Please note that diffs are not public domain; they are subject to the copyright notices on the relevant files. =================================================================== RCS file: /ftp/cvs/cvsroot/src/sys/netinet/ip_input.c,v rcsdiff: /ftp/cvs/cvsroot/src/sys/netinet/ip_input.c,v: warning: Unknown phrases like `commitid ...;' are present. retrieving revision 1.206 retrieving revision 1.257 diff -u -p -r1.206 -r1.257 --- src/sys/netinet/ip_input.c 2004/12/15 04:25:19 1.206 +++ src/sys/netinet/ip_input.c 2007/12/20 19:53:32 1.257 @@ -1,4 +1,4 @@ -/* $NetBSD: ip_input.c,v 1.206 2004/12/15 04:25:19 thorpej Exp $ */ +/* $NetBSD: ip_input.c,v 1.257 2007/12/20 19:53:32 dyoung Exp $ */ /* * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project. @@ -98,7 +98,7 @@ */ #include -__KERNEL_RCSID(0, "$NetBSD: ip_input.c,v 1.206 2004/12/15 04:25:19 thorpej Exp $"); +__KERNEL_RCSID(0, "$NetBSD: ip_input.c,v 1.257 2007/12/20 19:53:32 dyoung Exp $"); #include "opt_inet.h" #include "opt_gateway.h" @@ -121,6 +121,7 @@ __KERNEL_RCSID(0, "$NetBSD: ip_input.c,v #include #include #include +#include #include #include @@ -131,6 +132,7 @@ __KERNEL_RCSID(0, "$NetBSD: ip_input.c,v #include #include #include +#include #include #include #include @@ -200,7 +202,6 @@ int ipprintfs = 0; #endif int ip_do_randomid = 0; -int ip_do_loopback_cksum = 0; /* * XXX - Setting ip_checkinterface mostly implements the receive side of @@ -223,7 +224,7 @@ struct rttimer_queue *ip_mtudisc_timeout int ipqmaxlen = IFQ_MAXLEN; u_long in_ifaddrhash; /* size of hash table - 1 */ int in_ifaddrentries; /* total number of addrs */ -struct in_ifaddrhead in_ifaddrhead; +struct in_ifaddrhead in_ifaddrhead; struct in_ifaddrhashhead *in_ifaddrhashtbl; u_long in_multihash; /* size of hash table - 1 */ int in_multientries; /* total number of addrs */ @@ -241,7 +242,7 @@ struct pfil_head inet_pfil_hook; * recalculate IP parameters derived from nmbclusters. */ static int ip_nmbclusters; /* copy of nmbclusters */ -static void ip_nmbclusters_changed __P((void)); /* recalc limits */ +static void ip_nmbclusters_changed(void); /* recalc limits */ #define CHECK_NMBCLUSTER_PARAMS() \ do { \ @@ -257,7 +258,7 @@ do { \ (((((x) & 0xF) | ((((x) >> 8) & 0xF) << 4)) ^ (y)) & IPREASS_HMASK) struct ipqhead ipq[IPREASS_NHASH]; int ipq_locked; -static int ip_nfragpackets; /* packets in reass queue */ +static int ip_nfragpackets; /* packets in reass queue */ static int ip_nfrags; /* total fragments in reass queues */ int ip_maxfragpackets = 200; /* limit on packets. XXX sysctl */ @@ -267,7 +268,7 @@ int ip_maxfrags; /* limit on fr /* * Additive-Increase/Multiplicative-Decrease (AIMD) strategy for * IP reassembly queue buffer managment. - * + * * We keep a count of total IP fragments (NB: not fragmented packets!) * awaiting reassembly (ip_nfrags) and a limit (ip_maxfrags) on fragments. * If ip_nfrags exceeds ip_maxfrags the limit, we drop half the @@ -275,15 +276,15 @@ int ip_maxfrags; /* limit on fr * repeatedly deleting single packets under heavy fragmentation load * (e.g., from lossy NFS peers). */ -static u_int ip_reass_ttl_decr __P((u_int ticks)); -static void ip_reass_drophalf __P((void)); +static u_int ip_reass_ttl_decr(u_int ticks); +static void ip_reass_drophalf(void); -static __inline int ipq_lock_try __P((void)); -static __inline void ipq_unlock __P((void)); +static inline int ipq_lock_try(void); +static inline void ipq_unlock(void); -static __inline int -ipq_lock_try() +static inline int +ipq_lock_try(void) { int s; @@ -301,8 +302,8 @@ ipq_lock_try() return (1); } -static __inline void -ipq_unlock() +static inline void +ipq_unlock(void) { int s; @@ -333,8 +334,10 @@ do { \ #define IPQ_UNLOCK() ipq_unlock() -POOL_INIT(inmulti_pool, sizeof(struct in_multi), 0, 0, 0, "inmltpl", NULL); -POOL_INIT(ipqent_pool, sizeof(struct ipqent), 0, 0, 0, "ipqepl", NULL); +POOL_INIT(inmulti_pool, sizeof(struct in_multi), 0, 0, 0, "inmltpl", NULL, + IPL_SOFTNET); +POOL_INIT(ipqent_pool, sizeof(struct ipqent), 0, 0, 0, "ipqepl", NULL, + IPL_VM); #ifdef INET_CSUM_COUNTERS #include @@ -373,11 +376,11 @@ static struct ip_srcrt { struct in_addr route[MAX_IPOPTLEN/sizeof(struct in_addr)]; } ip_srcrt; -static void save_rte __P((u_char *, struct in_addr)); +static void save_rte(u_char *, struct in_addr); #ifdef MBUFTRACE -struct mowner ip_rx_mowner = { "internet", "rx" }; -struct mowner ip_tx_mowner = { "internet", "tx" }; +struct mowner ip_rx_mowner = MOWNER_INIT("internet", "rx"); +struct mowner ip_tx_mowner = MOWNER_INIT("internet", "tx"); #endif /* @@ -395,7 +398,7 @@ ip_nmbclusters_changed(void) * All protocols not implemented in kernel go to raw IP protocol handler. */ void -ip_init() +ip_init(void) { const struct protosw *pr; int i; @@ -414,7 +417,7 @@ ip_init() for (i = 0; i < IPREASS_NHASH; i++) LIST_INIT(&ipq[i]); - ip_id = time.tv_sec & 0xfffff; + ip_id = time_second & 0xfffff; ipintrq.ifq_maxlen = ipqmaxlen; ip_nmbclusters_changed(); @@ -426,7 +429,7 @@ ip_init() M_WAITOK, &in_multihash); ip_mtudisc_timeout_q = rt_timer_queue_create(ip_mtudisc_timeout); #ifdef GATEWAY - ipflow_init(); + ipflow_init(ip_hashsize); #endif #ifdef PFIL_HOOKS @@ -445,25 +448,27 @@ ip_init() #endif /* MBUFTRACE */ } -struct sockaddr_in ipaddr = { sizeof(ipaddr), AF_INET }; +struct sockaddr_in ipaddr = { + .sin_len = sizeof(ipaddr), + .sin_family = AF_INET, +}; struct route ipforward_rt; /* * IP software interrupt routine */ void -ipintr() +ipintr(void) { int s; struct mbuf *m; - while (1) { + while (!IF_IS_EMPTY(&ipintrq)) { s = splnet(); IF_DEQUEUE(&ipintrq, m); splx(s); if (m == 0) return; - MCLAIM(m, &ip_rx_mowner); ip_input(m); } } @@ -484,12 +489,13 @@ ip_input(struct mbuf *m) int downmatch; int checkif; int srcrt = 0; + int s; u_int hash; #ifdef FAST_IPSEC struct m_tag *mtag; struct tdb_ident *tdbi; struct secpolicy *sp; - int s, error; + int error; #endif /* FAST_IPSEC */ MCLAIM(m, &ip_rx_mowner); @@ -511,7 +517,7 @@ ip_input(struct mbuf *m) * it. Otherwise, if it is aligned, make sure the entire * base IP header is in the first mbuf of the chain. */ - if (IP_HDR_ALIGNED_P(mtod(m, caddr_t)) == 0) { + if (IP_HDR_ALIGNED_P(mtod(m, void *)) == 0) { if ((m = m_copyup(m, sizeof(struct ip), (max_linkhdr + 3) & ~3)) == NULL) { /* XXXJRT new stat, please */ @@ -668,7 +674,7 @@ ip_input(struct mbuf *m) * default route for hosts on 1.1.1.0/24. Of course this * also requires a "map tlp0 ..." to complete the story. * One might argue whether or not this kind of network config. - * should be supported in this manner... + * should be supported in this manner... */ srcrt = (odst.s_addr != ip->ip_dst.s_addr); } @@ -732,8 +738,8 @@ ip_input(struct mbuf *m) } if (ia != NULL) goto ours; - if (m->m_pkthdr.rcvif->if_flags & IFF_BROADCAST) { - TAILQ_FOREACH(ifa, &m->m_pkthdr.rcvif->if_addrlist, ifa_list) { + if (m->m_pkthdr.rcvif && m->m_pkthdr.rcvif->if_flags & IFF_BROADCAST) { + IFADDR_FOREACH(ifa, m->m_pkthdr.rcvif) { if (ifa->ifa_addr->sa_family != AF_INET) continue; ia = ifatoia(ifa); @@ -836,7 +842,7 @@ ip_input(struct mbuf *m) sp = ipsec_getpolicy(tdbi, IPSEC_DIR_INBOUND); } else { sp = ipsec_getpolicybyaddr(m, IPSEC_DIR_INBOUND, - IP_FORWARDING, &error); + IP_FORWARDING, &error); } if (sp == NULL) { /* NB: can happen if error */ splx(s); @@ -897,9 +903,7 @@ ours: */ IPQ_LOCK(); hash = IPREASS_HASH(ip->ip_src.s_addr, ip->ip_id); - /* XXX LIST_FOREACH(fp, &ipq[hash], ipq_q) */ - for (fp = LIST_FIRST(&ipq[hash]); fp != NULL; - fp = LIST_NEXT(fp, ipq_q)) { + LIST_FOREACH(fp, &ipq[hash], ipq_q) { if (ip->ip_id == fp->ipq_id && in_hosteq(ip->ip_src, fp->ipq_src) && in_hosteq(ip->ip_dst, fp->ipq_dst) && @@ -938,7 +942,9 @@ found: */ if (mff || ip->ip_off != htons(0)) { ipstat.ips_fragments++; + s = splvm(); ipqe = pool_get(&ipqent_pool, PR_NOWAIT); + splx(s); if (ipqe == NULL) { ipstat.ips_rcvmemdrop++; IPQ_UNLOCK(); @@ -974,7 +980,7 @@ found: goto bad; } #endif -#if FAST_IPSEC +#ifdef FAST_IPSEC /* * enforce IPsec policy checking if we are seeing last header. * note that we do not visit this with protocols with pcb layer @@ -986,7 +992,7 @@ found: * done. If so, then just pass it along. This tag gets * set during AH, ESP, etc. input handling, before the * packet is returned to the ip input queue for delivery. - */ + */ mtag = m_tag_find(m, PACKET_TAG_IPSEC_IN_DONE, NULL); s = splsoftnet(); if (mtag != NULL) { @@ -994,7 +1000,7 @@ found: sp = ipsec_getpolicy(tdbi, IPSEC_DIR_INBOUND); } else { sp = ipsec_getpolicybyaddr(m, IPSEC_DIR_INBOUND, - IP_FORWARDING, &error); + IP_FORWARDING, &error); } if (sp != NULL) { /* @@ -1006,7 +1012,6 @@ found: /* XXX error stat??? */ error = EINVAL; DPRINTF(("ip_input: no SP, packet discarded\n"));/*XXX*/ - goto bad; } splx(s); if (error) @@ -1044,17 +1049,14 @@ badcsum: * is given as fp; otherwise have to make a chain. */ struct mbuf * -ip_reass(ipqe, fp, ipqhead) - struct ipqent *ipqe; - struct ipq *fp; - struct ipqhead *ipqhead; +ip_reass(struct ipqent *ipqe, struct ipq *fp, struct ipqhead *ipqhead) { struct mbuf *m = ipqe->ipqe_m; struct ipqent *nq, *p, *q; struct ip *ip; struct mbuf *t; int hlen = ipqe->ipqe_ip->ip_hl << 2; - int i, next; + int i, next, s; IPQ_LOCK_CHECK(); @@ -1078,7 +1080,7 @@ ip_reass(ipqe, fp, ipqhead) * We are about to add a fragment; increment frag count. */ ip_nfrags++; - + /* * If first fragment to arrive, create a reassembly queue. */ @@ -1159,7 +1161,9 @@ ip_reass(ipqe, fp, ipqhead) nq = TAILQ_NEXT(q, ipqe_q); m_freem(q->ipqe_m); TAILQ_REMOVE(&fp->ipq_fragq, q, ipqe_q); + s = splvm(); pool_put(&ipqent_pool, q); + splx(s); fp->ipq_nfrags--; ip_nfrags--; } @@ -1200,11 +1204,15 @@ insert: m->m_next = 0; m_cat(m, t); nq = TAILQ_NEXT(q, ipqe_q); + s = splvm(); pool_put(&ipqent_pool, q); + splx(s); for (q = nq; q != NULL; q = nq) { t = q->ipqe_m; nq = TAILQ_NEXT(q, ipqe_q); + s = splvm(); pool_put(&ipqent_pool, q); + splx(s); m_cat(m, t); } ip_nfrags -= fp->ipq_nfrags; @@ -1229,6 +1237,7 @@ insert: for (t = m; t; t = t->m_next) plen += t->m_len; m->m_pkthdr.len = plen; + m->m_pkthdr.csum_flags = 0; } return (m); @@ -1238,7 +1247,9 @@ dropfrag: ip_nfrags--; ipstat.ips_fragdropped++; m_freem(m); + s = splvm(); pool_put(&ipqent_pool, ipqe); + splx(s); return (0); } @@ -1247,11 +1258,11 @@ dropfrag: * associated datagrams. */ void -ip_freef(fp) - struct ipq *fp; +ip_freef(struct ipq *fp) { struct ipqent *q, *p; u_int nfrags = 0; + int s; IPQ_LOCK_CHECK(); @@ -1260,7 +1271,9 @@ ip_freef(fp) m_freem(q->ipqe_m); nfrags++; TAILQ_REMOVE(&fp->ipq_fragq, q, ipqe_q); + s = splvm(); pool_put(&ipqent_pool, q); + splx(s); } if (nfrags != fp->ipq_nfrags) @@ -1289,10 +1302,10 @@ ip_reass_ttl_decr(u_int ticks) u_int nfrags, median, dropfraction, keepfraction; struct ipq *fp, *nfp; int i; - + nfrags = 0; memset(fragttl_histo, 0, sizeof fragttl_histo); - + for (i = 0; i < IPREASS_NHASH; i++) { for (fp = LIST_FIRST(&ipq[i]); fp != NULL; fp = nfp) { fp->ipq_ttl = ((fp->ipq_ttl <= ticks) ? @@ -1345,7 +1358,7 @@ ip_reass_drophalf(void) * queue, discard it. */ void -ip_slowtimo() +ip_slowtimo(void) { static u_int dropscanidx = 0; u_int i; @@ -1392,9 +1405,6 @@ ip_slowtimo() dropscanidx = i; } IPQ_UNLOCK(); -#ifdef GATEWAY - ipflow_slowtimo(); -#endif splx(s); } @@ -1402,7 +1412,7 @@ ip_slowtimo() * Drain off all datagram fragments. */ void -ip_drain() +ip_drain(void) { /* @@ -1429,8 +1439,7 @@ ip_drain() * 0 if the packet should be processed further. */ int -ip_dooptions(m) - struct mbuf *m; +ip_dooptions(struct mbuf *m) { struct ip *ip = mtod(m, struct ip *); u_char *cp, *cp0; @@ -1514,7 +1523,7 @@ ip_dooptions(m) /* * locate outgoing interface */ - bcopy((caddr_t)(cp + off), (caddr_t)&ipaddr.sin_addr, + bcopy((void *)(cp + off), (void *)&ipaddr.sin_addr, sizeof(ipaddr.sin_addr)); if (opt == IPOPT_SSRR) ia = ifatoia(ifa_ifwithladdr(sintosa(&ipaddr))); @@ -1526,8 +1535,8 @@ ip_dooptions(m) goto bad; } ip->ip_dst = ipaddr.sin_addr; - bcopy((caddr_t)&ia->ia_addr.sin_addr, - (caddr_t)(cp + off), sizeof(struct in_addr)); + bcopy((void *)&ia->ia_addr.sin_addr, + (void *)(cp + off), sizeof(struct in_addr)); cp[IPOPT_OFFSET] += sizeof(struct in_addr); /* * Let ip_intr's mcast routing check handle mcast pkts @@ -1550,7 +1559,7 @@ ip_dooptions(m) off--; /* 0 origin */ if ((off + sizeof(struct in_addr)) > optlen) break; - bcopy((caddr_t)(&ip->ip_dst), (caddr_t)&ipaddr.sin_addr, + bcopy((void *)(&ip->ip_dst), (void *)&ipaddr.sin_addr, sizeof(ipaddr.sin_addr)); /* * locate outgoing interface; if we're the destination, @@ -1563,8 +1572,8 @@ ip_dooptions(m) code = ICMP_UNREACH_HOST; goto bad; } - bcopy((caddr_t)&ia->ia_addr.sin_addr, - (caddr_t)(cp + off), sizeof(struct in_addr)); + bcopy((void *)&ia->ia_addr.sin_addr, + (void *)(cp + off), sizeof(struct in_addr)); cp[IPOPT_OFFSET] += sizeof(struct in_addr); break; @@ -1633,7 +1642,7 @@ ip_dooptions(m) } ntime = iptime(); cp0 = (u_char *) &ntime; /* XXX grumble, GCC... */ - bcopy(cp0, (caddr_t)cp + ipt->ipt_ptr - 1, + memmove((char *)cp + ipt->ipt_ptr - 1, cp0, sizeof(n_time)); ipt->ipt_ptr += sizeof(n_time); } @@ -1659,27 +1668,20 @@ bad: * return internet address info of interface to be used to get there. */ struct in_ifaddr * -ip_rtaddr(dst) - struct in_addr dst; +ip_rtaddr(struct in_addr dst) { - struct sockaddr_in *sin; + struct rtentry *rt; + union { + struct sockaddr dst; + struct sockaddr_in dst4; + } u; - sin = satosin(&ipforward_rt.ro_dst); + sockaddr_in_init(&u.dst4, &dst, 0); - if (ipforward_rt.ro_rt == 0 || !in_hosteq(dst, sin->sin_addr)) { - if (ipforward_rt.ro_rt) { - RTFREE(ipforward_rt.ro_rt); - ipforward_rt.ro_rt = 0; - } - sin->sin_family = AF_INET; - sin->sin_len = sizeof(*sin); - sin->sin_addr = dst; + if ((rt = rtcache_lookup(&ipforward_rt, &u.dst)) == NULL) + return NULL; - rtalloc(&ipforward_rt); - } - if (ipforward_rt.ro_rt == 0) - return ((struct in_ifaddr *)0); - return (ifatoia(ipforward_rt.ro_rt->rt_ifa)); + return ifatoia(rt->rt_ifa); } /* @@ -1687,9 +1689,7 @@ ip_rtaddr(dst) * to be picked up later by ip_srcroute if the receiver is interested. */ void -save_rte(option, dst) - u_char *option; - struct in_addr dst; +save_rte(u_char *option, struct in_addr dst) { unsigned olen; @@ -1700,7 +1700,7 @@ save_rte(option, dst) #endif /* 0 */ if (olen > sizeof(ip_srcrt) - (1 + sizeof(dst))) return; - bcopy((caddr_t)option, (caddr_t)ip_srcrt.srcopt, olen); + bcopy((void *)option, (void *)ip_srcrt.srcopt, olen); ip_nhops = (olen - IPOPT_OFFSET - 1) / sizeof(struct in_addr); ip_srcrt.dst = dst; } @@ -1711,16 +1711,16 @@ save_rte(option, dst) * The first hop is placed before the options, will be removed later. */ struct mbuf * -ip_srcroute() +ip_srcroute(void) { struct in_addr *p, *q; struct mbuf *m; if (ip_nhops == 0) - return ((struct mbuf *)0); + return NULL; m = m_get(M_DONTWAIT, MT_SOOPTS); if (m == 0) - return ((struct mbuf *)0); + return NULL; MCLAIM(m, &inetdomain.dom_mowner); #define OPTSIZ (sizeof(ip_srcrt.nop) + sizeof(ip_srcrt.srcopt)) @@ -1748,9 +1748,9 @@ ip_srcroute() */ ip_srcrt.nop = IPOPT_NOP; ip_srcrt.srcopt[IPOPT_OFFSET] = IPOPT_MINOFF; - bcopy((caddr_t)&ip_srcrt.nop, - mtod(m, caddr_t) + sizeof(struct in_addr), OPTSIZ); - q = (struct in_addr *)(mtod(m, caddr_t) + + memmove(mtod(m, char *) + sizeof(struct in_addr), &ip_srcrt.nop, + OPTSIZ); + q = (struct in_addr *)(mtod(m, char *) + sizeof(struct in_addr) + OPTSIZ); #undef OPTSIZ /* @@ -1775,41 +1775,16 @@ ip_srcroute() return (m); } -/* - * 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. - * XXX should be deleted; last arg currently ignored. - */ -void -ip_stripoptions(m, mopt) - struct mbuf *m; - struct mbuf *mopt; -{ - int i; - struct ip *ip = mtod(m, struct ip *); - caddr_t opts; - int olen; - - olen = (ip->ip_hl << 2) - sizeof (struct ip); - opts = (caddr_t)(ip + 1); - i = m->m_len - (sizeof (struct ip) + olen); - bcopy(opts + olen, opts, (unsigned)i); - m->m_len -= olen; - if (m->m_flags & M_PKTHDR) - m->m_pkthdr.len -= olen; - ip->ip_len = htons(ntohs(ip->ip_len) - olen); - ip->ip_hl = sizeof (struct ip) >> 2; -} - const int inetctlerrmap[PRC_NCMDS] = { - 0, 0, 0, 0, - 0, EMSGSIZE, EHOSTDOWN, EHOSTUNREACH, - EHOSTUNREACH, EHOSTUNREACH, ECONNREFUSED, ECONNREFUSED, - EMSGSIZE, EHOSTUNREACH, 0, 0, - 0, 0, 0, 0, - ENOPROTOOPT + [PRC_MSGSIZE] = EMSGSIZE, + [PRC_HOSTDEAD] = EHOSTDOWN, + [PRC_HOSTUNREACH] = EHOSTUNREACH, + [PRC_UNREACH_NET] = EHOSTUNREACH, + [PRC_UNREACH_HOST] = EHOSTUNREACH, + [PRC_UNREACH_PROTOCOL] = ECONNREFUSED, + [PRC_UNREACH_PORT] = ECONNREFUSED, + [PRC_UNREACH_SRCFAIL] = EHOSTUNREACH, + [PRC_PARAMPROB] = ENOPROTOOPT, }; /* @@ -1827,20 +1802,17 @@ const int inetctlerrmap[PRC_NCMDS] = { * via a source route. */ void -ip_forward(m, srcrt) - struct mbuf *m; - int srcrt; +ip_forward(struct mbuf *m, int srcrt) { struct ip *ip = mtod(m, struct ip *); - struct sockaddr_in *sin; struct rtentry *rt; - int error, type = 0, code = 0; + int error, type = 0, code = 0, destmtu = 0; struct mbuf *mcopy; n_long dest; - struct ifnet *destifp; -#if defined(IPSEC) || defined(FAST_IPSEC) - struct ifnet dummyifp; -#endif + union { + struct sockaddr dst; + struct sockaddr_in dst4; + } u; /* * We are now in the output path. @@ -1854,10 +1826,10 @@ ip_forward(m, srcrt) dest = 0; #ifdef DIAGNOSTIC - if (ipprintfs) - printf("forward: src %2.2x dst %2.2x ttl %x\n", - ntohl(ip->ip_src.s_addr), - ntohl(ip->ip_dst.s_addr), ip->ip_ttl); + if (ipprintfs) { + printf("forward: src %s ", inet_ntoa(ip->ip_src)); + printf("dst %s ttl %x\n", inet_ntoa(ip->ip_dst), ip->ip_ttl); + } #endif if (m->m_flags & (M_BCAST|M_MCAST) || in_canforward(ip->ip_dst) == 0) { ipstat.ips_cantforward++; @@ -1868,25 +1840,11 @@ ip_forward(m, srcrt) icmp_error(m, ICMP_TIMXCEED, ICMP_TIMXCEED_INTRANS, dest, 0); return; } - ip->ip_ttl -= IPTTLDEC; - sin = satosin(&ipforward_rt.ro_dst); - if ((rt = ipforward_rt.ro_rt) == 0 || - !in_hosteq(ip->ip_dst, sin->sin_addr)) { - if (ipforward_rt.ro_rt) { - RTFREE(ipforward_rt.ro_rt); - ipforward_rt.ro_rt = 0; - } - sin->sin_family = AF_INET; - sin->sin_len = sizeof(struct sockaddr_in); - sin->sin_addr = ip->ip_dst; - - rtalloc(&ipforward_rt); - if (ipforward_rt.ro_rt == 0) { - icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_HOST, dest, 0); - return; - } - rt = ipforward_rt.ro_rt; + sockaddr_in_init(&u.dst4, &ip->ip_dst, 0); + if ((rt = rtcache_lookup(&ipforward_rt, &u.dst)) == NULL) { + icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_NET, dest, 0); + return; } /* @@ -1898,6 +1856,8 @@ ip_forward(m, srcrt) if (mcopy) mcopy = m_pullup(mcopy, ip->ip_hl << 2); + ip->ip_ttl -= IPTTLDEC; + /* * If forwarding packet using same interface that it came in on, * perhaps should send a redirect to sender to shortcut a hop. @@ -1908,7 +1868,7 @@ ip_forward(m, srcrt) */ if (rt->rt_ifp == m->m_pkthdr.rcvif && (rt->rt_flags & (RTF_DYNAMIC|RTF_MODIFIED)) == 0 && - !in_nullhost(satosin(rt_key(rt))->sin_addr) && + !in_nullhost(satocsin(rt_getkey(rt))->sin_addr) && ipsendredirects && !srcrt) { if (rt->rt_ifa && (ip->ip_src.s_addr & ifatoia(rt->rt_ifa)->ia_subnetmask) == @@ -1931,7 +1891,7 @@ ip_forward(m, srcrt) } } - error = ip_output(m, (struct mbuf *)0, &ipforward_rt, + error = ip_output(m, NULL, &ipforward_rt, (IP_FORWARDING | (ip_directedbcast ? IP_ALLOWBROADCAST : 0)), (struct ip_moptions *)NULL, (struct socket *)NULL); @@ -1954,7 +1914,6 @@ ip_forward(m, srcrt) } if (mcopy == NULL) return; - destifp = NULL; switch (error) { @@ -1975,8 +1934,8 @@ ip_forward(m, srcrt) type = ICMP_UNREACH; code = ICMP_UNREACH_NEEDFRAG; #if !defined(IPSEC) && !defined(FAST_IPSEC) - if (ipforward_rt.ro_rt) - destifp = ipforward_rt.ro_rt->rt_ifp; + if ((rt = rtcache_getrt(&ipforward_rt)) != NULL) + destmtu = rt->rt_ifp->if_mtu; #else /* * If the packet is routed over IPsec tunnel, tell the @@ -1984,7 +1943,7 @@ ip_forward(m, srcrt) * tunnel MTU = if MTU - sizeof(IP) - ESP/AH hdrsiz * XXX quickhack!!! */ - if (ipforward_rt.ro_rt) { + if ((rt = rtcache_getrt(&ipforward_rt)) != NULL) { struct secpolicy *sp; int ipsecerror; size_t ipsechdr; @@ -1995,7 +1954,7 @@ ip_forward(m, srcrt) &ipsecerror); if (sp == NULL) - destifp = ipforward_rt.ro_rt->rt_ifp; + destmtu = rt->rt_ifp->if_mtu; else { /* count IPsec header size */ ipsechdr = ipsec4_hdrsiz(mcopy, @@ -2004,24 +1963,18 @@ ip_forward(m, srcrt) /* * find the correct route for outer IPv4 * header, compute tunnel MTU. - * - * XXX BUG ALERT - * The "dummyifp" code relies upon the fact - * that icmp_error() touches only ifp->if_mtu. */ - /*XXX*/ - destifp = NULL; + if (sp->req != NULL && sp->req->sav != NULL && sp->req->sav->sah != NULL) { ro = &sp->req->sav->sah->sa_route; - if (ro->ro_rt && ro->ro_rt->rt_ifp) { - dummyifp.if_mtu = - ro->ro_rt->rt_rmx.rmx_mtu ? - ro->ro_rt->rt_rmx.rmx_mtu : - ro->ro_rt->rt_ifp->if_mtu; - dummyifp.if_mtu -= ipsechdr; - destifp = &dummyifp; + if (rt && rt->rt_ifp) { + destmtu = + rt->rt_rmx.rmx_mtu ? + rt->rt_rmx.rmx_mtu : + rt->rt_ifp->if_mtu; + destmtu -= ipsechdr; } } @@ -2053,28 +2006,25 @@ ip_forward(m, srcrt) break; #endif } - icmp_error(mcopy, type, code, dest, destifp); + icmp_error(mcopy, type, code, dest, destmtu); } void -ip_savecontrol(inp, mp, ip, m) - struct inpcb *inp; - struct mbuf **mp; - struct ip *ip; - struct mbuf *m; +ip_savecontrol(struct inpcb *inp, struct mbuf **mp, struct ip *ip, + struct mbuf *m) { if (inp->inp_socket->so_options & SO_TIMESTAMP) { struct timeval tv; microtime(&tv); - *mp = sbcreatecontrol((caddr_t) &tv, sizeof(tv), + *mp = sbcreatecontrol((void *) &tv, sizeof(tv), SCM_TIMESTAMP, SOL_SOCKET); if (*mp) mp = &(*mp)->m_next; } if (inp->inp_flags & INP_RECVDSTADDR) { - *mp = sbcreatecontrol((caddr_t) &ip->ip_dst, + *mp = sbcreatecontrol((void *) &ip->ip_dst, sizeof(struct in_addr), IP_RECVDSTADDR, IPPROTO_IP); if (*mp) mp = &(*mp)->m_next; @@ -2088,14 +2038,14 @@ ip_savecontrol(inp, mp, ip, m) */ /* options were tossed already */ if (inp->inp_flags & INP_RECVOPTS) { - *mp = sbcreatecontrol((caddr_t) opts_deleted_above, + *mp = sbcreatecontrol((void *) opts_deleted_above, sizeof(struct in_addr), IP_RECVOPTS, IPPROTO_IP); if (*mp) mp = &(*mp)->m_next; } /* ip_srcroute doesn't do what we want here, need to fix */ if (inp->inp_flags & INP_RECVRETOPTS) { - *mp = sbcreatecontrol((caddr_t) ip_srcroute(), + *mp = sbcreatecontrol((void *) ip_srcroute(), sizeof(struct in_addr), IP_RECVRETOPTS, IPPROTO_IP); if (*mp) mp = &(*mp)->m_next; @@ -2104,19 +2054,43 @@ ip_savecontrol(inp, mp, ip, m) if (inp->inp_flags & INP_RECVIF) { struct sockaddr_dl sdl; - sdl.sdl_len = offsetof(struct sockaddr_dl, sdl_data[0]); - sdl.sdl_family = AF_LINK; - sdl.sdl_index = m->m_pkthdr.rcvif ? - m->m_pkthdr.rcvif->if_index : 0; - sdl.sdl_nlen = sdl.sdl_alen = sdl.sdl_slen = 0; - *mp = sbcreatecontrol((caddr_t) &sdl, sdl.sdl_len, - IP_RECVIF, IPPROTO_IP); + sockaddr_dl_init(&sdl, sizeof(sdl), + (m->m_pkthdr.rcvif != NULL) + ? m->m_pkthdr.rcvif->if_index + : 0, + 0, NULL, 0, NULL, 0); + *mp = sbcreatecontrol(&sdl, sdl.sdl_len, IP_RECVIF, IPPROTO_IP); if (*mp) mp = &(*mp)->m_next; } } /* + * sysctl helper routine for net.inet.ip.forwsrcrt. + */ +static int +sysctl_net_inet_ip_forwsrcrt(SYSCTLFN_ARGS) +{ + int error, tmp; + struct sysctlnode node; + + node = *rnode; + tmp = ip_forwsrcrt; + node.sysctl_data = &tmp; + error = sysctl_lookup(SYSCTLFN_CALL(&node)); + if (error || newp == NULL) + return (error); + + if (kauth_authorize_network(l->l_cred, KAUTH_NETWORK_FORWSRCRT, + 0, NULL, NULL, NULL)) + return (EPERM); + + ip_forwsrcrt = tmp; + + return (0); +} + +/* * sysctl helper routine for net.inet.ip.mtudisctimeout. checks the * range of the new value and tweaks timers if it changes. */ @@ -2143,8 +2117,7 @@ sysctl_net_inet_ip_pmtudto(SYSCTLFN_ARGS #ifdef GATEWAY /* - * sysctl helper routine for net.inet.ip.maxflows. apparently if - * maxflows is even looked up, we "reap flows". + * sysctl helper routine for net.inet.ip.maxflows. */ static int sysctl_net_inet_ip_maxflows(SYSCTLFN_ARGS) @@ -2152,15 +2125,44 @@ sysctl_net_inet_ip_maxflows(SYSCTLFN_ARG int s; s = sysctl_lookup(SYSCTLFN_CALL(rnode)); - if (s) + if (s || newp == NULL) return (s); - + s = splsoftnet(); ipflow_reap(0); splx(s); return (0); } + +static int +sysctl_net_inet_ip_hashsize(SYSCTLFN_ARGS) +{ + int error, tmp; + struct sysctlnode node; + + node = *rnode; + tmp = ip_hashsize; + node.sysctl_data = &tmp; + error = sysctl_lookup(SYSCTLFN_CALL(&node)); + if (error || newp == NULL) + return (error); + + if ((tmp & (tmp - 1)) == 0 && tmp != 0) { + /* + * Can only fail due to malloc() + */ + if (ipflow_invalidate_all(tmp)) + return ENOMEM; + } else { + /* + * EINVAL if not a power of 2 + */ + return EINVAL; + } + + return (0); +} #endif /* GATEWAY */ @@ -2185,7 +2187,7 @@ SYSCTL_SETUP(sysctl_net_inet_ip_setup, " SYSCTL_DESCR("IPv4 related settings"), NULL, 0, NULL, 0, CTL_NET, PF_INET, IPPROTO_IP, CTL_EOL); - + sysctl_createv(clog, 0, NULL, NULL, CTLFLAG_PERMANENT|CTLFLAG_READWRITE, CTLTYPE_INT, "forwarding", @@ -2217,11 +2219,11 @@ SYSCTL_SETUP(sysctl_net_inet_ip_setup, " IPCTL_DEFMTU, CTL_EOL); #endif /* IPCTL_DEFMTU */ sysctl_createv(clog, 0, NULL, NULL, - CTLFLAG_PERMANENT|CTLFLAG_READONLY1, + CTLFLAG_PERMANENT|CTLFLAG_READWRITE, CTLTYPE_INT, "forwsrcrt", SYSCTL_DESCR("Enable forwarding of source-routed " "datagrams"), - NULL, 0, &ip_forwsrcrt, 0, + sysctl_net_inet_ip_forwsrcrt, 0, &ip_forwsrcrt, 0, CTL_NET, PF_INET, IPPROTO_IP, IPCTL_FORWSRCRT, CTL_EOL); sysctl_createv(clog, 0, NULL, NULL, @@ -2282,6 +2284,13 @@ SYSCTL_SETUP(sysctl_net_inet_ip_setup, " sysctl_net_inet_ip_maxflows, 0, &ip_maxflows, 0, CTL_NET, PF_INET, IPPROTO_IP, IPCTL_MAXFLOWS, CTL_EOL); + sysctl_createv(clog, 0, NULL, NULL, + CTLFLAG_PERMANENT|CTLFLAG_READWRITE, + CTLTYPE_INT, "hashsize", + SYSCTL_DESCR("Size of hash table for fast forwarding (IPv4)"), + sysctl_net_inet_ip_hashsize, 0, &ip_hashsize, 0, + CTL_NET, PF_INET, IPPROTO_IP, + CTL_CREATE, CTL_EOL); #endif /* GATEWAY */ sysctl_createv(clog, 0, NULL, NULL, CTLFLAG_PERMANENT|CTLFLAG_READWRITE, @@ -2356,4 +2365,11 @@ SYSCTL_SETUP(sysctl_net_inet_ip_setup, " NULL, 0, &ip_do_loopback_cksum, 0, CTL_NET, PF_INET, IPPROTO_IP, IPCTL_LOOPBACKCKSUM, CTL_EOL); + sysctl_createv(clog, 0, NULL, NULL, + CTLFLAG_PERMANENT, + CTLTYPE_STRUCT, "stats", + SYSCTL_DESCR("IP statistics"), + NULL, 0, &ipstat, sizeof(ipstat), + CTL_NET, PF_INET, IPPROTO_IP, IPCTL_STATS, + CTL_EOL); }