keys: request_key() interception in containers

@@ -62,8 +62,10 @@ struct key_match_data {
  * kernel managed key type definition
  */
 struct key_type {
+	struct module *owner;
+
 	/* name of the type */
-	const char *name;
+	const char name[24];
 
 	/* default payload length for quota precalculation (optional)
 	 * - this can be used instead of calling key_payload_reserve(), that

@@ -73,6 +73,8 @@ struct user_namespace {
 	struct list_head	keyring_name_list;
 	struct key		*user_keyring_register;
 	struct rw_semaphore	keyring_sem;
+	struct hlist_head	request_key_services;
+	spinlock_t		request_key_services_lock;
 #endif
 
 	/* Register of per-UID persistent keyrings for this namespace */

@@ -36,6 +36,18 @@
 #define KEY_REQKEY_DEFL_GROUP_KEYRING		6
 #define KEY_REQKEY_DEFL_REQUESTOR_KEYRING	7
 
+/* Request_key service daemon namespace selection specifiers. */
+#define KEY_SERVICE_NS_UTS		0x0001
+#define KEY_SERVICE_NS_IPC		0x0002
+#define KEY_SERVICE_NS_MNT		0x0004
+#define KEY_SERVICE_NS_PID		0x0008
+#define KEY_SERVICE_NS_NET		0x0010
+#define KEY_SERVICE_NS_CGROUP		0x0020
+#define KEY_SERVICE___ALL_NS		0x003f
+
+#define KEY_SERVICE_FD_CLOEXEC		0x0001
+#define KEY_SERVICE_FD_NONBLOCK		0x0002
+
 /* keyctl commands */
 #define KEYCTL_GET_KEYRING_ID		0	/* ask for a keyring's ID */
 #define KEYCTL_JOIN_SESSION_KEYRING	1	/* join or start named session keyring */

@@ -70,6 +82,7 @@
 #define KEYCTL_MOVE			30	/* Move keys between keyrings */
 #define KEYCTL_CAPABILITIES		31	/* Find capabilities of keyrings subsystem */
 #define KEYCTL_WATCH_KEY		32	/* Watch a key or ring of keys for changes */
+#define KEYCTL_SERVICE_INTERCEPT	33	/* Intercept request_key services on a user_ns */
 
 /* keyctl structures */
 struct keyctl_dh_params {

@@ -71,6 +71,9 @@ struct user_namespace init_user_ns = {
 #ifdef CONFIG_KEYS
 	.keyring_name_list = LIST_HEAD_INIT(init_user_ns.keyring_name_list),
 	.keyring_sem = __RWSEM_INITIALIZER(init_user_ns.keyring_sem),
+	.request_key_services = HLIST_HEAD_INIT,
+	.request_key_services_lock =
+	__SPIN_LOCK_UNLOCKED(init_user_ns.request_key_services_lock),
 #endif
 };
 EXPORT_SYMBOL_GPL(init_user_ns);

@@ -130,6 +130,8 @@ int create_user_ns(struct cred *new)
 #ifdef CONFIG_KEYS
 	INIT_LIST_HEAD(&ns->keyring_name_list);
 	init_rwsem(&ns->keyring_sem);
+	INIT_HLIST_HEAD(&ns->request_key_services);
+	spin_lock_init(&ns->request_key_services_lock);
 #endif
 	ret = -ENOMEM;
 	if (!setup_userns_sysctls(ns))

@@ -1,4 +1,6 @@
 # SPDX-License-Identifier: GPL-2.0-only
 userprogs-always-y += watch_test
+userprogs-always-y += key_req_intercept
 
 userccflags += -I usr/include
+userldflags += -lkeyutils

@@ -0,0 +1,271 @@
+// SPDX-License-Identifier: GPL-2.0
+/* Intercept request_key upcalls
+ *
+ * Copyright (C) 2021 Red Hat, Inc. All Rights Reserved.
+ * Written by David Howells (dhowells@redhat.com)
+ */
+
+#define _GNU_SOURCE
+#include <stdbool.h>
+#include <stdarg.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <signal.h>
+#include <unistd.h>
+#include <errno.h>
+#include <sys/ioctl.h>
+#include <limits.h>
+#include <keyutils.h>
+#include <linux/watch_queue.h>
+#include <linux/unistd.h>
+
+#ifndef KEYCTL_WATCH_KEY
+#define KEYCTL_WATCH_KEY 32
+#endif
+#ifndef KEYCTL_SERVICE_INTERCEPT
+#define KEYCTL_SERVICE_INTERCEPT 33
+#endif
+#ifndef __NR_keyctl
+#define __NR_keyctl -1
+#endif
+
+#define BUF_SIZE 256
+
+typedef int key_serial_t;
+key_serial_t queue_keyring;
+
+static long keyctl_watch_key(int key, int watch_fd, int watch_id)
+{
+	return syscall(__NR_keyctl, KEYCTL_WATCH_KEY, key, watch_fd, watch_id);
+}
+
+static long keyctl_service_intercept(int queue_keyring, int userns_fd,
+				     const char *type_name, unsigned int ns_mask)
+{
+	return syscall(__NR_keyctl, KEYCTL_SERVICE_INTERCEPT,
+		       queue_keyring, userns_fd, type_name, ns_mask);
+}
+
+static const char auth_key_type[] = ".request_key_auth;";
+
+/*
+ * Instantiate a key.
+ */
+static void do_instantiation(key_serial_t key, char *desc)
+{
+	printf("INSTANTIATE %u '%s'\n", key, desc);
+
+	if (keyctl_assume_authority(key) == -1) {
+		perror("keyctl_assume_authority");
+		exit(1);
+	}
+
+	if (keyctl_reject(key, 20, ENOANO, 0) == -1) {
+		perror("keyctl_reject");
+		exit(1);
+	}
+}
+
+/*
+ * Process a notification.
+ */
+static void process_request(struct watch_notification *n, size_t len)
+{
+	struct key_notification *k = (struct key_notification *)n;
+	key_serial_t auth_key, key;
+	char desc[1024], *p;
+
+	if (len != sizeof(struct key_notification)) {
+		fprintf(stderr, "Incorrect key message length\n");
+		return;
+	}
+
+	auth_key = k->aux;
+	printf("REQUEST %d aux=%d\n", k->key_id, k->aux);
+
+	if (keyctl_describe(auth_key, desc, sizeof(desc)) == -1) {
+		perror("keyctl_describe(auth_key)");
+		exit(1);
+	}
+
+	printf("AUTH_KEY '%s'\n", desc);
+	if (memcmp(desc, auth_key_type, sizeof(auth_key_type) - 1) != 0) {
+		printf("NOT AUTH_KEY TYPE\n");
+	} else {
+		p = strrchr(desc, ';');
+		if (p) {
+			key = strtoul(p + 1, NULL, 16);
+			printf("KEY '%d'\n", key);
+
+			if (keyctl_describe(key, desc, sizeof(desc)) == -1) {
+				perror("keyctl_describe(key)");
+				exit(1);
+			}
+
+			do_instantiation(key, desc);
+			return;
+		}
+	}
+
+	/* Shouldn't need to do this if we successfully instantiated/rejected
+	 * the target key.
+	 */
+	if (keyctl_unlink(auth_key, queue_keyring) == -1)
+		perror("keyctl_unlink");
+}
+
+/*
+ * Consume and display events.
+ */
+static void consumer(int fd)
+{
+	unsigned char buffer[433], *p, *end;
+	union {
+		struct watch_notification n;
+		unsigned char buf1[128];
+	} n;
+	ssize_t buf_len;
+
+	for (;;) {
+		buf_len = read(fd, buffer, sizeof(buffer));
+		if (buf_len == -1) {
+			perror("read");
+			exit(1);
+		}
+
+		if (buf_len == 0) {
+			printf("-- END --\n");
+			return;
+		}
+
+		if (buf_len > sizeof(buffer)) {
+			fprintf(stderr, "Read buffer overrun: %zd\n", buf_len);
+			return;
+		}
+
+		printf("read() = %zd\n", buf_len);
+
+		p = buffer;
+		end = buffer + buf_len;
+		while (p < end) {
+			size_t largest, len;
+
+			largest = end - p;
+			if (largest > 128)
+				largest = 128;
+			if (largest < sizeof(struct watch_notification)) {
+				fprintf(stderr, "Short message header: %zu\n", largest);
+				return;
+			}
+			memcpy(&n, p, largest);
+
+			printf("NOTIFY[%03zx]: ty=%06x sy=%02x i=%08x\n",
+			       p - buffer, n.n.type, n.n.subtype, n.n.info);
+
+			len = n.n.info & WATCH_INFO_LENGTH;
+			if (len < sizeof(n.n) || len > largest) {
+				fprintf(stderr, "Bad message length: %zu/%zu\n", len, largest);
+				exit(1);
+			}
+
+			switch (n.n.type) {
+			case WATCH_TYPE_META:
+				switch (n.n.subtype) {
+				case WATCH_META_REMOVAL_NOTIFICATION:
+					printf("REMOVAL of watchpoint %08x\n",
+					       (n.n.info & WATCH_INFO_ID) >>
+					       WATCH_INFO_ID__SHIFT);
+					break;
+				case WATCH_META_LOSS_NOTIFICATION:
+					printf("-- LOSS --\n");
+					break;
+				default:
+					printf("other meta record\n");
+					break;
+				}
+				break;
+			case WATCH_TYPE_KEY_NOTIFY:
+				switch (n.n.subtype) {
+				case NOTIFY_KEY_LINKED:
+					process_request(&n.n, len);
+					break;
+				default:
+					printf("other key subtype\n");
+					break;
+				}
+				break;
+			default:
+				printf("other type\n");
+				break;
+			}
+
+			p += len;
+		}
+	}
+}
+
+static struct watch_notification_filter filter = {
+	.nr_filters	= 1,
+	.filters = {
+		[0]	= {
+			.type			= WATCH_TYPE_KEY_NOTIFY,
+			.subtype_filter[0]	= (1 << NOTIFY_KEY_LINKED),
+		},
+	},
+};
+
+static void cleanup(void)
+{
+	printf("--- clean up ---\n");
+	if (keyctl_service_intercept(0, -1, "user", 0) == -1)
+		perror("unintercept");
+	if (keyctl_clear(queue_keyring) == -1)
+		perror("clear");
+	if (keyctl_unlink(queue_keyring, KEY_SPEC_SESSION_KEYRING) == -1)
+		perror("unlink/q");
+}
+
+int main(int argc, char **argv)
+{
+	int pipefd[2], fd;
+
+	queue_keyring = add_key("keyring", "intercept", NULL, 0, KEY_SPEC_SESSION_KEYRING);
+	if (queue_keyring == -1) {
+		perror("add_key");
+		exit(1);
+	}
+
+	printf("QUEUE KEYRING %d\n", queue_keyring);
+
+	if (pipe2(pipefd, O_NOTIFICATION_PIPE) == -1) {
+		perror("pipe2");
+		exit(1);
+	}
+	fd = pipefd[0];
+
+	if (ioctl(fd, IOC_WATCH_QUEUE_SET_SIZE, BUF_SIZE) == -1) {
+		perror("watch_queue(size)");
+		exit(1);
+	}
+
+	if (ioctl(fd, IOC_WATCH_QUEUE_SET_FILTER, &filter) == -1) {
+		perror("watch_queue(filter)");
+		exit(1);
+	}
+
+	if (keyctl_watch_key(queue_keyring, fd, 0x01) == -1) {
+		perror("keyctl_watch_key");
+		exit(1);
+	}
+
+	if (keyctl_service_intercept(queue_keyring, -1, "user", 0) == -1) {
+		perror("keyctl_service_intercept");
+		exit(1);
+	}
+
+	atexit(cleanup);
+
+	consumer(fd);
+	exit(0);
+}

@@ -15,7 +15,9 @@ obj-y := \
 	process_keys.o \
 	request_key.o \
 	request_key_auth.o \
+	service.o \
 	user_defined.o
+
 compat-obj-$(CONFIG_KEY_DH_OPERATIONS) += compat_dh.o
 obj-$(CONFIG_COMPAT) += compat.o $(compat-obj-y)
 obj-$(CONFIG_PROC_FS) += proc.o

@@ -126,6 +126,9 @@ COMPAT_SYSCALL_DEFINE5(keyctl, u32, option,
 	case KEYCTL_WATCH_KEY:
 		return keyctl_watch_key(arg2, arg3, arg4);
 
+	case KEYCTL_SERVICE_INTERCEPT:
+		return keyctl_service_intercept(arg2, arg3, compat_ptr(arg4), arg5);
+
 	default:
 		return -EOPNOTSUPP;
 	}

@@ -150,6 +150,7 @@ extern struct key *find_keyring_by_name(const char *name, bool uid_keyring);
 
 extern int look_up_user_keyrings(struct key **, struct key **);
 extern struct key *get_user_session_keyring_rcu(const struct cred *);
+extern int install_thread_keyring(void);
 extern int install_thread_keyring_to_cred(struct cred *);
 extern int install_process_keyring_to_cred(struct cred *);
 extern int install_session_keyring_to_cred(struct cred *, struct key *);

@@ -183,6 +184,8 @@ extern void key_gc_keytype(struct key_type *ktype);
 extern int key_task_permission(const key_ref_t key_ref,
 			       const struct cred *cred,
 			       enum key_need_perm need_perm);
+extern int queue_request_key(struct key *key, struct key *auth_key);
+extern void clear_request_key_services(struct user_namespace *ns);
 
 static inline void notify_key(struct key *key,
 			      enum key_notification_subtype subtype, u32 aux)

@@ -265,6 +268,8 @@ extern long keyctl_invalidate_key(key_serial_t);
 extern long keyctl_restrict_keyring(key_serial_t id,
 				    const char __user *_type,
 				    const char __user *_restriction);
+extern long keyctl_service_intercept(int, int, const char __user *, unsigned int);
+
 #ifdef CONFIG_PERSISTENT_KEYRINGS
 extern long keyctl_get_persistent(uid_t, key_serial_t);
 extern unsigned persistent_keyring_expiry;

@@ -2015,6 +2015,12 @@ SYSCALL_DEFINE5(keyctl, int, option, unsigned long, arg2, unsigned long, arg3,
 	case KEYCTL_WATCH_KEY:
 		return keyctl_watch_key((key_serial_t)arg2, (int)arg3, (int)arg4);
 
+	case KEYCTL_SERVICE_INTERCEPT:
+		return keyctl_service_intercept((key_serial_t)arg2,
+						(int)arg3,
+						(const char __user *)arg4,
+						(unsigned int)arg5);
+
 	default:
 		return -EOPNOTSUPP;
 	}

@@ -60,6 +60,7 @@ void key_free_user_ns(struct user_namespace *ns)
 	list_del_init(&ns->keyring_name_list);
 	write_unlock(&keyring_name_lock);
 
+	clear_request_key_services(ns);
 	key_put(ns->user_keyring_register);
 #ifdef CONFIG_PERSISTENT_KEYRINGS
 	key_put(ns->persistent_keyring_register);

@@ -241,7 +241,7 @@ int install_thread_keyring_to_cred(struct cred *new)
  *
  * Return: 0 if a thread keyring is now present; -errno on failure.
  */
-static int install_thread_keyring(void)
+int install_thread_keyring(void)
 {
 	struct cred *new;
 	int ret;

@@ -14,6 +14,7 @@
 #include <linux/keyctl.h>
 #include <linux/slab.h>
 #include <net/net_namespace.h>
+#include <linux/init_task.h>
 #include "internal.h"
 #include <keys/request_key_auth-type.h>
 

@@ -112,7 +113,7 @@ static int call_usermodehelper_keys(const char *path, char **argv, char **envp,
  * Request userspace finish the construction of a key
  * - execute "/sbin/request-key <op> <key> <uid> <gid> <keyring> <keyring> <keyring>"
  */
-static int call_sbin_request_key(struct key *authkey, void *aux)
+static int call_sbin_request_key(struct key *authkey)
 {
 	static char const request_key[] = "/sbin/request-key";
 	struct request_key_auth *rka = get_request_key_auth(authkey);

@@ -224,7 +225,6 @@ static int construct_key(struct key *key, const void *callout_info,
 			 size_t callout_len, void *aux,
 			 struct key *dest_keyring)
 {
-	request_key_actor_t actor;
 	struct key *authkey;
 	int ret;
 

@@ -237,11 +237,13 @@ static int construct_key(struct key *key, const void *callout_info,
 		return PTR_ERR(authkey);
 
 	/* Make the call */
-	actor = call_sbin_request_key;
-	if (key->type->request_key)
-		actor = key->type->request_key;
-
-	ret = actor(authkey, aux);
+	if (key->type->request_key) {
+		ret = key->type->request_key(authkey, aux);
+	} else {
+		ret = queue_request_key(key, authkey);
+		if (ret == -ENOPARAM)
+			ret = call_sbin_request_key(authkey);
+	}
 
 	/* check that the actor called complete_request_key() prior to
 	 * returning an error */

@@ -152,6 +152,9 @@ static void request_key_auth_destroy(struct key *key)
 		rcu_assign_keypointer(key, NULL);
 		call_rcu(&rka->rcu, request_key_auth_rcu_disposal);
 	}
+
+	if (key_read_state(rka->target_key) == KEY_IS_UNINSTANTIATED)
+		key_reject_and_link(rka->target_key, 0, -ENOKEY, NULL, NULL);
 }
 
 /*

@@ -0,0 +1,337 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/* Service daemon interface
+ *
+ * Copyright (C) 2021 Red Hat, Inc. All Rights Reserved.
+ * Written by David Howells (dhowells@redhat.com)
+ */
+
+#include <linux/key.h>
+#include <linux/key-type.h>
+#include <linux/utsname.h>
+#include <linux/ipc_namespace.h>
+#include <linux/mnt_namespace.h>
+#include <linux/pid_namespace.h>
+#include <linux/cgroup.h>
+#include <net/net_namespace.h>
+#include "internal.h"
+#include "../fs/mount.h"
+
+/*
+ * Request service filter record.
+ */
+struct request_key_service {
+	struct hlist_node	user_ns_link;	/* Link in the user_ns service list */
+	struct key		*queue_keyring;	/* Keyring into which requests are placed */
+
+	/* The following fields define the selection criteria that we select
+	 * this record on.  All these references must be pinned just in case
+	 * the fd gets passed to another process or the owning process changes
+	 * its own namespaces.
+	 *
+	 * Most of the criteria can be NULL if that criterion is irrelevant to
+	 * the filter.
+	 */
+	char			type[24];	/* Key type of interest (or "") */
+	struct ns_tag		*uts_ns;	/* Matching UTS namespace (or NULL) */
+	struct ns_tag		*ipc_ns;	/* Matching IPC namespace (or NULL) */
+	struct ns_tag		*mnt_ns;	/* Matching mount namespace (or NULL) */
+	struct ns_tag		*pid_ns;	/* Matching process namespace (or NULL) */
+	struct ns_tag		*net_ns;	/* Matching network namespace (or NULL) */
+	struct ns_tag		*cgroup_ns;	/* Matching cgroup namespace (or NULL) */
+	u8			selectivity;	/* Number of exact-match fields */
+	bool			dead;
+};
+
+/*
+ * Free a request_key service record
+ */
+static void free_key_service(struct request_key_service *svc)
+{
+	if (svc) {
+		put_ns_tag(svc->uts_ns);
+		put_ns_tag(svc->ipc_ns);
+		put_ns_tag(svc->mnt_ns);
+		put_ns_tag(svc->pid_ns);
+		put_ns_tag(svc->net_ns);
+		put_ns_tag(svc->cgroup_ns);
+		key_put(svc->queue_keyring);
+		kfree(svc);
+	}
+}
+
+/*
+ * Allocate a service record.
+ */
+static struct request_key_service *alloc_key_service(key_serial_t queue_keyring,
+						     const char __user *type_name,
+						     unsigned int ns_mask)
+{
+	struct request_key_service *svc;
+	struct key_type *type;
+	key_ref_t key_ref;
+	int ret;
+	u8 selectivity = 0;
+
+	svc = kzalloc(sizeof(struct request_key_service), GFP_KERNEL);
+	if (!svc)
+		return ERR_PTR(-ENOMEM);
+
+	if (queue_keyring != 0) {
+		key_ref = lookup_user_key(queue_keyring, 0, KEY_NEED_SEARCH);
+		if (IS_ERR(key_ref)) {
+			ret = PTR_ERR(key_ref);
+			goto err_svc;
+		}
+
+		svc->queue_keyring = key_ref_to_ptr(key_ref);
+	}
+
+	/* Save the matching criteria.  Anything the caller doesn't care about
+	 * we leave as NULL.
+	 */
+	if (type_name) {
+		ret = strncpy_from_user(svc->type, type_name, sizeof(svc->type));
+		if (ret < 0)
+			goto err_keyring;
+		if (ret >= sizeof(svc->type)) {
+			ret = -EINVAL;
+			goto err_keyring;
+		}
+
+		type = key_type_lookup(type_name);
+		if (IS_ERR(type)) {
+			ret = -EINVAL;
+			goto err_keyring;
+		}
+		memcpy(svc->type, type->name, sizeof(svc->type));
+		key_type_put(type);
+	}
+
+	if (ns_mask & KEY_SERVICE_NS_UTS) {
+		svc->uts_ns = get_ns_tag(current->nsproxy->uts_ns->ns.tag);
+		selectivity++;
+	}
+	if (ns_mask & KEY_SERVICE_NS_IPC) {
+		svc->ipc_ns = get_ns_tag(current->nsproxy->ipc_ns->ns.tag);
+		selectivity++;
+	}
+	if (ns_mask & KEY_SERVICE_NS_MNT) {
+		svc->mnt_ns = get_ns_tag(current->nsproxy->mnt_ns->ns.tag);
+		selectivity++;
+	}
+	if (ns_mask & KEY_SERVICE_NS_PID) {
+		svc->pid_ns = get_ns_tag(task_active_pid_ns(current)->ns.tag);
+		selectivity++;
+	}
+	if (ns_mask & KEY_SERVICE_NS_NET) {
+		svc->net_ns = get_ns_tag(current->nsproxy->net_ns->ns.tag);
+		selectivity++;
+	}
+	if (ns_mask & KEY_SERVICE_NS_CGROUP) {
+		svc->cgroup_ns = get_ns_tag(current->nsproxy->cgroup_ns->ns.tag);
+		selectivity++;
+	}
+
+	svc->selectivity = selectivity;
+	return svc;
+
+err_keyring:
+	key_put(svc->queue_keyring);
+err_svc:
+	kfree(svc);
+	return ERR_PTR(ret);
+}
+
+/*
+ * Install a request_key service into the user namespace's list
+ */
+static int install_key_service(struct user_namespace *user_ns,
+			       struct request_key_service *svc)
+{
+	struct request_key_service *p;
+	struct hlist_node **pp;
+	int ret = 0;
+
+	spin_lock(&user_ns->request_key_services_lock);
+
+	/* The services list is kept in order of selectivity.  The more exact
+	 * matches a service requires, the earlier it is in the list.
+	 */
+	for (pp = &user_ns->request_key_services.first; *pp; pp = &(*pp)->next) {
+		p = hlist_entry(*pp, struct request_key_service, user_ns_link);
+		if (p->selectivity < svc->selectivity)
+			goto insert_before;
+		if (p->selectivity > svc->selectivity)
+			continue;
+		if (memcmp(p->type, svc->type, sizeof(p->type)) == 0 &&
+		    p->uts_ns == svc->uts_ns &&
+		    p->ipc_ns == svc->ipc_ns &&
+		    p->mnt_ns == svc->mnt_ns &&
+		    p->pid_ns == svc->pid_ns &&
+		    p->net_ns == svc->net_ns &&
+		    p->cgroup_ns == svc->cgroup_ns)
+			goto duplicate;
+	}
+
+	svc->user_ns_link.pprev = pp;
+	rcu_assign_pointer(*pp, &svc->user_ns_link);
+	goto out;
+
+insert_before:
+	hlist_add_before_rcu(&svc->user_ns_link, &p->user_ns_link);
+	goto out;
+
+duplicate:
+	free_key_service(svc);
+	ret = -EEXIST;
+out:
+	spin_unlock(&user_ns->request_key_services_lock);
+	return ret;
+}
+
+/*
+ * Remove a request_key service interception from the user namespace's list
+ */
+static int remove_key_service(struct user_namespace *user_ns,
+			      struct request_key_service *svc)
+{
+	struct request_key_service *p;
+	struct hlist_node **pp;
+	int ret = 0;
+
+	spin_lock(&user_ns->request_key_services_lock);
+
+	/* The services list is kept in order of selectivity.  The more exact
+	 * matches a service requires, the earlier it is in the list.
+	 */
+	for (pp = &user_ns->request_key_services.first; *pp; pp = &(*pp)->next) {
+		p = hlist_entry(*pp, struct request_key_service, user_ns_link);
+		if (p->selectivity < svc->selectivity)
+			break;
+		if (p->selectivity > svc->selectivity)
+			continue;
+		if (memcmp(p->type, svc->type, sizeof(p->type)) == 0 &&
+		    p->uts_ns == svc->uts_ns &&
+		    p->ipc_ns == svc->ipc_ns &&
+		    p->mnt_ns == svc->mnt_ns &&
+		    p->pid_ns == svc->pid_ns &&
+		    p->net_ns == svc->net_ns &&
+		    p->cgroup_ns == svc->cgroup_ns)
+			goto found;
+	}
+
+	p = NULL;
+	ret = -ENOENT;
+	goto out;
+
+found:
+	hlist_del_rcu(&p->user_ns_link);
+out:
+	spin_unlock(&user_ns->request_key_services_lock);
+	free_key_service(p);
+	free_key_service(svc);
+	return ret;
+}
+
+/*
+ * Add a request_key service handler for a subset of the calling process's
+ * particular set of namespaces.
+ */
+long keyctl_service_intercept(key_serial_t queue_keyring,
+			      int userns_fd,
+			      const char __user *type_name,
+			      unsigned int ns_mask)
+{
+	struct request_key_service *svc;
+	struct user_namespace *user_ns = current_user_ns();
+
+	if (ns_mask & ~KEY_SERVICE___ALL_NS)
+		return -EINVAL;
+	if (userns_fd != -1)
+		return -EINVAL; /* Not supported yet */
+
+	/* Require the caller to be the owner of the user namespace in which
+	 * the fd was created if they're not the sysadmin.  Possibly we should
+	 * be more strict about what namespaces one can select, but it's not
+	 * clear how best to do that.
+	 */
+	if (!capable(CAP_SYS_ADMIN) &&
+	    !uid_eq(user_ns->owner, current_cred()->euid))
+		return -EPERM;
+
+	svc = alloc_key_service(queue_keyring, type_name, ns_mask);
+	if (IS_ERR(svc))
+		return PTR_ERR(svc);
+
+	if (queue_keyring == 0)
+		return remove_key_service(user_ns, svc);
+
+	return install_key_service(user_ns, svc);
+}
+
+/*
+ * Queue a construction record if we can find a queue to punt it off to.
+ */
+int queue_request_key(struct key *key, struct key *auth_key)
+{
+	struct request_key_service *svc;
+	struct user_namespace *user_ns = current_user_ns();
+	struct pid_namespace *pid_ns = task_active_pid_ns(current);
+	struct nsproxy *nsproxy = current->nsproxy;
+	struct key *queue_keyring = NULL;
+	int ret;
+
+	if (hlist_empty(&user_ns->request_key_services))
+		return false;
+
+	rcu_read_lock();
+
+	hlist_for_each_entry_rcu(svc, &user_ns->request_key_services, user_ns_link) {
+		if (svc->type[0] &&
+		    memcmp(svc->type, key->type->name, sizeof(svc->type)) != 0)
+			continue;
+		if ((svc->uts_ns && svc->uts_ns != nsproxy->uts_ns->ns.tag) ||
+		    (svc->ipc_ns && svc->ipc_ns != nsproxy->ipc_ns->ns.tag) ||
+		    (svc->mnt_ns && svc->mnt_ns != nsproxy->mnt_ns->ns.tag) ||
+		    (svc->pid_ns && svc->pid_ns != pid_ns->ns.tag) ||
+		    (svc->net_ns && svc->net_ns != nsproxy->net_ns->ns.tag) ||
+		    (svc->cgroup_ns && svc->cgroup_ns != nsproxy->cgroup_ns->ns.tag))
+			continue;
+		goto found_match;
+	}
+
+	rcu_read_unlock();
+	return -ENOPARAM;
+
+found_match:
+	spin_lock(&user_ns->request_key_services_lock);
+	if (!svc->dead)
+		queue_keyring = key_get(svc->queue_keyring);
+	spin_unlock(&user_ns->request_key_services_lock);
+	rcu_read_unlock();
+
+	ret = -ENOPARAM;
+	if (queue_keyring) {
+		ret = key_link(queue_keyring, auth_key);
+		if (ret < 0)
+			key_reject_and_link(key, 0, ret, NULL, auth_key);
+		key_put(queue_keyring);
+	}
+
+	return ret;
+}
+
+/*
+ * Clear all the service intercept records on a user namespace.
+ */
+void clear_request_key_services(struct user_namespace *user_ns)
+{
+	struct request_key_service *svc;
+
+	while (!hlist_empty(&user_ns->request_key_services)) {
+		svc = hlist_entry(user_ns->request_key_services.first,
+				  struct request_key_service, user_ns_link);
+		hlist_del(&svc->user_ns_link);
+		free_key_service(svc);
+	}
+}

@@ -3238,10 +3238,9 @@ static void dec_mnt_namespaces(struct ucounts *ucounts)
 
 static void free_mnt_ns(struct mnt_namespace *ns)
 {
-	if (!is_anon_ns(ns))
-		ns_free_inum(&ns->ns);
 	dec_mnt_namespaces(ns->ucounts);
 	put_user_ns(ns->user_ns);
+	destroy_ns_common(&ns->ns);
 	kfree(ns);
 }
 

@@ -3269,18 +3268,17 @@ static struct mnt_namespace *alloc_mnt_ns(struct user_namespace *user_ns, bool a
 		dec_mnt_namespaces(ucounts);
 		return ERR_PTR(-ENOMEM);
 	}
-	if (!anon) {
-		ret = ns_alloc_inum(&new_ns->ns);
-		if (ret) {
-			kfree(new_ns);
-			dec_mnt_namespaces(ucounts);
-			return ERR_PTR(ret);
-		}
+
+	ret = init_ns_common(&new_ns->ns, anon);
+	if (ret) {
+		destroy_ns_common(&new_ns->ns);
+		kfree(new_ns);
+		dec_mnt_namespaces(ucounts);
+		return ERR_PTR(ret);
 	}
 	new_ns->ns.ops = &mntns_operations;
 	if (!anon)
 		new_ns->seq = atomic64_add_return(1, &mnt_ns_seq);
-	refcount_set(&new_ns->ns.count, 1);
 	INIT_LIST_HEAD(&new_ns->list);
 	init_waitqueue_head(&new_ns->poll);
 	spin_lock_init(&new_ns->ns_lock);

@@ -3,14 +3,37 @@
 #define _LINUX_NS_COMMON_H
 
 #include <linux/refcount.h>
+#include <linux/slab.h>
 
 struct proc_ns_operations;
 
+/*
+ * Comparable tag for namespaces so that namespaces don't have to be pinned by
+ * something that wishes to detect if a namespace matches a criterion.
+ */
+struct ns_tag {
+	refcount_t	usage;
+};
+
 struct ns_common {
 	atomic_long_t stashed;
 	const struct proc_ns_operations *ops;
+	struct ns_tag *tag;
 	unsigned int inum;
 	refcount_t count;
 };
 
+static inline struct ns_tag *get_ns_tag(struct ns_tag *tag)
+{
+	if (tag)
+		refcount_inc(&tag->usage);
+	return tag;
+}
+
+static inline void put_ns_tag(struct ns_tag *tag)
+{
+	if (tag && refcount_dec_and_test(&tag->usage))
+		kfree(tag);
+}
+
 #endif

@@ -64,13 +64,45 @@ static inline void proc_free_inum(unsigned int inum) {}
 
 #endif /* CONFIG_PROC_FS */
 
-static inline int ns_alloc_inum(struct ns_common *ns)
+/**
+ * init_ns_common - Initialise the common part of a namespace
+ * @ns: The namespace to initialise
+ * @anon: The namespace will be anonymous
+ *
+ * Set up the common part of a namespace, assigning an inode number and
+ * creating a tag.  Returns 0 on success and a negative error code on failure.
+ * On failure, the caller must call destroy_ns_common().
+ */
+static inline int init_ns_common(struct ns_common *ns, bool anon)
 {
+	struct ns_tag *tag;
+
+	tag = kzalloc(sizeof(*tag), GFP_KERNEL);
+	if (!tag)
+		return -ENOMEM;
+
+	refcount_set(&tag->usage, 1);
+	ns->tag = tag;
+	ns->inum = 0;
 	atomic_long_set(&ns->stashed, 0);
-	return proc_alloc_inum(&ns->inum);
+	refcount_set(&ns->count, 1);
+
+	return anon ? 0 : proc_alloc_inum(&ns->inum);
 }
 
-#define ns_free_inum(ns) proc_free_inum((ns)->inum)
+/**
+ * destroy_ns_common - Clean up the common part of a namespace
+ * @ns: The namespace to clean up
+ */
+static inline void destroy_ns_common(struct ns_common *ns)
+{
+	put_ns_tag(ns->tag);
+	ns->tag = NULL;
+	if (ns->inum) {
+		proc_free_inum(ns->inum);
+		ns->inum = 0;
+	}
+}
 
 extern struct file *proc_ns_fget(int fd);
 #define get_proc_ns(inode) ((struct ns_common *)(inode)->i_private)

@@ -24,8 +24,15 @@ extern int version_string(LINUX_VERSION_CODE);
 int version_string(LINUX_VERSION_CODE);
 #endif
 
+static struct ns_tag init_uts_ns_tag = {
+	.usage		= REFCOUNT_INIT(1),
+};
+
 struct uts_namespace init_uts_ns = {
-	.ns.count = REFCOUNT_INIT(2),
+	.ns = {
+		.count		= REFCOUNT_INIT(2),
+		.tag		= &init_uts_ns_tag,
+	},
 	.name = {
 		.sysname	= UTS_SYSNAME,
 		.nodename	= UTS_NODENAME,

@@ -20,13 +20,18 @@
 
 DEFINE_SPINLOCK(mq_lock);
 
+static struct ns_tag init_ipc_ns_tag = {
+	.usage		= REFCOUNT_INIT(1),
+};
+
 /*
  * The next 2 defines are here bc this is the only file
  * compiled when either CONFIG_SYSVIPC and CONFIG_POSIX_MQUEUE
  * and not CONFIG_IPC_NS.
  */
 struct ipc_namespace init_ipc_ns = {
-	.ns.count = REFCOUNT_INIT(1),
+	.ns.tag	= &init_ipc_ns_tag,
+	.ns.count = REFCOUNT_INIT(2),
 	.user_ns = &init_user_ns,
 	.ns.inum = PROC_IPC_INIT_INO,
 #ifdef CONFIG_IPC_NS

@@ -46,12 +46,10 @@ static struct ipc_namespace *create_ipc_ns(struct user_namespace *user_ns,
 	if (ns == NULL)
 		goto fail_dec;
 
-	err = ns_alloc_inum(&ns->ns);
+	err = init_ns_common(&ns->ns, false);
 	if (err)
 		goto fail_free;
 	ns->ns.ops = &ipcns_operations;
-
-	refcount_set(&ns->ns.count, 1);
 	ns->user_ns = get_user_ns(user_ns);
 	ns->ucounts = ucounts;
 

@@ -67,8 +65,8 @@ static struct ipc_namespace *create_ipc_ns(struct user_namespace *user_ns,
 
 fail_put:
 	put_user_ns(ns->user_ns);
-	ns_free_inum(&ns->ns);
 fail_free:
+	destroy_ns_common(&ns->ns);
 	kfree(ns);
 fail_dec:
 	dec_ipc_namespaces(ucounts);

@@ -127,7 +125,7 @@ static void free_ipc_ns(struct ipc_namespace *ns)
 
 	dec_ipc_namespaces(ns->ucounts);
 	put_user_ns(ns->user_ns);
-	ns_free_inum(&ns->ns);
+	destroy_ns_common(&ns->ns);
 	kfree(ns);
 }
 

@@ -197,8 +197,13 @@ static u16 have_exit_callback __read_mostly;
 static u16 have_release_callback __read_mostly;
 static u16 have_canfork_callback __read_mostly;
 
+static struct ns_tag init_cgroup_ns_tag = {
+	.usage		= REFCOUNT_INIT(1),
+};
+
 /* cgroup namespace for init task */
 struct cgroup_namespace init_cgroup_ns = {
+	.ns.tag		= &init_cgroup_ns_tag,
 	.ns.count	= REFCOUNT_INIT(2),
 	.user_ns	= &init_user_ns,
 	.ns.ops		= &cgroupns_operations,

@@ -27,12 +27,12 @@ static struct cgroup_namespace *alloc_cgroup_ns(void)
 	new_ns = kzalloc(sizeof(struct cgroup_namespace), GFP_KERNEL);
 	if (!new_ns)
 		return ERR_PTR(-ENOMEM);
-	ret = ns_alloc_inum(&new_ns->ns);
+	ret = init_ns_common(&new_ns->ns, false);
 	if (ret) {
+		destroy_ns_common(&new_ns->ns);
 		kfree(new_ns);
 		return ERR_PTR(ret);
 	}
-	refcount_set(&new_ns->ns.count, 1);
 	new_ns->ns.ops = &cgroupns_operations;
 	return new_ns;
 }

@@ -42,7 +42,7 @@ void free_cgroup_ns(struct cgroup_namespace *ns)
 	put_css_set(ns->root_cset);
 	dec_cgroup_namespaces(ns->ucounts);
 	put_user_ns(ns->user_ns);
-	ns_free_inum(&ns->ns);
+	destroy_ns_common(&ns->ns);
 	kfree(ns);
 }
 EXPORT_SYMBOL(free_cgroup_ns);

@@ -66,6 +66,10 @@ int pid_max = PID_MAX_DEFAULT;
 int pid_max_min = RESERVED_PIDS + 1;
 int pid_max_max = PID_MAX_LIMIT;
 
+static struct ns_tag init_pid_ns_tag = {
+	.usage		= REFCOUNT_INIT(1),
+};
+
 /*
  * PID-map pages start out as NULL, they get allocated upon
  * first use and are never deallocated. This way a low pid_max

@@ -73,6 +77,7 @@ int pid_max_max = PID_MAX_LIMIT;
  * the scheme scales to up to 4 million PIDs, runtime.
  */
 struct pid_namespace init_pid_ns = {
+	.ns.tag = &init_pid_ns_tag,
 	.ns.count = REFCOUNT_INIT(2),
 	.idr = IDR_INIT(init_pid_ns.idr),
 	.pid_allocated = PIDNS_ADDING,

@@ -93,16 +93,15 @@ static struct pid_namespace *create_pid_namespace(struct user_namespace *user_ns
 
 	idr_init(&ns->idr);
 
-	ns->pid_cachep = create_pid_cachep(level);
-	if (ns->pid_cachep == NULL)
-		goto out_free_idr;
-
-	err = ns_alloc_inum(&ns->ns);
+	err = init_ns_common(&ns->ns, false);
 	if (err)
-		goto out_free_idr;
+		goto out_free;
 	ns->ns.ops = &pidns_operations;
 
-	refcount_set(&ns->ns.count, 1);
+	ns->pid_cachep = create_pid_cachep(level);
+	if (ns->pid_cachep == NULL)
+		goto out_free;
+
 	ns->level = level;
 	ns->parent = get_pid_ns(parent_pid_ns);
 	ns->user_ns = get_user_ns(user_ns);

@@ -111,8 +110,9 @@ static struct pid_namespace *create_pid_namespace(struct user_namespace *user_ns
 
 	return ns;
 
-out_free_idr:
+out_free:
 	idr_destroy(&ns->idr);
+	destroy_ns_common(&ns->ns);
 	kmem_cache_free(pid_ns_cachep, ns);
 out_dec:
 	dec_pid_namespaces(ucounts);

@@ -132,7 +132,7 @@ static void delayed_free_pidns(struct rcu_head *p)
 
 static void destroy_pid_namespace(struct pid_namespace *ns)
 {
-	ns_free_inum(&ns->ns);
+	destroy_ns_common(&ns->ns);
 
 	idr_destroy(&ns->idr);
 	call_rcu(&ns->rcu, delayed_free_pidns);

@@ -92,16 +92,14 @@ static struct time_namespace *clone_time_ns(struct user_namespace *user_ns,
 	if (!ns)
 		goto fail_dec;
 
-	refcount_set(&ns->ns.count, 1);
+	err = init_ns_common(&ns->ns, false);
+	if (err)
+		goto fail_free;
 
 	ns->vvar_page = alloc_page(GFP_KERNEL | __GFP_ZERO);
 	if (!ns->vvar_page)
 		goto fail_free;
 
-	err = ns_alloc_inum(&ns->ns);
-	if (err)
-		goto fail_free_page;
-
 	ns->ucounts = ucounts;
 	ns->ns.ops = &timens_operations;
 	ns->user_ns = get_user_ns(user_ns);

@@ -109,9 +107,8 @@ static struct time_namespace *clone_time_ns(struct user_namespace *user_ns,
 	ns->frozen_offsets = false;
 	return ns;
 
-fail_free_page:
-	__free_page(ns->vvar_page);
 fail_free:
+	destroy_ns_common(&ns->ns);
 	kfree(ns);
 fail_dec:
 	dec_time_namespaces(ucounts);

@@ -230,7 +227,7 @@ void free_time_ns(struct time_namespace *ns)
 {
 	dec_time_namespaces(ns->ucounts);
 	put_user_ns(ns->user_ns);
-	ns_free_inum(&ns->ns);
+	destroy_ns_common(&ns->ns);
 	__free_page(ns->vvar_page);
 	kfree(ns);
 }

@@ -20,6 +20,10 @@
 #include <linux/user_namespace.h>
 #include <linux/proc_ns.h>
 
+static struct ns_tag init_user_ns_tag = {
+	.usage	= REFCOUNT_INIT(1),
+};
+
 /*
  * userns count is 1 for root user, 1 for init_uts_ns,
  * and 1 for... ?

@@ -55,6 +59,7 @@ struct user_namespace init_user_ns = {
 			},
 		},
 	},
+	.ns.tag	= &init_user_ns_tag,
 	.ns.count = REFCOUNT_INIT(3),
 	.owner = GLOBAL_ROOT_UID,
 	.group = GLOBAL_ROOT_GID,

@@ -106,12 +106,11 @@ int create_user_ns(struct cred *new)
 	if (!ns)
 		goto fail_dec;
 
-	ret = ns_alloc_inum(&ns->ns);
+	ret = init_ns_common(&ns->ns, false);
 	if (ret)
 		goto fail_free;
 	ns->ns.ops = &userns_operations;
 
-	refcount_set(&ns->ns.count, 1);
 	/* Leave the new->user_ns reference with the new user namespace. */
 	ns->parent = parent_ns;
 	ns->level = parent_ns->level + 1;

@@ -142,8 +141,8 @@ int create_user_ns(struct cred *new)
 #ifdef CONFIG_PERSISTENT_KEYRINGS
 	key_put(ns->persistent_keyring_register);
 #endif
-	ns_free_inum(&ns->ns);
 fail_free:
+	destroy_ns_common(&ns->ns);
 	kmem_cache_free(user_ns_cachep, ns);
 fail_dec:
 	dec_user_namespaces(ucounts);

@@ -193,7 +192,7 @@ static void free_user_ns(struct work_struct *work)
 		}
 		retire_userns_sysctls(ns);
 		key_free_user_ns(ns);
-		ns_free_inum(&ns->ns);
+		destroy_ns_common(&ns->ns);
 		kmem_cache_free(user_ns_cachep, ns);
 		dec_user_namespaces(ucounts);
 		ns = parent;

@@ -30,10 +30,17 @@ static void dec_uts_namespaces(struct ucounts *ucounts)
 static struct uts_namespace *create_uts_ns(void)
 {
 	struct uts_namespace *uts_ns;
+	int err;
 
 	uts_ns = kmem_cache_alloc(uts_ns_cache, GFP_KERNEL);
-	if (uts_ns)
-		refcount_set(&uts_ns->ns.count, 1);
+	if (uts_ns) {
+		err = init_ns_common(&uts_ns->ns, false);
+		if (err < 0) {
+			destroy_ns_common(&uts_ns->ns);
+			kmem_cache_free(uts_ns_cache, uts_ns);
+			return ERR_PTR(err);
+		}
+	}
 	return uts_ns;
 }
 

@@ -54,14 +61,11 @@ static struct uts_namespace *clone_uts_ns(struct user_namespace *user_ns,
 	if (!ucounts)
 		goto fail;
 
-	err = -ENOMEM;
 	ns = create_uts_ns();
-	if (!ns)
+	if (IS_ERR(ns)) {
+		err = PTR_ERR(ns);
 		goto fail_dec;
-
-	err = ns_alloc_inum(&ns->ns);
-	if (err)
-		goto fail_free;
+	}
 
 	ns->ucounts = ucounts;
 	ns->ns.ops = &utsns_operations;

@@ -72,8 +76,6 @@ static struct uts_namespace *clone_uts_ns(struct user_namespace *user_ns,
 	up_read(&uts_sem);
 	return ns;
 
-fail_free:
-	kmem_cache_free(uts_ns_cache, ns);
 fail_dec:
 	dec_uts_namespaces(ucounts);
 fail:

@@ -107,7 +109,7 @@ void free_uts_ns(struct uts_namespace *ns)
 {
 	dec_uts_namespaces(ns->ucounts);
 	put_user_ns(ns->user_ns);
-	ns_free_inum(&ns->ns);
+	destroy_ns_common(&ns->ns);
 	kmem_cache_free(uts_ns_cache, ns);
 }
 

@@ -44,8 +44,14 @@ EXPORT_SYMBOL_GPL(net_rwsem);
 static struct key_tag init_net_key_domain = { .usage = REFCOUNT_INIT(1) };
 #endif
 
+static struct ns_tag init_net_tag = {
+	.usage		= REFCOUNT_INIT(1),
+};
+
 struct net init_net = {
-	.ns.count	= REFCOUNT_INIT(1),
+	.ns.tag		= &init_net_tag,
+	.ns.count	= REFCOUNT_INIT(2),
+	.ns.ops		= &netns_operations,
 	.dev_base_head	= LIST_HEAD_INIT(init_net.dev_base_head),
 #ifdef CONFIG_KEYS
 	.key_domain	= &init_net_key_domain,

@@ -329,7 +335,6 @@ static __net_init int setup_net(struct net *net, struct user_namespace *user_ns)
 	int error = 0;
 	LIST_HEAD(net_exit_list);
 
-	refcount_set(&net->ns.count, 1);
 	refcount_set(&net->passive, 1);
 	get_random_bytes(&net->hash_mix, sizeof(u32));
 	net->dev_base_seq = 1;

@@ -419,6 +424,10 @@ static struct net *net_alloc(void)
 	if (!net)
 		goto out_free;
 
+	if (init_ns_common(&net->ns, false) < 0)
+		goto out_free_2;
+	net->ns.ops = &netns_operations;
+
 #ifdef CONFIG_KEYS
 	net->key_domain = kzalloc(sizeof(struct key_tag), GFP_KERNEL);
 	if (!net->key_domain)

@@ -432,6 +441,7 @@ static struct net *net_alloc(void)
 
 #ifdef CONFIG_KEYS
 out_free_2:
+	destroy_ns_common(&net->ns);
 	kmem_cache_free(net_cachep, net);
 	net = NULL;
 #endif

@@ -443,6 +453,7 @@ static struct net *net_alloc(void)
 static void net_free(struct net *net)
 {
 	kfree(rcu_access_pointer(net->gen));
+	destroy_ns_common(&net->ns);
 	kmem_cache_free(net_cachep, net);
 }
 

@@ -700,24 +711,6 @@ struct net *get_net_ns_by_pid(pid_t pid)
 }
 EXPORT_SYMBOL_GPL(get_net_ns_by_pid);
 
-static __net_init int net_ns_net_init(struct net *net)
-{
-#ifdef CONFIG_NET_NS
-	net->ns.ops = &netns_operations;
-#endif
-	return ns_alloc_inum(&net->ns);
-}
-
-static __net_exit void net_ns_net_exit(struct net *net)
-{
-	ns_free_inum(&net->ns);
-}
-
-static struct pernet_operations __net_initdata net_ns_ops = {
-	.init = net_ns_net_init,
-	.exit = net_ns_net_exit,
-};
-
 static const struct nla_policy rtnl_net_policy[NETNSA_MAX + 1] = {
 	[NETNSA_NONE]		= { .type = NLA_UNSPEC },
 	[NETNSA_NSID]		= { .type = NLA_S32 },

@@ -1097,6 +1090,8 @@ static int __init net_ns_init(void)
 		panic("Could not create netns workq");
 #endif
 
+	proc_alloc_inum(&init_net.ns.inum);
+
 	ng = net_alloc_generic();
 	if (!ng)
 		panic("Could not allocate generic netns");

@@ -1114,9 +1109,6 @@ static int __init net_ns_init(void)
 	init_net_initialized = true;
 	up_write(&pernet_ops_rwsem);
 
-	if (register_pernet_subsys(&net_ns_ops))
-		panic("Could not register network namespace subsystems");
-
 	rtnl_register(PF_UNSPEC, RTM_NEWNSID, rtnl_net_newid, NULL,
 		      RTNL_FLAG_DOIT_UNLOCKED);
 	rtnl_register(PF_UNSPEC, RTM_GETNSID, rtnl_net_getid, rtnl_net_dumpid,