af_packet.c 103.6 KB
Newer Older
L
Linus Torvalds 已提交
1 2 3 4 5 6 7
/*
 * INET		An implementation of the TCP/IP protocol suite for the LINUX
 *		operating system.  INET is implemented using the  BSD Socket
 *		interface as the means of communication with the user level.
 *
 *		PACKET - implements raw packet sockets.
 *
8
 * Authors:	Ross Biro
L
Linus Torvalds 已提交
9 10 11
 *		Fred N. van Kempen, <waltje@uWalt.NL.Mugnet.ORG>
 *		Alan Cox, <gw4pts@gw4pts.ampr.org>
 *
12
 * Fixes:
L
Linus Torvalds 已提交
13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34
 *		Alan Cox	:	verify_area() now used correctly
 *		Alan Cox	:	new skbuff lists, look ma no backlogs!
 *		Alan Cox	:	tidied skbuff lists.
 *		Alan Cox	:	Now uses generic datagram routines I
 *					added. Also fixed the peek/read crash
 *					from all old Linux datagram code.
 *		Alan Cox	:	Uses the improved datagram code.
 *		Alan Cox	:	Added NULL's for socket options.
 *		Alan Cox	:	Re-commented the code.
 *		Alan Cox	:	Use new kernel side addressing
 *		Rob Janssen	:	Correct MTU usage.
 *		Dave Platt	:	Counter leaks caused by incorrect
 *					interrupt locking and some slightly
 *					dubious gcc output. Can you read
 *					compiler: it said _VOLATILE_
 *	Richard Kooijman	:	Timestamp fixes.
 *		Alan Cox	:	New buffers. Use sk->mac.raw.
 *		Alan Cox	:	sendmsg/recvmsg support.
 *		Alan Cox	:	Protocol setting support
 *	Alexey Kuznetsov	:	Untied from IPv4 stack.
 *	Cyrus Durgin		:	Fixed kerneld for kmod.
 *	Michal Ostrowski        :       Module initialization cleanup.
35
 *         Ulises Alonso        :       Frame number limit removal and
L
Linus Torvalds 已提交
36
 *                                      packet_set_ring memory leak.
37 38 39
 *		Eric Biederman	:	Allow for > 8 byte hardware addresses.
 *					The convention is that longer addresses
 *					will simply extend the hardware address
40
 *					byte arrays at the end of sockaddr_ll
41
 *					and packet_mreq.
J
Johann Baudy 已提交
42
 *		Johann Baudy	:	Added TX RING.
43 44 45 46
 *		Chetan Loke	:	Implemented TPACKET_V3 block abstraction
 *					layer.
 *					Copyright (C) 2011, <lokec@ccs.neu.edu>
 *
L
Linus Torvalds 已提交
47 48 49 50 51 52 53
 *
 *		This program is free software; you can redistribute it and/or
 *		modify it under the terms of the GNU General Public License
 *		as published by the Free Software Foundation; either version
 *		2 of the License, or (at your option) any later version.
 *
 */
54

L
Linus Torvalds 已提交
55 56
#include <linux/types.h>
#include <linux/mm.h>
57
#include <linux/capability.h>
L
Linus Torvalds 已提交
58 59 60 61 62 63 64
#include <linux/fcntl.h>
#include <linux/socket.h>
#include <linux/in.h>
#include <linux/inet.h>
#include <linux/netdevice.h>
#include <linux/if_packet.h>
#include <linux/wireless.h>
65
#include <linux/kernel.h>
L
Linus Torvalds 已提交
66
#include <linux/kmod.h>
67
#include <linux/slab.h>
68
#include <linux/vmalloc.h>
69
#include <net/net_namespace.h>
L
Linus Torvalds 已提交
70 71 72 73 74 75 76 77 78
#include <net/ip.h>
#include <net/protocol.h>
#include <linux/skbuff.h>
#include <net/sock.h>
#include <linux/errno.h>
#include <linux/timer.h>
#include <asm/uaccess.h>
#include <asm/ioctls.h>
#include <asm/page.h>
A
Al Viro 已提交
79
#include <asm/cacheflush.h>
L
Linus Torvalds 已提交
80 81 82 83 84 85
#include <asm/io.h>
#include <linux/proc_fs.h>
#include <linux/seq_file.h>
#include <linux/poll.h>
#include <linux/module.h>
#include <linux/init.h>
86
#include <linux/mutex.h>
E
Eric Dumazet 已提交
87
#include <linux/if_vlan.h>
88
#include <linux/virtio_net.h>
89
#include <linux/errqueue.h>
90
#include <linux/net_tstamp.h>
91
#include <linux/percpu.h>
L
Linus Torvalds 已提交
92 93 94
#ifdef CONFIG_INET
#include <net/inet_common.h>
#endif
95
#include <linux/bpf.h>
L
Linus Torvalds 已提交
96

97 98
#include "internal.h"

L
Linus Torvalds 已提交
99 100 101 102 103 104 105 106 107 108 109 110 111 112 113
/*
   Assumptions:
   - if device has no dev->hard_header routine, it adds and removes ll header
     inside itself. In this case ll header is invisible outside of device,
     but higher levels still should reserve dev->hard_header_len.
     Some devices are enough clever to reallocate skb, when header
     will not fit to reserved space (tunnel), another ones are silly
     (PPP).
   - packet socket receives packets with pulled ll header,
     so that SOCK_RAW should push it back.

On receive:
-----------

Incoming, dev->hard_header!=NULL
114 115
   mac_header -> ll header
   data       -> data
L
Linus Torvalds 已提交
116 117

Outgoing, dev->hard_header!=NULL
118 119
   mac_header -> ll header
   data       -> ll header
L
Linus Torvalds 已提交
120 121

Incoming, dev->hard_header==NULL
122 123
   mac_header -> UNKNOWN position. It is very likely, that it points to ll
		 header.  PPP makes it, that is wrong, because introduce
124
		 assymetry between rx and tx paths.
125
   data       -> data
L
Linus Torvalds 已提交
126 127

Outgoing, dev->hard_header==NULL
128 129
   mac_header -> data. ll header is still not built!
   data       -> data
L
Linus Torvalds 已提交
130 131 132 133 134 135 136 137 138

Resume
  If dev->hard_header==NULL we are unlikely to restore sensible ll header.


On transmit:
------------

dev->hard_header != NULL
139 140
   mac_header -> ll header
   data       -> ll header
L
Linus Torvalds 已提交
141 142

dev->hard_header == NULL (ll header is added by device, we cannot control it)
143 144
   mac_header -> data
   data       -> data
L
Linus Torvalds 已提交
145 146 147 148 149 150 151

   We should set nh.raw on output to correct posistion,
   packet classifier depends on it.
 */

/* Private packet socket structures. */

152 153 154
/* identical to struct packet_mreq except it has
 * a longer address field.
 */
E
Eric Dumazet 已提交
155
struct packet_mreq_max {
156 157 158 159
	int		mr_ifindex;
	unsigned short	mr_type;
	unsigned short	mr_alen;
	unsigned char	mr_address[MAX_ADDR_LEN];
L
Linus Torvalds 已提交
160
};
161

162 163 164 165 166 167 168
union tpacket_uhdr {
	struct tpacket_hdr  *h1;
	struct tpacket2_hdr *h2;
	struct tpacket3_hdr *h3;
	void *raw;
};

169
static int packet_set_ring(struct sock *sk, union tpacket_req_u *req_u,
J
Johann Baudy 已提交
170 171
		int closing, int tx_ring);

172 173
#define V3_ALIGNMENT	(8)

174
#define BLK_HDR_LEN	(ALIGN(sizeof(struct tpacket_block_desc), V3_ALIGNMENT))
175 176 177 178 179

#define BLK_PLUS_PRIV(sz_of_priv) \
	(BLK_HDR_LEN + ALIGN((sz_of_priv), V3_ALIGNMENT))

#define PGV_FROM_VMALLOC 1
J
Johann Baudy 已提交
180

181 182 183 184 185 186 187 188
#define BLOCK_STATUS(x)	((x)->hdr.bh1.block_status)
#define BLOCK_NUM_PKTS(x)	((x)->hdr.bh1.num_pkts)
#define BLOCK_O2FP(x)		((x)->hdr.bh1.offset_to_first_pkt)
#define BLOCK_LEN(x)		((x)->hdr.bh1.blk_len)
#define BLOCK_SNUM(x)		((x)->hdr.bh1.seq_num)
#define BLOCK_O2PRIV(x)	((x)->offset_to_priv)
#define BLOCK_PRIV(x)		((void *)((char *)(x) + BLOCK_O2PRIV(x)))

J
Johann Baudy 已提交
189 190
struct packet_sock;
static int tpacket_snd(struct packet_sock *po, struct msghdr *msg);
191 192
static int tpacket_rcv(struct sk_buff *skb, struct net_device *dev,
		       struct packet_type *pt, struct net_device *orig_dev);
L
Linus Torvalds 已提交
193

194 195 196 197
static void *packet_previous_frame(struct packet_sock *po,
		struct packet_ring_buffer *rb,
		int status);
static void packet_increment_head(struct packet_ring_buffer *buff);
198 199 200
static int prb_curr_blk_in_use(struct tpacket_kbdq_core *,
			struct tpacket_block_desc *);
static void *prb_dispatch_next_block(struct tpacket_kbdq_core *,
201
			struct packet_sock *);
202
static void prb_retire_current_block(struct tpacket_kbdq_core *,
203
		struct packet_sock *, unsigned int status);
204 205 206
static int prb_queue_frozen(struct tpacket_kbdq_core *);
static void prb_open_block(struct tpacket_kbdq_core *,
		struct tpacket_block_desc *);
207
static void prb_retire_rx_blk_timer_expired(unsigned long);
208 209 210 211 212 213 214 215 216
static void _prb_refresh_rx_retire_blk_timer(struct tpacket_kbdq_core *);
static void prb_init_blk_timer(struct packet_sock *,
		struct tpacket_kbdq_core *,
		void (*func) (unsigned long));
static void prb_fill_rxhash(struct tpacket_kbdq_core *, struct tpacket3_hdr *);
static void prb_clear_rxhash(struct tpacket_kbdq_core *,
		struct tpacket3_hdr *);
static void prb_fill_vlan_info(struct tpacket_kbdq_core *,
		struct tpacket3_hdr *);
L
Linus Torvalds 已提交
217 218
static void packet_flush_mclist(struct sock *sk);

219 220 221
struct packet_skb_cb {
	union {
		struct sockaddr_pkt pkt;
222 223 224 225 226 227 228 229
		union {
			/* Trick: alias skb original length with
			 * ll.sll_family and ll.protocol in order
			 * to save room.
			 */
			unsigned int origlen;
			struct sockaddr_ll ll;
		};
230 231 232
	} sa;
};

233 234
#define vio_le() virtio_legacy_is_little_endian()

235
#define PACKET_SKB_CB(__skb)	((struct packet_skb_cb *)((__skb)->cb))
236

237
#define GET_PBDQC_FROM_RB(x)	((struct tpacket_kbdq_core *)(&(x)->prb_bdqc))
238
#define GET_PBLOCK_DESC(x, bid)	\
239
	((struct tpacket_block_desc *)((x)->pkbdq[(bid)].buffer))
240
#define GET_CURR_PBLOCK_DESC_FROM_CORE(x)	\
241
	((struct tpacket_block_desc *)((x)->pkbdq[(x)->kactive_blk_num].buffer))
242 243 244 245
#define GET_NEXT_PRB_BLK_NUM(x) \
	(((x)->kactive_blk_num < ((x)->knum_blocks-1)) ? \
	((x)->kactive_blk_num+1) : 0)

D
David S. Miller 已提交
246 247 248
static void __fanout_unlink(struct sock *sk, struct packet_sock *po);
static void __fanout_link(struct sock *sk, struct packet_sock *po);

249 250 251 252 253
static int packet_direct_xmit(struct sk_buff *skb)
{
	struct net_device *dev = skb->dev;
	netdev_features_t features;
	struct netdev_queue *txq;
254
	int ret = NETDEV_TX_BUSY;
255 256

	if (unlikely(!netif_running(dev) ||
257 258
		     !netif_carrier_ok(dev)))
		goto drop;
259 260 261

	features = netif_skb_features(skb);
	if (skb_needs_linearize(skb, features) &&
262 263
	    __skb_linearize(skb))
		goto drop;
264

265
	txq = skb_get_tx_queue(dev, skb);
266

267 268 269
	local_bh_disable();

	HARD_TX_LOCK(dev, txq, smp_processor_id());
270
	if (!netif_xmit_frozen_or_drv_stopped(txq))
271
		ret = netdev_start_xmit(skb, dev, txq, false);
272
	HARD_TX_UNLOCK(dev, txq);
273

274 275 276
	local_bh_enable();

	if (!dev_xmit_complete(ret))
277
		kfree_skb(skb);
278

279
	return ret;
280
drop:
281
	atomic_long_inc(&dev->tx_dropped);
282 283
	kfree_skb(skb);
	return NET_XMIT_DROP;
284 285
}

286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309
static struct net_device *packet_cached_dev_get(struct packet_sock *po)
{
	struct net_device *dev;

	rcu_read_lock();
	dev = rcu_dereference(po->cached_dev);
	if (likely(dev))
		dev_hold(dev);
	rcu_read_unlock();

	return dev;
}

static void packet_cached_dev_assign(struct packet_sock *po,
				     struct net_device *dev)
{
	rcu_assign_pointer(po->cached_dev, dev);
}

static void packet_cached_dev_reset(struct packet_sock *po)
{
	RCU_INIT_POINTER(po->cached_dev, NULL);
}

310 311 312 313 314
static bool packet_use_direct_xmit(const struct packet_sock *po)
{
	return po->xmit == packet_direct_xmit;
}

315
static u16 __packet_pick_tx_queue(struct net_device *dev, struct sk_buff *skb)
316
{
317
	return (u16) raw_smp_processor_id() % dev->real_num_tx_queues;
318 319
}

320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335
static void packet_pick_tx_queue(struct net_device *dev, struct sk_buff *skb)
{
	const struct net_device_ops *ops = dev->netdev_ops;
	u16 queue_index;

	if (ops->ndo_select_queue) {
		queue_index = ops->ndo_select_queue(dev, skb, NULL,
						    __packet_pick_tx_queue);
		queue_index = netdev_cap_txqueue(dev, queue_index);
	} else {
		queue_index = __packet_pick_tx_queue(dev, skb);
	}

	skb_set_queue_mapping(skb, queue_index);
}

336 337 338 339 340 341 342
/* register_prot_hook must be invoked with the po->bind_lock held,
 * or from a context in which asynchronous accesses to the packet
 * socket is not possible (packet_create()).
 */
static void register_prot_hook(struct sock *sk)
{
	struct packet_sock *po = pkt_sk(sk);
343

344
	if (!po->running) {
345
		if (po->fanout)
D
David S. Miller 已提交
346
			__fanout_link(sk, po);
347
		else
D
David S. Miller 已提交
348
			dev_add_pack(&po->prot_hook);
349

350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366
		sock_hold(sk);
		po->running = 1;
	}
}

/* {,__}unregister_prot_hook() must be invoked with the po->bind_lock
 * held.   If the sync parameter is true, we will temporarily drop
 * the po->bind_lock and do a synchronize_net to make sure no
 * asynchronous packet processing paths still refer to the elements
 * of po->prot_hook.  If the sync parameter is false, it is the
 * callers responsibility to take care of this.
 */
static void __unregister_prot_hook(struct sock *sk, bool sync)
{
	struct packet_sock *po = pkt_sk(sk);

	po->running = 0;
367 368

	if (po->fanout)
D
David S. Miller 已提交
369
		__fanout_unlink(sk, po);
370
	else
D
David S. Miller 已提交
371
		__dev_remove_pack(&po->prot_hook);
372

373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389
	__sock_put(sk);

	if (sync) {
		spin_unlock(&po->bind_lock);
		synchronize_net();
		spin_lock(&po->bind_lock);
	}
}

static void unregister_prot_hook(struct sock *sk, bool sync)
{
	struct packet_sock *po = pkt_sk(sk);

	if (po->running)
		__unregister_prot_hook(sk, sync);
}

M
Michael S. Tsirkin 已提交
390
static inline struct page * __pure pgv_to_page(void *addr)
391 392 393 394 395 396
{
	if (is_vmalloc_addr(addr))
		return vmalloc_to_page(addr);
	return virt_to_page(addr);
}

J
Johann Baudy 已提交
397
static void __packet_set_status(struct packet_sock *po, void *frame, int status)
L
Linus Torvalds 已提交
398
{
399
	union tpacket_uhdr h;
L
Linus Torvalds 已提交
400

J
Johann Baudy 已提交
401
	h.raw = frame;
402 403
	switch (po->tp_version) {
	case TPACKET_V1:
J
Johann Baudy 已提交
404
		h.h1->tp_status = status;
405
		flush_dcache_page(pgv_to_page(&h.h1->tp_status));
406 407
		break;
	case TPACKET_V2:
J
Johann Baudy 已提交
408
		h.h2->tp_status = status;
409
		flush_dcache_page(pgv_to_page(&h.h2->tp_status));
410
		break;
411
	case TPACKET_V3:
J
Johann Baudy 已提交
412
	default:
413
		WARN(1, "TPACKET version not supported.\n");
J
Johann Baudy 已提交
414
		BUG();
415
	}
J
Johann Baudy 已提交
416 417

	smp_wmb();
418 419
}

J
Johann Baudy 已提交
420
static int __packet_get_status(struct packet_sock *po, void *frame)
421
{
422
	union tpacket_uhdr h;
423

J
Johann Baudy 已提交
424 425
	smp_rmb();

426 427 428
	h.raw = frame;
	switch (po->tp_version) {
	case TPACKET_V1:
429
		flush_dcache_page(pgv_to_page(&h.h1->tp_status));
J
Johann Baudy 已提交
430
		return h.h1->tp_status;
431
	case TPACKET_V2:
432
		flush_dcache_page(pgv_to_page(&h.h2->tp_status));
J
Johann Baudy 已提交
433
		return h.h2->tp_status;
434
	case TPACKET_V3:
J
Johann Baudy 已提交
435
	default:
436
		WARN(1, "TPACKET version not supported.\n");
J
Johann Baudy 已提交
437 438
		BUG();
		return 0;
439
	}
L
Linus Torvalds 已提交
440
}
J
Johann Baudy 已提交
441

442 443
static __u32 tpacket_get_timestamp(struct sk_buff *skb, struct timespec *ts,
				   unsigned int flags)
444 445 446
{
	struct skb_shared_hwtstamps *shhwtstamps = skb_hwtstamps(skb);

447 448 449 450
	if (shhwtstamps &&
	    (flags & SOF_TIMESTAMPING_RAW_HARDWARE) &&
	    ktime_to_timespec_cond(shhwtstamps->hwtstamp, ts))
		return TP_STATUS_TS_RAW_HARDWARE;
451 452

	if (ktime_to_timespec_cond(skb->tstamp, ts))
453
		return TP_STATUS_TS_SOFTWARE;
454

455
	return 0;
456 457
}

458 459
static __u32 __packet_set_timestamp(struct packet_sock *po, void *frame,
				    struct sk_buff *skb)
460 461 462
{
	union tpacket_uhdr h;
	struct timespec ts;
463
	__u32 ts_status;
464

465 466
	if (!(ts_status = tpacket_get_timestamp(skb, &ts, po->tp_tstamp)))
		return 0;
467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486

	h.raw = frame;
	switch (po->tp_version) {
	case TPACKET_V1:
		h.h1->tp_sec = ts.tv_sec;
		h.h1->tp_usec = ts.tv_nsec / NSEC_PER_USEC;
		break;
	case TPACKET_V2:
		h.h2->tp_sec = ts.tv_sec;
		h.h2->tp_nsec = ts.tv_nsec;
		break;
	case TPACKET_V3:
	default:
		WARN(1, "TPACKET version not supported.\n");
		BUG();
	}

	/* one flush is safe, as both fields always lie on the same cacheline */
	flush_dcache_page(pgv_to_page(&h.h1->tp_sec));
	smp_wmb();
487 488

	return ts_status;
489 490
}

J
Johann Baudy 已提交
491 492 493 494 495 496
static void *packet_lookup_frame(struct packet_sock *po,
		struct packet_ring_buffer *rb,
		unsigned int position,
		int status)
{
	unsigned int pg_vec_pos, frame_offset;
497
	union tpacket_uhdr h;
J
Johann Baudy 已提交
498 499 500 501

	pg_vec_pos = position / rb->frames_per_block;
	frame_offset = position % rb->frames_per_block;

502 503
	h.raw = rb->pg_vec[pg_vec_pos].buffer +
		(frame_offset * rb->frame_size);
J
Johann Baudy 已提交
504 505 506 507 508 509 510

	if (status != __packet_get_status(po, h.raw))
		return NULL;

	return h.raw;
}

511
static void *packet_current_frame(struct packet_sock *po,
J
Johann Baudy 已提交
512 513 514 515 516 517
		struct packet_ring_buffer *rb,
		int status)
{
	return packet_lookup_frame(po, rb, rb->head, status);
}

518
static void prb_del_retire_blk_timer(struct tpacket_kbdq_core *pkc)
519 520 521 522 523 524 525
{
	del_timer_sync(&pkc->retire_blk_timer);
}

static void prb_shutdown_retire_blk_timer(struct packet_sock *po,
		struct sk_buff_head *rb_queue)
{
526
	struct tpacket_kbdq_core *pkc;
527

528
	pkc = GET_PBDQC_FROM_RB(&po->rx_ring);
529

530
	spin_lock_bh(&rb_queue->lock);
531
	pkc->delete_blk_timer = 1;
532
	spin_unlock_bh(&rb_queue->lock);
533 534 535 536 537

	prb_del_retire_blk_timer(pkc);
}

static void prb_init_blk_timer(struct packet_sock *po,
538
		struct tpacket_kbdq_core *pkc,
539 540 541 542 543 544 545 546
		void (*func) (unsigned long))
{
	init_timer(&pkc->retire_blk_timer);
	pkc->retire_blk_timer.data = (long)po;
	pkc->retire_blk_timer.function = func;
	pkc->retire_blk_timer.expires = jiffies;
}

547
static void prb_setup_retire_blk_timer(struct packet_sock *po)
548
{
549
	struct tpacket_kbdq_core *pkc;
550

551
	pkc = GET_PBDQC_FROM_RB(&po->rx_ring);
552 553 554 555 556 557 558 559
	prb_init_blk_timer(po, pkc, prb_retire_rx_blk_timer_expired);
}

static int prb_calc_retire_blk_tmo(struct packet_sock *po,
				int blk_size_in_bytes)
{
	struct net_device *dev;
	unsigned int mbits = 0, msec = 0, div = 0, tmo = 0;
560 561
	struct ethtool_cmd ecmd;
	int err;
562
	u32 speed;
563

564 565 566 567
	rtnl_lock();
	dev = __dev_get_by_index(sock_net(&po->sk), po->ifindex);
	if (unlikely(!dev)) {
		rtnl_unlock();
568
		return DEFAULT_PRB_RETIRE_TOV;
569 570
	}
	err = __ethtool_get_settings(dev, &ecmd);
571
	speed = ethtool_cmd_speed(&ecmd);
572 573 574 575 576 577
	rtnl_unlock();
	if (!err) {
		/*
		 * If the link speed is so slow you don't really
		 * need to worry about perf anyways
		 */
578
		if (speed < SPEED_1000 || speed == SPEED_UNKNOWN) {
579
			return DEFAULT_PRB_RETIRE_TOV;
580 581 582
		} else {
			msec = 1;
			div = speed / 1000;
583 584 585 586 587 588 589 590 591 592 593 594 595 596 597
		}
	}

	mbits = (blk_size_in_bytes * 8) / (1024 * 1024);

	if (div)
		mbits /= div;

	tmo = mbits * msec;

	if (div)
		return tmo+1;
	return tmo;
}

598
static void prb_init_ft_ops(struct tpacket_kbdq_core *p1,
599 600 601 602 603 604 605 606
			union tpacket_req_u *req_u)
{
	p1->feature_req_word = req_u->req3.tp_feature_req_word;
}

static void init_prb_bdqc(struct packet_sock *po,
			struct packet_ring_buffer *rb,
			struct pgv *pg_vec,
607
			union tpacket_req_u *req_u)
608
{
609
	struct tpacket_kbdq_core *p1 = GET_PBDQC_FROM_RB(rb);
610
	struct tpacket_block_desc *pbd;
611 612 613 614 615

	memset(p1, 0x0, sizeof(*p1));

	p1->knxt_seq_num = 1;
	p1->pkbdq = pg_vec;
616
	pbd = (struct tpacket_block_desc *)pg_vec[0].buffer;
J
Joe Perches 已提交
617
	p1->pkblk_start	= pg_vec[0].buffer;
618 619 620 621 622
	p1->kblk_size = req_u->req3.tp_block_size;
	p1->knum_blocks	= req_u->req3.tp_block_nr;
	p1->hdrlen = po->tp_hdrlen;
	p1->version = po->tp_version;
	p1->last_kactive_blk_num = 0;
623
	po->stats.stats3.tp_freeze_q_cnt = 0;
624 625 626 627 628 629 630 631
	if (req_u->req3.tp_retire_blk_tov)
		p1->retire_blk_tov = req_u->req3.tp_retire_blk_tov;
	else
		p1->retire_blk_tov = prb_calc_retire_blk_tmo(po,
						req_u->req3.tp_block_size);
	p1->tov_in_jiffies = msecs_to_jiffies(p1->retire_blk_tov);
	p1->blk_sizeof_priv = req_u->req3.tp_sizeof_priv;

632
	p1->max_frame_len = p1->kblk_size - BLK_PLUS_PRIV(p1->blk_sizeof_priv);
633
	prb_init_ft_ops(p1, req_u);
634
	prb_setup_retire_blk_timer(po);
635 636 637 638 639 640
	prb_open_block(p1, pbd);
}

/*  Do NOT update the last_blk_num first.
 *  Assumes sk_buff_head lock is held.
 */
641
static void _prb_refresh_rx_retire_blk_timer(struct tpacket_kbdq_core *pkc)
642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673
{
	mod_timer(&pkc->retire_blk_timer,
			jiffies + pkc->tov_in_jiffies);
	pkc->last_kactive_blk_num = pkc->kactive_blk_num;
}

/*
 * Timer logic:
 * 1) We refresh the timer only when we open a block.
 *    By doing this we don't waste cycles refreshing the timer
 *	  on packet-by-packet basis.
 *
 * With a 1MB block-size, on a 1Gbps line, it will take
 * i) ~8 ms to fill a block + ii) memcpy etc.
 * In this cut we are not accounting for the memcpy time.
 *
 * So, if the user sets the 'tmo' to 10ms then the timer
 * will never fire while the block is still getting filled
 * (which is what we want). However, the user could choose
 * to close a block early and that's fine.
 *
 * But when the timer does fire, we check whether or not to refresh it.
 * Since the tmo granularity is in msecs, it is not too expensive
 * to refresh the timer, lets say every '8' msecs.
 * Either the user can set the 'tmo' or we can derive it based on
 * a) line-speed and b) block-size.
 * prb_calc_retire_blk_tmo() calculates the tmo.
 *
 */
static void prb_retire_rx_blk_timer_expired(unsigned long data)
{
	struct packet_sock *po = (struct packet_sock *)data;
674
	struct tpacket_kbdq_core *pkc = GET_PBDQC_FROM_RB(&po->rx_ring);
675
	unsigned int frozen;
676
	struct tpacket_block_desc *pbd;
677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703

	spin_lock(&po->sk.sk_receive_queue.lock);

	frozen = prb_queue_frozen(pkc);
	pbd = GET_CURR_PBLOCK_DESC_FROM_CORE(pkc);

	if (unlikely(pkc->delete_blk_timer))
		goto out;

	/* We only need to plug the race when the block is partially filled.
	 * tpacket_rcv:
	 *		lock(); increment BLOCK_NUM_PKTS; unlock()
	 *		copy_bits() is in progress ...
	 *		timer fires on other cpu:
	 *		we can't retire the current block because copy_bits
	 *		is in progress.
	 *
	 */
	if (BLOCK_NUM_PKTS(pbd)) {
		while (atomic_read(&pkc->blk_fill_in_prog)) {
			/* Waiting for skb_copy_bits to finish... */
			cpu_relax();
		}
	}

	if (pkc->last_kactive_blk_num == pkc->kactive_blk_num) {
		if (!frozen) {
704 705 706 707
			if (!BLOCK_NUM_PKTS(pbd)) {
				/* An empty block. Just refresh the timer. */
				goto refresh_timer;
			}
708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743
			prb_retire_current_block(pkc, po, TP_STATUS_BLK_TMO);
			if (!prb_dispatch_next_block(pkc, po))
				goto refresh_timer;
			else
				goto out;
		} else {
			/* Case 1. Queue was frozen because user-space was
			 *	   lagging behind.
			 */
			if (prb_curr_blk_in_use(pkc, pbd)) {
				/*
				 * Ok, user-space is still behind.
				 * So just refresh the timer.
				 */
				goto refresh_timer;
			} else {
			       /* Case 2. queue was frozen,user-space caught up,
				* now the link went idle && the timer fired.
				* We don't have a block to close.So we open this
				* block and restart the timer.
				* opening a block thaws the queue,restarts timer
				* Thawing/timer-refresh is a side effect.
				*/
				prb_open_block(pkc, pbd);
				goto out;
			}
		}
	}

refresh_timer:
	_prb_refresh_rx_retire_blk_timer(pkc);

out:
	spin_unlock(&po->sk.sk_receive_queue.lock);
}

744
static void prb_flush_block(struct tpacket_kbdq_core *pkc1,
745
		struct tpacket_block_desc *pbd1, __u32 status)
746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786
{
	/* Flush everything minus the block header */

#if ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE == 1
	u8 *start, *end;

	start = (u8 *)pbd1;

	/* Skip the block header(we know header WILL fit in 4K) */
	start += PAGE_SIZE;

	end = (u8 *)PAGE_ALIGN((unsigned long)pkc1->pkblk_end);
	for (; start < end; start += PAGE_SIZE)
		flush_dcache_page(pgv_to_page(start));

	smp_wmb();
#endif

	/* Now update the block status. */

	BLOCK_STATUS(pbd1) = status;

	/* Flush the block header */

#if ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE == 1
	start = (u8 *)pbd1;
	flush_dcache_page(pgv_to_page(start));

	smp_wmb();
#endif
}

/*
 * Side effect:
 *
 * 1) flush the block
 * 2) Increment active_blk_num
 *
 * Note:We DONT refresh the timer on purpose.
 *	Because almost always the next block will be opened.
 */
787 788
static void prb_close_block(struct tpacket_kbdq_core *pkc1,
		struct tpacket_block_desc *pbd1,
789 790 791 792 793
		struct packet_sock *po, unsigned int stat)
{
	__u32 status = TP_STATUS_USER | stat;

	struct tpacket3_hdr *last_pkt;
794
	struct tpacket_hdr_v1 *h1 = &pbd1->hdr.bh1;
795
	struct sock *sk = &po->sk;
796

797
	if (po->stats.stats3.tp_drops)
798 799 800 801 802 803 804 805 806 807
		status |= TP_STATUS_LOSING;

	last_pkt = (struct tpacket3_hdr *)pkc1->prev;
	last_pkt->tp_next_offset = 0;

	/* Get the ts of the last pkt */
	if (BLOCK_NUM_PKTS(pbd1)) {
		h1->ts_last_pkt.ts_sec = last_pkt->tp_sec;
		h1->ts_last_pkt.ts_nsec	= last_pkt->tp_nsec;
	} else {
808 809 810 811 812
		/* Ok, we tmo'd - so get the current time.
		 *
		 * It shouldn't really happen as we don't close empty
		 * blocks. See prb_retire_rx_blk_timer_expired().
		 */
813 814 815 816 817 818 819 820 821 822 823
		struct timespec ts;
		getnstimeofday(&ts);
		h1->ts_last_pkt.ts_sec = ts.tv_sec;
		h1->ts_last_pkt.ts_nsec	= ts.tv_nsec;
	}

	smp_wmb();

	/* Flush the block */
	prb_flush_block(pkc1, pbd1, status);

824 825
	sk->sk_data_ready(sk);

826 827 828
	pkc1->kactive_blk_num = GET_NEXT_PRB_BLK_NUM(pkc1);
}

829
static void prb_thaw_queue(struct tpacket_kbdq_core *pkc)
830 831 832 833 834 835 836 837 838 839 840
{
	pkc->reset_pending_on_curr_blk = 0;
}

/*
 * Side effect of opening a block:
 *
 * 1) prb_queue is thawed.
 * 2) retire_blk_timer is refreshed.
 *
 */
841 842
static void prb_open_block(struct tpacket_kbdq_core *pkc1,
	struct tpacket_block_desc *pbd1)
843 844
{
	struct timespec ts;
845
	struct tpacket_hdr_v1 *h1 = &pbd1->hdr.bh1;
846 847 848

	smp_rmb();

849 850 851
	/* We could have just memset this but we will lose the
	 * flexibility of making the priv area sticky
	 */
852

853 854 855
	BLOCK_SNUM(pbd1) = pkc1->knxt_seq_num++;
	BLOCK_NUM_PKTS(pbd1) = 0;
	BLOCK_LEN(pbd1) = BLK_PLUS_PRIV(pkc1->blk_sizeof_priv);
856

857 858 859 860
	getnstimeofday(&ts);

	h1->ts_first_pkt.ts_sec = ts.tv_sec;
	h1->ts_first_pkt.ts_nsec = ts.tv_nsec;
861

862 863 864 865 866 867 868 869 870 871 872 873 874 875
	pkc1->pkblk_start = (char *)pbd1;
	pkc1->nxt_offset = pkc1->pkblk_start + BLK_PLUS_PRIV(pkc1->blk_sizeof_priv);

	BLOCK_O2FP(pbd1) = (__u32)BLK_PLUS_PRIV(pkc1->blk_sizeof_priv);
	BLOCK_O2PRIV(pbd1) = BLK_HDR_LEN;

	pbd1->version = pkc1->version;
	pkc1->prev = pkc1->nxt_offset;
	pkc1->pkblk_end = pkc1->pkblk_start + pkc1->kblk_size;

	prb_thaw_queue(pkc1);
	_prb_refresh_rx_retire_blk_timer(pkc1);

	smp_wmb();
876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900
}

/*
 * Queue freeze logic:
 * 1) Assume tp_block_nr = 8 blocks.
 * 2) At time 't0', user opens Rx ring.
 * 3) Some time past 't0', kernel starts filling blocks starting from 0 .. 7
 * 4) user-space is either sleeping or processing block '0'.
 * 5) tpacket_rcv is currently filling block '7', since there is no space left,
 *    it will close block-7,loop around and try to fill block '0'.
 *    call-flow:
 *    __packet_lookup_frame_in_block
 *      prb_retire_current_block()
 *      prb_dispatch_next_block()
 *        |->(BLOCK_STATUS == USER) evaluates to true
 *    5.1) Since block-0 is currently in-use, we just freeze the queue.
 * 6) Now there are two cases:
 *    6.1) Link goes idle right after the queue is frozen.
 *         But remember, the last open_block() refreshed the timer.
 *         When this timer expires,it will refresh itself so that we can
 *         re-open block-0 in near future.
 *    6.2) Link is busy and keeps on receiving packets. This is a simple
 *         case and __packet_lookup_frame_in_block will check if block-0
 *         is free and can now be re-used.
 */
901
static void prb_freeze_queue(struct tpacket_kbdq_core *pkc,
902 903 904
				  struct packet_sock *po)
{
	pkc->reset_pending_on_curr_blk = 1;
905
	po->stats.stats3.tp_freeze_q_cnt++;
906 907 908 909 910 911 912 913 914 915
}

#define TOTAL_PKT_LEN_INCL_ALIGN(length) (ALIGN((length), V3_ALIGNMENT))

/*
 * If the next block is free then we will dispatch it
 * and return a good offset.
 * Else, we will freeze the queue.
 * So, caller must check the return value.
 */
916
static void *prb_dispatch_next_block(struct tpacket_kbdq_core *pkc,
917 918
		struct packet_sock *po)
{
919
	struct tpacket_block_desc *pbd;
920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940

	smp_rmb();

	/* 1. Get current block num */
	pbd = GET_CURR_PBLOCK_DESC_FROM_CORE(pkc);

	/* 2. If this block is currently in_use then freeze the queue */
	if (TP_STATUS_USER & BLOCK_STATUS(pbd)) {
		prb_freeze_queue(pkc, po);
		return NULL;
	}

	/*
	 * 3.
	 * open this block and return the offset where the first packet
	 * needs to get stored.
	 */
	prb_open_block(pkc, pbd);
	return (void *)pkc->nxt_offset;
}

941
static void prb_retire_current_block(struct tpacket_kbdq_core *pkc,
942 943
		struct packet_sock *po, unsigned int status)
{
944
	struct tpacket_block_desc *pbd = GET_CURR_PBLOCK_DESC_FROM_CORE(pkc);
945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967

	/* retire/close the current block */
	if (likely(TP_STATUS_KERNEL == BLOCK_STATUS(pbd))) {
		/*
		 * Plug the case where copy_bits() is in progress on
		 * cpu-0 and tpacket_rcv() got invoked on cpu-1, didn't
		 * have space to copy the pkt in the current block and
		 * called prb_retire_current_block()
		 *
		 * We don't need to worry about the TMO case because
		 * the timer-handler already handled this case.
		 */
		if (!(status & TP_STATUS_BLK_TMO)) {
			while (atomic_read(&pkc->blk_fill_in_prog)) {
				/* Waiting for skb_copy_bits to finish... */
				cpu_relax();
			}
		}
		prb_close_block(pkc, pbd, po, status);
		return;
	}
}

968
static int prb_curr_blk_in_use(struct tpacket_kbdq_core *pkc,
969
				      struct tpacket_block_desc *pbd)
970 971 972 973
{
	return TP_STATUS_USER & BLOCK_STATUS(pbd);
}

974
static int prb_queue_frozen(struct tpacket_kbdq_core *pkc)
975 976 977 978
{
	return pkc->reset_pending_on_curr_blk;
}

979
static void prb_clear_blk_fill_status(struct packet_ring_buffer *rb)
980
{
981
	struct tpacket_kbdq_core *pkc  = GET_PBDQC_FROM_RB(rb);
982 983 984
	atomic_dec(&pkc->blk_fill_in_prog);
}

985
static void prb_fill_rxhash(struct tpacket_kbdq_core *pkc,
986 987
			struct tpacket3_hdr *ppd)
{
988
	ppd->hv1.tp_rxhash = skb_get_hash(pkc->skb);
989 990
}

991
static void prb_clear_rxhash(struct tpacket_kbdq_core *pkc,
992 993 994 995 996
			struct tpacket3_hdr *ppd)
{
	ppd->hv1.tp_rxhash = 0;
}

997
static void prb_fill_vlan_info(struct tpacket_kbdq_core *pkc,
998 999
			struct tpacket3_hdr *ppd)
{
1000 1001
	if (skb_vlan_tag_present(pkc->skb)) {
		ppd->hv1.tp_vlan_tci = skb_vlan_tag_get(pkc->skb);
1002 1003
		ppd->hv1.tp_vlan_tpid = ntohs(pkc->skb->vlan_proto);
		ppd->tp_status = TP_STATUS_VLAN_VALID | TP_STATUS_VLAN_TPID_VALID;
1004
	} else {
1005
		ppd->hv1.tp_vlan_tci = 0;
1006
		ppd->hv1.tp_vlan_tpid = 0;
1007
		ppd->tp_status = TP_STATUS_AVAILABLE;
1008 1009 1010
	}
}

1011
static void prb_run_all_ft_ops(struct tpacket_kbdq_core *pkc,
1012 1013
			struct tpacket3_hdr *ppd)
{
1014
	ppd->hv1.tp_padding = 0;
1015 1016 1017 1018 1019 1020 1021 1022
	prb_fill_vlan_info(pkc, ppd);

	if (pkc->feature_req_word & TP_FT_REQ_FILL_RXHASH)
		prb_fill_rxhash(pkc, ppd);
	else
		prb_clear_rxhash(pkc, ppd);
}

1023
static void prb_fill_curr_block(char *curr,
1024 1025
				struct tpacket_kbdq_core *pkc,
				struct tpacket_block_desc *pbd,
1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046
				unsigned int len)
{
	struct tpacket3_hdr *ppd;

	ppd  = (struct tpacket3_hdr *)curr;
	ppd->tp_next_offset = TOTAL_PKT_LEN_INCL_ALIGN(len);
	pkc->prev = curr;
	pkc->nxt_offset += TOTAL_PKT_LEN_INCL_ALIGN(len);
	BLOCK_LEN(pbd) += TOTAL_PKT_LEN_INCL_ALIGN(len);
	BLOCK_NUM_PKTS(pbd) += 1;
	atomic_inc(&pkc->blk_fill_in_prog);
	prb_run_all_ft_ops(pkc, ppd);
}

/* Assumes caller has the sk->rx_queue.lock */
static void *__packet_lookup_frame_in_block(struct packet_sock *po,
					    struct sk_buff *skb,
						int status,
					    unsigned int len
					    )
{
1047 1048
	struct tpacket_kbdq_core *pkc;
	struct tpacket_block_desc *pbd;
1049 1050
	char *curr, *end;

J
Joe Perches 已提交
1051
	pkc = GET_PBDQC_FROM_RB(&po->rx_ring);
1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076
	pbd = GET_CURR_PBLOCK_DESC_FROM_CORE(pkc);

	/* Queue is frozen when user space is lagging behind */
	if (prb_queue_frozen(pkc)) {
		/*
		 * Check if that last block which caused the queue to freeze,
		 * is still in_use by user-space.
		 */
		if (prb_curr_blk_in_use(pkc, pbd)) {
			/* Can't record this packet */
			return NULL;
		} else {
			/*
			 * Ok, the block was released by user-space.
			 * Now let's open that block.
			 * opening a block also thaws the queue.
			 * Thawing is a side effect.
			 */
			prb_open_block(pkc, pbd);
		}
	}

	smp_mb();
	curr = pkc->nxt_offset;
	pkc->skb = skb;
J
Joe Perches 已提交
1077
	end = (char *)pbd + pkc->kblk_size;
1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102

	/* first try the current block */
	if (curr+TOTAL_PKT_LEN_INCL_ALIGN(len) < end) {
		prb_fill_curr_block(curr, pkc, pbd, len);
		return (void *)curr;
	}

	/* Ok, close the current block */
	prb_retire_current_block(pkc, po, 0);

	/* Now, try to dispatch the next block */
	curr = (char *)prb_dispatch_next_block(pkc, po);
	if (curr) {
		pbd = GET_CURR_PBLOCK_DESC_FROM_CORE(pkc);
		prb_fill_curr_block(curr, pkc, pbd, len);
		return (void *)curr;
	}

	/*
	 * No free blocks are available.user_space hasn't caught up yet.
	 * Queue was just frozen and now this packet will get dropped.
	 */
	return NULL;
}

1103
static void *packet_current_rx_frame(struct packet_sock *po,
1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118
					    struct sk_buff *skb,
					    int status, unsigned int len)
{
	char *curr = NULL;
	switch (po->tp_version) {
	case TPACKET_V1:
	case TPACKET_V2:
		curr = packet_lookup_frame(po, &po->rx_ring,
					po->rx_ring.head, status);
		return curr;
	case TPACKET_V3:
		return __packet_lookup_frame_in_block(po, skb, status, len);
	default:
		WARN(1, "TPACKET version not supported\n");
		BUG();
1119
		return NULL;
1120 1121 1122
	}
}

1123
static void *prb_lookup_block(struct packet_sock *po,
1124
				     struct packet_ring_buffer *rb,
1125
				     unsigned int idx,
1126 1127
				     int status)
{
1128
	struct tpacket_kbdq_core *pkc  = GET_PBDQC_FROM_RB(rb);
1129
	struct tpacket_block_desc *pbd = GET_PBLOCK_DESC(pkc, idx);
1130 1131 1132 1133 1134 1135

	if (status != BLOCK_STATUS(pbd))
		return NULL;
	return pbd;
}

1136
static int prb_previous_blk_num(struct packet_ring_buffer *rb)
1137 1138 1139 1140 1141 1142 1143 1144 1145 1146
{
	unsigned int prev;
	if (rb->prb_bdqc.kactive_blk_num)
		prev = rb->prb_bdqc.kactive_blk_num-1;
	else
		prev = rb->prb_bdqc.knum_blocks-1;
	return prev;
}

/* Assumes caller has held the rx_queue.lock */
1147
static void *__prb_previous_block(struct packet_sock *po,
1148 1149 1150 1151 1152 1153 1154
					 struct packet_ring_buffer *rb,
					 int status)
{
	unsigned int previous = prb_previous_blk_num(rb);
	return prb_lookup_block(po, rb, previous, status);
}

1155
static void *packet_previous_rx_frame(struct packet_sock *po,
1156 1157 1158 1159 1160 1161 1162 1163 1164
					     struct packet_ring_buffer *rb,
					     int status)
{
	if (po->tp_version <= TPACKET_V2)
		return packet_previous_frame(po, rb, status);

	return __prb_previous_block(po, rb, status);
}

1165
static void packet_increment_rx_head(struct packet_sock *po,
1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179
					    struct packet_ring_buffer *rb)
{
	switch (po->tp_version) {
	case TPACKET_V1:
	case TPACKET_V2:
		return packet_increment_head(rb);
	case TPACKET_V3:
	default:
		WARN(1, "TPACKET version not supported.\n");
		BUG();
		return;
	}
}

1180
static void *packet_previous_frame(struct packet_sock *po,
J
Johann Baudy 已提交
1181 1182 1183 1184 1185 1186 1187
		struct packet_ring_buffer *rb,
		int status)
{
	unsigned int previous = rb->head ? rb->head - 1 : rb->frame_max;
	return packet_lookup_frame(po, rb, previous, status);
}

1188
static void packet_increment_head(struct packet_ring_buffer *buff)
J
Johann Baudy 已提交
1189 1190 1191 1192
{
	buff->head = buff->head != buff->frame_max ? buff->head+1 : 0;
}

1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233
static void packet_inc_pending(struct packet_ring_buffer *rb)
{
	this_cpu_inc(*rb->pending_refcnt);
}

static void packet_dec_pending(struct packet_ring_buffer *rb)
{
	this_cpu_dec(*rb->pending_refcnt);
}

static unsigned int packet_read_pending(const struct packet_ring_buffer *rb)
{
	unsigned int refcnt = 0;
	int cpu;

	/* We don't use pending refcount in rx_ring. */
	if (rb->pending_refcnt == NULL)
		return 0;

	for_each_possible_cpu(cpu)
		refcnt += *per_cpu_ptr(rb->pending_refcnt, cpu);

	return refcnt;
}

static int packet_alloc_pending(struct packet_sock *po)
{
	po->rx_ring.pending_refcnt = NULL;

	po->tx_ring.pending_refcnt = alloc_percpu(unsigned int);
	if (unlikely(po->tx_ring.pending_refcnt == NULL))
		return -ENOBUFS;

	return 0;
}

static void packet_free_pending(struct packet_sock *po)
{
	free_percpu(po->tx_ring.pending_refcnt);
}

1234 1235 1236 1237 1238 1239
#define ROOM_POW_OFF	2
#define ROOM_NONE	0x0
#define ROOM_LOW	0x1
#define ROOM_NORMAL	0x2

static bool __tpacket_has_room(struct packet_sock *po, int pow_off)
1240
{
1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263
	int idx, len;

	len = po->rx_ring.frame_max + 1;
	idx = po->rx_ring.head;
	if (pow_off)
		idx += len >> pow_off;
	if (idx >= len)
		idx -= len;
	return packet_lookup_frame(po, &po->rx_ring, idx, TP_STATUS_KERNEL);
}

static bool __tpacket_v3_has_room(struct packet_sock *po, int pow_off)
{
	int idx, len;

	len = po->rx_ring.prb_bdqc.knum_blocks;
	idx = po->rx_ring.prb_bdqc.kactive_blk_num;
	if (pow_off)
		idx += len >> pow_off;
	if (idx >= len)
		idx -= len;
	return prb_lookup_block(po, &po->rx_ring, idx, TP_STATUS_KERNEL);
}
1264

1265
static int __packet_rcv_has_room(struct packet_sock *po, struct sk_buff *skb)
1266 1267 1268 1269 1270 1271
{
	struct sock *sk = &po->sk;
	int ret = ROOM_NONE;

	if (po->prot_hook.func != tpacket_rcv) {
		int avail = sk->sk_rcvbuf - atomic_read(&sk->sk_rmem_alloc)
1272
					  - (skb ? skb->truesize : 0);
1273 1274 1275 1276 1277 1278 1279
		if (avail > (sk->sk_rcvbuf >> ROOM_POW_OFF))
			return ROOM_NORMAL;
		else if (avail > 0)
			return ROOM_LOW;
		else
			return ROOM_NONE;
	}
1280

1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291
	if (po->tp_version == TPACKET_V3) {
		if (__tpacket_v3_has_room(po, ROOM_POW_OFF))
			ret = ROOM_NORMAL;
		else if (__tpacket_v3_has_room(po, 0))
			ret = ROOM_LOW;
	} else {
		if (__tpacket_has_room(po, ROOM_POW_OFF))
			ret = ROOM_NORMAL;
		else if (__tpacket_has_room(po, 0))
			ret = ROOM_LOW;
	}
1292 1293 1294 1295 1296 1297 1298 1299 1300

	return ret;
}

static int packet_rcv_has_room(struct packet_sock *po, struct sk_buff *skb)
{
	int ret;
	bool has_room;

1301 1302
	spin_lock_bh(&po->sk.sk_receive_queue.lock);
	ret = __packet_rcv_has_room(po, skb);
1303 1304
	has_room = ret == ROOM_NORMAL;
	if (po->pressure == has_room)
1305 1306
		po->pressure = !has_room;
	spin_unlock_bh(&po->sk.sk_receive_queue.lock);
1307

1308
	return ret;
1309 1310
}

L
Linus Torvalds 已提交
1311 1312
static void packet_sock_destruct(struct sock *sk)
{
1313 1314
	skb_queue_purge(&sk->sk_error_queue);

1315 1316
	WARN_ON(atomic_read(&sk->sk_rmem_alloc));
	WARN_ON(atomic_read(&sk->sk_wmem_alloc));
L
Linus Torvalds 已提交
1317 1318

	if (!sock_flag(sk, SOCK_DEAD)) {
E
Eric Dumazet 已提交
1319
		pr_err("Attempt to release alive packet socket: %p\n", sk);
L
Linus Torvalds 已提交
1320 1321 1322
		return;
	}

1323
	sk_refcnt_debug_dec(sk);
L
Linus Torvalds 已提交
1324 1325
}

1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339
static bool fanout_flow_is_huge(struct packet_sock *po, struct sk_buff *skb)
{
	u32 rxhash;
	int i, count = 0;

	rxhash = skb_get_hash(skb);
	for (i = 0; i < ROLLOVER_HLEN; i++)
		if (po->rollover->history[i] == rxhash)
			count++;

	po->rollover->history[prandom_u32() % ROLLOVER_HLEN] = rxhash;
	return count > (ROLLOVER_HLEN >> 1);
}

1340 1341 1342
static unsigned int fanout_demux_hash(struct packet_fanout *f,
				      struct sk_buff *skb,
				      unsigned int num)
D
David S. Miller 已提交
1343
{
1344
	return reciprocal_scale(skb_get_hash(skb), num);
D
David S. Miller 已提交
1345 1346
}

1347 1348 1349
static unsigned int fanout_demux_lb(struct packet_fanout *f,
				    struct sk_buff *skb,
				    unsigned int num)
D
David S. Miller 已提交
1350
{
1351
	unsigned int val = atomic_inc_return(&f->rr_cur);
D
David S. Miller 已提交
1352

1353
	return val % num;
1354 1355 1356 1357 1358 1359 1360
}

static unsigned int fanout_demux_cpu(struct packet_fanout *f,
				     struct sk_buff *skb,
				     unsigned int num)
{
	return smp_processor_id() % num;
D
David S. Miller 已提交
1361 1362
}

1363 1364 1365 1366
static unsigned int fanout_demux_rnd(struct packet_fanout *f,
				     struct sk_buff *skb,
				     unsigned int num)
{
1367
	return prandom_u32_max(num);
1368 1369
}

1370 1371
static unsigned int fanout_demux_rollover(struct packet_fanout *f,
					  struct sk_buff *skb,
1372
					  unsigned int idx, bool try_self,
1373
					  unsigned int num)
1374
{
1375
	struct packet_sock *po, *po_next, *po_skip = NULL;
W
Willem de Bruijn 已提交
1376
	unsigned int i, j, room = ROOM_NONE;
1377

1378
	po = pkt_sk(f->arr[idx]);
1379 1380 1381 1382 1383 1384

	if (try_self) {
		room = packet_rcv_has_room(po, skb);
		if (room == ROOM_NORMAL ||
		    (room == ROOM_LOW && !fanout_flow_is_huge(po, skb)))
			return idx;
1385
		po_skip = po;
1386
	}
1387

1388
	i = j = min_t(int, po->rollover->sock, num - 1);
1389
	do {
1390
		po_next = pkt_sk(f->arr[i]);
1391
		if (po_next != po_skip && !po_next->pressure &&
1392
		    packet_rcv_has_room(po_next, skb) == ROOM_NORMAL) {
1393
			if (i != j)
1394
				po->rollover->sock = i;
W
Willem de Bruijn 已提交
1395 1396 1397
			atomic_long_inc(&po->rollover->num);
			if (room == ROOM_LOW)
				atomic_long_inc(&po->rollover->num_huge);
1398 1399
			return i;
		}
1400

1401 1402 1403 1404
		if (++i == num)
			i = 0;
	} while (i != j);

W
Willem de Bruijn 已提交
1405
	atomic_long_inc(&po->rollover->num_failed);
1406 1407 1408
	return idx;
}

1409 1410 1411 1412 1413 1414 1415
static unsigned int fanout_demux_qm(struct packet_fanout *f,
				    struct sk_buff *skb,
				    unsigned int num)
{
	return skb_get_queue_mapping(skb) % num;
}

1416 1417 1418 1419 1420 1421 1422 1423 1424 1425
static unsigned int fanout_demux_bpf(struct packet_fanout *f,
				     struct sk_buff *skb,
				     unsigned int num)
{
	struct bpf_prog *prog;
	unsigned int ret = 0;

	rcu_read_lock();
	prog = rcu_dereference(f->bpf_prog);
	if (prog)
1426
		ret = bpf_prog_run_clear_cb(prog, skb) % num;
1427 1428 1429 1430 1431
	rcu_read_unlock();

	return ret;
}

1432 1433 1434
static bool fanout_has_flag(struct packet_fanout *f, u16 flag)
{
	return f->flags & (flag >> 8);
1435 1436 1437 1438
}

static int packet_rcv_fanout(struct sk_buff *skb, struct net_device *dev,
			     struct packet_type *pt, struct net_device *orig_dev)
D
David S. Miller 已提交
1439 1440
{
	struct packet_fanout *f = pt->af_packet_priv;
1441
	unsigned int num = READ_ONCE(f->num_members);
1442
	struct net *net = read_pnet(&f->net);
D
David S. Miller 已提交
1443
	struct packet_sock *po;
1444
	unsigned int idx;
D
David S. Miller 已提交
1445

1446
	if (!net_eq(dev_net(dev), net) || !num) {
D
David S. Miller 已提交
1447 1448 1449 1450
		kfree_skb(skb);
		return 0;
	}

1451
	if (fanout_has_flag(f, PACKET_FANOUT_FLAG_DEFRAG)) {
1452
		skb = ip_check_defrag(net, skb, IP_DEFRAG_AF_PACKET);
1453 1454 1455
		if (!skb)
			return 0;
	}
1456 1457 1458
	switch (f->type) {
	case PACKET_FANOUT_HASH:
	default:
1459
		idx = fanout_demux_hash(f, skb, num);
1460 1461
		break;
	case PACKET_FANOUT_LB:
1462
		idx = fanout_demux_lb(f, skb, num);
1463 1464
		break;
	case PACKET_FANOUT_CPU:
1465 1466
		idx = fanout_demux_cpu(f, skb, num);
		break;
1467 1468 1469
	case PACKET_FANOUT_RND:
		idx = fanout_demux_rnd(f, skb, num);
		break;
1470 1471 1472
	case PACKET_FANOUT_QM:
		idx = fanout_demux_qm(f, skb, num);
		break;
1473
	case PACKET_FANOUT_ROLLOVER:
1474
		idx = fanout_demux_rollover(f, skb, 0, false, num);
1475
		break;
1476
	case PACKET_FANOUT_CBPF:
1477
	case PACKET_FANOUT_EBPF:
1478 1479
		idx = fanout_demux_bpf(f, skb, num);
		break;
D
David S. Miller 已提交
1480 1481
	}

1482 1483
	if (fanout_has_flag(f, PACKET_FANOUT_FLAG_ROLLOVER))
		idx = fanout_demux_rollover(f, skb, idx, true, num);
D
David S. Miller 已提交
1484

1485
	po = pkt_sk(f->arr[idx]);
D
David S. Miller 已提交
1486 1487 1488
	return po->prot_hook.func(skb, dev, &po->prot_hook, orig_dev);
}

1489 1490
DEFINE_MUTEX(fanout_mutex);
EXPORT_SYMBOL_GPL(fanout_mutex);
D
David S. Miller 已提交
1491 1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509 1510 1511 1512 1513 1514 1515 1516 1517 1518 1519
static LIST_HEAD(fanout_list);

static void __fanout_link(struct sock *sk, struct packet_sock *po)
{
	struct packet_fanout *f = po->fanout;

	spin_lock(&f->lock);
	f->arr[f->num_members] = sk;
	smp_wmb();
	f->num_members++;
	spin_unlock(&f->lock);
}

static void __fanout_unlink(struct sock *sk, struct packet_sock *po)
{
	struct packet_fanout *f = po->fanout;
	int i;

	spin_lock(&f->lock);
	for (i = 0; i < f->num_members; i++) {
		if (f->arr[i] == sk)
			break;
	}
	BUG_ON(i >= f->num_members);
	f->arr[i] = f->arr[f->num_members - 1];
	f->num_members--;
	spin_unlock(&f->lock);
}

1520
static bool match_fanout_group(struct packet_type *ptype, struct sock *sk)
1521
{
E
Eric Dumazet 已提交
1522 1523
	if (sk->sk_family != PF_PACKET)
		return false;
1524

E
Eric Dumazet 已提交
1525
	return ptype->af_packet_priv == pkt_sk(sk)->fanout;
1526 1527
}

1528 1529 1530 1531 1532 1533 1534
static void fanout_init_data(struct packet_fanout *f)
{
	switch (f->type) {
	case PACKET_FANOUT_LB:
		atomic_set(&f->rr_cur, 0);
		break;
	case PACKET_FANOUT_CBPF:
1535
	case PACKET_FANOUT_EBPF:
1536 1537 1538 1539 1540 1541 1542 1543 1544 1545 1546 1547 1548 1549 1550 1551 1552 1553 1554 1555 1556 1557 1558 1559 1560 1561 1562 1563 1564 1565 1566 1567 1568 1569
		RCU_INIT_POINTER(f->bpf_prog, NULL);
		break;
	}
}

static void __fanout_set_data_bpf(struct packet_fanout *f, struct bpf_prog *new)
{
	struct bpf_prog *old;

	spin_lock(&f->lock);
	old = rcu_dereference_protected(f->bpf_prog, lockdep_is_held(&f->lock));
	rcu_assign_pointer(f->bpf_prog, new);
	spin_unlock(&f->lock);

	if (old) {
		synchronize_net();
		bpf_prog_destroy(old);
	}
}

static int fanout_set_data_cbpf(struct packet_sock *po, char __user *data,
				unsigned int len)
{
	struct bpf_prog *new;
	struct sock_fprog fprog;
	int ret;

	if (sock_flag(&po->sk, SOCK_FILTER_LOCKED))
		return -EPERM;
	if (len != sizeof(fprog))
		return -EINVAL;
	if (copy_from_user(&fprog, data, len))
		return -EFAULT;

1570
	ret = bpf_prog_create_from_user(&new, &fprog, NULL, false);
1571 1572 1573 1574 1575 1576 1577
	if (ret)
		return ret;

	__fanout_set_data_bpf(po->fanout, new);
	return 0;
}

1578 1579 1580 1581 1582 1583 1584 1585 1586 1587 1588 1589 1590 1591 1592 1593 1594 1595 1596 1597 1598 1599 1600 1601 1602
static int fanout_set_data_ebpf(struct packet_sock *po, char __user *data,
				unsigned int len)
{
	struct bpf_prog *new;
	u32 fd;

	if (sock_flag(&po->sk, SOCK_FILTER_LOCKED))
		return -EPERM;
	if (len != sizeof(fd))
		return -EINVAL;
	if (copy_from_user(&fd, data, len))
		return -EFAULT;

	new = bpf_prog_get(fd);
	if (IS_ERR(new))
		return PTR_ERR(new);
	if (new->type != BPF_PROG_TYPE_SOCKET_FILTER) {
		bpf_prog_put(new);
		return -EINVAL;
	}

	__fanout_set_data_bpf(po->fanout, new);
	return 0;
}

1603 1604 1605 1606 1607 1608
static int fanout_set_data(struct packet_sock *po, char __user *data,
			   unsigned int len)
{
	switch (po->fanout->type) {
	case PACKET_FANOUT_CBPF:
		return fanout_set_data_cbpf(po, data, len);
1609 1610
	case PACKET_FANOUT_EBPF:
		return fanout_set_data_ebpf(po, data, len);
1611 1612 1613 1614 1615 1616 1617 1618 1619
	default:
		return -EINVAL;
	};
}

static void fanout_release_data(struct packet_fanout *f)
{
	switch (f->type) {
	case PACKET_FANOUT_CBPF:
1620
	case PACKET_FANOUT_EBPF:
1621 1622 1623 1624
		__fanout_set_data_bpf(f, NULL);
	};
}

1625
static int fanout_add(struct sock *sk, u16 id, u16 type_flags)
D
David S. Miller 已提交
1626 1627 1628
{
	struct packet_sock *po = pkt_sk(sk);
	struct packet_fanout *f, *match;
1629
	u8 type = type_flags & 0xff;
1630
	u8 flags = type_flags >> 8;
D
David S. Miller 已提交
1631 1632 1633
	int err;

	switch (type) {
1634 1635 1636
	case PACKET_FANOUT_ROLLOVER:
		if (type_flags & PACKET_FANOUT_FLAG_ROLLOVER)
			return -EINVAL;
D
David S. Miller 已提交
1637 1638
	case PACKET_FANOUT_HASH:
	case PACKET_FANOUT_LB:
1639
	case PACKET_FANOUT_CPU:
1640
	case PACKET_FANOUT_RND:
1641
	case PACKET_FANOUT_QM:
1642
	case PACKET_FANOUT_CBPF:
1643
	case PACKET_FANOUT_EBPF:
D
David S. Miller 已提交
1644 1645 1646 1647 1648 1649 1650 1651 1652 1653 1654
		break;
	default:
		return -EINVAL;
	}

	if (!po->running)
		return -EINVAL;

	if (po->fanout)
		return -EALREADY;

1655 1656
	if (type == PACKET_FANOUT_ROLLOVER ||
	    (type_flags & PACKET_FANOUT_FLAG_ROLLOVER)) {
1657 1658 1659
		po->rollover = kzalloc(sizeof(*po->rollover), GFP_KERNEL);
		if (!po->rollover)
			return -ENOMEM;
W
Willem de Bruijn 已提交
1660 1661 1662
		atomic_long_set(&po->rollover->num, 0);
		atomic_long_set(&po->rollover->num_huge, 0);
		atomic_long_set(&po->rollover->num_failed, 0);
1663 1664
	}

D
David S. Miller 已提交
1665 1666 1667 1668 1669 1670 1671 1672 1673
	mutex_lock(&fanout_mutex);
	match = NULL;
	list_for_each_entry(f, &fanout_list, list) {
		if (f->id == id &&
		    read_pnet(&f->net) == sock_net(sk)) {
			match = f;
			break;
		}
	}
E
Eric Dumazet 已提交
1674
	err = -EINVAL;
1675
	if (match && match->flags != flags)
E
Eric Dumazet 已提交
1676
		goto out;
D
David S. Miller 已提交
1677
	if (!match) {
E
Eric Dumazet 已提交
1678
		err = -ENOMEM;
D
David S. Miller 已提交
1679
		match = kzalloc(sizeof(*match), GFP_KERNEL);
E
Eric Dumazet 已提交
1680 1681 1682 1683 1684
		if (!match)
			goto out;
		write_pnet(&match->net, sock_net(sk));
		match->id = id;
		match->type = type;
1685
		match->flags = flags;
E
Eric Dumazet 已提交
1686 1687 1688
		INIT_LIST_HEAD(&match->list);
		spin_lock_init(&match->lock);
		atomic_set(&match->sk_ref, 0);
1689
		fanout_init_data(match);
E
Eric Dumazet 已提交
1690 1691 1692 1693
		match->prot_hook.type = po->prot_hook.type;
		match->prot_hook.dev = po->prot_hook.dev;
		match->prot_hook.func = packet_rcv_fanout;
		match->prot_hook.af_packet_priv = match;
1694
		match->prot_hook.id_match = match_fanout_group;
E
Eric Dumazet 已提交
1695 1696
		dev_add_pack(&match->prot_hook);
		list_add(&match->list, &fanout_list);
D
David S. Miller 已提交
1697
	}
E
Eric Dumazet 已提交
1698 1699 1700 1701 1702 1703 1704 1705 1706 1707 1708
	err = -EINVAL;
	if (match->type == type &&
	    match->prot_hook.type == po->prot_hook.type &&
	    match->prot_hook.dev == po->prot_hook.dev) {
		err = -ENOSPC;
		if (atomic_read(&match->sk_ref) < PACKET_FANOUT_MAX) {
			__dev_remove_pack(&po->prot_hook);
			po->fanout = match;
			atomic_inc(&match->sk_ref);
			__fanout_link(sk, po);
			err = 0;
D
David S. Miller 已提交
1709 1710
		}
	}
E
Eric Dumazet 已提交
1711
out:
D
David S. Miller 已提交
1712
	mutex_unlock(&fanout_mutex);
1713 1714 1715 1716
	if (err) {
		kfree(po->rollover);
		po->rollover = NULL;
	}
D
David S. Miller 已提交
1717 1718 1719 1720 1721 1722 1723 1724 1725 1726 1727 1728
	return err;
}

static void fanout_release(struct sock *sk)
{
	struct packet_sock *po = pkt_sk(sk);
	struct packet_fanout *f;

	f = po->fanout;
	if (!f)
		return;

1729
	mutex_lock(&fanout_mutex);
D
David S. Miller 已提交
1730 1731 1732 1733 1734
	po->fanout = NULL;

	if (atomic_dec_and_test(&f->sk_ref)) {
		list_del(&f->list);
		dev_remove_pack(&f->prot_hook);
1735
		fanout_release_data(f);
D
David S. Miller 已提交
1736 1737 1738
		kfree(f);
	}
	mutex_unlock(&fanout_mutex);
1739

1740 1741
	if (po->rollover)
		kfree_rcu(po->rollover, rcu);
D
David S. Miller 已提交
1742
}
L
Linus Torvalds 已提交
1743

1744 1745 1746 1747 1748 1749 1750 1751 1752 1753 1754 1755 1756 1757
static bool packet_extra_vlan_len_allowed(const struct net_device *dev,
					  struct sk_buff *skb)
{
	/* Earlier code assumed this would be a VLAN pkt, double-check
	 * this now that we have the actual packet in hand. We can only
	 * do this check on Ethernet devices.
	 */
	if (unlikely(dev->type != ARPHRD_ETHER))
		return false;

	skb_reset_mac_header(skb);
	return likely(eth_hdr(skb)->h_proto == htons(ETH_P_8021Q));
}

1758
static const struct proto_ops packet_ops;
L
Linus Torvalds 已提交
1759

1760
static const struct proto_ops packet_ops_spkt;
L
Linus Torvalds 已提交
1761

E
Eric Dumazet 已提交
1762 1763
static int packet_rcv_spkt(struct sk_buff *skb, struct net_device *dev,
			   struct packet_type *pt, struct net_device *orig_dev)
L
Linus Torvalds 已提交
1764 1765 1766 1767 1768 1769 1770 1771 1772 1773
{
	struct sock *sk;
	struct sockaddr_pkt *spkt;

	/*
	 *	When we registered the protocol we saved the socket in the data
	 *	field for just this event.
	 */

	sk = pt->af_packet_priv;
1774

L
Linus Torvalds 已提交
1775 1776 1777 1778 1779 1780 1781
	/*
	 *	Yank back the headers [hope the device set this
	 *	right or kerboom...]
	 *
	 *	Incoming packets have ll header pulled,
	 *	push it back.
	 *
1782
	 *	For outgoing ones skb->data == skb_mac_header(skb)
L
Linus Torvalds 已提交
1783 1784 1785 1786 1787 1788
	 *	so that this procedure is noop.
	 */

	if (skb->pkt_type == PACKET_LOOPBACK)
		goto out;

O
Octavian Purdila 已提交
1789
	if (!net_eq(dev_net(dev), sock_net(sk)))
1790 1791
		goto out;

E
Eric Dumazet 已提交
1792 1793
	skb = skb_share_check(skb, GFP_ATOMIC);
	if (skb == NULL)
L
Linus Torvalds 已提交
1794 1795 1796
		goto oom;

	/* drop any routing info */
E
Eric Dumazet 已提交
1797
	skb_dst_drop(skb);
L
Linus Torvalds 已提交
1798

1799 1800 1801
	/* drop conntrack reference */
	nf_reset(skb);

1802
	spkt = &PACKET_SKB_CB(skb)->sa.pkt;
L
Linus Torvalds 已提交
1803

1804
	skb_push(skb, skb->data - skb_mac_header(skb));
L
Linus Torvalds 已提交
1805 1806 1807 1808 1809 1810 1811 1812 1813 1814 1815 1816 1817 1818

	/*
	 *	The SOCK_PACKET socket receives _all_ frames.
	 */

	spkt->spkt_family = dev->type;
	strlcpy(spkt->spkt_device, dev->name, sizeof(spkt->spkt_device));
	spkt->spkt_protocol = skb->protocol;

	/*
	 *	Charge the memory to the socket. This is done specifically
	 *	to prevent sockets using all the memory up.
	 */

E
Eric Dumazet 已提交
1819
	if (sock_queue_rcv_skb(sk, skb) == 0)
L
Linus Torvalds 已提交
1820 1821 1822 1823 1824 1825 1826 1827 1828 1829 1830 1831 1832
		return 0;

out:
	kfree_skb(skb);
oom:
	return 0;
}


/*
 *	Output a raw packet to a device layer. This bypasses all the other
 *	protocol layers and you must therefore supply it with a complete frame
 */
1833

1834 1835
static int packet_sendmsg_spkt(struct socket *sock, struct msghdr *msg,
			       size_t len)
L
Linus Torvalds 已提交
1836 1837
{
	struct sock *sk = sock->sk;
1838
	DECLARE_SOCKADDR(struct sockaddr_pkt *, saddr, msg->msg_name);
1839
	struct sk_buff *skb = NULL;
L
Linus Torvalds 已提交
1840
	struct net_device *dev;
E
Eric Dumazet 已提交
1841
	__be16 proto = 0;
L
Linus Torvalds 已提交
1842
	int err;
1843
	int extra_len = 0;
1844

L
Linus Torvalds 已提交
1845
	/*
1846
	 *	Get and verify the address.
L
Linus Torvalds 已提交
1847 1848
	 */

E
Eric Dumazet 已提交
1849
	if (saddr) {
L
Linus Torvalds 已提交
1850
		if (msg->msg_namelen < sizeof(struct sockaddr))
E
Eric Dumazet 已提交
1851 1852 1853 1854 1855
			return -EINVAL;
		if (msg->msg_namelen == sizeof(struct sockaddr_pkt))
			proto = saddr->spkt_protocol;
	} else
		return -ENOTCONN;	/* SOCK_PACKET must be sent giving an address */
L
Linus Torvalds 已提交
1856 1857

	/*
1858
	 *	Find the device first to size check it
L
Linus Torvalds 已提交
1859 1860
	 */

1861
	saddr->spkt_device[sizeof(saddr->spkt_device) - 1] = 0;
1862
retry:
E
Eric Dumazet 已提交
1863 1864
	rcu_read_lock();
	dev = dev_get_by_name_rcu(sock_net(sk), saddr->spkt_device);
L
Linus Torvalds 已提交
1865 1866 1867
	err = -ENODEV;
	if (dev == NULL)
		goto out_unlock;
1868

1869 1870 1871 1872
	err = -ENETDOWN;
	if (!(dev->flags & IFF_UP))
		goto out_unlock;

L
Linus Torvalds 已提交
1873
	/*
E
Eric Dumazet 已提交
1874 1875
	 * You may not queue a frame bigger than the mtu. This is the lowest level
	 * raw protocol and you must do your own fragmentation at this level.
L
Linus Torvalds 已提交
1876
	 */
1877

1878 1879 1880 1881 1882 1883 1884 1885
	if (unlikely(sock_flag(sk, SOCK_NOFCS))) {
		if (!netif_supports_nofcs(dev)) {
			err = -EPROTONOSUPPORT;
			goto out_unlock;
		}
		extra_len = 4; /* We're doing our own CRC */
	}

L
Linus Torvalds 已提交
1886
	err = -EMSGSIZE;
1887
	if (len > dev->mtu + dev->hard_header_len + VLAN_HLEN + extra_len)
L
Linus Torvalds 已提交
1888 1889
		goto out_unlock;

1890 1891
	if (!skb) {
		size_t reserved = LL_RESERVED_SPACE(dev);
1892
		int tlen = dev->needed_tailroom;
1893 1894 1895
		unsigned int hhlen = dev->header_ops ? dev->hard_header_len : 0;

		rcu_read_unlock();
1896
		skb = sock_wmalloc(sk, len + reserved + tlen, 0, GFP_KERNEL);
1897 1898 1899 1900 1901 1902 1903 1904 1905 1906 1907 1908 1909 1910 1911 1912
		if (skb == NULL)
			return -ENOBUFS;
		/* FIXME: Save some space for broken drivers that write a hard
		 * header at transmission time by themselves. PPP is the notable
		 * one here. This should really be fixed at the driver level.
		 */
		skb_reserve(skb, reserved);
		skb_reset_network_header(skb);

		/* Try to align data part correctly */
		if (hhlen) {
			skb->data -= hhlen;
			skb->tail -= hhlen;
			if (len < hhlen)
				skb_reset_network_header(skb);
		}
A
Al Viro 已提交
1913
		err = memcpy_from_msg(skb_put(skb, len), msg, len);
1914 1915 1916
		if (err)
			goto out_free;
		goto retry;
L
Linus Torvalds 已提交
1917 1918
	}

1919 1920 1921 1922
	if (len > (dev->mtu + dev->hard_header_len + extra_len) &&
	    !packet_extra_vlan_len_allowed(dev, skb)) {
		err = -EMSGSIZE;
		goto out_unlock;
1923
	}
1924

L
Linus Torvalds 已提交
1925 1926 1927
	skb->protocol = proto;
	skb->dev = dev;
	skb->priority = sk->sk_priority;
1928
	skb->mark = sk->sk_mark;
1929 1930

	sock_tx_timestamp(sk, &skb_shinfo(skb)->tx_flags);
L
Linus Torvalds 已提交
1931

1932 1933 1934
	if (unlikely(extra_len == 4))
		skb->no_fcs = 1;

1935
	skb_probe_transport_header(skb, 0);
1936

L
Linus Torvalds 已提交
1937
	dev_queue_xmit(skb);
E
Eric Dumazet 已提交
1938
	rcu_read_unlock();
E
Eric Dumazet 已提交
1939
	return len;
L
Linus Torvalds 已提交
1940 1941

out_unlock:
E
Eric Dumazet 已提交
1942
	rcu_read_unlock();
1943 1944
out_free:
	kfree_skb(skb);
L
Linus Torvalds 已提交
1945 1946 1947
	return err;
}

1948 1949 1950
static unsigned int run_filter(struct sk_buff *skb,
			       const struct sock *sk,
			       unsigned int res)
L
Linus Torvalds 已提交
1951 1952
{
	struct sk_filter *filter;
1953

1954 1955
	rcu_read_lock();
	filter = rcu_dereference(sk->sk_filter);
D
David S. Miller 已提交
1956
	if (filter != NULL)
1957
		res = bpf_prog_run_clear_cb(filter->prog, skb);
1958
	rcu_read_unlock();
L
Linus Torvalds 已提交
1959

D
David S. Miller 已提交
1960
	return res;
L
Linus Torvalds 已提交
1961 1962 1963
}

/*
E
Eric Dumazet 已提交
1964 1965 1966 1967 1968 1969 1970 1971 1972
 * This function makes lazy skb cloning in hope that most of packets
 * are discarded by BPF.
 *
 * Note tricky part: we DO mangle shared skb! skb->data, skb->len
 * and skb->cb are mangled. It works because (and until) packets
 * falling here are owned by current CPU. Output packets are cloned
 * by dev_queue_xmit_nit(), input packets are processed by net_bh
 * sequencially, so that if we return skb to original state on exit,
 * we will not harm anyone.
L
Linus Torvalds 已提交
1973 1974
 */

E
Eric Dumazet 已提交
1975 1976
static int packet_rcv(struct sk_buff *skb, struct net_device *dev,
		      struct packet_type *pt, struct net_device *orig_dev)
L
Linus Torvalds 已提交
1977 1978 1979 1980
{
	struct sock *sk;
	struct sockaddr_ll *sll;
	struct packet_sock *po;
E
Eric Dumazet 已提交
1981
	u8 *skb_head = skb->data;
L
Linus Torvalds 已提交
1982
	int skb_len = skb->len;
D
David S. Miller 已提交
1983
	unsigned int snaplen, res;
L
Linus Torvalds 已提交
1984 1985 1986 1987 1988 1989 1990

	if (skb->pkt_type == PACKET_LOOPBACK)
		goto drop;

	sk = pt->af_packet_priv;
	po = pkt_sk(sk);

O
Octavian Purdila 已提交
1991
	if (!net_eq(dev_net(dev), sock_net(sk)))
1992 1993
		goto drop;

L
Linus Torvalds 已提交
1994 1995
	skb->dev = dev;

1996
	if (dev->header_ops) {
L
Linus Torvalds 已提交
1997
		/* The device has an explicit notion of ll header,
E
Eric Dumazet 已提交
1998 1999 2000 2001 2002
		 * exported to higher levels.
		 *
		 * Otherwise, the device hides details of its frame
		 * structure, so that corresponding packet head is
		 * never delivered to user.
L
Linus Torvalds 已提交
2003 2004
		 */
		if (sk->sk_type != SOCK_DGRAM)
2005
			skb_push(skb, skb->data - skb_mac_header(skb));
L
Linus Torvalds 已提交
2006 2007
		else if (skb->pkt_type == PACKET_OUTGOING) {
			/* Special case: outgoing packets have ll header at head */
2008
			skb_pull(skb, skb_network_offset(skb));
L
Linus Torvalds 已提交
2009 2010 2011 2012 2013
		}
	}

	snaplen = skb->len;

D
David S. Miller 已提交
2014 2015
	res = run_filter(skb, sk, snaplen);
	if (!res)
2016
		goto drop_n_restore;
D
David S. Miller 已提交
2017 2018
	if (snaplen > res)
		snaplen = res;
L
Linus Torvalds 已提交
2019

E
Eric Dumazet 已提交
2020
	if (atomic_read(&sk->sk_rmem_alloc) >= sk->sk_rcvbuf)
L
Linus Torvalds 已提交
2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031
		goto drop_n_acct;

	if (skb_shared(skb)) {
		struct sk_buff *nskb = skb_clone(skb, GFP_ATOMIC);
		if (nskb == NULL)
			goto drop_n_acct;

		if (skb_head != skb->data) {
			skb->data = skb_head;
			skb->len = skb_len;
		}
2032
		consume_skb(skb);
L
Linus Torvalds 已提交
2033 2034 2035
		skb = nskb;
	}

2036
	sock_skb_cb_check_size(sizeof(*PACKET_SKB_CB(skb)) + MAX_ADDR_LEN - 8);
2037 2038

	sll = &PACKET_SKB_CB(skb)->sa.ll;
L
Linus Torvalds 已提交
2039 2040
	sll->sll_hatype = dev->type;
	sll->sll_pkttype = skb->pkt_type;
2041
	if (unlikely(po->origdev))
2042 2043 2044
		sll->sll_ifindex = orig_dev->ifindex;
	else
		sll->sll_ifindex = dev->ifindex;
L
Linus Torvalds 已提交
2045

S
Stephen Hemminger 已提交
2046
	sll->sll_halen = dev_parse_header(skb, sll->sll_addr);
L
Linus Torvalds 已提交
2047

2048 2049 2050 2051
	/* sll->sll_family and sll->sll_protocol are set in packet_recvmsg().
	 * Use their space for storing the original skb length.
	 */
	PACKET_SKB_CB(skb)->sa.origlen = skb->len;
2052

L
Linus Torvalds 已提交
2053 2054 2055 2056 2057
	if (pskb_trim(skb, snaplen))
		goto drop_n_acct;

	skb_set_owner_r(skb, sk);
	skb->dev = NULL;
E
Eric Dumazet 已提交
2058
	skb_dst_drop(skb);
L
Linus Torvalds 已提交
2059

2060 2061 2062
	/* drop conntrack reference */
	nf_reset(skb);

L
Linus Torvalds 已提交
2063
	spin_lock(&sk->sk_receive_queue.lock);
2064
	po->stats.stats1.tp_packets++;
2065
	sock_skb_set_dropcount(sk, skb);
L
Linus Torvalds 已提交
2066 2067
	__skb_queue_tail(&sk->sk_receive_queue, skb);
	spin_unlock(&sk->sk_receive_queue.lock);
2068
	sk->sk_data_ready(sk);
L
Linus Torvalds 已提交
2069 2070 2071
	return 0;

drop_n_acct:
2072
	spin_lock(&sk->sk_receive_queue.lock);
2073
	po->stats.stats1.tp_drops++;
2074 2075
	atomic_inc(&sk->sk_drops);
	spin_unlock(&sk->sk_receive_queue.lock);
L
Linus Torvalds 已提交
2076 2077 2078 2079 2080 2081 2082

drop_n_restore:
	if (skb_head != skb->data && skb_shared(skb)) {
		skb->data = skb_head;
		skb->len = skb_len;
	}
drop:
2083
	consume_skb(skb);
L
Linus Torvalds 已提交
2084 2085 2086
	return 0;
}

E
Eric Dumazet 已提交
2087 2088
static int tpacket_rcv(struct sk_buff *skb, struct net_device *dev,
		       struct packet_type *pt, struct net_device *orig_dev)
L
Linus Torvalds 已提交
2089 2090 2091 2092
{
	struct sock *sk;
	struct packet_sock *po;
	struct sockaddr_ll *sll;
2093
	union tpacket_uhdr h;
E
Eric Dumazet 已提交
2094
	u8 *skb_head = skb->data;
L
Linus Torvalds 已提交
2095
	int skb_len = skb->len;
D
David S. Miller 已提交
2096
	unsigned int snaplen, res;
2097
	unsigned long status = TP_STATUS_USER;
2098
	unsigned short macoff, netoff, hdrlen;
L
Linus Torvalds 已提交
2099
	struct sk_buff *copy_skb = NULL;
2100
	struct timespec ts;
2101
	__u32 ts_status;
L
Linus Torvalds 已提交
2102

2103 2104 2105 2106 2107 2108 2109
	/* struct tpacket{2,3}_hdr is aligned to a multiple of TPACKET_ALIGNMENT.
	 * We may add members to them until current aligned size without forcing
	 * userspace to call getsockopt(..., PACKET_HDRLEN, ...).
	 */
	BUILD_BUG_ON(TPACKET_ALIGN(sizeof(*h.h2)) != 32);
	BUILD_BUG_ON(TPACKET_ALIGN(sizeof(*h.h3)) != 48);

L
Linus Torvalds 已提交
2110 2111 2112 2113 2114 2115
	if (skb->pkt_type == PACKET_LOOPBACK)
		goto drop;

	sk = pt->af_packet_priv;
	po = pkt_sk(sk);

O
Octavian Purdila 已提交
2116
	if (!net_eq(dev_net(dev), sock_net(sk)))
2117 2118
		goto drop;

2119
	if (dev->header_ops) {
L
Linus Torvalds 已提交
2120
		if (sk->sk_type != SOCK_DGRAM)
2121
			skb_push(skb, skb->data - skb_mac_header(skb));
L
Linus Torvalds 已提交
2122 2123
		else if (skb->pkt_type == PACKET_OUTGOING) {
			/* Special case: outgoing packets have ll header at head */
2124
			skb_pull(skb, skb_network_offset(skb));
L
Linus Torvalds 已提交
2125 2126 2127 2128 2129
		}
	}

	snaplen = skb->len;

D
David S. Miller 已提交
2130 2131
	res = run_filter(skb, sk, snaplen);
	if (!res)
2132
		goto drop_n_restore;
2133 2134 2135

	if (skb->ip_summed == CHECKSUM_PARTIAL)
		status |= TP_STATUS_CSUMNOTREADY;
2136 2137 2138 2139
	else if (skb->pkt_type != PACKET_OUTGOING &&
		 (skb->ip_summed == CHECKSUM_COMPLETE ||
		  skb_csum_unnecessary(skb)))
		status |= TP_STATUS_CSUM_VALID;
2140

D
David S. Miller 已提交
2141 2142
	if (snaplen > res)
		snaplen = res;
L
Linus Torvalds 已提交
2143 2144

	if (sk->sk_type == SOCK_DGRAM) {
2145 2146
		macoff = netoff = TPACKET_ALIGN(po->tp_hdrlen) + 16 +
				  po->tp_reserve;
L
Linus Torvalds 已提交
2147
	} else {
2148
		unsigned int maclen = skb_network_offset(skb);
2149
		netoff = TPACKET_ALIGN(po->tp_hdrlen +
2150 2151
				       (maclen < 16 ? 16 : maclen)) +
			po->tp_reserve;
L
Linus Torvalds 已提交
2152 2153
		macoff = netoff - maclen;
	}
2154 2155 2156
	if (po->tp_version <= TPACKET_V2) {
		if (macoff + snaplen > po->rx_ring.frame_size) {
			if (po->copy_thresh &&
E
Eric Dumazet 已提交
2157
			    atomic_read(&sk->sk_rmem_alloc) < sk->sk_rcvbuf) {
2158 2159 2160 2161 2162 2163 2164 2165
				if (skb_shared(skb)) {
					copy_skb = skb_clone(skb, GFP_ATOMIC);
				} else {
					copy_skb = skb_get(skb);
					skb_head = skb->data;
				}
				if (copy_skb)
					skb_set_owner_r(copy_skb, sk);
L
Linus Torvalds 已提交
2166
			}
2167 2168 2169
			snaplen = po->rx_ring.frame_size - macoff;
			if ((int)snaplen < 0)
				snaplen = 0;
L
Linus Torvalds 已提交
2170
		}
2171 2172 2173 2174 2175 2176 2177 2178 2179 2180 2181 2182
	} else if (unlikely(macoff + snaplen >
			    GET_PBDQC_FROM_RB(&po->rx_ring)->max_frame_len)) {
		u32 nval;

		nval = GET_PBDQC_FROM_RB(&po->rx_ring)->max_frame_len - macoff;
		pr_err_once("tpacket_rcv: packet too big, clamped from %u to %u. macoff=%u\n",
			    snaplen, nval, macoff);
		snaplen = nval;
		if (unlikely((int)snaplen < 0)) {
			snaplen = 0;
			macoff = GET_PBDQC_FROM_RB(&po->rx_ring)->max_frame_len;
		}
L
Linus Torvalds 已提交
2183 2184
	}
	spin_lock(&sk->sk_receive_queue.lock);
2185 2186
	h.raw = packet_current_rx_frame(po, skb,
					TP_STATUS_KERNEL, (macoff+snaplen));
2187
	if (!h.raw)
L
Linus Torvalds 已提交
2188
		goto ring_is_full;
2189 2190 2191 2192 2193 2194 2195 2196
	if (po->tp_version <= TPACKET_V2) {
		packet_increment_rx_head(po, &po->rx_ring);
	/*
	 * LOSING will be reported till you read the stats,
	 * because it's COR - Clear On Read.
	 * Anyways, moving it for V1/V2 only as V3 doesn't need this
	 * at packet level.
	 */
2197
		if (po->stats.stats1.tp_drops)
2198 2199
			status |= TP_STATUS_LOSING;
	}
2200
	po->stats.stats1.tp_packets++;
L
Linus Torvalds 已提交
2201 2202 2203 2204 2205 2206
	if (copy_skb) {
		status |= TP_STATUS_COPY;
		__skb_queue_tail(&sk->sk_receive_queue, copy_skb);
	}
	spin_unlock(&sk->sk_receive_queue.lock);

2207
	skb_copy_bits(skb, 0, h.raw + macoff, snaplen);
2208 2209

	if (!(ts_status = tpacket_get_timestamp(skb, &ts, po->tp_tstamp)))
2210
		getnstimeofday(&ts);
L
Linus Torvalds 已提交
2211

2212 2213
	status |= ts_status;

2214 2215 2216 2217 2218 2219
	switch (po->tp_version) {
	case TPACKET_V1:
		h.h1->tp_len = skb->len;
		h.h1->tp_snaplen = snaplen;
		h.h1->tp_mac = macoff;
		h.h1->tp_net = netoff;
2220 2221
		h.h1->tp_sec = ts.tv_sec;
		h.h1->tp_usec = ts.tv_nsec / NSEC_PER_USEC;
2222 2223 2224 2225 2226 2227 2228 2229 2230
		hdrlen = sizeof(*h.h1);
		break;
	case TPACKET_V2:
		h.h2->tp_len = skb->len;
		h.h2->tp_snaplen = snaplen;
		h.h2->tp_mac = macoff;
		h.h2->tp_net = netoff;
		h.h2->tp_sec = ts.tv_sec;
		h.h2->tp_nsec = ts.tv_nsec;
2231 2232
		if (skb_vlan_tag_present(skb)) {
			h.h2->tp_vlan_tci = skb_vlan_tag_get(skb);
2233 2234
			h.h2->tp_vlan_tpid = ntohs(skb->vlan_proto);
			status |= TP_STATUS_VLAN_VALID | TP_STATUS_VLAN_TPID_VALID;
2235 2236
		} else {
			h.h2->tp_vlan_tci = 0;
2237
			h.h2->tp_vlan_tpid = 0;
2238
		}
2239
		memset(h.h2->tp_padding, 0, sizeof(h.h2->tp_padding));
2240 2241
		hdrlen = sizeof(*h.h2);
		break;
2242 2243 2244 2245 2246 2247 2248 2249 2250 2251 2252
	case TPACKET_V3:
		/* tp_nxt_offset,vlan are already populated above.
		 * So DONT clear those fields here
		 */
		h.h3->tp_status |= status;
		h.h3->tp_len = skb->len;
		h.h3->tp_snaplen = snaplen;
		h.h3->tp_mac = macoff;
		h.h3->tp_net = netoff;
		h.h3->tp_sec  = ts.tv_sec;
		h.h3->tp_nsec = ts.tv_nsec;
2253
		memset(h.h3->tp_padding, 0, sizeof(h.h3->tp_padding));
2254 2255
		hdrlen = sizeof(*h.h3);
		break;
2256 2257 2258
	default:
		BUG();
	}
L
Linus Torvalds 已提交
2259

2260
	sll = h.raw + TPACKET_ALIGN(hdrlen);
S
Stephen Hemminger 已提交
2261
	sll->sll_halen = dev_parse_header(skb, sll->sll_addr);
L
Linus Torvalds 已提交
2262 2263 2264 2265
	sll->sll_family = AF_PACKET;
	sll->sll_hatype = dev->type;
	sll->sll_protocol = skb->protocol;
	sll->sll_pkttype = skb->pkt_type;
2266
	if (unlikely(po->origdev))
2267 2268 2269
		sll->sll_ifindex = orig_dev->ifindex;
	else
		sll->sll_ifindex = dev->ifindex;
L
Linus Torvalds 已提交
2270

R
Ralf Baechle 已提交
2271
	smp_mb();
2272

2273
#if ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE == 1
2274
	if (po->tp_version <= TPACKET_V2) {
2275 2276
		u8 *start, *end;

2277 2278 2279 2280 2281
		end = (u8 *) PAGE_ALIGN((unsigned long) h.raw +
					macoff + snaplen);

		for (start = h.raw; start < end; start += PAGE_SIZE)
			flush_dcache_page(pgv_to_page(start));
L
Linus Torvalds 已提交
2282
	}
2283
	smp_wmb();
2284
#endif
2285

2286
	if (po->tp_version <= TPACKET_V2) {
2287
		__packet_set_status(po, h.raw, status);
2288 2289
		sk->sk_data_ready(sk);
	} else {
2290
		prb_clear_blk_fill_status(&po->rx_ring);
2291
	}
L
Linus Torvalds 已提交
2292 2293 2294 2295 2296 2297 2298

drop_n_restore:
	if (skb_head != skb->data && skb_shared(skb)) {
		skb->data = skb_head;
		skb->len = skb_len;
	}
drop:
2299
	kfree_skb(skb);
L
Linus Torvalds 已提交
2300 2301 2302
	return 0;

ring_is_full:
2303
	po->stats.stats1.tp_drops++;
L
Linus Torvalds 已提交
2304 2305
	spin_unlock(&sk->sk_receive_queue.lock);

2306
	sk->sk_data_ready(sk);
2307
	kfree_skb(copy_skb);
L
Linus Torvalds 已提交
2308 2309 2310
	goto drop_n_restore;
}

J
Johann Baudy 已提交
2311 2312 2313
static void tpacket_destruct_skb(struct sk_buff *skb)
{
	struct packet_sock *po = pkt_sk(skb->sk);
L
Linus Torvalds 已提交
2314

J
Johann Baudy 已提交
2315
	if (likely(po->tx_ring.pg_vec)) {
2316
		void *ph;
2317 2318
		__u32 ts;

J
Johann Baudy 已提交
2319
		ph = skb_shinfo(skb)->destructor_arg;
2320
		packet_dec_pending(&po->tx_ring);
2321 2322 2323

		ts = __packet_set_timestamp(po, ph, skb);
		__packet_set_status(po, ph, TP_STATUS_AVAILABLE | ts);
J
Johann Baudy 已提交
2324 2325 2326 2327 2328
	}

	sock_wfree(skb);
}

2329 2330 2331 2332
static bool ll_header_truncated(const struct net_device *dev, int len)
{
	/* net device doesn't like empty head */
	if (unlikely(len <= dev->hard_header_len)) {
2333
		net_warn_ratelimited("%s: packet size is too short (%d <= %d)\n",
2334 2335 2336 2337 2338 2339 2340
				     current->comm, len, dev->hard_header_len);
		return true;
	}

	return false;
}

2341 2342 2343 2344 2345 2346 2347 2348 2349
static void tpacket_set_protocol(const struct net_device *dev,
				 struct sk_buff *skb)
{
	if (dev->type == ARPHRD_ETHER) {
		skb_reset_mac_header(skb);
		skb->protocol = eth_hdr(skb)->h_proto;
	}
}

E
Eric Dumazet 已提交
2350 2351
static int tpacket_fill_skb(struct packet_sock *po, struct sk_buff *skb,
		void *frame, struct net_device *dev, int size_max,
2352
		__be16 proto, unsigned char *addr, int hlen)
J
Johann Baudy 已提交
2353
{
2354
	union tpacket_uhdr ph;
2355
	int to_write, offset, len, tp_len, nr_frags, len_max;
J
Johann Baudy 已提交
2356 2357 2358 2359 2360 2361 2362 2363 2364 2365
	struct socket *sock = po->sk.sk_socket;
	struct page *page;
	void *data;
	int err;

	ph.raw = frame;

	skb->protocol = proto;
	skb->dev = dev;
	skb->priority = po->sk.sk_priority;
2366
	skb->mark = po->sk.sk_mark;
2367
	sock_tx_timestamp(&po->sk, &skb_shinfo(skb)->tx_flags);
J
Johann Baudy 已提交
2368 2369 2370 2371 2372 2373 2374 2375 2376 2377
	skb_shinfo(skb)->destructor_arg = ph.raw;

	switch (po->tp_version) {
	case TPACKET_V2:
		tp_len = ph.h2->tp_len;
		break;
	default:
		tp_len = ph.h1->tp_len;
		break;
	}
2378 2379 2380 2381
	if (unlikely(tp_len > size_max)) {
		pr_err("packet size is too long (%d > %d)\n", tp_len, size_max);
		return -EMSGSIZE;
	}
J
Johann Baudy 已提交
2382

2383
	skb_reserve(skb, hlen);
J
Johann Baudy 已提交
2384
	skb_reset_network_header(skb);
2385

2386
	if (unlikely(po->tp_tx_has_off)) {
2387 2388 2389 2390 2391 2392 2393 2394 2395 2396 2397 2398 2399 2400 2401 2402 2403 2404 2405 2406 2407 2408 2409 2410 2411 2412 2413 2414
		int off_min, off_max, off;
		off_min = po->tp_hdrlen - sizeof(struct sockaddr_ll);
		off_max = po->tx_ring.frame_size - tp_len;
		if (sock->type == SOCK_DGRAM) {
			switch (po->tp_version) {
			case TPACKET_V2:
				off = ph.h2->tp_net;
				break;
			default:
				off = ph.h1->tp_net;
				break;
			}
		} else {
			switch (po->tp_version) {
			case TPACKET_V2:
				off = ph.h2->tp_mac;
				break;
			default:
				off = ph.h1->tp_mac;
				break;
			}
		}
		if (unlikely((off < off_min) || (off_max < off)))
			return -EINVAL;
		data = ph.raw + off;
	} else {
		data = ph.raw + po->tp_hdrlen - sizeof(struct sockaddr_ll);
	}
J
Johann Baudy 已提交
2415 2416 2417 2418 2419 2420 2421
	to_write = tp_len;

	if (sock->type == SOCK_DGRAM) {
		err = dev_hard_header(skb, dev, ntohs(proto), addr,
				NULL, tp_len);
		if (unlikely(err < 0))
			return -EINVAL;
E
Eric Dumazet 已提交
2422
	} else if (dev->hard_header_len) {
2423
		if (ll_header_truncated(dev, tp_len))
J
Johann Baudy 已提交
2424 2425 2426 2427 2428 2429 2430
			return -EINVAL;

		skb_push(skb, dev->hard_header_len);
		err = skb_store_bits(skb, 0, data,
				dev->hard_header_len);
		if (unlikely(err))
			return err;
2431 2432
		if (!skb->protocol)
			tpacket_set_protocol(dev, skb);
J
Johann Baudy 已提交
2433 2434 2435 2436 2437 2438 2439 2440 2441 2442 2443 2444 2445 2446 2447 2448 2449 2450

		data += dev->hard_header_len;
		to_write -= dev->hard_header_len;
	}

	offset = offset_in_page(data);
	len_max = PAGE_SIZE - offset;
	len = ((to_write > len_max) ? len_max : to_write);

	skb->data_len = to_write;
	skb->len += to_write;
	skb->truesize += to_write;
	atomic_add(to_write, &po->sk.sk_wmem_alloc);

	while (likely(to_write)) {
		nr_frags = skb_shinfo(skb)->nr_frags;

		if (unlikely(nr_frags >= MAX_SKB_FRAGS)) {
E
Eric Dumazet 已提交
2451 2452
			pr_err("Packet exceed the number of skb frags(%lu)\n",
			       MAX_SKB_FRAGS);
J
Johann Baudy 已提交
2453 2454 2455
			return -EFAULT;
		}

2456 2457
		page = pgv_to_page(data);
		data += len;
J
Johann Baudy 已提交
2458 2459
		flush_dcache_page(page);
		get_page(page);
2460
		skb_fill_page_desc(skb, nr_frags, page, offset, len);
J
Johann Baudy 已提交
2461 2462 2463 2464 2465 2466
		to_write -= len;
		offset = 0;
		len_max = PAGE_SIZE;
		len = ((to_write > len_max) ? len_max : to_write);
	}

2467
	skb_probe_transport_header(skb, 0);
2468

J
Johann Baudy 已提交
2469 2470 2471 2472 2473 2474 2475 2476
	return tp_len;
}

static int tpacket_snd(struct packet_sock *po, struct msghdr *msg)
{
	struct sk_buff *skb;
	struct net_device *dev;
	__be16 proto;
2477
	int err, reserve = 0;
E
Eric Dumazet 已提交
2478
	void *ph;
2479
	DECLARE_SOCKADDR(struct sockaddr_ll *, saddr, msg->msg_name);
2480
	bool need_wait = !(msg->msg_flags & MSG_DONTWAIT);
J
Johann Baudy 已提交
2481 2482 2483
	int tp_len, size_max;
	unsigned char *addr;
	int len_sum = 0;
2484
	int status = TP_STATUS_AVAILABLE;
2485
	int hlen, tlen;
J
Johann Baudy 已提交
2486 2487 2488

	mutex_lock(&po->pg_vec_lock);

2489
	if (likely(saddr == NULL)) {
2490
		dev	= packet_cached_dev_get(po);
J
Johann Baudy 已提交
2491 2492 2493 2494 2495 2496 2497 2498 2499 2500 2501 2502
		proto	= po->num;
		addr	= NULL;
	} else {
		err = -EINVAL;
		if (msg->msg_namelen < sizeof(struct sockaddr_ll))
			goto out;
		if (msg->msg_namelen < (saddr->sll_halen
					+ offsetof(struct sockaddr_ll,
						sll_addr)))
			goto out;
		proto	= saddr->sll_protocol;
		addr	= saddr->sll_addr;
2503
		dev = dev_get_by_index(sock_net(&po->sk), saddr->sll_ifindex);
J
Johann Baudy 已提交
2504 2505 2506 2507 2508 2509 2510 2511 2512
	}

	err = -ENXIO;
	if (unlikely(dev == NULL))
		goto out;
	err = -ENETDOWN;
	if (unlikely(!(dev->flags & IFF_UP)))
		goto out_put;

2513 2514
	if (po->sk.sk_socket->type == SOCK_RAW)
		reserve = dev->hard_header_len;
J
Johann Baudy 已提交
2515
	size_max = po->tx_ring.frame_size
2516
		- (po->tp_hdrlen - sizeof(struct sockaddr_ll));
J
Johann Baudy 已提交
2517

2518 2519
	if (size_max > dev->mtu + reserve + VLAN_HLEN)
		size_max = dev->mtu + reserve + VLAN_HLEN;
2520

J
Johann Baudy 已提交
2521 2522
	do {
		ph = packet_current_frame(po, &po->tx_ring,
2523
					  TP_STATUS_SEND_REQUEST);
J
Johann Baudy 已提交
2524
		if (unlikely(ph == NULL)) {
2525 2526
			if (need_wait && need_resched())
				schedule();
J
Johann Baudy 已提交
2527 2528 2529 2530
			continue;
		}

		status = TP_STATUS_SEND_REQUEST;
2531 2532
		hlen = LL_RESERVED_SPACE(dev);
		tlen = dev->needed_tailroom;
J
Johann Baudy 已提交
2533
		skb = sock_alloc_send_skb(&po->sk,
2534
				hlen + tlen + sizeof(struct sockaddr_ll),
2535
				!need_wait, &err);
J
Johann Baudy 已提交
2536

2537 2538 2539 2540
		if (unlikely(skb == NULL)) {
			/* we assume the socket was initially writeable ... */
			if (likely(len_sum > 0))
				err = len_sum;
J
Johann Baudy 已提交
2541
			goto out_status;
2542
		}
J
Johann Baudy 已提交
2543
		tp_len = tpacket_fill_skb(po, skb, ph, dev, size_max, proto,
2544
					  addr, hlen);
2545
		if (likely(tp_len >= 0) &&
2546
		    tp_len > dev->mtu + reserve &&
2547 2548
		    !packet_extra_vlan_len_allowed(dev, skb))
			tp_len = -EMSGSIZE;
J
Johann Baudy 已提交
2549 2550 2551 2552 2553 2554 2555 2556 2557 2558 2559 2560 2561 2562 2563

		if (unlikely(tp_len < 0)) {
			if (po->tp_loss) {
				__packet_set_status(po, ph,
						TP_STATUS_AVAILABLE);
				packet_increment_head(&po->tx_ring);
				kfree_skb(skb);
				continue;
			} else {
				status = TP_STATUS_WRONG_FORMAT;
				err = tp_len;
				goto out_status;
			}
		}

2564 2565
		packet_pick_tx_queue(dev, skb);

J
Johann Baudy 已提交
2566 2567
		skb->destructor = tpacket_destruct_skb;
		__packet_set_status(po, ph, TP_STATUS_SENDING);
2568
		packet_inc_pending(&po->tx_ring);
J
Johann Baudy 已提交
2569 2570

		status = TP_STATUS_SEND_REQUEST;
2571
		err = po->xmit(skb);
2572 2573 2574 2575 2576 2577 2578 2579 2580 2581 2582 2583 2584 2585
		if (unlikely(err > 0)) {
			err = net_xmit_errno(err);
			if (err && __packet_get_status(po, ph) ==
				   TP_STATUS_AVAILABLE) {
				/* skb was destructed already */
				skb = NULL;
				goto out_status;
			}
			/*
			 * skb was dropped but not destructed yet;
			 * let's treat it like congestion or err < 0
			 */
			err = 0;
		}
J
Johann Baudy 已提交
2586 2587
		packet_increment_head(&po->tx_ring);
		len_sum += tp_len;
2588 2589 2590 2591 2592 2593 2594 2595
	} while (likely((ph != NULL) ||
		/* Note: packet_read_pending() might be slow if we have
		 * to call it as it's per_cpu variable, but in fast-path
		 * we already short-circuit the loop with the first
		 * condition, and luckily don't have to go that path
		 * anyway.
		 */
		 (need_wait && packet_read_pending(&po->tx_ring))));
J
Johann Baudy 已提交
2596 2597 2598 2599 2600 2601 2602 2603

	err = len_sum;
	goto out_put;

out_status:
	__packet_set_status(po, ph, status);
	kfree_skb(skb);
out_put:
2604
	dev_put(dev);
J
Johann Baudy 已提交
2605 2606 2607 2608 2609
out:
	mutex_unlock(&po->pg_vec_lock);
	return err;
}

2610 2611 2612 2613
static struct sk_buff *packet_alloc_skb(struct sock *sk, size_t prepad,
				        size_t reserve, size_t len,
				        size_t linear, int noblock,
				        int *err)
2614 2615 2616 2617 2618 2619 2620 2621
{
	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,
2622
				   err, 0);
2623 2624 2625 2626 2627 2628 2629 2630 2631 2632 2633
	if (!skb)
		return NULL;

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

	return skb;
}

2634
static int packet_snd(struct socket *sock, struct msghdr *msg, size_t len)
L
Linus Torvalds 已提交
2635 2636
{
	struct sock *sk = sock->sk;
2637
	DECLARE_SOCKADDR(struct sockaddr_ll *, saddr, msg->msg_name);
L
Linus Torvalds 已提交
2638 2639
	struct sk_buff *skb;
	struct net_device *dev;
A
Al Viro 已提交
2640
	__be16 proto;
L
Linus Torvalds 已提交
2641
	unsigned char *addr;
2642
	int err, reserve = 0;
2643
	struct sockcm_cookie sockc;
2644 2645 2646 2647 2648
	struct virtio_net_hdr vnet_hdr = { 0 };
	int offset = 0;
	int vnet_hdr_len;
	struct packet_sock *po = pkt_sk(sk);
	unsigned short gso_type = 0;
2649
	int hlen, tlen;
2650
	int extra_len = 0;
2651
	ssize_t n;
L
Linus Torvalds 已提交
2652 2653

	/*
2654
	 *	Get and verify the address.
L
Linus Torvalds 已提交
2655
	 */
2656

2657
	if (likely(saddr == NULL)) {
2658
		dev	= packet_cached_dev_get(po);
L
Linus Torvalds 已提交
2659 2660 2661 2662 2663 2664
		proto	= po->num;
		addr	= NULL;
	} else {
		err = -EINVAL;
		if (msg->msg_namelen < sizeof(struct sockaddr_ll))
			goto out;
2665 2666
		if (msg->msg_namelen < (saddr->sll_halen + offsetof(struct sockaddr_ll, sll_addr)))
			goto out;
L
Linus Torvalds 已提交
2667 2668
		proto	= saddr->sll_protocol;
		addr	= saddr->sll_addr;
2669
		dev = dev_get_by_index(sock_net(sk), saddr->sll_ifindex);
L
Linus Torvalds 已提交
2670 2671 2672
	}

	err = -ENXIO;
2673
	if (unlikely(dev == NULL))
L
Linus Torvalds 已提交
2674
		goto out_unlock;
2675
	err = -ENETDOWN;
2676
	if (unlikely(!(dev->flags & IFF_UP)))
2677 2678
		goto out_unlock;

2679 2680 2681 2682 2683 2684 2685
	sockc.mark = sk->sk_mark;
	if (msg->msg_controllen) {
		err = sock_cmsg_send(sk, msg, &sockc);
		if (unlikely(err))
			goto out_unlock;
	}

2686 2687
	if (sock->type == SOCK_RAW)
		reserve = dev->hard_header_len;
2688 2689 2690 2691 2692 2693 2694 2695 2696
	if (po->has_vnet_hdr) {
		vnet_hdr_len = sizeof(vnet_hdr);

		err = -EINVAL;
		if (len < vnet_hdr_len)
			goto out_unlock;

		len -= vnet_hdr_len;

2697
		err = -EFAULT;
A
Al Viro 已提交
2698
		n = copy_from_iter(&vnet_hdr, vnet_hdr_len, &msg->msg_iter);
2699
		if (n != vnet_hdr_len)
2700 2701 2702
			goto out_unlock;

		if ((vnet_hdr.flags & VIRTIO_NET_HDR_F_NEEDS_CSUM) &&
2703 2704 2705 2706 2707 2708
		    (__virtio16_to_cpu(vio_le(), vnet_hdr.csum_start) +
		     __virtio16_to_cpu(vio_le(), vnet_hdr.csum_offset) + 2 >
		      __virtio16_to_cpu(vio_le(), vnet_hdr.hdr_len)))
			vnet_hdr.hdr_len = __cpu_to_virtio16(vio_le(),
				 __virtio16_to_cpu(vio_le(), vnet_hdr.csum_start) +
				__virtio16_to_cpu(vio_le(), vnet_hdr.csum_offset) + 2);
2709 2710

		err = -EINVAL;
2711
		if (__virtio16_to_cpu(vio_le(), vnet_hdr.hdr_len) > len)
2712 2713 2714 2715 2716 2717 2718 2719 2720 2721 2722 2723 2724 2725 2726 2727 2728 2729 2730 2731 2732 2733 2734 2735 2736 2737
			goto out_unlock;

		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:
				goto out_unlock;
			}

			if (vnet_hdr.gso_type & VIRTIO_NET_HDR_GSO_ECN)
				gso_type |= SKB_GSO_TCP_ECN;

			if (vnet_hdr.gso_size == 0)
				goto out_unlock;

		}
	}

2738 2739 2740 2741 2742 2743 2744 2745
	if (unlikely(sock_flag(sk, SOCK_NOFCS))) {
		if (!netif_supports_nofcs(dev)) {
			err = -EPROTONOSUPPORT;
			goto out_unlock;
		}
		extra_len = 4; /* We're doing our own CRC */
	}

L
Linus Torvalds 已提交
2746
	err = -EMSGSIZE;
2747
	if (!gso_type && (len > dev->mtu + reserve + VLAN_HLEN + extra_len))
L
Linus Torvalds 已提交
2748 2749
		goto out_unlock;

2750
	err = -ENOBUFS;
2751 2752
	hlen = LL_RESERVED_SPACE(dev);
	tlen = dev->needed_tailroom;
M
Michael S. Tsirkin 已提交
2753
	skb = packet_alloc_skb(sk, hlen + tlen, hlen, len,
2754
			       __virtio16_to_cpu(vio_le(), vnet_hdr.hdr_len),
2755
			       msg->msg_flags & MSG_DONTWAIT, &err);
E
Eric Dumazet 已提交
2756
	if (skb == NULL)
L
Linus Torvalds 已提交
2757 2758
		goto out_unlock;

2759
	skb_set_network_header(skb, reserve);
L
Linus Torvalds 已提交
2760

2761
	err = -EINVAL;
2762 2763
	if (sock->type == SOCK_DGRAM) {
		offset = dev_hard_header(skb, dev, ntohs(proto), addr, NULL, len);
2764
		if (unlikely(offset < 0))
2765 2766 2767 2768 2769
			goto out_free;
	} else {
		if (ll_header_truncated(dev, len))
			goto out_free;
	}
L
Linus Torvalds 已提交
2770 2771

	/* Returns -EFAULT on error */
A
Al Viro 已提交
2772
	err = skb_copy_datagram_from_iter(skb, offset, &msg->msg_iter, len);
L
Linus Torvalds 已提交
2773 2774
	if (err)
		goto out_free;
2775 2776

	sock_tx_timestamp(sk, &skb_shinfo(skb)->tx_flags);
L
Linus Torvalds 已提交
2777

2778 2779 2780 2781
	if (!gso_type && (len > dev->mtu + reserve + extra_len) &&
	    !packet_extra_vlan_len_allowed(dev, skb)) {
		err = -EMSGSIZE;
		goto out_free;
2782 2783
	}

2784 2785
	skb->protocol = proto;
	skb->dev = dev;
L
Linus Torvalds 已提交
2786
	skb->priority = sk->sk_priority;
2787
	skb->mark = sockc.mark;
2788 2789

	packet_pick_tx_queue(dev, skb);
L
Linus Torvalds 已提交
2790

2791 2792
	if (po->has_vnet_hdr) {
		if (vnet_hdr.flags & VIRTIO_NET_HDR_F_NEEDS_CSUM) {
2793 2794
			u16 s = __virtio16_to_cpu(vio_le(), vnet_hdr.csum_start);
			u16 o = __virtio16_to_cpu(vio_le(), vnet_hdr.csum_offset);
M
Michael S. Tsirkin 已提交
2795
			if (!skb_partial_csum_set(skb, s, o)) {
2796 2797 2798 2799 2800
				err = -EINVAL;
				goto out_free;
			}
		}

M
Michael S. Tsirkin 已提交
2801
		skb_shinfo(skb)->gso_size =
2802
			__virtio16_to_cpu(vio_le(), vnet_hdr.gso_size);
2803 2804 2805 2806 2807 2808 2809 2810 2811
		skb_shinfo(skb)->gso_type = gso_type;

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

		len += vnet_hdr_len;
	}

2812 2813
	skb_probe_transport_header(skb, reserve);

2814 2815 2816
	if (unlikely(extra_len == 4))
		skb->no_fcs = 1;

2817
	err = po->xmit(skb);
L
Linus Torvalds 已提交
2818 2819 2820
	if (err > 0 && (err = net_xmit_errno(err)) != 0)
		goto out_unlock;

2821
	dev_put(dev);
L
Linus Torvalds 已提交
2822

E
Eric Dumazet 已提交
2823
	return len;
L
Linus Torvalds 已提交
2824 2825 2826 2827

out_free:
	kfree_skb(skb);
out_unlock:
2828
	if (dev)
L
Linus Torvalds 已提交
2829 2830 2831 2832 2833
		dev_put(dev);
out:
	return err;
}

2834
static int packet_sendmsg(struct socket *sock, struct msghdr *msg, size_t len)
J
Johann Baudy 已提交
2835 2836 2837
{
	struct sock *sk = sock->sk;
	struct packet_sock *po = pkt_sk(sk);
2838

J
Johann Baudy 已提交
2839 2840 2841 2842 2843 2844
	if (po->tx_ring.pg_vec)
		return tpacket_snd(po, msg);
	else
		return packet_snd(sock, msg, len);
}

L
Linus Torvalds 已提交
2845 2846 2847 2848 2849 2850 2851 2852 2853
/*
 *	Close a PACKET socket. This is fairly simple. We immediately go
 *	to 'closed' state and remove our protocol entry in the device list.
 */

static int packet_release(struct socket *sock)
{
	struct sock *sk = sock->sk;
	struct packet_sock *po;
2854
	struct net *net;
2855
	union tpacket_req_u req_u;
L
Linus Torvalds 已提交
2856 2857 2858 2859

	if (!sk)
		return 0;

2860
	net = sock_net(sk);
L
Linus Torvalds 已提交
2861 2862
	po = pkt_sk(sk);

2863
	mutex_lock(&net->packet.sklist_lock);
2864
	sk_del_node_init_rcu(sk);
2865 2866 2867
	mutex_unlock(&net->packet.sklist_lock);

	preempt_disable();
2868
	sock_prot_inuse_add(net, sk->sk_prot, -1);
2869
	preempt_enable();
L
Linus Torvalds 已提交
2870

2871
	spin_lock(&po->bind_lock);
2872
	unregister_prot_hook(sk, false);
2873 2874
	packet_cached_dev_reset(po);

2875 2876 2877 2878
	if (po->prot_hook.dev) {
		dev_put(po->prot_hook.dev);
		po->prot_hook.dev = NULL;
	}
2879
	spin_unlock(&po->bind_lock);
L
Linus Torvalds 已提交
2880 2881 2882

	packet_flush_mclist(sk);

2883 2884
	if (po->rx_ring.pg_vec) {
		memset(&req_u, 0, sizeof(req_u));
2885
		packet_set_ring(sk, &req_u, 1, 0);
2886
	}
J
Johann Baudy 已提交
2887

2888 2889
	if (po->tx_ring.pg_vec) {
		memset(&req_u, 0, sizeof(req_u));
2890
		packet_set_ring(sk, &req_u, 1, 1);
2891
	}
L
Linus Torvalds 已提交
2892

D
David S. Miller 已提交
2893 2894
	fanout_release(sk);

2895
	synchronize_net();
L
Linus Torvalds 已提交
2896 2897 2898 2899 2900 2901 2902 2903 2904
	/*
	 *	Now the socket is dead. No more input will appear.
	 */
	sock_orphan(sk);
	sock->sk = NULL;

	/* Purge queues */

	skb_queue_purge(&sk->sk_receive_queue);
2905
	packet_free_pending(po);
2906
	sk_refcnt_debug_release(sk);
L
Linus Torvalds 已提交
2907 2908 2909 2910 2911 2912 2913 2914 2915

	sock_put(sk);
	return 0;
}

/*
 *	Attach a packet hook.
 */

2916 2917
static int packet_do_bind(struct sock *sk, const char *name, int ifindex,
			  __be16 proto)
L
Linus Torvalds 已提交
2918 2919
{
	struct packet_sock *po = pkt_sk(sk);
2920
	struct net_device *dev_curr;
2921 2922
	__be16 proto_curr;
	bool need_rehook;
2923 2924 2925
	struct net_device *dev = NULL;
	int ret = 0;
	bool unlisted = false;
D
David S. Miller 已提交
2926

2927
	if (po->fanout)
D
David S. Miller 已提交
2928
		return -EINVAL;
L
Linus Torvalds 已提交
2929 2930 2931

	lock_sock(sk);
	spin_lock(&po->bind_lock);
2932 2933 2934 2935 2936 2937 2938 2939 2940 2941 2942 2943 2944 2945 2946 2947 2948 2949
	rcu_read_lock();

	if (name) {
		dev = dev_get_by_name_rcu(sock_net(sk), name);
		if (!dev) {
			ret = -ENODEV;
			goto out_unlock;
		}
	} else if (ifindex) {
		dev = dev_get_by_index_rcu(sock_net(sk), ifindex);
		if (!dev) {
			ret = -ENODEV;
			goto out_unlock;
		}
	}

	if (dev)
		dev_hold(dev);
2950

2951 2952 2953 2954 2955 2956
	proto_curr = po->prot_hook.type;
	dev_curr = po->prot_hook.dev;

	need_rehook = proto_curr != proto || dev_curr != dev;

	if (need_rehook) {
2957 2958 2959 2960 2961 2962 2963 2964 2965
		if (po->running) {
			rcu_read_unlock();
			__unregister_prot_hook(sk, true);
			rcu_read_lock();
			dev_curr = po->prot_hook.dev;
			if (dev)
				unlisted = !dev_get_by_index_rcu(sock_net(sk),
								 dev->ifindex);
		}
L
Linus Torvalds 已提交
2966

2967 2968 2969
		po->num = proto;
		po->prot_hook.type = proto;

2970 2971 2972 2973 2974 2975 2976 2977 2978 2979
		if (unlikely(unlisted)) {
			dev_put(dev);
			po->prot_hook.dev = NULL;
			po->ifindex = -1;
			packet_cached_dev_reset(po);
		} else {
			po->prot_hook.dev = dev;
			po->ifindex = dev ? dev->ifindex : 0;
			packet_cached_dev_assign(po, dev);
		}
2980
	}
2981 2982
	if (dev_curr)
		dev_put(dev_curr);
2983

2984
	if (proto == 0 || !need_rehook)
L
Linus Torvalds 已提交
2985 2986
		goto out_unlock;

2987
	if (!unlisted && (!dev || (dev->flags & IFF_UP))) {
2988
		register_prot_hook(sk);
2989 2990 2991 2992
	} else {
		sk->sk_err = ENETDOWN;
		if (!sock_flag(sk, SOCK_DEAD))
			sk->sk_error_report(sk);
L
Linus Torvalds 已提交
2993 2994 2995
	}

out_unlock:
2996
	rcu_read_unlock();
L
Linus Torvalds 已提交
2997 2998
	spin_unlock(&po->bind_lock);
	release_sock(sk);
2999
	return ret;
L
Linus Torvalds 已提交
3000 3001 3002 3003 3004 3005
}

/*
 *	Bind a packet socket to a device
 */

E
Eric Dumazet 已提交
3006 3007
static int packet_bind_spkt(struct socket *sock, struct sockaddr *uaddr,
			    int addr_len)
L
Linus Torvalds 已提交
3008
{
E
Eric Dumazet 已提交
3009
	struct sock *sk = sock->sk;
L
Linus Torvalds 已提交
3010
	char name[15];
3011

L
Linus Torvalds 已提交
3012 3013 3014
	/*
	 *	Check legality
	 */
3015

3016
	if (addr_len != sizeof(struct sockaddr))
L
Linus Torvalds 已提交
3017
		return -EINVAL;
E
Eric Dumazet 已提交
3018
	strlcpy(name, uaddr->sa_data, sizeof(name));
L
Linus Torvalds 已提交
3019

3020
	return packet_do_bind(sk, name, 0, pkt_sk(sk)->num);
L
Linus Torvalds 已提交
3021 3022 3023 3024
}

static int packet_bind(struct socket *sock, struct sockaddr *uaddr, int addr_len)
{
E
Eric Dumazet 已提交
3025 3026
	struct sockaddr_ll *sll = (struct sockaddr_ll *)uaddr;
	struct sock *sk = sock->sk;
L
Linus Torvalds 已提交
3027 3028 3029 3030

	/*
	 *	Check legality
	 */
3031

L
Linus Torvalds 已提交
3032 3033 3034 3035 3036
	if (addr_len < sizeof(struct sockaddr_ll))
		return -EINVAL;
	if (sll->sll_family != AF_PACKET)
		return -EINVAL;

3037 3038
	return packet_do_bind(sk, NULL, sll->sll_ifindex,
			      sll->sll_protocol ? : pkt_sk(sk)->num);
L
Linus Torvalds 已提交
3039 3040 3041 3042 3043 3044 3045 3046 3047
}

static struct proto packet_proto = {
	.name	  = "PACKET",
	.owner	  = THIS_MODULE,
	.obj_size = sizeof(struct packet_sock),
};

/*
3048
 *	Create a packet of type SOCK_PACKET.
L
Linus Torvalds 已提交
3049 3050
 */

3051 3052
static int packet_create(struct net *net, struct socket *sock, int protocol,
			 int kern)
L
Linus Torvalds 已提交
3053 3054 3055
{
	struct sock *sk;
	struct packet_sock *po;
A
Al Viro 已提交
3056
	__be16 proto = (__force __be16)protocol; /* weird, but documented */
L
Linus Torvalds 已提交
3057 3058
	int err;

3059
	if (!ns_capable(net->user_ns, CAP_NET_RAW))
L
Linus Torvalds 已提交
3060
		return -EPERM;
3061 3062
	if (sock->type != SOCK_DGRAM && sock->type != SOCK_RAW &&
	    sock->type != SOCK_PACKET)
L
Linus Torvalds 已提交
3063 3064 3065 3066 3067
		return -ESOCKTNOSUPPORT;

	sock->state = SS_UNCONNECTED;

	err = -ENOBUFS;
3068
	sk = sk_alloc(net, PF_PACKET, GFP_KERNEL, &packet_proto, kern);
L
Linus Torvalds 已提交
3069 3070 3071 3072 3073 3074
	if (sk == NULL)
		goto out;

	sock->ops = &packet_ops;
	if (sock->type == SOCK_PACKET)
		sock->ops = &packet_ops_spkt;
3075

L
Linus Torvalds 已提交
3076 3077 3078 3079
	sock_init_data(sock, sk);

	po = pkt_sk(sk);
	sk->sk_family = PF_PACKET;
A
Al Viro 已提交
3080
	po->num = proto;
3081
	po->xmit = dev_queue_xmit;
3082

3083 3084 3085 3086
	err = packet_alloc_pending(po);
	if (err)
		goto out2;

3087
	packet_cached_dev_reset(po);
L
Linus Torvalds 已提交
3088 3089

	sk->sk_destruct = packet_sock_destruct;
3090
	sk_refcnt_debug_inc(sk);
L
Linus Torvalds 已提交
3091 3092 3093 3094 3095 3096

	/*
	 *	Attach a protocol block
	 */

	spin_lock_init(&po->bind_lock);
3097
	mutex_init(&po->pg_vec_lock);
3098
	po->rollover = NULL;
L
Linus Torvalds 已提交
3099
	po->prot_hook.func = packet_rcv;
3100

L
Linus Torvalds 已提交
3101 3102
	if (sock->type == SOCK_PACKET)
		po->prot_hook.func = packet_rcv_spkt;
3103

L
Linus Torvalds 已提交
3104 3105
	po->prot_hook.af_packet_priv = sk;

A
Al Viro 已提交
3106 3107
	if (proto) {
		po->prot_hook.type = proto;
3108
		register_prot_hook(sk);
L
Linus Torvalds 已提交
3109 3110
	}

3111
	mutex_lock(&net->packet.sklist_lock);
3112
	sk_add_node_rcu(sk, &net->packet.sklist);
3113 3114 3115
	mutex_unlock(&net->packet.sklist_lock);

	preempt_disable();
3116
	sock_prot_inuse_add(net, &packet_proto, 1);
3117
	preempt_enable();
3118

E
Eric Dumazet 已提交
3119
	return 0;
3120 3121
out2:
	sk_free(sk);
L
Linus Torvalds 已提交
3122 3123 3124 3125 3126 3127 3128 3129 3130
out:
	return err;
}

/*
 *	Pull a packet from our receive queue and hand it to the user.
 *	If necessary we block.
 */

3131 3132
static int packet_recvmsg(struct socket *sock, struct msghdr *msg, size_t len,
			  int flags)
L
Linus Torvalds 已提交
3133 3134 3135 3136
{
	struct sock *sk = sock->sk;
	struct sk_buff *skb;
	int copied, err;
3137
	int vnet_hdr_len = 0;
3138
	unsigned int origlen = 0;
L
Linus Torvalds 已提交
3139 3140

	err = -EINVAL;
3141
	if (flags & ~(MSG_PEEK|MSG_DONTWAIT|MSG_TRUNC|MSG_CMSG_COMPAT|MSG_ERRQUEUE))
L
Linus Torvalds 已提交
3142 3143 3144 3145 3146 3147 3148 3149
		goto out;

#if 0
	/* What error should we return now? EUNATTACH? */
	if (pkt_sk(sk)->ifindex < 0)
		return -ENODEV;
#endif

3150
	if (flags & MSG_ERRQUEUE) {
3151 3152
		err = sock_recv_errqueue(sk, msg, len,
					 SOL_PACKET, PACKET_TX_TIMESTAMP);
3153 3154 3155
		goto out;
	}

L
Linus Torvalds 已提交
3156 3157 3158 3159 3160 3161 3162 3163 3164
	/*
	 *	Call the generic datagram receiver. This handles all sorts
	 *	of horrible races and re-entrancy so we can forget about it
	 *	in the protocol layers.
	 *
	 *	Now it will return ENETDOWN, if device have just gone down,
	 *	but then it will block.
	 */

E
Eric Dumazet 已提交
3165
	skb = skb_recv_datagram(sk, flags, flags & MSG_DONTWAIT, &err);
L
Linus Torvalds 已提交
3166 3167

	/*
3168
	 *	An error occurred so return it. Because skb_recv_datagram()
L
Linus Torvalds 已提交
3169 3170 3171 3172
	 *	handles the blocking we don't see and worry about blocking
	 *	retries.
	 */

3173
	if (skb == NULL)
L
Linus Torvalds 已提交
3174 3175
		goto out;

3176 3177 3178
	if (pkt_sk(sk)->pressure)
		packet_rcv_has_room(pkt_sk(sk), NULL);

3179 3180 3181 3182 3183
	if (pkt_sk(sk)->has_vnet_hdr) {
		struct virtio_net_hdr vnet_hdr = { 0 };

		err = -EINVAL;
		vnet_hdr_len = sizeof(vnet_hdr);
3184
		if (len < vnet_hdr_len)
3185 3186
			goto out_free;

3187 3188
		len -= vnet_hdr_len;

3189 3190 3191 3192
		if (skb_is_gso(skb)) {
			struct skb_shared_info *sinfo = skb_shinfo(skb);

			/* This is a hint as to how much should be linear. */
M
Michael S. Tsirkin 已提交
3193
			vnet_hdr.hdr_len =
3194
				__cpu_to_virtio16(vio_le(), skb_headlen(skb));
M
Michael S. Tsirkin 已提交
3195
			vnet_hdr.gso_size =
3196
				__cpu_to_virtio16(vio_le(), sinfo->gso_size);
3197 3198 3199 3200 3201 3202 3203 3204 3205 3206 3207 3208 3209 3210 3211 3212 3213
			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;
			else if (sinfo->gso_type & SKB_GSO_UDP)
				vnet_hdr.gso_type = VIRTIO_NET_HDR_GSO_UDP;
			else if (sinfo->gso_type & SKB_GSO_FCOE)
				goto out_free;
			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;
3214
			vnet_hdr.csum_start = __cpu_to_virtio16(vio_le(),
M
Michael S. Tsirkin 已提交
3215
					  skb_checksum_start_offset(skb));
3216
			vnet_hdr.csum_offset = __cpu_to_virtio16(vio_le(),
M
Michael S. Tsirkin 已提交
3217
							 skb->csum_offset);
3218 3219
		} else if (skb->ip_summed == CHECKSUM_UNNECESSARY) {
			vnet_hdr.flags = VIRTIO_NET_HDR_F_DATA_VALID;
3220 3221
		} /* else everything is zero */

A
Al Viro 已提交
3222
		err = memcpy_to_msg(msg, (void *)&vnet_hdr, vnet_hdr_len);
3223 3224 3225 3226
		if (err < 0)
			goto out_free;
	}

3227 3228 3229
	/* You lose any data beyond the buffer you gave. If it worries
	 * a user program they can ask the device for its MTU
	 * anyway.
L
Linus Torvalds 已提交
3230 3231
	 */
	copied = skb->len;
E
Eric Dumazet 已提交
3232 3233 3234
	if (copied > len) {
		copied = len;
		msg->msg_flags |= MSG_TRUNC;
L
Linus Torvalds 已提交
3235 3236
	}

3237
	err = skb_copy_datagram_msg(skb, 0, msg, copied);
L
Linus Torvalds 已提交
3238 3239 3240
	if (err)
		goto out_free;

3241 3242 3243 3244 3245 3246 3247 3248 3249
	if (sock->type != SOCK_PACKET) {
		struct sockaddr_ll *sll = &PACKET_SKB_CB(skb)->sa.ll;

		/* Original length was stored in sockaddr_ll fields */
		origlen = PACKET_SKB_CB(skb)->sa.origlen;
		sll->sll_family = AF_PACKET;
		sll->sll_protocol = skb->protocol;
	}

3250
	sock_recv_ts_and_drops(msg, sk, skb);
L
Linus Torvalds 已提交
3251

3252 3253 3254 3255 3256
	if (msg->msg_name) {
		/* If the address length field is there to be filled
		 * in, we fill it in now.
		 */
		if (sock->type == SOCK_PACKET) {
3257
			__sockaddr_check_size(sizeof(struct sockaddr_pkt));
3258 3259 3260
			msg->msg_namelen = sizeof(struct sockaddr_pkt);
		} else {
			struct sockaddr_ll *sll = &PACKET_SKB_CB(skb)->sa.ll;
3261

3262 3263 3264
			msg->msg_namelen = sll->sll_halen +
				offsetof(struct sockaddr_ll, sll_addr);
		}
3265 3266
		memcpy(msg->msg_name, &PACKET_SKB_CB(skb)->sa,
		       msg->msg_namelen);
3267
	}
L
Linus Torvalds 已提交
3268

3269
	if (pkt_sk(sk)->auxdata) {
3270 3271 3272 3273 3274
		struct tpacket_auxdata aux;

		aux.tp_status = TP_STATUS_USER;
		if (skb->ip_summed == CHECKSUM_PARTIAL)
			aux.tp_status |= TP_STATUS_CSUMNOTREADY;
3275 3276 3277 3278 3279
		else if (skb->pkt_type != PACKET_OUTGOING &&
			 (skb->ip_summed == CHECKSUM_COMPLETE ||
			  skb_csum_unnecessary(skb)))
			aux.tp_status |= TP_STATUS_CSUM_VALID;

3280
		aux.tp_len = origlen;
3281 3282
		aux.tp_snaplen = skb->len;
		aux.tp_mac = 0;
3283
		aux.tp_net = skb_network_offset(skb);
3284 3285
		if (skb_vlan_tag_present(skb)) {
			aux.tp_vlan_tci = skb_vlan_tag_get(skb);
3286 3287
			aux.tp_vlan_tpid = ntohs(skb->vlan_proto);
			aux.tp_status |= TP_STATUS_VLAN_VALID | TP_STATUS_VLAN_TPID_VALID;
3288 3289
		} else {
			aux.tp_vlan_tci = 0;
3290
			aux.tp_vlan_tpid = 0;
3291
		}
3292
		put_cmsg(msg, SOL_PACKET, PACKET_AUXDATA, sizeof(aux), &aux);
3293 3294
	}

L
Linus Torvalds 已提交
3295 3296 3297 3298
	/*
	 *	Free or return the buffer as appropriate. Again this
	 *	hides all the races and re-entrancy issues from us.
	 */
3299
	err = vnet_hdr_len + ((flags&MSG_TRUNC) ? skb->len : copied);
L
Linus Torvalds 已提交
3300 3301 3302 3303 3304 3305 3306 3307 3308 3309 3310 3311 3312 3313 3314 3315 3316

out_free:
	skb_free_datagram(sk, skb);
out:
	return err;
}

static int packet_getname_spkt(struct socket *sock, struct sockaddr *uaddr,
			       int *uaddr_len, int peer)
{
	struct net_device *dev;
	struct sock *sk	= sock->sk;

	if (peer)
		return -EOPNOTSUPP;

	uaddr->sa_family = AF_PACKET;
3317
	memset(uaddr->sa_data, 0, sizeof(uaddr->sa_data));
E
Eric Dumazet 已提交
3318 3319 3320
	rcu_read_lock();
	dev = dev_get_by_index_rcu(sock_net(sk), pkt_sk(sk)->ifindex);
	if (dev)
3321
		strlcpy(uaddr->sa_data, dev->name, sizeof(uaddr->sa_data));
E
Eric Dumazet 已提交
3322
	rcu_read_unlock();
L
Linus Torvalds 已提交
3323 3324 3325 3326 3327 3328 3329 3330 3331 3332 3333
	*uaddr_len = sizeof(*uaddr);

	return 0;
}

static int packet_getname(struct socket *sock, struct sockaddr *uaddr,
			  int *uaddr_len, int peer)
{
	struct net_device *dev;
	struct sock *sk = sock->sk;
	struct packet_sock *po = pkt_sk(sk);
3334
	DECLARE_SOCKADDR(struct sockaddr_ll *, sll, uaddr);
L
Linus Torvalds 已提交
3335 3336 3337 3338 3339 3340 3341

	if (peer)
		return -EOPNOTSUPP;

	sll->sll_family = AF_PACKET;
	sll->sll_ifindex = po->ifindex;
	sll->sll_protocol = po->num;
3342
	sll->sll_pkttype = 0;
E
Eric Dumazet 已提交
3343 3344
	rcu_read_lock();
	dev = dev_get_by_index_rcu(sock_net(sk), po->ifindex);
L
Linus Torvalds 已提交
3345 3346 3347 3348 3349 3350 3351 3352
	if (dev) {
		sll->sll_hatype = dev->type;
		sll->sll_halen = dev->addr_len;
		memcpy(sll->sll_addr, dev->dev_addr, dev->addr_len);
	} else {
		sll->sll_hatype = 0;	/* Bad: we have no ARPHRD_UNSPEC */
		sll->sll_halen = 0;
	}
E
Eric Dumazet 已提交
3353
	rcu_read_unlock();
3354
	*uaddr_len = offsetof(struct sockaddr_ll, sll_addr) + sll->sll_halen;
L
Linus Torvalds 已提交
3355 3356 3357 3358

	return 0;
}

3359 3360
static int packet_dev_mc(struct net_device *dev, struct packet_mclist *i,
			 int what)
L
Linus Torvalds 已提交
3361 3362 3363
{
	switch (i->type) {
	case PACKET_MR_MULTICAST:
3364 3365
		if (i->alen != dev->addr_len)
			return -EINVAL;
L
Linus Torvalds 已提交
3366
		if (what > 0)
3367
			return dev_mc_add(dev, i->addr);
L
Linus Torvalds 已提交
3368
		else
3369
			return dev_mc_del(dev, i->addr);
L
Linus Torvalds 已提交
3370 3371
		break;
	case PACKET_MR_PROMISC:
3372
		return dev_set_promiscuity(dev, what);
L
Linus Torvalds 已提交
3373
	case PACKET_MR_ALLMULTI:
3374
		return dev_set_allmulti(dev, what);
3375
	case PACKET_MR_UNICAST:
3376 3377
		if (i->alen != dev->addr_len)
			return -EINVAL;
3378
		if (what > 0)
3379
			return dev_uc_add(dev, i->addr);
3380
		else
3381
			return dev_uc_del(dev, i->addr);
3382
		break;
E
Eric Dumazet 已提交
3383 3384
	default:
		break;
L
Linus Torvalds 已提交
3385
	}
3386
	return 0;
L
Linus Torvalds 已提交
3387 3388
}

3389 3390
static void packet_dev_mclist_delete(struct net_device *dev,
				     struct packet_mclist **mlp)
L
Linus Torvalds 已提交
3391
{
3392 3393 3394 3395 3396 3397 3398 3399 3400
	struct packet_mclist *ml;

	while ((ml = *mlp) != NULL) {
		if (ml->ifindex == dev->ifindex) {
			packet_dev_mc(dev, ml, -1);
			*mlp = ml->next;
			kfree(ml);
		} else
			mlp = &ml->next;
L
Linus Torvalds 已提交
3401 3402 3403
	}
}

3404
static int packet_mc_add(struct sock *sk, struct packet_mreq_max *mreq)
L
Linus Torvalds 已提交
3405 3406 3407 3408 3409 3410 3411 3412 3413
{
	struct packet_sock *po = pkt_sk(sk);
	struct packet_mclist *ml, *i;
	struct net_device *dev;
	int err;

	rtnl_lock();

	err = -ENODEV;
3414
	dev = __dev_get_by_index(sock_net(sk), mreq->mr_ifindex);
L
Linus Torvalds 已提交
3415 3416 3417 3418
	if (!dev)
		goto done;

	err = -EINVAL;
3419
	if (mreq->mr_alen > dev->addr_len)
L
Linus Torvalds 已提交
3420 3421 3422
		goto done;

	err = -ENOBUFS;
3423
	i = kmalloc(sizeof(*i), GFP_KERNEL);
L
Linus Torvalds 已提交
3424 3425 3426 3427 3428 3429 3430 3431 3432 3433 3434 3435 3436 3437 3438 3439 3440 3441 3442 3443 3444 3445 3446
	if (i == NULL)
		goto done;

	err = 0;
	for (ml = po->mclist; ml; ml = ml->next) {
		if (ml->ifindex == mreq->mr_ifindex &&
		    ml->type == mreq->mr_type &&
		    ml->alen == mreq->mr_alen &&
		    memcmp(ml->addr, mreq->mr_address, ml->alen) == 0) {
			ml->count++;
			/* Free the new element ... */
			kfree(i);
			goto done;
		}
	}

	i->type = mreq->mr_type;
	i->ifindex = mreq->mr_ifindex;
	i->alen = mreq->mr_alen;
	memcpy(i->addr, mreq->mr_address, i->alen);
	i->count = 1;
	i->next = po->mclist;
	po->mclist = i;
3447 3448 3449 3450 3451
	err = packet_dev_mc(dev, i, 1);
	if (err) {
		po->mclist = i->next;
		kfree(i);
	}
L
Linus Torvalds 已提交
3452 3453 3454 3455 3456 3457

done:
	rtnl_unlock();
	return err;
}

3458
static int packet_mc_drop(struct sock *sk, struct packet_mreq_max *mreq)
L
Linus Torvalds 已提交
3459 3460 3461 3462 3463 3464 3465 3466 3467 3468 3469 3470 3471
{
	struct packet_mclist *ml, **mlp;

	rtnl_lock();

	for (mlp = &pkt_sk(sk)->mclist; (ml = *mlp) != NULL; mlp = &ml->next) {
		if (ml->ifindex == mreq->mr_ifindex &&
		    ml->type == mreq->mr_type &&
		    ml->alen == mreq->mr_alen &&
		    memcmp(ml->addr, mreq->mr_address, ml->alen) == 0) {
			if (--ml->count == 0) {
				struct net_device *dev;
				*mlp = ml->next;
3472 3473
				dev = __dev_get_by_index(sock_net(sk), ml->ifindex);
				if (dev)
L
Linus Torvalds 已提交
3474 3475 3476
					packet_dev_mc(dev, ml, -1);
				kfree(ml);
			}
3477
			break;
L
Linus Torvalds 已提交
3478 3479 3480
		}
	}
	rtnl_unlock();
3481
	return 0;
L
Linus Torvalds 已提交
3482 3483 3484 3485 3486 3487 3488 3489 3490 3491 3492 3493 3494 3495 3496
}

static void packet_flush_mclist(struct sock *sk)
{
	struct packet_sock *po = pkt_sk(sk);
	struct packet_mclist *ml;

	if (!po->mclist)
		return;

	rtnl_lock();
	while ((ml = po->mclist) != NULL) {
		struct net_device *dev;

		po->mclist = ml->next;
3497 3498
		dev = __dev_get_by_index(sock_net(sk), ml->ifindex);
		if (dev != NULL)
L
Linus Torvalds 已提交
3499 3500 3501 3502 3503 3504 3505
			packet_dev_mc(dev, ml, -1);
		kfree(ml);
	}
	rtnl_unlock();
}

static int
3506
packet_setsockopt(struct socket *sock, int level, int optname, char __user *optval, unsigned int optlen)
L
Linus Torvalds 已提交
3507 3508
{
	struct sock *sk = sock->sk;
3509
	struct packet_sock *po = pkt_sk(sk);
L
Linus Torvalds 已提交
3510 3511 3512 3513 3514
	int ret;

	if (level != SOL_PACKET)
		return -ENOPROTOOPT;

J
Johann Baudy 已提交
3515
	switch (optname) {
3516
	case PACKET_ADD_MEMBERSHIP:
L
Linus Torvalds 已提交
3517 3518
	case PACKET_DROP_MEMBERSHIP:
	{
3519 3520 3521 3522
		struct packet_mreq_max mreq;
		int len = optlen;
		memset(&mreq, 0, sizeof(mreq));
		if (len < sizeof(struct packet_mreq))
L
Linus Torvalds 已提交
3523
			return -EINVAL;
3524 3525
		if (len > sizeof(mreq))
			len = sizeof(mreq);
E
Eric Dumazet 已提交
3526
		if (copy_from_user(&mreq, optval, len))
L
Linus Torvalds 已提交
3527
			return -EFAULT;
3528 3529
		if (len < (mreq.mr_alen + offsetof(struct packet_mreq, mr_address)))
			return -EINVAL;
L
Linus Torvalds 已提交
3530 3531 3532 3533 3534 3535
		if (optname == PACKET_ADD_MEMBERSHIP)
			ret = packet_mc_add(sk, &mreq);
		else
			ret = packet_mc_drop(sk, &mreq);
		return ret;
	}
3536

L
Linus Torvalds 已提交
3537
	case PACKET_RX_RING:
J
Johann Baudy 已提交
3538
	case PACKET_TX_RING:
L
Linus Torvalds 已提交
3539
	{
3540 3541
		union tpacket_req_u req_u;
		int len;
L
Linus Torvalds 已提交
3542

3543 3544 3545 3546 3547 3548 3549 3550 3551 3552 3553
		switch (po->tp_version) {
		case TPACKET_V1:
		case TPACKET_V2:
			len = sizeof(req_u.req);
			break;
		case TPACKET_V3:
		default:
			len = sizeof(req_u.req3);
			break;
		}
		if (optlen < len)
L
Linus Torvalds 已提交
3554
			return -EINVAL;
3555 3556
		if (pkt_sk(sk)->has_vnet_hdr)
			return -EINVAL;
3557
		if (copy_from_user(&req_u.req, optval, len))
L
Linus Torvalds 已提交
3558
			return -EFAULT;
3559 3560
		return packet_set_ring(sk, &req_u, 0,
			optname == PACKET_TX_RING);
L
Linus Torvalds 已提交
3561 3562 3563 3564 3565
	}
	case PACKET_COPY_THRESH:
	{
		int val;

E
Eric Dumazet 已提交
3566
		if (optlen != sizeof(val))
L
Linus Torvalds 已提交
3567
			return -EINVAL;
E
Eric Dumazet 已提交
3568
		if (copy_from_user(&val, optval, sizeof(val)))
L
Linus Torvalds 已提交
3569 3570 3571 3572 3573
			return -EFAULT;

		pkt_sk(sk)->copy_thresh = val;
		return 0;
	}
3574 3575 3576 3577 3578 3579
	case PACKET_VERSION:
	{
		int val;

		if (optlen != sizeof(val))
			return -EINVAL;
J
Johann Baudy 已提交
3580
		if (po->rx_ring.pg_vec || po->tx_ring.pg_vec)
3581 3582 3583 3584 3585 3586
			return -EBUSY;
		if (copy_from_user(&val, optval, sizeof(val)))
			return -EFAULT;
		switch (val) {
		case TPACKET_V1:
		case TPACKET_V2:
3587
		case TPACKET_V3:
3588 3589 3590 3591 3592 3593
			po->tp_version = val;
			return 0;
		default:
			return -EINVAL;
		}
	}
3594 3595 3596 3597 3598 3599
	case PACKET_RESERVE:
	{
		unsigned int val;

		if (optlen != sizeof(val))
			return -EINVAL;
J
Johann Baudy 已提交
3600
		if (po->rx_ring.pg_vec || po->tx_ring.pg_vec)
3601 3602 3603 3604 3605 3606
			return -EBUSY;
		if (copy_from_user(&val, optval, sizeof(val)))
			return -EFAULT;
		po->tp_reserve = val;
		return 0;
	}
J
Johann Baudy 已提交
3607 3608 3609 3610 3611 3612 3613 3614 3615 3616 3617 3618 3619
	case PACKET_LOSS:
	{
		unsigned int val;

		if (optlen != sizeof(val))
			return -EINVAL;
		if (po->rx_ring.pg_vec || po->tx_ring.pg_vec)
			return -EBUSY;
		if (copy_from_user(&val, optval, sizeof(val)))
			return -EFAULT;
		po->tp_loss = !!val;
		return 0;
	}
3620 3621 3622 3623 3624 3625 3626 3627 3628 3629 3630 3631
	case PACKET_AUXDATA:
	{
		int val;

		if (optlen < sizeof(val))
			return -EINVAL;
		if (copy_from_user(&val, optval, sizeof(val)))
			return -EFAULT;

		po->auxdata = !!val;
		return 0;
	}
3632 3633 3634 3635 3636 3637 3638 3639 3640 3641 3642 3643
	case PACKET_ORIGDEV:
	{
		int val;

		if (optlen < sizeof(val))
			return -EINVAL;
		if (copy_from_user(&val, optval, sizeof(val)))
			return -EFAULT;

		po->origdev = !!val;
		return 0;
	}
3644 3645 3646 3647 3648 3649 3650 3651 3652 3653 3654 3655 3656 3657 3658 3659
	case PACKET_VNET_HDR:
	{
		int val;

		if (sock->type != SOCK_RAW)
			return -EINVAL;
		if (po->rx_ring.pg_vec || po->tx_ring.pg_vec)
			return -EBUSY;
		if (optlen < sizeof(val))
			return -EINVAL;
		if (copy_from_user(&val, optval, sizeof(val)))
			return -EFAULT;

		po->has_vnet_hdr = !!val;
		return 0;
	}
3660 3661 3662 3663 3664 3665 3666 3667 3668 3669 3670 3671
	case PACKET_TIMESTAMP:
	{
		int val;

		if (optlen != sizeof(val))
			return -EINVAL;
		if (copy_from_user(&val, optval, sizeof(val)))
			return -EFAULT;

		po->tp_tstamp = val;
		return 0;
	}
D
David S. Miller 已提交
3672 3673 3674 3675 3676 3677 3678 3679 3680 3681 3682
	case PACKET_FANOUT:
	{
		int val;

		if (optlen != sizeof(val))
			return -EINVAL;
		if (copy_from_user(&val, optval, sizeof(val)))
			return -EFAULT;

		return fanout_add(sk, val & 0xffff, val >> 16);
	}
3683 3684 3685 3686 3687 3688 3689
	case PACKET_FANOUT_DATA:
	{
		if (!po->fanout)
			return -EINVAL;

		return fanout_set_data(po, optval, optlen);
	}
3690 3691 3692 3693 3694 3695 3696 3697 3698 3699 3700 3701 3702
	case PACKET_TX_HAS_OFF:
	{
		unsigned int val;

		if (optlen != sizeof(val))
			return -EINVAL;
		if (po->rx_ring.pg_vec || po->tx_ring.pg_vec)
			return -EBUSY;
		if (copy_from_user(&val, optval, sizeof(val)))
			return -EFAULT;
		po->tp_tx_has_off = !!val;
		return 0;
	}
3703 3704 3705 3706 3707 3708 3709 3710 3711 3712 3713 3714
	case PACKET_QDISC_BYPASS:
	{
		int val;

		if (optlen != sizeof(val))
			return -EINVAL;
		if (copy_from_user(&val, optval, sizeof(val)))
			return -EFAULT;

		po->xmit = val ? packet_direct_xmit : dev_queue_xmit;
		return 0;
	}
L
Linus Torvalds 已提交
3715 3716 3717 3718 3719 3720 3721 3722 3723
	default:
		return -ENOPROTOOPT;
	}
}

static int packet_getsockopt(struct socket *sock, int level, int optname,
			     char __user *optval, int __user *optlen)
{
	int len;
3724
	int val, lv = sizeof(val);
L
Linus Torvalds 已提交
3725 3726
	struct sock *sk = sock->sk;
	struct packet_sock *po = pkt_sk(sk);
3727
	void *data = &val;
3728
	union tpacket_stats_u st;
W
Willem de Bruijn 已提交
3729
	struct tpacket_rollover_stats rstats;
L
Linus Torvalds 已提交
3730 3731 3732 3733

	if (level != SOL_PACKET)
		return -ENOPROTOOPT;

3734 3735
	if (get_user(len, optlen))
		return -EFAULT;
L
Linus Torvalds 已提交
3736 3737 3738

	if (len < 0)
		return -EINVAL;
3739

J
Johann Baudy 已提交
3740
	switch (optname) {
L
Linus Torvalds 已提交
3741 3742
	case PACKET_STATISTICS:
		spin_lock_bh(&sk->sk_receive_queue.lock);
3743 3744 3745 3746
		memcpy(&st, &po->stats, sizeof(st));
		memset(&po->stats, 0, sizeof(po->stats));
		spin_unlock_bh(&sk->sk_receive_queue.lock);

3747
		if (po->tp_version == TPACKET_V3) {
3748
			lv = sizeof(struct tpacket_stats_v3);
3749
			st.stats3.tp_packets += st.stats3.tp_drops;
3750
			data = &st.stats3;
3751
		} else {
3752
			lv = sizeof(struct tpacket_stats);
3753
			st.stats1.tp_packets += st.stats1.tp_drops;
3754
			data = &st.stats1;
3755
		}
3756

3757 3758 3759
		break;
	case PACKET_AUXDATA:
		val = po->auxdata;
3760 3761 3762
		break;
	case PACKET_ORIGDEV:
		val = po->origdev;
3763 3764 3765
		break;
	case PACKET_VNET_HDR:
		val = po->has_vnet_hdr;
L
Linus Torvalds 已提交
3766
		break;
3767 3768 3769 3770 3771 3772 3773 3774 3775 3776 3777 3778 3779 3780 3781
	case PACKET_VERSION:
		val = po->tp_version;
		break;
	case PACKET_HDRLEN:
		if (len > sizeof(int))
			len = sizeof(int);
		if (copy_from_user(&val, optval, len))
			return -EFAULT;
		switch (val) {
		case TPACKET_V1:
			val = sizeof(struct tpacket_hdr);
			break;
		case TPACKET_V2:
			val = sizeof(struct tpacket2_hdr);
			break;
3782 3783 3784
		case TPACKET_V3:
			val = sizeof(struct tpacket3_hdr);
			break;
3785 3786 3787 3788
		default:
			return -EINVAL;
		}
		break;
3789 3790 3791
	case PACKET_RESERVE:
		val = po->tp_reserve;
		break;
J
Johann Baudy 已提交
3792 3793 3794
	case PACKET_LOSS:
		val = po->tp_loss;
		break;
3795 3796 3797
	case PACKET_TIMESTAMP:
		val = po->tp_tstamp;
		break;
D
David S. Miller 已提交
3798 3799 3800
	case PACKET_FANOUT:
		val = (po->fanout ?
		       ((u32)po->fanout->id |
3801 3802
			((u32)po->fanout->type << 16) |
			((u32)po->fanout->flags << 24)) :
D
David S. Miller 已提交
3803 3804
		       0);
		break;
W
Willem de Bruijn 已提交
3805 3806 3807 3808 3809 3810 3811 3812 3813
	case PACKET_ROLLOVER_STATS:
		if (!po->rollover)
			return -EINVAL;
		rstats.tp_all = atomic_long_read(&po->rollover->num);
		rstats.tp_huge = atomic_long_read(&po->rollover->num_huge);
		rstats.tp_failed = atomic_long_read(&po->rollover->num_failed);
		data = &rstats;
		lv = sizeof(rstats);
		break;
3814 3815 3816
	case PACKET_TX_HAS_OFF:
		val = po->tp_tx_has_off;
		break;
3817 3818 3819
	case PACKET_QDISC_BYPASS:
		val = packet_use_direct_xmit(po);
		break;
L
Linus Torvalds 已提交
3820 3821 3822 3823
	default:
		return -ENOPROTOOPT;
	}

3824 3825
	if (len > lv)
		len = lv;
3826 3827
	if (put_user(len, optlen))
		return -EFAULT;
3828 3829
	if (copy_to_user(optval, data, len))
		return -EFAULT;
3830
	return 0;
L
Linus Torvalds 已提交
3831 3832 3833
}


3834 3835
static int packet_notifier(struct notifier_block *this,
			   unsigned long msg, void *ptr)
L
Linus Torvalds 已提交
3836 3837
{
	struct sock *sk;
3838
	struct net_device *dev = netdev_notifier_info_to_dev(ptr);
3839
	struct net *net = dev_net(dev);
L
Linus Torvalds 已提交
3840

3841
	rcu_read_lock();
3842
	sk_for_each_rcu(sk, &net->packet.sklist) {
L
Linus Torvalds 已提交
3843 3844 3845 3846 3847
		struct packet_sock *po = pkt_sk(sk);

		switch (msg) {
		case NETDEV_UNREGISTER:
			if (po->mclist)
3848
				packet_dev_mclist_delete(dev, &po->mclist);
3849 3850
			/* fallthrough */

L
Linus Torvalds 已提交
3851 3852 3853 3854
		case NETDEV_DOWN:
			if (dev->ifindex == po->ifindex) {
				spin_lock(&po->bind_lock);
				if (po->running) {
3855
					__unregister_prot_hook(sk, false);
L
Linus Torvalds 已提交
3856 3857 3858 3859 3860
					sk->sk_err = ENETDOWN;
					if (!sock_flag(sk, SOCK_DEAD))
						sk->sk_error_report(sk);
				}
				if (msg == NETDEV_UNREGISTER) {
3861
					packet_cached_dev_reset(po);
L
Linus Torvalds 已提交
3862
					po->ifindex = -1;
3863 3864
					if (po->prot_hook.dev)
						dev_put(po->prot_hook.dev);
L
Linus Torvalds 已提交
3865 3866 3867 3868 3869 3870
					po->prot_hook.dev = NULL;
				}
				spin_unlock(&po->bind_lock);
			}
			break;
		case NETDEV_UP:
3871 3872
			if (dev->ifindex == po->ifindex) {
				spin_lock(&po->bind_lock);
3873 3874
				if (po->num)
					register_prot_hook(sk);
3875
				spin_unlock(&po->bind_lock);
L
Linus Torvalds 已提交
3876 3877 3878 3879
			}
			break;
		}
	}
3880
	rcu_read_unlock();
L
Linus Torvalds 已提交
3881 3882 3883 3884 3885 3886 3887 3888 3889
	return NOTIFY_DONE;
}


static int packet_ioctl(struct socket *sock, unsigned int cmd,
			unsigned long arg)
{
	struct sock *sk = sock->sk;

J
Johann Baudy 已提交
3890
	switch (cmd) {
E
Eric Dumazet 已提交
3891 3892 3893
	case SIOCOUTQ:
	{
		int amount = sk_wmem_alloc_get(sk);
3894

E
Eric Dumazet 已提交
3895 3896 3897 3898 3899 3900 3901 3902 3903 3904 3905 3906 3907 3908 3909 3910 3911 3912
		return put_user(amount, (int __user *)arg);
	}
	case SIOCINQ:
	{
		struct sk_buff *skb;
		int amount = 0;

		spin_lock_bh(&sk->sk_receive_queue.lock);
		skb = skb_peek(&sk->sk_receive_queue);
		if (skb)
			amount = skb->len;
		spin_unlock_bh(&sk->sk_receive_queue.lock);
		return put_user(amount, (int __user *)arg);
	}
	case SIOCGSTAMP:
		return sock_get_timestamp(sk, (struct timeval __user *)arg);
	case SIOCGSTAMPNS:
		return sock_get_timestampns(sk, (struct timespec __user *)arg);
3913

L
Linus Torvalds 已提交
3914
#ifdef CONFIG_INET
E
Eric Dumazet 已提交
3915 3916 3917 3918 3919 3920 3921 3922 3923 3924 3925 3926 3927 3928 3929
	case SIOCADDRT:
	case SIOCDELRT:
	case SIOCDARP:
	case SIOCGARP:
	case SIOCSARP:
	case SIOCGIFADDR:
	case SIOCSIFADDR:
	case SIOCGIFBRDADDR:
	case SIOCSIFBRDADDR:
	case SIOCGIFNETMASK:
	case SIOCSIFNETMASK:
	case SIOCGIFDSTADDR:
	case SIOCSIFDSTADDR:
	case SIOCSIFFLAGS:
		return inet_dgram_ops.ioctl(sock, cmd, arg);
L
Linus Torvalds 已提交
3930 3931
#endif

E
Eric Dumazet 已提交
3932 3933
	default:
		return -ENOIOCTLCMD;
L
Linus Torvalds 已提交
3934 3935 3936 3937
	}
	return 0;
}

E
Eric Dumazet 已提交
3938
static unsigned int packet_poll(struct file *file, struct socket *sock,
L
Linus Torvalds 已提交
3939 3940 3941 3942 3943 3944 3945
				poll_table *wait)
{
	struct sock *sk = sock->sk;
	struct packet_sock *po = pkt_sk(sk);
	unsigned int mask = datagram_poll(file, sock, wait);

	spin_lock_bh(&sk->sk_receive_queue.lock);
J
Johann Baudy 已提交
3946
	if (po->rx_ring.pg_vec) {
3947 3948
		if (!packet_previous_rx_frame(po, &po->rx_ring,
			TP_STATUS_KERNEL))
L
Linus Torvalds 已提交
3949 3950
			mask |= POLLIN | POLLRDNORM;
	}
3951
	if (po->pressure && __packet_rcv_has_room(po, NULL) == ROOM_NORMAL)
3952
		po->pressure = 0;
L
Linus Torvalds 已提交
3953
	spin_unlock_bh(&sk->sk_receive_queue.lock);
J
Johann Baudy 已提交
3954 3955 3956 3957 3958 3959
	spin_lock_bh(&sk->sk_write_queue.lock);
	if (po->tx_ring.pg_vec) {
		if (packet_current_frame(po, &po->tx_ring, TP_STATUS_AVAILABLE))
			mask |= POLLOUT | POLLWRNORM;
	}
	spin_unlock_bh(&sk->sk_write_queue.lock);
L
Linus Torvalds 已提交
3960 3961 3962 3963 3964 3965 3966 3967 3968 3969 3970
	return mask;
}


/* Dirty? Well, I still did not learn better way to account
 * for user mmaps.
 */

static void packet_mm_open(struct vm_area_struct *vma)
{
	struct file *file = vma->vm_file;
E
Eric Dumazet 已提交
3971
	struct socket *sock = file->private_data;
L
Linus Torvalds 已提交
3972
	struct sock *sk = sock->sk;
3973

L
Linus Torvalds 已提交
3974 3975 3976 3977 3978 3979 3980
	if (sk)
		atomic_inc(&pkt_sk(sk)->mapped);
}

static void packet_mm_close(struct vm_area_struct *vma)
{
	struct file *file = vma->vm_file;
E
Eric Dumazet 已提交
3981
	struct socket *sock = file->private_data;
L
Linus Torvalds 已提交
3982
	struct sock *sk = sock->sk;
3983

L
Linus Torvalds 已提交
3984 3985 3986 3987
	if (sk)
		atomic_dec(&pkt_sk(sk)->mapped);
}

3988
static const struct vm_operations_struct packet_mmap_ops = {
E
Eric Dumazet 已提交
3989 3990
	.open	=	packet_mm_open,
	.close	=	packet_mm_close,
L
Linus Torvalds 已提交
3991 3992
};

3993 3994
static void free_pg_vec(struct pgv *pg_vec, unsigned int order,
			unsigned int len)
L
Linus Torvalds 已提交
3995 3996 3997
{
	int i;

3998
	for (i = 0; i < len; i++) {
3999
		if (likely(pg_vec[i].buffer)) {
C
Changli Gao 已提交
4000
			if (is_vmalloc_addr(pg_vec[i].buffer))
4001 4002 4003 4004 4005 4006
				vfree(pg_vec[i].buffer);
			else
				free_pages((unsigned long)pg_vec[i].buffer,
					   order);
			pg_vec[i].buffer = NULL;
		}
L
Linus Torvalds 已提交
4007 4008 4009 4010
	}
	kfree(pg_vec);
}

4011
static char *alloc_one_pg_vec_page(unsigned long order)
4012
{
4013
	char *buffer;
4014 4015 4016 4017 4018 4019 4020
	gfp_t gfp_flags = GFP_KERNEL | __GFP_COMP |
			  __GFP_ZERO | __GFP_NOWARN | __GFP_NORETRY;

	buffer = (char *) __get_free_pages(gfp_flags, order);
	if (buffer)
		return buffer;

4021
	/* __get_free_pages failed, fall back to vmalloc */
E
Eric Dumazet 已提交
4022
	buffer = vzalloc((1 << order) * PAGE_SIZE);
4023 4024 4025
	if (buffer)
		return buffer;

4026
	/* vmalloc failed, lets dig into swap here */
4027
	gfp_flags &= ~__GFP_NORETRY;
4028
	buffer = (char *) __get_free_pages(gfp_flags, order);
4029 4030 4031
	if (buffer)
		return buffer;

4032
	/* complete and utter failure */
4033
	return NULL;
4034 4035
}

4036
static struct pgv *alloc_pg_vec(struct tpacket_req *req, int order)
4037 4038
{
	unsigned int block_nr = req->tp_block_nr;
4039
	struct pgv *pg_vec;
4040 4041
	int i;

4042
	pg_vec = kcalloc(block_nr, sizeof(struct pgv), GFP_KERNEL);
4043 4044 4045 4046
	if (unlikely(!pg_vec))
		goto out;

	for (i = 0; i < block_nr; i++) {
C
Changli Gao 已提交
4047
		pg_vec[i].buffer = alloc_one_pg_vec_page(order);
4048
		if (unlikely(!pg_vec[i].buffer))
4049 4050 4051 4052 4053 4054 4055 4056 4057 4058 4059
			goto out_free_pgvec;
	}

out:
	return pg_vec;

out_free_pgvec:
	free_pg_vec(pg_vec, order, block_nr);
	pg_vec = NULL;
	goto out;
}
L
Linus Torvalds 已提交
4060

4061
static int packet_set_ring(struct sock *sk, union tpacket_req_u *req_u,
J
Johann Baudy 已提交
4062
		int closing, int tx_ring)
L
Linus Torvalds 已提交
4063
{
4064
	struct pgv *pg_vec = NULL;
L
Linus Torvalds 已提交
4065
	struct packet_sock *po = pkt_sk(sk);
A
Al Viro 已提交
4066
	int was_running, order = 0;
J
Johann Baudy 已提交
4067 4068
	struct packet_ring_buffer *rb;
	struct sk_buff_head *rb_queue;
A
Al Viro 已提交
4069
	__be16 num;
4070 4071 4072 4073 4074 4075 4076 4077 4078
	int err = -EINVAL;
	/* Added to avoid minimal code churn */
	struct tpacket_req *req = &req_u->req;

	/* Opening a Tx-ring is NOT supported in TPACKET_V3 */
	if (!closing && tx_ring && (po->tp_version > TPACKET_V2)) {
		WARN(1, "Tx-ring is not supported.\n");
		goto out;
	}
4079

J
Johann Baudy 已提交
4080 4081
	rb = tx_ring ? &po->tx_ring : &po->rx_ring;
	rb_queue = tx_ring ? &sk->sk_write_queue : &sk->sk_receive_queue;
L
Linus Torvalds 已提交
4082

J
Johann Baudy 已提交
4083 4084 4085 4086
	err = -EBUSY;
	if (!closing) {
		if (atomic_read(&po->mapped))
			goto out;
4087
		if (packet_read_pending(rb))
J
Johann Baudy 已提交
4088 4089
			goto out;
	}
L
Linus Torvalds 已提交
4090

J
Johann Baudy 已提交
4091 4092 4093 4094 4095
	if (req->tp_block_nr) {
		/* Sanity tests and some calculations */
		err = -EBUSY;
		if (unlikely(rb->pg_vec))
			goto out;
L
Linus Torvalds 已提交
4096

4097 4098 4099 4100 4101 4102 4103
		switch (po->tp_version) {
		case TPACKET_V1:
			po->tp_hdrlen = TPACKET_HDRLEN;
			break;
		case TPACKET_V2:
			po->tp_hdrlen = TPACKET2_HDRLEN;
			break;
4104 4105 4106
		case TPACKET_V3:
			po->tp_hdrlen = TPACKET3_HDRLEN;
			break;
4107 4108
		}

J
Johann Baudy 已提交
4109
		err = -EINVAL;
4110
		if (unlikely((int)req->tp_block_size <= 0))
J
Johann Baudy 已提交
4111
			goto out;
4112
		if (unlikely(req->tp_block_size & (PAGE_SIZE - 1)))
J
Johann Baudy 已提交
4113
			goto out;
4114 4115 4116 4117
		if (po->tp_version >= TPACKET_V3 &&
		    (int)(req->tp_block_size -
			  BLK_PLUS_PRIV(req_u->req3.tp_sizeof_priv)) <= 0)
			goto out;
4118
		if (unlikely(req->tp_frame_size < po->tp_hdrlen +
J
Johann Baudy 已提交
4119 4120
					po->tp_reserve))
			goto out;
4121
		if (unlikely(req->tp_frame_size & (TPACKET_ALIGNMENT - 1)))
J
Johann Baudy 已提交
4122
			goto out;
L
Linus Torvalds 已提交
4123

J
Johann Baudy 已提交
4124 4125 4126 4127 4128 4129
		rb->frames_per_block = req->tp_block_size/req->tp_frame_size;
		if (unlikely(rb->frames_per_block <= 0))
			goto out;
		if (unlikely((rb->frames_per_block * req->tp_block_nr) !=
					req->tp_frame_nr))
			goto out;
L
Linus Torvalds 已提交
4130 4131

		err = -ENOMEM;
4132 4133 4134
		order = get_order(req->tp_block_size);
		pg_vec = alloc_pg_vec(req, order);
		if (unlikely(!pg_vec))
L
Linus Torvalds 已提交
4135
			goto out;
4136 4137 4138 4139 4140 4141
		switch (po->tp_version) {
		case TPACKET_V3:
		/* Transmit path is not supported. We checked
		 * it above but just being paranoid
		 */
			if (!tx_ring)
4142
				init_prb_bdqc(po, rb, pg_vec, req_u);
4143
			break;
4144 4145 4146
		default:
			break;
		}
J
Johann Baudy 已提交
4147 4148 4149 4150
	}
	/* Done */
	else {
		err = -EINVAL;
4151
		if (unlikely(req->tp_frame_nr))
J
Johann Baudy 已提交
4152
			goto out;
L
Linus Torvalds 已提交
4153 4154 4155 4156 4157 4158 4159 4160 4161 4162
	}

	lock_sock(sk);

	/* Detach socket from network */
	spin_lock(&po->bind_lock);
	was_running = po->running;
	num = po->num;
	if (was_running) {
		po->num = 0;
4163
		__unregister_prot_hook(sk, false);
L
Linus Torvalds 已提交
4164 4165
	}
	spin_unlock(&po->bind_lock);
4166

L
Linus Torvalds 已提交
4167 4168 4169
	synchronize_net();

	err = -EBUSY;
4170
	mutex_lock(&po->pg_vec_lock);
L
Linus Torvalds 已提交
4171 4172
	if (closing || atomic_read(&po->mapped) == 0) {
		err = 0;
J
Johann Baudy 已提交
4173
		spin_lock_bh(&rb_queue->lock);
4174
		swap(rb->pg_vec, pg_vec);
J
Johann Baudy 已提交
4175 4176 4177 4178 4179
		rb->frame_max = (req->tp_frame_nr - 1);
		rb->head = 0;
		rb->frame_size = req->tp_frame_size;
		spin_unlock_bh(&rb_queue->lock);

4180 4181
		swap(rb->pg_vec_order, order);
		swap(rb->pg_vec_len, req->tp_block_nr);
J
Johann Baudy 已提交
4182 4183 4184 4185 4186

		rb->pg_vec_pages = req->tp_block_size/PAGE_SIZE;
		po->prot_hook.func = (po->rx_ring.pg_vec) ?
						tpacket_rcv : packet_rcv;
		skb_queue_purge(rb_queue);
L
Linus Torvalds 已提交
4187
		if (atomic_read(&po->mapped))
E
Eric Dumazet 已提交
4188 4189
			pr_err("packet_mmap: vma is busy: %d\n",
			       atomic_read(&po->mapped));
L
Linus Torvalds 已提交
4190
	}
4191
	mutex_unlock(&po->pg_vec_lock);
L
Linus Torvalds 已提交
4192 4193

	spin_lock(&po->bind_lock);
4194
	if (was_running) {
L
Linus Torvalds 已提交
4195
		po->num = num;
4196
		register_prot_hook(sk);
L
Linus Torvalds 已提交
4197 4198
	}
	spin_unlock(&po->bind_lock);
4199 4200 4201
	if (closing && (po->tp_version > TPACKET_V2)) {
		/* Because we don't support block-based V3 on tx-ring */
		if (!tx_ring)
4202
			prb_shutdown_retire_blk_timer(po, rb_queue);
4203
	}
L
Linus Torvalds 已提交
4204 4205 4206 4207 4208 4209 4210 4211
	release_sock(sk);

	if (pg_vec)
		free_pg_vec(pg_vec, order, req->tp_block_nr);
out:
	return err;
}

J
Johann Baudy 已提交
4212 4213
static int packet_mmap(struct file *file, struct socket *sock,
		struct vm_area_struct *vma)
L
Linus Torvalds 已提交
4214 4215 4216
{
	struct sock *sk = sock->sk;
	struct packet_sock *po = pkt_sk(sk);
J
Johann Baudy 已提交
4217 4218
	unsigned long size, expected_size;
	struct packet_ring_buffer *rb;
L
Linus Torvalds 已提交
4219 4220 4221 4222 4223 4224 4225
	unsigned long start;
	int err = -EINVAL;
	int i;

	if (vma->vm_pgoff)
		return -EINVAL;

4226
	mutex_lock(&po->pg_vec_lock);
J
Johann Baudy 已提交
4227 4228 4229 4230 4231 4232 4233 4234 4235 4236 4237

	expected_size = 0;
	for (rb = &po->rx_ring; rb <= &po->tx_ring; rb++) {
		if (rb->pg_vec) {
			expected_size += rb->pg_vec_len
						* rb->pg_vec_pages
						* PAGE_SIZE;
		}
	}

	if (expected_size == 0)
L
Linus Torvalds 已提交
4238
		goto out;
J
Johann Baudy 已提交
4239 4240 4241

	size = vma->vm_end - vma->vm_start;
	if (size != expected_size)
L
Linus Torvalds 已提交
4242 4243 4244
		goto out;

	start = vma->vm_start;
J
Johann Baudy 已提交
4245 4246 4247 4248 4249
	for (rb = &po->rx_ring; rb <= &po->tx_ring; rb++) {
		if (rb->pg_vec == NULL)
			continue;

		for (i = 0; i < rb->pg_vec_len; i++) {
4250 4251
			struct page *page;
			void *kaddr = rb->pg_vec[i].buffer;
J
Johann Baudy 已提交
4252 4253
			int pg_num;

C
Changli Gao 已提交
4254 4255
			for (pg_num = 0; pg_num < rb->pg_vec_pages; pg_num++) {
				page = pgv_to_page(kaddr);
J
Johann Baudy 已提交
4256 4257 4258 4259
				err = vm_insert_page(vma, start, page);
				if (unlikely(err))
					goto out;
				start += PAGE_SIZE;
4260
				kaddr += PAGE_SIZE;
J
Johann Baudy 已提交
4261
			}
4262
		}
L
Linus Torvalds 已提交
4263
	}
J
Johann Baudy 已提交
4264

4265
	atomic_inc(&po->mapped);
L
Linus Torvalds 已提交
4266 4267 4268 4269
	vma->vm_ops = &packet_mmap_ops;
	err = 0;

out:
4270
	mutex_unlock(&po->pg_vec_lock);
L
Linus Torvalds 已提交
4271 4272 4273
	return err;
}

4274
static const struct proto_ops packet_ops_spkt = {
L
Linus Torvalds 已提交
4275 4276 4277 4278 4279 4280 4281 4282 4283 4284 4285 4286 4287 4288 4289 4290 4291 4292 4293 4294
	.family =	PF_PACKET,
	.owner =	THIS_MODULE,
	.release =	packet_release,
	.bind =		packet_bind_spkt,
	.connect =	sock_no_connect,
	.socketpair =	sock_no_socketpair,
	.accept =	sock_no_accept,
	.getname =	packet_getname_spkt,
	.poll =		datagram_poll,
	.ioctl =	packet_ioctl,
	.listen =	sock_no_listen,
	.shutdown =	sock_no_shutdown,
	.setsockopt =	sock_no_setsockopt,
	.getsockopt =	sock_no_getsockopt,
	.sendmsg =	packet_sendmsg_spkt,
	.recvmsg =	packet_recvmsg,
	.mmap =		sock_no_mmap,
	.sendpage =	sock_no_sendpage,
};

4295
static const struct proto_ops packet_ops = {
L
Linus Torvalds 已提交
4296 4297 4298 4299 4300 4301 4302
	.family =	PF_PACKET,
	.owner =	THIS_MODULE,
	.release =	packet_release,
	.bind =		packet_bind,
	.connect =	sock_no_connect,
	.socketpair =	sock_no_socketpair,
	.accept =	sock_no_accept,
4303
	.getname =	packet_getname,
L
Linus Torvalds 已提交
4304 4305 4306 4307 4308 4309 4310 4311 4312 4313 4314 4315
	.poll =		packet_poll,
	.ioctl =	packet_ioctl,
	.listen =	sock_no_listen,
	.shutdown =	sock_no_shutdown,
	.setsockopt =	packet_setsockopt,
	.getsockopt =	packet_getsockopt,
	.sendmsg =	packet_sendmsg,
	.recvmsg =	packet_recvmsg,
	.mmap =		packet_mmap,
	.sendpage =	sock_no_sendpage,
};

4316
static const struct net_proto_family packet_family_ops = {
L
Linus Torvalds 已提交
4317 4318 4319 4320 4321 4322
	.family =	PF_PACKET,
	.create =	packet_create,
	.owner	=	THIS_MODULE,
};

static struct notifier_block packet_netdev_notifier = {
E
Eric Dumazet 已提交
4323
	.notifier_call =	packet_notifier,
L
Linus Torvalds 已提交
4324 4325 4326 4327 4328
};

#ifdef CONFIG_PROC_FS

static void *packet_seq_start(struct seq_file *seq, loff_t *pos)
4329
	__acquires(RCU)
L
Linus Torvalds 已提交
4330
{
4331
	struct net *net = seq_file_net(seq);
4332 4333 4334

	rcu_read_lock();
	return seq_hlist_start_head_rcu(&net->packet.sklist, *pos);
L
Linus Torvalds 已提交
4335 4336 4337 4338
}

static void *packet_seq_next(struct seq_file *seq, void *v, loff_t *pos)
{
4339
	struct net *net = seq_file_net(seq);
4340
	return seq_hlist_next_rcu(v, &net->packet.sklist, pos);
L
Linus Torvalds 已提交
4341 4342 4343
}

static void packet_seq_stop(struct seq_file *seq, void *v)
4344
	__releases(RCU)
L
Linus Torvalds 已提交
4345
{
4346
	rcu_read_unlock();
L
Linus Torvalds 已提交
4347 4348
}

4349
static int packet_seq_show(struct seq_file *seq, void *v)
L
Linus Torvalds 已提交
4350 4351 4352 4353
{
	if (v == SEQ_START_TOKEN)
		seq_puts(seq, "sk       RefCnt Type Proto  Iface R Rmem   User   Inode\n");
	else {
4354
		struct sock *s = sk_entry(v);
L
Linus Torvalds 已提交
4355 4356 4357
		const struct packet_sock *po = pkt_sk(s);

		seq_printf(seq,
D
Dan Rosenberg 已提交
4358
			   "%pK %-6d %-4d %04x   %-5d %1d %-6u %-6u %-6lu\n",
L
Linus Torvalds 已提交
4359 4360 4361 4362 4363 4364 4365
			   s,
			   atomic_read(&s->sk_refcnt),
			   s->sk_type,
			   ntohs(po->num),
			   po->ifindex,
			   po->running,
			   atomic_read(&s->sk_rmem_alloc),
4366
			   from_kuid_munged(seq_user_ns(seq), sock_i_uid(s)),
E
Eric Dumazet 已提交
4367
			   sock_i_ino(s));
L
Linus Torvalds 已提交
4368 4369 4370 4371 4372
	}

	return 0;
}

4373
static const struct seq_operations packet_seq_ops = {
L
Linus Torvalds 已提交
4374 4375 4376 4377 4378 4379 4380 4381
	.start	= packet_seq_start,
	.next	= packet_seq_next,
	.stop	= packet_seq_stop,
	.show	= packet_seq_show,
};

static int packet_seq_open(struct inode *inode, struct file *file)
{
4382 4383
	return seq_open_net(inode, file, &packet_seq_ops,
			    sizeof(struct seq_net_private));
L
Linus Torvalds 已提交
4384 4385
}

4386
static const struct file_operations packet_seq_fops = {
L
Linus Torvalds 已提交
4387 4388 4389 4390
	.owner		= THIS_MODULE,
	.open		= packet_seq_open,
	.read		= seq_read,
	.llseek		= seq_lseek,
4391
	.release	= seq_release_net,
L
Linus Torvalds 已提交
4392 4393 4394 4395
};

#endif

4396
static int __net_init packet_net_init(struct net *net)
4397
{
4398
	mutex_init(&net->packet.sklist_lock);
4399
	INIT_HLIST_HEAD(&net->packet.sklist);
4400

4401
	if (!proc_create("packet", 0, net->proc_net, &packet_seq_fops))
4402 4403 4404 4405 4406
		return -ENOMEM;

	return 0;
}

4407
static void __net_exit packet_net_exit(struct net *net)
4408
{
4409
	remove_proc_entry("packet", net->proc_net);
4410 4411 4412 4413 4414 4415 4416 4417
}

static struct pernet_operations packet_net_ops = {
	.init = packet_net_init,
	.exit = packet_net_exit,
};


L
Linus Torvalds 已提交
4418 4419 4420
static void __exit packet_exit(void)
{
	unregister_netdevice_notifier(&packet_netdev_notifier);
4421
	unregister_pernet_subsys(&packet_net_ops);
L
Linus Torvalds 已提交
4422 4423 4424 4425 4426 4427 4428 4429 4430 4431 4432 4433
	sock_unregister(PF_PACKET);
	proto_unregister(&packet_proto);
}

static int __init packet_init(void)
{
	int rc = proto_register(&packet_proto, 0);

	if (rc != 0)
		goto out;

	sock_register(&packet_family_ops);
4434
	register_pernet_subsys(&packet_net_ops);
L
Linus Torvalds 已提交
4435 4436 4437 4438 4439 4440 4441 4442 4443
	register_netdevice_notifier(&packet_netdev_notifier);
out:
	return rc;
}

module_init(packet_init);
module_exit(packet_exit);
MODULE_LICENSE("GPL");
MODULE_ALIAS_NETPROTO(PF_PACKET);