addr.c 14 KB
Newer Older
1
2
3
4
5
6
/*
 * Copyright (c) 2005 Voltaire Inc.  All rights reserved.
 * Copyright (c) 2002-2005, Network Appliance, Inc. All rights reserved.
 * Copyright (c) 1999-2005, Mellanox Technologies, Inc. All rights reserved.
 * Copyright (c) 2005 Intel Corporation.  All rights reserved.
 *
Sean Hefty's avatar
Sean Hefty committed
7
8
9
10
11
 * This software is available to you under a choice of one of two
 * licenses.  You may choose to be licensed under the terms of the GNU
 * General Public License (GPL) Version 2, available from the file
 * COPYING in the main directory of this source tree, or the
 * OpenIB.org BSD license below:
12
 *
Sean Hefty's avatar
Sean Hefty committed
13
14
15
 *     Redistribution and use in source and binary forms, with or
 *     without modification, are permitted provided that the following
 *     conditions are met:
16
 *
Sean Hefty's avatar
Sean Hefty committed
17
18
19
 *      - Redistributions of source code must retain the above
 *        copyright notice, this list of conditions and the following
 *        disclaimer.
20
 *
Sean Hefty's avatar
Sean Hefty committed
21
22
23
24
 *      - Redistributions in binary form must reproduce the above
 *        copyright notice, this list of conditions and the following
 *        disclaimer in the documentation and/or other materials
 *        provided with the distribution.
25
 *
Sean Hefty's avatar
Sean Hefty committed
26
27
28
29
30
31
32
33
 * 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.
34
35
36
37
 */

#include <linux/mutex.h>
#include <linux/inetdevice.h>
38
#include <linux/slab.h>
39
#include <linux/workqueue.h>
40
#include <linux/module.h>
41
42
43
#include <net/arp.h>
#include <net/neighbour.h>
#include <net/route.h>
44
#include <net/netevent.h>
45
46
#include <net/addrconf.h>
#include <net/ip6_route.h>
47
#include <rdma/ib_addr.h>
48
#include <rdma/ib.h>
49
50
51
52
53
54
55

MODULE_AUTHOR("Sean Hefty");
MODULE_DESCRIPTION("IB Address Translation");
MODULE_LICENSE("Dual BSD/GPL");

struct addr_req {
	struct list_head list;
56
57
	struct sockaddr_storage src_addr;
	struct sockaddr_storage dst_addr;
58
	struct rdma_dev_addr *addr;
59
	struct rdma_addr_client *client;
60
61
62
63
64
65
66
	void *context;
	void (*callback)(int status, struct sockaddr *src_addr,
			 struct rdma_dev_addr *addr, void *context);
	unsigned long timeout;
	int status;
};

David Howells's avatar
David Howells committed
67
static void process_req(struct work_struct *work);
68
69
70

static DEFINE_MUTEX(lock);
static LIST_HEAD(req_list);
David Howells's avatar
David Howells committed
71
static DECLARE_DELAYED_WORK(work, process_req);
72
73
static struct workqueue_struct *addr_wq;

74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
int rdma_addr_size(struct sockaddr *addr)
{
	switch (addr->sa_family) {
	case AF_INET:
		return sizeof(struct sockaddr_in);
	case AF_INET6:
		return sizeof(struct sockaddr_in6);
	case AF_IB:
		return sizeof(struct sockaddr_ib);
	default:
		return 0;
	}
}
EXPORT_SYMBOL(rdma_addr_size);

89
90
static struct rdma_addr_client self;

91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
void rdma_addr_register_client(struct rdma_addr_client *client)
{
	atomic_set(&client->refcount, 1);
	init_completion(&client->comp);
}
EXPORT_SYMBOL(rdma_addr_register_client);

static inline void put_client(struct rdma_addr_client *client)
{
	if (atomic_dec_and_test(&client->refcount))
		complete(&client->comp);
}

void rdma_addr_unregister_client(struct rdma_addr_client *client)
{
	put_client(client);
	wait_for_completion(&client->comp);
}
EXPORT_SYMBOL(rdma_addr_unregister_client);

Tom Tucker's avatar
Tom Tucker committed
111
112
int rdma_copy_addr(struct rdma_dev_addr *dev_addr, struct net_device *dev,
		     const unsigned char *dst_dev_addr)
113
{
114
	dev_addr->dev_type = dev->type;
115
116
117
118
	memcpy(dev_addr->src_dev_addr, dev->dev_addr, MAX_ADDR_LEN);
	memcpy(dev_addr->broadcast, dev->broadcast, MAX_ADDR_LEN);
	if (dst_dev_addr)
		memcpy(dev_addr->dst_dev_addr, dst_dev_addr, MAX_ADDR_LEN);
119
	dev_addr->bound_dev_if = dev->ifindex;
120
121
	return 0;
}
Tom Tucker's avatar
Tom Tucker committed
122
EXPORT_SYMBOL(rdma_copy_addr);
123

124
125
int rdma_translate_ip(struct sockaddr *addr, struct rdma_dev_addr *dev_addr,
		      u16 *vlan_id)
126
127
{
	struct net_device *dev;
128
	int ret = -EADDRNOTAVAIL;
129

130
	if (dev_addr->bound_dev_if) {
131
		dev = dev_get_by_index(dev_addr->net, dev_addr->bound_dev_if);
132
133
134
135
136
137
138
		if (!dev)
			return -ENODEV;
		ret = rdma_copy_addr(dev_addr, dev, NULL);
		dev_put(dev);
		return ret;
	}

139
140
	switch (addr->sa_family) {
	case AF_INET:
141
		dev = ip_dev_find(dev_addr->net,
142
143
144
145
			((struct sockaddr_in *) addr)->sin_addr.s_addr);

		if (!dev)
			return ret;
146

147
		ret = rdma_copy_addr(dev_addr, dev, NULL);
148
149
		if (vlan_id)
			*vlan_id = rdma_vlan_dev_vlan_id(dev);
150
151
		dev_put(dev);
		break;
152
#if IS_ENABLED(CONFIG_IPV6)
153
	case AF_INET6:
154
		rcu_read_lock();
155
156
		for_each_netdev_rcu(dev_addr->net, dev) {
			if (ipv6_chk_addr(dev_addr->net,
157
158
159
					  &((struct sockaddr_in6 *) addr)->sin6_addr,
					  dev, 1)) {
				ret = rdma_copy_addr(dev_addr, dev, NULL);
160
161
				if (vlan_id)
					*vlan_id = rdma_vlan_dev_vlan_id(dev);
162
163
164
				break;
			}
		}
165
		rcu_read_unlock();
166
		break;
167
#endif
168
	}
169
170
171
172
173
174
175
176
177
	return ret;
}
EXPORT_SYMBOL(rdma_translate_ip);

static void set_timeout(unsigned long time)
{
	unsigned long delay;

	delay = time - jiffies;
178
179
	if ((long)delay < 0)
		delay = 0;
180

181
	mod_delayed_work(addr_wq, &work, delay);
182
183
184
185
186
187
188
189
}

static void queue_req(struct addr_req *req)
{
	struct addr_req *temp_req;

	mutex_lock(&lock);
	list_for_each_entry_reverse(temp_req, &req_list, list) {
190
		if (time_after_eq(req->timeout, temp_req->timeout))
191
192
193
194
195
196
197
198
199
200
			break;
	}

	list_add(&req->list, &temp_req->list);

	if (req_list.next == &req->list)
		set_timeout(req->timeout);
	mutex_unlock(&lock);
}

201
static int dst_fetch_ha(struct dst_entry *dst, struct rdma_dev_addr *dev_addr, void *daddr)
202
203
204
205
{
	struct neighbour *n;
	int ret;

206
207
	n = dst_neigh_lookup(dst, daddr);

208
209
210
211
212
213
	rcu_read_lock();
	if (!n || !(n->nud_state & NUD_VALID)) {
		if (n)
			neigh_event_send(n, NULL);
		ret = -ENODATA;
	} else {
214
		ret = rdma_copy_addr(dev_addr, dst->dev, n->ha);
215
216
217
	}
	rcu_read_unlock();

218
219
220
	if (n)
		neigh_release(n);

221
222
223
	return ret;
}

224
225
226
static int addr4_resolve(struct sockaddr_in *src_in,
			 struct sockaddr_in *dst_in,
			 struct rdma_dev_addr *addr)
227
{
228
229
	__be32 src_ip = src_in->sin_addr.s_addr;
	__be32 dst_ip = dst_in->sin_addr.s_addr;
230
	struct rtable *rt;
231
	struct flowi4 fl4;
232
233
	int ret;

234
235
236
237
	memset(&fl4, 0, sizeof(fl4));
	fl4.daddr = dst_ip;
	fl4.saddr = src_ip;
	fl4.flowi4_oif = addr->bound_dev_if;
238
	rt = ip_route_output_key(addr->net, &fl4);
239
240
	if (IS_ERR(rt)) {
		ret = PTR_ERR(rt);
241
		goto out;
242
	}
243
	src_in->sin_family = AF_INET;
244
	src_in->sin_addr.s_addr = fl4.saddr;
245

Eric Dumazet's avatar
Eric Dumazet committed
246
	if (rt->dst.dev->flags & IFF_LOOPBACK) {
247
		ret = rdma_translate_ip((struct sockaddr *)dst_in, addr, NULL);
248
249
250
251
252
		if (!ret)
			memcpy(addr->dst_dev_addr, addr->src_dev_addr, MAX_ADDR_LEN);
		goto put;
	}

253
	/* If the device does ARP internally, return 'done' */
Eric Dumazet's avatar
Eric Dumazet committed
254
	if (rt->dst.dev->flags & IFF_NOARP) {
255
		ret = rdma_copy_addr(addr, rt->dst.dev, NULL);
256
257
258
		goto put;
	}

259
260
261
262
263
264
	/* If there's a gateway, we're definitely in RoCE v2 (as RoCE v1 isn't
	 * routable) and we could set the network type accordingly.
	 */
	if (rt->rt_uses_gateway)
		addr->network = RDMA_NETWORK_IPV4;

265
	ret = dst_fetch_ha(&rt->dst, addr, &fl4.daddr);
266
267
268
269
270
271
put:
	ip_rt_put(rt);
out:
	return ret;
}

272
#if IS_ENABLED(CONFIG_IPV6)
Sean Hefty's avatar
Sean Hefty committed
273
274
275
static int addr6_resolve(struct sockaddr_in6 *src_in,
			 struct sockaddr_in6 *dst_in,
			 struct rdma_dev_addr *addr)
276
{
277
	struct flowi6 fl6;
278
	struct dst_entry *dst;
279
	struct rt6_info *rt;
Sean Hefty's avatar
Sean Hefty committed
280
	int ret;
281

282
	memset(&fl6, 0, sizeof fl6);
Alexey Dobriyan's avatar
Alexey Dobriyan committed
283
284
	fl6.daddr = dst_in->sin6_addr;
	fl6.saddr = src_in->sin6_addr;
285
	fl6.flowi6_oif = addr->bound_dev_if;
286

287
	dst = ip6_route_output(addr->net, NULL, &fl6);
Sean Hefty's avatar
Sean Hefty committed
288
289
290
	if ((ret = dst->error))
		goto put;

291
	rt = (struct rt6_info *)dst;
292
	if (ipv6_addr_any(&fl6.saddr)) {
293
		ret = ipv6_dev_get_saddr(addr->net, ip6_dst_idev(dst)->dev,
294
					 &fl6.daddr, 0, &fl6.saddr);
Sean Hefty's avatar
Sean Hefty committed
295
296
		if (ret)
			goto put;
297

Sean Hefty's avatar
Sean Hefty committed
298
		src_in->sin6_family = AF_INET6;
Alexey Dobriyan's avatar
Alexey Dobriyan committed
299
		src_in->sin6_addr = fl6.saddr;
Sean Hefty's avatar
Sean Hefty committed
300
301
302
	}

	if (dst->dev->flags & IFF_LOOPBACK) {
303
		ret = rdma_translate_ip((struct sockaddr *)dst_in, addr, NULL);
Sean Hefty's avatar
Sean Hefty committed
304
305
306
307
308
309
		if (!ret)
			memcpy(addr->dst_dev_addr, addr->src_dev_addr, MAX_ADDR_LEN);
		goto put;
	}

	/* If the device does ARP internally, return 'done' */
310
311
	if (dst->dev->flags & IFF_NOARP) {
		ret = rdma_copy_addr(addr, dst->dev, NULL);
Sean Hefty's avatar
Sean Hefty committed
312
		goto put;
313
314
	}

315
316
317
318
319
320
	/* If there's a gateway, we're definitely in RoCE v2 (as RoCE v1 isn't
	 * routable) and we could set the network type accordingly.
	 */
	if (rt->rt6i_flags & RTF_GATEWAY)
		addr->network = RDMA_NETWORK_IPV6;

321
	ret = dst_fetch_ha(dst, addr, &fl6.daddr);
Sean Hefty's avatar
Sean Hefty committed
322
put:
323
324
325
	dst_release(dst);
	return ret;
}
326
#else
Sean Hefty's avatar
Sean Hefty committed
327
328
329
static int addr6_resolve(struct sockaddr_in6 *src_in,
			 struct sockaddr_in6 *dst_in,
			 struct rdma_dev_addr *addr)
330
331
332
333
{
	return -EADDRNOTAVAIL;
}
#endif
334

335
336
337
static int addr_resolve(struct sockaddr *src_in,
			struct sockaddr *dst_in,
			struct rdma_dev_addr *addr)
338
339
{
	if (src_in->sa_family == AF_INET) {
340
		return addr4_resolve((struct sockaddr_in *) src_in,
341
342
			(struct sockaddr_in *) dst_in, addr);
	} else
Sean Hefty's avatar
Sean Hefty committed
343
		return addr6_resolve((struct sockaddr_in6 *) src_in,
344
345
346
			(struct sockaddr_in6 *) dst_in, addr);
}

David Howells's avatar
David Howells committed
347
static void process_req(struct work_struct *work)
348
349
{
	struct addr_req *req, *temp_req;
350
	struct sockaddr *src_in, *dst_in;
351
352
353
354
355
356
	struct list_head done_list;

	INIT_LIST_HEAD(&done_list);

	mutex_lock(&lock);
	list_for_each_entry_safe(req, temp_req, &req_list, list) {
357
		if (req->status == -ENODATA) {
358
359
			src_in = (struct sockaddr *) &req->src_addr;
			dst_in = (struct sockaddr *) &req->dst_addr;
360
			req->status = addr_resolve(src_in, dst_in, req->addr);
361
362
363
364
			if (req->status && time_after_eq(jiffies, req->timeout))
				req->status = -ETIMEDOUT;
			else if (req->status == -ENODATA)
				continue;
365
		}
366
		list_move_tail(&req->list, &done_list);
367
368
369
370
371
372
373
374
375
376
	}

	if (!list_empty(&req_list)) {
		req = list_entry(req_list.next, struct addr_req, list);
		set_timeout(req->timeout);
	}
	mutex_unlock(&lock);

	list_for_each_entry_safe(req, temp_req, &done_list, list) {
		list_del(&req->list);
377
378
		req->callback(req->status, (struct sockaddr *) &req->src_addr,
			req->addr, req->context);
379
		put_client(req->client);
380
381
382
383
		kfree(req);
	}
}

384
385
int rdma_resolve_ip(struct rdma_addr_client *client,
		    struct sockaddr *src_addr, struct sockaddr *dst_addr,
386
387
388
389
390
		    struct rdma_dev_addr *addr, int timeout_ms,
		    void (*callback)(int status, struct sockaddr *src_addr,
				     struct rdma_dev_addr *addr, void *context),
		    void *context)
{
391
	struct sockaddr *src_in, *dst_in;
392
393
394
	struct addr_req *req;
	int ret = 0;

395
	req = kzalloc(sizeof *req, GFP_KERNEL);
396
397
398
	if (!req)
		return -ENOMEM;

399
400
401
402
403
404
405
406
407
	src_in = (struct sockaddr *) &req->src_addr;
	dst_in = (struct sockaddr *) &req->dst_addr;

	if (src_addr) {
		if (src_addr->sa_family != dst_addr->sa_family) {
			ret = -EINVAL;
			goto err;
		}

408
		memcpy(src_in, src_addr, rdma_addr_size(src_addr));
409
410
411
412
	} else {
		src_in->sa_family = dst_addr->sa_family;
	}

413
	memcpy(dst_in, dst_addr, rdma_addr_size(dst_addr));
414
415
416
	req->addr = addr;
	req->callback = callback;
	req->context = context;
417
418
	req->client = client;
	atomic_inc(&client->refcount);
419

Sean Hefty's avatar
Sean Hefty committed
420
	req->status = addr_resolve(src_in, dst_in, addr);
421
422
423
424
425
426
427
428
429
430
431
	switch (req->status) {
	case 0:
		req->timeout = jiffies;
		queue_req(req);
		break;
	case -ENODATA:
		req->timeout = msecs_to_jiffies(timeout_ms) + jiffies;
		queue_req(req);
		break;
	default:
		ret = req->status;
432
		atomic_dec(&client->refcount);
433
		goto err;
434
435
	}
	return ret;
436
437
438
err:
	kfree(req);
	return ret;
439
440
441
442
443
444
445
446
447
448
449
450
}
EXPORT_SYMBOL(rdma_resolve_ip);

void rdma_addr_cancel(struct rdma_dev_addr *addr)
{
	struct addr_req *req, *temp_req;

	mutex_lock(&lock);
	list_for_each_entry_safe(req, temp_req, &req_list, list) {
		if (req->addr == addr) {
			req->status = -ECANCELED;
			req->timeout = jiffies;
451
			list_move(&req->list, &req_list);
452
453
454
455
456
457
458
459
			set_timeout(req->timeout);
			break;
		}
	}
	mutex_unlock(&lock);
}
EXPORT_SYMBOL(rdma_addr_cancel);

460
461
462
463
464
465
466
467
468
469
470
471
472
struct resolve_cb_context {
	struct rdma_dev_addr *addr;
	struct completion comp;
};

static void resolve_cb(int status, struct sockaddr *src_addr,
	     struct rdma_dev_addr *addr, void *context)
{
	memcpy(((struct resolve_cb_context *)context)->addr, addr, sizeof(struct
				rdma_dev_addr));
	complete(&((struct resolve_cb_context *)context)->comp);
}

473
int rdma_addr_find_dmac_by_grh(const union ib_gid *sgid, const union ib_gid *dgid,
474
			       u8 *dmac, u16 *vlan_id, int if_index)
475
476
477
478
479
480
481
482
483
484
485
486
487
{
	int ret = 0;
	struct rdma_dev_addr dev_addr;
	struct resolve_cb_context ctx;
	struct net_device *dev;

	union {
		struct sockaddr     _sockaddr;
		struct sockaddr_in  _sockaddr_in;
		struct sockaddr_in6 _sockaddr_in6;
	} sgid_addr, dgid_addr;


488
489
	rdma_gid2ip(&sgid_addr._sockaddr, sgid);
	rdma_gid2ip(&dgid_addr._sockaddr, dgid);
490
491

	memset(&dev_addr, 0, sizeof(dev_addr));
492
	dev_addr.bound_dev_if = if_index;
493
	dev_addr.net = &init_net;
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524

	ctx.addr = &dev_addr;
	init_completion(&ctx.comp);
	ret = rdma_resolve_ip(&self, &sgid_addr._sockaddr, &dgid_addr._sockaddr,
			&dev_addr, 1000, resolve_cb, &ctx);
	if (ret)
		return ret;

	wait_for_completion(&ctx.comp);

	memcpy(dmac, dev_addr.dst_dev_addr, ETH_ALEN);
	dev = dev_get_by_index(&init_net, dev_addr.bound_dev_if);
	if (!dev)
		return -ENODEV;
	if (vlan_id)
		*vlan_id = rdma_vlan_dev_vlan_id(dev);
	dev_put(dev);
	return ret;
}
EXPORT_SYMBOL(rdma_addr_find_dmac_by_grh);

int rdma_addr_find_smac_by_sgid(union ib_gid *sgid, u8 *smac, u16 *vlan_id)
{
	int ret = 0;
	struct rdma_dev_addr dev_addr;
	union {
		struct sockaddr     _sockaddr;
		struct sockaddr_in  _sockaddr_in;
		struct sockaddr_in6 _sockaddr_in6;
	} gid_addr;

525
	rdma_gid2ip(&gid_addr._sockaddr, sgid);
526
527

	memset(&dev_addr, 0, sizeof(dev_addr));
528
	dev_addr.net = &init_net;
529
530
531
532
533
534
535
536
537
	ret = rdma_translate_ip(&gid_addr._sockaddr, &dev_addr, vlan_id);
	if (ret)
		return ret;

	memcpy(smac, dev_addr.src_dev_addr, ETH_ALEN);
	return ret;
}
EXPORT_SYMBOL(rdma_addr_find_smac_by_sgid);

Roland Dreier's avatar
Roland Dreier committed
538
static int netevent_callback(struct notifier_block *self, unsigned long event,
539
	void *ctx)
540
{
Roland Dreier's avatar
Roland Dreier committed
541
	if (event == NETEVENT_NEIGH_UPDATE) {
542
		struct neighbour *neigh = ctx;
543

544
		if (neigh->nud_state & NUD_VALID) {
545
546
547
			set_timeout(jiffies);
		}
	}
548
549
550
	return 0;
}

551
552
static struct notifier_block nb = {
	.notifier_call = netevent_callback
553
554
};

555
static int __init addr_init(void)
556
{
557
	addr_wq = create_singlethread_workqueue("ib_addr");
558
559
560
	if (!addr_wq)
		return -ENOMEM;

561
	register_netevent_notifier(&nb);
562
	rdma_addr_register_client(&self);
563
564
565
	return 0;
}

566
static void __exit addr_cleanup(void)
567
{
568
	rdma_addr_unregister_client(&self);
569
	unregister_netevent_notifier(&nb);
570
571
572
573
574
	destroy_workqueue(addr_wq);
}

module_init(addr_init);
module_exit(addr_cleanup);