On Mon, 2021-03-01 at 14:58 +0100, Matthieu Baerts wrote:
This report from syzbot mentioned an possible deadlock with MPTCP
socket. I don't know if the commit from nfsd is really causing that.
I'm then sharing this with the list just in case someone saw similar
Looks like the deadlock report is valid, we should not invoke
ipv6_sock_mc_close() under the socket lock.
I'm unsure why syzbot bisect to an nfs commit.
I'll test a fix via syzbot soon.