Re: [PATCH net-next v2 3/5] virtio_ring: add packed ring support

[Tiwei Bie <tiwei.bie@intel.com>]

On Thu, Nov 08, 2018 at 10:56:02AM -0500, Michael S. Tsirkin wrote:
> On Thu, Nov 08, 2018 at 07:51:48PM +0800, Tiwei Bie wrote:
> > On Thu, Nov 08, 2018 at 04:18:25PM +0800, Jason Wang wrote:
> > > 
> > > On 2018/11/8 上午9:38, Tiwei Bie wrote:
> > > > > > +
> > > > > > +	if (vq->vq.num_free < descs_used) {
> > > > > > +		pr_debug("Can't add buf len %i - avail = %i\n",
> > > > > > +			 descs_used, vq->vq.num_free);
> > > > > > +		/* FIXME: for historical reasons, we force a notify here if
> > > > > > +		 * there are outgoing parts to the buffer.  Presumably the
> > > > > > +		 * host should service the ring ASAP. */
> > > > > I don't think we have a reason to do this for packed ring.
> > > > > No historical baggage there, right?
> > > > Based on the original commit log, it seems that the notify here
> > > > is just an "optimization". But I don't quite understand what does
> > > > the "the heuristics which KVM uses" refer to. If it's safe to drop
> > > > this in packed ring, I'd like to do it.
> > > 
> > > 
> > > According to the commit log, it seems like a workaround of lguest networking
> > > backend.
> > 
> > Do you know why removing this notify in Tx will break "the
> > heuristics which KVM uses"? Or what does "the heuristics
> > which KVM uses" refer to?
> 
> Yes. QEMU has a mode where it disables notifications and processes TX
> ring periodically from a timer.  It's off by default but used to be on
> by default a long time ago. If ring becomes full this causes traffic
> stalls.  As a work-around Rusty put in this hack to kick on ring full
> even with notifications disabled.  It's easy enough to make sure QEMU
> does not combine devices with packed ring support with the timer hack.
> And I am guessing it's safe enough to also block that option completely
> e.g. when virtio 1.0 is enabled.

I see. Thanks!

> 
> > 
> > > I agree to drop it, we should not have such burden.
> > > 
> > > But we should notice that, with this removed, the compare between packed vs
> > > split is kind of unfair. Consider the removal of lguest support recently,
> > > maybe we can drop this for split ring as well?
> > > 
> > > Thanks
> > > 
> > > 
> > > > 
> > > > commit 44653eae1407f79dff6f52fcf594ae84cb165ec4
> > > > Author: Rusty Russell<rusty@rustcorp.com.au>
> > > > Date:   Fri Jul 25 12:06:04 2008 -0500
> > > > 
> > > >      virtio: don't always force a notification when ring is full
> > > >      We force notification when the ring is full, even if the host has
> > > >      indicated it doesn't want to know.  This seemed like a good idea at
> > > >      the time: if we fill the transmit ring, we should tell the host
> > > >      immediately.
> > > >      Unfortunately this logic also applies to the receiving ring, which is
> > > >      refilled constantly.  We should introduce real notification thesholds
> > > >      to replace this logic.  Meanwhile, removing the logic altogether breaks
> > > >      the heuristics which KVM uses, so we use a hack: only notify if there are
> > > >      outgoing parts of the new buffer.
> > > >      Here are the number of exits with lguest's crappy network implementation:
> > > >      Before:
> > > >              network xmit 7859051 recv 236420
> > > >      After:
> > > >              network xmit 7858610 recv 118136
> > > >      Signed-off-by: Rusty Russell<rusty@rustcorp.com.au>
> > > > 
> > > > diff --git a/drivers/virtio/virtio_ring.c b/drivers/virtio/virtio_ring.c
> > > > index 72bf8bc09014..21d9a62767af 100644
> > > > --- a/drivers/virtio/virtio_ring.c
> > > > +++ b/drivers/virtio/virtio_ring.c
> > > > @@ -87,8 +87,11 @@ static int vring_add_buf(struct virtqueue *_vq,
> > > >   	if (vq->num_free < out + in) {
> > > >   		pr_debug("Can't add buf len %i - avail = %i\n",
> > > >   			 out + in, vq->num_free);
> > > > -		/* We notify*even if*  VRING_USED_F_NO_NOTIFY is set here. */
> > > > -		vq->notify(&vq->vq);
> > > > +		/* FIXME: for historical reasons, we force a notify here if
> > > > +		 * there are outgoing parts to the buffer.  Presumably the
> > > > +		 * host should service the ring ASAP. */
> > > > +		if (out)
> > > > +			vq->notify(&vq->vq);
> > > >   		END_USE(vq);
> > > >   		return -ENOSPC;
> > > >   	}
> > > > 
> > > >