platform/upstream/gcc.git
10 years agore PR c++/60277 (Bogus "inline function virtual ..." used but never defined)
Jason Merrill [Fri, 21 Feb 2014 14:56:46 +0000 (09:56 -0500)]
re PR c++/60277 (Bogus "inline function virtual ..." used but never defined)

PR c++/60277
* call.c (build_array_conv): Don't crash on VLA.

From-SVN: r208001

10 years agore PR c++/60248 ([C++11] ICE specializing variadic template)
Jason Merrill [Fri, 21 Feb 2014 14:56:38 +0000 (09:56 -0500)]
re PR c++/60248 ([C++11] ICE specializing variadic template)

PR c++/60248
* mangle.c (mangle_decl): Don't make an alias for a TYPE_DECL.

From-SVN: r208000

10 years agore PR c++/60252 ([c++11] ICE with invalid variable-length array in lambda parameter)
Jason Merrill [Fri, 21 Feb 2014 14:56:31 +0000 (09:56 -0500)]
re PR c++/60252 ([c++11] ICE with invalid variable-length array in lambda parameter)

PR c++/60252
* lambda.c (maybe_resolve_dummy): Don't try to capture this
in declaration context.

From-SVN: r207999

10 years agoDR 1591 PR c++/60051
Jason Merrill [Fri, 21 Feb 2014 14:56:20 +0000 (09:56 -0500)]
DR 1591 PR c++/60051

DR 1591
PR c++/60051
* pt.c (unify): Only unify if deducible.  Handle 0-length list.

From-SVN: r207998

10 years agoRename testsuite directory shared_mutex to shared_timed_mutex for consistency.
Ed Smith-Rowland [Fri, 21 Feb 2014 14:40:26 +0000 (14:40 +0000)]
Rename testsuite directory shared_mutex to shared_timed_mutex for consistency.

2014-02-21  Ed Smith-Rowland  <3dw4rd@verizon.net>

Rename testsuite directory shared_mutex to shared_timed_mutex
for consistency.
* testsuite/30_threads/shared_mutex: Moved to...
* testsuite/30_threads/shared_timed_mutex: ...here

From-SVN: r207997

10 years agore PR c++/60250 ([c++1y] ICE using lambda for array size)
Jason Merrill [Fri, 21 Feb 2014 14:01:29 +0000 (09:01 -0500)]
re PR c++/60250 ([c++1y] ICE using lambda for array size)

PR c++/60250
* parser.c (cp_parser_direct_declarator): Don't wrap a
type-dependent expression in a NOP_EXPR.

From-SVN: r207996

10 years agore PR c++/60251 ([c++11] ICE capturing variable-length array)
Jason Merrill [Fri, 21 Feb 2014 14:01:20 +0000 (09:01 -0500)]
re PR c++/60251 ([c++11] ICE capturing variable-length array)

PR c++/60251
* lambda.c (is_normal_capture_proxy): Handle VLA capture.

From-SVN: r207995

10 years agore PR c++/60167 (Bogus error: conflicting declaration)
Jason Merrill [Fri, 21 Feb 2014 14:00:44 +0000 (09:00 -0500)]
re PR c++/60167 (Bogus error: conflicting declaration)

PR c++/60167
PR c++/60222
PR c++/58606
* parser.c (cp_parser_template_argument): Restore dereference.
* pt.c (template_parm_to_arg): Dereference non-pack expansions too.
(process_partial_specialization): Handle deref.
(unify): Likewise.

From-SVN: r207994

10 years agoinvoke.texi (mvirt, mno-virt): Document.
Catherine Moore [Fri, 21 Feb 2014 13:30:47 +0000 (08:30 -0500)]
invoke.texi (mvirt, mno-virt): Document.

2014-02-21  Catherine Moore  <clm@codesourcery.com>

* doc/invoke.texi (mvirt, mno-virt): Document.
* config/mips/mips.opt (mvirt): New option.
* config/mips/mips.h (ASM_SPEC): Pass mvirt to the assembler.

From-SVN: r207993

10 years agore PR tree-optimization/60276 (-O3 autovectorizer breaks on a particular loop)
Richard Biener [Fri, 21 Feb 2014 13:18:54 +0000 (13:18 +0000)]
re PR tree-optimization/60276 (-O3 autovectorizer breaks on a particular loop)

2014-02-21  Richard Biener  <rguenther@suse.de>

PR tree-optimization/60276
* tree-vectorizer.h (struct _stmt_vec_info): Add min_neg_dist field.
(STMT_VINFO_MIN_NEG_DIST): New macro.
* tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Record
STMT_VINFO_MIN_NEG_DIST.
* tree-vect-stmts.c (vectorizable_load): Verify if assumptions
made for negative dependence distances still hold.

* gcc.dg/vect/pr60276.c: New testcase.

From-SVN: r207992

10 years agore PR middle-end/60291 (slow compile times for any mode (-O0/-O1/-O2) on large .c...
Richard Biener [Fri, 21 Feb 2014 13:14:23 +0000 (13:14 +0000)]
re PR middle-end/60291 (slow compile times for any mode (-O0/-O1/-O2) on large .c source file (30MBs))

2014-02-21  Richard Biener  <rguenther@suse.de>

PR middle-end/60291
* tree-ssa-live.c (mark_all_vars_used_1): Do not walk
DECL_INITIAL for globals not in the current function context.

From-SVN: r207991

10 years agoAdded forgotten ChangeLog entry.
Rainer Orth [Fri, 21 Feb 2014 11:55:14 +0000 (11:55 +0000)]
Added forgotten ChangeLog entry.

From-SVN: r207989

10 years agore PR tree-optimization/56490 (-Wall triggering infinite loop)
Jakub Jelinek [Fri, 21 Feb 2014 09:53:56 +0000 (10:53 +0100)]
re PR tree-optimization/56490 (-Wall triggering infinite loop)

PR tree-optimization/56490
* params.def (PARAM_UNINIT_CONTROL_DEP_ATTEMPTS): New param.
* tree-ssa-uninit.c: Include params.h.
(compute_control_dep_chain): Add num_calls argument, return false
if it exceed PARAM_UNINIT_CONTROL_DEP_ATTEMPTS param, pass
num_calls to recursive call.
(find_predicates): Change dep_chain into normal array,
cur_chain into auto_vec<edge, MAX_CHAIN_LEN + 1>, add num_calls
variable and adjust compute_control_dep_chain caller.
(find_def_preds): Likewise.

From-SVN: r207988

10 years agoCorrect TDF_RAW pretty-printing of GIMPLE_OMP_FOR's GF_OMP_FOR_KIND_CILKSIMD.
Thomas Schwinge [Fri, 21 Feb 2014 09:18:15 +0000 (10:18 +0100)]
Correct TDF_RAW pretty-printing of GIMPLE_OMP_FOR's GF_OMP_FOR_KIND_CILKSIMD.

gcc/
* gimple-pretty-print.c (dump_gimple_omp_for) [flags & TDF_RAW]
<case GF_OMP_FOR_KIND_CILKSIMD>: Add missing break statement.

From-SVN: r207987

10 years agore PR fortran/60234 ([OOP] ICE in generate_finalization_wrapper at fortran/class...
Janus Weil [Fri, 21 Feb 2014 09:06:57 +0000 (10:06 +0100)]
re PR fortran/60234 ([OOP] ICE in generate_finalization_wrapper at fortran/class.c:1883)

2014-02-21  Janus Weil  <janus@gcc.gnu.org>

PR fortran/60234
* gfortran.h (gfc_build_class_symbol): Removed argument.
* class.c (gfc_add_component_ref): Fix up missing vtype if necessary.
(gfc_build_class_symbol): Remove argument 'delayed_vtab'. vtab is always
delayed now, except for unlimited polymorphics.
(comp_is_finalizable): Procedure pointer components are not finalizable.
* decl. (build_sym, build_struct, attr_decl1): Removed argument of
'gfc_build_class_symbol'.
* match.c (copy_ts_from_selector_to_associate, select_type_set_tmp):
Ditto.
* symbol.c (gfc_set_default_type): Ditto.

2014-02-21  Janus Weil  <janus@gcc.gnu.org>

PR fortran/60234
* gfortran.dg/finalize_23.f90: New.

From-SVN: r207986

10 years agostormy16.md (pushdqi1): Add mode to post_inc.
Nick Clifton [Fri, 21 Feb 2014 08:11:10 +0000 (08:11 +0000)]
stormy16.md (pushdqi1): Add mode to post_inc.

* config/stormy16/stormy16.md (pushdqi1): Add mode to post_inc.
(pushhi1): Likewise.
(popqi1): Add mode to pre_dec.
(pophi1): Likewise.

From-SVN: r207984

10 years agoi386.c (ix86_expand_vec_perm): Use V8SImode mode for mask of V8SFmode permutation.
Jakub Jelinek [Fri, 21 Feb 2014 07:48:07 +0000 (08:48 +0100)]
i386.c (ix86_expand_vec_perm): Use V8SImode mode for mask of V8SFmode permutation.

* config/i386/i386.c (ix86_expand_vec_perm): Use V8SImode
mode for mask of V8SFmode permutation.

From-SVN: r207981

10 years agoFix PR c++/60052 and PR c++/60053.
Adam Butcher [Fri, 21 Feb 2014 07:47:55 +0000 (07:47 +0000)]
Fix PR c++/60052 and PR c++/60053.

PR c++/60052
PR c++/60053
* parser.c (cp_parser_parameter_declaration_list): Correctly reset
implicit_template_scope upon leaving an out-of-line generic member
function definition.

PR c++/60052
PR c++/60053
* g++.dg/cpp1y/pr60052.C: New testcase.
* g++.dg/cpp1y/pr60053.C: New testcase.

From-SVN: r207980

10 years agore PR fortran/60286 (INQUIRE reports STDOUT as not writable)
Tobias Burnus [Fri, 21 Feb 2014 07:37:06 +0000 (08:37 +0100)]
re PR fortran/60286 (INQUIRE reports STDOUT as not writable)

2014-02-21  Tobias Burnus  <burnus@net-b.de>

        PR fortran/60286
        * libgfortran/io/inquire.c (yes, no): New static const char
        * vars.
        (inquire_via_unit): Use them. Use OPEN mode instead of using
        POSIX's access to query about write=, read= and readwrite=.

2014-02-21  Tobias Burnus  <burnus@net-b.de>

        PR fortran/60286
        * gfortran.dg/inquire_16.f90: New.

From-SVN: r207979

10 years agoruntime: Use a better heap location on arm64 systems.
Ian Lance Taylor [Fri, 21 Feb 2014 03:24:03 +0000 (03:24 +0000)]
runtime: Use a better heap location on arm64 systems.

Before this, the heap location used on a 64-bit system was not
available to user-space on arm64, so the "32-bit" strategy ended up
being used.  So use somewhere that is available, and for bonus points
is far away from where the kernel allocates address space by default.

From-SVN: r207977

10 years agoDaily bump.
GCC Administrator [Fri, 21 Feb 2014 00:16:45 +0000 (00:16 +0000)]
Daily bump.

From-SVN: r207976

10 years agore PR c++/60272 (atomic<>::compare_exchange_weak has spurious store and can cause...
Richard Henderson [Fri, 21 Feb 2014 00:11:43 +0000 (16:11 -0800)]
re PR c++/60272 (atomic<>::compare_exchange_weak has spurious store and can cause race conditions)

PR c++/60272

        * builtins.c (expand_builtin_atomic_compare_exchange): Always make
        a new pseudo for OLDVAL.

From-SVN: r207972

10 years agore PR target/57896 (ICE in expand_expr_real_2)
Jakub Jelinek [Thu, 20 Feb 2014 20:39:46 +0000 (21:39 +0100)]
re PR target/57896 (ICE in expand_expr_real_2)

PR target/57896
* config/i386/i386.c (expand_vec_perm_interleave2): Don't call
gen_reg_rtx if d->testing_p.
(expand_vec_perm_pshufb2, expand_vec_perm_broadcast_1): Return early
if d->testing_p and we will certainly return true.
(expand_vec_perm_even_odd_1): Likewise.  Don't call gen_reg_rtx
if d->testing_p.

From-SVN: r207969

10 years agoemit-rtl.c (gen_reg_rtx): Assert that crtl->emit.regno_pointer_align_length is non...
Uros Bizjak [Thu, 20 Feb 2014 20:21:42 +0000 (21:21 +0100)]
emit-rtl.c (gen_reg_rtx): Assert that crtl->emit.regno_pointer_align_length is non-zero.

* emit-rtl.c (gen_reg_rtx): Assert that
crtl->emit.regno_pointer_align_length is non-zero.

From-SVN: r207968

10 years agolto-partition.c (add_symbol_to_partition_1, [...]): Aliases have no defined size.
Jan Hubicka [Thu, 20 Feb 2014 19:40:14 +0000 (20:40 +0100)]
lto-partition.c (add_symbol_to_partition_1, [...]): Aliases have no defined size.

* lto-partition.c (add_symbol_to_partition_1,
undo_partition, lto_balanced_map): Aliases have no
defined size.
(lto_balanced_map): Do not follow refering variables
if they can be optimized out.

From-SVN: r207967

10 years agore PR c++/60272 (atomic<>::compare_exchange_weak has spurious store and can cause...
Richard Henderson [Thu, 20 Feb 2014 17:43:53 +0000 (09:43 -0800)]
re PR c++/60272 (atomic<>::compare_exchange_weak has spurious store and can cause race conditions)

PR c++/60272

gcc/
* builtins.c (expand_builtin_atomic_compare_exchange): Conditionalize
on failure the store back into EXPECT.
libatomic/
* cas_n.c (libat_compare_exchange): Conditionalize on failure
the store back to EPTR.

From-SVN: r207966

10 years agonios2.md (unspec): Add UNSPEC_PIC_GOTOFF_SYM enum.
Chung-Lin Tang [Thu, 20 Feb 2014 16:35:10 +0000 (16:35 +0000)]
nios2.md (unspec): Add UNSPEC_PIC_GOTOFF_SYM enum.

2014-02-20  Chung-Lin Tang  <cltang@codesourcery.com>
    Sandra Loosemore  <sandra@codesourcery.com>

gcc/
* config/nios2/nios2.md (unspec): Add UNSPEC_PIC_GOTOFF_SYM enum.
* config/nios2/nios2.c (nios2_function_profiler):
Add -fPIC (flag_pic == 2) support.
(nios2_handle_custom_fpu_cfg): Fix warning parameter.
(nios2_large_offset_p): New function.
(nios2_unspec_reloc_p): Move up position, update to use
nios2_large_offset_p.
(nios2_unspec_address): Remove function.
(nios2_unspec_offset): New function.
(nios2_large_got_address): New function.
(nios2_got_address): Add large offset support.
(nios2_legitimize_tls_address): Update usage of removed and new
functions.
(nios2_symbol_binds_local_p): New function.
(nios2_load_pic_address): Add -fPIC (flag_pic == 2) support.
(nios2_legitimize_address): Update to use nios2_large_offset_p.
(nios2_emit_move_sequence): Avoid legitimizing (const (unspec ...)).
(nios2_print_operand): Merge H/L processing, add hiadj/lo
processing for (const (unspec ...)).
(nios2_unspec_reloc_name): Add UNSPEC_PIC_GOTOFF_SYM case.

gcc/testsuite/
* gcc.target/nios2/biggot-1.c: New.
* gcc.target/nios2/biggot-2.c: New.

libgcc/
* config/nios2/t-nios2 (CRTSTUFF_T_CFLAGS): Add -mno-gpopt.
* config/nios2/crti.S: Remove .file directive.
* config/nios2/crtn.S: Likewise.

From-SVN: r207965

10 years agoRename shared_mutex to shared_timed_mutex per C++14 acceptance of N3891.
Ed Smith-Rowland [Thu, 20 Feb 2014 16:06:40 +0000 (16:06 +0000)]
Rename shared_mutex to shared_timed_mutex per C++14 acceptance of N3891.

2014-02-20  Ed Smith-Rowland  <3dw4rd@verizon.net>

Rename shared_mutex to shared_timed_mutex per C++14 acceptance of N3891.
* include/std/shared_mutex: Rename shared_mutex to shared_timed_mutex.
* testsuite/30_threads/shared_lock/locking/2.cc: Ditto.
* testsuite/30_threads/shared_lock/locking/4.cc: Ditto.
* testsuite/30_threads/shared_lock/locking/1.cc: Ditto.
* testsuite/30_threads/shared_lock/locking/3.cc: Ditto.
* testsuite/30_threads/shared_lock/requirements/
explicit_instantiation.cc: Ditto.
* testsuite/30_threads/shared_lock/requirements/typedefs.cc: Ditto.
* testsuite/30_threads/shared_lock/cons/2.cc: Ditto.
* testsuite/30_threads/shared_lock/cons/4.cc: Ditto.
* testsuite/30_threads/shared_lock/cons/1.cc: Ditto.
* testsuite/30_threads/shared_lock/cons/6.cc: Ditto.
* testsuite/30_threads/shared_lock/cons/3.cc: Ditto.
* testsuite/30_threads/shared_lock/cons/5.cc: Ditto.
* testsuite/30_threads/shared_lock/modifiers/2.cc: Ditto.
* testsuite/30_threads/shared_lock/modifiers/1.cc: Ditto.
* testsuite/30_threads/shared_mutex/requirements/
standard_layout.cc: Ditto.
* testsuite/30_threads/shared_mutex/cons/copy_neg.cc: Ditto.
* testsuite/30_threads/shared_mutex/cons/1.cc: Ditto.
* testsuite/30_threads/shared_mutex/cons/assign_neg.cc: Ditto.
* testsuite/30_threads/shared_mutex/try_lock/2.cc: Ditto.
* testsuite/30_threads/shared_mutex/try_lock/1.cc: Ditto.

From-SVN: r207964

10 years agore PR c++/58873 ([c++11] ICE with __underlying_type for broken enum)
Kai Tietz [Thu, 20 Feb 2014 16:02:24 +0000 (17:02 +0100)]
re PR c++/58873 ([c++11] ICE with __underlying_type for broken enum)

PR c++/58873
* parser.c (cp_parser_functional_cast): Treat NULL_TREE
valued type argument as error_mark_node.

From-SVN: r207961

10 years agore PR go/60288 (gccgo crashes compiling '*func_ptr(0)')
Ian Lance Taylor [Thu, 20 Feb 2014 15:20:26 +0000 (15:20 +0000)]
re PR go/60288 (gccgo crashes compiling '*func_ptr(0)')

PR go/60288
compiler: Avoid crash, give error for *&x when x is not addressable.

From-SVN: r207960

10 years agotree-cfg.c (replace_uses_by): Mark altered BBs before doing the substitution.
Richard Biener [Thu, 20 Feb 2014 14:56:27 +0000 (14:56 +0000)]
tree-cfg.c (replace_uses_by): Mark altered BBs before doing the substitution.

2014-02-20  Richard Biener  <rguenther@suse.de>

* tree-cfg.c (replace_uses_by): Mark altered BBs before
doing the substitution.
(verify_gimple_assign_single): Also verify bare MEM_REFs
on the lhs.

From-SVN: r207958

10 years agore PR c++/58835 (ICE with __PRETTY_FUNCTION__ in broken function)
Kai Tietz [Thu, 20 Feb 2014 14:28:16 +0000 (15:28 +0100)]
re PR c++/58835 (ICE with __PRETTY_FUNCTION__ in broken function)

PR c++/58835
* semantics.c (finish_fname): Handle error_mark_node.

From-SVN: r207955

10 years ago2014-02-20 Sergey Rybin <rybin@adacore.com frybin>
Arnaud Charlet [Thu, 20 Feb 2014 14:21:35 +0000 (15:21 +0100)]
2014-02-20  Sergey Rybin  <rybin@adacore.com frybin>

* gnat_ugn.texi, vms_data.ads (gnatelim): Add description of
project-specific options.

2014-02-20  Ed Schonberg  <schonberg@adacore.com>

* a-cbdlli.adb (Insert): When capacity exceeded, raise Capacity_Error,
not Constraint_Error.
* a-cbmutr.adb (Append_Child, Insert_Child, Prepend_Child): Ditto.
* sem_ch5.adb: Code clean up.

From-SVN: r207954

10 years ago[multiple changes]
Arnaud Charlet [Thu, 20 Feb 2014 14:19:23 +0000 (15:19 +0100)]
[multiple changes]

2014-02-20  Hristian Kirtchev  <kirtchev@adacore.com>

* sem_prag.adb (Usage_Error): Remove local
constant Typ. Remove the specialized diagnostics for unconstrained
or tagged items as those are not part of the explicit input set
of the related subprogram and should not be flagged.

2014-02-20  Ed Schonberg  <schonberg@adacore.com>

* sem_attr.adb: Add guard to preserve all errors.

2014-02-20  Vincent Celier  <celier@adacore.com>

* switch-m.adb (Normalize_Compiler_Switches): Take into account
switches that are recorded in ALI files: -gnateA, -gnateE,
-gnateF, -gnateinn, -gnateu, -gnateV and -gnateY.

2014-02-20  Ed Schonberg  <schonberg@adacore.com>

* sem_ch5.adb (Analyze_Iterator_Specification): Check legality
of an element iterator form over a formal container with an
Iterable aspect.
* exp_ch5.adb (Build_Formal_Container_Iteration): Utility
to create declaration and loop statements for both forms of
container iterators.
(Expand_Formal_Container_Element_Iterator): New procedure
to handle loops of the form  "for E of C" when C is a formal
container.
(Expand_Formal_Container_Iterator): Code cleanup.

From-SVN: r207953

10 years ago[multiple changes]
Arnaud Charlet [Thu, 20 Feb 2014 14:08:31 +0000 (15:08 +0100)]
[multiple changes]

2014-02-20  Hristian Kirtchev  <kirtchev@adacore.com>

* sem_prag.adb (Add_Item_To_Name_Buffer): New routine.
(Analyze_Contract_Case): Remove the use of
"may". Replace "aspect Contract_Cases" to avoid categorization
of aspect vs pragma.
(Analyze_External_Property_In_Decl_Part): Remove the use of "formal".
(Analyze_Global_Item): Remove
the use of "formal", specify the subprogram.  Split the
error message about a state with visible refinement into
two. Remove the use of "global" from "volatile global item".
(Analyze_Initialization_Item): Ensure that the SPARK RM reference
is on one line.
(Analyze_Input_Output): Update the call to
Check_Mode. Specify the duplicated item. Reword the error
message concerning an input of a null output list. Use "\"
for error message continuation.
(Analyze_Part_Of): Remove
the use of "may". Use "\" for error message continuation.
(Analyze_Refined_Depends_In_Decl_Part): Update the error
message concerning a useless refinement to match the format
of Refined_Global.
(Analyze_Refined_Global_In_Decl_Part): Reword the error message
concerning a useless refinement.
(Analyze_Refinement_Clause): Use "\" for error message continuation.
(Check_Constituent_Usage): Use "\" for error message continuation.
(Check_Dependency_Clause): Use "\" for error message continuation.
(Check_Matching_Constituent): Use "\" for error message continuation.
(Check_Missing_Part_Of): Use "\" for error message continuation.
(Check_Mode): Renamed to
Check_Role. Update the comment on usage. Redo the error reporting
to use Role_Error.
(Check_Mode_Restriction_In_Enclosing_Context): Use "\" for error
message continuation.
(Find_Mode): Renamed to Find_Role. Update the parameter profile along
with comment on usage. Update all occurrences of Is_Input and Is_Output.
(Inconsistent_Mode_Error): Use "\" for error message continuation.
(Input_Match): Use "\" for error message continuation.
(Role_Error): New routine.
(Set_Convention_From_Pragma): Use "\" for error message continuation.
(Usage_Error): Add local variable Error_Msg. Build specialized error
message showcasing the offending item kind. Redo the diagnostics for
unconstrained types.

2014-02-20  Ed Schonberg  <schonberg@adacore.com>

* exp_util.adb (Is_Iterated_Container): Use high-level primitives
to determine whether a type is a container type, rather than
examining only its type declaration, so that subtypes and derived
types are handled properly.

2014-02-20  Sergey Rybin  <rybin@adacore.com frybin>

* gnat_ugn.texi: gnatmetric: add description of project-specific
options.

2014-02-20  Ed Schonberg  <schonberg@adacore.com>

* sem_attr.adb (Analyze_Attribute, case 'Update): Verify that
all choices in an association for a record type have the same
type, as is done for record aggregates.

From-SVN: r207952

10 years agoXFAIL sourcelocation (PR libgcj/55637)
Rainer Orth [Thu, 20 Feb 2014 14:04:53 +0000 (14:04 +0000)]
XFAIL sourcelocation (PR libgcj/55637)

PR libgcj/55637
* testsuite/libjava.lang/sourcelocation.xfail: New file.

From-SVN: r207951

10 years ago[multiple changes]
Arnaud Charlet [Thu, 20 Feb 2014 14:04:38 +0000 (15:04 +0100)]
[multiple changes]

2014-02-20  Robert Dewar  <dewar@adacore.com>

* a-cborma.adb, a-cbhama.adb, a-cbdlli.adb, a-cbmutr.adb: Use pragma
Unmodified rather than Warnings (Off). Make comments
uniform in the four affected units.

2014-02-20  Robert Dewar  <dewar@adacore.com>

* sem_ch13.adb (Analyze_Attribute_Definition_Clause, case
Object_Size): For non-scalar types allow any value that is a
multiple of 8.
* gnat_rm.texi: Document Object_Size for composites more clearly.

2014-02-20  Yannick Moy  <moy@adacore.com>

* sem_util.ads, sem_util.adb (Default_Initialization): Remove function.

2014-02-20  Ed Schonberg  <schonberg@adacore.com>

* stand.ads: Raise_Type: new predefined entity, used as the type
of a Raise_Expression prior to resolution.
* cstand.adb: Build entity for Raise_Type.
* sem_ch11.adb (Analyze_Raise_Expression): use Raise_Type as the
initial type of the node.
* sem_type.adb (Covers): Raise_Type is compatible with all
other types.
* sem_res.adb (Resolve): Remove special handling of Any_Type on
Raise_Expression nodes.
(Resolve_Raise_Expression): Signal ambiguity if the type of the
context is still Raise_Type.

From-SVN: r207950

10 years ago[multiple changes]
Arnaud Charlet [Thu, 20 Feb 2014 14:02:27 +0000 (15:02 +0100)]
[multiple changes]

2014-02-20  Robert Dewar  <dewar@adacore.com>

* sem_ch12.adb (Validate_Access_Type_Instance): Add message if
mismatching predicates.
* sem_ch6.adb (Check_Conformance): Give better messages on
predicate mismatch.
* sem_eval.adb (Predicates_Match): Move to spec.
* sem_eval.ads (Predicates_Match): Moved here from body.

2014-02-20  Ed Schonberg  <schonberg@adacore.com>

* a-cbmutr.adb: Use default value in Insert_Child.

From-SVN: r207949

10 years ago[multiple changes]
Arnaud Charlet [Thu, 20 Feb 2014 13:54:18 +0000 (14:54 +0100)]
[multiple changes]

2014-02-20  Vincent Celier  <celier@adacore.com>

* gnatcmd.adb, make.adb, prj-makr.adb, clean.adb: Call
Stringt.Initialize in all project aware tools.

2014-02-20  Doug Rupp  <rupp@adacore.com>

* gnat_ugn.texi: Document gnatbind -X option.
* vms_data.ads: Minor warnings documentation reformatting.

2014-02-20  Ed Schonberg  <schonberg@adacore.com>

* exp_ch3.adb (Expand_Freeze_Array_Type): Only create invariant
procedure for a base type.

2014-02-20  Robert Dewar  <dewar@adacore.com>

* sem_ch4.adb (Analyze_Case_Expression): Get type from first
expression with type info.
* sem_ch13.adb: Minor reformatting.
* sem_eval.adb (Subtypes_Statically_Match): Make sure we return
False if predicates do not match on the two types.

2014-02-20  Arnaud Charlet  <charlet@adacore.com>

* sem_prag.adb (Analyze_Pragma [pragma Attach_Handler]):
In Relaxed_RM_Semantics mode, allow any static integer value,
for compatibility with other compilers.

From-SVN: r207948

10 years ago[multiple changes]
Arnaud Charlet [Thu, 20 Feb 2014 13:52:08 +0000 (14:52 +0100)]
[multiple changes]

2014-02-20  Vincent Celier  <celier@adacore.com>

* errutil.adb (Initialize): Properly initialize entry in table
Warnings when warnings are suppressed.

2014-02-20  Robert Dewar  <dewar@adacore.com>

* s-os_lib.ads: Add warning about non-portability of Rename_File.
* sem_util.ads, sem_util.adb (Is_Junk_Name): New function.
* sem_warn.adb: Use Is_Junk_Name to suppress some warnings.
* gnat_ugn.texi: Document no warnings on junk variable names.
* layout.adb: Minor code reorganization (use Ekind_In).
* stringt.adb: Move initialization of Null_String_Id to Initialize
routine.

2014-02-20  Ed Schonberg  <schonberg@adacore.com>

* sem_ch13.adb: Initialize optional Element_Id.

From-SVN: r207947

10 years ago[multiple changes]
Arnaud Charlet [Thu, 20 Feb 2014 13:48:32 +0000 (14:48 +0100)]
[multiple changes]

2014-02-20  Robert Dewar  <dewar@adacore.com>

* s-os_lib.ads (Rename_File): Minor commment addition.

2014-02-20  Thomas Quinot  <quinot@adacore.com>

* einfo.ads: Minor reformatting.

2014-02-20  Hristian Kirtchev  <kirtchev@adacore.com>

* aspects.adb (Exchange_Aspects): New routine.
* aspects.ads (Exchange_Aspects): New routine.
* atree.adb (Rewrite): Do not check whether the save node has
aspects as it never will, instead check the node about to be clobbered.
* einfo.adb (Write_Field25_Name): Abstract_States can appear in
entities of generic packages.
* sem_ch6.adb (Analyze_Expression_Function): Fix the parent
pointer of an aspect specification list after rewriting takes place.
* sem_ch7.adb (Analyze_Package_Body_Helper): Swap the aspect
specifications of the generic template and the copy used for analysis.
* sem_ch12.adb (Analyze_Generic_Package_Declaration): Swap
the aspect specifications of the generic template and the
copy used for analysis.
(Analyze_Package_Instantiation): Propagate the aspect specifications
from the generic template to the instantiation.
(Build_Instance_Compilation_Unit_Nodes): Propagate the aspect
specifications from the generic template to the instantiation.
* sem_ch13.adb (Analyze_Aspect_Specifications): Handle aspects
Abstract_State, Initializes and Initial_Condition when they
apply to a package instantiation.

2014-02-20  Robert Dewar  <dewar@adacore.com>

* stringt.adb: Add call to Initialize in package initialization.

From-SVN: r207946

10 years ago[multiple changes]
Arnaud Charlet [Thu, 20 Feb 2014 13:46:15 +0000 (14:46 +0100)]
[multiple changes]

2014-02-20  Robert Dewar  <dewar@adacore.com>

* a-crbtgk.adb, a-cihama.adb, a-coinve.adb, a-ciorse.adb, a-crbtgo.adb,
a-cidlli.adb, a-cimutr.adb, a-cihase.adb, a-cohama.adb, a-coorse.adb,
a-chtgke.adb, a-chtgop.adb, a-comutr.adb, a-ciorma.adb, a-cobove.adb,
a-convec.adb, a-cohase.adb, a-chtgbk.adb, a-chtgbo.adb: Minor
reformatting.

2014-02-20  Bob Duff  <duff@adacore.com>

* s-os_lib.ads: Minor: Remove incorrect comment.

2014-02-20  Robert Dewar  <dewar@adacore.com>

* sem_elab.adb (Check_Elab_Assign): Clearer warning message.

2014-02-20  Vincent Celier  <celier@adacore.com>

* gnat_rm.texi: Minor syntax error fix.

2014-02-20  Ed Schonberg  <schonberg@adacore.com>

* a-cborma.adb (Assign): When creating a node without a specified
element, insert an uninitialized element in the map, because
the instance may provide an element type with a default
initialization, e.g a scalar with a Default_Value aspect.
* a-cbhama.adb (Assign_Key): Remove useless Allocate procedure.
(Insert): In the version without explicit element, provide an
uninitialized element, as above.
* a-cbdlli.adb (Append): In the version without explicit element,
provide an uninitalized element, as above.
(Allocate): Remove unused version.

From-SVN: r207945

10 years ago[multiple changes]
Arnaud Charlet [Thu, 20 Feb 2014 13:44:20 +0000 (14:44 +0100)]
[multiple changes]

2014-02-20  Robert Dewar  <dewar@adacore.com>

* sem_elab.adb: Minor code reorganization (use Nkind_In).
* stringt.adb: Remove temporary pragma Warnings (Off).
* stringt.ads: Add pragma Elaborate_Body to ensure initialization
of Null_String_Id.

2014-02-20  Matthew Heaney  <heaney@adacore.com>

* a-chtgbk.adb (Replace): Use correct offset when calculating bucket
index.

From-SVN: r207944

10 years ago[multiple changes]
Arnaud Charlet [Thu, 20 Feb 2014 13:42:58 +0000 (14:42 +0100)]
[multiple changes]

2014-02-20  Ed Schonberg  <schonberg@adacore.com>

* sem_ch5.adb (Analyze_Iterator_Specification): Initialize
properly the cursor type for subsequent volatile testing in SPARK
mode, when domain is a formal container with an Iterabe aspect.

2014-02-20  Robert Dewar  <dewar@adacore.com>

* errout.adb (Set_Warnings_Mode_Off): Add Reason argument.
(Set_Specific_Warning_Off): Add Reason argument.
* errout.ads (Set_Warnings_Mode_Off): Add Reason argument.
(Set_Specific_Warning_Off): Add Reason argument.
* erroutc.adb (Warnings_Entry): Add Reason field
(Specific_Warning_Entry): Add Reason field.
(Warnings_Suppressed): return String_Id for Reason.
(Warning_Specifically_Suppressed): return String_Id for Reason.
* erroutc.ads (Warnings_Entry): Add Reason field.
(Specific_Warning_Entry): Add Reason field.
(Set_Specific_Warning_Off): Add Reason argument.
(Set_Warnings_Mode_Off): Add Reason argument.
(Warnings_Suppressed): return String_Id for Reason.
(Warning_Specifically_Suppressed): return String_Id for Reason.
* errutil.adb (Warnings_Suppressed): returns String_Id for Reason
(Warning_Specifically_Suppressed): returns String_Id for Reason
* gnat_rm.texi: Document that Warning parameter is string literal
or a concatenation of string literals.
* par-prag.adb: New handling for Reason argument.
* sem_prag.adb (Analyze_Pragma, case Warning): New handling
for Reason argument.
* sem_util.ads, sem_util.adb (Get_Reason_String): New procedure.
* sem_warn.ads (Warnings_Off_Entry): Add reason field.
* stringt.adb: Set Null_String_Id.
* stringt.ads (Null_String_Id): New constant.

From-SVN: r207943

10 years agoeinfo.ads: Minor comment addition: Etype of package is Standard_Void_Type.
Robert Dewar [Thu, 20 Feb 2014 13:38:06 +0000 (13:38 +0000)]
einfo.ads: Minor comment addition: Etype of package is Standard_Void_Type.

2014-02-20  Robert Dewar  <dewar@adacore.com>

* einfo.ads: Minor comment addition: Etype of package is
Standard_Void_Type.
* checks.adb, exp_aggr.adb, exp_atag.adb, exp_attr.adb, exp_ch13.adb,
exp_ch2.adb, exp_ch3.adb, exp_ch4.adb, exp_ch5.adb, exp_ch6.adb,
exp_ch7.adb, exp_ch9.adb, exp_dbug.adb, exp_disp.adb, exp_imgv.adb,
exp_intr.adb, exp_prag.adb, exp_sel.adb, exp_strm.adb, exp_util.adb,
freeze.adb, rtsfind.adb, sem_aggr.adb, sem_attr.adb, sem_ch10.adb,
sem_ch12.adb, sem_ch13.adb, sem_ch3.adb, sem_ch4.adb, sem_ch5.adb,
sem_ch6.adb, sem_ch8.adb, sem_dim.adb, sem_prag.adb, sem_res.adb,
sem_util.adb, tbuild.adb, tbuild.ads: Remove New_Reference_To.
Replace all calls with calls to New_Occurrence_Of.

2014-02-20  Robert Dewar  <dewar@adacore.com>

* par-util.adb (Ignore, case of right paren): Make this a
serious error.

From-SVN: r207942

10 years agore PR tree-optimization/55260 (ICE: in ipa_get_parm_lattices, at ipa-cp.c:263 with...
Martin Jambor [Thu, 20 Feb 2014 13:28:34 +0000 (14:28 +0100)]
re PR tree-optimization/55260 (ICE: in ipa_get_parm_lattices, at ipa-cp.c:263 with -O2 -fno-inline -fipa-cp-clone)

2014-02-20  Martin Jambor  <mjambor@suse.cz>

PR ipa/55260
* ipa-cp.c (cgraph_edge_brings_all_agg_vals_for_node): Uce correct
info when checking whether lattices are bottom.

testsuite/
* gcc.dg/ipa/pr55260.c: New test.

From-SVN: r207941

10 years agocrossmodule-indircall-1.c: Return 0 for execution test case.
Bin Cheng [Thu, 20 Feb 2014 10:41:03 +0000 (10:41 +0000)]
crossmodule-indircall-1.c: Return 0 for execution test case.

* gcc.dg/tree-prof/crossmodule-indircall-1.c: Return 0
for execution test case.

From-SVN: r207939

10 years agore PR middle-end/60221 (gcc -fexceptions generates unnecessary cleanup code)
Richard Biener [Thu, 20 Feb 2014 09:00:23 +0000 (09:00 +0000)]
re PR middle-end/60221 (gcc -fexceptions generates unnecessary cleanup code)

2014-02-20  Richard Biener  <rguenther@suse.de>

PR middle-end/60221
* tree-eh.c (execute_cleanup_eh_1): Also cleanup empty EH
regions at -O0.

From-SVN: r207937

10 years agore PR java/60261 (Weird java install with --enable-version-specific-runtime-libs)
Richard Biener [Thu, 20 Feb 2014 08:43:04 +0000 (08:43 +0000)]
re PR java/60261 (Weird java install with --enable-version-specific-runtime-libs)

2014-02-20  Richard Biener  <rguenther@suse.de>

PR libjava/60261
* configure.ac (dbexecdir): Remove redundant multilib addition.
* Makefile.am (gcc_version): Export.
* configure: Regenerate.
* Makefile.in: Likewise.

From-SVN: r207936

10 years agore PR middle-end/58555 (Floating point exception in want_inline_self_recursive_call_p)
Jan Hubicka [Thu, 20 Feb 2014 06:40:07 +0000 (07:40 +0100)]
re PR middle-end/58555 (Floating point exception in want_inline_self_recursive_call_p)

PR ipa/58555
* ipa-inline-transform.c (clone_inlined_nodes): Add freq_scale parameter
specifying the scaling.
(inline_call): Update.
(want_inline_recursively): Guard division by zero.
(recursive_inlining): Update.
* ipa-inline.h (clone_inlined_nodes): Update.
* testsuite/g++.dg/torture/pr58555.C: New testcase.

From-SVN: r207934

10 years agore PR target/60204 (struct with __m512i is mishandled in function parameter passing...
Ilya Tocar [Thu, 20 Feb 2014 06:32:21 +0000 (06:32 +0000)]
re PR target/60204 (struct with __m512i is mishandled in function parameter passing and return)

gcc/
PR target/60204
* config/i386/i386.c (classify_argument): Pass structures of size
64 bytes or less in register.

tessuite/gcc/
PR target/60204
* gcc.target/x86_64/abi/avx512f/test_passing_structs.c: Update to
reflect abi fix.
* gcc.target/x86_64/abi/avx512f/test_passing_unions.c: Ditto.

From-SVN: r207933

10 years agoavx512erintrin.h (_mm_rcp28_round_sd): Swap operands.
Ilya Tocar [Thu, 20 Feb 2014 06:26:38 +0000 (06:26 +0000)]
avx512erintrin.h (_mm_rcp28_round_sd): Swap operands.

gcc/
* config/i386/avx512erintrin.h (_mm_rcp28_round_sd): Swap operands.
(_mm_rcp28_round_ss): Ditto.
(_mm_rsqrt28_round_sd): Ditto.
(_mm_rsqrt28_round_ss): Ditto.
* config/i386/avx512erintrin.h (_mm_rcp14_round_sd): Ditto.
(_mm_rcp14_round_ss): Ditto.
(_mm_rsqrt14_round_sd): Ditto.
(_mm_rsqrt14_round_ss): Ditto.
* config/i386/sse.md (rsqrt14<mode>): Put nonimmediate operand as
the first input operand, get rid of match_dup.
(avx512er_exp2<mode><mask_name><round_saeonly_name>): Set type
attribute to sse.
(<mask_codefor>avx512er_rcp28<mode><mask_name><round_saeonly_name>):
Ditto.
(avx512er_vmrcp28<mode><round_saeonly_name>): Put nonimmediate
operand as the first input operand, set type attribute.
(<mask_codefor>avx512er_rsqrt28<mode><mask_name><round_saeonly_name>):
Set type attribute.
(avx512er_vmrsqrt28<mode><round_saeonly_name>): Put nonimmediate
operand as the first input operand, set type attribute.

testsuite/gcc/
* gcc.target/i386/avx512er-vrcp28sd-2.c: Distinguish src1 and src2.
* gcc.target/i386/avx512er-vrcp28ss-2.c: Call correct intrinsic.
* gcc.target/i386/avx512er-vrsqrt28sd-2.c: Distinguish src1 and src2.
* gcc.target/i386/avx512er-vrsqrt28ss-2.c: Ditto.
* gcc.target/i386/avx512f-vrcp14sd-2.c: Fix reference calculation.
* gcc.target/i386/avx512f-vrcp14ss-2.c: Fix reference calculation.

Co-Authored-By: Kirill Yukhin <kirill.yukhin@intel.com>
From-SVN: r207932

10 years agoDaily bump.
GCC Administrator [Thu, 20 Feb 2014 00:17:08 +0000 (00:17 +0000)]
Daily bump.

From-SVN: r207930

10 years agore PR c/37743 (Bogus printf format warning with __builtin_bswap32.)
Jakub Jelinek [Wed, 19 Feb 2014 22:27:40 +0000 (23:27 +0100)]
re PR c/37743 (Bogus printf format warning with __builtin_bswap32.)

PR c/37743
* c-common.c (c_common_nodes_and_builtins): When initializing
c_uint{16,32,64}_type_node, also set corresponding
uint{16,32,64}_type_node to the same value.

* g++.dg/ext/builtin-bswap1.C: New test.
* c-c++-common/pr37743.c: New test.

From-SVN: r207924

10 years agors6000.c (vspltis_constant): Fix most significant bit of zero.
Bill Schmidt [Wed, 19 Feb 2014 19:43:13 +0000 (19:43 +0000)]
rs6000.c (vspltis_constant): Fix most significant bit of zero.

2014-02-19  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>

* config/rs6000/rs6000.c (vspltis_constant): Fix most significant
bit of zero.

From-SVN: r207919

10 years agore PR c++/60046 (internal compiler error: in nothrow_spec_p, at cp/except.c:1280)
Jason Merrill [Wed, 19 Feb 2014 19:03:19 +0000 (14:03 -0500)]
re PR c++/60046 (internal compiler error: in nothrow_spec_p, at cp/except.c:1280)

PR c++/60046
* pt.c (maybe_instantiate_noexcept): Don't instantiate exception
spec from template context.

From-SVN: r207917

10 years agoc-parser.c (c_parser_declspecs): Replace call to error by error_at.
Prathamesh Kulkarni [Wed, 19 Feb 2014 18:51:57 +0000 (18:51 +0000)]
c-parser.c (c_parser_declspecs): Replace call to error by error_at.

* c-parser.c (c_parser_declspecs): Replace call to error by error_at.
(c_parser_parameter_declaration): Likewise.

From-SVN: r207916

10 years agore PR debug/56563 (no debuginfo for "explicit" operator)
Jakub Jelinek [Wed, 19 Feb 2014 18:12:31 +0000 (19:12 +0100)]
re PR debug/56563 (no debuginfo for "explicit" operator)

PR debug/56563
* cp-objcp-common.c (cp_function_decl_explicit_p): Remove
FUNCTION_FIRST_USER_PARMTYPE (decl) != void_list_node check.

From-SVN: r207915

10 years agore PR c++/60267 (ICE in c_pp_lookup_pragma, at c-family/c-pragma.c:1232; ICE in tsubs...
Jakub Jelinek [Wed, 19 Feb 2014 18:11:54 +0000 (19:11 +0100)]
re PR c++/60267 (ICE in c_pp_lookup_pragma, at c-family/c-pragma.c:1232; ICE in tsubst_copy, at cp/pt.c:12887)

PR c++/60267
* pt.c (tsubst_expr): Handle ANNOTATE_EXPR.

* g++.dg/ext/ivdep-1.C: New test.

From-SVN: r207914

10 years agoRemove TFmode check for X86_64_INTEGER_CLASS
H.J. Lu [Wed, 19 Feb 2014 18:10:04 +0000 (18:10 +0000)]
Remove TFmode check for X86_64_INTEGER_CLASS

PR target/60207
* config/i386/i386.c (construct_container): Remove TFmode check
for X86_64_INTEGER_CLASS.

From-SVN: r207913

10 years agore PR c++/60267 (ICE in c_pp_lookup_pragma, at c-family/c-pragma.c:1232; ICE in tsubs...
Jakub Jelinek [Wed, 19 Feb 2014 16:45:21 +0000 (17:45 +0100)]
re PR c++/60267 (ICE in c_pp_lookup_pragma, at c-family/c-pragma.c:1232; ICE in tsubst_copy, at cp/pt.c:12887)

PR c++/60267
* c-pragma.c (init_pragma): Don't call cpp_register_deferred_pragma
for PRAGMA_IVDEP if flag_preprocess_only.

* gcc.dg/pr60267.c: New test.

From-SVN: r207911

10 years agore PR target/59794 (i386 backend fails to detect MMX/SSE/AVX ABI changes)
Uros Bizjak [Wed, 19 Feb 2014 15:53:59 +0000 (16:53 +0100)]
re PR target/59794 (i386 backend fails to detect MMX/SSE/AVX ABI changes)

PR target/59794
* config/i386/i386.c (type_natural_mode): Warn for ABI changes
only when -Wpsabi is enabled.

testsuite/ChangeLog:

PR target/59794
* gcc.target/i386/pr39162.c: Add dg-prune-output.
(dg-options): Remove -Wno-psabi.
* gcc.target/i386/59794-2.c: Ditto.
* gcc.target/i386/60205-1.c: Ditto.
* gcc.target/i386/sse-5.c: Ditto.

From-SVN: r207910

10 years agopr46966.c: Only run on 32-bit+ targets.
Nick Clifton [Wed, 19 Feb 2014 15:44:11 +0000 (15:44 +0000)]
pr46966.c: Only run on 32-bit+ targets.

* gcc.dg/graphite/pr46966.c: Only run on 32-bit+ targets.
* gcc.dg/pr23623.c: Likewise.
* gcc.dg/pr48784-1.c: Likewise.
* gcc.dg/pr48784-2.c: Likewise.
* gcc.dg/pr56997-2.c: Likewise.
* gcc.dg/sms-6.c: Likewise.
* gcc.dg/torture/pr60183.c: Likewise.
* gcc.dg/torture/vec-cvt-1.c: Likewise.
* gcc.c-torture/execute/20061220-1.x: New.
* gcc.c-torture/execute/pr43220.x: New.
* gcc.c-torture/execute/pr51581-1.x: New.
* gcc.c-torture/execute/pr51581-2.x: New.
* gcc.c-torture/execute/pr58570.x: New.
* gcc.c-torture/unsorted/DFcmp.x: New.
* gcc.c-torture/unsorted/SFset.x: New.

From-SVN: r207909

10 years agore PR target/59799 (aarch64_pass_by_reference never passes arrays by value, contrary...
Michael Hudson-Doyle [Wed, 19 Feb 2014 15:32:54 +0000 (15:32 +0000)]
re PR target/59799 (aarch64_pass_by_reference never passes arrays by value, contrary to ABI documentation)

2014-02-19  Michael Hudson-Doyle  <michael.hudson@linaro.org>

 PR target/59799
* config/aarch64/aarch64.c (aarch64_pass_by_reference): The rules for
passing arrays in registers are the same as for structs, so remove the
special case for them.

From-SVN: r207908

10 years ago[multiple changes]
Arnaud Charlet [Wed, 19 Feb 2014 15:05:43 +0000 (16:05 +0100)]
[multiple changes]

2014-02-19  Robert Dewar  <dewar@adacore.com>

* sem_ch5.adb, sem_prag.adb, sem_attr.adb: Minor reformatting.
* debug.adb: Remove -gnatd.X switch, no longer used.
* styleg.adb (Check_Then): Remove check of -gnatd.X, check is
now always made.

2014-02-19  Bob Duff  <duff@adacore.com>

* g-socket.ads: Minor: Correct some comments.

From-SVN: r207907

10 years ago[multiple changes]
Arnaud Charlet [Wed, 19 Feb 2014 15:04:08 +0000 (16:04 +0100)]
[multiple changes]

2014-02-19  Ed Schonberg  <schonberg@adacore.com>

* sem_ch5.adb: Inhibit iterator rewriting in ASIS mode.

2014-02-19  Arnaud Charlet  <charlet@adacore.com>

* sem_ch13.adb (Analyze_Aspect_Specifications): Do not generate
an error on out of range priorities if Relaxed_RM_Semantics.
* sem_prag.adb (Analyze_Pragma): Ditto.

2014-02-19  Bob Duff  <duff@adacore.com>

* sem_attr.adb (Resolve_Attribute): Detect the case of F'Access
where F denotes the renaming of an enumeration literal, and
issue a specialized error message.

From-SVN: r207906

10 years ago[multiple changes]
Arnaud Charlet [Wed, 19 Feb 2014 14:59:32 +0000 (15:59 +0100)]
[multiple changes]

2014-02-19  Matthew Heaney  <heaney@adacore.com>

* a-chtgop.ads (Checked_Index): New operation.
(Next): Changed mode of hash table.
* a-chtgop.adb (Adjust, Delete_Node_Sans_Free): Detect tampering
(Generic_Read, Reserve_Capacity): Ditto.
(Generic_Equal): Detect tampering.
(Next): Changed mode of hash table, detect tampering.
* a-chtgke.ads (Checked_Index, Checked_Equivalent_Keys): New
operation.
(Find): Changed mode of hash table.
* a-chtgke.adb (Checked_Equivalent_Keys): New operation
(Delete_Key_Sans_Free, Generic_Conditional_Insert): Detect
tampering.
(Find): Changed mode of hash table, check for tampering.
(Generic_Replace_Element): Check for tampering.
* a-chtgbk.ads (Checked_Index, Checked_Equivalent_Keys): New operation.
* a-chtgbk.adb (Checked_Index, Checked_Equivalent_Keys): New
operation (Delete_Key_Sans_Free, Generic_Conditional_Insert):
Detect tampering.
(Find, Generic_Replace_Element): Check for tampering.
* a-chtgbo.ads (Checked_Index): New operation.
* a-chtgbo.adb (Checked_Index): New operation
(Delete_Node_Sans_Free, Generic_Equal): Detect tampering.
(Generic_Read, Next): Ditto.
* a-cohase.adb, a-cihase.adb (Is_In): Changed mode of hash
table (Difference, Intersection): Use variable view of
source, detect tampering (Find, Is_Subset, Overlap): Use
variable view of container (Symmetric_Difference, Union):
Detect tampering (Vet): Use Checked_Index to detect tampering
(Constant_Reference, Element, Find): Use variable view of
container.
(Update_Element_Preserving_Key): Detect tampering.
* a-cbhase.adb (Difference, Find, Is_In): Use variable view
of container.
(Is_Subset): Ditto.
(Equivalent_Sets, Overlap): Use Node's Next component.
(Vet): Use Checked_Index to detect tampering.
(Constant_Reference, Element, Find): Use variable view of container.
(Update_Element_Preserving_Key): Detect tampering.
* a-cohama.adb, a-cihama.adb, a-cbhama.adb (Constant_Reference,
Element, Find): Use variable view of container.
(Reference): Rename hash table component.
(Vet): Use Checked_Index to detect tampering.

2014-02-19  Arnaud Charlet  <charlet@adacore.com>

* adabkend.adb (Scan_Compiler_Arguments): Add missing handling
of -nostdinc.

2014-02-19  Thomas Quinot  <quinot@adacore.com>

* tbuild.adb (New_Occurrence_Of, New_Rerefence_To): Guard
against calls without Def_Id.

2014-02-19  Claire Dross  <dross@adacore.com>

* a-cfdlli.ads, a-cfhase.ads, a-cforma.ads, a-cfhama.ads, a-cforse.ads,
a-cofove.ads: Add global annotations to subprograms.

2014-02-19  Hristian Kirtchev  <kirtchev@adacore.com>

* sem_prag.adb (Analyze_Initial_Condition_In_Decl_Part): Remove
constants Errors, Pack_Id and Pack_Init. Remove variable Vars.
Initial_Condition no longer requires the presence of pragma
Initialized. Do not try to diagnose whether all variables mentioned in
pragma Initializes also appear in Initial_Condition.
(Collect_Variables): Removed.
(Match_Variable): Removed.
(Match_Variables): Removed.
(Report_Unused_Variables): Removed.

2014-02-19  Thomas Quinot  <quinot@adacore.com>

* gnat_rm.texi (pragma Stream_Convert): Minor rewording.

From-SVN: r207905

10 years ago[multiple changes]
Arnaud Charlet [Wed, 19 Feb 2014 14:48:32 +0000 (15:48 +0100)]
[multiple changes]

2014-02-19  Robert Dewar  <dewar@adacore.com>

* sem_util.adb, sem_util.ads, prj-conf.adb, s-os_lib.adb: Minor
reformatting.

2014-02-19  Vincent Celier  <celier@adacore.com>

* prj-part.adb (Parse_Single_Project): Use the fully resolved
project path, with all symbolic links resolved, to check if the
same project is imported with a different unresolved path.
* prj-tree.ads (Project_Name_And_Node): Component Canonical_Path
changed to Resolved_Path to reflect that all symbolic links
are resolved.

From-SVN: r207904

10 years ago[multiple changes]
Arnaud Charlet [Wed, 19 Feb 2014 14:46:15 +0000 (15:46 +0100)]
[multiple changes]

2014-02-19  Ed Schonberg  <schonberg@adacore.com>

* sem_util.ads, sem_util.adb (Get_Cursor_Type): Moved to sem_util
from sem_ch13, for use elsewhere.
* sem_ch13.adb (Get_Cursor_Type): Moved to sem_util.
* sem_ch5.adb (Analyze_Iterator_Specification): Set properly the
cursor type on the loop variable when the iteration is over o
formal container.

2014-02-19  Vincent Celier  <celier@adacore.com>

* prj-conf.adb (Add_Default_GNAT_Naming_Scheme): Add declaration
for an empty Target (Check_Target): Never fail when an empty
target is declared in the configuration project.

2014-02-19  Ed Schonberg  <schonberg@adacore.com>

* sem_prag.adb (Check_Arg_Is_Local_Name): Argument is local if
the pragma comes fron a predicate aspect and the context is a
record declaration within the scope that declares the type.

2014-02-19  Robert Dewar  <dewar@adacore.com>

* gnat_rm.texi: Minor clarifications.
* expander.adb, sem_aggr.adb: Add comments.

From-SVN: r207903

10 years agoexpr.c (expand_expr_real_1): For a bit-field destination type...
Eric Botcazou [Wed, 19 Feb 2014 14:45:16 +0000 (14:45 +0000)]
expr.c (expand_expr_real_1): For a bit-field destination type...

* expr.c (expand_expr_real_1) <case VIEW_CONVERT_EXPR>: For a bit-field
destination type, extract only the valid bits if the source type is not
integral and has a different mode.

From-SVN: r207902

10 years ago[multiple changes]
Arnaud Charlet [Wed, 19 Feb 2014 14:42:55 +0000 (15:42 +0100)]
[multiple changes]

2014-02-19  Ed Schonberg  <schonberg@adacore.com>

* sem_prag.adb (Check_Arg_Is_Local_Name): For an aspect that
applies to a subprogram body, the name is the current scope,
rather than being declared in the current scope.
(Analyze_Pragma, case No_Return): Handle properly a No_Return
aspect applied to a subprogram body.

2014-02-19  Ed Schonberg  <schonberg@adacore.com>

* sem_ch13.adb (Resolve_Iterable_Operation): Improve resolution
of overloaded subprograms, and fix bug in handling of primitive
operation Element.

2014-02-19  Pascal Obry  <obry@adacore.com>

* s-os_lib.adb: Minor reformatting.

2014-02-19  Yannick Moy  <moy@adacore.com>

* expander.adb (Expand): Do nothing inside generics.
* sem_aggr.adb (Aggregate_Constraint_Checks): Do nothing inside
generics.

2014-02-19  Yannick Moy  <moy@adacore.com>

* exp_ch2.adb: Remove useless 'with' of unit Uintp.

From-SVN: r207901

10 years ago[multiple changes]
Arnaud Charlet [Wed, 19 Feb 2014 14:40:26 +0000 (15:40 +0100)]
[multiple changes]

2014-02-19  Robert Dewar  <dewar@adacore.com>

* sem_ch3.adb, style.adb, sem_prag.adb, sem_ch13.adb: Minor reformatting
* a-sequio.adb: Remove useless pragma Extensions_Allowed (no
longer needed).

2014-02-19  Claire Dross  <dross@adacore.com>

* a-cfdlli.ads, a-cfhase.ads, a-cforma.ads, a-cfhama.ads,
a-cforse.ads, a-cofove.ads: Add Iterable aspect to formal containers.

From-SVN: r207900

10 years agore PR ipa/60243 (IPA is slow on large cgraph tree)
Richard Biener [Wed, 19 Feb 2014 14:25:47 +0000 (14:25 +0000)]
re PR ipa/60243 (IPA is slow on large cgraph tree)

2014-02-19  Richard Biener  <rguenther@suse.de>

PR ipa/60243
* tree-inline.c (estimate_num_insns): Avoid calling cgraph_get_node
for all calls.

From-SVN: r207899

10 years agoXFAIL gcc.dg/torture/pr60092.c execution on Solaris 11
Rainer Orth [Wed, 19 Feb 2014 12:28:15 +0000 (12:28 +0000)]
XFAIL gcc.dg/torture/pr60092.c execution on Solaris 11

* gcc.dg/torture/pr60092.c: xfail execution on *-*-solaris2.11* at -O0.

From-SVN: r207898

10 years agore PR fortran/60232 ([OOP] The rank of the element in the structure constructor does...
Janus Weil [Wed, 19 Feb 2014 11:52:39 +0000 (12:52 +0100)]
re PR fortran/60232 ([OOP] The rank of the element in the structure constructor does not match that of the component)

2014-02-19  Janus Weil  <janus@gcc.gnu.org>

PR fortran/60232
* expr.c (gfc_get_variable_expr): Don't add REF_ARRAY for dimensionful
functions, which are used as procedure pointer target.

2014-02-19  Janus Weil  <janus@gcc.gnu.org>

PR fortran/60232
* gfortran.dg/typebound_proc_33.f90: New.

From-SVN: r207896

10 years ago[multiple changes]
Arnaud Charlet [Wed, 19 Feb 2014 11:18:32 +0000 (12:18 +0100)]
[multiple changes]

2014-02-19  Hristian Kirtchev  <kirtchev@adacore.com>

* sem_ch3.adb (Analyze_Declarations): Analyze
a package contract at the end of the private declarations (if
applicable), otherwise analyze it and the end of the visible
declarations.

2014-02-19  Ed Schonberg  <schonberg@adacore.com>

* style.adb (Missing_Overriding): If subprogram is an
instantiation, place warning on the instance node itself,
without mention of the original generic. Do not emit message
if explicit Ada version is older than the introduction of the
overriding indicator.

From-SVN: r207895

10 years ago[multiple changes]
Arnaud Charlet [Wed, 19 Feb 2014 11:17:15 +0000 (12:17 +0100)]
[multiple changes]

2014-02-19  Yannick Moy  <moy@adacore.com>

* gnat_rm.texi: Doc clarifications.

2014-02-19  Yannick Moy  <moy@adacore.com>

* exp_util.adb (Remove_Side_Effects): Do not remove side-effects
inside a generic.

2014-02-19  Ed Schonberg  <schonberg@adacore.com>

* sem_ch13.adb (Get_Cursor_Type): Obtain cursor type from
specified First primitive, rather than by name.
(Validate_Iterable_Aspect, Resolve_Iterable_Operation): Use it,
and extend error checking for missing primitives and incorrect
signatures.

2014-02-19  Ed Schonberg  <schonberg@adacore.com>

* sem_ch3.adb (Check_Pragma_Implemented): Detect additional
errors when a Synchronization aspect on an overriding protected
operation does not match the given aspect on the overridden
operation of an ancestor interface.

2014-02-19  Hristian Kirtchev  <kirtchev@adacore.com>

* sem_prag.adb (Check_Loop_Pragma_Grouping): New routine.
(Check_Loop_Pragma_Placement): Update
comment on usage. Remove local variables Orig_Stmt and
Within_Same_Sequence. Check that the current Loop_Invariant or
Loop_Variant pragma is grouped together with other such pragmas.
(Is_Loop_Pragma): New routine.
(Prev_In_Loop): Removed.

From-SVN: r207894

10 years ago[multiple changes]
Arnaud Charlet [Wed, 19 Feb 2014 11:12:05 +0000 (12:12 +0100)]
[multiple changes]

2014-02-19  Robert Dewar  <dewar@adacore.com>

* par-ch6.adb (P_Return): For extended return, end column lines
up with RETURN.
* par.adb: Minor documentation clarification.

2014-02-19  Yannick Moy  <moy@adacore.com>

* sem_prag.adb (Check_Loop_Pragma_Placement): Add check
that Loop_Invariant and Loop_Variant appear consecutively.
* gnat_rm.texi Update documentation of Loop_Invariant and
Loop_Variant pragmas.

2014-02-19  Robert Dewar  <dewar@adacore.com>

* debug.adb: Document -gnatd.X.
* par-ch5.adb (P_If_Statement): Always check THEN, even if not
first token
(Check_Then_Column): Ditto.
* styleg.adb (Check_Then): Allow THEN on line after IF.
(Check_Then): Check THEN placement under control of -gnatd.X
* styleg.ads (Check_Then): Now called even if THEN is not first
token on line.
* stylesw.ads (Style_Check_If_Then_Layout): Document new
relaxed rules.
* gnat_ugn.texi: For -gnatyi, THEN can now be on line after IF.

2014-02-19  Robert Dewar  <dewar@adacore.com>

* a-cfhama.adb, a-cfhase.adb, a-cforse.adb, a-cofove.adb, a-ngcefu.adb,
a-teioed.adb, a-wtedit.adb, a-ztedit.adb, exp_ch5.adb, inline.adb,
prj-pp.adb, prj-tree.adb, sem_ch12.adb, sem_ch8.adb,
vms_conv.adb: Fix bad layout of IF statements

From-SVN: r207893

10 years agoexp_util.adb (Side_Effect_Free): Scalar if expressions can be SEF.
Robert Dewar [Wed, 19 Feb 2014 11:07:31 +0000 (11:07 +0000)]
exp_util.adb (Side_Effect_Free): Scalar if expressions can be SEF.

2014-02-19  Robert Dewar  <dewar@adacore.com>

* exp_util.adb (Side_Effect_Free): Scalar if expressions can be SEF.

From-SVN: r207892

10 years ago[multiple changes]
Arnaud Charlet [Wed, 19 Feb 2014 11:05:35 +0000 (12:05 +0100)]
[multiple changes]

2014-02-19  Robert Dewar  <dewar@adacore.com>

* exp_util.adb: Update comments.

2014-02-19  Doug Rupp  <rupp@adacore.com>

* bindgen.adb (Gen_Adainit) [VMS] New global Float_Format.
* init.c (__gl_float_format): [VMS] New global.
(__gnat_set_features): Call FP_CONTROL to set FPSR for the float
representation in effect.

2014-02-19  Hristian Kirtchev  <kirtchev@adacore.com>

* exp_ch6.adb Add with and use clause for Exp_Prag.
(Expand_Contract_Cases): Relocated to Exp_Prag.
* exp_ch6.ads (Expand_Contract_Cases): Relocated to Exp_Prag.
* exp_prag.adb Add with and use clauses for Checks and Validsw.
(Expand_Contract_Cases): Relocated from Exp_Ch6. Update the
structure of the expanded code to showcase the evaluation of
attribute 'Old prefixes. Add local variable Old_Evals. Expand
any attribute 'Old references found within a consequence. Add
circuitry to evaluate the prefixes of attribute 'Old that
belong to a selected consequence.
(Expand_Old_In_Consequence): New routine.
* exp_prag.ads (Expand_Contract_Cases): Relocated from Exp_Ch6.
* sem_attr.adb (Check_Use_In_Contract_Cases): Warn that a
potentially unevaluated prefix is always evaluated.

From-SVN: r207891

10 years ago[multiple changes]
Arnaud Charlet [Wed, 19 Feb 2014 11:02:48 +0000 (12:02 +0100)]
[multiple changes]

2014-02-19  Robert Dewar  <dewar@adacore.com>

* exp_attr.adb (Expand_Min_Max_Attribute): Use Insert_Declaration
(Expand_Min_Max_Attribute): Use Matching_Standard_Type.
* exp_ch4.adb (Expand_N_Expression_With_Actions): Remove special
handling for the case of Modify_Tree_For_C, this approach did
not work.
* exp_util.adb (Matching_Standard_Type): New function
(Side_Effect_Free): New top level functions (from
Remove_Side_Effects).
* exp_util.ads (Side_Effect_Free): New top level functions
(moved from body).
* sinfo.ads: Minor comment updates.

2014-02-19  Ed Schonberg  <schonberg@adacore.com>

* exp_ch6.adb (Expand_Simple_Function_Return): If return
type is unconstrained and uses the secondary stack, mark the
enclosing function accordingly, to ensure that the value is not
prematurely removed.

2014-02-19  Hristian Kirtchev  <kirtchev@adacore.com>

* par.adb Alphabetize the routines in Par.Sync.
(Resync_Past_Malformed_Aspect): New routine.
* par-ch13.adb (Get_Aspect_Specifications): Alphabetize local
variables. Code and comment reformatting. Detect missing
parentheses on aspects [Refined_]Global and [Refined_]Depends
with a non-null definition.
* par-sync.adb: Alphabetize all routines in this separate unit.
(Resync_Past_Malformed_Aspect): New routine.

From-SVN: r207890

10 years agosem_eval.ads, [...] (Subtypes_Statically_Match): Return False if Esize values do...
Robert Dewar [Wed, 19 Feb 2014 11:01:01 +0000 (11:01 +0000)]
sem_eval.ads, [...] (Subtypes_Statically_Match): Return False if Esize values do not match.

2014-02-19  Robert Dewar  <dewar@adacore.com>

* sem_eval.ads, sem_eval.adb (Subtypes_Statically_Match): Return False
if Esize values do not match.

From-SVN: r207889

10 years ago[multiple changes]
Arnaud Charlet [Wed, 19 Feb 2014 10:59:25 +0000 (11:59 +0100)]
[multiple changes]

2014-02-19  Yannick Moy  <moy@adacore.com>

* sinfo.ads: Minor comment update.

2014-02-19  Hristian Kirtchev  <kirtchev@adacore.com>

* atree.ads: Minor reformatting (change Entity_Info to Einfo).

2014-02-19  Thomas Quinot  <quinot@adacore.com>

* exp_ch7.adb (Find_Node_To_Be_Wrapped): An assignment statement
that has the No_Ctrl_Actions flag is a suitable node to be
wrapped if the assigned expression has no finalization actions.
* sem_eval.adb (Eval_Entity_Name): For a compile time known
boolean value, mark the corresponding condition SCO as constant.

2014-02-19  Robert Dewar  <dewar@adacore.com>

* exp_util.adb: Minor reformatting.
* exp_util.ads (Matching_Standard_Type): New function.
* exp_ch7.adb: Minor reformatting.

2014-02-19  Ed Schonberg  <schonberg@adacore.com>

* sem_ch13.adb (Get_Cursor_Type): Use scope of iterable type
to find declaration for Cursor, to handle properly the case of
a discriminated iterable type.

2014-02-19  Vincent Celier  <celier@adacore.com>

* gnatcmd.adb (GNATCmd): Always replace the object dirs of
imported library projects with the library ALI dirs, when setting
the object paths.
* prj-env.ads (Ada_Objects_Path): Correct comments about
argument Including_Libraries.

From-SVN: r207888

10 years ago[multiple changes]
Arnaud Charlet [Wed, 19 Feb 2014 10:55:33 +0000 (11:55 +0100)]
[multiple changes]

2014-02-19  Gary Dismukes  <dismukes@adacore.com>

* gnat_rm.texi: Minor spelling fixes.

2014-02-19  Doug Rupp  <rupp@adacore.com>

* init.c: Remove unneeded code.
* fe.h (Float_Format): New macro
* gcc-interface/trans.c (gigi): On VMS, set vms_float_format.

From-SVN: r207887

10 years ago[multiple changes]
Arnaud Charlet [Wed, 19 Feb 2014 10:51:26 +0000 (11:51 +0100)]
[multiple changes]

2014-02-19  Hristian Kirtchev  <kirtchev@adacore.com>

* sem_prag.adb (Check_Refined_Global_Item):
A state or variable acts as a constituent only it is part of an
encapsulating state and the state has visible refinement.

2014-02-19  Ed Schonberg  <schonberg@adacore.com>

* sem_ch6.adb (Analyze_Subprogram_Contract): Do not warn on a
postcondition for a function when the expression does not mention
'Result but the function has in-out parameters.

2014-02-19  Robert Dewar  <dewar@adacore.com>

* gnat_rm.texi: Add documentation on Value_Size forcing biased
representation.

From-SVN: r207886

10 years ago[multiple changes]
Arnaud Charlet [Wed, 19 Feb 2014 10:48:06 +0000 (11:48 +0100)]
[multiple changes]

2014-02-19  Hristian Kirtchev  <kirtchev@adacore.com>

* lib-xref.ads Remove the small table of letter and symbol usage as we
already have one.

2014-02-19  Hristian Kirtchev  <kirtchev@adacore.com>

* sem_ch6.adb (Analyze_Subprogram_Body_Helper): Emit specific error
messages depending on the offending misplaced aspect specifications.
(Diagnose_Misplaced_Aspect_Specifications): New routine.

2014-02-19  Ed Schonberg  <schonberg@adacore.com>

* sem_ch13.adb (Get_Cursor): Utility to retrieve cursor type
for iterable aspect primitives.
(Resolve_Iterable_Operation): Use expected signature of iterable
aspect to resolve primitive when overloading is present.
(Validate_Iterable_Aspect, Analyze_Aspects_At_Freeze_Point): use it.
(Check_Signature): Removed.

From-SVN: r207885

10 years ago[multiple changes]
Arnaud Charlet [Wed, 19 Feb 2014 10:44:33 +0000 (11:44 +0100)]
[multiple changes]

2014-02-19  Yannick Moy  <moy@adacore.com>

* sem_ch10.adb (Analyze_Proper_Body): Issue error on missing
subunit in GNATprove_Mode.
* sinfo.ads (GNATprove_Mode): Document error issued in GNATprove_Mode.

2014-02-19  Hristian Kirtchev  <kirtchev@adacore.com>

* lib-xref.ads Alphabetize the contents of table
Xref_Entity_Letters. Add an entry in table Xref_Entity_Letters
for E_Abstract_State. List all letters and symbols in use.
* sem_prag.adb (Analyze_Abstract_State): Update all calls
to Create_Abstract_State to reflect the new signature.
(Create_Abstract_State): Change subprogram profile and update
the comment on usage. Use the proper location of the state
declaration when creating a new abstract state entity. Do not
generate an external name, but simply reuse the name coming from
the state declaration.

From-SVN: r207884

10 years agoexp_ch4.adb (Expand_N_Expression_With_Actions): Make sure declarations get properly...
Robert Dewar [Wed, 19 Feb 2014 10:42:16 +0000 (10:42 +0000)]
exp_ch4.adb (Expand_N_Expression_With_Actions): Make sure declarations get properly inserted in Modify_Tree_For_C mode.

2014-02-19  Robert Dewar  <dewar@adacore.com>

* exp_ch4.adb (Expand_N_Expression_With_Actions): Make sure
declarations get properly inserted in Modify_Tree_For_C mode.
* sinfo.ads: Minor comment addition.

From-SVN: r207883

10 years agopar-ch9.adb, [...]: Minor reformatting.
Robert Dewar [Wed, 19 Feb 2014 10:32:17 +0000 (10:32 +0000)]
par-ch9.adb, [...]: Minor reformatting.

2014-02-19  Robert Dewar  <dewar@adacore.com>

* par-ch9.adb, exp_ch5.adb, sem_ch5.adb, exp_attr.adb, sem_util.adb,
sem_util.ads, sem_ch13.adb, sem_ch13.ads: Minor reformatting.

From-SVN: r207882

10 years agostyle.adb (Missing_Overriding): Warning does not apply in language versions prior...
Ed Schonberg [Wed, 19 Feb 2014 10:30:33 +0000 (10:30 +0000)]
style.adb (Missing_Overriding): Warning does not apply in language versions prior to Ada 2005.

2014-02-19  Ed Schonberg  <schonberg@adacore.com>

* style.adb (Missing_Overriding): Warning does not apply in
language versions prior to Ada 2005.
* snames.ads-tmpl: Add Name_Iterable and Attribute_Iterable.
* sem_attr.adb: Add Attribute_Iterable where needed.
* exp_attr.adb: ditto.
* exp_ch5.adb (Expand_Formal_Container_Loop): New procedure to
handle loops and quantified expressions over types that have an
iterable aspect.  Called from Expand_Iterator_Loop.
* sem_ch5.adb (Analyze_Iterator_Specification): Recognize types
with Iterable aspect.
* sem_ch13.adb (Validate_Iterable_Aspect): Verify that the
subprograms specified in the Iterable aspect have the proper
signature involving container and cursor.
(Check_Aspect_At_Freeze_Point): Analyze value of iterable aspect.
* sem_ch13.ads (Validate_Iterable_Aspect): New subprogram.
* sem_util.ads, sem_util.adb (Get_Iterable_Type_Primitive):
New procedure to retrieve one of the primitives First, Last,
or Has_Element, from the value of the iterable aspect of a
formal container.
(Is_Container_Element): Predicate to recognize expressions
that denote an element of one of the predefined containers,
for possible optimization.  This subprogram is not currently
used, pending ARG discussions on the legality of the proposed
optimization. Worth preserving for eventual use.
(Is_Iterator): Recognize formal container types.
* aspects.ads, aspects.adb: Add Aspect_Iterable where needed.

From-SVN: r207881

10 years ago[multiple changes]
Arnaud Charlet [Wed, 19 Feb 2014 10:25:53 +0000 (11:25 +0100)]
[multiple changes]

2014-02-19  Robert Dewar  <dewar@adacore.com>

* exp_attr.adb (Expand_Min_Max_Attribute): New procedure
(Expand_N_Attribute_Reference): Use this procedure for Min and Max.
* exp_ch4.adb (Expand_N_Expression_With_Actions): Remove object
declarations from list of actions.
* output.ads, output.adb (Delete_Last_Char): New procedure.
* sinfo.ads: Document handling of Mod and expression with actions
in Modify_Tree_For_C mode.

2014-02-19  Ed Schonberg  <schonberg@adacore.com>

* par-ch9.adb (P_Task): Add a null statement to produce a
well-formed task body when due to a previous syntax error the
statement list is empty.

2014-02-19  Hristian Kirtchev  <kirtchev@adacore.com>

* sem_prag.adb (Check_Dependency_Clause): Account
for the case where a state with a non-null refinement matches a
null output list. Comment reformatting.
(Inputs_Match): Copy a solitary input to avoid an assertion failure
when trying to match the same input in multiple clauses.

2014-02-19  Gary Dismukes  <dismukes@adacore.com>

* sem_attr.adb: Minor typo fix.

From-SVN: r207880

10 years agore PR ipa/60243 (IPA is slow on large cgraph tree)
Richard Biener [Wed, 19 Feb 2014 09:29:34 +0000 (09:29 +0000)]
re PR ipa/60243 (IPA is slow on large cgraph tree)

2014-02-19  Richard Biener  <rguenther@suse.de>

PR ipa/60243
* ipa-prop.c: Include stringpool.h and tree-ssanames.h.
(ipa_modify_call_arguments): Emit an argument load explicitely and
preserve virtual SSA form there and for the replacement call.
Do not update SSA form nor free dominance info.

From-SVN: r207879

10 years agoipa.c (function_and_variable_visibility): Also clear WEAK flag when disolving COMDAT_...
Jan Hubicka [Wed, 19 Feb 2014 07:45:00 +0000 (08:45 +0100)]
ipa.c (function_and_variable_visibility): Also clear WEAK flag when disolving COMDAT_GROUP.

* ipa.c (function_and_variable_visibility): Also clear WEAK
flag when disolving COMDAT_GROUP.

From-SVN: r207878

10 years agoipa-prop.h (ipa_ancestor_jf_data): Update ocmment.
Jan Hubicka [Wed, 19 Feb 2014 07:42:06 +0000 (08:42 +0100)]
ipa-prop.h (ipa_ancestor_jf_data): Update ocmment.

* ipa-prop.h (ipa_ancestor_jf_data): Update ocmment.
* ipa-prop.c (ipa_set_jf_known_type): Return early when
not devirtualizing.
(ipa_set_ancestor_jf): Set type to NULL hwen it is not preserved;
do more sanity checks.
(detect_type_change): Return true when giving up early.
(compute_complex_assign_jump_func): Fix type parameter of
ipa_set_ancestor_jf.
(compute_complex_ancestor_jump_func): Likewise.
(update_jump_functions_after_inlining): Fix updating of
ancestor function.
* ipa-cp.c (ipa_get_jf_ancestor_result): Be ready for type
to be NULL.

From-SVN: r207877

10 years agocgraph.c (cgraph_update_edges_for_call_stmt_node): Also remove inline clones when...
Jan Hubicka [Wed, 19 Feb 2014 07:33:45 +0000 (08:33 +0100)]
cgraph.c (cgraph_update_edges_for_call_stmt_node): Also remove inline clones when edge disappears.

* cgraph.c (cgraph_update_edges_for_call_stmt_node): Also remove
inline clones when edge disappears.

From-SVN: r207875

10 years agore PR c/60195 (Strange warnings using atomic types)
Marek Polacek [Wed, 19 Feb 2014 06:29:49 +0000 (06:29 +0000)]
re PR c/60195 (Strange warnings using atomic types)

PR c/60195
c/
* c-typeck.c (convert_lvalue_to_rvalue): Set TREE_NO_WARNING on tmp.
Call mark_exp_read on exp.value.
(build_atomic_assign): Set TREE_NO_WARNING on val and old.  Set
TREE_ADDRESSABLE on old instead of val.
(emit_side_effect_warnings): Warn only if RHS has !TREE_NO_WARNING.
testsuite/
* gcc.dg/pr60195.c: New test.

From-SVN: r207873

10 years agono-vfa-vect-depend-2.c (main1): Fix buffer overflow.
Paul Pluzhnikov [Wed, 19 Feb 2014 06:21:42 +0000 (22:21 -0800)]
no-vfa-vect-depend-2.c (main1): Fix buffer overflow.

2014-02-19  Paul Pluzhnikov  <ppluzhnikov@google.com>

* gcc.dg/vect/no-vfa-vect-depend-2.c (main1): Fix buffer
overflow.

From-SVN: r207872

10 years agore PR preprocessor/58844 (ICE with invalid use of ##)
Jakub Jelinek [Wed, 19 Feb 2014 06:05:55 +0000 (07:05 +0100)]
re PR preprocessor/58844 (ICE with invalid use of ##)

PR preprocessor/58844
* macro.c (enter_macro_context): Only push
macro_real_token_count (macro) tokens rather than
macro->count tokens, regardless of
CPP_OPTION (pfile, track-macro-expansion).

* c-c++-common/cpp/pr58844-1.c: New test.
* c-c++-common/cpp/pr58844-2.c: New test.

From-SVN: r207871

10 years ago+ * include/cilk/common.h (__CILKRTS_ABI_VERSION): Set the ABI version + to 1 instead...
Balaji V. Iyer [Wed, 19 Feb 2014 03:43:21 +0000 (03:43 +0000)]
+ * include/cilk/common.h (__CILKRTS_ABI_VERSION): Set the ABI version + to 1 instead of 0.

2014-02-18  Balaji V. Iyer  <balaji.v.iyer@intel.com>
+
+ * include/cilk/common.h (__CILKRTS_ABI_VERSION): Set the ABI version
+ to 1 instead of 0.
+ * runtime/cilk-abi-cilk-for.cpp (cilk_for_recursive): Added a fix to
+ initialize stack frame correctly.
+

From-SVN: r207870

10 years agore PR target/60203 (Support long double/_Decimal128 direct move on power8)
Michael Meissner [Wed, 19 Feb 2014 02:14:17 +0000 (02:14 +0000)]
re PR target/60203 (Support long double/_Decimal128 direct move on power8)

2014-02-18  Michael Meissner  <meissner@linux.vnet.ibm.com>

PR target/60203
* config/rs6000/rs6000.md (mov<mode>_64bit, TF/TDmode moves):
Split 64-bit moves into 2 patterns.  Do not allow the use of
direct move for TDmode in little endian, since the decimal value
has little endian bytes within a word, but the 64-bit pieces are
ordered in a big endian fashion, and normal subreg's of TDmode are
not allowed.
(mov<mode>_64bit_dm): Likewise.
(movtd_64bit_nodm): Likewise.

From-SVN: r207868