NetBSD/usr.sbin/rbootd/bpf.c

426 lines
11 KiB
C
Raw Normal View History

/* $NetBSD: bpf.c,v 1.6 1995/11/14 08:41:42 thorpej Exp $ */
1995-08-21 21:05:08 +04:00
1994-01-11 18:43:38 +03:00
/*
* Copyright (c) 1988, 1992 The University of Utah and the Center
* for Software Science (CSS).
* Copyright (c) 1992, 1993
* The Regents of the University of California. All rights reserved.
*
* This code is derived from software contributed to Berkeley by
* the Center for Software Science of the University of Utah Computer
* Science Department. CSS requests users of this software to return
* to css-dist@cs.utah.edu any improvements that they make and grant
* CSS redistribution rights.
*
* 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.
*
* from: @(#)bpf.c 8.1 (Berkeley) 6/4/93
1994-01-11 18:43:38 +03:00
*
* From: Utah Hdr: bpf.c 3.1 92/07/06
1994-01-11 18:43:38 +03:00
* Author: Jeff Forys, University of Utah CSS
*/
#ifndef lint
/*static char sccsid[] = "@(#)bpf.c 8.1 (Berkeley) 6/4/93";*/
static char rcsid[] = "$NetBSD: bpf.c,v 1.6 1995/11/14 08:41:42 thorpej Exp $";
1994-01-11 18:43:38 +03:00
#endif /* not lint */
#include <sys/param.h>
#include <sys/ioctl.h>
#include <sys/socket.h>
#include <net/if.h>
#include <net/bpf.h>
#include <ctype.h>
#include <errno.h>
#include <fcntl.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <syslog.h>
#include <unistd.h>
#include "defs.h"
#include "pathnames.h"
static int BpfFd = -1;
static unsigned BpfLen = 0;
static u_int8_t *BpfPkt = NULL;
1994-01-11 18:43:38 +03:00
/*
** BpfOpen -- Open and initialize a BPF device.
**
** Parameters:
** None.
**
** Returns:
** File descriptor of opened BPF device (for select() etc).
**
** Side Effects:
** If an error is encountered, the program terminates here.
*/
int
BpfOpen()
{
struct ifreq ifr;
char bpfdev[32];
int n = 0;
/*
* Open the first available BPF device.
*/
do {
(void) sprintf(bpfdev, _PATH_BPF, n++);
BpfFd = open(bpfdev, O_RDWR);
} while (BpfFd < 0 && (errno == EBUSY || errno == EPERM));
if (BpfFd < 0) {
syslog(LOG_ERR, "bpf: no available devices: %m");
Exit(0);
}
/*
* Set interface name for bpf device, get data link layer
* type and make sure it's type Ethernet.
*/
(void) strncpy(ifr.ifr_name, IntfName, sizeof(ifr.ifr_name));
if (ioctl(BpfFd, BIOCSETIF, (caddr_t)&ifr) < 0) {
syslog(LOG_ERR, "bpf: ioctl(BIOCSETIF,%s): %m", IntfName);
Exit(0);
}
/*
* Make sure we are dealing with an Ethernet device.
*/
if (ioctl(BpfFd, BIOCGDLT, (caddr_t)&n) < 0) {
syslog(LOG_ERR, "bpf: ioctl(BIOCGDLT): %m");
Exit(0);
}
if (n != DLT_EN10MB) {
syslog(LOG_ERR,"bpf: %s: data-link type %d unsupported",
IntfName, n);
Exit(0);
}
/*
* On read(), return packets immediately (do not buffer them).
*/
n = 1;
if (ioctl(BpfFd, BIOCIMMEDIATE, (caddr_t)&n) < 0) {
syslog(LOG_ERR, "bpf: ioctl(BIOCIMMEDIATE): %m");
Exit(0);
}
/*
* Try to enable the chip/driver's multicast address filter to
* grab our RMP address. If this fails, try promiscuous mode.
* If this fails, there's no way we are going to get any RMP
* packets so just exit here.
*/
#ifdef MSG_EOR
ifr.ifr_addr.sa_len = RMP_ADDRLEN + 2;
#endif
ifr.ifr_addr.sa_family = AF_UNSPEC;
bcopy(&RmpMcastAddr[0], (char *)&ifr.ifr_addr.sa_data[0], RMP_ADDRLEN);
if (ioctl(BpfFd, SIOCADDMULTI, (caddr_t)&ifr) < 0) {
syslog(LOG_WARNING,
"bpf: can't add mcast addr (%m), setting promiscuous mode");
if (ioctl(BpfFd, BIOCPROMISC, (caddr_t)0) < 0) {
syslog(LOG_ERR, "bpf: can't set promiscuous mode: %m");
Exit(0);
}
}
/*
* Ask BPF how much buffer space it requires and allocate one.
*/
if (ioctl(BpfFd, BIOCGBLEN, (caddr_t)&BpfLen) < 0) {
syslog(LOG_ERR, "bpf: ioctl(BIOCGBLEN): %m");
Exit(0);
}
if (BpfPkt == NULL)
BpfPkt = (u_int8_t *)malloc(BpfLen);
1994-01-11 18:43:38 +03:00
if (BpfPkt == NULL) {
syslog(LOG_ERR, "bpf: out of memory (%u bytes for bpfpkt)",
BpfLen);
Exit(0);
}
/*
* Write a little program to snarf RMP Boot packets and stuff
* it down BPF's throat (i.e. set up the packet filter).
*/
{
#define RMP ((struct rmp_packet *)0)
static struct bpf_insn bpf_insn[] = {
{ BPF_LD|BPF_B|BPF_ABS, 0, 0, (long)&RMP->hp_llc.dsap },
{ BPF_JMP|BPF_JEQ|BPF_K, 0, 5, IEEE_DSAP_HP },
{ BPF_LD|BPF_H|BPF_ABS, 0, 0, (long)&RMP->hp_llc.cntrl },
{ BPF_JMP|BPF_JEQ|BPF_K, 0, 3, IEEE_CNTL_HP },
{ BPF_LD|BPF_H|BPF_ABS, 0, 0, (long)&RMP->hp_llc.dxsap },
{ BPF_JMP|BPF_JEQ|BPF_K, 0, 1, HPEXT_DXSAP },
{ BPF_RET|BPF_K, 0, 0, RMP_MAX_PACKET },
{ BPF_RET|BPF_K, 0, 0, 0x0 }
};
#undef RMP
static struct bpf_program bpf_pgm = {
sizeof(bpf_insn)/sizeof(bpf_insn[0]), bpf_insn
};
if (ioctl(BpfFd, BIOCSETF, (caddr_t)&bpf_pgm) < 0) {
syslog(LOG_ERR, "bpf: ioctl(BIOCSETF): %m");
Exit(0);
}
}
return(BpfFd);
}
/*
** BPF GetIntfName -- Return the name of a network interface attached to
** the system, or 0 if none can be found. The interface
** must be configured up; the lowest unit number is
** preferred; loopback is ignored.
**
** Parameters:
** errmsg - if no network interface found, *errmsg explains why.
**
** Returns:
** A (static) pointer to interface name, or NULL on error.
**
** Side Effects:
** None.
*/
char *
BpfGetIntfName(errmsg)
char **errmsg;
{
struct ifreq ibuf[8], *ifrp, *ifend, *mp;
struct ifconf ifc;
int fd;
int minunit, n;
char *cp;
static char device[sizeof(ifrp->ifr_name)];
static char errbuf[128] = "No Error!";
if (errmsg != NULL)
*errmsg = errbuf;
if ((fd = socket(AF_INET, SOCK_DGRAM, 0)) < 0) {
(void) strcpy(errbuf, "bpf: socket: %m");
return(NULL);
}
ifc.ifc_len = sizeof ibuf;
ifc.ifc_buf = (caddr_t)ibuf;
#ifdef OSIOCGIFCONF
if (ioctl(fd, OSIOCGIFCONF, (char *)&ifc) < 0 ||
ifc.ifc_len < sizeof(struct ifreq)) {
(void) strcpy(errbuf, "bpf: ioctl(OSIOCGIFCONF): %m");
return(NULL);
}
#else
if (ioctl(fd, SIOCGIFCONF, (char *)&ifc) < 0 ||
ifc.ifc_len < sizeof(struct ifreq)) {
(void) strcpy(errbuf, "bpf: ioctl(SIOCGIFCONF): %m");
return(NULL);
}
#endif
ifrp = ibuf;
ifend = (struct ifreq *)((char *)ibuf + ifc.ifc_len);
mp = 0;
minunit = 666;
for (; ifrp < ifend; ++ifrp) {
if (ioctl(fd, SIOCGIFFLAGS, (char *)ifrp) < 0) {
(void) strcpy(errbuf, "bpf: ioctl(SIOCGIFFLAGS): %m");
return(NULL);
}
/*
* If interface is down or this is the loopback interface,
* ignore it.
*/
if ((ifrp->ifr_flags & IFF_UP) == 0 ||
#ifdef IFF_LOOPBACK
(ifrp->ifr_flags & IFF_LOOPBACK))
#else
(strcmp(ifrp->ifr_name, "lo0") == 0))
#endif
continue;
for (cp = ifrp->ifr_name; !isdigit(*cp); ++cp)
;
n = atoi(cp);
if (n < minunit) {
minunit = n;
mp = ifrp;
}
}
(void) close(fd);
if (mp == 0) {
(void) strcpy(errbuf, "bpf: no interfaces found");
return(NULL);
}
(void) strcpy(device, mp->ifr_name);
return(device);
}
/*
** BpfRead -- Read packets from a BPF device and fill in `rconn'.
**
** Parameters:
** rconn - filled in with next packet.
** doread - is True if we can issue a read() syscall.
**
** Returns:
** True if `rconn' contains a new packet, False otherwise.
**
** Side Effects:
** None.
*/
int
BpfRead(rconn, doread)
RMPCONN *rconn;
int doread;
{
register int datlen, caplen, hdrlen;
static u_int8_t *bp = NULL, *ep = NULL;
1994-01-11 18:43:38 +03:00
int cc;
/*
* The read() may block, or it may return one or more packets.
* We let the caller decide whether or not we can issue a read().
*/
if (doread) {
if ((cc = read(BpfFd, (char *)BpfPkt, (int)BpfLen)) < 0) {
syslog(LOG_ERR, "bpf: read: %m");
return(0);
} else {
bp = BpfPkt;
ep = BpfPkt + cc;
}
}
#define bhp ((struct bpf_hdr *)bp)
/*
* If there is a new packet in the buffer, stuff it into `rconn'
* and return a success indication.
*/
if (bp < ep) {
datlen = bhp->bh_datalen;
caplen = bhp->bh_caplen;
hdrlen = bhp->bh_hdrlen;
if (caplen != datlen)
syslog(LOG_ERR,
"bpf: short packet dropped (%d of %d bytes)",
caplen, datlen);
else if (caplen > sizeof(struct rmp_packet))
syslog(LOG_ERR, "bpf: large packet dropped (%d bytes)",
caplen);
else {
rconn->rmplen = caplen;
1994-01-11 18:43:38 +03:00
bcopy((char *)&bhp->bh_tstamp, (char *)&rconn->tstamp,
sizeof(struct timeval));
bcopy((char *)bp + hdrlen, (char *)&rconn->rmp, caplen);
}
bp += BPF_WORDALIGN(caplen + hdrlen);
return(1);
}
#undef bhp
return(0);
}
/*
** BpfWrite -- Write packet to BPF device.
**
** Parameters:
** rconn - packet to send.
**
** Returns:
** True if write succeeded, False otherwise.
**
** Side Effects:
** None.
*/
int
BpfWrite(rconn)
RMPCONN *rconn;
{
if (write(BpfFd, (char *)&rconn->rmp, rconn->rmplen) < 0) {
1994-01-11 18:43:38 +03:00
syslog(LOG_ERR, "write: %s: %m", EnetStr(rconn));
return(0);
}
return(1);
}
/*
** BpfClose -- Close a BPF device.
**
** Parameters:
** None.
**
** Returns:
** Nothing.
**
** Side Effects:
** None.
*/
void
BpfClose()
{
struct ifreq ifr;
if (BpfPkt != NULL) {
free((char *)BpfPkt);
BpfPkt = NULL;
}
if (BpfFd == -1)
return;
#ifdef MSG_EOR
ifr.ifr_addr.sa_len = RMP_ADDRLEN + 2;
#endif
ifr.ifr_addr.sa_family = AF_UNSPEC;
bcopy(&RmpMcastAddr[0], (char *)&ifr.ifr_addr.sa_data[0], RMP_ADDRLEN);
if (ioctl(BpfFd, SIOCDELMULTI, (caddr_t)&ifr) < 0)
(void) ioctl(BpfFd, BIOCPROMISC, (caddr_t)0);
(void) close(BpfFd);
BpfFd = -1;
}