In the case of v4.0 clients, we may call into the "create" client
tracking operation multiple times (once for each openowner). Upcalling
for each one of those is wasteful and slow however. We can skip doing
further "create" operations after the first one if we know that one has
already been done.
v4.1+ clients generally only call into this function once (on
RECLAIM_COMPLETE), and we can't skip upcalling on the create even if the
STABLE bit is set. Doing so would make it impossible for nfsdcltrack to
lift the grace period early.
Signed-off-by: Jeff Layton <jlayton@xxxxxxxxxxxxxxx>
---
fs/nfsd/nfs4recover.c | 16 ++++++++++++++++
1 file changed, 16 insertions(+)
diff --git a/fs/nfsd/nfs4recover.c b/fs/nfsd/nfs4recover.c
index 5697e9d2e875..6bd97d1d1880 100644
--- a/fs/nfsd/nfs4recover.c
+++ b/fs/nfsd/nfs4recover.c
@@ -1281,6 +1281,22 @@ nfsd4_umh_cltrack_create(struct nfs4_client *clp)
char *hexid, *reclaim_complete, *grace_start;
struct nfsd_net *nn = net_generic(clp->net, nfsd_net_id);
+ /*
+ * With v4.0 clients, there's little difference in outcome between a
+ * create and check operation, and we can end up calling into this
+ * function multiple times per client (once for each openowner). So,
+ * for v4.0 clients skip upcalling once the client has been recorded
+ * on stable storage.
+ *
+ * For v4.1+ clients, the outcome of the two operations is different,
+ * so we must ensure that we upcall for the create operation. v4.1+
+ * clients call this on RECLAIM_COMPLETE though, so we should only end
+ * up doing a single create upcall per client.
+ */
+ if (clp->cl_minorversion == 0 &&
+ test_bit(NFSD4_CLIENT_STABLE, &clp->cl_flags))
+ return;
+
hexid = bin_to_hex_dup(clp->cl_name.data, clp->cl_name.len);
if (!hexid) {
dprintk("%s: can't allocate memory for upcall!\n", __func__);
--
1.9.3
--
To unsubscribe from this list: send the line "unsubscribe linux-nfs" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
