Commit | Line | Data |
---|---|---|
1da177e4 LT |
1 | /* |
2 | * TUN - Universal TUN/TAP device driver. | |
3 | * Copyright (C) 1999-2002 Maxim Krasnyansky <maxk@qualcomm.com> | |
4 | * | |
5 | * This program is free software; you can redistribute it and/or modify | |
6 | * it under the terms of the GNU General Public License as published by | |
7 | * the Free Software Foundation; either version 2 of the License, or | |
8 | * (at your option) any later version. | |
9 | * | |
10 | * This program is distributed in the hope that it will be useful, | |
11 | * but WITHOUT ANY WARRANTY; without even the implied warranty of | |
12 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the | |
13 | * GNU General Public License for more details. | |
14 | * | |
15 | * $Id: tun.c,v 1.15 2002/03/01 02:44:24 maxk Exp $ | |
16 | */ | |
17 | ||
18 | /* | |
19 | * Changes: | |
20 | * | |
ff4cc3ac MK |
21 | * Mike Kershaw <dragorn@kismetwireless.net> 2005/08/14 |
22 | * Add TUNSETLINK ioctl to set the link encapsulation | |
23 | * | |
1da177e4 | 24 | * Mark Smith <markzzzsmith@yahoo.com.au> |
f271b2cc | 25 | * Use random_ether_addr() for tap MAC address. |
1da177e4 LT |
26 | * |
27 | * Harald Roelle <harald.roelle@ifi.lmu.de> 2004/04/20 | |
28 | * Fixes in packet dropping, queue length setting and queue wakeup. | |
29 | * Increased default tx queue length. | |
30 | * Added ethtool API. | |
31 | * Minor cleanups | |
32 | * | |
33 | * Daniel Podlejski <underley@underley.eu.org> | |
34 | * Modifications for 2.3.99-pre5 kernel. | |
35 | */ | |
36 | ||
37 | #define DRV_NAME "tun" | |
38 | #define DRV_VERSION "1.6" | |
39 | #define DRV_DESCRIPTION "Universal TUN/TAP device driver" | |
40 | #define DRV_COPYRIGHT "(C) 1999-2004 Max Krasnyansky <maxk@qualcomm.com>" | |
41 | ||
1da177e4 LT |
42 | #include <linux/module.h> |
43 | #include <linux/errno.h> | |
44 | #include <linux/kernel.h> | |
45 | #include <linux/major.h> | |
46 | #include <linux/slab.h> | |
fd3e05b6 | 47 | #include <linux/smp_lock.h> |
1da177e4 LT |
48 | #include <linux/poll.h> |
49 | #include <linux/fcntl.h> | |
50 | #include <linux/init.h> | |
51 | #include <linux/skbuff.h> | |
52 | #include <linux/netdevice.h> | |
53 | #include <linux/etherdevice.h> | |
54 | #include <linux/miscdevice.h> | |
55 | #include <linux/ethtool.h> | |
56 | #include <linux/rtnetlink.h> | |
57 | #include <linux/if.h> | |
58 | #include <linux/if_arp.h> | |
59 | #include <linux/if_ether.h> | |
60 | #include <linux/if_tun.h> | |
61 | #include <linux/crc32.h> | |
d647a591 | 62 | #include <linux/nsproxy.h> |
f43798c2 | 63 | #include <linux/virtio_net.h> |
881d966b | 64 | #include <net/net_namespace.h> |
79d17604 | 65 | #include <net/netns/generic.h> |
f019a7a5 | 66 | #include <net/rtnetlink.h> |
33dccbb0 | 67 | #include <net/sock.h> |
1da177e4 LT |
68 | |
69 | #include <asm/system.h> | |
70 | #include <asm/uaccess.h> | |
71 | ||
14daa021 RR |
72 | /* Uncomment to enable debugging */ |
73 | /* #define TUN_DEBUG 1 */ | |
74 | ||
1da177e4 LT |
75 | #ifdef TUN_DEBUG |
76 | static int debug; | |
14daa021 RR |
77 | |
78 | #define DBG if(tun->debug)printk | |
79 | #define DBG1 if(debug==2)printk | |
80 | #else | |
81 | #define DBG( a... ) | |
82 | #define DBG1( a... ) | |
83 | #endif | |
84 | ||
f271b2cc MK |
85 | #define FLT_EXACT_COUNT 8 |
86 | struct tap_filter { | |
87 | unsigned int count; /* Number of addrs. Zero means disabled */ | |
88 | u32 mask[2]; /* Mask of the hashed addrs */ | |
89 | unsigned char addr[FLT_EXACT_COUNT][ETH_ALEN]; | |
90 | }; | |
91 | ||
631ab46b | 92 | struct tun_file { |
c70f1829 | 93 | atomic_t count; |
631ab46b | 94 | struct tun_struct *tun; |
36b50bab | 95 | struct net *net; |
631ab46b EB |
96 | }; |
97 | ||
33dccbb0 HX |
98 | struct tun_sock; |
99 | ||
14daa021 | 100 | struct tun_struct { |
631ab46b | 101 | struct tun_file *tfile; |
f271b2cc | 102 | unsigned int flags; |
14daa021 RR |
103 | uid_t owner; |
104 | gid_t group; | |
105 | ||
14daa021 RR |
106 | struct sk_buff_head readq; |
107 | ||
108 | struct net_device *dev; | |
f271b2cc | 109 | struct fasync_struct *fasync; |
14daa021 | 110 | |
f271b2cc | 111 | struct tap_filter txflt; |
33dccbb0 HX |
112 | struct sock *sk; |
113 | struct socket socket; | |
14daa021 RR |
114 | |
115 | #ifdef TUN_DEBUG | |
116 | int debug; | |
1da177e4 | 117 | #endif |
14daa021 | 118 | }; |
1da177e4 | 119 | |
33dccbb0 HX |
120 | struct tun_sock { |
121 | struct sock sk; | |
122 | struct tun_struct *tun; | |
123 | }; | |
124 | ||
125 | static inline struct tun_sock *tun_sk(struct sock *sk) | |
126 | { | |
127 | return container_of(sk, struct tun_sock, sk); | |
128 | } | |
129 | ||
a7385ba2 EB |
130 | static int tun_attach(struct tun_struct *tun, struct file *file) |
131 | { | |
631ab46b | 132 | struct tun_file *tfile = file->private_data; |
a7385ba2 | 133 | const struct cred *cred = current_cred(); |
38231b7a | 134 | int err; |
a7385ba2 EB |
135 | |
136 | ASSERT_RTNL(); | |
137 | ||
a7385ba2 EB |
138 | /* Check permissions */ |
139 | if (((tun->owner != -1 && cred->euid != tun->owner) || | |
1bded710 | 140 | (tun->group != -1 && !in_egroup_p(tun->group))) && |
a7385ba2 EB |
141 | !capable(CAP_NET_ADMIN)) |
142 | return -EPERM; | |
143 | ||
38231b7a EB |
144 | netif_tx_lock_bh(tun->dev); |
145 | ||
146 | err = -EINVAL; | |
147 | if (tfile->tun) | |
148 | goto out; | |
149 | ||
150 | err = -EBUSY; | |
151 | if (tun->tfile) | |
152 | goto out; | |
153 | ||
154 | err = 0; | |
631ab46b EB |
155 | tfile->tun = tun; |
156 | tun->tfile = tfile; | |
c70f1829 | 157 | dev_hold(tun->dev); |
9c3fea6a | 158 | sock_hold(tun->sk); |
c70f1829 | 159 | atomic_inc(&tfile->count); |
a7385ba2 | 160 | |
38231b7a EB |
161 | out: |
162 | netif_tx_unlock_bh(tun->dev); | |
163 | return err; | |
a7385ba2 EB |
164 | } |
165 | ||
631ab46b EB |
166 | static void __tun_detach(struct tun_struct *tun) |
167 | { | |
631ab46b | 168 | /* Detach from net device */ |
38231b7a | 169 | netif_tx_lock_bh(tun->dev); |
631ab46b | 170 | tun->tfile = NULL; |
38231b7a | 171 | netif_tx_unlock_bh(tun->dev); |
631ab46b EB |
172 | |
173 | /* Drop read queue */ | |
174 | skb_queue_purge(&tun->readq); | |
c70f1829 EB |
175 | |
176 | /* Drop the extra count on the net device */ | |
177 | dev_put(tun->dev); | |
178 | } | |
179 | ||
180 | static void tun_detach(struct tun_struct *tun) | |
181 | { | |
182 | rtnl_lock(); | |
183 | __tun_detach(tun); | |
184 | rtnl_unlock(); | |
631ab46b EB |
185 | } |
186 | ||
187 | static struct tun_struct *__tun_get(struct tun_file *tfile) | |
188 | { | |
c70f1829 EB |
189 | struct tun_struct *tun = NULL; |
190 | ||
191 | if (atomic_inc_not_zero(&tfile->count)) | |
192 | tun = tfile->tun; | |
193 | ||
194 | return tun; | |
631ab46b EB |
195 | } |
196 | ||
197 | static struct tun_struct *tun_get(struct file *file) | |
198 | { | |
199 | return __tun_get(file->private_data); | |
200 | } | |
201 | ||
202 | static void tun_put(struct tun_struct *tun) | |
203 | { | |
c70f1829 EB |
204 | struct tun_file *tfile = tun->tfile; |
205 | ||
206 | if (atomic_dec_and_test(&tfile->count)) | |
207 | tun_detach(tfile->tun); | |
631ab46b EB |
208 | } |
209 | ||
f271b2cc MK |
210 | /* TAP filterting */ |
211 | static void addr_hash_set(u32 *mask, const u8 *addr) | |
212 | { | |
213 | int n = ether_crc(ETH_ALEN, addr) >> 26; | |
214 | mask[n >> 5] |= (1 << (n & 31)); | |
215 | } | |
216 | ||
217 | static unsigned int addr_hash_test(const u32 *mask, const u8 *addr) | |
218 | { | |
219 | int n = ether_crc(ETH_ALEN, addr) >> 26; | |
220 | return mask[n >> 5] & (1 << (n & 31)); | |
221 | } | |
222 | ||
223 | static int update_filter(struct tap_filter *filter, void __user *arg) | |
224 | { | |
225 | struct { u8 u[ETH_ALEN]; } *addr; | |
226 | struct tun_filter uf; | |
227 | int err, alen, n, nexact; | |
228 | ||
229 | if (copy_from_user(&uf, arg, sizeof(uf))) | |
230 | return -EFAULT; | |
231 | ||
232 | if (!uf.count) { | |
233 | /* Disabled */ | |
234 | filter->count = 0; | |
235 | return 0; | |
236 | } | |
237 | ||
238 | alen = ETH_ALEN * uf.count; | |
239 | addr = kmalloc(alen, GFP_KERNEL); | |
240 | if (!addr) | |
241 | return -ENOMEM; | |
242 | ||
243 | if (copy_from_user(addr, arg + sizeof(uf), alen)) { | |
244 | err = -EFAULT; | |
245 | goto done; | |
246 | } | |
247 | ||
248 | /* The filter is updated without holding any locks. Which is | |
249 | * perfectly safe. We disable it first and in the worst | |
250 | * case we'll accept a few undesired packets. */ | |
251 | filter->count = 0; | |
252 | wmb(); | |
253 | ||
254 | /* Use first set of addresses as an exact filter */ | |
255 | for (n = 0; n < uf.count && n < FLT_EXACT_COUNT; n++) | |
256 | memcpy(filter->addr[n], addr[n].u, ETH_ALEN); | |
257 | ||
258 | nexact = n; | |
259 | ||
cfbf84fc AW |
260 | /* Remaining multicast addresses are hashed, |
261 | * unicast will leave the filter disabled. */ | |
f271b2cc | 262 | memset(filter->mask, 0, sizeof(filter->mask)); |
cfbf84fc AW |
263 | for (; n < uf.count; n++) { |
264 | if (!is_multicast_ether_addr(addr[n].u)) { | |
265 | err = 0; /* no filter */ | |
266 | goto done; | |
267 | } | |
f271b2cc | 268 | addr_hash_set(filter->mask, addr[n].u); |
cfbf84fc | 269 | } |
f271b2cc MK |
270 | |
271 | /* For ALLMULTI just set the mask to all ones. | |
272 | * This overrides the mask populated above. */ | |
273 | if ((uf.flags & TUN_FLT_ALLMULTI)) | |
274 | memset(filter->mask, ~0, sizeof(filter->mask)); | |
275 | ||
276 | /* Now enable the filter */ | |
277 | wmb(); | |
278 | filter->count = nexact; | |
279 | ||
280 | /* Return the number of exact filters */ | |
281 | err = nexact; | |
282 | ||
283 | done: | |
284 | kfree(addr); | |
285 | return err; | |
286 | } | |
287 | ||
288 | /* Returns: 0 - drop, !=0 - accept */ | |
289 | static int run_filter(struct tap_filter *filter, const struct sk_buff *skb) | |
290 | { | |
291 | /* Cannot use eth_hdr(skb) here because skb_mac_hdr() is incorrect | |
292 | * at this point. */ | |
293 | struct ethhdr *eh = (struct ethhdr *) skb->data; | |
294 | int i; | |
295 | ||
296 | /* Exact match */ | |
297 | for (i = 0; i < filter->count; i++) | |
298 | if (!compare_ether_addr(eh->h_dest, filter->addr[i])) | |
299 | return 1; | |
300 | ||
301 | /* Inexact match (multicast only) */ | |
302 | if (is_multicast_ether_addr(eh->h_dest)) | |
303 | return addr_hash_test(filter->mask, eh->h_dest); | |
304 | ||
305 | return 0; | |
306 | } | |
307 | ||
308 | /* | |
309 | * Checks whether the packet is accepted or not. | |
310 | * Returns: 0 - drop, !=0 - accept | |
311 | */ | |
312 | static int check_filter(struct tap_filter *filter, const struct sk_buff *skb) | |
313 | { | |
314 | if (!filter->count) | |
315 | return 1; | |
316 | ||
317 | return run_filter(filter, skb); | |
318 | } | |
319 | ||
1da177e4 LT |
320 | /* Network device part of the driver */ |
321 | ||
7282d491 | 322 | static const struct ethtool_ops tun_ethtool_ops; |
1da177e4 | 323 | |
c70f1829 EB |
324 | /* Net device detach from fd. */ |
325 | static void tun_net_uninit(struct net_device *dev) | |
326 | { | |
327 | struct tun_struct *tun = netdev_priv(dev); | |
328 | struct tun_file *tfile = tun->tfile; | |
329 | ||
330 | /* Inform the methods they need to stop using the dev. | |
331 | */ | |
332 | if (tfile) { | |
c40af84a | 333 | wake_up_all(&tun->socket.wait); |
c70f1829 EB |
334 | if (atomic_dec_and_test(&tfile->count)) |
335 | __tun_detach(tun); | |
336 | } | |
337 | } | |
338 | ||
9c3fea6a HX |
339 | static void tun_free_netdev(struct net_device *dev) |
340 | { | |
341 | struct tun_struct *tun = netdev_priv(dev); | |
342 | ||
343 | sock_put(tun->sk); | |
344 | } | |
345 | ||
1da177e4 LT |
346 | /* Net device open. */ |
347 | static int tun_net_open(struct net_device *dev) | |
348 | { | |
349 | netif_start_queue(dev); | |
350 | return 0; | |
351 | } | |
352 | ||
353 | /* Net device close. */ | |
354 | static int tun_net_close(struct net_device *dev) | |
355 | { | |
356 | netif_stop_queue(dev); | |
357 | return 0; | |
358 | } | |
359 | ||
360 | /* Net device start xmit */ | |
361 | static int tun_net_xmit(struct sk_buff *skb, struct net_device *dev) | |
362 | { | |
363 | struct tun_struct *tun = netdev_priv(dev); | |
364 | ||
365 | DBG(KERN_INFO "%s: tun_net_xmit %d\n", tun->dev->name, skb->len); | |
366 | ||
367 | /* Drop packet if interface is not attached */ | |
631ab46b | 368 | if (!tun->tfile) |
1da177e4 LT |
369 | goto drop; |
370 | ||
f271b2cc MK |
371 | /* Drop if the filter does not like it. |
372 | * This is a noop if the filter is disabled. | |
373 | * Filter can be enabled only for the TAP devices. */ | |
374 | if (!check_filter(&tun->txflt, skb)) | |
375 | goto drop; | |
376 | ||
1da177e4 LT |
377 | if (skb_queue_len(&tun->readq) >= dev->tx_queue_len) { |
378 | if (!(tun->flags & TUN_ONE_QUEUE)) { | |
379 | /* Normal queueing mode. */ | |
380 | /* Packet scheduler handles dropping of further packets. */ | |
381 | netif_stop_queue(dev); | |
382 | ||
383 | /* We won't see all dropped packets individually, so overrun | |
384 | * error is more appropriate. */ | |
09f75cd7 | 385 | dev->stats.tx_fifo_errors++; |
1da177e4 LT |
386 | } else { |
387 | /* Single queue mode. | |
388 | * Driver handles dropping of all packets itself. */ | |
389 | goto drop; | |
390 | } | |
391 | } | |
392 | ||
f271b2cc | 393 | /* Enqueue packet */ |
1da177e4 LT |
394 | skb_queue_tail(&tun->readq, skb); |
395 | dev->trans_start = jiffies; | |
396 | ||
397 | /* Notify and wake up reader process */ | |
398 | if (tun->flags & TUN_FASYNC) | |
399 | kill_fasync(&tun->fasync, SIGIO, POLL_IN); | |
c40af84a | 400 | wake_up_interruptible(&tun->socket.wait); |
1da177e4 LT |
401 | return 0; |
402 | ||
403 | drop: | |
09f75cd7 | 404 | dev->stats.tx_dropped++; |
1da177e4 LT |
405 | kfree_skb(skb); |
406 | return 0; | |
407 | } | |
408 | ||
f271b2cc | 409 | static void tun_net_mclist(struct net_device *dev) |
1da177e4 | 410 | { |
f271b2cc MK |
411 | /* |
412 | * This callback is supposed to deal with mc filter in | |
413 | * _rx_ path and has nothing to do with the _tx_ path. | |
414 | * In rx path we always accept everything userspace gives us. | |
415 | */ | |
416 | return; | |
1da177e4 LT |
417 | } |
418 | ||
4885a504 ES |
419 | #define MIN_MTU 68 |
420 | #define MAX_MTU 65535 | |
421 | ||
422 | static int | |
423 | tun_net_change_mtu(struct net_device *dev, int new_mtu) | |
424 | { | |
425 | if (new_mtu < MIN_MTU || new_mtu + dev->hard_header_len > MAX_MTU) | |
426 | return -EINVAL; | |
427 | dev->mtu = new_mtu; | |
428 | return 0; | |
429 | } | |
430 | ||
758e43b7 | 431 | static const struct net_device_ops tun_netdev_ops = { |
c70f1829 | 432 | .ndo_uninit = tun_net_uninit, |
758e43b7 SH |
433 | .ndo_open = tun_net_open, |
434 | .ndo_stop = tun_net_close, | |
00829823 | 435 | .ndo_start_xmit = tun_net_xmit, |
758e43b7 | 436 | .ndo_change_mtu = tun_net_change_mtu, |
758e43b7 SH |
437 | }; |
438 | ||
439 | static const struct net_device_ops tap_netdev_ops = { | |
c70f1829 | 440 | .ndo_uninit = tun_net_uninit, |
758e43b7 SH |
441 | .ndo_open = tun_net_open, |
442 | .ndo_stop = tun_net_close, | |
00829823 | 443 | .ndo_start_xmit = tun_net_xmit, |
758e43b7 SH |
444 | .ndo_change_mtu = tun_net_change_mtu, |
445 | .ndo_set_multicast_list = tun_net_mclist, | |
446 | .ndo_set_mac_address = eth_mac_addr, | |
447 | .ndo_validate_addr = eth_validate_addr, | |
448 | }; | |
449 | ||
1da177e4 LT |
450 | /* Initialize net device. */ |
451 | static void tun_net_init(struct net_device *dev) | |
452 | { | |
453 | struct tun_struct *tun = netdev_priv(dev); | |
6aa20a22 | 454 | |
1da177e4 LT |
455 | switch (tun->flags & TUN_TYPE_MASK) { |
456 | case TUN_TUN_DEV: | |
758e43b7 SH |
457 | dev->netdev_ops = &tun_netdev_ops; |
458 | ||
1da177e4 LT |
459 | /* Point-to-Point TUN Device */ |
460 | dev->hard_header_len = 0; | |
461 | dev->addr_len = 0; | |
462 | dev->mtu = 1500; | |
463 | ||
464 | /* Zero header length */ | |
6aa20a22 | 465 | dev->type = ARPHRD_NONE; |
1da177e4 LT |
466 | dev->flags = IFF_POINTOPOINT | IFF_NOARP | IFF_MULTICAST; |
467 | dev->tx_queue_len = TUN_READQ_SIZE; /* We prefer our own queue length */ | |
468 | break; | |
469 | ||
470 | case TUN_TAP_DEV: | |
7a0a9608 | 471 | dev->netdev_ops = &tap_netdev_ops; |
1da177e4 | 472 | /* Ethernet TAP Device */ |
1da177e4 | 473 | ether_setup(dev); |
36226a8d | 474 | |
f271b2cc | 475 | random_ether_addr(dev->dev_addr); |
36226a8d | 476 | |
1da177e4 LT |
477 | dev->tx_queue_len = TUN_READQ_SIZE; /* We prefer our own queue length */ |
478 | break; | |
479 | } | |
480 | } | |
481 | ||
482 | /* Character device part */ | |
483 | ||
484 | /* Poll */ | |
485 | static unsigned int tun_chr_poll(struct file *file, poll_table * wait) | |
6aa20a22 | 486 | { |
b2430de3 EB |
487 | struct tun_file *tfile = file->private_data; |
488 | struct tun_struct *tun = __tun_get(tfile); | |
33dccbb0 HX |
489 | struct sock *sk = tun->sk; |
490 | unsigned int mask = 0; | |
1da177e4 LT |
491 | |
492 | if (!tun) | |
eac9e902 | 493 | return POLLERR; |
1da177e4 LT |
494 | |
495 | DBG(KERN_INFO "%s: tun_chr_poll\n", tun->dev->name); | |
496 | ||
c40af84a | 497 | poll_wait(file, &tun->socket.wait, wait); |
6aa20a22 | 498 | |
b03efcfb | 499 | if (!skb_queue_empty(&tun->readq)) |
1da177e4 LT |
500 | mask |= POLLIN | POLLRDNORM; |
501 | ||
33dccbb0 HX |
502 | if (sock_writeable(sk) || |
503 | (!test_and_set_bit(SOCK_ASYNC_NOSPACE, &sk->sk_socket->flags) && | |
504 | sock_writeable(sk))) | |
505 | mask |= POLLOUT | POLLWRNORM; | |
506 | ||
c70f1829 EB |
507 | if (tun->dev->reg_state != NETREG_REGISTERED) |
508 | mask = POLLERR; | |
509 | ||
631ab46b | 510 | tun_put(tun); |
1da177e4 LT |
511 | return mask; |
512 | } | |
513 | ||
f42157cb RR |
514 | /* prepad is the amount to reserve at front. len is length after that. |
515 | * linear is a hint as to how much to copy (usually headers). */ | |
33dccbb0 HX |
516 | static inline struct sk_buff *tun_alloc_skb(struct tun_struct *tun, |
517 | size_t prepad, size_t len, | |
518 | size_t linear, int noblock) | |
f42157cb | 519 | { |
33dccbb0 | 520 | struct sock *sk = tun->sk; |
f42157cb | 521 | struct sk_buff *skb; |
33dccbb0 | 522 | int err; |
f42157cb RR |
523 | |
524 | /* Under a page? Don't bother with paged skb. */ | |
0eca93bc | 525 | if (prepad + len < PAGE_SIZE || !linear) |
33dccbb0 | 526 | linear = len; |
f42157cb | 527 | |
33dccbb0 HX |
528 | skb = sock_alloc_send_pskb(sk, prepad + linear, len - linear, noblock, |
529 | &err); | |
f42157cb | 530 | if (!skb) |
33dccbb0 | 531 | return ERR_PTR(err); |
f42157cb RR |
532 | |
533 | skb_reserve(skb, prepad); | |
534 | skb_put(skb, linear); | |
33dccbb0 HX |
535 | skb->data_len = len - linear; |
536 | skb->len += len - linear; | |
f42157cb RR |
537 | |
538 | return skb; | |
539 | } | |
540 | ||
1da177e4 | 541 | /* Get packet from user space buffer */ |
33dccbb0 | 542 | static __inline__ ssize_t tun_get_user(struct tun_struct *tun, |
6f26c9a7 | 543 | const struct iovec *iv, size_t count, |
33dccbb0 | 544 | int noblock) |
1da177e4 | 545 | { |
09640e63 | 546 | struct tun_pi pi = { 0, cpu_to_be16(ETH_P_IP) }; |
1da177e4 LT |
547 | struct sk_buff *skb; |
548 | size_t len = count, align = 0; | |
f43798c2 | 549 | struct virtio_net_hdr gso = { 0 }; |
6f26c9a7 | 550 | int offset = 0; |
1da177e4 LT |
551 | |
552 | if (!(tun->flags & TUN_NO_PI)) { | |
553 | if ((len -= sizeof(pi)) > count) | |
554 | return -EINVAL; | |
555 | ||
6f26c9a7 | 556 | if (memcpy_fromiovecend((void *)&pi, iv, 0, sizeof(pi))) |
1da177e4 | 557 | return -EFAULT; |
6f26c9a7 | 558 | offset += sizeof(pi); |
1da177e4 LT |
559 | } |
560 | ||
f43798c2 RR |
561 | if (tun->flags & TUN_VNET_HDR) { |
562 | if ((len -= sizeof(gso)) > count) | |
563 | return -EINVAL; | |
564 | ||
6f26c9a7 | 565 | if (memcpy_fromiovecend((void *)&gso, iv, offset, sizeof(gso))) |
f43798c2 RR |
566 | return -EFAULT; |
567 | ||
4909122f HX |
568 | if ((gso.flags & VIRTIO_NET_HDR_F_NEEDS_CSUM) && |
569 | gso.csum_start + gso.csum_offset + 2 > gso.hdr_len) | |
570 | gso.hdr_len = gso.csum_start + gso.csum_offset + 2; | |
571 | ||
f43798c2 RR |
572 | if (gso.hdr_len > len) |
573 | return -EINVAL; | |
6f26c9a7 | 574 | offset += sizeof(pi); |
f43798c2 RR |
575 | } |
576 | ||
e01bf1c8 | 577 | if ((tun->flags & TUN_TYPE_MASK) == TUN_TAP_DEV) { |
1da177e4 | 578 | align = NET_IP_ALIGN; |
0eca93bc HX |
579 | if (unlikely(len < ETH_HLEN || |
580 | (gso.hdr_len && gso.hdr_len < ETH_HLEN))) | |
e01bf1c8 RR |
581 | return -EINVAL; |
582 | } | |
6aa20a22 | 583 | |
33dccbb0 HX |
584 | skb = tun_alloc_skb(tun, align, len, gso.hdr_len, noblock); |
585 | if (IS_ERR(skb)) { | |
586 | if (PTR_ERR(skb) != -EAGAIN) | |
587 | tun->dev->stats.rx_dropped++; | |
588 | return PTR_ERR(skb); | |
1da177e4 LT |
589 | } |
590 | ||
6f26c9a7 | 591 | if (skb_copy_datagram_from_iovec(skb, 0, iv, offset, len)) { |
09f75cd7 | 592 | tun->dev->stats.rx_dropped++; |
8f22757e | 593 | kfree_skb(skb); |
1da177e4 | 594 | return -EFAULT; |
8f22757e | 595 | } |
1da177e4 | 596 | |
f43798c2 RR |
597 | if (gso.flags & VIRTIO_NET_HDR_F_NEEDS_CSUM) { |
598 | if (!skb_partial_csum_set(skb, gso.csum_start, | |
599 | gso.csum_offset)) { | |
600 | tun->dev->stats.rx_frame_errors++; | |
601 | kfree_skb(skb); | |
602 | return -EINVAL; | |
603 | } | |
604 | } else if (tun->flags & TUN_NOCHECKSUM) | |
605 | skb->ip_summed = CHECKSUM_UNNECESSARY; | |
606 | ||
1da177e4 LT |
607 | switch (tun->flags & TUN_TYPE_MASK) { |
608 | case TUN_TUN_DEV: | |
f09f7ee2 AWC |
609 | if (tun->flags & TUN_NO_PI) { |
610 | switch (skb->data[0] & 0xf0) { | |
611 | case 0x40: | |
612 | pi.proto = htons(ETH_P_IP); | |
613 | break; | |
614 | case 0x60: | |
615 | pi.proto = htons(ETH_P_IPV6); | |
616 | break; | |
617 | default: | |
618 | tun->dev->stats.rx_dropped++; | |
619 | kfree_skb(skb); | |
620 | return -EINVAL; | |
621 | } | |
622 | } | |
623 | ||
459a98ed | 624 | skb_reset_mac_header(skb); |
1da177e4 | 625 | skb->protocol = pi.proto; |
4c13eb66 | 626 | skb->dev = tun->dev; |
1da177e4 LT |
627 | break; |
628 | case TUN_TAP_DEV: | |
629 | skb->protocol = eth_type_trans(skb, tun->dev); | |
630 | break; | |
631 | }; | |
632 | ||
f43798c2 RR |
633 | if (gso.gso_type != VIRTIO_NET_HDR_GSO_NONE) { |
634 | pr_debug("GSO!\n"); | |
635 | switch (gso.gso_type & ~VIRTIO_NET_HDR_GSO_ECN) { | |
636 | case VIRTIO_NET_HDR_GSO_TCPV4: | |
637 | skb_shinfo(skb)->gso_type = SKB_GSO_TCPV4; | |
638 | break; | |
639 | case VIRTIO_NET_HDR_GSO_TCPV6: | |
640 | skb_shinfo(skb)->gso_type = SKB_GSO_TCPV6; | |
641 | break; | |
642 | default: | |
643 | tun->dev->stats.rx_frame_errors++; | |
644 | kfree_skb(skb); | |
645 | return -EINVAL; | |
646 | } | |
647 | ||
648 | if (gso.gso_type & VIRTIO_NET_HDR_GSO_ECN) | |
649 | skb_shinfo(skb)->gso_type |= SKB_GSO_TCP_ECN; | |
650 | ||
651 | skb_shinfo(skb)->gso_size = gso.gso_size; | |
652 | if (skb_shinfo(skb)->gso_size == 0) { | |
653 | tun->dev->stats.rx_frame_errors++; | |
654 | kfree_skb(skb); | |
655 | return -EINVAL; | |
656 | } | |
657 | ||
658 | /* Header must be checked, and gso_segs computed. */ | |
659 | skb_shinfo(skb)->gso_type |= SKB_GSO_DODGY; | |
660 | skb_shinfo(skb)->gso_segs = 0; | |
661 | } | |
6aa20a22 | 662 | |
1da177e4 | 663 | netif_rx_ni(skb); |
6aa20a22 | 664 | |
09f75cd7 JG |
665 | tun->dev->stats.rx_packets++; |
666 | tun->dev->stats.rx_bytes += len; | |
1da177e4 LT |
667 | |
668 | return count; | |
6aa20a22 | 669 | } |
1da177e4 | 670 | |
ee0b3e67 BP |
671 | static ssize_t tun_chr_aio_write(struct kiocb *iocb, const struct iovec *iv, |
672 | unsigned long count, loff_t pos) | |
1da177e4 | 673 | { |
33dccbb0 | 674 | struct file *file = iocb->ki_filp; |
ab46d779 | 675 | struct tun_struct *tun = tun_get(file); |
631ab46b | 676 | ssize_t result; |
1da177e4 LT |
677 | |
678 | if (!tun) | |
679 | return -EBADFD; | |
680 | ||
681 | DBG(KERN_INFO "%s: tun_chr_write %ld\n", tun->dev->name, count); | |
682 | ||
6f26c9a7 | 683 | result = tun_get_user(tun, iv, iov_length(iv, count), |
33dccbb0 | 684 | file->f_flags & O_NONBLOCK); |
631ab46b EB |
685 | |
686 | tun_put(tun); | |
687 | return result; | |
1da177e4 LT |
688 | } |
689 | ||
1da177e4 LT |
690 | /* Put packet to the user space buffer */ |
691 | static __inline__ ssize_t tun_put_user(struct tun_struct *tun, | |
692 | struct sk_buff *skb, | |
43b39dcd | 693 | const struct iovec *iv, int len) |
1da177e4 LT |
694 | { |
695 | struct tun_pi pi = { 0, skb->protocol }; | |
696 | ssize_t total = 0; | |
697 | ||
698 | if (!(tun->flags & TUN_NO_PI)) { | |
699 | if ((len -= sizeof(pi)) < 0) | |
700 | return -EINVAL; | |
701 | ||
702 | if (len < skb->len) { | |
703 | /* Packet will be striped */ | |
704 | pi.flags |= TUN_PKT_STRIP; | |
705 | } | |
6aa20a22 | 706 | |
43b39dcd | 707 | if (memcpy_toiovecend(iv, (void *) &pi, 0, sizeof(pi))) |
1da177e4 LT |
708 | return -EFAULT; |
709 | total += sizeof(pi); | |
6aa20a22 | 710 | } |
1da177e4 | 711 | |
f43798c2 RR |
712 | if (tun->flags & TUN_VNET_HDR) { |
713 | struct virtio_net_hdr gso = { 0 }; /* no info leak */ | |
714 | if ((len -= sizeof(gso)) < 0) | |
715 | return -EINVAL; | |
716 | ||
717 | if (skb_is_gso(skb)) { | |
718 | struct skb_shared_info *sinfo = skb_shinfo(skb); | |
719 | ||
720 | /* This is a hint as to how much should be linear. */ | |
721 | gso.hdr_len = skb_headlen(skb); | |
722 | gso.gso_size = sinfo->gso_size; | |
723 | if (sinfo->gso_type & SKB_GSO_TCPV4) | |
724 | gso.gso_type = VIRTIO_NET_HDR_GSO_TCPV4; | |
725 | else if (sinfo->gso_type & SKB_GSO_TCPV6) | |
726 | gso.gso_type = VIRTIO_NET_HDR_GSO_TCPV6; | |
727 | else | |
728 | BUG(); | |
729 | if (sinfo->gso_type & SKB_GSO_TCP_ECN) | |
730 | gso.gso_type |= VIRTIO_NET_HDR_GSO_ECN; | |
731 | } else | |
732 | gso.gso_type = VIRTIO_NET_HDR_GSO_NONE; | |
733 | ||
734 | if (skb->ip_summed == CHECKSUM_PARTIAL) { | |
735 | gso.flags = VIRTIO_NET_HDR_F_NEEDS_CSUM; | |
736 | gso.csum_start = skb->csum_start - skb_headroom(skb); | |
737 | gso.csum_offset = skb->csum_offset; | |
738 | } /* else everything is zero */ | |
739 | ||
43b39dcd MT |
740 | if (unlikely(memcpy_toiovecend(iv, (void *)&gso, total, |
741 | sizeof(gso)))) | |
f43798c2 RR |
742 | return -EFAULT; |
743 | total += sizeof(gso); | |
744 | } | |
745 | ||
1da177e4 LT |
746 | len = min_t(int, skb->len, len); |
747 | ||
43b39dcd | 748 | skb_copy_datagram_const_iovec(skb, 0, iv, total, len); |
1da177e4 LT |
749 | total += len; |
750 | ||
09f75cd7 JG |
751 | tun->dev->stats.tx_packets++; |
752 | tun->dev->stats.tx_bytes += len; | |
1da177e4 LT |
753 | |
754 | return total; | |
755 | } | |
756 | ||
ee0b3e67 BP |
757 | static ssize_t tun_chr_aio_read(struct kiocb *iocb, const struct iovec *iv, |
758 | unsigned long count, loff_t pos) | |
1da177e4 | 759 | { |
ee0b3e67 | 760 | struct file *file = iocb->ki_filp; |
b2430de3 EB |
761 | struct tun_file *tfile = file->private_data; |
762 | struct tun_struct *tun = __tun_get(tfile); | |
1da177e4 LT |
763 | DECLARE_WAITQUEUE(wait, current); |
764 | struct sk_buff *skb; | |
765 | ssize_t len, ret = 0; | |
766 | ||
767 | if (!tun) | |
768 | return -EBADFD; | |
769 | ||
770 | DBG(KERN_INFO "%s: tun_chr_read\n", tun->dev->name); | |
771 | ||
52427c9d | 772 | len = iov_length(iv, count); |
631ab46b EB |
773 | if (len < 0) { |
774 | ret = -EINVAL; | |
775 | goto out; | |
776 | } | |
1da177e4 | 777 | |
c40af84a | 778 | add_wait_queue(&tun->socket.wait, &wait); |
1da177e4 | 779 | while (len) { |
1da177e4 LT |
780 | current->state = TASK_INTERRUPTIBLE; |
781 | ||
782 | /* Read frames from the queue */ | |
783 | if (!(skb=skb_dequeue(&tun->readq))) { | |
784 | if (file->f_flags & O_NONBLOCK) { | |
785 | ret = -EAGAIN; | |
786 | break; | |
787 | } | |
788 | if (signal_pending(current)) { | |
789 | ret = -ERESTARTSYS; | |
790 | break; | |
791 | } | |
c70f1829 EB |
792 | if (tun->dev->reg_state != NETREG_REGISTERED) { |
793 | ret = -EIO; | |
794 | break; | |
795 | } | |
1da177e4 LT |
796 | |
797 | /* Nothing to read, let's sleep */ | |
798 | schedule(); | |
799 | continue; | |
800 | } | |
801 | netif_wake_queue(tun->dev); | |
802 | ||
43b39dcd | 803 | ret = tun_put_user(tun, skb, iv, len); |
f271b2cc MK |
804 | kfree_skb(skb); |
805 | break; | |
1da177e4 LT |
806 | } |
807 | ||
808 | current->state = TASK_RUNNING; | |
c40af84a | 809 | remove_wait_queue(&tun->socket.wait, &wait); |
1da177e4 | 810 | |
631ab46b EB |
811 | out: |
812 | tun_put(tun); | |
1da177e4 LT |
813 | return ret; |
814 | } | |
815 | ||
1da177e4 LT |
816 | static void tun_setup(struct net_device *dev) |
817 | { | |
818 | struct tun_struct *tun = netdev_priv(dev); | |
819 | ||
820 | skb_queue_head_init(&tun->readq); | |
1da177e4 LT |
821 | |
822 | tun->owner = -1; | |
8c644623 | 823 | tun->group = -1; |
1da177e4 | 824 | |
1da177e4 | 825 | dev->ethtool_ops = &tun_ethtool_ops; |
9c3fea6a | 826 | dev->destructor = tun_free_netdev; |
1da177e4 LT |
827 | } |
828 | ||
f019a7a5 EB |
829 | /* Trivial set of netlink ops to allow deleting tun or tap |
830 | * device with netlink. | |
831 | */ | |
832 | static int tun_validate(struct nlattr *tb[], struct nlattr *data[]) | |
833 | { | |
834 | return -EINVAL; | |
835 | } | |
836 | ||
837 | static struct rtnl_link_ops tun_link_ops __read_mostly = { | |
838 | .kind = DRV_NAME, | |
839 | .priv_size = sizeof(struct tun_struct), | |
840 | .setup = tun_setup, | |
841 | .validate = tun_validate, | |
842 | }; | |
843 | ||
33dccbb0 HX |
844 | static void tun_sock_write_space(struct sock *sk) |
845 | { | |
846 | struct tun_struct *tun; | |
847 | ||
848 | if (!sock_writeable(sk)) | |
849 | return; | |
850 | ||
33dccbb0 HX |
851 | if (!test_and_clear_bit(SOCK_ASYNC_NOSPACE, &sk->sk_socket->flags)) |
852 | return; | |
853 | ||
c722c625 HX |
854 | if (sk->sk_sleep && waitqueue_active(sk->sk_sleep)) |
855 | wake_up_interruptible_sync(sk->sk_sleep); | |
856 | ||
33dccbb0 HX |
857 | tun = container_of(sk, struct tun_sock, sk)->tun; |
858 | kill_fasync(&tun->fasync, SIGIO, POLL_OUT); | |
859 | } | |
860 | ||
861 | static void tun_sock_destruct(struct sock *sk) | |
862 | { | |
9c3fea6a | 863 | free_netdev(container_of(sk, struct tun_sock, sk)->tun->dev); |
33dccbb0 HX |
864 | } |
865 | ||
866 | static struct proto tun_proto = { | |
867 | .name = "tun", | |
868 | .owner = THIS_MODULE, | |
869 | .obj_size = sizeof(struct tun_sock), | |
870 | }; | |
f019a7a5 | 871 | |
980c9e8c DW |
872 | static int tun_flags(struct tun_struct *tun) |
873 | { | |
874 | int flags = 0; | |
875 | ||
876 | if (tun->flags & TUN_TUN_DEV) | |
877 | flags |= IFF_TUN; | |
878 | else | |
879 | flags |= IFF_TAP; | |
880 | ||
881 | if (tun->flags & TUN_NO_PI) | |
882 | flags |= IFF_NO_PI; | |
883 | ||
884 | if (tun->flags & TUN_ONE_QUEUE) | |
885 | flags |= IFF_ONE_QUEUE; | |
886 | ||
887 | if (tun->flags & TUN_VNET_HDR) | |
888 | flags |= IFF_VNET_HDR; | |
889 | ||
890 | return flags; | |
891 | } | |
892 | ||
893 | static ssize_t tun_show_flags(struct device *dev, struct device_attribute *attr, | |
894 | char *buf) | |
895 | { | |
896 | struct tun_struct *tun = netdev_priv(to_net_dev(dev)); | |
897 | return sprintf(buf, "0x%x\n", tun_flags(tun)); | |
898 | } | |
899 | ||
900 | static ssize_t tun_show_owner(struct device *dev, struct device_attribute *attr, | |
901 | char *buf) | |
902 | { | |
903 | struct tun_struct *tun = netdev_priv(to_net_dev(dev)); | |
904 | return sprintf(buf, "%d\n", tun->owner); | |
905 | } | |
906 | ||
907 | static ssize_t tun_show_group(struct device *dev, struct device_attribute *attr, | |
908 | char *buf) | |
909 | { | |
910 | struct tun_struct *tun = netdev_priv(to_net_dev(dev)); | |
911 | return sprintf(buf, "%d\n", tun->group); | |
912 | } | |
913 | ||
914 | static DEVICE_ATTR(tun_flags, 0444, tun_show_flags, NULL); | |
915 | static DEVICE_ATTR(owner, 0444, tun_show_owner, NULL); | |
916 | static DEVICE_ATTR(group, 0444, tun_show_group, NULL); | |
917 | ||
d647a591 | 918 | static int tun_set_iff(struct net *net, struct file *file, struct ifreq *ifr) |
1da177e4 | 919 | { |
33dccbb0 | 920 | struct sock *sk; |
1da177e4 LT |
921 | struct tun_struct *tun; |
922 | struct net_device *dev; | |
923 | int err; | |
924 | ||
74a3e5a7 EB |
925 | dev = __dev_get_by_name(net, ifr->ifr_name); |
926 | if (dev) { | |
f85ba780 DW |
927 | if (ifr->ifr_flags & IFF_TUN_EXCL) |
928 | return -EBUSY; | |
74a3e5a7 EB |
929 | if ((ifr->ifr_flags & IFF_TUN) && dev->netdev_ops == &tun_netdev_ops) |
930 | tun = netdev_priv(dev); | |
931 | else if ((ifr->ifr_flags & IFF_TAP) && dev->netdev_ops == &tap_netdev_ops) | |
932 | tun = netdev_priv(dev); | |
933 | else | |
934 | return -EINVAL; | |
935 | ||
a7385ba2 EB |
936 | err = tun_attach(tun, file); |
937 | if (err < 0) | |
938 | return err; | |
6aa20a22 | 939 | } |
1da177e4 LT |
940 | else { |
941 | char *name; | |
942 | unsigned long flags = 0; | |
943 | ||
944 | err = -EINVAL; | |
945 | ||
ca6bb5d7 DW |
946 | if (!capable(CAP_NET_ADMIN)) |
947 | return -EPERM; | |
948 | ||
1da177e4 LT |
949 | /* Set dev type */ |
950 | if (ifr->ifr_flags & IFF_TUN) { | |
951 | /* TUN device */ | |
952 | flags |= TUN_TUN_DEV; | |
953 | name = "tun%d"; | |
954 | } else if (ifr->ifr_flags & IFF_TAP) { | |
955 | /* TAP device */ | |
956 | flags |= TUN_TAP_DEV; | |
957 | name = "tap%d"; | |
6aa20a22 | 958 | } else |
1da177e4 | 959 | goto failed; |
6aa20a22 | 960 | |
1da177e4 LT |
961 | if (*ifr->ifr_name) |
962 | name = ifr->ifr_name; | |
963 | ||
964 | dev = alloc_netdev(sizeof(struct tun_struct), name, | |
965 | tun_setup); | |
966 | if (!dev) | |
967 | return -ENOMEM; | |
968 | ||
fc54c658 | 969 | dev_net_set(dev, net); |
f019a7a5 | 970 | dev->rtnl_link_ops = &tun_link_ops; |
758e43b7 | 971 | |
1da177e4 LT |
972 | tun = netdev_priv(dev); |
973 | tun->dev = dev; | |
974 | tun->flags = flags; | |
f271b2cc | 975 | tun->txflt.count = 0; |
1da177e4 | 976 | |
33dccbb0 HX |
977 | err = -ENOMEM; |
978 | sk = sk_alloc(net, AF_UNSPEC, GFP_KERNEL, &tun_proto); | |
979 | if (!sk) | |
980 | goto err_free_dev; | |
981 | ||
c40af84a | 982 | init_waitqueue_head(&tun->socket.wait); |
33dccbb0 HX |
983 | sock_init_data(&tun->socket, sk); |
984 | sk->sk_write_space = tun_sock_write_space; | |
33dccbb0 | 985 | sk->sk_sndbuf = INT_MAX; |
33dccbb0 HX |
986 | |
987 | tun->sk = sk; | |
988 | container_of(sk, struct tun_sock, sk)->tun = tun; | |
989 | ||
1da177e4 LT |
990 | tun_net_init(dev); |
991 | ||
992 | if (strchr(dev->name, '%')) { | |
993 | err = dev_alloc_name(dev, dev->name); | |
994 | if (err < 0) | |
33dccbb0 | 995 | goto err_free_sk; |
1da177e4 LT |
996 | } |
997 | ||
33dccbb0 | 998 | err = -EINVAL; |
1da177e4 LT |
999 | err = register_netdevice(tun->dev); |
1000 | if (err < 0) | |
9c3fea6a HX |
1001 | goto err_free_sk; |
1002 | ||
980c9e8c DW |
1003 | if (device_create_file(&tun->dev->dev, &dev_attr_tun_flags) || |
1004 | device_create_file(&tun->dev->dev, &dev_attr_owner) || | |
1005 | device_create_file(&tun->dev->dev, &dev_attr_group)) | |
1006 | printk(KERN_ERR "Failed to create tun sysfs files\n"); | |
1007 | ||
9c3fea6a | 1008 | sk->sk_destruct = tun_sock_destruct; |
a7385ba2 EB |
1009 | |
1010 | err = tun_attach(tun, file); | |
1011 | if (err < 0) | |
9c3fea6a | 1012 | goto failed; |
1da177e4 LT |
1013 | } |
1014 | ||
1015 | DBG(KERN_INFO "%s: tun_set_iff\n", tun->dev->name); | |
1016 | ||
1017 | if (ifr->ifr_flags & IFF_NO_PI) | |
1018 | tun->flags |= TUN_NO_PI; | |
a26af1e0 NF |
1019 | else |
1020 | tun->flags &= ~TUN_NO_PI; | |
1da177e4 LT |
1021 | |
1022 | if (ifr->ifr_flags & IFF_ONE_QUEUE) | |
1023 | tun->flags |= TUN_ONE_QUEUE; | |
a26af1e0 NF |
1024 | else |
1025 | tun->flags &= ~TUN_ONE_QUEUE; | |
1da177e4 | 1026 | |
f43798c2 RR |
1027 | if (ifr->ifr_flags & IFF_VNET_HDR) |
1028 | tun->flags |= TUN_VNET_HDR; | |
1029 | else | |
1030 | tun->flags &= ~TUN_VNET_HDR; | |
1031 | ||
e35259a9 MK |
1032 | /* Make sure persistent devices do not get stuck in |
1033 | * xoff state. | |
1034 | */ | |
1035 | if (netif_running(tun->dev)) | |
1036 | netif_wake_queue(tun->dev); | |
1037 | ||
1da177e4 LT |
1038 | strcpy(ifr->ifr_name, tun->dev->name); |
1039 | return 0; | |
1040 | ||
33dccbb0 HX |
1041 | err_free_sk: |
1042 | sock_put(sk); | |
1da177e4 LT |
1043 | err_free_dev: |
1044 | free_netdev(dev); | |
1045 | failed: | |
1046 | return err; | |
1047 | } | |
1048 | ||
e3b99556 MM |
1049 | static int tun_get_iff(struct net *net, struct file *file, struct ifreq *ifr) |
1050 | { | |
631ab46b | 1051 | struct tun_struct *tun = tun_get(file); |
e3b99556 MM |
1052 | |
1053 | if (!tun) | |
1054 | return -EBADFD; | |
1055 | ||
1056 | DBG(KERN_INFO "%s: tun_get_iff\n", tun->dev->name); | |
1057 | ||
1058 | strcpy(ifr->ifr_name, tun->dev->name); | |
1059 | ||
980c9e8c | 1060 | ifr->ifr_flags = tun_flags(tun); |
e3b99556 | 1061 | |
631ab46b | 1062 | tun_put(tun); |
e3b99556 MM |
1063 | return 0; |
1064 | } | |
1065 | ||
5228ddc9 RR |
1066 | /* This is like a cut-down ethtool ops, except done via tun fd so no |
1067 | * privs required. */ | |
1068 | static int set_offload(struct net_device *dev, unsigned long arg) | |
1069 | { | |
1070 | unsigned int old_features, features; | |
1071 | ||
1072 | old_features = dev->features; | |
1073 | /* Unset features, set them as we chew on the arg. */ | |
1074 | features = (old_features & ~(NETIF_F_HW_CSUM|NETIF_F_SG|NETIF_F_FRAGLIST | |
1075 | |NETIF_F_TSO_ECN|NETIF_F_TSO|NETIF_F_TSO6)); | |
1076 | ||
1077 | if (arg & TUN_F_CSUM) { | |
1078 | features |= NETIF_F_HW_CSUM|NETIF_F_SG|NETIF_F_FRAGLIST; | |
1079 | arg &= ~TUN_F_CSUM; | |
1080 | ||
1081 | if (arg & (TUN_F_TSO4|TUN_F_TSO6)) { | |
1082 | if (arg & TUN_F_TSO_ECN) { | |
1083 | features |= NETIF_F_TSO_ECN; | |
1084 | arg &= ~TUN_F_TSO_ECN; | |
1085 | } | |
1086 | if (arg & TUN_F_TSO4) | |
1087 | features |= NETIF_F_TSO; | |
1088 | if (arg & TUN_F_TSO6) | |
1089 | features |= NETIF_F_TSO6; | |
1090 | arg &= ~(TUN_F_TSO4|TUN_F_TSO6); | |
1091 | } | |
1092 | } | |
1093 | ||
1094 | /* This gives the user a way to test for new features in future by | |
1095 | * trying to set them. */ | |
1096 | if (arg) | |
1097 | return -EINVAL; | |
1098 | ||
1099 | dev->features = features; | |
1100 | if (old_features != dev->features) | |
1101 | netdev_features_change(dev); | |
1102 | ||
1103 | return 0; | |
1104 | } | |
1105 | ||
6aa20a22 | 1106 | static int tun_chr_ioctl(struct inode *inode, struct file *file, |
1da177e4 LT |
1107 | unsigned int cmd, unsigned long arg) |
1108 | { | |
36b50bab | 1109 | struct tun_file *tfile = file->private_data; |
631ab46b | 1110 | struct tun_struct *tun; |
1da177e4 LT |
1111 | void __user* argp = (void __user*)arg; |
1112 | struct ifreq ifr; | |
33dccbb0 | 1113 | int sndbuf; |
f271b2cc | 1114 | int ret; |
1da177e4 LT |
1115 | |
1116 | if (cmd == TUNSETIFF || _IOC_TYPE(cmd) == 0x89) | |
1117 | if (copy_from_user(&ifr, argp, sizeof ifr)) | |
1118 | return -EFAULT; | |
1119 | ||
631ab46b EB |
1120 | if (cmd == TUNGETFEATURES) { |
1121 | /* Currently this just means: "what IFF flags are valid?". | |
1122 | * This is needed because we never checked for invalid flags on | |
1123 | * TUNSETIFF. */ | |
1124 | return put_user(IFF_TUN | IFF_TAP | IFF_NO_PI | IFF_ONE_QUEUE | | |
1125 | IFF_VNET_HDR, | |
1126 | (unsigned int __user*)argp); | |
1127 | } | |
1128 | ||
36b50bab | 1129 | tun = __tun_get(tfile); |
1da177e4 LT |
1130 | if (cmd == TUNSETIFF && !tun) { |
1131 | int err; | |
1132 | ||
1133 | ifr.ifr_name[IFNAMSIZ-1] = '\0'; | |
1134 | ||
1135 | rtnl_lock(); | |
36b50bab | 1136 | err = tun_set_iff(tfile->net, file, &ifr); |
1da177e4 LT |
1137 | rtnl_unlock(); |
1138 | ||
1139 | if (err) | |
1140 | return err; | |
1141 | ||
1142 | if (copy_to_user(argp, &ifr, sizeof(ifr))) | |
1143 | return -EFAULT; | |
1144 | return 0; | |
1145 | } | |
1146 | ||
07240fd0 | 1147 | |
1da177e4 LT |
1148 | if (!tun) |
1149 | return -EBADFD; | |
1150 | ||
1151 | DBG(KERN_INFO "%s: tun_chr_ioctl cmd %d\n", tun->dev->name, cmd); | |
1152 | ||
631ab46b | 1153 | ret = 0; |
1da177e4 | 1154 | switch (cmd) { |
e3b99556 MM |
1155 | case TUNGETIFF: |
1156 | ret = tun_get_iff(current->nsproxy->net_ns, file, &ifr); | |
1157 | if (ret) | |
631ab46b | 1158 | break; |
e3b99556 MM |
1159 | |
1160 | if (copy_to_user(argp, &ifr, sizeof(ifr))) | |
631ab46b | 1161 | ret = -EFAULT; |
e3b99556 MM |
1162 | break; |
1163 | ||
1da177e4 LT |
1164 | case TUNSETNOCSUM: |
1165 | /* Disable/Enable checksum */ | |
1166 | if (arg) | |
1167 | tun->flags |= TUN_NOCHECKSUM; | |
1168 | else | |
1169 | tun->flags &= ~TUN_NOCHECKSUM; | |
1170 | ||
1171 | DBG(KERN_INFO "%s: checksum %s\n", | |
1172 | tun->dev->name, arg ? "disabled" : "enabled"); | |
1173 | break; | |
1174 | ||
1175 | case TUNSETPERSIST: | |
1176 | /* Disable/Enable persist mode */ | |
1177 | if (arg) | |
1178 | tun->flags |= TUN_PERSIST; | |
1179 | else | |
1180 | tun->flags &= ~TUN_PERSIST; | |
1181 | ||
1182 | DBG(KERN_INFO "%s: persist %s\n", | |
c6e991de | 1183 | tun->dev->name, arg ? "enabled" : "disabled"); |
1da177e4 LT |
1184 | break; |
1185 | ||
1186 | case TUNSETOWNER: | |
1187 | /* Set owner of the device */ | |
1188 | tun->owner = (uid_t) arg; | |
1189 | ||
1190 | DBG(KERN_INFO "%s: owner set to %d\n", tun->dev->name, tun->owner); | |
1191 | break; | |
1192 | ||
8c644623 GG |
1193 | case TUNSETGROUP: |
1194 | /* Set group of the device */ | |
1195 | tun->group= (gid_t) arg; | |
1196 | ||
1197 | DBG(KERN_INFO "%s: group set to %d\n", tun->dev->name, tun->group); | |
1198 | break; | |
1199 | ||
ff4cc3ac MK |
1200 | case TUNSETLINK: |
1201 | /* Only allow setting the type when the interface is down */ | |
48abfe05 | 1202 | rtnl_lock(); |
ff4cc3ac MK |
1203 | if (tun->dev->flags & IFF_UP) { |
1204 | DBG(KERN_INFO "%s: Linktype set failed because interface is up\n", | |
1205 | tun->dev->name); | |
48abfe05 | 1206 | ret = -EBUSY; |
ff4cc3ac MK |
1207 | } else { |
1208 | tun->dev->type = (int) arg; | |
1209 | DBG(KERN_INFO "%s: linktype set to %d\n", tun->dev->name, tun->dev->type); | |
48abfe05 | 1210 | ret = 0; |
ff4cc3ac | 1211 | } |
48abfe05 | 1212 | rtnl_unlock(); |
631ab46b | 1213 | break; |
ff4cc3ac | 1214 | |
1da177e4 LT |
1215 | #ifdef TUN_DEBUG |
1216 | case TUNSETDEBUG: | |
1217 | tun->debug = arg; | |
1218 | break; | |
1219 | #endif | |
5228ddc9 | 1220 | case TUNSETOFFLOAD: |
5228ddc9 RR |
1221 | rtnl_lock(); |
1222 | ret = set_offload(tun->dev, arg); | |
1223 | rtnl_unlock(); | |
631ab46b | 1224 | break; |
5228ddc9 | 1225 | |
f271b2cc MK |
1226 | case TUNSETTXFILTER: |
1227 | /* Can be set only for TAPs */ | |
631ab46b | 1228 | ret = -EINVAL; |
f271b2cc | 1229 | if ((tun->flags & TUN_TYPE_MASK) != TUN_TAP_DEV) |
631ab46b | 1230 | break; |
f271b2cc | 1231 | rtnl_lock(); |
c0e5a8c2 | 1232 | ret = update_filter(&tun->txflt, (void __user *)arg); |
f271b2cc | 1233 | rtnl_unlock(); |
631ab46b | 1234 | break; |
1da177e4 LT |
1235 | |
1236 | case SIOCGIFHWADDR: | |
f271b2cc MK |
1237 | /* Get hw addres */ |
1238 | memcpy(ifr.ifr_hwaddr.sa_data, tun->dev->dev_addr, ETH_ALEN); | |
1239 | ifr.ifr_hwaddr.sa_family = tun->dev->type; | |
1240 | if (copy_to_user(argp, &ifr, sizeof ifr)) | |
631ab46b EB |
1241 | ret = -EFAULT; |
1242 | break; | |
1da177e4 LT |
1243 | |
1244 | case SIOCSIFHWADDR: | |
f271b2cc | 1245 | /* Set hw address */ |
e174961c JB |
1246 | DBG(KERN_DEBUG "%s: set hw address: %pM\n", |
1247 | tun->dev->name, ifr.ifr_hwaddr.sa_data); | |
40102371 KH |
1248 | |
1249 | rtnl_lock(); | |
1250 | ret = dev_set_mac_address(tun->dev, &ifr.ifr_hwaddr); | |
1251 | rtnl_unlock(); | |
631ab46b | 1252 | break; |
33dccbb0 HX |
1253 | |
1254 | case TUNGETSNDBUF: | |
1255 | sndbuf = tun->sk->sk_sndbuf; | |
1256 | if (copy_to_user(argp, &sndbuf, sizeof(sndbuf))) | |
1257 | ret = -EFAULT; | |
1258 | break; | |
1259 | ||
1260 | case TUNSETSNDBUF: | |
1261 | if (copy_from_user(&sndbuf, argp, sizeof(sndbuf))) { | |
1262 | ret = -EFAULT; | |
1263 | break; | |
1264 | } | |
1265 | ||
1266 | tun->sk->sk_sndbuf = sndbuf; | |
1267 | break; | |
1268 | ||
1da177e4 | 1269 | default: |
631ab46b EB |
1270 | ret = -EINVAL; |
1271 | break; | |
1da177e4 LT |
1272 | }; |
1273 | ||
631ab46b EB |
1274 | tun_put(tun); |
1275 | return ret; | |
1da177e4 LT |
1276 | } |
1277 | ||
1278 | static int tun_chr_fasync(int fd, struct file *file, int on) | |
1279 | { | |
631ab46b | 1280 | struct tun_struct *tun = tun_get(file); |
1da177e4 LT |
1281 | int ret; |
1282 | ||
1283 | if (!tun) | |
1284 | return -EBADFD; | |
1285 | ||
1286 | DBG(KERN_INFO "%s: tun_chr_fasync %d\n", tun->dev->name, on); | |
1287 | ||
9d319522 | 1288 | lock_kernel(); |
1da177e4 | 1289 | if ((ret = fasync_helper(fd, file, on, &tun->fasync)) < 0) |
9d319522 | 1290 | goto out; |
6aa20a22 | 1291 | |
1da177e4 | 1292 | if (on) { |
609d7fa9 | 1293 | ret = __f_setown(file, task_pid(current), PIDTYPE_PID, 0); |
1da177e4 | 1294 | if (ret) |
9d319522 | 1295 | goto out; |
1da177e4 | 1296 | tun->flags |= TUN_FASYNC; |
6aa20a22 | 1297 | } else |
1da177e4 | 1298 | tun->flags &= ~TUN_FASYNC; |
9d319522 JC |
1299 | ret = 0; |
1300 | out: | |
1301 | unlock_kernel(); | |
631ab46b | 1302 | tun_put(tun); |
9d319522 | 1303 | return ret; |
1da177e4 LT |
1304 | } |
1305 | ||
1306 | static int tun_chr_open(struct inode *inode, struct file * file) | |
1307 | { | |
631ab46b | 1308 | struct tun_file *tfile; |
fd3e05b6 | 1309 | cycle_kernel_lock(); |
1da177e4 | 1310 | DBG1(KERN_INFO "tunX: tun_chr_open\n"); |
631ab46b EB |
1311 | |
1312 | tfile = kmalloc(sizeof(*tfile), GFP_KERNEL); | |
1313 | if (!tfile) | |
1314 | return -ENOMEM; | |
c70f1829 | 1315 | atomic_set(&tfile->count, 0); |
631ab46b | 1316 | tfile->tun = NULL; |
36b50bab | 1317 | tfile->net = get_net(current->nsproxy->net_ns); |
631ab46b | 1318 | file->private_data = tfile; |
1da177e4 LT |
1319 | return 0; |
1320 | } | |
1321 | ||
1322 | static int tun_chr_close(struct inode *inode, struct file *file) | |
1323 | { | |
631ab46b EB |
1324 | struct tun_file *tfile = file->private_data; |
1325 | struct tun_struct *tun = __tun_get(tfile); | |
1da177e4 | 1326 | |
1da177e4 | 1327 | |
631ab46b EB |
1328 | if (tun) { |
1329 | DBG(KERN_INFO "%s: tun_chr_close\n", tun->dev->name); | |
1da177e4 | 1330 | |
631ab46b EB |
1331 | rtnl_lock(); |
1332 | __tun_detach(tun); | |
1da177e4 | 1333 | |
631ab46b | 1334 | /* If desireable, unregister the netdevice. */ |
9c3fea6a | 1335 | if (!(tun->flags & TUN_PERSIST)) |
631ab46b | 1336 | unregister_netdevice(tun->dev); |
1da177e4 | 1337 | |
631ab46b EB |
1338 | rtnl_unlock(); |
1339 | } | |
1da177e4 | 1340 | |
9c3fea6a HX |
1341 | tun = tfile->tun; |
1342 | if (tun) | |
1343 | sock_put(tun->sk); | |
1344 | ||
36b50bab | 1345 | put_net(tfile->net); |
631ab46b | 1346 | kfree(tfile); |
1da177e4 LT |
1347 | |
1348 | return 0; | |
1349 | } | |
1350 | ||
d54b1fdb | 1351 | static const struct file_operations tun_fops = { |
6aa20a22 | 1352 | .owner = THIS_MODULE, |
1da177e4 | 1353 | .llseek = no_llseek, |
ee0b3e67 BP |
1354 | .read = do_sync_read, |
1355 | .aio_read = tun_chr_aio_read, | |
1356 | .write = do_sync_write, | |
1357 | .aio_write = tun_chr_aio_write, | |
1da177e4 LT |
1358 | .poll = tun_chr_poll, |
1359 | .ioctl = tun_chr_ioctl, | |
1360 | .open = tun_chr_open, | |
1361 | .release = tun_chr_close, | |
6aa20a22 | 1362 | .fasync = tun_chr_fasync |
1da177e4 LT |
1363 | }; |
1364 | ||
1365 | static struct miscdevice tun_miscdev = { | |
1366 | .minor = TUN_MINOR, | |
1367 | .name = "tun", | |
1368 | .fops = &tun_fops, | |
1da177e4 LT |
1369 | }; |
1370 | ||
1371 | /* ethtool interface */ | |
1372 | ||
1373 | static int tun_get_settings(struct net_device *dev, struct ethtool_cmd *cmd) | |
1374 | { | |
1375 | cmd->supported = 0; | |
1376 | cmd->advertising = 0; | |
1377 | cmd->speed = SPEED_10; | |
1378 | cmd->duplex = DUPLEX_FULL; | |
1379 | cmd->port = PORT_TP; | |
1380 | cmd->phy_address = 0; | |
1381 | cmd->transceiver = XCVR_INTERNAL; | |
1382 | cmd->autoneg = AUTONEG_DISABLE; | |
1383 | cmd->maxtxpkt = 0; | |
1384 | cmd->maxrxpkt = 0; | |
1385 | return 0; | |
1386 | } | |
1387 | ||
1388 | static void tun_get_drvinfo(struct net_device *dev, struct ethtool_drvinfo *info) | |
1389 | { | |
1390 | struct tun_struct *tun = netdev_priv(dev); | |
1391 | ||
1392 | strcpy(info->driver, DRV_NAME); | |
1393 | strcpy(info->version, DRV_VERSION); | |
1394 | strcpy(info->fw_version, "N/A"); | |
1395 | ||
1396 | switch (tun->flags & TUN_TYPE_MASK) { | |
1397 | case TUN_TUN_DEV: | |
1398 | strcpy(info->bus_info, "tun"); | |
1399 | break; | |
1400 | case TUN_TAP_DEV: | |
1401 | strcpy(info->bus_info, "tap"); | |
1402 | break; | |
1403 | } | |
1404 | } | |
1405 | ||
1406 | static u32 tun_get_msglevel(struct net_device *dev) | |
1407 | { | |
1408 | #ifdef TUN_DEBUG | |
1409 | struct tun_struct *tun = netdev_priv(dev); | |
1410 | return tun->debug; | |
1411 | #else | |
1412 | return -EOPNOTSUPP; | |
1413 | #endif | |
1414 | } | |
1415 | ||
1416 | static void tun_set_msglevel(struct net_device *dev, u32 value) | |
1417 | { | |
1418 | #ifdef TUN_DEBUG | |
1419 | struct tun_struct *tun = netdev_priv(dev); | |
1420 | tun->debug = value; | |
1421 | #endif | |
1422 | } | |
1423 | ||
1424 | static u32 tun_get_link(struct net_device *dev) | |
1425 | { | |
1426 | struct tun_struct *tun = netdev_priv(dev); | |
631ab46b | 1427 | return !!tun->tfile; |
1da177e4 LT |
1428 | } |
1429 | ||
1430 | static u32 tun_get_rx_csum(struct net_device *dev) | |
1431 | { | |
1432 | struct tun_struct *tun = netdev_priv(dev); | |
1433 | return (tun->flags & TUN_NOCHECKSUM) == 0; | |
1434 | } | |
1435 | ||
1436 | static int tun_set_rx_csum(struct net_device *dev, u32 data) | |
1437 | { | |
1438 | struct tun_struct *tun = netdev_priv(dev); | |
1439 | if (data) | |
1440 | tun->flags &= ~TUN_NOCHECKSUM; | |
1441 | else | |
1442 | tun->flags |= TUN_NOCHECKSUM; | |
1443 | return 0; | |
1444 | } | |
1445 | ||
7282d491 | 1446 | static const struct ethtool_ops tun_ethtool_ops = { |
1da177e4 LT |
1447 | .get_settings = tun_get_settings, |
1448 | .get_drvinfo = tun_get_drvinfo, | |
1449 | .get_msglevel = tun_get_msglevel, | |
1450 | .set_msglevel = tun_set_msglevel, | |
1451 | .get_link = tun_get_link, | |
1452 | .get_rx_csum = tun_get_rx_csum, | |
1453 | .set_rx_csum = tun_set_rx_csum | |
1454 | }; | |
1455 | ||
79d17604 | 1456 | |
1da177e4 LT |
1457 | static int __init tun_init(void) |
1458 | { | |
1459 | int ret = 0; | |
1460 | ||
1461 | printk(KERN_INFO "tun: %s, %s\n", DRV_DESCRIPTION, DRV_VERSION); | |
1462 | printk(KERN_INFO "tun: %s\n", DRV_COPYRIGHT); | |
1463 | ||
f019a7a5 | 1464 | ret = rtnl_link_register(&tun_link_ops); |
79d17604 | 1465 | if (ret) { |
f019a7a5 EB |
1466 | printk(KERN_ERR "tun: Can't register link_ops\n"); |
1467 | goto err_linkops; | |
79d17604 PE |
1468 | } |
1469 | ||
1da177e4 | 1470 | ret = misc_register(&tun_miscdev); |
79d17604 | 1471 | if (ret) { |
1da177e4 | 1472 | printk(KERN_ERR "tun: Can't register misc device %d\n", TUN_MINOR); |
79d17604 PE |
1473 | goto err_misc; |
1474 | } | |
f019a7a5 | 1475 | return 0; |
79d17604 | 1476 | err_misc: |
f019a7a5 EB |
1477 | rtnl_link_unregister(&tun_link_ops); |
1478 | err_linkops: | |
1da177e4 LT |
1479 | return ret; |
1480 | } | |
1481 | ||
1482 | static void tun_cleanup(void) | |
1483 | { | |
6aa20a22 | 1484 | misc_deregister(&tun_miscdev); |
f019a7a5 | 1485 | rtnl_link_unregister(&tun_link_ops); |
1da177e4 LT |
1486 | } |
1487 | ||
1488 | module_init(tun_init); | |
1489 | module_exit(tun_cleanup); | |
1490 | MODULE_DESCRIPTION(DRV_DESCRIPTION); | |
1491 | MODULE_AUTHOR(DRV_COPYRIGHT); | |
1492 | MODULE_LICENSE("GPL"); | |
1493 | MODULE_ALIAS_MISCDEV(TUN_MINOR); |