inode.c 100.0 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 29
#include <linux/ext4_jbd2.h>
#include <linux/jbd2.h>
30 31 32 33 34 35 36 37 38 39 40 41 42 43 44
#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>
#include <linux/mpage.h>
#include <linux/uio.h>
#include <linux/bio.h>
#include "xattr.h"
#include "acl.h"

/*
 * Test whether an inode is a fast symlink.
 */
45
static int ext4_inode_is_fast_symlink(struct inode *inode)
46
{
47
	int ea_blocks = EXT4_I(inode)->i_file_acl ?
48 49 50 51 52 53
		(inode->i_sb->s_blocksize >> 9) : 0;

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

/*
54
 * The ext4 forget function must perform a revoke if we are freeing data
55 56 57 58 59 60 61
 * 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.
 */
62 63
int ext4_forget(handle_t *handle, int is_metadata, struct inode *inode,
			struct buffer_head *bh, ext4_fsblk_t blocknr)
64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80
{
	int err;

	might_sleep();

	BUFFER_TRACE(bh, "enter");

	jbd_debug(4, "forgetting bh %p: is_metadata = %d, mode %o, "
		  "data mode %lx\n",
		  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. */

81 82
	if (test_opt(inode->i_sb, DATA_FLAGS) == EXT4_MOUNT_JOURNAL_DATA ||
	    (!is_metadata && !ext4_should_journal_data(inode))) {
83
		if (bh) {
84
			BUFFER_TRACE(bh, "call jbd2_journal_forget");
85
			return ext4_journal_forget(handle, bh);
86 87 88 89 90 91 92
		}
		return 0;
	}

	/*
	 * data!=journal && (is_metadata || should_journal_data(inode))
	 */
93 94
	BUFFER_TRACE(bh, "call ext4_journal_revoke");
	err = ext4_journal_revoke(handle, blocknr, bh);
95
	if (err)
96
		ext4_abort(inode->i_sb, __FUNCTION__,
97 98 99 100 101 102 103 104 105 106 107
			   "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
108
	ext4_lblk_t needed;
109 110 111 112 113 114

	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
115
	 * like a regular file for ext4 to try to delete it.  Things
116 117 118 119 120 121 122
	 * 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. */
123 124
	if (needed > EXT4_MAX_TRANS_DATA)
		needed = EXT4_MAX_TRANS_DATA;
125

126
	return EXT4_DATA_TRANS_BLOCKS(inode->i_sb) + needed;
127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142
}

/*
 * 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;

143
	result = ext4_journal_start(inode, blocks_for_truncate(inode));
144 145 146
	if (!IS_ERR(result))
		return result;

147
	ext4_std_error(inode->i_sb, PTR_ERR(result));
148 149 150 151 152 153 154 155 156 157 158
	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)
{
159
	if (handle->h_buffer_credits > EXT4_RESERVE_TRANS_BLOCKS)
160
		return 0;
161
	if (!ext4_journal_extend(handle, blocks_for_truncate(inode)))
162 163 164 165 166 167 168 169 170
		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.
 */
171
static int ext4_journal_test_restart(handle_t *handle, struct inode *inode)
172 173
{
	jbd_debug(2, "restarting handle %p\n", handle);
174
	return ext4_journal_restart(handle, blocks_for_truncate(inode));
175 176 177 178 179
}

/*
 * Called at the last iput() if i_nlink is zero.
 */
180
void ext4_delete_inode (struct inode * inode)
181 182 183 184 185 186 187 188 189 190 191 192 193 194 195
{
	handle_t *handle;

	truncate_inode_pages(&inode->i_data, 0);

	if (is_bad_inode(inode))
		goto no_delete;

	handle = start_transaction(inode);
	if (IS_ERR(handle)) {
		/*
		 * 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.
		 */
196
		ext4_orphan_del(NULL, inode);
197 198 199 200 201 202 203
		goto no_delete;
	}

	if (IS_SYNC(inode))
		handle->h_sync = 1;
	inode->i_size = 0;
	if (inode->i_blocks)
204
		ext4_truncate(inode);
205
	/*
206
	 * Kill off the orphan record which ext4_truncate created.
207
	 * AKPM: I think this can be inside the above `if'.
208
	 * Note that ext4_orphan_del() has to be able to cope with the
209
	 * deletion of a non-existent orphan - this is because we don't
210
	 * know if ext4_truncate() actually created an orphan record.
211 212
	 * (Well, we could do this if we need to, but heck - it works)
	 */
213 214
	ext4_orphan_del(handle, inode);
	EXT4_I(inode)->i_dtime	= get_seconds();
215 216 217 218 219 220 221 222

	/*
	 * 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.
	 */
223
	if (ext4_mark_inode_dirty(handle, inode))
224 225 226
		/* If that failed, just do the required in-core inode clear. */
		clear_inode(inode);
	else
227 228
		ext4_free_inode(handle, inode);
	ext4_journal_stop(handle);
229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246
	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;
}

/**
247
 *	ext4_block_to_path - parse the block number into array of offsets
248 249 250
 *	@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
251 252
 *	@boundary: set this non-zero if the referred-to block is likely to be
 *	       followed (on disk) by an indirect block.
253
 *
254
 *	To store the locations of file's data ext4 uses a data structure common
255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276
 *	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.
 */

277
static int ext4_block_to_path(struct inode *inode,
A
Aneesh Kumar K.V 已提交
278 279
			ext4_lblk_t i_block,
			ext4_lblk_t offsets[4], int *boundary)
280
{
281 282 283
	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,
284 285 286 287 288 289
		indirect_blocks = ptrs,
		double_blocks = (1 << (ptrs_bits * 2));
	int n = 0;
	int final = 0;

	if (i_block < 0) {
290
		ext4_warning (inode->i_sb, "ext4_block_to_path", "block < 0");
291 292 293 294
	} else if (i_block < direct_blocks) {
		offsets[n++] = i_block;
		final = direct_blocks;
	} else if ( (i_block -= direct_blocks) < indirect_blocks) {
295
		offsets[n++] = EXT4_IND_BLOCK;
296 297 298
		offsets[n++] = i_block;
		final = ptrs;
	} else if ((i_block -= indirect_blocks) < double_blocks) {
299
		offsets[n++] = EXT4_DIND_BLOCK;
300 301 302 303
		offsets[n++] = i_block >> ptrs_bits;
		offsets[n++] = i_block & (ptrs - 1);
		final = ptrs;
	} else if (((i_block -= double_blocks) >> (ptrs_bits * 2)) < ptrs) {
304
		offsets[n++] = EXT4_TIND_BLOCK;
305 306 307 308 309
		offsets[n++] = i_block >> (ptrs_bits * 2);
		offsets[n++] = (i_block >> ptrs_bits) & (ptrs - 1);
		offsets[n++] = i_block & (ptrs - 1);
		final = ptrs;
	} else {
310
		ext4_warning(inode->i_sb, "ext4_block_to_path",
311
				"block %lu > max",
312 313
				i_block + direct_blocks +
				indirect_blocks + double_blocks);
314 315 316 317 318 319 320
	}
	if (boundary)
		*boundary = final - 1 - (i_block & (ptrs - 1));
	return n;
}

/**
321
 *	ext4_get_branch - read the chain of indirect blocks leading to data
322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345
 *	@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).
346 347
 *
 *      Need to be called with
348
 *      down_read(&EXT4_I(inode)->i_data_sem)
349
 */
A
Aneesh Kumar K.V 已提交
350 351
static Indirect *ext4_get_branch(struct inode *inode, int depth,
				 ext4_lblk_t  *offsets,
352 353 354 355 356 357 358 359
				 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 */
360
	add_chain (chain, NULL, EXT4_I(inode)->i_data + *offsets);
361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380
	if (!p->key)
		goto no_block;
	while (--depth) {
		bh = sb_bread(sb, le32_to_cpu(p->key));
		if (!bh)
			goto failure;
		add_chain(++p, bh, (__le32*)bh->b_data + *++offsets);
		/* Reader: end */
		if (!p->key)
			goto no_block;
	}
	return NULL;

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

/**
381
 *	ext4_find_near - find a place for allocation with sufficient locality
382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399
 *	@inode: owner
 *	@ind: descriptor of indirect block.
 *
 *	This function returns the prefered place for block allocation.
 *	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.
 */
400
static ext4_fsblk_t ext4_find_near(struct inode *inode, Indirect *ind)
401
{
402
	struct ext4_inode_info *ei = EXT4_I(inode);
403 404
	__le32 *start = ind->bh ? (__le32*) ind->bh->b_data : ei->i_data;
	__le32 *p;
405 406
	ext4_fsblk_t bg_start;
	ext4_grpblk_t colour;
407 408 409 410 411 412 413 414 415 416 417 418 419 420 421

	/* 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.
	 */
422
	bg_start = ext4_group_first_block_no(inode->i_sb, ei->i_block_group);
423
	colour = (current->pid % 16) *
424
			(EXT4_BLOCKS_PER_GROUP(inode->i_sb) / 16);
425 426 427 428
	return bg_start + colour;
}

/**
429
 *	ext4_find_goal - find a prefered place for allocation.
430 431 432 433 434 435 436 437 438 439
 *	@inode: owner
 *	@block:  block we want
 *	@chain:  chain of indirect blocks
 *	@partial: pointer to the last triple within a chain
 *	@goal:	place to store the result.
 *
 *	Normally this function find the prefered place for block allocation,
 *	stores it in *@goal and returns zero.
 */

A
Aneesh Kumar K.V 已提交
440
static ext4_fsblk_t ext4_find_goal(struct inode *inode, ext4_lblk_t block,
441 442
		Indirect chain[4], Indirect *partial)
{
443
	struct ext4_block_alloc_info *block_i;
444

445
	block_i =  EXT4_I(inode)->i_block_alloc_info;
446 447 448 449 450 451 452 453 454 455

	/*
	 * try the heuristic for sequential allocation,
	 * failing that at least try to get decent locality.
	 */
	if (block_i && (block == block_i->last_alloc_logical_block + 1)
		&& (block_i->last_alloc_physical_block != 0)) {
		return block_i->last_alloc_physical_block + 1;
	}

456
	return ext4_find_near(inode, partial);
457 458 459
}

/**
460
 *	ext4_blks_to_allocate: Look up the block map and count the number
461 462 463 464 465 466 467 468 469 470
 *	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.
 */
471
static int ext4_blks_to_allocate(Indirect *branch, int k, unsigned long blks,
472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497
		int blocks_to_boundary)
{
	unsigned long count = 0;

	/*
	 * 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;
}

/**
498
 *	ext4_alloc_blocks: multiple allocate blocks needed for a branch
499 500 501 502 503 504 505 506
 *	@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
 */
507 508 509
static int ext4_alloc_blocks(handle_t *handle, struct inode *inode,
			ext4_fsblk_t goal, int indirect_blks, int blks,
			ext4_fsblk_t new_blocks[4], int *err)
510 511 512 513
{
	int target, i;
	unsigned long count = 0;
	int index = 0;
514
	ext4_fsblk_t current_block = 0;
515 516 517 518 519 520 521 522 523 524 525 526 527 528 529
	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)
	 */
	target = blks + indirect_blks;

	while (1) {
		count = target;
		/* allocating blocks for indirect blocks and direct blocks */
530
		current_block = ext4_new_blocks(handle,inode,goal,&count,err);
531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553
		if (*err)
			goto failed_out;

		target -= count;
		/* allocate blocks for indirect blocks */
		while (index < indirect_blks && count) {
			new_blocks[index++] = current_block++;
			count--;
		}

		if (count > 0)
			break;
	}

	/* save the new block number for the first direct block */
	new_blocks[index] = current_block;

	/* total number of blocks allocated for direct blocks */
	ret = count;
	*err = 0;
	return ret;
failed_out:
	for (i = 0; i <index; i++)
554
		ext4_free_blocks(handle, inode, new_blocks[i], 1, 0);
555 556 557 558
	return ret;
}

/**
559
 *	ext4_alloc_branch - allocate and set up a chain of blocks.
560 561 562 563 564 565 566 567 568 569
 *	@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
570
 *	the same format as ext4_get_branch() would do. We are calling it after
571 572
 *	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
573
 *	picture as after the successful ext4_get_block(), except that in one
574 575 576 577 578 579
 *	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
580
 *	ext4_alloc_block() (normally -ENOSPC). Otherwise we set the chain
581 582
 *	as described above and return 0.
 */
583 584
static int ext4_alloc_branch(handle_t *handle, struct inode *inode,
			int indirect_blks, int *blks, ext4_fsblk_t goal,
A
Aneesh Kumar K.V 已提交
585
			ext4_lblk_t *offsets, Indirect *branch)
586 587 588 589 590 591
{
	int blocksize = inode->i_sb->s_blocksize;
	int i, n = 0;
	int err = 0;
	struct buffer_head *bh;
	int num;
592 593
	ext4_fsblk_t new_blocks[4];
	ext4_fsblk_t current_block;
594

595
	num = ext4_alloc_blocks(handle, inode, goal, indirect_blks,
596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613
				*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");
614
		err = ext4_journal_get_create_access(handle, bh);
615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638
		if (err) {
			unlock_buffer(bh);
			brelse(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;
		if ( n == indirect_blks) {
			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
			 */
			for (i=1; i < num; i++)
				*(branch[n].p + i) = cpu_to_le32(++current_block);
		}
		BUFFER_TRACE(bh, "marking uptodate");
		set_buffer_uptodate(bh);
		unlock_buffer(bh);

639 640
		BUFFER_TRACE(bh, "call ext4_journal_dirty_metadata");
		err = ext4_journal_dirty_metadata(handle, bh);
641 642 643 644 645 646 647 648
		if (err)
			goto failed;
	}
	*blks = num;
	return err;
failed:
	/* Allocation failed, free what we already allocated */
	for (i = 1; i <= n ; i++) {
649
		BUFFER_TRACE(branch[i].bh, "call jbd2_journal_forget");
650
		ext4_journal_forget(handle, branch[i].bh);
651 652
	}
	for (i = 0; i <indirect_blks; i++)
653
		ext4_free_blocks(handle, inode, new_blocks[i], 1, 0);
654

655
	ext4_free_blocks(handle, inode, new_blocks[i], num, 0);
656 657 658 659 660

	return err;
}

/**
661
 * ext4_splice_branch - splice the allocated branch onto inode.
662 663 664
 * @inode: owner
 * @block: (logical) number of block we are adding
 * @chain: chain of indirect blocks (with a missing link - see
665
 *	ext4_alloc_branch)
666 667 668 669 670 671 672 673
 * @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.
 */
674
static int ext4_splice_branch(handle_t *handle, struct inode *inode,
A
Aneesh Kumar K.V 已提交
675
			ext4_lblk_t block, Indirect *where, int num, int blks)
676 677 678
{
	int i;
	int err = 0;
679 680
	struct ext4_block_alloc_info *block_i;
	ext4_fsblk_t current_block;
681

682
	block_i = EXT4_I(inode)->i_block_alloc_info;
683 684 685 686 687 688 689
	/*
	 * 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");
690
		err = ext4_journal_get_write_access(handle, where->bh);
691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720
		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++)
			*(where->p + i ) = cpu_to_le32(current_block++);
	}

	/*
	 * update the most recently allocated logical & physical block
	 * in i_block_alloc_info, to assist find the proper goal block for next
	 * allocation
	 */
	if (block_i) {
		block_i->last_alloc_logical_block = block + blks - 1;
		block_i->last_alloc_physical_block =
				le32_to_cpu(where[num].key) + blks - 1;
	}

	/* We are done with atomic stuff, now do the rest of housekeeping */

K
Kalpak Shah 已提交
721
	inode->i_ctime = ext4_current_time(inode);
722
	ext4_mark_inode_dirty(handle, inode);
723 724 725 726 727 728 729 730 731

	/* 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
732
		 * generic_commit_write->__mark_inode_dirty->ext4_dirty_inode.
733 734
		 */
		jbd_debug(5, "splicing indirect only\n");
735 736
		BUFFER_TRACE(where->bh, "call ext4_journal_dirty_metadata");
		err = ext4_journal_dirty_metadata(handle, where->bh);
737 738 739 740 741 742 743 744 745 746 747 748 749
		if (err)
			goto err_out;
	} else {
		/*
		 * OK, we spliced it into the inode itself on a direct block.
		 * Inode was dirtied above.
		 */
		jbd_debug(5, "splicing direct\n");
	}
	return err;

err_out:
	for (i = 1; i <= num; i++) {
750
		BUFFER_TRACE(where[i].bh, "call jbd2_journal_forget");
751
		ext4_journal_forget(handle, where[i].bh);
752 753
		ext4_free_blocks(handle, inode,
					le32_to_cpu(where[i-1].key), 1, 0);
754
	}
755
	ext4_free_blocks(handle, inode, le32_to_cpu(where[num].key), blks, 0);
756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777

	return err;
}

/*
 * 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.
 *
 * The BKL may not be held on entry here.  Be sure to take it early.
 * return > 0, # of blocks mapped or allocated.
 * return = 0, if plain lookup failed.
 * return < 0, error case.
778 779 780
 *
 *
 * Need to be called with
781 782
 * down_read(&EXT4_I(inode)->i_data_sem) if not allocating file system block
 * (ie, create is zero). Otherwise down_write(&EXT4_I(inode)->i_data_sem)
783
 */
784
int ext4_get_blocks_handle(handle_t *handle, struct inode *inode,
A
Aneesh Kumar K.V 已提交
785
		ext4_lblk_t iblock, unsigned long maxblocks,
786 787 788 789
		struct buffer_head *bh_result,
		int create, int extend_disksize)
{
	int err = -EIO;
A
Aneesh Kumar K.V 已提交
790
	ext4_lblk_t offsets[4];
791 792
	Indirect chain[4];
	Indirect *partial;
793
	ext4_fsblk_t goal;
794 795 796
	int indirect_blks;
	int blocks_to_boundary = 0;
	int depth;
797
	struct ext4_inode_info *ei = EXT4_I(inode);
798
	int count = 0;
799
	ext4_fsblk_t first_block = 0;
800 801


A
Alex Tomas 已提交
802
	J_ASSERT(!(EXT4_I(inode)->i_flags & EXT4_EXTENTS_FL));
803
	J_ASSERT(handle != NULL || create == 0);
A
Aneesh Kumar K.V 已提交
804 805
	depth = ext4_block_to_path(inode, iblock, offsets,
					&blocks_to_boundary);
806 807 808 809

	if (depth == 0)
		goto out;

810
	partial = ext4_get_branch(inode, depth, offsets, chain, &err);
811 812 813 814 815 816 817 818

	/* 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) {
819
			ext4_fsblk_t blk;
820 821 822 823 824 825 826 827

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

			if (blk == first_block + count)
				count++;
			else
				break;
		}
828
		goto got_it;
829 830 831 832 833 834 835 836 837 838 839
	}

	/* Next simple case - plain lookup or failed read of indirect block */
	if (!create || err == -EIO)
		goto cleanup;

	/*
	 * Okay, we need to do block allocation.  Lazily initialize the block
	 * allocation info here if necessary
	*/
	if (S_ISREG(inode->i_mode) && (!ei->i_block_alloc_info))
840
		ext4_init_block_alloc_info(inode);
841

842
	goal = ext4_find_goal(inode, iblock, chain, partial);
843 844 845 846 847 848 849 850

	/* 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.
	 */
851
	count = ext4_blks_to_allocate(partial, indirect_blks,
852 853
					maxblocks, blocks_to_boundary);
	/*
854
	 * Block out ext4_truncate while we alter the tree
855
	 */
856
	err = ext4_alloc_branch(handle, inode, indirect_blks, &count, goal,
857 858 859
				offsets + (partial - chain), partial);

	/*
860
	 * The ext4_splice_branch call will free and forget any buffers
861 862 863 864 865 866
	 * 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)
867
		err = ext4_splice_branch(handle, inode, iblock,
868 869
					partial, indirect_blks, count);
	/*
870
	 * i_disksize growing is protected by i_data_sem.  Don't forget to
871
	 * protect it if you're about to implement concurrent
872
	 * ext4_get_block() -bzzz
873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897
	*/
	if (!err && extend_disksize && inode->i_size > ei->i_disksize)
		ei->i_disksize = inode->i_size;
	if (err)
		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;
}

898
#define DIO_CREDITS (EXT4_RESERVE_TRANS_BLOCKS + 32)
899

900 901 902 903 904
int ext4_get_blocks_wrap(handle_t *handle, struct inode *inode, sector_t block,
			unsigned long max_blocks, struct buffer_head *bh,
			int create, int extend_disksize)
{
	int retval;
905 906 907 908 909 910 911 912
	/*
	 * Try to see if we can get  the block without requesting
	 * for new file system block.
	 */
	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,
				bh, 0, 0);
913
	} else {
914 915
		retval = ext4_get_blocks_handle(handle,
				inode, block, max_blocks, bh, 0, 0);
916
	}
917 918 919 920 921 922 923 924 925 926 927 928 929
	up_read((&EXT4_I(inode)->i_data_sem));
	if (!create || (retval > 0))
		return retval;

	/*
	 * We need to allocate new blocks which will result
	 * in i_data update
	 */
	down_write((&EXT4_I(inode)->i_data_sem));
	/*
	 * We need to check for EXT4 here because migrate
	 * could have changed the inode type in between
	 */
930 931 932 933 934 935 936
	if (EXT4_I(inode)->i_flags & EXT4_EXTENTS_FL) {
		retval =  ext4_ext_get_blocks(handle, inode, block, max_blocks,
				bh, create, extend_disksize);
	} else {
		retval = ext4_get_blocks_handle(handle, inode, block,
				max_blocks, bh, create, extend_disksize);
	}
937
	up_write((&EXT4_I(inode)->i_data_sem));
938 939 940
	return retval;
}

941
static int ext4_get_block(struct inode *inode, sector_t iblock,
942 943
			struct buffer_head *bh_result, int create)
{
944
	handle_t *handle = ext4_journal_current_handle();
945 946 947 948 949 950 951 952 953 954 955 956 957 958
	int ret = 0;
	unsigned max_blocks = bh_result->b_size >> inode->i_blkbits;

	if (!create)
		goto get_block;		/* A read */

	if (max_blocks == 1)
		goto get_block;		/* A single block get */

	if (handle->h_transaction->t_state == T_LOCKED) {
		/*
		 * Huge direct-io writes can hold off commits for long
		 * periods of time.  Let this commit run.
		 */
959 960
		ext4_journal_stop(handle);
		handle = ext4_journal_start(inode, DIO_CREDITS);
961 962 963 964 965
		if (IS_ERR(handle))
			ret = PTR_ERR(handle);
		goto get_block;
	}

966
	if (handle->h_buffer_credits <= EXT4_RESERVE_TRANS_BLOCKS) {
967 968 969
		/*
		 * Getting low on buffer credits...
		 */
970
		ret = ext4_journal_extend(handle, DIO_CREDITS);
971 972 973 974
		if (ret > 0) {
			/*
			 * Couldn't extend the transaction.  Start a new one.
			 */
975
			ret = ext4_journal_restart(handle, DIO_CREDITS);
976 977 978 979 980
		}
	}

get_block:
	if (ret == 0) {
A
Alex Tomas 已提交
981
		ret = ext4_get_blocks_wrap(handle, inode, iblock,
982 983 984 985 986 987 988 989 990 991 992 993
					max_blocks, bh_result, create, 0);
		if (ret > 0) {
			bh_result->b_size = (ret << inode->i_blkbits);
			ret = 0;
		}
	}
	return ret;
}

/*
 * `handle' can be NULL if create is zero
 */
994
struct buffer_head *ext4_getblk(handle_t *handle, struct inode *inode,
A
Aneesh Kumar K.V 已提交
995
				ext4_lblk_t block, int create, int *errp)
996 997 998 999 1000 1001 1002 1003 1004
{
	struct buffer_head dummy;
	int fatal = 0, err;

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

	dummy.b_state = 0;
	dummy.b_blocknr = -1000;
	buffer_trace_init(&dummy.b_history);
A
Alex Tomas 已提交
1005
	err = ext4_get_blocks_wrap(handle, inode, block, 1,
1006 1007
					&dummy, create, 1);
	/*
1008
	 * ext4_get_blocks_handle() returns number of blocks
1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025
	 * mapped. 0 in case of a HOLE.
	 */
	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
1026
			J_ASSERT(handle != NULL);
1027 1028 1029 1030 1031

			/*
			 * 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
1032
			 * writes use ext4_get_block instead, so it's not a
1033 1034 1035 1036
			 * problem.
			 */
			lock_buffer(bh);
			BUFFER_TRACE(bh, "call get_create_access");
1037
			fatal = ext4_journal_get_create_access(handle, bh);
1038 1039 1040 1041 1042
			if (!fatal && !buffer_uptodate(bh)) {
				memset(bh->b_data,0,inode->i_sb->s_blocksize);
				set_buffer_uptodate(bh);
			}
			unlock_buffer(bh);
1043 1044
			BUFFER_TRACE(bh, "call ext4_journal_dirty_metadata");
			err = ext4_journal_dirty_metadata(handle, bh);
1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060
			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;
}

1061
struct buffer_head *ext4_bread(handle_t *handle, struct inode *inode,
A
Aneesh Kumar K.V 已提交
1062
			       ext4_lblk_t block, int create, int *err)
1063 1064 1065
{
	struct buffer_head * bh;

1066
	bh = ext4_getblk(handle, inode, block, create, err);
1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114
	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;
}

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))
{
	struct buffer_head *bh;
	unsigned block_start, block_end;
	unsigned blocksize = head->b_size;
	int err, ret = 0;
	struct buffer_head *next;

	for (	bh = head, block_start = 0;
		ret == 0 && (bh != head || !block_start);
		block_start = block_end, bh = next)
	{
		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
1115
 * close off a transaction and start a new one between the ext4_get_block()
1116
 * and the commit_write().  So doing the jbd2_journal_start at the start of
1117 1118
 * prepare_write() is the right place.
 *
1119 1120
 * Also, this function can nest inside ext4_writepage() ->
 * block_write_full_page(). In that case, we *know* that ext4_writepage()
1121 1122 1123 1124
 * 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.
 *
1125
 * By accident, ext4 can be reentered when a transaction is open via
1126 1127 1128 1129 1130 1131
 * 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.
 *
1132
 * So what we do is to rely on the fact that jbd2_journal_stop/journal_start
1133 1134 1135 1136 1137 1138 1139 1140 1141
 * 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,
					struct buffer_head *bh)
{
	if (!buffer_mapped(bh) || buffer_freed(bh))
		return 0;
1142
	return ext4_journal_get_write_access(handle, bh);
1143 1144
}

N
Nick Piggin 已提交
1145 1146 1147
static int ext4_write_begin(struct file *file, struct address_space *mapping,
				loff_t pos, unsigned len, unsigned flags,
				struct page **pagep, void **fsdata)
1148
{
N
Nick Piggin 已提交
1149
 	struct inode *inode = mapping->host;
1150
	int ret, needed_blocks = ext4_writepage_trans_blocks(inode);
1151 1152
	handle_t *handle;
	int retries = 0;
N
Nick Piggin 已提交
1153 1154 1155 1156 1157 1158 1159
 	struct page *page;
 	pgoff_t index;
 	unsigned from, to;

 	index = pos >> PAGE_CACHE_SHIFT;
 	from = pos & (PAGE_CACHE_SIZE - 1);
 	to = from + len;
1160 1161

retry:
N
Nick Piggin 已提交
1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172
 	page = __grab_cache_page(mapping, index);
 	if (!page)
 		return -ENOMEM;
 	*pagep = page;

  	handle = ext4_journal_start(inode, needed_blocks);
  	if (IS_ERR(handle)) {
 		unlock_page(page);
 		page_cache_release(page);
  		ret = PTR_ERR(handle);
  		goto out;
1173
	}
1174

N
Nick Piggin 已提交
1175 1176 1177 1178
	ret = block_write_begin(file, mapping, pos, len, flags, pagep, fsdata,
							ext4_get_block);

	if (!ret && ext4_should_journal_data(inode)) {
1179 1180 1181
		ret = walk_page_buffers(handle, page_buffers(page),
				from, to, NULL, do_journal_get_write_access);
	}
N
Nick Piggin 已提交
1182 1183

	if (ret) {
1184
		ext4_journal_stop(handle);
N
Nick Piggin 已提交
1185 1186 1187 1188
 		unlock_page(page);
 		page_cache_release(page);
	}

1189
	if (ret == -ENOSPC && ext4_should_retry_alloc(inode->i_sb, &retries))
1190
		goto retry;
1191
out:
1192 1193 1194
	return ret;
}

1195
int ext4_journal_dirty_data(handle_t *handle, struct buffer_head *bh)
1196
{
1197
	int err = jbd2_journal_dirty_data(handle, bh);
1198
	if (err)
1199
		ext4_journal_abort_handle(__FUNCTION__, __FUNCTION__,
N
Nick Piggin 已提交
1200
						bh, handle, err);
1201 1202 1203
	return err;
}

N
Nick Piggin 已提交
1204 1205
/* For write_end() in data=journal mode */
static int write_end_fn(handle_t *handle, struct buffer_head *bh)
1206 1207 1208 1209
{
	if (!buffer_mapped(bh) || buffer_freed(bh))
		return 0;
	set_buffer_uptodate(bh);
1210
	return ext4_journal_dirty_metadata(handle, bh);
1211 1212
}

N
Nick Piggin 已提交
1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235
/*
 * Generic write_end handler for ordered and writeback ext4 journal modes.
 * We can't use generic_write_end, because that unlocks the page and we need to
 * unlock the page after ext4_journal_stop, but ext4_journal_stop must run
 * after block_write_end.
 */
static int ext4_generic_write_end(struct file *file,
				struct address_space *mapping,
				loff_t pos, unsigned len, unsigned copied,
				struct page *page, void *fsdata)
{
	struct inode *inode = file->f_mapping->host;

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

	if (pos+copied > inode->i_size) {
		i_size_write(inode, pos+copied);
		mark_inode_dirty(inode);
	}

	return copied;
}

1236 1237 1238 1239
/*
 * 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().
 *
1240
 * ext4 never places buffers on inode->i_mapping->private_list.  metadata
1241 1242
 * buffers are managed internally.
 */
N
Nick Piggin 已提交
1243 1244 1245 1246
static int ext4_ordered_write_end(struct file *file,
				struct address_space *mapping,
				loff_t pos, unsigned len, unsigned copied,
				struct page *page, void *fsdata)
1247
{
1248
	handle_t *handle = ext4_journal_current_handle();
N
Nick Piggin 已提交
1249 1250
	struct inode *inode = file->f_mapping->host;
	unsigned from, to;
1251 1252
	int ret = 0, ret2;

N
Nick Piggin 已提交
1253 1254 1255
	from = pos & (PAGE_CACHE_SIZE - 1);
	to = from + len;

1256
	ret = walk_page_buffers(handle, page_buffers(page),
1257
		from, to, NULL, ext4_journal_dirty_data);
1258 1259 1260

	if (ret == 0) {
		/*
N
Nick Piggin 已提交
1261
		 * generic_write_end() will run mark_inode_dirty() if i_size
1262 1263 1264 1265 1266
		 * changes.  So let's piggyback the i_disksize mark_inode_dirty
		 * into that.
		 */
		loff_t new_i_size;

N
Nick Piggin 已提交
1267
		new_i_size = pos + copied;
1268 1269
		if (new_i_size > EXT4_I(inode)->i_disksize)
			EXT4_I(inode)->i_disksize = new_i_size;
N
Nick Piggin 已提交
1270 1271 1272 1273
		copied = ext4_generic_write_end(file, mapping, pos, len, copied,
							page, fsdata);
		if (copied < 0)
			ret = copied;
1274
	}
1275
	ret2 = ext4_journal_stop(handle);
1276 1277
	if (!ret)
		ret = ret2;
N
Nick Piggin 已提交
1278 1279 1280 1281
	unlock_page(page);
	page_cache_release(page);

	return ret ? ret : copied;
1282 1283
}

N
Nick Piggin 已提交
1284 1285 1286 1287
static int ext4_writeback_write_end(struct file *file,
				struct address_space *mapping,
				loff_t pos, unsigned len, unsigned copied,
				struct page *page, void *fsdata)
1288
{
1289
	handle_t *handle = ext4_journal_current_handle();
N
Nick Piggin 已提交
1290
	struct inode *inode = file->f_mapping->host;
1291 1292 1293
	int ret = 0, ret2;
	loff_t new_i_size;

N
Nick Piggin 已提交
1294
	new_i_size = pos + copied;
1295 1296
	if (new_i_size > EXT4_I(inode)->i_disksize)
		EXT4_I(inode)->i_disksize = new_i_size;
1297

N
Nick Piggin 已提交
1298 1299 1300 1301
	copied = ext4_generic_write_end(file, mapping, pos, len, copied,
							page, fsdata);
	if (copied < 0)
		ret = copied;
1302

1303
	ret2 = ext4_journal_stop(handle);
1304 1305
	if (!ret)
		ret = ret2;
N
Nick Piggin 已提交
1306 1307 1308 1309
	unlock_page(page);
	page_cache_release(page);

	return ret ? ret : copied;
1310 1311
}

N
Nick Piggin 已提交
1312 1313 1314 1315
static int ext4_journalled_write_end(struct file *file,
				struct address_space *mapping,
				loff_t pos, unsigned len, unsigned copied,
				struct page *page, void *fsdata)
1316
{
1317
	handle_t *handle = ext4_journal_current_handle();
N
Nick Piggin 已提交
1318
	struct inode *inode = mapping->host;
1319 1320
	int ret = 0, ret2;
	int partial = 0;
N
Nick Piggin 已提交
1321
	unsigned from, to;
1322

N
Nick Piggin 已提交
1323 1324 1325 1326 1327 1328 1329 1330
	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);
	}
1331 1332

	ret = walk_page_buffers(handle, page_buffers(page), from,
N
Nick Piggin 已提交
1333
				to, &partial, write_end_fn);
1334 1335
	if (!partial)
		SetPageUptodate(page);
N
Nick Piggin 已提交
1336 1337
	if (pos+copied > inode->i_size)
		i_size_write(inode, pos+copied);
1338 1339 1340 1341
	EXT4_I(inode)->i_state |= EXT4_STATE_JDATA;
	if (inode->i_size > EXT4_I(inode)->i_disksize) {
		EXT4_I(inode)->i_disksize = inode->i_size;
		ret2 = ext4_mark_inode_dirty(handle, inode);
1342 1343 1344
		if (!ret)
			ret = ret2;
	}
N
Nick Piggin 已提交
1345

1346
	ret2 = ext4_journal_stop(handle);
1347 1348
	if (!ret)
		ret = ret2;
N
Nick Piggin 已提交
1349 1350 1351 1352
	unlock_page(page);
	page_cache_release(page);

	return ret ? ret : copied;
1353 1354 1355 1356 1357 1358 1359
}

/*
 * 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
1360
 * journal.  If somebody makes a swapfile on an ext4 data-journaling
1361 1362 1363 1364 1365 1366 1367 1368
 * 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.
 */
1369
static sector_t ext4_bmap(struct address_space *mapping, sector_t block)
1370 1371 1372 1373 1374
{
	struct inode *inode = mapping->host;
	journal_t *journal;
	int err;

1375
	if (EXT4_I(inode)->i_state & EXT4_STATE_JDATA) {
1376 1377 1378 1379 1380 1381 1382 1383 1384 1385 1386
		/*
		 * 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.)
		 *
1387
		 * NB. EXT4_STATE_JDATA is not set on files other than
1388 1389 1390 1391 1392 1393
		 * 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.
		 */

1394 1395
		EXT4_I(inode)->i_state &= ~EXT4_STATE_JDATA;
		journal = EXT4_JOURNAL(inode);
1396 1397 1398
		jbd2_journal_lock_updates(journal);
		err = jbd2_journal_flush(journal);
		jbd2_journal_unlock_updates(journal);
1399 1400 1401 1402 1403

		if (err)
			return 0;
	}

1404
	return generic_block_bmap(mapping,block,ext4_get_block);
1405 1406 1407 1408 1409 1410 1411 1412 1413 1414 1415 1416 1417 1418
}

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;
}

1419
static int jbd2_journal_dirty_data_fn(handle_t *handle, struct buffer_head *bh)
1420 1421
{
	if (buffer_mapped(bh))
1422
		return ext4_journal_dirty_data(handle, bh);
1423 1424 1425 1426 1427 1428
	return 0;
}

/*
 * Note that we always start a transaction even if we're not journalling
 * data.  This is to preserve ordering: any hole instantiation within
1429
 * __block_write_full_page -> ext4_get_block() should be journalled
1430 1431 1432 1433 1434 1435 1436
 * along with the data so we don't crash and then get metadata which
 * refers to old data.
 *
 * In all journalling modes block_write_full_page() will start the I/O.
 *
 * Problem:
 *
1437 1438
 *	ext4_writepage() -> kmalloc() -> __alloc_pages() -> page_launder() ->
 *		ext4_writepage()
1439 1440 1441
 *
 * Similar for:
 *
1442
 *	ext4_file_write() -> generic_file_write() -> __alloc_pages() -> ...
1443
 *
1444
 * Same applies to ext4_get_block().  We will deadlock on various things like
1445
 * lock_journal and i_data_sem
1446 1447 1448 1449 1450 1451 1452 1453 1454 1455 1456 1457 1458 1459 1460 1461 1462 1463 1464 1465 1466 1467 1468 1469 1470 1471 1472 1473 1474 1475 1476 1477
 *
 * Setting PF_MEMALLOC here doesn't work - too many internal memory
 * allocations fail.
 *
 * 16May01: If we're reentered then journal_current_handle() will be
 *	    non-zero. We simply *return*.
 *
 * 1 July 2001: @@@ FIXME:
 *   In journalled data mode, a data buffer may be metadata against the
 *   current transaction.  But the same file is part of a shared mapping
 *   and someone does a writepage() on it.
 *
 *   We will move the buffer onto the async_data list, but *after* it has
 *   been dirtied. So there's a small window where we have dirty data on
 *   BJ_Metadata.
 *
 *   Note that this only applies to the last partial page in the file.  The
 *   bit which block_write_full_page() uses prepare/commit for.  (That's
 *   broken code anyway: it's wrong for msync()).
 *
 *   It's a rare case: affects the final partial page, for journalled data
 *   where the file is subject to bith write() and writepage() in the same
 *   transction.  To fix it we'll need a custom block_write_full_page().
 *   We'll probably need that anyway for journalling writepage() output.
 *
 * We don't honour synchronous mounts for writepage().  That would be
 * disastrous.  Any write() or metadata operation will sync the fs for
 * us.
 *
 * AKPM2: if all the page's buffers are mapped to disk and !data=journal,
 * we don't need to open a transaction here.
 */
1478
static int ext4_ordered_writepage(struct page *page,
1479 1480 1481 1482 1483 1484 1485 1486 1487 1488 1489 1490 1491 1492
				struct writeback_control *wbc)
{
	struct inode *inode = page->mapping->host;
	struct buffer_head *page_bufs;
	handle_t *handle = NULL;
	int ret = 0;
	int err;

	J_ASSERT(PageLocked(page));

	/*
	 * We give up here if we're reentered, because it might be for a
	 * different filesystem.
	 */
1493
	if (ext4_journal_current_handle())
1494 1495
		goto out_fail;

1496
	handle = ext4_journal_start(inode, ext4_writepage_trans_blocks(inode));
1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509 1510

	if (IS_ERR(handle)) {
		ret = PTR_ERR(handle);
		goto out_fail;
	}

	if (!page_has_buffers(page)) {
		create_empty_buffers(page, inode->i_sb->s_blocksize,
				(1 << BH_Dirty)|(1 << BH_Uptodate));
	}
	page_bufs = page_buffers(page);
	walk_page_buffers(handle, page_bufs, 0,
			PAGE_CACHE_SIZE, NULL, bget_one);

1511
	ret = block_write_full_page(page, ext4_get_block, wbc);
1512 1513 1514 1515 1516 1517 1518 1519 1520 1521 1522 1523 1524 1525 1526

	/*
	 * The page can become unlocked at any point now, and
	 * truncate can then come in and change things.  So we
	 * can't touch *page from now on.  But *page_bufs is
	 * safe due to elevated refcount.
	 */

	/*
	 * And attach them to the current transaction.  But only if
	 * block_write_full_page() succeeded.  Otherwise they are unmapped,
	 * and generally junk.
	 */
	if (ret == 0) {
		err = walk_page_buffers(handle, page_bufs, 0, PAGE_CACHE_SIZE,
1527
					NULL, jbd2_journal_dirty_data_fn);
1528 1529 1530 1531 1532
		if (!ret)
			ret = err;
	}
	walk_page_buffers(handle, page_bufs, 0,
			PAGE_CACHE_SIZE, NULL, bput_one);
1533
	err = ext4_journal_stop(handle);
1534 1535 1536 1537 1538 1539 1540 1541 1542 1543
	if (!ret)
		ret = err;
	return ret;

out_fail:
	redirty_page_for_writepage(wbc, page);
	unlock_page(page);
	return ret;
}

1544
static int ext4_writeback_writepage(struct page *page,
1545 1546 1547 1548 1549 1550 1551
				struct writeback_control *wbc)
{
	struct inode *inode = page->mapping->host;
	handle_t *handle = NULL;
	int ret = 0;
	int err;

1552
	if (ext4_journal_current_handle())
1553 1554
		goto out_fail;

1555
	handle = ext4_journal_start(inode, ext4_writepage_trans_blocks(inode));
1556 1557 1558 1559 1560
	if (IS_ERR(handle)) {
		ret = PTR_ERR(handle);
		goto out_fail;
	}

1561 1562
	if (test_opt(inode->i_sb, NOBH) && ext4_should_writeback_data(inode))
		ret = nobh_writepage(page, ext4_get_block, wbc);
1563
	else
1564
		ret = block_write_full_page(page, ext4_get_block, wbc);
1565

1566
	err = ext4_journal_stop(handle);
1567 1568 1569 1570 1571 1572 1573 1574 1575 1576
	if (!ret)
		ret = err;
	return ret;

out_fail:
	redirty_page_for_writepage(wbc, page);
	unlock_page(page);
	return ret;
}

1577
static int ext4_journalled_writepage(struct page *page,
1578 1579 1580 1581 1582 1583 1584
				struct writeback_control *wbc)
{
	struct inode *inode = page->mapping->host;
	handle_t *handle = NULL;
	int ret = 0;
	int err;

1585
	if (ext4_journal_current_handle())
1586 1587
		goto no_write;

1588
	handle = ext4_journal_start(inode, ext4_writepage_trans_blocks(inode));
1589 1590 1591 1592 1593 1594 1595 1596 1597 1598 1599 1600
	if (IS_ERR(handle)) {
		ret = PTR_ERR(handle);
		goto no_write;
	}

	if (!page_has_buffers(page) || PageChecked(page)) {
		/*
		 * It's mmapped pagecache.  Add buffers and journal it.  There
		 * doesn't seem much point in redirtying the page here.
		 */
		ClearPageChecked(page);
		ret = block_prepare_write(page, 0, PAGE_CACHE_SIZE,
1601
					ext4_get_block);
1602
		if (ret != 0) {
1603
			ext4_journal_stop(handle);
1604 1605 1606 1607 1608 1609
			goto out_unlock;
		}
		ret = walk_page_buffers(handle, page_buffers(page), 0,
			PAGE_CACHE_SIZE, NULL, do_journal_get_write_access);

		err = walk_page_buffers(handle, page_buffers(page), 0,
N
Nick Piggin 已提交
1610
				PAGE_CACHE_SIZE, NULL, write_end_fn);
1611 1612
		if (ret == 0)
			ret = err;
1613
		EXT4_I(inode)->i_state |= EXT4_STATE_JDATA;
1614 1615 1616 1617 1618 1619 1620
		unlock_page(page);
	} else {
		/*
		 * It may be a page full of checkpoint-mode buffers.  We don't
		 * really know unless we go poke around in the buffer_heads.
		 * But block_write_full_page will do the right thing.
		 */
1621
		ret = block_write_full_page(page, ext4_get_block, wbc);
1622
	}
1623
	err = ext4_journal_stop(handle);
1624 1625 1626 1627 1628 1629 1630 1631 1632 1633 1634 1635
	if (!ret)
		ret = err;
out:
	return ret;

no_write:
	redirty_page_for_writepage(wbc, page);
out_unlock:
	unlock_page(page);
	goto out;
}

1636
static int ext4_readpage(struct file *file, struct page *page)
1637
{
1638
	return mpage_readpage(page, ext4_get_block);
1639 1640 1641
}

static int
1642
ext4_readpages(struct file *file, struct address_space *mapping,
1643 1644
		struct list_head *pages, unsigned nr_pages)
{
1645
	return mpage_readpages(mapping, pages, nr_pages, ext4_get_block);
1646 1647
}

1648
static void ext4_invalidatepage(struct page *page, unsigned long offset)
1649
{
1650
	journal_t *journal = EXT4_JOURNAL(page->mapping->host);
1651 1652 1653 1654 1655 1656 1657

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

1658
	jbd2_journal_invalidatepage(journal, page, offset);
1659 1660
}

1661
static int ext4_releasepage(struct page *page, gfp_t wait)
1662
{
1663
	journal_t *journal = EXT4_JOURNAL(page->mapping->host);
1664 1665 1666 1667

	WARN_ON(PageChecked(page));
	if (!page_has_buffers(page))
		return 0;
1668
	return jbd2_journal_try_to_free_buffers(journal, page, wait);
1669 1670 1671 1672 1673 1674 1675 1676 1677 1678
}

/*
 * 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
 * crashes then stale disk data _may_ be exposed inside the file.
 */
1679
static ssize_t ext4_direct_IO(int rw, struct kiocb *iocb,
1680 1681 1682 1683 1684
			const struct iovec *iov, loff_t offset,
			unsigned long nr_segs)
{
	struct file *file = iocb->ki_filp;
	struct inode *inode = file->f_mapping->host;
1685
	struct ext4_inode_info *ei = EXT4_I(inode);
1686 1687 1688 1689 1690 1691 1692 1693
	handle_t *handle = NULL;
	ssize_t ret;
	int orphan = 0;
	size_t count = iov_length(iov, nr_segs);

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

1694
		handle = ext4_journal_start(inode, DIO_CREDITS);
1695 1696 1697 1698 1699
		if (IS_ERR(handle)) {
			ret = PTR_ERR(handle);
			goto out;
		}
		if (final_size > inode->i_size) {
1700
			ret = ext4_orphan_add(handle, inode);
1701 1702 1703 1704 1705 1706 1707 1708 1709
			if (ret)
				goto out_stop;
			orphan = 1;
			ei->i_disksize = inode->i_size;
		}
	}

	ret = blockdev_direct_IO(rw, iocb, inode, inode->i_sb->s_bdev, iov,
				 offset, nr_segs,
1710
				 ext4_get_block, NULL);
1711 1712

	/*
1713
	 * Reacquire the handle: ext4_get_block() can restart the transaction
1714
	 */
1715
	handle = ext4_journal_current_handle();
1716 1717 1718 1719 1720 1721

out_stop:
	if (handle) {
		int err;

		if (orphan && inode->i_nlink)
1722
			ext4_orphan_del(handle, inode);
1723 1724 1725 1726 1727 1728 1729 1730 1731
		if (orphan && ret > 0) {
			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
1732
				 * ext4_mark_inode_dirty() to userspace.  So
1733 1734
				 * ignore it.
				 */
1735
				ext4_mark_inode_dirty(handle, inode);
1736 1737
			}
		}
1738
		err = ext4_journal_stop(handle);
1739 1740 1741 1742 1743 1744 1745 1746
		if (ret == 0)
			ret = err;
	}
out:
	return ret;
}

/*
1747
 * Pages can be marked dirty completely asynchronously from ext4's journalling
1748 1749 1750 1751 1752 1753 1754 1755 1756 1757 1758
 * 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.
 */
1759
static int ext4_journalled_set_page_dirty(struct page *page)
1760 1761 1762 1763 1764
{
	SetPageChecked(page);
	return __set_page_dirty_nobuffers(page);
}

1765 1766 1767 1768
static const struct address_space_operations ext4_ordered_aops = {
	.readpage	= ext4_readpage,
	.readpages	= ext4_readpages,
	.writepage	= ext4_ordered_writepage,
1769
	.sync_page	= block_sync_page,
N
Nick Piggin 已提交
1770 1771
	.write_begin	= ext4_write_begin,
	.write_end	= ext4_ordered_write_end,
1772 1773 1774 1775
	.bmap		= ext4_bmap,
	.invalidatepage	= ext4_invalidatepage,
	.releasepage	= ext4_releasepage,
	.direct_IO	= ext4_direct_IO,
1776 1777 1778
	.migratepage	= buffer_migrate_page,
};

1779 1780 1781 1782
static const struct address_space_operations ext4_writeback_aops = {
	.readpage	= ext4_readpage,
	.readpages	= ext4_readpages,
	.writepage	= ext4_writeback_writepage,
1783
	.sync_page	= block_sync_page,
N
Nick Piggin 已提交
1784 1785
	.write_begin	= ext4_write_begin,
	.write_end	= ext4_writeback_write_end,
1786 1787 1788 1789
	.bmap		= ext4_bmap,
	.invalidatepage	= ext4_invalidatepage,
	.releasepage	= ext4_releasepage,
	.direct_IO	= ext4_direct_IO,
1790 1791 1792
	.migratepage	= buffer_migrate_page,
};

1793 1794 1795 1796
static const struct address_space_operations ext4_journalled_aops = {
	.readpage	= ext4_readpage,
	.readpages	= ext4_readpages,
	.writepage	= ext4_journalled_writepage,
1797
	.sync_page	= block_sync_page,
N
Nick Piggin 已提交
1798 1799
	.write_begin	= ext4_write_begin,
	.write_end	= ext4_journalled_write_end,
1800 1801 1802 1803
	.set_page_dirty	= ext4_journalled_set_page_dirty,
	.bmap		= ext4_bmap,
	.invalidatepage	= ext4_invalidatepage,
	.releasepage	= ext4_releasepage,
1804 1805
};

1806
void ext4_set_aops(struct inode *inode)
1807
{
1808 1809 1810 1811
	if (ext4_should_order_data(inode))
		inode->i_mapping->a_ops = &ext4_ordered_aops;
	else if (ext4_should_writeback_data(inode))
		inode->i_mapping->a_ops = &ext4_writeback_aops;
1812
	else
1813
		inode->i_mapping->a_ops = &ext4_journalled_aops;
1814 1815 1816
}

/*
1817
 * ext4_block_truncate_page() zeroes out a mapping from file offset `from'
1818 1819 1820 1821
 * 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.
 */
A
Alex Tomas 已提交
1822
int ext4_block_truncate_page(handle_t *handle, struct page *page,
1823 1824
		struct address_space *mapping, loff_t from)
{
1825
	ext4_fsblk_t index = from >> PAGE_CACHE_SHIFT;
1826
	unsigned offset = from & (PAGE_CACHE_SIZE-1);
A
Aneesh Kumar K.V 已提交
1827 1828
	unsigned blocksize, length, pos;
	ext4_lblk_t iblock;
1829 1830 1831 1832 1833 1834 1835 1836 1837 1838 1839 1840 1841
	struct inode *inode = mapping->host;
	struct buffer_head *bh;
	int err = 0;

	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) &&
1842
	     ext4_should_writeback_data(inode) && PageUptodate(page)) {
1843
		zero_user_page(page, offset, length, KM_USER0);
1844 1845 1846 1847 1848 1849 1850 1851 1852 1853 1854 1855 1856 1857 1858 1859 1860 1861 1862 1863 1864 1865 1866 1867
		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");
1868
		ext4_get_block(inode, iblock, bh, 0);
1869 1870 1871 1872 1873 1874 1875 1876 1877 1878 1879 1880 1881 1882 1883 1884 1885 1886 1887 1888
		/* 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;
	}

1889
	if (ext4_should_journal_data(inode)) {
1890
		BUFFER_TRACE(bh, "get write access");
1891
		err = ext4_journal_get_write_access(handle, bh);
1892 1893 1894 1895
		if (err)
			goto unlock;
	}

1896
	zero_user_page(page, offset, length, KM_USER0);
1897 1898 1899 1900

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

	err = 0;
1901 1902
	if (ext4_should_journal_data(inode)) {
		err = ext4_journal_dirty_metadata(handle, bh);
1903
	} else {
1904 1905
		if (ext4_should_order_data(inode))
			err = ext4_journal_dirty_data(handle, bh);
1906 1907 1908 1909 1910 1911 1912 1913 1914 1915 1916 1917 1918 1919 1920 1921 1922 1923 1924 1925 1926 1927 1928
		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;
}

/**
1929
 *	ext4_find_shared - find the indirect blocks for partial truncation.
1930 1931
 *	@inode:	  inode in question
 *	@depth:	  depth of the affected branch
1932
 *	@offsets: offsets of pointers in that branch (see ext4_block_to_path)
1933 1934 1935
 *	@chain:	  place to store the pointers to partial indirect blocks
 *	@top:	  place to the (detached) top of branch
 *
1936
 *	This is a helper function used by ext4_truncate().
1937 1938 1939 1940 1941 1942 1943
 *
 *	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
1944
 *	past the truncation point is possible until ext4_truncate()
1945 1946 1947 1948 1949 1950 1951 1952 1953 1954 1955 1956 1957 1958 1959 1960 1961 1962
 *	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).  */

1963
static Indirect *ext4_find_shared(struct inode *inode, int depth,
A
Aneesh Kumar K.V 已提交
1964
			ext4_lblk_t offsets[4], Indirect chain[4], __le32 *top)
1965 1966 1967 1968 1969 1970 1971 1972
{
	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--)
		;
1973
	partial = ext4_get_branch(inode, k, offsets, chain, &err);
1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995
	/* 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;
	for (p=partial; p>chain && all_zeroes((__le32*)p->bh->b_data,p->p); p--)
		;
	/*
	 * 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;
1996
		/* Nope, don't do this in ext4.  Must leave the tree intact */
1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018
#if 0
		*p->p = 0;
#endif
	}
	/* Writer: end */

	while(partial > p) {
		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.
 */
2019 2020
static void ext4_clear_blocks(handle_t *handle, struct inode *inode,
		struct buffer_head *bh, ext4_fsblk_t block_to_free,
2021 2022 2023 2024 2025
		unsigned long count, __le32 *first, __le32 *last)
{
	__le32 *p;
	if (try_to_extend_transaction(handle, inode)) {
		if (bh) {
2026 2027
			BUFFER_TRACE(bh, "call ext4_journal_dirty_metadata");
			ext4_journal_dirty_metadata(handle, bh);
2028
		}
2029 2030
		ext4_mark_inode_dirty(handle, inode);
		ext4_journal_test_restart(handle, inode);
2031 2032
		if (bh) {
			BUFFER_TRACE(bh, "retaking write access");
2033
			ext4_journal_get_write_access(handle, bh);
2034 2035 2036 2037 2038
		}
	}

	/*
	 * Any buffers which are on the journal will be in memory. We find
2039
	 * them on the hash table so jbd2_journal_revoke() will run jbd2_journal_forget()
2040
	 * on them.  We've already detached each block from the file, so
2041
	 * bforget() in jbd2_journal_forget() should be safe.
2042
	 *
2043
	 * AKPM: turn on bforget in jbd2_journal_forget()!!!
2044 2045 2046 2047
	 */
	for (p = first; p < last; p++) {
		u32 nr = le32_to_cpu(*p);
		if (nr) {
A
Aneesh Kumar K.V 已提交
2048
			struct buffer_head *tbh;
2049 2050

			*p = 0;
A
Aneesh Kumar K.V 已提交
2051 2052
			tbh = sb_find_get_block(inode->i_sb, nr);
			ext4_forget(handle, 0, inode, tbh, nr);
2053 2054 2055
		}
	}

2056
	ext4_free_blocks(handle, inode, block_to_free, count, 0);
2057 2058 2059
}

/**
2060
 * ext4_free_data - free a list of data blocks
2061 2062 2063 2064 2065 2066 2067 2068 2069 2070 2071 2072 2073 2074 2075 2076 2077
 * @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.
 */
2078
static void ext4_free_data(handle_t *handle, struct inode *inode,
2079 2080 2081
			   struct buffer_head *this_bh,
			   __le32 *first, __le32 *last)
{
2082
	ext4_fsblk_t block_to_free = 0;    /* Starting block # of a run */
2083 2084 2085 2086
	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 */
2087
	ext4_fsblk_t nr;		    /* Current block # */
2088 2089 2090 2091 2092 2093
	__le32 *p;			    /* Pointer into inode/ind
					       for current block */
	int err;

	if (this_bh) {				/* For indirect block */
		BUFFER_TRACE(this_bh, "get_write_access");
2094
		err = ext4_journal_get_write_access(handle, this_bh);
2095 2096 2097 2098 2099 2100 2101 2102 2103 2104 2105 2106 2107 2108 2109 2110 2111
		/* 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 {
2112
				ext4_clear_blocks(handle, inode, this_bh,
2113 2114 2115 2116 2117 2118 2119 2120 2121 2122
						  block_to_free,
						  count, block_to_free_p, p);
				block_to_free = nr;
				block_to_free_p = p;
				count = 1;
			}
		}
	}

	if (count > 0)
2123
		ext4_clear_blocks(handle, inode, this_bh, block_to_free,
2124 2125 2126
				  count, block_to_free_p, p);

	if (this_bh) {
2127 2128
		BUFFER_TRACE(this_bh, "call ext4_journal_dirty_metadata");
		ext4_journal_dirty_metadata(handle, this_bh);
2129 2130 2131 2132
	}
}

/**
2133
 *	ext4_free_branches - free an array of branches
2134 2135 2136 2137 2138 2139 2140 2141 2142 2143 2144
 *	@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.
 */
2145
static void ext4_free_branches(handle_t *handle, struct inode *inode,
2146 2147 2148
			       struct buffer_head *parent_bh,
			       __le32 *first, __le32 *last, int depth)
{
2149
	ext4_fsblk_t nr;
2150 2151 2152 2153 2154 2155 2156
	__le32 *p;

	if (is_handle_aborted(handle))
		return;

	if (depth--) {
		struct buffer_head *bh;
2157
		int addr_per_block = EXT4_ADDR_PER_BLOCK(inode->i_sb);
2158 2159 2160 2161 2162 2163 2164 2165 2166 2167 2168 2169 2170 2171
		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) {
2172
				ext4_error(inode->i_sb, "ext4_free_branches",
2173
					   "Read failure, inode=%lu, block=%llu",
2174 2175 2176 2177 2178 2179
					   inode->i_ino, nr);
				continue;
			}

			/* This zaps the entire block.  Bottom up. */
			BUFFER_TRACE(bh, "free child branches");
2180
			ext4_free_branches(handle, inode, bh,
2181 2182 2183 2184 2185 2186 2187 2188
					   (__le32*)bh->b_data,
					   (__le32*)bh->b_data + addr_per_block,
					   depth);

			/*
			 * 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
2189
			 * jbd2_journal_revoke().
2190 2191 2192
			 *
			 * That's easy if it's exclusively part of this
			 * transaction.  But if it's part of the committing
2193
			 * transaction then jbd2_journal_forget() will simply
2194
			 * brelse() it.  That means that if the underlying
2195
			 * block is reallocated in ext4_get_block(),
2196 2197 2198 2199 2200 2201 2202 2203
			 * 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.
			 */
2204
			ext4_forget(handle, 1, inode, bh, bh->b_blocknr);
2205 2206 2207 2208 2209 2210 2211 2212 2213 2214 2215 2216 2217 2218 2219 2220 2221 2222 2223 2224

			/*
			 * 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.
			 */
			if (is_handle_aborted(handle))
				return;
			if (try_to_extend_transaction(handle, inode)) {
2225 2226
				ext4_mark_inode_dirty(handle, inode);
				ext4_journal_test_restart(handle, inode);
2227 2228
			}

2229
			ext4_free_blocks(handle, inode, nr, 1, 1);
2230 2231 2232 2233 2234 2235 2236

			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");
2237
				if (!ext4_journal_get_write_access(handle,
2238 2239 2240
								   parent_bh)){
					*p = 0;
					BUFFER_TRACE(parent_bh,
2241 2242
					"call ext4_journal_dirty_metadata");
					ext4_journal_dirty_metadata(handle,
2243 2244 2245 2246 2247 2248 2249
								    parent_bh);
				}
			}
		}
	} else {
		/* We have reached the bottom of the tree. */
		BUFFER_TRACE(parent_bh, "free data blocks");
2250
		ext4_free_data(handle, inode, parent_bh, first, last);
2251 2252 2253 2254
	}
}

/*
2255
 * ext4_truncate()
2256
 *
2257 2258
 * We block out ext4_get_block() block instantiations across the entire
 * transaction, and VFS/VM ensures that ext4_truncate() cannot run
2259 2260 2261 2262 2263 2264 2265 2266 2267 2268 2269 2270 2271 2272 2273 2274
 * 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
2275
 * i_disksize in this case).  After a crash, ext4_orphan_cleanup() will see
2276
 * that this inode's truncate did not complete and it will again call
2277 2278
 * 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
2279
 * that's fine - as long as they are linked from the inode, the post-crash
2280
 * ext4_truncate() run will find them and release them.
2281
 */
2282
void ext4_truncate(struct inode *inode)
2283 2284
{
	handle_t *handle;
2285
	struct ext4_inode_info *ei = EXT4_I(inode);
2286
	__le32 *i_data = ei->i_data;
2287
	int addr_per_block = EXT4_ADDR_PER_BLOCK(inode->i_sb);
2288
	struct address_space *mapping = inode->i_mapping;
A
Aneesh Kumar K.V 已提交
2289
	ext4_lblk_t offsets[4];
2290 2291 2292 2293
	Indirect chain[4];
	Indirect *partial;
	__le32 nr = 0;
	int n;
A
Aneesh Kumar K.V 已提交
2294
	ext4_lblk_t last_block;
2295 2296 2297 2298 2299 2300
	unsigned blocksize = inode->i_sb->s_blocksize;
	struct page *page;

	if (!(S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
	    S_ISLNK(inode->i_mode)))
		return;
2301
	if (ext4_inode_is_fast_symlink(inode))
2302 2303 2304 2305 2306 2307
		return;
	if (IS_APPEND(inode) || IS_IMMUTABLE(inode))
		return;

	/*
	 * We have to lock the EOF page here, because lock_page() nests
2308
	 * outside jbd2_journal_start().
2309 2310 2311 2312 2313 2314 2315 2316 2317 2318 2319
	 */
	if ((inode->i_size & (blocksize - 1)) == 0) {
		/* Block boundary? Nothing to do */
		page = NULL;
	} else {
		page = grab_cache_page(mapping,
				inode->i_size >> PAGE_CACHE_SHIFT);
		if (!page)
			return;
	}

A
Aneesh Kumar K.V 已提交
2320 2321 2322 2323
	if (EXT4_I(inode)->i_flags & EXT4_EXTENTS_FL) {
		ext4_ext_truncate(inode, page);
		return;
	}
A
Alex Tomas 已提交
2324

2325 2326 2327 2328 2329 2330 2331 2332 2333 2334 2335 2336
	handle = start_transaction(inode);
	if (IS_ERR(handle)) {
		if (page) {
			clear_highpage(page);
			flush_dcache_page(page);
			unlock_page(page);
			page_cache_release(page);
		}
		return;		/* AKPM: return what? */
	}

	last_block = (inode->i_size + blocksize-1)
2337
					>> EXT4_BLOCK_SIZE_BITS(inode->i_sb);
2338 2339

	if (page)
2340
		ext4_block_truncate_page(handle, page, mapping, inode->i_size);
2341

2342
	n = ext4_block_to_path(inode, last_block, offsets, NULL);
2343 2344 2345 2346 2347 2348 2349 2350 2351 2352 2353 2354
	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.
	 */
2355
	if (ext4_orphan_add(handle, inode))
2356 2357 2358 2359 2360 2361 2362
		goto out_stop;

	/*
	 * 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
2363
	 * ext4 *really* writes onto the disk inode.
2364 2365 2366 2367
	 */
	ei->i_disksize = inode->i_size;

	/*
2368
	 * From here we block out all ext4_get_block() callers who want to
2369 2370
	 * modify the block allocation tree.
	 */
2371
	down_write(&ei->i_data_sem);
2372 2373

	if (n == 1) {		/* direct blocks */
2374 2375
		ext4_free_data(handle, inode, NULL, i_data+offsets[0],
			       i_data + EXT4_NDIR_BLOCKS);
2376 2377 2378
		goto do_indirects;
	}

2379
	partial = ext4_find_shared(inode, n, offsets, chain, &nr);
2380 2381 2382 2383
	/* Kill the top of shared branch (not detached) */
	if (nr) {
		if (partial == chain) {
			/* Shared branch grows from the inode */
2384
			ext4_free_branches(handle, inode, NULL,
2385 2386 2387 2388 2389 2390 2391 2392 2393
					   &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");
2394
			ext4_free_branches(handle, inode, partial->bh,
2395 2396 2397 2398 2399 2400
					partial->p,
					partial->p+1, (chain+n-1) - partial);
		}
	}
	/* Clear the ends of indirect blocks on the shared branch */
	while (partial > chain) {
2401
		ext4_free_branches(handle, inode, partial->bh, partial->p + 1,
2402 2403 2404 2405 2406 2407 2408 2409 2410 2411
				   (__le32*)partial->bh->b_data+addr_per_block,
				   (chain+n-1) - partial);
		BUFFER_TRACE(partial->bh, "call brelse");
		brelse (partial->bh);
		partial--;
	}
do_indirects:
	/* Kill the remaining (whole) subtrees */
	switch (offsets[0]) {
	default:
2412
		nr = i_data[EXT4_IND_BLOCK];
2413
		if (nr) {
2414 2415
			ext4_free_branches(handle, inode, NULL, &nr, &nr+1, 1);
			i_data[EXT4_IND_BLOCK] = 0;
2416
		}
2417 2418
	case EXT4_IND_BLOCK:
		nr = i_data[EXT4_DIND_BLOCK];
2419
		if (nr) {
2420 2421
			ext4_free_branches(handle, inode, NULL, &nr, &nr+1, 2);
			i_data[EXT4_DIND_BLOCK] = 0;
2422
		}
2423 2424
	case EXT4_DIND_BLOCK:
		nr = i_data[EXT4_TIND_BLOCK];
2425
		if (nr) {
2426 2427
			ext4_free_branches(handle, inode, NULL, &nr, &nr+1, 3);
			i_data[EXT4_TIND_BLOCK] = 0;
2428
		}
2429
	case EXT4_TIND_BLOCK:
2430 2431 2432
		;
	}

2433
	ext4_discard_reservation(inode);
2434

2435
	up_write(&ei->i_data_sem);
K
Kalpak Shah 已提交
2436
	inode->i_mtime = inode->i_ctime = ext4_current_time(inode);
2437
	ext4_mark_inode_dirty(handle, inode);
2438 2439 2440 2441 2442 2443 2444 2445 2446 2447 2448 2449

	/*
	 * In a multi-transaction truncate, we only make the final transaction
	 * synchronous
	 */
	if (IS_SYNC(inode))
		handle->h_sync = 1;
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
2450
	 * ext4_delete_inode(), and we allow that function to clean up the
2451 2452 2453
	 * orphan info for us.
	 */
	if (inode->i_nlink)
2454
		ext4_orphan_del(handle, inode);
2455

2456
	ext4_journal_stop(handle);
2457 2458
}

2459 2460
static ext4_fsblk_t ext4_get_inode_block(struct super_block *sb,
		unsigned long ino, struct ext4_iloc *iloc)
2461
{
2462 2463
	unsigned long desc, group_desc;
	ext4_group_t block_group;
2464
	unsigned long offset;
2465
	ext4_fsblk_t block;
2466
	struct buffer_head *bh;
2467
	struct ext4_group_desc * gdp;
2468

2469
	if (!ext4_valid_inum(sb, ino)) {
2470 2471 2472 2473 2474 2475 2476 2477
		/*
		 * This error is already checked for in namei.c unless we are
		 * looking at an NFS filehandle, in which case no error
		 * report is needed
		 */
		return 0;
	}

2478 2479 2480
	block_group = (ino - 1) / EXT4_INODES_PER_GROUP(sb);
	if (block_group >= EXT4_SB(sb)->s_groups_count) {
		ext4_error(sb,"ext4_get_inode_block","group >= groups count");
2481 2482 2483
		return 0;
	}
	smp_rmb();
2484 2485 2486
	group_desc = block_group >> EXT4_DESC_PER_BLOCK_BITS(sb);
	desc = block_group & (EXT4_DESC_PER_BLOCK(sb) - 1);
	bh = EXT4_SB(sb)->s_group_desc[group_desc];
2487
	if (!bh) {
2488
		ext4_error (sb, "ext4_get_inode_block",
2489 2490 2491 2492
			    "Descriptor not loaded");
		return 0;
	}

2493 2494
	gdp = (struct ext4_group_desc *)((__u8 *)bh->b_data +
		desc * EXT4_DESC_SIZE(sb));
2495 2496 2497
	/*
	 * Figure out the offset within the block group inode table
	 */
2498 2499
	offset = ((ino - 1) % EXT4_INODES_PER_GROUP(sb)) *
		EXT4_INODE_SIZE(sb);
2500 2501
	block = ext4_inode_table(sb, gdp) +
		(offset >> EXT4_BLOCK_SIZE_BITS(sb));
2502 2503

	iloc->block_group = block_group;
2504
	iloc->offset = offset & (EXT4_BLOCK_SIZE(sb) - 1);
2505 2506 2507 2508
	return block;
}

/*
2509
 * ext4_get_inode_loc returns with an extra refcount against the inode's
2510 2511 2512 2513
 * 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.
 */
2514 2515
static int __ext4_get_inode_loc(struct inode *inode,
				struct ext4_iloc *iloc, int in_mem)
2516
{
2517
	ext4_fsblk_t block;
2518 2519
	struct buffer_head *bh;

2520
	block = ext4_get_inode_block(inode->i_sb, inode->i_ino, iloc);
2521 2522 2523 2524 2525
	if (!block)
		return -EIO;

	bh = sb_getblk(inode->i_sb, block);
	if (!bh) {
2526
		ext4_error (inode->i_sb, "ext4_get_inode_loc",
2527
				"unable to read inode block - "
2528
				"inode=%lu, block=%llu",
2529 2530 2531 2532 2533 2534 2535 2536 2537 2538 2539 2540 2541 2542 2543 2544 2545 2546
				 inode->i_ino, block);
		return -EIO;
	}
	if (!buffer_uptodate(bh)) {
		lock_buffer(bh);
		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;
2547
			struct ext4_group_desc *desc;
2548 2549
			int inodes_per_buffer;
			int inode_offset, i;
2550
			ext4_group_t block_group;
2551 2552 2553
			int start;

			block_group = (inode->i_ino - 1) /
2554
					EXT4_INODES_PER_GROUP(inode->i_sb);
2555
			inodes_per_buffer = bh->b_size /
2556
				EXT4_INODE_SIZE(inode->i_sb);
2557
			inode_offset = ((inode->i_ino - 1) %
2558
					EXT4_INODES_PER_GROUP(inode->i_sb));
2559 2560 2561
			start = inode_offset & ~(inodes_per_buffer - 1);

			/* Is the inode bitmap in cache? */
2562
			desc = ext4_get_group_desc(inode->i_sb,
2563 2564 2565 2566 2567
						block_group, NULL);
			if (!desc)
				goto make_io;

			bitmap_bh = sb_getblk(inode->i_sb,
2568
				ext4_inode_bitmap(inode->i_sb, desc));
2569 2570 2571 2572 2573 2574 2575 2576 2577 2578 2579 2580 2581 2582 2583
			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;
			}
			for (i = start; i < start + inodes_per_buffer; i++) {
				if (i == inode_offset)
					continue;
2584
				if (ext4_test_bit(i, bitmap_bh->b_data))
2585 2586 2587 2588 2589 2590 2591 2592 2593 2594 2595 2596 2597 2598 2599 2600 2601 2602 2603 2604 2605 2606 2607
					break;
			}
			brelse(bitmap_bh);
			if (i == start + inodes_per_buffer) {
				/* 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:
		/*
		 * 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)) {
2608
			ext4_error(inode->i_sb, "ext4_get_inode_loc",
2609
					"unable to read inode block - "
2610
					"inode=%lu, block=%llu",
2611 2612 2613 2614 2615 2616 2617 2618 2619 2620
					inode->i_ino, block);
			brelse(bh);
			return -EIO;
		}
	}
has_buffer:
	iloc->bh = bh;
	return 0;
}

2621
int ext4_get_inode_loc(struct inode *inode, struct ext4_iloc *iloc)
2622 2623
{
	/* We have all inode data except xattrs in memory here. */
2624 2625
	return __ext4_get_inode_loc(inode, iloc,
		!(EXT4_I(inode)->i_state & EXT4_STATE_XATTR));
2626 2627
}

2628
void ext4_set_inode_flags(struct inode *inode)
2629
{
2630
	unsigned int flags = EXT4_I(inode)->i_flags;
2631 2632

	inode->i_flags &= ~(S_SYNC|S_APPEND|S_IMMUTABLE|S_NOATIME|S_DIRSYNC);
2633
	if (flags & EXT4_SYNC_FL)
2634
		inode->i_flags |= S_SYNC;
2635
	if (flags & EXT4_APPEND_FL)
2636
		inode->i_flags |= S_APPEND;
2637
	if (flags & EXT4_IMMUTABLE_FL)
2638
		inode->i_flags |= S_IMMUTABLE;
2639
	if (flags & EXT4_NOATIME_FL)
2640
		inode->i_flags |= S_NOATIME;
2641
	if (flags & EXT4_DIRSYNC_FL)
2642 2643 2644
		inode->i_flags |= S_DIRSYNC;
}

2645 2646 2647 2648 2649 2650 2651 2652 2653 2654 2655 2656 2657 2658 2659 2660 2661 2662
/* 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;
}
2663 2664 2665 2666
static blkcnt_t ext4_inode_blocks(struct ext4_inode *raw_inode,
					struct ext4_inode_info *ei)
{
	blkcnt_t i_blocks ;
A
Aneesh Kumar K.V 已提交
2667 2668
	struct inode *inode = &(ei->vfs_inode);
	struct super_block *sb = inode->i_sb;
2669 2670 2671 2672 2673 2674

	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
2675 2676 2677 2678 2679 2680
		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;
		}
2681 2682 2683 2684
	} else {
		return le32_to_cpu(raw_inode->i_blocks_lo);
	}
}
2685

2686
void ext4_read_inode(struct inode * inode)
2687
{
2688 2689 2690
	struct ext4_iloc iloc;
	struct ext4_inode *raw_inode;
	struct ext4_inode_info *ei = EXT4_I(inode);
2691 2692 2693
	struct buffer_head *bh;
	int block;

2694 2695 2696
#ifdef CONFIG_EXT4DEV_FS_POSIX_ACL
	ei->i_acl = EXT4_ACL_NOT_CACHED;
	ei->i_default_acl = EXT4_ACL_NOT_CACHED;
2697 2698 2699
#endif
	ei->i_block_alloc_info = NULL;

2700
	if (__ext4_get_inode_loc(inode, &iloc, 0))
2701 2702
		goto bad_inode;
	bh = iloc.bh;
2703
	raw_inode = ext4_raw_inode(&iloc);
2704 2705 2706 2707 2708 2709 2710 2711 2712 2713 2714 2715 2716 2717 2718 2719 2720 2721 2722
	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);
	if(!(test_opt (inode->i_sb, NO_UID32))) {
		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 ||
2723
		    !(EXT4_SB(inode->i_sb)->s_mount_state & EXT4_ORPHAN_FS)) {
2724 2725 2726 2727 2728 2729 2730 2731 2732 2733
			/* this inode is deleted */
			brelse (bh);
			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);
2734
	inode->i_blocks = ext4_inode_blocks(raw_inode, ei);
2735
	ei->i_file_acl = le32_to_cpu(raw_inode->i_file_acl_lo);
2736
	if (EXT4_SB(inode->i_sb)->s_es->s_creator_os !=
2737
	    cpu_to_le32(EXT4_OS_HURD)) {
B
Badari Pulavarty 已提交
2738 2739
		ei->i_file_acl |=
			((__u64)le16_to_cpu(raw_inode->i_file_acl_high)) << 32;
2740
	}
2741
	inode->i_size = ext4_isize(raw_inode);
2742 2743 2744 2745 2746 2747 2748
	ei->i_disksize = inode->i_size;
	inode->i_generation = le32_to_cpu(raw_inode->i_generation);
	ei->i_block_group = iloc.block_group;
	/*
	 * 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!
	 */
2749
	for (block = 0; block < EXT4_N_BLOCKS; block++)
2750 2751 2752
		ei->i_data[block] = raw_inode->i_block[block];
	INIT_LIST_HEAD(&ei->i_orphan);

2753 2754
	if (inode->i_ino >= EXT4_FIRST_INO(inode->i_sb) + 1 &&
	    EXT4_INODE_SIZE(inode->i_sb) > EXT4_GOOD_OLD_INODE_SIZE) {
2755 2756
		/*
		 * When mke2fs creates big inodes it does not zero out
2757
		 * the unused bytes above EXT4_GOOD_OLD_INODE_SIZE,
2758 2759 2760
		 * so ignore those first few inodes.
		 */
		ei->i_extra_isize = le16_to_cpu(raw_inode->i_extra_isize);
2761
		if (EXT4_GOOD_OLD_INODE_SIZE + ei->i_extra_isize >
2762 2763
		    EXT4_INODE_SIZE(inode->i_sb)) {
			brelse (bh);
2764
			goto bad_inode;
2765
		}
2766 2767
		if (ei->i_extra_isize == 0) {
			/* The extra space is currently unused. Use it. */
2768 2769
			ei->i_extra_isize = sizeof(struct ext4_inode) -
					    EXT4_GOOD_OLD_INODE_SIZE;
2770 2771
		} else {
			__le32 *magic = (void *)raw_inode +
2772
					EXT4_GOOD_OLD_INODE_SIZE +
2773
					ei->i_extra_isize;
2774 2775
			if (*magic == cpu_to_le32(EXT4_XATTR_MAGIC))
				 ei->i_state |= EXT4_STATE_XATTR;
2776 2777 2778 2779
		}
	} else
		ei->i_extra_isize = 0;

K
Kalpak Shah 已提交
2780 2781 2782 2783 2784
	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);

2785 2786 2787 2788 2789 2790 2791
	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;
	}

2792
	if (S_ISREG(inode->i_mode)) {
2793 2794 2795
		inode->i_op = &ext4_file_inode_operations;
		inode->i_fop = &ext4_file_operations;
		ext4_set_aops(inode);
2796
	} else if (S_ISDIR(inode->i_mode)) {
2797 2798
		inode->i_op = &ext4_dir_inode_operations;
		inode->i_fop = &ext4_dir_operations;
2799
	} else if (S_ISLNK(inode->i_mode)) {
2800 2801
		if (ext4_inode_is_fast_symlink(inode))
			inode->i_op = &ext4_fast_symlink_inode_operations;
2802
		else {
2803 2804
			inode->i_op = &ext4_symlink_inode_operations;
			ext4_set_aops(inode);
2805 2806
		}
	} else {
2807
		inode->i_op = &ext4_special_inode_operations;
2808 2809 2810 2811 2812 2813 2814 2815
		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])));
	}
	brelse (iloc.bh);
2816
	ext4_set_inode_flags(inode);
2817 2818 2819 2820 2821 2822 2823
	return;

bad_inode:
	make_bad_inode(inode);
	return;
}

2824 2825 2826 2827 2828 2829 2830 2831 2832 2833 2834 2835 2836 2837
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;
	int err = 0;

	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
2838
		raw_inode->i_blocks_lo   = cpu_to_le32(i_blocks);
2839
		raw_inode->i_blocks_high = 0;
A
Aneesh Kumar K.V 已提交
2840
		ei->i_flags &= ~EXT4_HUGE_FILE_FL;
2841 2842 2843 2844 2845 2846 2847 2848 2849 2850
	} else if (i_blocks <= 0xffffffffffffULL) {
		/*
		 * i_blocks can be represented in a 48 bit variable
		 * as multiple of 512 bytes
		 */
		err = ext4_update_rocompat_feature(handle, sb,
					    EXT4_FEATURE_RO_COMPAT_HUGE_FILE);
		if (err)
			goto  err_out;
		/* i_block is stored in the split  48 bit fields */
A
Aneesh Kumar K.V 已提交
2851
		raw_inode->i_blocks_lo   = cpu_to_le32(i_blocks);
2852
		raw_inode->i_blocks_high = cpu_to_le16(i_blocks >> 32);
A
Aneesh Kumar K.V 已提交
2853
		ei->i_flags &= ~EXT4_HUGE_FILE_FL;
2854
	} else {
A
Aneesh Kumar K.V 已提交
2855 2856 2857 2858 2859 2860 2861 2862 2863 2864 2865 2866 2867
		/*
		 * i_blocks should be represented in a 48 bit variable
		 * as multiple of  file system block size
		 */
		err = ext4_update_rocompat_feature(handle, sb,
					    EXT4_FEATURE_RO_COMPAT_HUGE_FILE);
		if (err)
			goto  err_out;
		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);
2868 2869 2870 2871 2872
	}
err_out:
	return err;
}

2873 2874 2875 2876 2877 2878 2879
/*
 * 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.
 */
2880
static int ext4_do_update_inode(handle_t *handle,
2881
				struct inode *inode,
2882
				struct ext4_iloc *iloc)
2883
{
2884 2885
	struct ext4_inode *raw_inode = ext4_raw_inode(iloc);
	struct ext4_inode_info *ei = EXT4_I(inode);
2886 2887 2888 2889 2890
	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. */
2891 2892
	if (ei->i_state & EXT4_STATE_NEW)
		memset(raw_inode, 0, EXT4_SB(inode->i_sb)->s_inode_size);
2893

2894
	ext4_get_inode_flags(ei);
2895 2896 2897 2898 2899 2900 2901 2902 2903 2904 2905 2906 2907 2908 2909 2910 2911 2912 2913 2914 2915 2916 2917 2918 2919 2920
	raw_inode->i_mode = cpu_to_le16(inode->i_mode);
	if(!(test_opt(inode->i_sb, NO_UID32))) {
		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
 */
		if(!ei->i_dtime) {
			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
2921 2922 2923 2924 2925 2926

	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);

2927 2928
	if (ext4_inode_blocks_set(handle, raw_inode, ei))
		goto out_brelse;
2929 2930
	raw_inode->i_dtime = cpu_to_le32(ei->i_dtime);
	raw_inode->i_flags = cpu_to_le32(ei->i_flags);
2931 2932
	if (EXT4_SB(inode->i_sb)->s_es->s_creator_os !=
	    cpu_to_le32(EXT4_OS_HURD))
B
Badari Pulavarty 已提交
2933 2934
		raw_inode->i_file_acl_high =
			cpu_to_le16(ei->i_file_acl >> 32);
2935
	raw_inode->i_file_acl_lo = cpu_to_le32(ei->i_file_acl);
2936 2937 2938 2939 2940 2941 2942 2943 2944 2945 2946 2947 2948 2949 2950 2951
	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,
2952
					EXT4_FEATURE_RO_COMPAT_LARGE_FILE);
2953 2954 2955 2956
			sb->s_dirt = 1;
			handle->h_sync = 1;
			err = ext4_journal_dirty_metadata(handle,
					EXT4_SB(sb)->s_sbh);
2957 2958 2959 2960 2961 2962 2963 2964 2965 2966 2967 2968 2969 2970
		}
	}
	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;
		}
2971
	} else for (block = 0; block < EXT4_N_BLOCKS; block++)
2972 2973
		raw_inode->i_block[block] = ei->i_data[block];

2974 2975 2976 2977 2978
	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);
2979
		raw_inode->i_extra_isize = cpu_to_le16(ei->i_extra_isize);
2980 2981
	}

2982

2983 2984
	BUFFER_TRACE(bh, "call ext4_journal_dirty_metadata");
	rc = ext4_journal_dirty_metadata(handle, bh);
2985 2986
	if (!err)
		err = rc;
2987
	ei->i_state &= ~EXT4_STATE_NEW;
2988 2989 2990

out_brelse:
	brelse (bh);
2991
	ext4_std_error(inode->i_sb, err);
2992 2993 2994 2995
	return err;
}

/*
2996
 * ext4_write_inode()
2997 2998 2999 3000 3001 3002 3003 3004 3005 3006 3007 3008 3009 3010 3011 3012
 *
 * 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
3013
 * ext4_mark_inode_dirty().  This is a correctness thing for O_SYNC and for
3014 3015 3016 3017 3018 3019 3020 3021 3022 3023 3024 3025 3026 3027 3028 3029
 * 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.
 */
3030
int ext4_write_inode(struct inode *inode, int wait)
3031 3032 3033 3034
{
	if (current->flags & PF_MEMALLOC)
		return 0;

3035
	if (ext4_journal_current_handle()) {
M
Mingming Cao 已提交
3036
		jbd_debug(1, "called recursively, non-PF_MEMALLOC!\n");
3037 3038 3039 3040 3041 3042 3043
		dump_stack();
		return -EIO;
	}

	if (!wait)
		return 0;

3044
	return ext4_force_commit(inode->i_sb);
3045 3046 3047
}

/*
3048
 * ext4_setattr()
3049 3050 3051 3052 3053 3054 3055 3056 3057 3058 3059 3060 3061 3062 3063
 *
 * 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.)
 *
 * Called with inode->sem down.
 */
3064
int ext4_setattr(struct dentry *dentry, struct iattr *attr)
3065 3066 3067 3068 3069 3070 3071 3072 3073 3074 3075 3076 3077 3078 3079
{
	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) */
3080 3081
		handle = ext4_journal_start(inode, 2*(EXT4_QUOTA_INIT_BLOCKS(inode->i_sb)+
					EXT4_QUOTA_DEL_BLOCKS(inode->i_sb))+3);
3082 3083 3084 3085 3086 3087
		if (IS_ERR(handle)) {
			error = PTR_ERR(handle);
			goto err_out;
		}
		error = DQUOT_TRANSFER(inode, attr) ? -EDQUOT : 0;
		if (error) {
3088
			ext4_journal_stop(handle);
3089 3090 3091 3092 3093 3094 3095 3096
			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;
3097 3098
		error = ext4_mark_inode_dirty(handle, inode);
		ext4_journal_stop(handle);
3099 3100
	}

3101 3102 3103 3104 3105 3106 3107 3108 3109 3110 3111
	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;
			}
		}
	}

3112 3113 3114 3115
	if (S_ISREG(inode->i_mode) &&
	    attr->ia_valid & ATTR_SIZE && attr->ia_size < inode->i_size) {
		handle_t *handle;

3116
		handle = ext4_journal_start(inode, 3);
3117 3118 3119 3120 3121
		if (IS_ERR(handle)) {
			error = PTR_ERR(handle);
			goto err_out;
		}

3122 3123 3124
		error = ext4_orphan_add(handle, inode);
		EXT4_I(inode)->i_disksize = attr->ia_size;
		rc = ext4_mark_inode_dirty(handle, inode);
3125 3126
		if (!error)
			error = rc;
3127
		ext4_journal_stop(handle);
3128 3129 3130 3131
	}

	rc = inode_setattr(inode, attr);

3132
	/* If inode_setattr's call to ext4_truncate failed to get a
3133 3134 3135
	 * transaction handle at all, we need to clean up the in-core
	 * orphan list manually. */
	if (inode->i_nlink)
3136
		ext4_orphan_del(NULL, inode);
3137 3138

	if (!rc && (ia_valid & ATTR_MODE))
3139
		rc = ext4_acl_chmod(inode);
3140 3141

err_out:
3142
	ext4_std_error(inode->i_sb, error);
3143 3144 3145 3146 3147 3148 3149 3150 3151 3152 3153 3154 3155 3156 3157 3158 3159 3160
	if (!error)
		error = rc;
	return error;
}


/*
 * How many blocks doth make a writepage()?
 *
 * With N blocks per page, it may be:
 * N data blocks
 * 2 indirect block
 * 2 dindirect
 * 1 tindirect
 * N+5 bitmap blocks (from the above)
 * N+5 group descriptor summary blocks
 * 1 inode block
 * 1 superblock.
3161
 * 2 * EXT4_SINGLEDATA_TRANS_BLOCKS for the quote files
3162
 *
3163
 * 3 * (N + 5) + 2 + 2 * EXT4_SINGLEDATA_TRANS_BLOCKS
3164 3165 3166 3167 3168 3169 3170 3171 3172 3173 3174 3175
 *
 * With ordered or writeback data it's the same, less the N data blocks.
 *
 * If the inode's direct blocks can hold an integral number of pages then a
 * page cannot straddle two indirect blocks, and we can only touch one indirect
 * and dindirect block, and the "5" above becomes "3".
 *
 * This still overestimates under most circumstances.  If we were to pass the
 * start and end offsets in here as well we could do block_to_path() on each
 * block and work out the exact number of indirects which are touched.  Pah.
 */

A
Alex Tomas 已提交
3176
int ext4_writepage_trans_blocks(struct inode *inode)
3177
{
3178 3179
	int bpp = ext4_journal_blocks_per_page(inode);
	int indirects = (EXT4_NDIR_BLOCKS % bpp) ? 5 : 3;
3180 3181
	int ret;

A
Alex Tomas 已提交
3182 3183 3184
	if (EXT4_I(inode)->i_flags & EXT4_EXTENTS_FL)
		return ext4_ext_writepage_trans_blocks(inode, bpp);

3185
	if (ext4_should_journal_data(inode))
3186 3187 3188 3189 3190 3191 3192
		ret = 3 * (bpp + indirects) + 2;
	else
		ret = 2 * (bpp + indirects) + 2;

#ifdef CONFIG_QUOTA
	/* We know that structure was already allocated during DQUOT_INIT so
	 * we will be updating only the data blocks + inodes */
3193
	ret += 2*EXT4_QUOTA_TRANS_BLOCKS(inode->i_sb);
3194 3195 3196 3197 3198 3199
#endif

	return ret;
}

/*
3200
 * The caller must have previously called ext4_reserve_inode_write().
3201 3202
 * Give this, we know that the caller already has write access to iloc->bh.
 */
3203 3204
int ext4_mark_iloc_dirty(handle_t *handle,
		struct inode *inode, struct ext4_iloc *iloc)
3205 3206 3207
{
	int err = 0;

3208 3209 3210
	if (test_opt(inode->i_sb, I_VERSION))
		inode_inc_iversion(inode);

3211 3212 3213
	/* the do_update_inode consumes one bh->b_count */
	get_bh(iloc->bh);

3214
	/* ext4_do_update_inode() does jbd2_journal_dirty_metadata */
3215
	err = ext4_do_update_inode(handle, inode, iloc);
3216 3217 3218 3219 3220 3221 3222 3223 3224 3225
	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
3226 3227
ext4_reserve_inode_write(handle_t *handle, struct inode *inode,
			 struct ext4_iloc *iloc)
3228 3229 3230
{
	int err = 0;
	if (handle) {
3231
		err = ext4_get_inode_loc(inode, iloc);
3232 3233
		if (!err) {
			BUFFER_TRACE(iloc->bh, "get_write_access");
3234
			err = ext4_journal_get_write_access(handle, iloc->bh);
3235 3236 3237 3238 3239 3240
			if (err) {
				brelse(iloc->bh);
				iloc->bh = NULL;
			}
		}
	}
3241
	ext4_std_error(inode->i_sb, err);
3242 3243 3244
	return err;
}

3245 3246 3247 3248
/*
 * 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
3249 3250 3251 3252
static int ext4_expand_extra_isize(struct inode *inode,
				   unsigned int new_extra_isize,
				   struct ext4_iloc iloc,
				   handle_t *handle)
3253 3254 3255 3256 3257 3258 3259 3260 3261 3262 3263 3264 3265 3266 3267 3268 3269 3270 3271 3272 3273 3274 3275 3276 3277 3278 3279
{
	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);
}

3280 3281 3282 3283 3284 3285 3286 3287 3288 3289 3290 3291 3292 3293 3294 3295 3296 3297 3298 3299 3300
/*
 * 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.
 */
3301
int ext4_mark_inode_dirty(handle_t *handle, struct inode *inode)
3302
{
3303
	struct ext4_iloc iloc;
3304 3305 3306
	struct ext4_sb_info *sbi = EXT4_SB(inode->i_sb);
	static unsigned int mnt_count;
	int err, ret;
3307 3308

	might_sleep();
3309
	err = ext4_reserve_inode_write(handle, inode, &iloc);
3310 3311 3312 3313 3314 3315 3316 3317 3318 3319 3320 3321 3322 3323 3324 3325
	if (EXT4_I(inode)->i_extra_isize < sbi->s_want_extra_isize &&
	    !(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
3326 3327
				if (mnt_count !=
					le16_to_cpu(sbi->s_es->s_mnt_count)) {
3328 3329 3330 3331
					ext4_warning(inode->i_sb, __FUNCTION__,
					"Unable to expand inode %lu. Delete"
					" some EAs or run e2fsck.",
					inode->i_ino);
A
Aneesh Kumar K.V 已提交
3332 3333
					mnt_count =
					  le16_to_cpu(sbi->s_es->s_mnt_count);
3334 3335 3336 3337
				}
			}
		}
	}
3338
	if (!err)
3339
		err = ext4_mark_iloc_dirty(handle, inode, &iloc);
3340 3341 3342 3343
	return err;
}

/*
3344
 * ext4_dirty_inode() is called from __mark_inode_dirty()
3345 3346 3347 3348 3349 3350 3351 3352 3353 3354 3355 3356
 *
 * 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.
 *
 * Also, DQUOT_ALLOC_SPACE() will always dirty the inode when blocks
 * 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.
 */
3357
void ext4_dirty_inode(struct inode *inode)
3358
{
3359
	handle_t *current_handle = ext4_journal_current_handle();
3360 3361
	handle_t *handle;

3362
	handle = ext4_journal_start(inode, 2);
3363 3364 3365 3366 3367 3368 3369 3370 3371 3372
	if (IS_ERR(handle))
		goto out;
	if (current_handle &&
		current_handle->h_transaction != handle->h_transaction) {
		/* This task has a transaction open against a different fs */
		printk(KERN_EMERG "%s: transactions do not match!\n",
		       __FUNCTION__);
	} else {
		jbd_debug(5, "marking dirty.  outer handle=%p\n",
				current_handle);
3373
		ext4_mark_inode_dirty(handle, inode);
3374
	}
3375
	ext4_journal_stop(handle);
3376 3377 3378 3379 3380 3381 3382 3383
out:
	return;
}

#if 0
/*
 * Bind an inode's backing buffer_head into this transaction, to prevent
 * it from being flushed to disk early.  Unlike
3384
 * ext4_reserve_inode_write, this leaves behind no bh reference and
3385 3386 3387
 * returns no iloc structure, so the caller needs to repeat the iloc
 * lookup to mark the inode dirty later.
 */
3388
static int ext4_pin_inode(handle_t *handle, struct inode *inode)
3389
{
3390
	struct ext4_iloc iloc;
3391 3392 3393

	int err = 0;
	if (handle) {
3394
		err = ext4_get_inode_loc(inode, &iloc);
3395 3396
		if (!err) {
			BUFFER_TRACE(iloc.bh, "get_write_access");
3397
			err = jbd2_journal_get_write_access(handle, iloc.bh);
3398
			if (!err)
3399
				err = ext4_journal_dirty_metadata(handle,
3400 3401 3402 3403
								  iloc.bh);
			brelse(iloc.bh);
		}
	}
3404
	ext4_std_error(inode->i_sb, err);
3405 3406 3407 3408
	return err;
}
#endif

3409
int ext4_change_inode_journal_flag(struct inode *inode, int val)
3410 3411 3412 3413 3414 3415 3416 3417 3418 3419 3420 3421 3422 3423 3424
{
	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.
	 */

3425
	journal = EXT4_JOURNAL(inode);
3426
	if (is_journal_aborted(journal))
3427 3428
		return -EROFS;

3429 3430
	jbd2_journal_lock_updates(journal);
	jbd2_journal_flush(journal);
3431 3432 3433 3434 3435 3436 3437 3438 3439 3440

	/*
	 * 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)
3441
		EXT4_I(inode)->i_flags |= EXT4_JOURNAL_DATA_FL;
3442
	else
3443 3444
		EXT4_I(inode)->i_flags &= ~EXT4_JOURNAL_DATA_FL;
	ext4_set_aops(inode);
3445

3446
	jbd2_journal_unlock_updates(journal);
3447 3448 3449

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

3450
	handle = ext4_journal_start(inode, 1);
3451 3452 3453
	if (IS_ERR(handle))
		return PTR_ERR(handle);

3454
	err = ext4_mark_inode_dirty(handle, inode);
3455
	handle->h_sync = 1;
3456 3457
	ext4_journal_stop(handle);
	ext4_std_error(inode->i_sb, err);
3458 3459 3460

	return err;
}