Re: [PATCH v5 3/5] virtio-scsi: default num_queues to -smp N

[Cornelia Huck <cohuck@redhat.com>]

[-- Attachment #1: Type: text/plain, Size: 1987 bytes --]

On Wed, 8 Jul 2020 14:05:26 +0100
Stefan Hajnoczi <stefanha@redhat.com> wrote:

> On Tue, Jul 07, 2020 at 05:44:53PM +0200, Cornelia Huck wrote:
> > On Mon,  6 Jul 2020 14:56:48 +0100
> > Stefan Hajnoczi <stefanha@redhat.com> wrote:

> > > diff --git a/hw/virtio/virtio-scsi-pci.c b/hw/virtio/virtio-scsi-pci.c
> > > index 3ff9eb7ef6..fa4b3bfb50 100644
> > > --- a/hw/virtio/virtio-scsi-pci.c
> > > +++ b/hw/virtio/virtio-scsi-pci.c
> > > @@ -46,13 +46,17 @@ static void virtio_scsi_pci_realize(VirtIOPCIProxy *vpci_dev, Error **errp)
> > >  {
> > >      VirtIOSCSIPCI *dev = VIRTIO_SCSI_PCI(vpci_dev);
> > >      DeviceState *vdev = DEVICE(&dev->vdev);
> > > -    VirtIOSCSICommon *vs = VIRTIO_SCSI_COMMON(vdev);
> > >      DeviceState *proxy = DEVICE(vpci_dev);
> > > +    VirtIOSCSIConf *conf = &dev->vdev.parent_obj.conf;
> > >      char *bus_name;
> > >  
> > > +    if (conf->num_queues == VIRTIO_SCSI_AUTO_NUM_QUEUES) {
> > > +        conf->num_queues =
> > > +            virtio_pci_optimal_num_queues(VIRTIO_SCSI_VQ_NUM_FIXED);
> > > +    }
> > > +
> > >      if (vpci_dev->nvectors == DEV_NVECTORS_UNSPECIFIED) {
> > > -        vpci_dev->nvectors = vs->conf.num_queues +
> > > -                             VIRTIO_SCSI_VQ_NUM_FIXED + 1;
> > > +        vpci_dev->nvectors = conf->num_queues + VIRTIO_SCSI_VQ_NUM_FIXED + 1;
> > >      }
> > >  
> > >      /*  
> > 
> > One corner case where the setup may end up being a bit odd is a
> > situation where nvectors was specified, but num_queues was not, and the
> > device suddenly ends up with more queues than vectors. But I don't see
> > a reason why you would want to specify nvectors but not num_queues in a
> > real word scenario, so I think we can ignore that corner case.  
> 
> I agree, I've ignored that case. Other options include printing a
> warning or even an error when num_queues disagrees with nvectors.

I think an error would be too harsh, but a warning sounds useful.

[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 833 bytes --]