On Wed, 25 Jan 2023 02:54:52 +0100 Toke Høiland-Jørgensen wrote: > However, because the skb->head in this case was allocated from a slab > allocator, taking a page refcount is not enough to prevent it from being > freed. > > I'm not sure how best to fix this. I guess we could try to detect this > case in the veth driver and copy the data, like we do for skb_share() > etc. However, I'm not sure how to actually detect this case... sbk->head_frag ?
