xref: /openbmc/linux/drivers/net/hyperv/netvsc_drv.c (revision b9ccfda2)
1 /*
2  * Copyright (c) 2009, Microsoft Corporation.
3  *
4  * This program is free software; you can redistribute it and/or modify it
5  * under the terms and conditions of the GNU General Public License,
6  * version 2, as published by the Free Software Foundation.
7  *
8  * This program is distributed in the hope it will be useful, but WITHOUT
9  * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
10  * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
11  * more details.
12  *
13  * You should have received a copy of the GNU General Public License along with
14  * this program; if not, write to the Free Software Foundation, Inc., 59 Temple
15  * Place - Suite 330, Boston, MA 02111-1307 USA.
16  *
17  * Authors:
18  *   Haiyang Zhang <haiyangz@microsoft.com>
19  *   Hank Janssen  <hjanssen@microsoft.com>
20  */
21 #define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
22 
23 #include <linux/init.h>
24 #include <linux/atomic.h>
25 #include <linux/module.h>
26 #include <linux/highmem.h>
27 #include <linux/device.h>
28 #include <linux/io.h>
29 #include <linux/delay.h>
30 #include <linux/netdevice.h>
31 #include <linux/inetdevice.h>
32 #include <linux/etherdevice.h>
33 #include <linux/skbuff.h>
34 #include <linux/in.h>
35 #include <linux/slab.h>
36 #include <net/arp.h>
37 #include <net/route.h>
38 #include <net/sock.h>
39 #include <net/pkt_sched.h>
40 
41 #include "hyperv_net.h"
42 
43 struct net_device_context {
44 	/* point back to our device context */
45 	struct hv_device *device_ctx;
46 	struct delayed_work dwork;
47 	struct work_struct work;
48 };
49 
50 
51 static int ring_size = 128;
52 module_param(ring_size, int, S_IRUGO);
53 MODULE_PARM_DESC(ring_size, "Ring buffer size (# of pages)");
54 
55 static void do_set_multicast(struct work_struct *w)
56 {
57 	struct net_device_context *ndevctx =
58 		container_of(w, struct net_device_context, work);
59 	struct netvsc_device *nvdev;
60 	struct rndis_device *rdev;
61 
62 	nvdev = hv_get_drvdata(ndevctx->device_ctx);
63 	if (nvdev == NULL || nvdev->ndev == NULL)
64 		return;
65 
66 	rdev = nvdev->extension;
67 	if (rdev == NULL)
68 		return;
69 
70 	if (nvdev->ndev->flags & IFF_PROMISC)
71 		rndis_filter_set_packet_filter(rdev,
72 			NDIS_PACKET_TYPE_PROMISCUOUS);
73 	else
74 		rndis_filter_set_packet_filter(rdev,
75 			NDIS_PACKET_TYPE_BROADCAST |
76 			NDIS_PACKET_TYPE_ALL_MULTICAST |
77 			NDIS_PACKET_TYPE_DIRECTED);
78 }
79 
80 static void netvsc_set_multicast_list(struct net_device *net)
81 {
82 	struct net_device_context *net_device_ctx = netdev_priv(net);
83 
84 	schedule_work(&net_device_ctx->work);
85 }
86 
87 static int netvsc_open(struct net_device *net)
88 {
89 	struct net_device_context *net_device_ctx = netdev_priv(net);
90 	struct hv_device *device_obj = net_device_ctx->device_ctx;
91 	int ret = 0;
92 
93 	/* Open up the device */
94 	ret = rndis_filter_open(device_obj);
95 	if (ret != 0) {
96 		netdev_err(net, "unable to open device (ret %d).\n", ret);
97 		return ret;
98 	}
99 
100 	netif_start_queue(net);
101 
102 	return ret;
103 }
104 
105 static int netvsc_close(struct net_device *net)
106 {
107 	struct net_device_context *net_device_ctx = netdev_priv(net);
108 	struct hv_device *device_obj = net_device_ctx->device_ctx;
109 	int ret;
110 
111 	netif_tx_disable(net);
112 
113 	/* Make sure netvsc_set_multicast_list doesn't re-enable filter! */
114 	cancel_work_sync(&net_device_ctx->work);
115 	ret = rndis_filter_close(device_obj);
116 	if (ret != 0)
117 		netdev_err(net, "unable to close device (ret %d).\n", ret);
118 
119 	return ret;
120 }
121 
122 static void netvsc_xmit_completion(void *context)
123 {
124 	struct hv_netvsc_packet *packet = (struct hv_netvsc_packet *)context;
125 	struct sk_buff *skb = (struct sk_buff *)
126 		(unsigned long)packet->completion.send.send_completion_tid;
127 
128 	kfree(packet);
129 
130 	if (skb)
131 		dev_kfree_skb_any(skb);
132 }
133 
134 static int netvsc_start_xmit(struct sk_buff *skb, struct net_device *net)
135 {
136 	struct net_device_context *net_device_ctx = netdev_priv(net);
137 	struct hv_netvsc_packet *packet;
138 	int ret;
139 	unsigned int i, num_pages, npg_data;
140 
141 	/* Add multipages for skb->data and additional 2 for RNDIS */
142 	npg_data = (((unsigned long)skb->data + skb_headlen(skb) - 1)
143 		>> PAGE_SHIFT) - ((unsigned long)skb->data >> PAGE_SHIFT) + 1;
144 	num_pages = skb_shinfo(skb)->nr_frags + npg_data + 2;
145 
146 	/* Allocate a netvsc packet based on # of frags. */
147 	packet = kzalloc(sizeof(struct hv_netvsc_packet) +
148 			 (num_pages * sizeof(struct hv_page_buffer)) +
149 			 sizeof(struct rndis_filter_packet) +
150 			 NDIS_VLAN_PPI_SIZE, GFP_ATOMIC);
151 	if (!packet) {
152 		/* out of memory, drop packet */
153 		netdev_err(net, "unable to allocate hv_netvsc_packet\n");
154 
155 		dev_kfree_skb(skb);
156 		net->stats.tx_dropped++;
157 		return NETDEV_TX_OK;
158 	}
159 
160 	packet->vlan_tci = skb->vlan_tci;
161 
162 	packet->extension = (void *)(unsigned long)packet +
163 				sizeof(struct hv_netvsc_packet) +
164 				    (num_pages * sizeof(struct hv_page_buffer));
165 
166 	/* If the rndis msg goes beyond 1 page, we will add 1 later */
167 	packet->page_buf_cnt = num_pages - 1;
168 
169 	/* Initialize it from the skb */
170 	packet->total_data_buflen = skb->len;
171 
172 	/* Start filling in the page buffers starting after RNDIS buffer. */
173 	packet->page_buf[1].pfn = virt_to_phys(skb->data) >> PAGE_SHIFT;
174 	packet->page_buf[1].offset
175 		= (unsigned long)skb->data & (PAGE_SIZE - 1);
176 	if (npg_data == 1)
177 		packet->page_buf[1].len = skb_headlen(skb);
178 	else
179 		packet->page_buf[1].len = PAGE_SIZE
180 			- packet->page_buf[1].offset;
181 
182 	for (i = 2; i <= npg_data; i++) {
183 		packet->page_buf[i].pfn = virt_to_phys(skb->data
184 			+ PAGE_SIZE * (i-1)) >> PAGE_SHIFT;
185 		packet->page_buf[i].offset = 0;
186 		packet->page_buf[i].len = PAGE_SIZE;
187 	}
188 	if (npg_data > 1)
189 		packet->page_buf[npg_data].len = (((unsigned long)skb->data
190 			+ skb_headlen(skb) - 1) & (PAGE_SIZE - 1)) + 1;
191 
192 	/* Additional fragments are after SKB data */
193 	for (i = 0; i < skb_shinfo(skb)->nr_frags; i++) {
194 		const skb_frag_t *f = &skb_shinfo(skb)->frags[i];
195 
196 		packet->page_buf[i+npg_data+1].pfn =
197 			page_to_pfn(skb_frag_page(f));
198 		packet->page_buf[i+npg_data+1].offset = f->page_offset;
199 		packet->page_buf[i+npg_data+1].len = skb_frag_size(f);
200 	}
201 
202 	/* Set the completion routine */
203 	packet->completion.send.send_completion = netvsc_xmit_completion;
204 	packet->completion.send.send_completion_ctx = packet;
205 	packet->completion.send.send_completion_tid = (unsigned long)skb;
206 
207 	ret = rndis_filter_send(net_device_ctx->device_ctx,
208 				  packet);
209 	if (ret == 0) {
210 		net->stats.tx_bytes += skb->len;
211 		net->stats.tx_packets++;
212 	} else {
213 		kfree(packet);
214 		if (ret != -EAGAIN) {
215 			dev_kfree_skb_any(skb);
216 			net->stats.tx_dropped++;
217 		}
218 	}
219 
220 	return (ret == -EAGAIN) ? NETDEV_TX_BUSY : NETDEV_TX_OK;
221 }
222 
223 /*
224  * netvsc_linkstatus_callback - Link up/down notification
225  */
226 void netvsc_linkstatus_callback(struct hv_device *device_obj,
227 				       unsigned int status)
228 {
229 	struct net_device *net;
230 	struct net_device_context *ndev_ctx;
231 	struct netvsc_device *net_device;
232 
233 	net_device = hv_get_drvdata(device_obj);
234 	net = net_device->ndev;
235 
236 	if (!net) {
237 		netdev_err(net, "got link status but net device "
238 				"not initialized yet\n");
239 		return;
240 	}
241 
242 	if (status == 1) {
243 		netif_carrier_on(net);
244 		netif_wake_queue(net);
245 		ndev_ctx = netdev_priv(net);
246 		schedule_delayed_work(&ndev_ctx->dwork, 0);
247 		schedule_delayed_work(&ndev_ctx->dwork, msecs_to_jiffies(20));
248 	} else {
249 		netif_carrier_off(net);
250 		netif_tx_disable(net);
251 	}
252 }
253 
254 /*
255  * netvsc_recv_callback -  Callback when we receive a packet from the
256  * "wire" on the specified device.
257  */
258 int netvsc_recv_callback(struct hv_device *device_obj,
259 				struct hv_netvsc_packet *packet)
260 {
261 	struct net_device *net;
262 	struct sk_buff *skb;
263 
264 	net = ((struct netvsc_device *)hv_get_drvdata(device_obj))->ndev;
265 	if (!net) {
266 		netdev_err(net, "got receive callback but net device"
267 			" not initialized yet\n");
268 		return 0;
269 	}
270 
271 	/* Allocate a skb - TODO direct I/O to pages? */
272 	skb = netdev_alloc_skb_ip_align(net, packet->total_data_buflen);
273 	if (unlikely(!skb)) {
274 		++net->stats.rx_dropped;
275 		return 0;
276 	}
277 
278 	/*
279 	 * Copy to skb. This copy is needed here since the memory pointed by
280 	 * hv_netvsc_packet cannot be deallocated
281 	 */
282 	memcpy(skb_put(skb, packet->total_data_buflen), packet->data,
283 		packet->total_data_buflen);
284 
285 	skb->protocol = eth_type_trans(skb, net);
286 	skb->ip_summed = CHECKSUM_NONE;
287 	skb->vlan_tci = packet->vlan_tci;
288 
289 	net->stats.rx_packets++;
290 	net->stats.rx_bytes += packet->total_data_buflen;
291 
292 	/*
293 	 * Pass the skb back up. Network stack will deallocate the skb when it
294 	 * is done.
295 	 * TODO - use NAPI?
296 	 */
297 	netif_rx(skb);
298 
299 	return 0;
300 }
301 
302 static void netvsc_get_drvinfo(struct net_device *net,
303 			       struct ethtool_drvinfo *info)
304 {
305 	strcpy(info->driver, KBUILD_MODNAME);
306 	strcpy(info->version, HV_DRV_VERSION);
307 	strcpy(info->fw_version, "N/A");
308 }
309 
310 static int netvsc_change_mtu(struct net_device *ndev, int mtu)
311 {
312 	struct net_device_context *ndevctx = netdev_priv(ndev);
313 	struct hv_device *hdev =  ndevctx->device_ctx;
314 	struct netvsc_device *nvdev = hv_get_drvdata(hdev);
315 	struct netvsc_device_info device_info;
316 	int limit = ETH_DATA_LEN;
317 
318 	if (nvdev == NULL || nvdev->destroy)
319 		return -ENODEV;
320 
321 	if (nvdev->nvsp_version == NVSP_PROTOCOL_VERSION_2)
322 		limit = NETVSC_MTU;
323 
324 	if (mtu < 68 || mtu > limit)
325 		return -EINVAL;
326 
327 	nvdev->start_remove = true;
328 	cancel_delayed_work_sync(&ndevctx->dwork);
329 	cancel_work_sync(&ndevctx->work);
330 	netif_tx_disable(ndev);
331 	rndis_filter_device_remove(hdev);
332 
333 	ndev->mtu = mtu;
334 
335 	ndevctx->device_ctx = hdev;
336 	hv_set_drvdata(hdev, ndev);
337 	device_info.ring_size = ring_size;
338 	rndis_filter_device_add(hdev, &device_info);
339 	netif_wake_queue(ndev);
340 
341 	return 0;
342 }
343 
344 
345 static int netvsc_set_mac_addr(struct net_device *ndev, void *p)
346 {
347 	struct net_device_context *ndevctx = netdev_priv(ndev);
348 	struct hv_device *hdev =  ndevctx->device_ctx;
349 	struct sockaddr *addr = p;
350 	char save_adr[14];
351 	unsigned char save_aatype;
352 	int err;
353 
354 	memcpy(save_adr, ndev->dev_addr, ETH_ALEN);
355 	save_aatype = ndev->addr_assign_type;
356 
357 	err = eth_mac_addr(ndev, p);
358 	if (err != 0)
359 		return err;
360 
361 	err = rndis_filter_set_device_mac(hdev, addr->sa_data);
362 	if (err != 0) {
363 		/* roll back to saved MAC */
364 		memcpy(ndev->dev_addr, save_adr, ETH_ALEN);
365 		ndev->addr_assign_type = save_aatype;
366 	}
367 
368 	return err;
369 }
370 
371 
372 static const struct ethtool_ops ethtool_ops = {
373 	.get_drvinfo	= netvsc_get_drvinfo,
374 	.get_link	= ethtool_op_get_link,
375 };
376 
377 static const struct net_device_ops device_ops = {
378 	.ndo_open =			netvsc_open,
379 	.ndo_stop =			netvsc_close,
380 	.ndo_start_xmit =		netvsc_start_xmit,
381 	.ndo_set_rx_mode =		netvsc_set_multicast_list,
382 	.ndo_change_mtu =		netvsc_change_mtu,
383 	.ndo_validate_addr =		eth_validate_addr,
384 	.ndo_set_mac_address =		netvsc_set_mac_addr,
385 };
386 
387 /*
388  * Send GARP packet to network peers after migrations.
389  * After Quick Migration, the network is not immediately operational in the
390  * current context when receiving RNDIS_STATUS_MEDIA_CONNECT event. So, add
391  * another netif_notify_peers() into a delayed work, otherwise GARP packet
392  * will not be sent after quick migration, and cause network disconnection.
393  */
394 static void netvsc_send_garp(struct work_struct *w)
395 {
396 	struct net_device_context *ndev_ctx;
397 	struct net_device *net;
398 	struct netvsc_device *net_device;
399 
400 	ndev_ctx = container_of(w, struct net_device_context, dwork.work);
401 	net_device = hv_get_drvdata(ndev_ctx->device_ctx);
402 	net = net_device->ndev;
403 	netif_notify_peers(net);
404 }
405 
406 
407 static int netvsc_probe(struct hv_device *dev,
408 			const struct hv_vmbus_device_id *dev_id)
409 {
410 	struct net_device *net = NULL;
411 	struct net_device_context *net_device_ctx;
412 	struct netvsc_device_info device_info;
413 	int ret;
414 
415 	net = alloc_etherdev(sizeof(struct net_device_context));
416 	if (!net)
417 		return -ENOMEM;
418 
419 	/* Set initial state */
420 	netif_carrier_off(net);
421 
422 	net_device_ctx = netdev_priv(net);
423 	net_device_ctx->device_ctx = dev;
424 	hv_set_drvdata(dev, net);
425 	INIT_DELAYED_WORK(&net_device_ctx->dwork, netvsc_send_garp);
426 	INIT_WORK(&net_device_ctx->work, do_set_multicast);
427 
428 	net->netdev_ops = &device_ops;
429 
430 	/* TODO: Add GSO and Checksum offload */
431 	net->hw_features = NETIF_F_SG;
432 	net->features = NETIF_F_SG | NETIF_F_HW_VLAN_TX;
433 
434 	SET_ETHTOOL_OPS(net, &ethtool_ops);
435 	SET_NETDEV_DEV(net, &dev->device);
436 
437 	ret = register_netdev(net);
438 	if (ret != 0) {
439 		pr_err("Unable to register netdev.\n");
440 		free_netdev(net);
441 		goto out;
442 	}
443 
444 	/* Notify the netvsc driver of the new device */
445 	device_info.ring_size = ring_size;
446 	ret = rndis_filter_device_add(dev, &device_info);
447 	if (ret != 0) {
448 		netdev_err(net, "unable to add netvsc device (ret %d)\n", ret);
449 		unregister_netdev(net);
450 		free_netdev(net);
451 		hv_set_drvdata(dev, NULL);
452 		return ret;
453 	}
454 	memcpy(net->dev_addr, device_info.mac_adr, ETH_ALEN);
455 
456 	netif_carrier_on(net);
457 
458 out:
459 	return ret;
460 }
461 
462 static int netvsc_remove(struct hv_device *dev)
463 {
464 	struct net_device *net;
465 	struct net_device_context *ndev_ctx;
466 	struct netvsc_device *net_device;
467 
468 	net_device = hv_get_drvdata(dev);
469 	net = net_device->ndev;
470 
471 	if (net == NULL) {
472 		dev_err(&dev->device, "No net device to remove\n");
473 		return 0;
474 	}
475 
476 	net_device->start_remove = true;
477 
478 	ndev_ctx = netdev_priv(net);
479 	cancel_delayed_work_sync(&ndev_ctx->dwork);
480 	cancel_work_sync(&ndev_ctx->work);
481 
482 	/* Stop outbound asap */
483 	netif_tx_disable(net);
484 
485 	unregister_netdev(net);
486 
487 	/*
488 	 * Call to the vsc driver to let it know that the device is being
489 	 * removed
490 	 */
491 	rndis_filter_device_remove(dev);
492 
493 	free_netdev(net);
494 	return 0;
495 }
496 
497 static const struct hv_vmbus_device_id id_table[] = {
498 	/* Network guid */
499 	{ VMBUS_DEVICE(0x63, 0x51, 0x61, 0xF8, 0x3E, 0xDF, 0xc5, 0x46,
500 		       0x91, 0x3F, 0xF2, 0xD2, 0xF9, 0x65, 0xED, 0x0E) },
501 	{ },
502 };
503 
504 MODULE_DEVICE_TABLE(vmbus, id_table);
505 
506 /* The one and only one */
507 static struct  hv_driver netvsc_drv = {
508 	.name = KBUILD_MODNAME,
509 	.id_table = id_table,
510 	.probe = netvsc_probe,
511 	.remove = netvsc_remove,
512 };
513 
514 static void __exit netvsc_drv_exit(void)
515 {
516 	vmbus_driver_unregister(&netvsc_drv);
517 }
518 
519 static int __init netvsc_drv_init(void)
520 {
521 	return vmbus_driver_register(&netvsc_drv);
522 }
523 
524 MODULE_LICENSE("GPL");
525 MODULE_VERSION(HV_DRV_VERSION);
526 MODULE_DESCRIPTION("Microsoft Hyper-V network driver");
527 
528 module_init(netvsc_drv_init);
529 module_exit(netvsc_drv_exit);
530