Re: [Xen-devel] [xen-unstable test] 11807: regressions - FAIL

[Ian Jackson <Ian.Jackson@xxxxxxxxxxxxx>]

xen.org writes ("[xen-unstable test] 11807: regressions - FAIL"):
> Tests which did not succeed and are blocking,
> including tests which could not be run:
>  test-amd64-i386-xl-credit2    9 guest-start            fail REGR. vs. 11643
>  test-amd64-i386-xl-win7-amd64  7 windows-install       fail REGR. vs. 11643

Both of these seem to be high-probability heisenbugs.  In both cases
LThe failure seems to be much more likely (perhaps even restricted to)
on certain hosts, and now that my tester is "sticky" when it sees a
fail these will give us more trouble.

For this one
>  test-amd64-i386-xl-win7-amd64  7 windows-install       fail REGR. vs. 11643
the failure mode is that Windows sits waiting at a login screen rather
than logging in the user which is supposed to run the test responder
which was installed with the image.  Most likely cause is that it has
stopped making progress for some reason.  None of the logs seem
interesting.

This one
>  test-amd64-i386-xl-credit2    9 guest-start            fail REGR. vs. 11643
which occasionally also manifests as various other failures of the
same test run, is this assertion:

Feb  2 02:34:49.293506 (XEN) Assertion 'test_bit(vector, 
desc->arch.used_vectors)' failed at irq.c:670

which I mentioned yesterday.  My bisector had a go at the
guest-saverestore test and produced inconclusive results but looking
at the revision graph output I suspect that something between
24612:54000bca7a6a (probably good) and 24620:926e0820d615 (definitely
bad) may be implicated.

Given that this bug seems to be pretty specific to a particular host,
which happens to be a dual-socket six-core AMD Opteron 2427, I'm
wondering whether 24613:04623ceabb72 "amd iommu: disable iommu
emulation on non-iommu systems" might be relevant.

Suggestions for either bug, and offers to debug these, gratefully
received.  Failing that we will probably get lucky and get a push
eventually.

Ian.

_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-devel