freebsd-skq/sys/net/if_stf.c
arr d502ffe4f0 - malloc should be passed M_WAITOK, not M_WAIT (a mbuf flag)
- make use of M_ZERO to remove a call to bzero()
2001-12-07 01:32:40 +00:00

759 lines
19 KiB
C

/* $FreeBSD$ */
/* $KAME: if_stf.c,v 1.62 2001/06/07 22:32:16 itojun Exp $ */
/*
* Copyright (C) 2000 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.
*/
/*
* 6to4 interface, based on RFC3056.
*
* 6to4 interface is NOT capable of link-layer (I mean, IPv4) multicasting.
* There is no address mapping defined from IPv6 multicast address to IPv4
* address. Therefore, we do not have IFF_MULTICAST on the interface.
*
* Due to the lack of address mapping for link-local addresses, we cannot
* throw packets toward link-local addresses (fe80::x). Also, we cannot throw
* packets to link-local multicast addresses (ff02::x).
*
* Here are interesting symptoms due to the lack of link-local address:
*
* Unicast routing exchange:
* - RIPng: Impossible. Uses link-local multicast packet toward ff02::9,
* and link-local addresses as nexthop.
* - OSPFv6: Impossible. OSPFv6 assumes that there's link-local address
* assigned to the link, and makes use of them. Also, HELLO packets use
* link-local multicast addresses (ff02::5 and ff02::6).
* - BGP4+: Maybe. You can only use global address as nexthop, and global
* address as TCP endpoint address.
*
* Multicast routing protocols:
* - PIM: Hello packet cannot be used to discover adjacent PIM routers.
* Adjacent PIM routers must be configured manually (is it really spec-wise
* correct thing to do?).
*
* ICMPv6:
* - Redirects cannot be used due to the lack of link-local address.
*
* stf interface does not have, and will not need, a link-local address.
* It seems to have no real benefit and does not help the above symptoms much.
* Even if we assign link-locals to interface, we cannot really
* use link-local unicast/multicast on top of 6to4 cloud (since there's no
* encapsulation defined for link-local address), and the above analysis does
* not change. RFC3056 does not mandate the assignment of link-local address
* either.
*
* 6to4 interface has security issues. Refer to
* http://playground.iijlab.net/i-d/draft-itojun-ipv6-transition-abuse-00.txt
* for details. The code tries to filter out some of malicious packets.
* Note that there is no way to be 100% secure.
*/
#include "opt_inet.h"
#include "opt_inet6.h"
#include <sys/param.h>
#include <sys/systm.h>
#include <sys/socket.h>
#include <sys/sockio.h>
#include <sys/mbuf.h>
#include <sys/errno.h>
#include <sys/kernel.h>
#include <sys/protosw.h>
#include <sys/queue.h>
#include <machine/bus.h> /* XXX: Shouldn't really be required! */
#include <sys/rman.h>
#include <machine/cpu.h>
#include <sys/malloc.h>
#include <net/if.h>
#include <net/route.h>
#include <net/netisr.h>
#include <net/if_types.h>
#include <net/if_stf.h>
#include <netinet/in.h>
#include <netinet/in_systm.h>
#include <netinet/ip.h>
#include <netinet/ip_var.h>
#include <netinet/in_var.h>
#include <netinet/ip6.h>
#include <netinet6/ip6_var.h>
#include <netinet6/in6_var.h>
#include <netinet/ip_ecn.h>
#include <netinet/ip_encap.h>
#include <machine/stdarg.h>
#include <net/net_osdep.h>
#include <net/bpf.h>
#define STFNAME "stf"
#define STF_MAXUNIT 0 /* only one is currently allowed */
#define IN6_IS_ADDR_6TO4(x) (ntohs((x)->s6_addr16[0]) == 0x2002)
#define GET_V4(x) ((struct in_addr *)(&(x)->s6_addr16[1]))
struct stf_softc {
struct ifnet sc_if; /* common area */
union {
struct route __sc_ro4;
struct route_in6 __sc_ro6; /* just for safety */
} __sc_ro46;
#define sc_ro __sc_ro46.__sc_ro4
const struct encaptab *encap_cookie;
struct resource *r_unit; /* resource allocated for this unit */
LIST_ENTRY(stf_softc) sc_list; /* all stf's are linked */
};
LIST_HEAD(, stf_softc) stf_softc_list;
static MALLOC_DEFINE(M_STF, STFNAME, "6to4 Tunnel Interface");
static struct rman stfunits[1];
static int ip_stf_ttl = 40;
extern struct domain inetdomain;
struct protosw in_stf_protosw =
{ SOCK_RAW, &inetdomain, IPPROTO_IPV6, PR_ATOMIC|PR_ADDR,
in_stf_input, rip_output, 0, rip_ctloutput,
0,
0, 0, 0, 0,
&rip_usrreqs
};
static int stfmodevent __P((module_t, int, void *));
static int stf_encapcheck __P((const struct mbuf *, int, int, void *));
static struct in6_ifaddr *stf_getsrcifa6 __P((struct ifnet *));
static int stf_output __P((struct ifnet *, struct mbuf *, struct sockaddr *,
struct rtentry *));
static int stf_checkaddr4 __P((struct stf_softc *, struct in_addr *,
struct ifnet *));
static int stf_checkaddr6 __P((struct stf_softc *, struct in6_addr *,
struct ifnet *));
static void stf_rtrequest __P((int, struct rtentry *, struct rt_addrinfo *));
static int stf_ioctl __P((struct ifnet *, u_long, caddr_t));
int stf_clone_create __P((struct if_clone *, int *));
void stf_clone_destroy __P((struct ifnet *));
struct if_clone stf_cloner =
IF_CLONE_INITIALIZER(STFNAME, stf_clone_create, stf_clone_destroy);
int
stf_clone_create(ifc, unit)
struct if_clone *ifc;
int *unit;
{
struct resource *r;
struct stf_softc *sc;
if (*unit > STF_MAXUNIT)
return (ENXIO);
if (*unit < 0) {
r = rman_reserve_resource(stfunits, 0, STF_MAXUNIT, 1,
RF_ALLOCATED | RF_ACTIVE, NULL);
if (r == NULL)
return (ENOSPC);
*unit = rman_get_start(r);
} else {
r = rman_reserve_resource(stfunits, *unit, *unit, 1,
RF_ALLOCATED | RF_ACTIVE, NULL);
if (r == NULL)
return (EEXIST);
}
sc = malloc(sizeof(struct stf_softc), M_STF, M_WAITOK | M_ZERO);
sc->sc_if.if_name = STFNAME;
sc->sc_if.if_unit = *unit;
sc->r_unit = r;
sc->encap_cookie = encap_attach_func(AF_INET, IPPROTO_IPV6,
stf_encapcheck, &in_stf_protosw, sc);
if (sc->encap_cookie == NULL) {
printf("%s: attach failed\n", if_name(&sc->sc_if));
free(sc, M_STF);
return (ENOMEM);
}
sc->sc_if.if_mtu = IPV6_MMTU;
sc->sc_if.if_ioctl = stf_ioctl;
sc->sc_if.if_output = stf_output;
sc->sc_if.if_type = IFT_STF;
sc->sc_if.if_snd.ifq_maxlen = IFQ_MAXLEN;
if_attach(&sc->sc_if);
bpfattach(&sc->sc_if, DLT_NULL, sizeof(u_int));
LIST_INSERT_HEAD(&stf_softc_list, sc, sc_list);
return (0);
}
void
stf_clone_destroy(ifp)
struct ifnet *ifp;
{
int err;
struct stf_softc *sc = (void *) ifp;
LIST_REMOVE(sc, sc_list);
err = encap_detach(sc->encap_cookie);
KASSERT(err == 0, ("Unexpected error detaching encap_cookie"));
bpfdetach(ifp);
if_detach(ifp);
err = rman_release_resource(sc->r_unit);
KASSERT(err == 0, ("Unexpected error freeing resource"));
free(sc, M_STF);
}
static int
stfmodevent(mod, type, data)
module_t mod;
int type;
void *data;
{
int err;
switch (type) {
case MOD_LOAD:
stfunits->rm_type = RMAN_ARRAY;
stfunits->rm_descr = "configurable if_stf units";
err = rman_init(stfunits);
if (err != 0)
return (err);
err = rman_manage_region(stfunits, 0, STF_MAXUNIT);
if (err != 0) {
printf("%s: stfunits: rman_manage_region: Failed %d\n",
STFNAME, err);
rman_fini(stfunits);
return (err);
}
LIST_INIT(&stf_softc_list);
if_clone_attach(&stf_cloner);
break;
case MOD_UNLOAD:
if_clone_detach(&stf_cloner);
while (!LIST_EMPTY(&stf_softc_list))
stf_clone_destroy(&LIST_FIRST(&stf_softc_list)->sc_if);
err = rman_fini(stfunits);
KASSERT(err == 0, ("Unexpected error freeing resource"));
break;
}
return (0);
}
static moduledata_t stf_mod = {
"if_stf",
stfmodevent,
0
};
DECLARE_MODULE(if_stf, stf_mod, SI_SUB_PSEUDO, SI_ORDER_ANY);
static int
stf_encapcheck(m, off, proto, arg)
const struct mbuf *m;
int off;
int proto;
void *arg;
{
struct ip ip;
struct in6_ifaddr *ia6;
struct stf_softc *sc;
struct in_addr a, b;
sc = (struct stf_softc *)arg;
if (sc == NULL)
return 0;
if ((sc->sc_if.if_flags & IFF_UP) == 0)
return 0;
/* IFF_LINK0 means "no decapsulation" */
if ((sc->sc_if.if_flags & IFF_LINK0) != 0)
return 0;
if (proto != IPPROTO_IPV6)
return 0;
/* LINTED const cast */
m_copydata((struct mbuf *)m, 0, sizeof(ip), (caddr_t)&ip);
if (ip.ip_v != 4)
return 0;
ia6 = stf_getsrcifa6(&sc->sc_if);
if (ia6 == NULL)
return 0;
/*
* check if IPv4 dst matches the IPv4 address derived from the
* local 6to4 address.
* success on: dst = 10.1.1.1, ia6->ia_addr = 2002:0a01:0101:...
*/
if (bcmp(GET_V4(&ia6->ia_addr.sin6_addr), &ip.ip_dst,
sizeof(ip.ip_dst)) != 0)
return 0;
/*
* check if IPv4 src matches the IPv4 address derived from the
* local 6to4 address masked by prefixmask.
* success on: src = 10.1.1.1, ia6->ia_addr = 2002:0a00:.../24
* fail on: src = 10.1.1.1, ia6->ia_addr = 2002:0b00:.../24
*/
bzero(&a, sizeof(a));
a.s_addr = GET_V4(&ia6->ia_addr.sin6_addr)->s_addr;
a.s_addr &= GET_V4(&ia6->ia_prefixmask.sin6_addr)->s_addr;
b = ip.ip_src;
b.s_addr &= GET_V4(&ia6->ia_prefixmask.sin6_addr)->s_addr;
if (a.s_addr != b.s_addr)
return 0;
/* stf interface makes single side match only */
return 32;
}
static struct in6_ifaddr *
stf_getsrcifa6(ifp)
struct ifnet *ifp;
{
struct ifaddr *ia;
struct in_ifaddr *ia4;
struct sockaddr_in6 *sin6;
struct in_addr in;
for (ia = TAILQ_FIRST(&ifp->if_addrlist);
ia;
ia = TAILQ_NEXT(ia, ifa_list))
{
if (ia->ifa_addr == NULL)
continue;
if (ia->ifa_addr->sa_family != AF_INET6)
continue;
sin6 = (struct sockaddr_in6 *)ia->ifa_addr;
if (!IN6_IS_ADDR_6TO4(&sin6->sin6_addr))
continue;
bcopy(GET_V4(&sin6->sin6_addr), &in, sizeof(in));
LIST_FOREACH(ia4, INADDR_HASH(in.s_addr), ia_hash)
if (ia4->ia_addr.sin_addr.s_addr == in.s_addr)
break;
if (ia4 == NULL)
continue;
return (struct in6_ifaddr *)ia;
}
return NULL;
}
static int
stf_output(ifp, m, dst, rt)
struct ifnet *ifp;
struct mbuf *m;
struct sockaddr *dst;
struct rtentry *rt;
{
struct stf_softc *sc;
struct sockaddr_in6 *dst6;
struct in_addr *in4;
struct sockaddr_in *dst4;
u_int8_t tos;
struct ip *ip;
struct ip6_hdr *ip6;
struct in6_ifaddr *ia6;
sc = (struct stf_softc*)ifp;
dst6 = (struct sockaddr_in6 *)dst;
/* just in case */
if ((ifp->if_flags & IFF_UP) == 0) {
m_freem(m);
return ENETDOWN;
}
/*
* If we don't have an ip4 address that match my inner ip6 address,
* we shouldn't generate output. Without this check, we'll end up
* using wrong IPv4 source.
*/
ia6 = stf_getsrcifa6(ifp);
if (ia6 == NULL) {
m_freem(m);
return ENETDOWN;
}
if (m->m_len < sizeof(*ip6)) {
m = m_pullup(m, sizeof(*ip6));
if (!m)
return ENOBUFS;
}
ip6 = mtod(m, struct ip6_hdr *);
tos = (ntohl(ip6->ip6_flow) >> 20) & 0xff;
/*
* Pickup the right outer dst addr from the list of candidates.
* ip6_dst has priority as it may be able to give us shorter IPv4 hops.
*/
if (IN6_IS_ADDR_6TO4(&ip6->ip6_dst))
in4 = GET_V4(&ip6->ip6_dst);
else if (IN6_IS_ADDR_6TO4(&dst6->sin6_addr))
in4 = GET_V4(&dst6->sin6_addr);
else {
m_freem(m);
return ENETUNREACH;
}
#if NBPFILTER > 0
if (ifp->if_bpf) {
/*
* We need to prepend the address family as
* a four byte field. Cons up a dummy header
* to pacify bpf. This is safe because bpf
* will only read from the mbuf (i.e., it won't
* try to free it or keep a pointer a to it).
*/
struct mbuf m0;
u_int32_t af = AF_INET6;
m0.m_next = m;
m0.m_len = 4;
m0.m_data = (char *)&af;
#ifdef HAVE_OLD_BPF
bpf_mtap(ifp, &m0);
#else
bpf_mtap(ifp->if_bpf, &m0);
#endif
}
#endif /*NBPFILTER > 0*/
M_PREPEND(m, sizeof(struct ip), M_DONTWAIT);
if (m && m->m_len < sizeof(struct ip))
m = m_pullup(m, sizeof(struct ip));
if (m == NULL)
return ENOBUFS;
ip = mtod(m, struct ip *);
bzero(ip, sizeof(*ip));
bcopy(GET_V4(&((struct sockaddr_in6 *)&ia6->ia_addr)->sin6_addr),
&ip->ip_src, sizeof(ip->ip_src));
bcopy(in4, &ip->ip_dst, sizeof(ip->ip_dst));
ip->ip_p = IPPROTO_IPV6;
ip->ip_ttl = ip_stf_ttl;
ip->ip_len = m->m_pkthdr.len; /*host order*/
if (ifp->if_flags & IFF_LINK1)
ip_ecn_ingress(ECN_ALLOWED, &ip->ip_tos, &tos);
else
ip_ecn_ingress(ECN_NOCARE, &ip->ip_tos, &tos);
dst4 = (struct sockaddr_in *)&sc->sc_ro.ro_dst;
if (dst4->sin_family != AF_INET ||
bcmp(&dst4->sin_addr, &ip->ip_dst, sizeof(ip->ip_dst)) != 0) {
/* cache route doesn't match */
dst4->sin_family = AF_INET;
dst4->sin_len = sizeof(struct sockaddr_in);
bcopy(&ip->ip_dst, &dst4->sin_addr, sizeof(dst4->sin_addr));
if (sc->sc_ro.ro_rt) {
RTFREE(sc->sc_ro.ro_rt);
sc->sc_ro.ro_rt = NULL;
}
}
if (sc->sc_ro.ro_rt == NULL) {
rtalloc(&sc->sc_ro);
if (sc->sc_ro.ro_rt == NULL) {
m_freem(m);
return ENETUNREACH;
}
}
return ip_output(m, NULL, &sc->sc_ro, 0, NULL);
}
static int
stf_checkaddr4(sc, in, inifp)
struct stf_softc *sc;
struct in_addr *in;
struct ifnet *inifp; /* incoming interface */
{
struct in_ifaddr *ia4;
/*
* reject packets with the following address:
* 224.0.0.0/4 0.0.0.0/8 127.0.0.0/8 255.0.0.0/8
*/
if (IN_MULTICAST(ntohl(in->s_addr)))
return -1;
switch ((ntohl(in->s_addr) & 0xff000000) >> 24) {
case 0: case 127: case 255:
return -1;
}
/*
* reject packets with broadcast
*/
for (ia4 = TAILQ_FIRST(&in_ifaddrhead);
ia4;
ia4 = TAILQ_NEXT(ia4, ia_link))
{
if ((ia4->ia_ifa.ifa_ifp->if_flags & IFF_BROADCAST) == 0)
continue;
if (in->s_addr == ia4->ia_broadaddr.sin_addr.s_addr)
return -1;
}
/*
* perform ingress filter
*/
if (sc && (sc->sc_if.if_flags & IFF_LINK2) == 0 && inifp) {
struct sockaddr_in sin;
struct rtentry *rt;
bzero(&sin, sizeof(sin));
sin.sin_family = AF_INET;
sin.sin_len = sizeof(struct sockaddr_in);
sin.sin_addr = *in;
rt = rtalloc1((struct sockaddr *)&sin, 0, 0UL);
if (!rt || rt->rt_ifp != inifp) {
#if 0
log(LOG_WARNING, "%s: packet from 0x%x dropped "
"due to ingress filter\n", if_name(&sc->sc_if),
(u_int32_t)ntohl(sin.sin_addr.s_addr));
#endif
if (rt)
rtfree(rt);
return -1;
}
rtfree(rt);
}
return 0;
}
static int
stf_checkaddr6(sc, in6, inifp)
struct stf_softc *sc;
struct in6_addr *in6;
struct ifnet *inifp; /* incoming interface */
{
/*
* check 6to4 addresses
*/
if (IN6_IS_ADDR_6TO4(in6))
return stf_checkaddr4(sc, GET_V4(in6), inifp);
/*
* reject anything that look suspicious. the test is implemented
* in ip6_input too, but we check here as well to
* (1) reject bad packets earlier, and
* (2) to be safe against future ip6_input change.
*/
if (IN6_IS_ADDR_V4COMPAT(in6) || IN6_IS_ADDR_V4MAPPED(in6))
return -1;
return 0;
}
void
in_stf_input(m, off)
struct mbuf *m;
int off;
{
int proto;
struct stf_softc *sc;
struct ip *ip;
struct ip6_hdr *ip6;
u_int8_t otos, itos;
int len, isr;
struct ifqueue *ifq = NULL;
struct ifnet *ifp;
proto = mtod(m, struct ip *)->ip_p;
if (proto != IPPROTO_IPV6) {
m_freem(m);
return;
}
ip = mtod(m, struct ip *);
sc = (struct stf_softc *)encap_getarg(m);
if (sc == NULL || (sc->sc_if.if_flags & IFF_UP) == 0) {
m_freem(m);
return;
}
ifp = &sc->sc_if;
/*
* perform sanity check against outer src/dst.
* for source, perform ingress filter as well.
*/
if (stf_checkaddr4(sc, &ip->ip_dst, NULL) < 0 ||
stf_checkaddr4(sc, &ip->ip_src, m->m_pkthdr.rcvif) < 0) {
m_freem(m);
return;
}
otos = ip->ip_tos;
m_adj(m, off);
if (m->m_len < sizeof(*ip6)) {
m = m_pullup(m, sizeof(*ip6));
if (!m)
return;
}
ip6 = mtod(m, struct ip6_hdr *);
/*
* perform sanity check against inner src/dst.
* for source, perform ingress filter as well.
*/
if (stf_checkaddr6(sc, &ip6->ip6_dst, NULL) < 0 ||
stf_checkaddr6(sc, &ip6->ip6_src, m->m_pkthdr.rcvif) < 0) {
m_freem(m);
return;
}
itos = (ntohl(ip6->ip6_flow) >> 20) & 0xff;
if ((ifp->if_flags & IFF_LINK1) != 0)
ip_ecn_egress(ECN_ALLOWED, &otos, &itos);
else
ip_ecn_egress(ECN_NOCARE, &otos, &itos);
ip6->ip6_flow &= ~htonl(0xff << 20);
ip6->ip6_flow |= htonl((u_int32_t)itos << 20);
m->m_pkthdr.rcvif = ifp;
if (ifp->if_bpf) {
/*
* We need to prepend the address family as
* a four byte field. Cons up a dummy header
* to pacify bpf. This is safe because bpf
* will only read from the mbuf (i.e., it won't
* try to free it or keep a pointer a to it).
*/
struct mbuf m0;
u_int32_t af = AF_INET6;
m0.m_next = m;
m0.m_len = 4;
m0.m_data = (char *)&af;
#ifdef HAVE_OLD_BPF
bpf_mtap(ifp, &m0);
#else
bpf_mtap(ifp->if_bpf, &m0);
#endif
}
/*
* Put the packet to the network layer input queue according to the
* specified address family.
* See net/if_gif.c for possible issues with packet processing
* reorder due to extra queueing.
*/
ifq = &ip6intrq;
isr = NETISR_IPV6;
len = m->m_pkthdr.len;
if (! IF_HANDOFF(ifq, m, NULL))
return;
schednetisr(isr);
ifp->if_ipackets++;
ifp->if_ibytes += len;
}
/* ARGSUSED */
static void
stf_rtrequest(cmd, rt, info)
int cmd;
struct rtentry *rt;
struct rt_addrinfo *info;
{
if (rt)
rt->rt_rmx.rmx_mtu = IPV6_MMTU;
}
static int
stf_ioctl(ifp, cmd, data)
struct ifnet *ifp;
u_long cmd;
caddr_t data;
{
struct ifaddr *ifa;
struct ifreq *ifr;
struct sockaddr_in6 *sin6;
int error;
error = 0;
switch (cmd) {
case SIOCSIFADDR:
ifa = (struct ifaddr *)data;
if (ifa == NULL || ifa->ifa_addr->sa_family != AF_INET6) {
error = EAFNOSUPPORT;
break;
}
sin6 = (struct sockaddr_in6 *)ifa->ifa_addr;
if (IN6_IS_ADDR_6TO4(&sin6->sin6_addr)) {
ifa->ifa_rtrequest = stf_rtrequest;
ifp->if_flags |= IFF_UP;
} else
error = EINVAL;
break;
case SIOCADDMULTI:
case SIOCDELMULTI:
ifr = (struct ifreq *)data;
if (ifr && ifr->ifr_addr.sa_family == AF_INET6)
;
else
error = EAFNOSUPPORT;
break;
default:
error = EINVAL;
break;
}
return error;
}