Steve Ellcey [Wed, 6 Oct 2004 16:07:03 +0000 (16:07 +0000)]
ia64.h (FUNCTION_ARG_BOUNDARY): Change macro to call ia64_function_arg_boundary.
* config/ia64/ia64.h (FUNCTION_ARG_BOUNDARY): Change macro to
call ia64_function_arg_boundary.
* config/ia64/ia64-protos.h (ia64_function_arg_boundary): New.
* config/ia64/ia64.c (ia64_function_arg_boundary): New.
(ia64_function_arg_advance): Do not put 128 bit floats into
FP registers.
Paul Brook [Wed, 6 Oct 2004 15:29:25 +0000 (15:29 +0000)]
trans-stmt.c (gfc_trans_simple_do): New function.
* trans-stmt.c (gfc_trans_simple_do): New function.
(gfc_trans_do): Use it. Evaluate iteration bounds before entering
loop. Update comments.
testsuite/
* gfortran.dg/do_1.f90: New test.
* builtins.c (expand_builtin_strcpy): Delete duplicate code.
Accept an expression instead of an arglist.
(expand_builtin_stpcpy): Accept an expression instead of an
arglist.
testsuite:
* gcc.c-torture/execute/builtins/lib/strcpy.c: Don't abort when
optimizing for size.
Eric Botcazou [Wed, 6 Oct 2004 11:01:43 +0000 (13:01 +0200)]
re PR target/16007 (Use of static template members results in broken executables)
PR target/16007
* doc/install.texi (*-*-solaris2*): Mention potential problem
with Sun assembler + GNU linker and C++ programs.
Document status of binutils 2.15 release.
Paolo Carlini [Wed, 6 Oct 2004 09:06:59 +0000 (09:06 +0000)]
[multiple changes]
2004-10-06 Paolo Carlini <pcarlini@suse.de>
* include/std/std_sstream.h (_M_sync): When the caller is
setbuf, don't trust _M_string.capacity() to be the size of
the buffer area, use _M_string.size() in this case.
* testsuite/27_io/basic_stringbuf/setbuf/char/4.cc: New.
* testsuite/27_io/basic_stringbuf/setbuf/wchar_t/4.cc: Likewise.
* include/bits/sstream.tcc (overflow): Avoid calling string::assign
unnecessarily when the current _M_string is empty.
* include/bits/stl_algo.h: Minor formatting tweaks.
2004-10-06 Christopher Jefferson <caj@cs.york.ac.uk>
* include/bits/stl_algo.h (__reverse(random_access_iterator_tag)):
Avoid iterator postincrement; fix swapping middle element with
itself on odd-length inputs.
Zack Weinberg [Wed, 6 Oct 2004 04:47:42 +0000 (04:47 +0000)]
pretty-print.c: Include tree.h.
* pretty-print.c: Include tree.h.
(pp_base_prepare_to_format): New function, logic from
text_specifies_location.
(pp_base_format_text): Use gcc_assert.
* pretty-print.h (pp_prepare_to_format): New macro.
(pp_base_prepare_to_format): Prototype.
* diagnostic.c (text_specifies_location): Delete.
(bug_report_request): Delete.
(diagnostic_set_info): Don't call text_specifies_location.
(diagnostic_action_after_output): Put text from
bug_report_request inline here. Use gcc_unreachable.
(diagnostic_report_current_function): Fix comment.
(diagnostic_report_diagnostic): Clarify logic for error recursion.
Call pp_prepare_to_format before diagnostic_starter.
(trim_filename): Use IS_DIR_SEPARATOR.
(fatal_error, internal_error): Use gcc_unreachable.
(error_recursion): Call diagnostic_action_after_output to
issue the bug_report_request message and exit.
* Makefile.in (diagnostic.o, pretty-print.o): Update dependencies.
* c-parse.in: Add list of diagnostic messages to insulate
translation template from version of yacc/bison used to
compile the grammar.
java:
* parse.y, parse-scan.y: Add list of diagnostic messages to
insulate translation template from version of yacc/bison used
to compile the grammar.
treelang:
* parse.y: Add list of diagnostic messages to insulate
translation template from version of yacc/bison used to
compile the grammar.
po:
* gcc.pot: Regenerate.
Joseph Myers [Tue, 5 Oct 2004 19:54:26 +0000 (20:54 +0100)]
c-decl.c (pushdecl): When an extern declaration at block scope refers to a visible entity with...
* c-decl.c (pushdecl): When an extern declaration at block scope
refers to a visible entity with internal linkage, use the old DECL
rather than the new one.
testsuite:
* gcc.c-torture/compile/20041005-1.c: New test.
Kazu Hirata [Tue, 5 Oct 2004 18:53:08 +0000 (18:53 +0000)]
tree-cfg.c (cleanup_tree_cfg): Don't call delete_unreachable_blosk() after thread_jumps().
* tree-cfg.c (cleanup_tree_cfg): Don't call
delete_unreachable_blosk() after thread_jumps().
(thread_jumps): Always remove basic blocks as they become
unreachable.
Kaveh R. Ghazi [Tue, 5 Oct 2004 17:10:42 +0000 (17:10 +0000)]
builtins.c (expand_builtin_memmove): Delete duplicate code and accept a tree type for the result.
* builtins.c (expand_builtin_memmove): Delete duplicate code
and accept a tree type for the result.
(expand_builtin_bcopy): Accept a tree type for the result.
(fold_builtin_memmove): Accept an arglist and tree type for
the result.
Ranjit Mathew [Tue, 5 Oct 2004 17:07:14 +0000 (17:07 +0000)]
Prepare for %q, %< and %> in diagnostic message strings.
* java-tree.h (parse_error_context): remove ATTRIBUTE_PRINTF_2.
Name second parameter 'msgid'.
* parse.y: Additionally include pretty-print.h and diagnostic.h.
(issue_warning_error_from_context): Use pretty-printer functions
instead of vsprintf for constructing formatted messages. Rename
parameter 'msg' to 'msgid'.
(parse_error_context): Rename parameter 'msg' to 'msgid'.
(parse_warning_context): Likewise.
testsuite/
merlin[1:17]% cd build
~
merlin[1:17]% ls
~/build
3 4 4.0.0 c++0x r tree
merlin[1:17]% cd 4.0.0
~/build
merlin[1:17]% cd gcc
~/build/4.0.0
merlin[1:17]% make check-g++
~/build/4.0.0/gcc
(rootme=`${PWDCMD-pwd}`; export rootme; \
srcdir=`cd /home/gdr/redhat/egcs/gcc; ${PWDCMD-pwd}` ; export srcdir ;
\
cd testsuite; \
EXPECT=`if [ -f ${rootme}/../expect/expect ] ; then echo
${rootme}/../expect/expect ; else echo expect ; fi` ; export EXPECT ;
\
if [ -f ${rootme}/../expect/expect ] ; then \
TCL_LIBRARY=`cd .. ; cd /home/gdr/redhat/egcs/gcc/../tcl/library ;
${PWDCMD-pwd}` ; \
export TCL_LIBRARY ; fi ; \
`if [ -f ${srcdir}/../dejagnu/runtest ] ; then echo
${srcdir}/../dejagnu/runtest ; else echo runtest; fi` --tool g++ )
Test Run By gdr on Tue Oct 5 01:18:04 2004
Native configuration is i686-pc-linux-gnu
=== g++ tests ===
Schedule of variations:
unix
Running target unix
Using /usr/share/dejagnu/baseboards/unix.exp as board description file
for target.
Using /usr/share/dejagnu/config/unix.exp as generic interface file for
target.
Using /home/gdr/redhat/egcs/gcc/testsuite/config/default.exp as
tool-and-target-specific interface file.
Running /home/gdr/redhat/egcs/gcc/testsuite/g++.dg/bprob/bprob.exp ...
Running /home/gdr/redhat/egcs/gcc/testsuite/g++.dg/charset/charset.exp
...Running /home/gdr/redhat/egcs/gcc/testsuite/g++.dg/compat/compat.exp
...Running /home/gdr/redhat/egcs/gcc/testsuite/g++.dg/debug/debug.exp ...
Running /home/gdr/redhat/egcs/gcc/testsuite/g++.dg/dg.exp ...
FAIL: g++.dg/other/classkey1.C (test for errors, line 9)
FAIL: g++.dg/other/classkey1.C (test for errors, line 11)
FAIL: g++.dg/other/classkey1.C (test for errors, line 14)
FAIL: g++.dg/other/classkey1.C (test for errors, line 17)
FAIL: g++.dg/other/classkey1.C (test for excess errors)
FAIL: g++.dg/parse/error15.C (test for errors, line 13)
FAIL: g++.dg/parse/error15.C (test for errors, line 14)
FAIL: g++.dg/parse/error15.C (test for errors, line 15)
FAIL: g++.dg/parse/error15.C (test for errors, line 21)
FAIL: g++.dg/parse/error15.C (test for errors, line 22)
FAIL: g++.dg/parse/error15.C (test for errors, line 23)
FAIL: g++.dg/parse/error15.C (test for errors, line 31)
FAIL: g++.dg/parse/error15.C (test for errors, line 32)
FAIL: g++.dg/parse/error15.C (test for errors, line 33)
FAIL: g++.dg/parse/error15.C (test for excess errors)
XPASS: g++.dg/rtti/tinfo1.C scan-assembler _ZTIP9CTemplateIhE:
XPASS: g++.dg/rtti/tinfo1.C scan-assembler-not .globl[
\t]+_ZTIP9CTemplateIhE
XPASS: g++.dg/rtti/tinfo1.C scan-assembler-not
.section[^\n\r]*_ZTIP9CTemplateIhE[^\n\r]*FAIL: g++.dg/template/meminit2.C (test for errors, line 10)
FAIL: g++.dg/template/meminit2.C (test for errors, line 11)
FAIL: g++.dg/template/meminit2.C (test for excess errors)
Running /home/gdr/redhat/egcs/gcc/testsuite/g++.dg/gcov/gcov.exp ...
FAIL: 102:is -:should be 1
FAIL: g++.dg/gcov/gcov-1.C gcov: 1 failures in line counts, 0 in
branch percentamerlin[1:27]% make check-g++
~/build/4.0.0/gcc
(rootme=`${PWDCMD-pwd}`; export rootme; \
srcdir=`cd /home/gdr/redhat/egcs/gcc; ${PWDCMD-pwd}` ; export srcdir ;
\
cd testsuite; \
EXPECT=`if [ -f ${rootme}/../expect/expect ] ; then echo
${rootme}/../expect/expect ; else echo expect ; fi` ; export EXPECT ;
\
if [ -f ${rootme}/../expect/expect ] ; then \
TCL_LIBRARY=`cd .. ; cd /home/gdr/redhat/egcs/gcc/../tcl/library ;
${PWDCMD-pwd}` ; \
export TCL_LIBRARY ; fi ; \
`if [ -f ${srcdir}/../dejagnu/runtest ] ; then echo
${srcdir}/../dejagnu/runtest ; else echo runtest; fi` --tool g++ )
Test Run By gdr on Tue Oct 5 01:27:39 2004
Native configuration is i686-pc-linux-gnu
=== g++ tests ===
Schedule of variations:
unix
Running target unix
Using /usr/share/dejagnu/baseboards/unix.exp as board description file
for target.
Using /usr/share/dejagnu/config/unix.exp as generic interface file for
tar method.c
cfns.gperf cxx-pretty-print.c method.c.~1.315.~
cfns.h cxx-pretty-print.h name-lookup.c
ChangeLog decl2.c name-lookup.c.~1.85.~
ChangeLog.1 decl.c name-lookup.h
ChangeLog.~1.4404.~ decl.h NEWS
ChangeLog.2 dump.c operators.def
ChangeLog.3 error.c optimize.c
ChangeLog.egcs except.c parser.c
ChangeLog.tree-ssa expr.c parser.c.~1.258.~
class.c friend.c pt.c
config-lang.in g++spec.c ptree.c
cp-gimplify.c init.c repo.c
cp-lang.c init.c.~1.396.~ rtti.c
cp-objcp-common.c lang-specs.h search.c
cp-objcp-common.h lex.c semantics.c
cp-tree.def lex.c.~1.349.~ tree.c
cp-tree.h Make-lang.in typeck2.c
CVS mangle.c typeck.c
merlin[7:37]% cvs -z9 diff -p > ~/d-5.patch
~/redhat/egcs/gcc/cp
Enter passphrase for RSA key '/home/gdr/.ssh/identity': cvs [diff
aborted]: received interrupt signal
Chao-ying Fu [Tue, 5 Oct 2004 07:15:03 +0000 (07:15 +0000)]
invoke.texi (-mpaired-single): Link to the new description of the built-in functions.
* doc/invoke.texi (-mpaired-single): Link to the new description of the
built-in functions. Document dependencies.
(-mips3d): Add link here too.
* doc/extend.texi (MIPS Paired-Single Support): New section.
Co-Authored-By: Richard Sandiford <rsandifo@redhat.com>
From-SVN: r88547
Tobias Schlüter [Mon, 4 Oct 2004 21:05:07 +0000 (23:05 +0200)]
re PR fortran/17708 (gfortran problem with goto inside loop)
fortran/
PR fortran/17708
* parse.c (accept_statement): Don't treat END DO like END IF and
END SELECT.
(parse_do_block): Generate possible END DO label inside END DO
block.
also, added ChangeLog entry for previous commit.
testsuite/
PR fortran/17708
* gfortran.dg/pr17708.f90: New test.
Tobias Schlüter [Mon, 4 Oct 2004 20:55:49 +0000 (22:55 +0200)]
trans-const.c (gfc_build_cstring_const): New function.
* trans-const.c (gfc_build_cstring_const): New function.
(gfc_init_cst): Use new function.
* trans-const.h (gfc_build_cstring_const): Add prototype.
* trans-io.c (set_string, set_error_locus): Use new function.
* trans-stmt.c (gfc_trans_goto): Use new function.
Tobias Schlüter [Mon, 4 Oct 2004 20:49:39 +0000 (22:49 +0200)]
re PR libfortran/17631 (libfortran: intrinsic subroutine MVBITS not implemented)
PR fortran/17631
fortran/
* intrinsic.c (add_sym_5): Remove.
(add_subroutines): Add resolution function for MVBITS.
* intrinsic.h (gfc_resolve_mvbits): Declare resolution function for
MVBITS
* iresolve.c (gfc_resolve_mvbits): New function.
(gfc_resolve_random_number): Remove empty line at end of function.
libgfortran/
* Makefile.am (gfor_helper_src): Add intrinsics/mvbits.h.
* intrinsics/mvbits.h: New file.
testsuite/
* gfortran.fortran-torture/execute/intrinsic_mvbits.f90: New test.
Tobias Schlüter [Mon, 4 Oct 2004 19:27:29 +0000 (21:27 +0200)]
re PR fortran/17283 (UNPACK issues)
PR fortran/17283
fortran/
* iresolve.c (gfc_resolve_pack): Choose function depending if mask is
scalar.
libgfortran/
* intrinsics/pack_generic.c (__pack): Allocate memory for return array
if not done by caller.
(__pack_s): New function.
* runtime/memory.c (internal_malloc, internal_malloc64): Allow
allocating zero memory.
testsuite/
* gfortran.fortran-torture/execute/intrinsic_pack.f90: Add more tests.
* s-tassta.adb (Task_Wrapper): Make it Convention C, which makes sense
in general and triggers stack alignment adjustment for thread entry
points on targets where this is necessary.
* init.c (__gnat_initialize): Call an Ada subprogram to perform the
table registration calls when need be. Ensures no reference to the crt
ctors symbol are issued in the SJLJ case, which avoids possible
undefined symbol errors in the case of modules to be statically linked
with the kernel.
Javier Miranda [Mon, 4 Oct 2004 15:01:03 +0000 (15:01 +0000)]
sem_ch4.adb (Try_Object_Operation): Reformat the code to expand in-line the code corresponding to subprogram...
2004-10-04 Javier Miranda <miranda@gnat.com>
* sem_ch4.adb (Try_Object_Operation): Reformat the code to expand
in-line the code corresponding to subprogram Analyze_Actuals. In
addition, analyze the actuals only in case of subprogram call.
Robert Dewar [Mon, 4 Oct 2004 15:00:06 +0000 (15:00 +0000)]
sem_ch5.adb (Unblocked_Exit_Count): Now used for blocks as well as IF and CASE.
2004-10-04 Robert Dewar <dewar@gnat.com>
* sem_ch5.adb (Unblocked_Exit_Count): Now used for blocks as well as
IF and CASE.
(Analyze_Block_Statement): Add circuitry to detect following dead code
(Check_Unreachable_Code): Handle case of block exit
Robert Dewar [Mon, 4 Oct 2004 14:59:10 +0000 (14:59 +0000)]
exp_ch4.adb (Is_Procedure_Actual): Correct so that this does not consider expressions buried within a procedure...
2004-10-04 Robert Dewar <dewar@gnat.com>
* exp_ch4.adb (Is_Procedure_Actual): Correct so that this does not
consider expressions buried within a procedure actual to be an actual.
This caused some blowups with uses of packed slices within a procedure
actual.
Robert Dewar [Mon, 4 Oct 2004 14:58:47 +0000 (14:58 +0000)]
exp_ch3.adb (Needs_Simple_Initialization): Modular packed arrays no longer need to be initialized to zero.
2004-10-04 Robert Dewar <dewar@gnat.com>
* exp_ch3.adb (Needs_Simple_Initialization): Modular packed arrays no
longer need to be initialized to zero.
(Get_Simple_Init_Val): Modular packed arrays no longer need to be
initialized to zero.
* checks.adb (Expr_Known_Valid): Packed arrays are now always
considered valid, even if the representation is modular. That's correct
now that we no longer initialize packed modular arrays to zero.
* exp_dbug.ads: Clarify documentation on handling of PAD and JM
suffixes. These are now documented as the only cases in which the
debugger ignores outer records.
Previously, the spec allowed arbitrary suffixes for this purpose.
Change name of LJM to JM for packed array pad records
Create separate section on packed array handling, and add a whole new
set of comments to this section describing the situation with packed
modular types and justification requirements depending on endianness.
Ed Schonberg [Mon, 4 Oct 2004 14:57:56 +0000 (16:57 +0200)]
sem_ch6.adb (Analyze_Subprogram_Body): Do not treat Inline as Inline_Always when in Configurable_Run_Time mode.
2004-10-04 Ed Schonberg <schonberg@gnat.com>
* sem_ch6.adb (Analyze_Subprogram_Body): Do not treat Inline as
Inline_Always when in Configurable_Run_Time mode.
* sem_prag.adb (Process_Convention): If entity is an inherited
subprogram, apply convention to parent subprogram if in same scope.
(Analyze_Pragma, case Inline): Do not treat Inline as Inline_Always
when in Configurable_Run_Time mode.
Ed Schonberg [Mon, 4 Oct 2004 14:57:11 +0000 (16:57 +0200)]
sem_ch12.adb (Check_Generic_Actuals): New predicate Denotes_Previous_Actual...
2004-10-04 Ed Schonberg <schonberg@gnat.com>
* sem_ch12.adb (Check_Generic_Actuals): New predicate
Denotes_Previous_Actual, to handle properly the case of a private
actual that is also the component type of a subsequent array actual.
The visibility status of the first actual is not affected when the
second is installed.
(Process_Nested_Formal): Subsidiary of Instantiate_Formal_Package, to
make fully recursive the treatment of formals of packages declared
with a box.
(Restore_Nested_Formal): Subsidiary of Restore_Private_Views, to undo
the above on exit from an instantiation.
(Denotes_Formal_Package): When called from Restore_Private_Views, ignore
current instantiation which is now complete.
(Analyze_Package_Instantiation): No instantiated body is needed if the
main unit is generic. Efficient, and avoid anomalies when a instance
appears in a package accessed through rtsfind.
Ed Schonberg [Mon, 4 Oct 2004 14:56:49 +0000 (16:56 +0200)]
exp_ch6.adb (Expand_N_Function_Call): If stack checking is enabled...
2004-10-04 Ed Schonberg <schonberg@gnat.com>
* exp_ch6.adb (Expand_N_Function_Call): If stack checking is enabled,
do not generate a declaration for a temporary if the call is part of a
library-level instantiation.
Ed Schonberg [Mon, 4 Oct 2004 14:56:27 +0000 (16:56 +0200)]
sem_util.adb (Explain_Limited_Type): Ignore internal components when searching for a limited component to flag.
2004-10-04 Ed Schonberg <schonberg@gnat.com>
* sem_util.adb (Explain_Limited_Type): Ignore internal components when
searching for a limited component to flag.
* exp_attr.adb (Freeze_Stream_Subprogram): Subsidiary procedure to
expansion of Input, to account for the fact that the implicit call
generated by the attribute reference must freeze the user-defined
stream subprogram. This is only relevant to 'Input, because it can
appear in an object declaration, prior to the body of the subprogram.
* sem_ch13.adb (Rep_Item_Too_Late): Make the error non-serious, so that
expansion can proceed and further errors uncovered.
(Minor clean up): Fix cases of using | instead of \ for continuation
messages.
Richard Kenner [Mon, 4 Oct 2004 14:56:04 +0000 (14:56 +0000)]
cuintp.c, [...]: Use gcc_assert and gcc_unreachable.
2004-10-04 Richard Kenner <kenner@vlsi1.ultra.nyu.edu>
* cuintp.c, decl.c, utils2.c: Use gcc_assert and gcc_unreachable.
* trans.c (assoc_to_constructor): Fix unused var warning if no checking.
(gnat_gimplify_expr, case ADDR_EXPR): Fix error in last change.
Use gcc_assert and gcc_unreachable.
* decl.c (gnat_to_gnu_entity, case object): Check and process a
specified alignment before validating size.
(gnat_to_gnu_entity) <E_Modular_Integer_Subtype>: Create a
stripped-down declaration for the type of the inner field when making
a JM type.
* utils.c (finish_record_type): Do not compute the size in units
incrementally. Instead compute it once for the rep clause case.
Use gcc_assert and gcc_unreachable.
Vincent Celier [Mon, 4 Oct 2004 14:55:06 +0000 (14:55 +0000)]
prj-part.adb (Parse_Single_Project): Call Is_Extending_All (Extended_Project) only if Extended_Project is defined...
2004-10-04 Vincent Celier <celier@gnat.com>
* prj-part.adb (Parse_Single_Project): Call Is_Extending_All
(Extended_Project) only if Extended_Project is defined, to avoid
assertion error.
(Post_Parse_Context_Clause): Always call Set_Path_Name_Of with a
resolved path.
(Parse_Single_Project): Ditto.
* prj-env.adb (Set_Ada_Paths.Add.Recursive_Add): Do not call
Add_To_Project_Path for virtual projects.
* mlib-prj.ads, mlib-prj.adb (Build_Library): Call
Build_Dynamic_Library with an empty Options_2.
* mlib-utl.ads, mlib-utl.adb (Gcc): Parameter Options_2 has no
default anymore.
* makegpr.adb (Get_Imported_Directories.add): Remove trailing
directory separator, if any.
(Gprmake): Do not allow mains on the command line for library projects.
Do not attempt to link when the project is a library project.
(Library_Opts): New table to store Library_Options.
(Build_Library): If Library_Options is specified, pass these options
when building a shared library.
Jose Ruiz [Mon, 4 Oct 2004 14:51:59 +0000 (16:51 +0200)]
s-tposen.adb (Service_Entry): The object must be always unlocked at the end of this procedure now that the...
2004-10-04 Jose Ruiz <ruiz@act-europe.fr>
* s-tposen.adb (Service_Entry): The object must be always unlocked at
the end of this procedure now that the unlock operation was inserted
by the expander.
Jose Ruiz [Mon, 4 Oct 2004 14:51:17 +0000 (16:51 +0200)]
targparm.ads, [...] (Targparm_Tags): Add PAS value corresponding to the Preallocated_Stacks flags in System.
2004-10-04 Jose Ruiz <ruiz@act-europe.fr>
* targparm.ads, targparm.adb (Targparm_Tags): Add PAS value
corresponding to the Preallocated_Stacks flags in System.
(Get_Target_Parameters): Including the processing for
Preallocated_Stacks.
* system.ads, system-vxworks-x86.ads, system-darwin-ppc.ads,
system-vms_64.ads, system-unixware.ads, system-linux-ia64.ads,
system-freebsd-x86.ads, system-lynxos-ppc.ads, system-lynxos-x86.ads,
system-linux-x86_64.ads, system-tru64.ads, system-aix.ads,
system-vxworks-sparcv9.ads, system-vxworks-xscale.ads,
system-solaris-x86.ads, system-irix-o32.ads, system-irix-n32.ads,
system-hpux.ads, system-vxworks-m68k.ads, system-linux-x86.ads,
system-vxworks-mips.ads, system-os2.ads, system-interix.ads,
system-solaris-sparc.ads, system-solaris-sparcv9.ads, system-vms.ads,
system-mingw.ads, system-vms-zcx.ads, system-vxworks-ppc.ads,
system-vxworks-alpha.ads: Add the flag Preallocated_Stacks, that is
used to signal whether the compiler creates the required stacks and
descriptors for the different tasks (when True) or it is done by the
underlying operating system at run time (when False).
It is initially set to False in all targets.
* exp_ch9.adb (Expand_N_Task_Type_Declaration): Create the task stack
if it is supported by the target.
(Make_Task_Create_Call): Pass the stack address if it has been
previously created. Otherwise pass a Null_Address.
* snames.adb: Add _stack.
* snames.ads: Add Name_uStack. Required to allow the expander to
statically allocated task stacks.
* s-tarest.ads, s-tarest.adb (Create_Restricted_Task): Add
Stack_Address argument.
Check that its value is equal to Null_Address because this target does
not support the static stack allocation.
Thomas Quinot [Mon, 4 Oct 2004 14:50:08 +0000 (16:50 +0200)]
exp_dist.adb: Split declaration of asynchronous flag out of Add_RACW_Read_Attribute.
2004-10-04 Thomas Quinot <quinot@act-europe.fr>
* exp_dist.adb: Split declaration of asynchronous flag out of
Add_RACW_Read_Attribute.
Minor reformatting for better alignment with PolyORB version.
Store the entity for the asynchronous flag of an RACW, rather than the
expression, in the asynchronous flags table. This will allow this flag
to be used in other subprograms beside Add_RACW_Read_Attribute.
Thomas Quinot [Mon, 4 Oct 2004 14:49:35 +0000 (16:49 +0200)]
g-socket.ads, [...]: Add new sockets constant MSG_NOSIGNAL (Linux-specific).
2004-10-04 Thomas Quinot <quinot@act-europe.fr>
* g-socket.ads, g-socket.adb, g-socthi.adb, socket.c,
g-soccon-aix.ads, g-soccon-irix.ads, g-soccon-hpux.ads,
g-soccon-interix.ads, g-soccon-solaris.ads, g-soccon-vms.adb,
g-soccon-mingw.ads, g-soccon-vxworks.ads, g-soccon-freebsd.ads,
g-soccon.ads, g-soccon-unixware.ads, g-soccon-tru64.ads: Add new
sockets constant MSG_NOSIGNAL (Linux-specific).
Add new sockets constant MSG_Forced_Flags, list of flags to be set on
all Send operations.
For Linux, set MSG_NOSIGNAL on all send operations to prevent them
from trigerring SIGPIPE.
Rename components to avoid clash with Ada 2005 possible reserved
word 'interface'.
(Check_Selector): When the select system call returns with an error
condition, propagate Socket_Error to the caller.