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 1.236.2.1 diff -u -p -r1.206 -r1.236.2.1 --- src/sys/netinet/ip_input.c 2004/12/15 04:25:19 1.206 +++ src/sys/netinet/ip_input.c 2007/09/16 15:34:59 1.236.2.1 @@ -1,4 +1,4 @@ -/* $NetBSD: ip_input.c,v 1.206 2004/12/15 04:25:19 thorpej Exp $ */ +/* $NetBSD: ip_input.c,v 1.236.2.1 2007/09/16 15:34:59 xtraeme 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 1.236.2.1 2007/09/16 15:34:59 xtraeme 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; @@ -373,11 +374,11 @@ 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 +396,7 @@ 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 +415,7 @@ 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(); @@ -445,14 +446,17 @@ 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; @@ -484,12 +488,13 @@ 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); @@ -668,7 +673,7 @@ 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 +737,8 @@ 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 +841,7 @@ 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); @@ -938,7 +943,9 @@ 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 +981,7 @@ 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 +993,7 @@ 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 +1001,7 @@ 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 +1013,6 @@ found: /* XXX error stat??? */ error = EINVAL; DPRINTF(("ip_input: no SP, packet discarded\n"));/*XXX*/ - goto bad; } splx(s); if (error) @@ -1044,17 +1050,14 @@ 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 +1081,7 @@ 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 +1162,9 @@ 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 +1205,15 @@ 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 +1238,7 @@ 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 +1248,9 @@ dropfrag: ip_nfrags--; ipstat.ips_fragdropped++; m_freem(m); + s = splvm(); pool_put(&ipqent_pool, ipqe); + splx(s); return (0); } @@ -1247,11 +1259,11 @@ 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 +1272,9 @@ 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 +1303,10 @@ 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 +1359,7 @@ ip_reass_drophalf(void) * queue, discard it. */ void -ip_slowtimo() +ip_slowtimo(void) { static u_int dropscanidx = 0; u_int i; @@ -1402,7 +1416,7 @@ ip_slowtimo() * Drain off all datagram fragments. */ void -ip_drain() +ip_drain(void) { /* @@ -1429,8 +1443,7 @@ 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; @@ -1659,8 +1672,7 @@ 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; @@ -1687,9 +1699,7 @@ 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; @@ -1711,7 +1721,7 @@ 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; @@ -1783,9 +1793,7 @@ ip_srcroute() * XXX should be deleted; last arg currently ignored. */ void -ip_stripoptions(m, mopt) - struct mbuf *m; - struct mbuf *mopt; +ip_stripoptions(struct mbuf *m, struct mbuf *mopt) { int i; struct ip *ip = mtod(m, struct ip *); @@ -1827,20 +1835,14 @@ 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 /* * We are now in the output path. @@ -1854,10 +1856,10 @@ 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++; @@ -1868,7 +1870,6 @@ 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 || @@ -1883,7 +1884,7 @@ ip_forward(m, srcrt) rtalloc(&ipforward_rt); if (ipforward_rt.ro_rt == 0) { - icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_HOST, dest, 0); + icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_NET, dest, 0); return; } rt = ipforward_rt.ro_rt; @@ -1898,6 +1899,8 @@ 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. @@ -1954,7 +1957,6 @@ ip_forward(m, srcrt) } if (mcopy == NULL) return; - destifp = NULL; switch (error) { @@ -1976,7 +1978,7 @@ ip_forward(m, srcrt) code = ICMP_UNREACH_NEEDFRAG; #if !defined(IPSEC) && !defined(FAST_IPSEC) if (ipforward_rt.ro_rt) - destifp = ipforward_rt.ro_rt->rt_ifp; + destmtu = ipforward_rt.ro_rt->rt_ifp->if_mtu; #else /* * If the packet is routed over IPsec tunnel, tell the @@ -1995,7 +1997,7 @@ ip_forward(m, srcrt) &ipsecerror); if (sp == NULL) - destifp = ipforward_rt.ro_rt->rt_ifp; + destmtu = ipforward_rt.ro_rt->rt_ifp->if_mtu; else { /* count IPsec header size */ ipsechdr = ipsec4_hdrsiz(mcopy, @@ -2004,24 +2006,18 @@ 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 = + destmtu = 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; + destmtu -= ipsechdr; } } @@ -2053,15 +2049,12 @@ 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) { @@ -2117,6 +2110,31 @@ ip_savecontrol(inp, mp, ip, m) } /* + * 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. */ @@ -2154,7 +2172,7 @@ sysctl_net_inet_ip_maxflows(SYSCTLFN_ARG s = sysctl_lookup(SYSCTLFN_CALL(rnode)); if (s) return (s); - + s = splsoftnet(); ipflow_reap(0); splx(s); @@ -2185,7 +2203,7 @@ 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 +2235,11 @@ 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, @@ -2356,4 +2374,11 @@ 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); }