dn_route.c

linux 内核源代码

	/*
	 * If we have padding, remove it.
	 */
	if (flags & DN_RT_F_PF) {
		padlen = flags & ~DN_RT_F_PF;
		if (!pskb_may_pull(skb, padlen + 1))
			goto dump_it;
		skb_pull(skb, padlen);
		flags = *skb->data;
	}

	skb_reset_network_header(skb);

	/*
	 * Weed out future version DECnet
	 */
	if (flags & DN_RT_F_VER)
		goto dump_it;

	cb->rt_flags = flags;

	if (decnet_debug_level & 1)
		printk(KERN_DEBUG
			"dn_route_rcv: got 0x%02x from %s [%d %d %d]\n",
			(int)flags, (dev) ? dev->name : "???", len, skb->len,
			padlen);

	if (flags & DN_RT_PKT_CNTL) {
		if (unlikely(skb_linearize(skb)))
			goto dump_it;

		switch(flags & DN_RT_CNTL_MSK) {
			case DN_RT_PKT_INIT:
				dn_dev_init_pkt(skb);
				break;
			case DN_RT_PKT_VERI:
				dn_dev_veri_pkt(skb);
				break;
		}

		if (dn->parms.state != DN_DEV_S_RU)
			goto dump_it;

		switch(flags & DN_RT_CNTL_MSK) {
			case DN_RT_PKT_HELO:
				return NF_HOOK(PF_DECnet, NF_DN_HELLO, skb, skb->dev, NULL, dn_route_ptp_hello);

			case DN_RT_PKT_L1RT:
			case DN_RT_PKT_L2RT:
				return NF_HOOK(PF_DECnet, NF_DN_ROUTE, skb, skb->dev, NULL, dn_route_discard);
			case DN_RT_PKT_ERTH:
				return NF_HOOK(PF_DECnet, NF_DN_HELLO, skb, skb->dev, NULL, dn_neigh_router_hello);

			case DN_RT_PKT_EEDH:
				return NF_HOOK(PF_DECnet, NF_DN_HELLO, skb, skb->dev, NULL, dn_neigh_endnode_hello);
		}
	} else {
		if (dn->parms.state != DN_DEV_S_RU)
			goto dump_it;

		skb_pull(skb, 1); /* Pull flags */

		switch(flags & DN_RT_PKT_MSK) {
			case DN_RT_PKT_LONG:
				return dn_route_rx_long(skb);
			case DN_RT_PKT_SHORT:
				return dn_route_rx_short(skb);
		}
	}

dump_it:
	kfree_skb(skb);
out:
	return NET_RX_DROP;
}

static int dn_output(struct sk_buff *skb)
{
	struct dst_entry *dst = skb->dst;
	struct dn_route *rt = (struct dn_route *)dst;
	struct net_device *dev = dst->dev;
	struct dn_skb_cb *cb = DN_SKB_CB(skb);
	struct neighbour *neigh;

	int err = -EINVAL;

	if ((neigh = dst->neighbour) == NULL)
		goto error;

	skb->dev = dev;

	cb->src = rt->rt_saddr;
	cb->dst = rt->rt_daddr;

	/*
	 * Always set the Intra-Ethernet bit on all outgoing packets
	 * originated on this node. Only valid flag from upper layers
	 * is return-to-sender-requested. Set hop count to 0 too.
	 */
	cb->rt_flags &= ~DN_RT_F_RQR;
	cb->rt_flags |= DN_RT_F_IE;
	cb->hops = 0;

	return NF_HOOK(PF_DECnet, NF_DN_LOCAL_OUT, skb, NULL, dev, neigh->output);

error:
	if (net_ratelimit())
		printk(KERN_DEBUG "dn_output: This should not happen\n");

	kfree_skb(skb);

	return err;
}

static int dn_forward(struct sk_buff *skb)
{
	struct dn_skb_cb *cb = DN_SKB_CB(skb);
	struct dst_entry *dst = skb->dst;
	struct dn_dev *dn_db = dst->dev->dn_ptr;
	struct dn_route *rt;
	struct neighbour *neigh = dst->neighbour;
	int header_len;
#ifdef CONFIG_NETFILTER
	struct net_device *dev = skb->dev;
#endif

	if (skb->pkt_type != PACKET_HOST)
		goto drop;

	/* Ensure that we have enough space for headers */
	rt = (struct dn_route *)skb->dst;
	header_len = dn_db->use_long ? 21 : 6;
	if (skb_cow(skb, LL_RESERVED_SPACE(rt->u.dst.dev)+header_len))
		goto drop;

	/*
	 * Hop count exceeded.
	 */
	if (++cb->hops > 30)
		goto drop;

	skb->dev = rt->u.dst.dev;

	/*
	 * If packet goes out same interface it came in on, then set
	 * the Intra-Ethernet bit. This has no effect for short
	 * packets, so we don't need to test for them here.
	 */
	cb->rt_flags &= ~DN_RT_F_IE;
	if (rt->rt_flags & RTCF_DOREDIRECT)
		cb->rt_flags |= DN_RT_F_IE;

	return NF_HOOK(PF_DECnet, NF_DN_FORWARD, skb, dev, skb->dev, neigh->output);

drop:
	kfree_skb(skb);
	return NET_RX_DROP;
}

/*
 * Drop packet. This is used for endnodes and for
 * when we should not be forwarding packets from
 * this dest.
 */
static int dn_blackhole(struct sk_buff *skb)
{
	kfree_skb(skb);
	return NET_RX_DROP;
}

/*
 * Used to catch bugs. This should never normally get
 * called.
 */
static int dn_rt_bug(struct sk_buff *skb)
{
	if (net_ratelimit()) {
		struct dn_skb_cb *cb = DN_SKB_CB(skb);

		printk(KERN_DEBUG "dn_rt_bug: skb from:%04x to:%04x\n",
				dn_ntohs(cb->src), dn_ntohs(cb->dst));
	}

	kfree_skb(skb);

	return NET_RX_BAD;
}

static int dn_rt_set_next_hop(struct dn_route *rt, struct dn_fib_res *res)
{
	struct dn_fib_info *fi = res->fi;
	struct net_device *dev = rt->u.dst.dev;
	struct neighbour *n;
	unsigned mss;

	if (fi) {
		if (DN_FIB_RES_GW(*res) &&
		    DN_FIB_RES_NH(*res).nh_scope == RT_SCOPE_LINK)
			rt->rt_gateway = DN_FIB_RES_GW(*res);
		memcpy(rt->u.dst.metrics, fi->fib_metrics,
		       sizeof(rt->u.dst.metrics));
	}
	rt->rt_type = res->type;

	if (dev != NULL && rt->u.dst.neighbour == NULL) {
		n = __neigh_lookup_errno(&dn_neigh_table, &rt->rt_gateway, dev);
		if (IS_ERR(n))
			return PTR_ERR(n);
		rt->u.dst.neighbour = n;
	}

	if (rt->u.dst.metrics[RTAX_MTU-1] == 0 ||
	    rt->u.dst.metrics[RTAX_MTU-1] > rt->u.dst.dev->mtu)
		rt->u.dst.metrics[RTAX_MTU-1] = rt->u.dst.dev->mtu;
	mss = dn_mss_from_pmtu(dev, dst_mtu(&rt->u.dst));
	if (rt->u.dst.metrics[RTAX_ADVMSS-1] == 0 ||
	    rt->u.dst.metrics[RTAX_ADVMSS-1] > mss)
		rt->u.dst.metrics[RTAX_ADVMSS-1] = mss;
	return 0;
}

static inline int dn_match_addr(__le16 addr1, __le16 addr2)
{
	__u16 tmp = dn_ntohs(addr1) ^ dn_ntohs(addr2);
	int match = 16;
	while(tmp) {
		tmp >>= 1;
		match--;
	}
	return match;
}

static __le16 dnet_select_source(const struct net_device *dev, __le16 daddr, int scope)
{
	__le16 saddr = 0;
	struct dn_dev *dn_db = dev->dn_ptr;
	struct dn_ifaddr *ifa;
	int best_match = 0;
	int ret;

	read_lock(&dev_base_lock);
	for(ifa = dn_db->ifa_list; ifa; ifa = ifa->ifa_next) {
		if (ifa->ifa_scope > scope)
			continue;
		if (!daddr) {
			saddr = ifa->ifa_local;
			break;
		}
		ret = dn_match_addr(daddr, ifa->ifa_local);
		if (ret > best_match)
			saddr = ifa->ifa_local;
		if (best_match == 0)
			saddr = ifa->ifa_local;
	}
	read_unlock(&dev_base_lock);

	return saddr;
}

static inline __le16 __dn_fib_res_prefsrc(struct dn_fib_res *res)
{
	return dnet_select_source(DN_FIB_RES_DEV(*res), DN_FIB_RES_GW(*res), res->scope);
}

static inline __le16 dn_fib_rules_map_destination(__le16 daddr, struct dn_fib_res *res)
{
	__le16 mask = dnet_make_mask(res->prefixlen);
	return (daddr&~mask)|res->fi->fib_nh->nh_gw;
}

static int dn_route_output_slow(struct dst_entry **pprt, const struct flowi *oldflp, int try_hard)
{
	struct flowi fl = { .nl_u = { .dn_u =
				      { .daddr = oldflp->fld_dst,
					.saddr = oldflp->fld_src,
					.scope = RT_SCOPE_UNIVERSE,
				     } },
			    .mark = oldflp->mark,
			    .iif = init_net.loopback_dev->ifindex,
			    .oif = oldflp->oif };
	struct dn_route *rt = NULL;
	struct net_device *dev_out = NULL, *dev;
	struct neighbour *neigh = NULL;
	unsigned hash;
	unsigned flags = 0;
	struct dn_fib_res res = { .fi = NULL, .type = RTN_UNICAST };
	int err;
	int free_res = 0;
	__le16 gateway = 0;

	if (decnet_debug_level & 16)
		printk(KERN_DEBUG
		       "dn_route_output_slow: dst=%04x src=%04x mark=%d"
		       " iif=%d oif=%d\n", dn_ntohs(oldflp->fld_dst),
		       dn_ntohs(oldflp->fld_src),
		       oldflp->mark, init_net.loopback_dev->ifindex, oldflp->oif);

	/* If we have an output interface, verify its a DECnet device */
	if (oldflp->oif) {
		dev_out = dev_get_by_index(&init_net, oldflp->oif);
		err = -ENODEV;
		if (dev_out && dev_out->dn_ptr == NULL) {
			dev_put(dev_out);
			dev_out = NULL;
		}
		if (dev_out == NULL)
			goto out;
	}

	/* If we have a source address, verify that its a local address */
	if (oldflp->fld_src) {
		err = -EADDRNOTAVAIL;

		if (dev_out) {
			if (dn_dev_islocal(dev_out, oldflp->fld_src))
				goto source_ok;
			dev_put(dev_out);
			goto out;
		}
		read_lock(&dev_base_lock);
		for_each_netdev(&init_net, dev) {
			if (!dev->dn_ptr)
				continue;
			if (!dn_dev_islocal(dev, oldflp->fld_src))
				continue;
			if ((dev->flags & IFF_LOOPBACK) &&
			    oldflp->fld_dst &&
			    !dn_dev_islocal(dev, oldflp->fld_dst))
				continue;

			dev_out = dev;
			break;
		}
		read_unlock(&dev_base_lock);
		if (dev_out == NULL)
			goto out;
		dev_hold(dev_out);
source_ok:
		;
	}

	/* No destination? Assume its local */
	if (!fl.fld_dst) {
		fl.fld_dst = fl.fld_src;

		err = -EADDRNOTAVAIL;
		if (dev_out)
			dev_put(dev_out);
		dev_out = init_net.loopback_dev;
		dev_hold(dev_out);
		if (!fl.fld_dst) {
			fl.fld_dst =
			fl.fld_src = dnet_select_source(dev_out, 0,
						       RT_SCOPE_HOST);
			if (!fl.fld_dst)
				goto out;
		}
		fl.oif = init_net.loopback_dev->ifindex;
		res.type = RTN_LOCAL;
		goto make_route;
	}

	if (decnet_debug_level & 16)
		printk(KERN_DEBUG
		       "dn_route_output_slow: initial checks complete."
		       " dst=%o4x src=%04x oif=%d try_hard=%d\n",
		       dn_ntohs(fl.fld_dst), dn_ntohs(fl.fld_src),
		       fl.oif, try_hard);

	/*
	 * N.B. If the kernel is compiled without router support then
	 * dn_fib_lookup() will evaluate to non-zero so this if () block
	 * will always be executed.
	 */
	err = -ESRCH;
	if (try_hard || (err = dn_fib_lookup(&fl, &res)) != 0) {
		struct dn_dev *dn_db;
		if (err != -ESRCH)
			goto out;
		/*
		 * Here the fallback is basically the standard algorithm for
		 * routing in endnodes which is described in the DECnet routing
		 * docs
		 *
		 * If we are not trying hard, look in neighbour cache.
		 * The result is tested to ensure that if a specific output
		 * device/source address was requested, then we honour that
		 * here
		 */
		if (!try_hard) {
			neigh = neigh_lookup_nodev(&dn_neigh_table, &fl.fld_dst);
			if (neigh) {
				if ((oldflp->oif &&
				    (neigh->dev->ifindex != oldflp->oif)) ||
				    (oldflp->fld_src &&
				    (!dn_dev_islocal(neigh->dev,
						      oldflp->fld_src)))) {
					neigh_release(neigh);
					neigh = NULL;
				} else {
					if (dev_out)
						dev_put(dev_out);
					if (dn_dev_islocal(neigh->dev, fl.fld_dst)) {
						dev_out = init_net.loopback_dev;
						res.type = RTN_LOCAL;
					} else {
						dev_out = neigh->dev;
					}
					dev_hold(dev_out);
					goto select_source;
				}
			}
		}

		/* Not there? Perhaps its a local address */
		if (dev_out == NULL)
			dev_out = dn_dev_get_default();
		err = -ENODEV;
		if (dev_out == NULL)
			goto out;
		dn_db = dev_out->dn_ptr;
		/* Possible improvement - check all devices for local addr */
		if (dn_dev_islocal(dev_out, fl.fld_dst)) {
			dev_put(dev_out);
			dev_out = init_net.loopback_dev;
			dev_hold(dev_out);
			res.type = RTN_LOCAL;
			goto select_source;
		}
		/* Not local either.... try sending it to the default router */
		neigh = neigh_clone(dn_db->router);
		BUG_ON(neigh && neigh->dev != dev_out);

		/* Ok then, we assume its directly connected and move on */
select_source:
		if (neigh)
			gateway = ((struct dn_neigh *)neigh)->addr;
		if (gateway == 0)
			gateway = fl.fld_dst;
		if (fl.fld_src == 0) {
			fl.fld_src = dnet_select_source(dev_out, gateway,
							 res.type == RTN_LOCAL ?
							 RT_SCOPE_HOST :
							 RT_SCOPE_LINK);
			if (fl.fld_src == 0 && res.type != RTN_LOCAL)
				goto e_addr;
		}
		fl.oif = dev_out->ifindex;
		goto make_route;
	}
	free_res = 1;

	if (res.type == RTN_NAT)
		goto e_inval;

	if (res.type == RTN_LOCAL) {
		if (!fl.fld_src)
			fl.fld_src = fl.fld_dst;
		if (dev_out)
			dev_put(dev_out);
		dev_out = init_net.loopback_dev;
		dev_hold(dev_out);
		fl.oif = dev_out->ifindex;
		if (res.fi)
			dn_fib_info_put(res.fi);
		res.fi = NULL;
		goto make_route;
	}

	if (res.fi->fib_nhs > 1 && fl.oif == 0)
		dn_fib_select_multipath(&fl, &res);

	/*
	 * We could add some logic to deal with default routes here and
	 * get rid of some of the special casing above.
	 */

	if (!fl.fld_src)
		fl.fld_src = DN_FIB_RES_PREFSRC(res);

	if (dev_out)
		dev_put(dev_out);
	dev_out = DN_FIB_RES_DEV(res);
	dev_hold(dev_out);
	fl.oif = dev_out->ifindex;
	gateway = DN_FIB_RES_GW(res);

make_route:
	if (dev_out->flags & IFF_LOOPBACK)
		flags |= RTCF_LOCAL;

	rt = dst_alloc(&dn_dst_ops);
	if (rt == NULL)
		goto e_nobufs;

	atomic_set(&rt->u.dst.__refcnt, 1);
	rt->u.dst.flags   = DST_HOST;

	rt->fl.fld_src    = oldflp->fld_src;
	rt->fl.fld_dst    = oldflp->fld_dst;
	rt->fl.oif        = oldflp->oif;
	rt->fl.iif        = 0;
	rt->fl.mark       = oldflp->mark;

	rt->rt_saddr      = fl.fld_src;
	rt->rt_daddr      = fl.fld_dst;
	rt->rt_gateway    = gateway ? gateway : fl.fld_dst;
	rt->rt_local_src  = fl.fld_src;

	rt->rt_dst_map    = fl.fld_dst;
	rt->rt_src_map    = fl.fld_src;

	rt->u.dst.dev = dev_out;
	dev_hold(dev_out);
	rt->u.dst.neighbour = neigh;
	neigh = NULL;

	rt->u.dst.lastuse = jiffies;
	rt->u.dst.output  = dn_output;
	rt->u.dst.input   = dn_rt_bug;
	rt->rt_flags      = flags;
	if (flags & RTCF_LOCAL)
		rt->u.dst.input = dn_nsp_rx;

	err = dn_rt_set_next_hop(rt, &res);
	if (err)
		goto e_neighbour;

	hash = dn_hash(rt->fl.fld_src, rt->fl.fld_dst);
	dn_insert_route(rt, hash, (struct dn_route **)pprt);

done:
	if (neigh)
		neigh_release(neigh);
	if (free_res)
		dn_fib_res_put(&res);
	if (dev_out)
		dev_put(dev_out);
out:
	return err;

e_addr:
	err = -EADDRNOTAVAIL;
	goto done;
e_inval:
	err = -EINVAL;
	goto done;
e_nobufs:
	err = -ENOBUFS;
	goto done;
e_neighbour:
	dst_free(&rt->u.dst);
	goto e_nobufs;
}


/*
 * N.B. The flags may be moved into the flowi at some future stage.
 */
static int __dn_route_output_key(struct dst_entry **pprt, const struct flowi *flp, int flags)
{
	unsigned hash = dn_hash(flp->fld_src, flp->fld_dst);
	struct dn_route *rt = NULL;

	if (!(flags & MSG_TRYHARD)) {
		rcu_read_lock_bh();
		for(rt = rcu_dereference(dn_rt_hash_table[hash].chain); rt;
			rt = rcu_dereference(rt->u.dst.dn_next)) {
			if ((flp->fld_dst == rt->fl.fld_dst) &&
			    (flp->fld_src == rt->fl.fld_src) &&
			    (flp->mark == rt->fl.mark) &&
			    (rt->fl.iif == 0) &&
			    (rt->fl.oif == flp->oif)) {
				dst_use(&rt->u.dst, jiffies);
				rcu_read_unlock_bh();
				*pprt = &rt->u.dst;
				return 0;
			}
		}
		rcu_read_unlock_bh();
	}

	return dn_route_output_slow(pprt, flp, flags);
}

static int dn_route_output_key(struct dst_entry **pprt, struct flowi *flp, int flags)
{
	int err;

	err = __dn_route_output_key(pprt, flp, flags);
	if (err == 0 && flp->proto) {
		err = xfrm_lookup(pprt, flp, NULL, 0);
	}
	return err;
}

int dn_route_output_sock(struct dst_entry **pprt, struct flowi *fl, struct sock *sk, int flags)
{
	int err;

	err = __dn_route_output_key(pprt, fl, flags & MSG_TRYHARD);
	if (err == 0 && fl->proto) {
		err = xfrm_lookup(pprt, fl, sk, !(flags & MSG_DONTWAIT));
	}
	return err;