[Lguest] A weird problem with a linux guest

[ron minnich]

On 8/1/07, Rusty Russell <rusty at rustcorp.com.au> wrote:
> On Wed, 2007-08-01 at 13:11 -0700, ron minnich wrote:
> > I have a linux busybox file system image. If I boot 2.6.23 lguest
> > guest kernel, it works fine to a point -- that point being the point
> > at which it locks up. Anyone seen anything like this?
>
> Hi Ron!
>
>         We used to have this problem with the jiffies clock, which is why the
> big clock change occurred: after some period of time the tick code would
> go into an infinite loop.  If your "struct lguest_data" has a "struct
> timespec time" field, it's the new one.
>
> > /proc/interrupts is odd:
> > cat /proc/interrupts
> >            CPU0
> >   1:         75    lguest-<NULL>    lgba
> >   2:          0    lguest-<NULL>    lguestnet
> >   3:         47    lguest-<NULL>    hvc_console
> > NMI:          0
> > LOC:          0
> > ERR:          0
> > MIS:          0
> > Where's the clock interrupt? It never increments past 75.
>
> Hmm, the clock isn't here (it'd be 0).  Perhaps I overrode it too hard
> 8)

I will send .config tomorrow -- one thing I did for a try, is put
printk in the linux guest interrupt handler. I'm getting clock
interrupts in the linux guest. I will dig some more and send you
.config tomorrow. I had this problem on both 2.6.22.1 and now.

Hardware is a thinkpad x60.

ron