summaryrefslogtreecommitdiffstats
path: root/drivers/md/persistent-data/dm-block-manager.c
blob: a7e8bf2963886dfa349a03644cf33ff1ded748c5 (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
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
/*
 * Copyright (C) 2011 Red Hat, Inc.
 *
 * This file is released under the GPL.
 */
#include "dm-block-manager.h"
#include "dm-persistent-data-internal.h"
#include "../dm-bufio.h"

#include <linux/crc32c.h>
#include <linux/module.h>
#include <linux/slab.h>
#include <linux/rwsem.h>
#include <linux/device-mapper.h>
#include <linux/stacktrace.h>

#define DM_MSG_PREFIX "block manager"

/*----------------------------------------------------------------*/

/*
 * This is a read/write semaphore with a couple of differences.
 *
 * i) There is a restriction on the number of concurrent read locks that
 * may be held at once.  This is just an implementation detail.
 *
 * ii) Recursive locking attempts are detected and return EINVAL.  A stack
 * trace is also emitted for the previous lock acquisition.
 *
 * iii) Priority is given to write locks.
 */
#define MAX_HOLDERS 4
#define MAX_STACK 10

typedef unsigned long stack_entries[MAX_STACK];

struct block_lock {
	spinlock_t lock;
	__s32 count;
	struct list_head waiters;
	struct task_struct *holders[MAX_HOLDERS];

#ifdef CONFIG_DM_DEBUG_BLOCK_STACK_TRACING
	struct stack_trace traces[MAX_HOLDERS];
	stack_entries entries[MAX_HOLDERS];
#endif
};

struct waiter {
	struct list_head list;
	struct task_struct *task;
	int wants_write;
};

static unsigned __find_holder(struct block_lock *lock,
			      struct task_struct *task)
{
	unsigned i;

	for (i = 0; i < MAX_HOLDERS; i++)
		if (lock->holders[i] == task)
			break;

	BUG_ON(i == MAX_HOLDERS);
	return i;
}

/* call this *after* you increment lock->count */
static void __add_holder(struct block_lock *lock, struct task_struct *task)
{
	unsigned h = __find_holder(lock, NULL);
#ifdef CONFIG_DM_DEBUG_BLOCK_STACK_TRACING
	struct stack_trace *t;
#endif

	get_task_struct(task);
	lock->holders[h] = task;

#ifdef CONFIG_DM_DEBUG_BLOCK_STACK_TRACING
	t = lock->traces + h;
	t->nr_entries = 0;
	t->max_entries = MAX_STACK;
	t->entries = lock->entries[h];
	t->skip = 2;
	save_stack_trace(t);
#endif
}

/* call this *before* you decrement lock->count */
static void __del_holder(struct block_lock *lock, struct task_struct *task)
{
	unsigned h = __find_holder(lock, task);
	lock->holders[h] = NULL;
	put_task_struct(task);
}

static int __check_holder(struct block_lock *lock)
{
	unsigned i;
#ifdef CONFIG_DM_DEBUG_BLOCK_STACK_TRACING
	static struct stack_trace t;
	static stack_entries entries;
#endif

	for (i = 0; i < MAX_HOLDERS; i++) {
		if (lock->holders[i] == current) {
			DMERR("recursive lock detected in pool metadata");
#ifdef CONFIG_DM_DEBUG_BLOCK_STACK_TRACING
			DMERR("previously held here:");
			print_stack_trace(lock->traces + i, 4);

			DMERR("subsequent acquisition attempted here:");
			t.nr_entries = 0;
			t.max_entries = MAX_STACK;
			t.entries = entries;
			t.skip = 3;
			save_stack_trace(&t);
			print_stack_trace(&t, 4);
#endif
			return -EINVAL;
		}
	}

	return 0;
}

static void __wait(struct waiter *w)
{
	for (;;) {
		set_task_state(current, TASK_UNINTERRUPTIBLE);

		if (!w->task)
			break;

		schedule();
	}

	set_task_state(current, TASK_RUNNING);
}

static void __wake_waiter(struct waiter *w)
{
	struct task_struct *task;

	list_del(&w->list);
	task = w->task;
	smp_mb();
	w->task = NULL;
	wake_up_process(task);
}

/*
 * We either wake a few readers or a single writer.
 */
static void __wake_many(struct block_lock *lock)
{
	struct waiter *w, *tmp;

	BUG_ON(lock->count < 0);
	list_for_each_entry_safe(w, tmp, &lock->waiters, list) {
		if (lock->count >= MAX_HOLDERS)
			return;

		if (w->wants_write) {
			if (lock->count > 0)
				return; /* still read locked */

			lock->count = -1;
			__add_holder(lock, w->task);
			__wake_waiter(w);
			return;
		}

		lock->count++;
		__add_holder(lock, w->task);
		__wake_waiter(w);
	}
}

static void bl_init(struct block_lock *lock)
{
	int i;

	spin_lock_init(&lock->lock);
	lock->count = 0;
	INIT_LIST_HEAD(&lock->waiters);
	for (i = 0; i < MAX_HOLDERS; i++)
		lock->holders[i] = NULL;
}

static int __available_for_read(struct block_lock *lock)
{
	return lock->count >= 0 &&
		lock->count < MAX_HOLDERS &&
		list_empty(&lock->waiters);
}

static int bl_down_read(struct block_lock *lock)
{
	int r;
	struct waiter w;

	spin_lock(&lock->lock);
	r = __check_holder(lock);
	if (r) {
		spin_unlock(&lock->lock);
		return r;
	}

	if (__available_for_read(lock)) {
		lock->count++;
		__add_holder(lock, current);
		spin_unlock(&lock->lock);
		return 0;
	}

	get_task_struct(current);

	w.task = current;
	w.wants_write = 0;
	list_add_tail(&w.list, &lock->waiters);
	spin_unlock(&lock->lock);

	__wait(&w);
	put_task_struct(current);
	return 0;
}

static int bl_down_read_nonblock(struct block_lock *lock)
{
	int r;

	spin_lock(&lock->lock);
	r = __check_holder(lock);
	if (r)
		goto out;

	if (__available_for_read(lock)) {
		lock->count++;
		__add_holder(lock, current);
		r = 0;
	} else
		r = -EWOULDBLOCK;

out:
	spin_unlock(&lock->lock);
	return r;
}

static void bl_up_read(struct block_lock *lock)
{
	spin_lock(&lock->lock);
	BUG_ON(lock->count <= 0);
	__del_holder(lock, current);
	--lock->count;
	if (!list_empty(&lock->waiters))
		__wake_many(lock);
	spin_unlock(&lock->lock);
}

static int bl_down_write(struct block_lock *lock)
{
	int r;
	struct waiter w;

	spin_lock(&lock->lock);
	r = __check_holder(lock);
	if (r) {
		spin_unlock(&lock->lock);
		return r;
	}

	if (lock->count == 0 && list_empty(&lock->waiters)) {
		lock->count = -1;
		__add_holder(lock, current);
		spin_unlock(&lock->lock);
		return 0;
	}

	get_task_struct(current);
	w.task = current;
	w.wants_write = 1;

	/*
	 * Writers given priority. We know there's only one mutator in the
	 * system, so ignoring the ordering reversal.
	 */
	list_add(&w.list, &lock->waiters);
	spin_unlock(&lock->lock);

	__wait(&w);
	put_task_struct(current);

	return 0;
}

static void bl_up_write(struct block_lock *lock)
{
	spin_lock(&lock->lock);
	__del_holder(lock, current);
	lock->count = 0;
	if (!list_empty(&lock->waiters))
		__wake_many(lock);
	spin_unlock(&lock->lock);
}

static void report_recursive_bug(dm_block_t b, int r)
{
	if (r == -EINVAL)
		DMERR("recursive acquisition of block %llu requested.",
		      (unsigned long long) b);
}

/*----------------------------------------------------------------*/

/*
 * Block manager is currently implemented using dm-bufio.  struct
 * dm_block_manager and struct dm_block map directly onto a couple of
 * structs in the bufio interface.  I want to retain the freedom to move
 * away from bufio in the future.  So these structs are just cast within
 * this .c file, rather than making it through to the public interface.
 */
static struct dm_buffer *to_buffer(struct dm_block *b)
{
	return (struct dm_buffer *) b;
}

dm_block_t dm_block_location(struct dm_block *b)
{
	return dm_bufio_get_block_number(to_buffer(b));
}
EXPORT_SYMBOL_GPL(dm_block_location);

void *dm_block_data(struct dm_block *b)
{
	return dm_bufio_get_block_data(to_buffer(b));
}
EXPORT_SYMBOL_GPL(dm_block_data);

struct buffer_aux {
	struct dm_block_validator *validator;
	struct block_lock lock;
	int write_locked;
};

static void dm_block_manager_alloc_callback(struct dm_buffer *buf)
{
	struct buffer_aux *aux = dm_bufio_get_aux_data(buf);
	aux->validator = NULL;
	bl_init(&aux->lock);
}

static void dm_block_manager_write_callback(struct dm_buffer *buf)
{
	struct buffer_aux *aux = dm_bufio_get_aux_data(buf);
	if (aux->validator) {
		aux->validator->prepare_for_write(aux->validator, (struct dm_block *) buf,
			 dm_bufio_get_block_size(dm_bufio_get_client(buf)));
	}
}

/*----------------------------------------------------------------
 * Public interface
 *--------------------------------------------------------------*/
struct dm_block_manager {
	struct dm_bufio_client *bufio;
	bool read_only:1;
};

struct dm_block_manager *dm_block_manager_create(struct block_device *bdev,
						 unsigned block_size,
						 unsigned cache_size,
						 unsigned max_held_per_thread)
{
	int r;
	struct dm_block_manager *bm;

	bm = kmalloc(sizeof(*bm), GFP_KERNEL);
	if (!bm) {
		r = -ENOMEM;
		goto bad;
	}

	bm->bufio = dm_bufio_client_create(bdev, block_size, max_held_per_thread,
					   sizeof(struct buffer_aux),
					   dm_block_manager_alloc_callback,
					   dm_block_manager_write_callback);
	if (IS_ERR(bm->bufio)) {
		r = PTR_ERR(bm->bufio);
		kfree(bm);
		goto bad;
	}

	bm->read_only = false;

	return bm;

bad:
	return ERR_PTR(r);
}
EXPORT_SYMBOL_GPL(dm_block_manager_create);

void dm_block_manager_destroy(struct dm_block_manager *bm)
{
	dm_bufio_client_destroy(bm->bufio);
	kfree(bm);
}
EXPORT_SYMBOL_GPL(dm_block_manager_destroy);

unsigned dm_bm_block_size(struct dm_block_manager *bm)
{
	return dm_bufio_get_block_size(bm->bufio);
}
EXPORT_SYMBOL_GPL(dm_bm_block_size);

dm_block_t dm_bm_nr_blocks(struct dm_block_manager *bm)
{
	return dm_bufio_get_device_size(bm->bufio);
}

static int dm_bm_validate_buffer(struct dm_block_manager *bm,
				 struct dm_buffer *buf,
				 struct buffer_aux *aux,
				 struct dm_block_validator *v)
{
	if (unlikely(!aux->validator)) {
		int r;
		if (!v)
			return 0;
		r = v->check(v, (struct dm_block *) buf, dm_bufio_get_block_size(bm->bufio));
		if (unlikely(r)) {
			DMERR_LIMIT("%s validator check failed for block %llu", v->name,
				    (unsigned long long) dm_bufio_get_block_number(buf));
			return r;
		}
		aux->validator = v;
	} else {
		if (unlikely(aux->validator != v)) {
			DMERR_LIMIT("validator mismatch (old=%s vs new=%s) for block %llu",
				    aux->validator->name, v ? v->name : "NULL",
				    (unsigned long long) dm_bufio_get_block_number(buf));
			return -EINVAL;
		}
	}

	return 0;
}
int dm_bm_read_lock(struct dm_block_manager *bm, dm_block_t b,
		    struct dm_block_validator *v,
		    struct dm_block **result)
{
	struct buffer_aux *aux;
	void *p;
	int r;

	p = dm_bufio_read(bm->bufio, b, (struct dm_buffer **) result);
	if (unlikely(IS_ERR(p)))
		return PTR_ERR(p);

	aux = dm_bufio_get_aux_data(to_buffer(*result));
	r = bl_down_read(&aux->lock);
	if (unlikely(r)) {
		dm_bufio_release(to_buffer(*result));
		report_recursive_bug(b, r);
		return r;
	}

	aux->write_locked = 0;

	r = dm_bm_validate_buffer(bm, to_buffer(*result), aux, v);
	if (unlikely(r)) {
		bl_up_read(&aux->lock);
		dm_bufio_release(to_buffer(*result));
		return r;
	}

	return 0;
}
EXPORT_SYMBOL_GPL(dm_bm_read_lock);

int dm_bm_write_lock(struct dm_block_manager *bm,
		     dm_block_t b, struct dm_block_validator *v,
		     struct dm_block **result)
{
	struct buffer_aux *aux;
	void *p;
	int r;

	if (bm->read_only)
		return -EPERM;

	p = dm_bufio_read(bm->bufio, b, (struct dm_buffer **) result);
	if (unlikely(IS_ERR(p)))
		return PTR_ERR(p);

	aux = dm_bufio_get_aux_data(to_buffer(*result));
	r = bl_down_write(&aux->lock);
	if (r) {
		dm_bufio_release(to_buffer(*result));
		report_recursive_bug(b, r);
		return r;
	}

	aux->write_locked = 1;

	r = dm_bm_validate_buffer(bm, to_buffer(*result), aux, v);
	if (unlikely(r)) {
		bl_up_write(&aux->lock);
		dm_bufio_release(to_buffer(*result));
		return r;
	}

	return 0;
}
EXPORT_SYMBOL_GPL(dm_bm_write_lock);

int dm_bm_read_try_lock(struct dm_block_manager *bm,
			dm_block_t b, struct dm_block_validator *v,
			struct dm_block **result)
{
	struct buffer_aux *aux;
	void *p;
	int r;

	p = dm_bufio_get(bm->bufio, b, (struct dm_buffer **) result);
	if (unlikely(IS_ERR(p)))
		return PTR_ERR(p);
	if (unlikely(!p))
		return -EWOULDBLOCK;

	aux = dm_bufio_get_aux_data(to_buffer(*result));
	r = bl_down_read_nonblock(&aux->lock);
	if (r < 0) {
		dm_bufio_release(to_buffer(*result));
		report_recursive_bug(b, r);
		return r;
	}
	aux->write_locked = 0;

	r = dm_bm_validate_buffer(bm, to_buffer(*result), aux, v);
	if (unlikely(r)) {
		bl_up_read(&aux->lock);
		dm_bufio_release(to_buffer(*result));
		return r;
	}

	return 0;
}

int dm_bm_write_lock_zero(struct dm_block_manager *bm,
			  dm_block_t b, struct dm_block_validator *v,
			  struct dm_block **result)
{
	int r;
	struct buffer_aux *aux;
	void *p;

	if (bm->read_only)
		return -EPERM;

	p = dm_bufio_new(bm->bufio, b, (struct dm_buffer **) result);
	if (unlikely(IS_ERR(p)))
		return PTR_ERR(p);

	memset(p, 0, dm_bm_block_size(bm));

	aux = dm_bufio_get_aux_data(to_buffer(*result));
	r = bl_down_write(&aux->lock);
	if (r) {
		dm_bufio_release(to_buffer(*result));
		return r;
	}

	aux->write_locked = 1;
	aux->validator = v;

	return 0;
}
EXPORT_SYMBOL_GPL(dm_bm_write_lock_zero);

int dm_bm_unlock(struct dm_block *b)
{
	struct buffer_aux *aux;
	aux = dm_bufio_get_aux_data(to_buffer(b));

	if (aux->write_locked) {
		dm_bufio_mark_buffer_dirty(to_buffer(b));
		bl_up_write(&aux->lock);
	} else
		bl_up_read(&aux->lock);

	dm_bufio_release(to_buffer(b));

	return 0;
}
EXPORT_SYMBOL_GPL(dm_bm_unlock);

int dm_bm_flush_and_unlock(struct dm_block_manager *bm,
			   struct dm_block *superblock)
{
	int r;

	if (bm->read_only)
		return -EPERM;

	r = dm_bufio_write_dirty_buffers(bm->bufio);
	if (unlikely(r)) {
		dm_bm_unlock(superblock);
		return r;
	}

	dm_bm_unlock(superblock);

	return dm_bufio_write_dirty_buffers(bm->bufio);
}
EXPORT_SYMBOL_GPL(dm_bm_flush_and_unlock);

void dm_bm_prefetch(struct dm_block_manager *bm, dm_block_t b)
{
	dm_bufio_prefetch(bm->bufio, b, 1);
}

void dm_bm_set_read_only(struct dm_block_manager *bm)
{
	bm->read_only = true;
}
EXPORT_SYMBOL_GPL(dm_bm_set_read_only);

u32 dm_bm_checksum(const void *data, size_t len, u32 init_xor)
{
	return crc32c(~(u32) 0, data, len) ^ init_xor;
}
EXPORT_SYMBOL_GPL(dm_bm_checksum);

/*----------------------------------------------------------------*/

MODULE_LICENSE("GPL");
MODULE_AUTHOR("Joe Thornber <dm-devel@redhat.com>");
MODULE_DESCRIPTION("Immutable metadata library for dm");

/*----------------------------------------------------------------*/
OpenPOWER on IntegriCloud