NFSv4: nfs4_stateid_is_current should return 'true' for an invalid stateid
authorTrond Myklebust <trond.myklebust@primarydata.com>
Wed, 5 Mar 2014 13:44:23 +0000 (08:44 -0500)
committerGreg Kroah-Hartman <gregkh@linuxfoundation.org>
Mon, 24 Mar 2014 04:38:17 +0000 (21:38 -0700)
commitc8dd8fdf0bd8c858d30ba3889104e226e865cade
tree63ec25df247bc7579d0ee43c5e6cb05d313125d6
parent40cb674a5a0cff49a46af150fb862229e545e908
NFSv4: nfs4_stateid_is_current should return 'true' for an invalid stateid

commit e1253be0ece1a95a02c7f5843194877471af8179 upstream.

When nfs4_set_rw_stateid() can fails by returning EIO to indicate that
the stateid is completely invalid, then it makes no sense to have it
trigger a retry of the READ or WRITE operation. Instead, we should just
have it fall through and attempt a recovery.

This fixes an infinite loop in which the client keeps replaying the same
bad stateid back to the server.

Reported-by: Andy Adamson <andros@netapp.com>
Link: http://lkml.kernel.org/r/1393954269-3974-1-git-send-email-andros@netapp.com
Signed-off-by: Trond Myklebust <trond.myklebust@primarydata.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
fs/nfs/nfs4proc.c