[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 <linux/slab.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;
	init_frag_mem_limit(nf);
	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;

	local_bh_disable();
	inet_frag_evictor(nf, f, true);
	local_bh_enable();
}
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 & INET_FRAG_COMPLETE)) {
		fq_unlink(fq, f);
		atomic_dec(&fq->refcnt);
		fq->last_in |= INET_FRAG_COMPLETE;
	}
}
EXPORT_SYMBOL(inet_frag_kill);

static inline void frag_kfree_skb(struct netns_frags *nf, struct inet_frags *f,
		struct sk_buff *skb)
{
	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;
	unsigned int sum, sum_truesize = 0;

	WARN_ON(!(q->last_in & INET_FRAG_COMPLETE));
	WARN_ON(del_timer(&q->timer) != 0);

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

		sum_truesize += fp->truesize;
		frag_kfree_skb(nf, f, fp);
		fp = xp;
	}
	sum = sum_truesize + f->qsize;
	if (work)
		*work -= sum;
	sub_frag_mem_limit(q, sum);

	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, bool force)
{
	struct inet_frag_queue *q;
	int work, evicted = 0;

	if (!force) {
		if (frag_mem_limit(nf) <= nf->high_thresh)
			return 0;
	}

	work = frag_mem_limit(nf) - 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 & INET_FRAG_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,
		void *arg)
{
	struct inet_frag_queue *qp;
#ifdef CONFIG_SMP
	struct hlist_node *n;
#endif
	unsigned int hash;

	write_lock(&f->lock);
	/*
	 * While we stayed w/o the lock other CPU could update
	 * the rnd seed, so we need to re-calculate the hash
	 * chain. Fortunatelly the qp_in can be used to get one.
	 */
	hash = f->hashfn(qp_in);
#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 |= INET_FRAG_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;

	q->net = nf;
	f->constructor(q, arg);
	add_frag_mem_limit(q, f->qsize);

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

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

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

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

	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);
}
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>
5a0e3ad6	22	#include <linux/slab.h>
7eb95156 PE	23
	24	#include <net/inet_frag.h>
	25
321a3a99 PE	26	static void inet_frag_secret_rebuild(unsigned long dummy)
	27	{
	28	struct inet_frags f = (struct inet_frags )dummy;
	29	unsigned long now = jiffies;
	30	int i;
	31
	32	write_lock(&f->lock);
	33	get_random_bytes(&f->rnd, sizeof(u32));
	34	for (i = 0; i < INETFRAGS_HASHSZ; i++) {
	35	struct inet_frag_queue *q;
	36	struct hlist_node p, n;
	37
	38	hlist_for_each_entry_safe(q, p, n, &f->hash[i], list) {
	39	unsigned int hval = f->hashfn(q);
	40
	41	if (hval != i) {
	42	hlist_del(&q->list);
	43
	44	/* Relink to new hash chain. */
	45	hlist_add_head(&q->list, &f->hash[hval]);
	46	}
	47	}
	48	}
	49	write_unlock(&f->lock);
	50
3b4bc4a2	51	mod_timer(&f->secret_timer, now + f->secret_interval);
321a3a99 PE	52	}
321a3a99 PE	53
7eb95156 PE	54	void inet_frags_init(struct inet_frags *f)
	55	{
	56	int i;
	57
	58	for (i = 0; i < INETFRAGS_HASHSZ; i++)
	59	INIT_HLIST_HEAD(&f->hash[i]);
	60
7eb95156 PE	61	rwlock_init(&f->lock);
	62
	63	f->rnd = (u32) ((num_physpages ^ (num_physpages>>7)) ^
	64	(jiffies ^ (jiffies >> 6)));
	65
b24b8a24 PE	66	setup_timer(&f->secret_timer, inet_frag_secret_rebuild,
b24b8a24 PE	67	(unsigned long)f);
3b4bc4a2	68	f->secret_timer.expires = jiffies + f->secret_interval;
321a3a99	69	add_timer(&f->secret_timer);
7eb95156 PE	70	}
	71	EXPORT_SYMBOL(inet_frags_init);
	72
e5a2bb84 PE	73	void inet_frags_init_net(struct netns_frags *nf)
	74	{
	75	nf->nqueues = 0;
d433673e	76	init_frag_mem_limit(nf);
3140c25c	77	INIT_LIST_HEAD(&nf->lru_list);
e5a2bb84 PE	78	}
	79	EXPORT_SYMBOL(inet_frags_init_net);
	80
7eb95156 PE	81	void inet_frags_fini(struct inet_frags *f)
7eb95156 PE	82	{
321a3a99	83	del_timer(&f->secret_timer);
7eb95156 PE	84	}
7eb95156 PE	85	EXPORT_SYMBOL(inet_frags_fini);
277e650d	86
81566e83 PE	87	void inet_frags_exit_net(struct netns_frags nf, struct inet_frags f)
	88	{
	89	nf->low_thresh = 0;
e8e16b70 DM	90
e8e16b70 DM	91	local_bh_disable();
6b102865	92	inet_frag_evictor(nf, f, true);
e8e16b70	93	local_bh_enable();
81566e83 PE	94	}
	95	EXPORT_SYMBOL(inet_frags_exit_net);
	96
277e650d PE	97	static inline void fq_unlink(struct inet_frag_queue fq, struct inet_frags f)
	98	{
	99	write_lock(&f->lock);
	100	hlist_del(&fq->list);
	101	list_del(&fq->lru_list);
e5a2bb84	102	fq->net->nqueues--;
277e650d PE	103	write_unlock(&f->lock);
	104	}
	105
	106	void inet_frag_kill(struct inet_frag_queue fq, struct inet_frags f)
	107	{
	108	if (del_timer(&fq->timer))
	109	atomic_dec(&fq->refcnt);
	110
bc578a54	111	if (!(fq->last_in & INET_FRAG_COMPLETE)) {
277e650d PE	112	fq_unlink(fq, f);
277e650d PE	113	atomic_dec(&fq->refcnt);
bc578a54	114	fq->last_in \|= INET_FRAG_COMPLETE;
277e650d PE	115	}
277e650d PE	116	}
277e650d	117	EXPORT_SYMBOL(inet_frag_kill);
1e4b8287	118
6ddc0822	119	static inline void frag_kfree_skb(struct netns_frags nf, struct inet_frags f,
d433673e	120	struct sk_buff *skb)
1e4b8287	121	{
1e4b8287 PE	122	if (f->skb_free)
	123	f->skb_free(skb);
	124	kfree_skb(skb);
	125	}
	126
	127	void inet_frag_destroy(struct inet_frag_queue q, struct inet_frags f,
	128	int *work)
	129	{
	130	struct sk_buff *fp;
6ddc0822	131	struct netns_frags *nf;
d433673e	132	unsigned int sum, sum_truesize = 0;
1e4b8287	133
547b792c IJ	134	WARN_ON(!(q->last_in & INET_FRAG_COMPLETE));
547b792c IJ	135	WARN_ON(del_timer(&q->timer) != 0);
1e4b8287 PE	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
d433673e JDB	143	sum_truesize += fp->truesize;
d433673e JDB	144	frag_kfree_skb(nf, f, fp);
1e4b8287 PE	145	fp = xp;
1e4b8287 PE	146	}
d433673e	147	sum = sum_truesize + f->qsize;
1e4b8287	148	if (work)
d433673e JDB	149	*work -= sum;
d433673e JDB	150	sub_frag_mem_limit(q, sum);
1e4b8287	151
c9547709 PE	152	if (f->destructor)
	153	f->destructor(q);
	154	kfree(q);
1e4b8287 PE	155
	156	}
	157	EXPORT_SYMBOL(inet_frag_destroy);
8e7999c4	158
6b102865	159	int inet_frag_evictor(struct netns_frags nf, struct inet_frags f, bool force)
8e7999c4 PE	160	{
	161	struct inet_frag_queue *q;
	162	int work, evicted = 0;
	163
6b102865	164	if (!force) {
d433673e	165	if (frag_mem_limit(nf) <= nf->high_thresh)
6b102865 AW	166	return 0;
	167	}
	168
d433673e	169	work = frag_mem_limit(nf) - nf->low_thresh;
8e7999c4 PE	170	while (work > 0) {
8e7999c4 PE	171	read_lock(&f->lock);
3140c25c	172	if (list_empty(&nf->lru_list)) {
8e7999c4 PE	173	read_unlock(&f->lock);
	174	break;
	175	}
	176
3140c25c	177	q = list_first_entry(&nf->lru_list,
8e7999c4 PE	178	struct inet_frag_queue, lru_list);
	179	atomic_inc(&q->refcnt);
	180	read_unlock(&f->lock);
	181
	182	spin_lock(&q->lock);
bc578a54	183	if (!(q->last_in & INET_FRAG_COMPLETE))
8e7999c4 PE	184	inet_frag_kill(q, f);
	185	spin_unlock(&q->lock);
	186
	187	if (atomic_dec_and_test(&q->refcnt))
	188	inet_frag_destroy(q, f, &work);
	189	evicted++;
	190	}
	191
	192	return evicted;
	193	}
	194	EXPORT_SYMBOL(inet_frag_evictor);
2588fe1d	195
ac18e750 PE	196	static struct inet_frag_queue inet_frag_intern(struct netns_frags nf,
ac18e750 PE	197	struct inet_frag_queue qp_in, struct inet_frags f,
9a375803	198	void *arg)
2588fe1d PE	199	{
	200	struct inet_frag_queue *qp;
	201	#ifdef CONFIG_SMP
	202	struct hlist_node *n;
	203	#endif
9a375803	204	unsigned int hash;
2588fe1d PE	205
2588fe1d PE	206	write_lock(&f->lock);
9a375803 PE	207	/*
	208	* While we stayed w/o the lock other CPU could update
	209	* the rnd seed, so we need to re-calculate the hash
	210	* chain. Fortunatelly the qp_in can be used to get one.
	211	*/
	212	hash = f->hashfn(qp_in);
2588fe1d PE	213	#ifdef CONFIG_SMP
	214	/* With SMP race we have to recheck hash table, because
	215	* such entry could be created on other cpu, while we
	216	* promoted read lock to write lock.
	217	*/
	218	hlist_for_each_entry(qp, n, &f->hash[hash], list) {
ac18e750	219	if (qp->net == nf && f->match(qp, arg)) {
2588fe1d PE	220	atomic_inc(&qp->refcnt);
2588fe1d PE	221	write_unlock(&f->lock);
bc578a54	222	qp_in->last_in \|= INET_FRAG_COMPLETE;
2588fe1d PE	223	inet_frag_put(qp_in, f);
	224	return qp;
	225	}
	226	}
	227	#endif
	228	qp = qp_in;
b2fd5321	229	if (!mod_timer(&qp->timer, jiffies + nf->timeout))
2588fe1d PE	230	atomic_inc(&qp->refcnt);
	231
	232	atomic_inc(&qp->refcnt);
	233	hlist_add_head(&qp->list, &f->hash[hash]);
3140c25c	234	list_add_tail(&qp->lru_list, &nf->lru_list);
e5a2bb84	235	nf->nqueues++;
2588fe1d PE	236	write_unlock(&f->lock);
	237	return qp;
	238	}
e521db9d	239
ac18e750 PE	240	static struct inet_frag_queue inet_frag_alloc(struct netns_frags nf,
ac18e750 PE	241	struct inet_frags f, void arg)
e521db9d PE	242	{
	243	struct inet_frag_queue *q;
	244
	245	q = kzalloc(f->qsize, GFP_ATOMIC);
	246	if (q == NULL)
	247	return NULL;
	248
54db0cc2	249	q->net = nf;
c6fda282	250	f->constructor(q, arg);
d433673e JDB	251	add_frag_mem_limit(q, f->qsize);
d433673e JDB	252
e521db9d PE	253	setup_timer(&q->timer, f->frag_expire, (unsigned long)q);
	254	spin_lock_init(&q->lock);
	255	atomic_set(&q->refcnt, 1);
	256
	257	return q;
	258	}
c6fda282	259
ac18e750	260	static struct inet_frag_queue inet_frag_create(struct netns_frags nf,
9a375803	261	struct inet_frags f, void arg)
c6fda282 PE	262	{
	263	struct inet_frag_queue *q;
	264
ac18e750	265	q = inet_frag_alloc(nf, f, arg);
c6fda282 PE	266	if (q == NULL)
	267	return NULL;
	268
9a375803	269	return inet_frag_intern(nf, q, f, arg);
c6fda282	270	}
abd6523d	271
ac18e750 PE	272	struct inet_frag_queue inet_frag_find(struct netns_frags nf,
ac18e750 PE	273	struct inet_frags f, void key, unsigned int hash)
56bca31f	274	__releases(&f->lock)
abd6523d PE	275	{
	276	struct inet_frag_queue *q;
	277	struct hlist_node *n;
	278
abd6523d	279	hlist_for_each_entry(q, n, &f->hash[hash], list) {
ac18e750	280	if (q->net == nf && f->match(q, key)) {
abd6523d PE	281	atomic_inc(&q->refcnt);
	282	read_unlock(&f->lock);
	283	return q;
	284	}
	285	}
	286	read_unlock(&f->lock);
	287
9a375803	288	return inet_frag_create(nf, f, key);
abd6523d PE	289	}
abd6523d PE	290	EXPORT_SYMBOL(inet_frag_find);