When the old nfsm_adv() macro was moved to nfsm_adv_xx(), a '>='

must have been inadvertently changed to '>'. This broke nfsm_adv()
in the case where the advancement count is equal to the amount of
data remaining in the current mbuf. Instead of moving the current
position N bytes forward, nfs_adv() could end up moving it back to
N bytes from the start of the mbuf data.

This should fix the client-side readdirplus problems that have been
reported since September.

1 files changed, 1 insertions, 1 deletions

diff --git a/sys/nfs/nfs_common.c b/sys/nfs/nfs_common.c
index d4a957e..4b9aa16 100644
--- a/sys/nfs/nfs_common.c
+++ b/sys/nfs/nfs_common.c
@@ -305,7 +305,7 @@ nfsm_adv_xx(int s, struct mbuf **md, caddr_t *dpos)
 	int t1;
 
 	t1 = mtod(*md, caddr_t) + (*md)->m_len - *dpos;
-	if (t1 > s) {
+	if (t1 >= s) {
 		*dpos += s;
 		return 0;
 	}