The recent NFS forced unmount improvements introduced a side-effect

where some client operations might be unexpectedly cancelled during
an unsuccessful non-forced unmount attempt. This causes problems
for amd(8), because it periodically attempts a non-forced unmount
to check if the filesystem is still in use.

Fix this by adding a new mountpoint flag MNTK_UNMOUNTF that is set
only during the operation of a forced unmount. Use this instead of
MNTK_UNMOUNT to trigger the cancellation of hung NFS operations.

Also correct a problem where dounmount() might inadvertently clear
the MNTK_UNMOUNT flag.

Reported by:	simokawa
MFC after:	1 week

1 files changed, 4 insertions, 4 deletions

diff --git a/sys/nfsclient/nfs_socket.c b/sys/nfsclient/nfs_socket.c
index 73ecf47..0a4f849 100644
--- a/sys/nfsclient/nfs_socket.c
+++ b/sys/nfsclient/nfs_socket.c
@@ -850,8 +850,8 @@ nfs_request(struct vnode *vp, struct mbuf *mrest, int procnum,
 	int trylater_delay = NQ_TRYLATERDEL, trylater_cnt = 0;
 	u_int32_t xid;
 
-	/* Reject requests while attempting to unmount. */
-	if (vp->v_mount->mnt_kern_flag & MNTK_UNMOUNT) {
+	/* Reject requests while attempting a forced unmount. */
+	if (vp->v_mount->mnt_kern_flag & MNTK_UNMOUNTF) {
 		m_freem(mrest);
 		return (ESTALE);
 	}
@@ -1230,8 +1230,8 @@ nfs_sigintr(struct nfsmount *nmp, struct nfsreq *rep, struct proc *p)
 
 	if (rep && (rep->r_flags & R_SOFTTERM))
 		return (EINTR);
-	/* Terminate all requests while attempting to unmount. */
-	if (nmp->nm_mountp->mnt_kern_flag & MNTK_UNMOUNT)
+	/* Terminate all requests while attempting a forced unmount. */
+	if (nmp->nm_mountp->mnt_kern_flag & MNTK_UNMOUNTF)
 		return (EINTR);
 	if (!(nmp->nm_flag & NFSMNT_INT))
 		return (0);