2 * Copyright (c) 1982, 1986, 1988, 1993
3 * The Regents of the University of California. All rights reserved.
5 * Redistribution and use in source and binary forms, with or without
6 * modification, are permitted provided that the following conditions
8 * 1. Redistributions of source code must retain the above copyright
9 * notice, this list of conditions and the following disclaimer.
10 * 2. Redistributions in binary form must reproduce the above copyright
11 * notice, this list of conditions and the following disclaimer in the
12 * documentation and/or other materials provided with the distribution.
13 * 4. Neither the name of the University nor the names of its contributors
14 * may be used to endorse or promote products derived from this software
15 * without specific prior written permission.
17 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
18 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
19 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
20 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
21 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
22 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
23 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
24 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
25 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
26 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
29 * @(#)raw_ip.c 8.7 (Berkeley) 5/15/95
33 #include "opt_inet6.h"
34 #include "opt_ipsec.h"
37 #include <sys/param.h>
39 #include <sys/kernel.h>
42 #include <sys/malloc.h>
45 #include <sys/protosw.h>
46 #include <sys/signalvar.h>
47 #include <sys/socket.h>
48 #include <sys/socketvar.h>
50 #include <sys/sysctl.h>
51 #include <sys/systm.h>
56 #include <net/route.h>
58 #include <netinet/in.h>
59 #include <netinet/in_systm.h>
60 #include <netinet/in_pcb.h>
61 #include <netinet/in_var.h>
62 #include <netinet/ip.h>
63 #include <netinet/ip_var.h>
64 #include <netinet/ip_mroute.h>
66 #include <netinet/ip_fw.h>
67 #include <netinet/ip_dummynet.h>
70 #include <netipsec/ipsec.h>
74 #include <netinet6/ipsec.h>
77 struct inpcbhead ripcb;
78 struct inpcbinfo ripcbinfo;
80 /* control hooks for ipfw and dummynet */
81 ip_fw_ctl_t *ip_fw_ctl_ptr = NULL;
82 ip_dn_ctl_t *ip_dn_ctl_ptr = NULL;
85 * hooks for multicast routing. They all default to NULL,
86 * so leave them not initialized and rely on BSS being set to 0.
89 /* The socket used to communicate with the multicast routing daemon. */
90 struct socket *ip_mrouter;
92 /* The various mrouter and rsvp functions */
93 int (*ip_mrouter_set)(struct socket *, struct sockopt *);
94 int (*ip_mrouter_get)(struct socket *, struct sockopt *);
95 int (*ip_mrouter_done)(void);
96 int (*ip_mforward)(struct ip *, struct ifnet *, struct mbuf *,
97 struct ip_moptions *);
98 int (*mrt_ioctl)(int, caddr_t);
99 int (*legal_vif_num)(int);
100 u_long (*ip_mcast_src)(int);
102 void (*rsvp_input_p)(struct mbuf *m, int off);
103 int (*ip_rsvp_vif)(struct socket *, struct sockopt *);
104 void (*ip_rsvp_force_done)(struct socket *);
107 * Nominal space allocated to a raw ip socket.
113 * Raw interface to IP protocol.
117 * Initialize raw connection block q.
122 INP_INFO_LOCK_INIT(&ripcbinfo, "rip");
124 ripcbinfo.listhead = &ripcb;
126 * XXX We don't use the hash list for raw IP, but it's easier
127 * to allocate a one entry hash list than it is to check all
128 * over the place for hashbase == NULL.
130 ripcbinfo.hashbase = hashinit(1, M_PCB, &ripcbinfo.hashmask);
131 ripcbinfo.porthashbase = hashinit(1, M_PCB, &ripcbinfo.porthashmask);
132 ripcbinfo.ipi_zone = uma_zcreate("ripcb", sizeof(struct inpcb),
133 NULL, NULL, NULL, NULL, UMA_ALIGN_PTR, UMA_ZONE_NOFREE);
134 uma_zone_set_max(ripcbinfo.ipi_zone, maxsockets);
137 static struct sockaddr_in ripsrc = { sizeof(ripsrc), AF_INET };
140 raw_append(struct inpcb *last, struct ip *ip, struct mbuf *n)
144 INP_LOCK_ASSERT(last);
146 #if defined(IPSEC) || defined(FAST_IPSEC)
147 /* check AH/ESP integrity. */
148 if (ipsec4_in_reject(n, last)) {
151 ipsecstat.in_polvio++;
153 /* do not inject data to pcb */
155 #endif /*IPSEC || FAST_IPSEC*/
157 if (!policyfail && mac_check_inpcb_deliver(last, n) != 0)
160 /* Check the minimum TTL for socket. */
161 if (last->inp_ip_minttl && last->inp_ip_minttl > ip->ip_ttl)
164 struct mbuf *opts = NULL;
167 so = last->inp_socket;
168 if ((last->inp_flags & INP_CONTROLOPTS) ||
169 (so->so_options & (SO_TIMESTAMP | SO_BINTIME)))
170 ip_savecontrol(last, &opts, ip, n);
171 SOCKBUF_LOCK(&so->so_rcv);
172 if (sbappendaddr_locked(&so->so_rcv,
173 (struct sockaddr *)&ripsrc, n, opts) == 0) {
174 /* should notify about lost packet */
178 SOCKBUF_UNLOCK(&so->so_rcv);
180 sorwakeup_locked(so);
187 * Setup generic address and protocol structures
188 * for raw_input routine, then pass them along with
192 rip_input(struct mbuf *m, int off)
194 struct ip *ip = mtod(m, struct ip *);
195 int proto = ip->ip_p;
196 struct inpcb *inp, *last;
198 INP_INFO_RLOCK(&ripcbinfo);
199 ripsrc.sin_addr = ip->ip_src;
201 LIST_FOREACH(inp, &ripcb, inp_list) {
203 if (inp->inp_ip_p && inp->inp_ip_p != proto) {
209 if ((inp->inp_vflag & INP_IPV4) == 0)
212 if (inp->inp_laddr.s_addr &&
213 inp->inp_laddr.s_addr != ip->ip_dst.s_addr)
215 if (inp->inp_faddr.s_addr &&
216 inp->inp_faddr.s_addr != ip->ip_src.s_addr)
218 if (jailed(inp->inp_socket->so_cred))
219 if (htonl(prison_getip(inp->inp_socket->so_cred)) !=
225 n = m_copy(m, 0, (int)M_COPYALL);
227 (void) raw_append(last, ip, n);
228 /* XXX count dropped packet */
234 if (raw_append(last, ip, m) != 0)
235 ipstat.ips_delivered--;
239 ipstat.ips_noproto++;
240 ipstat.ips_delivered--;
242 INP_INFO_RUNLOCK(&ripcbinfo);
246 * Generate IP header and pass packet to ip_output.
247 * Tack on options user may have setup with control call.
250 rip_output(struct mbuf *m, struct socket *so, u_long dst)
254 struct inpcb *inp = sotoinpcb(so);
255 int flags = ((so->so_options & SO_DONTROUTE) ? IP_ROUTETOIF : 0) |
259 * If the user handed us a complete IP packet, use it.
260 * Otherwise, allocate an mbuf for a header and fill it in.
262 if ((inp->inp_flags & INP_HDRINCL) == 0) {
263 if (m->m_pkthdr.len + sizeof(struct ip) > IP_MAXPACKET) {
267 M_PREPEND(m, sizeof(struct ip), M_DONTWAIT);
272 ip = mtod(m, struct ip *);
273 ip->ip_tos = inp->inp_ip_tos;
274 if (inp->inp_flags & INP_DONTFRAG)
278 ip->ip_p = inp->inp_ip_p;
279 ip->ip_len = m->m_pkthdr.len;
280 if (jailed(inp->inp_socket->so_cred))
282 htonl(prison_getip(inp->inp_socket->so_cred));
284 ip->ip_src = inp->inp_laddr;
285 ip->ip_dst.s_addr = dst;
286 ip->ip_ttl = inp->inp_ip_ttl;
288 if (m->m_pkthdr.len > IP_MAXPACKET) {
293 ip = mtod(m, struct ip *);
294 if (jailed(inp->inp_socket->so_cred)) {
295 if (ip->ip_src.s_addr !=
296 htonl(prison_getip(inp->inp_socket->so_cred))) {
302 /* don't allow both user specified and setsockopt options,
303 and don't allow packet length sizes that will crash */
304 if (((ip->ip_hl != (sizeof (*ip) >> 2))
306 || (ip->ip_len > m->m_pkthdr.len)
307 || (ip->ip_len < (ip->ip_hl << 2))) {
313 ip->ip_id = ip_newid();
314 /* XXX prevent ip_output from overwriting header fields */
315 flags |= IP_RAWOUTPUT;
319 if (inp->inp_vflag & INP_ONESBCAST)
320 flags |= IP_SENDONES;
323 mac_create_mbuf_from_inpcb(inp, m);
326 error = ip_output(m, inp->inp_options, NULL, flags,
327 inp->inp_moptions, inp);
333 * Raw IP socket option processing.
335 * IMPORTANT NOTE regarding access control: Traditionally, raw sockets could
336 * only be created by a privileged process, and as such, socket option
337 * operations to manage system properties on any raw socket were allowed to
338 * take place without explicit additional access control checks. However,
339 * raw sockets can now also be created in jail(), and therefore explicit
340 * checks are now required. Likewise, raw sockets can be used by a process
341 * after it gives up privilege, so some caution is required. For options
342 * passed down to the IP layer via ip_ctloutput(), checks are assumed to be
343 * performed in ip_ctloutput() and therefore no check occurs here.
344 * Unilaterally checking suser() here breaks normal IP socket option
345 * operations on raw sockets.
347 * When adding new socket options here, make sure to add access control
348 * checks here as necessary.
351 rip_ctloutput(struct socket *so, struct sockopt *sopt)
353 struct inpcb *inp = sotoinpcb(so);
356 if (sopt->sopt_level != IPPROTO_IP)
360 switch (sopt->sopt_dir) {
362 switch (sopt->sopt_name) {
364 optval = inp->inp_flags & INP_HDRINCL;
365 error = sooptcopyout(sopt, &optval, sizeof optval);
368 case IP_FW_ADD: /* ADD actually returns the body... */
370 case IP_FW_TABLE_GETSIZE:
371 case IP_FW_TABLE_LIST:
372 error = suser(curthread);
375 if (ip_fw_ctl_ptr != NULL)
376 error = ip_fw_ctl_ptr(sopt);
381 case IP_DUMMYNET_GET:
382 error = suser(curthread);
385 if (ip_dn_ctl_ptr != NULL)
386 error = ip_dn_ctl_ptr(sopt);
399 case MRT_API_SUPPORT:
401 case MRT_ADD_BW_UPCALL:
402 case MRT_DEL_BW_UPCALL:
403 error = suser(curthread);
406 error = ip_mrouter_get ? ip_mrouter_get(so, sopt) :
411 error = ip_ctloutput(so, sopt);
417 switch (sopt->sopt_name) {
419 error = sooptcopyin(sopt, &optval, sizeof optval,
424 inp->inp_flags |= INP_HDRINCL;
426 inp->inp_flags &= ~INP_HDRINCL;
434 case IP_FW_TABLE_ADD:
435 case IP_FW_TABLE_DEL:
436 case IP_FW_TABLE_FLUSH:
437 error = suser(curthread);
440 if (ip_fw_ctl_ptr != NULL)
441 error = ip_fw_ctl_ptr(sopt);
446 case IP_DUMMYNET_CONFIGURE:
447 case IP_DUMMYNET_DEL:
448 case IP_DUMMYNET_FLUSH:
449 error = suser(curthread);
452 if (ip_dn_ctl_ptr != NULL)
453 error = ip_dn_ctl_ptr(sopt);
455 error = ENOPROTOOPT ;
459 error = suser(curthread);
462 error = ip_rsvp_init(so);
466 error = suser(curthread);
469 error = ip_rsvp_done();
473 case IP_RSVP_VIF_OFF:
474 error = suser(curthread);
477 error = ip_rsvp_vif ?
478 ip_rsvp_vif(so, sopt) : EINVAL;
489 case MRT_API_SUPPORT:
491 case MRT_ADD_BW_UPCALL:
492 case MRT_DEL_BW_UPCALL:
493 error = suser(curthread);
496 error = ip_mrouter_set ? ip_mrouter_set(so, sopt) :
501 error = ip_ctloutput(so, sopt);
511 * This function exists solely to receive the PRC_IFDOWN messages which
512 * are sent by if_down(). It looks for an ifaddr whose ifa_addr is sa,
513 * and calls in_ifadown() to remove all routes corresponding to that address.
514 * It also receives the PRC_IFUP messages from if_up() and reinstalls the
518 rip_ctlinput(int cmd, struct sockaddr *sa, void *vip)
520 struct in_ifaddr *ia;
527 TAILQ_FOREACH(ia, &in_ifaddrhead, ia_link) {
528 if (ia->ia_ifa.ifa_addr == sa
529 && (ia->ia_flags & IFA_ROUTE)) {
531 * in_ifscrub kills the interface route.
533 in_ifscrub(ia->ia_ifp, ia);
535 * in_ifadown gets rid of all the rest of
536 * the routes. This is not quite the right
537 * thing to do, but at least if we are running
538 * a routing process they will come back.
540 in_ifadown(&ia->ia_ifa, 0);
547 TAILQ_FOREACH(ia, &in_ifaddrhead, ia_link) {
548 if (ia->ia_ifa.ifa_addr == sa)
551 if (ia == 0 || (ia->ia_flags & IFA_ROUTE))
554 ifp = ia->ia_ifa.ifa_ifp;
556 if ((ifp->if_flags & IFF_LOOPBACK)
557 || (ifp->if_flags & IFF_POINTOPOINT))
560 err = rtinit(&ia->ia_ifa, RTM_ADD, flags);
562 ia->ia_flags |= IFA_ROUTE;
567 u_long rip_sendspace = RIPSNDQ;
568 u_long rip_recvspace = RIPRCVQ;
570 SYSCTL_ULONG(_net_inet_raw, OID_AUTO, maxdgram, CTLFLAG_RW,
571 &rip_sendspace, 0, "Maximum outgoing raw IP datagram size");
572 SYSCTL_ULONG(_net_inet_raw, OID_AUTO, recvspace, CTLFLAG_RW,
573 &rip_recvspace, 0, "Maximum space for incoming raw IP datagrams");
576 rip_attach(struct socket *so, int proto, struct thread *td)
581 /* XXX why not lower? */
582 INP_INFO_WLOCK(&ripcbinfo);
585 /* XXX counter, printf */
586 INP_INFO_WUNLOCK(&ripcbinfo);
589 if (jailed(td->td_ucred) && !jail_allow_raw_sockets) {
590 INP_INFO_WUNLOCK(&ripcbinfo);
593 if ((error = suser_cred(td->td_ucred, SUSER_ALLOWJAIL)) != 0) {
594 INP_INFO_WUNLOCK(&ripcbinfo);
597 if (proto >= IPPROTO_MAX || proto < 0) {
598 INP_INFO_WUNLOCK(&ripcbinfo);
599 return EPROTONOSUPPORT;
602 error = soreserve(so, rip_sendspace, rip_recvspace);
604 INP_INFO_WUNLOCK(&ripcbinfo);
607 error = in_pcballoc(so, &ripcbinfo, "rawinp");
609 INP_INFO_WUNLOCK(&ripcbinfo);
612 inp = (struct inpcb *)so->so_pcb;
614 INP_INFO_WUNLOCK(&ripcbinfo);
615 inp->inp_vflag |= INP_IPV4;
616 inp->inp_ip_p = proto;
617 inp->inp_ip_ttl = ip_defttl;
623 rip_pcbdetach(struct socket *so, struct inpcb *inp)
626 INP_INFO_WLOCK_ASSERT(&ripcbinfo);
627 INP_LOCK_ASSERT(inp);
629 if (so == ip_mrouter && ip_mrouter_done)
631 if (ip_rsvp_force_done)
632 ip_rsvp_force_done(so);
639 rip_detach(struct socket *so)
643 INP_INFO_WLOCK(&ripcbinfo);
646 /* XXX counter, printf */
647 INP_INFO_WUNLOCK(&ripcbinfo);
651 rip_pcbdetach(so, inp);
652 INP_INFO_WUNLOCK(&ripcbinfo);
657 rip_abort(struct socket *so)
661 INP_INFO_WLOCK(&ripcbinfo);
664 INP_INFO_WUNLOCK(&ripcbinfo);
665 return EINVAL; /* ??? possible? panic instead? */
668 soisdisconnected(so);
669 if (so->so_state & SS_NOFDREF)
670 rip_pcbdetach(so, inp);
673 INP_INFO_WUNLOCK(&ripcbinfo);
678 rip_disconnect(struct socket *so)
680 if ((so->so_state & SS_ISCONNECTED) == 0)
682 return rip_abort(so);
686 rip_bind(struct socket *so, struct sockaddr *nam, struct thread *td)
688 struct sockaddr_in *addr = (struct sockaddr_in *)nam;
691 if (nam->sa_len != sizeof(*addr))
694 if (jailed(td->td_ucred)) {
695 if (addr->sin_addr.s_addr == INADDR_ANY)
696 addr->sin_addr.s_addr =
697 htonl(prison_getip(td->td_ucred));
698 if (htonl(prison_getip(td->td_ucred)) != addr->sin_addr.s_addr)
699 return (EADDRNOTAVAIL);
702 if (TAILQ_EMPTY(&ifnet) ||
703 (addr->sin_family != AF_INET && addr->sin_family != AF_IMPLINK) ||
704 (addr->sin_addr.s_addr &&
705 ifa_ifwithaddr((struct sockaddr *)addr) == 0))
706 return EADDRNOTAVAIL;
708 INP_INFO_WLOCK(&ripcbinfo);
711 INP_INFO_WUNLOCK(&ripcbinfo);
715 inp->inp_laddr = addr->sin_addr;
717 INP_INFO_WUNLOCK(&ripcbinfo);
722 rip_connect(struct socket *so, struct sockaddr *nam, struct thread *td)
724 struct sockaddr_in *addr = (struct sockaddr_in *)nam;
727 if (nam->sa_len != sizeof(*addr))
729 if (TAILQ_EMPTY(&ifnet))
730 return EADDRNOTAVAIL;
731 if (addr->sin_family != AF_INET && addr->sin_family != AF_IMPLINK)
734 INP_INFO_WLOCK(&ripcbinfo);
737 INP_INFO_WUNLOCK(&ripcbinfo);
741 inp->inp_faddr = addr->sin_addr;
744 INP_INFO_WUNLOCK(&ripcbinfo);
749 rip_shutdown(struct socket *so)
753 INP_INFO_RLOCK(&ripcbinfo);
756 INP_INFO_RUNLOCK(&ripcbinfo);
760 INP_INFO_RUNLOCK(&ripcbinfo);
767 rip_send(struct socket *so, int flags, struct mbuf *m, struct sockaddr *nam,
768 struct mbuf *control, struct thread *td)
774 INP_INFO_WLOCK(&ripcbinfo);
776 if (so->so_state & SS_ISCONNECTED) {
778 INP_INFO_WUNLOCK(&ripcbinfo);
782 dst = inp->inp_faddr.s_addr;
785 INP_INFO_WUNLOCK(&ripcbinfo);
789 dst = ((struct sockaddr_in *)nam)->sin_addr.s_addr;
791 ret = rip_output(m, so, dst);
792 INP_INFO_WUNLOCK(&ripcbinfo);
797 rip_pcblist(SYSCTL_HANDLER_ARGS)
800 struct inpcb *inp, **inp_list;
805 * The process of preparing the TCB list is too time-consuming and
806 * resource-intensive to repeat twice on every request.
808 if (req->oldptr == 0) {
809 n = ripcbinfo.ipi_count;
810 req->oldidx = 2 * (sizeof xig)
811 + (n + n/8) * sizeof(struct xinpcb);
815 if (req->newptr != 0)
819 * OK, now we're committed to doing something.
821 INP_INFO_RLOCK(&ripcbinfo);
822 gencnt = ripcbinfo.ipi_gencnt;
823 n = ripcbinfo.ipi_count;
824 INP_INFO_RUNLOCK(&ripcbinfo);
826 xig.xig_len = sizeof xig;
828 xig.xig_gen = gencnt;
829 xig.xig_sogen = so_gencnt;
830 error = SYSCTL_OUT(req, &xig, sizeof xig);
834 inp_list = malloc(n * sizeof *inp_list, M_TEMP, M_WAITOK);
838 INP_INFO_RLOCK(&ripcbinfo);
839 for (inp = LIST_FIRST(ripcbinfo.listhead), i = 0; inp && i < n;
840 inp = LIST_NEXT(inp, inp_list)) {
842 if (inp->inp_gencnt <= gencnt &&
843 cr_canseesocket(req->td->td_ucred, inp->inp_socket) == 0) {
844 /* XXX held references? */
849 INP_INFO_RUNLOCK(&ripcbinfo);
853 for (i = 0; i < n; i++) {
855 if (inp->inp_gencnt <= gencnt) {
857 bzero(&xi, sizeof(xi));
858 xi.xi_len = sizeof xi;
859 /* XXX should avoid extra copy */
860 bcopy(inp, &xi.xi_inp, sizeof *inp);
862 sotoxsocket(inp->inp_socket, &xi.xi_socket);
863 error = SYSCTL_OUT(req, &xi, sizeof xi);
868 * Give the user an updated idea of our state.
869 * If the generation differs from what we told
870 * her before, she knows that something happened
871 * while we were processing this request, and it
872 * might be necessary to retry.
874 INP_INFO_RLOCK(&ripcbinfo);
875 xig.xig_gen = ripcbinfo.ipi_gencnt;
876 xig.xig_sogen = so_gencnt;
877 xig.xig_count = ripcbinfo.ipi_count;
878 INP_INFO_RUNLOCK(&ripcbinfo);
879 error = SYSCTL_OUT(req, &xig, sizeof xig);
881 free(inp_list, M_TEMP);
886 * This is the wrapper function for in_setsockaddr. We just pass down
887 * the pcbinfo for in_setpeeraddr to lock.
890 rip_sockaddr(struct socket *so, struct sockaddr **nam)
892 return (in_setsockaddr(so, nam, &ripcbinfo));
896 * This is the wrapper function for in_setpeeraddr. We just pass down
897 * the pcbinfo for in_setpeeraddr to lock.
900 rip_peeraddr(struct socket *so, struct sockaddr **nam)
902 return (in_setpeeraddr(so, nam, &ripcbinfo));
906 SYSCTL_PROC(_net_inet_raw, OID_AUTO/*XXX*/, pcblist, CTLFLAG_RD, 0, 0,
907 rip_pcblist, "S,xinpcb", "List of active raw IP sockets");
909 struct pr_usrreqs rip_usrreqs = {
910 .pru_abort = rip_abort,
911 .pru_attach = rip_attach,
912 .pru_bind = rip_bind,
913 .pru_connect = rip_connect,
914 .pru_control = in_control,
915 .pru_detach = rip_detach,
916 .pru_disconnect = rip_disconnect,
917 .pru_peeraddr = rip_peeraddr,
918 .pru_send = rip_send,
919 .pru_shutdown = rip_shutdown,
920 .pru_sockaddr = rip_sockaddr,
921 .pru_sosetlabel = in_pcbsosetlabel