Re: [Xen-users] xen_platform_pci is stuck during load

[Wei Liu <wei.liu2@xxxxxxxxxx>]

On Mon, Mar 07, 2016 at 07:22:11PM +0200, Erez Zilber wrote:
> On Mon, Mar 7, 2016 at 1:33 PM, Wei Liu <wei.liu2@xxxxxxxxxx> wrote:
> > On Sun, Mar 06, 2016 at 09:18:14PM +0200, Erez Zilber wrote:
> >> Hi,
> >>
> >> I'm running a SUSE 11 sp4 VM over Xen with PV NIC. We use iPXE to boot
> >> the VM. During boot, iPXE holds the NIC and when the OS comes up, it
> >> signals the hypervisor to reset the NIC so it can be attached to the
> >> OS drivers.
> >>
> >> On a successful boot, the boot log looks like this:
> >>
> >> [    1.051785] Xen version 4.2.
> >> [    1.053635] Hypercall area is 1 pages.
> >> [    1.055753] xen-platform-pci 0000:00:03.0: I/O protocol version 1
> >> [    1.060436] XENBUS: frontend device/vif/0 Connected
> >> [    1.063270] XENBUS: backend /local/domain/0/backend/vif/18463/0 
> >> Connected
> >> [    1.067032] XENBUS: triggering reconnect on
> >> /local/domain/0/backend/vif/18463/0
> >> [    1.071909] XENBUS: backend
> >> /local/domain/0/backend/vif/18463/0/state Connected
> >> [    1.165918] XENBUS: backend /local/domain/0/backend/vif/18463/0/state 
> >> Closing
> >> [    1.174109] XENBUS: backend /local/domain/0/backend/vif/18463/0/state 
> >> Closed
> >> [    1.181906] XENBUS: backend
> >> /local/domain/0/backend/vif/18463/0/state InitWait
> >> [    1.187420] XENBUS: reconnect done on 
> >> /local/domain/0/backend/vif/18463/0
> >> [    1.192649] suspend: event channel 6
> >> [    1.198080] Unable to read sysrq code in control/sysrq
> >> [    1.208631] blkfront: device/vbd/768: ring-pages=1 nr_ents=32
> >>
> >> Sometimes, boot fails and then the boot log looks like this:
> >>
> >> [    1.284151] Xen version 4.2.
> >> [    1.286322] Hypercall area is 1 pages.
> >> [    1.288947] xen-platform-pci 0000:00:03.0: I/O protocol version 1
> >>
> >> I can see that the modprobe call (that loads xen_platform_pci) never 
> >> returns.
> >>
> >> Is this a known issue? Is there a workaround for this?
> >>
> >
> > I haven't seen similar report in upstream Xen.
> >
> > It's worth looking at the log of device model (qemu) to see if there is
> > anything useful.
> >
> 
> The VM is running in the cloud, so I don't have access to these logs.
> 

OK. Then it would be better to report this to service provider.

TBH there isn't much we can do other than saying "we don't see report
like this in upstream".

Wei.

> Erez

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