summaryrefslogtreecommitdiffstats
path: root/fs/nilfs2/mdt.c
blob: 0d7b71fbeff8fb5aa4915a652091d95f8f06dd04 (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
642
643
644
645
646
647
648
649
650
651
652
/*
 * mdt.c - meta data file for NILFS
 *
 * Copyright (C) 2005-2008 Nippon Telegraph and Telephone Corporation.
 *
 * 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 of the License, or
 * (at your option) any later version.
 *
 * This program is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 * GNU General Public License for more details.
 *
 * Written by Ryusuke Konishi.
 */

#include <linux/buffer_head.h>
#include <linux/mpage.h>
#include <linux/mm.h>
#include <linux/writeback.h>
#include <linux/backing-dev.h>
#include <linux/swap.h>
#include <linux/slab.h>
#include "nilfs.h"
#include "btnode.h"
#include "segment.h"
#include "page.h"
#include "mdt.h"
#include "alloc.h"		/* nilfs_palloc_destroy_cache() */

#include <trace/events/nilfs2.h>

#define NILFS_MDT_MAX_RA_BLOCKS		(16 - 1)


static int
nilfs_mdt_insert_new_block(struct inode *inode, unsigned long block,
			   struct buffer_head *bh,
			   void (*init_block)(struct inode *,
					      struct buffer_head *, void *))
{
	struct nilfs_inode_info *ii = NILFS_I(inode);
	void *kaddr;
	int ret;

	/* Caller exclude read accesses using page lock */

	/* set_buffer_new(bh); */
	bh->b_blocknr = 0;

	ret = nilfs_bmap_insert(ii->i_bmap, block, (unsigned long)bh);
	if (unlikely(ret))
		return ret;

	set_buffer_mapped(bh);

	kaddr = kmap_atomic(bh->b_page);
	memset(kaddr + bh_offset(bh), 0, 1 << inode->i_blkbits);
	if (init_block)
		init_block(inode, bh, kaddr);
	flush_dcache_page(bh->b_page);
	kunmap_atomic(kaddr);

	set_buffer_uptodate(bh);
	mark_buffer_dirty(bh);
	nilfs_mdt_mark_dirty(inode);

	trace_nilfs2_mdt_insert_new_block(inode, inode->i_ino, block);

	return 0;
}

static int nilfs_mdt_create_block(struct inode *inode, unsigned long block,
				  struct buffer_head **out_bh,
				  void (*init_block)(struct inode *,
						     struct buffer_head *,
						     void *))
{
	struct super_block *sb = inode->i_sb;
	struct nilfs_transaction_info ti;
	struct buffer_head *bh;
	int err;

	nilfs_transaction_begin(sb, &ti, 0);

	err = -ENOMEM;
	bh = nilfs_grab_buffer(inode, inode->i_mapping, block, 0);
	if (unlikely(!bh))
		goto failed_unlock;

	err = -EEXIST;
	if (buffer_uptodate(bh))
		goto failed_bh;

	wait_on_buffer(bh);
	if (buffer_uptodate(bh))
		goto failed_bh;

	bh->b_bdev = sb->s_bdev;
	err = nilfs_mdt_insert_new_block(inode, block, bh, init_block);
	if (likely(!err)) {
		get_bh(bh);
		*out_bh = bh;
	}

 failed_bh:
	unlock_page(bh->b_page);
	put_page(bh->b_page);
	brelse(bh);

 failed_unlock:
	if (likely(!err))
		err = nilfs_transaction_commit(sb);
	else
		nilfs_transaction_abort(sb);

	return err;
}

static int
nilfs_mdt_submit_block(struct inode *inode, unsigned long blkoff,
		       int mode, int mode_flags, struct buffer_head **out_bh)
{
	struct buffer_head *bh;
	__u64 blknum = 0;
	int ret = -ENOMEM;

	bh = nilfs_grab_buffer(inode, inode->i_mapping, blkoff, 0);
	if (unlikely(!bh))
		goto failed;

	ret = -EEXIST; /* internal code */
	if (buffer_uptodate(bh))
		goto out;

	if (mode_flags & REQ_RAHEAD) {
		if (!trylock_buffer(bh)) {
			ret = -EBUSY;
			goto failed_bh;
		}
	} else /* mode == READ */
		lock_buffer(bh);

	if (buffer_uptodate(bh)) {
		unlock_buffer(bh);
		goto out;
	}

	ret = nilfs_bmap_lookup(NILFS_I(inode)->i_bmap, blkoff, &blknum);
	if (unlikely(ret)) {
		unlock_buffer(bh);
		goto failed_bh;
	}
	map_bh(bh, inode->i_sb, (sector_t)blknum);

	bh->b_end_io = end_buffer_read_sync;
	get_bh(bh);
	submit_bh(mode, mode_flags, bh);
	ret = 0;

	trace_nilfs2_mdt_submit_block(inode, inode->i_ino, blkoff, mode);
 out:
	get_bh(bh);
	*out_bh = bh;

 failed_bh:
	unlock_page(bh->b_page);
	put_page(bh->b_page);
	brelse(bh);
 failed:
	return ret;
}

static int nilfs_mdt_read_block(struct inode *inode, unsigned long block,
				int readahead, struct buffer_head **out_bh)
{
	struct buffer_head *first_bh, *bh;
	unsigned long blkoff;
	int i, nr_ra_blocks = NILFS_MDT_MAX_RA_BLOCKS;
	int err;

	err = nilfs_mdt_submit_block(inode, block, REQ_OP_READ, 0, &first_bh);
	if (err == -EEXIST) /* internal code */
		goto out;

	if (unlikely(err))
		goto failed;

	if (readahead) {
		blkoff = block + 1;
		for (i = 0; i < nr_ra_blocks; i++, blkoff++) {
			err = nilfs_mdt_submit_block(inode, blkoff, REQ_OP_READ,
						     REQ_RAHEAD, &bh);
			if (likely(!err || err == -EEXIST))
				brelse(bh);
			else if (err != -EBUSY)
				break;
				/* abort readahead if bmap lookup failed */
			if (!buffer_locked(first_bh))
				goto out_no_wait;
		}
	}

	wait_on_buffer(first_bh);

 out_no_wait:
	err = -EIO;
	if (!buffer_uptodate(first_bh))
		goto failed_bh;
 out:
	*out_bh = first_bh;
	return 0;

 failed_bh:
	brelse(first_bh);
 failed:
	return err;
}

/**
 * nilfs_mdt_get_block - read or create a buffer on meta data file.
 * @inode: inode of the meta data file
 * @blkoff: block offset
 * @create: create flag
 * @init_block: initializer used for newly allocated block
 * @out_bh: output of a pointer to the buffer_head
 *
 * nilfs_mdt_get_block() looks up the specified buffer and tries to create
 * a new buffer if @create is not zero.  On success, the returned buffer is
 * assured to be either existing or formatted using a buffer lock on success.
 * @out_bh is substituted only when zero is returned.
 *
 * Return Value: On success, it returns 0. On error, the following negative
 * error code is returned.
 *
 * %-ENOMEM - Insufficient memory available.
 *
 * %-EIO - I/O error
 *
 * %-ENOENT - the specified block does not exist (hole block)
 *
 * %-EROFS - Read only filesystem (for create mode)
 */
int nilfs_mdt_get_block(struct inode *inode, unsigned long blkoff, int create,
			void (*init_block)(struct inode *,
					   struct buffer_head *, void *),
			struct buffer_head **out_bh)
{
	int ret;

	/* Should be rewritten with merging nilfs_mdt_read_block() */
 retry:
	ret = nilfs_mdt_read_block(inode, blkoff, !create, out_bh);
	if (!create || ret != -ENOENT)
		return ret;

	ret = nilfs_mdt_create_block(inode, blkoff, out_bh, init_block);
	if (unlikely(ret == -EEXIST)) {
		/* create = 0; */  /* limit read-create loop retries */
		goto retry;
	}
	return ret;
}

/**
 * nilfs_mdt_find_block - find and get a buffer on meta data file.
 * @inode: inode of the meta data file
 * @start: start block offset (inclusive)
 * @end: end block offset (inclusive)
 * @blkoff: block offset
 * @out_bh: place to store a pointer to buffer_head struct
 *
 * nilfs_mdt_find_block() looks up an existing block in range of
 * [@start, @end] and stores pointer to a buffer head of the block to
 * @out_bh, and block offset to @blkoff, respectively.  @out_bh and
 * @blkoff are substituted only when zero is returned.
 *
 * Return Value: On success, it returns 0. On error, the following negative
 * error code is returned.
 *
 * %-ENOMEM - Insufficient memory available.
 *
 * %-EIO - I/O error
 *
 * %-ENOENT - no block was found in the range
 */
int nilfs_mdt_find_block(struct inode *inode, unsigned long start,
			 unsigned long end, unsigned long *blkoff,
			 struct buffer_head **out_bh)
{
	__u64 next;
	int ret;

	if (unlikely(start > end))
		return -ENOENT;

	ret = nilfs_mdt_read_block(inode, start, true, out_bh);
	if (!ret) {
		*blkoff = start;
		goto out;
	}
	if (unlikely(ret != -ENOENT || start == ULONG_MAX))
		goto out;

	ret = nilfs_bmap_seek_key(NILFS_I(inode)->i_bmap, start + 1, &next);
	if (!ret) {
		if (next <= end) {
			ret = nilfs_mdt_read_block(inode, next, true, out_bh);
			if (!ret)
				*blkoff = next;
		} else {
			ret = -ENOENT;
		}
	}
out:
	return ret;
}

/**
 * nilfs_mdt_delete_block - make a hole on the meta data file.
 * @inode: inode of the meta data file
 * @block: block offset
 *
 * Return Value: On success, zero is returned.
 * On error, one of the following negative error code is returned.
 *
 * %-ENOMEM - Insufficient memory available.
 *
 * %-EIO - I/O error
 */
int nilfs_mdt_delete_block(struct inode *inode, unsigned long block)
{
	struct nilfs_inode_info *ii = NILFS_I(inode);
	int err;

	err = nilfs_bmap_delete(ii->i_bmap, block);
	if (!err || err == -ENOENT) {
		nilfs_mdt_mark_dirty(inode);
		nilfs_mdt_forget_block(inode, block);
	}
	return err;
}

/**
 * nilfs_mdt_forget_block - discard dirty state and try to remove the page
 * @inode: inode of the meta data file
 * @block: block offset
 *
 * nilfs_mdt_forget_block() clears a dirty flag of the specified buffer, and
 * tries to release the page including the buffer from a page cache.
 *
 * Return Value: On success, 0 is returned. On error, one of the following
 * negative error code is returned.
 *
 * %-EBUSY - page has an active buffer.
 *
 * %-ENOENT - page cache has no page addressed by the offset.
 */
int nilfs_mdt_forget_block(struct inode *inode, unsigned long block)
{
	pgoff_t index = (pgoff_t)block >>
		(PAGE_SHIFT - inode->i_blkbits);
	struct page *page;
	unsigned long first_block;
	int ret = 0;
	int still_dirty;

	page = find_lock_page(inode->i_mapping, index);
	if (!page)
		return -ENOENT;

	wait_on_page_writeback(page);

	first_block = (unsigned long)index <<
		(PAGE_SHIFT - inode->i_blkbits);
	if (page_has_buffers(page)) {
		struct buffer_head *bh;

		bh = nilfs_page_get_nth_block(page, block - first_block);
		nilfs_forget_buffer(bh);
	}
	still_dirty = PageDirty(page);
	unlock_page(page);
	put_page(page);

	if (still_dirty ||
	    invalidate_inode_pages2_range(inode->i_mapping, index, index) != 0)
		ret = -EBUSY;
	return ret;
}

int nilfs_mdt_fetch_dirty(struct inode *inode)
{
	struct nilfs_inode_info *ii = NILFS_I(inode);

	if (nilfs_bmap_test_and_clear_dirty(ii->i_bmap)) {
		set_bit(NILFS_I_DIRTY, &ii->i_state);
		return 1;
	}
	return test_bit(NILFS_I_DIRTY, &ii->i_state);
}

static int
nilfs_mdt_write_page(struct page *page, struct writeback_control *wbc)
{
	struct inode *inode = page->mapping->host;
	struct super_block *sb;
	int err = 0;

	if (inode && (inode->i_sb->s_flags & MS_RDONLY)) {
		/*
		 * It means that filesystem was remounted in read-only
		 * mode because of error or metadata corruption. But we
		 * have dirty pages that try to be flushed in background.
		 * So, here we simply discard this dirty page.
		 */
		nilfs_clear_dirty_page(page, false);
		unlock_page(page);
		return -EROFS;
	}

	redirty_page_for_writepage(wbc, page);
	unlock_page(page);

	if (!inode)
		return 0;

	sb = inode->i_sb;

	if (wbc->sync_mode == WB_SYNC_ALL)
		err = nilfs_construct_segment(sb);
	else if (wbc->for_reclaim)
		nilfs_flush_segment(sb, inode->i_ino);

	return err;
}


static const struct address_space_operations def_mdt_aops = {
	.writepage		= nilfs_mdt_write_page,
};

static const struct inode_operations def_mdt_iops;
static const struct file_operations def_mdt_fops;


int nilfs_mdt_init(struct inode *inode, gfp_t gfp_mask, size_t objsz)
{
	struct nilfs_mdt_info *mi;

	mi = kzalloc(max(sizeof(*mi), objsz), GFP_NOFS);
	if (!mi)
		return -ENOMEM;

	init_rwsem(&mi->mi_sem);
	inode->i_private = mi;

	inode->i_mode = S_IFREG;
	mapping_set_gfp_mask(inode->i_mapping, gfp_mask);

	inode->i_op = &def_mdt_iops;
	inode->i_fop = &def_mdt_fops;
	inode->i_mapping->a_ops = &def_mdt_aops;

	return 0;
}

/**
 * nilfs_mdt_clear - do cleanup for the metadata file
 * @inode: inode of the metadata file
 */
void nilfs_mdt_clear(struct inode *inode)
{
	struct nilfs_mdt_info *mdi = NILFS_MDT(inode);

	if (mdi->mi_palloc_cache)
		nilfs_palloc_destroy_cache(inode);
}

/**
 * nilfs_mdt_destroy - release resources used by the metadata file
 * @inode: inode of the metadata file
 */
void nilfs_mdt_destroy(struct inode *inode)
{
	struct nilfs_mdt_info *mdi = NILFS_MDT(inode);

	kfree(mdi->mi_bgl); /* kfree(NULL) is safe */
	kfree(mdi);
}

void nilfs_mdt_set_entry_size(struct inode *inode, unsigned int entry_size,
			      unsigned int header_size)
{
	struct nilfs_mdt_info *mi = NILFS_MDT(inode);

	mi->mi_entry_size = entry_size;
	mi->mi_entries_per_block = (1 << inode->i_blkbits) / entry_size;
	mi->mi_first_entry_offset = DIV_ROUND_UP(header_size, entry_size);
}

/**
 * nilfs_mdt_setup_shadow_map - setup shadow map and bind it to metadata file
 * @inode: inode of the metadata file
 * @shadow: shadow mapping
 */
int nilfs_mdt_setup_shadow_map(struct inode *inode,
			       struct nilfs_shadow_map *shadow)
{
	struct nilfs_mdt_info *mi = NILFS_MDT(inode);

	INIT_LIST_HEAD(&shadow->frozen_buffers);
	address_space_init_once(&shadow->frozen_data);
	nilfs_mapping_init(&shadow->frozen_data, inode);
	address_space_init_once(&shadow->frozen_btnodes);
	nilfs_mapping_init(&shadow->frozen_btnodes, inode);
	mi->mi_shadow = shadow;
	return 0;
}

/**
 * nilfs_mdt_save_to_shadow_map - copy bmap and dirty pages to shadow map
 * @inode: inode of the metadata file
 */
int nilfs_mdt_save_to_shadow_map(struct inode *inode)
{
	struct nilfs_mdt_info *mi = NILFS_MDT(inode);
	struct nilfs_inode_info *ii = NILFS_I(inode);
	struct nilfs_shadow_map *shadow = mi->mi_shadow;
	int ret;

	ret = nilfs_copy_dirty_pages(&shadow->frozen_data, inode->i_mapping);
	if (ret)
		goto out;

	ret = nilfs_copy_dirty_pages(&shadow->frozen_btnodes,
				     &ii->i_btnode_cache);
	if (ret)
		goto out;

	nilfs_bmap_save(ii->i_bmap, &shadow->bmap_store);
 out:
	return ret;
}

int nilfs_mdt_freeze_buffer(struct inode *inode, struct buffer_head *bh)
{
	struct nilfs_shadow_map *shadow = NILFS_MDT(inode)->mi_shadow;
	struct buffer_head *bh_frozen;
	struct page *page;
	int blkbits = inode->i_blkbits;

	page = grab_cache_page(&shadow->frozen_data, bh->b_page->index);
	if (!page)
		return -ENOMEM;

	if (!page_has_buffers(page))
		create_empty_buffers(page, 1 << blkbits, 0);

	bh_frozen = nilfs_page_get_nth_block(page, bh_offset(bh) >> blkbits);

	if (!buffer_uptodate(bh_frozen))
		nilfs_copy_buffer(bh_frozen, bh);
	if (list_empty(&bh_frozen->b_assoc_buffers)) {
		list_add_tail(&bh_frozen->b_assoc_buffers,
			      &shadow->frozen_buffers);
		set_buffer_nilfs_redirected(bh);
	} else {
		brelse(bh_frozen); /* already frozen */
	}

	unlock_page(page);
	put_page(page);
	return 0;
}

struct buffer_head *
nilfs_mdt_get_frozen_buffer(struct inode *inode, struct buffer_head *bh)
{
	struct nilfs_shadow_map *shadow = NILFS_MDT(inode)->mi_shadow;
	struct buffer_head *bh_frozen = NULL;
	struct page *page;
	int n;

	page = find_lock_page(&shadow->frozen_data, bh->b_page->index);
	if (page) {
		if (page_has_buffers(page)) {
			n = bh_offset(bh) >> inode->i_blkbits;
			bh_frozen = nilfs_page_get_nth_block(page, n);
		}
		unlock_page(page);
		put_page(page);
	}
	return bh_frozen;
}

static void nilfs_release_frozen_buffers(struct nilfs_shadow_map *shadow)
{
	struct list_head *head = &shadow->frozen_buffers;
	struct buffer_head *bh;

	while (!list_empty(head)) {
		bh = list_first_entry(head, struct buffer_head,
				      b_assoc_buffers);
		list_del_init(&bh->b_assoc_buffers);
		brelse(bh); /* drop ref-count to make it releasable */
	}
}

/**
 * nilfs_mdt_restore_from_shadow_map - restore dirty pages and bmap state
 * @inode: inode of the metadata file
 */
void nilfs_mdt_restore_from_shadow_map(struct inode *inode)
{
	struct nilfs_mdt_info *mi = NILFS_MDT(inode);
	struct nilfs_inode_info *ii = NILFS_I(inode);
	struct nilfs_shadow_map *shadow = mi->mi_shadow;

	down_write(&mi->mi_sem);

	if (mi->mi_palloc_cache)
		nilfs_palloc_clear_cache(inode);

	nilfs_clear_dirty_pages(inode->i_mapping, true);
	nilfs_copy_back_pages(inode->i_mapping, &shadow->frozen_data);

	nilfs_clear_dirty_pages(&ii->i_btnode_cache, true);
	nilfs_copy_back_pages(&ii->i_btnode_cache, &shadow->frozen_btnodes);

	nilfs_bmap_restore(ii->i_bmap, &shadow->bmap_store);

	up_write(&mi->mi_sem);
}

/**
 * nilfs_mdt_clear_shadow_map - truncate pages in shadow map caches
 * @inode: inode of the metadata file
 */
void nilfs_mdt_clear_shadow_map(struct inode *inode)
{
	struct nilfs_mdt_info *mi = NILFS_MDT(inode);
	struct nilfs_shadow_map *shadow = mi->mi_shadow;

	down_write(&mi->mi_sem);
	nilfs_release_frozen_buffers(shadow);
	truncate_inode_pages(&shadow->frozen_data, 0);
	truncate_inode_pages(&shadow->frozen_btnodes, 0);
	up_write(&mi->mi_sem);
}