[linux-2.6-block.git] / security / device_cgroup.c

/*
 * device_cgroup.c - device cgroup subsystem
 *
 * Copyright 2007 IBM Corp
 */

#include <linux/device_cgroup.h>
#include <linux/cgroup.h>
#include <linux/ctype.h>
#include <linux/list.h>
#include <linux/uaccess.h>
#include <linux/seq_file.h>
#include <linux/slab.h>
#include <linux/rcupdate.h>
#include <linux/mutex.h>

#define ACC_MKNOD 1
#define ACC_READ  2
#define ACC_WRITE 4
#define ACC_MASK (ACC_MKNOD | ACC_READ | ACC_WRITE)

#define DEV_BLOCK 1
#define DEV_CHAR  2
#define DEV_ALL   4  /* this represents all devices */

static DEFINE_MUTEX(devcgroup_mutex);

/*
 * exception list locking rules:
 * hold devcgroup_mutex for update/read.
 * hold rcu_read_lock() for read.
 */

struct dev_exception_item {
	u32 major, minor;
	short type;
	short access;
	struct list_head list;
	struct rcu_head rcu;
};

struct dev_cgroup {
	struct cgroup_subsys_state css;
	struct list_head exceptions;
	enum {
		DEVCG_DEFAULT_ALLOW,
		DEVCG_DEFAULT_DENY,
	} behavior;
};

static inline struct dev_cgroup *css_to_devcgroup(struct cgroup_subsys_state *s)
{
	return container_of(s, struct dev_cgroup, css);
}

static inline struct dev_cgroup *cgroup_to_devcgroup(struct cgroup *cgroup)
{
	return css_to_devcgroup(cgroup_subsys_state(cgroup, devices_subsys_id));
}

static inline struct dev_cgroup *task_devcgroup(struct task_struct *task)
{
	return css_to_devcgroup(task_subsys_state(task, devices_subsys_id));
}

struct cgroup_subsys devices_subsys;

static int devcgroup_can_attach(struct cgroup *new_cgrp,
				struct cgroup_taskset *set)
{
	struct task_struct *task = cgroup_taskset_first(set);

	if (current != task && !capable(CAP_SYS_ADMIN))
		return -EPERM;
	return 0;
}

/*
 * called under devcgroup_mutex
 */
static int dev_exceptions_copy(struct list_head *dest, struct list_head *orig)
{
	struct dev_exception_item *ex, *tmp, *new;

	list_for_each_entry(ex, orig, list) {
		new = kmemdup(ex, sizeof(*ex), GFP_KERNEL);
		if (!new)
			goto free_and_exit;
		list_add_tail(&new->list, dest);
	}

	return 0;

free_and_exit:
	list_for_each_entry_safe(ex, tmp, dest, list) {
		list_del(&ex->list);
		kfree(ex);
	}
	return -ENOMEM;
}

/*
 * called under devcgroup_mutex
 */
static int dev_exception_add(struct dev_cgroup *dev_cgroup,
			     struct dev_exception_item *ex)
{
	struct dev_exception_item *excopy, *walk;

	excopy = kmemdup(ex, sizeof(*ex), GFP_KERNEL);
	if (!excopy)
		return -ENOMEM;

	list_for_each_entry(walk, &dev_cgroup->exceptions, list) {
		if (walk->type != ex->type)
			continue;
		if (walk->major != ex->major)
			continue;
		if (walk->minor != ex->minor)
			continue;

		walk->access |= ex->access;
		kfree(excopy);
		excopy = NULL;
	}

	if (excopy != NULL)
		list_add_tail_rcu(&excopy->list, &dev_cgroup->exceptions);
	return 0;
}

/*
 * called under devcgroup_mutex
 */
static void dev_exception_rm(struct dev_cgroup *dev_cgroup,
			     struct dev_exception_item *ex)
{
	struct dev_exception_item *walk, *tmp;

	list_for_each_entry_safe(walk, tmp, &dev_cgroup->exceptions, list) {
		if (walk->type != ex->type)
			continue;
		if (walk->major != ex->major)
			continue;
		if (walk->minor != ex->minor)
			continue;

		walk->access &= ~ex->access;
		if (!walk->access) {
			list_del_rcu(&walk->list);
			kfree_rcu(walk, rcu);
		}
	}
}

/**
 * dev_exception_clean - frees all entries of the exception list
 * @dev_cgroup: dev_cgroup with the exception list to be cleaned
 *
 * called under devcgroup_mutex
 */
static void dev_exception_clean(struct dev_cgroup *dev_cgroup)
{
	struct dev_exception_item *ex, *tmp;

	list_for_each_entry_safe(ex, tmp, &dev_cgroup->exceptions, list) {
		list_del(&ex->list);
		kfree(ex);
	}
}

/*
 * called from kernel/cgroup.c with cgroup_lock() held.
 */
static struct cgroup_subsys_state *devcgroup_create(struct cgroup *cgroup)
{
	struct dev_cgroup *dev_cgroup, *parent_dev_cgroup;
	struct cgroup *parent_cgroup;
	int ret;

	dev_cgroup = kzalloc(sizeof(*dev_cgroup), GFP_KERNEL);
	if (!dev_cgroup)
		return ERR_PTR(-ENOMEM);
	INIT_LIST_HEAD(&dev_cgroup->exceptions);
	parent_cgroup = cgroup->parent;

	if (parent_cgroup == NULL)
		dev_cgroup->behavior = DEVCG_DEFAULT_ALLOW;
	else {
		parent_dev_cgroup = cgroup_to_devcgroup(parent_cgroup);
		mutex_lock(&devcgroup_mutex);
		ret = dev_exceptions_copy(&dev_cgroup->exceptions,
					  &parent_dev_cgroup->exceptions);
		dev_cgroup->behavior = parent_dev_cgroup->behavior;
		mutex_unlock(&devcgroup_mutex);
		if (ret) {
			kfree(dev_cgroup);
			return ERR_PTR(ret);
		}
	}

	return &dev_cgroup->css;
}

static void devcgroup_destroy(struct cgroup *cgroup)
{
	struct dev_cgroup *dev_cgroup;

	dev_cgroup = cgroup_to_devcgroup(cgroup);
	dev_exception_clean(dev_cgroup);
	kfree(dev_cgroup);
}

#define DEVCG_ALLOW 1
#define DEVCG_DENY 2
#define DEVCG_LIST 3

#define MAJMINLEN 13
#define ACCLEN 4

static void set_access(char *acc, short access)
{
	int idx = 0;
	memset(acc, 0, ACCLEN);
	if (access & ACC_READ)
		acc[idx++] = 'r';
	if (access & ACC_WRITE)
		acc[idx++] = 'w';
	if (access & ACC_MKNOD)
		acc[idx++] = 'm';
}

static char type_to_char(short type)
{
	if (type == DEV_ALL)
		return 'a';
	if (type == DEV_CHAR)
		return 'c';
	if (type == DEV_BLOCK)
		return 'b';
	return 'X';
}

static void set_majmin(char *str, unsigned m)
{
	if (m == ~0)
		strcpy(str, "*");
	else
		sprintf(str, "%u", m);
}

static int devcgroup_seq_read(struct cgroup *cgroup, struct cftype *cft,
				struct seq_file *m)
{
	struct dev_cgroup *devcgroup = cgroup_to_devcgroup(cgroup);
	struct dev_exception_item *ex;
	char maj[MAJMINLEN], min[MAJMINLEN], acc[ACCLEN];

	rcu_read_lock();
	/*
	 * To preserve the compatibility:
	 * - Only show the "all devices" when the default policy is to allow
	 * - List the exceptions in case the default policy is to deny
	 * This way, the file remains as a "whitelist of devices"
	 */
	if (devcgroup->behavior == DEVCG_DEFAULT_ALLOW) {
		set_access(acc, ACC_MASK);
		set_majmin(maj, ~0);
		set_majmin(min, ~0);
		seq_printf(m, "%c %s:%s %s\n", type_to_char(DEV_ALL),
			   maj, min, acc);
	} else {
		list_for_each_entry_rcu(ex, &devcgroup->exceptions, list) {
			set_access(acc, ex->access);
			set_majmin(maj, ex->major);
			set_majmin(min, ex->minor);
			seq_printf(m, "%c %s:%s %s\n", type_to_char(ex->type),
				   maj, min, acc);
		}
	}
	rcu_read_unlock();

	return 0;
}

/**
 * may_access - verifies if a new exception is part of what is allowed
 *		by a dev cgroup based on the default policy +
 *		exceptions. This is used to make sure a child cgroup
 *		won't have more privileges than its parent or to
 *		verify if a certain access is allowed.
 * @dev_cgroup: dev cgroup to be tested against
 * @refex: new exception
 */
static int may_access(struct dev_cgroup *dev_cgroup,
		      struct dev_exception_item *refex)
{
	struct dev_exception_item *ex;
	bool match = false;

	list_for_each_entry(ex, &dev_cgroup->exceptions, list) {
		if ((refex->type & DEV_BLOCK) && !(ex->type & DEV_BLOCK))
			continue;
		if ((refex->type & DEV_CHAR) && !(ex->type & DEV_CHAR))
			continue;
		if (ex->major != ~0 && ex->major != refex->major)
			continue;
		if (ex->minor != ~0 && ex->minor != refex->minor)
			continue;
		if (refex->access & (~ex->access))
			continue;
		match = true;
		break;
	}

	/*
	 * In two cases we'll consider this new exception valid:
	 * - the dev cgroup has its default policy to allow + exception list:
	 *   the new exception should *not* match any of the exceptions
	 *   (behavior == DEVCG_DEFAULT_ALLOW, !match)
	 * - the dev cgroup has its default policy to deny + exception list:
	 *   the new exception *should* match the exceptions
	 *   (behavior == DEVCG_DEFAULT_DENY, match)
	 */
	if ((dev_cgroup->behavior == DEVCG_DEFAULT_DENY) == match)
		return 1;
	return 0;
}

/*
 * parent_has_perm:
 * when adding a new allow rule to a device exception list, the rule
 * must be allowed in the parent device
 */
static int parent_has_perm(struct dev_cgroup *childcg,
				  struct dev_exception_item *ex)
{
	struct cgroup *pcg = childcg->css.cgroup->parent;
	struct dev_cgroup *parent;

	if (!pcg)
		return 1;
	parent = cgroup_to_devcgroup(pcg);
	return may_access(parent, ex);
}

/*
 * Modify the exception list using allow/deny rules.
 * CAP_SYS_ADMIN is needed for this.  It's at least separate from CAP_MKNOD
 * so we can give a container CAP_MKNOD to let it create devices but not
 * modify the exception list.
 * It seems likely we'll want to add a CAP_CONTAINER capability to allow
 * us to also grant CAP_SYS_ADMIN to containers without giving away the
 * device exception list controls, but for now we'll stick with CAP_SYS_ADMIN
 *
 * Taking rules away is always allowed (given CAP_SYS_ADMIN).  Granting
 * new access is only allowed if you're in the top-level cgroup, or your
 * parent cgroup has the access you're asking for.
 */
static int devcgroup_update_access(struct dev_cgroup *devcgroup,
				   int filetype, const char *buffer)
{
	const char *b;
	char temp[12];		/* 11 + 1 characters needed for a u32 */
	int count, rc;
	struct dev_exception_item ex;

	if (!capable(CAP_SYS_ADMIN))
		return -EPERM;

	memset(&ex, 0, sizeof(ex));
	b = buffer;

	switch (*b) {
	case 'a':
		switch (filetype) {
		case DEVCG_ALLOW:
			if (!parent_has_perm(devcgroup, &ex))
				return -EPERM;
			dev_exception_clean(devcgroup);
			devcgroup->behavior = DEVCG_DEFAULT_ALLOW;
			break;
		case DEVCG_DENY:
			dev_exception_clean(devcgroup);
			devcgroup->behavior = DEVCG_DEFAULT_DENY;
			break;
		default:
			return -EINVAL;
		}
		return 0;
	case 'b':
		ex.type = DEV_BLOCK;
		break;
	case 'c':
		ex.type = DEV_CHAR;
		break;
	default:
		return -EINVAL;
	}
	b++;
	if (!isspace(*b))
		return -EINVAL;
	b++;
	if (*b == '*') {
		ex.major = ~0;
		b++;
	} else if (isdigit(*b)) {
		memset(temp, 0, sizeof(temp));
		for (count = 0; count < sizeof(temp) - 1; count++) {
			temp[count] = *b;
			b++;
			if (!isdigit(*b))
				break;
		}
		rc = kstrtou32(temp, 10, &ex.major);
		if (rc)
			return -EINVAL;
	} else {
		return -EINVAL;
	}
	if (*b != ':')
		return -EINVAL;
	b++;

	/* read minor */
	if (*b == '*') {
		ex.minor = ~0;
		b++;
	} else if (isdigit(*b)) {
		memset(temp, 0, sizeof(temp));
		for (count = 0; count < sizeof(temp) - 1; count++) {
			temp[count] = *b;
			b++;
			if (!isdigit(*b))
				break;
		}
		rc = kstrtou32(temp, 10, &ex.minor);
		if (rc)
			return -EINVAL;
	} else {
		return -EINVAL;
	}
	if (!isspace(*b))
		return -EINVAL;
	for (b++, count = 0; count < 3; count++, b++) {
		switch (*b) {
		case 'r':
			ex.access |= ACC_READ;
			break;
		case 'w':
			ex.access |= ACC_WRITE;
			break;
		case 'm':
			ex.access |= ACC_MKNOD;
			break;
		case '\n':
		case '\0':
			count = 3;
			break;
		default:
			return -EINVAL;
		}
	}

	switch (filetype) {
	case DEVCG_ALLOW:
		if (!parent_has_perm(devcgroup, &ex))
			return -EPERM;
		/*
		 * If the default policy is to allow by default, try to remove
		 * an matching exception instead. And be silent about it: we
		 * don't want to break compatibility
		 */
		if (devcgroup->behavior == DEVCG_DEFAULT_ALLOW) {
			dev_exception_rm(devcgroup, &ex);
			return 0;
		}
		return dev_exception_add(devcgroup, &ex);
	case DEVCG_DENY:
		/*
		 * If the default policy is to deny by default, try to remove
		 * an matching exception instead. And be silent about it: we
		 * don't want to break compatibility
		 */
		if (devcgroup->behavior == DEVCG_DEFAULT_DENY) {
			dev_exception_rm(devcgroup, &ex);
			return 0;
		}
		return dev_exception_add(devcgroup, &ex);
	default:
		return -EINVAL;
	}
	return 0;
}

static int devcgroup_access_write(struct cgroup *cgrp, struct cftype *cft,
				  const char *buffer)
{
	int retval;

	mutex_lock(&devcgroup_mutex);
	retval = devcgroup_update_access(cgroup_to_devcgroup(cgrp),
					 cft->private, buffer);
	mutex_unlock(&devcgroup_mutex);
	return retval;
}

static struct cftype dev_cgroup_files[] = {
	{
		.name = "allow",
		.write_string  = devcgroup_access_write,
		.private = DEVCG_ALLOW,
	},
	{
		.name = "deny",
		.write_string = devcgroup_access_write,
		.private = DEVCG_DENY,
	},
	{
		.name = "list",
		.read_seq_string = devcgroup_seq_read,
		.private = DEVCG_LIST,
	},
	{ }	/* terminate */
};

struct cgroup_subsys devices_subsys = {
	.name = "devices",
	.can_attach = devcgroup_can_attach,
	.create = devcgroup_create,
	.destroy = devcgroup_destroy,
	.subsys_id = devices_subsys_id,
	.base_cftypes = dev_cgroup_files,

	/*
	 * While devices cgroup has the rudimentary hierarchy support which
	 * checks the parent's restriction, it doesn't properly propagates
	 * config changes in ancestors to their descendents.  A child
	 * should only be allowed to add more restrictions to the parent's
	 * configuration.  Fix it and remove the following.
	 */
	.broken_hierarchy = true,
};

/**
 * __devcgroup_check_permission - checks if an inode operation is permitted
 * @dev_cgroup: the dev cgroup to be tested against
 * @type: device type
 * @major: device major number
 * @minor: device minor number
 * @access: combination of ACC_WRITE, ACC_READ and ACC_MKNOD
 *
 * returns 0 on success, -EPERM case the operation is not permitted
 */
static int __devcgroup_check_permission(short type, u32 major, u32 minor,
				        short access)
{
	struct dev_cgroup *dev_cgroup;
	struct dev_exception_item ex;
	int rc;

	memset(&ex, 0, sizeof(ex));
	ex.type = type;
	ex.major = major;
	ex.minor = minor;
	ex.access = access;

	rcu_read_lock();
	dev_cgroup = task_devcgroup(current);
	rc = may_access(dev_cgroup, &ex);
	rcu_read_unlock();

	if (!rc)
		return -EPERM;

	return 0;
}

int __devcgroup_inode_permission(struct inode *inode, int mask)
{
	short type, access = 0;

	if (S_ISBLK(inode->i_mode))
		type = DEV_BLOCK;
	if (S_ISCHR(inode->i_mode))
		type = DEV_CHAR;
	if (mask & MAY_WRITE)
		access |= ACC_WRITE;
	if (mask & MAY_READ)
		access |= ACC_READ;

	return __devcgroup_check_permission(type, imajor(inode), iminor(inode),
			access);
}

int devcgroup_inode_mknod(int mode, dev_t dev)
{
	short type;

	if (!S_ISBLK(mode) && !S_ISCHR(mode))
		return 0;

	if (S_ISBLK(mode))
		type = DEV_BLOCK;
	else
		type = DEV_CHAR;

	return __devcgroup_check_permission(type, MAJOR(dev), MINOR(dev),
			ACC_MKNOD);

}
Commit	Line	Data
	1	/*
	2	* device_cgroup.c - device cgroup subsystem
	3	*
	4	* Copyright 2007 IBM Corp
	5	*/
	6
	7	#include <linux/device_cgroup.h>
	8	#include <linux/cgroup.h>
	9	#include <linux/ctype.h>
	10	#include <linux/list.h>
	11	#include <linux/uaccess.h>
	12	#include <linux/seq_file.h>
	13	#include <linux/slab.h>
	14	#include <linux/rcupdate.h>
	15	#include <linux/mutex.h>
	16
	17	#define ACC_MKNOD 1
	18	#define ACC_READ 2
	19	#define ACC_WRITE 4
	20	#define ACC_MASK (ACC_MKNOD \| ACC_READ \| ACC_WRITE)
	21
	22	#define DEV_BLOCK 1
	23	#define DEV_CHAR 2
	24	#define DEV_ALL 4 /* this represents all devices */
	25
	26	static DEFINE_MUTEX(devcgroup_mutex);
	27
	28	/*
	29	* exception list locking rules:
	30	* hold devcgroup_mutex for update/read.
	31	* hold rcu_read_lock() for read.
	32	*/
	33
	34	struct dev_exception_item {
	35	u32 major, minor;
	36	short type;
	37	short access;
	38	struct list_head list;
	39	struct rcu_head rcu;
	40	};
	41
	42	struct dev_cgroup {
	43	struct cgroup_subsys_state css;
	44	struct list_head exceptions;
	45	enum {
	46	DEVCG_DEFAULT_ALLOW,
	47	DEVCG_DEFAULT_DENY,
	48	} behavior;
	49	};
	50
	51	static inline struct dev_cgroup css_to_devcgroup(struct cgroup_subsys_state s)
	52	{
	53	return container_of(s, struct dev_cgroup, css);
	54	}
	55
	56	static inline struct dev_cgroup cgroup_to_devcgroup(struct cgroup cgroup)
	57	{
	58	return css_to_devcgroup(cgroup_subsys_state(cgroup, devices_subsys_id));
	59	}
	60
	61	static inline struct dev_cgroup task_devcgroup(struct task_struct task)
	62	{
	63	return css_to_devcgroup(task_subsys_state(task, devices_subsys_id));
	64	}
	65
	66	struct cgroup_subsys devices_subsys;
	67
	68	static int devcgroup_can_attach(struct cgroup *new_cgrp,
	69	struct cgroup_taskset *set)
	70	{
	71	struct task_struct *task = cgroup_taskset_first(set);
	72
	73	if (current != task && !capable(CAP_SYS_ADMIN))
	74	return -EPERM;
	75	return 0;
	76	}
	77
	78	/*
	79	* called under devcgroup_mutex
	80	*/
	81	static int dev_exceptions_copy(struct list_head dest, struct list_head orig)
	82	{
	83	struct dev_exception_item ex, tmp, *new;
	84
	85	list_for_each_entry(ex, orig, list) {
	86	new = kmemdup(ex, sizeof(*ex), GFP_KERNEL);
	87	if (!new)
	88	goto free_and_exit;
	89	list_add_tail(&new->list, dest);
	90	}
	91
	92	return 0;
	93
	94	free_and_exit:
	95	list_for_each_entry_safe(ex, tmp, dest, list) {
	96	list_del(&ex->list);
	97	kfree(ex);
	98	}
	99	return -ENOMEM;
	100	}
	101
	102	/*
	103	* called under devcgroup_mutex
	104	*/
	105	static int dev_exception_add(struct dev_cgroup *dev_cgroup,
	106	struct dev_exception_item *ex)
	107	{
	108	struct dev_exception_item excopy, walk;
	109
	110	excopy = kmemdup(ex, sizeof(*ex), GFP_KERNEL);
	111	if (!excopy)
	112	return -ENOMEM;
	113
	114	list_for_each_entry(walk, &dev_cgroup->exceptions, list) {
	115	if (walk->type != ex->type)
	116	continue;
	117	if (walk->major != ex->major)
	118	continue;
	119	if (walk->minor != ex->minor)
	120	continue;
	121
	122	walk->access \|= ex->access;
	123	kfree(excopy);
	124	excopy = NULL;
	125	}
	126
	127	if (excopy != NULL)
	128	list_add_tail_rcu(&excopy->list, &dev_cgroup->exceptions);
	129	return 0;
	130	}
	131
	132	/*
	133	* called under devcgroup_mutex
	134	*/
	135	static void dev_exception_rm(struct dev_cgroup *dev_cgroup,
	136	struct dev_exception_item *ex)
	137	{
	138	struct dev_exception_item walk, tmp;
	139
	140	list_for_each_entry_safe(walk, tmp, &dev_cgroup->exceptions, list) {
	141	if (walk->type != ex->type)
	142	continue;
	143	if (walk->major != ex->major)
	144	continue;
	145	if (walk->minor != ex->minor)
	146	continue;
	147
	148	walk->access &= ~ex->access;
	149	if (!walk->access) {
	150	list_del_rcu(&walk->list);
	151	kfree_rcu(walk, rcu);
	152	}
	153	}
	154	}
	155
	156	/**
	157	* dev_exception_clean - frees all entries of the exception list
	158	* @dev_cgroup: dev_cgroup with the exception list to be cleaned
	159	*
	160	* called under devcgroup_mutex
	161	*/
	162	static void dev_exception_clean(struct dev_cgroup *dev_cgroup)
	163	{
	164	struct dev_exception_item ex, tmp;
	165
	166	list_for_each_entry_safe(ex, tmp, &dev_cgroup->exceptions, list) {
	167	list_del(&ex->list);
	168	kfree(ex);
	169	}
	170	}
	171
	172	/*
	173	* called from kernel/cgroup.c with cgroup_lock() held.
	174	*/
	175	static struct cgroup_subsys_state devcgroup_create(struct cgroup cgroup)
	176	{
	177	struct dev_cgroup dev_cgroup, parent_dev_cgroup;
	178	struct cgroup *parent_cgroup;
	179	int ret;
	180
	181	dev_cgroup = kzalloc(sizeof(*dev_cgroup), GFP_KERNEL);
	182	if (!dev_cgroup)
	183	return ERR_PTR(-ENOMEM);
	184	INIT_LIST_HEAD(&dev_cgroup->exceptions);
	185	parent_cgroup = cgroup->parent;
	186
	187	if (parent_cgroup == NULL)
	188	dev_cgroup->behavior = DEVCG_DEFAULT_ALLOW;
	189	else {
	190	parent_dev_cgroup = cgroup_to_devcgroup(parent_cgroup);
	191	mutex_lock(&devcgroup_mutex);
	192	ret = dev_exceptions_copy(&dev_cgroup->exceptions,
	193	&parent_dev_cgroup->exceptions);
	194	dev_cgroup->behavior = parent_dev_cgroup->behavior;
	195	mutex_unlock(&devcgroup_mutex);
	196	if (ret) {
	197	kfree(dev_cgroup);
	198	return ERR_PTR(ret);
	199	}
	200	}
	201
	202	return &dev_cgroup->css;
	203	}
	204
	205	static void devcgroup_destroy(struct cgroup *cgroup)
	206	{
	207	struct dev_cgroup *dev_cgroup;
	208
	209	dev_cgroup = cgroup_to_devcgroup(cgroup);
	210	dev_exception_clean(dev_cgroup);
	211	kfree(dev_cgroup);
	212	}
	213
	214	#define DEVCG_ALLOW 1
	215	#define DEVCG_DENY 2
	216	#define DEVCG_LIST 3
	217
	218	#define MAJMINLEN 13
	219	#define ACCLEN 4
	220
	221	static void set_access(char *acc, short access)
	222	{
	223	int idx = 0;
	224	memset(acc, 0, ACCLEN);
	225	if (access & ACC_READ)
	226	acc[idx++] = 'r';
	227	if (access & ACC_WRITE)
	228	acc[idx++] = 'w';
	229	if (access & ACC_MKNOD)
	230	acc[idx++] = 'm';
	231	}
	232
	233	static char type_to_char(short type)
	234	{
	235	if (type == DEV_ALL)
	236	return 'a';
	237	if (type == DEV_CHAR)
	238	return 'c';
	239	if (type == DEV_BLOCK)
	240	return 'b';
	241	return 'X';
	242	}
	243
	244	static void set_majmin(char *str, unsigned m)
	245	{
	246	if (m == ~0)
	247	strcpy(str, "*");
	248	else
	249	sprintf(str, "%u", m);
	250	}
	251
	252	static int devcgroup_seq_read(struct cgroup cgroup, struct cftype cft,
	253	struct seq_file *m)
	254	{
	255	struct dev_cgroup *devcgroup = cgroup_to_devcgroup(cgroup);
	256	struct dev_exception_item *ex;
	257	char maj[MAJMINLEN], min[MAJMINLEN], acc[ACCLEN];
	258
	259	rcu_read_lock();
	260	/*
	261	* To preserve the compatibility:
	262	* - Only show the "all devices" when the default policy is to allow
	263	* - List the exceptions in case the default policy is to deny
	264	* This way, the file remains as a "whitelist of devices"
	265	*/
	266	if (devcgroup->behavior == DEVCG_DEFAULT_ALLOW) {
	267	set_access(acc, ACC_MASK);
	268	set_majmin(maj, ~0);
	269	set_majmin(min, ~0);
	270	seq_printf(m, "%c %s:%s %s\n", type_to_char(DEV_ALL),
	271	maj, min, acc);
	272	} else {
	273	list_for_each_entry_rcu(ex, &devcgroup->exceptions, list) {
	274	set_access(acc, ex->access);
	275	set_majmin(maj, ex->major);
	276	set_majmin(min, ex->minor);
	277	seq_printf(m, "%c %s:%s %s\n", type_to_char(ex->type),
	278	maj, min, acc);
	279	}
	280	}
	281	rcu_read_unlock();
	282
	283	return 0;
	284	}
	285
	286	/**
	287	* may_access - verifies if a new exception is part of what is allowed
	288	* by a dev cgroup based on the default policy +
	289	* exceptions. This is used to make sure a child cgroup
	290	* won't have more privileges than its parent or to
	291	* verify if a certain access is allowed.
	292	* @dev_cgroup: dev cgroup to be tested against
	293	* @refex: new exception
	294	*/
	295	static int may_access(struct dev_cgroup *dev_cgroup,
	296	struct dev_exception_item *refex)
	297	{
	298	struct dev_exception_item *ex;
	299	bool match = false;
	300
	301	list_for_each_entry(ex, &dev_cgroup->exceptions, list) {
	302	if ((refex->type & DEV_BLOCK) && !(ex->type & DEV_BLOCK))
	303	continue;
	304	if ((refex->type & DEV_CHAR) && !(ex->type & DEV_CHAR))
	305	continue;
	306	if (ex->major != ~0 && ex->major != refex->major)
	307	continue;
	308	if (ex->minor != ~0 && ex->minor != refex->minor)
	309	continue;
	310	if (refex->access & (~ex->access))
	311	continue;
	312	match = true;
	313	break;
	314	}
	315
	316	/*
	317	* In two cases we'll consider this new exception valid:
	318	* - the dev cgroup has its default policy to allow + exception list:
	319	* the new exception should not match any of the exceptions
	320	* (behavior == DEVCG_DEFAULT_ALLOW, !match)
	321	* - the dev cgroup has its default policy to deny + exception list:
	322	* the new exception should match the exceptions
	323	* (behavior == DEVCG_DEFAULT_DENY, match)
	324	*/
	325	if ((dev_cgroup->behavior == DEVCG_DEFAULT_DENY) == match)
	326	return 1;
	327	return 0;
	328	}
	329
	330	/*
	331	* parent_has_perm:
	332	* when adding a new allow rule to a device exception list, the rule
	333	* must be allowed in the parent device
	334	*/
	335	static int parent_has_perm(struct dev_cgroup *childcg,
	336	struct dev_exception_item *ex)
	337	{
	338	struct cgroup *pcg = childcg->css.cgroup->parent;
	339	struct dev_cgroup *parent;
	340
	341	if (!pcg)
	342	return 1;
	343	parent = cgroup_to_devcgroup(pcg);
	344	return may_access(parent, ex);
	345	}
	346
	347	/*
	348	* Modify the exception list using allow/deny rules.
	349	* CAP_SYS_ADMIN is needed for this. It's at least separate from CAP_MKNOD
	350	* so we can give a container CAP_MKNOD to let it create devices but not
	351	* modify the exception list.
	352	* It seems likely we'll want to add a CAP_CONTAINER capability to allow
	353	* us to also grant CAP_SYS_ADMIN to containers without giving away the
	354	* device exception list controls, but for now we'll stick with CAP_SYS_ADMIN
	355	*
	356	* Taking rules away is always allowed (given CAP_SYS_ADMIN). Granting
	357	* new access is only allowed if you're in the top-level cgroup, or your
	358	* parent cgroup has the access you're asking for.
	359	*/
	360	static int devcgroup_update_access(struct dev_cgroup *devcgroup,
	361	int filetype, const char *buffer)
	362	{
	363	const char *b;
	364	char temp[12]; /* 11 + 1 characters needed for a u32 */
	365	int count, rc;
	366	struct dev_exception_item ex;
	367
	368	if (!capable(CAP_SYS_ADMIN))
	369	return -EPERM;
	370
	371	memset(&ex, 0, sizeof(ex));
	372	b = buffer;
	373
	374	switch (*b) {
	375	case 'a':
	376	switch (filetype) {
	377	case DEVCG_ALLOW:
	378	if (!parent_has_perm(devcgroup, &ex))
	379	return -EPERM;
	380	dev_exception_clean(devcgroup);
	381	devcgroup->behavior = DEVCG_DEFAULT_ALLOW;
	382	break;
	383	case DEVCG_DENY:
	384	dev_exception_clean(devcgroup);
	385	devcgroup->behavior = DEVCG_DEFAULT_DENY;
	386	break;
	387	default:
	388	return -EINVAL;
	389	}
	390	return 0;
	391	case 'b':
	392	ex.type = DEV_BLOCK;
	393	break;
	394	case 'c':
	395	ex.type = DEV_CHAR;
	396	break;
	397	default:
	398	return -EINVAL;
	399	}
	400	b++;
	401	if (!isspace(*b))
	402	return -EINVAL;
	403	b++;
	404	if (b == '') {
	405	ex.major = ~0;
	406	b++;
	407	} else if (isdigit(*b)) {
	408	memset(temp, 0, sizeof(temp));
	409	for (count = 0; count < sizeof(temp) - 1; count++) {
	410	temp[count] = *b;
	411	b++;
	412	if (!isdigit(*b))
	413	break;
	414	}
	415	rc = kstrtou32(temp, 10, &ex.major);
	416	if (rc)
	417	return -EINVAL;
	418	} else {
	419	return -EINVAL;
	420	}
	421	if (*b != ':')
	422	return -EINVAL;
	423	b++;
	424
	425	/* read minor */
	426	if (b == '') {
	427	ex.minor = ~0;
	428	b++;
	429	} else if (isdigit(*b)) {
	430	memset(temp, 0, sizeof(temp));
	431	for (count = 0; count < sizeof(temp) - 1; count++) {
	432	temp[count] = *b;
	433	b++;
	434	if (!isdigit(*b))
	435	break;
	436	}
	437	rc = kstrtou32(temp, 10, &ex.minor);
	438	if (rc)
	439	return -EINVAL;
	440	} else {
	441	return -EINVAL;
	442	}
	443	if (!isspace(*b))
	444	return -EINVAL;
	445	for (b++, count = 0; count < 3; count++, b++) {
	446	switch (*b) {
	447	case 'r':
	448	ex.access \|= ACC_READ;
	449	break;
	450	case 'w':
	451	ex.access \|= ACC_WRITE;
	452	break;
	453	case 'm':
	454	ex.access \|= ACC_MKNOD;
	455	break;
	456	case '\n':
	457	case '\0':
	458	count = 3;
	459	break;
	460	default:
	461	return -EINVAL;
	462	}
	463	}
	464
	465	switch (filetype) {
	466	case DEVCG_ALLOW:
	467	if (!parent_has_perm(devcgroup, &ex))
	468	return -EPERM;
	469	/*
	470	* If the default policy is to allow by default, try to remove
	471	* an matching exception instead. And be silent about it: we
	472	* don't want to break compatibility
	473	*/
	474	if (devcgroup->behavior == DEVCG_DEFAULT_ALLOW) {
	475	dev_exception_rm(devcgroup, &ex);
	476	return 0;
	477	}
	478	return dev_exception_add(devcgroup, &ex);
	479	case DEVCG_DENY:
	480	/*
	481	* If the default policy is to deny by default, try to remove
	482	* an matching exception instead. And be silent about it: we
	483	* don't want to break compatibility
	484	*/
	485	if (devcgroup->behavior == DEVCG_DEFAULT_DENY) {
	486	dev_exception_rm(devcgroup, &ex);
	487	return 0;
	488	}
	489	return dev_exception_add(devcgroup, &ex);
	490	default:
	491	return -EINVAL;
	492	}
	493	return 0;
	494	}
	495
	496	static int devcgroup_access_write(struct cgroup cgrp, struct cftype cft,
	497	const char *buffer)
	498	{
	499	int retval;
	500
	501	mutex_lock(&devcgroup_mutex);
	502	retval = devcgroup_update_access(cgroup_to_devcgroup(cgrp),
	503	cft->private, buffer);
	504	mutex_unlock(&devcgroup_mutex);
	505	return retval;
	506	}
	507
	508	static struct cftype dev_cgroup_files[] = {
	509	{
	510	.name = "allow",
	511	.write_string = devcgroup_access_write,
	512	.private = DEVCG_ALLOW,
	513	},
	514	{
	515	.name = "deny",
	516	.write_string = devcgroup_access_write,
	517	.private = DEVCG_DENY,
	518	},
	519	{
	520	.name = "list",
	521	.read_seq_string = devcgroup_seq_read,
	522	.private = DEVCG_LIST,
	523	},
	524	{ } /* terminate */
	525	};
	526
	527	struct cgroup_subsys devices_subsys = {
	528	.name = "devices",
	529	.can_attach = devcgroup_can_attach,
	530	.create = devcgroup_create,
	531	.destroy = devcgroup_destroy,
	532	.subsys_id = devices_subsys_id,
	533	.base_cftypes = dev_cgroup_files,
	534
	535	/*
	536	* While devices cgroup has the rudimentary hierarchy support which
	537	* checks the parent's restriction, it doesn't properly propagates
	538	* config changes in ancestors to their descendents. A child
	539	* should only be allowed to add more restrictions to the parent's
	540	* configuration. Fix it and remove the following.
	541	*/
	542	.broken_hierarchy = true,
	543	};
	544
	545	/**
	546	* __devcgroup_check_permission - checks if an inode operation is permitted
	547	* @dev_cgroup: the dev cgroup to be tested against
	548	* @type: device type
	549	* @major: device major number
	550	* @minor: device minor number
	551	* @access: combination of ACC_WRITE, ACC_READ and ACC_MKNOD
	552	*
	553	* returns 0 on success, -EPERM case the operation is not permitted
	554	*/
	555	static int __devcgroup_check_permission(short type, u32 major, u32 minor,
	556	short access)
	557	{
	558	struct dev_cgroup *dev_cgroup;
	559	struct dev_exception_item ex;
	560	int rc;
	561
	562	memset(&ex, 0, sizeof(ex));
	563	ex.type = type;
	564	ex.major = major;
	565	ex.minor = minor;
	566	ex.access = access;
	567
	568	rcu_read_lock();
	569	dev_cgroup = task_devcgroup(current);
	570	rc = may_access(dev_cgroup, &ex);
	571	rcu_read_unlock();
	572
	573	if (!rc)
	574	return -EPERM;
	575
	576	return 0;
	577	}
	578
	579	int __devcgroup_inode_permission(struct inode *inode, int mask)
	580	{
	581	short type, access = 0;
	582
	583	if (S_ISBLK(inode->i_mode))
	584	type = DEV_BLOCK;
	585	if (S_ISCHR(inode->i_mode))
	586	type = DEV_CHAR;
	587	if (mask & MAY_WRITE)
	588	access \|= ACC_WRITE;
	589	if (mask & MAY_READ)
	590	access \|= ACC_READ;
	591
	592	return __devcgroup_check_permission(type, imajor(inode), iminor(inode),
	593	access);
	594	}
	595
	596	int devcgroup_inode_mknod(int mode, dev_t dev)
	597	{
	598	short type;
	599
	600	if (!S_ISBLK(mode) && !S_ISCHR(mode))
	601	return 0;
	602
	603	if (S_ISBLK(mode))
	604	type = DEV_BLOCK;
	605	else
	606	type = DEV_CHAR;
	607
	608	return __devcgroup_check_permission(type, MAJOR(dev), MINOR(dev),
	609	ACC_MKNOD);
	610
	611	}