[1/1] Fix SETATTR to always retry on bad_stateid-type error

Commit Message

Olga Kornievskaia June 10, 2016, 6:37 p.m. UTC

It's insufficient to only retry SETATTR if we have a delegation
and got a bad_stateid-type error. A DELEGRETURN could be sent
after SETATTR is sent and before the reply comes back thus
no delegation would be found and SETATTR not retried with
zero stateid and instead fail the operation to EIO.

Signed-off-by: Olga Kornievskaia <[e-mail]>
---
 fs/nfs/nfs4proc.c | 9 ++++++++-
 1 file changed, 8 insertions(+), 1 deletion(-)

Patch

diff --git a/fs/nfs/nfs4proc.c b/fs/nfs/nfs4proc.c
index 327b8c3..c40d895 100644
--- a/fs/nfs/nfs4proc.c
+++ b/fs/nfs/nfs4proc.c
@@ -2723,7 +2723,7 @@  static int nfs4_do_setattr(struct inode *inode, struct rpc_cred *cred,
 		.state = state,
 		.inode = inode,
 	};
-	int err;
+	int err, save_err;
 	do {
 		err = _nfs4_do_setattr(inode, cred, fattr, sattr, state, ilabel, olabel);
 		switch (err) {
@@ -2742,7 +2742,14 @@  static int nfs4_do_setattr(struct inode *inode, struct rpc_cred *cred,
 				goto out;
 			}
 		}
+		save_err = err;
 		err = nfs4_handle_exception(server, err, &exception);
+		switch (save_err) {
+		case -NFS4ERR_DELEG_REVOKED:
+		case -NFS4ERR_ADMIN_REVOKED:
+		case -NFS4ERR_BAD_STATEID:
+			exception.retry = 1;
+		}
 	} while (exception.retry);
 out:
 	return err;