1c398230bSWarner Losh /*- 2df8bae1dSRodney W. Grimes * Copyright (c) 1982, 1986, 1988, 1990, 1993 3df8bae1dSRodney W. Grimes * The Regents of the University of California. All rights reserved. 4df8bae1dSRodney W. Grimes * 5df8bae1dSRodney W. Grimes * Redistribution and use in source and binary forms, with or without 6df8bae1dSRodney W. Grimes * modification, are permitted provided that the following conditions 7df8bae1dSRodney W. Grimes * are met: 8df8bae1dSRodney W. Grimes * 1. Redistributions of source code must retain the above copyright 9df8bae1dSRodney W. Grimes * notice, this list of conditions and the following disclaimer. 10df8bae1dSRodney W. Grimes * 2. Redistributions in binary form must reproduce the above copyright 11df8bae1dSRodney W. Grimes * notice, this list of conditions and the following disclaimer in the 12df8bae1dSRodney W. Grimes * documentation and/or other materials provided with the distribution. 13df8bae1dSRodney W. Grimes * 4. Neither the name of the University nor the names of its contributors 14df8bae1dSRodney W. Grimes * may be used to endorse or promote products derived from this software 15df8bae1dSRodney W. Grimes * without specific prior written permission. 16df8bae1dSRodney W. Grimes * 17df8bae1dSRodney W. Grimes * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 18df8bae1dSRodney W. Grimes * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 19df8bae1dSRodney W. Grimes * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 20df8bae1dSRodney W. Grimes * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 21df8bae1dSRodney W. Grimes * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 22df8bae1dSRodney W. Grimes * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 23df8bae1dSRodney W. Grimes * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 24df8bae1dSRodney W. Grimes * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 25df8bae1dSRodney W. Grimes * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 26df8bae1dSRodney W. Grimes * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 27df8bae1dSRodney W. Grimes * SUCH DAMAGE. 28df8bae1dSRodney W. Grimes * 29df8bae1dSRodney W. Grimes * @(#)ip_output.c 8.3 (Berkeley) 1/21/94 30df8bae1dSRodney W. Grimes */ 31df8bae1dSRodney W. Grimes 324b421e2dSMike Silbersack #include <sys/cdefs.h> 334b421e2dSMike Silbersack __FBSDID("$FreeBSD$"); 344b421e2dSMike Silbersack 351f7e052cSJulian Elischer #include "opt_ipfw.h" 366a800098SYoshinobu Inoue #include "opt_ipsec.h" 3757f60867SMark Johnston #include "opt_kdtrace.h" 3853dcc544SMike Silbersack #include "opt_mbuf_stress_test.h" 39e440aed9SQing Li #include "opt_mpath.h" 4057f60867SMark Johnston #include "opt_route.h" 412f4afd21SRandall Stewart #include "opt_sctp.h" 42fbd1372aSJoerg Wunsch 43df8bae1dSRodney W. Grimes #include <sys/param.h> 4426f9a767SRodney W. Grimes #include <sys/systm.h> 45f0f6d643SLuigi Rizzo #include <sys/kernel.h> 46df8bae1dSRodney W. Grimes #include <sys/malloc.h> 47df8bae1dSRodney W. Grimes #include <sys/mbuf.h> 48acd3428bSRobert Watson #include <sys/priv.h> 49c26fe973SBjoern A. Zeeb #include <sys/proc.h> 50df8bae1dSRodney W. Grimes #include <sys/protosw.h> 5157f60867SMark Johnston #include <sys/sdt.h> 52df8bae1dSRodney W. Grimes #include <sys/socket.h> 53df8bae1dSRodney W. Grimes #include <sys/socketvar.h> 549d9edc56SMike Silbersack #include <sys/sysctl.h> 55c26fe973SBjoern A. Zeeb #include <sys/ucred.h> 56df8bae1dSRodney W. Grimes 57df8bae1dSRodney W. Grimes #include <net/if.h> 58*76039bc8SGleb Smirnoff #include <net/if_var.h> 593ef5e21dSQing Li #include <net/if_llatbl.h> 609b932e9eSAndre Oppermann #include <net/netisr.h> 61c21fd232SAndre Oppermann #include <net/pfil.h> 62df8bae1dSRodney W. Grimes #include <net/route.h> 6365111ec7SKip Macy #include <net/flowtable.h> 64e440aed9SQing Li #ifdef RADIX_MPATH 65e440aed9SQing Li #include <net/radix_mpath.h> 66e440aed9SQing Li #endif 674b79449eSBjoern A. Zeeb #include <net/vnet.h> 68df8bae1dSRodney W. Grimes 69df8bae1dSRodney W. Grimes #include <netinet/in.h> 7057f60867SMark Johnston #include <netinet/in_kdtrace.h> 71df8bae1dSRodney W. Grimes #include <netinet/in_systm.h> 72df8bae1dSRodney W. Grimes #include <netinet/ip.h> 73df8bae1dSRodney W. Grimes #include <netinet/in_pcb.h> 74df8bae1dSRodney W. Grimes #include <netinet/in_var.h> 75df8bae1dSRodney W. Grimes #include <netinet/ip_var.h> 76ef39adf0SAndre Oppermann #include <netinet/ip_options.h> 772f4afd21SRandall Stewart #ifdef SCTP 782f4afd21SRandall Stewart #include <netinet/sctp.h> 792f4afd21SRandall Stewart #include <netinet/sctp_crc32.h> 802f4afd21SRandall Stewart #endif 81df8bae1dSRodney W. Grimes 82b2630c29SGeorge V. Neville-Neil #ifdef IPSEC 832cb64cb2SGeorge V. Neville-Neil #include <netinet/ip_ipsec.h> 841dfcf0d2SAndre Oppermann #include <netipsec/ipsec.h> 85b2630c29SGeorge V. Neville-Neil #endif /* IPSEC*/ 866a800098SYoshinobu Inoue 871dfcf0d2SAndre Oppermann #include <machine/in_cksum.h> 881dfcf0d2SAndre Oppermann 89aed55708SRobert Watson #include <security/mac/mac_framework.h> 90aed55708SRobert Watson 91eddfbb76SRobert Watson VNET_DEFINE(u_short, ip_id); 92f23b4c91SGarrett Wollman 9353dcc544SMike Silbersack #ifdef MBUF_STRESS_TEST 9405b9d121SBjoern A. Zeeb static int mbuf_frag_size = 0; 959d9edc56SMike Silbersack SYSCTL_INT(_net_inet_ip, OID_AUTO, mbuf_frag_size, CTLFLAG_RW, 969d9edc56SMike Silbersack &mbuf_frag_size, 0, "Fragment outgoing mbufs to this size"); 979d9edc56SMike Silbersack #endif 989d9edc56SMike Silbersack 99df8bae1dSRodney W. Grimes static void ip_mloopback 1004d77a549SAlfred Perlstein (struct ifnet *, struct mbuf *, struct sockaddr_in *, int); 101afed1b49SDarren Reed 102afed1b49SDarren Reed 1038b889dbbSBruce M Simpson extern int in_mcast_loop; 10493e0e116SJulian Elischer extern struct protosw inetsw[]; 10593e0e116SJulian Elischer 106df8bae1dSRodney W. Grimes /* 107df8bae1dSRodney W. Grimes * IP output. The packet in mbuf chain m contains a skeletal IP 108df8bae1dSRodney W. Grimes * header (with len, off, ttl, proto, tos, src, dst). 109df8bae1dSRodney W. Grimes * The mbuf chain containing the packet will be freed. 110df8bae1dSRodney W. Grimes * The mbuf opt, if present, will not be freed. 111bf984051SGleb Smirnoff * If route ro is present and has ro_rt initialized, route lookup would be 112bf984051SGleb Smirnoff * skipped and ro->ro_rt would be used. If ro is present but ro->ro_rt is NULL, 113bf984051SGleb Smirnoff * then result of route lookup is stored in ro->ro_rt. 114bf984051SGleb Smirnoff * 1153de758d3SRobert Watson * In the IP forwarding case, the packet will arrive with options already 1163de758d3SRobert Watson * inserted, so must have a NULL opt pointer. 117df8bae1dSRodney W. Grimes */ 118df8bae1dSRodney W. Grimes int 119f2565d68SRobert Watson ip_output(struct mbuf *m, struct mbuf *opt, struct route *ro, int flags, 120f2565d68SRobert Watson struct ip_moptions *imo, struct inpcb *inp) 121df8bae1dSRodney W. Grimes { 1221e78ac21SJeffrey Hsu struct ip *ip; 123fc74d005SBjoern A. Zeeb struct ifnet *ifp = NULL; /* keep compiler happy */ 124ac9d7e26SMax Laier struct mbuf *m0; 12523bf9953SPoul-Henning Kamp int hlen = sizeof (struct ip); 1263ae2ad08SJohn-Mark Gurney int mtu; 127ca8b83b0SLuigi Rizzo int n; /* scratchpad */ 128ca8b83b0SLuigi Rizzo int error = 0; 129ca8b83b0SLuigi Rizzo struct sockaddr_in *dst; 1304c7a6059SGleb Smirnoff const struct sockaddr_in *gw; 1313c73180fSGleb Smirnoff struct in_ifaddr *ia; 13221d172a3SGleb Smirnoff int isbroadcast; 133078468edSGleb Smirnoff uint16_t ip_len, ip_off; 134e3f406b3SRuslan Ermilov struct route iproute; 135ec396e61SLuigi Rizzo struct rtentry *rte; /* cache for ro->ro_rt */ 1369b932e9eSAndre Oppermann struct in_addr odst; 1379b932e9eSAndre Oppermann struct m_tag *fwd_tag = NULL; 1381a499816SEdward Tomasz Napierala #ifdef IPSEC 1391a499816SEdward Tomasz Napierala int no_route_but_check_spd = 0; 1401a499816SEdward Tomasz Napierala #endif 141ac9d7e26SMax Laier M_ASSERTPKTHDR(m); 14236e8826fSMax Laier 143bc97ba51SJulian Elischer if (inp != NULL) { 144baa45840SRobert Watson INP_LOCK_ASSERT(inp); 14562e1ba83SRobert Watson M_SETFIB(m, inp->inp_inc.inc_fibnum); 14680cb9f21SKip Macy if (inp->inp_flags & (INP_HW_FLOWID|INP_SW_FLOWID)) { 14780cb9f21SKip Macy m->m_pkthdr.flowid = inp->inp_flowid; 14880cb9f21SKip Macy m->m_flags |= M_FLOWID; 14980cb9f21SKip Macy } 150bc97ba51SJulian Elischer } 15162e1ba83SRobert Watson 15262e1ba83SRobert Watson if (ro == NULL) { 15362e1ba83SRobert Watson ro = &iproute; 15462e1ba83SRobert Watson bzero(ro, sizeof (*ro)); 155bf984051SGleb Smirnoff } 15662e1ba83SRobert Watson 15753be8fcaSBjoern A. Zeeb #ifdef FLOWTABLE 158bf984051SGleb Smirnoff if (ro->ro_rt == NULL) { 159d4121a02SKip Macy struct flentry *fle; 160d4121a02SKip Macy 16162e1ba83SRobert Watson /* 16262e1ba83SRobert Watson * The flow table returns route entries valid for up to 30 16362e1ba83SRobert Watson * seconds; we rely on the remainder of ip_output() taking no 16462e1ba83SRobert Watson * longer than that long for the stability of ro_rt. The 16562e1ba83SRobert Watson * flow ID assignment must have happened before this point. 16662e1ba83SRobert Watson */ 167bf984051SGleb Smirnoff fle = flowtable_lookup_mbuf(V_ip_ft, m, AF_INET); 168bf984051SGleb Smirnoff if (fle != NULL) 169d4121a02SKip Macy flow_to_route(fle, ro); 170d4121a02SKip Macy } 17153be8fcaSBjoern A. Zeeb #endif 1722b25acc1SLuigi Rizzo 173df8bae1dSRodney W. Grimes if (opt) { 174ca8b83b0SLuigi Rizzo int len = 0; 175df8bae1dSRodney W. Grimes m = ip_insertoptions(m, opt, &len); 176cb7641e8SMaxim Konovalov if (len != 0) 177ca8b83b0SLuigi Rizzo hlen = len; /* ip->ip_hl is updated above */ 178df8bae1dSRodney W. Grimes } 179df8bae1dSRodney W. Grimes ip = mtod(m, struct ip *); 1808f134647SGleb Smirnoff ip_len = ntohs(ip->ip_len); 1818f134647SGleb Smirnoff ip_off = ntohs(ip->ip_off); 1823efc3014SJulian Elischer 183df8bae1dSRodney W. Grimes /* 1846e49b1feSGarrett Wollman * Fill in IP header. If we are not allowing fragmentation, 185e0f630eaSAndre Oppermann * then the ip_id field is meaningless, but we don't set it 186e0f630eaSAndre Oppermann * to zero. Doing so causes various problems when devices along 187e0f630eaSAndre Oppermann * the path (routers, load balancers, firewalls, etc.) illegally 188e0f630eaSAndre Oppermann * disable DF on our packet. Note that a 16-bit counter 1896e49b1feSGarrett Wollman * will wrap around in less than 10 seconds at 100 Mbit/s on a 1906e49b1feSGarrett Wollman * medium with MTU 1500. See Steven M. Bellovin, "A Technique 1916e49b1feSGarrett Wollman * for Counting NATted Hosts", Proc. IMW'02, available at 1924d09f5a0SGleb Smirnoff * <http://www.cs.columbia.edu/~smb/papers/fnat.pdf>. 193df8bae1dSRodney W. Grimes */ 194df8bae1dSRodney W. Grimes if ((flags & (IP_FORWARDING|IP_RAWOUTPUT)) == 0) { 19553be11f6SPoul-Henning Kamp ip->ip_v = IPVERSION; 19653be11f6SPoul-Henning Kamp ip->ip_hl = hlen >> 2; 1971f44b0a1SDavid Malone ip->ip_id = ip_newid(); 19886425c62SRobert Watson IPSTAT_INC(ips_localout); 199df8bae1dSRodney W. Grimes } else { 200ca8b83b0SLuigi Rizzo /* Header already set, fetch hlen from there */ 20153be11f6SPoul-Henning Kamp hlen = ip->ip_hl << 2; 202df8bae1dSRodney W. Grimes } 2039c9137eaSGarrett Wollman 2044c7a6059SGleb Smirnoff gw = dst = (struct sockaddr_in *)&ro->ro_dst; 2059b932e9eSAndre Oppermann again: 2063c73180fSGleb Smirnoff ia = NULL; 207df8bae1dSRodney W. Grimes /* 208df8bae1dSRodney W. Grimes * If there is a cached route, 209df8bae1dSRodney W. Grimes * check that it is to the same destination 210df8bae1dSRodney W. Grimes * and is still up. If not, free it and try again. 211a4a6e773SHajimu UMEMOTO * The address family should also be checked in case of sharing the 212a4a6e773SHajimu UMEMOTO * cache with IPv6. 213df8bae1dSRodney W. Grimes */ 214ec396e61SLuigi Rizzo rte = ro->ro_rt; 215ec396e61SLuigi Rizzo if (rte && ((rte->rt_flags & RTF_UP) == 0 || 216c7ea0aa6SQing Li rte->rt_ifp == NULL || 217c7ea0aa6SQing Li !RT_LINK_IS_UP(rte->rt_ifp) || 218a4a6e773SHajimu UMEMOTO dst->sin_family != AF_INET || 2199b932e9eSAndre Oppermann dst->sin_addr.s_addr != ip->ip_dst.s_addr)) { 220bf984051SGleb Smirnoff RO_RTFREE(ro); 221bf984051SGleb Smirnoff ro->ro_lle = NULL; 222bf984051SGleb Smirnoff rte = NULL; 223df8bae1dSRodney W. Grimes } 224ec396e61SLuigi Rizzo if (rte == NULL && fwd_tag == NULL) { 225a4a6e773SHajimu UMEMOTO bzero(dst, sizeof(*dst)); 226df8bae1dSRodney W. Grimes dst->sin_family = AF_INET; 227df8bae1dSRodney W. Grimes dst->sin_len = sizeof(*dst); 2289b932e9eSAndre Oppermann dst->sin_addr = ip->ip_dst; 229df8bae1dSRodney W. Grimes } 230df8bae1dSRodney W. Grimes /* 231a3fd02d8SBruce M Simpson * If routing to interface only, short circuit routing lookup. 232a3fd02d8SBruce M Simpson * The use of an all-ones broadcast address implies this; an 233a3fd02d8SBruce M Simpson * interface is specified by the broadcast address of an interface, 234a3fd02d8SBruce M Simpson * or the destination address of a ptp interface. 235df8bae1dSRodney W. Grimes */ 236a3fd02d8SBruce M Simpson if (flags & IP_SENDONES) { 237a3fd02d8SBruce M Simpson if ((ia = ifatoia(ifa_ifwithbroadaddr(sintosa(dst)))) == NULL && 238a3fd02d8SBruce M Simpson (ia = ifatoia(ifa_ifwithdstaddr(sintosa(dst)))) == NULL) { 23986425c62SRobert Watson IPSTAT_INC(ips_noroute); 240a3fd02d8SBruce M Simpson error = ENETUNREACH; 241a3fd02d8SBruce M Simpson goto bad; 242a3fd02d8SBruce M Simpson } 243a3fd02d8SBruce M Simpson ip->ip_dst.s_addr = INADDR_BROADCAST; 244a3fd02d8SBruce M Simpson dst->sin_addr = ip->ip_dst; 245a3fd02d8SBruce M Simpson ifp = ia->ia_ifp; 246a3fd02d8SBruce M Simpson ip->ip_ttl = 1; 247a3fd02d8SBruce M Simpson isbroadcast = 1; 248a3fd02d8SBruce M Simpson } else if (flags & IP_ROUTETOIF) { 2490b17fba7SAndre Oppermann if ((ia = ifatoia(ifa_ifwithdstaddr(sintosa(dst)))) == NULL && 2500ed6142bSQing Li (ia = ifatoia(ifa_ifwithnet(sintosa(dst), 0))) == NULL) { 25186425c62SRobert Watson IPSTAT_INC(ips_noroute); 252df8bae1dSRodney W. Grimes error = ENETUNREACH; 253df8bae1dSRodney W. Grimes goto bad; 254df8bae1dSRodney W. Grimes } 255df8bae1dSRodney W. Grimes ifp = ia->ia_ifp; 256df8bae1dSRodney W. Grimes ip->ip_ttl = 1; 2579f9b3dc4SGarrett Wollman isbroadcast = in_broadcast(dst->sin_addr, ifp); 2583afefa39SDaniel C. Sobral } else if (IN_MULTICAST(ntohl(ip->ip_dst.s_addr)) && 25938c1bc35SRuslan Ermilov imo != NULL && imo->imo_multicast_ifp != NULL) { 2603afefa39SDaniel C. Sobral /* 26138c1bc35SRuslan Ermilov * Bypass the normal routing lookup for multicast 26238c1bc35SRuslan Ermilov * packets if the interface is specified. 2633afefa39SDaniel C. Sobral */ 26438c1bc35SRuslan Ermilov ifp = imo->imo_multicast_ifp; 26538c1bc35SRuslan Ermilov IFP_TO_IA(ifp, ia); 26638c1bc35SRuslan Ermilov isbroadcast = 0; /* fool gcc */ 267df8bae1dSRodney W. Grimes } else { 2682c17fe93SGarrett Wollman /* 26997d8d152SAndre Oppermann * We want to do any cloning requested by the link layer, 27097d8d152SAndre Oppermann * as this is probably required in all cases for correct 27197d8d152SAndre Oppermann * operation (as it is for ARP). 2722c17fe93SGarrett Wollman */ 273ec396e61SLuigi Rizzo if (rte == NULL) { 274e440aed9SQing Li #ifdef RADIX_MPATH 2758b07e49aSJulian Elischer rtalloc_mpath_fib(ro, 2768b07e49aSJulian Elischer ntohl(ip->ip_src.s_addr ^ ip->ip_dst.s_addr), 2778b07e49aSJulian Elischer inp ? inp->inp_inc.inc_fibnum : M_GETFIB(m)); 278e440aed9SQing Li #else 2798b07e49aSJulian Elischer in_rtalloc_ign(ro, 0, 2808b07e49aSJulian Elischer inp ? inp->inp_inc.inc_fibnum : M_GETFIB(m)); 281e440aed9SQing Li #endif 282ec396e61SLuigi Rizzo rte = ro->ro_rt; 283ec396e61SLuigi Rizzo } 284c7ea0aa6SQing Li if (rte == NULL || 285c7ea0aa6SQing Li rte->rt_ifp == NULL || 286c7ea0aa6SQing Li !RT_LINK_IS_UP(rte->rt_ifp)) { 2871a499816SEdward Tomasz Napierala #ifdef IPSEC 2881a499816SEdward Tomasz Napierala /* 2891a499816SEdward Tomasz Napierala * There is no route for this packet, but it is 2901a499816SEdward Tomasz Napierala * possible that a matching SPD entry exists. 2911a499816SEdward Tomasz Napierala */ 2921a499816SEdward Tomasz Napierala no_route_but_check_spd = 1; 2931a499816SEdward Tomasz Napierala mtu = 0; /* Silence GCC warning. */ 2941a499816SEdward Tomasz Napierala goto sendit; 2951a499816SEdward Tomasz Napierala #endif 29686425c62SRobert Watson IPSTAT_INC(ips_noroute); 297df8bae1dSRodney W. Grimes error = EHOSTUNREACH; 298df8bae1dSRodney W. Grimes goto bad; 299df8bae1dSRodney W. Grimes } 300ec396e61SLuigi Rizzo ia = ifatoia(rte->rt_ifa); 3018c0fec80SRobert Watson ifa_ref(&ia->ia_ifa); 302ec396e61SLuigi Rizzo ifp = rte->rt_ifp; 303ec396e61SLuigi Rizzo rte->rt_rmx.rmx_pksent++; 304ec396e61SLuigi Rizzo if (rte->rt_flags & RTF_GATEWAY) 3054c7a6059SGleb Smirnoff gw = (struct sockaddr_in *)rte->rt_gateway; 306ec396e61SLuigi Rizzo if (rte->rt_flags & RTF_HOST) 307ec396e61SLuigi Rizzo isbroadcast = (rte->rt_flags & RTF_BROADCAST); 3089f9b3dc4SGarrett Wollman else 3094c7a6059SGleb Smirnoff isbroadcast = in_broadcast(gw->sin_addr, ifp); 310df8bae1dSRodney W. Grimes } 3113ae2ad08SJohn-Mark Gurney /* 3123ae2ad08SJohn-Mark Gurney * Calculate MTU. If we have a route that is up, use that, 3133ae2ad08SJohn-Mark Gurney * otherwise use the interface's MTU. 3143ae2ad08SJohn-Mark Gurney */ 315ec396e61SLuigi Rizzo if (rte != NULL && (rte->rt_flags & (RTF_UP|RTF_HOST))) { 3163ae2ad08SJohn-Mark Gurney /* 3173ae2ad08SJohn-Mark Gurney * This case can happen if the user changed the MTU 3183ae2ad08SJohn-Mark Gurney * of an interface after enabling IP on it. Because 3193ae2ad08SJohn-Mark Gurney * most netifs don't keep track of routes pointing to 3203ae2ad08SJohn-Mark Gurney * them, there is no way for one to update all its 3213ae2ad08SJohn-Mark Gurney * routes when the MTU is changed. 3223ae2ad08SJohn-Mark Gurney */ 323ec396e61SLuigi Rizzo if (rte->rt_rmx.rmx_mtu > ifp->if_mtu) 324ec396e61SLuigi Rizzo rte->rt_rmx.rmx_mtu = ifp->if_mtu; 325ec396e61SLuigi Rizzo mtu = rte->rt_rmx.rmx_mtu; 3263ae2ad08SJohn-Mark Gurney } else { 3273ae2ad08SJohn-Mark Gurney mtu = ifp->if_mtu; 3283ae2ad08SJohn-Mark Gurney } 329c744cde4SBjoern A. Zeeb /* Catch a possible divide by zero later. */ 330c744cde4SBjoern A. Zeeb KASSERT(mtu > 0, ("%s: mtu %d <= 0, rte=%p (rt_flags=0x%08x) ifp=%p", 331c744cde4SBjoern A. Zeeb __func__, mtu, rte, (rte != NULL) ? rte->rt_flags : 0, ifp)); 3329b932e9eSAndre Oppermann if (IN_MULTICAST(ntohl(ip->ip_dst.s_addr))) { 333df8bae1dSRodney W. Grimes m->m_flags |= M_MCAST; 334df8bae1dSRodney W. Grimes /* 335df8bae1dSRodney W. Grimes * See if the caller provided any multicast options 336df8bae1dSRodney W. Grimes */ 337df8bae1dSRodney W. Grimes if (imo != NULL) { 338df8bae1dSRodney W. Grimes ip->ip_ttl = imo->imo_multicast_ttl; 3391c5de19aSGarrett Wollman if (imo->imo_multicast_vif != -1) 3401c5de19aSGarrett Wollman ip->ip_src.s_addr = 341bbb4330bSLuigi Rizzo ip_mcast_src ? 342bbb4330bSLuigi Rizzo ip_mcast_src(imo->imo_multicast_vif) : 343bbb4330bSLuigi Rizzo INADDR_ANY; 344df8bae1dSRodney W. Grimes } else 345df8bae1dSRodney W. Grimes ip->ip_ttl = IP_DEFAULT_MULTICAST_TTL; 346df8bae1dSRodney W. Grimes /* 347df8bae1dSRodney W. Grimes * Confirm that the outgoing interface supports multicast. 348df8bae1dSRodney W. Grimes */ 3491c5de19aSGarrett Wollman if ((imo == NULL) || (imo->imo_multicast_vif == -1)) { 350df8bae1dSRodney W. Grimes if ((ifp->if_flags & IFF_MULTICAST) == 0) { 35186425c62SRobert Watson IPSTAT_INC(ips_noroute); 352df8bae1dSRodney W. Grimes error = ENETUNREACH; 353df8bae1dSRodney W. Grimes goto bad; 354df8bae1dSRodney W. Grimes } 3551c5de19aSGarrett Wollman } 356df8bae1dSRodney W. Grimes /* 357df8bae1dSRodney W. Grimes * If source address not specified yet, use address 358df8bae1dSRodney W. Grimes * of outgoing interface. 359df8bae1dSRodney W. Grimes */ 360df8bae1dSRodney W. Grimes if (ip->ip_src.s_addr == INADDR_ANY) { 36138c1bc35SRuslan Ermilov /* Interface may have no addresses. */ 36238c1bc35SRuslan Ermilov if (ia != NULL) 36338c1bc35SRuslan Ermilov ip->ip_src = IA_SIN(ia)->sin_addr; 364df8bae1dSRodney W. Grimes } 365df8bae1dSRodney W. Grimes 3668b889dbbSBruce M Simpson if ((imo == NULL && in_mcast_loop) || 3678b889dbbSBruce M Simpson (imo && imo->imo_multicast_loop)) { 368df8bae1dSRodney W. Grimes /* 3698b889dbbSBruce M Simpson * Loop back multicast datagram if not expressly 3708b889dbbSBruce M Simpson * forbidden to do so, even if we are not a member 3718b889dbbSBruce M Simpson * of the group; ip_input() will filter it later, 3728b889dbbSBruce M Simpson * thus deferring a hash lookup and mutex acquisition 3738b889dbbSBruce M Simpson * at the expense of a cheap copy using m_copym(). 374df8bae1dSRodney W. Grimes */ 37586b1d6d2SBill Fenner ip_mloopback(ifp, m, dst, hlen); 3768b889dbbSBruce M Simpson } else { 377df8bae1dSRodney W. Grimes /* 378df8bae1dSRodney W. Grimes * If we are acting as a multicast router, perform 379df8bae1dSRodney W. Grimes * multicast forwarding as if the packet had just 380df8bae1dSRodney W. Grimes * arrived on the interface to which we are about 381df8bae1dSRodney W. Grimes * to send. The multicast forwarding function 382df8bae1dSRodney W. Grimes * recursively calls this function, using the 383df8bae1dSRodney W. Grimes * IP_FORWARDING flag to prevent infinite recursion. 384df8bae1dSRodney W. Grimes * 385df8bae1dSRodney W. Grimes * Multicasts that are looped back by ip_mloopback(), 386df8bae1dSRodney W. Grimes * above, will be forwarded by the ip_input() routine, 387df8bae1dSRodney W. Grimes * if necessary. 388df8bae1dSRodney W. Grimes */ 389603724d3SBjoern A. Zeeb if (V_ip_mrouter && (flags & IP_FORWARDING) == 0) { 390f0068c4aSGarrett Wollman /* 391bbb4330bSLuigi Rizzo * If rsvp daemon is not running, do not 392f0068c4aSGarrett Wollman * set ip_moptions. This ensures that the packet 393f0068c4aSGarrett Wollman * is multicast and not just sent down one link 394f0068c4aSGarrett Wollman * as prescribed by rsvpd. 395f0068c4aSGarrett Wollman */ 396603724d3SBjoern A. Zeeb if (!V_rsvp_on) 397f0068c4aSGarrett Wollman imo = NULL; 398bbb4330bSLuigi Rizzo if (ip_mforward && 399bbb4330bSLuigi Rizzo ip_mforward(ip, ifp, m, imo) != 0) { 400df8bae1dSRodney W. Grimes m_freem(m); 401df8bae1dSRodney W. Grimes goto done; 402df8bae1dSRodney W. Grimes } 403df8bae1dSRodney W. Grimes } 404df8bae1dSRodney W. Grimes } 4055e9ae478SGarrett Wollman 406df8bae1dSRodney W. Grimes /* 407df8bae1dSRodney W. Grimes * Multicasts with a time-to-live of zero may be looped- 408df8bae1dSRodney W. Grimes * back, above, but must not be transmitted on a network. 409df8bae1dSRodney W. Grimes * Also, multicasts addressed to the loopback interface 410df8bae1dSRodney W. Grimes * are not sent -- the above call to ip_mloopback() will 4118b889dbbSBruce M Simpson * loop back a copy. ip_input() will drop the copy if 4128b889dbbSBruce M Simpson * this host does not belong to the destination group on 4138b889dbbSBruce M Simpson * the loopback interface. 414df8bae1dSRodney W. Grimes */ 415f5fea3ddSPaul Traina if (ip->ip_ttl == 0 || ifp->if_flags & IFF_LOOPBACK) { 416df8bae1dSRodney W. Grimes m_freem(m); 417df8bae1dSRodney W. Grimes goto done; 418df8bae1dSRodney W. Grimes } 419df8bae1dSRodney W. Grimes 420df8bae1dSRodney W. Grimes goto sendit; 421df8bae1dSRodney W. Grimes } 4226a7c943cSAndre Oppermann 423df8bae1dSRodney W. Grimes /* 4242b25acc1SLuigi Rizzo * If the source address is not specified yet, use the address 425cb459254SJohn-Mark Gurney * of the outoing interface. 426df8bae1dSRodney W. Grimes */ 427f9e354dfSJulian Elischer if (ip->ip_src.s_addr == INADDR_ANY) { 42838c1bc35SRuslan Ermilov /* Interface may have no addresses. */ 42938c1bc35SRuslan Ermilov if (ia != NULL) { 430df8bae1dSRodney W. Grimes ip->ip_src = IA_SIN(ia)->sin_addr; 431f9e354dfSJulian Elischer } 43238c1bc35SRuslan Ermilov } 4336a7c943cSAndre Oppermann 434ca7a789aSMax Laier /* 435ca7a789aSMax Laier * Verify that we have any chance at all of being able to queue the 436ca7a789aSMax Laier * packet or packet fragments, unless ALTQ is enabled on the given 437ca7a789aSMax Laier * interface in which case packetdrop should be done by queueing. 438ca7a789aSMax Laier */ 4398f134647SGleb Smirnoff n = ip_len / mtu + 1; /* how many fragments ? */ 440ca8b83b0SLuigi Rizzo if ( 44102b199f1SMax Laier #ifdef ALTQ 442ca8b83b0SLuigi Rizzo (!ALTQ_IS_ENABLED(&ifp->if_snd)) && 443ca7a789aSMax Laier #endif /* ALTQ */ 444ca8b83b0SLuigi Rizzo (ifp->if_snd.ifq_len + n) >= ifp->if_snd.ifq_maxlen ) { 445b5390296SDavid Greenman error = ENOBUFS; 44686425c62SRobert Watson IPSTAT_INC(ips_odropped); 447ca8b83b0SLuigi Rizzo ifp->if_snd.ifq_drops += n; 448b5390296SDavid Greenman goto bad; 449b5390296SDavid Greenman } 450b5390296SDavid Greenman 451b5390296SDavid Greenman /* 452df8bae1dSRodney W. Grimes * Look for broadcast address and 4538c3f5566SRuslan Ermilov * verify user is allowed to send 454df8bae1dSRodney W. Grimes * such a packet. 455df8bae1dSRodney W. Grimes */ 4569f9b3dc4SGarrett Wollman if (isbroadcast) { 457df8bae1dSRodney W. Grimes if ((ifp->if_flags & IFF_BROADCAST) == 0) { 458df8bae1dSRodney W. Grimes error = EADDRNOTAVAIL; 459df8bae1dSRodney W. Grimes goto bad; 460df8bae1dSRodney W. Grimes } 461df8bae1dSRodney W. Grimes if ((flags & IP_ALLOWBROADCAST) == 0) { 462df8bae1dSRodney W. Grimes error = EACCES; 463df8bae1dSRodney W. Grimes goto bad; 464df8bae1dSRodney W. Grimes } 465df8bae1dSRodney W. Grimes /* don't allow broadcast messages to be fragmented */ 4668f134647SGleb Smirnoff if (ip_len > mtu) { 467df8bae1dSRodney W. Grimes error = EMSGSIZE; 468df8bae1dSRodney W. Grimes goto bad; 469df8bae1dSRodney W. Grimes } 470df8bae1dSRodney W. Grimes m->m_flags |= M_BCAST; 4719f9b3dc4SGarrett Wollman } else { 472df8bae1dSRodney W. Grimes m->m_flags &= ~M_BCAST; 4739f9b3dc4SGarrett Wollman } 474df8bae1dSRodney W. Grimes 475f1743588SDarren Reed sendit: 476b2630c29SGeorge V. Neville-Neil #ifdef IPSEC 4774f7418a0SEdward Tomasz Napierala switch(ip_ipsec_output(&m, inp, &flags, &error)) { 4781dfcf0d2SAndre Oppermann case 1: 47933841545SHajimu UMEMOTO goto bad; 4801dfcf0d2SAndre Oppermann case -1: 4811dfcf0d2SAndre Oppermann goto done; 4821dfcf0d2SAndre Oppermann case 0: 48333841545SHajimu UMEMOTO default: 4841dfcf0d2SAndre Oppermann break; /* Continue with packet processing. */ 48533841545SHajimu UMEMOTO } 4861a499816SEdward Tomasz Napierala /* 4871a499816SEdward Tomasz Napierala * Check if there was a route for this packet; return error if not. 4881a499816SEdward Tomasz Napierala */ 4891a499816SEdward Tomasz Napierala if (no_route_but_check_spd) { 4901a499816SEdward Tomasz Napierala IPSTAT_INC(ips_noroute); 4911a499816SEdward Tomasz Napierala error = EHOSTUNREACH; 4921a499816SEdward Tomasz Napierala goto bad; 4931a499816SEdward Tomasz Napierala } 4941dfcf0d2SAndre Oppermann /* Update variables that are affected by ipsec4_output(). */ 49533841545SHajimu UMEMOTO ip = mtod(m, struct ip *); 49633841545SHajimu UMEMOTO hlen = ip->ip_hl << 2; 497b2630c29SGeorge V. Neville-Neil #endif /* IPSEC */ 49833841545SHajimu UMEMOTO 499c21fd232SAndre Oppermann /* Jump over all PFIL processing if hooks are not active. */ 5000b4b0b0fSJulian Elischer if (!PFIL_HOOKED(&V_inet_pfil_hook)) 501c21fd232SAndre Oppermann goto passout; 502c21fd232SAndre Oppermann 503c21fd232SAndre Oppermann /* Run through list of hooks for output packets. */ 5049b932e9eSAndre Oppermann odst.s_addr = ip->ip_dst.s_addr; 5050b4b0b0fSJulian Elischer error = pfil_run_hooks(&V_inet_pfil_hook, &m, ifp, PFIL_OUT, inp); 506134ea224SSam Leffler if (error != 0 || m == NULL) 507c4ac87eaSDarren Reed goto done; 5089b932e9eSAndre Oppermann 509c4ac87eaSDarren Reed ip = mtod(m, struct ip *); 510fed1c7e9SSøren Schmidt 5119b932e9eSAndre Oppermann /* See if destination IP address was changed by packet filter. */ 5129b932e9eSAndre Oppermann if (odst.s_addr != ip->ip_dst.s_addr) { 5139b932e9eSAndre Oppermann m->m_flags |= M_SKIP_FIREWALL; 514f4fca2d8SAndre Oppermann /* If destination is now ourself drop to ip_input(). */ 5159b932e9eSAndre Oppermann if (in_localip(ip->ip_dst)) { 5169b932e9eSAndre Oppermann m->m_flags |= M_FASTFWD_OURS; 517f9e354dfSJulian Elischer if (m->m_pkthdr.rcvif == NULL) 518603724d3SBjoern A. Zeeb m->m_pkthdr.rcvif = V_loif; 51920c822f3SJonathan Lemon if (m->m_pkthdr.csum_flags & CSUM_DELAY_DATA) { 52020c822f3SJonathan Lemon m->m_pkthdr.csum_flags |= 52120c822f3SJonathan Lemon CSUM_DATA_VALID | CSUM_PSEUDO_HDR; 522ac9d7e26SMax Laier m->m_pkthdr.csum_data = 0xffff; 52320c822f3SJonathan Lemon } 52420c822f3SJonathan Lemon m->m_pkthdr.csum_flags |= 52520c822f3SJonathan Lemon CSUM_IP_CHECKED | CSUM_IP_VALID; 5262f4afd21SRandall Stewart #ifdef SCTP 5272f4afd21SRandall Stewart if (m->m_pkthdr.csum_flags & CSUM_SCTP) 5282f4afd21SRandall Stewart m->m_pkthdr.csum_flags |= CSUM_SCTP_VALID; 5292f4afd21SRandall Stewart #endif 5309b932e9eSAndre Oppermann error = netisr_queue(NETISR_IP, m); 5319b932e9eSAndre Oppermann goto done; 5323c73180fSGleb Smirnoff } else { 5333c73180fSGleb Smirnoff if (ia != NULL) 5343c73180fSGleb Smirnoff ifa_free(&ia->ia_ifa); 535f4fca2d8SAndre Oppermann goto again; /* Redo the routing table lookup. */ 5369b932e9eSAndre Oppermann } 5373c73180fSGleb Smirnoff } 5389b932e9eSAndre Oppermann 5399b932e9eSAndre Oppermann /* See if local, if yes, send it to netisr with IP_FASTFWD_OURS. */ 5409b932e9eSAndre Oppermann if (m->m_flags & M_FASTFWD_OURS) { 5419b932e9eSAndre Oppermann if (m->m_pkthdr.rcvif == NULL) 542603724d3SBjoern A. Zeeb m->m_pkthdr.rcvif = V_loif; 5439b932e9eSAndre Oppermann if (m->m_pkthdr.csum_flags & CSUM_DELAY_DATA) { 5449b932e9eSAndre Oppermann m->m_pkthdr.csum_flags |= 5459b932e9eSAndre Oppermann CSUM_DATA_VALID | CSUM_PSEUDO_HDR; 5469b932e9eSAndre Oppermann m->m_pkthdr.csum_data = 0xffff; 5479b932e9eSAndre Oppermann } 5482f4afd21SRandall Stewart #ifdef SCTP 5492f4afd21SRandall Stewart if (m->m_pkthdr.csum_flags & CSUM_SCTP) 5502f4afd21SRandall Stewart m->m_pkthdr.csum_flags |= CSUM_SCTP_VALID; 5512f4afd21SRandall Stewart #endif 5529b932e9eSAndre Oppermann m->m_pkthdr.csum_flags |= 5539b932e9eSAndre Oppermann CSUM_IP_CHECKED | CSUM_IP_VALID; 5549b932e9eSAndre Oppermann 5559b932e9eSAndre Oppermann error = netisr_queue(NETISR_IP, m); 556f9e354dfSJulian Elischer goto done; 557f9e354dfSJulian Elischer } 5589b932e9eSAndre Oppermann /* Or forward to some other address? */ 559ffdbf9daSAndrey V. Elsukov if ((m->m_flags & M_IP_NEXTHOP) && 560ffdbf9daSAndrey V. Elsukov (fwd_tag = m_tag_find(m, PACKET_TAG_IPFORWARD, NULL)) != NULL) { 5619b932e9eSAndre Oppermann bcopy((fwd_tag+1), dst, sizeof(struct sockaddr_in)); 5629b932e9eSAndre Oppermann m->m_flags |= M_SKIP_FIREWALL; 563ffdbf9daSAndrey V. Elsukov m->m_flags &= ~M_IP_NEXTHOP; 5649b932e9eSAndre Oppermann m_tag_delete(m, fwd_tag); 5653c73180fSGleb Smirnoff if (ia != NULL) 5663c73180fSGleb Smirnoff ifa_free(&ia->ia_ifa); 5679b932e9eSAndre Oppermann goto again; 568099dd043SAndre Oppermann } 5692b25acc1SLuigi Rizzo 570c21fd232SAndre Oppermann passout: 57151c8ec4aSRuslan Ermilov /* 127/8 must not appear on wire - RFC1122. */ 57251c8ec4aSRuslan Ermilov if ((ntohl(ip->ip_dst.s_addr) >> IN_CLASSA_NSHIFT) == IN_LOOPBACKNET || 57351c8ec4aSRuslan Ermilov (ntohl(ip->ip_src.s_addr) >> IN_CLASSA_NSHIFT) == IN_LOOPBACKNET) { 57451c8ec4aSRuslan Ermilov if ((ifp->if_flags & IFF_LOOPBACK) == 0) { 57586425c62SRobert Watson IPSTAT_INC(ips_badaddr); 57651c8ec4aSRuslan Ermilov error = EADDRNOTAVAIL; 57751c8ec4aSRuslan Ermilov goto bad; 57851c8ec4aSRuslan Ermilov } 57951c8ec4aSRuslan Ermilov } 58051c8ec4aSRuslan Ermilov 581206a3274SRuslan Ermilov m->m_pkthdr.csum_flags |= CSUM_IP; 582078468edSGleb Smirnoff if (m->m_pkthdr.csum_flags & CSUM_DELAY_DATA & ~ifp->if_hwassist) { 583db4f9cc7SJonathan Lemon in_delayed_cksum(m); 584078468edSGleb Smirnoff m->m_pkthdr.csum_flags &= ~CSUM_DELAY_DATA; 585db4f9cc7SJonathan Lemon } 5862f4afd21SRandall Stewart #ifdef SCTP 587078468edSGleb Smirnoff if (m->m_pkthdr.csum_flags & CSUM_SCTP & ~ifp->if_hwassist) { 5881966e5b5SRandall Stewart sctp_delayed_cksum(m, (uint32_t)(ip->ip_hl << 2)); 589078468edSGleb Smirnoff m->m_pkthdr.csum_flags &= ~CSUM_SCTP; 5902f4afd21SRandall Stewart } 5912f4afd21SRandall Stewart #endif 592db4f9cc7SJonathan Lemon 593e7319babSPoul-Henning Kamp /* 594db4f9cc7SJonathan Lemon * If small enough for interface, or the interface will take 595233dcce1SAndre Oppermann * care of the fragmentation for us, we can just send directly. 596df8bae1dSRodney W. Grimes */ 59721d172a3SGleb Smirnoff if (ip_len <= mtu || 598233dcce1SAndre Oppermann (m->m_pkthdr.csum_flags & ifp->if_hwassist & CSUM_TSO) != 0 || 59921d172a3SGleb Smirnoff ((ip_off & IP_DF) == 0 && (ifp->if_hwassist & CSUM_FRAGMENT))) { 600df8bae1dSRodney W. Grimes ip->ip_sum = 0; 601078468edSGleb Smirnoff if (m->m_pkthdr.csum_flags & CSUM_IP & ~ifp->if_hwassist) { 602df8bae1dSRodney W. Grimes ip->ip_sum = in_cksum(m, hlen); 603078468edSGleb Smirnoff m->m_pkthdr.csum_flags &= ~CSUM_IP; 604078468edSGleb Smirnoff } 6055da9f8faSJosef Karthauser 606233dcce1SAndre Oppermann /* 607233dcce1SAndre Oppermann * Record statistics for this interface address. 608233dcce1SAndre Oppermann * With CSUM_TSO the byte/packet count will be slightly 609233dcce1SAndre Oppermann * incorrect because we count the IP+TCP headers only 610233dcce1SAndre Oppermann * once instead of for every generated packet. 611233dcce1SAndre Oppermann */ 61238c1bc35SRuslan Ermilov if (!(flags & IP_FORWARDING) && ia) { 613233dcce1SAndre Oppermann if (m->m_pkthdr.csum_flags & CSUM_TSO) 6147caf4ab7SGleb Smirnoff counter_u64_add(ia->ia_ifa.ifa_opackets, 6157caf4ab7SGleb Smirnoff m->m_pkthdr.len / m->m_pkthdr.tso_segsz); 616233dcce1SAndre Oppermann else 6177caf4ab7SGleb Smirnoff counter_u64_add(ia->ia_ifa.ifa_opackets, 1); 6187caf4ab7SGleb Smirnoff 6197caf4ab7SGleb Smirnoff counter_u64_add(ia->ia_ifa.ifa_obytes, m->m_pkthdr.len); 6205da9f8faSJosef Karthauser } 62153dcc544SMike Silbersack #ifdef MBUF_STRESS_TEST 6223390d476SMike Silbersack if (mbuf_frag_size && m->m_pkthdr.len > mbuf_frag_size) 623eb1b1807SGleb Smirnoff m = m_fragment(m, M_NOWAIT, mbuf_frag_size); 6249d9edc56SMike Silbersack #endif 625147f74d1SAndre Oppermann /* 626147f74d1SAndre Oppermann * Reset layer specific mbuf flags 627147f74d1SAndre Oppermann * to avoid confusing lower layers. 628147f74d1SAndre Oppermann */ 62986bd0491SAndre Oppermann m_clrprotoflags(m); 63057f60867SMark Johnston IP_PROBE(send, NULL, NULL, ip, ifp, ip, NULL); 63147e8d432SGleb Smirnoff error = (*ifp->if_output)(ifp, m, 63247e8d432SGleb Smirnoff (const struct sockaddr *)gw, ro); 633df8bae1dSRodney W. Grimes goto done; 634df8bae1dSRodney W. Grimes } 6351e78ac21SJeffrey Hsu 636233dcce1SAndre Oppermann /* Balk when DF bit is set or the interface didn't support TSO. */ 63721d172a3SGleb Smirnoff if ((ip_off & IP_DF) || (m->m_pkthdr.csum_flags & CSUM_TSO)) { 638df8bae1dSRodney W. Grimes error = EMSGSIZE; 63986425c62SRobert Watson IPSTAT_INC(ips_cantfrag); 640df8bae1dSRodney W. Grimes goto bad; 641df8bae1dSRodney W. Grimes } 6421e78ac21SJeffrey Hsu 6431e78ac21SJeffrey Hsu /* 6441e78ac21SJeffrey Hsu * Too large for interface; fragment if possible. If successful, 6451e78ac21SJeffrey Hsu * on return, m will point to a list of packets to be sent. 6461e78ac21SJeffrey Hsu */ 647078468edSGleb Smirnoff error = ip_fragment(ip, &m, mtu, ifp->if_hwassist); 6481e78ac21SJeffrey Hsu if (error) 649df8bae1dSRodney W. Grimes goto bad; 6501e78ac21SJeffrey Hsu for (; m; m = m0) { 651df8bae1dSRodney W. Grimes m0 = m->m_nextpkt; 652b375c9ecSLuigi Rizzo m->m_nextpkt = 0; 65307203494SDaniel C. Sobral if (error == 0) { 654fe937674SJosef Karthauser /* Record statistics for this interface address. */ 65507203494SDaniel C. Sobral if (ia != NULL) { 6567caf4ab7SGleb Smirnoff counter_u64_add(ia->ia_ifa.ifa_opackets, 1); 6577caf4ab7SGleb Smirnoff counter_u64_add(ia->ia_ifa.ifa_obytes, 6587caf4ab7SGleb Smirnoff m->m_pkthdr.len); 65907203494SDaniel C. Sobral } 660147f74d1SAndre Oppermann /* 661147f74d1SAndre Oppermann * Reset layer specific mbuf flags 662147f74d1SAndre Oppermann * to avoid confusing upper layers. 663147f74d1SAndre Oppermann */ 66486bd0491SAndre Oppermann m_clrprotoflags(m); 665fe937674SJosef Karthauser 66657f60867SMark Johnston IP_PROBE(send, NULL, NULL, ip, ifp, ip, NULL); 667df8bae1dSRodney W. Grimes error = (*ifp->if_output)(ifp, m, 66847e8d432SGleb Smirnoff (const struct sockaddr *)gw, ro); 669fe937674SJosef Karthauser } else 670df8bae1dSRodney W. Grimes m_freem(m); 671df8bae1dSRodney W. Grimes } 672df8bae1dSRodney W. Grimes 673df8bae1dSRodney W. Grimes if (error == 0) 67486425c62SRobert Watson IPSTAT_INC(ips_fragmented); 6751e78ac21SJeffrey Hsu 676df8bae1dSRodney W. Grimes done: 677bf984051SGleb Smirnoff if (ro == &iproute) 678bf984051SGleb Smirnoff RO_RTFREE(ro); 6798c0fec80SRobert Watson if (ia != NULL) 6808c0fec80SRobert Watson ifa_free(&ia->ia_ifa); 681df8bae1dSRodney W. Grimes return (error); 682df8bae1dSRodney W. Grimes bad: 6833528d68fSBill Paul m_freem(m); 684df8bae1dSRodney W. Grimes goto done; 685df8bae1dSRodney W. Grimes } 686df8bae1dSRodney W. Grimes 6871e78ac21SJeffrey Hsu /* 6881e78ac21SJeffrey Hsu * Create a chain of fragments which fit the given mtu. m_frag points to the 6891e78ac21SJeffrey Hsu * mbuf to be fragmented; on return it points to the chain with the fragments. 6901e78ac21SJeffrey Hsu * Return 0 if no error. If error, m_frag may contain a partially built 6911e78ac21SJeffrey Hsu * chain of fragments that should be freed by the caller. 6921e78ac21SJeffrey Hsu * 6931e78ac21SJeffrey Hsu * if_hwassist_flags is the hw offload capabilities (see if_data.ifi_hwassist) 6941e78ac21SJeffrey Hsu */ 6951e78ac21SJeffrey Hsu int 6961e78ac21SJeffrey Hsu ip_fragment(struct ip *ip, struct mbuf **m_frag, int mtu, 697078468edSGleb Smirnoff u_long if_hwassist_flags) 6981e78ac21SJeffrey Hsu { 6991e78ac21SJeffrey Hsu int error = 0; 7001e78ac21SJeffrey Hsu int hlen = ip->ip_hl << 2; 7011e78ac21SJeffrey Hsu int len = (mtu - hlen) & ~7; /* size of payload in each fragment */ 7021e78ac21SJeffrey Hsu int off; 7031e78ac21SJeffrey Hsu struct mbuf *m0 = *m_frag; /* the original packet */ 7041e78ac21SJeffrey Hsu int firstlen; 7051e78ac21SJeffrey Hsu struct mbuf **mnext; 7061e78ac21SJeffrey Hsu int nfrags; 70721d172a3SGleb Smirnoff uint16_t ip_len, ip_off; 7081e78ac21SJeffrey Hsu 70921d172a3SGleb Smirnoff ip_len = ntohs(ip->ip_len); 71021d172a3SGleb Smirnoff ip_off = ntohs(ip->ip_off); 71121d172a3SGleb Smirnoff 71221d172a3SGleb Smirnoff if (ip_off & IP_DF) { /* Fragmentation not allowed */ 71386425c62SRobert Watson IPSTAT_INC(ips_cantfrag); 7141e78ac21SJeffrey Hsu return EMSGSIZE; 7151e78ac21SJeffrey Hsu } 7161e78ac21SJeffrey Hsu 7171e78ac21SJeffrey Hsu /* 7181e78ac21SJeffrey Hsu * Must be able to put at least 8 bytes per fragment. 7191e78ac21SJeffrey Hsu */ 7201e78ac21SJeffrey Hsu if (len < 8) 7211e78ac21SJeffrey Hsu return EMSGSIZE; 7221e78ac21SJeffrey Hsu 7231e78ac21SJeffrey Hsu /* 7241e78ac21SJeffrey Hsu * If the interface will not calculate checksums on 7251e78ac21SJeffrey Hsu * fragmented packets, then do it here. 7261e78ac21SJeffrey Hsu */ 727da2299c5SAndre Oppermann if (m0->m_pkthdr.csum_flags & CSUM_DELAY_DATA) { 7281e78ac21SJeffrey Hsu in_delayed_cksum(m0); 7291e78ac21SJeffrey Hsu m0->m_pkthdr.csum_flags &= ~CSUM_DELAY_DATA; 7301e78ac21SJeffrey Hsu } 7312f4afd21SRandall Stewart #ifdef SCTP 732da2299c5SAndre Oppermann if (m0->m_pkthdr.csum_flags & CSUM_SCTP) { 7331966e5b5SRandall Stewart sctp_delayed_cksum(m0, hlen); 7342f4afd21SRandall Stewart m0->m_pkthdr.csum_flags &= ~CSUM_SCTP; 7352f4afd21SRandall Stewart } 7362f4afd21SRandall Stewart #endif 7371e78ac21SJeffrey Hsu if (len > PAGE_SIZE) { 7381e78ac21SJeffrey Hsu /* 7391e78ac21SJeffrey Hsu * Fragment large datagrams such that each segment 7401e78ac21SJeffrey Hsu * contains a multiple of PAGE_SIZE amount of data, 7411e78ac21SJeffrey Hsu * plus headers. This enables a receiver to perform 7421e78ac21SJeffrey Hsu * page-flipping zero-copy optimizations. 7431e78ac21SJeffrey Hsu * 7441e78ac21SJeffrey Hsu * XXX When does this help given that sender and receiver 7451e78ac21SJeffrey Hsu * could have different page sizes, and also mtu could 7461e78ac21SJeffrey Hsu * be less than the receiver's page size ? 7471e78ac21SJeffrey Hsu */ 7481e78ac21SJeffrey Hsu int newlen; 7491e78ac21SJeffrey Hsu struct mbuf *m; 7501e78ac21SJeffrey Hsu 7511e78ac21SJeffrey Hsu for (m = m0, off = 0; m && (off+m->m_len) <= mtu; m = m->m_next) 7521e78ac21SJeffrey Hsu off += m->m_len; 7531e78ac21SJeffrey Hsu 7541e78ac21SJeffrey Hsu /* 7551e78ac21SJeffrey Hsu * firstlen (off - hlen) must be aligned on an 7561e78ac21SJeffrey Hsu * 8-byte boundary 7571e78ac21SJeffrey Hsu */ 7581e78ac21SJeffrey Hsu if (off < hlen) 7591e78ac21SJeffrey Hsu goto smart_frag_failure; 7601e78ac21SJeffrey Hsu off = ((off - hlen) & ~7) + hlen; 7611e78ac21SJeffrey Hsu newlen = (~PAGE_MASK) & mtu; 7621e78ac21SJeffrey Hsu if ((newlen + sizeof (struct ip)) > mtu) { 7631e78ac21SJeffrey Hsu /* we failed, go back the default */ 7641e78ac21SJeffrey Hsu smart_frag_failure: 7651e78ac21SJeffrey Hsu newlen = len; 7661e78ac21SJeffrey Hsu off = hlen + len; 7671e78ac21SJeffrey Hsu } 7681e78ac21SJeffrey Hsu len = newlen; 7691e78ac21SJeffrey Hsu 7701e78ac21SJeffrey Hsu } else { 7711e78ac21SJeffrey Hsu off = hlen + len; 7721e78ac21SJeffrey Hsu } 7731e78ac21SJeffrey Hsu 7741e78ac21SJeffrey Hsu firstlen = off - hlen; 7751e78ac21SJeffrey Hsu mnext = &m0->m_nextpkt; /* pointer to next packet */ 7761e78ac21SJeffrey Hsu 7771e78ac21SJeffrey Hsu /* 7781e78ac21SJeffrey Hsu * Loop through length of segment after first fragment, 7791e78ac21SJeffrey Hsu * make new header and copy data of each part and link onto chain. 7801e78ac21SJeffrey Hsu * Here, m0 is the original packet, m is the fragment being created. 7811e78ac21SJeffrey Hsu * The fragments are linked off the m_nextpkt of the original 7821e78ac21SJeffrey Hsu * packet, which after processing serves as the first fragment. 7831e78ac21SJeffrey Hsu */ 78421d172a3SGleb Smirnoff for (nfrags = 1; off < ip_len; off += len, nfrags++) { 7851e78ac21SJeffrey Hsu struct ip *mhip; /* ip header on the fragment */ 7861e78ac21SJeffrey Hsu struct mbuf *m; 7871e78ac21SJeffrey Hsu int mhlen = sizeof (struct ip); 7881e78ac21SJeffrey Hsu 789dc4ad05eSGleb Smirnoff m = m_gethdr(M_NOWAIT, MT_DATA); 7900b17fba7SAndre Oppermann if (m == NULL) { 7911e78ac21SJeffrey Hsu error = ENOBUFS; 79286425c62SRobert Watson IPSTAT_INC(ips_odropped); 7931e78ac21SJeffrey Hsu goto done; 7941e78ac21SJeffrey Hsu } 795fb86dfcdSAndre Oppermann m->m_flags |= (m0->m_flags & M_MCAST); 7961e78ac21SJeffrey Hsu /* 7971e78ac21SJeffrey Hsu * In the first mbuf, leave room for the link header, then 7981e78ac21SJeffrey Hsu * copy the original IP header including options. The payload 7998b889dbbSBruce M Simpson * goes into an additional mbuf chain returned by m_copym(). 8001e78ac21SJeffrey Hsu */ 8011e78ac21SJeffrey Hsu m->m_data += max_linkhdr; 8021e78ac21SJeffrey Hsu mhip = mtod(m, struct ip *); 8031e78ac21SJeffrey Hsu *mhip = *ip; 8041e78ac21SJeffrey Hsu if (hlen > sizeof (struct ip)) { 8051e78ac21SJeffrey Hsu mhlen = ip_optcopy(ip, mhip) + sizeof (struct ip); 8061e78ac21SJeffrey Hsu mhip->ip_v = IPVERSION; 8071e78ac21SJeffrey Hsu mhip->ip_hl = mhlen >> 2; 8081e78ac21SJeffrey Hsu } 8091e78ac21SJeffrey Hsu m->m_len = mhlen; 81021d172a3SGleb Smirnoff /* XXX do we need to add ip_off below ? */ 81121d172a3SGleb Smirnoff mhip->ip_off = ((off - hlen) >> 3) + ip_off; 812fb86dfcdSAndre Oppermann if (off + len >= ip_len) 81321d172a3SGleb Smirnoff len = ip_len - off; 814fb86dfcdSAndre Oppermann else 8151e78ac21SJeffrey Hsu mhip->ip_off |= IP_MF; 8161e78ac21SJeffrey Hsu mhip->ip_len = htons((u_short)(len + mhlen)); 817eb1b1807SGleb Smirnoff m->m_next = m_copym(m0, off, len, M_NOWAIT); 8180b17fba7SAndre Oppermann if (m->m_next == NULL) { /* copy failed */ 8191e78ac21SJeffrey Hsu m_free(m); 8201e78ac21SJeffrey Hsu error = ENOBUFS; /* ??? */ 82186425c62SRobert Watson IPSTAT_INC(ips_odropped); 8221e78ac21SJeffrey Hsu goto done; 8231e78ac21SJeffrey Hsu } 8241e78ac21SJeffrey Hsu m->m_pkthdr.len = mhlen + len; 825fc74a9f9SBrooks Davis m->m_pkthdr.rcvif = NULL; 8261e78ac21SJeffrey Hsu #ifdef MAC 82730d239bcSRobert Watson mac_netinet_fragment(m0, m); 8281e78ac21SJeffrey Hsu #endif 8291e78ac21SJeffrey Hsu m->m_pkthdr.csum_flags = m0->m_pkthdr.csum_flags; 8301e78ac21SJeffrey Hsu mhip->ip_off = htons(mhip->ip_off); 8311e78ac21SJeffrey Hsu mhip->ip_sum = 0; 832078468edSGleb Smirnoff if (m->m_pkthdr.csum_flags & CSUM_IP & ~if_hwassist_flags) { 8331e78ac21SJeffrey Hsu mhip->ip_sum = in_cksum(m, mhlen); 834078468edSGleb Smirnoff m->m_pkthdr.csum_flags &= ~CSUM_IP; 835078468edSGleb Smirnoff } 8361e78ac21SJeffrey Hsu *mnext = m; 8371e78ac21SJeffrey Hsu mnext = &m->m_nextpkt; 8381e78ac21SJeffrey Hsu } 83986425c62SRobert Watson IPSTAT_ADD(ips_ofragments, nfrags); 8401e78ac21SJeffrey Hsu 8411e78ac21SJeffrey Hsu /* 8421e78ac21SJeffrey Hsu * Update first fragment by trimming what's been copied out 8431e78ac21SJeffrey Hsu * and updating header. 8441e78ac21SJeffrey Hsu */ 84521d172a3SGleb Smirnoff m_adj(m0, hlen + firstlen - ip_len); 8461e78ac21SJeffrey Hsu m0->m_pkthdr.len = hlen + firstlen; 8471e78ac21SJeffrey Hsu ip->ip_len = htons((u_short)m0->m_pkthdr.len); 84821d172a3SGleb Smirnoff ip->ip_off = htons(ip_off | IP_MF); 8491e78ac21SJeffrey Hsu ip->ip_sum = 0; 850078468edSGleb Smirnoff if (m0->m_pkthdr.csum_flags & CSUM_IP & ~if_hwassist_flags) { 8511e78ac21SJeffrey Hsu ip->ip_sum = in_cksum(m0, hlen); 852078468edSGleb Smirnoff m0->m_pkthdr.csum_flags &= ~CSUM_IP; 853078468edSGleb Smirnoff } 8541e78ac21SJeffrey Hsu 8551e78ac21SJeffrey Hsu done: 8561e78ac21SJeffrey Hsu *m_frag = m0; 8571e78ac21SJeffrey Hsu return error; 8581e78ac21SJeffrey Hsu } 8591e78ac21SJeffrey Hsu 8601c238475SJonathan Lemon void 861db4f9cc7SJonathan Lemon in_delayed_cksum(struct mbuf *m) 862db4f9cc7SJonathan Lemon { 863db4f9cc7SJonathan Lemon struct ip *ip; 86423e9c6dcSGleb Smirnoff uint16_t csum, offset, ip_len; 865db4f9cc7SJonathan Lemon 866db4f9cc7SJonathan Lemon ip = mtod(m, struct ip *); 86753be11f6SPoul-Henning Kamp offset = ip->ip_hl << 2 ; 86823e9c6dcSGleb Smirnoff ip_len = ntohs(ip->ip_len); 86923e9c6dcSGleb Smirnoff csum = in_cksum_skip(m, ip_len, offset); 870206a3274SRuslan Ermilov if (m->m_pkthdr.csum_flags & CSUM_UDP && csum == 0) 871206a3274SRuslan Ermilov csum = 0xffff; 872db4f9cc7SJonathan Lemon offset += m->m_pkthdr.csum_data; /* checksum offset */ 873db4f9cc7SJonathan Lemon 874db4f9cc7SJonathan Lemon if (offset + sizeof(u_short) > m->m_len) { 875db4f9cc7SJonathan Lemon printf("delayed m_pullup, m->len: %d off: %d p: %d\n", 876db4f9cc7SJonathan Lemon m->m_len, offset, ip->ip_p); 877db4f9cc7SJonathan Lemon /* 878db4f9cc7SJonathan Lemon * XXX 879db4f9cc7SJonathan Lemon * this shouldn't happen, but if it does, the 880db4f9cc7SJonathan Lemon * correct behavior may be to insert the checksum 8818f8d29f6SAndre Oppermann * in the appropriate next mbuf in the chain. 882db4f9cc7SJonathan Lemon */ 8838f8d29f6SAndre Oppermann return; 884db4f9cc7SJonathan Lemon } 885db4f9cc7SJonathan Lemon *(u_short *)(m->m_data + offset) = csum; 886db4f9cc7SJonathan Lemon } 887db4f9cc7SJonathan Lemon 888df8bae1dSRodney W. Grimes /* 889df8bae1dSRodney W. Grimes * IP socket option processing. 890df8bae1dSRodney W. Grimes */ 891df8bae1dSRodney W. Grimes int 892f2565d68SRobert Watson ip_ctloutput(struct socket *so, struct sockopt *sopt) 893df8bae1dSRodney W. Grimes { 894cfe8b629SGarrett Wollman struct inpcb *inp = sotoinpcb(so); 895cfe8b629SGarrett Wollman int error, optval; 896df8bae1dSRodney W. Grimes 897cfe8b629SGarrett Wollman error = optval = 0; 898cfe8b629SGarrett Wollman if (sopt->sopt_level != IPPROTO_IP) { 899fc06cd42SMikolaj Golub error = EINVAL; 900fc06cd42SMikolaj Golub 901fc06cd42SMikolaj Golub if (sopt->sopt_level == SOL_SOCKET && 902fc06cd42SMikolaj Golub sopt->sopt_dir == SOPT_SET) { 903fc06cd42SMikolaj Golub switch (sopt->sopt_name) { 904fc06cd42SMikolaj Golub case SO_REUSEADDR: 905fc06cd42SMikolaj Golub INP_WLOCK(inp); 906efdf104bSMikolaj Golub if ((so->so_options & SO_REUSEADDR) != 0) 907efdf104bSMikolaj Golub inp->inp_flags2 |= INP_REUSEADDR; 908fc06cd42SMikolaj Golub else 909efdf104bSMikolaj Golub inp->inp_flags2 &= ~INP_REUSEADDR; 910fc06cd42SMikolaj Golub INP_WUNLOCK(inp); 911fc06cd42SMikolaj Golub error = 0; 912fc06cd42SMikolaj Golub break; 913fc06cd42SMikolaj Golub case SO_REUSEPORT: 914fc06cd42SMikolaj Golub INP_WLOCK(inp); 915fc06cd42SMikolaj Golub if ((so->so_options & SO_REUSEPORT) != 0) 916fc06cd42SMikolaj Golub inp->inp_flags2 |= INP_REUSEPORT; 917fc06cd42SMikolaj Golub else 918fc06cd42SMikolaj Golub inp->inp_flags2 &= ~INP_REUSEPORT; 919fc06cd42SMikolaj Golub INP_WUNLOCK(inp); 920fc06cd42SMikolaj Golub error = 0; 921fc06cd42SMikolaj Golub break; 922fc06cd42SMikolaj Golub case SO_SETFIB: 923fc06cd42SMikolaj Golub INP_WLOCK(inp); 924fc06cd42SMikolaj Golub inp->inp_inc.inc_fibnum = so->so_fibnum; 925fc06cd42SMikolaj Golub INP_WUNLOCK(inp); 926fc06cd42SMikolaj Golub error = 0; 927fc06cd42SMikolaj Golub break; 928fc06cd42SMikolaj Golub default: 929fc06cd42SMikolaj Golub break; 930fc06cd42SMikolaj Golub } 931fc06cd42SMikolaj Golub } 932fc06cd42SMikolaj Golub return (error); 933cfe8b629SGarrett Wollman } 934df8bae1dSRodney W. Grimes 935cfe8b629SGarrett Wollman switch (sopt->sopt_dir) { 936cfe8b629SGarrett Wollman case SOPT_SET: 937cfe8b629SGarrett Wollman switch (sopt->sopt_name) { 938df8bae1dSRodney W. Grimes case IP_OPTIONS: 939df8bae1dSRodney W. Grimes #ifdef notyet 940df8bae1dSRodney W. Grimes case IP_RETOPTS: 941df8bae1dSRodney W. Grimes #endif 942cfe8b629SGarrett Wollman { 943cfe8b629SGarrett Wollman struct mbuf *m; 944cfe8b629SGarrett Wollman if (sopt->sopt_valsize > MLEN) { 945cfe8b629SGarrett Wollman error = EMSGSIZE; 946cfe8b629SGarrett Wollman break; 947cfe8b629SGarrett Wollman } 948dc4ad05eSGleb Smirnoff m = m_get(sopt->sopt_td ? M_WAITOK : M_NOWAIT, MT_DATA); 9490b17fba7SAndre Oppermann if (m == NULL) { 950cfe8b629SGarrett Wollman error = ENOBUFS; 951cfe8b629SGarrett Wollman break; 952cfe8b629SGarrett Wollman } 953cfe8b629SGarrett Wollman m->m_len = sopt->sopt_valsize; 954cfe8b629SGarrett Wollman error = sooptcopyin(sopt, mtod(m, char *), m->m_len, 955cfe8b629SGarrett Wollman m->m_len); 956635354c4SMaxim Konovalov if (error) { 957635354c4SMaxim Konovalov m_free(m); 958635354c4SMaxim Konovalov break; 959635354c4SMaxim Konovalov } 9608501a69cSRobert Watson INP_WLOCK(inp); 961993d9505SRobert Watson error = ip_pcbopts(inp, sopt->sopt_name, m); 9628501a69cSRobert Watson INP_WUNLOCK(inp); 963993d9505SRobert Watson return (error); 964cfe8b629SGarrett Wollman } 965df8bae1dSRodney W. Grimes 966f44270e7SPawel Jakub Dawidek case IP_BINDANY: 967f44270e7SPawel Jakub Dawidek if (sopt->sopt_td != NULL) { 968f44270e7SPawel Jakub Dawidek error = priv_check(sopt->sopt_td, 969f44270e7SPawel Jakub Dawidek PRIV_NETINET_BINDANY); 970f44270e7SPawel Jakub Dawidek if (error) 971be9347e3SAdrian Chadd break; 972be9347e3SAdrian Chadd } 973cef27294SAdrian Chadd /* FALLTHROUGH */ 974df8bae1dSRodney W. Grimes case IP_TOS: 975df8bae1dSRodney W. Grimes case IP_TTL: 976936cd18dSAndre Oppermann case IP_MINTTL: 977df8bae1dSRodney W. Grimes case IP_RECVOPTS: 978df8bae1dSRodney W. Grimes case IP_RECVRETOPTS: 979df8bae1dSRodney W. Grimes case IP_RECVDSTADDR: 9804957466bSMatthew N. Dodd case IP_RECVTTL: 98182c23ebaSBill Fenner case IP_RECVIF: 9826a800098SYoshinobu Inoue case IP_FAITH: 9838afa2304SBruce M Simpson case IP_ONESBCAST: 984b2828ad2SAndre Oppermann case IP_DONTFRAG: 9853cca425bSMichael Tuexen case IP_RECVTOS: 986cfe8b629SGarrett Wollman error = sooptcopyin(sopt, &optval, sizeof optval, 987cfe8b629SGarrett Wollman sizeof optval); 988cfe8b629SGarrett Wollman if (error) 989cfe8b629SGarrett Wollman break; 990df8bae1dSRodney W. Grimes 991cfe8b629SGarrett Wollman switch (sopt->sopt_name) { 992df8bae1dSRodney W. Grimes case IP_TOS: 993ca98b82cSDavid Greenman inp->inp_ip_tos = optval; 994df8bae1dSRodney W. Grimes break; 995df8bae1dSRodney W. Grimes 996df8bae1dSRodney W. Grimes case IP_TTL: 997ca98b82cSDavid Greenman inp->inp_ip_ttl = optval; 998df8bae1dSRodney W. Grimes break; 999936cd18dSAndre Oppermann 1000936cd18dSAndre Oppermann case IP_MINTTL: 1001a603c811SRobert Watson if (optval >= 0 && optval <= MAXTTL) 1002936cd18dSAndre Oppermann inp->inp_ip_minttl = optval; 1003936cd18dSAndre Oppermann else 1004936cd18dSAndre Oppermann error = EINVAL; 1005936cd18dSAndre Oppermann break; 1006936cd18dSAndre Oppermann 1007a138d217SRobert Watson #define OPTSET(bit) do { \ 10088501a69cSRobert Watson INP_WLOCK(inp); \ 1009df8bae1dSRodney W. Grimes if (optval) \ 1010df8bae1dSRodney W. Grimes inp->inp_flags |= bit; \ 1011df8bae1dSRodney W. Grimes else \ 1012a138d217SRobert Watson inp->inp_flags &= ~bit; \ 10138501a69cSRobert Watson INP_WUNLOCK(inp); \ 1014a138d217SRobert Watson } while (0) 1015df8bae1dSRodney W. Grimes 1016df8bae1dSRodney W. Grimes case IP_RECVOPTS: 1017df8bae1dSRodney W. Grimes OPTSET(INP_RECVOPTS); 1018df8bae1dSRodney W. Grimes break; 1019df8bae1dSRodney W. Grimes 1020df8bae1dSRodney W. Grimes case IP_RECVRETOPTS: 1021df8bae1dSRodney W. Grimes OPTSET(INP_RECVRETOPTS); 1022df8bae1dSRodney W. Grimes break; 1023df8bae1dSRodney W. Grimes 1024df8bae1dSRodney W. Grimes case IP_RECVDSTADDR: 1025df8bae1dSRodney W. Grimes OPTSET(INP_RECVDSTADDR); 1026df8bae1dSRodney W. Grimes break; 102782c23ebaSBill Fenner 10284957466bSMatthew N. Dodd case IP_RECVTTL: 10294957466bSMatthew N. Dodd OPTSET(INP_RECVTTL); 10304957466bSMatthew N. Dodd break; 10314957466bSMatthew N. Dodd 103282c23ebaSBill Fenner case IP_RECVIF: 103382c23ebaSBill Fenner OPTSET(INP_RECVIF); 103482c23ebaSBill Fenner break; 10356a800098SYoshinobu Inoue 10366a800098SYoshinobu Inoue case IP_FAITH: 10376a800098SYoshinobu Inoue OPTSET(INP_FAITH); 10386a800098SYoshinobu Inoue break; 10398afa2304SBruce M Simpson 10408afa2304SBruce M Simpson case IP_ONESBCAST: 10418afa2304SBruce M Simpson OPTSET(INP_ONESBCAST); 10428afa2304SBruce M Simpson break; 1043b2828ad2SAndre Oppermann case IP_DONTFRAG: 1044b2828ad2SAndre Oppermann OPTSET(INP_DONTFRAG); 1045b2828ad2SAndre Oppermann break; 1046f44270e7SPawel Jakub Dawidek case IP_BINDANY: 1047f44270e7SPawel Jakub Dawidek OPTSET(INP_BINDANY); 1048be9347e3SAdrian Chadd break; 10493cca425bSMichael Tuexen case IP_RECVTOS: 10503cca425bSMichael Tuexen OPTSET(INP_RECVTOS); 10513cca425bSMichael Tuexen break; 1052df8bae1dSRodney W. Grimes } 1053df8bae1dSRodney W. Grimes break; 1054df8bae1dSRodney W. Grimes #undef OPTSET 1055df8bae1dSRodney W. Grimes 105671498f30SBruce M Simpson /* 105771498f30SBruce M Simpson * Multicast socket options are processed by the in_mcast 105871498f30SBruce M Simpson * module. 105971498f30SBruce M Simpson */ 1060df8bae1dSRodney W. Grimes case IP_MULTICAST_IF: 1061f0068c4aSGarrett Wollman case IP_MULTICAST_VIF: 1062df8bae1dSRodney W. Grimes case IP_MULTICAST_TTL: 1063df8bae1dSRodney W. Grimes case IP_MULTICAST_LOOP: 1064df8bae1dSRodney W. Grimes case IP_ADD_MEMBERSHIP: 1065df8bae1dSRodney W. Grimes case IP_DROP_MEMBERSHIP: 106671498f30SBruce M Simpson case IP_ADD_SOURCE_MEMBERSHIP: 106771498f30SBruce M Simpson case IP_DROP_SOURCE_MEMBERSHIP: 106871498f30SBruce M Simpson case IP_BLOCK_SOURCE: 106971498f30SBruce M Simpson case IP_UNBLOCK_SOURCE: 107071498f30SBruce M Simpson case IP_MSFILTER: 107171498f30SBruce M Simpson case MCAST_JOIN_GROUP: 107271498f30SBruce M Simpson case MCAST_LEAVE_GROUP: 107371498f30SBruce M Simpson case MCAST_JOIN_SOURCE_GROUP: 107471498f30SBruce M Simpson case MCAST_LEAVE_SOURCE_GROUP: 107571498f30SBruce M Simpson case MCAST_BLOCK_SOURCE: 107671498f30SBruce M Simpson case MCAST_UNBLOCK_SOURCE: 107771498f30SBruce M Simpson error = inp_setmoptions(inp, sopt); 1078df8bae1dSRodney W. Grimes break; 1079df8bae1dSRodney W. Grimes 108033b3ac06SPeter Wemm case IP_PORTRANGE: 1081cfe8b629SGarrett Wollman error = sooptcopyin(sopt, &optval, sizeof optval, 1082cfe8b629SGarrett Wollman sizeof optval); 1083cfe8b629SGarrett Wollman if (error) 1084cfe8b629SGarrett Wollman break; 108533b3ac06SPeter Wemm 10868501a69cSRobert Watson INP_WLOCK(inp); 108733b3ac06SPeter Wemm switch (optval) { 108833b3ac06SPeter Wemm case IP_PORTRANGE_DEFAULT: 108933b3ac06SPeter Wemm inp->inp_flags &= ~(INP_LOWPORT); 109033b3ac06SPeter Wemm inp->inp_flags &= ~(INP_HIGHPORT); 109133b3ac06SPeter Wemm break; 109233b3ac06SPeter Wemm 109333b3ac06SPeter Wemm case IP_PORTRANGE_HIGH: 109433b3ac06SPeter Wemm inp->inp_flags &= ~(INP_LOWPORT); 109533b3ac06SPeter Wemm inp->inp_flags |= INP_HIGHPORT; 109633b3ac06SPeter Wemm break; 109733b3ac06SPeter Wemm 109833b3ac06SPeter Wemm case IP_PORTRANGE_LOW: 109933b3ac06SPeter Wemm inp->inp_flags &= ~(INP_HIGHPORT); 110033b3ac06SPeter Wemm inp->inp_flags |= INP_LOWPORT; 110133b3ac06SPeter Wemm break; 110233b3ac06SPeter Wemm 110333b3ac06SPeter Wemm default: 110433b3ac06SPeter Wemm error = EINVAL; 110533b3ac06SPeter Wemm break; 110633b3ac06SPeter Wemm } 11078501a69cSRobert Watson INP_WUNLOCK(inp); 1108ce8c72b1SPeter Wemm break; 110933b3ac06SPeter Wemm 1110b2630c29SGeorge V. Neville-Neil #ifdef IPSEC 11116a800098SYoshinobu Inoue case IP_IPSEC_POLICY: 11126a800098SYoshinobu Inoue { 11136a800098SYoshinobu Inoue caddr_t req; 11146a800098SYoshinobu Inoue struct mbuf *m; 11156a800098SYoshinobu Inoue 11166a800098SYoshinobu Inoue if ((error = soopt_getm(sopt, &m)) != 0) /* XXX */ 11176a800098SYoshinobu Inoue break; 11186a800098SYoshinobu Inoue if ((error = soopt_mcopyin(sopt, m)) != 0) /* XXX */ 11196a800098SYoshinobu Inoue break; 11206a800098SYoshinobu Inoue req = mtod(m, caddr_t); 112197aa4a51SBjoern A. Zeeb error = ipsec_set_policy(inp, sopt->sopt_name, req, 1122c26fe973SBjoern A. Zeeb m->m_len, (sopt->sopt_td != NULL) ? 1123c26fe973SBjoern A. Zeeb sopt->sopt_td->td_ucred : NULL); 11246a800098SYoshinobu Inoue m_freem(m); 11256a800098SYoshinobu Inoue break; 11266a800098SYoshinobu Inoue } 1127b2630c29SGeorge V. Neville-Neil #endif /* IPSEC */ 11286a800098SYoshinobu Inoue 1129df8bae1dSRodney W. Grimes default: 1130df8bae1dSRodney W. Grimes error = ENOPROTOOPT; 1131df8bae1dSRodney W. Grimes break; 1132df8bae1dSRodney W. Grimes } 1133df8bae1dSRodney W. Grimes break; 1134df8bae1dSRodney W. Grimes 1135cfe8b629SGarrett Wollman case SOPT_GET: 1136cfe8b629SGarrett Wollman switch (sopt->sopt_name) { 1137df8bae1dSRodney W. Grimes case IP_OPTIONS: 1138df8bae1dSRodney W. Grimes case IP_RETOPTS: 1139cfe8b629SGarrett Wollman if (inp->inp_options) 1140cfe8b629SGarrett Wollman error = sooptcopyout(sopt, 1141cfe8b629SGarrett Wollman mtod(inp->inp_options, 1142cfe8b629SGarrett Wollman char *), 1143cfe8b629SGarrett Wollman inp->inp_options->m_len); 1144cfe8b629SGarrett Wollman else 1145cfe8b629SGarrett Wollman sopt->sopt_valsize = 0; 1146df8bae1dSRodney W. Grimes break; 1147df8bae1dSRodney W. Grimes 1148df8bae1dSRodney W. Grimes case IP_TOS: 1149df8bae1dSRodney W. Grimes case IP_TTL: 1150936cd18dSAndre Oppermann case IP_MINTTL: 1151df8bae1dSRodney W. Grimes case IP_RECVOPTS: 1152df8bae1dSRodney W. Grimes case IP_RECVRETOPTS: 1153df8bae1dSRodney W. Grimes case IP_RECVDSTADDR: 11544957466bSMatthew N. Dodd case IP_RECVTTL: 115582c23ebaSBill Fenner case IP_RECVIF: 1156cfe8b629SGarrett Wollman case IP_PORTRANGE: 11576a800098SYoshinobu Inoue case IP_FAITH: 11588afa2304SBruce M Simpson case IP_ONESBCAST: 1159b2828ad2SAndre Oppermann case IP_DONTFRAG: 11605f6bf451SAttilio Rao case IP_BINDANY: 11613cca425bSMichael Tuexen case IP_RECVTOS: 1162cfe8b629SGarrett Wollman switch (sopt->sopt_name) { 1163df8bae1dSRodney W. Grimes 1164df8bae1dSRodney W. Grimes case IP_TOS: 1165ca98b82cSDavid Greenman optval = inp->inp_ip_tos; 1166df8bae1dSRodney W. Grimes break; 1167df8bae1dSRodney W. Grimes 1168df8bae1dSRodney W. Grimes case IP_TTL: 1169ca98b82cSDavid Greenman optval = inp->inp_ip_ttl; 1170df8bae1dSRodney W. Grimes break; 1171df8bae1dSRodney W. Grimes 1172936cd18dSAndre Oppermann case IP_MINTTL: 1173936cd18dSAndre Oppermann optval = inp->inp_ip_minttl; 1174936cd18dSAndre Oppermann break; 1175936cd18dSAndre Oppermann 1176df8bae1dSRodney W. Grimes #define OPTBIT(bit) (inp->inp_flags & bit ? 1 : 0) 1177df8bae1dSRodney W. Grimes 1178df8bae1dSRodney W. Grimes case IP_RECVOPTS: 1179df8bae1dSRodney W. Grimes optval = OPTBIT(INP_RECVOPTS); 1180df8bae1dSRodney W. Grimes break; 1181df8bae1dSRodney W. Grimes 1182df8bae1dSRodney W. Grimes case IP_RECVRETOPTS: 1183df8bae1dSRodney W. Grimes optval = OPTBIT(INP_RECVRETOPTS); 1184df8bae1dSRodney W. Grimes break; 1185df8bae1dSRodney W. Grimes 1186df8bae1dSRodney W. Grimes case IP_RECVDSTADDR: 1187df8bae1dSRodney W. Grimes optval = OPTBIT(INP_RECVDSTADDR); 1188df8bae1dSRodney W. Grimes break; 118982c23ebaSBill Fenner 11904957466bSMatthew N. Dodd case IP_RECVTTL: 11914957466bSMatthew N. Dodd optval = OPTBIT(INP_RECVTTL); 11924957466bSMatthew N. Dodd break; 11934957466bSMatthew N. Dodd 119482c23ebaSBill Fenner case IP_RECVIF: 119582c23ebaSBill Fenner optval = OPTBIT(INP_RECVIF); 119682c23ebaSBill Fenner break; 1197cfe8b629SGarrett Wollman 1198cfe8b629SGarrett Wollman case IP_PORTRANGE: 1199cfe8b629SGarrett Wollman if (inp->inp_flags & INP_HIGHPORT) 1200cfe8b629SGarrett Wollman optval = IP_PORTRANGE_HIGH; 1201cfe8b629SGarrett Wollman else if (inp->inp_flags & INP_LOWPORT) 1202cfe8b629SGarrett Wollman optval = IP_PORTRANGE_LOW; 1203cfe8b629SGarrett Wollman else 1204cfe8b629SGarrett Wollman optval = 0; 1205cfe8b629SGarrett Wollman break; 12066a800098SYoshinobu Inoue 12076a800098SYoshinobu Inoue case IP_FAITH: 12086a800098SYoshinobu Inoue optval = OPTBIT(INP_FAITH); 12096a800098SYoshinobu Inoue break; 12108afa2304SBruce M Simpson 12118afa2304SBruce M Simpson case IP_ONESBCAST: 12128afa2304SBruce M Simpson optval = OPTBIT(INP_ONESBCAST); 12138afa2304SBruce M Simpson break; 1214b2828ad2SAndre Oppermann case IP_DONTFRAG: 1215b2828ad2SAndre Oppermann optval = OPTBIT(INP_DONTFRAG); 1216b2828ad2SAndre Oppermann break; 12175f6bf451SAttilio Rao case IP_BINDANY: 12185f6bf451SAttilio Rao optval = OPTBIT(INP_BINDANY); 12195f6bf451SAttilio Rao break; 12203cca425bSMichael Tuexen case IP_RECVTOS: 12213cca425bSMichael Tuexen optval = OPTBIT(INP_RECVTOS); 12223cca425bSMichael Tuexen break; 1223df8bae1dSRodney W. Grimes } 1224cfe8b629SGarrett Wollman error = sooptcopyout(sopt, &optval, sizeof optval); 1225df8bae1dSRodney W. Grimes break; 1226df8bae1dSRodney W. Grimes 122771498f30SBruce M Simpson /* 122871498f30SBruce M Simpson * Multicast socket options are processed by the in_mcast 122971498f30SBruce M Simpson * module. 123071498f30SBruce M Simpson */ 1231df8bae1dSRodney W. Grimes case IP_MULTICAST_IF: 1232f0068c4aSGarrett Wollman case IP_MULTICAST_VIF: 1233df8bae1dSRodney W. Grimes case IP_MULTICAST_TTL: 1234df8bae1dSRodney W. Grimes case IP_MULTICAST_LOOP: 123571498f30SBruce M Simpson case IP_MSFILTER: 123671498f30SBruce M Simpson error = inp_getmoptions(inp, sopt); 123733b3ac06SPeter Wemm break; 123833b3ac06SPeter Wemm 1239b2630c29SGeorge V. Neville-Neil #ifdef IPSEC 12406a800098SYoshinobu Inoue case IP_IPSEC_POLICY: 12416a800098SYoshinobu Inoue { 1242f63e7634SYoshinobu Inoue struct mbuf *m = NULL; 12436a800098SYoshinobu Inoue caddr_t req = NULL; 1244686cdd19SJun-ichiro itojun Hagino size_t len = 0; 12456a800098SYoshinobu Inoue 1246686cdd19SJun-ichiro itojun Hagino if (m != 0) { 12476a800098SYoshinobu Inoue req = mtod(m, caddr_t); 1248686cdd19SJun-ichiro itojun Hagino len = m->m_len; 1249686cdd19SJun-ichiro itojun Hagino } 125097aa4a51SBjoern A. Zeeb error = ipsec_get_policy(sotoinpcb(so), req, len, &m); 12516a800098SYoshinobu Inoue if (error == 0) 12526a800098SYoshinobu Inoue error = soopt_mcopyout(sopt, m); /* XXX */ 1253f63e7634SYoshinobu Inoue if (error == 0) 12546a800098SYoshinobu Inoue m_freem(m); 12556a800098SYoshinobu Inoue break; 12566a800098SYoshinobu Inoue } 1257b2630c29SGeorge V. Neville-Neil #endif /* IPSEC */ 12586a800098SYoshinobu Inoue 1259df8bae1dSRodney W. Grimes default: 1260df8bae1dSRodney W. Grimes error = ENOPROTOOPT; 1261df8bae1dSRodney W. Grimes break; 1262df8bae1dSRodney W. Grimes } 1263df8bae1dSRodney W. Grimes break; 1264df8bae1dSRodney W. Grimes } 1265df8bae1dSRodney W. Grimes return (error); 1266df8bae1dSRodney W. Grimes } 1267df8bae1dSRodney W. Grimes 1268df8bae1dSRodney W. Grimes /* 1269df8bae1dSRodney W. Grimes * Routine called from ip_output() to loop back a copy of an IP multicast 1270df8bae1dSRodney W. Grimes * packet to the input queue of a specified interface. Note that this 1271df8bae1dSRodney W. Grimes * calls the output routine of the loopback "driver", but with an interface 1272f5fea3ddSPaul Traina * pointer that might NOT be a loopback interface -- evil, but easier than 1273f5fea3ddSPaul Traina * replicating that code here. 1274df8bae1dSRodney W. Grimes */ 1275df8bae1dSRodney W. Grimes static void 1276f2565d68SRobert Watson ip_mloopback(struct ifnet *ifp, struct mbuf *m, struct sockaddr_in *dst, 1277f2565d68SRobert Watson int hlen) 1278df8bae1dSRodney W. Grimes { 1279b375c9ecSLuigi Rizzo register struct ip *ip; 1280df8bae1dSRodney W. Grimes struct mbuf *copym; 1281df8bae1dSRodney W. Grimes 1282e4762f75SGeorge V. Neville-Neil /* 1283e4762f75SGeorge V. Neville-Neil * Make a deep copy of the packet because we're going to 1284e4762f75SGeorge V. Neville-Neil * modify the pack in order to generate checksums. 1285e4762f75SGeorge V. Neville-Neil */ 1286eb1b1807SGleb Smirnoff copym = m_dup(m, M_NOWAIT); 128786b1d6d2SBill Fenner if (copym != NULL && (copym->m_flags & M_EXT || copym->m_len < hlen)) 128886b1d6d2SBill Fenner copym = m_pullup(copym, hlen); 1289df8bae1dSRodney W. Grimes if (copym != NULL) { 1290390cdc6aSRuslan Ermilov /* If needed, compute the checksum and mark it as valid. */ 1291390cdc6aSRuslan Ermilov if (copym->m_pkthdr.csum_flags & CSUM_DELAY_DATA) { 1292390cdc6aSRuslan Ermilov in_delayed_cksum(copym); 1293390cdc6aSRuslan Ermilov copym->m_pkthdr.csum_flags &= ~CSUM_DELAY_DATA; 1294390cdc6aSRuslan Ermilov copym->m_pkthdr.csum_flags |= 1295390cdc6aSRuslan Ermilov CSUM_DATA_VALID | CSUM_PSEUDO_HDR; 1296390cdc6aSRuslan Ermilov copym->m_pkthdr.csum_data = 0xffff; 1297390cdc6aSRuslan Ermilov } 1298df8bae1dSRodney W. Grimes /* 1299df8bae1dSRodney W. Grimes * We don't bother to fragment if the IP length is greater 1300df8bae1dSRodney W. Grimes * than the interface's MTU. Can this possibly matter? 1301df8bae1dSRodney W. Grimes */ 13028f134647SGleb Smirnoff ip = mtod(copym, struct ip *); 1303df8bae1dSRodney W. Grimes ip->ip_sum = 0; 130486b1d6d2SBill Fenner ip->ip_sum = in_cksum(copym, hlen); 1305201c2527SJulian Elischer #if 1 /* XXX */ 1306201c2527SJulian Elischer if (dst->sin_family != AF_INET) { 13072b8a366cSJulian Elischer printf("ip_mloopback: bad address family %d\n", 1308201c2527SJulian Elischer dst->sin_family); 1309201c2527SJulian Elischer dst->sin_family = AF_INET; 1310201c2527SJulian Elischer } 1311201c2527SJulian Elischer #endif 131206a429a3SArchie Cobbs if_simloop(ifp, copym, dst->sin_family, 0); 1313df8bae1dSRodney W. Grimes } 1314df8bae1dSRodney W. Grimes } 1315