reassembly.c

Linux Kernel 2.6.9 for OMAP1710

/*
 *	IPv6 fragment reassembly
 *	Linux INET6 implementation 
 *
 *	Authors:
 *	Pedro Roque		<[e-mail]>	
 *
 *	$Id: reassembly.c,v 1.26 2001/03/07 22:00:57 davem Exp $
 *
 *	Based on: net/ipv4/ip_fragment.c
 *
 *	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.
 */

/* 
 *	Fixes:	
 *	Andi Kleen	Make it work with multiple hosts.
 *			More RFC compliance.
 *
 *      Horst von Brand Add missing #include <linux/string.h>
 *	Alexey Kuznetsov	SMP races, threading, cleanup.
 *	Patrick McHardy		LRU queue of frag heads for evictor.
 *	Mitsuru KANDA @USAGI	Register inet6_protocol{}.
 *	David Stevens and
 *	YOSHIFUJI,H. @USAGI	Always remove fragment header to
 *				calculate ICV correctly.
 */
#include <linux/config.h>
#include <linux/errno.h>
#include <linux/types.h>
#include <linux/string.h>
#include <linux/socket.h>
#include <linux/sockios.h>
#include <linux/jiffies.h>
#include <linux/net.h>
#include <linux/list.h>
#include <linux/netdevice.h>
#include <linux/in6.h>
#include <linux/ipv6.h>
#include <linux/icmpv6.h>
#include <linux/random.h>
#include <linux/jhash.h>

#include <net/sock.h>
#include <net/snmp.h>

#include <net/ipv6.h>
#include <net/protocol.h>
#include <net/transp_v6.h>
#include <net/rawv6.h>
#include <net/ndisc.h>
#include <net/addrconf.h>

int sysctl_ip6frag_high_thresh = 256*1024;
int sysctl_ip6frag_low_thresh = 192*1024;

int sysctl_ip6frag_time = IPV6_FRAG_TIMEOUT;

struct ip6frag_skb_cb
{
	struct inet6_skb_parm	h;
	int			offset;
};

#define FRAG6_CB(skb)	((struct ip6frag_skb_cb*)((skb)->cb))


/*
 *	Equivalent of ipv4 struct ipq
 */

struct frag_queue
{
	struct frag_queue	*next;
	struct list_head lru_list;		/* lru list member	*/

	__u32			id;		/* fragment id		*/
	struct in6_addr		saddr;
	struct in6_addr		daddr;

	spinlock_t		lock;
	atomic_t		refcnt;
	struct timer_list	timer;		/* expire timer		*/
	struct sk_buff		*fragments;
	int			len;
	int			meat;
	int			iif;
	struct timeval		stamp;
	unsigned int		csum;
	__u8			last_in;	/* has first/last segment arrived? */
#define COMPLETE		4
#define FIRST_IN		2
#define LAST_IN			1
	__u16			nhoffset;
	struct frag_queue	**pprev;
};

/* Hash table. */

#define IP6Q_HASHSZ	64

static struct frag_queue *ip6_frag_hash[IP6Q_HASHSZ];
static rwlock_t ip6_frag_lock = RW_LOCK_UNLOCKED;
static u32 ip6_frag_hash_rnd;
static LIST_HEAD(ip6_frag_lru_list);
int ip6_frag_nqueues = 0;

static __inline__ void __fq_unlink(struct frag_queue *fq)
{
	if(fq->next)
		fq->next->pprev = fq->pprev;
	*fq->pprev = fq->next;
	list_del(&fq->lru_list);
	ip6_frag_nqueues--;
}

static __inline__ void fq_unlink(struct frag_queue *fq)
{
	write_lock(&ip6_frag_lock);
	__fq_unlink(fq);
	write_unlock(&ip6_frag_lock);
}

static unsigned int ip6qhashfn(u32 id, struct in6_addr *saddr,
			       struct in6_addr *daddr)
{
	u32 a, b, c;

	a = saddr->s6_addr32[0];
	b = saddr->s6_addr32[1];
	c = saddr->s6_addr32[2];

	a += JHASH_GOLDEN_RATIO;
	b += JHASH_GOLDEN_RATIO;
	c += ip6_frag_hash_rnd;
	__jhash_mix(a, b, c);

	a += saddr->s6_addr32[3];
	b += daddr->s6_addr32[0];
	c += daddr->s6_addr32[1];
	__jhash_mix(a, b, c);

	a += daddr->s6_addr32[2];
	b += daddr->s6_addr32[3];
	c += id;
	__jhash_mix(a, b, c);

	return c & (IP6Q_HASHSZ - 1);
}

static struct timer_list ip6_frag_secret_timer;
int sysctl_ip6frag_secret_interval = 10 * 60 * HZ;

static void ip6_frag_secret_rebuild(unsigned long dummy)
{
	unsigned long now = jiffies;
	int i;

	write_lock(&ip6_frag_lock);
	get_random_bytes(&ip6_frag_hash_rnd, sizeof(u32));
	for (i = 0; i < IP6Q_HASHSZ; i++) {
		struct frag_queue *q;

		q = ip6_frag_hash[i];
		while (q) {
			struct frag_queue *next = q->next;
			unsigned int hval = ip6qhashfn(q->id,
						       &q->saddr,
						       &q->daddr);

			if (hval != i) {
				/* Unlink. */
				if (q->next)
					q->next->pprev = q->pprev;
				*q->pprev = q->next;

				/* Relink to new hash chain. */
				if ((q->next = ip6_frag_hash[hval]) != NULL)
					q->next->pprev = &q->next;
				ip6_frag_hash[hval] = q;
				q->pprev = &ip6_frag_hash[hval];
			}

			q = next;
		}
	}
	write_unlock(&ip6_frag_lock);

	mod_timer(&ip6_frag_secret_timer, now + sysctl_ip6frag_secret_interval);
}

atomic_t ip6_frag_mem = ATOMIC_INIT(0);

/* Memory Tracking Functions. */
static inline void frag_kfree_skb(struct sk_buff *skb, int *work)
{
	if (work)
		*work -= skb->truesize;
	atomic_sub(skb->truesize, &ip6_frag_mem);
	kfree_skb(skb);
}

static inline void frag_free_queue(struct frag_queue *fq, int *work)
{
	if (work)
		*work -= sizeof(struct frag_queue);
	atomic_sub(sizeof(struct frag_queue), &ip6_frag_mem);
	kfree(fq);
}

static inline struct frag_queue *frag_alloc_queue(void)
{
	struct frag_queue *fq = kmalloc(sizeof(struct frag_queue), GFP_ATOMIC);

	if(!fq)
		return NULL;
	atomic_add(sizeof(struct frag_queue), &ip6_frag_mem);
	return fq;
}

/* Destruction primitives. */

/* Complete destruction of fq. */
static void ip6_frag_destroy(struct frag_queue *fq, int *work)
{
	struct sk_buff *fp;

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

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

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

	frag_free_queue(fq, work);
}

static __inline__ void fq_put(struct frag_queue *fq, int *work)
{
	if (atomic_dec_and_test(&fq->refcnt))
		ip6_frag_destroy(fq, work);
}

/* Kill fq entry. It is not destroyed immediately,
 * because caller (and someone more) holds reference count.
 */
static __inline__ void fq_kill(struct frag_queue *fq)
{
	if (del_timer(&fq->timer))
		atomic_dec(&fq->refcnt);

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

static void ip6_evictor(void)
{
	struct frag_queue *fq;
	struct list_head *tmp;
	int work;

	work = atomic_read(&ip6_frag_mem) - sysctl_ip6frag_low_thresh;
	if (work <= 0)
		return;

	while(work > 0) {
		read_lock(&ip6_frag_lock);
		if (list_empty(&ip6_frag_lru_list)) {
			read_unlock(&ip6_frag_lock);
			return;
		}
		tmp = ip6_frag_lru_list.next;
		fq = list_entry(tmp, struct frag_queue, lru_list);
		atomic_inc(&fq->refcnt);
		read_unlock(&ip6_frag_lock);

		spin_lock(&fq->lock);
		if (!(fq->last_in&COMPLETE))
			fq_kill(fq);
		spin_unlock(&fq->lock);

		fq_put(fq, &work);
		IP6_INC_STATS_BH(IPSTATS_MIB_REASMFAILS);
	}
}

static void ip6_frag_expire(unsigned long data)
{
	struct frag_queue *fq = (struct frag_queue *) data;

	spin_lock(&fq->lock);

	if (fq->last_in & COMPLETE)
		goto out;

	fq_kill(fq);

	IP6_INC_STATS_BH(IPSTATS_MIB_REASMTIMEOUT);
	IP6_INC_STATS_BH(IPSTATS_MIB_REASMFAILS);

	/* Send error only if the first segment arrived. */
	if (fq->last_in&FIRST_IN && fq->fragments) {
		struct net_device *dev = dev_get_by_index(fq->iif);

		/*
		   But use as source device on which LAST ARRIVED
		   segment was received. And do not use fq->dev
		   pointer directly, device might already disappeared.
		 */
		if (dev) {
			fq->fragments->dev = dev;
			icmpv6_send(fq->fragments, ICMPV6_TIME_EXCEED, ICMPV6_EXC_FRAGTIME, 0,
				    dev);
			dev_put(dev);
		}
	}
out:
	spin_unlock(&fq->lock);
	fq_put(fq, NULL);
}

/* Creation primitives. */


static struct frag_queue *ip6_frag_intern(unsigned int hash,
					  struct frag_queue *fq_in)
{
	struct frag_queue *fq;

	write_lock(&ip6_frag_lock);
#ifdef CONFIG_SMP
	for (fq = ip6_frag_hash[hash]; fq; fq = fq->next) {
		if (fq->id == fq_in->id && 
		    !ipv6_addr_cmp(&fq_in->saddr, &fq->saddr) &&
		    !ipv6_addr_cmp(&fq_in->daddr, &fq->daddr)) {
			atomic_inc(&fq->refcnt);
			write_unlock(&ip6_frag_lock);
			fq_in->last_in |= COMPLETE;
			fq_put(fq_in, NULL);
			return fq;
		}
	}
#endif
	fq = fq_in;

	if (!mod_timer(&fq->timer, jiffies + sysctl_ip6frag_time))
		atomic_inc(&fq->refcnt);

	atomic_inc(&fq->refcnt);
	if((fq->next = ip6_frag_hash[hash]) != NULL)
		fq->next->pprev = &fq->next;
	ip6_frag_hash[hash] = fq;
	fq->pprev = &ip6_frag_hash[hash];
	INIT_LIST_HEAD(&fq->lru_list);
	list_add_tail(&fq->lru_list, &ip6_frag_lru_list);
	ip6_frag_nqueues++;
	write_unlock(&ip6_frag_lock);
	return fq;
}


static struct frag_queue *
ip6_frag_create(unsigned int hash, u32 id, struct in6_addr *src, struct in6_addr *dst)
{
	struct frag_queue *fq;

	if ((fq = frag_alloc_queue()) == NULL)
		goto oom;

	memset(fq, 0, sizeof(struct frag_queue));

	fq->id = id;
	ipv6_addr_copy(&fq->saddr, src);
	ipv6_addr_copy(&fq->daddr, dst);