xref: /dragonfly/usr.sbin/vknetd/vknetd.c (revision 25a2db75)
1 /*
2  * Copyright (c) 2008 The DragonFly Project.  All rights reserved.
3  *
4  * This code is derived from software contributed to The DragonFly Project
5  * by Matthew Dillon <dillon@backplane.com>
6  *
7  * Redistribution and use in source and binary forms, with or without
8  * modification, are permitted provided that the following conditions
9  * are met:
10  *
11  * 1. Redistributions of source code must retain the above copyright
12  *    notice, this list of conditions and the following disclaimer.
13  * 2. Redistributions in binary form must reproduce the above copyright
14  *    notice, this list of conditions and the following disclaimer in
15  *    the documentation and/or other materials provided with the
16  *    distribution.
17  * 3. Neither the name of The DragonFly Project nor the names of its
18  *    contributors may be used to endorse or promote products derived
19  *    from this software without specific, prior written permission.
20  *
21  * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
22  * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
23  * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS
24  * FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL THE
25  * COPYRIGHT HOLDERS OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT,
26  * INCIDENTAL, SPECIAL, EXEMPLARY OR CONSEQUENTIAL DAMAGES (INCLUDING,
27  * BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
28  * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED
29  * AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
30  * OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT
31  * OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
32  * SUCH DAMAGE.
33  */
34 /*
35  * vknet [-cdU] [-b bridgeN] [-p socket_path] [-t tapN] [address/cidrbits]
36  *
37  * Create a named unix-domain socket which userland vkernels can open
38  * to gain access to a local network.  All connections to the socket
39  * are bridged together and the local network can also be bridged onto
40  * a TAP interface by specifying the -t option.
41  */
42 #include "vknetd.h"
43 
44 static ioinfo_t vknet_tap(const char *tapName, const char *bridgeName);
45 static int vknet_listener(const char *pathName);
46 static void vknet_acceptor(int net_fd);
47 static void *vknet_io(void *arg);
48 static int vknet_connect(const char *pathName);
49 static void vknet_monitor(int net_fd);
50 static void usage(void);
51 static void writepid(void);
52 static void cleanup(int);
53 
54 pthread_mutex_t BridgeMutex;
55 
56 int SecureOpt = 1;
57 int DebugOpt = 0;
58 int SetAddrOpt = 0;
59 const char *pidfile = "/var/run/vknetd.pid";
60 
61 struct in_addr NetAddress;
62 struct in_addr NetMask;
63 
64 int
65 main(int ac, char **av)
66 {
67 	const char *pathName = "/var/run/vknet";
68 	const char *tapName = "auto";
69 	const char *bridgeName = NULL;
70 	int net_fd;
71 	int connectOpt = 0;
72 	int c;
73 	ioinfo_t tap_info;
74 	pthread_t dummy_td;
75 
76 	while ((c = getopt(ac, av, "b:cdp:i:t:U")) != -1) {
77 		switch (c) {
78 		case 'U':
79 			SecureOpt = 0;
80 			break;
81 		case 'b':
82 			bridgeName = optarg;
83 			break;
84 		case 'd':
85 			DebugOpt = 1;
86 			break;
87 		case 'p':
88 			pathName = optarg;
89 			break;
90 		case 'i':
91 			pidfile = optarg;
92 			break;
93 		case 't':
94 			tapName = optarg;
95 			break;
96 		case 'c':
97 			connectOpt = 1;
98 			break;
99 		default:
100 			usage();
101 		}
102 	}
103 	av += optind;
104 	ac -= optind;
105 	if (ac)
106 		SetAddrOpt = 1;
107 
108 	/*
109 	 * Ignore SIGPIPE to prevent write() races against disconnecting
110 	 * clients from killing vknetd.  Should be inherited by all I/O
111 	 * threads.
112 	 */
113 	signal(SIGPIPE, SIG_IGN);
114 
115 	/*
116 	 * Special connect/debug mode
117 	 */
118 	if (connectOpt) {
119 		net_fd = vknet_connect(pathName);
120 		if (net_fd < 0) {
121 			perror("connect");
122 			exit(1);
123 		}
124 		vknet_monitor(net_fd);
125 		exit(0);
126 	}
127 
128 	/*
129 	 * In secure mode (the default), a network address/mask must be
130 	 * specified.  e.g. 10.1.0.0/16.  Any traffic going out the TAP
131 	 * interface will be filtered.
132 	 *
133 	 * If non-secure mode the network address/mask is optional.
134 	 */
135 	if (SecureOpt || SetAddrOpt) {
136 		char *str;
137 		int masklen;
138 		u_int32_t mask;
139 
140 		if (ac == 0 || strchr(av[0], '/') == NULL)
141 			usage();
142 		str = strdup(av[0]);
143 		if (inet_pton(AF_INET, strtok(str, "/"), &NetAddress) <= 0)
144 			usage();
145 		masklen = strtoul(strtok(NULL, "/"), NULL, 10);
146 		mask = (1 << (32 - masklen)) - 1;
147 		NetMask.s_addr = htonl(~mask);
148 	}
149 
150 	/*
151 	 * Normal operation, create the tap/bridge and listener.  This
152 	 * part is not threaded.
153 	 */
154 	mac_init();
155 
156 	if ((tap_info = vknet_tap(tapName, bridgeName)) == NULL) {
157 		perror("tap: ");
158 		exit(1);
159 	}
160 	if ((net_fd = vknet_listener(pathName)) < 0) {
161 		perror("listener: ");
162 		exit(1);
163 	}
164 
165 	/*
166 	 * Now make us a demon and start the threads going.
167 	 */
168 	if (DebugOpt == 0)
169 		daemon(1, 0);
170 
171 	writepid();
172 
173 	signal(SIGINT, cleanup);
174 	signal(SIGHUP, cleanup);
175 	signal(SIGTERM, cleanup);
176 
177 	pthread_mutex_init(&BridgeMutex, NULL);
178 	pthread_create(&dummy_td, NULL, vknet_io, tap_info);
179 	vknet_acceptor(net_fd);
180 
181 	exit(0);
182 }
183 
184 #define TAPDEV_MINOR(x) ((int)((x) & 0xffff00ff))
185 
186 static ioinfo_t
187 vknet_tap(const char *tapName, const char *bridgeName)
188 {
189 	struct ifreq ifr;
190 	struct ifaliasreq ifra;
191 	struct stat st;
192 	char *buf = NULL;
193 	int tap_fd;
194 	int tap_unit;
195 	int i;
196 	int s;
197 	int flags;
198 	ioinfo_t info;
199 
200 	if (strcmp(tapName, "auto") == 0) {
201 		for (i = 0; ; ++i) {
202 			asprintf(&buf, "/dev/tap%d", i);
203 			tap_fd = open(buf, O_RDWR | O_NONBLOCK);
204 			free(buf);
205 			if (tap_fd >= 0 || errno == ENOENT)
206 				break;
207 		}
208 	} else if (strncmp(tapName, "tap", 3) == 0) {
209 		asprintf(&buf, "/dev/%s", tapName);
210 		tap_fd = open(buf, O_RDWR | O_NONBLOCK);
211 		free(buf);
212 	} else {
213 		tap_fd = open(tapName, O_RDWR | O_NONBLOCK);
214 	}
215 	if (tap_fd < 0)
216 		return(NULL);
217 
218 	/*
219 	 * Figure out the tap unit number
220 	 */
221 	if (fstat(tap_fd, &st) < 0) {
222 		close(tap_fd);
223 		return(NULL);
224 	}
225 	tap_unit = TAPDEV_MINOR(st.st_rdev);
226 
227 	/*
228 	 * Setup for ioctls
229 	 */
230 	fcntl(tap_fd, F_SETFL, 0);
231 	bzero(&ifr, sizeof(ifr));
232 	bzero(&ifra, sizeof(ifra));
233 	snprintf(ifr.ifr_name, sizeof(ifr.ifr_name), "tap%d", tap_unit);
234 	snprintf(ifra.ifra_name, sizeof(ifra.ifra_name), "tap%d", tap_unit);
235 
236 	s = socket(AF_INET, SOCK_DGRAM, 0);
237 
238 	/*
239 	 * Set the interface address if in Secure mode.
240 	 */
241 	if (SetAddrOpt) {
242 		struct sockaddr_in *in;
243 
244 		in = (void *)&ifra.ifra_addr;
245 		in->sin_family = AF_INET;
246 		in->sin_len = sizeof(ifra.ifra_addr);
247 		in->sin_addr = NetAddress;
248 		in = (void *)&ifra.ifra_mask;
249 		in->sin_family = AF_INET;
250 		in->sin_len = sizeof(ifra.ifra_mask);
251 		in->sin_addr = NetMask;
252 		if (ioctl(s, SIOCAIFADDR, &ifra) < 0) {
253 			perror("Unable to set address on tap interface");
254 			exit(1);
255 		}
256 	}
257 
258 	/*
259 	 * Turn up the interface
260 	 */
261 	flags = IFF_UP;
262 	if (ioctl(s, SIOCGIFFLAGS, &ifr) >= 0) {
263 		bzero(&ifr, sizeof(ifr));
264 		snprintf(ifr.ifr_name, sizeof(ifr.ifr_name), "tap%d", tap_unit);
265 		ifr.ifr_flags |= flags & 0xFFFF;
266 		ifr.ifr_flagshigh |= flags >> 16;
267 		if (ioctl(s, SIOCSIFFLAGS, &ifr) < 0) {
268 			perror("Unable to set IFF_UP on tap interface");
269 			exit(1);
270 		}
271 	}
272 
273 	if (bridgeName) {
274 		struct ifbreq ifbr;
275 		struct ifdrv ifd;
276 
277 		/*
278 		 * Create the bridge if necessary.
279 		 */
280 		bzero(&ifr, sizeof(ifr));
281 		snprintf(ifr.ifr_name, sizeof(ifr.ifr_name), "%s", bridgeName);
282 		if (ioctl(s, SIOCIFCREATE, &ifr) < 0) {
283 			if (errno != EEXIST) {
284 				perror("Unable to create bridge interface");
285 				exit(1);
286 			}
287 		}
288 
289 
290 		/*
291 		 * Add the tap interface to the bridge
292 		 */
293 		bzero(&ifbr, sizeof(ifbr));
294 		snprintf(ifbr.ifbr_ifsname, sizeof(ifbr.ifbr_ifsname),
295 			 "tap%d", tap_unit);
296 
297 		bzero(&ifd, sizeof(ifd));
298 		snprintf(ifd.ifd_name, sizeof(ifd.ifd_name), "%s", bridgeName);
299 		ifd.ifd_cmd = BRDGADD;
300 		ifd.ifd_len = sizeof(ifbr);
301 		ifd.ifd_data = &ifbr;
302 
303 		if (ioctl(s, SIOCSDRVSPEC, &ifd) < 0) {
304 			if (errno != EEXIST) {
305 				perror("Unable to add tap ifc to bridge!");
306 				exit(1);
307 			}
308 		}
309 	}
310 
311 	close(s);
312 	info = malloc(sizeof(*info));
313 	bzero(info, sizeof(*info));
314 	info->fd = tap_fd;
315 	info->istap = 1;
316 	return(info);
317 }
318 
319 #undef TAPDEV_MINOR
320 
321 static int
322 vknet_listener(const char *pathName)
323 {
324 	struct sockaddr_un sunx;
325 	int net_fd;
326 	int len;
327 	gid_t gid;
328 	struct group *grp;
329 
330 	/*
331 	 * Group access to our named unix domain socket.
332 	 */
333 	if ((grp = getgrnam("vknet")) == NULL) {
334 		fprintf(stderr, "The 'vknet' group must exist\n");
335 		exit(1);
336 	}
337 	gid = grp->gr_gid;
338 	endgrent();
339 
340 	/*
341 	 * Socket setup
342 	 */
343 	snprintf(sunx.sun_path, sizeof(sunx.sun_path), "%s", pathName);
344 	len = offsetof(struct sockaddr_un, sun_path[strlen(sunx.sun_path)]);
345 	++len;	/* include nul */
346 	sunx.sun_family = AF_UNIX;
347 	sunx.sun_len = len;
348 
349 	net_fd = socket(AF_UNIX, SOCK_SEQPACKET, 0);
350 	if (net_fd < 0)
351 		return(-1);
352 	remove(pathName);
353 	if (bind(net_fd, (void *)&sunx, len) < 0) {
354 		close(net_fd);
355 		return(-1);
356 	}
357 	if (listen(net_fd, 1024) < 0) {
358 		close(net_fd);
359 		return(-1);
360 	}
361 	if (chown(pathName, (uid_t)-1, gid) < 0) {
362 		close(net_fd);
363 		return(-1);
364 	}
365 	if (chmod(pathName, 0660) < 0) {
366 		close(net_fd);
367 		return(-1);
368 	}
369 	return(net_fd);
370 }
371 
372 static
373 void
374 vknet_acceptor(int net_fd)
375 {
376 	struct sockaddr_un sunx;
377 	pthread_t dummy_td;
378 	int sunx_len;
379 	int rfd;
380 	ioinfo_t info;
381 
382 	for (;;) {
383 		sunx_len = sizeof(sunx);
384 		rfd = accept(net_fd, (void *)&sunx, &sunx_len);
385 		if (rfd < 0)
386 			break;
387 		info = malloc(sizeof(*info));
388 		bzero(info, sizeof(*info));
389 		info->fd = rfd;
390 		info->istap = 0;
391 		pthread_create(&dummy_td, NULL, vknet_io, info);
392 	}
393 }
394 
395 /*
396  * This I/O thread implements the core of the bridging code.
397  */
398 static
399 void *
400 vknet_io(void *arg)
401 {
402 	ioinfo_t info = arg;
403 	bridge_t bridge;
404 	u_int8_t *pkt;
405 	int bytes;
406 
407 	pthread_detach(pthread_self());
408 
409 	/*
410 	 * Assign as a bridge slot using our thread id.
411 	 */
412 	pthread_mutex_lock(&BridgeMutex);
413 	bridge = bridge_add(info);
414 	pthread_mutex_unlock(&BridgeMutex);
415 
416 	/*
417 	 * Read packet loop.  Writing is handled by the bridge code.
418 	 */
419 	pkt = malloc(MAXPKT);
420 	while ((bytes = read(info->fd, pkt, MAXPKT)) > 0) {
421 		pthread_mutex_lock(&BridgeMutex);
422 		bridge_packet(bridge, pkt, bytes);
423 		pthread_mutex_unlock(&BridgeMutex);
424 	}
425 
426 	/*
427 	 * Cleanup
428 	 */
429 	pthread_mutex_lock(&BridgeMutex);
430 	bridge_del(bridge);
431 	pthread_mutex_unlock(&BridgeMutex);
432 
433 	close(info->fd);
434 	free(pkt);
435 	pthread_exit(NULL);
436 }
437 
438 /*
439  * Debugging
440  */
441 static int
442 vknet_connect(const char *pathName)
443 {
444 	struct sockaddr_un sunx;
445 	int len;
446 	int net_fd;
447 
448 	snprintf(sunx.sun_path, sizeof(sunx.sun_path), "%s", pathName);
449 	len = offsetof(struct sockaddr_un, sun_path[strlen(sunx.sun_path)]);
450 	++len;	/* include nul */
451 	sunx.sun_family = AF_UNIX;
452 	sunx.sun_len = len;
453 
454 	net_fd = socket(AF_UNIX, SOCK_SEQPACKET, 0);
455 	if (net_fd < 0)
456 		return(-1);
457 	if (connect(net_fd, (void *)&sunx, len) < 0) {
458 		close(net_fd);
459 		return(-1);
460 	}
461 	return(net_fd);
462 }
463 
464 static void
465 vknet_monitor(int net_fd)
466 {
467 	u_int8_t *pkt;
468 	int bytes;
469 	int i;
470 
471 	pkt = malloc(MAXPKT);
472 	while ((bytes = read(net_fd, pkt, MAXPKT)) > 0) {
473 		printf("%02x:%02x:%02x:%02x:%02x:%02x <- "
474 		       "%02x:%02x:%02x:%02x:%02x:%02x",
475 		       pkt[0], pkt[1], pkt[2], pkt[3], pkt[4], pkt[5],
476 		       pkt[6], pkt[7], pkt[8], pkt[9], pkt[10], pkt[11]);
477 		for (i = 12; i < bytes; ++i) {
478 			if (((i - 12) & 15) == 0) {
479 				printf("\n\t");
480 			}
481 			printf(" %02x", pkt[i]);
482 		}
483 		printf("\n");
484 	}
485 	free(pkt);
486 }
487 
488 /*
489  * Misc
490  */
491 static void
492 writepid(void)
493 {
494 	FILE *pf;
495 
496 	if ((pf = fopen(pidfile, "w+")) == NULL)
497 		errx(1, "Failed to create pidfile %s", pidfile);
498 
499 	if ((fprintf(pf, "%d\n", getpid())) < 1)
500 		err(1, "fprintf");
501 
502 	fclose(pf);
503 }
504 
505 static void
506 cleanup(int __unused sig)
507 {
508 	if (pidfile)
509 		unlink(pidfile);
510 }
511 
512 static
513 void
514 usage(void)
515 {
516 	fprintf(stderr, "usage: vknet [-cdU] [-b bridgeN] [-p socket_path] [-i pidfile] [-t tapN] [address/cidrbits]\n");
517 	fprintf(stderr, "address/cidrbits must be specified in default secure mode.\n");
518 	exit(1);
519 }
520 
521