[net-next-2.6.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->ctl->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]);

	INIT_LIST_HEAD(&f->lru_list);
	rwlock_init(&f->lock);

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

	f->nqueues = 0;
	atomic_set(&f->mem, 0);

	init_timer(&f->secret_timer);
	f->secret_timer.function = inet_frag_secret_rebuild;
	f->secret_timer.data = (unsigned long)f;
	f->secret_timer.expires = jiffies + f->ctl->secret_interval;
	add_timer(&f->secret_timer);
}
EXPORT_SYMBOL(inet_frags_init);

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

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);
	f->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 inet_frags *f, struct sk_buff *skb,
						int *work)
{
	if (work)
		*work -= skb->truesize;

	atomic_sub(skb->truesize, &f->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;

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

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

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

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

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

}
EXPORT_SYMBOL(inet_frag_destroy);

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

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

		q = list_first_entry(&f->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 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 (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 + f->ctl->timeout))
		atomic_inc(&qp->refcnt);

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

static struct inet_frag_queue *inet_frag_alloc(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, &f->mem);
	setup_timer(&q->timer, f->frag_expire, (unsigned long)q);
	spin_lock_init(&q->lock);
	atomic_set(&q->refcnt, 1);

	return q;
}

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

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

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

struct inet_frag_queue *inet_frag_find(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 (f->match(q, key)) {
			atomic_inc(&q->refcnt);
			read_unlock(&f->lock);
			return q;
		}
	}
	read_unlock(&f->lock);

	return inet_frag_create(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
	50	mod_timer(&f->secret_timer, now + f->ctl->secret_interval);
	51	}
	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
	60	INIT_LIST_HEAD(&f->lru_list);
	61	rwlock_init(&f->lock);
	62
	63	f->rnd = (u32) ((num_physpages ^ (num_physpages>>7)) ^
	64	(jiffies ^ (jiffies >> 6)));
	65
	66	f->nqueues = 0;
	67	atomic_set(&f->mem, 0);
	68
321a3a99 PE	69	init_timer(&f->secret_timer);
	70	f->secret_timer.function = inet_frag_secret_rebuild;
	71	f->secret_timer.data = (unsigned long)f;
	72	f->secret_timer.expires = jiffies + f->ctl->secret_interval;
	73	add_timer(&f->secret_timer);
7eb95156 PE	74	}
	75	EXPORT_SYMBOL(inet_frags_init);
	76
	77	void inet_frags_fini(struct inet_frags *f)
	78	{
321a3a99	79	del_timer(&f->secret_timer);
7eb95156 PE	80	}
7eb95156 PE	81	EXPORT_SYMBOL(inet_frags_fini);
277e650d PE	82
	83	static inline void fq_unlink(struct inet_frag_queue fq, struct inet_frags f)
	84	{
	85	write_lock(&f->lock);
	86	hlist_del(&fq->list);
	87	list_del(&fq->lru_list);
	88	f->nqueues--;
	89	write_unlock(&f->lock);
	90	}
	91
	92	void inet_frag_kill(struct inet_frag_queue fq, struct inet_frags f)
	93	{
	94	if (del_timer(&fq->timer))
	95	atomic_dec(&fq->refcnt);
	96
	97	if (!(fq->last_in & COMPLETE)) {
	98	fq_unlink(fq, f);
	99	atomic_dec(&fq->refcnt);
	100	fq->last_in \|= COMPLETE;
	101	}
	102	}
	103
	104	EXPORT_SYMBOL(inet_frag_kill);
1e4b8287 PE	105
	106	static inline void frag_kfree_skb(struct inet_frags f, struct sk_buff skb,
	107	int *work)
	108	{
	109	if (work)
	110	*work -= skb->truesize;
	111
	112	atomic_sub(skb->truesize, &f->mem);
	113	if (f->skb_free)
	114	f->skb_free(skb);
	115	kfree_skb(skb);
	116	}
	117
	118	void inet_frag_destroy(struct inet_frag_queue q, struct inet_frags f,
	119	int *work)
	120	{
	121	struct sk_buff *fp;
	122
	123	BUG_TRAP(q->last_in & COMPLETE);
	124	BUG_TRAP(del_timer(&q->timer) == 0);
	125
	126	/* Release all fragment data. */
	127	fp = q->fragments;
	128	while (fp) {
	129	struct sk_buff *xp = fp->next;
	130
	131	frag_kfree_skb(f, fp, work);
	132	fp = xp;
	133	}
	134
	135	if (work)
	136	*work -= f->qsize;
	137	atomic_sub(f->qsize, &f->mem);
	138
c9547709 PE	139	if (f->destructor)
	140	f->destructor(q);
	141	kfree(q);
1e4b8287 PE	142
	143	}
	144	EXPORT_SYMBOL(inet_frag_destroy);
8e7999c4 PE	145
	146	int inet_frag_evictor(struct inet_frags *f)
	147	{
	148	struct inet_frag_queue *q;
	149	int work, evicted = 0;
	150
	151	work = atomic_read(&f->mem) - f->ctl->low_thresh;
	152	while (work > 0) {
	153	read_lock(&f->lock);
	154	if (list_empty(&f->lru_list)) {
	155	read_unlock(&f->lock);
	156	break;
	157	}
	158
	159	q = list_first_entry(&f->lru_list,
	160	struct inet_frag_queue, lru_list);
	161	atomic_inc(&q->refcnt);
	162	read_unlock(&f->lock);
	163
	164	spin_lock(&q->lock);
	165	if (!(q->last_in & COMPLETE))
	166	inet_frag_kill(q, f);
	167	spin_unlock(&q->lock);
	168
	169	if (atomic_dec_and_test(&q->refcnt))
	170	inet_frag_destroy(q, f, &work);
	171	evicted++;
	172	}
	173
	174	return evicted;
	175	}
	176	EXPORT_SYMBOL(inet_frag_evictor);
2588fe1d	177
c6fda282	178	static struct inet_frag_queue inet_frag_intern(struct inet_frag_queue qp_in,
48d60056	179	struct inet_frags f, unsigned int hash, void arg)
2588fe1d PE	180	{
	181	struct inet_frag_queue *qp;
	182	#ifdef CONFIG_SMP
	183	struct hlist_node *n;
	184	#endif
	185
	186	write_lock(&f->lock);
	187	#ifdef CONFIG_SMP
	188	/* With SMP race we have to recheck hash table, because
	189	* such entry could be created on other cpu, while we
	190	* promoted read lock to write lock.
	191	*/
	192	hlist_for_each_entry(qp, n, &f->hash[hash], list) {
48d60056	193	if (f->match(qp, arg)) {
2588fe1d PE	194	atomic_inc(&qp->refcnt);
	195	write_unlock(&f->lock);
	196	qp_in->last_in \|= COMPLETE;
	197	inet_frag_put(qp_in, f);
	198	return qp;
	199	}
	200	}
	201	#endif
	202	qp = qp_in;
	203	if (!mod_timer(&qp->timer, jiffies + f->ctl->timeout))
	204	atomic_inc(&qp->refcnt);
	205
	206	atomic_inc(&qp->refcnt);
	207	hlist_add_head(&qp->list, &f->hash[hash]);
	208	list_add_tail(&qp->lru_list, &f->lru_list);
	209	f->nqueues++;
	210	write_unlock(&f->lock);
	211	return qp;
	212	}
e521db9d	213
c6fda282	214	static struct inet_frag_queue inet_frag_alloc(struct inet_frags f, void *arg)
e521db9d PE	215	{
	216	struct inet_frag_queue *q;
	217
	218	q = kzalloc(f->qsize, GFP_ATOMIC);
	219	if (q == NULL)
	220	return NULL;
	221
c6fda282	222	f->constructor(q, arg);
e521db9d PE	223	atomic_add(f->qsize, &f->mem);
	224	setup_timer(&q->timer, f->frag_expire, (unsigned long)q);
	225	spin_lock_init(&q->lock);
	226	atomic_set(&q->refcnt, 1);
	227
	228	return q;
	229	}
c6fda282	230
abd6523d PE	231	static struct inet_frag_queue inet_frag_create(struct inet_frags f,
abd6523d PE	232	void *arg, unsigned int hash)
c6fda282 PE	233	{
	234	struct inet_frag_queue *q;
	235
	236	q = inet_frag_alloc(f, arg);
	237	if (q == NULL)
	238	return NULL;
	239
48d60056	240	return inet_frag_intern(q, f, hash, arg);
c6fda282	241	}
abd6523d PE	242
	243	struct inet_frag_queue inet_frag_find(struct inet_frags f, void *key,
	244	unsigned int hash)
	245	{
	246	struct inet_frag_queue *q;
	247	struct hlist_node *n;
	248
	249	read_lock(&f->lock);
	250	hlist_for_each_entry(q, n, &f->hash[hash], list) {
	251	if (f->match(q, key)) {
	252	atomic_inc(&q->refcnt);
	253	read_unlock(&f->lock);
	254	return q;
	255	}
	256	}
	257	read_unlock(&f->lock);
	258
	259	return inet_frag_create(f, key, hash);
	260	}
	261	EXPORT_SYMBOL(inet_frag_find);