Skip to contentSkip to conversation

Results of random configuration builds

Buildroot Thomas Petazzoni 4 participants 11 messages Nov 6, 2010

lore
Share & tools
lore
#0 lore raw

Hello,

I've run a few random build tests yesterday, and analyzed the results a bit today. I've set up a small webpage that gathers the results, together with the logfile and the configuration for each test.

See http://free-electrons.com/~thomas/pub/buildroot-logs/

Most of the builds are failing. The worst problem is that libglib2 does not build in many situations (see
http://free-electrons.com/~thomas/pub/buildroot-logs/test-195/logfile-end).
There are also problems with logrotate, gvfs, libgcrypt, olsr, libmad, netperf, magiccube4d, libmpeg2, xfsprogs, imagemagick, squid, ltrace, tcpdump, bash, gmp, gdb.

It'd be great if some of the readers of this list could have a look at one or several of these build problems, so that we can collaboratively solve them.

Thanks!

Thomas

Signature · 4 lines
Thomas Petazzoni, Free Electrons
Kernel, drivers, real-time and embedded Linux
development, consulting, training and support.
http://free-electrons.com
#1 lore raw
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

"Thomas" == Thomas Petazzoni <[email protected]> writes:

Thomas> Hello,
Thomas> I've run a few random build tests yesterday, and analyzed the results a Thomas> bit today. I've set up a small webpage that gathers the results, Thomas> together with the logfile and the configuration for each test.

Thomas> See http://free-electrons.com/~thomas/pub/buildroot-logs/

Thanks, I also have a bunch of build logs on gcc10.fsffrance.org

Thomas> Most of the builds are failing. The worst problem is that libglib2 does Thomas> not build in many situations (see
Thomas> http://free-electrons.com/~thomas/pub/buildroot-logs/test-195/logfile-end).
Thomas> There are also problems with logrotate, gvfs, libgcrypt, olsr, libmad, Thomas> netperf, magiccube4d, libmpeg2, xfsprogs, imagemagick, squid, ltrace, Thomas> tcpdump, bash, gmp, gdb.

Thomas> It'd be great if some of the readers of this list could have a look at Thomas> one or several of these build problems, so that we can collaboratively Thomas> solve them.

I've noticed the tcpdump issue as well, and I'm fixing that now.

Signature · 1 line
Bye, Peter Korsgaard
#2 lore raw
Quoted 1 line · On Sat, 06 Nov 2010 13:48:00 +0100 Peter Korsgaard <[email protected]> wrote:

Thanks, I also have a bunch of build logs on gcc10.fsffrance.org

Are they publicly available / visible ?

Quoted 1 line

I've noticed the tcpdump issue as well, and I'm fixing that now.

Thanks, I've updated
http://free-electrons.com/~thomas/pub/buildroot-logs/ accordingly.

Thomas

Signature · 4 lines
Thomas Petazzoni, Free Electrons
Kernel, drivers, real-time and embedded Linux
development, consulting, training and support.
http://free-electrons.com
#3 lore raw
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

"Thomas" == Thomas Petazzoni <[email protected]> writes:

Thomas> On Sat, 06 Nov 2010 13:48:00 +0100

Quoted 1 line · Thomas> Peter Korsgaard <[email protected]> wrote:
Quoted 1 line

Thanks, I also have a bunch of build logs on gcc10.fsffrance.org

Thomas> Are they publicly available / visible ?

Sorry, so far only if you have a login to the gcc cluster (~peko/logs). Could you share your web scripts so I can put something online?

Quoted 1 line
Quoted 1 line

I've noticed the tcpdump issue as well, and I'm fixing that now.

Thomas> Thanks, I've updated
Thomas> http://free-electrons.com/~thomas/pub/buildroot-logs/ accordingly.

I also fixed the tslib issue yesterday.

Signature · 1 line
Bye, Peter Korsgaard
#4 lore raw

Le samedi 06 novembre 2010 ? 13:59 +0100, Peter Korsgaard a ?crit :

Quoted 5 lines
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

"Thomas" == Thomas Petazzoni <[email protected]> writes:

I also fixed the tslib issue yesterday.

I posted a patch for the directfb issue on sqrt yesterday too.

#5 lore raw
In reply to Thomas Petazzoni

Thomas,

It seems you've strange toolchains :)
It seems the glib breakage is due to a compiler not supporting builtin atomic operations, which is strange for x86.
I'm not able to reproduce with internal or external glibc/sh4 toolchains.

Can test with the attached patch to apply on glib2 ?

Regards,

--
Lionel Landwerlin

Le samedi 06 novembre 2010 ? 13:44 +0100, Thomas Petazzoni a ?crit :

Quoted 15 lines

Hello,

I've run a few random build tests yesterday, and analyzed the results a bit today. I've set up a small webpage that gathers the results, together with the logfile and the configuration for each test.

See http://free-electrons.com/~thomas/pub/buildroot-logs/

Most of the builds are failing. The worst problem is that libglib2 does not build in many situations (see
http://free-electrons.com/~thomas/pub/buildroot-logs/test-195/logfile-end).
There are also problems with logrotate, gvfs, libgcrypt, olsr, libmad, netperf, magiccube4d, libmpeg2, xfsprogs, imagemagick, squid, ltrace, tcpdump, bash, gmp, gdb.

Quoted 7 lines

It'd be great if some of the readers of this list could have a look at one or several of these build problems, so that we can collaboratively solve them.

Thanks!

Thomas

-------------- next part --------------
A non-text attachment was scrubbed...
Name: 0001-glib-fix-compilation-with-no-builtin-atomic-operatio.patch
Type: text/x-patch
Size: 797 bytes
Desc: not available
URL: <http://lists.busybox.net/pipermail/buildroot/attachments/20101106/dd5ae904/attachment.bin>

#8 lore raw
In reply to Thomas Petazzoni
Quoted 6 lines · Thomas Petazzoni

Hello,

I've run a few random build tests yesterday, and analyzed the results a bit today. I've set up a small webpage that gathers the results, together with the logfile and the configuration for each test.

Quoted 8 lines

See http://free-electrons.com/~thomas/pub/buildroot-logs/

Most of the builds are failing. The worst problem is that libglib2 does not build in many situations (see
http://free-electrons.com/~thomas/pub/buildroot-logs/test-195/logfile-end).
There are also problems with logrotate, gvfs, libgcrypt, olsr, libmad, netperf, magiccube4d, libmpeg2, xfsprogs, imagemagick, squid, ltrace, tcpdump, bash, gmp, gdb.

I've noticed ltrace breaks quite frequently because it is built from the debian archive, which doesn't guarantee that the package will remain available forever. I wonder whether we should be building anything from the debian archive directly and build from archive.debian.net instead.

#6 lore raw
In reply to Lionel Landwerlin

Hello,

Quoted 5 lines · On Sat, 06 Nov 2010 15:05:42 +0100 Lionel Landwerlin <[email protected]>…

It seems you've strange toolchains :)
It seems the glib breakage is due to a compiler not supporting builtin atomic operations, which is strange for x86.
I'm not able to reproduce with internal or external glibc/sh4 toolchains.

I was building for i386, and on i386 the __sync_bool_compare_and_swap_4() is not available.

Quoted 1 line

Can test with the attached patch to apply on glib2 ?

I confirm it fixes the problem on i386 toolchains. I haven't tested with my sh4 toolchain, but since the problem is the same, I suspect it also fixed the problem.

Thanks!

Thomas

Signature · 4 lines
Thomas Petazzoni, Free Electrons
Kernel, drivers, real-time and embedded Linux
development, consulting, training and support.
http://free-electrons.com
#9 lore raw
In reply to Will Newton
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

"Will" == Will Newton <[email protected]> writes:

Hi,

Will> I've noticed ltrace breaks quite frequently because it is built from Will> the debian archive, which doesn't guarantee that the package will Will> remain available forever. I wonder whether we should be building Will> anything from the debian archive directly and build from Will> archive.debian.net instead.

The archive is normally preferred, but a missing upstream shouldn't matter much with the backup mirror (sources.buildroot.net). Is that somehow not working for you?

Signature · 1 line
Bye, Peter Korsgaard
#10 lore raw
Quoted 10 lines · Peter Korsgaard
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

"Will" == Will Newton <[email protected]> writes:

Hi,

?Will> I've noticed ltrace breaks quite frequently because it is built from ?Will> the debian archive, which doesn't guarantee that the package will ?Will> remain available forever. I wonder whether we should be building ?Will> anything from the debian archive directly and build from ?Will> archive.debian.net instead.

Quoted 3 lines

The archive is normally preferred, but a missing upstream shouldn't matter much with the backup mirror (sources.buildroot.net). Is that somehow not working for you?

I just tested it and it works for me now (i386), so I'm not sure what the problem was that I saw previously.

#7 lore raw
In reply to Thomas Petazzoni

Le samedi 06 novembre 2010 ? 17:06 +0100, Thomas Petazzoni a ?crit :

Quoted 14 lines

Hello,

Quoted 5 lines · On Sat, 06 Nov 2010 15:05:42 +0100 Lionel Landwerlin <[email protected]>…

It seems you've strange toolchains :)
It seems the glib breakage is due to a compiler not supporting builtin atomic operations, which is strange for x86.
I'm not able to reproduce with internal or external glibc/sh4 toolchains.

I was building for i386, and on i386 the __sync_bool_compare_and_swap_4() is not available.

Quoted 7 lines
Quoted 1 line

Can test with the attached patch to apply on glib2 ?

I confirm it fixes the problem on i386 toolchains. I haven't tested with my sh4 toolchain, but since the problem is the same, I suspect it also fixed the problem.

Thanks!

I just found it has been reported on the glib bugzilla :

https://bugzilla.gnome.org/show_bug.cgi?id=631353

Regards,

--
Lionel Landwerlin