platform/upstream/gcc.git
8 years ago* de.po: Update.
Joseph Myers [Tue, 9 Feb 2016 16:20:46 +0000 (16:20 +0000)]
* de.po: Update.

From-SVN: r233246

8 years agovect-mask-store-move-1.c: Add missing space.
Richard Biener [Tue, 9 Feb 2016 15:04:48 +0000 (15:04 +0000)]
vect-mask-store-move-1.c: Add missing space.

2016-02-09  Richard Biener  <rguenther@suse.de>

* gcc.dg/vect/vect-mask-store-move-1.c: Add missing space.

From-SVN: r233245

8 years agoFix expectation in AVX512 logic instructions test.
Alexander Fomin [Tue, 9 Feb 2016 14:36:03 +0000 (14:36 +0000)]
Fix expectation in AVX512 logic instructions test.

gcc/testsuite

* gcc.target/i386/avx512f-klogic-2.c: Adjust expectation.

From-SVN: r233243

8 years agovect-mask-store-move-1.c: Fix test failures.
Yuri Rumyantsev [Tue, 9 Feb 2016 09:51:21 +0000 (09:51 +0000)]
vect-mask-store-move-1.c: Fix test failures.

gcc/testsuite/

2016-02-09  Yuri Rumyantsev  <ysrumyan@gmail.com>

* gcc.dg/vect/vect-mask-store-move-1.c: Fix test failures.

From-SVN: r233241

8 years agoFix GOMP/GOACC_parallel optimization in ipa-pta
Tom de Vries [Tue, 9 Feb 2016 08:52:26 +0000 (08:52 +0000)]
Fix GOMP/GOACC_parallel optimization in ipa-pta

2016-02-09  Tom de Vries  <tom@codesourcery.com>

PR tree-optimization/69599
* tree-ssa-structalias.c (fndecl_maybe_in_other_partition): New
function.
(find_func_aliases_for_builtin_call, find_func_clobbers)
(ipa_pta_execute):  Handle case that foo and foo._0 are not in same lto
partition.

* testsuite/libgomp.c/omp-nested-3.c: New test.
* testsuite/libgomp.c/pr46032-2.c: New test.
* testsuite/libgomp.oacc-c-c++-common/kernels-2.c: New test.
* testsuite/libgomp.oacc-c-c++-common/parallel-2.c: New test.

From-SVN: r233240

8 years agore PR middle-end/69715 (ICE: in store_bit_field_1, at expmed.c:839)
Richard Biener [Tue, 9 Feb 2016 08:35:22 +0000 (08:35 +0000)]
re PR middle-end/69715 (ICE: in store_bit_field_1, at expmed.c:839)

2016-02-09  Richard Biener  <rguenther@suse.de>

PR tree-optimization/69715
* tree-ssa.c (execute_update_addresses_taken): Mark non-decl
LHS on calls as non-rewritable.

* gcc.dg/torture/pr69715.c: New testcase.

From-SVN: r233239

8 years agoAdd missing entry
Bernd Schmidt [Tue, 9 Feb 2016 07:12:22 +0000 (07:12 +0000)]
Add missing entry

From-SVN: r233238

8 years agoHandle -fdiagnostics-color in lto
Tom de Vries [Tue, 9 Feb 2016 07:04:08 +0000 (07:04 +0000)]
Handle -fdiagnostics-color in lto

2016-02-09  Tom de Vries  <tom@codesourcery.com>

PR lto/69707
* lto-wrapper.c (append_diag_options): New function.
(compile_offload_image): Call append_diag_options.

* testsuite/libgomp.oacc-c-c++-common/parallel-dims-2.c: New test.

From-SVN: r233237

8 years agore PR other/69722 (gcc/doc/extend.texi:7526: warning: node `Constraints' is next...
Sandra Loosemore [Tue, 9 Feb 2016 03:25:30 +0000 (22:25 -0500)]
re PR other/69722 (gcc/doc/extend.texi:7526: warning: node `Constraints' is next for `Extended Asm' in menu but not in sectioning)

2016-02-08  Sandra Loosemore <sandra@codesourcery.com>

PR other/69722

gcc/
* doc/extend.texi (Flag Output Operands): Correct sectioning.
Minor copy-edit to fix verb tenses.

From-SVN: r233236

8 years agore PR go/69357 (libgo refers to _end in a non-weak way)
Ian Lance Taylor [Tue, 9 Feb 2016 00:34:55 +0000 (00:34 +0000)]
re PR go/69357 (libgo refers to _end in a non-weak way)

PR go/69537
    runtime: Don't refer to _end symbol in shared library.

    Fixes GCC PR 69357.

    Reviewed-on: https://go-review.googlesource.com/19362

From-SVN: r233235

8 years agoDaily bump.
GCC Administrator [Tue, 9 Feb 2016 00:16:13 +0000 (00:16 +0000)]
Daily bump.

From-SVN: r233234

8 years agoFix PR c++/69139 (deduction failure with trailing return type)
Patrick Palka [Mon, 8 Feb 2016 23:06:21 +0000 (23:06 +0000)]
Fix PR c++/69139 (deduction failure with trailing return type)

gcc/cp/ChangeLog:

PR c++/69139
* parser.c (cp_parser_simple_type_specifier): Make the check
for disambiguating between an 'auto' placeholder and an implicit
template parameter more robust.

gcc/testsuite/ChangeLog:

PR c++/69139
* g++.dg/cpp0x/trailing12.C: New test.
* g++.dg/cpp0x/trailing13.C: New test.

From-SVN: r233231

8 years agoFix PR c++/69283 (auto deduction fails when ADL is required)
Patrick Palka [Mon, 8 Feb 2016 23:02:50 +0000 (23:02 +0000)]
Fix PR c++/69283 (auto deduction fails when ADL is required)

gcc/cp/ChangeLog:

PR c++/69283
PR c++/67835
* decl2.c (mark_used): When given a TEMPLATE_DECL, return after
setting its TREE_USED flag.

gcc/testsuite/ChangeLog:

PR c++/69283
PR c++/67835
* g++.dg/cpp1y/auto-fn29.C: New test.
* g++.dg/cpp1y/auto-fn30.C: New test.

From-SVN: r233230

8 years agore PR c++/69657 (abs() not inlined after including math.h)
Jason Merrill [Mon, 8 Feb 2016 22:07:54 +0000 (17:07 -0500)]
re PR c++/69657 (abs() not inlined after including math.h)

PR c++/69657
* name-lookup.c (do_nonmember_using_decl): Leave anticipated
built-ins alone.

From-SVN: r233229

8 years agore PR tree-optimization/69209 (ICE at -Os and above on x86_64-linux-gnu (verify_gimpl...
Jakub Jelinek [Mon, 8 Feb 2016 20:07:56 +0000 (21:07 +0100)]
re PR tree-optimization/69209 (ICE at -Os and above on x86_64-linux-gnu (verify_gimple failed))

PR tree-optimization/69209
* ipa-split.c (split_function): If split part is not
returning retval, retval has gimple type but is not
gimple value, force it into a SSA_NAME first.

* gcc.c-torture/compile/pr69209.c: New test.

From-SVN: r233228

8 years agore PR tree-optimization/68541 (Path splitting causes if-conversion miss)
Jeff Law [Mon, 8 Feb 2016 19:52:50 +0000 (12:52 -0700)]
re PR tree-optimization/68541 (Path splitting causes if-conversion miss)

PR tree-optimization/68541
* gcc.dg/tree-ssa/split-path-1.c: Increase limit for number
of statements allowed in join block for path splitting.

From-SVN: r233227

8 years agore PR c++/59627 (ICE with OpenMP "declare reduction" and -flto)
Jakub Jelinek [Mon, 8 Feb 2016 18:23:03 +0000 (19:23 +0100)]
re PR c++/59627 (ICE with OpenMP "declare reduction" and -flto)

PR c++/59627
* parser.c (cp_parser_omp_declare_reduction): Set assembler name
of the DECL_OMP_DECLARE_REDUCTION_P decls.

* g++.dg/gomp/pr59627.C: New test.

From-SVN: r233225

8 years agore PR ipa/69239 (g++.dg/ipa/devirt-c-3.C FAILs with -O2 -fPIC --param=early-inlining...
Jakub Jelinek [Mon, 8 Feb 2016 17:41:35 +0000 (18:41 +0100)]
re PR ipa/69239 (g++.dg/ipa/devirt-c-3.C FAILs with -O2 -fPIC --param=early-inlining-insns=196)

PR ipa/69239
* g++.dg/ipa/pr69239.C: New test.

From-SVN: r233224

8 years agoPR preprocessor/69664: fix rich_location::override_column
David Malcolm [Mon, 8 Feb 2016 17:33:45 +0000 (17:33 +0000)]
PR preprocessor/69664: fix rich_location::override_column

gcc/testsuite/ChangeLog:
PR preprocessor/69664
* gcc.dg/cpp/trad/comment-2.c: Add expected column number.
* gcc.dg/cpp/warn-comments.c: Likewise.

libcpp/ChangeLog:
PR preprocessor/69664
* errors.c (cpp_diagnostic_with_line): Only call
rich_location::override_column if the column is non-zero.
* line-map.c (rich_location::override_column): Update columns
within m_ranges[0].  Add assertions to verify that doing so is
sane.

From-SVN: r233223

8 years agojit: fix build after r233218 (build_common_tree_nodes)
David Malcolm [Mon, 8 Feb 2016 17:18:57 +0000 (17:18 +0000)]
jit: fix build after r233218 (build_common_tree_nodes)

gcc/jit/ChangeLog:
* dummy-frontend.c (jit_langhook_init): Remove
second argument to build_common_tree_nodes to
track r233218.

From-SVN: r233222

8 years agotree-ssa.texi (Preserving the virtual SSA form): Remove outdated section.
Nicklas Bo Jensen [Mon, 8 Feb 2016 16:35:24 +0000 (16:35 +0000)]
tree-ssa.texi (Preserving the virtual SSA form): Remove outdated section.

* doc/tree-ssa.texi (Preserving the virtual SSA form): Remove
outdated section.

From-SVN: r233221

8 years agore PR c++/69688 (-Wsign-compare causes bogus error: size of array ‘uc_code’ is not...
Marek Polacek [Mon, 8 Feb 2016 15:40:33 +0000 (15:40 +0000)]
re PR c++/69688 (-Wsign-compare causes bogus error: size of array ‘uc_code’ is not an integral constant-expression)

PR c++/69688
* constexpr.c (clear_cv_and_fold_caches): Renamed from clear_cv_cache.
Call clear_fold_cache.
* cp-tree.h: Adjust declaration.
* decl.c (finish_enum_value_list): Call clear_cv_and_fold_caches
rather than clear_cv_cache and clear_fold_cache.
* typeck2.c (store_init_value): Call clear_cv_and_fold_caches.

* g++.dg/init/const12.C: New test.

From-SVN: r233220

8 years agoRemove accidentally added 'constexpr' in previous commit
Jonathan Wakely [Mon, 8 Feb 2016 15:37:59 +0000 (15:37 +0000)]
Remove accidentally added 'constexpr' in previous commit

* acinclude.m4 (GLIBCXX_CHECK_MATH11_PROTO): Remove accidentally
added 'constexpr' in previous commit.
* configure: Regenerate.

From-SVN: r233219

8 years agoRemove -fshort-double
Bernd Schmidt [Mon, 8 Feb 2016 15:36:16 +0000 (15:36 +0000)]
Remove -fshort-double

PR target/60410
* tree.c (build_common_tree_nodes): Remove short_double argument.
All callers changed.
* tree.h (build_common_tree_nodes): Adjust declaration.
* doc/invoke.texi (-fshort-double): Remove documentation.
* config/mips/t-img-elf (MULTILIB_OPTIONS, MULTILIB_DIRNAMES,
MULTILIB_EXCEPTIONS): Remove -fshort-double variant.
* lto-wrapper.c (merge_and_complain, append_compiler_options,
append_linker_options): Don't handle OPT_fshort_double.

c-family/
PR target/60410
* c.opt (fshort-double): Remove.

testsuite/
PR target/60410
* gcc.dg/lto/pr55113_0.c: Remove test.

From-SVN: r233218

8 years ago* convert.c (convert_to_integer_1): Re-indent.
Jason Merrill [Mon, 8 Feb 2016 15:31:53 +0000 (10:31 -0500)]
* convert.c (convert_to_integer_1): Re-indent.

From-SVN: r233217

8 years agore PR c++/69631 (Bogus overflow in constant expression error)
Jason Merrill [Mon, 8 Feb 2016 15:31:47 +0000 (10:31 -0500)]
re PR c++/69631 (Bogus overflow in constant expression error)

PR c++/69631

gcc/
* convert.c (convert_to_integer_1): Check dofold on truncation
distribution.
(convert_to_pointer_maybe_fold, convert_to_real_maybe_fold)
(convert_to_integer_maybe_fold, convert_to_complex_maybe_fold):
Rename from *_nofold.
* convert.h (convert_to_pointer_nofold, convert_to_integer_nofold)
(convert_to_real_nofold, convert_to_complex_nofold): New inlines.
gcc/cp/
* cp-tree.h (CONV_FOLD, CONV_BACKEND_CONVERT): New.
* cvt.c (convert): Pass CONV_BACKEND_CONVERT.
(ocp_convert): Use *_maybe_fold.
(cp_convert_to_pointer): Add dofold parameter.
* cp-gimplify.c (cp_fold) [CONVERT_EXPR]: Call convert.

From-SVN: r233216

8 years agoFix latent LRA remat issue (PR68730)
Bernd Schmidt [Mon, 8 Feb 2016 15:31:08 +0000 (15:31 +0000)]
Fix latent LRA remat issue (PR68730)

PR rtl-optimization/68730
* lra-remat.c (insn_to_cand_activation): New static variable.
(lra_remat): Allocate and free it.
(create_cand): New arg activation. Initialize a field in
insn_to_cand_activation if it is nonnull.
(create_cands): Pass the activation insn to create_cand when making
a candidate involving an output reload.  Reorganize code a little.
(do_remat): Keep track of active status of candidates in a separate
bitmap.

From-SVN: r233215

8 years agoEnable isinf/isnan checks for all targets
Jonathan Wakely [Mon, 8 Feb 2016 15:22:32 +0000 (15:22 +0000)]
Enable isinf/isnan checks for all targets

PR libstdc++/48891
* acinclude.m4 (GLIBCXX_CHECK_MATH11_PROTO): Enable isinf and isnan
checks for all targets except *-*-solaris2.* and ensure we find the
libc math.h header not our own.
* configure: Regenerate.

From-SVN: r233214

8 years agore PR tree-optimization/69719 (wrong code at -O3 on x86_64-linux-gnu)
Richard Biener [Mon, 8 Feb 2016 14:51:20 +0000 (14:51 +0000)]
re PR tree-optimization/69719 (wrong code at -O3 on x86_64-linux-gnu)

2016-02-08  Richard Biener  <rguenther@suse.de>

PR tree-optimization/69719
* tree-vect-data-refs.c (vect_prune_runtime_alias_test_list):
Properly use absolute of the difference of the two offsets to
compare or adjust the segment length.

* gcc.dg/torture/pr69719.c: New testcase.

From-SVN: r233212

8 years agore PR target/68273 (Wrong code on mips/mipsel due to (invalid?) peeking at alignments...
Richard Biener [Mon, 8 Feb 2016 10:04:18 +0000 (10:04 +0000)]
re PR target/68273 (Wrong code on mips/mipsel due to (invalid?) peeking at alignments in function_arg.)

2016-02-08  Richard Biener  <rguenther@suse.de>
Jeff Law  <law@redhat.com>

PR target/68273
* tree-ssanames.c (make_ssa_name_fn): Always use unqualified
types for anonymous SSA names.

* gcc.target/mips/pr68273.c: New testcase.

Co-Authored-By: Jeff Law <law@redhat.com>
From-SVN: r233211

8 years agoutils.c (create_var_decl): Set again DECL_COMMON and DECL_IGNORED_P last.
Eric Botcazou [Mon, 8 Feb 2016 09:40:46 +0000 (09:40 +0000)]
utils.c (create_var_decl): Set again DECL_COMMON and DECL_IGNORED_P last.

* gcc-interface/utils.c (create_var_decl): Set again DECL_COMMON and
DECL_IGNORED_P last.

From-SVN: r233210

8 years agore PR target/69274 (435.gromacs performance regression after r231814 on x86 Haswell...
Richard Biener [Mon, 8 Feb 2016 09:09:22 +0000 (09:09 +0000)]
re PR target/69274 (435.gromacs performance regression  after r231814 on x86 Haswell and bdver2)

2016-02-08   Richard Biener  <rguenther@suse.de>

PR rtl-optimization/69274
* ira.c (ira_setup_alts): Do not change recog_data.operand
order.

* gcc.target/i386/addr-sel-1.c: XFAIL.

From-SVN: r233209

8 years agore PR tree-optimization/68541 (Path splitting causes if-conversion miss)
Jeff Law [Mon, 8 Feb 2016 08:40:00 +0000 (01:40 -0700)]
re PR tree-optimization/68541 (Path splitting causes if-conversion miss)

        PR tree-optimization/68541
        * gcc.dg/tree-ssa/split-path-2.c: Make char array explicitly signed.

From-SVN: r233208

8 years agore PR tree-optimization/65917 (XFAIL: gcc.dg/tree-ssa/20030922-2.c scan-tree-dump...
Jeff Law [Mon, 8 Feb 2016 08:17:32 +0000 (01:17 -0700)]
re PR tree-optimization/65917 (XFAIL: gcc.dg/tree-ssa/20030922-2.c scan-tree-dump-times dom1 "if " 2)

PR tree-optimization/65917
* tree-ssa-dom.c (record_temporary_equivalences): Record both
equivalences from if (x == y) style conditionals.
(loop_depth_of_name): Remove.
(record_equality): Remove loop depth check.
* tree-ssa-scopedtables.h (const_and_copies): Refine comments.
(const_and_copies::record_const_or_copy_raw): New member function.
* tree-ssa-scopedtables.c
(const_and_copies::record_const_or_copy_raw): New, factored out of
(const_and_copies::record_const_or_copy): Call new member function.

        PR tree-optimization/65917
* gcc.dg/tree-ssa/20030922-2.c: No longer xfailed.

From-SVN: r233207

8 years agoDaily bump.
GCC Administrator [Mon, 8 Feb 2016 00:16:15 +0000 (00:16 +0000)]
Daily bump.

From-SVN: r233206

8 years agore PR fortran/50555 (synonymous namelist/statement function dummy argument not allowe...
Jerry DeLisle [Sun, 7 Feb 2016 20:15:55 +0000 (20:15 +0000)]
re PR fortran/50555 (synonymous namelist/statement function dummy argument not allowed (r178939))

2016-02-07  Jerry DeLisle  <jvdelisle@gcc.gnu.org>

PR fortran/50555
* primary.c (match_actual_arg): If symbol has attribute flavor of
namelist, generate an error. (gfc_match_rvalue): Likewise return
MATCH_ERROR.
* resolve.c (resolve_symbol): Scan arument list of procedures and
generate an error if a namelist is found.

PR fortran/50555
* gfortran.dg/namelist_args.f90: New test.

From-SVN: r233203

8 years ago* ru.po: Update.
Joseph Myers [Sun, 7 Feb 2016 14:37:20 +0000 (14:37 +0000)]
* ru.po: Update.

From-SVN: r233202

8 years ago* fi.po, sv.po: Update.
Joseph Myers [Sun, 7 Feb 2016 14:34:42 +0000 (14:34 +0000)]
* fi.po, sv.po: Update.

From-SVN: r233201

8 years agoDaily bump.
GCC Administrator [Sun, 7 Feb 2016 00:16:11 +0000 (00:16 +0000)]
Daily bump.

From-SVN: r233199

8 years agommap.c (MAP_FAILED): Define if not defined.
John David Anglin [Sat, 6 Feb 2016 22:27:34 +0000 (22:27 +0000)]
mmap.c (MAP_FAILED): Define if not defined.

* mmap.c (MAP_FAILED): Define if not defined.

From-SVN: r233196

8 years agoDaily bump.
GCC Administrator [Sat, 6 Feb 2016 00:16:13 +0000 (00:16 +0000)]
Daily bump.

From-SVN: r233194

8 years agore PR tree-optimization/68541 (Path splitting causes if-conversion miss)
Jeff Law [Fri, 5 Feb 2016 23:49:08 +0000 (16:49 -0700)]
re PR tree-optimization/68541 (Path splitting causes if-conversion miss)

PR tree-optimization/68541
* gimple-ssa-split-paths.c: Include tree-cfg.h and params.h.
(count_stmts_in_block): New function.
(poor_ifcvt_candidate_code): Likewise.
(is_feasible_trace): Add some heuristics to determine when path
splitting is profitable.
(find_block_to_duplicate_for_splitting_paths): Make sure the graph
is a diamond with a single exit.

        PR tree-optimization/68541
* gcc.dg/tree-ssa/split-path-2.c: New test.
* gcc.dg/tree-ssa/split-path-3.c: New test.
* gcc.dg/tree-ssa/split-path-4.c: New test.
* gcc.dg/tree-ssa/split-path-5.c: New test.
* gcc.dg/tree-ssa/split-path-6.c: New test.
* gcc.dg/tree-ssa/split-path-7.c: New test.

From-SVN: r233191

8 years agoPR c++/69662 - -Wplacement-new on allocated one element array members
Martin Sebor [Fri, 5 Feb 2016 22:27:37 +0000 (22:27 +0000)]
PR c++/69662 - -Wplacement-new on allocated one element array members

gcc/testsuite/ChangeLog:
PR c++/69662
* g++.dg/warn/Wplacement-new-size-1.C: New test.
* g++.dg/warn/Wplacement-new-size-2.C: New test.

gcc/cp/ChangeLog:
PR c++/69662
* init.c (find_field_init): New function.
(warn_placement_new_too_small): Call it.  Handle one-element arrays
        at ends of structures special.

gcc/c-family/ChangeLog:
PR c++/69662
* c.opt (Warning options): Update -Wplacement-new to take
        an optional argument.

gcc/ChangeLog:
PR c++/69662
* doc/invoke.texi: Update -Wplacement-new to take an optional
        argument.

From-SVN: r233190

8 years agore PR middle-end/69643 (Address space discarded)
Richard Henderson [Fri, 5 Feb 2016 22:05:17 +0000 (14:05 -0800)]
re PR middle-end/69643 (Address space discarded)

PR c/69643

  * tree.c (tree_nop_conversion_p): Do not strip casts into or
  out of non-standard address spaces.
testsuite/
  * gcc.target/i386/addr-space-4.c: New.
  * gcc.target/i386/addr-space-5.c: New.

From-SVN: r233189

8 years agoFix fortran scalar elemental dependency mishandling
Mikael Morin [Fri, 5 Feb 2016 21:41:15 +0000 (21:41 +0000)]
Fix fortran scalar elemental dependency mishandling

PR fortran/66089
gcc/fortran/
* trans-expr.c (expr_is_variable, gfc_expr_is_variable): Rename
the former to the latter and make it non-static.  Update callers.
* gfortran.h (gfc_expr_is_variable): New declaration.
(struct gfc_ss_info): Add field needs_temporary.
* trans-array.c (gfc_scalar_elemental_arg_saved_as_argument):
Tighten the condition on aggregate expressions with a check
that the expression is a variable and doesn't need a temporary.
(gfc_conv_resolve_dependency): Add intermediary reference variable.
Set the needs_temporary field.
gcc/testsuite/
* gfortran.dg/elemental_dependency_6.f90: New.

From-SVN: r233188

8 years agore PR rtl-optimization/69691 (wrong code at -O2 on x86_64-linux-gnu)
Jakub Jelinek [Fri, 5 Feb 2016 21:13:43 +0000 (22:13 +0100)]
re PR rtl-optimization/69691 (wrong code at -O2 on x86_64-linux-gnu)

PR rtl-optimization/69691
* lra-eliminations.c (move_plus_up): Don't add the addend twice.

* gcc.c-torture/execute/pr69691.c: New test.

From-SVN: r233187

8 years agore PR c++/69628 (Conditional jump or move depends on uninitialised value(s) in lex_ch...
Jakub Jelinek [Fri, 5 Feb 2016 19:39:48 +0000 (20:39 +0100)]
re PR c++/69628 (Conditional jump or move depends on uninitialised value(s) in lex_charconst(cpp_token const*) (c-lex.c:1252))

PR c++/69628
* charset.c (cpp_interpret_charconst): Clear *PCHARS_SEEN
and *UNSIGNEDP if bailing out early due to errors.

* g++.dg/parse/pr69628.C: New test.

From-SVN: r233186

8 years agoMake issues similar to PR c++/68948 fail loudly.
Jason Merrill [Fri, 5 Feb 2016 17:52:07 +0000 (12:52 -0500)]
Make issues similar to PR c++/68948 fail loudly.

* semantics.c (finish_expr_stmt): If expr is error_mark_node,
make sure we've seen_error().

From-SVN: r233183

8 years agoasan.c: Move to ...
Uros Bizjak [Fri, 5 Feb 2016 16:41:36 +0000 (17:41 +0100)]
asan.c: Move to ...

* gcc.dg/sancov/asan.c: Move to ...
* gcc.dg/asan/sancov-1.c: ... here.

From-SVN: r233182

8 years agopr69577.c: Require avx_runtime target.
Uros Bizjak [Fri, 5 Feb 2016 16:36:38 +0000 (17:36 +0100)]
pr69577.c: Require avx_runtime target.

* gcc.target/i386/pr69577.c: Require avx_runtime target.

From-SVN: r233181

8 years agoAdd a testcase for PR target/69677
H.J. Lu [Fri, 5 Feb 2016 16:24:06 +0000 (16:24 +0000)]
Add a testcase for PR target/69677

PR target/69677
* gcc.target/i386/pr69677.c: New test.

From-SVN: r233180

8 years agocrypto.md (crypto_vpermxor_<mode>): Correct insn type.
Pat Haugen [Fri, 5 Feb 2016 15:25:39 +0000 (15:25 +0000)]
crypto.md (crypto_vpermxor_<mode>): Correct insn type.

* config/rs6000/crypto.md (crypto_vpermxor_<mode>): Correct insn type.
* config/rs6000/rs6000.md (mov<mode>_hardfloat): Likewise.
(*ieee128_mfvsrd_64bit): Likewise.
(*ieee128_mfvsrd_32bit): Likewise.

From-SVN: r233179

8 years agore PR target/69369 (internal compiler error: in remove_unreachable_nodes, at ipa...
Ilya Enkovich [Fri, 5 Feb 2016 14:41:00 +0000 (14:41 +0000)]
re PR target/69369 (internal compiler error: in remove_unreachable_nodes, at ipa.c:457)

gcc/

2016-02-05  Ilya Enkovich  <enkovich.gnu@gmail.com>

PR target/69369
Revert r232560:
2016-01-19  Jan Hubicka  <hubicka@ucw.cz>

* cgraphunit.c (cgraph_node::reset): Clear thunk info and
instrumented_version.

From-SVN: r233177

8 years agoFix PR c++/68948 (wrong code generation due to invalid constructor call)
Patrick Palka [Fri, 5 Feb 2016 14:36:44 +0000 (14:36 +0000)]
Fix PR c++/68948 (wrong code generation due to invalid constructor call)

gcc/cp/ChangeLog:

PR c++/68948
* pt.c (tsubst_baselink): Diagnose an invalid constructor call
if lookup_fnfields returns NULL_TREE and the name being looked
up has the form A::A.

gcc/testsuite/ChangeLog:

PR c++/68948
* g++.dg/template/pr68948.C: New test.

From-SVN: r233176

8 years agoinvoke.texi (Optimize Options): In table of --param options rename second occurrence...
Jeremy Bennett [Fri, 5 Feb 2016 14:27:26 +0000 (14:27 +0000)]
invoke.texi (Optimize Options): In table of --param options rename second occurrence of tracer-min-branch-ratio...

2016-01-05  Jeremy Bennett  <jeremy.bennett@embecosm.com>

        * doc/invoke.texi (Optimize Options): In table of --param options
        rename second occurrence of tracer-min-branch-ratio to
        tracer-min-branch-probability, rename
        tracer-min-branch-ratio-feedback to
        tracer-min-branch-probability-feedback and clarify description,
        rename sched-spec-state-edge-prob-cutoff to
        sched-state-edge-prob-cutoff, rename selsched-max-insns-to-rename
        to selsched-insns-to-rename, rename lto-minpartition to
        lto-min-partition, delete reorder-blocks-duplicate and
        reorder-blocks-duplicate-feedback.

From-SVN: r233175

8 years ago* fi.po: Update.
Joseph Myers [Fri, 5 Feb 2016 13:24:28 +0000 (13:24 +0000)]
* fi.po: Update.

From-SVN: r233174

8 years agoS/390: Remove superfluous loops.
Andreas Krebbel [Fri, 5 Feb 2016 11:38:52 +0000 (11:38 +0000)]
S/390: Remove superfluous loops.

gcc/ChangeLog:

2016-02-05  Andreas Krebbel  <krebbel@linux.vnet.ibm.com>

* config/s390/s390.c (s390_register_info_set_ranges): Remove
superfluous loops.

From-SVN: r233173

8 years agoS/390: Correct documentation
Dominik Vogt [Fri, 5 Feb 2016 10:36:21 +0000 (10:36 +0000)]
S/390: Correct documentation

gcc/ChangeLog

* doc/extend.texi: S/390: Correct some typos.

From-SVN: r233172

8 years agoS/390: Do not require -march=z13 on s390 but only on s390x.
Dominik Vogt [Fri, 5 Feb 2016 10:33:21 +0000 (10:33 +0000)]
S/390: Do not require -march=z13 on s390 but only on s390x.

gcc/testsuite/ChangeLog

* gcc.dg/tree-ssa/ssa-dom-cse-2.c: Remove -march=z13 for s390
(only necessary on s390x).

From-SVN: r233171

8 years agolibstdc++: S/390: Add missing baseline_symbols.txt for s390x/-m31.
Dominik Vogt [Fri, 5 Feb 2016 10:25:08 +0000 (10:25 +0000)]
libstdc++: S/390: Add missing baseline_symbols.txt for s390x/-m31.

The attached patch copies the existing
libstdc++-v3/config/abi/post/s390-linux-gnu/baseline_symbols.txt
to .../s390x-linux-gnu/32/baseline_symbols.txt.  This fixes the
abi test failure on s390x with -m31.

libstdc++-v3/ChangeLog

* config/abi/post/s390x-linux-gnu/32/baseline_symbols.txt (FUNC):
New file.  Copied over from s390-linux-gnu.

From-SVN: r233170

8 years agoS/390: Fix 64 bit sibcall
Andreas Krebbel [Fri, 5 Feb 2016 10:10:35 +0000 (10:10 +0000)]
S/390: Fix 64 bit sibcall

This fixes a problem revealed during the split-stack work:
https://gcc.gnu.org/ml/gcc-patches/2016-02/msg00322.html

gcc/ChangeLog:

2016-02-05  Andreas Krebbel  <krebbel@linux.vnet.ibm.com>

* config/s390/s390.c (s390_emit_call): Add missing 64 bit check.

From-SVN: r233169

8 years agoS/390: Fix r6 vararg handling.
Andreas Krebbel [Fri, 5 Feb 2016 10:08:17 +0000 (10:08 +0000)]
S/390: Fix r6 vararg handling.

This patch fixes a problem introduced with the GPR into FPR slot save
feature for leaf functions.

r6 is argument register as well as call-saved.  Currently we might
decide that it will be a candidate for being saved into an FPR.  If it
turns out later that r6 also needs to be saved due to being required
for vararg we undo the FPR save decision and put it on the stack
again.  Unfortunately the code did not adjust the GPR restore range
accordingly so that the register does not get restored in the load
multiple.

This fixes the following testcases on s390x:

< FAIL: libgomp.c/doacross-1.c execution test
< FAIL: libgomp.c/doacross-2.c execution test
< FAIL: libgomp.c/doacross-3.c execution test
< FAIL: libgomp.c++/doacross-1.C execution test

gcc/ChangeLog:

2016-02-05  Andreas Krebbel  <krebbel@linux.vnet.ibm.com>

    PR target/69625
    * config/s390/s390.c (SAVE_SLOT_NONE, SAVE_SLOT_STACK): New
    defines.
    (s390_register_info_gprtofpr): Use new macros above.
    (s390_register_info_stdarg_fpr): Adjust max_fpr to better match
    its name.
    (s390_register_info_stdarg_gpr): Adjust max_gpr to better match
    its name.  Adjust restore and save gpr ranges.
    (s390_register_info_set_ranges): New function.
    (s390_register_info): Use new macros above.  Call
    s390_register_info_set_ranges.
    (s390_optimize_register_info): Likewise.
    (s390_hard_regno_rename_ok): Use new macros.
    (s390_hard_regno_scratch_ok): Likewise.
    (s390_emit_epilogue): Likewise.
    (s390_can_use_return_insn): Likewise.
    (s390_optimize_prologue): Likewise.
    * config/s390/s390.md (GPR2_REGNUM, GPR6_REGNUM): New constants.

From-SVN: r233168

8 years agore PR bootstrap/69677 (bootstrap failed with --with-arch=corei7 --with-cpu=corei7)
Jakub Jelinek [Fri, 5 Feb 2016 09:23:03 +0000 (10:23 +0100)]
re PR bootstrap/69677 (bootstrap failed with --with-arch=corei7 --with-cpu=corei7)

PR bootstrap/69677
* config/i386/i386.c (convert_scalars_to_vector): Readd stack
alignment fixes.
(ix86_option_override_internal): Disable TARGET_STV even for
-m{incoming,preferred}-stack-boundary=3.

From-SVN: r233167

8 years agoCompile pr69454-2.c with -msse2 -mno-accumulate-outgoing-args
H.J. Lu [Fri, 5 Feb 2016 03:06:40 +0000 (03:06 +0000)]
Compile pr69454-2.c with -msse2 -mno-accumulate-outgoing-args

2016-02-04  H.J. Lu  <hongjiu.lu@intel.com>

* gcc.target/i386/pr69454-2.c: Compile with -msse2
-mno-accumulate-outgoing-args.

From-SVN: r233166

8 years agoobsolete the deprecated rtems targets
Trevor Saunders [Fri, 5 Feb 2016 01:40:39 +0000 (01:40 +0000)]
obsolete the deprecated rtems targets

gcc/ChangeLog:

2016-02-03  Trevor Saunders  <tbsaunde+gcc@tbsaunde.org>

* config.gcc: Mark deprecated rtems targets as obsolete.

From-SVN: r233165

8 years agoDaily bump.
GCC Administrator [Fri, 5 Feb 2016 00:16:12 +0000 (00:16 +0000)]
Daily bump.

From-SVN: r233164

8 years agoTest for C99 stdlib.h functions with -std=c++98
Jonathan Wakely [Thu, 4 Feb 2016 23:47:21 +0000 (23:47 +0000)]
Test for C99 stdlib.h functions with -std=c++98

PR libstdc++/69626
* acinclude.m4 (GLIBCXX_ENABLE_C99): Check C99 stdlib.h functions
with -std=c++98 and define _GLIBCXX98_USE_C99_STDLIB.
* config.h.in: Regenerate.
* configure: Regenerate.
* testsuite/21_strings/c_strings/char/69626.cc: New.

From-SVN: r233161

8 years agocombine: distribute_notes again (PR69567, PR64682)
Segher Boessenkool [Thu, 4 Feb 2016 23:09:51 +0000 (00:09 +0100)]
combine: distribute_notes again (PR69567, PR64682)

As it happens the patch I did over a year ago for PR64682 isn't quite
correct.  This is PR69567.  This fixes it.

PR rtl-optimization/64682
PR rtl-optimization/69567
* combine.c (distribute_notes) <REG_DEAD>: Place the death note
before I2 only if the register is both used and set in I2.

From-SVN: r233159

8 years agoFix constexpr evaluation of comparisons involving pointer-to-members
Patrick Palka [Thu, 4 Feb 2016 22:54:52 +0000 (22:54 +0000)]
Fix constexpr evaluation of comparisons involving pointer-to-members

gcc/cp/ChangeLog:

* constexpr.c (cxx_eval_binary_expression): Fold equality
comparisons involving PTRMEM_CSTs.

gcc/testsuite/ChangeLog:

* g++.dg/cpp0x/constexpr-ptrmem5.C: New test.

From-SVN: r233158

8 years ago* config/msp430/msp430.c (msp430_start_function): Add function type.
DJ Delorie [Thu, 4 Feb 2016 22:54:30 +0000 (17:54 -0500)]
* config/msp430/msp430.c (msp430_start_function): Add function type.

From-SVN: r233157

8 years agolibgo: bump version number for upgrade to 1.6rc1
Ian Lance Taylor [Thu, 4 Feb 2016 22:54:20 +0000 (22:54 +0000)]
libgo: bump version number for upgrade to 1.6rc1

    Reviewed-on: https://go-review.googlesource.com/19233

From-SVN: r233156

8 years agore PR c/69669 (ICE with enum __attribute__((mode(QI))))
Jakub Jelinek [Thu, 4 Feb 2016 22:17:05 +0000 (23:17 +0100)]
re PR c/69669 (ICE with enum __attribute__((mode(QI))))

PR c/69669
* c-decl.c (finish_enum): When honoring mode attribute,
make sure to use proper TYPE_MIN_VALUE and TYPE_MAX_VALUE.

* c-c++-common/pr69669.c: New test.

From-SVN: r233154

8 years agore PR fortran/69368 (spec2006 test case 416.gamess fails with the g++ 6.0 compiler...
Jakub Jelinek [Thu, 4 Feb 2016 22:15:33 +0000 (23:15 +0100)]
re PR fortran/69368 (spec2006 test case 416.gamess fails with the g++ 6.0 compiler starting with r232508)

PR fortran/69368
* tree-dfa.c (get_ref_base_and_extent): Remove unreachable code.

From-SVN: r233153

8 years agore PR target/69577 (wrong code with -fno-forward-propagate -mavx and 128bit arithmeti...
Uros Bizjak [Thu, 4 Feb 2016 22:10:56 +0000 (23:10 +0100)]
re PR target/69577 (wrong code with -fno-forward-propagate -mavx and 128bit arithmetics since r215450)

PR rtl-optimization/69577
Revert:
2015-10-29  Richard Henderson  <rth@redhat.com>

PR target/68124
PR rtl-opt/67609
* config/i386/i386.c (ix86_cannot_change_mode_class): Tighten
sse check to the exact conditions of PR 67609.

From-SVN: r233152

8 years agoRegenerate front page of libstdc++ HTML docs
Jonathan Wakely [Thu, 4 Feb 2016 21:46:19 +0000 (21:46 +0000)]
Regenerate front page of libstdc++ HTML docs

* doc/html/index.html: Regenerate.

From-SVN: r233151

8 years agoUpdate copyright years in libstdc++ manual and add link
Jonathan Wakely [Thu, 4 Feb 2016 21:43:40 +0000 (21:43 +0000)]
Update copyright years in libstdc++ manual and add link

* doc/xml/manual/containers.xml: Add cross-reference to Dual ABI.
* doc/xml/manual/spine.xml: Update copyright years and author blurb.
* doc/html/*: Regenerate.

From-SVN: r233150

8 years agore PR target/69667 (ppc64le -mlra: ICE: Max. number of generated reload insns per...
Michael Meissner [Thu, 4 Feb 2016 21:05:14 +0000 (21:05 +0000)]
re PR target/69667 (ppc64le -mlra: ICE: Max. number of generated reload insns per insn is achieved (90))

[gcc]
2016-02-04  Michael Meissner  <meissner@linux.vnet.ibm.com>

PR target/69667
* config/rs6000/rs6000.md (mov<mode>_64bit_dm): Use 'd' constraint
instead of 'ws', and 'wh' instead of 'wm' since TFmode/IFmode are
not allowed into the traditional Altivec registers.
(movtd_64bit_nodm): Likewise.
(mov<mode>_32bit, FMOVE128_FPR iterator): Likewise.

[gcc/testsuite]
2016-02-04  Michael Meissner  <meissner@linux.vnet.ibm.com>

PR target/69667
* g++.dg/pr69667.C: New file.

From-SVN: r233147

8 years agoFix jit crash on aarch64
David Malcolm [Thu, 4 Feb 2016 19:30:28 +0000 (19:30 +0000)]
Fix jit crash on aarch64

gcc/ChangeLog:
* config/aarch64/cortex-a57-fma-steering.c
(aarch64_register_fma_steering): Remove "static" from arguments
to register_pass.

From-SVN: r233146

8 years agoThis patch fixes an exponential issue in ccmp.c.
Wilco Dijkstra [Thu, 4 Feb 2016 18:23:35 +0000 (18:23 +0000)]
This patch fixes an exponential issue in ccmp.c.

This patch fixes an exponential issue in ccmp.c.  When deciding which ccmp
expansion to use, the tree nodes gs0 and gs1 are fully expanded twice.  If
they contain more CCMP opportunities, their subtrees are also expanded twice.
When the trees are complex the expansion takes exponential time and memory.
As a workaround in GCC6 compute the cost of the first expansion early, and
only try the alternative expansion if the cost is low enough.  This rarely
affects real code, eg. SPECINT2006 has identical codesize.

2016-02-04  Wilco Dijkstra  <wdijkstr@arm.com>

    gcc/
PR target/69619
* ccmp.c (expand_ccmp_expr_1): Avoid evaluating gs0/gs1
twice when complex.

    gcc/testsuite/
PR target/69619
* gcc.dg/pr69619.c: Add new test.

From-SVN: r233145

8 years agogcc: invoke: delete -mno-fma4 docs
Mike Frysinger [Thu, 4 Feb 2016 17:32:11 +0000 (17:32 +0000)]
gcc: invoke: delete -mno-fma4 docs

We don't document the -mno-xxx variants for other flags here, and the
paragraph here specifically says "Each has a corresponding -mno- option
to disable use of these instructions".  Drop the -mno-fma4 line.

From-SVN: r233144

8 years agoPR 69577: Invalid RA of destination subregs
Richard Sandiford [Thu, 4 Feb 2016 15:01:15 +0000 (15:01 +0000)]
PR 69577: Invalid RA of destination subregs

In PR 69577 we have:

      A: (set (reg:V2TI X) ...)
      B: (set (subreg:TI (reg:V2TI X) 0) ...)

X gets allocated to an AVX register, as usual for V2TI.  The problem is
that the movti for B doesn't then preserve the other half of X, even
though the subreg semantics are supposed to guarantee that.

If instead the same value had been set by:

      A': (set (subreg:TI (reg:V2TI X) 16) ...)
      B: (set (subreg:TI (reg:V2TI X) 0) ...)

the subreg in A' would have prevented the use of AVX registers for X,
since you can't directly access the high part.

IMO these are really the same thing.  An alternative way to view it
is that the original sequence is equivalent to:

      A: (set (reg:V2TI X) ...)
      B1: (set (subreg:TI (reg:V2TI X) 0) ...)
      B2: (set (subreg:TI (reg:V2TI X) 16) (subreg:TI (reg:V2TI X) 16))

in which B2 is a no-op and therefore implicit.  The handling ought
to be the same regardless of whether there is an rtl insn that
explicitly assigns to (subreg:TI (reg:V2TI X) 16).

This patch implements that idea.  Hopefully the comments explain
what's going on.

Tested on x86_64-linux-gnu, aarch64-linux-gnu and arm-linux-gnueabihf.

gcc/
PR rtl-optimization/69577
* reginfo.c (record_subregs_of_mode): Add a partial_def parameter.
(find_subregs_of_mode): Update accordingly.  Iterate over partial
definitions.

gcc/testsuite/
PR rtl-optimization/69577
* gcc.target/i386/pr69577.c: New test.

From-SVN: r233143

8 years ago[ARM] Remove neon_reinterpret, use casts
Alan Lawrence [Thu, 4 Feb 2016 14:49:02 +0000 (14:49 +0000)]
[ARM] Remove neon_reinterpret, use casts

* config/arm/arm-protos.h (neon_reinterpret): Remove.
* config/arm/arm.c (neon_reinterpret): Remove.
* config/arm/arm_neon_builtins.def (vreinterpretv8qi, vreinterpretv4hi,
vreinterpretv2si, vreinterpretv2sf, vreinterpretdi, vreinterpretv16qi,
vreinterpretv8hi, vreinterpretv4si, vreinterpretv4sf, vreinterpretv2di,
vreinterpretti): Remove.
* config/arm/neon.md (neon_vreinterpretv8qi<mode>,
neon_vreinterpretv4hi<mode>, neon_vreinterpretv2si<mode>,
neon_vreinterpretv2sf<mode>, neon_vreinterpretdi<mode>,
neon_vreinterpretti<mode>, neon_vreinterpretv16qi<mode>,
neon_vreinterpretv8hi<mode>, neon_vreinterpretv4si<mode>,
neon_vreinterpretv4sf<mode>, neon_vreinterpretv2di<mode>): Remove.
* config/arm/arm_neon.h (vreinterpret_p8_p16, vreinterpret_p8_f32,
vreinterpret_p8_p64, vreinterpret_p8_s64, vreinterpret_p8_u64,
vreinterpret_p8_s8, vreinterpret_p8_s16, vreinterpret_p8_s32,
vreinterpret_p8_u8, vreinterpret_p8_u16, vreinterpret_p8_u32,
vreinterpret_p16_p8, vreinterpret_p16_f32, vreinterpret_p16_p64,
vreinterpret_p16_s64, vreinterpret_p16_u64, vreinterpret_p16_s8,
vreinterpret_p16_s16, vreinterpret_p16_s32, vreinterpret_p16_u8,
vreinterpret_p16_u16, vreinterpret_p16_u32, vreinterpret_f32_p8,
vreinterpret_f32_p16, vreinterpret_f32_p64, vreinterpret_f32_s64,
vreinterpret_f32_u64, vreinterpret_f32_s8, vreinterpret_f32_s16,
vreinterpret_f32_s32, vreinterpret_f32_u8, vreinterpret_f32_u16,
vreinterpret_f32_u32, vreinterpret_p64_p8, vreinterpret_p64_p16,
vreinterpret_p64_f32, vreinterpret_p64_s64, vreinterpret_p64_u64,
vreinterpret_p64_s8, vreinterpret_p64_s16, vreinterpret_p64_s32,
vreinterpret_p64_u8, vreinterpret_p64_u16, vreinterpret_p64_u32,
vreinterpret_s64_p8, vreinterpret_s64_p16, vreinterpret_s64_f32,
vreinterpret_s64_p64, vreinterpret_s64_u64, vreinterpret_s64_s8,
vreinterpret_s64_s16, vreinterpret_s64_s32, vreinterpret_s64_u8,
vreinterpret_s64_u16, vreinterpret_s64_u32, vreinterpret_u64_p8,
vreinterpret_u64_p16, vreinterpret_u64_f32, vreinterpret_u64_p64,
vreinterpret_u64_s64, vreinterpret_u64_s8, vreinterpret_u64_s16,
vreinterpret_u64_s32, vreinterpret_u64_u8, vreinterpret_u64_u16,
vreinterpret_u64_u32, vreinterpret_s8_p8, vreinterpret_s8_p16,
vreinterpret_s8_f32, vreinterpret_s8_p64, vreinterpret_s8_s64,
vreinterpret_s8_u64, vreinterpret_s8_s16, vreinterpret_s8_s32,
vreinterpret_s8_u8, vreinterpret_s8_u16, vreinterpret_s8_u32,
vreinterpret_s16_p8, vreinterpret_s16_p16, vreinterpret_s16_f32,
vreinterpret_s16_p64, vreinterpret_s16_s64, vreinterpret_s16_u64,
vreinterpret_s16_s8, vreinterpret_s16_s32, vreinterpret_s16_u8,
vreinterpret_s16_u16, vreinterpret_s16_u32, vreinterpret_s32_p8,
vreinterpret_s32_p16, vreinterpret_s32_f32, vreinterpret_s32_p64,
vreinterpret_s32_s64, vreinterpret_s32_u64, vreinterpret_s32_s8,
vreinterpret_s32_s16, vreinterpret_s32_u8, vreinterpret_s32_u16,
vreinterpret_s32_u32, vreinterpret_u8_p8, vreinterpret_u8_p16,
vreinterpret_u8_f32, vreinterpret_u8_p64, vreinterpret_u8_s64,
vreinterpret_u8_u64, vreinterpret_u8_s8, vreinterpret_u8_s16,
vreinterpret_u8_s32, vreinterpret_u8_u16, vreinterpret_u8_u32,
vreinterpret_u16_p8, vreinterpret_u16_p16, vreinterpret_u16_f32,
vreinterpret_u16_p64, vreinterpret_u16_s64, vreinterpret_u16_u64,
vreinterpret_u16_s8, vreinterpret_u16_s16, vreinterpret_u16_s32,
vreinterpret_u16_u8, vreinterpret_u16_u32, vreinterpret_u32_p8,
vreinterpret_u32_p16, vreinterpret_u32_f32, vreinterpret_u32_p64,
vreinterpret_u32_s64, vreinterpret_u32_u64, vreinterpret_u32_s8,
vreinterpret_u32_s16, vreinterpret_u32_s32, vreinterpret_u32_u8,
vreinterpret_u32_u16, vreinterpretq_p8_p16, vreinterpretq_p8_f32,
vreinterpretq_p8_p64, vreinterpretq_p8_p128, vreinterpretq_p8_s64,
vreinterpretq_p8_u64, vreinterpretq_p8_s8, vreinterpretq_p8_s16,
vreinterpretq_p8_s32, vreinterpretq_p8_u8, vreinterpretq_p8_u16,
vreinterpretq_p8_u32, vreinterpretq_p16_p8, vreinterpretq_p16_f32,
vreinterpretq_p16_p64, vreinterpretq_p16_p128, vreinterpretq_p16_s64,
vreinterpretq_p16_u64, vreinterpretq_p16_s8, vreinterpretq_p16_s16,
vreinterpretq_p16_s32, vreinterpretq_p16_u8, vreinterpretq_p16_u16,
vreinterpretq_p16_u32, vreinterpretq_f32_p8, vreinterpretq_f32_p16,
vreinterpretq_f32_p64, vreinterpretq_f32_p128, vreinterpretq_f32_s64,
vreinterpretq_f32_u64, vreinterpretq_f32_s8, vreinterpretq_f32_s16,
vreinterpretq_f32_s32, vreinterpretq_f32_u8, vreinterpretq_f32_u16,
vreinterpretq_f32_u32, vreinterpretq_p64_p8, vreinterpretq_p64_p16,
vreinterpretq_p64_f32, vreinterpretq_p64_p128, vreinterpretq_p64_s64,
vreinterpretq_p64_u64, vreinterpretq_p64_s8, vreinterpretq_p64_s16,
vreinterpretq_p64_s32, vreinterpretq_p64_u8, vreinterpretq_p64_u16,
vreinterpretq_p64_u32, vreinterpretq_p128_p8, vreinterpretq_p128_p16,
vreinterpretq_p128_f32, vreinterpretq_p128_p64, vreinterpretq_p128_s64,
vreinterpretq_p128_u64, vreinterpretq_p128_s8, vreinterpretq_p128_s16,
vreinterpretq_p128_s32, vreinterpretq_p128_u8, vreinterpretq_p128_u16,
vreinterpretq_p128_u32, vreinterpretq_s64_p8, vreinterpretq_s64_p16,
vreinterpretq_s64_f32, vreinterpretq_s64_p64, vreinterpretq_s64_p128,
vreinterpretq_s64_u64, vreinterpretq_s64_s8, vreinterpretq_s64_s16,
vreinterpretq_s64_s32, vreinterpretq_s64_u8, vreinterpretq_s64_u16,
vreinterpretq_s64_u32, vreinterpretq_u64_p8, vreinterpretq_u64_p16,
vreinterpretq_u64_f32, vreinterpretq_u64_p64, vreinterpretq_u64_p128,
vreinterpretq_u64_s64, vreinterpretq_u64_s8, vreinterpretq_u64_s16,
vreinterpretq_u64_s32, vreinterpretq_u64_u8, vreinterpretq_u64_u16,
vreinterpretq_u64_u32, vreinterpretq_s8_p8, vreinterpretq_s8_p16,
vreinterpretq_s8_f32, vreinterpretq_s8_p64, vreinterpretq_s8_p128,
vreinterpretq_s8_s64, vreinterpretq_s8_u64, vreinterpretq_s8_s16,
vreinterpretq_s8_s32, vreinterpretq_s8_u8, vreinterpretq_s8_u16,
vreinterpretq_s8_u32, vreinterpretq_s16_p8, vreinterpretq_s16_p16,
vreinterpretq_s16_f32, vreinterpretq_s16_p64, vreinterpretq_s16_p128,
vreinterpretq_s16_s64, vreinterpretq_s16_u64, vreinterpretq_s16_s8,
vreinterpretq_s16_s32, vreinterpretq_s16_u8, vreinterpretq_s16_u16,
vreinterpretq_s16_u32, vreinterpretq_s32_p8, vreinterpretq_s32_p16,
vreinterpretq_s32_f16, vreinterpretq_s32_f32, vreinterpretq_s32_p64,
vreinterpretq_s32_p128, vreinterpretq_s32_s64, vreinterpretq_s32_u64,
vreinterpretq_s32_s8, vreinterpretq_s32_s16, vreinterpretq_s32_u8,
vreinterpretq_s32_u16, vreinterpretq_s32_u32, vreinterpretq_u8_p8,
vreinterpretq_u8_p16, vreinterpretq_u8_f32, vreinterpretq_u8_p64,
vreinterpretq_u8_p128, vreinterpretq_u8_s64, vreinterpretq_u8_u64,
vreinterpretq_u8_s8, vreinterpretq_u8_s16, vreinterpretq_u8_s32,
vreinterpretq_u8_u16, vreinterpretq_u8_u32, vreinterpretq_u16_p8,
vreinterpretq_u16_p16, vreinterpretq_u16_f32, vreinterpretq_u16_p64,
vreinterpretq_u16_p128, vreinterpretq_u16_s64, vreinterpretq_u16_u64,
vreinterpretq_u16_s8, vreinterpretq_u16_s16, vreinterpretq_u16_s32,
vreinterpretq_u16_u8, vreinterpretq_u16_u32, vreinterpretq_u32_p8,
vreinterpretq_u32_p16, vreinterpretq_u32_f32, vreinterpretq_u32_p64,
vreinterpretq_u32_p128, vreinterpretq_u32_s64, vreinterpretq_u32_u64,
vreinterpretq_u32_s8, vreinterpretq_u32_s16, vreinterpretq_u32_s32,
vreinterpretq_u32_u8, vreinterpretq_u32_u16): Rewrite using casts.

From-SVN: r233142

8 years agoUpdate gcc .po files.
Joseph Myers [Thu, 4 Feb 2016 14:17:21 +0000 (14:17 +0000)]
Update gcc .po files.

* be.po, da.po, de.po, el.po, es.po, fi.po, fr.po, hr.po, id.po,
ja.po, nl.po, ru.po, sr.po, sv.po, tr.po, uk.po, vi.po, zh_CN.po,
zh_TW.po: Update.

From-SVN: r233141

8 years agoUpdate cpplib .po files.
Joseph Myers [Thu, 4 Feb 2016 14:15:08 +0000 (14:15 +0000)]
Update cpplib .po files.

* be.po, ca.po, da.po, de.po, el.po, eo.po, es.po, fi.po, fr.po,
id.po, ja.po, nl.po, pr_BR.po, ru.po, sr.po, sv.po, tr.po, uk.po,
vi.po, zh_CN.po, zh_TW.po: Update.

From-SVN: r233140

8 years agore PR sanitizer/69276 (Address sanitizer does not handle heap overflow)
Martin Liska [Thu, 4 Feb 2016 11:50:40 +0000 (12:50 +0100)]
re PR sanitizer/69276 (Address sanitizer does not handle heap overflow)

Fix PR sanitizer/69276

* g++.dg/asan/pr69276.C: New test.
PR sanitizer/PR69276
* asan.c (has_stmt_been_instrumented_p): Instrument gimple calls
that are gimple_store_p.
(maybe_instrument_call): Likewise.

From-SVN: r233137

8 years agoaarch64.c (aarch64_legitimize_address): Force register scaling out of memory referenc...
Bin Cheng [Thu, 4 Feb 2016 11:05:46 +0000 (11:05 +0000)]
aarch64.c (aarch64_legitimize_address): Force register scaling out of memory reference and comment why.

* config/aarch64/aarch64.c (aarch64_legitimize_address): Force
register scaling out of memory reference and comment why.

From-SVN: r233136

8 years agoclass.c (find_flexarrays): Don't declare dom variable.
Jakub Jelinek [Thu, 4 Feb 2016 11:04:07 +0000 (12:04 +0100)]
class.c (find_flexarrays): Don't declare dom variable.

* class.c (find_flexarrays): Don't declare dom variable.
(diagnose_flexarray): Likewise.

From-SVN: r233135

8 years ago[ARM][4/4] Adjust gcc.target/arm/wmul-[123].c tests
Kyrylo Tkachov [Thu, 4 Feb 2016 09:57:36 +0000 (09:57 +0000)]
[ARM][4/4] Adjust gcc.target/arm/wmul-[123].c tests

PR target/65932
PR target/67714
* gcc.target/arm/wmul-3.c: Simplify test to generate just
a single smulbb instruction.
* gcc.target/amr/wmul-1.c: Add -mtune=cortex-a9 to dg-options.
* gcc.target/amr/wmul-2.c: Likewise.

From-SVN: r233134

8 years ago[cse][3/4] Don't overwrite original rtx when folding source of set
Kyrylo Tkachov [Thu, 4 Feb 2016 09:56:13 +0000 (09:56 +0000)]
[cse][3/4] Don't overwrite original rtx when folding source of set

PR target/65932
PR target/67714
* cse.c (cse_insn): Pass NULL to fold_rtx when initially
folding the source of a SET.

From-SVN: r233133

8 years ago[ARM][2/4] Fix operand costing logic for SMUL[TB][TB]
Kyrylo Tkachov [Thu, 4 Feb 2016 09:54:37 +0000 (09:54 +0000)]
[ARM][2/4] Fix operand costing logic for SMUL[TB][TB]

PR target/65932
PR target/67714
* config/arm/arm.c (arm_new_rtx_costs, MULT case): Properly extract
the operands of the SIGN_EXTENDs from a SMUL[TB][TB] rtx.

From-SVN: r233132

8 years ago[ARM][1/4] PR target/65932: Add testcase
Kyrylo Tkachov [Thu, 4 Feb 2016 09:51:35 +0000 (09:51 +0000)]
[ARM][1/4] PR target/65932: Add testcase

PR target/65932
PR target/67714
* gcc.c-torture/execute/pr67714.c: New test.

From-SVN: r233131

8 years ago[ARM] PR target/65932: stop changing signedness in PROMOTE_MODE
Jim Wilson [Thu, 4 Feb 2016 09:50:12 +0000 (09:50 +0000)]
[ARM] PR target/65932: stop changing signedness in PROMOTE_MODE

2016-02-04  Jim Wilson  <jim.wilson@linaro.org>

PR target/65932
PR target/67714
* config/arm/arm.h (PROMOTE_MODE): Don't set UNSIGNEDP for QImode and
HImode.

From-SVN: r233130

8 years agoarm-c.c (arm_reset_previous_fndecl): Style fix and typo.
Christian Bruel [Thu, 4 Feb 2016 09:06:32 +0000 (10:06 +0100)]
arm-c.c (arm_reset_previous_fndecl): Style fix and typo.

2016-02-04  Christian Bruel  <christian.bruel@st.com>

* config/arm/arm-c.c (arm_reset_previous_fndecl): Style fix and typo.
* config/arm/arm.c (arm_set_current_function): Likewise.

From-SVN: r233129

8 years agore PR target/69454 (ix86_expand_prologue internal compiler error: Segmentation fault)
Jakub Jelinek [Thu, 4 Feb 2016 09:02:01 +0000 (10:02 +0100)]
re PR target/69454 (ix86_expand_prologue internal compiler error: Segmentation fault)

PR target/69454
* config/i386/i386.c (convert_scalars_to_vector): Remove
stack alignment fixes.
(ix86_option_override_internal): Disable TARGET_STV if stack
might not be aligned enough.
(ix86_minimum_alignment): Assert that TARGET_STV is false.

* gcc.target/i386/pr69454-1.c: New test.
* gcc.target/i386/pr69454-2.c: New test.

From-SVN: r233128

8 years agoDisable auto prefetcher for -march=znver1.
Victoria Stepanyan [Thu, 4 Feb 2016 07:52:08 +0000 (07:52 +0000)]
Disable auto prefetcher for -march=znver1.

2016-02-04  Victoria Stepanyan  <victoria.stepanyan@amd.com>

        * gcc/config/i386/x86-tune.def: Disable default prefetching
        for -march=znver1.

From-SVN: r233127

8 years agoPR c++/69251 - [6 Regression] ICE in unify_array_domain on a flexible array
Martin Sebor [Thu, 4 Feb 2016 04:50:42 +0000 (21:50 -0700)]
PR c++/69251 - [6 Regression] ICE in unify_array_domain on a flexible array

PR c++/69251 - [6 Regression] ICE in unify_array_domain on a flexible array
               member
PR c++/69253 - [6 Regression] ICE in cxx_incomplete_type_diagnostic initializing
               a flexible array member with empty string
PR c++/69290 - [6 Regression] ICE on invalid initialization of a flexible array
               member
PR c++/69277 - [6 Regression] ICE mangling a flexible array member
PR c++/69349 - template substitution error for flexible array members

gcc/testsuite/ChangeLog:
2016-02-03  Martin Sebor  <msebor@redhat.com>

PR c++/69251
PR c++/69253
PR c++/69290
PR c++/69277
PR c++/69349
* g++.dg/ext/flexarray-mangle-2.C: New test.
* g++.dg/ext/flexarray-mangle.C: New test.
* g++.dg/ext/flexarray-subst.C: New test.
* g++.dg/ext/flexary11.C: New test.
* g++.dg/ext/flexary12.C: New test.
* g++.dg/ext/flexary13.C: New test.
* g++.dg/ext/flexary14.C: New test.
* g++.dg/other/dump-ada-spec-2.C: Adjust.

gcc/cp/ChangeLog:
2016-02-03  Martain Sebor  <msebor@redhat.com>

PR c++/69251
PR c++/69253
PR c++/69290
PR c++/69277
PR c++/69349
* class.c (walk_subobject_offsets): Avoid testing the upper bound
of a flexible array member for equality to null.
(find_flexarrays): Remove spurious whitespace introduced in r231665.
(diagnose_flexarrays): Avoid checking the upper bound of arrays.
(check_flexarrays): Same.
* decl.c (compute_array_index_type): Avoid special case for flexible
array members.
(grokdeclarator): Avoid calling compute_array_index_type for flexible
array members.
* error.c (dump_type_suffix): Revert changes introduced in r231665
and rendered unnecessary by the changes above.
* pt.c (tsubst):  Same.
* tree.c (build_ctor_subob_ref): Handle flexible array members.
* typeck2.c (digest_init_r): Revert changes introduced in r231665.
(process_init_constructor_array): Same.
(process_init_constructor_record): Same.

From-SVN: r233126

8 years agoDefine check_union_passing6 only for CHECK_FLOAT128
H.J. Lu [Thu, 4 Feb 2016 01:46:17 +0000 (01:46 +0000)]
Define check_union_passing6 only for CHECK_FLOAT128

* gcc.target/i386/iamcu/test_passing_unions.c (check_union_passing6):
Define only if CHECK_FLOAT128 is defined.
(main): Properly initialize u5.

From-SVN: r233124

8 years agore PR target/69461 (ICE in lra_set_insn_recog_data, at lra.c:964)
Michael Meissner [Thu, 4 Feb 2016 00:39:34 +0000 (00:39 +0000)]
re PR target/69461 (ICE in lra_set_insn_recog_data, at lra.c:964)

2016-02-03  Michael Meissner  <meissner@linux.vnet.ibm.com>
    Vladimir Makarov  <vmakarov@redhat.com>

PR target/69461
* config/rs6000/rs6000.c (rs6000_legitimate_address_p): Fix thinko
in validating fused toc addresses.

Co-Authored-By: Vladimir Makarov <vmakarov@redhat.com>
From-SVN: r233120

8 years agoDaily bump.
GCC Administrator [Thu, 4 Feb 2016 00:16:12 +0000 (00:16 +0000)]
Daily bump.

From-SVN: r233119

8 years agore PR c/69627 (Conditional jump or move depends on uninitialised value(s) in (anonymo...
Jakub Jelinek [Wed, 3 Feb 2016 22:40:22 +0000 (23:40 +0100)]
re PR c/69627 (Conditional jump or move depends on uninitialised value(s) in (anonymous namespace)::layout::get_state_at_point)

PR c/69627
* diagnostic-show-locus.c (layout::get_state_at_point): Don't read
range->m_caret fields if range->m_show_caret_p is false.

* gcc.dg/pr69627.c: New test.

From-SVN: r233114