switch devtmpfs object creation/removal to separate kernel thread

... and give it a namespace where devtmpfs would be mounted on root,
thus avoiding abuses of vfs_path_lookup() (it was never intended to
be used with LOOKUP_PARENT).  Games with credentials are also gone.

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>

drivers/base/devtmpfs.c

@@ -21,12 +21,11 @@
 #include <linux/fs.h>
 #include <linux/shmem_fs.h>
 #include <linux/ramfs.h>
-#include <linux/cred.h>
 #include <linux/sched.h>
-#include <linux/init_task.h>
 #include <linux/slab.h>
+#include <linux/kthread.h>
 
-static struct vfsmount *dev_mnt;
+static struct task_struct *thread;
 
 #if defined CONFIG_DEVTMPFS_MOUNT
 static int mount_dev = 1;
@@ -34,7 +33,16 @@ static int mount_dev = 1;
 static int mount_dev;
 #endif
 
-static DEFINE_MUTEX(dirlock);
+static DEFINE_SPINLOCK(req_lock);
+
+static struct req {
+	struct req *next;
+	struct completion done;
+	int err;
+	const char *name;
+	mode_t mode;	/* 0 => delete */
+	struct device *dev;
+} *requests;
 
 static int __init mount_param(char *str)
 {
@@ -68,14 +76,79 @@ static inline int is_blockdev(struct device *dev)
 static inline int is_blockdev(struct device *dev) { return 0; }
 #endif
 
+int devtmpfs_create_node(struct device *dev)
+{
+	const char *tmp = NULL;
+	struct req req;
+
+	if (!thread)
+		return 0;
+
+	req.mode = 0;
+	req.name = device_get_devnode(dev, &req.mode, &tmp);
+	if (!req.name)
+		return -ENOMEM;
+
+	if (req.mode == 0)
+		req.mode = 0600;
+	if (is_blockdev(dev))
+		req.mode |= S_IFBLK;
+	else
+		req.mode |= S_IFCHR;
+
+	req.dev = dev;
+
+	init_completion(&req.done);
+
+	spin_lock(&req_lock);
+	req.next = requests;
+	requests = &req;
+	spin_unlock(&req_lock);
+
+	wake_up_process(thread);
+	wait_for_completion(&req.done);
+
+	kfree(tmp);
+
+	return req.err;
+}
+
+int devtmpfs_delete_node(struct device *dev)
+{
+	const char *tmp = NULL;
+	struct req req;
+
+	if (!thread)
+		return 0;
+
+	req.name = device_get_devnode(dev, NULL, &tmp);
+	if (!req.name)
+		return -ENOMEM;
+
+	req.mode = 0;
+	req.dev = dev;
+
+	init_completion(&req.done);
+
+	spin_lock(&req_lock);
+	req.next = requests;
+	requests = &req;
+	spin_unlock(&req_lock);
+
+	wake_up_process(thread);
+	wait_for_completion(&req.done);
+
+	kfree(tmp);
+	return req.err;
+}
+
 static int dev_mkdir(const char *name, mode_t mode)
 {
 	struct nameidata nd;
 	struct dentry *dentry;
 	int err;
 
-	err = vfs_path_lookup(dev_mnt->mnt_root, dev_mnt,
-			      name, LOOKUP_PARENT, &nd);
+	err = kern_path_parent(name, &nd);
 	if (err)
 		return err;
 
@@ -84,7 +157,7 @@ static int dev_mkdir(const char *name, mode_t mode)
 		err = vfs_mkdir(nd.path.dentry->d_inode, dentry, mode);
 		if (!err)
 			/* mark as kernel-created inode */
-			dentry->d_inode->i_private = &dev_mnt;
+			dentry->d_inode->i_private = &thread;
 		dput(dentry);
 	} else {
 		err = PTR_ERR(dentry);
@@ -99,7 +172,6 @@ static int create_path(const char *nodepath)
 {
 	int err;
 
-	mutex_lock(&dirlock);
 	err = dev_mkdir(nodepath, 0755);
 	if (err == -ENOENT) {
 		char *path;
@@ -126,45 +198,22 @@ static int create_path(const char *nodepath)
 		kfree(path);
 	}
 out:
-	mutex_unlock(&dirlock);
 	return err;
 }
 
-int devtmpfs_create_node(struct device *dev)
+static int handle_create(const char *nodename, mode_t mode, struct device *dev)
 {
-	const char *tmp = NULL;
-	const char *nodename;
-	const struct cred *curr_cred;
-	mode_t mode = 0;
 	struct nameidata nd;
 	struct dentry *dentry;
 	int err;
 
-	if (!dev_mnt)
-		return 0;
-
-	nodename = device_get_devnode(dev, &mode, &tmp);
-	if (!nodename)
-		return -ENOMEM;
-
-	if (mode == 0)
-		mode = 0600;
-	if (is_blockdev(dev))
-		mode |= S_IFBLK;
-	else
-		mode |= S_IFCHR;
-
-	curr_cred = override_creds(&init_cred);
-
-	err = vfs_path_lookup(dev_mnt->mnt_root, dev_mnt,
-			      nodename, LOOKUP_PARENT, &nd);
+	err = kern_path_parent(nodename, &nd);
 	if (err == -ENOENT) {
 		create_path(nodename);
-		err = vfs_path_lookup(dev_mnt->mnt_root, dev_mnt,
-				      nodename, LOOKUP_PARENT, &nd);
+		err = kern_path_parent(nodename, &nd);
 	}
 	if (err)
-		goto out;
+		return err;
 
 	dentry = lookup_create(&nd, 0);
 	if (!IS_ERR(dentry)) {
@@ -181,7 +230,7 @@ int devtmpfs_create_node(struct device *dev)
 			mutex_unlock(&dentry->d_inode->i_mutex);
 
 			/* mark as kernel-created inode */
-			dentry->d_inode->i_private = &dev_mnt;
+			dentry->d_inode->i_private = &thread;
 		}
 		dput(dentry);
 	} else {
@@ -190,9 +239,6 @@ int devtmpfs_create_node(struct device *dev)
 
 	mutex_unlock(&nd.path.dentry->d_inode->i_mutex);
 	path_put(&nd.path);
-out:
-	kfree(tmp);
-	revert_creds(curr_cred);
 	return err;
 }
 
@@ -202,8 +248,7 @@ static int dev_rmdir(const char *name)
 	struct dentry *dentry;
 	int err;
 
-	err = vfs_path_lookup(dev_mnt->mnt_root, dev_mnt,
-			      name, LOOKUP_PARENT, &nd);
+	err = kern_path_parent(name, &nd);
 	if (err)
 		return err;
 
@@ -211,7 +256,7 @@ static int dev_rmdir(const char *name)
 	dentry = lookup_one_len(nd.last.name, nd.path.dentry, nd.last.len);
 	if (!IS_ERR(dentry)) {
 		if (dentry->d_inode) {
-			if (dentry->d_inode->i_private == &dev_mnt)
+			if (dentry->d_inode->i_private == &thread)
 				err = vfs_rmdir(nd.path.dentry->d_inode,
 						dentry);
 			else
@@ -238,7 +283,6 @@ static int delete_path(const char *nodepath)
 	if (!path)
 		return -ENOMEM;
 
-	mutex_lock(&dirlock);
 	for (;;) {
 		char *base;
 
@@ -250,7 +294,6 @@ static int delete_path(const char *nodepath)
 		if (err)
 			break;
 	}
-	mutex_unlock(&dirlock);
 
 	kfree(path);
 	return err;
@@ -259,7 +302,7 @@ static int delete_path(const char *nodepath)
 static int dev_mynode(struct device *dev, struct inode *inode, struct kstat *stat)
 {
 	/* did we create it */
-	if (inode->i_private != &dev_mnt)
+	if (inode->i_private != &thread)
 		return 0;
 
 	/* does the dev_t match */
@@ -277,29 +320,17 @@ static int dev_mynode(struct device *dev, struct inode *inode, struct kstat *sta
 	return 1;
 }
 
-int devtmpfs_delete_node(struct device *dev)
+static int handle_remove(const char *nodename, struct device *dev)
 {
-	const char *tmp = NULL;
-	const char *nodename;
-	const struct cred *curr_cred;
 	struct nameidata nd;
 	struct dentry *dentry;
 	struct kstat stat;
 	int deleted = 1;
 	int err;
 
-	if (!dev_mnt)
-		return 0;
-
-	nodename = device_get_devnode(dev, NULL, &tmp);
-	if (!nodename)
-		return -ENOMEM;
-
-	curr_cred = override_creds(&init_cred);
-	err = vfs_path_lookup(dev_mnt->mnt_root, dev_mnt,
-			      nodename, LOOKUP_PARENT, &nd);
+	err = kern_path_parent(nodename, &nd);
 	if (err)
-		goto out;
+		return err;
 
 	mutex_lock_nested(&nd.path.dentry->d_inode->i_mutex, I_MUTEX_PARENT);
 	dentry = lookup_one_len(nd.last.name, nd.path.dentry, nd.last.len);
@@ -337,9 +368,6 @@ int devtmpfs_delete_node(struct device *dev)
 	path_put(&nd.path);
 	if (deleted && strchr(nodename, '/'))
 		delete_path(nodename);
-out:
-	kfree(tmp);
-	revert_creds(curr_cred);
 	return err;
 }
 
@@ -354,7 +382,7 @@ int devtmpfs_mount(const char *mntdir)
 	if (!mount_dev)
 		return 0;
 
-	if (!dev_mnt)
+	if (!thread)
 		return 0;
 
 	err = sys_mount("devtmpfs", (char *)mntdir, "devtmpfs", MS_SILENT, NULL);
@@ -365,31 +393,79 @@ int devtmpfs_mount(const char *mntdir)
 	return err;
 }
 
+static __initdata DECLARE_COMPLETION(setup_done);
+
+static int handle(const char *name, mode_t mode, struct device *dev)
+{
+	if (mode)
+		return handle_create(name, mode, dev);
+	else
+		return handle_remove(name, dev);
+}
+
+static int devtmpfsd(void *p)
+{
+	char options[] = "mode=0755";
+	int *err = p;
+	*err = sys_unshare(CLONE_NEWNS);
+	if (*err)
+		goto out;
+	*err = sys_mount("devtmpfs", "/", "devtmpfs", MS_SILENT, options);
+	if (*err)
+		goto out;
+	sys_chdir("/.."); /* will traverse into overmounted root */
+	sys_chroot(".");
+	complete(&setup_done);
+	while (1) {
+		spin_lock(&req_lock);
+		while (requests) {
+			struct req *req = requests;
+			requests = NULL;
+			spin_unlock(&req_lock);
+			while (req) {
+				req->err = handle(req->name, req->mode, req->dev);
+				complete(&req->done);
+				req = req->next;
+			}
+			spin_lock(&req_lock);
+		}
+		set_current_state(TASK_INTERRUPTIBLE);
+		spin_unlock(&req_lock);
+		schedule();
+		__set_current_state(TASK_RUNNING);
+	}
+	return 0;
+out:
+	complete(&setup_done);
+	return *err;
+}
+
 /*
  * Create devtmpfs instance, driver-core devices will add their device
  * nodes here.
  */
 int __init devtmpfs_init(void)
 {
-	int err;
-	struct vfsmount *mnt;
-	char options[] = "mode=0755";
-
-	err = register_filesystem(&dev_fs_type);
+	int err = register_filesystem(&dev_fs_type);
 	if (err) {
 		printk(KERN_ERR "devtmpfs: unable to register devtmpfs "
 		       "type %i\n", err);
 		return err;
 	}
 
-	mnt = kern_mount_data(&dev_fs_type, options);
-	if (IS_ERR(mnt)) {
-		err = PTR_ERR(mnt);
+	thread = kthread_run(devtmpfsd, &err, "kdevtmpfs");
+	if (!IS_ERR(thread)) {
+		wait_for_completion(&setup_done);
+	} else {
+		err = PTR_ERR(thread);
+		thread = NULL;
+	}
+
+	if (err) {
 		printk(KERN_ERR "devtmpfs: unable to create devtmpfs %i\n", err);
 		unregister_filesystem(&dev_fs_type);
 		return err;
 	}
-	dev_mnt = mnt;
 
 	printk(KERN_INFO "devtmpfs: initialized\n");
 	return 0;