aboutsummaryrefslogtreecommitdiff
path: root/drivers/md/faulty.c
blob: 6e83b38d931d6be62ad51e2855cf6008373abf80 (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
/*
 * faulty.c : Multiple Devices driver for Linux
 *
 * Copyright (C) 2004 Neil Brown
 *
 * fautly-device-simulator personality for md
 *
 *
 * This program is free software; you can redistribute it and/or modify
 * it under the terms of the GNU General Public License as published by
 * the Free Software Foundation; either version 2, or (at your option)
 * any later version.
 *
 * You should have received a copy of the GNU General Public License
 * (for example /usr/src/linux/COPYING); if not, write to the Free
 * Software Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
 */


/*
 * The "faulty" personality causes some requests to fail.
 *
 * Possible failure modes are:
 *   reads fail "randomly" but succeed on retry
 *   writes fail "randomly" but succeed on retry
 *   reads for some address fail and then persist until a write
 *   reads for some address fail and then persist irrespective of write
 *   writes for some address fail and persist
 *   all writes fail
 *
 * Different modes can be active at a time, but only
 * one can be set at array creation.  Others can be added later.
 * A mode can be one-shot or recurrent with the recurrance being
 * once in every N requests.
 * The bottom 5 bits of the "layout" indicate the mode.  The
 * remainder indicate a period, or 0 for one-shot.
 *
 * There is an implementation limit on the number of concurrently
 * persisting-faulty blocks. When a new fault is requested that would
 * exceed the limit, it is ignored.
 * All current faults can be clear using a layout of "0".
 *
 * Requests are always sent to the device.  If they are to fail,
 * we clone the bio and insert a new b_end_io into the chain.
 */

#define	WriteTransient	0
#define	ReadTransient	1
#define	WritePersistent	2
#define	ReadPersistent	3
#define	WriteAll	4 /* doesn't go to device */
#define	ReadFixable	5
#define	Modes	6

#define	ClearErrors	31
#define	ClearFaults	30

#define AllPersist	100 /* internal use only */
#define	NoPersist	101

#define	ModeMask	0x1f
#define	ModeShift	5

#define MaxFault	50
#include <linux/blkdev.h>
#include <linux/raid/md_u.h>
#include "md.h"
#include <linux/seq_file.h>


static void faulty_fail(struct bio *bio, int error)
{
	struct bio *b = bio->bi_private;

	b->bi_size = bio->bi_size;
	b->bi_sector = bio->bi_sector;

	bio_put(bio);

	bio_io_error(b);
}

typedef struct faulty_conf {
	int period[Modes];
	atomic_t counters[Modes];
	sector_t faults[MaxFault];
	int	modes[MaxFault];
	int nfaults;
	mdk_rdev_t *rdev;
} conf_t;

static int check_mode(conf_t *conf, int mode)
{
	if (conf->period[mode] == 0 &&
	    atomic_read(&conf->counters[mode]) <= 0)
		return 0; /* no failure, no decrement */


	if (atomic_dec_and_test(&conf->counters[mode])) {
		if (conf->period[mode])
			atomic_set(&conf->counters[mode], conf->period[mode]);
		return 1;
	}
	return 0;
}

static int check_sector(conf_t *conf, sector_t start, sector_t end, int dir)
{
	/* If we find a ReadFixable sector, we fix it ... */
	int i;
	for (i=0; i<conf->nfaults; i++)
		if (conf->faults[i] >= start &&
		    conf->faults[i] < end) {
			/* found it ... */
			switch (conf->modes[i] * 2 + dir) {
			case WritePersistent*2+WRITE: return 1;
			case ReadPersistent*2+READ: return 1;
			case ReadFixable*2+READ: return 1;
			case ReadFixable*2+WRITE:
				conf->modes[i] = NoPersist;
				return 0;
			case AllPersist*2+READ:
			case AllPersist*2+WRITE: return 1;
			default:
				return 0;
			}
		}
	return 0;
}

static void add_sector(conf_t *conf, sector_t start, int mode)
{
	int i;
	int n = conf->nfaults;
	for (i=0; i<conf->nfaults; i++)
		if (conf->faults[i] == start) {
			switch(mode) {
			case NoPersist: conf->modes[i] = mode; return;
			case WritePersistent:
				if (conf->modes[i] == ReadPersistent ||
				    conf->modes[i] == ReadFixable)
					conf->modes[i] = AllPersist;
				else
					conf->modes[i] = WritePersistent;
				return;
			case ReadPersistent:
				if (conf->modes[i] == WritePersistent)
					conf->modes[i] = AllPersist;
				else
					conf->modes[i] = ReadPersistent;
				return;
			case ReadFixable:
				if (conf->modes[i] == WritePersistent ||
				    conf->modes[i] == ReadPersistent)
					conf->modes[i] = AllPersist;
				else
					conf->modes[i] = ReadFixable;
				return;
			}
		} else if (conf->modes[i] == NoPersist)
			n = i;

	if (n >= MaxFault)
		return;
	conf->faults[n] = start;
	conf->modes[n] = mode;
	if (conf->nfaults == n)
		conf->nfaults = n+1;
}

static int make_request(struct request_queue *q, struct bio *bio)
{
	mddev_t *mddev = q->queuedata;
	conf_t *conf = (conf_t*)mddev->private;
	int failit = 0;

	if (bio_data_dir(bio) == WRITE) {
		/* write request */
		if (atomic_read(&conf->counters[WriteAll])) {
			/* special case - don't decrement, don't generic_make_request,
			 * just fail immediately
			 */
			bio_endio(bio, -EIO);
			return 0;
		}

		if (check_sector(conf, bio->bi_sector, bio->bi_sector+(bio->bi_size>>9),
				 WRITE))
			failit = 1;
		if (check_mode(conf, WritePersistent)) {
			add_sector(conf, bio->bi_sector, WritePersistent);
			failit = 1;
		}
		if (check_mode(conf, WriteTransient))
			failit = 1;
	} else {
		/* read request */
		if (check_sector(conf, bio->bi_sector, bio->bi_sector + (bio->bi_size>>9),
				 READ))
			failit = 1;
		if (check_mode(conf, ReadTransient))
			failit = 1;
		if (check_mode(conf, ReadPersistent)) {
			add_sector(conf, bio->bi_sector, ReadPersistent);
			failit = 1;
		}
		if (check_mode(conf, ReadFixable)) {
			add_sector(conf, bio->bi_sector, ReadFixable);
			failit = 1;
		}
	}
	if (failit) {
		struct bio *b = bio_clone(bio, GFP_NOIO);
		b->bi_bdev = conf->rdev->bdev;
		b->bi_private = bio;
		b->bi_end_io = faulty_fail;
		generic_make_request(b);
		return 0;
	} else {
		bio->bi_bdev = conf->rdev->bdev;
		return 1;
	}
}

static void status(struct seq_file *seq, mddev_t *mddev)
{
	conf_t *conf = (conf_t*)mddev->private;
	int n;

	if ((n=atomic_read(&conf->counters[WriteTransient])) != 0)
		seq_printf(seq, " WriteTransient=%d(%d)",
			   n, conf->period[WriteTransient]);

	if ((n=atomic_read(&conf->counters[ReadTransient])) != 0)
		seq_printf(seq, " ReadTransient=%d(%d)",
			   n, conf->period[ReadTransient]);

	if ((n=atomic_read(&conf->counters[WritePersistent])) != 0)
		seq_printf(seq, " WritePersistent=%d(%d)",
			   n, conf->period[WritePersistent]);

	if ((n=atomic_read(&conf->counters[ReadPersistent])) != 0)
		seq_printf(seq, " ReadPersistent=%d(%d)",
			   n, conf->period[ReadPersistent]);


	if ((n=atomic_read(&conf->counters[ReadFixable])) != 0)
		seq_printf(seq, " ReadFixable=%d(%d)",
			   n, conf->period[ReadFixable]);

	if ((n=atomic_read(&conf->counters[WriteAll])) != 0)
		seq_printf(seq, " WriteAll");

	seq_printf(seq, " nfaults=%d", conf->nfaults);
}


static int reshape(mddev_t *mddev)
{
	int mode = mddev->new_layout & ModeMask;
	int count = mddev->new_layout >> ModeShift;
	conf_t *conf = mddev->private;

	if (mddev->new_layout < 0)
		return 0;

	/* new layout */
	if (mode == ClearFaults)
		conf->nfaults = 0;
	else if (mode == ClearErrors) {
		int i;
		for (i=0 ; i < Modes ; i++) {
			conf->period[i] = 0;
			atomic_set(&conf->counters[i], 0);
		}
	} else if (mode < Modes) {
		conf->period[mode] = count;
		if (!count) count++;
		atomic_set(&conf->counters[mode], count);
	} else
		return -EINVAL;
	mddev->new_layout = -1;
	mddev->layout = -1; /* makes sure further changes come through */
	return 0;
}

static sector_t faulty_size(mddev_t *mddev, sector_t sectors, int raid_disks)
{
	WARN_ONCE(raid_disks,
		  "%s does not support generic reshape\n", __func__);

	if (sectors == 0)
		return mddev->dev_sectors;

	return sectors;
}

static int run(mddev_t *mddev)
{
	mdk_rdev_t *rdev;
	int i;

	conf_t *conf = kmalloc(sizeof(*conf), GFP_KERNEL);
	if (!conf)
		return -ENOMEM;

	for (i=0; i<Modes; i++) {
		atomic_set(&conf->counters[i], 0);
		conf->period[i] = 0;
	}
	conf->nfaults = 0;

	list_for_each_entry(rdev, &mddev->disks, same_set)
		conf->rdev = rdev;

	md_set_array_sectors(mddev, faulty_size(mddev, 0, 0));
	mddev->private = conf;

	reshape(mddev);

	return 0;
}

static int stop(mddev_t *mddev)
{
	conf_t *conf = (conf_t *)mddev->private;

	kfree(conf);
	mddev->private = NULL;
	return 0;
}

static struct mdk_personality faulty_personality =
{
	.name		= "faulty",
	.level		= LEVEL_FAULTY,
	.owner		= THIS_MODULE,
	.make_request	= make_request,
	.run		= run,
	.stop		= stop,
	.status		= status,
	.check_reshape	= reshape,
	.size		= faulty_size,
};

static int __init raid_init(void)
{
	return register_md_personality(&faulty_personality);
}

static void raid_exit(void)
{
	unregister_md_personality(&faulty_personality);
}

module_init(raid_init);
module_exit(raid_exit);
MODULE_LICENSE("GPL");
MODULE_ALIAS("md-personality-10"); /* faulty */
MODULE_ALIAS("md-faulty");
MODULE_ALIAS("md-level--5");