Re: [Xen-devel] [PATCH] [qemu] xen_be_init under stubdom

[Stefano Stabellini <stefano.stabellini@xxxxxxxxxxxxx>]

On Thu, 20 Jan 2011, Ian Jackson wrote:
> Kamala Narasimhan writes ("[Xen-devel] [PATCH] [qemu] xen_be_init under 
> stubdom"):
> > Do nothing in xen_be_init under stubdom plus a minor inconsequential 
> > cleanup.
> ...
> > -       goto cleanup;
> > +        return;
> ...
> > -cleanup:
> >      qemu_free(vec);
> >  }
> 
> I don't think this is a helpful change.  There is nothing wrong with
> calling qemu_free(0) and IMO in general functions that "goto cleanup"
> are to be preferred to ones that "return".
> 
> Furthermore, even if this patch were good, it's not a bugfix so is not
> acceptable at this stage of the release.
> 
> > @@ -646,6 +645,10 @@ static void xen_be_evtchn_event(void *opaque)
> > 
> >  int xen_be_init(void)
> >  {
> > +#ifdef CONFIG_STUBDOM
> > +    return 0;
> > +#endif
> 
> I don't understand this at all.  Why should stubdom not be able to
> make pv backends if it wants to ?  I agree that it probably doesn't
> want to but if something iswrongly causing it to then the right fix is
> to make it not do so.

the current xen_backend code in qemu cannot handle being run in a
stubdom, for example:

dom0 = xs_get_domain_path(xenstore, 0);

_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-devel