2 * TUN - Universal TUN/TAP device driver.
3 * Copyright (C) 1999-2002 Maxim Krasnyansky <maxk@qualcomm.com>
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.
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.
15 * $Id: tun.c,v 1.15 2002/03/01 02:44:24 maxk Exp $
21 * 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.
25 * Mike Kershaw <dragorn@kismetwireless.net> 2005/08/14
26 * Add TUNSETLINK ioctl to set the link encapsulation
28 * Mark Smith <markzzzsmith@yahoo.com.au>
29 * Use random_ether_addr() for tap MAC address.
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.
37 * Daniel Podlejski <underley@underley.eu.org>
38 * Modifications for 2.3.99-pre5 kernel.
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>"
46 #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>
61 #include <linux/if_arp.h>
62 #include <linux/if_ether.h>
63 #include <linux/if_tun.h>
64 #include <linux/crc32.h>
66 #include <asm/system.h>
67 #include <asm/uaccess.h>
73 /* Network device part of the driver */
75 static LIST_HEAD(tun_dev_list);
76 static const struct ethtool_ops tun_ethtool_ops;
78 /* Net device open. */
79 static int tun_net_open(struct net_device *dev)
81 netif_start_queue(dev);
85 /* Net device close. */
86 static int tun_net_close(struct net_device *dev)
88 netif_stop_queue(dev);
92 /* Net device start xmit */
93 static int tun_net_xmit(struct sk_buff *skb, struct net_device *dev)
95 struct tun_struct *tun = netdev_priv(dev);
97 DBG(KERN_INFO "%s: tun_net_xmit %d\n", tun->dev->name, skb->len);
99 /* Drop packet if interface is not attached */
103 /* Packet dropping */
104 if (skb_queue_len(&tun->readq) >= dev->tx_queue_len) {
105 if (!(tun->flags & TUN_ONE_QUEUE)) {
106 /* Normal queueing mode. */
107 /* Packet scheduler handles dropping of further packets. */
108 netif_stop_queue(dev);
110 /* We won't see all dropped packets individually, so overrun
111 * error is more appropriate. */
112 tun->stats.tx_fifo_errors++;
114 /* Single queue mode.
115 * Driver handles dropping of all packets itself. */
121 skb_queue_tail(&tun->readq, skb);
122 dev->trans_start = jiffies;
124 /* Notify and wake up reader process */
125 if (tun->flags & TUN_FASYNC)
126 kill_fasync(&tun->fasync, SIGIO, POLL_IN);
127 wake_up_interruptible(&tun->read_wait);
131 tun->stats.tx_dropped++;
136 /** Add the specified Ethernet address to this multicast filter. */
138 add_multi(u32* filter, const u8* addr)
140 int bit_nr = ether_crc(ETH_ALEN, addr) >> 26;
141 filter[bit_nr >> 5] |= 1 << (bit_nr & 31);
144 /** Remove the specified Ethernet addres from this multicast filter. */
146 del_multi(u32* filter, const u8* addr)
148 int bit_nr = ether_crc(ETH_ALEN, addr) >> 26;
149 filter[bit_nr >> 5] &= ~(1 << (bit_nr & 31));
152 /** Update the list of multicast groups to which the network device belongs.
153 * This list is used to filter packets being sent from the character device to
154 * the network device. */
156 tun_net_mclist(struct net_device *dev)
158 struct tun_struct *tun = netdev_priv(dev);
159 const struct dev_mc_list *mclist;
161 DBG(KERN_DEBUG "%s: tun_net_mclist: mc_count %d\n",
162 dev->name, dev->mc_count);
163 memset(tun->chr_filter, 0, sizeof tun->chr_filter);
164 for (i = 0, mclist = dev->mc_list; i < dev->mc_count && mclist != NULL;
165 i++, mclist = mclist->next) {
166 add_multi(tun->net_filter, mclist->dmi_addr);
167 DBG(KERN_DEBUG "%s: tun_net_mclist: %x:%x:%x:%x:%x:%x\n",
169 mclist->dmi_addr[0], mclist->dmi_addr[1], mclist->dmi_addr[2],
170 mclist->dmi_addr[3], mclist->dmi_addr[4], mclist->dmi_addr[5]);
174 static struct net_device_stats *tun_net_stats(struct net_device *dev)
176 struct tun_struct *tun = netdev_priv(dev);
180 /* Initialize net device. */
181 static void tun_net_init(struct net_device *dev)
183 struct tun_struct *tun = netdev_priv(dev);
185 switch (tun->flags & TUN_TYPE_MASK) {
187 /* Point-to-Point TUN Device */
188 dev->hard_header_len = 0;
192 /* Zero header length */
193 dev->type = ARPHRD_NONE;
194 dev->flags = IFF_POINTOPOINT | IFF_NOARP | IFF_MULTICAST;
195 dev->tx_queue_len = TUN_READQ_SIZE; /* We prefer our own queue length */
199 /* Ethernet TAP Device */
200 dev->set_multicast_list = tun_net_mclist;
204 /* random address already created for us by tun_set_iff, use it */
205 memcpy(dev->dev_addr, tun->dev_addr, min(sizeof(tun->dev_addr), sizeof(dev->dev_addr)) );
207 dev->tx_queue_len = TUN_READQ_SIZE; /* We prefer our own queue length */
212 /* Character device part */
215 static unsigned int tun_chr_poll(struct file *file, poll_table * wait)
217 struct tun_struct *tun = file->private_data;
218 unsigned int mask = POLLOUT | POLLWRNORM;
223 DBG(KERN_INFO "%s: tun_chr_poll\n", tun->dev->name);
225 poll_wait(file, &tun->read_wait, wait);
227 if (!skb_queue_empty(&tun->readq))
228 mask |= POLLIN | POLLRDNORM;
233 /* Get packet from user space buffer */
234 static __inline__ ssize_t tun_get_user(struct tun_struct *tun, struct iovec *iv, size_t count)
236 struct tun_pi pi = { 0, __constant_htons(ETH_P_IP) };
238 size_t len = count, align = 0;
240 if (!(tun->flags & TUN_NO_PI)) {
241 if ((len -= sizeof(pi)) > count)
244 if(memcpy_fromiovec((void *)&pi, iv, sizeof(pi)))
248 if ((tun->flags & TUN_TYPE_MASK) == TUN_TAP_DEV)
249 align = NET_IP_ALIGN;
251 if (!(skb = alloc_skb(len + align, GFP_KERNEL))) {
252 tun->stats.rx_dropped++;
257 skb_reserve(skb, align);
258 if (memcpy_fromiovec(skb_put(skb, len), iv, len)) {
259 tun->stats.rx_dropped++;
264 switch (tun->flags & TUN_TYPE_MASK) {
266 skb_reset_mac_header(skb);
267 skb->protocol = pi.proto;
271 skb->protocol = eth_type_trans(skb, tun->dev);
275 if (tun->flags & TUN_NOCHECKSUM)
276 skb->ip_summed = CHECKSUM_UNNECESSARY;
279 tun->dev->last_rx = jiffies;
281 tun->stats.rx_packets++;
282 tun->stats.rx_bytes += len;
287 static inline size_t iov_total(const struct iovec *iv, unsigned long count)
292 for (i = 0, len = 0; i < count; i++)
293 len += iv[i].iov_len;
298 static ssize_t tun_chr_aio_write(struct kiocb *iocb, const struct iovec *iv,
299 unsigned long count, loff_t pos)
301 struct tun_struct *tun = iocb->ki_filp->private_data;
306 DBG(KERN_INFO "%s: tun_chr_write %ld\n", tun->dev->name, count);
308 return tun_get_user(tun, (struct iovec *) iv, iov_total(iv, count));
311 /* Put packet to the user space buffer */
312 static __inline__ ssize_t tun_put_user(struct tun_struct *tun,
314 struct iovec *iv, int len)
316 struct tun_pi pi = { 0, skb->protocol };
319 if (!(tun->flags & TUN_NO_PI)) {
320 if ((len -= sizeof(pi)) < 0)
323 if (len < skb->len) {
324 /* Packet will be striped */
325 pi.flags |= TUN_PKT_STRIP;
328 if (memcpy_toiovec(iv, (void *) &pi, sizeof(pi)))
333 len = min_t(int, skb->len, len);
335 skb_copy_datagram_iovec(skb, 0, iv, len);
338 tun->stats.tx_packets++;
339 tun->stats.tx_bytes += len;
344 static ssize_t tun_chr_aio_read(struct kiocb *iocb, const struct iovec *iv,
345 unsigned long count, loff_t pos)
347 struct file *file = iocb->ki_filp;
348 struct tun_struct *tun = file->private_data;
349 DECLARE_WAITQUEUE(wait, current);
351 ssize_t len, ret = 0;
356 DBG(KERN_INFO "%s: tun_chr_read\n", tun->dev->name);
358 len = iov_total(iv, count);
362 add_wait_queue(&tun->read_wait, &wait);
364 const u8 ones[ ETH_ALEN] = { 0xff, 0xff, 0xff, 0xff, 0xff, 0xff };
368 current->state = TASK_INTERRUPTIBLE;
370 /* Read frames from the queue */
371 if (!(skb=skb_dequeue(&tun->readq))) {
372 if (file->f_flags & O_NONBLOCK) {
376 if (signal_pending(current)) {
381 /* Nothing to read, let's sleep */
385 netif_wake_queue(tun->dev);
387 /** Decide whether to accept this packet. This code is designed to
388 * behave identically to an Ethernet interface. Accept the packet if
389 * - we are promiscuous.
390 * - the packet is addressed to us.
391 * - the packet is broadcast.
392 * - the packet is multicast and
393 * - we are multicast promiscous.
394 * - we belong to the multicast group.
396 skb_copy_from_linear_data(skb, addr, min_t(size_t, sizeof addr,
398 bit_nr = ether_crc(sizeof addr, addr) >> 26;
399 if ((tun->if_flags & IFF_PROMISC) ||
400 memcmp(addr, tun->dev_addr, sizeof addr) == 0 ||
401 memcmp(addr, ones, sizeof addr) == 0 ||
402 (((addr[0] == 1 && addr[1] == 0 && addr[2] == 0x5e) ||
403 (addr[0] == 0x33 && addr[1] == 0x33)) &&
404 ((tun->if_flags & IFF_ALLMULTI) ||
405 (tun->chr_filter[bit_nr >> 5] & (1 << (bit_nr & 31)))))) {
406 DBG(KERN_DEBUG "%s: tun_chr_readv: accepted: %x:%x:%x:%x:%x:%x\n",
407 tun->dev->name, addr[0], addr[1], addr[2],
408 addr[3], addr[4], addr[5]);
409 ret = tun_put_user(tun, skb, (struct iovec *) iv, len);
413 DBG(KERN_DEBUG "%s: tun_chr_readv: rejected: %x:%x:%x:%x:%x:%x\n",
414 tun->dev->name, addr[0], addr[1], addr[2],
415 addr[3], addr[4], addr[5]);
421 current->state = TASK_RUNNING;
422 remove_wait_queue(&tun->read_wait, &wait);
427 static void tun_setup(struct net_device *dev)
429 struct tun_struct *tun = netdev_priv(dev);
431 skb_queue_head_init(&tun->readq);
432 init_waitqueue_head(&tun->read_wait);
436 SET_MODULE_OWNER(dev);
437 dev->open = tun_net_open;
438 dev->hard_start_xmit = tun_net_xmit;
439 dev->stop = tun_net_close;
440 dev->get_stats = tun_net_stats;
441 dev->ethtool_ops = &tun_ethtool_ops;
442 dev->destructor = free_netdev;
445 static struct tun_struct *tun_get_by_name(const char *name)
447 struct tun_struct *tun;
450 list_for_each_entry(tun, &tun_dev_list, list) {
451 if (!strncmp(tun->dev->name, name, IFNAMSIZ))
458 static int tun_set_iff(struct file *file, struct ifreq *ifr)
460 struct tun_struct *tun;
461 struct net_device *dev;
464 tun = tun_get_by_name(ifr->ifr_name);
469 /* Check permissions */
470 if (tun->owner != -1 &&
471 current->euid != tun->owner && !capable(CAP_NET_ADMIN))
474 else if (__dev_get_by_name(ifr->ifr_name))
478 unsigned long flags = 0;
482 if (!capable(CAP_NET_ADMIN))
486 if (ifr->ifr_flags & IFF_TUN) {
488 flags |= TUN_TUN_DEV;
490 } else if (ifr->ifr_flags & IFF_TAP) {
492 flags |= TUN_TAP_DEV;
498 name = ifr->ifr_name;
500 dev = alloc_netdev(sizeof(struct tun_struct), name,
505 tun = netdev_priv(dev);
508 /* Be promiscuous by default to maintain previous behaviour. */
509 tun->if_flags = IFF_PROMISC;
510 /* Generate random Ethernet address. */
511 *(u16 *)tun->dev_addr = htons(0x00FF);
512 get_random_bytes(tun->dev_addr + sizeof(u16), 4);
513 memset(tun->chr_filter, 0, sizeof tun->chr_filter);
517 if (strchr(dev->name, '%')) {
518 err = dev_alloc_name(dev, dev->name);
523 err = register_netdevice(tun->dev);
527 list_add(&tun->list, &tun_dev_list);
530 DBG(KERN_INFO "%s: tun_set_iff\n", tun->dev->name);
532 if (ifr->ifr_flags & IFF_NO_PI)
533 tun->flags |= TUN_NO_PI;
535 if (ifr->ifr_flags & IFF_ONE_QUEUE)
536 tun->flags |= TUN_ONE_QUEUE;
538 file->private_data = tun;
541 strcpy(ifr->ifr_name, tun->dev->name);
550 static int tun_chr_ioctl(struct inode *inode, struct file *file,
551 unsigned int cmd, unsigned long arg)
553 struct tun_struct *tun = file->private_data;
554 void __user* argp = (void __user*)arg;
557 if (cmd == TUNSETIFF || _IOC_TYPE(cmd) == 0x89)
558 if (copy_from_user(&ifr, argp, sizeof ifr))
561 if (cmd == TUNSETIFF && !tun) {
564 ifr.ifr_name[IFNAMSIZ-1] = '\0';
567 err = tun_set_iff(file, &ifr);
573 if (copy_to_user(argp, &ifr, sizeof(ifr)))
581 DBG(KERN_INFO "%s: tun_chr_ioctl cmd %d\n", tun->dev->name, cmd);
585 /* Disable/Enable checksum */
587 tun->flags |= TUN_NOCHECKSUM;
589 tun->flags &= ~TUN_NOCHECKSUM;
591 DBG(KERN_INFO "%s: checksum %s\n",
592 tun->dev->name, arg ? "disabled" : "enabled");
596 /* Disable/Enable persist mode */
598 tun->flags |= TUN_PERSIST;
600 tun->flags &= ~TUN_PERSIST;
602 DBG(KERN_INFO "%s: persist %s\n",
603 tun->dev->name, arg ? "disabled" : "enabled");
607 /* Set owner of the device */
608 tun->owner = (uid_t) arg;
610 DBG(KERN_INFO "%s: owner set to %d\n", tun->dev->name, tun->owner);
614 /* Only allow setting the type when the interface is down */
615 if (tun->dev->flags & IFF_UP) {
616 DBG(KERN_INFO "%s: Linktype set failed because interface is up\n",
620 tun->dev->type = (int) arg;
621 DBG(KERN_INFO "%s: linktype set to %d\n", tun->dev->name, tun->dev->type);
632 ifr.ifr_flags = tun->if_flags;
633 if (copy_to_user( argp, &ifr, sizeof ifr))
638 /** Set the character device's interface flags. Currently only
639 * IFF_PROMISC and IFF_ALLMULTI are used. */
640 tun->if_flags = ifr.ifr_flags;
641 DBG(KERN_INFO "%s: interface flags 0x%lx\n",
642 tun->dev->name, tun->if_flags);
646 /* Note: the actual net device's address may be different */
647 memcpy(ifr.ifr_hwaddr.sa_data, tun->dev_addr,
648 min(sizeof ifr.ifr_hwaddr.sa_data, sizeof tun->dev_addr));
649 if (copy_to_user( argp, &ifr, sizeof ifr))
655 /* try to set the actual net device's hw address */
656 int ret = dev_set_mac_address(tun->dev, &ifr.ifr_hwaddr);
659 /** Set the character device's hardware address. This is used when
660 * filtering packets being sent from the network device to the character
662 memcpy(tun->dev_addr, ifr.ifr_hwaddr.sa_data,
663 min(sizeof ifr.ifr_hwaddr.sa_data, sizeof tun->dev_addr));
664 DBG(KERN_DEBUG "%s: set hardware address: %x:%x:%x:%x:%x:%x\n",
666 tun->dev_addr[0], tun->dev_addr[1], tun->dev_addr[2],
667 tun->dev_addr[3], tun->dev_addr[4], tun->dev_addr[5]);
674 /** Add the specified group to the character device's multicast filter
676 add_multi(tun->chr_filter, ifr.ifr_hwaddr.sa_data);
677 DBG(KERN_DEBUG "%s: add multi: %x:%x:%x:%x:%x:%x\n",
679 (u8)ifr.ifr_hwaddr.sa_data[0], (u8)ifr.ifr_hwaddr.sa_data[1],
680 (u8)ifr.ifr_hwaddr.sa_data[2], (u8)ifr.ifr_hwaddr.sa_data[3],
681 (u8)ifr.ifr_hwaddr.sa_data[4], (u8)ifr.ifr_hwaddr.sa_data[5]);
685 /** Remove the specified group from the character device's multicast
687 del_multi(tun->chr_filter, ifr.ifr_hwaddr.sa_data);
688 DBG(KERN_DEBUG "%s: del multi: %x:%x:%x:%x:%x:%x\n",
690 (u8)ifr.ifr_hwaddr.sa_data[0], (u8)ifr.ifr_hwaddr.sa_data[1],
691 (u8)ifr.ifr_hwaddr.sa_data[2], (u8)ifr.ifr_hwaddr.sa_data[3],
692 (u8)ifr.ifr_hwaddr.sa_data[4], (u8)ifr.ifr_hwaddr.sa_data[5]);
702 static int tun_chr_fasync(int fd, struct file *file, int on)
704 struct tun_struct *tun = file->private_data;
710 DBG(KERN_INFO "%s: tun_chr_fasync %d\n", tun->dev->name, on);
712 if ((ret = fasync_helper(fd, file, on, &tun->fasync)) < 0)
716 ret = __f_setown(file, task_pid(current), PIDTYPE_PID, 0);
719 tun->flags |= TUN_FASYNC;
721 tun->flags &= ~TUN_FASYNC;
726 static int tun_chr_open(struct inode *inode, struct file * file)
728 DBG1(KERN_INFO "tunX: tun_chr_open\n");
729 file->private_data = NULL;
733 static int tun_chr_close(struct inode *inode, struct file *file)
735 struct tun_struct *tun = file->private_data;
740 DBG(KERN_INFO "%s: tun_chr_close\n", tun->dev->name);
742 tun_chr_fasync(-1, file, 0);
746 /* Detach from net device */
747 file->private_data = NULL;
750 /* Drop read queue */
751 skb_queue_purge(&tun->readq);
753 if (!(tun->flags & TUN_PERSIST)) {
754 list_del(&tun->list);
755 unregister_netdevice(tun->dev);
763 static const struct file_operations tun_fops = {
764 .owner = THIS_MODULE,
766 .read = do_sync_read,
767 .aio_read = tun_chr_aio_read,
768 .write = do_sync_write,
769 .aio_write = tun_chr_aio_write,
770 .poll = tun_chr_poll,
771 .ioctl = tun_chr_ioctl,
772 .open = tun_chr_open,
773 .release = tun_chr_close,
774 .fasync = tun_chr_fasync
777 static struct miscdevice tun_miscdev = {
783 /* ethtool interface */
785 static int tun_get_settings(struct net_device *dev, struct ethtool_cmd *cmd)
788 cmd->advertising = 0;
789 cmd->speed = SPEED_10;
790 cmd->duplex = DUPLEX_FULL;
792 cmd->phy_address = 0;
793 cmd->transceiver = XCVR_INTERNAL;
794 cmd->autoneg = AUTONEG_DISABLE;
800 static void tun_get_drvinfo(struct net_device *dev, struct ethtool_drvinfo *info)
802 struct tun_struct *tun = netdev_priv(dev);
804 strcpy(info->driver, DRV_NAME);
805 strcpy(info->version, DRV_VERSION);
806 strcpy(info->fw_version, "N/A");
808 switch (tun->flags & TUN_TYPE_MASK) {
810 strcpy(info->bus_info, "tun");
813 strcpy(info->bus_info, "tap");
818 static u32 tun_get_msglevel(struct net_device *dev)
821 struct tun_struct *tun = netdev_priv(dev);
828 static void tun_set_msglevel(struct net_device *dev, u32 value)
831 struct tun_struct *tun = netdev_priv(dev);
836 static u32 tun_get_link(struct net_device *dev)
838 struct tun_struct *tun = netdev_priv(dev);
839 return tun->attached;
842 static u32 tun_get_rx_csum(struct net_device *dev)
844 struct tun_struct *tun = netdev_priv(dev);
845 return (tun->flags & TUN_NOCHECKSUM) == 0;
848 static int tun_set_rx_csum(struct net_device *dev, u32 data)
850 struct tun_struct *tun = netdev_priv(dev);
852 tun->flags &= ~TUN_NOCHECKSUM;
854 tun->flags |= TUN_NOCHECKSUM;
858 static const struct ethtool_ops tun_ethtool_ops = {
859 .get_settings = tun_get_settings,
860 .get_drvinfo = tun_get_drvinfo,
861 .get_msglevel = tun_get_msglevel,
862 .set_msglevel = tun_set_msglevel,
863 .get_link = tun_get_link,
864 .get_rx_csum = tun_get_rx_csum,
865 .set_rx_csum = tun_set_rx_csum
868 static int __init tun_init(void)
872 printk(KERN_INFO "tun: %s, %s\n", DRV_DESCRIPTION, DRV_VERSION);
873 printk(KERN_INFO "tun: %s\n", DRV_COPYRIGHT);
875 ret = misc_register(&tun_miscdev);
877 printk(KERN_ERR "tun: Can't register misc device %d\n", TUN_MINOR);
881 static void tun_cleanup(void)
883 struct tun_struct *tun, *nxt;
885 misc_deregister(&tun_miscdev);
888 list_for_each_entry_safe(tun, nxt, &tun_dev_list, list) {
889 DBG(KERN_INFO "%s cleaned up\n", tun->dev->name);
890 unregister_netdevice(tun->dev);
896 module_init(tun_init);
897 module_exit(tun_cleanup);
898 MODULE_DESCRIPTION(DRV_DESCRIPTION);
899 MODULE_AUTHOR(DRV_COPYRIGHT);
900 MODULE_LICENSE("GPL");
901 MODULE_ALIAS_MISCDEV(TUN_MINOR);