nfsd: destroy percpu stats counters after reply cache shutdown

[ Upstream commit fd5e363 ]

Upon nfsd shutdown any pending DRC cache is freed. DRC cache use is
tracked via a percpu counter. In the current code the percpu counter
is destroyed before. If any pending cache is still present,
percpu_counter_add is called with a percpu counter==NULL. This causes
a kernel crash.
The solution is to destroy the percpu counter after the cache is freed.

Fixes: e567b98 (“nfsd: protect concurrent access to nfsd stats counters”)
Signed-off-by: Julian Schroeder <jumaco@amazon.com>
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

fs/nfsd/nfscache.c

@@ -212,7 +212,6 @@ void nfsd_reply_cache_shutdown(struct nfsd_net *nn)
 	struct svc_cacherep	*rp;
 	unsigned int i;
 
-	nfsd_reply_cache_stats_destroy(nn);
 	unregister_shrinker(&nn->nfsd_reply_cache_shrinker);
 
 	for (i = 0; i < nn->drc_hashsize; i++) {
@@ -223,6 +222,7 @@ void nfsd_reply_cache_shutdown(struct nfsd_net *nn)
 									rp, nn);
 		}
 	}
+	nfsd_reply_cache_stats_destroy(nn);
 
 	kvfree(nn->drc_hashtbl);
 	nn->drc_hashtbl = NULL;