summaryrefslogtreecommitdiff
path: root/fs/afs/server.c
blob: e900cd74361bc2d9fe8e3d46de096c970ed0864a (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
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
133
134
135
136
137
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
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
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
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
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
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
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
589
590
591
592
593
594
595
596
// SPDX-License-Identifier: GPL-2.0-or-later
/* AFS server record management
 *
 * Copyright (C) 2002, 2007 Red Hat, Inc. All Rights Reserved.
 * Written by David Howells (dhowells@redhat.com)
 */

#include <linux/sched.h>
#include <linux/slab.h>
#include "afs_fs.h"
#include "internal.h"
#include "protocol_yfs.h"

static unsigned afs_server_gc_delay = 10;	/* Server record timeout in seconds */
static unsigned afs_server_update_delay = 30;	/* Time till VLDB recheck in secs */

static void afs_inc_servers_outstanding(struct afs_net *net)
{
	atomic_inc(&net->servers_outstanding);
}

static void afs_dec_servers_outstanding(struct afs_net *net)
{
	if (atomic_dec_and_test(&net->servers_outstanding))
		wake_up_var(&net->servers_outstanding);
}

/*
 * Find a server by one of its addresses.
 */
struct afs_server *afs_find_server(struct afs_net *net,
				   const struct sockaddr_rxrpc *srx)
{
	const struct sockaddr_in6 *a = &srx->transport.sin6, *b;
	const struct afs_addr_list *alist;
	struct afs_server *server = NULL;
	unsigned int i;
	bool ipv6 = true;
	int seq = 0, diff;

	if (srx->transport.sin6.sin6_addr.s6_addr32[0] == 0 ||
	    srx->transport.sin6.sin6_addr.s6_addr32[1] == 0 ||
	    srx->transport.sin6.sin6_addr.s6_addr32[2] == htonl(0xffff))
		ipv6 = false;

	rcu_read_lock();

	do {
		if (server)
			afs_put_server(net, server);
		server = NULL;
		read_seqbegin_or_lock(&net->fs_addr_lock, &seq);

		if (ipv6) {
			hlist_for_each_entry_rcu(server, &net->fs_addresses6, addr6_link) {
				alist = rcu_dereference(server->addresses);
				for (i = alist->nr_ipv4; i < alist->nr_addrs; i++) {
					b = &alist->addrs[i].transport.sin6;
					diff = ((u16 __force)a->sin6_port -
						(u16 __force)b->sin6_port);
					if (diff == 0)
						diff = memcmp(&a->sin6_addr,
							      &b->sin6_addr,
							      sizeof(struct in6_addr));
					if (diff == 0)
						goto found;
				}
			}
		} else {
			hlist_for_each_entry_rcu(server, &net->fs_addresses4, addr4_link) {
				alist = rcu_dereference(server->addresses);
				for (i = 0; i < alist->nr_ipv4; i++) {
					b = &alist->addrs[i].transport.sin6;
					diff = ((u16 __force)a->sin6_port -
						(u16 __force)b->sin6_port);
					if (diff == 0)
						diff = ((u32 __force)a->sin6_addr.s6_addr32[3] -
							(u32 __force)b->sin6_addr.s6_addr32[3]);
					if (diff == 0)
						goto found;
				}
			}
		}

		server = NULL;
	found:
		if (server && !atomic_inc_not_zero(&server->usage))
			server = NULL;

	} while (need_seqretry(&net->fs_addr_lock, seq));

	done_seqretry(&net->fs_addr_lock, seq);

	rcu_read_unlock();
	return server;
}

/*
 * Look up a server by its UUID
 */
struct afs_server *afs_find_server_by_uuid(struct afs_net *net, const uuid_t *uuid)
{
	struct afs_server *server = NULL;
	struct rb_node *p;
	int diff, seq = 0;

	_enter("%pU", uuid);

	do {
		/* Unfortunately, rbtree walking doesn't give reliable results
		 * under just the RCU read lock, so we have to check for
		 * changes.
		 */
		if (server)
			afs_put_server(net, server);
		server = NULL;

		read_seqbegin_or_lock(&net->fs_lock, &seq);

		p = net->fs_servers.rb_node;
		while (p) {
			server = rb_entry(p, struct afs_server, uuid_rb);

			diff = memcmp(uuid, &server->uuid, sizeof(*uuid));
			if (diff < 0) {
				p = p->rb_left;
			} else if (diff > 0) {
				p = p->rb_right;
			} else {
				afs_get_server(server);
				break;
			}

			server = NULL;
		}
	} while (need_seqretry(&net->fs_lock, seq));

	done_seqretry(&net->fs_lock, seq);

	_leave(" = %p", server);
	return server;
}

/*
 * Install a server record in the namespace tree
 */
static struct afs_server *afs_install_server(struct afs_net *net,
					     struct afs_server *candidate)
{
	const struct afs_addr_list *alist;
	struct afs_server *server;
	struct rb_node **pp, *p;
	int ret = -EEXIST, diff;

	_enter("%p", candidate);

	write_seqlock(&net->fs_lock);

	/* Firstly install the server in the UUID lookup tree */
	pp = &net->fs_servers.rb_node;
	p = NULL;
	while (*pp) {
		p = *pp;
		_debug("- consider %p", p);
		server = rb_entry(p, struct afs_server, uuid_rb);
		diff = memcmp(&candidate->uuid, &server->uuid, sizeof(uuid_t));
		if (diff < 0)
			pp = &(*pp)->rb_left;
		else if (diff > 0)
			pp = &(*pp)->rb_right;
		else
			goto exists;
	}

	server = candidate;
	rb_link_node(&server->uuid_rb, p, pp);
	rb_insert_color(&server->uuid_rb, &net->fs_servers);
	hlist_add_head_rcu(&server->proc_link, &net->fs_proc);

	write_seqlock(&net->fs_addr_lock);
	alist = rcu_dereference_protected(server->addresses,
					  lockdep_is_held(&net->fs_addr_lock.lock));

	/* Secondly, if the server has any IPv4 and/or IPv6 addresses, install
	 * it in the IPv4 and/or IPv6 reverse-map lists.
	 *
	 * TODO: For speed we want to use something other than a flat list
	 * here; even sorting the list in terms of lowest address would help a
	 * bit, but anything we might want to do gets messy and memory
	 * intensive.
	 */
	if (alist->nr_ipv4 > 0)
		hlist_add_head_rcu(&server->addr4_link, &net->fs_addresses4);
	if (alist->nr_addrs > alist->nr_ipv4)
		hlist_add_head_rcu(&server->addr6_link, &net->fs_addresses6);

	write_sequnlock(&net->fs_addr_lock);
	ret = 0;

exists:
	afs_get_server(server);
	write_sequnlock(&net->fs_lock);
	return server;
}

/*
 * allocate a new server record
 */
static struct afs_server *afs_alloc_server(struct afs_net *net,
					   const uuid_t *uuid,
					   struct afs_addr_list *alist)
{
	struct afs_server *server;

	_enter("");

	server = kzalloc(sizeof(struct afs_server), GFP_KERNEL);
	if (!server)
		goto enomem;

	atomic_set(&server->usage, 1);
	RCU_INIT_POINTER(server->addresses, alist);
	server->addr_version = alist->version;
	server->uuid = *uuid;
	server->update_at = ktime_get_real_seconds() + afs_server_update_delay;
	rwlock_init(&server->fs_lock);
	INIT_HLIST_HEAD(&server->cb_volumes);
	rwlock_init(&server->cb_break_lock);
	init_waitqueue_head(&server->probe_wq);
	spin_lock_init(&server->probe_lock);

	afs_inc_servers_outstanding(net);
	_leave(" = %p", server);
	return server;

enomem:
	_leave(" = NULL [nomem]");
	return NULL;
}

/*
 * Look up an address record for a server
 */
static struct afs_addr_list *afs_vl_lookup_addrs(struct afs_cell *cell,
						 struct key *key, const uuid_t *uuid)
{
	struct afs_vl_cursor vc;
	struct afs_addr_list *alist = NULL;
	int ret;

	ret = -ERESTARTSYS;
	if (afs_begin_vlserver_operation(&vc, cell, key)) {
		while (afs_select_vlserver(&vc)) {
			if (test_bit(AFS_VLSERVER_FL_IS_YFS, &vc.server->flags))
				alist = afs_yfsvl_get_endpoints(&vc, uuid);
			else
				alist = afs_vl_get_addrs_u(&vc, uuid);
		}

		ret = afs_end_vlserver_operation(&vc);
	}

	return ret < 0 ? ERR_PTR(ret) : alist;
}

/*
 * Get or create a fileserver record.
 */
struct afs_server *afs_lookup_server(struct afs_cell *cell, struct key *key,
				     const uuid_t *uuid)
{
	struct afs_addr_list *alist;
	struct afs_server *server, *candidate;

	_enter("%p,%pU", cell->net, uuid);

	server = afs_find_server_by_uuid(cell->net, uuid);
	if (server)
		return server;

	alist = afs_vl_lookup_addrs(cell, key, uuid);
	if (IS_ERR(alist))
		return ERR_CAST(alist);

	candidate = afs_alloc_server(cell->net, uuid, alist);
	if (!candidate) {
		afs_put_addrlist(alist);
		return ERR_PTR(-ENOMEM);
	}

	server = afs_install_server(cell->net, candidate);
	if (server != candidate) {
		afs_put_addrlist(alist);
		kfree(candidate);
	}

	_leave(" = %p{%d}", server, atomic_read(&server->usage));
	return server;
}

/*
 * Set the server timer to fire after a given delay, assuming it's not already
 * set for an earlier time.
 */
static void afs_set_server_timer(struct afs_net *net, time64_t delay)
{
	if (net->live) {
		afs_inc_servers_outstanding(net);
		if (timer_reduce(&net->fs_timer, jiffies + delay * HZ))
			afs_dec_servers_outstanding(net);
	}
}

/*
 * Server management timer.  We have an increment on fs_outstanding that we
 * need to pass along to the work item.
 */
void afs_servers_timer(struct timer_list *timer)
{
	struct afs_net *net = container_of(timer, struct afs_net, fs_timer);

	_enter("");
	if (!queue_work(afs_wq, &net->fs_manager))
		afs_dec_servers_outstanding(net);
}

/*
 * Release a reference on a server record.
 */
void afs_put_server(struct afs_net *net, struct afs_server *server)
{
	unsigned int usage;

	if (!server)
		return;

	server->put_time = ktime_get_real_seconds();

	usage = atomic_dec_return(&server->usage);

	_enter("{%u}", usage);

	if (likely(usage > 0))
		return;

	afs_set_server_timer(net, afs_server_gc_delay);
}

static void afs_server_rcu(struct rcu_head *rcu)
{
	struct afs_server *server = container_of(rcu, struct afs_server, rcu);

	afs_put_addrlist(rcu_access_pointer(server->addresses));
	kfree(server);
}

/*
 * destroy a dead server
 */
static void afs_destroy_server(struct afs_net *net, struct afs_server *server)
{
	struct afs_addr_list *alist = rcu_access_pointer(server->addresses);
	struct afs_addr_cursor ac = {
		.alist	= alist,
		.index	= alist->preferred,
		.error	= 0,
	};
	_enter("%p", server);

	if (test_bit(AFS_SERVER_FL_MAY_HAVE_CB, &server->flags))
		afs_fs_give_up_all_callbacks(net, server, &ac, NULL);

	wait_var_event(&server->probe_outstanding,
		       atomic_read(&server->probe_outstanding) == 0);

	call_rcu(&server->rcu, afs_server_rcu);
	afs_dec_servers_outstanding(net);
}

/*
 * Garbage collect any expired servers.
 */
static void afs_gc_servers(struct afs_net *net, struct afs_server *gc_list)
{
	struct afs_server *server;
	bool deleted;
	int usage;

	while ((server = gc_list)) {
		gc_list = server->gc_next;

		write_seqlock(&net->fs_lock);
		usage = 1;
		deleted = atomic_try_cmpxchg(&server->usage, &usage, 0);
		if (deleted) {
			rb_erase(&server->uuid_rb, &net->fs_servers);
			hlist_del_rcu(&server->proc_link);
		}
		write_sequnlock(&net->fs_lock);

		if (deleted) {
			write_seqlock(&net->fs_addr_lock);
			if (!hlist_unhashed(&server->addr4_link))
				hlist_del_rcu(&server->addr4_link);
			if (!hlist_unhashed(&server->addr6_link))
				hlist_del_rcu(&server->addr6_link);
			write_sequnlock(&net->fs_addr_lock);
			afs_destroy_server(net, server);
		}
	}
}

/*
 * Manage the records of servers known to be within a network namespace.  This
 * includes garbage collecting unused servers.
 *
 * Note also that we were given an increment on net->servers_outstanding by
 * whoever queued us that we need to deal with before returning.
 */
void afs_manage_servers(struct work_struct *work)
{
	struct afs_net *net = container_of(work, struct afs_net, fs_manager);
	struct afs_server *gc_list = NULL;
	struct rb_node *cursor;
	time64_t now = ktime_get_real_seconds(), next_manage = TIME64_MAX;
	bool purging = !net->live;

	_enter("");

	/* Trawl the server list looking for servers that have expired from
	 * lack of use.
	 */
	read_seqlock_excl(&net->fs_lock);

	for (cursor = rb_first(&net->fs_servers); cursor; cursor = rb_next(cursor)) {
		struct afs_server *server =
			rb_entry(cursor, struct afs_server, uuid_rb);
		int usage = atomic_read(&server->usage);

		_debug("manage %pU %u", &server->uuid, usage);

		ASSERTCMP(usage, >=, 1);
		ASSERTIFCMP(purging, usage, ==, 1);

		if (usage == 1) {
			time64_t expire_at = server->put_time;

			if (!test_bit(AFS_SERVER_FL_VL_FAIL, &server->flags) &&
			    !test_bit(AFS_SERVER_FL_NOT_FOUND, &server->flags))
				expire_at += afs_server_gc_delay;
			if (purging || expire_at <= now) {
				server->gc_next = gc_list;
				gc_list = server;
			} else if (expire_at < next_manage) {
				next_manage = expire_at;
			}
		}
	}

	read_sequnlock_excl(&net->fs_lock);

	/* Update the timer on the way out.  We have to pass an increment on
	 * servers_outstanding in the namespace that we are in to the timer or
	 * the work scheduler.
	 */
	if (!purging && next_manage < TIME64_MAX) {
		now = ktime_get_real_seconds();

		if (next_manage - now <= 0) {
			if (queue_work(afs_wq, &net->fs_manager))
				afs_inc_servers_outstanding(net);
		} else {
			afs_set_server_timer(net, next_manage - now);
		}
	}

	afs_gc_servers(net, gc_list);

	afs_dec_servers_outstanding(net);
	_leave(" [%d]", atomic_read(&net->servers_outstanding));
}

static void afs_queue_server_manager(struct afs_net *net)
{
	afs_inc_servers_outstanding(net);
	if (!queue_work(afs_wq, &net->fs_manager))
		afs_dec_servers_outstanding(net);
}

/*
 * Purge list of servers.
 */
void afs_purge_servers(struct afs_net *net)
{
	_enter("");

	if (del_timer_sync(&net->fs_timer))
		atomic_dec(&net->servers_outstanding);

	afs_queue_server_manager(net);

	_debug("wait");
	wait_var_event(&net->servers_outstanding,
		       !atomic_read(&net->servers_outstanding));
	_leave("");
}

/*
 * Get an update for a server's address list.
 */
static noinline bool afs_update_server_record(struct afs_fs_cursor *fc, struct afs_server *server)
{
	struct afs_addr_list *alist, *discard;

	_enter("");

	alist = afs_vl_lookup_addrs(fc->vnode->volume->cell, fc->key,
				    &server->uuid);
	if (IS_ERR(alist)) {
		if ((PTR_ERR(alist) == -ERESTARTSYS ||
		     PTR_ERR(alist) == -EINTR) &&
		    !(fc->flags & AFS_FS_CURSOR_INTR) &&
		    server->addresses) {
			_leave(" = t [intr]");
			return true;
		}
		fc->error = PTR_ERR(alist);
		_leave(" = f [%d]", fc->error);
		return false;
	}

	discard = alist;
	if (server->addr_version != alist->version) {
		write_lock(&server->fs_lock);
		discard = rcu_dereference_protected(server->addresses,
						    lockdep_is_held(&server->fs_lock));
		rcu_assign_pointer(server->addresses, alist);
		server->addr_version = alist->version;
		write_unlock(&server->fs_lock);
	}

	server->update_at = ktime_get_real_seconds() + afs_server_update_delay;
	afs_put_addrlist(discard);
	_leave(" = t");
	return true;
}

/*
 * See if a server's address list needs updating.
 */
bool afs_check_server_record(struct afs_fs_cursor *fc, struct afs_server *server)
{
	time64_t now = ktime_get_real_seconds();
	long diff;
	bool success;
	int ret, retries = 0;

	_enter("");

	ASSERT(server);

retry:
	diff = READ_ONCE(server->update_at) - now;
	if (diff > 0) {
		_leave(" = t [not now %ld]", diff);
		return true;
	}

	if (!test_and_set_bit_lock(AFS_SERVER_FL_UPDATING, &server->flags)) {
		success = afs_update_server_record(fc, server);
		clear_bit_unlock(AFS_SERVER_FL_UPDATING, &server->flags);
		wake_up_bit(&server->flags, AFS_SERVER_FL_UPDATING);
		_leave(" = %d", success);
		return success;
	}

	ret = wait_on_bit(&server->flags, AFS_SERVER_FL_UPDATING,
			  TASK_INTERRUPTIBLE);
	if (ret == -ERESTARTSYS) {
		if (!(fc->flags & AFS_FS_CURSOR_INTR) && server->addresses) {
			_leave(" = t [intr]");
			return true;
		}
		fc->error = ret;
		_leave(" = f [intr]");
		return false;
	}

	retries++;
	if (retries == 4) {
		_leave(" = f [stale]");
		ret = -ESTALE;
		return false;
	}
	goto retry;
}