[deliverable/linux.git] / net / ipv4 / inet_fragment.c

/*
 * inet fragments management
 *
 *		This program is free software; you can redistribute it and/or
 *		modify it under the terms of the GNU General Public License
 *		as published by the Free Software Foundation; either version
 *		2 of the License, or (at your option) any later version.
 *
 * 		Authors:	Pavel Emelyanov <xemul@openvz.org>
 *				Started as consolidation of ipv4/ip_fragment.c,
 *				ipv6/reassembly. and ipv6 nf conntrack reassembly
 */

#include <linux/list.h>
#include <linux/spinlock.h>
#include <linux/module.h>
#include <linux/timer.h>
#include <linux/mm.h>
#include <linux/random.h>
#include <linux/skbuff.h>
#include <linux/rtnetlink.h>

#include <net/inet_frag.h>

static void inet_frag_secret_rebuild(unsigned long dummy)
{
	struct inet_frags *f = (struct inet_frags *)dummy;
	unsigned long now = jiffies;
	int i;

	write_lock(&f->lock);
	get_random_bytes(&f->rnd, sizeof(u32));
	for (i = 0; i < INETFRAGS_HASHSZ; i++) {
		struct inet_frag_queue *q;
		struct hlist_node *p, *n;

		hlist_for_each_entry_safe(q, p, n, &f->hash[i], list) {
			unsigned int hval = f->hashfn(q);

			if (hval != i) {
				hlist_del(&q->list);

				/* Relink to new hash chain. */
				hlist_add_head(&q->list, &f->hash[hval]);
			}
		}
	}
	write_unlock(&f->lock);

	mod_timer(&f->secret_timer, now + f->secret_interval);
}

void inet_frags_init(struct inet_frags *f)
{
	int i;

	for (i = 0; i < INETFRAGS_HASHSZ; i++)
		INIT_HLIST_HEAD(&f->hash[i]);

	rwlock_init(&f->lock);

	f->rnd = (u32) ((num_physpages ^ (num_physpages>>7)) ^
				   (jiffies ^ (jiffies >> 6)));

	setup_timer(&f->secret_timer, inet_frag_secret_rebuild,
			(unsigned long)f);
	f->secret_timer.expires = jiffies + f->secret_interval;
	add_timer(&f->secret_timer);
}
EXPORT_SYMBOL(inet_frags_init);

void inet_frags_init_net(struct netns_frags *nf)
{
	nf->nqueues = 0;
	atomic_set(&nf->mem, 0);
	INIT_LIST_HEAD(&nf->lru_list);
}
EXPORT_SYMBOL(inet_frags_init_net);

void inet_frags_fini(struct inet_frags *f)
{
	del_timer(&f->secret_timer);
}
EXPORT_SYMBOL(inet_frags_fini);

void inet_frags_exit_net(struct netns_frags *nf, struct inet_frags *f)
{
	nf->low_thresh = 0;
	inet_frag_evictor(nf, f);
}
EXPORT_SYMBOL(inet_frags_exit_net);

static inline void fq_unlink(struct inet_frag_queue *fq, struct inet_frags *f)
{
	write_lock(&f->lock);
	hlist_del(&fq->list);
	list_del(&fq->lru_list);
	fq->net->nqueues--;
	write_unlock(&f->lock);
}

void inet_frag_kill(struct inet_frag_queue *fq, struct inet_frags *f)
{
	if (del_timer(&fq->timer))
		atomic_dec(&fq->refcnt);

	if (!(fq->last_in & COMPLETE)) {
		fq_unlink(fq, f);
		atomic_dec(&fq->refcnt);
		fq->last_in |= COMPLETE;
	}
}

EXPORT_SYMBOL(inet_frag_kill);

static inline void frag_kfree_skb(struct netns_frags *nf, struct inet_frags *f,
		struct sk_buff *skb, int *work)
{
	if (work)
		*work -= skb->truesize;

	atomic_sub(skb->truesize, &nf->mem);
	if (f->skb_free)
		f->skb_free(skb);
	kfree_skb(skb);
}

void inet_frag_destroy(struct inet_frag_queue *q, struct inet_frags *f,
					int *work)
{
	struct sk_buff *fp;
	struct netns_frags *nf;

	BUG_TRAP(q->last_in & COMPLETE);
	BUG_TRAP(del_timer(&q->timer) == 0);

	/* Release all fragment data. */
	fp = q->fragments;
	nf = q->net;
	while (fp) {
		struct sk_buff *xp = fp->next;

		frag_kfree_skb(nf, f, fp, work);
		fp = xp;
	}

	if (work)
		*work -= f->qsize;
	atomic_sub(f->qsize, &nf->mem);

	if (f->destructor)
		f->destructor(q);
	kfree(q);

}
EXPORT_SYMBOL(inet_frag_destroy);

int inet_frag_evictor(struct netns_frags *nf, struct inet_frags *f)
{
	struct inet_frag_queue *q;
	int work, evicted = 0;

	work = atomic_read(&nf->mem) - nf->low_thresh;
	while (work > 0) {
		read_lock(&f->lock);
		if (list_empty(&nf->lru_list)) {
			read_unlock(&f->lock);
			break;
		}

		q = list_first_entry(&nf->lru_list,
				struct inet_frag_queue, lru_list);
		atomic_inc(&q->refcnt);
		read_unlock(&f->lock);

		spin_lock(&q->lock);
		if (!(q->last_in & COMPLETE))
			inet_frag_kill(q, f);
		spin_unlock(&q->lock);

		if (atomic_dec_and_test(&q->refcnt))
			inet_frag_destroy(q, f, &work);
		evicted++;
	}

	return evicted;
}
EXPORT_SYMBOL(inet_frag_evictor);

static struct inet_frag_queue *inet_frag_intern(struct netns_frags *nf,
		struct inet_frag_queue *qp_in, struct inet_frags *f,
		unsigned int hash, void *arg)
{
	struct inet_frag_queue *qp;
#ifdef CONFIG_SMP
	struct hlist_node *n;
#endif

	write_lock(&f->lock);
#ifdef CONFIG_SMP
	/* With SMP race we have to recheck hash table, because
	 * such entry could be created on other cpu, while we
	 * promoted read lock to write lock.
	 */
	hlist_for_each_entry(qp, n, &f->hash[hash], list) {
		if (qp->net == nf && f->match(qp, arg)) {
			atomic_inc(&qp->refcnt);
			write_unlock(&f->lock);
			qp_in->last_in |= COMPLETE;
			inet_frag_put(qp_in, f);
			return qp;
		}
	}
#endif
	qp = qp_in;
	if (!mod_timer(&qp->timer, jiffies + nf->timeout))
		atomic_inc(&qp->refcnt);

	atomic_inc(&qp->refcnt);
	hlist_add_head(&qp->list, &f->hash[hash]);
	list_add_tail(&qp->lru_list, &nf->lru_list);
	nf->nqueues++;
	write_unlock(&f->lock);
	return qp;
}

static struct inet_frag_queue *inet_frag_alloc(struct netns_frags *nf,
		struct inet_frags *f, void *arg)
{
	struct inet_frag_queue *q;

	q = kzalloc(f->qsize, GFP_ATOMIC);
	if (q == NULL)
		return NULL;

	f->constructor(q, arg);
	atomic_add(f->qsize, &nf->mem);
	setup_timer(&q->timer, f->frag_expire, (unsigned long)q);
	spin_lock_init(&q->lock);
	atomic_set(&q->refcnt, 1);
	q->net = nf;

	return q;
}

static struct inet_frag_queue *inet_frag_create(struct netns_frags *nf,
		struct inet_frags *f, void *arg, unsigned int hash)
{
	struct inet_frag_queue *q;

	q = inet_frag_alloc(nf, f, arg);
	if (q == NULL)
		return NULL;

	return inet_frag_intern(nf, q, f, hash, arg);
}

struct inet_frag_queue *inet_frag_find(struct netns_frags *nf,
		struct inet_frags *f, void *key, unsigned int hash)
{
	struct inet_frag_queue *q;
	struct hlist_node *n;

	read_lock(&f->lock);
	hlist_for_each_entry(q, n, &f->hash[hash], list) {
		if (q->net == nf && f->match(q, key)) {
			atomic_inc(&q->refcnt);
			read_unlock(&f->lock);
			return q;
		}
	}
	read_unlock(&f->lock);

	return inet_frag_create(nf, f, key, hash);
}
EXPORT_SYMBOL(inet_frag_find);
Commit	Line	Data
7eb95156 PE	1	/*
	2	* inet fragments management
	3	*
	4	* This program is free software; you can redistribute it and/or
	5	* modify it under the terms of the GNU General Public License
	6	* as published by the Free Software Foundation; either version
	7	* 2 of the License, or (at your option) any later version.
	8	*
	9	* Authors: Pavel Emelyanov <xemul@openvz.org>
	10	* Started as consolidation of ipv4/ip_fragment.c,
	11	* ipv6/reassembly. and ipv6 nf conntrack reassembly
	12	*/
	13
	14	#include <linux/list.h>
	15	#include <linux/spinlock.h>
	16	#include <linux/module.h>
	17	#include <linux/timer.h>
	18	#include <linux/mm.h>
321a3a99	19	#include <linux/random.h>
1e4b8287 PE	20	#include <linux/skbuff.h>
1e4b8287 PE	21	#include <linux/rtnetlink.h>
7eb95156 PE	22
	23	#include <net/inet_frag.h>
	24
321a3a99 PE	25	static void inet_frag_secret_rebuild(unsigned long dummy)
	26	{
	27	struct inet_frags f = (struct inet_frags )dummy;
	28	unsigned long now = jiffies;
	29	int i;
	30
	31	write_lock(&f->lock);
	32	get_random_bytes(&f->rnd, sizeof(u32));
	33	for (i = 0; i < INETFRAGS_HASHSZ; i++) {
	34	struct inet_frag_queue *q;
	35	struct hlist_node p, n;
	36
	37	hlist_for_each_entry_safe(q, p, n, &f->hash[i], list) {
	38	unsigned int hval = f->hashfn(q);
	39
	40	if (hval != i) {
	41	hlist_del(&q->list);
	42
	43	/* Relink to new hash chain. */
	44	hlist_add_head(&q->list, &f->hash[hval]);
	45	}
	46	}
	47	}
	48	write_unlock(&f->lock);
	49
3b4bc4a2	50	mod_timer(&f->secret_timer, now + f->secret_interval);
321a3a99 PE	51	}
321a3a99 PE	52
7eb95156 PE	53	void inet_frags_init(struct inet_frags *f)
	54	{
	55	int i;
	56
	57	for (i = 0; i < INETFRAGS_HASHSZ; i++)
	58	INIT_HLIST_HEAD(&f->hash[i]);
	59
7eb95156 PE	60	rwlock_init(&f->lock);
	61
	62	f->rnd = (u32) ((num_physpages ^ (num_physpages>>7)) ^
	63	(jiffies ^ (jiffies >> 6)));
	64
b24b8a24 PE	65	setup_timer(&f->secret_timer, inet_frag_secret_rebuild,
b24b8a24 PE	66	(unsigned long)f);
3b4bc4a2	67	f->secret_timer.expires = jiffies + f->secret_interval;
321a3a99	68	add_timer(&f->secret_timer);
7eb95156 PE	69	}
	70	EXPORT_SYMBOL(inet_frags_init);
	71
e5a2bb84 PE	72	void inet_frags_init_net(struct netns_frags *nf)
	73	{
	74	nf->nqueues = 0;
6ddc0822	75	atomic_set(&nf->mem, 0);
3140c25c	76	INIT_LIST_HEAD(&nf->lru_list);
e5a2bb84 PE	77	}
	78	EXPORT_SYMBOL(inet_frags_init_net);
	79
7eb95156 PE	80	void inet_frags_fini(struct inet_frags *f)
7eb95156 PE	81	{
321a3a99	82	del_timer(&f->secret_timer);
7eb95156 PE	83	}
7eb95156 PE	84	EXPORT_SYMBOL(inet_frags_fini);
277e650d	85
81566e83 PE	86	void inet_frags_exit_net(struct netns_frags nf, struct inet_frags f)
	87	{
	88	nf->low_thresh = 0;
	89	inet_frag_evictor(nf, f);
	90	}
	91	EXPORT_SYMBOL(inet_frags_exit_net);
	92
277e650d PE	93	static inline void fq_unlink(struct inet_frag_queue fq, struct inet_frags f)
	94	{
	95	write_lock(&f->lock);
	96	hlist_del(&fq->list);
	97	list_del(&fq->lru_list);
e5a2bb84	98	fq->net->nqueues--;
277e650d PE	99	write_unlock(&f->lock);
	100	}
	101
	102	void inet_frag_kill(struct inet_frag_queue fq, struct inet_frags f)
	103	{
	104	if (del_timer(&fq->timer))
	105	atomic_dec(&fq->refcnt);
	106
	107	if (!(fq->last_in & COMPLETE)) {
	108	fq_unlink(fq, f);
	109	atomic_dec(&fq->refcnt);
	110	fq->last_in \|= COMPLETE;
	111	}
	112	}
	113
	114	EXPORT_SYMBOL(inet_frag_kill);
1e4b8287	115
6ddc0822 PE	116	static inline void frag_kfree_skb(struct netns_frags nf, struct inet_frags f,
6ddc0822 PE	117	struct sk_buff skb, int work)
1e4b8287 PE	118	{
	119	if (work)
	120	*work -= skb->truesize;
	121
6ddc0822	122	atomic_sub(skb->truesize, &nf->mem);
1e4b8287 PE	123	if (f->skb_free)
	124	f->skb_free(skb);
	125	kfree_skb(skb);
	126	}
	127
	128	void inet_frag_destroy(struct inet_frag_queue q, struct inet_frags f,
	129	int *work)
	130	{
	131	struct sk_buff *fp;
6ddc0822	132	struct netns_frags *nf;
1e4b8287 PE	133
	134	BUG_TRAP(q->last_in & COMPLETE);
	135	BUG_TRAP(del_timer(&q->timer) == 0);
	136
	137	/* Release all fragment data. */
	138	fp = q->fragments;
6ddc0822	139	nf = q->net;
1e4b8287 PE	140	while (fp) {
	141	struct sk_buff *xp = fp->next;
	142
6ddc0822	143	frag_kfree_skb(nf, f, fp, work);
1e4b8287 PE	144	fp = xp;
	145	}
	146
	147	if (work)
	148	*work -= f->qsize;
6ddc0822	149	atomic_sub(f->qsize, &nf->mem);
1e4b8287	150
c9547709 PE	151	if (f->destructor)
	152	f->destructor(q);
	153	kfree(q);
1e4b8287 PE	154
	155	}
	156	EXPORT_SYMBOL(inet_frag_destroy);
8e7999c4	157
6ddc0822	158	int inet_frag_evictor(struct netns_frags nf, struct inet_frags f)
8e7999c4 PE	159	{
	160	struct inet_frag_queue *q;
	161	int work, evicted = 0;
	162
e31e0bdc	163	work = atomic_read(&nf->mem) - nf->low_thresh;
8e7999c4 PE	164	while (work > 0) {
8e7999c4 PE	165	read_lock(&f->lock);
3140c25c	166	if (list_empty(&nf->lru_list)) {
8e7999c4 PE	167	read_unlock(&f->lock);
	168	break;
	169	}
	170
3140c25c	171	q = list_first_entry(&nf->lru_list,
8e7999c4 PE	172	struct inet_frag_queue, lru_list);
	173	atomic_inc(&q->refcnt);
	174	read_unlock(&f->lock);
	175
	176	spin_lock(&q->lock);
	177	if (!(q->last_in & COMPLETE))
	178	inet_frag_kill(q, f);
	179	spin_unlock(&q->lock);
	180
	181	if (atomic_dec_and_test(&q->refcnt))
	182	inet_frag_destroy(q, f, &work);
	183	evicted++;
	184	}
	185
	186	return evicted;
	187	}
	188	EXPORT_SYMBOL(inet_frag_evictor);
2588fe1d	189
ac18e750 PE	190	static struct inet_frag_queue inet_frag_intern(struct netns_frags nf,
	191	struct inet_frag_queue qp_in, struct inet_frags f,
	192	unsigned int hash, void *arg)
2588fe1d PE	193	{
	194	struct inet_frag_queue *qp;
	195	#ifdef CONFIG_SMP
	196	struct hlist_node *n;
	197	#endif
	198
	199	write_lock(&f->lock);
	200	#ifdef CONFIG_SMP
	201	/* With SMP race we have to recheck hash table, because
	202	* such entry could be created on other cpu, while we
	203	* promoted read lock to write lock.
	204	*/
	205	hlist_for_each_entry(qp, n, &f->hash[hash], list) {
ac18e750	206	if (qp->net == nf && f->match(qp, arg)) {
2588fe1d PE	207	atomic_inc(&qp->refcnt);
	208	write_unlock(&f->lock);
	209	qp_in->last_in \|= COMPLETE;
	210	inet_frag_put(qp_in, f);
	211	return qp;
	212	}
	213	}
	214	#endif
	215	qp = qp_in;
b2fd5321	216	if (!mod_timer(&qp->timer, jiffies + nf->timeout))
2588fe1d PE	217	atomic_inc(&qp->refcnt);
	218
	219	atomic_inc(&qp->refcnt);
	220	hlist_add_head(&qp->list, &f->hash[hash]);
3140c25c	221	list_add_tail(&qp->lru_list, &nf->lru_list);
e5a2bb84	222	nf->nqueues++;
2588fe1d PE	223	write_unlock(&f->lock);
	224	return qp;
	225	}
e521db9d	226
ac18e750 PE	227	static struct inet_frag_queue inet_frag_alloc(struct netns_frags nf,
ac18e750 PE	228	struct inet_frags f, void arg)
e521db9d PE	229	{
	230	struct inet_frag_queue *q;
	231
	232	q = kzalloc(f->qsize, GFP_ATOMIC);
	233	if (q == NULL)
	234	return NULL;
	235
c6fda282	236	f->constructor(q, arg);
6ddc0822	237	atomic_add(f->qsize, &nf->mem);
e521db9d PE	238	setup_timer(&q->timer, f->frag_expire, (unsigned long)q);
	239	spin_lock_init(&q->lock);
	240	atomic_set(&q->refcnt, 1);
ac18e750	241	q->net = nf;
e521db9d PE	242
	243	return q;
	244	}
c6fda282	245
ac18e750 PE	246	static struct inet_frag_queue inet_frag_create(struct netns_frags nf,
ac18e750 PE	247	struct inet_frags f, void arg, unsigned int hash)
c6fda282 PE	248	{
	249	struct inet_frag_queue *q;
	250
ac18e750	251	q = inet_frag_alloc(nf, f, arg);
c6fda282 PE	252	if (q == NULL)
	253	return NULL;
	254
ac18e750	255	return inet_frag_intern(nf, q, f, hash, arg);
c6fda282	256	}
abd6523d	257
ac18e750 PE	258	struct inet_frag_queue inet_frag_find(struct netns_frags nf,
ac18e750 PE	259	struct inet_frags f, void key, unsigned int hash)
abd6523d PE	260	{
	261	struct inet_frag_queue *q;
	262	struct hlist_node *n;
	263
	264	read_lock(&f->lock);
	265	hlist_for_each_entry(q, n, &f->hash[hash], list) {
ac18e750	266	if (q->net == nf && f->match(q, key)) {
abd6523d PE	267	atomic_inc(&q->refcnt);
	268	read_unlock(&f->lock);
	269	return q;
	270	}
	271	}
	272	read_unlock(&f->lock);
	273
ac18e750	274	return inet_frag_create(nf, f, key, hash);
abd6523d PE	275	}
abd6523d PE	276	EXPORT_SYMBOL(inet_frag_find);