input.c revision 7c478bd95313f5f23a4c958a745db2134aa03244
/*
* Copyright 2004 Sun Microsystems, Inc. All rights reserved.
* Use is subject to license terms.
*
* Copyright (c) 1983, 1988, 1993
* The Regents of the University of California. 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. All advertising materials mentioning features or use of this software
* must display the following acknowledgment:
* This product includes software developed by the University of
* California, Berkeley and its contributors.
* 4. Neither the name of the University 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 REGENTS 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 REGENTS OR CONTRIBUTORS BE LIABLE
* FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
* SUCH DAMAGE.
*
* $FreeBSD: src/sbin/routed/input.c,v 1.9 2001/06/06 20:52:30 phk Exp $
*/
#pragma ident "%Z%%M% %I% %E% SMI"
#include "defs.h"
#include <md5.h>
/*
* The size of the control buffer passed to recvmsg() used to receive
* ancillary data.
*/
#define CONTROL_BUFSIZE 1024
static void input(struct sockaddr_in *, struct interface *, struct rip *, int);
static boolean_t ck_passwd(struct interface *, struct rip *, uint8_t *,
in_addr_t, struct msg_limit *);
/*
* Find the interface which received the given message.
*/
struct interface *
receiving_interface(struct msghdr *msg, boolean_t findremote)
{
struct interface *ifp, *ifp1, *ifp2;
struct sockaddr_in *from;
void *opt;
uint_t ifindex;
from = (struct sockaddr_in *)msg->msg_name;
/* First see if this packet came from a remote gateway. */
if (findremote && ((ifp = findremoteif(from->sin_addr.s_addr)) != NULL))
return (ifp);
/*
* It did not come from a remote gateway. Determine which
* physical interface this packet was received on by
* processing the message's ancillary data to find the
* IP_RECVIF option we requested.
*/
if ((opt = find_ancillary(msg, IP_RECVIF)) == NULL) {
msglog("unable to retrieve IP_RECVIF");
} else {
ifindex = *(uint_t *)opt;
if ((ifp = ifwithindex(ifindex, _B_TRUE)) != NULL) {
/* Find the best match of the aliases */
ifp2 = NULL;
for (ifp1 = ifp; ifp1 != NULL;
ifp1 = ifp1->int_ilist.hl_next) {
if (ifp1->int_addr == from->sin_addr.s_addr)
return (ifp1);
if ((ifp2 == NULL ||
(ifp2->int_state & IS_ALIAS)) &&
on_net(from->sin_addr.s_addr, ifp1->int_net,
ifp1->int_mask))
ifp2 = ifp1;
}
if (ifp2 != NULL)
ifp = ifp2;
return (ifp);
}
}
/*
* As a last resort (for some reason, ip didn't give us the
* IP_RECVIF index we requested), try to deduce the receiving
* interface based on the source address of the packet.
*/
return (iflookup(from->sin_addr.s_addr));
}
/*
* Process RIP input on rip_sock. Returns 0 for success, -1 for failure.
*/
int
read_rip()
{
struct sockaddr_in from;
struct interface *ifp;
int cc;
union pkt_buf inbuf;
struct msghdr msg;
struct iovec iov;
uint8_t ancillary_data[CONTROL_BUFSIZE];
iov.iov_base = &inbuf;
iov.iov_len = sizeof (inbuf);
msg.msg_iov = &iov;
msg.msg_iovlen = 1;
msg.msg_name = &from;
msg.msg_control = &ancillary_data;
for (;;) {
msg.msg_namelen = sizeof (from);
msg.msg_controllen = sizeof (ancillary_data);
cc = recvmsg(rip_sock, &msg, 0);
if (cc == 0)
return (-1);
if (cc < 0) {
if (errno == EWOULDBLOCK || errno == EINTR)
return (0);
LOGERR("recvmsg(rip_sock)");
return (-1);
}
/*
* ifp is the interface via which the packet arrived.
*/
ifp = receiving_interface(&msg, _B_TRUE);
input(&from, ifp, &inbuf.rip, cc);
}
}
/* Process a RIP packet */
static void
input(struct sockaddr_in *from, /* received from this IP address */
struct interface *ifp, /* interface of incoming socket */
struct rip *rip,
int cc)
{
#define FROM_NADDR from->sin_addr.s_addr
static struct msg_limit use_auth, bad_len, bad_mask;
static struct msg_limit unk_router, bad_router, bad_nhop;
struct rt_entry *rt;
struct rt_spare new;
struct netinfo *n, *lim;
struct interface *ifp1;
in_addr_t gate, mask, v1_mask, dst, ddst_h = 0;
struct auth *ap;
struct tgate *tg = NULL;
struct tgate_net *tn;
int i, j;
boolean_t poll_answer = _B_FALSE; /* Set to _B_TRUE if RIPCMD_POLL */
uint16_t rt_state = 0; /* Extra route state to pass to input_route() */
uint8_t metric;
(void) memset(&new, 0, sizeof (new));
/* Notice when we hear from a remote gateway */
if (ifp != NULL && (ifp->int_state & IS_REMOTE))
ifp->int_act_time = now.tv_sec;
trace_rip("Recv", "from", from, ifp, rip, cc);
if (ifp != NULL && (ifp->int_if_flags & IFF_NORTEXCH)) {
trace_misc("discard RIP packet received over %s (IFF_NORTEXCH)",
ifp->int_name);
return;
}
gate = ntohl(FROM_NADDR);
if (IN_EXPERIMENTAL(gate) || (gate >> IN_CLASSA_NSHIFT) == 0) {
msglim(&bad_router, FROM_NADDR, "source address %s unusable",
naddr_ntoa(FROM_NADDR));
return;
}
if (rip->rip_vers == 0) {
msglim(&bad_router, FROM_NADDR,
"RIP version 0, cmd %d, packet received from %s",
rip->rip_cmd, naddr_ntoa(FROM_NADDR));
return;
}
if (rip->rip_vers > RIPv2) {
msglim(&bad_router, FROM_NADDR,
"Treating RIP version %d packet received from %s as "
"version %d", rip->rip_vers, naddr_ntoa(FROM_NADDR),
RIPv2);
rip->rip_vers = RIPv2;
}
if (cc > (int)OVER_MAXPACKETSIZE) {
msglim(&bad_router, FROM_NADDR,
"packet at least %d bytes too long received from %s",
cc-MAXPACKETSIZE, naddr_ntoa(FROM_NADDR));
}
n = rip->rip_nets;
lim = n + (cc - 4) / sizeof (struct netinfo);
/*
* Notice authentication.
* As required by section 5.2 of RFC 2453, discard authenticated
* RIPv2 messages, but only if configured for that silliness.
*
* RIPv2 authentication is lame. Why authenticate queries?
* Why should a RIPv2 implementation with authentication disabled
* not be able to listen to RIPv2 packets with authentication, while
* RIPv1 systems will listen? Crazy!
*/
if (!auth_ok && rip->rip_vers == RIPv2 && n < lim &&
n->n_family == RIP_AF_AUTH) {
msglim(&use_auth, FROM_NADDR,
"RIPv2 message with authentication from %s discarded",
naddr_ntoa(FROM_NADDR));
return;
}
switch (rip->rip_cmd) {
case RIPCMD_POLL:
/*
* Similar to RIPCMD_REQUEST, this command is used to
* request either a full-table or a set of entries. Both
* silent processes and routers can respond to this
* command.
*/
poll_answer = _B_TRUE;
/* FALLTHRU */
case RIPCMD_REQUEST:
/* Are we talking to ourself or a remote gateway? */
ifp1 = ifwithaddr(FROM_NADDR, _B_FALSE, _B_TRUE);
if (ifp1 != NULL) {
if (ifp1->int_state & IS_REMOTE) {
/* remote gateway */
ifp = ifp1;
if (check_remote(ifp)) {
ifp->int_act_time = now.tv_sec;
if_ok(ifp, "remote ", _B_FALSE);
}
} else if (from->sin_port == htons(RIP_PORT)) {
trace_pkt(" discard our own RIP request");
return;
}
}
/* did the request come from a router? */
if (!poll_answer && (from->sin_port == htons(RIP_PORT))) {
/*
* yes, ignore the request if RIP is off so that
* the router does not depend on us.
*/
if (ripout_interfaces == 0 ||
(ifp != NULL && (IS_RIP_OUT_OFF(ifp->int_state) ||
!IS_IFF_ROUTING(ifp->int_if_flags)))) {
trace_pkt(" discard request while RIP off");
return;
}
}
/*
* According to RFC 2453 section 5.2, we should ignore
* unauthenticated queries when authentication is
* configured. That is too silly to bother with. Sheesh!
* Are forwarding tables supposed to be secret even though
* a bad guy can infer them with test traffic? RIP is
* still the most common router-discovery protocol, so
* hosts need to send queries that will be answered. What
* about `rtquery`? Maybe on firewalls you'd care, but not
* enough to give up the diagnostic facilities of remote
* probing.
*/
if (n >= lim) {
msglim(&bad_len, FROM_NADDR, "empty request from %s",
naddr_ntoa(FROM_NADDR));
return;
}
if (cc%sizeof (*n) != sizeof (struct rip)%sizeof (*n)) {
msglim(&bad_len, FROM_NADDR,
"request of bad length (%d) from %s",
cc, naddr_ntoa(FROM_NADDR));
}
if (rip->rip_vers == RIPv2 && (ifp == NULL ||
(ifp->int_state & IS_NO_RIPV1_OUT))) {
v12buf.buf->rip_vers = RIPv2;
/*
* If we have a secret but it is a cleartext secret,
* do not disclose our secret unless the other guy
* already knows it.
*/
ap = find_auth(ifp);
if (ap != NULL &&
(ulong_t)ap->end < (ulong_t)clk.tv_sec) {
/*
* Don't authenticate incoming packets
* using an expired key.
*/
msglim(&use_auth, FROM_NADDR,
"%s attempting to authenticate using "
"an expired password.",
naddr_ntoa(FROM_NADDR));
ap = NULL;
}
if (ap != NULL && ap->type == RIP_AUTH_PW &&
(n->n_family != RIP_AF_AUTH ||
!ck_passwd(ifp, rip, (uint8_t *)lim, FROM_NADDR,
&use_auth)))
ap = NULL;
} else {
v12buf.buf->rip_vers = RIPv1;
ap = NULL;
}
clr_ws_buf(&v12buf, ap);
do {
n->n_metric = ntohl(n->n_metric);
/*
* A single entry with family RIP_AF_UNSPEC and
* metric HOPCNT_INFINITY means "all routes".
* We respond to routers only if we are acting
* as a supplier, or to anyone other than a router
* (i.e. a query).
*/
if (n->n_family == RIP_AF_UNSPEC &&
n->n_metric == HOPCNT_INFINITY) {
/*
* Answer a full-table query from a utility
* program with all we know.
*/
if (poll_answer ||
(from->sin_port != htons(RIP_PORT))) {
supply(from, ifp, OUT_QUERY, 0,
rip->rip_vers, ap != NULL);
return;
}
/*
* A router is trying to prime its tables.
* Filter the answer in the same way
* broadcasts are filtered.
*
* Only answer a router if we are a supplier
* to keep an unwary host that is just starting
* from picking us as a router.
*/
if (ifp == NULL) {
trace_pkt("ignore distant router");
return;
}
if (IS_RIP_OFF(ifp->int_state) ||
!should_supply(ifp)) {
trace_pkt("ignore; not supplying");
return;
}
/*
* Do not answer a RIPv1 router if
* we are sending RIPv2. But do offer
* poor man's router discovery.
*/
if ((ifp->int_state & IS_NO_RIPV1_OUT) &&
rip->rip_vers == RIPv1) {
if (!(ifp->int_state & IS_PM_RDISC)) {
trace_pkt("ignore; sending RIPv2");
return;
}
v12buf.n->n_family = RIP_AF_INET;
v12buf.n->n_dst = RIP_DEFAULT;
metric = ifp->int_d_metric;
if (NULL !=
(rt = rtget(RIP_DEFAULT, 0)))
metric = MIN(metric,
(rt->rt_metric + 1));
v12buf.n->n_metric = htonl(metric);
v12buf.n++;
break;
}
/*
* Respond with RIPv1 instead of RIPv2 if
* that is what we are broadcasting on the
* interface to keep the remote router from
* getting the wrong initial idea of the
* routes we send.
*/
supply(from, ifp, OUT_UNICAST, 0,
(ifp->int_state & IS_NO_RIPV1_OUT)
? RIPv2 : RIPv1,
ap != NULL);
return;
}
/* Ignore authentication */
if (n->n_family == RIP_AF_AUTH)
continue;
if (n->n_family != RIP_AF_INET) {
msglim(&bad_router, FROM_NADDR,
"request from %s for unsupported"
" (af %d) %s",
naddr_ntoa(FROM_NADDR),
ntohs(n->n_family),
naddr_ntoa(n->n_dst));
return;
}
/* We are being asked about a specific destination. */
v12buf.n->n_dst = dst = n->n_dst;
v12buf.n->n_family = RIP_AF_INET;
if (!check_dst(dst)) {
msglim(&bad_router, FROM_NADDR,
"bad queried destination %s from %s",
naddr_ntoa(dst),
naddr_ntoa(FROM_NADDR));
v12buf.n->n_metric = HOPCNT_INFINITY;
goto rte_done;
}
/* decide what mask was intended */
if (rip->rip_vers == RIPv1 ||
0 == (mask = ntohl(n->n_mask)) ||
0 != (ntohl(dst) & ~mask))
mask = ripv1_mask_host(dst, ifp);
/*
* Try to find the answer. If we don't have an
* explicit route for the destination, use the best
* route to the destination.
*/
rt = rtget(dst, mask);
if (rt == NULL && dst != RIP_DEFAULT)
rt = rtfind(n->n_dst);
if (v12buf.buf->rip_vers != RIPv1)
v12buf.n->n_mask = htonl(mask);
if (rt == NULL) {
/* we do not have the answer */
v12buf.n->n_metric = HOPCNT_INFINITY;
goto rte_done;
}
/*
* we have the answer, so compute the right metric
* and next hop.
*/
v12buf.n->n_metric = rt->rt_metric + 1;
if (v12buf.n->n_metric > HOPCNT_INFINITY)
v12buf.n->n_metric = HOPCNT_INFINITY;
if (v12buf.buf->rip_vers != RIPv1) {
v12buf.n->n_tag = rt->rt_tag;
if (ifp != NULL &&
on_net(rt->rt_gate, ifp->int_net,
ifp->int_mask) &&
rt->rt_gate != ifp->int_addr)
v12buf.n->n_nhop = rt->rt_gate;
}
rte_done:
v12buf.n->n_metric = htonl(v12buf.n->n_metric);
/*
* Stop paying attention if we fill the output buffer.
*/
if (++v12buf.n >= v12buf.lim)
break;
} while (++n < lim);
/*
* If our response is authenticated with md5, complete the
* md5 computation.
*/
if (ap != NULL && ap->type == RIP_AUTH_MD5)
end_md5_auth(&v12buf, ap);
/*
* Diagnostic programs make specific requests
* from ports other than 520. Log other types
* of specific requests as suspicious.
*/
if (!poll_answer && (from->sin_port == htons(RIP_PORT))) {
writelog(LOG_WARNING,
"Received suspicious request from %s port %d",
naddr_ntoa(FROM_NADDR), RIP_PORT);
}
if (poll_answer || (from->sin_port != htons(RIP_PORT))) {
/* query */
(void) output(OUT_QUERY, from, ifp, v12buf.buf,
((char *)v12buf.n - (char *)v12buf.buf));
} else {
(void) output(OUT_UNICAST, from, ifp,
v12buf.buf, ((char *)v12buf.n -
(char *)v12buf.buf));
}
return;
case RIPCMD_TRACEON:
case RIPCMD_TRACEOFF:
/*
* Notice that trace messages are turned off for all possible
* abuse if PATH_TRACE is undefined in pathnames.h.
* Notice also that because of the way the trace file is
* handled in trace.c, no abuse is plausible even if
* PATH_TRACE is defined.
*
* First verify message came from a privileged port.
*/
if (ntohs(from->sin_port) > IPPORT_RESERVED) {
trace_pkt("trace command from untrusted port %d on %s",
ntohs(from->sin_port), naddr_ntoa(FROM_NADDR));
return;
}
if (ifp == NULL || !remote_address_ok(ifp, FROM_NADDR)) {
/*
* Use a message here to warn about strange
* messages from remote systems.
*/
msglim(&bad_router, FROM_NADDR,
"trace command from non-local host %s",
naddr_ntoa(FROM_NADDR));
return;
}
if (ifp->int_state & IS_DISTRUST) {
tg = tgates;
while (tg->tgate_addr != FROM_NADDR) {
tg = tg->tgate_next;
if (tg == NULL) {
trace_pkt("trace command from "
"untrusted host %s",
naddr_ntoa(FROM_NADDR));
return;
}
}
}
if (ifp->int_auth[0].type != RIP_AUTH_NONE) {
/*
* Technically, it would be fairly easy to add
* standard authentication to the existing
* trace commands -- just bracket the payload
* with the authentication information.
* However, the tracing message behavior
* itself is marginal enough that we don't
* actually care. Just discard if
* authentication is needed.
*/
trace_pkt("trace command unauthenticated from %s",
naddr_ntoa(FROM_NADDR));
return;
}
if (rip->rip_cmd == RIPCMD_TRACEON) {
rip->rip_tracefile[cc-4] = '\0';
set_tracefile(rip->rip_tracefile,
"trace command: %s\n", 0);
} else {
trace_off("tracing turned off by %s",
naddr_ntoa(FROM_NADDR));
}
return;
case RIPCMD_RESPONSE:
if (ifp != NULL && (ifp->int_if_flags & IFF_NOXMIT)) {
trace_misc("discard RIP response received over %s "
"(IFF_NOXMIT)", ifp->int_name);
return;
}
if (cc%sizeof (*n) != sizeof (struct rip)%sizeof (*n)) {
msglim(&bad_len, FROM_NADDR,
"response of bad length (%d) from %s",
cc, naddr_ntoa(FROM_NADDR));
}
if ((ntohl(FROM_NADDR) >> IN_CLASSA_NSHIFT) == IN_LOOPBACKNET) {
msglim(&bad_router, FROM_NADDR,
"discard RIP response from bad source address %s",
naddr_ntoa(FROM_NADDR));
return;
}
/* verify message came from a router */
if (from->sin_port != htons(RIP_PORT)) {
msglim(&bad_router, FROM_NADDR,
" discard RIP response from unknown port"
" %d on host %s", ntohs(from->sin_port),
naddr_ntoa(FROM_NADDR));
return;
}
if (!rip_enabled) {
trace_pkt(" discard response while RIP off");
return;
}
/* Are we talking to ourself or a remote gateway? */
ifp1 = ifwithaddr(FROM_NADDR, _B_FALSE, _B_TRUE);
if (ifp1 != NULL) {
if (ifp1->int_state & IS_REMOTE) {
/* remote gateway */
ifp = ifp1;
if (check_remote(ifp)) {
ifp->int_act_time = now.tv_sec;
if_ok(ifp, "remote ", _B_FALSE);
}
} else {
trace_pkt(" discard our own RIP response");
return;
}
} else {
/*
* If it's not a remote gateway, then the
* remote address *must* be directly
* connected. Make sure that it is.
*/
if (ifp != NULL &&
!remote_address_ok(ifp, FROM_NADDR)) {
msglim(&bad_router, FROM_NADDR,
"discard RIP response; source %s not on "
"interface %s", naddr_ntoa(FROM_NADDR),
ifp->int_name);
return;
}
}
/*
* Accept routing packets from routers directly connected
* via broadcast or point-to-point networks, and from
* those listed in /etc/gateways.
*/
if (ifp == NULL) {
msglim(&unk_router, FROM_NADDR,
" discard response from %s"
" via unexpected interface",
naddr_ntoa(FROM_NADDR));
return;
}
if (IS_RIP_IN_OFF(ifp->int_state)) {
trace_pkt(" discard RIPv%d response"
" via disabled interface %s",
rip->rip_vers, ifp->int_name);
return;
}
if (n >= lim) {
msglim(&bad_len, FROM_NADDR, "empty response from %s",
naddr_ntoa(FROM_NADDR));
return;
}
if (((ifp->int_state & IS_NO_RIPV1_IN) &&
rip->rip_vers == RIPv1) ||
((ifp->int_state & IS_NO_RIPV2_IN) &&
rip->rip_vers != RIPv1)) {
trace_pkt(" discard RIPv%d response",
rip->rip_vers);
return;
}
/*
* Continue to listen to routes via broken interfaces
* which might be declared IS_BROKE because of
* device-driver idiosyncracies, but might otherwise
* be perfectly healthy.
*/
if (ifp->int_state & IS_BROKE) {
trace_pkt("response via broken interface %s",
ifp->int_name);
}
/*
* If the interface cares, ignore bad routers.
* Trace but do not log this problem, because where it
* happens, it happens frequently.
*/
if (ifp->int_state & IS_DISTRUST) {
tg = tgates;
while (tg->tgate_addr != FROM_NADDR) {
tg = tg->tgate_next;
if (tg == NULL) {
trace_pkt(" discard RIP response"
" from untrusted router %s",
naddr_ntoa(FROM_NADDR));
return;
}
}
}
/*
* Authenticate the packet if we have a secret.
* If we do not have any secrets, ignore the error in
* RFC 1723 and accept it regardless.
*/
if (ifp->int_auth[0].type != RIP_AUTH_NONE &&
rip->rip_vers != RIPv1 &&
!ck_passwd(ifp, rip, (uint8_t *)lim, FROM_NADDR, &use_auth))
return;
/*
* Do this only if we're supplying routes to *nobody*.
*/
if (!should_supply(NULL) && save_space) {
/*
* "-S" option. Instead of entering all routes,
* only enter a default route for the sender of
* this RESPONSE message
*/
/* Should we trust this route from this router? */
if (tg != NULL && tg->tgate_nets->mask != 0) {
trace_pkt(" ignored unauthorized %s",
addrname(RIP_DEFAULT, 0, 0));
break;
}
new.rts_gate = FROM_NADDR;
new.rts_router = FROM_NADDR;
new.rts_metric = HOPCNT_INFINITY-1;
new.rts_tag = n->n_tag;
new.rts_time = now.tv_sec;
new.rts_ifp = ifp;
new.rts_de_ag = 0;
new.rts_origin = RO_RIP;
/*
* Add the newly generated default route, but don't
* propagate the madness. Treat it the same way as
* default routes learned from Router Discovery.
*/
input_route(RIP_DEFAULT, 0, &new, n, RS_NOPROPAGATE);
return;
}
if (!IS_IFF_ROUTING(ifp->int_if_flags)) {
/*
* We don't want to propagate routes which would
* result in a black-hole.
*/
rt_state = RS_NOPROPAGATE;
}
do {
if (n->n_family == RIP_AF_AUTH)
continue;
n->n_metric = ntohl(n->n_metric);
dst = n->n_dst;
if (n->n_family != RIP_AF_INET &&
(n->n_family != RIP_AF_UNSPEC ||
dst != RIP_DEFAULT)) {
msglim(&bad_router, FROM_NADDR,
"route from %s to unsupported"
" address family=%d destination=%s",
naddr_ntoa(FROM_NADDR), n->n_family,
naddr_ntoa(dst));
continue;
}
if (!check_dst(dst)) {
msglim(&bad_router, FROM_NADDR,
"bad destination %s from %s",
naddr_ntoa(dst),
naddr_ntoa(FROM_NADDR));
continue;
}
if (n->n_metric == 0 || n->n_metric > HOPCNT_INFINITY) {
msglim(&bad_router, FROM_NADDR,
"bad metric %d from %s"
" for destination %s",
n->n_metric, naddr_ntoa(FROM_NADDR),
naddr_ntoa(dst));
continue;
}
/*
* Notice the next-hop.
*/
gate = FROM_NADDR;
if (n->n_nhop != 0) {
if (rip->rip_vers == RIPv1) {
n->n_nhop = 0;
} else {
/* Use it only if it is valid. */
if (on_net(n->n_nhop,
ifp->int_net, ifp->int_mask) &&
check_dst(n->n_nhop)) {
gate = n->n_nhop;
} else {
msglim(&bad_nhop,
FROM_NADDR,
"router %s to %s"
" has bad next hop %s",
naddr_ntoa(FROM_NADDR),
naddr_ntoa(dst),
naddr_ntoa(n->n_nhop));
n->n_nhop = 0;
}
}
}
if (rip->rip_vers == RIPv1 ||
0 == (mask = ntohl(n->n_mask))) {
mask = ripv1_mask_host(dst, ifp);
} else if ((ntohl(dst) & ~mask) != 0) {
msglim(&bad_mask, FROM_NADDR,
"router %s sent bad netmask %s with %s",
naddr_ntoa(FROM_NADDR),
naddr_ntoa(htonl(mask)),
naddr_ntoa(dst));
continue;
}
if (mask == HOST_MASK &&
(ifp->int_state & IS_NO_HOST)) {
trace_pkt(" ignored host route %s",
addrname(dst, mask, 0));
continue;
}
if (rip->rip_vers == RIPv1)
n->n_tag = 0;
/*
* Adjust metric according to incoming interface cost.
* We intentionally don't drop incoming routes with
* metric 15 on the floor even though they will
* not be advertised to other routers. We can use
* such routes locally, resulting in a network with
* a maximum width of 15 hops rather than 14.
*/
n->n_metric += ifp->int_metric;
if (n->n_metric > HOPCNT_INFINITY)
n->n_metric = HOPCNT_INFINITY;
/*
* Should we trust this route from this router?
*/
if (tg != NULL && (tn = tg->tgate_nets)->mask != 0) {
for (i = 0; i < MAX_TGATE_NETS; i++, tn++) {
if (on_net(dst, tn->net, tn->mask) &&
tn->mask <= mask)
break;
}
if (i >= MAX_TGATE_NETS || tn->mask == 0) {
trace_pkt(" ignored unauthorized %s",
addrname(dst, mask, 0));
continue;
}
}
/*
* Recognize and ignore a default route we faked
* which is being sent back to us by a machine with
* broken split-horizon. Be a little more paranoid
* than that, and reject default routes with the
* same metric we advertised.
*/
if (ifp->int_d_metric != 0 && dst == RIP_DEFAULT &&
n->n_metric >= ifp->int_d_metric)
continue;
/*
* We can receive aggregated RIPv2 routes that must
* be broken down before they are transmitted by
* RIPv1 via an interface on a subnet. We might
* also receive the same routes aggregated via
* other RIPv2 interfaces. This could cause
* duplicate routes to be sent on the RIPv1
* interfaces. "Longest matching variable length
* netmasks" lets RIPv2 listeners understand, but
* breaking down the aggregated routes for RIPv1
* listeners can produce duplicate routes.
*
* Breaking down aggregated routes here bloats the
* daemon table, but does not hurt the kernel
* table, since routes are always aggregated for
* the kernel.
*
* Notice that this does not break down network
* routes corresponding to subnets. This is part of
* the defense against RS_NET_SYN.
*/
if (have_ripv1_out &&
(((rt = rtget(dst, mask)) == NULL ||
!(rt->rt_state & RS_NET_SYN))) &&
(v1_mask = ripv1_mask_net(dst, 0)) > mask) {
/* Get least significant set bit */
ddst_h = v1_mask & -v1_mask;
i = (v1_mask & ~mask)/ddst_h;
/*
* If you're going to make 512 or more
* routes, then that's just too many. The
* reason here is that breaking an old
* class B into /24 allocations is common
* enough that allowing for the creation of
* at least 256 deaggregated routes is
* good. The next power of 2 is 512.
*/
if (i >= 511) {
/*
* Punt if we would have to
* generate an unreasonable number
* of routes.
*/
if (TRACECONTENTS)
trace_misc("accept %s-->%s as 1"
" instead of %d routes",
addrname(dst, mask, 0),
naddr_ntoa(FROM_NADDR),
i + 1);
i = 0;
} else {
mask = v1_mask;
}
} else {
i = 0;
}
new.rts_gate = gate;
new.rts_router = FROM_NADDR;
new.rts_metric = n->n_metric;
new.rts_tag = n->n_tag;
new.rts_time = now.tv_sec;
new.rts_ifp = ifp;
new.rts_de_ag = i;
new.rts_origin = RO_RIP;
j = 0;
for (;;) {
input_route(dst, mask, &new, n, rt_state);
if (++j > i)
break;
dst = htonl(ntohl(dst) + ddst_h);
}
} while (++n < lim);
return;
case RIPCMD_POLLENTRY:
/*
* With this command one can request a single entry.
* Both silent processes and routers can respond to this
* command
*/
if (n >= lim) {
msglim(&bad_len, FROM_NADDR, "empty request from %s",
naddr_ntoa(FROM_NADDR));
return;
}
if (cc%sizeof (*n) != sizeof (struct rip)%sizeof (*n)) {
msglim(&bad_len, FROM_NADDR,
"request of bad length (%d) from %s",
cc, naddr_ntoa(FROM_NADDR));
}
if (rip->rip_vers == RIPv2 && (ifp == NULL ||
(ifp->int_state & IS_NO_RIPV1_OUT))) {
v12buf.buf->rip_vers = RIPv2;
} else {
v12buf.buf->rip_vers = RIPv1;
}
/* Dont bother with md5 authentication with POLLENTRY */
ap = NULL;
clr_ws_buf(&v12buf, ap);
n->n_metric = ntohl(n->n_metric);
if (n->n_family != RIP_AF_INET) {
msglim(&bad_router, FROM_NADDR,
"POLLENTRY request from %s for unsupported"
" (af %d) %s",
naddr_ntoa(FROM_NADDR),
ntohs(n->n_family),
naddr_ntoa(n->n_dst));
return;
}
/* We are being asked about a specific destination. */
v12buf.n->n_dst = dst = n->n_dst;
v12buf.n->n_family = RIP_AF_INET;
if (!check_dst(dst)) {
msglim(&bad_router, FROM_NADDR,
"bad queried destination %s from %s",
naddr_ntoa(dst),
naddr_ntoa(FROM_NADDR));
v12buf.n->n_metric = HOPCNT_INFINITY;
goto pollentry_done;
}
/* decide what mask was intended */
if (rip->rip_vers == RIPv1 ||
0 == (mask = ntohl(n->n_mask)) ||
0 != (ntohl(dst) & ~mask))
mask = ripv1_mask_host(dst, ifp);
/* try to find the answer */
rt = rtget(dst, mask);
if (rt == NULL && dst != RIP_DEFAULT)
rt = rtfind(n->n_dst);
if (v12buf.buf->rip_vers != RIPv1)
v12buf.n->n_mask = htonl(mask);
if (rt == NULL) {
/* we do not have the answer */
v12buf.n->n_metric = HOPCNT_INFINITY;
goto pollentry_done;
}
/*
* we have the answer, so compute the right metric and next
* hop.
*/
v12buf.n->n_metric = rt->rt_metric + 1;
if (v12buf.n->n_metric > HOPCNT_INFINITY)
v12buf.n->n_metric = HOPCNT_INFINITY;
if (v12buf.buf->rip_vers != RIPv1) {
v12buf.n->n_tag = rt->rt_tag;
if (ifp != NULL &&
on_net(rt->rt_gate, ifp->int_net, ifp->int_mask) &&
rt->rt_gate != ifp->int_addr)
v12buf.n->n_nhop = rt->rt_gate;
}
pollentry_done:
v12buf.n->n_metric = htonl(v12buf.n->n_metric);
/*
* Send the answer about specific routes.
*/
(void) output(OUT_QUERY, from, ifp, v12buf.buf,
((char *)v12buf.n - (char *)v12buf.buf));
break;
}
#undef FROM_NADDR
}
/*
* Process a single input route.
*/
void
input_route(in_addr_t dst, /* network order */
in_addr_t mask,
struct rt_spare *new,
struct netinfo *n,
uint16_t rt_state)
{
int i;
struct rt_entry *rt;
struct rt_spare *rts, *rts0;
struct interface *ifp1;
struct rt_spare *ptr;
size_t ptrsize;
/*
* See if we can already get there by a working interface. Ignore
* if so.
*/
ifp1 = ifwithaddr(dst, _B_TRUE, _B_FALSE);
if (ifp1 != NULL && (ifp1->int_state & IS_PASSIVE))
return;
/*
* Look for the route in our table.
*/
rt = rtget(dst, mask);
/* Consider adding the route if we do not already have it. */
if (rt == NULL) {
/* Ignore unknown routes being poisoned. */
if (new->rts_metric == HOPCNT_INFINITY)
return;
/* Ignore the route if it points to us */
if (n != NULL && n->n_nhop != 0 &&
NULL != ifwithaddr(n->n_nhop, _B_TRUE, _B_FALSE))
return;
/*
* If something has not gone crazy and tried to fill
* our memory, accept the new route.
*/
rtadd(dst, mask, rt_state, new);
return;
}
/*
* We already know about the route. Consider this update.
*
* If (rt->rt_state & RS_NET_SYN), then this route
* is the same as a network route we have inferred
* for subnets we know, in order to tell RIPv1 routers
* about the subnets.
*
* It is impossible to tell if the route is coming
* from a distant RIPv2 router with the standard
* netmask because that router knows about the entire
* network, or if it is a round-about echo of a
* synthetic, RIPv1 network route of our own.
* The worst is that both kinds of routes might be
* received, and the bad one might have the smaller
* metric. Partly solve this problem by never
* aggregating into such a route. Also keep it
* around as long as the interface exists.
*/
rts0 = rt->rt_spares;
trace_misc("rt 0x%lx num_spares %d", rt, rt->rt_num_spares);
for (rts = rts0, i = rt->rt_num_spares; i != 0; i--, rts++) {
if (rts->rts_router == new->rts_router)
break;
/*
* Note the worst slot to reuse,
* other than the current slot.
*/
if (BETTER_LINK(rt, rts0, rts))
rts0 = rts;
}
if (i != 0) {
/*
* Found a route from the router already in the table.
*/
/*
* If the new route is a route broken down from an
* aggregated route, and if the previous route is either
* not a broken down route or was broken down from a finer
* netmask, and if the previous route is current,
* then forget this one.
*/
if (new->rts_de_ag > rts->rts_de_ag &&
now_stale <= rts->rts_time)
return;
/*
* Keep poisoned routes around only long enough to pass
* the poison on. Use a new timestamp for good routes.
*/
if (rts->rts_metric == HOPCNT_INFINITY &&
new->rts_metric == HOPCNT_INFINITY)
new->rts_time = rts->rts_time;
/*
* If this is an update for the router we currently prefer,
* then note it.
*/
if (i == rt->rt_num_spares) {
rtchange(rt, rt->rt_state | rt_state, new, 0);
/*
* If the route got worse, check for something better.
*/
if (new->rts_metric != rts->rts_metric)
rtswitch(rt, 0);
return;
}
/*
* This is an update for a spare route.
* Finished if the route is unchanged.
*/
if (rts->rts_gate == new->rts_gate &&
rts->rts_metric == new->rts_metric &&
rts->rts_tag == new->rts_tag) {
if ((rt->rt_dst == RIP_DEFAULT) &&
(rts->rts_ifp != new->rts_ifp))
trace_misc("input_route update for spare");
trace_upslot(rt, rts, new);
*rts = *new;
return;
}
/*
* Forget it if it has gone bad.
*/
if (new->rts_metric == HOPCNT_INFINITY) {
rts_delete(rt, rts);
return;
}
} else {
/*
* The update is for a route we know about,
* but not from a familiar router.
*
* Ignore the route if it points to us.
*/
if (n != NULL && n->n_nhop != 0 &&
NULL != ifwithaddr(n->n_nhop, _B_TRUE, _B_FALSE))
return;
/* the loop above set rts0=worst spare */
if (rts0->rts_metric < HOPCNT_INFINITY) {
ptrsize = (rt->rt_num_spares + SPARE_INC) *
sizeof (struct rt_spare);
ptr = realloc(rt->rt_spares, ptrsize);
if (ptr != NULL) {
rt->rt_spares = ptr;
rts0 = &rt->rt_spares[rt->rt_num_spares];
(void) memset(rts0, 0,
SPARE_INC * sizeof (struct rt_spare));
rt->rt_num_spares += SPARE_INC;
for (rts = rts0, i = SPARE_INC;
i != 0; i--, rts++)
rts->rts_metric = HOPCNT_INFINITY;
}
}
rts = rts0;
/*
* Save the route as a spare only if it has
* a better metric than our worst spare.
* This also ignores poisoned routes (those
* received with metric HOPCNT_INFINITY).
*/
if (new->rts_metric >= rts->rts_metric)
return;
}
trace_upslot(rt, rts, new);
*rts = *new;
/* try to switch to a better route */
rtswitch(rt, rts);
}
/*
* Recorded information about peer's MD5 sequence numbers. This is
* used to validate that received sequence numbers are in
* non-decreasing order as per the RFC.
*/
struct peer_hash {
struct peer_hash *ph_next;
in_addr_t ph_addr;
time_t ph_heard;
uint32_t ph_seqno;
};
static struct peer_hash **peer_hashes;
static int ph_index;
static int ph_num_peers;
/*
* Get a peer_hash structure from the hash of known peers. Create a
* new one if not found. Returns NULL on unrecoverable allocation
* failure.
*/
static struct peer_hash *
get_peer_info(in_addr_t from)
{
struct peer_hash *php;
struct peer_hash *pnhp;
struct peer_hash **ph_pp;
struct peer_hash **ph2_pp;
struct peer_hash **ph3_pp;
int i;
static uint_t failed_count;
if (peer_hashes == NULL) {
peer_hashes = calloc(hash_table_sizes[0],
sizeof (peer_hashes[0]));
if (peer_hashes == NULL) {
if (++failed_count % 100 == 1)
msglog("no memory for peer hash");
return (NULL);
}
}
/* Search for peer in existing hash table */
ph_pp = peer_hashes + (from % hash_table_sizes[ph_index]);
for (php = ph_pp[0]; php != NULL; php = php->ph_next) {
if (php->ph_addr == from)
return (php);
}
/*
* Not found; we need to add this peer to the table. If there
* are already too many peers, then try to expand the table
* first. It's not a big deal if we can't expand the table
* right now due to memory constraints. We'll try again
* later.
*/
if (ph_num_peers >= hash_table_sizes[ph_index] * 5 &&
hash_table_sizes[ph_index + 1] != 0 &&
(ph_pp = calloc(hash_table_sizes[ph_index + 1],
sizeof (peer_hashes[0]))) != NULL) {
ph2_pp = peer_hashes;
for (i = hash_table_sizes[ph_index] - 1; i >= 0; i--) {
for (php = ph2_pp[i]; php != NULL; php = pnhp) {
pnhp = php->ph_next;
ph3_pp = ph_pp + (php->ph_addr %
hash_table_sizes[ph_index + 1]);
php->ph_next = ph3_pp[0];
ph3_pp[0] = php;
}
}
ph_index++;
free(peer_hashes);
peer_hashes = ph_pp;
ph_pp += from % hash_table_sizes[ph_index];
}
php = calloc(sizeof (*php), 1);
if (php == NULL) {
if (++failed_count % 100 == 1)
msglog("no memory for peer hash entry");
} else {
php->ph_addr = from;
php->ph_heard = now.tv_sec;
php->ph_next = ph_pp[0];
ph_pp[0] = php;
ph_num_peers++;
}
return (php);
}
/*
* Age out entries in the peer table. This is called every time we do
* a normal 30 second broadcast.
*/
void
age_peer_info(void)
{
struct peer_hash *php;
struct peer_hash *next_ph;
struct peer_hash *prev_ph;
struct peer_hash **ph_pp;
int i;
/*
* Scan through the list and remove peers that should not
* still have valid authenticated entries in the routing
* table.
*/
if ((ph_pp = peer_hashes) == NULL || ph_num_peers == 0)
return;
for (i = hash_table_sizes[ph_index] - 1; i >= 0; i--) {
prev_ph = NULL;
for (php = ph_pp[i]; php != NULL; php = next_ph) {
next_ph = php->ph_next;
if (php->ph_heard <= now_expire) {
if (prev_ph == NULL)
ph_pp[i] = next_ph;
else
prev_ph->ph_next = next_ph;
free(php);
if (--ph_num_peers == 0)
return;
} else {
prev_ph = php;
}
}
}
}
static boolean_t /* _B_FALSE if bad, _B_TRUE if good */
ck_passwd(struct interface *aifp,
struct rip *rip,
uint8_t *lim,
in_addr_t from,
struct msg_limit *use_authp)
{
#define NA (rip->rip_auths)
struct netauth *na2;
struct auth *ap;
MD5_CTX md5_ctx;
uchar_t hash[RIP_AUTH_PW_LEN];
int i, len;
struct peer_hash *php;
uint32_t seqno;
if ((uint8_t *)NA >= lim || NA->a_family != RIP_AF_AUTH) {
msglim(use_authp, from, "missing auth data from %s",
naddr_ntoa(from));
return (_B_FALSE);
}
/*
* Validate sequence number on RIPv2 responses using keyed MD5
* authentication per RFC 2082 section 3.2.2. Note that if we
* can't locate the peer information (due to transient
* allocation problems), then we don't do the test. Also note
* that we assume that all sequence numbers 0x80000000 or more
* away are "less than."
*
* We intentionally violate RFC 2082 with respect to one case:
* restablishing contact. The RFC says that you should
* continue to ignore old sequence numbers in this case but
* make a special allowance for 0. This is extremely foolish.
* The problem is that if the router has crashed, it's
* entirely possible that either we'll miss sequence zero (or
* that it might not even send it!) or that the peer doesn't
* remember what it last used for a sequence number. In
* either case, we'll create a failure state that persists
* until the sequence number happens to advance past the last
* one we saw. This is bad because it means that we may have
* to wait until the router has been up for at least as long
* as it was last time before we even pay attention to it.
* Meanwhile, other routers may listen to it if they hadn't
* seen it before (i.e., if they crashed in the meantime).
* This means -- perversely -- that stable systems that stay
* "up" for a long time pay a penalty for doing so.
*/
if (rip->rip_cmd == RIPCMD_RESPONSE && NA->a_type == RIP_AUTH_MD5 &&
(php = get_peer_info(from)) != NULL) {
/*
* If the entry that we find has been updated
* recently enough that the routes are known
* to still be good, but the sequence number
* looks bad, then discard the packet.
*/
seqno = ntohl(NA->au.a_md5.md5_seqno);
if (php->ph_heard > now_expire && php->ph_seqno != 0 &&
(seqno == 0 || ((seqno - php->ph_seqno) & 0x80000000ul))) {
msglim(use_authp, from,
"discarding sequence %x (older than %x)",
(unsigned)seqno, (unsigned)php->ph_seqno);
return (_B_FALSE);
}
php->ph_heard = now.tv_sec;
php->ph_seqno = seqno;
}
/*
* accept any current (+/- 24 hours) password
*/
for (ap = aifp->int_auth, i = 0; i < MAX_AUTH_KEYS; i++, ap++) {
if (ap->type != NA->a_type ||
(ulong_t)ap->start > (ulong_t)clk.tv_sec+DAY ||
(ulong_t)ap->end+DAY < (ulong_t)clk.tv_sec)
continue;
if (NA->a_type == RIP_AUTH_PW) {
if (0 == memcmp(NA->au.au_pw, ap->key, RIP_AUTH_PW_LEN))
return (_B_TRUE);
} else {
/*
* accept MD5 secret with the right key ID
*/
if (NA->au.a_md5.md5_keyid != ap->keyid)
continue;
len = ntohs(NA->au.a_md5.md5_pkt_len);
if ((len - sizeof (*rip)) % sizeof (*NA) != 0 ||
len > (lim - (uint8_t *)rip - sizeof (*NA))) {
msglim(use_authp, from,
"wrong MD5 RIPv2 packet length of %d"
" instead of %d from %s",
len, lim - (uint8_t *)rip - sizeof (*NA),
naddr_ntoa(from));
return (_B_FALSE);
}
na2 = (struct netauth *)(rip->rip_nets +
(len - 4) / sizeof (struct netinfo));
/*
* Given a good hash value, these are not security
* problems so be generous and accept the routes,
* after complaining.
*/
if (TRACEPACKETS) {
if (NA->au.a_md5.md5_auth_len !=
RIP_AUTH_MD5_LEN)
msglim(use_authp, from,
"unknown MD5 RIPv2 auth len %#x"
" instead of %#x from %s",
NA->au.a_md5.md5_auth_len,
RIP_AUTH_MD5_LEN,
naddr_ntoa(from));
if (na2->a_family != RIP_AF_AUTH)
msglim(use_authp, from,
"unknown MD5 RIPv2 family %#x"
" instead of %#x from %s",
na2->a_family, RIP_AF_AUTH,
naddr_ntoa(from));
if (na2->a_type != RIP_AUTH_TRAILER)
msglim(use_authp, from,
"MD5 RIPv2 hash has %#x"
" instead of %#x from %s",
ntohs(na2->a_type),
ntohs(RIP_AUTH_TRAILER),
naddr_ntoa(from));
}
MD5Init(&md5_ctx);
/*
* len+4 to include auth trailer's family/type in
* MD5 sum
*/
MD5Update(&md5_ctx, (uchar_t *)rip, len + 4);
MD5Update(&md5_ctx, ap->key, RIP_AUTH_MD5_LEN);
MD5Final(hash, &md5_ctx);
if (0 == memcmp(hash, na2->au.au_pw, sizeof (hash)))
return (_B_TRUE);
}
}
msglim(use_authp, from, "bad auth data from %s",
naddr_ntoa(from));
return (_B_FALSE);
#undef NA
}