path: root/sys/netinet/ip_fil.c

/*	$OpenBSD: ip_fil.c,v 1.10 1997/02/11 22:23:12 kstailey Exp $	*/
/*
 * (C)opyright 1993,1994,1995 by Darren Reed.
 *
 * Redistribution and use in source and binary forms are permitted
 * provided that this notice is preserved and due credit is given
 * to the original author and the contributors.
 */
#if !defined(lint) && defined(LIBC_SCCS)
static	char	sccsid[] = "@(#)ip_fil.c	2.41 6/5/96 (C) 1993-1995 Darren Reed";
static	char	rcsid[] = "Id: ip_fil.c,v 2.0.1.5 1997/01/29 13:41:45 darrenr Exp ";
#endif

#include <sys/param.h>
#include <sys/errno.h>
#include <sys/file.h>
#include <sys/ioctl.h>
#include <sys/mbuf.h>
#include <sys/protosw.h>
#include <sys/socket.h>
#include <sys/syslog.h>
#include <sys/systm.h>
#include <sys/uio.h>

#include <net/if.h>
#ifdef sun
#include <net/af.h>
#endif
#include <net/route.h>
#include <netinet/in.h>
#include <netinet/in_var.h>
#include <netinet/in_systm.h>
#include <netinet/ip.h>
#include <netinet/ip_var.h>
#include <netinet/tcp.h>
#include <netinet/udp.h>
#include <netinet/tcpip.h>
#include <netinet/ip_icmp.h>

#include "ip_fil_compat.h"
#include "ip_fil.h"
#include "ip_frag.h"
#include "ip_nat.h"
#include "ip_state.h"
#ifndef	MIN
#define	MIN(a,b)	(((a)<(b))?(a):(b))
#endif

extern	fr_flags, fr_active;
extern	struct	protosw	inetsw[];
extern	int	(*fr_checkp) __P((ip_t *, int, struct ifnet *, int,
				 struct mbuf **));
#if	BSD < 199306
extern	int	ipfr_slowtimer __P((void));
static	int	(*fr_saveslowtimo) __P((void));
extern	int	tcp_ttl;
#else
extern	void	ipfr_slowtimer __P((void));
static	void	(*fr_saveslowtimo) __P((void));
#endif

static	void	frzerostats __P((caddr_t));

int	ipl_inited = 0;
int	ipl_unreach = ICMP_UNREACH_FILTER_PROHIB;
int	send_reset __P((struct tcpiphdr *));

#ifdef	IPFILTER_LOG
# define LOGSIZE	8192
int	ipllog __P((u_int, ip_t *, register fr_info_t *, struct mbuf *));
char	iplbuf[LOGSIZE];
caddr_t	iplh = iplbuf, iplt = iplbuf;
static	int	iplused = 0;
#endif /* IPFILTER_LOG */
static	int	(*fr_savep) __P((ip_t *, int, struct ifnet *, int,
				struct mbuf **));
static	void	frflush __P((caddr_t));
static	int	frrequest __P((int, caddr_t, int));

#if _BSDI_VERSION >= 199501
# include <sys/device.h>
# include <sys/conf.h>

int	iplioctl __P((dev_t, int, caddr_t, int, struct proc *));
int	iplopen __P((dev_t, int, int, struct proc *));
int	iplclose __P((dev_t, int, int, struct proc *));
# ifdef IPFILTER_LOG
int	iplread __P((dev_t, struct uio *, int));
# else
#  define iplread	noread
# endif

struct cfdriver iplcd = {
	NULL, "ipl", NULL, NULL, DV_DULL, 0
};

struct devsw iplsw = {
	&iplcd,
	iplopen, iplclose, iplread, nowrite, iplioctl, noselect, nommap,
	nostrat, nodump, nopsize, 0,
	nostop
};
#else /* _BSDI_VERSION >= 199501 */

# ifdef IPFILTER_LOG
#  if BSD >= 199306
int	iplread __P((dev_t, struct uio *, int));
#  else
int	iplread __P((dev_t, struct uio *));
#  endif
# else
#  define iplread	noread
# endif
int	iplioctl __P((dev_t, int, caddr_t, int));
int	iplopen __P((dev_t, int));
int	iplclose __P((dev_t, int));
#endif /* _BSDI_VERSION >= 199501 */

int	iplattach __P((void));
int	ipldetach __P((void));

#ifdef	IPFILTER_LKM
int	iplidentify __P((char *));

int
iplidentify(s)
     char *s;
{
	if (strcmp(s, "ipl") == 0)
		return 1;
	return 0;
}
#else
void	iplinit __P((void));
#endif /* IPFILTER_LKM */

void	ipfr_fastroute __P((struct mbuf *, fr_info_t *, frdest_t *));

int
iplattach()
{
	int s;

	SPLNET(s);
	if (ipl_inited || (fr_checkp == fr_check)) {
		printf("IP Filter: already initialized\n");
		SPLX(s);
		return EBUSY;
	}
	ipl_inited = 1;
	bzero((char *)nat_table, sizeof(nat_t *) * NAT_SIZE * 2);
	fr_savep = fr_checkp;
	fr_checkp = fr_check;
	fr_saveslowtimo = inetsw[0].pr_slowtimo;
	inetsw[0].pr_slowtimo = ipfr_slowtimer;
	SPLX(s);
	return 0;
}


int
ipldetach()
{
	int s, i = FR_INQUE|FR_OUTQUE;

	SPLNET(s);
	if (!ipl_inited)
	{
		printf("IP Filter: not initialized\n");
		SPLX(s);
		return EBUSY;
	}

	fr_checkp = fr_savep;
	inetsw[0].pr_slowtimo = fr_saveslowtimo;
	frflush((caddr_t)&i);
	ipl_inited = 0;

	ipfr_unload();
	ip_natunload();
	fr_stateunload();

	SPLX(s);
	return 0;
}


static	void
frzerostats(data)
     caddr_t	data;
{
	struct	friostat	fio;

	bcopy((char *)frstats, (char *)fio.f_st,
		sizeof(struct filterstats) * 2);
	fio.f_fin[0] = ipfilter[0][0];
	fio.f_fin[1] = ipfilter[0][1];
	fio.f_fout[0] = ipfilter[1][0];
	fio.f_fout[1] = ipfilter[1][1];
	fio.f_acctin[0] = ipacct[0][0];
	fio.f_acctin[1] = ipacct[0][1];
	fio.f_acctout[0] = ipacct[1][0];
	fio.f_acctout[1] = ipacct[1][1];
	fio.f_active = fr_active;
	IWCOPY((caddr_t)&fio, data, sizeof(fio));
	bzero((char *)frstats, sizeof(*frstats) * 2);
}


static void
frflush(data)
     caddr_t data;
{
	struct frentry *f, **fp;
	int flags = *(int *)data, flushed = 0, set = fr_active;

	bzero((char *)frcache, sizeof(frcache[0]) * 2);

	if (flags & FR_INACTIVE)
		set = 1 - set;
	if (flags & FR_OUTQUE) {
		for (fp = &ipfilter[1][set]; (f = *fp); ) {
			*fp = f->fr_next;
			KFREE(f);
			flushed++;
		}
		for (fp = &ipacct[1][set]; (f = *fp); ) {
			*fp = f->fr_next;
			KFREE(f);
			flushed++;
		}
	}
	if (flags & FR_INQUE) {
		for (fp = &ipfilter[0][set]; (f = *fp); ) {
			*fp = f->fr_next;
			KFREE(f);
			flushed++;
		}
		for (fp = &ipacct[0][set]; (f = *fp); ) {
			*fp = f->fr_next;
			KFREE(f);
			flushed++;
		}
	}
	*(int *)data = flushed;
}


/*
 * Filter ioctl interface.
 */
int
iplioctl(dev, cmd, data, mode
#if _BSDI_VERSION >= 199501
   , p)
	struct proc *p;
#else
   )
#endif
     dev_t dev;
     int cmd;
     caddr_t data;
     int mode;
{
	int error = 0, s, unit;

	unit = minor(dev);
	if (unit != 0)
		return ENXIO;

	SPLNET(s);
	switch (cmd) {
	case FIONREAD :
#ifdef IPFILTER_LOG
		*(int *)data = iplused;
#endif
		break;
#ifndef	IPFILTER_LKM
	case SIOCFRENB :
	{
		u_int	enable;

		if (!(mode & FWRITE))
			error = EPERM;
		else {
			IRCOPY(data, (caddr_t)&enable, sizeof(enable));
			if (enable)
				error = iplattach();
			else
				error = ipldetach();
		}
		break;
	}
#endif
	case SIOCSETFF :
		if (!(mode & FWRITE))
			error = EPERM;
		else
			IRCOPY(data, (caddr_t)&fr_flags, sizeof(fr_flags));
		break;
	case SIOCGETFF :
		IWCOPY((caddr_t)&fr_flags, data, sizeof(fr_flags));
		break;
	case SIOCINAFR :
	case SIOCRMAFR :
	case SIOCADAFR :
	case SIOCZRLST :
		if (!(mode & FWRITE))
			error = EPERM;
		else
			error = frrequest(cmd, data, fr_active);
		break;
	case SIOCINIFR :
	case SIOCRMIFR :
	case SIOCADIFR :
		if (!(mode & FWRITE))
			error = EPERM;
		else
			error = frrequest(cmd, data, 1 - fr_active);
		break;
	case SIOCSWAPA :
		if (!(mode & FWRITE))
			error = EPERM;
		else {
			bzero((char *)frcache, sizeof(frcache[0]) * 2);
			*(u_int *)data = fr_active;
			fr_active = 1 - fr_active;
		}
		break;
	case SIOCGETFS :
	{
		struct	friostat	fio;

		bcopy((char *)frstats, (char *)fio.f_st,
			sizeof(struct filterstats) * 2);
		fio.f_fin[0] = ipfilter[0][0];
		fio.f_fin[1] = ipfilter[0][1];
		fio.f_fout[0] = ipfilter[1][0];
		fio.f_fout[1] = ipfilter[1][1];
		fio.f_acctin[0] = ipacct[0][0];
		fio.f_acctin[1] = ipacct[0][1];
		fio.f_acctout[0] = ipacct[1][0];
		fio.f_acctout[1] = ipacct[1][1];
		fio.f_active = fr_active;
		IWCOPY((caddr_t)&fio, data, sizeof(fio));
		break;
	}
	case	SIOCFRZST :
		if (!(mode & FWRITE))
			error = EPERM;
		else
			frzerostats(data);
		break;
	case	SIOCIPFFL :
		if (!(mode & FWRITE))
			error = EPERM;
		else
			frflush(data);
		break;
#ifdef	IPFILTER_LOG
	case	SIOCIPFFB :
		if (!(mode & FWRITE))
			error = EPERM;
		else {
			*(int *)data = iplused;
			iplh = iplt = iplbuf;
			iplused = 0;
		}
		break;
#endif /* IPFILTER_LOG */
	case SIOCADNAT :
	case SIOCRMNAT :
	case SIOCGNATS :
	case SIOCGNATL :
	case SIOCFLNAT :
	case SIOCCNATL :
		error = nat_ioctl(data, cmd, mode);
		break;
	case SIOCGFRST :
		IWCOPY((caddr_t)ipfr_fragstats(), data, sizeof(ipfrstat_t));
		break;
	case SIOCGIPST :
		IWCOPY((caddr_t)fr_statetstats(), data, sizeof(ips_stat_t));
		break;
	default :
		error = EINVAL;
		break;
	}
	SPLX(s);
	return error;
}


static int
frrequest(req, data, set)
     int req, set;
     caddr_t data;
{
	register frentry_t *fp, *f, **fprev;
	register frentry_t **ftail;
	frentry_t fr;
	frdest_t *fdp;
	struct frentry frd;
	int error = 0, in;

	fp = &fr;
	IRCOPY(data, (caddr_t)fp, sizeof(*fp));

	bzero((char *)frcache, sizeof(frcache[0]) * 2);

	in = (fp->fr_flags & FR_INQUE) ? 0 : 1;
	if (fp->fr_flags & FR_ACCOUNT) {
		ftail = fprev = &ipacct[in][set];
	} else if (fp->fr_flags & (FR_OUTQUE|FR_INQUE))
		ftail = fprev = &ipfilter[in][set];
	else
		return ESRCH;

	IRCOPY((char *)fp, (char *)&frd, sizeof(frd));
	fp = &frd;
	if (*fp->fr_ifname) {
		fp->fr_ifa = GETUNIT(fp->fr_ifname);
		if (!fp->fr_ifa)
			fp->fr_ifa = (struct ifnet *)-1;
	}

	fdp = &fp->fr_dif;
	fp->fr_flags &= ~FR_DUP;
	if (*fdp->fd_ifname) {
		fdp->fd_ifp = GETUNIT(fdp->fd_ifname);
		if (!fdp->fd_ifp)
			fdp->fd_ifp = (struct ifnet *)-1;
		else
			fp->fr_flags |= FR_DUP;
	}

	fdp = &fp->fr_tif;
	if (*fdp->fd_ifname) {
		fdp->fd_ifp = GETUNIT(fdp->fd_ifname);
		if (!fdp->fd_ifp)
			fdp->fd_ifp = (struct ifnet *)-1;
	}

	/*
	 * Look for a matching filter rule, but don't include the next or
	 * interface pointer in the comparison (fr_next, fr_ifa).
	 */
	for (; (f = *ftail); ftail = &f->fr_next)
		if (bcmp((char *)&f->fr_ip, (char *)&fp->fr_ip,
			 FR_CMPSIZ) == 0)
			break;

	/*
	 * If zero'ing statistics, copy current to caller and zero.
	 */
	if (req == SIOCZRLST) {
		if (!f)
			return ESRCH;
		IWCOPY((caddr_t)f, data, sizeof(*f));
		f->fr_hits = 0;
		f->fr_bytes = 0;
		return 0;
	}

	if (!f) {
		ftail = fprev;
		if (req != SIOCINAFR && req != SIOCINIFR)
			while ((f = *ftail))
				ftail = &f->fr_next;
		else if (fp->fr_hits)
			while (--fp->fr_hits && (f = *ftail))
				ftail = &f->fr_next;
		f = NULL;
	}

	if (req == SIOCDELFR || req == SIOCRMIFR) {
		if (!f)
			error = ESRCH;
		else {
			*ftail = f->fr_next;
			(void) KFREE(f);
		}
	} else {
		if (f)
			error = EEXIST;
		else {
			if ((f = (struct frentry *)KMALLOC(sizeof(*f)))) {
				bcopy((char *)fp, (char *)f, sizeof(*f));
				f->fr_hits = 0;
				f->fr_next = *ftail;
				*ftail = f;
			} else
				error = ENOMEM;
		}
	}
	return (error);
}


#if !defined(linux)
/*
 * routines below for saving IP headers to buffer
 */
int
iplopen(dev, flags
#if _BSDI_VERSION >= 199501
    , devtype, p)
	int devtype;
	struct proc *p;
#else
    )
#endif
     dev_t dev;
     int flags;
{
	u_int min = minor(dev);

	if (min)
		min = ENXIO;
	return min;
}


int
iplclose(dev, flags
#if _BSDI_VERSION >= 199501
    , devtype, p)
	int devtype;
	struct proc *p;
#else
    )
#endif
     dev_t dev;
     int flags;
{
	u_int	min = minor(dev);

	if (min)
		min = ENXIO;
	return min;
}

# ifdef	IPFILTER_LOG
/*
 * iplread/ipllog
 * both of these must operate with at least splnet() lest they be
 * called during packet processing and cause an inconsistancy to appear in
 * the filter lists.
 */
#  if BSD >= 199306
int
iplread(dev, uio, ioflag)
     int ioflag;
#  else
int
iplread(dev, uio)
#  endif
     dev_t dev;
     register struct uio *uio;
{
	register int ret, s;
	register size_t sz, sx;
	int error;

	if (!uio->uio_resid)
		return 0;
	while (!iplused) {
		error = SLEEP(iplbuf, "ipl sleep");
		if (error)
			return error;
	}
	SPLNET(s);

	sx = sz = MIN(uio->uio_resid, iplused);
	if (iplh < iplt)
		sz = MIN(sz, LOGSIZE - (iplt - iplbuf));
	sx -= sz;

#  if BSD >= 199306 || defined(__FreeBSD__)
	uio->uio_rw = UIO_READ;
#  endif
	if (!(ret = UIOMOVE(iplt, sz, UIO_READ, uio))) {
		iplt += sz;
		iplused -= sz;
		if ((iplh < iplt) && (iplt == iplbuf + LOGSIZE))
			iplt = iplbuf;

		if (sx && !(ret = UIOMOVE(iplt, sx, UIO_READ, uio))) {
			iplt += sx;
			iplused -= sx;
			if ((iplh < iplt) && (iplt == iplbuf + LOGSIZE))
				iplt = iplbuf;
		}
		if (!iplused)	/* minimise wrapping around the end */
			iplh = iplt = iplbuf;
	}
	SPLX(s);
	return ret;
}
# endif /* IPFILTER_LOG */
#endif /* linux */


#ifdef	IPFILTER_LOG
int
ipllog(flags, ip, fin, m)
     u_int flags;
     ip_t *ip;
     register fr_info_t *fin;
     struct mbuf *m;
{
	struct ipl_ci iplci;
	register int len, mlen, hlen;
	struct ifnet *ifp = fin->fin_ifp;

	hlen = fin->fin_hlen;
	if (ip->ip_p == IPPROTO_TCP || ip->ip_p == IPPROTO_UDP)
		hlen += MIN(sizeof(tcphdr_t), fin->fin_dlen);
	else if (ip->ip_p == IPPROTO_ICMP) {
		struct	icmp	*icmp = (struct icmp *)((char *)ip + hlen);

		switch (icmp->icmp_type) {
		case ICMP_UNREACH :
		case ICMP_SOURCEQUENCH :
		case ICMP_REDIRECT :
		case ICMP_TIMXCEED :
		case ICMP_PARAMPROB :
			hlen += MIN(sizeof(struct icmp) + 8, fin->fin_dlen);
			break;
		default :
			hlen += MIN(sizeof(struct icmp), fin->fin_dlen);
			break;
		}
	}

	mlen = (flags & FR_LOGBODY) ? MIN(ip->ip_len - hlen, 128) : 0;
	len = hlen + sizeof(iplci) + mlen;
	if (iplused + len > LOGSIZE)
		return 0;
	iplused += len;

# ifdef	sun
	uniqtime(&iplci);
# endif
# if BSD >= 199306 || defined(__FreeBSD__)
	microtime((struct timeval *)&iplci);
# endif
	iplci.flags = flags;
	iplci.hlen = (u_char)hlen;
	iplci.plen = (u_char)mlen;
	iplci.rule = fin->fin_rule;
# if (defined(NetBSD) && (NetBSD <= 1991011) && (NetBSD >= 199603)) || \
	(defined(OpenBSD) && (OpenBSD >= 199603))
	strncpy(iplci.ifname, ifp->if_xname, IFNAMSIZ);
# else
	iplci.unit = (u_char)ifp->if_unit;
	if ((iplci.ifname[0] = ifp->if_name[0]))
		if ((iplci.ifname[1] = ifp->if_name[1]))
			if ((iplci.ifname[2] = ifp->if_name[2]))
				iplci.ifname[3] = ifp->if_name[3];
# endif
	/*      
	 * Guaranteed to succeed from above
	 */     
	(void) fr_copytolog((char *)&iplci, sizeof(iplci));

	for (len -= sizeof(iplci); m && len > 0; m = m->m_next, len -= hlen) {
		hlen = MIN(len, m->m_len);
		if (fr_copytolog(mtod(m, char *), hlen))
			break;
	}

	wakeup(iplbuf);
	return 1;
}
#endif /* IPFILTER_LOG */

/*
 * send_reset - this could conceivably be a call to tcp_respond(), but that
 * requires a large amount of setting up and isn't any more efficient.
 */
int
send_reset(ti)
     struct tcpiphdr *ti;
{
	struct tcpiphdr *tp;
	struct ip *ip;
	struct tcphdr *tcp;
	struct mbuf *m;
	int tlen = 0;

	if (ti->ti_flags & TH_RST)
		return -1;		/* feedback loop */
#if	BSD < 199306
	m = m_get(M_DONTWAIT, MT_HEADER);
#else
	m = m_gethdr(M_DONTWAIT, MT_HEADER);
	m->m_data += max_linkhdr;
#endif
	if (m == NULL)
		return -1;

	if (ti->ti_flags & TH_SYN)
		tlen = 1;
	m->m_len = sizeof (struct tcpiphdr);
#if	BSD >= 199306
	m->m_pkthdr.len = sizeof (struct tcpiphdr);
	m->m_pkthdr.rcvif = (struct ifnet *)0;
#endif
	bzero(mtod(m, char *), sizeof(struct tcpiphdr));
	ip = mtod(m, struct ip *);
	tp = mtod(m, struct tcpiphdr *);
	tcp = (struct tcphdr *)((char *)ip + sizeof(struct ip));

	ip->ip_src.s_addr = ti->ti_dst.s_addr;
	ip->ip_dst.s_addr = ti->ti_src.s_addr;
	tcp->th_dport = ti->ti_sport;
	tcp->th_sport = ti->ti_dport;
	tcp->th_ack = htonl(ntohl(ti->ti_seq) + tlen);
	tcp->th_off = sizeof(struct tcphdr) >> 2;
	tcp->th_flags = TH_RST|TH_ACK;
	tp->ti_pr = ((struct ip *)ti)->ip_p;
	tp->ti_len = htons(sizeof(struct tcphdr));
	tcp->th_sum = in_cksum(m, sizeof(struct tcpiphdr));

	ip->ip_tos = ((struct ip *)ti)->ip_tos;
	ip->ip_p = ((struct ip *)ti)->ip_p;
	ip->ip_len = sizeof (struct tcpiphdr);
#if BSD < 199306
	ip->ip_ttl = tcp_ttl;
#else
	ip->ip_ttl = ip_defttl;
#endif

	/*
	 * extra 0 in case of multicast
	 */
	(void) ip_output(m, (struct mbuf *)0, 0, 0, 0);
	return 0;
}


#ifndef	IPFILTER_LKM
void
iplinit()
{
	(void) iplattach();
	ip_init();
	(void) ipldetach();	/* XXX */
}
#endif


void
ipfr_fastroute(m0, fin, fdp)
     struct mbuf *m0;
     fr_info_t *fin;
     frdest_t *fdp;
{
	register struct ip *ip, *mhip;
	register struct mbuf *m = m0;
	register struct route *ro;
	struct ifnet *ifp = fdp->fd_ifp;
	int len, off, error = 0;
	int hlen = fin->fin_hlen;
	struct route iproute;
	struct sockaddr_in *dst;

	ip = mtod(m0, struct ip *);
	/*
	 * Route packet.
	 */
	ro = &iproute;
	bzero((caddr_t)ro, sizeof (*ro));
	dst = (struct sockaddr_in *)&ro->ro_dst;
	dst->sin_family = AF_INET;
	dst->sin_addr = fdp->fd_ip.s_addr ? fdp->fd_ip : ip->ip_dst;
#if	(BSD >= 199306) && !defined(__NetBSD__) && !defined(__bsdi__) \
	&& !defined(__OpenBSD__)
# ifdef	RTF_CLONING
	rtalloc_ign(ro, RTF_CLONING);
# else
	rtalloc_ign(ro, RTF_PRCLONING);
# endif
#else
	rtalloc(ro);
#endif
	if (!ifp) {
		if (!(fin->fin_fr->fr_flags & FR_FASTROUTE)) {
			error = -2;
			goto bad;
		}
		if (ro->ro_rt == 0 || (ifp = ro->ro_rt->rt_ifp) == 0) {
			if (in_localaddr(ip->ip_dst))
				error = EHOSTUNREACH;
			else
				error = ENETUNREACH;
			goto bad;
		}
		if (ro->ro_rt->rt_flags & RTF_GATEWAY)
			dst = (struct sockaddr_in *)&ro->ro_rt->rt_gateway;
	}
	ro->ro_rt->rt_use++;

	/*
	 * For input packets which are being "fastrouted", they won't
	 * go back through output filtering and miss their chance to get
	 * NAT'd.
	 */
	(void) ip_natout(ip, hlen, fin);
	if (fin->fin_out)
		ip->ip_sum = 0;
	/*
	 * If small enough for interface, can just send directly.
	 */
	if (ip->ip_len <= ifp->if_mtu) {
#ifndef sparc
		ip->ip_id = htons(ip->ip_id);
		ip->ip_len = htons(ip->ip_len);
		ip->ip_off = htons(ip->ip_off);
#endif
		if (!ip->ip_sum)
			ip->ip_sum = in_cksum(m, hlen);
#if	BSD >= 199306
		error = (*ifp->if_output)(ifp, m, (struct sockaddr *)dst,
					  ro->ro_rt);

#else
		error = (*ifp->if_output)(ifp, m, (struct sockaddr *)dst);
#endif
		goto done;
	}
	/*
	 * Too large for interface; fragment if possible.
	 * Must be able to put at least 8 bytes per fragment.
	 */
	if (ip->ip_off & IP_DF) {
		error = EMSGSIZE;
		goto bad;
	}
	len = (ifp->if_mtu - hlen) &~ 7;
	if (len < 8) {
		error = EMSGSIZE;
		goto bad;
	}

    {
	int mhlen, firstlen = len;
	struct mbuf **mnext = &m->m_act;

	/*
	 * Loop through length of segment after first fragment,
	 * make new header and copy data of each part and link onto chain.
	 */
	m0 = m;
	mhlen = sizeof (struct ip);
	for (off = hlen + len; off < ip->ip_len; off += len) {
		MGET(m, M_DONTWAIT, MT_HEADER);
		if (m == 0) {
			error = ENOBUFS;
			goto bad;
		}
#if BSD >= 199306
		m->m_data += max_linkhdr;
#else
		m->m_off = MMAXOFF - hlen;
#endif
		mhip = mtod(m, struct ip *);
		bcopy((char *)ip, (char *)mhip, sizeof(*ip));
		if (hlen > sizeof (struct ip)) {
			mhlen = ip_optcopy(ip, mhip) + sizeof (struct ip);
			mhip->ip_hl = mhlen >> 2;
		}
		m->m_len = mhlen;
		mhip->ip_off = ((off - hlen) >> 3) + (ip->ip_off & ~IP_MF);
		if (ip->ip_off & IP_MF)
			mhip->ip_off |= IP_MF;
		if (off + len >= ip->ip_len)
			len = ip->ip_len - off;
		else
			mhip->ip_off |= IP_MF;
		mhip->ip_len = htons((u_short)(len + mhlen));
		m->m_next = m_copy(m0, off, len);
		if (m->m_next == 0) {
			error = ENOBUFS;	/* ??? */
			goto sendorfree;
		}
#ifndef sparc
		mhip->ip_off = htons((u_short)mhip->ip_off);
#endif
		mhip->ip_sum = 0;
		mhip->ip_sum = in_cksum(m, mhlen);
		*mnext = m;
		mnext = &m->m_act;
	}
	/*
	 * Update first fragment by trimming what's been copied out
	 * and updating header, then send each fragment (in order).
	 */
	m_adj(m0, hlen + firstlen - ip->ip_len);
	ip->ip_len = htons((u_short)(hlen + firstlen));
	ip->ip_off = htons((u_short)(ip->ip_off | IP_MF));
	ip->ip_sum = 0;
	ip->ip_sum = in_cksum(m0, hlen);
sendorfree:
	for (m = m0; m; m = m0) {
		m0 = m->m_act;
		m->m_act = 0;
		if (error == 0)
#if BSD >= 199306
			error = (*ifp->if_output)(ifp, m,
			    (struct sockaddr *)dst, ro->ro_rt);
#else
			error = (*ifp->if_output)(ifp, m,
			    (struct sockaddr *)dst);
#endif
		else
			m_freem(m);
	}
    }	
done:
	if (ro->ro_rt) {
		RTFREE(ro->ro_rt);
	}
	return;
bad:
	m_freem(m);
	goto done;
}