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>
65 #include <net/net_namespace.h>
67 #include <asm/system.h>
68 #include <asm/uaccess.h>
74 /* Network device part of the driver */
76 static LIST_HEAD(tun_dev_list);
77 static const struct ethtool_ops tun_ethtool_ops;
79 /* Net device open. */
80 static int tun_net_open(struct net_device *dev)
82 netif_start_queue(dev);
86 /* Net device close. */
87 static int tun_net_close(struct net_device *dev)
89 netif_stop_queue(dev);
93 /* Net device start xmit */
94 static int tun_net_xmit(struct sk_buff *skb, struct net_device *dev)
96 struct tun_struct *tun = netdev_priv(dev);
98 DBG(KERN_INFO "%s: tun_net_xmit %d\n", tun->dev->name, skb->len);
100 /* Drop packet if interface is not attached */
104 /* Packet dropping */
105 if (skb_queue_len(&tun->readq) >= dev->tx_queue_len) {
106 if (!(tun->flags & TUN_ONE_QUEUE)) {
107 /* Normal queueing mode. */
108 /* Packet scheduler handles dropping of further packets. */
109 netif_stop_queue(dev);
111 /* We won't see all dropped packets individually, so overrun
112 * error is more appropriate. */
113 dev->stats.tx_fifo_errors++;
115 /* Single queue mode.
116 * Driver handles dropping of all packets itself. */
122 skb_queue_tail(&tun->readq, skb);
123 dev->trans_start = jiffies;
125 /* Notify and wake up reader process */
126 if (tun->flags & TUN_FASYNC)
127 kill_fasync(&tun->fasync, SIGIO, POLL_IN);
128 wake_up_interruptible(&tun->read_wait);
132 dev->stats.tx_dropped++;
137 /** Add the specified Ethernet address to this multicast filter. */
139 add_multi(u32* filter, const u8* addr)
141 int bit_nr = ether_crc(ETH_ALEN, addr) >> 26;
142 filter[bit_nr >> 5] |= 1 << (bit_nr & 31);
145 /** Remove the specified Ethernet addres from this multicast filter. */
147 del_multi(u32* filter, const u8* addr)
149 int bit_nr = ether_crc(ETH_ALEN, addr) >> 26;
150 filter[bit_nr >> 5] &= ~(1 << (bit_nr & 31));
153 /** Update the list of multicast groups to which the network device belongs.
154 * This list is used to filter packets being sent from the character device to
155 * the network device. */
157 tun_net_mclist(struct net_device *dev)
159 struct tun_struct *tun = netdev_priv(dev);
160 const struct dev_mc_list *mclist;
162 DECLARE_MAC_BUF(mac);
163 DBG(KERN_DEBUG "%s: tun_net_mclist: mc_count %d\n",
164 dev->name, dev->mc_count);
165 memset(tun->chr_filter, 0, sizeof tun->chr_filter);
166 for (i = 0, mclist = dev->mc_list; i < dev->mc_count && mclist != NULL;
167 i++, mclist = mclist->next) {
168 add_multi(tun->net_filter, mclist->dmi_addr);
169 DBG(KERN_DEBUG "%s: tun_net_mclist: %s\n",
170 dev->name, print_mac(mac, mclist->dmi_addr));
175 #define MAX_MTU 65535
178 tun_net_change_mtu(struct net_device *dev, int new_mtu)
180 if (new_mtu < MIN_MTU || new_mtu + dev->hard_header_len > MAX_MTU)
186 /* Initialize net device. */
187 static void tun_net_init(struct net_device *dev)
189 struct tun_struct *tun = netdev_priv(dev);
191 switch (tun->flags & TUN_TYPE_MASK) {
193 /* Point-to-Point TUN Device */
194 dev->hard_header_len = 0;
197 dev->change_mtu = tun_net_change_mtu;
199 /* Zero header length */
200 dev->type = ARPHRD_NONE;
201 dev->flags = IFF_POINTOPOINT | IFF_NOARP | IFF_MULTICAST;
202 dev->tx_queue_len = TUN_READQ_SIZE; /* We prefer our own queue length */
206 /* Ethernet TAP Device */
207 dev->set_multicast_list = tun_net_mclist;
210 dev->change_mtu = tun_net_change_mtu;
212 /* random address already created for us by tun_set_iff, use it */
213 memcpy(dev->dev_addr, tun->dev_addr, min(sizeof(tun->dev_addr), sizeof(dev->dev_addr)) );
215 dev->tx_queue_len = TUN_READQ_SIZE; /* We prefer our own queue length */
220 /* Character device part */
223 static unsigned int tun_chr_poll(struct file *file, poll_table * wait)
225 struct tun_struct *tun = file->private_data;
226 unsigned int mask = POLLOUT | POLLWRNORM;
231 DBG(KERN_INFO "%s: tun_chr_poll\n", tun->dev->name);
233 poll_wait(file, &tun->read_wait, wait);
235 if (!skb_queue_empty(&tun->readq))
236 mask |= POLLIN | POLLRDNORM;
241 /* Get packet from user space buffer */
242 static __inline__ ssize_t tun_get_user(struct tun_struct *tun, struct iovec *iv, size_t count)
244 struct tun_pi pi = { 0, __constant_htons(ETH_P_IP) };
246 size_t len = count, align = 0;
248 if (!(tun->flags & TUN_NO_PI)) {
249 if ((len -= sizeof(pi)) > count)
252 if(memcpy_fromiovec((void *)&pi, iv, sizeof(pi)))
256 if ((tun->flags & TUN_TYPE_MASK) == TUN_TAP_DEV)
257 align = NET_IP_ALIGN;
259 if (!(skb = alloc_skb(len + align, GFP_KERNEL))) {
260 tun->dev->stats.rx_dropped++;
265 skb_reserve(skb, align);
266 if (memcpy_fromiovec(skb_put(skb, len), iv, len)) {
267 tun->dev->stats.rx_dropped++;
272 switch (tun->flags & TUN_TYPE_MASK) {
274 skb_reset_mac_header(skb);
275 skb->protocol = pi.proto;
279 skb->protocol = eth_type_trans(skb, tun->dev);
283 if (tun->flags & TUN_NOCHECKSUM)
284 skb->ip_summed = CHECKSUM_UNNECESSARY;
287 tun->dev->last_rx = jiffies;
289 tun->dev->stats.rx_packets++;
290 tun->dev->stats.rx_bytes += len;
295 static ssize_t tun_chr_aio_write(struct kiocb *iocb, const struct iovec *iv,
296 unsigned long count, loff_t pos)
298 struct tun_struct *tun = iocb->ki_filp->private_data;
303 DBG(KERN_INFO "%s: tun_chr_write %ld\n", tun->dev->name, count);
305 return tun_get_user(tun, (struct iovec *) iv, iov_length(iv, count));
308 /* Put packet to the user space buffer */
309 static __inline__ ssize_t tun_put_user(struct tun_struct *tun,
311 struct iovec *iv, int len)
313 struct tun_pi pi = { 0, skb->protocol };
316 if (!(tun->flags & TUN_NO_PI)) {
317 if ((len -= sizeof(pi)) < 0)
320 if (len < skb->len) {
321 /* Packet will be striped */
322 pi.flags |= TUN_PKT_STRIP;
325 if (memcpy_toiovec(iv, (void *) &pi, sizeof(pi)))
330 len = min_t(int, skb->len, len);
332 skb_copy_datagram_iovec(skb, 0, iv, len);
335 tun->dev->stats.tx_packets++;
336 tun->dev->stats.tx_bytes += len;
341 static ssize_t tun_chr_aio_read(struct kiocb *iocb, const struct iovec *iv,
342 unsigned long count, loff_t pos)
344 struct file *file = iocb->ki_filp;
345 struct tun_struct *tun = file->private_data;
346 DECLARE_WAITQUEUE(wait, current);
348 ssize_t len, ret = 0;
349 DECLARE_MAC_BUF(mac);
354 DBG(KERN_INFO "%s: tun_chr_read\n", tun->dev->name);
356 len = iov_length(iv, count);
360 add_wait_queue(&tun->read_wait, &wait);
362 const u8 ones[ ETH_ALEN] = { 0xff, 0xff, 0xff, 0xff, 0xff, 0xff };
366 current->state = TASK_INTERRUPTIBLE;
368 /* Read frames from the queue */
369 if (!(skb=skb_dequeue(&tun->readq))) {
370 if (file->f_flags & O_NONBLOCK) {
374 if (signal_pending(current)) {
379 /* Nothing to read, let's sleep */
383 netif_wake_queue(tun->dev);
385 /** Decide whether to accept this packet. This code is designed to
386 * behave identically to an Ethernet interface. Accept the packet if
387 * - we are promiscuous.
388 * - the packet is addressed to us.
389 * - the packet is broadcast.
390 * - the packet is multicast and
391 * - we are multicast promiscous.
392 * - we belong to the multicast group.
394 skb_copy_from_linear_data(skb, addr, min_t(size_t, sizeof addr,
396 bit_nr = ether_crc(sizeof addr, addr) >> 26;
397 if ((tun->if_flags & IFF_PROMISC) ||
398 memcmp(addr, tun->dev_addr, sizeof addr) == 0 ||
399 memcmp(addr, ones, sizeof addr) == 0 ||
400 (((addr[0] == 1 && addr[1] == 0 && addr[2] == 0x5e) ||
401 (addr[0] == 0x33 && addr[1] == 0x33)) &&
402 ((tun->if_flags & IFF_ALLMULTI) ||
403 (tun->chr_filter[bit_nr >> 5] & (1 << (bit_nr & 31)))))) {
404 DBG(KERN_DEBUG "%s: tun_chr_readv: accepted: %s\n",
405 tun->dev->name, print_mac(mac, addr));
406 ret = tun_put_user(tun, skb, (struct iovec *) iv, len);
410 DBG(KERN_DEBUG "%s: tun_chr_readv: rejected: %s\n",
411 tun->dev->name, print_mac(mac, addr));
417 current->state = TASK_RUNNING;
418 remove_wait_queue(&tun->read_wait, &wait);
423 static void tun_setup(struct net_device *dev)
425 struct tun_struct *tun = netdev_priv(dev);
427 skb_queue_head_init(&tun->readq);
428 init_waitqueue_head(&tun->read_wait);
433 dev->open = tun_net_open;
434 dev->hard_start_xmit = tun_net_xmit;
435 dev->stop = tun_net_close;
436 dev->ethtool_ops = &tun_ethtool_ops;
437 dev->destructor = free_netdev;
440 static struct tun_struct *tun_get_by_name(const char *name)
442 struct tun_struct *tun;
445 list_for_each_entry(tun, &tun_dev_list, list) {
446 if (!strncmp(tun->dev->name, name, IFNAMSIZ))
453 static int tun_set_iff(struct file *file, struct ifreq *ifr)
455 struct tun_struct *tun;
456 struct net_device *dev;
459 tun = tun_get_by_name(ifr->ifr_name);
464 /* Check permissions */
465 if (((tun->owner != -1 &&
466 current->euid != tun->owner) ||
468 current->egid != tun->group)) &&
469 !capable(CAP_NET_ADMIN))
472 else if (__dev_get_by_name(&init_net, ifr->ifr_name))
476 unsigned long flags = 0;
480 if (!capable(CAP_NET_ADMIN))
484 if (ifr->ifr_flags & IFF_TUN) {
486 flags |= TUN_TUN_DEV;
488 } else if (ifr->ifr_flags & IFF_TAP) {
490 flags |= TUN_TAP_DEV;
496 name = ifr->ifr_name;
498 dev = alloc_netdev(sizeof(struct tun_struct), name,
503 tun = netdev_priv(dev);
506 /* Be promiscuous by default to maintain previous behaviour. */
507 tun->if_flags = IFF_PROMISC;
508 /* Generate random Ethernet address. */
509 *(__be16 *)tun->dev_addr = htons(0x00FF);
510 get_random_bytes(tun->dev_addr + sizeof(u16), 4);
511 memset(tun->chr_filter, 0, sizeof tun->chr_filter);
515 if (strchr(dev->name, '%')) {
516 err = dev_alloc_name(dev, dev->name);
521 err = register_netdevice(tun->dev);
525 list_add(&tun->list, &tun_dev_list);
528 DBG(KERN_INFO "%s: tun_set_iff\n", tun->dev->name);
530 if (ifr->ifr_flags & IFF_NO_PI)
531 tun->flags |= TUN_NO_PI;
533 tun->flags &= ~TUN_NO_PI;
535 if (ifr->ifr_flags & IFF_ONE_QUEUE)
536 tun->flags |= TUN_ONE_QUEUE;
538 tun->flags &= ~TUN_ONE_QUEUE;
540 file->private_data = tun;
543 strcpy(ifr->ifr_name, tun->dev->name);
552 static int tun_chr_ioctl(struct inode *inode, struct file *file,
553 unsigned int cmd, unsigned long arg)
555 struct tun_struct *tun = file->private_data;
556 void __user* argp = (void __user*)arg;
558 DECLARE_MAC_BUF(mac);
560 if (cmd == TUNSETIFF || _IOC_TYPE(cmd) == 0x89)
561 if (copy_from_user(&ifr, argp, sizeof ifr))
564 if (cmd == TUNSETIFF && !tun) {
567 ifr.ifr_name[IFNAMSIZ-1] = '\0';
570 err = tun_set_iff(file, &ifr);
576 if (copy_to_user(argp, &ifr, sizeof(ifr)))
584 DBG(KERN_INFO "%s: tun_chr_ioctl cmd %d\n", tun->dev->name, cmd);
588 /* Disable/Enable checksum */
590 tun->flags |= TUN_NOCHECKSUM;
592 tun->flags &= ~TUN_NOCHECKSUM;
594 DBG(KERN_INFO "%s: checksum %s\n",
595 tun->dev->name, arg ? "disabled" : "enabled");
599 /* Disable/Enable persist mode */
601 tun->flags |= TUN_PERSIST;
603 tun->flags &= ~TUN_PERSIST;
605 DBG(KERN_INFO "%s: persist %s\n",
606 tun->dev->name, arg ? "enabled" : "disabled");
610 /* Set owner of the device */
611 tun->owner = (uid_t) arg;
613 DBG(KERN_INFO "%s: owner set to %d\n", tun->dev->name, tun->owner);
617 /* Set group of the device */
618 tun->group= (gid_t) arg;
620 DBG(KERN_INFO "%s: group set to %d\n", tun->dev->name, tun->group);
624 /* Only allow setting the type when the interface is down */
625 if (tun->dev->flags & IFF_UP) {
626 DBG(KERN_INFO "%s: Linktype set failed because interface is up\n",
630 tun->dev->type = (int) arg;
631 DBG(KERN_INFO "%s: linktype set to %d\n", tun->dev->name, tun->dev->type);
642 ifr.ifr_flags = tun->if_flags;
643 if (copy_to_user( argp, &ifr, sizeof ifr))
648 /** Set the character device's interface flags. Currently only
649 * IFF_PROMISC and IFF_ALLMULTI are used. */
650 tun->if_flags = ifr.ifr_flags;
651 DBG(KERN_INFO "%s: interface flags 0x%lx\n",
652 tun->dev->name, tun->if_flags);
656 /* Note: the actual net device's address may be different */
657 memcpy(ifr.ifr_hwaddr.sa_data, tun->dev_addr,
658 min(sizeof ifr.ifr_hwaddr.sa_data, sizeof tun->dev_addr));
659 if (copy_to_user( argp, &ifr, sizeof ifr))
665 /* try to set the actual net device's hw address */
669 ret = dev_set_mac_address(tun->dev, &ifr.ifr_hwaddr);
673 /** Set the character device's hardware address. This is used when
674 * filtering packets being sent from the network device to the character
676 memcpy(tun->dev_addr, ifr.ifr_hwaddr.sa_data,
677 min(sizeof ifr.ifr_hwaddr.sa_data, sizeof tun->dev_addr));
678 DBG(KERN_DEBUG "%s: set hardware address: %x:%x:%x:%x:%x:%x\n",
680 tun->dev_addr[0], tun->dev_addr[1], tun->dev_addr[2],
681 tun->dev_addr[3], tun->dev_addr[4], tun->dev_addr[5]);
688 /** Add the specified group to the character device's multicast filter
690 add_multi(tun->chr_filter, ifr.ifr_hwaddr.sa_data);
691 DBG(KERN_DEBUG "%s: add multi: %s\n",
692 tun->dev->name, print_mac(mac, ifr.ifr_hwaddr.sa_data));
696 /** Remove the specified group from the character device's multicast
698 del_multi(tun->chr_filter, ifr.ifr_hwaddr.sa_data);
699 DBG(KERN_DEBUG "%s: del multi: %s\n",
700 tun->dev->name, print_mac(mac, ifr.ifr_hwaddr.sa_data));
710 static int tun_chr_fasync(int fd, struct file *file, int on)
712 struct tun_struct *tun = file->private_data;
718 DBG(KERN_INFO "%s: tun_chr_fasync %d\n", tun->dev->name, on);
720 if ((ret = fasync_helper(fd, file, on, &tun->fasync)) < 0)
724 ret = __f_setown(file, task_pid(current), PIDTYPE_PID, 0);
727 tun->flags |= TUN_FASYNC;
729 tun->flags &= ~TUN_FASYNC;
734 static int tun_chr_open(struct inode *inode, struct file * file)
736 DBG1(KERN_INFO "tunX: tun_chr_open\n");
737 file->private_data = NULL;
741 static int tun_chr_close(struct inode *inode, struct file *file)
743 struct tun_struct *tun = file->private_data;
748 DBG(KERN_INFO "%s: tun_chr_close\n", tun->dev->name);
750 tun_chr_fasync(-1, file, 0);
754 /* Detach from net device */
755 file->private_data = NULL;
758 /* Drop read queue */
759 skb_queue_purge(&tun->readq);
761 if (!(tun->flags & TUN_PERSIST)) {
762 list_del(&tun->list);
763 unregister_netdevice(tun->dev);
771 static const struct file_operations tun_fops = {
772 .owner = THIS_MODULE,
774 .read = do_sync_read,
775 .aio_read = tun_chr_aio_read,
776 .write = do_sync_write,
777 .aio_write = tun_chr_aio_write,
778 .poll = tun_chr_poll,
779 .ioctl = tun_chr_ioctl,
780 .open = tun_chr_open,
781 .release = tun_chr_close,
782 .fasync = tun_chr_fasync
785 static struct miscdevice tun_miscdev = {
791 /* ethtool interface */
793 static int tun_get_settings(struct net_device *dev, struct ethtool_cmd *cmd)
796 cmd->advertising = 0;
797 cmd->speed = SPEED_10;
798 cmd->duplex = DUPLEX_FULL;
800 cmd->phy_address = 0;
801 cmd->transceiver = XCVR_INTERNAL;
802 cmd->autoneg = AUTONEG_DISABLE;
808 static void tun_get_drvinfo(struct net_device *dev, struct ethtool_drvinfo *info)
810 struct tun_struct *tun = netdev_priv(dev);
812 strcpy(info->driver, DRV_NAME);
813 strcpy(info->version, DRV_VERSION);
814 strcpy(info->fw_version, "N/A");
816 switch (tun->flags & TUN_TYPE_MASK) {
818 strcpy(info->bus_info, "tun");
821 strcpy(info->bus_info, "tap");
826 static u32 tun_get_msglevel(struct net_device *dev)
829 struct tun_struct *tun = netdev_priv(dev);
836 static void tun_set_msglevel(struct net_device *dev, u32 value)
839 struct tun_struct *tun = netdev_priv(dev);
844 static u32 tun_get_link(struct net_device *dev)
846 struct tun_struct *tun = netdev_priv(dev);
847 return tun->attached;
850 static u32 tun_get_rx_csum(struct net_device *dev)
852 struct tun_struct *tun = netdev_priv(dev);
853 return (tun->flags & TUN_NOCHECKSUM) == 0;
856 static int tun_set_rx_csum(struct net_device *dev, u32 data)
858 struct tun_struct *tun = netdev_priv(dev);
860 tun->flags &= ~TUN_NOCHECKSUM;
862 tun->flags |= TUN_NOCHECKSUM;
866 static const struct ethtool_ops tun_ethtool_ops = {
867 .get_settings = tun_get_settings,
868 .get_drvinfo = tun_get_drvinfo,
869 .get_msglevel = tun_get_msglevel,
870 .set_msglevel = tun_set_msglevel,
871 .get_link = tun_get_link,
872 .get_rx_csum = tun_get_rx_csum,
873 .set_rx_csum = tun_set_rx_csum
876 static int __init tun_init(void)
880 printk(KERN_INFO "tun: %s, %s\n", DRV_DESCRIPTION, DRV_VERSION);
881 printk(KERN_INFO "tun: %s\n", DRV_COPYRIGHT);
883 ret = misc_register(&tun_miscdev);
885 printk(KERN_ERR "tun: Can't register misc device %d\n", TUN_MINOR);
889 static void tun_cleanup(void)
891 struct tun_struct *tun, *nxt;
893 misc_deregister(&tun_miscdev);
896 list_for_each_entry_safe(tun, nxt, &tun_dev_list, list) {
897 DBG(KERN_INFO "%s cleaned up\n", tun->dev->name);
898 unregister_netdevice(tun->dev);
904 module_init(tun_init);
905 module_exit(tun_cleanup);
906 MODULE_DESCRIPTION(DRV_DESCRIPTION);
907 MODULE_AUTHOR(DRV_COPYRIGHT);
908 MODULE_LICENSE("GPL");
909 MODULE_ALIAS_MISCDEV(TUN_MINOR);