ipmr.c

linux 内核源代码

ipmr_cache_unresolved(vifi_t vifi, struct sk_buff *skb)
{
	int err;
	struct mfc_cache *c;
	const struct iphdr *iph = ip_hdr(skb);

	spin_lock_bh(&mfc_unres_lock);
	for (c=mfc_unres_queue; c; c=c->next) {
		if (c->mfc_mcastgrp == iph->daddr &&
		    c->mfc_origin == iph->saddr)
			break;
	}

	if (c == NULL) {
		/*
		 *	Create a new entry if allowable
		 */

		if (atomic_read(&cache_resolve_queue_len)>=10 ||
		    (c=ipmr_cache_alloc_unres())==NULL) {
			spin_unlock_bh(&mfc_unres_lock);

			kfree_skb(skb);
			return -ENOBUFS;
		}

		/*
		 *	Fill in the new cache entry
		 */
		c->mfc_parent	= -1;
		c->mfc_origin	= iph->saddr;
		c->mfc_mcastgrp	= iph->daddr;

		/*
		 *	Reflect first query at mrouted.
		 */
		if ((err = ipmr_cache_report(skb, vifi, IGMPMSG_NOCACHE))<0) {
			/* If the report failed throw the cache entry
			   out - Brad Parker
			 */
			spin_unlock_bh(&mfc_unres_lock);

			kmem_cache_free(mrt_cachep, c);
			kfree_skb(skb);
			return err;
		}

		atomic_inc(&cache_resolve_queue_len);
		c->next = mfc_unres_queue;
		mfc_unres_queue = c;

		mod_timer(&ipmr_expire_timer, c->mfc_un.unres.expires);
	}

	/*
	 *	See if we can append the packet
	 */
	if (c->mfc_un.unres.unresolved.qlen>3) {
		kfree_skb(skb);
		err = -ENOBUFS;
	} else {
		skb_queue_tail(&c->mfc_un.unres.unresolved,skb);
		err = 0;
	}

	spin_unlock_bh(&mfc_unres_lock);
	return err;
}

/*
 *	MFC cache manipulation by user space mroute daemon
 */

static int ipmr_mfc_delete(struct mfcctl *mfc)
{
	int line;
	struct mfc_cache *c, **cp;

	line=MFC_HASH(mfc->mfcc_mcastgrp.s_addr, mfc->mfcc_origin.s_addr);

	for (cp=&mfc_cache_array[line]; (c=*cp) != NULL; cp = &c->next) {
		if (c->mfc_origin == mfc->mfcc_origin.s_addr &&
		    c->mfc_mcastgrp == mfc->mfcc_mcastgrp.s_addr) {
			write_lock_bh(&mrt_lock);
			*cp = c->next;
			write_unlock_bh(&mrt_lock);

			kmem_cache_free(mrt_cachep, c);
			return 0;
		}
	}
	return -ENOENT;
}

static int ipmr_mfc_add(struct mfcctl *mfc, int mrtsock)
{
	int line;
	struct mfc_cache *uc, *c, **cp;

	line=MFC_HASH(mfc->mfcc_mcastgrp.s_addr, mfc->mfcc_origin.s_addr);

	for (cp=&mfc_cache_array[line]; (c=*cp) != NULL; cp = &c->next) {
		if (c->mfc_origin == mfc->mfcc_origin.s_addr &&
		    c->mfc_mcastgrp == mfc->mfcc_mcastgrp.s_addr)
			break;
	}

	if (c != NULL) {
		write_lock_bh(&mrt_lock);
		c->mfc_parent = mfc->mfcc_parent;
		ipmr_update_thresholds(c, mfc->mfcc_ttls);
		if (!mrtsock)
			c->mfc_flags |= MFC_STATIC;
		write_unlock_bh(&mrt_lock);
		return 0;
	}

	if (!MULTICAST(mfc->mfcc_mcastgrp.s_addr))
		return -EINVAL;

	c=ipmr_cache_alloc();
	if (c==NULL)
		return -ENOMEM;

	c->mfc_origin=mfc->mfcc_origin.s_addr;
	c->mfc_mcastgrp=mfc->mfcc_mcastgrp.s_addr;
	c->mfc_parent=mfc->mfcc_parent;
	ipmr_update_thresholds(c, mfc->mfcc_ttls);
	if (!mrtsock)
		c->mfc_flags |= MFC_STATIC;

	write_lock_bh(&mrt_lock);
	c->next = mfc_cache_array[line];
	mfc_cache_array[line] = c;
	write_unlock_bh(&mrt_lock);

	/*
	 *	Check to see if we resolved a queued list. If so we
	 *	need to send on the frames and tidy up.
	 */
	spin_lock_bh(&mfc_unres_lock);
	for (cp = &mfc_unres_queue; (uc=*cp) != NULL;
	     cp = &uc->next) {
		if (uc->mfc_origin == c->mfc_origin &&
		    uc->mfc_mcastgrp == c->mfc_mcastgrp) {
			*cp = uc->next;
			if (atomic_dec_and_test(&cache_resolve_queue_len))
				del_timer(&ipmr_expire_timer);
			break;
		}
	}
	spin_unlock_bh(&mfc_unres_lock);

	if (uc) {
		ipmr_cache_resolve(uc, c);
		kmem_cache_free(mrt_cachep, uc);
	}
	return 0;
}

/*
 *	Close the multicast socket, and clear the vif tables etc
 */

static void mroute_clean_tables(struct sock *sk)
{
	int i;

	/*
	 *	Shut down all active vif entries
	 */
	for (i=0; i<maxvif; i++) {
		if (!(vif_table[i].flags&VIFF_STATIC))
			vif_delete(i);
	}

	/*
	 *	Wipe the cache
	 */
	for (i=0;i<MFC_LINES;i++) {
		struct mfc_cache *c, **cp;

		cp = &mfc_cache_array[i];
		while ((c = *cp) != NULL) {
			if (c->mfc_flags&MFC_STATIC) {
				cp = &c->next;
				continue;
			}
			write_lock_bh(&mrt_lock);
			*cp = c->next;
			write_unlock_bh(&mrt_lock);

			kmem_cache_free(mrt_cachep, c);
		}
	}

	if (atomic_read(&cache_resolve_queue_len) != 0) {
		struct mfc_cache *c;

		spin_lock_bh(&mfc_unres_lock);
		while (mfc_unres_queue != NULL) {
			c = mfc_unres_queue;
			mfc_unres_queue = c->next;
			spin_unlock_bh(&mfc_unres_lock);

			ipmr_destroy_unres(c);

			spin_lock_bh(&mfc_unres_lock);
		}
		spin_unlock_bh(&mfc_unres_lock);
	}
}

static void mrtsock_destruct(struct sock *sk)
{
	rtnl_lock();
	if (sk == mroute_socket) {
		IPV4_DEVCONF_ALL(MC_FORWARDING)--;

		write_lock_bh(&mrt_lock);
		mroute_socket=NULL;
		write_unlock_bh(&mrt_lock);

		mroute_clean_tables(sk);
	}
	rtnl_unlock();
}

/*
 *	Socket options and virtual interface manipulation. The whole
 *	virtual interface system is a complete heap, but unfortunately
 *	that's how BSD mrouted happens to think. Maybe one day with a proper
 *	MOSPF/PIM router set up we can clean this up.
 */

int ip_mroute_setsockopt(struct sock *sk,int optname,char __user *optval,int optlen)
{
	int ret;
	struct vifctl vif;
	struct mfcctl mfc;

	if (optname != MRT_INIT) {
		if (sk != mroute_socket && !capable(CAP_NET_ADMIN))
			return -EACCES;
	}

	switch (optname) {
	case MRT_INIT:
		if (sk->sk_type != SOCK_RAW ||
		    inet_sk(sk)->num != IPPROTO_IGMP)
			return -EOPNOTSUPP;
		if (optlen!=sizeof(int))
			return -ENOPROTOOPT;

		rtnl_lock();
		if (mroute_socket) {
			rtnl_unlock();
			return -EADDRINUSE;
		}

		ret = ip_ra_control(sk, 1, mrtsock_destruct);
		if (ret == 0) {
			write_lock_bh(&mrt_lock);
			mroute_socket=sk;
			write_unlock_bh(&mrt_lock);

			IPV4_DEVCONF_ALL(MC_FORWARDING)++;
		}
		rtnl_unlock();
		return ret;
	case MRT_DONE:
		if (sk!=mroute_socket)
			return -EACCES;
		return ip_ra_control(sk, 0, NULL);
	case MRT_ADD_VIF:
	case MRT_DEL_VIF:
		if (optlen!=sizeof(vif))
			return -EINVAL;
		if (copy_from_user(&vif,optval,sizeof(vif)))
			return -EFAULT;
		if (vif.vifc_vifi >= MAXVIFS)
			return -ENFILE;
		rtnl_lock();
		if (optname==MRT_ADD_VIF) {
			ret = vif_add(&vif, sk==mroute_socket);
		} else {
			ret = vif_delete(vif.vifc_vifi);
		}
		rtnl_unlock();
		return ret;

		/*
		 *	Manipulate the forwarding caches. These live
		 *	in a sort of kernel/user symbiosis.
		 */
	case MRT_ADD_MFC:
	case MRT_DEL_MFC:
		if (optlen!=sizeof(mfc))
			return -EINVAL;
		if (copy_from_user(&mfc,optval, sizeof(mfc)))
			return -EFAULT;
		rtnl_lock();
		if (optname==MRT_DEL_MFC)
			ret = ipmr_mfc_delete(&mfc);
		else
			ret = ipmr_mfc_add(&mfc, sk==mroute_socket);
		rtnl_unlock();
		return ret;
		/*
		 *	Control PIM assert.
		 */
	case MRT_ASSERT:
	{
		int v;
		if (get_user(v,(int __user *)optval))
			return -EFAULT;
		mroute_do_assert=(v)?1:0;
		return 0;
	}
#ifdef CONFIG_IP_PIMSM
	case MRT_PIM:
	{
		int v, ret;
		if (get_user(v,(int __user *)optval))
			return -EFAULT;
		v = (v)?1:0;
		rtnl_lock();
		ret = 0;
		if (v != mroute_do_pim) {
			mroute_do_pim = v;
			mroute_do_assert = v;
#ifdef CONFIG_IP_PIMSM_V2
			if (mroute_do_pim)
				ret = inet_add_protocol(&pim_protocol,
							IPPROTO_PIM);
			else
				ret = inet_del_protocol(&pim_protocol,
							IPPROTO_PIM);
			if (ret < 0)
				ret = -EAGAIN;
#endif
		}
		rtnl_unlock();
		return ret;
	}
#endif
	/*
	 *	Spurious command, or MRT_VERSION which you cannot
	 *	set.
	 */
	default:
		return -ENOPROTOOPT;
	}
}

/*
 *	Getsock opt support for the multicast routing system.
 */

int ip_mroute_getsockopt(struct sock *sk,int optname,char __user *optval,int __user *optlen)
{
	int olr;
	int val;

	if (optname!=MRT_VERSION &&
#ifdef CONFIG_IP_PIMSM
	   optname!=MRT_PIM &&
#endif
	   optname!=MRT_ASSERT)
		return -ENOPROTOOPT;

	if (get_user(olr, optlen))
		return -EFAULT;

	olr = min_t(unsigned int, olr, sizeof(int));
	if (olr < 0)
		return -EINVAL;

	if (put_user(olr,optlen))
		return -EFAULT;
	if (optname==MRT_VERSION)
		val=0x0305;
#ifdef CONFIG_IP_PIMSM
	else if (optname==MRT_PIM)
		val=mroute_do_pim;
#endif
	else
		val=mroute_do_assert;
	if (copy_to_user(optval,&val,olr))
		return -EFAULT;
	return 0;
}

/*
 *	The IP multicast ioctl support routines.
 */

int ipmr_ioctl(struct sock *sk, int cmd, void __user *arg)
{
	struct sioc_sg_req sr;
	struct sioc_vif_req vr;
	struct vif_device *vif;
	struct mfc_cache *c;

	switch (cmd) {
	case SIOCGETVIFCNT:
		if (copy_from_user(&vr,arg,sizeof(vr)))
			return -EFAULT;
		if (vr.vifi>=maxvif)
			return -EINVAL;
		read_lock(&mrt_lock);
		vif=&vif_table[vr.vifi];
		if (VIF_EXISTS(vr.vifi))	{
			vr.icount=vif->pkt_in;
			vr.ocount=vif->pkt_out;
			vr.ibytes=vif->bytes_in;
			vr.obytes=vif->bytes_out;
			read_unlock(&mrt_lock);

			if (copy_to_user(arg,&vr,sizeof(vr)))
				return -EFAULT;
			return 0;
		}
		read_unlock(&mrt_lock);
		return -EADDRNOTAVAIL;
	case SIOCGETSGCNT:
		if (copy_from_user(&sr,arg,sizeof(sr)))
			return -EFAULT;

		read_lock(&mrt_lock);
		c = ipmr_cache_find(sr.src.s_addr, sr.grp.s_addr);
		if (c) {
			sr.pktcnt = c->mfc_un.res.pkt;
			sr.bytecnt = c->mfc_un.res.bytes;
			sr.wrong_if = c->mfc_un.res.wrong_if;
			read_unlock(&mrt_lock);

			if (copy_to_user(arg,&sr,sizeof(sr)))
				return -EFAULT;
			return 0;
		}
		read_unlock(&mrt_lock);
		return -EADDRNOTAVAIL;
	default:
		return -ENOIOCTLCMD;
	}
}


static int ipmr_device_event(struct notifier_block *this, unsigned long event, void *ptr)
{
	struct net_device *dev = ptr;
	struct vif_device *v;
	int ct;

	if (dev->nd_net != &init_net)
		return NOTIFY_DONE;

	if (event != NETDEV_UNREGISTER)
		return NOTIFY_DONE;
	v=&vif_table[0];
	for (ct=0;ct<maxvif;ct++,v++) {
		if (v->dev==dev)
			vif_delete(ct);
	}
	return NOTIFY_DONE;
}


static struct notifier_block ip_mr_notifier={
	.notifier_call = ipmr_device_event,
};

/*
 * 	Encapsulate a packet by attaching a valid IPIP header to it.
 *	This avoids tunnel drivers and other mess and gives us the speed so
 *	important for multicast video.
 */

static void ip_encap(struct sk_buff *skb, __be32 saddr, __be32 daddr)
{
	struct iphdr *iph;
	struct iphdr *old_iph = ip_hdr(skb);

	skb_push(skb, sizeof(struct iphdr));
	skb->transport_header = skb->network_header;
	skb_reset_network_header(skb);
	iph = ip_hdr(skb);

	iph->version	= 	4;
	iph->tos	=	old_iph->tos;
	iph->ttl	=	old_iph->ttl;
	iph->frag_off	=	0;
	iph->daddr	=	daddr;
	iph->saddr	=	saddr;
	iph->protocol	=	IPPROTO_IPIP;
	iph->ihl	=	5;
	iph->tot_len	=	htons(skb->len);
	ip_select_ident(iph, skb->dst, NULL);
	ip_send_check(iph);

	memset(&(IPCB(skb)->opt), 0, sizeof(IPCB(skb)->opt));
	nf_reset(skb);
}

static inline int ipmr_forward_finish(struct sk_buff *skb)
{
	struct ip_options * opt	= &(IPCB(skb)->opt);

	IP_INC_STATS_BH(IPSTATS_MIB_OUTFORWDATAGRAMS);

	if (unlikely(opt->optlen))
		ip_forward_options(skb);

	return dst_output(skb);
}

/*
 *	Processing handlers for ipmr_forward
 */

static void ipmr_queue_xmit(struct sk_buff *skb, struct mfc_cache *c, int vifi)
{
	const struct iphdr *iph = ip_hdr(skb);
	struct vif_device *vif = &vif_table[vifi];
	struct net_device *dev;
	struct rtable *rt;
	int    encap = 0;

	if (vif->dev == NULL)
		goto out_free;

#ifdef CONFIG_IP_PIMSM
	if (vif->flags & VIFF_REGISTER) {
		vif->pkt_out++;
		vif->bytes_out+=skb->len;
		((struct net_device_stats*)netdev_priv(vif->dev))->tx_bytes += skb->len;
		((struct net_device_stats*)netdev_priv(vif->dev))->tx_packets++;
		ipmr_cache_report(skb, vifi, IGMPMSG_WHOLEPKT);
		kfree_skb(skb);
		return;
	}
#endif

	if (vif->flags&VIFF_TUNNEL) {
		struct flowi fl = { .oif = vif->link,
				    .nl_u = { .ip4_u =
					      { .daddr = vif->remote,
						.saddr = vif->local,
						.tos = RT_TOS(iph->tos) } },
				    .proto = IPPROTO_IPIP };
		if (ip_route_output_key(&rt, &fl))
			goto out_free;
		encap = sizeof(struct iphdr);
	} else {
		struct flowi fl = { .oif = vif->link,
				    .nl_u = { .ip4_u =
					      { .daddr = iph->daddr,
						.tos = RT_TOS(iph->tos) } },
				    .proto = IPPROTO_IPIP };
		if (ip_route_output_key(&rt, &fl))
			goto out_free;
	}

	dev = rt->u.dst.dev;

	if (skb->len+encap > dst_mtu(&rt->u.dst) && (ntohs(iph->frag_off) & IP_DF)) {
		/* Do not fragment multicasts. Alas, IPv4 does not
		   allow to send ICMP, so that packets will disappear
		   to blackhole.
		 */

		IP_INC_STATS_BH(IPSTATS_MIB_FRAGFAILS);
		ip_rt_put(rt);
		goto out_free;
	}

	encap += LL_RESERVED_SPACE(dev) + rt->u.dst.header_len;

	if (skb_cow(skb, encap)) {
		ip_rt_put(rt);
		goto out_free;
	}

	vif->pkt_out++;
	vif->bytes_out+=skb->len;

	dst_release(skb->dst);
	skb->dst = &rt->u.dst;
	ip_decrease_ttl(ip_hdr(skb));

	/* FIXME: forward and output firewalls used to be called here.
	 * What do we do with netfilter? -- RR */
	if (vif->flags & VIFF_TUNNEL) {
		ip_encap(skb, vif->local, vif->remote);
		/* FIXME: extra output firewall step used to be here. --RR */
		((struct ip_tunnel *)netdev_priv(vif->dev))->stat.tx_packets++;
		((struct ip_tunnel *)netdev_priv(vif->dev))->stat.tx_bytes+=skb->len;
	}

	IPCB(skb)->flags |= IPSKB_FORWARDED;

	/*
	 * RFC1584 teaches, that DVMRP/PIM router must deliver packets locally
	 * not only before forwarding, but after forwarding on all output
	 * interfaces. It is clear, if mrouter runs a multicasting
	 * program, it should receive packets not depending to what interface
	 * program is joined.
	 * If we will not make it, the program will have to join on all
	 * interfaces. On the other hand, multihoming host (or router, but
	 * not mrouter) cannot join to more than one interface - it will
	 * result in receiving multiple packets.
	 */
	NF_HOOK(PF_INET, NF_IP_FORWARD, skb, skb->dev, dev,
		ipmr_forward_finish);
	return;

out_free:
	kfree_skb(skb);
	return;
}

static int ipmr_find_vif(struct net_device *dev)
{
	int ct;
	for (ct=maxvif-1; ct>=0; ct--) {
		if (vif_table[ct].dev == dev)
			break;
	}
	return ct;
}

/* "local" means that we should preserve one skb (for local delivery) */