inode.c 128 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 "ext4_extents.h"
46
#include "truncate.h"
47

48
49
#include <trace/events/ext4.h>

50
51
#define MPAGE_DA_EXTENT_TAIL 0x01

52
53
54
static inline int ext4_begin_ordered_truncate(struct inode *inode,
					      loff_t new_size)
{
55
	trace_ext4_begin_ordered_truncate(inode, new_size);
56
57
58
59
60
61
62
63
64
65
66
	/*
	 * 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);
67
68
}

69
static void ext4_invalidatepage(struct page *page, unsigned long offset);
70
71
72
73
74
75
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);
76

77
78
79
/*
 * Test whether an inode is a fast symlink.
 */
80
static int ext4_inode_is_fast_symlink(struct inode *inode)
81
{
82
	int ea_blocks = EXT4_I(inode)->i_file_acl ?
83
84
85
86
87
88
89
90
91
92
		(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.
 */
93
int ext4_truncate_restart_trans(handle_t *handle, struct inode *inode,
94
				 int nblocks)
95
{
96
97
98
	int ret;

	/*
99
	 * Drop i_data_sem to avoid deadlock with ext4_map_blocks.  At this
100
101
102
103
	 * 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.
	 */
104
	BUG_ON(EXT4_JOURNAL(inode) == NULL);
105
	jbd_debug(2, "restarting handle %p\n", handle);
106
	up_write(&EXT4_I(inode)->i_data_sem);
107
	ret = ext4_journal_restart(handle, nblocks);
108
	down_write(&EXT4_I(inode)->i_data_sem);
109
	ext4_discard_preallocations(inode);
110
111

	return ret;
112
113
114
115
116
}

/*
 * Called at the last iput() if i_nlink is zero.
 */
Al Viro's avatar
Al Viro committed
117
void ext4_evict_inode(struct inode *inode)
118
119
{
	handle_t *handle;
120
	int err;
121

122
	trace_ext4_evict_inode(inode);
123
124
125
126
127
128

	mutex_lock(&inode->i_mutex);
	ext4_flush_completed_IO(inode);
	mutex_unlock(&inode->i_mutex);
	ext4_ioend_wait(inode);

Al Viro's avatar
Al Viro committed
129
	if (inode->i_nlink) {
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
		/*
		 * 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
157
158
159
160
		truncate_inode_pages(&inode->i_data, 0);
		goto no_delete;
	}

161
	if (!is_bad_inode(inode))
162
		dquot_initialize(inode);
163

164
165
	if (ext4_should_order_data(inode))
		ext4_begin_ordered_truncate(inode, 0);
166
167
168
169
170
	truncate_inode_pages(&inode->i_data, 0);

	if (is_bad_inode(inode))
		goto no_delete;

171
	handle = ext4_journal_start(inode, ext4_blocks_for_truncate(inode)+3);
172
	if (IS_ERR(handle)) {
173
		ext4_std_error(inode->i_sb, PTR_ERR(handle));
174
175
176
177
178
		/*
		 * 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.
		 */
179
		ext4_orphan_del(NULL, inode);
180
181
182
183
		goto no_delete;
	}

	if (IS_SYNC(inode))
184
		ext4_handle_sync(handle);
185
	inode->i_size = 0;
186
187
	err = ext4_mark_inode_dirty(handle, inode);
	if (err) {
188
		ext4_warning(inode->i_sb,
189
190
191
			     "couldn't mark inode dirty (err %d)", err);
		goto stop_handle;
	}
192
	if (inode->i_blocks)
193
		ext4_truncate(inode);
194
195
196
197
198
199
200

	/*
	 * 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.
	 */
201
	if (!ext4_handle_has_enough_credits(handle, 3)) {
202
203
204
205
		err = ext4_journal_extend(handle, 3);
		if (err > 0)
			err = ext4_journal_restart(handle, 3);
		if (err != 0) {
206
			ext4_warning(inode->i_sb,
207
208
209
				     "couldn't extend journal (err %d)", err);
		stop_handle:
			ext4_journal_stop(handle);
210
			ext4_orphan_del(NULL, inode);
211
212
213
214
			goto no_delete;
		}
	}

215
	/*
216
	 * Kill off the orphan record which ext4_truncate created.
217
	 * AKPM: I think this can be inside the above `if'.
218
	 * Note that ext4_orphan_del() has to be able to cope with the
219
	 * deletion of a non-existent orphan - this is because we don't
220
	 * know if ext4_truncate() actually created an orphan record.
221
222
	 * (Well, we could do this if we need to, but heck - it works)
	 */
223
224
	ext4_orphan_del(handle, inode);
	EXT4_I(inode)->i_dtime	= get_seconds();
225
226
227
228
229
230
231
232

	/*
	 * 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.
	 */
233
	if (ext4_mark_inode_dirty(handle, inode))
234
		/* If that failed, just do the required in-core inode clear. */
Al Viro's avatar
Al Viro committed
235
		ext4_clear_inode(inode);
236
	else
237
238
		ext4_free_inode(handle, inode);
	ext4_journal_stop(handle);
239
240
	return;
no_delete:
Al Viro's avatar
Al Viro committed
241
	ext4_clear_inode(inode);	/* We must guarantee clearing of inode... */
242
243
}

244
245
#ifdef CONFIG_QUOTA
qsize_t *ext4_get_reserved_space(struct inode *inode)
246
{
247
	return &EXT4_I(inode)->i_reserved_quota;
248
}
249
#endif
250

251
252
/*
 * Calculate the number of metadata blocks need to reserve
253
 * to allocate a block located at @lblock
254
 */
255
static int ext4_calc_metadata_amount(struct inode *inode, ext4_lblk_t lblock)
256
{
257
	if (ext4_test_inode_flag(inode, EXT4_INODE_EXTENTS))
258
		return ext4_ext_calc_metadata_amount(inode, lblock);
259

260
	return ext4_ind_calc_metadata_amount(inode, lblock);
261
262
}

263
264
265
266
/*
 * Called with i_data_sem down, which is important since we can call
 * ext4_discard_preallocations() from here.
 */
267
268
void ext4_da_update_reserve_space(struct inode *inode,
					int used, int quota_claim)
269
270
{
	struct ext4_sb_info *sbi = EXT4_SB(inode->i_sb);
271
272
273
	struct ext4_inode_info *ei = EXT4_I(inode);

	spin_lock(&ei->i_block_reservation_lock);
274
	trace_ext4_da_update_reserve_space(inode, used);
275
276
277
278
279
280
281
282
	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;
	}
283

284
285
286
	/* Update per-inode reservations */
	ei->i_reserved_data_blocks -= used;
	ei->i_reserved_meta_blocks -= ei->i_allocated_meta_blocks;
287
288
	percpu_counter_sub(&sbi->s_dirtyblocks_counter,
			   used + ei->i_allocated_meta_blocks);
289
	ei->i_allocated_meta_blocks = 0;
290

291
292
293
294
295
296
	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.
		 */
297
298
		percpu_counter_sub(&sbi->s_dirtyblocks_counter,
				   ei->i_reserved_meta_blocks);
299
		ei->i_reserved_meta_blocks = 0;
300
		ei->i_da_metadata_calc_len = 0;
301
	}
302
	spin_unlock(&EXT4_I(inode)->i_block_reservation_lock);
303

304
305
	/* Update quota subsystem for data blocks */
	if (quota_claim)
306
		dquot_claim_block(inode, used);
307
	else {
308
309
310
		/*
		 * We did fallocate with an offset that is already delayed
		 * allocated. So on delayed allocated writeback we should
311
		 * not re-claim the quota for fallocated blocks.
312
		 */
313
		dquot_release_reservation_block(inode, used);
314
	}
315
316
317
318
319
320

	/*
	 * 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.
	 */
321
322
	if ((ei->i_reserved_data_blocks == 0) &&
	    (atomic_read(&inode->i_writecount) == 0))
323
		ext4_discard_preallocations(inode);
324
325
}

326
static int __check_block_validity(struct inode *inode, const char *func,
327
328
				unsigned int line,
				struct ext4_map_blocks *map)
329
{
330
331
	if (!ext4_data_block_valid(EXT4_SB(inode->i_sb), map->m_pblk,
				   map->m_len)) {
332
333
334
335
		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);
336
337
338
339
340
		return -EIO;
	}
	return 0;
}

341
#define check_block_validity(inode, map)	\
342
	__check_block_validity((inode), __func__, __LINE__, (map))
343

344
/*
345
346
 * Return the number of contiguous dirty pages in a given inode
 * starting at page frame idx.
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
 */
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;
			}
380
381
382
383
384
385
386
387
388
			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));
			}
389
390
391
392
393
			unlock_page(page);
			if (done)
				break;
			idx++;
			num++;
394
395
			if (num >= max_pages) {
				done = 1;
396
				break;
397
			}
398
399
400
401
402
403
		}
		pagevec_release(&pvec);
	}
	return num;
}

404
/*
405
 * The ext4_map_blocks() function tries to look up the requested blocks,
406
 * and returns if the blocks are already mapped.
407
408
409
410
411
 *
 * 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.
 *
412
413
 * If file type is extents based, it will call ext4_ext_map_blocks(),
 * Otherwise, call with ext4_ind_map_blocks() to handle indirect mapping
414
415
416
417
418
419
420
421
422
423
424
425
 * 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
 * that casem, buffer head is unmapped
 *
 * It returns the error in case of allocation failure.
 */
426
427
int ext4_map_blocks(handle_t *handle, struct inode *inode,
		    struct ext4_map_blocks *map, int flags)
428
429
{
	int retval;
430

431
432
433
434
	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);
435
	/*
436
437
	 * Try to see if we can get the block without requesting a new
	 * file system block.
438
439
	 */
	down_read((&EXT4_I(inode)->i_data_sem));
440
	if (ext4_test_inode_flag(inode, EXT4_INODE_EXTENTS)) {
441
		retval = ext4_ext_map_blocks(handle, inode, map, 0);
442
	} else {
443
		retval = ext4_ind_map_blocks(handle, inode, map, 0);
444
	}
445
	up_read((&EXT4_I(inode)->i_data_sem));
446

447
	if (retval > 0 && map->m_flags & EXT4_MAP_MAPPED) {
448
		int ret = check_block_validity(inode, map);
449
450
451
452
		if (ret != 0)
			return ret;
	}

453
	/* If it is only a block(s) look up */
454
	if ((flags & EXT4_GET_BLOCKS_CREATE) == 0)
455
456
457
458
459
460
461
462
463
		return retval;

	/*
	 * Returns if the blocks have already allocated
	 *
	 * Note that if blocks have been preallocated
	 * ext4_ext_get_block() returns th create = 0
	 * with buffer head unmapped.
	 */
464
	if (retval > 0 && map->m_flags & EXT4_MAP_MAPPED)
465
466
		return retval;

467
468
469
470
471
472
473
474
475
476
	/*
	 * 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.
	 */
477
	map->m_flags &= ~EXT4_MAP_UNWRITTEN;
478

479
	/*
480
481
482
483
	 * 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.
484
485
	 */
	down_write((&EXT4_I(inode)->i_data_sem));
486
487
488
489
490
491
492

	/*
	 * 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
	 */
493
	if (flags & EXT4_GET_BLOCKS_DELALLOC_RESERVE)
494
		ext4_set_inode_state(inode, EXT4_STATE_DELALLOC_RESERVED);
495
496
497
498
	/*
	 * We need to check for EXT4 here because migrate
	 * could have changed the inode type in between
	 */
499
	if (ext4_test_inode_flag(inode, EXT4_INODE_EXTENTS)) {
500
		retval = ext4_ext_map_blocks(handle, inode, map, flags);
501
	} else {
502
		retval = ext4_ind_map_blocks(handle, inode, map, flags);
503

504
		if (retval > 0 && map->m_flags & EXT4_MAP_NEW) {
505
506
507
508
509
			/*
			 * We allocated new blocks which will result in
			 * i_data's format changing.  Force the migrate
			 * to fail by clearing migrate flags
			 */
510
			ext4_clear_inode_state(inode, EXT4_STATE_EXT_MIGRATE);
511
		}
512

513
514
515
516
517
518
519
		/*
		 * 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) &&
520
			(flags & EXT4_GET_BLOCKS_DELALLOC_RESERVE))
521
522
			ext4_da_update_reserve_space(inode, retval, 1);
	}
523
	if (flags & EXT4_GET_BLOCKS_DELALLOC_RESERVE)
524
		ext4_clear_inode_state(inode, EXT4_STATE_DELALLOC_RESERVED);
525

526
	up_write((&EXT4_I(inode)->i_data_sem));
527
	if (retval > 0 && map->m_flags & EXT4_MAP_MAPPED) {
528
		int ret = check_block_validity(inode, map);
529
530
531
		if (ret != 0)
			return ret;
	}
532
533
534
	return retval;
}

535
536
537
/* Maximum number of blocks we map for direct IO at once. */
#define DIO_MAX_BLOCKS 4096

538
539
static int _ext4_get_block(struct inode *inode, sector_t iblock,
			   struct buffer_head *bh, int flags)
540
{
541
	handle_t *handle = ext4_journal_current_handle();
542
	struct ext4_map_blocks map;
Jan Kara's avatar
Jan Kara committed
543
	int ret = 0, started = 0;
544
	int dio_credits;
545

546
547
548
549
	map.m_lblk = iblock;
	map.m_len = bh->b_size >> inode->i_blkbits;

	if (flags && !handle) {
Jan Kara's avatar
Jan Kara committed
550
		/* Direct IO write... */
551
552
553
		if (map.m_len > DIO_MAX_BLOCKS)
			map.m_len = DIO_MAX_BLOCKS;
		dio_credits = ext4_chunk_trans_blocks(inode, map.m_len);
554
		handle = ext4_journal_start(inode, dio_credits);
Jan Kara's avatar
Jan Kara committed
555
		if (IS_ERR(handle)) {
556
			ret = PTR_ERR(handle);
557
			return ret;
558
		}
Jan Kara's avatar
Jan Kara committed
559
		started = 1;
560
561
	}

562
	ret = ext4_map_blocks(handle, inode, &map, flags);
Jan Kara's avatar
Jan Kara committed
563
	if (ret > 0) {
564
565
566
		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
567
		ret = 0;
568
	}
Jan Kara's avatar
Jan Kara committed
569
570
	if (started)
		ext4_journal_stop(handle);
571
572
573
	return ret;
}

574
575
576
577
578
579
580
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);
}

581
582
583
/*
 * `handle' can be NULL if create is zero
 */
584
struct buffer_head *ext4_getblk(handle_t *handle, struct inode *inode,
Aneesh Kumar K.V's avatar
Aneesh Kumar K.V committed
585
				ext4_lblk_t block, int create, int *errp)
586
{
587
588
	struct ext4_map_blocks map;
	struct buffer_head *bh;
589
590
591
592
	int fatal = 0, err;

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

593
594
595
596
	map.m_lblk = block;
	map.m_len = 1;
	err = ext4_map_blocks(handle, inode, &map,
			      create ? EXT4_GET_BLOCKS_CREATE : 0);
597

598
599
600
601
602
603
604
605
606
607
	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;
608
	}
609
610
611
	if (map.m_flags & EXT4_MAP_NEW) {
		J_ASSERT(create != 0);
		J_ASSERT(handle != NULL);
612

613
614
615
616
617
618
619
620
621
622
623
624
625
		/*
		 * 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);
626
		}
627
628
629
630
631
632
633
		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");
634
	}
635
636
637
638
639
640
	if (fatal) {
		*errp = fatal;
		brelse(bh);
		bh = NULL;
	}
	return bh;
641
642
}

643
struct buffer_head *ext4_bread(handle_t *handle, struct inode *inode,
Aneesh Kumar K.V's avatar
Aneesh Kumar K.V committed
644
			       ext4_lblk_t block, int create, int *err)
645
{
646
	struct buffer_head *bh;
647

648
	bh = ext4_getblk(handle, inode, block, create, err);
649
650
651
652
	if (!bh)
		return bh;
	if (buffer_uptodate(bh))
		return bh;
653
	ll_rw_block(READ | REQ_META, 1, &bh);
654
655
656
657
658
659
660
661
	wait_on_buffer(bh);
	if (buffer_uptodate(bh))
		return bh;
	put_bh(bh);
	*err = -EIO;
	return NULL;
}

662
663
664
665
666
667
668
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))
669
670
671
672
673
674
675
{
	struct buffer_head *bh;
	unsigned block_start, block_end;
	unsigned blocksize = head->b_size;
	int err, ret = 0;
	struct buffer_head *next;

676
677
	for (bh = head, block_start = 0;
	     ret == 0 && (bh != head || !block_start);
678
	     block_start = block_end, bh = next) {
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
		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
696
 * close off a transaction and start a new one between the ext4_get_block()
697
 * and the commit_write().  So doing the jbd2_journal_start at the start of
698
699
 * prepare_write() is the right place.
 *
700
701
 * Also, this function can nest inside ext4_writepage() ->
 * block_write_full_page(). In that case, we *know* that ext4_writepage()
702
703
704
705
 * 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.
 *
706
 * By accident, ext4 can be reentered when a transaction is open via
707
708
709
710
711
712
 * 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.
 *
713
 * So what we do is to rely on the fact that jbd2_journal_stop/journal_start
714
715
716
717
718
 * 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,
719
				       struct buffer_head *bh)
720
{
721
722
723
	int dirty = buffer_dirty(bh);
	int ret;

724
725
	if (!buffer_mapped(bh) || buffer_freed(bh))
		return 0;
726
	/*
727
	 * __block_write_begin() could have dirtied some buffers. Clean
728
729
	 * the dirty bit as jbd2_journal_get_write_access() could complain
	 * otherwise about fs integrity issues. Setting of the dirty bit
730
	 * by __block_write_begin() isn't a real problem here as we clear
731
732
733
734
735
736
737
738
739
	 * 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;
740
741
}

742
743
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
744
static int ext4_write_begin(struct file *file, struct address_space *mapping,
745
746
			    loff_t pos, unsigned len, unsigned flags,
			    struct page **pagep, void **fsdata)
747
{
748
	struct inode *inode = mapping->host;
749
	int ret, needed_blocks;
750
751
	handle_t *handle;
	int retries = 0;
752
	struct page *page;
753
	pgoff_t index;
754
	unsigned from, to;
Nick Piggin's avatar
Nick Piggin committed
755

756
	trace_ext4_write_begin(inode, pos, len, flags);
757
758
759
760
761
	/*
	 * 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;
762
	index = pos >> PAGE_CACHE_SHIFT;
763
764
	from = pos & (PAGE_CACHE_SIZE - 1);
	to = from + len;
765
766

retry:
767
768
769
770
	handle = ext4_journal_start(inode, needed_blocks);
	if (IS_ERR(handle)) {
		ret = PTR_ERR(handle);
		goto out;
771
	}
772

773
774
775
776
	/* We cannot recurse into the filesystem as the transaction is already
	 * started */
	flags |= AOP_FLAG_NOFS;

777
	page = grab_cache_page_write_begin(mapping, index, flags);
778
779
780
781
782
783
784
	if (!page) {
		ext4_journal_stop(handle);
		ret = -ENOMEM;
		goto out;
	}
	*pagep = page;

785
	if (ext4_should_dioread_nolock(inode))
786
		ret = __block_write_begin(page, pos, len, ext4_get_block_write);
787
	else
788
		ret = __block_write_begin(page, pos, len, ext4_get_block);
Nick Piggin's avatar
Nick Piggin committed
789
790

	if (!ret && ext4_should_journal_data(inode)) {
791
792
793
		ret = walk_page_buffers(handle, page_buffers(page),
				from, to, NULL, do_journal_get_write_access);
	}
Nick Piggin's avatar
Nick Piggin committed
794
795

	if (ret) {
796
797
		unlock_page(page);
		page_cache_release(page);
798
		/*
799
		 * __block_write_begin may have instantiated a few blocks
800
801
		 * outside i_size.  Trim these off again. Don't need
		 * i_size_read because we hold i_mutex.
802
803
804
		 *
		 * Add inode to orphan list in case we crash before
		 * truncate finishes
805
		 */
806
		if (pos + len > inode->i_size && ext4_can_truncate(inode))
807
808
809
810
			ext4_orphan_add(handle, inode);

		ext4_journal_stop(handle);
		if (pos + len > inode->i_size) {
811
			ext4_truncate_failed_write(inode);
812
			/*
813
			 * If truncate failed early the inode might
814
815
816
817
818
819
820
			 * 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
821
822
	}

823
	if (ret == -ENOSPC && ext4_should_retry_alloc(inode->i_sb, &retries))
824
		goto retry;
825
out:
826
827
828
	return ret;
}

Nick Piggin's avatar
Nick Piggin committed
829
830
/* For write_end() in data=journal mode */
static int write_end_fn(handle_t *handle, struct buffer_head *bh)
831
832
833
834
{
	if (!buffer_mapped(bh) || buffer_freed(bh))
		return 0;
	set_buffer_uptodate(bh);
835
	return ext4_handle_dirty_metadata(handle, NULL, bh);
836
837
}

838
static int ext4_generic_write_end(struct file *file,
839
840
841
				  struct address_space *mapping,
				  loff_t pos, unsigned len, unsigned copied,
				  struct page *page, void *fsdata)
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
{
	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;
}

884
885
886
887
/*
 * 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().
 *
888
 * ext4 never places buffers on inode->i_mapping->private_list.  metadata
889
890
 * buffers are managed internally.
 */
Nick Piggin's avatar
Nick Piggin committed
891
static int ext4_ordered_write_end(struct file *file,
892
893
894
				  struct address_space *mapping,
				  loff_t pos, unsigned len, unsigned copied,
				  struct page *page, void *fsdata)
895
{
896
	handle_t *handle = ext4_journal_current_handle();
897
	struct inode *inode = mapping->host;
898
899
	int ret = 0, ret2;

900
	trace_ext4_ordered_write_end(inode, pos, len, copied);
901
	ret = ext4_jbd2_file_inode(handle, inode);
902
903

	if (ret == 0) {
904
		ret2 = ext4_generic_write_end(file, mapping, pos, len, copied,
Nick Piggin's avatar
Nick Piggin committed
905
							page, fsdata);
906
		copied = ret2;
907
		if (pos + len > inode->i_size && ext4_can_truncate(inode))
908
909
910
911
912
			/* 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);
913
914
		if (ret2 < 0)
			ret = ret2;
915
	}
916
	ret2 = ext4_journal_stop(handle);
917
918
	if (!ret)
		ret = ret2;
Nick Piggin's avatar
Nick Piggin committed
919

920
	if (pos + len > inode->i_size) {
921
		ext4_truncate_failed_write(inode);
922
		/*
923
		 * If truncate failed early the inode might still be
924
925
926
927
928
929
930
931
		 * 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
932
	return ret ? ret : copied;
933
934
}

Nick Piggin's avatar
Nick Piggin committed
935
static int ext4_writeback_write_end(struct file *file,
936
937
938
				    struct address_space *mapping,
				    loff_t pos, unsigned len, unsigned copied,
				    struct page *page, void *fsdata)
939
{
940
	handle_t *handle = ext4_journal_current_handle();
941
	struct inode *inode = mapping->host;
942
943
	int ret = 0, ret2;

944
	trace_ext4_writeback_write_end(inode, pos, len, copied);
945
	ret2 = ext4_generic_write_end(file, mapping, pos, len, copied,
Nick Piggin's avatar
Nick Piggin committed
946
							page, fsdata);
947
	copied = ret2;
948
	if (pos + len > inode->i_size && ext4_can_truncate(inode))
949
950
951
952
953
954
		/* 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);

955
956
	if (ret2 < 0)
		ret = ret2;
957

958
	ret2 = ext4_journal_stop(handle);
959
960
	if (!ret)
		ret = ret2;
Nick Piggin's avatar
Nick Piggin committed
961

962
	if (pos + len > inode->i_size) {
963
		ext4_truncate_failed_write(inode);
964
		/*
965
		 * If truncate failed early the inode might still be
966
967
968
969
970
971
972
		 * 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
973
	return ret ? ret : copied;
974
975
}

Nick Piggin's avatar
Nick Piggin committed
976
static int ext4_journalled_write_end(struct file *file,
977
978
979
				     struct address_space *mapping,
				     loff_t pos, unsigned len, unsigned copied,
				     struct page *page, void *fsdata)
980
{
981
	handle_t *handle = ext4_journal_current_handle();
Nick Piggin's avatar
Nick Piggin committed
982
	struct inode *inode = mapping->host;
983
984
	int ret = 0, ret2;
	int partial = 0;
Nick Piggin's avatar
Nick Piggin committed
985
	unsigned from, to;
986
	loff_t new_i_size;
987

988
	trace_ext4_journalled_write_end(inode, pos, len, copied);
Nick Piggin's avatar
Nick Piggin committed
989
990
991
	from = pos & (PAGE_CACHE_SIZE - 1);
	to = from + len;

992
993
	BUG_ON(!ext4_handle_valid(handle));

Nick Piggin's avatar
Nick Piggin committed
994
995
996
997
998
	if (copied < len) {
		if (!PageUptodate(page))
			copied = 0;
		page_zero_new_buffers(page, from+copied, to);
	}
999
1000

	ret = walk_page_buffers(handle, page_buffers(page), from,
For faster browsing, not all history is shown. View entire blame