https://bugzilla.kernel.org/show_bug.cgi?id=197861 --- Comment #15 from Willem de Bruijn (willemb@xxxxxxxxxx) --- It is quite likely that this patch --or patchset-- is involved, as the hang happens in cleaning up zerocopy skb state and this patchset extends zerocopy. Because of conflicts I had to resolve this patch is not a pure revert, but the changes should be limited to replacing uarg->callback(uarg, false); skb_shinfo(skb)->tx_flags &= ~SKBTX_DEV_ZEROCOPY; with an skb_zcopy_clear(skb, false) that boils down to the same. One option would be to test while reverting the entire patchset. I may be able to do that, no idea how many conflicts I'd have to resolve. Perhaps first, could you try the other suggestion with the warnings. If those fire, then we know for sure that this is implicated. -- You are receiving this mail because: You are watching the assignee of the bug.
