tree-wide: Assorted spelling fixes
[safe/jmp/linux-2.6] / drivers / net / tun.c
index 4cda69b..e572ecc 100644 (file)
@@ -44,7 +44,6 @@
 #include <linux/kernel.h>
 #include <linux/major.h>
 #include <linux/slab.h>
-#include <linux/smp_lock.h>
 #include <linux/poll.h>
 #include <linux/fcntl.h>
 #include <linux/init.h>
@@ -54,6 +53,7 @@
 #include <linux/miscdevice.h>
 #include <linux/ethtool.h>
 #include <linux/rtnetlink.h>
+#include <linux/compat.h>
 #include <linux/if.h>
 #include <linux/if_arp.h>
 #include <linux/if_ether.h>
@@ -103,13 +103,10 @@ struct tun_struct {
        uid_t                   owner;
        gid_t                   group;
 
-       struct sk_buff_head     readq;
-
        struct net_device       *dev;
        struct fasync_struct    *fasync;
 
        struct tap_filter       txflt;
-       struct sock             *sk;
        struct socket           socket;
 
 #ifdef TUN_DEBUG
@@ -130,17 +127,10 @@ static inline struct tun_sock *tun_sk(struct sock *sk)
 static int tun_attach(struct tun_struct *tun, struct file *file)
 {
        struct tun_file *tfile = file->private_data;
-       const struct cred *cred = current_cred();
        int err;
 
        ASSERT_RTNL();
 
-       /* Check permissions */
-       if (((tun->owner != -1 && cred->euid != tun->owner) ||
-            (tun->group != -1 && !in_egroup_p(tun->group))) &&
-               !capable(CAP_NET_ADMIN))
-               return -EPERM;
-
        netif_tx_lock_bh(tun->dev);
 
        err = -EINVAL;
@@ -155,7 +145,7 @@ static int tun_attach(struct tun_struct *tun, struct file *file)
        tfile->tun = tun;
        tun->tfile = tfile;
        dev_hold(tun->dev);
-       sock_hold(tun->sk);
+       sock_hold(tun->socket.sk);
        atomic_inc(&tfile->count);
 
 out:
@@ -171,7 +161,7 @@ static void __tun_detach(struct tun_struct *tun)
        netif_tx_unlock_bh(tun->dev);
 
        /* Drop read queue */
-       skb_queue_purge(&tun->readq);
+       skb_queue_purge(&tun->socket.sk->sk_receive_queue);
 
        /* Drop the extra count on the net device */
        dev_put(tun->dev);
@@ -340,7 +330,7 @@ static void tun_free_netdev(struct net_device *dev)
 {
        struct tun_struct *tun = netdev_priv(dev);
 
-       sock_put(tun->sk);
+       sock_put(tun->socket.sk);
 }
 
 /* Net device open. */
@@ -358,7 +348,7 @@ static int tun_net_close(struct net_device *dev)
 }
 
 /* Net device start xmit */
-static int tun_net_xmit(struct sk_buff *skb, struct net_device *dev)
+static netdev_tx_t tun_net_xmit(struct sk_buff *skb, struct net_device *dev)
 {
        struct tun_struct *tun = netdev_priv(dev);
 
@@ -374,7 +364,7 @@ static int tun_net_xmit(struct sk_buff *skb, struct net_device *dev)
        if (!check_filter(&tun->txflt, skb))
                goto drop;
 
-       if (skb_queue_len(&tun->readq) >= dev->tx_queue_len) {
+       if (skb_queue_len(&tun->socket.sk->sk_receive_queue) >= dev->tx_queue_len) {
                if (!(tun->flags & TUN_ONE_QUEUE)) {
                        /* Normal queueing mode. */
                        /* Packet scheduler handles dropping of further packets. */
@@ -391,19 +381,19 @@ static int tun_net_xmit(struct sk_buff *skb, struct net_device *dev)
        }
 
        /* Enqueue packet */
-       skb_queue_tail(&tun->readq, skb);
+       skb_queue_tail(&tun->socket.sk->sk_receive_queue, skb);
        dev->trans_start = jiffies;
 
        /* Notify and wake up reader process */
        if (tun->flags & TUN_FASYNC)
                kill_fasync(&tun->fasync, SIGIO, POLL_IN);
        wake_up_interruptible(&tun->socket.wait);
-       return 0;
+       return NETDEV_TX_OK;
 
 drop:
        dev->stats.tx_dropped++;
        kfree_skb(skb);
-       return 0;
+       return NETDEV_TX_OK;
 }
 
 static void tun_net_mclist(struct net_device *dev)
@@ -486,17 +476,19 @@ static unsigned int tun_chr_poll(struct file *file, poll_table * wait)
 {
        struct tun_file *tfile = file->private_data;
        struct tun_struct *tun = __tun_get(tfile);
-       struct sock *sk = tun->sk;
+       struct sock *sk;
        unsigned int mask = 0;
 
        if (!tun)
                return POLLERR;
 
+       sk = tun->socket.sk;
+
        DBG(KERN_INFO "%s: tun_chr_poll\n", tun->dev->name);
 
        poll_wait(file, &tun->socket.wait, wait);
 
-       if (!skb_queue_empty(&tun->readq))
+       if (!skb_queue_empty(&sk->sk_receive_queue))
                mask |= POLLIN | POLLRDNORM;
 
        if (sock_writeable(sk) ||
@@ -517,7 +509,7 @@ static inline struct sk_buff *tun_alloc_skb(struct tun_struct *tun,
                                            size_t prepad, size_t len,
                                            size_t linear, int noblock)
 {
-       struct sock *sk = tun->sk;
+       struct sock *sk = tun->socket.sk;
        struct sk_buff *skb;
        int err;
 
@@ -565,9 +557,13 @@ static __inline__ ssize_t tun_get_user(struct tun_struct *tun,
                if (memcpy_fromiovecend((void *)&gso, iv, offset, sizeof(gso)))
                        return -EFAULT;
 
+               if ((gso.flags & VIRTIO_NET_HDR_F_NEEDS_CSUM) &&
+                   gso.csum_start + gso.csum_offset + 2 > gso.hdr_len)
+                       gso.hdr_len = gso.csum_start + gso.csum_offset + 2;
+
                if (gso.hdr_len > len)
                        return -EINVAL;
-               offset += sizeof(pi);
+               offset += sizeof(gso);
        }
 
        if ((tun->flags & TUN_TYPE_MASK) == TUN_TAP_DEV) {
@@ -635,6 +631,9 @@ static __inline__ ssize_t tun_get_user(struct tun_struct *tun,
                case VIRTIO_NET_HDR_GSO_TCPV6:
                        skb_shinfo(skb)->gso_type = SKB_GSO_TCPV6;
                        break;
+               case VIRTIO_NET_HDR_GSO_UDP:
+                       skb_shinfo(skb)->gso_type = SKB_GSO_UDP;
+                       break;
                default:
                        tun->dev->stats.rx_frame_errors++;
                        kfree_skb(skb);
@@ -720,6 +719,8 @@ static __inline__ ssize_t tun_put_user(struct tun_struct *tun,
                                gso.gso_type = VIRTIO_NET_HDR_GSO_TCPV4;
                        else if (sinfo->gso_type & SKB_GSO_TCPV6)
                                gso.gso_type = VIRTIO_NET_HDR_GSO_TCPV6;
+                       else if (sinfo->gso_type & SKB_GSO_UDP)
+                               gso.gso_type = VIRTIO_NET_HDR_GSO_UDP;
                        else
                                BUG();
                        if (sinfo->gso_type & SKB_GSO_TCP_ECN)
@@ -776,7 +777,7 @@ static ssize_t tun_chr_aio_read(struct kiocb *iocb, const struct iovec *iv,
                current->state = TASK_INTERRUPTIBLE;
 
                /* Read frames from the queue */
-               if (!(skb=skb_dequeue(&tun->readq))) {
+               if (!(skb=skb_dequeue(&tun->socket.sk->sk_receive_queue))) {
                        if (file->f_flags & O_NONBLOCK) {
                                ret = -EAGAIN;
                                break;
@@ -813,8 +814,6 @@ static void tun_setup(struct net_device *dev)
 {
        struct tun_struct *tun = netdev_priv(dev);
 
-       skb_queue_head_init(&tun->readq);
-
        tun->owner = -1;
        tun->group = -1;
 
@@ -844,19 +843,19 @@ static void tun_sock_write_space(struct sock *sk)
        if (!sock_writeable(sk))
                return;
 
-       if (sk->sk_sleep && waitqueue_active(sk->sk_sleep))
-               wake_up_interruptible_sync(sk->sk_sleep);
-
        if (!test_and_clear_bit(SOCK_ASYNC_NOSPACE, &sk->sk_socket->flags))
                return;
 
-       tun = container_of(sk, struct tun_sock, sk)->tun;
+       if (sk->sk_sleep && waitqueue_active(sk->sk_sleep))
+               wake_up_interruptible_sync(sk->sk_sleep);
+
+       tun = tun_sk(sk)->tun;
        kill_fasync(&tun->fasync, SIGIO, POLL_OUT);
 }
 
 static void tun_sock_destruct(struct sock *sk)
 {
-       free_netdev(container_of(sk, struct tun_sock, sk)->tun->dev);
+       free_netdev(tun_sk(sk)->tun->dev);
 }
 
 static struct proto tun_proto = {
@@ -920,6 +919,8 @@ static int tun_set_iff(struct net *net, struct file *file, struct ifreq *ifr)
 
        dev = __dev_get_by_name(net, ifr->ifr_name);
        if (dev) {
+               const struct cred *cred = current_cred();
+
                if (ifr->ifr_flags & IFF_TUN_EXCL)
                        return -EBUSY;
                if ((ifr->ifr_flags & IFF_TUN) && dev->netdev_ops == &tun_netdev_ops)
@@ -929,6 +930,14 @@ static int tun_set_iff(struct net *net, struct file *file, struct ifreq *ifr)
                else
                        return -EINVAL;
 
+               if (((tun->owner != -1 && cred->euid != tun->owner) ||
+                    (tun->group != -1 && !in_egroup_p(tun->group))) &&
+                   !capable(CAP_NET_ADMIN))
+                       return -EPERM;
+               err = security_tun_dev_attach(tun->socket.sk);
+               if (err < 0)
+                       return err;
+
                err = tun_attach(tun, file);
                if (err < 0)
                        return err;
@@ -937,10 +946,11 @@ static int tun_set_iff(struct net *net, struct file *file, struct ifreq *ifr)
                char *name;
                unsigned long flags = 0;
 
-               err = -EINVAL;
-
                if (!capable(CAP_NET_ADMIN))
                        return -EPERM;
+               err = security_tun_dev_create();
+               if (err < 0)
+                       return err;
 
                /* Set dev type */
                if (ifr->ifr_flags & IFF_TUN) {
@@ -952,7 +962,7 @@ static int tun_set_iff(struct net *net, struct file *file, struct ifreq *ifr)
                        flags |= TUN_TAP_DEV;
                        name = "tap%d";
                } else
-                       goto failed;
+                       return -EINVAL;
 
                if (*ifr->ifr_name)
                        name = ifr->ifr_name;
@@ -980,8 +990,9 @@ static int tun_set_iff(struct net *net, struct file *file, struct ifreq *ifr)
                sk->sk_write_space = tun_sock_write_space;
                sk->sk_sndbuf = INT_MAX;
 
-               tun->sk = sk;
-               container_of(sk, struct tun_sock, sk)->tun = tun;
+               tun_sk(sk)->tun = tun;
+
+               security_tun_dev_post_create(sk);
 
                tun_net_init(dev);
 
@@ -991,7 +1002,6 @@ static int tun_set_iff(struct net *net, struct file *file, struct ifreq *ifr)
                                goto err_free_sk;
                }
 
-               err = -EINVAL;
                err = register_netdevice(tun->dev);
                if (err < 0)
                        goto err_free_sk;
@@ -1042,20 +1052,15 @@ static int tun_set_iff(struct net *net, struct file *file, struct ifreq *ifr)
        return err;
 }
 
-static int tun_get_iff(struct net *net, struct file *file, struct ifreq *ifr)
+static int tun_get_iff(struct net *net, struct tun_struct *tun,
+                      struct ifreq *ifr)
 {
-       struct tun_struct *tun = tun_get(file);
-
-       if (!tun)
-               return -EBADFD;
-
        DBG(KERN_INFO "%s: tun_get_iff\n", tun->dev->name);
 
        strcpy(ifr->ifr_name, tun->dev->name);
 
        ifr->ifr_flags = tun_flags(tun);
 
-       tun_put(tun);
        return 0;
 }
 
@@ -1068,7 +1073,8 @@ static int set_offload(struct net_device *dev, unsigned long arg)
        old_features = dev->features;
        /* Unset features, set them as we chew on the arg. */
        features = (old_features & ~(NETIF_F_HW_CSUM|NETIF_F_SG|NETIF_F_FRAGLIST
-                                   |NETIF_F_TSO_ECN|NETIF_F_TSO|NETIF_F_TSO6));
+                                   |NETIF_F_TSO_ECN|NETIF_F_TSO|NETIF_F_TSO6
+                                   |NETIF_F_UFO));
 
        if (arg & TUN_F_CSUM) {
                features |= NETIF_F_HW_CSUM|NETIF_F_SG|NETIF_F_FRAGLIST;
@@ -1085,6 +1091,11 @@ static int set_offload(struct net_device *dev, unsigned long arg)
                                features |= NETIF_F_TSO6;
                        arg &= ~(TUN_F_TSO4|TUN_F_TSO6);
                }
+
+               if (arg & TUN_F_UFO) {
+                       features |= NETIF_F_UFO;
+                       arg &= ~TUN_F_UFO;
+               }
        }
 
        /* This gives the user a way to test for new features in future by
@@ -1099,8 +1110,8 @@ static int set_offload(struct net_device *dev, unsigned long arg)
        return 0;
 }
 
-static int tun_chr_ioctl(struct inode *inode, struct file *file,
-                        unsigned int cmd, unsigned long arg)
+static long __tun_chr_ioctl(struct file *file, unsigned int cmd,
+                           unsigned long arg, int ifreq_len)
 {
        struct tun_file *tfile = file->private_data;
        struct tun_struct *tun;
@@ -1110,7 +1121,7 @@ static int tun_chr_ioctl(struct inode *inode, struct file *file,
        int ret;
 
        if (cmd == TUNSETIFF || _IOC_TYPE(cmd) == 0x89)
-               if (copy_from_user(&ifr, argp, sizeof ifr))
+               if (copy_from_user(&ifr, argp, ifreq_len))
                        return -EFAULT;
 
        if (cmd == TUNGETFEATURES) {
@@ -1122,38 +1133,36 @@ static int tun_chr_ioctl(struct inode *inode, struct file *file,
                                (unsigned int __user*)argp);
        }
 
+       rtnl_lock();
+
        tun = __tun_get(tfile);
        if (cmd == TUNSETIFF && !tun) {
-               int err;
-
                ifr.ifr_name[IFNAMSIZ-1] = '\0';
 
-               rtnl_lock();
-               err = tun_set_iff(tfile->net, file, &ifr);
-               rtnl_unlock();
+               ret = tun_set_iff(tfile->net, file, &ifr);
 
-               if (err)
-                       return err;
+               if (ret)
+                       goto unlock;
 
-               if (copy_to_user(argp, &ifr, sizeof(ifr)))
-                       return -EFAULT;
-               return 0;
+               if (copy_to_user(argp, &ifr, ifreq_len))
+                       ret = -EFAULT;
+               goto unlock;
        }
 
-
+       ret = -EBADFD;
        if (!tun)
-               return -EBADFD;
+               goto unlock;
 
        DBG(KERN_INFO "%s: tun_chr_ioctl cmd %d\n", tun->dev->name, cmd);
 
        ret = 0;
        switch (cmd) {
        case TUNGETIFF:
-               ret = tun_get_iff(current->nsproxy->net_ns, file, &ifr);
+               ret = tun_get_iff(current->nsproxy->net_ns, tun, &ifr);
                if (ret)
                        break;
 
-               if (copy_to_user(argp, &ifr, sizeof(ifr)))
+               if (copy_to_user(argp, &ifr, ifreq_len))
                        ret = -EFAULT;
                break;
 
@@ -1195,7 +1204,6 @@ static int tun_chr_ioctl(struct inode *inode, struct file *file,
 
        case TUNSETLINK:
                /* Only allow setting the type when the interface is down */
-               rtnl_lock();
                if (tun->dev->flags & IFF_UP) {
                        DBG(KERN_INFO "%s: Linktype set failed because interface is up\n",
                                tun->dev->name);
@@ -1205,7 +1213,6 @@ static int tun_chr_ioctl(struct inode *inode, struct file *file,
                        DBG(KERN_INFO "%s: linktype set to %d\n", tun->dev->name, tun->dev->type);
                        ret = 0;
                }
-               rtnl_unlock();
                break;
 
 #ifdef TUN_DEBUG
@@ -1214,9 +1221,7 @@ static int tun_chr_ioctl(struct inode *inode, struct file *file,
                break;
 #endif
        case TUNSETOFFLOAD:
-               rtnl_lock();
                ret = set_offload(tun->dev, arg);
-               rtnl_unlock();
                break;
 
        case TUNSETTXFILTER:
@@ -1224,16 +1229,14 @@ static int tun_chr_ioctl(struct inode *inode, struct file *file,
                ret = -EINVAL;
                if ((tun->flags & TUN_TYPE_MASK) != TUN_TAP_DEV)
                        break;
-               rtnl_lock();
                ret = update_filter(&tun->txflt, (void __user *)arg);
-               rtnl_unlock();
                break;
 
        case SIOCGIFHWADDR:
                /* Get hw addres */
                memcpy(ifr.ifr_hwaddr.sa_data, tun->dev->dev_addr, ETH_ALEN);
                ifr.ifr_hwaddr.sa_family = tun->dev->type;
-               if (copy_to_user(argp, &ifr, sizeof ifr))
+               if (copy_to_user(argp, &ifr, ifreq_len))
                        ret = -EFAULT;
                break;
 
@@ -1242,13 +1245,11 @@ static int tun_chr_ioctl(struct inode *inode, struct file *file,
                DBG(KERN_DEBUG "%s: set hw address: %pM\n",
                        tun->dev->name, ifr.ifr_hwaddr.sa_data);
 
-               rtnl_lock();
                ret = dev_set_mac_address(tun->dev, &ifr.ifr_hwaddr);
-               rtnl_unlock();
                break;
 
        case TUNGETSNDBUF:
-               sndbuf = tun->sk->sk_sndbuf;
+               sndbuf = tun->socket.sk->sk_sndbuf;
                if (copy_to_user(argp, &sndbuf, sizeof(sndbuf)))
                        ret = -EFAULT;
                break;
@@ -1259,7 +1260,7 @@ static int tun_chr_ioctl(struct inode *inode, struct file *file,
                        break;
                }
 
-               tun->sk->sk_sndbuf = sndbuf;
+               tun->socket.sk->sk_sndbuf = sndbuf;
                break;
 
        default:
@@ -1267,10 +1268,48 @@ static int tun_chr_ioctl(struct inode *inode, struct file *file,
                break;
        };
 
-       tun_put(tun);
+unlock:
+       rtnl_unlock();
+       if (tun)
+               tun_put(tun);
        return ret;
 }
 
+static long tun_chr_ioctl(struct file *file,
+                         unsigned int cmd, unsigned long arg)
+{
+       return __tun_chr_ioctl(file, cmd, arg, sizeof (struct ifreq));
+}
+
+#ifdef CONFIG_COMPAT
+static long tun_chr_compat_ioctl(struct file *file,
+                        unsigned int cmd, unsigned long arg)
+{
+       switch (cmd) {
+       case TUNSETIFF:
+       case TUNGETIFF:
+       case TUNSETTXFILTER:
+       case TUNGETSNDBUF:
+       case TUNSETSNDBUF:
+       case SIOCGIFHWADDR:
+       case SIOCSIFHWADDR:
+               arg = (unsigned long)compat_ptr(arg);
+               break;
+       default:
+               arg = (compat_ulong_t)arg;
+               break;
+       }
+
+       /*
+        * compat_ifreq is shorter than ifreq, so we must not access beyond
+        * the end of that structure. All fields that are used in this
+        * driver are compatible though, we don't need to convert the
+        * contents.
+        */
+       return __tun_chr_ioctl(file, cmd, arg, sizeof(struct compat_ifreq));
+}
+#endif /* CONFIG_COMPAT */
+
 static int tun_chr_fasync(int fd, struct file *file, int on)
 {
        struct tun_struct *tun = tun_get(file);
@@ -1281,7 +1320,6 @@ static int tun_chr_fasync(int fd, struct file *file, int on)
 
        DBG(KERN_INFO "%s: tun_chr_fasync %d\n", tun->dev->name, on);
 
-       lock_kernel();
        if ((ret = fasync_helper(fd, file, on, &tun->fasync)) < 0)
                goto out;
 
@@ -1294,7 +1332,6 @@ static int tun_chr_fasync(int fd, struct file *file, int on)
                tun->flags &= ~TUN_FASYNC;
        ret = 0;
 out:
-       unlock_kernel();
        tun_put(tun);
        return ret;
 }
@@ -1302,7 +1339,7 @@ out:
 static int tun_chr_open(struct inode *inode, struct file * file)
 {
        struct tun_file *tfile;
-       cycle_kernel_lock();
+
        DBG1(KERN_INFO "tunX: tun_chr_open\n");
 
        tfile = kmalloc(sizeof(*tfile), GFP_KERNEL);
@@ -1318,25 +1355,28 @@ static int tun_chr_open(struct inode *inode, struct file * file)
 static int tun_chr_close(struct inode *inode, struct file *file)
 {
        struct tun_file *tfile = file->private_data;
-       struct tun_struct *tun = __tun_get(tfile);
-
+       struct tun_struct *tun;
 
+       tun = __tun_get(tfile);
        if (tun) {
-               DBG(KERN_INFO "%s: tun_chr_close\n", tun->dev->name);
+               struct net_device *dev = tun->dev;
 
-               rtnl_lock();
-               __tun_detach(tun);
+               DBG(KERN_INFO "%s: tun_chr_close\n", dev->name);
 
-               /* If desireable, unregister the netdevice. */
-               if (!(tun->flags & TUN_PERSIST))
-                       unregister_netdevice(tun->dev);
+               __tun_detach(tun);
 
-               rtnl_unlock();
+               /* If desirable, unregister the netdevice. */
+               if (!(tun->flags & TUN_PERSIST)) {
+                       rtnl_lock();
+                       if (dev->reg_state == NETREG_REGISTERED)
+                               unregister_netdevice(dev);
+                       rtnl_unlock();
+               }
        }
 
        tun = tfile->tun;
        if (tun)
-               sock_put(tun->sk);
+               sock_put(tun->socket.sk);
 
        put_net(tfile->net);
        kfree(tfile);
@@ -1352,7 +1392,10 @@ static const struct file_operations tun_fops = {
        .write = do_sync_write,
        .aio_write = tun_chr_aio_write,
        .poll   = tun_chr_poll,
-       .ioctl  = tun_chr_ioctl,
+       .unlocked_ioctl = tun_chr_ioctl,
+#ifdef CONFIG_COMPAT
+       .compat_ioctl = tun_chr_compat_ioctl,
+#endif
        .open   = tun_chr_open,
        .release = tun_chr_close,
        .fasync = tun_chr_fasync
@@ -1361,6 +1404,7 @@ static const struct file_operations tun_fops = {
 static struct miscdevice tun_miscdev = {
        .minor = TUN_MINOR,
        .name = "tun",
+       .nodename = "net/tun",
        .fops = &tun_fops,
 };