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 355d6d7e75SGleb Smirnoff #include "opt_inet.h" 361f7e052cSJulian Elischer #include "opt_ipfw.h" 376a800098SYoshinobu Inoue #include "opt_ipsec.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" 429c423972SAdrian Chadd #include "opt_rss.h" 43fbd1372aSJoerg Wunsch 44df8bae1dSRodney W. Grimes #include <sys/param.h> 4526f9a767SRodney W. Grimes #include <sys/systm.h> 46f0f6d643SLuigi Rizzo #include <sys/kernel.h> 47cc0a3c8cSAndrey V. Elsukov #include <sys/lock.h> 48df8bae1dSRodney W. Grimes #include <sys/malloc.h> 49df8bae1dSRodney W. Grimes #include <sys/mbuf.h> 50acd3428bSRobert Watson #include <sys/priv.h> 51c26fe973SBjoern A. Zeeb #include <sys/proc.h> 52df8bae1dSRodney W. Grimes #include <sys/protosw.h> 53cc0a3c8cSAndrey V. Elsukov #include <sys/rmlock.h> 5457f60867SMark Johnston #include <sys/sdt.h> 55df8bae1dSRodney W. Grimes #include <sys/socket.h> 56df8bae1dSRodney W. Grimes #include <sys/socketvar.h> 579d9edc56SMike Silbersack #include <sys/sysctl.h> 58c26fe973SBjoern A. Zeeb #include <sys/ucred.h> 59df8bae1dSRodney W. Grimes 60df8bae1dSRodney W. Grimes #include <net/if.h> 6176039bc8SGleb Smirnoff #include <net/if_var.h> 623ef5e21dSQing Li #include <net/if_llatbl.h> 639b932e9eSAndre Oppermann #include <net/netisr.h> 64c21fd232SAndre Oppermann #include <net/pfil.h> 65df8bae1dSRodney W. Grimes #include <net/route.h> 6665111ec7SKip Macy #include <net/flowtable.h> 67e440aed9SQing Li #ifdef RADIX_MPATH 68e440aed9SQing Li #include <net/radix_mpath.h> 69e440aed9SQing Li #endif 70b2bdc62aSAdrian Chadd #include <net/rss_config.h> 714b79449eSBjoern A. Zeeb #include <net/vnet.h> 72df8bae1dSRodney W. Grimes 73df8bae1dSRodney W. Grimes #include <netinet/in.h> 7457f60867SMark Johnston #include <netinet/in_kdtrace.h> 75df8bae1dSRodney W. Grimes #include <netinet/in_systm.h> 76df8bae1dSRodney W. Grimes #include <netinet/ip.h> 77df8bae1dSRodney W. Grimes #include <netinet/in_pcb.h> 789c423972SAdrian Chadd #include <netinet/in_rss.h> 79df8bae1dSRodney W. Grimes #include <netinet/in_var.h> 80df8bae1dSRodney W. Grimes #include <netinet/ip_var.h> 81ef39adf0SAndre Oppermann #include <netinet/ip_options.h> 822f4afd21SRandall Stewart #ifdef SCTP 832f4afd21SRandall Stewart #include <netinet/sctp.h> 842f4afd21SRandall Stewart #include <netinet/sctp_crc32.h> 852f4afd21SRandall Stewart #endif 86df8bae1dSRodney W. Grimes 87b2630c29SGeorge V. Neville-Neil #ifdef IPSEC 882cb64cb2SGeorge V. Neville-Neil #include <netinet/ip_ipsec.h> 891dfcf0d2SAndre Oppermann #include <netipsec/ipsec.h> 90b2630c29SGeorge V. Neville-Neil #endif /* IPSEC*/ 916a800098SYoshinobu Inoue 921dfcf0d2SAndre Oppermann #include <machine/in_cksum.h> 931dfcf0d2SAndre Oppermann 94aed55708SRobert Watson #include <security/mac/mac_framework.h> 95aed55708SRobert Watson 9653dcc544SMike Silbersack #ifdef MBUF_STRESS_TEST 9705b9d121SBjoern A. Zeeb static int mbuf_frag_size = 0; 989d9edc56SMike Silbersack SYSCTL_INT(_net_inet_ip, OID_AUTO, mbuf_frag_size, CTLFLAG_RW, 999d9edc56SMike Silbersack &mbuf_frag_size, 0, "Fragment outgoing mbufs to this size"); 1009d9edc56SMike Silbersack #endif 1019d9edc56SMike Silbersack 102331dff07SAlexander V. Chernikov static void ip_mloopback(struct ifnet *, const struct mbuf *, int); 103afed1b49SDarren Reed 104afed1b49SDarren Reed 1058b889dbbSBruce M Simpson extern int in_mcast_loop; 10693e0e116SJulian Elischer extern struct protosw inetsw[]; 10793e0e116SJulian Elischer 108d9f2a782SErmal Luçi static inline int 1098f980c01SMark Johnston ip_output_pfil(struct mbuf **mp, struct ifnet *ifp, struct inpcb *inp, 110d9f2a782SErmal Luçi struct sockaddr_in *dst, int *fibnum, int *error) 111d9f2a782SErmal Luçi { 112d9f2a782SErmal Luçi struct m_tag *fwd_tag = NULL; 1138f980c01SMark Johnston struct mbuf *m; 114d9f2a782SErmal Luçi struct in_addr odst; 115d9f2a782SErmal Luçi struct ip *ip; 116d9f2a782SErmal Luçi 1178f980c01SMark Johnston m = *mp; 118d9f2a782SErmal Luçi ip = mtod(m, struct ip *); 119d9f2a782SErmal Luçi 120d9f2a782SErmal Luçi /* Run through list of hooks for output packets. */ 121d9f2a782SErmal Luçi odst.s_addr = ip->ip_dst.s_addr; 1228f980c01SMark Johnston *error = pfil_run_hooks(&V_inet_pfil_hook, mp, ifp, PFIL_OUT, inp); 1238f980c01SMark Johnston m = *mp; 124d9f2a782SErmal Luçi if ((*error) != 0 || m == NULL) 125d9f2a782SErmal Luçi return 1; /* Finished */ 126d9f2a782SErmal Luçi 127d9f2a782SErmal Luçi ip = mtod(m, struct ip *); 128d9f2a782SErmal Luçi 129d9f2a782SErmal Luçi /* See if destination IP address was changed by packet filter. */ 130d9f2a782SErmal Luçi if (odst.s_addr != ip->ip_dst.s_addr) { 131d9f2a782SErmal Luçi m->m_flags |= M_SKIP_FIREWALL; 132d9f2a782SErmal Luçi /* If destination is now ourself drop to ip_input(). */ 133d9f2a782SErmal Luçi if (in_localip(ip->ip_dst)) { 134d9f2a782SErmal Luçi m->m_flags |= M_FASTFWD_OURS; 135d9f2a782SErmal Luçi if (m->m_pkthdr.rcvif == NULL) 136d9f2a782SErmal Luçi m->m_pkthdr.rcvif = V_loif; 137d9f2a782SErmal Luçi if (m->m_pkthdr.csum_flags & CSUM_DELAY_DATA) { 138d9f2a782SErmal Luçi m->m_pkthdr.csum_flags |= 139d9f2a782SErmal Luçi CSUM_DATA_VALID | CSUM_PSEUDO_HDR; 140d9f2a782SErmal Luçi m->m_pkthdr.csum_data = 0xffff; 141d9f2a782SErmal Luçi } 142d9f2a782SErmal Luçi m->m_pkthdr.csum_flags |= 143d9f2a782SErmal Luçi CSUM_IP_CHECKED | CSUM_IP_VALID; 144d9f2a782SErmal Luçi #ifdef SCTP 145d9f2a782SErmal Luçi if (m->m_pkthdr.csum_flags & CSUM_SCTP) 146d9f2a782SErmal Luçi m->m_pkthdr.csum_flags |= CSUM_SCTP_VALID; 147d9f2a782SErmal Luçi #endif 148d9f2a782SErmal Luçi *error = netisr_queue(NETISR_IP, m); 149d9f2a782SErmal Luçi return 1; /* Finished */ 150d9f2a782SErmal Luçi } 151d9f2a782SErmal Luçi 152d9f2a782SErmal Luçi bzero(dst, sizeof(*dst)); 153d9f2a782SErmal Luçi dst->sin_family = AF_INET; 154d9f2a782SErmal Luçi dst->sin_len = sizeof(*dst); 155d9f2a782SErmal Luçi dst->sin_addr = ip->ip_dst; 156d9f2a782SErmal Luçi 157d9f2a782SErmal Luçi return -1; /* Reloop */ 158d9f2a782SErmal Luçi } 159d9f2a782SErmal Luçi /* See if fib was changed by packet filter. */ 160d9f2a782SErmal Luçi if ((*fibnum) != M_GETFIB(m)) { 161d9f2a782SErmal Luçi m->m_flags |= M_SKIP_FIREWALL; 162d9f2a782SErmal Luçi *fibnum = M_GETFIB(m); 163d9f2a782SErmal Luçi return -1; /* Reloop for FIB change */ 164d9f2a782SErmal Luçi } 165d9f2a782SErmal Luçi 166d9f2a782SErmal Luçi /* See if local, if yes, send it to netisr with IP_FASTFWD_OURS. */ 167d9f2a782SErmal Luçi if (m->m_flags & M_FASTFWD_OURS) { 168d9f2a782SErmal Luçi if (m->m_pkthdr.rcvif == NULL) 169d9f2a782SErmal Luçi m->m_pkthdr.rcvif = V_loif; 170d9f2a782SErmal Luçi if (m->m_pkthdr.csum_flags & CSUM_DELAY_DATA) { 171d9f2a782SErmal Luçi m->m_pkthdr.csum_flags |= 172d9f2a782SErmal Luçi CSUM_DATA_VALID | CSUM_PSEUDO_HDR; 173d9f2a782SErmal Luçi m->m_pkthdr.csum_data = 0xffff; 174d9f2a782SErmal Luçi } 175d9f2a782SErmal Luçi #ifdef SCTP 176d9f2a782SErmal Luçi if (m->m_pkthdr.csum_flags & CSUM_SCTP) 177d9f2a782SErmal Luçi m->m_pkthdr.csum_flags |= CSUM_SCTP_VALID; 178d9f2a782SErmal Luçi #endif 179d9f2a782SErmal Luçi m->m_pkthdr.csum_flags |= 180d9f2a782SErmal Luçi CSUM_IP_CHECKED | CSUM_IP_VALID; 181d9f2a782SErmal Luçi 182d9f2a782SErmal Luçi *error = netisr_queue(NETISR_IP, m); 183d9f2a782SErmal Luçi return 1; /* Finished */ 184d9f2a782SErmal Luçi } 185d9f2a782SErmal Luçi /* Or forward to some other address? */ 186d9f2a782SErmal Luçi if ((m->m_flags & M_IP_NEXTHOP) && 187d9f2a782SErmal Luçi ((fwd_tag = m_tag_find(m, PACKET_TAG_IPFORWARD, NULL)) != NULL)) { 188d9f2a782SErmal Luçi bcopy((fwd_tag+1), dst, sizeof(struct sockaddr_in)); 189d9f2a782SErmal Luçi m->m_flags |= M_SKIP_FIREWALL; 190d9f2a782SErmal Luçi m->m_flags &= ~M_IP_NEXTHOP; 191d9f2a782SErmal Luçi m_tag_delete(m, fwd_tag); 192d9f2a782SErmal Luçi 193d9f2a782SErmal Luçi return -1; /* Reloop for CHANGE of dst */ 194d9f2a782SErmal Luçi } 195d9f2a782SErmal Luçi 196d9f2a782SErmal Luçi return 0; 197d9f2a782SErmal Luçi } 198d9f2a782SErmal Luçi 199df8bae1dSRodney W. Grimes /* 200df8bae1dSRodney W. Grimes * IP output. The packet in mbuf chain m contains a skeletal IP 201df8bae1dSRodney W. Grimes * header (with len, off, ttl, proto, tos, src, dst). 202df8bae1dSRodney W. Grimes * The mbuf chain containing the packet will be freed. 203df8bae1dSRodney W. Grimes * The mbuf opt, if present, will not be freed. 204bf984051SGleb Smirnoff * If route ro is present and has ro_rt initialized, route lookup would be 205bf984051SGleb Smirnoff * skipped and ro->ro_rt would be used. If ro is present but ro->ro_rt is NULL, 206bf984051SGleb Smirnoff * then result of route lookup is stored in ro->ro_rt. 207bf984051SGleb Smirnoff * 2083de758d3SRobert Watson * In the IP forwarding case, the packet will arrive with options already 2093de758d3SRobert Watson * inserted, so must have a NULL opt pointer. 210df8bae1dSRodney W. Grimes */ 211df8bae1dSRodney W. Grimes int 212f2565d68SRobert Watson ip_output(struct mbuf *m, struct mbuf *opt, struct route *ro, int flags, 213f2565d68SRobert Watson struct ip_moptions *imo, struct inpcb *inp) 214df8bae1dSRodney W. Grimes { 215cc0a3c8cSAndrey V. Elsukov struct rm_priotracker in_ifa_tracker; 2161e78ac21SJeffrey Hsu struct ip *ip; 217fc74d005SBjoern A. Zeeb struct ifnet *ifp = NULL; /* keep compiler happy */ 218ac9d7e26SMax Laier struct mbuf *m0; 21923bf9953SPoul-Henning Kamp int hlen = sizeof (struct ip); 2203ae2ad08SJohn-Mark Gurney int mtu; 221ca8b83b0SLuigi Rizzo int error = 0; 222ca8b83b0SLuigi Rizzo struct sockaddr_in *dst; 2234c7a6059SGleb Smirnoff const struct sockaddr_in *gw; 2243c73180fSGleb Smirnoff struct in_ifaddr *ia; 22521d172a3SGleb Smirnoff int isbroadcast; 226078468edSGleb Smirnoff uint16_t ip_len, ip_off; 227e3f406b3SRuslan Ermilov struct route iproute; 228ec396e61SLuigi Rizzo struct rtentry *rte; /* cache for ro->ro_rt */ 2299c57a5b6SHiroki Sato uint32_t fibnum; 230fe82cbe8SGleb Smirnoff int have_ia_ref; 2311a499816SEdward Tomasz Napierala #ifdef IPSEC 2321a499816SEdward Tomasz Napierala int no_route_but_check_spd = 0; 2331a499816SEdward Tomasz Napierala #endif 234ac9d7e26SMax Laier M_ASSERTPKTHDR(m); 23536e8826fSMax Laier 236bc97ba51SJulian Elischer if (inp != NULL) { 237baa45840SRobert Watson INP_LOCK_ASSERT(inp); 23862e1ba83SRobert Watson M_SETFIB(m, inp->inp_inc.inc_fibnum); 239c2529042SHans Petter Selasky if ((flags & IP_NODEFAULTFLOWID) == 0) { 24080cb9f21SKip Macy m->m_pkthdr.flowid = inp->inp_flowid; 2419c423972SAdrian Chadd M_HASHTYPE_SET(m, inp->inp_flowtype); 24280cb9f21SKip Macy } 243bc97ba51SJulian Elischer } 24462e1ba83SRobert Watson 24562e1ba83SRobert Watson if (ro == NULL) { 24662e1ba83SRobert Watson ro = &iproute; 24762e1ba83SRobert Watson bzero(ro, sizeof (*ro)); 248bf984051SGleb Smirnoff } 24962e1ba83SRobert Watson 25053be8fcaSBjoern A. Zeeb #ifdef FLOWTABLE 2510ff96b4fSGleb Smirnoff if (ro->ro_rt == NULL) 2520ff96b4fSGleb Smirnoff (void )flowtable_lookup(AF_INET, m, ro); 25353be8fcaSBjoern A. Zeeb #endif 2542b25acc1SLuigi Rizzo 255df8bae1dSRodney W. Grimes if (opt) { 256ca8b83b0SLuigi Rizzo int len = 0; 257df8bae1dSRodney W. Grimes m = ip_insertoptions(m, opt, &len); 258cb7641e8SMaxim Konovalov if (len != 0) 259ca8b83b0SLuigi Rizzo hlen = len; /* ip->ip_hl is updated above */ 260df8bae1dSRodney W. Grimes } 261df8bae1dSRodney W. Grimes ip = mtod(m, struct ip *); 2628f134647SGleb Smirnoff ip_len = ntohs(ip->ip_len); 2638f134647SGleb Smirnoff ip_off = ntohs(ip->ip_off); 2643efc3014SJulian Elischer 265df8bae1dSRodney W. Grimes if ((flags & (IP_FORWARDING|IP_RAWOUTPUT)) == 0) { 26653be11f6SPoul-Henning Kamp ip->ip_v = IPVERSION; 26753be11f6SPoul-Henning Kamp ip->ip_hl = hlen >> 2; 2686d947416SGleb Smirnoff ip_fillid(ip); 26986425c62SRobert Watson IPSTAT_INC(ips_localout); 270df8bae1dSRodney W. Grimes } else { 271ca8b83b0SLuigi Rizzo /* Header already set, fetch hlen from there */ 27253be11f6SPoul-Henning Kamp hlen = ip->ip_hl << 2; 273df8bae1dSRodney W. Grimes } 2749c9137eaSGarrett Wollman 275054692a4SAlexander V. Chernikov /* 276054692a4SAlexander V. Chernikov * dst/gw handling: 277054692a4SAlexander V. Chernikov * 2783c065f2fSGleb Smirnoff * dst can be rewritten but always points to &ro->ro_dst. 2793c065f2fSGleb Smirnoff * gw is readonly but can point either to dst OR rt_gateway, 2803c065f2fSGleb Smirnoff * therefore we need restore gw if we're redoing lookup. 281054692a4SAlexander V. Chernikov */ 2824c7a6059SGleb Smirnoff gw = dst = (struct sockaddr_in *)&ro->ro_dst; 2839c57a5b6SHiroki Sato fibnum = (inp != NULL) ? inp->inp_inc.inc_fibnum : M_GETFIB(m); 284d9f2a782SErmal Luçi rte = ro->ro_rt; 285df8bae1dSRodney W. Grimes /* 2863c065f2fSGleb Smirnoff * The address family should also be checked in case of sharing 2873c065f2fSGleb Smirnoff * the cache with IPv6. 288df8bae1dSRodney W. Grimes */ 289d9f2a782SErmal Luçi if (rte == NULL || dst->sin_family != AF_INET) { 290a4a6e773SHajimu UMEMOTO bzero(dst, sizeof(*dst)); 291df8bae1dSRodney W. Grimes dst->sin_family = AF_INET; 292df8bae1dSRodney W. Grimes dst->sin_len = sizeof(*dst); 2939b932e9eSAndre Oppermann dst->sin_addr = ip->ip_dst; 294df8bae1dSRodney W. Grimes } 295d9f2a782SErmal Luçi again: 296d9f2a782SErmal Luçi ia = NULL; 297d9f2a782SErmal Luçi have_ia_ref = 0; 298df8bae1dSRodney W. Grimes /* 299a3fd02d8SBruce M Simpson * If routing to interface only, short circuit routing lookup. 300a3fd02d8SBruce M Simpson * The use of an all-ones broadcast address implies this; an 301a3fd02d8SBruce M Simpson * interface is specified by the broadcast address of an interface, 302a3fd02d8SBruce M Simpson * or the destination address of a ptp interface. 303df8bae1dSRodney W. Grimes */ 304a3fd02d8SBruce M Simpson if (flags & IP_SENDONES) { 3054f8585e0SAlan Somers if ((ia = ifatoia(ifa_ifwithbroadaddr(sintosa(dst), 30658a39d8cSAlan Somers M_GETFIB(m)))) == NULL && 3074f8585e0SAlan Somers (ia = ifatoia(ifa_ifwithdstaddr(sintosa(dst), 30858a39d8cSAlan Somers M_GETFIB(m)))) == NULL) { 30986425c62SRobert Watson IPSTAT_INC(ips_noroute); 310a3fd02d8SBruce M Simpson error = ENETUNREACH; 311a3fd02d8SBruce M Simpson goto bad; 312a3fd02d8SBruce M Simpson } 313fe82cbe8SGleb Smirnoff have_ia_ref = 1; 314a3fd02d8SBruce M Simpson ip->ip_dst.s_addr = INADDR_BROADCAST; 315a3fd02d8SBruce M Simpson dst->sin_addr = ip->ip_dst; 316a3fd02d8SBruce M Simpson ifp = ia->ia_ifp; 317a3fd02d8SBruce M Simpson ip->ip_ttl = 1; 318a3fd02d8SBruce M Simpson isbroadcast = 1; 319a3fd02d8SBruce M Simpson } else if (flags & IP_ROUTETOIF) { 3204f8585e0SAlan Somers if ((ia = ifatoia(ifa_ifwithdstaddr(sintosa(dst), 32158a39d8cSAlan Somers M_GETFIB(m)))) == NULL && 3224f8585e0SAlan Somers (ia = ifatoia(ifa_ifwithnet(sintosa(dst), 0, 32358a39d8cSAlan Somers M_GETFIB(m)))) == NULL) { 32486425c62SRobert Watson IPSTAT_INC(ips_noroute); 325df8bae1dSRodney W. Grimes error = ENETUNREACH; 326df8bae1dSRodney W. Grimes goto bad; 327df8bae1dSRodney W. Grimes } 328fe82cbe8SGleb Smirnoff have_ia_ref = 1; 329df8bae1dSRodney W. Grimes ifp = ia->ia_ifp; 330df8bae1dSRodney W. Grimes ip->ip_ttl = 1; 3319f9b3dc4SGarrett Wollman isbroadcast = in_broadcast(dst->sin_addr, ifp); 3323afefa39SDaniel C. Sobral } else if (IN_MULTICAST(ntohl(ip->ip_dst.s_addr)) && 33338c1bc35SRuslan Ermilov imo != NULL && imo->imo_multicast_ifp != NULL) { 3343afefa39SDaniel C. Sobral /* 33538c1bc35SRuslan Ermilov * Bypass the normal routing lookup for multicast 33638c1bc35SRuslan Ermilov * packets if the interface is specified. 3373afefa39SDaniel C. Sobral */ 33838c1bc35SRuslan Ermilov ifp = imo->imo_multicast_ifp; 339cc0a3c8cSAndrey V. Elsukov IFP_TO_IA(ifp, ia, &in_ifa_tracker); 340fe82cbe8SGleb Smirnoff if (ia) 341fe82cbe8SGleb Smirnoff have_ia_ref = 1; 34238c1bc35SRuslan Ermilov isbroadcast = 0; /* fool gcc */ 343df8bae1dSRodney W. Grimes } else { 3442c17fe93SGarrett Wollman /* 34597d8d152SAndre Oppermann * We want to do any cloning requested by the link layer, 34697d8d152SAndre Oppermann * as this is probably required in all cases for correct 34797d8d152SAndre Oppermann * operation (as it is for ARP). 3482c17fe93SGarrett Wollman */ 349ec396e61SLuigi Rizzo if (rte == NULL) { 350e440aed9SQing Li #ifdef RADIX_MPATH 3518b07e49aSJulian Elischer rtalloc_mpath_fib(ro, 3528b07e49aSJulian Elischer ntohl(ip->ip_src.s_addr ^ ip->ip_dst.s_addr), 3539c57a5b6SHiroki Sato fibnum); 354e440aed9SQing Li #else 3559c57a5b6SHiroki Sato in_rtalloc_ign(ro, 0, fibnum); 356e440aed9SQing Li #endif 357ec396e61SLuigi Rizzo rte = ro->ro_rt; 358ec396e61SLuigi Rizzo } 359c7ea0aa6SQing Li if (rte == NULL || 360d9f2a782SErmal Luçi (rte->rt_flags & RTF_UP) == 0 || 361c7ea0aa6SQing Li rte->rt_ifp == NULL || 362c7ea0aa6SQing Li !RT_LINK_IS_UP(rte->rt_ifp)) { 3631a499816SEdward Tomasz Napierala #ifdef IPSEC 3641a499816SEdward Tomasz Napierala /* 3651a499816SEdward Tomasz Napierala * There is no route for this packet, but it is 3661a499816SEdward Tomasz Napierala * possible that a matching SPD entry exists. 3671a499816SEdward Tomasz Napierala */ 3681a499816SEdward Tomasz Napierala no_route_but_check_spd = 1; 3691a499816SEdward Tomasz Napierala mtu = 0; /* Silence GCC warning. */ 3701a499816SEdward Tomasz Napierala goto sendit; 3711a499816SEdward Tomasz Napierala #endif 37286425c62SRobert Watson IPSTAT_INC(ips_noroute); 373df8bae1dSRodney W. Grimes error = EHOSTUNREACH; 374df8bae1dSRodney W. Grimes goto bad; 375df8bae1dSRodney W. Grimes } 376ec396e61SLuigi Rizzo ia = ifatoia(rte->rt_ifa); 377ec396e61SLuigi Rizzo ifp = rte->rt_ifp; 378e3a7aa6fSGleb Smirnoff counter_u64_add(rte->rt_pksent, 1); 379ec396e61SLuigi Rizzo if (rte->rt_flags & RTF_GATEWAY) 3804c7a6059SGleb Smirnoff gw = (struct sockaddr_in *)rte->rt_gateway; 381ec396e61SLuigi Rizzo if (rte->rt_flags & RTF_HOST) 382ec396e61SLuigi Rizzo isbroadcast = (rte->rt_flags & RTF_BROADCAST); 3839f9b3dc4SGarrett Wollman else 3844c7a6059SGleb Smirnoff isbroadcast = in_broadcast(gw->sin_addr, ifp); 385df8bae1dSRodney W. Grimes } 386d9f2a782SErmal Luçi 3873ae2ad08SJohn-Mark Gurney /* 3883ae2ad08SJohn-Mark Gurney * Calculate MTU. If we have a route that is up, use that, 3893ae2ad08SJohn-Mark Gurney * otherwise use the interface's MTU. 3903ae2ad08SJohn-Mark Gurney */ 3917f948f12SAlexander V. Chernikov if (rte != NULL && (rte->rt_flags & (RTF_UP|RTF_HOST))) 392e3a7aa6fSGleb Smirnoff mtu = rte->rt_mtu; 3937f948f12SAlexander V. Chernikov else 3943ae2ad08SJohn-Mark Gurney mtu = ifp->if_mtu; 395c744cde4SBjoern A. Zeeb /* Catch a possible divide by zero later. */ 396c744cde4SBjoern A. Zeeb KASSERT(mtu > 0, ("%s: mtu %d <= 0, rte=%p (rt_flags=0x%08x) ifp=%p", 397c744cde4SBjoern A. Zeeb __func__, mtu, rte, (rte != NULL) ? rte->rt_flags : 0, ifp)); 398d9f2a782SErmal Luçi 3999b932e9eSAndre Oppermann if (IN_MULTICAST(ntohl(ip->ip_dst.s_addr))) { 400df8bae1dSRodney W. Grimes m->m_flags |= M_MCAST; 401df8bae1dSRodney W. Grimes /* 402183e1c86SGleb Smirnoff * IP destination address is multicast. Make sure "gw" 403183e1c86SGleb Smirnoff * still points to the address in "ro". (It may have been 404183e1c86SGleb Smirnoff * changed to point to a gateway address, above.) 405183e1c86SGleb Smirnoff */ 406183e1c86SGleb Smirnoff gw = dst; 407183e1c86SGleb Smirnoff /* 408df8bae1dSRodney W. Grimes * See if the caller provided any multicast options 409df8bae1dSRodney W. Grimes */ 410df8bae1dSRodney W. Grimes if (imo != NULL) { 411df8bae1dSRodney W. Grimes ip->ip_ttl = imo->imo_multicast_ttl; 4121c5de19aSGarrett Wollman if (imo->imo_multicast_vif != -1) 4131c5de19aSGarrett Wollman ip->ip_src.s_addr = 414bbb4330bSLuigi Rizzo ip_mcast_src ? 415bbb4330bSLuigi Rizzo ip_mcast_src(imo->imo_multicast_vif) : 416bbb4330bSLuigi Rizzo INADDR_ANY; 417df8bae1dSRodney W. Grimes } else 418df8bae1dSRodney W. Grimes ip->ip_ttl = IP_DEFAULT_MULTICAST_TTL; 419df8bae1dSRodney W. Grimes /* 420df8bae1dSRodney W. Grimes * Confirm that the outgoing interface supports multicast. 421df8bae1dSRodney W. Grimes */ 4221c5de19aSGarrett Wollman if ((imo == NULL) || (imo->imo_multicast_vif == -1)) { 423df8bae1dSRodney W. Grimes if ((ifp->if_flags & IFF_MULTICAST) == 0) { 42486425c62SRobert Watson IPSTAT_INC(ips_noroute); 425df8bae1dSRodney W. Grimes error = ENETUNREACH; 426df8bae1dSRodney W. Grimes goto bad; 427df8bae1dSRodney W. Grimes } 4281c5de19aSGarrett Wollman } 429df8bae1dSRodney W. Grimes /* 430df8bae1dSRodney W. Grimes * If source address not specified yet, use address 431df8bae1dSRodney W. Grimes * of outgoing interface. 432df8bae1dSRodney W. Grimes */ 433df8bae1dSRodney W. Grimes if (ip->ip_src.s_addr == INADDR_ANY) { 43438c1bc35SRuslan Ermilov /* Interface may have no addresses. */ 43538c1bc35SRuslan Ermilov if (ia != NULL) 43638c1bc35SRuslan Ermilov ip->ip_src = IA_SIN(ia)->sin_addr; 437df8bae1dSRodney W. Grimes } 438df8bae1dSRodney W. Grimes 4398b889dbbSBruce M Simpson if ((imo == NULL && in_mcast_loop) || 4408b889dbbSBruce M Simpson (imo && imo->imo_multicast_loop)) { 441df8bae1dSRodney W. Grimes /* 4428b889dbbSBruce M Simpson * Loop back multicast datagram if not expressly 4438b889dbbSBruce M Simpson * forbidden to do so, even if we are not a member 4448b889dbbSBruce M Simpson * of the group; ip_input() will filter it later, 4458b889dbbSBruce M Simpson * thus deferring a hash lookup and mutex acquisition 4468b889dbbSBruce M Simpson * at the expense of a cheap copy using m_copym(). 447df8bae1dSRodney W. Grimes */ 448331dff07SAlexander V. Chernikov ip_mloopback(ifp, m, hlen); 4498b889dbbSBruce M Simpson } else { 450df8bae1dSRodney W. Grimes /* 451df8bae1dSRodney W. Grimes * If we are acting as a multicast router, perform 452df8bae1dSRodney W. Grimes * multicast forwarding as if the packet had just 453df8bae1dSRodney W. Grimes * arrived on the interface to which we are about 454df8bae1dSRodney W. Grimes * to send. The multicast forwarding function 455df8bae1dSRodney W. Grimes * recursively calls this function, using the 456df8bae1dSRodney W. Grimes * IP_FORWARDING flag to prevent infinite recursion. 457df8bae1dSRodney W. Grimes * 458df8bae1dSRodney W. Grimes * Multicasts that are looped back by ip_mloopback(), 459df8bae1dSRodney W. Grimes * above, will be forwarded by the ip_input() routine, 460df8bae1dSRodney W. Grimes * if necessary. 461df8bae1dSRodney W. Grimes */ 462603724d3SBjoern A. Zeeb if (V_ip_mrouter && (flags & IP_FORWARDING) == 0) { 463f0068c4aSGarrett Wollman /* 464bbb4330bSLuigi Rizzo * If rsvp daemon is not running, do not 465f0068c4aSGarrett Wollman * set ip_moptions. This ensures that the packet 466f0068c4aSGarrett Wollman * is multicast and not just sent down one link 467f0068c4aSGarrett Wollman * as prescribed by rsvpd. 468f0068c4aSGarrett Wollman */ 469603724d3SBjoern A. Zeeb if (!V_rsvp_on) 470f0068c4aSGarrett Wollman imo = NULL; 471bbb4330bSLuigi Rizzo if (ip_mforward && 472bbb4330bSLuigi Rizzo ip_mforward(ip, ifp, m, imo) != 0) { 473df8bae1dSRodney W. Grimes m_freem(m); 474df8bae1dSRodney W. Grimes goto done; 475df8bae1dSRodney W. Grimes } 476df8bae1dSRodney W. Grimes } 477df8bae1dSRodney W. Grimes } 4785e9ae478SGarrett Wollman 479df8bae1dSRodney W. Grimes /* 480df8bae1dSRodney W. Grimes * Multicasts with a time-to-live of zero may be looped- 481df8bae1dSRodney W. Grimes * back, above, but must not be transmitted on a network. 482df8bae1dSRodney W. Grimes * Also, multicasts addressed to the loopback interface 483df8bae1dSRodney W. Grimes * are not sent -- the above call to ip_mloopback() will 4848b889dbbSBruce M Simpson * loop back a copy. ip_input() will drop the copy if 4858b889dbbSBruce M Simpson * this host does not belong to the destination group on 4868b889dbbSBruce M Simpson * the loopback interface. 487df8bae1dSRodney W. Grimes */ 488f5fea3ddSPaul Traina if (ip->ip_ttl == 0 || ifp->if_flags & IFF_LOOPBACK) { 489df8bae1dSRodney W. Grimes m_freem(m); 490df8bae1dSRodney W. Grimes goto done; 491df8bae1dSRodney W. Grimes } 492df8bae1dSRodney W. Grimes 493df8bae1dSRodney W. Grimes goto sendit; 494df8bae1dSRodney W. Grimes } 4956a7c943cSAndre Oppermann 496df8bae1dSRodney W. Grimes /* 4972b25acc1SLuigi Rizzo * If the source address is not specified yet, use the address 498cb459254SJohn-Mark Gurney * of the outoing interface. 499df8bae1dSRodney W. Grimes */ 500f9e354dfSJulian Elischer if (ip->ip_src.s_addr == INADDR_ANY) { 50138c1bc35SRuslan Ermilov /* Interface may have no addresses. */ 50238c1bc35SRuslan Ermilov if (ia != NULL) { 503df8bae1dSRodney W. Grimes ip->ip_src = IA_SIN(ia)->sin_addr; 504f9e354dfSJulian Elischer } 50538c1bc35SRuslan Ermilov } 5066a7c943cSAndre Oppermann 507ca7a789aSMax Laier /* 508df8bae1dSRodney W. Grimes * Look for broadcast address and 5098c3f5566SRuslan Ermilov * verify user is allowed to send 510df8bae1dSRodney W. Grimes * such a packet. 511df8bae1dSRodney W. Grimes */ 5129f9b3dc4SGarrett Wollman if (isbroadcast) { 513df8bae1dSRodney W. Grimes if ((ifp->if_flags & IFF_BROADCAST) == 0) { 514df8bae1dSRodney W. Grimes error = EADDRNOTAVAIL; 515df8bae1dSRodney W. Grimes goto bad; 516df8bae1dSRodney W. Grimes } 517df8bae1dSRodney W. Grimes if ((flags & IP_ALLOWBROADCAST) == 0) { 518df8bae1dSRodney W. Grimes error = EACCES; 519df8bae1dSRodney W. Grimes goto bad; 520df8bae1dSRodney W. Grimes } 521df8bae1dSRodney W. Grimes /* don't allow broadcast messages to be fragmented */ 5228f134647SGleb Smirnoff if (ip_len > mtu) { 523df8bae1dSRodney W. Grimes error = EMSGSIZE; 524df8bae1dSRodney W. Grimes goto bad; 525df8bae1dSRodney W. Grimes } 526df8bae1dSRodney W. Grimes m->m_flags |= M_BCAST; 5279f9b3dc4SGarrett Wollman } else { 528df8bae1dSRodney W. Grimes m->m_flags &= ~M_BCAST; 5299f9b3dc4SGarrett Wollman } 530df8bae1dSRodney W. Grimes 531f1743588SDarren Reed sendit: 532b2630c29SGeorge V. Neville-Neil #ifdef IPSEC 5330275b2e3SAndrey V. Elsukov switch(ip_ipsec_output(&m, inp, &error)) { 5341dfcf0d2SAndre Oppermann case 1: 53533841545SHajimu UMEMOTO goto bad; 5361dfcf0d2SAndre Oppermann case -1: 5371dfcf0d2SAndre Oppermann goto done; 5381dfcf0d2SAndre Oppermann case 0: 53933841545SHajimu UMEMOTO default: 5401dfcf0d2SAndre Oppermann break; /* Continue with packet processing. */ 54133841545SHajimu UMEMOTO } 5421a499816SEdward Tomasz Napierala /* 5431a499816SEdward Tomasz Napierala * Check if there was a route for this packet; return error if not. 5441a499816SEdward Tomasz Napierala */ 5451a499816SEdward Tomasz Napierala if (no_route_but_check_spd) { 5461a499816SEdward Tomasz Napierala IPSTAT_INC(ips_noroute); 5471a499816SEdward Tomasz Napierala error = EHOSTUNREACH; 5481a499816SEdward Tomasz Napierala goto bad; 5491a499816SEdward Tomasz Napierala } 5501dfcf0d2SAndre Oppermann /* Update variables that are affected by ipsec4_output(). */ 55133841545SHajimu UMEMOTO ip = mtod(m, struct ip *); 55233841545SHajimu UMEMOTO hlen = ip->ip_hl << 2; 553b2630c29SGeorge V. Neville-Neil #endif /* IPSEC */ 55433841545SHajimu UMEMOTO 555c21fd232SAndre Oppermann /* Jump over all PFIL processing if hooks are not active. */ 556d9f2a782SErmal Luçi if (PFIL_HOOKED(&V_inet_pfil_hook)) { 5578f980c01SMark Johnston switch (ip_output_pfil(&m, ifp, inp, dst, &fibnum, &error)) { 558d9f2a782SErmal Luçi case 1: /* Finished */ 559c4ac87eaSDarren Reed goto done; 5609b932e9eSAndre Oppermann 561d9f2a782SErmal Luçi case 0: /* Continue normally */ 562c4ac87eaSDarren Reed ip = mtod(m, struct ip *); 563d9f2a782SErmal Luçi break; 564fed1c7e9SSøren Schmidt 565d9f2a782SErmal Luçi case -1: /* Need to try again */ 566d9f2a782SErmal Luçi /* Reset everything for a new round */ 5679c57a5b6SHiroki Sato RO_RTFREE(ro); 568fe82cbe8SGleb Smirnoff if (have_ia_ref) 569fe82cbe8SGleb Smirnoff ifa_free(&ia->ia_ifa); 570*4fb3a820SAlexander V. Chernikov ro->ro_prepend = NULL; 571d9f2a782SErmal Luçi rte = NULL; 572d9f2a782SErmal Luçi gw = dst; 573d9f2a782SErmal Luçi ip = mtod(m, struct ip *); 5749b932e9eSAndre Oppermann goto again; 575d9f2a782SErmal Luçi 576d9f2a782SErmal Luçi } 577099dd043SAndre Oppermann } 5782b25acc1SLuigi Rizzo 57951c8ec4aSRuslan Ermilov /* 127/8 must not appear on wire - RFC1122. */ 58051c8ec4aSRuslan Ermilov if ((ntohl(ip->ip_dst.s_addr) >> IN_CLASSA_NSHIFT) == IN_LOOPBACKNET || 58151c8ec4aSRuslan Ermilov (ntohl(ip->ip_src.s_addr) >> IN_CLASSA_NSHIFT) == IN_LOOPBACKNET) { 58251c8ec4aSRuslan Ermilov if ((ifp->if_flags & IFF_LOOPBACK) == 0) { 58386425c62SRobert Watson IPSTAT_INC(ips_badaddr); 58451c8ec4aSRuslan Ermilov error = EADDRNOTAVAIL; 58551c8ec4aSRuslan Ermilov goto bad; 58651c8ec4aSRuslan Ermilov } 58751c8ec4aSRuslan Ermilov } 58851c8ec4aSRuslan Ermilov 589206a3274SRuslan Ermilov m->m_pkthdr.csum_flags |= CSUM_IP; 590078468edSGleb Smirnoff if (m->m_pkthdr.csum_flags & CSUM_DELAY_DATA & ~ifp->if_hwassist) { 591db4f9cc7SJonathan Lemon in_delayed_cksum(m); 592078468edSGleb Smirnoff m->m_pkthdr.csum_flags &= ~CSUM_DELAY_DATA; 593db4f9cc7SJonathan Lemon } 5942f4afd21SRandall Stewart #ifdef SCTP 595078468edSGleb Smirnoff if (m->m_pkthdr.csum_flags & CSUM_SCTP & ~ifp->if_hwassist) { 5961966e5b5SRandall Stewart sctp_delayed_cksum(m, (uint32_t)(ip->ip_hl << 2)); 597078468edSGleb Smirnoff m->m_pkthdr.csum_flags &= ~CSUM_SCTP; 5982f4afd21SRandall Stewart } 5992f4afd21SRandall Stewart #endif 600db4f9cc7SJonathan Lemon 601e7319babSPoul-Henning Kamp /* 602db4f9cc7SJonathan Lemon * If small enough for interface, or the interface will take 603233dcce1SAndre Oppermann * care of the fragmentation for us, we can just send directly. 604df8bae1dSRodney W. Grimes */ 60521d172a3SGleb Smirnoff if (ip_len <= mtu || 606bf7dcda3SGleb Smirnoff (m->m_pkthdr.csum_flags & ifp->if_hwassist & CSUM_TSO) != 0) { 607df8bae1dSRodney W. Grimes ip->ip_sum = 0; 608078468edSGleb Smirnoff if (m->m_pkthdr.csum_flags & CSUM_IP & ~ifp->if_hwassist) { 609df8bae1dSRodney W. Grimes ip->ip_sum = in_cksum(m, hlen); 610078468edSGleb Smirnoff m->m_pkthdr.csum_flags &= ~CSUM_IP; 611078468edSGleb Smirnoff } 6125da9f8faSJosef Karthauser 613233dcce1SAndre Oppermann /* 614233dcce1SAndre Oppermann * Record statistics for this interface address. 615233dcce1SAndre Oppermann * With CSUM_TSO the byte/packet count will be slightly 616233dcce1SAndre Oppermann * incorrect because we count the IP+TCP headers only 617233dcce1SAndre Oppermann * once instead of for every generated packet. 618233dcce1SAndre Oppermann */ 61938c1bc35SRuslan Ermilov if (!(flags & IP_FORWARDING) && ia) { 620233dcce1SAndre Oppermann if (m->m_pkthdr.csum_flags & CSUM_TSO) 6217caf4ab7SGleb Smirnoff counter_u64_add(ia->ia_ifa.ifa_opackets, 6227caf4ab7SGleb Smirnoff m->m_pkthdr.len / m->m_pkthdr.tso_segsz); 623233dcce1SAndre Oppermann else 6247caf4ab7SGleb Smirnoff counter_u64_add(ia->ia_ifa.ifa_opackets, 1); 6257caf4ab7SGleb Smirnoff 6267caf4ab7SGleb Smirnoff counter_u64_add(ia->ia_ifa.ifa_obytes, m->m_pkthdr.len); 6275da9f8faSJosef Karthauser } 62853dcc544SMike Silbersack #ifdef MBUF_STRESS_TEST 6293390d476SMike Silbersack if (mbuf_frag_size && m->m_pkthdr.len > mbuf_frag_size) 630eb1b1807SGleb Smirnoff m = m_fragment(m, M_NOWAIT, mbuf_frag_size); 6319d9edc56SMike Silbersack #endif 632147f74d1SAndre Oppermann /* 633147f74d1SAndre Oppermann * Reset layer specific mbuf flags 634147f74d1SAndre Oppermann * to avoid confusing lower layers. 635147f74d1SAndre Oppermann */ 63686bd0491SAndre Oppermann m_clrprotoflags(m); 63757f60867SMark Johnston IP_PROBE(send, NULL, NULL, ip, ifp, ip, NULL); 63847e8d432SGleb Smirnoff error = (*ifp->if_output)(ifp, m, 63947e8d432SGleb Smirnoff (const struct sockaddr *)gw, ro); 640df8bae1dSRodney W. Grimes goto done; 641df8bae1dSRodney W. Grimes } 6421e78ac21SJeffrey Hsu 643233dcce1SAndre Oppermann /* Balk when DF bit is set or the interface didn't support TSO. */ 64421d172a3SGleb Smirnoff if ((ip_off & IP_DF) || (m->m_pkthdr.csum_flags & CSUM_TSO)) { 645df8bae1dSRodney W. Grimes error = EMSGSIZE; 64686425c62SRobert Watson IPSTAT_INC(ips_cantfrag); 647df8bae1dSRodney W. Grimes goto bad; 648df8bae1dSRodney W. Grimes } 6491e78ac21SJeffrey Hsu 6501e78ac21SJeffrey Hsu /* 6511e78ac21SJeffrey Hsu * Too large for interface; fragment if possible. If successful, 6521e78ac21SJeffrey Hsu * on return, m will point to a list of packets to be sent. 6531e78ac21SJeffrey Hsu */ 654078468edSGleb Smirnoff error = ip_fragment(ip, &m, mtu, ifp->if_hwassist); 6551e78ac21SJeffrey Hsu if (error) 656df8bae1dSRodney W. Grimes goto bad; 6571e78ac21SJeffrey Hsu for (; m; m = m0) { 658df8bae1dSRodney W. Grimes m0 = m->m_nextpkt; 659b375c9ecSLuigi Rizzo m->m_nextpkt = 0; 66007203494SDaniel C. Sobral if (error == 0) { 661fe937674SJosef Karthauser /* Record statistics for this interface address. */ 66207203494SDaniel C. Sobral if (ia != NULL) { 6637caf4ab7SGleb Smirnoff counter_u64_add(ia->ia_ifa.ifa_opackets, 1); 6647caf4ab7SGleb Smirnoff counter_u64_add(ia->ia_ifa.ifa_obytes, 6657caf4ab7SGleb Smirnoff m->m_pkthdr.len); 66607203494SDaniel C. Sobral } 667147f74d1SAndre Oppermann /* 668147f74d1SAndre Oppermann * Reset layer specific mbuf flags 669147f74d1SAndre Oppermann * to avoid confusing upper layers. 670147f74d1SAndre Oppermann */ 67186bd0491SAndre Oppermann m_clrprotoflags(m); 672fe937674SJosef Karthauser 67357f60867SMark Johnston IP_PROBE(send, NULL, NULL, ip, ifp, ip, NULL); 674df8bae1dSRodney W. Grimes error = (*ifp->if_output)(ifp, m, 67547e8d432SGleb Smirnoff (const struct sockaddr *)gw, ro); 676fe937674SJosef Karthauser } else 677df8bae1dSRodney W. Grimes m_freem(m); 678df8bae1dSRodney W. Grimes } 679df8bae1dSRodney W. Grimes 680df8bae1dSRodney W. Grimes if (error == 0) 68186425c62SRobert Watson IPSTAT_INC(ips_fragmented); 6821e78ac21SJeffrey Hsu 683df8bae1dSRodney W. Grimes done: 684bf984051SGleb Smirnoff if (ro == &iproute) 685bf984051SGleb Smirnoff RO_RTFREE(ro); 6863c402323SErmal Luçi else if (rte == NULL) 6873c402323SErmal Luçi /* 6883c402323SErmal Luçi * If the caller supplied a route but somehow the reference 6893c402323SErmal Luçi * to it has been released need to prevent the caller 6903c402323SErmal Luçi * calling RTFREE on it again. 6913c402323SErmal Luçi */ 6923c402323SErmal Luçi ro->ro_rt = NULL; 693fe82cbe8SGleb Smirnoff if (have_ia_ref) 694fe82cbe8SGleb Smirnoff ifa_free(&ia->ia_ifa); 695df8bae1dSRodney W. Grimes return (error); 696df8bae1dSRodney W. Grimes bad: 6973528d68fSBill Paul m_freem(m); 698df8bae1dSRodney W. Grimes goto done; 699df8bae1dSRodney W. Grimes } 700df8bae1dSRodney W. Grimes 7011e78ac21SJeffrey Hsu /* 7021e78ac21SJeffrey Hsu * Create a chain of fragments which fit the given mtu. m_frag points to the 7031e78ac21SJeffrey Hsu * mbuf to be fragmented; on return it points to the chain with the fragments. 7041e78ac21SJeffrey Hsu * Return 0 if no error. If error, m_frag may contain a partially built 7051e78ac21SJeffrey Hsu * chain of fragments that should be freed by the caller. 7061e78ac21SJeffrey Hsu * 7071e78ac21SJeffrey Hsu * if_hwassist_flags is the hw offload capabilities (see if_data.ifi_hwassist) 7081e78ac21SJeffrey Hsu */ 7091e78ac21SJeffrey Hsu int 7101e78ac21SJeffrey Hsu ip_fragment(struct ip *ip, struct mbuf **m_frag, int mtu, 711078468edSGleb Smirnoff u_long if_hwassist_flags) 7121e78ac21SJeffrey Hsu { 7131e78ac21SJeffrey Hsu int error = 0; 7141e78ac21SJeffrey Hsu int hlen = ip->ip_hl << 2; 7151e78ac21SJeffrey Hsu int len = (mtu - hlen) & ~7; /* size of payload in each fragment */ 7161e78ac21SJeffrey Hsu int off; 7171e78ac21SJeffrey Hsu struct mbuf *m0 = *m_frag; /* the original packet */ 7181e78ac21SJeffrey Hsu int firstlen; 7191e78ac21SJeffrey Hsu struct mbuf **mnext; 7201e78ac21SJeffrey Hsu int nfrags; 72121d172a3SGleb Smirnoff uint16_t ip_len, ip_off; 7221e78ac21SJeffrey Hsu 72321d172a3SGleb Smirnoff ip_len = ntohs(ip->ip_len); 72421d172a3SGleb Smirnoff ip_off = ntohs(ip->ip_off); 72521d172a3SGleb Smirnoff 72621d172a3SGleb Smirnoff if (ip_off & IP_DF) { /* Fragmentation not allowed */ 72786425c62SRobert Watson IPSTAT_INC(ips_cantfrag); 7281e78ac21SJeffrey Hsu return EMSGSIZE; 7291e78ac21SJeffrey Hsu } 7301e78ac21SJeffrey Hsu 7311e78ac21SJeffrey Hsu /* 7321e78ac21SJeffrey Hsu * Must be able to put at least 8 bytes per fragment. 7331e78ac21SJeffrey Hsu */ 7341e78ac21SJeffrey Hsu if (len < 8) 7351e78ac21SJeffrey Hsu return EMSGSIZE; 7361e78ac21SJeffrey Hsu 7371e78ac21SJeffrey Hsu /* 7381e78ac21SJeffrey Hsu * If the interface will not calculate checksums on 7391e78ac21SJeffrey Hsu * fragmented packets, then do it here. 7401e78ac21SJeffrey Hsu */ 741da2299c5SAndre Oppermann if (m0->m_pkthdr.csum_flags & CSUM_DELAY_DATA) { 7421e78ac21SJeffrey Hsu in_delayed_cksum(m0); 7431e78ac21SJeffrey Hsu m0->m_pkthdr.csum_flags &= ~CSUM_DELAY_DATA; 7441e78ac21SJeffrey Hsu } 7452f4afd21SRandall Stewart #ifdef SCTP 746da2299c5SAndre Oppermann if (m0->m_pkthdr.csum_flags & CSUM_SCTP) { 7471966e5b5SRandall Stewart sctp_delayed_cksum(m0, hlen); 7482f4afd21SRandall Stewart m0->m_pkthdr.csum_flags &= ~CSUM_SCTP; 7492f4afd21SRandall Stewart } 7502f4afd21SRandall Stewart #endif 7511e78ac21SJeffrey Hsu if (len > PAGE_SIZE) { 7521e78ac21SJeffrey Hsu /* 7531e78ac21SJeffrey Hsu * Fragment large datagrams such that each segment 7541e78ac21SJeffrey Hsu * contains a multiple of PAGE_SIZE amount of data, 7551e78ac21SJeffrey Hsu * plus headers. This enables a receiver to perform 7561e78ac21SJeffrey Hsu * page-flipping zero-copy optimizations. 7571e78ac21SJeffrey Hsu * 7581e78ac21SJeffrey Hsu * XXX When does this help given that sender and receiver 7591e78ac21SJeffrey Hsu * could have different page sizes, and also mtu could 7601e78ac21SJeffrey Hsu * be less than the receiver's page size ? 7611e78ac21SJeffrey Hsu */ 7621e78ac21SJeffrey Hsu int newlen; 7631e78ac21SJeffrey Hsu 7649c0f6aa7SHans Petter Selasky off = MIN(mtu, m0->m_pkthdr.len); 7651e78ac21SJeffrey Hsu 7661e78ac21SJeffrey Hsu /* 7671e78ac21SJeffrey Hsu * firstlen (off - hlen) must be aligned on an 7681e78ac21SJeffrey Hsu * 8-byte boundary 7691e78ac21SJeffrey Hsu */ 7701e78ac21SJeffrey Hsu if (off < hlen) 7711e78ac21SJeffrey Hsu goto smart_frag_failure; 7721e78ac21SJeffrey Hsu off = ((off - hlen) & ~7) + hlen; 7731e78ac21SJeffrey Hsu newlen = (~PAGE_MASK) & mtu; 7741e78ac21SJeffrey Hsu if ((newlen + sizeof (struct ip)) > mtu) { 7751e78ac21SJeffrey Hsu /* we failed, go back the default */ 7761e78ac21SJeffrey Hsu smart_frag_failure: 7771e78ac21SJeffrey Hsu newlen = len; 7781e78ac21SJeffrey Hsu off = hlen + len; 7791e78ac21SJeffrey Hsu } 7801e78ac21SJeffrey Hsu len = newlen; 7811e78ac21SJeffrey Hsu 7821e78ac21SJeffrey Hsu } else { 7831e78ac21SJeffrey Hsu off = hlen + len; 7841e78ac21SJeffrey Hsu } 7851e78ac21SJeffrey Hsu 7861e78ac21SJeffrey Hsu firstlen = off - hlen; 7871e78ac21SJeffrey Hsu mnext = &m0->m_nextpkt; /* pointer to next packet */ 7881e78ac21SJeffrey Hsu 7891e78ac21SJeffrey Hsu /* 7901e78ac21SJeffrey Hsu * Loop through length of segment after first fragment, 7911e78ac21SJeffrey Hsu * make new header and copy data of each part and link onto chain. 7921e78ac21SJeffrey Hsu * Here, m0 is the original packet, m is the fragment being created. 7931e78ac21SJeffrey Hsu * The fragments are linked off the m_nextpkt of the original 7941e78ac21SJeffrey Hsu * packet, which after processing serves as the first fragment. 7951e78ac21SJeffrey Hsu */ 79621d172a3SGleb Smirnoff for (nfrags = 1; off < ip_len; off += len, nfrags++) { 7971e78ac21SJeffrey Hsu struct ip *mhip; /* ip header on the fragment */ 7981e78ac21SJeffrey Hsu struct mbuf *m; 7991e78ac21SJeffrey Hsu int mhlen = sizeof (struct ip); 8001e78ac21SJeffrey Hsu 801dc4ad05eSGleb Smirnoff m = m_gethdr(M_NOWAIT, MT_DATA); 8020b17fba7SAndre Oppermann if (m == NULL) { 8031e78ac21SJeffrey Hsu error = ENOBUFS; 80486425c62SRobert Watson IPSTAT_INC(ips_odropped); 8051e78ac21SJeffrey Hsu goto done; 8061e78ac21SJeffrey Hsu } 807c4c4346fSHans Petter Selasky /* 808c4c4346fSHans Petter Selasky * Make sure the complete packet header gets copied 809c4c4346fSHans Petter Selasky * from the originating mbuf to the newly created 810c4c4346fSHans Petter Selasky * mbuf. This also ensures that existing firewall 811c4c4346fSHans Petter Selasky * classification(s), VLAN tags and so on get copied 812c4c4346fSHans Petter Selasky * to the resulting fragmented packet(s): 813c4c4346fSHans Petter Selasky */ 814c4c4346fSHans Petter Selasky if (m_dup_pkthdr(m, m0, M_NOWAIT) == 0) { 815c4c4346fSHans Petter Selasky m_free(m); 816c4c4346fSHans Petter Selasky error = ENOBUFS; 817c4c4346fSHans Petter Selasky IPSTAT_INC(ips_odropped); 818c4c4346fSHans Petter Selasky goto done; 819c4c4346fSHans Petter Selasky } 8201e78ac21SJeffrey Hsu /* 8211e78ac21SJeffrey Hsu * In the first mbuf, leave room for the link header, then 8221e78ac21SJeffrey Hsu * copy the original IP header including options. The payload 8238b889dbbSBruce M Simpson * goes into an additional mbuf chain returned by m_copym(). 8241e78ac21SJeffrey Hsu */ 8251e78ac21SJeffrey Hsu m->m_data += max_linkhdr; 8261e78ac21SJeffrey Hsu mhip = mtod(m, struct ip *); 8271e78ac21SJeffrey Hsu *mhip = *ip; 8281e78ac21SJeffrey Hsu if (hlen > sizeof (struct ip)) { 8291e78ac21SJeffrey Hsu mhlen = ip_optcopy(ip, mhip) + sizeof (struct ip); 8301e78ac21SJeffrey Hsu mhip->ip_v = IPVERSION; 8311e78ac21SJeffrey Hsu mhip->ip_hl = mhlen >> 2; 8321e78ac21SJeffrey Hsu } 8331e78ac21SJeffrey Hsu m->m_len = mhlen; 83421d172a3SGleb Smirnoff /* XXX do we need to add ip_off below ? */ 83521d172a3SGleb Smirnoff mhip->ip_off = ((off - hlen) >> 3) + ip_off; 836fb86dfcdSAndre Oppermann if (off + len >= ip_len) 83721d172a3SGleb Smirnoff len = ip_len - off; 838fb86dfcdSAndre Oppermann else 8391e78ac21SJeffrey Hsu mhip->ip_off |= IP_MF; 8401e78ac21SJeffrey Hsu mhip->ip_len = htons((u_short)(len + mhlen)); 841eb1b1807SGleb Smirnoff m->m_next = m_copym(m0, off, len, M_NOWAIT); 8420b17fba7SAndre Oppermann if (m->m_next == NULL) { /* copy failed */ 8431e78ac21SJeffrey Hsu m_free(m); 8441e78ac21SJeffrey Hsu error = ENOBUFS; /* ??? */ 84586425c62SRobert Watson IPSTAT_INC(ips_odropped); 8461e78ac21SJeffrey Hsu goto done; 8471e78ac21SJeffrey Hsu } 8481e78ac21SJeffrey Hsu m->m_pkthdr.len = mhlen + len; 8491e78ac21SJeffrey Hsu #ifdef MAC 85030d239bcSRobert Watson mac_netinet_fragment(m0, m); 8511e78ac21SJeffrey Hsu #endif 8521e78ac21SJeffrey Hsu mhip->ip_off = htons(mhip->ip_off); 8531e78ac21SJeffrey Hsu mhip->ip_sum = 0; 854078468edSGleb Smirnoff if (m->m_pkthdr.csum_flags & CSUM_IP & ~if_hwassist_flags) { 8551e78ac21SJeffrey Hsu mhip->ip_sum = in_cksum(m, mhlen); 856078468edSGleb Smirnoff m->m_pkthdr.csum_flags &= ~CSUM_IP; 857078468edSGleb Smirnoff } 8581e78ac21SJeffrey Hsu *mnext = m; 8591e78ac21SJeffrey Hsu mnext = &m->m_nextpkt; 8601e78ac21SJeffrey Hsu } 86186425c62SRobert Watson IPSTAT_ADD(ips_ofragments, nfrags); 8621e78ac21SJeffrey Hsu 8631e78ac21SJeffrey Hsu /* 8641e78ac21SJeffrey Hsu * Update first fragment by trimming what's been copied out 8651e78ac21SJeffrey Hsu * and updating header. 8661e78ac21SJeffrey Hsu */ 86721d172a3SGleb Smirnoff m_adj(m0, hlen + firstlen - ip_len); 8681e78ac21SJeffrey Hsu m0->m_pkthdr.len = hlen + firstlen; 8691e78ac21SJeffrey Hsu ip->ip_len = htons((u_short)m0->m_pkthdr.len); 87021d172a3SGleb Smirnoff ip->ip_off = htons(ip_off | IP_MF); 8711e78ac21SJeffrey Hsu ip->ip_sum = 0; 872078468edSGleb Smirnoff if (m0->m_pkthdr.csum_flags & CSUM_IP & ~if_hwassist_flags) { 8731e78ac21SJeffrey Hsu ip->ip_sum = in_cksum(m0, hlen); 874078468edSGleb Smirnoff m0->m_pkthdr.csum_flags &= ~CSUM_IP; 875078468edSGleb Smirnoff } 8761e78ac21SJeffrey Hsu 8771e78ac21SJeffrey Hsu done: 8781e78ac21SJeffrey Hsu *m_frag = m0; 8791e78ac21SJeffrey Hsu return error; 8801e78ac21SJeffrey Hsu } 8811e78ac21SJeffrey Hsu 8821c238475SJonathan Lemon void 883db4f9cc7SJonathan Lemon in_delayed_cksum(struct mbuf *m) 884db4f9cc7SJonathan Lemon { 885db4f9cc7SJonathan Lemon struct ip *ip; 88623e9c6dcSGleb Smirnoff uint16_t csum, offset, ip_len; 887db4f9cc7SJonathan Lemon 888db4f9cc7SJonathan Lemon ip = mtod(m, struct ip *); 88953be11f6SPoul-Henning Kamp offset = ip->ip_hl << 2 ; 89023e9c6dcSGleb Smirnoff ip_len = ntohs(ip->ip_len); 89123e9c6dcSGleb Smirnoff csum = in_cksum_skip(m, ip_len, offset); 892206a3274SRuslan Ermilov if (m->m_pkthdr.csum_flags & CSUM_UDP && csum == 0) 893206a3274SRuslan Ermilov csum = 0xffff; 894db4f9cc7SJonathan Lemon offset += m->m_pkthdr.csum_data; /* checksum offset */ 895db4f9cc7SJonathan Lemon 8968e1d0a56SMichael Tuexen /* find the mbuf in the chain where the checksum starts*/ 8978e1d0a56SMichael Tuexen while ((m != NULL) && (offset >= m->m_len)) { 8988e1d0a56SMichael Tuexen offset -= m->m_len; 8998e1d0a56SMichael Tuexen m = m->m_next; 9008e1d0a56SMichael Tuexen } 90126461454SMichael Tuexen KASSERT(m != NULL, ("in_delayed_cksum: checksum outside mbuf chain.")); 90226461454SMichael Tuexen KASSERT(offset + sizeof(u_short) <= m->m_len, ("in_delayed_cksum: checksum split between mbufs.")); 903db4f9cc7SJonathan Lemon *(u_short *)(m->m_data + offset) = csum; 904db4f9cc7SJonathan Lemon } 905db4f9cc7SJonathan Lemon 906df8bae1dSRodney W. Grimes /* 907df8bae1dSRodney W. Grimes * IP socket option processing. 908df8bae1dSRodney W. Grimes */ 909df8bae1dSRodney W. Grimes int 910f2565d68SRobert Watson ip_ctloutput(struct socket *so, struct sockopt *sopt) 911df8bae1dSRodney W. Grimes { 912cfe8b629SGarrett Wollman struct inpcb *inp = sotoinpcb(so); 913cfe8b629SGarrett Wollman int error, optval; 914dc847eb6SAdrian Chadd #ifdef RSS 915dc847eb6SAdrian Chadd uint32_t rss_bucket; 916dc847eb6SAdrian Chadd int retval; 917dc847eb6SAdrian Chadd #endif 918df8bae1dSRodney W. Grimes 919cfe8b629SGarrett Wollman error = optval = 0; 920cfe8b629SGarrett Wollman if (sopt->sopt_level != IPPROTO_IP) { 921fc06cd42SMikolaj Golub error = EINVAL; 922fc06cd42SMikolaj Golub 923fc06cd42SMikolaj Golub if (sopt->sopt_level == SOL_SOCKET && 924fc06cd42SMikolaj Golub sopt->sopt_dir == SOPT_SET) { 925fc06cd42SMikolaj Golub switch (sopt->sopt_name) { 926fc06cd42SMikolaj Golub case SO_REUSEADDR: 927fc06cd42SMikolaj Golub INP_WLOCK(inp); 928efdf104bSMikolaj Golub if ((so->so_options & SO_REUSEADDR) != 0) 929efdf104bSMikolaj Golub inp->inp_flags2 |= INP_REUSEADDR; 930fc06cd42SMikolaj Golub else 931efdf104bSMikolaj Golub inp->inp_flags2 &= ~INP_REUSEADDR; 932fc06cd42SMikolaj Golub INP_WUNLOCK(inp); 933fc06cd42SMikolaj Golub error = 0; 934fc06cd42SMikolaj Golub break; 935fc06cd42SMikolaj Golub case SO_REUSEPORT: 936fc06cd42SMikolaj Golub INP_WLOCK(inp); 937fc06cd42SMikolaj Golub if ((so->so_options & SO_REUSEPORT) != 0) 938fc06cd42SMikolaj Golub inp->inp_flags2 |= INP_REUSEPORT; 939fc06cd42SMikolaj Golub else 940fc06cd42SMikolaj Golub inp->inp_flags2 &= ~INP_REUSEPORT; 941fc06cd42SMikolaj Golub INP_WUNLOCK(inp); 942fc06cd42SMikolaj Golub error = 0; 943fc06cd42SMikolaj Golub break; 944fc06cd42SMikolaj Golub case SO_SETFIB: 945fc06cd42SMikolaj Golub INP_WLOCK(inp); 946fc06cd42SMikolaj Golub inp->inp_inc.inc_fibnum = so->so_fibnum; 947fc06cd42SMikolaj Golub INP_WUNLOCK(inp); 948fc06cd42SMikolaj Golub error = 0; 949fc06cd42SMikolaj Golub break; 950fc06cd42SMikolaj Golub default: 951fc06cd42SMikolaj Golub break; 952fc06cd42SMikolaj Golub } 953fc06cd42SMikolaj Golub } 954fc06cd42SMikolaj Golub return (error); 955cfe8b629SGarrett Wollman } 956df8bae1dSRodney W. Grimes 957cfe8b629SGarrett Wollman switch (sopt->sopt_dir) { 958cfe8b629SGarrett Wollman case SOPT_SET: 959cfe8b629SGarrett Wollman switch (sopt->sopt_name) { 960df8bae1dSRodney W. Grimes case IP_OPTIONS: 961df8bae1dSRodney W. Grimes #ifdef notyet 962df8bae1dSRodney W. Grimes case IP_RETOPTS: 963df8bae1dSRodney W. Grimes #endif 964cfe8b629SGarrett Wollman { 965cfe8b629SGarrett Wollman struct mbuf *m; 966cfe8b629SGarrett Wollman if (sopt->sopt_valsize > MLEN) { 967cfe8b629SGarrett Wollman error = EMSGSIZE; 968cfe8b629SGarrett Wollman break; 969cfe8b629SGarrett Wollman } 970dc4ad05eSGleb Smirnoff m = m_get(sopt->sopt_td ? M_WAITOK : M_NOWAIT, MT_DATA); 9710b17fba7SAndre Oppermann if (m == NULL) { 972cfe8b629SGarrett Wollman error = ENOBUFS; 973cfe8b629SGarrett Wollman break; 974cfe8b629SGarrett Wollman } 975cfe8b629SGarrett Wollman m->m_len = sopt->sopt_valsize; 976cfe8b629SGarrett Wollman error = sooptcopyin(sopt, mtod(m, char *), m->m_len, 977cfe8b629SGarrett Wollman m->m_len); 978635354c4SMaxim Konovalov if (error) { 979635354c4SMaxim Konovalov m_free(m); 980635354c4SMaxim Konovalov break; 981635354c4SMaxim Konovalov } 9828501a69cSRobert Watson INP_WLOCK(inp); 983993d9505SRobert Watson error = ip_pcbopts(inp, sopt->sopt_name, m); 9848501a69cSRobert Watson INP_WUNLOCK(inp); 985993d9505SRobert Watson return (error); 986cfe8b629SGarrett Wollman } 987df8bae1dSRodney W. Grimes 988f44270e7SPawel Jakub Dawidek case IP_BINDANY: 989f44270e7SPawel Jakub Dawidek if (sopt->sopt_td != NULL) { 990f44270e7SPawel Jakub Dawidek error = priv_check(sopt->sopt_td, 991f44270e7SPawel Jakub Dawidek PRIV_NETINET_BINDANY); 992f44270e7SPawel Jakub Dawidek if (error) 993be9347e3SAdrian Chadd break; 994be9347e3SAdrian Chadd } 995cef27294SAdrian Chadd /* FALLTHROUGH */ 9960a100a6fSAdrian Chadd case IP_BINDMULTI: 9970a100a6fSAdrian Chadd #ifdef RSS 9980a100a6fSAdrian Chadd case IP_RSS_LISTEN_BUCKET: 9990a100a6fSAdrian Chadd #endif 1000df8bae1dSRodney W. Grimes case IP_TOS: 1001df8bae1dSRodney W. Grimes case IP_TTL: 1002936cd18dSAndre Oppermann case IP_MINTTL: 1003df8bae1dSRodney W. Grimes case IP_RECVOPTS: 1004df8bae1dSRodney W. Grimes case IP_RECVRETOPTS: 1005df8bae1dSRodney W. Grimes case IP_RECVDSTADDR: 10064957466bSMatthew N. Dodd case IP_RECVTTL: 100782c23ebaSBill Fenner case IP_RECVIF: 10088afa2304SBruce M Simpson case IP_ONESBCAST: 1009b2828ad2SAndre Oppermann case IP_DONTFRAG: 10103cca425bSMichael Tuexen case IP_RECVTOS: 10119d3ddf43SAdrian Chadd case IP_RECVFLOWID: 10129d3ddf43SAdrian Chadd #ifdef RSS 10139d3ddf43SAdrian Chadd case IP_RECVRSSBUCKETID: 10149d3ddf43SAdrian Chadd #endif 1015cfe8b629SGarrett Wollman error = sooptcopyin(sopt, &optval, sizeof optval, 1016cfe8b629SGarrett Wollman sizeof optval); 1017cfe8b629SGarrett Wollman if (error) 1018cfe8b629SGarrett Wollman break; 1019df8bae1dSRodney W. Grimes 1020cfe8b629SGarrett Wollman switch (sopt->sopt_name) { 1021df8bae1dSRodney W. Grimes case IP_TOS: 1022ca98b82cSDavid Greenman inp->inp_ip_tos = optval; 1023df8bae1dSRodney W. Grimes break; 1024df8bae1dSRodney W. Grimes 1025df8bae1dSRodney W. Grimes case IP_TTL: 1026ca98b82cSDavid Greenman inp->inp_ip_ttl = optval; 1027df8bae1dSRodney W. Grimes break; 1028936cd18dSAndre Oppermann 1029936cd18dSAndre Oppermann case IP_MINTTL: 1030a603c811SRobert Watson if (optval >= 0 && optval <= MAXTTL) 1031936cd18dSAndre Oppermann inp->inp_ip_minttl = optval; 1032936cd18dSAndre Oppermann else 1033936cd18dSAndre Oppermann error = EINVAL; 1034936cd18dSAndre Oppermann break; 1035936cd18dSAndre Oppermann 1036a138d217SRobert Watson #define OPTSET(bit) do { \ 10378501a69cSRobert Watson INP_WLOCK(inp); \ 1038df8bae1dSRodney W. Grimes if (optval) \ 1039df8bae1dSRodney W. Grimes inp->inp_flags |= bit; \ 1040df8bae1dSRodney W. Grimes else \ 1041a138d217SRobert Watson inp->inp_flags &= ~bit; \ 10428501a69cSRobert Watson INP_WUNLOCK(inp); \ 1043a138d217SRobert Watson } while (0) 1044df8bae1dSRodney W. Grimes 10450a100a6fSAdrian Chadd #define OPTSET2(bit, val) do { \ 10460a100a6fSAdrian Chadd INP_WLOCK(inp); \ 10470a100a6fSAdrian Chadd if (val) \ 10480a100a6fSAdrian Chadd inp->inp_flags2 |= bit; \ 10490a100a6fSAdrian Chadd else \ 10500a100a6fSAdrian Chadd inp->inp_flags2 &= ~bit; \ 10510a100a6fSAdrian Chadd INP_WUNLOCK(inp); \ 10520a100a6fSAdrian Chadd } while (0) 10530a100a6fSAdrian Chadd 1054df8bae1dSRodney W. Grimes case IP_RECVOPTS: 1055df8bae1dSRodney W. Grimes OPTSET(INP_RECVOPTS); 1056df8bae1dSRodney W. Grimes break; 1057df8bae1dSRodney W. Grimes 1058df8bae1dSRodney W. Grimes case IP_RECVRETOPTS: 1059df8bae1dSRodney W. Grimes OPTSET(INP_RECVRETOPTS); 1060df8bae1dSRodney W. Grimes break; 1061df8bae1dSRodney W. Grimes 1062df8bae1dSRodney W. Grimes case IP_RECVDSTADDR: 1063df8bae1dSRodney W. Grimes OPTSET(INP_RECVDSTADDR); 1064df8bae1dSRodney W. Grimes break; 106582c23ebaSBill Fenner 10664957466bSMatthew N. Dodd case IP_RECVTTL: 10674957466bSMatthew N. Dodd OPTSET(INP_RECVTTL); 10684957466bSMatthew N. Dodd break; 10694957466bSMatthew N. Dodd 107082c23ebaSBill Fenner case IP_RECVIF: 107182c23ebaSBill Fenner OPTSET(INP_RECVIF); 107282c23ebaSBill Fenner break; 10736a800098SYoshinobu Inoue 10748afa2304SBruce M Simpson case IP_ONESBCAST: 10758afa2304SBruce M Simpson OPTSET(INP_ONESBCAST); 10768afa2304SBruce M Simpson break; 1077b2828ad2SAndre Oppermann case IP_DONTFRAG: 1078b2828ad2SAndre Oppermann OPTSET(INP_DONTFRAG); 1079b2828ad2SAndre Oppermann break; 1080f44270e7SPawel Jakub Dawidek case IP_BINDANY: 1081f44270e7SPawel Jakub Dawidek OPTSET(INP_BINDANY); 1082be9347e3SAdrian Chadd break; 10833cca425bSMichael Tuexen case IP_RECVTOS: 10843cca425bSMichael Tuexen OPTSET(INP_RECVTOS); 10853cca425bSMichael Tuexen break; 10860a100a6fSAdrian Chadd case IP_BINDMULTI: 10870a100a6fSAdrian Chadd OPTSET2(INP_BINDMULTI, optval); 10880a100a6fSAdrian Chadd break; 10899d3ddf43SAdrian Chadd case IP_RECVFLOWID: 10909d3ddf43SAdrian Chadd OPTSET2(INP_RECVFLOWID, optval); 10919d3ddf43SAdrian Chadd break; 10920a100a6fSAdrian Chadd #ifdef RSS 10930a100a6fSAdrian Chadd case IP_RSS_LISTEN_BUCKET: 10940a100a6fSAdrian Chadd if ((optval >= 0) && 10950a100a6fSAdrian Chadd (optval < rss_getnumbuckets())) { 10960a100a6fSAdrian Chadd inp->inp_rss_listen_bucket = optval; 10970a100a6fSAdrian Chadd OPTSET2(INP_RSS_BUCKET_SET, 1); 10980a100a6fSAdrian Chadd } else { 10990a100a6fSAdrian Chadd error = EINVAL; 11000a100a6fSAdrian Chadd } 11010a100a6fSAdrian Chadd break; 11029d3ddf43SAdrian Chadd case IP_RECVRSSBUCKETID: 11039d3ddf43SAdrian Chadd OPTSET2(INP_RECVRSSBUCKETID, optval); 11049d3ddf43SAdrian Chadd break; 11050a100a6fSAdrian Chadd #endif 1106df8bae1dSRodney W. Grimes } 1107df8bae1dSRodney W. Grimes break; 1108df8bae1dSRodney W. Grimes #undef OPTSET 11090a100a6fSAdrian Chadd #undef OPTSET2 1110df8bae1dSRodney W. Grimes 111171498f30SBruce M Simpson /* 111271498f30SBruce M Simpson * Multicast socket options are processed by the in_mcast 111371498f30SBruce M Simpson * module. 111471498f30SBruce M Simpson */ 1115df8bae1dSRodney W. Grimes case IP_MULTICAST_IF: 1116f0068c4aSGarrett Wollman case IP_MULTICAST_VIF: 1117df8bae1dSRodney W. Grimes case IP_MULTICAST_TTL: 1118df8bae1dSRodney W. Grimes case IP_MULTICAST_LOOP: 1119df8bae1dSRodney W. Grimes case IP_ADD_MEMBERSHIP: 1120df8bae1dSRodney W. Grimes case IP_DROP_MEMBERSHIP: 112171498f30SBruce M Simpson case IP_ADD_SOURCE_MEMBERSHIP: 112271498f30SBruce M Simpson case IP_DROP_SOURCE_MEMBERSHIP: 112371498f30SBruce M Simpson case IP_BLOCK_SOURCE: 112471498f30SBruce M Simpson case IP_UNBLOCK_SOURCE: 112571498f30SBruce M Simpson case IP_MSFILTER: 112671498f30SBruce M Simpson case MCAST_JOIN_GROUP: 112771498f30SBruce M Simpson case MCAST_LEAVE_GROUP: 112871498f30SBruce M Simpson case MCAST_JOIN_SOURCE_GROUP: 112971498f30SBruce M Simpson case MCAST_LEAVE_SOURCE_GROUP: 113071498f30SBruce M Simpson case MCAST_BLOCK_SOURCE: 113171498f30SBruce M Simpson case MCAST_UNBLOCK_SOURCE: 113271498f30SBruce M Simpson error = inp_setmoptions(inp, sopt); 1133df8bae1dSRodney W. Grimes break; 1134df8bae1dSRodney W. Grimes 113533b3ac06SPeter Wemm case IP_PORTRANGE: 1136cfe8b629SGarrett Wollman error = sooptcopyin(sopt, &optval, sizeof optval, 1137cfe8b629SGarrett Wollman sizeof optval); 1138cfe8b629SGarrett Wollman if (error) 1139cfe8b629SGarrett Wollman break; 114033b3ac06SPeter Wemm 11418501a69cSRobert Watson INP_WLOCK(inp); 114233b3ac06SPeter Wemm switch (optval) { 114333b3ac06SPeter Wemm case IP_PORTRANGE_DEFAULT: 114433b3ac06SPeter Wemm inp->inp_flags &= ~(INP_LOWPORT); 114533b3ac06SPeter Wemm inp->inp_flags &= ~(INP_HIGHPORT); 114633b3ac06SPeter Wemm break; 114733b3ac06SPeter Wemm 114833b3ac06SPeter Wemm case IP_PORTRANGE_HIGH: 114933b3ac06SPeter Wemm inp->inp_flags &= ~(INP_LOWPORT); 115033b3ac06SPeter Wemm inp->inp_flags |= INP_HIGHPORT; 115133b3ac06SPeter Wemm break; 115233b3ac06SPeter Wemm 115333b3ac06SPeter Wemm case IP_PORTRANGE_LOW: 115433b3ac06SPeter Wemm inp->inp_flags &= ~(INP_HIGHPORT); 115533b3ac06SPeter Wemm inp->inp_flags |= INP_LOWPORT; 115633b3ac06SPeter Wemm break; 115733b3ac06SPeter Wemm 115833b3ac06SPeter Wemm default: 115933b3ac06SPeter Wemm error = EINVAL; 116033b3ac06SPeter Wemm break; 116133b3ac06SPeter Wemm } 11628501a69cSRobert Watson INP_WUNLOCK(inp); 1163ce8c72b1SPeter Wemm break; 116433b3ac06SPeter Wemm 1165b2630c29SGeorge V. Neville-Neil #ifdef IPSEC 11666a800098SYoshinobu Inoue case IP_IPSEC_POLICY: 11676a800098SYoshinobu Inoue { 11686a800098SYoshinobu Inoue caddr_t req; 11696a800098SYoshinobu Inoue struct mbuf *m; 11706a800098SYoshinobu Inoue 11716a800098SYoshinobu Inoue if ((error = soopt_getm(sopt, &m)) != 0) /* XXX */ 11726a800098SYoshinobu Inoue break; 11736a800098SYoshinobu Inoue if ((error = soopt_mcopyin(sopt, m)) != 0) /* XXX */ 11746a800098SYoshinobu Inoue break; 11756a800098SYoshinobu Inoue req = mtod(m, caddr_t); 117697aa4a51SBjoern A. Zeeb error = ipsec_set_policy(inp, sopt->sopt_name, req, 1177c26fe973SBjoern A. Zeeb m->m_len, (sopt->sopt_td != NULL) ? 1178c26fe973SBjoern A. Zeeb sopt->sopt_td->td_ucred : NULL); 11796a800098SYoshinobu Inoue m_freem(m); 11806a800098SYoshinobu Inoue break; 11816a800098SYoshinobu Inoue } 1182b2630c29SGeorge V. Neville-Neil #endif /* IPSEC */ 11836a800098SYoshinobu Inoue 1184df8bae1dSRodney W. Grimes default: 1185df8bae1dSRodney W. Grimes error = ENOPROTOOPT; 1186df8bae1dSRodney W. Grimes break; 1187df8bae1dSRodney W. Grimes } 1188df8bae1dSRodney W. Grimes break; 1189df8bae1dSRodney W. Grimes 1190cfe8b629SGarrett Wollman case SOPT_GET: 1191cfe8b629SGarrett Wollman switch (sopt->sopt_name) { 1192df8bae1dSRodney W. Grimes case IP_OPTIONS: 1193df8bae1dSRodney W. Grimes case IP_RETOPTS: 1194cfe8b629SGarrett Wollman if (inp->inp_options) 1195cfe8b629SGarrett Wollman error = sooptcopyout(sopt, 1196cfe8b629SGarrett Wollman mtod(inp->inp_options, 1197cfe8b629SGarrett Wollman char *), 1198cfe8b629SGarrett Wollman inp->inp_options->m_len); 1199cfe8b629SGarrett Wollman else 1200cfe8b629SGarrett Wollman sopt->sopt_valsize = 0; 1201df8bae1dSRodney W. Grimes break; 1202df8bae1dSRodney W. Grimes 1203df8bae1dSRodney W. Grimes case IP_TOS: 1204df8bae1dSRodney W. Grimes case IP_TTL: 1205936cd18dSAndre Oppermann case IP_MINTTL: 1206df8bae1dSRodney W. Grimes case IP_RECVOPTS: 1207df8bae1dSRodney W. Grimes case IP_RECVRETOPTS: 1208df8bae1dSRodney W. Grimes case IP_RECVDSTADDR: 12094957466bSMatthew N. Dodd case IP_RECVTTL: 121082c23ebaSBill Fenner case IP_RECVIF: 1211cfe8b629SGarrett Wollman case IP_PORTRANGE: 12128afa2304SBruce M Simpson case IP_ONESBCAST: 1213b2828ad2SAndre Oppermann case IP_DONTFRAG: 12145f6bf451SAttilio Rao case IP_BINDANY: 12153cca425bSMichael Tuexen case IP_RECVTOS: 12160a100a6fSAdrian Chadd case IP_BINDMULTI: 12179c423972SAdrian Chadd case IP_FLOWID: 12189c423972SAdrian Chadd case IP_FLOWTYPE: 12199d3ddf43SAdrian Chadd case IP_RECVFLOWID: 12200a100a6fSAdrian Chadd #ifdef RSS 12210a100a6fSAdrian Chadd case IP_RSSBUCKETID: 12229d3ddf43SAdrian Chadd case IP_RECVRSSBUCKETID: 12230a100a6fSAdrian Chadd #endif 1224cfe8b629SGarrett Wollman switch (sopt->sopt_name) { 1225df8bae1dSRodney W. Grimes 1226df8bae1dSRodney W. Grimes case IP_TOS: 1227ca98b82cSDavid Greenman optval = inp->inp_ip_tos; 1228df8bae1dSRodney W. Grimes break; 1229df8bae1dSRodney W. Grimes 1230df8bae1dSRodney W. Grimes case IP_TTL: 1231ca98b82cSDavid Greenman optval = inp->inp_ip_ttl; 1232df8bae1dSRodney W. Grimes break; 1233df8bae1dSRodney W. Grimes 1234936cd18dSAndre Oppermann case IP_MINTTL: 1235936cd18dSAndre Oppermann optval = inp->inp_ip_minttl; 1236936cd18dSAndre Oppermann break; 1237936cd18dSAndre Oppermann 1238df8bae1dSRodney W. Grimes #define OPTBIT(bit) (inp->inp_flags & bit ? 1 : 0) 12390a100a6fSAdrian Chadd #define OPTBIT2(bit) (inp->inp_flags2 & bit ? 1 : 0) 1240df8bae1dSRodney W. Grimes 1241df8bae1dSRodney W. Grimes case IP_RECVOPTS: 1242df8bae1dSRodney W. Grimes optval = OPTBIT(INP_RECVOPTS); 1243df8bae1dSRodney W. Grimes break; 1244df8bae1dSRodney W. Grimes 1245df8bae1dSRodney W. Grimes case IP_RECVRETOPTS: 1246df8bae1dSRodney W. Grimes optval = OPTBIT(INP_RECVRETOPTS); 1247df8bae1dSRodney W. Grimes break; 1248df8bae1dSRodney W. Grimes 1249df8bae1dSRodney W. Grimes case IP_RECVDSTADDR: 1250df8bae1dSRodney W. Grimes optval = OPTBIT(INP_RECVDSTADDR); 1251df8bae1dSRodney W. Grimes break; 125282c23ebaSBill Fenner 12534957466bSMatthew N. Dodd case IP_RECVTTL: 12544957466bSMatthew N. Dodd optval = OPTBIT(INP_RECVTTL); 12554957466bSMatthew N. Dodd break; 12564957466bSMatthew N. Dodd 125782c23ebaSBill Fenner case IP_RECVIF: 125882c23ebaSBill Fenner optval = OPTBIT(INP_RECVIF); 125982c23ebaSBill Fenner break; 1260cfe8b629SGarrett Wollman 1261cfe8b629SGarrett Wollman case IP_PORTRANGE: 1262cfe8b629SGarrett Wollman if (inp->inp_flags & INP_HIGHPORT) 1263cfe8b629SGarrett Wollman optval = IP_PORTRANGE_HIGH; 1264cfe8b629SGarrett Wollman else if (inp->inp_flags & INP_LOWPORT) 1265cfe8b629SGarrett Wollman optval = IP_PORTRANGE_LOW; 1266cfe8b629SGarrett Wollman else 1267cfe8b629SGarrett Wollman optval = 0; 1268cfe8b629SGarrett Wollman break; 12696a800098SYoshinobu Inoue 12708afa2304SBruce M Simpson case IP_ONESBCAST: 12718afa2304SBruce M Simpson optval = OPTBIT(INP_ONESBCAST); 12728afa2304SBruce M Simpson break; 1273b2828ad2SAndre Oppermann case IP_DONTFRAG: 1274b2828ad2SAndre Oppermann optval = OPTBIT(INP_DONTFRAG); 1275b2828ad2SAndre Oppermann break; 12765f6bf451SAttilio Rao case IP_BINDANY: 12775f6bf451SAttilio Rao optval = OPTBIT(INP_BINDANY); 12785f6bf451SAttilio Rao break; 12793cca425bSMichael Tuexen case IP_RECVTOS: 12803cca425bSMichael Tuexen optval = OPTBIT(INP_RECVTOS); 12813cca425bSMichael Tuexen break; 12829c423972SAdrian Chadd case IP_FLOWID: 12839c423972SAdrian Chadd optval = inp->inp_flowid; 12849c423972SAdrian Chadd break; 12859c423972SAdrian Chadd case IP_FLOWTYPE: 12869c423972SAdrian Chadd optval = inp->inp_flowtype; 12879c423972SAdrian Chadd break; 12889d3ddf43SAdrian Chadd case IP_RECVFLOWID: 12899d3ddf43SAdrian Chadd optval = OPTBIT2(INP_RECVFLOWID); 12909d3ddf43SAdrian Chadd break; 12919c423972SAdrian Chadd #ifdef RSS 12927847796aSAdrian Chadd case IP_RSSBUCKETID: 12937847796aSAdrian Chadd retval = rss_hash2bucket(inp->inp_flowid, 12947847796aSAdrian Chadd inp->inp_flowtype, 12957847796aSAdrian Chadd &rss_bucket); 12967847796aSAdrian Chadd if (retval == 0) 12977847796aSAdrian Chadd optval = rss_bucket; 12987847796aSAdrian Chadd else 12997847796aSAdrian Chadd error = EINVAL; 13009c423972SAdrian Chadd break; 13019d3ddf43SAdrian Chadd case IP_RECVRSSBUCKETID: 13029d3ddf43SAdrian Chadd optval = OPTBIT2(INP_RECVRSSBUCKETID); 13039d3ddf43SAdrian Chadd break; 13049c423972SAdrian Chadd #endif 13050a100a6fSAdrian Chadd case IP_BINDMULTI: 13060a100a6fSAdrian Chadd optval = OPTBIT2(INP_BINDMULTI); 13070a100a6fSAdrian Chadd break; 1308df8bae1dSRodney W. Grimes } 1309cfe8b629SGarrett Wollman error = sooptcopyout(sopt, &optval, sizeof optval); 1310df8bae1dSRodney W. Grimes break; 1311df8bae1dSRodney W. Grimes 131271498f30SBruce M Simpson /* 131371498f30SBruce M Simpson * Multicast socket options are processed by the in_mcast 131471498f30SBruce M Simpson * module. 131571498f30SBruce M Simpson */ 1316df8bae1dSRodney W. Grimes case IP_MULTICAST_IF: 1317f0068c4aSGarrett Wollman case IP_MULTICAST_VIF: 1318df8bae1dSRodney W. Grimes case IP_MULTICAST_TTL: 1319df8bae1dSRodney W. Grimes case IP_MULTICAST_LOOP: 132071498f30SBruce M Simpson case IP_MSFILTER: 132171498f30SBruce M Simpson error = inp_getmoptions(inp, sopt); 132233b3ac06SPeter Wemm break; 132333b3ac06SPeter Wemm 1324b2630c29SGeorge V. Neville-Neil #ifdef IPSEC 13256a800098SYoshinobu Inoue case IP_IPSEC_POLICY: 13266a800098SYoshinobu Inoue { 1327f63e7634SYoshinobu Inoue struct mbuf *m = NULL; 13286a800098SYoshinobu Inoue caddr_t req = NULL; 1329686cdd19SJun-ichiro itojun Hagino size_t len = 0; 13306a800098SYoshinobu Inoue 1331686cdd19SJun-ichiro itojun Hagino if (m != 0) { 13326a800098SYoshinobu Inoue req = mtod(m, caddr_t); 1333686cdd19SJun-ichiro itojun Hagino len = m->m_len; 1334686cdd19SJun-ichiro itojun Hagino } 133597aa4a51SBjoern A. Zeeb error = ipsec_get_policy(sotoinpcb(so), req, len, &m); 13366a800098SYoshinobu Inoue if (error == 0) 13376a800098SYoshinobu Inoue error = soopt_mcopyout(sopt, m); /* XXX */ 1338f63e7634SYoshinobu Inoue if (error == 0) 13396a800098SYoshinobu Inoue m_freem(m); 13406a800098SYoshinobu Inoue break; 13416a800098SYoshinobu Inoue } 1342b2630c29SGeorge V. Neville-Neil #endif /* IPSEC */ 13436a800098SYoshinobu Inoue 1344df8bae1dSRodney W. Grimes default: 1345df8bae1dSRodney W. Grimes error = ENOPROTOOPT; 1346df8bae1dSRodney W. Grimes break; 1347df8bae1dSRodney W. Grimes } 1348df8bae1dSRodney W. Grimes break; 1349df8bae1dSRodney W. Grimes } 1350df8bae1dSRodney W. Grimes return (error); 1351df8bae1dSRodney W. Grimes } 1352df8bae1dSRodney W. Grimes 1353df8bae1dSRodney W. Grimes /* 1354df8bae1dSRodney W. Grimes * Routine called from ip_output() to loop back a copy of an IP multicast 1355df8bae1dSRodney W. Grimes * packet to the input queue of a specified interface. Note that this 1356df8bae1dSRodney W. Grimes * calls the output routine of the loopback "driver", but with an interface 1357f5fea3ddSPaul Traina * pointer that might NOT be a loopback interface -- evil, but easier than 1358f5fea3ddSPaul Traina * replicating that code here. 1359df8bae1dSRodney W. Grimes */ 1360df8bae1dSRodney W. Grimes static void 1361331dff07SAlexander V. Chernikov ip_mloopback(struct ifnet *ifp, const struct mbuf *m, int hlen) 1362df8bae1dSRodney W. Grimes { 1363331dff07SAlexander V. Chernikov struct ip *ip; 1364df8bae1dSRodney W. Grimes struct mbuf *copym; 1365df8bae1dSRodney W. Grimes 1366e4762f75SGeorge V. Neville-Neil /* 1367e4762f75SGeorge V. Neville-Neil * Make a deep copy of the packet because we're going to 1368e4762f75SGeorge V. Neville-Neil * modify the pack in order to generate checksums. 1369e4762f75SGeorge V. Neville-Neil */ 1370eb1b1807SGleb Smirnoff copym = m_dup(m, M_NOWAIT); 1371f0cace5dSRobert Watson if (copym != NULL && (!M_WRITABLE(copym) || copym->m_len < hlen)) 137286b1d6d2SBill Fenner copym = m_pullup(copym, hlen); 1373df8bae1dSRodney W. Grimes if (copym != NULL) { 1374390cdc6aSRuslan Ermilov /* If needed, compute the checksum and mark it as valid. */ 1375390cdc6aSRuslan Ermilov if (copym->m_pkthdr.csum_flags & CSUM_DELAY_DATA) { 1376390cdc6aSRuslan Ermilov in_delayed_cksum(copym); 1377390cdc6aSRuslan Ermilov copym->m_pkthdr.csum_flags &= ~CSUM_DELAY_DATA; 1378390cdc6aSRuslan Ermilov copym->m_pkthdr.csum_flags |= 1379390cdc6aSRuslan Ermilov CSUM_DATA_VALID | CSUM_PSEUDO_HDR; 1380390cdc6aSRuslan Ermilov copym->m_pkthdr.csum_data = 0xffff; 1381390cdc6aSRuslan Ermilov } 1382df8bae1dSRodney W. Grimes /* 1383df8bae1dSRodney W. Grimes * We don't bother to fragment if the IP length is greater 1384df8bae1dSRodney W. Grimes * than the interface's MTU. Can this possibly matter? 1385df8bae1dSRodney W. Grimes */ 13868f134647SGleb Smirnoff ip = mtod(copym, struct ip *); 1387df8bae1dSRodney W. Grimes ip->ip_sum = 0; 138886b1d6d2SBill Fenner ip->ip_sum = in_cksum(copym, hlen); 1389331dff07SAlexander V. Chernikov if_simloop(ifp, copym, AF_INET, 0); 1390df8bae1dSRodney W. Grimes } 1391df8bae1dSRodney W. Grimes } 1392