[PATCH] intrinsic automount and mountpoint degradation support

[David Howells <dhowells@redhat.com>]

Hi Linus, Andrew,

Here's a patch that I worked out with Al Viro that adds support for a
filesystem (such as kAFS) to perform automounting intrinsically without the
need for a userspace daemon. It also adds support for such mountpoints to be
degraded at the filesystem's behest until they've been untouched long enough
that they'll be removed.

I've a patch (to follow) that removes some #ifdef's from fs/afs/* thus
allowing it to make use of this facility.

David


diff -uNr linux-2.6.7/fs/namei.c linux-2.6.7-auto/fs/namei.c
--- linux-2.6.7/fs/namei.c	2004-06-18 13:44:02.000000000 +0100
+++ linux-2.6.7-auto/fs/namei.c	2004-06-25 13:48:28.000000000 +0100
@@ -278,6 +278,12 @@
 	mntput(nd->mnt);
 }
 
+void path_release_on_umount(struct nameidata *nd)
+{
+	dput(nd->dentry);
+	_mntput(nd->mnt);
+}
+
 /*
  * Internal lookup() using the new generic dcache.
  * SMP-safe
diff -uNr linux-2.6.7/fs/namespace.c linux-2.6.7-auto/fs/namespace.c
--- linux-2.6.7/fs/namespace.c	2004-06-18 13:44:02.000000000 +0100
+++ linux-2.6.7-auto/fs/namespace.c	2004-06-30 19:43:14.565098078 +0100
@@ -60,6 +60,7 @@
 		INIT_LIST_HEAD(&mnt->mnt_child);
 		INIT_LIST_HEAD(&mnt->mnt_mounts);
 		INIT_LIST_HEAD(&mnt->mnt_list);
+		INIT_LIST_HEAD(&mnt->mnt_fslink);
 		if (name) {
 			int size = strlen(name)+1;
 			char *newname = kmalloc(size, GFP_KERNEL);
@@ -106,13 +107,9 @@
 
 EXPORT_SYMBOL(lookup_mnt);
 
-static int check_mnt(struct vfsmount *mnt)
+static inline int check_mnt(struct vfsmount *mnt)
 {
-	spin_lock(&vfsmount_lock);
-	while (mnt->mnt_parent != mnt)
-		mnt = mnt->mnt_parent;
-	spin_unlock(&vfsmount_lock);
-	return mnt == current->namespace->root;
+	return mnt->mnt_namespace == current->namespace;
 }
 
 static void detach_mnt(struct vfsmount *mnt, struct nameidata *old_nd)
@@ -164,6 +161,11 @@
 		mnt->mnt_root = dget(root);
 		mnt->mnt_mountpoint = mnt->mnt_root;
 		mnt->mnt_parent = mnt;
+		mnt->mnt_namespace = old->mnt_namespace;
+		spin_lock(&vfsmount_lock);
+		if (!list_empty(&old->mnt_fslink))
+			list_add(&mnt->mnt_fslink, &old->mnt_fslink);
+		spin_unlock(&vfsmount_lock);
 	}
 	return mnt;
 }
@@ -346,6 +348,7 @@
 	while (!list_empty(&kill)) {
 		mnt = list_entry(kill.next, struct vfsmount, mnt_list);
 		list_del_init(&mnt->mnt_list);
+		list_del_init(&mnt->mnt_fslink);
 		if (mnt->mnt_parent == mnt) {
 			spin_unlock(&vfsmount_lock);
 		} else {
@@ -369,6 +372,24 @@
 		return retval;
 
 	/*
+	 * If we've been asked to mark for expiry, we only _actually_ effect
+	 * the unmount if:
+	 *  (1) the mark is already set (the mark is cleared by mntput())
+	 *  (2) the usage count == 1 [parent vfsmount] + 1 [sys_umount]
+	 */
+	if (flags & MNT_EXPIRE) {
+		if (mnt == current->fs->rootmnt ||
+		    flags & (MNT_FORCE | MNT_DETACH))
+			return -EINVAL;
+
+		if (atomic_read(&mnt->mnt_count) != 2)
+			return -EBUSY;
+
+		if (!xchg(&mnt->mnt_expiry_mark, 1))
+			return -EAGAIN;
+	}
+
+	/*
 	 * If we may have to abort operations to get out of this
 	 * mount, and they will themselves hold resources we must
 	 * allow the fs to do things. In the Unix tradition of
@@ -461,7 +482,7 @@
 
 	retval = do_umount(nd.mnt, flags);
 dput_and_out:
-	path_release(&nd);
+	path_release_on_umount(&nd);
 out:
 	return retval;
 }
@@ -618,6 +639,10 @@
 	}
 
 	if (mnt) {
+		spin_lock(&vfsmount_lock);
+		list_del_init(&mnt->mnt_fslink);
+		spin_unlock(&vfsmount_lock);
+
 		err = graft_tree(mnt, nd);
 		if (err) {
 			spin_lock(&vfsmount_lock);
@@ -638,7 +663,8 @@
  * on it - tough luck.
  */
 
-static int do_remount(struct nameidata *nd,int flags,int mnt_flags,void *data)
+static int do_remount(struct nameidata *nd, int flags, int mnt_flags,
+		      void *data)
 {
 	int err;
 	struct super_block * sb = nd->mnt->mnt_sb;
@@ -710,6 +736,7 @@
 
 	detach_mnt(old_nd.mnt, &parent_nd);
 	attach_mnt(old_nd.mnt, nd);
+	list_del_init(&old_nd.mnt->mnt_fslink);
 out2:
 	spin_unlock(&vfsmount_lock);
 out1:
@@ -722,11 +749,10 @@
 	return err;
 }
 
-static int do_add_mount(struct nameidata *nd, char *type, int flags,
+static int do_new_mount(struct nameidata *nd, char *type, int flags,
 			int mnt_flags, char *name, void *data)
 {
 	struct vfsmount *mnt;
-	int err;
 
 	if (!type || !memchr(type, 0, PAGE_SIZE))
 		return -EINVAL;
@@ -736,9 +762,16 @@
 		return -EPERM;
 
 	mnt = do_kern_mount(type, flags, name, data);
-	err = PTR_ERR(mnt);
 	if (IS_ERR(mnt))
-		goto out;
+		return PTR_ERR(mnt);
+
+	return do_add_mount(mnt, nd, mnt_flags, NULL);
+}
+
+int do_add_mount(struct vfsmount *newmnt, struct nameidata *nd,
+		 int mnt_flags, struct list_head *fslist)
+{
+	int err;
 
 	down_write(&current->namespace->sem);
 	/* Something was mounted here while we slept */
@@ -750,22 +783,111 @@
 
 	/* Refuse the same filesystem on the same mount point */
 	err = -EBUSY;
-	if (nd->mnt->mnt_sb == mnt->mnt_sb && nd->mnt->mnt_root == nd->dentry)
+	if (nd->mnt->mnt_sb == newmnt->mnt_sb &&
+	    nd->mnt->mnt_root == nd->dentry)
 		goto unlock;
 
 	err = -EINVAL;
-	if (S_ISLNK(mnt->mnt_root->d_inode->i_mode))
+	if (S_ISLNK(newmnt->mnt_root->d_inode->i_mode))
 		goto unlock;
 
-	mnt->mnt_flags = mnt_flags;
-	err = graft_tree(mnt, nd);
+	newmnt->mnt_flags = mnt_flags;
+	err = graft_tree(newmnt, nd);
+
+	if (err == 0 && fslist) {
+		spin_lock(&vfsmount_lock);
+		list_add_tail(&newmnt->mnt_fslink, fslist);
+		spin_unlock(&vfsmount_lock);
+	}
+
 unlock:
 	up_write(&current->namespace->sem);
-	mntput(mnt);
-out:
+	mntput(newmnt);
 	return err;
 }
 
+EXPORT_SYMBOL_GPL(do_add_mount);
+
+void mark_mounts_for_expiry(struct list_head *mounts)
+{
+	struct namespace *namespace;
+	struct list_head graveyard, *_p, *_n;
+	struct vfsmount *mnt;
+
+	INIT_LIST_HEAD(&graveyard);
+
+	spin_lock(&vfsmount_lock);
+
+	list_for_each_safe(_p, _n, mounts) {
+		mnt = list_entry(_p, struct vfsmount, mnt_fslink);
+
+		if (!xchg(&mnt->mnt_expiry_mark, 1) ||
+		    atomic_read(&mnt->mnt_count) != 1)
+			continue;
+
+		mntget(mnt);
+		list_move(&mnt->mnt_fslink, &graveyard);
+	}
+
+	while (!list_empty(&graveyard)) {
+		mnt = list_entry(graveyard.next, struct vfsmount, mnt_fslink);
+		list_del_init(&mnt->mnt_fslink);
+
+		namespace = mnt->mnt_namespace;
+		if (!namespace || atomic_read(&namespace->count) <= 0)
+			continue;
+		get_namespace(namespace);
+
+		spin_unlock(&vfsmount_lock);
+		down_write(&namespace->sem);
+		spin_lock(&vfsmount_lock);
+
+		if (atomic_read(&mnt->mnt_count) == 2) {
+			struct vfsmount *xdmnt;
+			struct dentry *xdentry;
+
+			list_del_init(&mnt->mnt_list);
+			list_del_init(&mnt->mnt_child);
+			list_del_init(&mnt->mnt_hash);
+			mnt->mnt_mountpoint->d_mounted--;
+
+			xdentry = mnt->mnt_mountpoint;
+			mnt->mnt_mountpoint = mnt->mnt_root;
+			xdmnt = mnt->mnt_parent;
+			mnt->mnt_parent = mnt;
+			spin_unlock(&vfsmount_lock);
+
+			mntput(xdmnt);
+			dput(xdentry);
+
+			if (atomic_read(&mnt->mnt_sb->s_active) == 1) {
+				/* last instance - try to be smart */
+				lock_kernel();
+				DQUOT_OFF(mnt->mnt_sb);
+				acct_auto_close(mnt->mnt_sb);
+				unlock_kernel();
+			}
+
+			mntput(mnt);
+		}
+		else {
+			list_add_tail(&mnt->mnt_fslink, mounts);
+			spin_unlock(&vfsmount_lock);
+		}
+
+		up_write(&namespace->sem);
+
+		mntput(mnt);
+		put_namespace(namespace);
+
+		spin_lock(&vfsmount_lock);
+	}
+
+	spin_unlock(&vfsmount_lock);
+}
+
+EXPORT_SYMBOL_GPL(mark_mounts_for_expiry);
+
 int copy_mount_options (const void __user *data, unsigned long *where)
 {
 	int i;
@@ -860,7 +982,7 @@
 	else if (flags & MS_MOVE)
 		retval = do_move_mount(&nd, dev_name);
 	else
-		retval = do_add_mount(&nd, type_page, flags, mnt_flags,
+		retval = do_new_mount(&nd, type_page, flags, mnt_flags,
 				      dev_name, data_page);
 dput_out:
 	path_release(&nd);
@@ -1185,6 +1307,7 @@
 	init_rwsem(&namespace->sem);
 	list_add(&mnt->mnt_list, &namespace->list);
 	namespace->root = mnt;
+	mnt->mnt_namespace = namespace;
 
 	init_task.namespace = namespace;
 	read_lock(&tasklist_lock);
@@ -1252,8 +1375,15 @@
 
 void __put_namespace(struct namespace *namespace)
 {
+	struct vfsmount *mnt;
+
 	down_write(&namespace->sem);
 	spin_lock(&vfsmount_lock);
+
+	list_for_each_entry(mnt, &namespace->list, mnt_list) {
+		mnt->mnt_namespace = NULL;
+	}
+
 	umount_tree(namespace->root);
 	spin_unlock(&vfsmount_lock);
 	up_write(&namespace->sem);
diff -uNr linux-2.6.7/fs/super.c linux-2.6.7-auto/fs/super.c
--- linux-2.6.7/fs/super.c	2004-06-18 13:44:04.000000000 +0100
+++ linux-2.6.7-auto/fs/super.c	2004-06-25 13:50:04.000000000 +0100
@@ -788,6 +788,7 @@
 	mnt->mnt_root = dget(sb->s_root);
 	mnt->mnt_mountpoint = sb->s_root;
 	mnt->mnt_parent = mnt;
+	mnt->mnt_namespace = current->namespace;
 	up_write(&sb->s_umount);
 	put_filesystem(type);
 	return mnt;
@@ -804,6 +805,8 @@
 	return (struct vfsmount *)sb;
 }
 
+EXPORT_SYMBOL_GPL(do_kern_mount);
+
 struct vfsmount *kern_mount(struct file_system_type *type)
 {
 	return do_kern_mount(type->name, 0, type->name, NULL);
diff -uNr linux-2.6.7/include/linux/fs.h linux-2.6.7-auto/include/linux/fs.h
--- linux-2.6.7/include/linux/fs.h	2004-06-18 13:44:05.000000000 +0100
+++ linux-2.6.7-auto/include/linux/fs.h	2004-06-25 13:48:35.000000000 +0100
@@ -714,6 +714,7 @@
 
 #define MNT_FORCE	0x00000001	/* Attempt to forcibily umount */
 #define MNT_DETACH	0x00000002	/* Just detach from the tree */
+#define MNT_EXPIRE	0x00000004	/* Mark for expiry */
 
 extern struct list_head super_blocks;
 extern spinlock_t sb_lock;
diff -uNr linux-2.6.7/include/linux/mount.h linux-2.6.7-auto/include/linux/mount.h
--- linux-2.6.7/include/linux/mount.h	2004-06-18 13:42:15.000000000 +0100
+++ linux-2.6.7-auto/include/linux/mount.h	2004-06-25 13:48:35.000000000 +0100
@@ -29,8 +29,11 @@
 	struct list_head mnt_child;	/* and going through their mnt_child */
 	atomic_t mnt_count;
 	int mnt_flags;
+	int mnt_expiry_mark;		/* true if marked for expiry */
 	char *mnt_devname;		/* Name of device e.g. /dev/dsk/hda1 */
 	struct list_head mnt_list;
+	struct list_head mnt_fslink;	/* link in fs-specific expiry list */
+	struct namespace *mnt_namespace; /* containing namespace */
 };
 
 static inline struct vfsmount *mntget(struct vfsmount *mnt)
@@ -42,7 +45,7 @@
 
 extern void __mntput(struct vfsmount *mnt);
 
-static inline void mntput(struct vfsmount *mnt)
+static inline void _mntput(struct vfsmount *mnt)
 {
 	if (mnt) {
 		if (atomic_dec_and_test(&mnt->mnt_count))
@@ -50,10 +53,26 @@
 	}
 }
 
+static inline void mntput(struct vfsmount *mnt)
+{
+	if (mnt) {
+		mnt->mnt_expiry_mark = 0;
+		_mntput(mnt);
+	}
+}
+
 extern void free_vfsmnt(struct vfsmount *mnt);
 extern struct vfsmount *alloc_vfsmnt(const char *name);
 extern struct vfsmount *do_kern_mount(const char *fstype, int flags,
 				      const char *name, void *data);
+
+struct nameidata;
+
+extern int do_add_mount(struct vfsmount *newmnt, struct nameidata *nd,
+			int mnt_flags, struct list_head *fslist);
+
+extern void mark_mounts_for_expiry(struct list_head *mounts);
+
 extern spinlock_t vfsmount_lock;
 
 #endif
diff -uNr linux-2.6.7/include/linux/namei.h linux-2.6.7-auto/include/linux/namei.h
--- linux-2.6.7/include/linux/namei.h	2004-06-18 13:42:16.000000000 +0100
+++ linux-2.6.7-auto/include/linux/namei.h	2004-06-25 13:48:35.000000000 +0100
@@ -57,6 +57,7 @@
 extern int FASTCALL(path_walk(const char *, struct nameidata *));
 extern int FASTCALL(link_path_walk(const char *, struct nameidata *));
 extern void path_release(struct nameidata *);
+extern void path_release_on_umount(struct nameidata *);
 
 extern struct dentry * lookup_one_len(const char *, struct dentry *, int);
 extern struct dentry * lookup_hash(struct qstr *, struct dentry *);
diff -uNr linux-2.6.7/include/linux/namespace.h linux-2.6.7-auto/include/linux/namespace.h
--- linux-2.6.7/include/linux/namespace.h	2004-06-18 13:42:16.000000000 +0100
+++ linux-2.6.7-auto/include/linux/namespace.h	2004-06-25 13:48:35.000000000 +0100
@@ -14,7 +14,7 @@
 
 extern void umount_tree(struct vfsmount *);
 extern int copy_namespace(int, struct task_struct *);
-void __put_namespace(struct namespace *namespace);
+extern void __put_namespace(struct namespace *namespace);
 
 static inline void put_namespace(struct namespace *namespace)
 {