netpoll.c 19.4 KB
Newer Older
L
Linus Torvalds 已提交
1 2 3 4 5 6 7 8 9 10 11 12 13 14
/*
 * 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.
 */

#include <linux/netdevice.h>
#include <linux/etherdevice.h>
#include <linux/string.h>
15
#include <linux/if_arp.h>
L
Linus Torvalds 已提交
16 17 18 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>
#include <net/tcp.h>
#include <net/udp.h>
#include <asm/unaligned.h>
27
#include <trace/events/napi.h>
L
Linus Torvalds 已提交
28 29 30 31 32 33 34 35 36 37

/*
 * 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
#define MAX_QUEUE_DEPTH (MAX_SKBS / 2)

38
static struct sk_buff_head skb_pool;
L
Linus Torvalds 已提交
39 40 41

static atomic_t trapped;

S
Stephen Hemminger 已提交
42
#define USEC_PER_POLL	50
43 44
#define NETPOLL_RX_ENABLED  1
#define NETPOLL_RX_DROP     2
L
Linus Torvalds 已提交
45 46 47 48 49 50

#define MAX_SKB_SIZE \
		(MAX_UDP_CHUNK + sizeof(struct udphdr) + \
				sizeof(struct iphdr) + sizeof(struct ethhdr))

static void zap_completion_queue(void);
51
static void arp_reply(struct sk_buff *skb);
L
Linus Torvalds 已提交
52

D
David Howells 已提交
53
static void queue_process(struct work_struct *work)
L
Linus Torvalds 已提交
54
{
55 56
	struct netpoll_info *npinfo =
		container_of(work, struct netpoll_info, tx_work.work);
L
Linus Torvalds 已提交
57
	struct sk_buff *skb;
I
Ingo Molnar 已提交
58
	unsigned long flags;
L
Linus Torvalds 已提交
59

S
Stephen Hemminger 已提交
60 61
	while ((skb = skb_dequeue(&npinfo->txq))) {
		struct net_device *dev = skb->dev;
62
		const struct net_device_ops *ops = dev->netdev_ops;
63
		struct netdev_queue *txq;
L
Linus Torvalds 已提交
64

S
Stephen Hemminger 已提交
65 66 67 68
		if (!netif_device_present(dev) || !netif_running(dev)) {
			__kfree_skb(skb);
			continue;
		}
L
Linus Torvalds 已提交
69

70 71
		txq = netdev_get_tx_queue(dev, skb_get_queue_mapping(skb));

I
Ingo Molnar 已提交
72
		local_irq_save(flags);
73 74
		__netif_tx_lock(txq, smp_processor_id());
		if (netif_tx_queue_stopped(txq) ||
75
		    netif_tx_queue_frozen(txq) ||
76
		    ops->ndo_start_xmit(skb, dev) != NETDEV_TX_OK) {
S
Stephen Hemminger 已提交
77
			skb_queue_head(&npinfo->txq, skb);
78
			__netif_tx_unlock(txq);
I
Ingo Molnar 已提交
79
			local_irq_restore(flags);
L
Linus Torvalds 已提交
80

81
			schedule_delayed_work(&npinfo->tx_work, HZ/10);
S
Stephen Hemminger 已提交
82 83
			return;
		}
84
		__netif_tx_unlock(txq);
I
Ingo Molnar 已提交
85
		local_irq_restore(flags);
L
Linus Torvalds 已提交
86 87 88
	}
}

89 90
static __sum16 checksum_udp(struct sk_buff *skb, struct udphdr *uh,
			    unsigned short ulen, __be32 saddr, __be32 daddr)
L
Linus Torvalds 已提交
91
{
92
	__wsum psum;
93

94
	if (uh->check == 0 || skb_csum_unnecessary(skb))
L
Linus Torvalds 已提交
95 96
		return 0;

97 98
	psum = csum_tcpudp_nofold(saddr, daddr, ulen, IPPROTO_UDP, 0);

99
	if (skb->ip_summed == CHECKSUM_COMPLETE &&
100
	    !csum_fold(csum_add(psum, skb->csum)))
101
		return 0;
L
Linus Torvalds 已提交
102

103
	skb->csum = psum;
L
Linus Torvalds 已提交
104

105
	return __skb_checksum_complete(skb);
L
Linus Torvalds 已提交
106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123
}

/*
 * 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.
 */
124 125 126 127 128 129 130 131 132 133 134 135
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;

136
	npinfo->rx_flags |= NETPOLL_RX_DROP;
137
	atomic_inc(&trapped);
138
	set_bit(NAPI_STATE_NPSVC, &napi->state);
139 140

	work = napi->poll(napi, budget);
141
	trace_napi_poll(napi);
142

143
	clear_bit(NAPI_STATE_NPSVC, &napi->state);
144
	atomic_dec(&trapped);
145
	npinfo->rx_flags &= ~NETPOLL_RX_DROP;
146 147 148 149

	return budget - work;
}

150
static void poll_napi(struct net_device *dev)
L
Linus Torvalds 已提交
151
{
152
	struct napi_struct *napi;
L
Linus Torvalds 已提交
153 154
	int budget = 16;

155
	list_for_each_entry(napi, &dev->napi_list, dev_list) {
156
		if (napi->poll_owner != smp_processor_id() &&
157
		    spin_trylock(&napi->poll_lock)) {
158
			budget = poll_one_napi(dev->npinfo, napi, budget);
159
			spin_unlock(&napi->poll_lock);
160 161 162

			if (!budget)
				break;
163
		}
L
Linus Torvalds 已提交
164 165 166
	}
}

167 168
static void service_arp_queue(struct netpoll_info *npi)
{
169 170
	if (npi) {
		struct sk_buff *skb;
171

172 173
		while ((skb = skb_dequeue(&npi->arp_tx)))
			arp_reply(skb);
174 175 176
	}
}

L
Linus Torvalds 已提交
177 178
void netpoll_poll(struct netpoll *np)
{
179
	struct net_device *dev = np->dev;
180
	const struct net_device_ops *ops;
181

182 183 184 185 186
	if (!dev || !netif_running(dev))
		return;

	ops = dev->netdev_ops;
	if (!ops->ndo_poll_controller)
L
Linus Torvalds 已提交
187 188 189
		return;

	/* Process pending work on NIC */
190
	ops->ndo_poll_controller(dev);
191 192

	poll_napi(dev);
L
Linus Torvalds 已提交
193

194
	service_arp_queue(dev->npinfo);
195

L
Linus Torvalds 已提交
196 197 198 199 200 201 202 203
	zap_completion_queue();
}

static void refill_skbs(void)
{
	struct sk_buff *skb;
	unsigned long flags;

204 205
	spin_lock_irqsave(&skb_pool.lock, flags);
	while (skb_pool.qlen < MAX_SKBS) {
L
Linus Torvalds 已提交
206 207 208 209
		skb = alloc_skb(MAX_SKB_SIZE, GFP_ATOMIC);
		if (!skb)
			break;

210
		__skb_queue_tail(&skb_pool, skb);
L
Linus Torvalds 已提交
211
	}
212
	spin_unlock_irqrestore(&skb_pool.lock, flags);
L
Linus Torvalds 已提交
213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230
}

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;
231 232
			if (skb->destructor) {
				atomic_inc(&skb->users);
L
Linus Torvalds 已提交
233
				dev_kfree_skb_any(skb); /* put this one back */
234
			} else {
L
Linus Torvalds 已提交
235
				__kfree_skb(skb);
236
			}
L
Linus Torvalds 已提交
237 238 239 240 241 242
		}
	}

	put_cpu_var(softnet_data);
}

243
static struct sk_buff *find_skb(struct netpoll *np, int len, int reserve)
L
Linus Torvalds 已提交
244
{
245 246
	int count = 0;
	struct sk_buff *skb;
L
Linus Torvalds 已提交
247 248

	zap_completion_queue();
249
	refill_skbs();
L
Linus Torvalds 已提交
250 251 252
repeat:

	skb = alloc_skb(len, GFP_ATOMIC);
253 254
	if (!skb)
		skb = skb_dequeue(&skb_pool);
L
Linus Torvalds 已提交
255 256

	if (!skb) {
257 258 259
		if (++count < 10) {
			netpoll_poll(np);
			goto repeat;
L
Linus Torvalds 已提交
260
		}
261
		return NULL;
L
Linus Torvalds 已提交
262 263 264 265 266 267 268
	}

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

269 270 271 272 273 274 275 276 277 278 279
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;
}

L
Linus Torvalds 已提交
280 281
static void netpoll_send_skb(struct netpoll *np, struct sk_buff *skb)
{
S
Stephen Hemminger 已提交
282 283
	int status = NETDEV_TX_BUSY;
	unsigned long tries;
284
	struct net_device *dev = np->dev;
285
	const struct net_device_ops *ops = dev->netdev_ops;
286
	struct netpoll_info *npinfo = np->dev->npinfo;
S
Stephen Hemminger 已提交
287

288 289 290 291
	if (!npinfo || !netif_running(dev) || !netif_device_present(dev)) {
		__kfree_skb(skb);
		return;
	}
S
Stephen Hemminger 已提交
292 293

	/* don't get messages out of order, and no recursion */
294
	if (skb_queue_len(&npinfo->txq) == 0 && !netpoll_owner_active(dev)) {
295
		struct netdev_queue *txq;
296 297
		unsigned long flags;

298 299
		txq = netdev_get_tx_queue(dev, skb_get_queue_mapping(skb));

300
		local_irq_save(flags);
301 302 303
		/* try until next clock tick */
		for (tries = jiffies_to_usecs(1)/USEC_PER_POLL;
		     tries > 0; --tries) {
304
			if (__netif_tx_trylock(txq)) {
E
Eric Dumazet 已提交
305
				if (!netif_tx_queue_stopped(txq)) {
306
					status = ops->ndo_start_xmit(skb, dev);
E
Eric Dumazet 已提交
307 308 309
					if (status == NETDEV_TX_OK)
						txq_trans_update(txq);
				}
310
				__netif_tx_unlock(txq);
311 312 313 314 315

				if (status == NETDEV_TX_OK)
					break;

			}
316 317 318 319 320

			/* tickle device maybe there is some cleanup */
			netpoll_poll(np);

			udelay(USEC_PER_POLL);
M
Matt Mackall 已提交
321
		}
322
		local_irq_restore(flags);
L
Linus Torvalds 已提交
323 324
	}

S
Stephen Hemminger 已提交
325
	if (status != NETDEV_TX_OK) {
S
Stephen Hemminger 已提交
326
		skb_queue_tail(&npinfo->txq, skb);
327
		schedule_delayed_work(&npinfo->tx_work,0);
L
Linus Torvalds 已提交
328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346
	}
}

void netpoll_send_udp(struct netpoll *np, const char *msg, int len)
{
	int total_len, eth_len, ip_len, udp_len;
	struct sk_buff *skb;
	struct udphdr *udph;
	struct iphdr *iph;
	struct ethhdr *eth;

	udp_len = len + sizeof(*udph);
	ip_len = eth_len = udp_len + sizeof(*iph);
	total_len = eth_len + ETH_HLEN + NET_IP_ALIGN;

	skb = find_skb(np, total_len, total_len - len);
	if (!skb)
		return;

347
	skb_copy_to_linear_data(skb, msg, len);
L
Linus Torvalds 已提交
348 349
	skb->len += len;

350 351 352
	skb_push(skb, sizeof(*udph));
	skb_reset_transport_header(skb);
	udph = udp_hdr(skb);
L
Linus Torvalds 已提交
353 354 355 356
	udph->source = htons(np->local_port);
	udph->dest = htons(np->remote_port);
	udph->len = htons(udp_len);
	udph->check = 0;
357 358
	udph->check = csum_tcpudp_magic(np->local_ip,
					np->remote_ip,
359
					udp_len, IPPROTO_UDP,
360
					csum_partial(udph, udp_len, 0));
361
	if (udph->check == 0)
362
		udph->check = CSUM_MANGLED_0;
L
Linus Torvalds 已提交
363

364 365
	skb_push(skb, sizeof(*iph));
	skb_reset_network_header(skb);
366
	iph = ip_hdr(skb);
L
Linus Torvalds 已提交
367 368 369 370 371 372 373 374 375 376

	/* 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       = 0;
	iph->frag_off = 0;
	iph->ttl      = 64;
	iph->protocol = IPPROTO_UDP;
	iph->check    = 0;
377 378
	put_unaligned(np->local_ip, &(iph->saddr));
	put_unaligned(np->remote_ip, &(iph->daddr));
L
Linus Torvalds 已提交
379 380 381
	iph->check    = ip_fast_csum((unsigned char *)iph, iph->ihl);

	eth = (struct ethhdr *) skb_push(skb, ETH_HLEN);
382
	skb_reset_mac_header(skb);
383
	skb->protocol = eth->h_proto = htons(ETH_P_IP);
384 385
	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
386 387 388 389 390 391 392 393

	skb->dev = np->dev;

	netpoll_send_skb(np, skb);
}

static void arp_reply(struct sk_buff *skb)
{
394
	struct netpoll_info *npinfo = skb->dev->npinfo;
L
Linus Torvalds 已提交
395 396 397
	struct arphdr *arp;
	unsigned char *arp_ptr;
	int size, type = ARPOP_REPLY, ptype = ETH_P_ARP;
A
Al Viro 已提交
398
	__be32 sip, tip;
399
	unsigned char *sha;
L
Linus Torvalds 已提交
400
	struct sk_buff *send_skb;
401
	struct netpoll *np = NULL;
L
Linus Torvalds 已提交
402

403 404
	if (npinfo->rx_np && npinfo->rx_np->dev == skb->dev)
		np = npinfo->rx_np;
405 406
	if (!np)
		return;
L
Linus Torvalds 已提交
407 408 409 410 411

	/* No arp on this interface */
	if (skb->dev->flags & IFF_NOARP)
		return;

412
	if (!pskb_may_pull(skb, arp_hdr_len(skb->dev)))
L
Linus Torvalds 已提交
413 414
		return;

415
	skb_reset_network_header(skb);
416
	skb_reset_transport_header(skb);
417
	arp = arp_hdr(skb);
L
Linus Torvalds 已提交
418 419 420 421 422 423 424

	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;

425 426 427 428
	arp_ptr = (unsigned char *)(arp+1);
	/* save the location of the src hw addr */
	sha = arp_ptr;
	arp_ptr += skb->dev->addr_len;
L
Linus Torvalds 已提交
429
	memcpy(&sip, arp_ptr, 4);
430 431 432
	arp_ptr += 4;
	/* if we actually cared about dst hw addr, it would get copied here */
	arp_ptr += skb->dev->addr_len;
L
Linus Torvalds 已提交
433 434 435
	memcpy(&tip, arp_ptr, 4);

	/* Should we ignore arp? */
436
	if (tip != np->local_ip ||
437
	    ipv4_is_loopback(tip) || ipv4_is_multicast(tip))
L
Linus Torvalds 已提交
438 439
		return;

440
	size = arp_hdr_len(skb->dev);
441
	send_skb = find_skb(np, size + LL_ALLOCATED_SPACE(np->dev),
L
Linus Torvalds 已提交
442 443 444 445 446
			    LL_RESERVED_SPACE(np->dev));

	if (!send_skb)
		return;

447
	skb_reset_network_header(send_skb);
L
Linus Torvalds 已提交
448 449 450 451 452
	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 */
453
	if (dev_hard_header(send_skb, skb->dev, ptype,
454
			    sha, np->dev->dev_addr,
455
			    send_skb->len) < 0) {
L
Linus Torvalds 已提交
456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477
		kfree_skb(send_skb);
		return;
	}

	/*
	 * 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;
478
	memcpy(arp_ptr, sha, np->dev->addr_len);
L
Linus Torvalds 已提交
479 480 481 482 483 484 485 486 487 488 489
	arp_ptr += np->dev->addr_len;
	memcpy(arp_ptr, &sip, 4);

	netpoll_send_skb(np, send_skb);
}

int __netpoll_rx(struct sk_buff *skb)
{
	int proto, len, ulen;
	struct iphdr *iph;
	struct udphdr *uh;
490 491 492
	struct netpoll_info *npi = skb->dev->npinfo;
	struct netpoll *np = npi->rx_np;

493
	if (!np)
L
Linus Torvalds 已提交
494 495 496 497
		goto out;
	if (skb->dev->type != ARPHRD_ETHER)
		goto out;

498
	/* check if netpoll clients need ARP */
499
	if (skb->protocol == htons(ETH_P_ARP) &&
L
Linus Torvalds 已提交
500
	    atomic_read(&trapped)) {
501
		skb_queue_tail(&npi->arp_tx, skb);
L
Linus Torvalds 已提交
502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526
		return 1;
	}

	proto = ntohs(eth_hdr(skb)->h_proto);
	if (proto != ETH_P_IP)
		goto out;
	if (skb->pkt_type == PACKET_OTHERHOST)
		goto out;
	if (skb_shared(skb))
		goto out;

	iph = (struct iphdr *)skb->data;
	if (!pskb_may_pull(skb, sizeof(struct iphdr)))
		goto out;
	if (iph->ihl < 5 || iph->version != 4)
		goto out;
	if (!pskb_may_pull(skb, iph->ihl*4))
		goto out;
	if (ip_fast_csum((u8 *)iph, iph->ihl) != 0)
		goto out;

	len = ntohs(iph->tot_len);
	if (skb->len < len || len < iph->ihl*4)
		goto out;

527 528 529 530 531 532 533
	/*
	 * 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
534 535 536 537 538 539 540 541 542
	if (iph->protocol != IPPROTO_UDP)
		goto out;

	len -= iph->ihl*4;
	uh = (struct udphdr *)(((char *)iph) + iph->ihl*4);
	ulen = ntohs(uh->len);

	if (ulen != len)
		goto out;
543
	if (checksum_udp(skb, uh, ulen, iph->saddr, iph->daddr))
L
Linus Torvalds 已提交
544
		goto out;
545
	if (np->local_ip && np->local_ip != iph->daddr)
L
Linus Torvalds 已提交
546
		goto out;
547
	if (np->remote_ip && np->remote_ip != iph->saddr)
L
Linus Torvalds 已提交
548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567
		goto out;
	if (np->local_port && np->local_port != ntohs(uh->dest))
		goto out;

	np->rx_hook(np, ntohs(uh->source),
		    (char *)(uh+1),
		    ulen - sizeof(struct udphdr));

	kfree_skb(skb);
	return 1;

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

	return 0;
}

568 569 570 571
void netpoll_print_options(struct netpoll *np)
{
	printk(KERN_INFO "%s: local port %d\n",
			 np->name, np->local_port);
572 573
	printk(KERN_INFO "%s: local IP %pI4\n",
			 np->name, &np->local_ip);
574 575 576 577
	printk(KERN_INFO "%s: interface %s\n",
			 np->name, np->dev_name);
	printk(KERN_INFO "%s: remote port %d\n",
			 np->name, np->remote_port);
578 579
	printk(KERN_INFO "%s: remote IP %pI4\n",
			 np->name, &np->remote_ip);
J
Johannes Berg 已提交
580 581
	printk(KERN_INFO "%s: remote ethernet address %pM\n",
	                 np->name, np->remote_mac);
582 583
}

L
Linus Torvalds 已提交
584 585 586 587
int netpoll_parse_options(struct netpoll *np, char *opt)
{
	char *cur=opt, *delim;

588
	if (*cur != '@') {
L
Linus Torvalds 已提交
589 590
		if ((delim = strchr(cur, '@')) == NULL)
			goto parse_failed;
591 592 593
		*delim = 0;
		np->local_port = simple_strtol(cur, NULL, 10);
		cur = delim;
L
Linus Torvalds 已提交
594 595 596
	}
	cur++;

597
	if (*cur != '/') {
L
Linus Torvalds 已提交
598 599
		if ((delim = strchr(cur, '/')) == NULL)
			goto parse_failed;
600
		*delim = 0;
601
		np->local_ip = in_aton(cur);
602
		cur = delim;
L
Linus Torvalds 已提交
603 604 605
	}
	cur++;

606
	if (*cur != ',') {
L
Linus Torvalds 已提交
607 608 609
		/* parse out dev name */
		if ((delim = strchr(cur, ',')) == NULL)
			goto parse_failed;
610
		*delim = 0;
L
Linus Torvalds 已提交
611
		strlcpy(np->dev_name, cur, sizeof(np->dev_name));
612
		cur = delim;
L
Linus Torvalds 已提交
613 614 615
	}
	cur++;

616
	if (*cur != '@') {
L
Linus Torvalds 已提交
617 618 619
		/* dst port */
		if ((delim = strchr(cur, '@')) == NULL)
			goto parse_failed;
620 621 622
		*delim = 0;
		np->remote_port = simple_strtol(cur, NULL, 10);
		cur = delim;
L
Linus Torvalds 已提交
623 624 625 626 627 628
	}
	cur++;

	/* dst ip */
	if ((delim = strchr(cur, '/')) == NULL)
		goto parse_failed;
629
	*delim = 0;
630
	np->remote_ip = in_aton(cur);
631
	cur = delim + 1;
L
Linus Torvalds 已提交
632

633
	if (*cur != 0) {
L
Linus Torvalds 已提交
634 635 636
		/* MAC address */
		if ((delim = strchr(cur, ':')) == NULL)
			goto parse_failed;
637 638 639
		*delim = 0;
		np->remote_mac[0] = simple_strtol(cur, NULL, 16);
		cur = delim + 1;
L
Linus Torvalds 已提交
640 641
		if ((delim = strchr(cur, ':')) == NULL)
			goto parse_failed;
642 643 644
		*delim = 0;
		np->remote_mac[1] = simple_strtol(cur, NULL, 16);
		cur = delim + 1;
L
Linus Torvalds 已提交
645 646
		if ((delim = strchr(cur, ':')) == NULL)
			goto parse_failed;
647 648 649
		*delim = 0;
		np->remote_mac[2] = simple_strtol(cur, NULL, 16);
		cur = delim + 1;
L
Linus Torvalds 已提交
650 651
		if ((delim = strchr(cur, ':')) == NULL)
			goto parse_failed;
652 653 654
		*delim = 0;
		np->remote_mac[3] = simple_strtol(cur, NULL, 16);
		cur = delim + 1;
L
Linus Torvalds 已提交
655 656
		if ((delim = strchr(cur, ':')) == NULL)
			goto parse_failed;
657 658 659 660
		*delim = 0;
		np->remote_mac[4] = simple_strtol(cur, NULL, 16);
		cur = delim + 1;
		np->remote_mac[5] = simple_strtol(cur, NULL, 16);
L
Linus Torvalds 已提交
661 662
	}

663
	netpoll_print_options(np);
L
Linus Torvalds 已提交
664 665 666 667 668 669 670 671 672 673 674 675 676

	return 0;

 parse_failed:
	printk(KERN_INFO "%s: couldn't parse config at %s!\n",
	       np->name, cur);
	return -1;
}

int netpoll_setup(struct netpoll *np)
{
	struct net_device *ndev = NULL;
	struct in_device *in_dev;
677
	struct netpoll_info *npinfo;
678
	unsigned long flags;
S
Stephen Hemminger 已提交
679
	int err;
L
Linus Torvalds 已提交
680 681

	if (np->dev_name)
682
		ndev = dev_get_by_name(&init_net, np->dev_name);
L
Linus Torvalds 已提交
683 684 685
	if (!ndev) {
		printk(KERN_ERR "%s: %s doesn't exist, aborting.\n",
		       np->name, np->dev_name);
S
Stephen Hemminger 已提交
686
		return -ENODEV;
L
Linus Torvalds 已提交
687 688 689
	}

	np->dev = ndev;
690 691
	if (!ndev->npinfo) {
		npinfo = kmalloc(sizeof(*npinfo), GFP_KERNEL);
S
Stephen Hemminger 已提交
692 693
		if (!npinfo) {
			err = -ENOMEM;
694
			goto release;
S
Stephen Hemminger 已提交
695
		}
696

697
		npinfo->rx_flags = 0;
698
		npinfo->rx_np = NULL;
S
Stephen Hemminger 已提交
699

700
		spin_lock_init(&npinfo->rx_lock);
701
		skb_queue_head_init(&npinfo->arp_tx);
S
Stephen Hemminger 已提交
702
		skb_queue_head_init(&npinfo->txq);
703
		INIT_DELAYED_WORK(&npinfo->tx_work, queue_process);
S
Stephen Hemminger 已提交
704

S
Stephen Hemminger 已提交
705 706
		atomic_set(&npinfo->refcnt, 1);
	} else {
707
		npinfo = ndev->npinfo;
S
Stephen Hemminger 已提交
708 709
		atomic_inc(&npinfo->refcnt);
	}
L
Linus Torvalds 已提交
710

711
	if (!ndev->netdev_ops->ndo_poll_controller) {
L
Linus Torvalds 已提交
712 713
		printk(KERN_ERR "%s: %s doesn't support polling, aborting.\n",
		       np->name, np->dev_name);
S
Stephen Hemminger 已提交
714
		err = -ENOTSUPP;
L
Linus Torvalds 已提交
715 716 717 718 719 720 721 722 723
		goto release;
	}

	if (!netif_running(ndev)) {
		unsigned long atmost, atleast;

		printk(KERN_INFO "%s: device %s not up yet, forcing it\n",
		       np->name, np->dev_name);

724
		rtnl_lock();
S
Stephen Hemminger 已提交
725 726 727 728
		err = dev_open(ndev);
		rtnl_unlock();

		if (err) {
L
Linus Torvalds 已提交
729
			printk(KERN_ERR "%s: failed to open %s\n",
S
Stephen Hemminger 已提交
730
			       np->name, ndev->name);
L
Linus Torvalds 已提交
731 732 733 734
			goto release;
		}

		atleast = jiffies + HZ/10;
735
		atmost = jiffies + 4*HZ;
L
Linus Torvalds 已提交
736 737 738 739 740 741 742
		while (!netif_carrier_ok(ndev)) {
			if (time_after(jiffies, atmost)) {
				printk(KERN_NOTICE
				       "%s: timeout waiting for carrier\n",
				       np->name);
				break;
			}
743
			msleep(1);
L
Linus Torvalds 已提交
744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760
		}

		/* 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)) {
			printk(KERN_NOTICE "%s: carrier detect appears"
			       " untrustworthy, waiting 4 seconds\n",
			       np->name);
			msleep(4000);
		}
	}

	if (!np->local_ip) {
		rcu_read_lock();
761
		in_dev = __in_dev_get_rcu(ndev);
L
Linus Torvalds 已提交
762 763 764 765 766

		if (!in_dev || !in_dev->ifa_list) {
			rcu_read_unlock();
			printk(KERN_ERR "%s: no IP address for %s, aborting\n",
			       np->name, np->dev_name);
S
Stephen Hemminger 已提交
767
			err = -EDESTADDRREQ;
L
Linus Torvalds 已提交
768 769 770
			goto release;
		}

771
		np->local_ip = in_dev->ifa_list->ifa_local;
L
Linus Torvalds 已提交
772
		rcu_read_unlock();
773
		printk(KERN_INFO "%s: local IP %pI4\n", np->name, &np->local_ip);
L
Linus Torvalds 已提交
774 775
	}

776 777
	if (np->rx_hook) {
		spin_lock_irqsave(&npinfo->rx_lock, flags);
778
		npinfo->rx_flags |= NETPOLL_RX_ENABLED;
779 780 781
		npinfo->rx_np = np;
		spin_unlock_irqrestore(&npinfo->rx_lock, flags);
	}
I
Ingo Molnar 已提交
782 783 784 785

	/* fill up the skb queue */
	refill_skbs();

786
	/* last thing to do is link it to the net device structure */
787
	ndev->npinfo = npinfo;
L
Linus Torvalds 已提交
788

789 790 791
	/* avoid racing with NAPI reading npinfo */
	synchronize_rcu();

L
Linus Torvalds 已提交
792 793 794
	return 0;

 release:
795 796
	if (!ndev->npinfo)
		kfree(npinfo);
L
Linus Torvalds 已提交
797 798
	np->dev = NULL;
	dev_put(ndev);
S
Stephen Hemminger 已提交
799
	return err;
L
Linus Torvalds 已提交
800 801
}

802 803
static int __init netpoll_init(void)
{
804 805 806 807 808
	skb_queue_head_init(&skb_pool);
	return 0;
}
core_initcall(netpoll_init);

L
Linus Torvalds 已提交
809 810
void netpoll_cleanup(struct netpoll *np)
{
811 812 813
	struct netpoll_info *npinfo;
	unsigned long flags;

814
	if (np->dev) {
815
		npinfo = np->dev->npinfo;
S
Stephen Hemminger 已提交
816 817 818 819
		if (npinfo) {
			if (npinfo->rx_np == np) {
				spin_lock_irqsave(&npinfo->rx_lock, flags);
				npinfo->rx_np = NULL;
820
				npinfo->rx_flags &= ~NETPOLL_RX_ENABLED;
S
Stephen Hemminger 已提交
821 822 823 824 825
				spin_unlock_irqrestore(&npinfo->rx_lock, flags);
			}

			if (atomic_dec_and_test(&npinfo->refcnt)) {
				skb_queue_purge(&npinfo->arp_tx);
826
				skb_queue_purge(&npinfo->txq);
827
				cancel_rearming_delayed_work(&npinfo->tx_work);
S
Stephen Hemminger 已提交
828

829
				/* clean after last, unfinished work */
830
				__skb_queue_purge(&npinfo->txq);
S
Stephen Hemminger 已提交
831
				kfree(npinfo);
832
				np->dev->npinfo = NULL;
S
Stephen Hemminger 已提交
833
			}
834
		}
S
Stephen Hemminger 已提交
835

836 837
		dev_put(np->dev);
	}
838

L
Linus Torvalds 已提交
839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856
	np->dev = NULL;
}

int netpoll_trap(void)
{
	return atomic_read(&trapped);
}

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

EXPORT_SYMBOL(netpoll_set_trap);
EXPORT_SYMBOL(netpoll_trap);
857
EXPORT_SYMBOL(netpoll_print_options);
L
Linus Torvalds 已提交
858 859 860 861 862
EXPORT_SYMBOL(netpoll_parse_options);
EXPORT_SYMBOL(netpoll_setup);
EXPORT_SYMBOL(netpoll_cleanup);
EXPORT_SYMBOL(netpoll_send_udp);
EXPORT_SYMBOL(netpoll_poll);