xen-netfront.c 53.9 KB
Newer Older
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31
/*
 * Virtual network driver for conversing with remote driver backends.
 *
 * Copyright (c) 2002-2005, K A Fraser
 * Copyright (c) 2005, XenSource Ltd
 *
 * This program is free software; you can redistribute it and/or
 * modify it under the terms of the GNU General Public License version 2
 * as published by the Free Software Foundation; or, when distributed
 * separately from the Linux kernel or incorporated into other
 * software packages, subject to the following license:
 *
 * Permission is hereby granted, free of charge, to any person obtaining a copy
 * of this source file (the "Software"), to deal in the Software without
 * restriction, including without limitation the rights to use, copy, modify,
 * merge, publish, distribute, sublicense, and/or sell copies of the Software,
 * and to permit persons to whom the Software is furnished to do so, subject to
 * the following conditions:
 *
 * The above copyright notice and this permission notice shall be included in
 * all copies or substantial portions of the Software.
 *
 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
 * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
 * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
 * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
 * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
 * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
 * IN THE SOFTWARE.
 */

32 33
#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt

34 35 36 37 38 39 40
#include <linux/module.h>
#include <linux/kernel.h>
#include <linux/netdevice.h>
#include <linux/etherdevice.h>
#include <linux/skbuff.h>
#include <linux/ethtool.h>
#include <linux/if_ether.h>
41
#include <net/tcp.h>
42 43 44
#include <linux/udp.h>
#include <linux/moduleparam.h>
#include <linux/mm.h>
45
#include <linux/slab.h>
46 47
#include <net/ip.h>

48
#include <xen/xen.h>
49 50 51
#include <xen/xenbus.h>
#include <xen/events.h>
#include <xen/page.h>
52
#include <xen/platform_pci.h>
53 54 55 56 57 58
#include <xen/grant_table.h>

#include <xen/interface/io/netif.h>
#include <xen/interface/memory.h>
#include <xen/interface/grant_table.h>

59 60 61 62 63 64
/* Module parameters */
static unsigned int xennet_max_queues;
module_param_named(max_queues, xennet_max_queues, uint, 0644);
MODULE_PARM_DESC(max_queues,
		 "Maximum number of queues per virtual interface");

65
static const struct ethtool_ops xennet_ethtool_ops;
66 67

struct netfront_cb {
68
	int pull_to;
69 70 71 72 73 74 75 76
};

#define NETFRONT_SKB_CB(skb)	((struct netfront_cb *)((skb)->cb))

#define RX_COPY_THRESHOLD 256

#define GRANT_INVALID_REF	0

77 78
#define NET_TX_RING_SIZE __CONST_RING_SIZE(xen_netif_tx, XEN_PAGE_SIZE)
#define NET_RX_RING_SIZE __CONST_RING_SIZE(xen_netif_rx, XEN_PAGE_SIZE)
79 80 81

/* Minimum number of Rx slots (includes slot for GSO metadata). */
#define NET_RX_SLOTS_MIN (XEN_NETIF_NR_SLOTS_MIN + 1)
82

83 84 85 86 87 88
/* Queue name is interface name with "-qNNN" appended */
#define QUEUE_NAME_SIZE (IFNAMSIZ + 6)

/* IRQ name is queue name with "-tx" or "-rx" appended */
#define IRQ_NAME_SIZE (QUEUE_NAME_SIZE + 3)

89
struct netfront_stats {
90 91
	u64			packets;
	u64			bytes;
92 93 94
	struct u64_stats_sync	syncp;
};

95 96 97 98 99 100
struct netfront_info;

struct netfront_queue {
	unsigned int id; /* Queue ID, 0-based */
	char name[QUEUE_NAME_SIZE]; /* DEVNAME-qN */
	struct netfront_info *info;
101

102
	struct napi_struct napi;
103

104 105 106 107 108 109
	/* Split event channels support, tx_* == rx_* when using
	 * single event channel.
	 */
	unsigned int tx_evtchn, rx_evtchn;
	unsigned int tx_irq, rx_irq;
	/* Only used when split event channels support is enabled */
110 111
	char tx_irq_name[IRQ_NAME_SIZE]; /* DEVNAME-qN-tx */
	char rx_irq_name[IRQ_NAME_SIZE]; /* DEVNAME-qN-rx */
112

113 114 115
	spinlock_t   tx_lock;
	struct xen_netif_tx_front_ring tx;
	int tx_ring_ref;
116 117 118 119 120 121 122 123 124 125 126 127

	/*
	 * {tx,rx}_skbs store outstanding skbuffs. Free tx_skb entries
	 * are linked from tx_skb_freelist through skb_entry.link.
	 *
	 *  NB. Freelist index entries are always going to be less than
	 *  PAGE_OFFSET, whereas pointers to skbs will always be equal or
	 *  greater than PAGE_OFFSET: we use this property to distinguish
	 *  them.
	 */
	union skb_entry {
		struct sk_buff *skb;
128
		unsigned long link;
129 130 131
	} tx_skbs[NET_TX_RING_SIZE];
	grant_ref_t gref_tx_head;
	grant_ref_t grant_tx_ref[NET_TX_RING_SIZE];
132
	struct page *grant_tx_page[NET_TX_RING_SIZE];
133 134
	unsigned tx_skb_freelist;

135 136 137 138 139 140
	spinlock_t   rx_lock ____cacheline_aligned_in_smp;
	struct xen_netif_rx_front_ring rx;
	int rx_ring_ref;

	struct timer_list rx_refill_timer;

141 142 143
	struct sk_buff *rx_skbs[NET_RX_RING_SIZE];
	grant_ref_t gref_rx_head;
	grant_ref_t grant_rx_ref[NET_RX_RING_SIZE];
144 145 146 147 148 149 150 151 152 153
};

struct netfront_info {
	struct list_head list;
	struct net_device *netdev;

	struct xenbus_device *xbdev;

	/* Multi-queue support */
	struct netfront_queue *queues;
154 155

	/* Statistics */
156 157
	struct netfront_stats __percpu *rx_stats;
	struct netfront_stats __percpu *tx_stats;
158

159
	atomic_t rx_gso_checksum_fixup;
160 161 162 163 164 165 166
};

struct netfront_rx_info {
	struct xen_netif_rx_response rx;
	struct xen_netif_extra_info extras[XEN_NETIF_EXTRA_TYPE_MAX - 1];
};

167 168 169 170 171 172 173 174
static void skb_entry_set_link(union skb_entry *list, unsigned short id)
{
	list->link = id;
}

static int skb_entry_is_link(const union skb_entry *list)
{
	BUILD_BUG_ON(sizeof(list->skb) != sizeof(list->link));
175
	return (unsigned long)list->skb < PAGE_OFFSET;
176 177
}

178 179 180 181 182 183 184
/*
 * Access macros for acquiring freeing slots in tx_skbs[].
 */

static void add_id_to_freelist(unsigned *head, union skb_entry *list,
			       unsigned short id)
{
185
	skb_entry_set_link(&list[id], *head);
186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201
	*head = id;
}

static unsigned short get_id_from_freelist(unsigned *head,
					   union skb_entry *list)
{
	unsigned int id = *head;
	*head = list[id].link;
	return id;
}

static int xennet_rxidx(RING_IDX idx)
{
	return idx & (NET_RX_RING_SIZE - 1);
}

202
static struct sk_buff *xennet_get_rx_skb(struct netfront_queue *queue,
203 204 205
					 RING_IDX ri)
{
	int i = xennet_rxidx(ri);
206 207
	struct sk_buff *skb = queue->rx_skbs[i];
	queue->rx_skbs[i] = NULL;
208 209 210
	return skb;
}

211
static grant_ref_t xennet_get_rx_ref(struct netfront_queue *queue,
212 213 214
					    RING_IDX ri)
{
	int i = xennet_rxidx(ri);
215 216
	grant_ref_t ref = queue->grant_rx_ref[i];
	queue->grant_rx_ref[i] = GRANT_INVALID_REF;
217 218 219 220
	return ref;
}

#ifdef CONFIG_SYSFS
221
static const struct attribute_group xennet_dev_group;
222 223
#endif

224
static bool xennet_can_sg(struct net_device *dev)
225
{
226
	return dev->features & NETIF_F_SG;
227 228 229 230 231
}


static void rx_refill_timeout(unsigned long data)
{
232 233
	struct netfront_queue *queue = (struct netfront_queue *)data;
	napi_schedule(&queue->napi);
234 235
}

236
static int netfront_tx_slot_available(struct netfront_queue *queue)
237
{
238
	return (queue->tx.req_prod_pvt - queue->tx.rsp_cons) <
239
		(NET_TX_RING_SIZE - MAX_SKB_FRAGS - 2);
240 241
}

242
static void xennet_maybe_wake_tx(struct netfront_queue *queue)
243
{
244 245
	struct net_device *dev = queue->info->netdev;
	struct netdev_queue *dev_queue = netdev_get_tx_queue(dev, queue->id);
246

247 248
	if (unlikely(netif_tx_queue_stopped(dev_queue)) &&
	    netfront_tx_slot_available(queue) &&
249
	    likely(netif_running(dev)))
250
		netif_tx_wake_queue(netdev_get_tx_queue(dev, queue->id));
251 252
}

253 254

static struct sk_buff *xennet_alloc_one_rx_buffer(struct netfront_queue *queue)
255 256 257 258
{
	struct sk_buff *skb;
	struct page *page;

259 260 261 262 263
	skb = __netdev_alloc_skb(queue->info->netdev,
				 RX_COPY_THRESHOLD + NET_IP_ALIGN,
				 GFP_ATOMIC | __GFP_NOWARN);
	if (unlikely(!skb))
		return NULL;
264

265 266 267 268
	page = alloc_page(GFP_ATOMIC | __GFP_NOWARN);
	if (!page) {
		kfree_skb(skb);
		return NULL;
269
	}
270 271 272 273 274 275 276 277
	skb_add_rx_frag(skb, 0, page, 0, 0, PAGE_SIZE);

	/* Align ip header to a 16 bytes boundary */
	skb_reserve(skb, NET_IP_ALIGN);
	skb->dev = queue->info->netdev;

	return skb;
}
278

279 280 281 282 283 284 285

static void xennet_alloc_rx_buffers(struct netfront_queue *queue)
{
	RING_IDX req_prod = queue->rx.req_prod_pvt;
	int notify;

	if (unlikely(!netif_carrier_ok(queue->info->netdev)))
286 287
		return;

288 289 290 291 292 293
	for (req_prod = queue->rx.req_prod_pvt;
	     req_prod - queue->rx.rsp_cons < NET_RX_RING_SIZE;
	     req_prod++) {
		struct sk_buff *skb;
		unsigned short id;
		grant_ref_t ref;
294
		struct page *page;
295
		struct xen_netif_rx_request *req;
296

297 298
		skb = xennet_alloc_one_rx_buffer(queue);
		if (!skb)
299 300
			break;

301
		id = xennet_rxidx(req_prod);
302

303 304
		BUG_ON(queue->rx_skbs[id]);
		queue->rx_skbs[id] = skb;
305

306
		ref = gnttab_claim_grant_reference(&queue->gref_rx_head);
307
		WARN_ON_ONCE(IS_ERR_VALUE((unsigned long)(int)ref));
308
		queue->grant_rx_ref[id] = ref;
309

310
		page = skb_frag_page(&skb_shinfo(skb)->frags[0]);
311

312
		req = RING_GET_REQUEST(&queue->rx, req_prod);
313 314 315 316
		gnttab_page_grant_foreign_access_ref_one(ref,
							 queue->info->xbdev->otherend_id,
							 page,
							 0);
317 318 319 320
		req->id = id;
		req->gref = ref;
	}

321 322 323
	queue->rx.req_prod_pvt = req_prod;

	/* Not enough requests? Try again later. */
324
	if (req_prod - queue->rx.sring->req_prod < NET_RX_SLOTS_MIN) {
325 326 327 328
		mod_timer(&queue->rx_refill_timer, jiffies + (HZ/10));
		return;
	}

329
	wmb();		/* barrier so backend seens requests */
330

331
	RING_PUSH_REQUESTS_AND_CHECK_NOTIFY(&queue->rx, notify);
332
	if (notify)
333
		notify_remote_via_irq(queue->rx_irq);
334 335 336 337 338
}

static int xennet_open(struct net_device *dev)
{
	struct netfront_info *np = netdev_priv(dev);
339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354
	unsigned int num_queues = dev->real_num_tx_queues;
	unsigned int i = 0;
	struct netfront_queue *queue = NULL;

	for (i = 0; i < num_queues; ++i) {
		queue = &np->queues[i];
		napi_enable(&queue->napi);

		spin_lock_bh(&queue->rx_lock);
		if (netif_carrier_ok(dev)) {
			xennet_alloc_rx_buffers(queue);
			queue->rx.sring->rsp_event = queue->rx.rsp_cons + 1;
			if (RING_HAS_UNCONSUMED_RESPONSES(&queue->rx))
				napi_schedule(&queue->napi);
		}
		spin_unlock_bh(&queue->rx_lock);
355 356
	}

357
	netif_tx_start_all_queues(dev);
358 359 360 361

	return 0;
}

362
static void xennet_tx_buf_gc(struct netfront_queue *queue)
363 364 365 366
{
	RING_IDX cons, prod;
	unsigned short id;
	struct sk_buff *skb;
367
	bool more_to_do;
368

369
	BUG_ON(!netif_carrier_ok(queue->info->netdev));
370 371

	do {
372
		prod = queue->tx.sring->rsp_prod;
373 374
		rmb(); /* Ensure we see responses up to 'rp'. */

375
		for (cons = queue->tx.rsp_cons; cons != prod; cons++) {
376 377
			struct xen_netif_tx_response *txrsp;

378
			txrsp = RING_GET_RESPONSE(&queue->tx, cons);
I
Ian Campbell 已提交
379
			if (txrsp->status == XEN_NETIF_RSP_NULL)
380 381 382
				continue;

			id  = txrsp->id;
383
			skb = queue->tx_skbs[id].skb;
384
			if (unlikely(gnttab_query_foreign_access(
385
				queue->grant_tx_ref[id]) != 0)) {
386 387
				pr_alert("%s: warning -- grant still in use by backend domain\n",
					 __func__);
388 389 390
				BUG();
			}
			gnttab_end_foreign_access_ref(
391
				queue->grant_tx_ref[id], GNTMAP_readonly);
392
			gnttab_release_grant_reference(
393 394 395 396
				&queue->gref_tx_head, queue->grant_tx_ref[id]);
			queue->grant_tx_ref[id] = GRANT_INVALID_REF;
			queue->grant_tx_page[id] = NULL;
			add_id_to_freelist(&queue->tx_skb_freelist, queue->tx_skbs, id);
397 398 399
			dev_kfree_skb_irq(skb);
		}

400
		queue->tx.rsp_cons = prod;
401

402 403
		RING_FINAL_CHECK_FOR_RESPONSES(&queue->tx, more_to_do);
	} while (more_to_do);
404

405
	xennet_maybe_wake_tx(queue);
406 407
}

408 409 410 411 412 413 414 415 416 417
struct xennet_gnttab_make_txreq {
	struct netfront_queue *queue;
	struct sk_buff *skb;
	struct page *page;
	struct xen_netif_tx_request *tx; /* Last request */
	unsigned int size;
};

static void xennet_tx_setup_grant(unsigned long gfn, unsigned int offset,
				  unsigned int len, void *data)
418
{
419
	struct xennet_gnttab_make_txreq *info = data;
420
	unsigned int id;
421
	struct xen_netif_tx_request *tx;
422
	grant_ref_t ref;
423 424 425 426
	/* convenient aliases */
	struct page *page = info->page;
	struct netfront_queue *queue = info->queue;
	struct sk_buff *skb = info->skb;
427

428 429 430
	id = get_id_from_freelist(&queue->tx_skb_freelist, queue->tx_skbs);
	tx = RING_GET_REQUEST(&queue->tx, queue->tx.req_prod_pvt++);
	ref = gnttab_claim_grant_reference(&queue->gref_tx_head);
431
	WARN_ON_ONCE(IS_ERR_VALUE((unsigned long)(int)ref));
432

433 434
	gnttab_grant_foreign_access_ref(ref, queue->info->xbdev->otherend_id,
					gfn, GNTMAP_readonly);
435

436 437 438
	queue->tx_skbs[id].skb = skb;
	queue->grant_tx_page[id] = page;
	queue->grant_tx_ref[id] = ref;
439

440 441 442 443 444
	tx->id = id;
	tx->gref = ref;
	tx->offset = offset;
	tx->size = len;
	tx->flags = 0;
445

446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473
	info->tx = tx;
	info->size += tx->size;
}

static struct xen_netif_tx_request *xennet_make_first_txreq(
	struct netfront_queue *queue, struct sk_buff *skb,
	struct page *page, unsigned int offset, unsigned int len)
{
	struct xennet_gnttab_make_txreq info = {
		.queue = queue,
		.skb = skb,
		.page = page,
		.size = 0,
	};

	gnttab_for_one_grant(page, offset, len, xennet_tx_setup_grant, &info);

	return info.tx;
}

static void xennet_make_one_txreq(unsigned long gfn, unsigned int offset,
				  unsigned int len, void *data)
{
	struct xennet_gnttab_make_txreq *info = data;

	info->tx->flags |= XEN_NETTXF_more_data;
	skb_get(info->skb);
	xennet_tx_setup_grant(gfn, offset, len, data);
474
}
475

476 477 478 479 480
static struct xen_netif_tx_request *xennet_make_txreqs(
	struct netfront_queue *queue, struct xen_netif_tx_request *tx,
	struct sk_buff *skb, struct page *page,
	unsigned int offset, unsigned int len)
{
481 482 483 484 485 486
	struct xennet_gnttab_make_txreq info = {
		.queue = queue,
		.skb = skb,
		.tx = tx,
	};

487 488 489
	/* Skip unused frames from start of page */
	page += offset >> PAGE_SHIFT;
	offset &= ~PAGE_MASK;
490

491
	while (len) {
492 493 494 495 496 497 498
		info.page = page;
		info.size = 0;

		gnttab_foreach_grant_in_range(page, offset, len,
					      xennet_make_one_txreq,
					      &info);

499 500
		page++;
		offset = 0;
501
		len -= info.size;
502 503
	}

504
	return info.tx;
505 506
}

507
/*
508 509
 * Count how many ring slots are required to send this skb. Each frag
 * might be a compound page.
510
 */
511
static int xennet_count_skb_slots(struct sk_buff *skb)
512 513
{
	int i, frags = skb_shinfo(skb)->nr_frags;
514
	int slots;
515

516 517
	slots = gnttab_count_grant(offset_in_page(skb->data),
				   skb_headlen(skb));
518 519 520 521 522 523 524 525 526

	for (i = 0; i < frags; i++) {
		skb_frag_t *frag = skb_shinfo(skb)->frags + i;
		unsigned long size = skb_frag_size(frag);
		unsigned long offset = frag->page_offset;

		/* Skip unused frames from start of page */
		offset &= ~PAGE_MASK;

527
		slots += gnttab_count_grant(offset, size);
528 529
	}

530
	return slots;
531 532
}

533 534
static u16 xennet_select_queue(struct net_device *dev, struct sk_buff *skb,
			       void *accel_priv, select_queue_fallback_t fallback)
535
{
536 537 538 539 540 541 542 543 544 545 546 547 548
	unsigned int num_queues = dev->real_num_tx_queues;
	u32 hash;
	u16 queue_idx;

	/* First, check if there is only one queue */
	if (num_queues == 1) {
		queue_idx = 0;
	} else {
		hash = skb_get_hash(skb);
		queue_idx = hash % num_queues;
	}

	return queue_idx;
549 550
}

551 552
#define MAX_XEN_SKB_FRAGS (65536 / XEN_PAGE_SIZE + 1)

553 554 555
static int xennet_start_xmit(struct sk_buff *skb, struct net_device *dev)
{
	struct netfront_info *np = netdev_priv(dev);
556
	struct netfront_stats *tx_stats = this_cpu_ptr(np->tx_stats);
557 558
	struct xen_netif_tx_request *tx, *first_tx;
	unsigned int i;
559
	int notify;
560
	int slots;
561 562 563
	struct page *page;
	unsigned int offset;
	unsigned int len;
564
	unsigned long flags;
565 566 567
	struct netfront_queue *queue = NULL;
	unsigned int num_queues = dev->real_num_tx_queues;
	u16 queue_index;
568
	struct sk_buff *nskb;
569 570 571 572 573 574 575

	/* Drop the packet if no queues are set up */
	if (num_queues < 1)
		goto drop;
	/* Determine which queue to transmit this SKB on */
	queue_index = skb_get_queue_mapping(skb);
	queue = &np->queues[queue_index];
576

577 578 579 580 581 582 583 584 585 586
	/* If skb->len is too big for wire format, drop skb and alert
	 * user about misconfiguration.
	 */
	if (unlikely(skb->len > XEN_NETIF_MAX_TX_SIZE)) {
		net_alert_ratelimited(
			"xennet: skb->len = %u, too big for wire format\n",
			skb->len);
		goto drop;
	}

587
	slots = xennet_count_skb_slots(skb);
588
	if (unlikely(slots > MAX_XEN_SKB_FRAGS + 1)) {
589 590 591 592
		net_dbg_ratelimited("xennet: skb rides the rocket: %d slots, %d bytes\n",
				    slots, skb->len);
		if (skb_linearize(skb))
			goto drop;
593 594
	}

595 596
	page = virt_to_page(skb->data);
	offset = offset_in_page(skb->data);
597 598 599 600 601 602 603 604 605 606 607 608 609 610

	/* The first req should be at least ETH_HLEN size or the packet will be
	 * dropped by netback.
	 */
	if (unlikely(PAGE_SIZE - offset < ETH_HLEN)) {
		nskb = skb_copy(skb, GFP_ATOMIC);
		if (!nskb)
			goto drop;
		dev_kfree_skb_any(skb);
		skb = nskb;
		page = virt_to_page(skb->data);
		offset = offset_in_page(skb->data);
	}

611 612
	len = skb_headlen(skb);

613
	spin_lock_irqsave(&queue->tx_lock, flags);
614 615

	if (unlikely(!netif_carrier_ok(dev) ||
616
		     (slots > 1 && !xennet_can_sg(dev)) ||
617
		     netif_needs_gso(skb, netif_skb_features(skb)))) {
618
		spin_unlock_irqrestore(&queue->tx_lock, flags);
619 620 621
		goto drop;
	}

622
	/* First request for the linear area. */
623 624 625 626 627 628 629
	first_tx = tx = xennet_make_first_txreq(queue, skb,
						page, offset, len);
	offset += tx->size;
	if (offset == PAGE_SIZE) {
		page++;
		offset = 0;
	}
630
	len -= tx->size;
631 632 633

	if (skb->ip_summed == CHECKSUM_PARTIAL)
		/* local packet? */
I
Ian Campbell 已提交
634
		tx->flags |= XEN_NETTXF_csum_blank | XEN_NETTXF_data_validated;
635 636
	else if (skb->ip_summed == CHECKSUM_UNNECESSARY)
		/* remote but checksummed. */
I
Ian Campbell 已提交
637
		tx->flags |= XEN_NETTXF_data_validated;
638

639
	/* Optional extra info after the first request. */
640 641 642 643
	if (skb_shinfo(skb)->gso_size) {
		struct xen_netif_extra_info *gso;

		gso = (struct xen_netif_extra_info *)
644
			RING_GET_REQUEST(&queue->tx, queue->tx.req_prod_pvt++);
645

646
		tx->flags |= XEN_NETTXF_extra_info;
647 648

		gso->u.gso.size = skb_shinfo(skb)->gso_size;
649 650 651
		gso->u.gso.type = (skb_shinfo(skb)->gso_type & SKB_GSO_TCPV6) ?
			XEN_NETIF_GSO_TYPE_TCPV6 :
			XEN_NETIF_GSO_TYPE_TCPV4;
652 653 654 655 656 657 658
		gso->u.gso.pad = 0;
		gso->u.gso.features = 0;

		gso->type = XEN_NETIF_EXTRA_TYPE_GSO;
		gso->flags = 0;
	}

659 660 661 662 663 664 665 666 667 668
	/* Requests for the rest of the linear area. */
	tx = xennet_make_txreqs(queue, tx, skb, page, offset, len);

	/* Requests for all the frags. */
	for (i = 0; i < skb_shinfo(skb)->nr_frags; i++) {
		skb_frag_t *frag = &skb_shinfo(skb)->frags[i];
		tx = xennet_make_txreqs(queue, tx, skb,
					skb_frag_page(frag), frag->page_offset,
					skb_frag_size(frag));
	}
669

670 671
	/* First request has the packet length. */
	first_tx->size = skb->len;
672

673
	RING_PUSH_REQUESTS_AND_CHECK_NOTIFY(&queue->tx, notify);
674
	if (notify)
675
		notify_remote_via_irq(queue->tx_irq);
676

677 678 679 680
	u64_stats_update_begin(&tx_stats->syncp);
	tx_stats->bytes += skb->len;
	tx_stats->packets++;
	u64_stats_update_end(&tx_stats->syncp);
681 682

	/* Note: It is not safe to access skb after xennet_tx_buf_gc()! */
683
	xennet_tx_buf_gc(queue);
684

685 686
	if (!netfront_tx_slot_available(queue))
		netif_tx_stop_queue(netdev_get_tx_queue(dev, queue->id));
687

688
	spin_unlock_irqrestore(&queue->tx_lock, flags);
689

690
	return NETDEV_TX_OK;
691 692

 drop:
693
	dev->stats.tx_dropped++;
694
	dev_kfree_skb_any(skb);
695
	return NETDEV_TX_OK;
696 697 698 699 700
}

static int xennet_close(struct net_device *dev)
{
	struct netfront_info *np = netdev_priv(dev);
701 702 703 704 705 706 707 708
	unsigned int num_queues = dev->real_num_tx_queues;
	unsigned int i;
	struct netfront_queue *queue;
	netif_tx_stop_all_queues(np->netdev);
	for (i = 0; i < num_queues; ++i) {
		queue = &np->queues[i];
		napi_disable(&queue->napi);
	}
709 710 711
	return 0;
}

712
static void xennet_move_rx_slot(struct netfront_queue *queue, struct sk_buff *skb,
713 714
				grant_ref_t ref)
{
715 716 717 718 719 720 721 722
	int new = xennet_rxidx(queue->rx.req_prod_pvt);

	BUG_ON(queue->rx_skbs[new]);
	queue->rx_skbs[new] = skb;
	queue->grant_rx_ref[new] = ref;
	RING_GET_REQUEST(&queue->rx, queue->rx.req_prod_pvt)->id = new;
	RING_GET_REQUEST(&queue->rx, queue->rx.req_prod_pvt)->gref = ref;
	queue->rx.req_prod_pvt++;
723 724
}

725
static int xennet_get_extras(struct netfront_queue *queue,
726 727 728 729 730
			     struct xen_netif_extra_info *extras,
			     RING_IDX rp)

{
	struct xen_netif_extra_info *extra;
731 732
	struct device *dev = &queue->info->netdev->dev;
	RING_IDX cons = queue->rx.rsp_cons;
733 734 735 736 737 738 739 740 741 742 743 744 745 746
	int err = 0;

	do {
		struct sk_buff *skb;
		grant_ref_t ref;

		if (unlikely(cons + 1 == rp)) {
			if (net_ratelimit())
				dev_warn(dev, "Missing extra info\n");
			err = -EBADR;
			break;
		}

		extra = (struct xen_netif_extra_info *)
747
			RING_GET_RESPONSE(&queue->rx, ++cons);
748 749 750 751 752 753 754 755 756 757 758 759

		if (unlikely(!extra->type ||
			     extra->type >= XEN_NETIF_EXTRA_TYPE_MAX)) {
			if (net_ratelimit())
				dev_warn(dev, "Invalid extra type: %d\n",
					extra->type);
			err = -EINVAL;
		} else {
			memcpy(&extras[extra->type - 1], extra,
			       sizeof(*extra));
		}

760 761 762
		skb = xennet_get_rx_skb(queue, cons);
		ref = xennet_get_rx_ref(queue, cons);
		xennet_move_rx_slot(queue, skb, ref);
763 764
	} while (extra->flags & XEN_NETIF_EXTRA_FLAG_MORE);

765
	queue->rx.rsp_cons = cons;
766 767 768
	return err;
}

769
static int xennet_get_responses(struct netfront_queue *queue,
770 771 772 773 774
				struct netfront_rx_info *rinfo, RING_IDX rp,
				struct sk_buff_head *list)
{
	struct xen_netif_rx_response *rx = &rinfo->rx;
	struct xen_netif_extra_info *extras = rinfo->extras;
775 776 777 778
	struct device *dev = &queue->info->netdev->dev;
	RING_IDX cons = queue->rx.rsp_cons;
	struct sk_buff *skb = xennet_get_rx_skb(queue, cons);
	grant_ref_t ref = xennet_get_rx_ref(queue, cons);
779
	int max = MAX_SKB_FRAGS + (rx->status <= RX_COPY_THRESHOLD);
780
	int slots = 1;
781 782 783
	int err = 0;
	unsigned long ret;

I
Ian Campbell 已提交
784
	if (rx->flags & XEN_NETRXF_extra_info) {
785 786
		err = xennet_get_extras(queue, extras, rp);
		cons = queue->rx.rsp_cons;
787 788 789 790
	}

	for (;;) {
		if (unlikely(rx->status < 0 ||
791
			     rx->offset + rx->status > XEN_PAGE_SIZE)) {
792
			if (net_ratelimit())
793
				dev_warn(dev, "rx->offset: %u, size: %d\n",
794
					 rx->offset, rx->status);
795
			xennet_move_rx_slot(queue, skb, ref);
796 797 798 799 800 801 802
			err = -EINVAL;
			goto next;
		}

		/*
		 * This definitely indicates a bug, either in this driver or in
		 * the backend driver. In future this should flag the bad
803
		 * situation to the system controller to reboot the backend.
804 805 806 807 808 809 810 811 812 813 814 815
		 */
		if (ref == GRANT_INVALID_REF) {
			if (net_ratelimit())
				dev_warn(dev, "Bad rx response id %d.\n",
					 rx->id);
			err = -EINVAL;
			goto next;
		}

		ret = gnttab_end_foreign_access_ref(ref, 0);
		BUG_ON(!ret);

816
		gnttab_release_grant_reference(&queue->gref_rx_head, ref);
817 818 819 820

		__skb_queue_tail(list, skb);

next:
I
Ian Campbell 已提交
821
		if (!(rx->flags & XEN_NETRXF_more_data))
822 823
			break;

824
		if (cons + slots == rp) {
825
			if (net_ratelimit())
826
				dev_warn(dev, "Need more slots\n");
827 828 829 830
			err = -ENOENT;
			break;
		}

831 832 833
		rx = RING_GET_RESPONSE(&queue->rx, cons + slots);
		skb = xennet_get_rx_skb(queue, cons + slots);
		ref = xennet_get_rx_ref(queue, cons + slots);
834
		slots++;
835 836
	}

837
	if (unlikely(slots > max)) {
838
		if (net_ratelimit())
839
			dev_warn(dev, "Too many slots\n");
840 841 842 843
		err = -E2BIG;
	}

	if (unlikely(err))
844
		queue->rx.rsp_cons = cons + slots;
845 846 847 848 849 850 851 852 853

	return err;
}

static int xennet_set_skb_gso(struct sk_buff *skb,
			      struct xen_netif_extra_info *gso)
{
	if (!gso->u.gso.size) {
		if (net_ratelimit())
854
			pr_warn("GSO size must not be zero\n");
855 856 857
		return -EINVAL;
	}

858 859
	if (gso->u.gso.type != XEN_NETIF_GSO_TYPE_TCPV4 &&
	    gso->u.gso.type != XEN_NETIF_GSO_TYPE_TCPV6) {
860
		if (net_ratelimit())
861
			pr_warn("Bad GSO type %d\n", gso->u.gso.type);
862 863 864 865
		return -EINVAL;
	}

	skb_shinfo(skb)->gso_size = gso->u.gso.size;
866 867 868 869
	skb_shinfo(skb)->gso_type =
		(gso->u.gso.type == XEN_NETIF_GSO_TYPE_TCPV4) ?
		SKB_GSO_TCPV4 :
		SKB_GSO_TCPV6;
870 871 872 873 874 875 876 877

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

	return 0;
}

878
static RING_IDX xennet_fill_frags(struct netfront_queue *queue,
879 880 881 882
				  struct sk_buff *skb,
				  struct sk_buff_head *list)
{
	struct skb_shared_info *shinfo = skb_shinfo(skb);
883
	RING_IDX cons = queue->rx.rsp_cons;
884 885 886 887
	struct sk_buff *nskb;

	while ((nskb = __skb_dequeue(list))) {
		struct xen_netif_rx_response *rx =
888
			RING_GET_RESPONSE(&queue->rx, ++cons);
889
		skb_frag_t *nfrag = &skb_shinfo(nskb)->frags[0];
890

891 892
		if (shinfo->nr_frags == MAX_SKB_FRAGS) {
			unsigned int pull_to = NETFRONT_SKB_CB(skb)->pull_to;
893

894 895 896 897 898 899 900
			BUG_ON(pull_to <= skb_headlen(skb));
			__pskb_pull_tail(skb, pull_to - skb_headlen(skb));
		}
		BUG_ON(shinfo->nr_frags >= MAX_SKB_FRAGS);

		skb_add_rx_frag(skb, shinfo->nr_frags, skb_frag_page(nfrag),
				rx->offset, rx->status, PAGE_SIZE);
901 902 903 904 905 906 907 908

		skb_shinfo(nskb)->nr_frags = 0;
		kfree_skb(nskb);
	}

	return cons;
}

909
static int checksum_setup(struct net_device *dev, struct sk_buff *skb)
910
{
911
	bool recalculate_partial_csum = false;
912 913 914 915 916 917 918 919 920

	/*
	 * A GSO SKB must be CHECKSUM_PARTIAL. However some buggy
	 * peers can fail to set NETRXF_csum_blank when sending a GSO
	 * frame. In this case force the SKB to CHECKSUM_PARTIAL and
	 * recalculate the partial checksum.
	 */
	if (skb->ip_summed != CHECKSUM_PARTIAL && skb_is_gso(skb)) {
		struct netfront_info *np = netdev_priv(dev);
921
		atomic_inc(&np->rx_gso_checksum_fixup);
922
		skb->ip_summed = CHECKSUM_PARTIAL;
923
		recalculate_partial_csum = true;
924 925 926 927 928
	}

	/* A non-CHECKSUM_PARTIAL SKB does not require setup. */
	if (skb->ip_summed != CHECKSUM_PARTIAL)
		return 0;
929

930
	return skb_checksum_setup(skb, recalculate_partial_csum);
931 932
}

933
static int handle_incoming_queue(struct netfront_queue *queue,
934
				 struct sk_buff_head *rxq)
935
{
936
	struct netfront_stats *rx_stats = this_cpu_ptr(queue->info->rx_stats);
937 938 939 940
	int packets_dropped = 0;
	struct sk_buff *skb;

	while ((skb = __skb_dequeue(rxq)) != NULL) {
941
		int pull_to = NETFRONT_SKB_CB(skb)->pull_to;
942

943 944
		if (pull_to > skb_headlen(skb))
			__pskb_pull_tail(skb, pull_to - skb_headlen(skb));
945 946

		/* Ethernet work: Delayed to here as it peeks the header. */
947
		skb->protocol = eth_type_trans(skb, queue->info->netdev);
948
		skb_reset_network_header(skb);
949

950
		if (checksum_setup(queue->info->netdev, skb)) {
951 952
			kfree_skb(skb);
			packets_dropped++;
953
			queue->info->netdev->stats.rx_errors++;
954
			continue;
955 956
		}

957 958 959 960
		u64_stats_update_begin(&rx_stats->syncp);
		rx_stats->packets++;
		rx_stats->bytes += skb->len;
		u64_stats_update_end(&rx_stats->syncp);
961 962

		/* Pass it up. */
963
		napi_gro_receive(&queue->napi, skb);
964 965 966 967 968
	}

	return packets_dropped;
}

969
static int xennet_poll(struct napi_struct *napi, int budget)
970
{
971 972
	struct netfront_queue *queue = container_of(napi, struct netfront_queue, napi);
	struct net_device *dev = queue->info->netdev;
973 974 975 976 977
	struct sk_buff *skb;
	struct netfront_rx_info rinfo;
	struct xen_netif_rx_response *rx = &rinfo.rx;
	struct xen_netif_extra_info *extras = rinfo.extras;
	RING_IDX i, rp;
978
	int work_done;
979 980 981 982 983
	struct sk_buff_head rxq;
	struct sk_buff_head errq;
	struct sk_buff_head tmpq;
	int err;

984
	spin_lock(&queue->rx_lock);
985 986 987 988 989

	skb_queue_head_init(&rxq);
	skb_queue_head_init(&errq);
	skb_queue_head_init(&tmpq);

990
	rp = queue->rx.sring->rsp_prod;
991 992
	rmb(); /* Ensure we see queued responses up to 'rp'. */

993
	i = queue->rx.rsp_cons;
994 995
	work_done = 0;
	while ((i != rp) && (work_done < budget)) {
996
		memcpy(rx, RING_GET_RESPONSE(&queue->rx, i), sizeof(*rx));
997 998
		memset(extras, 0, sizeof(rinfo.extras));

999
		err = xennet_get_responses(queue, &rinfo, rp, &tmpq);
1000 1001 1002 1003 1004

		if (unlikely(err)) {
err:
			while ((skb = __skb_dequeue(&tmpq)))
				__skb_queue_tail(&errq, skb);
1005
			dev->stats.rx_errors++;
1006
			i = queue->rx.rsp_cons;
1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017
			continue;
		}

		skb = __skb_dequeue(&tmpq);

		if (extras[XEN_NETIF_EXTRA_TYPE_GSO - 1].type) {
			struct xen_netif_extra_info *gso;
			gso = &extras[XEN_NETIF_EXTRA_TYPE_GSO - 1];

			if (unlikely(xennet_set_skb_gso(skb, gso))) {
				__skb_queue_head(&tmpq, skb);
1018
				queue->rx.rsp_cons += skb_queue_len(&tmpq);
1019 1020 1021 1022
				goto err;
			}
		}

1023 1024 1025
		NETFRONT_SKB_CB(skb)->pull_to = rx->status;
		if (NETFRONT_SKB_CB(skb)->pull_to > RX_COPY_THRESHOLD)
			NETFRONT_SKB_CB(skb)->pull_to = RX_COPY_THRESHOLD;
1026

1027 1028 1029
		skb_shinfo(skb)->frags[0].page_offset = rx->offset;
		skb_frag_size_set(&skb_shinfo(skb)->frags[0], rx->status);
		skb->data_len = rx->status;
1030
		skb->len += rx->status;
1031

1032
		i = xennet_fill_frags(queue, skb, &tmpq);
1033

I
Ian Campbell 已提交
1034
		if (rx->flags & XEN_NETRXF_csum_blank)
1035
			skb->ip_summed = CHECKSUM_PARTIAL;
I
Ian Campbell 已提交
1036
		else if (rx->flags & XEN_NETRXF_data_validated)
1037 1038 1039 1040
			skb->ip_summed = CHECKSUM_UNNECESSARY;

		__skb_queue_tail(&rxq, skb);

1041
		queue->rx.rsp_cons = ++i;
1042 1043 1044
		work_done++;
	}

W
Wang Chen 已提交
1045
	__skb_queue_purge(&errq);
1046

1047
	work_done -= handle_incoming_queue(queue, &rxq);
1048

1049
	xennet_alloc_rx_buffers(queue);
1050 1051

	if (work_done < budget) {
1052 1053
		int more_to_do = 0;

1054
		napi_complete(napi);
1055

1056
		RING_FINAL_CHECK_FOR_RESPONSES(&queue->rx, more_to_do);
1057 1058
		if (more_to_do)
			napi_schedule(napi);
1059 1060
	}

1061
	spin_unlock(&queue->rx_lock);
1062

1063
	return work_done;
1064 1065 1066 1067
}

static int xennet_change_mtu(struct net_device *dev, int mtu)
{
1068
	int max = xennet_can_sg(dev) ? XEN_NETIF_MAX_TX_SIZE : ETH_DATA_LEN;
1069 1070 1071 1072 1073 1074 1075

	if (mtu > max)
		return -EINVAL;
	dev->mtu = mtu;
	return 0;
}

1076 1077 1078 1079 1080 1081 1082
static struct rtnl_link_stats64 *xennet_get_stats64(struct net_device *dev,
						    struct rtnl_link_stats64 *tot)
{
	struct netfront_info *np = netdev_priv(dev);
	int cpu;

	for_each_possible_cpu(cpu) {
1083 1084
		struct netfront_stats *rx_stats = per_cpu_ptr(np->rx_stats, cpu);
		struct netfront_stats *tx_stats = per_cpu_ptr(np->tx_stats, cpu);
1085 1086 1087 1088
		u64 rx_packets, rx_bytes, tx_packets, tx_bytes;
		unsigned int start;

		do {
1089 1090 1091 1092
			start = u64_stats_fetch_begin_irq(&tx_stats->syncp);
			tx_packets = tx_stats->packets;
			tx_bytes = tx_stats->bytes;
		} while (u64_stats_fetch_retry_irq(&tx_stats->syncp, start));
1093

1094 1095 1096 1097 1098
		do {
			start = u64_stats_fetch_begin_irq(&rx_stats->syncp);
			rx_packets = rx_stats->packets;
			rx_bytes = rx_stats->bytes;
		} while (u64_stats_fetch_retry_irq(&rx_stats->syncp, start));
1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111

		tot->rx_packets += rx_packets;
		tot->tx_packets += tx_packets;
		tot->rx_bytes   += rx_bytes;
		tot->tx_bytes   += tx_bytes;
	}

	tot->rx_errors  = dev->stats.rx_errors;
	tot->tx_dropped = dev->stats.tx_dropped;

	return tot;
}

1112
static void xennet_release_tx_bufs(struct netfront_queue *queue)
1113 1114 1115 1116 1117 1118
{
	struct sk_buff *skb;
	int i;

	for (i = 0; i < NET_TX_RING_SIZE; i++) {
		/* Skip over entries which are actually freelist references */
1119
		if (skb_entry_is_link(&queue->tx_skbs[i]))
1120 1121
			continue;

1122 1123 1124
		skb = queue->tx_skbs[i].skb;
		get_page(queue->grant_tx_page[i]);
		gnttab_end_foreign_access(queue->grant_tx_ref[i],
1125
					  GNTMAP_readonly,
1126 1127 1128 1129
					  (unsigned long)page_address(queue->grant_tx_page[i]));
		queue->grant_tx_page[i] = NULL;
		queue->grant_tx_ref[i] = GRANT_INVALID_REF;
		add_id_to_freelist(&queue->tx_skb_freelist, queue->tx_skbs, i);
1130 1131 1132 1133
		dev_kfree_skb_irq(skb);
	}
}

1134
static void xennet_release_rx_bufs(struct netfront_queue *queue)
1135 1136 1137
{
	int id, ref;

1138
	spin_lock_bh(&queue->rx_lock);
1139 1140

	for (id = 0; id < NET_RX_RING_SIZE; id++) {
1141 1142
		struct sk_buff *skb;
		struct page *page;
1143

1144
		skb = queue->rx_skbs[id];
1145
		if (!skb)
1146 1147
			continue;

1148
		ref = queue->grant_rx_ref[id];
1149 1150
		if (ref == GRANT_INVALID_REF)
			continue;
1151

1152
		page = skb_frag_page(&skb_shinfo(skb)->frags[0]);
1153

1154 1155 1156 1157 1158 1159
		/* gnttab_end_foreign_access() needs a page ref until
		 * foreign access is ended (which may be deferred).
		 */
		get_page(page);
		gnttab_end_foreign_access(ref, 0,
					  (unsigned long)page_address(page));
1160
		queue->grant_rx_ref[id] = GRANT_INVALID_REF;
1161

1162
		kfree_skb(skb);
1163 1164
	}

1165
	spin_unlock_bh(&queue->rx_lock);
1166 1167
}

1168 1169
static netdev_features_t xennet_fix_features(struct net_device *dev,
	netdev_features_t features)
1170 1171 1172
{
	struct netfront_info *np = netdev_priv(dev);

1173 1174 1175
	if (features & NETIF_F_SG &&
	    !xenbus_read_unsigned(np->xbdev->otherend, "feature-sg", 0))
		features &= ~NETIF_F_SG;
1176

1177 1178 1179 1180
	if (features & NETIF_F_IPV6_CSUM &&
	    !xenbus_read_unsigned(np->xbdev->otherend,
				  "feature-ipv6-csum-offload", 0))
		features &= ~NETIF_F_IPV6_CSUM;
1181

1182 1183 1184
	if (features & NETIF_F_TSO &&
	    !xenbus_read_unsigned(np->xbdev->otherend, "feature-gso-tcpv4", 0))
		features &= ~NETIF_F_TSO;
1185

1186 1187 1188
	if (features & NETIF_F_TSO6 &&
	    !xenbus_read_unsigned(np->xbdev->otherend, "feature-gso-tcpv6", 0))
		features &= ~NETIF_F_TSO6;
1189

1190 1191 1192
	return features;
}

1193 1194
static int xennet_set_features(struct net_device *dev,
	netdev_features_t features)
1195 1196 1197 1198 1199 1200 1201 1202 1203
{
	if (!(features & NETIF_F_SG) && dev->mtu > ETH_DATA_LEN) {
		netdev_info(dev, "Reducing MTU because no SG offload");
		dev->mtu = ETH_DATA_LEN;
	}

	return 0;
}

1204
static irqreturn_t xennet_tx_interrupt(int irq, void *dev_id)
1205
{
1206
	struct netfront_queue *queue = dev_id;
1207 1208
	unsigned long flags;

1209 1210 1211
	spin_lock_irqsave(&queue->tx_lock, flags);
	xennet_tx_buf_gc(queue);
	spin_unlock_irqrestore(&queue->tx_lock, flags);
1212

1213 1214 1215 1216 1217
	return IRQ_HANDLED;
}

static irqreturn_t xennet_rx_interrupt(int irq, void *dev_id)
{
1218 1219
	struct netfront_queue *queue = dev_id;
	struct net_device *dev = queue->info->netdev;
1220 1221

	if (likely(netif_carrier_ok(dev) &&
1222
		   RING_HAS_UNCONSUMED_RESPONSES(&queue->rx)))
1223
		napi_schedule(&queue->napi);
1224

1225 1226
	return IRQ_HANDLED;
}
1227

1228 1229 1230 1231
static irqreturn_t xennet_interrupt(int irq, void *dev_id)
{
	xennet_tx_interrupt(irq, dev_id);
	xennet_rx_interrupt(irq, dev_id);
1232 1233 1234 1235 1236 1237
	return IRQ_HANDLED;
}

#ifdef CONFIG_NET_POLL_CONTROLLER
static void xennet_poll_controller(struct net_device *dev)
{
1238 1239 1240 1241 1242 1243
	/* Poll each queue */
	struct netfront_info *info = netdev_priv(dev);
	unsigned int num_queues = dev->real_num_tx_queues;
	unsigned int i;
	for (i = 0; i < num_queues; ++i)
		xennet_interrupt(0, &info->queues[i]);
1244 1245 1246
}
#endif

1247 1248 1249 1250 1251
static const struct net_device_ops xennet_netdev_ops = {
	.ndo_open            = xennet_open,
	.ndo_stop            = xennet_close,
	.ndo_start_xmit      = xennet_start_xmit,
	.ndo_change_mtu	     = xennet_change_mtu,
1252
	.ndo_get_stats64     = xennet_get_stats64,
1253 1254
	.ndo_set_mac_address = eth_mac_addr,
	.ndo_validate_addr   = eth_validate_addr,
1255 1256
	.ndo_fix_features    = xennet_fix_features,
	.ndo_set_features    = xennet_set_features,
1257
	.ndo_select_queue    = xennet_select_queue,
1258 1259 1260
#ifdef CONFIG_NET_POLL_CONTROLLER
	.ndo_poll_controller = xennet_poll_controller,
#endif
1261 1262
};

1263 1264 1265 1266 1267 1268 1269 1270 1271
static void xennet_free_netdev(struct net_device *netdev)
{
	struct netfront_info *np = netdev_priv(netdev);

	free_percpu(np->rx_stats);
	free_percpu(np->tx_stats);
	free_netdev(netdev);
}

1272
static struct net_device *xennet_create_dev(struct xenbus_device *dev)
1273
{
1274
	int err;
1275 1276 1277
	struct net_device *netdev;
	struct netfront_info *np;

1278
	netdev = alloc_etherdev_mq(sizeof(struct netfront_info), xennet_max_queues);
1279
	if (!netdev)
1280 1281 1282 1283 1284
		return ERR_PTR(-ENOMEM);

	np                   = netdev_priv(netdev);
	np->xbdev            = dev;

1285
	np->queues = NULL;
1286

1287
	err = -ENOMEM;
1288 1289 1290 1291 1292
	np->rx_stats = netdev_alloc_pcpu_stats(struct netfront_stats);
	if (np->rx_stats == NULL)
		goto exit;
	np->tx_stats = netdev_alloc_pcpu_stats(struct netfront_stats);
	if (np->tx_stats == NULL)
1293 1294
		goto exit;

1295 1296
	netdev->netdev_ops	= &xennet_netdev_ops;

1297 1298
	netdev->features        = NETIF_F_IP_CSUM | NETIF_F_RXCSUM |
				  NETIF_F_GSO_ROBUST;
1299 1300 1301
	netdev->hw_features	= NETIF_F_SG |
				  NETIF_F_IPV6_CSUM |
				  NETIF_F_TSO | NETIF_F_TSO6;
1302

1303 1304 1305 1306 1307 1308 1309 1310
	/*
         * Assume that all hw features are available for now. This set
         * will be adjusted by the call to netdev_update_features() in
         * xennet_connect() which is the earliest point where we can
         * negotiate with the backend regarding supported features.
         */
	netdev->features |= netdev->hw_features;

1311
	netdev->ethtool_ops = &xennet_ethtool_ops;
1312 1313
	netdev->min_mtu = 0;
	netdev->max_mtu = XEN_NETIF_MAX_TX_SIZE;
1314 1315 1316 1317 1318 1319 1320 1321 1322
	SET_NETDEV_DEV(netdev, &dev->dev);

	np->netdev = netdev;

	netif_carrier_off(netdev);

	return netdev;

 exit:
1323
	xennet_free_netdev(netdev);
1324 1325 1326 1327 1328 1329 1330 1331
	return ERR_PTR(err);
}

/**
 * Entry point to this code when a new device is created.  Allocate the basic
 * structures and the ring buffers for communication with the backend, and
 * inform the backend of the appropriate details for those.
 */
1332
static int netfront_probe(struct xenbus_device *dev,
1333
			  const struct xenbus_device_id *id)
1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346
{
	int err;
	struct net_device *netdev;
	struct netfront_info *info;

	netdev = xennet_create_dev(dev);
	if (IS_ERR(netdev)) {
		err = PTR_ERR(netdev);
		xenbus_dev_fatal(dev, err, "creating netdev");
		return err;
	}

	info = netdev_priv(netdev);
1347
	dev_set_drvdata(&dev->dev, info);
1348 1349 1350
#ifdef CONFIG_SYSFS
	info->netdev->sysfs_groups[0] = &xennet_dev_group;
#endif
1351 1352
	err = register_netdev(info->netdev);
	if (err) {
1353
		pr_warn("%s: register_netdev err=%d\n", __func__, err);
1354 1355 1356 1357 1358 1359
		goto fail;
	}

	return 0;

 fail:
1360
	xennet_free_netdev(netdev);
1361
	dev_set_drvdata(&dev->dev, NULL);
1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373
	return err;
}

static void xennet_end_access(int ref, void *page)
{
	/* This frees the page as a side-effect */
	if (ref != GRANT_INVALID_REF)
		gnttab_end_foreign_access(ref, 0, (unsigned long)page);
}

static void xennet_disconnect_backend(struct netfront_info *info)
{
1374 1375 1376
	unsigned int i = 0;
	unsigned int num_queues = info->netdev->real_num_tx_queues;

1377 1378
	netif_carrier_off(info->netdev);

1379
	for (i = 0; i < num_queues && info->queues; ++i) {
1380 1381
		struct netfront_queue *queue = &info->queues[i];

1382 1383 1384 1385 1386 1387 1388 1389
		if (queue->tx_irq && (queue->tx_irq == queue->rx_irq))
			unbind_from_irqhandler(queue->tx_irq, queue);
		if (queue->tx_irq && (queue->tx_irq != queue->rx_irq)) {
			unbind_from_irqhandler(queue->tx_irq, queue);
			unbind_from_irqhandler(queue->rx_irq, queue);
		}
		queue->tx_evtchn = queue->rx_evtchn = 0;
		queue->tx_irq = queue->rx_irq = 0;
1390

1391 1392
		if (netif_running(info->netdev))
			napi_synchronize(&queue->napi);
1393

1394 1395 1396 1397 1398
		xennet_release_tx_bufs(queue);
		xennet_release_rx_bufs(queue);
		gnttab_free_grant_references(queue->gref_tx_head);
		gnttab_free_grant_references(queue->gref_rx_head);

1399 1400 1401
		/* End access and free the pages */
		xennet_end_access(queue->tx_ring_ref, queue->tx.sring);
		xennet_end_access(queue->rx_ring_ref, queue->rx.sring);
1402

1403 1404 1405 1406 1407
		queue->tx_ring_ref = GRANT_INVALID_REF;
		queue->rx_ring_ref = GRANT_INVALID_REF;
		queue->tx.sring = NULL;
		queue->rx.sring = NULL;
	}
1408 1409 1410 1411 1412 1413 1414 1415 1416 1417
}

/**
 * We are reconnecting to the backend, due to a suspend/resume, or a backend
 * driver restart.  We tear down our netif structure and recreate it, but
 * leave the device-layer structures intact so that this is transparent to the
 * rest of the kernel.
 */
static int netfront_resume(struct xenbus_device *dev)
{
1418
	struct netfront_info *info = dev_get_drvdata(&dev->dev);
1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 1440 1441 1442 1443 1444 1445 1446 1447

	dev_dbg(&dev->dev, "%s\n", dev->nodename);

	xennet_disconnect_backend(info);
	return 0;
}

static int xen_net_read_mac(struct xenbus_device *dev, u8 mac[])
{
	char *s, *e, *macstr;
	int i;

	macstr = s = xenbus_read(XBT_NIL, dev->nodename, "mac", NULL);
	if (IS_ERR(macstr))
		return PTR_ERR(macstr);

	for (i = 0; i < ETH_ALEN; i++) {
		mac[i] = simple_strtoul(s, &e, 16);
		if ((s == e) || (*e != ((i == ETH_ALEN-1) ? '\0' : ':'))) {
			kfree(macstr);
			return -ENOENT;
		}
		s = e+1;
	}

	kfree(macstr);
	return 0;
}

1448
static int setup_netfront_single(struct netfront_queue *queue)
1449 1450 1451
{
	int err;

1452
	err = xenbus_alloc_evtchn(queue->info->xbdev, &queue->tx_evtchn);
1453 1454 1455
	if (err < 0)
		goto fail;

1456
	err = bind_evtchn_to_irqhandler(queue->tx_evtchn,
1457
					xennet_interrupt,
1458
					0, queue->info->netdev->name, queue);
1459 1460
	if (err < 0)
		goto bind_fail;
1461 1462
	queue->rx_evtchn = queue->tx_evtchn;
	queue->rx_irq = queue->tx_irq = err;
1463 1464 1465 1466

	return 0;

bind_fail:
1467 1468
	xenbus_free_evtchn(queue->info->xbdev, queue->tx_evtchn);
	queue->tx_evtchn = 0;
1469 1470 1471 1472
fail:
	return err;
}

1473
static int setup_netfront_split(struct netfront_queue *queue)
1474 1475 1476
{
	int err;

1477
	err = xenbus_alloc_evtchn(queue->info->xbdev, &queue->tx_evtchn);
1478 1479
	if (err < 0)
		goto fail;
1480
	err = xenbus_alloc_evtchn(queue->info->xbdev, &queue->rx_evtchn);
1481 1482 1483
	if (err < 0)
		goto alloc_rx_evtchn_fail;

1484 1485 1486
	snprintf(queue->tx_irq_name, sizeof(queue->tx_irq_name),
		 "%s-tx", queue->name);
	err = bind_evtchn_to_irqhandler(queue->tx_evtchn,
1487
					xennet_tx_interrupt,
1488
					0, queue->tx_irq_name, queue);
1489 1490
	if (err < 0)
		goto bind_tx_fail;
1491
	queue->tx_irq = err;
1492

1493 1494 1495
	snprintf(queue->rx_irq_name, sizeof(queue->rx_irq_name),
		 "%s-rx", queue->name);
	err = bind_evtchn_to_irqhandler(queue->rx_evtchn,
1496
					xennet_rx_interrupt,
1497
					0, queue->rx_irq_name, queue);
1498 1499
	if (err < 0)
		goto bind_rx_fail;
1500
	queue->rx_irq = err;
1501 1502 1503 1504

	return 0;

bind_rx_fail:
1505 1506
	unbind_from_irqhandler(queue->tx_irq, queue);
	queue->tx_irq = 0;
1507
bind_tx_fail:
1508 1509
	xenbus_free_evtchn(queue->info->xbdev, queue->rx_evtchn);
	queue->rx_evtchn = 0;
1510
alloc_rx_evtchn_fail:
1511 1512
	xenbus_free_evtchn(queue->info->xbdev, queue->tx_evtchn);
	queue->tx_evtchn = 0;
1513 1514 1515 1516
fail:
	return err;
}

1517 1518
static int setup_netfront(struct xenbus_device *dev,
			struct netfront_queue *queue, unsigned int feature_split_evtchn)
1519 1520 1521
{
	struct xen_netif_tx_sring *txs;
	struct xen_netif_rx_sring *rxs;
1522
	grant_ref_t gref;
1523 1524
	int err;

1525 1526 1527 1528
	queue->tx_ring_ref = GRANT_INVALID_REF;
	queue->rx_ring_ref = GRANT_INVALID_REF;
	queue->rx.sring = NULL;
	queue->tx.sring = NULL;
1529

1530
	txs = (struct xen_netif_tx_sring *)get_zeroed_page(GFP_NOIO | __GFP_HIGH);
1531 1532 1533 1534 1535 1536
	if (!txs) {
		err = -ENOMEM;
		xenbus_dev_fatal(dev, err, "allocating tx ring page");
		goto fail;
	}
	SHARED_RING_INIT(txs);
1537
	FRONT_RING_INIT(&queue->tx, txs, XEN_PAGE_SIZE);
1538

1539
	err = xenbus_grant_ring(dev, txs, 1, &gref);
1540 1541
	if (err < 0)
		goto grant_tx_ring_fail;
1542
	queue->tx_ring_ref = gref;
1543

1544
	rxs = (struct xen_netif_rx_sring *)get_zeroed_page(GFP_NOIO | __GFP_HIGH);
1545 1546 1547
	if (!rxs) {
		err = -ENOMEM;
		xenbus_dev_fatal(dev, err, "allocating rx ring page");
1548
		goto alloc_rx_ring_fail;
1549 1550
	}
	SHARED_RING_INIT(rxs);
1551
	FRONT_RING_INIT(&queue->rx, rxs, XEN_PAGE_SIZE);
1552

1553
	err = xenbus_grant_ring(dev, rxs, 1, &gref);
1554 1555
	if (err < 0)
		goto grant_rx_ring_fail;
1556
	queue->rx_ring_ref = gref;
1557

1558
	if (feature_split_evtchn)
1559
		err = setup_netfront_split(queue);
1560 1561 1562 1563 1564
	/* setup single event channel if
	 *  a) feature-split-event-channels == 0
	 *  b) feature-split-event-channels == 1 but failed to setup
	 */
	if (!feature_split_evtchn || (feature_split_evtchn && err))
1565
		err = setup_netfront_single(queue);
1566

1567
	if (err)
1568
		goto alloc_evtchn_fail;
1569 1570 1571

	return 0;

1572 1573 1574 1575
	/* If we fail to setup netfront, it is safe to just revoke access to
	 * granted pages because backend is not accessing it at this point.
	 */
alloc_evtchn_fail:
1576
	gnttab_end_foreign_access_ref(queue->rx_ring_ref, 0);
1577 1578 1579
grant_rx_ring_fail:
	free_page((unsigned long)rxs);
alloc_rx_ring_fail:
1580
	gnttab_end_foreign_access_ref(queue->tx_ring_ref, 0);
1581 1582 1583
grant_tx_ring_fail:
	free_page((unsigned long)txs);
fail:
1584 1585 1586
	return err;
}

1587 1588 1589 1590 1591 1592 1593 1594 1595 1596 1597 1598
/* Queue-specific initialisation
 * This used to be done in xennet_create_dev() but must now
 * be run per-queue.
 */
static int xennet_init_queue(struct netfront_queue *queue)
{
	unsigned short i;
	int err = 0;

	spin_lock_init(&queue->tx_lock);
	spin_lock_init(&queue->rx_lock);

V
Vaishali Thakkar 已提交
1599 1600
	setup_timer(&queue->rx_refill_timer, rx_refill_timeout,
		    (unsigned long)queue);
1601

1602 1603 1604
	snprintf(queue->name, sizeof(queue->name), "%s-q%u",
		 queue->info->netdev->name, queue->id);

1605 1606 1607 1608 1609 1610 1611 1612 1613 1614 1615 1616 1617 1618 1619
	/* Initialise tx_skbs as a free chain containing every entry. */
	queue->tx_skb_freelist = 0;
	for (i = 0; i < NET_TX_RING_SIZE; i++) {
		skb_entry_set_link(&queue->tx_skbs[i], i+1);
		queue->grant_tx_ref[i] = GRANT_INVALID_REF;
		queue->grant_tx_page[i] = NULL;
	}

	/* Clear out rx_skbs */
	for (i = 0; i < NET_RX_RING_SIZE; i++) {
		queue->rx_skbs[i] = NULL;
		queue->grant_rx_ref[i] = GRANT_INVALID_REF;
	}

	/* A grant for every tx ring slot */
1620
	if (gnttab_alloc_grant_references(NET_TX_RING_SIZE,
1621 1622 1623 1624 1625 1626 1627
					  &queue->gref_tx_head) < 0) {
		pr_alert("can't alloc tx grant refs\n");
		err = -ENOMEM;
		goto exit;
	}

	/* A grant for every rx ring slot */
1628
	if (gnttab_alloc_grant_references(NET_RX_RING_SIZE,
1629 1630 1631 1632 1633 1634 1635 1636 1637 1638 1639 1640 1641 1642
					  &queue->gref_rx_head) < 0) {
		pr_alert("can't alloc rx grant refs\n");
		err = -ENOMEM;
		goto exit_free_tx;
	}

	return 0;

 exit_free_tx:
	gnttab_free_grant_references(queue->gref_tx_head);
 exit:
	return err;
}

1643 1644 1645 1646 1647 1648 1649 1650 1651 1652 1653 1654 1655 1656 1657 1658 1659 1660 1661 1662 1663 1664 1665 1666 1667 1668 1669 1670 1671 1672 1673 1674 1675 1676 1677 1678 1679 1680 1681 1682 1683 1684 1685 1686 1687 1688 1689 1690 1691 1692 1693 1694 1695 1696 1697 1698 1699 1700 1701 1702 1703 1704 1705 1706 1707 1708 1709 1710 1711 1712 1713 1714 1715 1716 1717 1718 1719 1720 1721 1722 1723 1724
static int write_queue_xenstore_keys(struct netfront_queue *queue,
			   struct xenbus_transaction *xbt, int write_hierarchical)
{
	/* Write the queue-specific keys into XenStore in the traditional
	 * way for a single queue, or in a queue subkeys for multiple
	 * queues.
	 */
	struct xenbus_device *dev = queue->info->xbdev;
	int err;
	const char *message;
	char *path;
	size_t pathsize;

	/* Choose the correct place to write the keys */
	if (write_hierarchical) {
		pathsize = strlen(dev->nodename) + 10;
		path = kzalloc(pathsize, GFP_KERNEL);
		if (!path) {
			err = -ENOMEM;
			message = "out of memory while writing ring references";
			goto error;
		}
		snprintf(path, pathsize, "%s/queue-%u",
				dev->nodename, queue->id);
	} else {
		path = (char *)dev->nodename;
	}

	/* Write ring references */
	err = xenbus_printf(*xbt, path, "tx-ring-ref", "%u",
			queue->tx_ring_ref);
	if (err) {
		message = "writing tx-ring-ref";
		goto error;
	}

	err = xenbus_printf(*xbt, path, "rx-ring-ref", "%u",
			queue->rx_ring_ref);
	if (err) {
		message = "writing rx-ring-ref";
		goto error;
	}

	/* Write event channels; taking into account both shared
	 * and split event channel scenarios.
	 */
	if (queue->tx_evtchn == queue->rx_evtchn) {
		/* Shared event channel */
		err = xenbus_printf(*xbt, path,
				"event-channel", "%u", queue->tx_evtchn);
		if (err) {
			message = "writing event-channel";
			goto error;
		}
	} else {
		/* Split event channels */
		err = xenbus_printf(*xbt, path,
				"event-channel-tx", "%u", queue->tx_evtchn);
		if (err) {
			message = "writing event-channel-tx";
			goto error;
		}

		err = xenbus_printf(*xbt, path,
				"event-channel-rx", "%u", queue->rx_evtchn);
		if (err) {
			message = "writing event-channel-rx";
			goto error;
		}
	}

	if (write_hierarchical)
		kfree(path);
	return 0;

error:
	if (write_hierarchical)
		kfree(path);
	xenbus_dev_fatal(dev, err, "%s", message);
	return err;
}

1725 1726 1727 1728 1729 1730 1731 1732 1733 1734 1735
static void xennet_destroy_queues(struct netfront_info *info)
{
	unsigned int i;

	rtnl_lock();

	for (i = 0; i < info->netdev->real_num_tx_queues; i++) {
		struct netfront_queue *queue = &info->queues[i];

		if (netif_running(info->netdev))
			napi_disable(&queue->napi);
1736
		del_timer_sync(&queue->rx_refill_timer);
1737 1738 1739 1740 1741 1742 1743 1744 1745 1746
		netif_napi_del(&queue->napi);
	}

	rtnl_unlock();

	kfree(info->queues);
	info->queues = NULL;
}

static int xennet_create_queues(struct netfront_info *info,
1747
				unsigned int *num_queues)
1748 1749 1750 1751
{
	unsigned int i;
	int ret;

1752
	info->queues = kcalloc(*num_queues, sizeof(struct netfront_queue),
1753 1754 1755 1756 1757 1758
			       GFP_KERNEL);
	if (!info->queues)
		return -ENOMEM;

	rtnl_lock();

1759
	for (i = 0; i < *num_queues; i++) {
1760 1761 1762 1763 1764 1765 1766
		struct netfront_queue *queue = &info->queues[i];

		queue->id = i;
		queue->info = info;

		ret = xennet_init_queue(queue);
		if (ret < 0) {
1767 1768
			dev_warn(&info->netdev->dev,
				 "only created %d queues\n", i);
1769
			*num_queues = i;
1770 1771 1772 1773 1774 1775 1776 1777 1778
			break;
		}

		netif_napi_add(queue->info->netdev, &queue->napi,
			       xennet_poll, 64);
		if (netif_running(info->netdev))
			napi_enable(&queue->napi);
	}

1779
	netif_set_real_num_tx_queues(info->netdev, *num_queues);
1780 1781 1782

	rtnl_unlock();

1783
	if (*num_queues == 0) {
1784 1785 1786 1787 1788 1789
		dev_err(&info->netdev->dev, "no queues\n");
		return -EINVAL;
	}
	return 0;
}

1790
/* Common code used when first setting up, and when resuming. */
1791
static int talk_to_netback(struct xenbus_device *dev,
1792 1793 1794 1795 1796
			   struct netfront_info *info)
{
	const char *message;
	struct xenbus_transaction xbt;
	int err;
1797 1798
	unsigned int feature_split_evtchn;
	unsigned int i = 0;
1799
	unsigned int max_queues = 0;
1800 1801
	struct netfront_queue *queue = NULL;
	unsigned int num_queues = 1;
1802

1803 1804
	info->netdev->irq = 0;

1805
	/* Check if backend supports multiple queues */
1806 1807
	max_queues = xenbus_read_unsigned(info->xbdev->otherend,
					  "multi-queue-max-queues", 1);
1808 1809
	num_queues = min(max_queues, xennet_max_queues);

1810
	/* Check feature-split-event-channels */
1811 1812
	feature_split_evtchn = xenbus_read_unsigned(info->xbdev->otherend,
					"feature-split-event-channels", 0);
1813 1814 1815 1816 1817 1818 1819 1820

	/* Read mac addr. */
	err = xen_net_read_mac(dev, info->netdev->dev_addr);
	if (err) {
		xenbus_dev_fatal(dev, err, "parsing %s/mac", dev->nodename);
		goto out;
	}

1821 1822 1823
	if (info->queues)
		xennet_destroy_queues(info);

1824
	err = xennet_create_queues(info, &num_queues);
1825 1826
	if (err < 0)
		goto destroy_ring;
1827 1828 1829 1830 1831 1832

	/* Create shared ring, alloc event channel -- for each queue */
	for (i = 0; i < num_queues; ++i) {
		queue = &info->queues[i];
		err = setup_netfront(dev, queue, feature_split_evtchn);
		if (err) {
1833 1834
			/* setup_netfront() will tidy up the current
			 * queue on error, but we need to clean up
1835 1836 1837 1838 1839 1840 1841 1842 1843 1844 1845 1846
			 * those already allocated.
			 */
			if (i > 0) {
				rtnl_lock();
				netif_set_real_num_tx_queues(info->netdev, i);
				rtnl_unlock();
				goto destroy_ring;
			} else {
				goto out;
			}
		}
	}
1847 1848 1849 1850 1851 1852 1853 1854

again:
	err = xenbus_transaction_start(&xbt);
	if (err) {
		xenbus_dev_fatal(dev, err, "starting transaction");
		goto destroy_ring;
	}

1855 1856
	if (xenbus_exists(XBT_NIL,
			  info->xbdev->otherend, "multi-queue-max-queues")) {
1857
		/* Write the number of queues */
1858 1859
		err = xenbus_printf(xbt, dev->nodename,
				    "multi-queue-num-queues", "%u", num_queues);
1860
		if (err) {
1861 1862
			message = "writing multi-queue-num-queues";
			goto abort_transaction_no_dev_fatal;
1863
		}
1864
	}
1865

1866 1867 1868 1869 1870
	if (num_queues == 1) {
		err = write_queue_xenstore_keys(&info->queues[0], &xbt, 0); /* flat */
		if (err)
			goto abort_transaction_no_dev_fatal;
	} else {
1871 1872 1873 1874 1875 1876
		/* Write the keys for each queue */
		for (i = 0; i < num_queues; ++i) {
			queue = &info->queues[i];
			err = write_queue_xenstore_keys(queue, &xbt, 1); /* hierarchical */
			if (err)
				goto abort_transaction_no_dev_fatal;
1877
		}
1878 1879
	}

1880
	/* The remaining keys are not queue-specific */
1881 1882 1883 1884 1885 1886 1887 1888 1889 1890 1891 1892 1893 1894 1895 1896 1897 1898 1899 1900 1901 1902 1903 1904 1905
	err = xenbus_printf(xbt, dev->nodename, "request-rx-copy", "%u",
			    1);
	if (err) {
		message = "writing request-rx-copy";
		goto abort_transaction;
	}

	err = xenbus_printf(xbt, dev->nodename, "feature-rx-notify", "%d", 1);
	if (err) {
		message = "writing feature-rx-notify";
		goto abort_transaction;
	}

	err = xenbus_printf(xbt, dev->nodename, "feature-sg", "%d", 1);
	if (err) {
		message = "writing feature-sg";
		goto abort_transaction;
	}

	err = xenbus_printf(xbt, dev->nodename, "feature-gso-tcpv4", "%d", 1);
	if (err) {
		message = "writing feature-gso-tcpv4";
		goto abort_transaction;
	}

1906 1907 1908 1909 1910 1911 1912 1913 1914 1915 1916 1917 1918
	err = xenbus_write(xbt, dev->nodename, "feature-gso-tcpv6", "1");
	if (err) {
		message = "writing feature-gso-tcpv6";
		goto abort_transaction;
	}

	err = xenbus_write(xbt, dev->nodename, "feature-ipv6-csum-offload",
			   "1");
	if (err) {
		message = "writing feature-ipv6-csum-offload";
		goto abort_transaction;
	}

1919 1920 1921 1922 1923 1924 1925 1926 1927 1928 1929 1930
	err = xenbus_transaction_end(xbt, 0);
	if (err) {
		if (err == -EAGAIN)
			goto again;
		xenbus_dev_fatal(dev, err, "completing transaction");
		goto destroy_ring;
	}

	return 0;

 abort_transaction:
	xenbus_dev_fatal(dev, err, "%s", message);
1931 1932
abort_transaction_no_dev_fatal:
	xenbus_transaction_end(xbt, 1);
1933 1934
 destroy_ring:
	xennet_disconnect_backend(info);
1935 1936
	kfree(info->queues);
	info->queues = NULL;
1937 1938 1939 1940 1941 1942 1943
 out:
	return err;
}

static int xennet_connect(struct net_device *dev)
{
	struct netfront_info *np = netdev_priv(dev);
1944
	unsigned int num_queues = 0;
1945
	int err;
1946 1947
	unsigned int j = 0;
	struct netfront_queue *queue = NULL;
1948

1949
	if (!xenbus_read_unsigned(np->xbdev->otherend, "feature-rx-copy", 0)) {
1950
		dev_info(&dev->dev,
1951
			 "backend does not support copying receive path\n");
1952 1953 1954
		return -ENODEV;
	}

1955
	err = talk_to_netback(np->xbdev, np);
1956 1957 1958
	if (err)
		return err;

1959 1960 1961
	/* talk_to_netback() sets the correct number of queues */
	num_queues = dev->real_num_tx_queues;

1962
	rtnl_lock();
1963
	netdev_update_features(dev);
1964
	rtnl_unlock();
1965 1966

	/*
1967
	 * All public and private state should now be sane.  Get
1968 1969 1970 1971 1972
	 * ready to start sending and receiving packets and give the driver
	 * domain a kick because we've probably just requeued some
	 * packets.
	 */
	netif_carrier_on(np->netdev);
1973 1974
	for (j = 0; j < num_queues; ++j) {
		queue = &np->queues[j];
1975

1976 1977 1978 1979
		notify_remote_via_irq(queue->tx_irq);
		if (queue->tx_irq != queue->rx_irq)
			notify_remote_via_irq(queue->rx_irq);

1980 1981
		spin_lock_irq(&queue->tx_lock);
		xennet_tx_buf_gc(queue);
1982
		spin_unlock_irq(&queue->tx_lock);
1983 1984 1985

		spin_lock_bh(&queue->rx_lock);
		xennet_alloc_rx_buffers(queue);
1986 1987
		spin_unlock_bh(&queue->rx_lock);
	}
1988 1989 1990 1991 1992 1993 1994

	return 0;
}

/**
 * Callback received when the backend's state changes.
 */
1995
static void netback_changed(struct xenbus_device *dev,
1996 1997
			    enum xenbus_state backend_state)
{
1998
	struct netfront_info *np = dev_get_drvdata(&dev->dev);
1999 2000 2001 2002 2003 2004 2005
	struct net_device *netdev = np->netdev;

	dev_dbg(&dev->dev, "%s\n", xenbus_strstate(backend_state));

	switch (backend_state) {
	case XenbusStateInitialising:
	case XenbusStateInitialised:
2006 2007
	case XenbusStateReconfiguring:
	case XenbusStateReconfigured:
2008 2009 2010 2011 2012 2013 2014 2015 2016
	case XenbusStateUnknown:
		break;

	case XenbusStateInitWait:
		if (dev->state != XenbusStateInitialising)
			break;
		if (xennet_connect(netdev) != 0)
			break;
		xenbus_switch_state(dev, XenbusStateConnected);
2017 2018 2019
		break;

	case XenbusStateConnected:
2020
		netdev_notify_peers(netdev);
2021 2022
		break;

2023 2024 2025 2026
	case XenbusStateClosed:
		if (dev->state == XenbusStateClosed)
			break;
		/* Missed the backend's CLOSING state -- fallthrough */
2027 2028 2029 2030 2031 2032
	case XenbusStateClosing:
		xenbus_frontend_closed(dev);
		break;
	}
}

2033 2034 2035 2036 2037 2038 2039 2040 2041 2042 2043 2044 2045 2046 2047 2048 2049 2050 2051 2052 2053 2054 2055 2056 2057 2058 2059
static const struct xennet_stat {
	char name[ETH_GSTRING_LEN];
	u16 offset;
} xennet_stats[] = {
	{
		"rx_gso_checksum_fixup",
		offsetof(struct netfront_info, rx_gso_checksum_fixup)
	},
};

static int xennet_get_sset_count(struct net_device *dev, int string_set)
{
	switch (string_set) {
	case ETH_SS_STATS:
		return ARRAY_SIZE(xennet_stats);
	default:
		return -EINVAL;
	}
}

static void xennet_get_ethtool_stats(struct net_device *dev,
				     struct ethtool_stats *stats, u64 * data)
{
	void *np = netdev_priv(dev);
	int i;

	for (i = 0; i < ARRAY_SIZE(xennet_stats); i++)
2060
		data[i] = atomic_read((atomic_t *)(np + xennet_stats[i].offset));
2061 2062 2063 2064 2065 2066 2067 2068 2069 2070 2071 2072 2073 2074 2075
}

static void xennet_get_strings(struct net_device *dev, u32 stringset, u8 * data)
{
	int i;

	switch (stringset) {
	case ETH_SS_STATS:
		for (i = 0; i < ARRAY_SIZE(xennet_stats); i++)
			memcpy(data + i * ETH_GSTRING_LEN,
			       xennet_stats[i].name, ETH_GSTRING_LEN);
		break;
	}
}

2076
static const struct ethtool_ops xennet_ethtool_ops =
2077 2078
{
	.get_link = ethtool_op_get_link,
2079 2080 2081 2082

	.get_sset_count = xennet_get_sset_count,
	.get_ethtool_stats = xennet_get_ethtool_stats,
	.get_strings = xennet_get_strings,
2083 2084 2085
};

#ifdef CONFIG_SYSFS
2086 2087
static ssize_t show_rxbuf(struct device *dev,
			  struct device_attribute *attr, char *buf)
2088
{
2089
	return sprintf(buf, "%lu\n", NET_RX_RING_SIZE);
2090 2091
}

2092 2093 2094
static ssize_t store_rxbuf(struct device *dev,
			   struct device_attribute *attr,
			   const char *buf, size_t len)
2095 2096 2097 2098 2099 2100 2101 2102 2103 2104 2105
{
	char *endp;
	unsigned long target;

	if (!capable(CAP_NET_ADMIN))
		return -EPERM;

	target = simple_strtoul(buf, &endp, 0);
	if (endp == buf)
		return -EBADMSG;

2106
	/* rxbuf_min and rxbuf_max are no longer configurable. */
2107 2108 2109 2110

	return len;
}

2111 2112 2113
static DEVICE_ATTR(rxbuf_min, S_IRUGO|S_IWUSR, show_rxbuf, store_rxbuf);
static DEVICE_ATTR(rxbuf_max, S_IRUGO|S_IWUSR, show_rxbuf, store_rxbuf);
static DEVICE_ATTR(rxbuf_cur, S_IRUGO, show_rxbuf, NULL);
2114

2115 2116 2117 2118 2119 2120
static struct attribute *xennet_dev_attrs[] = {
	&dev_attr_rxbuf_min.attr,
	&dev_attr_rxbuf_max.attr,
	&dev_attr_rxbuf_cur.attr,
	NULL
};
2121

2122 2123 2124
static const struct attribute_group xennet_dev_group = {
	.attrs = xennet_dev_attrs
};
2125 2126
#endif /* CONFIG_SYSFS */

2127
static int xennet_remove(struct xenbus_device *dev)
2128
{
2129
	struct netfront_info *info = dev_get_drvdata(&dev->dev);
2130 2131 2132 2133 2134

	dev_dbg(&dev->dev, "%s\n", dev->nodename);

	xennet_disconnect_backend(info);

2135 2136
	unregister_netdev(info->netdev);

2137 2138
	if (info->queues)
		xennet_destroy_queues(info);
2139
	xennet_free_netdev(info->netdev);
2140 2141 2142 2143

	return 0;
}

2144 2145 2146 2147 2148 2149 2150
static const struct xenbus_device_id netfront_ids[] = {
	{ "vif" },
	{ "" }
};

static struct xenbus_driver netfront_driver = {
	.ids = netfront_ids,
2151
	.probe = netfront_probe,
2152
	.remove = xennet_remove,
2153
	.resume = netfront_resume,
2154
	.otherend_changed = netback_changed,
2155
};
2156 2157 2158

static int __init netif_init(void)
{
2159
	if (!xen_domain())
2160 2161
		return -ENODEV;

2162
	if (!xen_has_pv_nic_devices())
2163 2164
		return -ENODEV;

2165
	pr_info("Initialising Xen virtual ethernet driver\n");
2166

2167 2168 2169 2170 2171
	/* Allow as many queues as there are CPUs if user has not
	 * specified a value.
	 */
	if (xennet_max_queues == 0)
		xennet_max_queues = num_online_cpus();
2172

2173
	return xenbus_register_frontend(&netfront_driver);
2174 2175 2176 2177 2178 2179
}
module_init(netif_init);


static void __exit netif_exit(void)
{
2180
	xenbus_unregister_driver(&netfront_driver);
2181 2182 2183 2184 2185
}
module_exit(netif_exit);

MODULE_DESCRIPTION("Xen virtual network device frontend");
MODULE_LICENSE("GPL");
2186
MODULE_ALIAS("xen:vif");
2187
MODULE_ALIAS("xennet");