nfs: fix a deadlock in nfs client initialization
The following deadlock can occur between a process waiting for a client
to initialize in while walking the client list during nfsv4 server trunking
detection and another process waiting for the nfs_clid_init_mutex so it
can initialize that client:
Process 1 Process 2
--------- ---------
spin_lock(&nn->nfs_client_lock);
list_add_tail(&CLIENTA->cl_share_link,
&nn->nfs_client_list);
spin_unlock(&nn->nfs_client_lock);
spin_lock(&nn->nfs_client_lock);
list_add_tail(&CLIENTB->cl_share_link,
&nn->nfs_client_list);
spin_unlock(&nn->nfs_client_lock);
mutex_lock(&nfs_clid_init_mutex);
nfs41_walk_client_list(clp, result, cred);
nfs_wait_client_init_complete(CLIENTA);
(waiting for nfs_clid_init_mutex)
Make sure nfs_match_client() only evaluates clients that have completed
initialization in order to prevent that deadlock.
This patch also fixes v4.0 trunking behavior by not marking the client
NFS_CS_READY until the clientid has been confirmed.
Signed-off-by: Scott Mayhew <smayhew@redhat.com>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
diff --git a/fs/nfs/client.c b/fs/nfs/client.c
index 0ac2fb1..b9129e2 100644
--- a/fs/nfs/client.c
+++ b/fs/nfs/client.c
@@ -291,12 +291,23 @@ static struct nfs_client *nfs_match_client(const struct nfs_client_initdata *dat
const struct sockaddr *sap = data->addr;
struct nfs_net *nn = net_generic(data->net, nfs_net_id);
+again:
list_for_each_entry(clp, &nn->nfs_client_list, cl_share_link) {
const struct sockaddr *clap = (struct sockaddr *)&clp->cl_addr;
/* Don't match clients that failed to initialise properly */
if (clp->cl_cons_state < 0)
continue;
+ /* If a client is still initializing then we need to wait */
+ if (clp->cl_cons_state > NFS_CS_READY) {
+ refcount_inc(&clp->cl_count);
+ spin_unlock(&nn->nfs_client_lock);
+ nfs_wait_client_init_complete(clp);
+ nfs_put_client(clp);
+ spin_lock(&nn->nfs_client_lock);
+ goto again;
+ }
+
/* Different NFS versions cannot share the same nfs_client */
if (clp->rpc_ops != data->nfs_mod->rpc_ops)
continue;