cgroups: implement device whitelist

Implement a cgroup to track and enforce open and mknod restrictions on device files. A device cgroup associates a device access whitelist with each cgroup. A whitelist entry has 4 fields. 'type' is a (all), c (char), or b (block). 'all' means it applies to all types and all major and minor numbers. Major and minor are either an integer or * for all. Access is a composition of r (read), w (write), and m (mknod). The root device cgroup starts with rwm to 'all'. A child devcg gets a copy of the parent. Admins can then remove devices from the whitelist or add new entries. A child cgroup can never receive a device access which is denied its parent. However when a device access is removed from a parent it will not also be removed from the child(ren). An entry is added using devices.allow, and removed using devices.deny. For instance echo 'c 1:3 mr' > /cgroups/1/devices.allow allows cgroup 1 to read and mknod the device usually known as /dev/null. Doing echo a > /cgroups/1/devices.deny will remove the default 'a *:* mrw' entry. CAP_SYS_ADMIN is needed to change permissions or move another task to a new cgroup. A cgroup may not be granted more permissions than the cgroup's parent has. Any task can move itself between cgroups. This won't be sufficient, but we can decide the best way to adequately restrict movement later. [akpm@linux-foundation.org: coding-style fixes] [akpm@linux-foundation.org: fix may-be-used-uninitialized warning] Signed-off-by: Serge E. Hallyn <serue@us.ibm.com> Acked-by: James Morris <jmorris@namei.org> Looks-good-to: Pavel Emelyanov <xemul@openvz.org> Cc: Daniel Hokka Zakrisson <daniel@hozac.com> Cc: Li Zefan <lizf@cn.fujitsu.com> Cc: Paul Menage <menage@google.com> Cc: Balbir Singh <balbir@in.ibm.com> Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

cgroups: implement device whitelist
Implement a cgroup to track and enforce open and mknod restrictions on device files. A device cgroup associates a device access whitelist with each cgroup. A whitelist entry has 4 fields. 'type' is a (all), c (char), or b (block). 'all' means it applies to all types and all major and minor numbers. Major and minor are either an integer or * for all. Access is a composition of r (read), w (write), and m (mknod). The root device cgroup starts with rwm to 'all'. A child devcg gets a copy of the parent. Admins can then remove devices from the whitelist or add new entries. A child cgroup can never receive a device access which is denied its parent. However when a device access is removed from a parent it will not also be removed from the child(ren). An entry is added using devices.allow, and removed using devices.deny. For instance echo 'c 1:3 mr' > /cgroups/1/devices.allow allows cgroup 1 to read and mknod the device usually known as /dev/null. Doing echo a > /cgroups/1/devices.deny will remove the default 'a *:* mrw' entry. CAP_SYS_ADMIN is needed to change permissions or move another task to a new cgroup. A cgroup may not be granted more permissions than the cgroup's parent has. Any task can move itself between cgroups. This won't be sufficient, but we can decide the best way to adequately restrict movement later. [akpm@linux-foundation.org: coding-style fixes] [akpm@linux-foundation.org: fix may-be-used-uninitialized warning] Signed-off-by: Serge E. Hallyn <serue@us.ibm.com> Acked-by: James Morris <jmorris@namei.org> Looks-good-to: Pavel Emelyanov <xemul@openvz.org> Cc: Daniel Hokka Zakrisson <daniel@hozac.com> Cc: Li Zefan <lizf@cn.fujitsu.com> Cc: Paul Menage <menage@google.com> Cc: Balbir Singh <balbir@in.ibm.com> Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Serge E. Hallyn · Linus Torvalds
1 parent d447ea2f30
Showing 7 changed files with 686 additions and 0 deletions Side-by-side Diff
Documentation/controllers/devices.txt
fs/namei.c
include/linux/cgroup_subsys.h
include/linux/device_cgroup.h
init/Kconfig
security/Makefile
security/device_cgroup.c
+Device Whitelist Controller
+
+1. Description:
+
+Implement a cgroup to track and enforce open and mknod restrictions
+on device files.  A device cgroup associates a device access
+whitelist with each cgroup.  A whitelist entry has 4 fields.
+'type' is a (all), c (char), or b (block).  'all' means it applies
+to all types and all major and minor numbers.  Major and minor are
+either an integer or * for all.  Access is a composition of r
+(read), w (write), and m (mknod).
+
+The root device cgroup starts with rwm to 'all'.  A child device
+cgroup gets a copy of the parent.  Administrators can then remove
+devices from the whitelist or add new entries.  A child cgroup can
+never receive a device access which is denied its parent.  However
+when a device access is removed from a parent it will not also be
+removed from the child(ren).
+
+2. User Interface
+
+An entry is added using devices.allow, and removed using
+devices.deny.  For instance
+
+	echo 'c 1:3 mr' > /cgroups/1/devices.allow
+
+allows cgroup 1 to read and mknod the device usually known as
+/dev/null.  Doing
+
+	echo a > /cgroups/1/devices.deny
+
+will remove the default 'a *:* mrw' entry.
+
+3. Security
+
+Any task can move itself between cgroups.  This clearly won't
+suffice, but we can decide the best way to adequately restrict
+movement as people get some experience with this.  We may just want
+to require CAP_SYS_ADMIN, which at least is a separate bit from
+CAP_MKNOD.  We may want to just refuse moving to a cgroup which
+isn't a descendent of the current one.  Or we may want to use
+CAP_MAC_ADMIN, since we really are trying to lock down root.
+
+CAP_SYS_ADMIN is needed to modify the whitelist or move another
+task to a new cgroup.  (Again we'll probably want to change that).
+
+A cgroup may not be granted more permissions than the cgroup's
+parent has.
@@ -30,6 +30,7 @@
 #include <linux/capability.h>
 #include <linux/file.h>
 #include <linux/fcntl.h>
+#include <linux/device_cgroup.h>
 #include <asm/namei.h>
 #include <asm/uaccess.h>
  
@@ -281,6 +282,10 @@
 	if (retval)
 		return retval;
  
+	retval = devcgroup_inode_permission(inode, mask);
+	if (retval)
+		return retval;
+
 	return security_inode_permission(inode, mask, nd);
 }
  
@@ -2027,6 +2032,10 @@
  
 	if (!dir->i_op || !dir->i_op->mknod)
 		return -EPERM;
+
+	error = devcgroup_inode_mknod(mode, dev);
+	if (error)
+		return error;
  
 	error = security_inode_mknod(dir, dentry, mode, dev);
 	if (error)
@@ -42,4 +42,10 @@
 #endif
  
 /* */
+
+#ifdef CONFIG_CGROUP_DEVICE
+SUBSYS(devices)
+#endif
+
+/* */
+#include <linux/module.h>
+#include <linux/fs.h>
+
+#ifdef CONFIG_CGROUP_DEVICE
+extern int devcgroup_inode_permission(struct inode *inode, int mask);
+extern int devcgroup_inode_mknod(int mode, dev_t dev);
+#else
+static inline int devcgroup_inode_permission(struct inode *inode, int mask)
+{ return 0; }
+static inline int devcgroup_inode_mknod(int mode, dev_t dev)
+{ return 0; }
+#endif
@@ -298,6 +298,13 @@
           for instance virtual servers and checkpoint/restart
           jobs.
  
+config CGROUP_DEVICE
+	bool "Device controller for cgroups"
+	depends on CGROUPS && EXPERIMENTAL
+	help
+	  Provides a cgroup implementing whitelists for devices which
+	  a process in the cgroup can mknod or open.
+
 config CPUSETS
 	bool "Cpuset support"
 	depends on SMP && CGROUPS
@@ -18,4 +18,5 @@
 obj-$(CONFIG_SECURITY_SMACK)		+= commoncap.o smack/built-in.o
 obj-$(CONFIG_SECURITY_CAPABILITIES)	+= commoncap.o capability.o
 obj-$(CONFIG_SECURITY_ROOTPLUG)		+= commoncap.o root_plug.o
+obj-$(CONFIG_CGROUP_DEVICE)		+= device_cgroup.o
+/*
+ * dev_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>
+
+#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 */
+
+/*
+ * whitelist locking rules:
+ * cgroup_lock() cannot be taken under dev_cgroup->lock.
+ * dev_cgroup->lock can be taken with or without cgroup_lock().
+ *
+ * modifications always require cgroup_lock
+ * modifications to a list which is visible require the
+ *   dev_cgroup->lock *and* cgroup_lock()
+ * walking the list requires dev_cgroup->lock or cgroup_lock().
+ *
+ * reasoning: dev_whitelist_copy() needs to kmalloc, so needs
+ *   a mutex, which the cgroup_lock() is.  Since modifying
+ *   a visible list requires both locks, either lock can be
+ *   taken for walking the list.
+ */
+
+struct dev_whitelist_item {
+	u32 major, minor;
+	short type;
+	short access;
+	struct list_head list;
+};
+
+struct dev_cgroup {
+	struct cgroup_subsys_state css;
+	struct list_head whitelist;
+	spinlock_t lock;
+};
+
+static inline struct dev_cgroup *cgroup_to_devcgroup(struct cgroup *cgroup)
+{
+	return container_of(cgroup_subsys_state(cgroup, devices_subsys_id),
+			    struct dev_cgroup, css);
+}
+
+struct cgroup_subsys devices_subsys;
+
+static int devcgroup_can_attach(struct cgroup_subsys *ss,
+		struct cgroup *new_cgroup, struct task_struct *task)
+{
+	if (current != task && !capable(CAP_SYS_ADMIN))
+			return -EPERM;
+
+	return 0;
+}
+
+/*
+ * called under cgroup_lock()
+ */
+static int dev_whitelist_copy(struct list_head *dest, struct list_head *orig)
+{
+	struct dev_whitelist_item *wh, *tmp, *new;
+
+	list_for_each_entry(wh, orig, list) {
+		new = kmalloc(sizeof(*wh), GFP_KERNEL);
+		if (!new)
+			goto free_and_exit;
+		new->major = wh->major;
+		new->minor = wh->minor;
+		new->type = wh->type;
+		new->access = wh->access;
+		list_add_tail(&new->list, dest);
+	}
+
+	return 0;
+
+free_and_exit:
+	list_for_each_entry_safe(wh, tmp, dest, list) {
+		list_del(&wh->list);
+		kfree(wh);
+	}
+	return -ENOMEM;
+}
+
+/* Stupid prototype - don't bother combining existing entries */
+/*
+ * called under cgroup_lock()
+ * since the list is visible to other tasks, we need the spinlock also
+ */
+static int dev_whitelist_add(struct dev_cgroup *dev_cgroup,
+			struct dev_whitelist_item *wh)
+{
+	struct dev_whitelist_item *whcopy;
+
+	whcopy = kmalloc(sizeof(*whcopy), GFP_KERNEL);
+	if (!whcopy)
+		return -ENOMEM;
+
+	memcpy(whcopy, wh, sizeof(*whcopy));
+	spin_lock(&dev_cgroup->lock);
+	list_add_tail(&whcopy->list, &dev_cgroup->whitelist);
+	spin_unlock(&dev_cgroup->lock);
+	return 0;
+}
+
+/*
+ * called under cgroup_lock()
+ * since the list is visible to other tasks, we need the spinlock also
+ */
+static void dev_whitelist_rm(struct dev_cgroup *dev_cgroup,
+			struct dev_whitelist_item *wh)
+{
+	struct dev_whitelist_item *walk, *tmp;
+
+	spin_lock(&dev_cgroup->lock);
+	list_for_each_entry_safe(walk, tmp, &dev_cgroup->whitelist, list) {
+		if (walk->type == DEV_ALL)
+			goto remove;
+		if (walk->type != wh->type)
+			continue;
+		if (walk->major != ~0 && walk->major != wh->major)
+			continue;
+		if (walk->minor != ~0 && walk->minor != wh->minor)
+			continue;
+
+remove:
+		walk->access &= ~wh->access;
+		if (!walk->access) {
+			list_del(&walk->list);
+			kfree(walk);
+		}
+	}
+	spin_unlock(&dev_cgroup->lock);
+}
+
+/*
+ * called from kernel/cgroup.c with cgroup_lock() held.
+ */
+static struct cgroup_subsys_state *devcgroup_create(struct cgroup_subsys *ss,
+						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->whitelist);
+	parent_cgroup = cgroup->parent;
+
+	if (parent_cgroup == NULL) {
+		struct dev_whitelist_item *wh;
+		wh = kmalloc(sizeof(*wh), GFP_KERNEL);
+		if (!wh) {
+			kfree(dev_cgroup);
+			return ERR_PTR(-ENOMEM);
+		}
+		wh->minor = wh->major = ~0;
+		wh->type = DEV_ALL;
+		wh->access = ACC_MKNOD | ACC_READ | ACC_WRITE;
+		list_add(&wh->list, &dev_cgroup->whitelist);
+	} else {
+		parent_dev_cgroup = cgroup_to_devcgroup(parent_cgroup);
+		ret = dev_whitelist_copy(&dev_cgroup->whitelist,
+				&parent_dev_cgroup->whitelist);
+		if (ret) {
+			kfree(dev_cgroup);
+			return ERR_PTR(ret);
+		}
+	}
+
+	spin_lock_init(&dev_cgroup->lock);
+	return &dev_cgroup->css;
+}
+
+static void devcgroup_destroy(struct cgroup_subsys *ss,
+			struct cgroup *cgroup)
+{
+	struct dev_cgroup *dev_cgroup;
+	struct dev_whitelist_item *wh, *tmp;
+
+	dev_cgroup = cgroup_to_devcgroup(cgroup);
+	list_for_each_entry_safe(wh, tmp, &dev_cgroup->whitelist, list) {
+		list_del(&wh->list);
+		kfree(wh);
+	}
+	kfree(dev_cgroup);
+}
+
+#define DEVCG_ALLOW 1
+#define DEVCG_DENY 2
+
+static void set_access(char *acc, short access)
+{
+	int idx = 0;
+	memset(acc, 0, 4);
+	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, int len, unsigned m)
+{
+	memset(str, 0, len);
+	if (m == ~0)
+		sprintf(str, "*");
+	else
+		snprintf(str, len, "%d", m);
+}
+
+static char *print_whitelist(struct dev_cgroup *devcgroup, int *len)
+{
+	char *buf, *s, acc[4];
+	struct dev_whitelist_item *wh;
+	int ret;
+	int count = 0;
+	char maj[10], min[10];
+
+	buf = kmalloc(4096, GFP_KERNEL);
+	if (!buf)
+		return ERR_PTR(-ENOMEM);
+	s = buf;
+	*s = '\0';
+	*len = 0;
+
+	spin_lock(&devcgroup->lock);
+	list_for_each_entry(wh, &devcgroup->whitelist, list) {
+		set_access(acc, wh->access);
+		set_majmin(maj, 10, wh->major);
+		set_majmin(min, 10, wh->minor);
+		ret = snprintf(s, 4095-(s-buf), "%c %s:%s %s\n",
+			type_to_char(wh->type), maj, min, acc);
+		if (s+ret >= buf+4095) {
+			kfree(buf);
+			buf = ERR_PTR(-ENOMEM);
+			break;
+		}
+		s += ret;
+		*len += ret;
+		count++;
+	}
+	spin_unlock(&devcgroup->lock);
+
+	return buf;
+}
+
+static ssize_t devcgroup_access_read(struct cgroup *cgroup,
+			struct cftype *cft, struct file *file,
+			char __user *userbuf, size_t nbytes, loff_t *ppos)
+{
+	struct dev_cgroup *devcgroup = cgroup_to_devcgroup(cgroup);
+	int filetype = cft->private;
+	char *buffer;
+	int uninitialized_var(len);
+	int retval;
+
+	if (filetype != DEVCG_ALLOW)
+		return -EINVAL;
+	buffer = print_whitelist(devcgroup, &len);
+	if (IS_ERR(buffer))
+		return PTR_ERR(buffer);
+
+	retval = simple_read_from_buffer(userbuf, nbytes, ppos, buffer, len);
+	kfree(buffer);
+	return retval;
+}
+
+/*
+ * may_access_whitelist:
+ * does the access granted to dev_cgroup c contain the access
+ * requested in whitelist item refwh.
+ * return 1 if yes, 0 if no.
+ * call with c->lock held
+ */
+static int may_access_whitelist(struct dev_cgroup *c,
+				       struct dev_whitelist_item *refwh)
+{
+	struct dev_whitelist_item *whitem;
+
+	list_for_each_entry(whitem, &c->whitelist, list) {
+		if (whitem->type & DEV_ALL)
+			return 1;
+		if ((refwh->type & DEV_BLOCK) && !(whitem->type & DEV_BLOCK))
+			continue;
+		if ((refwh->type & DEV_CHAR) && !(whitem->type & DEV_CHAR))
+			continue;
+		if (whitem->major != ~0 && whitem->major != refwh->major)
+			continue;
+		if (whitem->minor != ~0 && whitem->minor != refwh->minor)
+			continue;
+		if (refwh->access & (~(whitem->access | ACC_MASK)))
+			continue;
+		return 1;
+	}
+	return 0;
+}
+
+/*
+ * parent_has_perm:
+ * when adding a new allow rule to a device whitelist, the rule
+ * must be allowed in the parent device
+ */
+static int parent_has_perm(struct cgroup *childcg,
+				  struct dev_whitelist_item *wh)
+{
+	struct cgroup *pcg = childcg->parent;
+	struct dev_cgroup *parent;
+	int ret;
+
+	if (!pcg)
+		return 1;
+	parent = cgroup_to_devcgroup(pcg);
+	spin_lock(&parent->lock);
+	ret = may_access_whitelist(parent, wh);
+	spin_unlock(&parent->lock);
+	return ret;
+}
+
+/*
+ * Modify the whitelist 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 whitelist.
+ * 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 whitelist 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 ssize_t devcgroup_access_write(struct cgroup *cgroup, struct cftype *cft,
+				struct file *file, const char __user *userbuf,
+				size_t nbytes, loff_t *ppos)
+{
+	struct cgroup *cur_cgroup;
+	struct dev_cgroup *devcgroup, *cur_devcgroup;
+	int filetype = cft->private;
+	char *buffer, *b;
+	int retval = 0, count;
+	struct dev_whitelist_item wh;
+
+	if (!capable(CAP_SYS_ADMIN))
+		return -EPERM;
+
+	devcgroup = cgroup_to_devcgroup(cgroup);
+	cur_cgroup = task_cgroup(current, devices_subsys.subsys_id);
+	cur_devcgroup = cgroup_to_devcgroup(cur_cgroup);
+
+	buffer = kmalloc(nbytes+1, GFP_KERNEL);
+	if (!buffer)
+		return -ENOMEM;
+
+	if (copy_from_user(buffer, userbuf, nbytes)) {
+		retval = -EFAULT;
+		goto out1;
+	}
+	buffer[nbytes] = 0;	/* nul-terminate */
+
+	cgroup_lock();
+	if (cgroup_is_removed(cgroup)) {
+		retval = -ENODEV;
+		goto out2;
+	}
+
+	memset(&wh, 0, sizeof(wh));
+	b = buffer;
+
+	switch (*b) {
+	case 'a':
+		wh.type = DEV_ALL;
+		wh.access = ACC_MASK;
+		goto handle;
+	case 'b':
+		wh.type = DEV_BLOCK;
+		break;
+	case 'c':
+		wh.type = DEV_CHAR;
+		break;
+	default:
+		retval = -EINVAL;
+		goto out2;
+	}
+	b++;
+	if (!isspace(*b)) {
+		retval = -EINVAL;
+		goto out2;
+	}
+	b++;
+	if (*b == '*') {
+		wh.major = ~0;
+		b++;
+	} else if (isdigit(*b)) {
+		wh.major = 0;
+		while (isdigit(*b)) {
+			wh.major = wh.major*10+(*b-'0');
+			b++;
+		}
+	} else {
+		retval = -EINVAL;
+		goto out2;
+	}
+	if (*b != ':') {
+		retval = -EINVAL;
+		goto out2;
+	}
+	b++;
+
+	/* read minor */
+	if (*b == '*') {
+		wh.minor = ~0;
+		b++;
+	} else if (isdigit(*b)) {
+		wh.minor = 0;
+		while (isdigit(*b)) {
+			wh.minor = wh.minor*10+(*b-'0');
+			b++;
+		}
+	} else {
+		retval = -EINVAL;
+		goto out2;
+	}
+	if (!isspace(*b)) {
+		retval = -EINVAL;
+		goto out2;
+	}
+	for (b++, count = 0; count < 3; count++, b++) {
+		switch (*b) {
+		case 'r':
+			wh.access |= ACC_READ;
+			break;
+		case 'w':
+			wh.access |= ACC_WRITE;
+			break;
+		case 'm':
+			wh.access |= ACC_MKNOD;
+			break;
+		case '\n':
+		case '\0':
+			count = 3;
+			break;
+		default:
+			retval = -EINVAL;
+			goto out2;
+		}
+	}
+
+handle:
+	retval = 0;
+	switch (filetype) {
+	case DEVCG_ALLOW:
+		if (!parent_has_perm(cgroup, &wh))
+			retval = -EPERM;
+		else
+			retval = dev_whitelist_add(devcgroup, &wh);
+		break;
+	case DEVCG_DENY:
+		dev_whitelist_rm(devcgroup, &wh);
+		break;
+	default:
+		retval = -EINVAL;
+		goto out2;
+	}
+
+	if (retval == 0)
+		retval = nbytes;
+
+out2:
+	cgroup_unlock();
+out1:
+	kfree(buffer);
+	return retval;
+}
+
+static struct cftype dev_cgroup_files[] = {
+	{
+		.name = "allow",
+		.read = devcgroup_access_read,
+		.write  = devcgroup_access_write,
+		.private = DEVCG_ALLOW,
+	},
+	{
+		.name = "deny",
+		.write = devcgroup_access_write,
+		.private = DEVCG_DENY,
+	},
+};
+
+static int devcgroup_populate(struct cgroup_subsys *ss,
+				struct cgroup *cgroup)
+{
+	return cgroup_add_files(cgroup, ss, dev_cgroup_files,
+					ARRAY_SIZE(dev_cgroup_files));
+}
+
+struct cgroup_subsys devices_subsys = {
+	.name = "devices",
+	.can_attach = devcgroup_can_attach,
+	.create = devcgroup_create,
+	.destroy  = devcgroup_destroy,
+	.populate = devcgroup_populate,
+	.subsys_id = devices_subsys_id,
+};
+
+int devcgroup_inode_permission(struct inode *inode, int mask)
+{
+	struct cgroup *cgroup;
+	struct dev_cgroup *dev_cgroup;
+	struct dev_whitelist_item *wh;
+
+	dev_t device = inode->i_rdev;
+	if (!device)
+		return 0;
+	if (!S_ISBLK(inode->i_mode) && !S_ISCHR(inode->i_mode))
+		return 0;
+	cgroup = task_cgroup(current, devices_subsys.subsys_id);
+	dev_cgroup = cgroup_to_devcgroup(cgroup);
+	if (!dev_cgroup)
+		return 0;
+
+	spin_lock(&dev_cgroup->lock);
+	list_for_each_entry(wh, &dev_cgroup->whitelist, list) {
+		if (wh->type & DEV_ALL)
+			goto acc_check;
+		if ((wh->type & DEV_BLOCK) && !S_ISBLK(inode->i_mode))
+			continue;
+		if ((wh->type & DEV_CHAR) && !S_ISCHR(inode->i_mode))
+			continue;
+		if (wh->major != ~0 && wh->major != imajor(inode))
+			continue;
+		if (wh->minor != ~0 && wh->minor != iminor(inode))
+			continue;
+acc_check:
+		if ((mask & MAY_WRITE) && !(wh->access & ACC_WRITE))
+			continue;
+		if ((mask & MAY_READ) && !(wh->access & ACC_READ))
+			continue;
+		spin_unlock(&dev_cgroup->lock);
+		return 0;
+	}
+	spin_unlock(&dev_cgroup->lock);
+
+	return -EPERM;
+}
+
+int devcgroup_inode_mknod(int mode, dev_t dev)
+{
+	struct cgroup *cgroup;
+	struct dev_cgroup *dev_cgroup;
+	struct dev_whitelist_item *wh;
+
+	cgroup = task_cgroup(current, devices_subsys.subsys_id);
+	dev_cgroup = cgroup_to_devcgroup(cgroup);
+	if (!dev_cgroup)
+		return 0;
+
+	spin_lock(&dev_cgroup->lock);
+	list_for_each_entry(wh, &dev_cgroup->whitelist, list) {
+		if (wh->type & DEV_ALL)
+			goto acc_check;
+		if ((wh->type & DEV_BLOCK) && !S_ISBLK(mode))
+			continue;
+		if ((wh->type & DEV_CHAR) && !S_ISCHR(mode))
+			continue;
+		if (wh->major != ~0 && wh->major != MAJOR(dev))
+			continue;
+		if (wh->minor != ~0 && wh->minor != MINOR(dev))
+			continue;
+acc_check:
+		if (!(wh->access & ACC_MKNOD))
+			continue;
+		spin_unlock(&dev_cgroup->lock);
+		return 0;
+	}
+	spin_unlock(&dev_cgroup->lock);
+	return -EPERM;
+}
	1	+Device Whitelist Controller
	2	+
	3	+1. Description:
	4	+
	5	+Implement a cgroup to track and enforce open and mknod restrictions
	6	+on device files. A device cgroup associates a device access
	7	+whitelist with each cgroup. A whitelist entry has 4 fields.
	8	+'type' is a (all), c (char), or b (block). 'all' means it applies
	9	+to all types and all major and minor numbers. Major and minor are
	10	+either an integer or * for all. Access is a composition of r
	11	+(read), w (write), and m (mknod).
	12	+
	13	+The root device cgroup starts with rwm to 'all'. A child device
	14	+cgroup gets a copy of the parent. Administrators can then remove
	15	+devices from the whitelist or add new entries. A child cgroup can
	16	+never receive a device access which is denied its parent. However
	17	+when a device access is removed from a parent it will not also be
	18	+removed from the child(ren).
	19	+
	20	+2. User Interface
	21	+
	22	+An entry is added using devices.allow, and removed using
	23	+devices.deny. For instance
	24	+
	25	+ echo 'c 1:3 mr' > /cgroups/1/devices.allow
	26	+
	27	+allows cgroup 1 to read and mknod the device usually known as
	28	+/dev/null. Doing
	29	+
	30	+ echo a > /cgroups/1/devices.deny
	31	+
	32	+will remove the default 'a : mrw' entry.
	33	+
	34	+3. Security
	35	+
	36	+Any task can move itself between cgroups. This clearly won't
	37	+suffice, but we can decide the best way to adequately restrict
	38	+movement as people get some experience with this. We may just want
	39	+to require CAP_SYS_ADMIN, which at least is a separate bit from
	40	+CAP_MKNOD. We may want to just refuse moving to a cgroup which
	41	+isn't a descendent of the current one. Or we may want to use
	42	+CAP_MAC_ADMIN, since we really are trying to lock down root.
	43	+
	44	+CAP_SYS_ADMIN is needed to modify the whitelist or move another
	45	+task to a new cgroup. (Again we'll probably want to change that).
	46	+
	47	+A cgroup may not be granted more permissions than the cgroup's
	48	+parent has.
...	...	@@ -30,6 +30,7 @@
30	30	#include <linux/capability.h>
31	31	#include <linux/file.h>
32	32	#include <linux/fcntl.h>
	33	+#include <linux/device_cgroup.h>
33	34	#include <asm/namei.h>
34	35	#include <asm/uaccess.h>
35	36
...	...	@@ -281,6 +282,10 @@
281	282	if (retval)
282	283	return retval;
283	284
	285	+ retval = devcgroup_inode_permission(inode, mask);
	286	+ if (retval)
	287	+ return retval;
	288	+
284	289	return security_inode_permission(inode, mask, nd);
285	290	}
286	291
...	...	@@ -2027,6 +2032,10 @@
2027	2032
2028	2033	if (!dir->i_op \|\| !dir->i_op->mknod)
2029	2034	return -EPERM;
	2035	+
	2036	+ error = devcgroup_inode_mknod(mode, dev);
	2037	+ if (error)
	2038	+ return error;
2030	2039
2031	2040	error = security_inode_mknod(dir, dentry, mode, dev);
2032	2041	if (error)
...	...	@@ -42,4 +42,10 @@
42	42	#endif
43	43
44	44	/* */
	45	+
	46	+#ifdef CONFIG_CGROUP_DEVICE
	47	+SUBSYS(devices)
	48	+#endif
	49	+
	50	+/* */
	1	+#include <linux/module.h>
	2	+#include <linux/fs.h>
	3	+
	4	+#ifdef CONFIG_CGROUP_DEVICE
	5	+extern int devcgroup_inode_permission(struct inode *inode, int mask);
	6	+extern int devcgroup_inode_mknod(int mode, dev_t dev);
	7	+#else
	8	+static inline int devcgroup_inode_permission(struct inode *inode, int mask)
	9	+{ return 0; }
	10	+static inline int devcgroup_inode_mknod(int mode, dev_t dev)
	11	+{ return 0; }
	12	+#endif
...	...	@@ -298,6 +298,13 @@
298	298	for instance virtual servers and checkpoint/restart
299	299	jobs.
300	300
	301	+config CGROUP_DEVICE
	302	+ bool "Device controller for cgroups"
	303	+ depends on CGROUPS && EXPERIMENTAL
	304	+ help
	305	+ Provides a cgroup implementing whitelists for devices which
	306	+ a process in the cgroup can mknod or open.
	307	+
301	308	config CPUSETS
302	309	bool "Cpuset support"
303	310	depends on SMP && CGROUPS
...	...	@@ -18,4 +18,5 @@
18	18	obj-$(CONFIG_SECURITY_SMACK) += commoncap.o smack/built-in.o
19	19	obj-$(CONFIG_SECURITY_CAPABILITIES) += commoncap.o capability.o
20	20	obj-$(CONFIG_SECURITY_ROOTPLUG) += commoncap.o root_plug.o
	21	+obj-$(CONFIG_CGROUP_DEVICE) += device_cgroup.o
	1	+/*
	2	+ * dev_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	+
	13	+#define ACC_MKNOD 1
	14	+#define ACC_READ 2
	15	+#define ACC_WRITE 4
	16	+#define ACC_MASK (ACC_MKNOD \| ACC_READ \| ACC_WRITE)
	17	+
	18	+#define DEV_BLOCK 1
	19	+#define DEV_CHAR 2
	20	+#define DEV_ALL 4 /* this represents all devices */
	21	+
	22	+/*
	23	+ * whitelist locking rules:
	24	+ * cgroup_lock() cannot be taken under dev_cgroup->lock.
	25	+ * dev_cgroup->lock can be taken with or without cgroup_lock().
	26	+ *
	27	+ * modifications always require cgroup_lock
	28	+ * modifications to a list which is visible require the
	29	+ * dev_cgroup->lock and cgroup_lock()
	30	+ * walking the list requires dev_cgroup->lock or cgroup_lock().
	31	+ *
	32	+ * reasoning: dev_whitelist_copy() needs to kmalloc, so needs
	33	+ * a mutex, which the cgroup_lock() is. Since modifying
	34	+ * a visible list requires both locks, either lock can be
	35	+ * taken for walking the list.
	36	+ */
	37	+
	38	+struct dev_whitelist_item {
	39	+ u32 major, minor;
	40	+ short type;
	41	+ short access;
	42	+ struct list_head list;
	43	+};
	44	+
	45	+struct dev_cgroup {
	46	+ struct cgroup_subsys_state css;
	47	+ struct list_head whitelist;
	48	+ spinlock_t lock;
	49	+};
	50	+
	51	+static inline struct dev_cgroup cgroup_to_devcgroup(struct cgroup cgroup)
	52	+{
	53	+ return container_of(cgroup_subsys_state(cgroup, devices_subsys_id),
	54	+ struct dev_cgroup, css);
	55	+}
	56	+
	57	+struct cgroup_subsys devices_subsys;
	58	+
	59	+static int devcgroup_can_attach(struct cgroup_subsys *ss,
	60	+ struct cgroup new_cgroup, struct task_struct task)
	61	+{
	62	+ if (current != task && !capable(CAP_SYS_ADMIN))
	63	+ return -EPERM;
	64	+
	65	+ return 0;
	66	+}
	67	+
	68	+/*
	69	+ * called under cgroup_lock()
	70	+ */
	71	+static int dev_whitelist_copy(struct list_head dest, struct list_head orig)
	72	+{
	73	+ struct dev_whitelist_item wh, tmp, *new;
	74	+
	75	+ list_for_each_entry(wh, orig, list) {
	76	+ new = kmalloc(sizeof(*wh), GFP_KERNEL);
	77	+ if (!new)
	78	+ goto free_and_exit;
	79	+ new->major = wh->major;
	80	+ new->minor = wh->minor;
	81	+ new->type = wh->type;
	82	+ new->access = wh->access;
	83	+ list_add_tail(&new->list, dest);
	84	+ }
	85	+
	86	+ return 0;
	87	+
	88	+free_and_exit:
	89	+ list_for_each_entry_safe(wh, tmp, dest, list) {
	90	+ list_del(&wh->list);
	91	+ kfree(wh);
	92	+ }
	93	+ return -ENOMEM;
	94	+}
	95	+
	96	+/* Stupid prototype - don't bother combining existing entries */
	97	+/*
	98	+ * called under cgroup_lock()
	99	+ * since the list is visible to other tasks, we need the spinlock also
	100	+ */
	101	+static int dev_whitelist_add(struct dev_cgroup *dev_cgroup,
	102	+ struct dev_whitelist_item *wh)
	103	+{
	104	+ struct dev_whitelist_item *whcopy;
	105	+
	106	+ whcopy = kmalloc(sizeof(*whcopy), GFP_KERNEL);
	107	+ if (!whcopy)
	108	+ return -ENOMEM;
	109	+
	110	+ memcpy(whcopy, wh, sizeof(*whcopy));
	111	+ spin_lock(&dev_cgroup->lock);
	112	+ list_add_tail(&whcopy->list, &dev_cgroup->whitelist);
	113	+ spin_unlock(&dev_cgroup->lock);
	114	+ return 0;
	115	+}
	116	+
	117	+/*
	118	+ * called under cgroup_lock()
	119	+ * since the list is visible to other tasks, we need the spinlock also
	120	+ */
	121	+static void dev_whitelist_rm(struct dev_cgroup *dev_cgroup,
	122	+ struct dev_whitelist_item *wh)
	123	+{
	124	+ struct dev_whitelist_item walk, tmp;
	125	+
	126	+ spin_lock(&dev_cgroup->lock);
	127	+ list_for_each_entry_safe(walk, tmp, &dev_cgroup->whitelist, list) {
	128	+ if (walk->type == DEV_ALL)
	129	+ goto remove;
	130	+ if (walk->type != wh->type)
	131	+ continue;
	132	+ if (walk->major != ~0 && walk->major != wh->major)
	133	+ continue;
	134	+ if (walk->minor != ~0 && walk->minor != wh->minor)
	135	+ continue;
	136	+
	137	+remove:
	138	+ walk->access &= ~wh->access;
	139	+ if (!walk->access) {
	140	+ list_del(&walk->list);
	141	+ kfree(walk);
	142	+ }
	143	+ }
	144	+ spin_unlock(&dev_cgroup->lock);
	145	+}
	146	+
	147	+/*
	148	+ * called from kernel/cgroup.c with cgroup_lock() held.
	149	+ */
	150	+static struct cgroup_subsys_state devcgroup_create(struct cgroup_subsys ss,
	151	+ struct cgroup *cgroup)
	152	+{
	153	+ struct dev_cgroup dev_cgroup, parent_dev_cgroup;
	154	+ struct cgroup *parent_cgroup;
	155	+ int ret;
	156	+
	157	+ dev_cgroup = kzalloc(sizeof(*dev_cgroup), GFP_KERNEL);
	158	+ if (!dev_cgroup)
	159	+ return ERR_PTR(-ENOMEM);
	160	+ INIT_LIST_HEAD(&dev_cgroup->whitelist);
	161	+ parent_cgroup = cgroup->parent;
	162	+
	163	+ if (parent_cgroup == NULL) {
	164	+ struct dev_whitelist_item *wh;
	165	+ wh = kmalloc(sizeof(*wh), GFP_KERNEL);
	166	+ if (!wh) {
	167	+ kfree(dev_cgroup);
	168	+ return ERR_PTR(-ENOMEM);
	169	+ }
	170	+ wh->minor = wh->major = ~0;
	171	+ wh->type = DEV_ALL;
	172	+ wh->access = ACC_MKNOD \| ACC_READ \| ACC_WRITE;
	173	+ list_add(&wh->list, &dev_cgroup->whitelist);
	174	+ } else {
	175	+ parent_dev_cgroup = cgroup_to_devcgroup(parent_cgroup);
	176	+ ret = dev_whitelist_copy(&dev_cgroup->whitelist,
	177	+ &parent_dev_cgroup->whitelist);
	178	+ if (ret) {
	179	+ kfree(dev_cgroup);
	180	+ return ERR_PTR(ret);
	181	+ }
	182	+ }
	183	+
	184	+ spin_lock_init(&dev_cgroup->lock);
	185	+ return &dev_cgroup->css;
	186	+}
	187	+
	188	+static void devcgroup_destroy(struct cgroup_subsys *ss,
	189	+ struct cgroup *cgroup)
	190	+{
	191	+ struct dev_cgroup *dev_cgroup;
	192	+ struct dev_whitelist_item wh, tmp;
	193	+
	194	+ dev_cgroup = cgroup_to_devcgroup(cgroup);
	195	+ list_for_each_entry_safe(wh, tmp, &dev_cgroup->whitelist, list) {
	196	+ list_del(&wh->list);
	197	+ kfree(wh);
	198	+ }
	199	+ kfree(dev_cgroup);
	200	+}
	201	+
	202	+#define DEVCG_ALLOW 1
	203	+#define DEVCG_DENY 2
	204	+
	205	+static void set_access(char *acc, short access)
	206	+{
	207	+ int idx = 0;
	208	+ memset(acc, 0, 4);
	209	+ if (access & ACC_READ)
	210	+ acc[idx++] = 'r';
	211	+ if (access & ACC_WRITE)
	212	+ acc[idx++] = 'w';
	213	+ if (access & ACC_MKNOD)
	214	+ acc[idx++] = 'm';
	215	+}
	216	+
	217	+static char type_to_char(short type)
	218	+{
	219	+ if (type == DEV_ALL)
	220	+ return 'a';
	221	+ if (type == DEV_CHAR)
	222	+ return 'c';
	223	+ if (type == DEV_BLOCK)
	224	+ return 'b';
	225	+ return 'X';
	226	+}
	227	+
	228	+static void set_majmin(char *str, int len, unsigned m)
	229	+{
	230	+ memset(str, 0, len);
	231	+ if (m == ~0)
	232	+ sprintf(str, "*");
	233	+ else
	234	+ snprintf(str, len, "%d", m);
	235	+}
	236	+
	237	+static char print_whitelist(struct dev_cgroup devcgroup, int *len)
	238	+{
	239	+ char buf, s, acc[4];
	240	+ struct dev_whitelist_item *wh;
	241	+ int ret;
	242	+ int count = 0;
	243	+ char maj[10], min[10];
	244	+
	245	+ buf = kmalloc(4096, GFP_KERNEL);
	246	+ if (!buf)
	247	+ return ERR_PTR(-ENOMEM);
	248	+ s = buf;
	249	+ *s = '\0';
	250	+ *len = 0;
	251	+
	252	+ spin_lock(&devcgroup->lock);
	253	+ list_for_each_entry(wh, &devcgroup->whitelist, list) {
	254	+ set_access(acc, wh->access);
	255	+ set_majmin(maj, 10, wh->major);
	256	+ set_majmin(min, 10, wh->minor);
	257	+ ret = snprintf(s, 4095-(s-buf), "%c %s:%s %s\n",
	258	+ type_to_char(wh->type), maj, min, acc);
	259	+ if (s+ret >= buf+4095) {
	260	+ kfree(buf);
	261	+ buf = ERR_PTR(-ENOMEM);
	262	+ break;
	263	+ }
	264	+ s += ret;
	265	+ *len += ret;
	266	+ count++;
	267	+ }
	268	+ spin_unlock(&devcgroup->lock);
	269	+
	270	+ return buf;
	271	+}
	272	+
	273	+static ssize_t devcgroup_access_read(struct cgroup *cgroup,
	274	+ struct cftype cft, struct file file,
	275	+ char __user userbuf, size_t nbytes, loff_t ppos)
	276	+{
	277	+ struct dev_cgroup *devcgroup = cgroup_to_devcgroup(cgroup);
	278	+ int filetype = cft->private;
	279	+ char *buffer;
	280	+ int uninitialized_var(len);
	281	+ int retval;
	282	+
	283	+ if (filetype != DEVCG_ALLOW)
	284	+ return -EINVAL;
	285	+ buffer = print_whitelist(devcgroup, &len);
	286	+ if (IS_ERR(buffer))
	287	+ return PTR_ERR(buffer);
	288	+
	289	+ retval = simple_read_from_buffer(userbuf, nbytes, ppos, buffer, len);
	290	+ kfree(buffer);
	291	+ return retval;
	292	+}
	293	+
	294	+/*
	295	+ * may_access_whitelist:
	296	+ * does the access granted to dev_cgroup c contain the access
	297	+ * requested in whitelist item refwh.
	298	+ * return 1 if yes, 0 if no.
	299	+ * call with c->lock held
	300	+ */
	301	+static int may_access_whitelist(struct dev_cgroup *c,
	302	+ struct dev_whitelist_item *refwh)
	303	+{
	304	+ struct dev_whitelist_item *whitem;
	305	+
	306	+ list_for_each_entry(whitem, &c->whitelist, list) {
	307	+ if (whitem->type & DEV_ALL)
	308	+ return 1;
	309	+ if ((refwh->type & DEV_BLOCK) && !(whitem->type & DEV_BLOCK))
	310	+ continue;
	311	+ if ((refwh->type & DEV_CHAR) && !(whitem->type & DEV_CHAR))
	312	+ continue;
	313	+ if (whitem->major != ~0 && whitem->major != refwh->major)
	314	+ continue;
	315	+ if (whitem->minor != ~0 && whitem->minor != refwh->minor)
	316	+ continue;
	317	+ if (refwh->access & (~(whitem->access \| ACC_MASK)))
	318	+ continue;
	319	+ return 1;
	320	+ }
	321	+ return 0;
	322	+}
	323	+
	324	+/*
	325	+ * parent_has_perm:
	326	+ * when adding a new allow rule to a device whitelist, the rule
	327	+ * must be allowed in the parent device
	328	+ */
	329	+static int parent_has_perm(struct cgroup *childcg,
	330	+ struct dev_whitelist_item *wh)
	331	+{
	332	+ struct cgroup *pcg = childcg->parent;
	333	+ struct dev_cgroup *parent;
	334	+ int ret;
	335	+
	336	+ if (!pcg)
	337	+ return 1;
	338	+ parent = cgroup_to_devcgroup(pcg);
	339	+ spin_lock(&parent->lock);
	340	+ ret = may_access_whitelist(parent, wh);
	341	+ spin_unlock(&parent->lock);
	342	+ return ret;
	343	+}
	344	+
	345	+/*
	346	+ * Modify the whitelist using allow/deny rules.
	347	+ * CAP_SYS_ADMIN is needed for this. It's at least separate from CAP_MKNOD
	348	+ * so we can give a container CAP_MKNOD to let it create devices but not
	349	+ * modify the whitelist.
	350	+ * It seems likely we'll want to add a CAP_CONTAINER capability to allow
	351	+ * us to also grant CAP_SYS_ADMIN to containers without giving away the
	352	+ * device whitelist controls, but for now we'll stick with CAP_SYS_ADMIN
	353	+ *
	354	+ * Taking rules away is always allowed (given CAP_SYS_ADMIN). Granting
	355	+ * new access is only allowed if you're in the top-level cgroup, or your
	356	+ * parent cgroup has the access you're asking for.
	357	+ */
	358	+static ssize_t devcgroup_access_write(struct cgroup cgroup, struct cftype cft,
	359	+ struct file file, const char __user userbuf,
	360	+ size_t nbytes, loff_t *ppos)
	361	+{
	362	+ struct cgroup *cur_cgroup;
	363	+ struct dev_cgroup devcgroup, cur_devcgroup;
	364	+ int filetype = cft->private;
	365	+ char buffer, b;
	366	+ int retval = 0, count;
	367	+ struct dev_whitelist_item wh;
	368	+
	369	+ if (!capable(CAP_SYS_ADMIN))
	370	+ return -EPERM;
	371	+
	372	+ devcgroup = cgroup_to_devcgroup(cgroup);
	373	+ cur_cgroup = task_cgroup(current, devices_subsys.subsys_id);
	374	+ cur_devcgroup = cgroup_to_devcgroup(cur_cgroup);
	375	+
	376	+ buffer = kmalloc(nbytes+1, GFP_KERNEL);
	377	+ if (!buffer)
	378	+ return -ENOMEM;
	379	+
	380	+ if (copy_from_user(buffer, userbuf, nbytes)) {
	381	+ retval = -EFAULT;
	382	+ goto out1;
	383	+ }
	384	+ buffer[nbytes] = 0; /* nul-terminate */
	385	+
	386	+ cgroup_lock();
	387	+ if (cgroup_is_removed(cgroup)) {
	388	+ retval = -ENODEV;
	389	+ goto out2;
	390	+ }
	391	+
	392	+ memset(&wh, 0, sizeof(wh));
	393	+ b = buffer;
	394	+
	395	+ switch (*b) {
	396	+ case 'a':
	397	+ wh.type = DEV_ALL;
	398	+ wh.access = ACC_MASK;
	399	+ goto handle;
	400	+ case 'b':
	401	+ wh.type = DEV_BLOCK;
	402	+ break;
	403	+ case 'c':
	404	+ wh.type = DEV_CHAR;
	405	+ break;
	406	+ default:
	407	+ retval = -EINVAL;
	408	+ goto out2;
	409	+ }
	410	+ b++;
	411	+ if (!isspace(*b)) {
	412	+ retval = -EINVAL;
	413	+ goto out2;
	414	+ }
	415	+ b++;
	416	+ if (b == '') {
	417	+ wh.major = ~0;
	418	+ b++;
	419	+ } else if (isdigit(*b)) {
	420	+ wh.major = 0;
	421	+ while (isdigit(*b)) {
	422	+ wh.major = wh.major10+(b-'0');
	423	+ b++;
	424	+ }
	425	+ } else {
	426	+ retval = -EINVAL;
	427	+ goto out2;
	428	+ }
	429	+ if (*b != ':') {
	430	+ retval = -EINVAL;
	431	+ goto out2;
	432	+ }
	433	+ b++;
	434	+
	435	+ /* read minor */
	436	+ if (b == '') {
	437	+ wh.minor = ~0;
	438	+ b++;
	439	+ } else if (isdigit(*b)) {
	440	+ wh.minor = 0;
	441	+ while (isdigit(*b)) {
	442	+ wh.minor = wh.minor10+(b-'0');
	443	+ b++;
	444	+ }
	445	+ } else {
	446	+ retval = -EINVAL;
	447	+ goto out2;
	448	+ }
	449	+ if (!isspace(*b)) {
	450	+ retval = -EINVAL;
	451	+ goto out2;
	452	+ }
	453	+ for (b++, count = 0; count < 3; count++, b++) {
	454	+ switch (*b) {
	455	+ case 'r':
	456	+ wh.access \|= ACC_READ;
	457	+ break;
	458	+ case 'w':
	459	+ wh.access \|= ACC_WRITE;
	460	+ break;
	461	+ case 'm':
	462	+ wh.access \|= ACC_MKNOD;
	463	+ break;
	464	+ case '\n':
	465	+ case '\0':
	466	+ count = 3;
	467	+ break;
	468	+ default:
	469	+ retval = -EINVAL;
	470	+ goto out2;
	471	+ }
	472	+ }
	473	+
	474	+handle:
	475	+ retval = 0;
	476	+ switch (filetype) {
	477	+ case DEVCG_ALLOW:
	478	+ if (!parent_has_perm(cgroup, &wh))
	479	+ retval = -EPERM;
	480	+ else
	481	+ retval = dev_whitelist_add(devcgroup, &wh);
	482	+ break;
	483	+ case DEVCG_DENY:
	484	+ dev_whitelist_rm(devcgroup, &wh);
	485	+ break;
	486	+ default:
	487	+ retval = -EINVAL;
	488	+ goto out2;
	489	+ }
	490	+
	491	+ if (retval == 0)
	492	+ retval = nbytes;
	493	+
	494	+out2:
	495	+ cgroup_unlock();
	496	+out1:
	497	+ kfree(buffer);
	498	+ return retval;
	499	+}
	500	+
	501	+static struct cftype dev_cgroup_files[] = {
	502	+ {
	503	+ .name = "allow",
	504	+ .read = devcgroup_access_read,
	505	+ .write = devcgroup_access_write,
	506	+ .private = DEVCG_ALLOW,
	507	+ },
	508	+ {
	509	+ .name = "deny",
	510	+ .write = devcgroup_access_write,
	511	+ .private = DEVCG_DENY,
	512	+ },
	513	+};
	514	+
	515	+static int devcgroup_populate(struct cgroup_subsys *ss,
	516	+ struct cgroup *cgroup)
	517	+{
	518	+ return cgroup_add_files(cgroup, ss, dev_cgroup_files,
	519	+ ARRAY_SIZE(dev_cgroup_files));
	520	+}
	521	+
	522	+struct cgroup_subsys devices_subsys = {
	523	+ .name = "devices",
	524	+ .can_attach = devcgroup_can_attach,
	525	+ .create = devcgroup_create,
	526	+ .destroy = devcgroup_destroy,
	527	+ .populate = devcgroup_populate,
	528	+ .subsys_id = devices_subsys_id,
	529	+};
	530	+
	531	+int devcgroup_inode_permission(struct inode *inode, int mask)
	532	+{
	533	+ struct cgroup *cgroup;
	534	+ struct dev_cgroup *dev_cgroup;
	535	+ struct dev_whitelist_item *wh;
	536	+
	537	+ dev_t device = inode->i_rdev;
	538	+ if (!device)
	539	+ return 0;
	540	+ if (!S_ISBLK(inode->i_mode) && !S_ISCHR(inode->i_mode))
	541	+ return 0;
	542	+ cgroup = task_cgroup(current, devices_subsys.subsys_id);
	543	+ dev_cgroup = cgroup_to_devcgroup(cgroup);
	544	+ if (!dev_cgroup)
	545	+ return 0;
	546	+
	547	+ spin_lock(&dev_cgroup->lock);
	548	+ list_for_each_entry(wh, &dev_cgroup->whitelist, list) {
	549	+ if (wh->type & DEV_ALL)
	550	+ goto acc_check;
	551	+ if ((wh->type & DEV_BLOCK) && !S_ISBLK(inode->i_mode))
	552	+ continue;
	553	+ if ((wh->type & DEV_CHAR) && !S_ISCHR(inode->i_mode))
	554	+ continue;
	555	+ if (wh->major != ~0 && wh->major != imajor(inode))
	556	+ continue;
	557	+ if (wh->minor != ~0 && wh->minor != iminor(inode))
	558	+ continue;
	559	+acc_check:
	560	+ if ((mask & MAY_WRITE) && !(wh->access & ACC_WRITE))
	561	+ continue;
	562	+ if ((mask & MAY_READ) && !(wh->access & ACC_READ))
	563	+ continue;
	564	+ spin_unlock(&dev_cgroup->lock);
	565	+ return 0;
	566	+ }
	567	+ spin_unlock(&dev_cgroup->lock);
	568	+
	569	+ return -EPERM;
	570	+}
	571	+
	572	+int devcgroup_inode_mknod(int mode, dev_t dev)
	573	+{
	574	+ struct cgroup *cgroup;
	575	+ struct dev_cgroup *dev_cgroup;
	576	+ struct dev_whitelist_item *wh;
	577	+
	578	+ cgroup = task_cgroup(current, devices_subsys.subsys_id);
	579	+ dev_cgroup = cgroup_to_devcgroup(cgroup);
	580	+ if (!dev_cgroup)
	581	+ return 0;
	582	+
	583	+ spin_lock(&dev_cgroup->lock);
	584	+ list_for_each_entry(wh, &dev_cgroup->whitelist, list) {
	585	+ if (wh->type & DEV_ALL)
	586	+ goto acc_check;
	587	+ if ((wh->type & DEV_BLOCK) && !S_ISBLK(mode))
	588	+ continue;
	589	+ if ((wh->type & DEV_CHAR) && !S_ISCHR(mode))
	590	+ continue;
	591	+ if (wh->major != ~0 && wh->major != MAJOR(dev))
	592	+ continue;
	593	+ if (wh->minor != ~0 && wh->minor != MINOR(dev))
	594	+ continue;
	595	+acc_check:
	596	+ if (!(wh->access & ACC_MKNOD))
	597	+ continue;
	598	+ spin_unlock(&dev_cgroup->lock);
	599	+ return 0;
	600	+ }
	601	+ spin_unlock(&dev_cgroup->lock);
	602	+ return -EPERM;
	603	+}