Don't rely on NSM to help us forget about RPC client handles for

clients that have rebooted (or otherwise changed port numbers). If the
client is broken or has no active locks, it won't notify us. Fall back
on the two minute timeout logic used by the userland rpc.lockd code.

MFC after: 1 week

1 files changed, 17 insertions, 6 deletions

diff --git a/sys/nlm/nlm_prot_impl.c b/sys/nlm/nlm_prot_impl.c
index d48b8a5..8c00665 100644
--- a/sys/nlm/nlm_prot_impl.c
+++ b/sys/nlm/nlm_prot_impl.c
@@ -178,6 +178,7 @@ struct nlm_host {
 	int		nh_state;	 /* (s) last seen NSM state of host */
 	enum nlm_host_state nh_monstate; /* (s) local NSM monitoring state */
 	time_t		nh_idle_timeout; /* (s) Time at which host is idle */
+	time_t		nh_rpc_create_time; /* (s) Time we create RPC client */
 	struct sysctl_ctx_list nh_sysctl; /* (c) vfs.nlm.sysid nodes */
 	struct nlm_async_lock_list nh_pending; /* (l) pending async locks */
 	struct nlm_async_lock_list nh_finished; /* (l) finished async locks */
@@ -953,6 +954,21 @@ nlm_host_get_rpc(struct nlm_host *host)
 {
 	struct timeval zero;
 
+	/*
+ 	 * We can't hold onto RPC handles for too long - the async
+	 * call/reply protocol used by some NLM clients makes it hard
+	 * to tell when they change port numbers (e.g. after a
+	 * reboot). Note that if a client reboots while it isn't
+	 * holding any locks, it won't bother to notify us. We
+	 * expire the RPC handles after two minutes.
+	 */
+	if (host->nh_rpc && time_uptime > host->nh_rpc_create_time + 2*60) {
+		CLIENT *client;
+		client = host->nh_rpc;
+		host->nh_rpc = NULL;
+		CLNT_DESTROY(client);
+	}
+
 	if (host->nh_rpc)
 		return (host->nh_rpc);
 
@@ -968,12 +984,7 @@ nlm_host_get_rpc(struct nlm_host *host)
 		zero.tv_usec = 0;
 		CLNT_CONTROL(host->nh_rpc, CLSET_TIMEOUT, &zero);
 
-		/*
-		 * Monitor the host - if it reboots, the address of
-		 * its NSM might change so we must discard our RPC
-		 * handle.
-		 */
-		nlm_host_monitor(host, 0);
+		host->nh_rpc_create_time = time_uptime;
 	}
 
 	return (host->nh_rpc);