gcc.git
10 years agodecl.c (gnat_to_gnu_entity): Always build a variable for E_Variable with a pragma...
Tristan Gingold [Sat, 25 Jan 2014 10:42:04 +0000 (10:42 +0000)]
decl.c (gnat_to_gnu_entity): Always build a variable for E_Variable with a pragma Linker_Section.

* gcc-interface/decl.c (gnat_to_gnu_entity): Always build a variable
for E_Variable with a pragma Linker_Section.

From-SVN: r207072

10 years agodecl.c (gnat_to_gnu_param): Make sure an Out parameter with Default_Value aspect...
Robert Dewar [Sat, 25 Jan 2014 10:34:40 +0000 (10:34 +0000)]
decl.c (gnat_to_gnu_param): Make sure an Out parameter with Default_Value aspect is passed in by copy.

* gcc-interface/decl.c (gnat_to_gnu_param): Make sure an Out parameter
with Default_Value aspect is passed in by copy.

From-SVN: r207071

10 years agoDaily bump.
GCC Administrator [Sat, 25 Jan 2014 00:17:03 +0000 (00:17 +0000)]
Daily bump.

From-SVN: r207069

10 years agoprint-rtl.c (in_call_function_usage): New var.
Jakub Jelinek [Fri, 24 Jan 2014 23:18:26 +0000 (00:18 +0100)]
print-rtl.c (in_call_function_usage): New var.

* print-rtl.c (in_call_function_usage): New var.
(print_rtx): When in CALL_INSN_FUNCTION_USAGE, always print
EXPR_LIST mode as mode and not as reg note name.

From-SVN: r207066

10 years agore PR middle-end/59561 (warning: iteration 4 invokes undefined behavior)
Jakub Jelinek [Fri, 24 Jan 2014 23:17:25 +0000 (00:17 +0100)]
re PR middle-end/59561 (warning: iteration 4 invokes undefined behavior)

PR middle-end/59561
* cfgloopmanip.c (copy_loop_info): If
loop->warned_aggressive_loop_optimizations, make sure
the flag is set in target loop too.

From-SVN: r207065

10 years agoReplace flag_enable_cilkplus with flag_cilkplus.
Balaji V. Iyer [Fri, 24 Jan 2014 23:13:49 +0000 (23:13 +0000)]
Replace flag_enable_cilkplus with flag_cilkplus.

gcc/ChangeLog
+2014-01-24  Balaji V. Iyer  <balaji.v.iyer@intel.com>
+
+       * builtins.c (is_builtin_name): Renamed flag_enable_cilkplus to
+       flag_cilkplus.
+       * builtins.def: Likewise.
+       * cilk.h (fn_contains_cilk_spawn_p): Likewise.
+       * cppbuiltin.c (define_builtin_macros_for_compilation_flags): Likewise.
+       * ira.c (ira_setup_eliminable_regset): Likewise.
+       * omp-low.c (gate_expand_omp): Likewise.
+       (execute_lower_omp): Likewise.
+       (diagnose_sb_0): Likewise.
+       (gate_diagnose_omp_blocks): Likewise.
+       (simd_clone_clauses_extract): Likewise.
+       (gate): Likewise.
+

gcc/c-family/ChangeLog
+2014-01-24  Balaji V. Iyer  <balaji.v.iyer@intel.com>
+
+       * c-common.c (c_define_builtins): Replaced flag_enable_cilkplus with
+       flag_cilkplus.
+       * c-pragma.c (init_pragma): Likewise.
+       * c.opt: Likewise.
+

gcc/c/ChangeLog
+2014-01-24  Balaji V. Iyer  <balaji.v.iyer@intel.com>
+
+       * c-parser.c (c_parser_declaration_or_fndef): Replaced
+       flag_enable_cilkplus with flag_cilkplus.
+       (c_parser_direct_declarator_inner): Likewise.
+       (c_parser_attribute_any_word): Likewise.
+       (c_parser_attributes): Likewise.
+       (c_parser_compound_statement): Likewise.
+       (c_parser_statement_after_labels): Likewise.
+       (c_parser_if_statement): Likewise.
+       (c_parser_switch_statement): Likewise.
+       (c_parser_do_statement): Likewise.
+       (c_parser_for_statement): Likewise.
+       (c_parser_unary_expression): Likewise.
+       (c_parser_postfix_expression): Likewise.
+       (c_parser_postfix_expression_after_primary): Likewise.
+       (c_parser_postfix_expression_after_primary): Likewise.
+       (c_parser_omp_clause_name): Likewise.
+       (c_finish_omp_declare_simd): Likewise.
+       (c_parser_cilk_verify_simd): Likewise.
+       * c-typeck.c (build_array_ref): Likewise.
+       (build_function_call_vec): Likewise.
+       (convert_arguments): Likewise.
+       (build_compound_expr): Likewise.
+       (c_finish_return): Likewise.
+       (c_finish_if_stmt): Likewise.
+       (c_finish_loop): Likewise.
+       (build_binary_op): Likewise.
+

gcc/lto/ChangeLog
+2014-01-24  Balaji V. Iyer  <balaji.v.iyer@intel.com>
+
+       * lto-lang.c (lto_init): Replaced flag_enable_cilkplus with
+       flag_cilkplus.
+

From-SVN: r207064

10 years agocompiler: Use backend interface for unary expressions.
Chris Manghane [Fri, 24 Jan 2014 22:49:06 +0000 (22:49 +0000)]
compiler: Use backend interface for unary expressions.

* go-gcc.cc (Gcc_backend::unary_expression): New function.

From-SVN: r207063

10 years agors6000.c (rs6000_expand_vec_perm_const_1): Remove correction for little endian...
Bill Schmidt [Fri, 24 Jan 2014 21:32:38 +0000 (21:32 +0000)]
rs6000.c (rs6000_expand_vec_perm_const_1): Remove correction for little endian...

2014-01-24  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>

* config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Remove
correction for little endian...
* config/rs6000/vsx.md (vsx_xxpermdi2_<mode>_1): ...and move it to
here.

From-SVN: r207062

10 years agore PR tree-optimization/59919 (ICE in process_assert_insertions_for, at tree-vrp...
Jeff Law [Fri, 24 Jan 2014 20:51:22 +0000 (13:51 -0700)]
re PR tree-optimization/59919 (ICE in process_assert_insertions_for, at tree-vrp.c:6096)

PR tree-optimization/59919
* tree-vrp.c (find_assert_locations_1): Do not register asserts
for non-returning calls.

PR tree-optimization/59919
* gcc.c-torture/compile/pr59919.c: New test.

From-SVN: r207061

10 years agore PR libstdc++/59531 (string_view overrun in copy operation)
Ed Smith-Rowland [Fri, 24 Jan 2014 20:15:00 +0000 (20:15 +0000)]
re PR libstdc++/59531 (string_view overrun in copy operation)

2014-01-24  Ed Smith-Rowland  <3dw4rd@verizon.net>

PR libstdc++/59531
* testsuite/experimental/string_view/operations/copy/char/1.cc: New.
* testsuite/experimental/string_view/operations/copy/wchar_t/1.cc: New.

2014-01-24  Ed Smith-Rowland  <3dw4rd@verizon.net>
    Peter A. Bigot <pab@pabigot.com>

PR libstdc++/59531
* include/experimental/string_view
(copy(_CharT*, size_type, size_type) const): Correct throw string.
Correct copy start location.

2014-01-24  Ed Smith-Rowland  <3dw4rd@verizon.net>
    Peter A. Bigot <pab@pabigot.com>

PR libstdc++/59530
* include/experimental/string_view (operator[](size_type) const):
Fix one-off index error in debug check.
* testsuite/experimental/string_view/element_access/char/1.cc: Don't
test basic_string_view at size().
* testsuite/experimental/string_view/element_access/wchar_t/1.cc: Ditto.

2014-01-24  Ed Smith-Rowland  <3dw4rd@verizon.net>
    Peter A. Bigot <pab@pabigot.com>

PR libstdc++/59529
* include/experimental/string_view
(basic_string_view(const _CharT*, size_type)): Don't care if len == 0.
* testsuite/experimental/string_view/operations/substr/char/1.cc:
Comment out catch of out_of_range; No terminating null
in basic_string_view.  Check begin == end.
* testsuite/experimental/string_view/operations/substr/wchar_t/1.cc:
Ditto.

Co-Authored-By: Peter A. Bigot <pab@pabigot.com>
From-SVN: r207060

10 years agore PR libstdc++/59548 (Abort after copying std::unordered_map in debug mode)
Jonathan Wakely [Fri, 24 Jan 2014 20:08:20 +0000 (20:08 +0000)]
re PR libstdc++/59548 (Abort after copying std::unordered_map in debug mode)

PR libstdc++/59548
* include/debug/safe_base.h (_Safe_sequence_base): Define copy
constructor to prevent it being implicitly defined as deleted, but
do not copy anything.
* include/debug/safe_unordered_base.h (_Safe_unordered_container_base):
Define copy and move constructors similar to _Safe_sequence_base's.
* testsuite/23_containers/unordered_map/59548.cc: New.

From-SVN: r207059

10 years agore PR c++/58550 (][c++11] ICE with auto in function return type and lto)
Jason Merrill [Fri, 24 Jan 2014 19:05:39 +0000 (14:05 -0500)]
re PR c++/58550 (][c++11] ICE with auto in function return type and lto)

PR c++/58550
* decl.c (grokdeclarator): Turn pedwarn about auto return type in
c++11 into error.

From-SVN: r207055

10 years agotypo
Jason Merrill [Fri, 24 Jan 2014 19:05:32 +0000 (14:05 -0500)]
typo

From-SVN: r207054

10 years ago[AArch64] fix big.LITTLE spec rewriting
James Greenhalgh [Fri, 24 Jan 2014 17:14:35 +0000 (17:14 +0000)]
[AArch64] fix big.LITTLE spec rewriting

gcc/

* common/config/aarch64/aarch64-common.c
(aarch64_rewrite_mcpu): Handle multiple names.
* config/aarch64/aarch64.h
(BIG_LITTLE_SPEC): Do not discard mcpu switches.

From-SVN: r207053

10 years agore PR c++/59886 (C++ array init optimization results in RANGE_EXPRs in assignments)
Jason Merrill [Fri, 24 Jan 2014 17:09:07 +0000 (12:09 -0500)]
re PR c++/59886 (C++ array init optimization results in RANGE_EXPRs in assignments)

PR c++/59886
PR c++/59659
* g++.dg/opt/value-init2.C: Remove.

From-SVN: r207052

10 years agore PR c++/59886 (C++ array init optimization results in RANGE_EXPRs in assignments)
Jason Merrill [Fri, 24 Jan 2014 16:47:54 +0000 (11:47 -0500)]
re PR c++/59886 (C++ array init optimization results in RANGE_EXPRs in assignments)

PR c++/59886
PR c++/59659
* typeck2.c (process_init_constructor_array): Don't create
RANGE_EXPR yet.

From-SVN: r207051

10 years agotypeck2.c (split_nonconstant_init_1): Fix num_split_elts handling for RANGE_ARRAY...
Jakub Jelinek [Fri, 24 Jan 2014 16:47:31 +0000 (17:47 +0100)]
typeck2.c (split_nonconstant_init_1): Fix num_split_elts handling for RANGE_ARRAY case.

* typeck2.c (split_nonconstant_init_1): Fix num_split_elts
handling for RANGE_ARRAY case.

From-SVN: r207050

10 years ago[multiple changes]
Arnaud Charlet [Fri, 24 Jan 2014 16:22:02 +0000 (17:22 +0100)]
[multiple changes]

2014-01-24  Eric Botcazou  <ebotcazou@adacore.com>

* set_targ.adb: Set Short_Enums.
* gcc-interface/lang.opt (fshort-enums): New option.
* gcc-interface/misc.c (gnat_handle_option): Handle it.
(gnat_post_options): Do not modify the global settings.

2014-01-24  Robert Dewar  <dewar@adacore.com>

* g-rannum.ads, g-rannum.adb (Random_Ordinary_Fixed): New generic
function.
(Random_Decimal_Fixed): New generic function.
* s-rannum.ads: Minor comment clarifications.

From-SVN: r207049

10 years agore PR c++/57524 (ICE: in timevar_start, at timevar.c:343)
Paolo Carlini [Fri, 24 Jan 2014 15:45:14 +0000 (15:45 +0000)]
re PR c++/57524 (ICE: in timevar_start, at timevar.c:343)

/cp
2014-01-24  Paolo Carlini  <paolo.carlini@oracle.com>

PR c++/57524
* name-lookup.c (push_using_directive): Use timevar_cond_start.

/testsuite
2014-01-24  Paolo Carlini  <paolo.carlini@oracle.com>

PR c++/57524
* g++.dg/ext/timevar2.C: New.

From-SVN: r207047

10 years agoAvoid crashing when an input file could not be opened
Dodji Seketeli [Fri, 24 Jan 2014 15:34:03 +0000 (15:34 +0000)]
Avoid crashing when an input file could not be opened

* input.c (add_file_to_cache_tab): Handle the case where fopen
returns NULL.

Signed-off-by: Dodji Seketeli <dodji@seketeli.org>
From-SVN: r207046

10 years agoback_end.adb: Remove Short_Enums handling (handled in Ttypes/Get_Targ now) Minor...
Robert Dewar [Fri, 24 Jan 2014 15:18:46 +0000 (15:18 +0000)]
back_end.adb: Remove Short_Enums handling (handled in Ttypes/Get_Targ now) Minor added comments.

2014-01-24  Robert Dewar  <dewar@adacore.com>

* back_end.adb: Remove Short_Enums handling (handled in
Ttypes/Get_Targ now) Minor added comments.
* freeze.adb: Change name Short_Enums_On_Target to
Target_Short_Enums.
* get_targ.ads, get_targ.adb (Get_Short_Enums): New function.
* opt.ads: Minor comment updates.
* sem_ch13.adb: Change name Short_Enums_On_Target to
Target_Short_Enums.
* set_targ.adb: Set Short_Enums from gcc back end.
* set_targ.ads (Short_Enums): New variable.
* targparm.ads, targparm.adb: Remove Short_Enums entries (handled in
Ttypes/Get_Targ now).
* ttypes.ads (Target_Short_Enums): New constant boolean switch

From-SVN: r207045

10 years agog-sercom-mingw.adb: Fix serial port name for port number > 10.
Pascal Obry [Fri, 24 Jan 2014 15:16:49 +0000 (15:16 +0000)]
g-sercom-mingw.adb: Fix serial port name for port number > 10.

2014-01-24  Pascal Obry  <obry@adacore.com>

* g-sercom-mingw.adb: Fix serial port name for port number > 10.

From-SVN: r207044

10 years agoexp_disp.adb (Expand_Dispatching_Call): Call Unqualify on Param when comparing it...
Gary Dismukes [Fri, 24 Jan 2014 15:15:04 +0000 (15:15 +0000)]
exp_disp.adb (Expand_Dispatching_Call): Call Unqualify on Param when comparing it with Ctrl_Arg...

2014-01-24  Gary Dismukes  <dismukes@adacore.com>

* exp_disp.adb (Expand_Dispatching_Call): Call Unqualify on Param when
comparing it with Ctrl_Arg, since Ctrl_Arg may have had qualification
stripped off.

From-SVN: r207043

10 years ago[multiple changes]
Arnaud Charlet [Fri, 24 Jan 2014 15:02:48 +0000 (16:02 +0100)]
[multiple changes]

2014-01-24  Robert Dewar  <dewar@adacore.com>

* sinfo.ads, make.adb, prj-env.adb: Minor reformatting.

2014-01-24  Vincent Celier  <celier@adacore.com>

* prj.adb (Add_Aggregated_Project): Do not add a project in
the list if it is already there.

2014-01-24  Yannick Moy  <moy@adacore.com>

* lib-xref-spark_specific.adb (Enclosing_Subprogram_Or_Package):
Correct the search for a subrogram declaration to which a pragma is
attached.

2014-01-24  Bob Duff  <duff@adacore.com>

* gnat_ugn.texi: Document --decimal-grouping and
--based-grouping switches in gnatpp.

From-SVN: r207042

10 years agosinfo.ads: Documentation update.
Arnaud Charlet [Fri, 24 Jan 2014 15:00:26 +0000 (16:00 +0100)]
sinfo.ads: Documentation update.

2014-01-24  Ed Schonberg  <schonberg@adacore.com>

* sinfo.ads: Documentation update.

2014-01-24  Ed Schonberg  <schonberg@adacore.com>

* sem_ch3.adb (Constant_Redeclaration): New declaration is
illegal if previous one has an initial expression that is an
aggregate expanded into assignments.

2014-01-24  Ed Schonberg  <schonberg@adacore.com>

* sem_ch5.adb (Analyze_Loop_Parameter_Specification): Small
code reorganization to remove spurious warning on a loop with
an array element iterator that has a null range.

From-SVN: r207041

10 years agoprj.adb, [...]: Add comment, minor code clean ups.
Robert Dewar [Fri, 24 Jan 2014 14:53:30 +0000 (14:53 +0000)]
prj.adb, [...]: Add comment, minor code clean ups.

2014-01-24  Robert Dewar  <dewar@adacore.com>

* prj.adb, prj-env.adb, back_end.adb: Add comment, minor code clean ups.

From-SVN: r207040

10 years agosem_ch3.adb (Analyze_Declarations): At the end of an appropriate declarative part...
Ed Schonberg [Fri, 24 Jan 2014 14:51:19 +0000 (14:51 +0000)]
sem_ch3.adb (Analyze_Declarations): At the end of an appropriate declarative part...

2014-01-24  Ed Schonberg  <schonberg@adacore.com>

* sem_ch3.adb (Analyze_Declarations): At the end of an
appropriate declarative part, call Freeze_All from the first
declaration in the scope, not from the first unfrozen one. This
is necessary to apply visibility checks to entities with delayed
aspects. Otherwise, in the presence of instantiations and cleanups
that they may generate, the delayed aspects may be analyzed too
late and produce spurious visibility errors.
* sem_attr.adb: Place etype on range.
* sem_ch6.adb: Documentation expression functions.

From-SVN: r207039

10 years agomisc.c (flag_short_enums): Declare.
Tristan Gingold [Fri, 24 Jan 2014 14:48:48 +0000 (14:48 +0000)]
misc.c (flag_short_enums): Declare.

2014-01-24  Tristan Gingold  <gingold@adacore.com>

* gcc-interface/misc.c (flag_short_enums): Declare.
(gnat_post_options): Set it.

From-SVN: r207037

10 years ago[multiple changes]
Arnaud Charlet [Fri, 24 Jan 2014 14:47:12 +0000 (15:47 +0100)]
[multiple changes]

2014-01-24  Robert Dewar  <dewar@adacore.com>

* exp_ch7.adb: Minor change of Indices to Indexes (preferred
terminology in compiler).

2014-01-24  Robert Dewar  <dewar@adacore.com>

* scans.ads: Remove Tok_Raise from Sterm, Eterm, After_SM
categories, now that Ada 95 supports raise expressions.

2014-01-24  Robert Dewar  <dewar@adacore.com>

* freeze.adb (Freeze_Enumeration_Type): Use new target parameter
Short_Enums_On_Target.
* sem_ch13.adb (Set_Enum_Esize): Take Short_Enums_On_Target
into account.
* targparm.ads, targparm.adb: Add new target parameter Short_Enums.

2014-01-24  Ed Schonberg  <schonberg@adacore.com>

* sem_ch5.adb (Analyze_Iterator_Specification): If subtype
indication is given explicity, check that it matches the array
component type or the container element type of the domain
of iteration.

2014-01-24  Tristan Gingold  <gingold@adacore.com>

* back_end.adb (Scan_Compiler_Arguments): Set Short_Enums_On_Target.

2014-01-24  Vincent Celier  <celier@adacore.com>

* prj-env.adb (Ada_Objects_Path): Use Ada_Objects_Path_No_Libs
to cache the result when Including_Libraries is False.
* prj-env.ads (Ada_Objects_Path): Update documentation
* prj.adb (Free (Project_Id)): Also free Ada_Objects_Path_No_Libs
(Get_Object_Directory): Return the Library_Ali_Dir only when
when Including_Libraries is True.
* prj.ads (Get_Object_Directory): Fix and complete documentation
(Project_Data): New component Ada_Objects_Path_No_Libs

From-SVN: r207036

10 years ago[multiple changes]
Arnaud Charlet [Fri, 24 Jan 2014 14:35:45 +0000 (15:35 +0100)]
[multiple changes]

2014-01-24  Robert Dewar  <dewar@adacore.com>

* checks.adb (Expr_Known_Valid): Result of fpt operator never
considered valid.

2014-01-24  Eric Botcazou  <ebotcazou@adacore.com>

* back_end.adb: Minor fix in comment.

2014-01-24  Javier Miranda  <miranda@adacore.com>

* sem_ch3.adb (Check_Abstract_Overriding): Code reestructuration
required to report the error in case of task types.

2014-01-24  Ed Schonberg  <schonberg@adacore.com>

* sem_attr.adb: Additional index checking.

From-SVN: r207035

10 years ago[multiple changes]
Arnaud Charlet [Fri, 24 Jan 2014 14:27:22 +0000 (15:27 +0100)]
[multiple changes]

2014-01-24  Ed Schonberg  <schonberg@adacore.com>

* sem_attr.adb (Analyze_Attribute, case 'Update): Analyze
expressions in each component association, and for records note
the entity in each association choice, for subsequent resolution.
(Resolve_Attribute, case 'Update): Complete resolution of
expressions in each component association.

2014-01-24  Robert Dewar  <dewar@adacore.com>

* sem.adb (Sem): Avoid premature reference to Current_Sem_Unit
(this was causing Is_Main_Unit_Or_Main_Unit_Spec to be set wrong,
leading to wrong handling of SPARK_Mode for library units).

2014-01-24  Robert Dewar  <dewar@adacore.com>

* sem_ch6.adb (Analyze_Subprogram_Body_Helper): Set SPARK_Mode
on generic instances (do not consider them to be internally
generated)

2014-01-24  Doug Rupp  <rupp@adacore.com>

* s-osinte-android.ads (pthread_sigmask): Import sigprocmask
vice pthread_sigmask.

2014-01-24  Vincent Celier  <celier@adacore.com>

* prj.adb (Debug_Output (Str, Str2)): Output if verbosity is
not default.

2014-01-24  Vincent Celier  <celier@adacore.com>

* prj-ext.adb (Add): Do not output anything when Silent is True,
whatever the verbosity. When Source is From_External_Attribute,
set the corresponding environment variable if it is not already set.
* prj-ext.ads (Add): New Boolean parameter Silent, defaulted
to False
* prj-proc.adb (Process_Expression_For_Associative_Array):
For attribute External, call Prj.Ext.Add with Silent set to
True for the child environment, to avoid useless output in non
default verbosity.

2014-01-24  Ed Schonberg  <schonberg@adacore.com>

* sem_res.adb (Set_Slice_Subtype): Handle properly a discrete
range given by a subtype indication, and force evaluation of
the bounds, as for a simple range.
* exp_util.adb (Evaluate_Slice_Bounds): Utility to force evaluation
of bounds of slice for various kinds of discrete ranges.
(Evaluate_Name, Evaluate_Subtype_From_Expr): use
Evaluate_Slice_Bounds.

2014-01-24  Bob Duff  <duff@adacore.com>

* s-taskin.ads (Activator): Make this Atomic, because
Activation_Is_Complete reads it, and that can be called
from any task. Previously, this component was only
modified by the activator before activation, and by
Self after activation.
* a-taside.ads, a-taside.adb (Environment_Task,
Activation_Is_Complete): Implement these missing functions.

From-SVN: r207034

10 years ago[multiple changes]
Arnaud Charlet [Fri, 24 Jan 2014 14:23:03 +0000 (15:23 +0100)]
[multiple changes]

2014-01-24  Doug Rupp  <rupp@adacore.com>

* init.c: Add a handler section for Android.

2014-01-24  Arnaud Charlet  <charlet@adacore.com>

* i-cexten.ads (Unsigned_33..64, Unsigned_33..64): New types.

2014-01-24  Ed Schonberg  <schonberg@adacore.com>

* sem_ch4.adb (Operator_Check): If one operand is a
Raise_Expression, set its type to that of the other operand.
* sem_res.adb (Resolve_Raise_Expression): new procedure.
(Resolve_Actuals): For an actual that is a Raise_Expression,
set the type to that of the formal.
* sem_type.adb (Find_Unique_Type): If one of the operands is a
Raise_Expression, return type of the other operand.

2014-01-24  Ed Schonberg  <schonberg@adacore.com>

* sem_aggr.adb (Resolve_Record_Aggregate): If a scalar
component of the record has a type with a default aspect, and
the corresponding aggregate component is initiaized with a box,
use the default value in the rewritten aggregate.

2014-01-24  Tristan Gingold  <gingold@adacore.com>

* s-interr.ads, s-interr.adb, s-interr-hwint.adb, s-interr-vms.adb,
s-interr-sigaction.adb,
s-interr-dummy.adb (Install_Restricted_Handlers): Add Prio parameter.
* exp_ch9.adb (Make_Initialize_Protection): Add Prio parameter
to the call to Install_Restricted_Handlers.

2014-01-24  Emmanuel Briot  <briot@adacore.com>

* prj-nmsc.adb (Check_File): Add protection when the source is
not fully initialized.

From-SVN: r207033

10 years ago[multiple changes]
Arnaud Charlet [Fri, 24 Jan 2014 14:05:17 +0000 (15:05 +0100)]
[multiple changes]

2014-01-24  Ed Schonberg  <schonberg@adacore.com>

* sem_util.adb (Is_Post_State): In a postcondition, a selected
component that denotes an implicit dereference is a reference
to the post state of the subprogram.

2014-01-24  Robert Dewar  <dewar@adacore.com>

* sem_ch6.adb (Analyze_Subprogram_Body_Helper): SPARK_Mode OFF
for generated subprograms.
(Analyze_Subprogram_Specification): Ditto.

2014-01-24  Vincent Celier  <celier@adacore.com>

* prj-dect.adb (Check_Attribute_Allowed): Detect more forbidden
attributes in package Builder of aggregate and aggregate library
projects.
* prj-nmsc.adb (Process_Naming_Scheme.Check.Check_Aggregate):
Remove procedure (Process_Naming_Scheme.Check.Check_Aggregated):
Remove parameters.  Change error message from "... externally
build library ..." to "... externally built project ...".
(Process_Naming_Scheme.Check): Do not do any check in aggregate
project, as attribute Library_Dir and Library_Name have already
been detected as forbidden.

2014-01-24  Vincent Celier  <celier@adacore.com>

* prj-env.adb (Find_Project): If cached project path is not in
project directory, look in current directory first and use cached
project path only if project is not found in project directory.

From-SVN: r207032

10 years agosem_util.adb, [...]: Correct false positive warnings.
Robert Dewar [Fri, 24 Jan 2014 14:03:19 +0000 (14:03 +0000)]
sem_util.adb, [...]: Correct false positive warnings.

2014-01-24  Robert Dewar  <dewar@adacore.com>

* sem_util.adb, lib-xref.adb: Correct false positive warnings.

From-SVN: r207031

10 years ago[multiple changes]
Arnaud Charlet [Fri, 24 Jan 2014 14:01:57 +0000 (15:01 +0100)]
[multiple changes]

2014-01-24  Vincent Celier  <celier@adacore.com>

* projects.texi: Document that aggregate projects are only
supported by gprbuild, but not by gnatmake.  Document that the
only attribute Switches in package Binder of aggregate projects
will be ignored if its index is not others.  Document that
attribute Global_Config_File is allowed in package Binder of
aggregate projects.

2014-01-24  Robert Dewar  <dewar@adacore.com>

* sem_prag.adb: Minor code reorganization.
* sem_util.adb: Minor fix of potential latent bug in Is_LHS.

From-SVN: r207029

10 years ago[multiple changes]
Arnaud Charlet [Fri, 24 Jan 2014 14:00:34 +0000 (15:00 +0100)]
[multiple changes]

2014-01-24  Pascal Obry  <obry@adacore.com>

* prj-attr.adb, projects.texi, snames.ads-tmpl: Add Excluded_Patterns
attribute definition.

2014-01-24  Vincent Celier  <celier@adacore.com>

* makeutl.adb (Queue.Insert_No_Roots): In gprbuild, do not put
in the Queue the same source (same path, same multi-source index)
from the same project file, to avoid compiling several times
the same source.

2014-01-24  Eric Botcazou  <ebotcazou@adacore.com>

* einfo.ads (First_Rep_Item): Remove obsolete stuff.
(Has_Gigi_Rep_Item): Likewise.
* sem_prag.adb (Analyze_Pragma) <Pragma_Linker_Section>: Do not set
Has_Gigi_Rep_Item for objects.
* gcc-interface/decl.c (prepend_one_attribute_to): Rename into...
(prepend_one_attribute): ...this.
(prepend_one_attribute_pragma): New function extracted from...
(prepend_attributes): ...here.  Swap the parameters for consistency.
(gnat_to_gnu_entity): Adjust calls to prepend_one_attribute_to and to
prepend_attributes.
<object>: Deal with a pragma Linker_Section on a constant
or variable.  <E_Function>: Deal with a pragma Linker_Section
on a subprogram.
(get_minimal_subprog_decl): Adjust calls to prepend_one_attribute_to.

2014-01-24  Vincent Celier  <celier@adacore.com>

* opt.ads: Minor comment update.

From-SVN: r207028

10 years ago[multiple changes]
Arnaud Charlet [Fri, 24 Jan 2014 11:42:21 +0000 (12:42 +0100)]
[multiple changes]

2014-01-24  Robert Dewar  <dewar@adacore.com>

* sem_prag.adb (Analyze_Input_Output): Add missing error check
for junk operand.
* sem_util.adb (Is_Refined_State): Add defense against junk
tree from error.

2014-01-24  Pascal Obry  <obry@adacore.com>

* projects.texi: Removes Build_Slaves attribute documentation.

From-SVN: r207027

10 years ago* gcc.dg/pr59846.c (fn1, fn2): Use ULL suffix.
Marek Polacek [Fri, 24 Jan 2014 10:41:24 +0000 (10:41 +0000)]
* gcc.dg/pr59846.c (fn1, fn2): Use ULL suffix.

From-SVN: r207026

10 years agoGet stack adjustment from push operand in pushsf splitter
H.J. Lu [Fri, 24 Jan 2014 03:38:10 +0000 (03:38 +0000)]
Get stack adjustment from push operand in pushsf splitter

pushsf for -m64/-mx32 looks like

(set (mem:SF (pre_modify:SI (reg/f:SI 7 sp)
    (plus:SI (reg/f:SI 7 sp)
    (const_int -8))))
     (reg:SF 22 xmm1 [orig:84 D.1790 ] [84]))

Stack adjustment is in push operand and it isn't stack register mode size
which may be 4 bytes for -mx32.  This patch gets stack adjustment from
push operand if code of push isn't PRE_DEC.

gcc/

PR target/59929
* config/i386/i386.md (pushsf splitter): Get stack adjustment
from push operand if code of push isn't PRE_DEC.

gcc/testsuite/

PR target/59929
* gcc.target/i386/pr59929.c: New test.

From-SVN: r207023

10 years agore PR target/59909 (Quad memory bootstrap issues on little endian powerpc64 power8...
Michael Meissner [Fri, 24 Jan 2014 01:56:48 +0000 (01:56 +0000)]
re PR target/59909 (Quad memory bootstrap issues on little endian powerpc64 power8 systems)

[gcc]
2014-01-23  Michael Meissner  <meissner@linux.vnet.ibm.com>

PR target/59909
* doc/invoke.texi (RS/6000 and PowerPC Options): Document
-mquad-memory-atomic.  Update -mquad-memory documentation to say
it is only used for non-atomic loads/stores.

* config/rs6000/predicates.md (quad_int_reg_operand): Allow either
-mquad-memory or -mquad-memory-atomic switches.

* config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Add
-mquad-memory-atomic to ISA 2.07 support.

* config/rs6000/rs6000.opt (-mquad-memory-atomic): Add new switch
to separate support of normal quad word memory operations (ldq,
stq) from the atomic quad word memory operations.

* config/rs6000/rs6000.c (rs6000_option_override_internal): Add
support to separate non-atomic quad word operations from atomic
quad word operations.  Disable non-atomic quad word operations in
little endian mode so that we don't have to swap words after the
load and before the store.
(quad_load_store_p): Add comment about atomic quad word support.
(rs6000_opt_masks): Add -mquad-memory-atomic to the list of
options printed with -mdebug=reg.

* config/rs6000/rs6000.h (TARGET_SYNC_TI): Use
-mquad-memory-atomic as the test for whether we have quad word
atomic instructions.
(TARGET_SYNC_HI_QI): If either -mquad-memory-atomic,
-mquad-memory, or -mp8-vector are used, allow byte/half-word
atomic operations.

* config/rs6000/sync.md (load_lockedti): Insure that the address
is a proper indexed or indirect address for the lqarx instruction.
On little endian systems, swap the hi/lo registers after the lqarx
instruction.
(load_lockedpti): Use indexed_or_indirect_operand predicate to
insure the address is valid for the lqarx instruction.
(store_conditionalti): Insure that the address is a proper indexed
or indirect address for the stqcrx. instruction.  On little endian
systems, swap the hi/lo registers before doing the stqcrx.
instruction.
(store_conditionalpti): Use indexed_or_indirect_operand predicate to
insure the address is valid for the stqcrx. instruction.

* gcc/config/rs6000/rs6000-c.c (rs6000_target_modify_macros):
Define __QUAD_MEMORY__ and __QUAD_MEMORY_ATOMIC__ based on what
type of quad memory support is available.

[gcc/testsuite]
2014-01-23  Michael Meissner  <meissner@linux.vnet.ibm.com>

PR target/59909
* gcc.target/powerpc/quad-atomic.c: New file to test power8 quad
word atomic functions at runtime.

From-SVN: r207020

10 years agocompiler: Convert named types before flattening.
Ian Lance Taylor [Fri, 24 Jan 2014 01:01:47 +0000 (01:01 +0000)]
compiler: Convert named types before flattening.

From-SVN: r207018

10 years agoDaily bump.
GCC Administrator [Fri, 24 Jan 2014 00:16:28 +0000 (00:16 +0000)]
Daily bump.

From-SVN: r207015

10 years agoacinclude.m4 (GLIBCXX_CHECK_TMPNAM): New check for tmpnam function.
Bernhard Reutner-Fischer [Thu, 23 Jan 2014 21:17:15 +0000 (22:17 +0100)]
acinclude.m4 (GLIBCXX_CHECK_TMPNAM): New check for tmpnam function.

2014-01-23  Bernhard Reutner-Fischer  <aldot@gcc.gnu.org>
    Steve Ellcey  <sellcey@mips.com>

* acinclude.m4 (GLIBCXX_CHECK_TMPNAM): New check for tmpnam
function.
* configure.ac: Use GLIBCXX_CHECK_TMPNAM.
* (configure, config.h.in): Regenerate.
* include/c_global/cstdio: Guard ::tmpnam with _GLIBCXX_USE_TMPNAM

Co-Authored-By: Steve Ellcey <sellcey@mips.com>
From-SVN: r207009

10 years agore PR c/59846 (Imprecise column number for -Wtype-limits)
Marek Polacek [Thu, 23 Jan 2014 21:01:08 +0000 (21:01 +0000)]
re PR c/59846 (Imprecise column number for -Wtype-limits)

PR c/59846
c-family/
* c-common.c (shorten_compare): Add location_t parameter.
* c-common.h (shorten_binary_op): Adjust declaration.
cp/
* typeck.c (cp_build_binary_op): Pass location to shorten_compare.
c/
* c-typeck.c (parser_build_binary_op): Use location instead of
input_location.
(build_binary_op): Pass location to shorten_compare.
testsuite/
* gcc.dg/pr59846.c: New test.

From-SVN: r207008

10 years agore PR rtl-optimization/59915 (LRA ICE - Repeated looping over subreg reloads (gcc...
Vladimir Makarov [Thu, 23 Jan 2014 20:06:28 +0000 (20:06 +0000)]
re PR rtl-optimization/59915 (LRA ICE - Repeated looping over subreg reloads (gcc.c-torture/compile/simd-3.c))

2014-01-23  Vladimir Makarov  <vmakarov@redhat.com>

PR regression/59915
* lra-constraints.c (simplify_operand_subreg): Spill pseudo if
there is a danger of looping.

From-SVN: r207007

10 years agore PR c/58346 (ICE with SIGFPE at -O1 and above on x86_64-linux-gnu (affecting trunk...
Marek Polacek [Thu, 23 Jan 2014 19:18:49 +0000 (19:18 +0000)]
re PR c/58346 (ICE with SIGFPE at -O1 and above on x86_64-linux-gnu (affecting trunk, 4.8, 4.7, and 4.6))

PR c/58346
c-family/
* c-common.c (pointer_to_zero_sized_aggr_p): New function.
* c-common.h: Declare it.
cp/
* typeck.c (pointer_diff): Give an error on arithmetic on pointer to
an empty aggregate.
c/
* c-typeck.c (pointer_diff): Give an error on arithmetic on pointer to
an empty aggregate.
testsuite/
* c-c++-common/pr58346-1.c: New test.
* c-c++-common/pr58346-2.c: New test.
* c-c++-common/pr58346-3.c: New test.

From-SVN: r207004

10 years agore PR c/59871 (No unused value warning for comma expression)
Marek Polacek [Thu, 23 Jan 2014 19:04:29 +0000 (19:04 +0000)]
re PR c/59871 (No unused value warning for comma expression)

PR c/59871
c/
* c-typeck.c (build_compound_expr): Warn even for right-hand operand
of a comma expression.
(emit_side_effect_warnings): Likewise.
libdecnumber/
* decNumberLocal.h (UBFROMUS, UBFROMUI): Remove last argument.
testsuite/
* gcc.dg/20020220-2.c: Adjust dg-warning message.
* gcc.dg/pr59871.c: New test.

From-SVN: r207002

10 years agore PR c++/55189 (enable -Wreturn-type by default)
Jason Merrill [Thu, 23 Jan 2014 18:54:08 +0000 (13:54 -0500)]
re PR c++/55189 (enable -Wreturn-type by default)

PR c++/55189
* cp-tree.h (struct language_function): Add infinite_loop and
infinite_loops.
(current_function_infinite_loop): New.
* semantics.c (begin_maybe_infinite_loop, end_maybe_infinite_loop)
(break_maybe_infinite_loop): New.
(finish_while_stmt_cond, finish_while_stmt, begin_do_stmt)
(finish_do_stmt, finish_for_cond, finish_for_stmt)
(begin_range_for_stmt): Use them.
* decl.c (finish_function): Don't warn about missing return
if current_function_infinite_loop.
* pt.c (instantiate_decl): Copy current_function_infinite_loop.
* parser.c (cp_parser_jump_statement): Call break_maybe_infinite_loop.

From-SVN: r207001

10 years agocall.c (build_op_delete_call): Use make_tree_vector and release_tree_vector.
Jason Merrill [Thu, 23 Jan 2014 18:53:50 +0000 (13:53 -0500)]
call.c (build_op_delete_call): Use make_tree_vector and release_tree_vector.

* call.c (build_op_delete_call): Use make_tree_vector and
release_tree_vector.

From-SVN: r207000

10 years agore PR libstdc++/59872 (Cannot move std::map with move-only mapped_type)
Jonathan Wakely [Thu, 23 Jan 2014 17:17:04 +0000 (17:17 +0000)]
re PR libstdc++/59872 (Cannot move std::map with move-only mapped_type)

PR libstdc++/59872
* include/bits/stl_map.h (map::operator=(map&&)): Fix comment.
* include/bits/stl_multimap.h (multimap::operator=(multimap&&)):
Likewise.
* include/bits/stl_multiset.h (multiset::operator=(multiset&&)):
Likewise.
* include/bits/stl_set.h (set::operator=(set&&)): Likewise.
* include/bits/stl_tree.h (_Rb_tree::_M_move_data): New overloaded
functions to perform moving or copying of elements from rvalue tree.
(_Rb_tree::_Rb_tree(_Rb_tree&&)): Use _M_move_data.
(_Rb_tree::_Rb_tree(_Rb_tree&&, _Node_allocator&&)): Likewise.
* testsuite/23_containers/map/59872.cc: New.
* testsuite/23_containers/map/56613.cc: Remove duplicate include.

From-SVN: r206994

10 years ago[multiple changes]
Arnaud Charlet [Thu, 23 Jan 2014 17:06:29 +0000 (18:06 +0100)]
[multiple changes]

2014-01-23  Robert Dewar  <dewar@adacore.com>

* opt.adb (Register_Opt_Config_Switches): Save SPARK_Mode_Pragma
setting.

2014-01-23  Ed Schonberg  <schonberg@adacore.com>

* sem_util.adb (Is_Potentially_Unevaluated): Predicate only
applies to expressions that come from source.
* sem_attr.adb (Analyze_Attribute, case 'Old): Improve error
message.
(Analyze_Attribute, case 'Loop_Entry): Apply SPARK 2014 legality
rule regarding potentially unevaluated expressions, to prefix
of attribute.

From-SVN: r206993

10 years ago[multiple changes]
Arnaud Charlet [Thu, 23 Jan 2014 17:03:41 +0000 (18:03 +0100)]
[multiple changes]

2014-01-23  Ed Schonberg  <schonberg@adacore.com>

* exp_util.adb (Make_Invqriant_Call): If type of expression is
a private extension, get invariant from base type.

2014-01-23  Robert Dewar  <dewar@adacore.com>

* sem_util.adb, sem_attr.adb: Minor reformatting.

2014-01-23  Robert Dewar  <dewar@adacore.com>

* opt.adb (Save_Opt_Config_Switches): Save SPARK_Mode_Pragma
(Restore_Opt_Config_Switches): Restore SPARK_Mode_Pragma.
* sem.adb (Semantics): Remove save/restore of
SPARK_Mode[_Pragma]. Not needed since already done in
Save/Restore_Opt_Config_Switches.

2014-01-23  Robert Dewar  <dewar@adacore.com>

* gnat_rm.texi, einfo.adb, einfo.ads, sem_prag.adb, gnat_ugn.texi,
freeze.adb, repinfo.adb, aspects.adb, aspects.ads, sem_ch13.adb:
Linker_Section enhancements.

From-SVN: r206992

10 years agore PR c/59825 (Many cilkplus test failures)
Balaji V. Iyer [Thu, 23 Jan 2014 17:00:53 +0000 (17:00 +0000)]
re PR c/59825 (Many cilkplus test failures)

Fix for PR c/59825.
2014-01-23  Balaji V. Iyer  <balaji.v.iyer@intel.com>

        PR c/59825
        * c-array-notation.c (expand_array_notation_exprs): Rewrote this
        function to use walk_tree and moved a lot of its functionality to
        expand_array_notations.
        (expand_array_notations): New function.

From-SVN: r206991

10 years ago[multiple changes]
Arnaud Charlet [Thu, 23 Jan 2014 16:58:46 +0000 (17:58 +0100)]
[multiple changes]

2014-01-23  Tristan Gingold  <gingold@adacore.com>

* gnat_rm.texi: Minor editing.

2014-01-23  Robert Dewar  <dewar@adacore.com>

* opt.adb (Set_Opt_Config_Switches): Reset SPARK mode for
with'ed internal units.
* sem.adb (Semantics): Save and restore SPARK_Mode[_Pragma].

2014-01-23  Javier Miranda  <miranda@adacore.com>

* lib-xref.adb (Generate_Reference): As part of processing the
"end-of-spec" reference generate an extra reference to the first
private entity of the package.
* xr_tabls.adb (Add_Reference): No action needed for the extra
'E' reference associated; similar to the processing of the
'e' reference.

2014-01-23  Bob Duff  <duff@adacore.com>

* gnat_ugn.texi: Change "--&pp off" to "--!pp off".

2014-01-23  Ed Schonberg  <schonberg@adacore.com>

* sem_util.ads, sem_util.adb (Is_Potentially_Unevaluated): new
predicate to implement rule given in 6.1.1 (20/3).
* sem_attr.adb (Analyze_Attribute, case 'Old): Reject prefix of
'Old in a postcondition, if it is potentially unevaluated and
it is not an entity name.

From-SVN: r206990

10 years agoMinor editing.
Arnaud Charlet [Thu, 23 Jan 2014 16:56:32 +0000 (17:56 +0100)]
Minor editing.

From-SVN: r206989

10 years agognat_ugn.texi: Document the new "--&pp off" feature of gnatpp.
Bob Duff [Thu, 23 Jan 2014 16:55:46 +0000 (16:55 +0000)]
gnat_ugn.texi: Document the new "--&pp off" feature of gnatpp.

2014-01-23  Bob Duff  <duff@adacore.com>

* gnat_ugn.texi: Document the new "--&pp off" feature of gnatpp.

From-SVN: r206988

10 years ago[multiple changes]
Arnaud Charlet [Thu, 23 Jan 2014 16:54:50 +0000 (17:54 +0100)]
[multiple changes]

2014-01-23  Robert Dewar  <dewar@adacore.com>

* gnatlink.adb (Gnatlink): Fix problem of generating bad name
msg on VMS.

2014-01-23  Bob Duff  <duff@adacore.com>

* g-dynhta.ads: Minor comment fix.

2014-01-23  Yannick Moy  <moy@adacore.com>

* sem_ch6.adb (Analyze_Subprogram_Body_Helper): Inherit SPARK_Mode
from spec on body only when not already inherited on spec. Set
SPARK_Mode from context on body without previous spec. *
* sem_prag.adb (Analyze_Pragma): Check placement of pragma on
library-level entities.  Correct retrieval of entity from
declaration, for cases where the declaration is not a unit.
* sem_ch12.adb (Instantiate_Object): Avoid
calling Is_Volatile_Object on an empty node.

From-SVN: r206987

10 years agognatlink.adb (Gnatlink): Check for suspicious executable file names on windows.
Robert Dewar [Thu, 23 Jan 2014 16:53:34 +0000 (16:53 +0000)]
gnatlink.adb (Gnatlink): Check for suspicious executable file names on windows.

2014-01-23  Robert Dewar  <dewar@adacore.com>

* gnatlink.adb (Gnatlink): Check for suspicious executable file
names on windows.

2014-01-23  Robert Dewar  <dewar@adacore.com>

* a-ngelfu.ads: Remove bad uses of AND which should be AND THEN.
* sem_res.adb (Check_No_Direct_Boolean_Operators): Don't give
style errors in instances.
* g-dynhta.ads (Static_HTable): Comment updates.

From-SVN: r206986

10 years agoprj-conf.adb (Get_Or_Create_Configuration_File): Do not attempt to find a configurati...
Vincent Celier [Thu, 23 Jan 2014 16:52:00 +0000 (16:52 +0000)]
prj-conf.adb (Get_Or_Create_Configuration_File): Do not attempt to find a configuration project file when...

2014-01-23  Vincent Celier  <celier@adacore.com>

* prj-conf.adb (Get_Or_Create_Configuration_File): Do not attempt
to find a configuration project file when Config_File_Name is
No_Configuration_File.
* prj-conf.ads (No_Configuration_File): New constant String.
* prj-pars.adb (Parse): Call Get_Or_Create_Configuration_File
with Config_File_Name set to No_Configuration_File, so that
no existing configuration project file will be used, and the
configuration project will be only created in memory when
Add_Default_GNAT_Naming_Scheme is called.
* projects.texi: Minor reformatting.

From-SVN: r206985

10 years agoprj-conf.adb (Get_Or_Create_Configuration_File): Never parse a config project file...
Vincent Celier [Thu, 23 Jan 2014 16:39:02 +0000 (16:39 +0000)]
prj-conf.adb (Get_Or_Create_Configuration_File): Never parse a config project file if On_Load_Config is not null.

2014-01-23  Vincent Celier  <celier@adacore.com>

* prj-conf.adb (Get_Or_Create_Configuration_File): Never parse
a config project file if On_Load_Config is not null.
* prj-pars.adb: Minor comment changes.

From-SVN: r206984

10 years agoMinor reformatting.
Arnaud Charlet [Thu, 23 Jan 2014 16:38:09 +0000 (17:38 +0100)]
Minor reformatting.

From-SVN: r206983

10 years agors6000.c (rs6000_option_override_internal): Don't force flag_ira_loop_pressure if...
Pat Haugen [Thu, 23 Jan 2014 16:37:39 +0000 (16:37 +0000)]
rs6000.c (rs6000_option_override_internal): Don't force flag_ira_loop_pressure if set via command line.

* config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
force flag_ira_loop_pressure if set via command line.

From-SVN: r206982

10 years agolib-xref.adb (Output_References): Output progenitors of synchronized tagged types...
Ed Schonberg [Thu, 23 Jan 2014 16:37:31 +0000 (16:37 +0000)]
lib-xref.adb (Output_References): Output progenitors of synchronized tagged types, for source navigation.

2014-01-23  Ed Schonberg  <schonberg@adacore.com>

* lib-xref.adb (Output_References): Output progenitors of
synchronized tagged types, for source navigation.

From-SVN: r206981

10 years ago[multiple changes]
Arnaud Charlet [Thu, 23 Jan 2014 16:36:41 +0000 (17:36 +0100)]
[multiple changes]

2014-01-23  Robert Dewar  <dewar@adacore.com>

* exp_util.adb, sinfo.adb, sinfo.ads, sem.adb, sem_res.adb,
expander.adb, exp_ch11.adb, exp_ch11.ads, sem_ch11.adb, sem_ch11.ads,
sprint.adb, sprint.ads: Remove unused node N_Subprogram_Info.

2014-01-23  Emmanuel Briot  <briot@adacore.com>

* prj-conf.adb (Get_Or_Create_Configuration_File): call
On_Load_Config later.

2014-01-23  Hristian Kirtchev  <kirtchev@adacore.com>

* sem_ch3.adb (Analyze_Declarations): Do not
generate the spec of the late primitive in ASIS mode. Add two
comments to explain the special cases when the expansion is
not performed.

2014-01-23  Robert Dewar  <dewar@adacore.com>

* sem_util.adb (Note_Possible_Modification): Fix error of
misbehaving for implicit dereference cases in -gnatc mode.

2014-01-23  Emmanuel Briot  <briot@adacore.com>

* prj-pars.adb: Minor reformatting.

From-SVN: r206980

10 years agore PR c++/58980 (ICE with invalid enum declaration)
Paolo Carlini [Thu, 23 Jan 2014 16:31:52 +0000 (16:31 +0000)]
re PR c++/58980 (ICE with invalid enum declaration)

/cp
2014-01-23  Paolo Carlini  <paolo.carlini@oracle.com>

PR c++/58980
* parser.c (cp_parser_enum_specifier): Handle TYPENAME_TYPE as
nested_name_specifier.

/testsuite
2014-01-23  Paolo Carlini  <paolo.carlini@oracle.com>

PR c++/58980
* g++.dg/parse/enum11.C: New.

From-SVN: r206979

10 years ago[PATCH][AArch64] Vector shift by 64 fix
Alex Velenko [Thu, 23 Jan 2014 16:24:13 +0000 (16:24 +0000)]
[PATCH][AArch64] Vector shift by 64 fix

gcc/

* config/aarch64/aarch64-simd-builtins.def (ashr): DI mode removed.
(ashr_simd): New builtin handling DI mode.
* config/aarch64/aarch64-simd.md (aarch64_ashr_simddi): New pattern.
(aarch64_sshr_simddi): New match pattern.
* config/aarch64/arm_neon.h (vshr_n_s32): Builtin call modified.
(vshrd_n_s64): Likewise.
* config/aarch64/predicates.md (aarch64_shift_imm64_di): New predicate.

gcc/testsuite/

* gcc.target/aarch64/sshr64_1.c: New testcase.

From-SVN: r206978

10 years agomsp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
Nick Clifton [Thu, 23 Jan 2014 16:16:33 +0000 (16:16 +0000)]
msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.

* config/msp430/msp430.h (ASM_SPEC): Pass the -mcpu as -mcpu.
(LIB_SPEC): Drop use of memory.ld and peripherals.ld scripts in
favour of mcu specific scripts.
* config/msp430/t-msp430 (MULTILIB_MATCHES): Add more matches for
430x multilibs.

From-SVN: r206977

10 years agoAdded support for Cilk Plus SIMD-enabled functions for C++.
Balaji V. Iyer [Thu, 23 Jan 2014 15:21:42 +0000 (15:21 +0000)]
Added support for Cilk Plus SIMD-enabled functions for C++.

gcc/c/c-parser.c
2014-01-23  Balaji V. Iyer  <balaji.v.iyer@intel.com>

        * c-parser.c (c_finish_omp_declare_simd): Made "cilk simd function"
        attribute an attribute without value.

gcc/cp/ChangeLog
2014-01-23  Balaji V. Iyer  <balaji.v.iyer@intel.com>

        * parser.c (cp_parser_direct_declarator): When Cilk Plus is enabled
        see if there is an attribute after function decl.  If so, then
        parse them now.
        (cp_parser_late_return_type_opt): Handle parsing of Cilk Plus SIMD
        enabled function late parsing.
        (cp_parser_gnu_attribute_list): Parse all the tokens for the vector
        attribute for a SIMD-enabled function.
        (cp_parser_omp_all_clauses): Skip parsing to the end of pragma when
        the function is used by SIMD-enabled function (indicated by NULL
        pragma token).   Added 3 new clauses: PRAGMA_CILK_CLAUSE_MASK,
        PRAGMA_CILK_CLAUSE_NOMASK and PRAGMA_CILK_CLAUSE_VECTORLENGTH
        (cp_parser_cilk_simd_vectorlength): Modified this function to handle
        vectorlength clause in SIMD-enabled function and #pragma SIMD's
        vectorlength clause.  Added a new bool parameter to differentiate
        between the two.
        (cp_parser_cilk_simd_fn_vector_attrs): New function.
        (is_cilkplus_vector_p): Likewise.
        (cp_parser_late_parsing_elem_fn_info): Likewise.
        (cp_parser_omp_clause_name): Added a check for "mask", "nomask"
        and "vectorlength" clauses when Cilk Plus is enabled.
        (cp_parser_omp_clause_linear): Added a new parameter of type bool
        and emit a sorry message when step size is a parameter.
        * parser.h (cp_parser::cilk_simd_fn_info): New field.
        * decl.c (grokfndecl): Added flag_enable_cilkplus along with
        flag_openmp.
        * pt.c (apply_late_template_attributes): Likewise.

testsuite/ChangeLog
2014-01-23  Balaji V. Iyer  <balaji.v.iyer@intel.com>

        * g++.dg/cilk-plus/cilk-plus.exp: Called the C/C++ common tests for
        SIMD enabled function.
        * g++.dg/cilk-plus/ef_test.C: New test.
        * c-c++-common/cilk-plus/ef_error3.c: Made certain messages C specific
        and added C++ ones.
        * c-c++-common/cilk-plus/vlength_errors.c: Added new dg-error tags
        to differenciate C error messages from C++ ones.

From-SVN: r206975

10 years ago[AArch64_BE 4/4] Big-Endian lane numbering fix
James Greenhalgh [Thu, 23 Jan 2014 14:59:20 +0000 (14:59 +0000)]
[AArch64_BE 4/4] Big-Endian lane numbering fix

2014-01-23  James Greenhalgh  <james.greenhalgh@arm.com>
    Alex Velenko  <Alex.Velenko@arm.com>

* config/aarch64/arm_neon.h (vaddv_s8): __LANE0 cleanup.
(vaddv_s16): Likewise.
(vaddv_s32): Likewise.
(vaddv_u8): Likewise.
(vaddv_u16): Likewise.
(vaddv_u32): Likewise.
(vaddvq_s8): Likewise.
(vaddvq_s16): Likewise.
(vaddvq_s32): Likewise.
(vaddvq_s64): Likewise.
(vaddvq_u8): Likewise.
(vaddvq_u16): Likewise.
(vaddvq_u32): Likewise.
(vaddvq_u64): Likewise.
(vaddv_f32): Likewise.
(vaddvq_f32): Likewise.
(vaddvq_f64): Likewise.
(vmaxv_f32): Likewise.
(vmaxv_s8): Likewise.
(vmaxv_s16): Likewise.
(vmaxv_s32): Likewise.
(vmaxv_u8): Likewise.
(vmaxv_u16): Likewise.
(vmaxv_u32): Likewise.
(vmaxvq_f32): Likewise.
(vmaxvq_f64): Likewise.
(vmaxvq_s8): Likewise.
(vmaxvq_s16): Likewise.
(vmaxvq_s32): Likewise.
(vmaxvq_u8): Likewise.
(vmaxvq_u16): Likewise.
(vmaxvq_u32): Likewise.
(vmaxnmv_f32): Likewise.
(vmaxnmvq_f32): Likewise.
(vmaxnmvq_f64): Likewise.
(vminv_f32): Likewise.
(vminv_s8): Likewise.
(vminv_s16): Likewise.
(vminv_s32): Likewise.
(vminv_u8): Likewise.
(vminv_u16): Likewise.
(vminv_u32): Likewise.
(vminvq_f32): Likewise.
(vminvq_f64): Likewise.
(vminvq_s8): Likewise.
(vminvq_s16): Likewise.
(vminvq_s32): Likewise.
(vminvq_u8): Likewise.
(vminvq_u16): Likewise.
(vminvq_u32): Likewise.
(vminnmv_f32): Likewise.
(vminnmvq_f32): Likewise.
(vminnmvq_f64): Likewise.

Co-Authored-By: Alex Velenko <Alex.Velenko@arm.com>
From-SVN: r206973

10 years ago[AArch64_BE 3/4] Big-Endian lane numbering fix
James Greenhalgh [Thu, 23 Jan 2014 14:56:50 +0000 (14:56 +0000)]
[AArch64_BE 3/4] Big-Endian lane numbering fix

2014-01-23  James Greenhalgh  <james.greenhalgh@arm.com>

* config/aarch64/aarch64-simd.md
(aarch64_dup_lane<mode>): Correct lane number on big-endian.
(aarch64_dup_lane_<vswap_widthi_name><mode>): Likewise.
(*aarch64_mul3_elt<mode>): Likewise.
(*aarch64_mul3_elt<vswap_width_name><mode>): Likewise.
(*aarch64_mul3_elt_to_64v2df): Likewise.
(*aarch64_mla_elt<mode>): Likewise.
(*aarch64_mla_elt_<vswap_width_name><mode>): Likewise.
(*aarch64_mls_elt<mode>): Likewise.
(*aarch64_mls_elt_<vswap_width_name><mode>): Likewise.
(*aarch64_fma4_elt<mode>): Likewise.
(*aarch64_fma4_elt_<vswap_width_name><mode>): Likewise.
(*aarch64_fma4_elt_to_64v2df): Likewise.
(*aarch64_fnma4_elt<mode>): Likewise.
(*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
(*aarch64_fnma4_elt_to_64v2df): Likewise.
(aarch64_sq<r>dmulh_lane<mode>): Likewise.
(aarch64_sq<r>dmulh_laneq<mode>): Likewise.
(aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
(aarch64_sqdml<SBINQOPS:as>l_lane<mode>_internal): Likewise.
(aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
(aarch64_sqdmull_lane<mode>_internal): Likewise.
(aarch64_sqdmull2_lane<mode>_internal): Likewise.

From-SVN: r206972

10 years ago[AArch64_BE 2/4] Big-Endian lane numbering fix
Alex Velenko [Thu, 23 Jan 2014 14:48:40 +0000 (14:48 +0000)]
[AArch64_BE 2/4] Big-Endian lane numbering fix

2013-01-23  Alex Velenko  <Alex.Velenko@arm.com>

* config/aarch64/aarch64-simd.md
(aarch64_be_checked_get_lane<mode>): New define_expand.
* config/aarch64/aarch64-simd-builtins.def
(BUILTIN_VALL (GETLANE, be_checked_get_lane, 0):
New builtin definition.
* config/aarch64/arm_neon.h: (__aarch64_vget_lane_any):
Use new safe be builtin.

From-SVN: r206970

10 years ago[AArch64_BE 1/4] Big-Endian lane numbering fix
Alex Velenko [Thu, 23 Jan 2014 14:46:31 +0000 (14:46 +0000)]
[AArch64_BE 1/4] Big-Endian lane numbering fix

[gcc/]
2014-01-23  Alex Velenko  <Alex.Velenko@arm.com>

* config/aarch64/aarch64-simd.md (aarch64_be_ld1<mode>):
New define_insn.
(aarch64_be_st1<mode>): Likewise.
(aarch_ld1<VALL:mode>): Define_expand modified.
(aarch_st1<VALL:mode>): Likewise.
* config/aarch64/aarch64.md (UNSPEC_LD1): New unspec definition.
(UNSPEC_ST1): Likewise.

[gcc/testsuite/]
2014-01-23  Alex Velenko  <Alex.Velenko@arm.com>

* gcc.target/aarch64/vld1-vst1_1.c: New test_case.

From-SVN: r206968

10 years agomicroblaze.md: Add trap insn and attribute
David Holsgrove [Thu, 23 Jan 2014 14:37:44 +0000 (14:37 +0000)]
microblaze.md: Add trap insn and attribute

2014-01-23  David Holsgrove <david.holsgrove@xilinx.com>

* config/microblaze/microblaze.md: Add trap insn and attribute

From-SVN: r206967

10 years agore PR sanitizer/57316 (build failure in libsanitizer)
Yury Gribov [Thu, 23 Jan 2014 14:32:05 +0000 (14:32 +0000)]
re PR sanitizer/57316 (build failure in libsanitizer)

2014-01-23  Yury Gribov  <y.gribov@samsung.com>
    Jakub Jelinek  <jakub@redhat.com>

PR sanitizer/57316
* configure.ac: Check for missing syscalls.
* Makefile.am: Likewise.
* configure: Regenerate.
* Makefile.in: Regenerate.

Co-Authored-By: Jakub Jelinek <jakub@redhat.com>
From-SVN: r206966

10 years agore PR middle-end/58809 (ICE with complex variable in OpenMP reduction clause)
Jakub Jelinek [Thu, 23 Jan 2014 13:32:19 +0000 (14:32 +0100)]
re PR middle-end/58809 (ICE with complex variable in OpenMP reduction clause)

PR middle-end/58809
* c-typeck.c (c_finish_omp_clause): Reject MIN_EXPR, MAX_EXPR,
BIT_AND_EXPR, BIT_IOR_EXPR and BIT_XOR_EXPR on COMPLEX_TYPEs.

* semantics.c (finish_omp_reduction_clause): Reject
BIT_AND_EXPR, BIT_IOR_EXPR and BIT_XOR_EXPR on COMPLEX_TYPEs.

* c-c++-common/gomp/pr58809.c: New test.

From-SVN: r206962

10 years agore PR testsuite/59897 (FAIL: c-c++-common/asan/use-after-return-1.c -O* output...
Dominique d'Humieres [Thu, 23 Jan 2014 12:48:25 +0000 (14:48 +0200)]
re PR testsuite/59897 (FAIL: c-c++-common/asan/use-after-return-1.c  -O*  output pattern test, is 1: 0x2285020)

2014-01-23  Dominique Dhumieres  <dominiq@lps.ens.fr>

        PR sanitizer/59897
        * c-c++-common/asan/use-after-return-1.c: Fixed
        to pass on darwin.

From-SVN: r206961

10 years agofaq.xml (a-how_to_set_paths): Expand FAQ answer.
Jonathan Wakely [Thu, 23 Jan 2014 12:29:40 +0000 (12:29 +0000)]
faq.xml (a-how_to_set_paths): Expand FAQ answer.

* doc/xml/faq.xml (a-how_to_set_paths): Expand FAQ answer.
* doc/xml/manual/abi.xml (abi.versioning.history): Correct symver.

From-SVN: r206960

10 years agoavx512fintrin.h (_mm512_kmov): New.
Kirill Yukhin [Thu, 23 Jan 2014 09:25:11 +0000 (09:25 +0000)]
avx512fintrin.h (_mm512_kmov): New.

gcc/
* config/i386/avx512fintrin.h (_mm512_kmov): New.
* config/i386/i386.c (IX86_BUILTIN_KMOV16): Ditto.
(__builtin_ia32_kmov16): Ditto.
* config/i386/i386.md (UNSPEC_KMOV): New.
(kmovw): Ditto.

testsuite/gcc/
* gcc.target/i386/avx512f-kmovw-1.c: New.

Co-Authored-By: Ilya Tocar <ilya.tocar@intel.com>
From-SVN: r206958

10 years agoPR preprocessor/58580 - preprocessor goes OOM with warning for zero literals
Dodji Seketeli [Thu, 23 Jan 2014 09:13:08 +0000 (09:13 +0000)]
PR preprocessor/58580 - preprocessor goes OOM with warning for zero literals

In this problem report, the compiler is fed a (bogus) translation unit
in which some literals contain bytes whose value is zero.  The
preprocessor detects that and proceeds to emit diagnostics for that
king of bogus literals.  But then when the diagnostics machinery
re-reads the input file again to display the bogus literals with a
caret, it attempts to calculate the length of each of the lines it got
using fgets.  The line length calculation is done using strlen.  But
that doesn't work well when the content of the line can have several
zero bytes.  The result is that the read_line never sees the end of
the line because strlen repeatedly reports that the line ends before
the end-of-line character; so read_line thinks its buffer for reading
the line is too small; it thus increases the buffer, leading to a huge
memory consumption and disaster.

Here is what this patch does.

location_get_source_line is modified to return the length of a source
line that can now contain bytes with zero value.
diagnostic_show_locus() is then modified to consider that a line can
have characters of value zero, and so just shows a white space when
instructed to display one of these characters.

Additionally location_get_source_line is modified to avoid re-reading
each and every line from the beginning of the file until it reaches
the line number N that it is instructed to get; this was leading to
annoying quadratic behaviour when reading adjacent lines near the end
of (big) files.  So a cache is now associated to the file opened in
text mode.  When the content of the file is read, that content is
stashed in the file cache.  That file cache is searched for line
delimiters.  A number of line positions are saved in the cache and a
number of file caches are kept in memory.  That way when
location_get_source_line is asked to read line N + 1, it just has to
start reading from line N that it has already read.

libcpp/ChangeLog:

* include/line-map.h (linemap_get_file_highest_location): Declare
new function.
* line-map.c (linemap_get_file_highest_location): Define it.

gcc/ChangeLog:

* input.h (location_get_source_line): Take an additional line_size
parameter.
(void diagnostics_file_cache_fini): Declare new function.
* input.c (struct fcache): New type.
(fcache_tab_size, fcache_buffer_size, fcache_line_record_size):
New static constants.
(diagnostic_file_cache_init, total_lines_num)
(lookup_file_in_cache_tab, evicted_cache_tab_entry)
(add_file_to_cache_tab, lookup_or_add_file_to_cache_tab)
(needs_read, needs_grow, maybe_grow, read_data, maybe_read_data)
(get_next_line, read_next_line, goto_next_line, read_line_num):
New static function definitions.
(diagnostic_file_cache_fini): New function.
(location_get_source_line): Take an additional output line_len
parameter.  Re-write using lookup_or_add_file_to_cache_tab and
read_line_num.
* diagnostic.c (diagnostic_finish): Call
diagnostic_file_cache_fini.
(adjust_line): Take an additional input parameter for the length
of the line, rather than calculating it with strlen.
(diagnostic_show_locus): Adjust the use of
location_get_source_line and adjust_line with respect to their new
signature.  While displaying a line now, do not stop at the first
null byte.  Rather, display the zero byte as a space and keep
going until we reach the size of the line.
* Makefile.in: Add vec.o to OBJS-libcommon

gcc/testsuite/ChangeLog:

* c-c++-common/cpp/warning-zero-in-literals-1.c: New test file.

Signed-off-by: Dodji Seketeli <dodji@seketeli.org>
From-SVN: r206957

10 years agoavx512fintrin.h (_mm512_loadu_si512): Rename.
Kirill Yukhin [Thu, 23 Jan 2014 09:11:09 +0000 (09:11 +0000)]
avx512fintrin.h (_mm512_loadu_si512): Rename.

gcc/
        * config/i386/avx512fintrin.h (_mm512_loadu_si512): Rename.
        (_mm512_storeu_si512): Ditto.

testsuite/gcc/
        * gcc.target/i386/avx512f-vmovdqu32-1.c: Fix intrinsic name.
        * gcc.target/i386/avx512f-vmovdqu32-2.c: Ditto.
        * gcc.target/i386/avx512f-vpcmpd-2.c: Ditto.
        * gcc.target/i386/avx512f-vpcmpq-2.c: Ditto.
        * gcc.target/i386/avx512f-vpcmpud-2.c: Ditto.
        * gcc.target/i386/avx512f-vpcmpuq-2.c: Ditto.

From-SVN: r206956

10 years agore PR target/52125 (Problems with LO16 asm operands on MIPS)
Richard Sandiford [Thu, 23 Jan 2014 08:24:38 +0000 (08:24 +0000)]
re PR target/52125 (Problems with LO16 asm operands on MIPS)

gcc/
PR target/52125
* rtl.h (get_referenced_operands): Declare.
* recog.c (get_referenced_operands): New function.
* config/mips/mips.c (mips_reorg_process_insns): Check which asm
operands have been referenced when recording LO_SUM references.

gcc/testsuite/
PR target/52125
* gcc.dg/pr48774.c: Remove skip for mips_rel.
* gcc.target/mips/pr52125.c: New test.

From-SVN: r206955

10 years agomicroblaze.md: Correct bswaphi2 insn.
David Holsgrove [Thu, 23 Jan 2014 01:29:15 +0000 (01:29 +0000)]
microblaze.md: Correct bswaphi2 insn.

2014-01-22  David Holsgrove <david.holsgrove@xilinx.com>

* config/microblaze/microblaze.md: Correct bswaphi2 insn.

From-SVN: r206952

10 years agoDaily bump.
GCC Administrator [Thu, 23 Jan 2014 00:16:31 +0000 (00:16 +0000)]
Daily bump.

From-SVN: r206950

10 years agox86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS): Enable for generic and recent AMD...
Jan Hubicka [Thu, 23 Jan 2014 00:12:56 +0000 (01:12 +0100)]
x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS): Enable for generic and recent AMD targets.

* config/i386/x86-tune.def (X86_TUNE_ACCUMULATE_OUTGOING_ARGS):
Enable for generic and recent AMD targets.

From-SVN: r206947

10 years agoPR libstdc++/58764 (again)
Jonathan Wakely [Wed, 22 Jan 2014 23:50:58 +0000 (23:50 +0000)]
PR libstdc++/58764 (again)

PR libstdc++/58764 (again)
* include/bits/stl_list.h (list): Make default constructor's exception
specification conditional.
* include/bits/stl_vector.h (vector): Likewise.
* testsuite/util/testsuite_allocator.h (SimpleAllocator): Add noexcept
to default constructor.
* testsuite/23_containers/list/requirements/dr438/assign_neg.cc:
Adjust dg-error line number.
* testsuite/23_containers/list/requirements/dr438/constructor_1_neg.cc:
Likewise.
* testsuite/23_containers/list/requirements/dr438/constructor_2_neg.cc:
Likewise.
* testsuite/23_containers/list/requirements/dr438/insert_neg.cc:
Likewise.
* testsuite/23_containers/vector/requirements/dr438/assign_neg.cc:
Likewise.
* testsuite/23_containers/vector/requirements/dr438/
constructor_1_neg.cc: Likewise.
* testsuite/23_containers/vector/requirements/dr438/
constructor_2_neg.cc: Likewise.
* testsuite/23_containers/vector/requirements/dr438/insert_neg.cc:
Likewise.

From-SVN: r206946

10 years agore PR c/59891 (ICE on invalid code (with div-by-zero) at all optimization levels...
Marek Polacek [Wed, 22 Jan 2014 22:15:12 +0000 (22:15 +0000)]
re PR c/59891 (ICE on invalid code (with div-by-zero) at all optimization levels on x86_64-linux-gnu)

PR c/59891
c/
* c-typeck.c (build_conditional_expr): Call c_fully_fold instead
of remove_c_maybe_const_expr on op1 and op2.
testsuite/
* gcc.dg/torture/pr59891.c: New test.

From-SVN: r206944

10 years agocombine-stack-adj.c (combine_stack_adjustments_for_block): Remove ARG_SIZE note when...
Jan Hubicka [Wed, 22 Jan 2014 22:05:23 +0000 (23:05 +0100)]
combine-stack-adj.c (combine_stack_adjustments_for_block): Remove ARG_SIZE note when adjustment was eliminated.

* combine-stack-adj.c (combine_stack_adjustments_for_block): Remove
ARG_SIZE note when adjustment was eliminated.

From-SVN: r206943

10 years agore PR tree-optimization/59597 (Performance degradation on Coremark after r205074)
Jeff Law [Wed, 22 Jan 2014 21:27:38 +0000 (14:27 -0700)]
re PR tree-optimization/59597 (Performance degradation on Coremark after r205074)

PR tree-optimization/59597
* tree-ssa-threadupdate.c (dump_jump_thread_path): Move to earlier
in file.  Accept new argument REGISTERING and use it to modify
dump output appropriately.
(register_jump_thread): Corresponding changes.
(mark_threaded_blocks): Reinstate code to cancel unprofitable
thread paths involving joiner blocks.  Add code to dump cancelled
jump threading paths.

PR tree-optimization/59597
* gcc.dg/tree-ssa/pr59597.c: New test.

From-SVN: r206941

10 years agore PR libstdc++/58764 ([lwg/2193] error: converting to ‘const std::vector<std::basic_...
Jonathan Wakely [Wed, 22 Jan 2014 19:46:44 +0000 (19:46 +0000)]
re PR libstdc++/58764 ([lwg/2193] error: converting to ‘const std::vector<std::basic_string<char> >’ from initializer list would use explicit constructor)

PR libstdc++/58764
* include/bits/stl_deque.h (deque::deque(const allocator_type&):
Split into separate default constructor and constructor taking
allocator.
* include/bits/stl_list.h (list::list(const allocator_type&): Likewise.
* include/bits/stl_vector.h (vector::vector(const allocator_type&):
Likewise.
* include/debug/deque (deque::deque(const allocator_type&)): Likewise.
* include/debug/list (list::list(const _Allocator&)): Likewise.
* include/debug/map.h (map::map(const _Compare&, const _Allocator&)):
Likewise.
* include/debug/multimap.h
(multimap::multimap(const _Compare&, const _Allocator&)): Likewise.
* include/debug/set.h (set::set(const _Compare&, const _Allocator&)):
Likewise.
* include/debug/multiset.h
(multiset::multiset(const _Compare&, const _Allocator&)): Likewise.
* include/debug/vector (vector::vector(const allocator_type&)):
Likewise.
* include/profile/deque (deque::deque(const _Allocator&)): Likewise.
* include/profile/list (list::list(const _Allocator&)): Likewise.
* include/profile/map.h
(map::map(const _Compare&, const _Allocator&)): Likewise.
* include/profile/multimap.h
(multimap::multimap(const _Compare&, const _Allocator&)): Likewise.
* include/profile/set.h
(set::set(const _Compare&, const _Allocator&)): Likewise.
* include/profile/multiset.h
(multiset::multiset(const _Compare&, const _Allocator&)): Likewise.
* include/profile/vector (vector::vector(const _Allocator&)):
Likewise.
* testsuite/23_containers/deque/58764.cc: New.
* testsuite/23_containers/list/58764.cc: New.
* testsuite/23_containers/map/58764.cc: New.
* testsuite/23_containers/multimap/58764.cc: New.
* testsuite/23_containers/set/58764.cc: New.
* testsuite/23_containers/multiset/58764.cc: New.
* testsuite/23_containers/vector/58764.cc: New.
* testsuite/23_containers/deque/requirements/dr438/assign_neg.cc:
Adjust dg-error line number.
* testsuite/23_containers/deque/requirements/dr438/
constructor_1_neg.cc: Likewise.
* testsuite/23_containers/deque/requirements/dr438/
constructor_2_neg.cc: Likewise.
* testsuite/23_containers/deque/requirements/dr438/insert_neg.cc:
Likewise.
* testsuite/23_containers/list/requirements/dr438/assign_neg.cc:
Likewise.
* testsuite/23_containers/list/requirements/dr438/constructor_1_neg.cc:
Likewise.
* testsuite/23_containers/list/requirements/dr438/constructor_2_neg.cc:
Likewise.
* testsuite/23_containers/list/requirements/dr438/insert_neg.cc:
Likewise.
* testsuite/23_containers/vector/requirements/dr438/assign_neg.cc:
Likewise.
* testsuite/23_containers/vector/requirements/dr438/
constructor_1_neg.cc: Likewise.
* testsuite/23_containers/vector/requirements/dr438/
constructor_2_neg.cc: Likewise.
* testsuite/23_containers/vector/requirements/dr438/insert_neg.cc:
Likewise.

From-SVN: r206939

10 years agore PR rtl-optimization/59477 (ICE: in assign_by_spills, at lra-assigns.c:1281 with -O)
Vladimir Makarov [Wed, 22 Jan 2014 19:38:47 +0000 (19:38 +0000)]
re PR rtl-optimization/59477 (ICE: in assign_by_spills, at lra-assigns.c:1281 with -O)

2014-01-22  Vladimir Makarov  <vmakarov@redhat.com>

PR rtl-optimization/59477
* lra-constraints.c (inherit_in_ebb): Process call for living hard
regs.  Update reloads_num and potential_reload_hard_regs for all
insns.

2014-01-22  Vladimir Makarov  <vmakarov@redhat.com>

PR rtl-optimization/59477
* g++.dg/pr59477.C: New.

From-SVN: r206938

10 years agolibgo/configure: Test for gold with gccgo -Wl,--help, not ld --help.
Ian Lance Taylor [Wed, 22 Jan 2014 19:10:47 +0000 (19:10 +0000)]
libgo/configure: Test for gold with gccgo -Wl,--help, not ld --help.

From-SVN: r206937

10 years agopr44194-1.c: Match "insn " and "insn:", but not "insn/f".
Richard Sandiford [Wed, 22 Jan 2014 18:45:05 +0000 (18:45 +0000)]
pr44194-1.c: Match "insn " and "insn:", but not "insn/f".

gcc/testsuite/
* gcc.dg/pr44194-1.c: Match "insn " and "insn:", but not "insn/f".

From-SVN: r206936

10 years agore PR c++/59482 (A friend class cannot inherit a private nested class)
Ville Voutilainen [Wed, 22 Jan 2014 18:08:01 +0000 (20:08 +0200)]
re PR c++/59482 (A friend class cannot inherit a private nested class)

/cp
2014-01-22  Ville Voutilainen  <ville.voutilainen@gmail.com>

PR c++/59482
* parser.c (cp_parser_class_head): Push the class before parsing
the base-clause, pop after it.

/testsuite
2014-01-22  Ville Voutilainen  <ville.voutilainen@gmail.com>

PR c++/59482
* g++.dg/pr59482.C: New.

From-SVN: r206933

10 years agoi386-interix.h (i386_pe_unique_section): Don't use PARAMS.
Tom Tromey [Wed, 22 Jan 2014 17:10:10 +0000 (17:10 +0000)]
i386-interix.h (i386_pe_unique_section): Don't use PARAMS.

* config/i386/i386-interix.h (i386_pe_unique_section): Don't use
PARAMS.
* config/cr16/cr16-protos.h (notice_update_cc): Don't use PARAMS.

From-SVN: r206932

10 years ago[multiple changes]
Arnaud Charlet [Wed, 22 Jan 2014 17:04:57 +0000 (18:04 +0100)]
[multiple changes]

2014-01-22  Ed Schonberg  <schonberg@adacore.com>

* sem_ch6.adb (Analyze_Subprogram_Body_Helper): A subprogram
body generated for an expression function within a protected body
needs a set of renaming declarations if the expression function
comes from source.

2014-01-22  Ed Schonberg  <schonberg@adacore.com>

* lib-xref.adb (Get_Type_Reference): In semantics-only mode,
list interface progenitor of a tagged concurrent type, for better
source navigation.

2014-01-22  Robert Dewar  <dewar@adacore.com>

* lib.adb (In_Extended_Main_Code_Unit): Return False for
Standard_Location.
(In_Extended_Main_Source_Unit): Return False for Standard_Location.
* lib.ads (In_Extended_Main_Code_Unit): Add documentation on
treatment of Slocs No_Location and Standard_Location.
* restrict.adb (Check_Restriction_No_Dependence): Explicitly
check for entity with Standard_Location Sloc, rather than relying
on Lib routines to do that.
* sem_res.adb (Resolve_Call): Implement SPARK_05 restriction
that a call cannot occur before a later occuring body within
the same unit.

From-SVN: r206931

10 years ago[multiple changes]
Arnaud Charlet [Wed, 22 Jan 2014 16:53:24 +0000 (17:53 +0100)]
[multiple changes]

2014-01-22  Thomas Quinot  <quinot@adacore.com>

* rtsfind.adb: Update comment.

2014-01-22  Hristian Kirtchev  <kirtchev@adacore.com>

* sem_aux.ads, sem_aux.adb (Is_Body): New routine.
* sem_ch3.adb (Analyze_Declarations): Add local variable
Body_Seen. Generate the spec of a late controlled
primitive body that is about to freeze its related type.
(Handle_Late_Controlled_Primitive): New routine.

2014-01-22  Robert Dewar  <dewar@adacore.com>

* a-stream.adb: Minor reformatting.

2014-01-22  Ed Schonberg  <schonberg@adacore.com>

* sem_ch8.adb (From_Actual_Package): Introduce a recursive
sub-procedure Declared_In_Actual to handle properly the visibility
of actuals in actual packages, that are themselves actuals to a
actual package of the current instance. This mimics properly the
visibility of formals of formal packages declared with a box,
within the corresponding generic unit.

2014-01-22  Robert Dewar  <dewar@adacore.com>

* checks.adb: Do not assume that a volatile variable is valid.

2014-01-22  Thomas Quinot  <quinot@adacore.com>

* g-catiio.ads (Image, Value): Clarify that these functions
operate in the local time zone.  Minor documentation update.

2014-01-22  Thomas Quinot  <quinot@adacore.com>

* csets.adb, csets.ads, opt.ads: Minor documentation fixes.

From-SVN: r206930

10 years ago[multiple changes]
Arnaud Charlet [Wed, 22 Jan 2014 16:47:20 +0000 (17:47 +0100)]
[multiple changes]

2014-01-22  Robert Dewar  <dewar@adacore.com>

* sem_aux.adb, sem_aux.ads, sem_ch3.adb (Has_Unconstrained_Elements):
Moved to sem_aux.adb.

2014-01-22  Robert Dewar  <dewar@adacore.com>

* vms_data.ads: Minor reformatting.

2014-01-22  Robert Dewar  <dewar@adacore.com>

* debug.adb: Document messages affected by -gnatd.E including
the new ones that relate to late definition of equality.
* sem_ch6.adb (Check_Untagged_Equality): In Ada 2012 mode, if
debug flag -gnatd.E is set, then generate warnings rather than
errors.
(Check_Untagged_Equality): In earlier versions of Ada,
generate warnings if Warn_On_Ada_2012_Incompatibility flag is set.

2014-01-22  Robert Dewar  <dewar@adacore.com>

* sem_prag.adb (Usage_Error): Output additional messages for
unconstrained OUT parameters lacking an input dependency.

2014-01-22  Robert Dewar  <dewar@adacore.com>

* sem_ch4.adb: Minor reformatting.

2014-01-22  Robert Dewar  <dewar@adacore.com>

* restrict.ads: Minor reformatting.
* sem_res.adb (Resolve_Call): Check for SPARK_05 restriction that
forbids a call from within a subprogram to the same subprogram.

2014-01-22  Thomas Quinot  <quinot@adacore.com>

* a-stream.ads (Read_SEA, Write_SEA): New subprograms, optimized
stream attributes for Stream_Element_Array.
* a-stream.adb (Read_SEA, Write_SEA): Bodies for the above.
* rtsfind.adb (Check_CRT): Do not reject a reference to an entity
defined in the current scope.

From-SVN: r206929

10 years ago[multiple changes]
Arnaud Charlet [Wed, 22 Jan 2014 16:42:55 +0000 (17:42 +0100)]
[multiple changes]

2014-01-22  Robert Dewar  <dewar@adacore.com>

* debug.adb, exp_ch4.adb, erroutc.adb: Minor reformatting.

2014-01-22  Thomas Quinot  <quinot@adacore.com>

* sem_ch7.adb, sem_ch8.adb, exp_ch3.adb: Minor reformatting.

2014-01-22  Thomas Quinot  <quinot@adacore.com>

* sem_ch3.adb (Analyze_Object_Declaration): For a constant
declaration initialized with a function call, whose type
has variable size, need to remove side effects so that the
initialization expression becomes a dereference of a temporary
reference to the function result.

From-SVN: r206928

10 years agoerrout.adb (Initialize): Remove trick to add dummy entry in Warnings table.
Yannick Moy [Wed, 22 Jan 2014 16:40:30 +0000 (16:40 +0000)]
errout.adb (Initialize): Remove trick to add dummy entry in Warnings table.

2014-01-22  Yannick Moy  <moy@adacore.com>

* errout.adb (Initialize): Remove trick to add dummy entry
in Warnings table.
* erroutc.adb (Set_Warnings_Mode_Off,
Set_Warnings_Mode_On): Do nothing if warnings are suppressed by
command line switch.
* opt.ads (Warning_Mode): Document behavior
wrt pragma Warnings, in normal mode and in GNATprove_Mode.

From-SVN: r206927

10 years agoinsert-vsx-be-order.c: New.
William Schmidt [Wed, 22 Jan 2014 16:10:07 +0000 (16:10 +0000)]
insert-vsx-be-order.c: New.

2014-01-22  Bill Schmidt  <wschmidt@linux.vnet.ibm.com>

* gcc.dg/vmx/insert-vsx-be-order.c: New.
* gcc.dg/vmx/extract-vsx.c: New.
* gcc.dg/vmx/extract-vsx-be-order.c: New.
* gcc.dg/vmx/insert-vsx.c: New.

From-SVN: r206926