xref: /openbmc/linux/block/badblocks.c (revision bc5aa3a0)
1 /*
2  * Bad block management
3  *
4  * - Heavily based on MD badblocks code from Neil Brown
5  *
6  * Copyright (c) 2015, Intel Corporation.
7  *
8  * This program is free software; you can redistribute it and/or modify it
9  * under the terms and conditions of the GNU General Public License,
10  * version 2, as published by the Free Software Foundation.
11  *
12  * This program is distributed in the hope it will be useful, but WITHOUT
13  * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
14  * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
15  * more details.
16  */
17 
18 #include <linux/badblocks.h>
19 #include <linux/seqlock.h>
20 #include <linux/device.h>
21 #include <linux/kernel.h>
22 #include <linux/module.h>
23 #include <linux/stddef.h>
24 #include <linux/types.h>
25 #include <linux/slab.h>
26 
27 /**
28  * badblocks_check() - check a given range for bad sectors
29  * @bb:		the badblocks structure that holds all badblock information
30  * @s:		sector (start) at which to check for badblocks
31  * @sectors:	number of sectors to check for badblocks
32  * @first_bad:	pointer to store location of the first badblock
33  * @bad_sectors: pointer to store number of badblocks after @first_bad
34  *
35  * We can record which blocks on each device are 'bad' and so just
36  * fail those blocks, or that stripe, rather than the whole device.
37  * Entries in the bad-block table are 64bits wide.  This comprises:
38  * Length of bad-range, in sectors: 0-511 for lengths 1-512
39  * Start of bad-range, sector offset, 54 bits (allows 8 exbibytes)
40  *  A 'shift' can be set so that larger blocks are tracked and
41  *  consequently larger devices can be covered.
42  * 'Acknowledged' flag - 1 bit. - the most significant bit.
43  *
44  * Locking of the bad-block table uses a seqlock so badblocks_check
45  * might need to retry if it is very unlucky.
46  * We will sometimes want to check for bad blocks in a bi_end_io function,
47  * so we use the write_seqlock_irq variant.
48  *
49  * When looking for a bad block we specify a range and want to
50  * know if any block in the range is bad.  So we binary-search
51  * to the last range that starts at-or-before the given endpoint,
52  * (or "before the sector after the target range")
53  * then see if it ends after the given start.
54  *
55  * Return:
56  *  0: there are no known bad blocks in the range
57  *  1: there are known bad block which are all acknowledged
58  * -1: there are bad blocks which have not yet been acknowledged in metadata.
59  * plus the start/length of the first bad section we overlap.
60  */
61 int badblocks_check(struct badblocks *bb, sector_t s, int sectors,
62 			sector_t *first_bad, int *bad_sectors)
63 {
64 	int hi;
65 	int lo;
66 	u64 *p = bb->page;
67 	int rv;
68 	sector_t target = s + sectors;
69 	unsigned seq;
70 
71 	if (bb->shift > 0) {
72 		/* round the start down, and the end up */
73 		s >>= bb->shift;
74 		target += (1<<bb->shift) - 1;
75 		target >>= bb->shift;
76 		sectors = target - s;
77 	}
78 	/* 'target' is now the first block after the bad range */
79 
80 retry:
81 	seq = read_seqbegin(&bb->lock);
82 	lo = 0;
83 	rv = 0;
84 	hi = bb->count;
85 
86 	/* Binary search between lo and hi for 'target'
87 	 * i.e. for the last range that starts before 'target'
88 	 */
89 	/* INVARIANT: ranges before 'lo' and at-or-after 'hi'
90 	 * are known not to be the last range before target.
91 	 * VARIANT: hi-lo is the number of possible
92 	 * ranges, and decreases until it reaches 1
93 	 */
94 	while (hi - lo > 1) {
95 		int mid = (lo + hi) / 2;
96 		sector_t a = BB_OFFSET(p[mid]);
97 
98 		if (a < target)
99 			/* This could still be the one, earlier ranges
100 			 * could not.
101 			 */
102 			lo = mid;
103 		else
104 			/* This and later ranges are definitely out. */
105 			hi = mid;
106 	}
107 	/* 'lo' might be the last that started before target, but 'hi' isn't */
108 	if (hi > lo) {
109 		/* need to check all range that end after 's' to see if
110 		 * any are unacknowledged.
111 		 */
112 		while (lo >= 0 &&
113 		       BB_OFFSET(p[lo]) + BB_LEN(p[lo]) > s) {
114 			if (BB_OFFSET(p[lo]) < target) {
115 				/* starts before the end, and finishes after
116 				 * the start, so they must overlap
117 				 */
118 				if (rv != -1 && BB_ACK(p[lo]))
119 					rv = 1;
120 				else
121 					rv = -1;
122 				*first_bad = BB_OFFSET(p[lo]);
123 				*bad_sectors = BB_LEN(p[lo]);
124 			}
125 			lo--;
126 		}
127 	}
128 
129 	if (read_seqretry(&bb->lock, seq))
130 		goto retry;
131 
132 	return rv;
133 }
134 EXPORT_SYMBOL_GPL(badblocks_check);
135 
136 /**
137  * badblocks_set() - Add a range of bad blocks to the table.
138  * @bb:		the badblocks structure that holds all badblock information
139  * @s:		first sector to mark as bad
140  * @sectors:	number of sectors to mark as bad
141  * @acknowledged: weather to mark the bad sectors as acknowledged
142  *
143  * This might extend the table, or might contract it if two adjacent ranges
144  * can be merged. We binary-search to find the 'insertion' point, then
145  * decide how best to handle it.
146  *
147  * Return:
148  *  0: success
149  *  1: failed to set badblocks (out of space)
150  */
151 int badblocks_set(struct badblocks *bb, sector_t s, int sectors,
152 			int acknowledged)
153 {
154 	u64 *p;
155 	int lo, hi;
156 	int rv = 0;
157 	unsigned long flags;
158 
159 	if (bb->shift < 0)
160 		/* badblocks are disabled */
161 		return 0;
162 
163 	if (bb->shift) {
164 		/* round the start down, and the end up */
165 		sector_t next = s + sectors;
166 
167 		s >>= bb->shift;
168 		next += (1<<bb->shift) - 1;
169 		next >>= bb->shift;
170 		sectors = next - s;
171 	}
172 
173 	write_seqlock_irqsave(&bb->lock, flags);
174 
175 	p = bb->page;
176 	lo = 0;
177 	hi = bb->count;
178 	/* Find the last range that starts at-or-before 's' */
179 	while (hi - lo > 1) {
180 		int mid = (lo + hi) / 2;
181 		sector_t a = BB_OFFSET(p[mid]);
182 
183 		if (a <= s)
184 			lo = mid;
185 		else
186 			hi = mid;
187 	}
188 	if (hi > lo && BB_OFFSET(p[lo]) > s)
189 		hi = lo;
190 
191 	if (hi > lo) {
192 		/* we found a range that might merge with the start
193 		 * of our new range
194 		 */
195 		sector_t a = BB_OFFSET(p[lo]);
196 		sector_t e = a + BB_LEN(p[lo]);
197 		int ack = BB_ACK(p[lo]);
198 
199 		if (e >= s) {
200 			/* Yes, we can merge with a previous range */
201 			if (s == a && s + sectors >= e)
202 				/* new range covers old */
203 				ack = acknowledged;
204 			else
205 				ack = ack && acknowledged;
206 
207 			if (e < s + sectors)
208 				e = s + sectors;
209 			if (e - a <= BB_MAX_LEN) {
210 				p[lo] = BB_MAKE(a, e-a, ack);
211 				s = e;
212 			} else {
213 				/* does not all fit in one range,
214 				 * make p[lo] maximal
215 				 */
216 				if (BB_LEN(p[lo]) != BB_MAX_LEN)
217 					p[lo] = BB_MAKE(a, BB_MAX_LEN, ack);
218 				s = a + BB_MAX_LEN;
219 			}
220 			sectors = e - s;
221 		}
222 	}
223 	if (sectors && hi < bb->count) {
224 		/* 'hi' points to the first range that starts after 's'.
225 		 * Maybe we can merge with the start of that range
226 		 */
227 		sector_t a = BB_OFFSET(p[hi]);
228 		sector_t e = a + BB_LEN(p[hi]);
229 		int ack = BB_ACK(p[hi]);
230 
231 		if (a <= s + sectors) {
232 			/* merging is possible */
233 			if (e <= s + sectors) {
234 				/* full overlap */
235 				e = s + sectors;
236 				ack = acknowledged;
237 			} else
238 				ack = ack && acknowledged;
239 
240 			a = s;
241 			if (e - a <= BB_MAX_LEN) {
242 				p[hi] = BB_MAKE(a, e-a, ack);
243 				s = e;
244 			} else {
245 				p[hi] = BB_MAKE(a, BB_MAX_LEN, ack);
246 				s = a + BB_MAX_LEN;
247 			}
248 			sectors = e - s;
249 			lo = hi;
250 			hi++;
251 		}
252 	}
253 	if (sectors == 0 && hi < bb->count) {
254 		/* we might be able to combine lo and hi */
255 		/* Note: 's' is at the end of 'lo' */
256 		sector_t a = BB_OFFSET(p[hi]);
257 		int lolen = BB_LEN(p[lo]);
258 		int hilen = BB_LEN(p[hi]);
259 		int newlen = lolen + hilen - (s - a);
260 
261 		if (s >= a && newlen < BB_MAX_LEN) {
262 			/* yes, we can combine them */
263 			int ack = BB_ACK(p[lo]) && BB_ACK(p[hi]);
264 
265 			p[lo] = BB_MAKE(BB_OFFSET(p[lo]), newlen, ack);
266 			memmove(p + hi, p + hi + 1,
267 				(bb->count - hi - 1) * 8);
268 			bb->count--;
269 		}
270 	}
271 	while (sectors) {
272 		/* didn't merge (it all).
273 		 * Need to add a range just before 'hi'
274 		 */
275 		if (bb->count >= MAX_BADBLOCKS) {
276 			/* No room for more */
277 			rv = 1;
278 			break;
279 		} else {
280 			int this_sectors = sectors;
281 
282 			memmove(p + hi + 1, p + hi,
283 				(bb->count - hi) * 8);
284 			bb->count++;
285 
286 			if (this_sectors > BB_MAX_LEN)
287 				this_sectors = BB_MAX_LEN;
288 			p[hi] = BB_MAKE(s, this_sectors, acknowledged);
289 			sectors -= this_sectors;
290 			s += this_sectors;
291 		}
292 	}
293 
294 	bb->changed = 1;
295 	if (!acknowledged)
296 		bb->unacked_exist = 1;
297 	write_sequnlock_irqrestore(&bb->lock, flags);
298 
299 	return rv;
300 }
301 EXPORT_SYMBOL_GPL(badblocks_set);
302 
303 /**
304  * badblocks_clear() - Remove a range of bad blocks to the table.
305  * @bb:		the badblocks structure that holds all badblock information
306  * @s:		first sector to mark as bad
307  * @sectors:	number of sectors to mark as bad
308  *
309  * This may involve extending the table if we spilt a region,
310  * but it must not fail.  So if the table becomes full, we just
311  * drop the remove request.
312  *
313  * Return:
314  *  0: success
315  *  1: failed to clear badblocks
316  */
317 int badblocks_clear(struct badblocks *bb, sector_t s, int sectors)
318 {
319 	u64 *p;
320 	int lo, hi;
321 	sector_t target = s + sectors;
322 	int rv = 0;
323 
324 	if (bb->shift > 0) {
325 		/* When clearing we round the start up and the end down.
326 		 * This should not matter as the shift should align with
327 		 * the block size and no rounding should ever be needed.
328 		 * However it is better the think a block is bad when it
329 		 * isn't than to think a block is not bad when it is.
330 		 */
331 		s += (1<<bb->shift) - 1;
332 		s >>= bb->shift;
333 		target >>= bb->shift;
334 		sectors = target - s;
335 	}
336 
337 	write_seqlock_irq(&bb->lock);
338 
339 	p = bb->page;
340 	lo = 0;
341 	hi = bb->count;
342 	/* Find the last range that starts before 'target' */
343 	while (hi - lo > 1) {
344 		int mid = (lo + hi) / 2;
345 		sector_t a = BB_OFFSET(p[mid]);
346 
347 		if (a < target)
348 			lo = mid;
349 		else
350 			hi = mid;
351 	}
352 	if (hi > lo) {
353 		/* p[lo] is the last range that could overlap the
354 		 * current range.  Earlier ranges could also overlap,
355 		 * but only this one can overlap the end of the range.
356 		 */
357 		if (BB_OFFSET(p[lo]) + BB_LEN(p[lo]) > target) {
358 			/* Partial overlap, leave the tail of this range */
359 			int ack = BB_ACK(p[lo]);
360 			sector_t a = BB_OFFSET(p[lo]);
361 			sector_t end = a + BB_LEN(p[lo]);
362 
363 			if (a < s) {
364 				/* we need to split this range */
365 				if (bb->count >= MAX_BADBLOCKS) {
366 					rv = -ENOSPC;
367 					goto out;
368 				}
369 				memmove(p+lo+1, p+lo, (bb->count - lo) * 8);
370 				bb->count++;
371 				p[lo] = BB_MAKE(a, s-a, ack);
372 				lo++;
373 			}
374 			p[lo] = BB_MAKE(target, end - target, ack);
375 			/* there is no longer an overlap */
376 			hi = lo;
377 			lo--;
378 		}
379 		while (lo >= 0 &&
380 		       BB_OFFSET(p[lo]) + BB_LEN(p[lo]) > s) {
381 			/* This range does overlap */
382 			if (BB_OFFSET(p[lo]) < s) {
383 				/* Keep the early parts of this range. */
384 				int ack = BB_ACK(p[lo]);
385 				sector_t start = BB_OFFSET(p[lo]);
386 
387 				p[lo] = BB_MAKE(start, s - start, ack);
388 				/* now low doesn't overlap, so.. */
389 				break;
390 			}
391 			lo--;
392 		}
393 		/* 'lo' is strictly before, 'hi' is strictly after,
394 		 * anything between needs to be discarded
395 		 */
396 		if (hi - lo > 1) {
397 			memmove(p+lo+1, p+hi, (bb->count - hi) * 8);
398 			bb->count -= (hi - lo - 1);
399 		}
400 	}
401 
402 	bb->changed = 1;
403 out:
404 	write_sequnlock_irq(&bb->lock);
405 	return rv;
406 }
407 EXPORT_SYMBOL_GPL(badblocks_clear);
408 
409 /**
410  * ack_all_badblocks() - Acknowledge all bad blocks in a list.
411  * @bb:		the badblocks structure that holds all badblock information
412  *
413  * This only succeeds if ->changed is clear.  It is used by
414  * in-kernel metadata updates
415  */
416 void ack_all_badblocks(struct badblocks *bb)
417 {
418 	if (bb->page == NULL || bb->changed)
419 		/* no point even trying */
420 		return;
421 	write_seqlock_irq(&bb->lock);
422 
423 	if (bb->changed == 0 && bb->unacked_exist) {
424 		u64 *p = bb->page;
425 		int i;
426 
427 		for (i = 0; i < bb->count ; i++) {
428 			if (!BB_ACK(p[i])) {
429 				sector_t start = BB_OFFSET(p[i]);
430 				int len = BB_LEN(p[i]);
431 
432 				p[i] = BB_MAKE(start, len, 1);
433 			}
434 		}
435 		bb->unacked_exist = 0;
436 	}
437 	write_sequnlock_irq(&bb->lock);
438 }
439 EXPORT_SYMBOL_GPL(ack_all_badblocks);
440 
441 /**
442  * badblocks_show() - sysfs access to bad-blocks list
443  * @bb:		the badblocks structure that holds all badblock information
444  * @page:	buffer received from sysfs
445  * @unack:	weather to show unacknowledged badblocks
446  *
447  * Return:
448  *  Length of returned data
449  */
450 ssize_t badblocks_show(struct badblocks *bb, char *page, int unack)
451 {
452 	size_t len;
453 	int i;
454 	u64 *p = bb->page;
455 	unsigned seq;
456 
457 	if (bb->shift < 0)
458 		return 0;
459 
460 retry:
461 	seq = read_seqbegin(&bb->lock);
462 
463 	len = 0;
464 	i = 0;
465 
466 	while (len < PAGE_SIZE && i < bb->count) {
467 		sector_t s = BB_OFFSET(p[i]);
468 		unsigned int length = BB_LEN(p[i]);
469 		int ack = BB_ACK(p[i]);
470 
471 		i++;
472 
473 		if (unack && ack)
474 			continue;
475 
476 		len += snprintf(page+len, PAGE_SIZE-len, "%llu %u\n",
477 				(unsigned long long)s << bb->shift,
478 				length << bb->shift);
479 	}
480 	if (unack && len == 0)
481 		bb->unacked_exist = 0;
482 
483 	if (read_seqretry(&bb->lock, seq))
484 		goto retry;
485 
486 	return len;
487 }
488 EXPORT_SYMBOL_GPL(badblocks_show);
489 
490 /**
491  * badblocks_store() - sysfs access to bad-blocks list
492  * @bb:		the badblocks structure that holds all badblock information
493  * @page:	buffer received from sysfs
494  * @len:	length of data received from sysfs
495  * @unack:	weather to show unacknowledged badblocks
496  *
497  * Return:
498  *  Length of the buffer processed or -ve error.
499  */
500 ssize_t badblocks_store(struct badblocks *bb, const char *page, size_t len,
501 			int unack)
502 {
503 	unsigned long long sector;
504 	int length;
505 	char newline;
506 
507 	switch (sscanf(page, "%llu %d%c", &sector, &length, &newline)) {
508 	case 3:
509 		if (newline != '\n')
510 			return -EINVAL;
511 	case 2:
512 		if (length <= 0)
513 			return -EINVAL;
514 		break;
515 	default:
516 		return -EINVAL;
517 	}
518 
519 	if (badblocks_set(bb, sector, length, !unack))
520 		return -ENOSPC;
521 	else
522 		return len;
523 }
524 EXPORT_SYMBOL_GPL(badblocks_store);
525 
526 static int __badblocks_init(struct device *dev, struct badblocks *bb,
527 		int enable)
528 {
529 	bb->dev = dev;
530 	bb->count = 0;
531 	if (enable)
532 		bb->shift = 0;
533 	else
534 		bb->shift = -1;
535 	if (dev)
536 		bb->page = devm_kzalloc(dev, PAGE_SIZE, GFP_KERNEL);
537 	else
538 		bb->page = kzalloc(PAGE_SIZE, GFP_KERNEL);
539 	if (!bb->page) {
540 		bb->shift = -1;
541 		return -ENOMEM;
542 	}
543 	seqlock_init(&bb->lock);
544 
545 	return 0;
546 }
547 
548 /**
549  * badblocks_init() - initialize the badblocks structure
550  * @bb:		the badblocks structure that holds all badblock information
551  * @enable:	weather to enable badblocks accounting
552  *
553  * Return:
554  *  0: success
555  *  -ve errno: on error
556  */
557 int badblocks_init(struct badblocks *bb, int enable)
558 {
559 	return __badblocks_init(NULL, bb, enable);
560 }
561 EXPORT_SYMBOL_GPL(badblocks_init);
562 
563 int devm_init_badblocks(struct device *dev, struct badblocks *bb)
564 {
565 	if (!bb)
566 		return -EINVAL;
567 	return __badblocks_init(dev, bb, 1);
568 }
569 EXPORT_SYMBOL_GPL(devm_init_badblocks);
570 
571 /**
572  * badblocks_exit() - free the badblocks structure
573  * @bb:		the badblocks structure that holds all badblock information
574  */
575 void badblocks_exit(struct badblocks *bb)
576 {
577 	if (!bb)
578 		return;
579 	if (bb->dev)
580 		devm_kfree(bb->dev, bb->page);
581 	else
582 		kfree(bb->page);
583 	bb->page = NULL;
584 }
585 EXPORT_SYMBOL_GPL(badblocks_exit);
586