inode.c 135 KB
Newer Older
1
/*
2
 *  linux/fs/ext4/inode.c
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
 *
 * Copyright (C) 1992, 1993, 1994, 1995
 * Remy Card (card@masi.ibp.fr)
 * Laboratoire MASI - Institut Blaise Pascal
 * Universite Pierre et Marie Curie (Paris VI)
 *
 *  from
 *
 *  linux/fs/minix/inode.c
 *
 *  Copyright (C) 1991, 1992  Linus Torvalds
 *
 *  64-bit file support on 64-bit platforms by Jakub Jelinek
 *	(jj@sunsite.ms.mff.cuni.cz)
 *
18
 *  Assorted race fixes, rewrite of ext4_get_block() by Al Viro, 2000
19
20
21
22
23
 */

#include <linux/module.h>
#include <linux/fs.h>
#include <linux/time.h>
24
#include <linux/jbd2.h>
25
26
27
28
29
30
#include <linux/highuid.h>
#include <linux/pagemap.h>
#include <linux/quotaops.h>
#include <linux/string.h>
#include <linux/buffer_head.h>
#include <linux/writeback.h>
31
#include <linux/pagevec.h>
32
#include <linux/mpage.h>
33
#include <linux/namei.h>
34
35
#include <linux/uio.h>
#include <linux/bio.h>
36
#include <linux/workqueue.h>
37
#include <linux/kernel.h>
38
#include <linux/printk.h>
39
#include <linux/slab.h>
40
#include <linux/ratelimit.h>
41

42
#include "ext4_jbd2.h"
43
44
#include "xattr.h"
#include "acl.h"
45
#include "truncate.h"
46

47
48
#include <trace/events/ext4.h>

49
50
#define MPAGE_DA_EXTENT_TAIL 0x01

51
52
53
static inline int ext4_begin_ordered_truncate(struct inode *inode,
					      loff_t new_size)
{
54
	trace_ext4_begin_ordered_truncate(inode, new_size);
55
56
57
58
59
60
61
62
63
64
65
	/*
	 * If jinode is zero, then we never opened the file for
	 * writing, so there's no need to call
	 * jbd2_journal_begin_ordered_truncate() since there's no
	 * outstanding writes we need to flush.
	 */
	if (!EXT4_I(inode)->jinode)
		return 0;
	return jbd2_journal_begin_ordered_truncate(EXT4_JOURNAL(inode),
						   EXT4_I(inode)->jinode,
						   new_size);
66
67
}

68
static void ext4_invalidatepage(struct page *page, unsigned long offset);
69
70
71
72
73
74
static int noalloc_get_block_write(struct inode *inode, sector_t iblock,
				   struct buffer_head *bh_result, int create);
static int ext4_set_bh_endio(struct buffer_head *bh, struct inode *inode);
static void ext4_end_io_buffer_write(struct buffer_head *bh, int uptodate);
static int __ext4_journalled_writepage(struct page *page, unsigned int len);
static int ext4_bh_delay_or_unwritten(handle_t *handle, struct buffer_head *bh);
75

76
77
78
/*
 * Test whether an inode is a fast symlink.
 */
79
static int ext4_inode_is_fast_symlink(struct inode *inode)
80
{
81
	int ea_blocks = EXT4_I(inode)->i_file_acl ?
82
83
84
85
86
87
88
89
90
91
		(inode->i_sb->s_blocksize >> 9) : 0;

	return (S_ISLNK(inode->i_mode) && inode->i_blocks - ea_blocks == 0);
}

/*
 * Restart the transaction associated with *handle.  This does a commit,
 * so before we call here everything must be consistently dirtied against
 * this transaction.
 */
92
int ext4_truncate_restart_trans(handle_t *handle, struct inode *inode,
93
				 int nblocks)
94
{
95
96
97
	int ret;

	/*
98
	 * Drop i_data_sem to avoid deadlock with ext4_map_blocks.  At this
99
100
101
102
	 * moment, get_block can be called only for blocks inside i_size since
	 * page cache has been already dropped and writes are blocked by
	 * i_mutex. So we can safely drop the i_data_sem here.
	 */
103
	BUG_ON(EXT4_JOURNAL(inode) == NULL);
104
	jbd_debug(2, "restarting handle %p\n", handle);
105
	up_write(&EXT4_I(inode)->i_data_sem);
106
	ret = ext4_journal_restart(handle, nblocks);
107
	down_write(&EXT4_I(inode)->i_data_sem);
108
	ext4_discard_preallocations(inode);
109
110

	return ret;
111
112
113
114
115
}

/*
 * Called at the last iput() if i_nlink is zero.
 */
Al Viro's avatar
Al Viro committed
116
void ext4_evict_inode(struct inode *inode)
117
118
{
	handle_t *handle;
119
	int err;
120

121
	trace_ext4_evict_inode(inode);
122
123
124

	ext4_ioend_wait(inode);

Al Viro's avatar
Al Viro committed
125
	if (inode->i_nlink) {
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
		/*
		 * When journalling data dirty buffers are tracked only in the
		 * journal. So although mm thinks everything is clean and
		 * ready for reaping the inode might still have some pages to
		 * write in the running transaction or waiting to be
		 * checkpointed. Thus calling jbd2_journal_invalidatepage()
		 * (via truncate_inode_pages()) to discard these buffers can
		 * cause data loss. Also even if we did not discard these
		 * buffers, we would have no way to find them after the inode
		 * is reaped and thus user could see stale data if he tries to
		 * read them before the transaction is checkpointed. So be
		 * careful and force everything to disk here... We use
		 * ei->i_datasync_tid to store the newest transaction
		 * containing inode's data.
		 *
		 * Note that directories do not have this problem because they
		 * don't use page cache.
		 */
		if (ext4_should_journal_data(inode) &&
		    (S_ISLNK(inode->i_mode) || S_ISREG(inode->i_mode))) {
			journal_t *journal = EXT4_SB(inode->i_sb)->s_journal;
			tid_t commit_tid = EXT4_I(inode)->i_datasync_tid;

			jbd2_log_start_commit(journal, commit_tid);
			jbd2_log_wait_commit(journal, commit_tid);
			filemap_write_and_wait(&inode->i_data);
		}
Al Viro's avatar
Al Viro committed
153
154
155
156
		truncate_inode_pages(&inode->i_data, 0);
		goto no_delete;
	}

157
	if (!is_bad_inode(inode))
158
		dquot_initialize(inode);
159

160
161
	if (ext4_should_order_data(inode))
		ext4_begin_ordered_truncate(inode, 0);
162
163
164
165
166
	truncate_inode_pages(&inode->i_data, 0);

	if (is_bad_inode(inode))
		goto no_delete;

167
	handle = ext4_journal_start(inode, ext4_blocks_for_truncate(inode)+3);
168
	if (IS_ERR(handle)) {
169
		ext4_std_error(inode->i_sb, PTR_ERR(handle));
170
171
172
173
174
		/*
		 * If we're going to skip the normal cleanup, we still need to
		 * make sure that the in-core orphan linked list is properly
		 * cleaned up.
		 */
175
		ext4_orphan_del(NULL, inode);
176
177
178
179
		goto no_delete;
	}

	if (IS_SYNC(inode))
180
		ext4_handle_sync(handle);
181
	inode->i_size = 0;
182
183
	err = ext4_mark_inode_dirty(handle, inode);
	if (err) {
184
		ext4_warning(inode->i_sb,
185
186
187
			     "couldn't mark inode dirty (err %d)", err);
		goto stop_handle;
	}
188
	if (inode->i_blocks)
189
		ext4_truncate(inode);
190
191
192
193
194
195
196

	/*
	 * ext4_ext_truncate() doesn't reserve any slop when it
	 * restarts journal transactions; therefore there may not be
	 * enough credits left in the handle to remove the inode from
	 * the orphan list and set the dtime field.
	 */
197
	if (!ext4_handle_has_enough_credits(handle, 3)) {
198
199
200
201
		err = ext4_journal_extend(handle, 3);
		if (err > 0)
			err = ext4_journal_restart(handle, 3);
		if (err != 0) {
202
			ext4_warning(inode->i_sb,
203
204
205
				     "couldn't extend journal (err %d)", err);
		stop_handle:
			ext4_journal_stop(handle);
206
			ext4_orphan_del(NULL, inode);
207
208
209
210
			goto no_delete;
		}
	}

211
	/*
212
	 * Kill off the orphan record which ext4_truncate created.
213
	 * AKPM: I think this can be inside the above `if'.
214
	 * Note that ext4_orphan_del() has to be able to cope with the
215
	 * deletion of a non-existent orphan - this is because we don't
216
	 * know if ext4_truncate() actually created an orphan record.
217
218
	 * (Well, we could do this if we need to, but heck - it works)
	 */
219
220
	ext4_orphan_del(handle, inode);
	EXT4_I(inode)->i_dtime	= get_seconds();
221
222
223
224
225
226
227
228

	/*
	 * One subtle ordering requirement: if anything has gone wrong
	 * (transaction abort, IO errors, whatever), then we can still
	 * do these next steps (the fs will already have been marked as
	 * having errors), but we can't free the inode if the mark_dirty
	 * fails.
	 */
229
	if (ext4_mark_inode_dirty(handle, inode))
230
		/* If that failed, just do the required in-core inode clear. */
Al Viro's avatar
Al Viro committed
231
		ext4_clear_inode(inode);
232
	else
233
234
		ext4_free_inode(handle, inode);
	ext4_journal_stop(handle);
235
236
	return;
no_delete:
Al Viro's avatar
Al Viro committed
237
	ext4_clear_inode(inode);	/* We must guarantee clearing of inode... */
238
239
}

240
241
#ifdef CONFIG_QUOTA
qsize_t *ext4_get_reserved_space(struct inode *inode)
242
{
243
	return &EXT4_I(inode)->i_reserved_quota;
244
}
245
#endif
246

247
248
/*
 * Calculate the number of metadata blocks need to reserve
249
 * to allocate a block located at @lblock
250
 */
251
static int ext4_calc_metadata_amount(struct inode *inode, ext4_lblk_t lblock)
252
{
253
	if (ext4_test_inode_flag(inode, EXT4_INODE_EXTENTS))
254
		return ext4_ext_calc_metadata_amount(inode, lblock);
255

256
	return ext4_ind_calc_metadata_amount(inode, lblock);
257
258
}

259
260
261
262
/*
 * Called with i_data_sem down, which is important since we can call
 * ext4_discard_preallocations() from here.
 */
263
264
void ext4_da_update_reserve_space(struct inode *inode,
					int used, int quota_claim)
265
266
{
	struct ext4_sb_info *sbi = EXT4_SB(inode->i_sb);
267
268
269
	struct ext4_inode_info *ei = EXT4_I(inode);

	spin_lock(&ei->i_block_reservation_lock);
270
	trace_ext4_da_update_reserve_space(inode, used, quota_claim);
271
272
273
274
275
276
277
278
	if (unlikely(used > ei->i_reserved_data_blocks)) {
		ext4_msg(inode->i_sb, KERN_NOTICE, "%s: ino %lu, used %d "
			 "with only %d reserved data blocks\n",
			 __func__, inode->i_ino, used,
			 ei->i_reserved_data_blocks);
		WARN_ON(1);
		used = ei->i_reserved_data_blocks;
	}
279

280
281
282
	/* Update per-inode reservations */
	ei->i_reserved_data_blocks -= used;
	ei->i_reserved_meta_blocks -= ei->i_allocated_meta_blocks;
283
	percpu_counter_sub(&sbi->s_dirtyclusters_counter,
284
			   used + ei->i_allocated_meta_blocks);
285
	ei->i_allocated_meta_blocks = 0;
286

287
288
289
290
291
292
	if (ei->i_reserved_data_blocks == 0) {
		/*
		 * We can release all of the reserved metadata blocks
		 * only when we have written all of the delayed
		 * allocation blocks.
		 */
293
		percpu_counter_sub(&sbi->s_dirtyclusters_counter,
294
				   ei->i_reserved_meta_blocks);
295
		ei->i_reserved_meta_blocks = 0;
296
		ei->i_da_metadata_calc_len = 0;
297
	}
298
	spin_unlock(&EXT4_I(inode)->i_block_reservation_lock);
299

300
301
	/* Update quota subsystem for data blocks */
	if (quota_claim)
302
		dquot_claim_block(inode, EXT4_C2B(sbi, used));
303
	else {
304
305
306
		/*
		 * We did fallocate with an offset that is already delayed
		 * allocated. So on delayed allocated writeback we should
307
		 * not re-claim the quota for fallocated blocks.
308
		 */
309
		dquot_release_reservation_block(inode, EXT4_C2B(sbi, used));
310
	}
311
312
313
314
315
316

	/*
	 * If we have done all the pending block allocations and if
	 * there aren't any writers on the inode, we can discard the
	 * inode's preallocations.
	 */
317
318
	if ((ei->i_reserved_data_blocks == 0) &&
	    (atomic_read(&inode->i_writecount) == 0))
319
		ext4_discard_preallocations(inode);
320
321
}

322
static int __check_block_validity(struct inode *inode, const char *func,
323
324
				unsigned int line,
				struct ext4_map_blocks *map)
325
{
326
327
	if (!ext4_data_block_valid(EXT4_SB(inode->i_sb), map->m_pblk,
				   map->m_len)) {
328
329
330
331
		ext4_error_inode(inode, func, line, map->m_pblk,
				 "lblock %lu mapped to illegal pblock "
				 "(length %d)", (unsigned long) map->m_lblk,
				 map->m_len);
332
333
334
335
336
		return -EIO;
	}
	return 0;
}

337
#define check_block_validity(inode, map)	\
338
	__check_block_validity((inode), __func__, __LINE__, (map))
339

340
/*
341
342
 * Return the number of contiguous dirty pages in a given inode
 * starting at page frame idx.
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
 */
static pgoff_t ext4_num_dirty_pages(struct inode *inode, pgoff_t idx,
				    unsigned int max_pages)
{
	struct address_space *mapping = inode->i_mapping;
	pgoff_t	index;
	struct pagevec pvec;
	pgoff_t num = 0;
	int i, nr_pages, done = 0;

	if (max_pages == 0)
		return 0;
	pagevec_init(&pvec, 0);
	while (!done) {
		index = idx;
		nr_pages = pagevec_lookup_tag(&pvec, mapping, &index,
					      PAGECACHE_TAG_DIRTY,
					      (pgoff_t)PAGEVEC_SIZE);
		if (nr_pages == 0)
			break;
		for (i = 0; i < nr_pages; i++) {
			struct page *page = pvec.pages[i];
			struct buffer_head *bh, *head;

			lock_page(page);
			if (unlikely(page->mapping != mapping) ||
			    !PageDirty(page) ||
			    PageWriteback(page) ||
			    page->index != idx) {
				done = 1;
				unlock_page(page);
				break;
			}
376
377
378
379
380
381
382
383
384
			if (page_has_buffers(page)) {
				bh = head = page_buffers(page);
				do {
					if (!buffer_delay(bh) &&
					    !buffer_unwritten(bh))
						done = 1;
					bh = bh->b_this_page;
				} while (!done && (bh != head));
			}
385
386
387
388
389
			unlock_page(page);
			if (done)
				break;
			idx++;
			num++;
390
391
			if (num >= max_pages) {
				done = 1;
392
				break;
393
			}
394
395
396
397
398
399
		}
		pagevec_release(&pvec);
	}
	return num;
}

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
/*
 * Sets the BH_Da_Mapped bit on the buffer heads corresponding to the given map.
 */
static void set_buffers_da_mapped(struct inode *inode,
				   struct ext4_map_blocks *map)
{
	struct address_space *mapping = inode->i_mapping;
	struct pagevec pvec;
	int i, nr_pages;
	pgoff_t index, end;

	index = map->m_lblk >> (PAGE_CACHE_SHIFT - inode->i_blkbits);
	end = (map->m_lblk + map->m_len - 1) >>
		(PAGE_CACHE_SHIFT - inode->i_blkbits);

	pagevec_init(&pvec, 0);
	while (index <= end) {
		nr_pages = pagevec_lookup(&pvec, mapping, index,
					  min(end - index + 1,
					      (pgoff_t)PAGEVEC_SIZE));
		if (nr_pages == 0)
			break;
		for (i = 0; i < nr_pages; i++) {
			struct page *page = pvec.pages[i];
			struct buffer_head *bh, *head;

			if (unlikely(page->mapping != mapping) ||
			    !PageDirty(page))
				break;

			if (page_has_buffers(page)) {
				bh = head = page_buffers(page);
				do {
					set_buffer_da_mapped(bh);
					bh = bh->b_this_page;
				} while (bh != head);
			}
			index++;
		}
		pagevec_release(&pvec);
	}
}

443
/*
444
 * The ext4_map_blocks() function tries to look up the requested blocks,
445
 * and returns if the blocks are already mapped.
446
447
448
449
450
 *
 * Otherwise it takes the write lock of the i_data_sem and allocate blocks
 * and store the allocated blocks in the result buffer head and mark it
 * mapped.
 *
451
452
 * If file type is extents based, it will call ext4_ext_map_blocks(),
 * Otherwise, call with ext4_ind_map_blocks() to handle indirect mapping
453
454
455
456
457
458
459
460
 * based files
 *
 * On success, it returns the number of blocks being mapped or allocate.
 * if create==0 and the blocks are pre-allocated and uninitialized block,
 * the result buffer head is unmapped. If the create ==1, it will make sure
 * the buffer head is mapped.
 *
 * It returns 0 if plain look up failed (blocks have not been allocated), in
461
 * that case, buffer head is unmapped
462
463
464
 *
 * It returns the error in case of allocation failure.
 */
465
466
int ext4_map_blocks(handle_t *handle, struct inode *inode,
		    struct ext4_map_blocks *map, int flags)
467
468
{
	int retval;
469

470
471
472
473
	map->m_flags = 0;
	ext_debug("ext4_map_blocks(): inode %lu, flag %d, max_blocks %u,"
		  "logical block %lu\n", inode->i_ino, flags, map->m_len,
		  (unsigned long) map->m_lblk);
474
	/*
475
476
	 * Try to see if we can get the block without requesting a new
	 * file system block.
477
478
	 */
	down_read((&EXT4_I(inode)->i_data_sem));
479
	if (ext4_test_inode_flag(inode, EXT4_INODE_EXTENTS)) {
480
481
		retval = ext4_ext_map_blocks(handle, inode, map, flags &
					     EXT4_GET_BLOCKS_KEEP_SIZE);
482
	} else {
483
484
		retval = ext4_ind_map_blocks(handle, inode, map, flags &
					     EXT4_GET_BLOCKS_KEEP_SIZE);
485
	}
486
	up_read((&EXT4_I(inode)->i_data_sem));
487

488
	if (retval > 0 && map->m_flags & EXT4_MAP_MAPPED) {
489
		int ret = check_block_validity(inode, map);
490
491
492
493
		if (ret != 0)
			return ret;
	}

494
	/* If it is only a block(s) look up */
495
	if ((flags & EXT4_GET_BLOCKS_CREATE) == 0)
496
497
498
499
500
501
		return retval;

	/*
	 * Returns if the blocks have already allocated
	 *
	 * Note that if blocks have been preallocated
502
	 * ext4_ext_get_block() returns the create = 0
503
504
	 * with buffer head unmapped.
	 */
505
	if (retval > 0 && map->m_flags & EXT4_MAP_MAPPED)
506
507
		return retval;

508
509
510
511
512
513
514
515
516
517
	/*
	 * When we call get_blocks without the create flag, the
	 * BH_Unwritten flag could have gotten set if the blocks
	 * requested were part of a uninitialized extent.  We need to
	 * clear this flag now that we are committed to convert all or
	 * part of the uninitialized extent to be an initialized
	 * extent.  This is because we need to avoid the combination
	 * of BH_Unwritten and BH_Mapped flags being simultaneously
	 * set on the buffer_head.
	 */
518
	map->m_flags &= ~EXT4_MAP_UNWRITTEN;
519

520
	/*
521
522
523
524
	 * New blocks allocate and/or writing to uninitialized extent
	 * will possibly result in updating i_data, so we take
	 * the write lock of i_data_sem, and call get_blocks()
	 * with create == 1 flag.
525
526
	 */
	down_write((&EXT4_I(inode)->i_data_sem));
527
528
529
530
531
532
533

	/*
	 * if the caller is from delayed allocation writeout path
	 * we have already reserved fs blocks for allocation
	 * let the underlying get_block() function know to
	 * avoid double accounting
	 */
534
	if (flags & EXT4_GET_BLOCKS_DELALLOC_RESERVE)
535
		ext4_set_inode_state(inode, EXT4_STATE_DELALLOC_RESERVED);
536
537
538
539
	/*
	 * We need to check for EXT4 here because migrate
	 * could have changed the inode type in between
	 */
540
	if (ext4_test_inode_flag(inode, EXT4_INODE_EXTENTS)) {
541
		retval = ext4_ext_map_blocks(handle, inode, map, flags);
542
	} else {
543
		retval = ext4_ind_map_blocks(handle, inode, map, flags);
544

545
		if (retval > 0 && map->m_flags & EXT4_MAP_NEW) {
546
547
548
549
550
			/*
			 * We allocated new blocks which will result in
			 * i_data's format changing.  Force the migrate
			 * to fail by clearing migrate flags
			 */
551
			ext4_clear_inode_state(inode, EXT4_STATE_EXT_MIGRATE);
552
		}
553

554
555
556
557
558
559
560
		/*
		 * Update reserved blocks/metadata blocks after successful
		 * block allocation which had been deferred till now. We don't
		 * support fallocate for non extent files. So we can update
		 * reserve space here.
		 */
		if ((retval > 0) &&
561
			(flags & EXT4_GET_BLOCKS_DELALLOC_RESERVE))
562
563
			ext4_da_update_reserve_space(inode, retval, 1);
	}
564
	if (flags & EXT4_GET_BLOCKS_DELALLOC_RESERVE) {
565
		ext4_clear_inode_state(inode, EXT4_STATE_DELALLOC_RESERVED);
566

567
568
569
570
571
572
573
574
		/* If we have successfully mapped the delayed allocated blocks,
		 * set the BH_Da_Mapped bit on them. Its important to do this
		 * under the protection of i_data_sem.
		 */
		if (retval > 0 && map->m_flags & EXT4_MAP_MAPPED)
			set_buffers_da_mapped(inode, map);
	}

575
	up_write((&EXT4_I(inode)->i_data_sem));
576
	if (retval > 0 && map->m_flags & EXT4_MAP_MAPPED) {
577
		int ret = check_block_validity(inode, map);
578
579
580
		if (ret != 0)
			return ret;
	}
581
582
583
	return retval;
}

584
585
586
/* Maximum number of blocks we map for direct IO at once. */
#define DIO_MAX_BLOCKS 4096

587
588
static int _ext4_get_block(struct inode *inode, sector_t iblock,
			   struct buffer_head *bh, int flags)
589
{
590
	handle_t *handle = ext4_journal_current_handle();
591
	struct ext4_map_blocks map;
Jan Kara's avatar
Jan Kara committed
592
	int ret = 0, started = 0;
593
	int dio_credits;
594

595
596
597
598
	map.m_lblk = iblock;
	map.m_len = bh->b_size >> inode->i_blkbits;

	if (flags && !handle) {
Jan Kara's avatar
Jan Kara committed
599
		/* Direct IO write... */
600
601
602
		if (map.m_len > DIO_MAX_BLOCKS)
			map.m_len = DIO_MAX_BLOCKS;
		dio_credits = ext4_chunk_trans_blocks(inode, map.m_len);
603
		handle = ext4_journal_start(inode, dio_credits);
Jan Kara's avatar
Jan Kara committed
604
		if (IS_ERR(handle)) {
605
			ret = PTR_ERR(handle);
606
			return ret;
607
		}
Jan Kara's avatar
Jan Kara committed
608
		started = 1;
609
610
	}

611
	ret = ext4_map_blocks(handle, inode, &map, flags);
Jan Kara's avatar
Jan Kara committed
612
	if (ret > 0) {
613
614
615
		map_bh(bh, inode->i_sb, map.m_pblk);
		bh->b_state = (bh->b_state & ~EXT4_MAP_FLAGS) | map.m_flags;
		bh->b_size = inode->i_sb->s_blocksize * map.m_len;
Jan Kara's avatar
Jan Kara committed
616
		ret = 0;
617
	}
Jan Kara's avatar
Jan Kara committed
618
619
	if (started)
		ext4_journal_stop(handle);
620
621
622
	return ret;
}

623
624
625
626
627
628
629
int ext4_get_block(struct inode *inode, sector_t iblock,
		   struct buffer_head *bh, int create)
{
	return _ext4_get_block(inode, iblock, bh,
			       create ? EXT4_GET_BLOCKS_CREATE : 0);
}

630
631
632
/*
 * `handle' can be NULL if create is zero
 */
633
struct buffer_head *ext4_getblk(handle_t *handle, struct inode *inode,
Aneesh Kumar K.V's avatar
Aneesh Kumar K.V committed
634
				ext4_lblk_t block, int create, int *errp)
635
{
636
637
	struct ext4_map_blocks map;
	struct buffer_head *bh;
638
639
640
641
	int fatal = 0, err;

	J_ASSERT(handle != NULL || create == 0);

642
643
644
645
	map.m_lblk = block;
	map.m_len = 1;
	err = ext4_map_blocks(handle, inode, &map,
			      create ? EXT4_GET_BLOCKS_CREATE : 0);
646

647
648
649
650
651
652
653
654
655
656
	if (err < 0)
		*errp = err;
	if (err <= 0)
		return NULL;
	*errp = 0;

	bh = sb_getblk(inode->i_sb, map.m_pblk);
	if (!bh) {
		*errp = -EIO;
		return NULL;
657
	}
658
659
660
	if (map.m_flags & EXT4_MAP_NEW) {
		J_ASSERT(create != 0);
		J_ASSERT(handle != NULL);
661

662
663
664
665
666
667
668
669
670
671
672
673
674
		/*
		 * Now that we do not always journal data, we should
		 * keep in mind whether this should always journal the
		 * new buffer as metadata.  For now, regular file
		 * writes use ext4_get_block instead, so it's not a
		 * problem.
		 */
		lock_buffer(bh);
		BUFFER_TRACE(bh, "call get_create_access");
		fatal = ext4_journal_get_create_access(handle, bh);
		if (!fatal && !buffer_uptodate(bh)) {
			memset(bh->b_data, 0, inode->i_sb->s_blocksize);
			set_buffer_uptodate(bh);
675
		}
676
677
678
679
680
681
682
		unlock_buffer(bh);
		BUFFER_TRACE(bh, "call ext4_handle_dirty_metadata");
		err = ext4_handle_dirty_metadata(handle, inode, bh);
		if (!fatal)
			fatal = err;
	} else {
		BUFFER_TRACE(bh, "not a new buffer");
683
	}
684
685
686
687
688
689
	if (fatal) {
		*errp = fatal;
		brelse(bh);
		bh = NULL;
	}
	return bh;
690
691
}

692
struct buffer_head *ext4_bread(handle_t *handle, struct inode *inode,
Aneesh Kumar K.V's avatar
Aneesh Kumar K.V committed
693
			       ext4_lblk_t block, int create, int *err)
694
{
695
	struct buffer_head *bh;
696

697
	bh = ext4_getblk(handle, inode, block, create, err);
698
699
700
701
	if (!bh)
		return bh;
	if (buffer_uptodate(bh))
		return bh;
702
	ll_rw_block(READ | REQ_META | REQ_PRIO, 1, &bh);
703
704
705
706
707
708
709
710
	wait_on_buffer(bh);
	if (buffer_uptodate(bh))
		return bh;
	put_bh(bh);
	*err = -EIO;
	return NULL;
}

711
712
713
714
715
716
717
static int walk_page_buffers(handle_t *handle,
			     struct buffer_head *head,
			     unsigned from,
			     unsigned to,
			     int *partial,
			     int (*fn)(handle_t *handle,
				       struct buffer_head *bh))
718
719
720
721
722
723
724
{
	struct buffer_head *bh;
	unsigned block_start, block_end;
	unsigned blocksize = head->b_size;
	int err, ret = 0;
	struct buffer_head *next;

725
726
	for (bh = head, block_start = 0;
	     ret == 0 && (bh != head || !block_start);
727
	     block_start = block_end, bh = next) {
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
		next = bh->b_this_page;
		block_end = block_start + blocksize;
		if (block_end <= from || block_start >= to) {
			if (partial && !buffer_uptodate(bh))
				*partial = 1;
			continue;
		}
		err = (*fn)(handle, bh);
		if (!ret)
			ret = err;
	}
	return ret;
}

/*
 * To preserve ordering, it is essential that the hole instantiation and
 * the data write be encapsulated in a single transaction.  We cannot
745
 * close off a transaction and start a new one between the ext4_get_block()
746
 * and the commit_write().  So doing the jbd2_journal_start at the start of
747
748
 * prepare_write() is the right place.
 *
749
750
 * Also, this function can nest inside ext4_writepage() ->
 * block_write_full_page(). In that case, we *know* that ext4_writepage()
751
752
753
754
 * has generated enough buffer credits to do the whole page.  So we won't
 * block on the journal in that case, which is good, because the caller may
 * be PF_MEMALLOC.
 *
755
 * By accident, ext4 can be reentered when a transaction is open via
756
757
758
759
760
761
 * quota file writes.  If we were to commit the transaction while thus
 * reentered, there can be a deadlock - we would be holding a quota
 * lock, and the commit would never complete if another thread had a
 * transaction open and was blocking on the quota lock - a ranking
 * violation.
 *
762
 * So what we do is to rely on the fact that jbd2_journal_stop/journal_start
763
764
765
766
767
 * will _not_ run commit under these circumstances because handle->h_ref
 * is elevated.  We'll still have enough credits for the tiny quotafile
 * write.
 */
static int do_journal_get_write_access(handle_t *handle,
768
				       struct buffer_head *bh)
769
{
770
771
772
	int dirty = buffer_dirty(bh);
	int ret;

773
774
	if (!buffer_mapped(bh) || buffer_freed(bh))
		return 0;
775
	/*
776
	 * __block_write_begin() could have dirtied some buffers. Clean
777
778
	 * the dirty bit as jbd2_journal_get_write_access() could complain
	 * otherwise about fs integrity issues. Setting of the dirty bit
779
	 * by __block_write_begin() isn't a real problem here as we clear
780
781
782
783
784
785
786
787
788
	 * the bit before releasing a page lock and thus writeback cannot
	 * ever write the buffer.
	 */
	if (dirty)
		clear_buffer_dirty(bh);
	ret = ext4_journal_get_write_access(handle, bh);
	if (!ret && dirty)
		ret = ext4_handle_dirty_metadata(handle, NULL, bh);
	return ret;
789
790
}

791
792
static int ext4_get_block_write(struct inode *inode, sector_t iblock,
		   struct buffer_head *bh_result, int create);
Nick Piggin's avatar
Nick Piggin committed
793
static int ext4_write_begin(struct file *file, struct address_space *mapping,
794
795
			    loff_t pos, unsigned len, unsigned flags,
			    struct page **pagep, void **fsdata)
796
{
797
	struct inode *inode = mapping->host;
798
	int ret, needed_blocks;
799
800
	handle_t *handle;
	int retries = 0;
801
	struct page *page;
802
	pgoff_t index;
803
	unsigned from, to;
Nick Piggin's avatar
Nick Piggin committed
804

805
	trace_ext4_write_begin(inode, pos, len, flags);
806
807
808
809
810
	/*
	 * Reserve one block more for addition to orphan list in case
	 * we allocate blocks but write fails for some reason
	 */
	needed_blocks = ext4_writepage_trans_blocks(inode) + 1;
811
	index = pos >> PAGE_CACHE_SHIFT;
812
813
	from = pos & (PAGE_CACHE_SIZE - 1);
	to = from + len;
814
815

retry:
816
817
818
819
	handle = ext4_journal_start(inode, needed_blocks);
	if (IS_ERR(handle)) {
		ret = PTR_ERR(handle);
		goto out;
820
	}
821

822
823
824
825
	/* We cannot recurse into the filesystem as the transaction is already
	 * started */
	flags |= AOP_FLAG_NOFS;

826
	page = grab_cache_page_write_begin(mapping, index, flags);
827
828
829
830
831
832
833
	if (!page) {
		ext4_journal_stop(handle);
		ret = -ENOMEM;
		goto out;
	}
	*pagep = page;

834
	if (ext4_should_dioread_nolock(inode))
835
		ret = __block_write_begin(page, pos, len, ext4_get_block_write);
836
	else
837
		ret = __block_write_begin(page, pos, len, ext4_get_block);
Nick Piggin's avatar
Nick Piggin committed
838
839

	if (!ret && ext4_should_journal_data(inode)) {
840
841
842
		ret = walk_page_buffers(handle, page_buffers(page),
				from, to, NULL, do_journal_get_write_access);
	}
Nick Piggin's avatar
Nick Piggin committed
843
844

	if (ret) {
845
846
		unlock_page(page);
		page_cache_release(page);
847
		/*
848
		 * __block_write_begin may have instantiated a few blocks
849
850
		 * outside i_size.  Trim these off again. Don't need
		 * i_size_read because we hold i_mutex.
851
852
853
		 *
		 * Add inode to orphan list in case we crash before
		 * truncate finishes
854
		 */
855
		if (pos + len > inode->i_size && ext4_can_truncate(inode))
856
857
858
859
			ext4_orphan_add(handle, inode);

		ext4_journal_stop(handle);
		if (pos + len > inode->i_size) {
860
			ext4_truncate_failed_write(inode);
861
			/*
862
			 * If truncate failed early the inode might
863
864
865
866
867
868
869
			 * still be on the orphan list; we need to
			 * make sure the inode is removed from the
			 * orphan list in that case.
			 */
			if (inode->i_nlink)
				ext4_orphan_del(NULL, inode);
		}
Nick Piggin's avatar
Nick Piggin committed
870
871
	}

872
	if (ret == -ENOSPC && ext4_should_retry_alloc(inode->i_sb, &retries))
873
		goto retry;
874
out:
875
876
877
	return ret;
}

Nick Piggin's avatar
Nick Piggin committed
878
879
/* For write_end() in data=journal mode */
static int write_end_fn(handle_t *handle, struct buffer_head *bh)
880
881
882
883
{
	if (!buffer_mapped(bh) || buffer_freed(bh))
		return 0;
	set_buffer_uptodate(bh);
884
	return ext4_handle_dirty_metadata(handle, NULL, bh);
885
886
}

887
static int ext4_generic_write_end(struct file *file,
888
889
890
				  struct address_space *mapping,
				  loff_t pos, unsigned len, unsigned copied,
				  struct page *page, void *fsdata)
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
{
	int i_size_changed = 0;
	struct inode *inode = mapping->host;
	handle_t *handle = ext4_journal_current_handle();

	copied = block_write_end(file, mapping, pos, len, copied, page, fsdata);

	/*
	 * No need to use i_size_read() here, the i_size
	 * cannot change under us because we hold i_mutex.
	 *
	 * But it's important to update i_size while still holding page lock:
	 * page writeout could otherwise come in and zero beyond i_size.
	 */
	if (pos + copied > inode->i_size) {
		i_size_write(inode, pos + copied);
		i_size_changed = 1;
	}

	if (pos + copied >  EXT4_I(inode)->i_disksize) {
		/* We need to mark inode dirty even if
		 * new_i_size is less that inode->i_size
		 * bu greater than i_disksize.(hint delalloc)
		 */
		ext4_update_i_disksize(inode, (pos + copied));
		i_size_changed = 1;
	}
	unlock_page(page);
	page_cache_release(page);

	/*
	 * Don't mark the inode dirty under page lock. First, it unnecessarily
	 * makes the holding time of page lock longer. Second, it forces lock
	 * ordering of page lock and transaction start for journaling
	 * filesystems.
	 */
	if (i_size_changed)
		ext4_mark_inode_dirty(handle, inode);

	return copied;
}

933
934
935
936
/*
 * We need to pick up the new inode size which generic_commit_write gave us
 * `file' can be NULL - eg, when called from page_symlink().
 *
937
 * ext4 never places buffers on inode->i_mapping->private_list.  metadata
938
939
 * buffers are managed internally.
 */
Nick Piggin's avatar
Nick Piggin committed
940
static int ext4_ordered_write_end(struct file *file,
941
942
943
				  struct address_space *mapping,
				  loff_t pos, unsigned len, unsigned copied,
				  struct page *page, void *fsdata)
944
{
945
	handle_t *handle = ext4_journal_current_handle();
946
	struct inode *inode = mapping->host;
947
948
	int ret = 0, ret2;

949
	trace_ext4_ordered_write_end(inode, pos, len, copied);
950
	ret = ext4_jbd2_file_inode(handle, inode);
951
952

	if (ret == 0) {
953
		ret2 = ext4_generic_write_end(file, mapping, pos, len, copied,
Nick Piggin's avatar
Nick Piggin committed
954
							page, fsdata);
955
		copied = ret2;
956
		if (pos + len > inode->i_size && ext4_can_truncate(inode))
957
958
959
960
961
			/* if we have allocated more blocks and copied
			 * less. We will have blocks allocated outside
			 * inode->i_size. So truncate them
			 */
			ext4_orphan_add(handle, inode);
962
963
		if (ret2 < 0)
			ret = ret2;
964
965
966
	} else {
		unlock_page(page);
		page_cache_release(page);
967
	}
968

969
	ret2 = ext4_journal_stop(handle);
970
971
	if (!ret)
		ret = ret2;
Nick Piggin's avatar
Nick Piggin committed
972

973
	if (pos + len > inode->i_size) {
974
		ext4_truncate_failed_write(inode);
975
		/*
976
		 * If truncate failed early the inode might still be
977
978
979
980
981
982
983
984
		 * on the orphan list; we need to make sure the inode
		 * is removed from the orphan list in that case.
		 */
		if (inode->i_nlink)
			ext4_orphan_del(NULL, inode);
	}


Nick Piggin's avatar
Nick Piggin committed
985
	return ret ? ret : copied;
986
987
}

Nick Piggin's avatar
Nick Piggin committed
988
static int ext4_writeback_write_end(struct file *file,
989
990
991
				    struct address_space *mapping,
				    loff_t pos, unsigned len, unsigned copied,
				    struct page *page, void *fsdata)
992
{
993
	handle_t *handle = ext4_journal_current_handle();
994
	struct inode *inode = mapping->host;
995
996
	int ret = 0, ret2;

997
	trace_ext4_writeback_write_end(inode, pos, len, copied);
998
	ret2 = ext4_generic_write_end(file, mapping, pos, len, copied,
Nick Piggin's avatar
Nick Piggin committed
999
							page, fsdata);
1000
	copied = ret2;
For faster browsing, not all history is shown. View entire blame