FreeBSD/Linux Kernel Cross Reference
sys/netinet/raw_ip.c
1 /*-
2 * Copyright (c) 1982, 1986, 1988, 1993
3 * The Regents of the University of California. All rights reserved.
4 *
5 * Redistribution and use in source and binary forms, with or without
6 * modification, are permitted provided that the following conditions
7 * are met:
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.
16 *
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
27 * SUCH DAMAGE.
28 *
29 * @(#)raw_ip.c 8.7 (Berkeley) 5/15/95
30 * $FreeBSD$
31 */
32
33 #include "opt_inet6.h"
34 #include "opt_ipsec.h"
35 #include "opt_mac.h"
36
37 #include <sys/param.h>
38 #include <sys/jail.h>
39 #include <sys/kernel.h>
40 #include <sys/lock.h>
41 #include <sys/mac.h>
42 #include <sys/malloc.h>
43 #include <sys/mbuf.h>
44 #include <sys/proc.h>
45 #include <sys/protosw.h>
46 #include <sys/signalvar.h>
47 #include <sys/socket.h>
48 #include <sys/socketvar.h>
49 #include <sys/sx.h>
50 #include <sys/sysctl.h>
51 #include <sys/systm.h>
52
53 #include <vm/uma.h>
54
55 #include <net/if.h>
56 #include <net/route.h>
57
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>
65
66 #include <netinet/ip_fw.h>
67 #include <netinet/ip_dummynet.h>
68
69 #ifdef FAST_IPSEC
70 #include <netipsec/ipsec.h>
71 #endif /*FAST_IPSEC*/
72
73 #ifdef IPSEC
74 #include <netinet6/ipsec.h>
75 #endif /*IPSEC*/
76
77 struct inpcbhead ripcb;
78 struct inpcbinfo ripcbinfo;
79
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;
83
84 /*
85 * hooks for multicast routing. They all default to NULL,
86 * so leave them not initialized and rely on BSS being set to 0.
87 */
88
89 /* The socket used to communicate with the multicast routing daemon. */
90 struct socket *ip_mrouter;
91
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);
101
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 *);
105
106 /*
107 * Raw interface to IP protocol.
108 */
109
110 /*
111 * Initialize raw connection block q.
112 */
113 static void
114 rip_zone_change(void *tag)
115 {
116
117 uma_zone_set_max(ripcbinfo.ipi_zone, maxsockets);
118 }
119
120 static int
121 rip_inpcb_init(void *mem, int size, int flags)
122 {
123 struct inpcb *inp = mem;
124
125 INP_LOCK_INIT(inp, "inp", "rawinp");
126 return (0);
127 }
128
129 void
130 rip_init()
131 {
132 INP_INFO_LOCK_INIT(&ripcbinfo, "rip");
133 LIST_INIT(&ripcb);
134 ripcbinfo.listhead = &ripcb;
135 /*
136 * XXX We don't use the hash list for raw IP, but it's easier
137 * to allocate a one entry hash list than it is to check all
138 * over the place for hashbase == NULL.
139 */
140 ripcbinfo.hashbase = hashinit(1, M_PCB, &ripcbinfo.hashmask);
141 ripcbinfo.porthashbase = hashinit(1, M_PCB, &ripcbinfo.porthashmask);
142 ripcbinfo.ipi_zone = uma_zcreate("ripcb", sizeof(struct inpcb),
143 NULL, NULL, rip_inpcb_init, NULL, UMA_ALIGN_PTR, UMA_ZONE_NOFREE);
144 uma_zone_set_max(ripcbinfo.ipi_zone, maxsockets);
145 EVENTHANDLER_REGISTER(maxsockets_change, rip_zone_change,
146 NULL, EVENTHANDLER_PRI_ANY);
147 }
148
149 static struct sockaddr_in ripsrc = { sizeof(ripsrc), AF_INET };
150
151 static int
152 raw_append(struct inpcb *last, struct ip *ip, struct mbuf *n)
153 {
154 int policyfail = 0;
155
156 INP_LOCK_ASSERT(last);
157
158 #if defined(IPSEC) || defined(FAST_IPSEC)
159 /* check AH/ESP integrity. */
160 if (ipsec4_in_reject(n, last)) {
161 policyfail = 1;
162 #ifdef IPSEC
163 ipsecstat.in_polvio++;
164 #endif /*IPSEC*/
165 /* do not inject data to pcb */
166 }
167 #endif /*IPSEC || FAST_IPSEC*/
168 #ifdef MAC
169 if (!policyfail && mac_check_inpcb_deliver(last, n) != 0)
170 policyfail = 1;
171 #endif
172 /* Check the minimum TTL for socket. */
173 if (last->inp_ip_minttl && last->inp_ip_minttl > ip->ip_ttl)
174 policyfail = 1;
175 if (!policyfail) {
176 struct mbuf *opts = NULL;
177 struct socket *so;
178
179 so = last->inp_socket;
180 if ((last->inp_flags & INP_CONTROLOPTS) ||
181 (so->so_options & SO_TIMESTAMP))
182 ip_savecontrol(last, &opts, ip, n);
183 SOCKBUF_LOCK(&so->so_rcv);
184 if (sbappendaddr_locked(&so->so_rcv,
185 (struct sockaddr *)&ripsrc, n, opts) == 0) {
186 /* should notify about lost packet */
187 m_freem(n);
188 if (opts)
189 m_freem(opts);
190 SOCKBUF_UNLOCK(&so->so_rcv);
191 } else
192 sorwakeup_locked(so);
193 } else
194 m_freem(n);
195 return policyfail;
196 }
197
198 /*
199 * Setup generic address and protocol structures
200 * for raw_input routine, then pass them along with
201 * mbuf chain.
202 */
203 void
204 rip_input(struct mbuf *m, int off)
205 {
206 struct ip *ip = mtod(m, struct ip *);
207 int proto = ip->ip_p;
208 struct inpcb *inp, *last;
209
210 INP_INFO_RLOCK(&ripcbinfo);
211 ripsrc.sin_addr = ip->ip_src;
212 last = NULL;
213 LIST_FOREACH(inp, &ripcb, inp_list) {
214 INP_LOCK(inp);
215 if (inp->inp_ip_p && inp->inp_ip_p != proto) {
216 docontinue:
217 INP_UNLOCK(inp);
218 continue;
219 }
220 #ifdef INET6
221 if ((inp->inp_vflag & INP_IPV4) == 0)
222 goto docontinue;
223 #endif
224 if (inp->inp_laddr.s_addr &&
225 inp->inp_laddr.s_addr != ip->ip_dst.s_addr)
226 goto docontinue;
227 if (inp->inp_faddr.s_addr &&
228 inp->inp_faddr.s_addr != ip->ip_src.s_addr)
229 goto docontinue;
230 if (jailed(inp->inp_socket->so_cred))
231 if (htonl(prison_getip(inp->inp_socket->so_cred)) !=
232 ip->ip_dst.s_addr)
233 goto docontinue;
234 if (last) {
235 struct mbuf *n;
236
237 n = m_copy(m, 0, (int)M_COPYALL);
238 if (n != NULL)
239 (void) raw_append(last, ip, n);
240 /* XXX count dropped packet */
241 INP_UNLOCK(last);
242 }
243 last = inp;
244 }
245 if (last != NULL) {
246 if (raw_append(last, ip, m) != 0)
247 ipstat.ips_delivered--;
248 INP_UNLOCK(last);
249 } else {
250 m_freem(m);
251 ipstat.ips_noproto++;
252 ipstat.ips_delivered--;
253 }
254 INP_INFO_RUNLOCK(&ripcbinfo);
255 }
256
257 /*
258 * Generate IP header and pass packet to ip_output.
259 * Tack on options user may have setup with control call.
260 */
261 int
262 rip_output(struct mbuf *m, struct socket *so, u_long dst)
263 {
264 struct ip *ip;
265 int error;
266 struct inpcb *inp = sotoinpcb(so);
267 int flags = ((so->so_options & SO_DONTROUTE) ? IP_ROUTETOIF : 0) |
268 IP_ALLOWBROADCAST;
269
270 /*
271 * If the user handed us a complete IP packet, use it.
272 * Otherwise, allocate an mbuf for a header and fill it in.
273 */
274 if ((inp->inp_flags & INP_HDRINCL) == 0) {
275 if (m->m_pkthdr.len + sizeof(struct ip) > IP_MAXPACKET) {
276 m_freem(m);
277 return(EMSGSIZE);
278 }
279 M_PREPEND(m, sizeof(struct ip), M_DONTWAIT);
280 if (m == NULL)
281 return(ENOBUFS);
282
283 INP_LOCK(inp);
284 ip = mtod(m, struct ip *);
285 ip->ip_tos = inp->inp_ip_tos;
286 if (inp->inp_flags & INP_DONTFRAG)
287 ip->ip_off = IP_DF;
288 else
289 ip->ip_off = 0;
290 ip->ip_p = inp->inp_ip_p;
291 ip->ip_len = m->m_pkthdr.len;
292 if (jailed(inp->inp_socket->so_cred))
293 ip->ip_src.s_addr =
294 htonl(prison_getip(inp->inp_socket->so_cred));
295 else
296 ip->ip_src = inp->inp_laddr;
297 ip->ip_dst.s_addr = dst;
298 ip->ip_ttl = inp->inp_ip_ttl;
299 } else {
300 if (m->m_pkthdr.len > IP_MAXPACKET) {
301 m_freem(m);
302 return(EMSGSIZE);
303 }
304 INP_LOCK(inp);
305 ip = mtod(m, struct ip *);
306 if (jailed(inp->inp_socket->so_cred)) {
307 if (ip->ip_src.s_addr !=
308 htonl(prison_getip(inp->inp_socket->so_cred))) {
309 INP_UNLOCK(inp);
310 m_freem(m);
311 return (EPERM);
312 }
313 }
314 /* don't allow both user specified and setsockopt options,
315 and don't allow packet length sizes that will crash */
316 if (((ip->ip_hl != (sizeof (*ip) >> 2))
317 && inp->inp_options)
318 || (ip->ip_len > m->m_pkthdr.len)
319 || (ip->ip_len < (ip->ip_hl << 2))) {
320 INP_UNLOCK(inp);
321 m_freem(m);
322 return EINVAL;
323 }
324 if (ip->ip_id == 0)
325 ip->ip_id = ip_newid();
326 /* XXX prevent ip_output from overwriting header fields */
327 flags |= IP_RAWOUTPUT;
328 ipstat.ips_rawout++;
329 }
330
331 if (inp->inp_flags & INP_ONESBCAST)
332 flags |= IP_SENDONES;
333
334 #ifdef MAC
335 mac_create_mbuf_from_inpcb(inp, m);
336 #endif
337
338 error = ip_output(m, inp->inp_options, NULL, flags,
339 inp->inp_moptions, inp);
340 INP_UNLOCK(inp);
341 return error;
342 }
343
344 /*
345 * Raw IP socket option processing.
346 *
347 * IMPORTANT NOTE regarding access control: Traditionally, raw sockets could
348 * only be created by a privileged process, and as such, socket option
349 * operations to manage system properties on any raw socket were allowed to
350 * take place without explicit additional access control checks. However,
351 * raw sockets can now also be created in jail(), and therefore explicit
352 * checks are now required. Likewise, raw sockets can be used by a process
353 * after it gives up privilege, so some caution is required. For options
354 * passed down to the IP layer via ip_ctloutput(), checks are assumed to be
355 * performed in ip_ctloutput() and therefore no check occurs here.
356 * Unilaterally checking suser() here breaks normal IP socket option
357 * operations on raw sockets.
358 *
359 * When adding new socket options here, make sure to add access control
360 * checks here as necessary.
361 */
362 int
363 rip_ctloutput(struct socket *so, struct sockopt *sopt)
364 {
365 struct inpcb *inp = sotoinpcb(so);
366 int error, optval;
367
368 if (sopt->sopt_level != IPPROTO_IP)
369 return (EINVAL);
370
371 error = 0;
372 switch (sopt->sopt_dir) {
373 case SOPT_GET:
374 switch (sopt->sopt_name) {
375 case IP_HDRINCL:
376 optval = inp->inp_flags & INP_HDRINCL;
377 error = sooptcopyout(sopt, &optval, sizeof optval);
378 break;
379
380 case IP_FW_ADD: /* ADD actually returns the body... */
381 case IP_FW_GET:
382 case IP_FW_TABLE_GETSIZE:
383 case IP_FW_TABLE_LIST:
384 error = suser(curthread);
385 if (error != 0)
386 return (error);
387 if (ip_fw_ctl_ptr != NULL)
388 error = ip_fw_ctl_ptr(sopt);
389 else
390 error = ENOPROTOOPT;
391 break;
392
393 case IP_DUMMYNET_GET:
394 error = suser(curthread);
395 if (error != 0)
396 return (error);
397 if (ip_dn_ctl_ptr != NULL)
398 error = ip_dn_ctl_ptr(sopt);
399 else
400 error = ENOPROTOOPT;
401 break ;
402
403 case MRT_INIT:
404 case MRT_DONE:
405 case MRT_ADD_VIF:
406 case MRT_DEL_VIF:
407 case MRT_ADD_MFC:
408 case MRT_DEL_MFC:
409 case MRT_VERSION:
410 case MRT_ASSERT:
411 case MRT_API_SUPPORT:
412 case MRT_API_CONFIG:
413 case MRT_ADD_BW_UPCALL:
414 case MRT_DEL_BW_UPCALL:
415 error = suser(curthread);
416 if (error != 0)
417 return (error);
418 error = ip_mrouter_get ? ip_mrouter_get(so, sopt) :
419 EOPNOTSUPP;
420 break;
421
422 default:
423 error = ip_ctloutput(so, sopt);
424 break;
425 }
426 break;
427
428 case SOPT_SET:
429 switch (sopt->sopt_name) {
430 case IP_HDRINCL:
431 error = sooptcopyin(sopt, &optval, sizeof optval,
432 sizeof optval);
433 if (error)
434 break;
435 if (optval)
436 inp->inp_flags |= INP_HDRINCL;
437 else
438 inp->inp_flags &= ~INP_HDRINCL;
439 break;
440
441 case IP_FW_ADD:
442 case IP_FW_DEL:
443 case IP_FW_FLUSH:
444 case IP_FW_ZERO:
445 case IP_FW_RESETLOG:
446 case IP_FW_TABLE_ADD:
447 case IP_FW_TABLE_DEL:
448 case IP_FW_TABLE_FLUSH:
449 error = suser(curthread);
450 if (error != 0)
451 return (error);
452 if (ip_fw_ctl_ptr != NULL)
453 error = ip_fw_ctl_ptr(sopt);
454 else
455 error = ENOPROTOOPT;
456 break;
457
458 case IP_DUMMYNET_CONFIGURE:
459 case IP_DUMMYNET_DEL:
460 case IP_DUMMYNET_FLUSH:
461 error = suser(curthread);
462 if (error != 0)
463 return (error);
464 if (ip_dn_ctl_ptr != NULL)
465 error = ip_dn_ctl_ptr(sopt);
466 else
467 error = ENOPROTOOPT ;
468 break ;
469
470 case IP_RSVP_ON:
471 error = suser(curthread);
472 if (error != 0)
473 return (error);
474 error = ip_rsvp_init(so);
475 break;
476
477 case IP_RSVP_OFF:
478 error = suser(curthread);
479 if (error != 0)
480 return (error);
481 error = ip_rsvp_done();
482 break;
483
484 case IP_RSVP_VIF_ON:
485 case IP_RSVP_VIF_OFF:
486 error = suser(curthread);
487 if (error != 0)
488 return (error);
489 error = ip_rsvp_vif ?
490 ip_rsvp_vif(so, sopt) : EINVAL;
491 break;
492
493 case MRT_INIT:
494 case MRT_DONE:
495 case MRT_ADD_VIF:
496 case MRT_DEL_VIF:
497 case MRT_ADD_MFC:
498 case MRT_DEL_MFC:
499 case MRT_VERSION:
500 case MRT_ASSERT:
501 case MRT_API_SUPPORT:
502 case MRT_API_CONFIG:
503 case MRT_ADD_BW_UPCALL:
504 case MRT_DEL_BW_UPCALL:
505 error = suser(curthread);
506 if (error != 0)
507 return (error);
508 error = ip_mrouter_set ? ip_mrouter_set(so, sopt) :
509 EOPNOTSUPP;
510 break;
511
512 default:
513 error = ip_ctloutput(so, sopt);
514 break;
515 }
516 break;
517 }
518
519 return (error);
520 }
521
522 /*
523 * This function exists solely to receive the PRC_IFDOWN messages which
524 * are sent by if_down(). It looks for an ifaddr whose ifa_addr is sa,
525 * and calls in_ifadown() to remove all routes corresponding to that address.
526 * It also receives the PRC_IFUP messages from if_up() and reinstalls the
527 * interface routes.
528 */
529 void
530 rip_ctlinput(int cmd, struct sockaddr *sa, void *vip)
531 {
532 struct in_ifaddr *ia;
533 struct ifnet *ifp;
534 int err;
535 int flags;
536
537 switch (cmd) {
538 case PRC_IFDOWN:
539 TAILQ_FOREACH(ia, &in_ifaddrhead, ia_link) {
540 if (ia->ia_ifa.ifa_addr == sa
541 && (ia->ia_flags & IFA_ROUTE)) {
542 /*
543 * in_ifscrub kills the interface route.
544 */
545 in_ifscrub(ia->ia_ifp, ia);
546 /*
547 * in_ifadown gets rid of all the rest of
548 * the routes. This is not quite the right
549 * thing to do, but at least if we are running
550 * a routing process they will come back.
551 */
552 in_ifadown(&ia->ia_ifa, 0);
553 break;
554 }
555 }
556 break;
557
558 case PRC_IFUP:
559 TAILQ_FOREACH(ia, &in_ifaddrhead, ia_link) {
560 if (ia->ia_ifa.ifa_addr == sa)
561 break;
562 }
563 if (ia == 0 || (ia->ia_flags & IFA_ROUTE))
564 return;
565 flags = RTF_UP;
566 ifp = ia->ia_ifa.ifa_ifp;
567
568 if ((ifp->if_flags & IFF_LOOPBACK)
569 || (ifp->if_flags & IFF_POINTOPOINT))
570 flags |= RTF_HOST;
571
572 err = rtinit(&ia->ia_ifa, RTM_ADD, flags);
573 if (err == 0)
574 ia->ia_flags |= IFA_ROUTE;
575 break;
576 }
577 }
578
579 u_long rip_sendspace = 9216;
580 u_long rip_recvspace = 9216;
581
582 SYSCTL_INT(_net_inet_raw, OID_AUTO, maxdgram, CTLFLAG_RW,
583 &rip_sendspace, 0, "Maximum outgoing raw IP datagram size");
584 SYSCTL_INT(_net_inet_raw, OID_AUTO, recvspace, CTLFLAG_RW,
585 &rip_recvspace, 0, "Maximum space for incoming raw IP datagrams");
586
587 static int
588 rip_attach(struct socket *so, int proto, struct thread *td)
589 {
590 struct inpcb *inp;
591 int error;
592
593 /* XXX why not lower? */
594 INP_INFO_WLOCK(&ripcbinfo);
595 inp = sotoinpcb(so);
596 if (inp) {
597 /* XXX counter, printf */
598 INP_INFO_WUNLOCK(&ripcbinfo);
599 return EINVAL;
600 }
601 if (jailed(td->td_ucred) && !jail_allow_raw_sockets) {
602 INP_INFO_WUNLOCK(&ripcbinfo);
603 return (EPERM);
604 }
605 if ((error = suser_cred(td->td_ucred, SUSER_ALLOWJAIL)) != 0) {
606 INP_INFO_WUNLOCK(&ripcbinfo);
607 return error;
608 }
609 if (proto >= IPPROTO_MAX || proto < 0) {
610 INP_INFO_WUNLOCK(&ripcbinfo);
611 return EPROTONOSUPPORT;
612 }
613
614 error = soreserve(so, rip_sendspace, rip_recvspace);
615 if (error) {
616 INP_INFO_WUNLOCK(&ripcbinfo);
617 return error;
618 }
619 error = in_pcballoc(so, &ripcbinfo);
620 if (error) {
621 INP_INFO_WUNLOCK(&ripcbinfo);
622 return error;
623 }
624 inp = (struct inpcb *)so->so_pcb;
625 INP_INFO_WUNLOCK(&ripcbinfo);
626 inp->inp_vflag |= INP_IPV4;
627 inp->inp_ip_p = proto;
628 inp->inp_ip_ttl = ip_defttl;
629 INP_UNLOCK(inp);
630 return 0;
631 }
632
633 static void
634 rip_pcbdetach(struct socket *so, struct inpcb *inp)
635 {
636
637 INP_INFO_WLOCK_ASSERT(&ripcbinfo);
638 INP_LOCK_ASSERT(inp);
639
640 if (so == ip_mrouter && ip_mrouter_done)
641 ip_mrouter_done();
642 if (ip_rsvp_force_done)
643 ip_rsvp_force_done(so);
644 if (so == ip_rsvpd)
645 ip_rsvp_done();
646 in_pcbdetach(inp);
647 }
648
649 static int
650 rip_detach(struct socket *so)
651 {
652 struct inpcb *inp;
653
654 INP_INFO_WLOCK(&ripcbinfo);
655 inp = sotoinpcb(so);
656 if (inp == 0) {
657 /* XXX counter, printf */
658 INP_INFO_WUNLOCK(&ripcbinfo);
659 return EINVAL;
660 }
661 INP_LOCK(inp);
662 rip_pcbdetach(so, inp);
663 INP_INFO_WUNLOCK(&ripcbinfo);
664 return 0;
665 }
666
667 static int
668 rip_abort(struct socket *so)
669 {
670 struct inpcb *inp;
671
672 INP_INFO_WLOCK(&ripcbinfo);
673 inp = sotoinpcb(so);
674 if (inp == 0) {
675 INP_INFO_WUNLOCK(&ripcbinfo);
676 return EINVAL; /* ??? possible? panic instead? */
677 }
678 INP_LOCK(inp);
679 soisdisconnected(so);
680 if (so->so_state & SS_NOFDREF)
681 rip_pcbdetach(so, inp);
682 else
683 INP_UNLOCK(inp);
684 INP_INFO_WUNLOCK(&ripcbinfo);
685 return 0;
686 }
687
688 static int
689 rip_disconnect(struct socket *so)
690 {
691 if ((so->so_state & SS_ISCONNECTED) == 0)
692 return ENOTCONN;
693 return rip_abort(so);
694 }
695
696 static int
697 rip_bind(struct socket *so, struct sockaddr *nam, struct thread *td)
698 {
699 struct sockaddr_in *addr = (struct sockaddr_in *)nam;
700 struct inpcb *inp;
701
702 if (nam->sa_len != sizeof(*addr))
703 return EINVAL;
704
705 if (jailed(td->td_ucred)) {
706 if (addr->sin_addr.s_addr == INADDR_ANY)
707 addr->sin_addr.s_addr =
708 htonl(prison_getip(td->td_ucred));
709 if (htonl(prison_getip(td->td_ucred)) != addr->sin_addr.s_addr)
710 return (EADDRNOTAVAIL);
711 }
712
713 if (TAILQ_EMPTY(&ifnet) ||
714 (addr->sin_family != AF_INET && addr->sin_family != AF_IMPLINK) ||
715 (addr->sin_addr.s_addr &&
716 ifa_ifwithaddr((struct sockaddr *)addr) == 0))
717 return EADDRNOTAVAIL;
718
719 INP_INFO_WLOCK(&ripcbinfo);
720 inp = sotoinpcb(so);
721 if (inp == 0) {
722 INP_INFO_WUNLOCK(&ripcbinfo);
723 return EINVAL;
724 }
725 INP_LOCK(inp);
726 inp->inp_laddr = addr->sin_addr;
727 INP_UNLOCK(inp);
728 INP_INFO_WUNLOCK(&ripcbinfo);
729 return 0;
730 }
731
732 static int
733 rip_connect(struct socket *so, struct sockaddr *nam, struct thread *td)
734 {
735 struct sockaddr_in *addr = (struct sockaddr_in *)nam;
736 struct inpcb *inp;
737
738 if (nam->sa_len != sizeof(*addr))
739 return EINVAL;
740 if (TAILQ_EMPTY(&ifnet))
741 return EADDRNOTAVAIL;
742 if (addr->sin_family != AF_INET && addr->sin_family != AF_IMPLINK)
743 return EAFNOSUPPORT;
744
745 INP_INFO_WLOCK(&ripcbinfo);
746 inp = sotoinpcb(so);
747 if (inp == 0) {
748 INP_INFO_WUNLOCK(&ripcbinfo);
749 return EINVAL;
750 }
751 INP_LOCK(inp);
752 inp->inp_faddr = addr->sin_addr;
753 soisconnected(so);
754 INP_UNLOCK(inp);
755 INP_INFO_WUNLOCK(&ripcbinfo);
756 return 0;
757 }
758
759 static int
760 rip_shutdown(struct socket *so)
761 {
762 struct inpcb *inp;
763
764 INP_INFO_RLOCK(&ripcbinfo);
765 inp = sotoinpcb(so);
766 if (inp == 0) {
767 INP_INFO_RUNLOCK(&ripcbinfo);
768 return EINVAL;
769 }
770 INP_LOCK(inp);
771 INP_INFO_RUNLOCK(&ripcbinfo);
772 socantsendmore(so);
773 INP_UNLOCK(inp);
774 return 0;
775 }
776
777 static int
778 rip_send(struct socket *so, int flags, struct mbuf *m, struct sockaddr *nam,
779 struct mbuf *control, struct thread *td)
780 {
781 struct inpcb *inp;
782 u_long dst;
783 int ret;
784
785 INP_INFO_WLOCK(&ripcbinfo);
786 inp = sotoinpcb(so);
787 if (so->so_state & SS_ISCONNECTED) {
788 if (nam) {
789 INP_INFO_WUNLOCK(&ripcbinfo);
790 m_freem(m);
791 return EISCONN;
792 }
793 dst = inp->inp_faddr.s_addr;
794 } else {
795 if (nam == NULL) {
796 INP_INFO_WUNLOCK(&ripcbinfo);
797 m_freem(m);
798 return ENOTCONN;
799 }
800 dst = ((struct sockaddr_in *)nam)->sin_addr.s_addr;
801 }
802 ret = rip_output(m, so, dst);
803 INP_INFO_WUNLOCK(&ripcbinfo);
804 return ret;
805 }
806
807 static int
808 rip_pcblist(SYSCTL_HANDLER_ARGS)
809 {
810 int error, i, n;
811 struct inpcb *inp, **inp_list;
812 inp_gen_t gencnt;
813 struct xinpgen xig;
814
815 /*
816 * The process of preparing the TCB list is too time-consuming and
817 * resource-intensive to repeat twice on every request.
818 */
819 if (req->oldptr == 0) {
820 n = ripcbinfo.ipi_count;
821 req->oldidx = 2 * (sizeof xig)
822 + (n + n/8) * sizeof(struct xinpcb);
823 return 0;
824 }
825
826 if (req->newptr != 0)
827 return EPERM;
828
829 /*
830 * OK, now we're committed to doing something.
831 */
832 INP_INFO_RLOCK(&ripcbinfo);
833 gencnt = ripcbinfo.ipi_gencnt;
834 n = ripcbinfo.ipi_count;
835 INP_INFO_RUNLOCK(&ripcbinfo);
836
837 xig.xig_len = sizeof xig;
838 xig.xig_count = n;
839 xig.xig_gen = gencnt;
840 xig.xig_sogen = so_gencnt;
841 error = SYSCTL_OUT(req, &xig, sizeof xig);
842 if (error)
843 return error;
844
845 inp_list = malloc(n * sizeof *inp_list, M_TEMP, M_WAITOK);
846 if (inp_list == 0)
847 return ENOMEM;
848
849 INP_INFO_RLOCK(&ripcbinfo);
850 for (inp = LIST_FIRST(ripcbinfo.listhead), i = 0; inp && i < n;
851 inp = LIST_NEXT(inp, inp_list)) {
852 INP_LOCK(inp);
853 if (inp->inp_gencnt <= gencnt &&
854 cr_canseesocket(req->td->td_ucred, inp->inp_socket) == 0) {
855 /* XXX held references? */
856 inp_list[i++] = inp;
857 }
858 INP_UNLOCK(inp);
859 }
860 INP_INFO_RUNLOCK(&ripcbinfo);
861 n = i;
862
863 error = 0;
864 for (i = 0; i < n; i++) {
865 inp = inp_list[i];
866 INP_LOCK(inp);
867 if (inp->inp_gencnt <= gencnt) {
868 struct xinpcb xi;
869 bzero(&xi, sizeof(xi));
870 xi.xi_len = sizeof xi;
871 /* XXX should avoid extra copy */
872 bcopy(inp, &xi.xi_inp, sizeof *inp);
873 if (inp->inp_socket)
874 sotoxsocket(inp->inp_socket, &xi.xi_socket);
875 INP_UNLOCK(inp);
876 error = SYSCTL_OUT(req, &xi, sizeof xi);
877 } else
878 INP_UNLOCK(inp);
879 }
880 if (!error) {
881 /*
882 * Give the user an updated idea of our state.
883 * If the generation differs from what we told
884 * her before, she knows that something happened
885 * while we were processing this request, and it
886 * might be necessary to retry.
887 */
888 INP_INFO_RLOCK(&ripcbinfo);
889 xig.xig_gen = ripcbinfo.ipi_gencnt;
890 xig.xig_sogen = so_gencnt;
891 xig.xig_count = ripcbinfo.ipi_count;
892 INP_INFO_RUNLOCK(&ripcbinfo);
893 error = SYSCTL_OUT(req, &xig, sizeof xig);
894 }
895 free(inp_list, M_TEMP);
896 return error;
897 }
898
899 /*
900 * This is the wrapper function for in_setsockaddr. We just pass down
901 * the pcbinfo for in_setpeeraddr to lock.
902 */
903 static int
904 rip_sockaddr(struct socket *so, struct sockaddr **nam)
905 {
906 return (in_setsockaddr(so, nam, &ripcbinfo));
907 }
908
909 /*
910 * This is the wrapper function for in_setpeeraddr. We just pass down
911 * the pcbinfo for in_setpeeraddr to lock.
912 */
913 static int
914 rip_peeraddr(struct socket *so, struct sockaddr **nam)
915 {
916 return (in_setpeeraddr(so, nam, &ripcbinfo));
917 }
918
919
920 SYSCTL_PROC(_net_inet_raw, OID_AUTO/*XXX*/, pcblist, CTLFLAG_RD, 0, 0,
921 rip_pcblist, "S,xinpcb", "List of active raw IP sockets");
922
923 struct pr_usrreqs rip_usrreqs = {
924 .pru_abort = rip_abort,
925 .pru_attach = rip_attach,
926 .pru_bind = rip_bind,
927 .pru_connect = rip_connect,
928 .pru_control = in_control,
929 .pru_detach = rip_detach,
930 .pru_disconnect = rip_disconnect,
931 .pru_peeraddr = rip_peeraddr,
932 .pru_send = rip_send,
933 .pru_shutdown = rip_shutdown,
934 .pru_sockaddr = rip_sockaddr,
935 .pru_sosetlabel = in_pcbsosetlabel
936 };
Cache object: f5e5da7a8c244c51f169c3e48025bc28
|