-/* ip_input.c 1.10 81/10/31 */
+/* ip_input.c 1.56 82/10/23 */
#include "../h/param.h"
#include "../h/systm.h"
-#include "../h/clock.h"
#include "../h/mbuf.h"
-#include "../inet/cksum.h"
-#include "../inet/inet.h"
-#include "../inet/inet_systm.h"
-#include "../inet/imp.h"
-#include "../inet/ip.h" /* belongs before inet.h */
-#include "../inet/ip_icmp.h"
-#include "../inet/tcp.h"
-
-int nosum = 0;
-
-struct ip *ip_reass();
+#include "../h/protosw.h"
+#include "../h/socket.h"
+#include <errno.h>
+#include <time.h>
+#include "../h/kernel.h"
+
+#include "../net/if.h"
+#include "../net/route.h"
+#include "../netinet/in.h"
+#include "../netinet/in_pcb.h"
+#include "../netinet/in_systm.h"
+#include "../netinet/ip.h"
+#include "../netinet/ip_var.h"
+#include "../netinet/ip_icmp.h"
+#include "../netinet/tcp.h"
+
+u_char ip_protox[IPPROTO_MAX];
+int ipqmaxlen = IFQ_MAXLEN;
+struct ifnet *ifinet; /* first inet interface */
/*
- * Ip input routines.
+ * IP initialization: fill in IP protocol switch table.
+ * All protocols not implemented in kernel go to raw IP protocol handler.
*/
+ip_init()
+{
+ register struct protosw *pr;
+ register int i;
+
+ 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 &&
+ pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW)
+ ip_protox[pr->pr_protocol] = pr - protosw;
+ ipq.next = ipq.prev = &ipq;
+ 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
* try to reassamble. If complete and fragment queue exists, discard.
* Process options. Pass to next level.
*/
-ip_input(m0)
- struct mbuf *m0;
+ipintr()
{
- register struct ip *ip; /* known to be r11 in CKSUM below */
- register struct mbuf *m = m0;
+ register struct ip *ip;
+ register struct mbuf *m;
+ struct mbuf *m0;
register int i;
- register struct ipq *q;
register struct ipq *fp;
- int hlen;
+ int hlen, s;
-COUNT(IP_INPUT);
+next:
/*
- * Check header and byteswap.
+ * Get next datagram off input queue and get IP header
+ * in first mbuf.
*/
+ s = splimp();
+ IF_DEQUEUE(&ipintrq, m);
+ splx(s);
+ if (m == 0)
+ return;
+ if ((m->m_off > MMAXOFF || m->m_len < sizeof (struct ip)) &&
+ (m = m_pullup(m, sizeof (struct ip))) == 0)
+ return;
ip = mtod(m, struct ip *);
if ((hlen = ip->ip_hl << 2) > m->m_len) {
- printf("ip hdr ovflo\n");
- m_freem(m);
- return;
- }
- CKSUM_IPCHK(m, ip, r11, hlen);
- if (ip->ip_sum) {
- printf("ip_sum %x\n", ip->ip_sum);
- netstat.ip_badsum++;
- if (!nosum) {
- m_freem(m);
+ if ((m = m_pullup(m, hlen)) == 0)
return;
- }
+ ip = mtod(m, struct ip *);
}
- ip->ip_len = ntohs(ip->ip_len);
+ 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 || pdp11 || ns16032
+ /*
+ * Convert fields to host representation.
+ */
+ ip->ip_len = ntohs((u_short)ip->ip_len);
ip->ip_id = ntohs(ip->ip_id);
- ip->ip_off = ntohs(ip->ip_off);
+ ip->ip_off = ntohs((u_short)ip->ip_off);
+#endif
/*
* Check that the amount of data in the buffers
* Trim mbufs if longer than we expect.
* Drop packet if shorter than we expect.
*/
- i = 0;
- for (; m != NULL; m = m->m_next)
+ i = -ip->ip_len;
+ m0 = m;
+ for (;;) {
i += m->m_len;
- m = m0;
- if (i != ip->ip_len) {
- if (i < ip->ip_len) {
- printf("ip_input: short packet\n");
- m_freem(m);
- return;
+ if (m->m_next == 0)
+ break;
+ m = m->m_next;
+ }
+ if (i != 0) {
+ if (i < 0) {
+ ipstat.ips_tooshort++;
+ 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, hlen);
- if (ip->ip_dst.s_addr != n_lhost.s_addr) {
- if (--ip->ip_ttl == 0) {
- icmp_error(ip, ICMP_TIMXCEED);
- return;
- }
- ip_output(dtom(ip));
- return;
+ 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.
* convert offset of this to bytes.
*/
ip->ip_len -= hlen;
- ip->ip_mff = 0;
+ ((struct ipasfrag *)ip)->ipf_mff = 0;
if (ip->ip_off & IP_MF)
- ip->ip_mff = 1;
+ ((struct ipasfrag *)ip)->ipf_mff = 1;
ip->ip_off <<= 3;
/*
* or if this is not the first fragment,
* attempt reassembly; if it succeeds, proceed.
*/
- if (ip->ip_mff || ip->ip_off) {
- ip = ip_reass(ip, fp);
+ if (((struct ipasfrag *)ip)->ipf_mff || ip->ip_off) {
+ ip = ip_reass((struct ipasfrag *)ip, fp);
if (ip == 0)
- return;
+ goto next;
hlen = ip->ip_hl << 2;
m = dtom(ip);
} else
(void) ip_freef(fp);
/*
- * Switch out to protocol specific routine.
- * SHOULD GO THROUGH PROTOCOL SWITCH TABLE
+ * Switch out to protocol's input routine.
*/
- switch (ip->ip_p) {
-
- case IPPROTO_ICMP:
- icmp_input(m);
- break;
-
- case IPPROTO_TCP:
- if (hlen > sizeof (struct ip))
- ip_stripoptions(ip, hlen);
- tcp_input(m);
- break;
-
- case IPPROTO_UDP:
- if (hlen > sizeof (struct ip))
- ip_stripoptions(ip, hlen);
- udp_input(m);
- break;
-
- default:
- raw_input(m);
- break;
- }
+ (*protosw[ip_protox[ip->ip_p]].pr_input)(m);
+ goto next;
+bad:
+ m_freem(m);
+ goto next;
}
/*
* Take incoming datagram fragment and try to
- * reassamble it into whole datagram. If a chain for
+ * reassemble it into whole datagram. If a chain for
* reassembly of this datagram already exists, then it
* is given as fp; otherwise have to make a chain.
*/
struct ip *
ip_reass(ip, fp)
- register struct ip *ip;
+ register struct ipasfrag *ip;
register struct ipq *fp;
{
register struct mbuf *m = dtom(ip);
- register struct ip *q;
+ register struct ipasfrag *q;
struct mbuf *t;
int hlen = ip->ip_hl << 2;
int i, next;
* If first fragment to arrive, create a reassembly queue.
*/
if (fp == 0) {
- if ((t = m_get(1)) == NULL)
+ if ((t = m_get(M_WAIT)) == NULL)
goto dropfrag;
- t->m_off = MMINOFF;
fp = mtod(t, struct ipq *);
insque(fp, &ipq);
fp->ipq_ttl = IPFRAGTTL;
fp->ipq_p = ip->ip_p;
fp->ipq_id = ip->ip_id;
- fp->ipq_next = fp->ipq_prev = (struct ip *)fp;
- fp->ipq_src = ip->ip_src;
- fp->ipq_dst = ip->ip_dst;
+ fp->ipq_next = fp->ipq_prev = (struct ipasfrag *)fp;
+ fp->ipq_src = ((struct ip *)ip)->ip_src;
+ fp->ipq_dst = ((struct ip *)ip)->ip_dst;
+ q = (struct ipasfrag *)fp;
+ goto insert;
}
/*
* Find a segment which begins after this one does.
*/
- for (q = fp->ipq_next; q != (struct ip *)fp; q = q->ip_next)
+ for (q = fp->ipq_next; q != (struct ipasfrag *)fp; q = q->ipf_next)
if (q->ip_off > ip->ip_off)
break;
* our data already. If so, drop the data from the incoming
* segment. If it provides all of our data, drop us.
*/
- if (q->ip_prev != (struct ip *)fp) {
- i = q->ip_prev->ip_off + q->ip_prev->ip_len - ip->ip_off;
+ if (q->ipf_prev != (struct ipasfrag *)fp) {
+ i = q->ipf_prev->ip_off + q->ipf_prev->ip_len - ip->ip_off;
if (i > 0) {
if (i >= ip->ip_len)
goto dropfrag;
* While we overlap succeeding segments trim them or,
* if they are completely covered, dequeue them.
*/
- while (q != (struct ip *)fp && ip->ip_off + ip->ip_len > q->ip_off) {
+ while (q != (struct ipasfrag *)fp && ip->ip_off + ip->ip_len > q->ip_off) {
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;
}
- q = q->ip_next;
- m_freem(dtom(q->ip_prev));
- ip_deq(q->ip_prev);
+ q = q->ipf_next;
+ m_freem(dtom(q->ipf_prev));
+ ip_deq(q->ipf_prev);
}
+insert:
/*
* Stick new segment in its place;
* check for complete reassembly.
*/
- ip_enq(ip, q->ip_prev);
+ ip_enq(ip, q->ipf_prev);
next = 0;
- for (q = fp->ipq_next; q != (struct ip *)fp; q = q->ip_next) {
+ for (q = fp->ipq_next; q != (struct ipasfrag *)fp; q = q->ipf_next) {
if (q->ip_off != next)
return (0);
next += q->ip_len;
}
- if (q->ip_prev->ip_mff)
+ if (q->ipf_prev->ipf_mff)
return (0);
/*
t = m->m_next;
m->m_next = 0;
m_cat(m, t);
- while ((q = q->ip_next) != (struct ip *)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
*/
ip = fp->ipq_next;
ip->ip_len = next;
- ip->ip_src = fp->ipq_src;
- ip->ip_dst = fp->ipq_dst;
+ ((struct ip *)ip)->ip_src = fp->ipq_src;
+ ((struct ip *)ip)->ip_dst = fp->ipq_dst;
remque(fp);
- m_free(dtom(fp));
+ (void) m_free(dtom(fp));
m = dtom(ip);
- m->m_len += sizeof (struct ip);
- m->m_off -= sizeof (struct ip);
- return (ip);
+ m->m_len += sizeof (struct ipasfrag);
+ m->m_off -= sizeof (struct ipasfrag);
+ return ((struct ip *)ip);
dropfrag:
m_freem(m);
ip_freef(fp)
struct ipq *fp;
{
- register struct ip *q;
+ register struct ipasfrag *q;
struct mbuf *m;
- for (q = fp->ipq_next; q != (struct ip *)fp; q = q->ip_next)
+ for (q = fp->ipq_next; q != (struct ipasfrag *)fp; q = q->ipf_next)
m_freem(dtom(q));
m = dtom(fp);
fp = fp->next;
remque(fp->prev);
- m_free(m);
+ (void) m_free(m);
return (fp);
}
* Like insque, but pointers in middle of structure.
*/
ip_enq(p, prev)
- register struct ip *p;
- register struct ip *prev;
+ register struct ipasfrag *p, *prev;
{
-COUNT(IP_ENQ);
- p->ip_prev = prev;
- p->ip_next = prev->ip_next;
- prev->ip_next->ip_prev = p;
- prev->ip_next = p;
+ p->ipf_prev = prev;
+ p->ipf_next = prev->ipf_next;
+ prev->ipf_next->ipf_prev = p;
+ prev->ipf_next = p;
}
/*
* To ip_enq as remque is to insque.
*/
ip_deq(p)
- register struct ip *p;
+ register struct ipasfrag *p;
{
-COUNT(IP_DEQ);
- p->ip_prev->ip_next = p->ip_next;
- p->ip_next->ip_prev = p->ip_prev;
+ p->ipf_prev->ipf_next = p->ipf_next;
+ p->ipf_next->ipf_prev = p->ipf_prev;
}
/*
* if a timer expires on a reassembly
* queue, discard it.
*/
-ip_timeo()
+ip_slowtimo()
{
- register struct ip *q;
register struct ipq *fp;
int s = splnet();
-COUNT(IP_TIMEO);
- for (fp = ipq.next; fp != &ipq; )
+ 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;
- timeout(ip_timeo, 0, hz);
splx(s);
}
+/*
+ * Drain off all datagram fragments.
+ */
+ip_drain()
+{
+
+ while (ipq.next != &ipq)
+ (void) ip_freef(ipq.next);
+}
+
/*
* Do option processing on a datagram,
* possibly discarding it if bad options
struct ip *ip;
{
register u_char *cp;
- int opt, optlen, cnt, s;
- struct socket *sp;
+ int opt, optlen, cnt, code, type;
+ struct in_addr *sin;
+ register struct ip_timestamp *ipt;
+ register struct ifnet *ifp;
+ struct in_addr t;
cp = (u_char *)(ip + 1);
cnt = (ip->ip_hl << 2) - sizeof (struct ip);
default:
break;
+ /*
+ * Source routing with record.
+ * Find interface with current destination address.
+ * If none on this machine then drop if strictly routed,
+ * or do nothing if loosely routed.
+ * Record interface address and bring up next address
+ * component. If strictly routed make sure next
+ * address on directly accessible net.
+ */
case IPOPT_LSRR:
case IPOPT_SSRR:
- if (cp[2] < 4 || cp[2] > optlen - 3)
+ if (cp[2] < 4 || cp[2] > optlen - (sizeof (long) - 1))
+ break;
+ sin = (struct in_addr *)(cp + cp[2]);
+ 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;
break;
- sp = (struct socket *)(cp+cp[2]);
- if (n_lhost.s_addr == *(u_long *)sp) {
- if (opt == IPOPT_SSRR) {
- /* make sure *sp directly accessible*/
- }
- ip->ip_dst = *sp;
- *sp = n_lhost;
- cp[2] += 4;
}
+ t = ip->ip_dst; ip->ip_dst = *sin; *sin = t;
+ cp[2] += 4;
+ if (cp[2] > optlen - (sizeof (long) - 1))
+ break;
+ ip->ip_dst = sin[1];
+ if (opt == IPOPT_SSRR &&
+ if_ifonnetof(in_netof(ip->ip_dst)) == 0)
+ goto bad;
break;
case IPOPT_TS:
- if (cp[2] < 5)
+ code = cp - (u_char *)ip;
+ type = ICMP_PARAMPROB;
+ ipt = (struct ip_timestamp *)cp;
+ if (ipt->ipt_len < 5)
goto bad;
- if (cp[2] > cp[1] - 3) {
- if ((cp[3] & 0xf0) == 0xf0)
+ if (ipt->ipt_ptr > ipt->ipt_len - sizeof (long)) {
+ if (++ipt->ipt_oflw == 0)
goto bad;
- cp[3] += 0x10;
break;
}
- sp = (struct socket *)(cp+cp[2]);
- switch (cp[3] & 0xf) {
+ sin = (struct in_addr *)(cp+cp[2]);
+ switch (ipt->ipt_flg) {
case IPOPT_TS_TSONLY:
break;
case IPOPT_TS_TSANDADDR:
- if (cp[2] > cp[1] - 7)
+ if (ipt->ipt_ptr + 8 > ipt->ipt_len)
goto bad;
+ if (ifinet == 0)
+ goto bad; /* ??? */
+ *sin++ = ((struct sockaddr_in *)&ifinet->if_addr)->sin_addr;
break;
case IPOPT_TS_PRESPEC:
- if (*(u_long *)sp != n_lhost.s_addr)
- break;
- if (cp[2] > cp[1] - 7)
+ ipaddr.sin_addr = *sin;
+ if (!if_ifwithaddr((struct sockaddr *)&ipaddr))
+ continue;
+ if (ipt->ipt_ptr + 8 > ipt->ipt_len)
goto bad;
- cp[1] += 4;
+ ipt->ipt_ptr += 4;
break;
default:
goto bad;
}
- s = spl6();
- *(int *)sp = (time % SECDAY) * 1000 + (lbolt*1000/hz);
- splx(s);
- cp[1] += 4;
+ *(n_time *)sin = iptime();
+ ipt->ipt_ptr += 4;
}
}
return (0);
bad:
- /* SHOULD FORCE ICMP MESSAGE */
- return (-1);
+ icmp_error(ip, type, code);
+ return (1);
}
/*
- * Strip out IP options, e.g. before passing
- * to higher level protocol in the kernel.
+ * Strip out IP options, at higher
+ * level protocol in the kernel.
+ * Second argument is buffer to which options
+ * will be moved, and return value is their length.
*/
-ip_stripoptions(ip)
+ip_stripoptions(ip, mopt)
struct ip *ip;
+ struct mbuf *mopt;
{
register int i;
register struct mbuf *m;
- char *op;
int olen;
-COUNT(IP_OPT);
olen = (ip->ip_hl<<2) - sizeof (struct ip);
- op = (caddr_t)ip + olen;
- m = dtom(++ip);
+ m = dtom(ip);
+ ip++;
+ if (mopt) {
+ mopt->m_len = olen;
+ mopt->m_off = MMINOFF;
+ bcopy((caddr_t)ip, mtod(m, caddr_t), (unsigned)olen);
+ }
i = m->m_len - (sizeof (struct ip) + olen);
- bcopy((caddr_t)ip+olen, (caddr_t)ip, i);
- m->m_len -= i;
+ bcopy((caddr_t)ip+olen, (caddr_t)ip, (unsigned)i);
+ m->m_len -= olen;
}
-/* stubs */
-
-icmp_error(ip, error)
+u_char inetctlerrmap[] = {
+ ECONNABORTED, ECONNABORTED, 0, 0,
+ 0, 0,
+ EHOSTDOWN, EHOSTUNREACH, ENETUNREACH, EHOSTUNREACH,
+ ECONNREFUSED, ECONNREFUSED, EMSGSIZE, 0,
+ 0, 0, 0, 0
+};
+
+ip_ctlinput(cmd, arg)
+ int cmd;
+ caddr_t arg;
{
+ struct in_addr *in;
+ int tcp_abort(), udp_abort();
+ extern struct inpcb tcb, udb;
- m_freem(dtom(ip));
+ 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 */
+ in_pcbnotify(&tcb, in, (int)inetctlerrmap[cmd], tcp_abort);
+ in_pcbnotify(&udb, in, (int)inetctlerrmap[cmd], udp_abort);
}
-icmp_input(m)
- struct mbuf *m;
+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);
+ if (mopt == 0) {
+ m_freem(dtom(ip));
+ return;
+ }
- printf("icmp_input %x\n", m);
-}
+ /*
+ * 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);
-udp_input(m)
- struct mbuf *m;
-{
+ /* last 0 here means no directed broadcast */
+ if ((error = ip_output(dtom(ip), mopt, (struct route *)0, 0)) == 0) {
+ if (mcopy)
+ m_freem(mcopy);
+ return;
+ }
+ ip = mtod(mcopy, struct ip *);
+ type = ICMP_UNREACH, code = 0; /* need ``undefined'' */
+ switch (error) {
- printf("udp_input %x\n", m);
-}
+ case ENETUNREACH:
+ case ENETDOWN:
+ code = ICMP_UNREACH_NET;
+ break;
-raw_input(m)
- struct mbuf *m;
-{
+ case EMSGSIZE:
+ code = ICMP_UNREACH_NEEDFRAG;
+ break;
- printf("raw_input %x\n", m);
+ 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);
}