_stack_push_multiple, nds32_emit_stack_pop_multiple,
22978 nds32_emit_stack_v3push, nds32_emit_stack_v3pop,
22979 nds32_emit_adjust_frame, nds32_expand_prologue, nds32_expand_epilogue,
22980 nds32_expand_prologue_v3push, nds32_expand_epilogue_v3pop): Refine.
22981 * config/nds32/nds32.h (NDS32_FIRST_CALLEE_SAVE_GPR_REGNUM,
22982 NDS32_LAST_CALLEE_SAVE_GPR_REGNUM, NDS32_V3PUSH_AVAILABLE_P): New.
22983 * config/nds32/nds32.md (prologue, epilogue): Use macro
22984 NDS32_V3PUSH_AVAILABLE_P to do checking.
22986 2018-03-11 Jakub Jelinek <jakub@redhat.com>
22989 * dwarf2out.c (gen_enumeration_type_die): Don't guard adding
22990 DW_AT_declaration for ENUM_IS_OPAQUE on -gdwarf-4 or -gno-strict-dwarf,
22991 but on TYPE_SIZE. Don't do anything for ENUM_IS_OPAQUE if not creating
22992 a new die. Don't set TREE_ASM_WRITTEN if ENUM_IS_OPAQUE. Guard
22993 addition of most attributes on !orig_type_die or the attribute not
22994 being present already. Assert TYPE_VALUES is NULL for ENUM_IS_OPAQUE.
22996 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
22997 Chung-Ju Wu <jasonwucj@gmail.com>
22999 * config/nds32/nds32.c (nds32_cpu_cpp_builtins): Modify to define
23000 __NDS32_VH__ macro.
23001 * config/nds32/nds32.opt (mvh): New option.
23003 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
23004 Chung-Ju Wu <jasonwucj@gmail.com>
23006 * config/nds32/nds32-protos.h (nds32_cpu_cpp_builtins): Declare
23008 * config/nds32/nds32.c (nds32_cpu_cpp_builtins): New function.
23009 * config/nds32/nds32.h (TARGET_CPU_CPP_BUILTINS): Modify its
23012 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
23013 Chung-Ju Wu <jasonwucj@gmail.com>
23015 * config/nds32/nds32-memory-manipulation.c (nds32_expand_strlen): New
23017 * config/nds32/nds32-multiple.md (strlensi): New pattern.
23018 * config/nds32/nds32-protos.h (nds32_expand_strlen): Declare function.
23020 2018-03-11 Monk Chiang <sh.chiang04@gmail.com>
23021 Kito Cheng <kito.cheng@gmail.com>
23022 Chung-Ju Wu <jasonwucj@gmail.com>
23024 * config/nds32/constants.md (unspec_element): Add UNSPEC_FFB,
23025 UNSPEC_FFMISM and UNSPEC_FLMISM.
23026 * config/nds32/nds32-intrinsic.c (bdesc_2arg): Add builtin description
23027 for ffb, ffmism and flmism.
23028 * config/nds32/nds32-intrinsic.md (unspec_ffb): Define new pattern.
23029 (unspec_ffmism): Ditto.
23030 (unspec_flmism): Ditto.
23031 (nds32_expand_builtin_impl): Check if string extension is available.
23032 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_FFB,
23033 NDS32_BUILTIN_FFMISM and NDS32_BUILTIN_FLMISM.
23035 2018-03-10 Vladimir Makarov <vmakarov@redhat.com>
23038 2018-03-09 Vladimir Makarov <vmakarov@redhat.com>
23041 * lra-assigns.c (assign_by_spills): Return a flag of reload
23042 assignment failure. Do not process the reload assignment
23043 failures. Do not spill other reload pseudos if they has the same
23045 (lra_assign): Add a return arg. Set up from the result of
23046 assign_by_spills call.
23047 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
23048 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
23049 usage_insns if it is not NULL.
23050 (spill_hard_reg_in_range): New function.
23051 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
23052 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
23053 function prototypes.
23054 (lra_assign): Change prototype.
23055 * lra.c (lra): Add code to deal with fails by splitting hard reg
23058 2018-03-10 H.J. Lu <hongjiu.lu@intel.com>
23061 * config/i386/i386.opt: Replace Enforcment with Enforcement.
23063 2018-03-10 Alexandre Oliva <aoliva@redhat.com>
23066 * dwarf2out.h (dw_val_class): Add dw_val_class_symview.
23067 (dw_val_node): Add val_symbolic_view.
23068 * dwarf2out.c (dw_line_info_table): Add symviews_since_reset.
23069 (symview_upper_bound): New.
23070 (new_line_info_table): Initialize symviews_since_reset.
23071 (dwarf2out_source_line): Count symviews_since_reset and set
23072 symview_upper_bound.
23073 (dw_val_equal_p): Handle symview.
23074 (add_AT_symview): New.
23075 (print_dw_val): Handle symview.
23076 (attr_checksum, attr_checksum_ordered): Likewise.
23077 (same_dw_val_p, size_of_die): Likewise.
23078 (value_format, output_die): Likewise.
23079 (add_high_low_attributes): Use add_AT_symview for entry_view.
23080 (dwarf2out_finish): Reset symview_upper_bound, clear
23083 2018-03-09 Peter Bergner <bergner@vnet.ibm.com>
23086 * config/rs6000/rs6000.c (rs6000_offsettable_memref_p): New prototype.
23087 Add strict argument and use it.
23088 (rs6000_split_multireg_move): Update for new strict argument.
23089 (mem_operand_gpr): Disallow all non-offsettable addresses.
23090 * config/rs6000/rs6000.md (*movdi_internal64): Use YZ constraint.
23092 2018-03-09 Jakub Jelinek <jakub@redhat.com>
23095 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Mark va_arg_tmp
23096 temporary TREE_ADDRESSABLE before gimplification of BUILT_IN_MEMCPY.
23097 * config/powerpcspe/powerpcspe.c (rs6000_gimplify_va_arg): Likewise.
23100 * tree-inline.c (copy_tree_body_r): For INDIRECT_REF of a remapped
23101 decl, use remap_type if we want to use the type.
23103 2018-03-09 Martin Sebor <msebor@redhat.com>
23105 PR tree-optimization/84526
23106 * gimple-ssa-warn-restrict.c (builtin_memref::set_base_and_offset):
23108 (builtin_access::generic_overlap): Be prepared to handle non-array
23111 2018-03-09 Alexandre Oliva <aoliva@redhat.com>
23113 PR rtl-optimization/84682
23114 * lra-constraints.c (process_address_1): Check is_address flag
23115 for address constraints.
23116 (process_alt_operands): Likewise.
23117 * lra.c (lra_set_insn_recog_data): Pass asm operand locs to
23118 preprocess_constraints.
23119 * recog.h (preprocess_constraints): Add oploc parameter.
23121 * recog.c (preprocess_constraints): Test address_operand for
23122 CT_ADDRESS constraints.
23124 2018-03-09 Vladimir Makarov <vmakarov@redhat.com>
23127 * lra-assigns.c (assign_by_spills): Return a flag of reload
23128 assignment failure. Do not process the reload assignment
23129 failures. Do not spill other reload pseudos if they has the same
23131 (lra_assign): Add a return arg. Set up from the result of
23132 assign_by_spills call.
23133 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
23134 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
23135 usage_insns if it is not NULL.
23136 (spill_hard_reg_in_range): New function.
23137 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
23138 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
23139 function prototypes.
23140 (lra_assign): Change prototype.
23141 * lra.c (lra): Add code to deal with fails by splitting hard reg
23144 2018-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23147 * common/config/arm/arm-common.c (arm_parse_arch_option_name):
23148 Accept complain bool parameter. Only emit errors if it is true.
23149 (arm_parse_cpu_option_name): Likewise.
23150 (arm_target_thumb_only): Adjust callers of the above.
23151 * config/arm/arm-protos.h (arm_parse_cpu_option_name): Adjust
23152 prototype to take a default true bool parameter.
23153 (arm_parse_arch_option_name): Likewise.
23155 2018-03-09 David Malcolm <dmalcolm@redhat.com>
23156 Francois-Xavier Coudert <fxcoudert@gcc.gnu.org>
23160 * Makefile.in (LD_VERSION_SCRIPT_OPTION, LD_SONAME_OPTION): New.
23161 * configure: Regenerate.
23162 * configure.ac ("linker --version-script option"): New.
23163 ("linker soname option"): New.
23165 2018-03-09 Richard Biener <rguenther@suse.de>
23167 PR tree-optimization/84775
23168 * tree-if-conv.c (add_bb_predicate_gimplified_stmts): Delink
23169 immediate uses of predicate stmts and mark them modified.
23172 PR tree-optimization/84178
23173 * tree-if-conv.c (combine_blocks): Move insert_gimplified_predicates
23175 (version_loop_for_if_conversion): Delay update_ssa call.
23176 (tree_if_conversion): Delay update_ssa until after predicate
23179 2018-03-09 Eric Botcazou <ebotcazou@adacore.com>
23182 * config/i386/winnt.c (i386_pe_seh_cold_init): Use small pre-allocation
23183 when the function accesses prior frames.
23185 2018-03-08 Jakub Jelinek <jakub@redhat.com>
23188 * dwarf2out.c (dw_loc_list): If list && loc_list->first->next, call
23189 gen_llsym, otherwise call maybe_gen_llsym.
23191 PR inline-asm/84742
23192 * recog.c (asm_operand_ok): Return 0 if multi-character constraint
23193 has ',' character inside of it.
23195 2018-03-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23198 * config/aarch64/aarch64.md (*compare_cstore<mode>_insn): Mark pattern
23199 as clobbering CC_REGNUM.
23201 2018-03-08 Richard Biener <rguenther@suse.de>
23203 PR middle-end/84552
23204 * tree-scalar-evolution.c: Include tree-into-ssa.h.
23205 (follow_copies_to_constant): Do not follow SSA names registered
23208 2018-03-08 Richard Biener <rguenther@suse.de>
23210 PR tree-optimization/84178
23211 * tree-if-conv.c (combine_blocks): Move insert_gimplified_predicates
23213 (version_loop_for_if_conversion): Delay update_ssa call.
23214 (tree_if_conversion): Delay update_ssa until after predicate
23217 2018-03-08 David Malcolm <dmalcolm@redhat.com>
23219 PR tree-optimization/84178
23220 * tree-if-conv.c (release_bb_predicate): Remove the
23221 the assertion that the stmts have NULL use_ops.
23222 Discard the statements, asserting that they haven't
23223 yet been added to a BB.
23225 2018-03-08 Richard Biener <rguenther@suse.de>
23227 PR tree-optimization/84746
23228 * tree-ssa-pre.c (find_leader_in_sets): Deal with SET1 being NULL.
23229 (phi_translate): Pass in destination ANTIC_OUT set.
23230 (phi_translate_1): Likewise. For a simplified result lookup
23231 a leader in ANTIC_OUT and AVAIL_OUT, not the ANTIC_IN sets.
23232 (phi_translate_set): Adjust.
23233 (do_pre_regular_insertion): Likewise.
23234 (do_pre_partial_partial_insertion): Likewise.
23236 2018-03-08 Martin Liska <mliska@suse.cz>
23238 PR gcov-profile/84735
23239 * doc/gcov.texi: Document usage of profile files.
23240 * gcov-io.h: Document changes in the format.
23242 2018-03-08 Alexandre Oliva <aoliva@redhat.com>
23246 * dwarf2out.c (struct dw_line_info_table): Update comments for
23248 (FORCE_RESET_NEXT_VIEW): New.
23249 (FORCE_RESETTING_VIEW_P): New.
23250 (RESETTING_VIEW_P): Check for -1 too.
23251 (ZERO_VIEW_P): Likewise.
23252 (new_line_info_table): Force-reset next view.
23253 (dwarf2out_begin_function): Likewise.
23254 (dwarf2out_source_line): Simplify zero_view_p initialization.
23255 Test FORCE_RESETTING_VIEW_P and RESETTING_VIEW_P instead of
23256 view directly. Omit view when omitting .loc at line 0.
23258 2018-03-08 Jakub Jelinek <jakub@redhat.com>
23260 PR tree-optimization/84740
23261 * tree-switch-conversion.c (process_switch): Call build_constructors
23262 only if info.phi_count is non-zero.
23264 PR tree-optimization/84739
23265 * tree-tailcall.c (find_tail_calls): Check call arguments against
23266 DECL_ARGUMENTS (current_function_decl) rather than
23267 DECL_ARGUMENTS (func) when checking for tail recursion.
23269 2018-03-07 Jakub Jelinek <jakub@redhat.com>
23271 * doc/contrib.texi: Add entries for Martin Liska, David Malcolm,
23272 Marek Polacek, extend Vladimir Makarov's, Jonathan Wakely's and
23273 Volker Reichelt's entry and add entries for people that perform
23274 GCC fuzzy testing and report numerous bugs.
23276 2018-03-07 Segher Boessenkool <segher@kernel.crashing.org>
23279 * config/rs6000/rs6000.c (rs6000_elf_in_small_data_p): Don't put
23280 readonly data in sdata, if that is disabled.
23281 * config/rs6000/sysv4.opt (mreadonly-in-sdata): New option.
23282 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
23283 -mreadonly-in-sdata option.
23285 2018-03-07 Martin Sebor <msebor@redhat.com>
23287 PR tree-optimization/84468
23288 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Consider successor
23289 basic block when looking for nul assignment.
23291 2018-03-07 Eric Botcazou <ebotcazou@adacore.com>
23294 * except.h (output_function_exception_table): Adjust prototype.
23295 * except.c (output_function_exception_table): Remove FNNAME parameter
23296 and add SECTION parameter. Ouput one part of the table at a time.
23297 * final.c (final_scan_insn_1) <NOTE_INSN_SWITCH_TEXT_SECTIONS>: Output
23298 the first part of the exception table and emit unwind directives.
23299 * config/i386/i386-protos.h (i386_pe_end_cold_function): Declare.
23300 (i386_pe_seh_cold_init): Likewise.
23301 * config/i386/cygming.h (ASM_DECLARE_COLD_FUNCTION_NAME): New macro.
23302 (ASM_DECLARE_COLD_FUNCTION_SIZE): Likewise.
23303 * config/i386/i386.c (x86_expand_epilogue): Fix wording in comment.
23304 (ix86_output_call_insn): Emit a nop in one more case for SEH.
23305 * config/i386/winnt.c: Include except.h.
23306 (struct seh_frame_state): Add reg_offset, after_prologue and
23307 in_cold_section fields.
23308 (i386_pe_seh_end_prologue): Set seh->after_prologue.
23309 (i386_pe_seh_cold_init): New function.
23310 (i386_pe_seh_fini): Add COLD parameter and bail out if it is not equal
23311 to seh->in_cold_section.
23312 (seh_emit_push): Record the offset of the push.
23313 (seh_emit_save): Record the offet of the save.
23314 (i386_pe_seh_unwind_emit): Deal with NOTE_INSN_SWITCH_TEXT_SECTIONS.
23315 Test seh->after_prologue to disregard the epilogue.
23316 (i386_pe_end_function): Pass FALSE to i386_pe_seh_fini.
23317 (i386_pe_end_cold_function): New function.
23319 2018-03-07 Jakub Jelinek <jakub@redhat.com>
23322 * config/aarch64/predicates.md (aarch64_simd_reg_or_zero): Use
23323 aarch64_simd_or_scalar_imm_zero rather than aarch64_simd_imm_zero.
23326 * gimple-expr.c (create_tmp_var_raw): Set DECL_NAMELESS flag
23328 * tree-pretty-print.c (dump_decl_name): For TDF_COMPARE_DEBUG,
23329 don't print names of DECL_NAMELESS DECL_IGNORED_P decls.
23331 PR middle-end/84723
23332 * multiple_target.c: Include tree-inline.h and intl.h.
23333 (expand_target_clones): Diagnose and fail if node->definition and
23334 !tree_versionable_function_p (node->decl).
23336 2018-03-06 John David Anglin <danglin@gcc.gnu.org>
23338 * config/pa/pa.h (ASM_GENERATE_INTERNAL_LABEL): Revise to use
23340 (ASM_OUTPUT_ADDR_VEC_ELT): Revise for above change.
23341 (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
23342 * config/pa/pa64-hpux.h (ASM_GENERATE_INTERNAL_LABEL): Revise as above.
23344 2018-03-06 Jakub Jelinek <jakub@redhat.com>
23347 * combine.c (try_combine): Use reg_or_subregno instead of handling
23348 just paradoxical SUBREGs and REGs.
23350 2018-03-06 Claudiu Zissulescu <claziss@synopsys.com>
23352 * config/arc/arc.c (arc_finalize_pic): Remove function.
23353 (arc_must_save_register): We use single base PIC register, remove
23354 checks to save/restore the PIC register.
23355 (arc_expand_prologue): Likewise.
23356 * config/arc/arc-protos.h (arc_set_default_type_attributes):
23358 (arc_verify_short): Likewise.
23359 (arc_attr_type): Likewise.
23360 * config/arc/arc.c (arc_set_default_type_attributes): Remove.
23361 (walk_stores): Likewise.
23362 (arc_address_cost): Make it static.
23363 (arc_verify_short): Likewise.
23364 (branch_dest): Likewise.
23365 (arc_attr_type): Likewise.
23366 * config/arc/arc.c (TARGET_ADJUST_INSN_LENGTH): Remove.
23367 (TARGET_INSN_LENGTH_PARAMETERS): Likewise.
23368 (arc_final_prescan_insn): Remove inserting the nops due to
23369 hardware hazards. It is done in reorg step.
23370 (insn_length_variant_t): Remove.
23371 (insn_length_parameters_t): Likewise.
23372 (arc_insn_length_parameters): Likewise.
23373 (arc_get_insn_variants): Likewise.
23374 * config/arc/arc.h (TARGET_UPSIZE_DBR): Remove.
23376 2018-03-06 Jakub Jelinek <jakub@redhat.com>
23378 PR inline-asm/84683
23379 * reg-stack.c (move_for_stack_reg): If any_malformed_asm, avoid
23382 PR tree-optimization/84687
23383 * omp-simd-clone.c (simd_clone_create): Clear DECL_BUILT_IN_CLASS
23385 * match.pd (pow(C,x)*expN(y) -> expN(logN(C)*x+y)): New optimization.
23387 2018-03-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23389 * config/rs6000/rs6000-builtin.def (rs6000_speculation_barrier):
23390 Rename to ppc_speculation_barrier.
23391 * config/rs6000/rs6000.c (rs6000_init_builtins): Rename builtin to
23392 __builtin_ppc_speculation_barrier.
23394 2018-03-05 Jakub Jelinek <jakub@redhat.com>
23397 * combine.c (combine_simplify_rtx): Don't try to simplify if
23398 if_then_else_cond returned non-NULL, but either true_rtx or false_rtx
23401 2018-03-05 Segher Boessenkool <segher@kernel.crashing.org>
23403 * config/rs6000/rs6000.c (rs6000_loop_align): Don't align tiny loops
23404 to 32 bytes when compiling for POWER9.
23406 2018-03-05 Jakub Jelinek <jakub@redhat.com>
23409 * config/i386/i386.c (ix86_function_ok_for_sibcall): Check for
23410 regparm >= 3 with no arg reg available also for calls with
23411 flag_force_indirect_call. Pass decl to ix86_function_regparm.
23414 * config/i386/sse.md (*<code><mode>3): Replace <mask_prefix3> with
23416 (*<plusminus_insn><mode>3): Likewise. Remove <mask_operand3> uses.
23418 2018-03-05 Peter Bergner <bergner@vnet.ibm.com>
23421 * config/rs6000/vector.md (mov<mode>): Disallow altivec memory operands.
23423 2018-03-05 Richard Biener <rguenther@suse.de>
23425 PR tree-optimization/84486
23426 * tree-ssa-pre.c (create_expression_by_pieces): Remove dead code.
23427 When inserting a __builtin_assume_aligned call set the LHS
23428 SSA name alignment info accordingly.
23430 2018-03-05 Wilco Dijkstra <wdijkstr@arm.com>
23432 PR tree-optimization/84114
23433 * config/aarch64/aarch64.c (aarch64_reassociation_width)
23434 Avoid reassociation of FLOAT_MODE addition.
23436 2018-03-05 Olga Makhotina <olga.makhotina@intel.com>
23438 * common/config/i386/i386-common.c (OPTION_MASK_ISA_PCONFIG_SET,
23439 OPTION_MASK_ISA_PCONFIG_UNSET, OPTION_MASK_ISA_WBNOINVD_SET,
23440 OPTION_MASK_ISA_WBNOINVD_UNSET): New definitions.
23441 (ix86_handle_option): Handle -mpconfig and -mwbnoinvd.
23442 * config.gcc (pconfigintrin.h, wbnoinvdintrin.h) : Add headers.
23443 * config/i386/cpuid.h (bit_PCONFIG, bit_WBNOINVD): New.
23444 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mpconfig
23446 * config/i386/i386-builtin.def (__builtin_ia32_wbnoinvd,
23447 __builtin_ia32_wbinvd): New builtins.
23448 (SPECIAL_ARGS2): New.
23449 * config/i386/i386-c.c (__WBNOINVD__, __PCONFIG__): New.
23450 (SPECIAL_ARGS2): New.
23451 * config/i386/i386.c (ix86_target_string): Add -mpconfig and -mwbnoinvd.
23452 (ix86_valid_target_attribute_inner_p): Ditto.
23453 (ix86_init_mmx_sse_builtins): Add special_args2.
23454 * config/i386/i386.h (TARGET_PCONFIG, TARGET_PCONFIG_P, TARGET_WBNOINVD,
23455 TARGET_WBNOINVD_P): New.
23456 * config/i386/i386.md (UNSPECV_WBINVD, UNSPECV_WBNOINVD): New.
23457 (define_insn "wbinvd", define_insn "wbnoinvd"): New.
23458 * config/i386/i386.opt: Add -mpconfig and -mwbnoinvd.
23459 * config/i386/immintrin.h (_wbinvd): New intrinsic.
23460 * config/i386/pconfigintrin.h: New file.
23461 * config/i386/wbnoinvdintrin.h: Ditto.
23462 * config/i386/x86intrin.h: Add headers pconfigintrin.h and
23464 * doc/invoke.texi (-mpconfig, -mwbnoinvd): New.
23466 2018-03-05 Richard Biener <rguenther@suse.de>
23468 PR tree-optimization/84670
23469 * tree-ssa-pre.c (struct bb_bitmap_sets): Add visited_with_visited_succs
23471 (BB_VISITED_WITH_VISITED_SUCCS): New define.
23472 (compute_antic): Initialize BB_VISITED_WITH_VISITED_SUCCS.
23473 (compute_antic_aux): Only assert the number of values in ANTIC_IN
23474 doesn't grow if all successors (recursively) were visited at least
23477 2018-03-05 Richard Biener <rguenther@suse.de>
23479 PR tree-optimization/84650
23480 * tree-ssa-loop-im.c (pass_lim::execute): Reset the SCEV cache
23481 if executed in the loop pipeline.
23483 2018-03-05 Sandra Loosemore <sandra@codesourcery.com>
23485 * doc/configfiles.texi (Configuration Files): Move info about
23486 conditionalizing $target-protos.h to...
23487 * doc/sourcebuild.texi (Back End): Here. Explain how $target.h
23488 differs from $target-protos.h.
23490 2018-03-05 Kito Cheng <kito.cheng@gmail.com>
23491 Chung-Ju Wu <jasonwucj@gmail.com>
23493 * config/nds32/nds32-protos.h (nds32_expand_setmem): Declare.
23494 * config/nds32/nds32-multiple.md (setmemsi): Define.
23495 * config/nds32/nds32-memory-manipulation.c
23496 (nds32_gen_dup_4_byte_to_word_value): New.
23497 (emit_setmem_word_loop): New.
23498 (emit_setmem_byte_loop): New.
23499 (nds32_expand_setmem_loop): New.
23500 (nds32_expand_setmem_loop_v3m): New.
23501 (nds32_expand_setmem_unroll): New.
23502 (nds32_expand_setmem): New.
23504 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
23505 Chung-Ju Wu <jasonwucj@gmail.com>
23507 * config/nds32/nds32-memory-manipulation.c
23508 (nds32_emit_load_store): New.
23509 (nds32_emit_post_inc_load_store): New.
23510 (nds32_emit_mem_move): New.
23511 (nds32_emit_mem_move_block): New.
23512 (nds32_expand_movmemsi_loop_unknown_size): New.
23513 (nds32_expand_movmemsi_loop_known_size): New.
23514 (nds32_expand_movmemsi_loop): New.
23515 (nds32_expand_movmemsi_unroll): New.
23516 (nds32_expand_movmemqi): Rename ...
23517 (nds32_expand_movmemsi): ... to this.
23518 * config/nds32/nds32-multiple.md (movmemqi): Rename ...
23519 (movmemsi): ... to this.
23520 * config/nds32/nds32-protos.h (nds32_expand_movmemqi): Rename ...
23521 (nds32_expand_movmemsi): ... to this.
23523 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
23524 Monk Chiang <sh.chiang04@gmail.com>
23525 Chung-Ju Wu <jasonwucj@gmail.com>
23527 * config/nds32/nds32-protos.h
23528 (nds32_expand_load_multiple): New arguments.
23529 (nds32_expand_store_multiple): Ditto.
23530 (nds32_valid_multiple_load_store): Rename ...
23531 (nds32_valid_multiple_load_store_p): ... to this.
23532 * config/nds32/nds32-memory-manipulation.c
23533 (nds32_expand_load_multiple): Refine implementation.
23534 (nds32_expand_store_multiple): Ditto.
23535 * config/nds32/nds32-multiple.md
23536 (load_multiple): Update nds32_expand_load_multiple interface.
23537 (store_multiple): Update nds32_expand_store_multiple interface.
23538 * config/nds32/nds32-predicates.c
23539 (nds32_valid_multiple_load_store): Rename ...
23540 (nds32_valid_multiple_load_store_p): ... to this and refine
23542 * config/nds32/predicates.md
23543 (nds32_load_multiple_and_update_address_operation): New predicate.
23544 (nds32_store_multiple_and_update_address_operation): New predicate.
23546 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
23547 Chung-Ju Wu <jasonwucj@gmail.com>
23549 * config/nds32/nds32.md (type): Add load_multiple and store_multiple.
23550 (combo): New attribute.
23551 * config/nds32/nds32-multiple.md: Refine patterns with new attributes.
23553 2018-03-03 Chung-Ju Wu <jasonwucj@gmail.com>
23555 * config/nds32/nds32.opt: Change -mcmodel= default value.
23557 2018-03-03 Kito Cheng <kito.cheng@gmail.com>
23558 Monk Chiang <sh.chiang04@gmail.com>
23559 Chung-Ju Wu <jasonwucj@gmail.com>
23561 * config/nds32/constants.md (unspec_element): New enum.
23562 * config/nds32/constraints.md (Umw): New constraint.
23563 * config/nds32/nds32-intrinsic.c: Add more builtin functions.
23564 * config/nds32/nds32-intrinsic.md: Likewise.
23565 * config/nds32/nds32-md-auxiliary.c (nds32_regno_to_enable4): New.
23566 (nds32_valid_smw_lwm_base_p): New.
23567 (nds32_output_smw_single_word): New.
23568 (nds32_output_lmw_single_word): New.
23569 (nds32_expand_unaligned_load): New.
23570 (nds32_expand_unaligned_store): New.
23571 * config/nds32/nds32-protos.h (nds32_valid_smw_lwm_base_p): Declare.
23572 (nds32_output_smw_single_word): Declare.
23573 (nds32_output_lmw_single_word): Declare.
23574 (nds32_expand_unaligned_load): Declare.
23575 (nds32_expand_unaligned_store): Declare.
23576 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_UALOAD_HW,
23577 NDS32_BUILTIN_UALOAD_W, NDS32_BUILTIN_UALOAD_DW,
23578 NDS32_BUILTIN_UASTORE_HW, NDS32_BUILTIN_UASTORE_W,
23579 NDS32_BUILTIN_UASTORE_DW.
23580 * config/nds32/predicates.md (nds32_lmw_smw_base_operand): New
23583 2018-03-03 Monk Chiang <sh.chiang04@gmail.com>
23584 Kito Cheng <kito.cheng@gmail.com>
23585 Chung-Ju Wu <jasonwucj@gmail.com>
23587 * config/nds32/nds32-intrinsic.c
23588 (nds32_expand_builtin_null_ftype_reg): Delete.
23589 (nds32_expand_builtin_reg_ftype_imm): Ditto.
23590 (nds32_expand_builtin_null_ftype_reg_imm): Ditto.
23591 (nds32_read_argument): New.
23592 (nds32_legitimize_target): Ditto.
23593 (nds32_legitimize_argument): Ditto.
23594 (nds32_check_constant_argument): Ditto.
23595 (nds32_expand_unop_builtin): Ditto.
23596 (nds32_expand_unopimm_builtin): Ditto.
23597 (nds32_expand_binop_builtin): Ditto.
23598 (nds32_builtin_decl_impl): Ditto.
23599 (builtin_description): Ditto.
23600 (nds32_expand_builtin_impl): Rewrite with new infrastructure.
23601 (nds32_init_builtins_impl): Ditto.
23602 * config/nds32/nds32.c (TARGET_BUILTIN_DECL): Define.
23603 (nds32_builtin_decl): New.
23604 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_COUNT.
23605 * config/nds32/nds32-protos.h (nds32_builtin_decl_impl): Declare.
23607 2018-03-02 Jeff Law <law@redhat.com>
23609 * reorg.c (stop_search_p): Handle DEBUG_INSN.
23610 (redundant_insn, fill_simple_delay_slots): Likewise.
23611 (fill_slots_from_thread): Likewise.
23612 * resource.c (mark_referenced_resources): Likewise.
23613 (mark_set_resources, find_dead_or_set_registers): Likewise.
23615 2018-03-02 Jakub Jelinek <jakub@redhat.com>
23617 * substring-locations.h (format_warning_va): Formatting fix for
23618 ATTRIBUTE_GCC_DIAG.
23619 (format_warning_at_substring): Fix up ATTRIBUTE_GCC_DIAG second
23621 (format_warning_n_va, format_warning_at_substring_n): New prototypes.
23622 * substring-locations.c: Include intl.h.
23623 (format_warning_va): Turned into small wrapper around
23624 format_warning_n_va, renamed to ...
23625 (format_warning_n_va): ... this, add N and PLURAL_GMSGID arguments,
23626 rename GMSGID to SINGULAR_GMSGID, if SINGULAR_GMSGID != PLURAL_GMSGID,
23628 (format_warning_at_substring_n): New function.
23629 * gimple-ssa-sprintf.c: Remove GCC diagnostic ignored pragma.
23630 (fmtwarn): Add ATTRIBUTE_GCC_DIAG. Turn into a copy of
23631 format_warning_at_substring with just a shorter name instead of
23632 const function pointer.
23633 (fmtwarn_n): New function.
23634 (maybe_warn, format_directive, parse_directive): Use fmtwarn_n where
23635 appropriate, get rid of all the fmtstr temporaries, move conditionals
23636 with G_() wrapped string literals directly into fmtwarn arguments,
23637 cast dir.len to (int), formatting fixes.
23639 2018-03-02 Thomas Schwinge <thomas@codesourcery.com>
23641 * doc/invoke.texi: Remove "Cilk Plus" references.
23643 2018-03-02 Jakub Jelinek <jakub@redhat.com>
23644 Richard Biener <rguenther@suse.de>
23647 * expr.c (expand_expr_real_1) <case CALL_EXPR>: Don't emit diagnostics
23648 for error or warning attributes if CALL_FROM_THUNK_P is set.
23651 2018-03-02 Jakub Jelinek <jakub@redhat.com>
23654 * config/pa/pa.h (TARGET_CPU_CPP_BUILTINS): Predefine
23655 __SIZEOF_128__ macro if HPUX_LONG_DOUBLE_LIBRARY.
23658 * config/ia64/ia64.h (TARGET_CPU_CPP_BUILTINS): Predefine
23659 __SIZEOF_{FPREG,FLOAT{80,128}}__ macros.
23661 * predict.c (test_prediction_value_range): Use PROB_UNINITIALIZED
23662 instead of -1U in last predictors element's probability member.
23664 2018-03-02 Eric Botcazou <ebotcazou@adacore.com>
23667 * ipa-devirt.c (odr_subtypes_equivalent_p): Get the ODR type of both
23668 arguments if they are comparable.
23670 2018-03-02 Richard Sandiford <richard.sandiford@linaro.org>
23672 PR tree-optimization/84634
23673 * tree-vect-stmts.c (vectorizable_store, vectorizable_load): Replace
23674 masks and masked_loop_p with a single loop_masks, making sure it's
23675 null for bb vectorization.
23677 2018-03-02 Richard Sandiford <richard.sandiford@linaro.org>
23679 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence)
23680 (vect_analyze_data_ref_access): Use loop->safe_len rather than
23681 loop->force_vectorize to check whether there is no alias.
23683 2018-03-02 Jakub Jelinek <jakub@redhat.com>
23686 * rtl.h (prev_real_nondebug_insn, next_real_nondebug_insn): New
23688 * emit-rtl.c (next_real_insn, prev_real_insn): Fix up function
23690 (next_real_nondebug_insn, prev_real_nondebug_insn): New functions.
23691 * cfgcleanup.c (try_head_merge_bb): Use prev_real_nondebug_insn
23692 instead of a loop around prev_real_insn.
23693 * combine.c (move_deaths): Use prev_real_nondebug_insn instead of
23696 PR inline-asm/84625
23697 * config/i386/i386.c (ix86_print_operand): Use conditional
23698 output_operand_lossage instead of gcc_assert if CONST_VECTOR is not
23701 2018-03-02 Richard Biener <rguenther@suse.de>
23703 PR tree-optimization/84427
23704 * tree-ssa-pre.c (bitmap_remove_expr_from_set): Remove.
23705 (bitmap_set_subtract_values): Rewrite to handle multiple
23708 (prune_clobbered_mems): Likewise.
23709 (phi_translate): Take edge instead of pred/phiblock.
23710 (phi_translate_1): Likewise.
23711 (phi_translate_set): Likewise. Insert all translated
23712 exprs for a value into the set, keeping possibly multiple
23713 expressions per value.
23714 (compute_antic_aux): Adjust for phi_translate changes.
23715 When intersecting union the expressions and prune those
23716 not in the final value set, keeping possibly multiple
23717 expressions per value. Do not use value-insertion
23718 for unioning ANTIC_OUT U EXP_GEN - TMP_GEN but merge
23719 all expressions. Add verification that the value-sets
23720 only shrink during iteration.
23721 (compute_partial_antic_aux): Adjust for the phi_translate changes.
23722 (do_pre_regular_insertion): Likewise.
23723 (do_pre_partial_partial_insertion): Likewise.
23725 2018-03-02 Richard Biener <rguenther@suse.de>
23728 * config/darwin.c (saved_debug_info_level): New static global.
23729 (darwin_asm_lto_start): Disable debug info generation for LTO out.
23730 (darwin_asm_lto_end): Restore debug info generation settings.
23732 2018-03-01 Martin Liska <mliska@suse.cz>
23735 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
23736 volatile arguments.
23738 2018-03-01 Richard Biener <rguenther@suse.de>
23741 * dwarf2out.c (gen_variable_die): Properly handle late VLA
23742 type annotation with LTO when debug was disabled at compile-time.
23744 2018-03-01 Matthew Fortune <mfortune@gmail.com>
23746 * config/mips/mips.c (mips_final_prescan_insn): Fix incorrect
23748 (mips_final_postscan_insn): Likewise.
23750 2018-03-01 Richard Sandiford <richard.sandiford@linaro.org>
23752 PR rtl-optimization/84528
23753 * alias.c (init_alias_target): Add commentary.
23754 (init_alias_analysis): Only give HARD_FRAME_POINTER_REGNUM
23755 a unique base value if the frame pointer is not eliminated
23756 to the stack pointer.
23758 2018-03-01 Tom de Vries <tom@codesourcery.com>
23760 PR rtl-optimization/83327
23761 * lra-int.h (hard_regs_spilled_into): Declare.
23762 * lra.c (hard_regs_spilled_into): Define.
23763 (init_reg_info): Init hard_regs_spilled_into.
23764 * lra-spills.c (assign_spill_hard_regs): Update hard_regs_spilled_into.
23765 * lra-lives.c (make_hard_regno_born, make_hard_regno_dead)
23766 (process_bb_lives): Handle hard_regs_spilled_into.
23767 (lra_create_live_ranges_1): Before doing liveness propagation, clear
23768 regs in all_hard_regs_bitmap if set in hard_regs_spilled_into.
23770 2018-02-28 David Edelsohn <dje.gcc@gmail.com>
23772 * config.gcc (powerpc-ibm-aix7.1.*): New stanza.
23773 (powerpc-ibm-aix[789]*): Default to AIX 7.2.
23774 * config/rs6000/aix71.h (TARGET_DEFAULT): Revert to Power4 ISA.
23775 * config/rs6000/aix72.h: New file.
23777 2018-02-28 Jakub Jelinek <jakub@redhat.com>
23779 * gimple-ssa-warn-restrict.c (maybe_diag_overlap): Use warning_n
23780 instead of warning_at with conditional singular and plural messages
23784 * stor-layout.c (update_alignment_for_field): For
23785 targetm.ms_bitfield_layout_p (rli->t), if !is_bitfield
23786 && !DECL_PACKED (field), do the alignment update, just use
23787 only desired_align instead of MAX (type_align, desired_align)
23789 (place_field): Don't do known_align < desired_align handling
23790 early if targetm.ms_bitfield_layout_p (rli->t) and rli->prev_field
23791 is non-NULL, instead do it after rli->prev_field handling and
23792 only if not within a bitfield word. For DECL_PACKED (field)
23793 use type_align of BITS_PER_UNIT.
23795 2018-02-28 Eric Botcazou <ebotcazou@adacore.com>
23797 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Remove
23798 superfluous parentheses and trailing spaces.
23800 2018-02-28 Richard Biener <rguenther@suse.de>
23802 PR tree-optimization/84584
23803 * graphite-scop-detection.c (scop_detection::add_scop): Discard
23804 SCoPs with fake exit edge.
23806 2018-02-28 Martin Liska <mliska@suse.cz>
23809 * timevar.c (timer::print): Fix format to properly print 100%
23812 2018-02-28 Richard Biener <rguenther@suse.de>
23814 PR middle-end/84607
23815 * genmatch.c (capture_info::walk_match): Do not mark
23816 captured expressions without operands as expr_p given
23817 they act more like predicates and should be subject to
23818 "lost tail" side-effect preserving.
23820 2018-02-28 Alexandre Oliva <aoliva@redhat.com>
23822 PR rtl-optimization/81611
23823 * auto-inc-dec.c (attempt_change): Move dead note from
23824 mem_insn if it's the next use of regno
23825 (find_address): Take address use of reg holding
23826 non-incremented value. Add parm to limit search to the named
23828 (merge_in_block): Attempt to use a mem insn that is the next
23829 use of the original regno.
23831 2018-02-27 Martin Sebor <msebor@redhat.com>
23834 * doc/invoke.texi (-Wmissing-attributes): New option.
23835 * print-tree.c (print_node): Handle DECL_UNINLINABLE.
23837 2018-02-27 Martin Sebor <msebor@redhat.com>
23839 PR translation/84207
23840 * diagnostic-core.h (warning_n, error_n, inform_n): Change
23841 n argument to unsigned HOST_WIDE_INT.
23842 * diagnostic.c (warning_n, error_n, inform_n): Ditto.
23843 (diagnostic_n_impl): Ditto. Handle arguments in excess of LONG_MAX.
23844 * gimple-ssa-sprintf.c (format_directive): Simplify inform_n call.
23845 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Use warning_n.
23847 2018-02-27 Richard Biener <rguenther@suse.de>
23849 PR tree-optimization/84512
23850 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost):
23851 Do not use the estimate returned from record_stmt_cost for
23852 the scalar iteration cost but sum properly using add_stmt_cost.
23854 2018-02-27 Richard Biener <rguenther@suse.de>
23856 PR tree-optimization/84466
23857 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
23858 Adjust last change to less strictly validate use operands.
23860 2018-02-27 Martin Liska <mliska@suse.cz>
23862 PR gcov-profile/84548
23863 * gcov.c (process_file): Allow partial overlap and consider it
23864 also as group functions.
23865 (output_lines): Properly calculate range of lines for a group.
23867 2018-02-27 Martin Liska <mliska@suse.cz>
23869 * timevar.c (timer::print_row): Remove 'usr', 'sys', 'wall' and
23870 'ggc' suffixes. Change first column width.
23871 (timer::print): Fix formatting of the column.
23873 2018-02-27 Alexandre Oliva <aoliva@redhat.com>
23875 * tree-ssa-live.c (remove_unused_scope_block_p): Do not
23876 preserve inline entry blocks for the sake of debug inline
23877 entry point markers alone.
23878 (remove_unused_locals): Suggest in comments a better place to
23879 force the preservation of inline entry blocks that are
23880 otherwise unused, but do not preserve them.
23882 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
23884 * config/i386/i386.c (ix86_output_indirect_jmp): Update comments.
23886 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
23889 * config/i386/constraints.md (Bs): Replace
23890 ix86_indirect_branch_register with
23891 TARGET_INDIRECT_BRANCH_REGISTER.
23893 * config/i386/i386.md (indirect_jump): Likewise.
23894 (tablejump): Likewise.
23895 (*sibcall_memory): Likewise.
23896 (*sibcall_value_memory): Likewise.
23897 Peepholes of indirect call and jump via memory: Likewise.
23898 (*sibcall_GOT_32): Disallowed for TARGET_INDIRECT_BRANCH_REGISTER.
23899 (*sibcall_value_GOT_32): Likewise.
23900 * config/i386/predicates.md (indirect_branch_operand): Likewise.
23901 (GOT_memory_operand): Likewise.
23902 (call_insn_operand): Likewise.
23903 (sibcall_insn_operand): Likewise.
23904 (GOT32_symbol_operand): Likewise.
23905 * config/i386/i386.h (TARGET_INDIRECT_BRANCH_REGISTER): New.
23907 2018-02-26 Eric Botcazou <ebotcazou@adacore.com>
23909 PR rtl-optimization/83496
23910 * reorg.c (steal_delay_list_from_target): Change REDUNDANT array from
23911 booleans to RTXes. Call fix_reg_dead_note on every non-null element.
23912 (steal_delay_list_from_fallthrough): Call fix_reg_dead_note on a
23913 redundant insn, if any.
23914 (relax_delay_slots): Likewise.
23915 (update_reg_unused_notes): Rename REDUNDANT_INSN to OTHER_INSN.
23917 2018-02-26 Richard Sandiford <richard.sandiford@linaro.org>
23919 PR tree-optimization/83965
23920 * tree-vect-patterns.c (vect_reassociating_reduction_p): Assume
23921 that grouped statements are part of a reduction chain. Return
23922 true if the statement is not marked as a reduction itself but
23923 is part of a group.
23924 (vect_recog_dot_prod_pattern): Don't check whether the statement
23925 is part of a group here.
23926 (vect_recog_sad_pattern): Likewise.
23927 (vect_recog_widen_sum_pattern): Likewise.
23929 2018-02-26 Eric Botcazou <ebotcazou@adacore.com>
23932 * final.c (rest_of_clean_state): Also look for calls inside sequences.
23934 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
23937 * config/i386/i386-protos.h (ix86_output_indirect_jmp): Remove
23939 (ix86_output_indirect_function_return): New prototype.
23940 (ix86_split_simple_return_pop_internal): Likewise.
23941 * config/i386/i386.c (indirect_return_via_cx): New.
23942 (indirect_return_via_cx_bnd): Likewise.
23943 (indirect_thunk_name): Handle return va CX_REG.
23944 (output_indirect_thunk_function): Create alias for
23945 __x86_return_thunk_[re]cx and __x86_return_thunk_[re]cx_bnd.
23946 (ix86_output_indirect_jmp): Remove the bool argument.
23947 (ix86_output_indirect_function_return): New function.
23948 (ix86_split_simple_return_pop_internal): Likewise.
23949 * config/i386/i386.md (*indirect_jump): Don't pass false
23950 to ix86_output_indirect_jmp.
23951 (*tablejump_1): Likewise.
23952 (simple_return_pop_internal): Change it to define_insn_and_split.
23953 Call ix86_split_simple_return_pop_internal to split it for
23954 -mfunction-return=.
23955 (simple_return_indirect_internal): Call
23956 ix86_output_indirect_function_return instead of
23957 ix86_output_indirect_jmp.
23959 2018-02-26 Jakub Jelinek <jakub@redhat.com>
23962 * vec.h (vec_default_construct): For BROKEN_VALUE_INITIALIZATION use
23963 memset and value initialization afterwards.
23965 2018-02-26 Christophe Lyon <christophe.lyon@linaro.org>
23967 * Makefile.in (lto-wrapper): Use ALL_LINKERFLAGS.
23969 2018-02-26 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
23972 * common/config/aarch64/aarch64-common.c
23973 (aarch_option_optimization_table[]): Switch
23974 off fomit-frame-pointer
23976 2018-02-26 Kito Cheng <kito.cheng@gmail.com>
23977 Chung-Ju Wu <jasonwucj@gmail.com>
23979 * config/nds32/nds32-multiple.md (load_multiple): Disallow
23981 (store_multiple): Ditto.
23983 2018-02-26 Kito Cheng <kito.cheng@gmail.com>
23985 * config.gcc: Add --with-cpu support for nds32 target.
23986 * config/nds32/nds32-opts.h (nds32_cpu_type): New.
23987 * config/nds32/nds32.opt: Add -mcpu= option.
23989 2018-02-25 Segher Boessenkool <segher@kernel.crashing.org>
23991 * config/rs6000/rs6000.opt (mvrsave=no, mvrsave=yes, isel=no,
23992 isel=yes): Warn for these deprecated options.
23994 2018-02-23 David Edelsohn <dje.gcc@gmail.com>
23996 * config/rs6000/aix71.h (TARGET_DEFAULT): Change to
23997 ISA_2_5_MASKS_EMBEDDED.
23999 2018-02-23 Jakub Jelinek <jakub@redhat.com>
24001 * ipa-prop.c (ipa_vr_ggc_hash_traits::hash): Hash p->min and
24002 p->max as pointers rather than using iterative_hash_expr.
24004 2018-02-23 Carl Love <cel@us.ibm.com>
24006 * config/rs6000/rs6000-builtin.def: Change VSIGNED2 and VUNSIGNED2
24007 macro expansions from BU_VSX_2 to BU_P8V_VSX_2 and BU_VSX_OVERLOAD_2 to
24009 * config/rs6000/rs6000-c.c: Change VSX_BUILTIN_VEC_VSIGNED2 to
24010 P8V_BUILTIN_VEC_VSIGNED2. Change VSX_BUILTIN_VEC_VUNSIGNED2 to
24011 P8V_BUILTIN_VEC_VUNSIGNED2.
24013 2018-02-22 Vladimir Makarov <vmakarov@redhat.com>
24016 * lra-int.h (LRA_UNKNOWN_ALT, LRA_NON_CLOBBERED_ALT): New macros.
24017 * lra.c (lra_set_insn_recog_data, lra_update_insn_recog_data): Use
24019 * lra-constraints.c (curr_insn_transform): Set up
24020 LRA_NON_CLOBBERED_ALT for moves processed on the fast path. Use
24022 (remove_inheritance_pseudos): Use LRA_UNKNOWN_ALT.
24023 * lra-eliminations.c (spill_pseudos): Ditto.
24024 (process_insn_for_elimination): Ditto.
24025 * lra-lives.c (reg_early_clobber_p): Use the new macros.
24026 * lra-spills.c (spill_pseudos): Use LRA_UNKNOWN_ALT and
24027 LRA_NON_CLOBBERED_ALT.
24029 2018-02-22 Martin Sebor <msebor@redhat.com>
24031 PR tree-optimization/84480
24032 * gimple-fold.c (gimple_fold_builtin_strcpy): Move warnings
24033 to maybe_diag_stxncpy_trunc. Call it.
24034 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Integrate warnings
24035 from gimple_fold_builtin_strcpy. Print inlining stack.
24036 (handle_builtin_stxncpy): Print inlining stack.
24037 * tree-ssa-strlen.h (maybe_diag_stxncpy_trunc): Declare.
24039 2018-02-22 H.J. Lu <hongjiu.lu@intel.com>
24042 * config/i386/i386.c (ix86_set_indirect_branch_type): Issue an
24043 error when -mindirect-branch=thunk-extern, -fcf-protection=branch
24044 and -fcheck-pointer-bounds are used together.
24045 (indirect_thunk_prefix): New enum.
24046 (indirect_thunk_need_prefix): New function.
24047 (indirect_thunk_name): Replace need_bnd_p with need_prefix. Use
24048 "_nt" instead of "_bnd" for NOTRACK prefix.
24049 (output_indirect_thunk): Replace need_bnd_p with need_prefix.
24050 (output_indirect_thunk_function): Likewise.
24052 (ix86_code_end): Update output_indirect_thunk_function calls.
24053 (ix86_output_indirect_branch_via_reg): Replace
24054 ix86_bnd_prefixed_insn_p with indirect_thunk_need_prefix.
24055 (ix86_output_indirect_branch_via_push): Likewise.
24056 (ix86_output_function_return): Likewise.
24057 * doc/invoke.texi: Document -mindirect-branch=thunk-extern is
24058 incompatible with -fcf-protection=branch and
24059 -fcheck-pointer-bounds.
24061 2018-02-22 Steve Ellcey <sellcey@cavium.com>
24064 * config/aarch64/aarch64.c (aarch64_print_address_internal):
24065 Change gcc_assert call to output_operand_lossage.
24067 2018-02-22 Steve Ellcey <sellcey@cavium.com>
24069 * doc/extend.texi (__builtin_extend_pointer): Document builtin.
24071 2018-02-22 DJ Delorie <dj@redhat.com>
24072 Sebastian Perta <sebastian.perta@renesas.com>
24073 Oleg Endo <olegendo@gcc.gnu.org>
24075 * config/rx/rx.c (rx_rtx_costs): New function.
24076 (TARGET_RTX_COSTS): Override to use rx_rtx_costs.
24078 2018-02-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
24080 * config/arm/t-multilib: Map Armv8-R to Armv7 multilibs.
24082 2018-02-22 Martin Liska <mliska@suse.cz>
24085 * common/config/arm/arm-common.c (arm_print_hint_for_cpu_option):
24086 Add "native" as a possible value.
24088 2018-02-22 Martin Liska <mliska@suse.cz>
24091 * config/i386/i386.c (ix86_option_override_internal):
24092 Add "native" as a possible value for -march and -mtune.
24094 2018-02-22 Jakub Jelinek <jakub@redhat.com>
24097 * stor-layout.c (finalize_type_size): Propagate TYPE_EMPTY_P flag
24098 to all type variants.
24100 PR tree-optimization/84503
24101 * gimple-ssa-store-merging.c (merged_store_group::merge_into): Compute
24102 width as info->bitpos + info->bitsize - start.
24103 (merged_store_group::merge_overlapping): Simplify width computation.
24104 (check_no_overlap): New function.
24105 (imm_store_chain_info::try_coalesce_bswap): Compute expected
24106 start + width and last_order of the group, fail if check_no_overlap
24108 (imm_store_chain_info::coalesce_immediate_stores): Don't merge info
24109 to group if check_no_overlap fails.
24111 2018-02-21 Segher Boessenkool <segher@kernel.crashing.org>
24113 * config/rs6000/altivec.md: Delete contraint arguments to
24114 define_expand, define_split, and define_peephole2, and in
24115 define_insn_and_split if always unused.
24116 * config/rs6000/darwin.md: Ditto.
24117 * config/rs6000/dfp.md: Ditto.
24118 * config/rs6000/rs6000.md: Ditto.
24119 * config/rs6000/sync.md: Ditto.
24120 * config/rs6000/vector.md: Ditto.
24121 * config/rs6000/vsx.md: Ditto.
24123 2018-02-21 Segher Boessenkool <segher@kernel.crashing.org>
24125 * config/rs6000/altivec.md: Write output control strings as braced
24126 blocks instead of double-quoted strings.
24127 * config/rs6000/darwin.md: Ditto.
24128 * config/rs6000/rs6000.md: Ditto.
24129 * config/rs6000/vector.md: Ditto.
24130 * config/rs6000/vsx.md: Ditto.
24132 2018-02-21 Jason Merrill <jason@redhat.com>
24134 PR c++/84314 - ICE with templates and fastcall attribute.
24135 * attribs.c (build_type_attribute_qual_variant): Remove assert.
24137 2018-02-21 Jan Hubicka <hubicka@ucw.cz>
24139 * ipa-cp.c (determine_versionability): Fix comment typos.
24141 2018-02-21 Jan Hubicka <hubicka@ucw.cz>
24144 * ipa-cp.c (determine_versionability): Do not version functions caling
24147 2018-02-21 Martin Liska <mliska@suse.cz>
24150 * config/aarch64/aarch64.c (aarch64_print_hint_for_core_or_arch):
24151 Add "native" as a possible value.
24152 * config/aarch64/aarch64.h (HAVE_LOCAL_CPU_DETECT): Define
24153 the macro when native cpu detection is available.
24155 2018-02-21 Martin Liska <mliska@suse.cz>
24158 * common/config/arm/arm-common.c (arm_print_hint_for_arch_option):
24159 Add "native" as a possible value.
24160 * config/arm/arm.h (HAVE_LOCAL_CPU_DETECT): Define the macro
24161 when native cpu detection is available.
24163 2018-02-21 Jakub Jelinek <jakub@redhat.com>
24164 Martin Sebor <msebor@redhat.com>
24166 PR tree-optimization/84478
24167 * gimple-fold.h (get_range_strlen): Add a bool argument defaulted to
24169 * gimple-fold.c (get_range_strlen): Make minlen const and assume it
24170 can't be NULL. Change FUZZY from bool to int, for 1 add PHI/COND_EXPR
24171 support which is conservatively correct, for 2 only stay conservative
24172 for maxlen. Formatting and comment capitalization fixes. Add STRICT
24173 argument to the 2 argument get_range_strlen, adjust 6 arg
24174 get_range_strlen caller and clear minmaxlen[0] and [1] if it returned
24176 (get_maxval_strlen): Adjust 6 arg get_range_strlen caller.
24177 (gimple_fold_builtin_strlen): Pass true as last argument to
24180 2018-02-20 Martin Sebor <msebor@redhat.com>
24182 PR middle-end/84095
24183 * gimple-ssa-warn-restrict.c (builtin_memref::extend_offset_range): New.
24184 (builtin_memref::set_base_and_offset): Same. Handle inner references.
24185 (builtin_memref::builtin_memref): Factor out parts into
24186 set_base_and_offset and call it.
24188 2018-02-20 Richard Sandiford <richard.sandiford@linaro.org>
24190 PR middle-end/84406
24191 * optabs-query.c (find_widening_optab_handler_and_mode): If from_mode
24192 is a scalar_int_mode, assert that to_mode is a scalar_int_mode with
24193 greater precision. If to_mode is a MODE_PARTIAL_INT, stop the
24194 search at the associated MODE_INT.
24196 2018-02-20 Jeff Law <law@redhat.com>
24198 PR middle-end/82123
24199 PR tree-optimization/81592
24200 PR middle-end/79257
24201 * gimple-ssa-sprintf.c (format_integer): Query EVRP range analyzer
24202 for range data rather than using global data.
24203 * gimple-ssa-sprintf.c (get_int_range): Query EVRP range analyzer for
24204 range data rather than using global data.
24205 * gimple-ssa-sprintf.c (get_int_range): Accept vr_values parameter
24206 pass it to children as needed.
24207 (struct directive::fmtresult): Similarly.
24208 (struct directive::set_width): Similarly.
24209 (struct directive::set_precision): Similarly.
24210 (format_integer, format_directive, parse_directive): Similarly.
24211 (format_none): Accept unnamed vr_values parameter.
24212 (format_percent, format_floating, format_character): Similarly.
24213 (format_string, format_plain): Similarly.
24214 * gimple-ssa-sprintf.c (sprintf_dom_walker::handle_gimple_call): Query
24215 the EVRP range analyzer for range data rather than using global data.
24216 * gimple-ssa-sprintf.c: Include alloc-pool.h, vr-values.h and
24217 gimple-ssa-evrp-analyze.h
24218 (class sprintf_dom_walker): Add after_dom_children member function.
24219 Add evrp_range_analyzer member.
24220 (sprintf_dom_walker::before_dom_children): Call into the EVRP
24221 range analyzer as needed.
24222 (sprintf_dom_walker::after_dom_children): New member function.
24223 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::enter): Do nothing
24225 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
24226 (evrp_range_analyzer::pop_to_marker): Likewise.
24228 2018-02-20 Richard Sandiford <richard.sandiford@linaro.org>
24230 PR tree-optimization/84419
24231 * internal-fn.c (expand_call_mem_ref): Create a TARGET_MEM_REF
24232 with the required type if its current type is compatible but
24235 2018-02-20 Jakub Jelinek <jakub@redhat.com>
24237 PR middle-end/82004
24238 * match.pd (pow(C,x) -> exp(log(C)*x)): Delay all folding until
24239 after vectorization.
24241 2018-02-20 Martin Liska <mliska@suse.cz>
24244 * config/aarch64/aarch64.c (aarch64_print_hint_for_core_or_arch): Print
24245 possible values if we don't have a hint.
24247 2018-02-20 Martin Liska <mliska@suse.cz>
24251 * final.c (shorten_branches): Build align_tab array with one
24253 * opts.c (finish_options): Add alignment option limit check.
24254 (MAX_CODE_ALIGN): Likewise.
24255 (MAX_CODE_ALIGN_VALUE): Likewise.
24256 * doc/invoke.texi: Document maximum allowed option value for
24257 all -falign-* options.
24259 2018-02-19 Jakub Jelinek <jakub@redhat.com>
24262 * reg-notes.def (REG_CALL_ARG_LOCATION): New reg note.
24263 * insn-notes.def (NOTE_INSN_CALL_ARG_LOCATION): Remove.
24264 * var-tracking.c (emit_note_insn_var_location): Remove all references
24265 to NOTE_INSN_CALL_ARG_LOCATION.
24266 (emit_notes_in_bb): Emit arguments as REG_CALL_ARG_LOCATION note on
24267 the CALL_INSN rather than separate NOTE_INSN_CALL_ARG_LOCATION note.
24268 Use copy_rtx_if_shared.
24269 * dwarf2out.c (gen_subprogram_die): Use XEXP with 0 instead of
24270 NOTE_VAR_LOCATION on ca_loc->call_arg_loc_note.
24271 (dwarf2out_var_location): Remove handling of
24272 NOTE_INSN_CALL_ARG_LOCATION, instead handle REG_CALL_ARG_LOCATION note
24274 * final.c (final_scan_insn): Remove all references to
24275 NOTE_INSN_CALL_ARG_LOCATION.
24276 (rest_of_clean_state): Likewise. Remove REG_CALL_ARG_LOCATION notes
24277 before dumping final insns.
24278 * except.c (emit_note_eh_region_end): Remove all references to
24279 NOTE_INSN_CALL_ARG_LOCATION.
24280 * config/alpha/alpha.c (alpha_pad_function_end): Likewise.
24281 * config/c6x/c6x.c (c6x_gen_bundles): Likewise.
24282 * config/arc/arc.c (hwloop_optimize): Likewise.
24283 * config/arm/arm.c (create_fix_barrier): Likewise.
24284 * config/s390/s390.c (s390_chunkify_start): Likewise.
24285 * config/sh/sh.c (find_barrier): Likewise.
24286 * config/i386/i386.c (rest_of_insert_endbranch,
24287 ix86_seh_fixup_eh_fallthru): Likewise.
24288 * config/xtensa/xtensa.c (hwloop_optimize): Likewise.
24289 * config/iq2000/iq2000.c (final_prescan_insn): Likewise.
24290 * config/frv/frv.c (frv_function_prologue): Likewise.
24291 * emit-rtl.c (try_split): Likewise. Copy over REG_CALL_ARG_LOCATION
24293 (note_outside_basic_block_p): Remove all references to
24294 NOTE_INSN_CALL_ARG_LOCATION.
24295 * gengtype.c (adjust_field_rtx_def): Likewise.
24296 * print-rtl.c (rtx_writer::print_rtx_operand_code_0, print_insn):
24298 * jump.c (cleanup_barriers, delete_related_insns): Likewise.
24299 * cfgrtl.c (force_nonfallthru_and_redirect): Likewise.
24302 * builtins.c (builtin_mathfn_code): Don't check if CALL_EXPR_FN (t)
24305 PR tree-optimization/84452
24306 * tree-vect-patterns.c (vect_recog_pow_pattern): Don't call
24307 expand_simd_clones if targetm.simd_clone.compute_vecsize_and_simdlen
24310 2018-02-19 Martin Liska <mliska@suse.cz>
24313 * passes.def: Put pass_sancov_O0 before pass_lower_switch with -O0.
24315 2018-02-19 Martin Liska <mliska@suse.cz>
24316 Richard Sandiford <richard.sandiford@linaro.org>
24318 PR tree-optimization/82491
24319 * gimple-fold.c (get_base_constructor): Make earlier bail out
24322 2018-02-19 Carl Love <cel@us.ibm.com>
24324 * config/rs6000/rs6000-builtin.def: Change NEG macro expansions from
24325 BU_ALTIVEC_A to BU_P8V_AV_1 and BU_ALTIVEC_OVERLOAD_1 to
24327 * config/rs6000/rs6000-c.c: Change ALTIVEC_BUILTIN_VEC_NEG to
24328 P8V_BUILTIN_VEC_NEG.
24330 2018-02-19 Sebastian Perta <sebastian.perta@renesas.com>
24332 * config/rl78/rl78.md (movdf): New define expand.
24334 2018-02-19 Martin Liska <mliska@suse.cz>
24337 * doc/invoke.texi: Fix typo.
24338 * params.def (PARAM_MAX_LOOP_HEADER_INSNS): Likewise.
24340 2018-02-18 Segher Boessenkool <segher@kernel.crashing.org>
24342 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
24343 handle rs6000_single_float and rs6000_double_float specially for
24346 2018-02-16 Jeff Law <law@redhat.com>
24348 * config/rx/rx.c (add_pop_cfi_notes): New function.;
24349 (pop_regs): Use it.
24351 2018-02-16 Jakub Jelinek <jakub@redhat.com>
24354 * ipa-inline.c (inline_small_functions): Fix a typo.
24356 2018-02-16 Nathan Sidwell <nathan@acm.org>
24358 * doc/extend.texi (Backwards Compatibility): Americanize 'behaviour'.
24360 2018-02-16 Carl Love <cel@us.ibm.com>
24362 * config/rs6000/rs6000-builtin.def: Add BU_P8V_VSX_2 macro definition.
24363 Change FLOAT2 expansions from BU_VSX_2 to BU_P8V_VSX_2 and
24364 from BU_VSX_OVERLOAD_2 to BU_P8V_OVERLOAD_2.
24365 * config/rs6000/rs6000-c.c: Changed macro VSX_BUILTIN_VEC_FLOAT2
24366 expansion to P8V_BUILTIN_VEC_FLOAT2.
24368 2018-02-16 Vladimir Makarov <vmakarov@redhat.com>
24370 PR rtl-optimization/70023
24371 * lra-constraints.c (inherit_in_ebb): Take hard reg mode of
24372 src_regno into account.
24374 2018-02-16 Carl Love <cel@us.ibm.com>
24376 * config/rs6000/altivec.h: Remove vec_vextract4b and vec_vinsert4b.
24377 * config/rs6000/rs6000-builtin.def: Remove macro expansion for
24378 VEXTRACT4B, VINSERT4B, VINSERT4B_DI and VEXTRACT4B.
24379 * config/rs6000/rs6000.c: Remove case statements for
24380 P9V_BUILTIN_VEXTRACT4B, P9V_BUILTIN_VEC_VEXTRACT4B,
24381 P9V_BUILTIN_VINSERT4B, P9V_BUILTIN_VINSERT4B_DI,
24382 and P9V_BUILTIN_VEC_VINSERT4B.
24383 * config/rs6000/rs6000-c.c (altivec_expand_builtin): Remove entries for
24384 P9V_BUILTIN_VEC_VEXTRACT4B and P9V_BUILTIN_VEC_VINSERT4B.
24385 * config/rs6000/vsx.md:
24386 * doc/extend.texi: Remove vec_vextract4b, non ABI definitions for
24389 2018-02-16 Carl Love <cel@us.ibm.com>
24391 * config/rs6000/altivec.h: Add builtin names vec_extract4b
24393 * config/rs6000/rs6000-builtin.def: Add INSERT4B and EXTRACT4B
24395 * config/rs6000/rs6000-c.c: Add the definitions for
24396 P9V_BUILTIN_VEC_EXTRACT4B and P9V_BUILTIN_VEC_INSERT4B.
24397 * config/rs6000/rs6000.c (altivec_expand_builtin): Add
24398 P9V_BUILTIN_EXTRACT4B and P9V_BUILTIN_INSERT4B case statements.
24399 * config/rs6000/vsx.md: Add define_insn extract4b. Add define_expand
24400 definition for insert4b and define insn *insert3b_internal.
24401 * doc/extend.texi: Add documentation for vec_extract4b.
24403 2018-02-16 Nathan Sidwell <nathan@acm.org>
24405 * doc/extend.texi (Backwards Compatibility): Mention friend
24406 injection. Note for-scope is deprecated.
24407 * doc/invoke.texi (-ffriend-injection): Deprecate.
24409 2018-02-16 Segher Boessenkool <segher@kernel.crashing.org>
24411 * combine.c (try_combine): When adjusting LOG_LINKS for the destination
24412 that moved to I2, also allow destinations that are a paradoxical
24413 subreg (instead of a normal reg).
24415 2018-02-16 Oleg Endo <olegendo@gcc.gnu.org>
24418 * config/rx/rx.c (rx_fuse_in_memory_bitop): Convert shift operand
24421 2018-02-16 Richard Biener <rguenther@suse.de>
24423 PR tree-optimization/84037
24424 PR tree-optimization/84016
24426 * config/i386/i386.c (ix86_builtin_vectorization_cost):
24427 Adjust vec_construct for the fact we need additional higher latency
24428 128bit inserts for AVX256 and AVX512 vector builds.
24429 (ix86_add_stmt_cost): Scale vector construction cost for
24432 2018-02-16 Richard Biener <rguenther@suse.de>
24434 PR tree-optimization/84417
24435 * tree-ssa.c (non_rewritable_mem_ref_base): Properly constrain
24436 the MEM_REF offset when conversion to BIT_FIELD_REF is desired.
24437 (non_rewritable_lvalue_p): Likewise, use poly-ints.
24439 2018-02-16 Martin Liska <mliska@suse.cz>
24442 * internal-fn.def (ASAN_CHECK): Set proper flags.
24443 (ASAN_MARK): Likewise.
24445 2018-02-16 Julia Koval <julia.koval@intel.com>
24447 * config/i386/i386.c (ix86_option_override_internal): Remove PTA_CLWB
24448 from PTA_CANNONLAKE.
24450 2018-02-16 Jakub Jelinek <jakub@redhat.com>
24453 * config/aarch64/cortex-a57-fma-steering.c (fma_forest::merge_forest):
24454 Use ++iter rather than iter++ for std::list iterators.
24455 (func_fma_steering::dfs): Likewise. Don't delete nodes right away,
24456 defer deleting them until all nodes in the forest are processed. Do
24457 free even leaf nodes. Change to_process into auto_vec.
24460 * system.h (BROKEN_VALUE_INITIALIZATION): Define for GCC < 4.3.
24461 * vec.h (vec_default_construct): Use memset instead of placement new
24462 if BROKEN_VALUE_INITIALIZATION is defined.
24463 * hash-table.h (hash_table<Descriptor, Allocator>::empty_slow): Use
24464 memset instead of value initialization if BROKEN_VALUE_INITIALIZATION
24467 PR rtl-optimization/83723
24468 * lra-int.h (lra_substitute_pseudo): Add DEBUG_P argument.
24469 * lra.c (lra_substitute_pseudo): Likewise. If true, use
24470 gen_rtx_raw_SUBREG instead of gen_rtx_SUBREG. Pass DEBUG_P to
24472 (lra_substitute_pseudo_within_insn): Adjust lra_substitute_pseudo
24474 * lra-constraints.c (inherit_reload_reg, split_reg): Likewise.
24476 2018-02-16 Eric Botcazou <ebotcazou@adacore.com>
24478 PR rtl-optimization/81443
24479 * rtlanal.c (num_sign_bit_copies1) <SUBREG>: Do not propagate results
24480 from inner REGs to paradoxical SUBREGs.
24482 2018-02-16 Richard Biener <rguenther@suse.de>
24484 PR tree-optimization/84399
24485 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
24486 For operands we can analyze at their definition make sure we can
24487 analyze them at each use as well.
24489 2018-02-16 Richard Biener <rguenther@suse.de>
24491 PR tree-optimization/84190
24492 * tree-ssa.c (non_rewritable_mem_ref_base): Do not touch
24493 volatile accesses if the decl isn't volatile.
24495 2018-02-15 Jason Merrill <jason@redhat.com>
24497 PR c++/84314 - ICE with templates and fastcall attribute.
24498 * attribs.c (build_type_attribute_qual_variant): Don't clobber
24499 TYPE_CANONICAL on an existing type.
24501 2018-02-15 Jakub Jelinek <jakub@redhat.com>
24503 PR tree-optimization/84383
24504 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Don't look at
24505 dstoff nor call operand_equal_p if dstbase is NULL.
24507 PR tree-optimization/84334
24508 * match.pd ((A +- CST1) +- CST2 -> A + CST3): If A is
24509 also a CONSTANT_CLASS_P, punt.
24511 2018-02-14 Jim Wilson <jimw@sifive.com>
24513 * config/riscv/riscv.c (riscv_first_stack_step): Move locals after
24514 first SMALL_OPERAND check. New local min_second_step. Move assert
24515 to where locals are set. Add TARGET_RVC support.
24516 * config/riscv/riscv.h (C_SxSP_BITS, SWSP_REACH, SDSP_REACH): New.
24518 2018-02-14 Indu Bhagat <indu.bhagat@oracle.com>
24520 * doc/invoke.texi: Correct -Wformat-overflow code sample.
24522 2018-02-14 Martin Sebor <msebor@redhat.com>
24524 PR tree-optimization/83698
24525 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): For
24526 arrays constrain the offset range to their bounds.
24527 (builtin_access::strcat_overlap): Adjust the bounds of overlap offset.
24528 (builtin_access::overlap): Avoid setting the size of overlap if it's
24530 (maybe_diag_overlap): Also consider arrays when deciding what values
24531 of offsets to include in diagnostics.
24533 2018-02-14 Martin Sebor <msebor@redhat.com>
24536 * attribs.c (diag_attr_exclusions): Consider the exclusion(s)
24537 that correspond to the kind of a declaration.
24539 2018-02-14 John David Anglin <danglin@gcc.gnu.org>
24542 * config/pa/pa.md: Load address of PIC label using the linkage table
24543 if the label is nonlocal.
24545 2018-02-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
24547 * config/rs6000/rs6000.c (rs6000_option_override_internal): Issue
24548 warning message if user requests -maltivec=be.
24549 * doc/invoke.texi: Document deprecation of -maltivec=be.
24551 2018-02-14 Will Schmidt <will_schmidt@vnet.ibm.com>
24554 * config/rs6000/rs6000-c.c: Update definitions for
24555 ALTIVEC_BUILTIN_VEC_SLD, ALTIVEC_BUILTIN_VEC_SLDW,
24556 VEC_XXSLDWI and ALTIVEC_BUILTIN_VEC_XXPERMDI builtins.
24558 2018-02-14 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
24561 * config/i386/cetintrin.h: Remove _rdssp[d|q] and
24562 add _get_ssp intrinsics. Remove argument from
24563 __builtin_ia32_rdssp[d|q].
24564 * config/i386/i386-builtin-types.def: Add UINT_FTYPE_VOID.
24565 * config/i386/i386-builtin.def: Remove argument from
24566 __builtin_ia32_rdssp[d|q].
24567 * config/i386/i386.c: Use UINT_FTYPE_VOID. Use
24568 ix86_expand_special_args_builtin for _rdssp[d|q].
24569 * config/i386/i386.md: Remove argument from rdssp[si|di] insn.
24570 Clear register before usage.
24571 * doc/extend.texi: Remove argument from __builtin_ia32_rdssp[d|q].
24572 Add documentation for new _get_ssp and _inc_ssp intrinsics.
24574 2018-02-14 Richard Sandiford <richard.sandiford@linaro.org>
24576 PR tree-optimization/84357
24577 * tree-data-ref.c (object_address_invariant_in_loop_p): Check
24578 operand 1 of an ARRAY_REF too.
24580 2018-02-14 Oleg Endo <olegendo@gcc.gnu.org>
24583 * config/rx/rx-protos.h (rx_reg_dead_or_unused_after_insn,
24584 rx_copy_reg_dead_or_unused_notes, rx_fuse_in_memory_bitop): New
24586 (set_of_reg): New struct.
24587 (rx_find_set_of_reg, rx_find_use_of_reg): New functions.
24588 * config/rx/rx.c (rx_reg_dead_or_unused_after_insn,
24589 rx_copy_reg_dead_or_unused_notes, rx_fuse_in_memory_bitop): New
24591 * config/rx/rx.md (andsi3, iorsi3, xorsi3): Convert to insn_and_split.
24592 Split into bitclr, bitset, bitinvert patterns if appropriate.
24593 (*bitset, *bitinvert, *bitclr): Convert to named insn_and_split and
24594 use rx_fuse_in_memory_bitop.
24595 (*bitset_in_memory, *bitinvert_in_memory, *bitclr_in_memory): Convert
24596 to named insn, correct maximum insn length.
24598 2018-02-14 Jozef Lawrynowicz <jozefl.gcc@gmail.com>
24601 * machmode.def: Define a complex mode for PARTIAL_INT.
24602 * genmodes.c (complex_class): Return MODE_COMPLEX_INT for
24604 * doc/rtl.texi: Document CSPImode.
24605 * config/msp430/msp430.c (msp430_hard_regno_nregs): Add CPSImode
24607 (msp430_hard_regno_nregs_with_padding): Likewise.
24609 2018-02-13 Peter Bergner <bergner@vnet.ibm.com>
24612 * config/rs6000/rs6000.c (mem_operand_gpr): Disallow altivec addresses.
24614 2018-02-13 Segher Boessenkool <segher@kernel.crashing.org>
24616 PR rtl-optimization/84169
24617 * combine.c (try_combine): New variable split_i2i3. Set it to true if
24618 we generated a parallel as new i3 and we split that to new i2 and i3
24619 instructions. Handle split_i2i3 similar to swap_i2i3: scan the
24620 LOG_LINKs of i3 to see which of those need to link to i2 now. Link
24621 those to i2, not i1. Partially rewrite this scan code.
24623 2018-02-13 Jakub Jelinek <jakub@redhat.com>
24626 * stor-layout.c (place_field): For variable length fields, adjust
24627 offset_align afterwards not just based on the field's alignment,
24628 but also on the size.
24630 PR middle-end/84309
24631 * match.pd (pow(C,x) -> exp(log(C)*x)): Use exp2s and log2s instead
24632 of exps and logs in the use_exp2 case.
24634 2018-02-13 Jeff Law <law@redhat.com>
24636 * config/rl/rl78.c (rl78_attribute_table): Fix terminator and
24637 entry for "vector".
24639 * config/rl78/rl78.c (rl78_handle_func_attribute): Mark
24642 2018-02-13 Alexandre Oliva <aoliva@redhat.com>
24646 * common.opt (gas-loc-support, gas-locview-support): New.
24647 (ginline-points, ginternal-reset-location-views): New.
24648 * doc/invoke.texi: Document them. Use @itemx where intended.
24649 (gvariable-location-views): Adjust.
24650 * target.def (reset_location_view): New.
24651 * doc/tm.texi.in (DWARF2_ASM_VIEW_DEBUG_INFO): New.
24652 (TARGET_RESET_LOCATION_VIEW): New.
24653 * doc/tm.texi: Rebuilt.
24654 * dwarf2out.c (dwarf2out_default_as_loc_support): New.
24655 (dwarf2out_default_as_locview_support): New.
24656 (output_asm_line_debug_info): Use option variables.
24657 (dwarf2out_maybe_output_loclist_view_pair): Likewise.
24658 (output_loc_list): Likewise.
24659 (add_high_low_attributes): Check option variables.
24660 Don't output entry view attribute in strict mode.
24661 (gen_inlined_subroutine_die): Check option variables.
24662 (dwarf2out_inline_entry): Likewise.
24663 (init_sections_and_labels): Likewise.
24664 (dwarf2out_early_finish): Likewise.
24665 (maybe_reset_location_view): New, from...
24666 (dwarf2out_var_location): ... here. Call it.
24667 * debug.h (dwarf2out_default_as_loc_support): Declare.
24668 (dwarf2out_default_as_locview_support): Declare.
24669 * hooks.c (hook_int_rtx_insn_0): New.
24670 * hooks.h (hook_int_rtx_insn_0): Declare.
24671 * toplev.c (process_options): Take -gas-loc-support and
24672 -gas-locview-support from dwarf2out. Enable
24673 -gvariable-location-views by default only with locview
24674 assembler support. Enable -ginternal-reset-location-views by
24675 default only if the target defines the corresponding hook.
24676 Enable -ginline-points by default if location views are
24677 enabled; force it disabled if statement frontiers are
24679 * tree-inline.c (expand_call_inline): Check option variables.
24680 * tree-ssa-live.c (remove_unused_scope_block_p): Likewise.
24682 2018-02-13 Richard Sandiford <richard.sandiford@linaro.org>
24684 PR tree-optimization/84321
24685 * tree-vrp.c (intersect_range_with_nonzero_bits): Fix VR_ANTI_RANGE
24686 handling. Also check whether the anti-range contains any values
24687 that satisfy the mask; switch to a VR_RANGE if not.
24689 2018-02-13 Paolo Bonzini <bonzini@gnu.org>
24692 * internal-fn.def (ASAN_CHECK, ASAN_MARK): Revert changes to fnspec.
24694 2018-02-13 Martin Jambor <mjambor@suse.cz>
24697 * ipa-param-manipulation.c (ipa_modify_call_arguments): Use location
24698 of call statements, also set location of a load to a temporary.
24700 2018-02-13 Sebastian Perta <sebastian.perta@renesas.com>
24702 * config/rl78/rl78.c (add_vector_labels): New function.
24703 * config/rl78/rl78.c (rl78_handle_vector_attribute): New function.
24704 * config/rl78/rl78.c (rl78_start_function): Call add_vector_labels.
24705 * config/rl78/rl78.c (rl78_handle_func_attribute): Removed the assert
24706 which checks that no arguments are passed.
24707 * config/rl78/rl78.c (rl78_attribute_table): Add "vector" attribute.
24708 * doc/extend.texi: Documentation for the new attribute.
24710 2018-02-13 Andreas Schwab <schwab@suse.de>
24712 * config/riscv/linux.h (CPP_SPEC): Define.
24714 2018-02-13 Jakub Jelinek <jakub@redhat.com>
24717 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Pass
24718 OPTION_MASK_ISA_AES | OPTION_MASK_ISA_SSE2 instead of
24719 OPTION_MASK_ISA_AES as first argument to def_builtin_const
24720 for AES builtins. Pass OPTION_MASK_ISA_PCLMUL | OPTION_MASK_ISA_SSE2
24721 instead of OPTION_MASK_ISA_PCLMUL as first argument to
24722 def_builtin_const for __builtin_ia32_pclmulqdq128 builtin.
24723 * config/i386/wmmintrin.h: If __SSE2__ is not defined, enable it
24724 temporarily for AES and PCLMUL builtins.
24726 PR tree-optimization/84339
24727 * gimple-fold.c (get_range_strlen): Set *FLEXP to true when handling
24728 ARRAY_REF where first operand is array_at_struct_end_p COMPONENT_REF.
24731 PR middle-end/84309
24732 * match.pd (pow(C,x) -> exp(log(C)*x)): Optimize instead into
24733 exp2(log2(C)*x) if C is a power of 2 and c99 runtime is available.
24734 * generic-match-head.c (canonicalize_math_after_vectorization_p): New
24736 * gimple-match-head.c (canonicalize_math_after_vectorization_p): New
24738 * omp-simd-clone.h: New file.
24739 * omp-simd-clone.c: Include omp-simd-clone.h.
24740 (expand_simd_clones): No longer static.
24741 * tree-vect-patterns.c: Include fold-const-call.h, attribs.h,
24742 cgraph.h and omp-simd-clone.h.
24743 (vect_recog_pow_pattern): Optimize pow(C,x) to exp(log(C)*x).
24744 (vect_recog_widen_shift_pattern): Formatting fix.
24745 (vect_pattern_recog_1): Don't check optab for calls.
24748 * config/i386/sse.md (<avx512>_vpermi2var<mode>3_mask): Force
24749 operands[2] into a REG before using gen_lowpart on it.
24751 2018-02-12 Jeff Law <law@redhat.com>
24754 * config/sh/sh.c (find_barrier): Consider a sibling call
24757 * cse.c (try_back_substitute_reg): Move any REG_ARGS_SIZE note when
24758 successfully back substituting a reg.
24760 2018-02-12 Richard Biener <rguenther@suse.de>
24762 PR tree-optimization/84037
24763 * tree-vect-slp.c (vect_analyze_slp_cost): Add visited
24764 parameter, move visited init to caller.
24765 (vect_slp_analyze_operations): Separate cost from validity
24766 check, initialize visited once for all instances.
24767 (vect_schedule_slp): Analyze map to CSE vectorized nodes once
24769 * tree-vect-stmts.c (vect_model_simple_cost): Make early
24771 (vect_model_promotion_demotion_cost): Likewise.
24772 (vectorizable_bswap): Guard cost modeling with !slp_node
24773 instead of !PURE_SLP_STMT to avoid double-counting on hybrid
24775 (vectorizable_call): Likewise.
24776 (vectorizable_conversion): Likewise.
24777 (vectorizable_assignment): Likewise.
24778 (vectorizable_shift): Likewise.
24779 (vectorizable_operation): Likewise.
24780 (vectorizable_store): Likewise.
24781 (vectorizable_load): Likewise.
24782 (vectorizable_condition): Likewise.
24783 (vectorizable_comparison): Likewise.
24785 2018-02-12 Paolo Bonzini <bonzini@gnu.org>
24788 * internal-fn.def (ASAN_CHECK): Fix fnspec to account for return value.
24789 (ASAN_MARK): Fix fnspec to account for return value, change pointer
24790 argument from 'R' to 'W' so that the pointed-to datum is clobbered.
24792 2018-02-08 Jan Hubicka <hubicka@ucw.cz>
24794 PR middle-end/83665
24795 * params.def (inline-min-speedup): Increase from 8 to 15.
24796 (max-inline-insns-auto): Decrease from 40 to 30.
24797 * ipa-split.c (consider_split): Add some buffer for function to
24798 be considered inlining candidate.
24799 * invoke.texi (max-inline-insns-auto, inline-min-speedup): UPdate
24802 2018-02-12 Richard Biener <rguenther@suse.de>
24804 PR tree-optimization/84037
24805 * tree-vect-slp.c (vect_build_slp_tree_2): Try swapping the
24806 matched stmts if we cannot swap the non-matched ones.
24808 2018-02-12 Olga Makhotina <olga.makhotina@intel.com>
24810 * config/i386/avx512fintrin.h (_mm_mask_scalef_round_sd,
24811 _mm_maskz_scalef_round_sd, _mm_mask_scalef_round_ss,
24812 _mm_maskz_scalef_round_ss): New intrinsics.
24813 (__builtin_ia32_scalefsd_round, __builtin_ia32_scalefss_round): Fix.
24814 * config/i386/i386-builtin.def (__builtin_ia32_scalefsd_round,
24815 __builtin_ia32_scalefss_round): Remove.
24816 (__builtin_ia32_scalefsd_mask_round,
24817 __builtin_ia32_scalefss_mask_round): New intrinsics.
24818 * config/i386/sse.md (vmscalef<mode><round_name>): Renamed to ...
24819 (vmscalef<mode><mask_scalar_name><round_scalar_name>): ... this.
24820 ((match_operand:VF_128 2 "<round_nimm_predicate>"
24821 "<round_constraint>")): Changed to ...
24822 ((match_operand:VF_128 2 "<round_scalar_nimm_predicate>"
24823 "<round_scalar_constraint>")): ... this.
24824 ("vscalef<ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|
24825 %0, %1, %2<round_op3>}"): Changed to ...
24826 ("vscalef<ssescalarmodesuffix>\t{<round_scalar_mask_op3>%2, %1,
24827 %0<mask_scalar_operand3>|%0<mask_scalar_operand3>, %1,
24828 %2<round_scalar_mask_op3>}"): ... this.
24829 * config/i386/subst.md (round_scalar_nimm_predicate): New.
24831 2018-02-12 Olga Makhotina <olga.makhotina@intel.com>
24833 * config/i386/avx512fintrin.h (_mm_mask_sqrt_round_sd)
24834 (_mm_maskz_sqrt_round_sd, _mm_mask_sqrt_round_ss)
24835 (_mm_maskz_sqrt_round_ss): New intrinsics.
24836 (__builtin_ia32_sqrtsd_round, __builtin_ia32_sqrtss_round): Remove.
24837 (__builtin_ia32_sqrtsd_mask_round)
24838 (__builtin_ia32_sqrtss_mask_round): New builtins.
24839 * config/i386/i386-builtin.def (__builtin_ia32_sqrtsd_round)
24840 (__builtin_ia32_sqrtss_round): Remove.
24841 (__builtin_ia32_sqrtsd_mask_round)
24842 (__builtin_ia32_sqrtss_mask_round): New builtins.
24843 * config/i386/sse.md (vmsqrt<mode>2<round_name>): Renamed to ...
24844 (vmsqrt<mode>2<mask_scalar_name><round_scalar_name>): ... this.
24845 ((match_operand:VF_128 1 "vector_operand"
24846 "xBm,<round_constraint>")): Changed to ...
24847 ((match_operand:VF_128 1 "vector_operand"
24848 "xBm,<round_scalar_constraint>")): ... this.
24849 (vsqrt<ssescalarmodesuffix>\t{<round_op3>%1, %2, %0|
24850 %0, %2, %<iptr>1<round_op3>}): Changed to ...
24851 (vsqrt<ssescalarmodesuffix>\t{<round_scalar_mask_op3>%1, %2,
24852 %0<mask_scalar_operand3>|%0<mask_scalar_operand3>, %2,
24853 %<iptr>1<round_scalar_mask_op3>}): ... this.
24854 ((set_attr "prefix" "<round_prefix>")): Changed to ...
24855 ((set_attr "prefix" "<round_scalar_prefix>")): ... this.
24857 2018-02-11 Steven Munroe <munroesj@gcc.gnu.org>
24860 * config/rs6000/mmintrin.h (_mm_cmpeq_pi32 [_ARCH_PWR9]):
24861 Cast vec_cmpeq result to correct type.
24862 * config/rs6000/mmintrin.h (_mm_cmpgt_pi32 [_ARCH_PWR9]):
24863 Cast vec_cmpgt result to correct type.
24865 2018-02-11 Alexandre Oliva <aoliva@redhat.com>
24867 * final.c (final_scan_insn_1): Renamed from...
24868 (final_scan_insn): ... this. New wrapper, to recover
24869 seen from the outermost call in recursive ones.
24870 * config/sparc/sparc.c (output_return): Drop seen from call.
24871 (output_sibcall): Likewise.
24872 * config/visium/visium.c (output_branch): Likewise.
24874 2018-02-10 John David Anglin <danglin@gcc.gnu.org>
24876 * config/pa/pa.c (hppa_profile_hook): Mark SYMBOL_REF for _mcount as
24879 2018-02-10 Alan Modra <amodra@gmail.com>
24882 * config/rs6000/rs6000.md (split_stack_return): Remove (use ..).
24883 Specify LR as an input.
24885 2018-02-10 Jakub Jelinek <jakub@redhat.com>
24888 * omp-low.c (maybe_remove_omp_member_access_dummy_vars,
24889 remove_member_access_dummy_vars): New functions.
24890 (lower_omp_for, lower_omp_taskreg, lower_omp_target,
24891 lower_omp_1, execute_lower_omp): Use them.
24893 PR rtl-optimization/84308
24894 * shrink-wrap.c (spread_components): Release todo vector.
24896 2018-02-09 Vladimir Makarov <vmakarov@redhat.com>
24898 PR rtl-optimization/57193
24899 * ira-color.c (struct allocno_color_data): Add member
24900 conflict_allocno_hard_prefs.
24901 (update_conflict_allocno_hard_prefs): New.
24902 (bucket_allocno_compare_func): Add a preference based on
24903 conflict_allocno_hard_prefs.
24904 (push_allocno_to_stack): Update conflict_allocno_hard_prefs.
24905 (color_allocnos): Remove a dead code. Initiate
24906 conflict_allocno_hard_prefs. Call update_costs_from_prefs.
24908 2018-02-09 Jakub Jelinek <jakub@redhat.com>
24911 * config/rs6000/vsx.md (p9_xxbrq_v16qi): Change input operand
24912 constraint from =wa to wa. Avoid a subreg on the output operand,
24913 instead use a pseudo and subreg it in a move.
24914 (p9_xxbrd_<mode>): Changed to ...
24915 (p9_xxbrd_v2di): ... this insn, without VSX_D iterator.
24916 (p9_xxbrd_v2df): New expander.
24917 (p9_xxbrw_<mode>): Changed to ...
24918 (p9_xxbrw_v4si): ... this insn, without VSX_W iterator.
24919 (p9_xxbrw_v4sf): New expander.
24921 2018-02-09 Sebastian Perta <sebastian.perta@renesas.com>
24923 * config/rx/rx.md (movsicc): Update expander to be matched by GCC.
24925 2018-02-09 Peter Bergner <bergner@vnet.ibm.com>
24928 * config/rs6000/vsx.md (vsx_mul_v2di): Handle generating a 64-bit
24929 multiply in 32-bit mode.
24930 (vsx_div_v2di): Handle generating a 64-bit signed divide in 32-bit mode.
24931 (vsx_udiv_v2di): Handle generating a 64-bit unsigned divide in 32-bit
24934 2018-02-09 Sebastian Perta <sebastian.perta@renesas.com>
24936 * config/rx/constraints.md (CALL_OP_SYMBOL_REF): Added new constraint
24937 to allow or block "symbol_ref" depending on the value of TARGET_JSR.
24938 * config/rx/rx.md (call_internal): Use CALL_OP_SYMBOL_REF.
24939 * config/rx/rx.md (call_value_internal): Use CALL_OP_SYMBOL_REF.
24941 2018-02-09 Pierre-Marie de Rodat <derodat@adacore.com>
24944 * dwarf2out.c (is_trivial_indirect_ref): New function.
24945 (dwarf2out_late_global_decl): Do not generate a location
24946 attribute for variables that have a non-trivial DECL_VALUE_EXPR
24947 and that are not defined in the current unit.
24949 2018-02-09 Eric Botcazou <ebotcazou@adacore.com>
24951 * optabs.c (prepare_cmp_insn): Try harder to emit a direct comparison
24952 instead of a libcall for UNORDERED.
24954 2018-02-09 Tamar Christina <tamar.christina@arm.com>
24957 * config/arm/arm-c.c (arm_cpu_builtins): Un-define __ARM_FEATURE_LDREX,
24958 __ARM_ARCH_PROFILE, __ARM_ARCH_ISA_THUMB, __ARM_FP and __ARM_NEON_FP.
24960 2018-02-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24963 * config/s390/s390.c (s390_set_current_function): Invoke
24964 s390_indirect_branch_settings also if fndecl didn't change.
24966 2018-02-09 Alexandre Oliva <aoliva@redhat.com>
24968 * config/rs6000/rs6000.md (blockage): Set length to zero.
24970 2018-02-09 Eric Botcazou <ebotcazou@adacore.com>
24972 * expr.c (optimize_bitfield_assignment_op): Remove obsolete assertion.
24974 2018-02-09 Jakub Jelinek <jakub@redhat.com>
24977 * gcc.c (STATIC_LIBASAN_LIBS, STATIC_LIBTSAN_LIBS,
24978 STATIC_LIBLSAN_LIBS, STATIC_LIBUBSAN_LIBS): Handle -static like
24982 * var-tracking.c (vt_add_function_parameter): Punt for non-onepart
24983 PARALLEL incoming that failed vt_get_decl_and_offset check.
24985 PR middle-end/84237
24986 * output.h (bss_initializer_p): Add NAMED argument, defaulted to false.
24987 * varasm.c (bss_initializer_p): Add NAMED argument, if true, ignore
24989 (get_variable_section): For decls in named .bss* sections pass true as
24990 second argument to bss_initializer_p.
24992 2018-02-09 Marek Polacek <polacek@redhat.com>
24993 Jakub Jelinek <jakub@redhat.com>
24996 * fold-const.c (fold_indirect_ref_1): Use VECTOR_TYPE_P macro.
24997 Formatting fixes. Verify first that tree_fits_poly_int64_p (op01).
24998 Sync some changes from cxx_fold_indirect_ref.
25000 2018-02-09 Alexandre Oliva <aoliva@redhat.com>
25002 * cfgexpand.c (expand_gimple_basic_block): Handle inline entry
25004 * dwarf2out.c (dwarf2_debug_hooks): Enable inline_entry hook.
25005 (BLOCK_INLINE_ENTRY_LABEL): New.
25006 (dwarf2out_var_location): Disregard inline entry markers.
25007 (inline_entry_data): New struct.
25008 (inline_entry_data_hasher): New hashtable type.
25009 (inline_entry_data_hasher::hash): New.
25010 (inline_entry_data_hasher::equal): New.
25011 (inline_entry_data_table): New variable.
25012 (add_high_low_attributes): Add DW_AT_entry_pc and
25013 DW_AT_GNU_entry_view attributes if a pending entry is found
25014 in inline_entry_data_table. Add old entry_pc attribute only
25015 if debug nonbinding markers are disabled.
25016 (gen_inlined_subroutine_die): Set BLOCK_DIE if nonbinding
25017 markers are enabled.
25018 (block_within_block_p, dwarf2out_inline_entry): New.
25019 (dwarf2out_finish): Check that no entries remained in
25020 inline_entry_data_table.
25021 * final.c (reemit_insn_block_notes): Handle inline entry notes.
25022 (final_scan_insn, notice_source_line): Likewise.
25023 (rest_of_clean_state): Skip inline entry markers.
25024 * gimple-pretty-print.c (dump_gimple_debug): Handle inline entry
25026 * gimple.c (gimple_build_debug_inline_entry): New.
25027 * gimple.h (enum gimple_debug_subcode): Add
25028 GIMPLE_DEBUG_INLINE_ENTRY.
25029 (gimple_build_debug_inline_entry): Declare.
25030 (gimple_debug_inline_entry_p): New.
25031 (gimple_debug_nonbind_marker_p): Adjust.
25032 * insn-notes.def (INLINE_ENTRY): New.
25033 * print-rtl.c (rtx_writer::print_rtx_operand_code_0): Handle
25034 inline entry marker notes.
25035 (print_insn): Likewise.
25036 * rtl.h (NOTE_MARKER_P): Add INLINE_ENTRY support.
25037 (INSN_DEBUG_MARKER_KIND): Likewise.
25038 (GEN_RTX_DEBUG_MARKER_INLINE_ENTRY_PAT): New.
25039 * tree-inline.c (expand_call_inline): Build and insert
25040 debug_inline_entry stmt.
25041 * tree-ssa-live.c (remove_unused_scope_block_p): Preserve
25042 inline entry blocks early, if nonbind markers are enabled.
25043 (dump_scope_block): Dump fragment info.
25044 * var-tracking.c (reemit_marker_as_note): Handle inline entry note.
25045 * doc/gimple.texi (gimple_debug_inline_entry_p): New.
25046 (gimple_build_debug_inline_entry): New.
25047 * doc/invoke.texi (gstatement-frontiers, gno-statement-frontiers):
25048 Enable/disable inline entry points too.
25049 * doc/rtl.texi (NOTE_INSN_INLINE_ENTRY): New.
25050 (DEBUG_INSN): Describe inline entry markers.
25052 * common.opt (gvariable-location-views): New.
25053 (gvariable-location-views=incompat5): New.
25054 * config.in: Rebuilt.
25055 * configure: Rebuilt.
25056 * configure.ac: Test assembler for view support.
25057 * dwarf2asm.c (dw2_asm_output_symname_uleb128): New.
25058 * dwarf2asm.h (dw2_asm_output_symname_uleb128): Declare.
25059 * dwarf2out.c (var_loc_view): New typedef.
25060 (struct dw_loc_list_struct): Add vl_symbol, vbegin, vend.
25061 (dwarf2out_locviews_in_attribute): New.
25062 (dwarf2out_locviews_in_loclist): New.
25063 (dw_val_equal_p): Compare val_view_list of dw_val_class_view_lists.
25064 (enum dw_line_info_opcode): Add LI_adv_address.
25065 (struct dw_line_info_table): Add view.
25066 (RESET_NEXT_VIEW, RESETTING_VIEW_P): New macros.
25067 (DWARF2_ASM_VIEW_DEBUG_INFO): Define default.
25068 (zero_view_p): New variable.
25069 (ZERO_VIEW_P): New macro.
25070 (output_asm_line_debug_info): New.
25071 (struct var_loc_node): Add view.
25072 (add_AT_view_list, AT_loc_list): New.
25073 (add_var_loc_to_decl): Add view param. Test it against last.
25074 (new_loc_list): Add view params. Record them.
25075 (AT_loc_list_ptr): Handle loc and view lists.
25076 (view_list_to_loc_list_val_node): New.
25077 (print_dw_val): Handle dw_val_class_view_list.
25078 (size_of_die): Likewise.
25079 (value_format): Likewise.
25080 (loc_list_has_views): New.
25081 (gen_llsym): Set vl_symbol too.
25082 (maybe_gen_llsym, skip_loc_list_entry): New.
25083 (dwarf2out_maybe_output_loclist_view_pair): New.
25084 (output_loc_list): Output view list or entries too.
25085 (output_view_list_offset): New.
25086 (output_die): Handle dw_val_class_view_list.
25087 (output_dwarf_version): New.
25088 (output_compilation_unit_header): Use it.
25089 (output_skeleton_debug_sections): Likewise.
25090 (output_rnglists, output_line_info): Likewise.
25091 (output_pubnames, output_aranges): Update version comments.
25092 (output_one_line_info_table): Output view numbers in asm comments.
25093 (dw_loc_list): Determine current endview, pass it to new_loc_list.
25094 Call maybe_gen_llsym.
25095 (loc_list_from_tree_1): Adjust.
25096 (add_AT_location_description): Create view list attribute if
25097 needed, check it's absent otherwise.
25098 (convert_cfa_to_fb_loc_list): Adjust.
25099 (maybe_emit_file): Call output_asm_line_debug_info for test.
25100 (dwarf2out_var_location): Reset views as needed. Precompute
25101 add_var_loc_to_decl args. Call get_attr_min_length only if we have the
25102 attribute. Set view.
25103 (new_line_info_table): Reset next view.
25104 (set_cur_line_info_table): Call output_asm_line_debug_info for test.
25105 (dwarf2out_source_line): Likewise. Output view resets and labels to
25106 the assembler, or select appropriate line info opcodes.
25107 (prune_unused_types_walk_attribs): Handle dw_val_class_view_list.
25108 (optimize_string_length): Catch it. Adjust.
25109 (resolve_addr): Copy vl_symbol along with ll_symbol. Handle
25110 dw_val_class_view_list, and remove it if no longer needed.
25111 (hash_loc_list): Hash view numbers.
25112 (loc_list_hasher::equal): Compare them.
25113 (optimize_location_lists): Check whether a view list symbol is
25114 needed, and whether the locview attribute is present, and
25115 whether they match. Remove the locview attribute if no longer
25117 (index_location_lists): Call skip_loc_list_entry for test.
25118 (dwarf2out_finish): Call output_asm_line_debug_info for test.
25119 Use output_dwarf_version.
25120 * dwarf2out.h (enum dw_val_class): Add dw_val_class_view_list.
25121 (struct dw_val_node): Add val_view_list.
25122 * final.c (SEEN_NEXT_VIEW): New.
25123 (set_next_view_needed): New.
25124 (clear_next_view_needed): New.
25125 (maybe_output_next_view): New.
25126 (final_start_function): Rename to...
25127 (final_start_function_1): ... this. Take pointer to FIRST,
25128 add SEEN parameter. Emit param bindings in the initial view.
25129 (final_start_function): Reintroduce SEEN-less interface.
25130 (final): Rename to...
25131 (final_1): ... this. Take SEEN parameter. Output final pending
25132 next view at the end.
25133 (final): Reintroduce seen-less interface.
25134 (final_scan_insn): Output pending next view before switching
25135 sections or ending a block. Mark the next view as needed when
25136 outputting variable locations. Notify debug backend of section
25137 changes, and of location view changes.
25138 (rest_of_handle_final): Adjust.
25139 * toplev.c (process_options): Autodetect value for debug variable
25140 location views option. Warn on incompat5 without -gdwarf-5.
25141 * doc/invoke.texi (gvariable-location-views): New.
25142 (gvariable-location-views=incompat5): New.
25143 (gno-variable-location-views): New.
25145 2018-02-08 David Malcolm <dmalcolm@redhat.com>
25147 PR tree-optimization/84136
25148 * tree-cfg.c (find_taken_edge_computed_goto): Remove assertion
25149 that the result of find_edge is non-NULL.
25151 2018-02-08 Sergey Shalnov <sergey.shalnov@intel.com>
25154 * config/i386/x86-tune-costs.h (skylake_cost): Fix cost of
25155 storing integer register in SImode. Fix cost of 256 and 512
25156 byte aligned SSE register store.
25158 2018-02-08 Sergey Shalnov <sergey.shalnov@intel.com>
25160 * config/i386/i386.c (ix86_multiplication_cost): Fix
25161 multiplication cost for TARGET_AVX512DQ.
25163 2018-02-08 Marek Polacek <polacek@redhat.com>
25165 PR tree-optimization/84238
25166 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Verify the result of
25169 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
25171 PR tree-optimization/84265
25172 * tree-vect-stmts.c (vectorizable_store): Don't treat
25173 VMAT_CONTIGUOUS accesses as grouped.
25174 (vectorizable_load): Likewise.
25176 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
25178 PR tree-optimization/81635
25179 * wide-int.h (wi::round_down_for_mask, wi::round_up_for_mask): Declare.
25180 * wide-int.cc (wi::round_down_for_mask, wi::round_up_for_mask)
25181 (test_round_for_mask): New functions.
25182 (wide_int_cc_tests): Call test_round_for_mask.
25183 * tree-vrp.h (intersect_range_with_nonzero_bits): Declare.
25184 * tree-vrp.c (intersect_range_with_nonzero_bits): New function.
25185 * tree-data-ref.c (split_constant_offset_1): Use it to refine the
25186 range returned by get_range_info.
25188 2018-02-08 Jan Hubicka <hubicka@ucw.cz>
25191 * cgraph.h (symtab_node::output_to_lto_symbol_table_p): Declare
25192 * symtab.c: Include builtins.h
25193 (symtab_node::output_to_lto_symbol_table_p): Move here
25194 from lto-streamer-out.c:output_symbol_p.
25195 * lto-streamer-out.c (write_symbol): Turn early exit to assert.
25196 (output_symbol_p): Move all logic to symtab.c
25197 (produce_symtab): Update.
25199 2018-02-08 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25201 * config/s390/s390-opts.h (enum indirect_branch): Define.
25202 * config/s390/s390-protos.h (s390_return_addr_from_memory)
25203 (s390_indirect_branch_via_thunk)
25204 (s390_indirect_branch_via_inline_thunk): Add function prototypes.
25205 (enum s390_indirect_branch_type): Define.
25206 * config/s390/s390.c (struct s390_frame_layout, struct
25207 machine_function): Remove.
25208 (indirect_branch_prez10thunk_mask, indirect_branch_z10thunk_mask)
25209 (indirect_branch_table_label_no, indirect_branch_table_name):
25211 (INDIRECT_BRANCH_NUM_OPTIONS): Define macro.
25212 (enum s390_indirect_branch_option): Define.
25213 (s390_return_addr_from_memory): New function.
25214 (s390_handle_string_attribute): New function.
25215 (s390_attribute_table): Add new attribute handler.
25216 (s390_execute_label): Handle UNSPEC_EXECUTE_JUMP patterns.
25217 (s390_indirect_branch_via_thunk): New function.
25218 (s390_indirect_branch_via_inline_thunk): New function.
25219 (s390_function_ok_for_sibcall): When jumping via thunk disallow
25220 sibling call optimization for non z10 compiles.
25221 (s390_emit_call): Force indirect branch target to be a single
25222 register. Add r1 clobber for non-z10 compiles.
25223 (s390_emit_epilogue): Emit return jump via return_use expander.
25224 (s390_reorg): Handle JUMP_INSNs as execute targets.
25225 (s390_option_override_internal): Perform validity checks for the
25226 new command line options.
25227 (s390_indirect_branch_attrvalue): New function.
25228 (s390_indirect_branch_settings): New function.
25229 (s390_set_current_function): Invoke s390_indirect_branch_settings.
25230 (s390_output_indirect_thunk_function): New function.
25231 (s390_code_end): Implement target hook.
25232 (s390_case_values_threshold): Implement target hook.
25233 (TARGET_ASM_CODE_END, TARGET_CASE_VALUES_THRESHOLD): Define target
25235 * config/s390/s390.h (struct s390_frame_layout)
25236 (struct machine_function): Move here from s390.c.
25237 (TARGET_INDIRECT_BRANCH_NOBP_RET)
25238 (TARGET_INDIRECT_BRANCH_NOBP_JUMP)
25239 (TARGET_INDIRECT_BRANCH_NOBP_JUMP_THUNK)
25240 (TARGET_INDIRECT_BRANCH_NOBP_JUMP_INLINE_THUNK)
25241 (TARGET_INDIRECT_BRANCH_NOBP_CALL)
25242 (TARGET_DEFAULT_INDIRECT_BRANCH_TABLE)
25243 (TARGET_INDIRECT_BRANCH_THUNK_NAME_EXRL)
25244 (TARGET_INDIRECT_BRANCH_THUNK_NAME_EX)
25245 (TARGET_INDIRECT_BRANCH_TABLE): Define macros.
25246 * config/s390/s390.md (UNSPEC_EXECUTE_JUMP)
25247 (INDIRECT_BRANCH_THUNK_REGNUM): Define constants.
25248 (mnemonic attribute): Add values which aren't recognized
25250 ("*cjump_long", "*icjump_long", "*basr", "*basr_r"): Disable
25251 pattern for branch conversion. Fix mnemonic attribute.
25252 ("*c<code>", "*sibcall_br", "*sibcall_value_br", "*return"): Emit
25253 indirect branch via thunk if requested.
25254 ("indirect_jump", "<code>"): Expand patterns for branch conversion.
25255 ("*indirect_jump"): Disable for branch conversion using out of
25257 ("indirect_jump_via_thunk<mode>_z10")
25258 ("indirect_jump_via_thunk<mode>")
25259 ("indirect_jump_via_inlinethunk<mode>_z10")
25260 ("indirect_jump_via_inlinethunk<mode>", "*casesi_jump")
25261 ("casesi_jump_via_thunk<mode>_z10", "casesi_jump_via_thunk<mode>")
25262 ("casesi_jump_via_inlinethunk<mode>_z10")
25263 ("casesi_jump_via_inlinethunk<mode>", "*basr_via_thunk<mode>_z10")
25264 ("*basr_via_thunk<mode>", "*basr_r_via_thunk_z10")
25265 ("*basr_r_via_thunk", "return<mode>_prez10"): New pattern.
25266 ("*indirect2_jump"): Disable for branch conversion.
25267 ("casesi_jump"): Turn into expander and expand patterns for branch
25269 ("return_use"): New expander.
25270 ("*return"): Emit return via thunk and rename it to ...
25271 ("*return<mode>"): ... this one.
25272 * config/s390/s390.opt: Add new options and and enum for the
25275 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
25277 * lra-constraints.c (match_reload): Unconditionally use
25278 gen_lowpart_SUBREG, rather than selecting between that
25279 and equivalent gen_rtx_SUBREG code.
25281 2018-02-08 Richard Biener <rguenther@suse.de>
25283 PR tree-optimization/84233
25284 * tree-ssa-phiprop.c (propagate_with_phi): Use separate
25285 changed flag instead of boguously re-using phi_inserted.
25287 2018-02-08 Martin Jambor <mjambor@suse.cz>
25289 * hsa-gen.c (get_symbol_for_decl): Set program allocation for
25290 static local variables.
25292 2018-02-08 Richard Biener <rguenther@suse.de>
25294 PR tree-optimization/84278
25295 * tree-vect-stmts.c (vectorizable_store): When looking for
25296 smaller vector types to perform grouped strided loads/stores
25297 make sure the mode is supported by the target.
25298 (vectorizable_load): Likewise.
25300 2018-02-08 Wilco Dijkstra <wdijkstr@arm.com>
25302 * config/aarch64/aarch64.c (aarch64_components_for_bb):
25303 Increase LDP/STP opportunities by adding adjacent callee-saves.
25305 2018-02-08 Wilco Dijkstra <wdijkstr@arm.com>
25307 PR rtl-optimization/84068
25308 PR rtl-optimization/83459
25309 * haifa-sched.c (rank_for_schedule): Fix SCHED_PRESSURE_MODEL sorting.
25311 2018-02-08 Aldy Hernandez <aldyh@redhat.com>
25313 PR tree-optimization/84224
25314 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Remove assert.
25315 * calls.c (gimple_alloca_call_p): Only return TRUE when we have
25316 non-zero arguments.
25318 2018-02-07 Iain Sandoe <iain@codesourcery.com>
25321 * config/rs6000/altivec.md (*restore_world): Remove LR use.
25322 * config/rs6000/predicates.md (restore_world_operation): Adjust op
25323 count, remove one USE.
25325 2018-02-07 Michael Meissner <meissner@linux.vnet.ibm.com>
25327 * doc/install.texi (Configuration): Document the
25328 --with-long-double-format={ibm,ieee} PowerPC configuration
25332 * config/rs6000/rs6000.md (fix_trunc<SFDF:mode><QHI:mode>2):
25333 Convert from define_expand to be define_insn_and_split. Rework
25334 float/double/_Float128 conversions to QI/HI/SImode to work with
25335 both ISA 2.07 (power8) or ISA 3.0 (power9). Fix regression where
25336 conversions to QI/HImode types did a store and then a load to
25337 truncate the value. For conversions to VSX registers, don't split
25338 the insn, instead emit the code directly. Use the code iterator
25339 any_fix to combine signed and unsigned conversions.
25340 (fix<uns>_trunc<SFDF:mode>si2_p8): Likewise.
25341 (fixuns_trunc<SFDF:mode><QHI:mode>2): Likewise.
25342 (fix_trunc<IEEE128:mode><QHI:mode>2): Likewise.
25343 (fix<uns>_trunc<SFDF:mode><QHI:mode>2): Likewise.
25344 (fix_<mode>di2_hw): Likewise.
25345 (fixuns_<mode>di2_hw): Likewise.
25346 (fix_<mode>si2_hw): Likewise.
25347 (fixuns_<mode>si2_hw): Likewise.
25348 (fix<uns>_<IEEE128:mode><SDI:mode>2_hw): Likewise.
25349 (fix<uns>_trunc<IEEE128:mode><QHI:mode>2): Likewise.
25350 (fctiw<u>z_<mode>_smallint): Rename fctiw<u>z_<mode>_smallint to
25351 fix<uns>_trunc<SFDF:mode>si2_p8.
25352 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Delete, no longer used.
25353 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
25354 (fix<uns>_<mode>_mem): Likewise.
25355 (fctiw<u>z_<mode>_mem): Likewise.
25356 (fix<uns>_<mode>_mem): Likewise.
25357 (fix<uns>_trunc<SFDF:mode><QHSI:mode>2_mem): On ISA 3.0, prevent
25358 the register allocator from doing a direct move to the GPRs to do
25359 a store, and instead use the ISA 3.0 store byte/half-word from
25360 vector register instruction. For IEEE 128-bit floating point,
25361 also optimize stores of 32-bit ints.
25362 (fix<uns>_trunc<IEEE128:mode><QHSI:mode>2_mem): Likewise.
25364 2018-02-07 Alan Hayward <alan.hayward@arm.com>
25366 * genextract.c (push_pathstr_operand): New function to support
25368 (walk_rtx): Call push_pathstr_operand.
25369 (print_path): Support [a-zA-Z].
25371 2018-02-07 Richard Biener <rguenther@suse.de>
25373 PR tree-optimization/84037
25374 * tree-vectorizer.h (struct _loop_vec_info): Add ivexpr_map member.
25375 (cse_and_gimplify_to_preheader): Declare.
25376 (vect_get_place_in_interleaving_chain): Likewise.
25377 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
25379 (_loop_vec_info::~_loop_vec_info): Delete it.
25380 (cse_and_gimplify_to_preheader): New function.
25381 * tree-vect-slp.c (vect_get_place_in_interleaving_chain): Export.
25382 * tree-vect-stmts.c (vectorizable_store): CSE base and steps.
25383 (vectorizable_load): Likewise. For grouped stores always base
25384 the IV on the first element.
25385 * tree-vect-loop-manip.c (vect_loop_versioning): Unshare versioning
25386 condition before gimplifying.
25388 2018-02-07 Jakub Jelinek <jakub@redhat.com>
25390 * tree-eh.c (operation_could_trap_helper_p): Ignore honor_trapv for
25391 *DIV_EXPR and *MOD_EXPR.
25393 2018-02-07 H.J. Lu <hongjiu.lu@intel.com>
25396 * config/i386/i386.c (ix86_option_override_internal): Mask out
25397 the CF_SET bit when checking -fcf-protection.
25399 2018-02-07 Tom de Vries <tom@codesourcery.com>
25402 * omp-expand.c (expand_oacc_collapse_init): Ensure diff_type is large
25405 2018-02-07 Richard Biener <rguenther@suse.de>
25407 PR tree-optimization/84204
25408 * tree-chrec.c (chrec_fold_plus_1): Remove size limiting in
25411 PR tree-optimization/84205
25412 * graphite-isl-ast-to-gimple.c (binary_op_to_tree): Also
25413 special-case isl_ast_op_zdiv_r.
25415 PR tree-optimization/84223
25416 * graphite-scop-detection.c (gather_bbs::before_dom_children):
25417 Only add conditions from within the region.
25418 (gather_bbs::after_dom_children): Adjust.
25420 2018-02-07 Georg-Johann Lay <avr@gjlay.de>
25423 * config/avr/avr.h (GENERAL_REGNO_P, GENERAL_REG_P): New macros.
25424 * config/avr/avr.md: Only post-reload split REG-REG moves if
25425 either register is GENERAL_REG_P.
25427 2018-02-07 Jakub Jelinek <jakub@redhat.com>
25429 PR tree-optimization/84235
25430 * tree-ssa-scopedtables.c
25431 (avail_exprs_stack::simplify_binary_operation): Fir MINUS_EXPR, punt
25432 if the subtraction is performed in floating point type where NaNs are
25433 honored. For *DIV_EXPR, punt for ALL_FRACT_MODE_Ps where we can't
25434 build 1. Formatting fix.
25436 2018-02-06 Jakub Jelinek <jakub@redhat.com>
25439 * config/i386/i386.c (rest_of_insert_endbranch): Only skip
25440 NOTE_INSN_CALL_ARG_LOCATION after a call, not anything else,
25441 and skip it regardless of bb boundaries. Use CALL_P macro,
25442 don't test INSN_P (insn) together with CALL_P or JUMP_P check
25443 unnecessarily, formatting fix.
25445 2018-02-06 Michael Collison <michael.collison@arm.com>
25447 * config/arm/thumb2.md:
25448 (*thumb2_mov_negscc): Split only if TARGET_THUMB2 && !arm_restrict_it.
25449 (*thumb_mov_notscc): Ditto.
25451 2018-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
25454 * config/rs6000/rs6000.md (su code attribute): Use "u" for
25455 unsigned_fix, not "s".
25457 2018-02-06 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
25459 * configure.ac (gcc_fn_eh_frame_ro): New function.
25460 (gcc_cv_as_cfi_directive): Check both 32 and 64-bit assembler for
25461 correct .eh_frame permissions.
25462 * configure: Regenerate.
25464 2018-02-06 Andrew Jenner <andrew@codeourcery.com>
25466 * doc/invoke.texi: Add section for the PowerPC SPE backend. Remove
25467 irrelevant options.
25469 2018-02-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25471 * config/rs6000/rs6000.c (rs6000_option_override_internal):
25472 Display warning message for -mno-speculate-indirect-jumps.
25474 2018-02-06 Andrew Jenner <andrew@codesourcery.com>
25476 * config/powerpcspe/powerpcspe.opt: (msimple-fpu, mfpu) Add
25478 * config/powerpcspe/sysv4.opt (mbit-align): Likewise.
25480 2018-02-06 Aldy Hernandez <aldyh@redhat.com>
25482 PR tree-optimization/84225
25483 * tree-eh.c (find_trapping_overflow): Only call
25484 operation_no_trapping_overflow when ANY_INTEGRAL_TYPE_P.
25486 2018-02-06 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
25489 * config/i386/i386.c: Reimplement the check of possible options
25490 -mibt/-mshstk conbination. Change error messages.
25491 * doc/invoke.texi: Fix a typo: remove extra '='.
25493 2018-02-06 Marek Polacek <polacek@redhat.com>
25495 PR tree-optimization/84228
25496 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Skip debug statements.
25498 2018-02-06 Tamar Christina <tamar.christina@arm.com>
25501 * config/arm/arm.c (arm_print_asm_arch_directives): Record already
25502 emitted arch directives.
25503 * config/arm/arm-c.c (arm_cpu_builtins): Undefine __ARM_ARCH and
25504 __ARM_FEATURE_COPROC before changing architectures.
25506 2018-02-06 Richard Biener <rguenther@suse.de>
25508 * config/i386/i386.c (print_reg): Fix typo.
25509 (ix86_loop_unroll_adjust): Do not unroll beyond the original nunroll.
25511 2018-02-06 Eric Botcazou <ebotcazou@adacore.com>
25513 * configure: Regenerate.
25515 2018-02-05 Martin Sebor <msebor@redhat.com>
25517 PR tree-optimization/83369
25518 * tree-ssa-ccp.c (pass_post_ipa_warn::execute): Use %G to print
25521 2018-02-05 Martin Liska <mliska@suse.cz>
25523 * doc/invoke.texi: Cherry-pick upstream r323995.
25525 2018-02-05 Richard Sandiford <richard.sandiford@linaro.org>
25527 * ira.c (ira_init_register_move_cost): Adjust comment.
25529 2018-02-05 Martin Liska <mliska@suse.cz>
25531 PR gcov-profile/84137
25532 * doc/gcov.texi: Fix typo in documentation.
25534 2018-02-05 Martin Liska <mliska@suse.cz>
25536 PR gcov-profile/83879
25537 * doc/gcov.texi: Document necessity of --dynamic-list-data when
25538 using dlopen functionality.
25540 2018-02-05 Olga Makhotina <olga.makhotina@intel.com>
25542 * config/i386/avx512dqintrin.h (_mm_mask_range_sd, _mm_maskz_range_sd,
25543 _mm_mask_range_round_sd, _mm_maskz_range_round_sd, _mm_mask_range_ss,
25544 _mm_maskz_range_ss, _mm_mask_range_round_ss,
25545 _mm_maskz_range_round_ss): New intrinsics.
25546 (__builtin_ia32_rangesd128_round)
25547 (__builtin_ia32_rangess128_round): Remove.
25548 (__builtin_ia32_rangesd128_mask_round,
25549 __builtin_ia32_rangess128_mask_round): New builtins.
25550 * config/i386/i386-builtin.def (__builtin_ia32_rangesd128_round,
25551 __builtin_ia32_rangess128_round): Remove.
25552 (__builtin_ia32_rangesd128_mask_round,
25553 __builtin_ia32_rangess128_mask_round): New builtins.
25554 * config/i386/sse.md (ranges<mode><round_saeonly_name>): Renamed to ...
25555 (ranges<mode><mask_scalar_name><round_saeonly_scalar_name>): ... this.
25556 ((match_operand:VF_128 2 "<round_saeonly_nimm_predicate>"
25557 "<round_saeonly_constraint>")): Changed to ...
25558 ((match_operand:VF_128 2 "<round_saeonly_scalar_nimm_predicate>"
25559 "<round_saeonly_scalar_constraint>")): ... this.
25560 ("vrange<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
25561 %0, %1, %2<round_saeonly_op4>, %3}"): Changed to ...
25562 ("vrange<ssescalarmodesuffix>\t{%3, <round_saeonly_scalar_mask_op4>%2,
25563 %1, %0<mask_scalar_operand4>|%0<mask_scalar_operand4>, %1,
25564 %2<round_saeonly_scalar_mask_op4>, %3}"): ... this.
25566 2018-02-02 Andrew Jenner <andrew@codesourcery.com>
25568 * config/powerpcspe/powerpcspe.opt: Add Undocumented to irrelevant
25570 * config/powerpcspe/powerpcspe-tables.opt (rs6000_cpu_opt_value):
25571 Remove all values except native, 8540 and 8548.
25573 2018-02-02 H.J. Lu <hongjiu.lu@intel.com>
25575 * config/i386/i386.c (ix86_output_function_return): Pass
25576 INVALID_REGNUM, instead of -1, as invalid register number to
25577 indirect_thunk_name and output_indirect_thunk.
25579 2018-02-02 Julia Koval <julia.koval@intel.com>
25581 * config.gcc: Add -march=icelake.
25582 * config/i386/driver-i386.c (host_detect_local_cpu): Detect icelake.
25583 * config/i386/i386-c.c (ix86_target_macros_internal): Handle icelake.
25584 * config/i386/i386.c (processor_costs): Add m_ICELAKE.
25585 (PTA_ICELAKE, PTA_AVX512VNNI, PTA_GFNI, PTA_VAES, PTA_AVX512VBMI2,
25586 PTA_VPCLMULQDQ, PTA_RDPID, PTA_AVX512BITALG): New.
25587 (processor_target_table): Add icelake.
25588 (ix86_option_override_internal): Handle new PTAs.
25589 (get_builtin_code_for_version): Handle icelake.
25590 (M_INTEL_COREI7_ICELAKE): New.
25591 (fold_builtin_cpu): Handle icelake.
25592 * config/i386/i386.h (TARGET_ICELAKE, PROCESSOR_ICELAKE): New.
25593 * doc/invoke.texi: Add -march=icelake.
25595 2018-02-02 Julia Koval <julia.koval@intel.com>
25597 * config/i386/i386.c (ix86_option_override_internal): Change flags type
25598 to wide_int_bitmask.
25599 * wide-int-bitmask.h: New.
25601 2018-02-02 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
25604 * config/i386/i386.md: Replace Pmode with word_mode in
25605 builtin_setjmp_setup and builtin_longjmp to support x32.
25607 2018-02-01 Peter Bergner <bergner@vnet.ibm.com>
25611 * config/rs6000/driver-rs6000.c: #include "diagnostic.h".
25613 (rs6000_supported_cpu_names): New static variable.
25614 (linux_cpu_translation_table): Likewise.
25615 (elf_platform) <cpu>: Define new static variable and use it.
25616 Translate kernel AT_PLATFORM name to canonical name if needed.
25617 Error if platform name is unknown.
25619 2018-02-01 Aldy Hernandez <aldyh@redhat.com>
25622 * config/pa/predicates.md (base14_operand): Handle E_VOIDmode.
25624 2018-02-01 Jeff Law <law@redhat.com>
25627 * config/i386/i386.c (release_scratch_register_on_entry): Add new
25628 OFFSET and RELEASE_VIA_POP arguments. Use SP+OFFSET to restore
25629 the scratch if RELEASE_VIA_POP is false.
25630 (ix86_adjust_stack_and_probe_stack_clash): Un-constify SIZE.
25631 If we have to save a temporary register, decrement SIZE appropriately.
25632 Pass new arguments to release_scratch_register_on_entry.
25633 (ix86_adjust_stack_and_probe): Likewise.
25634 (ix86_emit_probe_stack_range): Pass new arguments to
25635 release_scratch_register_on_entry.
25637 2018-02-01 Uros Bizjak <ubizjak@gmail.com>
25639 PR rtl-optimization/84157
25640 * combine.c (change_zero_ext): Use REG_P predicate in
25641 front of HARD_REGISTER_P predicate.
25643 2018-02-01 Georg-Johann Lay <avr@gjlay.de>
25645 * config/avr/avr.c (avr_option_override): Move disabling of
25646 -fdelete-null-pointer-checks to...
25647 * common/config/avr/avr-common.c (avr_option_optimization_table):
25650 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
25652 PR tree-optimization/81635
25653 * tree-data-ref.c (split_constant_offset_1): For types that
25654 wrap on overflow, try to use range info to prove that wrapping
25657 2018-02-01 Renlin Li <renlin.li@arm.com>
25660 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
25661 TAILCALL_ADDR_REGS.
25662 (aarch64_register_move_cost): Likewise.
25663 * config/aarch64/aarch64.h (reg_class): Rename CALLER_SAVE_REGS to
25664 TAILCALL_ADDR_REGS.
25665 (REG_CLASS_NAMES): Likewise.
25666 (REG_CLASS_CONTENTS): Rename CALLER_SAVE_REGS to
25667 TAILCALL_ADDR_REGS. Remove IP registers.
25668 * config/aarch64/aarch64.md (Ucs): Update register constraint.
25670 2018-02-01 Richard Biener <rguenther@suse.de>
25672 * domwalk.h (dom_walker::dom_walker): Add additional constructor
25673 for specifying RPO order and allow NULL for that.
25674 * domwalk.c (dom_walker::dom_walker): Likewise.
25675 (dom_walker::walk): Handle NULL RPO order.
25676 * tree-into-ssa.c (rewrite_dom_walker): Do not walk dom children
25678 (rewrite_update_dom_walker): Likewise.
25679 (mark_def_dom_walker): Likewise.
25681 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
25683 * config/aarch64/aarch64-protos.h (aarch64_split_sve_subreg_move)
25684 (aarch64_maybe_expand_sve_subreg_move): Declare.
25685 * config/aarch64/aarch64.md (UNSPEC_REV_SUBREG): New unspec.
25686 * config/aarch64/predicates.md (aarch64_any_register_operand): New
25688 * config/aarch64/aarch64-sve.md (mov<mode>): Optimize subreg moves
25689 that are semantically a reverse operation.
25690 (*aarch64_sve_mov<mode>_subreg_be): New pattern.
25691 * config/aarch64/aarch64.c (aarch64_maybe_expand_sve_subreg_move):
25692 (aarch64_replace_reg_mode, aarch64_split_sve_subreg_move): New
25694 (aarch64_can_change_mode_class): For big-endian, forbid changes
25695 between two SVE modes if they have different element sizes.
25697 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
25699 * config/aarch64/aarch64.c (aarch64_expand_sve_const_vector): Prefer
25700 the TImode handling for big-endian targets.
25702 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
25704 * config/aarch64/aarch64-sve.md (sve_ld1rq): Replace with...
25705 (*sve_ld1rq<Vesize>): ... this new pattern. Handle all element sizes,
25707 * config/aarch64/aarch64.c (aarch64_expand_sve_widened_duplicate):
25708 Remove BSWAP handing for big-endian targets and use the form of
25709 LD1RQ appropariate for the mode.
25711 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
25713 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Handle
25714 all CONST_VECTOR_DUPLICATE_P vectors, not just those with a single
25715 duplicated element.
25717 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
25720 * config/aarch64/aarch64.c (aarch64_secondary_reload): Tighten
25721 check for operands that need to go through aarch64_sve_reload_be.
25723 2018-02-01 Jakub Jelinek <jakub@redhat.com>
25725 PR tree-optimization/81661
25726 PR tree-optimization/84117
25727 * tree-eh.h (rewrite_to_non_trapping_overflow): Declare.
25728 * tree-eh.c: Include gimplify.h.
25729 (find_trapping_overflow, replace_trapping_overflow,
25730 rewrite_to_non_trapping_overflow): New functions.
25731 * tree-vect-loop.c: Include tree-eh.h.
25732 (vect_get_loop_niters): Use rewrite_to_non_trapping_overflow.
25733 * tree-data-ref.c: Include tree-eh.h.
25734 (get_segment_min_max): Use rewrite_to_non_trapping_overflow.
25736 2018-01-31 Uros Bizjak <ubizjak@gmail.com>
25738 PR rtl-optimization/84123
25739 * combine.c (change_zero_ext): Check if hard register satisfies
25740 can_change_dest_mode before calling gen_lowpart_SUBREG.
25742 2018-01-31 Vladimir Makarov <vmakarov@redhat.com>
25745 * ira.c (ira_init_register_move_cost): Remove assert.
25747 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
25749 PR rtl-optimization/84071
25750 * doc/tm.texi.in (WORD_REGISTER_OPERATIONS): Add explicit case.
25751 * doc/tm.texi: Regenerate.
25753 2018-01-31 Richard Biener <rguenther@suse.de>
25755 PR tree-optimization/84132
25756 * tree-data-ref.c (analyze_miv_subscript): Properly
25757 check whether evolution_function_is_affine_multivariate_p
25758 before calling gcd_of_steps_may_divide_p.
25760 2018-01-31 Julia Koval <julia.koval@intel.com>
25763 * config/i386/i386.c (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
25764 * config/i386/i386.md (rdpid_rex64) New.
25765 (rdpid): Make 32bit only.
25767 2018-01-29 Aldy Hernandez <aldyh@redhat.com>
25770 * tree-pretty-print.c (dump_generic_node): Handle a TYPE_NAME with
25771 an IDENTIFIER_NODE for FUNCTION_TYPE's.
25773 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
25776 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
25778 * config/sparc/sparc.md (vxworks_load_got): Set the GOT register.
25780 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
25782 PR rtl-optimization/84071
25783 * combine.c (record_dead_and_set_regs_1): Record the source unmodified
25784 for a paradoxical SUBREG on a WORD_REGISTER_OPERATIONS target.
25786 2018-01-31 Claudiu Zissulescu <claziss@synopsys.com>
25788 * config/arc/arc.c (arc_handle_aux_attribute): New function.
25789 (arc_attribute_table): Add 'aux' attribute.
25790 (arc_in_small_data_p): Consider aux like variables.
25791 (arc_is_aux_reg_p): New function.
25792 (arc_asm_output_aligned_decl_local): Ignore 'aux' like variables.
25793 (arc_get_aux_arg): New function.
25794 (prepare_move_operands): Handle aux-register access.
25795 (arc_handle_aux_attribute): New function.
25796 * doc/extend.texi (ARC Variable attributes): Add subsection.
25798 2018-01-31 Claudiu Zissulescu <claziss@synopsys.com>
25800 * config/arc/arc-protos.h (arc_is_uncached_mem_p): Function proto.
25801 * config/arc/arc.c (arc_handle_uncached_attribute): New function.
25802 (arc_attribute_table): Add 'uncached' attribute.
25803 (arc_print_operand): Print '.di' flag for uncached memory
25805 (arc_in_small_data_p): Do not consider for small data the uncached
25807 (arc_is_uncached_mem_p): New function.
25808 * config/arc/predicates.md (compact_store_memory_operand): Check
25809 for uncached memory accesses.
25810 (nonvol_nonimm_operand): Likewise.
25811 * doc/extend.texi (ARC Type Attribute): New subsection.
25813 2018-01-31 Jakub Jelinek <jakub@redhat.com>
25816 * common.opt (falign-functions=, falign-jumps=, falign-labels=,
25817 falign-loops=): Add Optimization flag.
25819 2018-01-30 Jeff Law <law@redhat.com>
25822 * i386.c (ix86_adjust_stack_and_probe_stack_clash): New argument
25823 INT_REGISTERS_SAVED. Check it prior to calling
25824 get_scratch_register_on_entry.
25825 (ix86_adjust_stack_and_probe): Similarly.
25826 (ix86_emit_probe_stack_range): Similarly.
25827 (ix86_expand_prologue): Corresponding changes.
25829 2018-01-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
25832 * config/sol2.h (STARTFILE_ARCH_SPEC): Use -std=c*,
25833 -std=iso9899:199409 instead of -pedantic to select values-Xc.o.
25835 2018-01-30 Vladimir Makarov <vmakarov@redhat.com>
25838 * lra-constraints.c (curr_insn_transform): Process AND in the
25841 2018-01-30 Jakub Jelinek <jakub@redhat.com>
25843 PR rtl-optimization/83986
25844 * sched-deps.c (sched_analyze_insn): For frame related insns, add anti
25845 dependence against last_pending_memory_flush in addition to
25846 pending_jump_insns.
25848 2018-01-30 Alexandre Oliva <aoliva@redhat.com>
25850 PR tree-optimization/81611
25851 * tree-ssa-dom.c (simple_iv_increment_p): Skip intervening
25854 2018-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25857 * config/rs6000/rs6000.c (rs6000_internal_arg_pointer): Only return
25860 2018-01-30 Richard Biener <rguenther@suse.de>
25861 Jakub Jelinek <jakub@redhat.com>
25863 PR tree-optimization/84111
25864 * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1): Skip
25865 inner loops added during recursion, as they don't have up-to-date
25868 2018-01-30 Jan Hubicka <hubicka@ucw.cz>
25871 * ipa-inline.c (can_inline_edge_p): Break out late tests to...
25872 (can_inline_edge_by_limits_p): ... here.
25873 (can_early_inline_edge_p, check_callers,
25874 update_caller_keys, update_callee_keys, recursive_inlining,
25875 add_new_edges_to_heap, speculation_useful_p,
25876 inline_small_functions,
25877 inline_small_functions, flatten_function,
25878 inline_to_all_callers_1): Update.
25880 2018-01-30 Jan Hubicka <hubicka@ucw.cz>
25882 * profile-count.c (profile_count::combine_with_ipa_count): Handle
25885 2018-01-30 Richard Biener <rguenther@suse.de>
25887 PR tree-optimization/83008
25888 * tree-vect-slp.c (vect_analyze_slp_cost_1): Properly cost
25889 invariant and constant vector uses in stmts when they need
25890 more than one stmt.
25892 2018-01-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
25895 * configure.ac (gcc_cv_as_shf_merge): Disable on Solaris 10/x86.
25896 * configure: Regenerate.
25898 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
25900 * config/aarch64/aarch64-sve.md (*vec_extract<mode><Vel>_0): New
25902 (*vec_extract<mode><Vel>_v128): Require a nonzero lane number.
25903 Use gen_rtx_REG rather than gen_lowpart.
25905 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
25907 * lra-constraints.c (match_reload): Use subreg_lowpart_offset
25908 rather than 0 when creating partial subregs.
25910 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
25912 * vec-perm-indices.c (vec_perm_indices::series_p): Give examples
25915 2018-01-29 Michael Meissner <meissner@linux.vnet.ibm.com>
25918 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): If DFmode
25919 and SFmode can go in Altivec registers (-mcpu=power7 for DFmode,
25920 -mcpu=power8 for SFmode) don't set the PRE_INCDEC or PRE_MODIFY
25921 flags. This restores the settings used before the 2017-07-24.
25922 Turning off pre increment/decrement/modify allows IVOPTS to
25923 optimize DF/SF loops where the index is an int.
25925 2018-01-29 Richard Biener <rguenther@suse.de>
25926 Kelvin Nilsen <kelvin@gcc.gnu.org>
25929 * tree-vect-stmts.c (vectorizable_call): Don't call
25930 targetm.vectorize_builtin_md_vectorized_function if callee is
25933 2018-01-22 Carl Love <cel@us.ibm.com>
25935 * doc/extend.tex: Fix typo in second arg in
25936 __builtin_bcdadd_{lt|eq|gt|ov} and __builtin_bcdsub_{lt|eq|gt|ov}.
25938 2018-01-29 Richard Biener <rguenther@suse.de>
25940 PR tree-optimization/84086
25941 * tree-ssanames.c: Include cfgloop.h and tree-scalar-evolution.h.
25942 (flush_ssaname_freelist): When SSA names were released reset
25943 the SCEV hash table.
25945 2018-01-29 Richard Biener <rguenther@suse.de>
25947 PR tree-optimization/84057
25948 * tree-ssa-loop-ivcanon.c (unloop_loops): Deal with already
25949 removed paths when removing edges.
25951 2018-01-27 H.J. Lu <hongjiu.lu@intel.com>
25953 * doc/invoke.texi: Replace -mfunction-return==@var{choice} with
25954 -mfunction-return=@var{choice}.
25956 2018-01-27 Bernd Edlinger <bernd.edlinger@hotmail.de>
25958 PR diagnostic/84034
25959 * diagnostic-show-locus.c (get_line_width_without_trailing_whitespace):
25960 Handle CR like TAB.
25961 (layout::print_source_line): Likewise.
25962 (test_get_line_width_without_trailing_whitespace): Add test cases.
25964 2018-01-27 Jakub Jelinek <jakub@redhat.com>
25966 PR middle-end/84040
25967 * sched-deps.c (sched_macro_fuse_insns): Return immediately for
25970 2018-01-26 Jim Wilson <jimw@sifive.com>
25972 * config/riscv/riscv.h (MAX_FIXED_MODE_SIZE): New.
25974 * config/riscv/elf.h (LIB_SPEC): Don't include -lgloss when nosys.specs
25977 2018-01-26 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25979 * config/aarch64/aarch64.md: Add peepholes for CMP + SUB -> SUBS
25980 and CMP + SUB-immediate -> SUBS.
25982 2018-01-26 Martin Sebor <msebor@redhat.com>
25984 PR tree-optimization/83896
25985 * tree-ssa-strlen.c (get_string_len): Rename...
25986 (get_string_cst_length): ...to this. Return HOST_WIDE_INT.
25987 Avoid assuming length is constant.
25988 (handle_char_store): Use HOST_WIDE_INT for string length.
25990 2018-01-26 Uros Bizjak <ubizjak@gmail.com>
25993 * config/i386/i386.md (*andndi3_doubleword): Add earlyclobber
25994 to (=&r,r,rm) alternative. Add (=r,0,rm) and (=r,r,0) alternatives.
25996 2018-01-26 Richard Biener <rguenther@suse.de>
25998 PR rtl-optimization/84003
25999 * dse.c (record_store): Only record redundant stores when
26000 the earlier store aliases at least all accesses the later one does.
26002 2018-01-26 Jakub Jelinek <jakub@redhat.com>
26004 PR rtl-optimization/83985
26005 * dce.c (deletable_insn_p): Return false for separate shrink wrapping
26006 REG_CFA_RESTORE insns.
26007 (delete_unmarked_insns): Don't ignore separate shrink wrapping
26008 REG_CFA_RESTORE insns here.
26011 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Don't
26012 use SSA_NAME_VAR as base for SSA_NAMEs with non-NULL SSA_NAME_VAR.
26014 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
26016 * config/arc/arc-arch.h (arc_tune_attr): Add ARC_TUNE_CORE_3.
26017 * config/arc/arc.c (arc_sched_issue_rate): Use ARC_TUNE_... .
26018 (arc_init): Likewise.
26019 (arc_override_options): Likewise.
26020 (arc_file_start): Choose Tag_ARC_CPU_variation based on arc_tune
26022 (hwloop_fail): Use TARGET_DBNZ when we want to check for dbnz insn
26024 * config/arc/arc.h (TARGET_DBNZ): Define.
26025 * config/arc/arc.md (attr tune): Add core_3, use ARC_TUNE_... to
26026 properly set the tune attribute.
26027 (dbnz): Use TARGET_DBNZ guard.
26028 * config/arc/arc.opt (mtune): Add core3 option.
26030 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
26032 * config/arc/arc.c (arc_delegitimize_address_0): Refactored to
26033 recognize new pic like addresses.
26034 (arc_delegitimize_address): Clean up.
26036 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
26038 * config/arc/arc-arches.def: Option mrf16 valid for all
26040 * config/arc/arc-c.def (__ARC_RF16__): New predefined macro.
26041 * config/arc/arc-cpus.def (em_mini): New cpu with rf16 on.
26042 * config/arc/arc-options.def (FL_RF16): Add mrf16 option.
26043 * config/arc/arc-tables.opt: Regenerate.
26044 * config/arc/arc.c (arc_conditional_register_usage): Handle
26045 reduced register file case.
26046 (arc_file_start): Set must have build attributes.
26047 * config/arc/arc.h (MAX_ARC_PARM_REGS): Conditional define using
26048 mrf16 option value.
26049 * config/arc/arc.opt (mrf16): Add new option.
26050 * config/arc/elf.h (ATTRIBUTE_PCS): Define.
26051 * config/arc/genmultilib.awk: Handle new mrf16 option.
26052 * config/arc/linux.h (ATTRIBUTE_PCS): Define.
26053 * config/arc/t-multilib: Regenerate.
26054 * doc/invoke.texi (ARC Options): Document mrf16 option.
26056 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
26058 * config/arc/arc-protos.h: Add arc_is_secure_call_p proto.
26059 * config/arc/arc.c (arc_handle_secure_attribute): New function.
26060 (arc_attribute_table): Add 'secure_call' attribute.
26061 (arc_print_operand): Print secure call operand.
26062 (arc_function_ok_for_sibcall): Don't optimize tail calls when
26064 (arc_is_secure_call_p): New function. * config/arc/arc.md
26065 (call_i): Add support for sjli instruction.
26066 (call_value_i): Likewise.
26067 * config/arc/constraints.md (Csc): New constraint.
26069 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
26070 John Eric Martin <John.Martin@emmicro-us.com>
26072 * config/arc/arc-protos.h: Add arc_is_jli_call_p proto.
26073 * config/arc/arc.c (_arc_jli_section): New struct.
26074 (arc_jli_section): New type.
26075 (rc_jli_sections): New static variable.
26076 (arc_handle_jli_attribute): New function.
26077 (arc_attribute_table): Add jli_always and jli_fixed attribute.
26078 (arc_file_end): New function.
26079 (TARGET_ASM_FILE_END): Define.
26080 (arc_print_operand): Reuse 'S' letter for JLI output instruction.
26081 (arc_add_jli_section): New function.
26082 (jli_call_scan): Likewise.
26083 (arc_reorg): Call jli_call_scan.
26084 (arc_output_addsi): Remove 'S' from printing asm operand.
26085 (arc_is_jli_call_p): New function.
26086 * config/arc/arc.md (movqi_insn): Remove 'S' from printing asm
26088 (movhi_insn): Likewise.
26089 (movsi_insn): Likewise.
26090 (movsi_set_cc_insn): Likewise.
26091 (loadqi_update): Likewise.
26092 (load_zeroextendqisi_update): Likewise.
26093 (load_signextendqisi_update): Likewise.
26094 (loadhi_update): Likewise.
26095 (load_zeroextendhisi_update): Likewise.
26096 (load_signextendhisi_update): Likewise.
26097 (loadsi_update): Likewise.
26098 (loadsf_update): Likewise.
26099 (movsicc_insn): Likewise.
26100 (bset_insn): Likewise.
26101 (bxor_insn): Likewise.
26102 (bclr_insn): Likewise.
26103 (bmsk_insn): Likewise.
26104 (bicsi3_insn): Likewise.
26105 (cmpsi_cc_c_insn): Likewise.
26106 (movsi_ne): Likewise.
26107 (movsi_cond_exec): Likewise.
26108 (clrsbsi2): Likewise.
26109 (norm_f): Likewise.
26119 (call_i): Remove 'S' asm letter, add jli instruction.
26120 (call_value_i): Likewise.
26121 * config/arc/arc.op (mjli-always): New option.
26122 * config/arc/constraints.md (Cji): New constraint.
26123 * config/arc/fpx.md (addsf3_fpx): Remove 'S' from printing asm
26125 (subsf3_fpx): Likewise.
26126 (mulsf3_fpx): Likewise.
26127 * config/arc/simdext.md (vendrec_insn): Remove 'S' from printing
26129 * doc/extend.texi (ARC): Document 'jli-always' and 'jli-fixed'
26130 function attrbutes.
26131 * doc/invoke.texi (ARC): Document mjli-always option.
26133 2018-01-26 Sebastian Perta <sebastian.perta@renesas.com>
26135 * config/rl78/rl78.c (rl78_addsi3_internal): If operand 2 is const
26136 avoid addition with 0 and use incw and decw where possible.
26138 2018-01-26 Richard Biener <rguenther@suse.de>
26140 PR tree-optimization/81082
26141 * fold-const.c (fold_plusminus_mult_expr): Do not perform the
26142 association if it requires casting to unsigned.
26143 * match.pd ((A * C) +- (B * C) -> (A+-B)): New patterns derived
26144 from fold_plusminus_mult_expr to catch important cases late when
26145 range info is available.
26147 2018-01-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
26149 * config/i386/sol2.h (USE_HIDDEN_LINKONCE): Remove.
26150 * configure.ac (hidden_linkonce): New test.
26151 * configure: Regenerate.
26152 * config.in: Regenerate.
26154 2018-01-26 Julia Koval <julia.koval@intel.com>
26156 * config/i386/avx512bitalgintrin.h (_mm512_bitshuffle_epi64_mask,
26157 _mm512_mask_bitshuffle_epi64_mask, _mm256_bitshuffle_epi64_mask,
26158 _mm256_mask_bitshuffle_epi64_mask, _mm_bitshuffle_epi64_mask,
26159 _mm_mask_bitshuffle_epi64_mask): Fix type.
26160 * config/i386/i386-builtin-types.def (UHI_FTYPE_V2DI_V2DI_UHI,
26161 USI_FTYPE_V4DI_V4DI_USI): Remove.
26162 * config/i386/i386-builtin.def (__builtin_ia32_vpshufbitqmb512_mask,
26163 __builtin_ia32_vpshufbitqmb256_mask,
26164 __builtin_ia32_vpshufbitqmb128_mask): Fix types.
26165 * config/i386/i386.c (ix86_expand_args_builtin): Remove old types.
26166 * config/i386/sse.md (VI1_AVX512VLBW): Change types.
26168 2018-01-26 Alan Modra <amodra@gmail.com>
26171 * config/rs6000/rs6000-p8swap.c (rtx_is_swappable_p): Exclude
26172 UNSPEC_VBPERMQ. Sort other unspecs.
26174 2018-01-25 David Edelsohn <dje.gcc@gmail.com>
26176 * doc/invoke.texi (PowerPC Options): Document 'native' cpu type.
26178 2018-01-25 Jan Hubicka <hubicka@ucw.cz>
26180 PR middle-end/83055
26181 * predict.c (drop_profile): Do not push/pop cfun; update also
26183 (handle_missing_profiles): Fix logic looking for zero profiles.
26185 2018-01-25 Jakub Jelinek <jakub@redhat.com>
26187 PR middle-end/83977
26188 * ipa-fnsummary.c (compute_fn_summary): Clear can_change_signature
26189 on functions with #pragma omp declare simd or functions with simd
26191 * omp-simd-clone.c (expand_simd_clones): Revert 2018-01-24 change.
26192 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
26193 Remove trailing \n from warning_at calls.
26195 2018-01-25 Tom de Vries <tom@codesourcery.com>
26198 * config/nvptx/nvptx.c (nvptx_single): Add exit insn after noreturn call
26199 for neutered workers.
26201 2018-01-24 Joseph Myers <joseph@codesourcery.com>
26204 * config/m68k/m68k.c (m68k_promote_function_mode): New function.
26205 (TARGET_PROMOTE_FUNCTION_MODE): New macro.
26207 2018-01-24 Jeff Law <law@redhat.com>
26210 * i386.c (get_probe_interval): Move to earlier point.
26211 (ix86_compute_frame_layout): If -fstack-clash-protection and
26212 the frame is larger than the probe interval, then use pushes
26213 to save registers rather than reg->mem moves.
26214 (ix86_expand_prologue): Remove conditional for int_registers_saved
26217 2018-01-24 Vladimir Makarov <vmakarov@redhat.com>
26220 * ira-build.c (setup_min_max_allocno_live_range_point): Set up
26221 min/max for never referenced object.
26223 2018-01-24 Jakub Jelinek <jakub@redhat.com>
26225 PR middle-end/83977
26226 * tree.c (free_lang_data_in_decl): Don't clear DECL_ABSTRACT_ORIGIN
26228 * omp-low.c (create_omp_child_function): Remove "omp declare simd"
26229 attributes from DECL_ATTRIBUTES (decl) without affecting
26230 DECL_ATTRIBUTES (current_function_decl).
26231 * omp-simd-clone.c (expand_simd_clones): Ignore DECL_ARTIFICIAL
26232 functions with non-NULL DECL_ABSTRACT_ORIGIN.
26234 2018-01-24 Richard Sandiford <richard.sandiford@linaro.org>
26236 PR tree-optimization/83979
26237 * fold-const.c (fold_comparison): Use constant_boolean_node
26238 instead of boolean_{true,false}_node.
26240 2018-01-24 Jan Hubicka <hubicka@ucw.cz>
26242 * ipa-profile.c (ipa_propagate_frequency_1): Fix logic skipping calls
26245 2018-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26247 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
26248 Simplify the clause that sets the length attribute.
26249 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
26250 (*sibcall_nonlocal_sysv<mode>): Clean up code block; simplify the
26251 clause that sets the length attribute.
26252 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
26254 2018-01-24 Tom de Vries <tom@codesourcery.com>
26257 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG_2): Define to 1.
26258 (nvptx_pc_set, nvptx_condjump_label): New function. Copy from jump.c.
26259 Add strict parameter.
26260 (prevent_branch_around_nothing): Insert dummy insn between branch to
26261 label and label with no ptx insn inbetween.
26262 * config/nvptx/nvptx.md (define_insn "fake_nop"): New insn.
26264 2018-01-24 Tom de Vries <tom@codesourcery.com>
26267 * config/nvptx/nvptx.c (nvptx_single): Add exit insn after noreturn call
26268 for neutered threads in warp.
26269 * config/nvptx/nvptx.md (define_insn "exit"): New insn.
26271 2018-01-24 Richard Biener <rguenther@suse.de>
26273 PR tree-optimization/83176
26274 * tree-chrec.c (chrec_fold_plus_1): Handle (signed T){(T) .. }
26277 2018-01-24 Richard Biener <rguenther@suse.de>
26279 PR tree-optimization/82819
26280 * graphite-isl-ast-to-gimple.c (binary_op_to_tree): Avoid
26281 code generating pluses that are no-ops in the target precision.
26283 2018-01-24 Richard Biener <rguenther@suse.de>
26285 PR middle-end/84000
26286 * tree-cfg.c (replace_loop_annotate): Handle annot_expr_parallel_kind.
26288 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
26290 * cfgcleanup.c (try_crossjump_to_edge): Use combine_with_count
26291 to merge probabilities.
26292 * predict.c (probably_never_executed): Also mark as cold functions
26293 with global 0 profile and guessed local profile.
26294 * profile-count.c (profile_probability::combine_with_count): New
26296 * profile-count.h (profile_probability::operator*,
26297 profile_probability::operator*=, profile_probability::operator/,
26298 profile_probability::operator/=): Reduce precision to adjusted
26299 and set value to guessed on contradictory divisions.
26300 (profile_probability::combine_with_freq): Remove.
26301 (profile_probability::combine_wiht_count): Declare.
26302 (profile_count::force_nonzero):: Set to adjusted.
26303 (profile_count::probability_in):: Set quality to adjusted.
26304 * tree-ssa-tail-merge.c (replace_block_by): Use
26305 combine_with_count.
26307 2018-01-23 Andrew Waterman <andrew@sifive.com>
26308 Jim Wilson <jimw@sifive.com>
26310 * config/riscv/riscv.c (riscv_stack_boundary): New.
26311 (riscv_option_override): Set riscv_stack_boundary. Handle
26312 riscv_preferred_stack_boundary_arg.
26313 * config/riscv/riscv.h (MIN_STACK_BOUNDARY, ABI_STACK_BOUNDARY): New.
26314 (BIGGEST_ALIGNMENT): Set to STACK_BOUNDARY.
26315 (STACK_BOUNDARY): Set to riscv_stack_boundary.
26316 (RISCV_STACK_ALIGN): Use STACK_BOUNDARY.
26317 * config/riscv/riscv.opt (mpreferred-stack-boundary): New.
26318 * doc/invoke.tex (RISC-V Options): Add -mpreferred-stack-boundary.
26320 2018-01-23 H.J. Lu <hongjiu.lu@intel.com>
26323 * config/i386/i386.c (ix86_expand_prologue): Use cost reference
26324 of struct ix86_frame.
26325 (ix86_expand_epilogue): Likewise. Add a local variable for
26326 the reg_save_offset field in struct ix86_frame.
26328 2018-01-23 Bin Cheng <bin.cheng@arm.com>
26330 PR tree-optimization/82604
26331 * tree-loop-distribution.c (enum partition_kind): New enum item
26332 PKIND_PARTIAL_MEMSET.
26333 (partition_builtin_p): Support above new enum item.
26334 (generate_code_for_partition): Ditto.
26335 (compute_access_range): Differentiate cases that equality can be
26336 proven at all loops, the innermost loops or no loops.
26337 (classify_builtin_st, classify_builtin_ldst): Adjust call to above
26338 function. Set PKIND_PARTIAL_MEMSET for partition appropriately.
26339 (finalize_partitions, distribute_loop): Don't fuse partition of
26340 PKIND_PARTIAL_MEMSET kind when distributing 3-level loop nest.
26341 (prepare_perfect_loop_nest): Distribute 3-level loop nest only if
26342 parloop is enabled.
26344 2018-01-23 Martin Liska <mliska@suse.cz>
26346 * predict.def (PRED_INDIR_CALL): Set probability to PROB_EVEN in
26347 order to ignore the predictor.
26348 (PRED_POLYMORPHIC_CALL): Likewise.
26349 (PRED_RECURSIVE_CALL): Likewise.
26351 2018-01-23 Martin Liska <mliska@suse.cz>
26353 * tree-profile.c (tree_profiling): Print function header to
26354 aware reader which function we are working on.
26355 * value-prof.c (gimple_find_values_to_profile): Do not print
26356 not interesting value histograms.
26358 2018-01-23 Martin Liska <mliska@suse.cz>
26360 * profile-count.h (enum profile_quality): Add
26361 profile_uninitialized as the first value. Do not number values
26362 as they are zero based.
26363 (profile_count::verify): Update sanity check.
26364 (profile_probability::verify): Likewise.
26366 2018-01-23 Nathan Sidwell <nathan@acm.org>
26368 * doc/invoke.texi (ffor-scope): Deprecate.
26370 2018-01-23 David Malcolm <dmalcolm@redhat.com>
26372 PR tree-optimization/83510
26373 * domwalk.c (set_all_edges_as_executable): New function.
26374 (dom_walker::dom_walker): Convert bool param
26375 "skip_unreachable_blocks" to enum reachability. Move setup of
26376 edge flags to set_all_edges_as_executable and only do it when
26377 reachability is REACHABLE_BLOCKS.
26378 * domwalk.h (enum dom_walker::reachability): New enum.
26379 (dom_walker::dom_walker): Convert bool param
26380 "skip_unreachable_blocks" to enum reachability.
26381 (set_all_edges_as_executable): New decl.
26382 * graphite-scop-detection.c (gather_bbs::gather_bbs): Convert
26383 from false for "skip_unreachable_blocks" to ALL_BLOCKS for
26385 * tree-ssa-dom.c (dom_opt_dom_walker::dom_opt_dom_walker): Likewise,
26386 but converting true to REACHABLE_BLOCKS.
26387 * tree-ssa-sccvn.c (sccvn_dom_walker::sccvn_dom_walker): Likewise.
26389 (check_array_bounds_dom_walker::check_array_bounds_dom_walker):
26390 Likewise, but converting it to REACHABLE_BLOCKS_PRESERVING_FLAGS.
26391 (vrp_dom_walker::vrp_dom_walker): Likewise, but converting it to
26393 (vrp_prop::vrp_finalize): Call set_all_edges_as_executable
26394 if check_all_array_refs will be called.
26396 2018-01-23 David Malcolm <dmalcolm@redhat.com>
26398 * tree.c (selftest::test_location_wrappers): Add more test
26401 2018-01-23 David Malcolm <dmalcolm@redhat.com>
26403 * sbitmap.c (selftest::test_set_range): Fix memory leaks.
26404 (selftest::test_bit_in_range): Likewise.
26406 2018-01-23 Richard Sandiford <richard.sandiford@linaro.org>
26409 * doc/sourcebuild.texi (vect_float): Say that the selector
26410 only describes the situation when -funsafe-math-optimizations is on.
26411 (vect_float_strict): Document.
26413 2018-01-23 Richard Sandiford <richard.sandiford@linaro.org>
26415 PR tree-optimization/83965
26416 * tree-vect-patterns.c (vect_reassociating_reduction_p): New function.
26417 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern): Use it
26418 instead of checking only for a reduction.
26419 (vect_recog_widen_sum_pattern): Likewise.
26421 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
26423 * predict.c (probably_never_executed): Only use precise profile info.
26424 (compute_function_frequency): Skip after inlining hack since we now
26425 have quality checking.
26427 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
26429 * profile-count.h (profile_probability::very_unlikely,
26430 profile_probability::unlikely, profile_probability::even): Set
26431 precision to guessed.
26433 2018-01-23 Richard Biener <rguenther@suse.de>
26435 PR tree-optimization/83963
26436 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
26437 Properly terminate dominator walk when crossing the exit edge not
26438 when visiting its source block.
26440 2018-01-23 Jakub Jelinek <jakub@redhat.com>
26443 * tree.c (maybe_wrap_with_location): Use NON_LVALUE_EXPR rather than
26444 VIEW_CONVERT_EXPR to wrap CONST_DECLs.
26446 2018-01-22 Jakub Jelinek <jakub@redhat.com>
26448 PR tree-optimization/83957
26449 * omp-expand.c (expand_omp_for_generic): Ignore virtual PHIs. Remove
26450 semicolon after for body surrounded by braces.
26452 PR tree-optimization/83081
26453 * profile-count.h (profile_probability::split): New method.
26454 * dojump.c (do_jump_1) <case TRUTH_ANDIF_EXPR, case TRUTH_ORIF_EXPR>:
26455 Use profile_probability::split.
26456 (do_compare_rtx_and_jump): Fix adjustment of probabilities
26457 when splitting a single conditional jump into 2.
26459 2018-01-22 David Malcolm <dmalcolm@redhat.com>
26461 PR tree-optimization/69452
26462 * tree-ssa-loop-im.c (class move_computations_dom_walker): Remove
26465 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
26467 * config/rl78/rl78-expand.md (bswaphi2): New define_expand.
26468 * config/rl78/rl78-virt.md (*bswaphi2_virt): New define_insn.
26469 * config/rl78/rl78-real.md (*bswaphi2_real): New define_insn.
26471 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
26473 * config/rl78/rl78-protos.h (rl78_split_movdi): New function
26475 * config/rl78/rl78.md (movdi): New define_expand.
26476 * config/rl78/rl78.c (rl78_split_movdi): New function.
26478 2018-01-22 Michael Meissner <meissner@linux.vnet.ibm.com>
26481 * config/rs6000/rs6000-protos.h (rs6000_split_signbit): Delete,
26483 * config/rs6000/rs6000.c (rs6000_split_signbit): Likewise.
26484 * config/rs6000/rs6000.md (signbit<mode>2): Change code for IEEE
26485 128-bit to produce an UNSPEC move to get the double word with the
26486 signbit and then a shift directly to do signbit.
26487 (signbit<mode>2_dm): Replace old IEEE 128-bit signbit
26488 implementation with a new version that just does either a direct
26489 move or a regular move. Move memory interface to separate insns.
26490 Move insns so they are next to the expander.
26491 (signbit<mode>2_dm_mem_be): New combiner insns to combine load
26492 with signbit move. Split big and little endian case.
26493 (signbit<mode>2_dm_mem_le): Likewise.
26494 (signbit<mode>2_dm_<su>ext): Delete, no longer used.
26495 (signbit<mode>2_dm2): Likewise.
26497 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
26499 * config/rl78/rl78.md (anddi3): New define_expand.
26501 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
26503 * config/rl78/rl78.md (umindi3): New define_expand.
26505 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
26507 * config/rl78/rl78.md (smindi3): New define_expand.
26509 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
26511 * config/rl78/rl78.md (smaxdi3): New define_expand.
26513 2018-01-22 Carl Love <cel@us.ibm.com>
26515 * config/rs6000/rs6000-builtin.def (ST_ELEMREV_V1TI, LD_ELEMREV_V1TI,
26516 LVX_V1TI): Add macro expansion.
26517 * config/rs6000/rs6000-c.c (altivec_builtin_types): Add argument
26518 definitions for VSX_BUILTIN_VEC_XST_BE, VSX_BUILTIN_VEC_ST,
26519 VSX_BUILTIN_VEC_XL, LD_ELEMREV_V1TI builtins.
26520 * config/rs6000/rs6000-p8swap.c (insn_is_swappable_p);
26521 Change check to determine if the instruction is a byte reversing
26522 entry. Fix typo in comment.
26523 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case entry
26524 for VSX_BUILTIN_ST_ELEMREV_V1TI and VSX_BUILTIN_LD_ELEMREV_V1TI.
26525 Add def_builtin calls for new builtins.
26526 * config/rs6000/vsx.md (vsx_st_elemrev_v1ti, vsx_ld_elemrev_v1ti):
26527 Add define_insn expansion.
26529 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
26531 * config/rl78/rl78.md (umaxdi3): New define_expand.
26533 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
26535 * config/rl78/rl78.c (rl78_note_reg_set): Fixed dead reg check
26536 for non-QImode registers.
26538 2018-01-22 Richard Biener <rguenther@suse.de>
26540 PR tree-optimization/83963
26541 * graphite-scop-detection.c (scop_detection::get_sese): Delay
26542 including the loop exit block.
26543 (scop_detection::merge_sese): Likewise.
26544 (scop_detection::add_scop): Do it here instead.
26546 2018-01-22 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26548 * doc/sourcebuild.texi (arm_softfloat): Document.
26550 2018-01-21 John David Anglin <danglin@gcc.gnu.org>
26553 * config/pa/pa.c (pa_function_ok_for_sibcall): Use
26554 targetm.binds_local_p instead of TREE_PUBLIC to check local binding.
26555 Move TARGET_PORTABLE_RUNTIME check after TARGET_64BIT check.
26557 2018-01-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26558 David Edelsohn <dje.gcc@gmail.com>
26561 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
26562 Change "crset eq" to "crset 2".
26563 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
26564 (*call_indirect_aix<mode>_nospec): Likewise.
26565 (*call_value_indirect_aix<mode>_nospec): Likewise.
26566 (*call_indirect_elfv2<mode>_nospec): Likewise.
26567 (*call_value_indirect_elfv2<mode>_nospec): Likewise.
26568 (*sibcall_nonlocal_sysv<mode>): Change "crset eq" to "crset 2";
26569 change assembly output from . to $.
26570 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
26571 (indirect_jump<mode>_nospec): Change assembly output from . to $.
26572 (*tablejump<mode>_internal1_nospec): Likewise.
26574 2018-01-21 Oleg Endo <olegendo@gcc.gnu.org>
26577 * config/sh/sh_optimize_sett_clrt.cc:
26578 Use INCLUDE_ALGORITHM and INCLUDE_VECTOR instead of direct includes.
26580 2018-01-20 Richard Sandiford <richard.sandiford@linaro.org>
26582 PR tree-optimization/83940
26583 * tree-vect-stmts.c (vect_truncate_gather_scatter_offset): Set
26584 offset_dt to vect_constant_def rather than vect_unknown_def_type.
26585 (vect_check_load_store_mask): Add a mask_dt_out parameter and
26586 use it to pass back the definition type.
26587 (vect_check_store_rhs): Likewise rhs_dt_out.
26588 (vect_build_gather_load_calls): Add a mask_dt argument and use
26589 it instead of a call to vect_is_simple_use.
26590 (vectorizable_store): Update calls to vect_check_load_store_mask
26591 and vect_check_store_rhs. Use the dt returned by the latter instead
26592 of scatter_src_dt. Use the cached mask_dt and gs_info.offset_dt
26593 instead of calls to vect_is_simple_use. Pass the scalar rather
26594 than the vector operand to vect_is_simple_use when handling
26595 second and subsequent copies of an rhs value.
26596 (vectorizable_load): Update calls to vect_check_load_store_mask
26597 and vect_build_gather_load_calls. Use the cached mask_dt and
26598 gs_info.offset_dt instead of calls to vect_is_simple_use.
26600 2018-01-20 Jakub Jelinek <jakub@redhat.com>
26602 PR middle-end/83945
26603 * tree-emutls.c: Include gimplify.h.
26604 (lower_emutls_2): New function.
26605 (lower_emutls_1): If ADDR_EXPR is a gimple invariant and walk_tree
26606 with lower_emutls_2 callback finds some TLS decl in it, unshare_expr
26607 it before further processing.
26610 * simplify-rtx.c (simplify_binary_operation_1) <case UMOD>: Use
26611 UINTVAL (trueop1) instead of INTVAL (op1).
26613 2018-01-19 Jakub Jelinek <jakub@redhat.com>
26617 * dwarf2cfi.c (DEFAULT_INCOMING_FRAME_SP_OFFSET): Define to
26618 INCOMING_FRAME_SP_OFFSET if not defined.
26619 (scan_trace): Add ENTRY argument. If true and
26620 DEFAULT_INCOMING_FRAME_SP_OFFSET != INCOMING_FRAME_SP_OFFSET,
26621 emit a note to adjust the CFA offset.
26622 (create_cfi_notes): Adjust scan_trace callers.
26623 (create_cie_data): Use DEFAULT_INCOMING_FRAME_SP_OFFSET rather than
26624 INCOMING_FRAME_SP_OFFSET in the CIE.
26625 * config/i386/i386.h (DEFAULT_INCOMING_FRAME_SP_OFFSET): Define.
26626 * config/stormy16/stormy16.h (DEFAULT_INCOMING_FRAME_SP_OFFSET):
26628 * doc/tm.texi.in (DEFAULT_INCOMING_FRAME_SP_OFFSET): Document.
26629 * doc/tm.texi: Regenerated.
26631 2018-01-19 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26633 PR rtl-optimization/83147
26634 * lra-constraints.c (remove_inheritance_pseudos): Use
26635 lra_substitute_pseudo_within_insn.
26637 2018-01-19 Tom de Vries <tom@codesourcery.com>
26638 Cesar Philippidis <cesar@codesourcery.com>
26641 * config/nvptx/nvptx.c (nvptx_single): Fix jit workaround.
26643 2018-01-19 Cesar Philippidis <cesar@codesourcery.com>
26646 * config/nvptx/nvptx.c (output_init_frag): Don't use generic address
26647 spaces for function labels.
26649 2018-01-19 Martin Liska <mliska@suse.cz>
26651 * predict.def (PRED_LOOP_EXIT): Change from 85 to 89.
26652 (PRED_LOOP_EXIT_WITH_RECURSION): Change from 72 to 78.
26653 (PRED_LOOP_EXTRA_EXIT): Change from 83 to 67.
26654 (PRED_OPCODE_POSITIVE): Change from 64 to 59.
26655 (PRED_TREE_OPCODE_POSITIVE): Change from 64 to 59.
26656 (PRED_CONST_RETURN): Change from 69 to 65.
26657 (PRED_NULL_RETURN): Change from 91 to 71.
26658 (PRED_LOOP_IV_COMPARE_GUESS): Change from 98 to 64.
26659 (PRED_LOOP_GUARD): Change from 66 to 73.
26661 2018-01-19 Martin Liska <mliska@suse.cz>
26663 * predict.c (predict_insn_def): Add new assert.
26664 (struct branch_predictor): Change type to signed integer.
26665 (test_prediction_value_range): Amend test to cover
26666 PROB_UNINITIALIZED.
26667 * predict.def (PRED_LOOP_ITERATIONS): Use the new constant.
26668 (PRED_LOOP_ITERATIONS_GUESSED): Likewise.
26669 (PRED_LOOP_ITERATIONS_MAX): Likewise.
26670 (PRED_LOOP_IV_COMPARE): Likewise.
26671 * predict.h (PROB_UNINITIALIZED): Define new constant.
26673 2018-01-19 Martin Liska <mliska@suse.cz>
26675 * predict.c (dump_prediction): Add new format for
26676 analyze_brprob.py script which is enabled with -details
26678 * profile-count.h (precise_p): New function.
26680 2018-01-19 Richard Sandiford <richard.sandiford@linaro.org>
26682 PR tree-optimization/83922
26683 * tree-vect-loop.c (vect_verify_full_masking): Return false if
26684 there are no statements that need masking.
26685 (vect_active_double_reduction_p): New function.
26686 (vect_analyze_loop_operations): Use it when handling phis that
26687 are not in the loop header.
26689 2018-01-19 Richard Sandiford <richard.sandiford@linaro.org>
26691 PR tree-optimization/83914
26692 * tree-vect-loop.c (vectorizable_induction): Don't convert
26693 init_expr or apply the peeling adjustment for inductions
26694 that are nested within the vectorized loop.
26696 2018-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26698 * config/arm/thumb2.md (*thumb2_negsi2_short): Use RSB mnemonic
26701 2018-01-18 Jakub Jelinek <jakub@redhat.com>
26705 * function.h (gimplify_parameters): Add gimple_seq * argument.
26706 * function.c: Include gimple.h and options.h.
26707 (gimplify_parameters): Add cleanup argument, add CLOBBER stmts
26708 for the added local temporaries if needed.
26709 * gimplify.c (gimplify_body): Adjust gimplify_parameters caller,
26710 if there are any parameter cleanups, wrap whole body into a
26711 try/finally with the cleanups.
26713 2018-01-18 Wilco Dijkstra <wdijkstr@arm.com>
26716 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
26717 Use GET_MODE_CLASS for scalar floating point.
26719 2018-01-18 Jan Hubicka <hubicka@ucw.cz>
26723 * cgraphclones.c (cgraph_node::create_version_clone_with_body):
26724 Fix call of call_cgraph_insertion_hooks.
26726 2018-01-18 Martin Sebor <msebor@redhat.com>
26728 * doc/invoke.texi (-Wclass-memaccess): Tweak text.
26730 2018-01-18 Jan Hubicka <hubicka@ucw.cz>
26733 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Update edge
26736 2018-01-18 Boris Kolpackov <boris@codesynthesis.com>
26739 * common.opt: (-ffile-prefix-map): New option.
26740 * opts.c (common_handle_option): Defer it.
26741 * opts-global.c (handle_common_deferred_options): Handle it.
26742 * debug.h (remap_debug_filename, add_debug_prefix_map): Move to...
26743 * file-prefix-map.h: New file.
26744 (remap_debug_filename, add_debug_prefix_map): ...here.
26745 (add_macro_prefix_map, add_file_prefix_map, remap_macro_filename): New.
26746 * final.c (debug_prefix_map, add_debug_prefix_map
26747 remap_debug_filename): Move to...
26748 * file-prefix-map.c: New file.
26749 (file_prefix_map, add_prefix_map, remap_filename) ...here and rename,
26750 generalize, get rid of alloca(), use strrchr() instead of strchr().
26751 (add_macro_prefix_map, add_debug_prefix_map, add_file_prefix_map):
26752 Implement in terms of add_prefix_map().
26753 (remap_macro_filename, remap_debug_filename): Implement in term of
26755 * Makefile.in (OBJS, PLUGIN_HEADERS): Add new files.
26756 * builtins.c (fold_builtin_FILE): Call remap_macro_filename().
26757 * dbxout.c: Include file-prefix-map.h.
26758 * varasm.c: Likewise.
26759 * vmsdbgout.c: Likewise.
26760 * xcoffout.c: Likewise.
26761 * dwarf2out.c: Likewise plus omit new options from DW_AT_producer.
26762 * doc/cppopts.texi (-fmacro-prefix-map): Document.
26763 * doc/invoke.texi (-ffile-prefix-map): Document.
26764 (-fdebug-prefix-map): Update description.
26766 2018-01-18 Martin Liska <mliska@suse.cz>
26768 * config/i386/i386.c (indirect_thunk_name): Document that also
26770 (output_indirect_thunk): Document why both instructions
26771 (pause and lfence) are generated.
26773 2018-01-18 Richard Biener <rguenther@suse.de>
26775 PR tree-optimization/83887
26776 * graphite-scop-detection.c
26777 (scop_detection::get_nearest_dom_with_single_entry): Remove.
26778 (scop_detection::get_nearest_pdom_with_single_exit): Likewise.
26779 (scop_detection::merge_sese): Re-implement with a flood-fill
26780 algorithm that properly finds a SESE region if it exists.
26782 2018-01-18 Jakub Jelinek <jakub@redhat.com>
26785 * match.pd ((P + A) - P, P - (P + A), (P + A) - (P + B)): For
26786 pointer_diff optimizations use view_convert instead of convert.
26788 2018-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26790 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
26791 Generate different code for -mno-speculate-indirect-jumps.
26792 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
26793 (*call_indirect_aix<mode>): Disable for
26794 -mno-speculate-indirect-jumps.
26795 (*call_indirect_aix<mode>_nospec): New define_insn.
26796 (*call_value_indirect_aix<mode>): Disable for
26797 -mno-speculate-indirect-jumps.
26798 (*call_value_indirect_aix<mode>_nospec): New define_insn.
26799 (*sibcall_nonlocal_sysv<mode>): Generate different code for
26800 -mno-speculate-indirect-jumps.
26801 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
26803 2018-01-17 Michael Meissner <meissner@linux.vnet.ibm.com>
26805 * config/rs6000/rs6000.c (rs6000_emit_move): If we load or store a
26806 long double type, set the flags for noting the default long double
26807 type, even if we don't pass or return a long double type.
26809 2018-01-17 Jan Hubicka <hubicka@ucw.cz>
26812 * ipa-inline.c (flatten_function): Do not overwrite final inlining
26815 2018-01-17 Will Schmidt <will_schmidt@vnet.ibm.com>
26817 * config/rs6000/rs6000.c (rs6000_gimple_builtin): Add gimple folding
26818 support for merge[hl].
26819 (fold_mergehl_helper): New helper function.
26820 (tree-vector-builder.h): New #include for tree_vector_builder usage.
26821 * config/rs6000/altivec.md (altivec_vmrghw_direct): Add xxmrghw insn.
26822 (altivec_vmrglw_direct): Add xxmrglw insn.
26824 2018-01-17 Andrew Waterman <andrew@sifive.com>
26826 * config/riscv/riscv.c (riscv_conditional_register_usage): If
26827 UNITS_PER_FP_ARG is 0, set call_used_regs to 1 for all FP regs.
26829 2018-01-17 David Malcolm <dmalcolm@redhat.com>
26832 * ipa-devirt.c (add_type_duplicate): When comparing memory layout,
26833 call the lto_location_cache before reading the
26834 DECL_SOURCE_LOCATION of the types.
26836 2018-01-17 Wilco Dijkstra <wdijkstr@arm.com>
26837 Richard Sandiford <richard.sandiford@linaro.org>
26839 * config/aarch64/aarch64.md (movti_aarch64): Use Uti constraint.
26840 * config/aarch64/aarch64.c (aarch64_mov128_immediate): New function.
26841 (aarch64_legitimate_constant_p): Just support CONST_DOUBLE
26842 SF/DF/TF mode to avoid creating illegal CONST_WIDE_INT immediates.
26843 * config/aarch64/aarch64-protos.h (aarch64_mov128_immediate):
26845 * config/aarch64/constraints.md (aarch64_movti_operand):
26847 * config/aarch64/predicates.md (Uti): Add new constraint.
26849 2018-01-17 Carl Love <cel@us.ibm.com>
26851 * config/rs6000/vsx.md (define_expand xl_len_r,
26852 define_expand stxvl, define_expand *stxvl): Add match_dup argument.
26853 (define_insn): Add, match_dup 1 argument to define_insn stxvll and
26855 (define_expand, define_insn): Move the shift left from the
26856 define_insn to the define_expand for lxvl and stxvl instructions.
26857 * config/rs6000/rs6000-builtin.def (BU_P9V_64BIT_VSX_2): Change LXVL
26858 and XL_LEN_R definitions to PURE.
26860 2018-01-17 Uros Bizjak <ubizjak@gmail.com>
26862 * config/i386/i386.c (indirect_thunk_name): Declare regno
26863 as unsigned int. Compare regno with INVALID_REGNUM.
26864 (output_indirect_thunk): Ditto.
26865 (output_indirect_thunk_function): Ditto.
26866 (ix86_code_end): Declare regno as unsigned int. Use INVALID_REGNUM
26867 in the call to output_indirect_thunk_function.
26869 2018-01-17 Richard Sandiford <richard.sandiford@linaro.org>
26871 PR middle-end/83884
26872 * expr.c (expand_expr_real_1): Use the size of GET_MODE (op0)
26873 rather than the size of inner_type to determine the stack slot size
26874 when handling VIEW_CONVERT_EXPRs on strict-alignment targets.
26876 2018-01-16 Sebastian Peryt <sebastian.peryt@intel.com>
26879 * config/i386/i386.c (ix86_option_override_internal): Add PTA_RDRND
26882 2018-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
26884 * config.gcc (powerpc*-linux*-*): Add support for 64-bit little
26885 endian Linux systems to optionally enable multilibs for selecting
26886 the long double type if the user configured an explicit type.
26887 * config/rs6000/rs6000.h (TARGET_IEEEQUAD_MULTILIB): Indicate we
26888 have no long double multilibs if not defined.
26889 * config/rs6000/rs6000.c (rs6000_option_override_internal): Do not
26890 warn if the user used -mabi={ieee,ibm}longdouble and we built
26891 multilibs for long double.
26892 * config/rs6000/linux64.h (MULTILIB_DEFAULTS_IEEE): Define as the
26893 appropriate multilib option.
26894 (MULTILIB_DEFAULTS): Add MULTILIB_DEFAULTS_IEEE to the default
26896 * config/rs6000/t-ldouble-linux64le-ibm: New configuration files
26897 for building long double multilibs.
26898 * config/rs6000/t-ldouble-linux64le-ieee: Likewise.
26900 2018-01-16 John David Anglin <danglin@gcc.gnu.org>
26902 * config.gcc (hppa*-*-linux*): Change callee copies ABI to caller
26905 * config/pa.h (MALLOC_ABI_ALIGNMENT): Set 32-bit alignment default to
26907 * config/pa/pa32-linux.h (MALLOC_ABI_ALIGNMENT): Set alignment to
26910 * config/pa/som.h (ASM_DECLARE_FUNCTION_NAME): Cleanup type and mode
26913 * config/pa/pa.c (pa_function_arg_size): Apply CEIL to GET_MODE_SIZE
26916 2018-01-16 Eric Botcazou <ebotcazou@adacore.com>
26918 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): For an
26919 ADDR_EXPR, do not count the offset of a COMPONENT_REF twice.
26921 2018-01-16 Kelvin Nilsen <kelvin@gcc.gnu.org>
26923 * config/rs6000/rs6000-p8swap.c (rs6000_gen_stvx): Generate
26924 different rtl trees depending on TARGET_64BIT.
26925 (rs6000_gen_lvx): Likewise.
26927 2018-01-16 Eric Botcazou <ebotcazou@adacore.com>
26929 * config/visium/visium.md (nop): Tweak comment.
26930 (hazard_nop): Likewise.
26932 2018-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26934 * config/rs6000/rs6000.c (rs6000_opt_vars): Add entry for
26935 -mspeculate-indirect-jumps.
26936 * config/rs6000/rs6000.md (*call_indirect_elfv2<mode>): Disable
26937 for -mno-speculate-indirect-jumps.
26938 (*call_indirect_elfv2<mode>_nospec): New define_insn.
26939 (*call_value_indirect_elfv2<mode>): Disable for
26940 -mno-speculate-indirect-jumps.
26941 (*call_value_indirect_elfv2<mode>_nospec): New define_insn.
26942 (indirect_jump): Emit different RTL for
26943 -mno-speculate-indirect-jumps.
26944 (*indirect_jump<mode>): Disable for
26945 -mno-speculate-indirect-jumps.
26946 (*indirect_jump<mode>_nospec): New define_insn.
26947 (tablejump): Emit different RTL for
26948 -mno-speculate-indirect-jumps.
26949 (tablejumpsi): Disable for -mno-speculate-indirect-jumps.
26950 (tablejumpsi_nospec): New define_expand.
26951 (tablejumpdi): Disable for -mno-speculate-indirect-jumps.
26952 (tablejumpdi_nospec): New define_expand.
26953 (*tablejump<mode>_internal1): Disable for
26954 -mno-speculate-indirect-jumps.
26955 (*tablejump<mode>_internal1_nospec): New define_insn.
26956 * config/rs6000/rs6000.opt (mspeculate-indirect-jumps): New
26959 2018-01-16 Artyom Skrobov tyomitch@gmail.com
26961 * caller-save.c (insert_save): Drop unnecessary parameter. All
26964 2018-01-16 Jakub Jelinek <jakub@redhat.com>
26965 Richard Biener <rguenth@suse.de>
26968 * gimplify.c (gimplify_one_sizepos): For is_gimple_constant (expr)
26969 return early, inline manually is_gimple_sizepos. Make sure if we
26970 call gimplify_expr we don't end up with a gimple constant.
26971 * tree.c (variably_modified_type_p): Don't return true for
26972 is_gimple_constant (_t). Inline manually is_gimple_sizepos.
26973 * gimplify.h (is_gimple_sizepos): Remove.
26975 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
26977 PR tree-optimization/83857
26978 * tree-vect-loop.c (vect_analyze_loop_operations): Don't call
26979 vectorizable_live_operation for pure SLP statements.
26980 (vectorizable_live_operation): Handle PHIs.
26982 2018-01-16 Richard Biener <rguenther@suse.de>
26984 PR tree-optimization/83867
26985 * tree-vect-stmts.c (vect_transform_stmt): Precompute
26986 nested_in_vect_loop_p since the scalar stmt may get invalidated.
26988 2018-01-16 Jakub Jelinek <jakub@redhat.com>
26991 * stor-layout.c (handle_warn_if_not_align): Use byte_position and
26992 multiple_of_p instead of unchecked tree_to_uhwi and UHWI check.
26993 If off is not INTEGER_CST, issue a may not be aligned warning
26994 rather than isn't aligned. Use isn%'t rather than isn't.
26995 * fold-const.c (multiple_of_p) <case BIT_AND_EXPR>: Don't fall through
26997 <case MULT_EXPR>: Improve the case when bottom and one of the
26998 MULT_EXPR operands are INTEGER_CSTs and bottom is multiple of that
26999 operand, in that case check if the other operand is multiple of
27000 bottom divided by the INTEGER_CST operand.
27002 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
27005 * config/pa/pa.h (FUNCTION_ARG_SIZE): Delete.
27006 * config/pa/pa-protos.h (pa_function_arg_size): Declare.
27007 * config/pa/som.h (ASM_DECLARE_FUNCTION_NAME): Use
27008 pa_function_arg_size instead of FUNCTION_ARG_SIZE.
27009 * config/pa/pa.c (pa_function_arg_advance): Likewise.
27010 (pa_function_arg, pa_arg_partial_bytes): Likewise.
27011 (pa_function_arg_size): New function.
27013 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
27015 * fold-const.c (fold_ternary_loc): Construct the vec_perm_indices
27016 in a separate statement.
27018 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
27020 PR tree-optimization/83847
27021 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Don't
27022 group gathers and scatters.
27024 2018-01-16 Jakub Jelinek <jakub@redhat.com>
27026 PR rtl-optimization/86620
27027 * params.def (max-sched-ready-insns): Bump minimum value to 1.
27029 PR rtl-optimization/83213
27030 * recog.c (peep2_attempt): Copy over CROSSING_JUMP_P from peepinsn
27031 to last if both are JUMP_INSNs.
27033 PR tree-optimization/83843
27034 * gimple-ssa-store-merging.c
27035 (imm_store_chain_info::output_merged_store): Handle bit_not_p on
27036 store_immediate_info for bswap/nop orig_stores.
27038 2018-01-15 Andrew Waterman <andrew@sifive.com>
27040 * config/riscv/riscv.c (riscv_rtx_costs) <MULT>: Increase cost if
27042 <UDIV>: Increase cost if !TARGET_DIV.
27044 2018-01-15 Segher Boessenkool <segher@kernel.crashing.org>
27046 * config/rs6000/rs6000.md (define_attr "type"): Remove delayed_cr.
27047 (define_attr "cr_logical_3op"): New.
27048 (cceq_ior_compare): Adjust.
27049 (cceq_ior_compare_complement): Adjust.
27050 (*cceq_rev_compare): Adjust.
27051 * config/rs6000/rs6000.c (rs6000_adjust_cost): Adjust.
27052 (is_cracked_insn): Adjust.
27053 (insn_must_be_first_in_group): Adjust.
27054 * config/rs6000/40x.md: Adjust.
27055 * config/rs6000/440.md: Adjust.
27056 * config/rs6000/476.md: Adjust.
27057 * config/rs6000/601.md: Adjust.
27058 * config/rs6000/603.md: Adjust.
27059 * config/rs6000/6xx.md: Adjust.
27060 * config/rs6000/7450.md: Adjust.
27061 * config/rs6000/7xx.md: Adjust.
27062 * config/rs6000/8540.md: Adjust.
27063 * config/rs6000/cell.md: Adjust.
27064 * config/rs6000/e300c2c3.md: Adjust.
27065 * config/rs6000/e500mc.md: Adjust.
27066 * config/rs6000/e500mc64.md: Adjust.
27067 * config/rs6000/e5500.md: Adjust.
27068 * config/rs6000/e6500.md: Adjust.
27069 * config/rs6000/mpc.md: Adjust.
27070 * config/rs6000/power4.md: Adjust.
27071 * config/rs6000/power5.md: Adjust.
27072 * config/rs6000/power6.md: Adjust.
27073 * config/rs6000/power7.md: Adjust.
27074 * config/rs6000/power8.md: Adjust.
27075 * config/rs6000/power9.md: Adjust.
27076 * config/rs6000/rs64.md: Adjust.
27077 * config/rs6000/titan.md: Adjust.
27079 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
27081 * config/i386/predicates.md (indirect_branch_operand): Rewrite
27082 ix86_indirect_branch_register logic.
27084 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
27086 * config/i386/constraints.md (Bs): Update
27087 ix86_indirect_branch_register check. Don't check
27088 ix86_indirect_branch_register with GOT_memory_operand.
27090 * config/i386/predicates.md (GOT_memory_operand): Don't check
27091 ix86_indirect_branch_register here.
27092 (GOT32_symbol_operand): Likewise.
27094 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
27096 * config/i386/predicates.md (constant_call_address_operand):
27097 Rewrite ix86_indirect_branch_register logic.
27098 (sibcall_insn_operand): Likewise.
27100 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
27102 * config/i386/constraints.md (Bs): Replace
27103 ix86_indirect_branch_thunk_register with
27104 ix86_indirect_branch_register.
27106 * config/i386/i386.md (indirect_jump): Likewise.
27107 (tablejump): Likewise.
27108 (*sibcall_memory): Likewise.
27109 (*sibcall_value_memory): Likewise.
27110 Peepholes of indirect call and jump via memory: Likewise.
27111 * config/i386/i386.opt: Likewise.
27112 * config/i386/predicates.md (indirect_branch_operand): Likewise.
27113 (GOT_memory_operand): Likewise.
27114 (call_insn_operand): Likewise.
27115 (sibcall_insn_operand): Likewise.
27116 (GOT32_symbol_operand): Likewise.
27118 2018-01-15 Jakub Jelinek <jakub@redhat.com>
27120 PR middle-end/83837
27121 * omp-expand.c (expand_omp_atomic_pipeline): Use loaded_val
27122 type rather than type addr's type points to.
27123 (expand_omp_atomic_mutex): Likewise.
27124 (expand_omp_atomic): Likewise.
27126 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
27129 * config/i386/i386.c (output_indirect_thunk_function): Use
27130 ASM_OUTPUT_LABEL, instead of ASM_OUTPUT_DEF, for TARGET_MACHO
27131 for __x86_return_thunk.
27133 2018-01-15 Richard Biener <rguenther@suse.de>
27135 PR middle-end/83850
27136 * expmed.c (extract_bit_field_1): Fix typo.
27138 2018-01-15 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27141 * config/arm/iterators.md (VF): New mode iterator.
27142 * config/arm/neon.md (neon_vabd<mode>_2): Use the above.
27143 Remove integer-related logic from pattern.
27144 (neon_vabd<mode>_3): Likewise.
27146 2018-01-15 Jakub Jelinek <jakub@redhat.com>
27148 PR middle-end/82694
27149 * common.opt (fstrict-overflow): No longer an alias.
27150 (fwrapv-pointer): New option.
27151 * tree.h (TYPE_OVERFLOW_WRAPS, TYPE_OVERFLOW_UNDEFINED): Define
27152 also for pointer types based on flag_wrapv_pointer.
27153 * opts.c (common_handle_option) <case OPT_fstrict_overflow>: Set
27154 opts->x_flag_wrap[pv] to !value, clear opts->x_flag_trapv if
27155 opts->x_flag_wrapv got set.
27156 * fold-const.c (fold_comparison, fold_binary_loc): Revert 2017-08-01
27157 changes, just use TYPE_OVERFLOW_UNDEFINED on pointer type instead of
27158 POINTER_TYPE_OVERFLOW_UNDEFINED.
27159 * match.pd: Likewise in address comparison pattern.
27160 * doc/invoke.texi: Document -fwrapv and -fstrict-overflow.
27162 2018-01-15 Richard Biener <rguenther@suse.de>
27165 * tree.c (free_lang_data_in_type): Always unlink TYPE_DECLs
27166 from TYPE_FIELDS. Free TYPE_BINFO if not used by devirtualization.
27167 Reset type names to their identifier if their TYPE_DECL doesn't
27168 have linkage (and thus is used for ODR and devirt).
27169 (save_debug_info_for_decl): Remove.
27170 (save_debug_info_for_type): Likewise.
27171 (add_tree_to_fld_list): Adjust.
27172 * tree-pretty-print.c (dump_generic_node): Make dumping of
27173 type names more robust.
27175 2018-01-15 Richard Biener <rguenther@suse.de>
27177 * BASE-VER: Bump to 8.0.1.
27179 2018-01-14 Martin Sebor <msebor@redhat.com>
27182 * builtins.c (check_access): Avoid warning when the no-warning bit
27185 2018-01-14 Cory Fields <cory-nospam-@coryfields.com>
27187 * tree-ssa-loop-im.c (sort_bbs_in_loop_postorder_cmp): Stabilize sort.
27188 * ira-color (allocno_hard_regs_compare): Likewise.
27190 2018-01-14 Nathan Rossi <nathan@nathanrossi.com>
27193 * config/microblaze/microblaze.c (microblaze_asm_output_ident):
27194 Use .pushsection/.popsection.
27196 2018-01-14 Martin Sebor <msebor@redhat.com>
27199 * doc/invoke.texi (-Wlass-memaccess): Document suppression by casting.
27201 2018-01-14 Jakub Jelinek <jakub@redhat.com>
27203 * config.gcc (i[34567]86-*-*): Remove one duplicate gfniintrin.h
27204 entry from extra_headers.
27205 (x86_64-*-*): Remove two duplicate gfniintrin.h entries from
27206 extra_headers, make the list bitwise identical to the i?86-*-* one.
27208 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
27210 * config/i386/i386.c (ix86_set_indirect_branch_type): Disallow
27211 -mcmodel=large with -mindirect-branch=thunk,
27212 -mindirect-branch=thunk-extern, -mfunction-return=thunk and
27213 -mfunction-return=thunk-extern.
27214 * doc/invoke.texi: Document -mcmodel=large is incompatible with
27215 -mindirect-branch=thunk, -mindirect-branch=thunk-extern,
27216 -mfunction-return=thunk and -mfunction-return=thunk-extern.
27218 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
27220 * config/i386/i386.c (print_reg): Print the name of the full
27221 integer register without '%'.
27222 (ix86_print_operand): Handle 'V'.
27223 * doc/extend.texi: Document 'V' modifier.
27225 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
27227 * config/i386/constraints.md (Bs): Disallow memory operand for
27228 -mindirect-branch-register.
27230 * config/i386/predicates.md (indirect_branch_operand): Likewise.
27231 (GOT_memory_operand): Likewise.
27232 (call_insn_operand): Likewise.
27233 (sibcall_insn_operand): Likewise.
27234 (GOT32_symbol_operand): Likewise.
27235 * config/i386/i386.md (indirect_jump): Call convert_memory_address
27236 for -mindirect-branch-register.
27237 (tablejump): Likewise.
27238 (*sibcall_memory): Likewise.
27239 (*sibcall_value_memory): Likewise.
27240 Disallow peepholes of indirect call and jump via memory for
27241 -mindirect-branch-register.
27242 (*call_pop): Replace m with Bw.
27243 (*call_value_pop): Likewise.
27244 (*sibcall_pop_memory): Replace m with Bs.
27245 * config/i386/i386.opt (mindirect-branch-register): New option.
27246 * doc/invoke.texi: Document -mindirect-branch-register option.
27248 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
27250 * config/i386/i386-protos.h (ix86_output_function_return): New.
27251 * config/i386/i386.c (ix86_set_indirect_branch_type): Also
27252 set function_return_type.
27253 (indirect_thunk_name): Add ret_p to indicate thunk for function
27255 (output_indirect_thunk_function): Pass false to
27256 indirect_thunk_name.
27257 (ix86_output_indirect_branch_via_reg): Likewise.
27258 (ix86_output_indirect_branch_via_push): Likewise.
27259 (output_indirect_thunk_function): Create alias for function
27260 return thunk if regno < 0.
27261 (ix86_output_function_return): New function.
27262 (ix86_handle_fndecl_attribute): Handle function_return.
27263 (ix86_attribute_table): Add function_return.
27264 * config/i386/i386.h (machine_function): Add
27265 function_return_type.
27266 * config/i386/i386.md (simple_return_internal): Use
27267 ix86_output_function_return.
27268 (simple_return_internal_long): Likewise.
27269 * config/i386/i386.opt (mfunction-return=): New option.
27270 (indirect_branch): Mention -mfunction-return=.
27271 * doc/extend.texi: Document function_return function attribute.
27272 * doc/invoke.texi: Document -mfunction-return= option.
27274 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
27276 * config/i386/i386-opts.h (indirect_branch): New.
27277 * config/i386/i386-protos.h (ix86_output_indirect_jmp): Likewise.
27278 * config/i386/i386.c (ix86_using_red_zone): Disallow red-zone
27279 with local indirect jump when converting indirect call and jump.
27280 (ix86_set_indirect_branch_type): New.
27281 (ix86_set_current_function): Call ix86_set_indirect_branch_type.
27282 (indirectlabelno): New.
27283 (indirect_thunk_needed): Likewise.
27284 (indirect_thunk_bnd_needed): Likewise.
27285 (indirect_thunks_used): Likewise.
27286 (indirect_thunks_bnd_used): Likewise.
27287 (INDIRECT_LABEL): Likewise.
27288 (indirect_thunk_name): Likewise.
27289 (output_indirect_thunk): Likewise.
27290 (output_indirect_thunk_function): Likewise.
27291 (ix86_output_indirect_branch_via_reg): Likewise.
27292 (ix86_output_indirect_branch_via_push): Likewise.
27293 (ix86_output_indirect_branch): Likewise.
27294 (ix86_output_indirect_jmp): Likewise.
27295 (ix86_code_end): Call output_indirect_thunk_function if needed.
27296 (ix86_output_call_insn): Call ix86_output_indirect_branch if
27298 (ix86_handle_fndecl_attribute): Handle indirect_branch.
27299 (ix86_attribute_table): Add indirect_branch.
27300 * config/i386/i386.h (machine_function): Add indirect_branch_type
27301 and has_local_indirect_jump.
27302 * config/i386/i386.md (indirect_jump): Set has_local_indirect_jump
27304 (tablejump): Likewise.
27305 (*indirect_jump): Use ix86_output_indirect_jmp.
27306 (*tablejump_1): Likewise.
27307 (simple_return_indirect_internal): Likewise.
27308 * config/i386/i386.opt (mindirect-branch=): New option.
27309 (indirect_branch): New.
27312 (thunk-inline): Likewise.
27313 (thunk-extern): Likewise.
27314 * doc/extend.texi: Document indirect_branch function attribute.
27315 * doc/invoke.texi: Document -mindirect-branch= option.
27317 2018-01-14 Jan Hubicka <hubicka@ucw.cz>
27320 * ipa-inline.c (edge_badness): Tolerate roundoff errors.
27322 2018-01-14 Richard Sandiford <richard.sandiford@linaro.org>
27324 * ipa-inline.c (want_inline_small_function_p): Return false if
27325 inlining has already failed with CIF_FINAL_ERROR.
27326 (update_caller_keys): Call want_inline_small_function_p before
27328 (update_callee_keys): Likewise.
27330 2018-01-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
27332 * config/rs6000/rs6000-p8swap.c (rs6000_sum_of_two_registers_p):
27334 (rs6000_quadword_masked_address_p): Likewise.
27335 (quad_aligned_load_p): Likewise.
27336 (quad_aligned_store_p): Likewise.
27337 (const_load_sequence_p): Add comment to describe the outer-most loop.
27338 (mimic_memory_attributes_and_flags): New function.
27339 (rs6000_gen_stvx): Likewise.
27340 (replace_swapped_aligned_store): Likewise.
27341 (rs6000_gen_lvx): Likewise.
27342 (replace_swapped_aligned_load): Likewise.
27343 (replace_swapped_load_constant): Capitalize argument name in
27344 comment describing this function.
27345 (rs6000_analyze_swaps): Add a third pass to search for vector loads
27346 and stores that access quad-word aligned addresses and replace
27347 with stvx or lvx instructions when appropriate.
27348 * config/rs6000/rs6000-protos.h (rs6000_sum_of_two_registers_p):
27349 New function prototype.
27350 (rs6000_quadword_masked_address_p): Likewise.
27351 (rs6000_gen_lvx): Likewise.
27352 (rs6000_gen_stvx): Likewise.
27353 * config/rs6000/vsx.md (*vsx_le_perm_load_<mode>): For modes
27354 VSX_D (V2DF, V2DI), modify this split to select lvx instruction
27355 when memory address is aligned.
27356 (*vsx_le_perm_load_<mode>): For modes VSX_W (V4SF, V4SI), modify
27357 this split to select lvx instruction when memory address is aligned.
27358 (*vsx_le_perm_load_v8hi): Modify this split to select lvx
27359 instruction when memory address is aligned.
27360 (*vsx_le_perm_load_v16qi): Likewise.
27361 (four unnamed splitters): Modify to select the stvx instruction
27362 when memory is aligned.
27364 2018-01-13 Jan Hubicka <hubicka@ucw.cz>
27366 * predict.c (determine_unlikely_bbs): Handle correctly BBs
27367 which appears in the queue multiple times.
27369 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
27370 Alan Hayward <alan.hayward@arm.com>
27371 David Sherwood <david.sherwood@arm.com>
27373 * tree-vectorizer.h (vec_lower_bound): New structure.
27374 (_loop_vec_info): Add check_nonzero and lower_bounds.
27375 (LOOP_VINFO_CHECK_NONZERO): New macro.
27376 (LOOP_VINFO_LOWER_BOUNDS): Likewise.
27377 (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Check lower_bounds too.
27378 * tree-data-ref.h (dr_with_seg_len): Add access_size and align
27379 fields. Make seg_len the distance travelled, not including the
27381 (dr_direction_indicator): Declare.
27382 (dr_zero_step_indicator): Likewise.
27383 (dr_known_forward_stride_p): Likewise.
27384 * tree-data-ref.c: Include stringpool.h, tree-vrp.h and
27386 (runtime_alias_check_p): Allow runtime alias checks with
27388 (operator ==): Compare access_size and align.
27389 (prune_runtime_alias_test_list): Rework for new distinction between
27390 the access_size and seg_len.
27391 (create_intersect_range_checks_index): Likewise. Cope with polynomial
27393 (get_segment_min_max): New function.
27394 (create_intersect_range_checks): Use it.
27395 (dr_step_indicator): New function.
27396 (dr_direction_indicator): Likewise.
27397 (dr_zero_step_indicator): Likewise.
27398 (dr_known_forward_stride_p): Likewise.
27399 * tree-loop-distribution.c (data_ref_segment_size): Return
27400 DR_STEP * (niters - 1).
27401 (compute_alias_check_pairs): Update call to the dr_with_seg_len
27403 * tree-vect-data-refs.c (vect_check_nonzero_value): New function.
27404 (vect_preserves_scalar_order_p): New function, split out from...
27405 (vect_analyze_data_ref_dependence): ...here. Check for zero steps.
27406 (vect_vfa_segment_size): Return DR_STEP * (length_factor - 1).
27407 (vect_vfa_access_size): New function.
27408 (vect_vfa_align): Likewise.
27409 (vect_compile_time_alias): Take access_size_a and access_b arguments.
27410 (dump_lower_bound): New function.
27411 (vect_check_lower_bound): Likewise.
27412 (vect_small_gap_p): Likewise.
27413 (vectorizable_with_step_bound_p): Likewise.
27414 (vect_prune_runtime_alias_test_list): Ignore cross-iteration
27415 depencies if the vectorization factor is 1. Convert the checks
27416 for nonzero steps into checks on the bounds of DR_STEP. Try using
27417 a bunds check for variable steps if the minimum required step is
27418 relatively small. Update calls to the dr_with_seg_len
27419 constructor and to vect_compile_time_alias.
27420 * tree-vect-loop-manip.c (vect_create_cond_for_lower_bounds): New
27422 (vect_loop_versioning): Call it.
27423 * tree-vect-loop.c (vect_analyze_loop_2): Clear LOOP_VINFO_LOWER_BOUNDS
27425 (vect_estimate_min_profitable_iters): Account for any bounds checks.
27427 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
27428 Alan Hayward <alan.hayward@arm.com>
27429 David Sherwood <david.sherwood@arm.com>
27431 * doc/sourcebuild.texi (vect_scatter_store): Document.
27432 * optabs.def (scatter_store_optab, mask_scatter_store_optab): New
27434 * doc/md.texi (scatter_store@var{m}, mask_scatter_store@var{m}):
27436 * genopinit.c (main): Add supports_vec_scatter_store and
27437 supports_vec_scatter_store_cached to target_optabs.
27438 * gimple.h (gimple_expr_type): Handle IFN_SCATTER_STORE and
27439 IFN_MASK_SCATTER_STORE.
27440 * internal-fn.def (SCATTER_STORE, MASK_SCATTER_STORE): New internal
27442 * internal-fn.h (internal_store_fn_p): Declare.
27443 (internal_fn_stored_value_index): Likewise.
27444 * internal-fn.c (scatter_store_direct): New macro.
27445 (expand_scatter_store_optab_fn): New function.
27446 (direct_scatter_store_optab_supported_p): New macro.
27447 (internal_store_fn_p): New function.
27448 (internal_gather_scatter_fn_p): Handle IFN_SCATTER_STORE and
27449 IFN_MASK_SCATTER_STORE.
27450 (internal_fn_mask_index): Likewise.
27451 (internal_fn_stored_value_index): New function.
27452 (internal_gather_scatter_fn_supported_p): Adjust operand numbers
27453 for scatter stores.
27454 * optabs-query.h (supports_vec_scatter_store_p): Declare.
27455 * optabs-query.c (supports_vec_scatter_store_p): New function.
27456 * tree-vectorizer.h (vect_get_store_rhs): Declare.
27457 * tree-vect-data-refs.c (vect_analyze_data_ref_access): Return
27458 true for scatter stores.
27459 (vect_gather_scatter_fn_p): Handle scatter stores too.
27460 (vect_check_gather_scatter): Consider using scatter stores if
27461 supports_vec_scatter_store_p.
27462 * tree-vect-patterns.c (vect_try_gather_scatter_pattern): Handle
27463 scatter stores too.
27464 * tree-vect-stmts.c (exist_non_indexing_operands_for_use_p): Use
27465 internal_fn_stored_value_index.
27466 (check_load_store_masking): Handle scatter stores too.
27467 (vect_get_store_rhs): Make public.
27468 (vectorizable_call): Use internal_store_fn_p.
27469 (vectorizable_store): Handle scatter store internal functions.
27470 (vect_transform_stmt): Compare GROUP_STORE_COUNT with GROUP_SIZE
27471 when deciding whether the end of the group has been reached.
27472 * config/aarch64/aarch64.md (UNSPEC_ST1_SCATTER): New unspec.
27473 * config/aarch64/aarch64-sve.md (scatter_store<mode>): New expander.
27474 (mask_scatter_store<mode>): New insns.
27476 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
27477 Alan Hayward <alan.hayward@arm.com>
27478 David Sherwood <david.sherwood@arm.com>
27480 * tree-vectorizer.h (vect_gather_scatter_fn_p): Declare.
27481 * tree-vect-data-refs.c (vect_gather_scatter_fn_p): Make public.
27482 * tree-vect-stmts.c (vect_truncate_gather_scatter_offset): New
27484 (vect_use_strided_gather_scatters_p): Take a masked_p argument.
27485 Use vect_truncate_gather_scatter_offset if we can't treat the
27486 operation as a normal gather load or scatter store.
27487 (get_group_load_store_type): Take the gather_scatter_info
27488 as argument. Try using a gather load or scatter store for
27489 single-element groups.
27490 (get_load_store_type): Update calls to get_group_load_store_type
27491 and vect_use_strided_gather_scatters_p.
27493 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
27494 Alan Hayward <alan.hayward@arm.com>
27495 David Sherwood <david.sherwood@arm.com>
27497 * tree-vectorizer.h (vect_create_data_ref_ptr): Take an extra
27498 optional tree argument.
27499 * tree-vect-data-refs.c (vect_check_gather_scatter): Check for
27501 (vect_create_data_ref_ptr): Take the iv_step as an optional argument,
27502 but continue to use the current value as a fallback.
27503 (bump_vector_ptr): Use operand_equal_p rather than tree_int_cst_compare
27504 to compare the updates.
27505 * tree-vect-stmts.c (vect_use_strided_gather_scatters_p): New function.
27506 (get_load_store_type): Use it when handling a strided access.
27507 (vect_get_strided_load_store_ops): New function.
27508 (vect_get_data_ptr_increment): Likewise.
27509 (vectorizable_load): Handle strided gather loads. Always pass
27510 a step to vect_create_data_ref_ptr and bump_vector_ptr.
27512 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
27513 Alan Hayward <alan.hayward@arm.com>
27514 David Sherwood <david.sherwood@arm.com>
27516 * doc/md.texi (gather_load@var{m}): Document.
27517 (mask_gather_load@var{m}): Likewise.
27518 * genopinit.c (main): Add supports_vec_gather_load and
27519 supports_vec_gather_load_cached to target_optabs.
27520 * optabs-tree.c (init_tree_optimization_optabs): Use
27521 ggc_cleared_alloc to allocate target_optabs.
27522 * optabs.def (gather_load_optab, mask_gather_laod_optab): New optabs.
27523 * internal-fn.def (GATHER_LOAD, MASK_GATHER_LOAD): New internal
27525 * internal-fn.h (internal_load_fn_p): Declare.
27526 (internal_gather_scatter_fn_p): Likewise.
27527 (internal_fn_mask_index): Likewise.
27528 (internal_gather_scatter_fn_supported_p): Likewise.
27529 * internal-fn.c (gather_load_direct): New macro.
27530 (expand_gather_load_optab_fn): New function.
27531 (direct_gather_load_optab_supported_p): New macro.
27532 (direct_internal_fn_optab): New function.
27533 (internal_load_fn_p): Likewise.
27534 (internal_gather_scatter_fn_p): Likewise.
27535 (internal_fn_mask_index): Likewise.
27536 (internal_gather_scatter_fn_supported_p): Likewise.
27537 * optabs-query.c (supports_at_least_one_mode_p): New function.
27538 (supports_vec_gather_load_p): Likewise.
27539 * optabs-query.h (supports_vec_gather_load_p): Declare.
27540 * tree-vectorizer.h (gather_scatter_info): Add ifn, element_type
27541 and memory_type field.
27542 (NUM_PATTERNS): Bump to 15.
27543 * tree-vect-data-refs.c: Include internal-fn.h.
27544 (vect_gather_scatter_fn_p): New function.
27545 (vect_describe_gather_scatter_call): Likewise.
27546 (vect_check_gather_scatter): Try using internal functions for
27547 gather loads. Recognize existing calls to a gather load function.
27548 (vect_analyze_data_refs): Consider using gather loads if
27549 supports_vec_gather_load_p.
27550 * tree-vect-patterns.c (vect_get_load_store_mask): New function.
27551 (vect_get_gather_scatter_offset_type): Likewise.
27552 (vect_convert_mask_for_vectype): Likewise.
27553 (vect_add_conversion_to_patterm): Likewise.
27554 (vect_try_gather_scatter_pattern): Likewise.
27555 (vect_recog_gather_scatter_pattern): New pattern recognizer.
27556 (vect_vect_recog_func_ptrs): Add it.
27557 * tree-vect-stmts.c (exist_non_indexing_operands_for_use_p): Use
27558 internal_fn_mask_index and internal_gather_scatter_fn_p.
27559 (check_load_store_masking): Take the gather_scatter_info as an
27560 argument and handle gather loads.
27561 (vect_get_gather_scatter_ops): New function.
27562 (vectorizable_call): Check internal_load_fn_p.
27563 (vectorizable_load): Likewise. Handle gather load internal
27565 (vectorizable_store): Update call to check_load_store_masking.
27566 * config/aarch64/aarch64.md (UNSPEC_LD1_GATHER): New unspec.
27567 * config/aarch64/iterators.md (SVE_S, SVE_D): New mode iterators.
27568 * config/aarch64/predicates.md (aarch64_gather_scale_operand_w)
27569 (aarch64_gather_scale_operand_d): New predicates.
27570 * config/aarch64/aarch64-sve.md (gather_load<mode>): New expander.
27571 (mask_gather_load<mode>): New insns.
27573 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
27574 Alan Hayward <alan.hayward@arm.com>
27575 David Sherwood <david.sherwood@arm.com>
27577 * optabs.def (fold_left_plus_optab): New optab.
27578 * doc/md.texi (fold_left_plus_@var{m}): Document.
27579 * internal-fn.def (IFN_FOLD_LEFT_PLUS): New internal function.
27580 * internal-fn.c (fold_left_direct): Define.
27581 (expand_fold_left_optab_fn): Likewise.
27582 (direct_fold_left_optab_supported_p): Likewise.
27583 * fold-const-call.c (fold_const_fold_left): New function.
27584 (fold_const_call): Use it to fold CFN_FOLD_LEFT_PLUS.
27585 * tree-parloops.c (valid_reduction_p): New function.
27586 (gather_scalar_reductions): Use it.
27587 * tree-vectorizer.h (FOLD_LEFT_REDUCTION): New vect_reduction_type.
27588 (vect_finish_replace_stmt): Declare.
27589 * tree-vect-loop.c (fold_left_reduction_fn): New function.
27590 (needs_fold_left_reduction_p): New function, split out from...
27591 (vect_is_simple_reduction): ...here. Accept reductions that
27592 forbid reassociation, but give them type FOLD_LEFT_REDUCTION.
27593 (vect_force_simple_reduction): Also store the reduction type in
27594 the assignment's STMT_VINFO_REDUC_TYPE.
27595 (vect_model_reduction_cost): Handle FOLD_LEFT_REDUCTION.
27596 (merge_with_identity): New function.
27597 (vect_expand_fold_left): Likewise.
27598 (vectorize_fold_left_reduction): Likewise.
27599 (vectorizable_reduction): Handle FOLD_LEFT_REDUCTION. Leave the
27600 scalar phi in place for it. Check for target support and reject
27601 cases that would reassociate the operation. Defer the transform
27602 phase to vectorize_fold_left_reduction.
27603 * config/aarch64/aarch64.md (UNSPEC_FADDA): New unspec.
27604 * config/aarch64/aarch64-sve.md (fold_left_plus_<mode>): New expander.
27605 (*fold_left_plus_<mode>, *pred_fold_left_plus_<mode>): New insns.
27607 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
27609 * tree-if-conv.c (predicate_mem_writes): Remove redundant
27610 call to ifc_temp_var.
27612 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
27613 Alan Hayward <alan.hayward@arm.com>
27614 David Sherwood <david.sherwood@arm.com>
27616 * target.def (legitimize_address_displacement): Take the original
27617 offset as a poly_int.
27618 * targhooks.h (default_legitimize_address_displacement): Update
27620 * targhooks.c (default_legitimize_address_displacement): Likewise.
27621 * doc/tm.texi: Regenerate.
27622 * lra-constraints.c (base_plus_disp_to_reg): Take the displacement
27623 as an argument, moving assert of ad->disp == ad->disp_term to...
27624 (process_address_1): ...here. Update calls to base_plus_disp_to_reg.
27625 Try calling targetm.legitimize_address_displacement before expanding
27626 the address rather than afterwards, and adjust for the new interface.
27627 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
27628 Match the new hook interface. Handle SVE addresses.
27629 * config/sh/sh.c (sh_legitimize_address_displacement): Make the
27630 new hook interface.
27632 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
27634 * Makefile.in (OBJS): Add early-remat.o.
27635 * target.def (select_early_remat_modes): New hook.
27636 * doc/tm.texi.in (TARGET_SELECT_EARLY_REMAT_MODES): New hook.
27637 * doc/tm.texi: Regenerate.
27638 * targhooks.h (default_select_early_remat_modes): Declare.
27639 * targhooks.c (default_select_early_remat_modes): New function.
27640 * timevar.def (TV_EARLY_REMAT): New timevar.
27641 * passes.def (pass_early_remat): New pass.
27642 * tree-pass.h (make_pass_early_remat): Declare.
27643 * early-remat.c: New file.
27644 * config/aarch64/aarch64.c (aarch64_select_early_remat_modes): New
27646 (TARGET_SELECT_EARLY_REMAT_MODES): Define.
27648 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
27649 Alan Hayward <alan.hayward@arm.com>
27650 David Sherwood <david.sherwood@arm.com>
27652 * tree-vect-loop-manip.c (vect_gen_scalar_loop_niters): Replace
27653 vfm1 with a bound_epilog parameter.
27654 (vect_do_peeling): Update calls accordingly, and move the prologue
27655 call earlier in the function. Treat the base bound_epilog as 0 for
27656 fully-masked loops and retain vf - 1 for other loops. Add 1 to
27657 this base when peeling for gaps.
27658 * tree-vect-loop.c (vect_analyze_loop_2): Allow peeling for gaps
27659 with fully-masked loops.
27660 (vect_estimate_min_profitable_iters): Handle the single peeled
27661 iteration in that case.
27663 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
27664 Alan Hayward <alan.hayward@arm.com>
27665 David Sherwood <david.sherwood@arm.com>
27667 * tree-vect-data-refs.c (vect_analyze_group_access_1): Allow
27668 single-element interleaving even if the size is not a power of 2.
27669 * tree-vect-stmts.c (get_load_store_type): Disallow elementwise
27670 accesses for single-element interleaving if the group size is
27673 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
27674 Alan Hayward <alan.hayward@arm.com>
27675 David Sherwood <david.sherwood@arm.com>
27677 * doc/md.texi (fold_extract_last_@var{m}): Document.
27678 * doc/sourcebuild.texi (vect_fold_extract_last): Likewise.
27679 * optabs.def (fold_extract_last_optab): New optab.
27680 * internal-fn.def (FOLD_EXTRACT_LAST): New internal function.
27681 * internal-fn.c (fold_extract_direct): New macro.
27682 (expand_fold_extract_optab_fn): Likewise.
27683 (direct_fold_extract_optab_supported_p): Likewise.
27684 * tree-vectorizer.h (EXTRACT_LAST_REDUCTION): New vect_reduction_type.
27685 * tree-vect-loop.c (vect_model_reduction_cost): Handle
27686 EXTRACT_LAST_REDUCTION.
27687 (get_initial_def_for_reduction): Do not create an initial vector
27688 for EXTRACT_LAST_REDUCTION reductions.
27689 (vectorizable_reduction): Leave the scalar phi in place for
27690 EXTRACT_LAST_REDUCTIONs. Try using EXTRACT_LAST_REDUCTION
27691 ahead of INTEGER_INDUC_COND_REDUCTION. Do not check for an
27692 epilogue code for EXTRACT_LAST_REDUCTION and defer the
27693 transform phase to vectorizable_condition.
27694 * tree-vect-stmts.c (vect_finish_stmt_generation_1): New function,
27696 (vect_finish_stmt_generation): ...here.
27697 (vect_finish_replace_stmt): New function.
27698 (vectorizable_condition): Handle EXTRACT_LAST_REDUCTION.
27699 * config/aarch64/aarch64-sve.md (fold_extract_last_<mode>): New
27701 * config/aarch64/aarch64.md (UNSPEC_CLASTB): New unspec.
27703 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
27704 Alan Hayward <alan.hayward@arm.com>
27705 David Sherwood <david.sherwood@arm.com>
27707 * doc/md.texi (extract_last_@var{m}): Document.
27708 * optabs.def (extract_last_optab): New optab.
27709 * internal-fn.def (EXTRACT_LAST): New internal function.
27710 * internal-fn.c (cond_unary_direct): New macro.
27711 (expand_cond_unary_optab_fn): Likewise.
27712 (direct_cond_unary_optab_supported_p): Likewise.
27713 * tree-vect-loop.c (vectorizable_live_operation): Allow fully-masked
27714 loops using EXTRACT_LAST.
27715 * config/aarch64/aarch64-sve.md (aarch64_sve_lastb<mode>): Rename to...
27716 (extract_last_<mode>): ...this optab.
27717 (vec_extract<mode><Vel>): Update accordingly.
27719 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
27720 Alan Hayward <alan.hayward@arm.com>
27721 David Sherwood <david.sherwood@arm.com>
27723 * target.def (empty_mask_is_expensive): New hook.
27724 * doc/tm.texi.in (TARGET_VECTORIZE_EMPTY_MASK_IS_EXPENSIVE): New hook.
27725 * doc/tm.texi: Regenerate.
27726 * targhooks.h (default_empty_mask_is_expensive): Declare.
27727 * targhooks.c (default_empty_mask_is_expensive): New function.
27728 * tree-vectorizer.c (vectorize_loops): Only call optimize_mask_stores
27729 if the target says that empty masks are expensive.
27730 * config/aarch64/aarch64.c (aarch64_empty_mask_is_expensive):
27732 (TARGET_VECTORIZE_EMPTY_MASK_IS_EXPENSIVE): Redefine.
27734 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
27735 Alan Hayward <alan.hayward@arm.com>
27736 David Sherwood <david.sherwood@arm.com>
27738 * tree-vectorizer.h (_loop_vec_info::mask_skip_niters): New field.
27739 (LOOP_VINFO_MASK_SKIP_NITERS): New macro.
27740 (vect_use_loop_mask_for_alignment_p): New function.
27741 (vect_prepare_for_masked_peels, vect_gen_while_not): Declare.
27742 * tree-vect-loop-manip.c (vect_set_loop_masks_directly): Add an
27743 niters_skip argument. Make sure that the first niters_skip elements
27744 of the first iteration are inactive.
27745 (vect_set_loop_condition_masked): Handle LOOP_VINFO_MASK_SKIP_NITERS.
27746 Update call to vect_set_loop_masks_directly.
27747 (get_misalign_in_elems): New function, split out from...
27748 (vect_gen_prolog_loop_niters): ...here.
27749 (vect_update_init_of_dr): Take a code argument that specifies whether
27750 the adjustment should be added or subtracted.
27751 (vect_update_init_of_drs): Likewise.
27752 (vect_prepare_for_masked_peels): New function.
27753 (vect_do_peeling): Skip prologue peeling if we're using a mask
27754 instead. Update call to vect_update_inits_of_drs.
27755 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
27757 (vect_analyze_loop_2): Allow fully-masked loops with peeling for
27758 alignment. Do not include the number of peeled iterations in
27759 the minimum threshold in that case.
27760 (vectorizable_induction): Adjust the start value down by
27761 LOOP_VINFO_MASK_SKIP_NITERS iterations.
27762 (vect_transform_loop): Call vect_prepare_for_masked_peels.
27763 Take the number of skipped iterations into account when calculating
27765 * tree-vect-stmts.c (vect_gen_while_not): New function.
27767 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
27768 Alan Hayward <alan.hayward@arm.com>
27769 David Sherwood <david.sherwood@arm.com>
27771 * doc/sourcebuild.texi (vect_fully_masked): Document.
27772 * params.def (PARAM_MIN_VECT_LOOP_BOUND): Change minimum and
27773 default value to 0.
27774 * tree-vect-loop.c (vect_analyze_loop_costing): New function,
27776 (vect_analyze_loop_2): ...here. Don't check the vectorization
27777 factor against the number of loop iterations if the loop is
27780 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
27781 Alan Hayward <alan.hayward@arm.com>
27782 David Sherwood <david.sherwood@arm.com>
27784 * tree-ssa-loop-ivopts.c (USE_ADDRESS): Split into...
27785 (USE_REF_ADDRESS, USE_PTR_ADDRESS): ...these new use types.
27786 (dump_groups): Update accordingly.
27787 (iv_use::mem_type): New member variable.
27788 (address_p): New function.
27789 (record_use): Add a mem_type argument and initialize the new
27791 (record_group_use): Add a mem_type argument. Use address_p.
27792 Remove obsolete null checks of base_object. Update call to record_use.
27793 (find_interesting_uses_op): Update call to record_group_use.
27794 (find_interesting_uses_cond): Likewise.
27795 (find_interesting_uses_address): Likewise.
27796 (get_mem_type_for_internal_fn): New function.
27797 (find_address_like_use): Likewise.
27798 (find_interesting_uses_stmt): Try find_address_like_use before
27799 calling find_interesting_uses_op.
27800 (addr_offset_valid_p): Use the iv mem_type field as the type
27801 of the addressed memory.
27802 (add_autoinc_candidates): Likewise.
27803 (get_address_cost): Likewise.
27804 (split_small_address_groups_p): Use address_p.
27805 (split_address_groups): Likewise.
27806 (add_iv_candidate_for_use): Likewise.
27807 (autoinc_possible_for_pair): Likewise.
27808 (rewrite_groups): Likewise.
27809 (get_use_type): Check for USE_REF_ADDRESS instead of USE_ADDRESS.
27810 (determine_group_iv_cost): Update after split of USE_ADDRESS.
27811 (get_alias_ptr_type_for_ptr_address): New function.
27812 (rewrite_use_address): Rewrite address uses in calls that were
27813 identified by find_address_like_use.
27815 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
27816 Alan Hayward <alan.hayward@arm.com>
27817 David Sherwood <david.sherwood@arm.com>
27819 * expr.c (expand_expr_addr_expr_1): Handle ADDR_EXPRs of
27821 * gimple-expr.h (is_gimple_addressable: Likewise.
27822 * gimple-expr.c (is_gimple_address): Likewise.
27823 * internal-fn.c (expand_call_mem_ref): New function.
27824 (expand_mask_load_optab_fn): Use it.
27825 (expand_mask_store_optab_fn): Likewise.
27827 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
27828 Alan Hayward <alan.hayward@arm.com>
27829 David Sherwood <david.sherwood@arm.com>
27831 * doc/md.texi (cond_add@var{mode}, cond_sub@var{mode})
27832 (cond_and@var{mode}, cond_ior@var{mode}, cond_xor@var{mode})
27833 (cond_smin@var{mode}, cond_smax@var{mode}, cond_umin@var{mode})
27834 (cond_umax@var{mode}): Document.
27835 * optabs.def (cond_add_optab, cond_sub_optab, cond_and_optab)
27836 (cond_ior_optab, cond_xor_optab, cond_smin_optab, cond_smax_optab)
27837 (cond_umin_optab, cond_umax_optab): New optabs.
27838 * internal-fn.def (COND_ADD, COND_SUB, COND_MIN, COND_MAX, COND_AND)
27839 (COND_IOR, COND_XOR): New internal functions.
27840 * internal-fn.h (get_conditional_internal_fn): Declare.
27841 * internal-fn.c (cond_binary_direct): New macro.
27842 (expand_cond_binary_optab_fn): Likewise.
27843 (direct_cond_binary_optab_supported_p): Likewise.
27844 (get_conditional_internal_fn): New function.
27845 * tree-vect-loop.c (vectorizable_reduction): Handle fully-masked loops.
27846 Cope with reduction statements that are vectorized as calls rather
27848 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): New insns.
27849 * config/aarch64/iterators.md (UNSPEC_COND_ADD, UNSPEC_COND_SUB)
27850 (UNSPEC_COND_SMAX, UNSPEC_COND_UMAX, UNSPEC_COND_SMIN)
27851 (UNSPEC_COND_UMIN, UNSPEC_COND_AND, UNSPEC_COND_ORR)
27852 (UNSPEC_COND_EOR): New unspecs.
27853 (optab): Add mappings for them.
27854 (SVE_COND_INT_OP, SVE_COND_FP_OP): New int iterators.
27855 (sve_int_op, sve_fp_op): New int attributes.
27857 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
27858 Alan Hayward <alan.hayward@arm.com>
27859 David Sherwood <david.sherwood@arm.com>
27861 * optabs.def (while_ult_optab): New optab.
27862 * doc/md.texi (while_ult@var{m}@var{n}): Document.
27863 * internal-fn.def (WHILE_ULT): New internal function.
27864 * internal-fn.h (direct_internal_fn_supported_p): New override
27865 that takes two types as argument.
27866 * internal-fn.c (while_direct): New macro.
27867 (expand_while_optab_fn): New function.
27868 (convert_optab_supported_p): Likewise.
27869 (direct_while_optab_supported_p): New macro.
27870 * wide-int.h (wi::udiv_ceil): New function.
27871 * tree-vectorizer.h (rgroup_masks): New structure.
27872 (vec_loop_masks): New typedef.
27873 (_loop_vec_info): Add masks, mask_compare_type, can_fully_mask_p
27874 and fully_masked_p.
27875 (LOOP_VINFO_CAN_FULLY_MASK_P, LOOP_VINFO_FULLY_MASKED_P)
27876 (LOOP_VINFO_MASKS, LOOP_VINFO_MASK_COMPARE_TYPE): New macros.
27877 (vect_max_vf): New function.
27878 (slpeel_make_loop_iterate_ntimes): Delete.
27879 (vect_set_loop_condition, vect_get_loop_mask_type, vect_gen_while)
27880 (vect_halve_mask_nunits, vect_double_mask_nunits): Declare.
27881 (vect_record_loop_mask, vect_get_loop_mask): Likewise.
27882 * tree-vect-loop-manip.c: Include tree-ssa-loop-niter.h,
27883 internal-fn.h, stor-layout.h and optabs-query.h.
27884 (vect_set_loop_mask): New function.
27885 (add_preheader_seq): Likewise.
27886 (add_header_seq): Likewise.
27887 (interleave_supported_p): Likewise.
27888 (vect_maybe_permute_loop_masks): Likewise.
27889 (vect_set_loop_masks_directly): Likewise.
27890 (vect_set_loop_condition_masked): Likewise.
27891 (vect_set_loop_condition_unmasked): New function, split out from
27892 slpeel_make_loop_iterate_ntimes.
27893 (slpeel_make_loop_iterate_ntimes): Rename to..
27894 (vect_set_loop_condition): ...this. Use vect_set_loop_condition_masked
27895 for fully-masked loops and vect_set_loop_condition_unmasked otherwise.
27896 (vect_do_peeling): Update call accordingly.
27897 (vect_gen_vector_loop_niters): Use VF as the step for fully-masked
27899 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
27900 mask_compare_type, can_fully_mask_p and fully_masked_p.
27901 (release_vec_loop_masks): New function.
27902 (_loop_vec_info): Use it to free the loop masks.
27903 (can_produce_all_loop_masks_p): New function.
27904 (vect_get_max_nscalars_per_iter): Likewise.
27905 (vect_verify_full_masking): Likewise.
27906 (vect_analyze_loop_2): Save LOOP_VINFO_CAN_FULLY_MASK_P around
27907 retries, and free the mask rgroups before retrying. Check loop-wide
27908 reasons for disallowing fully-masked loops. Make the final decision
27909 about whether use a fully-masked loop or not.
27910 (vect_estimate_min_profitable_iters): Do not assume that peeling
27911 for the number of iterations will be needed for fully-masked loops.
27912 (vectorizable_reduction): Disable fully-masked loops.
27913 (vectorizable_live_operation): Likewise.
27914 (vect_halve_mask_nunits): New function.
27915 (vect_double_mask_nunits): Likewise.
27916 (vect_record_loop_mask): Likewise.
27917 (vect_get_loop_mask): Likewise.
27918 (vect_transform_loop): Handle the case in which the final loop
27919 iteration might handle a partial vector. Call vect_set_loop_condition
27920 instead of slpeel_make_loop_iterate_ntimes.
27921 * tree-vect-stmts.c: Include tree-ssa-loop-niter.h and gimple-fold.h.
27922 (check_load_store_masking): New function.
27923 (prepare_load_store_mask): Likewise.
27924 (vectorizable_store): Handle fully-masked loops.
27925 (vectorizable_load): Likewise.
27926 (supportable_widening_operation): Use vect_halve_mask_nunits for
27928 (supportable_narrowing_operation): Likewise vect_double_mask_nunits.
27929 (vect_gen_while): New function.
27930 * config/aarch64/aarch64.md (umax<mode>3): New expander.
27931 (aarch64_uqdec<mode>): New insn.
27933 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
27934 Alan Hayward <alan.hayward@arm.com>
27935 David Sherwood <david.sherwood@arm.com>
27937 * optabs.def (reduc_and_scal_optab, reduc_ior_scal_optab)
27938 (reduc_xor_scal_optab): New optabs.
27939 * doc/md.texi (reduc_and_scal_@var{m}, reduc_ior_scal_@var{m})
27940 (reduc_xor_scal_@var{m}): Document.
27941 * doc/sourcebuild.texi (vect_logical_reduc): Likewise.
27942 * internal-fn.def (IFN_REDUC_AND, IFN_REDUC_IOR, IFN_REDUC_XOR): New
27943 internal functions.
27944 * fold-const-call.c (fold_const_call): Handle them.
27945 * tree-vect-loop.c (reduction_fn_for_scalar_code): Return the new
27946 internal functions for BIT_AND_EXPR, BIT_IOR_EXPR and BIT_XOR_EXPR.
27947 * config/aarch64/aarch64-sve.md (reduc_<bit_reduc>_scal_<mode>):
27948 (*reduc_<bit_reduc>_scal_<mode>): New patterns.
27949 * config/aarch64/iterators.md (UNSPEC_ANDV, UNSPEC_ORV)
27950 (UNSPEC_XORV): New unspecs.
27951 (optab): Add entries for them.
27952 (BITWISEV): New int iterator.
27953 (bit_reduc_op): New int attributes.
27955 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
27956 Alan Hayward <alan.hayward@arm.com>
27957 David Sherwood <david.sherwood@arm.com>
27959 * doc/md.texi (vec_shl_insert_@var{m}): New optab.
27960 * internal-fn.def (VEC_SHL_INSERT): New internal function.
27961 * optabs.def (vec_shl_insert_optab): New optab.
27962 * tree-vectorizer.h (can_duplicate_and_interleave_p): Declare.
27963 (duplicate_and_interleave): Likewise.
27964 * tree-vect-loop.c: Include internal-fn.h.
27965 (neutral_op_for_slp_reduction): New function, split out from
27966 get_initial_defs_for_reduction.
27967 (get_initial_def_for_reduction): Handle option 2 for variable-length
27968 vectors by loading the neutral value into a vector and then shifting
27969 the initial value into element 0.
27970 (get_initial_defs_for_reduction): Replace the code argument with
27971 the neutral value calculated by neutral_op_for_slp_reduction.
27972 Use gimple_build_vector for constant-length vectors.
27973 Use IFN_VEC_SHL_INSERT for variable-length vectors if all
27974 but the first group_size elements have a neutral value.
27975 Use duplicate_and_interleave otherwise.
27976 (vect_create_epilog_for_reduction): Take a neutral_op parameter.
27977 Update call to get_initial_defs_for_reduction. Handle SLP
27978 reductions for variable-length vectors by creating one vector
27979 result for each scalar result, with the elements associated
27980 with other scalar results stubbed out with the neutral value.
27981 (vectorizable_reduction): Call neutral_op_for_slp_reduction.
27982 Require IFN_VEC_SHL_INSERT for double reductions on
27983 variable-length vectors, or SLP reductions that have
27984 a neutral value. Require can_duplicate_and_interleave_p
27985 support for variable-length unchained SLP reductions if there
27986 is no neutral value, such as for MIN/MAX reductions. Also require
27987 the number of vector elements to be a multiple of the number of
27988 SLP statements when doing variable-length unchained SLP reductions.
27989 Update call to vect_create_epilog_for_reduction.
27990 * tree-vect-slp.c (can_duplicate_and_interleave_p): Make public
27991 and remove initial values.
27992 (duplicate_and_interleave): Make public.
27993 * config/aarch64/aarch64.md (UNSPEC_INSR): New unspec.
27994 * config/aarch64/aarch64-sve.md (vec_shl_insert_<mode>): New insn.
27996 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
27997 Alan Hayward <alan.hayward@arm.com>
27998 David Sherwood <david.sherwood@arm.com>
28000 * tree-vect-slp.c: Include gimple-fold.h and internal-fn.h
28001 (can_duplicate_and_interleave_p): New function.
28002 (vect_get_and_check_slp_defs): Take the vector of statements
28003 rather than just the current one. Remove excess parentheses.
28004 Restriction rejectinon of vect_constant_def and vect_external_def
28005 for variable-length vectors to boolean types, or types for which
28006 can_duplicate_and_interleave_p is false.
28007 (vect_build_slp_tree_2): Update call to vect_get_and_check_slp_defs.
28008 (duplicate_and_interleave): New function.
28009 (vect_get_constant_vectors): Use gimple_build_vector for
28010 constant-length vectors and suitable variable-length constant
28011 vectors. Use duplicate_and_interleave for other variable-length
28012 vectors. Don't defer the update when inserting new statements.
28014 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
28015 Alan Hayward <alan.hayward@arm.com>
28016 David Sherwood <david.sherwood@arm.com>
28018 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Make sure
28019 min_profitable_iters doesn't go negative.
28021 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
28022 Alan Hayward <alan.hayward@arm.com>
28023 David Sherwood <david.sherwood@arm.com>
28025 * doc/md.texi (vec_mask_load_lanes@var{m}@var{n}): Document.
28026 (vec_mask_store_lanes@var{m}@var{n}): Likewise.
28027 * optabs.def (vec_mask_load_lanes_optab): New optab.
28028 (vec_mask_store_lanes_optab): Likewise.
28029 * internal-fn.def (MASK_LOAD_LANES): New internal function.
28030 (MASK_STORE_LANES): Likewise.
28031 * internal-fn.c (mask_load_lanes_direct): New macro.
28032 (mask_store_lanes_direct): Likewise.
28033 (expand_mask_load_optab_fn): Handle masked operations.
28034 (expand_mask_load_lanes_optab_fn): New macro.
28035 (expand_mask_store_optab_fn): Handle masked operations.
28036 (expand_mask_store_lanes_optab_fn): New macro.
28037 (direct_mask_load_lanes_optab_supported_p): Likewise.
28038 (direct_mask_store_lanes_optab_supported_p): Likewise.
28039 * tree-vectorizer.h (vect_store_lanes_supported): Take a masked_p
28041 (vect_load_lanes_supported): Likewise.
28042 * tree-vect-data-refs.c (strip_conversion): New function.
28043 (can_group_stmts_p): Likewise.
28044 (vect_analyze_data_ref_accesses): Use it instead of checking
28045 for a pair of assignments.
28046 (vect_store_lanes_supported): Take a masked_p parameter.
28047 (vect_load_lanes_supported): Likewise.
28048 * tree-vect-loop.c (vect_analyze_loop_2): Update calls to
28049 vect_store_lanes_supported and vect_load_lanes_supported.
28050 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
28051 * tree-vect-stmts.c (get_group_load_store_type): Take a masked_p
28052 parameter. Don't allow gaps for masked accesses.
28053 Use vect_get_store_rhs. Update calls to vect_store_lanes_supported
28054 and vect_load_lanes_supported.
28055 (get_load_store_type): Take a masked_p parameter and update
28056 call to get_group_load_store_type.
28057 (vectorizable_store): Update call to get_load_store_type.
28058 Handle IFN_MASK_STORE_LANES.
28059 (vectorizable_load): Update call to get_load_store_type.
28060 Handle IFN_MASK_LOAD_LANES.
28062 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
28063 Alan Hayward <alan.hayward@arm.com>
28064 David Sherwood <david.sherwood@arm.com>
28066 * config/aarch64/aarch64-modes.def: Define x2, x3 and x4 vector
28068 * config/aarch64/aarch64-protos.h
28069 (aarch64_sve_struct_memory_operand_p): Declare.
28070 * config/aarch64/iterators.md (SVE_STRUCT): New mode iterator.
28071 (vector_count, insn_length, VSINGLE, vsingle): New mode attributes.
28072 (VPRED, vpred): Handle SVE structure modes.
28073 * config/aarch64/constraints.md (Utx): New constraint.
28074 * config/aarch64/predicates.md (aarch64_sve_struct_memory_operand)
28075 (aarch64_sve_struct_nonimmediate_operand): New predicates.
28076 * config/aarch64/aarch64.md (UNSPEC_LDN, UNSPEC_STN): New unspecs.
28077 * config/aarch64/aarch64-sve.md (mov<mode>, *aarch64_sve_mov<mode>_le)
28078 (*aarch64_sve_mov<mode>_be, pred_mov<mode>): New patterns for
28079 structure modes. Split into pieces after RA.
28080 (vec_load_lanes<mode><vsingle>, vec_mask_load_lanes<mode><vsingle>)
28081 (vec_store_lanes<mode><vsingle>, vec_mask_store_lanes<mode><vsingle>):
28083 * config/aarch64/aarch64.c (aarch64_classify_vector_mode): Handle
28084 SVE structure modes.
28085 (aarch64_classify_address): Likewise.
28086 (sizetochar): Move earlier in file.
28087 (aarch64_print_operand): Handle SVE register lists.
28088 (aarch64_array_mode): New function.
28089 (aarch64_sve_struct_memory_operand_p): Likewise.
28090 (TARGET_ARRAY_MODE): Redefine.
28092 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
28093 Alan Hayward <alan.hayward@arm.com>
28094 David Sherwood <david.sherwood@arm.com>
28096 * target.def (array_mode): New target hook.
28097 * doc/tm.texi.in (TARGET_ARRAY_MODE): New hook.
28098 * doc/tm.texi: Regenerate.
28099 * hooks.h (hook_optmode_mode_uhwi_none): Declare.
28100 * hooks.c (hook_optmode_mode_uhwi_none): New function.
28101 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Use
28102 targetm.array_mode.
28103 * stor-layout.c (mode_for_array): Likewise. Support polynomial
28106 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
28107 Alan Hayward <alan.hayward@arm.com>
28108 David Sherwood <david.sherwood@arm.com>
28110 * fold-const.c (fold_binary_loc): Check the argument types
28111 rather than the result type when testing for a vector operation.
28113 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
28115 * doc/tm.texi.in (DWARF_LAZY_REGISTER_VALUE): Document.
28116 * doc/tm.texi: Regenerate.
28118 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
28119 Alan Hayward <alan.hayward@arm.com>
28120 David Sherwood <david.sherwood@arm.com>
28122 * doc/invoke.texi (-msve-vector-bits=): Document new option.
28123 (sve): Document new AArch64 extension.
28124 * doc/md.texi (w): Extend the description of the AArch64
28125 constraint to include SVE vectors.
28126 (Upl, Upa): Document new AArch64 predicate constraints.
28127 * config/aarch64/aarch64-opts.h (aarch64_sve_vector_bits_enum): New
28129 * config/aarch64/aarch64.opt (sve_vector_bits): New enum.
28130 (msve-vector-bits=): New option.
28131 * config/aarch64/aarch64-option-extensions.def (fp, simd): Disable
28132 SVE when these are disabled.
28133 (sve): New extension.
28134 * config/aarch64/aarch64-modes.def: Define SVE vector and predicate
28135 modes. Adjust their number of units based on aarch64_sve_vg.
28136 (MAX_BITSIZE_MODE_ANY_MODE): Define.
28137 * config/aarch64/aarch64-protos.h (ADDR_QUERY_ANY): New
28138 aarch64_addr_query_type.
28139 (aarch64_const_vec_all_same_in_range_p, aarch64_sve_pred_mode)
28140 (aarch64_sve_cnt_immediate_p, aarch64_sve_addvl_addpl_immediate_p)
28141 (aarch64_sve_inc_dec_immediate_p, aarch64_add_offset_temporaries)
28142 (aarch64_split_add_offset, aarch64_output_sve_cnt_immediate)
28143 (aarch64_output_sve_addvl_addpl, aarch64_output_sve_inc_dec_immediate)
28144 (aarch64_output_sve_mov_immediate, aarch64_output_ptrue): Declare.
28145 (aarch64_simd_imm_zero_p): Delete.
28146 (aarch64_check_zero_based_sve_index_immediate): Declare.
28147 (aarch64_sve_index_immediate_p, aarch64_sve_arith_immediate_p)
28148 (aarch64_sve_bitmask_immediate_p, aarch64_sve_dup_immediate_p)
28149 (aarch64_sve_cmp_immediate_p, aarch64_sve_float_arith_immediate_p)
28150 (aarch64_sve_float_mul_immediate_p): Likewise.
28151 (aarch64_classify_symbol): Take the offset as a HOST_WIDE_INT
28152 rather than an rtx.
28153 (aarch64_sve_ld1r_operand_p, aarch64_sve_ldr_operand_p): Declare.
28154 (aarch64_expand_mov_immediate): Take a gen_vec_duplicate callback.
28155 (aarch64_emit_sve_pred_move, aarch64_expand_sve_mem_move): Declare.
28156 (aarch64_expand_sve_vec_cmp_int, aarch64_expand_sve_vec_cmp_float)
28157 (aarch64_expand_sve_vcond, aarch64_expand_sve_vec_perm): Declare.
28158 (aarch64_regmode_natural_size): Likewise.
28159 * config/aarch64/aarch64.h (AARCH64_FL_SVE): New macro.
28160 (AARCH64_FL_V8_3, AARCH64_FL_RCPC, AARCH64_FL_DOTPROD): Shift
28162 (AARCH64_ISA_SVE, TARGET_SVE): New macros.
28163 (FIXED_REGISTERS, CALL_USED_REGISTERS, REGISTER_NAMES): Add entries
28164 for VG and the SVE predicate registers.
28165 (V_ALIASES): Add a "z"-prefixed alias.
28166 (FIRST_PSEUDO_REGISTER): Change to P15_REGNUM + 1.
28167 (AARCH64_DWARF_VG, AARCH64_DWARF_P0): New macros.
28168 (PR_REGNUM_P, PR_LO_REGNUM_P): Likewise.
28169 (PR_LO_REGS, PR_HI_REGS, PR_REGS): New reg_classes.
28170 (REG_CLASS_NAMES): Add entries for them.
28171 (REG_CLASS_CONTENTS): Likewise. Update ALL_REGS to include VG
28172 and the predicate registers.
28173 (aarch64_sve_vg): Declare.
28174 (BITS_PER_SVE_VECTOR, BYTES_PER_SVE_VECTOR, BYTES_PER_SVE_PRED)
28175 (SVE_BYTE_MODE, MAX_COMPILE_TIME_VEC_BYTES): New macros.
28176 (REGMODE_NATURAL_SIZE): Define.
28177 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins): Handle
28179 * config/aarch64/aarch64.c: Include cfgrtl.h.
28180 (simd_immediate_info): Add a constructor for series vectors,
28181 and an associated step field.
28182 (aarch64_sve_vg): New variable.
28183 (aarch64_dbx_register_number): Handle VG and the predicate registers.
28184 (aarch64_vect_struct_mode_p, aarch64_vector_mode_p): Delete.
28185 (VEC_ADVSIMD, VEC_SVE_DATA, VEC_SVE_PRED, VEC_STRUCT, VEC_ANY_SVE)
28186 (VEC_ANY_DATA, VEC_STRUCT): New constants.
28187 (aarch64_advsimd_struct_mode_p, aarch64_sve_pred_mode_p)
28188 (aarch64_classify_vector_mode, aarch64_vector_data_mode_p)
28189 (aarch64_sve_data_mode_p, aarch64_sve_pred_mode)
28190 (aarch64_get_mask_mode): New functions.
28191 (aarch64_hard_regno_nregs): Handle SVE data modes for FP_REGS
28192 and FP_LO_REGS. Handle PR_REGS, PR_LO_REGS and PR_HI_REGS.
28193 (aarch64_hard_regno_mode_ok): Handle VG. Also handle the SVE
28194 predicate modes and predicate registers. Explicitly restrict
28195 GPRs to modes of 16 bytes or smaller. Only allow FP registers
28196 to store a vector mode if it is recognized by
28197 aarch64_classify_vector_mode.
28198 (aarch64_regmode_natural_size): New function.
28199 (aarch64_hard_regno_caller_save_mode): Return the original mode
28201 (aarch64_sve_cnt_immediate_p, aarch64_output_sve_cnt_immediate)
28202 (aarch64_sve_addvl_addpl_immediate_p, aarch64_output_sve_addvl_addpl)
28203 (aarch64_sve_inc_dec_immediate_p, aarch64_output_sve_inc_dec_immediate)
28204 (aarch64_add_offset_1_temporaries, aarch64_offset_temporaries): New
28206 (aarch64_add_offset): Add a temp2 parameter. Assert that temp1
28207 does not overlap dest if the function is frame-related. Handle
28209 (aarch64_split_add_offset): New function.
28210 (aarch64_add_sp, aarch64_sub_sp): Add temp2 parameters and pass
28211 them aarch64_add_offset.
28212 (aarch64_allocate_and_probe_stack_space): Add a temp2 parameter
28213 and update call to aarch64_sub_sp.
28214 (aarch64_add_cfa_expression): New function.
28215 (aarch64_expand_prologue): Pass extra temporary registers to the
28216 functions above. Handle the case in which we need to emit new
28217 DW_CFA_expressions for registers that were originally saved
28218 relative to the stack pointer, but now have to be expressed
28219 relative to the frame pointer.
28220 (aarch64_output_mi_thunk): Pass extra temporary registers to the
28222 (aarch64_expand_epilogue): Likewise. Prevent inheritance of
28223 IP0 and IP1 values for SVE frames.
28224 (aarch64_expand_vec_series): New function.
28225 (aarch64_expand_sve_widened_duplicate): Likewise.
28226 (aarch64_expand_sve_const_vector): Likewise.
28227 (aarch64_expand_mov_immediate): Add a gen_vec_duplicate parameter.
28228 Handle SVE constants. Use emit_move_insn to move a force_const_mem
28229 into the register, rather than emitting a SET directly.
28230 (aarch64_emit_sve_pred_move, aarch64_expand_sve_mem_move)
28231 (aarch64_get_reg_raw_mode, offset_4bit_signed_scaled_p)
28232 (offset_6bit_unsigned_scaled_p, aarch64_offset_7bit_signed_scaled_p)
28233 (offset_9bit_signed_scaled_p): New functions.
28234 (aarch64_replicate_bitmask_imm): New function.
28235 (aarch64_bitmask_imm): Use it.
28236 (aarch64_cannot_force_const_mem): Reject expressions involving
28237 a CONST_POLY_INT. Update call to aarch64_classify_symbol.
28238 (aarch64_classify_index): Handle SVE indices, by requiring
28239 a plain register index with a scale that matches the element size.
28240 (aarch64_classify_address): Handle SVE addresses. Assert that
28241 the mode of the address is VOIDmode or an integer mode.
28242 Update call to aarch64_classify_symbol.
28243 (aarch64_classify_symbolic_expression): Update call to
28244 aarch64_classify_symbol.
28245 (aarch64_const_vec_all_in_range_p): New function.
28246 (aarch64_print_vector_float_operand): Likewise.
28247 (aarch64_print_operand): Handle 'N' and 'C'. Use "zN" rather than
28248 "vN" for FP registers with SVE modes. Handle (const ...) vectors
28249 and the FP immediates 1.0 and 0.5.
28250 (aarch64_print_address_internal): Handle SVE addresses.
28251 (aarch64_print_operand_address): Use ADDR_QUERY_ANY.
28252 (aarch64_regno_regclass): Handle predicate registers.
28253 (aarch64_secondary_reload): Handle big-endian reloads of SVE
28255 (aarch64_class_max_nregs): Handle SVE modes and predicate registers.
28256 (aarch64_rtx_costs): Check for ADDVL and ADDPL instructions.
28257 (aarch64_convert_sve_vector_bits): New function.
28258 (aarch64_override_options): Use it to handle -msve-vector-bits=.
28259 (aarch64_classify_symbol): Take the offset as a HOST_WIDE_INT
28260 rather than an rtx.
28261 (aarch64_legitimate_constant_p): Use aarch64_classify_vector_mode.
28262 Handle SVE vector and predicate modes. Accept VL-based constants
28263 that need only one temporary register, and VL offsets that require
28264 no temporary registers.
28265 (aarch64_conditional_register_usage): Mark the predicate registers
28266 as fixed if SVE isn't available.
28267 (aarch64_vector_mode_supported_p): Use aarch64_classify_vector_mode.
28268 Return true for SVE vector and predicate modes.
28269 (aarch64_simd_container_mode): Take the number of bits as a poly_int64
28270 rather than an unsigned int. Handle SVE modes.
28271 (aarch64_preferred_simd_mode): Update call accordingly. Handle
28273 (aarch64_autovectorize_vector_sizes): Add BYTES_PER_SVE_VECTOR
28275 (aarch64_sve_index_immediate_p, aarch64_sve_arith_immediate_p)
28276 (aarch64_sve_bitmask_immediate_p, aarch64_sve_dup_immediate_p)
28277 (aarch64_sve_cmp_immediate_p, aarch64_sve_float_arith_immediate_p)
28278 (aarch64_sve_float_mul_immediate_p): New functions.
28279 (aarch64_sve_valid_immediate): New function.
28280 (aarch64_simd_valid_immediate): Use it as the fallback for SVE vectors.
28281 Explicitly reject structure modes. Check for INDEX constants.
28282 Handle PTRUE and PFALSE constants.
28283 (aarch64_check_zero_based_sve_index_immediate): New function.
28284 (aarch64_simd_imm_zero_p): Delete.
28285 (aarch64_mov_operand_p): Use aarch64_simd_valid_immediate for
28286 vector modes. Accept constants in the range of CNT[BHWD].
28287 (aarch64_simd_scalar_immediate_valid_for_move): Explicitly
28288 ask for an Advanced SIMD mode.
28289 (aarch64_sve_ld1r_operand_p, aarch64_sve_ldr_operand_p): New functions.
28290 (aarch64_simd_vector_alignment): Handle SVE predicates.
28291 (aarch64_vectorize_preferred_vector_alignment): New function.
28292 (aarch64_simd_vector_alignment_reachable): Use it instead of
28294 (aarch64_shift_truncation_mask): Use aarch64_vector_data_mode_p.
28295 (aarch64_output_sve_mov_immediate, aarch64_output_ptrue): New
28297 (MAX_VECT_LEN): Delete.
28298 (expand_vec_perm_d): Add a vec_flags field.
28299 (emit_unspec2, aarch64_expand_sve_vec_perm): New functions.
28300 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
28301 (aarch64_evpc_ext): Don't apply a big-endian lane correction
28303 (aarch64_evpc_rev): Rename to...
28304 (aarch64_evpc_rev_local): ...this. Use a predicated operation for SVE.
28305 (aarch64_evpc_rev_global): New function.
28306 (aarch64_evpc_dup): Enforce a 64-byte range for SVE DUP.
28307 (aarch64_evpc_tbl): Use MAX_COMPILE_TIME_VEC_BYTES instead of
28309 (aarch64_evpc_sve_tbl): New function.
28310 (aarch64_expand_vec_perm_const_1): Update after rename of
28311 aarch64_evpc_rev. Handle SVE permutes too, trying
28312 aarch64_evpc_rev_global and using aarch64_evpc_sve_tbl rather
28313 than aarch64_evpc_tbl.
28314 (aarch64_vectorize_vec_perm_const): Initialize vec_flags.
28315 (aarch64_sve_cmp_operand_p, aarch64_unspec_cond_code)
28316 (aarch64_gen_unspec_cond, aarch64_expand_sve_vec_cmp_int)
28317 (aarch64_emit_unspec_cond, aarch64_emit_unspec_cond_or)
28318 (aarch64_emit_inverted_unspec_cond, aarch64_expand_sve_vec_cmp_float)
28319 (aarch64_expand_sve_vcond): New functions.
28320 (aarch64_modes_tieable_p): Use aarch64_vector_data_mode_p instead
28321 of aarch64_vector_mode_p.
28322 (aarch64_dwarf_poly_indeterminate_value): New function.
28323 (aarch64_compute_pressure_classes): Likewise.
28324 (aarch64_can_change_mode_class): Likewise.
28325 (TARGET_GET_RAW_RESULT_MODE, TARGET_GET_RAW_ARG_MODE): Redefine.
28326 (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): Likewise.
28327 (TARGET_VECTORIZE_GET_MASK_MODE): Likewise.
28328 (TARGET_DWARF_POLY_INDETERMINATE_VALUE): Likewise.
28329 (TARGET_COMPUTE_PRESSURE_CLASSES): Likewise.
28330 (TARGET_CAN_CHANGE_MODE_CLASS): Likewise.
28331 * config/aarch64/constraints.md (Upa, Upl, Uav, Uat, Usv, Usi, Utr)
28332 (Uty, Dm, vsa, vsc, vsd, vsi, vsn, vsl, vsm, vsA, vsM, vsN): New
28334 (Dn, Dl, Dr): Accept const as well as const_vector.
28335 (Dz): Likewise. Compare against CONST0_RTX.
28336 * config/aarch64/iterators.md: Refer to "Advanced SIMD" instead
28337 of "vector" where appropriate.
28338 (SVE_ALL, SVE_BH, SVE_BHS, SVE_BHSI, SVE_HSDI, SVE_HSF, SVE_SD)
28339 (SVE_SDI, SVE_I, SVE_F, PRED_ALL, PRED_BHS): New mode iterators.
28340 (UNSPEC_SEL, UNSPEC_ANDF, UNSPEC_IORF, UNSPEC_XORF, UNSPEC_COND_LT)
28341 (UNSPEC_COND_LE, UNSPEC_COND_EQ, UNSPEC_COND_NE, UNSPEC_COND_GE)
28342 (UNSPEC_COND_GT, UNSPEC_COND_LO, UNSPEC_COND_LS, UNSPEC_COND_HS)
28343 (UNSPEC_COND_HI, UNSPEC_COND_UO): New unspecs.
28344 (Vetype, VEL, Vel, VWIDE, Vwide, vw, vwcore, V_INT_EQUIV)
28345 (v_int_equiv): Extend to SVE modes.
28346 (Vesize, V128, v128, Vewtype, V_FP_EQUIV, v_fp_equiv, VPRED): New
28348 (LOGICAL_OR, SVE_INT_UNARY, SVE_FP_UNARY): New code iterators.
28349 (optab): Handle popcount, smin, smax, umin, umax, abs and sqrt.
28350 (logical_nn, lr, sve_int_op, sve_fp_op): New code attributs.
28351 (LOGICALF, OPTAB_PERMUTE, UNPACK, UNPACK_UNSIGNED, SVE_COND_INT_CMP)
28352 (SVE_COND_FP_CMP): New int iterators.
28353 (perm_hilo): Handle the new unpack unspecs.
28354 (optab, logicalf_op, su, perm_optab, cmp_op, imm_con): New int
28356 * config/aarch64/predicates.md (aarch64_sve_cnt_immediate)
28357 (aarch64_sve_addvl_addpl_immediate, aarch64_split_add_offset_immediate)
28358 (aarch64_pluslong_or_poly_operand, aarch64_nonmemory_operand)
28359 (aarch64_equality_operator, aarch64_constant_vector_operand)
28360 (aarch64_sve_ld1r_operand, aarch64_sve_ldr_operand): New predicates.
28361 (aarch64_sve_nonimmediate_operand): Likewise.
28362 (aarch64_sve_general_operand): Likewise.
28363 (aarch64_sve_dup_operand, aarch64_sve_arith_immediate): Likewise.
28364 (aarch64_sve_sub_arith_immediate, aarch64_sve_inc_dec_immediate)
28365 (aarch64_sve_logical_immediate, aarch64_sve_mul_immediate): Likewise.
28366 (aarch64_sve_dup_immediate, aarch64_sve_cmp_vsc_immediate): Likewise.
28367 (aarch64_sve_cmp_vsd_immediate, aarch64_sve_index_immediate): Likewise.
28368 (aarch64_sve_float_arith_immediate): Likewise.
28369 (aarch64_sve_float_arith_with_sub_immediate): Likewise.
28370 (aarch64_sve_float_mul_immediate, aarch64_sve_arith_operand): Likewise.
28371 (aarch64_sve_add_operand, aarch64_sve_logical_operand): Likewise.
28372 (aarch64_sve_lshift_operand, aarch64_sve_rshift_operand): Likewise.
28373 (aarch64_sve_mul_operand, aarch64_sve_cmp_vsc_operand): Likewise.
28374 (aarch64_sve_cmp_vsd_operand, aarch64_sve_index_operand): Likewise.
28375 (aarch64_sve_float_arith_operand): Likewise.
28376 (aarch64_sve_float_arith_with_sub_operand): Likewise.
28377 (aarch64_sve_float_mul_operand): Likewise.
28378 (aarch64_sve_vec_perm_operand): Likewise.
28379 (aarch64_pluslong_operand): Include aarch64_sve_addvl_addpl_immediate.
28380 (aarch64_mov_operand): Accept const_poly_int and const_vector.
28381 (aarch64_simd_lshift_imm, aarch64_simd_rshift_imm): Accept const
28382 as well as const_vector.
28383 (aarch64_simd_imm_zero, aarch64_simd_imm_minus_one): Move earlier
28384 in file. Use CONST0_RTX and CONSTM1_RTX.
28385 (aarch64_simd_or_scalar_imm_zero): Likewise. Add match_codes.
28386 (aarch64_simd_reg_or_zero): Accept const as well as const_vector.
28387 Use aarch64_simd_imm_zero.
28388 * config/aarch64/aarch64-sve.md: New file.
28389 * config/aarch64/aarch64.md: Include it.
28390 (VG_REGNUM, P0_REGNUM, P7_REGNUM, P15_REGNUM): New register numbers.
28391 (UNSPEC_REV, UNSPEC_LD1_SVE, UNSPEC_ST1_SVE, UNSPEC_MERGE_PTRUE)
28392 (UNSPEC_PTEST_PTRUE, UNSPEC_UNPACKSHI, UNSPEC_UNPACKUHI)
28393 (UNSPEC_UNPACKSLO, UNSPEC_UNPACKULO, UNSPEC_PACK)
28394 (UNSPEC_FLOAT_CONVERT, UNSPEC_WHILE_LO): New unspec constants.
28395 (sve): New attribute.
28396 (enabled): Disable instructions with the sve attribute unless
28398 (movqi, movhi): Pass CONST_POLY_INT operaneds through
28399 aarch64_expand_mov_immediate.
28400 (*mov<mode>_aarch64, *movsi_aarch64, *movdi_aarch64): Handle
28401 CNT[BHSD] immediates.
28402 (movti): Split CONST_POLY_INT moves into two halves.
28403 (add<mode>3): Accept aarch64_pluslong_or_poly_operand.
28404 Split additions that need a temporary here if the destination
28405 is the stack pointer.
28406 (*add<mode>3_aarch64): Handle ADDVL and ADDPL immediates.
28407 (*add<mode>3_poly_1): New instruction.
28408 (set_clobber_cc): New expander.
28410 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
28412 * simplify-rtx.c (simplify_immed_subreg): Add an inner_bytes
28413 parameter and use it instead of GET_MODE_SIZE (innermode). Use
28414 inner_bytes * BITS_PER_UNIT instead of GET_MODE_BITSIZE (innermode).
28415 Use CEIL (inner_bytes, GET_MODE_UNIT_SIZE (innermode)) instead of
28416 GET_MODE_NUNITS (innermode). Also add a first_elem parameter.
28417 Change innermode from fixed_mode_size to machine_mode.
28418 (simplify_subreg): Update call accordingly. Handle a constant-sized
28419 subreg of a variable-length CONST_VECTOR.
28421 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
28422 Alan Hayward <alan.hayward@arm.com>
28423 David Sherwood <david.sherwood@arm.com>
28425 * tree-ssa-address.c (mem_ref_valid_without_offset_p): New function.
28426 (add_offset_to_base): New function, split out from...
28427 (create_mem_ref): ...here. When handling a scale other than 1,
28428 check first whether the address is valid without the offset.
28429 Add it into the base if so, leaving the index and scale as-is.
28431 2018-01-12 Jakub Jelinek <jakub@redhat.com>
28434 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Call
28435 fold_for_warn before checking if arg2 is INTEGER_CST.
28437 2018-01-12 Segher Boessenkool <segher@kernel.crashing.org>
28439 * config/rs6000/predicates.md (load_multiple_operation): Delete.
28440 (store_multiple_operation): Delete.
28441 * config/rs6000/rs6000-cpus.def (601): Remove MASK_STRING.
28442 * config/rs6000/rs6000-protos.h (rs6000_output_load_multiple): Delete.
28443 * config/rs6000/rs6000-string.c (expand_block_move): Delete everything
28444 guarded by TARGET_STRING.
28445 (rs6000_output_load_multiple): Delete.
28446 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
28447 OPTION_MASK_STRING / TARGET_STRING handling.
28448 (print_operand) <'N', 'O'>: Add comment that these are unused now.
28449 (const rs6000_opt_masks) <"string">: Change mask to 0.
28450 * config/rs6000/rs6000.h (TARGET_DEFAULT): Remove MASK_STRING.
28451 (MASK_STRING): Delete.
28452 * config/rs6000/rs6000.md (*mov<mode>_string): Delete TARGET_STRING
28454 (load_multiple): Delete.
28461 (store_multiple): Delete.
28468 (movmemsi_8reg): Delete.
28469 (corresponding unnamed define_insn): Delete.
28470 (movmemsi_6reg): Delete.
28471 (corresponding unnamed define_insn): Delete.
28472 (movmemsi_4reg): Delete.
28473 (corresponding unnamed define_insn): Delete.
28474 (movmemsi_2reg): Delete.
28475 (corresponding unnamed define_insn): Delete.
28476 (movmemsi_1reg): Delete.
28477 (corresponding unnamed define_insn): Delete.
28478 * config/rs6000/rs6000.opt (mno-string): New.
28479 (mstring): Replace by deprecation warning stub.
28480 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mstring.
28482 2018-01-12 Jakub Jelinek <jakub@redhat.com>
28484 * regrename.c (regrename_do_replace): If replacing the same
28485 reg multiple times, try to reuse last created gen_raw_REG.
28488 * bb-reorder.c (pass_partition_blocks::gate): In lto don't partition
28489 main to workaround a bug in GDB.
28491 2018-01-12 Tom de Vries <tom@codesourcery.com>
28494 * config.gcc (nvptx*-*-*): Set use_gcc_stdint=wrap.
28496 2018-01-12 Vladimir Makarov <vmakarov@redhat.com>
28498 PR rtl-optimization/80481
28499 * ira-color.c (get_cap_member): New function.
28500 (allocnos_conflict_by_live_ranges_p): Use it.
28501 (slot_coalesced_allocno_live_ranges_intersect_p): Add assert.
28502 (setup_slot_coalesced_allocno_live_ranges): Ditto.
28504 2018-01-12 Uros Bizjak <ubizjak@gmail.com>
28507 * config/alpha/alpha.md (*saddsi_1): New insn_ans_split pattern.
28508 (*saddl_se_1): Ditto.
28509 (*ssubsi_1): Ditto.
28510 (*ssubl_se_1): Ditto.
28512 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
28514 * tree-predcom.c (aff_combination_dr_offset): Use wi::to_poly_widest
28515 rather than wi::to_widest for DR_INITs.
28516 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Use
28517 wi::to_poly_offset rather than wi::to_offset for DR_INIT.
28518 (vect_analyze_data_ref_accesses): Require both DR_INITs to be
28520 (vect_analyze_group_access_1): Note that here.
28522 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
28524 * tree-vectorizer.c (get_vec_alignment_for_array_type): Handle
28525 polynomial type sizes.
28527 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
28529 * gimplify.c (gimple_add_tmp_var_fn): Allow variables to have a
28530 poly_uint64 size, rather than requiring an unsigned HOST_WIDE_INT size.
28531 (gimple_add_tmp_var): Likewise.
28533 2018-01-12 Martin Liska <mliska@suse.cz>
28535 * gimple.c (gimple_alloc_counts): Use uint64_t instead of int.
28536 (gimple_alloc_sizes): Likewise.
28537 (dump_gimple_statistics): Use PRIu64 in printf format.
28538 * gimple.h: Change uint64_t to int.
28540 2018-01-12 Martin Liska <mliska@suse.cz>
28542 * tree-core.h: Use uint64_t instead of int.
28543 * tree.c (tree_node_counts): Likewise.
28544 (tree_node_sizes): Likewise.
28545 (dump_tree_statistics): Use PRIu64 in printf format.
28547 2018-01-12 Martin Liska <mliska@suse.cz>
28549 * Makefile.in: As qsort_chk is implemented in vec.c, add
28550 vec.o to linkage of gencfn-macros.
28551 * tree.c (build_new_poly_int_cst): Add CXX_MEM_STAT_INFO as it's
28552 passing the info to record_node_allocation_statistics.
28553 (test_vector_cst_patterns): Add CXX_MEM_STAT_INFO to declaration
28555 * ggc-common.c (struct ggc_usage): Add operator== and use
28556 it in operator< and compare function.
28557 * mem-stats.h (struct mem_usage): Likewise.
28558 * vec.c (struct vec_usage): Remove operator< and compare
28559 function. Can be simply inherited.
28561 2018-01-12 Martin Jambor <mjambor@suse.cz>
28564 * params.def: New parameter PARAM_AVOID_FMA_MAX_BITS.
28565 * tree-ssa-math-opts.c: Include domwalk.h.
28566 (convert_mult_to_fma_1): New function.
28567 (fma_transformation_info): New type.
28568 (fma_deferring_state): Likewise.
28569 (cancel_fma_deferring): New function.
28570 (result_of_phi): Likewise.
28571 (last_fma_candidate_feeds_initial_phi): Likewise.
28572 (convert_mult_to_fma): Added deferring logic, split actual
28573 transformation to convert_mult_to_fma_1.
28574 (math_opts_dom_walker): New type.
28575 (math_opts_dom_walker::after_dom_children): New method, body moved
28576 here from pass_optimize_widening_mul::execute, added deferring logic
28578 (pass_optimize_widening_mul::execute): Moved most of code to
28579 math_opts_dom_walker::after_dom_children.
28580 * config/i386/x86-tune.def (X86_TUNE_AVOID_128FMA_CHAINS): New.
28581 * config/i386/i386.c (ix86_option_override_internal): Added
28582 maybe_setting of PARAM_AVOID_FMA_MAX_BITS.
28584 2018-01-12 Richard Biener <rguenther@suse.de>
28587 * dwarf2out.c (gen_variable_die): Do not reset old_die for
28588 inline instance vars.
28590 2018-01-12 Oleg Endo <olegendo@gcc.gnu.org>
28593 * config/rx/rx.c (rx_is_restricted_memory_address):
28594 Handle SUBREG case.
28596 2018-01-12 Richard Biener <rguenther@suse.de>
28598 PR tree-optimization/80846
28599 * target.def (split_reduction): New target hook.
28600 * targhooks.c (default_split_reduction): New function.
28601 * targhooks.h (default_split_reduction): Declare.
28602 * tree-vect-loop.c (vect_create_epilog_for_reduction): If the
28603 target requests first reduce vectors by combining low and high
28605 * tree-vect-stmts.c (vect_gen_perm_mask_any): Adjust.
28606 (get_vectype_for_scalar_type_and_size): Export.
28607 * tree-vectorizer.h (get_vectype_for_scalar_type_and_size): Declare.
28608 * doc/tm.texi.in (TARGET_VECTORIZE_SPLIT_REDUCTION): Document.
28609 * doc/tm.texi: Regenerate.
28610 * config/i386/i386.c (ix86_split_reduction): Implement
28611 TARGET_VECTORIZE_SPLIT_REDUCTION.
28613 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
28616 * config/sparc/sparc.h (PIC_OFFSET_TABLE_REGNUM): Set to INVALID_REGNUM
28617 in PIC mode except for TARGET_VXWORKS_RTP.
28618 * config/sparc/sparc.c: Include cfgrtl.h.
28619 (TARGET_INIT_PIC_REG): Define.
28620 (TARGET_USE_PSEUDO_PIC_REG): Likewise.
28621 (sparc_pic_register_p): New predicate.
28622 (sparc_legitimate_address_p): Use it.
28623 (sparc_legitimize_pic_address): Likewise.
28624 (sparc_delegitimize_address): Likewise.
28625 (sparc_mode_dependent_address_p): Likewise.
28626 (gen_load_pcrel_sym): Remove 4th parameter.
28627 (load_got_register): Adjust call to above. Remove obsolete stuff.
28628 (sparc_expand_prologue): Do not call load_got_register here.
28629 (sparc_flat_expand_prologue): Likewise.
28630 (sparc_output_mi_thunk): Set the pic_offset_table_rtx object.
28631 (sparc_use_pseudo_pic_reg): New function.
28632 (sparc_init_pic_reg): Likewise.
28633 * config/sparc/sparc.md (vxworks_load_got): Set the GOT register.
28634 (builtin_setjmp_receiver): Enable only for TARGET_VXWORKS_RTP.
28636 2018-01-12 Christophe Lyon <christophe.lyon@linaro.org>
28638 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
28639 Add item for branch_cost.
28641 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
28643 PR rtl-optimization/83565
28644 * rtlanal.c (nonzero_bits1): On WORD_REGISTER_OPERATIONS machines, do
28645 not extend the result to a larger mode for rotate operations.
28646 (num_sign_bit_copies1): Likewise.
28648 2018-01-12 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
28651 * config/sol2.h (STARTFILE_ARCH_SPEC): Don't use with -shared or
28653 Use values-Xc.o for -pedantic.
28654 Link with values-xpg4.o for C90, values-xpg6.o otherwise.
28656 2018-01-12 Martin Liska <mliska@suse.cz>
28659 * ipa-devirt.c (final_warning_record::grow_type_warnings):
28661 (possible_polymorphic_call_targets): Use it.
28662 (ipa_devirt): Likewise.
28664 2018-01-12 Martin Liska <mliska@suse.cz>
28666 * profile-count.h (enum profile_quality): Use 0 as invalid
28667 enum value of profile_quality.
28669 2018-01-12 Chung-Ju Wu <jasonwucj@gmail.com>
28671 * doc/invoke.texi (NDS32 Options): Add -mext-perf, -mext-perf2 and
28672 -mext-string options.
28674 2018-01-12 Richard Biener <rguenther@suse.de>
28676 * lto-streamer-out.c (DFS::DFS_write_tree_body): Process
28677 DECL_DEBUG_EXPR conditional on DECL_HAS_DEBUG_EXPR_P.
28678 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
28680 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
28682 2018-01-11 Michael Meissner <meissner@linux.vnet.ibm.com>
28684 * configure.ac (--with-long-double-format): Add support for the
28685 configuration option to change the default long double format on
28687 * config.gcc (powerpc*-linux*-*): Likewise.
28688 * configure: Regenerate.
28689 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If long
28690 double is IEEE, define __KC__ and __KF__ to allow floatn.h to be
28691 used without modification.
28693 2018-01-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28695 * config/rs6000/rs6000-builtin.def (BU_P7_MISC_X): New #define.
28696 (SPEC_BARRIER): New instantiation of BU_P7_MISC_X.
28697 * config/rs6000/rs6000.c (rs6000_expand_builtin): Handle
28698 MISC_BUILTIN_SPEC_BARRIER.
28699 (rs6000_init_builtins): Likewise.
28700 * config/rs6000/rs6000.md (UNSPECV_SPEC_BARRIER): New UNSPECV
28702 (speculation_barrier): New define_insn.
28703 * doc/extend.texi: Document __builtin_speculation_barrier.
28705 2018-01-11 Jakub Jelinek <jakub@redhat.com>
28708 * config/i386/i386.c (ix86_expand_vector_init_one_nonzero): If one_var
28709 is 0, for V{8,16}S[IF] and V[48]D[IF]mode use gen_vec_set<mode>_0.
28710 * config/i386/sse.md (VI8_AVX_AVX512F, VI4F_256_512): New mode
28712 (ssescalarmodesuffix): Add 512-bit vectors. Use "d" or "q" for
28713 integral modes instead of "ss" and "sd".
28714 (vec_set<mode>_0): New define_insns for 256-bit and 512-bit
28715 vectors with 32-bit and 64-bit elements.
28716 (vecdupssescalarmodesuffix): New mode attribute.
28717 (vec_dup<mode>): Use it.
28719 2018-01-11 H.J. Lu <hongjiu.lu@intel.com>
28722 * config/i386/i386.c (ix86_compute_frame_layout): Align stack
28723 frame if argument is passed on stack.
28725 2018-01-11 Jakub Jelinek <jakub@redhat.com>
28728 * ree.c (combine_reaching_defs): Optimize also
28729 reg2=exp; reg1=reg2; reg2=any_extend(reg1); into
28730 reg2=any_extend(exp); reg1=reg2;, formatting fix.
28732 2018-01-11 Jan Hubicka <hubicka@ucw.cz>
28734 PR middle-end/83189
28735 * gimple-ssa-isolate-paths.c (isolate_path): Fix profile update.
28737 2018-01-11 Jan Hubicka <hubicka@ucw.cz>
28739 PR middle-end/83718
28740 * tree-inline.c (copy_cfg_body): Adjust num&den for scaling
28741 after they are computed.
28743 2018-01-11 Bin Cheng <bin.cheng@arm.com>
28745 PR tree-optimization/83695
28746 * gimple-loop-linterchange.cc
28747 (tree_loop_interchange::interchange_loops): Call scev_reset_htab to
28748 reset cached scev information after interchange.
28749 (pass_linterchange::execute): Remove call to scev_reset_htab.
28751 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28753 * config/arm/arm_neon.h (vfmlal_lane_low_u32, vfmlal_lane_high_u32,
28754 vfmlalq_laneq_low_u32, vfmlalq_lane_low_u32, vfmlal_laneq_low_u32,
28755 vfmlalq_laneq_high_u32, vfmlalq_lane_high_u32, vfmlal_laneq_high_u32,
28756 vfmlsl_lane_low_u32, vfmlsl_lane_high_u32, vfmlslq_laneq_low_u32,
28757 vfmlslq_lane_low_u32, vfmlsl_laneq_low_u32, vfmlslq_laneq_high_u32,
28758 vfmlslq_lane_high_u32, vfmlsl_laneq_high_u32): Define.
28759 * config/arm/arm_neon_builtins.def (vfmal_lane_low,
28760 vfmal_lane_lowv4hf, vfmal_lane_lowv8hf, vfmal_lane_high,
28761 vfmal_lane_highv4hf, vfmal_lane_highv8hf, vfmsl_lane_low,
28762 vfmsl_lane_lowv4hf, vfmsl_lane_lowv8hf, vfmsl_lane_high,
28763 vfmsl_lane_highv4hf, vfmsl_lane_highv8hf): New sets of builtins.
28764 * config/arm/iterators.md (VFMLSEL2, vfmlsel2): New mode attributes.
28765 (V_lane_reg): Likewise.
28766 * config/arm/neon.md (neon_vfm<vfml_op>l_lane_<vfml_half><VCVTF:mode>):
28768 (neon_vfm<vfml_op>l_lane_<vfml_half><vfmlsel2><mode>): Likewise.
28769 (vfmal_lane_low<mode>_intrinsic,
28770 vfmal_lane_low<vfmlsel2><mode>_intrinsic,
28771 vfmal_lane_high<vfmlsel2><mode>_intrinsic,
28772 vfmal_lane_high<mode>_intrinsic, vfmsl_lane_low<mode>_intrinsic,
28773 vfmsl_lane_low<vfmlsel2><mode>_intrinsic,
28774 vfmsl_lane_high<vfmlsel2><mode>_intrinsic,
28775 vfmsl_lane_high<mode>_intrinsic): New define_insns.
28777 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28779 * config/arm/arm-cpus.in (fp16fml): New feature.
28780 (ALL_SIMD): Add fp16fml.
28781 (armv8.2-a): Add fp16fml as an option.
28782 (armv8.3-a): Likewise.
28783 (armv8.4-a): Add fp16fml as part of fp16.
28784 * config/arm/arm.h (TARGET_FP16FML): Define.
28785 * config/arm/arm-c.c (arm_cpu_builtins): Define __ARM_FEATURE_FP16_FML
28787 * config/arm/arm-modes.def (V2HF): Define.
28788 * config/arm/arm_neon.h (vfmlal_low_u32, vfmlsl_low_u32,
28789 vfmlal_high_u32, vfmlsl_high_u32, vfmlalq_low_u32,
28790 vfmlslq_low_u32, vfmlalq_high_u32, vfmlslq_high_u32): Define.
28791 * config/arm/arm_neon_builtins.def (vfmal_low, vfmal_high,
28792 vfmsl_low, vfmsl_high): New set of builtins.
28793 * config/arm/iterators.md (PLUSMINUS): New code iterator.
28794 (vfml_op): New code attribute.
28795 (VFMLHALVES): New int iterator.
28796 (VFML, VFMLSEL): New mode attributes.
28797 (V_reg): Define mapping for V2HF.
28798 (V_hi, V_lo): New mode attributes.
28799 (VF_constraint): Likewise.
28800 (vfml_half, vfml_half_selector): New int attributes.
28801 * config/arm/neon.md (neon_vfm<vfml_op>l_<vfml_half><mode>): New
28803 (vfmal_low<mode>_intrinsic, vfmsl_high<mode>_intrinsic,
28804 vfmal_high<mode>_intrinsic, vfmsl_low<mode>_intrinsic):
28806 * config/arm/t-arm-elf (v8_fps): Add fp16fml.
28807 * config/arm/t-multilib (v8_2_a_simd_variants): Add fp16fml.
28808 * config/arm/unspecs.md (UNSPEC_VFML_LO, UNSPEC_VFML_HI): New unspecs.
28809 * doc/invoke.texi (ARM Options): Document fp16fml. Update armv8.4-a
28811 * doc/sourcebuild.texi (arm_fp16fml_neon_ok, arm_fp16fml_neon):
28812 Document new effective target and option set.
28814 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28816 * config/arm/arm-cpus.in (armv8_4): New feature.
28817 (ARMv8_4a): New fgroup.
28818 (armv8.4-a): New arch.
28819 * config/arm/arm-tables.opt: Regenerate.
28820 * config/arm/t-aprofile: Add matching rules for -march=armv8.4-a.
28821 * config/arm/t-arm-elf (all_v8_archs): Add armv8.4-a.
28822 * config/arm/t-multilib (v8_4_a_simd_variants): New variable.
28823 Add matching rules for -march=armv8.4-a and extensions.
28824 * doc/invoke.texi (ARM Options): Document -march=armv8.4-a.
28826 2018-01-11 Oleg Endo <olegendo@gcc.gnu.org>
28829 * config/rx/rx.md (BW): New mode attribute.
28830 (sync_lock_test_and_setsi): Add mode suffix to insn output.
28832 2018-01-11 Richard Biener <rguenther@suse.de>
28834 PR tree-optimization/83435
28835 * graphite.c (canonicalize_loop_form): Ignore fake loop exit edges.
28836 * graphite-scop-detection.c (scop_detection::get_sese): Likewise.
28837 * tree-vrp.c (add_assert_info): Drop TREE_OVERFLOW if they appear.
28839 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
28840 Alan Hayward <alan.hayward@arm.com>
28841 David Sherwood <david.sherwood@arm.com>
28843 * config/aarch64/aarch64.c (aarch64_address_info): Add a const_offset
28845 (aarch64_classify_address): Initialize it. Track polynomial offsets.
28846 (aarch64_print_address_internal): Use it to check for a zero offset.
28848 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
28849 Alan Hayward <alan.hayward@arm.com>
28850 David Sherwood <david.sherwood@arm.com>
28852 * config/aarch64/aarch64-modes.def (NUM_POLY_INT_COEFFS): Set to 2.
28853 * config/aarch64/aarch64-protos.h (aarch64_initial_elimination_offset):
28854 Return a poly_int64 rather than a HOST_WIDE_INT.
28855 (aarch64_offset_7bit_signed_scaled_p): Take the offset as a poly_int64
28856 rather than a HOST_WIDE_INT.
28857 * config/aarch64/aarch64.h (aarch64_frame): Protect with
28858 HAVE_POLY_INT_H rather than HOST_WIDE_INT. Change locals_offset,
28859 hard_fp_offset, frame_size, initial_adjust, callee_offset and
28860 final_offset from HOST_WIDE_INT to poly_int64.
28861 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args): Use
28862 to_constant when getting the number of units in an Advanced SIMD
28864 (aarch64_builtin_vectorized_function): Check for a constant number
28866 * config/aarch64/aarch64-simd.md (mov<mode>): Handle polynomial
28868 (aarch64_ld<VSTRUCT:nregs>_lane<VALLDIF:mode>): Use the nunits
28869 attribute instead of GET_MODE_NUNITS.
28870 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs)
28871 (aarch64_class_max_nregs): Use the constant_lowest_bound of the
28872 GET_MODE_SIZE for fixed-size registers.
28873 (aarch64_const_vec_all_same_in_range_p): Use const_vec_duplicate_p.
28874 (aarch64_hard_regno_call_part_clobbered, aarch64_classify_index)
28875 (aarch64_mode_valid_for_sched_fusion_p, aarch64_classify_address)
28876 (aarch64_legitimize_address_displacement, aarch64_secondary_reload)
28877 (aarch64_print_operand, aarch64_print_address_internal)
28878 (aarch64_address_cost, aarch64_rtx_costs, aarch64_register_move_cost)
28879 (aarch64_short_vector_p, aapcs_vfp_sub_candidate)
28880 (aarch64_simd_attr_length_rglist, aarch64_operands_ok_for_ldpstp):
28881 Handle polynomial GET_MODE_SIZE.
28882 (aarch64_hard_regno_caller_save_mode): Likewise. Return modes
28883 wider than SImode without modification.
28884 (tls_symbolic_operand_type): Use strip_offset instead of split_const.
28885 (aarch64_pass_by_reference, aarch64_layout_arg, aarch64_pad_reg_upward)
28886 (aarch64_gimplify_va_arg_expr): Assert that we don't yet handle
28887 passing and returning SVE modes.
28888 (aarch64_function_value, aarch64_layout_arg): Use gen_int_mode
28889 rather than GEN_INT.
28890 (aarch64_emit_probe_stack_range): Take the size as a poly_int64
28891 rather than a HOST_WIDE_INT, but call sorry if it isn't constant.
28892 (aarch64_allocate_and_probe_stack_space): Likewise.
28893 (aarch64_layout_frame): Cope with polynomial offsets.
28894 (aarch64_save_callee_saves, aarch64_restore_callee_saves): Take the
28895 start_offset as a poly_int64 rather than a HOST_WIDE_INT. Track
28896 polynomial offsets.
28897 (offset_9bit_signed_unscaled_p, offset_12bit_unsigned_scaled_p)
28898 (aarch64_offset_7bit_signed_scaled_p): Take the offset as a
28899 poly_int64 rather than a HOST_WIDE_INT.
28900 (aarch64_get_separate_components, aarch64_process_components)
28901 (aarch64_expand_prologue, aarch64_expand_epilogue)
28902 (aarch64_use_return_insn_p): Handle polynomial frame offsets.
28903 (aarch64_anchor_offset): New function, split out from...
28904 (aarch64_legitimize_address): ...here.
28905 (aarch64_builtin_vectorization_cost): Handle polynomial
28906 TYPE_VECTOR_SUBPARTS.
28907 (aarch64_simd_check_vect_par_cnst_half): Handle polynomial
28909 (aarch64_simd_make_constant, aarch64_expand_vector_init): Get the
28910 number of elements from the PARALLEL rather than the mode.
28911 (aarch64_shift_truncation_mask): Use GET_MODE_UNIT_BITSIZE
28912 rather than GET_MODE_BITSIZE.
28913 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_ext)
28914 (aarch64_evpc_rev, aarch64_evpc_dup, aarch64_evpc_zip)
28915 (aarch64_expand_vec_perm_const_1): Handle polynomial
28916 d->perm.length () and d->perm elements.
28917 (aarch64_evpc_tbl): Likewise. Use nelt rather than GET_MODE_NUNITS.
28918 Apply to_constant to d->perm elements.
28919 (aarch64_simd_valid_immediate, aarch64_vec_fpconst_pow_of_2): Handle
28920 polynomial CONST_VECTOR_NUNITS.
28921 (aarch64_move_pointer): Take amount as a poly_int64 rather
28923 (aarch64_progress_pointer): Avoid temporary variable.
28924 * config/aarch64/aarch64.md (aarch64_<crc_variant>): Use
28925 the mode attribute instead of GET_MODE.
28927 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
28928 Alan Hayward <alan.hayward@arm.com>
28929 David Sherwood <david.sherwood@arm.com>
28931 * config/aarch64/aarch64.c (aarch64_force_temporary): Assert that
28932 x exists before using it.
28933 (aarch64_add_constant_internal): Rename to...
28934 (aarch64_add_offset_1): ...this. Replace regnum with separate
28935 src and dest rtxes. Handle the case in which they're different,
28936 including when the offset is zero. Replace scratchreg with an rtx.
28937 Use 2 additions if there is no spare register into which we can
28938 move a 16-bit constant.
28939 (aarch64_add_constant): Delete.
28940 (aarch64_add_offset): Replace reg with separate src and dest
28941 rtxes. Take a poly_int64 offset instead of a HOST_WIDE_INT.
28942 Use aarch64_add_offset_1.
28943 (aarch64_add_sp, aarch64_sub_sp): Take the scratch register as
28944 an rtx rather than an int. Take the delta as a poly_int64
28945 rather than a HOST_WIDE_INT. Use aarch64_add_offset.
28946 (aarch64_expand_mov_immediate): Update uses of aarch64_add_offset.
28947 (aarch64_expand_prologue): Update calls to aarch64_sub_sp,
28948 aarch64_allocate_and_probe_stack_space and aarch64_add_offset.
28949 (aarch64_expand_epilogue): Update calls to aarch64_add_offset
28950 and aarch64_add_sp.
28951 (aarch64_output_mi_thunk): Use aarch64_add_offset rather than
28952 aarch64_add_constant.
28954 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
28956 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
28957 Use scalar_float_mode.
28959 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
28961 * config/aarch64/aarch64-simd.md
28962 (aarch64_fml<f16mac1>l<f16quad>_low<mode>): Avoid GET_MODE_NUNITS.
28963 (aarch64_fml<f16mac1>l<f16quad>_high<mode>): Likewise.
28964 (aarch64_fml<f16mac1>l_lane_lowv2sf): Likewise.
28965 (aarch64_fml<f16mac1>l_lane_highv2sf): Likewise.
28966 (aarch64_fml<f16mac1>lq_laneq_lowv4sf): Likewise.
28967 (aarch64_fml<f16mac1>lq_laneq_highv4sf): Likewise.
28968 (aarch64_fml<f16mac1>l_laneq_lowv2sf): Likewise.
28969 (aarch64_fml<f16mac1>l_laneq_highv2sf): Likewise.
28970 (aarch64_fml<f16mac1>lq_lane_lowv4sf): Likewise.
28971 (aarch64_fml<f16mac1>lq_lane_highv4sf): Likewise.
28973 2018-01-11 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
28976 * config/arm/arm.c (arm_declare_function_name): Set arch_to_print if
28977 targ_options->x_arm_arch_string is non NULL.
28979 2018-01-11 Tamar Christina <tamar.christina@arm.com>
28981 * config/aarch64/aarch64.h
28982 (AARCH64_FL_FOR_ARCH8_4): Add AARCH64_FL_DOTPROD.
28984 2018-01-11 Sudakshina Das <sudi.das@arm.com>
28987 * expmed.c (emit_store_flag_force): Swap if const op0
28988 and change VOIDmode to mode of op0.
28990 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
28992 PR rtl-optimization/83761
28993 * caller-save.c (replace_reg_with_saved_mem): Pass bits rather
28994 than bytes to mode_for_size.
28996 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
28998 PR middle-end/83189
28999 * gfortran.fortran-torture/compile/pr83189.f90: New testcase.
29000 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Handle zero
29003 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
29005 PR middle-end/83575
29006 * cfgrtl.c (rtl_verify_edges): Only verify fixability of partition
29007 when in layout mode.
29008 (cfg_layout_finalize): Do not verify cfg before we are out of layout.
29009 * cfgcleanup.c (try_optimize_cfg): Only verify flow info when doing
29012 2018-01-10 Michael Collison <michael.collison@arm.com>
29014 * config/aarch64/aarch64-modes.def (V2HF): New VECTOR_MODE.
29015 * config/aarch64/aarch64-option-extension.def: Add
29016 AARCH64_OPT_EXTENSION of 'fp16fml'.
29017 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
29018 (__ARM_FEATURE_FP16_FML): Define if TARGET_F16FML is true.
29019 * config/aarch64/predicates.md (aarch64_lane_imm3): New predicate.
29020 * config/aarch64/constraints.md (Ui7): New constraint.
29021 * config/aarch64/iterators.md (VFMLA_W): New mode iterator.
29022 (VFMLA_SEL_W): Ditto.
29025 (VFMLA16_LOW): New int iterator.
29026 (VFMLA16_HIGH): Ditto.
29027 (UNSPEC_FMLAL): New unspec.
29028 (UNSPEC_FMLSL): Ditto.
29029 (UNSPEC_FMLAL2): Ditto.
29030 (UNSPEC_FMLSL2): Ditto.
29031 (f16mac): New code attribute.
29032 * config/aarch64/aarch64-simd-builtins.def
29033 (aarch64_fmlal_lowv2sf): Ditto.
29034 (aarch64_fmlsl_lowv2sf): Ditto.
29035 (aarch64_fmlalq_lowv4sf): Ditto.
29036 (aarch64_fmlslq_lowv4sf): Ditto.
29037 (aarch64_fmlal_highv2sf): Ditto.
29038 (aarch64_fmlsl_highv2sf): Ditto.
29039 (aarch64_fmlalq_highv4sf): Ditto.
29040 (aarch64_fmlslq_highv4sf): Ditto.
29041 (aarch64_fmlal_lane_lowv2sf): Ditto.
29042 (aarch64_fmlsl_lane_lowv2sf): Ditto.
29043 (aarch64_fmlal_laneq_lowv2sf): Ditto.
29044 (aarch64_fmlsl_laneq_lowv2sf): Ditto.
29045 (aarch64_fmlalq_lane_lowv4sf): Ditto.
29046 (aarch64_fmlsl_lane_lowv4sf): Ditto.
29047 (aarch64_fmlalq_laneq_lowv4sf): Ditto.
29048 (aarch64_fmlsl_laneq_lowv4sf): Ditto.
29049 (aarch64_fmlal_lane_highv2sf): Ditto.
29050 (aarch64_fmlsl_lane_highv2sf): Ditto.
29051 (aarch64_fmlal_laneq_highv2sf): Ditto.
29052 (aarch64_fmlsl_laneq_highv2sf): Ditto.
29053 (aarch64_fmlalq_lane_highv4sf): Ditto.
29054 (aarch64_fmlsl_lane_highv4sf): Ditto.
29055 (aarch64_fmlalq_laneq_highv4sf): Ditto.
29056 (aarch64_fmlsl_laneq_highv4sf): Ditto.
29057 * config/aarch64/aarch64-simd.md:
29058 (aarch64_fml<f16mac1>l<f16quad>_low<mode>): New pattern.
29059 (aarch64_fml<f16mac1>l<f16quad>_high<mode>): Ditto.
29060 (aarch64_simd_fml<f16mac1>l<f16quad>_low<mode>): Ditto.
29061 (aarch64_simd_fml<f16mac1>l<f16quad>_high<mode>): Ditto.
29062 (aarch64_fml<f16mac1>l_lane_lowv2sf): Ditto.
29063 (aarch64_fml<f16mac1>l_lane_highv2sf): Ditto.
29064 (aarch64_simd_fml<f16mac>l_lane_lowv2sf): Ditto.
29065 (aarch64_simd_fml<f16mac>l_lane_highv2sf): Ditto.
29066 (aarch64_fml<f16mac1>lq_laneq_lowv4sf): Ditto.
29067 (aarch64_fml<f16mac1>lq_laneq_highv4sf): Ditto.
29068 (aarch64_simd_fml<f16mac>lq_laneq_lowv4sf): Ditto.
29069 (aarch64_simd_fml<f16mac>lq_laneq_highv4sf): Ditto.
29070 (aarch64_fml<f16mac1>l_laneq_lowv2sf): Ditto.
29071 (aarch64_fml<f16mac1>l_laneq_highv2sf): Ditto.
29072 (aarch64_simd_fml<f16mac>l_laneq_lowv2sf): Ditto.
29073 (aarch64_simd_fml<f16mac>l_laneq_highv2sf): Ditto.
29074 (aarch64_fml<f16mac1>lq_lane_lowv4sf): Ditto.
29075 (aarch64_fml<f16mac1>lq_lane_highv4sf): Ditto.
29076 (aarch64_simd_fml<f16mac>lq_lane_lowv4sf): Ditto.
29077 (aarch64_simd_fml<f16mac>lq_lane_highv4sf): Ditto.
29078 * config/aarch64/arm_neon.h (vfmlal_low_u32): New intrinsic.
29079 (vfmlsl_low_u32): Ditto.
29080 (vfmlalq_low_u32): Ditto.
29081 (vfmlslq_low_u32): Ditto.
29082 (vfmlal_high_u32): Ditto.
29083 (vfmlsl_high_u32): Ditto.
29084 (vfmlalq_high_u32): Ditto.
29085 (vfmlslq_high_u32): Ditto.
29086 (vfmlal_lane_low_u32): Ditto.
29087 (vfmlsl_lane_low_u32): Ditto.
29088 (vfmlal_laneq_low_u32): Ditto.
29089 (vfmlsl_laneq_low_u32): Ditto.
29090 (vfmlalq_lane_low_u32): Ditto.
29091 (vfmlslq_lane_low_u32): Ditto.
29092 (vfmlalq_laneq_low_u32): Ditto.
29093 (vfmlslq_laneq_low_u32): Ditto.
29094 (vfmlal_lane_high_u32): Ditto.
29095 (vfmlsl_lane_high_u32): Ditto.
29096 (vfmlal_laneq_high_u32): Ditto.
29097 (vfmlsl_laneq_high_u32): Ditto.
29098 (vfmlalq_lane_high_u32): Ditto.
29099 (vfmlslq_lane_high_u32): Ditto.
29100 (vfmlalq_laneq_high_u32): Ditto.
29101 (vfmlslq_laneq_high_u32): Ditto.
29102 * config/aarch64/aarch64.h (AARCH64_FL_F16SML): New flag.
29103 (AARCH64_FL_FOR_ARCH8_4): New.
29104 (AARCH64_ISA_F16FML): New ISA flag.
29105 (TARGET_F16FML): New feature flag for fp16fml.
29106 (doc/invoke.texi): Document new fp16fml option.
29108 2018-01-10 Michael Collison <michael.collison@arm.com>
29110 * config/aarch64/aarch64-builtins.c:
29111 (aarch64_types_ternopu_imm_qualifiers, TYPES_TERNOPUI): New.
29112 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
29113 (__ARM_FEATURE_SHA3): Define if TARGET_SHA3 is true.
29114 * config/aarch64/aarch64.h (AARCH64_FL_SHA3): New flags.
29115 (AARCH64_ISA_SHA3): New ISA flag.
29116 (TARGET_SHA3): New feature flag for sha3.
29117 * config/aarch64/iterators.md (sha512_op): New int attribute.
29118 (CRYPTO_SHA512): New int iterator.
29119 (UNSPEC_SHA512H): New unspec.
29120 (UNSPEC_SHA512H2): Ditto.
29121 (UNSPEC_SHA512SU0): Ditto.
29122 (UNSPEC_SHA512SU1): Ditto.
29123 * config/aarch64/aarch64-simd-builtins.def
29124 (aarch64_crypto_sha512hqv2di): New builtin.
29125 (aarch64_crypto_sha512h2qv2di): Ditto.
29126 (aarch64_crypto_sha512su0qv2di): Ditto.
29127 (aarch64_crypto_sha512su1qv2di): Ditto.
29128 (aarch64_eor3qv8hi): Ditto.
29129 (aarch64_rax1qv2di): Ditto.
29130 (aarch64_xarqv2di): Ditto.
29131 (aarch64_bcaxqv8hi): Ditto.
29132 * config/aarch64/aarch64-simd.md:
29133 (aarch64_crypto_sha512h<sha512_op>qv2di): New pattern.
29134 (aarch64_crypto_sha512su0qv2di): Ditto.
29135 (aarch64_crypto_sha512su1qv2di): Ditto.
29136 (aarch64_eor3qv8hi): Ditto.
29137 (aarch64_rax1qv2di): Ditto.
29138 (aarch64_xarqv2di): Ditto.
29139 (aarch64_bcaxqv8hi): Ditto.
29140 * config/aarch64/arm_neon.h (vsha512hq_u64): New intrinsic.
29141 (vsha512h2q_u64): Ditto.
29142 (vsha512su0q_u64): Ditto.
29143 (vsha512su1q_u64): Ditto.
29144 (veor3q_u16): Ditto.
29145 (vrax1q_u64): Ditto.
29146 (vxarq_u64): Ditto.
29147 (vbcaxq_u16): Ditto.
29148 * config/arm/types.md (crypto_sha512): New type attribute.
29149 (crypto_sha3): Ditto.
29150 (doc/invoke.texi): Document new sha3 option.
29152 2018-01-10 Michael Collison <michael.collison@arm.com>
29154 * config/aarch64/aarch64-builtins.c:
29155 (aarch64_types_quadopu_imm_qualifiers, TYPES_QUADOPUI): New.
29156 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
29157 (__ARM_FEATURE_SM3): Define if TARGET_SM4 is true.
29158 (__ARM_FEATURE_SM4): Define if TARGET_SM4 is true.
29159 * config/aarch64/aarch64.h (AARCH64_FL_SM4): New flags.
29160 (AARCH64_ISA_SM4): New ISA flag.
29161 (TARGET_SM4): New feature flag for sm4.
29162 * config/aarch64/aarch64-simd-builtins.def
29163 (aarch64_sm3ss1qv4si): Ditto.
29164 (aarch64_sm3tt1aq4si): Ditto.
29165 (aarch64_sm3tt1bq4si): Ditto.
29166 (aarch64_sm3tt2aq4si): Ditto.
29167 (aarch64_sm3tt2bq4si): Ditto.
29168 (aarch64_sm3partw1qv4si): Ditto.
29169 (aarch64_sm3partw2qv4si): Ditto.
29170 (aarch64_sm4eqv4si): Ditto.
29171 (aarch64_sm4ekeyqv4si): Ditto.
29172 * config/aarch64/aarch64-simd.md:
29173 (aarch64_sm3ss1qv4si): Ditto.
29174 (aarch64_sm3tt<sm3tt_op>qv4si): Ditto.
29175 (aarch64_sm3partw<sm3part_op>qv4si): Ditto.
29176 (aarch64_sm4eqv4si): Ditto.
29177 (aarch64_sm4ekeyqv4si): Ditto.
29178 * config/aarch64/iterators.md (sm3tt_op): New int iterator.
29179 (sm3part_op): Ditto.
29180 (CRYPTO_SM3TT): Ditto.
29181 (CRYPTO_SM3PART): Ditto.
29182 (UNSPEC_SM3SS1): New unspec.
29183 (UNSPEC_SM3TT1A): Ditto.
29184 (UNSPEC_SM3TT1B): Ditto.
29185 (UNSPEC_SM3TT2A): Ditto.
29186 (UNSPEC_SM3TT2B): Ditto.
29187 (UNSPEC_SM3PARTW1): Ditto.
29188 (UNSPEC_SM3PARTW2): Ditto.
29189 (UNSPEC_SM4E): Ditto.
29190 (UNSPEC_SM4EKEY): Ditto.
29191 * config/aarch64/constraints.md (Ui2): New constraint.
29192 * config/aarch64/predicates.md (aarch64_imm2): New predicate.
29193 * config/arm/types.md (crypto_sm3): New type attribute.
29194 (crypto_sm4): Ditto.
29195 * config/aarch64/arm_neon.h (vsm3ss1q_u32): New intrinsic.
29196 (vsm3tt1aq_u32): Ditto.
29197 (vsm3tt1bq_u32): Ditto.
29198 (vsm3tt2aq_u32): Ditto.
29199 (vsm3tt2bq_u32): Ditto.
29200 (vsm3partw1q_u32): Ditto.
29201 (vsm3partw2q_u32): Ditto.
29202 (vsm4eq_u32): Ditto.
29203 (vsm4ekeyq_u32): Ditto.
29204 (doc/invoke.texi): Document new sm4 option.
29206 2018-01-10 Michael Collison <michael.collison@arm.com>
29208 * config/aarch64/aarch64-arches.def (armv8.4-a): New architecture.
29209 * config/aarch64/aarch64.h (AARCH64_ISA_V8_4): New ISA flag.
29210 (AARCH64_FL_FOR_ARCH8_4): New.
29211 (AARCH64_FL_V8_4): New flag.
29212 (doc/invoke.texi): Document new armv8.4-a option.
29214 2018-01-10 Michael Collison <michael.collison@arm.com>
29216 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
29217 (__ARM_FEATURE_AES): Define if TARGET_AES is true.
29218 (__ARM_FEATURE_SHA2): Define if TARGET_SHA2 is true.
29219 * config/aarch64/aarch64-option-extension.def: Add
29220 AARCH64_OPT_EXTENSION of 'sha2'.
29221 (aes): Add AARCH64_OPT_EXTENSION of 'aes'.
29222 (crypto): Disable sha2 and aes if crypto disabled.
29223 (crypto): Enable aes and sha2 if enabled.
29224 (simd): Disable sha2 and aes if simd disabled.
29225 * config/aarch64/aarch64.h (AARCH64_FL_AES, AARCH64_FL_SHA2):
29227 (AARCH64_ISA_AES, AARCH64_ISA_SHA2): New ISA flags.
29228 (TARGET_SHA2): New feature flag for sha2.
29229 (TARGET_AES): New feature flag for aes.
29230 * config/aarch64/aarch64-simd.md:
29231 (aarch64_crypto_aes<aes_op>v16qi): Make pattern
29232 conditional on TARGET_AES.
29233 (aarch64_crypto_aes<aesmc_op>v16qi): Ditto.
29234 (aarch64_crypto_sha1hsi): Make pattern conditional
29236 (aarch64_crypto_sha1hv4si): Ditto.
29237 (aarch64_be_crypto_sha1hv4si): Ditto.
29238 (aarch64_crypto_sha1su1v4si): Ditto.
29239 (aarch64_crypto_sha1<sha1_op>v4si): Ditto.
29240 (aarch64_crypto_sha1su0v4si): Ditto.
29241 (aarch64_crypto_sha256h<sha256_op>v4si): Ditto.
29242 (aarch64_crypto_sha256su0v4si): Ditto.
29243 (aarch64_crypto_sha256su1v4si): Ditto.
29244 (doc/invoke.texi): Document new aes and sha2 options.
29246 2018-01-10 Martin Sebor <msebor@redhat.com>
29248 PR tree-optimization/83781
29249 * gimple-fold.c (get_range_strlen): Avoid treating arrays of pointers
29252 2018-01-11 Martin Sebor <msebor@gmail.com>
29253 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
29255 PR tree-optimization/83501
29256 PR tree-optimization/81703
29258 * tree-ssa-strlen.c (get_string_cst): Rename...
29259 (get_string_len): ...to this. Handle global constants.
29260 (handle_char_store): Adjust.
29262 2018-01-10 Kito Cheng <kito.cheng@gmail.com>
29263 Jim Wilson <jimw@sifive.com>
29265 * config/riscv/riscv-protos.h (riscv_output_return): New.
29266 * config/riscv/riscv.c (struct machine_function): New naked_p field.
29267 (riscv_attribute_table, riscv_output_return),
29268 (riscv_handle_fndecl_attribute, riscv_naked_function_p),
29269 (riscv_allocate_stack_slots_for_args, riscv_warn_func_return): New.
29270 (riscv_compute_frame_info): Only compute frame->mask if not a naked
29272 (riscv_expand_prologue): Add early return for naked function.
29273 (riscv_expand_epilogue): Likewise.
29274 (riscv_function_ok_for_sibcall): Return false for naked function.
29275 (riscv_set_current_function): New.
29276 (TARGET_SET_CURRENT_FUNCTION, TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS),
29277 (TARGET_ATTRIBUTE_TABLE, TARGET_WARN_FUNC_RETURN): New.
29278 * config/riscv/riscv.md (simple_return): Call riscv_output_return.
29279 * doc/extend.texi (RISC-V Function Attributes): New.
29281 2018-01-10 Michael Meissner <meissner@linux.vnet.ibm.com>
29283 * config/rs6000/rs6000.c (is_complex_IBM_long_double): Explicitly
29284 check for 128-bit long double before checking TCmode.
29285 * config/rs6000/rs6000.h (FLOAT128_IEEE_P): Explicitly check for
29286 128-bit long doubles before checking TFmode or TCmode.
29287 (FLOAT128_IBM_P): Likewise.
29289 2018-01-10 Martin Sebor <msebor@redhat.com>
29291 PR tree-optimization/83671
29292 * builtins.c (c_strlen): Unconditionally return zero for the empty
29294 Use -Warray-bounds for warnings.
29295 * gimple-fold.c (get_range_strlen): Handle non-constant lengths
29296 for non-constant array indices with COMPONENT_REF, arrays of
29297 arrays, and pointers to arrays.
29298 (gimple_fold_builtin_strlen): Determine and set length range for
29299 non-constant character arrays.
29301 2018-01-10 Aldy Hernandez <aldyh@redhat.com>
29303 PR middle-end/81897
29304 * tree-ssa-uninit.c (convert_control_dep_chain_into_preds): Skip
29307 2018-01-10 Eric Botcazou <ebotcazou@adacore.com>
29309 * dwarf2out.c (dwarf2out_var_location): Do not pass NULL to fprintf.
29311 2018-01-10 Peter Bergner <bergner@vnet.ibm.com>
29314 * config/rs6000/rs6000.c (print_operand) <'y'>: Use
29315 VECTOR_MEM_ALTIVEC_OR_VSX_P.
29316 * config/rs6000/vsx.md (*vsx_le_perm_load_<mode> for VSX_D): Use
29317 indexed_or_indirect_operand predicate.
29318 (*vsx_le_perm_load_<mode> for VSX_W): Likewise.
29319 (*vsx_le_perm_load_v8hi): Likewise.
29320 (*vsx_le_perm_load_v16qi): Likewise.
29321 (*vsx_le_perm_store_<mode> for VSX_D): Likewise.
29322 (*vsx_le_perm_store_<mode> for VSX_W): Likewise.
29323 (*vsx_le_perm_store_v8hi): Likewise.
29324 (*vsx_le_perm_store_v16qi): Likewise.
29325 (eight unnamed splitters): Likewise.
29327 2018-01-10 Peter Bergner <bergner@vnet.ibm.com>
29329 * config/rs6000/x86intrin.h: Change #warning to #error. Update message.
29330 * config/rs6000/emmintrin.h: Likewise.
29331 * config/rs6000/mmintrin.h: Likewise.
29332 * config/rs6000/xmmintrin.h: Likewise.
29334 2018-01-10 David Malcolm <dmalcolm@redhat.com>
29337 * tree-core.h: Document EXPR_LOCATION_WRAPPER_P's usage of
29339 * tree.c (tree_nop_conversion): Return true for location wrapper
29341 (maybe_wrap_with_location): New function.
29342 (selftest::check_strip_nops): New function.
29343 (selftest::test_location_wrappers): New function.
29344 (selftest::tree_c_tests): Call it.
29345 * tree.h (STRIP_ANY_LOCATION_WRAPPER): New macro.
29346 (maybe_wrap_with_location): New decl.
29347 (EXPR_LOCATION_WRAPPER_P): New macro.
29348 (location_wrapper_p): New inline function.
29349 (tree_strip_any_location_wrapper): New inline function.
29351 2018-01-10 H.J. Lu <hongjiu.lu@intel.com>
29354 * config/i386/i386.c (ix86_compute_frame_layout): Always adjust
29355 stack_realign_offset for the largest alignment of stack slot
29357 (ix86_find_max_used_stack_alignment): New function.
29358 (ix86_finalize_stack_frame_flags): Use it. Set
29359 max_used_stack_alignment if we don't realign stack.
29360 * config/i386/i386.h (machine_function): Add
29361 max_used_stack_alignment.
29363 2018-01-10 Christophe Lyon <christophe.lyon@linaro.org>
29365 * config/arm/arm.opt (-mbranch-cost): New option.
29366 * config/arm/arm.h (BRANCH_COST): Take arm_branch_cost into
29369 2018-01-10 Segher Boessenkool <segher@kernel.crashing.org>
29372 * config/rs6000/rs6000.md (load_toc_v4_PIC_2, load_toc_v4_PIC_3b,
29373 load_toc_v4_PIC_3c): Wrap const term in CONST RTL.
29375 2018-01-10 Richard Biener <rguenther@suse.de>
29378 * dwarf2out.c (gen_subprogram_die): Hoist old_die && declaration
29379 early out so it also covers the case where we have a non-NULL
29382 2018-01-10 Richard Sandiford <richard.sandiford@linaro.org>
29384 PR tree-optimization/83753
29385 * tree-vect-stmts.c (get_group_load_store_type): Use VMAT_CONTIGUOUS
29386 for non-strided grouped accesses if the number of elements is 1.
29388 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
29391 * i386.c (ix86_vectorize_builtin_gather): Check TARGET_USE_GATHER.
29392 * i386.h (TARGET_USE_GATHER): Define.
29393 * x86-tune.def (X86_TUNE_USE_GATHER): New.
29395 2018-01-10 Martin Liska <mliska@suse.cz>
29398 * basic-block.h (CLEANUP_NO_PARTITIONING): New define.
29399 * bb-reorder.c (pass_reorder_blocks::execute): Do not clean up
29401 * cfgcleanup.c (try_optimize_cfg): Fix up partitioning if
29402 CLEANUP_NO_PARTITIONING is not set.
29404 2018-01-10 Richard Sandiford <richard.sandiford@linaro.org>
29406 * doc/rtl.texi: Remove documentation of (const ...) wrappers
29407 for vectors, as a partial revert of r254296.
29408 * rtl.h (const_vec_p): Delete.
29409 (const_vec_duplicate_p): Don't test for vector CONSTs.
29410 (unwrap_const_vec_duplicate, const_vec_series_p): Likewise.
29411 * expmed.c (make_tree): Likewise.
29414 * common.md (E, F): Use CONSTANT_P instead of checking for
29416 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
29417 checking for CONST_VECTOR.
29419 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
29421 PR middle-end/83575
29422 * predict.c (force_edge_cold): Handle in more sane way edges
29423 with no prediction.
29425 2018-01-09 Carl Love <cel@us.ibm.com>
29427 * config/rs6002/altivec.md (p8_vmrgow): Add support for V2DI, V2DF,
29429 (p8_vmrgew): Add support for V2DI, V2DF, V4SF types.
29430 * config/rs6000/rs6000-builtin.def: Add definitions for FLOAT2_V2DF,
29431 VMRGEW_V2DI, VMRGEW_V2DF, VMRGEW_V4SF, VMRGOW_V4SI, VMRGOW_V4SF,
29432 VMRGOW_V2DI, VMRGOW_V2DF. Remove definition for VMRGOW.
29433 * config/rs6000/rs6000-c.c (VSX_BUILTIN_VEC_FLOAT2,
29434 P8V_BUILTIN_VEC_VMRGEW, P8V_BUILTIN_VEC_VMRGOW): Add definitions.
29435 * config/rs6000/rs6000-protos.h: Add extern defition for
29436 rs6000_generate_float2_double_code.
29437 * config/rs6000/rs6000.c (rs6000_generate_float2_double_code): Add
29439 * config/rs6000/vsx.md (vsx_xvcdpsp): Add define_insn.
29440 (float2_v2df): Add define_expand.
29442 2018-01-09 Uros Bizjak <ubizjak@gmail.com>
29445 * combine.c (force_int_to_mode) <case ASHIFT>: Use mode instead of
29446 op_mode in the force_to_mode call.
29448 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
29450 * config/aarch64/aarch64.c (aarch64_evpc_trn): Use d.perm.series_p
29451 instead of checking each element individually.
29452 (aarch64_evpc_uzp): Likewise.
29453 (aarch64_evpc_zip): Likewise.
29454 (aarch64_evpc_ext): Likewise.
29455 (aarch64_evpc_rev): Likewise.
29456 (aarch64_evpc_dup): Test the encoding for a single duplicated element,
29457 instead of checking each element individually. Return true without
29459 (aarch64_vectorize_vec_perm_const): Use all_from_input_p to test
29460 whether all selected elements come from the same input, instead of
29461 checking each element individually. Remove calls to gen_rtx_REG,
29462 start_sequence and end_sequence and instead assert that no rtl is
29465 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
29467 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p): Fix
29468 order of HIGH and CONST checks.
29470 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
29472 * tree-vect-stmts.c (permute_vec_elements): Create a fresh variable
29473 if the destination isn't an SSA_NAME.
29475 2018-01-09 Richard Biener <rguenther@suse.de>
29477 PR tree-optimization/83668
29478 * graphite.c (canonicalize_loop_closed_ssa): Add edge argument,
29480 (canonicalize_loop_form): ... here, renamed from ...
29481 (canonicalize_loop_closed_ssa_form): ... this and amended to
29482 swap successor edges for loop exit blocks to make us use
29483 the RPO order we need for initial schedule generation.
29485 2018-01-09 Joseph Myers <joseph@codesourcery.com>
29487 PR tree-optimization/64811
29488 * match.pd: When optimizing comparisons with Inf, avoid
29489 introducing or losing exceptions from comparisons with NaN.
29491 2018-01-09 Martin Liska <mliska@suse.cz>
29494 * asan.c (shadow_mem_size): Add gcc_assert.
29496 2018-01-09 Georg-Johann Lay <avr@gjlay.de>
29498 Don't save registers in main().
29501 * doc/invoke.texi (AVR Options) [-mmain-is-OS_task]: Document it.
29502 * config/avr/avr.opt (-mmain-is-OS_task): New target option.
29503 * config/avr/avr.c (avr_set_current_function): Don't error if
29504 naked, OS_task or OS_main are specified at the same time.
29505 (avr_function_ok_for_sibcall): Don't disable sibcalls for OS_task,
29507 (avr_insert_attributes) [-mmain-is-OS_task] <main>: Add OS_task
29509 * common/config/avr/avr-common.c (avr_option_optimization_table):
29510 Switch on -mmain-is-OS_task for optimizing compilations.
29512 2018-01-09 Richard Biener <rguenther@suse.de>
29514 PR tree-optimization/83572
29515 * graphite.c: Include cfganal.h.
29516 (graphite_transform_loops): Connect infinite loops to exit
29517 and remove fake edges at the end.
29519 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
29521 * ipa-inline.c (edge_badness): Revert accidental checkin.
29523 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
29526 * ipa-comdats.c (set_comdat_group): Only set comdat group of real
29527 symbols; not inline clones.
29529 2018-01-09 Jakub Jelinek <jakub@redhat.com>
29532 * modulo-sched.c (schedule_reg_moves): Punt if we'd need to move
29533 hard registers. Formatting fixes.
29535 PR preprocessor/83722
29536 * gcc.c (try_generate_repro): Pass
29537 &temp_stderr_files[RETRY_ICE_ATTEMPTS - 1] rather than
29538 &temp_stdout_files[RETRY_ICE_ATTEMPTS - 1] as last argument to
29541 2018-01-08 Monk Chiang <sh.chiang04@gmail.com>
29542 Kito Cheng <kito.cheng@gmail.com>
29544 * config/riscv/riscv.c (machine_function::is_leaf): Remove field.
29545 (riscv_leaf_function_p): Delete.
29546 (riscv_function_ok_for_sibcall): Return false when TARGET_SAVE_RESTORE.
29548 2018-01-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29550 * config/rs6000/rs6000-string.c (do_load_for_compare_from_addr): New
29552 (do_ifelse): New function.
29553 (do_isel): New function.
29554 (do_sub3): New function.
29555 (do_add3): New function.
29556 (do_load_mask_compare): New function.
29557 (do_overlap_load_compare): New function.
29558 (expand_compare_loop): New function.
29559 (expand_block_compare): Call expand_compare_loop() when appropriate.
29560 * config/rs6000/rs6000.opt (-mblock-compare-inline-limit): Change
29561 option description.
29562 (-mblock-compare-inline-loop-limit): New option.
29564 2018-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29567 * config/rs6000/altivec.md (*altivec_vpermr_<mode>_internal):
29568 Reverse order of second and third operands in first alternative.
29569 * config/rs6000/rs6000.c (rs6000_expand_vector_set): Reverse order
29570 of first and second elements in UNSPEC_VPERMR vector.
29571 (altivec_expand_vec_perm_le): Likewise.
29573 2018-01-08 Jeff Law <law@redhat.com>
29575 PR rtl-optimizatin/81308
29576 * tree-switch-conversion.c (cfg_altered): New file scoped static.
29577 (process_switch): If group_case_labels makes a change, then set
29579 (pass_convert_switch::execute): If a switch is converted, then
29580 set cfg_altered. Return TODO_cfg_cleanup if cfg_altered is true.
29582 PR rtl-optimization/81308
29583 * recog.c (split_all_insns): Conditionally cleanup the CFG after
29586 2018-01-08 Vidya Praveen <vidyapraveen@arm.com>
29588 PR target/83663 - Revert r255946
29589 * config/aarch64/aarch64.c (aarch64_expand_vector_init): Modify code
29590 generation for cases where splatting a value is not useful.
29591 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
29592 across a vec_duplicate and a paradoxical subreg forming a vector
29593 mode to a vec_concat.
29595 2018-01-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29597 * config/arm/t-aprofile (MULTILIB_MATCHES): Add mapping rules for
29598 -march=armv8.3-a variants.
29599 * config/arm/t-multilib: Likewise.
29600 * config/arm/t-arm-elf: Likewise. Handle dotprod extension.
29602 2018-01-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29604 * config/rs6000/rs6000.md (cceq_ior_compare): Remove * so I can use it
29606 (cceq_ior_compare_complement): Give it a name so I can use it, and
29607 change boolean_or_operator predicate to boolean_operator so it can
29608 be used to generate a crand.
29609 (eqne): New code iterator.
29610 (bd/bd_neg): New code_attrs.
29611 (<bd>_<mode>): New name for ctr<mode>_internal[12] now combined into
29612 a single define_insn.
29613 (<bd>tf_<mode>): A new insn pattern for the conditional form branch
29614 decrement (bdnzt/bdnzf/bdzt/bdzf).
29615 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Updated
29616 with the new names of the branch decrement patterns, and added the
29617 names of the branch decrement conditional patterns.
29619 2018-01-08 Richard Biener <rguenther@suse.de>
29621 PR tree-optimization/83563
29622 * graphite.c (canonicalize_loop_closed_ssa_form): Reset the SCEV
29625 2018-01-08 Richard Biener <rguenther@suse.de>
29627 PR middle-end/83713
29628 * convert.c (do_narrow): Properly guard TYPE_OVERFLOW_WRAPS checks.
29630 2018-01-08 Richard Biener <rguenther@suse.de>
29632 PR tree-optimization/83685
29633 * tree-ssa-pre.c (create_expression_by_pieces): Do not insert
29634 references to abnormals.
29636 2018-01-08 Richard Biener <rguenther@suse.de>
29639 * dwarf2out.c (output_indirect_strings): Handle empty
29640 skeleton_debug_str_hash.
29641 (dwarf2out_early_finish): Index strings for -gsplit-dwarf.
29643 2018-01-08 Claudiu Zissulescu <claziss@synopsys.com>
29645 * config/arc/arc.c (TARGET_TRAMPOLINE_ADJUST_ADDRESS): Delete.
29646 (emit_store_direct): Likewise.
29647 (arc_trampoline_adjust_address): Likewise.
29648 (arc_asm_trampoline_template): New function.
29649 (arc_initialize_trampoline): Use asm_trampoline_template.
29650 (TARGET_ASM_TRAMPOLINE_TEMPLATE): Define.
29651 * config/arc/arc.h (TRAMPOLINE_SIZE): Adjust to 16.
29652 * config/arc/arc.md (flush_icache): Delete pattern.
29654 2018-01-08 Claudiu Zissulescu <claziss@synopsys.com>
29656 * config/arc/arc-c.def (__ARC_UNALIGNED__): New define.
29657 * config/arc/arc.h (STRICT_ALIGNMENT): Control this macro using
29660 2018-01-08 Sebastian Huber <sebastian.huber@embedded-brains.de>
29663 * config/epiphany/epiphany.h (make_pass_mode_switch_use): Guard
29664 by not USED_FOR_TARGET.
29665 (make_pass_resolve_sw_modes): Likewise.
29667 2018-01-08 Sebastian Huber <sebastian.huber@embedded-brains.de>
29669 * config/nios2/nios2.h (nios2_section_threshold): Guard by not
29672 2018-01-08 Richard Biener <rguenther@suse.de>
29674 PR middle-end/83580
29675 * tree-data-ref.c (split_constant_offset): Remove STRIP_NOPS.
29677 2018-01-08 Richard Biener <rguenther@suse.de>
29679 PR middle-end/83517
29680 * match.pd ((t * 2) / 2) -> t): Add missing :c.
29682 2018-01-06 Aldy Hernandez <aldyh@redhat.com>
29684 PR middle-end/81897
29685 * tree-ssa-uninit.c (compute_control_dep_chain): Do not bail on
29686 basic blocks with a small number of successors.
29687 (convert_control_dep_chain_into_preds): Improve handling of
29689 (dump_predicates): Split apart into...
29690 (dump_pred_chain): ...here...
29691 (dump_pred_info): ...and here.
29692 (can_one_predicate_be_invalidated_p): Add debugging printfs.
29693 (can_chain_union_be_invalidated_p): Improve check for invalidation
29695 (uninit_uses_cannot_happen): Avoid unnecessary if
29696 convert_control_dep_chain_into_preds yielded nothing.
29698 2018-01-06 Martin Sebor <msebor@redhat.com>
29700 PR tree-optimization/83640
29701 * gimple-ssa-warn-restrict.c (builtin_access::builtin_access): Avoid
29702 subtracting negative offset from size.
29703 (builtin_access::overlap): Adjust offset bounds of the access to fall
29704 within the size of the object if possible.
29706 2018-01-06 Richard Sandiford <richard.sandiford@linaro.org>
29708 PR rtl-optimization/83699
29709 * expmed.c (extract_bit_field_1): Restrict the vector usage of
29710 extract_bit_field_as_subreg to cases in which the extracted
29711 value is also a vector.
29713 * lra-constraints.c (process_alt_operands): Test for the equivalence
29714 substitutions when detecting a possible reload cycle.
29716 2018-01-06 Jakub Jelinek <jakub@redhat.com>
29719 * toplev.c (process_options): Don't enable debug_nonbind_markers_p
29720 by default if flag_selective_schedling{,2}. Formatting fixes.
29722 PR rtl-optimization/83682
29723 * rtl.h (const_vec_duplicate_p): Only return true for VEC_DUPLICATE
29724 if it has non-VECTOR_MODE element mode.
29725 (vec_duplicate_p): Likewise.
29727 PR middle-end/83694
29728 * cfgexpand.c (expand_debug_expr): Punt if mode1 is VOIDmode
29729 and bitsize might be greater than MAX_BITSIZE_MODE_ANY_INT.
29731 2018-01-05 Jakub Jelinek <jakub@redhat.com>
29734 * config/i386/i386-builtin.def
29735 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
29736 __builtin_ia32_vgf2p8affineqb_v64qi, __builtin_ia32_vgf2p8mulb_v64qi):
29737 Require also OPTION_MASK_ISA_AVX512F in addition to
29738 OPTION_MASK_ISA_GFNI.
29739 (__builtin_ia32_vgf2p8affineinvqb_v16qi_mask,
29740 __builtin_ia32_vgf2p8affineqb_v16qi_mask): Require
29741 OPTION_MASK_ISA_AVX512VL instead of OPTION_MASK_ISA_SSE in addition
29742 to OPTION_MASK_ISA_GFNI.
29743 (__builtin_ia32_vgf2p8mulb_v32qi_mask): Require
29744 OPTION_MASK_ISA_AVX512VL in addition to OPTION_MASK_ISA_GFNI and
29745 OPTION_MASK_ISA_AVX512BW.
29746 (__builtin_ia32_vgf2p8mulb_v16qi_mask): Require
29747 OPTION_MASK_ISA_AVX512VL instead of OPTION_MASK_ISA_AVX512BW in
29748 addition to OPTION_MASK_ISA_GFNI.
29749 (__builtin_ia32_vgf2p8affineinvqb_v16qi,
29750 __builtin_ia32_vgf2p8affineqb_v16qi, __builtin_ia32_vgf2p8mulb_v16qi):
29751 Require OPTION_MASK_ISA_SSE2 instead of OPTION_MASK_ISA_SSE in addition
29752 to OPTION_MASK_ISA_GFNI.
29753 * config/i386/i386.c (def_builtin): Change to builtin isa/isa2 being
29754 a requirement for all ISAs rather than any of them with a few
29756 (ix86_add_new_builtins): Clear OPTION_MASK_ISA_64BIT from isa before
29758 (ix86_expand_builtin): Require all ISAs from builtin's isa and isa2
29759 bitmasks to be enabled with 3 exceptions, instead of requiring any
29760 enabled ISA with lots of exceptions.
29761 * config/i386/sse.md (vgf2p8affineinvqb_<mode><mask_name>,
29762 vgf2p8affineqb_<mode><mask_name>, vgf2p8mulb_<mode><mask_name>):
29763 Change avx512bw in isa attribute to avx512f.
29764 * config/i386/sgxintrin.h: Add license boilerplate.
29765 * config/i386/vaesintrin.h: Likewise. Fix macro spelling __AVX512F
29766 to __AVX512F__ and __AVX512VL to __AVX512VL__.
29767 (_mm256_aesdec_epi128, _mm256_aesdeclast_epi128, _mm256_aesenc_epi128,
29768 _mm256_aesenclast_epi128): Enable temporarily avx if __AVX__ is not
29770 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8,
29771 _mm_gf2p8affineinv_epi64_epi8, _mm_gf2p8affine_epi64_epi8): Enable
29772 temporarily sse2 rather than sse if not enabled already.
29775 * config/i386/sse.md (VI248_VLBW): Rename to ...
29776 (VI248_AVX512VL): ... this. Don't guard V32HI with TARGET_AVX512BW.
29777 (vpshrd_<mode><mask_name>, vpshld_<mode><mask_name>,
29778 vpshrdv_<mode>, vpshrdv_<mode>_mask, vpshrdv_<mode>_maskz,
29779 vpshrdv_<mode>_maskz_1, vpshldv_<mode>, vpshldv_<mode>_mask,
29780 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): Use VI248_AVX512VL
29781 mode iterator instead of VI248_VLBW.
29783 2018-01-05 Jan Hubicka <hubicka@ucw.cz>
29785 * ipa-fnsummary.c (record_modified_bb_info): Add OP.
29786 (record_modified): Skip clobbers; add debug output.
29787 (param_change_prob): Use sreal frequencies.
29789 2018-01-05 Richard Sandiford <richard.sandiford@linaro.org>
29791 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
29792 punt for user-aligned variables.
29794 2018-01-05 Richard Sandiford <richard.sandiford@linaro.org>
29796 * tree-chrec.c (chrec_contains_symbols): Return true for
29799 2018-01-05 Sudakshina Das <sudi.das@arm.com>
29802 * simplify-rtx.c (simplify_relational_operation_1): Add simplifications
29803 of (x|y) == x for BICS pattern.
29805 2018-01-05 Jakub Jelinek <jakub@redhat.com>
29807 PR tree-optimization/83605
29808 * gimple-ssa-strength-reduction.c: Include tree-eh.h.
29809 (find_candidates_dom_walker::before_dom_children): Ignore stmts that
29812 2018-01-05 Sebastian Huber <sebastian.huber@embedded-brains.de>
29814 * config.gcc (epiphany-*-elf*): Add (epiphany-*-rtems*) configuration.
29815 * config/epiphany/rtems.h: New file.
29817 2018-01-04 Jakub Jelinek <jakub@redhat.com>
29818 Uros Bizjak <ubizjak@gmail.com>
29821 * config/i386/i386.md (*<rotate_insn>hi3_1 splitter): Use
29822 QIreg_operand instead of register_operand predicate.
29823 * config/i386/i386.c (ix86_rop_should_change_byte_p,
29824 set_rop_modrm_reg_bits, ix86_mitigate_rop): Use -mmitigate-rop in
29825 comments instead of -fmitigate[-_]rop.
29827 2018-01-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
29830 * cgraphunit.c (symbol_table::compile): Switch to text_section
29831 before calling assembly_start debug hook.
29832 * run-rtl-passes.c (run_rtl_passes): Likewise.
29835 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
29837 * tree-vrp.c (extract_range_from_binary_expr_1): Check
29838 range_int_cst_p rather than !symbolic_range_p before calling
29839 extract_range_from_multiplicative_op_1.
29841 2018-01-04 Jeff Law <law@redhat.com>
29843 * tree-ssa-math-opts.c (execute_cse_reciprocals_1): Remove
29844 redundant test in assertion.
29846 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
29848 * doc/rtl.texi: Document machine_mode wrapper classes.
29850 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
29852 * fold-const.c (fold_ternary_loc): Check tree_fits_uhwi_p before
29853 using tree_to_uhwi.
29855 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
29857 * tree-ssa-forwprop.c (is_combined_permutation_identity): Allow
29858 the VEC_PERM_EXPR fold to fail.
29860 2018-01-04 Jakub Jelinek <jakub@redhat.com>
29863 * bb-reorder.c (insert_section_boundary_note): Set has_bb_partition
29864 to switched_sections.
29866 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
29869 * config/arm/arm.c (arm_vectorize_vec_perm_const): Fix inverted
29870 test for d.testing.
29872 2018-01-04 Peter Bergner <bergner@vnet.ibm.com>
29875 * config/rs6000/rs6000.c (rs6000_discover_homogeneous_aggregate): Do not
29876 allow arguments in FP registers if TARGET_HARD_FLOAT is false.
29878 2018-01-04 Jakub Jelinek <jakub@redhat.com>
29881 * cfgexpand.c (expand_debug_expr) <case BIT_FIELD_REF>: Punt if mode
29882 is BLKmode and bitpos not zero or mode change is needed.
29884 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
29887 * config/sparc/sparc.c (sparc_vectorize_vec_perm_const): Require
29890 2018-01-04 Uros Bizjak <ubizjak@gmail.com>
29893 * config/alpha/alpha.md (*sadd<modesuffix>): Use ASHIFT
29894 instead of MULT rtx. Update all corresponding splitters.
29895 (*saddl_se): Ditto.
29896 (*ssub<modesuffix>): Ditto.
29897 (*ssubl_se): Ditto.
29898 (*cmp_sadd_di): Update split patterns.
29899 (*cmp_sadd_si): Ditto.
29900 (*cmp_sadd_sidi): Ditto.
29901 (*cmp_ssub_di): Ditto.
29902 (*cmp_ssub_si): Ditto.
29903 (*cmp_ssub_sidi): Ditto.
29904 * config/alpha/predicates.md (const23_operand): New predicate.
29905 * config/alpha/alpha.c (alpha_rtx_costs) [PLUS, MINUS]:
29906 Look for ASHIFT, not MULT inner operand.
29907 (alpha_split_conditional_move): Update for *sadd<modesuffix> change.
29909 2018-01-04 Martin Liska <mliska@suse.cz>
29911 PR gcov-profile/83669
29912 * gcov.c (output_intermediate_file): Add version to intermediate
29914 * doc/gcov.texi: Document new field 'version' in intermediate
29915 file format. Fix location of '-k' option of gcov command.
29917 2018-01-04 Martin Liska <mliska@suse.cz>
29920 * ipa-icf.c (sem_function::merge): Do not cross comdat boundary.
29922 2018-01-04 Jakub Jelinek <jakub@redhat.com>
29924 * gimple-ssa-sprintf.c (parse_directive): Cast second dir.len to uhwi.
29926 2018-01-03 Martin Sebor <msebor@redhat.com>
29928 PR tree-optimization/83655
29929 * gimple-ssa-warn-restrict.c (wrestrict_dom_walker::check_call): Avoid
29930 checking calls with invalid arguments.
29932 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29934 * tree-vect-stmts.c (vect_get_store_rhs): New function.
29935 (vectorizable_mask_load_store): Delete.
29936 (vectorizable_call): Return false for masked loads and stores.
29937 (vectorizable_store): Handle IFN_MASK_STORE. Use vect_get_store_rhs
29938 instead of gimple_assign_rhs1.
29939 (vectorizable_load): Handle IFN_MASK_LOAD.
29940 (vect_transform_stmt): Don't set is_store for call_vec_info_type.
29942 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29944 * tree-vect-stmts.c (vect_build_gather_load_calls): New function,
29946 (vectorizable_mask_load_store): ...here.
29947 (vectorizable_load): ...and here.
29949 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29951 * tree-vect-stmts.c (vect_build_all_ones_mask)
29952 (vect_build_zero_merge_argument): New functions, split out from...
29953 (vectorizable_load): ...here.
29955 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29957 * tree-vect-stmts.c (vect_check_store_rhs): New function,
29959 (vectorizable_mask_load_store): ...here.
29960 (vectorizable_store): ...and here.
29962 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29964 * tree-vect-stmts.c (vect_check_load_store_mask): New function,
29966 (vectorizable_mask_load_store): ...here.
29968 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29970 * tree-vectorizer.h (vec_load_store_type): Moved from tree-vec-stmts.c
29971 (vect_model_store_cost): Take a vec_load_store_type instead of a
29973 * tree-vect-stmts.c (vec_load_store_type): Move to tree-vectorizer.h.
29974 (vect_model_store_cost): Take a vec_load_store_type instead of a
29976 (vectorizable_mask_load_store): Update accordingly.
29977 (vectorizable_store): Likewise.
29978 * tree-vect-slp.c (vect_analyze_slp_cost_1): Update accordingly.
29980 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29982 * tree-vect-loop.c (vect_transform_loop): Stub out scalar
29983 IFN_MASK_LOAD calls here rather than...
29984 * tree-vect-stmts.c (vectorizable_mask_load_store): ...here.
29986 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29987 Alan Hayward <alan.hayward@arm.com>
29988 David Sherwood <david.sherwood@arm.com>
29990 * expmed.c (extract_bit_field_1): For vector extracts,
29991 fall back to extract_bit_field_as_subreg if vec_extract
29994 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
29995 Alan Hayward <alan.hayward@arm.com>
29996 David Sherwood <david.sherwood@arm.com>
29998 * lra-spills.c (pseudo_reg_slot_compare): Sort slots by whether
29999 they are variable or constant sized.
30000 (assign_stack_slot_num_and_sort_pseudos): Don't reuse variable-sized
30001 slots for constant-sized data.
30003 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
30004 Alan Hayward <alan.hayward@arm.com>
30005 David Sherwood <david.sherwood@arm.com>
30007 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern): When
30008 handling COND_EXPRs with boolean comparisons, try to find a better
30009 basis for the mask type than the boolean itself.
30011 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
30013 * doc/rtl.texi (MAX_BITSIZE_MODE_ANY_MODE): Describe how the default
30014 is calculated and how it can be overridden.
30015 * genmodes.c (max_bitsize_mode_any_mode): New variable.
30016 (create_modes): Initialize it from MAX_BITSIZE_MODE_ANY_MODE,
30018 (emit_max_int): Use it to set the output MAX_BITSIZE_MODE_ANY_MODE,
30021 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
30022 Alan Hayward <alan.hayward@arm.com>
30023 David Sherwood <david.sherwood@arm.com>
30025 * config/aarch64/aarch64-protos.h (aarch64_output_simd_mov_immediate):
30026 Remove the mode argument.
30027 (aarch64_simd_valid_immediate): Remove the mode and inverse
30029 * config/aarch64/iterators.md (bitsize): New iterator.
30030 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>, and<mode>3)
30031 (ior<mode>3): Update calls to aarch64_output_simd_mov_immediate.
30032 * config/aarch64/constraints.md (Do, Db, Dn): Update calls to
30033 aarch64_simd_valid_immediate.
30034 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Likewise.
30035 (aarch64_reg_or_bic_imm): Likewise.
30036 * config/aarch64/aarch64.c (simd_immediate_info): Replace mvn
30037 with an insn_type enum and msl with a modifier_type enum.
30038 Replace element_width with a scalar_mode. Change the shift
30039 to unsigned int. Add constructors for scalar_float_mode and
30040 scalar_int_mode elements.
30041 (aarch64_vect_float_const_representable_p): Delete.
30042 (aarch64_can_const_movi_rtx_p)
30043 (aarch64_simd_scalar_immediate_valid_for_move)
30044 (aarch64_simd_make_constant): Update call to
30045 aarch64_simd_valid_immediate.
30046 (aarch64_advsimd_valid_immediate_hs): New function.
30047 (aarch64_advsimd_valid_immediate): Likewise.
30048 (aarch64_simd_valid_immediate): Remove mode and inverse
30049 arguments. Rewrite to use the above. Use const_vec_duplicate_p
30050 to detect duplicated constants and use aarch64_float_const_zero_rtx_p
30051 and aarch64_float_const_representable_p on the result.
30052 (aarch64_output_simd_mov_immediate): Remove mode argument.
30053 Update call to aarch64_simd_valid_immediate and use of
30054 simd_immediate_info.
30055 (aarch64_output_scalar_simd_mov_immediate): Update call
30058 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
30059 Alan Hayward <alan.hayward@arm.com>
30060 David Sherwood <david.sherwood@arm.com>
30062 * machmode.h (mode_precision): Prefix with CONST_MODE_PRECISION.
30063 (mode_nunits): Likewise CONST_MODE_NUNITS.
30064 * machmode.def (ADJUST_NUNITS): Document.
30065 * genmodes.c (mode_data::need_nunits_adj): New field.
30066 (blank_mode): Update accordingly.
30067 (adj_nunits): New variable.
30068 (print_maybe_const_decl): Replace CATEGORY with a NEEDS_ADJ
30070 (emit_mode_size_inline): Set need_bytesize_adj for all modes
30071 listed in adj_nunits.
30072 (emit_mode_nunits_inline): Set need_nunits_adj for all modes
30073 listed in adj_nunits. Don't emit case statements for such modes.
30074 (emit_insn_modes_h): Emit definitions of CONST_MODE_NUNITS
30075 and CONST_MODE_PRECISION. Make CONST_MODE_SIZE expand to
30076 nothing if adj_nunits is nonnull.
30077 (emit_mode_precision, emit_mode_nunits): Use print_maybe_const_decl.
30078 (emit_mode_unit_size, emit_mode_base_align, emit_mode_ibit)
30079 (emit_mode_fbit): Update use of print_maybe_const_decl.
30080 (emit_move_size): Likewise. Treat the array as non-const
30082 (emit_mode_adjustments): Handle adj_nunits.
30084 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
30086 * machmode.def (VECTOR_MODES_WITH_PREFIX): Document.
30087 * genmodes.c (VECTOR_MODES_WITH_PREFIX): New macro.
30088 (VECTOR_MODES): Use it.
30089 (make_vector_modes): Take the prefix as an argument.
30091 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
30092 Alan Hayward <alan.hayward@arm.com>
30093 David Sherwood <david.sherwood@arm.com>
30095 * mode-classes.def (MODE_VECTOR_BOOL): New mode class.
30096 * machmode.h (INTEGRAL_MODE_P, VECTOR_MODE_P): Return true
30097 for MODE_VECTOR_BOOL.
30098 * machmode.def (VECTOR_BOOL_MODE): Document.
30099 * genmodes.c (VECTOR_BOOL_MODE): New macro.
30100 (make_vector_bool_mode): New function.
30101 (complete_mode, emit_mode_wider, emit_mode_adjustments): Handle
30103 * lto-streamer-in.c (lto_input_mode_table): Likewise.
30104 * rtx-vector-builder.c (rtx_vector_builder::find_cached_value):
30106 * stor-layout.c (int_mode_for_mode): Likewise.
30107 * tree.c (build_vector_type_for_mode): Likewise.
30108 * varasm.c (output_constant_pool_2): Likewise.
30109 * emit-rtl.c (init_emit_once): Make sure that CONST1_RTX (BImode) and
30110 CONSTM1_RTX (BImode) are the same thing. Initialize const_tiny_rtx
30111 for MODE_VECTOR_BOOL.
30112 * expr.c (expand_expr_real_1): Use VECTOR_MODE_P instead of a list
30113 of mode class checks.
30114 * tree-vect-generic.c (expand_vector_operation): Use VECTOR_MODE_P
30115 instead of a list of mode class checks.
30116 (expand_vector_scalar_condition): Likewise.
30117 (type_for_widest_vector_mode): Handle BImode as an inner mode.
30119 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
30120 Alan Hayward <alan.hayward@arm.com>
30121 David Sherwood <david.sherwood@arm.com>
30123 * machmode.h (mode_size): Change from unsigned short to
30125 (mode_to_bytes): Return a poly_uint16 rather than an unsigned short.
30126 (GET_MODE_SIZE): Return a constant if ONLY_FIXED_SIZE_MODES,
30127 or if measurement_type is not polynomial.
30128 (fixed_size_mode::includes_p): Check for constant-sized modes.
30129 * genmodes.c (emit_mode_size_inline): Make mode_size_inline
30130 return a poly_uint16 rather than an unsigned short.
30131 (emit_mode_size): Change the type of mode_size from unsigned short
30132 to poly_uint16_pod. Use ZERO_COEFFS for the initializer.
30133 (emit_mode_adjustments): Cope with polynomial vector sizes.
30134 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
30136 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
30138 * auto-inc-dec.c (try_merge): Treat GET_MODE_SIZE as polynomial.
30139 * builtins.c (expand_ifn_atomic_compare_exchange_into_call): Likewise.
30140 * caller-save.c (setup_save_areas): Likewise.
30141 (replace_reg_with_saved_mem): Likewise.
30142 * calls.c (emit_library_call_value_1): Likewise.
30143 * combine-stack-adj.c (combine_stack_adjustments_for_block): Likewise.
30144 * combine.c (simplify_set, make_extraction, simplify_shift_const_1)
30145 (gen_lowpart_for_combine): Likewise.
30146 * convert.c (convert_to_integer_1): Likewise.
30147 * cse.c (equiv_constant, cse_insn): Likewise.
30148 * cselib.c (autoinc_split, cselib_hash_rtx): Likewise.
30149 (cselib_subst_to_values): Likewise.
30150 * dce.c (word_dce_process_block): Likewise.
30151 * df-problems.c (df_word_lr_mark_ref): Likewise.
30152 * dwarf2cfi.c (init_one_dwarf_reg_size): Likewise.
30153 * dwarf2out.c (multiple_reg_loc_descriptor, mem_loc_descriptor)
30154 (concat_loc_descriptor, concatn_loc_descriptor, loc_descriptor)
30155 (rtl_for_decl_location): Likewise.
30156 * emit-rtl.c (gen_highpart, widen_memory_access): Likewise.
30157 * expmed.c (extract_bit_field_1, extract_integral_bit_field): Likewise.
30158 * expr.c (emit_group_load_1, clear_storage_hints): Likewise.
30159 (emit_move_complex, emit_move_multi_word, emit_push_insn): Likewise.
30160 (expand_expr_real_1): Likewise.
30161 * function.c (assign_parm_setup_block_p, assign_parm_setup_block)
30162 (pad_below): Likewise.
30163 * gimple-fold.c (optimize_atomic_compare_exchange_p): Likewise.
30164 * gimple-ssa-store-merging.c (rhs_valid_for_store_merging_p): Likewise.
30165 * ira.c (get_subreg_tracking_sizes): Likewise.
30166 * ira-build.c (ira_create_allocno_objects): Likewise.
30167 * ira-color.c (coalesced_pseudo_reg_slot_compare): Likewise.
30168 (ira_sort_regnos_for_alter_reg): Likewise.
30169 * ira-costs.c (record_operand_costs): Likewise.
30170 * lower-subreg.c (interesting_mode_p, simplify_gen_subreg_concatn)
30171 (resolve_simple_move): Likewise.
30172 * lra-constraints.c (get_reload_reg, operands_match_p): Likewise.
30173 (process_addr_reg, simplify_operand_subreg, curr_insn_transform)
30174 (lra_constraints): Likewise.
30175 (CONST_POOL_OK_P): Reject variable-sized modes.
30176 * lra-spills.c (slot, assign_mem_slot, pseudo_reg_slot_compare)
30177 (add_pseudo_to_slot, lra_spill): Likewise.
30178 * omp-low.c (omp_clause_aligned_alignment): Likewise.
30179 * optabs-query.c (get_best_extraction_insn): Likewise.
30180 * optabs-tree.c (expand_vec_cond_expr_p): Likewise.
30181 * optabs.c (expand_vec_perm_var, expand_vec_cond_expr): Likewise.
30182 (expand_mult_highpart, valid_multiword_target_p): Likewise.
30183 * recog.c (offsettable_address_addr_space_p): Likewise.
30184 * regcprop.c (maybe_mode_change): Likewise.
30185 * reginfo.c (choose_hard_reg_mode, record_subregs_of_mode): Likewise.
30186 * regrename.c (build_def_use): Likewise.
30187 * regstat.c (dump_reg_info): Likewise.
30188 * reload.c (complex_word_subreg_p, push_reload, find_dummy_reload)
30189 (find_reloads, find_reloads_subreg_address): Likewise.
30190 * reload1.c (eliminate_regs_1): Likewise.
30191 * rtlanal.c (for_each_inc_dec_find_inc_dec, rtx_cost): Likewise.
30192 * simplify-rtx.c (avoid_constant_pool_reference): Likewise.
30193 (simplify_binary_operation_1, simplify_subreg): Likewise.
30194 * targhooks.c (default_function_arg_padding): Likewise.
30195 (default_hard_regno_nregs, default_class_max_nregs): Likewise.
30196 * tree-cfg.c (verify_gimple_assign_binary): Likewise.
30197 (verify_gimple_assign_ternary): Likewise.
30198 * tree-inline.c (estimate_move_cost): Likewise.
30199 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
30200 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Likewise.
30201 (get_address_cost_ainc): Likewise.
30202 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
30203 (vect_supportable_dr_alignment): Likewise.
30204 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
30205 (vectorizable_reduction): Likewise.
30206 * tree-vect-stmts.c (vectorizable_assignment, vectorizable_shift)
30207 (vectorizable_operation, vectorizable_load): Likewise.
30208 * tree.c (build_same_sized_truth_vector_type): Likewise.
30209 * valtrack.c (cleanup_auto_inc_dec): Likewise.
30210 * var-tracking.c (emit_note_insn_var_location): Likewise.
30211 * config/arc/arc.h (ASM_OUTPUT_CASE_END): Use as_a <scalar_int_mode>.
30212 (ADDR_VEC_ALIGN): Likewise.
30214 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
30215 Alan Hayward <alan.hayward@arm.com>
30216 David Sherwood <david.sherwood@arm.com>
30218 * machmode.h (mode_to_bits): Return a poly_uint16 rather than an
30220 (GET_MODE_BITSIZE): Return a constant if ONLY_FIXED_SIZE_MODES,
30221 or if measurement_type is polynomial.
30222 * calls.c (shift_return_value): Treat GET_MODE_BITSIZE as polynomial.
30223 * combine.c (make_extraction): Likewise.
30224 * dse.c (find_shift_sequence): Likewise.
30225 * dwarf2out.c (mem_loc_descriptor): Likewise.
30226 * expmed.c (store_integral_bit_field, extract_bit_field_1): Likewise.
30227 (extract_bit_field, extract_low_bits): Likewise.
30228 * expr.c (convert_move, convert_modes, emit_move_insn_1): Likewise.
30229 (optimize_bitfield_assignment_op, expand_assignment): Likewise.
30230 (store_expr_with_bounds, store_field, expand_expr_real_1): Likewise.
30231 * fold-const.c (optimize_bit_field_compare, merge_ranges): Likewise.
30232 * gimple-fold.c (optimize_atomic_compare_exchange_p): Likewise.
30233 * reload.c (find_reloads): Likewise.
30234 * reload1.c (alter_reg): Likewise.
30235 * stor-layout.c (bitwise_mode_for_mode, compute_record_mode): Likewise.
30236 * targhooks.c (default_secondary_memory_needed_mode): Likewise.
30237 * tree-if-conv.c (predicate_mem_writes): Likewise.
30238 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
30239 * tree-vect-patterns.c (adjust_bool_pattern): Likewise.
30240 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
30241 * valtrack.c (dead_debug_insert_temp): Likewise.
30242 * varasm.c (mergeable_constant_section): Likewise.
30243 * config/sh/sh.h (LOCAL_ALIGNMENT): Use as_a <fixed_size_mode>.
30245 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
30246 Alan Hayward <alan.hayward@arm.com>
30247 David Sherwood <david.sherwood@arm.com>
30249 * expr.c (expand_assignment): Cope with polynomial mode sizes
30250 when assigning to a CONCAT.
30252 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
30253 Alan Hayward <alan.hayward@arm.com>
30254 David Sherwood <david.sherwood@arm.com>
30256 * machmode.h (mode_precision): Change from unsigned short to
30258 (mode_to_precision): Return a poly_uint16 rather than an unsigned
30260 (GET_MODE_PRECISION): Return a constant if ONLY_FIXED_SIZE_MODES,
30261 or if measurement_type is not polynomial.
30262 (HWI_COMPUTABLE_MODE_P): Turn into a function. Optimize the case
30263 in which the mode is already known to be a scalar_int_mode.
30264 * genmodes.c (emit_mode_precision): Change the type of mode_precision
30265 from unsigned short to poly_uint16_pod. Use ZERO_COEFFS for the
30267 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
30268 for GET_MODE_PRECISION.
30269 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
30270 for GET_MODE_PRECISION.
30271 * combine.c (update_rsp_from_reg_equal): Treat GET_MODE_PRECISION
30273 (try_combine, find_split_point, combine_simplify_rtx): Likewise.
30274 (expand_field_assignment, make_extraction): Likewise.
30275 (make_compound_operation_int, record_dead_and_set_regs_1): Likewise.
30276 (get_last_value): Likewise.
30277 * convert.c (convert_to_integer_1): Likewise.
30278 * cse.c (cse_insn): Likewise.
30279 * expr.c (expand_expr_real_1): Likewise.
30280 * lra-constraints.c (simplify_operand_subreg): Likewise.
30281 * optabs-query.c (can_atomic_load_p): Likewise.
30282 * optabs.c (expand_atomic_load): Likewise.
30283 (expand_atomic_store): Likewise.
30284 * ree.c (combine_reaching_defs): Likewise.
30285 * rtl.h (partial_subreg_p, paradoxical_subreg_p): Likewise.
30286 * rtlanal.c (nonzero_bits1, lsb_bitfield_op_p): Likewise.
30287 * tree.h (type_has_mode_precision_p): Likewise.
30288 * ubsan.c (instrument_si_overflow): Likewise.
30290 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
30291 Alan Hayward <alan.hayward@arm.com>
30292 David Sherwood <david.sherwood@arm.com>
30294 * tree.h (TYPE_VECTOR_SUBPARTS): Turn into a function and handle
30295 polynomial numbers of units.
30296 (SET_TYPE_VECTOR_SUBPARTS): Likewise.
30297 (valid_vector_subparts_p): New function.
30298 (build_vector_type): Remove temporary shim and take the number
30299 of units as a poly_uint64 rather than an int.
30300 (build_opaque_vector_type): Take the number of units as a
30301 poly_uint64 rather than an int.
30302 * tree.c (build_vector_from_ctor): Handle polynomial
30303 TYPE_VECTOR_SUBPARTS.
30304 (type_hash_canon_hash, type_cache_hasher::equal): Likewise.
30305 (uniform_vector_p, vector_type_mode, build_vector): Likewise.
30306 (build_vector_from_val): If the number of units is variable,
30307 use build_vec_duplicate_cst for constant operands and
30308 VEC_DUPLICATE_EXPR otherwise.
30309 (make_vector_type): Remove temporary is_constant ().
30310 (build_vector_type, build_opaque_vector_type): Take the number of
30311 units as a poly_uint64 rather than an int.
30312 (check_vector_cst): Handle polynomial TYPE_VECTOR_SUBPARTS and
30314 * cfgexpand.c (expand_debug_expr): Likewise.
30315 * expr.c (count_type_elements, categorize_ctor_elements_1): Likewise.
30316 (store_constructor, expand_expr_real_1): Likewise.
30317 (const_scalar_mask_from_tree): Likewise.
30318 * fold-const-call.c (fold_const_reduction): Likewise.
30319 * fold-const.c (const_binop, const_unop, fold_convert_const): Likewise.
30320 (operand_equal_p, fold_vec_perm, fold_ternary_loc): Likewise.
30321 (native_encode_vector, vec_cst_ctor_to_array): Likewise.
30322 (fold_relational_const): Likewise.
30323 (native_interpret_vector): Likewise. Change the size from an
30324 int to an unsigned int.
30325 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Handle polynomial
30326 TYPE_VECTOR_SUBPARTS.
30327 (gimple_fold_indirect_ref, gimple_build_vector): Likewise.
30328 (gimple_build_vector_from_val): Use VEC_DUPLICATE_EXPR when
30329 duplicating a non-constant operand into a variable-length vector.
30330 * hsa-brig.c (hsa_op_immed::emit_to_buffer): Handle polynomial
30331 TYPE_VECTOR_SUBPARTS and VECTOR_CST_NELTS.
30332 * ipa-icf.c (sem_variable::equals): Likewise.
30333 * match.pd: Likewise.
30334 * omp-simd-clone.c (simd_clone_subparts): Likewise.
30335 * print-tree.c (print_node): Likewise.
30336 * stor-layout.c (layout_type): Likewise.
30337 * targhooks.c (default_builtin_vectorization_cost): Likewise.
30338 * tree-cfg.c (verify_gimple_comparison): Likewise.
30339 (verify_gimple_assign_binary): Likewise.
30340 (verify_gimple_assign_ternary): Likewise.
30341 (verify_gimple_assign_single): Likewise.
30342 * tree-pretty-print.c (dump_generic_node): Likewise.
30343 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
30344 (simplify_bitfield_ref, is_combined_permutation_identity): Likewise.
30345 * tree-vect-data-refs.c (vect_permute_store_chain): Likewise.
30346 (vect_grouped_load_supported, vect_permute_load_chain): Likewise.
30347 (vect_shift_permute_load_chain): Likewise.
30348 * tree-vect-generic.c (nunits_for_known_piecewise_op): Likewise.
30349 (expand_vector_condition, optimize_vector_constructor): Likewise.
30350 (lower_vec_perm, get_compute_type): Likewise.
30351 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
30352 (get_initial_defs_for_reduction, vect_transform_loop): Likewise.
30353 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
30354 (vect_recog_mask_conversion_pattern): Likewise.
30355 * tree-vect-slp.c (vect_supported_load_permutation_p): Likewise.
30356 (vect_get_constant_vectors, vect_transform_slp_perm_load): Likewise.
30357 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
30358 (get_group_load_store_type, vectorizable_mask_load_store): Likewise.
30359 (vectorizable_bswap, simd_clone_subparts, vectorizable_assignment)
30360 (vectorizable_shift, vectorizable_operation, vectorizable_store)
30361 (vectorizable_load, vect_is_simple_cond, vectorizable_comparison)
30362 (supportable_widening_operation): Likewise.
30363 (supportable_narrowing_operation): Likewise.
30364 * tree-vector-builder.c (tree_vector_builder::binary_encoded_nelts):
30366 * varasm.c (output_constant): Likewise.
30368 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
30369 Alan Hayward <alan.hayward@arm.com>
30370 David Sherwood <david.sherwood@arm.com>
30372 * tree-vect-data-refs.c (vect_permute_store_chain): Reorganize
30373 so that both the length == 3 and length != 3 cases set up their
30374 own permute vectors. Add comments explaining why we know the
30375 number of elements is constant.
30376 (vect_permute_load_chain): Likewise.
30378 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
30379 Alan Hayward <alan.hayward@arm.com>
30380 David Sherwood <david.sherwood@arm.com>
30382 * machmode.h (mode_nunits): Change from unsigned char to
30384 (ONLY_FIXED_SIZE_MODES): New macro.
30385 (pod_mode::measurement_type, scalar_int_mode::measurement_type)
30386 (scalar_float_mode::measurement_type, scalar_mode::measurement_type)
30387 (complex_mode::measurement_type, fixed_size_mode::measurement_type):
30389 (mode_to_nunits): Return a poly_uint16 rather than an unsigned short.
30390 (GET_MODE_NUNITS): Return a constant if ONLY_FIXED_SIZE_MODES,
30391 or if measurement_type is not polynomial.
30392 * genmodes.c (ZERO_COEFFS): New macro.
30393 (emit_mode_nunits_inline): Make mode_nunits_inline return a
30395 (emit_mode_nunits): Change the type of mode_nunits to poly_uint16_pod.
30396 Use ZERO_COEFFS when emitting initializers.
30397 * data-streamer.h (bp_pack_poly_value): New function.
30398 (bp_unpack_poly_value): Likewise.
30399 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
30400 for GET_MODE_NUNITS.
30401 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
30402 for GET_MODE_NUNITS.
30403 * tree.c (make_vector_type): Remove temporary shim and make
30404 the real function take the number of units as a poly_uint64
30405 rather than an int.
30406 (build_vector_type_for_mode): Handle polynomial nunits.
30407 * dwarf2out.c (loc_descriptor, add_const_value_attribute): Likewise.
30408 * emit-rtl.c (const_vec_series_p_1): Likewise.
30409 (gen_rtx_CONST_VECTOR): Likewise.
30410 * fold-const.c (test_vec_duplicate_folding): Likewise.
30411 * genrecog.c (validate_pattern): Likewise.
30412 * optabs-query.c (can_vec_perm_var_p, can_mult_highpart_p): Likewise.
30413 * optabs-tree.c (expand_vec_cond_expr_p): Likewise.
30414 * optabs.c (expand_vector_broadcast, expand_binop_directly): Likewise.
30415 (shift_amt_for_vec_perm_mask, expand_vec_perm_var): Likewise.
30416 (expand_vec_cond_expr, expand_mult_highpart): Likewise.
30417 * rtlanal.c (subreg_get_info): Likewise.
30418 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
30419 (vect_grouped_load_supported): Likewise.
30420 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
30421 * tree-vect-loop.c (have_whole_vector_shift): Likewise.
30422 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
30423 (simplify_const_unary_operation, simplify_binary_operation_1)
30424 (simplify_const_binary_operation, simplify_ternary_operation)
30425 (test_vector_ops_duplicate, test_vector_ops): Likewise.
30426 (simplify_immed_subreg): Use GET_MODE_NUNITS on a fixed_size_mode
30427 instead of CONST_VECTOR_NUNITS.
30428 * varasm.c (output_constant_pool_2): Likewise.
30429 * rtx-vector-builder.c (rtx_vector_builder::build): Only include the
30430 explicit-encoded elements in the XVEC for variable-length vectors.
30432 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
30434 * lra-constraints.c (curr_insn_transform): Use partial_subreg_p.
30436 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
30437 Alan Hayward <alan.hayward@arm.com>
30438 David Sherwood <david.sherwood@arm.com>
30440 * coretypes.h (fixed_size_mode): Declare.
30441 (fixed_size_mode_pod): New typedef.
30442 * builtins.h (target_builtins::x_apply_args_mode)
30443 (target_builtins::x_apply_result_mode): Change type to
30444 fixed_size_mode_pod.
30445 * builtins.c (apply_args_size, apply_result_size, result_vector)
30446 (expand_builtin_apply_args_1, expand_builtin_apply)
30447 (expand_builtin_return): Update accordingly.
30449 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
30451 * cse.c (hash_rtx_cb): Hash only the encoded elements.
30452 * cselib.c (cselib_hash_rtx): Likewise.
30453 * expmed.c (make_tree): Build VECTOR_CSTs directly from the
30454 CONST_VECTOR encoding.
30456 2018-01-03 Jakub Jelinek <jakub@redhat.com>
30457 Jeff Law <law@redhat.com>
30460 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): For
30461 noreturn probe, use gen_pop instead of ix86_emit_restore_reg_using_pop,
30462 only set RTX_FRAME_RELATED_P on both the push and pop if cfa_reg is sp
30463 and add REG_CFA_ADJUST_CFA notes in that case to both insns.
30466 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Do not
30467 explicitly probe *sp in a noreturn function if there were any callee
30468 register saves or frame pointer is needed.
30470 2018-01-03 Jakub Jelinek <jakub@redhat.com>
30473 * cfgexpand.c (expand_debug_expr): Return NULL if mode is
30474 BLKmode for ternary, binary or unary expressions.
30477 * var-tracking.c (delete_vta_debug_insn): New inline function.
30478 (delete_vta_debug_insns): Add USE_CFG argument, if true, walk just
30479 insns from get_insns () to NULL instead of each bb separately.
30480 Use delete_vta_debug_insn. No longer static.
30481 (vt_debug_insns_local, variable_tracking_main_1): Adjust
30482 delete_vta_debug_insns callers.
30483 * rtl.h (delete_vta_debug_insns): Declare.
30484 * final.c (rest_of_handle_final): Call delete_vta_debug_insns
30485 instead of variable_tracking_main.
30487 2018-01-03 Martin Sebor <msebor@redhat.com>
30489 PR tree-optimization/83603
30490 * calls.c (maybe_warn_nonstring_arg): Avoid accessing function
30491 arguments past the endof the argument list in functions declared
30492 without a prototype.
30493 * gimple-ssa-warn-restrict.c (wrestrict_dom_walker::check_call):
30494 Avoid checking when arguments are null.
30496 2018-01-03 Martin Sebor <msebor@redhat.com>
30499 * doc/extend.texi (attribute const): Fix a typo.
30500 * ipa-pure-const.c ((warn_function_const, warn_function_pure): Avoid
30501 issuing -Wsuggest-attribute for void functions.
30503 2018-01-03 Martin Sebor <msebor@redhat.com>
30505 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Use
30506 offset_int::from instead of wide_int::to_shwi.
30507 (maybe_diag_overlap): Remove assertion.
30508 Use HOST_WIDE_INT_PRINT_DEC instead of %lli.
30509 * gimple-ssa-sprintf.c (format_directive): Same.
30510 (parse_directive): Same.
30511 (sprintf_dom_walker::compute_format_length): Same.
30512 (try_substitute_return_value): Same.
30514 2018-01-03 Jeff Law <law@redhat.com>
30516 PR middle-end/83654
30517 * explow.c (anti_adjust_stack_and_probe_stack_clash): Test a
30518 non-constant residual for zero at runtime and avoid probing in
30519 that case. Reorganize code for trailing problem to mirror handling
30522 2018-01-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
30524 PR tree-optimization/83501
30525 * tree-ssa-strlen.c (get_string_cst): New.
30526 (handle_char_store): Call get_string_cst.
30528 2018-01-03 Martin Liska <mliska@suse.cz>
30530 PR tree-optimization/83593
30531 * tree-ssa-strlen.c: Include tree-cfg.h.
30532 (strlen_check_and_optimize_stmt): Add new argument cleanup_eh.
30533 (strlen_dom_walker): Add new member variable m_cleanup_cfg.
30534 (strlen_dom_walker::strlen_dom_walker): Initialize m_cleanup_cfg
30536 (strlen_dom_walker::before_dom_children): Call
30537 gimple_purge_dead_eh_edges. Dump tranformation with details
30539 (strlen_dom_walker::before_dom_children): Update call by adding
30540 new argument cleanup_eh.
30541 (pass_strlen::execute): Return TODO_cleanup_cfg if needed.
30543 2018-01-03 Martin Liska <mliska@suse.cz>
30546 * cif-code.def (VARIADIC_THUNK): New enum value.
30547 * ipa-fnsummary.c (compute_fn_summary): Do not inline variadic
30550 2018-01-03 Jan Beulich <jbeulich@suse.com>
30552 * sse.md (mov<mode>_internal): Tighten condition for when to use
30553 vmovdqu<ssescalarsize> for TI and OI modes.
30555 2018-01-03 Jakub Jelinek <jakub@redhat.com>
30557 Update copyright years.
30559 2018-01-03 Martin Liska <mliska@suse.cz>
30562 * ipa-visibility.c (function_and_variable_visibility): Skip
30563 functions with noipa attribure.
30565 2018-01-03 Jakub Jelinek <jakub@redhat.com>
30567 * gcc.c (process_command): Update copyright notice dates.
30568 * gcov-dump.c (print_version): Ditto.
30569 * gcov.c (print_version): Ditto.
30570 * gcov-tool.c (print_version): Ditto.
30571 * gengtype.c (create_file): Ditto.
30572 * doc/cpp.texi: Bump @copying's copyright year.
30573 * doc/cppinternals.texi: Ditto.
30574 * doc/gcc.texi: Ditto.
30575 * doc/gccint.texi: Ditto.
30576 * doc/gcov.texi: Ditto.
30577 * doc/install.texi: Ditto.
30578 * doc/invoke.texi: Ditto.
30580 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
30582 * vector-builder.h (vector_builder::m_full_nelts): Change from
30583 unsigned int to poly_uint64.
30584 (vector_builder::full_nelts): Update prototype accordingly.
30585 (vector_builder::new_vector): Likewise.
30586 (vector_builder::encoded_full_vector_p): Handle polynomial full_nelts.
30587 (vector_builder::operator ==): Likewise.
30588 (vector_builder::finalize): Likewise.
30589 * int-vector-builder.h (int_vector_builder::int_vector_builder):
30590 Take the number of elements as a poly_uint64 rather than an
30592 * vec-perm-indices.h (vec_perm_indices::m_nelts_per_input): Change
30593 from unsigned int to poly_uint64.
30594 (vec_perm_indices::vec_perm_indices): Update prototype accordingly.
30595 (vec_perm_indices::new_vector): Likewise.
30596 (vec_perm_indices::length): Likewise.
30597 (vec_perm_indices::nelts_per_input): Likewise.
30598 (vec_perm_indices::input_nelts): Likewise.
30599 * vec-perm-indices.c (vec_perm_indices::new_vector): Take the
30600 number of elements per input as a poly_uint64 rather than an
30601 unsigned int. Use the original encoding for variable-length
30602 vectors, rather than clamping each individual element.
30603 For the second and subsequent elements in each pattern,
30604 clamp the step and base before clamping their sum.
30605 (vec_perm_indices::series_p): Handle polynomial element counts.
30606 (vec_perm_indices::all_in_range_p): Likewise.
30607 (vec_perm_indices_to_tree): Likewise.
30608 (vec_perm_indices_to_rtx): Likewise.
30609 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
30610 * tree-vector-builder.c (tree_vector_builder::new_unary_operation)
30611 (tree_vector_builder::new_binary_operation): Handle polynomial
30612 element counts. Return false if we need to know the number
30613 of elements at compile time.
30614 * fold-const.c (fold_vec_perm): Punt if the number of elements
30615 isn't known at compile time.
30617 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
30619 * vec-perm-indices.h (vec_perm_builder): Change element type
30620 from HOST_WIDE_INT to poly_int64.
30621 (vec_perm_indices::element_type): Update accordingly.
30622 (vec_perm_indices::clamp): Handle polynomial element_types.
30623 * vec-perm-indices.c (vec_perm_indices::series_p): Likewise.
30624 (vec_perm_indices::all_in_range_p): Likewise.
30625 (tree_to_vec_perm_builder): Check for poly_int64 trees rather
30627 * vector-builder.h (vector_builder::stepped_sequence_p): Handle
30628 polynomial vec_perm_indices element types.
30629 * int-vector-builder.h (int_vector_builder::equal_p): Likewise.
30630 * fold-const.c (fold_vec_perm): Likewise.
30631 * optabs.c (shift_amt_for_vec_perm_mask): Likewise.
30632 * tree-vect-generic.c (lower_vec_perm): Likewise.
30633 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
30634 * config/aarch64/aarch64.c (aarch64_evpc_tbl): Cast d->perm
30635 element type to HOST_WIDE_INT.
30637 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
30638 Alan Hayward <alan.hayward@arm.com>
30639 David Sherwood <david.sherwood@arm.com>
30641 * alias.c (addr_side_effect_eval): Take the size as a poly_int64
30642 rather than an int. Use plus_constant.
30643 (memrefs_conflict_p): Take the sizes as poly_int64s rather than ints.
30644 Take the offset "c" as a poly_int64 rather than a HOST_WIDE_INT.
30646 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
30647 Alan Hayward <alan.hayward@arm.com>
30648 David Sherwood <david.sherwood@arm.com>
30650 * calls.c (emit_call_1, expand_call): Change struct_value_size from
30651 a HOST_WIDE_INT to a poly_int64.
30653 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
30654 Alan Hayward <alan.hayward@arm.com>
30655 David Sherwood <david.sherwood@arm.com>
30657 * calls.c (load_register_parameters): Cope with polynomial
30658 mode sizes. Require a constant size for BLKmode parameters
30659 that aren't described by a PARALLEL. If BLOCK_REG_PADDING
30660 forces a parameter to be padded at the lsb end in order to
30661 fill a complete number of words, require the parameter size
30662 to be ordered wrt UNITS_PER_WORD.
30664 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
30665 Alan Hayward <alan.hayward@arm.com>
30666 David Sherwood <david.sherwood@arm.com>
30668 * reload1.c (spill_stack_slot_width): Change element type
30669 from unsigned int to poly_uint64_pod.
30670 (alter_reg): Treat mode sizes as polynomial.
30672 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
30673 Alan Hayward <alan.hayward@arm.com>
30674 David Sherwood <david.sherwood@arm.com>
30676 * reload.c (complex_word_subreg_p): New function.
30677 (reload_inner_reg_of_subreg, push_reload): Use it.
30679 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
30680 Alan Hayward <alan.hayward@arm.com>
30681 David Sherwood <david.sherwood@arm.com>
30683 * lra-constraints.c (process_alt_operands): Reject matched
30684 operands whose sizes aren't ordered.
30685 (match_reload): Refer to this check here.
30687 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
30688 Alan Hayward <alan.hayward@arm.com>
30689 David Sherwood <david.sherwood@arm.com>
30691 * builtins.c (expand_ifn_atomic_compare_exchange_into_call): Assert
30692 that the mode size is in the set {1, 2, 4, 8, 16}.
30694 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
30695 Alan Hayward <alan.hayward@arm.com>
30696 David Sherwood <david.sherwood@arm.com>
30698 * var-tracking.c (adjust_mems): Treat mode sizes as polynomial.
30699 Use plus_constant instead of gen_rtx_PLUS.
30701 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
30702 Alan Hayward <alan.hayward@arm.com>
30703 David Sherwood <david.sherwood@arm.com>
30705 * config/cr16/cr16-protos.h (cr16_push_rounding): Declare.
30706 * config/cr16/cr16.h (PUSH_ROUNDING): Move implementation to...
30707 * config/cr16/cr16.c (cr16_push_rounding): ...this new function.
30708 * config/h8300/h8300-protos.h (h8300_push_rounding): Declare.
30709 * config/h8300/h8300.h (PUSH_ROUNDING): Move implementation to...
30710 * config/h8300/h8300.c (h8300_push_rounding): ...this new function.
30711 * config/i386/i386-protos.h (ix86_push_rounding): Declare.
30712 * config/i386/i386.h (PUSH_ROUNDING): Move implementation to...
30713 * config/i386/i386.c (ix86_push_rounding): ...this new function.
30714 * config/m32c/m32c-protos.h (m32c_push_rounding): Take and return
30716 * config/m32c/m32c.c (m32c_push_rounding): Likewise.
30717 * config/m68k/m68k-protos.h (m68k_push_rounding): Declare.
30718 * config/m68k/m68k.h (PUSH_ROUNDING): Move implementation to...
30719 * config/m68k/m68k.c (m68k_push_rounding): ...this new function.
30720 * config/pdp11/pdp11-protos.h (pdp11_push_rounding): Declare.
30721 * config/pdp11/pdp11.h (PUSH_ROUNDING): Move implementation to...
30722 * config/pdp11/pdp11.c (pdp11_push_rounding): ...this new function.
30723 * config/stormy16/stormy16-protos.h (xstormy16_push_rounding): Declare.
30724 * config/stormy16/stormy16.h (PUSH_ROUNDING): Move implementation to...
30725 * config/stormy16/stormy16.c (xstormy16_push_rounding): ...this new
30727 * expr.c (emit_move_resolve_push): Treat the input and result
30728 of PUSH_ROUNDING as a poly_int64.
30729 (emit_move_complex_push, emit_single_push_insn_1): Likewise.
30730 (emit_push_insn): Likewise.
30731 * lra-eliminations.c (mark_not_eliminable): Likewise.
30732 * recog.c (push_operand): Likewise.
30733 * reload1.c (elimination_effects): Likewise.
30734 * rtlanal.c (nonzero_bits1): Likewise.
30735 * calls.c (store_one_arg): Likewise. Require the padding to be
30736 known at compile time.
30738 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
30739 Alan Hayward <alan.hayward@arm.com>
30740 David Sherwood <david.sherwood@arm.com>
30742 * expr.c (emit_single_push_insn_1): Treat mode sizes as polynomial.
30743 Use plus_constant instead of gen_rtx_PLUS.
30745 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
30746 Alan Hayward <alan.hayward@arm.com>
30747 David Sherwood <david.sherwood@arm.com>
30749 * auto-inc-dec.c (set_inc_state): Take the mode size as a poly_int64
30750 rather than an int.
30752 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
30753 Alan Hayward <alan.hayward@arm.com>
30754 David Sherwood <david.sherwood@arm.com>
30756 * expr.c (expand_expr_real_1): Use tree_to_poly_uint64
30757 instead of int_size_in_bytes when handling VIEW_CONVERT_EXPRs
30758 via stack temporaries. Treat the mode size as polynomial too.
30760 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
30761 Alan Hayward <alan.hayward@arm.com>
30762 David Sherwood <david.sherwood@arm.com>
30764 * expr.c (expand_expr_real_2): When handling conversions involving
30765 unions, apply tree_to_poly_uint64 to the TYPE_SIZE rather than
30766 multiplying int_size_in_bytes by BITS_PER_UNIT. Treat GET_MODE_BISIZE
30767 as a poly_uint64 too.
30769 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
30770 Alan Hayward <alan.hayward@arm.com>
30771 David Sherwood <david.sherwood@arm.com>
30773 * rtlanal.c (subreg_get_info): Handle polynomial mode sizes.
30775 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
30776 Alan Hayward <alan.hayward@arm.com>
30777 David Sherwood <david.sherwood@arm.com>
30779 * combine.c (can_change_dest_mode): Handle polynomial
30780 REGMODE_NATURAL_SIZE.
30781 * expmed.c (store_bit_field_1): Likewise.
30782 * expr.c (store_constructor): Likewise.
30783 * emit-rtl.c (validate_subreg): Operate on polynomial mode sizes
30784 and polynomial REGMODE_NATURAL_SIZE.
30785 (gen_lowpart_common): Likewise.
30786 * reginfo.c (record_subregs_of_mode): Likewise.
30787 * rtlanal.c (read_modify_subreg_p): Likewise.
30789 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
30790 Alan Hayward <alan.hayward@arm.com>
30791 David Sherwood <david.sherwood@arm.com>
30793 * internal-fn.c (expand_vector_ubsan_overflow): Handle polynomial
30794 numbers of elements.
30796 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
30797 Alan Hayward <alan.hayward@arm.com>
30798 David Sherwood <david.sherwood@arm.com>
30800 * match.pd: Cope with polynomial numbers of vector elements.
30802 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
30803 Alan Hayward <alan.hayward@arm.com>
30804 David Sherwood <david.sherwood@arm.com>
30806 * fold-const.c (fold_indirect_ref_1): Handle polynomial offsets
30807 in a POINTER_PLUS_EXPR.
30809 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
30810 Alan Hayward <alan.hayward@arm.com>
30811 David Sherwood <david.sherwood@arm.com>
30813 * omp-simd-clone.c (simd_clone_subparts): New function.
30814 (simd_clone_init_simd_arrays): Use it instead of TYPE_VECTOR_SUBPARTS.
30815 (ipa_simd_modify_function_body): Likewise.
30817 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
30818 Alan Hayward <alan.hayward@arm.com>
30819 David Sherwood <david.sherwood@arm.com>
30821 * tree-vect-generic.c (nunits_for_known_piecewise_op): New function.
30822 (expand_vector_piecewise): Use it instead of TYPE_VECTOR_SUBPARTS.
30823 (expand_vector_addition, add_rshift, expand_vector_divmod): Likewise.
30824 (expand_vector_condition, vector_element): Likewise.
30825 (subparts_gt): New function.
30826 (get_compute_type): Use subparts_gt.
30827 (count_type_subparts): Delete.
30828 (expand_vector_operations_1): Use subparts_gt instead of
30829 count_type_subparts.
30831 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
30832 Alan Hayward <alan.hayward@arm.com>
30833 David Sherwood <david.sherwood@arm.com>
30835 * tree-vect-data-refs.c (vect_no_alias_p): Replace with...
30836 (vect_compile_time_alias): ...this new function. Do the calculation
30837 on poly_ints rather than trees.
30838 (vect_prune_runtime_alias_test_list): Update call accordingly.
30840 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
30841 Alan Hayward <alan.hayward@arm.com>
30842 David Sherwood <david.sherwood@arm.com>
30844 * tree-vect-slp.c (vect_build_slp_tree_1): Handle polynomial
30846 (vect_schedule_slp_instance): Likewise.
30848 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
30849 Alan Hayward <alan.hayward@arm.com>
30850 David Sherwood <david.sherwood@arm.com>
30852 * tree-vect-slp.c (vect_get_and_check_slp_defs): Reject
30853 constant and extern definitions for variable-length vectors.
30854 (vect_get_constant_vectors): Note that the number of units
30855 is known to be constant.
30857 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
30858 Alan Hayward <alan.hayward@arm.com>
30859 David Sherwood <david.sherwood@arm.com>
30861 * tree-vect-stmts.c (vectorizable_conversion): Treat the number
30862 of units as polynomial. Choose between WIDE and NARROW based
30865 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
30866 Alan Hayward <alan.hayward@arm.com>
30867 David Sherwood <david.sherwood@arm.com>
30869 * tree-vect-stmts.c (simd_clone_subparts): New function.
30870 (vectorizable_simd_clone_call): Use it instead of TYPE_VECTOR_SUBPARTS.
30872 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
30873 Alan Hayward <alan.hayward@arm.com>
30874 David Sherwood <david.sherwood@arm.com>
30876 * tree-vect-stmts.c (vectorizable_call): Treat the number of
30877 vectors as polynomial. Use build_index_vector for
30878 IFN_GOMP_SIMD_LANE.
30880 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
30881 Alan Hayward <alan.hayward@arm.com>
30882 David Sherwood <david.sherwood@arm.com>
30884 * tree-vect-stmts.c (get_load_store_type): Treat the number of
30885 units as polynomial. Reject VMAT_ELEMENTWISE and VMAT_STRIDED_SLP
30886 for variable-length vectors.
30887 (vectorizable_mask_load_store): Treat the number of units as
30888 polynomial, asserting that it is constant if the condition has
30889 already been enforced.
30890 (vectorizable_store, vectorizable_load): Likewise.
30892 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
30893 Alan Hayward <alan.hayward@arm.com>
30894 David Sherwood <david.sherwood@arm.com>
30896 * tree-vect-loop.c (vectorizable_live_operation): Treat the number
30897 of units as polynomial. Punt if we can't tell at compile time
30898 which vector contains the final result.
30900 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
30901 Alan Hayward <alan.hayward@arm.com>
30902 David Sherwood <david.sherwood@arm.com>
30904 * tree-vect-loop.c (vectorizable_induction): Treat the number
30905 of units as polynomial. Punt on SLP inductions. Use an integer
30906 VEC_SERIES_EXPR for variable-length integer reductions. Use a
30907 cast of such a series for variable-length floating-point
30910 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
30911 Alan Hayward <alan.hayward@arm.com>
30912 David Sherwood <david.sherwood@arm.com>
30914 * tree.h (build_index_vector): Declare.
30915 * tree.c (build_index_vector): New function.
30916 * tree-vect-loop.c (get_initial_defs_for_reduction): Treat the number
30917 of units as polynomial, forcibly converting it to a constant if
30918 vectorizable_reduction has already enforced the condition.
30919 (vect_create_epilog_for_reduction): Likewise. Use build_index_vector
30920 to create a {1,2,3,...} vector.
30921 (vectorizable_reduction): Treat the number of units as polynomial.
30922 Choose vectype_in based on the largest scalar element size rather
30923 than the smallest number of units. Enforce the restrictions
30926 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
30927 Alan Hayward <alan.hayward@arm.com>
30928 David Sherwood <david.sherwood@arm.com>
30930 * tree-vect-data-refs.c (vector_alignment_reachable_p): Treat the
30931 number of units as polynomial.
30933 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
30934 Alan Hayward <alan.hayward@arm.com>
30935 David Sherwood <david.sherwood@arm.com>
30937 * target.h (vector_sizes, auto_vector_sizes): New typedefs.
30938 * target.def (autovectorize_vector_sizes): Return the vector sizes
30939 by pointer, using vector_sizes rather than a bitmask.
30940 * targhooks.h (default_autovectorize_vector_sizes): Update accordingly.
30941 * targhooks.c (default_autovectorize_vector_sizes): Likewise.
30942 * config/aarch64/aarch64.c (aarch64_autovectorize_vector_sizes):
30944 * config/arc/arc.c (arc_autovectorize_vector_sizes): Likewise.
30945 * config/arm/arm.c (arm_autovectorize_vector_sizes): Likewise.
30946 * config/i386/i386.c (ix86_autovectorize_vector_sizes): Likewise.
30947 * config/mips/mips.c (mips_autovectorize_vector_sizes): Likewise.
30948 * omp-general.c (omp_max_vf): Likewise.
30949 * omp-low.c (omp_clause_aligned_alignment): Likewise.
30950 * optabs-query.c (can_vec_mask_load_store_p): Likewise.
30951 * tree-vect-loop.c (vect_analyze_loop): Likewise.
30952 * tree-vect-slp.c (vect_slp_bb): Likewise.
30953 * doc/tm.texi: Regenerate.
30954 * tree-vectorizer.h (current_vector_size): Change from an unsigned int
30956 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Take
30957 the vector size as a poly_uint64 rather than an unsigned int.
30958 (current_vector_size): Change from an unsigned int to a poly_uint64.
30959 (get_vectype_for_scalar_type): Update accordingly.
30960 * tree.h (build_truth_vector_type): Take the size and number of
30961 units as a poly_uint64 rather than an unsigned int.
30962 (build_vector_type): Add a temporary overload that takes
30963 the number of units as a poly_uint64 rather than an unsigned int.
30964 * tree.c (make_vector_type): Likewise.
30965 (build_truth_vector_type): Take the number of units as a poly_uint64
30966 rather than an unsigned int.
30968 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
30969 Alan Hayward <alan.hayward@arm.com>
30970 David Sherwood <david.sherwood@arm.com>
30972 * target.def (get_mask_mode): Take the number of units and length
30973 as poly_uint64s rather than unsigned ints.
30974 * targhooks.h (default_get_mask_mode): Update accordingly.
30975 * targhooks.c (default_get_mask_mode): Likewise.
30976 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
30977 * doc/tm.texi: Regenerate.
30979 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
30980 Alan Hayward <alan.hayward@arm.com>
30981 David Sherwood <david.sherwood@arm.com>
30983 * omp-general.h (omp_max_vf): Return a poly_uint64 instead of an int.
30984 * omp-general.c (omp_max_vf): Likewise.
30985 * omp-expand.c (omp_adjust_chunk_size): Update call to omp_max_vf.
30986 (expand_omp_simd): Handle polynomial safelen.
30987 * omp-low.c (omplow_simd_context): Add a default constructor.
30988 (omplow_simd_context::max_vf): Change from int to poly_uint64.
30989 (lower_rec_simd_input_clauses): Update accordingly.
30990 (lower_rec_input_clauses): Likewise.
30992 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
30993 Alan Hayward <alan.hayward@arm.com>
30994 David Sherwood <david.sherwood@arm.com>
30996 * tree-vectorizer.h (vect_nunits_for_cost): New function.
30997 * tree-vect-loop.c (vect_model_reduction_cost): Use it.
30998 * tree-vect-slp.c (vect_analyze_slp_cost_1): Likewise.
30999 (vect_analyze_slp_cost): Likewise.
31000 * tree-vect-stmts.c (vect_model_store_cost): Likewise.
31001 (vect_model_load_cost): Likewise.
31003 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31004 Alan Hayward <alan.hayward@arm.com>
31005 David Sherwood <david.sherwood@arm.com>
31007 * tree-vect-slp.c (vect_record_max_nunits, vect_build_slp_tree_1)
31008 (vect_build_slp_tree_2, vect_build_slp_tree): Change max_nunits
31009 from an unsigned int * to a poly_uint64_pod *.
31010 (calculate_unrolling_factor): New function.
31011 (vect_analyze_slp_instance): Use it. Track polynomial max_nunits.
31013 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31014 Alan Hayward <alan.hayward@arm.com>
31015 David Sherwood <david.sherwood@arm.com>
31017 * tree-vectorizer.h (_slp_instance::unrolling_factor): Change
31018 from an unsigned int to a poly_uint64.
31019 (_loop_vec_info::slp_unrolling_factor): Likewise.
31020 (_loop_vec_info::vectorization_factor): Change from an int
31022 (MAX_VECTORIZATION_FACTOR): Bump from 64 to INT_MAX.
31023 (vect_get_num_vectors): New function.
31024 (vect_update_max_nunits, vect_vf_for_cost): Likewise.
31025 (vect_get_num_copies): Use vect_get_num_vectors.
31026 (vect_analyze_data_ref_dependences): Change max_vf from an int *
31027 to an unsigned int *.
31028 (vect_analyze_data_refs): Change min_vf from an int * to a
31030 (vect_transform_slp_perm_load): Take the vf as a poly_uint64 rather
31031 than an unsigned HOST_WIDE_INT.
31032 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr)
31033 (vect_analyze_data_ref_dependence): Change max_vf from an int *
31034 to an unsigned int *.
31035 (vect_analyze_data_ref_dependences): Likewise.
31036 (vect_compute_data_ref_alignment): Handle polynomial vf.
31037 (vect_enhance_data_refs_alignment): Likewise.
31038 (vect_prune_runtime_alias_test_list): Likewise.
31039 (vect_shift_permute_load_chain): Likewise.
31040 (vect_supportable_dr_alignment): Likewise.
31041 (dependence_distance_ge_vf): Take the vectorization factor as a
31042 poly_uint64 rather than an unsigned HOST_WIDE_INT.
31043 (vect_analyze_data_refs): Change min_vf from an int * to a
31045 * tree-vect-loop-manip.c (vect_gen_scalar_loop_niters): Take
31046 vfm1 as a poly_uint64 rather than an int. Make the same change
31047 for the returned bound_scalar.
31048 (vect_gen_vector_loop_niters): Handle polynomial vf.
31049 (vect_do_peeling): Likewise. Update call to
31050 vect_gen_scalar_loop_niters and handle polynomial bound_scalars.
31051 (vect_gen_vector_loop_niters_mult_vf): Assert that the vf must
31053 * tree-vect-loop.c (vect_determine_vectorization_factor)
31054 (vect_update_vf_for_slp, vect_analyze_loop_2): Handle polynomial vf.
31055 (vect_get_known_peeling_cost): Likewise.
31056 (vect_estimate_min_profitable_iters, vectorizable_reduction): Likewise.
31057 (vect_worthwhile_without_simd_p, vectorizable_induction): Likewise.
31058 (vect_transform_loop): Likewise. Use the lowest possible VF when
31059 updating the upper bounds of the loop.
31060 (vect_min_worthwhile_factor): Make static. Return an unsigned int
31061 rather than an int.
31062 * tree-vect-slp.c (vect_attempt_slp_rearrange_stmts): Cope with
31063 polynomial unroll factors.
31064 (vect_analyze_slp_cost_1, vect_analyze_slp_instance): Likewise.
31065 (vect_make_slp_decision): Likewise.
31066 (vect_supported_load_permutation_p): Likewise, and polynomial
31068 (vect_analyze_slp_cost): Handle polynomial vf.
31069 (vect_slp_analyze_node_operations): Likewise.
31070 (vect_slp_analyze_bb_1): Likewise.
31071 (vect_transform_slp_perm_load): Take the vf as a poly_uint64 rather
31072 than an unsigned HOST_WIDE_INT.
31073 * tree-vect-stmts.c (vectorizable_simd_clone_call, vectorizable_store)
31074 (vectorizable_load): Handle polynomial vf.
31075 * tree-vectorizer.c (simduid_to_vf::vf): Change from an int to
31077 (adjust_simduid_builtins, shrink_simd_arrays): Update accordingly.
31079 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31080 Alan Hayward <alan.hayward@arm.com>
31081 David Sherwood <david.sherwood@arm.com>
31083 * match.pd: Handle bit operations involving three constants
31084 and try to fold one pair.
31086 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
31088 * tree-vect-loop-manip.c: Include gimple-fold.h.
31089 (slpeel_make_loop_iterate_ntimes): Add step, final_iv and
31090 niters_maybe_zero parameters. Handle other cases besides a step of 1.
31091 (vect_gen_vector_loop_niters): Add a step_vector_ptr parameter.
31092 Add a path that uses a step of VF instead of 1, but disable it
31094 (vect_do_peeling): Add step_vector, niters_vector_mult_vf_var
31095 and niters_no_overflow parameters. Update calls to
31096 slpeel_make_loop_iterate_ntimes and vect_gen_vector_loop_niters.
31097 Create a new SSA name if the latter choses to use a ste other
31098 than zero, and return it via niters_vector_mult_vf_var.
31099 * tree-vect-loop.c (vect_transform_loop): Update calls to
31100 vect_do_peeling, vect_gen_vector_loop_niters and
31101 slpeel_make_loop_iterate_ntimes.
31102 * tree-vectorizer.h (slpeel_make_loop_iterate_ntimes, vect_do_peeling)
31103 (vect_gen_vector_loop_niters): Update declarations after above changes.
31105 2018-01-02 Michael Meissner <meissner@linux.vnet.ibm.com>
31107 * config/rs6000/rs6000.md (floor<mode>2): Add support for IEEE
31108 128-bit round to integer instructions.
31109 (ceil<mode>2): Likewise.
31110 (btrunc<mode>2): Likewise.
31111 (round<mode>2): Likewise.
31113 2018-01-02 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31115 * config/rs6000/rs6000-string.c (expand_block_move): Allow the use of
31116 unaligned VSX load/store on P8/P9.
31117 (expand_block_clear): Allow the use of unaligned VSX
31118 load/store on P8/P9.
31120 2018-01-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31122 * config/rs6000/rs6000-p8swap.c (swap_feeds_both_load_and_store):
31124 (rs6000_analyze_swaps): Mark a web unoptimizable if it contains a
31125 swap associated with both a load and a store.
31127 2018-01-02 Andrew Waterman <andrew@sifive.com>
31129 * config/riscv/linux.h (ICACHE_FLUSH_FUNC): New.
31130 * config/riscv/riscv.md (clear_cache): Use it.
31132 2018-01-02 Artyom Skrobov <tyomitch@gmail.com>
31134 * web.c: Remove out-of-date comment.
31136 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
31138 * expr.c (fixup_args_size_notes): Check that any existing
31139 REG_ARGS_SIZE notes are correct, and don't try to re-add them.
31140 (emit_single_push_insn_1): Move stack_pointer_delta adjustment to...
31141 (emit_single_push_insn): ...here.
31143 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
31145 * rtl.h (CONST_VECTOR_ELT): Redefine to const_vector_elt.
31146 (const_vector_encoded_nelts): New function.
31147 (CONST_VECTOR_NUNITS): Redefine to use GET_MODE_NUNITS.
31148 (const_vector_int_elt, const_vector_elt): Declare.
31149 * emit-rtl.c (const_vector_int_elt_1): New function.
31150 (const_vector_elt): Likewise.
31151 * simplify-rtx.c (simplify_immed_subreg): Avoid taking the address
31152 of CONST_VECTOR_ELT.
31154 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
31156 * expr.c: Include rtx-vector-builder.h.
31157 (const_vector_mask_from_tree): Use rtx_vector_builder and operate
31158 directly on the tree encoding.
31159 (const_vector_from_tree): Likewise.
31160 * optabs.c: Include rtx-vector-builder.h.
31161 (expand_vec_perm_var): Use rtx_vector_builder and create a repeating
31162 sequence of "u" values.
31163 * vec-perm-indices.c: Include rtx-vector-builder.h.
31164 (vec_perm_indices_to_rtx): Use rtx_vector_builder and operate
31165 directly on the vec_perm_indices encoding.
31167 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
31169 * doc/rtl.texi (const_vector): Describe new encoding scheme.
31170 * Makefile.in (OBJS): Add rtx-vector-builder.o.
31171 * rtx-vector-builder.h: New file.
31172 * rtx-vector-builder.c: Likewise.
31173 * rtl.h (rtx_def::u2): Add a const_vector field.
31174 (CONST_VECTOR_NPATTERNS): New macro.
31175 (CONST_VECTOR_NELTS_PER_PATTERN): Likewise.
31176 (CONST_VECTOR_DUPLICATE_P): Likewise.
31177 (CONST_VECTOR_STEPPED_P): Likewise.
31178 (CONST_VECTOR_ENCODED_ELT): Likewise.
31179 (const_vec_duplicate_p): Check for a duplicated vector encoding.
31180 (unwrap_const_vec_duplicate): Likewise.
31181 (const_vec_series_p): Check for a non-duplicated vector encoding.
31182 Say that the function only returns true for integer vectors.
31183 * emit-rtl.c: Include rtx-vector-builder.h.
31184 (gen_const_vec_duplicate_1): Delete.
31185 (gen_const_vector): Call gen_const_vec_duplicate instead of
31186 gen_const_vec_duplicate_1.
31187 (const_vec_series_p_1): Operate directly on the CONST_VECTOR encoding.
31188 (gen_const_vec_duplicate): Use rtx_vector_builder.
31189 (gen_const_vec_series): Likewise.
31190 (gen_rtx_CONST_VECTOR): Likewise.
31191 * config/powerpcspe/powerpcspe.c: Include rtx-vector-builder.h.
31192 (swap_const_vector_halves): Take an rtx pointer rather than rtx.
31193 Build a new vector rather than modifying a CONST_VECTOR in-place.
31194 (handle_special_swappables): Update call accordingly.
31195 * config/rs6000/rs6000-p8swap.c: Include rtx-vector-builder.h.
31196 (swap_const_vector_halves): Take an rtx pointer rather than rtx.
31197 Build a new vector rather than modifying a CONST_VECTOR in-place.
31198 (handle_special_swappables): Update call accordingly.
31200 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
31202 * simplify-rtx.c (simplify_const_binary_operation): Use
31203 CONST_VECTOR_ELT instead of XVECEXP.
31205 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
31207 * tree-cfg.c (verify_gimple_assign_ternary): Allow the size of
31208 the selector elements to be different from the data elements
31209 if the selector is a VECTOR_CST.
31210 * tree-vect-stmts.c (vect_gen_perm_mask_any): Use a vector of
31211 ssizetype for the selector.
31213 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
31215 * optabs.c (shift_amt_for_vec_perm_mask): Try using series_p
31216 before testing each element individually.
31217 * tree-vect-generic.c (lower_vec_perm): Likewise.
31219 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
31221 * selftest.h (selftest::vec_perm_indices_c_tests): Declare.
31222 * selftest-run-tests.c (selftest::run_tests): Call it.
31223 * vector-builder.h (vector_builder::operator ==): New function.
31224 (vector_builder::operator !=): Likewise.
31225 * vec-perm-indices.h (vec_perm_indices::series_p): Declare.
31226 (vec_perm_indices::all_from_input_p): New function.
31227 * vec-perm-indices.c (vec_perm_indices::series_p): Likewise.
31228 (test_vec_perm_12, selftest::vec_perm_indices_c_tests): Likewise.
31229 * fold-const.c (fold_ternary_loc): Use tree_to_vec_perm_builder
31230 instead of reading the VECTOR_CST directly. Detect whether both
31231 vector inputs are the same before constructing the vec_perm_indices,
31232 and update the number of inputs argument accordingly. Use the
31233 utility functions added above. Only construct sel2 if we need to.
31235 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
31237 * optabs.c (expand_vec_perm_var): Use an explicit encoding for
31238 the broadcast of the low byte.
31239 (expand_mult_highpart): Use an explicit encoding for the permutes.
31240 * optabs-query.c (can_mult_highpart_p): Likewise.
31241 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
31242 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
31243 (vectorizable_bswap): Likewise.
31244 * tree-vect-data-refs.c (vect_grouped_store_supported): Use an
31245 explicit encoding for the power-of-2 permutes.
31246 (vect_permute_store_chain): Likewise.
31247 (vect_grouped_load_supported): Likewise.
31248 (vect_permute_load_chain): Likewise.
31250 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
31252 * vec-perm-indices.h (vec_perm_indices_to_tree): Declare.
31253 * vec-perm-indices.c (vec_perm_indices_to_tree): New function.
31254 * tree-ssa-forwprop.c (simplify_vector_constructor): Use it.
31255 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
31256 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
31257 (vect_gen_perm_mask_any): Likewise.
31259 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
31261 * int-vector-builder.h: New file.
31262 * vec-perm-indices.h: Include int-vector-builder.h.
31263 (vec_perm_indices): Redefine as an int_vector_builder.
31264 (auto_vec_perm_indices): Delete.
31265 (vec_perm_builder): Redefine as a stand-alone class.
31266 (vec_perm_indices::vec_perm_indices): New function.
31267 (vec_perm_indices::clamp): Likewise.
31268 * vec-perm-indices.c: Include fold-const.h and tree-vector-builder.h.
31269 (vec_perm_indices::new_vector): New function.
31270 (vec_perm_indices::new_expanded_vector): Update for new
31271 vec_perm_indices class.
31272 (vec_perm_indices::rotate_inputs): New function.
31273 (vec_perm_indices::all_in_range_p): Operate directly on the
31274 encoded form, without computing elided elements.
31275 (tree_to_vec_perm_builder): Operate directly on the VECTOR_CST
31276 encoding. Update for new vec_perm_indices class.
31277 * optabs.c (expand_vec_perm_const): Create a vec_perm_indices for
31278 the given vec_perm_builder.
31279 (expand_vec_perm_var): Update vec_perm_builder constructor.
31280 (expand_mult_highpart): Use vec_perm_builder instead of
31281 auto_vec_perm_indices.
31282 * optabs-query.c (can_mult_highpart_p): Use vec_perm_builder and
31283 vec_perm_indices instead of auto_vec_perm_indices. Use a single
31284 or double series encoding as appropriate.
31285 * fold-const.c (fold_ternary_loc): Use vec_perm_builder and
31286 vec_perm_indices instead of auto_vec_perm_indices.
31287 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
31288 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
31289 (vect_permute_store_chain): Likewise.
31290 (vect_grouped_load_supported): Likewise.
31291 (vect_permute_load_chain): Likewise.
31292 (vect_shift_permute_load_chain): Likewise.
31293 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
31294 (vect_transform_slp_perm_load): Likewise.
31295 (vect_schedule_slp_instance): Likewise.
31296 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
31297 (vectorizable_mask_load_store): Likewise.
31298 (vectorizable_bswap): Likewise.
31299 (vectorizable_store): Likewise.
31300 (vectorizable_load): Likewise.
31301 * tree-vect-generic.c (lower_vec_perm): Use vec_perm_builder and
31302 vec_perm_indices instead of auto_vec_perm_indices. Use
31303 tree_to_vec_perm_builder to read the vector from a tree.
31304 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Take a
31305 vec_perm_builder instead of a vec_perm_indices.
31306 (have_whole_vector_shift): Use vec_perm_builder and
31307 vec_perm_indices instead of auto_vec_perm_indices. Leave the
31308 truncation to calc_vec_perm_mask_for_shift.
31309 (vect_create_epilog_for_reduction): Likewise.
31310 * config/aarch64/aarch64.c (expand_vec_perm_d::perm): Change
31311 from auto_vec_perm_indices to vec_perm_indices.
31312 (aarch64_expand_vec_perm_const_1): Use rotate_inputs on d.perm
31313 instead of changing individual elements.
31314 (aarch64_vectorize_vec_perm_const): Use new_vector to install
31315 the vector in d.perm.
31316 * config/arm/arm.c (expand_vec_perm_d::perm): Change
31317 from auto_vec_perm_indices to vec_perm_indices.
31318 (arm_expand_vec_perm_const_1): Use rotate_inputs on d.perm
31319 instead of changing individual elements.
31320 (arm_vectorize_vec_perm_const): Use new_vector to install
31321 the vector in d.perm.
31322 * config/powerpcspe/powerpcspe.c (rs6000_expand_extract_even):
31323 Update vec_perm_builder constructor.
31324 (rs6000_expand_interleave): Likewise.
31325 * config/rs6000/rs6000.c (rs6000_expand_extract_even): Likewise.
31326 (rs6000_expand_interleave): Likewise.
31328 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
31330 * optabs-query.c (can_vec_perm_var_p): Check whether lowering
31331 to qimode could truncate the indices.
31332 * optabs.c (expand_vec_perm_var): Likewise.
31334 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
31336 * Makefile.in (OBJS): Add vec-perm-indices.o.
31337 * vec-perm-indices.h: New file.
31338 * vec-perm-indices.c: Likewise.
31339 * target.h (vec_perm_indices): Replace with a forward class
31341 (auto_vec_perm_indices): Move to vec-perm-indices.h.
31342 * optabs.h: Include vec-perm-indices.h.
31343 (expand_vec_perm): Delete.
31344 (selector_fits_mode_p, expand_vec_perm_var): Declare.
31345 (expand_vec_perm_const): Declare.
31346 * target.def (vec_perm_const_ok): Replace with...
31347 (vec_perm_const): ...this new hook.
31348 * doc/tm.texi.in (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Replace with...
31349 (TARGET_VECTORIZE_VEC_PERM_CONST): ...this new hook.
31350 * doc/tm.texi: Regenerate.
31351 * optabs.def (vec_perm_const): Delete.
31352 * doc/md.texi (vec_perm_const): Likewise.
31353 (vec_perm): Refer to TARGET_VECTORIZE_VEC_PERM_CONST.
31354 * expr.c (expand_expr_real_2): Use expand_vec_perm_const rather than
31355 expand_vec_perm for constant permutation vectors. Assert that
31356 the mode of variable permutation vectors is the integer equivalent
31357 of the mode that is being permuted.
31358 * optabs-query.h (selector_fits_mode_p): Declare.
31359 * optabs-query.c: Include vec-perm-indices.h.
31360 (selector_fits_mode_p): New function.
31361 (can_vec_perm_const_p): Check whether targetm.vectorize.vec_perm_const
31362 is defined, instead of checking whether the vec_perm_const_optab
31363 exists. Use targetm.vectorize.vec_perm_const instead of
31364 targetm.vectorize.vec_perm_const_ok. Check whether the indices
31365 fit in the vector mode before using a variable permute.
31366 * optabs.c (shift_amt_for_vec_perm_mask): Take a mode and a
31367 vec_perm_indices instead of an rtx.
31368 (expand_vec_perm): Replace with...
31369 (expand_vec_perm_const): ...this new function. Take the selector
31370 as a vec_perm_indices rather than an rtx. Also take the mode of
31371 the selector. Update call to shift_amt_for_vec_perm_mask.
31372 Use targetm.vectorize.vec_perm_const instead of vec_perm_const_optab.
31373 Use vec_perm_indices::new_expanded_vector to expand the original
31374 selector into bytes. Check whether the indices fit in the vector
31375 mode before using a variable permute.
31376 (expand_vec_perm_var): Make global.
31377 (expand_mult_highpart): Use expand_vec_perm_const.
31378 * fold-const.c: Includes vec-perm-indices.h.
31379 * tree-ssa-forwprop.c: Likewise.
31380 * tree-vect-data-refs.c: Likewise.
31381 * tree-vect-generic.c: Likewise.
31382 * tree-vect-loop.c: Likewise.
31383 * tree-vect-slp.c: Likewise.
31384 * tree-vect-stmts.c: Likewise.
31385 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm_const):
31387 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>): Delete.
31388 * config/aarch64/aarch64.c (aarch64_expand_vec_perm_const)
31389 (aarch64_vectorize_vec_perm_const_ok): Fuse into...
31390 (aarch64_vectorize_vec_perm_const): ...this new function.
31391 (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
31392 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
31393 * config/arm/arm-protos.h (arm_expand_vec_perm_const): Delete.
31394 * config/arm/vec-common.md (vec_perm_const<mode>): Delete.
31395 * config/arm/arm.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
31396 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
31397 (arm_expand_vec_perm_const, arm_vectorize_vec_perm_const_ok): Merge
31399 (arm_vectorize_vec_perm_const): ...this new function. Explicitly
31400 check for NEON modes.
31401 * config/i386/i386-protos.h (ix86_expand_vec_perm_const): Delete.
31402 * config/i386/sse.md (VEC_PERM_CONST, vec_perm_const<mode>): Delete.
31403 * config/i386/i386.c (ix86_expand_vec_perm_const_1): Update comment.
31404 (ix86_expand_vec_perm_const, ix86_vectorize_vec_perm_const_ok): Merge
31406 (ix86_vectorize_vec_perm_const): ...this new function. Incorporate
31407 the old VEC_PERM_CONST conditions.
31408 * config/ia64/ia64-protos.h (ia64_expand_vec_perm_const): Delete.
31409 * config/ia64/vect.md (vec_perm_const<mode>): Delete.
31410 * config/ia64/ia64.c (ia64_expand_vec_perm_const)
31411 (ia64_vectorize_vec_perm_const_ok): Merge into...
31412 (ia64_vectorize_vec_perm_const): ...this new function.
31413 * config/mips/loongson.md (vec_perm_const<mode>): Delete.
31414 * config/mips/mips-msa.md (vec_perm_const<mode>): Delete.
31415 * config/mips/mips-ps-3d.md (vec_perm_constv2sf): Delete.
31416 * config/mips/mips-protos.h (mips_expand_vec_perm_const): Delete.
31417 * config/mips/mips.c (mips_expand_vec_perm_const)
31418 (mips_vectorize_vec_perm_const_ok): Merge into...
31419 (mips_vectorize_vec_perm_const): ...this new function.
31420 * config/powerpcspe/altivec.md (vec_perm_constv16qi): Delete.
31421 * config/powerpcspe/paired.md (vec_perm_constv2sf): Delete.
31422 * config/powerpcspe/spe.md (vec_perm_constv2si): Delete.
31423 * config/powerpcspe/vsx.md (vec_perm_const<mode>): Delete.
31424 * config/powerpcspe/powerpcspe-protos.h (altivec_expand_vec_perm_const)
31425 (rs6000_expand_vec_perm_const): Delete.
31426 * config/powerpcspe/powerpcspe.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK):
31428 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
31429 (altivec_expand_vec_perm_const_le): Take each operand individually.
31430 Operate on constant selectors rather than rtxes.
31431 (altivec_expand_vec_perm_const): Likewise. Update call to
31432 altivec_expand_vec_perm_const_le.
31433 (rs6000_expand_vec_perm_const): Delete.
31434 (rs6000_vectorize_vec_perm_const_ok): Delete.
31435 (rs6000_vectorize_vec_perm_const): New function.
31436 (rs6000_do_expand_vec_perm): Take a vec_perm_builder instead of
31437 an element count and rtx array.
31438 (rs6000_expand_extract_even): Update call accordingly.
31439 (rs6000_expand_interleave): Likewise.
31440 * config/rs6000/altivec.md (vec_perm_constv16qi): Delete.
31441 * config/rs6000/paired.md (vec_perm_constv2sf): Delete.
31442 * config/rs6000/vsx.md (vec_perm_const<mode>): Delete.
31443 * config/rs6000/rs6000-protos.h (altivec_expand_vec_perm_const)
31444 (rs6000_expand_vec_perm_const): Delete.
31445 * config/rs6000/rs6000.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
31446 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
31447 (altivec_expand_vec_perm_const_le): Take each operand individually.
31448 Operate on constant selectors rather than rtxes.
31449 (altivec_expand_vec_perm_const): Likewise. Update call to
31450 altivec_expand_vec_perm_const_le.
31451 (rs6000_expand_vec_perm_const): Delete.
31452 (rs6000_vectorize_vec_perm_const_ok): Delete.
31453 (rs6000_vectorize_vec_perm_const): New function. Remove stray
31454 reference to the SPE evmerge intructions.
31455 (rs6000_do_expand_vec_perm): Take a vec_perm_builder instead of
31456 an element count and rtx array.
31457 (rs6000_expand_extract_even): Update call accordingly.
31458 (rs6000_expand_interleave): Likewise.
31459 * config/sparc/sparc.md (vec_perm_constv8qi): Delete in favor of...
31460 * config/sparc/sparc.c (sparc_vectorize_vec_perm_const): ...this
31462 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
31464 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
31466 * optabs.c (expand_vec_perm_1): Assert that SEL has an integer
31467 vector mode and that that mode matches the mode of the data
31469 (expand_vec_perm): Split handling of non-CONST_VECTOR selectors
31470 out into expand_vec_perm_var. Do all CONST_VECTOR handling here,
31471 directly using expand_vec_perm_1 when forcing selectors into
31473 (expand_vec_perm_var): New function, split out from expand_vec_perm.
31475 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
31477 * optabs-query.h (can_vec_perm_p): Delete.
31478 (can_vec_perm_var_p, can_vec_perm_const_p): Declare.
31479 * optabs-query.c (can_vec_perm_p): Split into...
31480 (can_vec_perm_var_p, can_vec_perm_const_p): ...these two functions.
31481 (can_mult_highpart_p): Use can_vec_perm_const_p to test whether a
31482 particular selector is valid.
31483 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
31484 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
31485 (vect_grouped_load_supported): Likewise.
31486 (vect_shift_permute_load_chain): Likewise.
31487 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
31488 (vect_transform_slp_perm_load): Likewise.
31489 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
31490 (vectorizable_bswap): Likewise.
31491 (vect_gen_perm_mask_checked): Likewise.
31492 * fold-const.c (fold_ternary_loc): Likewise. Don't take
31493 implementations of variable permutation vectors into account
31494 when deciding which selector to use.
31495 * tree-vect-loop.c (have_whole_vector_shift): Don't check whether
31496 vec_perm_const_optab is supported; instead use can_vec_perm_const_p
31497 with a false third argument.
31498 * tree-vect-generic.c (lower_vec_perm): Use can_vec_perm_const_p
31499 to test whether the constant selector is valid and can_vec_perm_var_p
31500 to test whether a variable selector is valid.
31502 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
31504 * optabs-query.h (can_vec_perm_p): Take a const vec_perm_indices *.
31505 * optabs-query.c (can_vec_perm_p): Likewise.
31506 * fold-const.c (fold_vec_perm): Take a const vec_perm_indices &
31507 instead of vec_perm_indices.
31508 * tree-vectorizer.h (vect_gen_perm_mask_any): Likewise,
31509 (vect_gen_perm_mask_checked): Likewise,
31510 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise,
31511 (vect_gen_perm_mask_checked): Likewise,
31513 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
31515 * optabs-query.h (qimode_for_vec_perm): Declare.
31516 * optabs-query.c (can_vec_perm_p): Split out qimode search to...
31517 (qimode_for_vec_perm): ...this new function.
31518 * optabs.c (expand_vec_perm): Use qimode_for_vec_perm.
31520 2018-01-02 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31522 * rtlanal.c (canonicalize_condition): Return 0 if final rtx
31523 does not have a conditional at the top.
31525 2018-01-02 Richard Biener <rguenther@suse.de>
31527 * ipa-inline.c (big_speedup_p): Fix expression.
31529 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
31532 * config/i386/x86-tune-costs.h: Increase cost of integer load costs
31535 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
31539 * x86-tune-costs.h (generic_cost): Reduce cost of FDIV 20->17,
31540 cost of sqrt 20->14, DIVSS 18->13, DIVSD 32->17, SQRtSS 30->14
31541 and SQRTsD 58->18, cond_not_taken_branch_cost. 2->1. Increase
31542 cond_taken_branch_cost 3->4.
31544 2018-01-01 Jakub Jelinek <jakub@redhat.com>
31546 PR tree-optimization/83581
31547 * tree-loop-distribution.c (pass_loop_distribution::execute): Return
31548 TODO_cleanup_cfg if any changes have been made.
31550 PR middle-end/83608
31551 * expr.c (store_expr_with_bounds): Use simplify_gen_subreg instead of
31552 convert_modes if target mode has the right side, but different mode
31555 PR middle-end/83609
31556 * expr.c (expand_assignment): Fix up a typo in simplify_gen_subreg
31557 last argument when extracting from CONCAT. If either from_real or
31558 from_imag is NULL, use expansion through memory. If result is not
31559 a CONCAT and simplify_gen_subreg fails, try to simplify_gen_subreg
31560 the parts directly to inner mode, if even that fails, use expansion
31563 PR middle-end/83623
31564 * expmed.c (expand_shift_1): For 2-byte rotates by BITS_PER_UNIT,
31565 check for bswap in mode rather than HImode and use that in expand_unop
31568 Copyright (C) 2018 Free Software Foundation, Inc.
31570 Copying and distribution of this file, with or without modification,
31571 are permitted in any medium without royalty provided the copyright
31572 notice and this notice are preserved.