nfsd: drop st_mutex before calling move_to_close_lru()
authorNeilBrown <neilb@suse.de>
Mon, 8 Apr 2024 02:09:18 +0000 (12:09 +1000)
committerChuck Lever <chuck.lever@oracle.com>
Mon, 6 May 2024 13:07:16 +0000 (09:07 -0400)
commit56c35f43eef013579c76c007ba1f386d8c2cac14
treea27365a76104714cfd5ffe6ec14087369565941f
parenteec7620800081e27dbf8019ac2e66259f0d5bf6f
nfsd: drop st_mutex before calling move_to_close_lru()

move_to_close_lru() is currently called with ->st_mutex held.
This can lead to a deadlock as move_to_close_lru() waits for sc_count to
drop to 2, and some threads holding a reference might be waiting for the
mutex.  These references will never be dropped so sc_count will never
reach 2.

There can be no harm in dropping ->st_mutex before
move_to_close_lru() because the only place that takes the mutex is
nfsd4_lock_ol_stateid(), and it quickly aborts if sc_type is
NFS4_CLOSED_STID, which it will be before move_to_close_lru() is called.

See also
 https://lore.kernel.org/lkml/4dd1fe21e11344e5969bb112e954affb@jd.com/T/
where this problem was raised but not successfully resolved.

Reviewed-by: Jeff Layton <jlayton@kernel.org>
Signed-off-by: NeilBrown <neilb@suse.de>
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
fs/nfsd/nfs4state.c