Re: [RFC PATCH net 2/3] virtio-net: allow usage of vrings smaller than MAX_SKB_FRAGS + 2

[Michael S. Tsirkin]

On Mon, May 01, 2023 at 11:59:42AM +0000, Alvaro Karsz wrote:
> > First up to 4k should not be a problem. Even jumbo frames e.g. 9k
> > is highly likely to succeed. And a probe time which is often boot
> > even 64k isn't a problem ...
> > 
> > Hmm. We could allocate large buffers at probe time. Reuse them and
> > copy data over.
> > 
> > IOW reusing  GOOD_COPY_LEN flow for this case.  Not yet sure how I feel
> > about this. OTOH it removes the need for the whole feature blocking
> > approach, does it not?
> > WDYT?
> > 
> 
> It could work..
> 
> In order to remove completely the feature blocking approach, we'll need to let the control commands fail (as you mentioned in the other patch).
> I'm not sure I like it, it means many warnings from virtnet..
> And it means accepting features that we know for sure that are not going to work.
>

Well they will work yes? just with an extra copy.

-- 
MST