On Sat, 28 Feb 2015 11:31:21 +0800
Jason Wang <jasow...@redhat.com> wrote:
>
>
> On Fri, Feb 27, 2015 at 5:49 PM, Cornelia Huck
> <cornelia.h...@de.ibm.com> wrote:
> > On Fri, 27 Feb 2015 06:42:57 +0008
> > Jason Wang <jasow...@redhat.com> wrote:
> >
> >> On Thu, Feb 26, 2015 at 9:05 PM, Cornelia Huck
> >> <cornelia.h...@de.ibm.com> wrote:
> >> > On Thu, 26 Feb 2015 15:04:41 +0800
> >> > Jason Wang <jasow...@redhat.com> wrote:
> >> >
> >
> >> >> typedef struct AdapterRoutes {
> >> >> AdapterInfo adapter;
> >> >> int num_routes;
> >> >> - int gsi[VIRTIO_PCI_QUEUE_MAX];
> >> >> + int gsi[VIRTIO_S390_QUEUE_MAX];
> >> >
> >> > Adapter routes are only applicable for the ccw transport, not for
> >> the
> >> > old s390 transport.
> >>
> >> Sure, will fix this.
> >>
> >> >
> >> >
> >> > (I'm also wondering whether this should be the generic limit
> >> instead.)
> >>
> >> As you pointed out in V1, there will be more issues if we just
> >> increase
> >> the generic limit. So I switch to use per transport limit. Since
> >> the
> >> limit was not changed for both s390 and ccw, it should be ok.
> >
> > I'm just wondering how many gsis we want to support for adapter
> > routes.
> > They were introduced for virtio-ccw, but recently s390 pci has started
> > to use them as well, so a virtio limit seems silly here. I'll switch
> > them to some kind of generic limit instead, I think.
>
> Get your point. My understanding is you can do this on top of this
> series.
Yup, that will work.
