xen-netback: Add support for multiple queues
[linux-2.6-block.git] / drivers / net / xen-netback / interface.c
CommitLineData
f942dc25
IC
1/*
2 * Network-device interface management.
3 *
4 * Copyright (c) 2004-2005, Keir Fraser
5 *
6 * This program is free software; you can redistribute it and/or
7 * modify it under the terms of the GNU General Public License version 2
8 * as published by the Free Software Foundation; or, when distributed
9 * separately from the Linux kernel or incorporated into other
10 * software packages, subject to the following license:
11 *
12 * Permission is hereby granted, free of charge, to any person obtaining a copy
13 * of this source file (the "Software"), to deal in the Software without
14 * restriction, including without limitation the rights to use, copy, modify,
15 * merge, publish, distribute, sublicense, and/or sell copies of the Software,
16 * and to permit persons to whom the Software is furnished to do so, subject to
17 * the following conditions:
18 *
19 * The above copyright notice and this permission notice shall be included in
20 * all copies or substantial portions of the Software.
21 *
22 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
23 * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
24 * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
25 * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
26 * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
27 * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
28 * IN THE SOFTWARE.
29 */
30
31#include "common.h"
32
b3f980bd 33#include <linux/kthread.h>
f942dc25
IC
34#include <linux/ethtool.h>
35#include <linux/rtnetlink.h>
36#include <linux/if_vlan.h>
37
38#include <xen/events.h>
39#include <asm/xen/hypercall.h>
f53c3fe8 40#include <xen/balloon.h>
f942dc25
IC
41
42#define XENVIF_QUEUE_LENGTH 32
b3f980bd 43#define XENVIF_NAPI_WEIGHT 64
f942dc25 44
e9ce7cb6
WL
45static inline void xenvif_stop_queue(struct xenvif_queue *queue)
46{
47 struct net_device *dev = queue->vif->dev;
48
49 if (!queue->vif->can_queue)
50 return;
51
52 netif_tx_stop_queue(netdev_get_tx_queue(dev, queue->id));
53}
54
f942dc25
IC
55int xenvif_schedulable(struct xenvif *vif)
56{
57 return netif_running(vif->dev) && netif_carrier_ok(vif->dev);
58}
59
e1f00a69 60static irqreturn_t xenvif_tx_interrupt(int irq, void *dev_id)
f942dc25 61{
e9ce7cb6 62 struct xenvif_queue *queue = dev_id;
f942dc25 63
e9ce7cb6
WL
64 if (RING_HAS_UNCONSUMED_REQUESTS(&queue->tx))
65 napi_schedule(&queue->napi);
f942dc25 66
e1f00a69
WL
67 return IRQ_HANDLED;
68}
69
e9ce7cb6 70int xenvif_poll(struct napi_struct *napi, int budget)
b3f980bd 71{
e9ce7cb6
WL
72 struct xenvif_queue *queue =
73 container_of(napi, struct xenvif_queue, napi);
b3f980bd
WL
74 int work_done;
75
e9d8b2c2
WL
76 /* This vif is rogue, we pretend we've there is nothing to do
77 * for this vif to deschedule it from NAPI. But this interface
78 * will be turned off in thread context later.
79 */
e9ce7cb6 80 if (unlikely(queue->vif->disabled)) {
e9d8b2c2
WL
81 napi_complete(napi);
82 return 0;
83 }
84
e9ce7cb6 85 work_done = xenvif_tx_action(queue, budget);
b3f980bd
WL
86
87 if (work_done < budget) {
0d08fceb 88 napi_complete(napi);
e9ce7cb6 89 xenvif_napi_schedule_or_enable_events(queue);
b3f980bd
WL
90 }
91
92 return work_done;
93}
94
e1f00a69
WL
95static irqreturn_t xenvif_rx_interrupt(int irq, void *dev_id)
96{
e9ce7cb6 97 struct xenvif_queue *queue = dev_id;
e1f00a69 98
e9ce7cb6 99 xenvif_kick_thread(queue);
f942dc25
IC
100
101 return IRQ_HANDLED;
102}
103
e1f00a69
WL
104static irqreturn_t xenvif_interrupt(int irq, void *dev_id)
105{
106 xenvif_tx_interrupt(irq, dev_id);
107 xenvif_rx_interrupt(irq, dev_id);
108
109 return IRQ_HANDLED;
110}
111
e9ce7cb6
WL
112int xenvif_queue_stopped(struct xenvif_queue *queue)
113{
114 struct net_device *dev = queue->vif->dev;
115 unsigned int id = queue->id;
116 return netif_tx_queue_stopped(netdev_get_tx_queue(dev, id));
117}
118
119void xenvif_wake_queue(struct xenvif_queue *queue)
120{
121 struct net_device *dev = queue->vif->dev;
122 unsigned int id = queue->id;
123 netif_tx_wake_queue(netdev_get_tx_queue(dev, id));
124}
125
126/* Callback to wake the queue and drain it on timeout */
127static void xenvif_wake_queue_callback(unsigned long data)
09350788 128{
e9ce7cb6
WL
129 struct xenvif_queue *queue = (struct xenvif_queue *)data;
130
131 if (xenvif_queue_stopped(queue)) {
132 netdev_err(queue->vif->dev, "draining TX queue\n");
133 queue->rx_queue_purge = true;
134 xenvif_kick_thread(queue);
135 xenvif_wake_queue(queue);
136 }
137}
09350788 138
e9ce7cb6
WL
139static u16 xenvif_select_queue(struct net_device *dev, struct sk_buff *skb,
140 void *accel_priv, select_queue_fallback_t fallback)
141{
e9ce7cb6
WL
142 unsigned int num_queues = dev->real_num_tx_queues;
143 u32 hash;
144 u16 queue_index;
145
146 /* First, check if there is only one queue to optimise the
147 * single-queue or old frontend scenario.
148 */
149 if (num_queues == 1) {
150 queue_index = 0;
151 } else {
152 /* Use skb_get_hash to obtain an L4 hash if available */
153 hash = skb_get_hash(skb);
154 queue_index = hash % num_queues;
09350788 155 }
e9ce7cb6
WL
156
157 return queue_index;
09350788
ZK
158}
159
f942dc25
IC
160static int xenvif_start_xmit(struct sk_buff *skb, struct net_device *dev)
161{
162 struct xenvif *vif = netdev_priv(dev);
e9ce7cb6
WL
163 struct xenvif_queue *queue = NULL;
164 unsigned int num_queues = dev->real_num_tx_queues;
165 u16 index;
ca2f09f2 166 int min_slots_needed;
f942dc25
IC
167
168 BUG_ON(skb->dev != dev);
169
e9ce7cb6
WL
170 /* Drop the packet if queues are not set up */
171 if (num_queues < 1)
172 goto drop;
173
174 /* Obtain the queue to be used to transmit this packet */
175 index = skb_get_queue_mapping(skb);
176 if (index >= num_queues) {
177 pr_warn_ratelimited("Invalid queue %hu for packet on interface %s\n.",
178 index, vif->dev->name);
179 index %= num_queues;
180 }
181 queue = &vif->queues[index];
182
183 /* Drop the packet if queue is not ready */
184 if (queue->task == NULL ||
185 queue->dealloc_task == NULL ||
f53c3fe8 186 !xenvif_schedulable(vif))
f942dc25
IC
187 goto drop;
188
ca2f09f2
PD
189 /* At best we'll need one slot for the header and one for each
190 * frag.
191 */
192 min_slots_needed = 1 + skb_shinfo(skb)->nr_frags;
f942dc25 193
ca2f09f2
PD
194 /* If the skb is GSO then we'll also need an extra slot for the
195 * metadata.
196 */
836fbaf4 197 if (skb_is_gso(skb))
ca2f09f2 198 min_slots_needed++;
f942dc25 199
ca2f09f2
PD
200 /* If the skb can't possibly fit in the remaining slots
201 * then turn off the queue to give the ring a chance to
202 * drain.
203 */
e9ce7cb6
WL
204 if (!xenvif_rx_ring_slots_available(queue, min_slots_needed)) {
205 queue->wake_queue.function = xenvif_wake_queue_callback;
206 queue->wake_queue.data = (unsigned long)queue;
207 xenvif_stop_queue(queue);
208 mod_timer(&queue->wake_queue,
09350788
ZK
209 jiffies + rx_drain_timeout_jiffies);
210 }
f942dc25 211
e9ce7cb6
WL
212 skb_queue_tail(&queue->rx_queue, skb);
213 xenvif_kick_thread(queue);
f942dc25
IC
214
215 return NETDEV_TX_OK;
216
217 drop:
218 vif->dev->stats.tx_dropped++;
219 dev_kfree_skb(skb);
220 return NETDEV_TX_OK;
221}
222
f942dc25
IC
223static struct net_device_stats *xenvif_get_stats(struct net_device *dev)
224{
225 struct xenvif *vif = netdev_priv(dev);
e9ce7cb6
WL
226 struct xenvif_queue *queue = NULL;
227 unsigned int num_queues = dev->real_num_tx_queues;
228 unsigned long rx_bytes = 0;
229 unsigned long rx_packets = 0;
230 unsigned long tx_bytes = 0;
231 unsigned long tx_packets = 0;
232 unsigned int index;
233
234 if (vif->queues == NULL)
235 goto out;
236
237 /* Aggregate tx and rx stats from each queue */
238 for (index = 0; index < num_queues; ++index) {
239 queue = &vif->queues[index];
240 rx_bytes += queue->stats.rx_bytes;
241 rx_packets += queue->stats.rx_packets;
242 tx_bytes += queue->stats.tx_bytes;
243 tx_packets += queue->stats.tx_packets;
244 }
245
246out:
247 vif->dev->stats.rx_bytes = rx_bytes;
248 vif->dev->stats.rx_packets = rx_packets;
249 vif->dev->stats.tx_bytes = tx_bytes;
250 vif->dev->stats.tx_packets = tx_packets;
251
f942dc25
IC
252 return &vif->dev->stats;
253}
254
255static void xenvif_up(struct xenvif *vif)
256{
e9ce7cb6
WL
257 struct xenvif_queue *queue = NULL;
258 unsigned int num_queues = vif->dev->real_num_tx_queues;
259 unsigned int queue_index;
260
261 for (queue_index = 0; queue_index < num_queues; ++queue_index) {
262 queue = &vif->queues[queue_index];
263 napi_enable(&queue->napi);
264 enable_irq(queue->tx_irq);
265 if (queue->tx_irq != queue->rx_irq)
266 enable_irq(queue->rx_irq);
267 xenvif_napi_schedule_or_enable_events(queue);
268 }
f942dc25
IC
269}
270
271static void xenvif_down(struct xenvif *vif)
272{
e9ce7cb6
WL
273 struct xenvif_queue *queue = NULL;
274 unsigned int num_queues = vif->dev->real_num_tx_queues;
275 unsigned int queue_index;
276
277 for (queue_index = 0; queue_index < num_queues; ++queue_index) {
278 queue = &vif->queues[queue_index];
279 napi_disable(&queue->napi);
280 disable_irq(queue->tx_irq);
281 if (queue->tx_irq != queue->rx_irq)
282 disable_irq(queue->rx_irq);
283 del_timer_sync(&queue->credit_timeout);
284 }
f942dc25
IC
285}
286
287static int xenvif_open(struct net_device *dev)
288{
289 struct xenvif *vif = netdev_priv(dev);
290 if (netif_carrier_ok(dev))
291 xenvif_up(vif);
e9ce7cb6 292 netif_tx_start_all_queues(dev);
f942dc25
IC
293 return 0;
294}
295
296static int xenvif_close(struct net_device *dev)
297{
298 struct xenvif *vif = netdev_priv(dev);
299 if (netif_carrier_ok(dev))
300 xenvif_down(vif);
e9ce7cb6 301 netif_tx_stop_all_queues(dev);
f942dc25
IC
302 return 0;
303}
304
305static int xenvif_change_mtu(struct net_device *dev, int mtu)
306{
307 struct xenvif *vif = netdev_priv(dev);
308 int max = vif->can_sg ? 65535 - VLAN_ETH_HLEN : ETH_DATA_LEN;
309
310 if (mtu > max)
311 return -EINVAL;
312 dev->mtu = mtu;
313 return 0;
314}
315
c8f44aff
MM
316static netdev_features_t xenvif_fix_features(struct net_device *dev,
317 netdev_features_t features)
f942dc25
IC
318{
319 struct xenvif *vif = netdev_priv(dev);
f942dc25 320
47103041
MM
321 if (!vif->can_sg)
322 features &= ~NETIF_F_SG;
82cada22 323 if (~(vif->gso_mask | vif->gso_prefix_mask) & GSO_BIT(TCPV4))
47103041 324 features &= ~NETIF_F_TSO;
82cada22
PD
325 if (~(vif->gso_mask | vif->gso_prefix_mask) & GSO_BIT(TCPV6))
326 features &= ~NETIF_F_TSO6;
146c8a77 327 if (!vif->ip_csum)
47103041 328 features &= ~NETIF_F_IP_CSUM;
146c8a77
PD
329 if (!vif->ipv6_csum)
330 features &= ~NETIF_F_IPV6_CSUM;
f942dc25 331
47103041 332 return features;
f942dc25
IC
333}
334
335static const struct xenvif_stat {
336 char name[ETH_GSTRING_LEN];
337 u16 offset;
338} xenvif_stats[] = {
339 {
340 "rx_gso_checksum_fixup",
e9ce7cb6 341 offsetof(struct xenvif_stats, rx_gso_checksum_fixup)
f942dc25 342 },
1bb332af
ZK
343 /* If (sent != success + fail), there are probably packets never
344 * freed up properly!
345 */
346 {
347 "tx_zerocopy_sent",
e9ce7cb6 348 offsetof(struct xenvif_stats, tx_zerocopy_sent),
1bb332af
ZK
349 },
350 {
351 "tx_zerocopy_success",
e9ce7cb6 352 offsetof(struct xenvif_stats, tx_zerocopy_success),
1bb332af
ZK
353 },
354 {
355 "tx_zerocopy_fail",
e9ce7cb6 356 offsetof(struct xenvif_stats, tx_zerocopy_fail)
1bb332af 357 },
e3377f36
ZK
358 /* Number of packets exceeding MAX_SKB_FRAG slots. You should use
359 * a guest with the same MAX_SKB_FRAG
360 */
361 {
362 "tx_frag_overflow",
e9ce7cb6 363 offsetof(struct xenvif_stats, tx_frag_overflow)
e3377f36 364 },
f942dc25
IC
365};
366
367static int xenvif_get_sset_count(struct net_device *dev, int string_set)
368{
369 switch (string_set) {
370 case ETH_SS_STATS:
371 return ARRAY_SIZE(xenvif_stats);
372 default:
373 return -EINVAL;
374 }
375}
376
377static void xenvif_get_ethtool_stats(struct net_device *dev,
378 struct ethtool_stats *stats, u64 * data)
379{
e9ce7cb6
WL
380 struct xenvif *vif = netdev_priv(dev);
381 unsigned int num_queues = dev->real_num_tx_queues;
f942dc25 382 int i;
e9ce7cb6
WL
383 unsigned int queue_index;
384 struct xenvif_stats *vif_stats;
385
386 for (i = 0; i < ARRAY_SIZE(xenvif_stats); i++) {
387 unsigned long accum = 0;
388 for (queue_index = 0; queue_index < num_queues; ++queue_index) {
389 vif_stats = &vif->queues[queue_index].stats;
390 accum += *(unsigned long *)(vif_stats + xenvif_stats[i].offset);
391 }
392 data[i] = accum;
393 }
f942dc25
IC
394}
395
396static void xenvif_get_strings(struct net_device *dev, u32 stringset, u8 * data)
397{
398 int i;
399
400 switch (stringset) {
401 case ETH_SS_STATS:
402 for (i = 0; i < ARRAY_SIZE(xenvif_stats); i++)
403 memcpy(data + i * ETH_GSTRING_LEN,
404 xenvif_stats[i].name, ETH_GSTRING_LEN);
405 break;
406 }
407}
408
813abbba 409static const struct ethtool_ops xenvif_ethtool_ops = {
f942dc25
IC
410 .get_link = ethtool_op_get_link,
411
412 .get_sset_count = xenvif_get_sset_count,
413 .get_ethtool_stats = xenvif_get_ethtool_stats,
414 .get_strings = xenvif_get_strings,
415};
416
813abbba 417static const struct net_device_ops xenvif_netdev_ops = {
f942dc25
IC
418 .ndo_start_xmit = xenvif_start_xmit,
419 .ndo_get_stats = xenvif_get_stats,
420 .ndo_open = xenvif_open,
421 .ndo_stop = xenvif_close,
422 .ndo_change_mtu = xenvif_change_mtu,
47103041 423 .ndo_fix_features = xenvif_fix_features,
4a633a60
MW
424 .ndo_set_mac_address = eth_mac_addr,
425 .ndo_validate_addr = eth_validate_addr,
e9ce7cb6 426 .ndo_select_queue = xenvif_select_queue,
f942dc25
IC
427};
428
429struct xenvif *xenvif_alloc(struct device *parent, domid_t domid,
430 unsigned int handle)
431{
432 int err;
433 struct net_device *dev;
434 struct xenvif *vif;
435 char name[IFNAMSIZ] = {};
436
437 snprintf(name, IFNAMSIZ - 1, "vif%u.%u", domid, handle);
8d3d53b3
AB
438 /* Allocate a netdev with the max. supported number of queues.
439 * When the guest selects the desired number, it will be updated
440 * via netif_set_real_num_tx_queues().
441 */
442 dev = alloc_netdev_mq(sizeof(struct xenvif), name, ether_setup,
443 xenvif_max_queues);
f942dc25 444 if (dev == NULL) {
b3f980bd 445 pr_warn("Could not allocate netdev for %s\n", name);
f942dc25
IC
446 return ERR_PTR(-ENOMEM);
447 }
448
449 SET_NETDEV_DEV(dev, parent);
450
451 vif = netdev_priv(dev);
ac3d5ac2 452
f942dc25
IC
453 vif->domid = domid;
454 vif->handle = handle;
f942dc25 455 vif->can_sg = 1;
146c8a77 456 vif->ip_csum = 1;
f942dc25 457 vif->dev = dev;
e9d8b2c2
WL
458 vif->disabled = false;
459
e9ce7cb6
WL
460 /* Start out with no queues. The call below does not require
461 * rtnl_lock() as it happens before register_netdev().
462 */
463 vif->queues = NULL;
464 netif_set_real_num_tx_queues(dev, 0);
09350788 465
f942dc25 466 dev->netdev_ops = &xenvif_netdev_ops;
146c8a77
PD
467 dev->hw_features = NETIF_F_SG |
468 NETIF_F_IP_CSUM | NETIF_F_IPV6_CSUM |
82cada22 469 NETIF_F_TSO | NETIF_F_TSO6;
7365bcfa 470 dev->features = dev->hw_features | NETIF_F_RXCSUM;
7ad24ea4 471 dev->ethtool_ops = &xenvif_ethtool_ops;
f942dc25
IC
472
473 dev->tx_queue_len = XENVIF_QUEUE_LENGTH;
474
475 /*
476 * Initialise a dummy MAC address. We choose the numerically
477 * largest non-broadcast address to prevent the address getting
478 * stolen by an Ethernet bridge for STP purposes.
479 * (FE:FF:FF:FF:FF:FF)
480 */
481 memset(dev->dev_addr, 0xFF, ETH_ALEN);
482 dev->dev_addr[0] &= ~0x01;
483
484 netif_carrier_off(dev);
485
486 err = register_netdev(dev);
487 if (err) {
488 netdev_warn(dev, "Could not register device: err=%d\n", err);
489 free_netdev(dev);
490 return ERR_PTR(err);
491 }
492
493 netdev_dbg(dev, "Successfully created xenvif\n");
279f438e
PD
494
495 __module_get(THIS_MODULE);
496
f942dc25
IC
497 return vif;
498}
499
e9ce7cb6
WL
500int xenvif_init_queue(struct xenvif_queue *queue)
501{
502 int err, i;
503
504 queue->credit_bytes = queue->remaining_credit = ~0UL;
505 queue->credit_usec = 0UL;
506 init_timer(&queue->credit_timeout);
507 queue->credit_window_start = get_jiffies_64();
508
509 skb_queue_head_init(&queue->rx_queue);
510 skb_queue_head_init(&queue->tx_queue);
511
512 queue->pending_cons = 0;
513 queue->pending_prod = MAX_PENDING_REQS;
514 for (i = 0; i < MAX_PENDING_REQS; ++i)
515 queue->pending_ring[i] = i;
516
517 spin_lock_init(&queue->callback_lock);
518 spin_lock_init(&queue->response_lock);
519
520 /* If ballooning is disabled, this will consume real memory, so you
521 * better enable it. The long term solution would be to use just a
522 * bunch of valid page descriptors, without dependency on ballooning
523 */
524 err = alloc_xenballooned_pages(MAX_PENDING_REQS,
525 queue->mmap_pages,
526 false);
527 if (err) {
528 netdev_err(queue->vif->dev, "Could not reserve mmap_pages\n");
529 return -ENOMEM;
530 }
531
532 for (i = 0; i < MAX_PENDING_REQS; i++) {
533 queue->pending_tx_info[i].callback_struct = (struct ubuf_info)
534 { .callback = xenvif_zerocopy_callback,
535 .ctx = NULL,
536 .desc = i };
537 queue->grant_tx_handle[i] = NETBACK_INVALID_HANDLE;
538 }
539
540 init_timer(&queue->wake_queue);
541
542 netif_napi_add(queue->vif->dev, &queue->napi, xenvif_poll,
543 XENVIF_NAPI_WEIGHT);
544
545 return 0;
546}
547
548void xenvif_carrier_on(struct xenvif *vif)
549{
550 rtnl_lock();
551 if (!vif->can_sg && vif->dev->mtu > ETH_DATA_LEN)
552 dev_set_mtu(vif->dev, ETH_DATA_LEN);
553 netdev_update_features(vif->dev);
554 netif_carrier_on(vif->dev);
555 if (netif_running(vif->dev))
556 xenvif_up(vif);
557 rtnl_unlock();
558}
559
560int xenvif_connect(struct xenvif_queue *queue, unsigned long tx_ring_ref,
e1f00a69
WL
561 unsigned long rx_ring_ref, unsigned int tx_evtchn,
562 unsigned int rx_evtchn)
f942dc25 563{
67fa3660 564 struct task_struct *task;
f942dc25
IC
565 int err = -ENOMEM;
566
e9ce7cb6
WL
567 BUG_ON(queue->tx_irq);
568 BUG_ON(queue->task);
569 BUG_ON(queue->dealloc_task);
f942dc25 570
e9ce7cb6 571 err = xenvif_map_frontend_rings(queue, tx_ring_ref, rx_ring_ref);
f942dc25
IC
572 if (err < 0)
573 goto err;
574
e9ce7cb6
WL
575 init_waitqueue_head(&queue->wq);
576 init_waitqueue_head(&queue->dealloc_wq);
ca2f09f2 577
e1f00a69
WL
578 if (tx_evtchn == rx_evtchn) {
579 /* feature-split-event-channels == 0 */
580 err = bind_interdomain_evtchn_to_irqhandler(
e9ce7cb6
WL
581 queue->vif->domid, tx_evtchn, xenvif_interrupt, 0,
582 queue->name, queue);
e1f00a69
WL
583 if (err < 0)
584 goto err_unmap;
e9ce7cb6
WL
585 queue->tx_irq = queue->rx_irq = err;
586 disable_irq(queue->tx_irq);
e1f00a69
WL
587 } else {
588 /* feature-split-event-channels == 1 */
e9ce7cb6
WL
589 snprintf(queue->tx_irq_name, sizeof(queue->tx_irq_name),
590 "%s-tx", queue->name);
e1f00a69 591 err = bind_interdomain_evtchn_to_irqhandler(
e9ce7cb6
WL
592 queue->vif->domid, tx_evtchn, xenvif_tx_interrupt, 0,
593 queue->tx_irq_name, queue);
e1f00a69
WL
594 if (err < 0)
595 goto err_unmap;
e9ce7cb6
WL
596 queue->tx_irq = err;
597 disable_irq(queue->tx_irq);
e1f00a69 598
e9ce7cb6
WL
599 snprintf(queue->rx_irq_name, sizeof(queue->rx_irq_name),
600 "%s-rx", queue->name);
e1f00a69 601 err = bind_interdomain_evtchn_to_irqhandler(
e9ce7cb6
WL
602 queue->vif->domid, rx_evtchn, xenvif_rx_interrupt, 0,
603 queue->rx_irq_name, queue);
e1f00a69
WL
604 if (err < 0)
605 goto err_tx_unbind;
e9ce7cb6
WL
606 queue->rx_irq = err;
607 disable_irq(queue->rx_irq);
e1f00a69 608 }
f942dc25 609
121fa4b7 610 task = kthread_create(xenvif_kthread_guest_rx,
e9ce7cb6 611 (void *)queue, "%s-guest-rx", queue->name);
67fa3660 612 if (IS_ERR(task)) {
e9ce7cb6 613 pr_warn("Could not allocate kthread for %s\n", queue->name);
67fa3660 614 err = PTR_ERR(task);
b3f980bd
WL
615 goto err_rx_unbind;
616 }
e9ce7cb6 617 queue->task = task;
67fa3660 618
f53c3fe8 619 task = kthread_create(xenvif_dealloc_kthread,
e9ce7cb6 620 (void *)queue, "%s-dealloc", queue->name);
f53c3fe8 621 if (IS_ERR(task)) {
e9ce7cb6 622 pr_warn("Could not allocate kthread for %s\n", queue->name);
f53c3fe8
ZK
623 err = PTR_ERR(task);
624 goto err_rx_unbind;
625 }
e9ce7cb6 626 queue->dealloc_task = task;
f53c3fe8 627
e9ce7cb6
WL
628 wake_up_process(queue->task);
629 wake_up_process(queue->dealloc_task);
b3f980bd 630
f942dc25 631 return 0;
b3f980bd
WL
632
633err_rx_unbind:
e9ce7cb6
WL
634 unbind_from_irqhandler(queue->rx_irq, queue);
635 queue->rx_irq = 0;
e1f00a69 636err_tx_unbind:
e9ce7cb6
WL
637 unbind_from_irqhandler(queue->tx_irq, queue);
638 queue->tx_irq = 0;
f942dc25 639err_unmap:
e9ce7cb6 640 xenvif_unmap_frontend_rings(queue);
f942dc25 641err:
b103f358 642 module_put(THIS_MODULE);
f942dc25
IC
643 return err;
644}
645
48856286 646void xenvif_carrier_off(struct xenvif *vif)
f942dc25
IC
647{
648 struct net_device *dev = vif->dev;
48856286
IC
649
650 rtnl_lock();
651 netif_carrier_off(dev); /* discard queued packets */
652 if (netif_running(dev))
653 xenvif_down(vif);
654 rtnl_unlock();
48856286
IC
655}
656
e9ce7cb6
WL
657static void xenvif_wait_unmap_timeout(struct xenvif_queue *queue,
658 unsigned int worst_case_skb_lifetime)
659{
660 int i, unmap_timeout = 0;
661
662 for (i = 0; i < MAX_PENDING_REQS; ++i) {
663 if (queue->grant_tx_handle[i] != NETBACK_INVALID_HANDLE) {
664 unmap_timeout++;
665 schedule_timeout(msecs_to_jiffies(1000));
666 if (unmap_timeout > worst_case_skb_lifetime &&
667 net_ratelimit())
668 netdev_err(queue->vif->dev,
669 "Page still granted! Index: %x\n",
670 i);
671 i = -1;
672 }
673 }
674}
675
48856286
IC
676void xenvif_disconnect(struct xenvif *vif)
677{
e9ce7cb6
WL
678 struct xenvif_queue *queue = NULL;
679 unsigned int num_queues = vif->dev->real_num_tx_queues;
680 unsigned int queue_index;
681
48856286
IC
682 if (netif_carrier_ok(vif->dev))
683 xenvif_carrier_off(vif);
f942dc25 684
e9ce7cb6
WL
685 for (queue_index = 0; queue_index < num_queues; ++queue_index) {
686 queue = &vif->queues[queue_index];
db739ef3 687
e9ce7cb6
WL
688 if (queue->task) {
689 del_timer_sync(&queue->wake_queue);
690 kthread_stop(queue->task);
691 queue->task = NULL;
692 }
f53c3fe8 693
e9ce7cb6
WL
694 if (queue->dealloc_task) {
695 kthread_stop(queue->dealloc_task);
696 queue->dealloc_task = NULL;
697 }
698
699 if (queue->tx_irq) {
700 if (queue->tx_irq == queue->rx_irq)
701 unbind_from_irqhandler(queue->tx_irq, queue);
702 else {
703 unbind_from_irqhandler(queue->tx_irq, queue);
704 unbind_from_irqhandler(queue->rx_irq, queue);
705 }
706 queue->tx_irq = 0;
e1f00a69 707 }
f942dc25 708
e9ce7cb6
WL
709 xenvif_unmap_frontend_rings(queue);
710 }
279f438e
PD
711}
712
8d3d53b3
AB
713/* Reverse the relevant parts of xenvif_init_queue().
714 * Used for queue teardown from xenvif_free(), and on the
715 * error handling paths in xenbus.c:connect().
716 */
717void xenvif_deinit_queue(struct xenvif_queue *queue)
718{
719 free_xenballooned_pages(MAX_PENDING_REQS, queue->mmap_pages);
720 netif_napi_del(&queue->napi);
721}
722
279f438e
PD
723void xenvif_free(struct xenvif *vif)
724{
e9ce7cb6
WL
725 struct xenvif_queue *queue = NULL;
726 unsigned int num_queues = vif->dev->real_num_tx_queues;
727 unsigned int queue_index;
0e59a4a5
ZK
728 /* Here we want to avoid timeout messages if an skb can be legitimately
729 * stuck somewhere else. Realistically this could be an another vif's
09350788
ZK
730 * internal or QDisc queue. That another vif also has this
731 * rx_drain_timeout_msecs timeout, but the timer only ditches the
732 * internal queue. After that, the QDisc queue can put in worst case
733 * XEN_NETIF_RX_RING_SIZE / MAX_SKB_FRAGS skbs into that another vif's
734 * internal queue, so we need several rounds of such timeouts until we
735 * can be sure that no another vif should have skb's from us. We are
0e59a4a5 736 * not sending more skb's, so newly stuck packets are not interesting
09350788
ZK
737 * for us here.
738 */
739 unsigned int worst_case_skb_lifetime = (rx_drain_timeout_msecs/1000) *
740 DIV_ROUND_UP(XENVIF_QUEUE_LENGTH, (XEN_NETIF_RX_RING_SIZE / MAX_SKB_FRAGS));
f53c3fe8 741
e9ce7cb6 742 unregister_netdev(vif->dev);
f53c3fe8 743
e9ce7cb6
WL
744 for (queue_index = 0; queue_index < num_queues; ++queue_index) {
745 queue = &vif->queues[queue_index];
e9ce7cb6 746 xenvif_wait_unmap_timeout(queue, worst_case_skb_lifetime);
8d3d53b3 747 xenvif_deinit_queue(queue);
e9ce7cb6
WL
748 }
749
750 /* Free the array of queues. The call below does not require
751 * rtnl_lock() because it happens after unregister_netdev().
752 */
753 netif_set_real_num_tx_queues(vif->dev, 0);
754 vfree(vif->queues);
755 vif->queues = NULL;
f942dc25 756
f942dc25 757 free_netdev(vif->dev);
b103f358 758
279f438e 759 module_put(THIS_MODULE);
f942dc25 760}