ip_mroute.c

完整的TCP/IP源代码

	vifp->v_lcl_grps[vifp->v_lcl_grps_n++] = gcp->lgc_gaddr;

	if (gcp->lgc_gaddr.s_addr == vifp->v_cached_group)
		vifp->v_cached_result = 1;

	splx(s);
	return (0);
}

/*
 * Delete the the local multicast group associated with the vif
 * indexed by gcp->lgc_vifi.
 */

static int
del_lgrp(gcp)
	register struct lgrplctl *gcp;
{
	register struct vif *vifp;
	register int i, error, s;

	if (gcp->lgc_vifi >= numvifs)
		return (EINVAL);
	vifp = viftable + gcp->lgc_vifi;
	if (vifp->v_lcl_addr.s_addr == 0 || (vifp->v_flags & VIFF_TUNNEL))
		return (EADDRNOTAVAIL);

	s = splnet();

	if (gcp->lgc_gaddr.s_addr == vifp->v_cached_group)
		vifp->v_cached_result = 0;

	error = EADDRNOTAVAIL;
	for (i = 0; i < vifp->v_lcl_grps_n; ++i)
		if (same(&gcp->lgc_gaddr, &vifp->v_lcl_grps[i])) {
			error = 0;
			vifp->v_lcl_grps_n--;
			bcopy((caddr_t)&vifp->v_lcl_grps[i + 1],
			    (caddr_t)&vifp->v_lcl_grps[i],
			    (vifp->v_lcl_grps_n - i) * sizeof(struct in_addr));
			error = 0;
			break;
		}

	splx(s);
	return (error);
}

/*
 * Return 1 if gaddr is a member of the local group list for vifp.
 */
static int
grplst_member(vifp, gaddr)
	register struct vif *vifp;
	struct in_addr gaddr;
{
	register int i, s;
	register u_long addr;

	mrtstat.mrts_grp_lookups++;

	addr = gaddr.s_addr;
	if (addr == vifp->v_cached_group)
		return (vifp->v_cached_result);

	mrtstat.mrts_grp_misses++;

	for (i = 0; i < vifp->v_lcl_grps_n; ++i)
		if (addr == vifp->v_lcl_grps[i].s_addr) {
			s = splnet();
			vifp->v_cached_group = addr;
			vifp->v_cached_result = 1;
			splx(s);
			return (1);
		}
	s = splnet();
	vifp->v_cached_group = addr;
	vifp->v_cached_result = 0;
	splx(s);
	return (0);
}

/*
 * A simple hash function: returns MRTHASHMOD of the low-order octet of
 * the argument's network or subnet number.
 */
static u_long
nethash(in)
	struct in_addr in;
{
	register u_long n;

	n = in_netof(in);
	while ((n & 0xff) == 0)
		n >>= 8;
	return (MRTHASHMOD(n));
}

/*
 * Add an mrt entry
 */
static int
add_mrt(mrtcp)
	register struct mrtctl *mrtcp;
{
	struct mrt *rt;
	u_long hash;
	int s;

	if (rt = mrtfind(mrtcp->mrtc_origin)) {
		/* Just update the route */
		s = splnet();
		rt->mrt_parent = mrtcp->mrtc_parent;
		VIFM_COPY(mrtcp->mrtc_children, rt->mrt_children);
		VIFM_COPY(mrtcp->mrtc_leaves, rt->mrt_leaves);
		splx(s);
		return (0);
	}

	s = splnet();
	MALLOC (rt, struct mrt *, sizeof(*rt), MT_MRTABLE, M_DONTWAIT); 
	if (rt == NULL) {
		splx(s);
		return (ENOBUFS);
	}

	/*
	 * insert new entry at head of hash chain
	 */
	rt->mrt_origin = mrtcp->mrtc_origin;
	rt->mrt_originmask = mrtcp->mrtc_originmask;
	rt->mrt_parent = mrtcp->mrtc_parent;
	VIFM_COPY(mrtcp->mrtc_children, rt->mrt_children);
	VIFM_COPY(mrtcp->mrtc_leaves, rt->mrt_leaves);
	/* link into table */
	hash = nethash(mrtcp->mrtc_origin);
	rt->mrt_next = mrttable[hash];
	mrttable[hash] = rt;

	splx(s);
	return (0);
}

/*
 * Delete an mrt entry
 */
static int
del_mrt(origin)
	register struct in_addr *origin;
{
	register struct mrt *rt, *prev_rt;
	register u_long hash = nethash(*origin);
	register int s;

	for (prev_rt = rt = mrttable[hash]; rt; prev_rt = rt, rt = rt->mrt_next)
		if (origin->s_addr == rt->mrt_origin.s_addr)
			break;
	if (!rt)
		return (ESRCH);

	s = splnet();

	if (rt == cached_mrt)
		cached_mrt = NULL;

	if (prev_rt == rt)
		mrttable[hash] = rt->mrt_next;
	else
		prev_rt->mrt_next = rt->mrt_next;
	FREE(rt, M_MRTABLE);
	splx(s);
	return (0);
}

/*
 * Find a route for a given origin IP address.
 */
static struct mrt *
mrtfind(origin)
	struct in_addr origin;
{
	register struct mrt *rt;
	register u_int hash;
	register int s;

	mrtstat.mrts_mrt_lookups++;

	if (cached_mrt != NULL &&
	    (origin.s_addr & cached_originmask) == cached_origin)
		return (cached_mrt);

	mrtstat.mrts_mrt_misses++;

	hash = nethash(origin);
	for (rt = mrttable[hash]; rt; rt = rt->mrt_next)
		if ((origin.s_addr & rt->mrt_originmask.s_addr) ==
		    rt->mrt_origin.s_addr) {
			s = splnet();
			cached_mrt = rt;
			cached_origin = rt->mrt_origin.s_addr;
			cached_originmask = rt->mrt_originmask.s_addr;
			splx(s);
			return (rt);
		}
	return (NULL);
}

/*
 * IP multicast forwarding function. This function assumes that the packet
 * pointed to by "ip" has arrived on (or is about to be sent to) the interface
 * pointed to by "ifp", and the packet is to be relayed to other networks
 * that have members of the packet's destination IP multicast group.
 *
 * The packet is returned unscathed to the caller, unless it is tunneled
 * or erroneous, in which case a non-zero return value tells the caller to
 * discard it.
 */

#define IP_HDR_LEN  20	/* # bytes of fixed IP header (excluding options) */
#define TUNNEL_LEN  12  /* # bytes of IP option for tunnel encapsulation  */

static int
ip_mforward(m, ifp)
	register struct mbuf *m;
	register struct ifnet *ifp;
{
	register struct ip *ip = mtod(m, struct ip *);
	register struct mrt *rt;
	register struct vif *vifp;
	register int vifi;
	register u_char *ipoptions;
	u_long tunnel_src;

	if (ip->ip_hl < (IP_HDR_LEN + TUNNEL_LEN) >> 2 ||
	    (ipoptions = (u_char *)(ip + 1))[1] != IPOPT_LSRR ) {
		/*
		 * Packet arrived via a physical interface.
		 */
		tunnel_src = 0;
	} else {
		/*
		 * Packet arrived through a tunnel.
		 *
		 * A tunneled packet has a single NOP option and a
		 * two-element loose-source-and-record-route (LSRR)
		 * option immediately following the fixed-size part of
		 * the IP header.  At this point in processing, the IP
		 * header should contain the following IP addresses:
		 *
		 * original source          - in the source address field
		 * destination group        - in the destination address field
		 * remote tunnel end-point  - in the first  element of LSRR
		 * one of this host's addrs - in the second element of LSRR
		 *
		 * NOTE: RFC-1075 would have the original source and
		 * remote tunnel end-point addresses swapped.  However,
		 * that could cause delivery of ICMP error messages to
		 * innocent applications on intermediate routing
		 * hosts!  Therefore, we hereby change the spec.
		 */

		/*
		 * Verify that the tunnel options are well-formed.
		 */
		if (ipoptions[0] != IPOPT_NOP ||
		    ipoptions[2] != 11 ||	/* LSRR option length   */
		    ipoptions[3] != 12 ||	/* LSRR address pointer */
		    (tunnel_src = *(u_long *)(&ipoptions[4])) == 0) {
			mrtstat.mrts_bad_tunnel++;
			return (1);
		}

		/*
		 * Delete the tunnel options from the packet.
		 */
		ovbcopy((caddr_t)(ipoptions + TUNNEL_LEN), (caddr_t)ipoptions,
		    (unsigned)(m->m_len - (IP_HDR_LEN + TUNNEL_LEN)));
		m->m_len -= TUNNEL_LEN;
		ip->ip_len -= TUNNEL_LEN;
		ip->ip_hl -= TUNNEL_LEN >> 2;
	}

	/*
	 * Don't forward a packet with time-to-live of zero or one,
	 * or a packet destined to a local-only group.
	 */
	if (ip->ip_ttl <= 1 ||
	    ntohl(ip->ip_dst.s_addr) <= INADDR_MAX_LOCAL_GROUP)
		return ((int)tunnel_src);

	/*
	 * Don't forward if we don't have a route for the packet's origin.
	 */
	if (!(rt = mrtfind(ip->ip_src))) {
		mrtstat.mrts_no_route++;
		return ((int)tunnel_src);
	}

	/*
	 * Don't forward if it didn't arrive from the parent vif for its origin.
	 */
	vifi = rt->mrt_parent;
	if (tunnel_src == 0 ) {
		if ((viftable[vifi].v_flags & VIFF_TUNNEL) ||
		    viftable[vifi].v_ifp != ifp )
			return ((int)tunnel_src);
	} else {
		if (!(viftable[vifi].v_flags & VIFF_TUNNEL) ||
		    viftable[vifi].v_rmt_addr.s_addr != tunnel_src )
			return ((int)tunnel_src);
	}

	/*
	 * For each vif, decide if a copy of the packet should be forwarded.
	 * Forward if:
	 *		- the ttl exceeds the vif's threshold AND
	 *		- the vif is a child in the origin's route AND
	 *		- ( the vif is not a leaf in the origin's route OR
	 *		    the destination group has members on the vif )
	 *
	 * (This might be speeded up with some sort of cache -- someday.)
	 */
	for (vifp = viftable, vifi = 0; vifi < numvifs; vifp++, vifi++) {
		if (ip->ip_ttl > vifp->v_threshold &&
		    VIFM_ISSET(vifi, rt->mrt_children) &&
		    (!VIFM_ISSET(vifi, rt->mrt_leaves) ||
		    grplst_member(vifp, ip->ip_dst))) {
			if (vifp->v_flags & VIFF_TUNNEL)
				tunnel_send(m, vifp);
			else
				phyint_send(m, vifp);
		}
	}

	return ((int)tunnel_src);
}

static void
phyint_send(m, vifp)
	register struct mbuf *m;
	register struct vif *vifp;
{
	register struct ip *ip = mtod(m, struct ip *);
	register struct mbuf *mb_copy;
	register struct ip_moptions *imo;
	register int error;
	struct ip_moptions simo;

	mb_copy = m_copy(m, 0, M_COPYALL);
	if (mb_copy == NULL)
		return;

	imo = &simo;
	imo->imo_multicast_ifp = vifp->v_ifp;
	imo->imo_multicast_ttl = ip->ip_ttl - 1;
	imo->imo_multicast_loop = 1;

	error = ip_output(mb_copy, NULL, NULL, IP_FORWARDING, imo);
}

static void
tunnel_send(m, vifp)
	register struct mbuf *m;
	register struct vif *vifp;
{
	register struct ip *ip = mtod(m, struct ip *);
	register struct mbuf *mb_copy, *mb_opts;
	register struct ip *ip_copy;
	register int error;
	register u_char *cp;

	/*
	 * Make sure that adding the tunnel options won't exceed the
	 * maximum allowed number of option bytes.
	 */
	if (ip->ip_hl > (60 - TUNNEL_LEN) >> 2) {
		mrtstat.mrts_cant_tunnel++;
		return;
	}

	/* 
	 * Get a private copy of the IP header so that changes to some 
	 * of the IP fields don't damage the original header, which is
	 * examined later in ip_input.c.
	 */
	mb_copy = m_copy(m, IP_HDR_LEN, M_COPYALL);
	if (mb_copy == NULL)
		return;
	mb_opts = mHdrClGet(M_DONTWAIT, MT_HEADER, CL_SIZE_128, TRUE);
	if (mb_opts == NULL) {
		m_freem(mb_copy);
		return;
	}
	/*
	 * Make mb_opts be the new head of the packet chain.
	 * Any options of the packet were left in the old packet chain head
	 */
	mb_opts->m_next = mb_copy;
	mb_opts->m_len = IP_HDR_LEN + TUNNEL_LEN;
	mb_opts->m_data += MSIZE - mb_opts->m_len;

	ip_copy = mtod(mb_opts, struct ip *);
	/*
	 * Copy the base ip header to the new head mbuf.
	 */
	*ip_copy = *ip;
	ip_copy->ip_ttl--;
	ip_copy->ip_dst = vifp->v_rmt_addr;	/* remote tunnel end-point */
	/*
	 * Adjust the ip header length to account for the tunnel options.
	 */
	ip_copy->ip_hl += TUNNEL_LEN >> 2;
	ip_copy->ip_len += TUNNEL_LEN;
	/*
	 * Add the NOP and LSRR after the base ip header
	 */
	cp = (u_char *)(ip_copy + 1);
	*cp++ = IPOPT_NOP;
	*cp++ = IPOPT_LSRR;
	*cp++ = 11;		/* LSRR option length */
	*cp++ = 8;		/* LSSR pointer to second element */
	*(u_long*)cp = vifp->v_lcl_addr.s_addr;	/* local tunnel end-point */
	cp += 4;
	*(u_long*)cp = ip->ip_dst.s_addr;		/* destination group */

	error = ip_output(mb_opts, NULL, NULL, IP_FORWARDING, NULL);
}