linaro-image-lng 4.8 build fails

[Zi Shen Lim]

https://ci.linaro.org/jenkins/job/openembedded-armv7a-rootfs/gcc_version=4.8,label=oe_persistent_cloud,rootfs=lng/

Failed > Console Output #104 Sep 26, 2013 12:14:33 AM
Failed > Console Output #103 Sep 25, 2013 12:14:20 AM
Failed > Console Output #102 Sep 24, 2013 12:38:38 AM
Failed > Console Output #101 Sep 23, 2013 12:13:38 AM

[Mike Holmes (mike-holmes)]

The LNG CI loop takes the last successful build, so the CI loops continue to work. I believe a race in bit-bake causes it to fail.

[Zi Shen Lim (zlim)]

I poked around a little. Looking at the log, it points to build failure in gcc/libatomic:
--8<---
09:41:57 | libtool: compile: arm-oe-linux-gnueabi-gcc -march=armv7-a -marm -mthumb-interwork -mfloat-abi=hard -mfpu=neon --sysroot=/mnt/ci_build/workspace/tmp/sysroots/genericarmv7a -march=armv7-a -marm -mthumb-interwork -mfloat-abi=hard -mfpu=neon -isystem/mnt/ci_build/workspace/tmp/sysroots/genericarmv7a/usr/include -I/mnt/ci_build/workspace/tmp/work/armv7ahf-vfp-neon-oe-linux-gnueabi/gcc/linaro-4.8-r2013.09/gcc-linaro-4.8-2013.09/build.arm-oe-linux-gnueabi.arm-oe-linux-gnueabi/gcc/include/ --sysroot=/mnt/ci_build/workspace/tmp/sysroots/genericarmv7a -DHAVE_CONFIG_H -I/mnt/ci_build/workspace/tmp/work-shared/gcc-linaro-4.8-r2013.09/gcc-linaro-4.8-2013.09/libatomic/config/arm -I/mnt/ci_build/workspace/tmp/work-shared/gcc-linaro-4.8-r2013.09/gcc-linaro-4.8-2013.09/libatomic/config/linux/arm -I/mnt/ci_build/workspace/tmp/work-shared/gcc-linaro-4.8-r2013.09/gcc-linaro-4.8-2013.09/libatomic/config/posix -I/mnt/ci_build/workspace/tmp/work-shared/gcc-linaro-4.8-r2013.09/gcc-linaro-4.8-2013.09/libatomic -I. -Wall -Werror -pthread -g -Os -MT load_1_.lo -MD -MP -MF .deps/load_1_.lo.Ppo -DN=1 -c /mnt/ci_build/workspace/tmp/work-shared/gcc-linaro-4.8-r2013.09/gcc-linaro-4.8-2013.09/libatomic/config/linux/arm/load_n.c -fPIC -DPIC -o .libs/load_1_.o
09:41:57 | /mnt/ci_build/workspace/tmp/work-shared/gcc-linaro-4.8-r2013.09/gcc-linaro-4.8-2013.09/libatomic/config/linux/arm/load_n.c:49:0: fatal error: opening dependency file .deps/load_1_.lo.Ppo: No such file or directory
09:41:57 | #include "../../../load_n.c"
09:41:57 | ^
09:41:57 | compilation terminated.
09:41:57 | make[4]: *** [load_1_.lo] Error 1
09:41:57 | make[4]: *** Waiting for unfinished jobs....
-->8---

As a quick test, I created a .deps directory, and now 'bitbake gcc' goes through just fine :)
The failure we're hitting doesn't seem related to "a race in bitbake".

[Riku Voipio (riku-voipio)]

Yes, gcc build is being racy - I haven't managed to debug what exactly
causes the race.

On 27 September 2013 20:16, Zi Shen Lim <email address hidden> wrote:
> I poked around a little. Looking at the log, it points to build failure in gcc/libatomic:
> --8<---
> 09:41:57 | libtool: compile: arm-oe-linux-gnueabi-gcc -march=armv7-a -marm -mthumb-interwork -mfloat-abi=hard -mfpu=neon --sysroot=/mnt/ci_build/workspace/tmp/sysroots/genericarmv7a -march=armv7-a -marm -mthumb-interwork -mfloat-abi=hard -mfpu=neon -isystem/mnt/ci_build/workspace/tmp/sysroots/genericarmv7a/usr/include -I/mnt/ci_build/workspace/tmp/work/armv7ahf-vfp-neon-oe-linux-gnueabi/gcc/linaro-4.8-r2013.09/gcc-linaro-4.8-2013.09/build.arm-oe-linux-gnueabi.arm-oe-linux-gnueabi/gcc/include/ --sysroot=/mnt/ci_build/workspace/tmp/sysroots/genericarmv7a -DHAVE_CONFIG_H -I/mnt/ci_build/workspace/tmp/work-shared/gcc-linaro-4.8-r2013.09/gcc-linaro-4.8-2013.09/libatomic/config/arm -I/mnt/ci_build/workspace/tmp/work-shared/gcc-linaro-4.8-r2013.09/gcc-linaro-4.8-2013.09/libatomic/config/linux/arm -I/mnt/ci_build/workspace/tmp/work-shared/gcc-linaro-4.8-r2013.09/gcc-linaro-4.8-2013.09/libatomic/config/posix -I/mnt/ci_build/workspace/tmp/work-shared/gcc-linaro-4.8-r2013.09/gcc-linaro-4.8-2013.09/libatomic -I. -Wall -Werror -pthread -g -Os -MT load_1_.lo -MD -MP -MF .deps/load_1_.lo.Ppo -DN=1 -c /mnt/ci_build/workspace/tmp/work-shared/gcc-linaro-4.8-r2013.09/gcc-linaro-4.8-2013.09/libatomic/config/linux/arm/load_n.c -fPIC -DPIC -o .libs/load_1_.o
> 09:41:57 | /mnt/ci_build/workspace/tmp/work-shared/gcc-linaro-4.8-r2013.09/gcc-linaro-4.8-2013.09/libatomic/config/linux/arm/load_n.c:49:0: fatal error: opening dependency file .deps/load_1_.lo.Ppo: No such file or directory
> 09:41:57 | #include "../../../load_n.c"
> 09:41:57 | ^
> 09:41:57 | compilation terminated.
> 09:41:57 | make[4]: *** [load_1_.lo] Error 1
> 09:41:57 | make[4]: *** Waiting for unfinished jobs....
> -->8---
>
> As a quick test, I created a .deps directory, and now 'bitbake gcc' goes through just fine :)
> The failure we're hitting doesn't seem related to "a race in bitbake".
>
> --
> You received this bug notification because you are subscribed to Linaro
> OpenEmbedded.
> Matching subscriptions: linaro-oe bugs
> https://bugs.launchpad.net/bugs/1231760
>
> Title:
> linaro-image-lng 4.8 build fails
>
> To manage notifications about this bug go to:
> https://bugs.launchpad.net/linaro-networking/+bug/1231760/+subscriptions

[Riku Voipio (riku-voipio)]

This patch has been mentioned on OE irc:

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55930

But I think there may be more issues here.

On 28 September 2013 19:47, Riku Voipio <email address hidden> wrote:
> Yes, gcc build is being racy - I haven't managed to debug what exactly
> causes the race.
>
> On 27 September 2013 20:16, Zi Shen Lim <email address hidden> wrote:
>> I poked around a little. Looking at the log, it points to build failure in gcc/libatomic:
>> --8<---
>> 09:41:57 | libtool: compile: arm-oe-linux-gnueabi-gcc -march=armv7-a -marm -mthumb-interwork -mfloat-abi=hard -mfpu=neon --sysroot=/mnt/ci_build/workspace/tmp/sysroots/genericarmv7a -march=armv7-a -marm -mthumb-interwork -mfloat-abi=hard -mfpu=neon -isystem/mnt/ci_build/workspace/tmp/sysroots/genericarmv7a/usr/include -I/mnt/ci_build/workspace/tmp/work/armv7ahf-vfp-neon-oe-linux-gnueabi/gcc/linaro-4.8-r2013.09/gcc-linaro-4.8-2013.09/build.arm-oe-linux-gnueabi.arm-oe-linux-gnueabi/gcc/include/ --sysroot=/mnt/ci_build/workspace/tmp/sysroots/genericarmv7a -DHAVE_CONFIG_H -I/mnt/ci_build/workspace/tmp/work-shared/gcc-linaro-4.8-r2013.09/gcc-linaro-4.8-2013.09/libatomic/config/arm -I/mnt/ci_build/workspace/tmp/work-shared/gcc-linaro-4.8-r2013.09/gcc-linaro-4.8-2013.09/libatomic/config/linux/arm -I/mnt/ci_build/workspace/tmp/work-shared/gcc-linaro-4.8-r2013.09/gcc-linaro-4.8-2013.09/libatomic/config/posix -I/mnt/ci_build/workspace/tmp/work-shared/gcc-linaro-4.8-r2013.09/gcc-linaro-4.8-2013.09/libatomic -I. -Wall -Werror -pthread -g -Os -MT load_1_.lo -MD -MP -MF .deps/load_1_.lo.Ppo -DN=1 -c /mnt/ci_build/workspace/tmp/work-shared/gcc-linaro-4.8-r2013.09/gcc-linaro-4.8-2013.09/libatomic/config/linux/arm/load_n.c -fPIC -DPIC -o .libs/load_1_.o
>> 09:41:57 | /mnt/ci_build/workspace/tmp/work-shared/gcc-linaro-4.8-r2013.09/gcc-linaro-4.8-2013.09/libatomic/config/linux/arm/load_n.c:49:0: fatal error: opening dependency file .deps/load_1_.lo.Ppo: No such file or directory
>> 09:41:57 | #include "../../../load_n.c"
>> 09:41:57 | ^
>> 09:41:57 | compilation terminated.
>> 09:41:57 | make[4]: *** [load_1_.lo] Error 1
>> 09:41:57 | make[4]: *** Waiting for unfinished jobs....
>> -->8---
>>
>> As a quick test, I created a .deps directory, and now 'bitbake gcc' goes through just fine :)
>> The failure we're hitting doesn't seem related to "a race in bitbake".
>>
>> --
>> You received this bug notification because you are subscribed to Linaro
>> OpenEmbedded.
>> Matching subscriptions: linaro-oe bugs
>> https://bugs.launchpad.net/bugs/1231760
>>
>> Title:
>> linaro-image-lng 4.8 build fails
>>
>> To manage notifications about this bug go to:
>> https://bugs.launchpad.net/linaro-networking/+bug/1231760/+subscriptions

[Riku Voipio (riku-voipio)]

Last armv7 lng images built fine. We now hopefully have a workaround for the racy gcc build. The bigendian armv7 builds are failing since the bigendian patches are not in linux-linaro at the moment.