2 * SPDX-License-Identifier: BSD-3-Clause
4 * Copyright (c) 1982, 1986, 1988, 1993
5 * The Regents of the University of California.
8 * Redistribution and use in source and binary forms, with or without
9 * modification, are permitted provided that the following conditions
11 * 1. Redistributions of source code must retain the above copyright
12 * notice, this list of conditions and the following disclaimer.
13 * 2. Redistributions in binary form must reproduce the above copyright
14 * notice, this list of conditions and the following disclaimer in the
15 * documentation and/or other materials provided with the distribution.
16 * 3. Neither the name of the University nor the names of its contributors
17 * may be used to endorse or promote products derived from this software
18 * without specific prior written permission.
20 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
21 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
22 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
23 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
24 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
25 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
26 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
27 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
28 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
29 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
32 * @(#)raw_ip.c 8.7 (Berkeley) 5/15/95
35 #include <sys/cdefs.h>
36 __FBSDID("$FreeBSD$");
39 #include "opt_inet6.h"
40 #include "opt_ipsec.h"
41 #include "opt_route.h"
43 #include <sys/param.h>
45 #include <sys/kernel.h>
46 #include <sys/eventhandler.h>
48 #include <sys/malloc.h>
52 #include <sys/protosw.h>
53 #include <sys/rwlock.h>
54 #include <sys/signalvar.h>
55 #include <sys/socket.h>
56 #include <sys/socketvar.h>
58 #include <sys/sysctl.h>
59 #include <sys/systm.h>
64 #include <net/if_var.h>
65 #include <net/route.h>
66 #include <net/route/route_ctl.h>
69 #include <netinet/in.h>
70 #include <netinet/in_systm.h>
71 #include <netinet/in_fib.h>
72 #include <netinet/in_pcb.h>
73 #include <netinet/in_var.h>
74 #include <netinet/if_ether.h>
75 #include <netinet/ip.h>
76 #include <netinet/ip_var.h>
77 #include <netinet/ip_mroute.h>
78 #include <netinet/ip_icmp.h>
80 #include <netipsec/ipsec_support.h>
82 #include <machine/stdarg.h>
83 #include <security/mac/mac_framework.h>
85 VNET_DEFINE(int, ip_defttl) = IPDEFTTL;
86 SYSCTL_INT(_net_inet_ip, IPCTL_DEFTTL, ttl, CTLFLAG_VNET | CTLFLAG_RW,
87 &VNET_NAME(ip_defttl), 0,
88 "Maximum TTL on IP packets");
90 VNET_DEFINE(struct inpcbinfo, ripcbinfo);
91 #define V_ripcbinfo VNET(ripcbinfo)
94 * Control and data hooks for ipfw, dummynet, divert and so on.
95 * The data hooks are not used here but it is convenient
96 * to keep them all in one place.
98 VNET_DEFINE(ip_fw_chk_ptr_t, ip_fw_chk_ptr) = NULL;
99 VNET_DEFINE(ip_fw_ctl_ptr_t, ip_fw_ctl_ptr) = NULL;
101 int (*ip_dn_ctl_ptr)(struct sockopt *);
102 int (*ip_dn_io_ptr)(struct mbuf **, struct ip_fw_args *);
103 void (*ip_divert_ptr)(struct mbuf *, bool);
104 int (*ng_ipfw_input_p)(struct mbuf **, struct ip_fw_args *, bool);
108 * Hooks for multicast routing. They all default to NULL, so leave them not
109 * initialized and rely on BSS being set to 0.
113 * The socket used to communicate with the multicast routing daemon.
115 VNET_DEFINE(struct socket *, ip_mrouter);
118 * The various mrouter and rsvp functions.
120 int (*ip_mrouter_set)(struct socket *, struct sockopt *);
121 int (*ip_mrouter_get)(struct socket *, struct sockopt *);
122 int (*ip_mrouter_done)(void);
123 int (*ip_mforward)(struct ip *, struct ifnet *, struct mbuf *,
124 struct ip_moptions *);
125 int (*mrt_ioctl)(u_long, caddr_t, int);
126 int (*legal_vif_num)(int);
127 u_long (*ip_mcast_src)(int);
129 int (*rsvp_input_p)(struct mbuf **, int *, int);
130 int (*ip_rsvp_vif)(struct socket *, struct sockopt *);
131 void (*ip_rsvp_force_done)(struct socket *);
134 extern struct protosw inetsw[];
136 u_long rip_sendspace = 9216;
137 SYSCTL_ULONG(_net_inet_raw, OID_AUTO, maxdgram, CTLFLAG_RW,
138 &rip_sendspace, 0, "Maximum outgoing raw IP datagram size");
140 u_long rip_recvspace = 9216;
141 SYSCTL_ULONG(_net_inet_raw, OID_AUTO, recvspace, CTLFLAG_RW,
142 &rip_recvspace, 0, "Maximum space for incoming raw IP datagrams");
148 #define INP_PCBHASH_RAW_SIZE 256
149 #define INP_PCBHASH_RAW(proto, laddr, faddr, mask) \
150 (((proto) + (laddr) + (faddr)) % (mask) + 1)
154 rip_inshash(struct inpcb *inp)
156 struct inpcbinfo *pcbinfo = inp->inp_pcbinfo;
157 struct inpcbhead *pcbhash;
160 INP_HASH_WLOCK_ASSERT(pcbinfo);
161 INP_WLOCK_ASSERT(inp);
163 if (inp->inp_ip_p != 0 &&
164 inp->inp_laddr.s_addr != INADDR_ANY &&
165 inp->inp_faddr.s_addr != INADDR_ANY) {
166 hash = INP_PCBHASH_RAW(inp->inp_ip_p, inp->inp_laddr.s_addr,
167 inp->inp_faddr.s_addr, pcbinfo->ipi_hashmask);
170 pcbhash = &pcbinfo->ipi_hashbase[hash];
171 CK_LIST_INSERT_HEAD(pcbhash, inp, inp_hash);
175 rip_delhash(struct inpcb *inp)
178 INP_HASH_WLOCK_ASSERT(inp->inp_pcbinfo);
179 INP_WLOCK_ASSERT(inp);
181 CK_LIST_REMOVE(inp, inp_hash);
185 INPCBSTORAGE_DEFINE(ripcbstor, "rawinp", "ripcb", "rip", "riphash");
188 rip_init(void *arg __unused)
191 in_pcbinfo_init(&V_ripcbinfo, &ripcbstor, INP_PCBHASH_RAW_SIZE, 1);
193 VNET_SYSINIT(rip_init, SI_SUB_PROTO_DOMAIN, SI_ORDER_THIRD, rip_init, NULL);
197 rip_destroy(void *unused __unused)
200 in_pcbinfo_destroy(&V_ripcbinfo);
202 VNET_SYSUNINIT(raw_ip, SI_SUB_PROTO_DOMAIN, SI_ORDER_FOURTH, rip_destroy, NULL);
207 rip_append(struct inpcb *inp, struct ip *ip, struct mbuf *m,
208 struct sockaddr_in *ripsrc)
210 struct socket *so = inp->inp_socket;
211 struct mbuf *n, *opts = NULL;
213 INP_LOCK_ASSERT(inp);
215 #if defined(IPSEC) || defined(IPSEC_SUPPORT)
216 /* check AH/ESP integrity. */
217 if (IPSEC_ENABLED(ipv4) && IPSEC_CHECK_POLICY(ipv4, m, inp) != 0)
221 if (mac_inpcb_check_deliver(inp, m) != 0)
224 /* Check the minimum TTL for socket. */
225 if (inp->inp_ip_minttl && inp->inp_ip_minttl > ip->ip_ttl)
228 if ((n = m_copym(m, 0, M_COPYALL, M_NOWAIT)) == NULL)
231 if ((inp->inp_flags & INP_CONTROLOPTS) ||
232 (so->so_options & (SO_TIMESTAMP | SO_BINTIME)))
233 ip_savecontrol(inp, &opts, ip, n);
234 SOCKBUF_LOCK(&so->so_rcv);
235 if (sbappendaddr_locked(&so->so_rcv,
236 (struct sockaddr *)ripsrc, n, opts) == 0) {
237 soroverflow_locked(so);
243 sorwakeup_locked(so);
248 struct rip_inp_match_ctx {
254 rip_inp_match1(const struct inpcb *inp, void *v)
256 struct rip_inp_match_ctx *ctx = v;
258 if (inp->inp_ip_p != ctx->proto)
261 /* XXX inp locking */
262 if ((inp->inp_vflag & INP_IPV4) == 0)
265 if (inp->inp_laddr.s_addr != ctx->ip->ip_dst.s_addr)
267 if (inp->inp_faddr.s_addr != ctx->ip->ip_src.s_addr)
273 rip_inp_match2(const struct inpcb *inp, void *v)
275 struct rip_inp_match_ctx *ctx = v;
277 if (inp->inp_ip_p && inp->inp_ip_p != ctx->proto)
280 /* XXX inp locking */
281 if ((inp->inp_vflag & INP_IPV4) == 0)
284 if (!in_nullhost(inp->inp_laddr) &&
285 !in_hosteq(inp->inp_laddr, ctx->ip->ip_dst))
287 if (!in_nullhost(inp->inp_faddr) &&
288 !in_hosteq(inp->inp_faddr, ctx->ip->ip_src))
294 * Setup generic address and protocol structures for raw_input routine, then
295 * pass them along with mbuf chain.
298 rip_input(struct mbuf **mp, int *offp, int proto)
300 struct rip_inp_match_ctx ctx = {
301 .ip = mtod(*mp, struct ip *),
304 struct inpcb_iterator inpi = INP_ITERATOR(&V_ripcbinfo,
305 INPLOOKUP_RLOCKPCB, rip_inp_match1, &ctx);
307 struct mbuf *m = *mp;
309 struct sockaddr_in ripsrc;
315 bzero(&ripsrc, sizeof(ripsrc));
316 ripsrc.sin_len = sizeof(ripsrc);
317 ripsrc.sin_family = AF_INET;
318 ripsrc.sin_addr = ctx.ip->ip_src;
320 ifp = m->m_pkthdr.rcvif;
322 inpi.hash = INP_PCBHASH_RAW(proto, ctx.ip->ip_src.s_addr,
323 ctx.ip->ip_dst.s_addr, V_ripcbinfo.ipi_hashmask);
324 while ((inp = inp_next(&inpi)) != NULL) {
325 INP_RLOCK_ASSERT(inp);
326 if (jailed_without_vnet(inp->inp_cred) &&
327 prison_check_ip4(inp->inp_cred, &ctx.ip->ip_dst) != 0) {
329 * XXX: If faddr was bound to multicast group,
330 * jailed raw socket will drop datagram.
334 appended += rip_append(inp, ctx.ip, m, &ripsrc);
338 inpi.match = rip_inp_match2;
339 MPASS(inpi.inp == NULL);
340 while ((inp = inp_next(&inpi)) != NULL) {
341 INP_RLOCK_ASSERT(inp);
342 if (jailed_without_vnet(inp->inp_cred) &&
343 !IN_MULTICAST(ntohl(ctx.ip->ip_dst.s_addr)) &&
344 prison_check_ip4(inp->inp_cred, &ctx.ip->ip_dst) != 0)
346 * Allow raw socket in jail to receive multicast;
347 * assume process had PRIV_NETINET_RAW at attach,
348 * and fall through into normal filter path if so.
352 * If this raw socket has multicast state, and we
353 * have received a multicast, check if this socket
354 * should receive it, as multicast filtering is now
355 * the responsibility of the transport layer.
357 if (inp->inp_moptions != NULL &&
358 IN_MULTICAST(ntohl(ctx.ip->ip_dst.s_addr))) {
360 * If the incoming datagram is for IGMP, allow it
361 * through unconditionally to the raw socket.
363 * In the case of IGMPv2, we may not have explicitly
364 * joined the group, and may have set IFF_ALLMULTI
365 * on the interface. imo_multi_filter() may discard
366 * control traffic we actually need to see.
368 * Userland multicast routing daemons should continue
369 * filter the control traffic appropriately.
373 blocked = MCAST_PASS;
374 if (proto != IPPROTO_IGMP) {
375 struct sockaddr_in group;
377 bzero(&group, sizeof(struct sockaddr_in));
378 group.sin_len = sizeof(struct sockaddr_in);
379 group.sin_family = AF_INET;
380 group.sin_addr = ctx.ip->ip_dst;
382 blocked = imo_multi_filter(inp->inp_moptions,
384 (struct sockaddr *)&group,
385 (struct sockaddr *)&ripsrc);
388 if (blocked != MCAST_PASS) {
389 IPSTAT_INC(ips_notmember);
393 appended += rip_append(inp, ctx.ip, m, &ripsrc);
396 inetsw[ip_protox[ctx.ip->ip_p]].pr_input == rip_input) {
397 IPSTAT_INC(ips_noproto);
398 IPSTAT_DEC(ips_delivered);
399 icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_PROTOCOL, 0, 0);
402 return (IPPROTO_DONE);
406 * Generate IP header and pass packet to ip_output. Tack on options user may
407 * have setup with control call.
410 rip_output(struct mbuf *m, struct socket *so, ...)
412 struct epoch_tracker et;
415 struct inpcb *inp = sotoinpcb(so);
418 int flags = ((so->so_options & SO_DONTROUTE) ? IP_ROUTETOIF : 0) |
421 u_char opttype, optlen, *cp;
424 dst = va_arg(ap, u_long);
428 * If the user handed us a complete IP packet, use it. Otherwise,
429 * allocate an mbuf for a header and fill it in.
431 if ((inp->inp_flags & INP_HDRINCL) == 0) {
432 if (m->m_pkthdr.len + sizeof(struct ip) > IP_MAXPACKET) {
436 M_PREPEND(m, sizeof(struct ip), M_NOWAIT);
441 ip = mtod(m, struct ip *);
442 ip->ip_tos = inp->inp_ip_tos;
443 if (inp->inp_flags & INP_DONTFRAG)
444 ip->ip_off = htons(IP_DF);
446 ip->ip_off = htons(0);
447 ip->ip_p = inp->inp_ip_p;
448 ip->ip_len = htons(m->m_pkthdr.len);
449 ip->ip_src = inp->inp_laddr;
450 ip->ip_dst.s_addr = dst;
452 if (CALC_FLOWID_OUTBOUND) {
453 uint32_t hash_type, hash_val;
455 hash_val = fib4_calc_software_hash(ip->ip_src,
456 ip->ip_dst, 0, 0, ip->ip_p, &hash_type);
457 m->m_pkthdr.flowid = hash_val;
458 M_HASHTYPE_SET(m, hash_type);
459 flags |= IP_NODEFAULTFLOWID;
462 if (jailed(inp->inp_cred)) {
464 * prison_local_ip4() would be good enough but would
465 * let a source of INADDR_ANY pass, which we do not
466 * want to see from jails.
468 if (ip->ip_src.s_addr == INADDR_ANY) {
470 error = in_pcbladdr(inp, &ip->ip_dst,
471 &ip->ip_src, inp->inp_cred);
474 error = prison_local_ip4(inp->inp_cred,
483 ip->ip_ttl = inp->inp_ip_ttl;
485 if (m->m_pkthdr.len > IP_MAXPACKET) {
489 if (m->m_pkthdr.len < sizeof(*ip)) {
493 m = m_pullup(m, sizeof(*ip));
496 ip = mtod(m, struct ip *);
497 hlen = ip->ip_hl << 2;
498 if (m->m_len < hlen) {
499 m = m_pullup(m, hlen);
502 ip = mtod(m, struct ip *);
505 if (CALC_FLOWID_OUTBOUND) {
506 uint32_t hash_type, hash_val;
508 hash_val = fib4_calc_software_hash(ip->ip_dst,
509 ip->ip_src, 0, 0, ip->ip_p, &hash_type);
510 m->m_pkthdr.flowid = hash_val;
511 M_HASHTYPE_SET(m, hash_type);
512 flags |= IP_NODEFAULTFLOWID;
517 * Don't allow both user specified and setsockopt options,
518 * and don't allow packet length sizes that will crash.
520 if ((hlen < sizeof (*ip))
521 || ((hlen > sizeof (*ip)) && inp->inp_options)
522 || (ntohs(ip->ip_len) != m->m_pkthdr.len)) {
527 error = prison_check_ip4(inp->inp_cred, &ip->ip_src);
534 * Don't allow IP options which do not have the required
535 * structure as specified in section 3.1 of RFC 791 on
538 cp = (u_char *)(ip + 1);
539 cnt = hlen - sizeof (struct ip);
540 for (; cnt > 0; cnt -= optlen, cp += optlen) {
541 opttype = cp[IPOPT_OPTVAL];
542 if (opttype == IPOPT_EOL)
544 if (opttype == IPOPT_NOP) {
548 if (cnt < IPOPT_OLEN + sizeof(u_char)) {
553 optlen = cp[IPOPT_OLEN];
554 if (optlen < IPOPT_OLEN + sizeof(u_char) ||
562 * This doesn't allow application to specify ID of zero,
563 * but we got this limitation from the beginning of history.
569 * XXX prevent ip_output from overwriting header fields.
571 flags |= IP_RAWOUTPUT;
572 IPSTAT_INC(ips_rawout);
575 if (inp->inp_flags & INP_ONESBCAST)
576 flags |= IP_SENDONES;
579 mac_inpcb_create_mbuf(inp, m);
583 error = ip_output(m, inp->inp_options, NULL, flags,
584 inp->inp_moptions, inp);
591 * Raw IP socket option processing.
593 * IMPORTANT NOTE regarding access control: Traditionally, raw sockets could
594 * only be created by a privileged process, and as such, socket option
595 * operations to manage system properties on any raw socket were allowed to
596 * take place without explicit additional access control checks. However,
597 * raw sockets can now also be created in jail(), and therefore explicit
598 * checks are now required. Likewise, raw sockets can be used by a process
599 * after it gives up privilege, so some caution is required. For options
600 * passed down to the IP layer via ip_ctloutput(), checks are assumed to be
601 * performed in ip_ctloutput() and therefore no check occurs here.
602 * Unilaterally checking priv_check() here breaks normal IP socket option
603 * operations on raw sockets.
605 * When adding new socket options here, make sure to add access control
606 * checks here as necessary.
608 * XXX-BZ inp locking?
611 rip_ctloutput(struct socket *so, struct sockopt *sopt)
613 struct inpcb *inp = sotoinpcb(so);
616 if (sopt->sopt_level != IPPROTO_IP) {
617 if ((sopt->sopt_level == SOL_SOCKET) &&
618 (sopt->sopt_name == SO_SETFIB)) {
619 inp->inp_inc.inc_fibnum = so->so_fibnum;
626 switch (sopt->sopt_dir) {
628 switch (sopt->sopt_name) {
630 optval = inp->inp_flags & INP_HDRINCL;
631 error = sooptcopyout(sopt, &optval, sizeof optval);
634 case IP_FW3: /* generic ipfw v.3 functions */
635 case IP_FW_ADD: /* ADD actually returns the body... */
637 case IP_FW_TABLE_GETSIZE:
638 case IP_FW_TABLE_LIST:
639 case IP_FW_NAT_GET_CONFIG:
640 case IP_FW_NAT_GET_LOG:
641 if (V_ip_fw_ctl_ptr != NULL)
642 error = V_ip_fw_ctl_ptr(sopt);
647 case IP_DUMMYNET3: /* generic dummynet v.3 functions */
648 case IP_DUMMYNET_GET:
649 if (ip_dn_ctl_ptr != NULL)
650 error = ip_dn_ctl_ptr(sopt);
663 case MRT_API_SUPPORT:
665 case MRT_ADD_BW_UPCALL:
666 case MRT_DEL_BW_UPCALL:
667 error = priv_check(curthread, PRIV_NETINET_MROUTE);
670 error = ip_mrouter_get ? ip_mrouter_get(so, sopt) :
675 error = ip_ctloutput(so, sopt);
681 switch (sopt->sopt_name) {
683 error = sooptcopyin(sopt, &optval, sizeof optval,
688 inp->inp_flags |= INP_HDRINCL;
690 inp->inp_flags &= ~INP_HDRINCL;
693 case IP_FW3: /* generic ipfw v.3 functions */
699 case IP_FW_TABLE_ADD:
700 case IP_FW_TABLE_DEL:
701 case IP_FW_TABLE_FLUSH:
704 if (V_ip_fw_ctl_ptr != NULL)
705 error = V_ip_fw_ctl_ptr(sopt);
710 case IP_DUMMYNET3: /* generic dummynet v.3 functions */
711 case IP_DUMMYNET_CONFIGURE:
712 case IP_DUMMYNET_DEL:
713 case IP_DUMMYNET_FLUSH:
714 if (ip_dn_ctl_ptr != NULL)
715 error = ip_dn_ctl_ptr(sopt);
717 error = ENOPROTOOPT ;
721 error = priv_check(curthread, PRIV_NETINET_MROUTE);
724 error = ip_rsvp_init(so);
728 error = priv_check(curthread, PRIV_NETINET_MROUTE);
731 error = ip_rsvp_done();
735 case IP_RSVP_VIF_OFF:
736 error = priv_check(curthread, PRIV_NETINET_MROUTE);
739 error = ip_rsvp_vif ?
740 ip_rsvp_vif(so, sopt) : EINVAL;
751 case MRT_API_SUPPORT:
753 case MRT_ADD_BW_UPCALL:
754 case MRT_DEL_BW_UPCALL:
755 error = priv_check(curthread, PRIV_NETINET_MROUTE);
758 error = ip_mrouter_set ? ip_mrouter_set(so, sopt) :
763 error = ip_ctloutput(so, sopt);
773 * This function exists solely to receive the PRC_IFDOWN messages which are
774 * sent by if_down(). It looks for an ifaddr whose ifa_addr is sa, and calls
775 * in_ifadown() to remove all routes corresponding to that address. It also
776 * receives the PRC_IFUP messages from if_up() and reinstalls the interface
780 rip_ctlinput(int cmd, struct sockaddr *sa, void *vip)
782 struct in_ifaddr *ia;
789 CK_STAILQ_FOREACH(ia, &V_in_ifaddrhead, ia_link) {
790 if (ia->ia_ifa.ifa_addr == sa
791 && (ia->ia_flags & IFA_ROUTE)) {
792 ifa_ref(&ia->ia_ifa);
794 * in_scrubprefix() kills the interface route.
796 in_scrubprefix(ia, 0);
798 * in_ifadown gets rid of all the rest of the
799 * routes. This is not quite the right thing
800 * to do, but at least if we are running a
801 * routing process they will come back.
803 in_ifadown(&ia->ia_ifa, 0);
804 ifa_free(&ia->ia_ifa);
811 CK_STAILQ_FOREACH(ia, &V_in_ifaddrhead, ia_link) {
812 if (ia->ia_ifa.ifa_addr == sa)
815 if (ia == NULL || (ia->ia_flags & IFA_ROUTE))
817 ifa_ref(&ia->ia_ifa);
819 err = ifa_del_loopback_route((struct ifaddr *)ia, sa);
821 rt_addrmsg(RTM_ADD, &ia->ia_ifa, ia->ia_ifp->if_fib);
822 err = in_handle_ifaddr_route(RTM_ADD, ia);
824 ia->ia_flags |= IFA_ROUTE;
826 err = ifa_add_loopback_route((struct ifaddr *)ia, sa);
828 ifa_free(&ia->ia_ifa);
830 #if defined(IPSEC) || defined(IPSEC_SUPPORT)
832 if (IPSEC_ENABLED(ipv4))
833 IPSEC_CTLINPUT(ipv4, cmd, sa, vip);
840 rip_attach(struct socket *so, int proto, struct thread *td)
846 KASSERT(inp == NULL, ("rip_attach: inp != NULL"));
848 error = priv_check(td, PRIV_NETINET_RAW);
851 if (proto >= IPPROTO_MAX || proto < 0)
852 return EPROTONOSUPPORT;
853 error = soreserve(so, rip_sendspace, rip_recvspace);
856 error = in_pcballoc(so, &V_ripcbinfo);
859 inp = (struct inpcb *)so->so_pcb;
860 inp->inp_vflag |= INP_IPV4;
861 inp->inp_ip_p = proto;
862 inp->inp_ip_ttl = V_ip_defttl;
863 INP_HASH_WLOCK(&V_ripcbinfo);
865 INP_HASH_WUNLOCK(&V_ripcbinfo);
871 rip_detach(struct socket *so)
876 KASSERT(inp != NULL, ("rip_detach: inp == NULL"));
877 KASSERT(inp->inp_faddr.s_addr == INADDR_ANY,
878 ("rip_detach: not closed"));
880 /* Disable mrouter first */
881 if (so == V_ip_mrouter && ip_mrouter_done)
885 INP_HASH_WLOCK(&V_ripcbinfo);
887 INP_HASH_WUNLOCK(&V_ripcbinfo);
889 if (ip_rsvp_force_done)
890 ip_rsvp_force_done(so);
891 if (so == V_ip_rsvpd)
898 rip_dodisconnect(struct socket *so, struct inpcb *inp)
900 struct inpcbinfo *pcbinfo;
902 pcbinfo = inp->inp_pcbinfo;
904 INP_HASH_WLOCK(pcbinfo);
906 inp->inp_faddr.s_addr = INADDR_ANY;
908 INP_HASH_WUNLOCK(pcbinfo);
910 so->so_state &= ~SS_ISCONNECTED;
916 rip_abort(struct socket *so)
921 KASSERT(inp != NULL, ("rip_abort: inp == NULL"));
923 rip_dodisconnect(so, inp);
927 rip_close(struct socket *so)
932 KASSERT(inp != NULL, ("rip_close: inp == NULL"));
934 rip_dodisconnect(so, inp);
938 rip_disconnect(struct socket *so)
942 if ((so->so_state & SS_ISCONNECTED) == 0)
946 KASSERT(inp != NULL, ("rip_disconnect: inp == NULL"));
948 rip_dodisconnect(so, inp);
953 rip_bind(struct socket *so, struct sockaddr *nam, struct thread *td)
955 struct sockaddr_in *addr = (struct sockaddr_in *)nam;
959 if (nam->sa_family != AF_INET)
960 return (EAFNOSUPPORT);
961 if (nam->sa_len != sizeof(*addr))
964 error = prison_check_ip4(td->td_ucred, &addr->sin_addr);
969 KASSERT(inp != NULL, ("rip_bind: inp == NULL"));
971 if (CK_STAILQ_EMPTY(&V_ifnet) ||
972 (addr->sin_family != AF_INET && addr->sin_family != AF_IMPLINK) ||
973 (addr->sin_addr.s_addr &&
974 (inp->inp_flags & INP_BINDANY) == 0 &&
975 ifa_ifwithaddr_check((struct sockaddr *)addr) == 0))
976 return (EADDRNOTAVAIL);
979 INP_HASH_WLOCK(&V_ripcbinfo);
981 inp->inp_laddr = addr->sin_addr;
983 INP_HASH_WUNLOCK(&V_ripcbinfo);
989 rip_connect(struct socket *so, struct sockaddr *nam, struct thread *td)
991 struct sockaddr_in *addr = (struct sockaddr_in *)nam;
994 if (nam->sa_len != sizeof(*addr))
996 if (CK_STAILQ_EMPTY(&V_ifnet))
997 return (EADDRNOTAVAIL);
998 if (addr->sin_family != AF_INET && addr->sin_family != AF_IMPLINK)
999 return (EAFNOSUPPORT);
1001 inp = sotoinpcb(so);
1002 KASSERT(inp != NULL, ("rip_connect: inp == NULL"));
1005 INP_HASH_WLOCK(&V_ripcbinfo);
1007 inp->inp_faddr = addr->sin_addr;
1009 INP_HASH_WUNLOCK(&V_ripcbinfo);
1016 rip_shutdown(struct socket *so)
1020 inp = sotoinpcb(so);
1021 KASSERT(inp != NULL, ("rip_shutdown: inp == NULL"));
1030 rip_send(struct socket *so, int flags, struct mbuf *m, struct sockaddr *nam,
1031 struct mbuf *control, struct thread *td)
1037 inp = sotoinpcb(so);
1038 KASSERT(inp != NULL, ("rip_send: inp == NULL"));
1040 if (control != NULL) {
1046 * Note: 'dst' reads below are unlocked.
1048 if (so->so_state & SS_ISCONNECTED) {
1053 dst = inp->inp_faddr.s_addr; /* Unlocked read. */
1058 else if (nam->sa_family != AF_INET)
1059 error = EAFNOSUPPORT;
1060 else if (nam->sa_len != sizeof(struct sockaddr_in))
1064 dst = ((struct sockaddr_in *)nam)->sin_addr.s_addr;
1066 return (rip_output(m, so, dst));
1075 rip_pcblist(SYSCTL_HANDLER_ARGS)
1077 struct inpcb_iterator inpi = INP_ALL_ITERATOR(&V_ripcbinfo,
1078 INPLOOKUP_RLOCKPCB);
1083 if (req->newptr != 0)
1086 if (req->oldptr == 0) {
1089 n = V_ripcbinfo.ipi_count;
1090 n += imax(n / 8, 10);
1091 req->oldidx = 2 * (sizeof xig) + n * sizeof(struct xinpcb);
1095 if ((error = sysctl_wire_old_buffer(req, 0)) != 0)
1098 bzero(&xig, sizeof(xig));
1099 xig.xig_len = sizeof xig;
1100 xig.xig_count = V_ripcbinfo.ipi_count;
1101 xig.xig_gen = V_ripcbinfo.ipi_gencnt;
1102 xig.xig_sogen = so_gencnt;
1103 error = SYSCTL_OUT(req, &xig, sizeof xig);
1107 while ((inp = inp_next(&inpi)) != NULL) {
1108 if (inp->inp_gencnt <= xig.xig_gen &&
1109 cr_canseeinpcb(req->td->td_ucred, inp) == 0) {
1112 in_pcbtoxinpcb(inp, &xi);
1113 error = SYSCTL_OUT(req, &xi, sizeof xi);
1123 * Give the user an updated idea of our state. If the
1124 * generation differs from what we told her before, she knows
1125 * that something happened while we were processing this
1126 * request, and it might be necessary to retry.
1128 xig.xig_gen = V_ripcbinfo.ipi_gencnt;
1129 xig.xig_sogen = so_gencnt;
1130 xig.xig_count = V_ripcbinfo.ipi_count;
1131 error = SYSCTL_OUT(req, &xig, sizeof xig);
1137 SYSCTL_PROC(_net_inet_raw, OID_AUTO/*XXX*/, pcblist,
1138 CTLTYPE_OPAQUE | CTLFLAG_RD | CTLFLAG_MPSAFE, NULL, 0,
1139 rip_pcblist, "S,xinpcb",
1140 "List of active raw IP sockets");
1143 struct pr_usrreqs rip_usrreqs = {
1144 .pru_abort = rip_abort,
1145 .pru_attach = rip_attach,
1146 .pru_bind = rip_bind,
1147 .pru_connect = rip_connect,
1148 .pru_control = in_control,
1149 .pru_detach = rip_detach,
1150 .pru_disconnect = rip_disconnect,
1151 .pru_peeraddr = in_getpeeraddr,
1152 .pru_send = rip_send,
1153 .pru_shutdown = rip_shutdown,
1154 .pru_sockaddr = in_getsockaddr,
1155 .pru_sosetlabel = in_pcbsosetlabel,
1156 .pru_close = rip_close,