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/netinet6/ip6_input.c,v rcsdiff: /ftp/cvs/cvsroot/src/sys/netinet6/ip6_input.c,v: warning: Unknown phrases like `commitid ...;' are present. retrieving revision 1.90.2.1.2.2 retrieving revision 1.101 diff -u -p -r1.90.2.1.2.2 -r1.101 --- src/sys/netinet6/ip6_input.c 2007/09/23 21:36:35 1.90.2.1.2.2 +++ src/sys/netinet6/ip6_input.c 2007/03/24 00:42:14 1.101 @@ -1,4 +1,4 @@ -/* $NetBSD: ip6_input.c,v 1.90.2.1.2.2 2007/09/23 21:36:35 wrstuden Exp $ */ +/* $NetBSD: ip6_input.c,v 1.101 2007/03/24 00:42:14 liamjfoy Exp $ */ /* $KAME: ip6_input.c,v 1.188 2001/03/29 05:34:31 itojun Exp $ */ /* @@ -62,7 +62,7 @@ */ #include -__KERNEL_RCSID(0, "$NetBSD: ip6_input.c,v 1.90.2.1.2.2 2007/09/23 21:36:35 wrstuden Exp $"); +__KERNEL_RCSID(0, "$NetBSD: ip6_input.c,v 1.101 2007/03/24 00:42:14 liamjfoy Exp $"); #include "opt_inet.h" #include "opt_inet6.h" @@ -182,6 +182,9 @@ ip6_init() ip6_desync_factor = arc4random() % MAX_TEMP_DESYNC_FACTOR; ip6_init2((void *)0); +#ifdef GATEWAY + ip6flow_init(ip6_hashsize); +#endif #ifdef PFIL_HOOKS /* Register our Packet Filter hook. */ @@ -237,8 +240,7 @@ ip6intr() extern struct route_in6 ip6_forward_rt; void -ip6_input(m) - struct mbuf *m; +ip6_input(struct mbuf *m) { struct ip6_hdr *ip6; int off = sizeof(struct ip6_hdr), nest; @@ -277,7 +279,7 @@ ip6_input(m) else ip6stat.ip6s_mext1++; } else { -#define M2MMAX (sizeof(ip6stat.ip6s_m2m)/sizeof(ip6stat.ip6s_m2m[0])) +#define M2MMAX __arraycount(ip6stat.ip6s_m2m) if (m->m_next) { if (m->m_flags & M_LOOP) { ip6stat.ip6s_m2m[lo0ifp->if_index]++; /* XXX */ @@ -299,7 +301,7 @@ ip6_input(m) * it. Otherwise, if it is aligned, make sure the entire base * IPv6 header is in the first mbuf of the chain. */ - if (IP6_HDR_ALIGNED_P(mtod(m, caddr_t)) == 0) { + if (IP6_HDR_ALIGNED_P(mtod(m, void *)) == 0) { struct ifnet *inifp = m->m_pkthdr.rcvif; if ((m = m_copyup(m, sizeof(struct ip6_hdr), (max_linkhdr + 3) & ~3)) == NULL) { @@ -325,6 +327,17 @@ ip6_input(m) goto bad; } +#if defined(IPSEC) + /* IPv6 fast forwarding is not compatible with IPsec. */ + m->m_flags &= ~M_CANFASTFWD; +#else + /* + * Assume that we can create a fast-forward IP flow entry + * based on this packet. + */ + m->m_flags |= M_CANFASTFWD; +#endif + #ifdef PFIL_HOOKS /* * Run through list of hooks for input packets. If there are any @@ -464,28 +477,26 @@ ip6_input(m) /* * Unicast check */ - if (ip6_forward_rt.ro_rt != NULL && - (ip6_forward_rt.ro_rt->rt_flags & RTF_UP) != 0 && - IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst, - &((struct sockaddr_in6 *)(&ip6_forward_rt.ro_dst))->sin6_addr)) + if (!IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst, + &((const struct sockaddr_in6 *)rtcache_getdst((const struct route *)&ip6_forward_rt))->sin6_addr)) + rtcache_free((struct route *)&ip6_forward_rt); + else + rtcache_check((struct route *)&ip6_forward_rt); + if (ip6_forward_rt.ro_rt != NULL) { + /* XXX Revalidated route is accounted wrongly. */ ip6stat.ip6s_forward_cachehit++; - else { + } else { struct sockaddr_in6 *dst6; - if (ip6_forward_rt.ro_rt) { - /* route is down or destination is different */ - ip6stat.ip6s_forward_cachemiss++; - RTFREE(ip6_forward_rt.ro_rt); - ip6_forward_rt.ro_rt = 0; - } + ip6stat.ip6s_forward_cachemiss++; - bzero(&ip6_forward_rt.ro_dst, sizeof(struct sockaddr_in6)); - dst6 = (struct sockaddr_in6 *)&ip6_forward_rt.ro_dst; + dst6 = &ip6_forward_rt.ro_dst; + memset(dst6, 0, sizeof(*dst6)); dst6->sin6_len = sizeof(struct sockaddr_in6); dst6->sin6_family = AF_INET6; dst6->sin6_addr = ip6->ip6_dst; - rtalloc((struct route *)&ip6_forward_rt); + rtcache_init((struct route *)&ip6_forward_rt); } #define rt6_key(r) ((struct sockaddr_in6 *)((r)->rt_nodes->rn_key)) @@ -499,7 +510,7 @@ ip6_input(m) * But we think it's even useful in some situations, e.g. when using * a special daemon which wants to intercept the packet. */ - if (ip6_forward_rt.ro_rt && + if (ip6_forward_rt.ro_rt != NULL && (ip6_forward_rt.ro_rt->rt_flags & (RTF_HOST|RTF_GATEWAY)) == RTF_HOST && !(ip6_forward_rt.ro_rt->rt_flags & RTF_CLONED) && @@ -542,7 +553,8 @@ ip6_input(m) */ #if defined(NFAITH) && 0 < NFAITH if (ip6_keepfaith) { - if (ip6_forward_rt.ro_rt && ip6_forward_rt.ro_rt->rt_ifp && + if (ip6_forward_rt.ro_rt != NULL && + ip6_forward_rt.ro_rt->rt_ifp != NULL && ip6_forward_rt.ro_rt->rt_ifp->if_type == IFT_FAITH) { /* XXX do we need more sanity checks? */ ours = 1; @@ -560,9 +572,7 @@ ip6_input(m) * working right. */ struct ifaddr *ifa; - for (ifa = m->m_pkthdr.rcvif->if_addrlist.tqh_first; - ifa; - ifa = ifa->ifa_list.tqe_next) { + TAILQ_FOREACH(ifa, &m->m_pkthdr.rcvif->if_addrlist, ifa_list) { if (ifa->ifa_addr == NULL) continue; /* just for safety */ if (ifa->ifa_addr->sa_family != AF_INET6) @@ -643,7 +653,7 @@ ip6_input(m) in6_ifstat_inc(m->m_pkthdr.rcvif, ifs6_in_hdrerr); icmp6_error(m, ICMP6_PARAM_PROB, ICMP6_PARAMPROB_HEADER, - (caddr_t)&ip6->ip6_plen - (caddr_t)ip6); + (char *)&ip6->ip6_plen - (char *)ip6); return; } IP6_EXTHDR_GET(hbh, struct ip6_hbh *, m, sizeof(struct ip6_hdr), @@ -802,6 +812,7 @@ ip6_input(m) /* XXX error stat??? */ error = EINVAL; DPRINTF(("ip6_input: no SP, packet discarded\n"));/*XXX*/ + goto bad; } splx(s); if (error) @@ -939,7 +950,7 @@ ip6_process_hopopts(m, opthead, hbhlen, return (-1); } optlen = IP6OPT_RTALERT_LEN; - bcopy((caddr_t)(opt + 2), (caddr_t)&rtalert_val, 2); + bcopy((void *)(opt + 2), (void *)&rtalert_val, 2); *rtalertp = ntohs(rtalert_val); break; case IP6OPT_JUMBO: @@ -1098,7 +1109,7 @@ ip6_savecontrol(in6p, mp, ip6, m) 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; @@ -1117,7 +1128,7 @@ ip6_savecontrol(in6p, mp, ip6, m) in6_clearscope(&pi6.ipi6_addr); /* XXX */ pi6.ipi6_ifindex = m->m_pkthdr.rcvif ? m->m_pkthdr.rcvif->if_index : 0; - *mp = sbcreatecontrol((caddr_t) &pi6, + *mp = sbcreatecontrol((void *) &pi6, sizeof(struct in6_pktinfo), IS2292(IPV6_2292PKTINFO, IPV6_PKTINFO), IPPROTO_IPV6); if (*mp) @@ -1127,7 +1138,7 @@ ip6_savecontrol(in6p, mp, ip6, m) if (in6p->in6p_flags & IN6P_HOPLIMIT) { int hlim = ip6->ip6_hlim & 0xff; - *mp = sbcreatecontrol((caddr_t) &hlim, sizeof(int), + *mp = sbcreatecontrol((void *) &hlim, sizeof(int), IS2292(IPV6_2292HOPLIMIT, IPV6_HOPLIMIT), IPPROTO_IPV6); if (*mp) mp = &(*mp)->m_next; @@ -1141,7 +1152,7 @@ ip6_savecontrol(in6p, mp, ip6, m) flowinfo >>= 20; tclass = flowinfo & 0xff; - *mp = sbcreatecontrol((caddr_t)&tclass, sizeof(tclass), + *mp = sbcreatecontrol((void *)&tclass, sizeof(tclass), IPV6_TCLASS, IPPROTO_IPV6); if (*mp) @@ -1189,7 +1200,7 @@ ip6_savecontrol(in6p, mp, ip6, m) * be removed before returning in the RFC 2292. * Note: this constraint is removed in RFC3542. */ - *mp = sbcreatecontrol((caddr_t)hbh, hbhlen, + *mp = sbcreatecontrol((void *)hbh, hbhlen, IS2292(IPV6_2292HOPOPTS, IPV6_HOPOPTS), IPPROTO_IPV6); if (*mp) @@ -1251,7 +1262,7 @@ ip6_savecontrol(in6p, mp, ip6, m) if (!in6p->in6p_flags & IN6P_DSTOPTS) break; - *mp = sbcreatecontrol((caddr_t)ip6e, elen, + *mp = sbcreatecontrol((void *)ip6e, elen, IS2292(IPV6_2292DSTOPTS, IPV6_DSTOPTS), IPPROTO_IPV6); if (*mp) @@ -1262,7 +1273,7 @@ ip6_savecontrol(in6p, mp, ip6, m) if (!in6p->in6p_flags & IN6P_RTHDR) break; - *mp = sbcreatecontrol((caddr_t)ip6e, elen, + *mp = sbcreatecontrol((void *)ip6e, elen, IS2292(IPV6_2292RTHDR, IPV6_RTHDR), IPPROTO_IPV6); if (*mp) @@ -1300,7 +1311,8 @@ ip6_savecontrol(in6p, mp, ip6, m) void -ip6_notify_pmtu(struct in6pcb *in6p, struct sockaddr_in6 *dst, uint32_t *mtu) +ip6_notify_pmtu(struct in6pcb *in6p, const struct sockaddr_in6 *dst, + uint32_t *mtu) { struct socket *so; struct mbuf *m_mtu; @@ -1322,11 +1334,11 @@ ip6_notify_pmtu(struct in6pcb *in6p, str if (sa6_recoverscope(&mtuctl.ip6m_addr)) return; - if ((m_mtu = sbcreatecontrol((caddr_t)&mtuctl, sizeof(mtuctl), + if ((m_mtu = sbcreatecontrol((void *)&mtuctl, sizeof(mtuctl), IPV6_PATHMTU, IPPROTO_IPV6)) == NULL) return; - if (sbappendaddr(&so->so_rcv, (struct sockaddr *)dst, NULL, m_mtu) + if (sbappendaddr(&so->so_rcv, (const struct sockaddr *)dst, NULL, m_mtu) == 0) { m_freem(m_mtu); /* XXX: should count statistics */ @@ -1362,7 +1374,7 @@ ip6_pullexthdr(m, off, nxt) } #endif - m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e); + m_copydata(m, off, sizeof(ip6e), (void *)&ip6e); if (nxt == IPPROTO_AH) elen = (ip6e.ip6e_len + 2) << 2; else @@ -1385,7 +1397,7 @@ ip6_pullexthdr(m, off, nxt) return NULL; } - m_copydata(m, off, elen, mtod(n, caddr_t)); + m_copydata(m, off, elen, mtod(n, void *)); n->m_len = elen; return n; } @@ -1418,7 +1430,7 @@ ip6_get_prevhdr(m, off) nxt = ip6->ip6_nxt; len = sizeof(struct ip6_hdr); while (len < off) { - ip6e = (struct ip6_ext *)(mtod(m, caddr_t) + len); + ip6e = (struct ip6_ext *)(mtod(m, char *) + len); switch (nxt) { case IPPROTO_FRAGMENT: @@ -1467,7 +1479,7 @@ ip6_nexthdr(m, off, proto, nxtp) return -1; if (m->m_pkthdr.len < off + sizeof(ip6)) return -1; - m_copydata(m, off, sizeof(ip6), (caddr_t)&ip6); + m_copydata(m, off, sizeof(ip6), (void *)&ip6); if (nxtp) *nxtp = ip6.ip6_nxt; off += sizeof(ip6); @@ -1480,7 +1492,7 @@ ip6_nexthdr(m, off, proto, nxtp) */ if (m->m_pkthdr.len < off + sizeof(fh)) return -1; - m_copydata(m, off, sizeof(fh), (caddr_t)&fh); + m_copydata(m, off, sizeof(fh), (void *)&fh); if ((fh.ip6f_offlg & IP6F_OFF_MASK) != 0) return -1; if (nxtp) @@ -1491,7 +1503,7 @@ ip6_nexthdr(m, off, proto, nxtp) case IPPROTO_AH: if (m->m_pkthdr.len < off + sizeof(ip6e)) return -1; - m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e); + m_copydata(m, off, sizeof(ip6e), (void *)&ip6e); if (nxtp) *nxtp = ip6e.ip6e_nxt; off += (ip6e.ip6e_len + 2) << 2; @@ -1504,7 +1516,7 @@ ip6_nexthdr(m, off, proto, nxtp) case IPPROTO_DSTOPTS: if (m->m_pkthdr.len < off + sizeof(ip6e)) return -1; - m_copydata(m, off, sizeof(ip6e), (caddr_t)&ip6e); + m_copydata(m, off, sizeof(ip6e), (void *)&ip6e); if (nxtp) *nxtp = ip6e.ip6e_nxt; off += (ip6e.ip6e_len + 1) << 3; @@ -1593,30 +1605,56 @@ ip6_delaux(m) m_tag_delete(m, mtag); } +#ifdef GATEWAY +/* + * sysctl helper routine for net.inet.ip6.maxflows. Since + * we could reduce this value, call ip6flow_reap(); + */ +static int +sysctl_net_inet6_ip6_maxflows(SYSCTLFN_ARGS) +{ + int s; + + s = sysctl_lookup(SYSCTLFN_CALL(rnode)); + if (s || newp == NULL) + return (s); + + s = splsoftnet(); + ip6flow_reap(0); + splx(s); + + return (0); +} + static int -sysctl_net_inet6_ip6_rht0(SYSCTLFN_ARGS) +sysctl_net_inet6_ip6_hashsize(SYSCTLFN_ARGS) { int error, tmp; struct sysctlnode node; node = *rnode; - tmp = ip6_rht0; + tmp = ip6_hashsize; node.sysctl_data = &tmp; error = sysctl_lookup(SYSCTLFN_CALL(&node)); if (error || newp == NULL) - return error; + return (error); - switch (tmp) { - case -1: /* disable processing */ - case 0: /* disable for host, enable for router */ - case 1: /* enable for all */ - break; - default: + if ((tmp & (tmp - 1)) == 0 && tmp != 0) { + /* + * Can only fail due to malloc() + */ + if (ip6flow_invalidate_all(tmp)) + return ENOMEM; + } else { + /* + * EINVAL if not a power of 2 + */ return EINVAL; - } - ip6_rht0 = tmp; - return 0; + } + + return (0); } +#endif /* GATEWAY */ /* * System control for IP6 @@ -1901,11 +1939,20 @@ SYSCTL_SETUP(sysctl_net_inet6_ip6_setup, NULL, 0, &ip6_mcast_pmtu, 0, CTL_NET, PF_INET6, IPPROTO_IPV6, CTL_CREATE, 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 (IPv6)"), + sysctl_net_inet6_ip6_maxflows, 0, &ip6_maxflows, 0, + CTL_NET, PF_INET6, IPPROTO_IPV6, + CTL_CREATE, CTL_EOL); sysctl_createv(clog, 0, NULL, NULL, CTLFLAG_PERMANENT|CTLFLAG_READWRITE, - CTLTYPE_INT, "rht0", - SYSCTL_DESCR("Processing of routing header type 0 (IPv6)"), - sysctl_net_inet6_ip6_rht0, 0, &ip6_rht0, 0, + CTLTYPE_INT, "hashsize", + SYSCTL_DESCR("Size of hash table for fast forwarding (IPv6)"), + sysctl_net_inet6_ip6_hashsize, 0, &ip6_hashsize, 0, CTL_NET, PF_INET6, IPPROTO_IPV6, CTL_CREATE, CTL_EOL); +#endif }