[OE-core] glibc build error when it is rebuild due hash mismatch

[Otavio Salvador]

On Tue, Oct 13, 2015 at 2:02 PM, Khem Raj <raj.khem at gmail.com> wrote:
>
>> On Oct 13, 2015, at 9:59 AM, Otavio Salvador <otavio.salvador at ossystems.com.br> wrote:
>>
>> On Tue, Oct 13, 2015 at 1:54 PM, Khem Raj <raj.khem at gmail.com> wrote:
>>>
>>>> On Oct 13, 2015, at 9:53 AM, Otavio Salvador <otavio.salvador at ossystems.com.br> wrote:
>>>>
>>>> On Tue, Oct 13, 2015 at 1:52 PM, Khem Raj <raj.khem at gmail.com> wrote:
>>>>>
>>>>>> On Oct 13, 2015, at 9:45 AM, Otavio Salvador <otavio.salvador at ossystems.com.br> wrote:
>>>>>>
>>>>>> Hello,
>>>>>>
>>>>>> From time to time I see following error:
>>>>>>
>>>>>> | /usr/bin/install -c -m 644
>>>>>> /home/otavio/src/yocto/build/tmp/work/cortexa9hf-vfp-neon-oel-linux-gnueabi/glibc/2.22-r0/build-arm-oel-linux-gnueabi/gnu/option-groups.h
>>>>>> /home/otavio/src/yocto/build/tmp/work/cortexa9hf-vfp-neon-oel-linux-gnueabi/glibc/2.22-r0/image/usr/include/gnu/option-groups.h
>>>>>> | /usr/bin/install -c -m 644
>>>>>> /home/otavio/src/yocto/build/tmp/work/cortexa9hf-vfp-neon-oel-linux-gnueabi/glibc/2.22-r0/build-arm-oel-linux-gnueabi/gnu/lib-names-hard.h
>>>>>> /home/otavio/src/yocto/build/tmp/work/cortexa9hf-vfp-neon-oel-linux-gnueabi/glibc/2.22-r0/image/usr/include/gnu/lib-names-hard.h
>>>>>> | msgfmt -o rw.mo rw.po
>>>>>> | /bin/bash: msgfmt: command not found
>>>>>> | Makefile:84: recipe for target 'rw.mo' failed
>>>>>> | make[2]: *** [rw.mo] Error 127
>>>>>> | make[2]: *** Waiting for unfinished jobs....
>>>>>> | make[2]: Leaving directory
>>>>>> '/home/otavio/src/yocto/build/tmp/work/cortexa9hf-vfp-neon-oel-linux-gnueabi/glibc/2.22-r0/git/po'
>>>>>> | Makefile:234: recipe for target 'po/subdir_install' failed
>>>>>>
>>>>>> and it seems to happen when a rebuild of glibc is triggered due a
>>>>>> change in the sstate hash. The host is a very minimal system so it
>>>>>> explains why it does not provide the msgfmt in host and also might
>>>>>> explain why it is not being catch in Yocto Project AB.
>>>>>
>>>>>
>>>>> are we missing a dependency on gettext-native ? it could be that AB slaves have gettext installed on host
>>>>> and you don’t.
>>>>
>>>> First build works fine; rebuild fails. So I am not sure this is the culprit.
>>>
>>> it might not be rebuilding message catalog first time.
>>
>> Right but in this case:
>>
>> - we fix it to not rebuild it
>> - we fix why it is not building it
>>
>> I am a complete ignorant in glibc build system. Can you help on this one?
>
> share the complete logs for first time and second time builds somewhere.

I will try to come up with  a way to reproduce it. I don't have the
logs anymore.

-- 
Otavio Salvador                             O.S. Systems
http://www.ossystems.com.br        http://code.ossystems.com.br
Mobile: +55 (53) 9981-7854            Mobile: +1 (347) 903-9750