hi,

> > > With:
> > > panic: lockmgr: no context
> > > It is in vndstrategy:
> > > 		vn_lock(vnd->sc_vp, LK_EXCLUSIVE | LK_RETRY | LK_CANRECURSE);
> > > 
> > > This means that I/O on the vnd block device expect to be called from
> > > a process context, while the xbd backend calls it from interrupt context.
> > 
> > do you have a backtrace?
> 
> I had, but the problem is quite clear: when writing to a vnd, the chain
> always end up calling vndstrategy() which always end up calling vn_lock().
> So we can't write to a vnd from interrupt context. The xbd backen initiates
> its I/O from interrupt context (the interrupt from the hypervisor, signaling
> that there is a request pending for this xbd).

i believe it's a problem in xbd, not in vnd.
doing vnode operations from interrupt context is evil.

(well, yes, vnd has its own problems.  having a thread for vnd is fine.
but it's another story.)

YAMAMOTO Takashi