virtio_net.c 33.2 KB
Newer Older
1
/* A network driver using virtio.
R
Rusty Russell 已提交
2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21
 *
 * Copyright 2007 Rusty Russell <rusty@rustcorp.com.au> IBM Corporation
 *
 * 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.
 *
 * This program is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 * GNU General Public License for more details.
 *
 * You should have received a copy of the GNU General Public License
 * along with this program; if not, write to the Free Software
 * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
 */
//#define DEBUG
#include <linux/netdevice.h>
#include <linux/etherdevice.h>
22
#include <linux/ethtool.h>
R
Rusty Russell 已提交
23 24 25 26
#include <linux/module.h>
#include <linux/virtio.h>
#include <linux/virtio_net.h>
#include <linux/scatterlist.h>
27
#include <linux/if_vlan.h>
28
#include <linux/slab.h>
R
Rusty Russell 已提交
29

30 31 32
static int napi_weight = 128;
module_param(napi_weight, int, 0444);

33
static bool csum = true, gso = true;
R
Rusty Russell 已提交
34 35 36
module_param(csum, bool, 0444);
module_param(gso, bool, 0444);

R
Rusty Russell 已提交
37
/* FIXME: MTU in config. */
38
#define MAX_PACKET_LEN (ETH_HLEN + VLAN_HLEN + ETH_DATA_LEN)
39
#define GOOD_COPY_LEN	128
R
Rusty Russell 已提交
40

41
#define VIRTNET_SEND_COMMAND_SG_MAX    2
42
#define VIRTNET_DRIVER_VERSION "1.0.0"
43

44
struct virtnet_stats {
45 46
	struct u64_stats_sync tx_syncp;
	struct u64_stats_sync rx_syncp;
47 48 49 50 51 52 53
	u64 tx_bytes;
	u64 tx_packets;

	u64 rx_bytes;
	u64 rx_packets;
};

54
struct virtnet_info {
R
Rusty Russell 已提交
55
	struct virtio_device *vdev;
56
	struct virtqueue *rvq, *svq, *cvq;
R
Rusty Russell 已提交
57 58
	struct net_device *dev;
	struct napi_struct napi;
59
	unsigned int status;
R
Rusty Russell 已提交
60 61 62 63

	/* Number of input buffers, and max we've ever had. */
	unsigned int num, max;

64 65 66
	/* I like... big packets and I cannot lie! */
	bool big_packets;

67 68 69
	/* Host will merge rx buffers for big packets (shake it! shake it!) */
	bool mergeable_rx_bufs;

70 71 72
	/* enable config space updates */
	bool config_enable;

73 74 75
	/* Active statistics */
	struct virtnet_stats __percpu *stats;

76 77 78
	/* Work struct for refilling if we run low on memory. */
	struct delayed_work refill;

79 80 81 82 83 84
	/* Work struct for config space updates */
	struct work_struct config_work;

	/* Lock for config space updates */
	struct mutex config_lock;

85 86
	/* Chain pages by the private ptr. */
	struct page *pages;
87 88 89 90

	/* fragments + linear part + virtio header */
	struct scatterlist rx_sg[MAX_SKB_FRAGS + 2];
	struct scatterlist tx_sg[MAX_SKB_FRAGS + 2];
R
Rusty Russell 已提交
91 92
};

93 94 95 96 97
struct skb_vnet_hdr {
	union {
		struct virtio_net_hdr hdr;
		struct virtio_net_hdr_mrg_rxbuf mhdr;
	};
98
	unsigned int num_sg;
99 100
};

101 102 103 104 105 106 107 108 109 110
struct padded_vnet_hdr {
	struct virtio_net_hdr hdr;
	/*
	 * virtio_net_hdr should be in a separated sg buffer because of a
	 * QEMU bug, and data sg buffer shares same page with this header sg.
	 * This padding makes next sg 16 byte aligned after virtio_net_hdr.
	 */
	char padding[6];
};

111
static inline struct skb_vnet_hdr *skb_vnet_hdr(struct sk_buff *skb)
R
Rusty Russell 已提交
112
{
113
	return (struct skb_vnet_hdr *)skb->cb;
R
Rusty Russell 已提交
114 115
}

116 117 118 119 120
/*
 * private is used to chain pages for big packets, put the whole
 * most recent used list in the beginning for reuse
 */
static void give_pages(struct virtnet_info *vi, struct page *page)
121
{
122
	struct page *end;
123

124 125 126 127
	/* Find end of list, sew whole thing into vi->pages. */
	for (end = page; end->private; end = (struct page *)end->private);
	end->private = (unsigned long)vi->pages;
	vi->pages = page;
128 129
}

130 131 132 133
static struct page *get_a_page(struct virtnet_info *vi, gfp_t gfp_mask)
{
	struct page *p = vi->pages;

134
	if (p) {
135
		vi->pages = (struct page *)p->private;
136 137 138
		/* clear private here, it is used to chain pages */
		p->private = 0;
	} else
139 140 141 142
		p = alloc_page(gfp_mask);
	return p;
}

143
static void skb_xmit_done(struct virtqueue *svq)
R
Rusty Russell 已提交
144
{
145
	struct virtnet_info *vi = svq->vdev->priv;
R
Rusty Russell 已提交
146

147
	/* Suppress further interrupts. */
148
	virtqueue_disable_cb(svq);
149

150
	/* We were probably waiting for more output buffers. */
R
Rusty Russell 已提交
151 152 153
	netif_wake_queue(vi->dev);
}

154 155
static void set_skb_frag(struct sk_buff *skb, struct page *page,
			 unsigned int offset, unsigned int *len)
R
Rusty Russell 已提交
156
{
157
	int size = min((unsigned)PAGE_SIZE - offset, *len);
158 159
	int i = skb_shinfo(skb)->nr_frags;

160
	__skb_fill_page_desc(skb, i, page, offset, size);
161

162 163
	skb->data_len += size;
	skb->len += size;
164
	skb->truesize += PAGE_SIZE;
165
	skb_shinfo(skb)->nr_frags++;
166
	*len -= size;
167
}
168

169
/* Called from bottom half context */
170 171 172 173 174 175 176
static struct sk_buff *page_to_skb(struct virtnet_info *vi,
				   struct page *page, unsigned int len)
{
	struct sk_buff *skb;
	struct skb_vnet_hdr *hdr;
	unsigned int copy, hdr_len, offset;
	char *p;
177

178
	p = page_address(page);
179

180 181 182 183
	/* copy small packet so we can reuse these pages for small data */
	skb = netdev_alloc_skb_ip_align(vi->dev, GOOD_COPY_LEN);
	if (unlikely(!skb))
		return NULL;
184

185
	hdr = skb_vnet_hdr(skb);
186

187 188 189 190 191 192 193
	if (vi->mergeable_rx_bufs) {
		hdr_len = sizeof hdr->mhdr;
		offset = hdr_len;
	} else {
		hdr_len = sizeof hdr->hdr;
		offset = sizeof(struct padded_vnet_hdr);
	}
194

195
	memcpy(hdr, p, hdr_len);
196

197 198
	len -= hdr_len;
	p += offset;
199

200 201 202 203
	copy = len;
	if (copy > skb_tailroom(skb))
		copy = skb_tailroom(skb);
	memcpy(skb_put(skb, copy), p, copy);
204

205 206
	len -= copy;
	offset += copy;
207

208 209 210 211 212 213 214 215 216 217 218 219 220
	/*
	 * Verify that we can indeed put this data into a skb.
	 * This is here to handle cases when the device erroneously
	 * tries to receive more than is possible. This is usually
	 * the case of a broken device.
	 */
	if (unlikely(len > MAX_SKB_FRAGS * PAGE_SIZE)) {
		if (net_ratelimit())
			pr_debug("%s: too much data\n", skb->dev->name);
		dev_kfree_skb(skb);
		return NULL;
	}

221 222 223 224 225
	while (len) {
		set_skb_frag(skb, page, offset, &len);
		page = (struct page *)page->private;
		offset = 0;
	}
226

227 228
	if (page)
		give_pages(vi, page);
229

230 231
	return skb;
}
232

233 234 235 236 237 238 239 240 241 242 243 244 245 246
static int receive_mergeable(struct virtnet_info *vi, struct sk_buff *skb)
{
	struct skb_vnet_hdr *hdr = skb_vnet_hdr(skb);
	struct page *page;
	int num_buf, i, len;

	num_buf = hdr->mhdr.num_buffers;
	while (--num_buf) {
		i = skb_shinfo(skb)->nr_frags;
		if (i >= MAX_SKB_FRAGS) {
			pr_debug("%s: packet too long\n", skb->dev->name);
			skb->dev->stats.rx_length_errors++;
			return -EINVAL;
		}
247
		page = virtqueue_get_buf(vi->rvq, &len);
248 249 250 251 252
		if (!page) {
			pr_debug("%s: rx error: %d buffers missing\n",
				 skb->dev->name, hdr->mhdr.num_buffers);
			skb->dev->stats.rx_length_errors++;
			return -EINVAL;
253
		}
254

255 256 257 258 259 260 261 262 263 264 265 266 267
		if (len > PAGE_SIZE)
			len = PAGE_SIZE;

		set_skb_frag(skb, page, 0, &len);

		--vi->num;
	}
	return 0;
}

static void receive_buf(struct net_device *dev, void *buf, unsigned int len)
{
	struct virtnet_info *vi = netdev_priv(dev);
E
Eric Dumazet 已提交
268
	struct virtnet_stats *stats = this_cpu_ptr(vi->stats);
269 270 271
	struct sk_buff *skb;
	struct page *page;
	struct skb_vnet_hdr *hdr;
272

273 274 275 276 277 278 279 280 281
	if (unlikely(len < sizeof(struct virtio_net_hdr) + ETH_HLEN)) {
		pr_debug("%s: short packet %i\n", dev->name, len);
		dev->stats.rx_length_errors++;
		if (vi->mergeable_rx_bufs || vi->big_packets)
			give_pages(vi, buf);
		else
			dev_kfree_skb(buf);
		return;
	}
282

283 284 285 286 287 288 289 290
	if (!vi->mergeable_rx_bufs && !vi->big_packets) {
		skb = buf;
		len -= sizeof(struct virtio_net_hdr);
		skb_trim(skb, len);
	} else {
		page = buf;
		skb = page_to_skb(vi, page, len);
		if (unlikely(!skb)) {
291
			dev->stats.rx_dropped++;
292 293
			give_pages(vi, page);
			return;
294
		}
295 296 297 298 299
		if (vi->mergeable_rx_bufs)
			if (receive_mergeable(vi, skb)) {
				dev_kfree_skb(skb);
				return;
			}
300
	}
301

302
	hdr = skb_vnet_hdr(skb);
303

304
	u64_stats_update_begin(&stats->rx_syncp);
305 306
	stats->rx_bytes += skb->len;
	stats->rx_packets++;
307
	u64_stats_update_end(&stats->rx_syncp);
R
Rusty Russell 已提交
308

309
	if (hdr->hdr.flags & VIRTIO_NET_HDR_F_NEEDS_CSUM) {
R
Rusty Russell 已提交
310
		pr_debug("Needs csum!\n");
311 312 313
		if (!skb_partial_csum_set(skb,
					  hdr->hdr.csum_start,
					  hdr->hdr.csum_offset))
R
Rusty Russell 已提交
314
			goto frame_err;
315 316
	} else if (hdr->hdr.flags & VIRTIO_NET_HDR_F_DATA_VALID) {
		skb->ip_summed = CHECKSUM_UNNECESSARY;
R
Rusty Russell 已提交
317 318
	}

319 320 321 322
	skb->protocol = eth_type_trans(skb, dev);
	pr_debug("Receiving skb proto 0x%04x len %i type %i\n",
		 ntohs(skb->protocol), skb->len, skb->pkt_type);

323
	if (hdr->hdr.gso_type != VIRTIO_NET_HDR_GSO_NONE) {
R
Rusty Russell 已提交
324
		pr_debug("GSO!\n");
325
		switch (hdr->hdr.gso_type & ~VIRTIO_NET_HDR_GSO_ECN) {
R
Rusty Russell 已提交
326 327 328 329 330 331 332 333 334 335 336 337
		case VIRTIO_NET_HDR_GSO_TCPV4:
			skb_shinfo(skb)->gso_type = SKB_GSO_TCPV4;
			break;
		case VIRTIO_NET_HDR_GSO_UDP:
			skb_shinfo(skb)->gso_type = SKB_GSO_UDP;
			break;
		case VIRTIO_NET_HDR_GSO_TCPV6:
			skb_shinfo(skb)->gso_type = SKB_GSO_TCPV6;
			break;
		default:
			if (net_ratelimit())
				printk(KERN_WARNING "%s: bad gso type %u.\n",
338
				       dev->name, hdr->hdr.gso_type);
R
Rusty Russell 已提交
339 340 341
			goto frame_err;
		}

342
		if (hdr->hdr.gso_type & VIRTIO_NET_HDR_GSO_ECN)
R
Rusty Russell 已提交
343 344
			skb_shinfo(skb)->gso_type |= SKB_GSO_TCP_ECN;

345
		skb_shinfo(skb)->gso_size = hdr->hdr.gso_size;
R
Rusty Russell 已提交
346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365
		if (skb_shinfo(skb)->gso_size == 0) {
			if (net_ratelimit())
				printk(KERN_WARNING "%s: zero gso size.\n",
				       dev->name);
			goto frame_err;
		}

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

	netif_receive_skb(skb);
	return;

frame_err:
	dev->stats.rx_frame_errors++;
	dev_kfree_skb(skb);
}

366
static int add_recvbuf_small(struct virtnet_info *vi, gfp_t gfp)
R
Rusty Russell 已提交
367 368
{
	struct sk_buff *skb;
369 370
	struct skb_vnet_hdr *hdr;
	int err;
371

372
	skb = __netdev_alloc_skb_ip_align(vi->dev, MAX_PACKET_LEN, gfp);
373 374
	if (unlikely(!skb))
		return -ENOMEM;
R
Rusty Russell 已提交
375

376
	skb_put(skb, MAX_PACKET_LEN);
377

378
	hdr = skb_vnet_hdr(skb);
379
	sg_set_buf(vi->rx_sg, &hdr->hdr, sizeof hdr->hdr);
380

381
	skb_to_sgvec(skb, vi->rx_sg + 1, 0, skb->len);
382

383
	err = virtqueue_add_buf(vi->rvq, vi->rx_sg, 0, 2, skb, gfp);
384 385
	if (err < 0)
		dev_kfree_skb(skb);
386

387 388
	return err;
}
389

390 391 392 393 394 395
static int add_recvbuf_big(struct virtnet_info *vi, gfp_t gfp)
{
	struct page *first, *list = NULL;
	char *p;
	int i, err, offset;

396
	/* page in vi->rx_sg[MAX_SKB_FRAGS + 1] is list tail */
397 398 399 400 401 402
	for (i = MAX_SKB_FRAGS + 1; i > 1; --i) {
		first = get_a_page(vi, gfp);
		if (!first) {
			if (list)
				give_pages(vi, list);
			return -ENOMEM;
403
		}
404
		sg_set_buf(&vi->rx_sg[i], page_address(first), PAGE_SIZE);
405

406 407 408 409
		/* chain new page in list head to match sg */
		first->private = (unsigned long)list;
		list = first;
	}
R
Rusty Russell 已提交
410

411 412 413 414 415 416 417
	first = get_a_page(vi, gfp);
	if (!first) {
		give_pages(vi, list);
		return -ENOMEM;
	}
	p = page_address(first);

418 419 420
	/* vi->rx_sg[0], vi->rx_sg[1] share the same page */
	/* a separated vi->rx_sg[0] for virtio_net_hdr only due to QEMU bug */
	sg_set_buf(&vi->rx_sg[0], p, sizeof(struct virtio_net_hdr));
421

422
	/* vi->rx_sg[1] for data packet, from offset */
423
	offset = sizeof(struct padded_vnet_hdr);
424
	sg_set_buf(&vi->rx_sg[1], p + offset, PAGE_SIZE - offset);
425 426 427

	/* chain first in list head */
	first->private = (unsigned long)list;
428 429
	err = virtqueue_add_buf(vi->rvq, vi->rx_sg, 0, MAX_SKB_FRAGS + 2,
				first, gfp);
430 431 432 433
	if (err < 0)
		give_pages(vi, first);

	return err;
R
Rusty Russell 已提交
434 435
}

436
static int add_recvbuf_mergeable(struct virtnet_info *vi, gfp_t gfp)
437
{
438
	struct page *page;
439 440
	int err;

441 442 443
	page = get_a_page(vi, gfp);
	if (!page)
		return -ENOMEM;
444

445
	sg_init_one(vi->rx_sg, page_address(page), PAGE_SIZE);
446

447
	err = virtqueue_add_buf(vi->rvq, vi->rx_sg, 0, 1, page, gfp);
448 449
	if (err < 0)
		give_pages(vi, page);
450

451 452
	return err;
}
453

454 455 456 457 458 459 460
/*
 * Returns false if we couldn't fill entirely (OOM).
 *
 * Normally run in the receive path, but can also be run from ndo_open
 * before we're receiving packets, or from refill_work which is
 * careful to disable receiving (using napi_disable).
 */
461 462 463
static bool try_fill_recv(struct virtnet_info *vi, gfp_t gfp)
{
	int err;
464
	bool oom;
465

466 467 468 469 470 471 472
	do {
		if (vi->mergeable_rx_bufs)
			err = add_recvbuf_mergeable(vi, gfp);
		else if (vi->big_packets)
			err = add_recvbuf_big(vi, gfp);
		else
			err = add_recvbuf_small(vi, gfp);
473

474 475
		oom = err == -ENOMEM;
		if (err < 0)
476
			break;
477
		++vi->num;
478
	} while (err > 0);
479 480
	if (unlikely(vi->num > vi->max))
		vi->max = vi->num;
481
	virtqueue_kick(vi->rvq);
482
	return !oom;
483 484
}

485
static void skb_recv_done(struct virtqueue *rvq)
R
Rusty Russell 已提交
486 487
{
	struct virtnet_info *vi = rvq->vdev->priv;
488
	/* Schedule NAPI, Suppress further interrupts if successful. */
489
	if (napi_schedule_prep(&vi->napi)) {
490
		virtqueue_disable_cb(rvq);
491
		__napi_schedule(&vi->napi);
492
	}
R
Rusty Russell 已提交
493 494
}

495 496 497 498 499 500 501 502 503 504
static void virtnet_napi_enable(struct virtnet_info *vi)
{
	napi_enable(&vi->napi);

	/* If all buffers were filled by other side before we napi_enabled, we
	 * won't get another interrupt, so process any outstanding packets
	 * now.  virtnet_poll wants re-enable the queue, so we disable here.
	 * We synchronize against interrupts via NAPI_STATE_SCHED */
	if (napi_schedule_prep(&vi->napi)) {
		virtqueue_disable_cb(vi->rvq);
505
		local_bh_disable();
506
		__napi_schedule(&vi->napi);
507
		local_bh_enable();
508 509 510
	}
}

511 512 513 514 515 516 517
static void refill_work(struct work_struct *work)
{
	struct virtnet_info *vi;
	bool still_empty;

	vi = container_of(work, struct virtnet_info, refill.work);
	napi_disable(&vi->napi);
518
	still_empty = !try_fill_recv(vi, GFP_KERNEL);
519
	virtnet_napi_enable(vi);
520 521 522 523

	/* In theory, this can happen: if we don't get any buffers in
	 * we will *never* try to fill again. */
	if (still_empty)
524
		queue_delayed_work(system_nrt_wq, &vi->refill, HZ/2);
525 526
}

R
Rusty Russell 已提交
527 528 529
static int virtnet_poll(struct napi_struct *napi, int budget)
{
	struct virtnet_info *vi = container_of(napi, struct virtnet_info, napi);
530
	void *buf;
R
Rusty Russell 已提交
531 532 533 534
	unsigned int len, received = 0;

again:
	while (received < budget &&
535
	       (buf = virtqueue_get_buf(vi->rvq, &len)) != NULL) {
536 537
		receive_buf(vi->dev, buf, len);
		--vi->num;
R
Rusty Russell 已提交
538 539 540
		received++;
	}

541 542
	if (vi->num < vi->max / 2) {
		if (!try_fill_recv(vi, GFP_ATOMIC))
543
			queue_delayed_work(system_nrt_wq, &vi->refill, 0);
544
	}
R
Rusty Russell 已提交
545

546 547
	/* Out of packets? */
	if (received < budget) {
548
		napi_complete(napi);
549
		if (unlikely(!virtqueue_enable_cb(vi->rvq)) &&
550
		    napi_schedule_prep(napi)) {
551
			virtqueue_disable_cb(vi->rvq);
552
			__napi_schedule(napi);
R
Rusty Russell 已提交
553
			goto again;
554
		}
R
Rusty Russell 已提交
555 556 557 558 559
	}

	return received;
}

560
static unsigned int free_old_xmit_skbs(struct virtnet_info *vi)
R
Rusty Russell 已提交
561 562
{
	struct sk_buff *skb;
563
	unsigned int len, tot_sgs = 0;
E
Eric Dumazet 已提交
564
	struct virtnet_stats *stats = this_cpu_ptr(vi->stats);
R
Rusty Russell 已提交
565

566
	while ((skb = virtqueue_get_buf(vi->svq, &len)) != NULL) {
R
Rusty Russell 已提交
567
		pr_debug("Sent skb %p\n", skb);
568

569
		u64_stats_update_begin(&stats->tx_syncp);
570 571
		stats->tx_bytes += skb->len;
		stats->tx_packets++;
572
		u64_stats_update_end(&stats->tx_syncp);
573

574
		tot_sgs += skb_vnet_hdr(skb)->num_sg;
575
		dev_kfree_skb_any(skb);
R
Rusty Russell 已提交
576
	}
577
	return tot_sgs;
R
Rusty Russell 已提交
578 579
}

580
static int xmit_skb(struct virtnet_info *vi, struct sk_buff *skb)
R
Rusty Russell 已提交
581
{
582
	struct skb_vnet_hdr *hdr = skb_vnet_hdr(skb);
R
Rusty Russell 已提交
583 584
	const unsigned char *dest = ((struct ethhdr *)skb->data)->h_dest;

J
Johannes Berg 已提交
585
	pr_debug("%s: xmit %p %pM\n", vi->dev->name, skb, dest);
R
Rusty Russell 已提交
586 587

	if (skb->ip_summed == CHECKSUM_PARTIAL) {
588
		hdr->hdr.flags = VIRTIO_NET_HDR_F_NEEDS_CSUM;
589
		hdr->hdr.csum_start = skb_checksum_start_offset(skb);
590
		hdr->hdr.csum_offset = skb->csum_offset;
R
Rusty Russell 已提交
591
	} else {
592 593
		hdr->hdr.flags = 0;
		hdr->hdr.csum_offset = hdr->hdr.csum_start = 0;
R
Rusty Russell 已提交
594 595 596
	}

	if (skb_is_gso(skb)) {
597 598
		hdr->hdr.hdr_len = skb_headlen(skb);
		hdr->hdr.gso_size = skb_shinfo(skb)->gso_size;
R
Rusty Russell 已提交
599
		if (skb_shinfo(skb)->gso_type & SKB_GSO_TCPV4)
600
			hdr->hdr.gso_type = VIRTIO_NET_HDR_GSO_TCPV4;
R
Rusty Russell 已提交
601
		else if (skb_shinfo(skb)->gso_type & SKB_GSO_TCPV6)
602
			hdr->hdr.gso_type = VIRTIO_NET_HDR_GSO_TCPV6;
R
Rusty Russell 已提交
603
		else if (skb_shinfo(skb)->gso_type & SKB_GSO_UDP)
604
			hdr->hdr.gso_type = VIRTIO_NET_HDR_GSO_UDP;
R
Rusty Russell 已提交
605 606
		else
			BUG();
R
Rusty Russell 已提交
607
		if (skb_shinfo(skb)->gso_type & SKB_GSO_TCP_ECN)
608
			hdr->hdr.gso_type |= VIRTIO_NET_HDR_GSO_ECN;
R
Rusty Russell 已提交
609
	} else {
610 611
		hdr->hdr.gso_type = VIRTIO_NET_HDR_GSO_NONE;
		hdr->hdr.gso_size = hdr->hdr.hdr_len = 0;
R
Rusty Russell 已提交
612 613
	}

614
	hdr->mhdr.num_buffers = 0;
615 616 617

	/* Encode metadata header at front. */
	if (vi->mergeable_rx_bufs)
618
		sg_set_buf(vi->tx_sg, &hdr->mhdr, sizeof hdr->mhdr);
619
	else
620
		sg_set_buf(vi->tx_sg, &hdr->hdr, sizeof hdr->hdr);
621

622
	hdr->num_sg = skb_to_sgvec(skb, vi->tx_sg + 1, 0, skb->len) + 1;
623
	return virtqueue_add_buf(vi->svq, vi->tx_sg, hdr->num_sg,
624
				 0, skb, GFP_ATOMIC);
625 626
}

627
static netdev_tx_t start_xmit(struct sk_buff *skb, struct net_device *dev)
628 629
{
	struct virtnet_info *vi = netdev_priv(dev);
630
	int capacity;
631 632 633

	/* Free up any pending old buffers before queueing new ones. */
	free_old_xmit_skbs(vi);
634

635
	/* Try to transmit */
636 637 638 639
	capacity = xmit_skb(vi, skb);

	/* This can happen with OOM and indirect buffers. */
	if (unlikely(capacity < 0)) {
640
		if (likely(capacity == -ENOMEM)) {
641
			if (net_ratelimit())
642 643
				dev_warn(&dev->dev,
					 "TX queue failure: out of memory\n");
644
		} else {
645 646
			dev->stats.tx_fifo_errors++;
			if (net_ratelimit())
647 648 649
				dev_warn(&dev->dev,
					 "Unexpected TX queue failure: %d\n",
					 capacity);
650
		}
651 652 653
		dev->stats.tx_dropped++;
		kfree_skb(skb);
		return NETDEV_TX_OK;
R
Rusty Russell 已提交
654
	}
655
	virtqueue_kick(vi->svq);
656

657 658 659 660 661 662 663 664
	/* Don't wait up for transmitted skbs to be freed. */
	skb_orphan(skb);
	nf_reset(skb);

	/* Apparently nice girls don't return TX_BUSY; stop the queue
	 * before it gets out of hand.  Naturally, this wastes entries. */
	if (capacity < 2+MAX_SKB_FRAGS) {
		netif_stop_queue(dev);
665
		if (unlikely(!virtqueue_enable_cb_delayed(vi->svq))) {
666 667 668 669
			/* More just got used, free them then recheck. */
			capacity += free_old_xmit_skbs(vi);
			if (capacity >= 2+MAX_SKB_FRAGS) {
				netif_start_queue(dev);
670
				virtqueue_disable_cb(vi->svq);
671 672
			}
		}
673
	}
674 675

	return NETDEV_TX_OK;
R
Rusty Russell 已提交
676 677
}

678 679 680 681
static int virtnet_set_mac_address(struct net_device *dev, void *p)
{
	struct virtnet_info *vi = netdev_priv(dev);
	struct virtio_device *vdev = vi->vdev;
682
	struct sockaddr *addr = p;
683

684 685 686 687
	if (!is_valid_ether_addr(addr->sa_data))
		return -EADDRNOTAVAIL;
	memcpy(dev->dev_addr, addr->sa_data, ETH_ALEN);
	dev->addr_assign_type &= ~NET_ADDR_RANDOM;
688

689 690 691
	if (virtio_has_feature(vdev, VIRTIO_NET_F_MAC))
		vdev->config->set(vdev, offsetof(struct virtio_net_config, mac),
		                  dev->dev_addr, dev->addr_len);
692 693 694 695

	return 0;
}

696 697 698 699 700 701 702 703
static struct rtnl_link_stats64 *virtnet_stats(struct net_device *dev,
					       struct rtnl_link_stats64 *tot)
{
	struct virtnet_info *vi = netdev_priv(dev);
	int cpu;
	unsigned int start;

	for_each_possible_cpu(cpu) {
E
Eric Dumazet 已提交
704
		struct virtnet_stats *stats = per_cpu_ptr(vi->stats, cpu);
705 706 707
		u64 tpackets, tbytes, rpackets, rbytes;

		do {
708
			start = u64_stats_fetch_begin(&stats->tx_syncp);
709 710
			tpackets = stats->tx_packets;
			tbytes   = stats->tx_bytes;
711 712 713 714
		} while (u64_stats_fetch_retry(&stats->tx_syncp, start));

		do {
			start = u64_stats_fetch_begin(&stats->rx_syncp);
715 716
			rpackets = stats->rx_packets;
			rbytes   = stats->rx_bytes;
717
		} while (u64_stats_fetch_retry(&stats->rx_syncp, start));
718 719 720 721 722 723 724 725

		tot->rx_packets += rpackets;
		tot->tx_packets += tpackets;
		tot->rx_bytes   += rbytes;
		tot->tx_bytes   += tbytes;
	}

	tot->tx_dropped = dev->stats.tx_dropped;
726
	tot->tx_fifo_errors = dev->stats.tx_fifo_errors;
727 728 729 730 731 732 733
	tot->rx_dropped = dev->stats.rx_dropped;
	tot->rx_length_errors = dev->stats.rx_length_errors;
	tot->rx_frame_errors = dev->stats.rx_frame_errors;

	return tot;
}

734 735 736 737 738 739 740 741 742
#ifdef CONFIG_NET_POLL_CONTROLLER
static void virtnet_netpoll(struct net_device *dev)
{
	struct virtnet_info *vi = netdev_priv(dev);

	napi_schedule(&vi->napi);
}
#endif

R
Rusty Russell 已提交
743 744 745 746
static int virtnet_open(struct net_device *dev)
{
	struct virtnet_info *vi = netdev_priv(dev);

747 748
	/* Make sure we have some buffers: if oom use wq. */
	if (!try_fill_recv(vi, GFP_KERNEL))
749
		queue_delayed_work(system_nrt_wq, &vi->refill, 0);
750

751
	virtnet_napi_enable(vi);
R
Rusty Russell 已提交
752 753 754
	return 0;
}

755 756 757 758 759 760 761 762
/*
 * Send command via the control virtqueue and check status.  Commands
 * supported by the hypervisor, as indicated by feature bits, should
 * never fail unless improperly formated.
 */
static bool virtnet_send_command(struct virtnet_info *vi, u8 class, u8 cmd,
				 struct scatterlist *data, int out, int in)
{
763
	struct scatterlist *s, sg[VIRTNET_SEND_COMMAND_SG_MAX + 2];
764 765 766
	struct virtio_net_ctrl_hdr ctrl;
	virtio_net_ctrl_ack status = ~0;
	unsigned int tmp;
767
	int i;
768

769 770 771
	/* Caller should know better */
	BUG_ON(!virtio_has_feature(vi->vdev, VIRTIO_NET_F_CTRL_VQ) ||
		(out + in > VIRTNET_SEND_COMMAND_SG_MAX));
772 773 774 775 776 777 778 779 780 781

	out++; /* Add header */
	in++; /* Add return status */

	ctrl.class = class;
	ctrl.cmd = cmd;

	sg_init_table(sg, out + in);

	sg_set_buf(&sg[0], &ctrl, sizeof(ctrl));
782 783
	for_each_sg(data, s, out + in - 2, i)
		sg_set_buf(&sg[i + 1], sg_virt(s), s->length);
784 785
	sg_set_buf(&sg[out + in - 1], &status, sizeof(status));

786
	BUG_ON(virtqueue_add_buf(vi->cvq, sg, out, in, vi, GFP_ATOMIC) < 0);
787

788
	virtqueue_kick(vi->cvq);
789 790 791 792 793

	/*
	 * Spin for a response, the kick causes an ioport write, trapping
	 * into the hypervisor, so the request should be handled immediately.
	 */
794
	while (!virtqueue_get_buf(vi->cvq, &tmp))
795 796 797 798 799
		cpu_relax();

	return status == VIRTIO_NET_OK;
}

800 801 802 803 804 805 806 807 808 809
static void virtnet_ack_link_announce(struct virtnet_info *vi)
{
	rtnl_lock();
	if (!virtnet_send_command(vi, VIRTIO_NET_CTRL_ANNOUNCE,
				  VIRTIO_NET_CTRL_ANNOUNCE_ACK, NULL,
				  0, 0))
		dev_warn(&vi->dev->dev, "Failed to ack link announce.\n");
	rtnl_unlock();
}

R
Rusty Russell 已提交
810 811 812 813
static int virtnet_close(struct net_device *dev)
{
	struct virtnet_info *vi = netdev_priv(dev);

814 815
	/* Make sure refill_work doesn't re-enable napi! */
	cancel_delayed_work_sync(&vi->refill);
R
Rusty Russell 已提交
816 817 818 819 820
	napi_disable(&vi->napi);

	return 0;
}

821 822 823
static void virtnet_set_rx_mode(struct net_device *dev)
{
	struct virtnet_info *vi = netdev_priv(dev);
824
	struct scatterlist sg[2];
825
	u8 promisc, allmulti;
826
	struct virtio_net_ctrl_mac *mac_data;
J
Jiri Pirko 已提交
827
	struct netdev_hw_addr *ha;
828
	int uc_count;
829
	int mc_count;
830 831
	void *buf;
	int i;
832 833 834 835 836

	/* We can't dynamicaly set ndo_set_rx_mode, so return gracefully */
	if (!virtio_has_feature(vi->vdev, VIRTIO_NET_F_CTRL_RX))
		return;

837 838
	promisc = ((dev->flags & IFF_PROMISC) != 0);
	allmulti = ((dev->flags & IFF_ALLMULTI) != 0);
839

840
	sg_init_one(sg, &promisc, sizeof(promisc));
841 842 843

	if (!virtnet_send_command(vi, VIRTIO_NET_CTRL_RX,
				  VIRTIO_NET_CTRL_RX_PROMISC,
844
				  sg, 1, 0))
845 846 847
		dev_warn(&dev->dev, "Failed to %sable promisc mode.\n",
			 promisc ? "en" : "dis");

848
	sg_init_one(sg, &allmulti, sizeof(allmulti));
849 850 851

	if (!virtnet_send_command(vi, VIRTIO_NET_CTRL_RX,
				  VIRTIO_NET_CTRL_RX_ALLMULTI,
852
				  sg, 1, 0))
853 854
		dev_warn(&dev->dev, "Failed to %sable allmulti mode.\n",
			 allmulti ? "en" : "dis");
855

856
	uc_count = netdev_uc_count(dev);
857
	mc_count = netdev_mc_count(dev);
858
	/* MAC filter - use one buffer for both lists */
859 860 861
	buf = kzalloc(((uc_count + mc_count) * ETH_ALEN) +
		      (2 * sizeof(mac_data->entries)), GFP_ATOMIC);
	mac_data = buf;
862 863 864 865 866
	if (!buf) {
		dev_warn(&dev->dev, "No memory for MAC address buffer\n");
		return;
	}

867 868
	sg_init_table(sg, 2);

869
	/* Store the unicast list and count in the front of the buffer */
870
	mac_data->entries = uc_count;
J
Jiri Pirko 已提交
871
	i = 0;
872
	netdev_for_each_uc_addr(ha, dev)
J
Jiri Pirko 已提交
873
		memcpy(&mac_data->macs[i++][0], ha->addr, ETH_ALEN);
874 875

	sg_set_buf(&sg[0], mac_data,
876
		   sizeof(mac_data->entries) + (uc_count * ETH_ALEN));
877 878

	/* multicast list and count fill the end */
879
	mac_data = (void *)&mac_data->macs[uc_count][0];
880

881
	mac_data->entries = mc_count;
882
	i = 0;
883 884
	netdev_for_each_mc_addr(ha, dev)
		memcpy(&mac_data->macs[i++][0], ha->addr, ETH_ALEN);
885 886

	sg_set_buf(&sg[1], mac_data,
887
		   sizeof(mac_data->entries) + (mc_count * ETH_ALEN));
888 889 890 891 892 893 894

	if (!virtnet_send_command(vi, VIRTIO_NET_CTRL_MAC,
				  VIRTIO_NET_CTRL_MAC_TABLE_SET,
				  sg, 2, 0))
		dev_warn(&dev->dev, "Failed to set MAC fitler table.\n");

	kfree(buf);
895 896
}

897
static int virtnet_vlan_rx_add_vid(struct net_device *dev, u16 vid)
898 899 900 901
{
	struct virtnet_info *vi = netdev_priv(dev);
	struct scatterlist sg;

902
	sg_init_one(&sg, &vid, sizeof(vid));
903 904 905 906

	if (!virtnet_send_command(vi, VIRTIO_NET_CTRL_VLAN,
				  VIRTIO_NET_CTRL_VLAN_ADD, &sg, 1, 0))
		dev_warn(&dev->dev, "Failed to add VLAN ID %d.\n", vid);
907
	return 0;
908 909
}

910
static int virtnet_vlan_rx_kill_vid(struct net_device *dev, u16 vid)
911 912 913 914
{
	struct virtnet_info *vi = netdev_priv(dev);
	struct scatterlist sg;

915
	sg_init_one(&sg, &vid, sizeof(vid));
916 917 918 919

	if (!virtnet_send_command(vi, VIRTIO_NET_CTRL_VLAN,
				  VIRTIO_NET_CTRL_VLAN_DEL, &sg, 1, 0))
		dev_warn(&dev->dev, "Failed to kill VLAN ID %d.\n", vid);
920
	return 0;
921 922
}

R
Rick Jones 已提交
923 924 925 926 927 928 929 930 931 932 933 934
static void virtnet_get_ringparam(struct net_device *dev,
				struct ethtool_ringparam *ring)
{
	struct virtnet_info *vi = netdev_priv(dev);

	ring->rx_max_pending = virtqueue_get_vring_size(vi->rvq);
	ring->tx_max_pending = virtqueue_get_vring_size(vi->svq);
	ring->rx_pending = ring->rx_max_pending;
	ring->tx_pending = ring->tx_max_pending;

}

935 936 937 938 939 940 941 942 943 944 945 946 947

static void virtnet_get_drvinfo(struct net_device *dev,
				struct ethtool_drvinfo *info)
{
	struct virtnet_info *vi = netdev_priv(dev);
	struct virtio_device *vdev = vi->vdev;

	strlcpy(info->driver, KBUILD_MODNAME, sizeof(info->driver));
	strlcpy(info->version, VIRTNET_DRIVER_VERSION, sizeof(info->version));
	strlcpy(info->bus_info, virtio_bus_name(vdev), sizeof(info->bus_info));

}

948
static const struct ethtool_ops virtnet_ethtool_ops = {
949
	.get_drvinfo = virtnet_get_drvinfo,
950
	.get_link = ethtool_op_get_link,
R
Rick Jones 已提交
951
	.get_ringparam = virtnet_get_ringparam,
952 953
};

M
Mark McLoughlin 已提交
954 955 956 957 958 959 960 961 962 963 964
#define MIN_MTU 68
#define MAX_MTU 65535

static int virtnet_change_mtu(struct net_device *dev, int new_mtu)
{
	if (new_mtu < MIN_MTU || new_mtu > MAX_MTU)
		return -EINVAL;
	dev->mtu = new_mtu;
	return 0;
}

965 966 967 968 969
static const struct net_device_ops virtnet_netdev = {
	.ndo_open            = virtnet_open,
	.ndo_stop   	     = virtnet_close,
	.ndo_start_xmit      = start_xmit,
	.ndo_validate_addr   = eth_validate_addr,
970
	.ndo_set_mac_address = virtnet_set_mac_address,
971
	.ndo_set_rx_mode     = virtnet_set_rx_mode,
972
	.ndo_change_mtu	     = virtnet_change_mtu,
973
	.ndo_get_stats64     = virtnet_stats,
974 975
	.ndo_vlan_rx_add_vid = virtnet_vlan_rx_add_vid,
	.ndo_vlan_rx_kill_vid = virtnet_vlan_rx_kill_vid,
976 977 978 979 980
#ifdef CONFIG_NET_POLL_CONTROLLER
	.ndo_poll_controller = virtnet_netpoll,
#endif
};

981
static void virtnet_config_changed_work(struct work_struct *work)
982
{
983 984
	struct virtnet_info *vi =
		container_of(work, struct virtnet_info, config_work);
985 986
	u16 v;

987 988 989 990
	mutex_lock(&vi->config_lock);
	if (!vi->config_enable)
		goto done;

991
	if (virtio_config_val(vi->vdev, VIRTIO_NET_F_STATUS,
992
			      offsetof(struct virtio_net_config, status),
993
			      &v) < 0)
994 995 996 997 998 999
		goto done;

	if (v & VIRTIO_NET_S_ANNOUNCE) {
		netif_notify_peers(vi->dev);
		virtnet_ack_link_announce(vi);
	}
1000 1001 1002 1003 1004

	/* Ignore unknown (future) status bits */
	v &= VIRTIO_NET_S_LINK_UP;

	if (vi->status == v)
1005
		goto done;
1006 1007 1008 1009 1010 1011 1012 1013 1014 1015

	vi->status = v;

	if (vi->status & VIRTIO_NET_S_LINK_UP) {
		netif_carrier_on(vi->dev);
		netif_wake_queue(vi->dev);
	} else {
		netif_carrier_off(vi->dev);
		netif_stop_queue(vi->dev);
	}
1016 1017
done:
	mutex_unlock(&vi->config_lock);
1018 1019 1020 1021 1022 1023
}

static void virtnet_config_changed(struct virtio_device *vdev)
{
	struct virtnet_info *vi = vdev->priv;

1024
	queue_work(system_nrt_wq, &vi->config_work);
1025 1026
}

1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053
static int init_vqs(struct virtnet_info *vi)
{
	struct virtqueue *vqs[3];
	vq_callback_t *callbacks[] = { skb_recv_done, skb_xmit_done, NULL};
	const char *names[] = { "input", "output", "control" };
	int nvqs, err;

	/* We expect two virtqueues, receive then send,
	 * and optionally control. */
	nvqs = virtio_has_feature(vi->vdev, VIRTIO_NET_F_CTRL_VQ) ? 3 : 2;

	err = vi->vdev->config->find_vqs(vi->vdev, nvqs, vqs, callbacks, names);
	if (err)
		return err;

	vi->rvq = vqs[0];
	vi->svq = vqs[1];

	if (virtio_has_feature(vi->vdev, VIRTIO_NET_F_CTRL_VQ)) {
		vi->cvq = vqs[2];

		if (virtio_has_feature(vi->vdev, VIRTIO_NET_F_CTRL_VLAN))
			vi->dev->features |= NETIF_F_HW_VLAN_FILTER;
	}
	return 0;
}

R
Rusty Russell 已提交
1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065
static int virtnet_probe(struct virtio_device *vdev)
{
	int err;
	struct net_device *dev;
	struct virtnet_info *vi;

	/* Allocate ourselves a network device with room for our info */
	dev = alloc_etherdev(sizeof(struct virtnet_info));
	if (!dev)
		return -ENOMEM;

	/* Set up network device as normal. */
1066
	dev->priv_flags |= IFF_UNICAST_FLT;
1067
	dev->netdev_ops = &virtnet_netdev;
R
Rusty Russell 已提交
1068
	dev->features = NETIF_F_HIGHDMA;
1069

1070
	SET_ETHTOOL_OPS(dev, &virtnet_ethtool_ops);
R
Rusty Russell 已提交
1071 1072 1073
	SET_NETDEV_DEV(dev, &vdev->dev);

	/* Do we support "hardware" checksums? */
1074
	if (virtio_has_feature(vdev, VIRTIO_NET_F_CSUM)) {
R
Rusty Russell 已提交
1075
		/* This opens up the world of extra features. */
1076 1077 1078 1079 1080 1081
		dev->hw_features |= NETIF_F_HW_CSUM|NETIF_F_SG|NETIF_F_FRAGLIST;
		if (csum)
			dev->features |= NETIF_F_HW_CSUM|NETIF_F_SG|NETIF_F_FRAGLIST;

		if (virtio_has_feature(vdev, VIRTIO_NET_F_GSO)) {
			dev->hw_features |= NETIF_F_TSO | NETIF_F_UFO
R
Rusty Russell 已提交
1082 1083
				| NETIF_F_TSO_ECN | NETIF_F_TSO6;
		}
1084
		/* Individual feature bits: what can host handle? */
1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096
		if (virtio_has_feature(vdev, VIRTIO_NET_F_HOST_TSO4))
			dev->hw_features |= NETIF_F_TSO;
		if (virtio_has_feature(vdev, VIRTIO_NET_F_HOST_TSO6))
			dev->hw_features |= NETIF_F_TSO6;
		if (virtio_has_feature(vdev, VIRTIO_NET_F_HOST_ECN))
			dev->hw_features |= NETIF_F_TSO_ECN;
		if (virtio_has_feature(vdev, VIRTIO_NET_F_HOST_UFO))
			dev->hw_features |= NETIF_F_UFO;

		if (gso)
			dev->features |= dev->hw_features & (NETIF_F_ALL_TSO|NETIF_F_UFO);
		/* (!csum && gso) case will be fixed by register_netdev() */
R
Rusty Russell 已提交
1097 1098 1099
	}

	/* Configuration may specify what MAC to use.  Otherwise random. */
1100
	if (virtio_config_val_len(vdev, VIRTIO_NET_F_MAC,
1101
				  offsetof(struct virtio_net_config, mac),
1102
				  dev->dev_addr, dev->addr_len) < 0)
1103
		eth_hw_addr_random(dev);
R
Rusty Russell 已提交
1104 1105 1106

	/* Set up our device-specific information */
	vi = netdev_priv(dev);
1107
	netif_napi_add(dev, &vi->napi, virtnet_poll, napi_weight);
R
Rusty Russell 已提交
1108 1109
	vi->dev = dev;
	vi->vdev = vdev;
1110
	vdev->priv = vi;
1111
	vi->pages = NULL;
1112 1113 1114 1115 1116
	vi->stats = alloc_percpu(struct virtnet_stats);
	err = -ENOMEM;
	if (vi->stats == NULL)
		goto free;

1117
	INIT_DELAYED_WORK(&vi->refill, refill_work);
1118 1119 1120
	mutex_init(&vi->config_lock);
	vi->config_enable = true;
	INIT_WORK(&vi->config_work, virtnet_config_changed_work);
1121 1122
	sg_init_table(vi->rx_sg, ARRAY_SIZE(vi->rx_sg));
	sg_init_table(vi->tx_sg, ARRAY_SIZE(vi->tx_sg));
R
Rusty Russell 已提交
1123

1124
	/* If we can receive ANY GSO packets, we must allocate large ones. */
1125 1126 1127
	if (virtio_has_feature(vdev, VIRTIO_NET_F_GUEST_TSO4) ||
	    virtio_has_feature(vdev, VIRTIO_NET_F_GUEST_TSO6) ||
	    virtio_has_feature(vdev, VIRTIO_NET_F_GUEST_ECN))
1128 1129
		vi->big_packets = true;

1130 1131 1132
	if (virtio_has_feature(vdev, VIRTIO_NET_F_MRG_RXBUF))
		vi->mergeable_rx_bufs = true;

1133
	err = init_vqs(vi);
1134
	if (err)
1135
		goto free_stats;
R
Rusty Russell 已提交
1136 1137 1138 1139

	err = register_netdev(dev);
	if (err) {
		pr_debug("virtio_net: registering device failed\n");
1140
		goto free_vqs;
R
Rusty Russell 已提交
1141
	}
1142 1143

	/* Last of all, set up some receive buffers. */
1144
	try_fill_recv(vi, GFP_KERNEL);
1145 1146 1147 1148 1149 1150 1151

	/* If we didn't even get one input buffer, we're useless. */
	if (vi->num == 0) {
		err = -ENOMEM;
		goto unregister;
	}

J
Jason Wang 已提交
1152 1153 1154 1155
	/* Assume link up if device can't report link status,
	   otherwise get link status from config. */
	if (virtio_has_feature(vi->vdev, VIRTIO_NET_F_STATUS)) {
		netif_carrier_off(dev);
1156
		queue_work(system_nrt_wq, &vi->config_work);
J
Jason Wang 已提交
1157 1158 1159 1160
	} else {
		vi->status = VIRTIO_NET_S_LINK_UP;
		netif_carrier_on(dev);
	}
1161

R
Rusty Russell 已提交
1162 1163 1164
	pr_debug("virtnet: registered device %s\n", dev->name);
	return 0;

1165 1166
unregister:
	unregister_netdev(dev);
1167 1168
free_vqs:
	vdev->config->del_vqs(vdev);
1169 1170
free_stats:
	free_percpu(vi->stats);
R
Rusty Russell 已提交
1171 1172 1173 1174 1175
free:
	free_netdev(dev);
	return err;
}

1176 1177 1178
static void free_unused_bufs(struct virtnet_info *vi)
{
	void *buf;
S
Shirley Ma 已提交
1179
	while (1) {
1180
		buf = virtqueue_detach_unused_buf(vi->svq);
S
Shirley Ma 已提交
1181 1182 1183 1184
		if (!buf)
			break;
		dev_kfree_skb(buf);
	}
1185
	while (1) {
1186
		buf = virtqueue_detach_unused_buf(vi->rvq);
1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197
		if (!buf)
			break;
		if (vi->mergeable_rx_bufs || vi->big_packets)
			give_pages(vi, buf);
		else
			dev_kfree_skb(buf);
		--vi->num;
	}
	BUG_ON(vi->num != 0);
}

1198
static void remove_vq_common(struct virtnet_info *vi)
R
Rusty Russell 已提交
1199
{
1200
	vi->vdev->config->reset(vi->vdev);
S
Shirley Ma 已提交
1201 1202

	/* Free unused buffers in both send and recv, if any. */
1203
	free_unused_bufs(vi);
1204

1205
	vi->vdev->config->del_vqs(vi->vdev);
1206

1207 1208
	while (vi->pages)
		__free_pages(get_a_page(vi, GFP_KERNEL), 0);
1209 1210 1211 1212 1213 1214
}

static void __devexit virtnet_remove(struct virtio_device *vdev)
{
	struct virtnet_info *vi = vdev->priv;

1215 1216 1217 1218 1219
	/* Prevent config work handler from accessing the device. */
	mutex_lock(&vi->config_lock);
	vi->config_enable = false;
	mutex_unlock(&vi->config_lock);

1220 1221 1222
	unregister_netdev(vi->dev);

	remove_vq_common(vi);
1223

1224 1225
	flush_work(&vi->config_work);

1226
	free_percpu(vi->stats);
1227
	free_netdev(vi->dev);
R
Rusty Russell 已提交
1228 1229
}

1230 1231 1232 1233 1234
#ifdef CONFIG_PM
static int virtnet_freeze(struct virtio_device *vdev)
{
	struct virtnet_info *vi = vdev->priv;

1235 1236 1237 1238 1239
	/* Prevent config work handler from accessing the device */
	mutex_lock(&vi->config_lock);
	vi->config_enable = false;
	mutex_unlock(&vi->config_lock);

1240 1241 1242 1243 1244 1245 1246 1247
	netif_device_detach(vi->dev);
	cancel_delayed_work_sync(&vi->refill);

	if (netif_running(vi->dev))
		napi_disable(&vi->napi);

	remove_vq_common(vi);

1248 1249
	flush_work(&vi->config_work);

1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269
	return 0;
}

static int virtnet_restore(struct virtio_device *vdev)
{
	struct virtnet_info *vi = vdev->priv;
	int err;

	err = init_vqs(vi);
	if (err)
		return err;

	if (netif_running(vi->dev))
		virtnet_napi_enable(vi);

	netif_device_attach(vi->dev);

	if (!try_fill_recv(vi, GFP_KERNEL))
		queue_delayed_work(system_nrt_wq, &vi->refill, 0);

1270 1271 1272 1273
	mutex_lock(&vi->config_lock);
	vi->config_enable = true;
	mutex_unlock(&vi->config_lock);

1274 1275 1276 1277
	return 0;
}
#endif

R
Rusty Russell 已提交
1278 1279 1280 1281 1282
static struct virtio_device_id id_table[] = {
	{ VIRTIO_ID_NET, VIRTIO_DEV_ANY_ID },
	{ 0 },
};

1283
static unsigned int features[] = {
1284 1285
	VIRTIO_NET_F_CSUM, VIRTIO_NET_F_GUEST_CSUM,
	VIRTIO_NET_F_GSO, VIRTIO_NET_F_MAC,
1286
	VIRTIO_NET_F_HOST_TSO4, VIRTIO_NET_F_HOST_UFO, VIRTIO_NET_F_HOST_TSO6,
1287
	VIRTIO_NET_F_HOST_ECN, VIRTIO_NET_F_GUEST_TSO4, VIRTIO_NET_F_GUEST_TSO6,
1288
	VIRTIO_NET_F_GUEST_ECN, VIRTIO_NET_F_GUEST_UFO,
1289
	VIRTIO_NET_F_MRG_RXBUF, VIRTIO_NET_F_STATUS, VIRTIO_NET_F_CTRL_VQ,
1290
	VIRTIO_NET_F_CTRL_RX, VIRTIO_NET_F_CTRL_VLAN,
1291
	VIRTIO_NET_F_GUEST_ANNOUNCE,
1292 1293
};

1294
static struct virtio_driver virtio_net_driver = {
1295 1296
	.feature_table = features,
	.feature_table_size = ARRAY_SIZE(features),
R
Rusty Russell 已提交
1297 1298 1299 1300 1301
	.driver.name =	KBUILD_MODNAME,
	.driver.owner =	THIS_MODULE,
	.id_table =	id_table,
	.probe =	virtnet_probe,
	.remove =	__devexit_p(virtnet_remove),
1302
	.config_changed = virtnet_config_changed,
1303 1304 1305 1306
#ifdef CONFIG_PM
	.freeze =	virtnet_freeze,
	.restore =	virtnet_restore,
#endif
R
Rusty Russell 已提交
1307 1308 1309 1310
};

static int __init init(void)
{
1311
	return register_virtio_driver(&virtio_net_driver);
R
Rusty Russell 已提交
1312 1313 1314 1315
}

static void __exit fini(void)
{
1316
	unregister_virtio_driver(&virtio_net_driver);
R
Rusty Russell 已提交
1317 1318 1319 1320 1321 1322 1323
}
module_init(init);
module_exit(fini);

MODULE_DEVICE_TABLE(virtio, id_table);
MODULE_DESCRIPTION("Virtio network driver");
MODULE_LICENSE("GPL");