Re: [Xen-devel] [xen-unstable test] 24366: tolerable trouble: broken/fail/pass

[Ian Campbell <Ian.Campbell@xxxxxxxxxx>]

On Thu, 2014-01-16 at 14:22 +0000, Stefano Stabellini wrote:
> On Mon, 13 Jan 2014, Ian Campbell wrote:
> > On Mon, 2014-01-13 at 08:19 +0000, xen.org wrote:
> > > flight 24366 xen-unstable real [real]
> > > http://www.chiark.greenend.org.uk/~xensrcts/logs/24366/
> > > [...]
> > > Tests which did not succeed, but are not blocking:
> > >  test-armhf-armhf-xl           9 guest-start                  fail   
> > > never pass
> > AKA
> > http://www.chiark.greenend.org.uk/~xensrcts/logs/24366/test-armhf-armhf-xl/info.html
> > 
> > We are getting there (slowly), the new failure after making EXT4
> > available is:
> > http://www.chiark.greenend.org.uk/~xensrcts/logs/24366/test-armhf-armhf-xl/marilith-n5---var-log-xen-console-guest-debian.guest.osstest.log
> >         [    0.087330] Registering SWP/SWPB emulation handler
> >         [    0.089780] blkfront: xvda2: flush diskcache: enabled; 
> > persistent grants: enabled; indirect descriptors: enabled;
> >         [    0.099255] blkfront: xvda1: flush diskcache: enabled; 
> > persistent grants: enabled; indirect descriptors: enabled;
> >         [    0.179907] drivers/rtc/hctosys.c: unable to open rtc device 
> > (rtc0)
> >         [    0.193307] List of all partitions:
> >         [    0.193325] ca02         4194304 xvda2  driver: vbd
> >         [    0.193340] ca01         1024000 xvda1  driver: vbd
> >         [    0.193352] No filesystem could mount root, tried:  ext3 ext2 
> > ext4
> >         [    0.193376] Kernel panic - not syncing: VFS: Unable to mount 
> > root fs on unknown-block(202,2)
> >         
> > The disk is on LVM and
> > http://www.chiark.greenend.org.uk/~xensrcts/logs/24366/test-armhf-armhf-xl/marilith-n5-output-xenstore-ls_-fp
> > shows that front and backend are in state 4/connected. I don't see any
> > smoking guns in the logs.
> > 
> > Julien/Stefano -- do you see anything?
> 
> LVM works for me, but I am not using udev at the moment.

I instrumented the guest f/s and blkfront and it seems like reads are
returning buffers full of 0xc2c2c2c2, which is the pattern that Xen
scrubs pages with in a debug build.

So either there is a cache coherency issue or perhaps something to do
with the dom0 swiotlb doing direct i/o to guest pages and sending them
to the wrong place.

Ian.


_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
http://lists.xen.org/xen-devel