Re: netfront for review

On Wed, 2007-05-02 at 13:51 +1000, Herbert Xu wrote:

On Wed, May 02, 2007 at 01:37:13PM +1000, Rusty Russell wrote:

quoted

quoted

+static int xennet_change_mtu(struct net_device *dev, int mtu)
+{
+	int max = xennet_can_sg(dev) ? 65535 - ETH_HLEN : ETH_DATA_LEN;

This seems odd to me: just because a device does TSO should we really
allow huge mtu settings?  Herbert?

Actually this has nothing to do with TSO/GSO.  This driver inherently
supports arbitrary MTUs that's only limited by our network stack.  If
the physical network device can't handle it you'll just get an ICMP
error back or fragmentation.

Oops, I misread "xennet_can_sg" as a test for GSO.

Thanks for the clarification,
Rusty.