NFSv4: Retry LOCK on OLD_STATEID during delegation return
authorBenjamin Coddington <bcodding@redhat.com>
Wed, 19 Oct 2022 16:09:18 +0000 (12:09 -0400)
committerGreg Kroah-Hartman <gregkh@linuxfoundation.org>
Sat, 26 Nov 2022 08:24:31 +0000 (09:24 +0100)
[ Upstream commit f5ea16137a3fa2858620dc9084466491c128535f ]

There's a small window where a LOCK sent during a delegation return can
race with another OPEN on client, but the open stateid has not yet been
updated.  In this case, the client doesn't handle the OLD_STATEID error
from the server and will lose this lock, emitting:
"NFS: nfs4_handle_delegation_recall_error: unhandled error -10024".

Fix this by sending the task through the nfs4 error handling in
nfs4_lock_done() when we may have to reconcile our stateid with what the
server believes it to be.  For this case, the result is a retry of the
LOCK operation with the updated stateid.

Reported-by: Gonzalo Siero Humet <gsierohu@redhat.com>
Signed-off-by: Benjamin Coddington <bcodding@redhat.com>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
fs/nfs/nfs4proc.c

index b42e332..dc03924 100644 (file)
@@ -7118,6 +7118,7 @@ static void nfs4_lock_done(struct rpc_task *task, void *calldata)
 {
        struct nfs4_lockdata *data = calldata;
        struct nfs4_lock_state *lsp = data->lsp;
+       struct nfs_server *server = NFS_SERVER(d_inode(data->ctx->dentry));
 
        dprintk("%s: begin!\n", __func__);
 
@@ -7127,8 +7128,7 @@ static void nfs4_lock_done(struct rpc_task *task, void *calldata)
        data->rpc_status = task->tk_status;
        switch (task->tk_status) {
        case 0:
-               renew_lease(NFS_SERVER(d_inode(data->ctx->dentry)),
-                               data->timestamp);
+               renew_lease(server, data->timestamp);
                if (data->arg.new_lock && !data->cancelled) {
                        data->fl.fl_flags &= ~(FL_SLEEP | FL_ACCESS);
                        if (locks_lock_inode_wait(lsp->ls_state->inode, &data->fl) < 0)
@@ -7149,6 +7149,8 @@ static void nfs4_lock_done(struct rpc_task *task, void *calldata)
                        if (!nfs4_stateid_match(&data->arg.open_stateid,
                                                &lsp->ls_state->open_stateid))
                                goto out_restart;
+                       else if (nfs4_async_handle_error(task, server, lsp->ls_state, NULL) == -EAGAIN)
+                               goto out_restart;
                } else if (!nfs4_stateid_match(&data->arg.lock_stateid,
                                                &lsp->ls_stateid))
                                goto out_restart;