freebsd-nq/sys/netiso/iso.c
1995-05-30 08:16:23 +00:00

920 lines
23 KiB
C

/*-
* Copyright (c) 1991, 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 acknowledgement:
* 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.
*
* @(#)iso.c 8.2 (Berkeley) 11/15/93
* $Id: iso.c,v 1.3 1995/04/26 18:10:58 pst Exp $
*/
/***********************************************************
Copyright IBM Corporation 1987
All Rights Reserved
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 appear in all copies and that
both that copyright notice and this permission notice appear in
supporting documentation, and that the name of IBM not be
used in advertising or publicity pertaining to distribution of the
software without specific, written prior permission.
IBM DISCLAIMS ALL WARRANTIES WITH REGARD TO THIS SOFTWARE, INCLUDING
ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS, IN NO EVENT SHALL
IBM BE LIABLE FOR ANY SPECIAL, INDIRECT OR CONSEQUENTIAL DAMAGES OR
ANY DAMAGES WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS,
WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION,
ARISING OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS
SOFTWARE.
******************************************************************/
/*
* ARGO Project, Computer Sciences Dept., University of Wisconsin - Madison
*/
/*
* $Header: /home/ncvs/src/sys/netiso/iso.c,v 1.3 1995/04/26 18:10:58 pst Exp $
* $Source: /home/ncvs/src/sys/netiso/iso.c,v $
*
* iso.c: miscellaneous routines to support the iso address family
*/
#include <sys/param.h>
#include <sys/systm.h>
#include <sys/ioctl.h>
#include <sys/mbuf.h>
#include <sys/domain.h>
#include <sys/protosw.h>
#include <sys/socket.h>
#include <sys/socketvar.h>
#include <sys/errno.h>
#include <net/if.h>
#include <net/route.h>
#include <netiso/iso.h>
#include <netiso/iso_var.h>
#include <netiso/iso_snpac.h>
#include <netiso/iso_pcb.h>
#include <netiso/clnp.h>
#include <netiso/argo_debug.h>
#ifdef TUBA
#include <netiso/tuba_table.h>
#endif
#ifdef ISO
int iso_interfaces = 0; /* number of external interfaces */
int ether_output();
void llc_rtrequest();
/*
* FUNCTION: iso_addrmatch1
*
* PURPOSE: decide if the two iso_addrs passed are equal
*
* RETURNS: true if the addrs match, false if they do not
*
* SIDE EFFECTS:
*
* NOTES:
*/
iso_addrmatch1(isoaa, isoab)
register struct iso_addr *isoaa, *isoab; /* addresses to check */
{
u_int compare_len;
IFDEBUG(D_ROUTE)
printf("iso_addrmatch1: comparing lengths: %d to %d\n", isoaa->isoa_len,
isoab->isoa_len);
printf("a:\n");
dump_buf(isoaa->isoa_genaddr, isoaa->isoa_len);
printf("b:\n");
dump_buf(isoab->isoa_genaddr, isoab->isoa_len);
ENDDEBUG
if ((compare_len = isoaa->isoa_len) != isoab->isoa_len) {
IFDEBUG(D_ROUTE)
printf("iso_addrmatch1: returning false because of lengths\n");
ENDDEBUG
return 0;
}
#ifdef notdef
/* TODO : generalize this to all afis with masks */
if( isoaa->isoa_afi == AFI_37 ) {
/* must not compare 2 least significant digits, or for
* that matter, the DSP
*/
compare_len = ADDR37_IDI_LEN - 1;
}
#endif
IFDEBUG(D_ROUTE)
int i;
char *a, *b;
a = isoaa->isoa_genaddr;
b = isoab->isoa_genaddr;
for (i=0; i<compare_len; i++) {
printf("<%x=%x>", a[i]&0xff, b[i]&0xff);
if (a[i] != b[i]) {
printf("\naddrs are not equal at byte %d\n", i);
return(0);
}
}
printf("\n");
printf("addrs are equal\n");
return (1);
ENDDEBUG
return (!bcmp(isoaa->isoa_genaddr, isoab->isoa_genaddr, compare_len));
}
/*
* FUNCTION: iso_addrmatch
*
* PURPOSE: decide if the two sockadrr_isos passed are equal
*
* RETURNS: true if the addrs match, false if they do not
*
* SIDE EFFECTS:
*
* NOTES:
*/
iso_addrmatch(sisoa, sisob)
struct sockaddr_iso *sisoa, *sisob; /* addresses to check */
{
return(iso_addrmatch1(&sisoa->siso_addr, &sisob->siso_addr));
}
#ifdef notdef
/*
* FUNCTION: iso_netmatch
*
* PURPOSE: similar to iso_addrmatch but takes sockaddr_iso
* as argument.
*
* RETURNS: true if same net, false if not
*
* SIDE EFFECTS:
*
* NOTES:
*/
iso_netmatch(sisoa, sisob)
struct sockaddr_iso *sisoa, *sisob;
{
u_char bufa[sizeof(struct sockaddr_iso)];
u_char bufb[sizeof(struct sockaddr_iso)];
register int lena, lenb;
lena = iso_netof(&sisoa->siso_addr, bufa);
lenb = iso_netof(&sisob->siso_addr, bufb);
IFDEBUG(D_ROUTE)
printf("iso_netmatch: comparing lengths: %d to %d\n", lena, lenb);
printf("a:\n");
dump_buf(bufa, lena);
printf("b:\n");
dump_buf(bufb, lenb);
ENDDEBUG
return ((lena == lenb) && (!bcmp(bufa, bufb, lena)));
}
#endif /* notdef */
/*
* FUNCTION: iso_hashchar
*
* PURPOSE: Hash all character in the buffer specified into
* a long. Return the long.
*
* RETURNS: The hash value.
*
* SIDE EFFECTS:
*
* NOTES: The hash is achieved by exclusive ORing 4 byte
* quantities.
*/
u_long
iso_hashchar(buf, len)
register caddr_t buf; /* buffer to pack from */
register int len; /* length of buffer */
{
register u_long h = 0;
register int i;
for (i=0; i<len; i+=4) {
register u_long l = 0;
if ((len - i) < 4) {
/* buffer not multiple of 4 */
switch (len - i) {
case 3:
l |= buf[i+2] << 8;
case 2:
l |= buf[i+1] << 16;
case 1:
l |= buf[i] << 24;
break;
default:
printf("iso_hashchar: unexpected value x%x\n", len - i);
break;
}
} else {
l |= buf[i] << 24;
l |= buf[i+1] << 16;
l |= buf[i+2] << 8;
l |= buf[i+3];
}
h ^= l;
}
h ^= (u_long) (len % 4);
return(h);
}
#ifdef notdef
/*
* FUNCTION: iso_hash
*
* PURPOSE: Fill in fields of afhash structure based upon addr passed.
*
* RETURNS: none
*
* SIDE EFFECTS:
*
* NOTES:
*/
iso_hash(siso, hp)
struct sockaddr_iso *siso; /* address to perform hash on */
struct afhash *hp; /* RETURN: hash info here */
{
u_long buf[sizeof(struct sockaddr_iso)+1/4];
register int bufsize;
bzero(buf, sizeof(buf));
bufsize = iso_netof(&siso->siso_addr, buf);
hp->afh_nethash = iso_hashchar((caddr_t)buf, bufsize);
IFDEBUG(D_ROUTE)
printf("iso_hash: iso_netof: bufsize = %d\n", bufsize);
ENDDEBUG
hp->afh_hosthash = iso_hashchar((caddr_t)&siso->siso_addr,
siso->siso_addr.isoa_len);
IFDEBUG(D_ROUTE)
printf("iso_hash: %s: nethash = x%x, hosthash = x%x\n",
clnp_iso_addrp(&siso->siso_addr), hp->afh_nethash,
hp->afh_hosthash);
ENDDEBUG
}
/*
* FUNCTION: iso_netof
*
* PURPOSE: Extract the network portion of the iso address.
* The network portion of the iso address varies depending
* on the type of address. The network portion of the
* address will include the IDP. The network portion is:
*
* TYPE DESC
* t37 The AFI and x.121 (IDI)
* osinet The AFI, orgid, snetid
* rfc986 The AFI, vers and network part of
* internet address.
*
* RETURNS: number of bytes placed into buf.
*
* SIDE EFFECTS:
*
* NOTES: Buf is assumed to be big enough
*/
iso_netof(isoa, buf)
struct iso_addr *isoa; /* address */
caddr_t buf; /* RESULT: network portion of address here */
{
u_int len = 1; /* length of afi */
switch (isoa->isoa_afi) {
case AFI_37:
/*
* Due to classic x.25 tunnel vision, there is no
* net portion of an x.121 address. For our purposes
* the AFI will do, so that all x.25 -type addresses
* map to the single x.25 SNPA. (Cannot have more than
* one, obviously).
*/
break;
/* case AFI_OSINET:*/
case AFI_RFC986: {
u_short idi; /* value of idi */
/* osinet and rfc986 have idi in the same place */
CTOH(isoa->rfc986_idi[0], isoa->rfc986_idi[1], idi);
if (idi == IDI_OSINET)
/*
* Network portion of OSINET address can only be the IDI. Clearly,
* with one x25 interface, one could get to several orgids, and
* several snetids.
len += (ADDROSINET_IDI_LEN + OVLOSINET_ORGID_LEN +
OVLOSINET_SNETID_LEN);
*/
len += ADDROSINET_IDI_LEN;
else if (idi == IDI_RFC986) {
u_long inetaddr;
struct ovl_rfc986 *o986 = (struct ovl_rfc986 *)isoa;
/* bump len to include idi and version (1 byte) */
len += ADDRRFC986_IDI_LEN + 1;
/* get inet addr long aligned */
bcopy(o986->o986_inetaddr, &inetaddr, sizeof(inetaddr));
inetaddr = ntohl(inetaddr); /* convert to host byte order */
IFDEBUG(D_ROUTE)
printf("iso_netof: isoa ");
dump_buf(isoa, sizeof(*isoa));
printf("iso_netof: inetaddr 0x%x ", inetaddr);
ENDDEBUG
/* bump len by size of network portion of inet address */
if (IN_CLASSA(inetaddr)) {
len += 4-IN_CLASSA_NSHIFT/8;
IFDEBUG(D_ROUTE)
printf("iso_netof: class A net len is now %d\n", len);
ENDDEBUG
} else if (IN_CLASSB(inetaddr)) {
len += 4-IN_CLASSB_NSHIFT/8;
IFDEBUG(D_ROUTE)
printf("iso_netof: class B net len is now %d\n", len);
ENDDEBUG
} else {
len += 4-IN_CLASSC_NSHIFT/8;
IFDEBUG(D_ROUTE)
printf("iso_netof: class C net len is now %d\n", len);
ENDDEBUG
}
} else
len = 0;
} break;
default:
len = 0;
}
bcopy((caddr_t)isoa, buf, len);
IFDEBUG(D_ROUTE)
printf("iso_netof: isoa ");
dump_buf(isoa, len);
printf("iso_netof: net ");
dump_buf(buf, len);
ENDDEBUG
return len;
}
#endif /* notdef */
/*
* Generic iso control operations (ioctl's).
* Ifp is 0 if not an interface-specific ioctl.
*/
/* ARGSUSED */
iso_control(so, cmd, data, ifp)
struct socket *so;
int cmd;
caddr_t data;
register struct ifnet *ifp;
{
register struct iso_ifreq *ifr = (struct iso_ifreq *)data;
register struct iso_ifaddr *ia = 0;
register struct ifaddr *ifa;
struct iso_ifaddr *oia;
struct iso_aliasreq *ifra = (struct iso_aliasreq *)data;
int error, hostIsNew, maskIsNew;
/*
* Find address for this interface, if it exists.
*/
if (ifp)
for (ia = iso_ifaddr; ia; ia = ia->ia_next)
if (ia->ia_ifp == ifp)
break;
switch (cmd) {
case SIOCAIFADDR_ISO:
case SIOCDIFADDR_ISO:
if (ifra->ifra_addr.siso_family == AF_ISO)
for (oia = ia; ia; ia = ia->ia_next) {
if (ia->ia_ifp == ifp &&
SAME_ISOADDR(&ia->ia_addr, &ifra->ifra_addr))
break;
}
if ((so->so_state & SS_PRIV) == 0)
return (EPERM);
if (ifp == 0)
panic("iso_control");
if (ia == (struct iso_ifaddr *)0) {
struct iso_ifaddr *nia;
if (cmd == SIOCDIFADDR_ISO)
return (EADDRNOTAVAIL);
#ifdef TUBA
/* XXXXXX can't be done in the proto init routines */
if (tuba_tree == 0)
tuba_table_init();
#endif
MALLOC(nia, struct iso_ifaddr *, sizeof(*nia),
M_IFADDR, M_WAITOK);
if (nia == (struct iso_ifaddr *)0)
return (ENOBUFS);
bzero((caddr_t)nia, sizeof(*nia));
if (ia = iso_ifaddr) {
for ( ; ia->ia_next; ia = ia->ia_next)
;
ia->ia_next = nia;
} else
iso_ifaddr = nia;
ia = nia;
if (ifa = ifp->if_addrlist) {
for ( ; ifa->ifa_next; ifa = ifa->ifa_next)
;
ifa->ifa_next = (struct ifaddr *) ia;
} else
ifp->if_addrlist = (struct ifaddr *) ia;
ia->ia_ifa.ifa_addr = (struct sockaddr *)&ia->ia_addr;
ia->ia_ifa.ifa_dstaddr
= (struct sockaddr *)&ia->ia_dstaddr;
ia->ia_ifa.ifa_netmask
= (struct sockaddr *)&ia->ia_sockmask;
ia->ia_ifp = ifp;
if (!(ifp->if_flags & IFF_LOOPBACK))
iso_interfaces++;
}
break;
#define cmdbyte(x) (((x) >> 8) & 0xff)
default:
if (cmdbyte(cmd) == 'a')
return (snpac_ioctl(so, cmd, data));
if (ia == (struct iso_ifaddr *)0)
return (EADDRNOTAVAIL);
break;
}
switch (cmd) {
case SIOCGIFADDR_ISO:
ifr->ifr_Addr = ia->ia_addr;
break;
case SIOCGIFDSTADDR_ISO:
if ((ifp->if_flags & IFF_POINTOPOINT) == 0)
return (EINVAL);
ifr->ifr_Addr = ia->ia_dstaddr;
break;
case SIOCGIFNETMASK_ISO:
ifr->ifr_Addr = ia->ia_sockmask;
break;
case SIOCAIFADDR_ISO:
maskIsNew = 0; hostIsNew = 1; error = 0;
if (ia->ia_addr.siso_family == AF_ISO) {
if (ifra->ifra_addr.siso_len == 0) {
ifra->ifra_addr = ia->ia_addr;
hostIsNew = 0;
} else if (SAME_ISOADDR(&ia->ia_addr, &ifra->ifra_addr))
hostIsNew = 0;
}
if (ifra->ifra_mask.siso_len) {
iso_ifscrub(ifp, ia);
ia->ia_sockmask = ifra->ifra_mask;
maskIsNew = 1;
}
if ((ifp->if_flags & IFF_POINTOPOINT) &&
(ifra->ifra_dstaddr.siso_family == AF_ISO)) {
iso_ifscrub(ifp, ia);
ia->ia_dstaddr = ifra->ifra_dstaddr;
maskIsNew = 1; /* We lie; but the effect's the same */
}
if (ifra->ifra_addr.siso_family == AF_ISO &&
(hostIsNew || maskIsNew)) {
error = iso_ifinit(ifp, ia, &ifra->ifra_addr, 0);
}
if (ifra->ifra_snpaoffset)
ia->ia_snpaoffset = ifra->ifra_snpaoffset;
return (error);
case SIOCDIFADDR_ISO:
iso_ifscrub(ifp, ia);
if ((ifa = ifp->if_addrlist) == (struct ifaddr *)ia)
ifp->if_addrlist = ifa->ifa_next;
else {
while (ifa->ifa_next &&
(ifa->ifa_next != (struct ifaddr *)ia))
ifa = ifa->ifa_next;
if (ifa->ifa_next)
ifa->ifa_next = ((struct ifaddr *)ia)->ifa_next;
else
printf("Couldn't unlink isoifaddr from ifp\n");
}
oia = ia;
if (oia == (ia = iso_ifaddr)) {
iso_ifaddr = ia->ia_next;
} else {
while (ia->ia_next && (ia->ia_next != oia)) {
ia = ia->ia_next;
}
if (ia->ia_next)
ia->ia_next = oia->ia_next;
else
printf("Didn't unlink isoifadr from list\n");
}
IFAFREE((&oia->ia_ifa));
break;
default:
if (ifp == 0 || ifp->if_ioctl == 0)
return (EOPNOTSUPP);
return ((*ifp->if_ioctl)(ifp, cmd, data));
}
return (0);
}
/*
* Delete any existing route for an interface.
*/
iso_ifscrub(ifp, ia)
register struct ifnet *ifp;
register struct iso_ifaddr *ia;
{
int nsellength = ia->ia_addr.siso_tlen;
if ((ia->ia_flags & IFA_ROUTE) == 0)
return;
ia->ia_addr.siso_tlen = 0;
if (ifp->if_flags & IFF_LOOPBACK)
rtinit(&(ia->ia_ifa), (int)RTM_DELETE, RTF_HOST);
else if (ifp->if_flags & IFF_POINTOPOINT)
rtinit(&(ia->ia_ifa), (int)RTM_DELETE, RTF_HOST);
else {
rtinit(&(ia->ia_ifa), (int)RTM_DELETE, 0);
}
ia->ia_addr.siso_tlen = nsellength;
ia->ia_flags &= ~IFA_ROUTE;
}
/*
* Initialize an interface's internet address
* and routing table entry.
*/
iso_ifinit(ifp, ia, siso, scrub)
register struct ifnet *ifp;
register struct iso_ifaddr *ia;
struct sockaddr_iso *siso;
{
struct sockaddr_iso oldaddr;
int s = splimp(), error, nsellength;
oldaddr = ia->ia_addr;
ia->ia_addr = *siso;
/*
* Give the interface a chance to initialize
* if this is its first address,
* and to validate the address if necessary.
*/
if (ifp->if_ioctl &&
(error = (*ifp->if_ioctl)(ifp, SIOCSIFADDR, (caddr_t)ia))) {
splx(s);
ia->ia_addr = oldaddr;
return (error);
}
if (scrub) {
ia->ia_ifa.ifa_addr = (struct sockaddr *)&oldaddr;
iso_ifscrub(ifp, ia);
ia->ia_ifa.ifa_addr = (struct sockaddr *)&ia->ia_addr;
}
/* XXX -- The following is here temporarily out of laziness
in not changing every ethernet driver's if_ioctl routine */
if (ifp->if_output == ether_output) {
ia->ia_ifa.ifa_rtrequest = llc_rtrequest;
ia->ia_ifa.ifa_flags |= RTF_CLONING;
}
/*
* Add route for the network.
*/
nsellength = ia->ia_addr.siso_tlen;
ia->ia_addr.siso_tlen = 0;
if (ifp->if_flags & IFF_LOOPBACK) {
ia->ia_ifa.ifa_dstaddr = ia->ia_ifa.ifa_addr;
error = rtinit(&(ia->ia_ifa), (int)RTM_ADD, RTF_HOST|RTF_UP);
} else if (ifp->if_flags & IFF_POINTOPOINT &&
ia->ia_dstaddr.siso_family == AF_ISO)
error = rtinit(&(ia->ia_ifa), (int)RTM_ADD, RTF_HOST|RTF_UP);
else {
rt_maskedcopy(ia->ia_ifa.ifa_addr, ia->ia_ifa.ifa_dstaddr,
ia->ia_ifa.ifa_netmask);
ia->ia_dstaddr.siso_nlen =
min(ia->ia_addr.siso_nlen, (ia->ia_sockmask.siso_len - 6));
error = rtinit(&(ia->ia_ifa), (int)RTM_ADD, RTF_UP);
}
ia->ia_addr.siso_tlen = nsellength;
ia->ia_flags |= IFA_ROUTE;
splx(s);
return (error);
}
#ifdef notdef
struct ifaddr *
iso_ifwithidi(addr)
register struct sockaddr *addr;
{
register struct ifnet *ifp;
register struct ifaddr *ifa;
register u_int af = addr->sa_family;
if (af != AF_ISO)
return (0);
IFDEBUG(D_ROUTE)
printf(">>> iso_ifwithidi addr\n");
dump_isoaddr( (struct sockaddr_iso *)(addr));
printf("\n");
ENDDEBUG
for (ifp = ifnet; ifp; ifp = ifp->if_next) {
IFDEBUG(D_ROUTE)
printf("iso_ifwithidi ifnet %s\n", ifp->if_name);
ENDDEBUG
for (ifa = ifp->if_addrlist; ifa; ifa = ifa->ifa_next) {
IFDEBUG(D_ROUTE)
printf("iso_ifwithidi address ");
dump_isoaddr( (struct sockaddr_iso *)(ifa->ifa_addr));
ENDDEBUG
if (ifa->ifa_addr->sa_family != addr->sa_family)
continue;
#define IFA_SIS(ifa)\
((struct sockaddr_iso *)((ifa)->ifa_addr))
IFDEBUG(D_ROUTE)
printf(" af same, args to iso_eqtype:\n");
printf("0x%x ", IFA_SIS(ifa)->siso_addr);
printf(" 0x%x\n",
&(((struct sockaddr_iso *)addr)->siso_addr));
ENDDEBUG
if (iso_eqtype(&(IFA_SIS(ifa)->siso_addr),
&(((struct sockaddr_iso *)addr)->siso_addr))) {
IFDEBUG(D_ROUTE)
printf("ifa_ifwithidi: ifa found\n");
ENDDEBUG
return (ifa);
}
IFDEBUG(D_ROUTE)
printf(" iso_eqtype failed\n");
ENDDEBUG
}
}
return ((struct ifaddr *)0);
}
#endif /* notdef */
/*
* FUNCTION: iso_ck_addr
*
* PURPOSE: return true if the iso_addr passed is
* within the legal size limit for an iso address.
*
* RETURNS: true or false
*
* SIDE EFFECTS:
*
*/
iso_ck_addr(isoa)
struct iso_addr *isoa; /* address to check */
{
return (isoa->isoa_len <= 20);
}
#ifdef notdef
/*
* FUNCTION: iso_eqtype
*
* PURPOSE: Determine if two iso addresses are of the same type.
* This is flaky. Really we should consider all type 47 addrs to be the
* same - but there do exist different structures for 47 addrs.
* Gosip adds a 3rd.
*
* RETURNS: true if the addresses are the same type
*
* SIDE EFFECTS:
*
* NOTES: By type, I mean rfc986, t37, or osinet
*
* This will first compare afis. If they match, then
* if the addr is not t37, the idis must be compared.
*/
iso_eqtype(isoaa, isoab)
struct iso_addr *isoaa; /* first addr to check */
struct iso_addr *isoab; /* other addr to check */
{
if (isoaa->isoa_afi == isoab->isoa_afi) {
if (isoaa->isoa_afi == AFI_37)
return(1);
else
return (!bcmp(&isoaa->isoa_u, &isoab->isoa_u, 2));
}
return(0);
}
#endif /* notdef */
/*
* FUNCTION: iso_localifa()
*
* PURPOSE: Find an interface addresss having a given destination
* or at least matching the net.
*
* RETURNS: ptr to an interface address
*
* SIDE EFFECTS:
*
* NOTES:
*/
struct iso_ifaddr *
iso_localifa(siso)
register struct sockaddr_iso *siso;
{
register struct iso_ifaddr *ia;
register char *cp1, *cp2, *cp3;
register struct ifnet *ifp;
struct iso_ifaddr *ia_maybe = 0;
/*
* We make one pass looking for both net matches and an exact
* dst addr.
*/
for (ia = iso_ifaddr; ia; ia = ia->ia_next) {
if ((ifp = ia->ia_ifp) == 0 || ((ifp->if_flags & IFF_UP) == 0))
continue;
if (ifp->if_flags & IFF_POINTOPOINT) {
if ((ia->ia_dstaddr.siso_family == AF_ISO) &&
SAME_ISOADDR(&ia->ia_dstaddr, siso))
return (ia);
else
if (SAME_ISOADDR(&ia->ia_addr, siso))
ia_maybe = ia;
continue;
}
if (ia->ia_sockmask.siso_len) {
char *cplim = ia->ia_sockmask.siso_len + (char *)&ia->ia_sockmask;
cp1 = ia->ia_sockmask.siso_data;
cp2 = siso->siso_data;
cp3 = ia->ia_addr.siso_data;
while (cp1 < cplim)
if (*cp1++ & (*cp2++ ^ *cp3++))
goto next;
ia_maybe = ia;
}
if (SAME_ISOADDR(&ia->ia_addr, siso))
return ia;
next:;
}
return ia_maybe;
}
#ifdef TPCONS
#include <netiso/cons.h>
#endif /* TPCONS */
/*
* FUNCTION: iso_nlctloutput
*
* PURPOSE: Set options at the network level
*
* RETURNS: E*
*
* SIDE EFFECTS:
*
* NOTES: This could embody some of the functions of
* rclnp_ctloutput and cons_ctloutput.
*/
iso_nlctloutput(cmd, optname, pcb, m)
int cmd; /* command:set or get */
int optname; /* option of interest */
caddr_t pcb; /* nl pcb */
struct mbuf *m; /* data for set, buffer for get */
{
struct isopcb *isop = (struct isopcb *)pcb;
int error = 0; /* return value */
caddr_t data; /* data for option */
int data_len; /* data's length */
IFDEBUG(D_ISO)
printf("iso_nlctloutput: cmd %x, opt %x, pcb %x, m %x\n",
cmd, optname, pcb, m);
ENDDEBUG
if ((cmd != PRCO_GETOPT) && (cmd != PRCO_SETOPT))
return(EOPNOTSUPP);
data = mtod(m, caddr_t);
data_len = (m)->m_len;
IFDEBUG(D_ISO)
printf("iso_nlctloutput: data is:\n");
dump_buf(data, data_len);
ENDDEBUG
switch (optname) {
#ifdef TPCONS
case CONSOPT_X25CRUD:
if (cmd == PRCO_GETOPT) {
error = EOPNOTSUPP;
break;
}
if (data_len > MAXX25CRUDLEN) {
error = EINVAL;
break;
}
IFDEBUG(D_ISO)
printf("iso_nlctloutput: setting x25 crud\n");
ENDDEBUG
bcopy(data, (caddr_t)isop->isop_x25crud, (unsigned)data_len);
isop->isop_x25crud_len = data_len;
break;
#endif /* TPCONS */
default:
error = EOPNOTSUPP;
}
if (cmd == PRCO_SETOPT)
m_freem(m);
return error;
}
#endif /* ISO */
#ifdef ARGO_DEBUG
/*
* FUNCTION: dump_isoaddr
*
* PURPOSE: debugging
*
* RETURNS: nada
*
*/
dump_isoaddr(s)
struct sockaddr_iso *s;
{
char *clnp_saddr_isop();
register int i;
if( s->siso_family == AF_ISO) {
printf("ISO address: suffixlen %d, %s\n",
s->siso_tlen, clnp_saddr_isop(s));
} else if( s->siso_family == AF_INET) {
/* hack */
struct sockaddr_in *sin = (struct sockaddr_in *)s;
printf("%d.%d.%d.%d: %d",
(sin->sin_addr.s_addr>>24)&0xff,
(sin->sin_addr.s_addr>>16)&0xff,
(sin->sin_addr.s_addr>>8)&0xff,
(sin->sin_addr.s_addr)&0xff,
sin->sin_port);
}
}
#endif /* ARGO_DEBUG */