On 01/09/2023 21:28, Jay Vosburgh wrote:

quoted

quoted

The resulting zero-length SKBs cause crashes in GRO since commit
 1d11fa696733 ("net-gro: remove GRO_DROP"), so add a check to the driver
 to detect these zero-length RX events and discard the packet.

[...]

	Should this have included

Fixes: 1d11fa696733 ("net-gro: remove GRO_DROP")

	to queue the patch for -stable?  We have users running into this
issue on 5.15 series kernels.

I didn't think Fixes: was appropriate (for various abstruse reasons I
 won't go into), but this does want to go to -stable.  I expect Sasha's
 robot will select it, but feel free to submit it explicitly.

-ed