Re: [Xen-devel] Re: [kvm-devel] [PATCH RFC 0/3] Virtio draft III

[Rusty Russell <rusty@xxxxxxxxxxxxxxx>]

On Sun, 2007-06-17 at 17:14 +0300, Avi Kivity wrote:
> Rusty Russell wrote:
> > In this episode, Rusty tries to NAPI-ize the driver and discovers that
> > virtio callbacks are a bad idea: NAPI needs to turn interrupts off and
> > still be able to query for new incoming packets.
> >
> > Changes to core:
> > 1) Back to "interrupt" model with get_inbuf()/get_outbuf() calls.
> >   
> 
> Seems to be these are just two different ways of iterating over the 
> pending buffers, and one could be implemented in terms of the other in a 
> few lines.  I think that new new way is better, though, due to less 
> indirection.

Yes, the lguest implementation points them to a common routine.  We
could do some batching here too, but it seemed like premature
optimization.

> btw, I'm not sure that the name 'virtio' is correct.  Probably any 
> network or block driver which has support for batching (the vast 
> majority) could use this to good effect.

Trying to avoid boiling the ocean 8)

Cheers,
Rusty.



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