addr.c 20.1 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
#include <rdma/rdma_netlink.h>
#include <net/netlink.h>

#include "core_priv.h"
53
54
55

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
	void *context;
	void (*callback)(int status, struct sockaddr *src_addr,
			 struct rdma_dev_addr *addr, void *context);
	unsigned long timeout;
64
	struct delayed_work work;
65
	int status;
66
	u32 seq;
67
68
};

69
70
static atomic_t ib_nl_addr_request_seq = ATOMIC_INIT(0);

David Howells's avatar
David Howells committed
71
static void process_req(struct work_struct *work);
72
73
74

static DEFINE_MUTEX(lock);
static LIST_HEAD(req_list);
David Howells's avatar
David Howells committed
75
static DECLARE_DELAYED_WORK(work, process_req);
76
77
static struct workqueue_struct *addr_wq;

78
79
80
81
82
83
84
85
86
87
88
89
90
91
static const struct nla_policy ib_nl_addr_policy[LS_NLA_TYPE_MAX] = {
	[LS_NLA_TYPE_DGID] = {.type = NLA_BINARY,
		.len = sizeof(struct rdma_nla_ls_gid)},
};

static inline bool ib_nl_is_good_ip_resp(const struct nlmsghdr *nlh)
{
	struct nlattr *tb[LS_NLA_TYPE_MAX] = {};
	int ret;

	if (nlh->nlmsg_flags & RDMA_NL_LS_F_ERR)
		return false;

	ret = nla_parse(tb, LS_NLA_TYPE_MAX - 1, nlmsg_data(nlh),
92
			nlmsg_len(nlh), ib_nl_addr_policy, NULL);
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
	if (ret)
		return false;

	return true;
}

static void ib_nl_process_good_ip_rsep(const struct nlmsghdr *nlh)
{
	const struct nlattr *head, *curr;
	union ib_gid gid;
	struct addr_req *req;
	int len, rem;
	int found = 0;

	head = (const struct nlattr *)nlmsg_data(nlh);
	len = nlmsg_len(nlh);

	nla_for_each_attr(curr, head, len, rem) {
		if (curr->nla_type == LS_NLA_TYPE_DGID)
			memcpy(&gid, nla_data(curr), nla_len(curr));
	}

	mutex_lock(&lock);
	list_for_each_entry(req, &req_list, list) {
		if (nlh->nlmsg_seq != req->seq)
			continue;
		/* We set the DGID part, the rest was set earlier */
		rdma_addr_set_dgid(req->addr, &gid);
		req->status = 0;
		found = 1;
		break;
	}
	mutex_unlock(&lock);

	if (!found)
		pr_info("Couldn't find request waiting for DGID: %pI6\n",
			&gid);
}

int ib_nl_handle_ip_res_resp(struct sk_buff *skb,
133
134
			     struct nlmsghdr *nlh,
			     struct netlink_ext_ack *extack)
135
136
{
	if ((nlh->nlmsg_flags & NLM_F_REQUEST) ||
137
	    !(NETLINK_CB(skb).sk))
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
		return -EPERM;

	if (ib_nl_is_good_ip_resp(nlh))
		ib_nl_process_good_ip_rsep(nlh);

	return skb->len;
}

static int ib_nl_ip_send_msg(struct rdma_dev_addr *dev_addr,
			     const void *daddr,
			     u32 seq, u16 family)
{
	struct sk_buff *skb = NULL;
	struct nlmsghdr *nlh;
	struct rdma_ls_ip_resolve_header *header;
	void *data;
	size_t size;
	int attrtype;
	int len;

	if (family == AF_INET) {
		size = sizeof(struct in_addr);
		attrtype = RDMA_NLA_F_MANDATORY | LS_NLA_TYPE_IPV4;
	} else {
		size = sizeof(struct in6_addr);
		attrtype = RDMA_NLA_F_MANDATORY | LS_NLA_TYPE_IPV6;
	}

	len = nla_total_size(sizeof(size));
	len += NLMSG_ALIGN(sizeof(*header));

	skb = nlmsg_new(len, GFP_KERNEL);
	if (!skb)
		return -ENOMEM;

	data = ibnl_put_msg(skb, &nlh, seq, 0, RDMA_NL_LS,
			    RDMA_NL_LS_OP_IP_RESOLVE, NLM_F_REQUEST);
	if (!data) {
		nlmsg_free(skb);
		return -ENODATA;
	}

	/* Construct the family header first */
181
	header = skb_put(skb, NLMSG_ALIGN(sizeof(*header)));
182
183
184
185
186
	header->ifindex = dev_addr->bound_dev_if;
	nla_put(skb, attrtype, size, daddr);

	/* Repair the nlmsg header length */
	nlmsg_end(skb, nlh);
187
	rdma_nl_multicast(skb, RDMA_NL_GROUP_LS, GFP_KERNEL);
188
189
190
191
192
193
194

	/* Make the request retry, so when we get the response from userspace
	 * we will have something.
	 */
	return -ENODATA;
}

195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
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);

210
211
static struct rdma_addr_client self;

212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
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);

232
233
234
void rdma_copy_addr(struct rdma_dev_addr *dev_addr,
		    const struct net_device *dev,
		    const unsigned char *dst_dev_addr)
235
{
236
	dev_addr->dev_type = dev->type;
237
238
239
240
	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);
241
	dev_addr->bound_dev_if = dev->ifindex;
242
}
Tom Tucker's avatar
Tom Tucker committed
243
EXPORT_SYMBOL(rdma_copy_addr);
244

245
int rdma_translate_ip(const struct sockaddr *addr,
246
		      struct rdma_dev_addr *dev_addr)
247
248
249
{
	struct net_device *dev;

250
	if (dev_addr->bound_dev_if) {
251
		dev = dev_get_by_index(dev_addr->net, dev_addr->bound_dev_if);
252
253
		if (!dev)
			return -ENODEV;
254
		rdma_copy_addr(dev_addr, dev, NULL);
255
		dev_put(dev);
256
		return 0;
257
258
	}

259
260
	switch (addr->sa_family) {
	case AF_INET:
261
		dev = ip_dev_find(dev_addr->net,
262
			((const struct sockaddr_in *)addr)->sin_addr.s_addr);
263
264

		if (!dev)
265
			return -EADDRNOTAVAIL;
266

267
		rdma_copy_addr(dev_addr, dev, NULL);
268
269
		dev_put(dev);
		break;
270
#if IS_ENABLED(CONFIG_IPV6)
271
	case AF_INET6:
272
		rcu_read_lock();
273
274
		for_each_netdev_rcu(dev_addr->net, dev) {
			if (ipv6_chk_addr(dev_addr->net,
275
					  &((const struct sockaddr_in6 *)addr)->sin6_addr,
276
					  dev, 1)) {
277
				rdma_copy_addr(dev_addr, dev, NULL);
278
279
280
				break;
			}
		}
281
		rcu_read_unlock();
282
		break;
283
#endif
284
	}
285
	return 0;
286
287
288
}
EXPORT_SYMBOL(rdma_translate_ip);

289
static void set_timeout(struct delayed_work *delayed_work, unsigned long time)
290
291
292
293
{
	unsigned long delay;

	delay = time - jiffies;
294
295
	if ((long)delay < 0)
		delay = 0;
296

297
	mod_delayed_work(addr_wq, delayed_work, delay);
298
299
300
301
302
303
304
305
}

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) {
306
		if (time_after_eq(req->timeout, temp_req->timeout))
307
308
309
310
311
			break;
	}

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

312
	set_timeout(&req->work, req->timeout);
313
314
315
	mutex_unlock(&lock);
}

316
317
318
static int ib_nl_fetch_ha(struct dst_entry *dst, struct rdma_dev_addr *dev_addr,
			  const void *daddr, u32 seq, u16 family)
{
319
	if (rdma_nl_chk_listeners(RDMA_NL_GROUP_LS))
320
321
322
323
324
325
326
		return -EADDRNOTAVAIL;

	/* We fill in what we can, the response will fill the rest */
	rdma_copy_addr(dev_addr, dst->dev, NULL);
	return ib_nl_ip_send_msg(dev_addr, daddr, seq, family);
}

327
328
static int dst_fetch_ha(struct dst_entry *dst, struct rdma_dev_addr *dev_addr,
			const void *daddr)
329
330
{
	struct neighbour *n;
331
	int ret = 0;
332

333
334
	n = dst_neigh_lookup(dst, daddr);

335
336
337
338
339
340
	rcu_read_lock();
	if (!n || !(n->nud_state & NUD_VALID)) {
		if (n)
			neigh_event_send(n, NULL);
		ret = -ENODATA;
	} else {
341
		rdma_copy_addr(dev_addr, dst->dev, n->ha);
342
343
344
	}
	rcu_read_unlock();

345
346
347
	if (n)
		neigh_release(n);

348
349
350
	return ret;
}

351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
static bool has_gateway(struct dst_entry *dst, sa_family_t family)
{
	struct rtable *rt;
	struct rt6_info *rt6;

	if (family == AF_INET) {
		rt = container_of(dst, struct rtable, dst);
		return rt->rt_uses_gateway;
	}

	rt6 = container_of(dst, struct rt6_info, dst);
	return rt6->rt6i_flags & RTF_GATEWAY;
}

static int fetch_ha(struct dst_entry *dst, struct rdma_dev_addr *dev_addr,
		    const struct sockaddr *dst_in, u32 seq)
{
	const struct sockaddr_in *dst_in4 =
		(const struct sockaddr_in *)dst_in;
	const struct sockaddr_in6 *dst_in6 =
		(const struct sockaddr_in6 *)dst_in;
	const void *daddr = (dst_in->sa_family == AF_INET) ?
		(const void *)&dst_in4->sin_addr.s_addr :
		(const void *)&dst_in6->sin6_addr;
	sa_family_t family = dst_in->sa_family;

	/* Gateway + ARPHRD_INFINIBAND -> IB router */
	if (has_gateway(dst, family) && dst->dev->type == ARPHRD_INFINIBAND)
		return ib_nl_fetch_ha(dst, dev_addr, daddr, seq, family);
	else
		return dst_fetch_ha(dst, dev_addr, daddr);
}

384
static int addr4_resolve(struct sockaddr_in *src_in,
385
386
387
			 const struct sockaddr_in *dst_in,
			 struct rdma_dev_addr *addr,
			 struct rtable **prt)
388
{
389
390
	__be32 src_ip = src_in->sin_addr.s_addr;
	__be32 dst_ip = dst_in->sin_addr.s_addr;
391
	struct rtable *rt;
392
	struct flowi4 fl4;
393
394
	int ret;

395
396
397
398
	memset(&fl4, 0, sizeof(fl4));
	fl4.daddr = dst_ip;
	fl4.saddr = src_ip;
	fl4.flowi4_oif = addr->bound_dev_if;
399
	rt = ip_route_output_key(addr->net, &fl4);
400
401
402
403
	ret = PTR_ERR_OR_ZERO(rt);
	if (ret)
		return ret;

404
	src_in->sin_family = AF_INET;
405
	src_in->sin_addr.s_addr = fl4.saddr;
406

407
408
409
	/* If there's a gateway and type of device not ARPHRD_INFINIBAND, we're
	 * definitely in RoCE v2 (as RoCE v1 isn't routable) set the network
	 * type accordingly.
410
	 */
411
	if (rt->rt_uses_gateway && rt->dst.dev->type != ARPHRD_INFINIBAND)
412
413
		addr->network = RDMA_NETWORK_IPV4;

414
415
	addr->hoplimit = ip4_dst_hoplimit(&rt->dst);

416
417
	*prt = rt;
	return 0;
418
419
}

420
#if IS_ENABLED(CONFIG_IPV6)
Sean Hefty's avatar
Sean Hefty committed
421
static int addr6_resolve(struct sockaddr_in6 *src_in,
422
423
424
			 const struct sockaddr_in6 *dst_in,
			 struct rdma_dev_addr *addr,
			 struct dst_entry **pdst)
425
{
426
	struct flowi6 fl6;
427
	struct dst_entry *dst;
428
	struct rt6_info *rt;
Sean Hefty's avatar
Sean Hefty committed
429
	int ret;
430

431
	memset(&fl6, 0, sizeof fl6);
Alexey Dobriyan's avatar
Alexey Dobriyan committed
432
433
	fl6.daddr = dst_in->sin6_addr;
	fl6.saddr = src_in->sin6_addr;
434
	fl6.flowi6_oif = addr->bound_dev_if;
435

436
437
	ret = ipv6_stub->ipv6_dst_lookup(addr->net, NULL, &dst, &fl6);
	if (ret < 0)
438
		return ret;
Sean Hefty's avatar
Sean Hefty committed
439

440
	rt = (struct rt6_info *)dst;
441
	if (ipv6_addr_any(&src_in->sin6_addr)) {
Sean Hefty's avatar
Sean Hefty committed
442
		src_in->sin6_family = AF_INET6;
Alexey Dobriyan's avatar
Alexey Dobriyan committed
443
		src_in->sin6_addr = fl6.saddr;
Sean Hefty's avatar
Sean Hefty committed
444
445
	}

446
447
448
	/* If there's a gateway and type of device not ARPHRD_INFINIBAND, we're
	 * definitely in RoCE v2 (as RoCE v1 isn't routable) set the network
	 * type accordingly.
449
	 */
450
451
	if (rt->rt6i_flags & RTF_GATEWAY &&
	    ip6_dst_idev(dst)->dev->type != ARPHRD_INFINIBAND)
452
453
		addr->network = RDMA_NETWORK_IPV6;

454
455
	addr->hoplimit = ip6_dst_hoplimit(dst);

456
457
	*pdst = dst;
	return 0;
458
}
459
#else
Sean Hefty's avatar
Sean Hefty committed
460
static int addr6_resolve(struct sockaddr_in6 *src_in,
461
462
463
			 const struct sockaddr_in6 *dst_in,
			 struct rdma_dev_addr *addr,
			 struct dst_entry **pdst)
464
465
466
467
{
	return -EADDRNOTAVAIL;
}
#endif
468

469
470
static int addr_resolve_neigh(struct dst_entry *dst,
			      const struct sockaddr *dst_in,
471
472
			      struct rdma_dev_addr *addr,
			      u32 seq)
473
474
475
476
{
	if (dst->dev->flags & IFF_LOOPBACK) {
		int ret;

477
		ret = rdma_translate_ip(dst_in, addr);
478
479
480
481
482
483
484
485
		if (!ret)
			memcpy(addr->dst_dev_addr, addr->src_dev_addr,
			       MAX_ADDR_LEN);

		return ret;
	}

	/* If the device doesn't do ARP internally */
486
487
	if (!(dst->dev->flags & IFF_NOARP))
		return fetch_ha(dst, addr, dst_in, seq);
488

489
490
491
	rdma_copy_addr(addr, dst->dev, NULL);

	return 0;
492
493
}

494
static int addr_resolve(struct sockaddr *src_in,
495
496
			const struct sockaddr *dst_in,
			struct rdma_dev_addr *addr,
497
498
			bool resolve_neigh,
			u32 seq)
499
{
500
501
502
503
	struct net_device *ndev;
	struct dst_entry *dst;
	int ret;

504
505
506
507
508
	if (!addr->net) {
		pr_warn_ratelimited("%s: missing namespace\n", __func__);
		return -EINVAL;
	}

509
	if (src_in->sa_family == AF_INET) {
510
511
512
513
514
515
516
517
518
519
		struct rtable *rt = NULL;
		const struct sockaddr_in *dst_in4 =
			(const struct sockaddr_in *)dst_in;

		ret = addr4_resolve((struct sockaddr_in *)src_in,
				    dst_in4, addr, &rt);
		if (ret)
			return ret;

		if (resolve_neigh)
520
			ret = addr_resolve_neigh(&rt->dst, dst_in, addr, seq);
521

522
523
524
525
526
527
		if (addr->bound_dev_if) {
			ndev = dev_get_by_index(addr->net, addr->bound_dev_if);
		} else {
			ndev = rt->dst.dev;
			dev_hold(ndev);
		}
528
529
530
531
532
533
534
535
536
537
538
539
540

		ip_rt_put(rt);
	} else {
		const struct sockaddr_in6 *dst_in6 =
			(const struct sockaddr_in6 *)dst_in;

		ret = addr6_resolve((struct sockaddr_in6 *)src_in,
				    dst_in6, addr,
				    &dst);
		if (ret)
			return ret;

		if (resolve_neigh)
541
			ret = addr_resolve_neigh(dst, dst_in, addr, seq);
542

543
544
545
546
547
548
		if (addr->bound_dev_if) {
			ndev = dev_get_by_index(addr->net, addr->bound_dev_if);
		} else {
			ndev = dst->dev;
			dev_hold(ndev);
		}
549
550
551
552

		dst_release(dst);
	}

553
554
555
556
557
	if (ndev) {
		if (ndev->flags & IFF_LOOPBACK)
			ret = rdma_translate_ip(dst_in, addr);
		else
			addr->bound_dev_if = ndev->ifindex;
558
559
		dev_put(ndev);
	}
560
561

	return ret;
562
563
}

564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
static void process_one_req(struct work_struct *_work)
{
	struct addr_req *req;
	struct sockaddr *src_in, *dst_in;

	mutex_lock(&lock);
	req = container_of(_work, struct addr_req, work.work);

	if (req->status == -ENODATA) {
		src_in = (struct sockaddr *)&req->src_addr;
		dst_in = (struct sockaddr *)&req->dst_addr;
		req->status = addr_resolve(src_in, dst_in, req->addr,
					   true, req->seq);
		if (req->status && time_after_eq(jiffies, req->timeout)) {
			req->status = -ETIMEDOUT;
		} else if (req->status == -ENODATA) {
			/* requeue the work for retrying again */
			set_timeout(&req->work, req->timeout);
			mutex_unlock(&lock);
			return;
		}
	}
	list_del(&req->list);
	mutex_unlock(&lock);

589
590
591
592
593
594
595
596
597
	/*
	 * Although the work will normally have been canceled by the
	 * workqueue, it can still be requeued as long as it is on the
	 * req_list, so it could have been requeued before we grabbed &lock.
	 * We need to cancel it after it is removed from req_list to really be
	 * sure it is safe to free.
	 */
	cancel_delayed_work(&req->work);

598
599
600
601
602
603
	req->callback(req->status, (struct sockaddr *)&req->src_addr,
		req->addr, req->context);
	put_client(req->client);
	kfree(req);
}

David Howells's avatar
David Howells committed
604
static void process_req(struct work_struct *work)
605
606
{
	struct addr_req *req, *temp_req;
607
	struct sockaddr *src_in, *dst_in;
608
609
610
611
612
613
	struct list_head done_list;

	INIT_LIST_HEAD(&done_list);

	mutex_lock(&lock);
	list_for_each_entry_safe(req, temp_req, &req_list, list) {
614
		if (req->status == -ENODATA) {
615
616
			src_in = (struct sockaddr *) &req->src_addr;
			dst_in = (struct sockaddr *) &req->dst_addr;
617
			req->status = addr_resolve(src_in, dst_in, req->addr,
618
						   true, req->seq);
619
620
			if (req->status && time_after_eq(jiffies, req->timeout))
				req->status = -ETIMEDOUT;
621
622
			else if (req->status == -ENODATA) {
				set_timeout(&req->work, req->timeout);
623
				continue;
624
			}
625
		}
626
		list_move_tail(&req->list, &done_list);
627
628
629
630
631
632
	}

	mutex_unlock(&lock);

	list_for_each_entry_safe(req, temp_req, &done_list, list) {
		list_del(&req->list);
633
634
635
636
637
		/* It is safe to cancel other work items from this work item
		 * because at a time there can be only one work item running
		 * with this single threaded work queue.
		 */
		cancel_delayed_work(&req->work);
638
639
		req->callback(req->status, (struct sockaddr *) &req->src_addr,
			req->addr, req->context);
640
		put_client(req->client);
641
642
643
644
		kfree(req);
	}
}

645
646
int rdma_resolve_ip(struct rdma_addr_client *client,
		    struct sockaddr *src_addr, struct sockaddr *dst_addr,
647
648
649
650
651
		    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)
{
652
	struct sockaddr *src_in, *dst_in;
653
654
655
	struct addr_req *req;
	int ret = 0;

656
	req = kzalloc(sizeof *req, GFP_KERNEL);
657
658
659
	if (!req)
		return -ENOMEM;

660
661
662
663
664
665
666
667
668
	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;
		}

669
		memcpy(src_in, src_addr, rdma_addr_size(src_addr));
670
671
672
673
	} else {
		src_in->sa_family = dst_addr->sa_family;
	}

674
	memcpy(dst_in, dst_addr, rdma_addr_size(dst_addr));
675
676
677
	req->addr = addr;
	req->callback = callback;
	req->context = context;
678
679
	req->client = client;
	atomic_inc(&client->refcount);
680
	INIT_DELAYED_WORK(&req->work, process_one_req);
681
	req->seq = (u32)atomic_inc_return(&ib_nl_addr_request_seq);
682

683
	req->status = addr_resolve(src_in, dst_in, addr, true, req->seq);
684
685
686
687
688
689
690
691
692
693
694
	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;
695
		atomic_dec(&client->refcount);
696
		goto err;
697
698
	}
	return ret;
699
700
701
err:
	kfree(req);
	return ret;
702
703
704
}
EXPORT_SYMBOL(rdma_resolve_ip);

705
706
707
708
709
710
711
int rdma_resolve_ip_route(struct sockaddr *src_addr,
			  const struct sockaddr *dst_addr,
			  struct rdma_dev_addr *addr)
{
	struct sockaddr_storage ssrc_addr = {};
	struct sockaddr *src_in = (struct sockaddr *)&ssrc_addr;

712
713
714
	if (src_addr) {
		if (src_addr->sa_family != dst_addr->sa_family)
			return -EINVAL;
715
716

		memcpy(src_in, src_addr, rdma_addr_size(src_addr));
717
	} else {
718
		src_in->sa_family = dst_addr->sa_family;
719
	}
720

721
	return addr_resolve(src_in, dst_addr, addr, false, 0);
722
723
724
}
EXPORT_SYMBOL(rdma_resolve_ip_route);

725
726
727
728
729
730
731
732
733
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;
734
			list_move(&req->list, &req_list);
735
			set_timeout(&req->work, req->timeout);
736
737
738
739
740
741
742
			break;
		}
	}
	mutex_unlock(&lock);
}
EXPORT_SYMBOL(rdma_addr_cancel);

743
744
struct resolve_cb_context {
	struct completion comp;
745
	int status;
746
747
748
749
750
};

static void resolve_cb(int status, struct sockaddr *src_addr,
	     struct rdma_dev_addr *addr, void *context)
{
751
	((struct resolve_cb_context *)context)->status = status;
752
753
754
	complete(&((struct resolve_cb_context *)context)->comp);
}

755
756
int rdma_addr_find_l2_eth_by_grh(const union ib_gid *sgid,
				 const union ib_gid *dgid,
757
				 u8 *dmac, const struct net_device *ndev,
758
				 int *hoplimit)
759
760
761
762
763
764
765
766
{
	struct rdma_dev_addr dev_addr;
	struct resolve_cb_context ctx;
	union {
		struct sockaddr     _sockaddr;
		struct sockaddr_in  _sockaddr_in;
		struct sockaddr_in6 _sockaddr_in6;
	} sgid_addr, dgid_addr;
767
	int ret;
768

769
770
	rdma_gid2ip(&sgid_addr._sockaddr, sgid);
	rdma_gid2ip(&dgid_addr._sockaddr, dgid);
771
772

	memset(&dev_addr, 0, sizeof(dev_addr));
773
	dev_addr.bound_dev_if = ndev->ifindex;
774
	dev_addr.net = &init_net;
775
776
777
778
779
780
781
782
783

	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);

784
785
786
787
	ret = ctx.status;
	if (ret)
		return ret;

788
	memcpy(dmac, dev_addr.dst_dev_addr, ETH_ALEN);
789
790
	*hoplimit = dev_addr.hoplimit;
	return 0;
791
792
}

Roland Dreier's avatar
Roland Dreier committed
793
static int netevent_callback(struct notifier_block *self, unsigned long event,
794
	void *ctx)
795
{
Roland Dreier's avatar
Roland Dreier committed
796
	if (event == NETEVENT_NEIGH_UPDATE) {
797
		struct neighbour *neigh = ctx;
798

799
800
		if (neigh->nud_state & NUD_VALID)
			set_timeout(&work, jiffies);
801
	}
802
803
804
	return 0;
}

805
806
static struct notifier_block nb = {
	.notifier_call = netevent_callback
807
808
};

809
int addr_init(void)
810
{
811
	addr_wq = alloc_ordered_workqueue("ib_addr", 0);
812
813
814
	if (!addr_wq)
		return -ENOMEM;

815
	register_netevent_notifier(&nb);
816
	rdma_addr_register_client(&self);
817

818
819
820
	return 0;
}

821
void addr_cleanup(void)
822
{
823
	rdma_addr_unregister_client(&self);
824
	unregister_netevent_notifier(&nb);
825
826
	destroy_workqueue(addr_wq);
}