cgroups: implement device whitelist
authorSerge E. Hallyn <serue@us.ibm.com>
Tue, 29 Apr 2008 08:00:10 +0000 (01:00 -0700)
committerLinus Torvalds <torvalds@linux-foundation.org>
Tue, 29 Apr 2008 15:06:09 +0000 (08:06 -0700)
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>
Documentation/controllers/devices.txt [new file with mode: 0644]
fs/namei.c
include/linux/cgroup_subsys.h
include/linux/device_cgroup.h [new file with mode: 0644]
init/Kconfig
security/Makefile
security/device_cgroup.c [new file with mode: 0644]

diff --git a/Documentation/controllers/devices.txt b/Documentation/controllers/devices.txt
new file mode 100644 (file)
index 0000000..4dcea42
--- /dev/null
@@ -0,0 +1,48 @@
+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.
index e179f71bfcb058df613f83cb1ff923d47ffc4618..32fd9655485b7fc9b8c8616ad6a61f6ac243b328 100644 (file)
@@ -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 @@ int permission(struct inode *inode, int mask, struct nameidata *nd)
        if (retval)
                return retval;
 
+       retval = devcgroup_inode_permission(inode, mask);
+       if (retval)
+               return retval;
+
        return security_inode_permission(inode, mask, nd);
 }
 
@@ -2028,6 +2033,10 @@ int vfs_mknod(struct inode *dir, struct dentry *dentry, int mode, dev_t dev)
        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)
                return error;
index 1ddebfc52565f3eb4894c422c1fc5dd831ae523c..e2877454ec82df3205c0251f9c9759db9334dedb 100644 (file)
@@ -42,3 +42,9 @@ SUBSYS(mem_cgroup)
 #endif
 
 /* */
+
+#ifdef CONFIG_CGROUP_DEVICE
+SUBSYS(devices)
+#endif
+
+/* */
diff --git a/include/linux/device_cgroup.h b/include/linux/device_cgroup.h
new file mode 100644 (file)
index 0000000..0b0d9c3
--- /dev/null
@@ -0,0 +1,12 @@
+#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
index 6ce16bdbec7608a8a4151d71e9567b74397dcc6e..a3457926342a5115ae494e53918fc273ca926f47 100644 (file)
@@ -298,6 +298,13 @@ config CGROUP_NS
           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
index 9e8b025250144ce739e80583a095b3059d89110c..7ef1107a72875412e3cdc52f48fe609f7dfbc764 100644 (file)
@@ -18,3 +18,4 @@ obj-$(CONFIG_SECURITY_SELINUX)                += selinux/built-in.o
 obj-$(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
diff --git a/security/device_cgroup.c b/security/device_cgroup.c
new file mode 100644 (file)
index 0000000..4237b19
--- /dev/null
@@ -0,0 +1,603 @@
+/*
+ * 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;
+}