blob: 3bb991fd2b512b9bd20bb22066982e03f3b3245a [file] [log] [blame]
Linus Torvalds1da177e2005-04-16 15:20:36 -07001/*
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 *
Brian Braunstein36226a82007-04-26 01:00:55 -070021 * Brian Braunstein <linuxkernel@bristyle.com> 2007/03/23
22 * Fixed hw address handling. Now net_device.dev_addr is kept consistent
23 * with tun.dev_addr when the address is set by this module.
24 *
Mike Kershawff4cc3a2005-09-01 17:40:05 -070025 * Mike Kershaw <dragorn@kismetwireless.net> 2005/08/14
26 * Add TUNSETLINK ioctl to set the link encapsulation
27 *
Linus Torvalds1da177e2005-04-16 15:20:36 -070028 * Mark Smith <markzzzsmith@yahoo.com.au>
29 * Use random_ether_addr() for tap MAC address.
30 *
31 * Harald Roelle <harald.roelle@ifi.lmu.de> 2004/04/20
32 * Fixes in packet dropping, queue length setting and queue wakeup.
33 * Increased default tx queue length.
34 * Added ethtool API.
35 * Minor cleanups
36 *
37 * Daniel Podlejski <underley@underley.eu.org>
38 * Modifications for 2.3.99-pre5 kernel.
39 */
40
41#define DRV_NAME "tun"
42#define DRV_VERSION "1.6"
43#define DRV_DESCRIPTION "Universal TUN/TAP device driver"
44#define DRV_COPYRIGHT "(C) 1999-2004 Max Krasnyansky <maxk@qualcomm.com>"
45
Linus Torvalds1da177e2005-04-16 15:20:36 -070046#include <linux/module.h>
47#include <linux/errno.h>
48#include <linux/kernel.h>
49#include <linux/major.h>
50#include <linux/slab.h>
51#include <linux/poll.h>
52#include <linux/fcntl.h>
53#include <linux/init.h>
54#include <linux/skbuff.h>
55#include <linux/netdevice.h>
56#include <linux/etherdevice.h>
57#include <linux/miscdevice.h>
58#include <linux/ethtool.h>
59#include <linux/rtnetlink.h>
60#include <linux/if.h>
61#include <linux/if_arp.h>
62#include <linux/if_ether.h>
63#include <linux/if_tun.h>
64#include <linux/crc32.h>
Pavel Emelyanovd647a592008-04-16 00:41:16 -070065#include <linux/nsproxy.h>
Eric W. Biederman881d9662007-09-17 11:56:21 -070066#include <net/net_namespace.h>
Pavel Emelyanov79d17602008-04-16 00:40:46 -070067#include <net/netns/generic.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070068
69#include <asm/system.h>
70#include <asm/uaccess.h>
71
Rusty Russell14daa022008-04-12 18:48:58 -070072/* Uncomment to enable debugging */
73/* #define TUN_DEBUG 1 */
74
Linus Torvalds1da177e2005-04-16 15:20:36 -070075#ifdef TUN_DEBUG
76static int debug;
Rusty Russell14daa022008-04-12 18:48:58 -070077
78#define DBG if(tun->debug)printk
79#define DBG1 if(debug==2)printk
80#else
81#define DBG( a... )
82#define DBG1( a... )
Linus Torvalds1da177e2005-04-16 15:20:36 -070083#endif
84
Rusty Russell14daa022008-04-12 18:48:58 -070085struct tun_struct {
86 struct list_head list;
87 unsigned long flags;
88 int attached;
89 uid_t owner;
90 gid_t group;
91
92 wait_queue_head_t read_wait;
93 struct sk_buff_head readq;
94
95 struct net_device *dev;
96
97 struct fasync_struct *fasync;
98
99 unsigned long if_flags;
100 u8 dev_addr[ETH_ALEN];
101 u32 chr_filter[2];
102 u32 net_filter[2];
103
104#ifdef TUN_DEBUG
105 int debug;
106#endif
107};
108
Linus Torvalds1da177e2005-04-16 15:20:36 -0700109/* Network device part of the driver */
110
Pavel Emelyanov79d17602008-04-16 00:40:46 -0700111static unsigned int tun_net_id;
112struct tun_net {
113 struct list_head dev_list;
114};
115
Jeff Garzik7282d492006-09-13 14:30:00 -0400116static const struct ethtool_ops tun_ethtool_ops;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700117
118/* Net device open. */
119static int tun_net_open(struct net_device *dev)
120{
121 netif_start_queue(dev);
122 return 0;
123}
124
125/* Net device close. */
126static int tun_net_close(struct net_device *dev)
127{
128 netif_stop_queue(dev);
129 return 0;
130}
131
132/* Net device start xmit */
133static int tun_net_xmit(struct sk_buff *skb, struct net_device *dev)
134{
135 struct tun_struct *tun = netdev_priv(dev);
136
137 DBG(KERN_INFO "%s: tun_net_xmit %d\n", tun->dev->name, skb->len);
138
139 /* Drop packet if interface is not attached */
140 if (!tun->attached)
141 goto drop;
142
143 /* Packet dropping */
144 if (skb_queue_len(&tun->readq) >= dev->tx_queue_len) {
145 if (!(tun->flags & TUN_ONE_QUEUE)) {
146 /* Normal queueing mode. */
147 /* Packet scheduler handles dropping of further packets. */
148 netif_stop_queue(dev);
149
150 /* We won't see all dropped packets individually, so overrun
151 * error is more appropriate. */
Jeff Garzik09f75cd2007-10-03 17:41:50 -0700152 dev->stats.tx_fifo_errors++;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700153 } else {
154 /* Single queue mode.
155 * Driver handles dropping of all packets itself. */
156 goto drop;
157 }
158 }
159
160 /* Queue packet */
161 skb_queue_tail(&tun->readq, skb);
162 dev->trans_start = jiffies;
163
164 /* Notify and wake up reader process */
165 if (tun->flags & TUN_FASYNC)
166 kill_fasync(&tun->fasync, SIGIO, POLL_IN);
167 wake_up_interruptible(&tun->read_wait);
168 return 0;
169
170drop:
Jeff Garzik09f75cd2007-10-03 17:41:50 -0700171 dev->stats.tx_dropped++;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700172 kfree_skb(skb);
173 return 0;
174}
175
176/** Add the specified Ethernet address to this multicast filter. */
177static void
178add_multi(u32* filter, const u8* addr)
179{
180 int bit_nr = ether_crc(ETH_ALEN, addr) >> 26;
181 filter[bit_nr >> 5] |= 1 << (bit_nr & 31);
182}
183
184/** Remove the specified Ethernet addres from this multicast filter. */
185static void
186del_multi(u32* filter, const u8* addr)
187{
188 int bit_nr = ether_crc(ETH_ALEN, addr) >> 26;
189 filter[bit_nr >> 5] &= ~(1 << (bit_nr & 31));
190}
191
192/** Update the list of multicast groups to which the network device belongs.
193 * This list is used to filter packets being sent from the character device to
194 * the network device. */
195static void
196tun_net_mclist(struct net_device *dev)
197{
198 struct tun_struct *tun = netdev_priv(dev);
199 const struct dev_mc_list *mclist;
200 int i;
Joe Perches0795af52007-10-03 17:59:30 -0700201 DECLARE_MAC_BUF(mac);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700202 DBG(KERN_DEBUG "%s: tun_net_mclist: mc_count %d\n",
203 dev->name, dev->mc_count);
204 memset(tun->chr_filter, 0, sizeof tun->chr_filter);
205 for (i = 0, mclist = dev->mc_list; i < dev->mc_count && mclist != NULL;
206 i++, mclist = mclist->next) {
207 add_multi(tun->net_filter, mclist->dmi_addr);
Joe Perches0795af52007-10-03 17:59:30 -0700208 DBG(KERN_DEBUG "%s: tun_net_mclist: %s\n",
209 dev->name, print_mac(mac, mclist->dmi_addr));
Linus Torvalds1da177e2005-04-16 15:20:36 -0700210 }
211}
212
Ed Swierk4885a502007-09-16 12:21:38 -0700213#define MIN_MTU 68
214#define MAX_MTU 65535
215
216static int
217tun_net_change_mtu(struct net_device *dev, int new_mtu)
218{
219 if (new_mtu < MIN_MTU || new_mtu + dev->hard_header_len > MAX_MTU)
220 return -EINVAL;
221 dev->mtu = new_mtu;
222 return 0;
223}
224
Linus Torvalds1da177e2005-04-16 15:20:36 -0700225/* Initialize net device. */
226static void tun_net_init(struct net_device *dev)
227{
228 struct tun_struct *tun = netdev_priv(dev);
Jeff Garzik6aa20a22006-09-13 13:24:59 -0400229
Linus Torvalds1da177e2005-04-16 15:20:36 -0700230 switch (tun->flags & TUN_TYPE_MASK) {
231 case TUN_TUN_DEV:
232 /* Point-to-Point TUN Device */
233 dev->hard_header_len = 0;
234 dev->addr_len = 0;
235 dev->mtu = 1500;
Ed Swierk4885a502007-09-16 12:21:38 -0700236 dev->change_mtu = tun_net_change_mtu;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700237
238 /* Zero header length */
Jeff Garzik6aa20a22006-09-13 13:24:59 -0400239 dev->type = ARPHRD_NONE;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700240 dev->flags = IFF_POINTOPOINT | IFF_NOARP | IFF_MULTICAST;
241 dev->tx_queue_len = TUN_READQ_SIZE; /* We prefer our own queue length */
242 break;
243
244 case TUN_TAP_DEV:
245 /* Ethernet TAP Device */
246 dev->set_multicast_list = tun_net_mclist;
247
248 ether_setup(dev);
Ed Swierk4885a502007-09-16 12:21:38 -0700249 dev->change_mtu = tun_net_change_mtu;
Brian Braunstein36226a82007-04-26 01:00:55 -0700250
251 /* random address already created for us by tun_set_iff, use it */
252 memcpy(dev->dev_addr, tun->dev_addr, min(sizeof(tun->dev_addr), sizeof(dev->dev_addr)) );
253
Linus Torvalds1da177e2005-04-16 15:20:36 -0700254 dev->tx_queue_len = TUN_READQ_SIZE; /* We prefer our own queue length */
255 break;
256 }
257}
258
259/* Character device part */
260
261/* Poll */
262static unsigned int tun_chr_poll(struct file *file, poll_table * wait)
Jeff Garzik6aa20a22006-09-13 13:24:59 -0400263{
Linus Torvalds1da177e2005-04-16 15:20:36 -0700264 struct tun_struct *tun = file->private_data;
265 unsigned int mask = POLLOUT | POLLWRNORM;
266
267 if (!tun)
268 return -EBADFD;
269
270 DBG(KERN_INFO "%s: tun_chr_poll\n", tun->dev->name);
271
272 poll_wait(file, &tun->read_wait, wait);
Jeff Garzik6aa20a22006-09-13 13:24:59 -0400273
David S. Millerb03efcf2005-07-08 14:57:23 -0700274 if (!skb_queue_empty(&tun->readq))
Linus Torvalds1da177e2005-04-16 15:20:36 -0700275 mask |= POLLIN | POLLRDNORM;
276
277 return mask;
278}
279
280/* Get packet from user space buffer */
281static __inline__ ssize_t tun_get_user(struct tun_struct *tun, struct iovec *iv, size_t count)
282{
283 struct tun_pi pi = { 0, __constant_htons(ETH_P_IP) };
284 struct sk_buff *skb;
285 size_t len = count, align = 0;
286
287 if (!(tun->flags & TUN_NO_PI)) {
288 if ((len -= sizeof(pi)) > count)
289 return -EINVAL;
290
291 if(memcpy_fromiovec((void *)&pi, iv, sizeof(pi)))
292 return -EFAULT;
293 }
294
Rusty Russelle01bf1c2008-04-12 18:49:30 -0700295 if ((tun->flags & TUN_TYPE_MASK) == TUN_TAP_DEV) {
Linus Torvalds1da177e2005-04-16 15:20:36 -0700296 align = NET_IP_ALIGN;
Rusty Russelle01bf1c2008-04-12 18:49:30 -0700297 if (unlikely(len < ETH_HLEN))
298 return -EINVAL;
299 }
Jeff Garzik6aa20a22006-09-13 13:24:59 -0400300
Linus Torvalds1da177e2005-04-16 15:20:36 -0700301 if (!(skb = alloc_skb(len + align, GFP_KERNEL))) {
Jeff Garzik09f75cd2007-10-03 17:41:50 -0700302 tun->dev->stats.rx_dropped++;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700303 return -ENOMEM;
304 }
305
306 if (align)
307 skb_reserve(skb, align);
Dave Jones8f227572006-03-11 18:49:13 -0800308 if (memcpy_fromiovec(skb_put(skb, len), iv, len)) {
Jeff Garzik09f75cd2007-10-03 17:41:50 -0700309 tun->dev->stats.rx_dropped++;
Dave Jones8f227572006-03-11 18:49:13 -0800310 kfree_skb(skb);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700311 return -EFAULT;
Dave Jones8f227572006-03-11 18:49:13 -0800312 }
Linus Torvalds1da177e2005-04-16 15:20:36 -0700313
Linus Torvalds1da177e2005-04-16 15:20:36 -0700314 switch (tun->flags & TUN_TYPE_MASK) {
315 case TUN_TUN_DEV:
Ang Way Chuangf09f7ee2008-06-17 21:10:33 -0700316 if (tun->flags & TUN_NO_PI) {
317 switch (skb->data[0] & 0xf0) {
318 case 0x40:
319 pi.proto = htons(ETH_P_IP);
320 break;
321 case 0x60:
322 pi.proto = htons(ETH_P_IPV6);
323 break;
324 default:
325 tun->dev->stats.rx_dropped++;
326 kfree_skb(skb);
327 return -EINVAL;
328 }
329 }
330
Arnaldo Carvalho de Melo459a98e2007-03-19 15:30:44 -0700331 skb_reset_mac_header(skb);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700332 skb->protocol = pi.proto;
Arnaldo Carvalho de Melo4c13eb62007-04-25 17:40:23 -0700333 skb->dev = tun->dev;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700334 break;
335 case TUN_TAP_DEV:
336 skb->protocol = eth_type_trans(skb, tun->dev);
337 break;
338 };
339
340 if (tun->flags & TUN_NOCHECKSUM)
341 skb->ip_summed = CHECKSUM_UNNECESSARY;
Jeff Garzik6aa20a22006-09-13 13:24:59 -0400342
Linus Torvalds1da177e2005-04-16 15:20:36 -0700343 netif_rx_ni(skb);
344 tun->dev->last_rx = jiffies;
Jeff Garzik6aa20a22006-09-13 13:24:59 -0400345
Jeff Garzik09f75cd2007-10-03 17:41:50 -0700346 tun->dev->stats.rx_packets++;
347 tun->dev->stats.rx_bytes += len;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700348
349 return count;
Jeff Garzik6aa20a22006-09-13 13:24:59 -0400350}
Linus Torvalds1da177e2005-04-16 15:20:36 -0700351
Badari Pulavartyee0b3e62006-09-30 23:28:47 -0700352static ssize_t tun_chr_aio_write(struct kiocb *iocb, const struct iovec *iv,
353 unsigned long count, loff_t pos)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700354{
Badari Pulavartyee0b3e62006-09-30 23:28:47 -0700355 struct tun_struct *tun = iocb->ki_filp->private_data;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700356
357 if (!tun)
358 return -EBADFD;
359
360 DBG(KERN_INFO "%s: tun_chr_write %ld\n", tun->dev->name, count);
361
Akinobu Mita52427c92007-11-19 22:46:51 -0800362 return tun_get_user(tun, (struct iovec *) iv, iov_length(iv, count));
Linus Torvalds1da177e2005-04-16 15:20:36 -0700363}
364
Linus Torvalds1da177e2005-04-16 15:20:36 -0700365/* Put packet to the user space buffer */
366static __inline__ ssize_t tun_put_user(struct tun_struct *tun,
367 struct sk_buff *skb,
368 struct iovec *iv, int len)
369{
370 struct tun_pi pi = { 0, skb->protocol };
371 ssize_t total = 0;
372
373 if (!(tun->flags & TUN_NO_PI)) {
374 if ((len -= sizeof(pi)) < 0)
375 return -EINVAL;
376
377 if (len < skb->len) {
378 /* Packet will be striped */
379 pi.flags |= TUN_PKT_STRIP;
380 }
Jeff Garzik6aa20a22006-09-13 13:24:59 -0400381
Linus Torvalds1da177e2005-04-16 15:20:36 -0700382 if (memcpy_toiovec(iv, (void *) &pi, sizeof(pi)))
383 return -EFAULT;
384 total += sizeof(pi);
Jeff Garzik6aa20a22006-09-13 13:24:59 -0400385 }
Linus Torvalds1da177e2005-04-16 15:20:36 -0700386
387 len = min_t(int, skb->len, len);
388
389 skb_copy_datagram_iovec(skb, 0, iv, len);
390 total += len;
391
Jeff Garzik09f75cd2007-10-03 17:41:50 -0700392 tun->dev->stats.tx_packets++;
393 tun->dev->stats.tx_bytes += len;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700394
395 return total;
396}
397
Badari Pulavartyee0b3e62006-09-30 23:28:47 -0700398static ssize_t tun_chr_aio_read(struct kiocb *iocb, const struct iovec *iv,
399 unsigned long count, loff_t pos)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700400{
Badari Pulavartyee0b3e62006-09-30 23:28:47 -0700401 struct file *file = iocb->ki_filp;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700402 struct tun_struct *tun = file->private_data;
403 DECLARE_WAITQUEUE(wait, current);
404 struct sk_buff *skb;
405 ssize_t len, ret = 0;
Joe Perches0795af52007-10-03 17:59:30 -0700406 DECLARE_MAC_BUF(mac);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700407
408 if (!tun)
409 return -EBADFD;
410
411 DBG(KERN_INFO "%s: tun_chr_read\n", tun->dev->name);
412
Akinobu Mita52427c92007-11-19 22:46:51 -0800413 len = iov_length(iv, count);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700414 if (len < 0)
415 return -EINVAL;
416
417 add_wait_queue(&tun->read_wait, &wait);
418 while (len) {
419 const u8 ones[ ETH_ALEN] = { 0xff, 0xff, 0xff, 0xff, 0xff, 0xff };
420 u8 addr[ ETH_ALEN];
421 int bit_nr;
422
423 current->state = TASK_INTERRUPTIBLE;
424
425 /* Read frames from the queue */
426 if (!(skb=skb_dequeue(&tun->readq))) {
427 if (file->f_flags & O_NONBLOCK) {
428 ret = -EAGAIN;
429 break;
430 }
431 if (signal_pending(current)) {
432 ret = -ERESTARTSYS;
433 break;
434 }
435
436 /* Nothing to read, let's sleep */
437 schedule();
438 continue;
439 }
440 netif_wake_queue(tun->dev);
441
442 /** Decide whether to accept this packet. This code is designed to
443 * behave identically to an Ethernet interface. Accept the packet if
444 * - we are promiscuous.
445 * - the packet is addressed to us.
446 * - the packet is broadcast.
447 * - the packet is multicast and
448 * - we are multicast promiscous.
449 * - we belong to the multicast group.
450 */
Arnaldo Carvalho de Melod626f622007-03-27 18:55:52 -0300451 skb_copy_from_linear_data(skb, addr, min_t(size_t, sizeof addr,
452 skb->len));
Linus Torvalds1da177e2005-04-16 15:20:36 -0700453 bit_nr = ether_crc(sizeof addr, addr) >> 26;
454 if ((tun->if_flags & IFF_PROMISC) ||
455 memcmp(addr, tun->dev_addr, sizeof addr) == 0 ||
456 memcmp(addr, ones, sizeof addr) == 0 ||
457 (((addr[0] == 1 && addr[1] == 0 && addr[2] == 0x5e) ||
458 (addr[0] == 0x33 && addr[1] == 0x33)) &&
459 ((tun->if_flags & IFF_ALLMULTI) ||
460 (tun->chr_filter[bit_nr >> 5] & (1 << (bit_nr & 31)))))) {
Joe Perches0795af52007-10-03 17:59:30 -0700461 DBG(KERN_DEBUG "%s: tun_chr_readv: accepted: %s\n",
462 tun->dev->name, print_mac(mac, addr));
Linus Torvalds1da177e2005-04-16 15:20:36 -0700463 ret = tun_put_user(tun, skb, (struct iovec *) iv, len);
464 kfree_skb(skb);
465 break;
466 } else {
Joe Perches0795af52007-10-03 17:59:30 -0700467 DBG(KERN_DEBUG "%s: tun_chr_readv: rejected: %s\n",
468 tun->dev->name, print_mac(mac, addr));
Linus Torvalds1da177e2005-04-16 15:20:36 -0700469 kfree_skb(skb);
470 continue;
471 }
472 }
473
474 current->state = TASK_RUNNING;
475 remove_wait_queue(&tun->read_wait, &wait);
476
477 return ret;
478}
479
Linus Torvalds1da177e2005-04-16 15:20:36 -0700480static void tun_setup(struct net_device *dev)
481{
482 struct tun_struct *tun = netdev_priv(dev);
483
484 skb_queue_head_init(&tun->readq);
485 init_waitqueue_head(&tun->read_wait);
486
487 tun->owner = -1;
Guido Guenther8c644622007-07-02 22:50:25 -0700488 tun->group = -1;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700489
Linus Torvalds1da177e2005-04-16 15:20:36 -0700490 dev->open = tun_net_open;
491 dev->hard_start_xmit = tun_net_xmit;
492 dev->stop = tun_net_close;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700493 dev->ethtool_ops = &tun_ethtool_ops;
494 dev->destructor = free_netdev;
Pavel Emelyanovfc54c652008-04-16 00:41:53 -0700495 dev->features |= NETIF_F_NETNS_LOCAL;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700496}
497
Pavel Emelyanovd647a592008-04-16 00:41:16 -0700498static struct tun_struct *tun_get_by_name(struct tun_net *tn, const char *name)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700499{
500 struct tun_struct *tun;
501
502 ASSERT_RTNL();
Pavel Emelyanovd647a592008-04-16 00:41:16 -0700503 list_for_each_entry(tun, &tn->dev_list, list) {
Linus Torvalds1da177e2005-04-16 15:20:36 -0700504 if (!strncmp(tun->dev->name, name, IFNAMSIZ))
505 return tun;
506 }
507
508 return NULL;
509}
510
Pavel Emelyanovd647a592008-04-16 00:41:16 -0700511static int tun_set_iff(struct net *net, struct file *file, struct ifreq *ifr)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700512{
Pavel Emelyanovd647a592008-04-16 00:41:16 -0700513 struct tun_net *tn;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700514 struct tun_struct *tun;
515 struct net_device *dev;
516 int err;
517
Pavel Emelyanovd647a592008-04-16 00:41:16 -0700518 tn = net_generic(net, tun_net_id);
519 tun = tun_get_by_name(tn, ifr->ifr_name);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700520 if (tun) {
521 if (tun->attached)
522 return -EBUSY;
523
524 /* Check permissions */
Guido Guenther8c644622007-07-02 22:50:25 -0700525 if (((tun->owner != -1 &&
526 current->euid != tun->owner) ||
527 (tun->group != -1 &&
528 current->egid != tun->group)) &&
529 !capable(CAP_NET_ADMIN))
Linus Torvalds1da177e2005-04-16 15:20:36 -0700530 return -EPERM;
Jeff Garzik6aa20a22006-09-13 13:24:59 -0400531 }
Pavel Emelyanovd647a592008-04-16 00:41:16 -0700532 else if (__dev_get_by_name(net, ifr->ifr_name))
Linus Torvalds1da177e2005-04-16 15:20:36 -0700533 return -EINVAL;
534 else {
535 char *name;
536 unsigned long flags = 0;
537
538 err = -EINVAL;
539
David Woodhouseca6bb5d2006-06-22 16:07:52 -0700540 if (!capable(CAP_NET_ADMIN))
541 return -EPERM;
542
Linus Torvalds1da177e2005-04-16 15:20:36 -0700543 /* Set dev type */
544 if (ifr->ifr_flags & IFF_TUN) {
545 /* TUN device */
546 flags |= TUN_TUN_DEV;
547 name = "tun%d";
548 } else if (ifr->ifr_flags & IFF_TAP) {
549 /* TAP device */
550 flags |= TUN_TAP_DEV;
551 name = "tap%d";
Jeff Garzik6aa20a22006-09-13 13:24:59 -0400552 } else
Linus Torvalds1da177e2005-04-16 15:20:36 -0700553 goto failed;
Jeff Garzik6aa20a22006-09-13 13:24:59 -0400554
Linus Torvalds1da177e2005-04-16 15:20:36 -0700555 if (*ifr->ifr_name)
556 name = ifr->ifr_name;
557
558 dev = alloc_netdev(sizeof(struct tun_struct), name,
559 tun_setup);
560 if (!dev)
561 return -ENOMEM;
562
Pavel Emelyanovfc54c652008-04-16 00:41:53 -0700563 dev_net_set(dev, net);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700564 tun = netdev_priv(dev);
565 tun->dev = dev;
566 tun->flags = flags;
567 /* Be promiscuous by default to maintain previous behaviour. */
568 tun->if_flags = IFF_PROMISC;
569 /* Generate random Ethernet address. */
Al Viroa3edb082007-12-22 17:52:42 +0000570 *(__be16 *)tun->dev_addr = htons(0x00FF);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700571 get_random_bytes(tun->dev_addr + sizeof(u16), 4);
572 memset(tun->chr_filter, 0, sizeof tun->chr_filter);
573
574 tun_net_init(dev);
575
576 if (strchr(dev->name, '%')) {
577 err = dev_alloc_name(dev, dev->name);
578 if (err < 0)
579 goto err_free_dev;
580 }
581
582 err = register_netdevice(tun->dev);
583 if (err < 0)
584 goto err_free_dev;
Jeff Garzik6aa20a22006-09-13 13:24:59 -0400585
Pavel Emelyanovd647a592008-04-16 00:41:16 -0700586 list_add(&tun->list, &tn->dev_list);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700587 }
588
589 DBG(KERN_INFO "%s: tun_set_iff\n", tun->dev->name);
590
591 if (ifr->ifr_flags & IFF_NO_PI)
592 tun->flags |= TUN_NO_PI;
Nathaniel Filardoa26af1e2008-02-05 03:05:07 -0800593 else
594 tun->flags &= ~TUN_NO_PI;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700595
596 if (ifr->ifr_flags & IFF_ONE_QUEUE)
597 tun->flags |= TUN_ONE_QUEUE;
Nathaniel Filardoa26af1e2008-02-05 03:05:07 -0800598 else
599 tun->flags &= ~TUN_ONE_QUEUE;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700600
601 file->private_data = tun;
602 tun->attached = 1;
Pavel Emelyanovfc54c652008-04-16 00:41:53 -0700603 get_net(dev_net(tun->dev));
Linus Torvalds1da177e2005-04-16 15:20:36 -0700604
605 strcpy(ifr->ifr_name, tun->dev->name);
606 return 0;
607
608 err_free_dev:
609 free_netdev(dev);
610 failed:
611 return err;
612}
613
Jeff Garzik6aa20a22006-09-13 13:24:59 -0400614static int tun_chr_ioctl(struct inode *inode, struct file *file,
Linus Torvalds1da177e2005-04-16 15:20:36 -0700615 unsigned int cmd, unsigned long arg)
616{
617 struct tun_struct *tun = file->private_data;
618 void __user* argp = (void __user*)arg;
619 struct ifreq ifr;
Joe Perches0795af52007-10-03 17:59:30 -0700620 DECLARE_MAC_BUF(mac);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700621
622 if (cmd == TUNSETIFF || _IOC_TYPE(cmd) == 0x89)
623 if (copy_from_user(&ifr, argp, sizeof ifr))
624 return -EFAULT;
625
626 if (cmd == TUNSETIFF && !tun) {
627 int err;
628
629 ifr.ifr_name[IFNAMSIZ-1] = '\0';
630
631 rtnl_lock();
Pavel Emelyanovd647a592008-04-16 00:41:16 -0700632 err = tun_set_iff(current->nsproxy->net_ns, file, &ifr);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700633 rtnl_unlock();
634
635 if (err)
636 return err;
637
638 if (copy_to_user(argp, &ifr, sizeof(ifr)))
639 return -EFAULT;
640 return 0;
641 }
642
Rusty Russell07240fd2008-07-03 03:45:32 -0700643 if (cmd == TUNGETFEATURES) {
644 /* Currently this just means: "what IFF flags are valid?".
645 * This is needed because we never checked for invalid flags on
646 * TUNSETIFF. */
647 return put_user(IFF_TUN | IFF_TAP | IFF_NO_PI | IFF_ONE_QUEUE,
648 (unsigned int __user*)argp);
649 }
650
Linus Torvalds1da177e2005-04-16 15:20:36 -0700651 if (!tun)
652 return -EBADFD;
653
654 DBG(KERN_INFO "%s: tun_chr_ioctl cmd %d\n", tun->dev->name, cmd);
655
656 switch (cmd) {
657 case TUNSETNOCSUM:
658 /* Disable/Enable checksum */
659 if (arg)
660 tun->flags |= TUN_NOCHECKSUM;
661 else
662 tun->flags &= ~TUN_NOCHECKSUM;
663
664 DBG(KERN_INFO "%s: checksum %s\n",
665 tun->dev->name, arg ? "disabled" : "enabled");
666 break;
667
668 case TUNSETPERSIST:
669 /* Disable/Enable persist mode */
670 if (arg)
671 tun->flags |= TUN_PERSIST;
672 else
673 tun->flags &= ~TUN_PERSIST;
674
675 DBG(KERN_INFO "%s: persist %s\n",
Toyo Abec6e991d2007-12-24 21:29:35 -0800676 tun->dev->name, arg ? "enabled" : "disabled");
Linus Torvalds1da177e2005-04-16 15:20:36 -0700677 break;
678
679 case TUNSETOWNER:
680 /* Set owner of the device */
681 tun->owner = (uid_t) arg;
682
683 DBG(KERN_INFO "%s: owner set to %d\n", tun->dev->name, tun->owner);
684 break;
685
Guido Guenther8c644622007-07-02 22:50:25 -0700686 case TUNSETGROUP:
687 /* Set group of the device */
688 tun->group= (gid_t) arg;
689
690 DBG(KERN_INFO "%s: group set to %d\n", tun->dev->name, tun->group);
691 break;
692
Mike Kershawff4cc3a2005-09-01 17:40:05 -0700693 case TUNSETLINK:
David S. Miller48abfe02008-04-23 19:37:58 -0700694 {
695 int ret;
696
Mike Kershawff4cc3a2005-09-01 17:40:05 -0700697 /* Only allow setting the type when the interface is down */
David S. Miller48abfe02008-04-23 19:37:58 -0700698 rtnl_lock();
Mike Kershawff4cc3a2005-09-01 17:40:05 -0700699 if (tun->dev->flags & IFF_UP) {
700 DBG(KERN_INFO "%s: Linktype set failed because interface is up\n",
701 tun->dev->name);
David S. Miller48abfe02008-04-23 19:37:58 -0700702 ret = -EBUSY;
Mike Kershawff4cc3a2005-09-01 17:40:05 -0700703 } else {
704 tun->dev->type = (int) arg;
705 DBG(KERN_INFO "%s: linktype set to %d\n", tun->dev->name, tun->dev->type);
David S. Miller48abfe02008-04-23 19:37:58 -0700706 ret = 0;
Mike Kershawff4cc3a2005-09-01 17:40:05 -0700707 }
David S. Miller48abfe02008-04-23 19:37:58 -0700708 rtnl_unlock();
709 return ret;
710 }
Mike Kershawff4cc3a2005-09-01 17:40:05 -0700711
Linus Torvalds1da177e2005-04-16 15:20:36 -0700712#ifdef TUN_DEBUG
713 case TUNSETDEBUG:
714 tun->debug = arg;
715 break;
716#endif
717
718 case SIOCGIFFLAGS:
719 ifr.ifr_flags = tun->if_flags;
720 if (copy_to_user( argp, &ifr, sizeof ifr))
721 return -EFAULT;
722 return 0;
723
724 case SIOCSIFFLAGS:
725 /** Set the character device's interface flags. Currently only
726 * IFF_PROMISC and IFF_ALLMULTI are used. */
727 tun->if_flags = ifr.ifr_flags;
728 DBG(KERN_INFO "%s: interface flags 0x%lx\n",
729 tun->dev->name, tun->if_flags);
730 return 0;
731
732 case SIOCGIFHWADDR:
Brian Braunstein36226a82007-04-26 01:00:55 -0700733 /* Note: the actual net device's address may be different */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700734 memcpy(ifr.ifr_hwaddr.sa_data, tun->dev_addr,
735 min(sizeof ifr.ifr_hwaddr.sa_data, sizeof tun->dev_addr));
736 if (copy_to_user( argp, &ifr, sizeof ifr))
737 return -EFAULT;
738 return 0;
739
740 case SIOCSIFHWADDR:
Brian Braunstein36226a82007-04-26 01:00:55 -0700741 {
742 /* try to set the actual net device's hw address */
Kim B. Heino40102372008-02-29 12:26:21 -0800743 int ret;
744
745 rtnl_lock();
746 ret = dev_set_mac_address(tun->dev, &ifr.ifr_hwaddr);
747 rtnl_unlock();
Brian Braunstein36226a82007-04-26 01:00:55 -0700748
749 if (ret == 0) {
750 /** Set the character device's hardware address. This is used when
751 * filtering packets being sent from the network device to the character
752 * device. */
753 memcpy(tun->dev_addr, ifr.ifr_hwaddr.sa_data,
754 min(sizeof ifr.ifr_hwaddr.sa_data, sizeof tun->dev_addr));
755 DBG(KERN_DEBUG "%s: set hardware address: %x:%x:%x:%x:%x:%x\n",
756 tun->dev->name,
757 tun->dev_addr[0], tun->dev_addr[1], tun->dev_addr[2],
758 tun->dev_addr[3], tun->dev_addr[4], tun->dev_addr[5]);
759 }
760
761 return ret;
762 }
Linus Torvalds1da177e2005-04-16 15:20:36 -0700763
764 case SIOCADDMULTI:
765 /** Add the specified group to the character device's multicast filter
766 * list. */
David S. Miller9edb74c2008-04-24 03:44:43 -0700767 rtnl_lock();
768 netif_tx_lock_bh(tun->dev);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700769 add_multi(tun->chr_filter, ifr.ifr_hwaddr.sa_data);
David S. Miller9edb74c2008-04-24 03:44:43 -0700770 netif_tx_unlock_bh(tun->dev);
771 rtnl_unlock();
772
Joe Perches0795af52007-10-03 17:59:30 -0700773 DBG(KERN_DEBUG "%s: add multi: %s\n",
774 tun->dev->name, print_mac(mac, ifr.ifr_hwaddr.sa_data));
Linus Torvalds1da177e2005-04-16 15:20:36 -0700775 return 0;
776
777 case SIOCDELMULTI:
778 /** Remove the specified group from the character device's multicast
779 * filter list. */
David S. Miller9edb74c2008-04-24 03:44:43 -0700780 rtnl_lock();
781 netif_tx_lock_bh(tun->dev);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700782 del_multi(tun->chr_filter, ifr.ifr_hwaddr.sa_data);
David S. Miller9edb74c2008-04-24 03:44:43 -0700783 netif_tx_unlock_bh(tun->dev);
784 rtnl_unlock();
785
Joe Perches0795af52007-10-03 17:59:30 -0700786 DBG(KERN_DEBUG "%s: del multi: %s\n",
787 tun->dev->name, print_mac(mac, ifr.ifr_hwaddr.sa_data));
Linus Torvalds1da177e2005-04-16 15:20:36 -0700788 return 0;
789
790 default:
791 return -EINVAL;
792 };
793
794 return 0;
795}
796
797static int tun_chr_fasync(int fd, struct file *file, int on)
798{
799 struct tun_struct *tun = file->private_data;
800 int ret;
801
802 if (!tun)
803 return -EBADFD;
804
805 DBG(KERN_INFO "%s: tun_chr_fasync %d\n", tun->dev->name, on);
806
807 if ((ret = fasync_helper(fd, file, on, &tun->fasync)) < 0)
Jeff Garzik6aa20a22006-09-13 13:24:59 -0400808 return ret;
809
Linus Torvalds1da177e2005-04-16 15:20:36 -0700810 if (on) {
Eric W. Biederman609d7fa2006-10-02 02:17:15 -0700811 ret = __f_setown(file, task_pid(current), PIDTYPE_PID, 0);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700812 if (ret)
813 return ret;
814 tun->flags |= TUN_FASYNC;
Jeff Garzik6aa20a22006-09-13 13:24:59 -0400815 } else
Linus Torvalds1da177e2005-04-16 15:20:36 -0700816 tun->flags &= ~TUN_FASYNC;
817
818 return 0;
819}
820
821static int tun_chr_open(struct inode *inode, struct file * file)
822{
823 DBG1(KERN_INFO "tunX: tun_chr_open\n");
824 file->private_data = NULL;
825 return 0;
826}
827
828static int tun_chr_close(struct inode *inode, struct file *file)
829{
830 struct tun_struct *tun = file->private_data;
831
832 if (!tun)
833 return 0;
834
835 DBG(KERN_INFO "%s: tun_chr_close\n", tun->dev->name);
836
837 tun_chr_fasync(-1, file, 0);
838
839 rtnl_lock();
840
841 /* Detach from net device */
842 file->private_data = NULL;
843 tun->attached = 0;
Pavel Emelyanovfc54c652008-04-16 00:41:53 -0700844 put_net(dev_net(tun->dev));
Linus Torvalds1da177e2005-04-16 15:20:36 -0700845
846 /* Drop read queue */
847 skb_queue_purge(&tun->readq);
848
849 if (!(tun->flags & TUN_PERSIST)) {
850 list_del(&tun->list);
851 unregister_netdevice(tun->dev);
852 }
853
854 rtnl_unlock();
855
856 return 0;
857}
858
Arjan van de Vend54b1fd2007-02-12 00:55:34 -0800859static const struct file_operations tun_fops = {
Jeff Garzik6aa20a22006-09-13 13:24:59 -0400860 .owner = THIS_MODULE,
Linus Torvalds1da177e2005-04-16 15:20:36 -0700861 .llseek = no_llseek,
Badari Pulavartyee0b3e62006-09-30 23:28:47 -0700862 .read = do_sync_read,
863 .aio_read = tun_chr_aio_read,
864 .write = do_sync_write,
865 .aio_write = tun_chr_aio_write,
Linus Torvalds1da177e2005-04-16 15:20:36 -0700866 .poll = tun_chr_poll,
867 .ioctl = tun_chr_ioctl,
868 .open = tun_chr_open,
869 .release = tun_chr_close,
Jeff Garzik6aa20a22006-09-13 13:24:59 -0400870 .fasync = tun_chr_fasync
Linus Torvalds1da177e2005-04-16 15:20:36 -0700871};
872
873static struct miscdevice tun_miscdev = {
874 .minor = TUN_MINOR,
875 .name = "tun",
876 .fops = &tun_fops,
Linus Torvalds1da177e2005-04-16 15:20:36 -0700877};
878
879/* ethtool interface */
880
881static int tun_get_settings(struct net_device *dev, struct ethtool_cmd *cmd)
882{
883 cmd->supported = 0;
884 cmd->advertising = 0;
885 cmd->speed = SPEED_10;
886 cmd->duplex = DUPLEX_FULL;
887 cmd->port = PORT_TP;
888 cmd->phy_address = 0;
889 cmd->transceiver = XCVR_INTERNAL;
890 cmd->autoneg = AUTONEG_DISABLE;
891 cmd->maxtxpkt = 0;
892 cmd->maxrxpkt = 0;
893 return 0;
894}
895
896static void tun_get_drvinfo(struct net_device *dev, struct ethtool_drvinfo *info)
897{
898 struct tun_struct *tun = netdev_priv(dev);
899
900 strcpy(info->driver, DRV_NAME);
901 strcpy(info->version, DRV_VERSION);
902 strcpy(info->fw_version, "N/A");
903
904 switch (tun->flags & TUN_TYPE_MASK) {
905 case TUN_TUN_DEV:
906 strcpy(info->bus_info, "tun");
907 break;
908 case TUN_TAP_DEV:
909 strcpy(info->bus_info, "tap");
910 break;
911 }
912}
913
914static u32 tun_get_msglevel(struct net_device *dev)
915{
916#ifdef TUN_DEBUG
917 struct tun_struct *tun = netdev_priv(dev);
918 return tun->debug;
919#else
920 return -EOPNOTSUPP;
921#endif
922}
923
924static void tun_set_msglevel(struct net_device *dev, u32 value)
925{
926#ifdef TUN_DEBUG
927 struct tun_struct *tun = netdev_priv(dev);
928 tun->debug = value;
929#endif
930}
931
932static u32 tun_get_link(struct net_device *dev)
933{
934 struct tun_struct *tun = netdev_priv(dev);
935 return tun->attached;
936}
937
938static u32 tun_get_rx_csum(struct net_device *dev)
939{
940 struct tun_struct *tun = netdev_priv(dev);
941 return (tun->flags & TUN_NOCHECKSUM) == 0;
942}
943
944static int tun_set_rx_csum(struct net_device *dev, u32 data)
945{
946 struct tun_struct *tun = netdev_priv(dev);
947 if (data)
948 tun->flags &= ~TUN_NOCHECKSUM;
949 else
950 tun->flags |= TUN_NOCHECKSUM;
951 return 0;
952}
953
Jeff Garzik7282d492006-09-13 14:30:00 -0400954static const struct ethtool_ops tun_ethtool_ops = {
Linus Torvalds1da177e2005-04-16 15:20:36 -0700955 .get_settings = tun_get_settings,
956 .get_drvinfo = tun_get_drvinfo,
957 .get_msglevel = tun_get_msglevel,
958 .set_msglevel = tun_set_msglevel,
959 .get_link = tun_get_link,
960 .get_rx_csum = tun_get_rx_csum,
961 .set_rx_csum = tun_set_rx_csum
962};
963
Pavel Emelyanov79d17602008-04-16 00:40:46 -0700964static int tun_init_net(struct net *net)
965{
966 struct tun_net *tn;
967
968 tn = kmalloc(sizeof(*tn), GFP_KERNEL);
969 if (tn == NULL)
970 return -ENOMEM;
971
972 INIT_LIST_HEAD(&tn->dev_list);
973
974 if (net_assign_generic(net, tun_net_id, tn)) {
975 kfree(tn);
976 return -ENOMEM;
977 }
978
979 return 0;
980}
981
982static void tun_exit_net(struct net *net)
983{
984 struct tun_net *tn;
Pavel Emelyanovd647a592008-04-16 00:41:16 -0700985 struct tun_struct *tun, *nxt;
Pavel Emelyanov79d17602008-04-16 00:40:46 -0700986
987 tn = net_generic(net, tun_net_id);
Pavel Emelyanovd647a592008-04-16 00:41:16 -0700988
989 rtnl_lock();
990 list_for_each_entry_safe(tun, nxt, &tn->dev_list, list) {
991 DBG(KERN_INFO "%s cleaned up\n", tun->dev->name);
992 unregister_netdevice(tun->dev);
993 }
994 rtnl_unlock();
995
Pavel Emelyanov79d17602008-04-16 00:40:46 -0700996 kfree(tn);
997}
998
999static struct pernet_operations tun_net_ops = {
1000 .init = tun_init_net,
1001 .exit = tun_exit_net,
1002};
1003
Linus Torvalds1da177e2005-04-16 15:20:36 -07001004static int __init tun_init(void)
1005{
1006 int ret = 0;
1007
1008 printk(KERN_INFO "tun: %s, %s\n", DRV_DESCRIPTION, DRV_VERSION);
1009 printk(KERN_INFO "tun: %s\n", DRV_COPYRIGHT);
1010
Pavel Emelyanov79d17602008-04-16 00:40:46 -07001011 ret = register_pernet_gen_device(&tun_net_id, &tun_net_ops);
1012 if (ret) {
1013 printk(KERN_ERR "tun: Can't register pernet ops\n");
1014 goto err_pernet;
1015 }
1016
Linus Torvalds1da177e2005-04-16 15:20:36 -07001017 ret = misc_register(&tun_miscdev);
Pavel Emelyanov79d17602008-04-16 00:40:46 -07001018 if (ret) {
Linus Torvalds1da177e2005-04-16 15:20:36 -07001019 printk(KERN_ERR "tun: Can't register misc device %d\n", TUN_MINOR);
Pavel Emelyanov79d17602008-04-16 00:40:46 -07001020 goto err_misc;
1021 }
1022 return 0;
1023
1024err_misc:
1025 unregister_pernet_gen_device(tun_net_id, &tun_net_ops);
1026err_pernet:
Linus Torvalds1da177e2005-04-16 15:20:36 -07001027 return ret;
1028}
1029
1030static void tun_cleanup(void)
1031{
Jeff Garzik6aa20a22006-09-13 13:24:59 -04001032 misc_deregister(&tun_miscdev);
Pavel Emelyanov79d17602008-04-16 00:40:46 -07001033 unregister_pernet_gen_device(tun_net_id, &tun_net_ops);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001034}
1035
1036module_init(tun_init);
1037module_exit(tun_cleanup);
1038MODULE_DESCRIPTION(DRV_DESCRIPTION);
1039MODULE_AUTHOR(DRV_COPYRIGHT);
1040MODULE_LICENSE("GPL");
1041MODULE_ALIAS_MISCDEV(TUN_MINOR);