inode.c 140 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
 */

#include <linux/fs.h>
#include <linux/time.h>
23
#include <linux/jbd2.h>
24
25
26
27
28
29
#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>
30
#include <linux/pagevec.h>
31
#include <linux/mpage.h>
32
#include <linux/namei.h>
33
34
#include <linux/uio.h>
#include <linux/bio.h>
35
#include <linux/workqueue.h>
36
#include <linux/kernel.h>
37
#include <linux/printk.h>
38
#include <linux/slab.h>
39
#include <linux/ratelimit.h>
40

41
#include "ext4_jbd2.h"
42
43
#include "xattr.h"
#include "acl.h"
44
#include "truncate.h"
45

46
47
#include <trace/events/ext4.h>

48
49
#define MPAGE_DA_EXTENT_TAIL 0x01

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
static __u32 ext4_inode_csum(struct inode *inode, struct ext4_inode *raw,
			      struct ext4_inode_info *ei)
{
	struct ext4_sb_info *sbi = EXT4_SB(inode->i_sb);
	__u16 csum_lo;
	__u16 csum_hi = 0;
	__u32 csum;

	csum_lo = raw->i_checksum_lo;
	raw->i_checksum_lo = 0;
	if (EXT4_INODE_SIZE(inode->i_sb) > EXT4_GOOD_OLD_INODE_SIZE &&
	    EXT4_FITS_IN_INODE(raw, ei, i_checksum_hi)) {
		csum_hi = raw->i_checksum_hi;
		raw->i_checksum_hi = 0;
	}

	csum = ext4_chksum(sbi, ei->i_csum_seed, (__u8 *)raw,
			   EXT4_INODE_SIZE(inode->i_sb));

	raw->i_checksum_lo = csum_lo;
	if (EXT4_INODE_SIZE(inode->i_sb) > EXT4_GOOD_OLD_INODE_SIZE &&
	    EXT4_FITS_IN_INODE(raw, ei, i_checksum_hi))
		raw->i_checksum_hi = csum_hi;

	return csum;
}

static int ext4_inode_csum_verify(struct inode *inode, struct ext4_inode *raw,
				  struct ext4_inode_info *ei)
{
	__u32 provided, calculated;

	if (EXT4_SB(inode->i_sb)->s_es->s_creator_os !=
	    cpu_to_le32(EXT4_OS_LINUX) ||
	    !EXT4_HAS_RO_COMPAT_FEATURE(inode->i_sb,
		EXT4_FEATURE_RO_COMPAT_METADATA_CSUM))
		return 1;

	provided = le16_to_cpu(raw->i_checksum_lo);
	calculated = ext4_inode_csum(inode, raw, ei);
	if (EXT4_INODE_SIZE(inode->i_sb) > EXT4_GOOD_OLD_INODE_SIZE &&
	    EXT4_FITS_IN_INODE(raw, ei, i_checksum_hi))
		provided |= ((__u32)le16_to_cpu(raw->i_checksum_hi)) << 16;
	else
		calculated &= 0xFFFF;

	return provided == calculated;
}

static void ext4_inode_csum_set(struct inode *inode, struct ext4_inode *raw,
				struct ext4_inode_info *ei)
{
	__u32 csum;

	if (EXT4_SB(inode->i_sb)->s_es->s_creator_os !=
	    cpu_to_le32(EXT4_OS_LINUX) ||
	    !EXT4_HAS_RO_COMPAT_FEATURE(inode->i_sb,
		EXT4_FEATURE_RO_COMPAT_METADATA_CSUM))
		return;

	csum = ext4_inode_csum(inode, raw, ei);
	raw->i_checksum_lo = cpu_to_le16(csum & 0xFFFF);
	if (EXT4_INODE_SIZE(inode->i_sb) > EXT4_GOOD_OLD_INODE_SIZE &&
	    EXT4_FITS_IN_INODE(raw, ei, i_checksum_hi))
		raw->i_checksum_hi = cpu_to_le16(csum >> 16);
}

117
118
119
static inline int ext4_begin_ordered_truncate(struct inode *inode,
					      loff_t new_size)
{
120
	trace_ext4_begin_ordered_truncate(inode, new_size);
121
122
123
124
125
126
127
128
129
130
131
	/*
	 * 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);
132
133
}

134
static void ext4_invalidatepage(struct page *page, unsigned long offset);
135
136
137
138
139
140
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);
Eric Sandeen's avatar
Eric Sandeen committed
141
142
143
static int ext4_discard_partial_page_buffers_no_lock(handle_t *handle,
		struct inode *inode, struct page *page, loff_t from,
		loff_t length, int flags);
144

145
146
147
/*
 * Test whether an inode is a fast symlink.
 */
148
static int ext4_inode_is_fast_symlink(struct inode *inode)
149
{
150
	int ea_blocks = EXT4_I(inode)->i_file_acl ?
151
152
153
154
155
156
157
158
159
160
		(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.
 */
161
int ext4_truncate_restart_trans(handle_t *handle, struct inode *inode,
162
				 int nblocks)
163
{
164
165
166
	int ret;

	/*
167
	 * Drop i_data_sem to avoid deadlock with ext4_map_blocks.  At this
168
169
170
171
	 * 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.
	 */
172
	BUG_ON(EXT4_JOURNAL(inode) == NULL);
173
	jbd_debug(2, "restarting handle %p\n", handle);
174
	up_write(&EXT4_I(inode)->i_data_sem);
175
	ret = ext4_journal_restart(handle, nblocks);
176
	down_write(&EXT4_I(inode)->i_data_sem);
177
	ext4_discard_preallocations(inode);
178
179

	return ret;
180
181
182
183
184
}

/*
 * Called at the last iput() if i_nlink is zero.
 */
Al Viro's avatar
Al Viro committed
185
void ext4_evict_inode(struct inode *inode)
186
187
{
	handle_t *handle;
188
	int err;
189

190
	trace_ext4_evict_inode(inode);
191
192
193

	ext4_ioend_wait(inode);

Al Viro's avatar
Al Viro committed
194
	if (inode->i_nlink) {
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
		/*
		 * 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
222
223
224
225
		truncate_inode_pages(&inode->i_data, 0);
		goto no_delete;
	}

226
	if (!is_bad_inode(inode))
227
		dquot_initialize(inode);
228

229
230
	if (ext4_should_order_data(inode))
		ext4_begin_ordered_truncate(inode, 0);
231
232
233
234
235
	truncate_inode_pages(&inode->i_data, 0);

	if (is_bad_inode(inode))
		goto no_delete;

236
	handle = ext4_journal_start(inode, ext4_blocks_for_truncate(inode)+3);
237
	if (IS_ERR(handle)) {
238
		ext4_std_error(inode->i_sb, PTR_ERR(handle));
239
240
241
242
243
		/*
		 * 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.
		 */
244
		ext4_orphan_del(NULL, inode);
245
246
247
248
		goto no_delete;
	}

	if (IS_SYNC(inode))
249
		ext4_handle_sync(handle);
250
	inode->i_size = 0;
251
252
	err = ext4_mark_inode_dirty(handle, inode);
	if (err) {
253
		ext4_warning(inode->i_sb,
254
255
256
			     "couldn't mark inode dirty (err %d)", err);
		goto stop_handle;
	}
257
	if (inode->i_blocks)
258
		ext4_truncate(inode);
259
260
261
262
263
264
265

	/*
	 * 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.
	 */
266
	if (!ext4_handle_has_enough_credits(handle, 3)) {
267
268
269
270
		err = ext4_journal_extend(handle, 3);
		if (err > 0)
			err = ext4_journal_restart(handle, 3);
		if (err != 0) {
271
			ext4_warning(inode->i_sb,
272
273
274
				     "couldn't extend journal (err %d)", err);
		stop_handle:
			ext4_journal_stop(handle);
275
			ext4_orphan_del(NULL, inode);
276
277
278
279
			goto no_delete;
		}
	}

280
	/*
281
	 * Kill off the orphan record which ext4_truncate created.
282
	 * AKPM: I think this can be inside the above `if'.
283
	 * Note that ext4_orphan_del() has to be able to cope with the
284
	 * deletion of a non-existent orphan - this is because we don't
285
	 * know if ext4_truncate() actually created an orphan record.
286
287
	 * (Well, we could do this if we need to, but heck - it works)
	 */
288
289
	ext4_orphan_del(handle, inode);
	EXT4_I(inode)->i_dtime	= get_seconds();
290
291
292
293
294
295
296
297

	/*
	 * 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.
	 */
298
	if (ext4_mark_inode_dirty(handle, inode))
299
		/* If that failed, just do the required in-core inode clear. */
Al Viro's avatar
Al Viro committed
300
		ext4_clear_inode(inode);
301
	else
302
303
		ext4_free_inode(handle, inode);
	ext4_journal_stop(handle);
304
305
	return;
no_delete:
Al Viro's avatar
Al Viro committed
306
	ext4_clear_inode(inode);	/* We must guarantee clearing of inode... */
307
308
}

309
310
#ifdef CONFIG_QUOTA
qsize_t *ext4_get_reserved_space(struct inode *inode)
311
{
312
	return &EXT4_I(inode)->i_reserved_quota;
313
}
314
#endif
315

316
317
/*
 * Calculate the number of metadata blocks need to reserve
318
 * to allocate a block located at @lblock
319
 */
320
static int ext4_calc_metadata_amount(struct inode *inode, ext4_lblk_t lblock)
321
{
322
	if (ext4_test_inode_flag(inode, EXT4_INODE_EXTENTS))
323
		return ext4_ext_calc_metadata_amount(inode, lblock);
324

325
	return ext4_ind_calc_metadata_amount(inode, lblock);
326
327
}

328
329
330
331
/*
 * Called with i_data_sem down, which is important since we can call
 * ext4_discard_preallocations() from here.
 */
332
333
void ext4_da_update_reserve_space(struct inode *inode,
					int used, int quota_claim)
334
335
{
	struct ext4_sb_info *sbi = EXT4_SB(inode->i_sb);
336
337
338
	struct ext4_inode_info *ei = EXT4_I(inode);

	spin_lock(&ei->i_block_reservation_lock);
339
	trace_ext4_da_update_reserve_space(inode, used, quota_claim);
340
341
	if (unlikely(used > ei->i_reserved_data_blocks)) {
		ext4_msg(inode->i_sb, KERN_NOTICE, "%s: ino %lu, used %d "
342
			 "with only %d reserved data blocks",
343
344
345
346
347
			 __func__, inode->i_ino, used,
			 ei->i_reserved_data_blocks);
		WARN_ON(1);
		used = ei->i_reserved_data_blocks;
	}
348

349
350
351
	/* Update per-inode reservations */
	ei->i_reserved_data_blocks -= used;
	ei->i_reserved_meta_blocks -= ei->i_allocated_meta_blocks;
352
	percpu_counter_sub(&sbi->s_dirtyclusters_counter,
353
			   used + ei->i_allocated_meta_blocks);
354
	ei->i_allocated_meta_blocks = 0;
355

356
357
358
359
360
361
	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.
		 */
362
		percpu_counter_sub(&sbi->s_dirtyclusters_counter,
363
				   ei->i_reserved_meta_blocks);
364
		ei->i_reserved_meta_blocks = 0;
365
		ei->i_da_metadata_calc_len = 0;
366
	}
367
	spin_unlock(&EXT4_I(inode)->i_block_reservation_lock);
368

369
370
	/* Update quota subsystem for data blocks */
	if (quota_claim)
371
		dquot_claim_block(inode, EXT4_C2B(sbi, used));
372
	else {
373
374
375
		/*
		 * We did fallocate with an offset that is already delayed
		 * allocated. So on delayed allocated writeback we should
376
		 * not re-claim the quota for fallocated blocks.
377
		 */
378
		dquot_release_reservation_block(inode, EXT4_C2B(sbi, used));
379
	}
380
381
382
383
384
385

	/*
	 * 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.
	 */
386
387
	if ((ei->i_reserved_data_blocks == 0) &&
	    (atomic_read(&inode->i_writecount) == 0))
388
		ext4_discard_preallocations(inode);
389
390
}

391
static int __check_block_validity(struct inode *inode, const char *func,
392
393
				unsigned int line,
				struct ext4_map_blocks *map)
394
{
395
396
	if (!ext4_data_block_valid(EXT4_SB(inode->i_sb), map->m_pblk,
				   map->m_len)) {
397
398
399
400
		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);
401
402
403
404
405
		return -EIO;
	}
	return 0;
}

406
#define check_block_validity(inode, map)	\
407
	__check_block_validity((inode), __func__, __LINE__, (map))
408

409
/*
410
411
 * Return the number of contiguous dirty pages in a given inode
 * starting at page frame idx.
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
 */
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;
			}
445
446
447
448
449
450
451
452
453
			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));
			}
454
455
456
457
458
			unlock_page(page);
			if (done)
				break;
			idx++;
			num++;
459
460
			if (num >= max_pages) {
				done = 1;
461
				break;
462
			}
463
464
465
466
467
468
		}
		pagevec_release(&pvec);
	}
	return num;
}

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
/*
 * 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);
	}
}

512
/*
513
 * The ext4_map_blocks() function tries to look up the requested blocks,
514
 * and returns if the blocks are already mapped.
515
516
517
518
519
 *
 * 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.
 *
520
521
 * If file type is extents based, it will call ext4_ext_map_blocks(),
 * Otherwise, call with ext4_ind_map_blocks() to handle indirect mapping
522
523
524
525
526
527
528
529
 * 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
530
 * that case, buffer head is unmapped
531
532
533
 *
 * It returns the error in case of allocation failure.
 */
534
535
int ext4_map_blocks(handle_t *handle, struct inode *inode,
		    struct ext4_map_blocks *map, int flags)
536
537
{
	int retval;
538

539
540
541
542
	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);
543
	/*
544
545
	 * Try to see if we can get the block without requesting a new
	 * file system block.
546
	 */
547
548
	if (!(flags & EXT4_GET_BLOCKS_NO_LOCK))
		down_read((&EXT4_I(inode)->i_data_sem));
549
	if (ext4_test_inode_flag(inode, EXT4_INODE_EXTENTS)) {
550
551
		retval = ext4_ext_map_blocks(handle, inode, map, flags &
					     EXT4_GET_BLOCKS_KEEP_SIZE);
552
	} else {
553
554
		retval = ext4_ind_map_blocks(handle, inode, map, flags &
					     EXT4_GET_BLOCKS_KEEP_SIZE);
555
	}
556
557
	if (!(flags & EXT4_GET_BLOCKS_NO_LOCK))
		up_read((&EXT4_I(inode)->i_data_sem));
558

559
	if (retval > 0 && map->m_flags & EXT4_MAP_MAPPED) {
560
		int ret = check_block_validity(inode, map);
561
562
563
564
		if (ret != 0)
			return ret;
	}

565
	/* If it is only a block(s) look up */
566
	if ((flags & EXT4_GET_BLOCKS_CREATE) == 0)
567
568
569
570
571
572
		return retval;

	/*
	 * Returns if the blocks have already allocated
	 *
	 * Note that if blocks have been preallocated
573
	 * ext4_ext_get_block() returns the create = 0
574
575
	 * with buffer head unmapped.
	 */
576
	if (retval > 0 && map->m_flags & EXT4_MAP_MAPPED)
577
578
		return retval;

579
580
581
582
583
584
585
586
587
588
	/*
	 * 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.
	 */
589
	map->m_flags &= ~EXT4_MAP_UNWRITTEN;
590

591
	/*
592
593
594
595
	 * 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.
596
597
	 */
	down_write((&EXT4_I(inode)->i_data_sem));
598
599
600
601
602
603
604

	/*
	 * 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
	 */
605
	if (flags & EXT4_GET_BLOCKS_DELALLOC_RESERVE)
606
		ext4_set_inode_state(inode, EXT4_STATE_DELALLOC_RESERVED);
607
608
609
610
	/*
	 * We need to check for EXT4 here because migrate
	 * could have changed the inode type in between
	 */
611
	if (ext4_test_inode_flag(inode, EXT4_INODE_EXTENTS)) {
612
		retval = ext4_ext_map_blocks(handle, inode, map, flags);
613
	} else {
614
		retval = ext4_ind_map_blocks(handle, inode, map, flags);
615

616
		if (retval > 0 && map->m_flags & EXT4_MAP_NEW) {
617
618
619
620
621
			/*
			 * We allocated new blocks which will result in
			 * i_data's format changing.  Force the migrate
			 * to fail by clearing migrate flags
			 */
622
			ext4_clear_inode_state(inode, EXT4_STATE_EXT_MIGRATE);
623
		}
624

625
626
627
628
629
630
631
		/*
		 * 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) &&
632
			(flags & EXT4_GET_BLOCKS_DELALLOC_RESERVE))
633
634
			ext4_da_update_reserve_space(inode, retval, 1);
	}
635
	if (flags & EXT4_GET_BLOCKS_DELALLOC_RESERVE) {
636
		ext4_clear_inode_state(inode, EXT4_STATE_DELALLOC_RESERVED);
637

638
639
640
641
642
643
644
645
		/* 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);
	}

646
	up_write((&EXT4_I(inode)->i_data_sem));
647
	if (retval > 0 && map->m_flags & EXT4_MAP_MAPPED) {
648
		int ret = check_block_validity(inode, map);
649
650
651
		if (ret != 0)
			return ret;
	}
652
653
654
	return retval;
}

655
656
657
/* Maximum number of blocks we map for direct IO at once. */
#define DIO_MAX_BLOCKS 4096

658
659
static int _ext4_get_block(struct inode *inode, sector_t iblock,
			   struct buffer_head *bh, int flags)
660
{
661
	handle_t *handle = ext4_journal_current_handle();
662
	struct ext4_map_blocks map;
Jan Kara's avatar
Jan Kara committed
663
	int ret = 0, started = 0;
664
	int dio_credits;
665

666
667
668
669
	map.m_lblk = iblock;
	map.m_len = bh->b_size >> inode->i_blkbits;

	if (flags && !handle) {
Jan Kara's avatar
Jan Kara committed
670
		/* Direct IO write... */
671
672
673
		if (map.m_len > DIO_MAX_BLOCKS)
			map.m_len = DIO_MAX_BLOCKS;
		dio_credits = ext4_chunk_trans_blocks(inode, map.m_len);
674
		handle = ext4_journal_start(inode, dio_credits);
Jan Kara's avatar
Jan Kara committed
675
		if (IS_ERR(handle)) {
676
			ret = PTR_ERR(handle);
677
			return ret;
678
		}
Jan Kara's avatar
Jan Kara committed
679
		started = 1;
680
681
	}

682
	ret = ext4_map_blocks(handle, inode, &map, flags);
Jan Kara's avatar
Jan Kara committed
683
	if (ret > 0) {
684
685
686
		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
687
		ret = 0;
688
	}
Jan Kara's avatar
Jan Kara committed
689
690
	if (started)
		ext4_journal_stop(handle);
691
692
693
	return ret;
}

694
695
696
697
698
699
700
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);
}

701
702
703
/*
 * `handle' can be NULL if create is zero
 */
704
struct buffer_head *ext4_getblk(handle_t *handle, struct inode *inode,
Aneesh Kumar K.V's avatar
Aneesh Kumar K.V committed
705
				ext4_lblk_t block, int create, int *errp)
706
{
707
708
	struct ext4_map_blocks map;
	struct buffer_head *bh;
709
710
711
712
	int fatal = 0, err;

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

713
714
715
716
	map.m_lblk = block;
	map.m_len = 1;
	err = ext4_map_blocks(handle, inode, &map,
			      create ? EXT4_GET_BLOCKS_CREATE : 0);
717

718
719
720
721
722
723
724
725
726
727
	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;
728
	}
729
730
731
	if (map.m_flags & EXT4_MAP_NEW) {
		J_ASSERT(create != 0);
		J_ASSERT(handle != NULL);
732

733
734
735
736
737
738
739
740
741
742
743
744
745
		/*
		 * 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);
746
		}
747
748
749
750
751
752
753
		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");
754
	}
755
756
757
758
759
760
	if (fatal) {
		*errp = fatal;
		brelse(bh);
		bh = NULL;
	}
	return bh;
761
762
}

763
struct buffer_head *ext4_bread(handle_t *handle, struct inode *inode,
Aneesh Kumar K.V's avatar
Aneesh Kumar K.V committed
764
			       ext4_lblk_t block, int create, int *err)
765
{
766
	struct buffer_head *bh;
767

768
	bh = ext4_getblk(handle, inode, block, create, err);
769
770
771
772
	if (!bh)
		return bh;
	if (buffer_uptodate(bh))
		return bh;
773
	ll_rw_block(READ | REQ_META | REQ_PRIO, 1, &bh);
774
775
776
777
778
779
780
781
	wait_on_buffer(bh);
	if (buffer_uptodate(bh))
		return bh;
	put_bh(bh);
	*err = -EIO;
	return NULL;
}

782
783
784
785
786
787
788
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))
789
790
791
792
793
794
795
{
	struct buffer_head *bh;
	unsigned block_start, block_end;
	unsigned blocksize = head->b_size;
	int err, ret = 0;
	struct buffer_head *next;

796
797
	for (bh = head, block_start = 0;
	     ret == 0 && (bh != head || !block_start);
798
	     block_start = block_end, bh = next) {
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
		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
816
 * close off a transaction and start a new one between the ext4_get_block()
817
 * and the commit_write().  So doing the jbd2_journal_start at the start of
818
819
 * prepare_write() is the right place.
 *
820
821
 * Also, this function can nest inside ext4_writepage() ->
 * block_write_full_page(). In that case, we *know* that ext4_writepage()
822
823
824
825
 * 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.
 *
826
 * By accident, ext4 can be reentered when a transaction is open via
827
828
829
830
831
832
 * 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.
 *
833
 * So what we do is to rely on the fact that jbd2_journal_stop/journal_start
834
835
836
837
838
 * 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,
839
				       struct buffer_head *bh)
840
{
841
842
843
	int dirty = buffer_dirty(bh);
	int ret;

844
845
	if (!buffer_mapped(bh) || buffer_freed(bh))
		return 0;
846
	/*
847
	 * __block_write_begin() could have dirtied some buffers. Clean
848
849
	 * the dirty bit as jbd2_journal_get_write_access() could complain
	 * otherwise about fs integrity issues. Setting of the dirty bit
850
	 * by __block_write_begin() isn't a real problem here as we clear
851
852
853
854
855
856
857
858
859
	 * 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;
860
861
}

862
863
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
864
static int ext4_write_begin(struct file *file, struct address_space *mapping,
865
866
			    loff_t pos, unsigned len, unsigned flags,
			    struct page **pagep, void **fsdata)
867
{
868
	struct inode *inode = mapping->host;
869
	int ret, needed_blocks;
870
871
	handle_t *handle;
	int retries = 0;
872
	struct page *page;
873
	pgoff_t index;
874
	unsigned from, to;
Nick Piggin's avatar
Nick Piggin committed
875

876
	trace_ext4_write_begin(inode, pos, len, flags);
877
878
879
880
881
	/*
	 * 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;
882
	index = pos >> PAGE_CACHE_SHIFT;
883
884
	from = pos & (PAGE_CACHE_SIZE - 1);
	to = from + len;
885
886

retry:
887
888
889
890
	handle = ext4_journal_start(inode, needed_blocks);
	if (IS_ERR(handle)) {
		ret = PTR_ERR(handle);
		goto out;
891
	}
892

893
894
895
896
	/* We cannot recurse into the filesystem as the transaction is already
	 * started */
	flags |= AOP_FLAG_NOFS;

897
	page = grab_cache_page_write_begin(mapping, index, flags);
898
899
900
901
902
903
904
	if (!page) {
		ext4_journal_stop(handle);
		ret = -ENOMEM;
		goto out;
	}
	*pagep = page;

905
	if (ext4_should_dioread_nolock(inode))
906
		ret = __block_write_begin(page, pos, len, ext4_get_block_write);
907
	else
908
		ret = __block_write_begin(page, pos, len, ext4_get_block);
Nick Piggin's avatar
Nick Piggin committed
909
910

	if (!ret && ext4_should_journal_data(inode)) {
911
912
913
		ret = walk_page_buffers(handle, page_buffers(page),
				from, to, NULL, do_journal_get_write_access);
	}
Nick Piggin's avatar
Nick Piggin committed
914
915

	if (ret) {
916
917
		unlock_page(page);
		page_cache_release(page);
918
		/*
919
		 * __block_write_begin may have instantiated a few blocks
920
921
		 * outside i_size.  Trim these off again. Don't need
		 * i_size_read because we hold i_mutex.
922
923
924
		 *
		 * Add inode to orphan list in case we crash before
		 * truncate finishes
925
		 */
926
		if (pos + len > inode->i_size && ext4_can_truncate(inode))
927
928
929
930
			ext4_orphan_add(handle, inode);

		ext4_journal_stop(handle);
		if (pos + len > inode->i_size) {
931
			ext4_truncate_failed_write(inode);
932
			/*
933
			 * If truncate failed early the inode might
934
935
936
937
938
939
940
			 * 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
941
942
	}

943
	if (ret == -ENOSPC && ext4_should_retry_alloc(inode->i_sb, &retries))
944
		goto retry;
945
out:
946
947
948
	return ret;
}

Nick Piggin's avatar
Nick Piggin committed
949
950
/* For write_end() in data=journal mode */
static int write_end_fn(handle_t *handle, struct buffer_head *bh)
951
952
953
954
{
	if (!buffer_mapped(bh) || buffer_freed(bh))
		return 0;
	set_buffer_uptodate(bh);
955
	return ext4_handle_dirty_metadata(handle, NULL, bh);
956
957
}

958
static int ext4_generic_write_end(struct file *file,
959
960
961
				  struct address_space *mapping,
				  loff_t pos, unsigned len, unsigned copied,
				  struct page *page, void *fsdata)
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
{
	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);

For faster browsing, not all history is shown. View entire blame