dn_route.c

linux 内核源代码

/*
 * DECnet       An implementation of the DECnet protocol suite for the LINUX
 *              operating system.  DECnet is implemented using the  BSD Socket
 *              interface as the means of communication with the user level.
 *
 *              DECnet Routing Functions (Endnode and Router)
 *
 * Authors:     Steve Whitehouse <[e-mail]>
 *              Eduardo Marcelo Serrat <[e-mail]>
 *
 * Changes:
 *              Steve Whitehouse : Fixes to allow "intra-ethernet" and
 *                                 "return-to-sender" bits on outgoing
 *                                 packets.
 *		Steve Whitehouse : Timeouts for cached routes.
 *              Steve Whitehouse : Use dst cache for input routes too.
 *              Steve Whitehouse : Fixed error values in dn_send_skb.
 *              Steve Whitehouse : Rework routing functions to better fit
 *                                 DECnet routing design
 *              Alexey Kuznetsov : New SMP locking
 *              Steve Whitehouse : More SMP locking changes & dn_cache_dump()
 *              Steve Whitehouse : Prerouting NF hook, now really is prerouting.
 *				   Fixed possible skb leak in rtnetlink funcs.
 *              Steve Whitehouse : Dave Miller's dynamic hash table sizing and
 *                                 Alexey Kuznetsov's finer grained locking
 *                                 from ipv4/route.c.
 *              Steve Whitehouse : Routing is now starting to look like a
 *                                 sensible set of code now, mainly due to
 *                                 my copying the IPv4 routing code. The
 *                                 hooks here are modified and will continue
 *                                 to evolve for a while.
 *              Steve Whitehouse : Real SMP at last :-) Also new netfilter
 *                                 stuff. Look out raw sockets your days
 *                                 are numbered!
 *              Steve Whitehouse : Added return-to-sender functions. Added
 *                                 backlog congestion level return codes.
 *		Steve Whitehouse : Fixed bug where routes were set up with
 *                                 no ref count on net devices.
 *              Steve Whitehouse : RCU for the route cache
 *              Steve Whitehouse : Preparations for the flow cache
 *              Steve Whitehouse : Prepare for nonlinear skbs
 */

/******************************************************************************
    (c) 1995-1998 E.M. Serrat		[e-mail]

    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
    any later version.

    This program is distributed in the hope that it will be useful,
    but WITHOUT ANY WARRANTY; without even the implied warranty of
    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
    GNU General Public License for more details.
*******************************************************************************/

#include <linux/errno.h>
#include <linux/types.h>
#include <linux/socket.h>
#include <linux/in.h>
#include <linux/kernel.h>
#include <linux/sockios.h>
#include <linux/net.h>
#include <linux/netdevice.h>
#include <linux/inet.h>
#include <linux/route.h>
#include <linux/in_route.h>
#include <net/sock.h>
#include <linux/mm.h>
#include <linux/proc_fs.h>
#include <linux/seq_file.h>
#include <linux/init.h>
#include <linux/rtnetlink.h>
#include <linux/string.h>
#include <linux/netfilter_decnet.h>
#include <linux/rcupdate.h>
#include <linux/times.h>
#include <asm/errno.h>
#include <net/net_namespace.h>
#include <net/netlink.h>
#include <net/neighbour.h>
#include <net/dst.h>
#include <net/flow.h>
#include <net/fib_rules.h>
#include <net/dn.h>
#include <net/dn_dev.h>
#include <net/dn_nsp.h>
#include <net/dn_route.h>
#include <net/dn_neigh.h>
#include <net/dn_fib.h>

struct dn_rt_hash_bucket
{
	struct dn_route *chain;
	spinlock_t lock;
} __attribute__((__aligned__(8)));

extern struct neigh_table dn_neigh_table;


static unsigned char dn_hiord_addr[6] = {0xAA,0x00,0x04,0x00,0x00,0x00};

static const int dn_rt_min_delay = 2 * HZ;
static const int dn_rt_max_delay = 10 * HZ;
static const int dn_rt_mtu_expires = 10 * 60 * HZ;

static unsigned long dn_rt_deadline;

static int dn_dst_gc(void);
static struct dst_entry *dn_dst_check(struct dst_entry *, __u32);
static struct dst_entry *dn_dst_negative_advice(struct dst_entry *);
static void dn_dst_link_failure(struct sk_buff *);
static void dn_dst_update_pmtu(struct dst_entry *dst, u32 mtu);
static int dn_route_input(struct sk_buff *);
static void dn_run_flush(unsigned long dummy);

static struct dn_rt_hash_bucket *dn_rt_hash_table;
static unsigned dn_rt_hash_mask;

static struct timer_list dn_route_timer;
static DEFINE_TIMER(dn_rt_flush_timer, dn_run_flush, 0, 0);
int decnet_dst_gc_interval = 2;

static struct dst_ops dn_dst_ops = {
	.family =		PF_DECnet,
	.protocol =		__constant_htons(ETH_P_DNA_RT),
	.gc_thresh =		128,
	.gc =			dn_dst_gc,
	.check =		dn_dst_check,
	.negative_advice =	dn_dst_negative_advice,
	.link_failure =		dn_dst_link_failure,
	.update_pmtu =		dn_dst_update_pmtu,
	.entry_size =		sizeof(struct dn_route),
	.entries =		ATOMIC_INIT(0),
};

static __inline__ unsigned dn_hash(__le16 src, __le16 dst)
{
	__u16 tmp = (__u16 __force)(src ^ dst);
	tmp ^= (tmp >> 3);
	tmp ^= (tmp >> 5);
	tmp ^= (tmp >> 10);
	return dn_rt_hash_mask & (unsigned)tmp;
}

static inline void dnrt_free(struct dn_route *rt)
{
	call_rcu_bh(&rt->u.dst.rcu_head, dst_rcu_free);
}

static inline void dnrt_drop(struct dn_route *rt)
{
	dst_release(&rt->u.dst);
	call_rcu_bh(&rt->u.dst.rcu_head, dst_rcu_free);
}

static void dn_dst_check_expire(unsigned long dummy)
{
	int i;
	struct dn_route *rt, **rtp;
	unsigned long now = jiffies;
	unsigned long expire = 120 * HZ;

	for(i = 0; i <= dn_rt_hash_mask; i++) {
		rtp = &dn_rt_hash_table[i].chain;

		spin_lock(&dn_rt_hash_table[i].lock);
		while((rt=*rtp) != NULL) {
			if (atomic_read(&rt->u.dst.__refcnt) ||
					(now - rt->u.dst.lastuse) < expire) {
				rtp = &rt->u.dst.dn_next;
				continue;
			}
			*rtp = rt->u.dst.dn_next;
			rt->u.dst.dn_next = NULL;
			dnrt_free(rt);
		}
		spin_unlock(&dn_rt_hash_table[i].lock);

		if ((jiffies - now) > 0)
			break;
	}

	mod_timer(&dn_route_timer, now + decnet_dst_gc_interval * HZ);
}

static int dn_dst_gc(void)
{
	struct dn_route *rt, **rtp;
	int i;
	unsigned long now = jiffies;
	unsigned long expire = 10 * HZ;

	for(i = 0; i <= dn_rt_hash_mask; i++) {

		spin_lock_bh(&dn_rt_hash_table[i].lock);
		rtp = &dn_rt_hash_table[i].chain;

		while((rt=*rtp) != NULL) {
			if (atomic_read(&rt->u.dst.__refcnt) ||
					(now - rt->u.dst.lastuse) < expire) {
				rtp = &rt->u.dst.dn_next;
				continue;
			}
			*rtp = rt->u.dst.dn_next;
			rt->u.dst.dn_next = NULL;
			dnrt_drop(rt);
			break;
		}
		spin_unlock_bh(&dn_rt_hash_table[i].lock);
	}

	return 0;
}

/*
 * The decnet standards don't impose a particular minimum mtu, what they
 * do insist on is that the routing layer accepts a datagram of at least
 * 230 bytes long. Here we have to subtract the routing header length from
 * 230 to get the minimum acceptable mtu. If there is no neighbour, then we
 * assume the worst and use a long header size.
 *
 * We update both the mtu and the advertised mss (i.e. the segment size we
 * advertise to the other end).
 */
static void dn_dst_update_pmtu(struct dst_entry *dst, u32 mtu)
{
	u32 min_mtu = 230;
	struct dn_dev *dn = dst->neighbour ?
			    (struct dn_dev *)dst->neighbour->dev->dn_ptr : NULL;

	if (dn && dn->use_long == 0)
		min_mtu -= 6;
	else
		min_mtu -= 21;

	if (dst->metrics[RTAX_MTU-1] > mtu && mtu >= min_mtu) {
		if (!(dst_metric_locked(dst, RTAX_MTU))) {
			dst->metrics[RTAX_MTU-1] = mtu;
			dst_set_expires(dst, dn_rt_mtu_expires);
		}
		if (!(dst_metric_locked(dst, RTAX_ADVMSS))) {
			u32 mss = mtu - DN_MAX_NSP_DATA_HEADER;
			if (dst->metrics[RTAX_ADVMSS-1] > mss)
				dst->metrics[RTAX_ADVMSS-1] = mss;
		}
	}
}

/*
 * When a route has been marked obsolete. (e.g. routing cache flush)
 */
static struct dst_entry *dn_dst_check(struct dst_entry *dst, __u32 cookie)
{
	return NULL;
}

static struct dst_entry *dn_dst_negative_advice(struct dst_entry *dst)
{
	dst_release(dst);
	return NULL;
}

static void dn_dst_link_failure(struct sk_buff *skb)
{
	return;
}

static inline int compare_keys(struct flowi *fl1, struct flowi *fl2)
{
	return ((fl1->nl_u.dn_u.daddr ^ fl2->nl_u.dn_u.daddr) |
		(fl1->nl_u.dn_u.saddr ^ fl2->nl_u.dn_u.saddr) |
		(fl1->mark ^ fl2->mark) |
		(fl1->nl_u.dn_u.scope ^ fl2->nl_u.dn_u.scope) |
		(fl1->oif ^ fl2->oif) |
		(fl1->iif ^ fl2->iif)) == 0;
}

static int dn_insert_route(struct dn_route *rt, unsigned hash, struct dn_route **rp)
{
	struct dn_route *rth, **rthp;
	unsigned long now = jiffies;

	rthp = &dn_rt_hash_table[hash].chain;

	spin_lock_bh(&dn_rt_hash_table[hash].lock);
	while((rth = *rthp) != NULL) {
		if (compare_keys(&rth->fl, &rt->fl)) {
			/* Put it first */
			*rthp = rth->u.dst.dn_next;
			rcu_assign_pointer(rth->u.dst.dn_next,
					   dn_rt_hash_table[hash].chain);
			rcu_assign_pointer(dn_rt_hash_table[hash].chain, rth);

			dst_use(&rth->u.dst, now);
			spin_unlock_bh(&dn_rt_hash_table[hash].lock);

			dnrt_drop(rt);
			*rp = rth;
			return 0;
		}
		rthp = &rth->u.dst.dn_next;
	}

	rcu_assign_pointer(rt->u.dst.dn_next, dn_rt_hash_table[hash].chain);
	rcu_assign_pointer(dn_rt_hash_table[hash].chain, rt);

	dst_use(&rt->u.dst, now);
	spin_unlock_bh(&dn_rt_hash_table[hash].lock);
	*rp = rt;
	return 0;
}

void dn_run_flush(unsigned long dummy)
{
	int i;
	struct dn_route *rt, *next;

	for(i = 0; i < dn_rt_hash_mask; i++) {
		spin_lock_bh(&dn_rt_hash_table[i].lock);

		if ((rt = xchg(&dn_rt_hash_table[i].chain, NULL)) == NULL)
			goto nothing_to_declare;

		for(; rt; rt=next) {
			next = rt->u.dst.dn_next;
			rt->u.dst.dn_next = NULL;
			dst_free((struct dst_entry *)rt);
		}

nothing_to_declare:
		spin_unlock_bh(&dn_rt_hash_table[i].lock);
	}
}

static DEFINE_SPINLOCK(dn_rt_flush_lock);

void dn_rt_cache_flush(int delay)
{
	unsigned long now = jiffies;
	int user_mode = !in_interrupt();

	if (delay < 0)
		delay = dn_rt_min_delay;

	spin_lock_bh(&dn_rt_flush_lock);

	if (del_timer(&dn_rt_flush_timer) && delay > 0 && dn_rt_deadline) {
		long tmo = (long)(dn_rt_deadline - now);

		if (user_mode && tmo < dn_rt_max_delay - dn_rt_min_delay)
			tmo = 0;

		if (delay > tmo)
			delay = tmo;
	}

	if (delay <= 0) {
		spin_unlock_bh(&dn_rt_flush_lock);
		dn_run_flush(0);
		return;
	}

	if (dn_rt_deadline == 0)
		dn_rt_deadline = now + dn_rt_max_delay;

	dn_rt_flush_timer.expires = now + delay;
	add_timer(&dn_rt_flush_timer);
	spin_unlock_bh(&dn_rt_flush_lock);
}

/**
 * dn_return_short - Return a short packet to its sender
 * @skb: The packet to return
 *
 */
static int dn_return_short(struct sk_buff *skb)
{
	struct dn_skb_cb *cb;
	unsigned char *ptr;
	__le16 *src;
	__le16 *dst;
	__le16 tmp;

	/* Add back headers */
	skb_push(skb, skb->data - skb_network_header(skb));

	if ((skb = skb_unshare(skb, GFP_ATOMIC)) == NULL)
		return NET_RX_DROP;

	cb = DN_SKB_CB(skb);
	/* Skip packet length and point to flags */
	ptr = skb->data + 2;
	*ptr++ = (cb->rt_flags & ~DN_RT_F_RQR) | DN_RT_F_RTS;

	dst = (__le16 *)ptr;
	ptr += 2;
	src = (__le16 *)ptr;
	ptr += 2;
	*ptr = 0; /* Zero hop count */

	/* Swap source and destination */
	tmp  = *src;
	*src = *dst;
	*dst = tmp;

	skb->pkt_type = PACKET_OUTGOING;
	dn_rt_finish_output(skb, NULL, NULL);
	return NET_RX_SUCCESS;
}

/**
 * dn_return_long - Return a long packet to its sender
 * @skb: The long format packet to return
 *
 */
static int dn_return_long(struct sk_buff *skb)
{
	struct dn_skb_cb *cb;
	unsigned char *ptr;
	unsigned char *src_addr, *dst_addr;
	unsigned char tmp[ETH_ALEN];

	/* Add back all headers */
	skb_push(skb, skb->data - skb_network_header(skb));

	if ((skb = skb_unshare(skb, GFP_ATOMIC)) == NULL)
		return NET_RX_DROP;

	cb = DN_SKB_CB(skb);
	/* Ignore packet length and point to flags */
	ptr = skb->data + 2;

	/* Skip padding */
	if (*ptr & DN_RT_F_PF) {
		char padlen = (*ptr & ~DN_RT_F_PF);
		ptr += padlen;
	}

	*ptr++ = (cb->rt_flags & ~DN_RT_F_RQR) | DN_RT_F_RTS;
	ptr += 2;
	dst_addr = ptr;
	ptr += 8;
	src_addr = ptr;
	ptr += 6;
	*ptr = 0; /* Zero hop count */

	/* Swap source and destination */
	memcpy(tmp, src_addr, ETH_ALEN);
	memcpy(src_addr, dst_addr, ETH_ALEN);
	memcpy(dst_addr, tmp, ETH_ALEN);

	skb->pkt_type = PACKET_OUTGOING;
	dn_rt_finish_output(skb, dst_addr, src_addr);
	return NET_RX_SUCCESS;
}

/**
 * dn_route_rx_packet - Try and find a route for an incoming packet
 * @skb: The packet to find a route for
 *
 * Returns: result of input function if route is found, error code otherwise
 */
static int dn_route_rx_packet(struct sk_buff *skb)
{
	struct dn_skb_cb *cb = DN_SKB_CB(skb);
	int err;

	if ((err = dn_route_input(skb)) == 0)
		return dst_input(skb);

	if (decnet_debug_level & 4) {
		char *devname = skb->dev ? skb->dev->name : "???";
		struct dn_skb_cb *cb = DN_SKB_CB(skb);
		printk(KERN_DEBUG
			"DECnet: dn_route_rx_packet: rt_flags=0x%02x dev=%s len=%d src=0x%04hx dst=0x%04hx err=%d type=%d\n",
			(int)cb->rt_flags, devname, skb->len,
			dn_ntohs(cb->src), dn_ntohs(cb->dst),
			err, skb->pkt_type);
	}

	if ((skb->pkt_type == PACKET_HOST) && (cb->rt_flags & DN_RT_F_RQR)) {
		switch(cb->rt_flags & DN_RT_PKT_MSK) {
			case DN_RT_PKT_SHORT:
				return dn_return_short(skb);
			case DN_RT_PKT_LONG:
				return dn_return_long(skb);
		}
	}

	kfree_skb(skb);
	return NET_RX_DROP;
}

static int dn_route_rx_long(struct sk_buff *skb)
{
	struct dn_skb_cb *cb = DN_SKB_CB(skb);
	unsigned char *ptr = skb->data;

	if (!pskb_may_pull(skb, 21)) /* 20 for long header, 1 for shortest nsp */
		goto drop_it;

	skb_pull(skb, 20);
	skb_reset_transport_header(skb);

	/* Destination info */
	ptr += 2;
	cb->dst = dn_eth2dn(ptr);
	if (memcmp(ptr, dn_hiord_addr, 4) != 0)
		goto drop_it;
	ptr += 6;


	/* Source info */
	ptr += 2;
	cb->src = dn_eth2dn(ptr);
	if (memcmp(ptr, dn_hiord_addr, 4) != 0)
		goto drop_it;
	ptr += 6;
	/* Other junk */
	ptr++;
	cb->hops = *ptr++; /* Visit Count */

	return NF_HOOK(PF_DECnet, NF_DN_PRE_ROUTING, skb, skb->dev, NULL, dn_route_rx_packet);

drop_it:
	kfree_skb(skb);
	return NET_RX_DROP;
}



static int dn_route_rx_short(struct sk_buff *skb)
{
	struct dn_skb_cb *cb = DN_SKB_CB(skb);
	unsigned char *ptr = skb->data;

	if (!pskb_may_pull(skb, 6)) /* 5 for short header + 1 for shortest nsp */
		goto drop_it;

	skb_pull(skb, 5);
	skb_reset_transport_header(skb);

	cb->dst = *(__le16 *)ptr;
	ptr += 2;
	cb->src = *(__le16 *)ptr;
	ptr += 2;
	cb->hops = *ptr & 0x3f;

	return NF_HOOK(PF_DECnet, NF_DN_PRE_ROUTING, skb, skb->dev, NULL, dn_route_rx_packet);

drop_it:
	kfree_skb(skb);
	return NET_RX_DROP;
}

static int dn_route_discard(struct sk_buff *skb)
{
	/*
	 * I know we drop the packet here, but thats considered success in
	 * this case
	 */
	kfree_skb(skb);
	return NET_RX_SUCCESS;
}

static int dn_route_ptp_hello(struct sk_buff *skb)
{
	dn_dev_hello(skb);
	dn_neigh_pointopoint_hello(skb);
	return NET_RX_SUCCESS;
}

int dn_route_rcv(struct sk_buff *skb, struct net_device *dev, struct packet_type *pt, struct net_device *orig_dev)
{
	struct dn_skb_cb *cb;
	unsigned char flags = 0;
	__u16 len = dn_ntohs(*(__le16 *)skb->data);
	struct dn_dev *dn = (struct dn_dev *)dev->dn_ptr;
	unsigned char padlen = 0;

	if (dev->nd_net != &init_net)
		goto dump_it;

	if (dn == NULL)
		goto dump_it;

	if ((skb = skb_share_check(skb, GFP_ATOMIC)) == NULL)
		goto out;

	if (!pskb_may_pull(skb, 3))
		goto dump_it;

	skb_pull(skb, 2);

	if (len > skb->len)
		goto dump_it;

	skb_trim(skb, len);

	flags = *skb->data;

	cb = DN_SKB_CB(skb);
	cb->stamp = jiffies;
	cb->iif = dev->ifindex;