FUSE: Notifying the kernel of deletion.

Allows a FUSE file-system to tell the kernel when a file or directory is
deleted. If the specified dentry has the specified inode number, the kernel will
unhash it.

The current 'fuse_notify_inval_entry' does not cause the kernel to clean up
directories that are in use properly, and as a result the users of those
directories see incorrect semantics from the file-system. The error condition
seen when 'fuse_notify_inval_entry' is used to notify of a deleted directory is
avoided when 'fuse_notify_delete' is used instead.

The following scenario demonstrates the difference:
1. User A chdirs into 'testdir' and starts reading 'testfile'.
2. User B rm -rf 'testdir'.
3. User B creates 'testdir'.
4. User C chdirs into 'testdir'.

If you run the above within the same machine on any file-system (including fuse
file-systems), there is no problem: user C is able to chdir into the new
testdir. The old testdir is removed from the dentry tree, but still open by user
A.

If operations 2 and 3 are performed via the network such that the fuse
file-system uses one of the notify functions to tell the kernel that the nodes
are gone, then the following error occurs for user C while user A holds the
original directory open:

muirj@empacher:~> ls /test/testdir
ls: cannot access /test/testdir: No such file or directory

The issue here is that the kernel still has a dentry for testdir, and so it is
requesting the attributes for the old directory, while the file-system is
responding that the directory no longer exists.

If on the other hand, if the file-system can notify the kernel that the
directory is deleted using the new 'fuse_notify_delete' function, then the above
ls will find the new directory as expected.

Signed-off-by: John Muir <john@jmuir.com>
Signed-off-by: Miklos Szeredi <mszeredi@suse.cz>

fs/fuse/dev.c

@@ -1378,7 +1378,59 @@ static int fuse_notify_inval_entry(struct fuse_conn *fc, unsigned int size,
 	down_read(&fc->killsb);
 	err = -ENOENT;
 	if (fc->sb)
-		err = fuse_reverse_inval_entry(fc->sb, outarg.parent, &name);
+		err = fuse_reverse_inval_entry(fc->sb, outarg.parent, 0, &name);
+	up_read(&fc->killsb);
+	kfree(buf);
+	return err;
+
+err:
+	kfree(buf);
+	fuse_copy_finish(cs);
+	return err;
+}
+
+static int fuse_notify_delete(struct fuse_conn *fc, unsigned int size,
+			      struct fuse_copy_state *cs)
+{
+	struct fuse_notify_delete_out outarg;
+	int err = -ENOMEM;
+	char *buf;
+	struct qstr name;
+
+	buf = kzalloc(FUSE_NAME_MAX + 1, GFP_KERNEL);
+	if (!buf)
+		goto err;
+
+	err = -EINVAL;
+	if (size < sizeof(outarg))
+		goto err;
+
+	err = fuse_copy_one(cs, &outarg, sizeof(outarg));
+	if (err)
+		goto err;
+
+	err = -ENAMETOOLONG;
+	if (outarg.namelen > FUSE_NAME_MAX)
+		goto err;
+
+	err = -EINVAL;
+	if (size != sizeof(outarg) + outarg.namelen + 1)
+		goto err;
+
+	name.name = buf;
+	name.len = outarg.namelen;
+	err = fuse_copy_one(cs, buf, outarg.namelen + 1);
+	if (err)
+		goto err;
+	fuse_copy_finish(cs);
+	buf[outarg.namelen] = 0;
+	name.hash = full_name_hash(name.name, name.len);
+
+	down_read(&fc->killsb);
+	err = -ENOENT;
+	if (fc->sb)
+		err = fuse_reverse_inval_entry(fc->sb, outarg.parent,
+					       outarg.child, &name);
 	up_read(&fc->killsb);
 	kfree(buf);
 	return err;
@@ -1597,6 +1649,9 @@ static int fuse_notify(struct fuse_conn *fc, enum fuse_notify_code code,
 	case FUSE_NOTIFY_RETRIEVE:
 		return fuse_notify_retrieve(fc, size, cs);
 
+	case FUSE_NOTIFY_DELETE:
+		return fuse_notify_delete(fc, size, cs);
+
 	default:
 		fuse_copy_finish(cs);
 		return -EINVAL;

fs/fuse/dir.c

@@ -868,7 +868,7 @@ int fuse_update_attributes(struct inode *inode, struct kstat *stat,
 }
 
 int fuse_reverse_inval_entry(struct super_block *sb, u64 parent_nodeid,
-			     struct qstr *name)
+			     u64 child_nodeid, struct qstr *name)
 {
 	int err = -ENOTDIR;
 	struct inode *parent;
@@ -895,8 +895,36 @@ int fuse_reverse_inval_entry(struct super_block *sb, u64 parent_nodeid,
 
 	fuse_invalidate_attr(parent);
 	fuse_invalidate_entry(entry);
-	dput(entry);
+
+	if (child_nodeid != 0 && entry->d_inode) {
+		mutex_lock(&entry->d_inode->i_mutex);
+		if (get_node_id(entry->d_inode) != child_nodeid) {
+			err = -ENOENT;
+			goto badentry;
+		}
+		if (d_mountpoint(entry)) {
+			err = -EBUSY;
+			goto badentry;
+		}
+		if (S_ISDIR(entry->d_inode->i_mode)) {
+			shrink_dcache_parent(entry);
+			if (!simple_empty(entry)) {
+				err = -ENOTEMPTY;
+				goto badentry;
+			}
+			entry->d_inode->i_flags |= S_DEAD;
+		}
+		dont_mount(entry);
+		clear_nlink(entry->d_inode);
 		err = 0;
+ badentry:
+		mutex_unlock(&entry->d_inode->i_mutex);
+		if (!err)
+			d_delete(entry);
+	} else {
+		err = 0;
+	}
+	dput(entry);
 
  unlock:
 	mutex_unlock(&parent->i_mutex);

fs/fuse/fuse_i.h

@@ -755,9 +755,15 @@ int fuse_reverse_inval_inode(struct super_block *sb, u64 nodeid,
 /**
  * File-system tells the kernel to invalidate parent attributes and
  * the dentry matching parent/name.
+ *
+ * If the child_nodeid is non-zero and:
+ *    - matches the inode number for the dentry matching parent/name,
+ *    - is not a mount point
+ *    - is a file or oan empty directory
+ * then the dentry is unhashed (d_delete()).
  */
 int fuse_reverse_inval_entry(struct super_block *sb, u64 parent_nodeid,
-			     struct qstr *name);
+			     u64 child_nodeid, struct qstr *name);
 
 int fuse_do_open(struct fuse_conn *fc, u64 nodeid, struct file *file,
 		 bool isdir);

include/linux/fuse.h

@@ -53,6 +53,7 @@
  *
  * 7.18
  *  - add FUSE_IOCTL_DIR flag
+ *  - add FUSE_NOTIFY_DELETE
  */
 
 #ifndef _LINUX_FUSE_H
@@ -288,6 +289,7 @@ enum fuse_notify_code {
 	FUSE_NOTIFY_INVAL_ENTRY = 3,
 	FUSE_NOTIFY_STORE = 4,
 	FUSE_NOTIFY_RETRIEVE = 5,
+	FUSE_NOTIFY_DELETE = 6,
 	FUSE_NOTIFY_CODE_MAX,
 };
 
@@ -611,6 +613,13 @@ struct fuse_notify_inval_entry_out {
 	__u32	padding;
 };
 
+struct fuse_notify_delete_out {
+	__u64	parent;
+	__u64	child;
+	__u32	namelen;
+	__u32	padding;
+};
+
 struct fuse_notify_store_out {
 	__u64	nodeid;
 	__u64	offset;