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 1.117 retrieving revision 1.233 diff -u -p -r1.117 -r1.233 --- src/sys/netinet/ip_input.c 2000/08/25 13:35:05 1.117 +++ src/sys/netinet/ip_input.c 2006/10/05 17:35:19 1.233 @@ -1,9 +1,9 @@ -/* $NetBSD: ip_input.c,v 1.117 2000/08/25 13:35:05 tron Exp $ */ +/* $NetBSD: ip_input.c,v 1.233 2006/10/05 17:35:19 tls Exp $ */ /* * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project. * All rights reserved. - * + * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: @@ -15,7 +15,7 @@ * 3. Neither the name of the project nor the names of its contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. - * + * * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE @@ -78,11 +78,7 @@ * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. - * 3. All advertising materials mentioning features or use of this software - * must display the following acknowledgement: - * This product includes software developed by the University of - * California, Berkeley and its contributors. - * 4. Neither the name of the University nor the names of its contributors + * 3. Neither the name of the University nor the names of its contributors * may be used to endorse or promote products derived from this software * without specific prior written permission. * @@ -101,10 +97,16 @@ * @(#)ip_input.c 8.2 (Berkeley) 1/4/94 */ +#include +__KERNEL_RCSID(0, "$NetBSD: ip_input.c,v 1.233 2006/10/05 17:35:19 tls Exp $"); + +#include "opt_inet.h" #include "opt_gateway.h" #include "opt_pfil_hooks.h" #include "opt_ipsec.h" #include "opt_mrouting.h" +#include "opt_mbuftrace.h" +#include "opt_inet_csum.h" #include #include @@ -117,12 +119,9 @@ #include #include #include -#include #include - -#include - #include +#include #include #include @@ -133,12 +132,15 @@ #include #include #include +#include #include #include #include /* just for gif_ttl */ #include #include "gif.h" +#include +#include "gre.h" #ifdef MROUTING #include @@ -148,6 +150,10 @@ #include #include #endif +#ifdef FAST_IPSEC +#include +#include +#endif /* FAST_IPSEC*/ #ifndef IPFORWARDING #ifdef GATEWAY @@ -166,7 +172,7 @@ #define IPALLOWSRCRT 1 /* allow source-routed packets */ #endif #ifndef IPMTUDISC -#define IPMTUDISC 0 +#define IPMTUDISC 1 #endif #ifndef IPMTUDISCTIMEOUT #define IPMTUDISCTIMEOUT (10 * 60) /* as per RFC 1191 */ @@ -190,33 +196,103 @@ int ip_forwsrcrt = IPFORWSRCRT; int ip_directedbcast = IPDIRECTEDBCAST; int ip_allowsrcrt = IPALLOWSRCRT; int ip_mtudisc = IPMTUDISC; -u_int ip_mtudisc_timeout = IPMTUDISCTIMEOUT; +int ip_mtudisc_timeout = IPMTUDISCTIMEOUT; #ifdef DIAGNOSTIC int ipprintfs = 0; #endif +int ip_do_randomid = 0; + +/* + * XXX - Setting ip_checkinterface mostly implements the receive side of + * the Strong ES model described in RFC 1122, but since the routing table + * and transmit implementation do not implement the Strong ES model, + * setting this to 1 results in an odd hybrid. + * + * XXX - ip_checkinterface currently must be disabled if you use ipnat + * to translate the destination address to another local interface. + * + * XXX - ip_checkinterface must be disabled if you add IP aliases + * to the loopback interface instead of the interface where the + * packets for those addresses are received. + */ +int ip_checkinterface = 0; + + struct rttimer_queue *ip_mtudisc_timeout_q = NULL; -extern struct domain inetdomain; int ipqmaxlen = IFQ_MAXLEN; -struct in_ifaddrhead in_ifaddr; +u_long in_ifaddrhash; /* size of hash table - 1 */ +int in_ifaddrentries; /* total number of addrs */ +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 */ +struct in_multihashhead *in_multihashtbl; struct ifqueue ipintrq; struct ipstat ipstat; -u_int16_t ip_id; +uint16_t ip_id; + +#ifdef PFIL_HOOKS +struct pfil_head inet_pfil_hook; +#endif + +/* + * Cached copy of nmbclusters. If nbclusters is different, + * recalculate IP parameters derived from nmbclusters. + */ +static int ip_nmbclusters; /* copy of nmbclusters */ +static void ip_nmbclusters_changed(void); /* recalc limits */ -struct ipqhead ipq; +#define CHECK_NMBCLUSTER_PARAMS() \ +do { \ + if (__predict_false(ip_nmbclusters != nmbclusters)) \ + ip_nmbclusters_changed(); \ +} while (/*CONSTCOND*/0) + +/* IP datagram reassembly queues (hashed) */ +#define IPREASS_NHASH_LOG2 6 +#define IPREASS_NHASH (1 << IPREASS_NHASH_LOG2) +#define IPREASS_HMASK (IPREASS_NHASH - 1) +#define IPREASS_HASH(x,y) \ + (((((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_nfrags; /* total fragments in reass queues */ -static __inline int ipq_lock_try __P((void)); -static __inline void ipq_unlock __P((void)); +int ip_maxfragpackets = 200; /* limit on packets. XXX sysctl */ +int ip_maxfrags; /* limit on fragments. XXX sysctl */ -static __inline int -ipq_lock_try() + +/* + * 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 + * total fragments in reassembly queues.This AIMD policy avoids + * repeatedly deleting single packets under heavy fragmentation load + * (e.g., from lossy NFS peers). + */ +static u_int ip_reass_ttl_decr(u_int ticks); +static void ip_reass_drophalf(void); + + +static inline int ipq_lock_try(void); +static inline void ipq_unlock(void); + +static inline int +ipq_lock_try(void) { int s; - s = splimp(); + /* + * Use splvm() -- we're blocking things that would cause + * mbuf allocation. + */ + s = splvm(); if (ipq_locked) { splx(s); return (0); @@ -226,12 +302,12 @@ ipq_lock_try() return (1); } -static __inline void -ipq_unlock() +static inline void +ipq_unlock(void) { int s; - s = splimp(); + s = splvm(); ipq_locked = 0; splx(s); } @@ -243,14 +319,14 @@ do { \ printf("%s:%d: ipq already locked\n", __FILE__, __LINE__); \ panic("ipq_lock"); \ } \ -} while (0) +} while (/*CONSTCOND*/ 0) #define IPQ_LOCK_CHECK() \ do { \ if (ipq_locked == 0) { \ printf("%s:%d: ipq lock not held\n", __FILE__, __LINE__); \ panic("ipq lock check"); \ } \ -} while (0) +} while (/*CONSTCOND*/ 0) #else #define IPQ_LOCK() (void) ipq_lock_try() #define IPQ_LOCK_CHECK() /* nothing */ @@ -258,7 +334,30 @@ do { \ #define IPQ_UNLOCK() ipq_unlock() -struct pool ipqent_pool; +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); + +#ifdef INET_CSUM_COUNTERS +#include + +struct evcnt ip_hwcsum_bad = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, + NULL, "inet", "hwcsum bad"); +struct evcnt ip_hwcsum_ok = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, + NULL, "inet", "hwcsum ok"); +struct evcnt ip_swcsum = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, + NULL, "inet", "swcsum"); + +#define INET_CSUM_COUNTER_INCR(ev) (ev)->ev_count++ + +EVCNT_ATTACH_STATIC(ip_hwcsum_bad); +EVCNT_ATTACH_STATIC(ip_hwcsum_ok); +EVCNT_ATTACH_STATIC(ip_swcsum); + +#else + +#define INET_CSUM_COUNTER_INCR(ev) /* nothing */ + +#endif /* INET_CSUM_COUNTERS */ /* * We need to save the IP options in case a protocol wants to respond @@ -275,21 +374,33 @@ 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" }; +#endif + +/* + * Compute IP limits derived from the value of nmbclusters. + */ +static void +ip_nmbclusters_changed(void) +{ + ip_maxfrags = nmbclusters / 4; + ip_nmbclusters = nmbclusters; +} /* * IP initialization: fill in IP protocol switch table. * All protocols not implemented in kernel go to raw IP protocol handler. */ void -ip_init() +ip_init(void) { - struct protosw *pr; + const struct protosw *pr; int i; - pool_init(&ipqent_pool, sizeof(struct ipqent), 0, 0, 0, "ipqepl", - 0, NULL, NULL, M_IPQ); - pr = pffindproto(PF_INET, IPPROTO_RAW, SOCK_RAW); if (pr == 0) panic("ip_init"); @@ -300,38 +411,63 @@ ip_init() if (pr->pr_domain->dom_family == PF_INET && pr->pr_protocol && pr->pr_protocol != IPPROTO_RAW) ip_protox[pr->pr_protocol] = pr - inetsw; - LIST_INIT(&ipq); - ip_id = time.tv_sec & 0xffff; + + for (i = 0; i < IPREASS_NHASH; i++) + LIST_INIT(&ipq[i]); + + ip_id = time_second & 0xfffff; + ipintrq.ifq_maxlen = ipqmaxlen; - TAILQ_INIT(&in_ifaddr); - in_ifaddrhashtbl = - hashinit(IN_IFADDR_HASH_SIZE, M_IFADDR, M_WAITOK, &in_ifaddrhash); - if (ip_mtudisc != 0) - ip_mtudisc_timeout_q = - rt_timer_queue_create(ip_mtudisc_timeout); + ip_nmbclusters_changed(); + + TAILQ_INIT(&in_ifaddrhead); + in_ifaddrhashtbl = hashinit(IN_IFADDR_HASH_SIZE, HASH_LIST, M_IFADDR, + M_WAITOK, &in_ifaddrhash); + in_multihashtbl = hashinit(IN_IFADDR_HASH_SIZE, HASH_LIST, M_IPMADDR, + M_WAITOK, &in_multihash); + ip_mtudisc_timeout_q = rt_timer_queue_create(ip_mtudisc_timeout); #ifdef GATEWAY ipflow_init(); #endif + +#ifdef PFIL_HOOKS + /* Register our Packet Filter hook. */ + inet_pfil_hook.ph_type = PFIL_TYPE_AF; + inet_pfil_hook.ph_af = AF_INET; + i = pfil_head_register(&inet_pfil_hook); + if (i != 0) + printf("ip_init: WARNING: unable to register pfil hook, " + "error %d\n", i); +#endif /* PFIL_HOOKS */ + +#ifdef MBUFTRACE + MOWNER_ATTACH(&ip_tx_mowner); + MOWNER_ATTACH(&ip_rx_mowner); +#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) { - s = splimp(); + s = splnet(); IF_DEQUEUE(&ipintrq, m); splx(s); if (m == 0) return; + MCLAIM(m, &ip_rx_mowner); ip_input(m); } } @@ -350,37 +486,48 @@ ip_input(struct mbuf *m) struct ipqent *ipqe; int hlen = 0, mff, len; int downmatch; -#ifdef PFIL_HOOKS - struct packet_filter_hook *pfh; - struct mbuf *m0; - int rv; -#endif /* PFIL_HOOKS */ + 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 error; +#endif /* FAST_IPSEC */ + MCLAIM(m, &ip_rx_mowner); #ifdef DIAGNOSTIC if ((m->m_flags & M_PKTHDR) == 0) panic("ipintr no HDR"); #endif -#ifdef IPSEC - /* - * should the inner packet be considered authentic? - * see comment in ah4_input(). - */ - if (m) { - m->m_flags &= ~M_AUTHIPHDR; - m->m_flags &= ~M_AUTHIPDGM; - } -#endif + /* * If no IP addresses have been set yet but the interfaces * are receiving, can't do anything with incoming packets yet. */ - if (in_ifaddr.tqh_first == 0) + if (TAILQ_FIRST(&in_ifaddrhead) == 0) goto bad; ipstat.ips_total++; - if (m->m_len < sizeof (struct ip) && - (m = m_pullup(m, sizeof (struct ip))) == 0) { - ipstat.ips_toosmall++; - return; + /* + * If the IP header is not aligned, slurp it up into a new + * mbuf with space for link headers, in the event we forward + * 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 ((m = m_copyup(m, sizeof(struct ip), + (max_linkhdr + 3) & ~3)) == NULL) { + /* XXXJRT new stat, please */ + ipstat.ips_toosmall++; + return; + } + } else if (__predict_false(m->m_len < sizeof (struct ip))) { + if ((m = m_pullup(m, sizeof (struct ip))) == NULL) { + ipstat.ips_toosmall++; + return; + } } ip = mtod(m, struct ip *); if (ip->ip_v != IPVERSION) { @@ -405,21 +552,47 @@ ip_input(struct mbuf *m) * not allowed. */ if (IN_MULTICAST(ip->ip_src.s_addr)) { - /* XXX stat */ + ipstat.ips_badaddr++; goto bad; } - if (in_cksum(m, hlen) != 0) { - ipstat.ips_badsum++; - goto bad; + /* 127/8 must not appear on wire - RFC1122 */ + if ((ntohl(ip->ip_dst.s_addr) >> IN_CLASSA_NSHIFT) == IN_LOOPBACKNET || + (ntohl(ip->ip_src.s_addr) >> IN_CLASSA_NSHIFT) == IN_LOOPBACKNET) { + if ((m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) == 0) { + ipstat.ips_badaddr++; + goto bad; + } } - /* - * Convert fields to host representation. - */ - NTOHS(ip->ip_len); - NTOHS(ip->ip_off); - len = ip->ip_len; + switch (m->m_pkthdr.csum_flags & + ((m->m_pkthdr.rcvif->if_csum_flags_rx & M_CSUM_IPv4) | + M_CSUM_IPv4_BAD)) { + case M_CSUM_IPv4|M_CSUM_IPv4_BAD: + INET_CSUM_COUNTER_INCR(&ip_hwcsum_bad); + goto badcsum; + + case M_CSUM_IPv4: + /* Checksum was okay. */ + INET_CSUM_COUNTER_INCR(&ip_hwcsum_ok); + break; + + default: + /* + * Must compute it ourselves. Maybe skip checksum on + * loopback interfaces. + */ + if (__predict_true(!(m->m_pkthdr.rcvif->if_flags & + IFF_LOOPBACK) || ip_do_loopback_cksum)) { + INET_CSUM_COUNTER_INCR(&ip_swcsum); + if (in_cksum(m, hlen) != 0) + goto badcsum; + } + break; + } + + /* Retrieve the packet length. */ + len = ntohs(ip->ip_len); /* * Check for additional length bogosity @@ -447,8 +620,8 @@ ip_input(struct mbuf *m) m_adj(m, len - m->m_pkthdr.len); } -#ifdef IPSEC - /* ipflow (IP fast fowarding) is not compatible with IPsec. */ +#if defined(IPSEC) + /* ipflow (IP fast forwarding) is not compatible with IPsec. */ m->m_flags &= ~M_CANFASTFWD; #else /* @@ -466,21 +639,54 @@ ip_input(struct mbuf *m) * Note that filters must _never_ set this flag, as another filter * in the list may have previously cleared it. */ - m0 = m; - pfh = pfil_hook_get(PFIL_IN, &inetsw[ip_protox[IPPROTO_IP]].pr_pfh); - for (; pfh; pfh = pfh->pfil_link.tqe_next) - if (pfh->pfil_func) { - rv = pfh->pfil_func(ip, hlen, - m->m_pkthdr.rcvif, 0, &m0); - if (rv) - return; - m = m0; - if (m == NULL) - return; - ip = mtod(m, struct ip *); - } + /* + * let ipfilter look at packet on the wire, + * not the decapsulated packet. + */ +#ifdef IPSEC + if (!ipsec_getnhist(m)) +#elif defined(FAST_IPSEC) + if (!ipsec_indone(m)) +#else + if (1) +#endif + { + struct in_addr odst; + + odst = ip->ip_dst; + if (pfil_run_hooks(&inet_pfil_hook, &m, m->m_pkthdr.rcvif, + PFIL_IN) != 0) + return; + if (m == NULL) + return; + ip = mtod(m, struct ip *); + hlen = ip->ip_hl << 2; + /* + * XXX The setting of "srcrt" here is to prevent ip_forward() + * from generating ICMP redirects for packets that have + * been redirected by a hook back out on to the same LAN that + * they came from and is not an indication that the packet + * is being inffluenced by source routing options. This + * allows things like + * "rdr tlp0 0/0 port 80 -> 1.1.1.200 3128 tcp" + * where tlp0 is both on the 1.1.1.0/24 network and is the + * 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... + */ + srcrt = (odst.s_addr != ip->ip_dst.s_addr); + } #endif /* PFIL_HOOKS */ +#ifdef ALTQ + /* XXX Temporary until ALTQ is changed to use a pfil hook */ + if (altq_input != NULL && (*altq_input)(m, AF_INET) == 0) { + /* packet dropped by traffic conditioner */ + return; + } +#endif + /* * Process options and, if not destined for us, * ship it on. ip_dooptions returns 1 when an @@ -492,6 +698,26 @@ ip_input(struct mbuf *m) return; /* + * Enable a consistency check between the destination address + * and the arrival interface for a unicast packet (the RFC 1122 + * strong ES model) if IP forwarding is disabled and the packet + * is not locally generated. + * + * XXX - Checking also should be disabled if the destination + * address is ipnat'ed to a different interface. + * + * XXX - Checking is incompatible with IP aliases added + * to the loopback interface instead of the interface where + * the packets are received. + * + * XXX - We need to add a per ifaddr flag for this so that + * we get finer grain control. + */ + checkif = ip_checkinterface && (ipforwarding == 0) && + (m->m_pkthdr.rcvif != NULL) && + ((m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) == 0); + + /* * Check our list of addresses, to see if the packet is for us. * * Traditional 4.4BSD did not consult IFF_UP at all. @@ -499,10 +725,10 @@ ip_input(struct mbuf *m) * as not mine. */ downmatch = 0; - for (ia = IN_IFADDR_HASH(ip->ip_dst.s_addr).lh_first; - ia != NULL; - ia = ia->ia_hash.le_next) { + LIST_FOREACH(ia, &IN_IFADDR_HASH(ip->ip_dst.s_addr), ia_hash) { if (in_hosteq(ia->ia_addr.sin_addr, ip->ip_dst)) { + if (checkif && ia->ia_ifp != m->m_pkthdr.rcvif) + continue; if ((ia->ia_ifp->if_flags & IFF_UP) != 0) break; else @@ -511,10 +737,10 @@ ip_input(struct mbuf *m) } if (ia != NULL) goto ours; - if (m->m_pkthdr.rcvif->if_flags & IFF_BROADCAST) { - for (ifa = m->m_pkthdr.rcvif->if_addrlist.tqh_first; - ifa != NULL; ifa = ifa->ifa_list.tqe_next) { - if (ifa->ifa_addr->sa_family != AF_INET) continue; + 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); if (in_hosteq(ip->ip_dst, ia->ia_broadaddr.sin_addr) || in_hosteq(ip->ip_dst, ia->ia_netbroadcast) || @@ -538,14 +764,6 @@ ip_input(struct mbuf *m) #ifdef MROUTING extern struct socket *ip_mrouter; - if (m->m_flags & M_EXT) { - if ((m = m_pullup(m, hlen)) == 0) { - ipstat.ips_toosmall++; - return; - } - ip = mtod(m, struct ip *); - } - if (ip_mrouter) { /* * If we are acting as a multicast router, all @@ -609,7 +827,62 @@ ip_input(struct mbuf *m) ipstat.ips_cantforward++; return; } - ip_forward(m, 0); +#ifdef IPSEC + if (ipsec4_in_reject(m, NULL)) { + ipsecstat.in_polvio++; + goto bad; + } +#endif +#ifdef FAST_IPSEC + mtag = m_tag_find(m, PACKET_TAG_IPSEC_IN_DONE, NULL); + s = splsoftnet(); + if (mtag != NULL) { + tdbi = (struct tdb_ident *)(mtag + 1); + sp = ipsec_getpolicy(tdbi, IPSEC_DIR_INBOUND); + } else { + sp = ipsec_getpolicybyaddr(m, IPSEC_DIR_INBOUND, + IP_FORWARDING, &error); + } + if (sp == NULL) { /* NB: can happen if error */ + splx(s); + /*XXX error stat???*/ + DPRINTF(("ip_input: no SP for forwarding\n")); /*XXX*/ + goto bad; + } + + /* + * Check security policy against packet attributes. + */ + error = ipsec_in_reject(sp, m); + KEY_FREESP(&sp); + splx(s); + if (error) { + ipstat.ips_cantforward++; + goto bad; + } + + /* + * Peek at the outbound SP for this packet to determine if + * it's a Fast Forward candidate. + */ + mtag = m_tag_find(m, PACKET_TAG_IPSEC_PENDING_TDB, NULL); + if (mtag != NULL) + m->m_flags &= ~M_CANFASTFWD; + else { + s = splsoftnet(); + sp = ipsec4_checkpolicy(m, IPSEC_DIR_OUTBOUND, + (IP_FORWARDING | + (ip_directedbcast ? IP_ALLOWBROADCAST : 0)), + &error, NULL); + if (sp != NULL) { + m->m_flags &= ~M_CANFASTFWD; + KEY_FREESP(&sp); + } + splx(s); + } +#endif /* FAST_IPSEC */ + + ip_forward(m, srcrt); } return; @@ -621,18 +894,24 @@ ours: * if the packet was previously fragmented, * but it's not worth the time; just let them time out.) */ - if (ip->ip_off & ~(IP_DF|IP_RF)) { + if (ip->ip_off & ~htons(IP_DF|IP_RF)) { + /* * Look for queue of fragments * of this datagram. */ IPQ_LOCK(); - for (fp = ipq.lh_first; fp != NULL; fp = fp->ipq_q.le_next) + 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)) { if (ip->ip_id == fp->ipq_id && in_hosteq(ip->ip_src, fp->ipq_src) && in_hosteq(ip->ip_dst, fp->ipq_dst) && ip->ip_p == fp->ipq_p) goto found; + + } fp = 0; found: @@ -641,29 +920,32 @@ found: * set ipqe_mff if more fragments are expected, * convert offset of this to bytes. */ - ip->ip_len -= hlen; - mff = (ip->ip_off & IP_MF) != 0; + ip->ip_len = htons(ntohs(ip->ip_len) - hlen); + mff = (ip->ip_off & htons(IP_MF)) != 0; if (mff) { /* * Make sure that fragments have a data length * that's a non-zero multiple of 8 bytes. */ - if (ip->ip_len == 0 || (ip->ip_len & 0x7) != 0) { + if (ntohs(ip->ip_len) == 0 || + (ntohs(ip->ip_len) & 0x7) != 0) { ipstat.ips_badfrags++; IPQ_UNLOCK(); goto bad; } } - ip->ip_off <<= 3; + ip->ip_off = htons((ntohs(ip->ip_off) & IP_OFFMASK) << 3); /* * If datagram marked as having more fragments * or if this is not the first fragment, * attempt reassembly; if it succeeds, proceed. */ - if (mff || ip->ip_off) { + 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(); @@ -672,7 +954,7 @@ found: ipqe->ipqe_mff = mff; ipqe->ipqe_m = m; ipqe->ipqe_ip = ip; - m = ip_reass(ipqe, fp); + m = ip_reass(ipqe, fp, &ipq[hash]); if (m == 0) { IPQ_UNLOCK(); return; @@ -680,18 +962,71 @@ found: ipstat.ips_reassembled++; ip = mtod(m, struct ip *); hlen = ip->ip_hl << 2; - ip->ip_len += hlen; + ip->ip_len = htons(ntohs(ip->ip_len) + hlen); } else if (fp) ip_freef(fp); IPQ_UNLOCK(); } +#if defined(IPSEC) + /* + * enforce IPsec policy checking if we are seeing last header. + * note that we do not visit this with protocols with pcb layer + * code - like udp/tcp/raw ip. + */ + if ((inetsw[ip_protox[ip->ip_p]].pr_flags & PR_LASTHDR) != 0 && + ipsec4_in_reject(m, NULL)) { + ipsecstat.in_polvio++; + goto bad; + } +#endif +#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 + * code - like udp/tcp/raw ip. + */ + if ((inetsw[ip_protox[ip->ip_p]].pr_flags & PR_LASTHDR) != 0) { + /* + * Check if the packet has already had IPsec processing + * 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) { + tdbi = (struct tdb_ident *)(mtag + 1); + sp = ipsec_getpolicy(tdbi, IPSEC_DIR_INBOUND); + } else { + sp = ipsec_getpolicybyaddr(m, IPSEC_DIR_INBOUND, + IP_FORWARDING, &error); + } + if (sp != NULL) { + /* + * Check security policy against packet attributes. + */ + error = ipsec_in_reject(sp, m); + KEY_FREESP(&sp); + } else { + /* XXX error stat??? */ + error = EINVAL; +DPRINTF(("ip_input: no SP, packet discarded\n"));/*XXX*/ + goto bad; + } + splx(s); + if (error) + goto bad; + } +#endif /* FAST_IPSEC */ + /* * Switch out to protocol's input routine. */ #if IFA_STATS - ia->ia_ifa.ifa_data.ifad_inbytes += ip->ip_len; + if (ia && ip) + ia->ia_ifa.ifa_data.ifad_inbytes += ntohs(ip->ip_len); #endif ipstat.ips_delivered++; { @@ -702,6 +1037,11 @@ found: } bad: m_freem(m); + return; + +badcsum: + ipstat.ips_badsum++; + m_freem(m); } /* @@ -711,16 +1051,14 @@ bad: * is given as fp; otherwise have to make a chain. */ struct mbuf * -ip_reass(ipqe, fp) - struct ipqent *ipqe; - struct ipq *fp; +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(); @@ -731,31 +1069,59 @@ ip_reass(ipqe, fp) m->m_data += hlen; m->m_len -= hlen; +#ifdef notyet + /* make sure fragment limit is up-to-date */ + CHECK_NMBCLUSTER_PARAMS(); + + /* If we have too many fragments, drop the older half. */ + if (ip_nfrags >= ip_maxfrags) + ip_reass_drophalf(void); +#endif + + /* + * We are about to add a fragment; increment frag count. + */ + ip_nfrags++; + /* * If first fragment to arrive, create a reassembly queue. */ if (fp == 0) { + /* + * Enforce upper bound on number of fragmented packets + * for which we attempt reassembly; + * If maxfrag is 0, never accept fragments. + * If maxfrag is -1, accept all fragments without limitation. + */ + if (ip_maxfragpackets < 0) + ; + else if (ip_nfragpackets >= ip_maxfragpackets) + goto dropfrag; + ip_nfragpackets++; MALLOC(fp, struct ipq *, sizeof (struct ipq), M_FTABLE, M_NOWAIT); if (fp == NULL) goto dropfrag; - LIST_INSERT_HEAD(&ipq, fp, ipq_q); + LIST_INSERT_HEAD(ipqhead, fp, ipq_q); + fp->ipq_nfrags = 1; fp->ipq_ttl = IPFRAGTTL; fp->ipq_p = ipqe->ipqe_ip->ip_p; fp->ipq_id = ipqe->ipqe_ip->ip_id; - LIST_INIT(&fp->ipq_fragq); + TAILQ_INIT(&fp->ipq_fragq); fp->ipq_src = ipqe->ipqe_ip->ip_src; fp->ipq_dst = ipqe->ipqe_ip->ip_dst; p = NULL; goto insert; + } else { + fp->ipq_nfrags++; } /* * Find a segment which begins after this one does. */ - for (p = NULL, q = fp->ipq_fragq.lh_first; q != NULL; - p = q, q = q->ipqe_q.le_next) - if (q->ipqe_ip->ip_off > ipqe->ipqe_ip->ip_off) + for (p = NULL, q = TAILQ_FIRST(&fp->ipq_fragq); q != NULL; + p = q, q = TAILQ_NEXT(q, ipqe_q)) + if (ntohs(q->ipqe_ip->ip_off) > ntohs(ipqe->ipqe_ip->ip_off)) break; /* @@ -764,14 +1130,16 @@ ip_reass(ipqe, fp) * segment. If it provides all of our data, drop us. */ if (p != NULL) { - i = p->ipqe_ip->ip_off + p->ipqe_ip->ip_len - - ipqe->ipqe_ip->ip_off; + i = ntohs(p->ipqe_ip->ip_off) + ntohs(p->ipqe_ip->ip_len) - + ntohs(ipqe->ipqe_ip->ip_off); if (i > 0) { - if (i >= ipqe->ipqe_ip->ip_len) + if (i >= ntohs(ipqe->ipqe_ip->ip_len)) goto dropfrag; m_adj(ipqe->ipqe_m, i); - ipqe->ipqe_ip->ip_off += i; - ipqe->ipqe_ip->ip_len -= i; + ipqe->ipqe_ip->ip_off = + htons(ntohs(ipqe->ipqe_ip->ip_off) + i); + ipqe->ipqe_ip->ip_len = + htons(ntohs(ipqe->ipqe_ip->ip_len) - i); } } @@ -779,20 +1147,27 @@ ip_reass(ipqe, fp) * While we overlap succeeding segments trim them or, * if they are completely covered, dequeue them. */ - for (; q != NULL && ipqe->ipqe_ip->ip_off + ipqe->ipqe_ip->ip_len > - q->ipqe_ip->ip_off; q = nq) { - i = (ipqe->ipqe_ip->ip_off + ipqe->ipqe_ip->ip_len) - - q->ipqe_ip->ip_off; - if (i < q->ipqe_ip->ip_len) { - q->ipqe_ip->ip_len -= i; - q->ipqe_ip->ip_off += i; + for (; q != NULL && + ntohs(ipqe->ipqe_ip->ip_off) + ntohs(ipqe->ipqe_ip->ip_len) > + ntohs(q->ipqe_ip->ip_off); q = nq) { + i = (ntohs(ipqe->ipqe_ip->ip_off) + + ntohs(ipqe->ipqe_ip->ip_len)) - ntohs(q->ipqe_ip->ip_off); + if (i < ntohs(q->ipqe_ip->ip_len)) { + q->ipqe_ip->ip_len = + htons(ntohs(q->ipqe_ip->ip_len) - i); + q->ipqe_ip->ip_off = + htons(ntohs(q->ipqe_ip->ip_off) + i); m_adj(q->ipqe_m, i); break; } - nq = q->ipqe_q.le_next; + nq = TAILQ_NEXT(q, ipqe_q); m_freem(q->ipqe_m); - LIST_REMOVE(q, ipqe_q); + TAILQ_REMOVE(&fp->ipq_fragq, q, ipqe_q); + s = splvm(); pool_put(&ipqent_pool, q); + splx(s); + fp->ipq_nfrags--; + ip_nfrags--; } insert: @@ -801,16 +1176,16 @@ insert: * check for complete reassembly. */ if (p == NULL) { - LIST_INSERT_HEAD(&fp->ipq_fragq, ipqe, ipqe_q); + TAILQ_INSERT_HEAD(&fp->ipq_fragq, ipqe, ipqe_q); } else { - LIST_INSERT_AFTER(p, ipqe, ipqe_q); + TAILQ_INSERT_AFTER(&fp->ipq_fragq, p, ipqe, ipqe_q); } next = 0; - for (p = NULL, q = fp->ipq_fragq.lh_first; q != NULL; - p = q, q = q->ipqe_q.le_next) { - if (q->ipqe_ip->ip_off != next) + for (p = NULL, q = TAILQ_FIRST(&fp->ipq_fragq); q != NULL; + p = q, q = TAILQ_NEXT(q, ipqe_q)) { + if (ntohs(q->ipqe_ip->ip_off) != next) return (0); - next += q->ipqe_ip->ip_len; + next += ntohs(q->ipqe_ip->ip_len); } if (p->ipqe_mff) return (0); @@ -819,7 +1194,7 @@ insert: * Reassembly is complete. Check for a bogus message size and * concatenate fragments. */ - q = fp->ipq_fragq.lh_first; + q = TAILQ_FIRST(&fp->ipq_fragq); ip = q->ipqe_ip; if ((next + (ip->ip_hl << 2)) > IP_MAXPACKET) { ipstat.ips_toolong++; @@ -830,14 +1205,19 @@ insert: t = m->m_next; m->m_next = 0; m_cat(m, t); - nq = q->ipqe_q.le_next; + 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 = q->ipqe_q.le_next; + nq = TAILQ_NEXT(q, ipqe_q); + s = splvm(); pool_put(&ipqent_pool, q); + splx(s); m_cat(m, t); } + ip_nfrags -= fp->ipq_nfrags; /* * Create header for new ip packet by @@ -845,11 +1225,12 @@ insert: * dequeue and discard fragment reassembly header. * Make header visible. */ - ip->ip_len = next; + ip->ip_len = htons(next); ip->ip_src = fp->ipq_src; ip->ip_dst = fp->ipq_dst; LIST_REMOVE(fp, ipq_q); FREE(fp, M_FTABLE); + ip_nfragpackets--; m->m_len += (ip->ip_hl << 2); m->m_data -= (ip->ip_hl << 2); /* some debugging cruft by sklower, below, will go away soon */ @@ -858,13 +1239,19 @@ 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); dropfrag: + if (fp != 0) + fp->ipq_nfrags--; + ip_nfrags--; ipstat.ips_fragdropped++; m_freem(m); + s = splvm(); pool_put(&ipqent_pool, ipqe); + splx(s); return (0); } @@ -873,21 +1260,98 @@ 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(); - for (q = fp->ipq_fragq.lh_first; q != NULL; q = p) { - p = q->ipqe_q.le_next; + for (q = TAILQ_FIRST(&fp->ipq_fragq); q != NULL; q = p) { + p = TAILQ_NEXT(q, ipqe_q); m_freem(q->ipqe_m); - LIST_REMOVE(q, ipqe_q); + nfrags++; + TAILQ_REMOVE(&fp->ipq_fragq, q, ipqe_q); + s = splvm(); pool_put(&ipqent_pool, q); + splx(s); } + + if (nfrags != fp->ipq_nfrags) + printf("ip_freef: nfrags %d != %d\n", fp->ipq_nfrags, nfrags); + ip_nfrags -= nfrags; LIST_REMOVE(fp, ipq_q); FREE(fp, M_FTABLE); + ip_nfragpackets--; +} + +/* + * IP reassembly TTL machinery for multiplicative drop. + */ +static u_int fragttl_histo[(IPFRAGTTL+1)]; + + +/* + * Decrement TTL of all reasembly queue entries by `ticks'. + * Count number of distinct fragments (as opposed to partial, fragmented + * datagrams) in the reassembly queue. While we traverse the entire + * reassembly queue, compute and return the median TTL over all fragments. + */ +static u_int +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) ? + 0 : fp->ipq_ttl - ticks); + nfp = LIST_NEXT(fp, ipq_q); + if (fp->ipq_ttl == 0) { + ipstat.ips_fragtimeout++; + ip_freef(fp); + } else { + nfrags += fp->ipq_nfrags; + fragttl_histo[fp->ipq_ttl] += fp->ipq_nfrags; + } + } + } + + KASSERT(ip_nfrags == nfrags); + + /* Find median (or other drop fraction) in histogram. */ + dropfraction = (ip_nfrags / 2); + keepfraction = ip_nfrags - dropfraction; + for (i = IPFRAGTTL, median = 0; i >= 0; i--) { + median += fragttl_histo[i]; + if (median >= keepfraction) + break; + } + + /* Return TTL of median (or other fraction). */ + return (u_int)i; +} + +void +ip_reass_drophalf(void) +{ + + u_int median_ticks; + /* + * Compute median TTL of all fragments, and count frags + * with that TTL or lower (roughly half of all fragments). + */ + median_ticks = ip_reass_ttl_decr(0); + + /* Drop half. */ + median_ticks = ip_reass_ttl_decr(median_ticks); + } /* @@ -896,18 +1360,51 @@ ip_freef(fp) * queue, discard it. */ void -ip_slowtimo() +ip_slowtimo(void) { - struct ipq *fp, *nfp; + static u_int dropscanidx = 0; + u_int i; + u_int median_ttl; int s = splsoftnet(); IPQ_LOCK(); - for (fp = ipq.lh_first; fp != NULL; fp = nfp) { - nfp = fp->ipq_q.le_next; - if (--fp->ipq_ttl == 0) { - ipstat.ips_fragtimeout++; - ip_freef(fp); + + /* Age TTL of all fragments by 1 tick .*/ + median_ttl = ip_reass_ttl_decr(1); + + /* make sure fragment limit is up-to-date */ + CHECK_NMBCLUSTER_PARAMS(); + + /* If we have too many fragments, drop the older half. */ + if (ip_nfrags > ip_maxfrags) + ip_reass_ttl_decr(median_ttl); + + /* + * If we are over the maximum number of fragmented packets + * (due to the limit being lowered), drain off + * enough to get down to the new limit. Start draining + * from the reassembly hashqueue most recently drained. + */ + if (ip_maxfragpackets < 0) + ; + else { + int wrapped = 0; + + i = dropscanidx; + while (ip_nfragpackets > ip_maxfragpackets && wrapped == 0) { + while (LIST_FIRST(&ipq[i]) != NULL) + ip_freef(LIST_FIRST(&ipq[i])); + if (++i >= IPREASS_NHASH) { + i = 0; + } + /* + * Dont scan forever even if fragment counters are + * wrong: stop after scanning entire reassembly queue. + */ + if (i == dropscanidx) + wrapped = 1; } + dropscanidx = i; } IPQ_UNLOCK(); #ifdef GATEWAY @@ -920,7 +1417,7 @@ ip_slowtimo() * Drain off all datagram fragments. */ void -ip_drain() +ip_drain(void) { /* @@ -930,10 +1427,11 @@ ip_drain() if (ipq_lock_try() == 0) return; - while (ipq.lh_first != NULL) { - ipstat.ips_fragdropped++; - ip_freef(ipq.lh_first); - } + /* + * Drop half the total fragments now. If more mbufs are needed, + * we will be called again soon. + */ + ip_reass_drophalf(); IPQ_UNLOCK(); } @@ -946,8 +1444,7 @@ 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; @@ -1034,7 +1531,7 @@ ip_dooptions(m) bcopy((caddr_t)(cp + off), (caddr_t)&ipaddr.sin_addr, sizeof(ipaddr.sin_addr)); if (opt == IPOPT_SSRR) - ia = ifatoia(ifa_ifwithaddr(sintosa(&ipaddr))); + ia = ifatoia(ifa_ifwithladdr(sintosa(&ipaddr))); else ia = ip_rtaddr(ipaddr.sin_addr); if (ia == 0) { @@ -1176,8 +1673,7 @@ 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; @@ -1204,9 +1700,7 @@ 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; @@ -1228,7 +1722,7 @@ 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; @@ -1239,6 +1733,7 @@ ip_srcroute() if (m == 0) return ((struct mbuf *)0); + MCLAIM(m, &inetdomain.dom_mowner); #define OPTSIZ (sizeof(ip_srcrt.nop) + sizeof(ip_srcrt.srcopt)) /* length is (nhops+1)*sizeof(addr) + sizeof(nop + srcrt header) */ @@ -1299,9 +1794,7 @@ 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 *); @@ -1315,11 +1808,11 @@ ip_stripoptions(m, mopt) m->m_len -= olen; if (m->m_flags & M_PKTHDR) m->m_pkthdr.len -= olen; - ip->ip_len -= olen; + ip->ip_len = htons(ntohs(ip->ip_len) - olen); ip->ip_hl = sizeof (struct ip) >> 2; } -int inetctlerrmap[PRC_NCMDS] = { +const int inetctlerrmap[PRC_NCMDS] = { 0, 0, 0, 0, 0, EMSGSIZE, EHOSTDOWN, EHOSTUNREACH, EHOSTUNREACH, EHOSTUNREACH, ECONNREFUSED, ECONNREFUSED, @@ -1343,27 +1836,31 @@ 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; -#ifdef IPSEC - struct ifnet dummyifp; -#endif + + /* + * We are now in the output path. + */ + MCLAIM(m, &ip_tx_mowner); + + /* + * Clear any in-bound checksum flags for this packet. + */ + m->m_pkthdr.csum_flags = 0; 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++; @@ -1374,7 +1871,6 @@ 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 || @@ -1389,7 +1885,7 @@ 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; @@ -1398,8 +1894,13 @@ ip_forward(m, srcrt) /* * Save at most 68 bytes of the packet in case * we need to generate an ICMP message to the src. + * Pullup to avoid sharing mbuf cluster between m and mcopy. */ - mcopy = m_copy(m, 0, imin((int)ip->ip_len, 68)); + mcopy = m_copym(m, 0, imin(ntohs(ip->ip_len), 68), M_DONTWAIT); + 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, @@ -1434,12 +1935,10 @@ ip_forward(m, srcrt) } } -#ifdef IPSEC - /* Don't lookup socket in forwading case */ - ipsec_setsocket(m, NULL); -#endif error = ip_output(m, (struct mbuf *)0, &ipforward_rt, - (IP_FORWARDING | (ip_directedbcast ? IP_ALLOWBROADCAST : 0)), 0); + (IP_FORWARDING | (ip_directedbcast ? IP_ALLOWBROADCAST : 0)), + (struct ip_moptions *)NULL, (struct socket *)NULL); + if (error) ipstat.ips_cantforward++; else { @@ -1459,7 +1958,6 @@ ip_forward(m, srcrt) } if (mcopy == NULL) return; - destifp = NULL; switch (error) { @@ -1479,9 +1977,9 @@ ip_forward(m, srcrt) case EMSGSIZE: type = ICMP_UNREACH; code = ICMP_UNREACH_NEEDFRAG; -#ifndef IPSEC +#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 @@ -1496,41 +1994,39 @@ ip_forward(m, srcrt) struct route *ro; sp = ipsec4_getpolicybyaddr(mcopy, - IPSEC_DIR_OUTBOUND, - IP_FORWARDING, - &ipsecerror); + IPSEC_DIR_OUTBOUND, IP_FORWARDING, + &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, - IPSEC_DIR_OUTBOUND, - NULL); + IPSEC_DIR_OUTBOUND, NULL); /* * 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; } } +#ifdef IPSEC key_freesp(sp); +#else + KEY_FREESP(&sp); +#endif } } #endif /*IPSEC*/ @@ -1538,19 +2034,28 @@ ip_forward(m, srcrt) break; case ENOBUFS: +#if 1 + /* + * a router should not generate ICMP_SOURCEQUENCH as + * required in RFC1812 Requirements for IP Version 4 Routers. + * source quench could be a big problem under DoS attacks, + * or if the underlying interface is rate-limited. + */ + if (mcopy) + m_freem(mcopy); + return; +#else type = ICMP_SOURCEQUENCH; code = 0; 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) { @@ -1605,143 +2110,276 @@ ip_savecontrol(inp, mp, ip, m) } } -int -ip_sysctl(name, namelen, oldp, oldlenp, newp, newlen) - int *name; - u_int namelen; - void *oldp; - size_t *oldlenp; - void *newp; - size_t newlen; +/* + * sysctl helper routine for net.inet.ip.forwsrcrt. + */ +static int +sysctl_net_inet_ip_forwsrcrt(SYSCTLFN_ARGS) { - extern int subnetsarelocal, hostzeroisbroadcast; - - int error, old; + int error, tmp; + struct sysctlnode node; - /* All sysctl names at this level are terminal. */ - if (namelen != 1) - return (ENOTDIR); - - switch (name[0]) { - case IPCTL_FORWARDING: - return (sysctl_int(oldp, oldlenp, newp, newlen, &ipforwarding)); - case IPCTL_SENDREDIRECTS: - return (sysctl_int(oldp, oldlenp, newp, newlen, - &ipsendredirects)); - case IPCTL_DEFTTL: - return (sysctl_int(oldp, oldlenp, newp, newlen, &ip_defttl)); -#ifdef notyet - case IPCTL_DEFMTU: - return (sysctl_int(oldp, oldlenp, newp, newlen, &ip_mtu)); -#endif - case IPCTL_FORWSRCRT: - /* Don't allow this to change in a secure environment. */ - if (securelevel > 0) - return (sysctl_rdint(oldp, oldlenp, newp, - ip_forwsrcrt)); - else - return (sysctl_int(oldp, oldlenp, newp, newlen, - &ip_forwsrcrt)); - case IPCTL_DIRECTEDBCAST: - return (sysctl_int(oldp, oldlenp, newp, newlen, - &ip_directedbcast)); - case IPCTL_ALLOWSRCRT: - return (sysctl_int(oldp, oldlenp, newp, newlen, - &ip_allowsrcrt)); - case IPCTL_SUBNETSARELOCAL: - return (sysctl_int(oldp, oldlenp, newp, newlen, - &subnetsarelocal)); - case IPCTL_MTUDISC: - error = sysctl_int(oldp, oldlenp, newp, newlen, - &ip_mtudisc); - if (ip_mtudisc != 0 && ip_mtudisc_timeout_q == NULL) { - ip_mtudisc_timeout_q = - rt_timer_queue_create(ip_mtudisc_timeout); - } else if (ip_mtudisc == 0 && ip_mtudisc_timeout_q != NULL) { - rt_timer_queue_destroy(ip_mtudisc_timeout_q, TRUE); - ip_mtudisc_timeout_q = NULL; - } - return error; - case IPCTL_ANONPORTMIN: - old = anonportmin; - error = sysctl_int(oldp, oldlenp, newp, newlen, &anonportmin); - if (anonportmin >= anonportmax || anonportmin > 65535 -#ifndef IPNOPRIVPORTS - || anonportmin < IPPORT_RESERVED -#endif - ) { - anonportmin = old; - return (EINVAL); - } + node = *rnode; + tmp = ip_forwsrcrt; + node.sysctl_data = &tmp; + error = sysctl_lookup(SYSCTLFN_CALL(&node)); + if (error || newp == NULL) return (error); - case IPCTL_ANONPORTMAX: - old = anonportmax; - error = sysctl_int(oldp, oldlenp, newp, newlen, &anonportmax); - if (anonportmin >= anonportmax || anonportmax > 65535 -#ifndef IPNOPRIVPORTS - || anonportmax < IPPORT_RESERVED -#endif - ) { - anonportmax = old; - return (EINVAL); - } - return (error); - case IPCTL_MTUDISCTIMEOUT: - error = sysctl_int(oldp, oldlenp, newp, newlen, - &ip_mtudisc_timeout); - if (ip_mtudisc_timeout_q != NULL) - rt_timer_queue_change(ip_mtudisc_timeout_q, - ip_mtudisc_timeout); + + 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. + */ +static int +sysctl_net_inet_ip_pmtudto(SYSCTLFN_ARGS) +{ + int error, tmp; + struct sysctlnode node; + + node = *rnode; + tmp = ip_mtudisc_timeout; + node.sysctl_data = &tmp; + error = sysctl_lookup(SYSCTLFN_CALL(&node)); + if (error || newp == NULL) return (error); + if (tmp < 0) + return (EINVAL); + + ip_mtudisc_timeout = tmp; + rt_timer_queue_change(ip_mtudisc_timeout_q, ip_mtudisc_timeout); + + return (0); +} + #ifdef GATEWAY - case IPCTL_MAXFLOWS: - { - int s; +/* + * sysctl helper routine for net.inet.ip.maxflows. apparently if + * maxflows is even looked up, we "reap flows". + */ +static int +sysctl_net_inet_ip_maxflows(SYSCTLFN_ARGS) +{ + int s; - error = sysctl_int(oldp, oldlenp, newp, newlen, - &ip_maxflows); - s = splsoftnet(); - ipflow_reap(0); - splx(s); - return (error); - } -#endif - case IPCTL_HOSTZEROBROADCAST: - return (sysctl_int(oldp, oldlenp, newp, newlen, - &hostzeroisbroadcast)); -#if NGIF > 0 - case IPCTL_GIF_TTL: - return(sysctl_int(oldp, oldlenp, newp, newlen, - &ip_gif_ttl)); -#endif + s = sysctl_lookup(SYSCTLFN_CALL(rnode)); + if (s) + return (s); -#ifndef IPNOPRIVPORTS - case IPCTL_LOWPORTMIN: - old = lowportmin; - error = sysctl_int(oldp, oldlenp, newp, newlen, &lowportmin); - if (lowportmin >= lowportmax - || lowportmin > IPPORT_RESERVEDMAX - || lowportmin < IPPORT_RESERVEDMIN - ) { - lowportmin = old; - return (EINVAL); - } - return (error); - case IPCTL_LOWPORTMAX: - old = lowportmax; - error = sysctl_int(oldp, oldlenp, newp, newlen, &lowportmax); - if (lowportmin >= lowportmax - || lowportmax > IPPORT_RESERVEDMAX - || lowportmax < IPPORT_RESERVEDMIN - ) { - lowportmax = old; - return (EINVAL); - } - return (error); -#endif + s = splsoftnet(); + ipflow_reap(0); + splx(s); - default: - return (EOPNOTSUPP); - } - /* NOTREACHED */ + return (0); +} +#endif /* GATEWAY */ + + +SYSCTL_SETUP(sysctl_net_inet_ip_setup, "sysctl net.inet.ip subtree setup") +{ + extern int subnetsarelocal, hostzeroisbroadcast; + + sysctl_createv(clog, 0, NULL, NULL, + CTLFLAG_PERMANENT, + CTLTYPE_NODE, "net", NULL, + NULL, 0, NULL, 0, + CTL_NET, CTL_EOL); + sysctl_createv(clog, 0, NULL, NULL, + CTLFLAG_PERMANENT, + CTLTYPE_NODE, "inet", + SYSCTL_DESCR("PF_INET related settings"), + NULL, 0, NULL, 0, + CTL_NET, PF_INET, CTL_EOL); + sysctl_createv(clog, 0, NULL, NULL, + CTLFLAG_PERMANENT, + CTLTYPE_NODE, "ip", + 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", + SYSCTL_DESCR("Enable forwarding of INET datagrams"), + NULL, 0, &ipforwarding, 0, + CTL_NET, PF_INET, IPPROTO_IP, + IPCTL_FORWARDING, CTL_EOL); + sysctl_createv(clog, 0, NULL, NULL, + CTLFLAG_PERMANENT|CTLFLAG_READWRITE, + CTLTYPE_INT, "redirect", + SYSCTL_DESCR("Enable sending of ICMP redirect messages"), + NULL, 0, &ipsendredirects, 0, + CTL_NET, PF_INET, IPPROTO_IP, + IPCTL_SENDREDIRECTS, CTL_EOL); + sysctl_createv(clog, 0, NULL, NULL, + CTLFLAG_PERMANENT|CTLFLAG_READWRITE, + CTLTYPE_INT, "ttl", + SYSCTL_DESCR("Default TTL for an INET datagram"), + NULL, 0, &ip_defttl, 0, + CTL_NET, PF_INET, IPPROTO_IP, + IPCTL_DEFTTL, CTL_EOL); +#ifdef IPCTL_DEFMTU + sysctl_createv(clog, 0, NULL, NULL, + CTLFLAG_PERMANENT /* |CTLFLAG_READWRITE? */, + CTLTYPE_INT, "mtu", + SYSCTL_DESCR("Default MTA for an INET route"), + NULL, 0, &ip_mtu, 0, + CTL_NET, PF_INET, IPPROTO_IP, + IPCTL_DEFMTU, CTL_EOL); +#endif /* IPCTL_DEFMTU */ + sysctl_createv(clog, 0, NULL, NULL, + CTLFLAG_PERMANENT|CTLFLAG_READWRITE, + CTLTYPE_INT, "forwsrcrt", + SYSCTL_DESCR("Enable forwarding of source-routed " + "datagrams"), + 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, + CTLFLAG_PERMANENT|CTLFLAG_READWRITE, + CTLTYPE_INT, "directed-broadcast", + SYSCTL_DESCR("Enable forwarding of broadcast datagrams"), + NULL, 0, &ip_directedbcast, 0, + CTL_NET, PF_INET, IPPROTO_IP, + IPCTL_DIRECTEDBCAST, CTL_EOL); + sysctl_createv(clog, 0, NULL, NULL, + CTLFLAG_PERMANENT|CTLFLAG_READWRITE, + CTLTYPE_INT, "allowsrcrt", + SYSCTL_DESCR("Accept source-routed datagrams"), + NULL, 0, &ip_allowsrcrt, 0, + CTL_NET, PF_INET, IPPROTO_IP, + IPCTL_ALLOWSRCRT, CTL_EOL); + sysctl_createv(clog, 0, NULL, NULL, + CTLFLAG_PERMANENT|CTLFLAG_READWRITE, + CTLTYPE_INT, "subnetsarelocal", + SYSCTL_DESCR("Whether logical subnets are considered " + "local"), + NULL, 0, &subnetsarelocal, 0, + CTL_NET, PF_INET, IPPROTO_IP, + IPCTL_SUBNETSARELOCAL, CTL_EOL); + sysctl_createv(clog, 0, NULL, NULL, + CTLFLAG_PERMANENT|CTLFLAG_READWRITE, + CTLTYPE_INT, "mtudisc", + SYSCTL_DESCR("Use RFC1191 Path MTU Discovery"), + NULL, 0, &ip_mtudisc, 0, + CTL_NET, PF_INET, IPPROTO_IP, + IPCTL_MTUDISC, CTL_EOL); + sysctl_createv(clog, 0, NULL, NULL, + CTLFLAG_PERMANENT|CTLFLAG_READWRITE, + CTLTYPE_INT, "anonportmin", + SYSCTL_DESCR("Lowest ephemeral port number to assign"), + sysctl_net_inet_ip_ports, 0, &anonportmin, 0, + CTL_NET, PF_INET, IPPROTO_IP, + IPCTL_ANONPORTMIN, CTL_EOL); + sysctl_createv(clog, 0, NULL, NULL, + CTLFLAG_PERMANENT|CTLFLAG_READWRITE, + CTLTYPE_INT, "anonportmax", + SYSCTL_DESCR("Highest ephemeral port number to assign"), + sysctl_net_inet_ip_ports, 0, &anonportmax, 0, + CTL_NET, PF_INET, IPPROTO_IP, + IPCTL_ANONPORTMAX, CTL_EOL); + sysctl_createv(clog, 0, NULL, NULL, + CTLFLAG_PERMANENT|CTLFLAG_READWRITE, + CTLTYPE_INT, "mtudisctimeout", + SYSCTL_DESCR("Lifetime of a Path MTU Discovered route"), + sysctl_net_inet_ip_pmtudto, 0, &ip_mtudisc_timeout, 0, + CTL_NET, PF_INET, IPPROTO_IP, + IPCTL_MTUDISCTIMEOUT, CTL_EOL); +#ifdef GATEWAY + sysctl_createv(clog, 0, NULL, NULL, + CTLFLAG_PERMANENT|CTLFLAG_READWRITE, + CTLTYPE_INT, "maxflows", + SYSCTL_DESCR("Number of flows for fast forwarding"), + sysctl_net_inet_ip_maxflows, 0, &ip_maxflows, 0, + CTL_NET, PF_INET, IPPROTO_IP, + IPCTL_MAXFLOWS, CTL_EOL); +#endif /* GATEWAY */ + sysctl_createv(clog, 0, NULL, NULL, + CTLFLAG_PERMANENT|CTLFLAG_READWRITE, + CTLTYPE_INT, "hostzerobroadcast", + SYSCTL_DESCR("All zeroes address is broadcast address"), + NULL, 0, &hostzeroisbroadcast, 0, + CTL_NET, PF_INET, IPPROTO_IP, + IPCTL_HOSTZEROBROADCAST, CTL_EOL); +#if NGIF > 0 + sysctl_createv(clog, 0, NULL, NULL, + CTLFLAG_PERMANENT|CTLFLAG_READWRITE, + CTLTYPE_INT, "gifttl", + SYSCTL_DESCR("Default TTL for a gif tunnel datagram"), + NULL, 0, &ip_gif_ttl, 0, + CTL_NET, PF_INET, IPPROTO_IP, + IPCTL_GIF_TTL, CTL_EOL); +#endif /* NGIF */ +#ifndef IPNOPRIVPORTS + sysctl_createv(clog, 0, NULL, NULL, + CTLFLAG_PERMANENT|CTLFLAG_READWRITE, + CTLTYPE_INT, "lowportmin", + SYSCTL_DESCR("Lowest privileged ephemeral port number " + "to assign"), + sysctl_net_inet_ip_ports, 0, &lowportmin, 0, + CTL_NET, PF_INET, IPPROTO_IP, + IPCTL_LOWPORTMIN, CTL_EOL); + sysctl_createv(clog, 0, NULL, NULL, + CTLFLAG_PERMANENT|CTLFLAG_READWRITE, + CTLTYPE_INT, "lowportmax", + SYSCTL_DESCR("Highest privileged ephemeral port number " + "to assign"), + sysctl_net_inet_ip_ports, 0, &lowportmax, 0, + CTL_NET, PF_INET, IPPROTO_IP, + IPCTL_LOWPORTMAX, CTL_EOL); +#endif /* IPNOPRIVPORTS */ + sysctl_createv(clog, 0, NULL, NULL, + CTLFLAG_PERMANENT|CTLFLAG_READWRITE, + CTLTYPE_INT, "maxfragpackets", + SYSCTL_DESCR("Maximum number of fragments to retain for " + "possible reassembly"), + NULL, 0, &ip_maxfragpackets, 0, + CTL_NET, PF_INET, IPPROTO_IP, + IPCTL_MAXFRAGPACKETS, CTL_EOL); +#if NGRE > 0 + sysctl_createv(clog, 0, NULL, NULL, + CTLFLAG_PERMANENT|CTLFLAG_READWRITE, + CTLTYPE_INT, "grettl", + SYSCTL_DESCR("Default TTL for a gre tunnel datagram"), + NULL, 0, &ip_gre_ttl, 0, + CTL_NET, PF_INET, IPPROTO_IP, + IPCTL_GRE_TTL, CTL_EOL); +#endif /* NGRE */ + sysctl_createv(clog, 0, NULL, NULL, + CTLFLAG_PERMANENT|CTLFLAG_READWRITE, + CTLTYPE_INT, "checkinterface", + SYSCTL_DESCR("Enable receive side of Strong ES model " + "from RFC1122"), + NULL, 0, &ip_checkinterface, 0, + CTL_NET, PF_INET, IPPROTO_IP, + IPCTL_CHECKINTERFACE, CTL_EOL); + sysctl_createv(clog, 0, NULL, NULL, + CTLFLAG_PERMANENT|CTLFLAG_READWRITE, + CTLTYPE_INT, "random_id", + SYSCTL_DESCR("Assign random ip_id values"), + NULL, 0, &ip_do_randomid, 0, + CTL_NET, PF_INET, IPPROTO_IP, + IPCTL_RANDOMID, CTL_EOL); + sysctl_createv(clog, 0, NULL, NULL, + CTLFLAG_PERMANENT|CTLFLAG_READWRITE, + CTLTYPE_INT, "do_loopback_cksum", + SYSCTL_DESCR("Perform IP checksum on loopback"), + 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); }