Re: 3 Failures in RC3

[Saul Wold <sgw@linux.intel.com>]

On 08/07/2011 11:11 AM, Flanagan, Elizabeth wrote:
> On Sat, Aug 6, 2011 at 4:20 PM, Saul Wold<sgw@linux.intel.com>  wrote:
>> On 08/06/2011 11:23 AM, Saul Wold wrote:
>>>
>>> Folks,
>>>
>>> We currently have 3 failures on the RC3 build for M3. They are all
>>> related to LSB builds. If you are on the TO: line please review your
>>> issues as we need a prompt fix.
>>>
>>> 1348: Qt3 Fails to build for ARM LSB
>>>
>>> 1349: Shared State Information gets lost for Mips FSTests
>>> I am trying to reproduce this on the Autobuilder
>>>
>> I am still working on reproducing this one, might be autobuilder related as
>> my local autobuilder build seems to be OK right now.
>> I have 2 sets of sstate for fstests, which may mean that something is
>> different between the poky and poky-lsb builds.
>>
>> On the autobuilder, the sstate directory was empty, so I can't verify
>> short of restarting the build again.
>
> Look at http://autobuilder.yoctoproject.org/nightly/CURRENT/sstate-cache
>
> That is the saved sstate for that build.
>
I found the correct location in the autobuilder, but it seems that the 
sstate-cache is bad and should be removed, I thought we where removing 
sstate before starting a nightly?

The bad sets are:
sstate-fstests-mips-poky-linux-0.0+svnr426-r1-mips-2-1ac2de42c742acfe5b35e44bfe68e7c1_package.*
sstate-fstests-mips-poky-linux-0.0+svnr426-r1-mips-2-44cb8fcc9b29e00d9d339e641c333d1e_package.*

sstate-fstests-mips-poky-linux-0.0+svnr426-r1-mips-2-ad481871e7622c756a3262632dfb088c_deploy-rpm.*
sstate-fstests-mips-poky-linux-0.0+svnr426-r1-mips-2-79fefc65e6e34db9ae3bfbfdff6f9522_deploy-rpm.*

I think if those are removed or the sstate is reset for each nightly 
build the fstest issue will go away.

>>
>>> 1450: ccache directory does not get created correct
>>
>> oops 1350 is the correct bugid!
>>
>>> I can reproduce this one very easily by building, saving the
>>> sstate-cache directory and then building again with a clean tmp
>>> directory. The ccache/x86-64-poky-linux directory does not get created
>>> and the kernel-configme task fail during the HOSTCC execution.
>>>
>> I have a fix for this one, in MUT and oe-contrib/sgw/stage.
>>
>
> Is this also what is occurring during the meta-intel builds?
>
Yes, and as you may have noticed it's been fixed and I pushed it to the 
RC3 branch and re-built the meta-intel builds.

>>> Please take a look or if you know of these issues, please let me know.
>>>
>>> Thanks
>>>
>>
>