[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 *n;

		hlist_for_each_entry_safe(q, 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);
	spin_lock_init(&nf->lru_lock);
}
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();

	percpu_counter_destroy(&nf->mem);
}
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);
	fq->net->nqueues--;
	write_unlock(&f->lock);
	inet_frag_lru_del(fq);
}

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) {
		spin_lock(&nf->lru_lock);

		if (list_empty(&nf->lru_list)) {
			spin_unlock(&nf->lru_lock);
			break;
		}

		q = list_first_entry(&nf->lru_list,
				struct inet_frag_queue, lru_list);
		atomic_inc(&q->refcnt);
		spin_unlock(&nf->lru_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
#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, &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]);
	nf->nqueues++;
	write_unlock(&f->lock);
	inet_frag_lru_add(nf, qp);
	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;

	hlist_for_each_entry(q, &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;
b67bfe0d	36	struct hlist_node *n;
321a3a99	37
b67bfe0d	38	hlist_for_each_entry_safe(q, n, &f->hash[i], list) {
321a3a99 PE	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);
3ef0eb0d	78	spin_lock_init(&nf->lru_lock);
e5a2bb84 PE	79	}
	80	EXPORT_SYMBOL(inet_frags_init_net);
	81
7eb95156 PE	82	void inet_frags_fini(struct inet_frags *f)
7eb95156 PE	83	{
321a3a99	84	del_timer(&f->secret_timer);
7eb95156 PE	85	}
7eb95156 PE	86	EXPORT_SYMBOL(inet_frags_fini);
277e650d	87
81566e83 PE	88	void inet_frags_exit_net(struct netns_frags nf, struct inet_frags f)
	89	{
	90	nf->low_thresh = 0;
e8e16b70 DM	91
e8e16b70 DM	92	local_bh_disable();
6b102865	93	inet_frag_evictor(nf, f, true);
e8e16b70	94	local_bh_enable();
6d7b857d JDB	95
6d7b857d JDB	96	percpu_counter_destroy(&nf->mem);
81566e83 PE	97	}
	98	EXPORT_SYMBOL(inet_frags_exit_net);
	99
277e650d PE	100	static inline void fq_unlink(struct inet_frag_queue fq, struct inet_frags f)
	101	{
	102	write_lock(&f->lock);
	103	hlist_del(&fq->list);
e5a2bb84	104	fq->net->nqueues--;
277e650d	105	write_unlock(&f->lock);
3ef0eb0d	106	inet_frag_lru_del(fq);
277e650d PE	107	}
	108
	109	void inet_frag_kill(struct inet_frag_queue fq, struct inet_frags f)
	110	{
	111	if (del_timer(&fq->timer))
	112	atomic_dec(&fq->refcnt);
	113
bc578a54	114	if (!(fq->last_in & INET_FRAG_COMPLETE)) {
277e650d PE	115	fq_unlink(fq, f);
277e650d PE	116	atomic_dec(&fq->refcnt);
bc578a54	117	fq->last_in \|= INET_FRAG_COMPLETE;
277e650d PE	118	}
277e650d PE	119	}
277e650d	120	EXPORT_SYMBOL(inet_frag_kill);
1e4b8287	121
6ddc0822	122	static inline void frag_kfree_skb(struct netns_frags nf, struct inet_frags f,
d433673e	123	struct sk_buff *skb)
1e4b8287	124	{
1e4b8287 PE	125	if (f->skb_free)
	126	f->skb_free(skb);
	127	kfree_skb(skb);
	128	}
	129
	130	void inet_frag_destroy(struct inet_frag_queue q, struct inet_frags f,
	131	int *work)
	132	{
	133	struct sk_buff *fp;
6ddc0822	134	struct netns_frags *nf;
d433673e	135	unsigned int sum, sum_truesize = 0;
1e4b8287	136
547b792c IJ	137	WARN_ON(!(q->last_in & INET_FRAG_COMPLETE));
547b792c IJ	138	WARN_ON(del_timer(&q->timer) != 0);
1e4b8287 PE	139
	140	/* Release all fragment data. */
	141	fp = q->fragments;
6ddc0822	142	nf = q->net;
1e4b8287 PE	143	while (fp) {
	144	struct sk_buff *xp = fp->next;
	145
d433673e JDB	146	sum_truesize += fp->truesize;
d433673e JDB	147	frag_kfree_skb(nf, f, fp);
1e4b8287 PE	148	fp = xp;
1e4b8287 PE	149	}
d433673e	150	sum = sum_truesize + f->qsize;
1e4b8287	151	if (work)
d433673e JDB	152	*work -= sum;
d433673e JDB	153	sub_frag_mem_limit(q, sum);
1e4b8287	154
c9547709 PE	155	if (f->destructor)
	156	f->destructor(q);
	157	kfree(q);
1e4b8287 PE	158
	159	}
	160	EXPORT_SYMBOL(inet_frag_destroy);
8e7999c4	161
6b102865	162	int inet_frag_evictor(struct netns_frags nf, struct inet_frags f, bool force)
8e7999c4 PE	163	{
	164	struct inet_frag_queue *q;
	165	int work, evicted = 0;
	166
6b102865	167	if (!force) {
d433673e	168	if (frag_mem_limit(nf) <= nf->high_thresh)
6b102865 AW	169	return 0;
	170	}
	171
d433673e	172	work = frag_mem_limit(nf) - nf->low_thresh;
8e7999c4	173	while (work > 0) {
3ef0eb0d JDB	174	spin_lock(&nf->lru_lock);
3ef0eb0d JDB	175
3140c25c	176	if (list_empty(&nf->lru_list)) {
3ef0eb0d	177	spin_unlock(&nf->lru_lock);
8e7999c4 PE	178	break;
	179	}
	180
3140c25c	181	q = list_first_entry(&nf->lru_list,
8e7999c4 PE	182	struct inet_frag_queue, lru_list);
8e7999c4 PE	183	atomic_inc(&q->refcnt);
3ef0eb0d	184	spin_unlock(&nf->lru_lock);
8e7999c4 PE	185
8e7999c4 PE	186	spin_lock(&q->lock);
bc578a54	187	if (!(q->last_in & INET_FRAG_COMPLETE))
8e7999c4 PE	188	inet_frag_kill(q, f);
	189	spin_unlock(&q->lock);
	190
	191	if (atomic_dec_and_test(&q->refcnt))
	192	inet_frag_destroy(q, f, &work);
	193	evicted++;
	194	}
	195
	196	return evicted;
	197	}
	198	EXPORT_SYMBOL(inet_frag_evictor);
2588fe1d	199
ac18e750 PE	200	static struct inet_frag_queue inet_frag_intern(struct netns_frags nf,
ac18e750 PE	201	struct inet_frag_queue qp_in, struct inet_frags f,
9a375803	202	void *arg)
2588fe1d PE	203	{
	204	struct inet_frag_queue *qp;
	205	#ifdef CONFIG_SMP
2588fe1d	206	#endif
9a375803	207	unsigned int hash;
2588fe1d PE	208
2588fe1d PE	209	write_lock(&f->lock);
9a375803 PE	210	/*
	211	* While we stayed w/o the lock other CPU could update
	212	* the rnd seed, so we need to re-calculate the hash
	213	* chain. Fortunatelly the qp_in can be used to get one.
	214	*/
	215	hash = f->hashfn(qp_in);
2588fe1d PE	216	#ifdef CONFIG_SMP
	217	/* With SMP race we have to recheck hash table, because
	218	* such entry could be created on other cpu, while we
	219	* promoted read lock to write lock.
	220	*/
b67bfe0d	221	hlist_for_each_entry(qp, &f->hash[hash], list) {
ac18e750	222	if (qp->net == nf && f->match(qp, arg)) {
2588fe1d PE	223	atomic_inc(&qp->refcnt);
2588fe1d PE	224	write_unlock(&f->lock);
bc578a54	225	qp_in->last_in \|= INET_FRAG_COMPLETE;
2588fe1d PE	226	inet_frag_put(qp_in, f);
	227	return qp;
	228	}
	229	}
	230	#endif
	231	qp = qp_in;
b2fd5321	232	if (!mod_timer(&qp->timer, jiffies + nf->timeout))
2588fe1d PE	233	atomic_inc(&qp->refcnt);
	234
	235	atomic_inc(&qp->refcnt);
	236	hlist_add_head(&qp->list, &f->hash[hash]);
e5a2bb84	237	nf->nqueues++;
2588fe1d	238	write_unlock(&f->lock);
3ef0eb0d	239	inet_frag_lru_add(nf, qp);
2588fe1d PE	240	return qp;
2588fe1d PE	241	}
e521db9d	242
ac18e750 PE	243	static struct inet_frag_queue inet_frag_alloc(struct netns_frags nf,
ac18e750 PE	244	struct inet_frags f, void arg)
e521db9d PE	245	{
	246	struct inet_frag_queue *q;
	247
	248	q = kzalloc(f->qsize, GFP_ATOMIC);
	249	if (q == NULL)
	250	return NULL;
	251
54db0cc2	252	q->net = nf;
c6fda282	253	f->constructor(q, arg);
d433673e JDB	254	add_frag_mem_limit(q, f->qsize);
d433673e JDB	255
e521db9d PE	256	setup_timer(&q->timer, f->frag_expire, (unsigned long)q);
	257	spin_lock_init(&q->lock);
	258	atomic_set(&q->refcnt, 1);
	259
	260	return q;
	261	}
c6fda282	262
ac18e750	263	static struct inet_frag_queue inet_frag_create(struct netns_frags nf,
9a375803	264	struct inet_frags f, void arg)
c6fda282 PE	265	{
	266	struct inet_frag_queue *q;
	267
ac18e750	268	q = inet_frag_alloc(nf, f, arg);
c6fda282 PE	269	if (q == NULL)
	270	return NULL;
	271
9a375803	272	return inet_frag_intern(nf, q, f, arg);
c6fda282	273	}
abd6523d	274
ac18e750 PE	275	struct inet_frag_queue inet_frag_find(struct netns_frags nf,
ac18e750 PE	276	struct inet_frags f, void key, unsigned int hash)
56bca31f	277	__releases(&f->lock)
abd6523d PE	278	{
abd6523d PE	279	struct inet_frag_queue *q;
abd6523d	280
b67bfe0d	281	hlist_for_each_entry(q, &f->hash[hash], list) {
ac18e750	282	if (q->net == nf && f->match(q, key)) {
abd6523d PE	283	atomic_inc(&q->refcnt);
	284	read_unlock(&f->lock);
	285	return q;
	286	}
	287	}
	288	read_unlock(&f->lock);
	289
9a375803	290	return inet_frag_create(nf, f, key);
abd6523d PE	291	}
abd6523d PE	292	EXPORT_SYMBOL(inet_frag_find);