inode.c 167.9 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 18 19 20 21
 *
 * 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
 *
 *  Goal-directed block allocation by Stephen Tweedie
 *	(sct@redhat.com), 1993, 1998
 *  Big-endian to little-endian byte-swapping/bitmaps by
 *        David S. Miller (davem@caip.rutgers.edu), 1995
 *  64-bit file support on 64-bit platforms by Jakub Jelinek
 *	(jj@sunsite.ms.mff.cuni.cz)
 *
22
 *  Assorted race fixes, rewrite of ext4_get_block() by Al Viro, 2000
23 24 25 26 27
 */

#include <linux/module.h>
#include <linux/fs.h>
#include <linux/time.h>
28
#include <linux/jbd2.h>
29 30 31 32 33 34
#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>
35
#include <linux/pagevec.h>
36
#include <linux/mpage.h>
37
#include <linux/namei.h>
38 39
#include <linux/uio.h>
#include <linux/bio.h>
40
#include <linux/workqueue.h>
41

42
#include "ext4_jbd2.h"
43 44
#include "xattr.h"
#include "acl.h"
45
#include "ext4_extents.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 55 56 57
	return jbd2_journal_begin_ordered_truncate(
					EXT4_SB(inode->i_sb)->s_journal,
					&EXT4_I(inode)->jinode,
					new_size);
58 59
}

60 61
static void ext4_invalidatepage(struct page *page, unsigned long offset);

62 63 64
/*
 * Test whether an inode is a fast symlink.
 */
65
static int ext4_inode_is_fast_symlink(struct inode *inode)
66
{
67
	int ea_blocks = EXT4_I(inode)->i_file_acl ?
68 69 70 71 72 73
		(inode->i_sb->s_blocksize >> 9) : 0;

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

/*
74
 * The ext4 forget function must perform a revoke if we are freeing data
75 76 77 78 79 80
 * which has been journaled.  Metadata (eg. indirect blocks) must be
 * revoked in all cases.
 *
 * "bh" may be NULL: a metadata block may have been freed from memory
 * but there may still be a record of it in the journal, and that record
 * still needs to be revoked.
81
 *
82 83
 * If the handle isn't valid we're not journaling, but we still need to
 * call into ext4_journal_revoke() to put the buffer head.
84
 */
85
int ext4_forget(handle_t *handle, int is_metadata, struct inode *inode,
86
		struct buffer_head *bh, ext4_fsblk_t blocknr)
87 88 89 90 91 92 93 94
{
	int err;

	might_sleep();

	BUFFER_TRACE(bh, "enter");

	jbd_debug(4, "forgetting bh %p: is_metadata = %d, mode %o, "
95
		  "data mode %x\n",
96 97 98 99 100 101 102 103
		  bh, is_metadata, inode->i_mode,
		  test_opt(inode->i_sb, DATA_FLAGS));

	/* Never use the revoke function if we are doing full data
	 * journaling: there is no need to, and a V1 superblock won't
	 * support it.  Otherwise, only skip the revoke on un-journaled
	 * data blocks. */

104 105
	if (test_opt(inode->i_sb, DATA_FLAGS) == EXT4_MOUNT_JOURNAL_DATA ||
	    (!is_metadata && !ext4_should_journal_data(inode))) {
106
		if (bh) {
107
			BUFFER_TRACE(bh, "call jbd2_journal_forget");
108
			return ext4_journal_forget(handle, bh);
109 110 111 112 113 114 115
		}
		return 0;
	}

	/*
	 * data!=journal && (is_metadata || should_journal_data(inode))
	 */
116 117
	BUFFER_TRACE(bh, "call ext4_journal_revoke");
	err = ext4_journal_revoke(handle, blocknr, bh);
118
	if (err)
119
		ext4_abort(inode->i_sb, __func__,
120 121 122 123 124 125 126 127 128 129 130
			   "error %d when attempting revoke", err);
	BUFFER_TRACE(bh, "exit");
	return err;
}

/*
 * Work out how many blocks we need to proceed with the next chunk of a
 * truncate transaction.
 */
static unsigned long blocks_for_truncate(struct inode *inode)
{
A
Aneesh Kumar K.V 已提交
131
	ext4_lblk_t needed;
132 133 134 135 136 137

	needed = inode->i_blocks >> (inode->i_sb->s_blocksize_bits - 9);

	/* Give ourselves just enough room to cope with inodes in which
	 * i_blocks is corrupt: we've seen disk corruptions in the past
	 * which resulted in random data in an inode which looked enough
138
	 * like a regular file for ext4 to try to delete it.  Things
139 140 141 142 143 144 145
	 * will go a bit crazy if that happens, but at least we should
	 * try not to panic the whole kernel. */
	if (needed < 2)
		needed = 2;

	/* But we need to bound the transaction so we don't overflow the
	 * journal. */
146 147
	if (needed > EXT4_MAX_TRANS_DATA)
		needed = EXT4_MAX_TRANS_DATA;
148

149
	return EXT4_DATA_TRANS_BLOCKS(inode->i_sb) + needed;
150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165
}

/*
 * Truncate transactions can be complex and absolutely huge.  So we need to
 * be able to restart the transaction at a conventient checkpoint to make
 * sure we don't overflow the journal.
 *
 * start_transaction gets us a new handle for a truncate transaction,
 * and extend_transaction tries to extend the existing one a bit.  If
 * extend fails, we need to propagate the failure up and restart the
 * transaction in the top-level truncate loop. --sct
 */
static handle_t *start_transaction(struct inode *inode)
{
	handle_t *result;

166
	result = ext4_journal_start(inode, blocks_for_truncate(inode));
167 168 169
	if (!IS_ERR(result))
		return result;

170
	ext4_std_error(inode->i_sb, PTR_ERR(result));
171 172 173 174 175 176 177 178 179 180 181
	return result;
}

/*
 * Try to extend this transaction for the purposes of truncation.
 *
 * Returns 0 if we managed to create more room.  If we can't create more
 * room, and the transaction must be restarted we return 1.
 */
static int try_to_extend_transaction(handle_t *handle, struct inode *inode)
{
182 183 184
	if (!ext4_handle_valid(handle))
		return 0;
	if (ext4_handle_has_enough_credits(handle, EXT4_RESERVE_TRANS_BLOCKS+1))
185
		return 0;
186
	if (!ext4_journal_extend(handle, blocks_for_truncate(inode)))
187 188 189 190 191 192 193 194 195
		return 0;
	return 1;
}

/*
 * Restart the transaction associated with *handle.  This does a commit,
 * so before we call here everything must be consistently dirtied against
 * this transaction.
 */
196
int ext4_truncate_restart_trans(handle_t *handle, struct inode *inode,
197
				 int nblocks)
198
{
199 200 201 202 203 204 205 206
	int ret;

	/*
	 * Drop i_data_sem to avoid deadlock with ext4_get_blocks At this
	 * 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.
	 */
207
	BUG_ON(EXT4_JOURNAL(inode) == NULL);
208
	jbd_debug(2, "restarting handle %p\n", handle);
209 210 211
	up_write(&EXT4_I(inode)->i_data_sem);
	ret = ext4_journal_restart(handle, blocks_for_truncate(inode));
	down_write(&EXT4_I(inode)->i_data_sem);
212
	ext4_discard_preallocations(inode);
213 214

	return ret;
215 216 217 218 219
}

/*
 * Called at the last iput() if i_nlink is zero.
 */
220
void ext4_delete_inode(struct inode *inode)
221 222
{
	handle_t *handle;
223
	int err;
224

225 226
	if (ext4_should_order_data(inode))
		ext4_begin_ordered_truncate(inode, 0);
227 228 229 230 231
	truncate_inode_pages(&inode->i_data, 0);

	if (is_bad_inode(inode))
		goto no_delete;

232
	handle = ext4_journal_start(inode, blocks_for_truncate(inode)+3);
233
	if (IS_ERR(handle)) {
234
		ext4_std_error(inode->i_sb, PTR_ERR(handle));
235 236 237 238 239
		/*
		 * 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.
		 */
240
		ext4_orphan_del(NULL, inode);
241 242 243 244
		goto no_delete;
	}

	if (IS_SYNC(inode))
245
		ext4_handle_sync(handle);
246
	inode->i_size = 0;
247 248 249 250 251 252
	err = ext4_mark_inode_dirty(handle, inode);
	if (err) {
		ext4_warning(inode->i_sb, __func__,
			     "couldn't mark inode dirty (err %d)", err);
		goto stop_handle;
	}
253
	if (inode->i_blocks)
254
		ext4_truncate(inode);
255 256 257 258 259 260 261

	/*
	 * 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.
	 */
262
	if (!ext4_handle_has_enough_credits(handle, 3)) {
263 264 265 266 267 268 269 270 271 272 273 274
		err = ext4_journal_extend(handle, 3);
		if (err > 0)
			err = ext4_journal_restart(handle, 3);
		if (err != 0) {
			ext4_warning(inode->i_sb, __func__,
				     "couldn't extend journal (err %d)", err);
		stop_handle:
			ext4_journal_stop(handle);
			goto no_delete;
		}
	}

275
	/*
276
	 * Kill off the orphan record which ext4_truncate created.
277
	 * AKPM: I think this can be inside the above `if'.
278
	 * Note that ext4_orphan_del() has to be able to cope with the
279
	 * deletion of a non-existent orphan - this is because we don't
280
	 * know if ext4_truncate() actually created an orphan record.
281 282
	 * (Well, we could do this if we need to, but heck - it works)
	 */
283 284
	ext4_orphan_del(handle, inode);
	EXT4_I(inode)->i_dtime	= get_seconds();
285 286 287 288 289 290 291 292

	/*
	 * 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.
	 */
293
	if (ext4_mark_inode_dirty(handle, inode))
294 295 296
		/* If that failed, just do the required in-core inode clear. */
		clear_inode(inode);
	else
297 298
		ext4_free_inode(handle, inode);
	ext4_journal_stop(handle);
299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316
	return;
no_delete:
	clear_inode(inode);	/* We must guarantee clearing of inode... */
}

typedef struct {
	__le32	*p;
	__le32	key;
	struct buffer_head *bh;
} Indirect;

static inline void add_chain(Indirect *p, struct buffer_head *bh, __le32 *v)
{
	p->key = *(p->p = v);
	p->bh = bh;
}

/**
317
 *	ext4_block_to_path - parse the block number into array of offsets
318 319 320
 *	@inode: inode in question (we are only interested in its superblock)
 *	@i_block: block number to be parsed
 *	@offsets: array to store the offsets in
D
Dave Kleikamp 已提交
321 322
 *	@boundary: set this non-zero if the referred-to block is likely to be
 *	       followed (on disk) by an indirect block.
323
 *
324
 *	To store the locations of file's data ext4 uses a data structure common
325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346
 *	for UNIX filesystems - tree of pointers anchored in the inode, with
 *	data blocks at leaves and indirect blocks in intermediate nodes.
 *	This function translates the block number into path in that tree -
 *	return value is the path length and @offsets[n] is the offset of
 *	pointer to (n+1)th node in the nth one. If @block is out of range
 *	(negative or too large) warning is printed and zero returned.
 *
 *	Note: function doesn't find node addresses, so no IO is needed. All
 *	we need to know is the capacity of indirect blocks (taken from the
 *	inode->i_sb).
 */

/*
 * Portability note: the last comparison (check that we fit into triple
 * indirect block) is spelled differently, because otherwise on an
 * architecture with 32-bit longs and 8Kb pages we might get into trouble
 * if our filesystem had 8Kb blocks. We might use long long, but that would
 * kill us on x86. Oh, well, at least the sign propagation does not matter -
 * i_block would have to be negative in the very beginning, so we would not
 * get there at all.
 */

347
static int ext4_block_to_path(struct inode *inode,
348 349
			      ext4_lblk_t i_block,
			      ext4_lblk_t offsets[4], int *boundary)
350
{
351 352 353
	int ptrs = EXT4_ADDR_PER_BLOCK(inode->i_sb);
	int ptrs_bits = EXT4_ADDR_PER_BLOCK_BITS(inode->i_sb);
	const long direct_blocks = EXT4_NDIR_BLOCKS,
354 355 356 357 358
		indirect_blocks = ptrs,
		double_blocks = (1 << (ptrs_bits * 2));
	int n = 0;
	int final = 0;

359
	if (i_block < direct_blocks) {
360 361
		offsets[n++] = i_block;
		final = direct_blocks;
362
	} else if ((i_block -= direct_blocks) < indirect_blocks) {
363
		offsets[n++] = EXT4_IND_BLOCK;
364 365 366
		offsets[n++] = i_block;
		final = ptrs;
	} else if ((i_block -= indirect_blocks) < double_blocks) {
367
		offsets[n++] = EXT4_DIND_BLOCK;
368 369 370 371
		offsets[n++] = i_block >> ptrs_bits;
		offsets[n++] = i_block & (ptrs - 1);
		final = ptrs;
	} else if (((i_block -= double_blocks) >> (ptrs_bits * 2)) < ptrs) {
372
		offsets[n++] = EXT4_TIND_BLOCK;
373 374 375 376 377
		offsets[n++] = i_block >> (ptrs_bits * 2);
		offsets[n++] = (i_block >> ptrs_bits) & (ptrs - 1);
		offsets[n++] = i_block & (ptrs - 1);
		final = ptrs;
	} else {
378
		ext4_warning(inode->i_sb, "ext4_block_to_path",
379 380 381
			     "block %lu > max in inode %lu",
			     i_block + direct_blocks +
			     indirect_blocks + double_blocks, inode->i_ino);
382 383 384 385 386 387
	}
	if (boundary)
		*boundary = final - 1 - (i_block & (ptrs - 1));
	return n;
}

388
static int __ext4_check_blockref(const char *function, struct inode *inode,
389 390
				 __le32 *p, unsigned int max)
{
391
	__le32 *bref = p;
392 393
	unsigned int blk;

394
	while (bref < p+max) {
395
		blk = le32_to_cpu(*bref++);
396 397
		if (blk &&
		    unlikely(!ext4_data_block_valid(EXT4_SB(inode->i_sb),
398
						    blk, 1))) {
399
			ext4_error(inode->i_sb, function,
400 401
				   "invalid block reference %u "
				   "in inode #%lu", blk, inode->i_ino);
402 403 404 405
			return -EIO;
		}
	}
	return 0;
406 407 408 409
}


#define ext4_check_indirect_blockref(inode, bh)                         \
410
	__ext4_check_blockref(__func__, inode, (__le32 *)(bh)->b_data,  \
411 412 413
			      EXT4_ADDR_PER_BLOCK((inode)->i_sb))

#define ext4_check_inode_blockref(inode)                                \
414
	__ext4_check_blockref(__func__, inode, EXT4_I(inode)->i_data,   \
415 416
			      EXT4_NDIR_BLOCKS)

417
/**
418
 *	ext4_get_branch - read the chain of indirect blocks leading to data
419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442
 *	@inode: inode in question
 *	@depth: depth of the chain (1 - direct pointer, etc.)
 *	@offsets: offsets of pointers in inode/indirect blocks
 *	@chain: place to store the result
 *	@err: here we store the error value
 *
 *	Function fills the array of triples <key, p, bh> and returns %NULL
 *	if everything went OK or the pointer to the last filled triple
 *	(incomplete one) otherwise. Upon the return chain[i].key contains
 *	the number of (i+1)-th block in the chain (as it is stored in memory,
 *	i.e. little-endian 32-bit), chain[i].p contains the address of that
 *	number (it points into struct inode for i==0 and into the bh->b_data
 *	for i>0) and chain[i].bh points to the buffer_head of i-th indirect
 *	block for i>0 and NULL for i==0. In other words, it holds the block
 *	numbers of the chain, addresses they were taken from (and where we can
 *	verify that chain did not change) and buffer_heads hosting these
 *	numbers.
 *
 *	Function stops when it stumbles upon zero pointer (absent block)
 *		(pointer to last triple returned, *@err == 0)
 *	or when it gets an IO error reading an indirect block
 *		(ditto, *@err == -EIO)
 *	or when it reads all @depth-1 indirect blocks successfully and finds
 *	the whole chain, all way to the data (returns %NULL, *err == 0).
443 444
 *
 *      Need to be called with
445
 *      down_read(&EXT4_I(inode)->i_data_sem)
446
 */
A
Aneesh Kumar K.V 已提交
447 448
static Indirect *ext4_get_branch(struct inode *inode, int depth,
				 ext4_lblk_t  *offsets,
449 450 451 452 453 454 455 456
				 Indirect chain[4], int *err)
{
	struct super_block *sb = inode->i_sb;
	Indirect *p = chain;
	struct buffer_head *bh;

	*err = 0;
	/* i_data is not going away, no lock needed */
457
	add_chain(chain, NULL, EXT4_I(inode)->i_data + *offsets);
458 459 460
	if (!p->key)
		goto no_block;
	while (--depth) {
461 462
		bh = sb_getblk(sb, le32_to_cpu(p->key));
		if (unlikely(!bh))
463
			goto failure;
464

465 466 467 468 469 470 471 472 473 474 475
		if (!bh_uptodate_or_lock(bh)) {
			if (bh_submit_read(bh) < 0) {
				put_bh(bh);
				goto failure;
			}
			/* validate block references */
			if (ext4_check_indirect_blockref(inode, bh)) {
				put_bh(bh);
				goto failure;
			}
		}
476

477
		add_chain(++p, bh, (__le32 *)bh->b_data + *++offsets);
478 479 480 481 482 483 484 485 486 487 488 489 490
		/* Reader: end */
		if (!p->key)
			goto no_block;
	}
	return NULL;

failure:
	*err = -EIO;
no_block:
	return p;
}

/**
491
 *	ext4_find_near - find a place for allocation with sufficient locality
492 493 494
 *	@inode: owner
 *	@ind: descriptor of indirect block.
 *
495
 *	This function returns the preferred place for block allocation.
496 497 498 499 500 501 502 503 504 505 506 507 508 509
 *	It is used when heuristic for sequential allocation fails.
 *	Rules are:
 *	  + if there is a block to the left of our position - allocate near it.
 *	  + if pointer will live in indirect block - allocate near that block.
 *	  + if pointer will live in inode - allocate in the same
 *	    cylinder group.
 *
 * In the latter case we colour the starting block by the callers PID to
 * prevent it from clashing with concurrent allocations for a different inode
 * in the same block group.   The PID is used here so that functionally related
 * files will be close-by on-disk.
 *
 *	Caller must make sure that @ind is valid and will stay that way.
 */
510
static ext4_fsblk_t ext4_find_near(struct inode *inode, Indirect *ind)
511
{
512
	struct ext4_inode_info *ei = EXT4_I(inode);
513
	__le32 *start = ind->bh ? (__le32 *) ind->bh->b_data : ei->i_data;
514
	__le32 *p;
515
	ext4_fsblk_t bg_start;
516
	ext4_fsblk_t last_block;
517
	ext4_grpblk_t colour;
518 519
	ext4_group_t block_group;
	int flex_size = ext4_flex_bg_size(EXT4_SB(inode->i_sb));
520 521 522 523 524 525 526 527 528 529 530 531 532 533 534

	/* Try to find previous block */
	for (p = ind->p - 1; p >= start; p--) {
		if (*p)
			return le32_to_cpu(*p);
	}

	/* No such thing, so let's try location of indirect block */
	if (ind->bh)
		return ind->bh->b_blocknr;

	/*
	 * It is going to be referred to from the inode itself? OK, just put it
	 * into the same cylinder group then.
	 */
535 536 537 538 539 540 541
	block_group = ei->i_block_group;
	if (flex_size >= EXT4_FLEX_SIZE_DIR_ALLOC_SCHEME) {
		block_group &= ~(flex_size-1);
		if (S_ISREG(inode->i_mode))
			block_group++;
	}
	bg_start = ext4_group_first_block_no(inode->i_sb, block_group);
542 543
	last_block = ext4_blocks_count(EXT4_SB(inode->i_sb)->s_es) - 1;

544 545 546 547 548 549 550
	/*
	 * If we are doing delayed allocation, we don't need take
	 * colour into account.
	 */
	if (test_opt(inode->i_sb, DELALLOC))
		return bg_start;

551 552
	if (bg_start + EXT4_BLOCKS_PER_GROUP(inode->i_sb) <= last_block)
		colour = (current->pid % 16) *
553
			(EXT4_BLOCKS_PER_GROUP(inode->i_sb) / 16);
554 555
	else
		colour = (current->pid % 16) * ((last_block - bg_start) / 16);
556 557 558 559
	return bg_start + colour;
}

/**
560
 *	ext4_find_goal - find a preferred place for allocation.
561 562 563 564
 *	@inode: owner
 *	@block:  block we want
 *	@partial: pointer to the last triple within a chain
 *
565
 *	Normally this function find the preferred place for block allocation,
566
 *	returns it.
567 568
 *	Because this is only used for non-extent files, we limit the block nr
 *	to 32 bits.
569
 */
A
Aneesh Kumar K.V 已提交
570
static ext4_fsblk_t ext4_find_goal(struct inode *inode, ext4_lblk_t block,
571
				   Indirect *partial)
572
{
573 574
	ext4_fsblk_t goal;

575
	/*
576
	 * XXX need to get goal block from mballoc's data structures
577 578
	 */

579 580 581
	goal = ext4_find_near(inode, partial);
	goal = goal & EXT4_MAX_BLOCK_FILE_PHYS;
	return goal;
582 583 584
}

/**
585
 *	ext4_blks_to_allocate: Look up the block map and count the number
586 587 588 589 590 591 592 593 594 595
 *	of direct blocks need to be allocated for the given branch.
 *
 *	@branch: chain of indirect blocks
 *	@k: number of blocks need for indirect blocks
 *	@blks: number of data blocks to be mapped.
 *	@blocks_to_boundary:  the offset in the indirect block
 *
 *	return the total number of blocks to be allocate, including the
 *	direct and indirect blocks.
 */
596
static int ext4_blks_to_allocate(Indirect *branch, int k, unsigned int blks,
597
				 int blocks_to_boundary)
598
{
599
	unsigned int count = 0;
600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622

	/*
	 * Simple case, [t,d]Indirect block(s) has not allocated yet
	 * then it's clear blocks on that path have not allocated
	 */
	if (k > 0) {
		/* right now we don't handle cross boundary allocation */
		if (blks < blocks_to_boundary + 1)
			count += blks;
		else
			count += blocks_to_boundary + 1;
		return count;
	}

	count++;
	while (count < blks && count <= blocks_to_boundary &&
		le32_to_cpu(*(branch[0].p + count)) == 0) {
		count++;
	}
	return count;
}

/**
623
 *	ext4_alloc_blocks: multiple allocate blocks needed for a branch
624 625 626 627 628 629 630 631
 *	@indirect_blks: the number of blocks need to allocate for indirect
 *			blocks
 *
 *	@new_blocks: on return it will store the new block numbers for
 *	the indirect blocks(if needed) and the first direct block,
 *	@blks:	on return it will store the total number of allocated
 *		direct blocks
 */
632
static int ext4_alloc_blocks(handle_t *handle, struct inode *inode,
633 634 635
			     ext4_lblk_t iblock, ext4_fsblk_t goal,
			     int indirect_blks, int blks,
			     ext4_fsblk_t new_blocks[4], int *err)
636
{
637
	struct ext4_allocation_request ar;
638
	int target, i;
639
	unsigned long count = 0, blk_allocated = 0;
640
	int index = 0;
641
	ext4_fsblk_t current_block = 0;
642 643 644 645 646 647 648 649 650 651
	int ret = 0;

	/*
	 * Here we try to allocate the requested multiple blocks at once,
	 * on a best-effort basis.
	 * To build a branch, we should allocate blocks for
	 * the indirect blocks(if not allocated yet), and at least
	 * the first direct block of this branch.  That's the
	 * minimum number of blocks need to allocate(required)
	 */
652 653 654
	/* first we try to allocate the indirect blocks */
	target = indirect_blks;
	while (target > 0) {
655 656
		count = target;
		/* allocating blocks for indirect blocks and direct blocks */
657 658
		current_block = ext4_new_meta_blocks(handle, inode,
							goal, &count, err);
659 660 661
		if (*err)
			goto failed_out;

662 663
		BUG_ON(current_block + count > EXT4_MAX_BLOCK_FILE_PHYS);

664 665 666 667 668 669
		target -= count;
		/* allocate blocks for indirect blocks */
		while (index < indirect_blks && count) {
			new_blocks[index++] = current_block++;
			count--;
		}
670 671 672 673 674 675 676 677 678
		if (count > 0) {
			/*
			 * save the new block number
			 * for the first direct block
			 */
			new_blocks[index] = current_block;
			printk(KERN_INFO "%s returned more blocks than "
						"requested\n", __func__);
			WARN_ON(1);
679
			break;
680
		}
681 682
	}

683 684 685 686 687
	target = blks - count ;
	blk_allocated = count;
	if (!target)
		goto allocated;
	/* Now allocate data blocks */
688 689 690 691 692 693 694 695 696 697
	memset(&ar, 0, sizeof(ar));
	ar.inode = inode;
	ar.goal = goal;
	ar.len = target;
	ar.logical = iblock;
	if (S_ISREG(inode->i_mode))
		/* enable in-core preallocation only for regular files */
		ar.flags = EXT4_MB_HINT_DATA;

	current_block = ext4_mb_new_blocks(handle, &ar, err);
698
	BUG_ON(current_block + ar.len > EXT4_MAX_BLOCK_FILE_PHYS);
699

700 701 702 703 704 705 706 707 708
	if (*err && (target == blks)) {
		/*
		 * if the allocation failed and we didn't allocate
		 * any blocks before
		 */
		goto failed_out;
	}
	if (!*err) {
		if (target == blks) {
709 710 711 712
			/*
			 * save the new block number
			 * for the first direct block
			 */
713 714
			new_blocks[index] = current_block;
		}
715
		blk_allocated += ar.len;
716 717
	}
allocated:
718
	/* total number of blocks allocated for direct blocks */
719
	ret = blk_allocated;
720 721 722
	*err = 0;
	return ret;
failed_out:
723
	for (i = 0; i < index; i++)
724
		ext4_free_blocks(handle, inode, new_blocks[i], 1, 0);
725 726 727 728
	return ret;
}

/**
729
 *	ext4_alloc_branch - allocate and set up a chain of blocks.
730 731 732 733 734 735 736 737 738 739
 *	@inode: owner
 *	@indirect_blks: number of allocated indirect blocks
 *	@blks: number of allocated direct blocks
 *	@offsets: offsets (in the blocks) to store the pointers to next.
 *	@branch: place to store the chain in.
 *
 *	This function allocates blocks, zeroes out all but the last one,
 *	links them into chain and (if we are synchronous) writes them to disk.
 *	In other words, it prepares a branch that can be spliced onto the
 *	inode. It stores the information about that chain in the branch[], in
740
 *	the same format as ext4_get_branch() would do. We are calling it after
741 742
 *	we had read the existing part of chain and partial points to the last
 *	triple of that (one with zero ->key). Upon the exit we have the same
743
 *	picture as after the successful ext4_get_block(), except that in one
744 745 746 747 748 749
 *	place chain is disconnected - *branch->p is still zero (we did not
 *	set the last link), but branch->key contains the number that should
 *	be placed into *branch->p to fill that gap.
 *
 *	If allocation fails we free all blocks we've allocated (and forget
 *	their buffer_heads) and return the error value the from failed
750
 *	ext4_alloc_block() (normally -ENOSPC). Otherwise we set the chain
751 752
 *	as described above and return 0.
 */
753
static int ext4_alloc_branch(handle_t *handle, struct inode *inode,
754 755 756
			     ext4_lblk_t iblock, int indirect_blks,
			     int *blks, ext4_fsblk_t goal,
			     ext4_lblk_t *offsets, Indirect *branch)
757 758 759 760 761 762
{
	int blocksize = inode->i_sb->s_blocksize;
	int i, n = 0;
	int err = 0;
	struct buffer_head *bh;
	int num;
763 764
	ext4_fsblk_t new_blocks[4];
	ext4_fsblk_t current_block;
765

766
	num = ext4_alloc_blocks(handle, inode, iblock, goal, indirect_blks,
767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784
				*blks, new_blocks, &err);
	if (err)
		return err;

	branch[0].key = cpu_to_le32(new_blocks[0]);
	/*
	 * metadata blocks and data blocks are allocated.
	 */
	for (n = 1; n <= indirect_blks;  n++) {
		/*
		 * Get buffer_head for parent block, zero it out
		 * and set the pointer to new one, then send
		 * parent to disk.
		 */
		bh = sb_getblk(inode->i_sb, new_blocks[n-1]);
		branch[n].bh = bh;
		lock_buffer(bh);
		BUFFER_TRACE(bh, "call get_create_access");
785
		err = ext4_journal_get_create_access(handle, bh);
786
		if (err) {
787 788
			/* Don't brelse(bh) here; it's done in
			 * ext4_journal_forget() below */
789 790 791 792 793 794 795 796
			unlock_buffer(bh);
			goto failed;
		}

		memset(bh->b_data, 0, blocksize);
		branch[n].p = (__le32 *) bh->b_data + offsets[n];
		branch[n].key = cpu_to_le32(new_blocks[n]);
		*branch[n].p = branch[n].key;
797
		if (n == indirect_blks) {
798 799 800 801 802 803
			current_block = new_blocks[n];
			/*
			 * End of chain, update the last new metablock of
			 * the chain to point to the new allocated
			 * data blocks numbers
			 */
804
			for (i = 1; i < num; i++)
805 806 807 808 809 810
				*(branch[n].p + i) = cpu_to_le32(++current_block);
		}
		BUFFER_TRACE(bh, "marking uptodate");
		set_buffer_uptodate(bh);
		unlock_buffer(bh);

811 812
		BUFFER_TRACE(bh, "call ext4_handle_dirty_metadata");
		err = ext4_handle_dirty_metadata(handle, inode, bh);
813 814 815 816 817 818 819 820
		if (err)
			goto failed;
	}
	*blks = num;
	return err;
failed:
	/* Allocation failed, free what we already allocated */
	for (i = 1; i <= n ; i++) {
821
		BUFFER_TRACE(branch[i].bh, "call jbd2_journal_forget");
822
		ext4_journal_forget(handle, branch[i].bh);
823
	}
824
	for (i = 0; i < indirect_blks; i++)
825
		ext4_free_blocks(handle, inode, new_blocks[i], 1, 0);
826

827
	ext4_free_blocks(handle, inode, new_blocks[i], num, 0);
828 829 830 831 832

	return err;
}

/**
833
 * ext4_splice_branch - splice the allocated branch onto inode.
834 835 836
 * @inode: owner
 * @block: (logical) number of block we are adding
 * @chain: chain of indirect blocks (with a missing link - see
837
 *	ext4_alloc_branch)
838 839 840 841 842 843 844 845
 * @where: location of missing link
 * @num:   number of indirect blocks we are adding
 * @blks:  number of direct blocks we are adding
 *
 * This function fills the missing link and does all housekeeping needed in
 * inode (->i_blocks, etc.). In case of success we end up with the full
 * chain to new block and return 0.
 */
846
static int ext4_splice_branch(handle_t *handle, struct inode *inode,
847 848
			      ext4_lblk_t block, Indirect *where, int num,
			      int blks)
849 850 851
{
	int i;
	int err = 0;
852
	ext4_fsblk_t current_block;
853 854 855 856 857 858 859 860

	/*
	 * If we're splicing into a [td]indirect block (as opposed to the
	 * inode) then we need to get write access to the [td]indirect block
	 * before the splice.
	 */
	if (where->bh) {
		BUFFER_TRACE(where->bh, "get_write_access");
861
		err = ext4_journal_get_write_access(handle, where->bh);
862 863 864 865 866 867 868 869 870 871 872 873 874 875
		if (err)
			goto err_out;
	}
	/* That's it */

	*where->p = where->key;

	/*
	 * Update the host buffer_head or inode to point to more just allocated
	 * direct blocks blocks
	 */
	if (num == 0 && blks > 1) {
		current_block = le32_to_cpu(where->key) + 1;
		for (i = 1; i < blks; i++)
876
			*(where->p + i) = cpu_to_le32(current_block++);
877 878 879 880 881 882 883 884 885 886 887
	}

	/* We are done with atomic stuff, now do the rest of housekeeping */
	/* had we spliced it onto indirect block? */
	if (where->bh) {
		/*
		 * If we spliced it onto an indirect block, we haven't
		 * altered the inode.  Note however that if it is being spliced
		 * onto an indirect block at the very end of the file (the
		 * file is growing) then we *will* alter the inode to reflect
		 * the new i_size.  But that is not done here - it is done in
888
		 * generic_commit_write->__mark_inode_dirty->ext4_dirty_inode.
889 890
		 */
		jbd_debug(5, "splicing indirect only\n");
891 892
		BUFFER_TRACE(where->bh, "call ext4_handle_dirty_metadata");
		err = ext4_handle_dirty_metadata(handle, inode, where->bh);
893 894 895 896 897 898
		if (err)
			goto err_out;
	} else {
		/*
		 * OK, we spliced it into the inode itself on a direct block.
		 */
899
		ext4_mark_inode_dirty(handle, inode);
900 901 902 903 904 905
		jbd_debug(5, "splicing direct\n");
	}
	return err;

err_out:
	for (i = 1; i <= num; i++) {
906
		BUFFER_TRACE(where[i].bh, "call jbd2_journal_forget");
907
		ext4_journal_forget(handle, where[i].bh);
908 909
		ext4_free_blocks(handle, inode,
					le32_to_cpu(where[i-1].key), 1, 0);
910
	}
911
	ext4_free_blocks(handle, inode, le32_to_cpu(where[num].key), blks, 0);
912 913 914 915 916

	return err;
}

/*
917 918 919 920
 * The ext4_ind_get_blocks() function handles non-extents inodes
 * (i.e., using the traditional indirect/double-indirect i_blocks
 * scheme) for ext4_get_blocks().
 *
921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936
 * Allocation strategy is simple: if we have to allocate something, we will
 * have to go the whole way to leaf. So let's do it before attaching anything
 * to tree, set linkage between the newborn blocks, write them if sync is
 * required, recheck the path, free and repeat if check fails, otherwise
 * set the last missing link (that will protect us from any truncate-generated
 * removals - all blocks on the path are immune now) and possibly force the
 * write on the parent block.
 * That has a nice additional property: no special recovery from the failed
 * allocations is needed - we simply release blocks and do not touch anything
 * reachable from inode.
 *
 * `handle' can be NULL if create == 0.
 *
 * return > 0, # of blocks mapped or allocated.
 * return = 0, if plain lookup failed.
 * return < 0, error case.
937
 *
938 939 940 941 942
 * The ext4_ind_get_blocks() function should be called with
 * down_write(&EXT4_I(inode)->i_data_sem) if allocating filesystem
 * blocks (i.e., flags has EXT4_GET_BLOCKS_CREATE set) or
 * down_read(&EXT4_I(inode)->i_data_sem) if not allocating file system
 * blocks.
943
 */
944
static int ext4_ind_get_blocks(handle_t *handle, struct inode *inode,
945 946 947
			       ext4_lblk_t iblock, unsigned int maxblocks,
			       struct buffer_head *bh_result,
			       int flags)
948 949
{
	int err = -EIO;
A
Aneesh Kumar K.V 已提交
950
	ext4_lblk_t offsets[4];
951 952
	Indirect chain[4];
	Indirect *partial;
953
	ext4_fsblk_t goal;
954 955 956 957
	int indirect_blks;
	int blocks_to_boundary = 0;
	int depth;
	int count = 0;
958
	ext4_fsblk_t first_block = 0;
959

A
Alex Tomas 已提交
960
	J_ASSERT(!(EXT4_I(inode)->i_flags & EXT4_EXTENTS_FL));
961
	J_ASSERT(handle != NULL || (flags & EXT4_GET_BLOCKS_CREATE) == 0);
A
Aneesh Kumar K.V 已提交
962
	depth = ext4_block_to_path(inode, iblock, offsets,
963
				   &blocks_to_boundary);
964 965 966 967

	if (depth == 0)
		goto out;

968
	partial = ext4_get_branch(inode, depth, offsets, chain, &err);
969 970 971 972 973 974 975 976

	/* Simplest case - block found, no allocation needed */
	if (!partial) {
		first_block = le32_to_cpu(chain[depth - 1].key);
		clear_buffer_new(bh_result);
		count++;
		/*map more blocks*/
		while (count < maxblocks && count <= blocks_to_boundary) {
977
			ext4_fsblk_t blk;
978 979 980 981 982 983 984 985

			blk = le32_to_cpu(*(chain[depth-1].p + count));

			if (blk == first_block + count)
				count++;
			else
				break;
		}
986
		goto got_it;
987 988 989
	}

	/* Next simple case - plain lookup or failed read of indirect block */
990
	if ((flags & EXT4_GET_BLOCKS_CREATE) == 0 || err == -EIO)
991 992 993
		goto cleanup;

	/*
994
	 * Okay, we need to do block allocation.
995
	*/
996
	goal = ext4_find_goal(inode, iblock, partial);
997 998 999 1000 1001 1002 1003 1004

	/* the number of blocks need to allocate for [d,t]indirect blocks */
	indirect_blks = (chain + depth) - partial - 1;

	/*
	 * Next look up the indirect map to count the totoal number of
	 * direct blocks to allocate for this branch.
	 */
1005
	count = ext4_blks_to_allocate(partial, indirect_blks,
1006 1007
					maxblocks, blocks_to_boundary);
	/*
1008
	 * Block out ext4_truncate while we alter the tree
1009
	 */
1010
	err = ext4_alloc_branch(handle, inode, iblock, indirect_blks,
1011 1012
				&count, goal,
				offsets + (partial - chain), partial);
1013 1014

	/*
1015
	 * The ext4_splice_branch call will free and forget any buffers
1016 1017 1018 1019 1020 1021
	 * on the new chain if there is a failure, but that risks using
	 * up transaction credits, especially for bitmaps where the
	 * credits cannot be returned.  Can we handle this somehow?  We
	 * may need to return -EAGAIN upwards in the worst case.  --sct
	 */
	if (!err)
1022
		err = ext4_splice_branch(handle, inode, iblock,
1023 1024
					 partial, indirect_blks, count);
	else
1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045
		goto cleanup;

	set_buffer_new(bh_result);
got_it:
	map_bh(bh_result, inode->i_sb, le32_to_cpu(chain[depth-1].key));
	if (count > blocks_to_boundary)
		set_buffer_boundary(bh_result);
	err = count;
	/* Clean up and exit */
	partial = chain + depth - 1;	/* the whole chain */
cleanup:
	while (partial > chain) {
		BUFFER_TRACE(partial->bh, "call brelse");
		brelse(partial->bh);
		partial--;
	}
	BUFFER_TRACE(bh_result, "returned");
out:
	return err;
}

1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056
qsize_t ext4_get_reserved_space(struct inode *inode)
{
	unsigned long long total;

	spin_lock(&EXT4_I(inode)->i_block_reservation_lock);
	total = EXT4_I(inode)->i_reserved_data_blocks +
		EXT4_I(inode)->i_reserved_meta_blocks;
	spin_unlock(&EXT4_I(inode)->i_block_reservation_lock);

	return total;
}
1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081
/*
 * Calculate the number of metadata blocks need to reserve
 * to allocate @blocks for non extent file based file
 */
static int ext4_indirect_calc_metadata_amount(struct inode *inode, int blocks)
{
	int icap = EXT4_ADDR_PER_BLOCK(inode->i_sb);
	int ind_blks, dind_blks, tind_blks;

	/* number of new indirect blocks needed */
	ind_blks = (blocks + icap - 1) / icap;

	dind_blks = (ind_blks + icap - 1) / icap;

	tind_blks = 1;

	return ind_blks + dind_blks + tind_blks;
}

/*
 * Calculate the number of metadata blocks need to reserve
 * to allocate given number of blocks
 */
static int ext4_calc_metadata_amount(struct inode *inode, int blocks)
{
1082 1083 1084
	if (!blocks)
		return 0;

1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104
	if (EXT4_I(inode)->i_flags & EXT4_EXTENTS_FL)
		return ext4_ext_calc_metadata_amount(inode, blocks);

	return ext4_indirect_calc_metadata_amount(inode, blocks);
}

static void ext4_da_update_reserve_space(struct inode *inode, int used)
{
	struct ext4_sb_info *sbi = EXT4_SB(inode->i_sb);
	int total, mdb, mdb_free;

	spin_lock(&EXT4_I(inode)->i_block_reservation_lock);
	/* recalculate the number of metablocks still need to be reserved */
	total = EXT4_I(inode)->i_reserved_data_blocks - used;
	mdb = ext4_calc_metadata_amount(inode, total);

	/* figure out how many metablocks to release */
	BUG_ON(mdb > EXT4_I(inode)->i_reserved_meta_blocks);
	mdb_free = EXT4_I(inode)->i_reserved_meta_blocks - mdb;

1105 1106 1107 1108 1109 1110 1111 1112 1113
	if (mdb_free) {
		/* Account for allocated meta_blocks */
		mdb_free -= EXT4_I(inode)->i_allocated_meta_blocks;

		/* update fs dirty blocks counter */
		percpu_counter_sub(&sbi->s_dirtyblocks_counter, mdb_free);
		EXT4_I(inode)->i_allocated_meta_blocks = 0;
		EXT4_I(inode)->i_reserved_meta_blocks = mdb;
	}
1114 1115 1116 1117 1118

	/* update per-inode reservations */
	BUG_ON(used  > EXT4_I(inode)->i_reserved_data_blocks);
	EXT4_I(inode)->i_reserved_data_blocks -= used;
	spin_unlock(&EXT4_I(inode)->i_block_reservation_lock);
1119 1120 1121 1122 1123 1124

	/*
	 * free those over-booking quota for metadata blocks
	 */
	if (mdb_free)
		vfs_dq_release_reservation_block(inode, mdb_free);
1125 1126 1127 1128 1129 1130 1131 1132

	/*
	 * 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.
	 */
	if (!total && (atomic_read(&inode->i_writecount) == 0))
		ext4_discard_preallocations(inode);
1133 1134
}

1135 1136
static int check_block_validity(struct inode *inode, const char *msg,
				sector_t logical, sector_t phys, int len)
1137 1138
{
	if (!ext4_data_block_valid(EXT4_SB(inode->i_sb), phys, len)) {
1139
		ext4_error(inode->i_sb, msg,
1140 1141 1142 1143 1144 1145 1146 1147 1148
			   "inode #%lu logical block %llu mapped to %llu "
			   "(size %d)", inode->i_ino,
			   (unsigned long long) logical,
			   (unsigned long long) phys, len);
		return -EIO;
	}
	return 0;
}

1149
/*
1150 1151
 * Return the number of contiguous dirty pages in a given inode
 * starting at page frame idx.
1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184
 */
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;
			}
1185 1186 1187 1188 1189 1190 1191 1192 1193
			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));
			}
1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206
			unlock_page(page);
			if (done)
				break;
			idx++;
			num++;
			if (num >= max_pages)
				break;
		}
		pagevec_release(&pvec);
	}
	return num;
}

1207
/*
1208
 * The ext4_get_blocks() function tries to look up the requested blocks,
1209
 * and returns if the blocks are already mapped.
1210 1211 1212 1213 1214 1215
 *
 * 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.
 *
 * If file type is extents based, it will call ext4_ext_get_blocks(),
1216
 * Otherwise, call with ext4_ind_get_blocks() to handle indirect mapping
1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228
 * 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.
 */
1229 1230
int ext4_get_blocks(handle_t *handle, struct inode *inode, sector_t block,
		    unsigned int max_blocks, struct buffer_head *bh,
1231
		    int flags)
1232 1233
{
	int retval;
1234 1235

	clear_buffer_mapped(bh);
1236
	clear_buffer_unwritten(bh);
1237

1238 1239 1240
	ext_debug("ext4_get_blocks(): inode %lu, flag %d, max_blocks %u,"
		  "logical block %lu\n", inode->i_ino, flags, max_blocks,
		  (unsigned long)block);
1241
	/*
1242 1243
	 * Try to see if we can get the block without requesting a new
	 * file system block.
1244 1245 1246 1247
	 */
	down_read((&EXT4_I(inode)->i_data_sem));
	if (EXT4_I(inode)->i_flags & EXT4_EXTENTS_FL) {
		retval =  ext4_ext_get_blocks(handle, inode, block, max_blocks,
1248
				bh, 0);
1249
	} else {
1250
		retval = ext4_ind_get_blocks(handle, inode, block, max_blocks,
1251
					     bh, 0);
1252
	}
1253
	up_read((&EXT4_I(inode)->i_data_sem));
1254

1255
	if (retval > 0 && buffer_mapped(bh)) {
1256 1257
		int ret = check_block_validity(inode, "file system corruption",
					       block, bh->b_blocknr, retval);
1258 1259 1260 1261
		if (ret != 0)
			return ret;
	}

1262
	/* If it is only a block(s) look up */
1263
	if ((flags & EXT4_GET_BLOCKS_CREATE) == 0)
1264 1265 1266 1267 1268 1269 1270 1271 1272 1273
		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.
	 */
	if (retval > 0 && buffer_mapped(bh))
1274 1275
		return retval;

1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287
	/*
	 * 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.
	 */
	clear_buffer_unwritten(bh);

1288
	/*
1289 1290 1291 1292
	 * 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.
1293 1294
	 */
	down_write((&EXT4_I(inode)->i_data_sem));
1295 1296 1297 1298 1299 1300 1301

	/*
	 * 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
	 */
1302
	if (flags & EXT4_GET_BLOCKS_DELALLOC_RESERVE)
1303
		EXT4_I(inode)->i_delalloc_reserved_flag = 1;
1304 1305 1306 1307
	/*
	 * We need to check for EXT4 here because migrate
	 * could have changed the inode type in between
	 */
1308 1309
	if (EXT4_I(inode)->i_flags & EXT4_EXTENTS_FL) {
		retval =  ext4_ext_get_blocks(handle, inode, block, max_blocks,
1310
					      bh, flags);
1311
	} else {
1312
		retval = ext4_ind_get_blocks(handle, inode, block,
1313
					     max_blocks, bh, flags);
1314 1315 1316 1317 1318 1319 1320

		if (retval > 0 && buffer_new(bh)) {
			/*
			 * We allocated new blocks which will result in
			 * i_data's format changing.  Force the migrate
			 * to fail by clearing migrate flags
			 */
1321
			EXT4_I(inode)->i_state &= ~EXT4_STATE_EXT_MIGRATE;
1322
		}
1323
	}
1324

1325
	if (flags & EXT4_GET_BLOCKS_DELALLOC_RESERVE)
1326
		EXT4_I(inode)->i_delalloc_reserved_flag = 0;
1327 1328 1329 1330 1331 1332 1333

	/*
	 * Update reserved blocks/metadata blocks after successful
	 * block allocation which had been deferred till now.
	 */
	if ((retval > 0) && (flags & EXT4_GET_BLOCKS_UPDATE_RESERVE_SPACE))
		ext4_da_update_reserve_space(inode, retval);
1334

1335
	up_write((&EXT4_I(inode)->i_data_sem));
1336
	if (retval > 0 && buffer_mapped(bh)) {
1337 1338 1339
		int ret = check_block_validity(inode, "file system "
					       "corruption after allocation",
					       block, bh->b_blocknr, retval);
1340 1341 1342
		if (ret != 0)
			return ret;
	}
1343 1344 1345
	return retval;
}

1346 1347 1348
/* Maximum number of blocks we map for direct IO at once. */
#define DIO_MAX_BLOCKS 4096

1349 1350
int ext4_get_block(struct inode *inode, sector_t iblock,
		   struct buffer_head *bh_result, int create)
1351
{
1352
	handle_t *handle = ext4_journal_current_handle();
J
Jan Kara 已提交
1353
	int ret = 0, started = 0;
1354
	unsigned max_blocks = bh_result->b_size >> inode->i_blkbits;
1355
	int dio_credits;
1356

J
Jan Kara 已提交
1357 1358 1359 1360
	if (create && !handle) {
		/* Direct IO write... */
		if (max_blocks > DIO_MAX_BLOCKS)
			max_blocks = DIO_MAX_BLOCKS;
1361 1362
		dio_credits = ext4_chunk_trans_blocks(inode, max_blocks);
		handle = ext4_journal_start(inode, dio_credits);
J
Jan Kara 已提交
1363
		if (IS_ERR(handle)) {
1364
			ret = PTR_ERR(handle);
J
Jan Kara 已提交
1365
			goto out;
1366
		}
J
Jan Kara 已提交
1367
		started = 1;
1368 1369
	}

1370
	ret = ext4_get_blocks(handle, inode, iblock, max_blocks, bh_result,
1371
			      create ? EXT4_GET_BLOCKS_CREATE : 0);
J
Jan Kara 已提交
1372 1373 1374
	if (ret > 0) {
		bh_result->b_size = (ret << inode->i_blkbits);
		ret = 0;
1375
	}
J
Jan Kara 已提交
1376 1377 1378
	if (started)
		ext4_journal_stop(handle);
out:
1379 1380 1381 1382 1383 1384
	return ret;
}

/*
 * `handle' can be NULL if create is zero
 */
1385
struct buffer_head *ext4_getblk(handle_t *handle, struct inode *inode,
A
Aneesh Kumar K.V 已提交
1386
				ext4_lblk_t block, int create, int *errp)
1387 1388 1389
{
	struct buffer_head dummy;
	int fatal = 0, err;
1390
	int flags = 0;
1391 1392 1393 1394 1395 1396

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

	dummy.b_state = 0;
	dummy.b_blocknr = -1000;
	buffer_trace_init(&dummy.b_history);
1397 1398 1399
	if (create)
		flags |= EXT4_GET_BLOCKS_CREATE;
	err = ext4_get_blocks(handle, inode, block, 1, &dummy, flags);
1400
	/*
1401 1402
	 * ext4_get_blocks() returns number of blocks mapped. 0 in
	 * case of a HOLE.
1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413 1414 1415 1416 1417 1418
	 */
	if (err > 0) {
		if (err > 1)
			WARN_ON(1);
		err = 0;
	}
	*errp = err;
	if (!err && buffer_mapped(&dummy)) {
		struct buffer_head *bh;
		bh = sb_getblk(inode->i_sb, dummy.b_blocknr);
		if (!bh) {
			*errp = -EIO;
			goto err;
		}
		if (buffer_new(&dummy)) {
			J_ASSERT(create != 0);
A
Aneesh Kumar K.V 已提交
1419
			J_ASSERT(handle != NULL);
1420 1421 1422 1423 1424

			/*
			 * 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
1425
			 * writes use ext4_get_block instead, so it's not a
1426 1427 1428 1429
			 * problem.
			 */
			lock_buffer(bh);
			BUFFER_TRACE(bh, "call get_create_access");
1430
			fatal = ext4_journal_get_create_access(handle, bh);
1431
			if (!fatal && !buffer_uptodate(bh)) {
1432
				memset(bh->b_data, 0, inode->i_sb->s_blocksize);
1433 1434 1435
				set_buffer_uptodate(bh);
			}
			unlock_buffer(bh);
1436 1437
			BUFFER_TRACE(bh, "call ext4_handle_dirty_metadata");
			err = ext4_handle_dirty_metadata(handle, inode, bh);
1438 1439 1440 1441 1442 1443 1444 1445 1446 1447 1448 1449 1450 1451 1452 1453
			if (!fatal)
				fatal = err;
		} else {
			BUFFER_TRACE(bh, "not a new buffer");
		}
		if (fatal) {
			*errp = fatal;
			brelse(bh);
			bh = NULL;
		}
		return bh;
	}
err:
	return NULL;
}

1454
struct buffer_head *ext4_bread(handle_t *handle, struct inode *inode,
A
Aneesh Kumar K.V 已提交
1455
			       ext4_lblk_t block, int create, int *err)
1456
{
1457
	struct buffer_head *bh;
1458

1459
	bh = ext4_getblk(handle, inode, block, create, err);
1460 1461 1462 1463 1464 1465 1466 1467 1468 1469 1470 1471 1472
	if (!bh)
		return bh;
	if (buffer_uptodate(bh))
		return bh;
	ll_rw_block(READ_META, 1, &bh);
	wait_on_buffer(bh);
	if (buffer_uptodate(bh))
		return bh;
	put_bh(bh);
	*err = -EIO;
	return NULL;
}

1473 1474 1475 1476 1477 1478 1479
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))
1480 1481 1482 1483 1484 1485 1486
{
	struct buffer_head *bh;
	unsigned block_start, block_end;
	unsigned blocksize = head->b_size;
	int err, ret = 0;
	struct buffer_head *next;

1487 1488
	for (bh = head, block_start = 0;
	     ret == 0 && (bh != head || !block_start);
1489
	     block_start = block_end, bh = next) {
1490 1491 1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506
		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
1507
 * close off a transaction and start a new one between the ext4_get_block()
1508
 * and the commit_write().  So doing the jbd2_journal_start at the start of
1509 1510
 * prepare_write() is the right place.
 *
1511 1512
 * Also, this function can nest inside ext4_writepage() ->
 * block_write_full_page(). In that case, we *know* that ext4_writepage()
1513 1514 1515 1516
 * 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.
 *
1517
 * By accident, ext4 can be reentered when a transaction is open via
1518 1519 1520 1521 1522 1523
 * 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.
 *
1524
 * So what we do is to rely on the fact that jbd2_journal_stop/journal_start
1525 1526 1527 1528 1529
 * 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,
1530
				       struct buffer_head *bh)
1531 1532 1533
{
	if (!buffer_mapped(bh) || buffer_freed(bh))
		return 0;
1534
	return ext4_journal_get_write_access(handle, bh);
1535 1536
}

N
Nick Piggin 已提交
1537
static int ext4_write_begin(struct file *file, struct address_space *mapping,
1538 1539
			    loff_t pos, unsigned len, unsigned flags,
			    struct page **pagep, void **fsdata)
1540
{
1541
	struct inode *inode = mapping->host;
1542
	int ret, needed_blocks;
1543 1544
	handle_t *handle;
	int retries = 0;
1545
	struct page *page;
1546
	pgoff_t index;
1547
	unsigned from, to;
N
Nick Piggin 已提交
1548

1549
	trace_ext4_write_begin(inode, pos, len, flags);
1550 1551 1552 1553 1554
	/*
	 * 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;
1555
	index = pos >> PAGE_CACHE_SHIFT;
1556 1557
	from = pos & (PAGE_CACHE_SIZE - 1);
	to = from + len;
1558 1559

retry:
1560 1561 1562 1563
	handle = ext4_journal_start(inode, needed_blocks);
	if (IS_ERR(handle)) {
		ret = PTR_ERR(handle);
		goto out;
1564
	}
1565

1566 1567 1568 1569
	/* We cannot recurse into the filesystem as the transaction is already
	 * started */
	flags |= AOP_FLAG_NOFS;

1570
	page = grab_cache_page_write_begin(mapping, index, flags);
1571 1572 1573 1574 1575 1576 1577
	if (!page) {
		ext4_journal_stop(handle);
		ret = -ENOMEM;
		goto out;
	}
	*pagep = page;

N
Nick Piggin 已提交
1578
	ret = block_write_begin(file, mapping, pos, len, flags, pagep, fsdata,
1579
				ext4_get_block);
N
Nick Piggin 已提交
1580 1581

	if (!ret && ext4_should_journal_data(inode)) {
1582 1583 1584
		ret = walk_page_buffers(handle, page_buffers(page),
				from, to, NULL, do_journal_get_write_access);
	}
N
Nick Piggin 已提交
1585 1586

	if (ret) {
1587 1588
		unlock_page(page);
		page_cache_release(page);
1589 1590 1591 1592
		/*
		 * block_write_begin may have instantiated a few blocks
		 * outside i_size.  Trim these off again. Don't need
		 * i_size_read because we hold i_mutex.
1593 1594 1595
		 *
		 * Add inode to orphan list in case we crash before
		 * truncate finishes
1596
		 */
1597
		if (pos + len > inode->i_size && ext4_can_truncate(inode))
1598 1599 1600 1601
			ext4_orphan_add(handle, inode);

		ext4_journal_stop(handle);
		if (pos + len > inode->i_size) {
1602
			ext4_truncate(inode);
1603
			/*
1604
			 * If truncate failed early the inode might
1605 1606 1607 1608 1609 1610 1611
			 * 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);
		}
N
Nick Piggin 已提交
1612 1613
	}

1614
	if (ret == -ENOSPC && ext4_should_retry_alloc(inode->i_sb, &retries))
1615
		goto retry;
1616
out:
1617 1618 1619
	return ret;
}

N
Nick Piggin 已提交
1620 1621
/* For write_end() in data=journal mode */
static int write_end_fn(handle_t *handle, struct buffer_head *bh)
1622 1623 1624 1625
{
	if (!buffer_mapped(bh) || buffer_freed(bh))
		return 0;
	set_buffer_uptodate(bh);
1626
	return ext4_handle_dirty_metadata(handle, NULL, bh);
1627 1628
}

1629
static int ext4_generic_write_end(struct file *file,
1630 1631 1632
				  struct address_space *mapping,
				  loff_t pos, unsigned len, unsigned copied,
				  struct page *page, void *fsdata)
1633 1634 1635 1636 1637 1638 1639 1640 1641 1642 1643 1644 1645 1646 1647 1648 1649 1650 1651 1652 1653 1654 1655 1656 1657 1658 1659 1660 1661 1662 1663 1664 1665 1666 1667 1668 1669 1670 1671 1672 1673 1674
{
	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;
}

1675 1676 1677 1678
/*
 * 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().
 *
1679
 * ext4 never places buffers on inode->i_mapping->private_list.  metadata
1680 1681
 * buffers are managed internally.
 */
N
Nick Piggin 已提交
1682
static int ext4_ordered_write_end(struct file *file,
1683 1684 1685
				  struct address_space *mapping,
				  loff_t pos, unsigned len, unsigned copied,
				  struct page *page, void *fsdata)
1686
{
1687
	handle_t *handle = ext4_journal_current_handle();
1688
	struct inode *inode = mapping->host;
1689 1690
	int ret = 0, ret2;

1691
	trace_ext4_ordered_write_end(inode, pos, len, copied);
1692
	ret = ext4_jbd2_file_inode(handle, inode);
1693 1694

	if (ret == 0) {
1695
		ret2 = ext4_generic_write_end(file, mapping, pos, len, copied,
N
Nick Piggin 已提交
1696
							page, fsdata);
1697
		copied = ret2;
1698
		if (pos + len > inode->i_size && ext4_can_truncate(inode))
1699 1700 1701 1702 1703
			/* 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);
1704 1705
		if (ret2 < 0)
			ret = ret2;
1706
	}
1707
	ret2 = ext4_journal_stop(handle);
1708 1709
	if (!ret)
		ret = ret2;
N
Nick Piggin 已提交
1710

1711
	if (pos + len > inode->i_size) {
1712
		ext4_truncate(inode);
1713
		/*
1714
		 * If truncate failed early the inode might still be
1715 1716 1717 1718 1719 1720 1721 1722
		 * 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);
	}


N
Nick Piggin 已提交
1723
	return ret ? ret : copied;
1724 1725
}

N
Nick Piggin 已提交
1726
static int ext4_writeback_write_end(struct file *file,
1727 1728 1729
				    struct address_space *mapping,
				    loff_t pos, unsigned len, unsigned copied,
				    struct page *page, void *fsdata)
1730
{
1731
	handle_t *handle = ext4_journal_current_handle();
1732
	struct inode *inode = mapping->host;
1733 1734
	int ret = 0, ret2;

1735
	trace_ext4_writeback_write_end(inode, pos, len, copied);
1736
	ret2 = ext4_generic_write_end(file, mapping, pos, len, copied,
N
Nick Piggin 已提交
1737
							page, fsdata);
1738
	copied = ret2;
1739
	if (pos + len > inode->i_size && ext4_can_truncate(inode))
1740 1741 1742 1743 1744 1745
		/* 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);

1746 1747
	if (ret2 < 0)
		ret = ret2;
1748

1749
	ret2 = ext4_journal_stop(handle);
1750 1751
	if (!ret)
		ret = ret2;
N
Nick Piggin 已提交
1752

1753
	if (pos + len > inode->i_size) {
1754
		ext4_truncate(inode);
1755
		/*
1756
		 * If truncate failed early the inode might still be
1757 1758 1759 1760 1761 1762 1763
		 * 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);
	}

N
Nick Piggin 已提交
1764
	return ret ? ret : copied;
1765 1766
}

N
Nick Piggin 已提交
1767
static int ext4_journalled_write_end(struct file *file,
1768 1769 1770
				     struct address_space *mapping,
				     loff_t pos, unsigned len, unsigned copied,
				     struct page *page, void *fsdata)
1771
{
1772
	handle_t *handle = ext4_journal_current_handle();
N
Nick Piggin 已提交
1773
	struct inode *inode = mapping->host;
1774 1775
	int ret = 0, ret2;
	int partial = 0;
N
Nick Piggin 已提交
1776
	unsigned from, to;
1777
	loff_t new_i_size;
1778

1779
	trace_ext4_journalled_write_end(inode, pos, len, copied);
N
Nick Piggin 已提交
1780 1781 1782 1783 1784 1785 1786 1787
	from = pos & (PAGE_CACHE_SIZE - 1);
	to = from + len;

	if (copied < len) {
		if (!PageUptodate(page))
			copied = 0;
		page_zero_new_buffers(page, from+copied, to);
	}
1788 1789

	ret = walk_page_buffers(handle, page_buffers(page), from,
N
Nick Piggin 已提交
1790
				to, &partial, write_end_fn);
1791 1792
	if (!partial)
		SetPageUptodate(page);
1793 1794
	new_i_size = pos + copied;
	if (new_i_size > inode->i_size)
N
Nick Piggin 已提交
1795
		i_size_write(inode, pos+copied);
1796
	EXT4_I(inode)->i_state |= EXT4_STATE_JDATA;
1797 1798
	if (new_i_size > EXT4_I(inode)->i_disksize) {
		ext4_update_i_disksize(inode, new_i_size);
1799
		ret2 = ext4_mark_inode_dirty(handle, inode);
1800 1801 1802
		if (!ret)
			ret = ret2;
	}
N
Nick Piggin 已提交
1803

1804
	unlock_page(page);
1805
	page_cache_release(page);
1806
	if (pos + len > inode->i_size && ext4_can_truncate(inode))
1807 1808 1809 1810 1811 1812
		/* 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);

1813
	ret2 = ext4_journal_stop(handle);
1814 1815
	if (!ret)
		ret = ret2;
1816
	if (pos + len > inode->i_size) {
1817
		ext4_truncate(inode);
1818
		/*
1819
		 * If truncate failed early the inode might still be
1820 1821 1822 1823 1824 1825
		 * 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);
	}
N
Nick Piggin 已提交
1826 1827

	return ret ? ret : copied;
1828
}
1829 1830 1831

static int ext4_da_reserve_space(struct inode *inode, int nrblocks)
{
A
Aneesh Kumar K.V 已提交
1832
	int retries = 0;
1833 1834
	struct ext4_sb_info *sbi = EXT4_SB(inode->i_sb);
	unsigned long md_needed, mdblocks, total = 0;
1835 1836 1837 1838 1839 1840

	/*
	 * recalculate the amount of metadata blocks to reserve
	 * in order to allocate nrblocks
	 * worse case is one extent per block
	 */
A
Aneesh Kumar K.V 已提交
1841
repeat:
1842 1843 1844 1845 1846 1847 1848 1849
	spin_lock(&EXT4_I(inode)->i_block_reservation_lock);
	total = EXT4_I(inode)->i_reserved_data_blocks + nrblocks;
	mdblocks = ext4_calc_metadata_amount(inode, total);
	BUG_ON(mdblocks < EXT4_I(inode)->i_reserved_meta_blocks);

	md_needed = mdblocks - EXT4_I(inode)->i_reserved_meta_blocks;
	total = md_needed + nrblocks;

1850 1851 1852 1853 1854 1855 1856 1857 1858 1859
	/*
	 * Make quota reservation here to prevent quota overflow
	 * later. Real quota accounting is done at pages writeout
	 * time.
	 */
	if (vfs_dq_reserve_block(inode, total)) {
		spin_unlock(&EXT4_I(inode)->i_block_reservation_lock);
		return -EDQUOT;
	}

1860
	if (ext4_claim_free_blocks(sbi, total)) {
1861
		spin_unlock(&EXT4_I(inode)->i_block_reservation_lock);
1862
		vfs_dq_release_reservation_block(inode, total);
A
Aneesh Kumar K.V 已提交
1863 1864 1865 1866
		if (ext4_should_retry_alloc(inode->i_sb, &retries)) {
			yield();
			goto repeat;
		}
1867 1868 1869 1870 1871 1872 1873 1874 1875
		return -ENOSPC;
	}
	EXT4_I(inode)->i_reserved_data_blocks += nrblocks;
	EXT4_I(inode)->i_reserved_meta_blocks = mdblocks;

	spin_unlock(&EXT4_I(inode)->i_block_reservation_lock);
	return 0;       /* success */
}

1876
static void ext4_da_release_space(struct inode *inode, int to_free)
1877 1878 1879 1880
{
	struct ext4_sb_info *sbi = EXT4_SB(inode->i_sb);
	int total, mdb, mdb_free, release;

1881 1882 1883
	if (!to_free)
		return;		/* Nothing to release, exit */

1884
	spin_lock(&EXT4_I(inode)->i_block_reservation_lock);
1885 1886 1887 1888 1889 1890 1891 1892 1893 1894 1895 1896 1897 1898 1899

	if (!EXT4_I(inode)->i_reserved_data_blocks) {
		/*
		 * if there is no reserved blocks, but we try to free some
		 * then the counter is messed up somewhere.
		 * but since this function is called from invalidate
		 * page, it's harmless to return without any action
		 */
		printk(KERN_INFO "ext4 delalloc try to release %d reserved "
			    "blocks for inode %lu, but there is no reserved "
			    "data blocks\n", to_free, inode->i_ino);
		spin_unlock(&EXT4_I(inode)->i_block_reservation_lock);
		return;
	}

1900
	/* recalculate the number of metablocks still need to be reserved */
1901
	total = EXT4_I(inode)->i_reserved_data_blocks - to_free;
1902 1903 1904 1905 1906 1907 1908 1909
	mdb = ext4_calc_metadata_amount(inode, total);

	/* figure out how many metablocks to release */
	BUG_ON(mdb > EXT4_I(inode)->i_reserved_meta_blocks);
	mdb_free = EXT4_I(inode)->i_reserved_meta_blocks - mdb;

	release = to_free + mdb_free;

1910 1911
	/* update fs dirty blocks counter for truncate case */
	percpu_counter_sub(&sbi->s_dirtyblocks_counter, release);
1912 1913

	/* update per-inode reservations */
1914 1915
	BUG_ON(to_free > EXT4_I(inode)->i_reserved_data_blocks);
	EXT4_I(inode)->i_reserved_data_blocks -= to_free;
1916 1917 1918 1919

	BUG_ON(mdb > EXT4_I(inode)->i_reserved_meta_blocks);
	EXT4_I(inode)->i_reserved_meta_blocks = mdb;
	spin_unlock(&EXT4_I(inode)->i_block_reservation_lock);
1920 1921

	vfs_dq_release_reservation_block(inode, release);
1922 1923 1924
}

static void ext4_da_page_release_reservation(struct page *page,
1925
					     unsigned long offset)
1926 1927 1928 1929 1930 1931 1932 1933 1934 1935 1936 1937 1938 1939 1940 1941
{
	int to_release = 0;
	struct buffer_head *head, *bh;
	unsigned int curr_off = 0;

	head = page_buffers(page);
	bh = head;
	do {
		unsigned int next_off = curr_off + bh->b_size;

		if ((offset <= curr_off) && (buffer_delay(bh))) {
			to_release++;
			clear_buffer_delay(bh);
		}
		curr_off = next_off;
	} while ((bh = bh->b_this_page) != head);
1942
	ext4_da_release_space(page->mapping->host, to_release);
1943
}
1944

1945 1946 1947 1948 1949 1950
/*
 * Delayed allocation stuff
 */

/*
 * mpage_da_submit_io - walks through extent of pages and try to write
1951
 * them with writepage() call back
1952 1953 1954 1955 1956 1957 1958 1959 1960 1961 1962 1963
 *
 * @mpd->inode: inode
 * @mpd->first_page: first page of the extent
 * @mpd->next_page: page after the last page of the extent
 *
 * By the time mpage_da_submit_io() is called we expect all blocks
 * to be allocated. this may be wrong if allocation failed.
 *
 * As pages are already locked by write_cache_pages(), we can't use it
 */
static int mpage_da_submit_io(struct mpage_da_data *mpd)
{
1964
	long pages_skipped;
1965 1966 1967 1968 1969
	struct pagevec pvec;
	unsigned long index, end;
	int ret = 0, err, nr_pages, i;
	struct inode *inode = mpd->inode;
	struct address_space *mapping = inode->i_mapping;
1970 1971

	BUG_ON(mpd->next_page <= mpd->first_page);
1972 1973 1974
	/*
	 * We need to start from the first_page to the next_page - 1
	 * to make sure we also write the mapped dirty buffer_heads.
1975
	 * If we look at mpd->b_blocknr we would only be looking
1976 1977
	 * at the currently mapped buffer_heads.
	 */
1978 1979 1980
	index = mpd->first_page;
	end = mpd->next_page - 1;

1981
	pagevec_init(&pvec, 0);
1982
	while (index <= end) {
1983
		nr_pages = pagevec_lookup(&pvec, mapping, index, PAGEVEC_SIZE);
1984 1985 1986 1987 1988
		if (nr_pages == 0)
			break;
		for (i = 0; i < nr_pages; i++) {
			struct page *page = pvec.pages[i];

1989 1990 1991 1992 1993 1994 1995 1996
			index = page->index;
			if (index > end)
				break;
			index++;

			BUG_ON(!PageLocked(page));
			BUG_ON(PageWriteback(page));

1997
			pages_skipped = mpd->wbc->pages_skipped;
1998
			err = mapping->a_ops->writepage(page, mpd->wbc);
1999 2000 2001 2002 2003
			if (!err && (pages_skipped == mpd->wbc->pages_skipped))
				/*
				 * have successfully written the page
				 * without skipping the same
				 */
2004
				mpd->pages_written++;
2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026
			/*
			 * In error case, we have to continue because
			 * remaining pages are still locked
			 * XXX: unlock and re-dirty them?
			 */
			if (ret == 0)
				ret = err;
		}
		pagevec_release(&pvec);
	}
	return ret;
}

/*
 * mpage_put_bnr_to_bhs - walk blocks and assign them actual numbers
 *
 * @mpd->inode - inode to walk through
 * @exbh->b_blocknr - first block on a disk
 * @exbh->b_size - amount of space in bytes
 * @logical - first logical block to start assignment with
 *
 * the function goes through all passed space and put actual disk
2027
 * block numbers into buffer heads, dropping BH_Delay and BH_Unwritten
2028 2029 2030 2031 2032 2033 2034 2035 2036
 */
static void mpage_put_bnr_to_bhs(struct mpage_da_data *mpd, sector_t logical,
				 struct buffer_head *exbh)
{
	struct inode *inode = mpd->inode;
	struct address_space *mapping = inode->i_mapping;
	int blocks = exbh->b_size >> inode->i_blkbits;
	sector_t pblock = exbh->b_blocknr, cur_logical;
	struct buffer_head *head, *bh;
2037
	pgoff_t index, end;
2038 2039 2040 2041 2042 2043 2044 2045 2046 2047 2048 2049 2050 2051 2052 2053 2054 2055 2056 2057 2058 2059 2060 2061 2062 2063 2064 2065 2066 2067 2068 2069 2070 2071 2072 2073 2074 2075 2076
	struct pagevec pvec;
	int nr_pages, i;

	index = logical >> (PAGE_CACHE_SHIFT - inode->i_blkbits);
	end = (logical + blocks - 1) >> (PAGE_CACHE_SHIFT - inode->i_blkbits);
	cur_logical = index << (PAGE_CACHE_SHIFT - inode->i_blkbits);

	pagevec_init(&pvec, 0);

	while (index <= end) {
		/* XXX: optimize tail */
		nr_pages = pagevec_lookup(&pvec, mapping, index, PAGEVEC_SIZE);
		if (nr_pages == 0)
			break;
		for (i = 0; i < nr_pages; i++) {
			struct page *page = pvec.pages[i];

			index = page->index;
			if (index > end)
				break;
			index++;

			BUG_ON(!PageLocked(page));
			BUG_ON(PageWriteback(page));
			BUG_ON(!page_has_buffers(page));

			bh = page_buffers(page);
			head = bh;

			/* skip blocks out of the range */
			do {
				if (cur_logical >= logical)
					break;
				cur_logical++;
			} while ((bh = bh->b_this_page) != head);

			do {
				if (cur_logical >= logical + blocks)
					break;
2077 2078 2079 2080 2081 2082 2083 2084 2085 2086 2087 2088 2089 2090 2091 2092 2093 2094

				if (buffer_delay(bh) ||
						buffer_unwritten(bh)) {

					BUG_ON(bh->b_bdev != inode->i_sb->s_bdev);

					if (buffer_delay(bh)) {
						clear_buffer_delay(bh);
						bh->b_blocknr = pblock;
					} else {
						/*
						 * unwritten already should have
						 * blocknr assigned. Verify that
						 */
						clear_buffer_unwritten(bh);
						BUG_ON(bh->b_blocknr != pblock);
					}

2095
				} else if (buffer_mapped(bh))
2096 2097 2098 2099 2100 2101 2102 2103 2104 2105 2106 2107 2108 2109 2110 2111 2112 2113 2114 2115 2116 2117 2118 2119 2120 2121
					BUG_ON(bh->b_blocknr != pblock);

				cur_logical++;
				pblock++;
			} while ((bh = bh->b_this_page) != head);
		}
		pagevec_release(&pvec);
	}
}


/*
 * __unmap_underlying_blocks - just a helper function to unmap
 * set of blocks described by @bh
 */
static inline void __unmap_underlying_blocks(struct inode *inode,
					     struct buffer_head *bh)
{
	struct block_device *bdev = inode->i_sb->s_bdev;
	int blocks, i;

	blocks = bh->b_size >> inode->i_blkbits;
	for (i = 0; i < blocks; i++)
		unmap_underlying_metadata(bdev, bh->b_blocknr + i);
}

2122 2123 2124 2125 2126 2127 2128 2129 2130 2131 2132 2133 2134 2135 2136 2137 2138 2139 2140 2141 2142 2143 2144 2145 2146 2147 2148 2149 2150 2151 2152 2153 2154
static void ext4_da_block_invalidatepages(struct mpage_da_data *mpd,
					sector_t logical, long blk_cnt)
{
	int nr_pages, i;
	pgoff_t index, end;
	struct pagevec pvec;
	struct inode *inode = mpd->inode;
	struct address_space *mapping = inode->i_mapping;

	index = logical >> (PAGE_CACHE_SHIFT - inode->i_blkbits);
	end   = (logical + blk_cnt - 1) >>
				(PAGE_CACHE_SHIFT - inode->i_blkbits);
	while (index <= end) {
		nr_pages = pagevec_lookup(&pvec, mapping, index, PAGEVEC_SIZE);
		if (nr_pages == 0)
			break;
		for (i = 0; i < nr_pages; i++) {
			struct page *page = pvec.pages[i];
			index = page->index;
			if (index > end)
				break;
			index++;

			BUG_ON(!PageLocked(page));
			BUG_ON(PageWriteback(page));
			block_invalidatepage(page, 0);
			ClearPageUptodate(page);
			unlock_page(page);
		}
	}
	return;
}

2155 2156 2157
static void ext4_print_free_blocks(struct inode *inode)
{
	struct ext4_sb_info *sbi = EXT4_SB(inode->i_sb);
2158 2159 2160 2161 2162 2163 2164 2165 2166 2167 2168 2169
	printk(KERN_CRIT "Total free blocks count %lld\n",
	       ext4_count_free_blocks(inode->i_sb));
	printk(KERN_CRIT "Free/Dirty block details\n");
	printk(KERN_CRIT "free_blocks=%lld\n",
	       (long long) percpu_counter_sum(&sbi->s_freeblocks_counter));
	printk(KERN_CRIT "dirty_blocks=%lld\n",
	       (long long) percpu_counter_sum(&sbi->s_dirtyblocks_counter));
	printk(KERN_CRIT "Block reservation details\n");
	printk(KERN_CRIT "i_reserved_data_blocks=%u\n",
	       EXT4_I(inode)->i_reserved_data_blocks);
	printk(KERN_CRIT "i_reserved_meta_blocks=%u\n",
	       EXT4_I(inode)->i_reserved_meta_blocks);
2170 2171 2172
	return;
}

2173 2174 2175
/*
 * mpage_da_map_blocks - go through given space
 *
2176
 * @mpd - bh describing space
2177 2178 2179 2180
 *
 * The function skips space we know is already mapped to disk blocks.
 *
 */
2181
static int mpage_da_map_blocks(struct mpage_da_data *mpd)
2182
{
2183
	int err, blks, get_blocks_flags;
A
Aneesh Kumar K.V 已提交
2184
	struct buffer_head new;
2185 2186 2187 2188
	sector_t next = mpd->b_blocknr;
	unsigned max_blocks = mpd->b_size >> mpd->inode->i_blkbits;
	loff_t disksize = EXT4_I(mpd->inode)->i_disksize;
	handle_t *handle = NULL;
2189 2190 2191 2192

	/*
	 * We consider only non-mapped and non-allocated blocks
	 */
2193
	if ((mpd->b_state  & (1 << BH_Mapped)) &&
2194 2195
		!(mpd->b_state & (1 << BH_Delay)) &&
		!(mpd->b_state & (1 << BH_Unwritten)))
2196
		return 0;
2197 2198 2199 2200 2201 2202 2203 2204 2205 2206

	/*
	 * If we didn't accumulate anything to write simply return
	 */
	if (!mpd->b_size)
		return 0;

	handle = ext4_journal_current_handle();
	BUG_ON(!handle);

2207
	/*
2208 2209 2210 2211 2212 2213 2214 2215 2216 2217 2218 2219 2220 2221 2222 2223
	 * Call ext4_get_blocks() to allocate any delayed allocation
	 * blocks, or to convert an uninitialized extent to be
	 * initialized (in the case where we have written into
	 * one or more preallocated blocks).
	 *
	 * We pass in the magic EXT4_GET_BLOCKS_DELALLOC_RESERVE to
	 * indicate that we are on the delayed allocation path.  This
	 * affects functions in many different parts of the allocation
	 * call path.  This flag exists primarily because we don't
	 * want to change *many* call functions, so ext4_get_blocks()
	 * will set the magic i_delalloc_reserved_flag once the
	 * inode's allocation semaphore is taken.
	 *
	 * If the blocks in questions were delalloc blocks, set
	 * EXT4_GET_BLOCKS_DELALLOC_RESERVE so the delalloc accounting
	 * variables are updated after the blocks have been allocated.
2224
	 */
2225 2226 2227 2228 2229
	new.b_state = 0;
	get_blocks_flags = (EXT4_GET_BLOCKS_CREATE |
			    EXT4_GET_BLOCKS_DELALLOC_RESERVE);
	if (mpd->b_state & (1 << BH_Delay))
		get_blocks_flags |= EXT4_GET_BLOCKS_UPDATE_RESERVE_SPACE;
2230
	blks = ext4_get_blocks(handle, mpd->inode, next, max_blocks,
2231
			       &new, get_blocks_flags);
2232 2233
	if (blks < 0) {
		err = blks;
2234 2235 2236 2237
		/*
		 * If get block returns with error we simply
		 * return. Later writepage will redirty the page and
		 * writepages will find the dirty page again
2238 2239 2240
		 */
		if (err == -EAGAIN)
			return 0;
2241 2242

		if (err == -ENOSPC &&
2243
		    ext4_count_free_blocks(mpd->inode->i_sb)) {
2244 2245 2246 2247
			mpd->retval = err;
			return 0;
		}

2248
		/*
2249 2250 2251 2252 2253
		 * get block failure will cause us to loop in
		 * writepages, because a_ops->writepage won't be able
		 * to make progress. The page will be redirtied by
		 * writepage and writepages will again try to write
		 * the same.
2254
		 */
2255 2256 2257 2258 2259 2260 2261 2262
		ext4_msg(mpd->inode->i_sb, KERN_CRIT,
			 "delayed block allocation failed for inode %lu at "
			 "logical offset %llu with max blocks %zd with "
			 "error %d\n", mpd->inode->i_ino,
			 (unsigned long long) next,
			 mpd->b_size >> mpd->inode->i_blkbits, err);
		printk(KERN_CRIT "This should not happen!!  "
		       "Data will be lost\n");
A
Aneesh Kumar K.V 已提交
2263
		if (err == -ENOSPC) {
2264
			ext4_print_free_blocks(mpd->inode);
A
Aneesh Kumar K.V 已提交
2265
		}
2266
		/* invalidate all the pages */
2267
		ext4_da_block_invalidatepages(mpd, next,
2268
				mpd->b_size >> mpd->inode->i_blkbits);
2269 2270
		return err;
	}
2271 2272 2273
	BUG_ON(blks == 0);

	new.b_size = (blks << mpd->inode->i_blkbits);
2274

2275 2276
	if (buffer_new(&new))
		__unmap_underlying_blocks(mpd->inode, &new);
2277

2278 2279 2280 2281
	/*
	 * If blocks are delayed marked, we need to
	 * put actual blocknr and drop delayed bit
	 */
2282 2283
	if ((mpd->b_state & (1 << BH_Delay)) ||
	    (mpd->b_state & (1 << BH_Unwritten)))
2284
		mpage_put_bnr_to_bhs(mpd, next, &new);
2285

2286 2287 2288 2289 2290 2291 2292
	if (ext4_should_order_data(mpd->inode)) {
		err = ext4_jbd2_file_inode(handle, mpd->inode);
		if (err)
			return err;
	}

	/*
2293
	 * Update on-disk size along with block allocation.
2294 2295 2296 2297 2298 2299 2300 2301 2302
	 */
	disksize = ((loff_t) next + blks) << mpd->inode->i_blkbits;
	if (disksize > i_size_read(mpd->inode))
		disksize = i_size_read(mpd->inode);
	if (disksize > EXT4_I(mpd->inode)->i_disksize) {
		ext4_update_i_disksize(mpd->inode, disksize);
		return ext4_mark_inode_dirty(handle, mpd->inode);
	}

2303
	return 0;
2304 2305
}

2306 2307
#define BH_FLAGS ((1 << BH_Uptodate) | (1 << BH_Mapped) | \
		(1 << BH_Delay) | (1 << BH_Unwritten))
2308 2309 2310 2311 2312 2313 2314 2315 2316 2317 2318

/*
 * mpage_add_bh_to_extent - try to add one more block to extent of blocks
 *
 * @mpd->lbh - extent of blocks
 * @logical - logical number of the block in the file
 * @bh - bh of the block (used to access block's state)
 *
 * the function is used to collect contig. blocks in same state
 */
static void mpage_add_bh_to_extent(struct mpage_da_data *mpd,
2319 2320
				   sector_t logical, size_t b_size,
				   unsigned long b_state)
2321 2322
{
	sector_t next;
2323
	int nrblocks = mpd->b_size >> mpd->inode->i_blkbits;
2324

2325 2326 2327 2328 2329 2330 2331 2332 2333 2334 2335 2336 2337 2338 2339 2340 2341 2342 2343 2344 2345 2346
	/* check if thereserved journal credits might overflow */
	if (!(EXT4_I(mpd->inode)->i_flags & EXT4_EXTENTS_FL)) {
		if (nrblocks >= EXT4_MAX_TRANS_DATA) {
			/*
			 * With non-extent format we are limited by the journal
			 * credit available.  Total credit needed to insert
			 * nrblocks contiguous blocks is dependent on the
			 * nrblocks.  So limit nrblocks.
			 */
			goto flush_it;
		} else if ((nrblocks + (b_size >> mpd->inode->i_blkbits)) >
				EXT4_MAX_TRANS_DATA) {
			/*
			 * Adding the new buffer_head would make it cross the
			 * allowed limit for which we have journal credit
			 * reserved. So limit the new bh->b_size
			 */
			b_size = (EXT4_MAX_TRANS_DATA - nrblocks) <<
						mpd->inode->i_blkbits;
			/* we will do mpage_da_submit_io in the next loop */
		}
	}
2347 2348 2349
	/*
	 * First block in the extent
	 */
2350 2351 2352 2353
	if (mpd->b_size == 0) {
		mpd->b_blocknr = logical;
		mpd->b_size = b_size;
		mpd->b_state = b_state & BH_FLAGS;
2354 2355 2356
		return;
	}

2357
	next = mpd->b_blocknr + nrblocks;
2358 2359 2360
	/*
	 * Can we merge the block to our big extent?
	 */
2361 2362
	if (logical == next && (b_state & BH_FLAGS) == mpd->b_state) {
		mpd->b_size += b_size;
2363 2364 2365
		return;
	}

2366
flush_it:
2367 2368 2369 2370
	/*
	 * We couldn't merge the block to our extent, so we
	 * need to flush current  extent and start new one
	 */
2371 2372
	if (mpage_da_map_blocks(mpd) == 0)
		mpage_da_submit_io(mpd);
2373 2374
	mpd->io_done = 1;
	return;
2375 2376
}

2377
static int ext4_bh_delay_or_unwritten(handle_t *handle, struct buffer_head *bh)
2378
{
2379
	return (buffer_delay(bh) || buffer_unwritten(bh)) && buffer_dirty(bh);
2380 2381
}

2382 2383 2384 2385 2386 2387 2388 2389 2390 2391 2392 2393 2394 2395
/*
 * __mpage_da_writepage - finds extent of pages and blocks
 *
 * @page: page to consider
 * @wbc: not used, we just follow rules
 * @data: context
 *
 * The function finds extents of pages and scan them for all blocks.
 */
static int __mpage_da_writepage(struct page *page,
				struct writeback_control *wbc, void *data)
{
	struct mpage_da_data *mpd = data;
	struct inode *inode = mpd->inode;
2396
	struct buffer_head *bh, *head;
2397 2398
	sector_t logical;

2399 2400 2401 2402
	if (mpd->io_done) {
		/*
		 * Rest of the page in the page_vec
		 * redirty then and skip then. We will
2403
		 * try to write them again after
2404 2405 2406 2407 2408 2409
		 * starting a new transaction
		 */
		redirty_page_for_writepage(wbc, page);
		unlock_page(page);
		return MPAGE_DA_EXTENT_TAIL;
	}
2410 2411 2412 2413 2414 2415
	/*
	 * Can we merge this page to current extent?
	 */
	if (mpd->next_page != page->index) {
		/*
		 * Nope, we can't. So, we map non-allocated blocks
2416
		 * and start IO on them using writepage()
2417 2418
		 */
		if (mpd->next_page != mpd->first_page) {
2419 2420
			if (mpage_da_map_blocks(mpd) == 0)
				mpage_da_submit_io(mpd);
2421 2422 2423 2424 2425 2426 2427
			/*
			 * skip rest of the page in the page_vec
			 */
			mpd->io_done = 1;
			redirty_page_for_writepage(wbc, page);
			unlock_page(page);
			return MPAGE_DA_EXTENT_TAIL;
2428 2429 2430 2431 2432 2433 2434 2435 2436 2437
		}

		/*
		 * Start next extent of pages ...
		 */
		mpd->first_page = page->index;

		/*
		 * ... and blocks
		 */
2438 2439 2440
		mpd->b_size = 0;
		mpd->b_state = 0;
		mpd->b_blocknr = 0;
2441 2442 2443 2444 2445 2446 2447
	}

	mpd->next_page = page->index + 1;
	logical = (sector_t) page->index <<
		  (PAGE_CACHE_SHIFT - inode->i_blkbits);

	if (!page_has_buffers(page)) {
2448 2449
		mpage_add_bh_to_extent(mpd, logical, PAGE_CACHE_SIZE,
				       (1 << BH_Dirty) | (1 << BH_Uptodate));
2450 2451
		if (mpd->io_done)
			return MPAGE_DA_EXTENT_TAIL;
2452 2453 2454 2455 2456 2457 2458 2459
	} else {
		/*
		 * Page with regular buffer heads, just add all dirty ones
		 */
		head = page_buffers(page);
		bh = head;
		do {
			BUG_ON(buffer_locked(bh));
2460 2461 2462 2463
			/*
			 * We need to try to allocate
			 * unmapped blocks in the same page.
			 * Otherwise we won't make progress
2464
			 * with the page in ext4_writepage
2465
			 */
2466
			if (ext4_bh_delay_or_unwritten(NULL, bh)) {
2467 2468 2469
				mpage_add_bh_to_extent(mpd, logical,
						       bh->b_size,
						       bh->b_state);
2470 2471
				if (mpd->io_done)
					return MPAGE_DA_EXTENT_TAIL;
2472 2473 2474 2475 2476 2477 2478 2479 2480
			} else if (buffer_dirty(bh) && (buffer_mapped(bh))) {
				/*
				 * mapped dirty buffer. We need to update
				 * the b_state because we look at
				 * b_state in mpage_da_map_blocks. We don't
				 * update b_size because if we find an
				 * unmapped buffer_head later we need to
				 * use the b_state flag of that buffer_head.
				 */
2481 2482
				if (mpd->b_size == 0)
					mpd->b_state = bh->b_state & BH_FLAGS;
2483
			}
2484 2485 2486 2487 2488 2489 2490 2491
			logical++;
		} while ((bh = bh->b_this_page) != head);
	}

	return 0;
}

/*
2492 2493 2494
 * This is a special get_blocks_t callback which is used by
 * ext4_da_write_begin().  It will either return mapped block or
 * reserve space for a single block.
2495 2496 2497 2498 2499 2500 2501
 *
 * For delayed buffer_head we have BH_Mapped, BH_New, BH_Delay set.
 * We also have b_blocknr = -1 and b_bdev initialized properly
 *
 * For unwritten buffer_head we have BH_Mapped, BH_New, BH_Unwritten set.
 * We also have b_blocknr = physicalblock mapping unwritten extent and b_bdev
 * initialized properly.
2502 2503 2504 2505 2506
 */
static int ext4_da_get_block_prep(struct inode *inode, sector_t iblock,
				  struct buffer_head *bh_result, int create)
{
	int ret = 0;
2507 2508 2509 2510
	sector_t invalid_block = ~((sector_t) 0xffff);

	if (invalid_block < ext4_blocks_count(EXT4_SB(inode->i_sb)->s_es))
		invalid_block = ~0;
2511 2512 2513 2514 2515 2516 2517 2518 2519

	BUG_ON(create == 0);
	BUG_ON(bh_result->b_size != inode->i_sb->s_blocksize);

	/*
	 * first, we need to know whether the block is allocated already
	 * preallocated blocks are unmapped but should treated
	 * the same as allocated blocks.
	 */
2520
	ret = ext4_get_blocks(NULL, inode, iblock, 1,  bh_result, 0);
2521 2522
	if ((ret == 0) && !buffer_delay(bh_result)) {
		/* the block isn't (pre)allocated yet, let's reserve space */
2523 2524 2525 2526
		/*
		 * XXX: __block_prepare_write() unmaps passed block,
		 * is it OK?
		 */
2527 2528 2529 2530 2531
		ret = ext4_da_reserve_space(inode, 1);
		if (ret)
			/* not enough space to reserve */
			return ret;

2532
		map_bh(bh_result, inode->i_sb, invalid_block);
2533 2534 2535 2536
		set_buffer_new(bh_result);
		set_buffer_delay(bh_result);
	} else if (ret > 0) {
		bh_result->b_size = (ret << inode->i_blkbits);
2537 2538 2539 2540 2541 2542 2543 2544
		if (buffer_unwritten(bh_result)) {
			/* A delayed write to unwritten bh should
			 * be marked new and mapped.  Mapped ensures
			 * that we don't do get_block multiple times
			 * when we write to the same offset and new
			 * ensures that we do proper zero out for
			 * partial write.
			 */
2545
			set_buffer_new(bh_result);
2546 2547
			set_buffer_mapped(bh_result);
		}
2548 2549 2550 2551 2552
		ret = 0;
	}

	return ret;
}
2553

2554 2555 2556 2557 2558 2559 2560 2561 2562 2563 2564 2565 2566 2567 2568 2569 2570
/*
 * This function is used as a standard get_block_t calback function
 * when there is no desire to allocate any blocks.  It is used as a
 * callback function for block_prepare_write(), nobh_writepage(), and
 * block_write_full_page().  These functions should only try to map a
 * single block at a time.
 *
 * Since this function doesn't do block allocations even if the caller
 * requests it by passing in create=1, it is critically important that
 * any caller checks to make sure that any buffer heads are returned
 * by this function are either all already mapped or marked for
 * delayed allocation before calling nobh_writepage() or
 * block_write_full_page().  Otherwise, b_blocknr could be left
 * unitialized, and the page write functions will be taken by
 * surprise.
 */
static int noalloc_get_block_write(struct inode *inode, sector_t iblock,
2571 2572 2573 2574 2575
				   struct buffer_head *bh_result, int create)
{
	int ret = 0;
	unsigned max_blocks = bh_result->b_size >> inode->i_blkbits;

2576 2577
	BUG_ON(bh_result->b_size != inode->i_sb->s_blocksize);

2578 2579 2580 2581
	/*
	 * we don't want to do block allocation in writepage
	 * so call get_block_wrap with create = 0
	 */
2582
	ret = ext4_get_blocks(NULL, inode, iblock, max_blocks, bh_result, 0);
2583 2584 2585 2586 2587
	if (ret > 0) {
		bh_result->b_size = (ret << inode->i_blkbits);
		ret = 0;
	}
	return ret;
2588 2589
}

2590 2591 2592 2593 2594 2595 2596 2597 2598 2599 2600 2601 2602 2603 2604 2605 2606 2607 2608 2609 2610 2611 2612 2613 2614 2615 2616 2617 2618 2619 2620 2621 2622 2623 2624 2625 2626 2627 2628 2629 2630 2631 2632 2633 2634 2635 2636 2637 2638 2639 2640 2641 2642
static int bget_one(handle_t *handle, struct buffer_head *bh)
{
	get_bh(bh);
	return 0;
}

static int bput_one(handle_t *handle, struct buffer_head *bh)
{
	put_bh(bh);
	return 0;
}

static int __ext4_journalled_writepage(struct page *page,
				       struct writeback_control *wbc,
				       unsigned int len)
{
	struct address_space *mapping = page->mapping;
	struct inode *inode = mapping->host;
	struct buffer_head *page_bufs;
	handle_t *handle = NULL;
	int ret = 0;
	int err;

	page_bufs = page_buffers(page);
	BUG_ON(!page_bufs);
	walk_page_buffers(handle, page_bufs, 0, len, NULL, bget_one);
	/* As soon as we unlock the page, it can go away, but we have
	 * references to buffers so we are safe */
	unlock_page(page);

	handle = ext4_journal_start(inode, ext4_writepage_trans_blocks(inode));
	if (IS_ERR(handle)) {
		ret = PTR_ERR(handle);
		goto out;
	}

	ret = walk_page_buffers(handle, page_bufs, 0, len, NULL,
				do_journal_get_write_access);

	err = walk_page_buffers(handle, page_bufs, 0, len, NULL,
				write_end_fn);
	if (ret == 0)
		ret = err;
	err = ext4_journal_stop(handle);
	if (!ret)
		ret = err;

	walk_page_buffers(handle, page_bufs, 0, len, NULL, bput_one);
	EXT4_I(inode)->i_state |= EXT4_STATE_JDATA;
out:
	return ret;
}

2643
/*
2644 2645 2646 2647 2648 2649 2650 2651 2652
 * Note that we don't need to start a transaction unless we're journaling data
 * because we should have holes filled from ext4_page_mkwrite(). We even don't
 * need to file the inode to the transaction's list in ordered mode because if
 * we are writing back data added by write(), the inode is already there and if
 * we are writing back data modified via mmap(), noone guarantees in which
 * transaction the data will hit the disk. In case we are journaling data, we
 * cannot start transaction directly because transaction start ranks above page
 * lock so we have to do some magic.
 *
2653 2654 2655 2656 2657
 * This function can get called via...
 *   - ext4_da_writepages after taking page lock (have journal handle)
 *   - journal_submit_inode_data_buffers (no journal handle)
 *   - shrink_page_list via pdflush (no journal handle)
 *   - grab_page_cache when doing write_begin (have journal handle)
2658 2659 2660 2661 2662 2663 2664 2665 2666 2667 2668 2669 2670 2671 2672 2673 2674 2675 2676 2677 2678 2679 2680 2681 2682
 *
 * We don't do any block allocation in this function. If we have page with
 * multiple blocks we need to write those buffer_heads that are mapped. This
 * is important for mmaped based write. So if we do with blocksize 1K
 * truncate(f, 1024);
 * a = mmap(f, 0, 4096);
 * a[0] = 'a';
 * truncate(f, 4096);
 * we have in the page first buffer_head mapped via page_mkwrite call back
 * but other bufer_heads would be unmapped but dirty(dirty done via the
 * do_wp_page). So writepage should write the first block. If we modify
 * the mmap area beyond 1024 we will again get a page_fault and the
 * page_mkwrite callback will do the block allocation and mark the
 * buffer_heads mapped.
 *
 * We redirty the page if we have any buffer_heads that is either delay or
 * unwritten in the page.
 *
 * We can get recursively called as show below.
 *
 *	ext4_writepage() -> kmalloc() -> __alloc_pages() -> page_launder() ->
 *		ext4_writepage()
 *
 * But since we don't do any block allocation we should not deadlock.
 * Page also have the dirty flag cleared so we don't get recurive page_lock.
2683
 */
2684
static int ext4_writepage(struct page *page,
2685
			  struct writeback_control *wbc)
2686 2687
{
	int ret = 0;
2688
	loff_t size;
2689
	unsigned int len;
2690 2691 2692
	struct buffer_head *page_bufs;
	struct inode *inode = page->mapping->host;

2693
	trace_ext4_writepage(inode, page);
2694 2695 2696 2697 2698
	size = i_size_read(inode);
	if (page->index == size >> PAGE_CACHE_SHIFT)
		len = size & ~PAGE_CACHE_MASK;
	else
		len = PAGE_CACHE_SIZE;
2699

2700
	if (page_has_buffers(page)) {
2701
		page_bufs = page_buffers(page);
2702
		if (walk_page_buffers(NULL, page_bufs, 0, len, NULL,
2703
					ext4_bh_delay_or_unwritten)) {
2704
			/*
2705 2706
			 * We don't want to do  block allocation
			 * So redirty the page and return
2707 2708 2709
			 * We may reach here when we do a journal commit
			 * via journal_submit_inode_data_buffers.
			 * If we don't have mapping block we just ignore
2710 2711 2712 2713 2714 2715 2716 2717 2718 2719 2720 2721 2722 2723 2724 2725 2726 2727 2728 2729
			 * them. We can also reach here via shrink_page_list
			 */
			redirty_page_for_writepage(wbc, page);
			unlock_page(page);
			return 0;
		}
	} else {
		/*
		 * The test for page_has_buffers() is subtle:
		 * We know the page is dirty but it lost buffers. That means
		 * that at some moment in time after write_begin()/write_end()
		 * has been called all buffers have been clean and thus they
		 * must have been written at least once. So they are all
		 * mapped and we can happily proceed with mapping them
		 * and writing the page.
		 *
		 * Try to initialize the buffer_heads and check whether
		 * all are mapped and non delay. We don't want to
		 * do block allocation here.
		 */
2730
		ret = block_prepare_write(page, 0, len,
2731
					  noalloc_get_block_write);
2732 2733 2734 2735
		if (!ret) {
			page_bufs = page_buffers(page);
			/* check whether all are mapped and non delay */
			if (walk_page_buffers(NULL, page_bufs, 0, len, NULL,
2736
						ext4_bh_delay_or_unwritten)) {
2737 2738 2739 2740 2741 2742 2743 2744 2745
				redirty_page_for_writepage(wbc, page);
				unlock_page(page);
				return 0;
			}
		} else {
			/*
			 * We can't do block allocation here
			 * so just redity the page and unlock
			 * and return
2746 2747 2748 2749 2750
			 */
			redirty_page_for_writepage(wbc, page);
			unlock_page(page);
			return 0;
		}
2751
		/* now mark the buffer_heads as dirty and uptodate */
2752
		block_commit_write(page, 0, len);
2753 2754
	}

2755 2756 2757 2758 2759 2760 2761 2762 2763
	if (PageChecked(page) && ext4_should_journal_data(inode)) {
		/*
		 * It's mmapped pagecache.  Add buffers and journal it.  There
		 * doesn't seem much point in redirtying the page here.
		 */
		ClearPageChecked(page);
		return __ext4_journalled_writepage(page, wbc, len);
	}

2764
	if (test_opt(inode->i_sb, NOBH) && ext4_should_writeback_data(inode))
2765
		ret = nobh_writepage(page, noalloc_get_block_write, wbc);
2766
	else
2767 2768
		ret = block_write_full_page(page, noalloc_get_block_write,
					    wbc);
2769 2770 2771 2772

	return ret;
}

2773
/*
2774 2775 2776 2777 2778
 * This is called via ext4_da_writepages() to
 * calulate the total number of credits to reserve to fit
 * a single extent allocation into a single transaction,
 * ext4_da_writpeages() will loop calling this before
 * the block allocation.
2779
 */
2780 2781 2782 2783 2784 2785 2786 2787 2788 2789 2790 2791 2792 2793 2794 2795 2796

static int ext4_da_writepages_trans_blocks(struct inode *inode)
{
	int max_blocks = EXT4_I(inode)->i_reserved_data_blocks;

	/*
	 * With non-extent format the journal credit needed to
	 * insert nrblocks contiguous block is dependent on
	 * number of contiguous block. So we will limit
	 * number of contiguous block to a sane value
	 */
	if (!(inode->i_flags & EXT4_EXTENTS_FL) &&
	    (max_blocks > EXT4_MAX_TRANS_DATA))
		max_blocks = EXT4_MAX_TRANS_DATA;

	return ext4_chunk_trans_blocks(inode, max_blocks);
}
2797

2798
static int ext4_da_writepages(struct address_space *mapping,
2799
			      struct writeback_control *wbc)
2800
{
2801 2802
	pgoff_t	index;
	int range_whole = 0;
2803
	handle_t *handle = NULL;
2804
	struct mpage_da_data mpd;
2805
	struct inode *inode = mapping->host;
2806
	int no_nrwrite_index_update;
2807 2808
	int pages_written = 0;
	long pages_skipped;
2809
	unsigned int max_pages;
2810
	int range_cyclic, cycled = 1, io_done = 0;
2811 2812
	int needed_blocks, ret = 0;
	long desired_nr_to_write, nr_to_writebump = 0;
2813
	loff_t range_start = wbc->range_start;
2814
	struct ext4_sb_info *sbi = EXT4_SB(mapping->host->i_sb);
2815

2816
	trace_ext4_da_writepages(inode, wbc);
2817

2818 2819 2820 2821 2822
	/*
	 * No pages to write? This is mainly a kludge to avoid starting
	 * a transaction for special inodes like journal inode on last iput()
	 * because that could violate lock ordering on umount
	 */
2823
	if (!mapping->nrpages || !mapping_tagged(mapping, PAGECACHE_TAG_DIRTY))
2824
		return 0;
2825 2826 2827 2828 2829

	/*
	 * If the filesystem has aborted, it is read-only, so return
	 * right away instead of dumping stack traces later on that
	 * will obscure the real source of the problem.  We test
2830
	 * EXT4_MF_FS_ABORTED instead of sb->s_flag's MS_RDONLY because
2831 2832 2833 2834 2835
	 * the latter could be true if the filesystem is mounted
	 * read-only, and in that case, ext4_da_writepages should
	 * *never* be called, so if that ever happens, we would want
	 * the stack trace.
	 */
2836
	if (unlikely(sbi->s_mount_flags & EXT4_MF_FS_ABORTED))
2837 2838
		return -EROFS;

2839 2840
	if (wbc->range_start == 0 && wbc->range_end == LLONG_MAX)
		range_whole = 1;
2841

2842 2843
	range_cyclic = wbc->range_cyclic;
	if (wbc->range_cyclic) {
2844
		index = mapping->writeback_index;
2845 2846 2847 2848 2849 2850
		if (index)
			cycled = 0;
		wbc->range_start = index << PAGE_CACHE_SHIFT;
		wbc->range_end  = LLONG_MAX;
		wbc->range_cyclic = 0;
	} else
2851
		index = wbc->range_start >> PAGE_CACHE_SHIFT;
2852

2853 2854 2855 2856 2857 2858 2859 2860 2861 2862 2863 2864 2865 2866 2867 2868 2869 2870 2871 2872 2873 2874 2875 2876 2877 2878 2879 2880 2881 2882
	/*
	 * This works around two forms of stupidity.  The first is in
	 * the writeback code, which caps the maximum number of pages
	 * written to be 1024 pages.  This is wrong on multiple
	 * levels; different architectues have a different page size,
	 * which changes the maximum amount of data which gets
	 * written.  Secondly, 4 megabytes is way too small.  XFS
	 * forces this value to be 16 megabytes by multiplying
	 * nr_to_write parameter by four, and then relies on its
	 * allocator to allocate larger extents to make them
	 * contiguous.  Unfortunately this brings us to the second
	 * stupidity, which is that ext4's mballoc code only allocates
	 * at most 2048 blocks.  So we force contiguous writes up to
	 * the number of dirty blocks in the inode, or
	 * sbi->max_writeback_mb_bump whichever is smaller.
	 */
	max_pages = sbi->s_max_writeback_mb_bump << (20 - PAGE_CACHE_SHIFT);
	if (!range_cyclic && range_whole)
		desired_nr_to_write = wbc->nr_to_write * 8;
	else
		desired_nr_to_write = ext4_num_dirty_pages(inode, index,
							   max_pages);
	if (desired_nr_to_write > max_pages)
		desired_nr_to_write = max_pages;

	if (wbc->nr_to_write < desired_nr_to_write) {
		nr_to_writebump = desired_nr_to_write - wbc->nr_to_write;
		wbc->nr_to_write = desired_nr_to_write;
	}

2883 2884 2885
	mpd.wbc = wbc;
	mpd.inode = mapping->host;

2886 2887 2888 2889 2890 2891 2892 2893
	/*
	 * we don't want write_cache_pages to update
	 * nr_to_write and writeback_index
	 */
	no_nrwrite_index_update = wbc->no_nrwrite_index_update;
	wbc->no_nrwrite_index_update = 1;
	pages_skipped = wbc->pages_skipped;

2894
retry:
2895
	while (!ret && wbc->nr_to_write > 0) {
2896 2897 2898 2899 2900 2901 2902 2903

		/*
		 * we  insert one extent at a time. So we need
		 * credit needed for single extent allocation.
		 * journalled mode is currently not supported
		 * by delalloc
		 */
		BUG_ON(ext4_should_journal_data(inode));
2904
		needed_blocks = ext4_da_writepages_trans_blocks(inode);
2905

2906 2907 2908 2909
		/* start a new transaction*/
		handle = ext4_journal_start(inode, needed_blocks);
		if (IS_ERR(handle)) {
			ret = PTR_ERR(handle);
2910
			ext4_msg(inode->i_sb, KERN_CRIT, "%s: jbd2_start: "
2911 2912
			       "%ld pages, ino %lu; err %d\n", __func__,
				wbc->nr_to_write, inode->i_ino, ret);
2913 2914
			goto out_writepages;
		}
2915 2916 2917 2918 2919 2920 2921 2922 2923 2924 2925 2926 2927 2928 2929 2930 2931 2932 2933 2934 2935 2936 2937 2938 2939 2940 2941 2942 2943 2944 2945

		/*
		 * Now call __mpage_da_writepage to find the next
		 * contiguous region of logical blocks that need
		 * blocks to be allocated by ext4.  We don't actually
		 * submit the blocks for I/O here, even though
		 * write_cache_pages thinks it will, and will set the
		 * pages as clean for write before calling
		 * __mpage_da_writepage().
		 */
		mpd.b_size = 0;
		mpd.b_state = 0;
		mpd.b_blocknr = 0;
		mpd.first_page = 0;
		mpd.next_page = 0;
		mpd.io_done = 0;
		mpd.pages_written = 0;
		mpd.retval = 0;
		ret = write_cache_pages(mapping, wbc, __mpage_da_writepage,
					&mpd);
		/*
		 * If we have a contigous extent of pages and we
		 * haven't done the I/O yet, map the blocks and submit
		 * them for I/O.
		 */
		if (!mpd.io_done && mpd.next_page != mpd.first_page) {
			if (mpage_da_map_blocks(&mpd) == 0)
				mpage_da_submit_io(&mpd);
			mpd.io_done = 1;
			ret = MPAGE_DA_EXTENT_TAIL;
		}
2946
		trace_ext4_da_write_pages(inode, &mpd);
2947
		wbc->nr_to_write -= mpd.pages_written;
2948

2949
		ext4_journal_stop(handle);
2950

2951
		if ((mpd.retval == -ENOSPC) && sbi->s_journal) {
2952 2953 2954 2955
			/* commit the transaction which would
			 * free blocks released in the transaction
			 * and try again
			 */
2956
			jbd2_journal_force_commit_nested(sbi->s_journal);
2957 2958 2959
			wbc->pages_skipped = pages_skipped;
			ret = 0;
		} else if (ret == MPAGE_DA_EXTENT_TAIL) {
2960 2961 2962 2963
			/*
			 * got one extent now try with
			 * rest of the pages
			 */
2964 2965
			pages_written += mpd.pages_written;
			wbc->pages_skipped = pages_skipped;
2966
			ret = 0;
2967
			io_done = 1;
2968
		} else if (wbc->nr_to_write)
2969 2970 2971 2972 2973 2974
			/*
			 * There is no more writeout needed
			 * or we requested for a noblocking writeout
			 * and we found the device congested
			 */
			break;
2975
	}
2976 2977 2978 2979 2980 2981 2982
	if (!io_done && !cycled) {
		cycled = 1;
		index = 0;
		wbc->range_start = index << PAGE_CACHE_SHIFT;
		wbc->range_end  = mapping->writeback_index - 1;
		goto retry;
	}
2983
	if (pages_skipped != wbc->pages_skipped)
2984 2985 2986 2987
		ext4_msg(inode->i_sb, KERN_CRIT,
			 "This should not happen leaving %s "
			 "with nr_to_write = %ld ret = %d\n",
			 __func__, wbc->nr_to_write, ret);
2988 2989 2990

	/* Update index */
	index += pages_written;
2991
	wbc->range_cyclic = range_cyclic;
2992 2993 2994 2995 2996 2997
	if (wbc->range_cyclic || (range_whole && wbc->nr_to_write > 0))
		/*
		 * set the writeback_index so that range_cyclic
		 * mode will write it back later
		 */
		mapping->writeback_index = index;
2998

2999
out_writepages:
3000 3001
	if (!no_nrwrite_index_update)
		wbc->no_nrwrite_index_update = 0;
3002 3003
	if (wbc->nr_to_write > nr_to_writebump)
		wbc->nr_to_write -= nr_to_writebump;
3004
	wbc->range_start = range_start;
3005
	trace_ext4_da_writepages_result(inode, wbc, ret, pages_written);
3006
	return ret;
3007 3008
}

3009 3010 3011 3012 3013 3014 3015 3016 3017
#define FALL_BACK_TO_NONDELALLOC 1
static int ext4_nonda_switch(struct super_block *sb)
{
	s64 free_blocks, dirty_blocks;
	struct ext4_sb_info *sbi = EXT4_SB(sb);

	/*
	 * switch to non delalloc mode if we are running low
	 * on free block. The free block accounting via percpu
3018
	 * counters can get slightly wrong with percpu_counter_batch getting
3019 3020 3021 3022 3023 3024 3025 3026 3027 3028 3029 3030 3031 3032 3033 3034 3035
	 * accumulated on each CPU without updating global counters
	 * Delalloc need an accurate free block accounting. So switch
	 * to non delalloc when we are near to error range.
	 */
	free_blocks  = percpu_counter_read_positive(&sbi->s_freeblocks_counter);
	dirty_blocks = percpu_counter_read_positive(&sbi->s_dirtyblocks_counter);
	if (2 * free_blocks < 3 * dirty_blocks ||
		free_blocks < (dirty_blocks + EXT4_FREEBLOCKS_WATERMARK)) {
		/*
		 * free block count is less that 150% of dirty blocks
		 * or free blocks is less that watermark
		 */
		return 1;
	}
	return 0;
}

3036
static int ext4_da_write_begin(struct file *file, struct address_space *mapping,
3037 3038
			       loff_t pos, unsigned len, unsigned flags,
			       struct page **pagep, void **fsdata)
3039
{
3040
	int ret, retries = 0;
3041 3042 3043 3044 3045 3046 3047 3048 3049
	struct page *page;
	pgoff_t index;
	unsigned from, to;
	struct inode *inode = mapping->host;
	handle_t *handle;

	index = pos >> PAGE_CACHE_SHIFT;
	from = pos & (PAGE_CACHE_SIZE - 1);
	to = from + len;
3050 3051 3052 3053 3054 3055 3056

	if (ext4_nonda_switch(inode->i_sb)) {
		*fsdata = (void *)FALL_BACK_TO_NONDELALLOC;
		return ext4_write_begin(file, mapping, pos,
					len, flags, pagep, fsdata);
	}
	*fsdata = (void *)0;
3057
	trace_ext4_da_write_begin(inode, pos, len, flags);
3058
retry:
3059 3060 3061 3062 3063 3064 3065 3066 3067 3068 3069
	/*
	 * With delayed allocation, we don't log the i_disksize update
	 * if there is delayed block allocation. But we still need
	 * to journalling the i_disksize update if writes to the end
	 * of file which has an already mapped buffer.
	 */
	handle = ext4_journal_start(inode, 1);
	if (IS_ERR(handle)) {
		ret = PTR_ERR(handle);
		goto out;
	}
3070 3071 3072
	/* We cannot recurse into the filesystem as the transaction is already
	 * started */
	flags |= AOP_FLAG_NOFS;
3073

3074
	page = grab_cache_page_write_begin(mapping, index, flags);
3075 3076 3077 3078 3079
	if (!page) {
		ext4_journal_stop(handle);
		ret = -ENOMEM;
		goto out;
	}
3080 3081 3082
	*pagep = page;

	ret = block_write_begin(file, mapping, pos, len, flags, pagep, fsdata,
3083
				ext4_da_get_block_prep);
3084 3085 3086 3087
	if (ret < 0) {
		unlock_page(page);
		ext4_journal_stop(handle);
		page_cache_release(page);
3088 3089 3090 3091 3092 3093
		/*
		 * block_write_begin may have instantiated a few blocks
		 * outside i_size.  Trim these off again. Don't need
		 * i_size_read because we hold i_mutex.
		 */
		if (pos + len > inode->i_size)
3094
			ext4_truncate(inode);
3095 3096
	}

3097 3098
	if (ret == -ENOSPC && ext4_should_retry_alloc(inode->i_sb, &retries))
		goto retry;
3099 3100 3101 3102
out:
	return ret;
}

3103 3104 3105 3106 3107
/*
 * Check if we should update i_disksize
 * when write to the end of file but not require block allocation
 */
static int ext4_da_should_update_i_disksize(struct page *page,
3108
					    unsigned long offset)
3109 3110 3111 3112 3113 3114 3115 3116 3117
{
	struct buffer_head *bh;
	struct inode *inode = page->mapping->host;
	unsigned int idx;
	int i;

	bh = page_buffers(page);
	idx = offset >> inode->i_blkbits;

3118
	for (i = 0; i < idx; i++)
3119 3120
		bh = bh->b_this_page;

3121
	if (!buffer_mapped(bh) || (buffer_delay(bh)) || buffer_unwritten(bh))
3122 3123 3124 3125
		return 0;
	return 1;
}

3126
static int ext4_da_write_end(struct file *file,
3127 3128 3129
			     struct address_space *mapping,
			     loff_t pos, unsigned len, unsigned copied,
			     struct page *page, void *fsdata)
3130 3131 3132 3133 3134
{
	struct inode *inode = mapping->host;
	int ret = 0, ret2;
	handle_t *handle = ext4_journal_current_handle();
	loff_t new_i_size;
3135
	unsigned long start, end;
3136 3137 3138 3139 3140 3141 3142 3143 3144 3145 3146 3147 3148
	int write_mode = (int)(unsigned long)fsdata;

	if (write_mode == FALL_BACK_TO_NONDELALLOC) {
		if (ext4_should_order_data(inode)) {
			return ext4_ordered_write_end(file, mapping, pos,
					len, copied, page, fsdata);
		} else if (ext4_should_writeback_data(inode)) {
			return ext4_writeback_write_end(file, mapping, pos,
					len, copied, page, fsdata);
		} else {
			BUG();
		}
	}
3149

3150
	trace_ext4_da_write_end(inode, pos, len, copied);
3151
	start = pos & (PAGE_CACHE_SIZE - 1);
3152
	end = start + copied - 1;
3153 3154 3155 3156 3157 3158 3159 3160

	/*
	 * generic_write_end() will run mark_inode_dirty() if i_size
	 * changes.  So let's piggyback the i_disksize mark_inode_dirty
	 * into that.
	 */

	new_i_size = pos + copied;
3161 3162 3163 3164 3165 3166 3167 3168 3169 3170 3171
	if (new_i_size > EXT4_I(inode)->i_disksize) {
		if (ext4_da_should_update_i_disksize(page, end)) {
			down_write(&EXT4_I(inode)->i_data_sem);
			if (new_i_size > EXT4_I(inode)->i_disksize) {
				/*
				 * Updating i_disksize when extending file
				 * without needing block allocation
				 */
				if (ext4_should_order_data(inode))
					ret = ext4_jbd2_file_inode(handle,
								   inode);
3172

3173 3174 3175
				EXT4_I(inode)->i_disksize = new_i_size;
			}
			up_write(&EXT4_I(inode)->i_data_sem);
3176 3177 3178 3179 3180
			/* 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_mark_inode_dirty(handle, inode);
3181
		}
3182
	}
3183 3184 3185 3186 3187 3188 3189 3190 3191 3192 3193 3194 3195 3196 3197 3198 3199 3200 3201 3202 3203
	ret2 = generic_write_end(file, mapping, pos, len, copied,
							page, fsdata);
	copied = ret2;
	if (ret2 < 0)
		ret = ret2;
	ret2 = ext4_journal_stop(handle);
	if (!ret)
		ret = ret2;

	return ret ? ret : copied;
}

static void ext4_da_invalidatepage(struct page *page, unsigned long offset)
{
	/*
	 * Drop reserved blocks
	 */
	BUG_ON(!PageLocked(page));
	if (!page_has_buffers(page))
		goto out;

3204
	ext4_da_page_release_reservation(page, offset);
3205 3206 3207 3208 3209 3210 3211

out:
	ext4_invalidatepage(page, offset);

	return;
}

3212 3213 3214 3215 3216
/*
 * Force all delayed allocation blocks to be allocated for a given inode.
 */
int ext4_alloc_da_blocks(struct inode *inode)
{
3217 3218
	trace_ext4_alloc_da_blocks(inode);

3219 3220 3221 3222 3223 3224 3225 3226 3227 3228
	if (!EXT4_I(inode)->i_reserved_data_blocks &&
	    !EXT4_I(inode)->i_reserved_meta_blocks)
		return 0;

	/*
	 * We do something simple for now.  The filemap_flush() will
	 * also start triggering a write of the data blocks, which is
	 * not strictly speaking necessary (and for users of
	 * laptop_mode, not even desirable).  However, to do otherwise
	 * would require replicating code paths in:
3229
	 *
3230 3231 3232 3233 3234 3235 3236 3237 3238 3239 3240 3241 3242 3243 3244 3245 3246 3247 3248
	 * ext4_da_writepages() ->
	 *    write_cache_pages() ---> (via passed in callback function)
	 *        __mpage_da_writepage() -->
	 *           mpage_add_bh_to_extent()
	 *           mpage_da_map_blocks()
	 *
	 * The problem is that write_cache_pages(), located in
	 * mm/page-writeback.c, marks pages clean in preparation for
	 * doing I/O, which is not desirable if we're not planning on
	 * doing I/O at all.
	 *
	 * We could call write_cache_pages(), and then redirty all of
	 * the pages by calling redirty_page_for_writeback() but that
	 * would be ugly in the extreme.  So instead we would need to
	 * replicate parts of the code in the above functions,
	 * simplifying them becuase we wouldn't actually intend to
	 * write out the pages, but rather only collect contiguous
	 * logical block extents, call the multi-block allocator, and
	 * then update the buffer heads with the block allocations.
3249
	 *
3250 3251 3252 3253 3254 3255
	 * For now, though, we'll cheat by calling filemap_flush(),
	 * which will map the blocks, and start the I/O, but not
	 * actually wait for the I/O to complete.
	 */
	return filemap_flush(inode->i_mapping);
}
3256

3257 3258 3259 3260 3261
/*
 * bmap() is special.  It gets used by applications such as lilo and by
 * the swapper to find the on-disk block of a specific piece of data.
 *
 * Naturally, this is dangerous if the block concerned is still in the
3262
 * journal.  If somebody makes a swapfile on an ext4 data-journaling
3263 3264 3265 3266 3267 3268 3269 3270
 * filesystem and enables swap, then they may get a nasty shock when the
 * data getting swapped to that swapfile suddenly gets overwritten by
 * the original zero's written out previously to the journal and
 * awaiting writeback in the kernel's buffer cache.
 *
 * So, if we see any bmap calls here on a modified, data-journaled file,
 * take extra steps to flush any blocks which might be in the cache.
 */
3271
static sector_t ext4_bmap(struct address_space *mapping, sector_t block)
3272 3273 3274 3275 3276
{
	struct inode *inode = mapping->host;
	journal_t *journal;
	int err;

3277 3278 3279 3280 3281 3282 3283 3284 3285 3286
	if (mapping_tagged(mapping, PAGECACHE_TAG_DIRTY) &&
			test_opt(inode->i_sb, DELALLOC)) {
		/*
		 * With delalloc we want to sync the file
		 * so that we can make sure we allocate
		 * blocks for file
		 */
		filemap_write_and_wait(mapping);
	}

3287
	if (EXT4_JOURNAL(inode) && EXT4_I(inode)->i_state & EXT4_STATE_JDATA) {
3288 3289 3290 3291 3292 3293 3294 3295 3296 3297 3298
		/*
		 * This is a REALLY heavyweight approach, but the use of
		 * bmap on dirty files is expected to be extremely rare:
		 * only if we run lilo or swapon on a freshly made file
		 * do we expect this to happen.
		 *
		 * (bmap requires CAP_SYS_RAWIO so this does not
		 * represent an unprivileged user DOS attack --- we'd be
		 * in trouble if mortal users could trigger this path at
		 * will.)
		 *
3299
		 * NB. EXT4_STATE_JDATA is not set on files other than
3300 3301 3302 3303 3304 3305
		 * regular files.  If somebody wants to bmap a directory
		 * or symlink and gets confused because the buffer
		 * hasn't yet been flushed to disk, they deserve
		 * everything they get.
		 */

3306 3307
		EXT4_I(inode)->i_state &= ~EXT4_STATE_JDATA;
		journal = EXT4_JOURNAL(inode);
3308 3309 3310
		jbd2_journal_lock_updates(journal);
		err = jbd2_journal_flush(journal);
		jbd2_journal_unlock_updates(journal);
3311 3312 3313 3314 3315

		if (err)
			return 0;
	}

3316
	return generic_block_bmap(mapping, block, ext4_get_block);
3317 3318
}

3319
static int ext4_readpage(struct file *file, struct page *page)
3320
{
3321
	return mpage_readpage(page, ext4_get_block);
3322 3323 3324
}

static int
3325
ext4_readpages(struct file *file, struct address_space *mapping,
3326 3327
		struct list_head *pages, unsigned nr_pages)
{
3328
	return mpage_readpages(mapping, pages, nr_pages, ext4_get_block);
3329 3330
}

3331
static void ext4_invalidatepage(struct page *page, unsigned long offset)
3332
{
3333
	journal_t *journal = EXT4_JOURNAL(page->mapping->host);
3334 3335 3336 3337 3338 3339 3340

	/*
	 * If it's a full truncate we just forget about the pending dirtying
	 */
	if (offset == 0)
		ClearPageChecked(page);

3341 3342 3343 3344
	if (journal)
		jbd2_journal_invalidatepage(journal, page, offset);
	else
		block_invalidatepage(page, offset);
3345 3346
}

3347
static int ext4_releasepage(struct page *page, gfp_t wait)
3348
{
3349
	journal_t *journal = EXT4_JOURNAL(page->mapping->host);
3350 3351 3352 3353

	WARN_ON(PageChecked(page));
	if (!page_has_buffers(page))
		return 0;
3354 3355 3356 3357
	if (journal)
		return jbd2_journal_try_to_free_buffers(journal, page, wait);
	else
		return try_to_free_buffers(page);
3358 3359 3360
}

/*
3361 3362
 * O_DIRECT for ext3 (or indirect map) based files
 *
3363 3364 3365 3366 3367
 * If the O_DIRECT write will extend the file then add this inode to the
 * orphan list.  So recovery will truncate it back to the original size
 * if the machine crashes during the write.
 *
 * If the O_DIRECT write is intantiating holes inside i_size and the machine
J
Jan Kara 已提交
3368 3369
 * crashes then stale disk data _may_ be exposed inside the file. But current
 * VFS code falls back into buffered path in that case so we are safe.
3370
 */
3371
static ssize_t ext4_ind_direct_IO(int rw, struct kiocb *iocb,
3372 3373
			      const struct iovec *iov, loff_t offset,
			      unsigned long nr_segs)
3374 3375 3376
{
	struct file *file = iocb->ki_filp;
	struct inode *inode = file->f_mapping->host;
3377
	struct ext4_inode_info *ei = EXT4_I(inode);
J
Jan Kara 已提交
3378
	handle_t *handle;
3379 3380 3381
	ssize_t ret;
	int orphan = 0;
	size_t count = iov_length(iov, nr_segs);
3382
	int retries = 0;
3383 3384 3385 3386 3387

	if (rw == WRITE) {
		loff_t final_size = offset + count;

		if (final_size > inode->i_size) {
J
Jan Kara 已提交
3388 3389 3390 3391 3392 3393
			/* Credits for sb + inode write */
			handle = ext4_journal_start(inode, 2);
			if (IS_ERR(handle)) {
				ret = PTR_ERR(handle);
				goto out;
			}
3394
			ret = ext4_orphan_add(handle, inode);
J
Jan Kara 已提交
3395 3396 3397 3398
			if (ret) {
				ext4_journal_stop(handle);
				goto out;
			}
3399 3400
			orphan = 1;
			ei->i_disksize = inode->i_size;
J
Jan Kara 已提交
3401
			ext4_journal_stop(handle);
3402 3403 3404
		}
	}

3405
retry:
3406 3407
	ret = blockdev_direct_IO(rw, iocb, inode, inode->i_sb->s_bdev, iov,
				 offset, nr_segs,
3408
				 ext4_get_block, NULL);
3409 3410
	if (ret == -ENOSPC && ext4_should_retry_alloc(inode->i_sb, &retries))
		goto retry;
3411

J
Jan Kara 已提交
3412
	if (orphan) {
3413 3414
		int err;

J
Jan Kara 已提交
3415 3416 3417 3418 3419 3420 3421 3422 3423 3424
		/* Credits for sb + inode write */
		handle = ext4_journal_start(inode, 2);
		if (IS_ERR(handle)) {
			/* This is really bad luck. We've written the data
			 * but cannot extend i_size. Bail out and pretend
			 * the write failed... */
			ret = PTR_ERR(handle);
			goto out;
		}
		if (inode->i_nlink)
3425
			ext4_orphan_del(handle, inode);
J
Jan Kara 已提交
3426
		if (ret > 0) {
3427 3428 3429 3430 3431 3432 3433 3434
			loff_t end = offset + ret;
			if (end > inode->i_size) {
				ei->i_disksize = end;
				i_size_write(inode, end);
				/*
				 * We're going to return a positive `ret'
				 * here due to non-zero-length I/O, so there's
				 * no way of reporting error returns from
3435
				 * ext4_mark_inode_dirty() to userspace.  So
3436 3437
				 * ignore it.
				 */
3438
				ext4_mark_inode_dirty(handle, inode);
3439 3440
			}
		}
3441
		err = ext4_journal_stop(handle);
3442 3443 3444 3445 3446 3447 3448
		if (ret == 0)
			ret = err;
	}
out:
	return ret;
}

3449 3450 3451 3452 3453 3454 3455 3456 3457 3458
/* Maximum number of blocks we map for direct IO at once. */

static int ext4_get_block_dio_write(struct inode *inode, sector_t iblock,
		   struct buffer_head *bh_result, int create)
{
	handle_t *handle = NULL;
	int ret = 0;
	unsigned max_blocks = bh_result->b_size >> inode->i_blkbits;
	int dio_credits;

3459 3460
	ext4_debug("ext4_get_block_dio_write: inode %lu, create flag %d\n",
		   inode->i_ino, create);
3461 3462 3463 3464 3465 3466 3467 3468 3469 3470 3471 3472 3473 3474 3475 3476 3477 3478 3479 3480 3481 3482 3483 3484 3485 3486 3487 3488 3489 3490 3491 3492 3493 3494 3495 3496 3497 3498 3499 3500 3501 3502
	/*
	 * DIO VFS code passes create = 0 flag for write to
	 * the middle of file. It does this to avoid block
	 * allocation for holes, to prevent expose stale data
	 * out when there is parallel buffered read (which does
	 * not hold the i_mutex lock) while direct IO write has
	 * not completed. DIO request on holes finally falls back
	 * to buffered IO for this reason.
	 *
	 * For ext4 extent based file, since we support fallocate,
	 * new allocated extent as uninitialized, for holes, we
	 * could fallocate blocks for holes, thus parallel
	 * buffered IO read will zero out the page when read on
	 * a hole while parallel DIO write to the hole has not completed.
	 *
	 * when we come here, we know it's a direct IO write to
	 * to the middle of file (<i_size)
	 * so it's safe to override the create flag from VFS.
	 */
	create = EXT4_GET_BLOCKS_DIO_CREATE_EXT;

	if (max_blocks > DIO_MAX_BLOCKS)
		max_blocks = DIO_MAX_BLOCKS;
	dio_credits = ext4_chunk_trans_blocks(inode, max_blocks);
	handle = ext4_journal_start(inode, dio_credits);
	if (IS_ERR(handle)) {
		ret = PTR_ERR(handle);
		goto out;
	}
	ret = ext4_get_blocks(handle, inode, iblock, max_blocks, bh_result,
			      create);
	if (ret > 0) {
		bh_result->b_size = (ret << inode->i_blkbits);
		ret = 0;
	}
	ext4_journal_stop(handle);
out:
	return ret;
}

static void ext4_free_io_end(ext4_io_end_t *io)
{
3503 3504
	BUG_ON(!io);
	iput(io->inode);
3505 3506
	kfree(io);
}
3507 3508 3509 3510 3511 3512 3513 3514 3515 3516 3517 3518 3519 3520 3521 3522 3523 3524 3525 3526 3527 3528 3529 3530
static void dump_aio_dio_list(struct inode * inode)
{
#ifdef	EXT4_DEBUG
	struct list_head *cur, *before, *after;
	ext4_io_end_t *io, *io0, *io1;

	if (list_empty(&EXT4_I(inode)->i_aio_dio_complete_list)){
		ext4_debug("inode %lu aio dio list is empty\n", inode->i_ino);
		return;
	}

	ext4_debug("Dump inode %lu aio_dio_completed_IO list \n", inode->i_ino);
	list_for_each_entry(io, &EXT4_I(inode)->i_aio_dio_complete_list, list){
		cur = &io->list;
		before = cur->prev;
		io0 = container_of(before, ext4_io_end_t, list);
		after = cur->next;
		io1 = container_of(after, ext4_io_end_t, list);

		ext4_debug("io 0x%p from inode %lu,prev 0x%p,next 0x%p\n",
			    io, inode->i_ino, io0, io1);
	}
#endif
}
3531 3532 3533 3534

/*
 * check a range of space and convert unwritten extents to written.
 */
3535
static int ext4_end_aio_dio_nolock(ext4_io_end_t *io)
3536 3537 3538 3539 3540 3541
{
	struct inode *inode = io->inode;
	loff_t offset = io->offset;
	size_t size = io->size;
	int ret = 0;

3542 3543 3544 3545 3546 3547 3548 3549 3550 3551
	ext4_debug("end_aio_dio_onlock: io 0x%p from inode %lu,list->next 0x%p,"
		   "list->prev 0x%p\n",
	           io, inode->i_ino, io->list.next, io->list.prev);

	if (list_empty(&io->list))
		return ret;

	if (io->flag != DIO_AIO_UNWRITTEN)
		return ret;

3552 3553 3554
	if (offset + size <= i_size_read(inode))
		ret = ext4_convert_unwritten_extents(inode, offset, size);

3555
	if (ret < 0) {
3556
		printk(KERN_EMERG "%s: failed to convert unwritten"
3557 3558 3559 3560 3561
			"extents to written extents, error is %d"
			" io is still on inode %lu aio dio list\n",
                       __func__, ret, inode->i_ino);
		return ret;
	}
3562

3563 3564 3565
	/* clear the DIO AIO unwritten flag */
	io->flag = 0;
	return ret;
3566
}
3567 3568 3569 3570 3571 3572 3573 3574
/*
 * work on completed aio dio IO, to convert unwritten extents to extents
 */
static void ext4_end_aio_dio_work(struct work_struct *work)
{
	ext4_io_end_t *io  = container_of(work, ext4_io_end_t, work);
	struct inode *inode = io->inode;
	int ret = 0;
3575

3576 3577 3578 3579 3580 3581 3582 3583 3584 3585 3586 3587 3588 3589 3590 3591 3592 3593 3594 3595 3596 3597 3598 3599 3600 3601 3602 3603 3604 3605 3606 3607 3608 3609 3610 3611 3612 3613 3614 3615 3616 3617 3618 3619 3620 3621 3622 3623 3624 3625 3626 3627 3628 3629 3630 3631 3632
	mutex_lock(&inode->i_mutex);
	ret = ext4_end_aio_dio_nolock(io);
	if (ret >= 0) {
		if (!list_empty(&io->list))
			list_del_init(&io->list);
		ext4_free_io_end(io);
	}
	mutex_unlock(&inode->i_mutex);
}
/*
 * This function is called from ext4_sync_file().
 *
 * When AIO DIO IO is completed, the work to convert unwritten
 * extents to written is queued on workqueue but may not get immediately
 * scheduled. When fsync is called, we need to ensure the
 * conversion is complete before fsync returns.
 * The inode keeps track of a list of completed AIO from DIO path
 * that might needs to do the conversion. This function walks through
 * the list and convert the related unwritten extents to written.
 */
int flush_aio_dio_completed_IO(struct inode *inode)
{
	ext4_io_end_t *io;
	int ret = 0;
	int ret2 = 0;

	if (list_empty(&EXT4_I(inode)->i_aio_dio_complete_list))
		return ret;

	dump_aio_dio_list(inode);
	while (!list_empty(&EXT4_I(inode)->i_aio_dio_complete_list)){
		io = list_entry(EXT4_I(inode)->i_aio_dio_complete_list.next,
				ext4_io_end_t, list);
		/*
		 * Calling ext4_end_aio_dio_nolock() to convert completed
		 * IO to written.
		 *
		 * When ext4_sync_file() is called, run_queue() may already
		 * about to flush the work corresponding to this io structure.
		 * It will be upset if it founds the io structure related
		 * to the work-to-be schedule is freed.
		 *
		 * Thus we need to keep the io structure still valid here after
		 * convertion finished. The io structure has a flag to
		 * avoid double converting from both fsync and background work
		 * queue work.
		 */
		ret = ext4_end_aio_dio_nolock(io);
		if (ret < 0)
			ret2 = ret;
		else
			list_del_init(&io->list);
	}
	return (ret2 < 0) ? ret2 : 0;
}

static ext4_io_end_t *ext4_init_io_end (struct inode *inode)
3633 3634 3635 3636 3637 3638
{
	ext4_io_end_t *io = NULL;

	io = kmalloc(sizeof(*io), GFP_NOFS);

	if (io) {
3639
		igrab(inode);
3640
		io->inode = inode;
3641
		io->flag = 0;
3642 3643 3644
		io->offset = 0;
		io->size = 0;
		io->error = 0;
3645 3646
		INIT_WORK(&io->work, ext4_end_aio_dio_work);
		INIT_LIST_HEAD(&io->list);
3647 3648 3649 3650 3651 3652 3653 3654 3655 3656 3657
	}

	return io;
}

static void ext4_end_io_dio(struct kiocb *iocb, loff_t offset,
			    ssize_t size, void *private)
{
        ext4_io_end_t *io_end = iocb->private;
	struct workqueue_struct *wq;

3658 3659 3660 3661 3662 3663 3664 3665 3666 3667 3668 3669
	ext_debug("ext4_end_io_dio(): io_end 0x%p"
		  "for inode %lu, iocb 0x%p, offset %llu, size %llu\n",
 		  iocb->private, io_end->inode->i_ino, iocb, offset,
		  size);
	/* if not async direct IO or dio with 0 bytes write, just return */
	if (!io_end || !size)
		return;

	/* if not aio dio with unwritten extents, just free io and return */
	if (io_end->flag != DIO_AIO_UNWRITTEN){
		ext4_free_io_end(io_end);
		iocb->private = NULL;
3670
		return;
3671 3672
	}

3673 3674 3675 3676
	io_end->offset = offset;
	io_end->size = size;
	wq = EXT4_SB(io_end->inode->i_sb)->dio_unwritten_wq;

3677
	/* queue the work to convert unwritten extents to written */
3678 3679
	queue_work(wq, &io_end->work);

3680 3681 3682
	/* Add the io_end to per-inode completed aio dio list*/
	list_add_tail(&io_end->list,
		 &EXT4_I(io_end->inode)->i_aio_dio_complete_list);
3683 3684 3685 3686 3687 3688 3689 3690 3691 3692 3693
	iocb->private = NULL;
}
/*
 * For ext4 extent files, ext4 will do direct-io write to holes,
 * preallocated extents, and those write extend the file, no need to
 * fall back to buffered IO.
 *
 * For holes, we fallocate those blocks, mark them as unintialized
 * If those blocks were preallocated, we mark sure they are splited, but
 * still keep the range to write as unintialized.
 *
3694 3695 3696 3697
 * The unwrritten extents will be converted to written when DIO is completed.
 * For async direct IO, since the IO may still pending when return, we
 * set up an end_io call back function, which will do the convertion
 * when async direct IO completed.
3698 3699 3700 3701 3702 3703 3704 3705 3706 3707 3708 3709 3710 3711 3712 3713 3714 3715
 *
 * If the O_DIRECT write will extend the file then add this inode to the
 * orphan list.  So recovery will truncate it back to the original size
 * if the machine crashes during the write.
 *
 */
static ssize_t ext4_ext_direct_IO(int rw, struct kiocb *iocb,
			      const struct iovec *iov, loff_t offset,
			      unsigned long nr_segs)
{
	struct file *file = iocb->ki_filp;
	struct inode *inode = file->f_mapping->host;
	ssize_t ret;
	size_t count = iov_length(iov, nr_segs);

	loff_t final_size = offset + count;
	if (rw == WRITE && final_size <= inode->i_size) {
		/*
3716 3717 3718
 		 * We could direct write to holes and fallocate.
		 *
 		 * Allocated blocks to fill the hole are marked as uninitialized
3719 3720
 		 * to prevent paralel buffered read to expose the stale data
 		 * before DIO complete the data IO.
3721 3722
		 *
 		 * As to previously fallocated extents, ext4 get_block
3723 3724 3725
 		 * will just simply mark the buffer mapped but still
 		 * keep the extents uninitialized.
 		 *
3726 3727 3728 3729 3730 3731 3732 3733
		 * for non AIO case, we will convert those unwritten extents
		 * to written after return back from blockdev_direct_IO.
		 *
		 * for async DIO, the conversion needs to be defered when
		 * the IO is completed. The ext4 end_io callback function
		 * will be called to take care of the conversion work.
		 * Here for async case, we allocate an io_end structure to
		 * hook to the iocb.
3734
 		 */
3735 3736 3737 3738 3739 3740 3741 3742 3743 3744 3745 3746 3747 3748 3749 3750
		iocb->private = NULL;
		EXT4_I(inode)->cur_aio_dio = NULL;
		if (!is_sync_kiocb(iocb)) {
			iocb->private = ext4_init_io_end(inode);
			if (!iocb->private)
				return -ENOMEM;
			/*
			 * we save the io structure for current async
			 * direct IO, so that later ext4_get_blocks()
			 * could flag the io structure whether there
			 * is a unwritten extents needs to be converted
			 * when IO is completed.
			 */
			EXT4_I(inode)->cur_aio_dio = iocb->private;
		}

3751 3752 3753 3754 3755
		ret = blockdev_direct_IO(rw, iocb, inode,
					 inode->i_sb->s_bdev, iov,
					 offset, nr_segs,
					 ext4_get_block_dio_write,
					 ext4_end_io_dio);
3756 3757 3758 3759 3760 3761 3762 3763 3764 3765 3766 3767 3768 3769 3770 3771 3772 3773 3774
		if (iocb->private)
			EXT4_I(inode)->cur_aio_dio = NULL;
		/*
		 * The io_end structure takes a reference to the inode,
		 * that structure needs to be destroyed and the
		 * reference to the inode need to be dropped, when IO is
		 * complete, even with 0 byte write, or failed.
		 *
		 * In the successful AIO DIO case, the io_end structure will be
		 * desctroyed and the reference to the inode will be dropped
		 * after the end_io call back function is called.
		 *
		 * In the case there is 0 byte write, or error case, since
		 * VFS direct IO won't invoke the end_io call back function,
		 * we need to free the end_io structure here.
		 */
		if (ret != -EIOCBQUEUED && ret <= 0 && iocb->private) {
			ext4_free_io_end(iocb->private);
			iocb->private = NULL;
3775 3776
		} else if (ret > 0) {
			int err;
3777 3778 3779 3780
			/*
			 * for non AIO case, since the IO is already
			 * completed, we could do the convertion right here
			 */
3781 3782 3783 3784 3785
			err = ext4_convert_unwritten_extents(inode,
							     offset, ret);
			if (err < 0)
				ret = err;
		}
3786 3787
		return ret;
	}
3788 3789

	/* for write the the end of file case, we fall back to old way */
3790 3791 3792 3793 3794 3795 3796 3797 3798 3799 3800 3801 3802 3803 3804 3805
	return ext4_ind_direct_IO(rw, iocb, iov, offset, nr_segs);
}

static ssize_t ext4_direct_IO(int rw, struct kiocb *iocb,
			      const struct iovec *iov, loff_t offset,
			      unsigned long nr_segs)
{
	struct file *file = iocb->ki_filp;
	struct inode *inode = file->f_mapping->host;

	if (EXT4_I(inode)->i_flags & EXT4_EXTENTS_FL)
		return ext4_ext_direct_IO(rw, iocb, iov, offset, nr_segs);

	return ext4_ind_direct_IO(rw, iocb, iov, offset, nr_segs);
}

3806
/*
3807
 * Pages can be marked dirty completely asynchronously from ext4's journalling
3808 3809 3810 3811 3812 3813 3814 3815 3816 3817 3818
 * activity.  By filemap_sync_pte(), try_to_unmap_one(), etc.  We cannot do
 * much here because ->set_page_dirty is called under VFS locks.  The page is
 * not necessarily locked.
 *
 * We cannot just dirty the page and leave attached buffers clean, because the
 * buffers' dirty state is "definitive".  We cannot just set the buffers dirty
 * or jbddirty because all the journalling code will explode.
 *
 * So what we do is to mark the page "pending dirty" and next time writepage
 * is called, propagate that into the buffers appropriately.
 */
3819
static int ext4_journalled_set_page_dirty(struct page *page)
3820 3821 3822 3823 3824
{
	SetPageChecked(page);
	return __set_page_dirty_nobuffers(page);
}

3825
static const struct address_space_operations ext4_ordered_aops = {
3826 3827
	.readpage		= ext4_readpage,
	.readpages		= ext4_readpages,
3828
	.writepage		= ext4_writepage,
3829 3830 3831 3832 3833 3834 3835 3836 3837
	.sync_page		= block_sync_page,
	.write_begin		= ext4_write_begin,
	.write_end		= ext4_ordered_write_end,
	.bmap			= ext4_bmap,
	.invalidatepage		= ext4_invalidatepage,
	.releasepage		= ext4_releasepage,
	.direct_IO		= ext4_direct_IO,
	.migratepage		= buffer_migrate_page,
	.is_partially_uptodate  = block_is_partially_uptodate,
3838
	.error_remove_page	= generic_error_remove_page,
3839 3840
};

3841
static const struct address_space_operations ext4_writeback_aops = {
3842 3843
	.readpage		= ext4_readpage,
	.readpages		= ext4_readpages,
3844
	.writepage		= ext4_writepage,
3845 3846 3847 3848 3849 3850 3851 3852 3853
	.sync_page		= block_sync_page,
	.write_begin		= ext4_write_begin,
	.write_end		= ext4_writeback_write_end,
	.bmap			= ext4_bmap,
	.invalidatepage		= ext4_invalidatepage,
	.releasepage		= ext4_releasepage,
	.direct_IO		= ext4_direct_IO,
	.migratepage		= buffer_migrate_page,
	.is_partially_uptodate  = block_is_partially_uptodate,
3854
	.error_remove_page	= generic_error_remove_page,
3855 3856
};

3857
static const struct address_space_operations ext4_journalled_aops = {
3858 3859
	.readpage		= ext4_readpage,
	.readpages		= ext4_readpages,
3860
	.writepage		= ext4_writepage,
3861 3862 3863 3864 3865 3866 3867 3868
	.sync_page		= block_sync_page,
	.write_begin		= ext4_write_begin,
	.write_end		= ext4_journalled_write_end,
	.set_page_dirty		= ext4_journalled_set_page_dirty,
	.bmap			= ext4_bmap,
	.invalidatepage		= ext4_invalidatepage,
	.releasepage		= ext4_releasepage,
	.is_partially_uptodate  = block_is_partially_uptodate,
3869
	.error_remove_page	= generic_error_remove_page,
3870 3871
};

3872
static const struct address_space_operations ext4_da_aops = {
3873 3874
	.readpage		= ext4_readpage,
	.readpages		= ext4_readpages,
3875
	.writepage		= ext4_writepage,
3876 3877 3878 3879 3880 3881 3882 3883 3884 3885
	.writepages		= ext4_da_writepages,
	.sync_page		= block_sync_page,
	.write_begin		= ext4_da_write_begin,
	.write_end		= ext4_da_write_end,
	.bmap			= ext4_bmap,
	.invalidatepage		= ext4_da_invalidatepage,
	.releasepage		= ext4_releasepage,
	.direct_IO		= ext4_direct_IO,
	.migratepage		= buffer_migrate_page,
	.is_partially_uptodate  = block_is_partially_uptodate,
3886
	.error_remove_page	= generic_error_remove_page,
3887 3888
};

3889
void ext4_set_aops(struct inode *inode)
3890
{
3891 3892 3893 3894
	if (ext4_should_order_data(inode) &&
		test_opt(inode->i_sb, DELALLOC))
		inode->i_mapping->a_ops = &ext4_da_aops;
	else if (ext4_should_order_data(inode))
3895
		inode->i_mapping->a_ops = &ext4_ordered_aops;
3896 3897 3898
	else if (ext4_should_writeback_data(inode) &&
		 test_opt(inode->i_sb, DELALLOC))
		inode->i_mapping->a_ops = &ext4_da_aops;
3899 3900
	else if (ext4_should_writeback_data(inode))
		inode->i_mapping->a_ops = &ext4_writeback_aops;
3901
	else
3902
		inode->i_mapping->a_ops = &ext4_journalled_aops;
3903 3904 3905
}

/*
3906
 * ext4_block_truncate_page() zeroes out a mapping from file offset `from'
3907 3908 3909 3910
 * up to the end of the block which corresponds to `from'.
 * This required during truncate. We need to physically zero the tail end
 * of that block so it doesn't yield old data if the file is later grown.
 */
3911
int ext4_block_truncate_page(handle_t *handle,
3912 3913
		struct address_space *mapping, loff_t from)
{
3914
	ext4_fsblk_t index = from >> PAGE_CACHE_SHIFT;
3915
	unsigned offset = from & (PAGE_CACHE_SIZE-1);
A
Aneesh Kumar K.V 已提交
3916 3917
	unsigned blocksize, length, pos;
	ext4_lblk_t iblock;
3918 3919
	struct inode *inode = mapping->host;
	struct buffer_head *bh;
3920
	struct page *page;
3921 3922
	int err = 0;

3923 3924
	page = find_or_create_page(mapping, from >> PAGE_CACHE_SHIFT,
				   mapping_gfp_mask(mapping) & ~__GFP_FS);
3925 3926 3927
	if (!page)
		return -EINVAL;

3928 3929 3930 3931 3932 3933 3934 3935 3936
	blocksize = inode->i_sb->s_blocksize;
	length = blocksize - (offset & (blocksize - 1));
	iblock = index << (PAGE_CACHE_SHIFT - inode->i_sb->s_blocksize_bits);

	/*
	 * For "nobh" option,  we can only work if we don't need to
	 * read-in the page - otherwise we create buffers to do the IO.
	 */
	if (!page_has_buffers(page) && test_opt(inode->i_sb, NOBH) &&
3937
	     ext4_should_writeback_data(inode) && PageUptodate(page)) {
3938
		zero_user(page, offset, length);
3939 3940 3941 3942 3943 3944 3945 3946 3947 3948 3949 3950 3951 3952 3953 3954 3955 3956 3957 3958 3959 3960 3961 3962
		set_page_dirty(page);
		goto unlock;
	}

	if (!page_has_buffers(page))
		create_empty_buffers(page, blocksize, 0);

	/* Find the buffer that contains "offset" */
	bh = page_buffers(page);
	pos = blocksize;
	while (offset >= pos) {
		bh = bh->b_this_page;
		iblock++;
		pos += blocksize;
	}

	err = 0;
	if (buffer_freed(bh)) {
		BUFFER_TRACE(bh, "freed: skip");
		goto unlock;
	}

	if (!buffer_mapped(bh)) {
		BUFFER_TRACE(bh, "unmapped");
3963
		ext4_get_block(inode, iblock, bh, 0);
3964 3965 3966 3967 3968 3969 3970 3971 3972 3973 3974 3975 3976 3977 3978 3979 3980 3981 3982 3983
		/* unmapped? It's a hole - nothing to do */
		if (!buffer_mapped(bh)) {
			BUFFER_TRACE(bh, "still unmapped");
			goto unlock;
		}
	}

	/* Ok, it's mapped. Make sure it's up-to-date */
	if (PageUptodate(page))
		set_buffer_uptodate(bh);

	if (!buffer_uptodate(bh)) {
		err = -EIO;
		ll_rw_block(READ, 1, &bh);
		wait_on_buffer(bh);
		/* Uhhuh. Read error. Complain and punt. */
		if (!buffer_uptodate(bh))
			goto unlock;
	}

3984
	if (ext4_should_journal_data(inode)) {
3985
		BUFFER_TRACE(bh, "get write access");
3986
		err = ext4_journal_get_write_access(handle, bh);
3987 3988 3989 3990
		if (err)
			goto unlock;
	}

3991
	zero_user(page, offset, length);
3992 3993 3994 3995

	BUFFER_TRACE(bh, "zeroed end of block");

	err = 0;
3996
	if (ext4_should_journal_data(inode)) {
3997
		err = ext4_handle_dirty_metadata(handle, inode, bh);
3998
	} else {
3999
		if (ext4_should_order_data(inode))
4000
			err = ext4_jbd2_file_inode(handle, inode);
4001 4002 4003 4004 4005 4006 4007 4008 4009 4010 4011 4012 4013 4014 4015 4016 4017 4018 4019 4020 4021 4022 4023
		mark_buffer_dirty(bh);
	}

unlock:
	unlock_page(page);
	page_cache_release(page);
	return err;
}

/*
 * Probably it should be a library function... search for first non-zero word
 * or memcmp with zero_page, whatever is better for particular architecture.
 * Linus?
 */
static inline int all_zeroes(__le32 *p, __le32 *q)
{
	while (p < q)
		if (*p++)
			return 0;
	return 1;
}

/**
4024
 *	ext4_find_shared - find the indirect blocks for partial truncation.
4025 4026
 *	@inode:	  inode in question
 *	@depth:	  depth of the affected branch
4027
 *	@offsets: offsets of pointers in that branch (see ext4_block_to_path)
4028 4029 4030
 *	@chain:	  place to store the pointers to partial indirect blocks
 *	@top:	  place to the (detached) top of branch
 *
4031
 *	This is a helper function used by ext4_truncate().
4032 4033 4034 4035 4036 4037 4038
 *
 *	When we do truncate() we may have to clean the ends of several
 *	indirect blocks but leave the blocks themselves alive. Block is
 *	partially truncated if some data below the new i_size is refered
 *	from it (and it is on the path to the first completely truncated
 *	data block, indeed).  We have to free the top of that path along
 *	with everything to the right of the path. Since no allocation
4039
 *	past the truncation point is possible until ext4_truncate()
4040 4041 4042 4043 4044 4045 4046 4047 4048 4049 4050 4051 4052 4053 4054 4055 4056 4057
 *	finishes, we may safely do the latter, but top of branch may
 *	require special attention - pageout below the truncation point
 *	might try to populate it.
 *
 *	We atomically detach the top of branch from the tree, store the
 *	block number of its root in *@top, pointers to buffer_heads of
 *	partially truncated blocks - in @chain[].bh and pointers to
 *	their last elements that should not be removed - in
 *	@chain[].p. Return value is the pointer to last filled element
 *	of @chain.
 *
 *	The work left to caller to do the actual freeing of subtrees:
 *		a) free the subtree starting from *@top
 *		b) free the subtrees whose roots are stored in
 *			(@chain[i].p+1 .. end of @chain[i].bh->b_data)
 *		c) free the subtrees growing from the inode past the @chain[0].
 *			(no partially truncated stuff there).  */

4058
static Indirect *ext4_find_shared(struct inode *inode, int depth,
4059 4060
				  ext4_lblk_t offsets[4], Indirect chain[4],
				  __le32 *top)
4061 4062 4063 4064 4065 4066 4067 4068
{
	Indirect *partial, *p;
	int k, err;

	*top = 0;
	/* Make k index the deepest non-null offest + 1 */
	for (k = depth; k > 1 && !offsets[k-1]; k--)
		;
4069
	partial = ext4_get_branch(inode, k, offsets, chain, &err);
4070 4071 4072 4073 4074 4075 4076 4077 4078 4079
	/* Writer: pointers */
	if (!partial)
		partial = chain + k-1;
	/*
	 * If the branch acquired continuation since we've looked at it -
	 * fine, it should all survive and (new) top doesn't belong to us.
	 */
	if (!partial->key && *partial->p)
		/* Writer: end */
		goto no_top;
4080
	for (p = partial; (p > chain) && all_zeroes((__le32 *) p->bh->b_data, p->p); p--)
4081 4082 4083 4084 4085 4086 4087 4088 4089 4090 4091
		;
	/*
	 * OK, we've found the last block that must survive. The rest of our
	 * branch should be detached before unlocking. However, if that rest
	 * of branch is all ours and does not grow immediately from the inode
	 * it's easier to cheat and just decrement partial->p.
	 */
	if (p == chain + k - 1 && p > chain) {
		p->p--;
	} else {
		*top = *p->p;
4092
		/* Nope, don't do this in ext4.  Must leave the tree intact */
4093 4094 4095 4096 4097 4098
#if 0
		*p->p = 0;
#endif
	}
	/* Writer: end */

4099
	while (partial > p) {
4100 4101 4102 4103 4104 4105 4106 4107 4108 4109 4110 4111 4112 4113 4114
		brelse(partial->bh);
		partial--;
	}
no_top:
	return partial;
}

/*
 * Zero a number of block pointers in either an inode or an indirect block.
 * If we restart the transaction we must again get write access to the
 * indirect block for further modification.
 *
 * We release `count' blocks on disk, but (last - first) may be greater
 * than `count' because there can be holes in there.
 */
4115
static void ext4_clear_blocks(handle_t *handle, struct inode *inode,
4116 4117 4118 4119
			      struct buffer_head *bh,
			      ext4_fsblk_t block_to_free,
			      unsigned long count, __le32 *first,
			      __le32 *last)
4120 4121 4122 4123
{
	__le32 *p;
	if (try_to_extend_transaction(handle, inode)) {
		if (bh) {
4124 4125
			BUFFER_TRACE(bh, "call ext4_handle_dirty_metadata");
			ext4_handle_dirty_metadata(handle, inode, bh);
4126
		}
4127
		ext4_mark_inode_dirty(handle, inode);
4128 4129
		ext4_truncate_restart_trans(handle, inode,
					    blocks_for_truncate(inode));
4130 4131
		if (bh) {
			BUFFER_TRACE(bh, "retaking write access");
4132
			ext4_journal_get_write_access(handle, bh);
4133 4134 4135 4136
		}
	}

	/*
4137 4138 4139 4140 4141
	 * Any buffers which are on the journal will be in memory. We
	 * find them on the hash table so jbd2_journal_revoke() will
	 * run jbd2_journal_forget() on them.  We've already detached
	 * each block from the file, so bforget() in
	 * jbd2_journal_forget() should be safe.
4142
	 *
4143
	 * AKPM: turn on bforget in jbd2_journal_forget()!!!
4144 4145 4146 4147
	 */
	for (p = first; p < last; p++) {
		u32 nr = le32_to_cpu(*p);
		if (nr) {
A
Aneesh Kumar K.V 已提交
4148
			struct buffer_head *tbh;
4149 4150

			*p = 0;
A
Aneesh Kumar K.V 已提交
4151 4152
			tbh = sb_find_get_block(inode->i_sb, nr);
			ext4_forget(handle, 0, inode, tbh, nr);
4153 4154 4155
		}
	}

4156
	ext4_free_blocks(handle, inode, block_to_free, count, 0);
4157 4158 4159
}

/**
4160
 * ext4_free_data - free a list of data blocks
4161 4162 4163 4164 4165 4166 4167 4168 4169 4170 4171 4172 4173 4174 4175 4176 4177
 * @handle:	handle for this transaction
 * @inode:	inode we are dealing with
 * @this_bh:	indirect buffer_head which contains *@first and *@last
 * @first:	array of block numbers
 * @last:	points immediately past the end of array
 *
 * We are freeing all blocks refered from that array (numbers are stored as
 * little-endian 32-bit) and updating @inode->i_blocks appropriately.
 *
 * We accumulate contiguous runs of blocks to free.  Conveniently, if these
 * blocks are contiguous then releasing them at one time will only affect one
 * or two bitmap blocks (+ group descriptor(s) and superblock) and we won't
 * actually use a lot of journal space.
 *
 * @this_bh will be %NULL if @first and @last point into the inode's direct
 * block pointers.
 */
4178
static void ext4_free_data(handle_t *handle, struct inode *inode,
4179 4180 4181
			   struct buffer_head *this_bh,
			   __le32 *first, __le32 *last)
{
4182
	ext4_fsblk_t block_to_free = 0;    /* Starting block # of a run */
4183 4184 4185 4186
	unsigned long count = 0;	    /* Number of blocks in the run */
	__le32 *block_to_free_p = NULL;	    /* Pointer into inode/ind
					       corresponding to
					       block_to_free */
4187
	ext4_fsblk_t nr;		    /* Current block # */
4188 4189 4190 4191 4192 4193
	__le32 *p;			    /* Pointer into inode/ind
					       for current block */
	int err;

	if (this_bh) {				/* For indirect block */
		BUFFER_TRACE(this_bh, "get_write_access");
4194
		err = ext4_journal_get_write_access(handle, this_bh);
4195 4196 4197 4198 4199 4200 4201 4202 4203 4204 4205 4206 4207 4208 4209 4210 4211
		/* Important: if we can't update the indirect pointers
		 * to the blocks, we can't free them. */
		if (err)
			return;
	}

	for (p = first; p < last; p++) {
		nr = le32_to_cpu(*p);
		if (nr) {
			/* accumulate blocks to free if they're contiguous */
			if (count == 0) {
				block_to_free = nr;
				block_to_free_p = p;
				count = 1;
			} else if (nr == block_to_free + count) {
				count++;
			} else {
4212
				ext4_clear_blocks(handle, inode, this_bh,
4213 4214 4215 4216 4217 4218 4219 4220 4221 4222
						  block_to_free,
						  count, block_to_free_p, p);
				block_to_free = nr;
				block_to_free_p = p;
				count = 1;
			}
		}
	}

	if (count > 0)
4223
		ext4_clear_blocks(handle, inode, this_bh, block_to_free,
4224 4225 4226
				  count, block_to_free_p, p);

	if (this_bh) {
4227
		BUFFER_TRACE(this_bh, "call ext4_handle_dirty_metadata");
4228 4229 4230 4231 4232 4233 4234

		/*
		 * The buffer head should have an attached journal head at this
		 * point. However, if the data is corrupted and an indirect
		 * block pointed to itself, it would have been detached when
		 * the block was cleared. Check for this instead of OOPSing.
		 */
4235
		if ((EXT4_JOURNAL(inode) == NULL) || bh2jh(this_bh))
4236
			ext4_handle_dirty_metadata(handle, inode, this_bh);
4237 4238 4239 4240 4241 4242
		else
			ext4_error(inode->i_sb, __func__,
				   "circular indirect block detected, "
				   "inode=%lu, block=%llu",
				   inode->i_ino,
				   (unsigned long long) this_bh->b_blocknr);
4243 4244 4245 4246
	}
}

/**
4247
 *	ext4_free_branches - free an array of branches
4248 4249 4250 4251 4252 4253 4254 4255 4256 4257 4258
 *	@handle: JBD handle for this transaction
 *	@inode:	inode we are dealing with
 *	@parent_bh: the buffer_head which contains *@first and *@last
 *	@first:	array of block numbers
 *	@last:	pointer immediately past the end of array
 *	@depth:	depth of the branches to free
 *
 *	We are freeing all blocks refered from these branches (numbers are
 *	stored as little-endian 32-bit) and updating @inode->i_blocks
 *	appropriately.
 */
4259
static void ext4_free_branches(handle_t *handle, struct inode *inode,
4260 4261 4262
			       struct buffer_head *parent_bh,
			       __le32 *first, __le32 *last, int depth)
{
4263
	ext4_fsblk_t nr;
4264 4265
	__le32 *p;

4266
	if (ext4_handle_is_aborted(handle))
4267 4268 4269 4270
		return;

	if (depth--) {
		struct buffer_head *bh;
4271
		int addr_per_block = EXT4_ADDR_PER_BLOCK(inode->i_sb);
4272 4273 4274 4275 4276 4277 4278 4279 4280 4281 4282 4283 4284 4285
		p = last;
		while (--p >= first) {
			nr = le32_to_cpu(*p);
			if (!nr)
				continue;		/* A hole */

			/* Go read the buffer for the next level down */
			bh = sb_bread(inode->i_sb, nr);

			/*
			 * A read failure? Report error and clear slot
			 * (should be rare).
			 */
			if (!bh) {
4286
				ext4_error(inode->i_sb, "ext4_free_branches",
4287
					   "Read failure, inode=%lu, block=%llu",
4288 4289 4290 4291 4292 4293
					   inode->i_ino, nr);
				continue;
			}

			/* This zaps the entire block.  Bottom up. */
			BUFFER_TRACE(bh, "free child branches");
4294
			ext4_free_branches(handle, inode, bh,
4295 4296 4297
					(__le32 *) bh->b_data,
					(__le32 *) bh->b_data + addr_per_block,
					depth);
4298 4299 4300 4301 4302

			/*
			 * We've probably journalled the indirect block several
			 * times during the truncate.  But it's no longer
			 * needed and we now drop it from the transaction via
4303
			 * jbd2_journal_revoke().
4304 4305 4306
			 *
			 * That's easy if it's exclusively part of this
			 * transaction.  But if it's part of the committing
4307
			 * transaction then jbd2_journal_forget() will simply
4308
			 * brelse() it.  That means that if the underlying
4309
			 * block is reallocated in ext4_get_block(),
4310 4311 4312 4313 4314 4315 4316 4317
			 * unmap_underlying_metadata() will find this block
			 * and will try to get rid of it.  damn, damn.
			 *
			 * If this block has already been committed to the
			 * journal, a revoke record will be written.  And
			 * revoke records must be emitted *before* clearing
			 * this block's bit in the bitmaps.
			 */
4318
			ext4_forget(handle, 1, inode, bh, bh->b_blocknr);
4319 4320 4321 4322 4323 4324 4325 4326 4327 4328 4329 4330 4331 4332 4333 4334 4335

			/*
			 * Everything below this this pointer has been
			 * released.  Now let this top-of-subtree go.
			 *
			 * We want the freeing of this indirect block to be
			 * atomic in the journal with the updating of the
			 * bitmap block which owns it.  So make some room in
			 * the journal.
			 *
			 * We zero the parent pointer *after* freeing its
			 * pointee in the bitmaps, so if extend_transaction()
			 * for some reason fails to put the bitmap changes and
			 * the release into the same transaction, recovery
			 * will merely complain about releasing a free block,
			 * rather than leaking blocks.
			 */
4336
			if (ext4_handle_is_aborted(handle))
4337 4338
				return;
			if (try_to_extend_transaction(handle, inode)) {
4339
				ext4_mark_inode_dirty(handle, inode);
4340 4341
				ext4_truncate_restart_trans(handle, inode,
					    blocks_for_truncate(inode));
4342 4343
			}

4344
			ext4_free_blocks(handle, inode, nr, 1, 1);
4345 4346 4347 4348 4349 4350 4351

			if (parent_bh) {
				/*
				 * The block which we have just freed is
				 * pointed to by an indirect block: journal it
				 */
				BUFFER_TRACE(parent_bh, "get_write_access");
4352
				if (!ext4_journal_get_write_access(handle,
4353 4354 4355
								   parent_bh)){
					*p = 0;
					BUFFER_TRACE(parent_bh,
4356 4357 4358 4359
					"call ext4_handle_dirty_metadata");
					ext4_handle_dirty_metadata(handle,
								   inode,
								   parent_bh);
4360 4361 4362 4363 4364 4365
				}
			}
		}
	} else {
		/* We have reached the bottom of the tree. */
		BUFFER_TRACE(parent_bh, "free data blocks");
4366
		ext4_free_data(handle, inode, parent_bh, first, last);
4367 4368 4369
	}
}

4370 4371 4372 4373 4374 4375 4376 4377 4378 4379 4380 4381 4382
int ext4_can_truncate(struct inode *inode)
{
	if (IS_APPEND(inode) || IS_IMMUTABLE(inode))
		return 0;
	if (S_ISREG(inode->i_mode))
		return 1;
	if (S_ISDIR(inode->i_mode))
		return 1;
	if (S_ISLNK(inode->i_mode))
		return !ext4_inode_is_fast_symlink(inode);
	return 0;
}

4383
/*
4384
 * ext4_truncate()
4385
 *
4386 4387
 * We block out ext4_get_block() block instantiations across the entire
 * transaction, and VFS/VM ensures that ext4_truncate() cannot run
4388 4389 4390 4391 4392 4393 4394 4395 4396 4397 4398 4399 4400 4401 4402 4403
 * simultaneously on behalf of the same inode.
 *
 * As we work through the truncate and commmit bits of it to the journal there
 * is one core, guiding principle: the file's tree must always be consistent on
 * disk.  We must be able to restart the truncate after a crash.
 *
 * The file's tree may be transiently inconsistent in memory (although it
 * probably isn't), but whenever we close off and commit a journal transaction,
 * the contents of (the filesystem + the journal) must be consistent and
 * restartable.  It's pretty simple, really: bottom up, right to left (although
 * left-to-right works OK too).
 *
 * Note that at recovery time, journal replay occurs *before* the restart of
 * truncate against the orphan inode list.
 *
 * The committed inode has the new, desired i_size (which is the same as
4404
 * i_disksize in this case).  After a crash, ext4_orphan_cleanup() will see
4405
 * that this inode's truncate did not complete and it will again call
4406 4407
 * ext4_truncate() to have another go.  So there will be instantiated blocks
 * to the right of the truncation point in a crashed ext4 filesystem.  But
4408
 * that's fine - as long as they are linked from the inode, the post-crash
4409
 * ext4_truncate() run will find them and release them.
4410
 */
4411
void ext4_truncate(struct inode *inode)
4412 4413
{
	handle_t *handle;
4414
	struct ext4_inode_info *ei = EXT4_I(inode);
4415
	__le32 *i_data = ei->i_data;
4416
	int addr_per_block = EXT4_ADDR_PER_BLOCK(inode->i_sb);
4417
	struct address_space *mapping = inode->i_mapping;
A
Aneesh Kumar K.V 已提交
4418
	ext4_lblk_t offsets[4];
4419 4420 4421 4422
	Indirect chain[4];
	Indirect *partial;
	__le32 nr = 0;
	int n;
A
Aneesh Kumar K.V 已提交
4423
	ext4_lblk_t last_block;
4424 4425
	unsigned blocksize = inode->i_sb->s_blocksize;

4426
	if (!ext4_can_truncate(inode))
4427 4428
		return;

4429
	if (inode->i_size == 0 && !test_opt(inode->i_sb, NO_AUTO_DA_ALLOC))
4430 4431
		ei->i_state |= EXT4_STATE_DA_ALLOC_CLOSE;

A
Aneesh Kumar K.V 已提交
4432
	if (EXT4_I(inode)->i_flags & EXT4_EXTENTS_FL) {
4433
		ext4_ext_truncate(inode);
A
Aneesh Kumar K.V 已提交
4434 4435
		return;
	}
A
Alex Tomas 已提交
4436

4437
	handle = start_transaction(inode);
4438
	if (IS_ERR(handle))
4439 4440 4441
		return;		/* AKPM: return what? */

	last_block = (inode->i_size + blocksize-1)
4442
					>> EXT4_BLOCK_SIZE_BITS(inode->i_sb);
4443

4444 4445 4446
	if (inode->i_size & (blocksize - 1))
		if (ext4_block_truncate_page(handle, mapping, inode->i_size))
			goto out_stop;
4447

4448
	n = ext4_block_to_path(inode, last_block, offsets, NULL);
4449 4450 4451 4452 4453 4454 4455 4456 4457 4458 4459 4460
	if (n == 0)
		goto out_stop;	/* error */

	/*
	 * OK.  This truncate is going to happen.  We add the inode to the
	 * orphan list, so that if this truncate spans multiple transactions,
	 * and we crash, we will resume the truncate when the filesystem
	 * recovers.  It also marks the inode dirty, to catch the new size.
	 *
	 * Implication: the file must always be in a sane, consistent
	 * truncatable state while each transaction commits.
	 */
4461
	if (ext4_orphan_add(handle, inode))
4462 4463
		goto out_stop;

4464 4465 4466 4467 4468
	/*
	 * From here we block out all ext4_get_block() callers who want to
	 * modify the block allocation tree.
	 */
	down_write(&ei->i_data_sem);
4469

4470
	ext4_discard_preallocations(inode);
4471

4472 4473 4474 4475 4476
	/*
	 * The orphan list entry will now protect us from any crash which
	 * occurs before the truncate completes, so it is now safe to propagate
	 * the new, shorter inode size (held for now in i_size) into the
	 * on-disk inode. We do this via i_disksize, which is the value which
4477
	 * ext4 *really* writes onto the disk inode.
4478 4479 4480 4481
	 */
	ei->i_disksize = inode->i_size;

	if (n == 1) {		/* direct blocks */
4482 4483
		ext4_free_data(handle, inode, NULL, i_data+offsets[0],
			       i_data + EXT4_NDIR_BLOCKS);
4484 4485 4486
		goto do_indirects;
	}

4487
	partial = ext4_find_shared(inode, n, offsets, chain, &nr);
4488 4489 4490 4491
	/* Kill the top of shared branch (not detached) */
	if (nr) {
		if (partial == chain) {
			/* Shared branch grows from the inode */
4492
			ext4_free_branches(handle, inode, NULL,
4493 4494 4495 4496 4497 4498 4499 4500 4501
					   &nr, &nr+1, (chain+n-1) - partial);
			*partial->p = 0;
			/*
			 * We mark the inode dirty prior to restart,
			 * and prior to stop.  No need for it here.
			 */
		} else {
			/* Shared branch grows from an indirect block */
			BUFFER_TRACE(partial->bh, "get_write_access");
4502
			ext4_free_branches(handle, inode, partial->bh,
4503 4504 4505 4506 4507 4508
					partial->p,
					partial->p+1, (chain+n-1) - partial);
		}
	}
	/* Clear the ends of indirect blocks on the shared branch */
	while (partial > chain) {
4509
		ext4_free_branches(handle, inode, partial->bh, partial->p + 1,
4510 4511 4512
				   (__le32*)partial->bh->b_data+addr_per_block,
				   (chain+n-1) - partial);
		BUFFER_TRACE(partial->bh, "call brelse");
4513
		brelse(partial->bh);
4514 4515 4516 4517 4518 4519
		partial--;
	}
do_indirects:
	/* Kill the remaining (whole) subtrees */
	switch (offsets[0]) {
	default:
4520
		nr = i_data[EXT4_IND_BLOCK];
4521
		if (nr) {
4522 4523
			ext4_free_branches(handle, inode, NULL, &nr, &nr+1, 1);
			i_data[EXT4_IND_BLOCK] = 0;
4524
		}
4525 4526
	case EXT4_IND_BLOCK:
		nr = i_data[EXT4_DIND_BLOCK];
4527
		if (nr) {
4528 4529
			ext4_free_branches(handle, inode, NULL, &nr, &nr+1, 2);
			i_data[EXT4_DIND_BLOCK] = 0;
4530
		}
4531 4532
	case EXT4_DIND_BLOCK:
		nr = i_data[EXT4_TIND_BLOCK];
4533
		if (nr) {
4534 4535
			ext4_free_branches(handle, inode, NULL, &nr, &nr+1, 3);
			i_data[EXT4_TIND_BLOCK] = 0;
4536
		}
4537
	case EXT4_TIND_BLOCK:
4538 4539 4540
		;
	}

4541
	up_write(&ei->i_data_sem);
K
Kalpak Shah 已提交
4542
	inode->i_mtime = inode->i_ctime = ext4_current_time(inode);
4543
	ext4_mark_inode_dirty(handle, inode);
4544 4545 4546 4547 4548 4549

	/*
	 * In a multi-transaction truncate, we only make the final transaction
	 * synchronous
	 */
	if (IS_SYNC(inode))
4550
		ext4_handle_sync(handle);
4551 4552 4553 4554 4555
out_stop:
	/*
	 * If this was a simple ftruncate(), and the file will remain alive
	 * then we need to clear up the orphan record which we created above.
	 * However, if this was a real unlink then we were called by
4556
	 * ext4_delete_inode(), and we allow that function to clean up the
4557 4558 4559
	 * orphan info for us.
	 */
	if (inode->i_nlink)
4560
		ext4_orphan_del(handle, inode);
4561

4562
	ext4_journal_stop(handle);
4563 4564 4565
}

/*
4566
 * ext4_get_inode_loc returns with an extra refcount against the inode's
4567 4568 4569 4570
 * underlying buffer_head on success. If 'in_mem' is true, we have all
 * data in memory that is needed to recreate the on-disk version of this
 * inode.
 */
4571 4572
static int __ext4_get_inode_loc(struct inode *inode,
				struct ext4_iloc *iloc, int in_mem)
4573
{
4574 4575 4576 4577 4578 4579
	struct ext4_group_desc	*gdp;
	struct buffer_head	*bh;
	struct super_block	*sb = inode->i_sb;
	ext4_fsblk_t		block;
	int			inodes_per_block, inode_offset;

A
Aneesh Kumar K.V 已提交
4580
	iloc->bh = NULL;
4581 4582
	if (!ext4_valid_inum(sb, inode->i_ino))
		return -EIO;
4583

4584 4585 4586
	iloc->block_group = (inode->i_ino - 1) / EXT4_INODES_PER_GROUP(sb);
	gdp = ext4_get_group_desc(sb, iloc->block_group, NULL);
	if (!gdp)
4587 4588
		return -EIO;

4589 4590 4591 4592 4593 4594 4595 4596 4597 4598
	/*
	 * Figure out the offset within the block group inode table
	 */
	inodes_per_block = (EXT4_BLOCK_SIZE(sb) / EXT4_INODE_SIZE(sb));
	inode_offset = ((inode->i_ino - 1) %
			EXT4_INODES_PER_GROUP(sb));
	block = ext4_inode_table(sb, gdp) + (inode_offset / inodes_per_block);
	iloc->offset = (inode_offset % inodes_per_block) * EXT4_INODE_SIZE(sb);

	bh = sb_getblk(sb, block);
4599
	if (!bh) {
4600 4601 4602
		ext4_error(sb, "ext4_get_inode_loc", "unable to read "
			   "inode block - inode=%lu, block=%llu",
			   inode->i_ino, block);
4603 4604 4605 4606
		return -EIO;
	}
	if (!buffer_uptodate(bh)) {
		lock_buffer(bh);
4607 4608 4609 4610 4611 4612 4613 4614 4615 4616

		/*
		 * If the buffer has the write error flag, we have failed
		 * to write out another inode in the same block.  In this
		 * case, we don't have to read the block because we may
		 * read the old inode data successfully.
		 */
		if (buffer_write_io_error(bh) && !buffer_uptodate(bh))
			set_buffer_uptodate(bh);

4617 4618 4619 4620 4621 4622 4623 4624 4625 4626 4627 4628 4629
		if (buffer_uptodate(bh)) {
			/* someone brought it uptodate while we waited */
			unlock_buffer(bh);
			goto has_buffer;
		}

		/*
		 * If we have all information of the inode in memory and this
		 * is the only valid inode in the block, we need not read the
		 * block.
		 */
		if (in_mem) {
			struct buffer_head *bitmap_bh;
4630
			int i, start;
4631

4632
			start = inode_offset & ~(inodes_per_block - 1);
4633

4634 4635
			/* Is the inode bitmap in cache? */
			bitmap_bh = sb_getblk(sb, ext4_inode_bitmap(sb, gdp));
4636 4637 4638 4639 4640 4641 4642 4643 4644 4645 4646 4647
			if (!bitmap_bh)
				goto make_io;

			/*
			 * If the inode bitmap isn't in cache then the
			 * optimisation may end up performing two reads instead
			 * of one, so skip it.
			 */
			if (!buffer_uptodate(bitmap_bh)) {
				brelse(bitmap_bh);
				goto make_io;
			}
4648
			for (i = start; i < start + inodes_per_block; i++) {
4649 4650
				if (i == inode_offset)
					continue;
4651
				if (ext4_test_bit(i, bitmap_bh->b_data))
4652 4653 4654
					break;
			}
			brelse(bitmap_bh);
4655
			if (i == start + inodes_per_block) {
4656 4657 4658 4659 4660 4661 4662 4663 4664
				/* all other inodes are free, so skip I/O */
				memset(bh->b_data, 0, bh->b_size);
				set_buffer_uptodate(bh);
				unlock_buffer(bh);
				goto has_buffer;
			}
		}

make_io:
4665 4666 4667 4668 4669 4670 4671 4672 4673
		/*
		 * If we need to do any I/O, try to pre-readahead extra
		 * blocks from the inode table.
		 */
		if (EXT4_SB(sb)->s_inode_readahead_blks) {
			ext4_fsblk_t b, end, table;
			unsigned num;

			table = ext4_inode_table(sb, gdp);
T
Theodore Ts'o 已提交
4674
			/* s_inode_readahead_blks is always a power of 2 */
4675 4676 4677 4678 4679 4680 4681
			b = block & ~(EXT4_SB(sb)->s_inode_readahead_blks-1);
			if (table > b)
				b = table;
			end = b + EXT4_SB(sb)->s_inode_readahead_blks;
			num = EXT4_INODES_PER_GROUP(sb);
			if (EXT4_HAS_RO_COMPAT_FEATURE(sb,
				       EXT4_FEATURE_RO_COMPAT_GDT_CSUM))
4682
				num -= ext4_itable_unused_count(sb, gdp);
4683 4684 4685 4686 4687 4688 4689
			table += num / inodes_per_block;
			if (end > table)
				end = table;
			while (b <= end)
				sb_breadahead(sb, b++);
		}

4690 4691 4692 4693 4694 4695 4696 4697 4698 4699
		/*
		 * There are other valid inodes in the buffer, this inode
		 * has in-inode xattrs, or we don't have this inode in memory.
		 * Read the block from disk.
		 */
		get_bh(bh);
		bh->b_end_io = end_buffer_read_sync;
		submit_bh(READ_META, bh);
		wait_on_buffer(bh);
		if (!buffer_uptodate(bh)) {
4700 4701 4702
			ext4_error(sb, __func__,
				   "unable to read inode block - inode=%lu, "
				   "block=%llu", inode->i_ino, block);
4703 4704 4705 4706 4707 4708 4709 4710 4711
			brelse(bh);
			return -EIO;
		}
	}
has_buffer:
	iloc->bh = bh;
	return 0;
}

4712
int ext4_get_inode_loc(struct inode *inode, struct ext4_iloc *iloc)
4713 4714
{
	/* We have all inode data except xattrs in memory here. */
4715 4716
	return __ext4_get_inode_loc(inode, iloc,
		!(EXT4_I(inode)->i_state & EXT4_STATE_XATTR));
4717 4718
}

4719
void ext4_set_inode_flags(struct inode *inode)
4720
{
4721
	unsigned int flags = EXT4_I(inode)->i_flags;
4722 4723

	inode->i_flags &= ~(S_SYNC|S_APPEND|S_IMMUTABLE|S_NOATIME|S_DIRSYNC);
4724
	if (flags & EXT4_SYNC_FL)
4725
		inode->i_flags |= S_SYNC;
4726
	if (flags & EXT4_APPEND_FL)
4727
		inode->i_flags |= S_APPEND;
4728
	if (flags & EXT4_IMMUTABLE_FL)
4729
		inode->i_flags |= S_IMMUTABLE;
4730
	if (flags & EXT4_NOATIME_FL)
4731
		inode->i_flags |= S_NOATIME;
4732
	if (flags & EXT4_DIRSYNC_FL)
4733 4734 4735
		inode->i_flags |= S_DIRSYNC;
}

4736 4737 4738 4739 4740 4741 4742 4743 4744 4745 4746 4747 4748 4749 4750 4751 4752 4753
/* Propagate flags from i_flags to EXT4_I(inode)->i_flags */
void ext4_get_inode_flags(struct ext4_inode_info *ei)
{
	unsigned int flags = ei->vfs_inode.i_flags;

	ei->i_flags &= ~(EXT4_SYNC_FL|EXT4_APPEND_FL|
			EXT4_IMMUTABLE_FL|EXT4_NOATIME_FL|EXT4_DIRSYNC_FL);
	if (flags & S_SYNC)
		ei->i_flags |= EXT4_SYNC_FL;
	if (flags & S_APPEND)
		ei->i_flags |= EXT4_APPEND_FL;
	if (flags & S_IMMUTABLE)
		ei->i_flags |= EXT4_IMMUTABLE_FL;
	if (flags & S_NOATIME)
		ei->i_flags |= EXT4_NOATIME_FL;
	if (flags & S_DIRSYNC)
		ei->i_flags |= EXT4_DIRSYNC_FL;
}
4754

4755
static blkcnt_t ext4_inode_blocks(struct ext4_inode *raw_inode,
4756
				  struct ext4_inode_info *ei)
4757 4758
{
	blkcnt_t i_blocks ;
A
Aneesh Kumar K.V 已提交
4759 4760
	struct inode *inode = &(ei->vfs_inode);
	struct super_block *sb = inode->i_sb;
4761 4762 4763 4764 4765 4766

	if (EXT4_HAS_RO_COMPAT_FEATURE(sb,
				EXT4_FEATURE_RO_COMPAT_HUGE_FILE)) {
		/* we are using combined 48 bit field */
		i_blocks = ((u64)le16_to_cpu(raw_inode->i_blocks_high)) << 32 |
					le32_to_cpu(raw_inode->i_blocks_lo);
A
Aneesh Kumar K.V 已提交
4767 4768 4769 4770 4771 4772
		if (ei->i_flags & EXT4_HUGE_FILE_FL) {
			/* i_blocks represent file system block size */
			return i_blocks  << (inode->i_blkbits - 9);
		} else {
			return i_blocks;
		}
4773 4774 4775 4776
	} else {
		return le32_to_cpu(raw_inode->i_blocks_lo);
	}
}
4777

4778
struct inode *ext4_iget(struct super_block *sb, unsigned long ino)
4779
{
4780 4781
	struct ext4_iloc iloc;
	struct ext4_inode *raw_inode;
4782
	struct ext4_inode_info *ei;
4783
	struct buffer_head *bh;
4784 4785
	struct inode *inode;
	long ret;
4786 4787
	int block;

4788 4789 4790 4791 4792 4793 4794
	inode = iget_locked(sb, ino);
	if (!inode)
		return ERR_PTR(-ENOMEM);
	if (!(inode->i_state & I_NEW))
		return inode;

	ei = EXT4_I(inode);
4795

4796 4797
	ret = __ext4_get_inode_loc(inode, &iloc, 0);
	if (ret < 0)
4798 4799
		goto bad_inode;
	bh = iloc.bh;
4800
	raw_inode = ext4_raw_inode(&iloc);
4801 4802 4803
	inode->i_mode = le16_to_cpu(raw_inode->i_mode);
	inode->i_uid = (uid_t)le16_to_cpu(raw_inode->i_uid_low);
	inode->i_gid = (gid_t)le16_to_cpu(raw_inode->i_gid_low);
4804
	if (!(test_opt(inode->i_sb, NO_UID32))) {
4805 4806 4807 4808 4809 4810 4811 4812 4813 4814 4815 4816 4817 4818 4819
		inode->i_uid |= le16_to_cpu(raw_inode->i_uid_high) << 16;
		inode->i_gid |= le16_to_cpu(raw_inode->i_gid_high) << 16;
	}
	inode->i_nlink = le16_to_cpu(raw_inode->i_links_count);

	ei->i_state = 0;
	ei->i_dir_start_lookup = 0;
	ei->i_dtime = le32_to_cpu(raw_inode->i_dtime);
	/* We now have enough fields to check if the inode was active or not.
	 * This is needed because nfsd might try to access dead inodes
	 * the test is that same one that e2fsck uses
	 * NeilBrown 1999oct15
	 */
	if (inode->i_nlink == 0) {
		if (inode->i_mode == 0 ||
4820
		    !(EXT4_SB(inode->i_sb)->s_mount_state & EXT4_ORPHAN_FS)) {
4821
			/* this inode is deleted */
4822
			brelse(bh);
4823
			ret = -ESTALE;
4824 4825 4826 4827 4828 4829 4830 4831
			goto bad_inode;
		}
		/* The only unlinked inodes we let through here have
		 * valid i_mode and are being read by the orphan
		 * recovery code: that's fine, we're about to complete
		 * the process of deleting those. */
	}
	ei->i_flags = le32_to_cpu(raw_inode->i_flags);
4832
	inode->i_blocks = ext4_inode_blocks(raw_inode, ei);
4833
	ei->i_file_acl = le32_to_cpu(raw_inode->i_file_acl_lo);
4834
	if (EXT4_HAS_INCOMPAT_FEATURE(sb, EXT4_FEATURE_INCOMPAT_64BIT))
B
Badari Pulavarty 已提交
4835 4836
		ei->i_file_acl |=
			((__u64)le16_to_cpu(raw_inode->i_file_acl_high)) << 32;
4837
	inode->i_size = ext4_isize(raw_inode);
4838 4839 4840
	ei->i_disksize = inode->i_size;
	inode->i_generation = le32_to_cpu(raw_inode->i_generation);
	ei->i_block_group = iloc.block_group;
4841
	ei->i_last_alloc_group = ~0;
4842 4843 4844 4845
	/*
	 * NOTE! The in-memory inode i_data array is in little-endian order
	 * even on big-endian machines: we do NOT byteswap the block numbers!
	 */
4846
	for (block = 0; block < EXT4_N_BLOCKS; block++)
4847 4848 4849
		ei->i_data[block] = raw_inode->i_block[block];
	INIT_LIST_HEAD(&ei->i_orphan);

4850
	if (EXT4_INODE_SIZE(inode->i_sb) > EXT4_GOOD_OLD_INODE_SIZE) {
4851
		ei->i_extra_isize = le16_to_cpu(raw_inode->i_extra_isize);
4852
		if (EXT4_GOOD_OLD_INODE_SIZE + ei->i_extra_isize >
4853
		    EXT4_INODE_SIZE(inode->i_sb)) {
4854
			brelse(bh);
4855
			ret = -EIO;
4856
			goto bad_inode;
4857
		}
4858 4859
		if (ei->i_extra_isize == 0) {
			/* The extra space is currently unused. Use it. */
4860 4861
			ei->i_extra_isize = sizeof(struct ext4_inode) -
					    EXT4_GOOD_OLD_INODE_SIZE;
4862 4863
		} else {
			__le32 *magic = (void *)raw_inode +
4864
					EXT4_GOOD_OLD_INODE_SIZE +
4865
					ei->i_extra_isize;
4866
			if (*magic == cpu_to_le32(EXT4_XATTR_MAGIC))
4867
				ei->i_state |= EXT4_STATE_XATTR;
4868 4869 4870 4871
		}
	} else
		ei->i_extra_isize = 0;

K
Kalpak Shah 已提交
4872 4873 4874 4875 4876
	EXT4_INODE_GET_XTIME(i_ctime, inode, raw_inode);
	EXT4_INODE_GET_XTIME(i_mtime, inode, raw_inode);
	EXT4_INODE_GET_XTIME(i_atime, inode, raw_inode);
	EXT4_EINODE_GET_XTIME(i_crtime, ei, raw_inode);

4877 4878 4879 4880 4881 4882 4883
	inode->i_version = le32_to_cpu(raw_inode->i_disk_version);
	if (EXT4_INODE_SIZE(inode->i_sb) > EXT4_GOOD_OLD_INODE_SIZE) {
		if (EXT4_FITS_IN_INODE(raw_inode, ei, i_version_hi))
			inode->i_version |=
			(__u64)(le32_to_cpu(raw_inode->i_version_hi)) << 32;
	}

4884
	ret = 0;
4885
	if (ei->i_file_acl &&
4886
	    ((ei->i_file_acl <
4887 4888 4889 4890 4891 4892 4893 4894 4895
	      (le32_to_cpu(EXT4_SB(sb)->s_es->s_first_data_block) +
	       EXT4_SB(sb)->s_gdb_count)) ||
	     (ei->i_file_acl >= ext4_blocks_count(EXT4_SB(sb)->s_es)))) {
		ext4_error(sb, __func__,
			   "bad extended attribute block %llu in inode #%lu",
			   ei->i_file_acl, inode->i_ino);
		ret = -EIO;
		goto bad_inode;
	} else if (ei->i_flags & EXT4_EXTENTS_FL) {
4896 4897 4898 4899 4900
		if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
		    (S_ISLNK(inode->i_mode) &&
		     !ext4_inode_is_fast_symlink(inode)))
			/* Validate extent which is part of inode */
			ret = ext4_ext_check_inode(inode);
4901
	} else if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
4902 4903
		   (S_ISLNK(inode->i_mode) &&
		    !ext4_inode_is_fast_symlink(inode))) {
4904
		/* Validate block references which are part of inode */
4905 4906 4907
		ret = ext4_check_inode_blockref(inode);
	}
	if (ret) {
4908 4909
		brelse(bh);
		goto bad_inode;
4910 4911
	}

4912
	if (S_ISREG(inode->i_mode)) {
4913 4914 4915
		inode->i_op = &ext4_file_inode_operations;
		inode->i_fop = &ext4_file_operations;
		ext4_set_aops(inode);
4916
	} else if (S_ISDIR(inode->i_mode)) {
4917 4918
		inode->i_op = &ext4_dir_inode_operations;
		inode->i_fop = &ext4_dir_operations;
4919
	} else if (S_ISLNK(inode->i_mode)) {
4920
		if (ext4_inode_is_fast_symlink(inode)) {
4921
			inode->i_op = &ext4_fast_symlink_inode_operations;
4922 4923 4924
			nd_terminate_link(ei->i_data, inode->i_size,
				sizeof(ei->i_data) - 1);
		} else {
4925 4926
			inode->i_op = &ext4_symlink_inode_operations;
			ext4_set_aops(inode);
4927
		}
4928 4929
	} else if (S_ISCHR(inode->i_mode) || S_ISBLK(inode->i_mode) ||
	      S_ISFIFO(inode->i_mode) || S_ISSOCK(inode->i_mode)) {
4930
		inode->i_op = &ext4_special_inode_operations;
4931 4932 4933 4934 4935 4936
		if (raw_inode->i_block[0])
			init_special_inode(inode, inode->i_mode,
			   old_decode_dev(le32_to_cpu(raw_inode->i_block[0])));
		else
			init_special_inode(inode, inode->i_mode,
			   new_decode_dev(le32_to_cpu(raw_inode->i_block[1])));
4937 4938 4939
	} else {
		brelse(bh);
		ret = -EIO;
4940
		ext4_error(inode->i_sb, __func__,
4941 4942 4943
			   "bogus i_mode (%o) for inode=%lu",
			   inode->i_mode, inode->i_ino);
		goto bad_inode;
4944
	}
4945
	brelse(iloc.bh);
4946
	ext4_set_inode_flags(inode);
4947 4948
	unlock_new_inode(inode);
	return inode;
4949 4950

bad_inode:
4951 4952
	iget_failed(inode);
	return ERR_PTR(ret);
4953 4954
}

4955 4956 4957 4958 4959 4960 4961 4962 4963 4964 4965 4966 4967
static int ext4_inode_blocks_set(handle_t *handle,
				struct ext4_inode *raw_inode,
				struct ext4_inode_info *ei)
{
	struct inode *inode = &(ei->vfs_inode);
	u64 i_blocks = inode->i_blocks;
	struct super_block *sb = inode->i_sb;

	if (i_blocks <= ~0U) {
		/*
		 * i_blocks can be represnted in a 32 bit variable
		 * as multiple of 512 bytes
		 */
A
Aneesh Kumar K.V 已提交
4968
		raw_inode->i_blocks_lo   = cpu_to_le32(i_blocks);
4969
		raw_inode->i_blocks_high = 0;
A
Aneesh Kumar K.V 已提交
4970
		ei->i_flags &= ~EXT4_HUGE_FILE_FL;
4971 4972 4973 4974 4975 4976
		return 0;
	}
	if (!EXT4_HAS_RO_COMPAT_FEATURE(sb, EXT4_FEATURE_RO_COMPAT_HUGE_FILE))
		return -EFBIG;

	if (i_blocks <= 0xffffffffffffULL) {
4977 4978 4979 4980
		/*
		 * i_blocks can be represented in a 48 bit variable
		 * as multiple of 512 bytes
		 */
A
Aneesh Kumar K.V 已提交
4981
		raw_inode->i_blocks_lo   = cpu_to_le32(i_blocks);
4982
		raw_inode->i_blocks_high = cpu_to_le16(i_blocks >> 32);
A
Aneesh Kumar K.V 已提交
4983
		ei->i_flags &= ~EXT4_HUGE_FILE_FL;
4984
	} else {
A
Aneesh Kumar K.V 已提交
4985 4986 4987 4988 4989
		ei->i_flags |= EXT4_HUGE_FILE_FL;
		/* i_block is stored in file system block size */
		i_blocks = i_blocks >> (inode->i_blkbits - 9);
		raw_inode->i_blocks_lo   = cpu_to_le32(i_blocks);
		raw_inode->i_blocks_high = cpu_to_le16(i_blocks >> 32);
4990
	}
4991
	return 0;
4992 4993
}

4994 4995 4996 4997 4998 4999 5000
/*
 * Post the struct inode info into an on-disk inode location in the
 * buffer-cache.  This gobbles the caller's reference to the
 * buffer_head in the inode location struct.
 *
 * The caller must have write access to iloc->bh.
 */
5001
static int ext4_do_update_inode(handle_t *handle,
5002
				struct inode *inode,
5003
				struct ext4_iloc *iloc)
5004
{
5005 5006
	struct ext4_inode *raw_inode = ext4_raw_inode(iloc);
	struct ext4_inode_info *ei = EXT4_I(inode);
5007 5008 5009 5010 5011
	struct buffer_head *bh = iloc->bh;
	int err = 0, rc, block;

	/* For fields not not tracking in the in-memory inode,
	 * initialise them to zero for new inodes. */
5012 5013
	if (ei->i_state & EXT4_STATE_NEW)
		memset(raw_inode, 0, EXT4_SB(inode->i_sb)->s_inode_size);
5014

5015
	ext4_get_inode_flags(ei);
5016
	raw_inode->i_mode = cpu_to_le16(inode->i_mode);
5017
	if (!(test_opt(inode->i_sb, NO_UID32))) {
5018 5019 5020 5021 5022 5023
		raw_inode->i_uid_low = cpu_to_le16(low_16_bits(inode->i_uid));
		raw_inode->i_gid_low = cpu_to_le16(low_16_bits(inode->i_gid));
/*
 * Fix up interoperability with old kernels. Otherwise, old inodes get
 * re-used with the upper 16 bits of the uid/gid intact
 */
5024
		if (!ei->i_dtime) {
5025 5026 5027 5028 5029 5030 5031 5032 5033 5034 5035 5036 5037 5038 5039 5040 5041
			raw_inode->i_uid_high =
				cpu_to_le16(high_16_bits(inode->i_uid));
			raw_inode->i_gid_high =
				cpu_to_le16(high_16_bits(inode->i_gid));
		} else {
			raw_inode->i_uid_high = 0;
			raw_inode->i_gid_high = 0;
		}
	} else {
		raw_inode->i_uid_low =
			cpu_to_le16(fs_high2lowuid(inode->i_uid));
		raw_inode->i_gid_low =
			cpu_to_le16(fs_high2lowgid(inode->i_gid));
		raw_inode->i_uid_high = 0;
		raw_inode->i_gid_high = 0;
	}
	raw_inode->i_links_count = cpu_to_le16(inode->i_nlink);
K
Kalpak Shah 已提交
5042 5043 5044 5045 5046 5047

	EXT4_INODE_SET_XTIME(i_ctime, inode, raw_inode);
	EXT4_INODE_SET_XTIME(i_mtime, inode, raw_inode);
	EXT4_INODE_SET_XTIME(i_atime, inode, raw_inode);
	EXT4_EINODE_SET_XTIME(i_crtime, ei, raw_inode);

5048 5049
	if (ext4_inode_blocks_set(handle, raw_inode, ei))
		goto out_brelse;
5050
	raw_inode->i_dtime = cpu_to_le32(ei->i_dtime);
5051
	raw_inode->i_flags = cpu_to_le32(ei->i_flags);
5052 5053
	if (EXT4_SB(inode->i_sb)->s_es->s_creator_os !=
	    cpu_to_le32(EXT4_OS_HURD))
B
Badari Pulavarty 已提交
5054 5055
		raw_inode->i_file_acl_high =
			cpu_to_le16(ei->i_file_acl >> 32);
5056
	raw_inode->i_file_acl_lo = cpu_to_le32(ei->i_file_acl);
5057 5058 5059 5060 5061 5062 5063 5064 5065 5066 5067 5068 5069 5070 5071 5072
	ext4_isize_set(raw_inode, ei->i_disksize);
	if (ei->i_disksize > 0x7fffffffULL) {
		struct super_block *sb = inode->i_sb;
		if (!EXT4_HAS_RO_COMPAT_FEATURE(sb,
				EXT4_FEATURE_RO_COMPAT_LARGE_FILE) ||
				EXT4_SB(sb)->s_es->s_rev_level ==
				cpu_to_le32(EXT4_GOOD_OLD_REV)) {
			/* If this is the first large file
			 * created, add a flag to the superblock.
			 */
			err = ext4_journal_get_write_access(handle,
					EXT4_SB(sb)->s_sbh);
			if (err)
				goto out_brelse;
			ext4_update_dynamic_rev(sb);
			EXT4_SET_RO_COMPAT_FEATURE(sb,
5073
					EXT4_FEATURE_RO_COMPAT_LARGE_FILE);
5074
			sb->s_dirt = 1;
5075 5076
			ext4_handle_sync(handle);
			err = ext4_handle_dirty_metadata(handle, inode,
5077
					EXT4_SB(sb)->s_sbh);
5078 5079 5080 5081 5082 5083 5084 5085 5086 5087 5088 5089 5090 5091
		}
	}
	raw_inode->i_generation = cpu_to_le32(inode->i_generation);
	if (S_ISCHR(inode->i_mode) || S_ISBLK(inode->i_mode)) {
		if (old_valid_dev(inode->i_rdev)) {
			raw_inode->i_block[0] =
				cpu_to_le32(old_encode_dev(inode->i_rdev));
			raw_inode->i_block[1] = 0;
		} else {
			raw_inode->i_block[0] = 0;
			raw_inode->i_block[1] =
				cpu_to_le32(new_encode_dev(inode->i_rdev));
			raw_inode->i_block[2] = 0;
		}
5092 5093 5094
	} else
		for (block = 0; block < EXT4_N_BLOCKS; block++)
			raw_inode->i_block[block] = ei->i_data[block];
5095

5096 5097 5098 5099 5100
	raw_inode->i_disk_version = cpu_to_le32(inode->i_version);
	if (ei->i_extra_isize) {
		if (EXT4_FITS_IN_INODE(raw_inode, ei, i_version_hi))
			raw_inode->i_version_hi =
			cpu_to_le32(inode->i_version >> 32);
5101
		raw_inode->i_extra_isize = cpu_to_le16(ei->i_extra_isize);
5102 5103
	}

5104 5105 5106 5107
	BUFFER_TRACE(bh, "call ext4_handle_dirty_metadata");
	rc = ext4_handle_dirty_metadata(handle, inode, bh);
	if (!err)
		err = rc;
5108
	ei->i_state &= ~EXT4_STATE_NEW;
5109 5110

out_brelse:
5111
	brelse(bh);
5112
	ext4_std_error(inode->i_sb, err);
5113 5114 5115 5116
	return err;
}

/*
5117
 * ext4_write_inode()
5118 5119 5120 5121 5122 5123 5124 5125 5126 5127 5128 5129 5130 5131 5132 5133
 *
 * We are called from a few places:
 *
 * - Within generic_file_write() for O_SYNC files.
 *   Here, there will be no transaction running. We wait for any running
 *   trasnaction to commit.
 *
 * - Within sys_sync(), kupdate and such.
 *   We wait on commit, if tol to.
 *
 * - Within prune_icache() (PF_MEMALLOC == true)
 *   Here we simply return.  We can't afford to block kswapd on the
 *   journal commit.
 *
 * In all cases it is actually safe for us to return without doing anything,
 * because the inode has been copied into a raw inode buffer in
5134
 * ext4_mark_inode_dirty().  This is a correctness thing for O_SYNC and for
5135 5136 5137 5138 5139 5140 5141 5142 5143 5144 5145 5146 5147 5148 5149 5150
 * knfsd.
 *
 * Note that we are absolutely dependent upon all inode dirtiers doing the
 * right thing: they *must* call mark_inode_dirty() after dirtying info in
 * which we are interested.
 *
 * It would be a bug for them to not do this.  The code:
 *
 *	mark_inode_dirty(inode)
 *	stuff();
 *	inode->i_size = expr;
 *
 * is in error because a kswapd-driven write_inode() could occur while
 * `stuff()' is running, and the new i_size will be lost.  Plus the inode
 * will no longer be on the superblock's dirty inode list.
 */
5151
int ext4_write_inode(struct inode *inode, int wait)
5152
{
5153 5154
	int err;

5155 5156 5157
	if (current->flags & PF_MEMALLOC)
		return 0;

5158 5159 5160 5161 5162 5163
	if (EXT4_SB(inode->i_sb)->s_journal) {
		if (ext4_journal_current_handle()) {
			jbd_debug(1, "called recursively, non-PF_MEMALLOC!\n");
			dump_stack();
			return -EIO;
		}
5164

5165 5166 5167 5168 5169 5170
		if (!wait)
			return 0;

		err = ext4_force_commit(inode->i_sb);
	} else {
		struct ext4_iloc iloc;
5171

5172 5173 5174
		err = ext4_get_inode_loc(inode, &iloc);
		if (err)
			return err;
5175 5176 5177 5178 5179 5180 5181 5182 5183 5184
		if (wait)
			sync_dirty_buffer(iloc.bh);
		if (buffer_req(iloc.bh) && !buffer_uptodate(iloc.bh)) {
			ext4_error(inode->i_sb, __func__,
				   "IO error syncing inode, "
				   "inode=%lu, block=%llu",
				   inode->i_ino,
				   (unsigned long long)iloc.bh->b_blocknr);
			err = -EIO;
		}
5185 5186
	}
	return err;
5187 5188 5189
}

/*
5190
 * ext4_setattr()
5191 5192 5193 5194 5195 5196 5197 5198 5199 5200 5201 5202 5203
 *
 * Called from notify_change.
 *
 * We want to trap VFS attempts to truncate the file as soon as
 * possible.  In particular, we want to make sure that when the VFS
 * shrinks i_size, we put the inode on the orphan list and modify
 * i_disksize immediately, so that during the subsequent flushing of
 * dirty pages and freeing of disk blocks, we can guarantee that any
 * commit will leave the blocks being flushed in an unused state on
 * disk.  (On recovery, the inode will get truncated and the blocks will
 * be freed, so we have a strong guarantee that no future commit will
 * leave these blocks visible to the user.)
 *
5204 5205 5206 5207 5208 5209 5210 5211
 * Another thing we have to assure is that if we are in ordered mode
 * and inode is still attached to the committing transaction, we must
 * we start writeout of all the dirty pages which are being truncated.
 * This way we are sure that all the data written in the previous
 * transaction are already on disk (truncate waits for pages under
 * writeback).
 *
 * Called with inode->i_mutex down.
5212
 */
5213
int ext4_setattr(struct dentry *dentry, struct iattr *attr)
5214 5215 5216 5217 5218 5219 5220 5221 5222 5223 5224 5225 5226 5227 5228
{
	struct inode *inode = dentry->d_inode;
	int error, rc = 0;
	const unsigned int ia_valid = attr->ia_valid;

	error = inode_change_ok(inode, attr);
	if (error)
		return error;

	if ((ia_valid & ATTR_UID && attr->ia_uid != inode->i_uid) ||
		(ia_valid & ATTR_GID && attr->ia_gid != inode->i_gid)) {
		handle_t *handle;

		/* (user+group)*(old+new) structure, inode write (sb,
		 * inode block, ? - but truncate inode update has it) */
5229 5230
		handle = ext4_journal_start(inode, 2*(EXT4_QUOTA_INIT_BLOCKS(inode->i_sb)+
					EXT4_QUOTA_DEL_BLOCKS(inode->i_sb))+3);
5231 5232 5233 5234
		if (IS_ERR(handle)) {
			error = PTR_ERR(handle);
			goto err_out;
		}
5235
		error = vfs_dq_transfer(inode, attr) ? -EDQUOT : 0;
5236
		if (error) {
5237
			ext4_journal_stop(handle);
5238 5239 5240 5241 5242 5243 5244 5245
			return error;
		}
		/* Update corresponding info in inode so that everything is in
		 * one transaction */
		if (attr->ia_valid & ATTR_UID)
			inode->i_uid = attr->ia_uid;
		if (attr->ia_valid & ATTR_GID)
			inode->i_gid = attr->ia_gid;
5246 5247
		error = ext4_mark_inode_dirty(handle, inode);
		ext4_journal_stop(handle);
5248 5249
	}

5250 5251 5252 5253 5254 5255 5256 5257 5258 5259 5260
	if (attr->ia_valid & ATTR_SIZE) {
		if (!(EXT4_I(inode)->i_flags & EXT4_EXTENTS_FL)) {
			struct ext4_sb_info *sbi = EXT4_SB(inode->i_sb);

			if (attr->ia_size > sbi->s_bitmap_maxbytes) {
				error = -EFBIG;
				goto err_out;
			}
		}
	}

5261 5262 5263 5264
	if (S_ISREG(inode->i_mode) &&
	    attr->ia_valid & ATTR_SIZE && attr->ia_size < inode->i_size) {
		handle_t *handle;

5265
		handle = ext4_journal_start(inode, 3);
5266 5267 5268 5269 5270
		if (IS_ERR(handle)) {
			error = PTR_ERR(handle);
			goto err_out;
		}

5271 5272 5273
		error = ext4_orphan_add(handle, inode);
		EXT4_I(inode)->i_disksize = attr->ia_size;
		rc = ext4_mark_inode_dirty(handle, inode);
5274 5275
		if (!error)
			error = rc;
5276
		ext4_journal_stop(handle);
5277 5278 5279 5280 5281 5282 5283 5284 5285 5286 5287 5288 5289 5290 5291 5292

		if (ext4_should_order_data(inode)) {
			error = ext4_begin_ordered_truncate(inode,
							    attr->ia_size);
			if (error) {
				/* Do as much error cleanup as possible */
				handle = ext4_journal_start(inode, 3);
				if (IS_ERR(handle)) {
					ext4_orphan_del(NULL, inode);
					goto err_out;
				}
				ext4_orphan_del(handle, inode);
				ext4_journal_stop(handle);
				goto err_out;
			}
		}
5293 5294 5295 5296
	}

	rc = inode_setattr(inode, attr);

5297
	/* If inode_setattr's call to ext4_truncate failed to get a
5298 5299 5300
	 * transaction handle at all, we need to clean up the in-core
	 * orphan list manually. */
	if (inode->i_nlink)
5301
		ext4_orphan_del(NULL, inode);
5302 5303

	if (!rc && (ia_valid & ATTR_MODE))
5304
		rc = ext4_acl_chmod(inode);
5305 5306

err_out:
5307
	ext4_std_error(inode->i_sb, error);
5308 5309 5310 5311 5312
	if (!error)
		error = rc;
	return error;
}

5313 5314 5315 5316 5317 5318 5319 5320 5321 5322 5323 5324 5325 5326 5327 5328 5329 5330 5331 5332 5333 5334 5335 5336 5337 5338
int ext4_getattr(struct vfsmount *mnt, struct dentry *dentry,
		 struct kstat *stat)
{
	struct inode *inode;
	unsigned long delalloc_blocks;

	inode = dentry->d_inode;
	generic_fillattr(inode, stat);

	/*
	 * We can't update i_blocks if the block allocation is delayed
	 * otherwise in the case of system crash before the real block
	 * allocation is done, we will have i_blocks inconsistent with
	 * on-disk file blocks.
	 * We always keep i_blocks updated together with real
	 * allocation. But to not confuse with user, stat
	 * will return the blocks that include the delayed allocation
	 * blocks for this file.
	 */
	spin_lock(&EXT4_I(inode)->i_block_reservation_lock);
	delalloc_blocks = EXT4_I(inode)->i_reserved_data_blocks;
	spin_unlock(&EXT4_I(inode)->i_block_reservation_lock);

	stat->blocks += (delalloc_blocks << inode->i_sb->s_blocksize_bits)>>9;
	return 0;
}
5339

5340 5341 5342 5343 5344 5345 5346 5347 5348 5349 5350 5351 5352 5353 5354 5355 5356 5357 5358 5359 5360 5361 5362 5363 5364 5365 5366 5367
static int ext4_indirect_trans_blocks(struct inode *inode, int nrblocks,
				      int chunk)
{
	int indirects;

	/* if nrblocks are contiguous */
	if (chunk) {
		/*
		 * With N contiguous data blocks, it need at most
		 * N/EXT4_ADDR_PER_BLOCK(inode->i_sb) indirect blocks
		 * 2 dindirect blocks
		 * 1 tindirect block
		 */
		indirects = nrblocks / EXT4_ADDR_PER_BLOCK(inode->i_sb);
		return indirects + 3;
	}
	/*
	 * if nrblocks are not contiguous, worse case, each block touch
	 * a indirect block, and each indirect block touch a double indirect
	 * block, plus a triple indirect block
	 */
	indirects = nrblocks * 2 + 1;
	return indirects;
}

static int ext4_index_trans_blocks(struct inode *inode, int nrblocks, int chunk)
{
	if (!(EXT4_I(inode)->i_flags & EXT4_EXTENTS_FL))
5368 5369
		return ext4_indirect_trans_blocks(inode, nrblocks, chunk);
	return ext4_ext_index_trans_blocks(inode, nrblocks, chunk);
5370
}
5371

5372
/*
5373 5374 5375
 * Account for index blocks, block groups bitmaps and block group
 * descriptor blocks if modify datablocks and index blocks
 * worse case, the indexs blocks spread over different block groups
5376
 *
5377 5378 5379
 * If datablocks are discontiguous, they are possible to spread over
 * different block groups too. If they are contiugous, with flexbg,
 * they could still across block group boundary.
5380
 *
5381 5382 5383 5384
 * Also account for superblock, inode, quota and xattr blocks
 */
int ext4_meta_trans_blocks(struct inode *inode, int nrblocks, int chunk)
{
5385 5386
	ext4_group_t groups, ngroups = ext4_get_groups_count(inode->i_sb);
	int gdpblocks;
5387 5388 5389 5390 5391 5392 5393 5394 5395 5396 5397 5398 5399 5400 5401 5402 5403 5404 5405 5406 5407 5408 5409 5410 5411 5412
	int idxblocks;
	int ret = 0;

	/*
	 * How many index blocks need to touch to modify nrblocks?
	 * The "Chunk" flag indicating whether the nrblocks is
	 * physically contiguous on disk
	 *
	 * For Direct IO and fallocate, they calls get_block to allocate
	 * one single extent at a time, so they could set the "Chunk" flag
	 */
	idxblocks = ext4_index_trans_blocks(inode, nrblocks, chunk);

	ret = idxblocks;

	/*
	 * Now let's see how many group bitmaps and group descriptors need
	 * to account
	 */
	groups = idxblocks;
	if (chunk)
		groups += 1;
	else
		groups += nrblocks;

	gdpblocks = groups;
5413 5414
	if (groups > ngroups)
		groups = ngroups;
5415 5416 5417 5418 5419 5420 5421 5422 5423 5424 5425 5426 5427 5428
	if (groups > EXT4_SB(inode->i_sb)->s_gdb_count)
		gdpblocks = EXT4_SB(inode->i_sb)->s_gdb_count;

	/* bitmaps and block group descriptor blocks */
	ret += groups + gdpblocks;

	/* Blocks for super block, inode, quota and xattr blocks */
	ret += EXT4_META_TRANS_BLOCKS(inode->i_sb);

	return ret;
}

/*
 * Calulate the total number of credits to reserve to fit
5429 5430
 * the modification of a single pages into a single transaction,
 * which may include multiple chunks of block allocations.
5431
 *
5432
 * This could be called via ext4_write_begin()
5433
 *
5434
 * We need to consider the worse case, when
5435
 * one new block per extent.
5436
 */
A
Alex Tomas 已提交
5437
int ext4_writepage_trans_blocks(struct inode *inode)
5438
{
5439
	int bpp = ext4_journal_blocks_per_page(inode);
5440 5441
	int ret;

5442
	ret = ext4_meta_trans_blocks(inode, bpp, 0);
A
Alex Tomas 已提交
5443

5444
	/* Account for data blocks for journalled mode */
5445
	if (ext4_should_journal_data(inode))
5446
		ret += bpp;
5447 5448
	return ret;
}
5449 5450 5451 5452 5453

/*
 * Calculate the journal credits for a chunk of data modification.
 *
 * This is called from DIO, fallocate or whoever calling
5454
 * ext4_get_blocks() to map/allocate a chunk of contigous disk blocks.
5455 5456 5457 5458 5459 5460 5461 5462 5463
 *
 * journal buffers for data blocks are not included here, as DIO
 * and fallocate do no need to journal data buffers.
 */
int ext4_chunk_trans_blocks(struct inode *inode, int nrblocks)
{
	return ext4_meta_trans_blocks(inode, nrblocks, 1);
}

5464
/*
5465
 * The caller must have previously called ext4_reserve_inode_write().
5466 5467
 * Give this, we know that the caller already has write access to iloc->bh.
 */
5468
int ext4_mark_iloc_dirty(handle_t *handle,
5469
			 struct inode *inode, struct ext4_iloc *iloc)
5470 5471 5472
{
	int err = 0;

5473 5474 5475
	if (test_opt(inode->i_sb, I_VERSION))
		inode_inc_iversion(inode);

5476 5477 5478
	/* the do_update_inode consumes one bh->b_count */
	get_bh(iloc->bh);

5479
	/* ext4_do_update_inode() does jbd2_journal_dirty_metadata */
5480
	err = ext4_do_update_inode(handle, inode, iloc);
5481 5482 5483 5484 5485 5486 5487 5488 5489 5490
	put_bh(iloc->bh);
	return err;
}

/*
 * On success, We end up with an outstanding reference count against
 * iloc->bh.  This _must_ be cleaned up later.
 */

int
5491 5492
ext4_reserve_inode_write(handle_t *handle, struct inode *inode,
			 struct ext4_iloc *iloc)
5493
{
5494 5495 5496 5497 5498 5499 5500 5501 5502
	int err;

	err = ext4_get_inode_loc(inode, iloc);
	if (!err) {
		BUFFER_TRACE(iloc->bh, "get_write_access");
		err = ext4_journal_get_write_access(handle, iloc->bh);
		if (err) {
			brelse(iloc->bh);
			iloc->bh = NULL;
5503 5504
		}
	}
5505
	ext4_std_error(inode->i_sb, err);
5506 5507 5508
	return err;
}

5509 5510 5511 5512
/*
 * Expand an inode by new_extra_isize bytes.
 * Returns 0 on success or negative error number on failure.
 */
A
Aneesh Kumar K.V 已提交
5513 5514 5515 5516
static int ext4_expand_extra_isize(struct inode *inode,
				   unsigned int new_extra_isize,
				   struct ext4_iloc iloc,
				   handle_t *handle)
5517 5518 5519 5520 5521 5522 5523 5524 5525 5526 5527 5528 5529 5530 5531 5532 5533 5534 5535 5536 5537 5538 5539 5540 5541 5542 5543
{
	struct ext4_inode *raw_inode;
	struct ext4_xattr_ibody_header *header;
	struct ext4_xattr_entry *entry;

	if (EXT4_I(inode)->i_extra_isize >= new_extra_isize)
		return 0;

	raw_inode = ext4_raw_inode(&iloc);

	header = IHDR(inode, raw_inode);
	entry = IFIRST(header);

	/* No extended attributes present */
	if (!(EXT4_I(inode)->i_state & EXT4_STATE_XATTR) ||
		header->h_magic != cpu_to_le32(EXT4_XATTR_MAGIC)) {
		memset((void *)raw_inode + EXT4_GOOD_OLD_INODE_SIZE, 0,
			new_extra_isize);
		EXT4_I(inode)->i_extra_isize = new_extra_isize;
		return 0;
	}

	/* try to expand with EAs present */
	return ext4_expand_extra_isize_ea(inode, new_extra_isize,
					  raw_inode, handle);
}

5544 5545 5546 5547 5548 5549 5550 5551 5552 5553 5554 5555 5556 5557 5558 5559 5560 5561 5562 5563 5564
/*
 * What we do here is to mark the in-core inode as clean with respect to inode
 * dirtiness (it may still be data-dirty).
 * This means that the in-core inode may be reaped by prune_icache
 * without having to perform any I/O.  This is a very good thing,
 * because *any* task may call prune_icache - even ones which
 * have a transaction open against a different journal.
 *
 * Is this cheating?  Not really.  Sure, we haven't written the
 * inode out, but prune_icache isn't a user-visible syncing function.
 * Whenever the user wants stuff synced (sys_sync, sys_msync, sys_fsync)
 * we start and wait on commits.
 *
 * Is this efficient/effective?  Well, we're being nice to the system
 * by cleaning up our inodes proactively so they can be reaped
 * without I/O.  But we are potentially leaving up to five seconds'
 * worth of inodes floating about which prune_icache wants us to
 * write out.  One way to fix that would be to get prune_icache()
 * to do a write_super() to free up some memory.  It has the desired
 * effect.
 */
5565
int ext4_mark_inode_dirty(handle_t *handle, struct inode *inode)
5566
{
5567
	struct ext4_iloc iloc;
5568 5569 5570
	struct ext4_sb_info *sbi = EXT4_SB(inode->i_sb);
	static unsigned int mnt_count;
	int err, ret;
5571 5572

	might_sleep();
5573
	err = ext4_reserve_inode_write(handle, inode, &iloc);
5574 5575
	if (ext4_handle_valid(handle) &&
	    EXT4_I(inode)->i_extra_isize < sbi->s_want_extra_isize &&
5576 5577 5578 5579 5580 5581 5582 5583 5584 5585 5586 5587 5588 5589 5590
	    !(EXT4_I(inode)->i_state & EXT4_STATE_NO_EXPAND)) {
		/*
		 * We need extra buffer credits since we may write into EA block
		 * with this same handle. If journal_extend fails, then it will
		 * only result in a minor loss of functionality for that inode.
		 * If this is felt to be critical, then e2fsck should be run to
		 * force a large enough s_min_extra_isize.
		 */
		if ((jbd2_journal_extend(handle,
			     EXT4_DATA_TRANS_BLOCKS(inode->i_sb))) == 0) {
			ret = ext4_expand_extra_isize(inode,
						      sbi->s_want_extra_isize,
						      iloc, handle);
			if (ret) {
				EXT4_I(inode)->i_state |= EXT4_STATE_NO_EXPAND;
A
Aneesh Kumar K.V 已提交
5591 5592
				if (mnt_count !=
					le16_to_cpu(sbi->s_es->s_mnt_count)) {
5593
					ext4_warning(inode->i_sb, __func__,
5594 5595 5596
					"Unable to expand inode %lu. Delete"
					" some EAs or run e2fsck.",
					inode->i_ino);
A
Aneesh Kumar K.V 已提交
5597 5598
					mnt_count =
					  le16_to_cpu(sbi->s_es->s_mnt_count);
5599 5600 5601 5602
				}
			}
		}
	}
5603
	if (!err)
5604
		err = ext4_mark_iloc_dirty(handle, inode, &iloc);
5605 5606 5607 5608
	return err;
}

/*
5609
 * ext4_dirty_inode() is called from __mark_inode_dirty()
5610 5611 5612 5613 5614
 *
 * We're really interested in the case where a file is being extended.
 * i_size has been changed by generic_commit_write() and we thus need
 * to include the updated inode in the current transaction.
 *
5615
 * Also, vfs_dq_alloc_block() will always dirty the inode when blocks
5616 5617 5618 5619 5620 5621
 * are allocated to the file.
 *
 * If the inode is marked synchronous, we don't honour that here - doing
 * so would cause a commit on atime updates, which we don't bother doing.
 * We handle synchronous inodes at the highest possible level.
 */
5622
void ext4_dirty_inode(struct inode *inode)
5623 5624 5625
{
	handle_t *handle;

5626
	handle = ext4_journal_start(inode, 2);
5627 5628
	if (IS_ERR(handle))
		goto out;
5629 5630 5631

	ext4_mark_inode_dirty(handle, inode);

5632
	ext4_journal_stop(handle);
5633 5634 5635 5636 5637 5638 5639 5640
out:
	return;
}

#if 0
/*
 * Bind an inode's backing buffer_head into this transaction, to prevent
 * it from being flushed to disk early.  Unlike
5641
 * ext4_reserve_inode_write, this leaves behind no bh reference and
5642 5643 5644
 * returns no iloc structure, so the caller needs to repeat the iloc
 * lookup to mark the inode dirty later.
 */
5645
static int ext4_pin_inode(handle_t *handle, struct inode *inode)
5646
{
5647
	struct ext4_iloc iloc;
5648 5649 5650

	int err = 0;
	if (handle) {
5651
		err = ext4_get_inode_loc(inode, &iloc);
5652 5653
		if (!err) {
			BUFFER_TRACE(iloc.bh, "get_write_access");
5654
			err = jbd2_journal_get_write_access(handle, iloc.bh);
5655
			if (!err)
5656 5657 5658
				err = ext4_handle_dirty_metadata(handle,
								 inode,
								 iloc.bh);
5659 5660 5661
			brelse(iloc.bh);
		}
	}
5662
	ext4_std_error(inode->i_sb, err);
5663 5664 5665 5666
	return err;
}
#endif

5667
int ext4_change_inode_journal_flag(struct inode *inode, int val)
5668 5669 5670 5671 5672 5673 5674 5675 5676 5677 5678 5679 5680 5681 5682
{
	journal_t *journal;
	handle_t *handle;
	int err;

	/*
	 * We have to be very careful here: changing a data block's
	 * journaling status dynamically is dangerous.  If we write a
	 * data block to the journal, change the status and then delete
	 * that block, we risk forgetting to revoke the old log record
	 * from the journal and so a subsequent replay can corrupt data.
	 * So, first we make sure that the journal is empty and that
	 * nobody is changing anything.
	 */

5683
	journal = EXT4_JOURNAL(inode);
5684 5685
	if (!journal)
		return 0;
5686
	if (is_journal_aborted(journal))
5687 5688
		return -EROFS;

5689 5690
	jbd2_journal_lock_updates(journal);
	jbd2_journal_flush(journal);
5691 5692 5693 5694 5695 5696 5697 5698 5699 5700

	/*
	 * OK, there are no updates running now, and all cached data is
	 * synced to disk.  We are now in a completely consistent state
	 * which doesn't have anything in the journal, and we know that
	 * no filesystem updates are running, so it is safe to modify
	 * the inode's in-core data-journaling state flag now.
	 */

	if (val)
5701
		EXT4_I(inode)->i_flags |= EXT4_JOURNAL_DATA_FL;
5702
	else
5703 5704
		EXT4_I(inode)->i_flags &= ~EXT4_JOURNAL_DATA_FL;
	ext4_set_aops(inode);
5705

5706
	jbd2_journal_unlock_updates(journal);
5707 5708 5709

	/* Finally we can mark the inode as dirty. */

5710
	handle = ext4_journal_start(inode, 1);
5711 5712 5713
	if (IS_ERR(handle))
		return PTR_ERR(handle);

5714
	err = ext4_mark_inode_dirty(handle, inode);
5715
	ext4_handle_sync(handle);
5716 5717
	ext4_journal_stop(handle);
	ext4_std_error(inode->i_sb, err);
5718 5719 5720

	return err;
}
5721 5722 5723 5724 5725 5726

static int ext4_bh_unmapped(handle_t *handle, struct buffer_head *bh)
{
	return !buffer_mapped(bh);
}

5727
int ext4_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf)
5728
{
5729
	struct page *page = vmf->page;
5730 5731 5732
	loff_t size;
	unsigned long len;
	int ret = -EINVAL;
5733
	void *fsdata;
5734 5735 5736 5737 5738 5739 5740 5741 5742 5743 5744 5745 5746 5747 5748 5749 5750 5751 5752 5753 5754 5755 5756 5757
	struct file *file = vma->vm_file;
	struct inode *inode = file->f_path.dentry->d_inode;
	struct address_space *mapping = inode->i_mapping;

	/*
	 * Get i_alloc_sem to stop truncates messing with the inode. We cannot
	 * get i_mutex because we are already holding mmap_sem.
	 */
	down_read(&inode->i_alloc_sem);
	size = i_size_read(inode);
	if (page->mapping != mapping || size <= page_offset(page)
	    || !PageUptodate(page)) {
		/* page got truncated from under us? */
		goto out_unlock;
	}
	ret = 0;
	if (PageMappedToDisk(page))
		goto out_unlock;

	if (page->index == size >> PAGE_CACHE_SHIFT)
		len = size & ~PAGE_CACHE_MASK;
	else
		len = PAGE_CACHE_SIZE;

5758 5759 5760 5761 5762 5763 5764
	lock_page(page);
	/*
	 * return if we have all the buffers mapped. This avoid
	 * the need to call write_begin/write_end which does a
	 * journal_start/journal_stop which can block and take
	 * long time
	 */
5765 5766
	if (page_has_buffers(page)) {
		if (!walk_page_buffers(NULL, page_buffers(page), 0, len, NULL,
5767 5768
					ext4_bh_unmapped)) {
			unlock_page(page);
5769
			goto out_unlock;
5770
		}
5771
	}
5772
	unlock_page(page);
5773 5774 5775 5776 5777 5778 5779 5780
	/*
	 * OK, we need to fill the hole... Do write_begin write_end
	 * to do block allocation/reservation.We are not holding
	 * inode.i__mutex here. That allow * parallel write_begin,
	 * write_end call. lock_page prevent this from happening
	 * on the same page though
	 */
	ret = mapping->a_ops->write_begin(file, mapping, page_offset(page),
5781
			len, AOP_FLAG_UNINTERRUPTIBLE, &page, &fsdata);
5782 5783 5784
	if (ret < 0)
		goto out_unlock;
	ret = mapping->a_ops->write_end(file, mapping, page_offset(page),
5785
			len, len, page, fsdata);
5786 5787 5788 5789
	if (ret < 0)
		goto out_unlock;
	ret = 0;
out_unlock:
5790 5791
	if (ret)
		ret = VM_FAULT_SIGBUS;
5792 5793 5794
	up_read(&inode->i_alloc_sem);
	return ret;
}