Re: [PATCH V3] xen: Fix BUFIOREQ evtchn init for a stubdom.

[Keir Fraser <keir.xen@gmail.com>]

On 02/07/2012 17:01, "Jan Beulich" <JBeulich@suse.com> wrote:

>>>> On 02.07.12 at 17:58, "Jan Beulich" <JBeulich@suse.com> wrote:
>>> @@ -3775,19 +3790,16 @@ long do_hvm_op(unsigned long op,
>>> XEN_GUEST_HANDLE(void) arg)
>>>                  rc = 0;
>>>                  domain_pause(d); /* safe to change per-vcpu xen_port */
>>>                  iorp = &d->arch.hvm_domain.ioreq;
>>> +                if (d->vcpu[0])
>>> +                    hvm_replace_event_channel(d->vcpu[0], a.value,
>>> +              
>>> (int*)&d->vcpu[0]->domain->arch.hvm_domain.params[HVM_PARAM_BUFIOREQ_EVTCHN]
>>> );
>> 
>> Did I overlook this in v2? You clearly need to handle the error
>> case here (it is being handled, albeit - but that's not your patch's
>> fault - only in a rudimentary way, inside the loop).
> 
> Probably it'll be easiest if I - or Keir if he's faster - add this while
> committing, to save you from posting another version.

I'm travelling so please just go ahead, Jan.

 K.

> Jan
>