1
0
mirror of https://git.FreeBSD.org/src.git synced 2024-12-26 11:47:31 +00:00
freebsd/sys/netatalk/ddp_input.c
Robert Watson 1761672b75 Inline umich license from COPYRIGHT to make it clear what license the
umich copyright is asserting.

Clarify that the copyright I'm asserting is the standard Berkeley
license.

Remove Giant assertions from AARP and DDP input routines.
2004-08-10 03:23:05 +00:00

503 lines
13 KiB
C

/*
* Copyright (c) 2004 Robert N. M. Watson
* 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.
*
* THIS SOFTWARE IS PROVIDED BY THE AUTHOR 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 AUTHOR 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) 1990,1994 Regents of The University of Michigan.
*
* Permission to use, copy, modify, and distribute this software and
* its documentation for any purpose and without fee is hereby granted,
* provided that the above copyright notice appears in all copies and
* that both that copyright notice and this permission notice appear
* in supporting documentation, and that the name of The University
* of Michigan not be used in advertising or publicity pertaining to
* distribution of the software without specific, written prior
* permission. This software is supplied as is without expressed or
* implied warranties of any kind.
*
* This product includes software developed by the University of
* California, Berkeley and its contributors.
*
* Research Systems Unix Group
* The University of Michigan
* c/o Wesley Craig
* 535 W. William Street
* Ann Arbor, Michigan
* +1-313-764-2278
* netatalk@umich.edu
*
* $FreeBSD$
*/
#include "opt_mac.h"
#include <sys/param.h>
#include <sys/kernel.h>
#include <sys/lock.h>
#include <sys/mac.h>
#include <sys/mbuf.h>
#include <sys/signalvar.h>
#include <sys/socket.h>
#include <sys/socketvar.h>
#include <sys/sx.h>
#include <sys/systm.h>
#include <net/if.h>
#include <net/route.h>
#include <netatalk/at.h>
#include <netatalk/at_var.h>
#include <netatalk/ddp.h>
#include <netatalk/ddp_var.h>
#include <netatalk/ddp_pcb.h>
#include <netatalk/at_extern.h>
static volatile int ddp_forward = 1;
static volatile int ddp_firewall = 0;
static struct ddpstat ddpstat;
static struct route forwro;
static void ddp_input(struct mbuf *, struct ifnet *, struct elaphdr *, int);
/*
* Could probably merge these two code segments a little better...
*/
void
at2intr(struct mbuf *m)
{
/*
* Phase 2 packet handling
*/
ddp_input(m, m->m_pkthdr.rcvif, NULL, 2);
return;
}
void
at1intr(struct mbuf *m)
{
struct elaphdr *elhp, elh;
/*
* Phase 1 packet handling
*/
if (m->m_len < SZ_ELAPHDR && ((m = m_pullup(m, SZ_ELAPHDR)) == NULL)) {
ddpstat.ddps_tooshort++;
return;
}
/*
* This seems a little dubious, but I don't know phase 1 so leave it.
*/
elhp = mtod(m, struct elaphdr *);
m_adj(m, SZ_ELAPHDR);
if (elhp->el_type == ELAP_DDPEXTEND) {
ddp_input(m, m->m_pkthdr.rcvif, NULL, 1);
} else {
bcopy((caddr_t)elhp, (caddr_t)&elh, SZ_ELAPHDR);
ddp_input(m, m->m_pkthdr.rcvif, &elh, 1);
}
return;
}
static void
ddp_input(m, ifp, elh, phase)
struct mbuf *m;
struct ifnet *ifp;
struct elaphdr *elh;
int phase;
{
struct sockaddr_at from, to;
struct ddpshdr *dsh, ddps;
struct at_ifaddr *aa;
struct ddpehdr *deh = NULL, ddpe;
struct ddpcb *ddp;
int dlen, mlen;
u_short cksum = 0;
bzero((caddr_t)&from, sizeof(struct sockaddr_at));
bzero((caddr_t)&to, sizeof(struct sockaddr_at));
if (elh != NULL) {
/*
* Extract the information in the short header.
* netowrk information is defaulted to ATADDR_ANYNET
* and node information comes from the elh info.
* We must be phase 1.
*/
ddpstat.ddps_short++;
if (m->m_len < sizeof(struct ddpshdr) &&
((m = m_pullup(m, sizeof(struct ddpshdr))) == NULL)) {
ddpstat.ddps_tooshort++;
return;
}
dsh = mtod(m, struct ddpshdr *);
bcopy((caddr_t)dsh, (caddr_t)&ddps, sizeof(struct ddpshdr));
ddps.dsh_bytes = ntohl(ddps.dsh_bytes);
dlen = ddps.dsh_len;
to.sat_addr.s_net = ATADDR_ANYNET;
to.sat_addr.s_node = elh->el_dnode;
to.sat_port = ddps.dsh_dport;
from.sat_addr.s_net = ATADDR_ANYNET;
from.sat_addr.s_node = elh->el_snode;
from.sat_port = ddps.dsh_sport;
/*
* Make sure that we point to the phase1 ifaddr info
* and that it's valid for this packet.
*/
for (aa = at_ifaddr_list; aa != NULL; aa = aa->aa_next) {
if ((aa->aa_ifp == ifp)
&& ((aa->aa_flags & AFA_PHASE2) == 0)
&& ((to.sat_addr.s_node == AA_SAT(aa)->sat_addr.s_node)
|| (to.sat_addr.s_node == ATADDR_BCAST))) {
break;
}
}
/*
* maybe we got a broadcast not meant for us.. ditch it.
*/
if (aa == NULL) {
m_freem(m);
return;
}
} else {
/*
* There was no 'elh' passed on. This could still be
* either phase1 or phase2.
* We have a long header, but we may be running on a phase 1 net.
* Extract out all the info regarding this packet's src & dst.
*/
ddpstat.ddps_long++;
if (m->m_len < sizeof(struct ddpehdr) &&
((m = m_pullup(m, sizeof(struct ddpehdr))) == NULL)) {
ddpstat.ddps_tooshort++;
return;
}
deh = mtod(m, struct ddpehdr *);
bcopy((caddr_t)deh, (caddr_t)&ddpe, sizeof(struct ddpehdr));
ddpe.deh_bytes = ntohl(ddpe.deh_bytes);
dlen = ddpe.deh_len;
if ((cksum = ddpe.deh_sum) == 0) {
ddpstat.ddps_nosum++;
}
from.sat_addr.s_net = ddpe.deh_snet;
from.sat_addr.s_node = ddpe.deh_snode;
from.sat_port = ddpe.deh_sport;
to.sat_addr.s_net = ddpe.deh_dnet;
to.sat_addr.s_node = ddpe.deh_dnode;
to.sat_port = ddpe.deh_dport;
if (to.sat_addr.s_net == ATADDR_ANYNET) {
/*
* The TO address doesn't specify a net,
* So by definition it's for this net.
* Try find ifaddr info with the right phase,
* the right interface, and either to our node, a broadcast,
* or looped back (though that SHOULD be covered in the other
* cases).
*
* XXX If we have multiple interfaces, then the first with
* this node number will match (which may NOT be what we want,
* but it's probably safe in 99.999% of cases.
*/
for (aa = at_ifaddr_list; aa != NULL; aa = aa->aa_next) {
if (phase == 1 && (aa->aa_flags & AFA_PHASE2)) {
continue;
}
if (phase == 2 && (aa->aa_flags & AFA_PHASE2) == 0) {
continue;
}
if ((aa->aa_ifp == ifp)
&& ((to.sat_addr.s_node == AA_SAT(aa)->sat_addr.s_node)
|| (to.sat_addr.s_node == ATADDR_BCAST)
|| (ifp->if_flags & IFF_LOOPBACK))) {
break;
}
}
} else {
/*
* A destination network was given. We just try to find
* which ifaddr info matches it.
*/
for (aa = at_ifaddr_list; aa != NULL; aa = aa->aa_next) {
/*
* This is a kludge. Accept packets that are
* for any router on a local netrange.
*/
if (to.sat_addr.s_net == aa->aa_firstnet &&
to.sat_addr.s_node == 0) {
break;
}
/*
* Don't use ifaddr info for which we are totally outside the
* netrange, and it's not a startup packet.
* Startup packets are always implicitly allowed on to
* the next test.
*/
if (((ntohs(to.sat_addr.s_net) < ntohs(aa->aa_firstnet))
|| (ntohs(to.sat_addr.s_net) > ntohs(aa->aa_lastnet)))
&& ((ntohs(to.sat_addr.s_net) < 0xff00)
|| (ntohs(to.sat_addr.s_net) > 0xfffe))) {
continue;
}
/*
* Don't record a match either if we just don't have a match
* in the node address. This can have if the interface
* is in promiscuous mode for example.
*/
if ((to.sat_addr.s_node != AA_SAT(aa)->sat_addr.s_node)
&& (to.sat_addr.s_node != ATADDR_BCAST)) {
continue;
}
break;
}
}
}
/*
* Adjust the length, removing any padding that may have been added
* at a link layer. We do this before we attempt to forward a packet,
* possibly on a different media.
*/
mlen = m->m_pkthdr.len;
if (mlen < dlen) {
ddpstat.ddps_toosmall++;
m_freem(m);
return;
}
if (mlen > dlen) {
m_adj(m, dlen - mlen);
}
/*
* If it aint for a net on any of our interfaces,
* or it IS for a net on a different interface than it came in on,
* (and it is not looped back) then consider if we should forward it.
* As we are not really a router this is a bit cheeky, but it may be
* useful some day.
*/
if ((aa == NULL)
|| ((to.sat_addr.s_node == ATADDR_BCAST)
&& (aa->aa_ifp != ifp)
&& ((ifp->if_flags & IFF_LOOPBACK) == 0))) {
/*
* If we've explicitly disabled it, don't route anything
*/
if (ddp_forward == 0) {
m_freem(m);
return;
}
/*
* If the cached forwarding route is still valid, use it.
*/
if (forwro.ro_rt
&& (satosat(&forwro.ro_dst)->sat_addr.s_net != to.sat_addr.s_net
|| satosat(&forwro.ro_dst)->sat_addr.s_node != to.sat_addr.s_node)) {
RTFREE(forwro.ro_rt);
forwro.ro_rt = NULL;
}
/*
* If we don't have a cached one (any more) or it's useless,
* Then get a new route.
* XXX this could cause a 'route leak'. check this!
*/
if (forwro.ro_rt == NULL || forwro.ro_rt->rt_ifp == NULL) {
forwro.ro_dst.sa_len = sizeof(struct sockaddr_at);
forwro.ro_dst.sa_family = AF_APPLETALK;
satosat(&forwro.ro_dst)->sat_addr.s_net = to.sat_addr.s_net;
satosat(&forwro.ro_dst)->sat_addr.s_node = to.sat_addr.s_node;
rtalloc(&forwro);
}
/*
* If it's not going to get there on this hop, and it's
* already done too many hops, then throw it away.
*/
if ((to.sat_addr.s_net != satosat(&forwro.ro_dst)->sat_addr.s_net)
&& (ddpe.deh_hops == DDP_MAXHOPS)) {
m_freem(m);
return;
}
/*
* A ddp router might use the same interface
* to forward the packet, which this would not effect.
* Don't allow packets to cross from one interface to another however.
*/
if (ddp_firewall
&& ((forwro.ro_rt == NULL)
|| (forwro.ro_rt->rt_ifp != ifp))) {
m_freem(m);
return;
}
/*
* Adjust the header.
* If it was a short header then it would have not gotten here,
* so we can assume there is room to drop the header in.
* XXX what about promiscuous mode, etc...
*/
ddpe.deh_hops++;
ddpe.deh_bytes = htonl(ddpe.deh_bytes);
bcopy((caddr_t)&ddpe, (caddr_t)deh, sizeof(u_short)); /* XXX deh? */
if (ddp_route(m, &forwro)) {
ddpstat.ddps_cantforward++;
} else {
ddpstat.ddps_forward++;
}
return;
}
/*
* It was for us, and we have an ifaddr to use with it.
*/
from.sat_len = sizeof(struct sockaddr_at);
from.sat_family = AF_APPLETALK;
/*
* We are no longer interested in the link layer.
* so cut it off.
*/
if (elh != NULL) {
m_adj(m, sizeof(struct ddpshdr));
} else {
if (ddp_cksum && cksum && cksum != at_cksum(m, sizeof(int))) {
ddpstat.ddps_badsum++;
m_freem(m);
return;
}
m_adj(m, sizeof(struct ddpehdr));
}
/*
* Search for ddp protocol control blocks that match these
* addresses.
*/
DDP_LIST_SLOCK();
if ((ddp = ddp_search(&from, &to, aa)) == NULL) {
goto out;
}
#ifdef MAC
SOCK_LOCK(ddp->ddp_socket);
if (mac_check_socket_deliver(ddp->ddp_socket, m) != 0) {
SOCK_UNLOCK(ddp->ddp_socket);
goto out;
}
SOCK_UNLOCK(ddp->ddp_socket);
#endif
/*
* If we found one, deliver th epacket to the socket
*/
if (sbappendaddr(&ddp->ddp_socket->so_rcv, (struct sockaddr *)&from,
m, NULL) == 0) {
/*
* If the socket is full (or similar error) dump the packet.
*/
ddpstat.ddps_nosockspace++;
goto out;
}
/*
* And wake up whatever might be waiting for it
*/
sorwakeup(ddp->ddp_socket);
m = NULL;
out:
DDP_LIST_SUNLOCK();
if (m != NULL)
m_freem(m);
}
#if 0
/* As if we haven't got enough of this sort of think floating
around the kernel :) */
#define BPXLEN 48
#define BPALEN 16
#include <ctype.h>
char hexdig[] = "0123456789ABCDEF";
static void
bprint(char *data, int len)
{
char xout[ BPXLEN ], aout[ BPALEN ];
int i = 0;
bzero(xout, BPXLEN);
bzero(aout, BPALEN);
for (;;) {
if (len < 1) {
if (i != 0) {
printf("%s\t%s\n", xout, aout);
}
printf("%s\n", "(end)");
break;
}
xout[ (i*3) ] = hexdig[ (*data & 0xf0) >> 4 ];
xout[ (i*3) + 1 ] = hexdig[ *data & 0x0f ];
if ((u_char)*data < 0x7f && (u_char)*data > 0x20) {
aout[ i ] = *data;
} else {
aout[ i ] = '.';
}
xout[ (i*3) + 2 ] = ' ';
i++;
len--;
data++;
if (i > BPALEN - 2) {
printf("%s\t%s\n", xout, aout);
bzero(xout, BPXLEN);
bzero(aout, BPALEN);
i = 0;
continue;
}
}
}
static void
m_printm(struct mbuf *m)
{
for (; m; m = m->m_next) {
bprint(mtod(m, char *), m->m_len);
}
}
#endif