macvtap.c 30.8 KB
Newer Older
Arnd Bergmann's avatar
Arnd Bergmann committed
1 2
#include <linux/etherdevice.h>
#include <linux/if_macvlan.h>
3
#include <linux/if_vlan.h>
Arnd Bergmann's avatar
Arnd Bergmann committed
4 5 6 7 8 9 10 11 12
#include <linux/interrupt.h>
#include <linux/nsproxy.h>
#include <linux/compat.h>
#include <linux/if_tun.h>
#include <linux/module.h>
#include <linux/skbuff.h>
#include <linux/cache.h>
#include <linux/sched.h>
#include <linux/types.h>
13
#include <linux/slab.h>
Arnd Bergmann's avatar
Arnd Bergmann committed
14 15
#include <linux/wait.h>
#include <linux/cdev.h>
Al Viro's avatar
Al Viro committed
16
#include <linux/idr.h>
Arnd Bergmann's avatar
Arnd Bergmann committed
17
#include <linux/fs.h>
Herbert Xu's avatar
Herbert Xu committed
18
#include <linux/uio.h>
Arnd Bergmann's avatar
Arnd Bergmann committed
19 20 21 22

#include <net/net_namespace.h>
#include <net/rtnetlink.h>
#include <net/sock.h>
23
#include <linux/virtio_net.h>
Arnd Bergmann's avatar
Arnd Bergmann committed
24 25 26 27 28 29 30 31 32 33 34 35 36 37

/*
 * A macvtap queue is the central object of this driver, it connects
 * an open character device to a macvlan interface. There can be
 * multiple queues on one interface, which map back to queues
 * implemented in hardware on the underlying device.
 *
 * macvtap_proto is used to allocate queues through the sock allocation
 * mechanism.
 *
 */
struct macvtap_queue {
	struct sock sk;
	struct socket sock;
38
	struct socket_wq wq;
39
	int vnet_hdr_sz;
40
	struct macvlan_dev __rcu *vlan;
Arnd Bergmann's avatar
Arnd Bergmann committed
41
	struct file *file;
42
	unsigned int flags;
43
	u16 queue_index;
Jason Wang's avatar
Jason Wang committed
44 45
	bool enabled;
	struct list_head next;
Arnd Bergmann's avatar
Arnd Bergmann committed
46 47
};

48 49 50
#define MACVTAP_FEATURES (IFF_VNET_HDR | IFF_MULTI_QUEUE)

#define MACVTAP_VNET_LE 0x80000000
51 52 53

static inline u16 macvtap16_to_cpu(struct macvtap_queue *q, __virtio16 val)
{
54
	return __virtio16_to_cpu(q->flags & MACVTAP_VNET_LE, val);
55 56 57 58
}

static inline __virtio16 cpu_to_macvtap16(struct macvtap_queue *q, u16 val)
{
59
	return __cpu_to_virtio16(q->flags & MACVTAP_VNET_LE, val);
60 61
}

Arnd Bergmann's avatar
Arnd Bergmann committed
62 63 64 65 66 67 68
static struct proto macvtap_proto = {
	.name = "macvtap",
	.owner = THIS_MODULE,
	.obj_size = sizeof (struct macvtap_queue),
};

/*
69
 * Variables for dealing with macvtaps device numbers.
Arnd Bergmann's avatar
Arnd Bergmann committed
70
 */
71
static dev_t macvtap_major;
72 73 74 75
#define MACVTAP_NUM_DEVS (1U << MINORBITS)
static DEFINE_MUTEX(minor_lock);
static DEFINE_IDR(minor_idr);

76
#define GOODCOPY_LEN 128
Arnd Bergmann's avatar
Arnd Bergmann committed
77 78 79
static struct class *macvtap_class;
static struct cdev macvtap_cdev;

80 81
static const struct proto_ops macvtap_socket_ops;

82
#define TUN_OFFLOADS (NETIF_F_HW_CSUM | NETIF_F_TSO_ECN | NETIF_F_TSO | \
83
		      NETIF_F_TSO6 | NETIF_F_UFO)
84
#define RX_OFFLOADS (NETIF_F_GRO | NETIF_F_LRO)
85 86
#define TAP_FEATURES (NETIF_F_GSO | NETIF_F_SG)

87 88 89 90 91
static struct macvlan_dev *macvtap_get_vlan_rcu(const struct net_device *dev)
{
	return rcu_dereference(dev->rx_handler_data);
}

Arnd Bergmann's avatar
Arnd Bergmann committed
92 93
/*
 * RCU usage:
94 95
 * The macvtap_queue and the macvlan_dev are loosely coupled, the
 * pointers from one to the other can only be read while rcu_read_lock
96
 * or rtnl is held.
Arnd Bergmann's avatar
Arnd Bergmann committed
97
 *
98 99 100 101
 * Both the file and the macvlan_dev hold a reference on the macvtap_queue
 * through sock_hold(&q->sk). When the macvlan_dev goes away first,
 * q->vlan becomes inaccessible. When the files gets closed,
 * macvtap_get_queue() fails.
Arnd Bergmann's avatar
Arnd Bergmann committed
102
 *
103 104 105 106
 * There may still be references to the struct sock inside of the
 * queue from outbound SKBs, but these never reference back to the
 * file or the dev. The data structure is freed through __sk_free
 * when both our references and any pending SKBs are gone.
Arnd Bergmann's avatar
Arnd Bergmann committed
107 108
 */

Jason Wang's avatar
Jason Wang committed
109
static int macvtap_enable_queue(struct net_device *dev, struct file *file,
Arnd Bergmann's avatar
Arnd Bergmann committed
110
				struct macvtap_queue *q)
Jason Wang's avatar
Jason Wang committed
111 112 113 114
{
	struct macvlan_dev *vlan = netdev_priv(dev);
	int err = -EINVAL;

115
	ASSERT_RTNL();
Jason Wang's avatar
Jason Wang committed
116 117 118 119 120 121 122 123 124 125 126 127 128 129

	if (q->enabled)
		goto out;

	err = 0;
	rcu_assign_pointer(vlan->taps[vlan->numvtaps], q);
	q->queue_index = vlan->numvtaps;
	q->enabled = true;

	vlan->numvtaps++;
out:
	return err;
}

130
/* Requires RTNL */
Jason Wang's avatar
Jason Wang committed
131 132
static int macvtap_set_queue(struct net_device *dev, struct file *file,
			     struct macvtap_queue *q)
Arnd Bergmann's avatar
Arnd Bergmann committed
133 134 135
{
	struct macvlan_dev *vlan = netdev_priv(dev);

Jason Wang's avatar
Jason Wang committed
136
	if (vlan->numqueues == MAX_MACVTAP_QUEUES)
137
		return -EBUSY;
Arnd Bergmann's avatar
Arnd Bergmann committed
138

139
	rcu_assign_pointer(q->vlan, vlan);
140
	rcu_assign_pointer(vlan->taps[vlan->numvtaps], q);
141
	sock_hold(&q->sk);
Arnd Bergmann's avatar
Arnd Bergmann committed
142 143

	q->file = file;
144
	q->queue_index = vlan->numvtaps;
Jason Wang's avatar
Jason Wang committed
145
	q->enabled = true;
146
	file->private_data = q;
Jason Wang's avatar
Jason Wang committed
147
	list_add_tail(&q->next, &vlan->queue_list);
Arnd Bergmann's avatar
Arnd Bergmann committed
148

149
	vlan->numvtaps++;
Jason Wang's avatar
Jason Wang committed
150
	vlan->numqueues++;
151

152
	return 0;
Arnd Bergmann's avatar
Arnd Bergmann committed
153 154
}

155
static int macvtap_disable_queue(struct macvtap_queue *q)
Jason Wang's avatar
Jason Wang committed
156 157 158 159
{
	struct macvlan_dev *vlan;
	struct macvtap_queue *nq;

160
	ASSERT_RTNL();
Jason Wang's avatar
Jason Wang committed
161 162 163
	if (!q->enabled)
		return -EINVAL;

164 165
	vlan = rtnl_dereference(q->vlan);

Jason Wang's avatar
Jason Wang committed
166 167 168
	if (vlan) {
		int index = q->queue_index;
		BUG_ON(index >= vlan->numvtaps);
169
		nq = rtnl_dereference(vlan->taps[vlan->numvtaps - 1]);
Jason Wang's avatar
Jason Wang committed
170 171 172 173 174 175 176 177 178 179 180 181
		nq->queue_index = index;

		rcu_assign_pointer(vlan->taps[index], nq);
		RCU_INIT_POINTER(vlan->taps[vlan->numvtaps - 1], NULL);
		q->enabled = false;

		vlan->numvtaps--;
	}

	return 0;
}

Arnd Bergmann's avatar
Arnd Bergmann committed
182
/*
183 184 185
 * The file owning the queue got closed, give up both
 * the reference that the files holds as well as the
 * one from the macvlan_dev if that still exists.
Arnd Bergmann's avatar
Arnd Bergmann committed
186 187 188 189
 *
 * Using the spinlock makes sure that we don't get
 * to the queue again after destroying it.
 */
190
static void macvtap_put_queue(struct macvtap_queue *q)
Arnd Bergmann's avatar
Arnd Bergmann committed
191
{
192
	struct macvlan_dev *vlan;
Arnd Bergmann's avatar
Arnd Bergmann committed
193

194 195 196
	rtnl_lock();
	vlan = rtnl_dereference(q->vlan);

197
	if (vlan) {
Jason Wang's avatar
Jason Wang committed
198
		if (q->enabled)
199
			BUG_ON(macvtap_disable_queue(q));
200

Jason Wang's avatar
Jason Wang committed
201
		vlan->numqueues--;
202
		RCU_INIT_POINTER(q->vlan, NULL);
203
		sock_put(&q->sk);
Jason Wang's avatar
Jason Wang committed
204
		list_del_init(&q->next);
Arnd Bergmann's avatar
Arnd Bergmann committed
205 206
	}

207
	rtnl_unlock();
Arnd Bergmann's avatar
Arnd Bergmann committed
208 209 210 211 212 213

	synchronize_rcu();
	sock_put(&q->sk);
}

/*
214 215 216 217 218
 * Select a queue based on the rxq of the device on which this packet
 * arrived. If the incoming device is not mq, calculate a flow hash
 * to select a queue. If all fails, find the first available queue.
 * Cache vlan->numvtaps since it can become zero during the execution
 * of this function.
Arnd Bergmann's avatar
Arnd Bergmann committed
219 220 221 222 223
 */
static struct macvtap_queue *macvtap_get_queue(struct net_device *dev,
					       struct sk_buff *skb)
{
	struct macvlan_dev *vlan = netdev_priv(dev);
224
	struct macvtap_queue *tap = NULL;
Jason Wang's avatar
Jason Wang committed
225 226 227 228 229
	/* Access to taps array is protected by rcu, but access to numvtaps
	 * isn't. Below we use it to lookup a queue, but treat it as a hint
	 * and validate that the result isn't NULL - in case we are
	 * racing against queue removal.
	 */
230
	int numvtaps = ACCESS_ONCE(vlan->numvtaps);
231 232 233 234 235
	__u32 rxq;

	if (!numvtaps)
		goto out;

236
	/* Check if we can use flow to select a queue */
237
	rxq = skb_get_hash(skb);
238 239
	if (rxq) {
		tap = rcu_dereference(vlan->taps[rxq % numvtaps]);
240
		goto out;
241 242
	}

243 244
	if (likely(skb_rx_queue_recorded(skb))) {
		rxq = skb_get_rx_queue(skb);
Arnd Bergmann's avatar
Arnd Bergmann committed
245

246 247 248 249
		while (unlikely(rxq >= numvtaps))
			rxq -= numvtaps;

		tap = rcu_dereference(vlan->taps[rxq]);
250
		goto out;
251 252
	}

253
	tap = rcu_dereference(vlan->taps[0]);
254 255
out:
	return tap;
Arnd Bergmann's avatar
Arnd Bergmann committed
256 257
}

258 259
/*
 * The net_device is going away, give up the reference
260 261
 * that it holds on all queues and safely set the pointer
 * from the queues to NULL.
262
 */
Arnd Bergmann's avatar
Arnd Bergmann committed
263 264 265
static void macvtap_del_queues(struct net_device *dev)
{
	struct macvlan_dev *vlan = netdev_priv(dev);
266
	struct macvtap_queue *q, *tmp;
267

268
	ASSERT_RTNL();
Jason Wang's avatar
Jason Wang committed
269 270
	list_for_each_entry_safe(q, tmp, &vlan->queue_list, next) {
		list_del_init(&q->next);
271
		RCU_INIT_POINTER(q->vlan, NULL);
Jason Wang's avatar
Jason Wang committed
272 273 274
		if (q->enabled)
			vlan->numvtaps--;
		vlan->numqueues--;
275
		sock_put(&q->sk);
276
	}
Jason Wang's avatar
Jason Wang committed
277 278
	BUG_ON(vlan->numvtaps);
	BUG_ON(vlan->numqueues);
279 280
	/* guarantee that any future macvtap_set_queue will fail */
	vlan->numvtaps = MAX_MACVTAP_QUEUES;
Arnd Bergmann's avatar
Arnd Bergmann committed
281 282
}

283
static rx_handler_result_t macvtap_handle_frame(struct sk_buff **pskb)
Arnd Bergmann's avatar
Arnd Bergmann committed
284
{
285 286 287 288
	struct sk_buff *skb = *pskb;
	struct net_device *dev = skb->dev;
	struct macvlan_dev *vlan;
	struct macvtap_queue *q;
289 290
	netdev_features_t features = TAP_FEATURES;

291 292 293 294 295
	vlan = macvtap_get_vlan_rcu(dev);
	if (!vlan)
		return RX_HANDLER_PASS;

	q = macvtap_get_queue(dev, skb);
Arnd Bergmann's avatar
Arnd Bergmann committed
296
	if (!q)
297
		return RX_HANDLER_PASS;
298 299 300

	if (skb_queue_len(&q->sk.sk_receive_queue) >= dev->tx_queue_len)
		goto drop;
Arnd Bergmann's avatar
Arnd Bergmann committed
301

302 303
	skb_push(skb, ETH_HLEN);

304
	/* Apply the forward feature mask so that we perform segmentation
305 306
	 * according to users wishes.  This only works if VNET_HDR is
	 * enabled.
307
	 */
308 309
	if (q->flags & IFF_VNET_HDR)
		features |= vlan->tap_features;
310
	if (netif_needs_gso(skb, features)) {
311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329
		struct sk_buff *segs = __skb_gso_segment(skb, features, false);

		if (IS_ERR(segs))
			goto drop;

		if (!segs) {
			skb_queue_tail(&q->sk.sk_receive_queue, skb);
			goto wake_up;
		}

		kfree_skb(skb);
		while (segs) {
			struct sk_buff *nskb = segs->next;

			segs->next = NULL;
			skb_queue_tail(&q->sk.sk_receive_queue, segs);
			segs = nskb;
		}
	} else {
330 331 332 333 334 335 336 337 338
		/* If we receive a partial checksum and the tap side
		 * doesn't support checksum offload, compute the checksum.
		 * Note: it doesn't matter which checksum feature to
		 *        check, we either support them all or none.
		 */
		if (skb->ip_summed == CHECKSUM_PARTIAL &&
		    !(features & NETIF_F_ALL_CSUM) &&
		    skb_checksum_help(skb))
			goto drop;
339 340 341 342
		skb_queue_tail(&q->sk.sk_receive_queue, skb);
	}

wake_up:
Eric Dumazet's avatar
Eric Dumazet committed
343
	wake_up_interruptible_poll(sk_sleep(&q->sk), POLLIN | POLLRDNORM | POLLRDBAND);
344
	return RX_HANDLER_CONSUMED;
345 346

drop:
347 348
	/* Count errors/drops only here, thus don't care about args. */
	macvlan_count_rx(vlan, 0, 0, 0);
349
	kfree_skb(skb);
350
	return RX_HANDLER_CONSUMED;
Arnd Bergmann's avatar
Arnd Bergmann committed
351 352
}

353 354 355 356 357
static int macvtap_get_minor(struct macvlan_dev *vlan)
{
	int retval = -ENOMEM;

	mutex_lock(&minor_lock);
Tejun Heo's avatar
Tejun Heo committed
358 359 360 361
	retval = idr_alloc(&minor_idr, vlan, 1, MACVTAP_NUM_DEVS, GFP_KERNEL);
	if (retval >= 0) {
		vlan->minor = retval;
	} else if (retval == -ENOSPC) {
362 363 364 365
		printk(KERN_ERR "too many macvtap devices\n");
		retval = -EINVAL;
	}
	mutex_unlock(&minor_lock);
Tejun Heo's avatar
Tejun Heo committed
366
	return retval < 0 ? retval : 0;
367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393
}

static void macvtap_free_minor(struct macvlan_dev *vlan)
{
	mutex_lock(&minor_lock);
	if (vlan->minor) {
		idr_remove(&minor_idr, vlan->minor);
		vlan->minor = 0;
	}
	mutex_unlock(&minor_lock);
}

static struct net_device *dev_get_by_macvtap_minor(int minor)
{
	struct net_device *dev = NULL;
	struct macvlan_dev *vlan;

	mutex_lock(&minor_lock);
	vlan = idr_find(&minor_idr, minor);
	if (vlan) {
		dev = vlan->dev;
		dev_hold(dev);
	}
	mutex_unlock(&minor_lock);
	return dev;
}

Arnd Bergmann's avatar
Arnd Bergmann committed
394 395 396 397 398
static int macvtap_newlink(struct net *src_net,
			   struct net_device *dev,
			   struct nlattr *tb[],
			   struct nlattr *data[])
{
Jason Wang's avatar
Jason Wang committed
399
	struct macvlan_dev *vlan = netdev_priv(dev);
400 401
	int err;

Jason Wang's avatar
Jason Wang committed
402 403
	INIT_LIST_HEAD(&vlan->queue_list);

404 405 406 407 408
	/* Since macvlan supports all offloads by default, make
	 * tap support all offloads also.
	 */
	vlan->tap_features = TUN_OFFLOADS;

409 410 411 412
	err = netdev_rx_handler_register(dev, macvtap_handle_frame, vlan);
	if (err)
		return err;

413 414 415
	/* Don't put anything that may fail after macvlan_common_newlink
	 * because we can't undo what it does.
	 */
416
	return macvlan_common_newlink(src_net, dev, tb, data);
Arnd Bergmann's avatar
Arnd Bergmann committed
417 418 419 420 421
}

static void macvtap_dellink(struct net_device *dev,
			    struct list_head *head)
{
422
	netdev_rx_handler_unregister(dev);
Arnd Bergmann's avatar
Arnd Bergmann committed
423 424 425 426
	macvtap_del_queues(dev);
	macvlan_dellink(dev, head);
}

427 428 429 430 431 432
static void macvtap_setup(struct net_device *dev)
{
	macvlan_common_setup(dev);
	dev->tx_queue_len = TUN_READQ_SIZE;
}

Arnd Bergmann's avatar
Arnd Bergmann committed
433 434
static struct rtnl_link_ops macvtap_link_ops __read_mostly = {
	.kind		= "macvtap",
435
	.setup		= macvtap_setup,
Arnd Bergmann's avatar
Arnd Bergmann committed
436 437 438 439 440 441 442
	.newlink	= macvtap_newlink,
	.dellink	= macvtap_dellink,
};


static void macvtap_sock_write_space(struct sock *sk)
{
443 444
	wait_queue_head_t *wqueue;

Arnd Bergmann's avatar
Arnd Bergmann committed
445 446 447 448
	if (!sock_writeable(sk) ||
	    !test_and_clear_bit(SOCK_ASYNC_NOSPACE, &sk->sk_socket->flags))
		return;

449 450 451
	wqueue = sk_sleep(sk);
	if (wqueue && waitqueue_active(wqueue))
		wake_up_interruptible_poll(wqueue, POLLOUT | POLLWRNORM | POLLWRBAND);
Arnd Bergmann's avatar
Arnd Bergmann committed
452 453
}

454 455 456 457 458
static void macvtap_sock_destruct(struct sock *sk)
{
	skb_queue_purge(&sk->sk_receive_queue);
}

Arnd Bergmann's avatar
Arnd Bergmann committed
459 460 461
static int macvtap_open(struct inode *inode, struct file *file)
{
	struct net *net = current->nsproxy->net_ns;
462
	struct net_device *dev;
Arnd Bergmann's avatar
Arnd Bergmann committed
463
	struct macvtap_queue *q;
464
	int err = -ENODEV;
Arnd Bergmann's avatar
Arnd Bergmann committed
465

466 467
	rtnl_lock();
	dev = dev_get_by_macvtap_minor(iminor(inode));
Arnd Bergmann's avatar
Arnd Bergmann committed
468 469 470 471 472
	if (!dev)
		goto out;

	err = -ENOMEM;
	q = (struct macvtap_queue *)sk_alloc(net, AF_UNSPEC, GFP_KERNEL,
473
					     &macvtap_proto, 0);
Arnd Bergmann's avatar
Arnd Bergmann committed
474 475 476
	if (!q)
		goto out;

Jason Wang's avatar
Jason Wang committed
477
	RCU_INIT_POINTER(q->sock.wq, &q->wq);
478
	init_waitqueue_head(&q->wq.wait);
Arnd Bergmann's avatar
Arnd Bergmann committed
479 480
	q->sock.type = SOCK_RAW;
	q->sock.state = SS_CONNECTED;
481 482
	q->sock.file = file;
	q->sock.ops = &macvtap_socket_ops;
Arnd Bergmann's avatar
Arnd Bergmann committed
483 484
	sock_init_data(&q->sock, &q->sk);
	q->sk.sk_write_space = macvtap_sock_write_space;
485
	q->sk.sk_destruct = macvtap_sock_destruct;
486
	q->flags = IFF_VNET_HDR | IFF_NO_PI | IFF_TAP;
487
	q->vnet_hdr_sz = sizeof(struct virtio_net_hdr);
Arnd Bergmann's avatar
Arnd Bergmann committed
488

489 490 491
	/*
	 * so far only KVM virtio_net uses macvtap, enable zero copy between
	 * guest kernel and host kernel when lower device supports zerocopy
492 493 494
	 *
	 * The macvlan supports zerocopy iff the lower device supports zero
	 * copy so we don't have to look at the lower device directly.
495
	 */
496 497
	if ((dev->features & NETIF_F_HIGHDMA) && (dev->features & NETIF_F_SG))
		sock_set_flag(&q->sk, SOCK_ZEROCOPY);
498

Arnd Bergmann's avatar
Arnd Bergmann committed
499 500 501 502 503 504 505 506
	err = macvtap_set_queue(dev, file, q);
	if (err)
		sock_put(&q->sk);

out:
	if (dev)
		dev_put(dev);

507
	rtnl_unlock();
Arnd Bergmann's avatar
Arnd Bergmann committed
508 509 510 511 512
	return err;
}

static int macvtap_release(struct inode *inode, struct file *file)
{
513 514
	struct macvtap_queue *q = file->private_data;
	macvtap_put_queue(q);
Arnd Bergmann's avatar
Arnd Bergmann committed
515 516 517 518 519
	return 0;
}

static unsigned int macvtap_poll(struct file *file, poll_table * wait)
{
520
	struct macvtap_queue *q = file->private_data;
Arnd Bergmann's avatar
Arnd Bergmann committed
521 522 523 524 525 526
	unsigned int mask = POLLERR;

	if (!q)
		goto out;

	mask = 0;
527
	poll_wait(file, &q->wq.wait, wait);
Arnd Bergmann's avatar
Arnd Bergmann committed
528 529 530 531 532 533 534 535 536 537 538 539 540

	if (!skb_queue_empty(&q->sk.sk_receive_queue))
		mask |= POLLIN | POLLRDNORM;

	if (sock_writeable(&q->sk) ||
	    (!test_and_set_bit(SOCK_ASYNC_NOSPACE, &q->sock.flags) &&
	     sock_writeable(&q->sk)))
		mask |= POLLOUT | POLLWRNORM;

out:
	return mask;
}

541 542 543 544 545 546 547 548 549 550 551
static inline struct sk_buff *macvtap_alloc_skb(struct sock *sk, size_t prepad,
						size_t len, size_t linear,
						int noblock, int *err)
{
	struct sk_buff *skb;

	/* Under a page?  Don't bother with paged skb. */
	if (prepad + len < PAGE_SIZE || !linear)
		linear = len;

	skb = sock_alloc_send_pskb(sk, prepad + linear, len - linear, noblock,
552
				   err, 0);
553 554 555 556 557 558 559 560 561 562 563 564 565 566 567
	if (!skb)
		return NULL;

	skb_reserve(skb, prepad);
	skb_put(skb, linear);
	skb->data_len = len - linear;
	skb->len += len - linear;

	return skb;
}

/*
 * macvtap_skb_from_vnet_hdr and macvtap_skb_to_vnet_hdr should
 * be shared with the tun/tap driver.
 */
568 569
static int macvtap_skb_from_vnet_hdr(struct macvtap_queue *q,
				     struct sk_buff *skb,
570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595
				     struct virtio_net_hdr *vnet_hdr)
{
	unsigned short gso_type = 0;
	if (vnet_hdr->gso_type != VIRTIO_NET_HDR_GSO_NONE) {
		switch (vnet_hdr->gso_type & ~VIRTIO_NET_HDR_GSO_ECN) {
		case VIRTIO_NET_HDR_GSO_TCPV4:
			gso_type = SKB_GSO_TCPV4;
			break;
		case VIRTIO_NET_HDR_GSO_TCPV6:
			gso_type = SKB_GSO_TCPV6;
			break;
		case VIRTIO_NET_HDR_GSO_UDP:
			gso_type = SKB_GSO_UDP;
			break;
		default:
			return -EINVAL;
		}

		if (vnet_hdr->gso_type & VIRTIO_NET_HDR_GSO_ECN)
			gso_type |= SKB_GSO_TCP_ECN;

		if (vnet_hdr->gso_size == 0)
			return -EINVAL;
	}

	if (vnet_hdr->flags & VIRTIO_NET_HDR_F_NEEDS_CSUM) {
596 597
		if (!skb_partial_csum_set(skb, macvtap16_to_cpu(q, vnet_hdr->csum_start),
					  macvtap16_to_cpu(q, vnet_hdr->csum_offset)))
598 599 600 601
			return -EINVAL;
	}

	if (vnet_hdr->gso_type != VIRTIO_NET_HDR_GSO_NONE) {
602
		skb_shinfo(skb)->gso_size = macvtap16_to_cpu(q, vnet_hdr->gso_size);
603
		skb_shinfo(skb)->gso_type = gso_type;
604 605 606 607 608 609 610 611

		/* Header must be checked, and gso_segs computed. */
		skb_shinfo(skb)->gso_type |= SKB_GSO_DODGY;
		skb_shinfo(skb)->gso_segs = 0;
	}
	return 0;
}

612 613 614
static void macvtap_skb_to_vnet_hdr(struct macvtap_queue *q,
				    const struct sk_buff *skb,
				    struct virtio_net_hdr *vnet_hdr)
615 616 617 618 619 620 621
{
	memset(vnet_hdr, 0, sizeof(*vnet_hdr));

	if (skb_is_gso(skb)) {
		struct skb_shared_info *sinfo = skb_shinfo(skb);

		/* This is a hint as to how much should be linear. */
622 623
		vnet_hdr->hdr_len = cpu_to_macvtap16(q, skb_headlen(skb));
		vnet_hdr->gso_size = cpu_to_macvtap16(q, sinfo->gso_size);
624 625 626 627
		if (sinfo->gso_type & SKB_GSO_TCPV4)
			vnet_hdr->gso_type = VIRTIO_NET_HDR_GSO_TCPV4;
		else if (sinfo->gso_type & SKB_GSO_TCPV6)
			vnet_hdr->gso_type = VIRTIO_NET_HDR_GSO_TCPV6;
628 629
		else if (sinfo->gso_type & SKB_GSO_UDP)
			vnet_hdr->gso_type = VIRTIO_NET_HDR_GSO_UDP;
630 631 632 633 634 635 636 637 638
		else
			BUG();
		if (sinfo->gso_type & SKB_GSO_TCP_ECN)
			vnet_hdr->gso_type |= VIRTIO_NET_HDR_GSO_ECN;
	} else
		vnet_hdr->gso_type = VIRTIO_NET_HDR_GSO_NONE;

	if (skb->ip_summed == CHECKSUM_PARTIAL) {
		vnet_hdr->flags = VIRTIO_NET_HDR_F_NEEDS_CSUM;
639
		if (skb_vlan_tag_present(skb))
640 641 642 643 644 645
			vnet_hdr->csum_start = cpu_to_macvtap16(q,
				skb_checksum_start_offset(skb) + VLAN_HLEN);
		else
			vnet_hdr->csum_start = cpu_to_macvtap16(q,
				skb_checksum_start_offset(skb));
		vnet_hdr->csum_offset = cpu_to_macvtap16(q, skb->csum_offset);
646 647
	} else if (skb->ip_summed == CHECKSUM_UNNECESSARY) {
		vnet_hdr->flags = VIRTIO_NET_HDR_F_DATA_VALID;
648 649 650
	} /* else everything is zero */
}

651 652 653
/* Neighbour code has some assumptions on HH_DATA_MOD alignment */
#define MACVTAP_RESERVE HH_DATA_OFF(ETH_HLEN)

Arnd Bergmann's avatar
Arnd Bergmann committed
654
/* Get packet from user space buffer */
655
static ssize_t macvtap_get_user(struct macvtap_queue *q, struct msghdr *m,
656
				struct iov_iter *from, int noblock)
Arnd Bergmann's avatar
Arnd Bergmann committed
657
{
658
	int good_linear = SKB_MAX_HEAD(MACVTAP_RESERVE);
Arnd Bergmann's avatar
Arnd Bergmann committed
659
	struct sk_buff *skb;
660
	struct macvlan_dev *vlan;
661
	unsigned long total_len = iov_iter_count(from);
662
	unsigned long len = total_len;
Arnd Bergmann's avatar
Arnd Bergmann committed
663
	int err;
664 665
	struct virtio_net_hdr vnet_hdr = { 0 };
	int vnet_hdr_len = 0;
666
	int copylen = 0;
667
	bool zerocopy = false;
668
	size_t linear;
669
	ssize_t n;
670 671

	if (q->flags & IFF_VNET_HDR) {
672
		vnet_hdr_len = q->vnet_hdr_sz;
673 674

		err = -EINVAL;
675
		if (len < vnet_hdr_len)
676
			goto err;
677
		len -= vnet_hdr_len;
678

679 680 681
		err = -EFAULT;
		n = copy_from_iter(&vnet_hdr, sizeof(vnet_hdr), from);
		if (n != sizeof(vnet_hdr))
682
			goto err;
683
		iov_iter_advance(from, vnet_hdr_len - sizeof(vnet_hdr));
684
		if ((vnet_hdr.flags & VIRTIO_NET_HDR_F_NEEDS_CSUM) &&
685 686 687 688 689 690
		     macvtap16_to_cpu(q, vnet_hdr.csum_start) +
		     macvtap16_to_cpu(q, vnet_hdr.csum_offset) + 2 >
			     macvtap16_to_cpu(q, vnet_hdr.hdr_len))
			vnet_hdr.hdr_len = cpu_to_macvtap16(q,
				 macvtap16_to_cpu(q, vnet_hdr.csum_start) +
				 macvtap16_to_cpu(q, vnet_hdr.csum_offset) + 2);
691
		err = -EINVAL;
692
		if (macvtap16_to_cpu(q, vnet_hdr.hdr_len) > len)
693 694
			goto err;
	}
Arnd Bergmann's avatar
Arnd Bergmann committed
695

696
	err = -EINVAL;
Arnd Bergmann's avatar
Arnd Bergmann committed
697
	if (unlikely(len < ETH_HLEN))
698
		goto err;
Arnd Bergmann's avatar
Arnd Bergmann committed
699

700
	if (m && m->msg_control && sock_flag(&q->sk, SOCK_ZEROCOPY)) {
701 702
		struct iov_iter i;

703 704
		copylen = vnet_hdr.hdr_len ?
			macvtap16_to_cpu(q, vnet_hdr.hdr_len) : GOODCOPY_LEN;
705 706
		if (copylen > good_linear)
			copylen = good_linear;
707
		linear = copylen;
708 709 710
		i = *from;
		iov_iter_advance(&i, copylen);
		if (iov_iter_npages(&i, INT_MAX) <= MAX_SKB_FRAGS)
711 712 713 714
			zerocopy = true;
	}

	if (!zerocopy) {
715
		copylen = len;
716
		if (macvtap16_to_cpu(q, vnet_hdr.hdr_len) > good_linear)
717 718
			linear = good_linear;
		else
719
			linear = macvtap16_to_cpu(q, vnet_hdr.hdr_len);
720
	}
721

722
	skb = macvtap_alloc_skb(&q->sk, MACVTAP_RESERVE, copylen,
723
				linear, noblock, &err);
724 725
	if (!skb)
		goto err;
Arnd Bergmann's avatar
Arnd Bergmann committed
726

727
	if (zerocopy)
728
		err = zerocopy_sg_from_iter(skb, from);
729
	else {
730
		err = skb_copy_datagram_from_iter(skb, 0, from, len);
731 732 733 734 735 736
		if (!err && m && m->msg_control) {
			struct ubuf_info *uarg = m->msg_control;
			uarg->callback(uarg, false);
		}
	}

737
	if (err)
738
		goto err_kfree;
Arnd Bergmann's avatar
Arnd Bergmann committed
739 740

	skb_set_network_header(skb, ETH_HLEN);
741 742 743 744
	skb_reset_mac_header(skb);
	skb->protocol = eth_hdr(skb)->h_proto;

	if (vnet_hdr_len) {
745
		err = macvtap_skb_from_vnet_hdr(q, skb, &vnet_hdr);
746 747 748 749
		if (err)
			goto err_kfree;
	}

750
	skb_probe_transport_header(skb, ETH_HLEN);
751

752 753
	rcu_read_lock();
	vlan = rcu_dereference(q->vlan);
754
	/* copy skb_ubuf_info for callback when skb has no error */
755
	if (zerocopy) {
756
		skb_shinfo(skb)->destructor_arg = m->msg_control;
757
		skb_shinfo(skb)->tx_flags |= SKBTX_DEV_ZEROCOPY;
758
		skb_shinfo(skb)->tx_flags |= SKBTX_SHARED_FRAG;
759
	}
Eric Dumazet's avatar
Eric Dumazet committed
760
	if (vlan) {
761 762
		skb->dev = vlan->dev;
		dev_queue_xmit(skb);
Eric Dumazet's avatar
Eric Dumazet committed
763
	} else {
764
		kfree_skb(skb);
Eric Dumazet's avatar
Eric Dumazet committed
765
	}
766
	rcu_read_unlock();
Arnd Bergmann's avatar
Arnd Bergmann committed
767

768
	return total_len;
769

770 771 772
err_kfree:
	kfree_skb(skb);

773
err:
774 775
	rcu_read_lock();
	vlan = rcu_dereference(q->vlan);
776
	if (vlan)
777
		this_cpu_inc(vlan->pcpu_stats->tx_dropped);
778
	rcu_read_unlock();
779 780

	return err;
Arnd Bergmann's avatar
Arnd Bergmann committed
781 782
}

783
static ssize_t macvtap_write_iter(struct kiocb *iocb, struct iov_iter *from)
Arnd Bergmann's avatar
Arnd Bergmann committed
784 785
{
	struct file *file = iocb->ki_filp;
786
	struct macvtap_queue *q = file->private_data;
Arnd Bergmann's avatar
Arnd Bergmann committed
787

788
	return macvtap_get_user(q, NULL, from, file->f_flags & O_NONBLOCK);
Arnd Bergmann's avatar
Arnd Bergmann committed
789 790 791 792 793
}

/* Put packet to the user space buffer */
static ssize_t macvtap_put_user(struct macvtap_queue *q,
				const struct sk_buff *skb,
Herbert Xu's avatar
Herbert Xu committed
794
				struct iov_iter *iter)
Arnd Bergmann's avatar
Arnd Bergmann committed
795 796
{
	int ret;
797
	int vnet_hdr_len = 0;
798
	int vlan_offset = 0;
Herbert Xu's avatar
Herbert Xu committed
799
	int total;
800 801 802

	if (q->flags & IFF_VNET_HDR) {
		struct virtio_net_hdr vnet_hdr;
803
		vnet_hdr_len = q->vnet_hdr_sz;
Herbert Xu's avatar
Herbert Xu committed
804
		if (iov_iter_count(iter) < vnet_hdr_len)
805 806
			return -EINVAL;

807
		macvtap_skb_to_vnet_hdr(q, skb, &vnet_hdr);
808

Herbert Xu's avatar
Herbert Xu committed
809 810
		if (copy_to_iter(&vnet_hdr, sizeof(vnet_hdr), iter) !=
		    sizeof(vnet_hdr))
811
			return -EFAULT;
812 813

		iov_iter_advance(iter, vnet_hdr_len - sizeof(vnet_hdr));
814
	}
Herbert Xu's avatar
Herbert Xu committed
815
	total = vnet_hdr_len;
816
	total += skb->len;
817

818
	if (skb_vlan_tag_present(skb)) {
819 820 821 822
		struct {
			__be16 h_vlan_proto;
			__be16 h_vlan_TCI;
		} veth;
823
		veth.h_vlan_proto = skb->vlan_proto;
824
		veth.h_vlan_TCI = htons(skb_vlan_tag_get(skb));
825 826

		vlan_offset = offsetof(struct vlan_ethhdr, h_vlan_proto);
827
		total += VLAN_HLEN;
828

Herbert Xu's avatar
Herbert Xu committed
829 830
		ret = skb_copy_datagram_iter(skb, 0, iter, vlan_offset);
		if (ret || !iov_iter_count(iter))
831 832
			goto done;

Herbert Xu's avatar
Herbert Xu committed
833 834
		ret = copy_to_iter(&veth, sizeof(veth), iter);
		if (ret != sizeof(veth) || !iov_iter_count(iter))
835 836
			goto done;
	}
Arnd Bergmann's avatar
Arnd Bergmann committed
837

Herbert Xu's avatar
Herbert Xu committed
838 839
	ret = skb_copy_datagram_iter(skb, vlan_offset, iter,
				     skb->len - vlan_offset);
Arnd Bergmann's avatar
Arnd Bergmann committed
840

841
done:
842
	return ret ? ret : total;
Arnd Bergmann's avatar
Arnd Bergmann committed
843 844
}

845
static ssize_t macvtap_do_read(struct macvtap_queue *q,
Al Viro's avatar
Al Viro committed
846
			       struct iov_iter *to,
847
			       int noblock)
Arnd Bergmann's avatar
Arnd Bergmann committed
848
{
849
	DEFINE_WAIT(wait);
Arnd Bergmann's avatar
Arnd Bergmann committed
850
	struct sk_buff *skb;
851
	ssize_t ret = 0;
Arnd Bergmann's avatar
Arnd Bergmann committed
852

Al Viro's avatar
Al Viro committed
853 854 855 856
	if (!iov_iter_count(to))
		return 0;

	while (1) {
857 858 859
		if (!noblock)
			prepare_to_wait(sk_sleep(&q->sk), &wait,
					TASK_INTERRUPTIBLE);
Arnd Bergmann's avatar
Arnd Bergmann committed
860 861 862

		/* Read frames from the queue */
		skb = skb_dequeue(&q->sk.sk_receive_queue);
Al Viro's avatar
Al Viro committed
863 864 865 866 867
		if (skb)
			break;
		if (noblock) {
			ret = -EAGAIN;
			break;
Arnd Bergmann's avatar
Arnd Bergmann committed
868
		}
Al Viro's avatar
Al Viro committed
869 870 871 872 873 874 875 876 877
		if (signal_pending(current)) {
			ret = -ERESTARTSYS;
			break;
		}
		/* Nothing to read, let's sleep */
		schedule();
	}
	if (skb) {
		ret = macvtap_put_user(q, skb, to);
878 879 880 881
		if (unlikely(ret < 0))
			kfree_skb(skb);
		else
			consume_skb(skb);
Arnd Bergmann's avatar
Arnd Bergmann committed
882
	}
883 884
	if (!noblock)
		finish_wait(sk_sleep(&q->sk), &wait);
885 886 887
	return ret;
}

Al Viro's avatar
Al Viro committed
888
static ssize_t macvtap_read_iter(struct kiocb *iocb, struct iov_iter *to)
889 890 891
{
	struct file *file = iocb->ki_filp;
	struct macvtap_queue *q = file->private_data;
Al Viro's avatar
Al Viro committed
892
	ssize_t len = iov_iter_count(to), ret;
Arnd Bergmann's avatar
Arnd Bergmann committed
893

Al Viro's avatar
Al Viro committed
894
	ret = macvtap_do_read(q, to, file->f_flags & O_NONBLOCK);
895
	ret = min_t(ssize_t, ret, len);
896 897
	if (ret > 0)
		iocb->ki_pos = ret;
Arnd Bergmann's avatar
Arnd Bergmann committed
898 899 900
	return ret;
}

901 902 903 904
static struct macvlan_dev *macvtap_get_vlan(struct macvtap_queue *q)
{
	struct macvlan_dev *vlan;

905 906
	ASSERT_RTNL();
	vlan = rtnl_dereference(q->vlan);
907 908 909 910 911 912 913 914 915 916 917
	if (vlan)
		dev_hold(vlan->dev);

	return vlan;
}

static void macvtap_put_vlan(struct macvlan_dev *vlan)
{
	dev_put(vlan->dev);
}

Jason Wang's avatar
Jason Wang committed
918 919 920 921 922 923 924 925 926 927 928 929 930 931
static int macvtap_ioctl_set_queue(struct file *file, unsigned int flags)
{
	struct macvtap_queue *q = file->private_data;
	struct macvlan_dev *vlan;
	int ret;

	vlan = macvtap_get_vlan(q);
	if (!vlan)
		return -EINVAL;

	if (flags & IFF_ATTACH_QUEUE)
		ret = macvtap_enable_queue(vlan->dev, file, q);
	else if (flags & IFF_DETACH_QUEUE)
		ret = macvtap_disable_queue(q);
932 933
	else
		ret = -EINVAL;
Jason Wang's avatar
Jason Wang committed
934 935 936 937 938

	macvtap_put_vlan(vlan);
	return ret;
}

939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961
static int set_offload(struct macvtap_queue *q, unsigned long arg)
{
	struct macvlan_dev *vlan;
	netdev_features_t features;
	netdev_features_t feature_mask = 0;

	vlan = rtnl_dereference(q->vlan);
	if (!vlan)
		return -ENOLINK;

	features = vlan->dev->features;

	if (arg & TUN_F_CSUM) {
		feature_mask = NETIF_F_HW_CSUM;

		if (arg & (TUN_F_TSO4 | TUN_F_TSO6)) {
			if (arg & TUN_F_TSO_ECN)
				feature_mask |= NETIF_F_TSO_ECN;
			if (arg & TUN_F_TSO4)
				feature_mask |= NETIF_F_TSO;
			if (arg & TUN_F_TSO6)
				feature_mask |= NETIF_F_TSO6;
		}
962 963 964

		if (arg & TUN_F_UFO)
			feature_mask |= NETIF_F_UFO;
965 966 967 968 969 970 971 972 973 974
	}

	/* tun/tap driver inverts the usage for TSO offloads, where
	 * setting the TSO bit means that the userspace wants to
	 * accept TSO frames and turning it off means that user space
	 * does not support TSO.
	 * For macvtap, we have to invert it to mean the same thing.
	 * When user space turns off TSO, we turn off GSO/LRO so that
	 * user-space will not receive TSO frames.
	 */
975
	if (feature_mask & (NETIF_F_TSO | NETIF_F_TSO6 | NETIF_F_UFO))
976 977 978 979 980 981 982
		features |= RX_OFFLOADS;
	else
		features &= ~RX_OFFLOADS;

	/* tap_features are the same as features on tun/tap and
	 * reflect user expectations.
	 */
983
	vlan->tap_features = feature_mask;
984 985 986 987 988 989
	vlan->set_features = features;
	netdev_update_features(vlan->dev);

	return 0;
}

Arnd Bergmann's avatar
Arnd Bergmann committed
990 991 992 993 994 995
/*
 * provide compatibility with generic tun/tap interface
 */
static long macvtap_ioctl(struct file *file, unsigned int cmd,
			  unsigned long arg)
{
996 997
	struct macvtap_queue *q = file->private_data;
	struct macvlan_dev *vlan;
Arnd Bergmann's avatar
Arnd Bergmann committed
998 999 1000
	void __user *argp = (void __user *)arg;
	struct ifreq __user *ifr = argp;
	unsigned int __user *up = argp;
1001
	unsigned short u;
1002
	int __user *sp = argp;
1003
	struct sockaddr sa;
1004
	int s;
1005
	int ret;
Arnd Bergmann's avatar
Arnd Bergmann committed
1006 1007 1008 1009 1010 1011

	switch (cmd) {
	case TUNSETIFF:
		/* ignore the name, just look at flags */
		if (get_user(u, &ifr->ifr_flags))
			return -EFAULT;
1012 1013

		ret = 0;
1014
		if ((u & ~MACVTAP_FEATURES) != (IFF_NO_PI | IFF_TAP))
1015 1016
			ret = -EINVAL;
		else
1017
			q->flags = (q->flags & ~MACVTAP_FEATURES) | u;
1018 1019

		return ret;
Arnd Bergmann's avatar
Arnd Bergmann committed
1020 1021

	case TUNGETIFF:
1022
		rtnl_lock();
1023
		vlan = macvtap_get_vlan(q);
1024 1025
		if (!vlan) {
			rtnl_unlock();
Arnd Bergmann's avatar
Arnd Bergmann committed
1026
			return -ENOLINK;
1027
		}
Arnd Bergmann's avatar
Arnd Bergmann committed
1028

1029
		ret = 0;
1030
		u = q->flags;
1031
		if (copy_to_user(&ifr->ifr_name, vlan->dev->name, IFNAMSIZ) ||
1032
		    put_user(u, &ifr->ifr_flags))
1033
			ret = -EFAULT;
1034
		macvtap_put_vlan(vlan);
1035
		rtnl_unlock();
1036
		return ret;
Arnd Bergmann's avatar
Arnd Bergmann committed
1037

Jason Wang's avatar
Jason Wang committed
1038 1039 1040
	case TUNSETQUEUE:
		if (get_user(u, &ifr->ifr_flags))
			return -EFAULT;
1041 1042 1043
		rtnl_lock();
		ret = macvtap_ioctl_set_queue(file, u);
		rtnl_unlock();
1044
		return ret;
Jason Wang's avatar
Jason Wang committed
1045

Arnd Bergmann's avatar
Arnd Bergmann committed
1046
	case TUNGETFEATURES:
1047
		if (put_user(IFF_TAP | IFF_NO_PI | MACVTAP_FEATURES, up))
Arnd Bergmann's avatar
Arnd Bergmann committed
1048 1049 1050 1051 1052 1053 1054 1055 1056 1057
			return -EFAULT;
		return 0;

	case TUNSETSNDBUF:
		if (get_user(u, up))
			return -EFAULT;

		q->sk.sk_sndbuf = u;
		return 0;

1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072
	case TUNGETVNETHDRSZ:
		s = q->vnet_hdr_sz;
		if (put_user(s, sp))
			return -EFAULT;
		return 0;

	case TUNSETVNETHDRSZ:
		if (get_user(s, sp))
			return -EFAULT;
		if (s < (int)sizeof(struct virtio_net_hdr))
			return -EINVAL;

		q->vnet_hdr_sz = s;
		return 0;

1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087
	case TUNGETVNETLE:
		s = !!(q->flags & MACVTAP_VNET_LE);
		if (put_user(s, sp))
			return -EFAULT;
		return 0;

	case TUNSETVNETLE:
		if (get_user(s, sp))
			return -EFAULT;
		if (s)
			q->flags |= MACVTAP_VNET_LE;
		else
			q->flags &= ~MACVTAP_VNET_LE;
		return 0;

Arnd Bergmann's avatar
Arnd Bergmann committed
1088 1089 1090
	case TUNSETOFFLOAD:
		/* let the user check for future flags */
		if (arg & ~(TUN_F_CSUM | TUN_F_TSO4 | TUN_F_TSO6 |
1091
			    TUN_F_TSO_ECN | TUN_F_UFO))
Arnd Bergmann's avatar
Arnd Bergmann committed
1092 1093
			return -EINVAL;

1094 1095 1096 1097
		rtnl_lock();
		ret = set_offload(q, arg);
		rtnl_unlock();
		return ret;
Arnd Bergmann's avatar
Arnd Bergmann committed
1098

1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116
	case SIOCGIFHWADDR:
		rtnl_lock();
		vlan = macvtap_get_vlan(q);
		if (!vlan) {
			rtnl_unlock();
			return -ENOLINK;
		}
		ret = 0;
		u = vlan->dev->type;
		if (copy_to_user(&ifr->ifr_name, vlan->dev->name, IFNAMSIZ) ||
		    copy_to_user(&ifr->ifr_hwaddr.sa_data, vlan->dev->dev_addr, ETH_ALEN) ||
		    put_user(u, &ifr->ifr_hwaddr.sa_family))
			ret = -EFAULT;
		macvtap_put_vlan(vlan);
		rtnl_unlock();
		return ret;

	case SIOCSIFHWADDR:
1117 1118
		if (copy_from_user(&sa, &ifr->ifr_hwaddr, sizeof(sa)))
			return -EFAULT;
1119 1120 1121 1122 1123 1124
		rtnl_lock();
		vlan = macvtap_get_vlan(q);
		if (!vlan) {
			rtnl_unlock();
			return -ENOLINK;
		}
1125
		ret = dev_set_mac_address(vlan->dev, &sa);
1126 1127 1128 1129
		macvtap_put_vlan(vlan);
		rtnl_unlock();
		return ret;

Arnd Bergmann's avatar
Arnd Bergmann committed
1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146
	default:
		return -EINVAL;
	}
}

#ifdef CONFIG_COMPAT
static long macvtap_compat_ioctl(struct file *file, unsigned int cmd,
				 unsigned long arg)
{
	return macvtap_ioctl(file, cmd, (unsigned long)compat_ptr(arg));
}
#endif

static const struct file_operations macvtap_fops = {
	.owner		= THIS_MODULE,
	.open		= macvtap_open,
	.release	= macvtap_release,
Al Viro's avatar
Al Viro committed
1147
	.read_iter	= macvtap_read_iter,
1148
	.write_iter	= macvtap_write_iter,
Arnd Bergmann's avatar
Arnd Bergmann committed
1149 1150 1151 1152 1153 1154 1155 1156
	.poll		= macvtap_poll,
	.llseek		= no_llseek,
	.unlocked_ioctl	= macvtap_ioctl,
#ifdef CONFIG_COMPAT
	.compat_ioctl	= macvtap_compat_ioctl,
#endif
};

1157 1158
static int macvtap_sendmsg(struct socket *sock, struct msghdr *m,
			   size_t total_len)
1159 1160
{
	struct macvtap_queue *q = container_of(sock, struct macvtap_queue, sock);
Al Viro's avatar
Al Viro committed
1161
	return macvtap_get_user(q, m, &m->msg_iter, m->msg_flags & MSG_DONTWAIT);
1162 1163
}

1164 1165
static int macvtap_recvmsg(struct socket *sock, struct msghdr *m,
			   size_t total_len, int flags)
1166 1167 1168 1169 1170
{
	struct macvtap_queue *q = container_of(sock, struct macvtap_queue, sock);
	int ret;
	if (flags & ~(MSG_DONTWAIT|MSG_TRUNC))
		return -EINVAL;
Al Viro's avatar
Al Viro committed
1171
	ret = macvtap_do_read(q, &m->msg_iter, flags & MSG_DONTWAIT);
1172 1173 1174 1175
	if (ret > total_len) {
		m->msg_flags |= MSG_TRUNC;
		ret = flags & MSG_TRUNC ? ret : total_len;
	}
1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200
	return ret;
}

/* Ops structure to mimic raw sockets with tun */
static const struct proto_ops macvtap_socket_ops = {
	.sendmsg = macvtap_sendmsg,
	.recvmsg = macvtap_recvmsg,
};

/* Get an underlying socket object from tun file.  Returns error unless file is
 * attached to a device.  The returned object works like a packet socket, it
 * can be used for sock_sendmsg/sock_recvmsg.  The caller is responsible for
 * holding a reference to the file for as long as the socket is in use. */
struct socket *macvtap_get_socket(struct file *file)
{
	struct macvtap_queue *q;
	if (file->f_op != &macvtap_fops)
		return ERR_PTR(-EINVAL);
	q = file->private_data;
	if (!q)
		return ERR_PTR(-EBADFD);
	return &q->sock;
}
EXPORT_SYMBOL_GPL(macvtap_get_socket);

1201 1202 1203
static int macvtap_device_event(struct notifier_block *unused,
				unsigned long event, void *ptr)
{
1204
	struct net_device *dev = netdev_notifier_info_to_dev(ptr);
1205
	struct macvlan_dev *vlan;
1206 1207
	struct device *classdev;
	dev_t devt;
1208
	int err;
1209 1210 1211 1212

	if (dev->rtnl_link_ops != &macvtap_link_ops)
		return NOTIFY_DONE;

1213
	vlan = netdev_priv(dev);
1214 1215 1216 1217 1218 1219 1220

	switch (event) {
	case NETDEV_REGISTER:
		/* Create the device node here after the network device has
		 * been registered but before register_netdevice has
		 * finished running.
		 */
1221 1222 1223 1224 1225
		err = macvtap_get_minor(vlan);
		if (err)
			return notifier_from_errno(err);

		devt = MKDEV(MAJOR(macvtap_major), vlan->minor);
1226 1227
		classdev = device_create(macvtap_class, &dev->dev, devt,
					 dev, "tap%d", dev->ifindex);
1228 1229
		if (IS_ERR(classdev)) {
			macvtap_free_minor(vlan);
1230
			return notifier_from_errno(PTR_ERR(classdev));
1231
		}
1232 1233
		break;
	case NETDEV_UNREGISTER:
1234
		devt = MKDEV(MAJOR(macvtap_major), vlan->minor);
1235
		device_destroy(macvtap_class, devt);
1236
		macvtap_free_minor(vlan);
1237 1238 1239 1240 1241 1242 1243 1244 1245 1246
		break;
	}

	return NOTIFY_DONE;
}

static struct notifier_block macvtap_notifier_block __read_mostly = {
	.notifier_call	= macvtap_device_event,
};

Arnd Bergmann's avatar
Arnd Bergmann committed
1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266
static int macvtap_init(void)
{
	int err;

	err = alloc_chrdev_region(&macvtap_major, 0,
				MACVTAP_NUM_DEVS, "macvtap");
	if (err)
		goto out1;

	cdev_init(&macvtap_cdev, &macvtap_fops);
	err = cdev_add(&macvtap_cdev, macvtap_major, MACVTAP_NUM_DEVS);
	if (err)
		goto out2;

	macvtap_class = class_create(THIS_MODULE, "macvtap");
	if (IS_ERR(macvtap_class)) {
		err = PTR_ERR(macvtap_class);
		goto out3;
	}

1267
	err = register_netdevice_notifier(&macvtap_notifier_block);
Arnd Bergmann's avatar
Arnd Bergmann committed
1268 1269 1270
	if (err)
		goto out4;

1271 1272 1273 1274
	err = macvlan_link_register(&macvtap_link_ops);
	if (err)
		goto out5;

Arnd Bergmann's avatar
Arnd Bergmann committed
1275 1276
	return 0;

1277 1278
out5:
	unregister_netdevice_notifier(&macvtap_notifier_block);
Arnd Bergmann's avatar
Arnd Bergmann committed
1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292
out4:
	class_unregister(macvtap_class);
out3:
	cdev_del(&macvtap_cdev);
out2:
	unregister_chrdev_region(macvtap_major, MACVTAP_NUM_DEVS);
out1:
	return err;
}
module_init(macvtap_init);

static void macvtap_exit(void)
{
	rtnl_link_unregister(&macvtap_link_ops);
1293
	unregister_netdevice_notifier(&macvtap_notifier_block);
Arnd Bergmann's avatar
Arnd Bergmann committed
1294 1295 1296
	class_unregister(macvtap_class);
	cdev_del(&macvtap_cdev);
	unregister_chrdev_region(macvtap_major, MACVTAP_NUM_DEVS);
1297
	idr_destroy(&minor_idr);
Arnd Bergmann's avatar
Arnd Bergmann committed
1298 1299 1300 1301 1302 1303
}
module_exit(macvtap_exit);

MODULE_ALIAS_RTNL_LINK("macvtap");
MODULE_AUTHOR("Arnd Bergmann <arnd@arndb.de>");
MODULE_LICENSE("GPL");