X-Git-Url: https://git.subgeniuskitty.com/unix-history/.git/blobdiff_plain/e6b33a0337d0d220510616ff5ecf3b428600ed0f..36bb5f9457b331be70c5595ae9f76489369de5ed:/usr/src/sys/netinet/ip_input.c

diff --git a/usr/src/sys/netinet/ip_input.c b/usr/src/sys/netinet/ip_input.c
index f0e6057595..1867ef30ed 100644
--- a/usr/src/sys/netinet/ip_input.c
+++ b/usr/src/sys/netinet/ip_input.c
@@ -1,20 +1,29 @@
-/*	ip_input.c	1.30	82/02/15	*/
-
-#include "../h/param.h"
-#include "../h/systm.h"
-#include "../h/clock.h"
-#include "../h/mbuf.h"
-#include "../h/protosw.h"
-#include "../h/socket.h"
-#include "../net/in.h"
-#include "../net/in_systm.h"
+/*	ip_input.c	6.8	84/12/20	*/
+
+#include "param.h"
+#include "systm.h"
+#include "mbuf.h"
+#include "domain.h"
+#include "protosw.h"
+#include "socket.h"
+#include "errno.h"
+#include "time.h"
+#include "kernel.h"
+
 #include "../net/if.h"
-#include "../net/ip.h"			/* belongs before in.h */
-#include "../net/ip_var.h"
-#include "../net/ip_icmp.h"
-#include "../net/tcp.h"
+#include "../net/route.h"
+
+#include "in.h"
+#include "in_pcb.h"
+#include "in_systm.h"
+#include "ip.h"
+#include "ip_var.h"
+#include "ip_icmp.h"
+#include "tcp.h"
 
 u_char	ip_protox[IPPROTO_MAX];
+int	ipqmaxlen = IFQ_MAXLEN;
+struct	ifnet *ifinet;			/* first inet interface */
 
 /*
  * IP initialization: fill in IP protocol switch table.
@@ -25,22 +34,25 @@ ip_init()
 	register struct protosw *pr;
 	register int i;
 
-COUNT(IP_INIT);
 	pr = pffindproto(PF_INET, IPPROTO_RAW);
 	if (pr == 0)
 		panic("ip_init");
 	for (i = 0; i < IPPROTO_MAX; i++)
-		ip_protox[i] = pr - protosw;
-	for (pr = protosw; pr <= protoswLAST; pr++)
-		if (pr->pr_family == PF_INET &&
+		ip_protox[i] = pr - inetsw;
+	for (pr = inetdomain.dom_protosw;
+	    pr < inetdomain.dom_protoswNPROTOSW; pr++)
+		if (pr->pr_domain->dom_family == PF_INET &&
 		    pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW)
-			ip_protox[pr->pr_protocol] = pr - protosw;
+			ip_protox[pr->pr_protocol] = pr - inetsw;
 	ipq.next = ipq.prev = &ipq;
-	ip_id = time & 0xffff;
+	ip_id = time.tv_sec & 0xffff;
+	ipintrq.ifq_maxlen = ipqmaxlen;
+	ifinet = if_ifwithaf(AF_INET);
 }
 
 u_char	ipcksum = 1;
 struct	ip *ip_reass();
+struct	sockaddr_in ipaddr = { AF_INET };
 
 /*
  * Ip input routine.  Checksum and byte swap header.  If fragmented
@@ -51,12 +63,11 @@ ipintr()
 {
 	register struct ip *ip;
 	register struct mbuf *m;
-	struct mbuf *m0, *mopt;
+	struct mbuf *m0;
 	register int i;
 	register struct ipq *fp;
 	int hlen, s;
 
-COUNT(IPINTR);
 next:
 	/*
 	 * Get next datagram off input queue and get IP header
@@ -68,29 +79,34 @@ next:
 	if (m == 0)
 		return;
 	if ((m->m_off > MMAXOFF || m->m_len < sizeof (struct ip)) &&
-	    (m = m_pullup(m, sizeof (struct ip))) == 0)
-		return;
+	    (m = m_pullup(m, sizeof (struct ip))) == 0) {
+		ipstat.ips_toosmall++;
+		goto next;
+	}
 	ip = mtod(m, struct ip *);
 	if ((hlen = ip->ip_hl << 2) > m->m_len) {
-		if ((m = m_pullup(m, hlen)) == 0)
-			return;
+		if ((m = m_pullup(m, hlen)) == 0) {
+			ipstat.ips_badhlen++;
+			goto next;
+		}
 		ip = mtod(m, struct ip *);
 	}
 	if (ipcksum)
 		if (ip->ip_sum = in_cksum(m, hlen)) {
-			printf("ip_sum %x\n", ip->ip_sum);	/* XXX */
 			ipstat.ips_badsum++;
 			goto bad;
 		}
 
-#if vax
 	/*
 	 * Convert fields to host representation.
 	 */
 	ip->ip_len = ntohs((u_short)ip->ip_len);
+	if (ip->ip_len < hlen) {
+		ipstat.ips_badlen++;
+		goto bad;
+	}
 	ip->ip_id = ntohs(ip->ip_id);
 	ip->ip_off = ntohs((u_short)ip->ip_off);
-#endif
 
 	/*
 	 * Check that the amount of data in the buffers
@@ -98,40 +114,58 @@ next:
 	 * Trim mbufs if longer than we expect.
 	 * Drop packet if shorter than we expect.
 	 */
-	i = 0;
+	i = -ip->ip_len;
 	m0 = m;
-	for (; m != NULL; m = m->m_next)
+	for (;;) {
 		i += m->m_len;
-	m = m0;
-	if (i != ip->ip_len) {
-		if (i < ip->ip_len) {
+		if (m->m_next == 0)
+			break;
+		m = m->m_next;
+	}
+	if (i != 0) {
+		if (i < 0) {
 			ipstat.ips_tooshort++;
+			m = m0;
 			goto bad;
 		}
-		m_adj(m, ip->ip_len - i);
+		if (i <= m->m_len)
+			m->m_len -= i;
+		else
+			m_adj(m0, -i);
 	}
+	m = m0;
 
 	/*
 	 * Process options and, if not destined for us,
-	 * ship it on.
+	 * ship it on.  ip_dooptions returns 1 when an
+	 * error was detected (causing an icmp message
+	 * to be sent).
 	 */
-	if (hlen > sizeof (struct ip))
-		ip_dooptions(ip);
-	if (ifnet && ip->ip_dst.s_addr != ifnet->if_addr.s_addr &&
-	    if_ifwithaddr(ip->ip_dst) == 0) {
-printf("ip->ip_dst %x ip->ip_ttl %x\n", ip->ip_dst, ip->ip_ttl);
-		if (--ip->ip_ttl == 0) {
-			icmp_error(ip, ICMP_TIMXCEED, 0);
-			goto next;
-		}
-		mopt = m_get(M_DONTWAIT);
-		if (mopt == 0)
-			goto bad;
-		ip_stripoptions(ip, mopt);
-		(void) ip_output(m0, mopt);
+	if (hlen > sizeof (struct ip) && ip_dooptions(ip))
+		goto next;
+
+	/*
+	 * Fast check on the first internet
+	 * interface in the list.
+	 */
+	if (ifinet) {
+		struct sockaddr_in *sin;
+
+		sin = (struct sockaddr_in *)&ifinet->if_addr;
+		if (sin->sin_addr.s_addr == ip->ip_dst.s_addr)
+			goto ours;
+		sin = (struct sockaddr_in *)&ifinet->if_broadaddr;
+		if ((ifinet->if_flags & IFF_BROADCAST) &&
+		    sin->sin_addr.s_addr == ip->ip_dst.s_addr)
+			goto ours;
+	}
+	ipaddr.sin_addr = ip->ip_dst;
+	if (if_ifwithaddr((struct sockaddr *)&ipaddr) == 0) {
+		ip_forward(ip);
 		goto next;
 	}
 
+ours:
 	/*
 	 * Look for queue of fragments
 	 * of this datagram.
@@ -169,12 +203,12 @@ found:
 		m = dtom(ip);
 	} else
 		if (fp)
-			(void) ip_freef(fp);
+			ip_freef(fp);
 
 	/*
 	 * Switch out to protocol's input routine.
 	 */
-	(*protosw[ip_protox[ip->ip_p]].pr_input)(m);
+	(*inetsw[ip_protox[ip->ip_p]].pr_input)(m);
 	goto next;
 bad:
 	m_freem(m);
@@ -197,7 +231,6 @@ ip_reass(ip, fp)
 	struct mbuf *t;
 	int hlen = ip->ip_hl << 2;
 	int i, next;
-COUNT(IP_REASS);
 
 	/*
 	 * Presence of header sizes in mbufs
@@ -210,9 +243,8 @@ COUNT(IP_REASS);
 	 * If first fragment to arrive, create a reassembly queue.
 	 */
 	if (fp == 0) {
-		if ((t = m_get(M_WAIT)) == NULL)
+		if ((t = m_get(M_WAIT, MT_FTABLE)) == NULL)
 			goto dropfrag;
-		t->m_off = MMINOFF;
 		fp = mtod(t, struct ipq *);
 		insque(fp, &ipq);
 		fp->ipq_ttl = IPFRAGTTL;
@@ -256,6 +288,7 @@ COUNT(IP_REASS);
 		i = (ip->ip_off + ip->ip_len) - q->ip_off;
 		if (i < q->ip_len) {
 			q->ip_len -= i;
+			q->ip_off += i;
 			m_adj(dtom(q), i);
 			break;
 		}
@@ -287,8 +320,12 @@ insert:
 	t = m->m_next;
 	m->m_next = 0;
 	m_cat(m, t);
-	while ((q = q->ipf_next) != (struct ipasfrag *)fp)
-		m_cat(m, dtom(q));
+	q = q->ipf_next;
+	while (q != (struct ipasfrag *)fp) {
+		t = dtom(q);
+		q = q->ipf_next;
+		m_cat(m, t);
+	}
 
 	/*
 	 * Create header for new ip packet by
@@ -316,21 +353,18 @@ dropfrag:
  * Free a fragment reassembly header and all
  * associated datagrams.
  */
-struct ipq *
 ip_freef(fp)
 	struct ipq *fp;
 {
-	register struct ipasfrag *q;
-	struct mbuf *m;
-COUNT(IP_FREEF);
+	register struct ipasfrag *q, *p;
 
-	for (q = fp->ipq_next; q != (struct ipasfrag *)fp; q = q->ipf_next)
+	for (q = fp->ipq_next; q != (struct ipasfrag *)fp; q = p) {
+		p = q->ipf_next;
+		ip_deq(q);
 		m_freem(dtom(q));
-	m = dtom(fp);
-	fp = fp->next;
-	remque(fp->prev);
-	(void) m_free(m);
-	return (fp);
+	}
+	remque(fp);
+	(void) m_free(dtom(fp));
 }
 
 /*
@@ -341,7 +375,6 @@ ip_enq(p, prev)
 	register struct ipasfrag *p, *prev;
 {
 
-COUNT(IP_ENQ);
 	p->ipf_prev = prev;
 	p->ipf_next = prev->ipf_next;
 	prev->ipf_next->ipf_prev = p;
@@ -355,7 +388,6 @@ ip_deq(p)
 	register struct ipasfrag *p;
 {
 
-COUNT(IP_DEQ);
 	p->ipf_prev->ipf_next = p->ipf_next;
 	p->ipf_next->ipf_prev = p->ipf_prev;
 }
@@ -370,17 +402,17 @@ ip_slowtimo()
 	register struct ipq *fp;
 	int s = splnet();
 
-COUNT(IP_SLOWTIMO);
 	fp = ipq.next;
 	if (fp == 0) {
 		splx(s);
 		return;
 	}
-	while (fp != &ipq)
-		if (--fp->ipq_ttl == 0)
-			fp = ip_freef(fp);
-		else
-			fp = fp->next;
+	while (fp != &ipq) {
+		--fp->ipq_ttl;
+		fp = fp->next;
+		if (fp->prev->ipq_ttl == 0)
+			ip_freef(fp->prev);
+	}
 	splx(s);
 }
 
@@ -390,9 +422,8 @@ COUNT(IP_SLOWTIMO);
 ip_drain()
 {
 
-COUNT(IP_DRAIN);
 	while (ipq.next != &ipq)
-		(void) ip_freef(ipq.next);
+		ip_freef(ipq.next);
 }
 
 /*
@@ -404,13 +435,12 @@ ip_dooptions(ip)
 	struct ip *ip;
 {
 	register u_char *cp;
-	int opt, optlen, cnt;
+	int opt, optlen, cnt, code, type;
 	struct in_addr *sin;
 	register struct ip_timestamp *ipt;
 	register struct ifnet *ifp;
 	struct in_addr t;
 
-COUNT(IP_DOOPTIONS);
 	cp = (u_char *)(ip + 1);
 	cnt = (ip->ip_hl << 2) - sizeof (struct ip);
 	for (; cnt > 0; cnt -= optlen, cp += optlen) {
@@ -419,8 +449,11 @@ COUNT(IP_DOOPTIONS);
 			break;
 		if (opt == IPOPT_NOP)
 			optlen = 1;
-		else
+		else {
 			optlen = cp[1];
+			if (optlen <= 0 || optlen >= cnt)
+				goto bad;
+		}
 		switch (opt) {
 
 		default:
@@ -436,10 +469,13 @@ COUNT(IP_DOOPTIONS);
 		 * address on directly accessible net.
 		 */
 		case IPOPT_LSRR:
+		case IPOPT_SSRR:
 			if (cp[2] < 4 || cp[2] > optlen - (sizeof (long) - 1))
 				break;
 			sin = (struct in_addr *)(cp + cp[2]);
-			ifp = if_ifwithaddr(*sin);
+			ipaddr.sin_addr = *sin;
+			ifp = if_ifwithaddr((struct sockaddr *)&ipaddr);
+			type = ICMP_UNREACH, code = ICMP_UNREACH_SRCFAIL;
 			if (ifp == 0) {
 				if (opt == IPOPT_SSRR)
 					goto bad;
@@ -450,11 +486,14 @@ COUNT(IP_DOOPTIONS);
 			if (cp[2] > optlen - (sizeof (long) - 1))
 				break;
 			ip->ip_dst = sin[1];
-			if (opt == IPOPT_SSRR && if_ifonnetof(ip->ip_dst)==0)
+			if (opt == IPOPT_SSRR &&
+			    if_ifonnetof(in_netof(ip->ip_dst)) == 0)
 				goto bad;
 			break;
 
 		case IPOPT_TS:
+			code = cp - (u_char *)ip;
+			type = ICMP_PARAMPROB;
 			ipt = (struct ip_timestamp *)cp;
 			if (ipt->ipt_len < 5)
 				goto bad;
@@ -472,12 +511,14 @@ COUNT(IP_DOOPTIONS);
 			case IPOPT_TS_TSANDADDR:
 				if (ipt->ipt_ptr + 8 > ipt->ipt_len)
 					goto bad;
-				/* stamp with ``first'' interface address */
-				*sin++ = ifnet->if_addr;
+				if (ifinet == 0)
+					goto bad;	/* ??? */
+				*sin++ = ((struct sockaddr_in *)&ifinet->if_addr)->sin_addr;
 				break;
 
 			case IPOPT_TS_PRESPEC:
-				if (if_ifwithaddr(*sin) == 0)
+				ipaddr.sin_addr = *sin;
+				if (if_ifwithaddr((struct sockaddr *)&ipaddr) == 0)
 					continue;
 				if (ipt->ipt_ptr + 8 > ipt->ipt_len)
 					goto bad;
@@ -491,10 +532,10 @@ COUNT(IP_DOOPTIONS);
 			ipt->ipt_ptr += 4;
 		}
 	}
-	return;
+	return (0);
 bad:
-	/* SHOULD FORCE ICMP MESSAGE */
-	return;
+	icmp_error(ip, type, code);
+	return (1);
 }
 
 /*
@@ -510,7 +551,6 @@ ip_stripoptions(ip, mopt)
 	register int i;
 	register struct mbuf *m;
 	int olen;
-COUNT(IP_STRIPOPTIONS);
 
 	olen = (ip->ip_hl<<2) - sizeof (struct ip);
 	m = dtom(ip);
@@ -518,9 +558,123 @@ COUNT(IP_STRIPOPTIONS);
 	if (mopt) {
 		mopt->m_len = olen;
 		mopt->m_off = MMINOFF;
-		bcopy((caddr_t)ip, mtod(m, caddr_t), (unsigned)olen);
+		bcopy((caddr_t)ip, mtod(mopt, caddr_t), (unsigned)olen);
 	}
 	i = m->m_len - (sizeof (struct ip) + olen);
 	bcopy((caddr_t)ip+olen, (caddr_t)ip, (unsigned)i);
 	m->m_len -= olen;
 }
+
+u_char inetctlerrmap[PRC_NCMDS] = {
+	ECONNABORTED,	ECONNABORTED,	0,		0,
+	0,		0,		EHOSTDOWN,	EHOSTUNREACH,
+	ENETUNREACH,	EHOSTUNREACH,	ECONNREFUSED,	ECONNREFUSED,
+	EMSGSIZE,	0,		0,		0,
+	0,		0,		0,		0
+};
+
+ip_ctlinput(cmd, arg)
+	int cmd;
+	caddr_t arg;
+{
+	struct in_addr *in;
+	int in_rtchange(), tcp_abort(), udp_abort();
+	extern struct inpcb tcb, udb;
+
+	if (cmd < 0 || cmd > PRC_NCMDS)
+		return;
+	if (inetctlerrmap[cmd] == 0)
+		return;		/* XXX */
+	if (cmd == PRC_IFDOWN)
+		in = &((struct sockaddr_in *)arg)->sin_addr;
+	else if (cmd == PRC_HOSTDEAD || cmd == PRC_HOSTUNREACH)
+		in = (struct in_addr *)arg;
+	else
+		in = &((struct icmp *)arg)->icmp_ip.ip_dst;
+	/* THIS IS VERY QUESTIONABLE, SHOULD HIT ALL PROTOCOLS */
+	if (cmd == PRC_REDIRECT_NET || cmd == PRC_REDIRECT_HOST) {
+		in_pcbnotify(&tcb, in, 0, in_rtchange);
+		in_pcbnotify(&udb, in, 0, in_rtchange);
+	} else {
+		in_pcbnotify(&tcb, in, (int)inetctlerrmap[cmd], tcp_abort);
+		in_pcbnotify(&udb, in, (int)inetctlerrmap[cmd], udp_abort);
+	}
+}
+
+int	ipprintfs = 0;
+int	ipforwarding = 1;
+/*
+ * Forward a packet.  If some error occurs return the sender
+ * and icmp packet.  Note we can't always generate a meaningful
+ * icmp message because icmp doesn't have a large enough repetoire
+ * of codes and types.
+ */
+ip_forward(ip)
+	register struct ip *ip;
+{
+	register int error, type, code;
+	struct mbuf *mopt, *mcopy;
+
+	if (ipprintfs)
+		printf("forward: src %x dst %x ttl %x\n", ip->ip_src,
+			ip->ip_dst, ip->ip_ttl);
+	if (ipforwarding == 0) {
+		/* can't tell difference between net and host */
+		type = ICMP_UNREACH, code = ICMP_UNREACH_NET;
+		goto sendicmp;
+	}
+	if (ip->ip_ttl < IPTTLDEC) {
+		type = ICMP_TIMXCEED, code = ICMP_TIMXCEED_INTRANS;
+		goto sendicmp;
+	}
+	ip->ip_ttl -= IPTTLDEC;
+	mopt = m_get(M_DONTWAIT, MT_DATA);
+	if (mopt == NULL) {
+		m_freem(dtom(ip));
+		return;
+	}
+
+	/*
+	 * Save at most 64 bytes of the packet in case
+	 * we need to generate an ICMP message to the src.
+	 */
+	mcopy = m_copy(dtom(ip), 0, imin(ip->ip_len, 64));
+	ip_stripoptions(ip, mopt);
+
+	error = ip_output(dtom(ip), mopt, (struct route *)0, IP_FORWARDING);
+	if (error == 0) {
+		if (mcopy)
+			m_freem(mcopy);
+		return;
+	}
+	if (mcopy == NULL)
+		return;
+	ip = mtod(mcopy, struct ip *);
+	type = ICMP_UNREACH, code = 0;		/* need ``undefined'' */
+	switch (error) {
+
+	case ENETUNREACH:
+	case ENETDOWN:
+		code = ICMP_UNREACH_NET;
+		break;
+
+	case EMSGSIZE:
+		code = ICMP_UNREACH_NEEDFRAG;
+		break;
+
+	case EPERM:
+		code = ICMP_UNREACH_PORT;
+		break;
+
+	case ENOBUFS:
+		type = ICMP_SOURCEQUENCH;
+		break;
+
+	case EHOSTDOWN:
+	case EHOSTUNREACH:
+		code = ICMP_UNREACH_HOST;
+		break;
+	}
+sendicmp:
+	icmp_error(ip, type, code);
+}