2 * Copyright (c) 1982, 1986, 1988, 1993
3 * The Regents of the University of California.
6 * Redistribution and use in source and binary forms, with or without
7 * modification, are permitted provided that the following conditions
9 * 1. Redistributions of source code must retain the above copyright
10 * notice, this list of conditions and the following disclaimer.
11 * 2. Redistributions in binary form must reproduce the above copyright
12 * notice, this list of conditions and the following disclaimer in the
13 * documentation and/or other materials provided with the distribution.
14 * 4. Neither the name of the University nor the names of its contributors
15 * may be used to endorse or promote products derived from this software
16 * without specific prior written permission.
18 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
19 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
20 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
21 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
22 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
23 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
24 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
25 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
26 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
27 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
30 * @(#)raw_ip.c 8.7 (Berkeley) 5/15/95
33 #include <sys/cdefs.h>
34 __FBSDID("$FreeBSD$");
37 #include "opt_inet6.h"
38 #include "opt_ipsec.h"
40 #include <sys/param.h>
42 #include <sys/kernel.h>
44 #include <sys/malloc.h>
48 #include <sys/protosw.h>
49 #include <sys/rwlock.h>
50 #include <sys/signalvar.h>
51 #include <sys/socket.h>
52 #include <sys/socketvar.h>
54 #include <sys/sysctl.h>
55 #include <sys/systm.h>
60 #include <net/if_var.h>
61 #include <net/route.h>
64 #include <netinet/in.h>
65 #include <netinet/in_systm.h>
66 #include <netinet/in_pcb.h>
67 #include <netinet/in_var.h>
68 #include <netinet/if_ether.h>
69 #include <netinet/ip.h>
70 #include <netinet/ip_var.h>
71 #include <netinet/ip_mroute.h>
74 #include <netipsec/ipsec.h>
77 #include <security/mac/mac_framework.h>
79 VNET_DEFINE(int, ip_defttl) = IPDEFTTL;
80 SYSCTL_VNET_INT(_net_inet_ip, IPCTL_DEFTTL, ttl, CTLFLAG_RW,
81 &VNET_NAME(ip_defttl), 0,
82 "Maximum TTL on IP packets");
84 VNET_DEFINE(struct inpcbhead, ripcb);
85 VNET_DEFINE(struct inpcbinfo, ripcbinfo);
87 #define V_ripcb VNET(ripcb)
88 #define V_ripcbinfo VNET(ripcbinfo)
91 * Control and data hooks for ipfw, dummynet, divert and so on.
92 * The data hooks are not used here but it is convenient
93 * to keep them all in one place.
95 VNET_DEFINE(ip_fw_chk_ptr_t, ip_fw_chk_ptr) = NULL;
96 VNET_DEFINE(ip_fw_ctl_ptr_t, ip_fw_ctl_ptr) = NULL;
98 int (*ip_dn_ctl_ptr)(struct sockopt *);
99 int (*ip_dn_io_ptr)(struct mbuf **, int, struct ip_fw_args *);
100 void (*ip_divert_ptr)(struct mbuf *, int);
101 int (*ng_ipfw_input_p)(struct mbuf **, int,
102 struct ip_fw_args *, int);
106 * Hooks for multicast routing. They all default to NULL, so leave them not
107 * initialized and rely on BSS being set to 0.
111 * The socket used to communicate with the multicast routing daemon.
113 VNET_DEFINE(struct socket *, ip_mrouter);
116 * The various mrouter and rsvp functions.
118 int (*ip_mrouter_set)(struct socket *, struct sockopt *);
119 int (*ip_mrouter_get)(struct socket *, struct sockopt *);
120 int (*ip_mrouter_done)(void);
121 int (*ip_mforward)(struct ip *, struct ifnet *, struct mbuf *,
122 struct ip_moptions *);
123 int (*mrt_ioctl)(u_long, caddr_t, int);
124 int (*legal_vif_num)(int);
125 u_long (*ip_mcast_src)(int);
127 int (*rsvp_input_p)(struct mbuf **, int *, int);
128 int (*ip_rsvp_vif)(struct socket *, struct sockopt *);
129 void (*ip_rsvp_force_done)(struct socket *);
132 u_long rip_sendspace = 9216;
133 SYSCTL_ULONG(_net_inet_raw, OID_AUTO, maxdgram, CTLFLAG_RW,
134 &rip_sendspace, 0, "Maximum outgoing raw IP datagram size");
136 u_long rip_recvspace = 9216;
137 SYSCTL_ULONG(_net_inet_raw, OID_AUTO, recvspace, CTLFLAG_RW,
138 &rip_recvspace, 0, "Maximum space for incoming raw IP datagrams");
144 #define INP_PCBHASH_RAW_SIZE 256
145 #define INP_PCBHASH_RAW(proto, laddr, faddr, mask) \
146 (((proto) + (laddr) + (faddr)) % (mask) + 1)
150 rip_inshash(struct inpcb *inp)
152 struct inpcbinfo *pcbinfo = inp->inp_pcbinfo;
153 struct inpcbhead *pcbhash;
156 INP_INFO_WLOCK_ASSERT(pcbinfo);
157 INP_WLOCK_ASSERT(inp);
159 if (inp->inp_ip_p != 0 &&
160 inp->inp_laddr.s_addr != INADDR_ANY &&
161 inp->inp_faddr.s_addr != INADDR_ANY) {
162 hash = INP_PCBHASH_RAW(inp->inp_ip_p, inp->inp_laddr.s_addr,
163 inp->inp_faddr.s_addr, pcbinfo->ipi_hashmask);
166 pcbhash = &pcbinfo->ipi_hashbase[hash];
167 LIST_INSERT_HEAD(pcbhash, inp, inp_hash);
171 rip_delhash(struct inpcb *inp)
174 INP_INFO_WLOCK_ASSERT(inp->inp_pcbinfo);
175 INP_WLOCK_ASSERT(inp);
177 LIST_REMOVE(inp, inp_hash);
182 * Raw interface to IP protocol.
186 * Initialize raw connection block q.
189 rip_zone_change(void *tag)
192 uma_zone_set_max(V_ripcbinfo.ipi_zone, maxsockets);
196 rip_inpcb_init(void *mem, int size, int flags)
198 struct inpcb *inp = mem;
200 INP_LOCK_INIT(inp, "inp", "rawinp");
208 in_pcbinfo_init(&V_ripcbinfo, "rip", &V_ripcb, INP_PCBHASH_RAW_SIZE,
209 1, "ripcb", rip_inpcb_init, NULL, UMA_ZONE_NOFREE,
210 IPI_HASHFIELDS_NONE);
211 EVENTHANDLER_REGISTER(maxsockets_change, rip_zone_change, NULL,
212 EVENTHANDLER_PRI_ANY);
220 in_pcbinfo_destroy(&V_ripcbinfo);
226 rip_append(struct inpcb *last, struct ip *ip, struct mbuf *n,
227 struct sockaddr_in *ripsrc)
231 INP_LOCK_ASSERT(last);
234 /* check AH/ESP integrity. */
235 if (ipsec4_in_reject(n, last)) {
240 if (!policyfail && mac_inpcb_check_deliver(last, n) != 0)
243 /* Check the minimum TTL for socket. */
244 if (last->inp_ip_minttl && last->inp_ip_minttl > ip->ip_ttl)
247 struct mbuf *opts = NULL;
250 so = last->inp_socket;
251 if ((last->inp_flags & INP_CONTROLOPTS) ||
252 (so->so_options & (SO_TIMESTAMP | SO_BINTIME)))
253 ip_savecontrol(last, &opts, ip, n);
254 SOCKBUF_LOCK(&so->so_rcv);
255 if (sbappendaddr_locked(&so->so_rcv,
256 (struct sockaddr *)ripsrc, n, opts) == 0) {
257 /* should notify about lost packet */
261 SOCKBUF_UNLOCK(&so->so_rcv);
263 sorwakeup_locked(so);
270 * Setup generic address and protocol structures for raw_input routine, then
271 * pass them along with mbuf chain.
274 rip_input(struct mbuf **mp, int *offp, int proto)
277 struct mbuf *m = *mp;
278 struct ip *ip = mtod(m, struct ip *);
279 struct inpcb *inp, *last;
280 struct sockaddr_in ripsrc;
285 bzero(&ripsrc, sizeof(ripsrc));
286 ripsrc.sin_len = sizeof(ripsrc);
287 ripsrc.sin_family = AF_INET;
288 ripsrc.sin_addr = ip->ip_src;
291 ifp = m->m_pkthdr.rcvif;
293 * Applications on raw sockets expect host byte order.
295 ip->ip_len = ntohs(ip->ip_len);
296 ip->ip_off = ntohs(ip->ip_off);
298 hash = INP_PCBHASH_RAW(proto, ip->ip_src.s_addr,
299 ip->ip_dst.s_addr, V_ripcbinfo.ipi_hashmask);
300 INP_INFO_RLOCK(&V_ripcbinfo);
301 LIST_FOREACH(inp, &V_ripcbinfo.ipi_hashbase[hash], inp_hash) {
302 if (inp->inp_ip_p != proto)
305 /* XXX inp locking */
306 if ((inp->inp_vflag & INP_IPV4) == 0)
309 if (inp->inp_laddr.s_addr != ip->ip_dst.s_addr)
311 if (inp->inp_faddr.s_addr != ip->ip_src.s_addr)
313 if (jailed_without_vnet(inp->inp_cred)) {
315 * XXX: If faddr was bound to multicast group,
316 * jailed raw socket will drop datagram.
318 if (prison_check_ip4(inp->inp_cred, &ip->ip_dst) != 0)
324 n = m_copy(m, 0, (int)M_COPYALL);
326 (void) rip_append(last, ip, n, &ripsrc);
327 /* XXX count dropped packet */
333 LIST_FOREACH(inp, &V_ripcbinfo.ipi_hashbase[0], inp_hash) {
334 if (inp->inp_ip_p && inp->inp_ip_p != proto)
337 /* XXX inp locking */
338 if ((inp->inp_vflag & INP_IPV4) == 0)
341 if (!in_nullhost(inp->inp_laddr) &&
342 !in_hosteq(inp->inp_laddr, ip->ip_dst))
344 if (!in_nullhost(inp->inp_faddr) &&
345 !in_hosteq(inp->inp_faddr, ip->ip_src))
347 if (jailed_without_vnet(inp->inp_cred)) {
349 * Allow raw socket in jail to receive multicast;
350 * assume process had PRIV_NETINET_RAW at attach,
351 * and fall through into normal filter path if so.
353 if (!IN_MULTICAST(ntohl(ip->ip_dst.s_addr)) &&
354 prison_check_ip4(inp->inp_cred, &ip->ip_dst) != 0)
358 * If this raw socket has multicast state, and we
359 * have received a multicast, check if this socket
360 * should receive it, as multicast filtering is now
361 * the responsibility of the transport layer.
363 if (inp->inp_moptions != NULL &&
364 IN_MULTICAST(ntohl(ip->ip_dst.s_addr))) {
366 * If the incoming datagram is for IGMP, allow it
367 * through unconditionally to the raw socket.
369 * In the case of IGMPv2, we may not have explicitly
370 * joined the group, and may have set IFF_ALLMULTI
371 * on the interface. imo_multi_filter() may discard
372 * control traffic we actually need to see.
374 * Userland multicast routing daemons should continue
375 * filter the control traffic appropriately.
379 blocked = MCAST_PASS;
380 if (proto != IPPROTO_IGMP) {
381 struct sockaddr_in group;
383 bzero(&group, sizeof(struct sockaddr_in));
384 group.sin_len = sizeof(struct sockaddr_in);
385 group.sin_family = AF_INET;
386 group.sin_addr = ip->ip_dst;
388 blocked = imo_multi_filter(inp->inp_moptions,
390 (struct sockaddr *)&group,
391 (struct sockaddr *)&ripsrc);
394 if (blocked != MCAST_PASS) {
395 IPSTAT_INC(ips_notmember);
402 n = m_copy(m, 0, (int)M_COPYALL);
404 (void) rip_append(last, ip, n, &ripsrc);
405 /* XXX count dropped packet */
411 INP_INFO_RUNLOCK(&V_ripcbinfo);
413 if (rip_append(last, ip, m, &ripsrc) != 0)
414 IPSTAT_INC(ips_delivered);
418 IPSTAT_INC(ips_noproto);
419 IPSTAT_DEC(ips_delivered);
421 return (IPPROTO_DONE);
425 * Generate IP header and pass packet to ip_output. Tack on options user may
426 * have setup with control call.
429 rip_output(struct mbuf *m, struct socket *so, u_long dst)
433 struct inpcb *inp = sotoinpcb(so);
434 int flags = ((so->so_options & SO_DONTROUTE) ? IP_ROUTETOIF : 0) |
438 * If the user handed us a complete IP packet, use it. Otherwise,
439 * allocate an mbuf for a header and fill it in.
441 if ((inp->inp_flags & INP_HDRINCL) == 0) {
442 if (m->m_pkthdr.len + sizeof(struct ip) > IP_MAXPACKET) {
446 M_PREPEND(m, sizeof(struct ip), M_NOWAIT);
451 ip = mtod(m, struct ip *);
452 ip->ip_tos = inp->inp_ip_tos;
453 if (inp->inp_flags & INP_DONTFRAG)
454 ip->ip_off = htons(IP_DF);
456 ip->ip_off = htons(0);
457 ip->ip_p = inp->inp_ip_p;
458 ip->ip_len = htons(m->m_pkthdr.len);
459 ip->ip_src = inp->inp_laddr;
460 ip->ip_dst.s_addr = dst;
461 if (jailed(inp->inp_cred)) {
463 * prison_local_ip4() would be good enough but would
464 * let a source of INADDR_ANY pass, which we do not
465 * want to see from jails.
467 if (ip->ip_src.s_addr == INADDR_ANY) {
468 error = in_pcbladdr(inp, &ip->ip_dst, &ip->ip_src,
471 error = prison_local_ip4(inp->inp_cred,
480 ip->ip_ttl = inp->inp_ip_ttl;
482 if (m->m_pkthdr.len > IP_MAXPACKET) {
487 ip = mtod(m, struct ip *);
488 error = prison_check_ip4(inp->inp_cred, &ip->ip_src);
496 * Don't allow both user specified and setsockopt options,
497 * and don't allow packet length sizes that will crash.
499 if (((ip->ip_hl != (sizeof (*ip) >> 2)) && inp->inp_options)
500 || (ip->ip_len > m->m_pkthdr.len)
501 || (ip->ip_len < (ip->ip_hl << 2))) {
507 ip->ip_id = ip_newid();
510 * Applications on raw sockets pass us packets
511 * in host byte order.
513 ip->ip_len = htons(ip->ip_len);
514 ip->ip_off = htons(ip->ip_off);
517 * XXX prevent ip_output from overwriting header fields.
519 flags |= IP_RAWOUTPUT;
520 IPSTAT_INC(ips_rawout);
523 if (inp->inp_flags & INP_ONESBCAST)
524 flags |= IP_SENDONES;
527 mac_inpcb_create_mbuf(inp, m);
530 error = ip_output(m, inp->inp_options, NULL, flags,
531 inp->inp_moptions, inp);
537 * Raw IP socket option processing.
539 * IMPORTANT NOTE regarding access control: Traditionally, raw sockets could
540 * only be created by a privileged process, and as such, socket option
541 * operations to manage system properties on any raw socket were allowed to
542 * take place without explicit additional access control checks. However,
543 * raw sockets can now also be created in jail(), and therefore explicit
544 * checks are now required. Likewise, raw sockets can be used by a process
545 * after it gives up privilege, so some caution is required. For options
546 * passed down to the IP layer via ip_ctloutput(), checks are assumed to be
547 * performed in ip_ctloutput() and therefore no check occurs here.
548 * Unilaterally checking priv_check() here breaks normal IP socket option
549 * operations on raw sockets.
551 * When adding new socket options here, make sure to add access control
552 * checks here as necessary.
554 * XXX-BZ inp locking?
557 rip_ctloutput(struct socket *so, struct sockopt *sopt)
559 struct inpcb *inp = sotoinpcb(so);
562 if (sopt->sopt_level != IPPROTO_IP) {
563 if ((sopt->sopt_level == SOL_SOCKET) &&
564 (sopt->sopt_name == SO_SETFIB)) {
565 inp->inp_inc.inc_fibnum = so->so_fibnum;
572 switch (sopt->sopt_dir) {
574 switch (sopt->sopt_name) {
576 optval = inp->inp_flags & INP_HDRINCL;
577 error = sooptcopyout(sopt, &optval, sizeof optval);
580 case IP_FW3: /* generic ipfw v.3 functions */
581 case IP_FW_ADD: /* ADD actually returns the body... */
583 case IP_FW_TABLE_GETSIZE:
584 case IP_FW_TABLE_LIST:
585 case IP_FW_NAT_GET_CONFIG:
586 case IP_FW_NAT_GET_LOG:
587 if (V_ip_fw_ctl_ptr != NULL)
588 error = V_ip_fw_ctl_ptr(sopt);
593 case IP_DUMMYNET3: /* generic dummynet v.3 functions */
594 case IP_DUMMYNET_GET:
595 if (ip_dn_ctl_ptr != NULL)
596 error = ip_dn_ctl_ptr(sopt);
609 case MRT_API_SUPPORT:
611 case MRT_ADD_BW_UPCALL:
612 case MRT_DEL_BW_UPCALL:
613 error = priv_check(curthread, PRIV_NETINET_MROUTE);
616 error = ip_mrouter_get ? ip_mrouter_get(so, sopt) :
621 error = ip_ctloutput(so, sopt);
627 switch (sopt->sopt_name) {
629 error = sooptcopyin(sopt, &optval, sizeof optval,
634 inp->inp_flags |= INP_HDRINCL;
636 inp->inp_flags &= ~INP_HDRINCL;
639 case IP_FW3: /* generic ipfw v.3 functions */
645 case IP_FW_TABLE_ADD:
646 case IP_FW_TABLE_DEL:
647 case IP_FW_TABLE_FLUSH:
650 if (V_ip_fw_ctl_ptr != NULL)
651 error = V_ip_fw_ctl_ptr(sopt);
656 case IP_DUMMYNET3: /* generic dummynet v.3 functions */
657 case IP_DUMMYNET_CONFIGURE:
658 case IP_DUMMYNET_DEL:
659 case IP_DUMMYNET_FLUSH:
660 if (ip_dn_ctl_ptr != NULL)
661 error = ip_dn_ctl_ptr(sopt);
663 error = ENOPROTOOPT ;
667 error = priv_check(curthread, PRIV_NETINET_MROUTE);
670 error = ip_rsvp_init(so);
674 error = priv_check(curthread, PRIV_NETINET_MROUTE);
677 error = ip_rsvp_done();
681 case IP_RSVP_VIF_OFF:
682 error = priv_check(curthread, PRIV_NETINET_MROUTE);
685 error = ip_rsvp_vif ?
686 ip_rsvp_vif(so, sopt) : EINVAL;
697 case MRT_API_SUPPORT:
699 case MRT_ADD_BW_UPCALL:
700 case MRT_DEL_BW_UPCALL:
701 error = priv_check(curthread, PRIV_NETINET_MROUTE);
704 error = ip_mrouter_set ? ip_mrouter_set(so, sopt) :
709 error = ip_ctloutput(so, sopt);
719 * This function exists solely to receive the PRC_IFDOWN messages which are
720 * sent by if_down(). It looks for an ifaddr whose ifa_addr is sa, and calls
721 * in_ifadown() to remove all routes corresponding to that address. It also
722 * receives the PRC_IFUP messages from if_up() and reinstalls the interface
726 rip_ctlinput(int cmd, struct sockaddr *sa, void *vip)
728 struct in_ifaddr *ia;
736 TAILQ_FOREACH(ia, &V_in_ifaddrhead, ia_link) {
737 if (ia->ia_ifa.ifa_addr == sa
738 && (ia->ia_flags & IFA_ROUTE)) {
739 ifa_ref(&ia->ia_ifa);
742 * in_scrubprefix() kills the interface route.
744 in_scrubprefix(ia, 0);
746 * in_ifadown gets rid of all the rest of the
747 * routes. This is not quite the right thing
748 * to do, but at least if we are running a
749 * routing process they will come back.
751 in_ifadown(&ia->ia_ifa, 0);
752 ifa_free(&ia->ia_ifa);
756 if (ia == NULL) /* If ia matched, already unlocked. */
762 TAILQ_FOREACH(ia, &V_in_ifaddrhead, ia_link) {
763 if (ia->ia_ifa.ifa_addr == sa)
766 if (ia == NULL || (ia->ia_flags & IFA_ROUTE)) {
770 ifa_ref(&ia->ia_ifa);
773 ifp = ia->ia_ifa.ifa_ifp;
775 if ((ifp->if_flags & IFF_LOOPBACK)
776 || (ifp->if_flags & IFF_POINTOPOINT))
779 err = ifa_del_loopback_route((struct ifaddr *)ia, sa);
781 err = rtinit(&ia->ia_ifa, RTM_ADD, flags);
783 ia->ia_flags |= IFA_ROUTE;
785 err = ifa_add_loopback_route((struct ifaddr *)ia, sa);
787 ifa_free(&ia->ia_ifa);
793 rip_attach(struct socket *so, int proto, struct thread *td)
799 KASSERT(inp == NULL, ("rip_attach: inp != NULL"));
801 error = priv_check(td, PRIV_NETINET_RAW);
804 if (proto >= IPPROTO_MAX || proto < 0)
805 return EPROTONOSUPPORT;
806 error = soreserve(so, rip_sendspace, rip_recvspace);
809 INP_INFO_WLOCK(&V_ripcbinfo);
810 error = in_pcballoc(so, &V_ripcbinfo);
812 INP_INFO_WUNLOCK(&V_ripcbinfo);
815 inp = (struct inpcb *)so->so_pcb;
816 inp->inp_vflag |= INP_IPV4;
817 inp->inp_ip_p = proto;
818 inp->inp_ip_ttl = V_ip_defttl;
820 INP_INFO_WUNLOCK(&V_ripcbinfo);
826 rip_detach(struct socket *so)
831 KASSERT(inp != NULL, ("rip_detach: inp == NULL"));
832 KASSERT(inp->inp_faddr.s_addr == INADDR_ANY,
833 ("rip_detach: not closed"));
835 INP_INFO_WLOCK(&V_ripcbinfo);
838 if (so == V_ip_mrouter && ip_mrouter_done)
840 if (ip_rsvp_force_done)
841 ip_rsvp_force_done(so);
842 if (so == V_ip_rsvpd)
846 INP_INFO_WUNLOCK(&V_ripcbinfo);
850 rip_dodisconnect(struct socket *so, struct inpcb *inp)
852 struct inpcbinfo *pcbinfo;
854 pcbinfo = inp->inp_pcbinfo;
855 INP_INFO_WLOCK(pcbinfo);
858 inp->inp_faddr.s_addr = INADDR_ANY;
861 so->so_state &= ~SS_ISCONNECTED;
864 INP_INFO_WUNLOCK(pcbinfo);
868 rip_abort(struct socket *so)
873 KASSERT(inp != NULL, ("rip_abort: inp == NULL"));
875 rip_dodisconnect(so, inp);
879 rip_close(struct socket *so)
884 KASSERT(inp != NULL, ("rip_close: inp == NULL"));
886 rip_dodisconnect(so, inp);
890 rip_disconnect(struct socket *so)
894 if ((so->so_state & SS_ISCONNECTED) == 0)
898 KASSERT(inp != NULL, ("rip_disconnect: inp == NULL"));
900 rip_dodisconnect(so, inp);
905 rip_bind(struct socket *so, struct sockaddr *nam, struct thread *td)
907 struct sockaddr_in *addr = (struct sockaddr_in *)nam;
911 if (nam->sa_len != sizeof(*addr))
914 error = prison_check_ip4(td->td_ucred, &addr->sin_addr);
919 KASSERT(inp != NULL, ("rip_bind: inp == NULL"));
921 if (TAILQ_EMPTY(&V_ifnet) ||
922 (addr->sin_family != AF_INET && addr->sin_family != AF_IMPLINK) ||
923 (addr->sin_addr.s_addr &&
924 (inp->inp_flags & INP_BINDANY) == 0 &&
925 ifa_ifwithaddr_check((struct sockaddr *)addr) == 0))
926 return (EADDRNOTAVAIL);
928 INP_INFO_WLOCK(&V_ripcbinfo);
931 inp->inp_laddr = addr->sin_addr;
934 INP_INFO_WUNLOCK(&V_ripcbinfo);
939 rip_connect(struct socket *so, struct sockaddr *nam, struct thread *td)
941 struct sockaddr_in *addr = (struct sockaddr_in *)nam;
944 if (nam->sa_len != sizeof(*addr))
946 if (TAILQ_EMPTY(&V_ifnet))
947 return (EADDRNOTAVAIL);
948 if (addr->sin_family != AF_INET && addr->sin_family != AF_IMPLINK)
949 return (EAFNOSUPPORT);
952 KASSERT(inp != NULL, ("rip_connect: inp == NULL"));
954 INP_INFO_WLOCK(&V_ripcbinfo);
957 inp->inp_faddr = addr->sin_addr;
961 INP_INFO_WUNLOCK(&V_ripcbinfo);
966 rip_shutdown(struct socket *so)
971 KASSERT(inp != NULL, ("rip_shutdown: inp == NULL"));
980 rip_send(struct socket *so, int flags, struct mbuf *m, struct sockaddr *nam,
981 struct mbuf *control, struct thread *td)
987 KASSERT(inp != NULL, ("rip_send: inp == NULL"));
990 * Note: 'dst' reads below are unlocked.
992 if (so->so_state & SS_ISCONNECTED) {
997 dst = inp->inp_faddr.s_addr; /* Unlocked read. */
1003 dst = ((struct sockaddr_in *)nam)->sin_addr.s_addr;
1005 return (rip_output(m, so, dst));
1010 rip_pcblist(SYSCTL_HANDLER_ARGS)
1013 struct inpcb *inp, **inp_list;
1018 * The process of preparing the TCB list is too time-consuming and
1019 * resource-intensive to repeat twice on every request.
1021 if (req->oldptr == 0) {
1022 n = V_ripcbinfo.ipi_count;
1023 n += imax(n / 8, 10);
1024 req->oldidx = 2 * (sizeof xig) + n * sizeof(struct xinpcb);
1028 if (req->newptr != 0)
1032 * OK, now we're committed to doing something.
1034 INP_INFO_RLOCK(&V_ripcbinfo);
1035 gencnt = V_ripcbinfo.ipi_gencnt;
1036 n = V_ripcbinfo.ipi_count;
1037 INP_INFO_RUNLOCK(&V_ripcbinfo);
1039 xig.xig_len = sizeof xig;
1041 xig.xig_gen = gencnt;
1042 xig.xig_sogen = so_gencnt;
1043 error = SYSCTL_OUT(req, &xig, sizeof xig);
1047 inp_list = malloc(n * sizeof *inp_list, M_TEMP, M_WAITOK);
1051 INP_INFO_RLOCK(&V_ripcbinfo);
1052 for (inp = LIST_FIRST(V_ripcbinfo.ipi_listhead), i = 0; inp && i < n;
1053 inp = LIST_NEXT(inp, inp_list)) {
1055 if (inp->inp_gencnt <= gencnt &&
1056 cr_canseeinpcb(req->td->td_ucred, inp) == 0) {
1058 inp_list[i++] = inp;
1062 INP_INFO_RUNLOCK(&V_ripcbinfo);
1066 for (i = 0; i < n; i++) {
1069 if (inp->inp_gencnt <= gencnt) {
1072 bzero(&xi, sizeof(xi));
1073 xi.xi_len = sizeof xi;
1074 /* XXX should avoid extra copy */
1075 bcopy(inp, &xi.xi_inp, sizeof *inp);
1076 if (inp->inp_socket)
1077 sotoxsocket(inp->inp_socket, &xi.xi_socket);
1079 error = SYSCTL_OUT(req, &xi, sizeof xi);
1083 INP_INFO_WLOCK(&V_ripcbinfo);
1084 for (i = 0; i < n; i++) {
1087 if (!in_pcbrele_rlocked(inp))
1090 INP_INFO_WUNLOCK(&V_ripcbinfo);
1094 * Give the user an updated idea of our state. If the
1095 * generation differs from what we told her before, she knows
1096 * that something happened while we were processing this
1097 * request, and it might be necessary to retry.
1099 INP_INFO_RLOCK(&V_ripcbinfo);
1100 xig.xig_gen = V_ripcbinfo.ipi_gencnt;
1101 xig.xig_sogen = so_gencnt;
1102 xig.xig_count = V_ripcbinfo.ipi_count;
1103 INP_INFO_RUNLOCK(&V_ripcbinfo);
1104 error = SYSCTL_OUT(req, &xig, sizeof xig);
1106 free(inp_list, M_TEMP);
1110 SYSCTL_PROC(_net_inet_raw, OID_AUTO/*XXX*/, pcblist,
1111 CTLTYPE_OPAQUE | CTLFLAG_RD, NULL, 0,
1112 rip_pcblist, "S,xinpcb", "List of active raw IP sockets");
1115 struct pr_usrreqs rip_usrreqs = {
1116 .pru_abort = rip_abort,
1117 .pru_attach = rip_attach,
1118 .pru_bind = rip_bind,
1119 .pru_connect = rip_connect,
1120 .pru_control = in_control,
1121 .pru_detach = rip_detach,
1122 .pru_disconnect = rip_disconnect,
1123 .pru_peeraddr = in_getpeeraddr,
1124 .pru_send = rip_send,
1125 .pru_shutdown = rip_shutdown,
1126 .pru_sockaddr = in_getsockaddr,
1127 .pru_sosetlabel = in_pcbsosetlabel,
1128 .pru_close = rip_close,