Patchwork strace: Version bump to 4.7

login
register
mail settings
Submitter Markos Chandras
Date Sept. 28, 2012, 8:43 a.m.
Message ID <1348821796-28798-1-git-send-email-markos.chandras@gmail.com>
Download mbox | patch
Permalink /patch/187736/
State Accepted
Headers show

Comments

Markos Chandras - Sept. 28, 2012, 8:43 a.m.
From: Markos Chandras <markos.chandras@imgtec.com>

This also fixes a compilation problem with kernel headers 3.5
http://autobuild.buildroot.net/results/bb66a3a06d26f558e1c4c0593bb68e7af1d82398/build-end.log

Signed-off-by: Markos Chandras <markos.chandras@imgtec.com>
---
 package/strace/strace.mk |    4 ++--
 1 files changed, 2 insertions(+), 2 deletions(-)
Peter Korsgaard - Oct. 4, 2012, 8:48 p.m.
>>>>>   <markos.chandras@gmail.com> writes:

 > From: Markos Chandras <markos.chandras@imgtec.com>
 > This also fixes a compilation problem with kernel headers 3.5
 > http://autobuild.buildroot.net/results/bb66a3a06d26f558e1c4c0593bb68e7af1d82398/build-end.log

Committed, thanks.
Peter Korsgaard - Oct. 5, 2012, 5:35 a.m.
>>>>> "Peter" == Peter Korsgaard <jacmet@uclibc.org> writes:

>>>>>   <markos.chandras@gmail.com> writes:
 >> From: Markos Chandras <markos.chandras@imgtec.com>
 >> This also fixes a compilation problem with kernel headers 3.5
 >> http://autobuild.buildroot.net/results/bb66a3a06d26f558e1c4c0593bb68e7af1d82398/build-end.log

 Peter> Committed, thanks.

It unfortunately seems to be broken on largefile configurations:

http://autobuild.buildroot.net/results/db48ce61cdb53669958b8b301fd24d7ecc33b236/build-end.log

Any idea how to fix?
Markos Chandras - Oct. 5, 2012, 8:20 a.m.
On Fri, Oct 5, 2012 at 6:35 AM, Peter Korsgaard <jacmet@sunsite.dk> wrote:
>>>>>> "Peter" == Peter Korsgaard <jacmet@uclibc.org> writes:
>
>>>>>>   <markos.chandras@gmail.com> writes:
>  >> From: Markos Chandras <markos.chandras@imgtec.com>
>  >> This also fixes a compilation problem with kernel headers 3.5
>  >> http://autobuild.buildroot.net/results/bb66a3a06d26f558e1c4c0593bb68e7af1d82398/build-end.log
>
>  Peter> Committed, thanks.
>
> It unfortunately seems to be broken on largefile configurations:
>
> http://autobuild.buildroot.net/results/db48ce61cdb53669958b8b301fd24d7ecc33b236/build-end.log
>
> Any idea how to fix?
>
> --
> Bye, Peter Korsgaard

Hi Peter,

I will have a look and I will let you know.
Markos Chandras - Oct. 5, 2012, 1:20 p.m.
On Fri, Oct 5, 2012 at 9:20 AM, Markos Chandras
<markos.chandras@gmail.com> wrote:
> On Fri, Oct 5, 2012 at 6:35 AM, Peter Korsgaard <jacmet@sunsite.dk> wrote:
>>>>>>> "Peter" == Peter Korsgaard <jacmet@uclibc.org> writes:
>>
>>>>>>>   <markos.chandras@gmail.com> writes:
>>  >> From: Markos Chandras <markos.chandras@imgtec.com>
>>  >> This also fixes a compilation problem with kernel headers 3.5
>>  >> http://autobuild.buildroot.net/results/bb66a3a06d26f558e1c4c0593bb68e7af1d82398/build-end.log
>>
>>  Peter> Committed, thanks.
>>
>> It unfortunately seems to be broken on largefile configurations:
>>
>> http://autobuild.buildroot.net/results/db48ce61cdb53669958b8b301fd24d7ecc33b236/build-end.log
>>
>> Any idea how to fix?
>>
>> --
>> Bye, Peter Korsgaard
>
> Hi Peter,
>
> I will have a look and I will let you know.
>
> --
> Regards,
> Markos

Hi Peter,

This also happens with the latest strace git code. It seems like
-D_FILE_OFFSET_BITS=64 is causing it. But there are also some other
problems in the LFS support. I will see if I can write some patches
and send them upstream.
Peter Korsgaard - Oct. 5, 2012, 1:58 p.m.
>>>>> "Markos" == Markos Chandras <markos.chandras@gmail.com> writes:

 Markos> Hi Peter,

 Markos> This also happens with the latest strace git code. It seems like
 Markos> -D_FILE_OFFSET_BITS=64 is causing it. But there are also some other
 Markos> problems in the LFS support. I will see if I can write some patches
 Markos> and send them upstream.

Great, thanks!
Arnout Vandecappelle - Oct. 5, 2012, 11:18 p.m.
On 05/10/12 15:58, Peter Korsgaard wrote:
>>>>>> "Markos" == Markos Chandras<markos.chandras@gmail.com>  writes:
>
>   Markos>  Hi Peter,
>
>   Markos>  This also happens with the latest strace git code. It seems like
>   Markos>  -D_FILE_OFFSET_BITS=64 is causing it. But there are also some other
>   Markos>  problems in the LFS support. I will see if I can write some patches
>   Markos>  and send them upstream.
>
> Great, thanks!

  Not sure if it is related, but the previous version was already broken for me
on x86_64 builds. Hm, probably not related, because it failed with:

file.c: In function 'printstat64':
file.c:1073:16: error: storage size of 'statbuf' isn't known
file.c:1073:16: warning: unused variable 'statbuf' [-Wunused-variable]


  Regards,
  Arnout
Markos Chandras - Oct. 10, 2012, 12:49 p.m.
On Sat, Oct 6, 2012 at 12:18 AM, Arnout Vandecappelle <arnout@mind.be> wrote:
> On 05/10/12 15:58, Peter Korsgaard wrote:
>>>>>>>
>>>>>>> "Markos" == Markos Chandras<markos.chandras@gmail.com>  writes:
>>
>>
>>   Markos>  Hi Peter,
>>
>>   Markos>  This also happens with the latest strace git code. It seems
>> like
>>   Markos>  -D_FILE_OFFSET_BITS=64 is causing it. But there are also some
>> other
>>   Markos>  problems in the LFS support. I will see if I can write some
>> patches
>>   Markos>  and send them upstream.
>>
>> Great, thanks!
>
>
>  Not sure if it is related, but the previous version was already broken for
> me
> on x86_64 builds. Hm, probably not related, because it failed with:
>
> file.c: In function 'printstat64':
> file.c:1073:16: error: storage size of 'statbuf' isn't known
> file.c:1073:16: warning: unused variable 'statbuf' [-Wunused-variable]
>
>
>  Regards,
>  Arnout
> --
> Arnout Vandecappelle                               arnout at mind be
> Senior Embedded Software Architect                 +32-16-286540
> Essensium/Mind                                     http://www.mind.be
> G.Geenslaan 9, 3001 Leuven, Belgium                BE 872 984 063 RPR Leuven
> LinkedIn profile: http://www.linkedin.com/in/arnoutvandecappelle
> GPG fingerprint:  7CB5 E4CC 6C2E EFD4 6E3D A754 F963 ECAB 2450 2F1F

Hi Arnout,

I reported this problem upstream about an hour ago (with a proposed
patch included)

http://sourceforge.net/mailarchive/message.php?msg_id=29947462
Arnout Vandecappelle - Oct. 11, 2012, 9:26 p.m.
On 10/10/12 14:49, Markos Chandras wrote:
> On Sat, Oct 6, 2012 at 12:18 AM, Arnout Vandecappelle<arnout@mind.be>  wrote:
>>   Not sure if it is related, but the previous version was already broken for me
>> on x86_64 builds. Hm, probably not related, because it failed with:
>>
>> file.c: In function 'printstat64':
>> file.c:1073:16: error: storage size of 'statbuf' isn't known
>> file.c:1073:16: warning: unused variable 'statbuf' [-Wunused-variable]
>>

> I reported this problem upstream about an hour ago (with a proposed
> patch included)
>
> http://sourceforge.net/mailarchive/message.php?msg_id=29947462

  Is this only on x86_64 or also on other 64-bit platforms?

  Regards,
  Arnout
Markos Chandras - Oct. 12, 2012, 8:04 a.m.
On Thu, Oct 11, 2012 at 10:26 PM, Arnout Vandecappelle <arnout@mind.be> wrote:
> On 10/10/12 14:49, Markos Chandras wrote:
>>
>> On Sat, Oct 6, 2012 at 12:18 AM, Arnout Vandecappelle<arnout@mind.be>
>> wrote:
>>>
>>>   Not sure if it is related, but the previous version was already broken
>>> for me
>>>
>>> on x86_64 builds. Hm, probably not related, because it failed with:
>>>
>>> file.c: In function 'printstat64':
>>> file.c:1073:16: error: storage size of 'statbuf' isn't known
>>> file.c:1073:16: warning: unused variable 'statbuf' [-Wunused-variable]
>>>
>
>> I reported this problem upstream about an hour ago (with a proposed
>> patch included)
>>
>> http://sourceforge.net/mailarchive/message.php?msg_id=29947462
>
>
>  Is this only on x86_64 or also on other 64-bit platforms?
>
>
I haven't tested it with other 64-bit platforms but I think the rest
of them provide a struct stat64. But even with this patch, it does not
go much further as it fails to build again due to undefined syscalls
like Peter posted before

http://autobuild.buildroot.net/results/db48ce61cdb53669958b8b301fd24d7ecc33b236/build-end.log

Another way to workaround the stat64 problem on x86_64 is to
ac_cv_have_long_long_off_t=yes on strace.mk when
BR2_LARGEFILE is enabled.
Alexander Khryukin - Oct. 12, 2012, 12:14 p.m.
2012/10/12 Markos Chandras <markos.chandras@gmail.com>

> On Thu, Oct 11, 2012 at 10:26 PM, Arnout Vandecappelle <arnout@mind.be>
> wrote:
> > On 10/10/12 14:49, Markos Chandras wrote:
> >>
> >> On Sat, Oct 6, 2012 at 12:18 AM, Arnout Vandecappelle<arnout@mind.be>
> >> wrote:
> >>>
> >>>   Not sure if it is related, but the previous version was already
> broken
> >>> for me
> >>>
> >>> on x86_64 builds. Hm, probably not related, because it failed with:
> >>>
> >>> file.c: In function 'printstat64':
> >>> file.c:1073:16: error: storage size of 'statbuf' isn't known
> >>> file.c:1073:16: warning: unused variable 'statbuf' [-Wunused-variable]
> >>>
> >
> >> I reported this problem upstream about an hour ago (with a proposed
> >> patch included)
> >>
> >> http://sourceforge.net/mailarchive/message.php?msg_id=29947462
> >
> >
> >  Is this only on x86_64 or also on other 64-bit platforms?
> >
> >
> I haven't tested it with other 64-bit platforms but I think the rest
> of them provide a struct stat64. But even with this patch, it does not
> go much further as it fails to build again due to undefined syscalls
> like Peter posted before
>
>
> http://autobuild.buildroot.net/results/db48ce61cdb53669958b8b301fd24d7ecc33b236/build-end.log
>
> Another way to workaround the stat64 problem on x86_64 is to
> ac_cv_have_long_long_off_t=yes on strace.mk when
> BR2_LARGEFILE is enabled.
>
> --
> Regards,
> Markos
> _______________________________________________
> buildroot mailing list
> buildroot@busybox.net
> http://lists.busybox.net/mailman/listinfo/buildroot
>


Hi!

i'm trying to buiid strace

See here http://pastie.org/5043151




syscall.o:(.rodata+0x128): undefined reference to `sys_oldstat'
syscall.o:(.rodata+0x1c8): undefined reference to `sys_oldfstat'
syscall.o:(.rodata+0x548): undefined reference to `sys_oldlstat'
syscall.o:(.rodata+0x5a8): undefined reference to `sys_old_mmap'
syscall.o:(.rodata+0xc18): undefined reference to `sys_truncate64'
syscall.o:(.rodata+0xc28): undefined reference to `sys_ftruncate64'
syscall.o:(.rodata+0xc38): undefined reference to `sys_stat64'
syscall.o:(.rodata+0xc48): undefined reference to `sys_lstat64'
syscall.o:(.rodata+0xc58): undefined reference to `sys_fstat64'
Peter Korsgaard - Oct. 16, 2012, 7:57 p.m.
>>>>> "Markos" == Markos Chandras <markos.chandras@gmail.com> writes:
 Markos> Hi Peter,

 Markos> This also happens with the latest strace git code. It seems like
 Markos> -D_FILE_OFFSET_BITS=64 is causing it. But there are also some other
 Markos> problems in the LFS support. I will see if I can write some patches
 Markos> and send them upstream.

 Peter> Great, thanks!

Any progress on this? All the strace failures on the autobuilders are
not so nice. Alternatively I can revert until we have a fix.
Markos Chandras - Oct. 16, 2012, 8:48 p.m.
On Tue, Oct 16, 2012 at 8:57 PM, Peter Korsgaard <jacmet@sunsite.dk> wrote:
>>>>>> "Markos" == Markos Chandras <markos.chandras@gmail.com> writes:
>  Markos> Hi Peter,
>
>  Markos> This also happens with the latest strace git code. It seems like
>  Markos> -D_FILE_OFFSET_BITS=64 is causing it. But there are also some other
>  Markos> problems in the LFS support. I will see if I can write some patches
>  Markos> and send them upstream.
>
>  Peter> Great, thanks!
>
> Any progress on this? All the strace failures on the autobuilders are
> not so nice. Alternatively I can revert until we have a fix.
>
> --
> Bye, Peter Korsgaard

Hi Peter,

I haven't managed to find a solution yet and I think all the LFS
problems are buildroot specific.
I don't have a native LFS toolchain so I can try an out-of-buildroot build.
You can either revert the commit or disable the LFS support in the
package for now.
Peter Korsgaard - Oct. 16, 2012, 9:12 p.m.
>>>>> "Markos" == Markos Chandras <markos.chandras@gmail.com> writes:

Hi,

 Markos> I haven't managed to find a solution yet and I think all the
 Markos> LFS problems are buildroot specific.  I don't have a native LFS
 Markos> toolchain so I can try an out-of-buildroot build.  You can
 Markos> either revert the commit or disable the LFS support in the
 Markos> package for now.

ok, thanks.

I'll try to take another look, and revert if I don't see an easy
solution.
Peter Korsgaard - Oct. 29, 2012, 8:55 p.m.
>>>>> "Peter" == Peter Korsgaard <jacmet@sunsite.dk> writes:

Hi,

 Markos> I haven't managed to find a solution yet and I think all the
 Markos> LFS problems are buildroot specific.  I don't have a native LFS
 Markos> toolchain so I can try an out-of-buildroot build.  You can
 Markos> either revert the commit or disable the LFS support in the
 Markos> package for now.

 Peter> ok, thanks.

 Peter> I'll try to take another look, and revert if I don't see an easy
 Peter> solution.

I luckily found a solution (23f6455bff), which I've applied.
Markos Chandras - Oct. 30, 2012, 10:07 a.m.
On Mon, Oct 29, 2012 at 8:55 PM, Peter Korsgaard <jacmet@sunsite.dk> wrote:
>>>>>> "Peter" == Peter Korsgaard <jacmet@sunsite.dk> writes:
>
> Hi,
>
>  Markos> I haven't managed to find a solution yet and I think all the
>  Markos> LFS problems are buildroot specific.  I don't have a native LFS
>  Markos> toolchain so I can try an out-of-buildroot build.  You can
>  Markos> either revert the commit or disable the LFS support in the
>  Markos> package for now.
>
>  Peter> ok, thanks.
>
>  Peter> I'll try to take another look, and revert if I don't see an easy
>  Peter> solution.
>
> I luckily found a solution (23f6455bff), which I've applied.
>
> --
> Bye, Peter Korsgaard

Hi Peter,

Hmm I thought that -D_FILE_OFFSET_BITS=64 was required when building
with LFS. I guess I was wrong.
Thanks for fixing it.

Patch

diff --git a/package/strace/strace.mk b/package/strace/strace.mk
index 2d95137..5c2d33c 100644
--- a/package/strace/strace.mk
+++ b/package/strace/strace.mk
@@ -4,8 +4,8 @@ 
 #
 #############################################################
 
-STRACE_VERSION = 4.5.20
-STRACE_SOURCE = strace-$(STRACE_VERSION).tar.bz2
+STRACE_VERSION = 4.7
+STRACE_SOURCE = strace-$(STRACE_VERSION).tar.xz
 STRACE_SITE = http://downloads.sourceforge.net/project/strace/strace/$(STRACE_VERSION)
 
 STRACE_CONF_ENV = ac_cv_header_linux_if_packet_h=yes \