tree-pass.h: Added comment about not dumping passes with name starting with star in struct...
2008-08-01 Basile Starynkevitch <basile@starynkevitch>
* gcc/tree-pass.h: Added comment about not dumping passes with name
starting with star in struct opt_pass.
* gcc/passes.c (register_dump_files_1): Don't do dump for a pass with
name starting with star.
* gcc/doc/passes.texi (Pass manager): Mention pass names and special
meaning of star prefix to avoid dump.
Chris Fairles [Thu, 31 Jul 2008 19:37:21 +0000 (19:37 +0000)]
chrono (duration): Use explicitly defaulted ctor, cctor, dtor and assignment.
2008-07-31 Chris Fairles <chris.fairles@gmail.com>
* include/std/chrono (duration): Use explicitly defaulted ctor, cctor,
dtor and assignment. Add diagnostics as per 20.8.3 paragraphs 2, 3
and 4 in WD. Other minor tweaks.
* testsuite/20_util/duration/cons/1_neg.cc: Adjust line numbers.
* testsuite/20_util/duration/requirements/typedefs_neg1.cc: New.
* testsuite/20_util/duration/requirements/typedefs_neg2.cc: Likewise.
* testsuite/20_util/duration/requirements/typedefs_neg3.cc: Likewise.
Jakub Jelinek [Thu, 31 Jul 2008 19:12:14 +0000 (21:12 +0200)]
re PR preprocessor/36649 (-H option doesn't work as expected)
PR preprocessor/36649
* files.c (struct report_missing_guard_data): New type.
(report_missing_guard): Put paths into an array instead of printing
them right away. Return 1 rather than 0.
(report_missing_guard_cmp): New function.
(_cpp_report_missing_guards): Sort and print paths gathered by
report_missing_guard callback.
* gcc.dg/pch/cpp-3.hs: Add include guards.
* gcc.dg/pch/cpp-3a.h: Likewise.
* gcc.dg/pch/cpp-3b.h: Likewise.
* gcc.dg/cpp/mi8.c: New test.
* gcc.dg/cpp/mi8a.h: New file.
* gcc.dg/cpp/mi8b.h: New file.
* gcc.dg/cpp/mi8c.h: New file.
* gcc.dg/cpp/mi8d.h: New file.
mkstemps.c (mkstemps): If open failed with errno other than EEXIST, return immediately.
* mkstemps.c (mkstemps): If open failed with errno other than
EEXIST, return immediately.
* make-temp-file.c: Include errno.h.
(make_temp_file): If mkstemps failed, print an error message
before aborting.
Paolo Carlini [Thu, 31 Jul 2008 18:43:08 +0000 (18:43 +0000)]
libstdc++.exp (libstdc++_init): Set v3-libgomp.
2008-07-31 Paolo Carlini <paolo.carlini@oracle.com>
* testsuite/lib/libstdc++.exp (libstdc++_init): Set v3-libgomp.
(check_v3_target_parallel_mode): Robustify, just follow the
structure of testsuite/Makefile.am.
Jakub Jelinek [Thu, 31 Jul 2008 18:08:36 +0000 (20:08 +0200)]
re PR rtl-optimization/36419 (Wrong unwind info with -Os -fasynchronous-unwind-tables)
PR rtl-optimization/36419
* dwarf2out.c (barrier_args_size): New variable.
(compute_barrier_args_size, compute_barrier_args_size_1): New
functions.
(dwarf2out_stack_adjust): For BARRIERs call compute_barrier_args_size
if not called yet in the current function, use barrier_args_size
array to find the new args_size value.
(dwarf2out_frame_debug): Free and clear barrier_args_size.
Jose Ruiz [Thu, 31 Jul 2008 14:41:01 +0000 (16:41 +0200)]
2008-07-31 Jose Ruiz <ruiz@adacore.com>
* s-parame-vxworks.adb
Document that this body is used for RTX in RTSS (kernel) mode.
* gcc-interface/Makefile.in
(LIBGNAT_TARGET_PAIRS for the rtx_rtss run time): Use the
s-parame-vxworks.adb body in order to have reasonable stack sizes in
RTX RTSS kernel mode. Virtual memory is not used in that case, so we
cannot ask for too big values.
Gary Dismukes [Thu, 31 Jul 2008 13:53:55 +0000 (15:53 +0200)]
2008-07-31 Gary Dismukes <dismukes@adacore.com>
* sem_aggr.adb:
(Resolve_Record_Aggregate): Bypass error that a type without
components must have a "null record" aggregate when compiling for Ada
2005, since it's legal to give an aggregate of form (others => <>)
for such a type.
Ed Schonberg [Thu, 31 Jul 2008 13:31:37 +0000 (15:31 +0200)]
sem_ch10.adb (Build_Unit_Name): If the unit name in a with_clause has the form A.B.C and B is a unit renaming...
2008-07-31 Ed Schonberg <schonberg@adacore.com>
* sem_ch10.adb (Build_Unit_Name): If the unit name in a with_clause
has the form A.B.C and B is a unit renaming, analyze its compilation
unit and add a with_clause on A.b to the context.
* bindgen.adb Comment reformatting. Update the list of run-time globals.
(Gen_Adainit_Ada): Add the declaration, import and value set for
configuration flag Canonical_Streams.
(Gen_Adainit_C): Add the declaration and initial value of external
symbol __gl_canonical_streams.
* init.c: Update the list of global values computed by the binder.
* opt.ads: Add flag Canonical_Streams.
* par-prag.adb (Prag): Include Pragma_Canonical_Streams to the list of
semantically handled pragmas.
* sem_prag.adb: Add an entry into enumeration type Sig_Flags.
(Analyze_Pragma): Add case for pragma Canonical_Streams.
* snames.adb: Add character value for name Canonical_Streams.
* snames.ads:
Add Name_Canonical_Streams to the list of configuration pragmas.
Add Pragma_Canonical_Streams to enumeration type Pragma_Id.
* snames.h: Add a definition for Pragma_Canonical_Streams.
* s-ststop.adb:
Add a flag and import to seize the value of external symbol
__gl_canonical_streams. Update comment and initial value of constant
Use_Block_IO.
* gnat_rm.texi: Add section of pragma Canonical_Streams.
* gnat_ugn.texi:
Add pragma Canonical_Streams to the list of configuration pragmas.
* s-stausa.ad? (Fill_Stack): Stack_Used_When_Filling is now stored
anymore - just used internally.
Added handling of very small tasks - when the theoretical size is
already full at the point of the call.
(Report_Result): Fixed result computation, Stack_Used_When_Filling does
not need to be added to the result.
Javier Miranda [Thu, 31 Jul 2008 13:25:48 +0000 (15:25 +0200)]
sem_ch4.adb (Valid_First_Argument_Of): Complete its functionality to handle synchronized types.
2008-07-31 Javier Miranda <miranda@adacore.com>
* sem_ch4.adb (Valid_First_Argument_Of): Complete its functionality to
handle synchronized types. Required to handle well the object.operation
notation applied to synchronized types.
Javier Miranda [Thu, 31 Jul 2008 12:46:35 +0000 (14:46 +0200)]
sem_type.adb (Has_Compatible_Type): Complete support for synchronized types when...
2008-07-31 Javier Miranda <miranda@adacore.com>
* sem_type.adb (Has_Compatible_Type): Complete support for synchronized
types when the candidate type is a synchronized type.
* sem_res.adb (Resolve_Actuals): Reorganize code handling synchronized
types, and complete management of synchronized types adding missing
code to handle formal that is a synchronized type.
* sem_ch4.adb (Try_Primitive_Operation): Avoid testing attributes that
are not available and cause the compiler to blowup. Found compiling
test with switch -gnatc
* sem_ch6.adb (Check_Synchronized_Overriding): Remove local subprogram
Has_Correct_Formal_Mode plus code cleanup.
* checks.adb (Apply_Accessibility_Check): Insert the check on
Insert_Node.
* exp_attr.adb:
(Expand_N_Attribute_Refernce, Attribute_Access): Pass attribute node
to new parameter Insert_Node on call to Apply_Accessibility_Check.
Necessary to distinguish the insertion node because the dereferenced
formal may come from a rename, but the check must be inserted in
front of the attribute.
* exp_ch4.adb:
(Expand_N_Allocator): Pass actual for new Insert_Node parameter on
call to Apply_Accessibility_Check.
(Expand_N_Type_Conversion): Pass actual for new Insert_Node parameter
on call to Apply_Accessibility_Check.
Minor reformatting
Vincent Celier [Thu, 31 Jul 2008 12:37:42 +0000 (14:37 +0200)]
prj-part.adb, [...]: Move back spec of Parse_Single_Project to body...
2008-07-31 Vincent Celier <celier@adacore.com>
* prj-part.adb, prj-part.ads, prj.adb, prj.ads, prj-env.adb:
Move back spec of Parse_Single_Project to body, as it is not called
outside of package Prj.Part.
(Project_Data): Remove components Linker_Name, Linker_Path and
Minimum_Linker_Options as they are no longer set.
Remove function There_Are_Ada_Sources from package Prj and move code
in the only place it was used, in Prj.Env.Set_Ada_Paths.
Ed Schonberg [Thu, 31 Jul 2008 12:37:33 +0000 (14:37 +0200)]
exp_aggr.adb (Aggr_Size_OK): If the aggregate has a single component and the context is an object...
2008-07-31 Ed Schonberg <schonberg@adacore.com>
* exp_aggr.adb (Aggr_Size_OK): If the aggregate has a single component
and the context is an object declaration with non-static bounds, treat
the aggregate as non-static.
Gary Dismukes [Thu, 31 Jul 2008 11:04:10 +0000 (13:04 +0200)]
exp_attr.adb (Enclosing_Object): New function local to handling of access attributes...
2008-07-31 Gary Dismukes <dismukes@adacore.com>
* exp_attr.adb (Enclosing_Object): New function local to handling of
access attributes,
for retrieving the innermost enclosing object prefix of a compound name.
(Expand_N_Attribute_Reference, N_Attribute_Access): In the case where an
Access attribute has a prefix that is a dereference of an access
parameter (or the prefix is a subcomponent selected from such a
dereference), apply an accessibility check to the access parameter.
Replaces code that rewrote the prefix as a type conversion (and that
didn't handle subcomponent cases).
Also, this is now only applied in the case of 'Access.
* exp_ch6.adb (Expand_Call): Add handling for the case of an access
discriminant passed as an actual to an access formal, passing the
Object_Access_Level of the object containing the access discriminant.
Ed Schonberg [Thu, 31 Jul 2008 10:27:42 +0000 (12:27 +0200)]
exp_ch6.adb (Make_Build_In_Place_Call_In_Object_Declaration): Use Insert_Actions to place the pointer declaration in the code...
2008-07-31 Ed Schonberg <schonberg@adacore.com>
* exp_ch6.adb (Make_Build_In_Place_Call_In_Object_Declaration): Use
Insert_Actions to place the pointer declaration in the code, rather
than Insert_Before_And_Analyze, so that insertions of temporaries are
kept in the proper order when transient scopes are present.
* gcc-interface/utils.c:
(init_gigi_decls): Add initialization of mulv64_decl
* gcc-interface/trans.c:
(build_unary_op_trapv): New function
(build_binary_op_trapv): New function
(gnat_to_gnu): Use the above functions instead of
build_{unary,binary}_op
* exp_attr.adb (Find_Stream_Subprogram): Check the base type instead
of the type when looking for stream subprograms for type String,
Wide_String and Wide_Wide_String.
* s-ststop.adb: Change the initialization expression of constant
Use_Block_IO.
* passes.c (init_optimization_passes): Always call
pass_early_warn_uninitialized.
* opts.c (decode_options): Do not warn about -Wuninitialized
at -O0.
* doc/invoke.texi (-Wuninitialized): Correct for enabling at -O0.
* doc/passes.texi (Warn for uninitialized variables): Adjust.
Ed Schonberg [Thu, 31 Jul 2008 09:42:51 +0000 (11:42 +0200)]
sem_ch6.adb (Analyze_Generic_Subprogram_Body): After analysis...
2008-07-31 Ed Schonberg <schonberg@adacore.com>
* sem_ch6.adb (Analyze_Generic_Subprogram_Body): After analysis,
transfer pre/postconditions from generic copy to original tree, so that
they will appear in each instance.
(Process_PPCs): Do not transform postconditions into a procedure in a
generic context, to prevent double expansion of check pragmas.
* sem_attr.adb: In an instance, the prefix of the 'result attribute
can be the renaming of the
current instance, so check validity of the name accordingly.
Ed Schonberg [Thu, 31 Jul 2008 08:17:31 +0000 (10:17 +0200)]
sem_attr.adb: 'Result can have an ambiguous prefix, and is resolved from context.
2008-07-31 Ed Schonberg <schonberg@adacore.com>
sem_attr.adb: 'Result can have an ambiguous prefix, and is resolved
from context. This attribute must be usable in Ada95 mode.
The attribute can appear in the body of a function marked
Inline_Always, but in this case the postocondition is not enforced.
sem_prag.adb (Check_Precondition_Postcondition): within the expansion
of an inlined call pre- and postconditions are legal
Jakub Jelinek [Thu, 31 Jul 2008 08:02:49 +0000 (10:02 +0200)]
re PR c/36970 (GCC should display a warning when trying to free a static array)
PR c/36970
* builtins.c (maybe_emit_free_warning): New function.
(expand_builtin): Process BUILT_IN_FREE even at -O0. Call
maybe_emit_free_warning for BUILT_IN_FREE.
* gcc.dg/free-1.c: New test.
* gcc.dg/free-2.c: New test.
Jakub Jelinek [Thu, 31 Jul 2008 08:01:25 +0000 (10:01 +0200)]
re PR debug/36278 (ICE with typedef void in namespace and using the defined type in another when compiling with "-g")
PR debug/36278
* dwarf2out.c (get_context_die): New function.
(force_decl_die, force_type_die): Use it.
(dwarf2out_imported_module_or_decl): Likewise. If base_type_die
returns NULL, force generation of DW_TAG_typedef and put that into
DW_AT_import.
Eric Botcazou [Wed, 30 Jul 2008 23:54:56 +0000 (23:54 +0000)]
re PR ada/36554 (verify_flow_info ICE can not throw but has EH edges)
PR ada/36554
* dwarf2out.c (is_subrange_type): Deal with BOOLEAN_TYPE.
ada/
* back_end.adb (Call_Back_End): Pass Standard_Boolean to gigi.
* gcc-interface/gigi.h (gigi): Take new standard_boolean parameter.
* gcc-interface/decl.c (gnat_to_gnu_entity) <E_Enumeration_Subtype>:
Set precision to 1 for subtype of BOOLEAN_TYPE.
(set_rm_size): Set TYPE_RM_SIZE_NUM for BOOLEAN_TYPE.
(make_type_from_size): Deal with BOOLEAN_TYPE.
* gcc-interface/misc.c (gnat_print_type): Likewise.
* gcc-interface/trans.c (gigi): Take new standard_boolean parameter.
Set boolean_type_node as its translation in the table, as well
as boolean_false_node for False and boolean_true_node for True.
* gcc-interface/utils.c (gnat_init_decl_processing): Create custom
8-bit boolean_type_node and set its TYPE_RM_SIZE_NUM.
(create_param_decl): Deal with BOOLEAN_TYPE.
(build_vms_descriptor): Likewise.
(build_vms_descriptor64): Likewise.
(convert): Deal with BOOLEAN_TYPE like with ENUMERAL_TYPE.
projects / thirdparty / gcc.git / log
