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: releng/5.3/sys/netinet/raw_ip.c 145954 2005-05-06 02:50:35Z cperciva $
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 * Nominal space allocated to a raw ip socket.
108 */
109 #define RIPSNDQ 8192
110 #define RIPRCVQ 8192
111
112 /*
113 * Raw interface to IP protocol.
114 */
115
116 /*
117 * Initialize raw connection block q.
118 */
119 void
120 rip_init()
121 {
122 INP_INFO_LOCK_INIT(&ripcbinfo, "rip");
123 LIST_INIT(&ripcb);
124 ripcbinfo.listhead = &ripcb;
125 /*
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.
129 */
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);
135 }
136
137 static struct sockaddr_in ripsrc = { sizeof(ripsrc), AF_INET };
138
139 static int
140 raw_append(struct inpcb *last, struct ip *ip, struct mbuf *n)
141 {
142 int policyfail = 0;
143
144 INP_LOCK_ASSERT(last);
145
146 #if defined(IPSEC) || defined(FAST_IPSEC)
147 /* check AH/ESP integrity. */
148 if (ipsec4_in_reject(n, last)) {
149 policyfail = 1;
150 #ifdef IPSEC
151 ipsecstat.in_polvio++;
152 #endif /*IPSEC*/
153 /* do not inject data to pcb */
154 }
155 #endif /*IPSEC || FAST_IPSEC*/
156 #ifdef MAC
157 if (!policyfail && mac_check_inpcb_deliver(last, n) != 0)
158 policyfail = 1;
159 #endif
160 if (!policyfail) {
161 struct mbuf *opts = NULL;
162 struct socket *so;
163
164 so = last->inp_socket;
165 if ((last->inp_flags & INP_CONTROLOPTS) ||
166 (so->so_options & SO_TIMESTAMP))
167 ip_savecontrol(last, &opts, ip, n);
168 SOCKBUF_LOCK(&so->so_rcv);
169 if (sbappendaddr_locked(&so->so_rcv,
170 (struct sockaddr *)&ripsrc, n, opts) == 0) {
171 /* should notify about lost packet */
172 m_freem(n);
173 if (opts)
174 m_freem(opts);
175 SOCKBUF_UNLOCK(&so->so_rcv);
176 } else
177 sorwakeup_locked(so);
178 } else
179 m_freem(n);
180 return policyfail;
181 }
182
183 /*
184 * Setup generic address and protocol structures
185 * for raw_input routine, then pass them along with
186 * mbuf chain.
187 */
188 void
189 rip_input(struct mbuf *m, int off)
190 {
191 struct ip *ip = mtod(m, struct ip *);
192 int proto = ip->ip_p;
193 struct inpcb *inp, *last;
194
195 INP_INFO_RLOCK(&ripcbinfo);
196 ripsrc.sin_addr = ip->ip_src;
197 last = NULL;
198 LIST_FOREACH(inp, &ripcb, inp_list) {
199 INP_LOCK(inp);
200 if (inp->inp_ip_p && inp->inp_ip_p != proto) {
201 docontinue:
202 INP_UNLOCK(inp);
203 continue;
204 }
205 #ifdef INET6
206 if ((inp->inp_vflag & INP_IPV4) == 0)
207 goto docontinue;
208 #endif
209 if (inp->inp_laddr.s_addr &&
210 inp->inp_laddr.s_addr != ip->ip_dst.s_addr)
211 goto docontinue;
212 if (inp->inp_faddr.s_addr &&
213 inp->inp_faddr.s_addr != ip->ip_src.s_addr)
214 goto docontinue;
215 if (jailed(inp->inp_socket->so_cred))
216 if (htonl(prison_getip(inp->inp_socket->so_cred)) !=
217 ip->ip_dst.s_addr)
218 goto docontinue;
219 if (last) {
220 struct mbuf *n;
221
222 n = m_copy(m, 0, (int)M_COPYALL);
223 if (n != NULL)
224 (void) raw_append(last, ip, n);
225 /* XXX count dropped packet */
226 INP_UNLOCK(last);
227 }
228 last = inp;
229 }
230 if (last != NULL) {
231 if (raw_append(last, ip, m) != 0)
232 ipstat.ips_delivered--;
233 INP_UNLOCK(last);
234 } else {
235 m_freem(m);
236 ipstat.ips_noproto++;
237 ipstat.ips_delivered--;
238 }
239 INP_INFO_RUNLOCK(&ripcbinfo);
240 }
241
242 /*
243 * Generate IP header and pass packet to ip_output.
244 * Tack on options user may have setup with control call.
245 */
246 int
247 rip_output(struct mbuf *m, struct socket *so, u_long dst)
248 {
249 struct ip *ip;
250 int error;
251 struct inpcb *inp = sotoinpcb(so);
252 int flags = (so->so_options & SO_DONTROUTE) | IP_ALLOWBROADCAST;
253
254 /*
255 * If the user handed us a complete IP packet, use it.
256 * Otherwise, allocate an mbuf for a header and fill it in.
257 */
258 if ((inp->inp_flags & INP_HDRINCL) == 0) {
259 if (m->m_pkthdr.len + sizeof(struct ip) > IP_MAXPACKET) {
260 m_freem(m);
261 return(EMSGSIZE);
262 }
263 M_PREPEND(m, sizeof(struct ip), M_DONTWAIT);
264 if (m == NULL)
265 return(ENOBUFS);
266
267 INP_LOCK(inp);
268 ip = mtod(m, struct ip *);
269 ip->ip_tos = inp->inp_ip_tos;
270 ip->ip_off = 0;
271 ip->ip_p = inp->inp_ip_p;
272 ip->ip_len = m->m_pkthdr.len;
273 if (jailed(inp->inp_socket->so_cred))
274 ip->ip_src.s_addr =
275 htonl(prison_getip(inp->inp_socket->so_cred));
276 else
277 ip->ip_src = inp->inp_laddr;
278 ip->ip_dst.s_addr = dst;
279 ip->ip_ttl = inp->inp_ip_ttl;
280 } else {
281 if (m->m_pkthdr.len > IP_MAXPACKET) {
282 m_freem(m);
283 return(EMSGSIZE);
284 }
285 INP_LOCK(inp);
286 ip = mtod(m, struct ip *);
287 if (jailed(inp->inp_socket->so_cred)) {
288 if (ip->ip_src.s_addr !=
289 htonl(prison_getip(inp->inp_socket->so_cred))) {
290 INP_UNLOCK(inp);
291 m_freem(m);
292 return (EPERM);
293 }
294 }
295 /* don't allow both user specified and setsockopt options,
296 and don't allow packet length sizes that will crash */
297 if (((ip->ip_hl != (sizeof (*ip) >> 2))
298 && inp->inp_options)
299 || (ip->ip_len > m->m_pkthdr.len)
300 || (ip->ip_len < (ip->ip_hl << 2))) {
301 INP_UNLOCK(inp);
302 m_freem(m);
303 return EINVAL;
304 }
305 if (ip->ip_id == 0)
306 ip->ip_id = ip_newid();
307 /* XXX prevent ip_output from overwriting header fields */
308 flags |= IP_RAWOUTPUT;
309 ipstat.ips_rawout++;
310 }
311
312 if (inp->inp_flags & INP_ONESBCAST)
313 flags |= IP_SENDONES;
314
315 #ifdef MAC
316 mac_create_mbuf_from_inpcb(inp, m);
317 #endif
318
319 error = ip_output(m, inp->inp_options, NULL, flags,
320 inp->inp_moptions, inp);
321 INP_UNLOCK(inp);
322 return error;
323 }
324
325 /*
326 * Raw IP socket option processing.
327 *
328 * IMPORTANT NOTE regarding access control: Traditionally, raw sockets could
329 * only be created by a privileged process, and as such, socket option
330 * operations to manage system properties on any raw socket were allowed to
331 * take place without explicit additional access control checks. However,
332 * raw sockets can now also be created in jail(), and therefore explicit
333 * checks are now required. Likewise, raw sockets can be used by a process
334 * after it gives up privilege, so some caution is required. For options
335 * passed down to the IP layer via ip_ctloutput(), checks are assumed to be
336 * performed in ip_ctloutput() and therefore no check occurs here.
337 * Unilaterally checking suser() here breaks normal IP socket option
338 * operations on raw sockets.
339 *
340 * When adding new socket options here, make sure to add access control
341 * checks here as necessary.
342 */
343 int
344 rip_ctloutput(struct socket *so, struct sockopt *sopt)
345 {
346 struct inpcb *inp = sotoinpcb(so);
347 int error, optval;
348
349 if (sopt->sopt_level != IPPROTO_IP)
350 return (EINVAL);
351
352 error = 0;
353 switch (sopt->sopt_dir) {
354 case SOPT_GET:
355 switch (sopt->sopt_name) {
356 case IP_HDRINCL:
357 optval = inp->inp_flags & INP_HDRINCL;
358 error = sooptcopyout(sopt, &optval, sizeof optval);
359 break;
360
361 case IP_FW_ADD: /* ADD actually returns the body... */
362 case IP_FW_GET:
363 case IP_FW_TABLE_GETSIZE:
364 case IP_FW_TABLE_LIST:
365 error = suser(curthread);
366 if (error != 0)
367 return (error);
368 if (ip_fw_ctl_ptr != NULL)
369 error = ip_fw_ctl_ptr(sopt);
370 else
371 error = ENOPROTOOPT;
372 break;
373
374 case IP_DUMMYNET_GET:
375 error = suser(curthread);
376 if (error != 0)
377 return (error);
378 if (ip_dn_ctl_ptr != NULL)
379 error = ip_dn_ctl_ptr(sopt);
380 else
381 error = ENOPROTOOPT;
382 break ;
383
384 case MRT_INIT:
385 case MRT_DONE:
386 case MRT_ADD_VIF:
387 case MRT_DEL_VIF:
388 case MRT_ADD_MFC:
389 case MRT_DEL_MFC:
390 case MRT_VERSION:
391 case MRT_ASSERT:
392 case MRT_API_SUPPORT:
393 case MRT_API_CONFIG:
394 case MRT_ADD_BW_UPCALL:
395 case MRT_DEL_BW_UPCALL:
396 error = suser(curthread);
397 if (error != 0)
398 return (error);
399 error = ip_mrouter_get ? ip_mrouter_get(so, sopt) :
400 EOPNOTSUPP;
401 break;
402
403 default:
404 error = ip_ctloutput(so, sopt);
405 break;
406 }
407 break;
408
409 case SOPT_SET:
410 switch (sopt->sopt_name) {
411 case IP_HDRINCL:
412 error = sooptcopyin(sopt, &optval, sizeof optval,
413 sizeof optval);
414 if (error)
415 break;
416 if (optval)
417 inp->inp_flags |= INP_HDRINCL;
418 else
419 inp->inp_flags &= ~INP_HDRINCL;
420 break;
421
422 case IP_FW_ADD:
423 case IP_FW_DEL:
424 case IP_FW_FLUSH:
425 case IP_FW_ZERO:
426 case IP_FW_RESETLOG:
427 case IP_FW_TABLE_ADD:
428 case IP_FW_TABLE_DEL:
429 case IP_FW_TABLE_FLUSH:
430 error = suser(curthread);
431 if (error != 0)
432 return (error);
433 if (ip_fw_ctl_ptr != NULL)
434 error = ip_fw_ctl_ptr(sopt);
435 else
436 error = ENOPROTOOPT;
437 break;
438
439 case IP_DUMMYNET_CONFIGURE:
440 case IP_DUMMYNET_DEL:
441 case IP_DUMMYNET_FLUSH:
442 error = suser(curthread);
443 if (error != 0)
444 return (error);
445 if (ip_dn_ctl_ptr != NULL)
446 error = ip_dn_ctl_ptr(sopt);
447 else
448 error = ENOPROTOOPT ;
449 break ;
450
451 case IP_RSVP_ON:
452 error = suser(curthread);
453 if (error != 0)
454 return (error);
455 error = ip_rsvp_init(so);
456 break;
457
458 case IP_RSVP_OFF:
459 error = suser(curthread);
460 if (error != 0)
461 return (error);
462 error = ip_rsvp_done();
463 break;
464
465 case IP_RSVP_VIF_ON:
466 case IP_RSVP_VIF_OFF:
467 error = suser(curthread);
468 if (error != 0)
469 return (error);
470 error = ip_rsvp_vif ?
471 ip_rsvp_vif(so, sopt) : EINVAL;
472 break;
473
474 case MRT_INIT:
475 case MRT_DONE:
476 case MRT_ADD_VIF:
477 case MRT_DEL_VIF:
478 case MRT_ADD_MFC:
479 case MRT_DEL_MFC:
480 case MRT_VERSION:
481 case MRT_ASSERT:
482 case MRT_API_SUPPORT:
483 case MRT_API_CONFIG:
484 case MRT_ADD_BW_UPCALL:
485 case MRT_DEL_BW_UPCALL:
486 error = suser(curthread);
487 if (error != 0)
488 return (error);
489 error = ip_mrouter_set ? ip_mrouter_set(so, sopt) :
490 EOPNOTSUPP;
491 break;
492
493 default:
494 error = ip_ctloutput(so, sopt);
495 break;
496 }
497 break;
498 }
499
500 return (error);
501 }
502
503 /*
504 * This function exists solely to receive the PRC_IFDOWN messages which
505 * are sent by if_down(). It looks for an ifaddr whose ifa_addr is sa,
506 * and calls in_ifadown() to remove all routes corresponding to that address.
507 * It also receives the PRC_IFUP messages from if_up() and reinstalls the
508 * interface routes.
509 */
510 void
511 rip_ctlinput(int cmd, struct sockaddr *sa, void *vip)
512 {
513 struct in_ifaddr *ia;
514 struct ifnet *ifp;
515 int err;
516 int flags;
517
518 switch (cmd) {
519 case PRC_IFDOWN:
520 TAILQ_FOREACH(ia, &in_ifaddrhead, ia_link) {
521 if (ia->ia_ifa.ifa_addr == sa
522 && (ia->ia_flags & IFA_ROUTE)) {
523 /*
524 * in_ifscrub kills the interface route.
525 */
526 in_ifscrub(ia->ia_ifp, ia);
527 /*
528 * in_ifadown gets rid of all the rest of
529 * the routes. This is not quite the right
530 * thing to do, but at least if we are running
531 * a routing process they will come back.
532 */
533 in_ifadown(&ia->ia_ifa, 0);
534 break;
535 }
536 }
537 break;
538
539 case PRC_IFUP:
540 TAILQ_FOREACH(ia, &in_ifaddrhead, ia_link) {
541 if (ia->ia_ifa.ifa_addr == sa)
542 break;
543 }
544 if (ia == 0 || (ia->ia_flags & IFA_ROUTE))
545 return;
546 flags = RTF_UP;
547 ifp = ia->ia_ifa.ifa_ifp;
548
549 if ((ifp->if_flags & IFF_LOOPBACK)
550 || (ifp->if_flags & IFF_POINTOPOINT))
551 flags |= RTF_HOST;
552
553 err = rtinit(&ia->ia_ifa, RTM_ADD, flags);
554 if (err == 0)
555 ia->ia_flags |= IFA_ROUTE;
556 break;
557 }
558 }
559
560 u_long rip_sendspace = RIPSNDQ;
561 u_long rip_recvspace = RIPRCVQ;
562
563 SYSCTL_INT(_net_inet_raw, OID_AUTO, maxdgram, CTLFLAG_RW,
564 &rip_sendspace, 0, "Maximum outgoing raw IP datagram size");
565 SYSCTL_INT(_net_inet_raw, OID_AUTO, recvspace, CTLFLAG_RW,
566 &rip_recvspace, 0, "Maximum space for incoming raw IP datagrams");
567
568 static int
569 rip_attach(struct socket *so, int proto, struct thread *td)
570 {
571 struct inpcb *inp;
572 int error;
573
574 /* XXX why not lower? */
575 INP_INFO_WLOCK(&ripcbinfo);
576 inp = sotoinpcb(so);
577 if (inp) {
578 /* XXX counter, printf */
579 INP_INFO_WUNLOCK(&ripcbinfo);
580 return EINVAL;
581 }
582 if (td && jailed(td->td_ucred) && !jail_allow_raw_sockets) {
583 INP_INFO_WUNLOCK(&ripcbinfo);
584 return (EPERM);
585 }
586 if (td && (error = suser_cred(td->td_ucred, SUSER_ALLOWJAIL)) != 0) {
587 INP_INFO_WUNLOCK(&ripcbinfo);
588 return error;
589 }
590 if (proto >= IPPROTO_MAX || proto < 0) {
591 INP_INFO_WUNLOCK(&ripcbinfo);
592 return EPROTONOSUPPORT;
593 }
594
595 error = soreserve(so, rip_sendspace, rip_recvspace);
596 if (error) {
597 INP_INFO_WUNLOCK(&ripcbinfo);
598 return error;
599 }
600 error = in_pcballoc(so, &ripcbinfo, "rawinp");
601 if (error) {
602 INP_INFO_WUNLOCK(&ripcbinfo);
603 return error;
604 }
605 inp = (struct inpcb *)so->so_pcb;
606 INP_LOCK(inp);
607 INP_INFO_WUNLOCK(&ripcbinfo);
608 inp->inp_vflag |= INP_IPV4;
609 inp->inp_ip_p = proto;
610 inp->inp_ip_ttl = ip_defttl;
611 INP_UNLOCK(inp);
612 return 0;
613 }
614
615 static void
616 rip_pcbdetach(struct socket *so, struct inpcb *inp)
617 {
618 INP_INFO_WLOCK_ASSERT(&ripcbinfo);
619 INP_LOCK_ASSERT(inp);
620
621 if (so == ip_mrouter && ip_mrouter_done)
622 ip_mrouter_done();
623 if (ip_rsvp_force_done)
624 ip_rsvp_force_done(so);
625 if (so == ip_rsvpd)
626 ip_rsvp_done();
627 in_pcbdetach(inp);
628 }
629
630 static int
631 rip_detach(struct socket *so)
632 {
633 struct inpcb *inp;
634
635 INP_INFO_WLOCK(&ripcbinfo);
636 inp = sotoinpcb(so);
637 if (inp == 0) {
638 /* XXX counter, printf */
639 INP_INFO_WUNLOCK(&ripcbinfo);
640 return EINVAL;
641 }
642 INP_LOCK(inp);
643 rip_pcbdetach(so, inp);
644 INP_INFO_WUNLOCK(&ripcbinfo);
645 return 0;
646 }
647
648 static int
649 rip_abort(struct socket *so)
650 {
651 struct inpcb *inp;
652
653 INP_INFO_WLOCK(&ripcbinfo);
654 inp = sotoinpcb(so);
655 if (inp == 0) {
656 INP_INFO_WUNLOCK(&ripcbinfo);
657 return EINVAL; /* ??? possible? panic instead? */
658 }
659 INP_LOCK(inp);
660 soisdisconnected(so);
661 if (so->so_state & SS_NOFDREF)
662 rip_pcbdetach(so, inp);
663 else
664 INP_UNLOCK(inp);
665 INP_INFO_WUNLOCK(&ripcbinfo);
666 return 0;
667 }
668
669 static int
670 rip_disconnect(struct socket *so)
671 {
672 if ((so->so_state & SS_ISCONNECTED) == 0)
673 return ENOTCONN;
674 return rip_abort(so);
675 }
676
677 static int
678 rip_bind(struct socket *so, struct sockaddr *nam, struct thread *td)
679 {
680 struct sockaddr_in *addr = (struct sockaddr_in *)nam;
681 struct inpcb *inp;
682
683 if (nam->sa_len != sizeof(*addr))
684 return EINVAL;
685
686 if (jailed(td->td_ucred)) {
687 if (addr->sin_addr.s_addr == INADDR_ANY)
688 addr->sin_addr.s_addr =
689 htonl(prison_getip(td->td_ucred));
690 if (htonl(prison_getip(td->td_ucred)) != addr->sin_addr.s_addr)
691 return (EADDRNOTAVAIL);
692 }
693
694 if (TAILQ_EMPTY(&ifnet) ||
695 (addr->sin_family != AF_INET && addr->sin_family != AF_IMPLINK) ||
696 (addr->sin_addr.s_addr &&
697 ifa_ifwithaddr((struct sockaddr *)addr) == 0))
698 return EADDRNOTAVAIL;
699
700 INP_INFO_WLOCK(&ripcbinfo);
701 inp = sotoinpcb(so);
702 if (inp == 0) {
703 INP_INFO_WUNLOCK(&ripcbinfo);
704 return EINVAL;
705 }
706 INP_LOCK(inp);
707 inp->inp_laddr = addr->sin_addr;
708 INP_UNLOCK(inp);
709 INP_INFO_WUNLOCK(&ripcbinfo);
710 return 0;
711 }
712
713 static int
714 rip_connect(struct socket *so, struct sockaddr *nam, struct thread *td)
715 {
716 struct sockaddr_in *addr = (struct sockaddr_in *)nam;
717 struct inpcb *inp;
718
719 if (nam->sa_len != sizeof(*addr))
720 return EINVAL;
721 if (TAILQ_EMPTY(&ifnet))
722 return EADDRNOTAVAIL;
723 if (addr->sin_family != AF_INET && addr->sin_family != AF_IMPLINK)
724 return EAFNOSUPPORT;
725
726 INP_INFO_WLOCK(&ripcbinfo);
727 inp = sotoinpcb(so);
728 if (inp == 0) {
729 INP_INFO_WUNLOCK(&ripcbinfo);
730 return EINVAL;
731 }
732 INP_LOCK(inp);
733 inp->inp_faddr = addr->sin_addr;
734 soisconnected(so);
735 INP_UNLOCK(inp);
736 INP_INFO_WUNLOCK(&ripcbinfo);
737 return 0;
738 }
739
740 static int
741 rip_shutdown(struct socket *so)
742 {
743 struct inpcb *inp;
744
745 INP_INFO_RLOCK(&ripcbinfo);
746 inp = sotoinpcb(so);
747 if (inp == 0) {
748 INP_INFO_RUNLOCK(&ripcbinfo);
749 return EINVAL;
750 }
751 INP_LOCK(inp);
752 INP_INFO_RUNLOCK(&ripcbinfo);
753 socantsendmore(so);
754 INP_UNLOCK(inp);
755 return 0;
756 }
757
758 static int
759 rip_send(struct socket *so, int flags, struct mbuf *m, struct sockaddr *nam,
760 struct mbuf *control, struct thread *td)
761 {
762 struct inpcb *inp;
763 u_long dst;
764 int ret;
765
766 INP_INFO_WLOCK(&ripcbinfo);
767 inp = sotoinpcb(so);
768 if (so->so_state & SS_ISCONNECTED) {
769 if (nam) {
770 INP_INFO_WUNLOCK(&ripcbinfo);
771 m_freem(m);
772 return EISCONN;
773 }
774 dst = inp->inp_faddr.s_addr;
775 } else {
776 if (nam == NULL) {
777 INP_INFO_WUNLOCK(&ripcbinfo);
778 m_freem(m);
779 return ENOTCONN;
780 }
781 dst = ((struct sockaddr_in *)nam)->sin_addr.s_addr;
782 }
783 ret = rip_output(m, so, dst);
784 INP_INFO_WUNLOCK(&ripcbinfo);
785 return ret;
786 }
787
788 static int
789 rip_pcblist(SYSCTL_HANDLER_ARGS)
790 {
791 int error, i, n;
792 struct inpcb *inp, **inp_list;
793 inp_gen_t gencnt;
794 struct xinpgen xig;
795
796 /*
797 * The process of preparing the TCB list is too time-consuming and
798 * resource-intensive to repeat twice on every request.
799 */
800 if (req->oldptr == 0) {
801 n = ripcbinfo.ipi_count;
802 req->oldidx = 2 * (sizeof xig)
803 + (n + n/8) * sizeof(struct xinpcb);
804 return 0;
805 }
806
807 if (req->newptr != 0)
808 return EPERM;
809
810 /*
811 * OK, now we're committed to doing something.
812 */
813 INP_INFO_RLOCK(&ripcbinfo);
814 gencnt = ripcbinfo.ipi_gencnt;
815 n = ripcbinfo.ipi_count;
816 INP_INFO_RUNLOCK(&ripcbinfo);
817
818 xig.xig_len = sizeof xig;
819 xig.xig_count = n;
820 xig.xig_gen = gencnt;
821 xig.xig_sogen = so_gencnt;
822 error = SYSCTL_OUT(req, &xig, sizeof xig);
823 if (error)
824 return error;
825
826 inp_list = malloc(n * sizeof *inp_list, M_TEMP, M_WAITOK);
827 if (inp_list == 0)
828 return ENOMEM;
829
830 INP_INFO_RLOCK(&ripcbinfo);
831 for (inp = LIST_FIRST(ripcbinfo.listhead), i = 0; inp && i < n;
832 inp = LIST_NEXT(inp, inp_list)) {
833 INP_LOCK(inp);
834 if (inp->inp_gencnt <= gencnt &&
835 cr_canseesocket(req->td->td_ucred, inp->inp_socket) == 0) {
836 /* XXX held references? */
837 inp_list[i++] = inp;
838 }
839 INP_UNLOCK(inp);
840 }
841 INP_INFO_RUNLOCK(&ripcbinfo);
842 n = i;
843
844 error = 0;
845 for (i = 0; i < n; i++) {
846 inp = inp_list[i];
847 if (inp->inp_gencnt <= gencnt) {
848 struct xinpcb xi;
849 bzero(&xi, sizeof(xi));
850 xi.xi_len = sizeof xi;
851 /* XXX should avoid extra copy */
852 bcopy(inp, &xi.xi_inp, sizeof *inp);
853 if (inp->inp_socket)
854 sotoxsocket(inp->inp_socket, &xi.xi_socket);
855 error = SYSCTL_OUT(req, &xi, sizeof xi);
856 }
857 }
858 if (!error) {
859 /*
860 * Give the user an updated idea of our state.
861 * If the generation differs from what we told
862 * her before, she knows that something happened
863 * while we were processing this request, and it
864 * might be necessary to retry.
865 */
866 INP_INFO_RLOCK(&ripcbinfo);
867 xig.xig_gen = ripcbinfo.ipi_gencnt;
868 xig.xig_sogen = so_gencnt;
869 xig.xig_count = ripcbinfo.ipi_count;
870 INP_INFO_RUNLOCK(&ripcbinfo);
871 error = SYSCTL_OUT(req, &xig, sizeof xig);
872 }
873 free(inp_list, M_TEMP);
874 return error;
875 }
876
877 /*
878 * This is the wrapper function for in_setsockaddr. We just pass down
879 * the pcbinfo for in_setpeeraddr to lock.
880 */
881 static int
882 rip_sockaddr(struct socket *so, struct sockaddr **nam)
883 {
884 return (in_setsockaddr(so, nam, &ripcbinfo));
885 }
886
887 /*
888 * This is the wrapper function for in_setpeeraddr. We just pass down
889 * the pcbinfo for in_setpeeraddr to lock.
890 */
891 static int
892 rip_peeraddr(struct socket *so, struct sockaddr **nam)
893 {
894 return (in_setpeeraddr(so, nam, &ripcbinfo));
895 }
896
897
898 SYSCTL_PROC(_net_inet_raw, OID_AUTO/*XXX*/, pcblist, CTLFLAG_RD, 0, 0,
899 rip_pcblist, "S,xinpcb", "List of active raw IP sockets");
900
901 struct pr_usrreqs rip_usrreqs = {
902 rip_abort, pru_accept_notsupp, rip_attach, rip_bind, rip_connect,
903 pru_connect2_notsupp, in_control, rip_detach, rip_disconnect,
904 pru_listen_notsupp, rip_peeraddr, pru_rcvd_notsupp,
905 pru_rcvoob_notsupp, rip_send, pru_sense_null, rip_shutdown,
906 rip_sockaddr, sosend, soreceive, sopoll, in_pcbsosetlabel
907 };
Cache object: a9ac4ecefc734175b9f4a62049d4df9f
|