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/udp_usrreq.c,v rcsdiff: /ftp/cvs/cvsroot/src/sys/netinet/udp_usrreq.c,v: warning: Unknown phrases like `commitid ...;' are present. retrieving revision 1.32 retrieving revision 1.88 diff -u -p -r1.32 -r1.88 --- src/sys/netinet/udp_usrreq.c 1996/05/23 17:03:31 1.32 +++ src/sys/netinet/udp_usrreq.c 2001/11/02 02:37:50 1.88 @@ -1,7 +1,36 @@ -/* $NetBSD: udp_usrreq.c,v 1.32 1996/05/23 17:03:31 mycroft Exp $ */ +/* $NetBSD: udp_usrreq.c,v 1.88 2001/11/02 02:37:50 itojun Exp $ */ /* - * Copyright (c) 1982, 1986, 1988, 1990, 1993 + * 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: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 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. 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 + * ARE DISCLAIMED. IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE + * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL + * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS + * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) + * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT + * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY + * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF + * SUCH DAMAGE. + */ + +/* + * Copyright (c) 1982, 1986, 1988, 1990, 1993, 1995 * The Regents of the University of California. All rights reserved. * * Redistribution and use in source and binary forms, with or without @@ -32,9 +61,14 @@ * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF * SUCH DAMAGE. * - * @(#)udp_usrreq.c 8.4 (Berkeley) 1/21/94 + * @(#)udp_usrreq.c 8.6 (Berkeley) 5/23/95 */ +#include "opt_inet.h" +#include "opt_ipsec.h" +#include "opt_inet_csum.h" +#include "opt_ipkdb.h" + #include #include #include @@ -45,8 +79,7 @@ #include #include #include - -#include +#include #include #include @@ -62,8 +95,37 @@ #include #include +#ifdef INET6 +#include +#include +#include +#include +#include +#endif + +#ifdef PULLDOWN_TEST +#ifndef INET6 +/* always need ip6.h for IP6_EXTHDR_GET */ +#include +#endif +#endif + +#include "faith.h" +#if defined(NFAITH) && NFAITH > 0 +#include +#endif + #include +#ifdef IPSEC +#include +#include +#endif /*IPSEC*/ + +#ifdef IPKDB +#include +#endif + /* * UDP protocol implementation. * Per RFC 768, August, 1980. @@ -74,23 +136,66 @@ int udpcksum = 1; int udpcksum = 0; /* XXX */ #endif -struct sockaddr_in udp_in = { sizeof(udp_in), AF_INET }; - +#ifdef INET +static void udp4_sendup __P((struct mbuf *, int, struct sockaddr *, + struct socket *)); +static int udp4_realinput __P((struct sockaddr_in *, struct sockaddr_in *, + struct mbuf *, int)); +#endif +#ifdef INET6 +static void udp6_sendup __P((struct mbuf *, int, struct sockaddr *, + struct socket *)); +static int in6_mcmatch __P((struct in6pcb *, struct in6_addr *, + struct ifnet *)); +static int udp6_realinput __P((int, struct sockaddr_in6 *, + struct sockaddr_in6 *, struct mbuf *, int)); +#endif +#ifdef INET static void udp_notify __P((struct inpcb *, int)); -static struct mbuf *udp_saveopt __P((caddr_t, int, int)); +#endif #ifndef UDBHASHSIZE #define UDBHASHSIZE 128 #endif int udbhashsize = UDBHASHSIZE; +#ifdef UDP_CSUM_COUNTERS +#include + +struct evcnt udp_hwcsum_bad = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, + NULL, "udp", "hwcsum bad"); +struct evcnt udp_hwcsum_ok = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, + NULL, "udp", "hwcsum ok"); +struct evcnt udp_hwcsum_data = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, + NULL, "udp", "hwcsum data"); +struct evcnt udp_swcsum = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, + NULL, "udp", "swcsum"); + +#define UDP_CSUM_COUNTER_INCR(ev) (ev)->ev_count++ + +#else + +#define UDP_CSUM_COUNTER_INCR(ev) /* nothing */ + +#endif /* UDP_CSUM_COUNTERS */ + void udp_init() { - in_pcbinit(&udbtable, udbhashsize); +#ifdef INET + in_pcbinit(&udbtable, udbhashsize, udbhashsize); +#endif + +#ifdef UDP_CSUM_COUNTERS + evcnt_attach_static(&udp_hwcsum_bad); + evcnt_attach_static(&udp_hwcsum_ok); + evcnt_attach_static(&udp_hwcsum_data); + evcnt_attach_static(&udp_swcsum); +#endif /* UDP_CSUM_COUNTERS */ } +#ifdef INET void #if __STDC__ udp_input(struct mbuf *m, ...) @@ -100,21 +205,22 @@ udp_input(m, va_alist) va_dcl #endif { - register struct ip *ip; - register struct udphdr *uh; - register struct inpcb *inp; - struct mbuf *opts = 0; - int len; - struct ip save_ip; - int iphlen; va_list ap; + struct sockaddr_in src, dst; + struct ip *ip; + struct udphdr *uh; + int iphlen, proto; + int len; + int n; va_start(ap, m); iphlen = va_arg(ap, int); + proto = va_arg(ap, int); va_end(ap); udpstat.udps_ipackets++; +#ifndef PULLDOWN_TEST /* * Strip IP options, if any; should skip this, * make available to user, and use on returned packets, @@ -125,11 +231,18 @@ udp_input(m, va_alist) ip_stripoptions(m, (struct mbuf *)0); iphlen = sizeof(struct ip); } +#else + /* + * we may enable the above code if we save and pass IPv4 options + * to the userland. + */ +#endif /* * Get IP and UDP header together in first mbuf. */ ip = mtod(m, struct ip *); +#ifndef PULLDOWN_TEST if (m->m_len < iphlen + sizeof(struct udphdr)) { if ((m = m_pullup(m, iphlen + sizeof(struct udphdr))) == 0) { udpstat.udps_hdrops++; @@ -138,43 +251,367 @@ udp_input(m, va_alist) ip = mtod(m, struct ip *); } uh = (struct udphdr *)((caddr_t)ip + iphlen); +#else + IP6_EXTHDR_GET(uh, struct udphdr *, m, iphlen, sizeof(struct udphdr)); + if (uh == NULL) { + udpstat.udps_hdrops++; + return; + } +#endif + + /* destination port of 0 is illegal, based on RFC768. */ + if (uh->uh_dport == 0) + goto bad; /* * Make mbuf data length reflect UDP length. * If not enough data to reflect UDP length, drop. */ len = ntohs((u_int16_t)uh->uh_ulen); - if (ip->ip_len != len) { - if (len > ip->ip_len) { + if (ip->ip_len != iphlen + len) { + if (ip->ip_len < iphlen + len || len < sizeof(struct udphdr)) { udpstat.udps_badlen++; goto bad; } - m_adj(m, len - ip->ip_len); - /* ip->ip_len = len; */ + m_adj(m, iphlen + len - ip->ip_len); + } + + /* + * Checksum extended UDP header and data. + */ + if (uh->uh_sum) { + switch (m->m_pkthdr.csum_flags & + ((m->m_pkthdr.rcvif->if_csum_flags_rx & M_CSUM_UDPv4) | + M_CSUM_TCP_UDP_BAD | M_CSUM_DATA)) { + case M_CSUM_UDPv4|M_CSUM_TCP_UDP_BAD: + UDP_CSUM_COUNTER_INCR(&udp_hwcsum_bad); + goto badcsum; + + case M_CSUM_UDPv4|M_CSUM_DATA: + UDP_CSUM_COUNTER_INCR(&udp_hwcsum_data); + if ((m->m_pkthdr.csum_data ^ 0xffff) != 0) + goto badcsum; + break; + + case M_CSUM_UDPv4: + /* Checksum was okay. */ + UDP_CSUM_COUNTER_INCR(&udp_hwcsum_ok); + break; + + default: + /* Need to compute it ourselves. */ + UDP_CSUM_COUNTER_INCR(&udp_swcsum); + if (in4_cksum(m, IPPROTO_UDP, iphlen, len) != 0) + goto badcsum; + break; + } + } + + /* construct source and dst sockaddrs. */ + bzero(&src, sizeof(src)); + src.sin_family = AF_INET; + src.sin_len = sizeof(struct sockaddr_in); + bcopy(&ip->ip_src, &src.sin_addr, sizeof(src.sin_addr)); + src.sin_port = uh->uh_sport; + bzero(&dst, sizeof(dst)); + dst.sin_family = AF_INET; + dst.sin_len = sizeof(struct sockaddr_in); + bcopy(&ip->ip_dst, &dst.sin_addr, sizeof(dst.sin_addr)); + dst.sin_port = uh->uh_dport; + + n = udp4_realinput(&src, &dst, m, iphlen); +#ifdef INET6 + if (IN_MULTICAST(ip->ip_dst.s_addr) || n == 0) { + struct sockaddr_in6 src6, dst6; + + bzero(&src6, sizeof(src6)); + src6.sin6_family = AF_INET6; + src6.sin6_len = sizeof(struct sockaddr_in6); + src6.sin6_addr.s6_addr[10] = src6.sin6_addr.s6_addr[11] = 0xff; + bcopy(&ip->ip_src, &src6.sin6_addr.s6_addr[12], + sizeof(ip->ip_src)); + src6.sin6_port = uh->uh_sport; + bzero(&dst6, sizeof(dst6)); + dst6.sin6_family = AF_INET6; + dst6.sin6_len = sizeof(struct sockaddr_in6); + dst6.sin6_addr.s6_addr[10] = dst6.sin6_addr.s6_addr[11] = 0xff; + bcopy(&ip->ip_dst, &dst6.sin6_addr.s6_addr[12], + sizeof(ip->ip_dst)); + dst6.sin6_port = uh->uh_dport; + + n += udp6_realinput(AF_INET, &src6, &dst6, m, iphlen); + } +#endif + + if (n == 0) { + if (m->m_flags & (M_BCAST | M_MCAST)) { + udpstat.udps_noportbcast++; + goto bad; + } + udpstat.udps_noport++; +#ifdef IPKDB + if (checkipkdb(&ip->ip_src, uh->uh_sport, uh->uh_dport, + m, iphlen + sizeof(struct udphdr), + m->m_pkthdr.len - iphlen - sizeof(struct udphdr))) { + /* + * It was a debugger connect packet, + * just drop it now + */ + goto bad; + } +#endif + icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_PORT, 0, 0); + m = NULL; } + +bad: + if (m) + m_freem(m); + return; + +badcsum: + m_freem(m); + udpstat.udps_badsum++; +} +#endif + +#ifdef INET6 +int +udp6_input(mp, offp, proto) + struct mbuf **mp; + int *offp, proto; +{ + struct mbuf *m = *mp; + int off = *offp; + struct sockaddr_in6 src, dst; + struct ip6_hdr *ip6; + struct udphdr *uh; + u_int32_t plen, ulen; + +#ifndef PULLDOWN_TEST + IP6_EXTHDR_CHECK(m, off, sizeof(struct udphdr), IPPROTO_DONE); +#endif + ip6 = mtod(m, struct ip6_hdr *); + +#if defined(NFAITH) && 0 < NFAITH + if (faithprefix(&ip6->ip6_dst)) { + /* send icmp6 host unreach? */ + m_freem(m); + return IPPROTO_DONE; + } +#endif + + udp6stat.udp6s_ipackets++; + + /* check for jumbogram is done in ip6_input. we can trust pkthdr.len */ + plen = m->m_pkthdr.len - off; +#ifndef PULLDOWN_TEST + uh = (struct udphdr *)((caddr_t)ip6 + off); +#else + IP6_EXTHDR_GET(uh, struct udphdr *, m, off, sizeof(struct udphdr)); + if (uh == NULL) { + ip6stat.ip6s_tooshort++; + return IPPROTO_DONE; + } +#endif + ulen = ntohs((u_short)uh->uh_ulen); /* - * Save a copy of the IP header in case we want restore it - * for sending an ICMP error message in response. + * RFC2675 section 4: jumbograms will have 0 in the UDP header field, + * iff payload length > 0xffff. */ - save_ip = *ip; + if (ulen == 0 && plen > 0xffff) + ulen = plen; + + if (plen != ulen) { + udp6stat.udp6s_badlen++; + goto bad; + } + + /* destination port of 0 is illegal, based on RFC768. */ + if (uh->uh_dport == 0) + goto bad; + + /* Be proactive about malicious use of IPv4 mapped address */ + if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) || + IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) { + /* XXX stat */ + goto bad; + } /* * Checksum extended UDP header and data. */ - if (uh->uh_sum) { - bzero(((struct ipovly *)ip)->ih_x1, - sizeof ((struct ipovly *)ip)->ih_x1); - ((struct ipovly *)ip)->ih_len = uh->uh_ulen; - if ((uh->uh_sum = in_cksum(m, len + sizeof (struct ip))) != 0) { - udpstat.udps_badsum++; - m_freem(m); - return; + if (uh->uh_sum == 0) + udp6stat.udp6s_nosum++; + else if (in6_cksum(m, IPPROTO_UDP, off, ulen) != 0) { + udp6stat.udp6s_badsum++; + goto bad; + } + + /* + * Construct source and dst sockaddrs. + * Note that ifindex (s6_addr16[1]) is already filled. + */ + bzero(&src, sizeof(src)); + src.sin6_family = AF_INET6; + src.sin6_len = sizeof(struct sockaddr_in6); + /* KAME hack: recover scopeid */ + (void)in6_recoverscope(&src, &ip6->ip6_src, m->m_pkthdr.rcvif); + src.sin6_port = uh->uh_sport; + bzero(&dst, sizeof(dst)); + dst.sin6_family = AF_INET6; + dst.sin6_len = sizeof(struct sockaddr_in6); + /* KAME hack: recover scopeid */ + (void)in6_recoverscope(&dst, &ip6->ip6_dst, m->m_pkthdr.rcvif); + dst.sin6_port = uh->uh_dport; + + if (udp6_realinput(AF_INET6, &src, &dst, m, off) == 0) { + if (m->m_flags & M_MCAST) { + udp6stat.udp6s_noportmcast++; + goto bad; } + udp6stat.udp6s_noport++; + icmp6_error(m, ICMP6_DST_UNREACH, ICMP6_DST_UNREACH_NOPORT, 0); + m = NULL; + } + +bad: + if (m) + m_freem(m); + return IPPROTO_DONE; +} +#endif + +#ifdef INET +static void +udp4_sendup(m, off, src, so) + struct mbuf *m; + int off; /* offset of data portion */ + struct sockaddr *src; + struct socket *so; +{ + struct mbuf *opts = NULL; + struct mbuf *n; + struct inpcb *inp = NULL; +#ifdef INET6 + struct in6pcb *in6p = NULL; +#endif + + if (!so) + return; + switch (so->so_proto->pr_domain->dom_family) { + case AF_INET: + inp = sotoinpcb(so); + break; +#ifdef INET6 + case AF_INET6: + in6p = sotoin6pcb(so); + break; +#endif + default: + return; } - if (IN_MULTICAST(ip->ip_dst.s_addr) || - in_broadcast(ip->ip_dst, m->m_pkthdr.rcvif)) { - struct socket *last; +#ifdef IPSEC + /* check AH/ESP integrity. */ + if (so != NULL && ipsec4_in_reject_so(m, so)) { + ipsecstat.in_polvio++; + return; + } +#endif /*IPSEC*/ + + if ((n = m_copy(m, 0, M_COPYALL)) != NULL) { + if (inp && (inp->inp_flags & INP_CONTROLOPTS + || so->so_options & SO_TIMESTAMP)) { + struct ip *ip = mtod(n, struct ip *); + ip_savecontrol(inp, &opts, ip, n); + } + + m_adj(n, off); + if (sbappendaddr(&so->so_rcv, src, n, + opts) == 0) { + m_freem(n); + if (opts) + m_freem(opts); + udpstat.udps_fullsock++; + } else + sorwakeup(so); + } +} +#endif + +#ifdef INET6 +static void +udp6_sendup(m, off, src, so) + struct mbuf *m; + int off; /* offset of data portion */ + struct sockaddr *src; + struct socket *so; +{ + struct mbuf *opts = NULL; + struct mbuf *n; + struct in6pcb *in6p = NULL; + + if (!so) + return; + if (so->so_proto->pr_domain->dom_family != AF_INET6) + return; + in6p = sotoin6pcb(so); + +#ifdef IPSEC + /* check AH/ESP integrity. */ + if (so != NULL && ipsec6_in_reject_so(m, so)) { + ipsec6stat.in_polvio++; + return; + } +#endif /*IPSEC*/ + + if ((n = m_copy(m, 0, M_COPYALL)) != NULL) { + if (in6p && (in6p->in6p_flags & IN6P_CONTROLOPTS + || in6p->in6p_socket->so_options & SO_TIMESTAMP)) { + struct ip6_hdr *ip6 = mtod(n, struct ip6_hdr *); + ip6_savecontrol(in6p, &opts, ip6, n); + } + + m_adj(n, off); + if (sbappendaddr(&so->so_rcv, src, n, opts) == 0) { + m_freem(n); + if (opts) + m_freem(opts); + udp6stat.udp6s_fullsock++; + } else + sorwakeup(so); + } +} +#endif + +#ifdef INET +static int +udp4_realinput(src, dst, m, off) + struct sockaddr_in *src; + struct sockaddr_in *dst; + struct mbuf *m; + int off; /* offset of udphdr */ +{ + u_int16_t *sport, *dport; + int rcvcnt; + struct in_addr *src4, *dst4; + struct inpcb *inp; + + rcvcnt = 0; + off += sizeof(struct udphdr); /* now, offset of payload */ + + if (src->sin_family != AF_INET || dst->sin_family != AF_INET) + goto bad; + + src4 = &src->sin_addr; + sport = &src->sin_port; + dst4 = &dst->sin_addr; + dport = &dst->sin_port; + + if (IN_MULTICAST(dst4->s_addr) || + in_broadcast(*dst4, m->m_pkthdr.rcvif)) { + struct inpcb *last; /* * Deliver a multicast or broadcast datagram to *all* sockets * for which the local and remote addresses and ports match @@ -192,182 +629,218 @@ udp_input(m, va_alist) */ /* - * Construct sockaddr format source address. + * KAME note: usually we drop udpiphdr from mbuf here. + * we need udpiphdr for IPsec processing so we do that later. */ - udp_in.sin_port = uh->uh_sport; - udp_in.sin_addr = ip->ip_src; - m->m_len -= sizeof (struct udpiphdr); - m->m_data += sizeof (struct udpiphdr); /* * Locate pcb(s) for datagram. - * (Algorithm copied from raw_intr().) */ - last = NULL; for (inp = udbtable.inpt_queue.cqh_first; inp != (struct inpcb *)&udbtable.inpt_queue; inp = inp->inp_queue.cqe_next) { - if (inp->inp_lport != uh->uh_dport) + if (inp->inp_lport != *dport) continue; - if (inp->inp_laddr.s_addr != INADDR_ANY) { - if (inp->inp_laddr.s_addr != - ip->ip_dst.s_addr) + if (!in_nullhost(inp->inp_laddr)) { + if (!in_hosteq(inp->inp_laddr, *dst4)) continue; } - if (inp->inp_faddr.s_addr != INADDR_ANY) { - if (inp->inp_faddr.s_addr != - ip->ip_src.s_addr || - inp->inp_fport != uh->uh_sport) + if (!in_nullhost(inp->inp_faddr)) { + if (!in_hosteq(inp->inp_faddr, *src4) || + inp->inp_fport != *sport) continue; } - if (last != NULL) { - struct mbuf *n; + last = inp; + udp4_sendup(m, off, (struct sockaddr *)src, + inp->inp_socket); + rcvcnt++; - if ((n = m_copy(m, 0, M_COPYALL)) != NULL) { - if (sbappendaddr(&last->so_rcv, - sintosa(&udp_in), n, - (struct mbuf *)0) == 0) { - m_freem(n); - udpstat.udps_fullsock++; - } else - sorwakeup(last); - } - } - last = inp->inp_socket; /* * Don't look for additional matches if this one does * not have either the SO_REUSEPORT or SO_REUSEADDR * socket options set. This heuristic avoids searching * through all pcbs in the common case of a non-shared - * port. It * assumes that an application will never + * port. It assumes that an application will never * clear these options after setting them. */ - if ((last->so_options&(SO_REUSEPORT|SO_REUSEADDR)) == 0) + if ((inp->inp_socket->so_options & + (SO_REUSEPORT|SO_REUSEADDR)) == 0) break; } - - if (last == NULL) { - /* - * No matching pcb found; discard datagram. - * (No need to send an ICMP Port Unreachable - * for a broadcast or multicast datgram.) - */ - udpstat.udps_noportbcast++; - goto bad; - } - if (sbappendaddr(&last->so_rcv, sintosa(&udp_in), m, - (struct mbuf *)0) == 0) { - udpstat.udps_fullsock++; - goto bad; - } - sorwakeup(last); - return; - } - /* - * Locate pcb for datagram. - */ - inp = in_pcbhashlookup(&udbtable, ip->ip_src, uh->uh_sport, - ip->ip_dst, uh->uh_dport); - if (inp == 0) { - ++udpstat.udps_pcbhashmiss; - inp = in_pcblookup(&udbtable, ip->ip_src, uh->uh_sport, - ip->ip_dst, uh->uh_dport, INPLOOKUP_WILDCARD); + } else { + /* + * Locate pcb for datagram. + */ + inp = in_pcblookup_connect(&udbtable, *src4, *sport, *dst4, *dport); if (inp == 0) { - udpstat.udps_noport++; - if (m->m_flags & (M_BCAST | M_MCAST)) { - udpstat.udps_noportbcast++; - goto bad; - } - *ip = save_ip; - ip->ip_len += iphlen; - icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_PORT, 0, 0); - return; + ++udpstat.udps_pcbhashmiss; + inp = in_pcblookup_bind(&udbtable, *dst4, *dport); + if (inp == 0) + return rcvcnt; } - } - /* - * Construct sockaddr format source address. - * Stuff source address and datagram in user buffer. - */ - udp_in.sin_port = uh->uh_sport; - udp_in.sin_addr = ip->ip_src; - if (inp->inp_flags & INP_CONTROLOPTS) { - struct mbuf **mp = &opts; - - if (inp->inp_flags & INP_RECVDSTADDR) { - *mp = udp_saveopt((caddr_t) &ip->ip_dst, - sizeof(struct in_addr), IP_RECVDSTADDR); - if (*mp) - mp = &(*mp)->m_next; - } -#ifdef notyet - /* options were tossed above */ - if (inp->inp_flags & INP_RECVOPTS) { - *mp = udp_saveopt((caddr_t) opts_deleted_above, - sizeof(struct in_addr), IP_RECVOPTS); - 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 = udp_saveopt((caddr_t) ip_srcroute(), - sizeof(struct in_addr), IP_RECVRETOPTS); - if (*mp) - mp = &(*mp)->m_next; - } -#endif - } - iphlen += sizeof(struct udphdr); - m->m_len -= iphlen; - m->m_pkthdr.len -= iphlen; - m->m_data += iphlen; - if (sbappendaddr(&inp->inp_socket->so_rcv, sintosa(&udp_in), m, - opts) == 0) { - udpstat.udps_fullsock++; - goto bad; + udp4_sendup(m, off, (struct sockaddr *)src, inp->inp_socket); + rcvcnt++; } - sorwakeup(inp->inp_socket); - return; + bad: - m_freem(m); - if (opts) - m_freem(opts); + return rcvcnt; } +#endif -/* - * Create a "control" mbuf containing the specified data - * with the specified type for presentation with a datagram. - */ -struct mbuf * -udp_saveopt(p, size, type) - caddr_t p; - register int size; - int type; +#ifdef INET6 +static int +in6_mcmatch(in6p, ia6, ifp) + struct in6pcb *in6p; + struct in6_addr *ia6; + struct ifnet *ifp; { - register struct cmsghdr *cp; + struct ip6_moptions *im6o = in6p->in6p_moptions; + struct in6_multi_mship *imm; + + if (im6o == NULL) + return 0; + + for (imm = im6o->im6o_memberships.lh_first; imm != NULL; + imm = imm->i6mm_chain.le_next) { + if ((ifp == NULL || + imm->i6mm_maddr->in6m_ifp == ifp) && + IN6_ARE_ADDR_EQUAL(&imm->i6mm_maddr->in6m_addr, + ia6)) + return 1; + } + return 0; +} + +static int +udp6_realinput(af, src, dst, m, off) + int af; /* af on packet */ + struct sockaddr_in6 *src; + struct sockaddr_in6 *dst; struct mbuf *m; + int off; /* offset of udphdr */ +{ + u_int16_t sport, dport; + int rcvcnt; + struct in6_addr src6, dst6; + const struct in_addr *dst4; + struct in6pcb *in6p; + + rcvcnt = 0; + off += sizeof(struct udphdr); /* now, offset of payload */ + + if (af != AF_INET && af != AF_INET6) + goto bad; + if (src->sin6_family != AF_INET6 || dst->sin6_family != AF_INET6) + goto bad; - if ((m = m_get(M_DONTWAIT, MT_CONTROL)) == NULL) - return ((struct mbuf *) NULL); - cp = (struct cmsghdr *) mtod(m, struct cmsghdr *); - bcopy(p, CMSG_DATA(cp), size); - size += sizeof(*cp); - m->m_len = size; - cp->cmsg_len = size; - cp->cmsg_level = IPPROTO_IP; - cp->cmsg_type = type; - return (m); + in6_embedscope(&src6, src, NULL, NULL); + sport = src->sin6_port; + in6_embedscope(&dst6, dst, NULL, NULL); + dport = dst->sin6_port; + dst4 = (struct in_addr *)&dst->sin6_addr.s6_addr[12]; + + if (IN6_IS_ADDR_MULTICAST(&dst6) || + (af == AF_INET && IN_MULTICAST(dst4->s_addr))) { + struct in6pcb *last; + /* + * Deliver a multicast or broadcast datagram to *all* sockets + * for which the local and remote addresses and ports match + * those of the incoming datagram. This allows more than + * one process to receive multi/broadcasts on the same port. + * (This really ought to be done for unicast datagrams as + * well, but that would cause problems with existing + * applications that open both address-specific sockets and + * a wildcard socket listening to the same port -- they would + * end up receiving duplicates of every unicast datagram. + * Those applications open the multiple sockets to overcome an + * inadequacy of the UDP socket interface, but for backwards + * compatibility we avoid the problem here rather than + * fixing the interface. Maybe 4.5BSD will remedy this?) + */ + + /* + * KAME note: usually we drop udpiphdr from mbuf here. + * we need udpiphdr for IPsec processing so we do that later. + */ + /* + * Locate pcb(s) for datagram. + */ + for (in6p = udb6.in6p_next; in6p != &udb6; + in6p = in6p->in6p_next) { + if (in6p->in6p_lport != dport) + continue; + if (!IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_laddr)) { + if (!IN6_ARE_ADDR_EQUAL(&in6p->in6p_laddr, &dst6) && + !in6_mcmatch(in6p, &dst6, m->m_pkthdr.rcvif)) + continue; + } + else { + if (IN6_IS_ADDR_V4MAPPED(&dst6) && + (in6p->in6p_flags & IN6P_IPV6_V6ONLY)) + continue; + } + if (!IN6_IS_ADDR_UNSPECIFIED(&in6p->in6p_faddr)) { + if (!IN6_ARE_ADDR_EQUAL(&in6p->in6p_faddr, + &src6) || in6p->in6p_fport != sport) + continue; + } + else { + if (IN6_IS_ADDR_V4MAPPED(&src6) && + (in6p->in6p_flags & IN6P_IPV6_V6ONLY)) + continue; + } + + last = in6p; + udp6_sendup(m, off, (struct sockaddr *)src, + in6p->in6p_socket); + rcvcnt++; + + /* + * Don't look for additional matches if this one does + * not have either the SO_REUSEPORT or SO_REUSEADDR + * socket options set. This heuristic avoids searching + * through all pcbs in the common case of a non-shared + * port. It assumes that an application will never + * clear these options after setting them. + */ + if ((in6p->in6p_socket->so_options & + (SO_REUSEPORT|SO_REUSEADDR)) == 0) + break; + } + } else { + /* + * Locate pcb for datagram. + */ + in6p = in6_pcblookup_connect(&udb6, &src6, sport, + &dst6, dport, 0); + if (in6p == 0) { + ++udpstat.udps_pcbhashmiss; + in6p = in6_pcblookup_bind(&udb6, &dst6, dport, 0); + if (in6p == 0) + return rcvcnt; + } + + udp6_sendup(m, off, (struct sockaddr *)src, in6p->in6p_socket); + rcvcnt++; + } + +bad: + return rcvcnt; } +#endif +#ifdef INET /* * Notify a udp user of an asynchronous error; * just wake up so that he can collect error status. */ static void udp_notify(inp, errno) - register struct inpcb *inp; + struct inpcb *inp; int errno; { + inp->inp_socket->so_error = errno; sorwakeup(inp->inp_socket); sowwakeup(inp->inp_socket); @@ -379,12 +852,14 @@ udp_ctlinput(cmd, sa, v) struct sockaddr *sa; void *v; { - register struct ip *ip = v; - register struct udphdr *uh; - extern int inetctlerrmap[]; + struct ip *ip = v; + struct udphdr *uh; void (*notify) __P((struct inpcb *, int)) = udp_notify; int errno; + if (sa->sa_family != AF_INET + || sa->sa_len != sizeof(struct sockaddr_in)) + return NULL; if ((unsigned)cmd >= PRC_NCMDS) return NULL; errno = inetctlerrmap[cmd]; @@ -396,10 +871,13 @@ udp_ctlinput(cmd, sa, v) return NULL; if (ip) { uh = (struct udphdr *)((caddr_t)ip + (ip->ip_hl << 2)); - in_pcbnotify(&udbtable, sa, uh->uh_dport, ip->ip_src, - uh->uh_sport, errno, notify); + in_pcbnotify(&udbtable, satosin(sa)->sin_addr, uh->uh_dport, + ip->ip_src, uh->uh_sport, errno, notify); + + /* XXX mapped address case */ } else - in_pcbnotifyall(&udbtable, sa, errno, notify); + in_pcbnotifyall(&udbtable, satosin(sa)->sin_addr, errno, + notify); return NULL; } @@ -412,21 +890,16 @@ udp_output(m, va_alist) va_dcl #endif { - struct mbuf *control; - register struct inpcb *inp; - register struct udpiphdr *ui; - register int len = m->m_pkthdr.len; + struct inpcb *inp; + struct udpiphdr *ui; + int len = m->m_pkthdr.len; int error = 0; va_list ap; va_start(ap, m); - control = va_arg(ap, struct mbuf *); inp = va_arg(ap, struct inpcb *); va_end(ap); - if (control) - m_freem(control); /* XXX */ - /* * Calculate data length and get a mbuf * for UDP and IP headers. @@ -438,31 +911,53 @@ udp_output(m, va_alist) } /* + * Compute the packet length of the IP header, and + * punt if the length looks bogus. + */ + if ((len + sizeof(struct udpiphdr)) > IP_MAXPACKET) { + error = EMSGSIZE; + goto release; + } + + /* * Fill in mbuf with extended UDP header * and addresses and length put into network format. */ ui = mtod(m, struct udpiphdr *); - bzero(ui->ui_x1, sizeof ui->ui_x1); ui->ui_pr = IPPROTO_UDP; - ui->ui_len = htons((u_int16_t)len + sizeof (struct udphdr)); ui->ui_src = inp->inp_laddr; ui->ui_dst = inp->inp_faddr; ui->ui_sport = inp->inp_lport; ui->ui_dport = inp->inp_fport; - ui->ui_ulen = ui->ui_len; + ui->ui_ulen = htons((u_int16_t)len + sizeof(struct udphdr)); /* - * Stuff checksum and output datagram. + * Set up checksum and output datagram. */ - ui->ui_sum = 0; if (udpcksum) { - if ((ui->ui_sum = in_cksum(m, sizeof (struct udpiphdr) + len)) == 0) - ui->ui_sum = 0xffff; - } + /* + * XXX Cache pseudo-header checksum part for + * XXX "connected" UDP sockets. + */ + ui->ui_sum = in_cksum_phdr(ui->ui_src.s_addr, + ui->ui_dst.s_addr, htons((u_int16_t)len + + sizeof(struct udphdr) + IPPROTO_UDP)); + m->m_pkthdr.csum_flags = M_CSUM_UDPv4; + m->m_pkthdr.csum_data = offsetof(struct udphdr, uh_sum); + } else + ui->ui_sum = 0; ((struct ip *)ui)->ip_len = sizeof (struct udpiphdr) + len; ((struct ip *)ui)->ip_ttl = inp->inp_ip.ip_ttl; /* XXX */ ((struct ip *)ui)->ip_tos = inp->inp_ip.ip_tos; /* XXX */ udpstat.udps_opackets++; + +#ifdef IPSEC + if (ipsec_setsocket(m, inp->inp_socket) != 0) { + error = ENOBUFS; + goto release; + } +#endif /*IPSEC*/ + return (ip_output(m, inp->inp_options, &inp->inp_route, inp->inp_socket->so_options & (SO_DONTROUTE | SO_BROADCAST), inp->inp_moptions)); @@ -472,8 +967,8 @@ release: return (error); } -u_long udp_sendspace = 9216; /* really max datagram size */ -u_long udp_recvspace = 40 * (1024 + sizeof(struct sockaddr_in)); +int udp_sendspace = 9216; /* really max datagram size */ +int udp_recvspace = 40 * (1024 + sizeof(struct sockaddr_in)); /* 40 1K datagrams */ /*ARGSUSED*/ @@ -484,14 +979,21 @@ udp_usrreq(so, req, m, nam, control, p) struct mbuf *m, *nam, *control; struct proc *p; { - register struct inpcb *inp; + struct inpcb *inp; int s; - register int error = 0; + int error = 0; if (req == PRU_CONTROL) return (in_control(so, (long)m, (caddr_t)nam, (struct ifnet *)control, p)); + if (req == PRU_PURGEIF) { + in_pcbpurgeif0(&udbtable, (struct ifnet *)control); + in_purgeif((struct ifnet *)control); + in_pcbpurgeif(&udbtable, (struct ifnet *)control); + return (0); + } + s = splsoftnet(); inp = sotoinpcb(so); #ifdef DIAGNOSTIC @@ -553,7 +1055,8 @@ udp_usrreq(so, req, m, nam, control, p) /*soisdisconnected(so);*/ so->so_state &= ~SS_ISCONNECTED; /* XXX */ in_pcbdisconnect(inp); - inp->inp_laddr.s_addr = INADDR_ANY; /* XXX */ + inp->inp_laddr = zeroin_addr; /* XXX */ + in_pcbstate(inp, INP_BOUND); /* XXX */ break; case PRU_SHUTDOWN: @@ -572,10 +1075,10 @@ udp_usrreq(so, req, m, nam, control, p) break; } { - struct in_addr laddr; + struct in_addr laddr; /* XXX */ if (nam) { - laddr = inp->inp_laddr; + laddr = inp->inp_laddr; /* XXX */ if ((so->so_state & SS_ISCONNECTED) != 0) { error = EISCONN; goto die; @@ -592,10 +1095,11 @@ udp_usrreq(so, req, m, nam, control, p) goto die; } } - error = udp_output(m, control, inp); + error = udp_output(m, inp); if (nam) { in_pcbdisconnect(inp); - inp->inp_laddr = laddr; + inp->inp_laddr = laddr; /* XXX */ + in_pcbstate(inp, INP_BOUND); /* XXX */ } } break; @@ -653,8 +1157,15 @@ udp_sysctl(name, namelen, oldp, oldlenp, switch (name[0]) { case UDPCTL_CHECKSUM: return (sysctl_int(oldp, oldlenp, newp, newlen, &udpcksum)); + case UDPCTL_SENDSPACE: + return (sysctl_int(oldp, oldlenp, newp, newlen, + &udp_sendspace)); + case UDPCTL_RECVSPACE: + return (sysctl_int(oldp, oldlenp, newp, newlen, + &udp_recvspace)); default: return (ENOPROTOOPT); } /* NOTREACHED */ } +#endif