Merge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/ebiederm/user-namespace

Pull userns fixes from Eric W Biederman:
"The bulk of the changes are fixing the worst consequences of the user
namespace design oversight in not considering what happens when one
namespace starts off as a clone of another namespace, as happens with
the mount namespace.

The rest of the changes are just plain bug fixes.

Many thanks to Andy Lutomirski for pointing out many of these issues."

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/ebiederm/user-namespace:
userns: Restrict when proc and sysfs can be mounted
ipc: Restrict mounting the mqueue filesystem
vfs: Carefully propogate mounts across user namespaces
vfs: Add a mount flag to lock read only bind mounts
userns: Don't allow creation if the user is chrooted
yama: Better permission check for ptraceme
pid: Handle the exit of a multi-threaded init.
scm: Require CAP_SYS_ADMIN over the current pidns to spoof pids.

fs/namespace.c

···
 	}
 
 	mnt->mnt.mnt_flags = old->mnt.mnt_flags & ~MNT_WRITE_HOLD;
+	/* Don't allow unprivileged users to change mount flags */
+	if ((flag & CL_UNPRIVILEGED) && (mnt->mnt.mnt_flags & MNT_READONLY))
+		mnt->mnt.mnt_flags |= MNT_LOCK_READONLY;
+
 	atomic_inc(&sb->s_active);
 	mnt->mnt.mnt_sb = sb;
 	mnt->mnt.mnt_root = dget(root);
···
 	if (readonly_request == __mnt_is_readonly(mnt))
 		return 0;
 
+	if (mnt->mnt_flags & MNT_LOCK_READONLY)
+		return -EPERM;
+
 	if (readonly_request)
 		error = mnt_make_readonly(real_mount(mnt));
 	else
···
 	/* First pass: copy the tree topology */
 	copy_flags = CL_COPY_ALL | CL_EXPIRE;
 	if (user_ns != mnt_ns->user_ns)
-		copy_flags |= CL_SHARED_TO_SLAVE;
+		copy_flags |= CL_SHARED_TO_SLAVE | CL_UNPRIVILEGED;
 	new = copy_tree(old, old->mnt.mnt_root, copy_flags);
 	if (IS_ERR(new)) {
 		up_write(&namespace_sem);
···
 bool our_mnt(struct vfsmount *mnt)
 {
 	return check_mnt(real_mount(mnt));
+}
+
+bool current_chrooted(void)
+{
+	/* Does the current process have a non-standard root */
+	struct path ns_root;
+	struct path fs_root;
+	bool chrooted;
+
+	/* Find the namespace root */
+	ns_root.mnt = &current->nsproxy->mnt_ns->root->mnt;
+	ns_root.dentry = ns_root.mnt->mnt_root;
+	path_get(&ns_root);
+	while (d_mountpoint(ns_root.dentry) && follow_down_one(&ns_root))
+		;
+
+	get_fs_root(current->fs, &fs_root);
+
+	chrooted = !path_equal(&fs_root, &ns_root);
+
+	path_put(&fs_root);
+	path_put(&ns_root);
+
+	return chrooted;
+}
+
+void update_mnt_policy(struct user_namespace *userns)
+{
+	struct mnt_namespace *ns = current->nsproxy->mnt_ns;
+	struct mount *mnt;
+
+	down_read(&namespace_sem);
+	list_for_each_entry(mnt, &ns->list, mnt_list) {
+		switch (mnt->mnt.mnt_sb->s_magic) {
+		case SYSFS_MAGIC:
+			userns->may_mount_sysfs = true;
+			break;
+		case PROC_SUPER_MAGIC:
+			userns->may_mount_proc = true;
+			break;
+		}
+		if (userns->may_mount_sysfs && userns->may_mount_proc)
+			break;
+	}
+	up_read(&namespace_sem);
 }
 
 static void *mntns_get(struct task_struct *task)

fs/pnode.c

···
 #include <linux/mnt_namespace.h>
 #include <linux/mount.h>
 #include <linux/fs.h>
+#include <linux/nsproxy.h>
 #include "internal.h"
 #include "pnode.h"
 
···
 int propagate_mnt(struct mount *dest_mnt, struct dentry *dest_dentry,
 		    struct mount *source_mnt, struct list_head *tree_list)
 {
+	struct user_namespace *user_ns = current->nsproxy->mnt_ns->user_ns;
 	struct mount *m, *child;
 	int ret = 0;
 	struct mount *prev_dest_mnt = dest_mnt;
···
 			continue;
 
 		source =  get_source(m, prev_dest_mnt, prev_src_mnt, &type);
+
+		/* Notice when we are propagating across user namespaces */
+		if (m->mnt_ns->user_ns != user_ns)
+			type |= CL_UNPRIVILEGED;
 
 		child = copy_tree(source, source->mnt.mnt_root, type);
 		if (IS_ERR(child)) {

fs/pnode.h

···
 #define CL_MAKE_SHARED 		0x08
 #define CL_PRIVATE 		0x10
 #define CL_SHARED_TO_SLAVE	0x20
+#define CL_UNPRIVILEGED		0x40
 
 static inline void set_mnt_shared(struct mount *mnt)
 {

fs/proc/root.c

···
 #include <linux/sched.h>
 #include <linux/module.h>
 #include <linux/bitops.h>
+#include <linux/user_namespace.h>
 #include <linux/mount.h>
 #include <linux/pid_namespace.h>
 #include <linux/parser.h>
···
 	} else {
 		ns = task_active_pid_ns(current);
 		options = data;
+
+		if (!current_user_ns()->may_mount_proc)
+			return ERR_PTR(-EPERM);
 	}
 
 	sb = sget(fs_type, proc_test_super, proc_set_super, flags, ns);

fs/sysfs/mount.c

···
 #include <linux/module.h>
 #include <linux/magic.h>
 #include <linux/slab.h>
+#include <linux/user_namespace.h>
 
 #include "sysfs.h"
 
···
 	enum kobj_ns_type type;
 	struct super_block *sb;
 	int error;
+
+	if (!(flags & MS_KERNMOUNT) && !current_user_ns()->may_mount_sysfs)
+		return ERR_PTR(-EPERM);
 
 	info = kzalloc(sizeof(*info), GFP_KERNEL);
 	if (!info)

include/linux/fs_struct.h

···
 	spin_unlock(&fs->lock);
 }
 
+extern bool current_chrooted(void);
+
 #endif /* _LINUX_FS_STRUCT_H */

include/linux/mount.h

···
 
 #define MNT_INTERNAL	0x4000
 
+#define MNT_LOCK_READONLY	0x400000
+
 struct vfsmount {
 	struct dentry *mnt_root;	/* root of the mounted tree */
 	struct super_block *mnt_sb;	/* pointer to superblock */

include/linux/user_namespace.h

···
 	kuid_t			owner;
 	kgid_t			group;
 	unsigned int		proc_inum;
+	bool			may_mount_sysfs;
+	bool			may_mount_proc;
 };
 
 extern struct user_namespace init_user_ns;
···
 }
 
 #endif
+
+void update_mnt_policy(struct user_namespace *userns);
 
 #endif /* _LINUX_USER_H */

ipc/mqueue.c

···
 			 int flags, const char *dev_name,
 			 void *data)
 {
-	if (!(flags & MS_KERNMOUNT))
-		data = current->nsproxy->ipc_ns;
+	if (!(flags & MS_KERNMOUNT)) {
+		struct ipc_namespace *ns = current->nsproxy->ipc_ns;
+		/* Don't allow mounting unless the caller has CAP_SYS_ADMIN
+		 * over the ipc namespace.
+		 */
+		if (!ns_capable(ns->user_ns, CAP_SYS_ADMIN))
+			return ERR_PTR(-EPERM);
+
+		data = ns;
+	}
 	return mount_ns(fs_type, flags, data, mqueue_fill_super);
 }
 

kernel/pid_namespace.c

···
 	int nr;
 	int rc;
 	struct task_struct *task, *me = current;
+	int init_pids = thread_group_leader(me) ? 1 : 2;
 
 	/* Don't allow any more processes into the pid namespace */
 	disable_pid_allocation(pid_ns);
···
 	 */
 	for (;;) {
 		set_current_state(TASK_UNINTERRUPTIBLE);
-		if (pid_ns->nr_hashed == 1)
+		if (pid_ns->nr_hashed == init_pids)
 			break;
 		schedule();
 	}

kernel/user.c

···
 	.owner = GLOBAL_ROOT_UID,
 	.group = GLOBAL_ROOT_GID,
 	.proc_inum = PROC_USER_INIT_INO,
+	.may_mount_sysfs = true,
+	.may_mount_proc = true,
 };
 EXPORT_SYMBOL_GPL(init_user_ns);
 

kernel/user_namespace.c

···
 	kgid_t group = new->egid;
 	int ret;
 
+	/*
+	 * Verify that we can not violate the policy of which files
+	 * may be accessed that is specified by the root directory,
+	 * by verifing that the root directory is at the root of the
+	 * mount namespace which allows all files to be accessed.
+	 */
+	if (current_chrooted())
+		return -EPERM;
+
 	/* The creator needs a mapping in the parent user namespace
 	 * or else we won't be able to reasonably tell userspace who
 	 * created a user_namespace.
···
 	ns->group = group;
 
 	set_cred_user_ns(new, ns);
+
+	update_mnt_policy(ns);
 
 	return 0;
 }

net/core/scm.c

···
 #include <linux/interrupt.h>
 #include <linux/netdevice.h>
 #include <linux/security.h>
+#include <linux/pid_namespace.h>
 #include <linux/pid.h>
 #include <linux/nsproxy.h>
 #include <linux/slab.h>
···
 	if (!uid_valid(uid) || !gid_valid(gid))
 		return -EINVAL;
 
-	if ((creds->pid == task_tgid_vnr(current) || nsown_capable(CAP_SYS_ADMIN)) &&
+	if ((creds->pid == task_tgid_vnr(current) ||
+	     ns_capable(current->nsproxy->pid_ns->user_ns, CAP_SYS_ADMIN)) &&
 	    ((uid_eq(uid, cred->uid)   || uid_eq(uid, cred->euid) ||
 	      uid_eq(uid, cred->suid)) || nsown_capable(CAP_SETUID)) &&
 	    ((gid_eq(gid, cred->gid)   || gid_eq(gid, cred->egid) ||

security/yama/yama_lsm.c

···
 	/* Only disallow PTRACE_TRACEME on more aggressive settings. */
 	switch (ptrace_scope) {
 	case YAMA_SCOPE_CAPABILITY:
-		rcu_read_lock();
-		if (!ns_capable(__task_cred(parent)->user_ns, CAP_SYS_PTRACE))
+		if (!has_ns_capability(parent, current_user_ns(), CAP_SYS_PTRACE))
 			rc = -EPERM;
-		rcu_read_unlock();
 		break;
 	case YAMA_SCOPE_NO_ATTACH:
 		rc = -EPERM;