freebsd-dev/sys/netinet/ip_mroute.c
1994-10-21 02:38:29 +00:00

1922 lines
48 KiB
C

/*
* IP multicast forwarding procedures
*
* Written by David Waitzman, BBN Labs, August 1988.
* Modified by Steve Deering, Stanford, February 1989.
* Modified by Mark J. Steiglitz, Stanford, May, 1991
* Modified by Van Jacobson, LBL, January 1993
* Modified by Ajit Thyagarajan, PARC, August 1993
*
* MROUTING 1.8
*/
#include <sys/param.h>
#include <sys/systm.h>
#include <sys/mbuf.h>
#include <sys/socket.h>
#include <sys/socketvar.h>
#include <sys/protosw.h>
#include <sys/errno.h>
#include <sys/time.h>
#include <sys/ioctl.h>
#include <sys/syslog.h>
#include <net/if.h>
#include <net/route.h>
#include <net/raw_cb.h>
#include <netinet/in.h>
#include <netinet/in_systm.h>
#include <netinet/ip.h>
#include <netinet/ip_var.h>
#include <netinet/in_pcb.h>
#include <netinet/in_var.h>
#include <netinet/igmp.h>
#include <netinet/igmp_var.h>
#include <netinet/ip_mroute.h>
#ifndef NTOHL
#if BYTE_ORDER != BIG_ENDIAN
#define NTOHL(d) ((d) = ntohl((d)))
#define NTOHS(d) ((d) = ntohs((u_short)(d)))
#define HTONL(d) ((d) = htonl((d)))
#define HTONS(d) ((d) = htons((u_short)(d)))
#else
#define NTOHL(d)
#define NTOHS(d)
#define HTONL(d)
#define HTONS(d)
#endif
#endif
#ifndef MROUTING
/*
* Dummy routines and globals used when multicast routing is not compiled in.
*/
u_int ip_mrtproto = 0;
struct socket *ip_mrouter = NULL;
struct mrtstat mrtstat;
int
_ip_mrouter_cmd(cmd, so, m)
int cmd;
struct socket *so;
struct mbuf *m;
{
return(EOPNOTSUPP);
}
int (*ip_mrouter_cmd)(int, struct socket *, struct mbuf *) = _ip_mrouter_cmd;
int
_ip_mrouter_done()
{
return(0);
}
int (*ip_mrouter_done)(void) = _ip_mrouter_done;
int
_ip_mforward(ip, ifp, m, imo)
struct ip *ip;
struct ifnet *ifp;
struct mbuf *m;
struct ip_moptions *imo;
{
return(0);
}
int (*ip_mforward)(struct ip *, struct ifnet *, struct mbuf *,
struct ip_moptions *) = _ip_mforward;
int
_mrt_ioctl(int req, caddr_t data, struct proc *p)
{
return EOPNOTSUPP;
}
int (*mrt_ioctl)(int, caddr_t, struct proc *) = _mrt_ioctl;
void multiencap_decap(struct mbuf *m) { /* XXX must fixup manually */
rip_input(m);
}
int (*legal_vif_num)(int) = 0;
#else
#define INSIZ sizeof(struct in_addr)
#define same(a1, a2) \
(bcmp((caddr_t)(a1), (caddr_t)(a2), INSIZ) == 0)
#define MT_MRTABLE MT_RTABLE /* since nothing else uses it */
/*
* Globals. All but ip_mrouter and ip_mrtproto could be static,
* except for netstat or debugging purposes.
*/
#ifndef MROUTE_LKM
struct socket *ip_mrouter = NULL;
struct mrtstat mrtstat;
int ip_mrtproto = IGMP_DVMRP; /* for netstat only */
#else
extern struct mrtstat mrtstat;
extern int ip_mrtproto;
#endif
#define NO_RTE_FOUND 0x1
#define RTE_FOUND 0x2
struct mbuf *mfctable[MFCTBLSIZ];
struct vif viftable[MAXVIFS];
u_int mrtdebug = 0; /* debug level */
u_int tbfdebug = 0; /* tbf debug level */
u_long timeout_val = 0; /* count of outstanding upcalls */
/*
* Define the token bucket filter structures
* tbftable -> each vif has one of these for storing info
* qtable -> each interface has an associated queue of pkts
*/
struct tbf tbftable[MAXVIFS];
struct pkt_queue qtable[MAXVIFS][MAXQSIZE];
/*
* '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
* the system ifnet list & no routes point to them. I.e., packets
* can't be sent this way. They only exist as a placeholder for
* multicast source verification.
*/
struct ifnet multicast_decap_if[MAXVIFS];
#define ENCAP_TTL 64
#define ENCAP_PROTO 4
/* prototype IP hdr for encapsulated packets */
struct ip multicast_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, ENCAP_PROTO,
0, /* checksum */
};
/*
* Private variables.
*/
static vifi_t numvifs = 0;
/*
* one-back cache used by multiencap_decap to locate a tunnel's vif
* given a datagram's src ip address.
*/
static u_long last_encap_src;
static struct vif *last_encap_vif;
static u_long nethash_fc(u_long, u_long);
static struct mfc *mfcfind(u_long, u_long);
int get_sg_cnt(struct sioc_sg_req *);
int get_vif_cnt(struct sioc_vif_req *);
int get_vifs(caddr_t);
static int add_vif(struct vifctl *);
static int del_vif(vifi_t *);
static int add_mfc(struct mfcctl *);
static int del_mfc(struct delmfcctl *);
static void cleanup_cache(void *);
static int ip_mdq(struct mbuf *, struct ifnet *, u_long, struct mfc *,
struct ip_moptions *);
extern int (*legal_vif_num)(int);
static void phyint_send(struct ip *, struct vif *, struct mbuf *);
static void srcrt_send(struct ip *, struct vif *, struct mbuf *);
static void encap_send(struct ip *, struct vif *, struct mbuf *);
void tbf_control(struct vif *, struct mbuf *, struct ip *, u_long,
struct ip_moptions *);
void tbf_queue(struct vif *, struct mbuf *, struct ip *, struct ip_moptions *);
void tbf_process_q(struct vif *);
void tbf_dequeue(struct vif *, int);
void tbf_reprocess_q(void *);
int tbf_dq_sel(struct vif *, struct ip *);
void tbf_send_packet(struct vif *, struct mbuf *, struct ip_moptions *);
void tbf_update_tokens(struct vif *);
static int priority(struct vif *, struct ip *);
static int ip_mrouter_init(struct socket *);
/*
* A simple hash function: returns MFCHASHMOD of the low-order octet of
* the argument's network or subnet number and the multicast group assoc.
*/
static u_long
nethash_fc(m,n)
register u_long m;
register u_long n;
{
struct in_addr in1;
struct in_addr in2;
in1.s_addr = m;
m = in_netof(in1);
while ((m & 0xff) == 0) m >>= 8;
in2.s_addr = n;
n = in_netof(in2);
while ((n & 0xff) == 0) n >>= 8;
return (MFCHASHMOD(m) ^ MFCHASHMOD(n));
}
/*
* this is a direct-mapped cache used to speed the mapping from a
* datagram source address to the associated multicast route. Note
* that unlike mrttable, the hash is on IP address, not IP net number.
*/
#define MFCHASHSIZ 1024
#define MFCHASH(a, g) ((((a) >> 20) ^ ((a) >> 10) ^ (a) ^ \
((g) >> 20) ^ ((g) >> 10) ^ (g)) & (MFCHASHSIZ-1))
struct mfc *mfchash[MFCHASHSIZ];
/*
* Find a route for a given origin IP address and Multicast group address
* Type of service parameter to be added in the future!!!
*/
#define MFCFIND(o, g, rt) { \
register u_int _mrhasho = o; \
register u_int _mrhashg = g; \
_mrhasho = MFCHASH(_mrhasho, _mrhashg); \
++mrtstat.mrts_mfc_lookups; \
rt = mfchash[_mrhasho]; \
if ((rt == NULL) || \
((o & rt->mfc_originmask.s_addr) != rt->mfc_origin.s_addr) || \
(g != rt->mfc_mcastgrp.s_addr)) \
if ((rt = mfcfind(o, g)) != NULL) \
mfchash[_mrhasho] = rt; \
}
/*
* Find route by examining hash table entries
*/
static struct mfc *
mfcfind(origin, mcastgrp)
u_long origin;
u_long mcastgrp;
{
register struct mbuf *mb_rt;
register struct mfc *rt;
register u_long hash;
hash = nethash_fc(origin, mcastgrp);
for (mb_rt = mfctable[hash]; mb_rt; mb_rt = mb_rt->m_next) {
rt = mtod(mb_rt, struct mfc *);
if (((origin & rt->mfc_originmask.s_addr) == rt->mfc_origin.s_addr) &&
(mcastgrp == rt->mfc_mcastgrp.s_addr) &&
(mb_rt->m_act == NULL))
return (rt);
}
mrtstat.mrts_mfc_misses++;
return NULL;
}
/*
* Macros to compute elapsed time efficiently
* Borrowed from Van Jacobson's scheduling code
*/
#define TV_DELTA(a, b, delta) { \
register int xxs; \
\
delta = (a).tv_usec - (b).tv_usec; \
if ((xxs = (a).tv_sec - (b).tv_sec)) { \
switch (xxs) { \
case 2: \
delta += 1000000; \
/* fall through */ \
case 1: \
delta += 1000000; \
break; \
default: \
delta += (1000000 * xxs); \
} \
} \
}
#define TV_LT(a, b) (((a).tv_usec < (b).tv_usec && \
(a).tv_sec <= (b).tv_sec) || (a).tv_sec < (b).tv_sec)
/*
* Handle DVMRP setsockopt commands to modify the multicast routing tables.
*/
int
X_ip_mrouter_cmd(cmd, so, m)
int cmd;
struct socket *so;
struct mbuf *m;
{
if (cmd != DVMRP_INIT && so != ip_mrouter) return EACCES;
switch (cmd) {
case DVMRP_INIT: return ip_mrouter_init(so);
case DVMRP_DONE: return ip_mrouter_done();
case DVMRP_ADD_VIF: return add_vif (mtod(m, struct vifctl *));
case DVMRP_DEL_VIF: return del_vif (mtod(m, vifi_t *));
case DVMRP_ADD_MFC: return add_mfc (mtod(m, struct mfcctl *));
case DVMRP_DEL_MFC: return del_mfc (mtod(m, struct delmfcctl *));
default: return EOPNOTSUPP;
}
}
#ifndef MROUTE_LKM
int (*ip_mrouter_cmd)(int, struct socket *, struct mbuf *) = X_ip_mrouter_cmd;
#endif
/*
* Handle ioctl commands to obtain information from the cache
*/
int
X_mrt_ioctl(cmd, data)
int cmd;
caddr_t data;
{
int error = 0;
switch (cmd) {
case (SIOCGETVIFINF): /* Read Virtual Interface (m/cast) */
return (get_vifs(data));
break;
case (SIOCGETVIFCNT):
return (get_vif_cnt((struct sioc_vif_req *)data));
break;
case (SIOCGETSGCNT):
return (get_sg_cnt((struct sioc_sg_req *)data));
break;
default:
return (EINVAL);
break;
}
return error;
}
#ifndef MROUTE_LKM
int (*mrt_ioctl)(int, caddr_t, struct proc *) = X_mrt_ioctl;
#else
extern int (*mrt_ioctl)(int, caddr_t, struct proc *);
#endif
/*
* returns the packet count for the source group provided
*/
int
get_sg_cnt(req)
register struct sioc_sg_req *req;
{
register struct mfc *rt;
int s;
s = splnet();
MFCFIND(req->src.s_addr, req->grp.s_addr, rt);
splx(s);
if (rt != NULL)
req->count = rt->mfc_pkt_cnt;
else
req->count = 0xffffffff;
return 0;
}
/*
* returns the input and output packet counts on the interface provided
*/
int
get_vif_cnt(req)
register struct sioc_vif_req *req;
{
register vifi_t vifi = req->vifi;
req->icount = viftable[vifi].v_pkt_in;
req->ocount = viftable[vifi].v_pkt_out;
return 0;
}
int
get_vifs(data)
char *data;
{
struct vif_conf *vifc = (struct vif_conf *)data;
struct vif_req *vifrp, vifr;
int space, error=0;
vifi_t vifi;
int s;
space = vifc->vifc_len;
vifrp = vifc->vifc_req;
s = splnet();
vifc->vifc_num=numvifs;
for (vifi = 0; vifi < numvifs; vifi++, vifrp++) {
if (viftable[vifi].v_lcl_addr.s_addr != 0) {
vifr.v_flags=viftable[vifi].v_flags;
vifr.v_threshold=viftable[vifi].v_threshold;
vifr.v_lcl_addr=viftable[vifi].v_lcl_addr;
vifr.v_rmt_addr=viftable[vifi].v_rmt_addr;
strncpy(vifr.v_if_name,viftable[vifi].v_ifp->if_name,IFNAMSIZ);
if ((space -= sizeof(vifr)) < 0) {
splx(s);
return(ENOSPC);
}
error = copyout((caddr_t)&vifr,(caddr_t)vifrp,(u_int)(sizeof vifr));
if (error) {
splx(s);
return(error);
}
}
}
splx(s);
return 0;
}
/*
* Enable multicast routing
*/
static int
ip_mrouter_init(so)
struct socket *so;
{
if (so->so_type != SOCK_RAW ||
so->so_proto->pr_protocol != IPPROTO_IGMP) return EOPNOTSUPP;
if (ip_mrouter != NULL) return EADDRINUSE;
ip_mrouter = so;
if (mrtdebug)
log(LOG_DEBUG, "ip_mrouter_init");
return 0;
}
/*
* Disable multicast routing
*/
int
X_ip_mrouter_done()
{
vifi_t vifi;
int i;
struct ifnet *ifp;
struct ifreq ifr;
struct mbuf *mb_rt;
struct mbuf *m;
struct rtdetq *rte;
int s;
s = splnet();
/*
* For each phyint in use, disable promiscuous reception of all IP
* multicasts.
*/
for (vifi = 0; vifi < numvifs; vifi++) {
if (viftable[vifi].v_lcl_addr.s_addr != 0 &&
!(viftable[vifi].v_flags & VIFF_TUNNEL)) {
((struct sockaddr_in *)&(ifr.ifr_addr))->sin_family = AF_INET;
((struct sockaddr_in *)&(ifr.ifr_addr))->sin_addr.s_addr
= INADDR_ANY;
ifp = viftable[vifi].v_ifp;
(*ifp->if_ioctl)(ifp, SIOCDELMULTI, (caddr_t)&ifr);
}
}
bzero((caddr_t)qtable, sizeof(qtable));
bzero((caddr_t)tbftable, sizeof(tbftable));
bzero((caddr_t)viftable, sizeof(viftable));
numvifs = 0;
/*
* Check if any outstanding timeouts remain
*/
if (timeout_val != 0)
for (i = 0; i < MFCTBLSIZ; i++) {
mb_rt = mfctable[i];
while (mb_rt) {
if ( mb_rt->m_act != NULL) {
untimeout(cleanup_cache, (caddr_t)mb_rt);
while (mb_rt->m_act) {
m = mb_rt->m_act;
mb_rt->m_act = m->m_act;
rte = mtod(m, struct rtdetq *);
m_freem(rte->m);
m_free(m);
}
timeout_val--;
}
mb_rt = mb_rt->m_next;
}
if (timeout_val == 0)
break;
}
/*
* Free all multicast forwarding cache entries.
*/
for (i = 0; i < MFCTBLSIZ; i++)
m_freem(mfctable[i]);
bzero((caddr_t)mfctable, sizeof(mfctable));
bzero((caddr_t)mfchash, sizeof(mfchash));
/*
* Reset de-encapsulation cache
*/
last_encap_src = NULL;
last_encap_vif = NULL;
ip_mrouter = NULL;
splx(s);
if (mrtdebug)
log(LOG_DEBUG, "ip_mrouter_done");
return 0;
}
#ifndef MROUTE_LKM
int (*ip_mrouter_done)(void) = X_ip_mrouter_done;
#endif
/*
* Add a vif to the vif table
*/
static int
add_vif(vifcp)
register struct vifctl *vifcp;
{
register struct vif *vifp = viftable + vifcp->vifc_vifi;
static struct sockaddr_in sin = {sizeof sin, AF_INET};
struct ifaddr *ifa;
struct ifnet *ifp;
struct ifreq ifr;
int error, s;
struct tbf *v_tbf = tbftable + vifcp->vifc_vifi;
if (vifcp->vifc_vifi >= MAXVIFS) return EINVAL;
if (vifp->v_lcl_addr.s_addr != 0) return EADDRINUSE;
/* 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 == 0) return EADDRNOTAVAIL;
ifp = ifa->ifa_ifp;
if (vifcp->vifc_flags & VIFF_TUNNEL) {
if ((vifcp->vifc_flags & VIFF_SRCRT) == 0) {
static int inited = 0;
if(!inited) {
for (s = 0; s < MAXVIFS; ++s) {
multicast_decap_if[s].if_name = "mdecap";
multicast_decap_if[s].if_unit = s;
}
inited = 1;
}
ifp = &multicast_decap_if[vifcp->vifc_vifi];
} else {
ifp = 0;
}
} else {
/* Make sure the interface supports multicast */
if ((ifp->if_flags & IFF_MULTICAST) == 0)
return EOPNOTSUPP;
/* Enable promiscuous reception of all IP multicasts from the if */
((struct sockaddr_in *)&(ifr.ifr_addr))->sin_family = AF_INET;
((struct sockaddr_in *)&(ifr.ifr_addr))->sin_addr.s_addr = INADDR_ANY;
s = splnet();
error = (*ifp->if_ioctl)(ifp, SIOCADDMULTI, (caddr_t)&ifr);
splx(s);
if (error)
return error;
}
s = splnet();
/* define parameters for the tbf structure */
vifp->v_tbf = v_tbf;
vifp->v_tbf->q_len = 0;
vifp->v_tbf->n_tok = 0;
vifp->v_tbf->last_pkt_t = 0;
vifp->v_flags = vifcp->vifc_flags;
vifp->v_threshold = vifcp->vifc_threshold;
vifp->v_lcl_addr = vifcp->vifc_lcl_addr;
vifp->v_rmt_addr = vifcp->vifc_rmt_addr;
vifp->v_ifp = ifp;
vifp->v_rate_limit= vifcp->vifc_rate_limit;
/* initialize per vif pkt counters */
vifp->v_pkt_in = 0;
vifp->v_pkt_out = 0;
splx(s);
/* Adjust numvifs up if the vifi is higher than numvifs */
if (numvifs <= vifcp->vifc_vifi) numvifs = vifcp->vifc_vifi + 1;
if (mrtdebug)
log(LOG_DEBUG, "add_vif #%d, lcladdr %x, %s %x, thresh %x, rate %d",
vifcp->vifc_vifi,
ntohl(vifcp->vifc_lcl_addr.s_addr),
(vifcp->vifc_flags & VIFF_TUNNEL) ? "rmtaddr" : "mask",
ntohl(vifcp->vifc_rmt_addr.s_addr),
vifcp->vifc_threshold,
vifcp->vifc_rate_limit);
return 0;
}
/*
* Delete a vif from the vif table
*/
static int
del_vif(vifip)
vifi_t *vifip;
{
register struct vif *vifp = viftable + *vifip;
register vifi_t vifi;
struct ifnet *ifp;
struct ifreq ifr;
int s;
if (*vifip >= numvifs) return EINVAL;
if (vifp->v_lcl_addr.s_addr == 0) return EADDRNOTAVAIL;
s = splnet();
if (!(vifp->v_flags & VIFF_TUNNEL)) {
((struct sockaddr_in *)&(ifr.ifr_addr))->sin_family = AF_INET;
((struct sockaddr_in *)&(ifr.ifr_addr))->sin_addr.s_addr = INADDR_ANY;
ifp = vifp->v_ifp;
(*ifp->if_ioctl)(ifp, SIOCDELMULTI, (caddr_t)&ifr);
}
if (vifp == last_encap_vif) {
last_encap_vif = 0;
last_encap_src = 0;
}
bzero((caddr_t)qtable[*vifip],
sizeof(qtable[*vifip]));
bzero((caddr_t)vifp->v_tbf, sizeof(*(vifp->v_tbf)));
bzero((caddr_t)vifp, sizeof (*vifp));
/* Adjust numvifs down */
for (vifi = numvifs; vifi > 0; vifi--)
if (viftable[vifi-1].v_lcl_addr.s_addr != 0) break;
numvifs = vifi;
splx(s);
if (mrtdebug)
log(LOG_DEBUG, "del_vif %d, numvifs %d", *vifip, numvifs);
return 0;
}
/*
* Add an mfc entry
*/
static int
add_mfc(mfccp)
struct mfcctl *mfccp;
{
struct mfc *rt;
struct mfc *rt1 = 0;
register struct mbuf *mb_rt;
struct mbuf *prev_mb_rt;
u_long hash;
struct mbuf *mb_ntry;
struct rtdetq *rte;
register u_short nstl;
int s;
int i;
rt = mfcfind(mfccp->mfcc_origin.s_addr, mfccp->mfcc_mcastgrp.s_addr);
/* If an entry already exists, just update the fields */
if (rt) {
if (mrtdebug)
log(LOG_DEBUG,"add_mfc update o %x g %x m %x p %x",
ntohl(mfccp->mfcc_origin.s_addr),
ntohl(mfccp->mfcc_mcastgrp.s_addr),
ntohl(mfccp->mfcc_originmask.s_addr),
mfccp->mfcc_parent);
s = splnet();
rt->mfc_parent = mfccp->mfcc_parent;
for (i = 0; i < numvifs; i++)
VIFM_COPY(mfccp->mfcc_ttls[i], rt->mfc_ttls[i]);
splx(s);
return 0;
}
/*
* Find the entry for which the upcall was made and update
*/
s = splnet();
hash = nethash_fc(mfccp->mfcc_origin.s_addr, mfccp->mfcc_mcastgrp.s_addr);
for (prev_mb_rt = mb_rt = mfctable[hash], nstl = 0;
mb_rt; prev_mb_rt = mb_rt, mb_rt = mb_rt->m_next) {
rt = mtod(mb_rt, struct mfc *);
if (((rt->mfc_origin.s_addr & mfccp->mfcc_originmask.s_addr)
== mfccp->mfcc_origin.s_addr) &&
(rt->mfc_mcastgrp.s_addr == mfccp->mfcc_mcastgrp.s_addr) &&
(mb_rt->m_act != NULL)) {
if (!nstl++) {
if (mrtdebug)
log(LOG_DEBUG,"add_mfc o %x g %x m %x p %x dbg %x",
ntohl(mfccp->mfcc_origin.s_addr),
ntohl(mfccp->mfcc_mcastgrp.s_addr),
ntohl(mfccp->mfcc_originmask.s_addr),
mfccp->mfcc_parent, mb_rt->m_act);
rt->mfc_origin = mfccp->mfcc_origin;
rt->mfc_originmask = mfccp->mfcc_originmask;
rt->mfc_mcastgrp = mfccp->mfcc_mcastgrp;
rt->mfc_parent = mfccp->mfcc_parent;
for (i = 0; i < numvifs; i++)
VIFM_COPY(mfccp->mfcc_ttls[i], rt->mfc_ttls[i]);
/* initialize pkt counters per src-grp */
rt->mfc_pkt_cnt = 0;
rt1 = rt;
}
/* prevent cleanup of cache entry */
untimeout(cleanup_cache, (caddr_t)mb_rt);
timeout_val--;
/* free packets Qed at the end of this entry */
while (mb_rt->m_act) {
mb_ntry = mb_rt->m_act;
rte = mtod(mb_ntry, struct rtdetq *);
ip_mdq(rte->m, rte->ifp, rte->tunnel_src,
rt1, rte->imo);
mb_rt->m_act = mb_ntry->m_act;
m_freem(rte->m);
m_free(mb_ntry);
}
/*
* If more than one entry was created for a single upcall
* delete that entry
*/
if (nstl > 1) {
MFREE(mb_rt, prev_mb_rt->m_next);
mb_rt = prev_mb_rt;
}
}
}
/*
* It is possible that an entry is being inserted without an upcall
*/
if (nstl == 0) {
if (mrtdebug)
log(LOG_DEBUG,"add_mfc no upcall h %d o %x g %x m %x p %x",
hash, ntohl(mfccp->mfcc_origin.s_addr),
ntohl(mfccp->mfcc_mcastgrp.s_addr),
ntohl(mfccp->mfcc_originmask.s_addr),
mfccp->mfcc_parent);
for (prev_mb_rt = mb_rt = mfctable[hash];
mb_rt; prev_mb_rt = mb_rt, mb_rt = mb_rt->m_next) {
rt = mtod(mb_rt, struct mfc *);
if (((rt->mfc_origin.s_addr & mfccp->mfcc_originmask.s_addr)
== mfccp->mfcc_origin.s_addr) &&
(rt->mfc_mcastgrp.s_addr == mfccp->mfcc_mcastgrp.s_addr)) {
rt->mfc_origin = mfccp->mfcc_origin;
rt->mfc_originmask = mfccp->mfcc_originmask;
rt->mfc_mcastgrp = mfccp->mfcc_mcastgrp;
rt->mfc_parent = mfccp->mfcc_parent;
for (i = 0; i < numvifs; i++)
VIFM_COPY(mfccp->mfcc_ttls[i], rt->mfc_ttls[i]);
/* initialize pkt counters per src-grp */
rt->mfc_pkt_cnt = 0;
}
}
if (mb_rt == NULL) {
/* no upcall, so make a new entry */
MGET(mb_rt, M_DONTWAIT, MT_MRTABLE);
if (mb_rt == NULL) {
splx(s);
return ENOBUFS;
}
rt = mtod(mb_rt, struct mfc *);
/* insert new entry at head of hash chain */
rt->mfc_origin = mfccp->mfcc_origin;
rt->mfc_originmask = mfccp->mfcc_originmask;
rt->mfc_mcastgrp = mfccp->mfcc_mcastgrp;
rt->mfc_parent = mfccp->mfcc_parent;
for (i = 0; i < numvifs; i++)
VIFM_COPY(mfccp->mfcc_ttls[i], rt->mfc_ttls[i]);
/* initialize pkt counters per src-grp */
rt->mfc_pkt_cnt = 0;
/* link into table */
mb_rt->m_next = mfctable[hash];
mfctable[hash] = mb_rt;
mb_rt->m_act = NULL;
}
}
splx(s);
return 0;
}
/*
* Delete an mfc entry
*/
static int
del_mfc(mfccp)
struct delmfcctl *mfccp;
{
struct in_addr origin;
struct in_addr mcastgrp;
struct mfc *rt;
struct mbuf *mb_rt;
struct mbuf *prev_mb_rt;
u_long hash;
struct mfc **cmfc;
struct mfc **cmfcend;
int s;
origin = mfccp->mfcc_origin;
mcastgrp = mfccp->mfcc_mcastgrp;
hash = nethash_fc(origin.s_addr, mcastgrp.s_addr);
if (mrtdebug)
log(LOG_DEBUG,"del_mfc orig %x mcastgrp %x",
ntohl(origin.s_addr), ntohl(mcastgrp.s_addr));
for (prev_mb_rt = mb_rt = mfctable[hash]
; mb_rt
; prev_mb_rt = mb_rt, mb_rt = mb_rt->m_next) {
rt = mtod(mb_rt, struct mfc *);
if (origin.s_addr == rt->mfc_origin.s_addr &&
mcastgrp.s_addr == rt->mfc_mcastgrp.s_addr &&
mb_rt->m_act == NULL)
break;
}
if (mb_rt == NULL) {
return ESRCH;
}
s = splnet();
cmfc = mfchash;
cmfcend = cmfc + MFCHASHSIZ;
for ( ; cmfc < cmfcend; ++cmfc)
if (*cmfc == rt)
*cmfc = 0;
if (prev_mb_rt != mb_rt) { /* if moved past head of list */
MFREE(mb_rt, prev_mb_rt->m_next);
} else /* delete head of list, it is in the table */
mfctable[hash] = m_free(mb_rt);
splx(s);
return 0;
}
/*
* IP multicast forwarding function. This function assumes that the packet
* pointed to by "ip" has arrived on (or is about to be sent to) the interface
* pointed to by "ifp", and the packet is to be relayed to other networks
* that have members of the packet's destination IP multicast group.
*
* The packet is returned unscathed to the caller, unless it is tunneled
* or erroneous, in which case a non-zero return value tells the caller to
* discard it.
*/
#define IP_HDR_LEN 20 /* # bytes of fixed IP header (excluding options) */
#define TUNNEL_LEN 12 /* # bytes of IP option for tunnel encapsulation */
int
X_ip_mforward(ip, ifp, m, imo)
register struct ip *ip;
struct ifnet *ifp;
struct mbuf *m;
struct ip_moptions *imo;
{
register struct mfc *rt;
register u_char *ipoptions;
u_long tunnel_src;
static struct sockproto k_igmpproto = { AF_INET, IPPROTO_IGMP };
static struct sockaddr_in k_igmpsrc = { sizeof k_igmpsrc, AF_INET };
static struct sockaddr_in k_igmpdst = { sizeof k_igmpdst, AF_INET };
register struct mbuf *mm;
register struct ip *k_data;
int s;
if (mrtdebug > 1)
log(LOG_DEBUG, "ip_mforward: src %x, dst %x, ifp %x",
ntohl(ip->ip_src.s_addr), ntohl(ip->ip_dst.s_addr), ifp);
if (ip->ip_hl < (IP_HDR_LEN + TUNNEL_LEN) >> 2 ||
(ipoptions = (u_char *)(ip + 1))[1] != IPOPT_LSRR ) {
/*
* Packet arrived via a physical interface.
*/
tunnel_src = 0;
} else {
/*
* Packet arrived through a source-route tunnel.
*
* A source-route tunneled packet has a single NOP option and a
* two-element
* loose-source-and-record-route (LSRR) option immediately following
* the fixed-size part of the IP header. At this point in processing,
* the IP header should contain the following IP addresses:
*
* original source - in the source address field
* destination group - in the destination address field
* remote tunnel end-point - in the first element of LSRR
* one of this host's addrs - in the second element of LSRR
*
* NOTE: RFC-1075 would have the original source and remote tunnel
* end-point addresses swapped. However, that could cause
* delivery of ICMP error messages to innocent applications
* on intermediate routing hosts! Therefore, we hereby
* change the spec.
*/
/*
* Verify that the tunnel options are well-formed.
*/
if (ipoptions[0] != IPOPT_NOP ||
ipoptions[2] != 11 || /* LSRR option length */
ipoptions[3] != 12 || /* LSRR address pointer */
(tunnel_src = *(u_long *)(&ipoptions[4])) == 0) {
mrtstat.mrts_bad_tunnel++;
if (mrtdebug)
log(LOG_DEBUG,
"ip_mforward: bad tunnel from %u (%x %x %x %x %x %x)",
ntohl(ip->ip_src.s_addr),
ipoptions[0], ipoptions[1], ipoptions[2], ipoptions[3],
*(u_long *)(&ipoptions[4]), *(u_long *)(&ipoptions[8]));
return 1;
}
/*
* Delete the tunnel options from the packet.
*/
ovbcopy((caddr_t)(ipoptions + TUNNEL_LEN), (caddr_t)ipoptions,
(unsigned)(m->m_len - (IP_HDR_LEN + TUNNEL_LEN)));
m->m_len -= TUNNEL_LEN;
ip->ip_len -= TUNNEL_LEN;
ip->ip_hl -= TUNNEL_LEN >> 2;
ifp = 0;
}
/*
* Don't forward a packet with time-to-live of zero or one,
* or a packet destined to a local-only group.
*/
if (ip->ip_ttl <= 1 ||
ntohl(ip->ip_dst.s_addr) <= INADDR_MAX_LOCAL_GROUP)
return (int)tunnel_src;
/*
* Determine forwarding vifs from the forwarding cache table
*/
s = splnet();
MFCFIND(ip->ip_src.s_addr, ip->ip_dst.s_addr, rt);
/* Entry exists, so forward if necessary */
if (rt != NULL) {
splx(s);
return (ip_mdq(m, ifp, tunnel_src, rt, imo));
}
else {
/*
* If we don't have a route for packet's origin,
* Make a copy of the packet &
* send message to routing daemon
*/
register struct mbuf *mb_rt;
register struct mbuf *mb_ntry;
register struct mbuf *mb0;
register struct rtdetq *rte;
register struct mbuf *rte_m;
register u_long hash;
mrtstat.mrts_no_route++;
if (mrtdebug)
log(LOG_DEBUG, "ip_mforward: no rte s %x g %x",
ntohl(ip->ip_src.s_addr),
ntohl(ip->ip_dst.s_addr));
/* is there an upcall waiting for this packet? */
hash = nethash_fc(ip->ip_src.s_addr, ip->ip_dst.s_addr);
for (mb_rt = mfctable[hash]; mb_rt; mb_rt = mb_rt->m_next) {
rt = mtod(mb_rt, struct mfc *);
if (((ip->ip_src.s_addr & rt->mfc_originmask.s_addr) ==
rt->mfc_origin.s_addr) &&
(ip->ip_dst.s_addr == rt->mfc_mcastgrp.s_addr) &&
(mb_rt->m_act != NULL))
break;
}
if (mb_rt == NULL) {
/* no upcall, so make a new entry */
MGET(mb_rt, M_DONTWAIT, MT_MRTABLE);
if (mb_rt == NULL) {
splx(s);
return ENOBUFS;
}
rt = mtod(mb_rt, struct mfc *);
/* insert new entry at head of hash chain */
rt->mfc_origin.s_addr = ip->ip_src.s_addr;
rt->mfc_originmask.s_addr = (u_long)0xffffffff;
rt->mfc_mcastgrp.s_addr = ip->ip_dst.s_addr;
/* link into table */
hash = nethash_fc(rt->mfc_origin.s_addr, rt->mfc_mcastgrp.s_addr);
mb_rt->m_next = mfctable[hash];
mfctable[hash] = mb_rt;
mb_rt->m_act = NULL;
}
/* determine if q has overflowed */
for (rte_m = mb_rt, hash = 0; rte_m->m_act; rte_m = rte_m->m_act)
hash++;
if (hash > MAX_UPQ) {
mrtstat.mrts_upq_ovflw++;
splx(s);
return 0;
}
/* add this packet and timing, ifp info to m_act */
MGET(mb_ntry, M_DONTWAIT, MT_DATA);
if (mb_ntry == NULL) {
splx(s);
return ENOBUFS;
}
mb_ntry->m_act = NULL;
rte = mtod(mb_ntry, struct rtdetq *);
mb0 = m_copy(m, 0, M_COPYALL);
if (mb0 == NULL) {
splx(s);
return ENOBUFS;
}
rte->m = mb0;
rte->ifp = ifp;
rte->tunnel_src = tunnel_src;
rte->imo = imo;
rte_m->m_act = mb_ntry;
splx(s);
if (hash == 0) {
/*
* Send message to routing daemon to install
* a route into the kernel table
*/
k_igmpsrc.sin_addr = ip->ip_src;
k_igmpdst.sin_addr = ip->ip_dst;
mm = m_copy(m, 0, M_COPYALL);
if (mm == NULL) {
splx(s);
return ENOBUFS;
}
k_data = mtod(mm, struct ip *);
k_data->ip_p = 0;
mrtstat.mrts_upcalls++;
raw_input(mm, &k_igmpproto,
(struct sockaddr *)&k_igmpsrc,
(struct sockaddr *)&k_igmpdst);
/* set timer to cleanup entry if upcall is lost */
timeout(cleanup_cache, (caddr_t)mb_rt, 100);
timeout_val++;
}
return 0;
}
}
#ifndef MROUTE_LKM
int (*ip_mforward)(struct ip *, struct ifnet *, struct mbuf *,
struct ip_moptions *) = X_ip_mforward;
#endif
/*
* Clean up the cache entry if upcall is not serviced
*/
static void
cleanup_cache(xmb_rt)
void *xmb_rt;
{
struct mbuf *mb_rt = xmb_rt;
struct mfc *rt;
u_long hash;
struct mbuf *prev_m0;
struct mbuf *m0;
struct mbuf *m;
struct rtdetq *rte;
int s;
rt = mtod(mb_rt, struct mfc *);
hash = nethash_fc(rt->mfc_origin.s_addr, rt->mfc_mcastgrp.s_addr);
if (mrtdebug)
log(LOG_DEBUG, "ip_mforward: cleanup ipm %d h %d s %x g %x",
ip_mrouter, hash, ntohl(rt->mfc_origin.s_addr),
ntohl(rt->mfc_mcastgrp.s_addr));
mrtstat.mrts_cache_cleanups++;
/*
* determine entry to be cleaned up in cache table
*/
s = splnet();
for (prev_m0 = m0 = mfctable[hash]; m0; prev_m0 = m0, m0 = m0->m_next)
if (m0 == mb_rt)
break;
/*
* drop all the packets
* free the mbuf with the pkt, if, timing info
*/
while (mb_rt->m_act) {
m = mb_rt->m_act;
mb_rt->m_act = m->m_act;
rte = mtod(m, struct rtdetq *);
m_freem(rte->m);
m_free(m);
}
/*
* Delete the entry from the cache
*/
if (prev_m0 != m0) { /* if moved past head of list */
MFREE(m0, prev_m0->m_next);
} else /* delete head of list, it is in the table */
mfctable[hash] = m_free(m0);
timeout_val--;
splx(s);
}
/*
* Packet forwarding routine once entry in the cache is made
*/
static int
ip_mdq(m, ifp, tunnel_src, rt, imo)
register struct mbuf *m;
register struct ifnet *ifp;
register u_long tunnel_src;
register struct mfc *rt;
register struct ip_moptions *imo;
{
register struct ip *ip = mtod(m, struct ip *);
register vifi_t vifi;
register struct vif *vifp;
/*
* Don't forward if it didn't arrive from the parent vif for its origin.
* Notes: v_ifp is zero for src route tunnels, multicast_decap_if
* for encapsulated tunnels and a real ifnet for non-tunnels so
* the first part of the if catches wrong physical interface or
* tunnel type; v_rmt_addr is zero for non-tunneled packets so
* the 2nd part catches both packets that arrive via a tunnel
* that shouldn't and packets that arrive via the wrong tunnel.
*/
vifi = rt->mfc_parent;
if (viftable[vifi].v_ifp != ifp ||
(ifp == 0 && viftable[vifi].v_rmt_addr.s_addr != tunnel_src)) {
/* came in the wrong interface */
if (mrtdebug)
log(LOG_DEBUG, "wrong if: ifp %x vifi %d",
ifp, vifi);
++mrtstat.mrts_wrong_if;
return (int)tunnel_src;
}
/* increment the interface and s-g counters */
viftable[vifi].v_pkt_in++;
rt->mfc_pkt_cnt++;
/*
* For each vif, decide if a copy of the packet should be forwarded.
* Forward if:
* - the ttl exceeds the vif's threshold
* - there are group members downstream on interface
*/
#define MC_SEND(ip,vifp,m) { \
(vifp)->v_pkt_out++; \
if ((vifp)->v_flags & VIFF_SRCRT) \
srcrt_send((ip), (vifp), (m)); \
else if ((vifp)->v_flags & VIFF_TUNNEL) \
encap_send((ip), (vifp), (m)); \
else \
phyint_send((ip), (vifp), (m)); \
}
/* If no options or the imo_multicast_vif option is 0, don't do this part
*/
if ((imo != NULL) &&
(( vifi = imo->imo_multicast_vif - 1) < numvifs) /*&& (vifi>=0)*/)
{
MC_SEND(ip,viftable+vifi,m);
return (1); /* make sure we are done: No more physical sends */
}
for (vifp = viftable, vifi = 0; vifi < numvifs; vifp++, vifi++)
if ((rt->mfc_ttls[vifi] > 0) &&
(ip->ip_ttl > rt->mfc_ttls[vifi]))
MC_SEND(ip, vifp, m);
return 0;
}
/* check if a vif number is legal/ok. This is used by ip_output, to export
* numvifs there,
*/
int
X_legal_vif_num(vif)
int vif;
{ if (vif>=0 && vif<=numvifs)
return(1);
else
return(0);
}
#ifndef MROUTE_LKM
int (*legal_vif_num)(int) = X_legal_vif_num;
#endif
static void
phyint_send(ip, vifp, m)
struct ip *ip;
struct vif *vifp;
struct mbuf *m;
{
register struct mbuf *mb_copy;
int hlen = ip->ip_hl << 2;
register struct ip_moptions *imo;
if ((mb_copy = m_copy(m, 0, M_COPYALL)) == NULL)
return;
/*
* Make sure the header isn't in an cluster, because the sharing
* in clusters defeats the whole purpose of making the copy above.
*/
mb_copy = m_pullup(mb_copy, hlen);
if (mb_copy == NULL)
return;
MALLOC(imo, struct ip_moptions *, sizeof *imo, M_IPMOPTS, M_NOWAIT);
if (imo == NULL) {
m_freem(mb_copy);
return;
}
imo->imo_multicast_ifp = vifp->v_ifp;
imo->imo_multicast_ttl = ip->ip_ttl - 1;
imo->imo_multicast_loop = 1;
if (vifp->v_rate_limit <= 0)
tbf_send_packet(vifp, mb_copy, imo);
else
tbf_control(vifp, mb_copy, mtod(mb_copy, struct ip *), ip->ip_len,
imo);
}
static void
srcrt_send(ip, vifp, m)
struct ip *ip;
struct vif *vifp;
struct mbuf *m;
{
struct mbuf *mb_copy, *mb_opts;
int hlen = ip->ip_hl << 2;
register struct ip *ip_copy;
u_char *cp;
/*
* Make sure that adding the tunnel options won't exceed the
* maximum allowed number of option bytes.
*/
if (ip->ip_hl > (60 - TUNNEL_LEN) >> 2) {
mrtstat.mrts_cant_tunnel++;
if (mrtdebug)
log(LOG_DEBUG, "srcrt_send: no room for tunnel options, from %u",
ntohl(ip->ip_src.s_addr));
return;
}
if ((mb_copy = m_copy(m, 0, M_COPYALL)) == NULL)
return;
MGETHDR(mb_opts, M_DONTWAIT, MT_HEADER);
if (mb_opts == NULL) {
m_freem(mb_copy);
return;
}
/*
* 'Delete' the base ip header from the mb_copy chain
*/
mb_copy->m_len -= hlen;
mb_copy->m_data += hlen;
/*
* Make mb_opts be the new head of the packet chain.
* Any options of the packet were left in the old packet chain head
*/
mb_opts->m_next = mb_copy;
mb_opts->m_len = hlen + TUNNEL_LEN;
mb_opts->m_data += MSIZE - mb_opts->m_len;
mb_opts->m_pkthdr.len = mb_copy->m_pkthdr.len + TUNNEL_LEN;
/*
* Copy the base ip header from the mb_copy chain to the new head mbuf
*/
ip_copy = mtod(mb_opts, struct ip *);
bcopy((caddr_t)ip_copy, mtod(mb_opts, caddr_t), hlen);
ip_copy->ip_ttl--;
ip_copy->ip_dst = vifp->v_rmt_addr; /* remote tunnel end-point */
/*
* Adjust the ip header length to account for the tunnel options.
*/
ip_copy->ip_hl += TUNNEL_LEN >> 2;
ip_copy->ip_len += TUNNEL_LEN;
/*
* Add the NOP and LSRR after the base ip header
*/
cp = mtod(mb_opts, u_char *) + IP_HDR_LEN;
*cp++ = IPOPT_NOP;
*cp++ = IPOPT_LSRR;
*cp++ = 11; /* LSRR option length */
*cp++ = 8; /* LSSR pointer to second element */
*(u_long*)cp = vifp->v_lcl_addr.s_addr; /* local tunnel end-point */
cp += 4;
*(u_long*)cp = ip->ip_dst.s_addr; /* destination group */
if (vifp->v_rate_limit <= 0)
tbf_send_packet(vifp, mb_opts, 0);
else
tbf_control(vifp, mb_opts,
mtod(mb_opts, struct ip *), ip_copy->ip_len, 0);
}
static void
encap_send(ip, vifp, m)
register struct ip *ip;
register struct vif *vifp;
register struct mbuf *m;
{
register struct mbuf *mb_copy;
register struct ip *ip_copy;
int hlen = ip->ip_hl << 2;
register int i, len = ip->ip_len;
/*
* copy the old packet & pullup it's IP header into the
* new mbuf so we can modify it. Try to fill the new
* mbuf since if we don't the ethernet driver will.
*/
MGET(mb_copy, M_DONTWAIT, MT_DATA);
if (mb_copy == NULL)
return;
mb_copy->m_data += 16;
mb_copy->m_len = sizeof(multicast_encap_iphdr);
if ((mb_copy->m_next = m_copy(m, 0, M_COPYALL)) == NULL) {
m_freem(mb_copy);
return;
}
i = MHLEN - M_LEADINGSPACE(mb_copy);
if (i > len)
i = len;
mb_copy = m_pullup(mb_copy, i);
if (mb_copy == NULL)
return;
mb_copy->m_pkthdr.len = len + sizeof(multicast_encap_iphdr);
/*
* fill in the encapsulating IP header.
*/
ip_copy = mtod(mb_copy, struct ip *);
*ip_copy = multicast_encap_iphdr;
ip_copy->ip_id = htons(ip_id++);
ip_copy->ip_len += len;
ip_copy->ip_src = vifp->v_lcl_addr;
ip_copy->ip_dst = vifp->v_rmt_addr;
/*
* turn the encapsulated IP header back into a valid one.
*/
ip = (struct ip *)((caddr_t)ip_copy + sizeof(multicast_encap_iphdr));
--ip->ip_ttl;
HTONS(ip->ip_len);
HTONS(ip->ip_off);
ip->ip_sum = 0;
#if defined(LBL) && !defined(ultrix)
ip->ip_sum = ~oc_cksum((caddr_t)ip, ip->ip_hl << 2, 0);
#else
mb_copy->m_data += sizeof(multicast_encap_iphdr);
ip->ip_sum = in_cksum(mb_copy, ip->ip_hl << 2);
mb_copy->m_data -= sizeof(multicast_encap_iphdr);
#endif
if (vifp->v_rate_limit <= 0)
tbf_send_packet(vifp, mb_copy, 0);
else
tbf_control(vifp, mb_copy, ip, ip_copy->ip_len, 0);
}
/*
* De-encapsulate a packet and feed it back through ip input (this
* routine is called whenever IP gets a packet with proto type
* ENCAP_PROTO and a local destination address).
*/
void
#ifdef MROUTE_LKM
X_multiencap_decap(m)
#else
multiencap_decap(m)
#endif
register struct mbuf *m;
{
struct ifnet *ifp = m->m_pkthdr.rcvif;
register struct ip *ip = mtod(m, struct ip *);
register int hlen = ip->ip_hl << 2;
register int s;
register struct ifqueue *ifq;
register struct vif *vifp;
if (ip->ip_p != ENCAP_PROTO) {
rip_input(m);
return;
}
/*
* dump the packet if it's not to a multicast destination or if
* we don't have an encapsulating tunnel with the source.
* Note: This code assumes that the remote site IP address
* uniquely identifies the tunnel (i.e., that this site has
* at most one tunnel with the remote site).
*/
if (! IN_MULTICAST(ntohl(((struct ip *)((char *)ip + hlen))->ip_dst.s_addr))) {
++mrtstat.mrts_bad_tunnel;
m_freem(m);
return;
}
if (ip->ip_src.s_addr != last_encap_src) {
register struct vif *vife;
vifp = viftable;
vife = vifp + numvifs;
last_encap_src = ip->ip_src.s_addr;
last_encap_vif = 0;
for ( ; vifp < vife; ++vifp)
if (vifp->v_rmt_addr.s_addr == ip->ip_src.s_addr) {
if ((vifp->v_flags & (VIFF_TUNNEL|VIFF_SRCRT))
== VIFF_TUNNEL)
last_encap_vif = vifp;
break;
}
}
if ((vifp = last_encap_vif) == 0) {
last_encap_src = 0;
mrtstat.mrts_cant_tunnel++; /*XXX*/
m_freem(m);
if (mrtdebug)
log(LOG_DEBUG, "ip_mforward: no tunnel with %u",
ntohl(ip->ip_src.s_addr));
return;
}
ifp = vifp->v_ifp;
hlen -= sizeof(struct ifnet *);
m->m_data += hlen;
m->m_len -= hlen;
*(mtod(m, struct ifnet **)) = ifp;
ifq = &ipintrq;
s = splimp();
if (IF_QFULL(ifq)) {
IF_DROP(ifq);
m_freem(m);
} else {
IF_ENQUEUE(ifq, m);
/*
* normally we would need a "schednetisr(NETISR_IP)"
* here but we were called by ip_input and it is going
* to loop back & try to dequeue the packet we just
* queued as soon as we return so we avoid the
* unnecessary software interrrupt.
*/
}
splx(s);
}
/*
* Token bucket filter module
*/
void
tbf_control(vifp, m, ip, p_len, imo)
register struct vif *vifp;
register struct mbuf *m;
register struct ip *ip;
register u_long p_len;
struct ip_moptions *imo;
{
tbf_update_tokens(vifp);
/* if there are enough tokens,
* and the queue is empty,
* send this packet out
*/
if (vifp->v_tbf->q_len == 0) {
if (p_len <= vifp->v_tbf->n_tok) {
vifp->v_tbf->n_tok -= p_len;
tbf_send_packet(vifp, m, imo);
} else if (p_len > MAX_BKT_SIZE) {
/* drop if packet is too large */
mrtstat.mrts_pkt2large++;
m_freem(m);
return;
} else {
/* queue packet and timeout till later */
tbf_queue(vifp, m, ip, imo);
timeout(tbf_reprocess_q, (caddr_t)vifp, 1);
}
} else if (vifp->v_tbf->q_len < MAXQSIZE) {
/* finite queue length, so queue pkts and process queue */
tbf_queue(vifp, m, ip, imo);
tbf_process_q(vifp);
} else {
/* queue length too much, try to dq and queue and process */
if (!tbf_dq_sel(vifp, ip)) {
mrtstat.mrts_q_overflow++;
m_freem(m);
return;
} else {
tbf_queue(vifp, m, ip, imo);
tbf_process_q(vifp);
}
}
return;
}
/*
* adds a packet to the queue at the interface
*/
void
tbf_queue(vifp, m, ip, imo)
register struct vif *vifp;
register struct mbuf *m;
register struct ip *ip;
struct ip_moptions *imo;
{
register u_long ql;
register int index = (vifp - viftable);
register int s = splnet();
ql = vifp->v_tbf->q_len;
qtable[index][ql].pkt_m = m;
qtable[index][ql].pkt_len = (mtod(m, struct ip *))->ip_len;
qtable[index][ql].pkt_ip = ip;
qtable[index][ql].pkt_imo = imo;
vifp->v_tbf->q_len++;
splx(s);
}
/*
* processes the queue at the interface
*/
void
tbf_process_q(vifp)
register struct vif *vifp;
{
register struct pkt_queue pkt_1;
register int index = (vifp - viftable);
register int s = splnet();
/* loop through the queue at the interface and send as many packets
* as possible
*/
while (vifp->v_tbf->q_len > 0) {
/* locate the first packet */
pkt_1.pkt_len = ((qtable[index][0]).pkt_len);
pkt_1.pkt_m = (qtable[index][0]).pkt_m;
pkt_1.pkt_ip = (qtable[index][0]).pkt_ip;
pkt_1.pkt_imo = (qtable[index][0]).pkt_imo;
/* determine if the packet can be sent */
if (pkt_1.pkt_len <= vifp->v_tbf->n_tok) {
/* if so,
* reduce no of tokens, dequeue the queue,
* send the packet.
*/
vifp->v_tbf->n_tok -= pkt_1.pkt_len;
tbf_dequeue(vifp, 0);
tbf_send_packet(vifp, pkt_1.pkt_m, pkt_1.pkt_imo);
} else break;
}
splx(s);
}
/*
* removes the jth packet from the queue at the interface
*/
void
tbf_dequeue(vifp,j)
register struct vif *vifp;
register int j;
{
register u_long index = vifp - viftable;
register int i;
for (i=j+1; i <= vifp->v_tbf->q_len - 1; i++) {
qtable[index][i-1].pkt_m = qtable[index][i].pkt_m;
qtable[index][i-1].pkt_len = qtable[index][i].pkt_len;
qtable[index][i-1].pkt_ip = qtable[index][i].pkt_ip;
qtable[index][i-1].pkt_imo = qtable[index][i].pkt_imo;
}
qtable[index][i-1].pkt_m = NULL;
qtable[index][i-1].pkt_len = NULL;
qtable[index][i-1].pkt_ip = NULL;
qtable[index][i-1].pkt_imo = NULL;
vifp->v_tbf->q_len--;
if (tbfdebug > 1)
log(LOG_DEBUG, "tbf_dequeue: vif# %d qlen %d",vifp-viftable, i-1);
}
void
tbf_reprocess_q(xvifp)
void *xvifp;
{
register struct vif *vifp = xvifp;
if (ip_mrouter == NULL)
return;
tbf_update_tokens(vifp);
tbf_process_q(vifp);
if (vifp->v_tbf->q_len)
timeout(tbf_reprocess_q, (caddr_t)vifp, 1);
}
/* function that will selectively discard a member of the queue
* based on the precedence value and the priority obtained through
* a lookup table - not yet implemented accurately!
*/
int
tbf_dq_sel(vifp, ip)
register struct vif *vifp;
register struct ip *ip;
{
register int i;
register int s = splnet();
register u_int p;
p = priority(vifp, ip);
for(i=vifp->v_tbf->q_len-1;i >= 0;i--) {
if (p > priority(vifp, qtable[vifp-viftable][i].pkt_ip)) {
m_freem(qtable[vifp-viftable][i].pkt_m);
tbf_dequeue(vifp,i);
splx(s);
mrtstat.mrts_drop_sel++;
return(1);
}
}
splx(s);
return(0);
}
void
tbf_send_packet(vifp, m, imo)
register struct vif *vifp;
register struct mbuf *m;
struct ip_moptions *imo;
{
int error;
int s = splnet();
/* if source route tunnels */
if (vifp->v_flags & VIFF_SRCRT) {
error = ip_output(m, (struct mbuf *)0, (struct route *)0,
IP_FORWARDING, imo);
if (mrtdebug > 1)
log(LOG_DEBUG, "srcrt_send on vif %d err %d", vifp-viftable, error);
} else if (vifp->v_flags & VIFF_TUNNEL) {
/* If tunnel options */
ip_output(m, (struct mbuf *)0, (struct route *)0,
IP_FORWARDING, imo);
} else {
/* if physical interface option, extract the options and then send */
error = ip_output(m, (struct mbuf *)0, (struct route *)0,
IP_FORWARDING, imo);
FREE(imo, M_IPMOPTS);
if (mrtdebug > 1)
log(LOG_DEBUG, "phyint_send on vif %d err %d", vifp-viftable, error);
}
splx(s);
}
/* determine the current time and then
* the elapsed time (between the last time and time now)
* in milliseconds & update the no. of tokens in the bucket
*/
void
tbf_update_tokens(vifp)
register struct vif *vifp;
{
struct timeval tp;
register u_long t;
register u_long elapsed;
register int s = splnet();
GET_TIME(tp);
t = tp.tv_sec*1000 + tp.tv_usec/1000;
elapsed = (t - vifp->v_tbf->last_pkt_t) * vifp->v_rate_limit /8;
vifp->v_tbf->n_tok += elapsed;
vifp->v_tbf->last_pkt_t = t;
if (vifp->v_tbf->n_tok > MAX_BKT_SIZE)
vifp->v_tbf->n_tok = MAX_BKT_SIZE;
splx(s);
}
static int
priority(vifp, ip)
register struct vif *vifp;
register struct ip *ip;
{
register u_long graddr;
register int prio;
/* temporary hack; will add general packet classifier some day */
prio = 50; /* default priority */
/* check for source route options and add option length to get dst */
if (vifp->v_flags & VIFF_SRCRT)
graddr = ntohl((ip+8)->ip_dst.s_addr);
else
graddr = ntohl(ip->ip_dst.s_addr);
switch (graddr & 0xf) {
case 0x0: break;
case 0x1: if (graddr == 0xe0020001) prio = 65; /* MBone Audio */
break;
case 0x2: break;
case 0x3: break;
case 0x4: break;
case 0x5: break;
case 0x6: break;
case 0x7: break;
case 0x8: break;
case 0x9: break;
case 0xa: if (graddr == 0xe000010a) prio = 85; /* IETF Low Audio 1 */
break;
case 0xb: if (graddr == 0xe000010b) prio = 75; /* IETF Audio 1 */
break;
case 0xc: if (graddr == 0xe000010c) prio = 60; /* IETF Video 1 */
break;
case 0xd: if (graddr == 0xe000010d) prio = 80; /* IETF Low Audio 2 */
break;
case 0xe: if (graddr == 0xe000010e) prio = 70; /* IETF Audio 2 */
break;
case 0xf: if (graddr == 0xe000010f) prio = 55; /* IETF Video 2 */
break;
}
if (tbfdebug > 1) log(LOG_DEBUG, "graddr%x prio%d", graddr, prio);
return prio;
}
/*
* End of token bucket filter modifications
*/
#ifdef MROUTE_LKM
#include <sys/conf.h>
#include <sys/exec.h>
#include <sys/sysent.h>
#include <sys/lkm.h>
MOD_MISC("ip_mroute_mod")
static int
ip_mroute_mod_handle(struct lkm_table *lkmtp, int cmd)
{
int i;
struct lkm_misc *args = lkmtp->private.lkm_misc;
int err = 0;
switch(cmd) {
static int (*old_ip_mrouter_cmd)();
static int (*old_ip_mrouter_done)();
static int (*old_ip_mforward)();
static int (*old_mrt_ioctl)();
static int (*old_proto4_input)();
static int (*old_legal_vif_num)();
extern u_char ip_protox[];
extern struct protosw inetsw[];
case LKM_E_LOAD:
if(lkmexists(lkmtp) || ip_mrtproto)
return(EEXIST);
old_ip_mrouter_cmd = ip_mrouter_cmd;
ip_mrouter_cmd = X_ip_mrouter_cmd;
old_ip_mrouter_done = ip_mrouter_done;
ip_mrouter_done = X_ip_mrouter_done;
old_ip_mforward = ip_mforward;
ip_mforward = X_ip_mforward;
old_mrt_ioctl = mrt_ioctl;
mrt_ioctl = X_mrt_ioctl;
old_proto4_input = inetsw[ip_protox[IPPROTO_ENCAP]].pr_input;
inetsw[ip_protox[IPPROTO_ENCAP]].pr_input = X_multiencap_decap;
old_legal_vif_num = legal_vif_num;
legal_vif_num = X_legal_vif_num;
ip_mrtproto = IGMP_DVMRP;
printf("\nIP multicast routing loaded\n");
break;
case LKM_E_UNLOAD:
if (ip_mrouter)
return EINVAL;
ip_mrouter_cmd = old_ip_mrouter_cmd;
ip_mrouter_done = old_ip_mrouter_done;
ip_mforward = old_ip_mforward;
mrt_ioctl = old_mrt_ioctl;
inetsw[ip_protox[IPPROTO_ENCAP]].pr_input = old_proto4_input;
legal_vif_num = old_legal_vif_num;
ip_mrtproto = 0;
break;
default:
err = EINVAL;
break;
}
return(err);
}
int
ip_mroute_mod(struct lkm_table *lkmtp, int cmd, int ver) {
DISPATCH(lkmtp, cmd, ver, ip_mroute_mod_handle, ip_mroute_mod_handle,
nosys);
}
#endif /* MROUTE_LKM */
#endif /* MROUTING */