Re: [PATCH 2/3] vhost: better detection of available buffers

[Michael S. Tsirkin]

On Tue, Nov 15, 2016 at 04:00:21PM +0800, Jason Wang wrote:
> 
> 
> On 2016å11æ15æ 11:28, Michael S. Tsirkin wrote:
> > On Tue, Nov 15, 2016 at 11:16:59AM +0800, Jason Wang wrote:
> > > 
> > > On 2016å11æ12æ 00:20, Michael S. Tsirkin wrote:
> > > > On Fri, Nov 11, 2016 at 12:18:50PM +0800, Jason Wang wrote:
> > > > > On 2016å11æ11æ 11:41, Michael S. Tsirkin wrote:
> > > > > > On Fri, Nov 11, 2016 at 10:18:37AM +0800, Jason Wang wrote:
> > > > > > > > On 2016å11æ10æ 03:57, Michael S. Tsirkin wrote:
> > > > > > > > > > On Wed, Nov 09, 2016 at 03:38:32PM +0800, Jason Wang wrote:
> > > > > > > > > > > > We should use vq->last_avail_idx instead of vq->avail_idx in the
> > > > > > > > > > > > checking of vhost_vq_avail_empty() since latter is the cached avail
> > > > > > > > > > > > index from guest but we want to know if there's pending available
> > > > > > > > > > > > buffers in the virtqueue.
> > > > > > > > > > > > 
> > > > > > > > > > > > Signed-off-by: Jason Wang<jasowang@xxxxxxxxxx>
> > > > > > > > > > I'm not sure why is this patch here. Is it related to
> > > > > > > > > > batching somehow?
> > > > > > > > Yes, we need to know whether or not there's still buffers left in the
> > > > > > > > virtqueue, so need to check last_avail_idx. Otherwise, we're checking if
> > > > > > > > guest has submitted new buffers.
> > > > > > > > 
> > > > > > > > > > > > ---
> > > > > > > > > > > >     drivers/vhost/vhost.c | 2 +-
> > > > > > > > > > > >     1 file changed, 1 insertion(+), 1 deletion(-)
> > > > > > > > > > > > 
> > > > > > > > > > > > diff --git a/drivers/vhost/vhost.c b/drivers/vhost/vhost.c
> > > > > > > > > > > > index c6f2d89..fdf4cdf 100644
> > > > > > > > > > > > --- a/drivers/vhost/vhost.c
> > > > > > > > > > > > +++ b/drivers/vhost/vhost.c
> > > > > > > > > > > > @@ -2230,7 +2230,7 @@ bool vhost_vq_avail_empty(struct vhost_dev *dev, struct vhost_virtqueue *vq)
> > > > > > > > > > > >     	if (r)
> > > > > > > > > > > >     		return false;
> > > > > > > > > > > > -	return vhost16_to_cpu(vq, avail_idx) == vq->avail_idx;
> > > > > > > > > > > > +	return vhost16_to_cpu(vq, avail_idx) == vq->last_avail_idx;
> > > > > > > > > > > >     }
> > > > > > > > > > > >     EXPORT_SYMBOL_GPL(vhost_vq_avail_empty);
> > > > > > > > > > That might be OK for TX but it's probably wrong for RX
> > > > > > > > > > where the fact that used != avail does not mean
> > > > > > > > > > we have enough space to store the packet.
> > > > > > > > Right, but it's no harm since it was just a hint, handle_rx() can handle
> > > > > > > > this situation.
> > > > > > Means busy polling will cause useless load on the CPU though.
> > > > > > 
> > > > > Right, but,it's not easy to have 100% correct hint here. Needs more thought.
> > > > What's wrong with what we have? It polls until value changes.
> > > > 
> > > But as you said, this does not mean (in mergeable cases) we have enough
> > > space to store the packet.
> > Absolutely but it checks once and then only re-checks after value
> > changes again.
> > 
> 
> Since get_rx_bufs() does not get enough buffers, we will wait for the kick
> in this case. For busy polling, we probably want to stay in the busy loop
> here.

That's what I'm saying. You don't want to re-poll the queue
if available idx was unchanged.

-- 
MST