- Fix nullfs vnode reference leak in nullfs_reclaim_lowervp(). The

  null_hashget() obtains the reference on the nullfs vnode, which must
  be dropped.

- Fix a wart which existed from the introduction of the nullfs
  caching, do not unlock lower vnode in the nullfs_reclaim_lowervp().
  It should be innocent, but now it is also formally safe.  Inform the
  nullfs_reclaim() about this using the NULLV_NOUNLOCK flag set on
  nullfs inode.

- Add a callback to the upper filesystems for the lower vnode
  unlinking. When inactivating a nullfs vnode, check if the lower
  vnode was unlinked, indicated by nullfs flag NULLV_DROP or VV_NOSYNC
  on the lower vnode, and reclaim upper vnode if so.  This allows
  nullfs to purge cached vnodes for the unlinked lower vnode, avoiding
  excessive caching.

Reported by:	G??ran L??wkrantz <goran.lowkrantz@ismobile.com>
Tested by:	pho
Sponsored by:	The FreeBSD Foundation
MFC after:	2 weeks

2 files changed, 20 insertions, 7 deletions

diff --git a/sys/kern/vfs_subr.c b/sys/kern/vfs_subr.c
index d9cbde2..7f72ed1 100644
--- a/sys/kern/vfs_subr.c
+++ b/sys/kern/vfs_subr.c
@@ -2700,19 +2700,20 @@ vgone(struct vnode *vp)
 }
 
 static void
-vgonel_reclaim_lowervp_vfs(struct mount *mp __unused,
+notify_lowervp_vfs_dummy(struct mount *mp __unused,
     struct vnode *lowervp __unused)
 {
 }
 
 /*
- * Notify upper mounts about reclaimed vnode.
+ * Notify upper mounts about reclaimed or unlinked vnode.
  */
-static void
-vgonel_reclaim_lowervp(struct vnode *vp)
+void
+vfs_notify_upper(struct vnode *vp, int event)
 {
 	static struct vfsops vgonel_vfsops = {
-		.vfs_reclaim_lowervp = vgonel_reclaim_lowervp_vfs
+		.vfs_reclaim_lowervp = notify_lowervp_vfs_dummy,
+		.vfs_unlink_lowervp = notify_lowervp_vfs_dummy,
 	};
 	struct mount *mp, *ump, *mmp;
 
@@ -2736,7 +2737,17 @@ vgonel_reclaim_lowervp(struct vnode *vp)
 		}
 		TAILQ_INSERT_AFTER(&mp->mnt_uppers, ump, mmp, mnt_upper_link);
 		MNT_IUNLOCK(mp);
-		VFS_RECLAIM_LOWERVP(ump, vp);
+		switch (event) {
+		case VFS_NOTIFY_UPPER_RECLAIM:
+			VFS_RECLAIM_LOWERVP(ump, vp);
+			break;
+		case VFS_NOTIFY_UPPER_UNLINK:
+			VFS_UNLINK_LOWERVP(ump, vp);
+			break;
+		default:
+			KASSERT(0, ("invalid event %d", event));
+			break;
+		}
 		MNT_ILOCK(mp);
 		ump = TAILQ_NEXT(mmp, mnt_upper_link);
 		TAILQ_REMOVE(&mp->mnt_uppers, mmp, mnt_upper_link);
@@ -2783,7 +2794,7 @@ vgonel(struct vnode *vp)
 	active = vp->v_usecount;
 	oweinact = (vp->v_iflag & VI_OWEINACT);
 	VI_UNLOCK(vp);
-	vgonel_reclaim_lowervp(vp);
+	vfs_notify_upper(vp, VFS_NOTIFY_UPPER_RECLAIM);
 
 	/*
 	 * Clean out any buffers associated with the vnode.
diff --git a/sys/kern/vfs_syscalls.c b/sys/kern/vfs_syscalls.c
index 29cb7bd..a004ea0 100644
--- a/sys/kern/vfs_syscalls.c
+++ b/sys/kern/vfs_syscalls.c
@@ -1846,6 +1846,7 @@ restart:
 		if (error)
 			goto out;
 #endif
+		vfs_notify_upper(vp, VFS_NOTIFY_UPPER_UNLINK);
 		error = VOP_REMOVE(nd.ni_dvp, vp, &nd.ni_cnd);
 #ifdef MAC
 out:
@@ -3825,6 +3826,7 @@ restart:
 			return (error);
 		goto restart;
 	}
+	vfs_notify_upper(vp, VFS_NOTIFY_UPPER_UNLINK);
 	error = VOP_RMDIR(nd.ni_dvp, nd.ni_vp, &nd.ni_cnd);
 	vn_finished_write(mp);
 out: