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 1.254.2.1 retrieving revision 1.262.6.4 diff -u -p -r1.254.2.1 -r1.262.6.4 --- src/sys/netinet/ip_input.c 2007/11/13 16:02:51 1.254.2.1 +++ src/sys/netinet/ip_input.c 2008/10/05 20:11:33 1.262.6.4 @@ -1,4 +1,4 @@ -/* $NetBSD: ip_input.c,v 1.254.2.1 2007/11/13 16:02:51 bouyer Exp $ */ +/* $NetBSD: ip_input.c,v 1.262.6.4 2008/10/05 20:11:33 mjf Exp $ */ /* * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project. @@ -45,13 +45,6 @@ * 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 NetBSD - * Foundation, Inc. and its contributors. - * 4. Neither the name of The NetBSD Foundation 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 NETBSD FOUNDATION, INC. AND CONTRIBUTORS * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED @@ -98,7 +91,7 @@ */ #include -__KERNEL_RCSID(0, "$NetBSD: ip_input.c,v 1.254.2.1 2007/11/13 16:02:51 bouyer Exp $"); +__KERNEL_RCSID(0, "$NetBSD: ip_input.c,v 1.262.6.4 2008/10/05 20:11:33 mjf Exp $"); #include "opt_inet.h" #include "opt_gateway.h" @@ -135,6 +128,7 @@ __KERNEL_RCSID(0, "$NetBSD: ip_input.c,v #include #include #include +#include #include /* just for gif_ttl */ #include @@ -148,6 +142,7 @@ __KERNEL_RCSID(0, "$NetBSD: ip_input.c,v #ifdef IPSEC #include +#include #include #endif #ifdef FAST_IPSEC @@ -230,9 +225,10 @@ u_long in_multihash; /* size of hash int in_multientries; /* total number of addrs */ struct in_multihashhead *in_multihashtbl; struct ifqueue ipintrq; -struct ipstat ipstat; uint16_t ip_id; +percpu_t *ipstat_percpu; + #ifdef PFIL_HOOKS struct pfil_head inet_pfil_hook; #endif @@ -334,10 +330,8 @@ do { \ #define IPQ_UNLOCK() ipq_unlock() -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); +struct pool inmulti_pool; +struct pool ipqent_pool; #ifdef INET_CSUM_COUNTERS #include @@ -403,6 +397,11 @@ ip_init(void) const struct protosw *pr; int i; + 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); + pr = pffindproto(PF_INET, IPPROTO_RAW, SOCK_RAW); if (pr == 0) panic("ip_init"); @@ -417,16 +416,17 @@ ip_init(void) for (i = 0; i < IPREASS_NHASH; i++) LIST_INIT(&ipq[i]); + ip_initid(); ip_id = time_second & 0xfffff; ipintrq.ifq_maxlen = ipqmaxlen; 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); + in_ifaddrhashtbl = hashinit(IN_IFADDR_HASH_SIZE, HASH_LIST, true, + &in_ifaddrhash); + in_multihashtbl = hashinit(IN_IFADDR_HASH_SIZE, HASH_LIST, true, + &in_multihash); ip_mtudisc_timeout_q = rt_timer_queue_create(ip_mtudisc_timeout); #ifdef GATEWAY ipflow_init(ip_hashsize); @@ -446,6 +446,8 @@ ip_init(void) MOWNER_ATTACH(&ip_tx_mowner); MOWNER_ATTACH(&ip_rx_mowner); #endif /* MBUFTRACE */ + + ipstat_percpu = percpu_alloc(sizeof(uint64_t) * IP_NSTATS); } struct sockaddr_in ipaddr = { @@ -463,14 +465,18 @@ ipintr(void) int s; struct mbuf *m; + mutex_enter(softnet_lock); + KERNEL_LOCK(1, NULL); while (!IF_IS_EMPTY(&ipintrq)) { s = splnet(); IF_DEQUEUE(&ipintrq, m); splx(s); - if (m == 0) - return; + if (m == NULL) + break; ip_input(m); } + KERNEL_UNLOCK_ONE(NULL); + mutex_exit(softnet_lock); } /* @@ -510,7 +516,7 @@ ip_input(struct mbuf *m) */ if (TAILQ_FIRST(&in_ifaddrhead) == 0) goto bad; - ipstat.ips_total++; + IP_STATINC(IP_STAT_TOTAL); /* * If the IP header is not aligned, slurp it up into a new * mbuf with space for link headers, in the event we forward @@ -521,28 +527,28 @@ ip_input(struct mbuf *m) if ((m = m_copyup(m, sizeof(struct ip), (max_linkhdr + 3) & ~3)) == NULL) { /* XXXJRT new stat, please */ - ipstat.ips_toosmall++; + IP_STATINC(IP_STAT_TOOSMALL); return; } } else if (__predict_false(m->m_len < sizeof (struct ip))) { if ((m = m_pullup(m, sizeof (struct ip))) == NULL) { - ipstat.ips_toosmall++; + IP_STATINC(IP_STAT_TOOSMALL); return; } } ip = mtod(m, struct ip *); if (ip->ip_v != IPVERSION) { - ipstat.ips_badvers++; + IP_STATINC(IP_STAT_BADVERS); goto bad; } hlen = ip->ip_hl << 2; if (hlen < sizeof(struct ip)) { /* minimum header length */ - ipstat.ips_badhlen++; + IP_STATINC(IP_STAT_BADHLEN); goto bad; } if (hlen > m->m_len) { if ((m = m_pullup(m, hlen)) == 0) { - ipstat.ips_badhlen++; + IP_STATINC(IP_STAT_BADHLEN); return; } ip = mtod(m, struct ip *); @@ -553,7 +559,7 @@ ip_input(struct mbuf *m) * not allowed. */ if (IN_MULTICAST(ip->ip_src.s_addr)) { - ipstat.ips_badaddr++; + IP_STATINC(IP_STAT_BADADDR); goto bad; } @@ -561,7 +567,7 @@ ip_input(struct mbuf *m) 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++; + IP_STATINC(IP_STAT_BADADDR); goto bad; } } @@ -599,7 +605,7 @@ ip_input(struct mbuf *m) * Check for additional length bogosity */ if (len < hlen) { - ipstat.ips_badlen++; + IP_STATINC(IP_STAT_BADLEN); goto bad; } @@ -610,7 +616,7 @@ ip_input(struct mbuf *m) * Drop packet if shorter than we expect. */ if (m->m_pkthdr.len < len) { - ipstat.ips_tooshort++; + IP_STATINC(IP_STAT_TOOSHORT); goto bad; } if (m->m_pkthdr.len > len) { @@ -779,7 +785,7 @@ ip_input(struct mbuf *m) * ip_output().) */ if (ip_mforward(m, m->m_pkthdr.rcvif) != 0) { - ipstat.ips_cantforward++; + IP_STATINC(IP_STAT_CANTFORWARD); m_freem(m); return; } @@ -791,7 +797,7 @@ ip_input(struct mbuf *m) */ if (ip->ip_p == IPPROTO_IGMP) goto ours; - ipstat.ips_forward++; + IP_STATINC(IP_STAT_CANTFORWARD); } #endif /* @@ -800,7 +806,7 @@ ip_input(struct mbuf *m) */ IN_LOOKUP_MULTI(ip->ip_dst, m->m_pkthdr.rcvif, inm); if (inm == NULL) { - ipstat.ips_cantforward++; + IP_STATINC(IP_STAT_CANTFORWARD); m_freem(m); return; } @@ -814,7 +820,7 @@ ip_input(struct mbuf *m) * Not for us; forward if possible and desirable. */ if (ipforwarding == 0) { - ipstat.ips_cantforward++; + IP_STATINC(IP_STAT_CANTFORWARD); m_freem(m); } else { /* @@ -825,12 +831,12 @@ ip_input(struct mbuf *m) */ if (downmatch) { icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_HOST, 0, 0); - ipstat.ips_cantforward++; + IP_STATINC(IP_STAT_CANTFORWARD); return; } #ifdef IPSEC if (ipsec4_in_reject(m, NULL)) { - ipsecstat.in_polvio++; + IPSEC_STATINC(IPSEC_STAT_IN_POLVIO); goto bad; } #endif @@ -858,7 +864,7 @@ ip_input(struct mbuf *m) KEY_FREESP(&sp); splx(s); if (error) { - ipstat.ips_cantforward++; + IP_STATINC(IP_STAT_CANTFORWARD); goto bad; } @@ -896,7 +902,18 @@ ours: * but it's not worth the time; just let them time out.) */ if (ip->ip_off & ~htons(IP_DF|IP_RF)) { - + uint16_t off; + /* + * Prevent TCP blind data attacks by not allowing non-initial + * fragments to start at less than 68 bytes (minimal fragment + * size) and making sure the first fragment is at least 68 + * bytes. + */ + off = (ntohs(ip->ip_off) & IP_OFFMASK) << 3; + if ((off > 0 ? off + hlen : len) < IP_MINFRAGSIZE - 1) { + IP_STATINC(IP_STAT_BADFRAGS); + goto bad; + } /* * Look for queue of fragments * of this datagram. @@ -907,9 +924,17 @@ ours: 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) + ip->ip_p == fp->ipq_p) { + /* + * Make sure the TOS is matches previous + * fragments. + */ + if (ip->ip_tos != fp->ipq_tos) { + IP_STATINC(IP_STAT_BADFRAGS); + goto bad; + } goto found; - + } } fp = 0; found: @@ -928,7 +953,7 @@ found: */ if (ntohs(ip->ip_len) == 0 || (ntohs(ip->ip_len) & 0x7) != 0) { - ipstat.ips_badfrags++; + IP_STATINC(IP_STAT_BADFRAGS); IPQ_UNLOCK(); goto bad; } @@ -941,12 +966,12 @@ found: * attempt reassembly; if it succeeds, proceed. */ if (mff || ip->ip_off != htons(0)) { - ipstat.ips_fragments++; + IP_STATINC(IP_STAT_FRAGMENTS); s = splvm(); ipqe = pool_get(&ipqent_pool, PR_NOWAIT); splx(s); if (ipqe == NULL) { - ipstat.ips_rcvmemdrop++; + IP_STATINC(IP_STAT_RCVMEMDROP); IPQ_UNLOCK(); goto bad; } @@ -958,7 +983,7 @@ found: IPQ_UNLOCK(); return; } - ipstat.ips_reassembled++; + IP_STATINC(IP_STAT_REASSEMBLED); ip = mtod(m, struct ip *); hlen = ip->ip_hl << 2; ip->ip_len = htons(ntohs(ip->ip_len) + hlen); @@ -976,7 +1001,7 @@ found: */ if ((inetsw[ip_protox[ip->ip_p]].pr_flags & PR_LASTHDR) != 0 && ipsec4_in_reject(m, NULL)) { - ipsecstat.in_polvio++; + IPSEC_STATINC(IPSEC_STAT_IN_POLVIO); goto bad; } #endif @@ -1026,7 +1051,7 @@ DPRINTF(("ip_input: no SP, packet discar if (ia && ip) ia->ia_ifa.ifa_data.ifad_inbytes += ntohs(ip->ip_len); #endif - ipstat.ips_delivered++; + IP_STATINC(IP_STAT_DELIVERED); { int off = hlen, nh = ip->ip_p; @@ -1038,7 +1063,7 @@ bad: return; badcsum: - ipstat.ips_badsum++; + IP_STATINC(IP_STAT_BADSUM); m_freem(m); } @@ -1105,6 +1130,7 @@ ip_reass(struct ipqent *ipqe, struct ipq fp->ipq_ttl = IPFRAGTTL; fp->ipq_p = ipqe->ipqe_ip->ip_p; fp->ipq_id = ipqe->ipqe_ip->ip_id; + fp->ipq_tos = ipqe->ipqe_ip->ip_tos; TAILQ_INIT(&fp->ipq_fragq); fp->ipq_src = ipqe->ipqe_ip->ip_src; fp->ipq_dst = ipqe->ipqe_ip->ip_dst; @@ -1195,7 +1221,7 @@ insert: q = TAILQ_FIRST(&fp->ipq_fragq); ip = q->ipqe_ip; if ((next + (ip->ip_hl << 2)) > IP_MAXPACKET) { - ipstat.ips_toolong++; + IP_STATINC(IP_STAT_TOOLONG); ip_freef(fp); return (0); } @@ -1245,7 +1271,7 @@ dropfrag: if (fp != 0) fp->ipq_nfrags--; ip_nfrags--; - ipstat.ips_fragdropped++; + IP_STATINC(IP_STAT_FRAGDROPPED); m_freem(m); s = splvm(); pool_put(&ipqent_pool, ipqe); @@ -1312,7 +1338,7 @@ ip_reass_ttl_decr(u_int ticks) 0 : fp->ipq_ttl - ticks); nfp = LIST_NEXT(fp, ipq_q); if (fp->ipq_ttl == 0) { - ipstat.ips_fragtimeout++; + IP_STATINC(IP_STAT_FRAGTIMEOUT); ip_freef(fp); } else { nfrags += fp->ipq_nfrags; @@ -1363,7 +1389,9 @@ ip_slowtimo(void) static u_int dropscanidx = 0; u_int i; u_int median_ttl; - int s = splsoftnet(); + + mutex_enter(softnet_lock); + KERNEL_LOCK(1, NULL); IPQ_LOCK(); @@ -1405,30 +1433,35 @@ ip_slowtimo(void) dropscanidx = i; } IPQ_UNLOCK(); - splx(s); + + KERNEL_UNLOCK_ONE(NULL); + mutex_exit(softnet_lock); } /* - * Drain off all datagram fragments. + * Drain off all datagram fragments. Don't acquire softnet_lock as + * can be called from hardware interrupt context. */ void ip_drain(void) { + KERNEL_LOCK(1, NULL); + /* * We may be called from a device's interrupt context. If * the ipq is already busy, just bail out now. */ - if (ipq_lock_try() == 0) - return; - - /* - * Drop half the total fragments now. If more mbufs are needed, - * we will be called again soon. - */ - ip_reass_drophalf(); + if (ipq_lock_try() != 0) { + /* + * Drop half the total fragments now. If more mbufs are + * needed, we will be called again soon. + */ + ip_reass_drophalf(); + IPQ_UNLOCK(); + } - IPQ_UNLOCK(); + KERNEL_UNLOCK_ONE(NULL); } /* @@ -1659,7 +1692,7 @@ ip_dooptions(struct mbuf *m) return (0); bad: icmp_error(m, type, code, 0, 0); - ipstat.ips_badoptions++; + IP_STATINC(IP_STAT_BADOPTIONS); return (1); } @@ -1776,12 +1809,15 @@ ip_srcroute(void) } 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, }; /* @@ -1829,7 +1865,7 @@ ip_forward(struct mbuf *m, int srcrt) } #endif if (m->m_flags & (M_BCAST|M_MCAST) || in_canforward(ip->ip_dst) == 0) { - ipstat.ips_cantforward++; + IP_STATINC(IP_STAT_CANTFORWARD); m_freem(m); return; } @@ -1893,12 +1929,15 @@ ip_forward(struct mbuf *m, int srcrt) (struct ip_moptions *)NULL, (struct socket *)NULL); if (error) - ipstat.ips_cantforward++; + IP_STATINC(IP_STAT_CANTFORWARD); else { - ipstat.ips_forward++; - if (type) - ipstat.ips_redirectsent++; - else { + uint64_t *ips = IP_STAT_GETREF(); + ips[IP_STAT_FORWARD]++; + if (type) { + ips[IP_STAT_REDIRECTSENT]++; + IP_STAT_PUTREF(); + } else { + IP_STAT_PUTREF(); if (mcopy) { #ifdef GATEWAY if (mcopy->m_flags & M_CANFASTFWD) @@ -1930,17 +1969,19 @@ ip_forward(struct mbuf *m, int srcrt) case EMSGSIZE: type = ICMP_UNREACH; code = ICMP_UNREACH_NEEDFRAG; -#if !defined(IPSEC) && !defined(FAST_IPSEC) - if (ipforward_rt.ro_rt != NULL) - destmtu = ipforward_rt.ro_rt->rt_ifp->if_mtu; -#else - /* - * If the packet is routed over IPsec tunnel, tell the - * originator the tunnel MTU. - * tunnel MTU = if MTU - sizeof(IP) - ESP/AH hdrsiz - * XXX quickhack!!! - */ - if (ipforward_rt.ro_rt != NULL) { + + if ((rt = rtcache_validate(&ipforward_rt)) != NULL) + destmtu = rt->rt_ifp->if_mtu; + +#if defined(IPSEC) || defined(FAST_IPSEC) + { + /* + * If the packet is routed over IPsec tunnel, tell the + * originator the tunnel MTU. + * tunnel MTU = if MTU - sizeof(IP) - ESP/AH hdrsiz + * XXX quickhack!!! + */ + struct secpolicy *sp; int ipsecerror; size_t ipsechdr; @@ -1950,9 +1991,7 @@ ip_forward(struct mbuf *m, int srcrt) IPSEC_DIR_OUTBOUND, IP_FORWARDING, &ipsecerror); - if (sp == NULL) - destmtu = ipforward_rt.ro_rt->rt_ifp->if_mtu; - else { + if (sp != NULL) { /* count IPsec header size */ ipsechdr = ipsec4_hdrsiz(mcopy, IPSEC_DIR_OUTBOUND, NULL); @@ -1966,11 +2005,12 @@ ip_forward(struct mbuf *m, int srcrt) && 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) { + rt = rtcache_validate(ro); + if (rt && rt->rt_ifp) { destmtu = - ro->ro_rt->rt_rmx.rmx_mtu ? - ro->ro_rt->rt_rmx.rmx_mtu : - ro->ro_rt->rt_ifp->if_mtu; + rt->rt_rmx.rmx_mtu ? + rt->rt_rmx.rmx_mtu : + rt->rt_ifp->if_mtu; destmtu -= ipsechdr; } } @@ -1982,8 +2022,8 @@ ip_forward(struct mbuf *m, int srcrt) #endif } } -#endif /*IPSEC*/ - ipstat.ips_cantfrag++; +#endif /*defined(IPSEC) || defined(FAST_IPSEC)*/ + IP_STATINC(IP_STAT_CANTFRAG); break; case ENOBUFS: @@ -2106,9 +2146,13 @@ sysctl_net_inet_ip_pmtudto(SYSCTLFN_ARGS if (tmp < 0) return (EINVAL); + mutex_enter(softnet_lock); + ip_mtudisc_timeout = tmp; rt_timer_queue_change(ip_mtudisc_timeout_q, ip_mtudisc_timeout); + mutex_exit(softnet_lock); + return (0); } @@ -2119,15 +2163,19 @@ sysctl_net_inet_ip_pmtudto(SYSCTLFN_ARGS static int sysctl_net_inet_ip_maxflows(SYSCTLFN_ARGS) { - int s; + int error; - s = sysctl_lookup(SYSCTLFN_CALL(rnode)); - if (s || newp == NULL) - return (s); + error = sysctl_lookup(SYSCTLFN_CALL(rnode)); + if (error || newp == NULL) + return (error); - s = splsoftnet(); - ipflow_reap(0); - splx(s); + mutex_enter(softnet_lock); + KERNEL_LOCK(1, NULL); + + ipflow_prune(); + + KERNEL_UNLOCK_ONE(NULL); + mutex_exit(softnet_lock); return (0); } @@ -2149,19 +2197,31 @@ sysctl_net_inet_ip_hashsize(SYSCTLFN_ARG /* * Can only fail due to malloc() */ - if (ipflow_invalidate_all(tmp)) - return ENOMEM; + mutex_enter(softnet_lock); + KERNEL_LOCK(1, NULL); + + error = ipflow_invalidate_all(tmp); + + KERNEL_UNLOCK_ONE(NULL); + mutex_exit(softnet_lock); + } else { /* * EINVAL if not a power of 2 */ - return EINVAL; + error = EINVAL; } - return (0); + return error; } #endif /* GATEWAY */ +static int +sysctl_net_inet_ip_stats(SYSCTLFN_ARGS) +{ + + return (NETSTAT_SYSCTL(ipstat_percpu, IP_NSTATS)); +} SYSCTL_SETUP(sysctl_net_inet_ip_setup, "sysctl net.inet.ip subtree setup") { @@ -2366,7 +2426,15 @@ SYSCTL_SETUP(sysctl_net_inet_ip_setup, " CTLFLAG_PERMANENT, CTLTYPE_STRUCT, "stats", SYSCTL_DESCR("IP statistics"), - NULL, 0, &ipstat, sizeof(ipstat), + sysctl_net_inet_ip_stats, 0, NULL, 0, CTL_NET, PF_INET, IPPROTO_IP, IPCTL_STATS, CTL_EOL); } + +void +ip_statinc(u_int stat) +{ + + KASSERT(stat < IP_NSTATS); + IP_STATINC(stat); +}