[deliverable/linux.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 <linux/proc_fs.h>
#include <linux/file.h>
#include <linux/export.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 unsigned int max_gen_ptrs = INITIAL_NET_GEN_PTRS;

static struct net_generic *net_alloc_generic(void)
{
	struct net_generic *ng;
	size_t generic_size = offsetof(struct net_generic, ptr[max_gen_ptrs]);

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

	return ng;
}

static 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);

	old_ng = rcu_dereference_protected(net->gen,
					   lockdep_is_held(&net_mutex));
	ng = old_ng;
	if (old_ng->len >= id)
		goto assign;

	ng = net_alloc_generic();
	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.
	 */

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

	rcu_assign_pointer(net->gen, ng);
	kfree_rcu(old_ng, rcu);
assign:
	ng->ptr[id - 1] = data;
	return 0;
}

static int ops_init(const struct pernet_operations *ops, struct net *net)
{
	int err = -ENOMEM;
	void *data = NULL;

	if (ops->id && ops->size) {
		data = kzalloc(ops->size, GFP_KERNEL);
		if (!data)
			goto out;

		err = net_assign_generic(net, *ops->id, data);
		if (err)
			goto cleanup;
	}
	err = 0;
	if (ops->init)
		err = ops->init(net);
	if (!err)
		return 0;

cleanup:
	kfree(data);

out:
	return err;
}

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);
	atomic_set(&net->passive, 1);
	net->dev_base_seq = 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;
}


#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);
}

void net_drop_ns(void *p)
{
	struct net *ns = p;
	if (ns && atomic_dec_and_test(&ns->passive))
		net_free(ns);
}

struct net *copy_net_ns(unsigned long flags, struct net *old_net)
{
	struct net *net;
	int rv;

	if (!(flags & CLONE_NEWNET))
		return get_net(old_net);

	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_drop_ns(net);
		return ERR_PTR(rv);
	}
	return net;
}

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_drop_ns(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);

struct net *get_net_ns_by_fd(int fd)
{
	struct proc_inode *ei;
	struct file *file;
	struct net *net;

	file = proc_ns_fget(fd);
	if (IS_ERR(file))
		return ERR_CAST(file);

	ei = PROC_I(file->f_dentry->d_inode);
	if (ei->ns_ops == &netns_operations)
		net = get_net(ei->ns);
	else
		net = ERR_PTR(-EINVAL);

	fput(file);
	return 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;
}

struct net *get_net_ns_by_fd(int fd)
{
	return ERR_PTR(-EINVAL);
}
#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)
{
	return ops_init(ops, &init_net);
}

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;
		}
		max_gen_ptrs = max_t(unsigned int, max_gen_ptrs, *ops->id);
	}
	error = __register_pernet_operations(list, ops);
	if (error) {
		rcu_barrier();
		if (ops->id)
			ida_remove(&net_generic_ids, *ops->id);
	}

	return error;
}

static void unregister_pernet_operations(struct pernet_operations *ops)
{
	
	__unregister_pernet_operations(ops);
	rcu_barrier();
	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 *ops)
{
	mutex_lock(&net_mutex);
	unregister_pernet_operations(ops);
	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);

#ifdef CONFIG_NET_NS
static void *netns_get(struct task_struct *task)
{
	struct net *net = NULL;
	struct nsproxy *nsproxy;

	rcu_read_lock();
	nsproxy = task_nsproxy(task);
	if (nsproxy)
		net = get_net(nsproxy->net_ns);
	rcu_read_unlock();

	return net;
}

static void netns_put(void *ns)
{
	put_net(ns);
}

static int netns_install(struct nsproxy *nsproxy, void *ns)
{
	put_net(nsproxy->net_ns);
	nsproxy->net_ns = get_net(ns);
	return 0;
}

const struct proc_ns_operations netns_operations = {
	.name		= "net",
	.type		= CLONE_NEWNET,
	.get		= netns_get,
	.put		= netns_put,
	.install	= netns_install,
};
#endif
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>
f0630529 EB	11	#include <linux/proc_fs.h>
f0630529 EB	12	#include <linux/file.h>
bc3b2d7f	13	#include <linux/export.h>
5f256bec	14	#include <net/net_namespace.h>
dec827d1	15	#include <net/netns/generic.h>
5f256bec EB	16
	17	/*
	18	* Our network namespace constructor/destructor lists
	19	*/
	20
	21	static LIST_HEAD(pernet_list);
	22	static struct list_head *first_device = &pernet_list;
	23	static DEFINE_MUTEX(net_mutex);
	24
5f256bec	25	LIST_HEAD(net_namespace_list);
b76a461f	26	EXPORT_SYMBOL_GPL(net_namespace_list);
5f256bec	27
5f256bec	28	struct net init_net;
ff4b9502	29	EXPORT_SYMBOL(init_net);
5f256bec	30
dec827d1 PE	31	#define INITIAL_NET_GEN_PTRS 13 /* +1 for len +2 for rcu_head */
dec827d1 PE	32
073862ba ED	33	static unsigned int max_gen_ptrs = INITIAL_NET_GEN_PTRS;
	34
	35	static struct net_generic *net_alloc_generic(void)
	36	{
	37	struct net_generic *ng;
	38	size_t generic_size = offsetof(struct net_generic, ptr[max_gen_ptrs]);
	39
	40	ng = kzalloc(generic_size, GFP_KERNEL);
	41	if (ng)
	42	ng->len = max_gen_ptrs;
	43
	44	return ng;
	45	}
	46
05fceb4a JP	47	static int net_assign_generic(struct net net, int id, void data)
	48	{
	49	struct net_generic ng, old_ng;
	50
	51	BUG_ON(!mutex_is_locked(&net_mutex));
	52	BUG_ON(id == 0);
	53
1c87733d ED	54	old_ng = rcu_dereference_protected(net->gen,
	55	lockdep_is_held(&net_mutex));
	56	ng = old_ng;
05fceb4a JP	57	if (old_ng->len >= id)
	58	goto assign;
	59
073862ba	60	ng = net_alloc_generic();
05fceb4a JP	61	if (ng == NULL)
	62	return -ENOMEM;
	63
	64	/*
	65	* Some synchronisation notes:
	66	*
	67	* The net_generic explores the net->gen array inside rcu
	68	* read section. Besides once set the net->gen->ptr[x]
	69	* pointer never changes (see rules in netns/generic.h).
	70	*
	71	* That said, we simply duplicate this array and schedule
	72	* the old copy for kfree after a grace period.
	73	*/
	74
05fceb4a JP	75	memcpy(&ng->ptr, &old_ng->ptr, old_ng->len * sizeof(void*));
	76
	77	rcu_assign_pointer(net->gen, ng);
04d4dfed	78	kfree_rcu(old_ng, rcu);
05fceb4a JP	79	assign:
	80	ng->ptr[id - 1] = data;
	81	return 0;
	82	}
	83
f875bae0 EB	84	static int ops_init(const struct pernet_operations ops, struct net net)
f875bae0 EB	85	{
b922934d JA	86	int err = -ENOMEM;
	87	void *data = NULL;
	88
f875bae0	89	if (ops->id && ops->size) {
b922934d	90	data = kzalloc(ops->size, GFP_KERNEL);
f875bae0	91	if (!data)
b922934d	92	goto out;
f875bae0 EB	93
f875bae0 EB	94	err = net_assign_generic(net, *ops->id, data);
b922934d JA	95	if (err)
b922934d JA	96	goto cleanup;
f875bae0	97	}
b922934d	98	err = 0;
f875bae0	99	if (ops->init)
b922934d JA	100	err = ops->init(net);
	101	if (!err)
	102	return 0;
	103
	104	cleanup:
	105	kfree(data);
	106
	107	out:
	108	return err;
f875bae0 EB	109	}
	110
	111	static void ops_free(const struct pernet_operations ops, struct net net)
	112	{
	113	if (ops->id && ops->size) {
	114	int id = *ops->id;
	115	kfree(net_generic(net, id));
	116	}
	117	}
	118
72ad937a EB	119	static void ops_exit_list(const struct pernet_operations *ops,
	120	struct list_head *net_exit_list)
	121	{
	122	struct net *net;
	123	if (ops->exit) {
	124	list_for_each_entry(net, net_exit_list, exit_list)
	125	ops->exit(net);
	126	}
72ad937a EB	127	if (ops->exit_batch)
	128	ops->exit_batch(net_exit_list);
	129	}
	130
	131	static void ops_free_list(const struct pernet_operations *ops,
	132	struct list_head *net_exit_list)
	133	{
	134	struct net *net;
	135	if (ops->size && ops->id) {
	136	list_for_each_entry(net, net_exit_list, exit_list)
	137	ops_free(ops, net);
	138	}
	139	}
	140
5f256bec EB	141	/*
	142	* setup_net runs the initializers for the network namespace object.
	143	*/
1a2ee93d	144	static __net_init int setup_net(struct net *net)
5f256bec EB	145	{
5f256bec EB	146	/* Must be called with net_mutex held */
f875bae0	147	const struct pernet_operations ops, saved_ops;
486a87f1	148	int error = 0;
72ad937a	149	LIST_HEAD(net_exit_list);
5f256bec	150
5f256bec	151	atomic_set(&net->count, 1);
a685e089	152	atomic_set(&net->passive, 1);
4e985ada	153	net->dev_base_seq = 1;
486a87f1	154
5d1e4468	155	#ifdef NETNS_REFCNT_DEBUG
5f256bec	156	atomic_set(&net->use_count, 0);
5d1e4468	157	#endif
5f256bec	158
768f3591	159	list_for_each_entry(ops, &pernet_list, list) {
f875bae0 EB	160	error = ops_init(ops, net);
	161	if (error < 0)
	162	goto out_undo;
5f256bec EB	163	}
	164	out:
	165	return error;
768f3591	166
5f256bec EB	167	out_undo:
	168	/* Walk through the list backwards calling the exit functions
	169	* for the pernet modules whose init functions did not fail.
	170	*/
72ad937a	171	list_add(&net->exit_list, &net_exit_list);
f875bae0	172	saved_ops = ops;
72ad937a EB	173	list_for_each_entry_continue_reverse(ops, &pernet_list, list)
	174	ops_exit_list(ops, &net_exit_list);
	175
f875bae0 EB	176	ops = saved_ops;
f875bae0 EB	177	list_for_each_entry_continue_reverse(ops, &pernet_list, list)
72ad937a	178	ops_free_list(ops, &net_exit_list);
310928d9 DL	179
310928d9 DL	180	rcu_barrier();
5f256bec EB	181	goto out;
	182	}
	183
6a1a3b9f	184
ebe47d47 CN	185	#ifdef CONFIG_NET_NS
	186	static struct kmem_cache *net_cachep;
	187	static struct workqueue_struct *netns_wq;
	188
486a87f1	189	static struct net *net_alloc(void)
45a19b0a	190	{
486a87f1 DL	191	struct net *net = NULL;
	192	struct net_generic *ng;
	193
	194	ng = net_alloc_generic();
	195	if (!ng)
	196	goto out;
	197
	198	net = kmem_cache_zalloc(net_cachep, GFP_KERNEL);
45a19b0a	199	if (!net)
486a87f1	200	goto out_free;
45a19b0a	201
486a87f1 DL	202	rcu_assign_pointer(net->gen, ng);
	203	out:
	204	return net;
	205
	206	out_free:
	207	kfree(ng);
	208	goto out;
	209	}
	210
	211	static void net_free(struct net *net)
	212	{
5d1e4468	213	#ifdef NETNS_REFCNT_DEBUG
45a19b0a JFS	214	if (unlikely(atomic_read(&net->use_count) != 0)) {
	215	printk(KERN_EMERG "network namespace not free! Usage: %d\n",
	216	atomic_read(&net->use_count));
	217	return;
	218	}
5d1e4468	219	#endif
4ef079cc	220	kfree(net->gen);
45a19b0a JFS	221	kmem_cache_free(net_cachep, net);
	222	}
	223
a685e089 AV	224	void net_drop_ns(void *p)
	225	{
	226	struct net *ns = p;
	227	if (ns && atomic_dec_and_test(&ns->passive))
	228	net_free(ns);
	229	}
	230
911cb193	231	struct net copy_net_ns(unsigned long flags, struct net old_net)
9dd776b6	232	{
088eb2d9 AD	233	struct net *net;
088eb2d9 AD	234	int rv;
9dd776b6	235
911cb193 RL	236	if (!(flags & CLONE_NEWNET))
	237	return get_net(old_net);
	238
088eb2d9 AD	239	net = net_alloc();
	240	if (!net)
	241	return ERR_PTR(-ENOMEM);
9dd776b6	242	mutex_lock(&net_mutex);
088eb2d9 AD	243	rv = setup_net(net);
088eb2d9 AD	244	if (rv == 0) {
486a87f1	245	rtnl_lock();
11a28d37	246	list_add_tail_rcu(&net->list, &net_namespace_list);
486a87f1 DL	247	rtnl_unlock();
486a87f1 DL	248	}
9dd776b6	249	mutex_unlock(&net_mutex);
088eb2d9	250	if (rv < 0) {
a685e089	251	net_drop_ns(net);
088eb2d9 AD	252	return ERR_PTR(rv);
	253	}
	254	return net;
	255	}
486a87f1	256
2b035b39 EB	257	static DEFINE_SPINLOCK(cleanup_list_lock);
	258	static LIST_HEAD(cleanup_list); /* Must hold cleanup_list_lock to touch */
	259
6a1a3b9f PE	260	static void cleanup_net(struct work_struct *work)
6a1a3b9f PE	261	{
f875bae0	262	const struct pernet_operations *ops;
2b035b39 EB	263	struct net net, tmp;
2b035b39 EB	264	LIST_HEAD(net_kill_list);
72ad937a	265	LIST_HEAD(net_exit_list);
6a1a3b9f	266
2b035b39 EB	267	/* Atomically snapshot the list of namespaces to cleanup */
	268	spin_lock_irq(&cleanup_list_lock);
	269	list_replace_init(&cleanup_list, &net_kill_list);
	270	spin_unlock_irq(&cleanup_list_lock);
6a1a3b9f PE	271
	272	mutex_lock(&net_mutex);
	273
	274	/* Don't let anyone else find us. */
	275	rtnl_lock();
72ad937a	276	list_for_each_entry(net, &net_kill_list, cleanup_list) {
2b035b39	277	list_del_rcu(&net->list);
72ad937a EB	278	list_add_tail(&net->exit_list, &net_exit_list);
72ad937a EB	279	}
6a1a3b9f PE	280	rtnl_unlock();
6a1a3b9f PE	281
11a28d37 JB	282	/*
	283	* Another CPU might be rcu-iterating the list, wait for it.
	284	* This needs to be before calling the exit() notifiers, so
	285	* the rcu_barrier() below isn't sufficient alone.
	286	*/
	287	synchronize_rcu();
	288
6a1a3b9f	289	/* Run all of the network namespace exit methods */
72ad937a EB	290	list_for_each_entry_reverse(ops, &pernet_list, list)
	291	ops_exit_list(ops, &net_exit_list);
	292
f875bae0	293	/* Free the net generic variables */
72ad937a EB	294	list_for_each_entry_reverse(ops, &pernet_list, list)
72ad937a EB	295	ops_free_list(ops, &net_exit_list);
6a1a3b9f PE	296
	297	mutex_unlock(&net_mutex);
	298
	299	/* Ensure there are no outstanding rcu callbacks using this
	300	* network namespace.
	301	*/
	302	rcu_barrier();
	303
	304	/* Finally it is safe to free my network namespace structure */
72ad937a EB	305	list_for_each_entry_safe(net, tmp, &net_exit_list, exit_list) {
72ad937a EB	306	list_del_init(&net->exit_list);
a685e089	307	net_drop_ns(net);
2b035b39	308	}
6a1a3b9f	309	}
2b035b39	310	static DECLARE_WORK(net_cleanup_work, cleanup_net);
6a1a3b9f PE	311
	312	void __put_net(struct net *net)
	313	{
	314	/* Cleanup the network namespace in process context */
2b035b39 EB	315	unsigned long flags;
	316
	317	spin_lock_irqsave(&cleanup_list_lock, flags);
	318	list_add(&net->cleanup_list, &cleanup_list);
	319	spin_unlock_irqrestore(&cleanup_list_lock, flags);
	320
	321	queue_work(netns_wq, &net_cleanup_work);
6a1a3b9f PE	322	}
	323	EXPORT_SYMBOL_GPL(__put_net);
	324
956c9207 SR	325	struct net *get_net_ns_by_fd(int fd)
	326	{
	327	struct proc_inode *ei;
	328	struct file *file;
	329	struct net *net;
	330
956c9207	331	file = proc_ns_fget(fd);
c316e6a3 AV	332	if (IS_ERR(file))
c316e6a3 AV	333	return ERR_CAST(file);
956c9207 SR	334
956c9207 SR	335	ei = PROC_I(file->f_dentry->d_inode);
c316e6a3 AV	336	if (ei->ns_ops == &netns_operations)
	337	net = get_net(ei->ns);
	338	else
	339	net = ERR_PTR(-EINVAL);
956c9207	340
c316e6a3	341	fput(file);
956c9207 SR	342	return net;
	343	}
	344
6a1a3b9f PE	345	#else
	346	struct net copy_net_ns(unsigned long flags, struct net old_net)
	347	{
	348	if (flags & CLONE_NEWNET)
	349	return ERR_PTR(-EINVAL);
	350	return old_net;
	351	}
956c9207 SR	352
	353	struct net *get_net_ns_by_fd(int fd)
	354	{
	355	return ERR_PTR(-EINVAL);
	356	}
6a1a3b9f PE	357	#endif
6a1a3b9f PE	358
30ffee84 JB	359	struct net *get_net_ns_by_pid(pid_t pid)
	360	{
	361	struct task_struct *tsk;
	362	struct net *net;
	363
	364	/* Lookup the network namespace */
	365	net = ERR_PTR(-ESRCH);
	366	rcu_read_lock();
	367	tsk = find_task_by_vpid(pid);
	368	if (tsk) {
	369	struct nsproxy *nsproxy;
	370	nsproxy = task_nsproxy(tsk);
	371	if (nsproxy)
	372	net = get_net(nsproxy->net_ns);
	373	}
	374	rcu_read_unlock();
	375	return net;
	376	}
	377	EXPORT_SYMBOL_GPL(get_net_ns_by_pid);
	378
5f256bec EB	379	static int __init net_ns_init(void)
5f256bec EB	380	{
486a87f1	381	struct net_generic *ng;
5f256bec	382
d57a9212	383	#ifdef CONFIG_NET_NS
5f256bec EB	384	net_cachep = kmem_cache_create("net_namespace", sizeof(struct net),
	385	SMP_CACHE_BYTES,
	386	SLAB_PANIC, NULL);
3ef1355d BT	387
	388	/* Create workqueue for cleanup */
	389	netns_wq = create_singlethread_workqueue("netns");
	390	if (!netns_wq)
	391	panic("Could not create netns workq");
d57a9212	392	#endif
3ef1355d	393
486a87f1 DL	394	ng = net_alloc_generic();
	395	if (!ng)
	396	panic("Could not allocate generic netns");
	397
	398	rcu_assign_pointer(init_net.gen, ng);
	399
5f256bec	400	mutex_lock(&net_mutex);
ca0f3112 SH	401	if (setup_net(&init_net))
ca0f3112 SH	402	panic("Could not setup the initial network namespace");
5f256bec	403
f4618d39	404	rtnl_lock();
11a28d37	405	list_add_tail_rcu(&init_net.list, &net_namespace_list);
f4618d39	406	rtnl_unlock();
5f256bec EB	407
5f256bec EB	408	mutex_unlock(&net_mutex);
5f256bec EB	409
	410	return 0;
	411	}
	412
	413	pure_initcall(net_ns_init);
	414
ed160e83	415	#ifdef CONFIG_NET_NS
f875bae0 EB	416	static int __register_pernet_operations(struct list_head *list,
f875bae0 EB	417	struct pernet_operations *ops)
5f256bec	418	{
72ad937a	419	struct net *net;
5f256bec	420	int error;
72ad937a	421	LIST_HEAD(net_exit_list);
5f256bec	422
5f256bec	423	list_add_tail(&ops->list, list);
f875bae0	424	if (ops->init \|\| (ops->id && ops->size)) {
1dba323b	425	for_each_net(net) {
f875bae0	426	error = ops_init(ops, net);
5f256bec EB	427	if (error)
5f256bec EB	428	goto out_undo;
72ad937a	429	list_add_tail(&net->exit_list, &net_exit_list);
5f256bec EB	430	}
5f256bec EB	431	}
1dba323b	432	return 0;
5f256bec EB	433
	434	out_undo:
	435	/* If I have an error cleanup all namespaces I initialized */
	436	list_del(&ops->list);
72ad937a EB	437	ops_exit_list(ops, &net_exit_list);
72ad937a EB	438	ops_free_list(ops, &net_exit_list);
1dba323b	439	return error;
5f256bec EB	440	}
5f256bec EB	441
f875bae0	442	static void __unregister_pernet_operations(struct pernet_operations *ops)
5f256bec EB	443	{
5f256bec EB	444	struct net *net;
72ad937a	445	LIST_HEAD(net_exit_list);
5f256bec EB	446
5f256bec EB	447	list_del(&ops->list);
72ad937a EB	448	for_each_net(net)
	449	list_add_tail(&net->exit_list, &net_exit_list);
	450	ops_exit_list(ops, &net_exit_list);
	451	ops_free_list(ops, &net_exit_list);
5f256bec EB	452	}
5f256bec EB	453
ed160e83 DL	454	#else
ed160e83 DL	455
f875bae0 EB	456	static int __register_pernet_operations(struct list_head *list,
f875bae0 EB	457	struct pernet_operations *ops)
ed160e83	458	{
b922934d	459	return ops_init(ops, &init_net);
ed160e83 DL	460	}
ed160e83 DL	461
f875bae0	462	static void __unregister_pernet_operations(struct pernet_operations *ops)
ed160e83	463	{
72ad937a EB	464	LIST_HEAD(net_exit_list);
	465	list_add(&init_net.exit_list, &net_exit_list);
	466	ops_exit_list(ops, &net_exit_list);
	467	ops_free_list(ops, &net_exit_list);
ed160e83	468	}
f875bae0 EB	469
f875bae0 EB	470	#endif /* CONFIG_NET_NS */
ed160e83	471
c93cf61f PE	472	static DEFINE_IDA(net_generic_ids);
c93cf61f PE	473
f875bae0 EB	474	static int register_pernet_operations(struct list_head *list,
	475	struct pernet_operations *ops)
	476	{
	477	int error;
	478
	479	if (ops->id) {
	480	again:
	481	error = ida_get_new_above(&net_generic_ids, 1, ops->id);
	482	if (error < 0) {
	483	if (error == -EAGAIN) {
	484	ida_pre_get(&net_generic_ids, GFP_KERNEL);
	485	goto again;
	486	}
	487	return error;
	488	}
073862ba	489	max_gen_ptrs = max_t(unsigned int, max_gen_ptrs, *ops->id);
f875bae0 EB	490	}
f875bae0 EB	491	error = __register_pernet_operations(list, ops);
3a765eda EB	492	if (error) {
	493	rcu_barrier();
	494	if (ops->id)
	495	ida_remove(&net_generic_ids, *ops->id);
	496	}
f875bae0 EB	497
	498	return error;
	499	}
	500
	501	static void unregister_pernet_operations(struct pernet_operations *ops)
	502	{
	503
	504	__unregister_pernet_operations(ops);
3a765eda	505	rcu_barrier();
f875bae0 EB	506	if (ops->id)
	507	ida_remove(&net_generic_ids, *ops->id);
	508	}
	509
5f256bec EB	510	/**
	511	* register_pernet_subsys - register a network namespace subsystem
	512	* @ops: pernet operations structure for the subsystem
	513	*
	514	* Register a subsystem which has init and exit functions
	515	* that are called when network namespaces are created and
	516	* destroyed respectively.
	517	*
	518	* When registered all network namespace init functions are
	519	* called for every existing network namespace. Allowing kernel
	520	* modules to have a race free view of the set of network namespaces.
	521	*
	522	* When a new network namespace is created all of the init
	523	* methods are called in the order in which they were registered.
	524	*
	525	* When a network namespace is destroyed all of the exit methods
	526	* are called in the reverse of the order with which they were
	527	* registered.
	528	*/
	529	int register_pernet_subsys(struct pernet_operations *ops)
	530	{
	531	int error;
	532	mutex_lock(&net_mutex);
	533	error = register_pernet_operations(first_device, ops);
	534	mutex_unlock(&net_mutex);
	535	return error;
	536	}
	537	EXPORT_SYMBOL_GPL(register_pernet_subsys);
	538
	539	/**
	540	* unregister_pernet_subsys - unregister a network namespace subsystem
	541	* @ops: pernet operations structure to manipulate
	542	*
	543	* Remove the pernet operations structure from the list to be
53379e57	544	* used when network namespaces are created or destroyed. In
5f256bec EB	545	* addition run the exit method for all existing network
	546	* namespaces.
	547	*/
b3c981d2	548	void unregister_pernet_subsys(struct pernet_operations *ops)
5f256bec EB	549	{
5f256bec EB	550	mutex_lock(&net_mutex);
b3c981d2	551	unregister_pernet_operations(ops);
5f256bec EB	552	mutex_unlock(&net_mutex);
	553	}
	554	EXPORT_SYMBOL_GPL(unregister_pernet_subsys);
	555
	556	/**
	557	* register_pernet_device - register a network namespace device
	558	* @ops: pernet operations structure for the subsystem
	559	*
	560	* Register a device which has init and exit functions
	561	* that are called when network namespaces are created and
	562	* destroyed respectively.
	563	*
	564	* When registered all network namespace init functions are
	565	* called for every existing network namespace. Allowing kernel
	566	* modules to have a race free view of the set of network namespaces.
	567	*
	568	* When a new network namespace is created all of the init
	569	* methods are called in the order in which they were registered.
	570	*
	571	* When a network namespace is destroyed all of the exit methods
	572	* are called in the reverse of the order with which they were
	573	* registered.
	574	*/
	575	int register_pernet_device(struct pernet_operations *ops)
	576	{
	577	int error;
	578	mutex_lock(&net_mutex);
	579	error = register_pernet_operations(&pernet_list, ops);
	580	if (!error && (first_device == &pernet_list))
	581	first_device = &ops->list;
	582	mutex_unlock(&net_mutex);
	583	return error;
	584	}
	585	EXPORT_SYMBOL_GPL(register_pernet_device);
	586
	587	/**
	588	* unregister_pernet_device - unregister a network namespace netdevice
	589	* @ops: pernet operations structure to manipulate
	590	*
	591	* Remove the pernet operations structure from the list to be
53379e57	592	* used when network namespaces are created or destroyed. In
5f256bec EB	593	* addition run the exit method for all existing network
	594	* namespaces.
	595	*/
	596	void unregister_pernet_device(struct pernet_operations *ops)
	597	{
	598	mutex_lock(&net_mutex);
	599	if (&ops->list == first_device)
	600	first_device = first_device->next;
	601	unregister_pernet_operations(ops);
	602	mutex_unlock(&net_mutex);
	603	}
	604	EXPORT_SYMBOL_GPL(unregister_pernet_device);
13b6f576 EB	605
	606	#ifdef CONFIG_NET_NS
	607	static void netns_get(struct task_struct task)
	608	{
f0630529 EB	609	struct net *net = NULL;
	610	struct nsproxy *nsproxy;
	611
13b6f576	612	rcu_read_lock();
f0630529 EB	613	nsproxy = task_nsproxy(task);
	614	if (nsproxy)
	615	net = get_net(nsproxy->net_ns);
13b6f576	616	rcu_read_unlock();
f0630529	617
13b6f576 EB	618	return net;
	619	}
	620
	621	static void netns_put(void *ns)
	622	{
	623	put_net(ns);
	624	}
	625
	626	static int netns_install(struct nsproxy nsproxy, void ns)
	627	{
	628	put_net(nsproxy->net_ns);
	629	nsproxy->net_ns = get_net(ns);
	630	return 0;
	631	}
	632
	633	const struct proc_ns_operations netns_operations = {
	634	.name = "net",
	635	.type = CLONE_NEWNET,
	636	.get = netns_get,
	637	.put = netns_put,
	638	.install = netns_install,
	639	};
	640	#endif