On 2016-05-10 18:02, Andreas Henriksson wrote:
> Hello mipsel porters!
Hi,
> The libsecret package currently suffers from FTBFS on mipsel
> because of two testsuite failures and your help fixing these
> would be much appreciated so we can move on elsewhere.
>
> The problem is tracked in debian bug #821347 (CCed).
>
> >From
> >https://buildd.debian.org/status/fetch.php?pkg=libsecret&arch=mipsel&ver=0.18.5-1&stamp=1461171471
> > :
>
> FAIL: test-session 4 /session/ensure-async-aes
This test is fully reproducible on Loongson 3 machines (almost all our
mips*el build daemons, and the porterbox), but I have not been able to
reproduce it on an Octeon machine, nor on a CI20 board nor under QEMU. I
therefore believe there is a bug specific to the Loongson 3 there.
Compared to the other machines they have 16kB pages (instead of 4kB), I
don't know if it could be the issue as I haven't investigated more yet.
> FAIL: test-collection 18 /collection/delete-sync
This one fail intermittently only, though again I have only seen it
failing on Loongson 3 machines. It looks like tests are done using a
random seed, so the result might depends on that.
> Please note that the second failure also happens on s390x.
> I've only contacted you in hope that your fixes might positively
> effect the s390x build, but you might want to contact and join
> forces with the s390x porters.
This looks like a different issue, as it doesn't fail on s390x, but
instead hangs, and in addition this is fully reproducible not
intermittent.
A workaround might be to always build libsecret on the octeon build
daemons. Currently that means the package might stay much longer in
the build queue before being built, but we should deploy more of such
machines in the next weeks. Otherwise it needs a deeper investigation
which will take more time, as one has to understand the code to be
able to debug it.
Aurelien
--
Aurelien Jarno GPG: 4096R/1DDD8C9B
aurel...@aurel32.net http://www.aurel32.net
