VirtualBox

source: vbox/trunk/src/VBox/Devices/Network/slirp/ip_icmp.c@ 52343

Last change on this file since 52343 was 50986, checked in by vboxsync, 11 years ago

NAT: Update function comment for icmp_error to reflect r93155 (it was
already out of sync with the code after r91460).

  • Property svn:eol-style set to native
  • Property svn:keywords set to Author Date Id Revision
File size: 25.3 KB
Line 
1/* $Id: ip_icmp.c 50986 2014-04-07 16:12:55Z vboxsync $ */
2/** @file
3 * NAT - IP/ICMP handling.
4 */
5
6/*
7 * Copyright (C) 2006-2012 Oracle Corporation
8 *
9 * This file is part of VirtualBox Open Source Edition (OSE), as
10 * available from http://www.virtualbox.org. This file is free software;
11 * you can redistribute it and/or modify it under the terms of the GNU
12 * General Public License (GPL) as published by the Free Software
13 * Foundation, in version 2 as it comes in the "COPYING" file of the
14 * VirtualBox OSE distribution. VirtualBox OSE is distributed in the
15 * hope that it will be useful, but WITHOUT ANY WARRANTY of any kind.
16 */
17
18/*
19 * This code is based on:
20 *
21 * Copyright (c) 1982, 1986, 1988, 1993
22 * The Regents of the University of California. All rights reserved.
23 *
24 * Redistribution and use in source and binary forms, with or without
25 * modification, are permitted provided that the following conditions
26 * are met:
27 * 1. Redistributions of source code must retain the above copyright
28 * notice, this list of conditions and the following disclaimer.
29 * 2. Redistributions in binary form must reproduce the above copyright
30 * notice, this list of conditions and the following disclaimer in the
31 * documentation and/or other materials provided with the distribution.
32 * 3. All advertising materials mentioning features or use of this software
33 * must display the following acknowledgement:
34 * This product includes software developed by the University of
35 * California, Berkeley and its contributors.
36 * 4. Neither the name of the University nor the names of its contributors
37 * may be used to endorse or promote products derived from this software
38 * without specific prior written permission.
39 *
40 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
41 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
42 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
43 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
44 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
45 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
46 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
47 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
48 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
49 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
50 * SUCH DAMAGE.
51 *
52 * @(#)ip_icmp.c 8.2 (Berkeley) 1/4/94
53 * ip_icmp.c,v 1.7 1995/05/30 08:09:42 rgrimes Exp
54 */
55
56#include "slirp.h"
57#include "ip_icmp.h"
58#ifdef RT_OS_WINDOWS
59# include <Icmpapi.h>
60# include <Iphlpapi.h>
61# include <iprt/ldr.h>
62#endif
63
64/* The message sent when emulating PING */
65/* Be nice and tell them it's just a psuedo-ping packet */
66static const char icmp_ping_msg[] = "This is a psuedo-PING packet used by Slirp to emulate ICMP ECHO-REQUEST packets.\n";
67
68/* list of actions for icmp_error() on RX of an icmp message */
69static const int icmp_flush[19] =
70{
71/* ECHO REPLY (0) */ 0,
72 1,
73 1,
74/* DEST UNREACH (3) */ 1,
75/* SOURCE QUENCH (4)*/ 1,
76/* REDIRECT (5) */ 1,
77 1,
78 1,
79/* ECHO (8) */ 0,
80/* ROUTERADVERT (9) */ 1,
81/* ROUTERSOLICIT (10) */ 1,
82/* TIME EXCEEDED (11) */ 1,
83/* PARAMETER PROBLEM (12) */ 1,
84/* TIMESTAMP (13) */ 0,
85/* TIMESTAMP REPLY (14) */ 0,
86/* INFO (15) */ 0,
87/* INFO REPLY (16) */ 0,
88/* ADDR MASK (17) */ 0,
89/* ADDR MASK REPLY (18) */ 0
90};
91
92static void icmp_cache_clean(PNATState pData, int iEntries);
93
94int
95icmp_init(PNATState pData, int iIcmpCacheLimit)
96{
97 pData->icmp_socket.so_type = IPPROTO_ICMP;
98 pData->icmp_socket.so_state = SS_ISFCONNECTED;
99 if (iIcmpCacheLimit < 0)
100 {
101 LogRel(("NAT: iIcmpCacheLimit is invalid %d, will be alter to default value 100\n", iIcmpCacheLimit));
102 iIcmpCacheLimit = 100;
103 }
104 pData->iIcmpCacheLimit = iIcmpCacheLimit;
105#ifndef RT_OS_WINDOWS
106# ifndef RT_OS_DARWIN
107 pData->icmp_socket.s = socket(PF_INET, SOCK_RAW, IPPROTO_ICMP);
108# else /* !RT_OS_DARWIN */
109 pData->icmp_socket.s = socket(AF_INET, SOCK_DGRAM, IPPROTO_ICMP);
110# endif /* RT_OS_DARWIN */
111 if (pData->icmp_socket.s == -1)
112 {
113 int rc = RTErrConvertFromErrno(errno);
114 LogRel(("NAT: ICMP/ping not available (could not open ICMP socket, error %Rrc)\n", rc));
115 return 1;
116 }
117 fd_nonblock(pData->icmp_socket.s);
118 NSOCK_INC();
119
120#else /* RT_OS_WINDOWS */
121 /* Resolve symbols we need. */
122 {
123 RTLDRMOD hLdrMod;
124 int rc = RTLdrLoadSystem("Iphlpapi.dll", true /*fNoUnload*/, &hLdrMod);
125 if (RT_SUCCESS(rc))
126 {
127 pData->pfIcmpParseReplies = (long (WINAPI *)(void *, long))RTLdrGetFunction(hLdrMod, "IcmpParseReplies");
128 pData->pfIcmpCloseHandle = (BOOL (WINAPI *)(HANDLE))RTLdrGetFunction(hLdrMod, "IcmpCloseHandle");
129 rc = RTLdrGetSymbol(hLdrMod, "GetAdaptersAddresses", (void **)&pData->pfGetAdaptersAddresses);
130 if (RT_FAILURE(rc))
131 LogRel(("NAT: Can't find GetAdapterAddresses in Iphlpapi.dll\n"));
132 RTLdrClose(hLdrMod);
133 }
134
135 if (pData->pfIcmpParseReplies == NULL)
136 {
137 int rc = RTLdrLoadSystem("Icmp.dll", true /*fNoUnload*/, &hLdrMod);
138 if (RT_FAILURE(rc))
139 {
140 LogRel(("NAT: Icmp.dll could not be loaded: %Rrc\n", rc));
141 return 1;
142 }
143 pData->pfIcmpParseReplies = (long (WINAPI *)(void *, long))RTLdrGetFunction(hLdrMod, "IcmpParseReplies");
144 pData->pfIcmpCloseHandle = (BOOL (WINAPI *)(HANDLE))RTLdrGetFunction(hLdrMod, "IcmpCloseHandle");
145 RTLdrClose(hLdrMod);
146 }
147 }
148 if (pData->pfIcmpParseReplies == NULL)
149 {
150 LogRel(("NAT: Can't find IcmpParseReplies symbol\n"));
151 return 1;
152 }
153 if (pData->pfIcmpCloseHandle == NULL)
154 {
155 LogRel(("NAT: Can't find IcmpCloseHandle symbol\n"));
156 return 1;
157 }
158
159 pData->icmp_socket.sh = IcmpCreateFile();
160 pData->phEvents[VBOX_ICMP_EVENT_INDEX] = CreateEvent(NULL, FALSE, FALSE, NULL);
161 pData->cbIcmpBuffer = sizeof(ICMP_ECHO_REPLY) * 10;
162 pData->pvIcmpBuffer = RTMemAlloc(pData->cbIcmpBuffer);
163#endif /* RT_OS_WINDOWS */
164
165 LIST_INIT(&pData->icmp_msg_head);
166 return 0;
167}
168
169/**
170 * Cleans ICMP cache.
171 */
172void
173icmp_finit(PNATState pData)
174{
175 icmp_cache_clean(pData, -1);
176#ifdef RT_OS_WINDOWS
177 pData->pfIcmpCloseHandle(pData->icmp_socket.sh);
178 RTMemFree(pData->pvIcmpBuffer);
179#else
180 closesocket(pData->icmp_socket.s);
181#endif
182}
183
184/*
185 * ip here is ip header + 64bytes readed from ICMP packet
186 */
187struct icmp_msg *
188icmp_find_original_mbuf(PNATState pData, struct ip *ip)
189{
190 struct mbuf *m0;
191 struct ip *ip0;
192 struct icmp *icp, *icp0;
193 struct icmp_msg *icm = NULL;
194 int found = 0;
195 struct udphdr *udp;
196 struct tcphdr *tcp;
197 struct socket *head_socket = NULL;
198 struct socket *last_socket = NULL;
199 struct socket *so = NULL;
200 struct in_addr faddr;
201 u_short lport, fport;
202
203 faddr.s_addr = ~0;
204
205 lport = ~0;
206 fport = ~0;
207
208
209 LogFlowFunc(("ENTER: ip->ip_p:%d\n", ip->ip_p));
210 switch (ip->ip_p)
211 {
212 case IPPROTO_ICMP:
213 icp = (struct icmp *)((char *)ip + (ip->ip_hl << 2));
214 LIST_FOREACH(icm, &pData->icmp_msg_head, im_list)
215 {
216 m0 = icm->im_m;
217 ip0 = mtod(m0, struct ip *);
218 if (ip0->ip_p != IPPROTO_ICMP)
219 {
220 /* try next item */
221 continue;
222 }
223 icp0 = (struct icmp *)((char *)ip0 + (ip0->ip_hl << 2));
224 /*
225 * IP could pointer to ICMP_REPLY datagram (1)
226 * or pointer IP header in ICMP payload in case of
227 * ICMP_TIMXCEED or ICMP_UNREACH (2)
228 *
229 * if (1) and then ICMP (type should be ICMP_ECHOREPLY) and we need check that
230 * IP.IP_SRC == IP0.IP_DST received datagramm comes from destination.
231 *
232 * if (2) then check that payload ICMP has got type ICMP_ECHO and
233 * IP.IP_DST == IP0.IP_DST destination of returned datagram is the same as
234 * one was sent.
235 */
236 if ( ( (icp->icmp_type != ICMP_ECHO && ip->ip_src.s_addr == ip0->ip_dst.s_addr)
237 || (icp->icmp_type == ICMP_ECHO && ip->ip_dst.s_addr == ip0->ip_dst.s_addr))
238 && icp->icmp_id == icp0->icmp_id
239 && icp->icmp_seq == icp0->icmp_seq)
240 {
241 found = 1;
242 Log(("Have found %R[natsock]\n", icm->im_so));
243 break;
244 }
245 Log(("Have found nothing\n"));
246 }
247 break;
248
249 /*
250 * for TCP and UDP logic little bit reverted, we try to find the HOST socket
251 * from which the IP package has been sent.
252 */
253 case IPPROTO_UDP:
254 head_socket = &udb;
255 udp = (struct udphdr *)((char *)ip + (ip->ip_hl << 2));
256 faddr.s_addr = ip->ip_dst.s_addr;
257 fport = udp->uh_dport;
258 lport = udp->uh_sport;
259 last_socket = udp_last_so;
260 /* fall through */
261
262 case IPPROTO_TCP:
263 if (head_socket == NULL)
264 {
265 tcp = (struct tcphdr *)((char *)ip + (ip->ip_hl << 2));
266 head_socket = &tcb; /* head_socket could be initialized with udb*/
267 faddr.s_addr = ip->ip_dst.s_addr;
268 fport = tcp->th_dport;
269 lport = tcp->th_sport;
270 last_socket = tcp_last_so;
271 }
272 /* check last socket first */
273 if ( last_socket->so_faddr.s_addr == faddr.s_addr
274 && last_socket->so_fport == fport
275 && last_socket->so_hlport == lport)
276 {
277 found = 1;
278 so = last_socket;
279 goto sofound;
280 }
281 for (so = head_socket->so_prev; so != head_socket; so = so->so_prev)
282 {
283 /* Should be reaplaced by hash here */
284 Log(("trying:%R[natsock] against %RTnaipv4:%d lport=%d hlport=%d\n", so, &faddr, fport, lport, so->so_hlport));
285 if ( so->so_faddr.s_addr == faddr.s_addr
286 && so->so_fport == fport
287 && so->so_hlport == lport)
288 {
289 found = 1;
290 break;
291 }
292 }
293 break;
294
295 default:
296 Log(("NAT:ICMP: unsupported protocol(%d)\n", ip->ip_p));
297 }
298 sofound:
299 if (found == 1 && icm == NULL)
300 {
301 if (so->so_state == SS_NOFDREF)
302 {
303 /* socket is shutdowning we've already sent ICMP on it.*/
304 Log(("NAT: Received icmp on shutdowning socket (probably corresponding ICMP socket has been already sent)\n"));
305 return NULL;
306 }
307 icm = RTMemAlloc(sizeof(struct icmp_msg));
308 icm->im_m = so->so_m;
309 icm->im_so = so;
310 found = 1;
311 Log(("hit:%R[natsock]\n", so));
312 /*XXX: this storage not very long,
313 * better add flag if it should removed from lis
314 */
315 LIST_INSERT_HEAD(&pData->icmp_msg_head, icm, im_list);
316 pData->cIcmpCacheSize++;
317 if (pData->cIcmpCacheSize > pData->iIcmpCacheLimit)
318 icmp_cache_clean(pData, pData->iIcmpCacheLimit/2);
319 LogFlowFunc(("LEAVE: icm:%p\n", icm));
320 return (icm);
321 }
322 if (found == 1)
323 {
324 LogFlowFunc(("LEAVE: icm:%p\n", icm));
325 return icm;
326 }
327
328 LogFlowFunc(("LEAVE: NULL\n"));
329 return NULL;
330}
331
332/**
333 * iEntries how many entries to leave, if iEntries < 0, clean all
334 */
335static void icmp_cache_clean(PNATState pData, int iEntries)
336{
337 int iIcmpCount = 0;
338 struct icmp_msg *icm = NULL;
339 LogFlowFunc(("iEntries:%d\n", iEntries));
340 if (iEntries > pData->cIcmpCacheSize)
341 {
342 LogFlowFuncLeave();
343 return;
344 }
345 while(!LIST_EMPTY(&pData->icmp_msg_head))
346 {
347 icm = LIST_FIRST(&pData->icmp_msg_head);
348 if ( iEntries > 0
349 && iIcmpCount < iEntries)
350 {
351 iIcmpCount++;
352 continue;
353 }
354
355 LIST_REMOVE(icm, im_list);
356 if (icm->im_m)
357 {
358 pData->cIcmpCacheSize--;
359 m_freem(pData, icm->im_m);
360 }
361 RTMemFree(icm);
362 }
363 LogFlowFuncLeave();
364}
365
366static int
367icmp_attach(PNATState pData, struct mbuf *m)
368{
369 struct icmp_msg *icm;
370 struct ip *ip;
371 ip = mtod(m, struct ip *);
372 Assert(ip->ip_p == IPPROTO_ICMP);
373 icm = RTMemAlloc(sizeof(struct icmp_msg));
374 icm->im_m = m;
375 icm->im_so = m->m_so;
376 LIST_INSERT_HEAD(&pData->icmp_msg_head, icm, im_list);
377 pData->cIcmpCacheSize++;
378 if (pData->cIcmpCacheSize > pData->iIcmpCacheLimit)
379 icmp_cache_clean(pData, pData->iIcmpCacheLimit/2);
380 return 0;
381}
382
383/*
384 * Process a received ICMP message.
385 */
386void
387icmp_input(PNATState pData, struct mbuf *m, int hlen)
388{
389 register struct icmp *icp;
390 void *icp_buf = NULL;
391 register struct ip *ip = mtod(m, struct ip *);
392 int icmplen = ip->ip_len;
393 int status;
394 uint32_t dst;
395#if !defined(RT_OS_WINDOWS)
396 int ttl;
397#endif
398
399 /* int code; */
400
401 LogFlowFunc(("ENTER: m = %lx, m_len = %d\n", (long)m, m ? m->m_len : 0));
402
403 icmpstat.icps_received++;
404
405 /*
406 * Locate icmp structure in mbuf, and check
407 * that its not corrupted and of at least minimum length.
408 */
409 if (icmplen < ICMP_MINLEN)
410 {
411 /* min 8 bytes payload */
412 icmpstat.icps_tooshort++;
413 goto end_error_free_m;
414 }
415
416 m->m_len -= hlen;
417 m->m_data += hlen;
418
419 if (cksum(m, icmplen))
420 {
421 icmpstat.icps_checksum++;
422 goto end_error_free_m;
423 }
424
425 if (m->m_next)
426 {
427 icp_buf = RTMemAlloc(icmplen);
428 if (!icp_buf)
429 {
430 Log(("NAT: not enought memory to allocate the buffer\n"));
431 goto end_error_free_m;
432 }
433 m_copydata(m, 0, icmplen, icp_buf);
434 icp = (struct icmp *)icp_buf;
435 }
436 else
437 icp = mtod(m, struct icmp *);
438
439 m->m_len += hlen;
440 m->m_data -= hlen;
441
442 /* icmpstat.icps_inhist[icp->icmp_type]++; */
443 /* code = icp->icmp_code; */
444
445 LogFlow(("icmp_type = %d\n", icp->icmp_type));
446 switch (icp->icmp_type)
447 {
448 case ICMP_ECHO:
449 ip->ip_len += hlen; /* since ip_input subtracts this */
450 dst = ip->ip_dst.s_addr;
451 if ( CTL_CHECK(dst, CTL_ALIAS)
452 || CTL_CHECK(dst, CTL_DNS)
453 || CTL_CHECK(dst, CTL_TFTP))
454 {
455 icp->icmp_type = ICMP_ECHOREPLY;
456 ip->ip_dst.s_addr = ip->ip_src.s_addr;
457 ip->ip_src.s_addr = dst;
458 icmp_reflect(pData, m);
459 goto done;
460 }
461 else
462 {
463 struct sockaddr_in addr;
464#ifdef RT_OS_WINDOWS
465 IP_OPTION_INFORMATION ipopt;
466 int error;
467#endif
468 addr.sin_family = AF_INET;
469 if ((ip->ip_dst.s_addr & RT_H2N_U32(pData->netmask)) == pData->special_addr.s_addr)
470 {
471 /* It's an alias */
472 switch (RT_N2H_U32(ip->ip_dst.s_addr) & ~pData->netmask)
473 {
474 case CTL_DNS:
475 case CTL_ALIAS:
476 default:
477 addr.sin_addr = loopback_addr;
478 break;
479 }
480 }
481 else
482 addr.sin_addr.s_addr = ip->ip_dst.s_addr;
483#ifndef RT_OS_WINDOWS
484 if (pData->icmp_socket.s != -1)
485 {
486 ssize_t rc;
487 static bool fIcmpSocketErrorReported;
488 ttl = ip->ip_ttl;
489 Log(("NAT/ICMP: try to set TTL(%d)\n", ttl));
490 status = setsockopt(pData->icmp_socket.s, IPPROTO_IP, IP_TTL,
491 (void *)&ttl, sizeof(ttl));
492 if (status < 0)
493 Log(("NAT: Error (%s) occurred while setting TTL attribute of IP packet\n",
494 strerror(errno)));
495 rc = sendto(pData->icmp_socket.s, icp, icmplen, 0,
496 (struct sockaddr *)&addr, sizeof(addr));
497 if (rc >= 0)
498 {
499 m->m_so = &pData->icmp_socket;
500 icmp_attach(pData, m);
501 /* don't let m_freem at the end free atached buffer */
502 goto done;
503 }
504
505
506 if (!fIcmpSocketErrorReported)
507 {
508 LogRel(("icmp_input udp sendto tx errno = %d (%s)\n",
509 errno, strerror(errno)));
510 fIcmpSocketErrorReported = true;
511 }
512 icmp_error(pData, m, ICMP_UNREACH, ICMP_UNREACH_NET, 0, strerror(errno));
513 }
514#else /* RT_OS_WINDOWS */
515 pData->icmp_socket.so_laddr.s_addr = ip->ip_src.s_addr; /* XXX: hack*/
516 pData->icmp_socket.so_icmp_id = icp->icmp_id;
517 pData->icmp_socket.so_icmp_seq = icp->icmp_seq;
518 memset(&ipopt, 0, sizeof(IP_OPTION_INFORMATION));
519 ipopt.Ttl = ip->ip_ttl;
520 status = IcmpSendEcho2(pData->icmp_socket.sh /*=handle*/,
521 pData->phEvents[VBOX_ICMP_EVENT_INDEX] /*=Event*/,
522 NULL /*=ApcRoutine*/,
523 NULL /*=ApcContext*/,
524 addr.sin_addr.s_addr /*=DestinationAddress*/,
525 icp->icmp_data /*=RequestData*/,
526 icmplen - ICMP_MINLEN /*=RequestSize*/,
527 &ipopt /*=RequestOptions*/,
528 pData->pvIcmpBuffer /*=ReplyBuffer*/,
529 pData->cbIcmpBuffer /*=ReplySize*/,
530 1 /*=Timeout in ms*/);
531 error = GetLastError();
532 if ( status != 0
533 || error == ERROR_IO_PENDING)
534 {
535 /* no error! */
536 m->m_so = &pData->icmp_socket;
537 icmp_attach(pData, m);
538 /* don't let m_freem at the end free atached buffer */
539 goto done;
540 }
541 Log(("NAT: Error (%d) occurred while sending ICMP (", error));
542 switch (error)
543 {
544 case ERROR_INVALID_PARAMETER:
545 Log(("icmp_socket:%lx is invalid)\n", pData->icmp_socket.s));
546 break;
547 case ERROR_NOT_SUPPORTED:
548 Log(("operation is unsupported)\n"));
549 break;
550 case ERROR_NOT_ENOUGH_MEMORY:
551 Log(("OOM!!!)\n"));
552 break;
553 case IP_BUF_TOO_SMALL:
554 Log(("Buffer too small)\n"));
555 break;
556 default:
557 Log(("Other error!!!)\n"));
558 break;
559 }
560#endif /* RT_OS_WINDOWS */
561 } /* if ip->ip_dst.s_addr == alias_addr.s_addr */
562 break;
563 case ICMP_UNREACH:
564 case ICMP_TIMXCEED:
565 /* @todo(vvl): both up cases comes from guest,
566 * indeed right solution would be find the socket
567 * corresponding to ICMP data and close it.
568 */
569 case ICMP_PARAMPROB:
570 case ICMP_SOURCEQUENCH:
571 case ICMP_TSTAMP:
572 case ICMP_MASKREQ:
573 case ICMP_REDIRECT:
574 icmpstat.icps_notsupp++;
575 break;
576
577 default:
578 icmpstat.icps_badtype++;
579 } /* switch */
580
581end_error_free_m:
582 m_freem(pData, m);
583
584done:
585 if (icp_buf)
586 RTMemFree(icp_buf);
587}
588
589
590/**
591 * Send an ICMP message in response to a situation
592 *
593 * RFC 1122: 3.2.2 MUST send at least the IP header and 8 bytes of header. MAY send more (we do).
594 * MUST NOT change this header information.
595 * MUST NOT reply to a multicast/broadcast IP address.
596 * MUST NOT reply to a multicast/broadcast MAC address.
597 * MUST reply to only the first fragment.
598 *
599 * Send ICMP_UNREACH back to the source regarding msrc.
600 * It is reported as the bad ip packet. The header should
601 * be fully correct and in host byte order.
602 * ICMP fragmentation is illegal.
603 *
604 * @note: implementation note: MSIZE is 256 bytes (minimal buffer).
605 * We always truncate original payload to 8 bytes required by the RFC,
606 * so the largest possible datagram is 14 (ethernet) + 20 (ip) +
607 * 8 (icmp) + 60 (max original ip with options) + 8 (original payload)
608 * = 110 bytes which fits into sinlge mbuf.
609 *
610 * @note This function will free msrc!
611 */
612
613void icmp_error(PNATState pData, struct mbuf *msrc, u_char type, u_char code, int minsize, const char *message)
614{
615 unsigned ohlen, olen;
616 struct mbuf *m;
617 struct ip *oip, *ip;
618 struct icmp *icp;
619 void *payload;
620
621 LogFlow(("icmp_error: msrc = %p, msrc_len = %d\n",
622 (void *)msrc, msrc ? msrc->m_len : 0));
623
624 if (RT_UNLIKELY(msrc == NULL))
625 goto end_error;
626
627 M_ASSERTPKTHDR(msrc);
628
629 if ( type != ICMP_UNREACH
630 && type != ICMP_TIMXCEED
631 && type != ICMP_SOURCEQUENCH)
632 goto end_error;
633
634 oip = mtod(msrc, struct ip *);
635 LogFunc(("msrc: %RTnaipv4 -> %RTnaipv4\n", oip->ip_src, oip->ip_dst));
636
637 if (oip->ip_src.s_addr == INADDR_ANY)
638 goto end_error;
639
640 if (oip->ip_off & IP_OFFMASK)
641 goto end_error; /* Only reply to fragment 0 */
642
643 ohlen = oip->ip_hl * 4;
644 AssertStmt(ohlen >= sizeof(struct ip), goto end_error);
645
646 olen = oip->ip_len;
647 AssertStmt(olen >= ohlen, goto end_error);
648
649 if (oip->ip_p == IPPROTO_ICMP)
650 {
651 struct icmp *oicp = (struct icmp *)((char *)oip + ohlen);
652 /*
653 * Assume any unknown ICMP type is an error. This isn't
654 * specified by the RFC, but think about it..
655 */
656 if (oicp->icmp_type > ICMP_MAXTYPE || icmp_flush[oicp->icmp_type])
657 goto end_error;
658 }
659
660 /* undo byte order conversions done in ip_input() */
661 HTONS(oip->ip_len);
662 HTONS(oip->ip_id);
663 HTONS(oip->ip_off);
664
665 m = m_gethdr(pData, M_NOWAIT, MT_HEADER);
666 if (RT_UNLIKELY(m == NULL))
667 goto end_error;
668
669 m->m_flags |= M_SKIP_FIREWALL;
670 m->m_data += if_maxlinkhdr;
671
672 ip = mtod(m, struct ip *);
673 m->m_pkthdr.header = (void *)ip;
674
675 /* fill in ip (ip_output0() does the boilerplate for us) */
676 ip->ip_tos = ((oip->ip_tos & 0x1E) | 0xC0); /* high priority for errors */
677 /* ip->ip_len will be set later */
678 ip->ip_off = 0;
679 ip->ip_ttl = MAXTTL;
680 ip->ip_p = IPPROTO_ICMP;
681 ip->ip_src = alias_addr;
682 ip->ip_dst = oip->ip_src;
683
684 /* fill in icmp */
685 icp = (struct icmp *)((char *)ip + sizeof(*ip));
686 icp->icmp_type = type;
687 icp->icmp_code = code;
688 icp->icmp_id = 0;
689 icp->icmp_seq = 0;
690
691 /* fill in icmp payload: original ip header plus 8 bytes of its payload */
692 if (olen > ohlen + 8)
693 olen = ohlen + 8;
694 payload = (void *)((char *)icp + ICMP_MINLEN);
695 memcpy(payload, oip, olen);
696
697 /*
698 * Original code appended this message after the payload. This
699 * might have been a good idea for real slirp, as it provided a
700 * communication channel with the remote host. But 90s are over.
701 */
702 NOREF(message);
703
704 /* hide ip header for icmp checksum calculation */
705 m->m_data += sizeof(struct ip);
706 m->m_len = ICMP_MINLEN + /* truncated */ olen;
707
708 icp->icmp_cksum = 0;
709 icp->icmp_cksum = cksum(m, m->m_len);
710
711 /* reveal ip header */
712 m->m_data -= sizeof(struct ip);
713 m->m_len += sizeof(struct ip);
714 ip->ip_len = m->m_len;
715
716 (void) ip_output0(pData, (struct socket *)NULL, m, 1);
717
718 icmpstat.icps_reflect++;
719
720 /* clear source datagramm in positive branch */
721 m_freem(pData, msrc);
722 LogFlowFuncLeave();
723 return;
724
725end_error:
726
727 /*
728 * clear source datagramm in case if some of requirement haven't been met.
729 */
730 if (msrc)
731 m_freem(pData, msrc);
732
733 {
734 static bool fIcmpErrorReported;
735 if (!fIcmpErrorReported)
736 {
737 LogRel(("NAT: error occurred while sending ICMP error message\n"));
738 fIcmpErrorReported = true;
739 }
740 }
741 LogFlowFuncLeave();
742}
743
744/*
745 * Reflect the ip packet back to the source
746 * Note: m isn't duplicated by this method and more delivered to ip_output then.
747 */
748void
749icmp_reflect(PNATState pData, struct mbuf *m)
750{
751 register struct ip *ip = mtod(m, struct ip *);
752 int hlen = ip->ip_hl << 2;
753 register struct icmp *icp;
754 LogFlowFunc(("ENTER: m:%p\n", m));
755
756 /*
757 * Send an icmp packet back to the ip level,
758 * after supplying a checksum.
759 */
760 m->m_data += hlen;
761 m->m_len -= hlen;
762 icp = mtod(m, struct icmp *);
763
764 icp->icmp_cksum = 0;
765 icp->icmp_cksum = cksum(m, ip->ip_len - hlen);
766
767 m->m_data -= hlen;
768 m->m_len += hlen;
769
770 (void) ip_output(pData, (struct socket *)NULL, m);
771
772 icmpstat.icps_reflect++;
773 LogFlowFuncLeave();
774}
Note: See TracBrowser for help on using the repository browser.

© 2024 Oracle Support Privacy / Do Not Sell My Info Terms of Use Trademark Policy Automated Access Etiquette