Re: [5/6] [VIRTIO] net: Add ethtool ops for SG/GSO

[Rusty Russell <rusty@rustcorp.com.au>]

On Tuesday 22 April 2008 11:15:02 Herbert Xu wrote:
> On Tue, Apr 22, 2008 at 05:01:46AM +1000, Rusty Russell wrote:
> > On Friday 18 April 2008 13:21:42 Herbert Xu wrote:
> > > +static int virtnet_change_mtu(struct net_device *dev, int mtu)
> > > +{
> > > +	int max = 65535 - ETH_HLEN;
> > > +
> > > +	if (mtu > max)
> > > +		return -EINVAL;
> > > +	dev->mtu = mtu;
> > > +	return 0;
> > > +}
> >
> > Hi Herbert!
> >
> >     I removed this part; useful for testing, but we need a feature bit
> > for MTU size in general.  And to change it on the fly either requires a
> > reset & re-init, or some protocol (and feature bit!) for renegotiating
> > MTU.
>
> BTW Rusty this was just a work-in-progress.  When I submit them I
> will add sign-offs.

OK.  Meanwhile I stole it for my own testing :)

> However, the MTU part should be fine as long as the other end supports
> SG.  The operative word in MTU is T :)

I hadn't really thought about it; you're right, MTU is merely a formality.  If 
the other end has big enough receive buffers, we might as well use them.

I still feel oddly nervous about surprising the other end tho...

Thanks,
Rusty.