]>
Commit | Line | Data |
---|---|---|
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> |
0881e7bd | 34 | #include <linux/sched/task.h> |
f942dc25 IC |
35 | #include <linux/ethtool.h> |
36 | #include <linux/rtnetlink.h> | |
37 | #include <linux/if_vlan.h> | |
e7b599d7 | 38 | #include <linux/vmalloc.h> |
f942dc25 IC |
39 | |
40 | #include <xen/events.h> | |
41 | #include <asm/xen/hypercall.h> | |
f53c3fe8 | 42 | #include <xen/balloon.h> |
f942dc25 IC |
43 | |
44 | #define XENVIF_QUEUE_LENGTH 32 | |
b3f980bd | 45 | #define XENVIF_NAPI_WEIGHT 64 |
f942dc25 | 46 | |
f48da8b1 DV |
47 | /* Number of bytes allowed on the internal guest Rx queue. */ |
48 | #define XENVIF_RX_QUEUE_BYTES (XEN_NETIF_RX_RING_SIZE/2 * PAGE_SIZE) | |
49 | ||
a64bd934 WL |
50 | /* This function is used to set SKBTX_DEV_ZEROCOPY as well as |
51 | * increasing the inflight counter. We need to increase the inflight | |
52 | * counter because core driver calls into xenvif_zerocopy_callback | |
53 | * which calls xenvif_skb_zerocopy_complete. | |
54 | */ | |
55 | void xenvif_skb_zerocopy_prepare(struct xenvif_queue *queue, | |
56 | struct sk_buff *skb) | |
57 | { | |
58 | skb_shinfo(skb)->tx_flags |= SKBTX_DEV_ZEROCOPY; | |
59 | atomic_inc(&queue->inflight_packets); | |
60 | } | |
61 | ||
62 | void xenvif_skb_zerocopy_complete(struct xenvif_queue *queue) | |
63 | { | |
64 | atomic_dec(&queue->inflight_packets); | |
57b22906 RL |
65 | |
66 | /* Wake the dealloc thread _after_ decrementing inflight_packets so | |
67 | * that if kthread_stop() has already been called, the dealloc thread | |
68 | * does not wait forever with nothing to wake it. | |
69 | */ | |
70 | wake_up(&queue->dealloc_wq); | |
a64bd934 WL |
71 | } |
72 | ||
f942dc25 IC |
73 | int xenvif_schedulable(struct xenvif *vif) |
74 | { | |
3d1af1df | 75 | return netif_running(vif->dev) && |
f48da8b1 DV |
76 | test_bit(VIF_STATUS_CONNECTED, &vif->status) && |
77 | !vif->disabled; | |
f942dc25 IC |
78 | } |
79 | ||
e1f00a69 | 80 | static irqreturn_t xenvif_tx_interrupt(int irq, void *dev_id) |
f942dc25 | 81 | { |
e9ce7cb6 | 82 | struct xenvif_queue *queue = dev_id; |
f942dc25 | 83 | |
e9ce7cb6 WL |
84 | if (RING_HAS_UNCONSUMED_REQUESTS(&queue->tx)) |
85 | napi_schedule(&queue->napi); | |
f942dc25 | 86 | |
e1f00a69 WL |
87 | return IRQ_HANDLED; |
88 | } | |
89 | ||
38741d50 | 90 | static int xenvif_poll(struct napi_struct *napi, int budget) |
b3f980bd | 91 | { |
e9ce7cb6 WL |
92 | struct xenvif_queue *queue = |
93 | container_of(napi, struct xenvif_queue, napi); | |
b3f980bd WL |
94 | int work_done; |
95 | ||
e9d8b2c2 WL |
96 | /* This vif is rogue, we pretend we've there is nothing to do |
97 | * for this vif to deschedule it from NAPI. But this interface | |
98 | * will be turned off in thread context later. | |
99 | */ | |
2561cc15 | 100 | if (unlikely(queue->vif->disabled)) { |
e9d8b2c2 WL |
101 | napi_complete(napi); |
102 | return 0; | |
103 | } | |
104 | ||
e9ce7cb6 | 105 | work_done = xenvif_tx_action(queue, budget); |
b3f980bd WL |
106 | |
107 | if (work_done < budget) { | |
6ad20165 | 108 | napi_complete_done(napi, work_done); |
dfa523ae WL |
109 | /* If the queue is rate-limited, it shall be |
110 | * rescheduled in the timer callback. | |
111 | */ | |
112 | if (likely(!queue->rate_limited)) | |
113 | xenvif_napi_schedule_or_enable_events(queue); | |
b3f980bd WL |
114 | } |
115 | ||
116 | return work_done; | |
117 | } | |
118 | ||
e1f00a69 WL |
119 | static irqreturn_t xenvif_rx_interrupt(int irq, void *dev_id) |
120 | { | |
e9ce7cb6 | 121 | struct xenvif_queue *queue = dev_id; |
e1f00a69 | 122 | |
e9ce7cb6 | 123 | xenvif_kick_thread(queue); |
f942dc25 IC |
124 | |
125 | return IRQ_HANDLED; | |
126 | } | |
127 | ||
f51de243 | 128 | irqreturn_t xenvif_interrupt(int irq, void *dev_id) |
e1f00a69 WL |
129 | { |
130 | xenvif_tx_interrupt(irq, dev_id); | |
131 | xenvif_rx_interrupt(irq, dev_id); | |
132 | ||
133 | return IRQ_HANDLED; | |
134 | } | |
135 | ||
e9ce7cb6 WL |
136 | int xenvif_queue_stopped(struct xenvif_queue *queue) |
137 | { | |
138 | struct net_device *dev = queue->vif->dev; | |
139 | unsigned int id = queue->id; | |
140 | return netif_tx_queue_stopped(netdev_get_tx_queue(dev, id)); | |
141 | } | |
142 | ||
143 | void xenvif_wake_queue(struct xenvif_queue *queue) | |
144 | { | |
145 | struct net_device *dev = queue->vif->dev; | |
146 | unsigned int id = queue->id; | |
147 | netif_tx_wake_queue(netdev_get_tx_queue(dev, id)); | |
148 | } | |
149 | ||
40d8abde PD |
150 | static u16 xenvif_select_queue(struct net_device *dev, struct sk_buff *skb, |
151 | void *accel_priv, | |
152 | select_queue_fallback_t fallback) | |
153 | { | |
154 | struct xenvif *vif = netdev_priv(dev); | |
155 | unsigned int size = vif->hash.size; | |
37248764 ID |
156 | unsigned int num_queues; |
157 | ||
158 | /* If queues are not set up internally - always return 0 | |
159 | * as the packet going to be dropped anyway */ | |
160 | num_queues = READ_ONCE(vif->num_queues); | |
161 | if (num_queues < 1) | |
162 | return 0; | |
40d8abde | 163 | |
912e27e8 PD |
164 | if (vif->hash.alg == XEN_NETIF_CTRL_HASH_ALGORITHM_NONE) |
165 | return fallback(dev, skb) % dev->real_num_tx_queues; | |
40d8abde PD |
166 | |
167 | xenvif_set_skb_hash(vif, skb); | |
168 | ||
169 | if (size == 0) | |
170 | return skb_get_hash_raw(skb) % dev->real_num_tx_queues; | |
171 | ||
172 | return vif->hash.mapping[skb_get_hash_raw(skb) % size]; | |
173 | } | |
174 | ||
a13b09fb Y |
175 | static netdev_tx_t |
176 | xenvif_start_xmit(struct sk_buff *skb, struct net_device *dev) | |
f942dc25 IC |
177 | { |
178 | struct xenvif *vif = netdev_priv(dev); | |
e9ce7cb6 | 179 | struct xenvif_queue *queue = NULL; |
b17075d5 | 180 | unsigned int num_queues; |
e9ce7cb6 | 181 | u16 index; |
f48da8b1 | 182 | struct xenvif_rx_cb *cb; |
f942dc25 IC |
183 | |
184 | BUG_ON(skb->dev != dev); | |
185 | ||
b17075d5 ID |
186 | /* Drop the packet if queues are not set up. |
187 | * This handler should be called inside an RCU read section | |
188 | * so we don't need to enter it here explicitly. | |
189 | */ | |
190 | num_queues = READ_ONCE(vif->num_queues); | |
e9ce7cb6 WL |
191 | if (num_queues < 1) |
192 | goto drop; | |
193 | ||
194 | /* Obtain the queue to be used to transmit this packet */ | |
195 | index = skb_get_queue_mapping(skb); | |
196 | if (index >= num_queues) { | |
cc10f871 | 197 | pr_warn_ratelimited("Invalid queue %hu for packet on interface %s\n", |
e9ce7cb6 WL |
198 | index, vif->dev->name); |
199 | index %= num_queues; | |
200 | } | |
201 | queue = &vif->queues[index]; | |
202 | ||
203 | /* Drop the packet if queue is not ready */ | |
204 | if (queue->task == NULL || | |
205 | queue->dealloc_task == NULL || | |
f53c3fe8 | 206 | !xenvif_schedulable(vif)) |
f942dc25 IC |
207 | goto drop; |
208 | ||
210c34dc PD |
209 | if (vif->multicast_control && skb->pkt_type == PACKET_MULTICAST) { |
210 | struct ethhdr *eth = (struct ethhdr *)skb->data; | |
211 | ||
212 | if (!xenvif_mcast_match(vif, eth->h_dest)) | |
213 | goto drop; | |
214 | } | |
215 | ||
f48da8b1 | 216 | cb = XENVIF_RX_CB(skb); |
26c0e102 | 217 | cb->expires = jiffies + vif->drain_timeout; |
f942dc25 | 218 | |
912e27e8 PD |
219 | /* If there is no hash algorithm configured then make sure there |
220 | * is no hash information in the socket buffer otherwise it | |
221 | * would be incorrectly forwarded to the frontend. | |
222 | */ | |
223 | if (vif->hash.alg == XEN_NETIF_CTRL_HASH_ALGORITHM_NONE) | |
224 | skb_clear_hash(skb); | |
225 | ||
f48da8b1 | 226 | xenvif_rx_queue_tail(queue, skb); |
e9ce7cb6 | 227 | xenvif_kick_thread(queue); |
f942dc25 IC |
228 | |
229 | return NETDEV_TX_OK; | |
230 | ||
231 | drop: | |
232 | vif->dev->stats.tx_dropped++; | |
233 | dev_kfree_skb(skb); | |
234 | return NETDEV_TX_OK; | |
235 | } | |
236 | ||
f942dc25 IC |
237 | static struct net_device_stats *xenvif_get_stats(struct net_device *dev) |
238 | { | |
239 | struct xenvif *vif = netdev_priv(dev); | |
e9ce7cb6 | 240 | struct xenvif_queue *queue = NULL; |
b17075d5 | 241 | unsigned int num_queues; |
ebf692f8 MS |
242 | u64 rx_bytes = 0; |
243 | u64 rx_packets = 0; | |
244 | u64 tx_bytes = 0; | |
245 | u64 tx_packets = 0; | |
e9ce7cb6 WL |
246 | unsigned int index; |
247 | ||
b17075d5 ID |
248 | rcu_read_lock(); |
249 | num_queues = READ_ONCE(vif->num_queues); | |
e9ce7cb6 WL |
250 | |
251 | /* Aggregate tx and rx stats from each queue */ | |
b17075d5 | 252 | for (index = 0; index < num_queues; ++index) { |
e9ce7cb6 WL |
253 | queue = &vif->queues[index]; |
254 | rx_bytes += queue->stats.rx_bytes; | |
255 | rx_packets += queue->stats.rx_packets; | |
256 | tx_bytes += queue->stats.tx_bytes; | |
257 | tx_packets += queue->stats.tx_packets; | |
258 | } | |
259 | ||
b17075d5 | 260 | rcu_read_unlock(); |
f16f1df6 | 261 | |
e9ce7cb6 WL |
262 | vif->dev->stats.rx_bytes = rx_bytes; |
263 | vif->dev->stats.rx_packets = rx_packets; | |
264 | vif->dev->stats.tx_bytes = tx_bytes; | |
265 | vif->dev->stats.tx_packets = tx_packets; | |
266 | ||
f942dc25 IC |
267 | return &vif->dev->stats; |
268 | } | |
269 | ||
270 | static void xenvif_up(struct xenvif *vif) | |
271 | { | |
e9ce7cb6 | 272 | struct xenvif_queue *queue = NULL; |
f7b50c4e | 273 | unsigned int num_queues = vif->num_queues; |
e9ce7cb6 WL |
274 | unsigned int queue_index; |
275 | ||
276 | for (queue_index = 0; queue_index < num_queues; ++queue_index) { | |
277 | queue = &vif->queues[queue_index]; | |
278 | napi_enable(&queue->napi); | |
279 | enable_irq(queue->tx_irq); | |
280 | if (queue->tx_irq != queue->rx_irq) | |
281 | enable_irq(queue->rx_irq); | |
282 | xenvif_napi_schedule_or_enable_events(queue); | |
283 | } | |
f942dc25 IC |
284 | } |
285 | ||
286 | static void xenvif_down(struct xenvif *vif) | |
287 | { | |
e9ce7cb6 | 288 | struct xenvif_queue *queue = NULL; |
f7b50c4e | 289 | unsigned int num_queues = vif->num_queues; |
e9ce7cb6 WL |
290 | unsigned int queue_index; |
291 | ||
292 | for (queue_index = 0; queue_index < num_queues; ++queue_index) { | |
293 | queue = &vif->queues[queue_index]; | |
e9ce7cb6 WL |
294 | disable_irq(queue->tx_irq); |
295 | if (queue->tx_irq != queue->rx_irq) | |
296 | disable_irq(queue->rx_irq); | |
8fe78989 | 297 | napi_disable(&queue->napi); |
e9ce7cb6 WL |
298 | del_timer_sync(&queue->credit_timeout); |
299 | } | |
f942dc25 IC |
300 | } |
301 | ||
302 | static int xenvif_open(struct net_device *dev) | |
303 | { | |
304 | struct xenvif *vif = netdev_priv(dev); | |
3d1af1df | 305 | if (test_bit(VIF_STATUS_CONNECTED, &vif->status)) |
f942dc25 | 306 | xenvif_up(vif); |
e9ce7cb6 | 307 | netif_tx_start_all_queues(dev); |
f942dc25 IC |
308 | return 0; |
309 | } | |
310 | ||
311 | static int xenvif_close(struct net_device *dev) | |
312 | { | |
313 | struct xenvif *vif = netdev_priv(dev); | |
3d1af1df | 314 | if (test_bit(VIF_STATUS_CONNECTED, &vif->status)) |
f942dc25 | 315 | xenvif_down(vif); |
e9ce7cb6 | 316 | netif_tx_stop_all_queues(dev); |
f942dc25 IC |
317 | return 0; |
318 | } | |
319 | ||
320 | static int xenvif_change_mtu(struct net_device *dev, int mtu) | |
321 | { | |
322 | struct xenvif *vif = netdev_priv(dev); | |
d0c2c997 | 323 | int max = vif->can_sg ? ETH_MAX_MTU - VLAN_ETH_HLEN : ETH_DATA_LEN; |
f942dc25 IC |
324 | |
325 | if (mtu > max) | |
326 | return -EINVAL; | |
327 | dev->mtu = mtu; | |
328 | return 0; | |
329 | } | |
330 | ||
c8f44aff MM |
331 | static netdev_features_t xenvif_fix_features(struct net_device *dev, |
332 | netdev_features_t features) | |
f942dc25 IC |
333 | { |
334 | struct xenvif *vif = netdev_priv(dev); | |
f942dc25 | 335 | |
47103041 MM |
336 | if (!vif->can_sg) |
337 | features &= ~NETIF_F_SG; | |
fedbc8c1 | 338 | if (~(vif->gso_mask) & GSO_BIT(TCPV4)) |
47103041 | 339 | features &= ~NETIF_F_TSO; |
fedbc8c1 | 340 | if (~(vif->gso_mask) & GSO_BIT(TCPV6)) |
82cada22 | 341 | features &= ~NETIF_F_TSO6; |
146c8a77 | 342 | if (!vif->ip_csum) |
47103041 | 343 | features &= ~NETIF_F_IP_CSUM; |
146c8a77 PD |
344 | if (!vif->ipv6_csum) |
345 | features &= ~NETIF_F_IPV6_CSUM; | |
f942dc25 | 346 | |
47103041 | 347 | return features; |
f942dc25 IC |
348 | } |
349 | ||
350 | static const struct xenvif_stat { | |
351 | char name[ETH_GSTRING_LEN]; | |
352 | u16 offset; | |
353 | } xenvif_stats[] = { | |
354 | { | |
355 | "rx_gso_checksum_fixup", | |
e9ce7cb6 | 356 | offsetof(struct xenvif_stats, rx_gso_checksum_fixup) |
f942dc25 | 357 | }, |
1bb332af ZK |
358 | /* If (sent != success + fail), there are probably packets never |
359 | * freed up properly! | |
360 | */ | |
361 | { | |
362 | "tx_zerocopy_sent", | |
e9ce7cb6 | 363 | offsetof(struct xenvif_stats, tx_zerocopy_sent), |
1bb332af ZK |
364 | }, |
365 | { | |
366 | "tx_zerocopy_success", | |
e9ce7cb6 | 367 | offsetof(struct xenvif_stats, tx_zerocopy_success), |
1bb332af ZK |
368 | }, |
369 | { | |
370 | "tx_zerocopy_fail", | |
e9ce7cb6 | 371 | offsetof(struct xenvif_stats, tx_zerocopy_fail) |
1bb332af | 372 | }, |
e3377f36 ZK |
373 | /* Number of packets exceeding MAX_SKB_FRAG slots. You should use |
374 | * a guest with the same MAX_SKB_FRAG | |
375 | */ | |
376 | { | |
377 | "tx_frag_overflow", | |
e9ce7cb6 | 378 | offsetof(struct xenvif_stats, tx_frag_overflow) |
e3377f36 | 379 | }, |
f942dc25 IC |
380 | }; |
381 | ||
382 | static int xenvif_get_sset_count(struct net_device *dev, int string_set) | |
383 | { | |
384 | switch (string_set) { | |
385 | case ETH_SS_STATS: | |
386 | return ARRAY_SIZE(xenvif_stats); | |
387 | default: | |
388 | return -EINVAL; | |
389 | } | |
390 | } | |
391 | ||
392 | static void xenvif_get_ethtool_stats(struct net_device *dev, | |
393 | struct ethtool_stats *stats, u64 * data) | |
394 | { | |
e9ce7cb6 | 395 | struct xenvif *vif = netdev_priv(dev); |
b17075d5 | 396 | unsigned int num_queues; |
f942dc25 | 397 | int i; |
e9ce7cb6 | 398 | unsigned int queue_index; |
e9ce7cb6 | 399 | |
b17075d5 ID |
400 | rcu_read_lock(); |
401 | num_queues = READ_ONCE(vif->num_queues); | |
402 | ||
e9ce7cb6 WL |
403 | for (i = 0; i < ARRAY_SIZE(xenvif_stats); i++) { |
404 | unsigned long accum = 0; | |
405 | for (queue_index = 0; queue_index < num_queues; ++queue_index) { | |
d63951d7 | 406 | void *vif_stats = &vif->queues[queue_index].stats; |
e9ce7cb6 WL |
407 | accum += *(unsigned long *)(vif_stats + xenvif_stats[i].offset); |
408 | } | |
409 | data[i] = accum; | |
410 | } | |
b17075d5 ID |
411 | |
412 | rcu_read_unlock(); | |
f942dc25 IC |
413 | } |
414 | ||
415 | static void xenvif_get_strings(struct net_device *dev, u32 stringset, u8 * data) | |
416 | { | |
417 | int i; | |
418 | ||
419 | switch (stringset) { | |
420 | case ETH_SS_STATS: | |
421 | for (i = 0; i < ARRAY_SIZE(xenvif_stats); i++) | |
422 | memcpy(data + i * ETH_GSTRING_LEN, | |
423 | xenvif_stats[i].name, ETH_GSTRING_LEN); | |
424 | break; | |
425 | } | |
426 | } | |
427 | ||
813abbba | 428 | static const struct ethtool_ops xenvif_ethtool_ops = { |
f942dc25 IC |
429 | .get_link = ethtool_op_get_link, |
430 | ||
431 | .get_sset_count = xenvif_get_sset_count, | |
432 | .get_ethtool_stats = xenvif_get_ethtool_stats, | |
433 | .get_strings = xenvif_get_strings, | |
434 | }; | |
435 | ||
813abbba | 436 | static const struct net_device_ops xenvif_netdev_ops = { |
40d8abde | 437 | .ndo_select_queue = xenvif_select_queue, |
f942dc25 IC |
438 | .ndo_start_xmit = xenvif_start_xmit, |
439 | .ndo_get_stats = xenvif_get_stats, | |
440 | .ndo_open = xenvif_open, | |
441 | .ndo_stop = xenvif_close, | |
442 | .ndo_change_mtu = xenvif_change_mtu, | |
47103041 | 443 | .ndo_fix_features = xenvif_fix_features, |
4a633a60 MW |
444 | .ndo_set_mac_address = eth_mac_addr, |
445 | .ndo_validate_addr = eth_validate_addr, | |
f942dc25 IC |
446 | }; |
447 | ||
448 | struct xenvif *xenvif_alloc(struct device *parent, domid_t domid, | |
449 | unsigned int handle) | |
450 | { | |
451 | int err; | |
452 | struct net_device *dev; | |
453 | struct xenvif *vif; | |
454 | char name[IFNAMSIZ] = {}; | |
455 | ||
456 | snprintf(name, IFNAMSIZ - 1, "vif%u.%u", domid, handle); | |
8d3d53b3 AB |
457 | /* Allocate a netdev with the max. supported number of queues. |
458 | * When the guest selects the desired number, it will be updated | |
f7b50c4e | 459 | * via netif_set_real_num_*_queues(). |
8d3d53b3 | 460 | */ |
c835a677 TG |
461 | dev = alloc_netdev_mq(sizeof(struct xenvif), name, NET_NAME_UNKNOWN, |
462 | ether_setup, xenvif_max_queues); | |
f942dc25 | 463 | if (dev == NULL) { |
b3f980bd | 464 | pr_warn("Could not allocate netdev for %s\n", name); |
f942dc25 IC |
465 | return ERR_PTR(-ENOMEM); |
466 | } | |
467 | ||
468 | SET_NETDEV_DEV(dev, parent); | |
469 | ||
470 | vif = netdev_priv(dev); | |
ac3d5ac2 | 471 | |
f942dc25 IC |
472 | vif->domid = domid; |
473 | vif->handle = handle; | |
f942dc25 | 474 | vif->can_sg = 1; |
146c8a77 | 475 | vif->ip_csum = 1; |
f942dc25 | 476 | vif->dev = dev; |
e9d8b2c2 | 477 | vif->disabled = false; |
26c0e102 DV |
478 | vif->drain_timeout = msecs_to_jiffies(rx_drain_timeout_msecs); |
479 | vif->stall_timeout = msecs_to_jiffies(rx_stall_timeout_msecs); | |
e9d8b2c2 | 480 | |
f7b50c4e | 481 | /* Start out with no queues. */ |
e9ce7cb6 | 482 | vif->queues = NULL; |
f7b50c4e | 483 | vif->num_queues = 0; |
09350788 | 484 | |
ecf08d2d | 485 | spin_lock_init(&vif->lock); |
210c34dc | 486 | INIT_LIST_HEAD(&vif->fe_mcast_addr); |
ecf08d2d | 487 | |
f942dc25 | 488 | dev->netdev_ops = &xenvif_netdev_ops; |
146c8a77 PD |
489 | dev->hw_features = NETIF_F_SG | |
490 | NETIF_F_IP_CSUM | NETIF_F_IPV6_CSUM | | |
2167ca02 | 491 | NETIF_F_TSO | NETIF_F_TSO6 | NETIF_F_FRAGLIST; |
7365bcfa | 492 | dev->features = dev->hw_features | NETIF_F_RXCSUM; |
7ad24ea4 | 493 | dev->ethtool_ops = &xenvif_ethtool_ops; |
f942dc25 IC |
494 | |
495 | dev->tx_queue_len = XENVIF_QUEUE_LENGTH; | |
496 | ||
e1043a4b | 497 | dev->min_mtu = ETH_MIN_MTU; |
d0c2c997 JW |
498 | dev->max_mtu = ETH_MAX_MTU - VLAN_ETH_HLEN; |
499 | ||
f942dc25 IC |
500 | /* |
501 | * Initialise a dummy MAC address. We choose the numerically | |
502 | * largest non-broadcast address to prevent the address getting | |
503 | * stolen by an Ethernet bridge for STP purposes. | |
504 | * (FE:FF:FF:FF:FF:FF) | |
505 | */ | |
3b6ed26d | 506 | eth_broadcast_addr(dev->dev_addr); |
f942dc25 IC |
507 | dev->dev_addr[0] &= ~0x01; |
508 | ||
509 | netif_carrier_off(dev); | |
510 | ||
511 | err = register_netdev(dev); | |
512 | if (err) { | |
513 | netdev_warn(dev, "Could not register device: err=%d\n", err); | |
514 | free_netdev(dev); | |
515 | return ERR_PTR(err); | |
516 | } | |
517 | ||
518 | netdev_dbg(dev, "Successfully created xenvif\n"); | |
279f438e PD |
519 | |
520 | __module_get(THIS_MODULE); | |
521 | ||
f942dc25 IC |
522 | return vif; |
523 | } | |
524 | ||
e9ce7cb6 WL |
525 | int xenvif_init_queue(struct xenvif_queue *queue) |
526 | { | |
527 | int err, i; | |
528 | ||
529 | queue->credit_bytes = queue->remaining_credit = ~0UL; | |
530 | queue->credit_usec = 0UL; | |
cac6a8f9 | 531 | timer_setup(&queue->credit_timeout, xenvif_tx_credit_callback, 0); |
e9ce7cb6 WL |
532 | queue->credit_window_start = get_jiffies_64(); |
533 | ||
f48da8b1 DV |
534 | queue->rx_queue_max = XENVIF_RX_QUEUE_BYTES; |
535 | ||
e9ce7cb6 WL |
536 | skb_queue_head_init(&queue->rx_queue); |
537 | skb_queue_head_init(&queue->tx_queue); | |
538 | ||
539 | queue->pending_cons = 0; | |
540 | queue->pending_prod = MAX_PENDING_REQS; | |
541 | for (i = 0; i < MAX_PENDING_REQS; ++i) | |
542 | queue->pending_ring[i] = i; | |
543 | ||
544 | spin_lock_init(&queue->callback_lock); | |
545 | spin_lock_init(&queue->response_lock); | |
546 | ||
547 | /* If ballooning is disabled, this will consume real memory, so you | |
548 | * better enable it. The long term solution would be to use just a | |
549 | * bunch of valid page descriptors, without dependency on ballooning | |
550 | */ | |
ff4b156f DV |
551 | err = gnttab_alloc_pages(MAX_PENDING_REQS, |
552 | queue->mmap_pages); | |
e9ce7cb6 WL |
553 | if (err) { |
554 | netdev_err(queue->vif->dev, "Could not reserve mmap_pages\n"); | |
555 | return -ENOMEM; | |
556 | } | |
557 | ||
558 | for (i = 0; i < MAX_PENDING_REQS; i++) { | |
559 | queue->pending_tx_info[i].callback_struct = (struct ubuf_info) | |
560 | { .callback = xenvif_zerocopy_callback, | |
cc8737a5 WB |
561 | { { .ctx = NULL, |
562 | .desc = i } } }; | |
e9ce7cb6 WL |
563 | queue->grant_tx_handle[i] = NETBACK_INVALID_HANDLE; |
564 | } | |
565 | ||
e9ce7cb6 WL |
566 | return 0; |
567 | } | |
568 | ||
569 | void xenvif_carrier_on(struct xenvif *vif) | |
570 | { | |
571 | rtnl_lock(); | |
572 | if (!vif->can_sg && vif->dev->mtu > ETH_DATA_LEN) | |
573 | dev_set_mtu(vif->dev, ETH_DATA_LEN); | |
574 | netdev_update_features(vif->dev); | |
3d1af1df | 575 | set_bit(VIF_STATUS_CONNECTED, &vif->status); |
e9ce7cb6 WL |
576 | if (netif_running(vif->dev)) |
577 | xenvif_up(vif); | |
578 | rtnl_unlock(); | |
579 | } | |
580 | ||
4e15ee2c PD |
581 | int xenvif_connect_ctrl(struct xenvif *vif, grant_ref_t ring_ref, |
582 | unsigned int evtchn) | |
583 | { | |
584 | struct net_device *dev = vif->dev; | |
585 | void *addr; | |
586 | struct xen_netif_ctrl_sring *shared; | |
0364a882 | 587 | int err; |
4e15ee2c PD |
588 | |
589 | err = xenbus_map_ring_valloc(xenvif_to_xenbus_device(vif), | |
590 | &ring_ref, 1, &addr); | |
591 | if (err) | |
592 | goto err; | |
593 | ||
594 | shared = (struct xen_netif_ctrl_sring *)addr; | |
595 | BACK_RING_INIT(&vif->ctrl, shared, XEN_PAGE_SIZE); | |
596 | ||
0364a882 | 597 | err = bind_interdomain_evtchn_to_irq(vif->domid, evtchn); |
4e15ee2c PD |
598 | if (err < 0) |
599 | goto err_unmap; | |
600 | ||
601 | vif->ctrl_irq = err; | |
602 | ||
40d8abde PD |
603 | xenvif_init_hash(vif); |
604 | ||
0364a882 JG |
605 | err = request_threaded_irq(vif->ctrl_irq, NULL, xenvif_ctrl_irq_fn, |
606 | IRQF_ONESHOT, "xen-netback-ctrl", vif); | |
607 | if (err) { | |
608 | pr_warn("Could not setup irq handler for %s\n", dev->name); | |
4e15ee2c PD |
609 | goto err_deinit; |
610 | } | |
611 | ||
4e15ee2c PD |
612 | return 0; |
613 | ||
614 | err_deinit: | |
40d8abde | 615 | xenvif_deinit_hash(vif); |
4e15ee2c PD |
616 | unbind_from_irqhandler(vif->ctrl_irq, vif); |
617 | vif->ctrl_irq = 0; | |
618 | ||
619 | err_unmap: | |
620 | xenbus_unmap_ring_vfree(xenvif_to_xenbus_device(vif), | |
621 | vif->ctrl.sring); | |
622 | vif->ctrl.sring = NULL; | |
623 | ||
624 | err: | |
625 | return err; | |
626 | } | |
627 | ||
628 | int xenvif_connect_data(struct xenvif_queue *queue, | |
629 | unsigned long tx_ring_ref, | |
630 | unsigned long rx_ring_ref, | |
631 | unsigned int tx_evtchn, | |
632 | unsigned int rx_evtchn) | |
f942dc25 | 633 | { |
67fa3660 | 634 | struct task_struct *task; |
f942dc25 IC |
635 | int err = -ENOMEM; |
636 | ||
e9ce7cb6 WL |
637 | BUG_ON(queue->tx_irq); |
638 | BUG_ON(queue->task); | |
639 | BUG_ON(queue->dealloc_task); | |
f942dc25 | 640 | |
4e15ee2c PD |
641 | err = xenvif_map_frontend_data_rings(queue, tx_ring_ref, |
642 | rx_ring_ref); | |
f942dc25 IC |
643 | if (err < 0) |
644 | goto err; | |
645 | ||
e9ce7cb6 WL |
646 | init_waitqueue_head(&queue->wq); |
647 | init_waitqueue_head(&queue->dealloc_wq); | |
a64bd934 | 648 | atomic_set(&queue->inflight_packets, 0); |
ca2f09f2 | 649 | |
e24f8191 WL |
650 | netif_napi_add(queue->vif->dev, &queue->napi, xenvif_poll, |
651 | XENVIF_NAPI_WEIGHT); | |
652 | ||
e1f00a69 WL |
653 | if (tx_evtchn == rx_evtchn) { |
654 | /* feature-split-event-channels == 0 */ | |
655 | err = bind_interdomain_evtchn_to_irqhandler( | |
e9ce7cb6 WL |
656 | queue->vif->domid, tx_evtchn, xenvif_interrupt, 0, |
657 | queue->name, queue); | |
e1f00a69 WL |
658 | if (err < 0) |
659 | goto err_unmap; | |
e9ce7cb6 WL |
660 | queue->tx_irq = queue->rx_irq = err; |
661 | disable_irq(queue->tx_irq); | |
e1f00a69 WL |
662 | } else { |
663 | /* feature-split-event-channels == 1 */ | |
e9ce7cb6 WL |
664 | snprintf(queue->tx_irq_name, sizeof(queue->tx_irq_name), |
665 | "%s-tx", queue->name); | |
e1f00a69 | 666 | err = bind_interdomain_evtchn_to_irqhandler( |
e9ce7cb6 WL |
667 | queue->vif->domid, tx_evtchn, xenvif_tx_interrupt, 0, |
668 | queue->tx_irq_name, queue); | |
e1f00a69 WL |
669 | if (err < 0) |
670 | goto err_unmap; | |
e9ce7cb6 WL |
671 | queue->tx_irq = err; |
672 | disable_irq(queue->tx_irq); | |
e1f00a69 | 673 | |
e9ce7cb6 WL |
674 | snprintf(queue->rx_irq_name, sizeof(queue->rx_irq_name), |
675 | "%s-rx", queue->name); | |
e1f00a69 | 676 | err = bind_interdomain_evtchn_to_irqhandler( |
e9ce7cb6 WL |
677 | queue->vif->domid, rx_evtchn, xenvif_rx_interrupt, 0, |
678 | queue->rx_irq_name, queue); | |
e1f00a69 WL |
679 | if (err < 0) |
680 | goto err_tx_unbind; | |
e9ce7cb6 WL |
681 | queue->rx_irq = err; |
682 | disable_irq(queue->rx_irq); | |
e1f00a69 | 683 | } |
f942dc25 | 684 | |
ecf08d2d DV |
685 | queue->stalled = true; |
686 | ||
121fa4b7 | 687 | task = kthread_create(xenvif_kthread_guest_rx, |
e9ce7cb6 | 688 | (void *)queue, "%s-guest-rx", queue->name); |
67fa3660 | 689 | if (IS_ERR(task)) { |
e9ce7cb6 | 690 | pr_warn("Could not allocate kthread for %s\n", queue->name); |
67fa3660 | 691 | err = PTR_ERR(task); |
b3f980bd WL |
692 | goto err_rx_unbind; |
693 | } | |
e9ce7cb6 | 694 | queue->task = task; |
42b5212f | 695 | get_task_struct(task); |
67fa3660 | 696 | |
f53c3fe8 | 697 | task = kthread_create(xenvif_dealloc_kthread, |
e9ce7cb6 | 698 | (void *)queue, "%s-dealloc", queue->name); |
f53c3fe8 | 699 | if (IS_ERR(task)) { |
e9ce7cb6 | 700 | pr_warn("Could not allocate kthread for %s\n", queue->name); |
f53c3fe8 ZK |
701 | err = PTR_ERR(task); |
702 | goto err_rx_unbind; | |
703 | } | |
e9ce7cb6 | 704 | queue->dealloc_task = task; |
f53c3fe8 | 705 | |
e9ce7cb6 WL |
706 | wake_up_process(queue->task); |
707 | wake_up_process(queue->dealloc_task); | |
b3f980bd | 708 | |
f942dc25 | 709 | return 0; |
b3f980bd WL |
710 | |
711 | err_rx_unbind: | |
e9ce7cb6 WL |
712 | unbind_from_irqhandler(queue->rx_irq, queue); |
713 | queue->rx_irq = 0; | |
e1f00a69 | 714 | err_tx_unbind: |
e9ce7cb6 WL |
715 | unbind_from_irqhandler(queue->tx_irq, queue); |
716 | queue->tx_irq = 0; | |
f942dc25 | 717 | err_unmap: |
4e15ee2c | 718 | xenvif_unmap_frontend_data_rings(queue); |
4a658527 | 719 | netif_napi_del(&queue->napi); |
f942dc25 IC |
720 | err: |
721 | return err; | |
722 | } | |
723 | ||
48856286 | 724 | void xenvif_carrier_off(struct xenvif *vif) |
f942dc25 IC |
725 | { |
726 | struct net_device *dev = vif->dev; | |
48856286 IC |
727 | |
728 | rtnl_lock(); | |
3d1af1df ZK |
729 | if (test_and_clear_bit(VIF_STATUS_CONNECTED, &vif->status)) { |
730 | netif_carrier_off(dev); /* discard queued packets */ | |
731 | if (netif_running(dev)) | |
732 | xenvif_down(vif); | |
733 | } | |
48856286 | 734 | rtnl_unlock(); |
48856286 IC |
735 | } |
736 | ||
4e15ee2c | 737 | void xenvif_disconnect_data(struct xenvif *vif) |
48856286 | 738 | { |
e9ce7cb6 | 739 | struct xenvif_queue *queue = NULL; |
f7b50c4e | 740 | unsigned int num_queues = vif->num_queues; |
e9ce7cb6 WL |
741 | unsigned int queue_index; |
742 | ||
3d1af1df | 743 | xenvif_carrier_off(vif); |
f942dc25 | 744 | |
e9ce7cb6 WL |
745 | for (queue_index = 0; queue_index < num_queues; ++queue_index) { |
746 | queue = &vif->queues[queue_index]; | |
db739ef3 | 747 | |
ea2c5e13 WL |
748 | netif_napi_del(&queue->napi); |
749 | ||
e9ce7cb6 | 750 | if (queue->task) { |
e9ce7cb6 | 751 | kthread_stop(queue->task); |
42b5212f | 752 | put_task_struct(queue->task); |
e9ce7cb6 WL |
753 | queue->task = NULL; |
754 | } | |
f53c3fe8 | 755 | |
e9ce7cb6 WL |
756 | if (queue->dealloc_task) { |
757 | kthread_stop(queue->dealloc_task); | |
758 | queue->dealloc_task = NULL; | |
759 | } | |
760 | ||
761 | if (queue->tx_irq) { | |
762 | if (queue->tx_irq == queue->rx_irq) | |
763 | unbind_from_irqhandler(queue->tx_irq, queue); | |
764 | else { | |
765 | unbind_from_irqhandler(queue->tx_irq, queue); | |
766 | unbind_from_irqhandler(queue->rx_irq, queue); | |
767 | } | |
768 | queue->tx_irq = 0; | |
e1f00a69 | 769 | } |
f942dc25 | 770 | |
4e15ee2c | 771 | xenvif_unmap_frontend_data_rings(queue); |
e9ce7cb6 | 772 | } |
210c34dc PD |
773 | |
774 | xenvif_mcast_addr_list_free(vif); | |
279f438e PD |
775 | } |
776 | ||
4e15ee2c PD |
777 | void xenvif_disconnect_ctrl(struct xenvif *vif) |
778 | { | |
4e15ee2c | 779 | if (vif->ctrl_irq) { |
c0fcded2 | 780 | xenvif_deinit_hash(vif); |
4e15ee2c PD |
781 | unbind_from_irqhandler(vif->ctrl_irq, vif); |
782 | vif->ctrl_irq = 0; | |
783 | } | |
784 | ||
785 | if (vif->ctrl.sring) { | |
786 | xenbus_unmap_ring_vfree(xenvif_to_xenbus_device(vif), | |
787 | vif->ctrl.sring); | |
788 | vif->ctrl.sring = NULL; | |
789 | } | |
790 | } | |
791 | ||
8d3d53b3 AB |
792 | /* Reverse the relevant parts of xenvif_init_queue(). |
793 | * Used for queue teardown from xenvif_free(), and on the | |
794 | * error handling paths in xenbus.c:connect(). | |
795 | */ | |
796 | void xenvif_deinit_queue(struct xenvif_queue *queue) | |
797 | { | |
ff4b156f | 798 | gnttab_free_pages(MAX_PENDING_REQS, queue->mmap_pages); |
8d3d53b3 AB |
799 | } |
800 | ||
279f438e PD |
801 | void xenvif_free(struct xenvif *vif) |
802 | { | |
9c6f3ffe | 803 | struct xenvif_queue *queues = vif->queues; |
f7b50c4e | 804 | unsigned int num_queues = vif->num_queues; |
e9ce7cb6 | 805 | unsigned int queue_index; |
f53c3fe8 | 806 | |
e9ce7cb6 | 807 | unregister_netdev(vif->dev); |
f942dc25 | 808 | free_netdev(vif->dev); |
b103f358 | 809 | |
9c6f3ffe DV |
810 | for (queue_index = 0; queue_index < num_queues; ++queue_index) |
811 | xenvif_deinit_queue(&queues[queue_index]); | |
812 | vfree(queues); | |
813 | ||
279f438e | 814 | module_put(THIS_MODULE); |
f942dc25 | 815 | } |