Add kernel support for Protocol Independant Multicast (PIM)

Information: http://netweb.usc.edu/pim/

From Pavlin Radoslavov <pavlin@icir.org>

ok deraadt@ brad@

8 files changed, 1794 insertions, 32 deletions

diff --git a/sys/conf/GENERIC b/sys/conf/GENERIC
index a97d69965ab..c7debd1dc21 100644
--- a/sys/conf/GENERIC
+++ b/sys/conf/GENERIC
@@ -1,4 +1,4 @@
-#	$OpenBSD: GENERIC,v 1.112 2004/11/28 23:39:45 canacar Exp $
+#	$OpenBSD: GENERIC,v 1.113 2005/01/14 14:51:27 mcbride Exp $
 #
 #	Machine-independent option; used by all architectures for their
 #	GENERIC kernel
@@ -74,6 +74,7 @@ option		IPSEC		# IPsec
 option		PPP_BSDCOMP	# PPP BSD compression
 option		PPP_DEFLATE
 #option		MROUTING	# Multicast router
+#option		PIM		# Protocol Independent Multicast
 
 pseudo-device	pf		# packet filter
 pseudo-device	pflog		# pf log if
diff --git a/sys/netinet/in.h b/sys/netinet/in.h
index 254b939205f..366583281c5 100644
--- a/sys/netinet/in.h
+++ b/sys/netinet/in.h
@@ -1,4 +1,4 @@
-/*	$OpenBSD: in.h,v 1.63 2004/11/24 01:25:42 mcbride Exp $	*/
+/*	$OpenBSD: in.h,v 1.64 2005/01/14 14:51:27 mcbride Exp $	*/
 /*	$NetBSD: in.h,v 1.20 1996/02/13 23:41:47 christos Exp $	*/
 
 /*
@@ -431,7 +431,7 @@ struct ip_mreq {
 	{ 0, 0 }, \
 	{ 0, 0 }, \
 	{ 0, 0 }, \
-	{ 0, 0 }, \
+	{ "pim", CTLTYPE_NODE }, \
 	{ 0, 0 }, \
 	{ 0, 0 }, \
 	{ 0, 0 }, \
diff --git a/sys/netinet/in_proto.c b/sys/netinet/in_proto.c
index 68a35c622b6..61b3c1ca6ac 100644
--- a/sys/netinet/in_proto.c
+++ b/sys/netinet/in_proto.c
@@ -1,4 +1,4 @@
-/*	$OpenBSD: in_proto.c,v 1.42 2004/12/07 20:38:47 mcbride Exp $	*/
+/*	$OpenBSD: in_proto.c,v 1.43 2005/01/14 14:51:27 mcbride Exp $	*/
 /*	$NetBSD: in_proto.c,v 1.14 1996/02/18 18:58:32 christos Exp $	*/
 
 /*
@@ -125,6 +125,9 @@
 #endif
 
 #include <netinet/igmp_var.h>
+#ifdef PIM
+#include <netinet/pim_var.h>
+#endif
 #include <netinet/tcp.h>
 #include <netinet/tcp_timer.h>
 #include <netinet/tcp_var.h>
@@ -242,6 +245,13 @@ struct protosw inetsw[] = {
   rip_usrreq,
   igmp_init,	igmp_fasttimo,	igmp_slowtimo,	0,
 },
+#ifdef PIM
+{ SOCK_RAW,	&inetdomain,	IPPROTO_PIM,	PR_ATOMIC|PR_ADDR,
+  pim_input,	rip_output,	0,		rip_ctloutput,
+  rip_usrreq,
+  0,		0,		0,		0,
+},
+#endif /* PIM */
 #ifdef IPXIP
 { SOCK_RAW,	&inetdomain,	IPPROTO_IDP,	PR_ATOMIC|PR_ADDR,
   ipxip_input,	rip_output,	ipxip_ctlinput,	0,
diff --git a/sys/netinet/ip_mroute.c b/sys/netinet/ip_mroute.c
index 54c695e3312..113d5236b30 100644
--- a/sys/netinet/ip_mroute.c
+++ b/sys/netinet/ip_mroute.c
@@ -1,4 +1,4 @@
-/*	$OpenBSD: ip_mroute.c,v 1.38 2004/11/24 01:25:42 mcbride Exp $	*/
+/*	$OpenBSD: ip_mroute.c,v 1.39 2005/01/14 14:51:27 mcbride Exp $	*/
 /*	$NetBSD: ip_mroute.c,v 1.85 2004/04/26 01:31:57 matt Exp $	*/
 
 /*
@@ -46,10 +46,21 @@
  * Modified by Ajit Thyagarajan, PARC, August 1993
  * Modified by Bill Fenner, PARC, April 1994
  * Modified by Charles M. Hannum, NetBSD, May 1995.
+ * Modified by Ahmed Helmy, SGI, June 1996
+ * Modified by George Edmond Eddy (Rusty), ISI, February 1998
+ * Modified by Pavlin Radoslavov, USC/ISI, May 1998, August 1999, October 2000
+ * Modified by Hitoshi Asaeda, WIDE, August 2000
+ * Modified by Pavlin Radoslavov, ICSI, October 2002
  *
  * MROUTING Revision: 1.2
+ * and PIM-SMv2 and PIM-DM support, advanced API support,
+ * bandwidth metering and signaling
  */
 
+#ifdef PIM
+#define _PIM_VT 1
+#endif
+
 #include <sys/param.h>
 #include <sys/systm.h>
 #include <sys/mbuf.h>
@@ -77,6 +88,10 @@
 #include <netinet/igmp.h>
 #include <netinet/igmp_var.h>
 #include <netinet/ip_mroute.h>
+#ifdef PIM
+#include <netinet/pim.h>
+#include <netinet/pim_var.h>
+#endif
 
 #include <sys/stdarg.h>
 
@@ -111,6 +126,10 @@ u_int		mrtdebug = 0;	  /* debug level 	*/
 #define		DEBUG_FORWARD	0x04
 #define		DEBUG_EXPIRE	0x08
 #define		DEBUG_XMIT	0x10
+#define		DEBUG_PIM	0x20
+
+#define		VIFI_INVALID	((vifi_t) -1)
+
 u_int       	tbfdebug = 0;     /* tbf debug level 	*/
 #ifdef RSVP_ISI
 u_int		rsvpdebug = 0;	  /* rsvp debug level   */
@@ -136,14 +155,17 @@ static int set_assert(struct mbuf *);
 static int get_assert(struct mbuf *);
 static int add_vif(struct mbuf *);
 static int del_vif(struct mbuf *);
-static void update_mfc_params(struct mfc *, struct mfcctl *);
-static void init_mfc_params(struct mfc *, struct mfcctl *);
+static void update_mfc_params(struct mfc *, struct mfcctl2 *);
+static void init_mfc_params(struct mfc *, struct mfcctl2 *);
 static void expire_mfc(struct mfc *);
 static int add_mfc(struct mbuf *);
 #ifdef UPCALL_TIMING
 static void collate(struct timeval *);
 #endif
 static int del_mfc(struct mbuf *);
+static int set_api_config(struct mbuf *); /* chose API capabilities */
+static int get_api_support(struct mbuf *);
+static int get_api_config(struct mbuf *);
 static int socket_send(struct socket *, struct mbuf *,
 			    struct sockaddr_in *);
 static void expire_upcalls(void *);
@@ -165,6 +187,31 @@ static void tbf_update_tokens(struct vif *);
 static int priority(struct vif *, struct ip *);
 
 /*
+ * Bandwidth monitoring
+ */
+static void free_bw_list(struct bw_meter *);
+static int add_bw_upcall(struct mbuf *);
+static int del_bw_upcall(struct mbuf *);
+static void bw_meter_receive_packet(struct bw_meter *, int , struct timeval *);
+static void bw_meter_prepare_upcall(struct bw_meter *, struct timeval *);
+static void bw_upcalls_send(void);
+static void schedule_bw_meter(struct bw_meter *, struct timeval *);
+static void unschedule_bw_meter(struct bw_meter *);
+static void bw_meter_process(void);
+static void expire_bw_upcalls_send(void *);
+static void expire_bw_meter_process(void *);
+
+#ifdef PIM
+static int pim_register_send(struct ip *, struct vif *,
+		struct mbuf *, struct mfc *);
+static int pim_register_send_rp(struct ip *, struct vif *,
+		struct mbuf *, struct mfc *);
+static int pim_register_send_upcall(struct ip *, struct vif *,
+		struct mbuf *, struct mfc *);
+static struct mbuf *pim_register_prepare(struct ip *, struct mbuf *);
+#endif
+
+/*
  * 'Interfaces' associated with decapsulator (so we can tell
  * packets that went through it from ones that get reflected
  * by a broken gateway).  These interfaces are never linked into
@@ -195,6 +242,78 @@ struct ip multicast_encap_iphdr = {
 };
 
 /*
+ * Bandwidth meter variables and constants
+ */
+
+/*
+ * Pending timeouts are stored in a hash table, the key being the
+ * expiration time. Periodically, the entries are analysed and processed.
+ */
+#define BW_METER_BUCKETS	1024
+static struct bw_meter *bw_meter_timers[BW_METER_BUCKETS];
+struct timeout bw_meter_ch;
+#define BW_METER_PERIOD (hz)		/* periodical handling of bw meters */
+
+/*
+ * Pending upcalls are stored in a vector which is flushed when
+ * full, or periodically
+ */
+static struct bw_upcall	bw_upcalls[BW_UPCALLS_MAX];
+static u_int	bw_upcalls_n; /* # of pending upcalls */
+struct timeout	bw_upcalls_ch;
+#define BW_UPCALLS_PERIOD (hz)		/* periodical flush of bw upcalls */
+
+#ifdef PIM
+struct pimstat pimstat;
+
+/*
+ * Note: the PIM Register encapsulation adds the following in front of a
+ * data packet:
+ *
+ * struct pim_encap_hdr {
+ *    struct ip ip;
+ *    struct pim_encap_pimhdr  pim;
+ * }
+ *
+ */
+
+struct pim_encap_pimhdr {
+	struct pim pim;
+	uint32_t   flags;
+};
+
+static struct ip pim_encap_iphdr = {
+#if BYTE_ORDER == LITTLE_ENDIAN
+	sizeof(struct ip) >> 2,
+	IPVERSION,
+#else
+	IPVERSION,
+	sizeof(struct ip) >> 2,
+#endif
+	0,			/* tos */
+	sizeof(struct ip),	/* total length */
+	0,			/* id */
+	0,			/* frag offset */ 
+	ENCAP_TTL,
+	IPPROTO_PIM,
+	0,			/* checksum */
+};
+
+static struct pim_encap_pimhdr pim_encap_pimhdr = {
+    {
+	PIM_MAKE_VT(PIM_VERSION, PIM_REGISTER), /* PIM vers and message type */
+	0,			/* reserved */
+	0,			/* checksum */
+    },
+    0				/* flags */
+};
+
+static struct ifnet multicast_register_if;
+static vifi_t reg_vif_num = VIFI_INVALID;
+#endif /* PIM */
+
+
+/*
  * Private variables.
  */
 static vifi_t	   numvifs = 0;
@@ -217,6 +336,17 @@ static int pim_assert;
 #define ASSERT_MSG_TIME		3000000
 
 /*
+ * Kernel multicast routing API capabilities and setup.
+ * If more API capabilities are added to the kernel, they should be
+ * recorded in `mrt_api_support'.
+ */
+static const u_int32_t mrt_api_support = (MRT_MFC_FLAGS_DISABLE_WRONGVIF |
+					  MRT_MFC_FLAGS_BORDER_VIF |
+					  MRT_MFC_RP |
+					  MRT_MFC_BW_UPCALL);
+static u_int32_t mrt_api_config = 0;
+
+/*
  * Find a route for a given origin IP address and Multicast group address
  * Type of service parameter to be added in the future!!!
  * Statistics are updated by the caller if needed
@@ -297,6 +427,15 @@ ip_mrouter_set(struct socket *so, int optname, struct mbuf **m)
 		case MRT_ASSERT:
 			error = set_assert(*m);
 			break;
+		case MRT_API_CONFIG:
+			error = set_api_config(*m);
+			break;
+		case MRT_ADD_BW_UPCALL:
+			error = add_bw_upcall(*m);
+			break;
+		case MRT_DEL_BW_UPCALL:
+			error = del_bw_upcall(*m);
+			break;
 		default:
 			error = ENOPROTOOPT;
 			break;
@@ -327,6 +466,12 @@ ip_mrouter_get(struct socket *so, int optname, struct mbuf **m)
 		case MRT_ASSERT:
 			error = get_assert(*m);
 			break;
+		case MRT_API_SUPPORT:
+			error = get_api_support(*m);
+			break;
+		case MRT_API_CONFIG:
+			error = get_api_config(*m);
+			break;
 		default:
 			error = ENOPROTOOPT;
 			break;
@@ -445,6 +590,12 @@ ip_mrouter_init(struct socket *so, struct mbuf *m)
 	timeout_set(&expire_upcalls_ch, expire_upcalls, NULL);
 	timeout_add(&expire_upcalls_ch, EXPIRE_TIMEOUT);
 
+	timeout_set(&bw_upcalls_ch, expire_bw_upcalls_send, NULL);
+	timeout_add(&bw_upcalls_ch, BW_UPCALLS_PERIOD);
+
+	timeout_set(&bw_meter_ch, expire_bw_meter_process, NULL);
+	timeout_add(&bw_meter_ch, BW_METER_PERIOD);
+
 	if (mrtdebug)
 		log(LOG_DEBUG, "ip_mrouter_init\n");
 
@@ -473,8 +624,11 @@ ip_mrouter_done()
 
 	numvifs = 0;
 	pim_assert = 0;
+	mrt_api_config = 0;
 
 	timeout_del(&expire_upcalls_ch);
+	timeout_del(&bw_upcalls_ch);
+	timeout_del(&bw_meter_ch);
 
 	/*
 	 * Free all multicast forwarding cache entries.
@@ -493,6 +647,9 @@ ip_mrouter_done()
 	free(mfchashtbl, M_MRTABLE);
 	mfchashtbl = NULL;
 
+	bw_upcalls_n = 0;
+	bzero(bw_meter_timers, sizeof(bw_meter_timers));
+
 	/* Reset de-encapsulation cache. */
 	have_encap_tunnel = 0;
 
@@ -571,6 +728,84 @@ get_assert(struct mbuf *m)
 	return (0);
 }
 
+/*
+ * Configure API capabilities
+ */
+static int
+set_api_config(struct mbuf *m)
+{
+	int i;
+	u_int32_t *apival;
+
+	if (m == NULL || m->m_len < sizeof(u_int32_t))
+		return (EINVAL);
+
+	apival = mtod(m, u_int32_t *);
+
+	/*
+	 * We can set the API capabilities only if it is the first operation
+	 * after MRT_INIT. I.e.:
+	 *  - there are no vifs installed
+	 *  - pim_assert is not enabled
+	 *  - the MFC table is empty
+	 */
+	if (numvifs > 0) {
+		*apival = 0;
+		return (EPERM);
+	}
+	if (pim_assert) {
+		*apival = 0;
+		return (EPERM);
+	}
+	for (i = 0; i < MFCTBLSIZ; i++) {
+		if (LIST_FIRST(&mfchashtbl[i]) != NULL) {
+			*apival = 0;
+			return (EPERM);
+		}
+	}
+
+	mrt_api_config = *apival & mrt_api_support;
+	*apival = mrt_api_config;
+
+	return (0);
+}
+
+/*
+ * Get API capabilities
+ */
+static int
+get_api_support(struct mbuf *m)
+{
+	u_int32_t *apival;
+
+	if (m == NULL || m->m_len < sizeof(u_int32_t))
+		return (EINVAL);
+
+	apival = mtod(m, u_int32_t *);
+
+	*apival = mrt_api_support;
+
+	return (0);
+}
+
+/*
+ * Get API configured capabilities
+ */
+static int
+get_api_config(struct mbuf *m)
+{
+	u_int32_t *apival;
+
+	if (m == NULL || m->m_len < sizeof(u_int32_t))
+		return (EINVAL);
+
+	apival = mtod(m, u_int32_t *);
+
+	*apival = mrt_api_config;
+
+	return (0);
+}
+
 static struct sockaddr_in sin = { sizeof(sin), AF_INET };
 
 /*
@@ -600,11 +835,23 @@ add_vif(struct mbuf *m)
 		return (EADDRINUSE);
 
 	/* Find the interface with an address in AF_INET family. */
-	sin.sin_addr = vifcp->vifc_lcl_addr;
-	ifa = ifa_ifwithaddr(sintosa(&sin));
-	if (ifa == NULL)
-		return (EADDRNOTAVAIL);
-	ifp = ifa->ifa_ifp;
+#ifdef PIM
+	if (vifcp->vifc_flags & VIFF_REGISTER) {
+		/*
+		 * XXX: Because VIFF_REGISTER does not really need a valid
+		 * local interface (e.g. it could be 127.0.0.2), we don't
+		 * check its address.
+		 */
+	    ifp = NULL;
+	} else
+#endif
+	{
+		sin.sin_addr = vifcp->vifc_lcl_addr;
+		ifa = ifa_ifwithaddr(sintosa(&sin));
+		if (ifa == NULL)
+			return (EADDRNOTAVAIL);
+		ifp = ifa->ifa_ifp;
+	}
 
 	if (vifcp->vifc_flags & VIFF_TUNNEL) {
 		if (vifcp->vifc_flags & VIFF_SRCRT) {
@@ -626,6 +873,21 @@ add_vif(struct mbuf *m)
 		 * encapsulated packets.
 		 */
 		have_encap_tunnel = 1;
+#ifdef PIM
+	} else if (vifcp->vifc_flags & VIFF_REGISTER) {
+		ifp = &multicast_register_if;
+		if (mrtdebug)
+			log(LOG_DEBUG, "Adding a register vif, ifp: %p\n",
+			    (void *)ifp);
+		if (reg_vif_num == VIFI_INVALID) {
+			bzero(ifp, sizeof(*ifp));
+			snprintf(ifp->if_xname, sizeof ifp->if_xname,
+				 "register_vif");
+			ifp->if_flags = IFF_LOOPBACK;
+			bzero(&vifp->v_route, sizeof(vifp->v_route));
+			reg_vif_num = vifcp->vifc_vifi;
+		}
+#endif
 	} else {
 		/* Use the physical interface associated with the address. */
 		ifp = ifa->ifa_ifp;
@@ -714,6 +976,11 @@ reset_vif(struct vif *vifp)
 			last_encap_vif = NULL;
 			last_encap_src = zeroin_addr;
 		}
+	} else if (vifp->v_flags & VIFF_REGISTER) {
+#ifdef PIM
+		if (vifp->v_flags & VIFF_REGISTER)
+			reg_vif_num = VIFI_INVALID;
+#endif
 	} else {
 		satosin(&ifr.ifr_addr)->sin_len = sizeof(struct sockaddr_in);
 		satosin(&ifr.ifr_addr)->sin_family = AF_INET;
@@ -799,21 +1066,28 @@ vif_delete(struct ifnet *ifp)
  * update an mfc entry without resetting counters and S,G addresses.
  */
 static void
-update_mfc_params(struct mfc *rt, struct mfcctl *mfccp)
+update_mfc_params(struct mfc *rt, struct mfcctl2 *mfccp)
 {
 	int i;
 
 	rt->mfc_parent = mfccp->mfcc_parent;
 	for (i = 0; i < numvifs; i++) {
 		rt->mfc_ttls[i] = mfccp->mfcc_ttls[i];
+		rt->mfc_flags[i] = mfccp->mfcc_flags[i] & mrt_api_config &
+			MRT_MFC_FLAGS_ALL;
 	}
+	/* set the RP address */
+	if (mrt_api_config & MRT_MFC_RP)
+		rt->mfc_rp = mfccp->mfcc_rp;
+	else
+		rt->mfc_rp = zeroin_addr;
 }
 
 /*
  * fully initialize an mfc entry from the parameter.
  */
 static void
-init_mfc_params(struct mfc *rt, struct mfcctl *mfccp)
+init_mfc_params(struct mfc *rt, struct mfcctl2 *mfccp)
 {
 	rt->mfc_origin     = mfccp->mfcc_origin;
 	rt->mfc_mcastgrp   = mfccp->mfcc_mcastgrp;
@@ -832,6 +1106,8 @@ expire_mfc(struct mfc *rt)
 {
 	struct rtdetq *rte, *nrte;
 
+	free_bw_list(rt->mfc_bw_meter);
+
 	for (rte = rt->mfc_stall; rte != NULL; rte = nrte) {
 		nrte = rte->next;
 		m_freem(rte->m);
@@ -848,17 +1124,34 @@ expire_mfc(struct mfc *rt)
 static int
 add_mfc(struct mbuf *m)
 {
-	struct mfcctl *mfccp;
+	struct mfcctl2 mfcctl2;
+	struct mfcctl2 *mfccp;
 	struct mfc *rt;
 	u_int32_t hash = 0;
 	struct rtdetq *rte, *nrte;
 	u_short nstl;
 	int s;
+	int mfcctl_size = sizeof(struct mfcctl);
+
+	if (mrt_api_config & MRT_API_FLAGS_ALL)
+		mfcctl_size = sizeof(struct mfcctl2);
 
-	if (m == NULL || m->m_len < sizeof(struct mfcctl))
+	if (m == NULL || m->m_len < mfcctl_size)
 		return (EINVAL);
 
-	mfccp = mtod(m, struct mfcctl *);
+	/*
+	 * select data size depending on API version.
+	 */
+	if (mrt_api_config & MRT_API_FLAGS_ALL) {
+		struct mfcctl2 *mp2 = mtod(m, struct mfcctl2 *);
+		bcopy(mp2, (caddr_t)&mfcctl2, sizeof(*mp2));
+	} else {
+		struct mfcctl *mp = mtod(m, struct mfcctl *);
+		bcopy(mp, (caddr_t)&mfcctl2, sizeof(*mp));
+		bzero((caddr_t)&mfcctl2 + sizeof(struct mfcctl),
+		      sizeof(mfcctl2) - sizeof(struct mfcctl));
+	}
+	mfccp = &mfcctl2;
 
 	s = splsoftnet();
 	rt = mfc_find(&mfccp->mfcc_origin, &mfccp->mfcc_mcastgrp);
@@ -959,6 +1252,7 @@ add_mfc(struct mbuf *m)
 			init_mfc_params(rt, mfccp);
 			rt->mfc_expire	= 0;
 			rt->mfc_stall	= NULL;
+			rt->mfc_bw_meter = NULL;
 
 			/* insert new entry at head of hash chain */
 			LIST_INSERT_HEAD(&mfchashtbl[hash], rt, mfc_hash);
@@ -1000,14 +1294,26 @@ collate(struct timeval *t)
 static int
 del_mfc(struct mbuf *m)
 {
-	struct mfcctl *mfccp;
+	struct mfcctl2 mfcctl2;
+	struct mfcctl2 *mfccp;
 	struct mfc *rt;
 	int s;
+	int mfcctl_size = sizeof(struct mfcctl);
+	struct mfcctl *mp = mtod(m, struct mfcctl *);
+
+	/*
+	 * XXX: for deleting MFC entries the information in entries
+	 * of size "struct mfcctl" is sufficient.
+	 */
 
-	if (m == NULL || m->m_len < sizeof(struct mfcctl))
+	if (m == NULL || m->m_len < mfcctl_size)
 		return (EINVAL);
 
-	mfccp = mtod(m, struct mfcctl *);
+	bcopy(mp, (caddr_t)&mfcctl2, sizeof(*mp));
+	bzero((caddr_t)&mfcctl2 + sizeof(struct mfcctl),
+	      sizeof(mfcctl2) - sizeof(struct mfcctl));
+
+	mfccp = &mfcctl2;
 
 	if (mrtdebug & DEBUG_MFC)
 		log(LOG_DEBUG, "del_mfc origin %x mcastgrp %x\n",
@@ -1022,6 +1328,12 @@ del_mfc(struct mbuf *m)
 		return (EADDRNOTAVAIL);
 	}
 
+	/*
+	 * free the bw_meter entries
+	 */
+	free_bw_list(rt->mfc_bw_meter);
+	rt->mfc_bw_meter = NULL;
+
 	LIST_REMOVE(rt, mfc_hash);
 	free(rt, M_MRTABLE);
 
@@ -1079,7 +1391,7 @@ ip_mforward(struct mbuf *m, struct ifnet *ifp)
 	    ((u_char *)(ip + 1))[1] != IPOPT_LSRR) {
 		/*
 		 * Packet arrived via a physical interface or
-		 * an encapuslated tunnel.
+		 * an encapsulated tunnel or a register_vif.
 		 */
 	} else {
 		/*
@@ -1248,10 +1560,17 @@ ip_mforward(struct mbuf *m, struct ifnet *ifp)
 			rt->mfc_wrong_if = 0;
 			rt->mfc_expire = UPCALL_EXPIRE;
 			nexpire[hash]++;
-			for (i = 0; i < numvifs; i++)
+			for (i = 0; i < numvifs; i++) {
 				rt->mfc_ttls[i] = 0;
+				rt->mfc_flags[i] = 0;
+			}
 			rt->mfc_parent = -1;
 
+			/* clear the RP address */
+			rt->mfc_rp = zeroin_addr;
+
+			rt->mfc_bw_meter = NULL;
+
 			/* link into table */
 			LIST_INSERT_HEAD(&mfchashtbl[hash], rt, mfc_hash);
 			/* Add this entry to the end of the queue */
@@ -1316,6 +1635,16 @@ expire_upcalls(void *v)
 				continue;
 			nexpire[i]--;
 
+			/*
+			 * free the bw_meter entries
+			 */
+			while (rt->mfc_bw_meter != NULL) {
+				struct bw_meter *x = rt->mfc_bw_meter;
+
+				rt->mfc_bw_meter = x->bm_mfc_next;
+				free(x, M_BWMETER);
+			}
+
 			++mrtstat.mrts_cache_cleanups;
 			if (mrtdebug & DEBUG_EXPIRE)
 				log(LOG_DEBUG,
@@ -1365,6 +1694,11 @@ ip_mdq(struct mbuf *m, struct ifnet *ifp, struct mfc *rt)
 	 * (since vifi_t is u_short, -1 becomes MAXUSHORT, which > numvifs.
 	 */
 	if (xmt_vif < numvifs) {
+#ifdef PIM
+		if (viftable[xmt_vif].v_flags & VIFF_REGISTER)
+			pim_register_send(ip, viftable + xmt_vif, m, rt);
+		else
+#endif
 		MC_SEND(ip, viftable + xmt_vif, m);
 		return (1);
 	}
@@ -1383,16 +1717,22 @@ ip_mdq(struct mbuf *m, struct ifnet *ifp, struct mfc *rt)
 		++mrtstat.mrts_wrong_if;
 		++rt->mfc_wrong_if;
 		/*
-		 * If we are doing PIM assert processing, and we are forwarding
-		 * packets on this interface, and it is a broadcast medium
-		 * interface (and not a tunnel), send a message to the routing daemon.
+		 * If we are doing PIM assert processing, send a message
+		 * to the routing daemon.
+		 *
+		 * XXX: A PIM-SM router needs the WRONGVIF detection so it
+		 * can complete the SPT switch, regardless of the type
+		 * of the iif (broadcast media, GRE tunnel, etc).
 		 */
-		if (pim_assert && rt->mfc_ttls[vifi] &&
-		    (ifp->if_flags & IFF_BROADCAST) &&
-		    !(viftable[vifi].v_flags & VIFF_TUNNEL)) {
+		if (pim_assert && (vifi < numvifs) && viftable[vifi].v_ifp) {
 			struct timeval now;
 			u_int32_t delta;
 
+#ifdef PIM
+			if (ifp == &multicast_register_if)
+				pimstat.pims_rcv_registers_wrongiif++;
+#endif
+
 			/* Get vifi for the incoming packet */
 			for (vifi = 0;
 			     vifi < numvifs && viftable[vifi].v_ifp != ifp;
@@ -1403,6 +1743,12 @@ ip_mdq(struct mbuf *m, struct ifnet *ifp, struct mfc *rt)
 				return (0);
 			}
 
+			if (rt->mfc_flags[vifi] &
+			    MRT_MFC_FLAGS_DISABLE_WRONGVIF) {
+				/* WRONGVIF disabled: ignore the packet */
+				return (0);
+			}
+
 			microtime(&now);
 
 			TV_DELTA(rt->mfc_last_assert, now, delta);
@@ -1459,9 +1805,26 @@ ip_mdq(struct mbuf *m, struct ifnet *ifp, struct mfc *rt)
 			(ip->ip_ttl > rt->mfc_ttls[vifi])) {
 			vifp->v_pkt_out++;
 			vifp->v_bytes_out += plen;
+#ifdef PIM
+			if (vifp->v_flags & VIFF_REGISTER)
+				pim_register_send(ip, vifp, m, rt);
+			else
+#endif
 			MC_SEND(ip, vifp, m);
 		}
 
+	/*
+	 * Perform upcall-related bw measuring.
+	 */
+	if (rt->mfc_bw_meter != NULL) {
+		struct bw_meter *x;
+		struct timeval now;
+
+		microtime(&now);
+		for (x = rt->mfc_bw_meter; x != NULL; x = x->bm_mfc_next)
+			bw_meter_receive_packet(x, plen, &now);
+	}
+
 	return (0);
 }
 
@@ -2103,3 +2466,1053 @@ rsvp_input(struct mbuf *m, struct ifnet *ifp)
 	splx(s);
 }
 #endif /* RSVP_ISI */
+
+/*
+ * Code for bandwidth monitors
+ */
+
+/*
+ * Define common interface for timeval-related methods
+ */
+#define	BW_TIMEVALCMP(tvp, uvp, cmp) timercmp((tvp), (uvp), cmp)
+#define	BW_TIMEVALDECR(vvp, uvp) timersub((vvp), (uvp), (vvp))
+#define	BW_TIMEVALADD(vvp, uvp) timeradd((vvp), (uvp), (vvp))
+
+static uint32_t
+compute_bw_meter_flags(struct bw_upcall *req)
+{
+    uint32_t flags = 0;
+
+    if (req->bu_flags & BW_UPCALL_UNIT_PACKETS)
+	flags |= BW_METER_UNIT_PACKETS;
+    if (req->bu_flags & BW_UPCALL_UNIT_BYTES)
+	flags |= BW_METER_UNIT_BYTES;
+    if (req->bu_flags & BW_UPCALL_GEQ)
+	flags |= BW_METER_GEQ;
+    if (req->bu_flags & BW_UPCALL_LEQ)
+	flags |= BW_METER_LEQ;
+    
+    return flags;
+}
+ 
+/*
+ * Add a bw_meter entry
+ */
+static int
+add_bw_upcall(struct mbuf *m)
+{
+    int s;
+    struct mfc *mfc;
+    struct timeval delta = { BW_UPCALL_THRESHOLD_INTERVAL_MIN_SEC,
+		BW_UPCALL_THRESHOLD_INTERVAL_MIN_USEC };
+    struct timeval now;
+    struct bw_meter *x;
+    uint32_t flags;
+    struct bw_upcall *req;
+
+    if (m == NULL || m->m_len < sizeof(struct bw_upcall))
+	return EINVAL;
+
+    req = mtod(m, struct bw_upcall *);
+
+    if (!(mrt_api_config & MRT_MFC_BW_UPCALL))
+	return EOPNOTSUPP;
+
+    /* Test if the flags are valid */
+    if (!(req->bu_flags & (BW_UPCALL_UNIT_PACKETS | BW_UPCALL_UNIT_BYTES)))
+	return EINVAL;
+    if (!(req->bu_flags & (BW_UPCALL_GEQ | BW_UPCALL_LEQ)))
+	return EINVAL;
+    if ((req->bu_flags & (BW_UPCALL_GEQ | BW_UPCALL_LEQ))
+	    == (BW_UPCALL_GEQ | BW_UPCALL_LEQ))
+	return EINVAL;
+
+    /* Test if the threshold time interval is valid */
+    if (BW_TIMEVALCMP(&req->bu_threshold.b_time, &delta, <))
+	return EINVAL;
+
+    flags = compute_bw_meter_flags(req);
+
+    /*
+     * Find if we have already same bw_meter entry
+     */
+    s = splsoftnet();
+    mfc = mfc_find(&req->bu_src, &req->bu_dst);
+    if (mfc == NULL) {
+	splx(s);
+	return EADDRNOTAVAIL;
+    }
+    for (x = mfc->mfc_bw_meter; x != NULL; x = x->bm_mfc_next) {
+	if ((BW_TIMEVALCMP(&x->bm_threshold.b_time,
+			   &req->bu_threshold.b_time, ==)) &&
+	    (x->bm_threshold.b_packets == req->bu_threshold.b_packets) &&
+	    (x->bm_threshold.b_bytes == req->bu_threshold.b_bytes) &&
+	    (x->bm_flags & BW_METER_USER_FLAGS) == flags)  {
+	    splx(s);
+	    return 0;		/* XXX Already installed */
+	}
+    }
+
+    /* Allocate the new bw_meter entry */
+    x = (struct bw_meter *)malloc(sizeof(*x), M_BWMETER, M_NOWAIT);
+    if (x == NULL) {
+	splx(s);
+	return ENOBUFS;
+    }
+
+    /* Set the new bw_meter entry */
+    x->bm_threshold.b_time = req->bu_threshold.b_time;
+    microtime(&now);
+    x->bm_start_time = now;
+    x->bm_threshold.b_packets = req->bu_threshold.b_packets;
+    x->bm_threshold.b_bytes = req->bu_threshold.b_bytes;
+    x->bm_measured.b_packets = 0;
+    x->bm_measured.b_bytes = 0;
+    x->bm_flags = flags;
+    x->bm_time_next = NULL;
+    x->bm_time_hash = BW_METER_BUCKETS;
+
+    /* Add the new bw_meter entry to the front of entries for this MFC */
+    x->bm_mfc = mfc;
+    x->bm_mfc_next = mfc->mfc_bw_meter;
+    mfc->mfc_bw_meter = x;
+    schedule_bw_meter(x, &now);
+    splx(s);
+
+    return 0;
+}
+
+static void
+free_bw_list(struct bw_meter *list)
+{
+    while (list != NULL) {
+	struct bw_meter *x = list;
+
+	list = list->bm_mfc_next;
+	unschedule_bw_meter(x);
+	free(x, M_BWMETER);
+    }
+}
+
+/*
+ * Delete one or multiple bw_meter entries
+ */
+static int
+del_bw_upcall(struct mbuf *m)
+{
+    int s;
+    struct mfc *mfc;
+    struct bw_meter *x;
+    struct bw_upcall *req;
+
+    if (m == NULL || m->m_len < sizeof(struct bw_upcall))
+	return EINVAL;
+
+    req = mtod(m, struct bw_upcall *);
+    
+    if (!(mrt_api_config & MRT_MFC_BW_UPCALL))
+	return EOPNOTSUPP;
+
+    s = splsoftnet();    
+    /* Find the corresponding MFC entry */
+    mfc = mfc_find(&req->bu_src, &req->bu_dst);
+    if (mfc == NULL) {
+	splx(s);
+	return EADDRNOTAVAIL;
+    } else if (req->bu_flags & BW_UPCALL_DELETE_ALL) {
+	/*
+	 * Delete all bw_meter entries for this mfc
+	 */
+	struct bw_meter *list;
+
+	list = mfc->mfc_bw_meter;
+	mfc->mfc_bw_meter = NULL;
+	free_bw_list(list);
+	splx(s);
+	return 0;
+    } else {			/* Delete a single bw_meter entry */
+	struct bw_meter *prev;
+	uint32_t flags = 0;
+
+	flags = compute_bw_meter_flags(req);
+
+	/* Find the bw_meter entry to delete */
+	for (prev = NULL, x = mfc->mfc_bw_meter; x != NULL;
+	     prev = x, x = x->bm_mfc_next) {
+	    if ((BW_TIMEVALCMP(&x->bm_threshold.b_time,
+			       &req->bu_threshold.b_time, ==)) &&
+		(x->bm_threshold.b_packets == req->bu_threshold.b_packets) &&
+		(x->bm_threshold.b_bytes == req->bu_threshold.b_bytes) &&
+		(x->bm_flags & BW_METER_USER_FLAGS) == flags)
+		break;
+	}
+	if (x != NULL) { /* Delete entry from the list for this MFC */
+	    if (prev != NULL)
+		prev->bm_mfc_next = x->bm_mfc_next;	/* remove from middle*/
+	    else
+		x->bm_mfc->mfc_bw_meter = x->bm_mfc_next;/* new head of list */
+
+	    unschedule_bw_meter(x);
+	    splx(s);
+	    /* Free the bw_meter entry */
+	    free(x, M_BWMETER);
+	    return 0;
+	} else {
+	    splx(s);
+	    return EINVAL;
+	}
+    }
+    /* NOTREACHED */
+}
+
+/*
+ * Perform bandwidth measurement processing that may result in an upcall
+ */
+static void
+bw_meter_receive_packet(struct bw_meter *x, int plen, struct timeval *nowp)
+{
+    struct timeval delta;
+
+    delta = *nowp;
+    BW_TIMEVALDECR(&delta, &x->bm_start_time);
+
+    if (x->bm_flags & BW_METER_GEQ) {
+	/*
+	 * Processing for ">=" type of bw_meter entry
+	 */
+	if (BW_TIMEVALCMP(&delta, &x->bm_threshold.b_time, >)) {
+	    /* Reset the bw_meter entry */
+	    x->bm_start_time = *nowp;
+	    x->bm_measured.b_packets = 0;
+	    x->bm_measured.b_bytes = 0;
+	    x->bm_flags &= ~BW_METER_UPCALL_DELIVERED;
+	}
+
+	/* Record that a packet is received */
+	x->bm_measured.b_packets++;
+	x->bm_measured.b_bytes += plen;
+
+	/*
+	 * Test if we should deliver an upcall
+	 */
+	if (!(x->bm_flags & BW_METER_UPCALL_DELIVERED)) {	
+	    if (((x->bm_flags & BW_METER_UNIT_PACKETS) &&
+		 (x->bm_measured.b_packets >= x->bm_threshold.b_packets)) ||
+		((x->bm_flags & BW_METER_UNIT_BYTES) &&
+		 (x->bm_measured.b_bytes >= x->bm_threshold.b_bytes))) {
+		/* Prepare an upcall for delivery */
+		bw_meter_prepare_upcall(x, nowp);
+		x->bm_flags |= BW_METER_UPCALL_DELIVERED;
+	    }
+	}
+    } else if (x->bm_flags & BW_METER_LEQ) {
+	/*
+	 * Processing for "<=" type of bw_meter entry
+	 */
+	if (BW_TIMEVALCMP(&delta, &x->bm_threshold.b_time, >)) {
+	    /*
+	     * We are behind time with the multicast forwarding table
+	     * scanning for "<=" type of bw_meter entries, so test now
+	     * if we should deliver an upcall.
+	     */
+	    if (((x->bm_flags & BW_METER_UNIT_PACKETS) &&
+		 (x->bm_measured.b_packets <= x->bm_threshold.b_packets)) ||
+		((x->bm_flags & BW_METER_UNIT_BYTES) &&
+		 (x->bm_measured.b_bytes <= x->bm_threshold.b_bytes))) {
+		/* Prepare an upcall for delivery */
+		bw_meter_prepare_upcall(x, nowp);
+	    }
+	    /* Reschedule the bw_meter entry */
+	    unschedule_bw_meter(x);
+	    schedule_bw_meter(x, nowp);
+	}
+	
+	/* Record that a packet is received */
+	x->bm_measured.b_packets++;
+	x->bm_measured.b_bytes += plen;
+
+	/*
+	 * Test if we should restart the measuring interval
+	 */
+	if ((x->bm_flags & BW_METER_UNIT_PACKETS &&
+	     x->bm_measured.b_packets <= x->bm_threshold.b_packets) ||
+	    (x->bm_flags & BW_METER_UNIT_BYTES &&
+	     x->bm_measured.b_bytes <= x->bm_threshold.b_bytes)) {
+	    /* Don't restart the measuring interval */
+	} else {
+	    /* Do restart the measuring interval */
+	    /*
+	     * XXX: note that we don't unschedule and schedule, because this
+	     * might be too much overhead per packet. Instead, when we process
+	     * all entries for a given timer hash bin, we check whether it is
+	     * really a timeout. If not, we reschedule at that time.
+	     */
+	    x->bm_start_time = *nowp;
+	    x->bm_measured.b_packets = 0;
+	    x->bm_measured.b_bytes = 0;
+	    x->bm_flags &= ~BW_METER_UPCALL_DELIVERED;
+	}
+    }
+}
+
+/*
+ * Prepare a bandwidth-related upcall
+ */
+static void
+bw_meter_prepare_upcall(struct bw_meter *x, struct timeval *nowp)
+{
+    struct timeval delta;
+    struct bw_upcall *u;
+
+    /*
+     * Compute the measured time interval 
+     */
+    delta = *nowp;
+    BW_TIMEVALDECR(&delta, &x->bm_start_time);
+
+    /*
+     * If there are too many pending upcalls, deliver them now
+     */
+    if (bw_upcalls_n >= BW_UPCALLS_MAX)
+	bw_upcalls_send();
+
+    /*
+     * Set the bw_upcall entry
+     */
+    u = &bw_upcalls[bw_upcalls_n++];
+    u->bu_src = x->bm_mfc->mfc_origin;
+    u->bu_dst = x->bm_mfc->mfc_mcastgrp;
+    u->bu_threshold.b_time = x->bm_threshold.b_time;
+    u->bu_threshold.b_packets = x->bm_threshold.b_packets;
+    u->bu_threshold.b_bytes = x->bm_threshold.b_bytes;
+    u->bu_measured.b_time = delta;
+    u->bu_measured.b_packets = x->bm_measured.b_packets;
+    u->bu_measured.b_bytes = x->bm_measured.b_bytes;
+    u->bu_flags = 0;
+    if (x->bm_flags & BW_METER_UNIT_PACKETS)
+	u->bu_flags |= BW_UPCALL_UNIT_PACKETS;
+    if (x->bm_flags & BW_METER_UNIT_BYTES)
+	u->bu_flags |= BW_UPCALL_UNIT_BYTES;
+    if (x->bm_flags & BW_METER_GEQ)
+	u->bu_flags |= BW_UPCALL_GEQ;
+    if (x->bm_flags & BW_METER_LEQ)
+	u->bu_flags |= BW_UPCALL_LEQ;
+}
+
+/*
+ * Send the pending bandwidth-related upcalls
+ */
+static void
+bw_upcalls_send(void)
+{
+    struct mbuf *m;
+    int len = bw_upcalls_n * sizeof(bw_upcalls[0]);
+    struct sockaddr_in k_igmpsrc = { sizeof k_igmpsrc, AF_INET };
+    static struct igmpmsg igmpmsg = { 0,		/* unused1 */
+				      0,		/* unused2 */
+				      IGMPMSG_BW_UPCALL,/* im_msgtype */
+				      0,		/* im_mbz  */
+				      0,		/* im_vif  */
+				      0,		/* unused3 */
+				      { 0 },		/* im_src  */
+				      { 0 } };		/* im_dst  */
+    
+    if (bw_upcalls_n == 0)
+	return;			/* No pending upcalls */
+
+    bw_upcalls_n = 0;
+
+    /*
+     * Allocate a new mbuf, initialize it with the header and
+     * the payload for the pending calls.
+     */
+    MGETHDR(m, M_DONTWAIT, MT_HEADER);
+    if (m == NULL) {
+	log(LOG_WARNING, "bw_upcalls_send: cannot allocate mbuf\n");
+	return;
+    }
+
+    m->m_len = m->m_pkthdr.len = 0;
+    m_copyback(m, 0, sizeof(struct igmpmsg), (caddr_t)&igmpmsg);
+    m_copyback(m, sizeof(struct igmpmsg), len, (caddr_t)&bw_upcalls[0]);
+
+    /*
+     * Send the upcalls
+     * XXX do we need to set the address in k_igmpsrc ?
+     */
+    mrtstat.mrts_upcalls++;
+    if (socket_send(ip_mrouter, m, &k_igmpsrc) < 0) {
+	log(LOG_WARNING, "bw_upcalls_send: ip_mrouter socket queue full\n");
+	++mrtstat.mrts_upq_sockfull;
+    }
+}
+
+/*
+ * Compute the timeout hash value for the bw_meter entries
+ */
+#define	BW_METER_TIMEHASH(bw_meter, hash)				\
+    do {								\
+	struct timeval next_timeval = (bw_meter)->bm_start_time;	\
+									\
+	BW_TIMEVALADD(&next_timeval, &(bw_meter)->bm_threshold.b_time); \
+	(hash) = next_timeval.tv_sec;					\
+	if (next_timeval.tv_usec)					\
+	    (hash)++; /* XXX: make sure we don't timeout early */	\
+	(hash) %= BW_METER_BUCKETS;					\
+    } while (/*CONSTCOND*/ 0)
+
+/*
+ * Schedule a timer to process periodically bw_meter entry of type "<="
+ * by linking the entry in the proper hash bucket.
+ */
+static void
+schedule_bw_meter(struct bw_meter *x, struct timeval *nowp)
+{
+    int time_hash;
+
+    if (!(x->bm_flags & BW_METER_LEQ))
+	return;		/* XXX: we schedule timers only for "<=" entries */
+
+    /*
+     * Reset the bw_meter entry
+     */
+    x->bm_start_time = *nowp;
+    x->bm_measured.b_packets = 0;
+    x->bm_measured.b_bytes = 0;
+    x->bm_flags &= ~BW_METER_UPCALL_DELIVERED;
+
+    /*
+     * Compute the timeout hash value and insert the entry
+     */
+    BW_METER_TIMEHASH(x, time_hash);
+    x->bm_time_next = bw_meter_timers[time_hash];
+    bw_meter_timers[time_hash] = x;
+    x->bm_time_hash = time_hash;
+}
+
+/*
+ * Unschedule the periodic timer that processes bw_meter entry of type "<="
+ * by removing the entry from the proper hash bucket.
+ */
+static void
+unschedule_bw_meter(struct bw_meter *x)
+{
+    int time_hash;
+    struct bw_meter *prev, *tmp;
+
+    if (!(x->bm_flags & BW_METER_LEQ))
+	return;		/* XXX: we schedule timers only for "<=" entries */
+
+    /*
+     * Compute the timeout hash value and delete the entry
+     */
+    time_hash = x->bm_time_hash;
+    if (time_hash >= BW_METER_BUCKETS)
+	return;		/* Entry was not scheduled */
+
+    for (prev = NULL, tmp = bw_meter_timers[time_hash];
+	     tmp != NULL; prev = tmp, tmp = tmp->bm_time_next)
+	if (tmp == x)
+	    break;
+
+    if (tmp == NULL)
+	panic("unschedule_bw_meter: bw_meter entry not found");
+
+    if (prev != NULL)
+	prev->bm_time_next = x->bm_time_next;
+    else
+	bw_meter_timers[time_hash] = x->bm_time_next;
+
+    x->bm_time_next = NULL;
+    x->bm_time_hash = BW_METER_BUCKETS;
+}
+
+/*
+ * Process all "<=" type of bw_meter that should be processed now,
+ * and for each entry prepare an upcall if necessary. Each processed
+ * entry is rescheduled again for the (periodic) processing.
+ *
+ * This is run periodically (once per second normally). On each round,
+ * all the potentially matching entries are in the hash slot that we are
+ * looking at.
+ */
+static void
+bw_meter_process()
+{
+    int s;
+    static uint32_t last_tv_sec;	/* last time we processed this */
+
+    uint32_t loops;
+    int i;
+    struct timeval now, process_endtime;
+
+    microtime(&now);
+    if (last_tv_sec == now.tv_sec)
+	return;		/* nothing to do */
+
+    loops = now.tv_sec - last_tv_sec;
+    last_tv_sec = now.tv_sec;
+    if (loops > BW_METER_BUCKETS)
+	loops = BW_METER_BUCKETS;
+
+    s = splsoftnet();
+    /*
+     * Process all bins of bw_meter entries from the one after the last
+     * processed to the current one. On entry, i points to the last bucket
+     * visited, so we need to increment i at the beginning of the loop.
+     */
+    for (i = (now.tv_sec - loops) % BW_METER_BUCKETS; loops > 0; loops--) {
+	struct bw_meter *x, *tmp_list;
+
+	if (++i >= BW_METER_BUCKETS)
+	    i = 0;
+
+	/* Disconnect the list of bw_meter entries from the bin */
+	tmp_list = bw_meter_timers[i];
+	bw_meter_timers[i] = NULL;
+
+	/* Process the list of bw_meter entries */
+	while (tmp_list != NULL) {
+	    x = tmp_list;
+	    tmp_list = tmp_list->bm_time_next;
+
+	    /* Test if the time interval is over */
+	    process_endtime = x->bm_start_time;
+	    BW_TIMEVALADD(&process_endtime, &x->bm_threshold.b_time);
+	    if (BW_TIMEVALCMP(&process_endtime, &now, >)) {
+		/* Not yet: reschedule, but don't reset */
+		int time_hash;
+
+		BW_METER_TIMEHASH(x, time_hash);
+		if (time_hash == i && process_endtime.tv_sec == now.tv_sec) {
+		    /*
+		     * XXX: somehow the bin processing is a bit ahead of time.
+		     * Put the entry in the next bin.
+		     */
+		    if (++time_hash >= BW_METER_BUCKETS)
+			time_hash = 0;
+		}
+		x->bm_time_next = bw_meter_timers[time_hash];
+		bw_meter_timers[time_hash] = x;
+		x->bm_time_hash = time_hash;
+
+		continue;
+	    }
+
+	    /*
+	     * Test if we should deliver an upcall
+	     */
+	    if (((x->bm_flags & BW_METER_UNIT_PACKETS) &&
+		 (x->bm_measured.b_packets <= x->bm_threshold.b_packets)) ||
+		((x->bm_flags & BW_METER_UNIT_BYTES) &&
+		 (x->bm_measured.b_bytes <= x->bm_threshold.b_bytes))) {
+		/* Prepare an upcall for delivery */
+		bw_meter_prepare_upcall(x, &now);
+	    }
+
+	    /*
+	     * Reschedule for next processing
+	     */
+	    schedule_bw_meter(x, &now);
+	}
+    }
+
+    /* Send all upcalls that are pending delivery */
+    bw_upcalls_send();
+
+    splx(s);
+}
+
+/*
+ * A periodic function for sending all upcalls that are pending delivery
+ */
+static void
+expire_bw_upcalls_send(void *unused)
+{
+    int s;
+
+    s = splsoftnet();
+    bw_upcalls_send();
+    splx(s);
+
+    timeout_add(&bw_upcalls_ch, BW_UPCALLS_PERIOD);
+}
+
+/*
+ * A periodic function for periodic scanning of the multicast forwarding
+ * table for processing all "<=" bw_meter entries.
+ */
+static void
+expire_bw_meter_process(void *unused)
+{
+    if (mrt_api_config & MRT_MFC_BW_UPCALL)
+	bw_meter_process();
+
+    timeout_add(&bw_meter_ch, BW_METER_PERIOD);
+}
+
+/*
+ * End of bandwidth monitoring code
+ */
+
+#ifdef PIM
+/*
+ * Send the packet up to the user daemon, or eventually do kernel encapsulation
+ */
+static int
+pim_register_send(struct ip *ip, struct vif *vifp,
+	struct mbuf *m, struct mfc *rt)
+{
+    struct mbuf *mb_copy, *mm;
+
+    if (mrtdebug & DEBUG_PIM)
+        log(LOG_DEBUG, "pim_register_send: ");
+
+    mb_copy = pim_register_prepare(ip, m);
+    if (mb_copy == NULL)
+	return ENOBUFS;
+
+    /*
+     * Send all the fragments. Note that the mbuf for each fragment
+     * is freed by the sending machinery.
+     */
+    for (mm = mb_copy; mm; mm = mb_copy) {
+	mb_copy = mm->m_nextpkt;
+	mm->m_nextpkt = NULL;
+	mm = m_pullup(mm, sizeof(struct ip));
+	if (mm != NULL) {
+	    ip = mtod(mm, struct ip *);
+	    if ((mrt_api_config & MRT_MFC_RP) &&
+		!in_nullhost(rt->mfc_rp)) {
+		pim_register_send_rp(ip, vifp, mm, rt);
+	    } else {
+		pim_register_send_upcall(ip, vifp, mm, rt);
+	    }
+	}
+    }
+
+    return 0;
+}
+
+/*
+ * Return a copy of the data packet that is ready for PIM Register
+ * encapsulation.
+ * XXX: Note that in the returned copy the IP header is a valid one.
+ */
+static struct mbuf *
+pim_register_prepare(struct ip *ip, struct mbuf *m)
+{
+    struct mbuf *mb_copy = NULL;
+    int mtu;
+
+    /* Take care of delayed checksums */
+    if (m->m_pkthdr.csum & (M_TCPV4_CSUM_OUT | M_UDPV4_CSUM_OUT)) {
+	in_delayed_cksum(m);
+	m->m_pkthdr.csum &= ~(M_UDPV4_CSUM_OUT | M_TCPV4_CSUM_OUT);
+    }
+
+    /*
+     * Copy the old packet & pullup its IP header into the
+     * new mbuf so we can modify it.
+     */
+    mb_copy = m_copy(m, 0, M_COPYALL);
+    if (mb_copy == NULL)
+	return NULL;
+    mb_copy = m_pullup(mb_copy, ip->ip_hl << 2);
+    if (mb_copy == NULL)
+	return NULL;
+
+    /* take care of the TTL */
+    ip = mtod(mb_copy, struct ip *);
+    --ip->ip_ttl;
+
+    /* Compute the MTU after the PIM Register encapsulation */
+    mtu = 0xffff - sizeof(pim_encap_iphdr) - sizeof(pim_encap_pimhdr);
+
+    if (ntohs(ip->ip_len) <= mtu) {
+	/* Turn the IP header into a valid one */
+	ip->ip_sum = 0;
+	ip->ip_sum = in_cksum(mb_copy, ip->ip_hl << 2);
+    } else {
+	/* Fragment the packet */
+	if (ip_fragment(mb_copy, NULL, mtu) != 0) {
+	    /* XXX: mb_copy was freed by ip_fragment() */
+	    return NULL;
+	}
+    }
+    return mb_copy;
+}
+
+/*
+ * Send an upcall with the data packet to the user-level process.
+ */
+static int
+pim_register_send_upcall(struct ip *ip, struct vif *vifp,
+	struct mbuf *mb_copy, struct mfc *rt)
+{
+    struct mbuf *mb_first;
+    int len = ntohs(ip->ip_len);
+    struct igmpmsg *im;
+    struct sockaddr_in k_igmpsrc = { sizeof k_igmpsrc, AF_INET };
+
+    /*
+     * Add a new mbuf with an upcall header
+     */
+    MGETHDR(mb_first, M_DONTWAIT, MT_HEADER);
+    if (mb_first == NULL) {
+	m_freem(mb_copy);
+	return ENOBUFS;
+    }
+    mb_first->m_data += max_linkhdr;
+    mb_first->m_pkthdr.len = len + sizeof(struct igmpmsg);
+    mb_first->m_len = sizeof(struct igmpmsg);
+    mb_first->m_next = mb_copy;
+
+    /* Send message to routing daemon */
+    im = mtod(mb_first, struct igmpmsg *);
+    im->im_msgtype	= IGMPMSG_WHOLEPKT;
+    im->im_mbz		= 0;
+    im->im_vif		= vifp - viftable;
+    im->im_src		= ip->ip_src;
+    im->im_dst		= ip->ip_dst;
+
+    k_igmpsrc.sin_addr	= ip->ip_src;
+
+    mrtstat.mrts_upcalls++;
+
+    if (socket_send(ip_mrouter, mb_first, &k_igmpsrc) < 0) {
+	if (mrtdebug & DEBUG_PIM)
+	    log(LOG_WARNING,
+		"mcast: pim_register_send_upcall: ip_mrouter socket queue full");
+	++mrtstat.mrts_upq_sockfull;
+	return ENOBUFS;
+    }
+
+    /* Keep statistics */
+    pimstat.pims_snd_registers_msgs++;
+    pimstat.pims_snd_registers_bytes += len;
+
+    return 0;
+}
+
+/*
+ * Encapsulate the data packet in PIM Register message and send it to the RP.
+ */
+static int
+pim_register_send_rp(struct ip *ip, struct vif *vifp,
+	struct mbuf *mb_copy, struct mfc *rt)
+{
+    struct mbuf *mb_first;
+    struct ip *ip_outer;
+    struct pim_encap_pimhdr *pimhdr;
+    int len = ntohs(ip->ip_len);
+    vifi_t vifi = rt->mfc_parent;
+
+    if ((vifi >= numvifs) || in_nullhost(viftable[vifi].v_lcl_addr)) {
+	m_freem(mb_copy);
+	return EADDRNOTAVAIL;		/* The iif vif is invalid */
+    }
+
+    /*
+     * Add a new mbuf with the encapsulating header
+     */
+    MGETHDR(mb_first, M_DONTWAIT, MT_HEADER);
+    if (mb_first == NULL) {
+	m_freem(mb_copy);
+	return ENOBUFS;
+    }
+    mb_first->m_data += max_linkhdr;
+    mb_first->m_len = sizeof(pim_encap_iphdr) + sizeof(pim_encap_pimhdr);
+    mb_first->m_next = mb_copy;
+
+    mb_first->m_pkthdr.len = len + mb_first->m_len;
+
+    /*
+     * Fill in the encapsulating IP and PIM header
+     */
+    ip_outer = mtod(mb_first, struct ip *);
+    *ip_outer = pim_encap_iphdr;
+    ip_outer->ip_id = htons(ip_randomid());
+    ip_outer->ip_len = htons(len + sizeof(pim_encap_iphdr) +
+			     sizeof(pim_encap_pimhdr));
+    ip_outer->ip_src = viftable[vifi].v_lcl_addr;
+    ip_outer->ip_dst = rt->mfc_rp;
+    /*
+     * Copy the inner header TOS to the outer header, and take care of the
+     * IP_DF bit.
+     */
+    ip_outer->ip_tos = ip->ip_tos;
+    if (ntohs(ip->ip_off) & IP_DF)
+	ip_outer->ip_off |= IP_DF;
+    pimhdr = (struct pim_encap_pimhdr *)((caddr_t)ip_outer
+					 + sizeof(pim_encap_iphdr));
+    *pimhdr = pim_encap_pimhdr;
+    /* If the iif crosses a border, set the Border-bit */
+    if (rt->mfc_flags[vifi] & MRT_MFC_FLAGS_BORDER_VIF & mrt_api_config)
+	pimhdr->flags |= htonl(PIM_BORDER_REGISTER);
+
+    mb_first->m_data += sizeof(pim_encap_iphdr);
+    pimhdr->pim.pim_cksum = in_cksum(mb_first, sizeof(pim_encap_pimhdr));
+    mb_first->m_data -= sizeof(pim_encap_iphdr);
+
+    if (vifp->v_rate_limit == 0)
+	tbf_send_packet(vifp, mb_first);
+    else
+	tbf_control(vifp, mb_first, ip, ntohs(ip_outer->ip_len));
+
+    /* Keep statistics */
+    pimstat.pims_snd_registers_msgs++;
+    pimstat.pims_snd_registers_bytes += len;
+
+    return 0;
+}
+
+/*
+ * PIM-SMv2 and PIM-DM messages processing.
+ * Receives and verifies the PIM control messages, and passes them
+ * up to the listening socket, using rip_input().
+ * The only message with special processing is the PIM_REGISTER message
+ * (used by PIM-SM): the PIM header is stripped off, and the inner packet
+ * is passed to if_simloop().
+ */
+void
+pim_input(struct mbuf *m, ...)
+{
+    struct ip *ip = mtod(m, struct ip *);
+    struct pim *pim;
+    int minlen;
+    int datalen;
+    int ip_tos;
+    int iphlen;
+    va_list ap;
+
+    va_start(ap, m);
+    iphlen = va_arg(ap, int);
+    va_end(ap);
+
+    datalen = ntohs(ip->ip_len) - iphlen;
+
+    /* Keep statistics */
+    pimstat.pims_rcv_total_msgs++;
+    pimstat.pims_rcv_total_bytes += datalen;
+
+    /*
+     * Validate lengths
+     */
+    if (datalen < PIM_MINLEN) {
+	pimstat.pims_rcv_tooshort++;
+	log(LOG_ERR, "pim_input: packet size too small %d from %lx\n",
+	    datalen, (u_long)ip->ip_src.s_addr);
+	m_freem(m);
+	return;
+    }
+
+    /*
+     * If the packet is at least as big as a REGISTER, go agead
+     * and grab the PIM REGISTER header size, to avoid another
+     * possible m_pullup() later.
+     * 
+     * PIM_MINLEN       == pimhdr + u_int32_t == 4 + 4 = 8
+     * PIM_REG_MINLEN   == pimhdr + reghdr + encap_iphdr == 4 + 4 + 20 = 28
+     */
+    minlen = iphlen + (datalen >= PIM_REG_MINLEN ? PIM_REG_MINLEN : PIM_MINLEN);
+    /*
+     * Get the IP and PIM headers in contiguous memory, and
+     * possibly the PIM REGISTER header.
+     */
+    if ((m->m_flags & M_EXT || m->m_len < minlen) &&
+	(m = m_pullup(m, minlen)) == NULL) {
+	log(LOG_ERR, "pim_input: m_pullup failure\n");
+	return;
+    }
+    /* m_pullup() may have given us a new mbuf so reset ip. */
+    ip = mtod(m, struct ip *);
+    ip_tos = ip->ip_tos;
+
+    /* adjust mbuf to point to the PIM header */
+    m->m_data += iphlen;
+    m->m_len  -= iphlen;
+    pim = mtod(m, struct pim *);
+
+    /*
+     * Validate checksum. If PIM REGISTER, exclude the data packet.
+     *
+     * XXX: some older PIMv2 implementations don't make this distinction,
+     * so for compatibility reason perform the checksum over part of the
+     * message, and if error, then over the whole message.
+     */
+    if (PIM_VT_T(pim->pim_vt) == PIM_REGISTER && in_cksum(m, PIM_MINLEN) == 0) {
+	/* do nothing, checksum okay */
+    } else if (in_cksum(m, datalen)) {
+	pimstat.pims_rcv_badsum++;
+	if (mrtdebug & DEBUG_PIM)
+	    log(LOG_DEBUG, "pim_input: invalid checksum");
+	m_freem(m);
+	return;
+    }
+
+    /* PIM version check */
+    if (PIM_VT_V(pim->pim_vt) < PIM_VERSION) {
+	pimstat.pims_rcv_badversion++;
+	log(LOG_ERR, "pim_input: incorrect version %d, expecting %d\n",
+	    PIM_VT_V(pim->pim_vt), PIM_VERSION);
+	m_freem(m);
+	return;
+    }
+
+    /* restore mbuf back to the outer IP */
+    m->m_data -= iphlen;
+    m->m_len  += iphlen;
+
+    if (PIM_VT_T(pim->pim_vt) == PIM_REGISTER) {
+	/*
+	 * Since this is a REGISTER, we'll make a copy of the register
+	 * headers ip + pim + u_int32 + encap_ip, to be passed up to the
+	 * routing daemon.
+	 */
+	int s;
+	struct sockaddr_in dst = { sizeof(dst), AF_INET };
+	struct mbuf *mcp;
+	struct ip *encap_ip;
+	u_int32_t *reghdr;
+	struct ifnet *vifp;
+
+	s = splsoftnet();
+	if ((reg_vif_num >= numvifs) || (reg_vif_num == VIFI_INVALID)) {
+	    splx(s);
+	    if (mrtdebug & DEBUG_PIM)
+		log(LOG_DEBUG,
+		    "pim_input: register vif not set: %d\n", reg_vif_num);
+	    m_freem(m);
+	    return;
+	}
+	/* XXX need refcnt? */
+	vifp = viftable[reg_vif_num].v_ifp;
+	splx(s);
+
+	/*
+	 * Validate length
+	 */
+	if (datalen < PIM_REG_MINLEN) {
+	    pimstat.pims_rcv_tooshort++;
+	    pimstat.pims_rcv_badregisters++;
+	    log(LOG_ERR,
+		"pim_input: register packet size too small %d from %lx\n",
+		datalen, (u_long)ip->ip_src.s_addr);
+	    m_freem(m);
+	    return;
+	}
+
+	reghdr = (u_int32_t *)(pim + 1);
+	encap_ip = (struct ip *)(reghdr + 1);
+
+	if (mrtdebug & DEBUG_PIM) {
+	    log(LOG_DEBUG,
+		"pim_input[register], encap_ip: %lx -> %lx, encap_ip len %d\n",
+		(u_long)ntohl(encap_ip->ip_src.s_addr),
+		(u_long)ntohl(encap_ip->ip_dst.s_addr),
+		ntohs(encap_ip->ip_len));
+	}
+
+	/* verify the version number of the inner packet */
+	if (encap_ip->ip_v != IPVERSION) {
+	    pimstat.pims_rcv_badregisters++;
+	    if (mrtdebug & DEBUG_PIM) {
+		log(LOG_DEBUG, "pim_input: invalid IP version (%d) "
+		    "of the inner packet\n", encap_ip->ip_v);
+	    }
+	    m_freem(m);
+	    return;
+	}
+
+	/* verify the inner packet is destined to a mcast group */
+	if (!IN_MULTICAST(encap_ip->ip_dst.s_addr)) {
+	    pimstat.pims_rcv_badregisters++;
+	    if (mrtdebug & DEBUG_PIM)
+		log(LOG_DEBUG,
+		    "pim_input: inner packet of register is not "
+		    "multicast %lx\n",
+		    (u_long)ntohl(encap_ip->ip_dst.s_addr));
+	    m_freem(m);
+	    return;
+	}
+
+	/* If a NULL_REGISTER, pass it to the daemon */
+	if ((ntohl(*reghdr) & PIM_NULL_REGISTER))
+	    goto pim_input_to_daemon;
+
+	/*
+	 * Copy the TOS from the outer IP header to the inner IP header.
+	 */
+	if (encap_ip->ip_tos != ip_tos) {
+	    /* Outer TOS -> inner TOS */
+	    encap_ip->ip_tos = ip_tos;
+	    /* Recompute the inner header checksum. Sigh... */
+
+	    /* adjust mbuf to point to the inner IP header */
+	    m->m_data += (iphlen + PIM_MINLEN);
+	    m->m_len  -= (iphlen + PIM_MINLEN);
+
+	    encap_ip->ip_sum = 0;
+	    encap_ip->ip_sum = in_cksum(m, encap_ip->ip_hl << 2);
+
+	    /* restore mbuf to point back to the outer IP header */
+	    m->m_data -= (iphlen + PIM_MINLEN);
+	    m->m_len  += (iphlen + PIM_MINLEN);
+	}
+
+	/*
+	 * Decapsulate the inner IP packet and loopback to forward it
+	 * as a normal multicast packet. Also, make a copy of the 
+	 *     outer_iphdr + pimhdr + reghdr + encap_iphdr
+	 * to pass to the daemon later, so it can take the appropriate
+	 * actions (e.g., send back PIM_REGISTER_STOP).
+	 * XXX: here m->m_data points to the outer IP header.
+	 */
+	mcp = m_copy(m, 0, iphlen + PIM_REG_MINLEN);
+	if (mcp == NULL) {
+	    log(LOG_ERR,
+		"pim_input: pim register: could not copy register head\n");
+	    m_freem(m);
+	    return;
+	}
+
+	/* Keep statistics */
+	/* XXX: registers_bytes include only the encap. mcast pkt */
+	pimstat.pims_rcv_registers_msgs++;
+	pimstat.pims_rcv_registers_bytes += ntohs(encap_ip->ip_len);
+
+	/*
+	 * forward the inner ip packet; point m_data at the inner ip.
+	 */
+	m_adj(m, iphlen + PIM_MINLEN);
+
+	if (mrtdebug & DEBUG_PIM) {
+	    log(LOG_DEBUG,
+		"pim_input: forwarding decapsulated register: "
+		"src %lx, dst %lx, vif %d\n",
+		(u_long)ntohl(encap_ip->ip_src.s_addr),
+		(u_long)ntohl(encap_ip->ip_dst.s_addr),
+		reg_vif_num);
+	}
+	/* NB: vifp was collected above; can it change on us? */
+	looutput(vifp, m, (struct sockaddr *)&dst, (struct rtentry *)NULL);
+
+	/* prepare the register head to send to the mrouting daemon */
+	m = mcp;
+    }
+
+pim_input_to_daemon:
+    /*
+     * Pass the PIM message up to the daemon; if it is a Register message,
+     * pass the 'head' only up to the daemon. This includes the
+     * outer IP header, PIM header, PIM-Register header and the
+     * inner IP header.
+     * XXX: the outer IP header pkt size of a Register is not adjust to
+     * reflect the fact that the inner multicast data is truncated.
+     */
+    rip_input(m);
+
+    return;
+}
+#endif /* PIM */
diff --git a/sys/netinet/ip_mroute.h b/sys/netinet/ip_mroute.h
index 747960e8bf6..4ab30d9c610 100644
--- a/sys/netinet/ip_mroute.h
+++ b/sys/netinet/ip_mroute.h
@@ -1,4 +1,4 @@
-/*	$OpenBSD: ip_mroute.h,v 1.11 2004/11/24 01:25:42 mcbride Exp $	*/
+/*	$OpenBSD: ip_mroute.h,v 1.12 2005/01/14 14:51:28 mcbride Exp $	*/
 /*	$NetBSD: ip_mroute.h,v 1.23 2004/04/21 17:49:46 itojun Exp $	*/
 
 #ifndef _NETINET_IP_MROUTE_H_
@@ -11,8 +11,12 @@
  * Modified by Steve Deering, Stanford, February 1989.
  * Modified by Ajit Thyagarajan, PARC, August 1993.
  * Modified by Ajit Thyagarajan, PARC, August 1994.
+ * Modified by Ahmed Helmy, SGI, June 1996.
+ * Modified by Pavlin Radoslavov, ICSI, October 2002.
  *
  * MROUTING Revision: 1.2
+ * and PIM-SMv2 and PIM-DM support, advanced API support,
+ * bandwidth metering and signaling.
  */
 
 #include <sys/queue.h>
@@ -29,6 +33,11 @@
 #define	MRT_DEL_MFC		105	/* delete forwarding cache entry */
 #define	MRT_VERSION		106	/* get kernel version number */
 #define	MRT_ASSERT		107	/* enable assert processing */
+#define	MRT_PIM			MRT_ASSERT /* enable PIM processing */
+#define	MRT_API_SUPPORT		109	/* supported MRT API */
+#define	MRT_API_CONFIG		110	/* config MRT API */
+#define	MRT_ADD_BW_UPCALL	111	/* create bandwidth monitor */
+#define	MRT_DEL_BW_UPCALL	112	/* delete bandwidth monitor */
 
 
 /*
@@ -48,6 +57,7 @@ typedef u_int16_t vifi_t;		/* type of a vif index */
 
 #define	VIFF_TUNNEL	0x1		/* vif represents a tunnel end-point */
 #define	VIFF_SRCRT	0x2		/* tunnel uses IP src routing */
+#define	VIFF_REGISTER	0x4		/* used for PIM Register encap/decap */
 
 /*
  * Argument structure for MRT_ADD_VIF.
@@ -64,6 +74,7 @@ struct vifctl {
 
 /*
  * Argument structure for MRT_ADD_MFC and MRT_DEL_MFC.
+ * XXX if you change this, make sure to change struct mfcctl2 as well.
  */
 struct mfcctl {
 	struct	 in_addr mfcc_origin;	/* ip origin of mcasts */
@@ -73,6 +84,94 @@ struct mfcctl {
 };
 
 /*
+ * The new argument structure for MRT_ADD_MFC and MRT_DEL_MFC overlays
+ * and extends the old struct mfcctl.
+ */
+struct mfcctl2 {
+	/* the mfcctl fields */
+	struct in_addr	mfcc_origin;		/* ip origin of mcasts	     */
+	struct in_addr	mfcc_mcastgrp;		/* multicast group associated*/
+	vifi_t		mfcc_parent;		/* incoming vif		     */
+	u_int8_t	mfcc_ttls[MAXVIFS]; 	/* forwarding ttls on vifs   */
+
+	/* extension fields */
+	u_int8_t	mfcc_flags[MAXVIFS];	/* the MRT_MFC_FLAGS_* flags */
+	struct in_addr	mfcc_rp;		/* the RP address            */
+};
+/*
+ * The advanced-API flags.
+ *
+ * The MRT_MFC_FLAGS_XXX API flags are also used as flags
+ * for the mfcc_flags field.
+ */
+#define	MRT_MFC_FLAGS_DISABLE_WRONGVIF	(1 << 0) /* disable WRONGVIF signals */
+#define	MRT_MFC_FLAGS_BORDER_VIF	(1 << 1) /* border vif		     */
+#define	MRT_MFC_RP			(1 << 8) /* enable RP address	     */
+#define	MRT_MFC_BW_UPCALL		(1 << 9) /* enable bw upcalls	     */
+#define	MRT_MFC_FLAGS_ALL		(MRT_MFC_FLAGS_DISABLE_WRONGVIF |    \
+					 MRT_MFC_FLAGS_BORDER_VIF)
+#define	MRT_API_FLAGS_ALL		(MRT_MFC_FLAGS_ALL |		     \
+					 MRT_MFC_RP |			     \
+					 MRT_MFC_BW_UPCALL)
+
+/*
+ * Structure for installing or delivering an upcall if the
+ * measured bandwidth is above or below a threshold.
+ *
+ * User programs (e.g. daemons) may have a need to know when the
+ * bandwidth used by some data flow is above or below some threshold.
+ * This interface allows the userland to specify the threshold (in
+ * bytes and/or packets) and the measurement interval. Flows are
+ * all packet with the same source and destination IP address.
+ * At the moment the code is only used for multicast destinations
+ * but there is nothing that prevents its use for unicast.
+ *
+ * The measurement interval cannot be shorter than some Tmin (currently, 3s).
+ * The threshold is set in packets and/or bytes per_interval.
+ *
+ * Measurement works as follows:
+ *
+ * For >= measurements: 
+ * The first packet marks the start of a measurement interval.
+ * During an interval we count packets and bytes, and when we
+ * pass the threshold we deliver an upcall and we are done.
+ * The first packet after the end of the interval resets the
+ * count and restarts the measurement.
+ *
+ * For <= measurement:
+ * We start a timer to fire at the end of the interval, and
+ * then for each incoming packet we count packets and bytes.
+ * When the timer fires, we compare the value with the threshold,
+ * schedule an upcall if we are below, and restart the measurement
+ * (reschedule timer and zero counters).
+ */
+
+struct bw_data {
+	struct timeval	b_time;
+	u_int64_t	b_packets;
+	u_int64_t	b_bytes;
+};
+
+struct bw_upcall {
+	struct in_addr	bu_src;			/* source address            */
+	struct in_addr	bu_dst;			/* destination address       */
+	u_int32_t	bu_flags;		/* misc flags (see below)    */
+#define	BW_UPCALL_UNIT_PACKETS   (1 << 0)	/* threshold (in packets)    */
+#define	BW_UPCALL_UNIT_BYTES     (1 << 1)	/* threshold (in bytes)      */
+#define	BW_UPCALL_GEQ            (1 << 2)	/* upcall if bw >= threshold */
+#define	BW_UPCALL_LEQ            (1 << 3)	/* upcall if bw <= threshold */
+#define	BW_UPCALL_DELETE_ALL     (1 << 4)	/* delete all upcalls for s,d*/
+	struct bw_data	bu_threshold;		/* the bw threshold	     */
+	struct bw_data	bu_measured;		/* the measured bw	     */
+};
+
+/* max. number of upcalls to deliver together */
+#define	BW_UPCALLS_MAX				128
+/* min. threshold time interval for bandwidth measurement */
+#define	BW_UPCALL_THRESHOLD_INTERVAL_MIN_SEC	3
+#define	BW_UPCALL_THRESHOLD_INTERVAL_MIN_USEC	0
+
+/*
  * Argument structure used by mrouted to get src-grp pkt counts.
  */
 struct sioc_sg_req {
@@ -162,6 +261,9 @@ struct mfc {
 	int	 mfc_expire;			/* time to clean entry up */
 	struct	 timeval mfc_last_assert;	/* last time I sent an assert */
 	struct	 rtdetq *mfc_stall;		/* pkts waiting for route */
+	u_int8_t mfc_flags[MAXVIFS];		/* the MRT_MFC_FLAGS_* flags */
+	struct in_addr	mfc_rp;			/* the RP address	     */
+	struct bw_meter	*mfc_bw_meter;		/* list of bandwidth meters  */
 };
 
 /*
@@ -172,8 +274,10 @@ struct igmpmsg {
 	u_int32_t unused1;
 	u_int32_t unused2;
 	u_int8_t  im_msgtype;		/* what type of message */
-#define IGMPMSG_NOCACHE		1	/* no MFC in the kernel		    */
-#define IGMPMSG_WRONGVIF	2	/* packet came from wrong interface */
+#define	IGMPMSG_NOCACHE		1	/* no MFC in the kernel		    */
+#define	IGMPMSG_WRONGVIF	2	/* packet came from wrong interface */
+#define	IGMPMSG_WHOLEPKT	3	/* PIM pkt for user level encap.    */
+#define	IGMPMSG_BW_UPCALL	4	/* BW monitoring upcall		    */
 	u_int8_t  im_mbz;		/* must be zero */
 	u_int8_t  im_vif;		/* vif rec'd on */
 	u_int8_t  unused3;
@@ -201,6 +305,32 @@ struct rtdetq {
 #define	MAX_BKT_SIZE    10000		/* 10K bytes size */
 #define	MAXQSIZE        10		/* max. no of pkts in token queue */
 
+/*
+ * Structure for measuring the bandwidth and sending an upcall if the
+ * measured bandwidth is above or below a threshold.
+ */
+struct bw_meter {
+	struct bw_meter	*bm_mfc_next;		/* next bw meter (same mfc)  */
+	struct bw_meter	*bm_time_next;		/* next bw meter (same time) */
+	uint32_t	bm_time_hash;		/* the time hash value       */
+	struct mfc	*bm_mfc;		/* the corresponding mfc     */
+	uint32_t	bm_flags;		/* misc flags (see below)    */
+#define	BW_METER_UNIT_PACKETS	(1 << 0)	/* threshold (in packets)    */
+#define	BW_METER_UNIT_BYTES	(1 << 1)	/* threshold (in bytes)      */
+#define	BW_METER_GEQ		(1 << 2)	/* upcall if bw >= threshold */
+#define	BW_METER_LEQ		(1 << 3)	/* upcall if bw <= threshold */
+#define	BW_METER_USER_FLAGS 	(BW_METER_UNIT_PACKETS |		\
+				 BW_METER_UNIT_BYTES |			\
+				 BW_METER_GEQ |				\
+				 BW_METER_LEQ)
+
+#define	BW_METER_UPCALL_DELIVERED (1 << 24)	/* upcall was delivered      */
+
+	struct bw_data	bm_threshold;		/* the upcall threshold	     */
+	struct bw_data	bm_measured;		/* the measured bw	     */
+	struct timeval	bm_start_time;		/* abs. time		     */
+};
+
 int	ip_mrouter_set(struct socket *, int, struct mbuf **);
 int	ip_mrouter_get(struct socket *, int, struct mbuf **);
 int	mrt_ioctl(struct socket *, u_long, caddr_t);
diff --git a/sys/netinet/pim.h b/sys/netinet/pim.h
new file mode 100644
index 00000000000..03c5ea76a25
--- /dev/null
+++ b/sys/netinet/pim.h
@@ -0,0 +1,122 @@
+/*	$OpenBSD: pim.h,v 1.1 2005/01/14 14:51:28 mcbride Exp $	*/
+/*	$NetBSD: pim.h,v 1.1 2004/09/04 23:32:29 manu Exp $	*/
+
+/*
+ * Copyright (c) 1996-2000
+ * University of Southern California/Information Sciences Institute.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ * 1. Redistributions of source code must retain the above copyright
+ *    notice, this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright
+ *    notice, this list of conditions and the following disclaimer in the
+ *    documentation and/or other materials provided with the distribution.
+ * 3. Neither the name of the project nor the names of its contributors
+ *    may be used to endorse or promote products derived from this software
+ *    without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND
+ * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+ * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
+ * ARE DISCLAIMED.  IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE
+ * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
+ * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
+ * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
+ * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
+ * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
+ * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
+ * SUCH DAMAGE.
+ *
+ * $FreeBSD: /repoman/r/ncvs/src/sys/netinet/pim.h,v 1.2 2004/03/08 07:45:32 hsu Exp $
+ */
+
+#ifndef _NETINET_PIM_H_
+#define _NETINET_PIM_H_
+
+/*
+ * Protocol Independent Multicast (PIM) definitions.
+ * RFC 2362, June 1998.
+ *
+ * Written by Ahmed Helmy, USC/SGI, July 1996.
+ * Modified by George Edmond Eddy (Rusty), ISI, February 1998.
+ * Modified by Pavlin Radoslavov, USC/ISI, May 1998, October 2000.
+ */
+
+#include <sys/types.h>
+
+#ifndef _PIM_VT
+#ifndef BYTE_ORDER
+# error BYTE_ORDER is not defined!
+#endif
+#if (BYTE_ORDER != BIG_ENDIAN) && (BYTE_ORDER != LITTLE_ENDIAN)
+# error BYTE_ORDER must be defined to either BIG_ENDIAN or LITTLE_ENDIAN
+#endif
+#endif /* ! _PIM_VT */
+
+/*
+ * PIM packet header
+ */
+struct pim {
+#ifdef _PIM_VT
+	uint8_t		pim_vt;		/* PIM version and message type	*/
+#else /* ! _PIM_VT   */
+#if BYTE_ORDER == BIG_ENDIAN
+	u_int		pim_vers:4,	/* PIM protocol version		*/
+			pim_type:4;	/* PIM message type		*/
+#endif
+#if BYTE_ORDER == LITTLE_ENDIAN
+	u_int		pim_type:4,	/* PIM message type		*/
+			pim_vers:4;	/* PIM protocol version		*/
+#endif
+#endif /* ! _PIM_VT  */
+	uint8_t		pim_reserved;	/* Reserved			*/
+	uint16_t	pim_cksum;	/* IP-style checksum		*/
+};
+/* KAME-related name backward compatibility */
+#define pim_ver pim_vers
+#define pim_rsv pim_reserved
+
+#ifdef _PIM_VT
+#define PIM_MAKE_VT(v, t)	(0xff & (((v) << 4) | (0x0f & (t))))
+#define PIM_VT_V(x)		(((x) >> 4) & 0x0f)
+#define PIM_VT_T(x)		((x) & 0x0f)
+#endif /* _PIM_VT */
+
+#define PIM_VERSION		2
+#define PIM_MINLEN		8	/* PIM message min. length	*/
+#define PIM_REG_MINLEN	(PIM_MINLEN+20)	/* PIM Register hdr + inner IPv4 hdr */
+#define PIM6_REG_MINLEN	(PIM_MINLEN+40)	/* PIM Register hdr + inner IPv6 hdr */
+
+/*
+ * PIM message types
+ */
+#define PIM_HELLO		0x0	/* PIM-SM and PIM-DM		*/
+#define PIM_REGISTER		0x1	/* PIM-SM only			*/
+#define PIM_REGISTER_STOP	0x2	/* PIM-SM only			*/
+#define PIM_JOIN_PRUNE		0x3	/* PIM-SM and PIM-DM		*/
+#define PIM_BOOTSTRAP		0x4	/* PIM-SM only			*/
+#define PIM_ASSERT		0x5	/* PIM-SM and PIM-DM		*/
+#define PIM_GRAFT		0x6	/* PIM-DM only			*/
+#define PIM_GRAFT_ACK		0x7	/* PIM-DM only			*/
+#define PIM_CAND_RP_ADV		0x8	/* PIM-SM only			*/
+#define PIM_ALL_DF_ELECTION	0xa	/* Bidir-PIM-SM only		*/
+
+/*
+ * PIM-Register message flags
+ */
+#define PIM_BORDER_REGISTER 0x80000000U	/* The Border bit (host-order)	*/
+#define PIM_NULL_REGISTER   0x40000000U	/* The Null-Register bit (host-order)*/
+
+/*
+ * All-PIM-Routers IPv4 and IPv6 multicast addresses
+ */
+#define INADDR_ALLPIM_ROUTERS_GROUP	(uint32_t)0xe000000dU  /* 224.0.0.13 */
+#define IN6ADDR_LINKLOCAL_ALLPIM_ROUTERS	"ff02::d"
+#define IN6ADDR_LINKLOCAL_ALLPIM_ROUTERS_INIT				\
+	{{{ 0xff, 0x02, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,		\
+	    0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x0d }}}
+
+#endif /* _NETINET_PIM_H_ */
diff --git a/sys/netinet/pim_var.h b/sys/netinet/pim_var.h
new file mode 100644
index 00000000000..2dc8c84ba75
--- /dev/null
+++ b/sys/netinet/pim_var.h
@@ -0,0 +1,82 @@
+/*	$OpenBSD: pim_var.h,v 1.1 2005/01/14 14:51:28 mcbride Exp $	*/
+/*	$NetBSD: pim_var.h,v 1.1 2004/09/04 23:32:29 manu Exp $	*/
+
+/*
+ * Copyright (c) 1998-2000
+ * University of Southern California/Information Sciences Institute.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ * 1. Redistributions of source code must retain the above copyright
+ *    notice, this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright
+ *    notice, this list of conditions and the following disclaimer in the
+ *    documentation and/or other materials provided with the distribution.
+ * 3. Neither the name of the project nor the names of its contributors
+ *    may be used to endorse or promote products derived from this software
+ *    without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND
+ * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+ * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
+ * ARE DISCLAIMED.  IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE
+ * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
+ * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
+ * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
+ * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
+ * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
+ * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
+ * SUCH DAMAGE.
+ *
+ * $FreeBSD: /repoman/r/ncvs/src/sys/netinet/pim_var.h,v 1.1 2003/08/07 18:17:43 hsu Exp $
+ */
+
+#ifndef _NETINET_PIM_VAR_H_
+#define _NETINET_PIM_VAR_H_
+
+/*
+ * Protocol Independent Multicast (PIM),
+ * kernel variables and implementation-specific definitions.
+ *
+ * Written by George Edmond Eddy (Rusty), ISI, February 1998.
+ * Modified by Pavlin Radoslavov, USC/ISI, May 1998, Aug 1999, October 2000.
+ * Modified by Hitoshi Asaeda, WIDE, August 1998.
+ */
+
+/*
+ * PIM statistics kept in the kernel
+ */
+struct pimstat {
+	u_quad_t pims_rcv_total_msgs;	   /* total PIM messages received    */
+	u_quad_t pims_rcv_total_bytes;	   /* total PIM bytes received	     */
+	u_quad_t pims_rcv_tooshort;	   /* rcvd with too few bytes	     */
+	u_quad_t pims_rcv_badsum;	   /* rcvd with bad checksum	     */
+	u_quad_t pims_rcv_badversion;	   /* rcvd bad PIM version	     */
+	u_quad_t pims_rcv_registers_msgs;  /* rcvd regs. msgs (data only)    */
+	u_quad_t pims_rcv_registers_bytes; /* rcvd regs. bytes (data only)   */
+	u_quad_t pims_rcv_registers_wrongiif; /* rcvd regs. on wrong iif     */
+	u_quad_t pims_rcv_badregisters;	   /* rcvd invalid registers	     */
+	u_quad_t pims_snd_registers_msgs;  /* sent regs. msgs (data only)    */
+	u_quad_t pims_snd_registers_bytes; /* sent regs. bytes (data only)   */
+};
+
+/*
+ * Names for PIM sysctl objects
+ */
+#define PIMCTL_STATS		1	/* statistics (read-only) */
+#define PIMCTL_MAXID		2
+
+#define PIMCTL_NAMES {			\
+	{ 0, 0 },			\
+	{ "stats", CTLTYPE_STRUCT },	\
+}
+
+#ifdef _KERNEL
+extern struct pimstat pimstat;
+
+void	pim_input(struct mbuf *, ...);
+#endif
+
+#endif /* _NETINET_PIM_VAR_H_ */
diff --git a/sys/netinet/raw_ip.c b/sys/netinet/raw_ip.c
index ef809eab541..1aa90898f3b 100644
--- a/sys/netinet/raw_ip.c
+++ b/sys/netinet/raw_ip.c
@@ -1,4 +1,4 @@
-/*	$OpenBSD: raw_ip.c,v 1.35 2004/12/08 16:37:11 henning Exp $	*/
+/*	$OpenBSD: raw_ip.c,v 1.36 2005/01/14 14:51:28 mcbride Exp $	*/
 /*	$NetBSD: raw_ip.c,v 1.25 1996/02/18 18:58:33 christos Exp $	*/
 
 /*
@@ -294,6 +294,10 @@ rip_ctloutput(op, so, level, optname, m)
 	case MRT_DEL_MFC:
 	case MRT_VERSION:
 	case MRT_ASSERT:
+	case MRT_API_SUPPORT:
+	case MRT_API_CONFIG:
+	case MRT_ADD_BW_UPCALL:
+	case MRT_DEL_BW_UPCALL:
 #ifdef MROUTING
 		switch (op) {
 		case PRCO_SETOPT: