diff --git a/sys/netinet/in.h b/sys/netinet/in.h
index a4e9f0141126..63bd1cedfe17 100644
--- a/sys/netinet/in.h
+++ b/sys/netinet/in.h
@@ -486,6 +486,22 @@ struct ip_mreq {
 	{ 0, 0 }, \
 	{ 0, 0 }, \
 	{ "ipsec", CTLTYPE_NODE }, \
+	{ 0, 0 }, \
+	{ 0, 0 }, \
+	{ 0, 0 }, \
+	{ 0, 0 }, { 0, 0 }, { 0, 0 }, { 0, 0 }, { 0, 0 }, \
+	{ 0, 0 }, { 0, 0 }, { 0, 0 }, { 0, 0 }, { 0, 0 }, \
+	{ 0, 0 }, { 0, 0 }, { 0, 0 }, { 0, 0 }, { 0, 0 }, \
+	{ 0, 0 }, { 0, 0 }, { 0, 0 }, { 0, 0 }, { 0, 0 }, \
+	{ 0, 0 }, { 0, 0 }, { 0, 0 }, { 0, 0 }, { 0, 0 }, \
+	{ 0, 0 }, { 0, 0 }, { 0, 0 }, { 0, 0 }, { 0, 0 }, \
+	{ 0, 0 }, { 0, 0 }, { 0, 0 }, { 0, 0 }, { 0, 0 }, \
+	{ 0, 0 }, { 0, 0 }, { 0, 0 }, { 0, 0 }, { 0, 0 }, \
+	{ 0, 0 }, { 0, 0 }, { 0, 0 }, { 0, 0 }, { 0, 0 }, \
+	{ 0, 0 }, \
+	{ 0, 0 }, \
+	{ 0, 0 }, \
+	{ "pim", CTLTYPE_NODE }, \
 }
 
 /*
diff --git a/sys/netinet/in_proto.c b/sys/netinet/in_proto.c
index 47c8b2463486..8199df80115d 100644
--- a/sys/netinet/in_proto.c
+++ b/sys/netinet/in_proto.c
@@ -36,6 +36,7 @@
 
 #include "opt_ipdivert.h"
 #include "opt_ipx.h"
+#include "opt_mrouting.h"
 #include "opt_ipsec.h"
 #include "opt_inet6.h"
 
@@ -57,6 +58,9 @@
 #include <netinet/ip_var.h>
 #include <netinet/ip_icmp.h>
 #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>
@@ -216,6 +220,14 @@ struct protosw inetsw[] = {
   &rip_usrreqs
 },
 #endif
+#ifdef PIM
+{ SOCK_RAW,	&inetdomain,	IPPROTO_PIM,	PR_ATOMIC|PR_ADDR|PR_LASTHDR,
+  pim_input,	0,		0,		rip_ctloutput,
+  0,
+  0,		0,		0,		0,
+  &rip_usrreqs
+},
+#endif	/* PIM */
 	/* raw wildcard */
 { SOCK_RAW,	&inetdomain,	0,		PR_ATOMIC|PR_ADDR,
   rip_input,	0,		0,		rip_ctloutput,
@@ -260,4 +272,6 @@ SYSCTL_NODE(_net_inet, IPPROTO_RAW,	raw,	CTLFLAG_RW, 0,	"RAW");
 #ifdef IPDIVERT
 SYSCTL_NODE(_net_inet, IPPROTO_DIVERT,	divert,	CTLFLAG_RW, 0,	"DIVERT");
 #endif
-
+#ifdef PIM
+SYSCTL_NODE(_net_inet, IPPROTO_PIM,    pim,    CTLFLAG_RW, 0,  "PIM");
+#endif
diff --git a/sys/netinet/ip_mroute.c b/sys/netinet/ip_mroute.c
index f0f2471370c7..50ffdcabf93e 100644
--- a/sys/netinet/ip_mroute.c
+++ b/sys/netinet/ip_mroute.c
@@ -7,8 +7,16 @@
  * Modified by Van Jacobson, LBL, January 1993
  * Modified by Ajit Thyagarajan, PARC, August 1993
  * Modified by Bill Fenner, PARC, April 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: 3.5
+ * and PIM-SMv2 and PIM-DM support, advanced API support,
+ * bandwidth metering and signaling
+ *
  * $FreeBSD$
  */
 
@@ -16,6 +24,10 @@
 #include "opt_mrouting.h"
 #include "opt_random_ip_id.h"
 
+#ifdef PIM
+#define _PIM_VT 1
+#endif
+
 #include <sys/param.h>
 #include <sys/kernel.h>
 #include <sys/lock.h>
@@ -43,6 +55,10 @@
 #include <netinet/ip_encap.h>
 #include <netinet/ip_mroute.h>
 #include <netinet/ip_var.h>
+#ifdef PIM
+#include <netinet/pim.h>
+#include <netinet/pim_var.h>
+#endif
 #include <netinet/udp.h>
 #include <machine/in_cksum.h>
 
@@ -57,6 +73,9 @@ static u_int	mrtdebug;		/* any set of the flags below	*/
 #define		DEBUG_FORWARD	0x04
 #define		DEBUG_EXPIRE	0x08
 #define		DEBUG_XMIT	0x10
+#define		DEBUG_PIM	0x20
+
+#define		VIFI_INVALID	((vifi_t) -1)
 
 #define M_HASCL(m)	((m)->m_flags & M_EXT)
 
@@ -120,6 +139,80 @@ static struct ip multicast_encap_iphdr = {
 	0,				/* checksum */
 };
 
+/*
+ * Bandwidth meter variables and constants
+ */
+static MALLOC_DEFINE(M_BWMETER, "bwmeter", "multicast upcall bw meters");
+/*
+ * 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];
+static struct callout_handle 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 */
+static struct callout_handle bw_upcalls_ch;
+#define BW_UPCALLS_PERIOD (hz)		/* periodical flush of bw upcalls */
+
+#ifdef PIM
+static struct pimstat pimstat;
+SYSCTL_STRUCT(_net_inet_pim, PIMCTL_STATS, stats, CTLFLAG_RD,
+    &pimstat, pimstat,
+    "PIM Statistics (struct pimstat, netinet/pim_var.h)");
+
+/*
+ * 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.
  */
@@ -147,8 +240,9 @@ static int get_vif_cnt(struct sioc_vif_req *);
 static int ip_mrouter_init(struct socket *, int);
 static int add_vif(struct vifctl *);
 static int del_vif(vifi_t);
-static int add_mfc(struct mfcctl *);
-static int del_mfc(struct mfcctl *);
+static int add_mfc(struct mfcctl2 *);
+static int del_mfc(struct mfcctl2 *);
+static int set_api_config(uint32_t *); /* chose API capabilities */
 static int socket_send(struct socket *, struct mbuf *, struct sockaddr_in *);
 static int set_assert(int);
 static void expire_upcalls(void *);
@@ -164,6 +258,32 @@ static void tbf_send_packet(struct vif *, struct mbuf *);
 static void tbf_update_tokens(struct vif *);
 static int priority(struct vif *, struct ip *);
 
+/*
+ * Bandwidth monitoring
+ */
+static void free_bw_list(struct bw_meter *list);
+static int add_bw_upcall(struct bw_upcall *);
+static int del_bw_upcall(struct bw_upcall *);
+static void bw_meter_receive_packet(struct bw_meter *x, int plen,
+		struct timeval *nowp);
+static void bw_meter_prepare_upcall(struct bw_meter *x, struct timeval *nowp);
+static void bw_upcalls_send(void);
+static void schedule_bw_meter(struct bw_meter *x, struct timeval *nowp);
+static void unschedule_bw_meter(struct bw_meter *x);
+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
+
 /*
  * whether or not special PIM assert processing is enabled.
  */
@@ -173,6 +293,17 @@ 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 uint32_t mrt_api_support = (MRT_MFC_FLAGS_DISABLE_WRONGVIF |
+					 MRT_MFC_FLAGS_BORDER_VIF |
+					 MRT_MFC_RP |
+					 MRT_MFC_BW_UPCALL);
+static uint32_t mrt_api_config = 0;
+
 /*
  * Hash function for a source, group entry
  */
@@ -230,7 +361,9 @@ X_ip_mrouter_set(struct socket *so, struct sockopt *sopt)
     int	error, optval;
     vifi_t	vifi;
     struct	vifctl vifc;
-    struct	mfcctl mfc;
+    struct	mfcctl2 mfc;
+    struct	bw_upcall bw_upcall;
+    uint32_t	i;
 
     if (so != ip_mrouter && sopt->sopt_name != MRT_INIT)
 	return EPERM;
@@ -264,7 +397,20 @@ X_ip_mrouter_set(struct socket *so, struct sockopt *sopt)
 
     case MRT_ADD_MFC:
     case MRT_DEL_MFC:
-	error = sooptcopyin(sopt, &mfc, sizeof mfc, sizeof mfc);
+	bzero((caddr_t)&mfc, sizeof(mfc)); /* zero out all fields */
+	/*
+	 * select data size depending on API version.
+	 */
+	if (sopt->sopt_name == MRT_ADD_MFC &&
+		mrt_api_config & MRT_API_FLAGS_ALL) {
+	    error = sooptcopyin(sopt, &mfc, sizeof(struct mfcctl2),
+				sizeof(struct mfcctl2));
+	} else {
+	    error = sooptcopyin(sopt, &mfc, sizeof(struct mfcctl),
+				sizeof(struct mfcctl));
+	    bzero((caddr_t)&mfc + sizeof(struct mfcctl),
+			sizeof(mfc) - sizeof(struct mfcctl));
+	}
 	if (error)
 	    break;
 	if (sopt->sopt_name == MRT_ADD_MFC)
@@ -280,6 +426,26 @@ X_ip_mrouter_set(struct socket *so, struct sockopt *sopt)
 	set_assert(optval);
 	break;
 
+    case MRT_API_CONFIG:
+	error = sooptcopyin(sopt, &i, sizeof i, sizeof i);
+	if (!error)
+	    error = set_api_config(&i);
+	if (!error)
+	    error = sooptcopyout(sopt, &i, sizeof i);
+	break;
+
+    case MRT_ADD_BW_UPCALL:
+    case MRT_DEL_BW_UPCALL:
+	error = sooptcopyin(sopt, &bw_upcall, sizeof bw_upcall,
+				sizeof bw_upcall);
+	if (error)
+	    break;
+	if (sopt->sopt_name == MRT_ADD_BW_UPCALL)
+	    error = add_bw_upcall(&bw_upcall);
+	else
+	    error = del_bw_upcall(&bw_upcall);
+	break;
+
     default:
 	error = EOPNOTSUPP;
 	break;
@@ -305,6 +471,14 @@ X_ip_mrouter_get(struct socket *so, struct sockopt *sopt)
 	error = sooptcopyout(sopt, &pim_assert, sizeof pim_assert);
 	break;
 
+    case MRT_API_SUPPORT:
+	error = sooptcopyout(sopt, &mrt_api_support, sizeof mrt_api_support);
+	break;
+
+    case MRT_API_CONFIG:
+	error = sooptcopyout(sopt, &mrt_api_config, sizeof mrt_api_config);
+	break;
+
     default:
 	error = EOPNOTSUPP;
 	break;
@@ -405,6 +579,13 @@ ip_mrouter_init(struct socket *so, int version)
 
     expire_upcalls_ch = timeout(expire_upcalls, NULL, EXPIRE_TIMEOUT);
 
+    bw_upcalls_n = 0;
+    bzero((caddr_t)bw_meter_timers, sizeof(bw_meter_timers));
+    bw_upcalls_ch = timeout(expire_bw_upcalls_send, NULL, BW_UPCALLS_PERIOD);
+    bw_meter_ch = timeout(expire_bw_meter_process, NULL, BW_METER_PERIOD);
+
+    mrt_api_config = 0;
+
     if (mrtdebug)
 	log(LOG_DEBUG, "ip_mrouter_init\n");
 
@@ -433,7 +614,7 @@ X_ip_mrouter_done(void)
      */
     for (vifi = 0; vifi < numvifs; vifi++) {
 	if (viftable[vifi].v_lcl_addr.s_addr != 0 &&
-		!(viftable[vifi].v_flags & VIFF_TUNNEL)) {
+		!(viftable[vifi].v_flags & (VIFF_TUNNEL | VIFF_REGISTER))) {
 	    struct sockaddr_in *so = (struct sockaddr_in *)&(ifr.ifr_addr);
 
 	    so->sin_len = sizeof(struct sockaddr_in);
@@ -450,6 +631,11 @@ X_ip_mrouter_done(void)
 
     untimeout(expire_upcalls, NULL, expire_upcalls_ch);
 
+    mrt_api_config = 0;
+    bw_upcalls_n = 0;
+    untimeout(expire_bw_upcalls_send, NULL, bw_upcalls_ch);
+    untimeout(expire_bw_meter_process, NULL, bw_meter_ch);
+
     /*
      * Free all multicast forwarding cache entries.
      */
@@ -464,6 +650,7 @@ X_ip_mrouter_done(void)
 		free(rte, M_MRTABLE);
 		rte = n;
 	    }
+	    free_bw_list(rt->mfc_bw_meter);
 	    free(rt, M_MRTABLE);
 	    rt = nr;
 	}
@@ -471,11 +658,17 @@ X_ip_mrouter_done(void)
 
     bzero((caddr_t)mfctable, sizeof(mfctable));
 
+    bzero(bw_meter_timers, sizeof(bw_meter_timers));
+
     /*
      * Reset de-encapsulation cache
      */
     last_encap_src = INADDR_ANY;
     last_encap_vif = NULL;
+#ifdef PIM
+    reg_vif_num = VIFI_INVALID;
+#endif
+
     if (encap_cookie) {
 	encap_detach(encap_cookie);
 	encap_cookie = NULL;
@@ -505,6 +698,42 @@ set_assert(int i)
     return 0;
 }
 
+/*
+ * Configure API capabilities
+ */
+int
+set_api_config(uint32_t *apival)
+{
+    int i;
+
+    /*
+     * 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 (mfctable[i] != NULL) {
+	    *apival = 0;
+	    return EPERM;
+	}
+    }
+
+    mrt_api_config = *apival & mrt_api_support;
+    *apival = mrt_api_config;
+
+    return 0;
+}
+
 /*
  * Decide if a packet is from a tunnelled peer.
  * Return 0 if not, 64 if so.  XXX yuck.. 64 ???
@@ -607,11 +836,23 @@ add_vif(struct vifctl *vifcp)
 	return EADDRNOTAVAIL;
 
     /* Find the interface with an address in AF_INET family */
-    sin.sin_addr = vifcp->vifc_lcl_addr;
-    ifa = ifa_ifwithaddr((struct sockaddr *)&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((struct sockaddr *)&sin);
+	if (ifa == NULL)
+	    return EADDRNOTAVAIL;
+	ifp = ifa->ifa_ifp;
+    }
 
     if (vifcp->vifc_flags & VIFF_TUNNEL) {
 	if ((vifcp->vifc_flags & VIFF_SRCRT) == 0) {
@@ -646,6 +887,20 @@ add_vif(struct vifctl *vifcp)
 	    log(LOG_ERR, "source routed tunnels not supported\n");
 	    return EOPNOTSUPP;
 	}
+#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 *)&multicast_register_if);
+	if (reg_vif_num == VIFI_INVALID) {
+	    multicast_register_if.if_name = "register_vif";
+	    multicast_register_if.if_unit = 0;
+	    multicast_register_if.if_flags = IFF_LOOPBACK;
+	    bzero(&vifp->v_route, sizeof(vifp->v_route));
+	    reg_vif_num = vifcp->vifc_vifi;
+	}
+#endif
     } else {		/* Make sure the interface supports multicast */
 	if ((ifp->if_flags & IFF_MULTICAST) == 0)
 	    return EOPNOTSUPP;
@@ -715,7 +970,7 @@ del_vif(vifi_t vifi)
 
     s = splnet();
 
-    if (!(vifp->v_flags & VIFF_TUNNEL))
+    if (!(vifp->v_flags & (VIFF_TUNNEL | VIFF_REGISTER)))
 	if_allmulti(vifp->v_ifp, 0);
 
     if (vifp == last_encap_vif) {
@@ -733,6 +988,11 @@ del_vif(vifi_t vifi)
 	m_freem(m);
     }
 
+#ifdef PIM
+    if (vifp->v_flags & VIFF_REGISTER)
+	reg_vif_num = VIFI_INVALID;
+#endif
+
     bzero((caddr_t)vifp->v_tbf, sizeof(*(vifp->v_tbf)));
     bzero((caddr_t)vifp, sizeof (*vifp));
 
@@ -754,20 +1014,28 @@ del_vif(vifi_t vifi)
  * 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++)
+    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.s_addr = INADDR_ANY;
 }
 
 /*
  * 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;
@@ -786,7 +1054,7 @@ init_mfc_params(struct mfc *rt, struct mfcctl *mfccp)
  * Add an mfc entry
  */
 static int
-add_mfc(struct mfcctl *mfccp)
+add_mfc(struct mfcctl2 *mfccp)
 {
     struct mfc *rt;
     u_long hash;
@@ -883,6 +1151,7 @@ add_mfc(struct mfcctl *mfccp)
 	    rt->mfc_expire     = 0;
 	    rt->mfc_stall      = NULL;
 	    
+	    rt->mfc_bw_meter = NULL;
 	    /* insert new entry at head of hash chain */
 	    rt->mfc_next = mfctable[hash];
 	    mfctable[hash] = rt;
@@ -896,7 +1165,7 @@ add_mfc(struct mfcctl *mfccp)
  * Delete an mfc entry
  */
 static int
-del_mfc(struct mfcctl *mfccp)
+del_mfc(struct mfcctl2 *mfccp)
 {
     struct in_addr 	origin;
     struct in_addr 	mcastgrp;
@@ -904,6 +1173,7 @@ del_mfc(struct mfcctl *mfccp)
     struct mfc	 	**nptr;
     u_long 		hash;
     int s;
+    struct bw_meter	*list;
 
     origin = mfccp->mfcc_origin;
     mcastgrp = mfccp->mfcc_mcastgrp;
@@ -926,10 +1196,19 @@ del_mfc(struct mfcctl *mfccp)
     }
 
     *nptr = rt->mfc_next;
+
+    /*
+     * free the bw_meter entries
+     */
+    list = rt->mfc_bw_meter;
+    rt->mfc_bw_meter = NULL;
+
     free(rt, M_MRTABLE);
 
     splx(s);
 
+    free_bw_list(list);
+
     return 0;
 }
 
@@ -963,8 +1242,8 @@ socket_send(struct socket *s, struct mbuf *mm, struct sockaddr_in *src)
 #define TUNNEL_LEN  12  /* # bytes of IP option for tunnel encapsulation  */
 
 static int
-X_ip_mforward(struct ip *ip, struct ifnet *ifp,
-	struct mbuf *m, struct ip_moptions *imo)
+X_ip_mforward(struct ip *ip, struct ifnet *ifp, struct mbuf *m,
+    struct ip_moptions *imo)
 {
     struct mfc *rt;
     int s;
@@ -979,7 +1258,7 @@ X_ip_mforward(struct ip *ip, struct ifnet *ifp,
 		((u_char *)(ip + 1))[1] != IPOPT_LSRR ) {
 	/*
 	 * Packet arrived via a physical interface or
-	 * an encapsulated tunnel.
+	 * an encapsulated tunnel or a register_vif.
 	 */
     } else {
 	/*
@@ -996,7 +1275,7 @@ X_ip_mforward(struct ip *ip, struct ifnet *ifp,
 	return 1;
     }
 
-    if ((imo) && ((vifi = imo->imo_multicast_vif) < numvifs)) {
+    if (imo && ((vifi = imo->imo_multicast_vif) < numvifs)) {
 	if (ip->ip_ttl < 255)
 	    ip->ip_ttl++;	/* compensate for -1 in *_send routines */
 	if (rsvpdebug && ip->ip_p == IPPROTO_RSVP) {
@@ -1064,7 +1343,7 @@ X_ip_mforward(struct ip *ip, struct ifnet *ifp,
 	    splx(s);
 	    return ENOBUFS;
 	}
-	mb0 = m_copy(m, 0, M_COPYALL);
+	mb0 = m_copypacket(m, M_DONTWAIT);
 	if (mb0 && (M_HASCL(mb0) || mb0->m_len < hlen))
 	    mb0 = m_pullup(mb0, hlen);
 	if (mb0 == NULL) {
@@ -1136,10 +1415,16 @@ fail:
 	    rt->mfc_mcastgrp.s_addr   = ip->ip_dst.s_addr;
 	    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;
 
+	    rt->mfc_rp.s_addr = INADDR_ANY; /* clear the RP address */
+
+	    rt->mfc_bw_meter = NULL;
+
 	    /* link into table */
 	    rt->mfc_next   = mfctable[hash];
 	    mfctable[hash] = rt;
@@ -1223,6 +1508,16 @@ expire_upcalls(void *unused)
 		++mrtstat.mrts_cache_cleanups;
 		nexpire[i]--;
 
+		/*
+		 * free the bw_meter entries
+		 */
+		while (mfc->mfc_bw_meter != NULL) {
+		    struct bw_meter *x = mfc->mfc_bw_meter;
+
+		    mfc->mfc_bw_meter = x->bm_mfc_next;
+		    free(x, M_BWMETER);
+		}
+
 		*nptr = mfc->mfc_next;
 		free(mfc, M_MRTABLE);
 	    } else {
@@ -1262,6 +1557,11 @@ ip_mdq(struct mbuf *m, struct ifnet *ifp, struct mfc *rt, vifi_t xmt_vif)
      * (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;
     }
@@ -1278,21 +1578,30 @@ ip_mdq(struct mbuf *m, struct ifnet *ifp, struct mfc *rt, vifi_t xmt_vif)
 	++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_long 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; vifi++)
 		;
 	    if (vifi >= numvifs)
-		return 0;	/* if not found: ignore the packet */
+		return 0;	/* The iif is not found: ignore the packet. */
+
+	    if (rt->mfc_flags[vifi] & MRT_MFC_FLAGS_DISABLE_WRONGVIF)
+		return 0;	/* WRONGVIF disabled: ignore the packet */
 
 	    GET_TIME(now);
 
@@ -1351,9 +1660,26 @@ ip_mdq(struct mbuf *m, struct ifnet *ifp, struct mfc *rt, vifi_t xmt_vif)
 	if ((rt->mfc_ttls[vifi] > 0) && (ip->ip_ttl > rt->mfc_ttls[vifi])) {
 	    viftable[vifi].v_pkt_out++;
 	    viftable[vifi].v_bytes_out += plen;
+#ifdef PIM
+	    if (viftable[vifi].v_flags & VIFF_REGISTER)
+		pim_register_send(ip, viftable + vifi, m, rt);
+	    else
+#endif
 	    MC_SEND(ip, viftable+vifi, m);
 	}
 
+    /*
+     * Perform upcall-related bw measuring.
+     */
+    if (rt->mfc_bw_meter != NULL) {
+	struct bw_meter *x;
+	struct timeval now;
+
+	GET_TIME(now);
+	for (x = rt->mfc_bw_meter; x != NULL; x = x->bm_mfc_next)
+	    bw_meter_receive_packet(x, plen, &now);
+    }
+
     return 0;
 }
 
@@ -1389,7 +1715,7 @@ phyint_send(struct ip *ip, struct vif *vifp, struct mbuf *m)
      * the IP header is actually copied, not just referenced,
      * so that ip_output() only scribbles on the copy.
      */
-    mb_copy = m_copy(m, 0, M_COPYALL);
+    mb_copy = m_copypacket(m, M_DONTWAIT);
     if (mb_copy && (M_HASCL(mb_copy) || mb_copy->m_len < hlen))
 	mb_copy = m_pullup(mb_copy, hlen);
     if (mb_copy == NULL)
@@ -1408,16 +1734,6 @@ encap_send(struct ip *ip, struct vif *vifp, struct mbuf *m)
     struct ip *ip_copy;
     int i, len = ip->ip_len;
 
-    /*
-     * XXX: take care of delayed checksums.
-     * XXX: if network interfaces are capable of computing checksum for
-     * encapsulated multicast data packets, we need to reconsider this.
-     */
-    if (m->m_pkthdr.csum_flags & CSUM_DELAY_DATA) {
-	in_delayed_cksum(m);
-	m->m_pkthdr.csum_flags &= ~CSUM_DELAY_DATA;
-    }
-
     /*
      * copy the old packet & pullup its IP header into the
      * new mbuf so we can modify it.  Try to fill the new
@@ -1432,7 +1748,7 @@ encap_send(struct ip *ip, struct vif *vifp, struct mbuf *m)
     mb_copy->m_data += max_linkhdr;
     mb_copy->m_len = sizeof(multicast_encap_iphdr);
 
-    if ((mb_copy->m_next = m_copy(m, 0, M_COPYALL)) == NULL) {
+    if ((mb_copy->m_next = m_copypacket(m, M_DONTWAIT)) == NULL) {
 	m_freem(mb_copy);
 	return;
     }
@@ -1888,6 +2204,1023 @@ X_rsvp_input(struct mbuf *m, int off)
     splx(s);
 }
 
+/*
+ * Code for bandwidth monitors
+ */
+
+/*
+ * Define common interface for timeval-related methods
+ */
+#define	BW_TIMEVALCMP(tvp, uvp, cmp) timevalcmp((tvp), (uvp), cmp)
+#define	BW_TIMEVALDECR(vvp, uvp) timevalsub((vvp), (uvp))
+#define	BW_TIMEVALADD(vvp, uvp) timevaladd((vvp), (uvp))
+
+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 bw_upcall *req)
+{
+    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;
+    int s;
+    
+    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 = splnet();
+    mfc = mfc_find(req->bu_src.s_addr, req->bu_dst.s_addr);
+    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 */
+	}
+    }
+    splx(s);
+    
+    /* Allocate the new bw_meter entry */
+    x = (struct bw_meter *)malloc(sizeof(*x), M_BWMETER, M_NOWAIT);
+    if (x == NULL)
+	return ENOBUFS;
+    
+    /* Set the new bw_meter entry */
+    x->bm_threshold.b_time = req->bu_threshold.b_time;
+    GET_TIME(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 */
+    s = splnet();
+    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 bw_upcall *req)
+{
+    struct mfc *mfc;
+    struct bw_meter *x;
+    int s;
+    
+    if (!(mrt_api_config & MRT_MFC_BW_UPCALL))
+	return EOPNOTSUPP;
+    
+    s = splnet();
+    /* Find the corresponding MFC entry */
+    mfc = mfc_find(req->bu_src.s_addr, req->bu_dst.s_addr);
+    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;
+	splx(s);
+	free_bw_list(list);
+	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;
+	     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 */
+	    splx(s);
+
+	    unschedule_bw_meter(x);
+	    /* 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;
+    int s;
+    
+    s = splnet();
+    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;
+	}
+    }
+    splx(s);
+}
+
+/*
+ * 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;
+    int s;
+    
+    s = splnet();
+    
+    /*
+     * 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;
+    
+    splx(s);
+}
+
+/*
+ * 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 (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, s;
+    
+    if (!(x->bm_flags & BW_METER_LEQ))
+	return;		/* XXX: we schedule timers only for "<=" entries */
+    
+    /*
+     * Reset the bw_meter entry
+     */
+    s = splnet();
+    x->bm_start_time = *nowp;
+    x->bm_measured.b_packets = 0;
+    x->bm_measured.b_bytes = 0;
+    x->bm_flags &= ~BW_METER_UPCALL_DELIVERED;
+    splx(s);
+    
+    /*
+     * 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()
+{
+    static uint32_t last_tv_sec;	/* last time we processed this */
+
+    uint32_t loops;
+    int i, s;
+    struct timeval now, process_endtime;
+    
+    GET_TIME(now);
+    if (last_tv_sec == now.tv_sec)
+	return;		/* nothing to do */
+
+    s = splnet();
+    loops = now.tv_sec - last_tv_sec;
+    last_tv_sec = now.tv_sec;
+    if (loops > BW_METER_BUCKETS)
+	loops = BW_METER_BUCKETS;
+
+    /*
+     * 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 + 1) % BW_METER_BUCKETS; loops > 0; loops--) {
+	struct bw_meter *x, *tmp_list;
+	
+	if (++i >= BW_METER_BUCKETS)
+	    i = 0;
+	
+	tmp_list = bw_meter_timers[i];
+	bw_meter_timers[i] = NULL;
+	
+	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);
+		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);
+	}
+    }
+    splx(s);
+    
+    /* Send all upcalls that are pending delivery */
+    bw_upcalls_send();
+}
+
+/*
+ * A periodic function for sending all upcalls that are pending delivery
+ */
+static void
+expire_bw_upcalls_send(void *unused)
+{
+    bw_upcalls_send();
+    
+    bw_upcalls_ch = timeout(expire_bw_upcalls_send, NULL, 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();
+    
+    bw_meter_ch = timeout(expire_bw_meter_process, NULL, 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 = 0;
+	mm = m_pullup(mm, sizeof(struct ip));
+	if (mm != NULL) {
+	    ip = mtod(mm, struct ip *);
+	    if ((mrt_api_config & MRT_MFC_RP) &&
+		(rt->mfc_rp.s_addr != INADDR_ANY)) {
+		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;
+    
+    /*
+     * XXX: take care of delayed checksums.
+     * XXX: if network interfaces are capable of computing checksum for
+     * encapsulated multicast data packets, we need to reconsider this.
+     */
+    if (m->m_pkthdr.csum_flags & CSUM_DELAY_DATA) {
+	in_delayed_cksum(m);
+	m->m_pkthdr.csum_flags &= ~CSUM_DELAY_DATA;
+    }
+    
+    /*
+     * Copy the old packet & pullup its IP header into the
+     * new mbuf so we can modify it.
+     */
+    mb_copy = m_copypacket(m, M_DONTWAIT);
+    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 (ip_fragment(ip, &mb_copy, mtu, 0, 0) != 0) {
+	m_freem(mb_copy);
+	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) || (viftable[vifi].v_lcl_addr.s_addr == 0)) {
+	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;
+#ifdef RANDOM_IP_ID
+    ip_outer->ip_id = ip_randomid();
+#else
+    ip_outer->ip_id = htons(ip_id++);
+#endif
+    ip_outer->ip_len = 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, 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, int off)
+{
+    struct ip *ip = mtod(m, struct ip *);
+    struct pim *pim;
+    int minlen;
+    int datalen = ip->ip_len;
+    int ip_tos;
+    int iphlen = off;
+    
+    /* 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)) == 0) {
+	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.
+	 */
+	struct sockaddr_in dst = { sizeof(dst), AF_INET };
+	struct mbuf *mcp;
+	struct ip *encap_ip;
+	u_int32_t *reghdr;
+	
+	if ((reg_vif_num >= numvifs) || (reg_vif_num == VIFI_INVALID)) {
+	    if (mrtdebug & DEBUG_PIM)
+		log(LOG_DEBUG,
+		    "pim_input: register vif not set: %d\n", reg_vif_num);
+	    m_freem(m);
+	    return;
+	}
+	
+	/*
+	 * 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(ntohl(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;
+	}
+	
+	/*
+	 * 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);
+	}
+	
+	/* If a NULL_REGISTER, pass it to the daemon */
+	if ((ntohl(*reghdr) & PIM_NULL_REGISTER))
+	    goto pim_input_to_daemon;
+	
+	/*
+	 * 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);
+	}
+	if_simloop(viftable[reg_vif_num].v_ifp, m, dst.sin_family, 0);
+	
+	/* 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, iphlen);
+
+    return;
+}
+#endif /* PIM */
+
 static int
 ip_mroute_modevent(module_t mod, int type, void *unused)
 {
diff --git a/sys/netinet/ip_mroute.h b/sys/netinet/ip_mroute.h
index 1045ebbfccd4..58b1b6137023 100644
--- a/sys/netinet/ip_mroute.h
+++ b/sys/netinet/ip_mroute.h
@@ -48,8 +48,12 @@
  * 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: 3.3.1.3
+ * and PIM-SMv2 and PIM-DM support, advanced API support,
+ * bandwidth metering and signaling.
  */
 
 
@@ -63,7 +67,12 @@
 #define MRT_ADD_MFC	104	/* insert forwarding cache entry */
 #define MRT_DEL_MFC	105	/* delete forwarding cache entry */
 #define MRT_VERSION	106	/* get kernel version number */
-#define MRT_ASSERT      107     /* enable PIM assert processing */
+#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 */
 
 
 #define GET_TIME(t)	microtime(&t)
@@ -99,10 +108,11 @@ struct vifctl {
 
 #define	VIFF_TUNNEL	0x1		/* vif represents a tunnel end-point */
 #define VIFF_SRCRT	0x2		/* tunnel uses IP source routing */
+#define VIFF_REGISTER	0x4		/* used for PIM Register encap/decap */
 
 /*
  * Argument structure for MRT_ADD_MFC and MRT_DEL_MFC
- * (mfcc_tos to be added at a future point)
+ * XXX if you change this, make sure to change struct mfcctl2 as well.
  */
 struct mfcctl {
     struct in_addr  mfcc_origin;		/* ip origin of mcasts       */
@@ -111,6 +121,94 @@ struct mfcctl {
     u_char	    mfcc_ttls[MAXVIFS]; 	/* forwarding ttls on vifs   */
 };
 
+/*
+ * 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_char		mfcc_ttls[MAXVIFS]; 	/* forwarding ttls on vifs   */
+
+	/* extension fields */
+	uint8_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;
+	uint64_t	b_packets;
+	uint64_t	b_bytes;
+};
+
+struct bw_upcall {
+	struct in_addr	bu_src;			/* source address            */
+	struct in_addr	bu_dst;			/* destination address       */
+	uint32_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
+
 /*
  * The kernel's multicast routing statistics.
  */
@@ -179,17 +277,20 @@ struct vif {
  * at a future point)
  */
 struct mfc {
-    struct in_addr  mfc_origin;	 		/* IP origin of mcasts   */
-    struct in_addr  mfc_mcastgrp;  		/* multicast group associated*/
-    vifi_t	    mfc_parent; 		/* incoming vif              */
-    u_char	    mfc_ttls[MAXVIFS]; 		/* forwarding ttls on vifs   */
-    u_long	    mfc_pkt_cnt;		/* pkt count for src-grp     */
-    u_long	    mfc_byte_cnt;		/* byte count for src-grp    */
-    u_long	    mfc_wrong_if;		/* wrong if for src-grp	     */
-    int		    mfc_expire;			/* time to clean entry up    */
-    struct timeval  mfc_last_assert;		/* last time I sent an assert*/
-    struct rtdetq  *mfc_stall;			/* q of packets awaiting mfc */
-    struct mfc     *mfc_next;			/* next mfc entry            */
+	struct in_addr	mfc_origin;		/* IP origin of mcasts	     */
+	struct in_addr  mfc_mcastgrp;  		/* multicast group associated*/
+	vifi_t		mfc_parent; 		/* incoming vif              */
+	u_char		mfc_ttls[MAXVIFS]; 	/* forwarding ttls on vifs   */
+	u_long		mfc_pkt_cnt;		/* pkt count for src-grp     */
+	u_long		mfc_byte_cnt;		/* byte count for src-grp    */
+	u_long		mfc_wrong_if;		/* wrong if for src-grp	     */
+	int		mfc_expire;		/* time to clean entry up    */
+	struct timeval	mfc_last_assert;	/* last time I sent an assert*/
+	struct rtdetq	*mfc_stall;		/* q of packets awaiting mfc */
+	struct mfc	*mfc_next;		/* next mfc entry            */
+	uint8_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  */
 };
 
 /*
@@ -200,8 +301,10 @@ struct igmpmsg {
     u_long	    unused1;
     u_long	    unused2;
     u_char	    im_msgtype;			/* what type of message	    */
-#define IGMPMSG_NOCACHE		1
-#define IGMPMSG_WRONGVIF	2
+#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_char	    im_mbz;			/* must be zero		    */
     u_char	    im_vif;			/* vif rec'd on		    */
     u_char	    unused3;
@@ -246,6 +349,32 @@ struct tbf
     struct mbuf *tbf_t;		/* tail-insertion pointer	*/
 };
 
+/*
+ * 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		     */
+};
+
 #ifdef _KERNEL
 
 struct sockopt;
diff --git a/sys/netinet/ip_output.c b/sys/netinet/ip_output.c
index 1c6d8942b48a..60d5b1ddaf8d 100644
--- a/sys/netinet/ip_output.c
+++ b/sys/netinet/ip_output.c
@@ -124,15 +124,10 @@ extern	struct protosw inetsw[];
  * The mbuf opt, if present, will not be freed.
  */
 int
-ip_output(m0, opt, ro, flags, imo, inp)
-	struct mbuf *m0;
-	struct mbuf *opt;
-	struct route *ro;
-	int flags;
-	struct ip_moptions *imo;
-	struct inpcb *inp;
+ip_output(struct mbuf *m0, struct mbuf *opt, struct route *ro,
+	int flags, struct ip_moptions *imo, struct inpcb *inp)
 {
-	struct ip *ip, *mhip;
+	struct ip *ip;
 	struct ifnet *ifp = NULL;	/* keep compiler happy */
 	struct mbuf *m;
 	int hlen = sizeof (struct ip);
@@ -478,7 +473,7 @@ ip_output(m0, opt, ro, flags, imo, inp)
 			goto bad;
 		}
 		/* don't allow broadcast messages to be fragmented */
-		if ((u_short)ip->ip_len > ifp->if_mtu) {
+		if (ip->ip_len > ifp->if_mtu) {
 			error = EMSGSIZE;
 			goto bad;
 		}
@@ -1014,8 +1009,7 @@ pass:
 	 * If small enough for interface, or the interface will take
 	 * care of the fragmentation for us, can just send directly.
 	 */
-	if ((u_short)ip->ip_len <= ifp->if_mtu ||
-	    ifp->if_hwassist & CSUM_FRAGMENT) {
+	if (ip->ip_len <= ifp->if_mtu || ifp->if_hwassist & CSUM_FRAGMENT) {
 		ip->ip_len = htons(ip->ip_len);
 		ip->ip_off = htons(ip->ip_off);
 		ip->ip_sum = 0;
@@ -1057,10 +1051,7 @@ pass:
 				(struct sockaddr *)dst, ro->ro_rt);
 		goto done;
 	}
-	/*
-	 * Too large for interface; fragment if possible.
-	 * Must be able to put at least 8 bytes per fragment.
-	 */
+
 	if (ip->ip_off & IP_DF) {
 		error = EMSGSIZE;
 		/*
@@ -1070,149 +1061,23 @@ pass:
 		 * them, there is no way for one to update all its
 		 * routes when the MTU is changed.
 		 */
-		if ((ro->ro_rt->rt_flags & (RTF_UP | RTF_HOST))
-		    && !(ro->ro_rt->rt_rmx.rmx_locks & RTV_MTU)
-		    && (ro->ro_rt->rt_rmx.rmx_mtu > ifp->if_mtu)) {
+		if ((ro->ro_rt->rt_flags & (RTF_UP | RTF_HOST)) &&
+		    !(ro->ro_rt->rt_rmx.rmx_locks & RTV_MTU) &&
+		    (ro->ro_rt->rt_rmx.rmx_mtu > ifp->if_mtu)) {
 			ro->ro_rt->rt_rmx.rmx_mtu = ifp->if_mtu;
 		}
 		ipstat.ips_cantfrag++;
 		goto bad;
 	}
-	len = (ifp->if_mtu - hlen) &~ 7;
-	if (len < 8) {
-		error = EMSGSIZE;
+
+	/*
+	 * Too large for interface; fragment if possible. If successful,
+	 * on return, m will point to a list of packets to be sent.
+	 */
+	error = ip_fragment(ip, &m, ifp->if_mtu, ifp->if_hwassist, sw_csum);
+	if (error)
 		goto bad;
-	}
-
-	/*
-	 * if the interface will not calculate checksums on
-	 * fragmented packets, then do it here.
-	 */
-	if (m->m_pkthdr.csum_flags & CSUM_DELAY_DATA &&
-	    (ifp->if_hwassist & CSUM_IP_FRAGS) == 0) {
-		in_delayed_cksum(m);
-		m->m_pkthdr.csum_flags &= ~CSUM_DELAY_DATA;
-	}
-
-	if (len > PAGE_SIZE) {
-		/* 
-		 * Fragement large datagrams such that each segment 
-		 * contains a multiple of PAGE_SIZE amount of data, 
-		 * plus headers. This enables a receiver to perform 
-		 * page-flipping zero-copy optimizations.
-		 */
-
-		int newlen;
-		struct mbuf *mtmp;
-
-		for (mtmp = m, off = 0; 
-		     mtmp && ((off + mtmp->m_len) <= ifp->if_mtu);
-		     mtmp = mtmp->m_next) {
-			off += mtmp->m_len;
-		}
-		/*
-		 * firstlen (off - hlen) must be aligned on an 
-		 * 8-byte boundary
-		 */
-		if (off < hlen)
-			goto smart_frag_failure;
-		off = ((off - hlen) & ~7) + hlen;
-		newlen = (~PAGE_MASK) & ifp->if_mtu;
-		if ((newlen + sizeof (struct ip)) > ifp->if_mtu) {
-			/* we failed, go back the default */
-smart_frag_failure:
-			newlen = len;
-			off = hlen + len;
-		}
-
-/*		printf("ipfrag: len = %d, hlen = %d, mhlen = %d, newlen = %d, off = %d\n",
-		len, hlen, sizeof (struct ip), newlen, off);*/
-
-		len = newlen;
-
-	} else {
-		off = hlen + len;
-	}
-
-
-
-    {
-	int mhlen, firstlen = off - hlen;
-	struct mbuf **mnext = &m->m_nextpkt;
-	int nfrags = 1;
-
-	/*
-	 * Loop through length of segment after first fragment,
-	 * make new header and copy data of each part and link onto chain.
-	 */
-	m0 = m;
-	mhlen = sizeof (struct ip);
-	for (; off < (u_short)ip->ip_len; off += len) {
-		MGETHDR(m, M_DONTWAIT, MT_HEADER);
-		if (m == 0) {
-			error = ENOBUFS;
-			ipstat.ips_odropped++;
-			goto sendorfree;
-		}
-		m->m_flags |= (m0->m_flags & M_MCAST) | M_FRAG;
-		m->m_data += max_linkhdr;
-		mhip = mtod(m, struct ip *);
-		*mhip = *ip;
-		if (hlen > sizeof (struct ip)) {
-			mhlen = ip_optcopy(ip, mhip) + sizeof (struct ip);
-			mhip->ip_v = IPVERSION;
-			mhip->ip_hl = mhlen >> 2;
-		}
-		m->m_len = mhlen;
-		mhip->ip_off = ((off - hlen) >> 3) + ip->ip_off;
-		if (off + len >= (u_short)ip->ip_len)
-			len = (u_short)ip->ip_len - off;
-		else
-			mhip->ip_off |= IP_MF;
-		mhip->ip_len = htons((u_short)(len + mhlen));
-		m->m_next = m_copy(m0, off, len);
-		if (m->m_next == 0) {
-			(void) m_free(m);
-			error = ENOBUFS;	/* ??? */
-			ipstat.ips_odropped++;
-			goto sendorfree;
-		}
-		m->m_pkthdr.len = mhlen + len;
-		m->m_pkthdr.rcvif = (struct ifnet *)0;
-#ifdef MAC
-		mac_create_fragment(m0, m);
-#endif
-		m->m_pkthdr.csum_flags = m0->m_pkthdr.csum_flags;
-		mhip->ip_off = htons(mhip->ip_off);
-		mhip->ip_sum = 0;
-		if (sw_csum & CSUM_DELAY_IP)
-			mhip->ip_sum = in_cksum(m, mhlen);
-		*mnext = m;
-		mnext = &m->m_nextpkt;
-		nfrags++;
-	}
-	ipstat.ips_ofragments += nfrags;
-
-	/* set first/last markers for fragment chain */
-	m->m_flags |= M_LASTFRAG;
-	m0->m_flags |= M_FIRSTFRAG | M_FRAG;
-	m0->m_pkthdr.csum_data = nfrags;
-
-	/*
-	 * Update first fragment by trimming what's been copied out
-	 * and updating header, then send each fragment (in order).
-	 */
-	m = m0;
-	m_adj(m, hlen + firstlen - (u_short)ip->ip_len);
-	m->m_pkthdr.len = hlen + firstlen;
-	ip->ip_len = htons((u_short)m->m_pkthdr.len);
-	ip->ip_off |= IP_MF;
-	ip->ip_off = htons(ip->ip_off);
-	ip->ip_sum = 0;
-	if (sw_csum & CSUM_DELAY_IP)
-		ip->ip_sum = in_cksum(m, hlen);
-sendorfree:
-	for (m = m0; m; m = m0) {
+	for (; m; m = m0) {
 		m0 = m->m_nextpkt;
 		m->m_nextpkt = 0;
 #ifdef IPSEC
@@ -1234,7 +1099,7 @@ sendorfree:
 
 	if (error == 0)
 		ipstat.ips_fragmented++;
-    }
+
 done:
 #ifdef IPSEC
 	if (ro == &iproute && ro->ro_rt) {
@@ -1246,7 +1111,7 @@ done:
 			printf("DP ip_output call free SP:%p\n", sp));
 		key_freesp(sp);
 	}
-#endif /* IPSEC */
+#endif
 #ifdef FAST_IPSEC
 	if (ro == &iproute && ro->ro_rt) {
 		RTFREE(ro->ro_rt);
@@ -1254,13 +1119,181 @@ done:
 	}
 	if (sp != NULL)
 		KEY_FREESP(&sp);
-#endif /* FAST_IPSEC */
+#endif
 	return (error);
 bad:
 	m_freem(m);
 	goto done;
 }
 
+/*
+ * Create a chain of fragments which fit the given mtu. m_frag points to the
+ * mbuf to be fragmented; on return it points to the chain with the fragments.
+ * Return 0 if no error. If error, m_frag may contain a partially built
+ * chain of fragments that should be freed by the caller.
+ *
+ * if_hwassist_flags is the hw offload capabilities (see if_data.ifi_hwassist)
+ * sw_csum contains the delayed checksums flags (e.g., CSUM_DELAY_IP).
+ */
+int
+ip_fragment(struct ip *ip, struct mbuf **m_frag, int mtu,
+	    u_long if_hwassist_flags, int sw_csum)
+{
+	int error = 0;
+	int hlen = ip->ip_hl << 2;
+	int len = (mtu - hlen) & ~7;	/* size of payload in each fragment */
+	int off;
+	struct mbuf *m0 = *m_frag;	/* the original packet		*/
+	int firstlen;
+	struct mbuf **mnext;
+	int nfrags;
+
+	if (ip->ip_off & IP_DF) {	/* Fragmentation not allowed */
+		ipstat.ips_cantfrag++;
+		return EMSGSIZE;
+	}
+
+	/*
+	 * Must be able to put at least 8 bytes per fragment.
+	 */
+	if (len < 8)
+		return EMSGSIZE;
+
+	/*
+	 * If the interface will not calculate checksums on
+	 * fragmented packets, then do it here.
+	 */
+	if (m0->m_pkthdr.csum_flags & CSUM_DELAY_DATA &&
+	    (if_hwassist_flags & CSUM_IP_FRAGS) == 0) {
+		in_delayed_cksum(m0);
+		m0->m_pkthdr.csum_flags &= ~CSUM_DELAY_DATA;
+	}
+
+	if (len > PAGE_SIZE) {
+		/* 
+		 * Fragment large datagrams such that each segment 
+		 * contains a multiple of PAGE_SIZE amount of data, 
+		 * plus headers. This enables a receiver to perform 
+		 * page-flipping zero-copy optimizations.
+		 *
+		 * XXX When does this help given that sender and receiver
+		 * could have different page sizes, and also mtu could
+		 * be less than the receiver's page size ?
+		 */
+		int newlen;
+		struct mbuf *m;
+
+		for (m = m0, off = 0; m && (off+m->m_len) <= mtu; m = m->m_next)
+			off += m->m_len;
+
+		/*
+		 * firstlen (off - hlen) must be aligned on an 
+		 * 8-byte boundary
+		 */
+		if (off < hlen)
+			goto smart_frag_failure;
+		off = ((off - hlen) & ~7) + hlen;
+		newlen = (~PAGE_MASK) & mtu;
+		if ((newlen + sizeof (struct ip)) > mtu) {
+			/* we failed, go back the default */
+smart_frag_failure:
+			newlen = len;
+			off = hlen + len;
+		}
+		len = newlen;
+
+	} else {
+		off = hlen + len;
+	}
+
+	firstlen = off - hlen;
+	mnext = &m0->m_nextpkt;		/* pointer to next packet */
+
+	/*
+	 * Loop through length of segment after first fragment,
+	 * make new header and copy data of each part and link onto chain.
+	 * Here, m0 is the original packet, m is the fragment being created.
+	 * The fragments are linked off the m_nextpkt of the original
+	 * packet, which after processing serves as the first fragment.
+	 */
+	for (nfrags = 1; off < ip->ip_len; off += len, nfrags++) {
+		struct ip *mhip;	/* ip header on the fragment */
+		struct mbuf *m;
+		int mhlen = sizeof (struct ip);
+
+		MGETHDR(m, M_DONTWAIT, MT_HEADER);
+		if (m == 0) {
+			error = ENOBUFS;
+			ipstat.ips_odropped++;
+			goto done;
+		}
+		m->m_flags |= (m0->m_flags & M_MCAST) | M_FRAG;
+		/*
+		 * In the first mbuf, leave room for the link header, then
+		 * copy the original IP header including options. The payload
+		 * goes into an additional mbuf chain returned by m_copy().
+		 */
+		m->m_data += max_linkhdr;
+		mhip = mtod(m, struct ip *);
+		*mhip = *ip;
+		if (hlen > sizeof (struct ip)) {
+			mhlen = ip_optcopy(ip, mhip) + sizeof (struct ip);
+			mhip->ip_v = IPVERSION;
+			mhip->ip_hl = mhlen >> 2;
+		}
+		m->m_len = mhlen;
+		/* XXX do we need to add ip->ip_off below ? */
+		mhip->ip_off = ((off - hlen) >> 3) + ip->ip_off;
+		if (off + len >= ip->ip_len) {	/* last fragment */
+			len = ip->ip_len - off;
+			m->m_flags |= M_LASTFRAG;
+		} else
+			mhip->ip_off |= IP_MF;
+		mhip->ip_len = htons((u_short)(len + mhlen));
+		m->m_next = m_copy(m0, off, len);
+		if (m->m_next == 0) {		/* copy failed */
+			m_free(m);
+			error = ENOBUFS;	/* ??? */
+			ipstat.ips_odropped++;
+			goto done;
+		}
+		m->m_pkthdr.len = mhlen + len;
+		m->m_pkthdr.rcvif = (struct ifnet *)0;
+#ifdef MAC
+		mac_create_fragment(m0, m);
+#endif
+		m->m_pkthdr.csum_flags = m0->m_pkthdr.csum_flags;
+		mhip->ip_off = htons(mhip->ip_off);
+		mhip->ip_sum = 0;
+		if (sw_csum & CSUM_DELAY_IP)
+			mhip->ip_sum = in_cksum(m, mhlen);
+		*mnext = m;
+		mnext = &m->m_nextpkt;
+	}
+	ipstat.ips_ofragments += nfrags;
+
+	/* set first marker for fragment chain */
+	m0->m_flags |= M_FIRSTFRAG | M_FRAG;
+	m0->m_pkthdr.csum_data = nfrags;
+
+	/*
+	 * Update first fragment by trimming what's been copied out
+	 * and updating header.
+	 */
+	m_adj(m0, hlen + firstlen - ip->ip_len);
+	m0->m_pkthdr.len = hlen + firstlen;
+	ip->ip_len = htons((u_short)m0->m_pkthdr.len);
+	ip->ip_off |= IP_MF;
+	ip->ip_off = htons(ip->ip_off);
+	ip->ip_sum = 0;
+	if (sw_csum & CSUM_DELAY_IP)
+		ip->ip_sum = in_cksum(m0, hlen);
+
+done:
+	*m_frag = m0;
+	return error;
+}
+
 void
 in_delayed_cksum(struct mbuf *m)
 {
@@ -1307,7 +1340,7 @@ ip_insertoptions(m, opt, phlen)
 	unsigned optlen;
 
 	optlen = opt->m_len - sizeof(p->ipopt_dst);
-	if (optlen + (u_short)ip->ip_len > IP_MAXPACKET) {
+	if (optlen + ip->ip_len > IP_MAXPACKET) {
 		*phlen = 0;
 		return (m);		/* XXX should fail */
 	}
diff --git a/sys/netinet/ip_var.h b/sys/netinet/ip_var.h
index c8df2e673e4a..5a9a1ca1fe78 100644
--- a/sys/netinet/ip_var.h
+++ b/sys/netinet/ip_var.h
@@ -164,6 +164,8 @@ extern struct	pr_usrreqs rip_usrreqs;
 
 int	 ip_ctloutput(struct socket *, struct sockopt *sopt);
 void	 ip_drain(void);
+int	 ip_fragment(struct ip *ip, struct mbuf **m_frag, int mtu,
+	    u_long if_hwassist_flags, int sw_csum);
 void	 ip_freemoptions(struct ip_moptions *);
 void	 ip_init(void);
 extern int	 (*ip_mforward)(struct ip *, struct ifnet *, struct mbuf *,
diff --git a/sys/netinet/raw_ip.c b/sys/netinet/raw_ip.c
index dd40eb092d47..05447b4712e9 100644
--- a/sys/netinet/raw_ip.c
+++ b/sys/netinet/raw_ip.c
@@ -383,6 +383,10 @@ rip_ctloutput(so, sopt)
 		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:
 			error = ip_mrouter_get ? ip_mrouter_get(so, sopt) :
 				EOPNOTSUPP;
 			break;
@@ -448,6 +452,10 @@ rip_ctloutput(so, sopt)
 		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:
 			error = ip_mrouter_set ? ip_mrouter_set(so, sopt) :
 					EOPNOTSUPP;
 			break;