| 
Gumstix

OE BUILD

View: New views
10 Messages — Rating Filter:   Alert me  

OE BUILD

by Marshall Lawrence :: Rate this Message:

Reply to Author | View Threaded | Show Only this Message

Some parts of this message have been removed. Learn more about Nabble's security policy.

Hi folks,

 

I just tried to use the basic directions from the gumstix.net web site to build a clean basic OS.  I’m using Fedora 9 as my build machine.

 

When I run Bitbake as required, things seem to run fine until it gets to building GSM.  I get “multiple declaration” errors and then bitbake errors out.

 

Any suggestions?

 

Thanks,

Marshall


-------------------------------------------------------------------------
This SF.net email is sponsored by: Microsoft
Defy all challenges. Microsoft(R) Visual Studio 2008.
http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/
_______________________________________________
gumstix-users mailing list
gumstix-users@...
https://lists.sourceforge.net/lists/listinfo/gumstix-users

Re: OE BUILD

by walt0x007 :: Rate this Message:

Reply to Author | View Threaded | Show Only this Message

Marshall,

Did you get resolution on this issue?  I'm having a similar problem with "multiple declaration" errors and am also using fedora 9.   I posted a note about it here:  http://www.nabble.com/anyone-using-fedora-9----tt17419326.html 

Thanks,
Walt



Marshall Lawrence wrote:
Hi folks,

 

I just tried to use the basic directions from the gumstix.net web site to
build a clean basic OS.  I'm using Fedora 9 as my build machine.

 

When I run Bitbake as required, things seem to run fine until it gets to
building GSM.  I get "multiple declaration" errors and then bitbake errors
out.

 

Any suggestions?

 

Thanks,

Marshall


-------------------------------------------------------------------------
This SF.net email is sponsored by: Microsoft
Defy all challenges. Microsoft(R) Visual Studio 2008.
http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/
_______________________________________________
gumstix-users mailing list
gumstix-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/gumstix-users

Re: OE BUILD

by peteb :: Rate this Message:

Reply to Author | View Threaded | Show Only this Message

Some parts of this message have been removed. Learn more about Nabble's security policy.
Looks like the OE certain packages don't like Fedora 9. For me, it dies almost immediately when building gmp-native for the cross compiler. Anyone else seen this and know how to work around it?

Marshall Lawrence wrote:

Hi folks,

 

I just tried to use the basic directions from the gumstix.net web site to build a clean basic OS.  I’m using Fedora 9 as my build machine.

 

When I run Bitbake as required, things seem to run fine until it gets to building GSM.  I get “multiple declaration” errors and then bitbake errors out.

 

Any suggestions?

 

Thanks,

Marshall


-------------------------------------------------------------------------
This SF.net email is sponsored by: Microsoft 
Defy all challenges. Microsoft(R) Visual Studio 2008. 
http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/

_______________________________________________
gumstix-users mailing list
gumstix-users@...
https://lists.sourceforge.net/lists/listinfo/gumstix-users
  


-- 
Pete

Funny=Truth
Truth=Beauty
So take it as a complement when people say you look Funny.

-------------------------------------------------------------------------
This SF.net email is sponsored by: Microsoft
Defy all challenges. Microsoft(R) Visual Studio 2008.
http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/
_______________________________________________
gumstix-users mailing list
gumstix-users@...
https://lists.sourceforge.net/lists/listinfo/gumstix-users

Re: OE BUILD

by peteb :: Rate this Message:

Reply to Author | View Threaded | Show Only this Message

Some parts of this message have been removed. Learn more about Nabble's security policy.
I think the problem is with the -std=gnu99 handling of inline functions in the new Fedora 9 compiler. Not sure exactly how to fix this though. I don't know how it's getting turned on in the OE version of gmp, because I can't find the configure options. Building gmp natively without that option and it works just fine. Does anyone know how this works, or how to disable the gnu99 check? I can't figure it out for OE.

Peter Buelow wrote:
Looks like the OE certain packages don't like Fedora 9. For me, it dies almost immediately when building gmp-native for the cross compiler. Anyone else seen this and know how to work around it?

Marshall Lawrence wrote:

Hi folks,

 

I just tried to use the basic directions from the gumstix.net web site to build a clean basic OS.  I’m using Fedora 9 as my build machine.

 

When I run Bitbake as required, things seem to run fine until it gets to building GSM.  I get “multiple declaration” errors and then bitbake errors out.

 

Any suggestions?

 

Thanks,

Marshall


-------------------------------------------------------------------------
This SF.net email is sponsored by: Microsoft 
Defy all challenges. Microsoft(R) Visual Studio 2008. 
http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/

_______________________________________________
gumstix-users mailing list
gumstix-users@...
https://lists.sourceforge.net/lists/listinfo/gumstix-users
  


-- 
Pete

Funny=Truth
Truth=Beauty
So take it as a complement when people say you look Funny.

  

-------------------------------------------------------------------------
This SF.net email is sponsored by: Microsoft
Defy all challenges. Microsoft(R) Visual Studio 2008.
http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/

_______________________________________________
gumstix-users mailing list
gumstix-users@...
https://lists.sourceforge.net/lists/listinfo/gumstix-users
  


-- 
Pete

Funny=Truth
Truth=Beauty
So take it as a complement when people say you look Funny.

-------------------------------------------------------------------------
This SF.net email is sponsored by: Microsoft
Defy all challenges. Microsoft(R) Visual Studio 2008.
http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/
_______________________________________________
gumstix-users mailing list
gumstix-users@...
https://lists.sourceforge.net/lists/listinfo/gumstix-users

Re: OE BUILD

by Terry Kemp-2 :: Rate this Message:

Reply to Author | View Threaded | Show Only this Message

OE.dev has gmp-4.2.2 recipes and patches that fix this.

to override angstrom (2007) preferred version (4.2.1) add in
gumstix-oe/com.gumstix.collection/conf/machine/include...

PREFERRED_VERSION_gmp = "4.2.2"
PREFERRED_VERSION_gmp-native = "4.2.2"


next the kernel will fail...

scripts/mod/sumversion.c: In function 'get_src_version':
| scripts/mod/sumversion.c:384: error: 'PATH_MAX' undeclared (first use
in this function)


in scripts/mod/sumversion.c you need to add
#include <limits.h>

This got me through bitbake gumstix-kernel (from a blown away tmp) OK.

Let me know if you want the gmp 4.2.2 recipes and patches, although it
would be better to get Steve to add them into svn from dev.

Terry

On Sun, 2008-05-25 at 13:20 -0500, Peter Buelow wrote:

> I think the problem is with the -std=gnu99 handling of inline
> functions in the new Fedora 9 compiler. Not sure exactly how to fix
> this though. I don't know how it's getting turned on in the OE version
> of gmp, because I can't find the configure options. Building gmp
> natively without that option and it works just fine. Does anyone know
> how this works, or how to disable the gnu99 check? I can't figure it
> out for OE.
>
> Peter Buelow wrote:
> > Looks like the OE certain packages don't like Fedora 9. For me, it
> > dies almost immediately when building gmp-native for the cross
> > compiler. Anyone else seen this and know how to work around it?
> >
> > Marshall Lawrence wrote:
> > > Hi folks,
> > >
> > >  
> > >
> > > I just tried to use the basic directions from the gumstix.net web
> > > site to build a clean basic OS.  I’m using Fedora 9 as my build
> > > machine.
> > >
> > >  
> > >
> > > When I run Bitbake as required, things seem to run fine until it
> > > gets to building GSM.  I get “multiple declaration” errors and
> > > then bitbake errors out.
> > >
> > >  
> > >
> > > Any suggestions?
> > >
> > >  
> > >
> > > Thanks,
> > >
> > > Marshall
> > >
> > >
> > >
> > > __________________________________________________________________
> > >
> > > -------------------------------------------------------------------------
> > > This SF.net email is sponsored by: Microsoft
> > > Defy all challenges. Microsoft(R) Visual Studio 2008.
> > > http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/
> > >
> > > __________________________________________________________________
> > >
> > > _______________________________________________
> > > gumstix-users mailing list
> > > gumstix-users@...
> > > https://lists.sourceforge.net/lists/listinfo/gumstix-users
> > >  
> >
> >
> > --
> > Pete
> >
> > Funny=Truth
> > Truth=Beauty
> > So take it as a complement when people say you look Funny.
> >
> >  
> >
> > ____________________________________________________________________
> >
> > -------------------------------------------------------------------------
> > This SF.net email is sponsored by: Microsoft
> > Defy all challenges. Microsoft(R) Visual Studio 2008.
> > http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/
> >
> > ____________________________________________________________________
> >
> > _______________________________________________
> > gumstix-users mailing list
> > gumstix-users@...
> > https://lists.sourceforge.net/lists/listinfo/gumstix-users
> >  
>
>
> --
> Pete
>
> Funny=Truth
> Truth=Beauty
> So take it as a complement when people say you look Funny.
>
> -------------------------------------------------------------------------
> This SF.net email is sponsored by: Microsoft
> Defy all challenges. Microsoft(R) Visual Studio 2008.
> http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/
> _______________________________________________ gumstix-users mailing list gumstix-users@... https://lists.sourceforge.net/lists/listinfo/gumstix-users


-------------------------------------------------------------------------
This SF.net email is sponsored by: Microsoft
Defy all challenges. Microsoft(R) Visual Studio 2008.
http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/
_______________________________________________
gumstix-users mailing list
gumstix-users@...
https://lists.sourceforge.net/lists/listinfo/gumstix-users

Re: OE BUILD

by peteb :: Rate this Message:

Reply to Author | View Threaded | Show Only this Message

Terry Kemp wrote:

> OE.dev has gmp-4.2.2 recipes and patches that fix this.
>
> to override angstrom (2007) preferred version (4.2.1) add in
> gumstix-oe/com.gumstix.collection/conf/machine/include...
>
> PREFERRED_VERSION_gmp = "4.2.2"
> PREFERRED_VERSION_gmp-native = "4.2.2"
>
>
> next the kernel will fail...
>
> scripts/mod/sumversion.c: In function 'get_src_version':
> | scripts/mod/sumversion.c:384: error: 'PATH_MAX' undeclared (first use
> in this function)
>
>
> in scripts/mod/sumversion.c you need to add
> #include <limits.h>
>
> This got me through bitbake gumstix-kernel (from a blown away tmp) OK.
>
> Let me know if you want the gmp 4.2.2 recipes and patches, although it
> would be better to get Steve to add them into svn from dev.
>
> Terry
>
> On Sun, 2008-05-25 at 13:20 -0500, Peter Buelow wrote:
>  
Interesting, good catch on that. Going to the next revision was my next
thought, but Qemu won out instead. Until all the patches are in that
make it work, I've got Qemu running Fedora 8 which lets me build
(slowly, but it works). I'll try this out as well to see what happens.

Whether it matters or not, with the same configure options, gmp 4.2.1
builds just fine outside of OE in Fedora 9.

--
Pete

Funny=Truth
Truth=Beauty
So take it as a complement when people say you look Funny.



-------------------------------------------------------------------------
This SF.net email is sponsored by: Microsoft
Defy all challenges. Microsoft(R) Visual Studio 2008.
http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/
_______________________________________________
gumstix-users mailing list
gumstix-users@...
https://lists.sourceforge.net/lists/listinfo/gumstix-users

Re: OE BUILD

by peteb :: Rate this Message:

Reply to Author | View Threaded | Show Only this Message

Terry Kemp wrote:

> OE.dev has gmp-4.2.2 recipes and patches that fix this.
>
> to override angstrom (2007) preferred version (4.2.1) add in
> gumstix-oe/com.gumstix.collection/conf/machine/include...
>
> PREFERRED_VERSION_gmp = "4.2.2"
> PREFERRED_VERSION_gmp-native = "4.2.2"
>
>
> next the kernel will fail...
>
> scripts/mod/sumversion.c: In function 'get_src_version':
> | scripts/mod/sumversion.c:384: error: 'PATH_MAX' undeclared (first use
> in this function)
>
>
> in scripts/mod/sumversion.c you need to add
> #include <limits.h>
>
> This got me through bitbake gumstix-kernel (from a blown away tmp) OK.
>
> Let me know if you want the gmp 4.2.2 recipes and patches, although it
> would be better to get Steve to add them into svn from dev.
>
> Terry
>
>  
Adding them to svn would be best, but I'm looking through OE.dev and
can't find them. I've gone back through May now searching the dev
tracker. Can you give me an approximate day to look for? Or just attach
them to an email?

Also, would it make sense to add the include change into svn as well?
Now that F9 is out, it's probably going to get more usage, and the
include change is a pretty big non-issue for other platforms (or should
be anyway).

--
Pete

Funny=Truth
Truth=Beauty
So take it as a complement when people say you look Funny.



-------------------------------------------------------------------------
This SF.net email is sponsored by: Microsoft
Defy all challenges. Microsoft(R) Visual Studio 2008.
http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/
_______________________________________________
gumstix-users mailing list
gumstix-users@...
https://lists.sourceforge.net/lists/listinfo/gumstix-users

Re: OE BUILD

by Terry Kemp-2 :: Rate this Message:

Reply to Author | View Threaded | Show Only this Message


On Tue, 2008-05-27 at 10:46 -0500, Peter Buelow wrote:

> >  
> Adding them to svn would be best, but I'm looking through OE.dev and
> can't find them. I've gone back through May now searching the dev
> tracker. Can you give me an approximate day to look for? Or just attach
> them to an email?
>
> Also, would it make sense to add the include change into svn as well?
> Now that F9 is out, it's probably going to get more usage, and the
> include change is a pretty big non-issue for other platforms (or should
> be anyway).
>
from
mtn log | grep -C 10 gmp | less

| | |   merge of '53dccbcb3344fa9a96f7f2c544d3dfec1b100fca'
| | |        and 'fea7dedc0a856ea56969779e97c888a16ffc63de'
| o |
-----------------------------------------------------------------
| | |   Revision: fea7dedc0a856ea56969779e97c888a16ffc63de
| | |   Ancestor: 1563949c5add46370d8282542dbdf5184048c725
| | |   Author: woglinde2@...
| | |   Date: 2008-05-07T19:57:32
| | |   Branch: org.openembedded.dev
| | |  
| | |   Added files:
| | |           packages/gmp/gmp-4.2.2/02_noexec-stack.diff
| | |           packages/gmp/gmp-4.2.2/03_gnu89-inline.diff
| | |           packages/gmp/gmp-4.2.2/amd64.patch
| | |           packages/gmp/gmp-4.2.2/configure.patch
| | |           packages/gmp/gmp-4.2.2/mpf_set_str_c.diff
| | |           packages/gmp/gmp-4.2.2/sh4-asmfix.patch
| | |           packages/gmp/gmp-4.2.2/x86-fat.diff
| | |           packages/gmp/gmp-native_4.2.2.bb
packages/gmp/gmp_4.2.2.bb
| | |   Added directories:
| | |           packages/gmp/gmp-4.2.2
| | |  
| | |   ChangeLog:
| | |  
| | |   gmp: update to version 4.2.2
| | |   * seems to work
| | |   * add patches 02_noexec-stack.diff and 03_gnu89-inline.diff from
debian
| | |   * add mpf_set_str_c.diff from the gmp-website
o | |
-----------------------------------------------------------------

attached files for 4.2.2, layout like this...
[tkemp@sulphur gmp]$ tree
.
|-- files
|   |-- sh4-asmfix.patch
|   `-- x86-fat.diff
|-- gmp-4.1.2
|   |-- amd64.patch
|   |-- configure.patch
|   `-- gcc-compile.patch
|-- gmp-4.1.4
|   |-- amd64.patch
|   `-- configure.patch
|-- gmp-4.2.1
|   |-- amd64.patch
|   `-- configure.patch
|-- gmp-4.2.2
|   |-- 02_noexec-stack.diff
|   |-- 03_gnu89-inline.diff
|   |-- amd64.patch
|   |-- configure.patch
|   `-- mpf_set_str_c.diff
|-- gmp-native.inc
|-- gmp-native_4.1.2.bb
|-- gmp-native_4.2.1.bb
|-- gmp-native_4.2.2.bb
|-- gmp.inc
|-- gmp_4.1.2.bb
|-- gmp_4.1.4.bb
|-- gmp_4.2.1.bb
`-- gmp_4.2.2.bb

5 directories, 23 files

Terry

PR = "r1"

SRC_URI_append += "file://02_noexec-stack.diff;patch=1 \
                   file://03_gnu89-inline.diff;patch=1 \
                   file://mpf_set_str_c.diff;patch=1 \
                   file://sh4-asmfix.patch;patch=1 \
                  "
require gmp.inc

PV = "4.2.2"
require gmp-native.inc

[02_noexec-stack.diff]
Index: acinclude.m4
===================================================================
RCS file: /cvsroot/pkg-gmp/gmp/acinclude.m4,v
retrieving revision 1.3
diff -u -b -B -r1.3 acinclude.m4
--- gmp-4.2.2.orig/acinclude.m4 14 Apr 2006 22:48:55 -0000 1.3
+++ gmp-4.2.2/acinclude.m4 15 Apr 2006 06:11:48 -0000
@@ -1613,6 +1613,34 @@
 ])
 
 
+dnl Checks whether the stack can be marked nonexecutable by passing an option
+dnl to the C-compiler when acting on .s files. Appends that option to ASFLAGS.
+dnl This macro is adapted from one found in GLIBC-2.3.5.
+AC_DEFUN([CL_AS_NOEXECSTACK],[
+dnl AC_REQUIRE([AC_PROG_CC]) GMP uses something else
+AC_CACHE_CHECK([whether assembler supports --noexecstack option],
+cl_cv_as_noexecstack, [dnl
+  cat > conftest.c <<EOF
+void foo() {}
+EOF
+  if AC_TRY_COMMAND([${CC} $CFLAGS $CPPFLAGS
+                     -S -o conftest.s conftest.c >/dev/null]) \
+     && grep -q .note.GNU-stack conftest.s \
+     && AC_TRY_COMMAND([${CC} $CFLAGS $CPPFLAGS -Wa,--noexecstack
+                       -c -o conftest.o conftest.s >/dev/null])
+  then
+    cl_cv_as_noexecstack=yes
+  else
+    cl_cv_as_noexecstack=no
+  fi
+  rm -f conftest*])
+  if test "$cl_cv_as_noexecstack" = yes; then
+    ASMFLAGS="$ASMFLAGS -Wa,--noexecstack"
+  fi
+  AC_SUBST(ASMFLAGS)
+])
+
+
 dnl  GMP_ASM_LABEL_SUFFIX
 dnl  --------------------
 dnl  : - is usual.
Index: configure.in
===================================================================
RCS file: /cvsroot/pkg-gmp/gmp/configure.in,v
retrieving revision 1.3
diff -u -b -B -r1.3 configure.in
--- gmp-4.2.2.orig/configure.in 14 Apr 2006 22:48:55 -0000 1.3
+++ gmp-4.2.2/configure.in 15 Apr 2006 06:11:49 -0000
@@ -2024,6 +2024,8 @@
 # Automake ansi2knr support.
 AM_C_PROTOTYPES
 
+CL_AS_NOEXECSTACK
+
 GMP_PROG_AR
 GMP_PROG_NM
 
Index: mpn/Makeasm.am
===================================================================
RCS file: /cvsroot/pkg-gmp/gmp/mpn/Makeasm.am,v
retrieving revision 1.5
diff -u -b -B -r1.5 Makeasm.am
--- gmp-4.2.2.orig/mpn/Makeasm.am 14 Apr 2006 22:48:59 -0000 1.5
+++ gmp-4.2.2/mpn/Makeasm.am 15 Apr 2006 06:11:49 -0000
@@ -24,7 +24,7 @@
 # COMPILE minus CC.
 #
 COMPILE_FLAGS = $(DEFS) $(DEFAULT_INCLUDES) $(INCLUDES) $(AM_CPPFLAGS) \
- $(CPPFLAGS) $(AM_CFLAGS) $(CFLAGS)
+ $(CPPFLAGS) $(AM_CFLAGS) $(CFLAGS) $(ASMFLAGS)
 
 # Flags used for preprocessing (in ansi2knr rules).
 #

[03_gnu89-inline.diff]
Index: gmp-h.in
===================================================================
RCS file: /cvsroot/pkg-gmp/gmp/gmp-h.in,v
retrieving revision 1.4
diff -u -b -B -r1.4 gmp-h.in
--- gmp-4.2.2.orig/gmp-h.in 15 Sep 2007 23:05:28 -0000 1.4
+++ gmp-4.2.2/gmp-h.in 9 Apr 2008 05:30:42 -0000
@@ -418,9 +418,13 @@
     for an inline too, so as to correctly specify "dllimport" on windows, in
     case the function is called rather than inlined.
     GCC 4.3 and above with -std=c99 or -std=gnu99 implements ISO C99
-    inline semantics, unless -fgnu89-inline is used.  */
+    inline semantics, unless -fgnu89-inline is used.
+
+   With GCC 4.2, `__GNUC_STDC_INLINE__' is never defined (because C99 inline
+   semantics are not supported), but a warning is issued in C99 mode if
+   `__gnu_inline__' is not used.  */
 #ifdef __GNUC__
-#ifdef __GNUC_STDC_INLINE__
+#if (defined __GNUC_STDC_INLINE__) || (__GNUC__ == 4 && __GNUC_MINOR__ == 2)
 #define __GMP_EXTERN_INLINE extern __inline__ __attribute__ ((__gnu_inline__))
 #else
 #define __GMP_EXTERN_INLINE      extern __inline__

[amd64.patch]
--- gmp-4.1.4/longlong.h.orig 2004-04-22 00:34:28.000000000 +0200
+++ gmp-4.1.4/longlong.h 2005-07-18 01:13:06.000000000 +0200
@@ -738,8 +738,10 @@
    count is only an int. */
 #define count_trailing_zeros(count, x) \
   do { \
+    UDItype __cbtmp; \
     ASSERT ((x) != 0); \
-    __asm__ ("bsfq %1,%q0" : "=r" (count) : "rm" ((UDItype)(x))); \
+    __asm__ ("bsfq %1,%0" : "=r" (__cbtmp) : "rm" ((UDItype)(x))); \
+    (count) = __cbtmp; \
   } while (0)
 #endif /* x86_64 */
 

[configure.patch]
 acinclude.m4 |   26 ++++++++++++++------------
 configure.in |   24 +++++++++---------------
 2 files changed, 23 insertions(+), 27 deletions(-)

Index: gmp-4.2.2/acinclude.m4
===================================================================
--- gmp-4.2.2.orig/acinclude.m4 2008-05-01 11:07:43.000000000 +0200
+++ gmp-4.2.2/acinclude.m4 2008-05-01 11:07:48.000000000 +0200
@@ -30,20 +30,20 @@
 dnl    conftest.exe - various DOS compilers
 
 
-define(IA64_PATTERN,
+define([IA64_PATTERN],
 [[ia64*-*-* | itanium-*-* | itanium2-*-*]])
 
 dnl  Need to be careful not to match m6811, m6812, m68hc11 and m68hc12, all
 dnl  of which config.sub accepts.  (Though none of which are likely to work
 dnl  with GMP.)
 dnl
-define(M68K_PATTERN,
+define([M68K_PATTERN],
 [[m68k-*-* | m68[0-9][0-9][0-9]-*-*]])
 
-define(POWERPC64_PATTERN,
+define([POWERPC64_PATTERN],
 [[powerpc64-*-* | powerpc64le-*-* | powerpc620-*-* | powerpc630-*-* | powerpc970-*-* | power[3-9]-*-*]])
 
-define(X86_PATTERN,
+define([X86_PATTERN],
 [[i?86*-*-* | k[5-8]*-*-* | pentium*-*-* | athlon-*-* | viac3*-*-*]])
 
 
@@ -62,7 +62,7 @@
 dnl      x86/k6      ->  k6
 dnl      x86/k6/mmx  ->  k6_mmx
 
-define(GMP_FAT_SUFFIX,
+define([GMP_FAT_SUFFIX],
 [[$1=`echo $2 | sed -e '/\//s:^[^/]*/::' -e 's:[\\/]:_:g'`]])
 
 
@@ -71,7 +71,7 @@
 dnl  Emit code to remove any occurance of ITEM from $LISTVAR.  ITEM can be a
 dnl  shell expression like $foo if desired.
 
-define(GMP_REMOVE_FROM_LIST,
+define([GMP_REMOVE_FROM_LIST],
 [remove_from_list_tmp=
 for remove_from_list_i in $[][$1]; do
   if test $remove_from_list_i = [$2]; then :;
@@ -87,12 +87,12 @@
 dnl  ----------------------
 dnl  Strip entries subdir from $path and $fat_path.
 
-define(GMP_STRIP_PATH,
+define([GMP_STRIP_PATH],
 [GMP_STRIP_PATH_VAR(path, [$1])
 GMP_STRIP_PATH_VAR(fat_path, [$1])
 ])
 
-define(GMP_STRIP_PATH_VAR,
+define([GMP_STRIP_PATH_VAR],
 [tmp_path=
 for i in $[][$1]; do
   case $i in
@@ -113,7 +113,7 @@
 dnl  Dummy values for __GMP_BITS_PER_MP_LIMB and GMP_LIMB_BITS are enough
 dnl  for all current configure-time uses of gmp.h.
 
-define(GMP_INCLUDE_GMP_H,
+define([GMP_INCLUDE_GMP_H],
 [[#define __GMP_WITHIN_CONFIGURE 1   /* ignore template stuff */
 #define GMP_NAIL_BITS $GMP_NAIL_BITS
 #define __GMP_BITS_PER_MP_LIMB 123 /* dummy for GMP_NUMB_BITS etc */
@@ -129,7 +129,7 @@
 dnl  FILE.  The regexps here aren't very rugged, but are enough for gmp.
 dnl  /dev/null as a parameter prevents a hang if $2 is accidentally omitted.
 
-define(GMP_HEADER_GETVAL,
+define([GMP_HEADER_GETVAL],
 [patsubst(patsubst(
 esyscmd([grep "^#define $1 " $2 /dev/null 2>/dev/null]),
 [^.*$1[ ]+],[]),
@@ -143,7 +143,7 @@
 dnl  autoconf time.  Two digits like 3.0 if patchlevel <= 0, or three digits
 dnl  like 3.0.1 if patchlevel > 0.
 
-define(GMP_VERSION,
+define([GMP_VERSION],
 [GMP_HEADER_GETVAL(__GNU_MP_VERSION,gmp-h.in)[]dnl
 .GMP_HEADER_GETVAL(__GNU_MP_VERSION_MINOR,gmp-h.in)[]dnl
 ifelse(m4_eval(GMP_HEADER_GETVAL(__GNU_MP_VERSION_PATCHLEVEL,gmp-h.in) > 0),1,
@@ -1489,7 +1489,9 @@
 echo ["define(<CONFIG_TOP_SRCDIR>,<\`$tmp'>)"] >>$gmp_tmpconfigm4
 
 # All CPUs use asm-defs.m4
-echo ["include][(CONFIG_TOP_SRCDIR\`/mpn/asm-defs.m4')"] >>$gmp_tmpconfigm4i
+echo -n ["include("] >>$gmp_tmpconfigm4i
+echo -n ["CONFIG_TOP_SRCDIR\`/mpn/asm-defs.m4'"] >>$gmp_tmpconfigm4i
+echo [")"] >>$gmp_tmpconfigm4i
 ])
 
 
Index: gmp-4.2.2/configure.in
===================================================================
--- gmp-4.2.2.orig/configure.in 2008-05-01 11:07:43.000000000 +0200
+++ gmp-4.2.2/configure.in 2008-05-01 11:07:48.000000000 +0200
@@ -29,12 +29,6 @@
 AC_PREREQ(2.59)
 AC_INIT(GNU MP, GMP_VERSION, gmp-bugs@..., gmp)
 AC_CONFIG_SRCDIR(gmp-impl.h)
-m4_pattern_forbid([^[ \t]*GMP_])
-m4_pattern_allow(GMP_LDFLAGS)
-m4_pattern_allow(GMP_LIMB_BITS)
-m4_pattern_allow(GMP_MPARAM_H_SUGGEST)
-m4_pattern_allow(GMP_NAIL_BITS)
-m4_pattern_allow(GMP_NUMB_BITS)
 
 # If --target is not used then $target_alias is empty, but if say
 # "./configure athlon-pc-freebsd3.5" is used, then all three of
@@ -275,7 +269,7 @@
 # After GMP specific searches and tests, the standard autoconf AC_PROG_CC is
 # called.  User selections of CC etc are respected.
 #
-# Care is taken not to use macros like AC_TRY_COMPILE during the GMP
+# Care is taken not to use macros like AC_COMPILE_IFELSE([AC_LANG_PROGRAM([[]], [[]])],[],[]) during the GMP
 # pre-testing, since they of course depend on AC_PROG_CC, and also some of
 # them cache their results, which is not wanted.
 #
@@ -367,7 +361,7 @@
 # FIXME: We'd like to prefer an ANSI compiler, perhaps by preferring
 # c89 over cc here.  But note that on HP-UX c89 provides a castrated
 # environment, and would want to be excluded somehow.  Maybe
-# AC_PROG_CC_STDC already does enough to stick cc into ANSI mode and
+#  already does enough to stick cc into ANSI mode and
 # we don't need to worry.
 #
 cclist="gcc cc"
@@ -1518,7 +1512,7 @@
 CFLAGS_or_unset=${CFLAGS-'(unset)'}
 CPPFLAGS_or_unset=${CPPFLAGS-'(unset)'}
 
-cat >&AC_FD_CC <<EOF
+cat >&AS_MESSAGE_LOG_FD() <<EOF
 User:
 ABI=$ABI
 CC=$CC
@@ -1889,7 +1883,6 @@
 
 # The C compiler and preprocessor, put into ANSI mode if possible.
 AC_PROG_CC
-AC_PROG_CC_STDC
 AC_PROG_CPP
 GMP_H_ANSI
 
@@ -1912,11 +1905,11 @@
 
 # The C++ compiler, if desired.
 want_cxx=no
+AC_PROG_CXX
 if test $enable_cxx != no; then
   test_CXXFLAGS=${CXXFLAGS+set}
-  AC_PROG_CXX
 
-  echo "CXXFLAGS chosen by autoconf: $CXXFLAGS" >&AC_FD_CC
+  echo "CXXFLAGS chosen by autoconf: $CXXFLAGS" >&AS_MESSAGE_LOG_FD()
   cxxflags_ac_prog_cxx=$CXXFLAGS
   cxxflags_list=ac_prog_cxx
 
@@ -2022,7 +2015,7 @@
 esac
 
 
-cat >&AC_FD_CC <<EOF
+cat >&AS_MESSAGE_LOG_FD() <<EOF
 Decided:
 ABI=$ABI
 CC=$CC
@@ -3199,7 +3192,7 @@
 # FIXME: Upcoming version of autoconf/automake may not like broken lines.
 #        Right now automake isn't accepting the new AC_CONFIG_FILES scheme.
 
-AC_OUTPUT(Makefile \
+AC_CONFIG_FILES([Makefile \
   mpbsd/Makefile mpf/Makefile mpn/Makefile mpq/Makefile \
   mpz/Makefile printf/Makefile scanf/Makefile cxx/Makefile \
   tests/Makefile tests/devel/Makefile tests/mpbsd/Makefile \
@@ -3208,4 +3201,5 @@
   tests/cxx/Makefile \
   doc/Makefile tune/Makefile \
   demos/Makefile demos/calc/Makefile demos/expr/Makefile \
-  gmp.h:gmp-h.in mp.h:mp-h.in)
+  gmp.h:gmp-h.in mp.h:mp-h.in])
+AC_OUTPUT
Index: gmp-4.2.2/gmp-h.in
===================================================================
--- gmp-4.2.2.orig/gmp-h.in 2007-09-03 18:05:40.000000000 +0200
+++ gmp-4.2.2/gmp-h.in 2008-05-01 11:10:42.000000000 +0200
@@ -516,6 +516,7 @@
 
 #if defined (__cplusplus)
 extern "C" {
+#include <cstdio>
 #ifdef _GMP_H_HAVE_FILE
 using std::FILE;
 #endif

[mpf_set_str_c.diff]
*** gmp-4.2.2.orig/mpf/set_str.c 30 Aug 2007 18:19:41 -0000
--- gmp-4.2.2/mpf/set_str.c 10 Dec 2007 04:47:18 -0000
*************** mpf_set_str (mpf_ptr x, const char *str,
*** 272,277 ****
 
      if (expptr != 0)
!       /* FIXME: Should do some error checking here.  */
!       exp_in_base = strtol (expptr, (char **) 0, exp_base);
      else
        exp_in_base = 0;
--- 272,298 ----
 
      if (expptr != 0)
!       {
! /* Scan and convert the exponent, in base exp_base.  */
! long dig, neg = -(long) ('-' == expptr[0]);
! expptr -= neg; /* conditional increment */
! c = (unsigned char) *expptr++;
! dig = digit_value[c];
! if (dig >= exp_base)
!  {
!    TMP_FREE;
!    return -1;
!  }
! exp_in_base = dig;
! c = (unsigned char) *expptr++;
! dig = digit_value[c];
! while (dig < exp_base)
!  {
!    exp_in_base = exp_in_base * exp_base;
!    exp_in_base += dig;
!    c = (unsigned char) *expptr++;
!    dig = digit_value[c];
!  }
! exp_in_base = (exp_in_base ^ neg) - neg; /* conditional negation */
!       }
      else
        exp_in_base = 0;

-------------------------------------------------------------------------
This SF.net email is sponsored by: Microsoft
Defy all challenges. Microsoft(R) Visual Studio 2008.
http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/
_______________________________________________
gumstix-users mailing list
gumstix-users@...
https://lists.sourceforge.net/lists/listinfo/gumstix-users

Re: OE BUILD

by Terry Kemp-2 :: Rate this Message:

Reply to Author | View Threaded | Show Only this Message

On Wed, 2008-05-28 at 08:40 +1200, Terry Kemp wrote:

> On Tue, 2008-05-27 at 10:46 -0500, Peter Buelow wrote:
>
> > >  
> > Adding them to svn would be best, but I'm looking through OE.dev and
> > can't find them. I've gone back through May now searching the dev
> > tracker. Can you give me an approximate day to look for? Or just attach
> > them to an email?
> >
> > Also, would it make sense to add the include change into svn as well?
> > Now that F9 is out, it's probably going to get more usage, and the
> > include change is a pretty big non-issue for other platforms (or should
> > be anyway).
> >

Now building gumstix-directfb-image there is a prob with dbus. this one
is a bit harder. The newer version from dev contains
$STAGING_DATADIR_NATIVE which is not defined (along with lots of other
changes) in the bitbake.conf file. I updated this file and bitbake which
gets me further but dbus-glib is using autoreconf from the host system
(bad bad bad) so more work to do here.
Might start another thread called Fedora 9 issues soon.

Terry


-------------------------------------------------------------------------
This SF.net email is sponsored by: Microsoft
Defy all challenges. Microsoft(R) Visual Studio 2008.
http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/
_______________________________________________
gumstix-users mailing list
gumstix-users@...
https://lists.sourceforge.net/lists/listinfo/gumstix-users

Re: OE BUILD

by peteb :: Rate this Message:

Reply to Author | View Threaded | Show Only this Message

I was considering a Fedora 9 wiki entry. I'm at the dbus stage as well. Similar issues, but no resolution. Looks like it's having trouble with some of the #ifdef's around the code. Doesn't seem to be picking up every header quite right.

Terry Kemp wrote: 
On Wed, 2008-05-28 at 08:40 +1200, Terry Kemp wrote:
  
On Tue, 2008-05-27 at 10:46 -0500, Peter Buelow wrote:

    
  
        
Adding them to svn would be best, but I'm looking through OE.dev and 
can't find them. I've gone back through May now searching the dev 
tracker. Can you give me an approximate day to look for? Or just attach 
them to an email?

Also, would it make sense to add the include change into svn as well? 
Now that F9 is out, it's probably going to get more usage, and the 
include change is a pretty big non-issue for other platforms (or should 
be anyway).

      

Now building gumstix-directfb-image there is a prob with dbus. this one
is a bit harder. The newer version from dev contains
$STAGING_DATADIR_NATIVE which is not defined (along with lots of other
changes) in the bitbake.conf file. I updated this file and bitbake which
gets me further but dbus-glib is using autoreconf from the host system
(bad bad bad) so more work to do here.
Might start another thread called Fedora 9 issues soon.

Terry 


-------------------------------------------------------------------------
This SF.net email is sponsored by: Microsoft
Defy all challenges. Microsoft(R) Visual Studio 2008.
http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/
_______________________________________________
gumstix-users mailing list
gumstix-users@...
https://lists.sourceforge.net/lists/listinfo/gumstix-users
  


-- 
Pete

Funny=Truth
Truth=Beauty
So take it as a complement when people say you look Funny.

-------------------------------------------------------------------------
This SF.net email is sponsored by: Microsoft
Defy all challenges. Microsoft(R) Visual Studio 2008.
http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/
_______________________________________________
gumstix-users mailing list
gumstix-users@...
https://lists.sourceforge.net/lists/listinfo/gumstix-users
Free embeddable forum powered by Nabble Forum Help