netpoll.c 30.6 KB
Newer Older
L
Linus Torvalds 已提交
1 2 3 4 5 6 7 8 9 10 11
/*
 * Common framework for low-level network console, dump, and debugger code
 *
 * Sep 8 2003  Matt Mackall <mpm@selenic.com>
 *
 * based on the netconsole code from:
 *
 * Copyright (C) 2001  Ingo Molnar <mingo@redhat.com>
 * Copyright (C) 2002  Red Hat, Inc.
 */

12 13
#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt

14
#include <linux/moduleparam.h>
L
Linus Torvalds 已提交
15 16 17
#include <linux/netdevice.h>
#include <linux/etherdevice.h>
#include <linux/string.h>
18
#include <linux/if_arp.h>
L
Linus Torvalds 已提交
19 20 21 22 23 24 25 26
#include <linux/inetdevice.h>
#include <linux/inet.h>
#include <linux/interrupt.h>
#include <linux/netpoll.h>
#include <linux/sched.h>
#include <linux/delay.h>
#include <linux/rcupdate.h>
#include <linux/workqueue.h>
27
#include <linux/slab.h>
28
#include <linux/export.h>
29
#include <linux/if_vlan.h>
L
Linus Torvalds 已提交
30 31
#include <net/tcp.h>
#include <net/udp.h>
C
Cong Wang 已提交
32 33 34
#include <net/addrconf.h>
#include <net/ndisc.h>
#include <net/ip6_checksum.h>
L
Linus Torvalds 已提交
35
#include <asm/unaligned.h>
36
#include <trace/events/napi.h>
L
Linus Torvalds 已提交
37 38 39 40 41 42 43 44 45

/*
 * We maintain a small pool of fully-sized skbs, to make sure the
 * message gets out even in extreme OOM situations.
 */

#define MAX_UDP_CHUNK 1460
#define MAX_SKBS 32

46
static struct sk_buff_head skb_pool;
L
Linus Torvalds 已提交
47 48 49

static atomic_t trapped;

50
DEFINE_STATIC_SRCU(netpoll_srcu);
51

S
Stephen Hemminger 已提交
52
#define USEC_PER_POLL	50
53 54
#define NETPOLL_RX_ENABLED  1
#define NETPOLL_RX_DROP     2
L
Linus Torvalds 已提交
55

J
Joe Perches 已提交
56 57 58 59 60
#define MAX_SKB_SIZE							\
	(sizeof(struct ethhdr) +					\
	 sizeof(struct iphdr) +						\
	 sizeof(struct udphdr) +					\
	 MAX_UDP_CHUNK)
L
Linus Torvalds 已提交
61

62
static void zap_completion_queue(void);
C
Cong Wang 已提交
63
static void netpoll_neigh_reply(struct sk_buff *skb, struct netpoll_info *npinfo);
64
static void netpoll_async_cleanup(struct work_struct *work);
L
Linus Torvalds 已提交
65

66 67 68
static unsigned int carrier_timeout = 4;
module_param(carrier_timeout, uint, 0644);

69 70 71 72 73 74 75
#define np_info(np, fmt, ...)				\
	pr_info("%s: " fmt, np->name, ##__VA_ARGS__)
#define np_err(np, fmt, ...)				\
	pr_err("%s: " fmt, np->name, ##__VA_ARGS__)
#define np_notice(np, fmt, ...)				\
	pr_notice("%s: " fmt, np->name, ##__VA_ARGS__)

D
David Howells 已提交
76
static void queue_process(struct work_struct *work)
L
Linus Torvalds 已提交
77
{
78 79
	struct netpoll_info *npinfo =
		container_of(work, struct netpoll_info, tx_work.work);
L
Linus Torvalds 已提交
80
	struct sk_buff *skb;
I
Ingo Molnar 已提交
81
	unsigned long flags;
L
Linus Torvalds 已提交
82

S
Stephen Hemminger 已提交
83 84
	while ((skb = skb_dequeue(&npinfo->txq))) {
		struct net_device *dev = skb->dev;
85
		const struct net_device_ops *ops = dev->netdev_ops;
86
		struct netdev_queue *txq;
L
Linus Torvalds 已提交
87

S
Stephen Hemminger 已提交
88 89 90 91
		if (!netif_device_present(dev) || !netif_running(dev)) {
			__kfree_skb(skb);
			continue;
		}
L
Linus Torvalds 已提交
92

93 94
		txq = netdev_get_tx_queue(dev, skb_get_queue_mapping(skb));

I
Ingo Molnar 已提交
95
		local_irq_save(flags);
96
		__netif_tx_lock(txq, smp_processor_id());
97
		if (netif_xmit_frozen_or_stopped(txq) ||
98
		    ops->ndo_start_xmit(skb, dev) != NETDEV_TX_OK) {
S
Stephen Hemminger 已提交
99
			skb_queue_head(&npinfo->txq, skb);
100
			__netif_tx_unlock(txq);
I
Ingo Molnar 已提交
101
			local_irq_restore(flags);
L
Linus Torvalds 已提交
102

103
			schedule_delayed_work(&npinfo->tx_work, HZ/10);
S
Stephen Hemminger 已提交
104 105
			return;
		}
106
		__netif_tx_unlock(txq);
I
Ingo Molnar 已提交
107
		local_irq_restore(flags);
L
Linus Torvalds 已提交
108 109 110
	}
}

111 112
static __sum16 checksum_udp(struct sk_buff *skb, struct udphdr *uh,
			    unsigned short ulen, __be32 saddr, __be32 daddr)
L
Linus Torvalds 已提交
113
{
114
	__wsum psum;
115

116
	if (uh->check == 0 || skb_csum_unnecessary(skb))
L
Linus Torvalds 已提交
117 118
		return 0;

119 120
	psum = csum_tcpudp_nofold(saddr, daddr, ulen, IPPROTO_UDP, 0);

121
	if (skb->ip_summed == CHECKSUM_COMPLETE &&
122
	    !csum_fold(csum_add(psum, skb->csum)))
123
		return 0;
L
Linus Torvalds 已提交
124

125
	skb->csum = psum;
L
Linus Torvalds 已提交
126

127
	return __skb_checksum_complete(skb);
L
Linus Torvalds 已提交
128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145
}

/*
 * Check whether delayed processing was scheduled for our NIC. If so,
 * we attempt to grab the poll lock and use ->poll() to pump the card.
 * If this fails, either we've recursed in ->poll() or it's already
 * running on another CPU.
 *
 * Note: we don't mask interrupts with this lock because we're using
 * trylock here and interrupts are already disabled in the softirq
 * case. Further, we test the poll_owner to avoid recursion on UP
 * systems where the lock doesn't exist.
 *
 * In cases where there is bi-directional communications, reading only
 * one message at a time can lead to packets being dropped by the
 * network adapter, forcing superfluous retries and possibly timeouts.
 * Thus, we set our budget to greater than 1.
 */
146 147 148 149 150 151 152 153 154 155 156 157
static int poll_one_napi(struct netpoll_info *npinfo,
			 struct napi_struct *napi, int budget)
{
	int work;

	/* net_rx_action's ->poll() invocations and our's are
	 * synchronized by this test which is only made while
	 * holding the napi->poll_lock.
	 */
	if (!test_bit(NAPI_STATE_SCHED, &napi->state))
		return budget;

158
	npinfo->rx_flags |= NETPOLL_RX_DROP;
159
	atomic_inc(&trapped);
160
	set_bit(NAPI_STATE_NPSVC, &napi->state);
161 162

	work = napi->poll(napi, budget);
163
	trace_napi_poll(napi);
164

165
	clear_bit(NAPI_STATE_NPSVC, &napi->state);
166
	atomic_dec(&trapped);
167
	npinfo->rx_flags &= ~NETPOLL_RX_DROP;
168 169 170 171

	return budget - work;
}

172
static void poll_napi(struct net_device *dev)
L
Linus Torvalds 已提交
173
{
174
	struct napi_struct *napi;
L
Linus Torvalds 已提交
175 176
	int budget = 16;

177
	list_for_each_entry(napi, &dev->napi_list, dev_list) {
178
		if (napi->poll_owner != smp_processor_id() &&
179
		    spin_trylock(&napi->poll_lock)) {
180 181
			budget = poll_one_napi(rcu_dereference_bh(dev->npinfo),
					       napi, budget);
182
			spin_unlock(&napi->poll_lock);
183

184
			if (!budget)
185
				break;
186
		}
L
Linus Torvalds 已提交
187 188 189
	}
}

C
Cong Wang 已提交
190
static void service_neigh_queue(struct netpoll_info *npi)
191
{
192 193
	if (npi) {
		struct sk_buff *skb;
194

C
Cong Wang 已提交
195 196
		while ((skb = skb_dequeue(&npi->neigh_tx)))
			netpoll_neigh_reply(skb, npi);
197 198 199
	}
}

200
static void netpoll_poll_dev(struct net_device *dev)
L
Linus Torvalds 已提交
201
{
202
	const struct net_device_ops *ops;
203
	struct netpoll_info *ni = rcu_dereference_bh(dev->npinfo);
204

205 206 207 208
	/* Don't do any rx activity if the dev_lock mutex is held
	 * the dev_open/close paths use this to block netpoll activity
	 * while changing device state
	 */
209
	if (down_trylock(&ni->dev_lock))
210 211
		return;

212
	if (!netif_running(dev)) {
213
		up(&ni->dev_lock);
214
		return;
215
	}
216 217

	ops = dev->netdev_ops;
218
	if (!ops->ndo_poll_controller) {
219
		up(&ni->dev_lock);
L
Linus Torvalds 已提交
220
		return;
221
	}
L
Linus Torvalds 已提交
222 223

	/* Process pending work on NIC */
224
	ops->ndo_poll_controller(dev);
225 226

	poll_napi(dev);
L
Linus Torvalds 已提交
227

228
	up(&ni->dev_lock);
229

230
	if (dev->flags & IFF_SLAVE) {
231
		if (ni) {
232
			struct net_device *bond_dev;
233
			struct sk_buff *skb;
234 235 236 237
			struct netpoll_info *bond_ni;

			bond_dev = netdev_master_upper_dev_get_rcu(dev);
			bond_ni = rcu_dereference_bh(bond_dev->npinfo);
C
Cong Wang 已提交
238
			while ((skb = skb_dequeue(&ni->neigh_tx))) {
239
				skb->dev = bond_dev;
C
Cong Wang 已提交
240
				skb_queue_tail(&bond_ni->neigh_tx, skb);
241 242 243 244
			}
		}
	}

C
Cong Wang 已提交
245
	service_neigh_queue(ni);
246

247
	zap_completion_queue();
L
Linus Torvalds 已提交
248 249
}

250 251 252 253 254 255 256 257
int netpoll_rx_disable(struct net_device *dev)
{
	struct netpoll_info *ni;
	int idx;
	might_sleep();
	idx = srcu_read_lock(&netpoll_srcu);
	ni = srcu_dereference(dev->npinfo, &netpoll_srcu);
	if (ni)
258
		down(&ni->dev_lock);
259 260 261 262 263 264 265 266 267 268 269
	srcu_read_unlock(&netpoll_srcu, idx);
	return 0;
}
EXPORT_SYMBOL(netpoll_rx_disable);

void netpoll_rx_enable(struct net_device *dev)
{
	struct netpoll_info *ni;
	rcu_read_lock();
	ni = rcu_dereference(dev->npinfo);
	if (ni)
270
		up(&ni->dev_lock);
271 272 273 274
	rcu_read_unlock();
}
EXPORT_SYMBOL(netpoll_rx_enable);

L
Linus Torvalds 已提交
275 276 277 278 279
static void refill_skbs(void)
{
	struct sk_buff *skb;
	unsigned long flags;

280 281
	spin_lock_irqsave(&skb_pool.lock, flags);
	while (skb_pool.qlen < MAX_SKBS) {
L
Linus Torvalds 已提交
282 283 284 285
		skb = alloc_skb(MAX_SKB_SIZE, GFP_ATOMIC);
		if (!skb)
			break;

286
		__skb_queue_tail(&skb_pool, skb);
L
Linus Torvalds 已提交
287
	}
288
	spin_unlock_irqrestore(&skb_pool.lock, flags);
L
Linus Torvalds 已提交
289 290
}

291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318
static void zap_completion_queue(void)
{
	unsigned long flags;
	struct softnet_data *sd = &get_cpu_var(softnet_data);

	if (sd->completion_queue) {
		struct sk_buff *clist;

		local_irq_save(flags);
		clist = sd->completion_queue;
		sd->completion_queue = NULL;
		local_irq_restore(flags);

		while (clist != NULL) {
			struct sk_buff *skb = clist;
			clist = clist->next;
			if (skb->destructor) {
				atomic_inc(&skb->users);
				dev_kfree_skb_any(skb); /* put this one back */
			} else {
				__kfree_skb(skb);
			}
		}
	}

	put_cpu_var(softnet_data);
}

319
static struct sk_buff *find_skb(struct netpoll *np, int len, int reserve)
L
Linus Torvalds 已提交
320
{
321 322
	int count = 0;
	struct sk_buff *skb;
L
Linus Torvalds 已提交
323

324
	zap_completion_queue();
325
	refill_skbs();
L
Linus Torvalds 已提交
326 327 328
repeat:

	skb = alloc_skb(len, GFP_ATOMIC);
329 330
	if (!skb)
		skb = skb_dequeue(&skb_pool);
L
Linus Torvalds 已提交
331 332

	if (!skb) {
333
		if (++count < 10) {
334
			netpoll_poll_dev(np->dev);
335
			goto repeat;
L
Linus Torvalds 已提交
336
		}
337
		return NULL;
L
Linus Torvalds 已提交
338 339 340 341 342 343 344
	}

	atomic_set(&skb->users, 1);
	skb_reserve(skb, reserve);
	return skb;
}

345 346 347 348 349 350 351 352 353 354 355
static int netpoll_owner_active(struct net_device *dev)
{
	struct napi_struct *napi;

	list_for_each_entry(napi, &dev->napi_list, dev_list) {
		if (napi->poll_owner == smp_processor_id())
			return 1;
	}
	return 0;
}

356
/* call with IRQ disabled */
357 358
void netpoll_send_skb_on_dev(struct netpoll *np, struct sk_buff *skb,
			     struct net_device *dev)
L
Linus Torvalds 已提交
359
{
S
Stephen Hemminger 已提交
360 361
	int status = NETDEV_TX_BUSY;
	unsigned long tries;
362
	const struct net_device_ops *ops = dev->netdev_ops;
H
Herbert Xu 已提交
363
	/* It is up to the caller to keep npinfo alive. */
364
	struct netpoll_info *npinfo;
S
Stephen Hemminger 已提交
365

366 367 368
	WARN_ON_ONCE(!irqs_disabled());

	npinfo = rcu_dereference_bh(np->dev->npinfo);
369 370 371 372
	if (!npinfo || !netif_running(dev) || !netif_device_present(dev)) {
		__kfree_skb(skb);
		return;
	}
S
Stephen Hemminger 已提交
373 374

	/* don't get messages out of order, and no recursion */
375
	if (skb_queue_len(&npinfo->txq) == 0 && !netpoll_owner_active(dev)) {
376
		struct netdev_queue *txq;
377

378
		txq = netdev_pick_tx(dev, skb);
379

380 381 382
		/* try until next clock tick */
		for (tries = jiffies_to_usecs(1)/USEC_PER_POLL;
		     tries > 0; --tries) {
383
			if (__netif_tx_trylock(txq)) {
384
				if (!netif_xmit_stopped(txq)) {
385
					if (vlan_tx_tag_present(skb) &&
386 387 388
					    !vlan_hw_offload_capable(netif_skb_features(skb),
								     skb->vlan_proto)) {
						skb = __vlan_put_tag(skb, skb->vlan_proto, vlan_tx_tag_get(skb));
389 390 391 392 393
						if (unlikely(!skb))
							break;
						skb->vlan_tci = 0;
					}

394
					status = ops->ndo_start_xmit(skb, dev);
E
Eric Dumazet 已提交
395 396 397
					if (status == NETDEV_TX_OK)
						txq_trans_update(txq);
				}
398
				__netif_tx_unlock(txq);
399 400 401 402 403

				if (status == NETDEV_TX_OK)
					break;

			}
404 405

			/* tickle device maybe there is some cleanup */
406
			netpoll_poll_dev(np->dev);
407 408

			udelay(USEC_PER_POLL);
M
Matt Mackall 已提交
409
		}
410 411

		WARN_ONCE(!irqs_disabled(),
412
			"netpoll_send_skb_on_dev(): %s enabled interrupts in poll (%pF)\n",
413 414
			dev->name, ops->ndo_start_xmit);

L
Linus Torvalds 已提交
415 416
	}

S
Stephen Hemminger 已提交
417
	if (status != NETDEV_TX_OK) {
S
Stephen Hemminger 已提交
418
		skb_queue_tail(&npinfo->txq, skb);
419
		schedule_delayed_work(&npinfo->tx_work,0);
L
Linus Torvalds 已提交
420 421
	}
}
422
EXPORT_SYMBOL(netpoll_send_skb_on_dev);
L
Linus Torvalds 已提交
423 424 425

void netpoll_send_udp(struct netpoll *np, const char *msg, int len)
{
426
	int total_len, ip_len, udp_len;
L
Linus Torvalds 已提交
427 428 429 430
	struct sk_buff *skb;
	struct udphdr *udph;
	struct iphdr *iph;
	struct ethhdr *eth;
431
	static atomic_t ip_ident;
C
Cong Wang 已提交
432
	struct ipv6hdr *ip6h;
L
Linus Torvalds 已提交
433 434

	udp_len = len + sizeof(*udph);
C
Cong Wang 已提交
435 436 437
	if (np->ipv6)
		ip_len = udp_len + sizeof(*ip6h);
	else
C
Cong Wang 已提交
438 439
		ip_len = udp_len + sizeof(*iph);

440
	total_len = ip_len + LL_RESERVED_SPACE(np->dev);
L
Linus Torvalds 已提交
441

442 443
	skb = find_skb(np, total_len + np->dev->needed_tailroom,
		       total_len - len);
L
Linus Torvalds 已提交
444 445 446
	if (!skb)
		return;

447
	skb_copy_to_linear_data(skb, msg, len);
448
	skb_put(skb, len);
L
Linus Torvalds 已提交
449

450 451 452
	skb_push(skb, sizeof(*udph));
	skb_reset_transport_header(skb);
	udph = udp_hdr(skb);
L
Linus Torvalds 已提交
453 454 455
	udph->source = htons(np->local_port);
	udph->dest = htons(np->remote_port);
	udph->len = htons(udp_len);
C
Cong Wang 已提交
456

C
Cong Wang 已提交
457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485
	if (np->ipv6) {
		udph->check = 0;
		udph->check = csum_ipv6_magic(&np->local_ip.in6,
					      &np->remote_ip.in6,
					      udp_len, IPPROTO_UDP,
					      csum_partial(udph, udp_len, 0));
		if (udph->check == 0)
			udph->check = CSUM_MANGLED_0;

		skb_push(skb, sizeof(*ip6h));
		skb_reset_network_header(skb);
		ip6h = ipv6_hdr(skb);

		/* ip6h->version = 6; ip6h->priority = 0; */
		put_unaligned(0x60, (unsigned char *)ip6h);
		ip6h->flow_lbl[0] = 0;
		ip6h->flow_lbl[1] = 0;
		ip6h->flow_lbl[2] = 0;

		ip6h->payload_len = htons(sizeof(struct udphdr) + len);
		ip6h->nexthdr = IPPROTO_UDP;
		ip6h->hop_limit = 32;
		ip6h->saddr = np->local_ip.in6;
		ip6h->daddr = np->remote_ip.in6;

		eth = (struct ethhdr *) skb_push(skb, ETH_HLEN);
		skb_reset_mac_header(skb);
		skb->protocol = eth->h_proto = htons(ETH_P_IPV6);
	} else {
C
Cong Wang 已提交
486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515
		udph->check = 0;
		udph->check = csum_tcpudp_magic(np->local_ip.ip,
						np->remote_ip.ip,
						udp_len, IPPROTO_UDP,
						csum_partial(udph, udp_len, 0));
		if (udph->check == 0)
			udph->check = CSUM_MANGLED_0;

		skb_push(skb, sizeof(*iph));
		skb_reset_network_header(skb);
		iph = ip_hdr(skb);

		/* iph->version = 4; iph->ihl = 5; */
		put_unaligned(0x45, (unsigned char *)iph);
		iph->tos      = 0;
		put_unaligned(htons(ip_len), &(iph->tot_len));
		iph->id       = htons(atomic_inc_return(&ip_ident));
		iph->frag_off = 0;
		iph->ttl      = 64;
		iph->protocol = IPPROTO_UDP;
		iph->check    = 0;
		put_unaligned(np->local_ip.ip, &(iph->saddr));
		put_unaligned(np->remote_ip.ip, &(iph->daddr));
		iph->check    = ip_fast_csum((unsigned char *)iph, iph->ihl);

		eth = (struct ethhdr *) skb_push(skb, ETH_HLEN);
		skb_reset_mac_header(skb);
		skb->protocol = eth->h_proto = htons(ETH_P_IP);
	}

516 517
	memcpy(eth->h_source, np->dev->dev_addr, ETH_ALEN);
	memcpy(eth->h_dest, np->remote_mac, ETH_ALEN);
L
Linus Torvalds 已提交
518 519 520 521 522

	skb->dev = np->dev;

	netpoll_send_skb(np, skb);
}
E
Eric Dumazet 已提交
523
EXPORT_SYMBOL(netpoll_send_udp);
L
Linus Torvalds 已提交
524

C
Cong Wang 已提交
525
static void netpoll_neigh_reply(struct sk_buff *skb, struct netpoll_info *npinfo)
L
Linus Torvalds 已提交
526
{
C
Cong Wang 已提交
527
	int size, type = ARPOP_REPLY;
A
Al Viro 已提交
528
	__be32 sip, tip;
529
	unsigned char *sha;
L
Linus Torvalds 已提交
530
	struct sk_buff *send_skb;
531 532
	struct netpoll *np, *tmp;
	unsigned long flags;
533
	int hlen, tlen;
C
Cong Wang 已提交
534
	int hits = 0, proto;
535 536 537 538 539 540 541 542 543 544 545 546

	if (list_empty(&npinfo->rx_np))
		return;

	/* Before checking the packet, we do some early
	   inspection whether this is interesting at all */
	spin_lock_irqsave(&npinfo->rx_lock, flags);
	list_for_each_entry_safe(np, tmp, &npinfo->rx_np, rx) {
		if (np->dev == skb->dev)
			hits++;
	}
	spin_unlock_irqrestore(&npinfo->rx_lock, flags);
L
Linus Torvalds 已提交
547

548 549
	/* No netpoll struct is using this dev */
	if (!hits)
550
		return;
L
Linus Torvalds 已提交
551

C
Cong Wang 已提交
552 553
	proto = ntohs(eth_hdr(skb)->h_proto);
	if (proto == ETH_P_IP) {
C
Cong Wang 已提交
554 555
		struct arphdr *arp;
		unsigned char *arp_ptr;
C
Cong Wang 已提交
556 557 558
		/* No arp on this interface */
		if (skb->dev->flags & IFF_NOARP)
			return;
L
Linus Torvalds 已提交
559

C
Cong Wang 已提交
560 561
		if (!pskb_may_pull(skb, arp_hdr_len(skb->dev)))
			return;
L
Linus Torvalds 已提交
562

C
Cong Wang 已提交
563 564 565
		skb_reset_network_header(skb);
		skb_reset_transport_header(skb);
		arp = arp_hdr(skb);
L
Linus Torvalds 已提交
566

C
Cong Wang 已提交
567 568 569 570 571
		if ((arp->ar_hrd != htons(ARPHRD_ETHER) &&
		     arp->ar_hrd != htons(ARPHRD_IEEE802)) ||
		    arp->ar_pro != htons(ETH_P_IP) ||
		    arp->ar_op != htons(ARPOP_REQUEST))
			return;
L
Linus Torvalds 已提交
572

C
Cong Wang 已提交
573 574 575 576 577 578 579 580 581 582
		arp_ptr = (unsigned char *)(arp+1);
		/* save the location of the src hw addr */
		sha = arp_ptr;
		arp_ptr += skb->dev->addr_len;
		memcpy(&sip, arp_ptr, 4);
		arp_ptr += 4;
		/* If we actually cared about dst hw addr,
		   it would get copied here */
		arp_ptr += skb->dev->addr_len;
		memcpy(&tip, arp_ptr, 4);
L
Linus Torvalds 已提交
583

C
Cong Wang 已提交
584 585 586
		/* Should we ignore arp? */
		if (ipv4_is_loopback(tip) || ipv4_is_multicast(tip))
			return;
L
Linus Torvalds 已提交
587

C
Cong Wang 已提交
588
		size = arp_hdr_len(skb->dev);
L
Linus Torvalds 已提交
589

C
Cong Wang 已提交
590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606
		spin_lock_irqsave(&npinfo->rx_lock, flags);
		list_for_each_entry_safe(np, tmp, &npinfo->rx_np, rx) {
			if (tip != np->local_ip.ip)
				continue;

			hlen = LL_RESERVED_SPACE(np->dev);
			tlen = np->dev->needed_tailroom;
			send_skb = find_skb(np, size + hlen + tlen, hlen);
			if (!send_skb)
				continue;

			skb_reset_network_header(send_skb);
			arp = (struct arphdr *) skb_put(send_skb, size);
			send_skb->dev = skb->dev;
			send_skb->protocol = htons(ETH_P_ARP);

			/* Fill the device header for the ARP frame */
C
Cong Wang 已提交
607
			if (dev_hard_header(send_skb, skb->dev, ETH_P_ARP,
C
Cong Wang 已提交
608 609 610 611 612
					    sha, np->dev->dev_addr,
					    send_skb->len) < 0) {
				kfree_skb(send_skb);
				continue;
			}
L
Linus Torvalds 已提交
613

C
Cong Wang 已提交
614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641
			/*
			 * Fill out the arp protocol part.
			 *
			 * we only support ethernet device type,
			 * which (according to RFC 1390) should
			 * always equal 1 (Ethernet).
			 */

			arp->ar_hrd = htons(np->dev->type);
			arp->ar_pro = htons(ETH_P_IP);
			arp->ar_hln = np->dev->addr_len;
			arp->ar_pln = 4;
			arp->ar_op = htons(type);

			arp_ptr = (unsigned char *)(arp + 1);
			memcpy(arp_ptr, np->dev->dev_addr, np->dev->addr_len);
			arp_ptr += np->dev->addr_len;
			memcpy(arp_ptr, &tip, 4);
			arp_ptr += 4;
			memcpy(arp_ptr, sha, np->dev->addr_len);
			arp_ptr += np->dev->addr_len;
			memcpy(arp_ptr, &sip, 4);

			netpoll_send_skb(np, send_skb);

			/* If there are several rx_hooks for the same address,
			   we're fine by sending a single reply */
			break;
642
		}
C
Cong Wang 已提交
643
		spin_unlock_irqrestore(&npinfo->rx_lock, flags);
C
Cong Wang 已提交
644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679
	} else if( proto == ETH_P_IPV6) {
#if IS_ENABLED(CONFIG_IPV6)
		struct nd_msg *msg;
		u8 *lladdr = NULL;
		struct ipv6hdr *hdr;
		struct icmp6hdr *icmp6h;
		const struct in6_addr *saddr;
		const struct in6_addr *daddr;
		struct inet6_dev *in6_dev = NULL;
		struct in6_addr *target;

		in6_dev = in6_dev_get(skb->dev);
		if (!in6_dev || !in6_dev->cnf.accept_ra)
			return;

		if (!pskb_may_pull(skb, skb->len))
			return;

		msg = (struct nd_msg *)skb_transport_header(skb);

		__skb_push(skb, skb->data - skb_transport_header(skb));

		if (ipv6_hdr(skb)->hop_limit != 255)
			return;
		if (msg->icmph.icmp6_code != 0)
			return;
		if (msg->icmph.icmp6_type != NDISC_NEIGHBOUR_SOLICITATION)
			return;

		saddr = &ipv6_hdr(skb)->saddr;
		daddr = &ipv6_hdr(skb)->daddr;

		size = sizeof(struct icmp6hdr) + sizeof(struct in6_addr);

		spin_lock_irqsave(&npinfo->rx_lock, flags);
		list_for_each_entry_safe(np, tmp, &npinfo->rx_np, rx) {
680
			if (!ipv6_addr_equal(daddr, &np->local_ip.in6))
C
Cong Wang 已提交
681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710
				continue;

			hlen = LL_RESERVED_SPACE(np->dev);
			tlen = np->dev->needed_tailroom;
			send_skb = find_skb(np, size + hlen + tlen, hlen);
			if (!send_skb)
				continue;

			send_skb->protocol = htons(ETH_P_IPV6);
			send_skb->dev = skb->dev;

			skb_reset_network_header(send_skb);
			skb_put(send_skb, sizeof(struct ipv6hdr));
			hdr = ipv6_hdr(send_skb);

			*(__be32*)hdr = htonl(0x60000000);

			hdr->payload_len = htons(size);
			hdr->nexthdr = IPPROTO_ICMPV6;
			hdr->hop_limit = 255;
			hdr->saddr = *saddr;
			hdr->daddr = *daddr;

			send_skb->transport_header = send_skb->tail;
			skb_put(send_skb, size);

			icmp6h = (struct icmp6hdr *)skb_transport_header(skb);
			icmp6h->icmp6_type = NDISC_NEIGHBOUR_ADVERTISEMENT;
			icmp6h->icmp6_router = 0;
			icmp6h->icmp6_solicited = 1;
711
			target = (struct in6_addr *)(skb_transport_header(send_skb) + sizeof(struct icmp6hdr));
C
Cong Wang 已提交
712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732
			*target = msg->target;
			icmp6h->icmp6_cksum = csum_ipv6_magic(saddr, daddr, size,
							      IPPROTO_ICMPV6,
							      csum_partial(icmp6h,
									   size, 0));

			if (dev_hard_header(send_skb, skb->dev, ETH_P_IPV6,
					    lladdr, np->dev->dev_addr,
					    send_skb->len) < 0) {
				kfree_skb(send_skb);
				continue;
			}

			netpoll_send_skb(np, send_skb);

			/* If there are several rx_hooks for the same address,
			   we're fine by sending a single reply */
			break;
		}
		spin_unlock_irqrestore(&npinfo->rx_lock, flags);
#endif
733
	}
L
Linus Torvalds 已提交
734 735
}

C
Cong Wang 已提交
736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761
static bool pkt_is_ns(struct sk_buff *skb)
{
	struct nd_msg *msg;
	struct ipv6hdr *hdr;

	if (skb->protocol != htons(ETH_P_ARP))
		return false;
	if (!pskb_may_pull(skb, sizeof(struct ipv6hdr) + sizeof(struct nd_msg)))
		return false;

	msg = (struct nd_msg *)skb_transport_header(skb);
	__skb_push(skb, skb->data - skb_transport_header(skb));
	hdr = ipv6_hdr(skb);

	if (hdr->nexthdr != IPPROTO_ICMPV6)
		return false;
	if (hdr->hop_limit != 255)
		return false;
	if (msg->icmph.icmp6_code != 0)
		return false;
	if (msg->icmph.icmp6_type != NDISC_NEIGHBOUR_SOLICITATION)
		return false;

	return true;
}

762
int __netpoll_rx(struct sk_buff *skb, struct netpoll_info *npinfo)
L
Linus Torvalds 已提交
763 764
{
	int proto, len, ulen;
765
	int hits = 0;
766
	const struct iphdr *iph;
L
Linus Torvalds 已提交
767
	struct udphdr *uh;
768
	struct netpoll *np, *tmp;
769

770
	if (list_empty(&npinfo->rx_np))
L
Linus Torvalds 已提交
771
		goto out;
772

L
Linus Torvalds 已提交
773 774 775
	if (skb->dev->type != ARPHRD_ETHER)
		goto out;

776
	/* check if netpoll clients need ARP */
C
Cong Wang 已提交
777 778 779 780
	if (skb->protocol == htons(ETH_P_ARP) && atomic_read(&trapped)) {
		skb_queue_tail(&npinfo->neigh_tx, skb);
		return 1;
	} else if (pkt_is_ns(skb) && atomic_read(&trapped)) {
C
Cong Wang 已提交
781
		skb_queue_tail(&npinfo->neigh_tx, skb);
L
Linus Torvalds 已提交
782 783 784
		return 1;
	}

785 786 787 788 789 790
	if (skb->protocol == cpu_to_be16(ETH_P_8021Q)) {
		skb = vlan_untag(skb);
		if (unlikely(!skb))
			goto out;
	}

L
Linus Torvalds 已提交
791
	proto = ntohs(eth_hdr(skb)->h_proto);
C
Cong Wang 已提交
792
	if (proto != ETH_P_IP && proto != ETH_P_IPV6)
L
Linus Torvalds 已提交
793 794 795 796 797 798
		goto out;
	if (skb->pkt_type == PACKET_OTHERHOST)
		goto out;
	if (skb_shared(skb))
		goto out;

C
Cong Wang 已提交
799 800 801 802 803 804 805 806 807 808 809
	if (proto == ETH_P_IP) {
		if (!pskb_may_pull(skb, sizeof(struct iphdr)))
			goto out;
		iph = (struct iphdr *)skb->data;
		if (iph->ihl < 5 || iph->version != 4)
			goto out;
		if (!pskb_may_pull(skb, iph->ihl*4))
			goto out;
		iph = (struct iphdr *)skb->data;
		if (ip_fast_csum((u8 *)iph, iph->ihl) != 0)
			goto out;
810

C
Cong Wang 已提交
811 812 813
		len = ntohs(iph->tot_len);
		if (skb->len < len || len < iph->ihl*4)
			goto out;
L
Linus Torvalds 已提交
814

C
Cong Wang 已提交
815 816 817 818 819 820
		/*
		 * Our transport medium may have padded the buffer out.
		 * Now We trim to the true length of the frame.
		 */
		if (pskb_trim_rcsum(skb, len))
			goto out;
L
Linus Torvalds 已提交
821

C
Cong Wang 已提交
822 823 824
		iph = (struct iphdr *)skb->data;
		if (iph->protocol != IPPROTO_UDP)
			goto out;
L
Linus Torvalds 已提交
825

C
Cong Wang 已提交
826 827 828
		len -= iph->ihl*4;
		uh = (struct udphdr *)(((char *)iph) + iph->ihl*4);
		ulen = ntohs(uh->len);
829

C
Cong Wang 已提交
830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846
		if (ulen != len)
			goto out;
		if (checksum_udp(skb, uh, ulen, iph->saddr, iph->daddr))
			goto out;
		list_for_each_entry_safe(np, tmp, &npinfo->rx_np, rx) {
			if (np->local_ip.ip && np->local_ip.ip != iph->daddr)
				continue;
			if (np->remote_ip.ip && np->remote_ip.ip != iph->saddr)
				continue;
			if (np->local_port && np->local_port != ntohs(uh->dest))
				continue;

			np->rx_hook(np, ntohs(uh->source),
				       (char *)(uh+1),
				       ulen - sizeof(struct udphdr));
			hits++;
		}
C
Cong Wang 已提交
847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872
	} else {
#if IS_ENABLED(CONFIG_IPV6)
		const struct ipv6hdr *ip6h;

		if (!pskb_may_pull(skb, sizeof(struct ipv6hdr)))
			goto out;
		ip6h = (struct ipv6hdr *)skb->data;
		if (ip6h->version != 6)
			goto out;
		len = ntohs(ip6h->payload_len);
		if (!len)
			goto out;
		if (len + sizeof(struct ipv6hdr) > skb->len)
			goto out;
		if (pskb_trim_rcsum(skb, len + sizeof(struct ipv6hdr)))
			goto out;
		ip6h = ipv6_hdr(skb);
		if (!pskb_may_pull(skb, sizeof(struct udphdr)))
			goto out;
		uh = udp_hdr(skb);
		ulen = ntohs(uh->len);
		if (ulen != skb->len)
			goto out;
		if (udp6_csum_init(skb, uh, IPPROTO_UDP))
			goto out;
		list_for_each_entry_safe(np, tmp, &npinfo->rx_np, rx) {
873
			if (!ipv6_addr_equal(&np->local_ip.in6, &ip6h->daddr))
C
Cong Wang 已提交
874
				continue;
875
			if (!ipv6_addr_equal(&np->remote_ip.in6, &ip6h->saddr))
C
Cong Wang 已提交
876 877 878 879 880 881 882 883 884 885
				continue;
			if (np->local_port && np->local_port != ntohs(uh->dest))
				continue;

			np->rx_hook(np, ntohs(uh->source),
				       (char *)(uh+1),
				       ulen - sizeof(struct udphdr));
			hits++;
		}
#endif
886 887 888 889
	}

	if (!hits)
		goto out;
L
Linus Torvalds 已提交
890 891 892 893 894 895 896 897 898 899 900 901 902

	kfree_skb(skb);
	return 1;

out:
	if (atomic_read(&trapped)) {
		kfree_skb(skb);
		return 1;
	}

	return 0;
}

903 904
void netpoll_print_options(struct netpoll *np)
{
905
	np_info(np, "local port %d\n", np->local_port);
C
Cong Wang 已提交
906 907 908
	if (np->ipv6)
		np_info(np, "local IPv6 address %pI6c\n", &np->local_ip.in6);
	else
C
Cong Wang 已提交
909
		np_info(np, "local IPv4 address %pI4\n", &np->local_ip.ip);
910 911
	np_info(np, "interface '%s'\n", np->dev_name);
	np_info(np, "remote port %d\n", np->remote_port);
C
Cong Wang 已提交
912 913 914
	if (np->ipv6)
		np_info(np, "remote IPv6 address %pI6c\n", &np->remote_ip.in6);
	else
C
Cong Wang 已提交
915
		np_info(np, "remote IPv4 address %pI4\n", &np->remote_ip.ip);
916
	np_info(np, "remote ethernet address %pM\n", np->remote_mac);
917
}
E
Eric Dumazet 已提交
918
EXPORT_SYMBOL(netpoll_print_options);
919

C
Cong Wang 已提交
920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939
static int netpoll_parse_ip_addr(const char *str, union inet_addr *addr)
{
	const char *end;

	if (!strchr(str, ':') &&
	    in4_pton(str, -1, (void *)addr, -1, &end) > 0) {
		if (!*end)
			return 0;
	}
	if (in6_pton(str, -1, addr->in6.s6_addr, -1, &end) > 0) {
#if IS_ENABLED(CONFIG_IPV6)
		if (!*end)
			return 1;
#else
		return -1;
#endif
	}
	return -1;
}

L
Linus Torvalds 已提交
940 941 942
int netpoll_parse_options(struct netpoll *np, char *opt)
{
	char *cur=opt, *delim;
C
Cong Wang 已提交
943
	int ipv6;
L
Linus Torvalds 已提交
944

945
	if (*cur != '@') {
L
Linus Torvalds 已提交
946 947
		if ((delim = strchr(cur, '@')) == NULL)
			goto parse_failed;
948
		*delim = 0;
949 950
		if (kstrtou16(cur, 10, &np->local_port))
			goto parse_failed;
951
		cur = delim;
L
Linus Torvalds 已提交
952 953 954
	}
	cur++;

955
	if (*cur != '/') {
L
Linus Torvalds 已提交
956 957
		if ((delim = strchr(cur, '/')) == NULL)
			goto parse_failed;
958
		*delim = 0;
C
Cong Wang 已提交
959 960 961 962 963
		ipv6 = netpoll_parse_ip_addr(cur, &np->local_ip);
		if (ipv6 < 0)
			goto parse_failed;
		else
			np->ipv6 = (bool)ipv6;
964
		cur = delim;
L
Linus Torvalds 已提交
965 966 967
	}
	cur++;

968
	if (*cur != ',') {
L
Linus Torvalds 已提交
969 970 971
		/* parse out dev name */
		if ((delim = strchr(cur, ',')) == NULL)
			goto parse_failed;
972
		*delim = 0;
L
Linus Torvalds 已提交
973
		strlcpy(np->dev_name, cur, sizeof(np->dev_name));
974
		cur = delim;
L
Linus Torvalds 已提交
975 976 977
	}
	cur++;

978
	if (*cur != '@') {
L
Linus Torvalds 已提交
979 980 981
		/* dst port */
		if ((delim = strchr(cur, '@')) == NULL)
			goto parse_failed;
982
		*delim = 0;
983
		if (*cur == ' ' || *cur == '\t')
984
			np_info(np, "warning: whitespace is not allowed\n");
985 986
		if (kstrtou16(cur, 10, &np->remote_port))
			goto parse_failed;
987
		cur = delim;
L
Linus Torvalds 已提交
988 989 990 991 992 993
	}
	cur++;

	/* dst ip */
	if ((delim = strchr(cur, '/')) == NULL)
		goto parse_failed;
994
	*delim = 0;
C
Cong Wang 已提交
995 996 997 998 999 1000 1001
	ipv6 = netpoll_parse_ip_addr(cur, &np->remote_ip);
	if (ipv6 < 0)
		goto parse_failed;
	else if (np->ipv6 != (bool)ipv6)
		goto parse_failed;
	else
		np->ipv6 = (bool)ipv6;
1002
	cur = delim + 1;
L
Linus Torvalds 已提交
1003

1004
	if (*cur != 0) {
L
Linus Torvalds 已提交
1005
		/* MAC address */
1006
		if (!mac_pton(cur, np->remote_mac))
L
Linus Torvalds 已提交
1007 1008 1009
			goto parse_failed;
	}

1010
	netpoll_print_options(np);
L
Linus Torvalds 已提交
1011 1012 1013 1014

	return 0;

 parse_failed:
1015
	np_info(np, "couldn't parse config at '%s'!\n", cur);
L
Linus Torvalds 已提交
1016 1017
	return -1;
}
E
Eric Dumazet 已提交
1018
EXPORT_SYMBOL(netpoll_parse_options);
L
Linus Torvalds 已提交
1019

1020
int __netpoll_setup(struct netpoll *np, struct net_device *ndev, gfp_t gfp)
L
Linus Torvalds 已提交
1021
{
1022
	struct netpoll_info *npinfo;
H
Herbert Xu 已提交
1023
	const struct net_device_ops *ops;
1024
	unsigned long flags;
S
Stephen Hemminger 已提交
1025
	int err;
L
Linus Torvalds 已提交
1026

1027 1028
	np->dev = ndev;
	strlcpy(np->dev_name, ndev->name, IFNAMSIZ);
1029
	INIT_WORK(&np->cleanup_work, netpoll_async_cleanup);
1030

1031 1032
	if ((ndev->priv_flags & IFF_DISABLE_NETPOLL) ||
	    !ndev->netdev_ops->ndo_poll_controller) {
1033 1034
		np_err(np, "%s doesn't support polling, aborting\n",
		       np->dev_name);
1035 1036 1037 1038 1039
		err = -ENOTSUPP;
		goto out;
	}

	if (!ndev->npinfo) {
1040
		npinfo = kmalloc(sizeof(*npinfo), gfp);
1041 1042 1043 1044 1045 1046 1047 1048 1049
		if (!npinfo) {
			err = -ENOMEM;
			goto out;
		}

		npinfo->rx_flags = 0;
		INIT_LIST_HEAD(&npinfo->rx_np);

		spin_lock_init(&npinfo->rx_lock);
1050
		sema_init(&npinfo->dev_lock, 1);
C
Cong Wang 已提交
1051
		skb_queue_head_init(&npinfo->neigh_tx);
1052 1053 1054 1055 1056 1057 1058
		skb_queue_head_init(&npinfo->txq);
		INIT_DELAYED_WORK(&npinfo->tx_work, queue_process);

		atomic_set(&npinfo->refcnt, 1);

		ops = np->dev->netdev_ops;
		if (ops->ndo_netpoll_setup) {
1059
			err = ops->ndo_netpoll_setup(ndev, npinfo, gfp);
1060 1061 1062 1063
			if (err)
				goto free_npinfo;
		}
	} else {
N
Neil Horman 已提交
1064
		npinfo = rtnl_dereference(ndev->npinfo);
1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077
		atomic_inc(&npinfo->refcnt);
	}

	npinfo->netpoll = np;

	if (np->rx_hook) {
		spin_lock_irqsave(&npinfo->rx_lock, flags);
		npinfo->rx_flags |= NETPOLL_RX_ENABLED;
		list_add_tail(&np->rx, &npinfo->rx_np);
		spin_unlock_irqrestore(&npinfo->rx_lock, flags);
	}

	/* last thing to do is link it to the net device structure */
1078
	rcu_assign_pointer(ndev->npinfo, npinfo);
1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094

	return 0;

free_npinfo:
	kfree(npinfo);
out:
	return err;
}
EXPORT_SYMBOL_GPL(__netpoll_setup);

int netpoll_setup(struct netpoll *np)
{
	struct net_device *ndev = NULL;
	struct in_device *in_dev;
	int err;

1095
	rtnl_lock();
1096 1097 1098 1099
	if (np->dev_name) {
		struct net *net = current->nsproxy->net_ns;
		ndev = __dev_get_by_name(net, np->dev_name);
	}
L
Linus Torvalds 已提交
1100
	if (!ndev) {
1101
		np_err(np, "%s doesn't exist, aborting\n", np->dev_name);
1102 1103
		err = -ENODEV;
		goto unlock;
L
Linus Torvalds 已提交
1104
	}
1105
	dev_hold(ndev);
L
Linus Torvalds 已提交
1106

1107
	if (netdev_master_upper_dev_get(ndev)) {
1108
		np_err(np, "%s is a slave device, aborting\n", np->dev_name);
1109 1110
		err = -EBUSY;
		goto put;
1111 1112
	}

L
Linus Torvalds 已提交
1113 1114 1115
	if (!netif_running(ndev)) {
		unsigned long atmost, atleast;

1116
		np_info(np, "device %s not up yet, forcing it\n", np->dev_name);
L
Linus Torvalds 已提交
1117

S
Stephen Hemminger 已提交
1118 1119 1120
		err = dev_open(ndev);

		if (err) {
1121
			np_err(np, "failed to open %s\n", ndev->name);
1122
			goto put;
L
Linus Torvalds 已提交
1123 1124
		}

1125
		rtnl_unlock();
L
Linus Torvalds 已提交
1126
		atleast = jiffies + HZ/10;
1127
		atmost = jiffies + carrier_timeout * HZ;
L
Linus Torvalds 已提交
1128 1129
		while (!netif_carrier_ok(ndev)) {
			if (time_after(jiffies, atmost)) {
1130
				np_notice(np, "timeout waiting for carrier\n");
L
Linus Torvalds 已提交
1131 1132
				break;
			}
1133
			msleep(1);
L
Linus Torvalds 已提交
1134 1135 1136 1137 1138 1139 1140 1141
		}

		/* If carrier appears to come up instantly, we don't
		 * trust it and pause so that we don't pump all our
		 * queued console messages into the bitbucket.
		 */

		if (time_before(jiffies, atleast)) {
1142
			np_notice(np, "carrier detect appears untrustworthy, waiting 4 seconds\n");
L
Linus Torvalds 已提交
1143 1144
			msleep(4000);
		}
1145
		rtnl_lock();
L
Linus Torvalds 已提交
1146 1147
	}

C
Cong Wang 已提交
1148 1149
	if (!np->local_ip.ip) {
		if (!np->ipv6) {
1150
			in_dev = __in_dev_get_rtnl(ndev);
C
Cong Wang 已提交
1151 1152 1153 1154 1155 1156 1157 1158 1159 1160

			if (!in_dev || !in_dev->ifa_list) {
				np_err(np, "no IP address for %s, aborting\n",
				       np->dev_name);
				err = -EDESTADDRREQ;
				goto put;
			}

			np->local_ip.ip = in_dev->ifa_list->ifa_local;
			np_info(np, "local IP %pI4\n", &np->local_ip.ip);
C
Cong Wang 已提交
1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 1187 1188
		} else {
#if IS_ENABLED(CONFIG_IPV6)
			struct inet6_dev *idev;

			err = -EDESTADDRREQ;
			idev = __in6_dev_get(ndev);
			if (idev) {
				struct inet6_ifaddr *ifp;

				read_lock_bh(&idev->lock);
				list_for_each_entry(ifp, &idev->addr_list, if_list) {
					if (ipv6_addr_type(&ifp->addr) & IPV6_ADDR_LINKLOCAL)
						continue;
					np->local_ip.in6 = ifp->addr;
					err = 0;
					break;
				}
				read_unlock_bh(&idev->lock);
			}
			if (err) {
				np_err(np, "no IPv6 address for %s, aborting\n",
				       np->dev_name);
				goto put;
			} else
				np_info(np, "local IPv6 %pI6c\n", &np->local_ip.in6);
#else
			np_err(np, "IPv6 is not supported %s, aborting\n",
			       np->dev_name);
1189
			err = -EINVAL;
C
Cong Wang 已提交
1190 1191
			goto put;
#endif
L
Linus Torvalds 已提交
1192 1193 1194
		}
	}

1195 1196 1197
	/* fill up the skb queue */
	refill_skbs();

1198
	err = __netpoll_setup(np, ndev, GFP_KERNEL);
1199 1200 1201
	if (err)
		goto put;

1202
	rtnl_unlock();
L
Linus Torvalds 已提交
1203 1204
	return 0;

1205
put:
L
Linus Torvalds 已提交
1206
	dev_put(ndev);
1207 1208
unlock:
	rtnl_unlock();
S
Stephen Hemminger 已提交
1209
	return err;
L
Linus Torvalds 已提交
1210
}
E
Eric Dumazet 已提交
1211
EXPORT_SYMBOL(netpoll_setup);
L
Linus Torvalds 已提交
1212

1213 1214
static int __init netpoll_init(void)
{
1215 1216 1217 1218 1219
	skb_queue_head_init(&skb_pool);
	return 0;
}
core_initcall(netpoll_init);

1220 1221 1222 1223 1224
static void rcu_cleanup_netpoll_info(struct rcu_head *rcu_head)
{
	struct netpoll_info *npinfo =
			container_of(rcu_head, struct netpoll_info, rcu);

C
Cong Wang 已提交
1225
	skb_queue_purge(&npinfo->neigh_tx);
1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237
	skb_queue_purge(&npinfo->txq);

	/* we can't call cancel_delayed_work_sync here, as we are in softirq */
	cancel_delayed_work(&npinfo->tx_work);

	/* clean after last, unfinished work */
	__skb_queue_purge(&npinfo->txq);
	/* now cancel it again */
	cancel_delayed_work(&npinfo->tx_work);
	kfree(npinfo);
}

1238
void __netpoll_cleanup(struct netpoll *np)
L
Linus Torvalds 已提交
1239
{
1240 1241 1242
	struct netpoll_info *npinfo;
	unsigned long flags;

N
Neil Horman 已提交
1243 1244 1245 1246 1247
	/* rtnl_dereference would be preferable here but
	 * rcu_cleanup_netpoll path can put us in here safely without
	 * holding the rtnl, so plain rcu_dereference it is
	 */
	npinfo = rtnl_dereference(np->dev->npinfo);
1248
	if (!npinfo)
1249
		return;
S
Stephen Hemminger 已提交
1250

1251 1252 1253 1254 1255 1256 1257
	if (!list_empty(&npinfo->rx_np)) {
		spin_lock_irqsave(&npinfo->rx_lock, flags);
		list_del(&np->rx);
		if (list_empty(&npinfo->rx_np))
			npinfo->rx_flags &= ~NETPOLL_RX_ENABLED;
		spin_unlock_irqrestore(&npinfo->rx_lock, flags);
	}
H
Herbert Xu 已提交
1258

1259 1260
	synchronize_srcu(&netpoll_srcu);

1261 1262
	if (atomic_dec_and_test(&npinfo->refcnt)) {
		const struct net_device_ops *ops;
H
Herbert Xu 已提交
1263

1264 1265 1266
		ops = np->dev->netdev_ops;
		if (ops->ndo_netpoll_cleanup)
			ops->ndo_netpoll_cleanup(np->dev);
H
Herbert Xu 已提交
1267

1268
		rcu_assign_pointer(np->dev->npinfo, NULL);
1269 1270 1271 1272
		call_rcu_bh(&npinfo->rcu, rcu_cleanup_netpoll_info);
	}
}
EXPORT_SYMBOL_GPL(__netpoll_cleanup);
H
Herbert Xu 已提交
1273

1274
static void netpoll_async_cleanup(struct work_struct *work)
1275
{
1276
	struct netpoll *np = container_of(work, struct netpoll, cleanup_work);
S
Stephen Hemminger 已提交
1277

1278
	rtnl_lock();
1279
	__netpoll_cleanup(np);
1280
	rtnl_unlock();
1281 1282
	kfree(np);
}
S
Stephen Hemminger 已提交
1283

1284
void __netpoll_free_async(struct netpoll *np)
1285
{
1286
	schedule_work(&np->cleanup_work);
1287
}
1288
EXPORT_SYMBOL_GPL(__netpoll_free_async);
1289

1290 1291 1292 1293
void netpoll_cleanup(struct netpoll *np)
{
	if (!np->dev)
		return;
1294

1295 1296 1297 1298 1299
	rtnl_lock();
	__netpoll_cleanup(np);
	rtnl_unlock();

	dev_put(np->dev);
L
Linus Torvalds 已提交
1300 1301
	np->dev = NULL;
}
E
Eric Dumazet 已提交
1302
EXPORT_SYMBOL(netpoll_cleanup);
L
Linus Torvalds 已提交
1303 1304 1305 1306 1307

int netpoll_trap(void)
{
	return atomic_read(&trapped);
}
E
Eric Dumazet 已提交
1308
EXPORT_SYMBOL(netpoll_trap);
L
Linus Torvalds 已提交
1309 1310 1311 1312 1313 1314 1315 1316 1317

void netpoll_set_trap(int trap)
{
	if (trap)
		atomic_inc(&trapped);
	else
		atomic_dec(&trapped);
}
EXPORT_SYMBOL(netpoll_set_trap);