[linux-2.6-block.git] / drivers / base / devtmpfs.c

/*
 * devtmpfs - kernel-maintained tmpfs-based /dev
 *
 * Copyright (C) 2009, Kay Sievers <kay.sievers@vrfy.org>
 *
 * During bootup, before any driver core device is registered,
 * devtmpfs, a tmpfs-based filesystem is created. Every driver-core
 * device which requests a device node, will add a node in this
 * filesystem.
 * By default, all devices are named after the the name of the
 * device, owned by root and have a default mode of 0600. Subsystems
 * can overwrite the default setting if needed.
 */

#include <linux/kernel.h>
#include <linux/syscalls.h>
#include <linux/mount.h>
#include <linux/device.h>
#include <linux/genhd.h>
#include <linux/namei.h>
#include <linux/fs.h>
#include <linux/shmem_fs.h>
#include <linux/cred.h>
#include <linux/sched.h>
#include <linux/init_task.h>

static struct vfsmount *dev_mnt;

#if defined CONFIG_DEVTMPFS_MOUNT
static int dev_mount = 1;
#else
static int dev_mount;
#endif

static DEFINE_MUTEX(dirlock);

static int __init mount_param(char *str)
{
	dev_mount = simple_strtoul(str, NULL, 0);
	return 1;
}
__setup("devtmpfs.mount=", mount_param);

static int dev_get_sb(struct file_system_type *fs_type, int flags,
		      const char *dev_name, void *data, struct vfsmount *mnt)
{
	return get_sb_single(fs_type, flags, data, shmem_fill_super, mnt);
}

static struct file_system_type dev_fs_type = {
	.name = "devtmpfs",
	.get_sb = dev_get_sb,
	.kill_sb = kill_litter_super,
};

#ifdef CONFIG_BLOCK
static inline int is_blockdev(struct device *dev)
{
	return dev->class == &block_class;
}
#else
static inline int is_blockdev(struct device *dev) { return 0; }
#endif

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);
	if (err)
		return err;

	dentry = lookup_create(&nd, 1);
	if (!IS_ERR(dentry)) {
		err = vfs_mkdir(nd.path.dentry->d_inode, dentry, mode);
		if (!err)
			/* mark as kernel-created inode */
			dentry->d_inode->i_private = &dev_mnt;
		dput(dentry);
	} else {
		err = PTR_ERR(dentry);
	}

	mutex_unlock(&nd.path.dentry->d_inode->i_mutex);
	path_put(&nd.path);
	return err;
}

static int create_path(const char *nodepath)
{
	int err;

	mutex_lock(&dirlock);
	err = dev_mkdir(nodepath, 0755);
	if (err == -ENOENT) {
		char *path;
		char *s;

		/* parent directories do not exist, create them */
		path = kstrdup(nodepath, GFP_KERNEL);
		if (!path) {
			err = -ENOMEM;
			goto out;
		}
		s = path;
		for (;;) {
			s = strchr(s, '/');
			if (!s)
				break;
			s[0] = '\0';
			err = dev_mkdir(path, 0755);
			if (err && err != -EEXIST)
				break;
			s[0] = '/';
			s++;
		}
		kfree(path);
	}
out:
	mutex_unlock(&dirlock);
	return err;
}

int devtmpfs_create_node(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);
	if (err == -ENOENT) {
		create_path(nodename);
		err = vfs_path_lookup(dev_mnt->mnt_root, dev_mnt,
				      nodename, LOOKUP_PARENT, &nd);
	}
	if (err)
		goto out;

	dentry = lookup_create(&nd, 0);
	if (!IS_ERR(dentry)) {
		err = vfs_mknod(nd.path.dentry->d_inode,
				dentry, mode, dev->devt);
		if (!err) {
			struct iattr newattrs;

			/* fixup possibly umasked mode */
			newattrs.ia_mode = mode;
			newattrs.ia_valid = ATTR_MODE;
			mutex_lock(&dentry->d_inode->i_mutex);
			notify_change(dentry, &newattrs);
			mutex_unlock(&dentry->d_inode->i_mutex);

			/* mark as kernel-created inode */
			dentry->d_inode->i_private = &dev_mnt;
		}
		dput(dentry);
	} else {
		err = PTR_ERR(dentry);
	}

	mutex_unlock(&nd.path.dentry->d_inode->i_mutex);
	path_put(&nd.path);
out:
	kfree(tmp);
	revert_creds(curr_cred);
	return err;
}

static int dev_rmdir(const char *name)
{
	struct nameidata nd;
	struct dentry *dentry;
	int err;

	err = vfs_path_lookup(dev_mnt->mnt_root, dev_mnt,
			      name, LOOKUP_PARENT, &nd);
	if (err)
		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);
	if (!IS_ERR(dentry)) {
		if (dentry->d_inode) {
			if (dentry->d_inode->i_private == &dev_mnt)
				err = vfs_rmdir(nd.path.dentry->d_inode,
						dentry);
			else
				err = -EPERM;
		} else {
			err = -ENOENT;
		}
		dput(dentry);
	} else {
		err = PTR_ERR(dentry);
	}

	mutex_unlock(&nd.path.dentry->d_inode->i_mutex);
	path_put(&nd.path);
	return err;
}

static int delete_path(const char *nodepath)
{
	const char *path;
	int err = 0;

	path = kstrdup(nodepath, GFP_KERNEL);
	if (!path)
		return -ENOMEM;

	mutex_lock(&dirlock);
	for (;;) {
		char *base;

		base = strrchr(path, '/');
		if (!base)
			break;
		base[0] = '\0';
		err = dev_rmdir(path);
		if (err)
			break;
	}
	mutex_unlock(&dirlock);

	kfree(path);
	return err;
}

static int dev_mynode(struct device *dev, struct inode *inode, struct kstat *stat)
{
	/* did we create it */
	if (inode->i_private != &dev_mnt)
		return 0;

	/* does the dev_t match */
	if (is_blockdev(dev)) {
		if (!S_ISBLK(stat->mode))
			return 0;
	} else {
		if (!S_ISCHR(stat->mode))
			return 0;
	}
	if (stat->rdev != dev->devt)
		return 0;

	/* ours */
	return 1;
}

int devtmpfs_delete_node(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);
	if (err)
		goto out;

	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);
	if (!IS_ERR(dentry)) {
		if (dentry->d_inode) {
			err = vfs_getattr(nd.path.mnt, dentry, &stat);
			if (!err && dev_mynode(dev, dentry->d_inode, &stat)) {
				err = vfs_unlink(nd.path.dentry->d_inode,
						 dentry);
				if (!err || err == -ENOENT)
					deleted = 1;
			}
		} else {
			err = -ENOENT;
		}
		dput(dentry);
	} else {
		err = PTR_ERR(dentry);
	}
	mutex_unlock(&nd.path.dentry->d_inode->i_mutex);

	path_put(&nd.path);
	if (deleted && strchr(nodename, '/'))
		delete_path(nodename);
out:
	kfree(tmp);
	revert_creds(curr_cred);
	return err;
}

/*
 * If configured, or requested by the commandline, devtmpfs will be
 * auto-mounted after the kernel mounted the root filesystem.
 */
int devtmpfs_mount(const char *mntdir)
{
	int err;

	if (!dev_mount)
		return 0;

	if (!dev_mnt)
		return 0;

	err = sys_mount("devtmpfs", (char *)mntdir, "devtmpfs", MS_SILENT, NULL);
	if (err)
		printk(KERN_INFO "devtmpfs: error mounting %i\n", err);
	else
		printk(KERN_INFO "devtmpfs: mounted\n");
	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);
	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);
		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;
}
Commit	Line	Data
2b2af54a KS	1	/*
	2	* devtmpfs - kernel-maintained tmpfs-based /dev
	3	*
	4	* Copyright (C) 2009, Kay Sievers <kay.sievers@vrfy.org>
	5	*
	6	* During bootup, before any driver core device is registered,
	7	* devtmpfs, a tmpfs-based filesystem is created. Every driver-core
	8	* device which requests a device node, will add a node in this
e454cea2 KS	9	* filesystem.
	10	* By default, all devices are named after the the name of the
	11	* device, owned by root and have a default mode of 0600. Subsystems
	12	* can overwrite the default setting if needed.
2b2af54a KS	13	*/
	14
	15	#include <linux/kernel.h>
	16	#include <linux/syscalls.h>
	17	#include <linux/mount.h>
	18	#include <linux/device.h>
	19	#include <linux/genhd.h>
	20	#include <linux/namei.h>
	21	#include <linux/fs.h>
	22	#include <linux/shmem_fs.h>
	23	#include <linux/cred.h>
e454cea2	24	#include <linux/sched.h>
2b2af54a KS	25	#include <linux/init_task.h>
	26
	27	static struct vfsmount *dev_mnt;
	28
	29	#if defined CONFIG_DEVTMPFS_MOUNT
	30	static int dev_mount = 1;
	31	#else
	32	static int dev_mount;
	33	#endif
	34
f1f76f86	35	static DEFINE_MUTEX(dirlock);
ed413ae6	36
2b2af54a KS	37	static int __init mount_param(char *str)
	38	{
	39	dev_mount = simple_strtoul(str, NULL, 0);
	40	return 1;
	41	}
	42	__setup("devtmpfs.mount=", mount_param);
	43
	44	static int dev_get_sb(struct file_system_type *fs_type, int flags,
	45	const char dev_name, void data, struct vfsmount *mnt)
	46	{
	47	return get_sb_single(fs_type, flags, data, shmem_fill_super, mnt);
	48	}
	49
	50	static struct file_system_type dev_fs_type = {
	51	.name = "devtmpfs",
	52	.get_sb = dev_get_sb,
	53	.kill_sb = kill_litter_super,
	54	};
	55
	56	#ifdef CONFIG_BLOCK
	57	static inline int is_blockdev(struct device *dev)
	58	{
	59	return dev->class == &block_class;
	60	}
	61	#else
	62	static inline int is_blockdev(struct device *dev) { return 0; }
	63	#endif
	64
	65	static int dev_mkdir(const char *name, mode_t mode)
	66	{
	67	struct nameidata nd;
	68	struct dentry *dentry;
	69	int err;
	70
	71	err = vfs_path_lookup(dev_mnt->mnt_root, dev_mnt,
	72	name, LOOKUP_PARENT, &nd);
	73	if (err)
	74	return err;
	75
	76	dentry = lookup_create(&nd, 1);
	77	if (!IS_ERR(dentry)) {
	78	err = vfs_mkdir(nd.path.dentry->d_inode, dentry, mode);
015bf43b KS	79	if (!err)
	80	/* mark as kernel-created inode */
	81	dentry->d_inode->i_private = &dev_mnt;
2b2af54a KS	82	dput(dentry);
	83	} else {
	84	err = PTR_ERR(dentry);
	85	}
2b2af54a	86
015bf43b	87	mutex_unlock(&nd.path.dentry->d_inode->i_mutex);
2b2af54a KS	88	path_put(&nd.path);
	89	return err;
	90	}
	91
	92	static int create_path(const char *nodepath)
	93	{
015bf43b	94	int err;
2b2af54a	95
f1f76f86	96	mutex_lock(&dirlock);
015bf43b KS	97	err = dev_mkdir(nodepath, 0755);
015bf43b KS	98	if (err == -ENOENT) {
ed413ae6	99	char *path;
2b2af54a KS	100	char *s;
	101
	102	/* parent directories do not exist, create them */
ed413ae6	103	path = kstrdup(nodepath, GFP_KERNEL);
80422738 KS	104	if (!path) {
	105	err = -ENOMEM;
	106	goto out;
	107	}
2b2af54a	108	s = path;
ed413ae6	109	for (;;) {
2b2af54a KS	110	s = strchr(s, '/');
	111	if (!s)
	112	break;
	113	s[0] = '\0';
	114	err = dev_mkdir(path, 0755);
	115	if (err && err != -EEXIST)
	116	break;
	117	s[0] = '/';
	118	s++;
	119	}
ed413ae6	120	kfree(path);
2b2af54a	121	}
80422738	122	out:
f1f76f86	123	mutex_unlock(&dirlock);
2b2af54a KS	124	return err;
	125	}
	126
	127	int devtmpfs_create_node(struct device *dev)
	128	{
	129	const char *tmp = NULL;
	130	const char *nodename;
	131	const struct cred *curr_cred;
e454cea2	132	mode_t mode = 0;
2b2af54a KS	133	struct nameidata nd;
	134	struct dentry *dentry;
	135	int err;
	136
	137	if (!dev_mnt)
	138	return 0;
	139
e454cea2	140	nodename = device_get_devnode(dev, &mode, &tmp);
2b2af54a KS	141	if (!nodename)
	142	return -ENOMEM;
	143
e454cea2 KS	144	if (mode == 0)
e454cea2 KS	145	mode = 0600;
2b2af54a	146	if (is_blockdev(dev))
e454cea2	147	mode \|= S_IFBLK;
2b2af54a	148	else
e454cea2	149	mode \|= S_IFCHR;
2b2af54a KS	150
2b2af54a KS	151	curr_cred = override_creds(&init_cred);
00926996	152
2b2af54a KS	153	err = vfs_path_lookup(dev_mnt->mnt_root, dev_mnt,
	154	nodename, LOOKUP_PARENT, &nd);
	155	if (err == -ENOENT) {
2b2af54a KS	156	create_path(nodename);
	157	err = vfs_path_lookup(dev_mnt->mnt_root, dev_mnt,
	158	nodename, LOOKUP_PARENT, &nd);
2b2af54a	159	}
00926996 KS	160	if (err)
00926996 KS	161	goto out;
2b2af54a KS	162
	163	dentry = lookup_create(&nd, 0);
	164	if (!IS_ERR(dentry)) {
	165	err = vfs_mknod(nd.path.dentry->d_inode,
	166	dentry, mode, dev->devt);
00926996 KS	167	if (!err) {
	168	struct iattr newattrs;
	169
	170	/* fixup possibly umasked mode */
	171	newattrs.ia_mode = mode;
	172	newattrs.ia_valid = ATTR_MODE;
	173	mutex_lock(&dentry->d_inode->i_mutex);
	174	notify_change(dentry, &newattrs);
	175	mutex_unlock(&dentry->d_inode->i_mutex);
	176
	177	/* mark as kernel-created inode */
2b2af54a	178	dentry->d_inode->i_private = &dev_mnt;
00926996	179	}
2b2af54a KS	180	dput(dentry);
	181	} else {
	182	err = PTR_ERR(dentry);
	183	}
2b2af54a	184
00926996	185	mutex_unlock(&nd.path.dentry->d_inode->i_mutex);
2b2af54a KS	186	path_put(&nd.path);
	187	out:
	188	kfree(tmp);
	189	revert_creds(curr_cred);
	190	return err;
	191	}
	192
	193	static int dev_rmdir(const char *name)
	194	{
	195	struct nameidata nd;
	196	struct dentry *dentry;
	197	int err;
	198
	199	err = vfs_path_lookup(dev_mnt->mnt_root, dev_mnt,
	200	name, LOOKUP_PARENT, &nd);
	201	if (err)
	202	return err;
	203
	204	mutex_lock_nested(&nd.path.dentry->d_inode->i_mutex, I_MUTEX_PARENT);
	205	dentry = lookup_one_len(nd.last.name, nd.path.dentry, nd.last.len);
	206	if (!IS_ERR(dentry)) {
015bf43b KS	207	if (dentry->d_inode) {
	208	if (dentry->d_inode->i_private == &dev_mnt)
	209	err = vfs_rmdir(nd.path.dentry->d_inode,
	210	dentry);
	211	else
	212	err = -EPERM;
	213	} else {
2b2af54a	214	err = -ENOENT;
015bf43b	215	}
2b2af54a KS	216	dput(dentry);
	217	} else {
	218	err = PTR_ERR(dentry);
	219	}
2b2af54a	220
015bf43b	221	mutex_unlock(&nd.path.dentry->d_inode->i_mutex);
2b2af54a KS	222	path_put(&nd.path);
	223	return err;
	224	}
	225
	226	static int delete_path(const char *nodepath)
	227	{
	228	const char *path;
	229	int err = 0;
	230
	231	path = kstrdup(nodepath, GFP_KERNEL);
	232	if (!path)
	233	return -ENOMEM;
	234
f1f76f86	235	mutex_lock(&dirlock);
ed413ae6	236	for (;;) {
2b2af54a KS	237	char *base;
	238
	239	base = strrchr(path, '/');
	240	if (!base)
	241	break;
	242	base[0] = '\0';
	243	err = dev_rmdir(path);
	244	if (err)
	245	break;
	246	}
f1f76f86	247	mutex_unlock(&dirlock);
2b2af54a KS	248
	249	kfree(path);
	250	return err;
	251	}
	252
	253	static int dev_mynode(struct device dev, struct inode inode, struct kstat *stat)
	254	{
	255	/* did we create it */
	256	if (inode->i_private != &dev_mnt)
	257	return 0;
	258
	259	/* does the dev_t match */
	260	if (is_blockdev(dev)) {
	261	if (!S_ISBLK(stat->mode))
	262	return 0;
	263	} else {
	264	if (!S_ISCHR(stat->mode))
	265	return 0;
	266	}
	267	if (stat->rdev != dev->devt)
	268	return 0;
	269
	270	/* ours */
	271	return 1;
	272	}
	273
	274	int devtmpfs_delete_node(struct device *dev)
	275	{
	276	const char *tmp = NULL;
	277	const char *nodename;
	278	const struct cred *curr_cred;
	279	struct nameidata nd;
	280	struct dentry *dentry;
	281	struct kstat stat;
	282	int deleted = 1;
	283	int err;
	284
	285	if (!dev_mnt)
	286	return 0;
	287
e454cea2	288	nodename = device_get_devnode(dev, NULL, &tmp);
2b2af54a KS	289	if (!nodename)
	290	return -ENOMEM;
	291
	292	curr_cred = override_creds(&init_cred);
	293	err = vfs_path_lookup(dev_mnt->mnt_root, dev_mnt,
	294	nodename, LOOKUP_PARENT, &nd);
	295	if (err)
	296	goto out;
	297
	298	mutex_lock_nested(&nd.path.dentry->d_inode->i_mutex, I_MUTEX_PARENT);
	299	dentry = lookup_one_len(nd.last.name, nd.path.dentry, nd.last.len);
	300	if (!IS_ERR(dentry)) {
	301	if (dentry->d_inode) {
	302	err = vfs_getattr(nd.path.mnt, dentry, &stat);
	303	if (!err && dev_mynode(dev, dentry->d_inode, &stat)) {
	304	err = vfs_unlink(nd.path.dentry->d_inode,
	305	dentry);
	306	if (!err \|\| err == -ENOENT)
	307	deleted = 1;
	308	}
	309	} else {
	310	err = -ENOENT;
	311	}
	312	dput(dentry);
	313	} else {
	314	err = PTR_ERR(dentry);
	315	}
	316	mutex_unlock(&nd.path.dentry->d_inode->i_mutex);
	317
	318	path_put(&nd.path);
	319	if (deleted && strchr(nodename, '/'))
	320	delete_path(nodename);
	321	out:
	322	kfree(tmp);
	323	revert_creds(curr_cred);
	324	return err;
	325	}
	326
	327	/*
	328	* If configured, or requested by the commandline, devtmpfs will be
	329	* auto-mounted after the kernel mounted the root filesystem.
	330	*/
073120cc	331	int devtmpfs_mount(const char *mntdir)
2b2af54a	332	{
2b2af54a KS	333	int err;
	334
	335	if (!dev_mount)
	336	return 0;
	337
	338	if (!dev_mnt)
	339	return 0;
	340
073120cc	341	err = sys_mount("devtmpfs", (char *)mntdir, "devtmpfs", MS_SILENT, NULL);
2b2af54a KS	342	if (err)
	343	printk(KERN_INFO "devtmpfs: error mounting %i\n", err);
	344	else
	345	printk(KERN_INFO "devtmpfs: mounted\n");
2b2af54a KS	346	return err;
	347	}
	348
	349	/*
	350	* Create devtmpfs instance, driver-core devices will add their device
	351	* nodes here.
	352	*/
	353	int __init devtmpfs_init(void)
	354	{
	355	int err;
	356	struct vfsmount *mnt;
f776c5ec	357	char options[] = "mode=0755";
2b2af54a KS	358
	359	err = register_filesystem(&dev_fs_type);
	360	if (err) {
	361	printk(KERN_ERR "devtmpfs: unable to register devtmpfs "
	362	"type %i\n", err);
	363	return err;
	364	}
	365
f776c5ec	366	mnt = kern_mount_data(&dev_fs_type, options);
2b2af54a KS	367	if (IS_ERR(mnt)) {
	368	err = PTR_ERR(mnt);
	369	printk(KERN_ERR "devtmpfs: unable to create devtmpfs %i\n", err);
	370	unregister_filesystem(&dev_fs_type);
	371	return err;
	372	}
	373	dev_mnt = mnt;
	374
	375	printk(KERN_INFO "devtmpfs: initialized\n");
	376	return 0;
	377	}