Various cleanups towards make nullfs functional (it is still broken

at this point):

Replace all '#ifdef DEBUG' with '#ifdef NULLFS_DEBUG' and add NULLFSDEBUG
macro.

Protect nullfs hash table with lockmgr.

Use proper order of operations when freeing mnt_data.

Return correct fsid in the null_getattr().

Add null_open() function to catch MNT_NODEV (obtained from NetBSD).

Add null_rename() to catch cross-fs rename operations (submitted by
Ustimenko Semen <semen@iclub.nsu.ru>)

Remove duplicate $FreeBSD$ tags.

4 files changed, 130 insertions, 38 deletions

diff --git a/sys/fs/nullfs/null.h b/sys/fs/nullfs/null.h
index f825ee8..a5ebf87 100644
--- a/sys/fs/nullfs/null.h
+++ b/sys/fs/nullfs/null.h
@@ -62,6 +62,7 @@ struct null_node {
 #define	NULLTOV(xp) ((xp)->null_vnode)
 
 int nullfs_init(struct vfsconf *vfsp);
+int nullfs_uninit(struct vfsconf *vfsp);
 int null_node_create(struct mount *mp, struct vnode *target, struct vnode **vpp);
 int null_bypass(struct vop_generic_args *ap);
 
@@ -73,5 +74,16 @@ struct vnode *null_checkvp(struct vnode *vp, char *fil, int lno);
 #endif
 
 extern vop_t **null_vnodeop_p;
+extern struct lock null_hashlock;
+
+#ifdef MALLOC_DECLARE
+MALLOC_DECLARE(M_NULLFSNODE);
+#endif
+
+#ifdef NULLFS_DEBUG
+#define NULLFSDEBUG(format, args...) printf(format ,## args)
+#else
+#define NULLFSDEBUG(format, args...)
+#endif /* NULLFS_DEBUG */
 
 #endif /* _KERNEL */
diff --git a/sys/fs/nullfs/null_subr.c b/sys/fs/nullfs/null_subr.c
index 900bd15..3671f0a 100644
--- a/sys/fs/nullfs/null_subr.c
+++ b/sys/fs/nullfs/null_subr.c
@@ -40,6 +40,7 @@
 
 #include <sys/param.h>
 #include <sys/systm.h>
+#include <sys/kernel.h>
 #include <sys/proc.h>
 #include <sys/vnode.h>
 #include <sys/mount.h>
@@ -59,8 +60,13 @@
 
 #define	NULL_NHASH(vp) \
 	(&null_node_hashtbl[(((uintptr_t)vp)>>LOG2_SIZEVNODE) & null_node_hash])
+
 static LIST_HEAD(null_node_hashhead, null_node) *null_node_hashtbl;
 static u_long null_node_hash;
+struct lock null_hashlock;
+
+static MALLOC_DEFINE(M_NULLFSHASH, "NULLFS hash", "NULLFS hash table");
+MALLOC_DEFINE(M_NULLFSNODE, "NULLFS node", "NULLFS vnode private part");
 
 static int	null_node_alloc(struct mount *mp, struct vnode *lowervp,
 				     struct vnode **vpp);
@@ -75,10 +81,19 @@ nullfs_init(vfsp)
 	struct vfsconf *vfsp;
 {
 
-#ifdef DEBUG
-	printf("nullfs_init\n");		/* printed during system boot */
-#endif
-	null_node_hashtbl = hashinit(NNULLNODECACHE, M_CACHE, &null_node_hash);
+	NULLFSDEBUG("nullfs_init\n");		/* printed during system boot */
+	null_node_hashtbl = hashinit(NNULLNODECACHE, M_NULLFSHASH, &null_node_hash);
+	lockinit(&null_hashlock, PVFS, "nullhs", 0, 0);
+	return (0);
+}
+
+int
+nullfs_uninit(vfsp)
+	struct vfsconf *vfsp;
+{
+
+        if (null_node_hashtbl)
+		free(null_node_hashtbl, M_NULLFSHASH);
 	return (0);
 }
 
@@ -103,9 +118,11 @@ null_node_find(mp, lowervp)
 	 */
 	hd = NULL_NHASH(lowervp);
 loop:
+	lockmgr(&null_hashlock, LK_EXCLUSIVE, NULL, p);
 	for (a = hd->lh_first; a != 0; a = a->null_hash.le_next) {
 		if (a->null_lowervp == lowervp && NULLTOV(a)->v_mount == mp) {
 			vp = NULLTOV(a);
+			lockmgr(&null_hashlock, LK_RELEASE, NULL, p);
 			/*
 			 * We need vget for the VXLOCK
 			 * stuff, but we don't want to lock
@@ -118,6 +135,7 @@ loop:
 			return (vp);
 		}
 	}
+	lockmgr(&null_hashlock, LK_RELEASE, NULL, p);
 
 	return NULLVP;
 }
@@ -134,6 +152,7 @@ null_node_alloc(mp, lowervp, vpp)
 	struct vnode *lowervp;
 	struct vnode **vpp;
 {
+	struct proc *p = curproc;	/* XXX */
 	struct null_node_hashhead *hd;
 	struct null_node *xp;
 	struct vnode *othervp, *vp;
@@ -144,11 +163,12 @@ null_node_alloc(mp, lowervp, vpp)
 	 * might cause a bogus v_data pointer to get dereferenced
 	 * elsewhere if MALLOC should block.
 	 */
-	MALLOC(xp, struct null_node *, sizeof(struct null_node), M_TEMP, M_WAITOK);
+	MALLOC(xp, struct null_node *, sizeof(struct null_node),
+	    M_NULLFSNODE, M_WAITOK);
 
 	error = getnewvnode(VT_NULL, mp, null_vnodeop_p, vpp);
 	if (error) {
-		FREE(xp, M_TEMP);
+		FREE(xp, M_NULLFSNODE);
 		return (error);
 	}
 	vp = *vpp;
@@ -164,15 +184,17 @@ null_node_alloc(mp, lowervp, vpp)
 	 */
 	othervp = null_node_find(mp, lowervp);
 	if (othervp) {
-		FREE(xp, M_TEMP);
+		FREE(xp, M_NULLFSNODE);
 		vp->v_type = VBAD;	/* node is discarded */
 		vp->v_usecount = 0;	/* XXX */
 		*vpp = othervp;
 		return 0;
 	};
+	lockmgr(&null_hashlock, LK_EXCLUSIVE, NULL, p);
 	VREF(lowervp);   /* Extra VREF will be vrele'd in null_node_create */
 	hd = NULL_NHASH(lowervp);
 	LIST_INSERT_HEAD(hd, xp, null_hash);
+	lockmgr(&null_hashlock, LK_RELEASE, NULL, p);
 	return 0;
 }
 
@@ -196,7 +218,7 @@ null_node_create(mp, lowervp, newvpp)
 		 * null_node_find has taken another reference
 		 * to the alias vnode.
 		 */
-#ifdef DEBUG
+#ifdef NULLFS_DEBUG
 		vprint("null_node_create: exists", aliasvp);
 #endif
 		/* VREF(aliasvp); --- done in null_node_find */
@@ -206,9 +228,7 @@ null_node_create(mp, lowervp, newvpp)
 		/*
 		 * Get new vnode.
 		 */
-#ifdef DEBUG
-		printf("null_node_create: create new alias vnode\n");
-#endif
+		NULLFSDEBUG("null_node_create: create new alias vnode\n");
 
 		/*
 		 * Make new vnode reference the null_node.
@@ -233,7 +253,7 @@ null_node_create(mp, lowervp, newvpp)
 	};
 #endif
 
-#ifdef DEBUG
+#ifdef NULLFS_DEBUG
 	vprint("null_node_create: alias", aliasvp);
 	vprint("null_node_create: lower", lowervp);
 #endif
diff --git a/sys/fs/nullfs/null_vfsops.c b/sys/fs/nullfs/null_vfsops.c
index 754a121..5b49fbb 100644
--- a/sys/fs/nullfs/null_vfsops.c
+++ b/sys/fs/nullfs/null_vfsops.c
@@ -93,9 +93,7 @@ nullfs_mount(mp, path, data, ndp, p)
 	u_int size;
 	int isvnunlocked = 0;
 
-#ifdef DEBUG
-	printf("nullfs_mount(mp = %p)\n", (void *)mp);
-#endif
+	NULLFSDEBUG("nullfs_mount(mp = %p)\n", (void *)mp);
 
 	/*
 	 * Update is a no-op
@@ -149,9 +147,7 @@ nullfs_mount(mp, path, data, ndp, p)
 	 * Check multi null mount to avoid `lock against myself' panic.
 	 */
 	if (lowerrootvp == VTONULL(mp->mnt_vnodecovered)->null_lowervp) {
-#ifdef DEBUG
-		printf("nullfs_mount: multi null mount?\n");
-#endif
+		NULLFSDEBUG("nullfs_mount: multi null mount?\n");
 		return (EDEADLK);
 	}
 
@@ -199,10 +195,8 @@ nullfs_mount(mp, path, data, ndp, p)
 	    &size);
 	bzero(mp->mnt_stat.f_mntfromname + size, MNAMELEN - size);
 	(void)nullfs_statfs(mp, &mp->mnt_stat, p);
-#ifdef DEBUG
-	printf("nullfs_mount: lower %s, alias at %s\n",
+	NULLFSDEBUG("nullfs_mount: lower %s, alias at %s\n",
 		mp->mnt_stat.f_mntfromname, mp->mnt_stat.f_mntonname);
-#endif
 	return (0);
 }
 
@@ -231,12 +225,11 @@ nullfs_unmount(mp, mntflags, p)
 	struct proc *p;
 {
 	struct vnode *nullm_rootvp = MOUNTTONULLMOUNT(mp)->nullm_rootvp;
+	void *mntdata;
 	int error;
 	int flags = 0;
 
-#ifdef DEBUG
-	printf("nullfs_unmount(mp = %p)\n", (void *)mp);
-#endif
+	NULLFSDEBUG("nullfs_unmount: mp = %p\n", (void *)mp);
 
 	if (mntflags & MNT_FORCE)
 		flags |= FORCECLOSE;
@@ -257,7 +250,7 @@ nullfs_unmount(mp, mntflags, p)
 	if (error)
 		return (error);
 
-#ifdef DEBUG
+#ifdef NULLFS_DEBUG
 	vprint("alias root of lower", nullm_rootvp);
 #endif
 	/*
@@ -271,8 +264,9 @@ nullfs_unmount(mp, mntflags, p)
 	/*
 	 * Finally, throw away the null_mount structure
 	 */
-	free(mp->mnt_data, M_NULLFSMNT);	/* XXX */
+	mntdata = mp->mnt_data;
 	mp->mnt_data = 0;
+	free(mntdata, M_NULLFSMNT);	/* XXX */
 	return 0;
 }
 
@@ -284,29 +278,27 @@ nullfs_root(mp, vpp)
 	struct proc *p = curproc;	/* XXX */
 	struct vnode *vp;
 
-#ifdef DEBUG
-	printf("nullfs_root(mp = %p, vp = %p->%p)\n", (void *)mp,
+	NULLFSDEBUG("nullfs_root(mp = %p, vp = %p->%p)\n", (void *)mp,
 	    (void *)MOUNTTONULLMOUNT(mp)->nullm_rootvp,
 	    (void *)NULLVPTOLOWERVP(MOUNTTONULLMOUNT(mp)->nullm_rootvp));
-#endif
 
 	/*
 	 * Return locked reference to root.
 	 */
 	vp = MOUNTTONULLMOUNT(mp)->nullm_rootvp;
 	VREF(vp);
+#ifdef NULLFS_DEBUG
 	if (VOP_ISLOCKED(vp, NULL)) {
 		/*
 		 * XXX
 		 * Should we check type of node?
 		 */
-#ifdef DEBUG
 		printf("nullfs_root: multi null mount?\n");
-#endif
 		vrele(vp);
 		return (EDEADLK);
-	} else
-		vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, p);
+	}
+#endif
+	vn_lock(vp, LK_EXCLUSIVE | LK_RETRY, p);
 	*vpp = vp;
 	return 0;
 }
@@ -331,11 +323,9 @@ nullfs_statfs(mp, sbp, p)
 	int error;
 	struct statfs mstat;
 
-#ifdef DEBUG
-	printf("nullfs_statfs(mp = %p, vp = %p->%p)\n", (void *)mp,
+	NULLFSDEBUG("nullfs_statfs(mp = %p, vp = %p->%p)\n", (void *)mp,
 	    (void *)MOUNTTONULLMOUNT(mp)->nullm_rootvp,
 	    (void *)NULLVPTOLOWERVP(MOUNTTONULLMOUNT(mp)->nullm_rootvp));
-#endif
 
 	bzero(&mstat, sizeof(mstat));
 
@@ -440,7 +430,7 @@ static struct vfsops null_vfsops = {
 	nullfs_checkexp,
 	nullfs_vptofh,
 	nullfs_init,
-	vfs_stduninit,
+	nullfs_uninit,
 	nullfs_extattrctl,
 };
 
diff --git a/sys/fs/nullfs/null_vnops.c b/sys/fs/nullfs/null_vnops.c
index 8329d5e..5692df6 100644
--- a/sys/fs/nullfs/null_vnops.c
+++ b/sys/fs/nullfs/null_vnops.c
@@ -37,7 +37,6 @@
  *
  * Ancestors:
  *	@(#)lofs_vnops.c	1.2 (Berkeley) 6/18/92
- * $FreeBSD$
  *	...and...
  *	@(#)null_vnodeops.c 1.20 92/07/07 UCLA Ficus project
  *
@@ -193,8 +192,10 @@ static int	null_getattr(struct vop_getattr_args *ap);
 static int	null_inactive(struct vop_inactive_args *ap);
 static int	null_lock(struct vop_lock_args *ap);
 static int	null_lookup(struct vop_lookup_args *ap);
+static int	null_open(struct vop_open_args *ap);
 static int	null_print(struct vop_print_args *ap);
 static int	null_reclaim(struct vop_reclaim_args *ap);
+static int	null_rename(struct vop_rename_args *ap);
 static int	null_setattr(struct vop_setattr_args *ap);
 static int	null_unlock(struct vop_unlock_args *ap);
 
@@ -448,6 +449,8 @@ null_getattr(ap)
 
 	if ((error = null_bypass((struct vop_generic_args *)ap)) != 0)
 		return (error);
+
+	ap->a_vap->va_fsid = ap->a_vp->v_mount->mnt_stat.f_fsid.val[0];
 	return (0);
 }
 
@@ -484,6 +487,66 @@ null_access(ap)
 }
 
 /*
+ * We must handle open to be able to catch MNT_NODEV and friends.
+ */
+static int
+null_open(ap)
+	struct vop_open_args /* {
+		struct vnode *a_vp;
+		int  a_mode;
+		struct ucred *a_cred;
+		struct proc *a_p;
+	} */ *ap;
+{
+	struct vnode *vp = ap->a_vp;
+	struct vnode *lvp = NULLVPTOLOWERVP(ap->a_vp);
+
+	if ((vp->v_mount->mnt_flag & MNT_NODEV) &&
+	    (lvp->v_type == VBLK || lvp->v_type == VCHR))
+		return ENXIO;
+
+	return (null_bypass((struct vop_generic_args *)ap));
+}
+
+/*
+ * We handle this to eliminate null FS to lower FS
+ * file moving. Don't know why we don't allow this,
+ * possibly we should.
+ */
+static int
+null_rename(ap)
+	struct vop_rename_args /* {
+		struct vnode *a_fdvp;
+		struct vnode *a_fvp;
+		struct componentname *a_fcnp;
+		struct vnode *a_tdvp;
+		struct vnode *a_tvp;
+		struct componentname *a_tcnp;
+	} */ *ap;
+{
+	struct vnode *tdvp = ap->a_tdvp;
+	struct vnode *fvp = ap->a_fvp;
+	struct vnode *fdvp = ap->a_fdvp;
+	struct vnode *tvp = ap->a_tvp;
+
+	/* Check for cross-device rename. */
+	if ((fvp->v_mount != tdvp->v_mount) ||
+	    (tvp && (fvp->v_mount != tvp->v_mount))) {
+		if (tdvp == tvp)
+			vrele(tdvp);
+		else
+			vput(tdvp);
+		if (tvp)
+			vput(tvp);
+		vrele(fdvp);
+		vrele(fvp);
+		return (EXDEV);
+	}
+	
+	return (null_bypass((struct vop_generic_args *)ap));
+}
+
+/*
  * We need to process our own vnode lock and then clear the
  * interlock flag as it applies only to our vnode, not the
  * vnodes below us on the stack.
@@ -568,7 +631,9 @@ null_reclaim(ap)
 	 */
 	/* After this assignment, this node will not be re-used. */
 	xp->null_lowervp = NULLVP;
+	lockmgr(&null_hashlock, LK_EXCLUSIVE, NULL, ap->a_p);
 	LIST_REMOVE(xp, null_hash);
+	lockmgr(&null_hashlock, LK_RELEASE, NULL, ap->a_p);
 	FREE(vp->v_data, M_TEMP);
 	vp->v_data = NULL;
 	vrele (lowervp);
@@ -593,13 +658,18 @@ vop_t **null_vnodeop_p;
 static struct vnodeopv_entry_desc null_vnodeop_entries[] = {
 	{ &vop_default_desc,		(vop_t *) null_bypass },
 	{ &vop_access_desc,		(vop_t *) null_access },
+	{ &vop_bmap_desc,		(vop_t *) vop_eopnotsupp },
 	{ &vop_getattr_desc,		(vop_t *) null_getattr },
+	{ &vop_getwritemount_desc,	(vop_t *) vop_stdgetwritemount},
 	{ &vop_inactive_desc,		(vop_t *) null_inactive },
 	{ &vop_lock_desc,		(vop_t *) null_lock },
 	{ &vop_lookup_desc,		(vop_t *) null_lookup },
+	{ &vop_open_desc,		(vop_t *) null_open },
 	{ &vop_print_desc,		(vop_t *) null_print },
 	{ &vop_reclaim_desc,		(vop_t *) null_reclaim },
+	{ &vop_rename_desc,		(vop_t *) null_rename },
 	{ &vop_setattr_desc,		(vop_t *) null_setattr },
+	{ &vop_strategy_desc,		(vop_t *) vop_eopnotsupp },
 	{ &vop_unlock_desc,		(vop_t *) null_unlock },
 	{ NULL, NULL }
 };