NetBSD/usr.sbin/pim6dd/mld6.c
itojun 8912746994 update IPv6 MLD API to new one. various corrections including
timeout processing, node-local/link-local handling (ignore them),
and more.  (sync with kame)
nroff fixes to manpage.
2000-05-19 10:43:36 +00:00

548 lines
17 KiB
C

/* $NetBSD: mld6.c,v 1.6 2000/05/19 10:43:38 itojun Exp $ */
/*
* Copyright (C) 1998 WIDE Project.
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
* 3. Neither the name of the project nor the names of its contributors
* may be used to endorse or promote products derived from this software
* without specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE
* FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
* SUCH DAMAGE.
*/
/*
* Copyright (c) 1998 by the University of Southern California.
* All rights reserved.
*
* Permission to use, copy, modify, and distribute this software and
* its documentation in source and binary forms for lawful
* purposes and without fee is hereby granted, provided
* that the above copyright notice appear in all copies and that both
* the copyright notice and this permission notice appear in supporting
* documentation, and that any documentation, advertising materials,
* and other materials related to such distribution and use acknowledge
* that the software was developed by the University of Southern
* California and/or Information Sciences Institute.
* The name of the University of Southern California may not
* be used to endorse or promote products derived from this software
* without specific prior written permission.
*
* THE UNIVERSITY OF SOUTHERN CALIFORNIA DOES NOT MAKE ANY REPRESENTATIONS
* ABOUT THE SUITABILITY OF THIS SOFTWARE FOR ANY PURPOSE. THIS SOFTWARE IS
* PROVIDED "AS IS" AND WITHOUT ANY EXPRESS OR IMPLIED WARRANTIES,
* INCLUDING, WITHOUT LIMITATION, THE IMPLIED WARRANTIES OF
* MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE, TITLE, AND
* NON-INFRINGEMENT.
*
* IN NO EVENT SHALL USC, OR ANY OTHER CONTRIBUTOR BE LIABLE FOR ANY
* SPECIAL, INDIRECT OR CONSEQUENTIAL DAMAGES, WHETHER IN CONTRACT,
* TORT, OR OTHER FORM OF ACTION, ARISING OUT OF OR IN CONNECTION WITH,
* THE USE OR PERFORMANCE OF THIS SOFTWARE.
*
* Other copyrights might apply to parts of this software and are so
* noted when applicable.
*/
/*
* Questions concerning this software should be directed to
* Pavlin Ivanov Radoslavov (pavlin@catarina.usc.edu)
*
* KAME Id: mld6.c,v 1.13 2000/04/12 07:34:38 jinmei Exp
*/
/*
* Part of this program has been derived from mrouted.
* The mrouted program is covered by the license in the accompanying file
* named "LICENSE.mrouted".
*
* The mrouted program is COPYRIGHT 1989 by The Board of Trustees of
* Leland Stanford Junior University.
*
*/
#include "defs.h"
#include <sys/uio.h>
/*
* Exported variables.
*/
char *mld6_recv_buf; /* input packet buffer */
char *mld6_send_buf; /* output packet buffer */
int mld6_socket; /* socket for all network I/O */
struct sockaddr_in6 allrouters_group = {sizeof(struct sockaddr_in6), AF_INET6};
struct sockaddr_in6 allnodes_group = {sizeof(struct sockaddr_in6), AF_INET6};
/* Extenals */
extern struct in6_addr in6addr_linklocal_allnodes;
/* local variables. */
static struct sockaddr_in6 dst = {sizeof(dst), AF_INET6};
static struct msghdr sndmh,
rcvmh;
static struct iovec sndiov[2];
static struct iovec rcviov[2];
static struct sockaddr_in6 from;
static u_char *rcvcmsgbuf = NULL;
static int rcvcmsglen;
#ifndef USE_RFC2292BIS
u_int8_t raopt[IP6OPT_RTALERT_LEN];
#endif
static char *sndcmsgbuf;
static int ctlbuflen = 0;
static u_short rtalert_code;
/* local functions */
static void mld6_read __P((int i, fd_set * fds));
static void accept_mld6 __P((int len));
static void make_mld6_msg __P((int, int, struct sockaddr_in6 *,
struct sockaddr_in6 *, struct in6_addr *, int, int, int, int));
#ifndef IP6OPT_ROUTER_ALERT /* XXX to be compatible older systems */
#define IP6OPT_ROUTER_ALERT IP6OPT_RTALERT
#endif
/*
* Open and initialize the MLD socket.
*/
void
init_mld6()
{
struct icmp6_filter filt;
int on;
rtalert_code = htons(IP6OPT_RTALERT_MLD);
if (!mld6_recv_buf && (mld6_recv_buf = malloc(RECV_BUF_SIZE)) == NULL)
log(LOG_ERR, 0, "malloc failed");
if (!mld6_send_buf && (mld6_send_buf = malloc(RECV_BUF_SIZE)) == NULL)
log(LOG_ERR, 0, "malloc failed");
rcvcmsglen = CMSG_SPACE(sizeof(struct in6_pktinfo)) +
CMSG_SPACE(sizeof(int));
if (rcvcmsgbuf == NULL && (rcvcmsgbuf = malloc(rcvcmsglen)) == NULL)
log(LOG_ERR, 0,"malloc failed");
IF_DEBUG(DEBUG_KERN)
log(LOG_DEBUG,0,"%d octets allocated for the emit/recept buffer mld6",RECV_BUF_SIZE);
if ((mld6_socket = socket(AF_INET6, SOCK_RAW, IPPROTO_ICMPV6)) < 0)
log(LOG_ERR, errno, "MLD6 socket");
k_set_rcvbuf(mld6_socket, SO_RECV_BUF_SIZE_MAX,
SO_RECV_BUF_SIZE_MIN); /* lots of input buffering */
k_set_hlim(mld6_socket, MINHLIM); /* restrict multicasts to one hop */
k_set_loop(mld6_socket, FALSE); /* disable multicast loopback */
/* address initialization */
allnodes_group.sin6_addr = in6addr_linklocal_allnodes;
if (inet_pton(AF_INET6, "ff02::2",
(void *) &allrouters_group.sin6_addr) != 1)
log(LOG_ERR, 0, "inet_pton failed for ff02::2");
/* filter all non-MLD ICMP messages */
ICMP6_FILTER_SETBLOCKALL(&filt);
ICMP6_FILTER_SETPASS(ICMP6_MEMBERSHIP_QUERY, &filt);
ICMP6_FILTER_SETPASS(ICMP6_MEMBERSHIP_REPORT, &filt);
ICMP6_FILTER_SETPASS(ICMP6_MEMBERSHIP_REDUCTION, &filt);
ICMP6_FILTER_SETPASS(MLD6_MTRACE_RESP, &filt);
ICMP6_FILTER_SETPASS(MLD6_MTRACE, &filt);
if (setsockopt(mld6_socket, IPPROTO_ICMPV6, ICMP6_FILTER, &filt,
sizeof(filt)) < 0)
log(LOG_ERR, errno, "setsockopt(ICMP6_FILTER)");
/* specify to tell receiving interface */
on = 1;
#ifdef IPV6_RECVPKTINFO
if (setsockopt(mld6_socket, IPPROTO_IPV6, IPV6_RECVPKTINFO, &on,
sizeof(on)) < 0)
log(LOG_ERR, errno, "setsockopt(IPV6_RECVPKTINFO)");
#else /* old adv. API */
if (setsockopt(mld6_socket, IPPROTO_IPV6, IPV6_PKTINFO, &on,
sizeof(on)) < 0)
log(LOG_ERR, errno, "setsockopt(IPV6_PKTINFO)");
#endif
on = 1;
/* specify to tell value of hoplimit field of received IP6 hdr */
#ifdef IPV6_RECVHOPLIMIT
if (setsockopt(mld6_socket, IPPROTO_IPV6, IPV6_RECVHOPLIMIT, &on,
sizeof(on)) < 0)
log(LOG_ERR, errno, "setsockopt(IPV6_RECVHOPLIMIT)");
#else /* old adv. API */
if (setsockopt(mld6_socket, IPPROTO_IPV6, IPV6_HOPLIMIT, &on,
sizeof(on)) < 0)
log(LOG_ERR, errno, "setsockopt(IPV6_HOPLIMIT)");
#endif
/* initialize msghdr for receiving packets */
rcviov[0].iov_base = (caddr_t) mld6_recv_buf;
rcviov[0].iov_len = RECV_BUF_SIZE;
rcvmh.msg_name = (caddr_t) & from;
rcvmh.msg_namelen = sizeof(from);
rcvmh.msg_iov = rcviov;
rcvmh.msg_iovlen = 1;
rcvmh.msg_control = (caddr_t) rcvcmsgbuf;
rcvmh.msg_controllen = rcvcmsglen;
/* initialize msghdr for sending packets */
sndiov[0].iov_base = (caddr_t)mld6_send_buf;
sndmh.msg_namelen = sizeof(struct sockaddr_in6);
sndmh.msg_iov = sndiov;
sndmh.msg_iovlen = 1;
/* specifiy to insert router alert option in a hop-by-hop opt hdr. */
#ifndef USE_RFC2292BIS
raopt[0] = IP6OPT_ROUTER_ALERT;
raopt[1] = IP6OPT_RTALERT_LEN - 2;
memcpy(&raopt[2], (caddr_t) & rtalert_code, sizeof(u_short));
#endif
/* register MLD message handler */
if (register_input_handler(mld6_socket, mld6_read) < 0)
log(LOG_ERR, 0,
"Couldn't register mld6_read as an input handler");
}
/* Read an MLD message */
static void
mld6_read(i, rfd)
int i;
fd_set *rfd;
{
register int mld6_recvlen;
mld6_recvlen = recvmsg(mld6_socket, &rcvmh, 0);
if (mld6_recvlen < 0)
{
if (errno != EINTR)
log(LOG_ERR, errno, "MLD6 recvmsg");
return;
}
/* TODO: make it as a thread in the future releases */
accept_mld6(mld6_recvlen);
}
/*
* Process a newly received MLD6 packet that is sitting in the input packet
* buffer.
*/
static void
accept_mld6(recvlen)
int recvlen;
{
struct in6_addr *group, *dst = NULL;
struct mld6_hdr *mldh;
struct cmsghdr *cm;
struct in6_pktinfo *pi = NULL;
int *hlimp = NULL;
int ifindex = 0;
struct sockaddr_in6 *src = (struct sockaddr_in6 *) rcvmh.msg_name;
if (recvlen < sizeof(struct mld6_hdr))
{
log(LOG_WARNING, 0,
"received packet too short (%u bytes) for MLD header",
recvlen);
return;
}
mldh = (struct mld6_hdr *) rcvmh.msg_iov[0].iov_base;
/*
* Packets sent up from kernel to daemon have ICMPv6 type = 0.
* Note that we set filters on the mld6_socket, so we should never
* see a "normal" ICMPv6 packet with type 0 of ICMPv6 type.
*/
if (mldh->mld6_type == 0) {
/* XXX: msg_controllen must be reset in this case. */
rcvmh.msg_controllen = rcvcmsglen;
process_kernel_call();
return;
}
group = &mldh->mld6_addr;
/* extract optional information via Advanced API */
for (cm = (struct cmsghdr *) CMSG_FIRSTHDR(&rcvmh);
cm;
cm = (struct cmsghdr *) CMSG_NXTHDR(&rcvmh, cm))
{
if (cm->cmsg_level == IPPROTO_IPV6 &&
cm->cmsg_type == IPV6_PKTINFO &&
cm->cmsg_len == CMSG_LEN(sizeof(struct in6_pktinfo)))
{
pi = (struct in6_pktinfo *) (CMSG_DATA(cm));
ifindex = pi->ipi6_ifindex;
dst = &pi->ipi6_addr;
}
if (cm->cmsg_level == IPPROTO_IPV6 &&
cm->cmsg_type == IPV6_HOPLIMIT &&
cm->cmsg_len == CMSG_LEN(sizeof(int)))
hlimp = (int *) CMSG_DATA(cm);
}
if (hlimp == NULL)
{
log(LOG_WARNING, 0,
"failed to get receiving hop limit");
return;
}
/* TODO: too noisy. Remove it? */
//#define NOSUCHDEF
#ifdef NOSUCHDEF
IF_DEBUG(DEBUG_PKT | debug_kind(IPPROTO_ICMPV6, mldh->mld6_type,
mldh->mld6_code))
log(LOG_DEBUG, 0, "RECV %s from %s to %s",
packet_kind(IPPROTO_ICMPV6,
mldh->mld6_type, mldh->mld6_code),
inet6_fmt(&src->sin6_addr), inet6_fmt(dst));
#endif /* NOSUCHDEF */
/* for an mtrace message, we don't need strict checks */
if (mldh->mld6_type == MLD6_MTRACE) {
accept_mtrace(src, dst, group, ifindex, (char *)(mldh + 1),
mldh->mld6_code, recvlen - sizeof(struct mld6_hdr));
return;
}
/* hop limit check */
if (*hlimp != 1)
{
log(LOG_WARNING, 0,
"received an MLD6 message with illegal hop limit(%d) from %s",
*hlimp, inet6_fmt(&src->sin6_addr));
/* but accept the packet */
}
if (ifindex == 0)
{
log(LOG_WARNING, 0, "failed to get receiving interface");
return;
}
/* scope check */
if (IN6_IS_ADDR_MC_NODELOCAL(&mldh->mld6_addr))
{
log(LOG_INFO, 0,
"RECV %s with an invalid scope: %s from %s",
inet6_fmt(&mldh->mld6_addr),
inet6_fmt(&src->sin6_addr));
return; /* discard */
}
/* source address check */
if (!IN6_IS_ADDR_LINKLOCAL(&src->sin6_addr))
{
log(LOG_INFO, 0,
"RECV %s from a non link local address: %s",
packet_kind(IPPROTO_ICMPV6, mldh->mld6_type,
mldh->mld6_code),
inet6_fmt(&src->sin6_addr));
return;
}
switch (mldh->mld6_type)
{
case MLD6_LISTENER_QUERY:
accept_listener_query(src, dst, group,
ntohs(mldh->mld6_maxdelay));
return;
case MLD6_LISTENER_REPORT:
accept_listener_report(src, dst, group);
return;
case MLD6_LISTENER_DONE:
accept_listener_done(src, dst, group);
return;
default:
/* This must be impossible since we set a type filter */
log(LOG_INFO, 0,
"ignoring unknown ICMPV6 message type %x from %s to %s",
mldh->mld6_type, inet6_fmt(&src->sin6_addr),
inet6_fmt(dst));
return;
}
}
static void
make_mld6_msg(type, code, src, dst, group, ifindex, delay, datalen, alert)
int type, code, ifindex, delay, datalen, alert;
struct sockaddr_in6 *src, *dst;
struct in6_addr *group;
{
static struct sockaddr_in6 dst_sa = {sizeof(dst_sa), AF_INET6};
struct mld6_hdr *mhp = (struct mld6_hdr *)mld6_send_buf;
int ctllen, hbhlen = 0;
switch(type) {
case MLD6_MTRACE:
case MLD6_MTRACE_RESP:
sndmh.msg_name = (caddr_t)dst;
break;
default:
if (IN6_IS_ADDR_UNSPECIFIED(group))
dst_sa.sin6_addr = allnodes_group.sin6_addr;
else
dst_sa.sin6_addr = *group;
sndmh.msg_name = (caddr_t)&dst_sa;
datalen = sizeof(struct mld6_hdr);
break;
}
bzero(mhp, sizeof(*mhp));
mhp->mld6_type = type;
mhp->mld6_code = code;
mhp->mld6_maxdelay = htons(delay);
mhp->mld6_addr = *group;
sndiov[0].iov_len = datalen;
/* estimate total ancillary data length */
ctllen = 0;
if (ifindex != -1 || src)
ctllen += CMSG_SPACE(sizeof(struct in6_pktinfo));
if (alert) {
#ifdef USE_RFC2292BIS
if ((hbhlen = inet6_opt_init(NULL, 0)) == -1)
log(LOG_ERR, 0, "inet6_opt_init(0) failed");
if ((hbhlen = inet6_opt_append(NULL, 0, hbhlen, IP6OPT_ROUTER_ALERT, 2,
2, NULL)) == -1)
log(LOG_ERR, 0, "inet6_opt_append(0) failed");
if ((hbhlen = inet6_opt_finish(NULL, 0, hbhlen)) == -1)
log(LOG_ERR, 0, "inet6_opt_finish(0) failed");
ctllen += CMSG_SPACE(hbhlen);
#else /* old advanced API */
hbhlen = inet6_option_space(sizeof(raopt));
ctllen += hbhlen;
#endif
}
/* extend ancillary data space (if necessary) */
if (ctlbuflen < ctllen) {
if (sndcmsgbuf)
free(sndcmsgbuf);
if ((sndcmsgbuf = malloc(ctllen)) == NULL)
log(LOG_ERR, 0, "make_mld6_msg: malloc failed"); /* assert */
ctlbuflen = ctllen;
}
/* store ancillary data */
if ((sndmh.msg_controllen = ctllen) > 0) {
struct cmsghdr *cmsgp;
sndmh.msg_control = sndcmsgbuf;
cmsgp = CMSG_FIRSTHDR(&sndmh);
if (ifindex != -1 || src) {
struct in6_pktinfo *pktinfo;
cmsgp->cmsg_len = CMSG_SPACE(sizeof(struct in6_pktinfo));
cmsgp->cmsg_level = IPPROTO_IPV6;
cmsgp->cmsg_type = IPV6_PKTINFO;
pktinfo = (struct in6_pktinfo *)CMSG_DATA(cmsgp);
memset((caddr_t)pktinfo, 0, sizeof(*pktinfo));
if (ifindex != -1)
pktinfo->ipi6_ifindex = ifindex;
if (src)
pktinfo->ipi6_addr = src->sin6_addr;
cmsgp = CMSG_NXTHDR(&sndmh, cmsgp);
}
if (alert) {
#ifdef USE_RFC2292BIS
int currentlen;
void *hbhbuf, *optp = NULL;
cmsgp->cmsg_len = CMSG_SPACE(hbhlen);
cmsgp->cmsg_level = IPPROTO_IPV6;
cmsgp->cmsg_type = IPV6_HOPOPTS;
hbhbuf = CMSG_DATA(cmsgp);
if ((currentlen = inet6_opt_init(hbhbuf, hbhlen)) == -1)
log(LOG_ERR, 0, "inet6_opt_init(len = %d) failed",
hbhlen);
if ((currentlen = inet6_opt_append(hbhbuf, hbhlen,
currentlen,
IP6OPT_ROUTER_ALERT, 2,
2, &optp)) == -1)
log(LOG_ERR, 0,
"inet6_opt_append(len = %d) failed",
currentlen, hbhlen);
(void)inet6_opt_set_val(optp, 0, &rtalert_code,
sizeof(rtalert_code));
if (inet6_opt_finish(hbhbuf, hbhlen, currentlen) == -1)
log(LOG_ERR, 0, "inet6_opt_finish(buf) failed");
#else /* old advanced API */
if (inet6_option_init((void *)cmsgp, &cmsgp, IPV6_HOPOPTS))
log(LOG_ERR, 0, /* assert */
"make_mld6_msg: inet6_option_init failed");
if (inet6_option_append(cmsgp, raopt, 4, 0))
log(LOG_ERR, 0, /* assert */
"make_mld6_msg: inet6_option_append failed");
#endif
cmsgp = CMSG_NXTHDR(&sndmh, cmsgp);
}
}
else
sndmh.msg_control = NULL; /* clear for safety */
}
void
send_mld6(type, code, src, dst, group, index, delay, datalen, alert)
int type;
int code; /* for trace packets only */
struct sockaddr_in6 *src;
struct sockaddr_in6 *dst; /* may be NULL */
struct in6_addr *group;
int index, delay, alert;
int datalen; /* for trace packets only */
{
int setloop = 0;
struct sockaddr_in6 *dstp;
make_mld6_msg(type, code, src, dst, group, index, delay, datalen, alert);
dstp = (struct sockaddr_in6 *)sndmh.msg_name;
if (IN6_ARE_ADDR_EQUAL(&dstp->sin6_addr, &allnodes_group.sin6_addr)) {
setloop = 1;
k_set_loop(mld6_socket, TRUE);
}
if (sendmsg(mld6_socket, &sndmh, 0) < 0) {
if (errno == ENETDOWN)
check_vif_state();
else
log(log_level(IPPROTO_ICMPV6, type, 0), errno,
"sendmsg to %s with src %s on %s",
inet6_fmt(&dstp->sin6_addr),
src ? inet6_fmt(&src->sin6_addr) : "(unspec)",
ifindex2str(index));
if (setloop)
k_set_loop(mld6_socket, FALSE);
return;
}
IF_DEBUG(DEBUG_PKT|debug_kind(IPPROTO_IGMP, type, 0))
log(LOG_DEBUG, 0, "SENT %s from %-15s to %s",
packet_kind(IPPROTO_ICMPV6, type, 0),
src ? inet6_fmt(&src->sin6_addr) : "unspec",
inet6_fmt(&dstp->sin6_addr));
}