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 1.223.6.2 retrieving revision 1.254 diff -u -p -r1.223.6.2 -r1.254 --- src/sys/netinet/ip_input.c 2006/04/22 11:40:10 1.223.6.2 +++ src/sys/netinet/ip_input.c 2007/10/02 20:35:04 1.254 @@ -1,4 +1,4 @@ -/* $NetBSD: ip_input.c,v 1.223.6.2 2006/04/22 11:40:10 simonb Exp $ */ +/* $NetBSD: ip_input.c,v 1.254 2007/10/02 20:35:04 dyoung Exp $ */ /* * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project. @@ -98,7 +98,7 @@ */ #include -__KERNEL_RCSID(0, "$NetBSD: ip_input.c,v 1.223.6.2 2006/04/22 11:40:10 simonb Exp $"); +__KERNEL_RCSID(0, "$NetBSD: ip_input.c,v 1.254 2007/10/02 20:35:04 dyoung Exp $"); #include "opt_inet.h" #include "opt_gateway.h" @@ -121,6 +121,7 @@ __KERNEL_RCSID(0, "$NetBSD: ip_input.c,v #include #include #include +#include #include #include @@ -333,8 +334,10 @@ do { \ #define IPQ_UNLOCK() ipq_unlock() -POOL_INIT(inmulti_pool, sizeof(struct in_multi), 0, 0, 0, "inmltpl", NULL); -POOL_INIT(ipqent_pool, sizeof(struct ipqent), 0, 0, 0, "ipqepl", NULL); +POOL_INIT(inmulti_pool, sizeof(struct in_multi), 0, 0, 0, "inmltpl", NULL, + IPL_SOFTNET); +POOL_INIT(ipqent_pool, sizeof(struct ipqent), 0, 0, 0, "ipqepl", NULL, + IPL_VM); #ifdef INET_CSUM_COUNTERS #include @@ -376,8 +379,8 @@ static struct ip_srcrt { 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 /* @@ -426,7 +429,7 @@ ip_init(void) M_WAITOK, &in_multihash); ip_mtudisc_timeout_q = rt_timer_queue_create(ip_mtudisc_timeout); #ifdef GATEWAY - ipflow_init(); + ipflow_init(ip_hashsize); #endif #ifdef PFIL_HOOKS @@ -445,7 +448,10 @@ ip_init(void) #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; /* @@ -457,7 +463,7 @@ ipintr(void) int s; struct mbuf *m; - while (1) { + while (!IF_IS_EMPTY(&ipintrq)) { s = splnet(); IF_DEQUEUE(&ipintrq, m); splx(s); @@ -484,12 +490,13 @@ ip_input(struct mbuf *m) int downmatch; int checkif; int srcrt = 0; + int s; u_int hash; #ifdef FAST_IPSEC struct m_tag *mtag; struct tdb_ident *tdbi; struct secpolicy *sp; - int s, error; + int error; #endif /* FAST_IPSEC */ MCLAIM(m, &ip_rx_mowner); @@ -511,7 +518,7 @@ ip_input(struct mbuf *m) * it. Otherwise, if it is aligned, make sure the entire * base IP header is in the first mbuf of the chain. */ - if (IP_HDR_ALIGNED_P(mtod(m, caddr_t)) == 0) { + if (IP_HDR_ALIGNED_P(mtod(m, void *)) == 0) { if ((m = m_copyup(m, sizeof(struct ip), (max_linkhdr + 3) & ~3)) == NULL) { /* XXXJRT new stat, please */ @@ -897,9 +904,7 @@ ours: */ IPQ_LOCK(); hash = IPREASS_HASH(ip->ip_src.s_addr, ip->ip_id); - /* XXX LIST_FOREACH(fp, &ipq[hash], ipq_q) */ - for (fp = LIST_FIRST(&ipq[hash]); fp != NULL; - fp = LIST_NEXT(fp, ipq_q)) { + LIST_FOREACH(fp, &ipq[hash], ipq_q) { if (ip->ip_id == fp->ipq_id && in_hosteq(ip->ip_src, fp->ipq_src) && in_hosteq(ip->ip_dst, fp->ipq_dst) && @@ -938,7 +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 @@ -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) @@ -1051,7 +1057,7 @@ ip_reass(struct ipqent *ipqe, struct ipq struct ip *ip; struct mbuf *t; int hlen = ipqe->ipqe_ip->ip_hl << 2; - int i, next; + int i, next, s; IPQ_LOCK_CHECK(); @@ -1156,7 +1162,9 @@ ip_reass(struct ipqent *ipqe, struct ipq 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--; } @@ -1197,11 +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; @@ -1236,7 +1248,9 @@ dropfrag: ip_nfrags--; ipstat.ips_fragdropped++; m_freem(m); + s = splvm(); pool_put(&ipqent_pool, ipqe); + splx(s); return (0); } @@ -1249,6 +1263,7 @@ ip_freef(struct ipq *fp) { struct ipqent *q, *p; u_int nfrags = 0; + int s; IPQ_LOCK_CHECK(); @@ -1257,7 +1272,9 @@ ip_freef(struct ipq *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) @@ -1389,9 +1406,6 @@ ip_slowtimo(void) dropscanidx = i; } IPQ_UNLOCK(); -#ifdef GATEWAY - ipflow_slowtimo(); -#endif splx(s); } @@ -1510,7 +1524,7 @@ ip_dooptions(struct mbuf *m) /* * locate outgoing interface */ - bcopy((caddr_t)(cp + off), (caddr_t)&ipaddr.sin_addr, + bcopy((void *)(cp + off), (void *)&ipaddr.sin_addr, sizeof(ipaddr.sin_addr)); if (opt == IPOPT_SSRR) ia = ifatoia(ifa_ifwithladdr(sintosa(&ipaddr))); @@ -1522,8 +1536,8 @@ ip_dooptions(struct mbuf *m) goto bad; } ip->ip_dst = ipaddr.sin_addr; - bcopy((caddr_t)&ia->ia_addr.sin_addr, - (caddr_t)(cp + off), sizeof(struct in_addr)); + bcopy((void *)&ia->ia_addr.sin_addr, + (void *)(cp + off), sizeof(struct in_addr)); cp[IPOPT_OFFSET] += sizeof(struct in_addr); /* * Let ip_intr's mcast routing check handle mcast pkts @@ -1546,7 +1560,7 @@ ip_dooptions(struct mbuf *m) off--; /* 0 origin */ if ((off + sizeof(struct in_addr)) > optlen) break; - bcopy((caddr_t)(&ip->ip_dst), (caddr_t)&ipaddr.sin_addr, + bcopy((void *)(&ip->ip_dst), (void *)&ipaddr.sin_addr, sizeof(ipaddr.sin_addr)); /* * locate outgoing interface; if we're the destination, @@ -1559,8 +1573,8 @@ ip_dooptions(struct mbuf *m) code = ICMP_UNREACH_HOST; goto bad; } - bcopy((caddr_t)&ia->ia_addr.sin_addr, - (caddr_t)(cp + off), sizeof(struct in_addr)); + bcopy((void *)&ia->ia_addr.sin_addr, + (void *)(cp + off), sizeof(struct in_addr)); cp[IPOPT_OFFSET] += sizeof(struct in_addr); break; @@ -1629,7 +1643,7 @@ ip_dooptions(struct mbuf *m) } ntime = iptime(); cp0 = (u_char *) &ntime; /* XXX grumble, GCC... */ - bcopy(cp0, (caddr_t)cp + ipt->ipt_ptr - 1, + memmove((char *)cp + ipt->ipt_ptr - 1, cp0, sizeof(n_time)); ipt->ipt_ptr += sizeof(n_time); } @@ -1657,24 +1671,18 @@ bad: struct in_ifaddr * ip_rtaddr(struct in_addr dst) { - struct sockaddr_in *sin; + struct rtentry *rt; + union { + struct sockaddr dst; + struct sockaddr_in dst4; + } u; - sin = satosin(&ipforward_rt.ro_dst); + sockaddr_in_init(&u.dst4, &dst, 0); - if (ipforward_rt.ro_rt == 0 || !in_hosteq(dst, sin->sin_addr)) { - if (ipforward_rt.ro_rt) { - RTFREE(ipforward_rt.ro_rt); - ipforward_rt.ro_rt = 0; - } - sin->sin_family = AF_INET; - sin->sin_len = sizeof(*sin); - sin->sin_addr = dst; + if ((rt = rtcache_lookup(&ipforward_rt, &u.dst)) == NULL) + return NULL; - rtalloc(&ipforward_rt); - } - if (ipforward_rt.ro_rt == 0) - return ((struct in_ifaddr *)0); - return (ifatoia(ipforward_rt.ro_rt->rt_ifa)); + return ifatoia(rt->rt_ifa); } /* @@ -1693,7 +1701,7 @@ save_rte(u_char *option, struct in_addr #endif /* 0 */ if (olen > sizeof(ip_srcrt) - (1 + sizeof(dst))) return; - bcopy((caddr_t)option, (caddr_t)ip_srcrt.srcopt, olen); + bcopy((void *)option, (void *)ip_srcrt.srcopt, olen); ip_nhops = (olen - IPOPT_OFFSET - 1) / sizeof(struct in_addr); ip_srcrt.dst = dst; } @@ -1710,10 +1718,10 @@ ip_srcroute(void) struct mbuf *m; if (ip_nhops == 0) - return ((struct mbuf *)0); + return NULL; m = m_get(M_DONTWAIT, MT_SOOPTS); if (m == 0) - return ((struct mbuf *)0); + return NULL; MCLAIM(m, &inetdomain.dom_mowner); #define OPTSIZ (sizeof(ip_srcrt.nop) + sizeof(ip_srcrt.srcopt)) @@ -1741,9 +1749,9 @@ ip_srcroute(void) */ ip_srcrt.nop = IPOPT_NOP; ip_srcrt.srcopt[IPOPT_OFFSET] = IPOPT_MINOFF; - bcopy((caddr_t)&ip_srcrt.nop, - mtod(m, caddr_t) + sizeof(struct in_addr), OPTSIZ); - q = (struct in_addr *)(mtod(m, caddr_t) + + memmove(mtod(m, char *) + sizeof(struct in_addr), &ip_srcrt.nop, + OPTSIZ); + q = (struct in_addr *)(mtod(m, char *) + sizeof(struct in_addr) + OPTSIZ); #undef OPTSIZ /* @@ -1768,32 +1776,6 @@ ip_srcroute(void) return (m); } -/* - * Strip out IP options, at higher - * level protocol in the kernel. - * Second argument is buffer to which options - * will be moved, and return value is their length. - * XXX should be deleted; last arg currently ignored. - */ -void -ip_stripoptions(struct mbuf *m, struct mbuf *mopt) -{ - int i; - struct ip *ip = mtod(m, struct ip *); - caddr_t opts; - int olen; - - olen = (ip->ip_hl << 2) - sizeof (struct ip); - opts = (caddr_t)(ip + 1); - i = m->m_len - (sizeof (struct ip) + olen); - bcopy(opts + olen, opts, (unsigned)i); - m->m_len -= olen; - if (m->m_flags & M_PKTHDR) - m->m_pkthdr.len -= olen; - ip->ip_len = htons(ntohs(ip->ip_len) - olen); - ip->ip_hl = sizeof (struct ip) >> 2; -} - const int inetctlerrmap[PRC_NCMDS] = { 0, 0, 0, 0, 0, EMSGSIZE, EHOSTDOWN, EHOSTUNREACH, @@ -1821,11 +1803,14 @@ void 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, destmtu = 0; struct mbuf *mcopy; n_long dest; + union { + struct sockaddr dst; + struct sockaddr_in dst4; + } u; /* * We are now in the output path. @@ -1854,23 +1839,10 @@ ip_forward(struct mbuf *m, int srcrt) return; } - sin = satosin(&ipforward_rt.ro_dst); - if ((rt = ipforward_rt.ro_rt) == 0 || - !in_hosteq(ip->ip_dst, sin->sin_addr)) { - if (ipforward_rt.ro_rt) { - RTFREE(ipforward_rt.ro_rt); - ipforward_rt.ro_rt = 0; - } - sin->sin_family = AF_INET; - sin->sin_len = sizeof(struct sockaddr_in); - sin->sin_addr = ip->ip_dst; - - rtalloc(&ipforward_rt); - if (ipforward_rt.ro_rt == 0) { - icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_NET, dest, 0); - return; - } - rt = ipforward_rt.ro_rt; + sockaddr_in_init(&u.dst4, &ip->ip_dst, 0); + if ((rt = rtcache_lookup(&ipforward_rt, &u.dst)) == NULL) { + icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_NET, dest, 0); + return; } /* @@ -1894,7 +1866,7 @@ ip_forward(struct mbuf *m, int srcrt) */ if (rt->rt_ifp == m->m_pkthdr.rcvif && (rt->rt_flags & (RTF_DYNAMIC|RTF_MODIFIED)) == 0 && - !in_nullhost(satosin(rt_key(rt))->sin_addr) && + !in_nullhost(satocsin(rt_getkey(rt))->sin_addr) && ipsendredirects && !srcrt) { if (rt->rt_ifa && (ip->ip_src.s_addr & ifatoia(rt->rt_ifa)->ia_subnetmask) == @@ -1917,7 +1889,7 @@ ip_forward(struct mbuf *m, int srcrt) } } - error = ip_output(m, (struct mbuf *)0, &ipforward_rt, + error = ip_output(m, NULL, &ipforward_rt, (IP_FORWARDING | (ip_directedbcast ? IP_ALLOWBROADCAST : 0)), (struct ip_moptions *)NULL, (struct socket *)NULL); @@ -1960,7 +1932,7 @@ ip_forward(struct mbuf *m, int srcrt) type = ICMP_UNREACH; code = ICMP_UNREACH_NEEDFRAG; #if !defined(IPSEC) && !defined(FAST_IPSEC) - if (ipforward_rt.ro_rt) + if (ipforward_rt.ro_rt != NULL) destmtu = ipforward_rt.ro_rt->rt_ifp->if_mtu; #else /* @@ -1969,7 +1941,7 @@ ip_forward(struct mbuf *m, int srcrt) * tunnel MTU = if MTU - sizeof(IP) - ESP/AH hdrsiz * XXX quickhack!!! */ - if (ipforward_rt.ro_rt) { + if (ipforward_rt.ro_rt != NULL) { struct secpolicy *sp; int ipsecerror; size_t ipsechdr; @@ -2044,13 +2016,13 @@ ip_savecontrol(struct inpcb *inp, struct struct timeval tv; microtime(&tv); - *mp = sbcreatecontrol((caddr_t) &tv, sizeof(tv), + *mp = sbcreatecontrol((void *) &tv, sizeof(tv), SCM_TIMESTAMP, SOL_SOCKET); if (*mp) mp = &(*mp)->m_next; } if (inp->inp_flags & INP_RECVDSTADDR) { - *mp = sbcreatecontrol((caddr_t) &ip->ip_dst, + *mp = sbcreatecontrol((void *) &ip->ip_dst, sizeof(struct in_addr), IP_RECVDSTADDR, IPPROTO_IP); if (*mp) mp = &(*mp)->m_next; @@ -2064,14 +2036,14 @@ ip_savecontrol(struct inpcb *inp, struct */ /* options were tossed already */ if (inp->inp_flags & INP_RECVOPTS) { - *mp = sbcreatecontrol((caddr_t) opts_deleted_above, + *mp = sbcreatecontrol((void *) opts_deleted_above, sizeof(struct in_addr), IP_RECVOPTS, IPPROTO_IP); if (*mp) mp = &(*mp)->m_next; } /* ip_srcroute doesn't do what we want here, need to fix */ if (inp->inp_flags & INP_RECVRETOPTS) { - *mp = sbcreatecontrol((caddr_t) ip_srcroute(), + *mp = sbcreatecontrol((void *) ip_srcroute(), sizeof(struct in_addr), IP_RECVRETOPTS, IPPROTO_IP); if (*mp) mp = &(*mp)->m_next; @@ -2080,19 +2052,43 @@ ip_savecontrol(struct inpcb *inp, struct if (inp->inp_flags & INP_RECVIF) { struct sockaddr_dl sdl; - sdl.sdl_len = offsetof(struct sockaddr_dl, sdl_data[0]); - sdl.sdl_family = AF_LINK; - sdl.sdl_index = m->m_pkthdr.rcvif ? - m->m_pkthdr.rcvif->if_index : 0; - sdl.sdl_nlen = sdl.sdl_alen = sdl.sdl_slen = 0; - *mp = sbcreatecontrol((caddr_t) &sdl, sdl.sdl_len, - IP_RECVIF, IPPROTO_IP); + sockaddr_dl_init(&sdl, sizeof(sdl), + (m->m_pkthdr.rcvif != NULL) + ? m->m_pkthdr.rcvif->if_index + : 0, + 0, NULL, 0, NULL, 0); + *mp = sbcreatecontrol(&sdl, sdl.sdl_len, IP_RECVIF, IPPROTO_IP); if (*mp) mp = &(*mp)->m_next; } } /* + * sysctl helper routine for net.inet.ip.forwsrcrt. + */ +static int +sysctl_net_inet_ip_forwsrcrt(SYSCTLFN_ARGS) +{ + int error, tmp; + struct sysctlnode node; + + node = *rnode; + tmp = ip_forwsrcrt; + node.sysctl_data = &tmp; + error = sysctl_lookup(SYSCTLFN_CALL(&node)); + if (error || newp == NULL) + return (error); + + if (kauth_authorize_network(l->l_cred, KAUTH_NETWORK_FORWSRCRT, + 0, NULL, NULL, NULL)) + return (EPERM); + + ip_forwsrcrt = tmp; + + return (0); +} + +/* * sysctl helper routine for net.inet.ip.mtudisctimeout. checks the * range of the new value and tweaks timers if it changes. */ @@ -2119,8 +2115,7 @@ sysctl_net_inet_ip_pmtudto(SYSCTLFN_ARGS #ifdef GATEWAY /* - * sysctl helper routine for net.inet.ip.maxflows. apparently if - * maxflows is even looked up, we "reap flows". + * sysctl helper routine for net.inet.ip.maxflows. */ static int sysctl_net_inet_ip_maxflows(SYSCTLFN_ARGS) @@ -2128,7 +2123,7 @@ sysctl_net_inet_ip_maxflows(SYSCTLFN_ARG int s; s = sysctl_lookup(SYSCTLFN_CALL(rnode)); - if (s) + if (s || newp == NULL) return (s); s = splsoftnet(); @@ -2137,6 +2132,35 @@ sysctl_net_inet_ip_maxflows(SYSCTLFN_ARG return (0); } + +static int +sysctl_net_inet_ip_hashsize(SYSCTLFN_ARGS) +{ + int error, tmp; + struct sysctlnode node; + + node = *rnode; + tmp = ip_hashsize; + node.sysctl_data = &tmp; + error = sysctl_lookup(SYSCTLFN_CALL(&node)); + if (error || newp == NULL) + return (error); + + if ((tmp & (tmp - 1)) == 0 && tmp != 0) { + /* + * Can only fail due to malloc() + */ + if (ipflow_invalidate_all(tmp)) + return ENOMEM; + } else { + /* + * EINVAL if not a power of 2 + */ + return EINVAL; + } + + return (0); +} #endif /* GATEWAY */ @@ -2193,11 +2217,11 @@ 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, @@ -2258,6 +2282,13 @@ SYSCTL_SETUP(sysctl_net_inet_ip_setup, " sysctl_net_inet_ip_maxflows, 0, &ip_maxflows, 0, CTL_NET, PF_INET, IPPROTO_IP, IPCTL_MAXFLOWS, CTL_EOL); + sysctl_createv(clog, 0, NULL, NULL, + CTLFLAG_PERMANENT|CTLFLAG_READWRITE, + CTLTYPE_INT, "hashsize", + SYSCTL_DESCR("Size of hash table for fast forwarding (IPv4)"), + sysctl_net_inet_ip_hashsize, 0, &ip_hashsize, 0, + CTL_NET, PF_INET, IPPROTO_IP, + CTL_CREATE, CTL_EOL); #endif /* GATEWAY */ sysctl_createv(clog, 0, NULL, NULL, CTLFLAG_PERMANENT|CTLFLAG_READWRITE,