[mirror_ubuntu-artful-kernel.git] / net / core / net_namespace.c

#include <linux/workqueue.h>
#include <linux/rtnetlink.h>
#include <linux/cache.h>
#include <linux/slab.h>
#include <linux/list.h>
#include <linux/delay.h>
#include <linux/sched.h>
#include <linux/idr.h>
#include <linux/rculist.h>
#include <linux/nsproxy.h>
#include <net/net_namespace.h>
#include <net/netns/generic.h>

/*
 *	Our network namespace constructor/destructor lists
 */

static LIST_HEAD(pernet_list);
static struct list_head *first_device = &pernet_list;
static DEFINE_MUTEX(net_mutex);

LIST_HEAD(net_namespace_list);
EXPORT_SYMBOL_GPL(net_namespace_list);

struct net init_net;
EXPORT_SYMBOL(init_net);

#define INITIAL_NET_GEN_PTRS	13 /* +1 for len +2 for rcu_head */

static int ops_init(const struct pernet_operations *ops, struct net *net)
{
	int err;
	if (ops->id && ops->size) {
		void *data = kzalloc(ops->size, GFP_KERNEL);
		if (!data)
			return -ENOMEM;

		err = net_assign_generic(net, *ops->id, data);
		if (err) {
			kfree(data);
			return err;
		}
	}
	if (ops->init)
		return ops->init(net);
	return 0;
}

static void ops_free(const struct pernet_operations *ops, struct net *net)
{
	if (ops->id && ops->size) {
		int id = *ops->id;
		kfree(net_generic(net, id));
	}
}

static void ops_exit_list(const struct pernet_operations *ops,
			  struct list_head *net_exit_list)
{
	struct net *net;
	if (ops->exit) {
		list_for_each_entry(net, net_exit_list, exit_list)
			ops->exit(net);
	}
	if (ops->exit_batch)
		ops->exit_batch(net_exit_list);
}

static void ops_free_list(const struct pernet_operations *ops,
			  struct list_head *net_exit_list)
{
	struct net *net;
	if (ops->size && ops->id) {
		list_for_each_entry(net, net_exit_list, exit_list)
			ops_free(ops, net);
	}
}

/*
 * setup_net runs the initializers for the network namespace object.
 */
static __net_init int setup_net(struct net *net)
{
	/* Must be called with net_mutex held */
	const struct pernet_operations *ops, *saved_ops;
	int error = 0;
	LIST_HEAD(net_exit_list);

	atomic_set(&net->count, 1);

#ifdef NETNS_REFCNT_DEBUG
	atomic_set(&net->use_count, 0);
#endif

	list_for_each_entry(ops, &pernet_list, list) {
		error = ops_init(ops, net);
		if (error < 0)
			goto out_undo;
	}
out:
	return error;

out_undo:
	/* Walk through the list backwards calling the exit functions
	 * for the pernet modules whose init functions did not fail.
	 */
	list_add(&net->exit_list, &net_exit_list);
	saved_ops = ops;
	list_for_each_entry_continue_reverse(ops, &pernet_list, list)
		ops_exit_list(ops, &net_exit_list);

	ops = saved_ops;
	list_for_each_entry_continue_reverse(ops, &pernet_list, list)
		ops_free_list(ops, &net_exit_list);

	rcu_barrier();
	goto out;
}

static struct net_generic *net_alloc_generic(void)
{
	struct net_generic *ng;
	size_t generic_size = sizeof(struct net_generic) +
		INITIAL_NET_GEN_PTRS * sizeof(void *);

	ng = kzalloc(generic_size, GFP_KERNEL);
	if (ng)
		ng->len = INITIAL_NET_GEN_PTRS;

	return ng;
}

#ifdef CONFIG_NET_NS
static struct kmem_cache *net_cachep;
static struct workqueue_struct *netns_wq;

static struct net *net_alloc(void)
{
	struct net *net = NULL;
	struct net_generic *ng;

	ng = net_alloc_generic();
	if (!ng)
		goto out;

	net = kmem_cache_zalloc(net_cachep, GFP_KERNEL);
	if (!net)
		goto out_free;

	rcu_assign_pointer(net->gen, ng);
out:
	return net;

out_free:
	kfree(ng);
	goto out;
}

static void net_free(struct net *net)
{
#ifdef NETNS_REFCNT_DEBUG
	if (unlikely(atomic_read(&net->use_count) != 0)) {
		printk(KERN_EMERG "network namespace not free! Usage: %d\n",
			atomic_read(&net->use_count));
		return;
	}
#endif
	kfree(net->gen);
	kmem_cache_free(net_cachep, net);
}

static struct net *net_create(void)
{
	struct net *net;
	int rv;

	net = net_alloc();
	if (!net)
		return ERR_PTR(-ENOMEM);
	mutex_lock(&net_mutex);
	rv = setup_net(net);
	if (rv == 0) {
		rtnl_lock();
		list_add_tail_rcu(&net->list, &net_namespace_list);
		rtnl_unlock();
	}
	mutex_unlock(&net_mutex);
	if (rv < 0) {
		net_free(net);
		return ERR_PTR(rv);
	}
	return net;
}

struct net *copy_net_ns(unsigned long flags, struct net *old_net)
{
	if (!(flags & CLONE_NEWNET))
		return get_net(old_net);
	return net_create();
}

static DEFINE_SPINLOCK(cleanup_list_lock);
static LIST_HEAD(cleanup_list);  /* Must hold cleanup_list_lock to touch */

static void cleanup_net(struct work_struct *work)
{
	const struct pernet_operations *ops;
	struct net *net, *tmp;
	LIST_HEAD(net_kill_list);
	LIST_HEAD(net_exit_list);

	/* Atomically snapshot the list of namespaces to cleanup */
	spin_lock_irq(&cleanup_list_lock);
	list_replace_init(&cleanup_list, &net_kill_list);
	spin_unlock_irq(&cleanup_list_lock);

	mutex_lock(&net_mutex);

	/* Don't let anyone else find us. */
	rtnl_lock();
	list_for_each_entry(net, &net_kill_list, cleanup_list) {
		list_del_rcu(&net->list);
		list_add_tail(&net->exit_list, &net_exit_list);
	}
	rtnl_unlock();

	/*
	 * Another CPU might be rcu-iterating the list, wait for it.
	 * This needs to be before calling the exit() notifiers, so
	 * the rcu_barrier() below isn't sufficient alone.
	 */
	synchronize_rcu();

	/* Run all of the network namespace exit methods */
	list_for_each_entry_reverse(ops, &pernet_list, list)
		ops_exit_list(ops, &net_exit_list);

	/* Free the net generic variables */
	list_for_each_entry_reverse(ops, &pernet_list, list)
		ops_free_list(ops, &net_exit_list);

	mutex_unlock(&net_mutex);

	/* Ensure there are no outstanding rcu callbacks using this
	 * network namespace.
	 */
	rcu_barrier();

	/* Finally it is safe to free my network namespace structure */
	list_for_each_entry_safe(net, tmp, &net_exit_list, exit_list) {
		list_del_init(&net->exit_list);
		net_free(net);
	}
}
static DECLARE_WORK(net_cleanup_work, cleanup_net);

void __put_net(struct net *net)
{
	/* Cleanup the network namespace in process context */
	unsigned long flags;

	spin_lock_irqsave(&cleanup_list_lock, flags);
	list_add(&net->cleanup_list, &cleanup_list);
	spin_unlock_irqrestore(&cleanup_list_lock, flags);

	queue_work(netns_wq, &net_cleanup_work);
}
EXPORT_SYMBOL_GPL(__put_net);

#else
struct net *copy_net_ns(unsigned long flags, struct net *old_net)
{
	if (flags & CLONE_NEWNET)
		return ERR_PTR(-EINVAL);
	return old_net;
}
#endif

struct net *get_net_ns_by_pid(pid_t pid)
{
	struct task_struct *tsk;
	struct net *net;

	/* Lookup the network namespace */
	net = ERR_PTR(-ESRCH);
	rcu_read_lock();
	tsk = find_task_by_vpid(pid);
	if (tsk) {
		struct nsproxy *nsproxy;
		nsproxy = task_nsproxy(tsk);
		if (nsproxy)
			net = get_net(nsproxy->net_ns);
	}
	rcu_read_unlock();
	return net;
}
EXPORT_SYMBOL_GPL(get_net_ns_by_pid);

static int __init net_ns_init(void)
{
	struct net_generic *ng;

#ifdef CONFIG_NET_NS
	net_cachep = kmem_cache_create("net_namespace", sizeof(struct net),
					SMP_CACHE_BYTES,
					SLAB_PANIC, NULL);

	/* Create workqueue for cleanup */
	netns_wq = create_singlethread_workqueue("netns");
	if (!netns_wq)
		panic("Could not create netns workq");
#endif

	ng = net_alloc_generic();
	if (!ng)
		panic("Could not allocate generic netns");

	rcu_assign_pointer(init_net.gen, ng);

	mutex_lock(&net_mutex);
	if (setup_net(&init_net))
		panic("Could not setup the initial network namespace");

	rtnl_lock();
	list_add_tail_rcu(&init_net.list, &net_namespace_list);
	rtnl_unlock();

	mutex_unlock(&net_mutex);

	return 0;
}

pure_initcall(net_ns_init);

#ifdef CONFIG_NET_NS
static int __register_pernet_operations(struct list_head *list,
					struct pernet_operations *ops)
{
	struct net *net;
	int error;
	LIST_HEAD(net_exit_list);

	list_add_tail(&ops->list, list);
	if (ops->init || (ops->id && ops->size)) {
		for_each_net(net) {
			error = ops_init(ops, net);
			if (error)
				goto out_undo;
			list_add_tail(&net->exit_list, &net_exit_list);
		}
	}
	return 0;

out_undo:
	/* If I have an error cleanup all namespaces I initialized */
	list_del(&ops->list);
	ops_exit_list(ops, &net_exit_list);
	ops_free_list(ops, &net_exit_list);
	return error;
}

static void __unregister_pernet_operations(struct pernet_operations *ops)
{
	struct net *net;
	LIST_HEAD(net_exit_list);

	list_del(&ops->list);
	for_each_net(net)
		list_add_tail(&net->exit_list, &net_exit_list);
	ops_exit_list(ops, &net_exit_list);
	ops_free_list(ops, &net_exit_list);
}

#else

static int __register_pernet_operations(struct list_head *list,
					struct pernet_operations *ops)
{
	int err = 0;
	err = ops_init(ops, &init_net);
	if (err)
		ops_free(ops, &init_net);
	return err;
	
}

static void __unregister_pernet_operations(struct pernet_operations *ops)
{
	LIST_HEAD(net_exit_list);
	list_add(&init_net.exit_list, &net_exit_list);
	ops_exit_list(ops, &net_exit_list);
	ops_free_list(ops, &net_exit_list);
}

#endif /* CONFIG_NET_NS */

static DEFINE_IDA(net_generic_ids);

static int register_pernet_operations(struct list_head *list,
				      struct pernet_operations *ops)
{
	int error;

	if (ops->id) {
again:
		error = ida_get_new_above(&net_generic_ids, 1, ops->id);
		if (error < 0) {
			if (error == -EAGAIN) {
				ida_pre_get(&net_generic_ids, GFP_KERNEL);
				goto again;
			}
			return error;
		}
	}
	error = __register_pernet_operations(list, ops);
	if (error && ops->id)
		ida_remove(&net_generic_ids, *ops->id);

	return error;
}

static void unregister_pernet_operations(struct pernet_operations *ops)
{
	
	__unregister_pernet_operations(ops);
	if (ops->id)
		ida_remove(&net_generic_ids, *ops->id);
}

/**
 *      register_pernet_subsys - register a network namespace subsystem
 *	@ops:  pernet operations structure for the subsystem
 *
 *	Register a subsystem which has init and exit functions
 *	that are called when network namespaces are created and
 *	destroyed respectively.
 *
 *	When registered all network namespace init functions are
 *	called for every existing network namespace.  Allowing kernel
 *	modules to have a race free view of the set of network namespaces.
 *
 *	When a new network namespace is created all of the init
 *	methods are called in the order in which they were registered.
 *
 *	When a network namespace is destroyed all of the exit methods
 *	are called in the reverse of the order with which they were
 *	registered.
 */
int register_pernet_subsys(struct pernet_operations *ops)
{
	int error;
	mutex_lock(&net_mutex);
	error =  register_pernet_operations(first_device, ops);
	mutex_unlock(&net_mutex);
	return error;
}
EXPORT_SYMBOL_GPL(register_pernet_subsys);

/**
 *      unregister_pernet_subsys - unregister a network namespace subsystem
 *	@ops: pernet operations structure to manipulate
 *
 *	Remove the pernet operations structure from the list to be
 *	used when network namespaces are created or destroyed.  In
 *	addition run the exit method for all existing network
 *	namespaces.
 */
void unregister_pernet_subsys(struct pernet_operations *module)
{
	mutex_lock(&net_mutex);
	unregister_pernet_operations(module);
	mutex_unlock(&net_mutex);
}
EXPORT_SYMBOL_GPL(unregister_pernet_subsys);

/**
 *      register_pernet_device - register a network namespace device
 *	@ops:  pernet operations structure for the subsystem
 *
 *	Register a device which has init and exit functions
 *	that are called when network namespaces are created and
 *	destroyed respectively.
 *
 *	When registered all network namespace init functions are
 *	called for every existing network namespace.  Allowing kernel
 *	modules to have a race free view of the set of network namespaces.
 *
 *	When a new network namespace is created all of the init
 *	methods are called in the order in which they were registered.
 *
 *	When a network namespace is destroyed all of the exit methods
 *	are called in the reverse of the order with which they were
 *	registered.
 */
int register_pernet_device(struct pernet_operations *ops)
{
	int error;
	mutex_lock(&net_mutex);
	error = register_pernet_operations(&pernet_list, ops);
	if (!error && (first_device == &pernet_list))
		first_device = &ops->list;
	mutex_unlock(&net_mutex);
	return error;
}
EXPORT_SYMBOL_GPL(register_pernet_device);

/**
 *      unregister_pernet_device - unregister a network namespace netdevice
 *	@ops: pernet operations structure to manipulate
 *
 *	Remove the pernet operations structure from the list to be
 *	used when network namespaces are created or destroyed.  In
 *	addition run the exit method for all existing network
 *	namespaces.
 */
void unregister_pernet_device(struct pernet_operations *ops)
{
	mutex_lock(&net_mutex);
	if (&ops->list == first_device)
		first_device = first_device->next;
	unregister_pernet_operations(ops);
	mutex_unlock(&net_mutex);
}
EXPORT_SYMBOL_GPL(unregister_pernet_device);

static void net_generic_release(struct rcu_head *rcu)
{
	struct net_generic *ng;

	ng = container_of(rcu, struct net_generic, rcu);
	kfree(ng);
}

int net_assign_generic(struct net *net, int id, void *data)
{
	struct net_generic *ng, *old_ng;

	BUG_ON(!mutex_is_locked(&net_mutex));
	BUG_ON(id == 0);

	ng = old_ng = net->gen;
	if (old_ng->len >= id)
		goto assign;

	ng = kzalloc(sizeof(struct net_generic) +
			id * sizeof(void *), GFP_KERNEL);
	if (ng == NULL)
		return -ENOMEM;

	/*
	 * Some synchronisation notes:
	 *
	 * The net_generic explores the net->gen array inside rcu
	 * read section. Besides once set the net->gen->ptr[x]
	 * pointer never changes (see rules in netns/generic.h).
	 *
	 * That said, we simply duplicate this array and schedule
	 * the old copy for kfree after a grace period.
	 */

	ng->len = id;
	memcpy(&ng->ptr, &old_ng->ptr, old_ng->len * sizeof(void*));

	rcu_assign_pointer(net->gen, ng);
	call_rcu(&old_ng->rcu, net_generic_release);
assign:
	ng->ptr[id - 1] = data;
	return 0;
}
EXPORT_SYMBOL_GPL(net_assign_generic);
Commit	Line	Data
5f256bec EB	1	#include <linux/workqueue.h>
	2	#include <linux/rtnetlink.h>
	3	#include <linux/cache.h>
	4	#include <linux/slab.h>
	5	#include <linux/list.h>
	6	#include <linux/delay.h>
9dd776b6	7	#include <linux/sched.h>
c93cf61f	8	#include <linux/idr.h>
11a28d37	9	#include <linux/rculist.h>
30ffee84	10	#include <linux/nsproxy.h>
5f256bec	11	#include <net/net_namespace.h>
dec827d1	12	#include <net/netns/generic.h>
5f256bec EB	13
	14	/*
	15	* Our network namespace constructor/destructor lists
	16	*/
	17
	18	static LIST_HEAD(pernet_list);
	19	static struct list_head *first_device = &pernet_list;
	20	static DEFINE_MUTEX(net_mutex);
	21
5f256bec	22	LIST_HEAD(net_namespace_list);
b76a461f	23	EXPORT_SYMBOL_GPL(net_namespace_list);
5f256bec	24
5f256bec	25	struct net init_net;
ff4b9502	26	EXPORT_SYMBOL(init_net);
5f256bec	27
dec827d1 PE	28	#define INITIAL_NET_GEN_PTRS 13 /* +1 for len +2 for rcu_head */
dec827d1 PE	29
f875bae0 EB	30	static int ops_init(const struct pernet_operations ops, struct net net)
	31	{
	32	int err;
	33	if (ops->id && ops->size) {
	34	void *data = kzalloc(ops->size, GFP_KERNEL);
	35	if (!data)
	36	return -ENOMEM;
	37
	38	err = net_assign_generic(net, *ops->id, data);
	39	if (err) {
	40	kfree(data);
	41	return err;
	42	}
	43	}
	44	if (ops->init)
	45	return ops->init(net);
	46	return 0;
	47	}
	48
	49	static void ops_free(const struct pernet_operations ops, struct net net)
	50	{
	51	if (ops->id && ops->size) {
	52	int id = *ops->id;
	53	kfree(net_generic(net, id));
	54	}
	55	}
	56
72ad937a EB	57	static void ops_exit_list(const struct pernet_operations *ops,
	58	struct list_head *net_exit_list)
	59	{
	60	struct net *net;
	61	if (ops->exit) {
	62	list_for_each_entry(net, net_exit_list, exit_list)
	63	ops->exit(net);
	64	}
72ad937a EB	65	if (ops->exit_batch)
	66	ops->exit_batch(net_exit_list);
	67	}
	68
	69	static void ops_free_list(const struct pernet_operations *ops,
	70	struct list_head *net_exit_list)
	71	{
	72	struct net *net;
	73	if (ops->size && ops->id) {
	74	list_for_each_entry(net, net_exit_list, exit_list)
	75	ops_free(ops, net);
	76	}
	77	}
	78
5f256bec EB	79	/*
	80	* setup_net runs the initializers for the network namespace object.
	81	*/
1a2ee93d	82	static __net_init int setup_net(struct net *net)
5f256bec EB	83	{
5f256bec EB	84	/* Must be called with net_mutex held */
f875bae0	85	const struct pernet_operations ops, saved_ops;
486a87f1	86	int error = 0;
72ad937a	87	LIST_HEAD(net_exit_list);
5f256bec	88
5f256bec	89	atomic_set(&net->count, 1);
486a87f1	90
5d1e4468	91	#ifdef NETNS_REFCNT_DEBUG
5f256bec	92	atomic_set(&net->use_count, 0);
5d1e4468	93	#endif
5f256bec	94
768f3591	95	list_for_each_entry(ops, &pernet_list, list) {
f875bae0 EB	96	error = ops_init(ops, net);
	97	if (error < 0)
	98	goto out_undo;
5f256bec EB	99	}
	100	out:
	101	return error;
768f3591	102
5f256bec EB	103	out_undo:
	104	/* Walk through the list backwards calling the exit functions
	105	* for the pernet modules whose init functions did not fail.
	106	*/
72ad937a	107	list_add(&net->exit_list, &net_exit_list);
f875bae0	108	saved_ops = ops;
72ad937a EB	109	list_for_each_entry_continue_reverse(ops, &pernet_list, list)
	110	ops_exit_list(ops, &net_exit_list);
	111
f875bae0 EB	112	ops = saved_ops;
f875bae0 EB	113	list_for_each_entry_continue_reverse(ops, &pernet_list, list)
72ad937a	114	ops_free_list(ops, &net_exit_list);
310928d9 DL	115
310928d9 DL	116	rcu_barrier();
5f256bec EB	117	goto out;
	118	}
	119
486a87f1	120	static struct net_generic *net_alloc_generic(void)
6a1a3b9f	121	{
486a87f1 DL	122	struct net_generic *ng;
	123	size_t generic_size = sizeof(struct net_generic) +
	124	INITIAL_NET_GEN_PTRS * sizeof(void *);
	125
	126	ng = kzalloc(generic_size, GFP_KERNEL);
	127	if (ng)
	128	ng->len = INITIAL_NET_GEN_PTRS;
	129
	130	return ng;
6a1a3b9f PE	131	}
6a1a3b9f PE	132
ebe47d47 CN	133	#ifdef CONFIG_NET_NS
	134	static struct kmem_cache *net_cachep;
	135	static struct workqueue_struct *netns_wq;
	136
486a87f1	137	static struct net *net_alloc(void)
45a19b0a	138	{
486a87f1 DL	139	struct net *net = NULL;
	140	struct net_generic *ng;
	141
	142	ng = net_alloc_generic();
	143	if (!ng)
	144	goto out;
	145
	146	net = kmem_cache_zalloc(net_cachep, GFP_KERNEL);
45a19b0a	147	if (!net)
486a87f1	148	goto out_free;
45a19b0a	149
486a87f1 DL	150	rcu_assign_pointer(net->gen, ng);
	151	out:
	152	return net;
	153
	154	out_free:
	155	kfree(ng);
	156	goto out;
	157	}
	158
	159	static void net_free(struct net *net)
	160	{
5d1e4468	161	#ifdef NETNS_REFCNT_DEBUG
45a19b0a JFS	162	if (unlikely(atomic_read(&net->use_count) != 0)) {
	163	printk(KERN_EMERG "network namespace not free! Usage: %d\n",
	164	atomic_read(&net->use_count));
	165	return;
	166	}
5d1e4468	167	#endif
4ef079cc	168	kfree(net->gen);
45a19b0a JFS	169	kmem_cache_free(net_cachep, net);
	170	}
	171
088eb2d9	172	static struct net *net_create(void)
9dd776b6	173	{
088eb2d9 AD	174	struct net *net;
088eb2d9 AD	175	int rv;
9dd776b6	176
088eb2d9 AD	177	net = net_alloc();
	178	if (!net)
	179	return ERR_PTR(-ENOMEM);
9dd776b6	180	mutex_lock(&net_mutex);
088eb2d9 AD	181	rv = setup_net(net);
088eb2d9 AD	182	if (rv == 0) {
486a87f1	183	rtnl_lock();
11a28d37	184	list_add_tail_rcu(&net->list, &net_namespace_list);
486a87f1 DL	185	rtnl_unlock();
486a87f1 DL	186	}
9dd776b6	187	mutex_unlock(&net_mutex);
088eb2d9 AD	188	if (rv < 0) {
	189	net_free(net);
	190	return ERR_PTR(rv);
	191	}
	192	return net;
	193	}
486a87f1	194
088eb2d9 AD	195	struct net copy_net_ns(unsigned long flags, struct net old_net)
	196	{
	197	if (!(flags & CLONE_NEWNET))
	198	return get_net(old_net);
	199	return net_create();
9dd776b6 EB	200	}
9dd776b6 EB	201
2b035b39 EB	202	static DEFINE_SPINLOCK(cleanup_list_lock);
	203	static LIST_HEAD(cleanup_list); /* Must hold cleanup_list_lock to touch */
	204
6a1a3b9f PE	205	static void cleanup_net(struct work_struct *work)
6a1a3b9f PE	206	{
f875bae0	207	const struct pernet_operations *ops;
2b035b39 EB	208	struct net net, tmp;
2b035b39 EB	209	LIST_HEAD(net_kill_list);
72ad937a	210	LIST_HEAD(net_exit_list);
6a1a3b9f	211
2b035b39 EB	212	/* Atomically snapshot the list of namespaces to cleanup */
	213	spin_lock_irq(&cleanup_list_lock);
	214	list_replace_init(&cleanup_list, &net_kill_list);
	215	spin_unlock_irq(&cleanup_list_lock);
6a1a3b9f PE	216
	217	mutex_lock(&net_mutex);
	218
	219	/* Don't let anyone else find us. */
	220	rtnl_lock();
72ad937a	221	list_for_each_entry(net, &net_kill_list, cleanup_list) {
2b035b39	222	list_del_rcu(&net->list);
72ad937a EB	223	list_add_tail(&net->exit_list, &net_exit_list);
72ad937a EB	224	}
6a1a3b9f PE	225	rtnl_unlock();
6a1a3b9f PE	226
11a28d37 JB	227	/*
	228	* Another CPU might be rcu-iterating the list, wait for it.
	229	* This needs to be before calling the exit() notifiers, so
	230	* the rcu_barrier() below isn't sufficient alone.
	231	*/
	232	synchronize_rcu();
	233
6a1a3b9f	234	/* Run all of the network namespace exit methods */
72ad937a EB	235	list_for_each_entry_reverse(ops, &pernet_list, list)
	236	ops_exit_list(ops, &net_exit_list);
	237
f875bae0	238	/* Free the net generic variables */
72ad937a EB	239	list_for_each_entry_reverse(ops, &pernet_list, list)
72ad937a EB	240	ops_free_list(ops, &net_exit_list);
6a1a3b9f PE	241
	242	mutex_unlock(&net_mutex);
	243
	244	/* Ensure there are no outstanding rcu callbacks using this
	245	* network namespace.
	246	*/
	247	rcu_barrier();
	248
	249	/* Finally it is safe to free my network namespace structure */
72ad937a EB	250	list_for_each_entry_safe(net, tmp, &net_exit_list, exit_list) {
72ad937a EB	251	list_del_init(&net->exit_list);
2b035b39 EB	252	net_free(net);
2b035b39 EB	253	}
6a1a3b9f	254	}
2b035b39	255	static DECLARE_WORK(net_cleanup_work, cleanup_net);
6a1a3b9f PE	256
	257	void __put_net(struct net *net)
	258	{
	259	/* Cleanup the network namespace in process context */
2b035b39 EB	260	unsigned long flags;
	261
	262	spin_lock_irqsave(&cleanup_list_lock, flags);
	263	list_add(&net->cleanup_list, &cleanup_list);
	264	spin_unlock_irqrestore(&cleanup_list_lock, flags);
	265
	266	queue_work(netns_wq, &net_cleanup_work);
6a1a3b9f PE	267	}
	268	EXPORT_SYMBOL_GPL(__put_net);
	269
	270	#else
	271	struct net copy_net_ns(unsigned long flags, struct net old_net)
	272	{
	273	if (flags & CLONE_NEWNET)
	274	return ERR_PTR(-EINVAL);
	275	return old_net;
	276	}
	277	#endif
	278
30ffee84 JB	279	struct net *get_net_ns_by_pid(pid_t pid)
	280	{
	281	struct task_struct *tsk;
	282	struct net *net;
	283
	284	/* Lookup the network namespace */
	285	net = ERR_PTR(-ESRCH);
	286	rcu_read_lock();
	287	tsk = find_task_by_vpid(pid);
	288	if (tsk) {
	289	struct nsproxy *nsproxy;
	290	nsproxy = task_nsproxy(tsk);
	291	if (nsproxy)
	292	net = get_net(nsproxy->net_ns);
	293	}
	294	rcu_read_unlock();
	295	return net;
	296	}
	297	EXPORT_SYMBOL_GPL(get_net_ns_by_pid);
	298
5f256bec EB	299	static int __init net_ns_init(void)
5f256bec EB	300	{
486a87f1	301	struct net_generic *ng;
5f256bec	302
d57a9212	303	#ifdef CONFIG_NET_NS
5f256bec EB	304	net_cachep = kmem_cache_create("net_namespace", sizeof(struct net),
	305	SMP_CACHE_BYTES,
	306	SLAB_PANIC, NULL);
3ef1355d BT	307
	308	/* Create workqueue for cleanup */
	309	netns_wq = create_singlethread_workqueue("netns");
	310	if (!netns_wq)
	311	panic("Could not create netns workq");
d57a9212	312	#endif
3ef1355d	313
486a87f1 DL	314	ng = net_alloc_generic();
	315	if (!ng)
	316	panic("Could not allocate generic netns");
	317
	318	rcu_assign_pointer(init_net.gen, ng);
	319
5f256bec	320	mutex_lock(&net_mutex);
ca0f3112 SH	321	if (setup_net(&init_net))
ca0f3112 SH	322	panic("Could not setup the initial network namespace");
5f256bec	323
f4618d39	324	rtnl_lock();
11a28d37	325	list_add_tail_rcu(&init_net.list, &net_namespace_list);
f4618d39	326	rtnl_unlock();
5f256bec EB	327
5f256bec EB	328	mutex_unlock(&net_mutex);
5f256bec EB	329
	330	return 0;
	331	}
	332
	333	pure_initcall(net_ns_init);
	334
ed160e83	335	#ifdef CONFIG_NET_NS
f875bae0 EB	336	static int __register_pernet_operations(struct list_head *list,
f875bae0 EB	337	struct pernet_operations *ops)
5f256bec	338	{
72ad937a	339	struct net *net;
5f256bec	340	int error;
72ad937a	341	LIST_HEAD(net_exit_list);
5f256bec	342
5f256bec	343	list_add_tail(&ops->list, list);
f875bae0	344	if (ops->init \|\| (ops->id && ops->size)) {
1dba323b	345	for_each_net(net) {
f875bae0	346	error = ops_init(ops, net);
5f256bec EB	347	if (error)
5f256bec EB	348	goto out_undo;
72ad937a	349	list_add_tail(&net->exit_list, &net_exit_list);
5f256bec EB	350	}
5f256bec EB	351	}
1dba323b	352	return 0;
5f256bec EB	353
	354	out_undo:
	355	/* If I have an error cleanup all namespaces I initialized */
	356	list_del(&ops->list);
72ad937a EB	357	ops_exit_list(ops, &net_exit_list);
72ad937a EB	358	ops_free_list(ops, &net_exit_list);
1dba323b	359	return error;
5f256bec EB	360	}
5f256bec EB	361
f875bae0	362	static void __unregister_pernet_operations(struct pernet_operations *ops)
5f256bec EB	363	{
5f256bec EB	364	struct net *net;
72ad937a	365	LIST_HEAD(net_exit_list);
5f256bec EB	366
5f256bec EB	367	list_del(&ops->list);
72ad937a EB	368	for_each_net(net)
	369	list_add_tail(&net->exit_list, &net_exit_list);
	370	ops_exit_list(ops, &net_exit_list);
	371	ops_free_list(ops, &net_exit_list);
5f256bec EB	372	}
5f256bec EB	373
ed160e83 DL	374	#else
ed160e83 DL	375
f875bae0 EB	376	static int __register_pernet_operations(struct list_head *list,
f875bae0 EB	377	struct pernet_operations *ops)
ed160e83	378	{
f875bae0 EB	379	int err = 0;
	380	err = ops_init(ops, &init_net);
	381	if (err)
	382	ops_free(ops, &init_net);
	383	return err;
	384
ed160e83 DL	385	}
ed160e83 DL	386
f875bae0	387	static void __unregister_pernet_operations(struct pernet_operations *ops)
ed160e83	388	{
72ad937a EB	389	LIST_HEAD(net_exit_list);
	390	list_add(&init_net.exit_list, &net_exit_list);
	391	ops_exit_list(ops, &net_exit_list);
	392	ops_free_list(ops, &net_exit_list);
ed160e83	393	}
f875bae0 EB	394
f875bae0 EB	395	#endif /* CONFIG_NET_NS */
ed160e83	396
c93cf61f PE	397	static DEFINE_IDA(net_generic_ids);
c93cf61f PE	398
f875bae0 EB	399	static int register_pernet_operations(struct list_head *list,
	400	struct pernet_operations *ops)
	401	{
	402	int error;
	403
	404	if (ops->id) {
	405	again:
	406	error = ida_get_new_above(&net_generic_ids, 1, ops->id);
	407	if (error < 0) {
	408	if (error == -EAGAIN) {
	409	ida_pre_get(&net_generic_ids, GFP_KERNEL);
	410	goto again;
	411	}
	412	return error;
	413	}
	414	}
	415	error = __register_pernet_operations(list, ops);
	416	if (error && ops->id)
	417	ida_remove(&net_generic_ids, *ops->id);
	418
	419	return error;
	420	}
	421
	422	static void unregister_pernet_operations(struct pernet_operations *ops)
	423	{
	424
	425	__unregister_pernet_operations(ops);
	426	if (ops->id)
	427	ida_remove(&net_generic_ids, *ops->id);
	428	}
	429
5f256bec EB	430	/**
	431	* register_pernet_subsys - register a network namespace subsystem
	432	* @ops: pernet operations structure for the subsystem
	433	*
	434	* Register a subsystem which has init and exit functions
	435	* that are called when network namespaces are created and
	436	* destroyed respectively.
	437	*
	438	* When registered all network namespace init functions are
	439	* called for every existing network namespace. Allowing kernel
	440	* modules to have a race free view of the set of network namespaces.
	441	*
	442	* When a new network namespace is created all of the init
	443	* methods are called in the order in which they were registered.
	444	*
	445	* When a network namespace is destroyed all of the exit methods
	446	* are called in the reverse of the order with which they were
	447	* registered.
	448	*/
	449	int register_pernet_subsys(struct pernet_operations *ops)
	450	{
	451	int error;
	452	mutex_lock(&net_mutex);
	453	error = register_pernet_operations(first_device, ops);
	454	mutex_unlock(&net_mutex);
	455	return error;
	456	}
	457	EXPORT_SYMBOL_GPL(register_pernet_subsys);
	458
	459	/**
	460	* unregister_pernet_subsys - unregister a network namespace subsystem
	461	* @ops: pernet operations structure to manipulate
	462	*
	463	* Remove the pernet operations structure from the list to be
53379e57	464	* used when network namespaces are created or destroyed. In
5f256bec EB	465	* addition run the exit method for all existing network
	466	* namespaces.
	467	*/
	468	void unregister_pernet_subsys(struct pernet_operations *module)
	469	{
	470	mutex_lock(&net_mutex);
	471	unregister_pernet_operations(module);
	472	mutex_unlock(&net_mutex);
	473	}
	474	EXPORT_SYMBOL_GPL(unregister_pernet_subsys);
	475
	476	/**
	477	* register_pernet_device - register a network namespace device
	478	* @ops: pernet operations structure for the subsystem
	479	*
	480	* Register a device which has init and exit functions
	481	* that are called when network namespaces are created and
	482	* destroyed respectively.
	483	*
	484	* When registered all network namespace init functions are
	485	* called for every existing network namespace. Allowing kernel
	486	* modules to have a race free view of the set of network namespaces.
	487	*
	488	* When a new network namespace is created all of the init
	489	* methods are called in the order in which they were registered.
	490	*
	491	* When a network namespace is destroyed all of the exit methods
	492	* are called in the reverse of the order with which they were
	493	* registered.
	494	*/
	495	int register_pernet_device(struct pernet_operations *ops)
	496	{
	497	int error;
	498	mutex_lock(&net_mutex);
	499	error = register_pernet_operations(&pernet_list, ops);
	500	if (!error && (first_device == &pernet_list))
	501	first_device = &ops->list;
	502	mutex_unlock(&net_mutex);
	503	return error;
	504	}
	505	EXPORT_SYMBOL_GPL(register_pernet_device);
	506
	507	/**
	508	* unregister_pernet_device - unregister a network namespace netdevice
	509	* @ops: pernet operations structure to manipulate
	510	*
	511	* Remove the pernet operations structure from the list to be
53379e57	512	* used when network namespaces are created or destroyed. In
5f256bec EB	513	* addition run the exit method for all existing network
	514	* namespaces.
	515	*/
	516	void unregister_pernet_device(struct pernet_operations *ops)
	517	{
	518	mutex_lock(&net_mutex);
	519	if (&ops->list == first_device)
	520	first_device = first_device->next;
	521	unregister_pernet_operations(ops);
	522	mutex_unlock(&net_mutex);
	523	}
	524	EXPORT_SYMBOL_GPL(unregister_pernet_device);
c93cf61f	525
dec827d1 PE	526	static void net_generic_release(struct rcu_head *rcu)
	527	{
	528	struct net_generic *ng;
	529
	530	ng = container_of(rcu, struct net_generic, rcu);
	531	kfree(ng);
	532	}
	533
	534	int net_assign_generic(struct net net, int id, void data)
	535	{
	536	struct net_generic ng, old_ng;
	537
	538	BUG_ON(!mutex_is_locked(&net_mutex));
	539	BUG_ON(id == 0);
	540
	541	ng = old_ng = net->gen;
	542	if (old_ng->len >= id)
	543	goto assign;
	544
	545	ng = kzalloc(sizeof(struct net_generic) +
	546	id * sizeof(void *), GFP_KERNEL);
	547	if (ng == NULL)
	548	return -ENOMEM;
	549
	550	/*
	551	* Some synchronisation notes:
	552	*
	553	* The net_generic explores the net->gen array inside rcu
	554	* read section. Besides once set the net->gen->ptr[x]
	555	* pointer never changes (see rules in netns/generic.h).
	556	*
	557	* That said, we simply duplicate this array and schedule
	558	* the old copy for kfree after a grace period.
	559	*/
	560
	561	ng->len = id;
14458630	562	memcpy(&ng->ptr, &old_ng->ptr, old_ng->len * sizeof(void*));
dec827d1 PE	563
	564	rcu_assign_pointer(net->gen, ng);
	565	call_rcu(&old_ng->rcu, net_generic_release);
	566	assign:
	567	ng->ptr[id - 1] = data;
	568	return 0;
	569	}
	570	EXPORT_SYMBOL_GPL(net_assign_generic);