extents.c 130.2 KB
Newer Older
A
Alex Tomas 已提交
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34
/*
 * Copyright (c) 2003-2006, Cluster File Systems, Inc, info@clusterfs.com
 * Written by Alex Tomas <alex@clusterfs.com>
 *
 * Architecture independence:
 *   Copyright (c) 2005, Bull S.A.
 *   Written by Pierre Peiffer <pierre.peiffer@bull.net>
 *
 * This program is free software; you can redistribute it and/or modify
 * it under the terms of the GNU General Public License version 2 as
 * published by the Free Software Foundation.
 *
 * This program is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 * GNU General Public License for more details.
 *
 * You should have received a copy of the GNU General Public Licens
 * along with this program; if not, write to the Free Software
 * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-
 */

/*
 * Extents support for EXT4
 *
 * TODO:
 *   - ext4*_error() should be used in some situations
 *   - analyze all BUG()/BUG_ON(), use -EIO where appropriate
 *   - smart tree reduction
 */

#include <linux/module.h>
#include <linux/fs.h>
#include <linux/time.h>
35
#include <linux/jbd2.h>
A
Alex Tomas 已提交
36 37 38 39 40
#include <linux/highuid.h>
#include <linux/pagemap.h>
#include <linux/quotaops.h>
#include <linux/string.h>
#include <linux/slab.h>
A
Amit Arora 已提交
41
#include <linux/falloc.h>
A
Alex Tomas 已提交
42
#include <asm/uaccess.h>
43
#include <linux/fiemap.h>
44
#include "ext4_jbd2.h"
A
Alex Tomas 已提交
45

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

A
Allison Henderson 已提交
48 49 50 51 52 53 54
static int ext4_split_extent(handle_t *handle,
				struct inode *inode,
				struct ext4_ext_path *path,
				struct ext4_map_blocks *map,
				int split_flag,
				int flags);

55 56 57
static int ext4_ext_truncate_extend_restart(handle_t *handle,
					    struct inode *inode,
					    int needed)
A
Alex Tomas 已提交
58 59 60
{
	int err;

61 62
	if (!ext4_handle_valid(handle))
		return 0;
A
Alex Tomas 已提交
63
	if (handle->h_buffer_credits > needed)
64 65
		return 0;
	err = ext4_journal_extend(handle, needed);
66
	if (err <= 0)
67
		return err;
68
	err = ext4_truncate_restart_trans(handle, inode, needed);
69 70
	if (err == 0)
		err = -EAGAIN;
71 72

	return err;
A
Alex Tomas 已提交
73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97
}

/*
 * could return:
 *  - EROFS
 *  - ENOMEM
 */
static int ext4_ext_get_access(handle_t *handle, struct inode *inode,
				struct ext4_ext_path *path)
{
	if (path->p_bh) {
		/* path points to block */
		return ext4_journal_get_write_access(handle, path->p_bh);
	}
	/* path points to leaf/index in inode body */
	/* we use in-core data, no need to protect them */
	return 0;
}

/*
 * could return:
 *  - EROFS
 *  - ENOMEM
 *  - EIO
 */
98 99 100 101 102
#define ext4_ext_dirty(handle, inode, path) \
		__ext4_ext_dirty(__func__, __LINE__, (handle), (inode), (path))
static int __ext4_ext_dirty(const char *where, unsigned int line,
			    handle_t *handle, struct inode *inode,
			    struct ext4_ext_path *path)
A
Alex Tomas 已提交
103 104 105 106
{
	int err;
	if (path->p_bh) {
		/* path points to block */
107 108
		err = __ext4_handle_dirty_metadata(where, line, handle,
						   inode, path->p_bh);
A
Alex Tomas 已提交
109 110 111 112 113 114 115
	} else {
		/* path points to leaf/index in inode body */
		err = ext4_mark_inode_dirty(handle, inode);
	}
	return err;
}

116
static ext4_fsblk_t ext4_ext_find_goal(struct inode *inode,
A
Alex Tomas 已提交
117
			      struct ext4_ext_path *path,
A
Aneesh Kumar K.V 已提交
118
			      ext4_lblk_t block)
A
Alex Tomas 已提交
119 120 121 122 123 124 125
{
	int depth;

	if (path) {
		struct ext4_extent *ex;
		depth = path->p_depth;

126 127 128 129 130 131 132 133 134 135 136 137 138 139
		/*
		 * Try to predict block placement assuming that we are
		 * filling in a file which will eventually be
		 * non-sparse --- i.e., in the case of libbfd writing
		 * an ELF object sections out-of-order but in a way
		 * the eventually results in a contiguous object or
		 * executable file, or some database extending a table
		 * space file.  However, this is actually somewhat
		 * non-ideal if we are writing a sparse file such as
		 * qemu or KVM writing a raw image file that is going
		 * to stay fairly sparse, since it will end up
		 * fragmenting the file system's free space.  Maybe we
		 * should have some hueristics or some way to allow
		 * userspace to pass a hint to file system,
T
Tao Ma 已提交
140
		 * especially if the latter case turns out to be
141 142
		 * common.
		 */
143
		ex = path[depth].p_ext;
144 145 146 147 148 149 150 151 152
		if (ex) {
			ext4_fsblk_t ext_pblk = ext4_ext_pblock(ex);
			ext4_lblk_t ext_block = le32_to_cpu(ex->ee_block);

			if (block > ext_block)
				return ext_pblk + (block - ext_block);
			else
				return ext_pblk - (ext_block - block);
		}
A
Alex Tomas 已提交
153

154 155
		/* it looks like index is empty;
		 * try to find starting block from index itself */
A
Alex Tomas 已提交
156 157 158 159 160
		if (path[depth].p_bh)
			return path[depth].p_bh->b_blocknr;
	}

	/* OK. use inode's group */
161
	return ext4_inode_to_goal_block(inode);
A
Alex Tomas 已提交
162 163
}

A
Aneesh Kumar K.V 已提交
164 165 166
/*
 * Allocation for a meta data block
 */
167
static ext4_fsblk_t
A
Aneesh Kumar K.V 已提交
168
ext4_ext_new_meta_block(handle_t *handle, struct inode *inode,
A
Alex Tomas 已提交
169
			struct ext4_ext_path *path,
170
			struct ext4_extent *ex, int *err, unsigned int flags)
A
Alex Tomas 已提交
171
{
172
	ext4_fsblk_t goal, newblock;
A
Alex Tomas 已提交
173 174

	goal = ext4_ext_find_goal(inode, path, le32_to_cpu(ex->ee_block));
175 176
	newblock = ext4_new_meta_blocks(handle, inode, goal, flags,
					NULL, err);
A
Alex Tomas 已提交
177 178 179
	return newblock;
}

180
static inline int ext4_ext_space_block(struct inode *inode, int check)
A
Alex Tomas 已提交
181 182 183 184 185
{
	int size;

	size = (inode->i_sb->s_blocksize - sizeof(struct ext4_extent_header))
			/ sizeof(struct ext4_extent);
186
	if (!check) {
187
#ifdef AGGRESSIVE_TEST
188 189
		if (size > 6)
			size = 6;
A
Alex Tomas 已提交
190
#endif
191
	}
A
Alex Tomas 已提交
192 193 194
	return size;
}

195
static inline int ext4_ext_space_block_idx(struct inode *inode, int check)
A
Alex Tomas 已提交
196 197 198 199 200
{
	int size;

	size = (inode->i_sb->s_blocksize - sizeof(struct ext4_extent_header))
			/ sizeof(struct ext4_extent_idx);
201
	if (!check) {
202
#ifdef AGGRESSIVE_TEST
203 204
		if (size > 5)
			size = 5;
A
Alex Tomas 已提交
205
#endif
206
	}
A
Alex Tomas 已提交
207 208 209
	return size;
}

210
static inline int ext4_ext_space_root(struct inode *inode, int check)
A
Alex Tomas 已提交
211 212 213 214 215 216
{
	int size;

	size = sizeof(EXT4_I(inode)->i_data);
	size -= sizeof(struct ext4_extent_header);
	size /= sizeof(struct ext4_extent);
217
	if (!check) {
218
#ifdef AGGRESSIVE_TEST
219 220
		if (size > 3)
			size = 3;
A
Alex Tomas 已提交
221
#endif
222
	}
A
Alex Tomas 已提交
223 224 225
	return size;
}

226
static inline int ext4_ext_space_root_idx(struct inode *inode, int check)
A
Alex Tomas 已提交
227 228 229 230 231 232
{
	int size;

	size = sizeof(EXT4_I(inode)->i_data);
	size -= sizeof(struct ext4_extent_header);
	size /= sizeof(struct ext4_extent_idx);
233
	if (!check) {
234
#ifdef AGGRESSIVE_TEST
235 236
		if (size > 4)
			size = 4;
A
Alex Tomas 已提交
237
#endif
238
	}
A
Alex Tomas 已提交
239 240 241
	return size;
}

242 243 244 245 246
/*
 * Calculate the number of metadata blocks needed
 * to allocate @blocks
 * Worse case is one block per extent
 */
247
int ext4_ext_calc_metadata_amount(struct inode *inode, ext4_lblk_t lblock)
248
{
249 250
	struct ext4_inode_info *ei = EXT4_I(inode);
	int idxs, num = 0;
251

252 253
	idxs = ((inode->i_sb->s_blocksize - sizeof(struct ext4_extent_header))
		/ sizeof(struct ext4_extent_idx));
254 255

	/*
256 257 258 259 260 261
	 * If the new delayed allocation block is contiguous with the
	 * previous da block, it can share index blocks with the
	 * previous block, so we only need to allocate a new index
	 * block every idxs leaf blocks.  At ldxs**2 blocks, we need
	 * an additional index block, and at ldxs**3 blocks, yet
	 * another index blocks.
262
	 */
263 264 265 266 267 268 269 270 271 272 273 274 275 276
	if (ei->i_da_metadata_calc_len &&
	    ei->i_da_metadata_calc_last_lblock+1 == lblock) {
		if ((ei->i_da_metadata_calc_len % idxs) == 0)
			num++;
		if ((ei->i_da_metadata_calc_len % (idxs*idxs)) == 0)
			num++;
		if ((ei->i_da_metadata_calc_len % (idxs*idxs*idxs)) == 0) {
			num++;
			ei->i_da_metadata_calc_len = 0;
		} else
			ei->i_da_metadata_calc_len++;
		ei->i_da_metadata_calc_last_lblock++;
		return num;
	}
277

278 279 280 281 282 283 284
	/*
	 * In the worst case we need a new set of index blocks at
	 * every level of the inode's extent tree.
	 */
	ei->i_da_metadata_calc_len = 1;
	ei->i_da_metadata_calc_last_lblock = lblock;
	return ext_depth(inode) + 1;
285 286
}

287 288 289 290 291 292 293
static int
ext4_ext_max_entries(struct inode *inode, int depth)
{
	int max;

	if (depth == ext_depth(inode)) {
		if (depth == 0)
294
			max = ext4_ext_space_root(inode, 1);
295
		else
296
			max = ext4_ext_space_root_idx(inode, 1);
297 298
	} else {
		if (depth == 0)
299
			max = ext4_ext_space_block(inode, 1);
300
		else
301
			max = ext4_ext_space_block_idx(inode, 1);
302 303 304 305 306
	}

	return max;
}

307 308
static int ext4_valid_extent(struct inode *inode, struct ext4_extent *ext)
{
309
	ext4_fsblk_t block = ext4_ext_pblock(ext);
310
	int len = ext4_ext_get_actual_len(ext);
311

312
	return ext4_data_block_valid(EXT4_SB(inode->i_sb), block, len);
313 314 315 316 317
}

static int ext4_valid_extent_idx(struct inode *inode,
				struct ext4_extent_idx *ext_idx)
{
318
	ext4_fsblk_t block = ext4_idx_pblock(ext_idx);
319

320
	return ext4_data_block_valid(EXT4_SB(inode->i_sb), block, 1);
321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355
}

static int ext4_valid_extent_entries(struct inode *inode,
				struct ext4_extent_header *eh,
				int depth)
{
	struct ext4_extent *ext;
	struct ext4_extent_idx *ext_idx;
	unsigned short entries;
	if (eh->eh_entries == 0)
		return 1;

	entries = le16_to_cpu(eh->eh_entries);

	if (depth == 0) {
		/* leaf entries */
		ext = EXT_FIRST_EXTENT(eh);
		while (entries) {
			if (!ext4_valid_extent(inode, ext))
				return 0;
			ext++;
			entries--;
		}
	} else {
		ext_idx = EXT_FIRST_INDEX(eh);
		while (entries) {
			if (!ext4_valid_extent_idx(inode, ext_idx))
				return 0;
			ext_idx++;
			entries--;
		}
	}
	return 1;
}

356 357 358
static int __ext4_ext_check(const char *function, unsigned int line,
			    struct inode *inode, struct ext4_extent_header *eh,
			    int depth)
359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383
{
	const char *error_msg;
	int max = 0;

	if (unlikely(eh->eh_magic != EXT4_EXT_MAGIC)) {
		error_msg = "invalid magic";
		goto corrupted;
	}
	if (unlikely(le16_to_cpu(eh->eh_depth) != depth)) {
		error_msg = "unexpected eh_depth";
		goto corrupted;
	}
	if (unlikely(eh->eh_max == 0)) {
		error_msg = "invalid eh_max";
		goto corrupted;
	}
	max = ext4_ext_max_entries(inode, depth);
	if (unlikely(le16_to_cpu(eh->eh_max) > max)) {
		error_msg = "too large eh_max";
		goto corrupted;
	}
	if (unlikely(le16_to_cpu(eh->eh_entries) > le16_to_cpu(eh->eh_max))) {
		error_msg = "invalid eh_entries";
		goto corrupted;
	}
384 385 386 387
	if (!ext4_valid_extent_entries(inode, eh, depth)) {
		error_msg = "invalid extent entries";
		goto corrupted;
	}
388 389 390
	return 0;

corrupted:
391
	ext4_error_inode(inode, function, line, 0,
392
			"bad header/extent: %s - magic %x, "
393
			"entries %u, max %u(%u), depth %u(%u)",
394
			error_msg, le16_to_cpu(eh->eh_magic),
395 396 397 398 399 400
			le16_to_cpu(eh->eh_entries), le16_to_cpu(eh->eh_max),
			max, le16_to_cpu(eh->eh_depth), depth);

	return -EIO;
}

401
#define ext4_ext_check(inode, eh, depth)	\
402
	__ext4_ext_check(__func__, __LINE__, inode, eh, depth)
403

404 405 406 407 408
int ext4_ext_check_inode(struct inode *inode)
{
	return ext4_ext_check(inode, ext_inode_hdr(inode), ext_depth(inode));
}

A
Alex Tomas 已提交
409 410 411 412 413 414 415 416
#ifdef EXT_DEBUG
static void ext4_ext_show_path(struct inode *inode, struct ext4_ext_path *path)
{
	int k, l = path->p_depth;

	ext_debug("path:");
	for (k = 0; k <= l; k++, path++) {
		if (path->p_idx) {
417
		  ext_debug("  %d->%llu", le32_to_cpu(path->p_idx->ei_block),
418
			    ext4_idx_pblock(path->p_idx));
A
Alex Tomas 已提交
419
		} else if (path->p_ext) {
420
			ext_debug("  %d:[%d]%d:%llu ",
A
Alex Tomas 已提交
421
				  le32_to_cpu(path->p_ext->ee_block),
422
				  ext4_ext_is_uninitialized(path->p_ext),
A
Amit Arora 已提交
423
				  ext4_ext_get_actual_len(path->p_ext),
424
				  ext4_ext_pblock(path->p_ext));
A
Alex Tomas 已提交
425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443
		} else
			ext_debug("  []");
	}
	ext_debug("\n");
}

static void ext4_ext_show_leaf(struct inode *inode, struct ext4_ext_path *path)
{
	int depth = ext_depth(inode);
	struct ext4_extent_header *eh;
	struct ext4_extent *ex;
	int i;

	if (!path)
		return;

	eh = path[depth].p_hdr;
	ex = EXT_FIRST_EXTENT(eh);

444 445
	ext_debug("Displaying leaf extents for inode %lu\n", inode->i_ino);

A
Alex Tomas 已提交
446
	for (i = 0; i < le16_to_cpu(eh->eh_entries); i++, ex++) {
447 448
		ext_debug("%d:[%d]%d:%llu ", le32_to_cpu(ex->ee_block),
			  ext4_ext_is_uninitialized(ex),
449
			  ext4_ext_get_actual_len(ex), ext4_ext_pblock(ex));
A
Alex Tomas 已提交
450 451 452
	}
	ext_debug("\n");
}
453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485

static void ext4_ext_show_move(struct inode *inode, struct ext4_ext_path *path,
			ext4_fsblk_t newblock, int level)
{
	int depth = ext_depth(inode);
	struct ext4_extent *ex;

	if (depth != level) {
		struct ext4_extent_idx *idx;
		idx = path[level].p_idx;
		while (idx <= EXT_MAX_INDEX(path[level].p_hdr)) {
			ext_debug("%d: move %d:%llu in new index %llu\n", level,
					le32_to_cpu(idx->ei_block),
					ext4_idx_pblock(idx),
					newblock);
			idx++;
		}

		return;
	}

	ex = path[depth].p_ext;
	while (ex <= EXT_MAX_EXTENT(path[depth].p_hdr)) {
		ext_debug("move %d:%llu:[%d]%d in new leaf %llu\n",
				le32_to_cpu(ex->ee_block),
				ext4_ext_pblock(ex),
				ext4_ext_is_uninitialized(ex),
				ext4_ext_get_actual_len(ex),
				newblock);
		ex++;
	}
}

A
Alex Tomas 已提交
486
#else
487 488
#define ext4_ext_show_path(inode, path)
#define ext4_ext_show_leaf(inode, path)
489
#define ext4_ext_show_move(inode, path, newblock, level)
A
Alex Tomas 已提交
490 491
#endif

492
void ext4_ext_drop_refs(struct ext4_ext_path *path)
A
Alex Tomas 已提交
493 494 495 496 497 498 499 500 501 502 503 504
{
	int depth = path->p_depth;
	int i;

	for (i = 0; i <= depth; i++, path++)
		if (path->p_bh) {
			brelse(path->p_bh);
			path->p_bh = NULL;
		}
}

/*
505 506
 * ext4_ext_binsearch_idx:
 * binary search for the closest index of the given block
507
 * the header must be checked before calling this
A
Alex Tomas 已提交
508 509
 */
static void
A
Aneesh Kumar K.V 已提交
510 511
ext4_ext_binsearch_idx(struct inode *inode,
			struct ext4_ext_path *path, ext4_lblk_t block)
A
Alex Tomas 已提交
512 513 514 515 516
{
	struct ext4_extent_header *eh = path->p_hdr;
	struct ext4_extent_idx *r, *l, *m;


517
	ext_debug("binsearch for %u(idx):  ", block);
A
Alex Tomas 已提交
518 519

	l = EXT_FIRST_INDEX(eh) + 1;
D
Dmitry Monakhov 已提交
520
	r = EXT_LAST_INDEX(eh);
A
Alex Tomas 已提交
521 522 523 524 525 526
	while (l <= r) {
		m = l + (r - l) / 2;
		if (block < le32_to_cpu(m->ei_block))
			r = m - 1;
		else
			l = m + 1;
527 528 529
		ext_debug("%p(%u):%p(%u):%p(%u) ", l, le32_to_cpu(l->ei_block),
				m, le32_to_cpu(m->ei_block),
				r, le32_to_cpu(r->ei_block));
A
Alex Tomas 已提交
530 531 532
	}

	path->p_idx = l - 1;
533
	ext_debug("  -> %d->%lld ", le32_to_cpu(path->p_idx->ei_block),
534
		  ext4_idx_pblock(path->p_idx));
A
Alex Tomas 已提交
535 536 537 538 539 540 541 542 543 544

#ifdef CHECK_BINSEARCH
	{
		struct ext4_extent_idx *chix, *ix;
		int k;

		chix = ix = EXT_FIRST_INDEX(eh);
		for (k = 0; k < le16_to_cpu(eh->eh_entries); k++, ix++) {
		  if (k != 0 &&
		      le32_to_cpu(ix->ei_block) <= le32_to_cpu(ix[-1].ei_block)) {
545 546 547 548
				printk(KERN_DEBUG "k=%d, ix=0x%p, "
				       "first=0x%p\n", k,
				       ix, EXT_FIRST_INDEX(eh));
				printk(KERN_DEBUG "%u <= %u\n",
A
Alex Tomas 已提交
549 550 551 552
				       le32_to_cpu(ix->ei_block),
				       le32_to_cpu(ix[-1].ei_block));
			}
			BUG_ON(k && le32_to_cpu(ix->ei_block)
D
Dave Kleikamp 已提交
553
					   <= le32_to_cpu(ix[-1].ei_block));
A
Alex Tomas 已提交
554 555 556 557 558 559 560 561 562 563 564
			if (block < le32_to_cpu(ix->ei_block))
				break;
			chix = ix;
		}
		BUG_ON(chix != path->p_idx);
	}
#endif

}

/*
565 566
 * ext4_ext_binsearch:
 * binary search for closest extent of the given block
567
 * the header must be checked before calling this
A
Alex Tomas 已提交
568 569
 */
static void
A
Aneesh Kumar K.V 已提交
570 571
ext4_ext_binsearch(struct inode *inode,
		struct ext4_ext_path *path, ext4_lblk_t block)
A
Alex Tomas 已提交
572 573 574 575 576 577
{
	struct ext4_extent_header *eh = path->p_hdr;
	struct ext4_extent *r, *l, *m;

	if (eh->eh_entries == 0) {
		/*
578 579
		 * this leaf is empty:
		 * we get such a leaf in split/add case
A
Alex Tomas 已提交
580 581 582 583
		 */
		return;
	}

584
	ext_debug("binsearch for %u:  ", block);
A
Alex Tomas 已提交
585 586

	l = EXT_FIRST_EXTENT(eh) + 1;
D
Dmitry Monakhov 已提交
587
	r = EXT_LAST_EXTENT(eh);
A
Alex Tomas 已提交
588 589 590 591 592 593 594

	while (l <= r) {
		m = l + (r - l) / 2;
		if (block < le32_to_cpu(m->ee_block))
			r = m - 1;
		else
			l = m + 1;
595 596 597
		ext_debug("%p(%u):%p(%u):%p(%u) ", l, le32_to_cpu(l->ee_block),
				m, le32_to_cpu(m->ee_block),
				r, le32_to_cpu(r->ee_block));
A
Alex Tomas 已提交
598 599 600
	}

	path->p_ext = l - 1;
601
	ext_debug("  -> %d:%llu:[%d]%d ",
D
Dave Kleikamp 已提交
602
			le32_to_cpu(path->p_ext->ee_block),
603
			ext4_ext_pblock(path->p_ext),
604
			ext4_ext_is_uninitialized(path->p_ext),
A
Amit Arora 已提交
605
			ext4_ext_get_actual_len(path->p_ext));
A
Alex Tomas 已提交
606 607 608 609 610 611 612 613 614

#ifdef CHECK_BINSEARCH
	{
		struct ext4_extent *chex, *ex;
		int k;

		chex = ex = EXT_FIRST_EXTENT(eh);
		for (k = 0; k < le16_to_cpu(eh->eh_entries); k++, ex++) {
			BUG_ON(k && le32_to_cpu(ex->ee_block)
D
Dave Kleikamp 已提交
615
					  <= le32_to_cpu(ex[-1].ee_block));
A
Alex Tomas 已提交
616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633
			if (block < le32_to_cpu(ex->ee_block))
				break;
			chex = ex;
		}
		BUG_ON(chex != path->p_ext);
	}
#endif

}

int ext4_ext_tree_init(handle_t *handle, struct inode *inode)
{
	struct ext4_extent_header *eh;

	eh = ext_inode_hdr(inode);
	eh->eh_depth = 0;
	eh->eh_entries = 0;
	eh->eh_magic = EXT4_EXT_MAGIC;
634
	eh->eh_max = cpu_to_le16(ext4_ext_space_root(inode, 0));
A
Alex Tomas 已提交
635 636 637 638 639 640
	ext4_mark_inode_dirty(handle, inode);
	ext4_ext_invalidate_cache(inode);
	return 0;
}

struct ext4_ext_path *
A
Aneesh Kumar K.V 已提交
641 642
ext4_ext_find_extent(struct inode *inode, ext4_lblk_t block,
					struct ext4_ext_path *path)
A
Alex Tomas 已提交
643 644 645 646 647 648
{
	struct ext4_extent_header *eh;
	struct buffer_head *bh;
	short int depth, i, ppos = 0, alloc = 0;

	eh = ext_inode_hdr(inode);
649
	depth = ext_depth(inode);
A
Alex Tomas 已提交
650 651 652

	/* account possible depth increase */
	if (!path) {
A
Avantika Mathur 已提交
653
		path = kzalloc(sizeof(struct ext4_ext_path) * (depth + 2),
A
Alex Tomas 已提交
654 655 656 657 658 659
				GFP_NOFS);
		if (!path)
			return ERR_PTR(-ENOMEM);
		alloc = 1;
	}
	path[0].p_hdr = eh;
660
	path[0].p_bh = NULL;
A
Alex Tomas 已提交
661

662
	i = depth;
A
Alex Tomas 已提交
663 664
	/* walk through the tree */
	while (i) {
665 666
		int need_to_validate = 0;

A
Alex Tomas 已提交
667 668
		ext_debug("depth %d: num %d, max %d\n",
			  ppos, le16_to_cpu(eh->eh_entries), le16_to_cpu(eh->eh_max));
669

A
Alex Tomas 已提交
670
		ext4_ext_binsearch_idx(inode, path + ppos, block);
671
		path[ppos].p_block = ext4_idx_pblock(path[ppos].p_idx);
A
Alex Tomas 已提交
672 673 674
		path[ppos].p_depth = i;
		path[ppos].p_ext = NULL;

675 676
		bh = sb_getblk(inode->i_sb, path[ppos].p_block);
		if (unlikely(!bh))
A
Alex Tomas 已提交
677
			goto err;
678
		if (!bh_uptodate_or_lock(bh)) {
679 680
			trace_ext4_ext_load_extent(inode, block,
						path[ppos].p_block);
681 682 683 684 685 686 687
			if (bh_submit_read(bh) < 0) {
				put_bh(bh);
				goto err;
			}
			/* validate the extent entries */
			need_to_validate = 1;
		}
A
Alex Tomas 已提交
688 689
		eh = ext_block_hdr(bh);
		ppos++;
690 691 692 693 694 695
		if (unlikely(ppos > depth)) {
			put_bh(bh);
			EXT4_ERROR_INODE(inode,
					 "ppos %d > depth %d", ppos, depth);
			goto err;
		}
A
Alex Tomas 已提交
696 697 698 699
		path[ppos].p_bh = bh;
		path[ppos].p_hdr = eh;
		i--;

700
		if (need_to_validate && ext4_ext_check(inode, eh, i))
A
Alex Tomas 已提交
701 702 703 704 705 706 707 708 709
			goto err;
	}

	path[ppos].p_depth = i;
	path[ppos].p_ext = NULL;
	path[ppos].p_idx = NULL;

	/* find extent */
	ext4_ext_binsearch(inode, path + ppos, block);
710 711
	/* if not an empty leaf */
	if (path[ppos].p_ext)
712
		path[ppos].p_block = ext4_ext_pblock(path[ppos].p_ext);
A
Alex Tomas 已提交
713 714 715 716 717 718 719 720 721 722 723 724 725

	ext4_ext_show_path(inode, path);

	return path;

err:
	ext4_ext_drop_refs(path);
	if (alloc)
		kfree(path);
	return ERR_PTR(-EIO);
}

/*
726 727 728
 * ext4_ext_insert_index:
 * insert new index [@logical;@ptr] into the block at @curp;
 * check where to insert: before @curp or after @curp
A
Alex Tomas 已提交
729
 */
730 731 732
static int ext4_ext_insert_index(handle_t *handle, struct inode *inode,
				 struct ext4_ext_path *curp,
				 int logical, ext4_fsblk_t ptr)
A
Alex Tomas 已提交
733 734 735 736
{
	struct ext4_extent_idx *ix;
	int len, err;

737 738
	err = ext4_ext_get_access(handle, inode, curp);
	if (err)
A
Alex Tomas 已提交
739 740
		return err;

741 742 743 744 745 746
	if (unlikely(logical == le32_to_cpu(curp->p_idx->ei_block))) {
		EXT4_ERROR_INODE(inode,
				 "logical %d == ei_block %d!",
				 logical, le32_to_cpu(curp->p_idx->ei_block));
		return -EIO;
	}
747 748 749 750 751 752 753 754 755 756

	if (unlikely(le16_to_cpu(curp->p_hdr->eh_entries)
			     >= le16_to_cpu(curp->p_hdr->eh_max))) {
		EXT4_ERROR_INODE(inode,
				 "eh_entries %d >= eh_max %d!",
				 le16_to_cpu(curp->p_hdr->eh_entries),
				 le16_to_cpu(curp->p_hdr->eh_max));
		return -EIO;
	}

A
Alex Tomas 已提交
757 758 759 760 761 762
	len = EXT_MAX_INDEX(curp->p_hdr) - curp->p_idx;
	if (logical > le32_to_cpu(curp->p_idx->ei_block)) {
		/* insert after */
		if (curp->p_idx != EXT_LAST_INDEX(curp->p_hdr)) {
			len = (len - 1) * sizeof(struct ext4_extent_idx);
			len = len < 0 ? 0 : len;
763
			ext_debug("insert new index %d after: %llu. "
A
Alex Tomas 已提交
764 765 766 767 768 769 770 771 772 773
					"move %d from 0x%p to 0x%p\n",
					logical, ptr, len,
					(curp->p_idx + 1), (curp->p_idx + 2));
			memmove(curp->p_idx + 2, curp->p_idx + 1, len);
		}
		ix = curp->p_idx + 1;
	} else {
		/* insert before */
		len = len * sizeof(struct ext4_extent_idx);
		len = len < 0 ? 0 : len;
774
		ext_debug("insert new index %d before: %llu. "
A
Alex Tomas 已提交
775 776 777 778 779 780 781
				"move %d from 0x%p to 0x%p\n",
				logical, ptr, len,
				curp->p_idx, (curp->p_idx + 1));
		memmove(curp->p_idx + 1, curp->p_idx, len);
		ix = curp->p_idx;
	}

782 783 784 785 786
	if (unlikely(ix > EXT_MAX_INDEX(curp->p_hdr))) {
		EXT4_ERROR_INODE(inode, "ix > EXT_MAX_INDEX!");
		return -EIO;
	}

A
Alex Tomas 已提交
787
	ix->ei_block = cpu_to_le32(logical);
788
	ext4_idx_store_pblock(ix, ptr);
M
Marcin Slusarz 已提交
789
	le16_add_cpu(&curp->p_hdr->eh_entries, 1);
A
Alex Tomas 已提交
790

791 792 793 794
	if (unlikely(ix > EXT_LAST_INDEX(curp->p_hdr))) {
		EXT4_ERROR_INODE(inode, "ix > EXT_LAST_INDEX!");
		return -EIO;
	}
A
Alex Tomas 已提交
795 796 797 798 799 800 801 802

	err = ext4_ext_dirty(handle, inode, curp);
	ext4_std_error(inode->i_sb, err);

	return err;
}

/*
803 804 805 806 807 808 809 810
 * ext4_ext_split:
 * inserts new subtree into the path, using free index entry
 * at depth @at:
 * - allocates all needed blocks (new leaf and all intermediate index blocks)
 * - makes decision where to split
 * - moves remaining extents and index entries (right to the split point)
 *   into the newly allocated blocks
 * - initializes subtree
A
Alex Tomas 已提交
811 812
 */
static int ext4_ext_split(handle_t *handle, struct inode *inode,
813 814 815
			  unsigned int flags,
			  struct ext4_ext_path *path,
			  struct ext4_extent *newext, int at)
A
Alex Tomas 已提交
816 817 818 819 820 821
{
	struct buffer_head *bh = NULL;
	int depth = ext_depth(inode);
	struct ext4_extent_header *neh;
	struct ext4_extent_idx *fidx;
	int i = at, k, m, a;
822
	ext4_fsblk_t newblock, oldblock;
A
Alex Tomas 已提交
823
	__le32 border;
824
	ext4_fsblk_t *ablocks = NULL; /* array of allocated blocks */
A
Alex Tomas 已提交
825 826 827
	int err = 0;

	/* make decision: where to split? */
828
	/* FIXME: now decision is simplest: at current extent */
A
Alex Tomas 已提交
829

830
	/* if current leaf will be split, then we should use
A
Alex Tomas 已提交
831
	 * border from split point */
832 833 834 835
	if (unlikely(path[depth].p_ext > EXT_MAX_EXTENT(path[depth].p_hdr))) {
		EXT4_ERROR_INODE(inode, "p_ext > EXT_MAX_EXTENT!");
		return -EIO;
	}
A
Alex Tomas 已提交
836 837
	if (path[depth].p_ext != EXT_MAX_EXTENT(path[depth].p_hdr)) {
		border = path[depth].p_ext[1].ee_block;
838
		ext_debug("leaf will be split."
A
Alex Tomas 已提交
839
				" next leaf starts at %d\n",
D
Dave Kleikamp 已提交
840
				  le32_to_cpu(border));
A
Alex Tomas 已提交
841 842 843 844
	} else {
		border = newext->ee_block;
		ext_debug("leaf will be added."
				" next leaf starts at %d\n",
D
Dave Kleikamp 已提交
845
				le32_to_cpu(border));
A
Alex Tomas 已提交
846 847 848
	}

	/*
849 850
	 * If error occurs, then we break processing
	 * and mark filesystem read-only. index won't
A
Alex Tomas 已提交
851
	 * be inserted and tree will be in consistent
852
	 * state. Next mount will repair buffers too.
A
Alex Tomas 已提交
853 854 855
	 */

	/*
856 857 858
	 * Get array to track all allocated blocks.
	 * We need this to handle errors and free blocks
	 * upon them.
A
Alex Tomas 已提交
859
	 */
A
Avantika Mathur 已提交
860
	ablocks = kzalloc(sizeof(ext4_fsblk_t) * depth, GFP_NOFS);
A
Alex Tomas 已提交
861 862 863 864 865 866
	if (!ablocks)
		return -ENOMEM;

	/* allocate all needed blocks */
	ext_debug("allocate %d blocks for indexes/leaf\n", depth - at);
	for (a = 0; a < depth - at; a++) {
A
Aneesh Kumar K.V 已提交
867
		newblock = ext4_ext_new_meta_block(handle, inode, path,
868
						   newext, &err, flags);
A
Alex Tomas 已提交
869 870 871 872 873 874 875
		if (newblock == 0)
			goto cleanup;
		ablocks[a] = newblock;
	}

	/* initialize new leaf */
	newblock = ablocks[--a];
876 877 878 879 880
	if (unlikely(newblock == 0)) {
		EXT4_ERROR_INODE(inode, "newblock == 0!");
		err = -EIO;
		goto cleanup;
	}
A
Alex Tomas 已提交
881 882 883 884 885 886 887
	bh = sb_getblk(inode->i_sb, newblock);
	if (!bh) {
		err = -EIO;
		goto cleanup;
	}
	lock_buffer(bh);

888 889
	err = ext4_journal_get_create_access(handle, bh);
	if (err)
A
Alex Tomas 已提交
890 891 892 893
		goto cleanup;

	neh = ext_block_hdr(bh);
	neh->eh_entries = 0;
894
	neh->eh_max = cpu_to_le16(ext4_ext_space_block(inode, 0));
A
Alex Tomas 已提交
895 896 897
	neh->eh_magic = EXT4_EXT_MAGIC;
	neh->eh_depth = 0;

898
	/* move remainder of path[depth] to the new leaf */
899 900 901 902 903 904 905 906
	if (unlikely(path[depth].p_hdr->eh_entries !=
		     path[depth].p_hdr->eh_max)) {
		EXT4_ERROR_INODE(inode, "eh_entries %d != eh_max %d!",
				 path[depth].p_hdr->eh_entries,
				 path[depth].p_hdr->eh_max);
		err = -EIO;
		goto cleanup;
	}
A
Alex Tomas 已提交
907
	/* start copy from next extent */
908 909
	m = EXT_MAX_EXTENT(path[depth].p_hdr) - path[depth].p_ext++;
	ext4_ext_show_move(inode, path, newblock, depth);
A
Alex Tomas 已提交
910
	if (m) {
911 912 913
		struct ext4_extent *ex;
		ex = EXT_FIRST_EXTENT(neh);
		memmove(ex, path[depth].p_ext, sizeof(struct ext4_extent) * m);
M
Marcin Slusarz 已提交
914
		le16_add_cpu(&neh->eh_entries, m);
A
Alex Tomas 已提交
915 916 917 918 919
	}

	set_buffer_uptodate(bh);
	unlock_buffer(bh);

920
	err = ext4_handle_dirty_metadata(handle, inode, bh);
921
	if (err)
A
Alex Tomas 已提交
922 923 924 925 926 927
		goto cleanup;
	brelse(bh);
	bh = NULL;

	/* correct old leaf */
	if (m) {
928 929
		err = ext4_ext_get_access(handle, inode, path + depth);
		if (err)
A
Alex Tomas 已提交
930
			goto cleanup;
M
Marcin Slusarz 已提交
931
		le16_add_cpu(&path[depth].p_hdr->eh_entries, -m);
932 933
		err = ext4_ext_dirty(handle, inode, path + depth);
		if (err)
A
Alex Tomas 已提交
934 935 936 937 938 939
			goto cleanup;

	}

	/* create intermediate indexes */
	k = depth - at - 1;
940 941 942 943 944
	if (unlikely(k < 0)) {
		EXT4_ERROR_INODE(inode, "k %d < 0!", k);
		err = -EIO;
		goto cleanup;
	}
A
Alex Tomas 已提交
945 946 947 948 949 950 951 952
	if (k)
		ext_debug("create %d intermediate indices\n", k);
	/* insert new index into current index block */
	/* current depth stored in i var */
	i = depth - 1;
	while (k--) {
		oldblock = newblock;
		newblock = ablocks[--a];
953
		bh = sb_getblk(inode->i_sb, newblock);
A
Alex Tomas 已提交
954 955 956 957 958 959
		if (!bh) {
			err = -EIO;
			goto cleanup;
		}
		lock_buffer(bh);

960 961
		err = ext4_journal_get_create_access(handle, bh);
		if (err)
A
Alex Tomas 已提交
962 963 964 965 966
			goto cleanup;

		neh = ext_block_hdr(bh);
		neh->eh_entries = cpu_to_le16(1);
		neh->eh_magic = EXT4_EXT_MAGIC;
967
		neh->eh_max = cpu_to_le16(ext4_ext_space_block_idx(inode, 0));
A
Alex Tomas 已提交
968 969 970
		neh->eh_depth = cpu_to_le16(depth - i);
		fidx = EXT_FIRST_INDEX(neh);
		fidx->ei_block = border;
971
		ext4_idx_store_pblock(fidx, oldblock);
A
Alex Tomas 已提交
972

973 974
		ext_debug("int.index at %d (block %llu): %u -> %llu\n",
				i, newblock, le32_to_cpu(border), oldblock);
A
Alex Tomas 已提交
975

976
		/* move remainder of path[i] to the new index block */
977 978 979 980 981 982 983 984
		if (unlikely(EXT_MAX_INDEX(path[i].p_hdr) !=
					EXT_LAST_INDEX(path[i].p_hdr))) {
			EXT4_ERROR_INODE(inode,
					 "EXT_MAX_INDEX != EXT_LAST_INDEX ee_block %d!",
					 le32_to_cpu(path[i].p_ext->ee_block));
			err = -EIO;
			goto cleanup;
		}
985 986 987 988 989
		/* start copy indexes */
		m = EXT_MAX_INDEX(path[i].p_hdr) - path[i].p_idx++;
		ext_debug("cur 0x%p, last 0x%p\n", path[i].p_idx,
				EXT_MAX_INDEX(path[i].p_hdr));
		ext4_ext_show_move(inode, path, newblock, i);
A
Alex Tomas 已提交
990
		if (m) {
991
			memmove(++fidx, path[i].p_idx,
A
Alex Tomas 已提交
992
				sizeof(struct ext4_extent_idx) * m);
M
Marcin Slusarz 已提交
993
			le16_add_cpu(&neh->eh_entries, m);
A
Alex Tomas 已提交
994 995 996 997
		}
		set_buffer_uptodate(bh);
		unlock_buffer(bh);

998
		err = ext4_handle_dirty_metadata(handle, inode, bh);
999
		if (err)
A
Alex Tomas 已提交
1000 1001 1002 1003 1004 1005 1006 1007 1008
			goto cleanup;
		brelse(bh);
		bh = NULL;

		/* correct old index */
		if (m) {
			err = ext4_ext_get_access(handle, inode, path + i);
			if (err)
				goto cleanup;
M
Marcin Slusarz 已提交
1009
			le16_add_cpu(&path[i].p_hdr->eh_entries, -m);
A
Alex Tomas 已提交
1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033
			err = ext4_ext_dirty(handle, inode, path + i);
			if (err)
				goto cleanup;
		}

		i--;
	}

	/* insert new index */
	err = ext4_ext_insert_index(handle, inode, path + at,
				    le32_to_cpu(border), newblock);

cleanup:
	if (bh) {
		if (buffer_locked(bh))
			unlock_buffer(bh);
		brelse(bh);
	}

	if (err) {
		/* free all allocated blocks in error case */
		for (i = 0; i < depth; i++) {
			if (!ablocks[i])
				continue;
1034
			ext4_free_blocks(handle, inode, NULL, ablocks[i], 1,
1035
					 EXT4_FREE_BLOCKS_METADATA);
A
Alex Tomas 已提交
1036 1037 1038 1039 1040 1041 1042 1043
		}
	}
	kfree(ablocks);

	return err;
}

/*
1044 1045 1046 1047 1048 1049
 * ext4_ext_grow_indepth:
 * implements tree growing procedure:
 * - allocates new block
 * - moves top-level data (index block or leaf) into the new block
 * - initializes new top-level, creating index that points to the
 *   just created block
A
Alex Tomas 已提交
1050 1051
 */
static int ext4_ext_grow_indepth(handle_t *handle, struct inode *inode,
1052 1053
				 unsigned int flags,
				 struct ext4_extent *newext)
A
Alex Tomas 已提交
1054 1055 1056
{
	struct ext4_extent_header *neh;
	struct buffer_head *bh;
1057
	ext4_fsblk_t newblock;
A
Alex Tomas 已提交
1058 1059
	int err = 0;

1060
	newblock = ext4_ext_new_meta_block(handle, inode, NULL,
1061
		newext, &err, flags);
A
Alex Tomas 已提交
1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072
	if (newblock == 0)
		return err;

	bh = sb_getblk(inode->i_sb, newblock);
	if (!bh) {
		err = -EIO;
		ext4_std_error(inode->i_sb, err);
		return err;
	}
	lock_buffer(bh);

1073 1074
	err = ext4_journal_get_create_access(handle, bh);
	if (err) {
A
Alex Tomas 已提交
1075 1076 1077 1078 1079
		unlock_buffer(bh);
		goto out;
	}

	/* move top-level index/leaf into new block */
1080 1081
	memmove(bh->b_data, EXT4_I(inode)->i_data,
		sizeof(EXT4_I(inode)->i_data));
A
Alex Tomas 已提交
1082 1083 1084 1085 1086 1087

	/* set size of new block */
	neh = ext_block_hdr(bh);
	/* old root could have indexes or leaves
	 * so calculate e_max right way */
	if (ext_depth(inode))
1088
		neh->eh_max = cpu_to_le16(ext4_ext_space_block_idx(inode, 0));
A
Alex Tomas 已提交
1089
	else
1090
		neh->eh_max = cpu_to_le16(ext4_ext_space_block(inode, 0));
A
Alex Tomas 已提交
1091 1092 1093 1094
	neh->eh_magic = EXT4_EXT_MAGIC;
	set_buffer_uptodate(bh);
	unlock_buffer(bh);

1095
	err = ext4_handle_dirty_metadata(handle, inode, bh);
1096
	if (err)
A
Alex Tomas 已提交
1097 1098
		goto out;

1099
	/* Update top-level index: num,max,pointer */
A
Alex Tomas 已提交
1100
	neh = ext_inode_hdr(inode);
1101 1102 1103 1104 1105 1106 1107 1108
	neh->eh_entries = cpu_to_le16(1);
	ext4_idx_store_pblock(EXT_FIRST_INDEX(neh), newblock);
	if (neh->eh_depth == 0) {
		/* Root extent block becomes index block */
		neh->eh_max = cpu_to_le16(ext4_ext_space_root_idx(inode, 0));
		EXT_FIRST_INDEX(neh)->ei_block =
			EXT_FIRST_EXTENT(neh)->ee_block;
	}
1109
	ext_debug("new root: num %d(%d), lblock %d, ptr %llu\n",
A
Alex Tomas 已提交
1110
		  le16_to_cpu(neh->eh_entries), le16_to_cpu(neh->eh_max),
1111
		  le32_to_cpu(EXT_FIRST_INDEX(neh)->ei_block),
1112
		  ext4_idx_pblock(EXT_FIRST_INDEX(neh)));
A
Alex Tomas 已提交
1113

1114 1115
	neh->eh_depth = cpu_to_le16(neh->eh_depth + 1);
	ext4_mark_inode_dirty(handle, inode);
A
Alex Tomas 已提交
1116 1117 1118 1119 1120 1121 1122
out:
	brelse(bh);

	return err;
}

/*
1123 1124 1125
 * ext4_ext_create_new_leaf:
 * finds empty index and adds new leaf.
 * if no free index is found, then it requests in-depth growing.
A
Alex Tomas 已提交
1126 1127
 */
static int ext4_ext_create_new_leaf(handle_t *handle, struct inode *inode,
1128 1129 1130
				    unsigned int flags,
				    struct ext4_ext_path *path,
				    struct ext4_extent *newext)
A
Alex Tomas 已提交
1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144
{
	struct ext4_ext_path *curp;
	int depth, i, err = 0;

repeat:
	i = depth = ext_depth(inode);

	/* walk up to the tree and look for free index entry */
	curp = path + depth;
	while (i > 0 && !EXT_HAS_FREE_INDEX(curp)) {
		i--;
		curp--;
	}

1145 1146
	/* we use already allocated block for index block,
	 * so subsequent data blocks should be contiguous */
A
Alex Tomas 已提交
1147 1148 1149
	if (EXT_HAS_FREE_INDEX(curp)) {
		/* if we found index with free entry, then use that
		 * entry: create all needed subtree and add new leaf */
1150
		err = ext4_ext_split(handle, inode, flags, path, newext, i);
1151 1152
		if (err)
			goto out;
A
Alex Tomas 已提交
1153 1154 1155 1156

		/* refill path */
		ext4_ext_drop_refs(path);
		path = ext4_ext_find_extent(inode,
A
Aneesh Kumar K.V 已提交
1157 1158
				    (ext4_lblk_t)le32_to_cpu(newext->ee_block),
				    path);
A
Alex Tomas 已提交
1159 1160 1161 1162
		if (IS_ERR(path))
			err = PTR_ERR(path);
	} else {
		/* tree is full, time to grow in depth */
1163
		err = ext4_ext_grow_indepth(handle, inode, flags, newext);
A
Alex Tomas 已提交
1164 1165 1166 1167 1168 1169
		if (err)
			goto out;

		/* refill path */
		ext4_ext_drop_refs(path);
		path = ext4_ext_find_extent(inode,
A
Aneesh Kumar K.V 已提交
1170 1171
				   (ext4_lblk_t)le32_to_cpu(newext->ee_block),
				    path);
A
Alex Tomas 已提交
1172 1173 1174 1175 1176 1177
		if (IS_ERR(path)) {
			err = PTR_ERR(path);
			goto out;
		}

		/*
1178 1179
		 * only first (depth 0 -> 1) produces free space;
		 * in all other cases we have to split the grown tree
A
Alex Tomas 已提交
1180 1181 1182
		 */
		depth = ext_depth(inode);
		if (path[depth].p_hdr->eh_entries == path[depth].p_hdr->eh_max) {
1183
			/* now we need to split */
A
Alex Tomas 已提交
1184 1185 1186 1187 1188 1189 1190 1191
			goto repeat;
		}
	}

out:
	return err;
}

1192 1193 1194 1195 1196 1197 1198
/*
 * search the closest allocated block to the left for *logical
 * and returns it at @logical + it's physical address at @phys
 * if *logical is the smallest allocated block, the function
 * returns 0 at @phys
 * return value contains 0 (success) or error code
 */
1199 1200 1201
static int ext4_ext_search_left(struct inode *inode,
				struct ext4_ext_path *path,
				ext4_lblk_t *logical, ext4_fsblk_t *phys)
1202 1203 1204
{
	struct ext4_extent_idx *ix;
	struct ext4_extent *ex;
1205
	int depth, ee_len;
1206

1207 1208 1209 1210
	if (unlikely(path == NULL)) {
		EXT4_ERROR_INODE(inode, "path == NULL *logical %d!", *logical);
		return -EIO;
	}
1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221
	depth = path->p_depth;
	*phys = 0;

	if (depth == 0 && path->p_ext == NULL)
		return 0;

	/* usually extent in the path covers blocks smaller
	 * then *logical, but it can be that extent is the
	 * first one in the file */

	ex = path[depth].p_ext;
1222
	ee_len = ext4_ext_get_actual_len(ex);
1223
	if (*logical < le32_to_cpu(ex->ee_block)) {
1224 1225 1226 1227 1228 1229
		if (unlikely(EXT_FIRST_EXTENT(path[depth].p_hdr) != ex)) {
			EXT4_ERROR_INODE(inode,
					 "EXT_FIRST_EXTENT != ex *logical %d ee_block %d!",
					 *logical, le32_to_cpu(ex->ee_block));
			return -EIO;
		}
1230 1231
		while (--depth >= 0) {
			ix = path[depth].p_idx;
1232 1233 1234
			if (unlikely(ix != EXT_FIRST_INDEX(path[depth].p_hdr))) {
				EXT4_ERROR_INODE(inode,
				  "ix (%d) != EXT_FIRST_INDEX (%d) (depth %d)!",
1235
				  ix != NULL ? le32_to_cpu(ix->ei_block) : 0,
1236
				  EXT_FIRST_INDEX(path[depth].p_hdr) != NULL ?
1237
		le32_to_cpu(EXT_FIRST_INDEX(path[depth].p_hdr)->ei_block) : 0,
1238 1239 1240
				  depth);
				return -EIO;
			}
1241 1242 1243 1244
		}
		return 0;
	}

1245 1246 1247 1248 1249 1250
	if (unlikely(*logical < (le32_to_cpu(ex->ee_block) + ee_len))) {
		EXT4_ERROR_INODE(inode,
				 "logical %d < ee_block %d + ee_len %d!",
				 *logical, le32_to_cpu(ex->ee_block), ee_len);
		return -EIO;
	}
1251

1252
	*logical = le32_to_cpu(ex->ee_block) + ee_len - 1;
1253
	*phys = ext4_ext_pblock(ex) + ee_len - 1;
1254 1255 1256 1257 1258 1259
	return 0;
}

/*
 * search the closest allocated block to the right for *logical
 * and returns it at @logical + it's physical address at @phys
1260
 * if *logical is the largest allocated block, the function
1261 1262 1263
 * returns 0 at @phys
 * return value contains 0 (success) or error code
 */
1264 1265
static int ext4_ext_search_right(struct inode *inode,
				 struct ext4_ext_path *path,
1266 1267
				 ext4_lblk_t *logical, ext4_fsblk_t *phys,
				 struct ext4_extent **ret_ex)
1268 1269 1270 1271 1272 1273
{
	struct buffer_head *bh = NULL;
	struct ext4_extent_header *eh;
	struct ext4_extent_idx *ix;
	struct ext4_extent *ex;
	ext4_fsblk_t block;
1274 1275
	int depth;	/* Note, NOT eh_depth; depth from top of tree */
	int ee_len;
1276

1277 1278 1279 1280
	if (unlikely(path == NULL)) {
		EXT4_ERROR_INODE(inode, "path == NULL *logical %d!", *logical);
		return -EIO;
	}
1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291
	depth = path->p_depth;
	*phys = 0;

	if (depth == 0 && path->p_ext == NULL)
		return 0;

	/* usually extent in the path covers blocks smaller
	 * then *logical, but it can be that extent is the
	 * first one in the file */

	ex = path[depth].p_ext;
1292
	ee_len = ext4_ext_get_actual_len(ex);
1293
	if (*logical < le32_to_cpu(ex->ee_block)) {
1294 1295 1296 1297 1298 1299
		if (unlikely(EXT_FIRST_EXTENT(path[depth].p_hdr) != ex)) {
			EXT4_ERROR_INODE(inode,
					 "first_extent(path[%d].p_hdr) != ex",
					 depth);
			return -EIO;
		}
1300 1301
		while (--depth >= 0) {
			ix = path[depth].p_idx;
1302 1303 1304 1305 1306 1307
			if (unlikely(ix != EXT_FIRST_INDEX(path[depth].p_hdr))) {
				EXT4_ERROR_INODE(inode,
						 "ix != EXT_FIRST_INDEX *logical %d!",
						 *logical);
				return -EIO;
			}
1308
		}
1309
		goto found_extent;
1310 1311
	}

1312 1313 1314 1315 1316 1317
	if (unlikely(*logical < (le32_to_cpu(ex->ee_block) + ee_len))) {
		EXT4_ERROR_INODE(inode,
				 "logical %d < ee_block %d + ee_len %d!",
				 *logical, le32_to_cpu(ex->ee_block), ee_len);
		return -EIO;
	}
1318 1319 1320 1321

	if (ex != EXT_LAST_EXTENT(path[depth].p_hdr)) {
		/* next allocated block in this leaf */
		ex++;
1322
		goto found_extent;
1323 1324 1325 1326 1327 1328
	}

	/* go up and search for index to the right */
	while (--depth >= 0) {
		ix = path[depth].p_idx;
		if (ix != EXT_LAST_INDEX(path[depth].p_hdr))
W
Wu Fengguang 已提交
1329
			goto got_index;
1330 1331
	}

W
Wu Fengguang 已提交
1332 1333
	/* we've gone up to the root and found no index to the right */
	return 0;
1334

W
Wu Fengguang 已提交
1335
got_index:
1336 1337 1338 1339
	/* we've found index to the right, let's
	 * follow it and find the closest allocated
	 * block to the right */
	ix++;
1340
	block = ext4_idx_pblock(ix);
1341 1342 1343 1344 1345
	while (++depth < path->p_depth) {
		bh = sb_bread(inode->i_sb, block);
		if (bh == NULL)
			return -EIO;
		eh = ext_block_hdr(bh);
1346
		/* subtract from p_depth to get proper eh_depth */
1347
		if (ext4_ext_check(inode, eh, path->p_depth - depth)) {
1348 1349 1350 1351
			put_bh(bh);
			return -EIO;
		}
		ix = EXT_FIRST_INDEX(eh);
1352
		block = ext4_idx_pblock(ix);
1353 1354 1355 1356 1357 1358 1359
		put_bh(bh);
	}

	bh = sb_bread(inode->i_sb, block);
	if (bh == NULL)
		return -EIO;
	eh = ext_block_hdr(bh);
1360
	if (ext4_ext_check(inode, eh, path->p_depth - depth)) {
1361 1362 1363 1364
		put_bh(bh);
		return -EIO;
	}
	ex = EXT_FIRST_EXTENT(eh);
1365
found_extent:
1366
	*logical = le32_to_cpu(ex->ee_block);
1367
	*phys = ext4_ext_pblock(ex);
1368 1369 1370
	*ret_ex = ex;
	if (bh)
		put_bh(bh);
1371 1372 1373
	return 0;
}

A
Alex Tomas 已提交
1374
/*
1375
 * ext4_ext_next_allocated_block:
1376
 * returns allocated block in subsequent extent or EXT_MAX_BLOCKS.
1377 1378 1379
 * NOTE: it considers block number from index entry as
 * allocated block. Thus, index entries have to be consistent
 * with leaves.
A
Alex Tomas 已提交
1380
 */
A
Aneesh Kumar K.V 已提交
1381
static ext4_lblk_t
A
Alex Tomas 已提交
1382 1383 1384 1385 1386 1387 1388 1389
ext4_ext_next_allocated_block(struct ext4_ext_path *path)
{
	int depth;

	BUG_ON(path == NULL);
	depth = path->p_depth;

	if (depth == 0 && path->p_ext == NULL)
1390
		return EXT_MAX_BLOCKS;
A
Alex Tomas 已提交
1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406

	while (depth >= 0) {
		if (depth == path->p_depth) {
			/* leaf */
			if (path[depth].p_ext !=
					EXT_LAST_EXTENT(path[depth].p_hdr))
			  return le32_to_cpu(path[depth].p_ext[1].ee_block);
		} else {
			/* index */
			if (path[depth].p_idx !=
					EXT_LAST_INDEX(path[depth].p_hdr))
			  return le32_to_cpu(path[depth].p_idx[1].ei_block);
		}
		depth--;
	}

1407
	return EXT_MAX_BLOCKS;
A
Alex Tomas 已提交
1408 1409 1410
}

/*
1411
 * ext4_ext_next_leaf_block:
1412
 * returns first allocated block from next leaf or EXT_MAX_BLOCKS
A
Alex Tomas 已提交
1413
 */
1414
static ext4_lblk_t ext4_ext_next_leaf_block(struct ext4_ext_path *path)
A
Alex Tomas 已提交
1415 1416 1417 1418 1419 1420 1421 1422
{
	int depth;

	BUG_ON(path == NULL);
	depth = path->p_depth;

	/* zero-tree has no leaf blocks at all */
	if (depth == 0)
1423
		return EXT_MAX_BLOCKS;
A
Alex Tomas 已提交
1424 1425 1426 1427 1428 1429 1430

	/* go to index block */
	depth--;

	while (depth >= 0) {
		if (path[depth].p_idx !=
				EXT_LAST_INDEX(path[depth].p_hdr))
A
Aneesh Kumar K.V 已提交
1431 1432
			return (ext4_lblk_t)
				le32_to_cpu(path[depth].p_idx[1].ei_block);
A
Alex Tomas 已提交
1433 1434 1435
		depth--;
	}

1436
	return EXT_MAX_BLOCKS;
A
Alex Tomas 已提交
1437 1438 1439
}

/*
1440 1441 1442
 * ext4_ext_correct_indexes:
 * if leaf gets modified and modified extent is first in the leaf,
 * then we have to correct all indexes above.
A
Alex Tomas 已提交
1443 1444
 * TODO: do we need to correct tree in all cases?
 */
A
Aneesh Kumar K.V 已提交
1445
static int ext4_ext_correct_indexes(handle_t *handle, struct inode *inode,
A
Alex Tomas 已提交
1446 1447 1448 1449 1450 1451 1452 1453 1454 1455
				struct ext4_ext_path *path)
{
	struct ext4_extent_header *eh;
	int depth = ext_depth(inode);
	struct ext4_extent *ex;
	__le32 border;
	int k, err = 0;

	eh = path[depth].p_hdr;
	ex = path[depth].p_ext;
1456 1457 1458 1459 1460 1461

	if (unlikely(ex == NULL || eh == NULL)) {
		EXT4_ERROR_INODE(inode,
				 "ex %p == NULL or eh %p == NULL", ex, eh);
		return -EIO;
	}
A
Alex Tomas 已提交
1462 1463 1464 1465 1466 1467 1468 1469 1470 1471 1472 1473

	if (depth == 0) {
		/* there is no tree at all */
		return 0;
	}

	if (ex != EXT_FIRST_EXTENT(eh)) {
		/* we correct tree if first leaf got modified only */
		return 0;
	}

	/*
1474
	 * TODO: we need correction if border is smaller than current one
A
Alex Tomas 已提交
1475 1476 1477
	 */
	k = depth - 1;
	border = path[depth].p_ext->ee_block;
1478 1479
	err = ext4_ext_get_access(handle, inode, path + k);
	if (err)
A
Alex Tomas 已提交
1480 1481
		return err;
	path[k].p_idx->ei_block = border;
1482 1483
	err = ext4_ext_dirty(handle, inode, path + k);
	if (err)
A
Alex Tomas 已提交
1484 1485 1486 1487 1488 1489
		return err;

	while (k--) {
		/* change all left-side indexes */
		if (path[k+1].p_idx != EXT_FIRST_INDEX(path[k+1].p_hdr))
			break;
1490 1491
		err = ext4_ext_get_access(handle, inode, path + k);
		if (err)
A
Alex Tomas 已提交
1492 1493
			break;
		path[k].p_idx->ei_block = border;
1494 1495
		err = ext4_ext_dirty(handle, inode, path + k);
		if (err)
A
Alex Tomas 已提交
1496 1497 1498 1499 1500 1501
			break;
	}

	return err;
}

1502
int
A
Alex Tomas 已提交
1503 1504 1505
ext4_can_extents_be_merged(struct inode *inode, struct ext4_extent *ex1,
				struct ext4_extent *ex2)
{
1506
	unsigned short ext1_ee_len, ext2_ee_len, max_len;
A
Amit Arora 已提交
1507 1508 1509 1510 1511 1512 1513 1514

	/*
	 * Make sure that either both extents are uninitialized, or
	 * both are _not_.
	 */
	if (ext4_ext_is_uninitialized(ex1) ^ ext4_ext_is_uninitialized(ex2))
		return 0;

1515 1516 1517 1518 1519
	if (ext4_ext_is_uninitialized(ex1))
		max_len = EXT_UNINIT_MAX_LEN;
	else
		max_len = EXT_INIT_MAX_LEN;

A
Amit Arora 已提交
1520 1521 1522 1523
	ext1_ee_len = ext4_ext_get_actual_len(ex1);
	ext2_ee_len = ext4_ext_get_actual_len(ex2);

	if (le32_to_cpu(ex1->ee_block) + ext1_ee_len !=
A
Andrew Morton 已提交
1524
			le32_to_cpu(ex2->ee_block))
A
Alex Tomas 已提交
1525 1526
		return 0;

1527 1528 1529
	/*
	 * To allow future support for preallocated extents to be added
	 * as an RO_COMPAT feature, refuse to merge to extents if
1530
	 * this can result in the top bit of ee_len being set.
1531
	 */
1532
	if (ext1_ee_len + ext2_ee_len > max_len)
1533
		return 0;
1534
#ifdef AGGRESSIVE_TEST
1535
	if (ext1_ee_len >= 4)
A
Alex Tomas 已提交
1536 1537 1538
		return 0;
#endif

1539
	if (ext4_ext_pblock(ex1) + ext1_ee_len == ext4_ext_pblock(ex2))
A
Alex Tomas 已提交
1540 1541 1542 1543
		return 1;
	return 0;
}

1544 1545 1546 1547 1548 1549 1550
/*
 * This function tries to merge the "ex" extent to the next extent in the tree.
 * It always tries to merge towards right. If you want to merge towards
 * left, pass "ex - 1" as argument instead of "ex".
 * Returns 0 if the extents (ex and ex+1) were _not_ merged and returns
 * 1 if they got merged.
 */
1551
static int ext4_ext_try_to_merge_right(struct inode *inode,
1552 1553
				 struct ext4_ext_path *path,
				 struct ext4_extent *ex)
1554 1555 1556 1557 1558 1559 1560 1561 1562 1563 1564 1565 1566 1567 1568 1569 1570 1571 1572 1573 1574 1575 1576 1577 1578 1579
{
	struct ext4_extent_header *eh;
	unsigned int depth, len;
	int merge_done = 0;
	int uninitialized = 0;

	depth = ext_depth(inode);
	BUG_ON(path[depth].p_hdr == NULL);
	eh = path[depth].p_hdr;

	while (ex < EXT_LAST_EXTENT(eh)) {
		if (!ext4_can_extents_be_merged(inode, ex, ex + 1))
			break;
		/* merge with next extent! */
		if (ext4_ext_is_uninitialized(ex))
			uninitialized = 1;
		ex->ee_len = cpu_to_le16(ext4_ext_get_actual_len(ex)
				+ ext4_ext_get_actual_len(ex + 1));
		if (uninitialized)
			ext4_ext_mark_uninitialized(ex);

		if (ex + 1 < EXT_LAST_EXTENT(eh)) {
			len = (EXT_LAST_EXTENT(eh) - ex - 1)
				* sizeof(struct ext4_extent);
			memmove(ex + 1, ex + 2, len);
		}
M
Marcin Slusarz 已提交
1580
		le16_add_cpu(&eh->eh_entries, -1);
1581 1582 1583
		merge_done = 1;
		WARN_ON(eh->eh_entries == 0);
		if (!eh->eh_entries)
1584
			EXT4_ERROR_INODE(inode, "eh->eh_entries = 0!");
1585 1586 1587 1588 1589
	}

	return merge_done;
}

1590 1591 1592 1593 1594 1595 1596 1597 1598 1599 1600 1601 1602 1603 1604 1605 1606 1607 1608 1609 1610 1611 1612 1613 1614
/*
 * This function tries to merge the @ex extent to neighbours in the tree.
 * return 1 if merge left else 0.
 */
static int ext4_ext_try_to_merge(struct inode *inode,
				  struct ext4_ext_path *path,
				  struct ext4_extent *ex) {
	struct ext4_extent_header *eh;
	unsigned int depth;
	int merge_done = 0;
	int ret = 0;

	depth = ext_depth(inode);
	BUG_ON(path[depth].p_hdr == NULL);
	eh = path[depth].p_hdr;

	if (ex > EXT_FIRST_EXTENT(eh))
		merge_done = ext4_ext_try_to_merge_right(inode, path, ex - 1);

	if (!merge_done)
		ret = ext4_ext_try_to_merge_right(inode, path, ex);

	return ret;
}

A
Amit Arora 已提交
1615 1616 1617 1618 1619 1620 1621 1622
/*
 * check if a portion of the "newext" extent overlaps with an
 * existing extent.
 *
 * If there is an overlap discovered, it updates the length of the newext
 * such that there will be no overlap, and then returns 1.
 * If there is no overlap found, it returns 0.
 */
1623 1624
static unsigned int ext4_ext_check_overlap(struct ext4_sb_info *sbi,
					   struct inode *inode,
1625 1626
					   struct ext4_extent *newext,
					   struct ext4_ext_path *path)
A
Amit Arora 已提交
1627
{
A
Aneesh Kumar K.V 已提交
1628
	ext4_lblk_t b1, b2;
A
Amit Arora 已提交
1629 1630 1631 1632
	unsigned int depth, len1;
	unsigned int ret = 0;

	b1 = le32_to_cpu(newext->ee_block);
A
Amit Arora 已提交
1633
	len1 = ext4_ext_get_actual_len(newext);
A
Amit Arora 已提交
1634 1635 1636 1637
	depth = ext_depth(inode);
	if (!path[depth].p_ext)
		goto out;
	b2 = le32_to_cpu(path[depth].p_ext->ee_block);
1638
	b2 &= ~(sbi->s_cluster_ratio - 1);
A
Amit Arora 已提交
1639 1640 1641

	/*
	 * get the next allocated block if the extent in the path
1642
	 * is before the requested block(s)
A
Amit Arora 已提交
1643 1644 1645
	 */
	if (b2 < b1) {
		b2 = ext4_ext_next_allocated_block(path);
1646
		if (b2 == EXT_MAX_BLOCKS)
A
Amit Arora 已提交
1647
			goto out;
1648
		b2 &= ~(sbi->s_cluster_ratio - 1);
A
Amit Arora 已提交
1649 1650
	}

A
Aneesh Kumar K.V 已提交
1651
	/* check for wrap through zero on extent logical start block*/
A
Amit Arora 已提交
1652
	if (b1 + len1 < b1) {
1653
		len1 = EXT_MAX_BLOCKS - b1;
A
Amit Arora 已提交
1654 1655 1656 1657 1658 1659 1660 1661 1662 1663 1664 1665 1666
		newext->ee_len = cpu_to_le16(len1);
		ret = 1;
	}

	/* check for overlap */
	if (b1 + len1 > b2) {
		newext->ee_len = cpu_to_le16(b2 - b1);
		ret = 1;
	}
out:
	return ret;
}

A
Alex Tomas 已提交
1667
/*
1668 1669 1670 1671
 * ext4_ext_insert_extent:
 * tries to merge requsted extent into the existing extent or
 * inserts requested extent as new one into the tree,
 * creating new leaf in the no-space case.
A
Alex Tomas 已提交
1672 1673 1674
 */
int ext4_ext_insert_extent(handle_t *handle, struct inode *inode,
				struct ext4_ext_path *path,
1675
				struct ext4_extent *newext, int flag)
A
Alex Tomas 已提交
1676
{
1677
	struct ext4_extent_header *eh;
A
Alex Tomas 已提交
1678 1679 1680
	struct ext4_extent *ex, *fex;
	struct ext4_extent *nearex; /* nearest extent */
	struct ext4_ext_path *npath = NULL;
A
Aneesh Kumar K.V 已提交
1681 1682
	int depth, len, err;
	ext4_lblk_t next;
A
Amit Arora 已提交
1683
	unsigned uninitialized = 0;
1684
	int flags = 0;
A
Alex Tomas 已提交
1685

1686 1687 1688 1689
	if (unlikely(ext4_ext_get_actual_len(newext) == 0)) {
		EXT4_ERROR_INODE(inode, "ext4_ext_get_actual_len(newext) == 0");
		return -EIO;
	}
A
Alex Tomas 已提交
1690 1691
	depth = ext_depth(inode);
	ex = path[depth].p_ext;
1692 1693 1694 1695
	if (unlikely(path[depth].p_hdr == NULL)) {
		EXT4_ERROR_INODE(inode, "path[%d].p_hdr == NULL", depth);
		return -EIO;
	}
A
Alex Tomas 已提交
1696 1697

	/* try to insert block into found extent and return */
1698
	if (ex && !(flag & EXT4_GET_BLOCKS_PRE_IO)
1699
		&& ext4_can_extents_be_merged(inode, ex, newext)) {
1700
		ext_debug("append [%d]%d block to %d:[%d]%d (from %llu)\n",
1701 1702 1703 1704 1705 1706
			  ext4_ext_is_uninitialized(newext),
			  ext4_ext_get_actual_len(newext),
			  le32_to_cpu(ex->ee_block),
			  ext4_ext_is_uninitialized(ex),
			  ext4_ext_get_actual_len(ex),
			  ext4_ext_pblock(ex));
1707 1708
		err = ext4_ext_get_access(handle, inode, path + depth);
		if (err)
A
Alex Tomas 已提交
1709
			return err;
A
Amit Arora 已提交
1710 1711 1712 1713 1714 1715 1716 1717 1718 1719 1720 1721

		/*
		 * ext4_can_extents_be_merged should have checked that either
		 * both extents are uninitialized, or both aren't. Thus we
		 * need to check only one of them here.
		 */
		if (ext4_ext_is_uninitialized(ex))
			uninitialized = 1;
		ex->ee_len = cpu_to_le16(ext4_ext_get_actual_len(ex)
					+ ext4_ext_get_actual_len(newext));
		if (uninitialized)
			ext4_ext_mark_uninitialized(ex);
A
Alex Tomas 已提交
1722 1723 1724 1725 1726 1727 1728 1729 1730 1731 1732 1733
		eh = path[depth].p_hdr;
		nearex = ex;
		goto merge;
	}

	depth = ext_depth(inode);
	eh = path[depth].p_hdr;
	if (le16_to_cpu(eh->eh_entries) < le16_to_cpu(eh->eh_max))
		goto has_space;

	/* probably next leaf has space for us? */
	fex = EXT_LAST_EXTENT(eh);
1734 1735
	next = EXT_MAX_BLOCKS;
	if (le32_to_cpu(newext->ee_block) > le32_to_cpu(fex->ee_block))
1736
		next = ext4_ext_next_leaf_block(path);
1737
	if (next != EXT_MAX_BLOCKS) {
A
Alex Tomas 已提交
1738 1739 1740 1741 1742 1743 1744 1745
		ext_debug("next leaf block - %d\n", next);
		BUG_ON(npath != NULL);
		npath = ext4_ext_find_extent(inode, next, NULL);
		if (IS_ERR(npath))
			return PTR_ERR(npath);
		BUG_ON(npath->p_depth != path->p_depth);
		eh = npath[depth].p_hdr;
		if (le16_to_cpu(eh->eh_entries) < le16_to_cpu(eh->eh_max)) {
L
Lucas De Marchi 已提交
1746
			ext_debug("next leaf isn't full(%d)\n",
A
Alex Tomas 已提交
1747 1748
				  le16_to_cpu(eh->eh_entries));
			path = npath;
1749
			goto has_space;
A
Alex Tomas 已提交
1750 1751 1752 1753 1754 1755
		}
		ext_debug("next leaf has no free space(%d,%d)\n",
			  le16_to_cpu(eh->eh_entries), le16_to_cpu(eh->eh_max));
	}

	/*
1756 1757
	 * There is no free space in the found leaf.
	 * We're gonna add a new leaf in the tree.
A
Alex Tomas 已提交
1758
	 */
1759 1760 1761
	if (flag & EXT4_GET_BLOCKS_PUNCH_OUT_EXT)
		flags = EXT4_MB_USE_ROOT_BLOCKS;
	err = ext4_ext_create_new_leaf(handle, inode, flags, path, newext);
A
Alex Tomas 已提交
1762 1763 1764 1765 1766 1767 1768 1769
	if (err)
		goto cleanup;
	depth = ext_depth(inode);
	eh = path[depth].p_hdr;

has_space:
	nearex = path[depth].p_ext;

1770 1771
	err = ext4_ext_get_access(handle, inode, path + depth);
	if (err)
A
Alex Tomas 已提交
1772 1773 1774 1775
		goto cleanup;

	if (!nearex) {
		/* there is no extent in this leaf, create first one */
1776
		ext_debug("first extent in the leaf: %d:%llu:[%d]%d\n",
D
Dave Kleikamp 已提交
1777
				le32_to_cpu(newext->ee_block),
1778
				ext4_ext_pblock(newext),
1779
				ext4_ext_is_uninitialized(newext),
A
Amit Arora 已提交
1780
				ext4_ext_get_actual_len(newext));
A
Alex Tomas 已提交
1781 1782
		path[depth].p_ext = EXT_FIRST_EXTENT(eh);
	} else if (le32_to_cpu(newext->ee_block)
D
Dave Kleikamp 已提交
1783
			   > le32_to_cpu(nearex->ee_block)) {
A
Alex Tomas 已提交
1784 1785 1786 1787 1788
/*		BUG_ON(newext->ee_block == nearex->ee_block); */
		if (nearex != EXT_LAST_EXTENT(eh)) {
			len = EXT_MAX_EXTENT(eh) - nearex;
			len = (len - 1) * sizeof(struct ext4_extent);
			len = len < 0 ? 0 : len;
1789
			ext_debug("insert %d:%llu:[%d]%d after: nearest 0x%p, "
A
Alex Tomas 已提交
1790
					"move %d from 0x%p to 0x%p\n",
D
Dave Kleikamp 已提交
1791
					le32_to_cpu(newext->ee_block),
1792
					ext4_ext_pblock(newext),
1793
					ext4_ext_is_uninitialized(newext),
A
Amit Arora 已提交
1794
					ext4_ext_get_actual_len(newext),
A
Alex Tomas 已提交
1795 1796 1797 1798 1799 1800 1801 1802
					nearex, len, nearex + 1, nearex + 2);
			memmove(nearex + 2, nearex + 1, len);
		}
		path[depth].p_ext = nearex + 1;
	} else {
		BUG_ON(newext->ee_block == nearex->ee_block);
		len = (EXT_MAX_EXTENT(eh) - nearex) * sizeof(struct ext4_extent);
		len = len < 0 ? 0 : len;
1803
		ext_debug("insert %d:%llu:[%d]%d before: nearest 0x%p, "
A
Alex Tomas 已提交
1804 1805
				"move %d from 0x%p to 0x%p\n",
				le32_to_cpu(newext->ee_block),
1806
				ext4_ext_pblock(newext),
1807
				ext4_ext_is_uninitialized(newext),
A
Amit Arora 已提交
1808
				ext4_ext_get_actual_len(newext),
1809
				nearex, len, nearex, nearex + 1);
A
Alex Tomas 已提交
1810 1811 1812 1813
		memmove(nearex + 1, nearex, len);
		path[depth].p_ext = nearex;
	}

M
Marcin Slusarz 已提交
1814
	le16_add_cpu(&eh->eh_entries, 1);
A
Alex Tomas 已提交
1815 1816
	nearex = path[depth].p_ext;
	nearex->ee_block = newext->ee_block;
1817
	ext4_ext_store_pblock(nearex, ext4_ext_pblock(newext));
A
Alex Tomas 已提交
1818 1819 1820 1821
	nearex->ee_len = newext->ee_len;

merge:
	/* try to merge extents to the right */
1822
	if (!(flag & EXT4_GET_BLOCKS_PRE_IO))
1823
		ext4_ext_try_to_merge(inode, path, nearex);
A
Alex Tomas 已提交
1824 1825 1826 1827 1828 1829 1830 1831 1832 1833 1834 1835 1836 1837 1838 1839 1840 1841 1842

	/* try to merge extents to the left */

	/* time to correct all indexes above */
	err = ext4_ext_correct_indexes(handle, inode, path);
	if (err)
		goto cleanup;

	err = ext4_ext_dirty(handle, inode, path + depth);

cleanup:
	if (npath) {
		ext4_ext_drop_refs(npath);
		kfree(npath);
	}
	ext4_ext_invalidate_cache(inode);
	return err;
}

1843 1844 1845
static int ext4_ext_walk_space(struct inode *inode, ext4_lblk_t block,
			       ext4_lblk_t num, ext_prepare_callback func,
			       void *cbdata)
1846 1847 1848 1849 1850 1851 1852 1853 1854 1855 1856
{
	struct ext4_ext_path *path = NULL;
	struct ext4_ext_cache cbex;
	struct ext4_extent *ex;
	ext4_lblk_t next, start = 0, end = 0;
	ext4_lblk_t last = block + num;
	int depth, exists, err = 0;

	BUG_ON(func == NULL);
	BUG_ON(inode == NULL);

1857
	while (block < last && block != EXT_MAX_BLOCKS) {
1858 1859
		num = last - block;
		/* find extent for this block */
1860
		down_read(&EXT4_I(inode)->i_data_sem);
1861
		path = ext4_ext_find_extent(inode, block, path);
1862
		up_read(&EXT4_I(inode)->i_data_sem);
1863 1864 1865 1866 1867 1868 1869
		if (IS_ERR(path)) {
			err = PTR_ERR(path);
			path = NULL;
			break;
		}

		depth = ext_depth(inode);
1870 1871 1872 1873 1874
		if (unlikely(path[depth].p_hdr == NULL)) {
			EXT4_ERROR_INODE(inode, "path[%d].p_hdr == NULL", depth);
			err = -EIO;
			break;
		}
1875 1876 1877 1878 1879 1880 1881 1882 1883 1884 1885 1886 1887 1888 1889 1890 1891 1892 1893 1894 1895 1896 1897 1898 1899 1900 1901 1902 1903 1904 1905 1906 1907 1908 1909 1910 1911 1912 1913 1914 1915 1916 1917 1918 1919
		ex = path[depth].p_ext;
		next = ext4_ext_next_allocated_block(path);

		exists = 0;
		if (!ex) {
			/* there is no extent yet, so try to allocate
			 * all requested space */
			start = block;
			end = block + num;
		} else if (le32_to_cpu(ex->ee_block) > block) {
			/* need to allocate space before found extent */
			start = block;
			end = le32_to_cpu(ex->ee_block);
			if (block + num < end)
				end = block + num;
		} else if (block >= le32_to_cpu(ex->ee_block)
					+ ext4_ext_get_actual_len(ex)) {
			/* need to allocate space after found extent */
			start = block;
			end = block + num;
			if (end >= next)
				end = next;
		} else if (block >= le32_to_cpu(ex->ee_block)) {
			/*
			 * some part of requested space is covered
			 * by found extent
			 */
			start = block;
			end = le32_to_cpu(ex->ee_block)
				+ ext4_ext_get_actual_len(ex);
			if (block + num < end)
				end = block + num;
			exists = 1;
		} else {
			BUG();
		}
		BUG_ON(end <= start);

		if (!exists) {
			cbex.ec_block = start;
			cbex.ec_len = end - start;
			cbex.ec_start = 0;
		} else {
			cbex.ec_block = le32_to_cpu(ex->ee_block);
			cbex.ec_len = ext4_ext_get_actual_len(ex);
1920
			cbex.ec_start = ext4_ext_pblock(ex);
1921 1922
		}

1923 1924 1925 1926 1927
		if (unlikely(cbex.ec_len == 0)) {
			EXT4_ERROR_INODE(inode, "cbex.ec_len == 0");
			err = -EIO;
			break;
		}
1928
		err = func(inode, next, &cbex, ex, cbdata);
1929 1930 1931 1932 1933 1934 1935 1936 1937 1938 1939 1940 1941 1942 1943 1944 1945 1946 1947 1948 1949 1950 1951 1952 1953 1954 1955 1956 1957
		ext4_ext_drop_refs(path);

		if (err < 0)
			break;

		if (err == EXT_REPEAT)
			continue;
		else if (err == EXT_BREAK) {
			err = 0;
			break;
		}

		if (ext_depth(inode) != depth) {
			/* depth was changed. we have to realloc path */
			kfree(path);
			path = NULL;
		}

		block = cbex.ec_block + cbex.ec_len;
	}

	if (path) {
		ext4_ext_drop_refs(path);
		kfree(path);
	}

	return err;
}

1958
static void
A
Aneesh Kumar K.V 已提交
1959
ext4_ext_put_in_cache(struct inode *inode, ext4_lblk_t block,
1960
			__u32 len, ext4_fsblk_t start)
A
Alex Tomas 已提交
1961 1962 1963
{
	struct ext4_ext_cache *cex;
	BUG_ON(len == 0);
1964
	spin_lock(&EXT4_I(inode)->i_block_reservation_lock);
1965
	trace_ext4_ext_put_in_cache(inode, block, len, start);
A
Alex Tomas 已提交
1966 1967 1968 1969
	cex = &EXT4_I(inode)->i_cached_extent;
	cex->ec_block = block;
	cex->ec_len = len;
	cex->ec_start = start;
1970
	spin_unlock(&EXT4_I(inode)->i_block_reservation_lock);
A
Alex Tomas 已提交
1971 1972 1973
}

/*
1974 1975
 * ext4_ext_put_gap_in_cache:
 * calculate boundaries of the gap that the requested block fits into
A
Alex Tomas 已提交
1976 1977
 * and cache this gap
 */
1978
static void
A
Alex Tomas 已提交
1979
ext4_ext_put_gap_in_cache(struct inode *inode, struct ext4_ext_path *path,
A
Aneesh Kumar K.V 已提交
1980
				ext4_lblk_t block)
A
Alex Tomas 已提交
1981 1982
{
	int depth = ext_depth(inode);
A
Aneesh Kumar K.V 已提交
1983 1984
	unsigned long len;
	ext4_lblk_t lblock;
A
Alex Tomas 已提交
1985 1986 1987 1988 1989 1990
	struct ext4_extent *ex;

	ex = path[depth].p_ext;
	if (ex == NULL) {
		/* there is no extent yet, so gap is [0;-] */
		lblock = 0;
1991
		len = EXT_MAX_BLOCKS;
A
Alex Tomas 已提交
1992 1993 1994 1995
		ext_debug("cache gap(whole file):");
	} else if (block < le32_to_cpu(ex->ee_block)) {
		lblock = block;
		len = le32_to_cpu(ex->ee_block) - block;
1996 1997 1998 1999
		ext_debug("cache gap(before): %u [%u:%u]",
				block,
				le32_to_cpu(ex->ee_block),
				 ext4_ext_get_actual_len(ex));
A
Alex Tomas 已提交
2000
	} else if (block >= le32_to_cpu(ex->ee_block)
A
Amit Arora 已提交
2001
			+ ext4_ext_get_actual_len(ex)) {
A
Aneesh Kumar K.V 已提交
2002
		ext4_lblk_t next;
D
Dave Kleikamp 已提交
2003
		lblock = le32_to_cpu(ex->ee_block)
A
Amit Arora 已提交
2004
			+ ext4_ext_get_actual_len(ex);
A
Aneesh Kumar K.V 已提交
2005 2006

		next = ext4_ext_next_allocated_block(path);
2007 2008 2009 2010
		ext_debug("cache gap(after): [%u:%u] %u",
				le32_to_cpu(ex->ee_block),
				ext4_ext_get_actual_len(ex),
				block);
A
Aneesh Kumar K.V 已提交
2011 2012
		BUG_ON(next == lblock);
		len = next - lblock;
A
Alex Tomas 已提交
2013 2014 2015 2016 2017
	} else {
		lblock = len = 0;
		BUG();
	}

2018
	ext_debug(" -> %u:%lu\n", lblock, len);
2019
	ext4_ext_put_in_cache(inode, lblock, len, 0);
A
Alex Tomas 已提交
2020 2021
}

2022
/*
2023
 * ext4_ext_check_cache()
2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035
 * Checks to see if the given block is in the cache.
 * If it is, the cached extent is stored in the given
 * cache extent pointer.  If the cached extent is a hole,
 * this routine should be used instead of
 * ext4_ext_in_cache if the calling function needs to
 * know the size of the hole.
 *
 * @inode: The files inode
 * @block: The block to look for in the cache
 * @ex:    Pointer where the cached extent will be stored
 *         if it contains block
 *
2036 2037
 * Return 0 if cache is invalid; 1 if the cache is valid
 */
2038 2039
static int ext4_ext_check_cache(struct inode *inode, ext4_lblk_t block,
	struct ext4_ext_cache *ex){
A
Alex Tomas 已提交
2040
	struct ext4_ext_cache *cex;
2041
	struct ext4_sb_info *sbi;
2042
	int ret = 0;
A
Alex Tomas 已提交
2043

2044
	/*
2045 2046 2047
	 * We borrow i_block_reservation_lock to protect i_cached_extent
	 */
	spin_lock(&EXT4_I(inode)->i_block_reservation_lock);
A
Alex Tomas 已提交
2048
	cex = &EXT4_I(inode)->i_cached_extent;
2049
	sbi = EXT4_SB(inode->i_sb);
A
Alex Tomas 已提交
2050 2051

	/* has cache valid data? */
2052
	if (cex->ec_len == 0)
2053
		goto errout;
A
Alex Tomas 已提交
2054

2055
	if (in_range(block, cex->ec_block, cex->ec_len)) {
2056
		memcpy(ex, cex, sizeof(struct ext4_ext_cache));
2057 2058 2059
		ext_debug("%u cached by %u:%u:%llu\n",
				block,
				cex->ec_block, cex->ec_len, cex->ec_start);
2060
		ret = 1;
A
Alex Tomas 已提交
2061
	}
2062
errout:
2063 2064 2065 2066
	if (!ret)
		sbi->extent_cache_misses++;
	else
		sbi->extent_cache_hits++;
2067
	trace_ext4_ext_in_cache(inode, block, ret);
2068 2069
	spin_unlock(&EXT4_I(inode)->i_block_reservation_lock);
	return ret;
A
Alex Tomas 已提交
2070 2071
}

2072 2073 2074 2075 2076 2077 2078 2079 2080 2081 2082 2083 2084 2085 2086 2087 2088 2089 2090 2091 2092 2093 2094 2095 2096 2097 2098 2099 2100 2101 2102
/*
 * ext4_ext_in_cache()
 * Checks to see if the given block is in the cache.
 * If it is, the cached extent is stored in the given
 * extent pointer.
 *
 * @inode: The files inode
 * @block: The block to look for in the cache
 * @ex:    Pointer where the cached extent will be stored
 *         if it contains block
 *
 * Return 0 if cache is invalid; 1 if the cache is valid
 */
static int
ext4_ext_in_cache(struct inode *inode, ext4_lblk_t block,
			struct ext4_extent *ex)
{
	struct ext4_ext_cache cex;
	int ret = 0;

	if (ext4_ext_check_cache(inode, block, &cex)) {
		ex->ee_block = cpu_to_le32(cex.ec_block);
		ext4_ext_store_pblock(ex, cex.ec_start);
		ex->ee_len = cpu_to_le16(cex.ec_len);
		ret = 1;
	}

	return ret;
}


A
Alex Tomas 已提交
2103
/*
2104 2105
 * ext4_ext_rm_idx:
 * removes index from the index block.
A
Alex Tomas 已提交
2106
 */
A
Aneesh Kumar K.V 已提交
2107
static int ext4_ext_rm_idx(handle_t *handle, struct inode *inode,
A
Alex Tomas 已提交
2108 2109 2110
			struct ext4_ext_path *path)
{
	int err;
2111
	ext4_fsblk_t leaf;
A
Alex Tomas 已提交
2112 2113 2114

	/* free index block */
	path--;
2115
	leaf = ext4_idx_pblock(path->p_idx);
2116 2117 2118 2119
	if (unlikely(path->p_hdr->eh_entries == 0)) {
		EXT4_ERROR_INODE(inode, "path->p_hdr->eh_entries == 0");
		return -EIO;
	}
2120 2121
	err = ext4_ext_get_access(handle, inode, path);
	if (err)
A
Alex Tomas 已提交
2122
		return err;
2123 2124 2125 2126 2127 2128 2129

	if (path->p_idx != EXT_LAST_INDEX(path->p_hdr)) {
		int len = EXT_LAST_INDEX(path->p_hdr) - path->p_idx;
		len *= sizeof(struct ext4_extent_idx);
		memmove(path->p_idx, path->p_idx + 1, len);
	}

M
Marcin Slusarz 已提交
2130
	le16_add_cpu(&path->p_hdr->eh_entries, -1);
2131 2132
	err = ext4_ext_dirty(handle, inode, path);
	if (err)
A
Alex Tomas 已提交
2133
		return err;
2134
	ext_debug("index is empty, remove it, free block %llu\n", leaf);
2135 2136
	trace_ext4_ext_rm_idx(inode, leaf);

2137
	ext4_free_blocks(handle, inode, NULL, leaf, 1,
2138
			 EXT4_FREE_BLOCKS_METADATA | EXT4_FREE_BLOCKS_FORGET);
A
Alex Tomas 已提交
2139 2140 2141 2142
	return err;
}

/*
2143 2144 2145 2146 2147
 * ext4_ext_calc_credits_for_single_extent:
 * This routine returns max. credits that needed to insert an extent
 * to the extent tree.
 * When pass the actual path, the caller should calculate credits
 * under i_data_sem.
A
Alex Tomas 已提交
2148
 */
2149
int ext4_ext_calc_credits_for_single_extent(struct inode *inode, int nrblocks,
A
Alex Tomas 已提交
2150 2151 2152
						struct ext4_ext_path *path)
{
	if (path) {
2153
		int depth = ext_depth(inode);
2154
		int ret = 0;
2155

A
Alex Tomas 已提交
2156 2157
		/* probably there is space in leaf? */
		if (le16_to_cpu(path[depth].p_hdr->eh_entries)
2158
				< le16_to_cpu(path[depth].p_hdr->eh_max)) {
A
Alex Tomas 已提交
2159

2160 2161 2162 2163 2164
			/*
			 *  There are some space in the leaf tree, no
			 *  need to account for leaf block credit
			 *
			 *  bitmaps and block group descriptor blocks
2165
			 *  and other metadata blocks still need to be
2166 2167
			 *  accounted.
			 */
2168
			/* 1 bitmap, 1 block group descriptor */
2169
			ret = 2 + EXT4_META_TRANS_BLOCKS(inode->i_sb);
2170
			return ret;
2171 2172
		}
	}
A
Alex Tomas 已提交
2173

2174
	return ext4_chunk_trans_blocks(inode, nrblocks);
2175
}
A
Alex Tomas 已提交
2176

2177 2178 2179 2180 2181 2182 2183 2184 2185 2186 2187
/*
 * How many index/leaf blocks need to change/allocate to modify nrblocks?
 *
 * if nrblocks are fit in a single extent (chunk flag is 1), then
 * in the worse case, each tree level index/leaf need to be changed
 * if the tree split due to insert a new extent, then the old tree
 * index/leaf need to be updated too
 *
 * If the nrblocks are discontiguous, they could cause
 * the whole tree split more than once, but this is really rare.
 */
2188
int ext4_ext_index_trans_blocks(struct inode *inode, int nrblocks, int chunk)
2189 2190 2191
{
	int index;
	int depth = ext_depth(inode);
A
Alex Tomas 已提交
2192

2193 2194 2195 2196
	if (chunk)
		index = depth * 2;
	else
		index = depth * 3;
A
Alex Tomas 已提交
2197

2198
	return index;
A
Alex Tomas 已提交
2199 2200 2201
}

static int ext4_remove_blocks(handle_t *handle, struct inode *inode,
2202 2203 2204
			      struct ext4_extent *ex,
			      ext4_fsblk_t *partial_cluster,
			      ext4_lblk_t from, ext4_lblk_t to)
A
Alex Tomas 已提交
2205
{
2206
	struct ext4_sb_info *sbi = EXT4_SB(inode->i_sb);
A
Amit Arora 已提交
2207
	unsigned short ee_len =  ext4_ext_get_actual_len(ex);
2208
	ext4_fsblk_t pblk;
2209
	int flags = EXT4_FREE_BLOCKS_FORGET;
A
Alex Tomas 已提交
2210

2211
	if (S_ISDIR(inode->i_mode) || S_ISLNK(inode->i_mode))
2212
		flags |= EXT4_FREE_BLOCKS_METADATA;
2213 2214 2215 2216 2217 2218 2219 2220 2221
	/*
	 * For bigalloc file systems, we never free a partial cluster
	 * at the beginning of the extent.  Instead, we make a note
	 * that we tried freeing the cluster, and check to see if we
	 * need to free it on a subsequent call to ext4_remove_blocks,
	 * or at the end of the ext4_truncate() operation.
	 */
	flags |= EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER;

2222
	trace_ext4_remove_blocks(inode, ex, from, to, *partial_cluster);
2223 2224 2225 2226 2227 2228 2229 2230 2231 2232 2233 2234 2235
	/*
	 * If we have a partial cluster, and it's different from the
	 * cluster of the last block, we need to explicitly free the
	 * partial cluster here.
	 */
	pblk = ext4_ext_pblock(ex) + ee_len - 1;
	if (*partial_cluster && (EXT4_B2C(sbi, pblk) != *partial_cluster)) {
		ext4_free_blocks(handle, inode, NULL,
				 EXT4_C2B(sbi, *partial_cluster),
				 sbi->s_cluster_ratio, flags);
		*partial_cluster = 0;
	}

A
Alex Tomas 已提交
2236 2237 2238 2239 2240 2241 2242 2243 2244 2245 2246 2247 2248 2249 2250 2251
#ifdef EXTENTS_STATS
	{
		struct ext4_sb_info *sbi = EXT4_SB(inode->i_sb);
		spin_lock(&sbi->s_ext_stats_lock);
		sbi->s_ext_blocks += ee_len;
		sbi->s_ext_extents++;
		if (ee_len < sbi->s_ext_min)
			sbi->s_ext_min = ee_len;
		if (ee_len > sbi->s_ext_max)
			sbi->s_ext_max = ee_len;
		if (ext_depth(inode) > sbi->s_depth_max)
			sbi->s_depth_max = ext_depth(inode);
		spin_unlock(&sbi->s_ext_stats_lock);
	}
#endif
	if (from >= le32_to_cpu(ex->ee_block)
A
Amit Arora 已提交
2252
	    && to == le32_to_cpu(ex->ee_block) + ee_len - 1) {
A
Alex Tomas 已提交
2253
		/* tail removal */
A
Aneesh Kumar K.V 已提交
2254 2255
		ext4_lblk_t num;

A
Amit Arora 已提交
2256
		num = le32_to_cpu(ex->ee_block) + ee_len - from;
2257 2258 2259 2260 2261 2262 2263 2264 2265 2266 2267 2268 2269 2270 2271 2272
		pblk = ext4_ext_pblock(ex) + ee_len - num;
		ext_debug("free last %u blocks starting %llu\n", num, pblk);
		ext4_free_blocks(handle, inode, NULL, pblk, num, flags);
		/*
		 * If the block range to be freed didn't start at the
		 * beginning of a cluster, and we removed the entire
		 * extent, save the partial cluster here, since we
		 * might need to delete if we determine that the
		 * truncate operation has removed all of the blocks in
		 * the cluster.
		 */
		if (pblk & (sbi->s_cluster_ratio - 1) &&
		    (ee_len == num))
			*partial_cluster = EXT4_B2C(sbi, pblk);
		else
			*partial_cluster = 0;
A
Alex Tomas 已提交
2273
	} else if (from == le32_to_cpu(ex->ee_block)
A
Amit Arora 已提交
2274
		   && to <= le32_to_cpu(ex->ee_block) + ee_len - 1) {
A
Allison Henderson 已提交
2275 2276 2277 2278 2279 2280 2281 2282
		/* head removal */
		ext4_lblk_t num;
		ext4_fsblk_t start;

		num = to - from;
		start = ext4_ext_pblock(ex);

		ext_debug("free first %u blocks starting %llu\n", num, start);
2283
		ext4_free_blocks(handle, inode, NULL, start, num, flags);
A
Allison Henderson 已提交
2284

A
Alex Tomas 已提交
2285
	} else {
A
Aneesh Kumar K.V 已提交
2286 2287 2288
		printk(KERN_INFO "strange request: removal(2) "
				"%u-%u from %u:%u\n",
				from, to, le32_to_cpu(ex->ee_block), ee_len);
A
Alex Tomas 已提交
2289 2290 2291 2292
	}
	return 0;
}

A
Allison Henderson 已提交
2293 2294 2295 2296 2297 2298 2299 2300 2301 2302 2303 2304

/*
 * ext4_ext_rm_leaf() Removes the extents associated with the
 * blocks appearing between "start" and "end", and splits the extents
 * if "start" and "end" appear in the same extent
 *
 * @handle: The journal handle
 * @inode:  The files inode
 * @path:   The path to the leaf
 * @start:  The first block to remove
 * @end:   The last block to remove
 */
A
Alex Tomas 已提交
2305 2306
static int
ext4_ext_rm_leaf(handle_t *handle, struct inode *inode,
2307 2308
		 struct ext4_ext_path *path, ext4_fsblk_t *partial_cluster,
		 ext4_lblk_t start, ext4_lblk_t end)
A
Alex Tomas 已提交
2309
{
2310
	struct ext4_sb_info *sbi = EXT4_SB(inode->i_sb);
A
Alex Tomas 已提交
2311 2312 2313
	int err = 0, correct_index = 0;
	int depth = ext_depth(inode), credits;
	struct ext4_extent_header *eh;
2314
	ext4_lblk_t a, b;
A
Aneesh Kumar K.V 已提交
2315 2316
	unsigned num;
	ext4_lblk_t ex_ee_block;
A
Alex Tomas 已提交
2317
	unsigned short ex_ee_len;
A
Amit Arora 已提交
2318
	unsigned uninitialized = 0;
A
Alex Tomas 已提交
2319 2320
	struct ext4_extent *ex;

2321
	/* the header must be checked already in ext4_ext_remove_space() */
A
Aneesh Kumar K.V 已提交
2322
	ext_debug("truncate since %u in leaf\n", start);
A
Alex Tomas 已提交
2323 2324 2325
	if (!path[depth].p_hdr)
		path[depth].p_hdr = ext_block_hdr(path[depth].p_bh);
	eh = path[depth].p_hdr;
2326 2327 2328 2329
	if (unlikely(path[depth].p_hdr == NULL)) {
		EXT4_ERROR_INODE(inode, "path[%d].p_hdr == NULL", depth);
		return -EIO;
	}
A
Alex Tomas 已提交
2330 2331 2332 2333
	/* find where to start removing */
	ex = EXT_LAST_EXTENT(eh);

	ex_ee_block = le32_to_cpu(ex->ee_block);
A
Amit Arora 已提交
2334
	ex_ee_len = ext4_ext_get_actual_len(ex);
A
Alex Tomas 已提交
2335

2336 2337
	trace_ext4_ext_rm_leaf(inode, start, ex, *partial_cluster);

A
Alex Tomas 已提交
2338 2339
	while (ex >= EXT_FIRST_EXTENT(eh) &&
			ex_ee_block + ex_ee_len > start) {
2340 2341 2342 2343 2344 2345

		if (ext4_ext_is_uninitialized(ex))
			uninitialized = 1;
		else
			uninitialized = 0;

2346 2347
		ext_debug("remove ext %u:[%d]%d\n", ex_ee_block,
			 uninitialized, ex_ee_len);
A
Alex Tomas 已提交
2348 2349 2350
		path[depth].p_ext = ex;

		a = ex_ee_block > start ? ex_ee_block : start;
A
Allison Henderson 已提交
2351 2352
		b = ex_ee_block+ex_ee_len - 1 < end ?
			ex_ee_block+ex_ee_len - 1 : end;
A
Alex Tomas 已提交
2353 2354 2355

		ext_debug("  border %u:%u\n", a, b);

A
Allison Henderson 已提交
2356 2357 2358 2359 2360 2361
		/* If this extent is beyond the end of the hole, skip it */
		if (end <= ex_ee_block) {
			ex--;
			ex_ee_block = le32_to_cpu(ex->ee_block);
			ex_ee_len = ext4_ext_get_actual_len(ex);
			continue;
2362 2363 2364 2365 2366
		} else if (b != ex_ee_block + ex_ee_len - 1) {
			EXT4_ERROR_INODE(inode,"  bad truncate %u:%u\n",
					 start, end);
			err = -EIO;
			goto out;
A
Alex Tomas 已提交
2367 2368
		} else if (a != ex_ee_block) {
			/* remove tail of the extent */
2369
			num = a - ex_ee_block;
A
Alex Tomas 已提交
2370 2371 2372 2373
		} else {
			/* remove whole extent: excellent! */
			num = 0;
		}
2374 2375 2376 2377 2378 2379 2380
		/*
		 * 3 for leaf, sb, and inode plus 2 (bmap and group
		 * descriptor) for each block group; assume two block
		 * groups plus ex_ee_len/blocks_per_block_group for
		 * the worst case
		 */
		credits = 7 + 2*(ex_ee_len/EXT4_BLOCKS_PER_GROUP(inode->i_sb));
A
Alex Tomas 已提交
2381 2382 2383 2384
		if (ex == EXT_FIRST_EXTENT(eh)) {
			correct_index = 1;
			credits += (ext_depth(inode)) + 1;
		}
D
Dmitry Monakhov 已提交
2385
		credits += EXT4_MAXQUOTAS_TRANS_BLOCKS(inode->i_sb);
A
Alex Tomas 已提交
2386

2387
		err = ext4_ext_truncate_extend_restart(handle, inode, credits);
2388
		if (err)
A
Alex Tomas 已提交
2389 2390 2391 2392 2393 2394
			goto out;

		err = ext4_ext_get_access(handle, inode, path + depth);
		if (err)
			goto out;

2395 2396
		err = ext4_remove_blocks(handle, inode, ex, partial_cluster,
					 a, b);
A
Alex Tomas 已提交
2397 2398 2399
		if (err)
			goto out;

2400
		if (num == 0)
2401
			/* this extent is removed; mark slot entirely unused */
2402
			ext4_ext_store_pblock(ex, 0);
A
Alex Tomas 已提交
2403 2404

		ex->ee_len = cpu_to_le16(num);
2405 2406 2407 2408 2409
		/*
		 * Do not mark uninitialized if all the blocks in the
		 * extent have been removed.
		 */
		if (uninitialized && num)
A
Amit Arora 已提交
2410
			ext4_ext_mark_uninitialized(ex);
A
Allison Henderson 已提交
2411 2412 2413 2414 2415
		/*
		 * If the extent was completely released,
		 * we need to remove it from the leaf
		 */
		if (num == 0) {
2416
			if (end != EXT_MAX_BLOCKS - 1) {
A
Allison Henderson 已提交
2417 2418 2419 2420 2421 2422 2423 2424 2425 2426 2427 2428 2429
				/*
				 * For hole punching, we need to scoot all the
				 * extents up when an extent is removed so that
				 * we dont have blank extents in the middle
				 */
				memmove(ex, ex+1, (EXT_LAST_EXTENT(eh) - ex) *
					sizeof(struct ext4_extent));

				/* Now get rid of the one at the end */
				memset(EXT_LAST_EXTENT(eh), 0,
					sizeof(struct ext4_extent));
			}
			le16_add_cpu(&eh->eh_entries, -1);
2430 2431
		} else
			*partial_cluster = 0;
A
Allison Henderson 已提交
2432

2433 2434 2435 2436
		err = ext4_ext_dirty(handle, inode, path + depth);
		if (err)
			goto out;

2437
		ext_debug("new extent: %u:%u:%llu\n", block, num,
2438
				ext4_ext_pblock(ex));
A
Alex Tomas 已提交
2439 2440
		ex--;
		ex_ee_block = le32_to_cpu(ex->ee_block);
A
Amit Arora 已提交
2441
		ex_ee_len = ext4_ext_get_actual_len(ex);
A
Alex Tomas 已提交
2442 2443 2444 2445 2446
	}

	if (correct_index && eh->eh_entries)
		err = ext4_ext_correct_indexes(handle, inode, path);

2447 2448 2449 2450 2451 2452 2453 2454 2455 2456 2457 2458 2459 2460 2461 2462 2463 2464 2465
	/*
	 * If there is still a entry in the leaf node, check to see if
	 * it references the partial cluster.  This is the only place
	 * where it could; if it doesn't, we can free the cluster.
	 */
	if (*partial_cluster && ex >= EXT_FIRST_EXTENT(eh) &&
	    (EXT4_B2C(sbi, ext4_ext_pblock(ex) + ex_ee_len - 1) !=
	     *partial_cluster)) {
		int flags = EXT4_FREE_BLOCKS_FORGET;

		if (S_ISDIR(inode->i_mode) || S_ISLNK(inode->i_mode))
			flags |= EXT4_FREE_BLOCKS_METADATA;

		ext4_free_blocks(handle, inode, NULL,
				 EXT4_C2B(sbi, *partial_cluster),
				 sbi->s_cluster_ratio, flags);
		*partial_cluster = 0;
	}

A
Alex Tomas 已提交
2466 2467 2468 2469 2470 2471 2472 2473 2474 2475
	/* if this leaf is free, then we should
	 * remove it from index block above */
	if (err == 0 && eh->eh_entries == 0 && path[depth].p_bh != NULL)
		err = ext4_ext_rm_idx(handle, inode, path + depth);

out:
	return err;
}

/*
2476 2477
 * ext4_ext_more_to_rm:
 * returns 1 if current index has to be freed (even partial)
A
Alex Tomas 已提交
2478
 */
2479
static int
A
Alex Tomas 已提交
2480 2481 2482 2483 2484 2485 2486 2487
ext4_ext_more_to_rm(struct ext4_ext_path *path)
{
	BUG_ON(path->p_idx == NULL);

	if (path->p_idx < EXT_FIRST_INDEX(path->p_hdr))
		return 0;

	/*
2488
	 * if truncate on deeper level happened, it wasn't partial,
A
Alex Tomas 已提交
2489 2490 2491 2492 2493 2494 2495
	 * so we have to consider current index for truncation
	 */
	if (le16_to_cpu(path->p_hdr->eh_entries) == path->p_block)
		return 0;
	return 1;
}

2496
static int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start)
A
Alex Tomas 已提交
2497 2498 2499 2500
{
	struct super_block *sb = inode->i_sb;
	int depth = ext_depth(inode);
	struct ext4_ext_path *path;
2501
	ext4_fsblk_t partial_cluster = 0;
A
Alex Tomas 已提交
2502
	handle_t *handle;
2503
	int i, err;
A
Alex Tomas 已提交
2504

A
Aneesh Kumar K.V 已提交
2505
	ext_debug("truncate since %u\n", start);
A
Alex Tomas 已提交
2506 2507 2508 2509 2510 2511

	/* probably first extent we're gonna free will be last in block */
	handle = ext4_journal_start(inode, depth + 1);
	if (IS_ERR(handle))
		return PTR_ERR(handle);

2512
again:
A
Alex Tomas 已提交
2513 2514
	ext4_ext_invalidate_cache(inode);

2515 2516
	trace_ext4_ext_remove_space(inode, start, depth);

A
Alex Tomas 已提交
2517
	/*
2518 2519
	 * We start scanning from right side, freeing all the blocks
	 * after i_size and walking into the tree depth-wise.
A
Alex Tomas 已提交
2520
	 */
2521
	depth = ext_depth(inode);
2522
	path = kzalloc(sizeof(struct ext4_ext_path) * (depth + 1), GFP_NOFS);
A
Alex Tomas 已提交
2523 2524 2525 2526
	if (path == NULL) {
		ext4_journal_stop(handle);
		return -ENOMEM;
	}
2527
	path[0].p_depth = depth;
A
Alex Tomas 已提交
2528
	path[0].p_hdr = ext_inode_hdr(inode);
2529
	if (ext4_ext_check(inode, path[0].p_hdr, depth)) {
A
Alex Tomas 已提交
2530 2531 2532
		err = -EIO;
		goto out;
	}
2533
	i = err = 0;
A
Alex Tomas 已提交
2534 2535 2536 2537

	while (i >= 0 && err == 0) {
		if (i == depth) {
			/* this is leaf block */
A
Allison Henderson 已提交
2538
			err = ext4_ext_rm_leaf(handle, inode, path,
2539 2540
					       &partial_cluster, start,
					       EXT_MAX_BLOCKS - 1);
2541
			/* root level has p_bh == NULL, brelse() eats this */
A
Alex Tomas 已提交
2542 2543 2544 2545 2546 2547 2548 2549 2550 2551 2552 2553 2554
			brelse(path[i].p_bh);
			path[i].p_bh = NULL;
			i--;
			continue;
		}

		/* this is index block */
		if (!path[i].p_hdr) {
			ext_debug("initialize header\n");
			path[i].p_hdr = ext_block_hdr(path[i].p_bh);
		}

		if (!path[i].p_idx) {
2555
			/* this level hasn't been touched yet */
A
Alex Tomas 已提交
2556 2557 2558 2559 2560 2561
			path[i].p_idx = EXT_LAST_INDEX(path[i].p_hdr);
			path[i].p_block = le16_to_cpu(path[i].p_hdr->eh_entries)+1;
			ext_debug("init index ptr: hdr 0x%p, num %d\n",
				  path[i].p_hdr,
				  le16_to_cpu(path[i].p_hdr->eh_entries));
		} else {
2562
			/* we were already here, see at next index */
A
Alex Tomas 已提交
2563 2564 2565 2566 2567 2568 2569
			path[i].p_idx--;
		}

		ext_debug("level %d - index, first 0x%p, cur 0x%p\n",
				i, EXT_FIRST_INDEX(path[i].p_hdr),
				path[i].p_idx);
		if (ext4_ext_more_to_rm(path + i)) {
2570
			struct buffer_head *bh;
A
Alex Tomas 已提交
2571
			/* go to the next level */
2572
			ext_debug("move to level %d (block %llu)\n",
2573
				  i + 1, ext4_idx_pblock(path[i].p_idx));
A
Alex Tomas 已提交
2574
			memset(path + i + 1, 0, sizeof(*path));
2575
			bh = sb_bread(sb, ext4_idx_pblock(path[i].p_idx));
2576
			if (!bh) {
A
Alex Tomas 已提交
2577 2578 2579 2580
				/* should we reset i_size? */
				err = -EIO;
				break;
			}
2581 2582 2583 2584
			if (WARN_ON(i + 1 > depth)) {
				err = -EIO;
				break;
			}
2585
			if (ext4_ext_check(inode, ext_block_hdr(bh),
2586 2587 2588 2589 2590
							depth - i - 1)) {
				err = -EIO;
				break;
			}
			path[i + 1].p_bh = bh;
A
Alex Tomas 已提交
2591

2592 2593
			/* save actual number of indexes since this
			 * number is changed at the next iteration */
A
Alex Tomas 已提交
2594 2595 2596
			path[i].p_block = le16_to_cpu(path[i].p_hdr->eh_entries);
			i++;
		} else {
2597
			/* we finished processing this index, go up */
A
Alex Tomas 已提交
2598
			if (path[i].p_hdr->eh_entries == 0 && i > 0) {
2599
				/* index is empty, remove it;
A
Alex Tomas 已提交
2600 2601 2602 2603
				 * handle must be already prepared by the
				 * truncatei_leaf() */
				err = ext4_ext_rm_idx(handle, inode, path + i);
			}
2604
			/* root level has p_bh == NULL, brelse() eats this */
A
Alex Tomas 已提交
2605 2606 2607 2608 2609 2610 2611
			brelse(path[i].p_bh);
			path[i].p_bh = NULL;
			i--;
			ext_debug("return to level %d\n", i);
		}
	}

2612 2613 2614
	trace_ext4_ext_remove_space_done(inode, start, depth, partial_cluster,
			path->p_hdr->eh_entries);

2615 2616 2617 2618 2619 2620 2621 2622 2623 2624 2625 2626 2627 2628 2629
	/* If we still have something in the partial cluster and we have removed
	 * even the first extent, then we should free the blocks in the partial
	 * cluster as well. */
	if (partial_cluster && path->p_hdr->eh_entries == 0) {
		int flags = EXT4_FREE_BLOCKS_FORGET;

		if (S_ISDIR(inode->i_mode) || S_ISLNK(inode->i_mode))
			flags |= EXT4_FREE_BLOCKS_METADATA;

		ext4_free_blocks(handle, inode, NULL,
				 EXT4_C2B(EXT4_SB(sb), partial_cluster),
				 EXT4_SB(sb)->s_cluster_ratio, flags);
		partial_cluster = 0;
	}

A
Alex Tomas 已提交
2630 2631 2632
	/* TODO: flexible tree reduction should be here */
	if (path->p_hdr->eh_entries == 0) {
		/*
2633 2634
		 * truncate to zero freed all the tree,
		 * so we need to correct eh_depth
A
Alex Tomas 已提交
2635 2636 2637 2638 2639
		 */
		err = ext4_ext_get_access(handle, inode, path);
		if (err == 0) {
			ext_inode_hdr(inode)->eh_depth = 0;
			ext_inode_hdr(inode)->eh_max =
2640
				cpu_to_le16(ext4_ext_space_root(inode, 0));
A
Alex Tomas 已提交
2641 2642 2643 2644 2645 2646
			err = ext4_ext_dirty(handle, inode, path);
		}
	}
out:
	ext4_ext_drop_refs(path);
	kfree(path);
2647 2648
	if (err == -EAGAIN)
		goto again;
A
Alex Tomas 已提交
2649 2650 2651 2652 2653 2654 2655 2656 2657 2658 2659 2660 2661 2662
	ext4_journal_stop(handle);

	return err;
}

/*
 * called at mount time
 */
void ext4_ext_init(struct super_block *sb)
{
	/*
	 * possible initialization would be here
	 */

2663
	if (EXT4_HAS_INCOMPAT_FEATURE(sb, EXT4_FEATURE_INCOMPAT_EXTENTS)) {
2664
#if defined(AGGRESSIVE_TEST) || defined(CHECK_BINSEARCH) || defined(EXTENTS_STATS)
2665
		printk(KERN_INFO "EXT4-fs: file extents enabled");
2666 2667
#ifdef AGGRESSIVE_TEST
		printk(", aggressive tests");
A
Alex Tomas 已提交
2668 2669 2670 2671 2672 2673 2674 2675
#endif
#ifdef CHECK_BINSEARCH
		printk(", check binsearch");
#endif
#ifdef EXTENTS_STATS
		printk(", stats");
#endif
		printk("\n");
2676
#endif
A
Alex Tomas 已提交
2677 2678 2679 2680 2681 2682 2683 2684 2685 2686 2687 2688 2689
#ifdef EXTENTS_STATS
		spin_lock_init(&EXT4_SB(sb)->s_ext_stats_lock);
		EXT4_SB(sb)->s_ext_min = 1 << 30;
		EXT4_SB(sb)->s_ext_max = 0;
#endif
	}
}

/*
 * called at umount time
 */
void ext4_ext_release(struct super_block *sb)
{
2690
	if (!EXT4_HAS_INCOMPAT_FEATURE(sb, EXT4_FEATURE_INCOMPAT_EXTENTS))
A
Alex Tomas 已提交
2691 2692 2693 2694 2695 2696 2697 2698 2699 2700 2701 2702 2703 2704
		return;

#ifdef EXTENTS_STATS
	if (EXT4_SB(sb)->s_ext_blocks && EXT4_SB(sb)->s_ext_extents) {
		struct ext4_sb_info *sbi = EXT4_SB(sb);
		printk(KERN_ERR "EXT4-fs: %lu blocks in %lu extents (%lu ave)\n",
			sbi->s_ext_blocks, sbi->s_ext_extents,
			sbi->s_ext_blocks / sbi->s_ext_extents);
		printk(KERN_ERR "EXT4-fs: extents: %lu min, %lu max, max depth %lu\n",
			sbi->s_ext_min, sbi->s_ext_max, sbi->s_depth_max);
	}
#endif
}

2705 2706 2707
/* FIXME!! we need to try to merge to left or right after zero-out  */
static int ext4_ext_zeroout(struct inode *inode, struct ext4_extent *ex)
{
2708 2709
	ext4_fsblk_t ee_pblock;
	unsigned int ee_len;
2710
	int ret;
2711 2712

	ee_len    = ext4_ext_get_actual_len(ex);
2713
	ee_pblock = ext4_ext_pblock(ex);
2714

2715
	ret = sb_issue_zeroout(inode->i_sb, ee_pblock, ee_len, GFP_NOFS);
2716 2717
	if (ret > 0)
		ret = 0;
2718

2719
	return ret;
2720 2721
}

2722 2723 2724 2725 2726 2727 2728 2729 2730 2731 2732 2733 2734 2735 2736 2737 2738 2739 2740 2741 2742 2743 2744 2745 2746 2747 2748 2749 2750 2751 2752 2753 2754 2755 2756 2757 2758 2759 2760 2761 2762 2763 2764 2765 2766 2767 2768 2769 2770 2771 2772 2773 2774 2775 2776 2777 2778 2779 2780 2781 2782 2783 2784 2785 2786 2787 2788 2789 2790 2791 2792 2793 2794 2795 2796 2797 2798 2799 2800 2801 2802 2803 2804 2805 2806 2807 2808 2809 2810 2811 2812 2813 2814 2815 2816 2817 2818 2819 2820 2821 2822 2823 2824 2825 2826 2827 2828 2829 2830 2831 2832 2833 2834 2835 2836 2837 2838 2839 2840 2841 2842 2843 2844 2845 2846 2847 2848 2849 2850 2851 2852 2853 2854 2855 2856 2857 2858 2859 2860 2861 2862 2863 2864 2865 2866 2867 2868 2869 2870 2871 2872 2873 2874 2875 2876 2877 2878 2879 2880 2881 2882 2883
/*
 * used by extent splitting.
 */
#define EXT4_EXT_MAY_ZEROOUT	0x1  /* safe to zeroout if split fails \
					due to ENOSPC */
#define EXT4_EXT_MARK_UNINIT1	0x2  /* mark first half uninitialized */
#define EXT4_EXT_MARK_UNINIT2	0x4  /* mark second half uninitialized */

/*
 * ext4_split_extent_at() splits an extent at given block.
 *
 * @handle: the journal handle
 * @inode: the file inode
 * @path: the path to the extent
 * @split: the logical block where the extent is splitted.
 * @split_flags: indicates if the extent could be zeroout if split fails, and
 *		 the states(init or uninit) of new extents.
 * @flags: flags used to insert new extent to extent tree.
 *
 *
 * Splits extent [a, b] into two extents [a, @split) and [@split, b], states
 * of which are deterimined by split_flag.
 *
 * There are two cases:
 *  a> the extent are splitted into two extent.
 *  b> split is not needed, and just mark the extent.
 *
 * return 0 on success.
 */
static int ext4_split_extent_at(handle_t *handle,
			     struct inode *inode,
			     struct ext4_ext_path *path,
			     ext4_lblk_t split,
			     int split_flag,
			     int flags)
{
	ext4_fsblk_t newblock;
	ext4_lblk_t ee_block;
	struct ext4_extent *ex, newex, orig_ex;
	struct ext4_extent *ex2 = NULL;
	unsigned int ee_len, depth;
	int err = 0;

	ext_debug("ext4_split_extents_at: inode %lu, logical"
		"block %llu\n", inode->i_ino, (unsigned long long)split);

	ext4_ext_show_leaf(inode, path);

	depth = ext_depth(inode);
	ex = path[depth].p_ext;
	ee_block = le32_to_cpu(ex->ee_block);
	ee_len = ext4_ext_get_actual_len(ex);
	newblock = split - ee_block + ext4_ext_pblock(ex);

	BUG_ON(split < ee_block || split >= (ee_block + ee_len));

	err = ext4_ext_get_access(handle, inode, path + depth);
	if (err)
		goto out;

	if (split == ee_block) {
		/*
		 * case b: block @split is the block that the extent begins with
		 * then we just change the state of the extent, and splitting
		 * is not needed.
		 */
		if (split_flag & EXT4_EXT_MARK_UNINIT2)
			ext4_ext_mark_uninitialized(ex);
		else
			ext4_ext_mark_initialized(ex);

		if (!(flags & EXT4_GET_BLOCKS_PRE_IO))
			ext4_ext_try_to_merge(inode, path, ex);

		err = ext4_ext_dirty(handle, inode, path + depth);
		goto out;
	}

	/* case a */
	memcpy(&orig_ex, ex, sizeof(orig_ex));
	ex->ee_len = cpu_to_le16(split - ee_block);
	if (split_flag & EXT4_EXT_MARK_UNINIT1)
		ext4_ext_mark_uninitialized(ex);

	/*
	 * path may lead to new leaf, not to original leaf any more
	 * after ext4_ext_insert_extent() returns,
	 */
	err = ext4_ext_dirty(handle, inode, path + depth);
	if (err)
		goto fix_extent_len;

	ex2 = &newex;
	ex2->ee_block = cpu_to_le32(split);
	ex2->ee_len   = cpu_to_le16(ee_len - (split - ee_block));
	ext4_ext_store_pblock(ex2, newblock);
	if (split_flag & EXT4_EXT_MARK_UNINIT2)
		ext4_ext_mark_uninitialized(ex2);

	err = ext4_ext_insert_extent(handle, inode, path, &newex, flags);
	if (err == -ENOSPC && (EXT4_EXT_MAY_ZEROOUT & split_flag)) {
		err = ext4_ext_zeroout(inode, &orig_ex);
		if (err)
			goto fix_extent_len;
		/* update the extent length and mark as initialized */
		ex->ee_len = cpu_to_le32(ee_len);
		ext4_ext_try_to_merge(inode, path, ex);
		err = ext4_ext_dirty(handle, inode, path + depth);
		goto out;
	} else if (err)
		goto fix_extent_len;

out:
	ext4_ext_show_leaf(inode, path);
	return err;

fix_extent_len:
	ex->ee_len = orig_ex.ee_len;
	ext4_ext_dirty(handle, inode, path + depth);
	return err;
}

/*
 * ext4_split_extents() splits an extent and mark extent which is covered
 * by @map as split_flags indicates
 *
 * It may result in splitting the extent into multiple extents (upto three)
 * There are three possibilities:
 *   a> There is no split required
 *   b> Splits in two extents: Split is happening at either end of the extent
 *   c> Splits in three extents: Somone is splitting in middle of the extent
 *
 */
static int ext4_split_extent(handle_t *handle,
			      struct inode *inode,
			      struct ext4_ext_path *path,
			      struct ext4_map_blocks *map,
			      int split_flag,
			      int flags)
{
	ext4_lblk_t ee_block;
	struct ext4_extent *ex;
	unsigned int ee_len, depth;
	int err = 0;
	int uninitialized;
	int split_flag1, flags1;

	depth = ext_depth(inode);
	ex = path[depth].p_ext;
	ee_block = le32_to_cpu(ex->ee_block);
	ee_len = ext4_ext_get_actual_len(ex);
	uninitialized = ext4_ext_is_uninitialized(ex);

	if (map->m_lblk + map->m_len < ee_block + ee_len) {
		split_flag1 = split_flag & EXT4_EXT_MAY_ZEROOUT ?
			      EXT4_EXT_MAY_ZEROOUT : 0;
		flags1 = flags | EXT4_GET_BLOCKS_PRE_IO;
		if (uninitialized)
			split_flag1 |= EXT4_EXT_MARK_UNINIT1 |
				       EXT4_EXT_MARK_UNINIT2;
		err = ext4_split_extent_at(handle, inode, path,
				map->m_lblk + map->m_len, split_flag1, flags1);
2884 2885
		if (err)
			goto out;
2886 2887 2888 2889 2890 2891 2892 2893 2894 2895 2896 2897 2898 2899 2900 2901 2902 2903 2904 2905 2906 2907 2908 2909 2910
	}

	ext4_ext_drop_refs(path);
	path = ext4_ext_find_extent(inode, map->m_lblk, path);
	if (IS_ERR(path))
		return PTR_ERR(path);

	if (map->m_lblk >= ee_block) {
		split_flag1 = split_flag & EXT4_EXT_MAY_ZEROOUT ?
			      EXT4_EXT_MAY_ZEROOUT : 0;
		if (uninitialized)
			split_flag1 |= EXT4_EXT_MARK_UNINIT1;
		if (split_flag & EXT4_EXT_MARK_UNINIT2)
			split_flag1 |= EXT4_EXT_MARK_UNINIT2;
		err = ext4_split_extent_at(handle, inode, path,
				map->m_lblk, split_flag1, flags);
		if (err)
			goto out;
	}

	ext4_ext_show_leaf(inode, path);
out:
	return err ? err : map->m_len;
}

2911
#define EXT4_EXT_ZERO_LEN 7
2912
/*
2913
 * This function is called by ext4_ext_map_blocks() if someone tries to write
2914
 * to an uninitialized extent. It may result in splitting the uninitialized
L
Lucas De Marchi 已提交
2915
 * extent into multiple extents (up to three - one initialized and two
2916 2917 2918 2919 2920 2921
 * uninitialized).
 * There are three possibilities:
 *   a> There is no split required: Entire extent should be initialized
 *   b> Splits in two extents: Write is happening at either end of the extent
 *   c> Splits in three extents: Somone is writing in middle of the extent
 */
A
Aneesh Kumar K.V 已提交
2922
static int ext4_ext_convert_to_initialized(handle_t *handle,
2923 2924 2925
					   struct inode *inode,
					   struct ext4_map_blocks *map,
					   struct ext4_ext_path *path)
2926
{
2927 2928 2929
	struct ext4_map_blocks split_map;
	struct ext4_extent zero_ex;
	struct ext4_extent *ex;
2930
	ext4_lblk_t ee_block, eof_block;
A
Aneesh Kumar K.V 已提交
2931
	unsigned int allocated, ee_len, depth;
2932
	int err = 0;
2933
	int split_flag = 0;
2934 2935 2936

	ext_debug("ext4_ext_convert_to_initialized: inode %lu, logical"
		"block %llu, max_blocks %u\n", inode->i_ino,
2937
		(unsigned long long)map->m_lblk, map->m_len);
2938 2939 2940

	eof_block = (inode->i_size + inode->i_sb->s_blocksize - 1) >>
		inode->i_sb->s_blocksize_bits;
2941 2942
	if (eof_block < map->m_lblk + map->m_len)
		eof_block = map->m_lblk + map->m_len;
2943 2944 2945 2946 2947

	depth = ext_depth(inode);
	ex = path[depth].p_ext;
	ee_block = le32_to_cpu(ex->ee_block);
	ee_len = ext4_ext_get_actual_len(ex);
2948
	allocated = ee_len - (map->m_lblk - ee_block);
2949

2950
	WARN_ON(map->m_lblk < ee_block);
2951 2952 2953 2954
	/*
	 * It is safe to convert extent to initialized via explicit
	 * zeroout only if extent is fully insde i_size or new_size.
	 */
2955
	split_flag |= ee_block + ee_len <= eof_block ? EXT4_EXT_MAY_ZEROOUT : 0;
2956

2957
	/* If extent has less than 2*EXT4_EXT_ZERO_LEN zerout directly */
2958 2959 2960
	if (ee_len <= 2*EXT4_EXT_ZERO_LEN &&
	    (EXT4_EXT_MAY_ZEROOUT & split_flag)) {
		err = ext4_ext_zeroout(inode, ex);
2961
		if (err)
2962 2963 2964 2965 2966
			goto out;

		err = ext4_ext_get_access(handle, inode, path + depth);
		if (err)
			goto out;
2967 2968 2969 2970
		ext4_ext_mark_initialized(ex);
		ext4_ext_try_to_merge(inode, path, ex);
		err = ext4_ext_dirty(handle, inode, path + depth);
		goto out;
2971
	}
2972

2973
	/*
2974 2975 2976 2977 2978
	 * four cases:
	 * 1. split the extent into three extents.
	 * 2. split the extent into two extents, zeroout the first half.
	 * 3. split the extent into two extents, zeroout the second half.
	 * 4. split the extent into two extents with out zeroout.
2979
	 */
2980 2981 2982 2983 2984 2985 2986 2987
	split_map.m_lblk = map->m_lblk;
	split_map.m_len = map->m_len;

	if (allocated > map->m_len) {
		if (allocated <= EXT4_EXT_ZERO_LEN &&
		    (EXT4_EXT_MAY_ZEROOUT & split_flag)) {
			/* case 3 */
			zero_ex.ee_block =
2988 2989
					 cpu_to_le32(map->m_lblk);
			zero_ex.ee_len = cpu_to_le16(allocated);
2990 2991 2992
			ext4_ext_store_pblock(&zero_ex,
				ext4_ext_pblock(ex) + map->m_lblk - ee_block);
			err = ext4_ext_zeroout(inode, &zero_ex);
2993 2994
			if (err)
				goto out;
2995 2996 2997 2998 2999 3000 3001 3002 3003 3004 3005 3006 3007 3008 3009 3010 3011 3012
			split_map.m_lblk = map->m_lblk;
			split_map.m_len = allocated;
		} else if ((map->m_lblk - ee_block + map->m_len <
			   EXT4_EXT_ZERO_LEN) &&
			   (EXT4_EXT_MAY_ZEROOUT & split_flag)) {
			/* case 2 */
			if (map->m_lblk != ee_block) {
				zero_ex.ee_block = ex->ee_block;
				zero_ex.ee_len = cpu_to_le16(map->m_lblk -
							ee_block);
				ext4_ext_store_pblock(&zero_ex,
						      ext4_ext_pblock(ex));
				err = ext4_ext_zeroout(inode, &zero_ex);
				if (err)
					goto out;
			}

			split_map.m_lblk = ee_block;
3013 3014
			split_map.m_len = map->m_lblk - ee_block + map->m_len;
			allocated = map->m_len;
3015 3016
		}
	}
3017 3018 3019 3020 3021 3022

	allocated = ext4_split_extent(handle, inode, path,
				       &split_map, split_flag, 0);
	if (allocated < 0)
		err = allocated;

3023 3024 3025 3026
out:
	return err ? err : allocated;
}

3027
/*
3028
 * This function is called by ext4_ext_map_blocks() from
3029 3030 3031
 * ext4_get_blocks_dio_write() when DIO to write
 * to an uninitialized extent.
 *
P
Paul Bolle 已提交
3032
 * Writing to an uninitialized extent may result in splitting the uninitialized
3033
 * extent into multiple /initialized uninitialized extents (up to three)
3034 3035 3036 3037 3038 3039
 * There are three possibilities:
 *   a> There is no split required: Entire extent should be uninitialized
 *   b> Splits in two extents: Write is happening at either end of the extent
 *   c> Splits in three extents: Somone is writing in middle of the extent
 *
 * One of more index blocks maybe needed if the extent tree grow after
3040
 * the uninitialized extent split. To prevent ENOSPC occur at the IO
3041
 * complete, we need to split the uninitialized extent before DIO submit
3042
 * the IO. The uninitialized extent called at this time will be split
3043 3044 3045
 * into three uninitialized extent(at most). After IO complete, the part
 * being filled will be convert to initialized by the end_io callback function
 * via ext4_convert_unwritten_extents().
3046 3047
 *
 * Returns the size of uninitialized extent to be written on success.
3048 3049 3050
 */
static int ext4_split_unwritten_extents(handle_t *handle,
					struct inode *inode,
3051
					struct ext4_map_blocks *map,
3052 3053 3054
					struct ext4_ext_path *path,
					int flags)
{
3055 3056 3057 3058 3059
	ext4_lblk_t eof_block;
	ext4_lblk_t ee_block;
	struct ext4_extent *ex;
	unsigned int ee_len;
	int split_flag = 0, depth;
3060 3061 3062

	ext_debug("ext4_split_unwritten_extents: inode %lu, logical"
		"block %llu, max_blocks %u\n", inode->i_ino,
3063
		(unsigned long long)map->m_lblk, map->m_len);
3064 3065 3066

	eof_block = (inode->i_size + inode->i_sb->s_blocksize - 1) >>
		inode->i_sb->s_blocksize_bits;
3067 3068
	if (eof_block < map->m_lblk + map->m_len)
		eof_block = map->m_lblk + map->m_len;
3069 3070 3071 3072
	/*
	 * It is safe to convert extent to initialized via explicit
	 * zeroout only if extent is fully insde i_size or new_size.
	 */
3073 3074 3075 3076
	depth = ext_depth(inode);
	ex = path[depth].p_ext;
	ee_block = le32_to_cpu(ex->ee_block);
	ee_len = ext4_ext_get_actual_len(ex);
3077

3078 3079
	split_flag |= ee_block + ee_len <= eof_block ? EXT4_EXT_MAY_ZEROOUT : 0;
	split_flag |= EXT4_EXT_MARK_UNINIT2;
3080

3081 3082
	flags |= EXT4_GET_BLOCKS_PRE_IO;
	return ext4_split_extent(handle, inode, path, map, split_flag, flags);
3083
}
3084

3085
static int ext4_convert_unwritten_extents_endio(handle_t *handle,
3086 3087 3088 3089 3090 3091 3092 3093 3094 3095
					      struct inode *inode,
					      struct ext4_ext_path *path)
{
	struct ext4_extent *ex;
	int depth;
	int err = 0;

	depth = ext_depth(inode);
	ex = path[depth].p_ext;

3096 3097 3098 3099 3100
	ext_debug("ext4_convert_unwritten_extents_endio: inode %lu, logical"
		"block %llu, max_blocks %u\n", inode->i_ino,
		(unsigned long long)le32_to_cpu(ex->ee_block),
		ext4_ext_get_actual_len(ex));

3101 3102 3103 3104 3105 3106
	err = ext4_ext_get_access(handle, inode, path + depth);
	if (err)
		goto out;
	/* first mark the extent as initialized */
	ext4_ext_mark_initialized(ex);

3107 3108
	/* note: ext4_ext_correct_indexes() isn't needed here because
	 * borders are not changed
3109
	 */
3110 3111
	ext4_ext_try_to_merge(inode, path, ex);

3112 3113 3114 3115 3116 3117 3118
	/* Mark modified extent as dirty */
	err = ext4_ext_dirty(handle, inode, path + depth);
out:
	ext4_ext_show_leaf(inode, path);
	return err;
}

3119 3120 3121 3122 3123 3124 3125 3126
static void unmap_underlying_metadata_blocks(struct block_device *bdev,
			sector_t block, int count)
{
	int i;
	for (i = 0; i < count; i++)
                unmap_underlying_metadata(bdev, block + i);
}

T
Theodore Ts'o 已提交
3127 3128 3129 3130
/*
 * Handle EOFBLOCKS_FL flag, clearing it if necessary
 */
static int check_eofblocks_fl(handle_t *handle, struct inode *inode,
3131
			      ext4_lblk_t lblk,
T
Theodore Ts'o 已提交
3132 3133 3134 3135 3136
			      struct ext4_ext_path *path,
			      unsigned int len)
{
	int i, depth;
	struct ext4_extent_header *eh;
3137
	struct ext4_extent *last_ex;
T
Theodore Ts'o 已提交
3138 3139 3140 3141 3142 3143 3144 3145 3146 3147 3148 3149 3150 3151 3152 3153 3154 3155 3156 3157 3158 3159

	if (!ext4_test_inode_flag(inode, EXT4_INODE_EOFBLOCKS))
		return 0;

	depth = ext_depth(inode);
	eh = path[depth].p_hdr;

	if (unlikely(!eh->eh_entries)) {
		EXT4_ERROR_INODE(inode, "eh->eh_entries == 0 and "
				 "EOFBLOCKS_FL set");
		return -EIO;
	}
	last_ex = EXT_LAST_EXTENT(eh);
	/*
	 * We should clear the EOFBLOCKS_FL flag if we are writing the
	 * last block in the last extent in the file.  We test this by
	 * first checking to see if the caller to
	 * ext4_ext_get_blocks() was interested in the last block (or
	 * a block beyond the last block) in the current extent.  If
	 * this turns out to be false, we can bail out from this
	 * function immediately.
	 */
3160
	if (lblk + len < le32_to_cpu(last_ex->ee_block) +
T
Theodore Ts'o 已提交
3161 3162 3163 3164 3165 3166 3167 3168 3169 3170 3171 3172 3173 3174 3175 3176
	    ext4_ext_get_actual_len(last_ex))
		return 0;
	/*
	 * If the caller does appear to be planning to write at or
	 * beyond the end of the current extent, we then test to see
	 * if the current extent is the last extent in the file, by
	 * checking to make sure it was reached via the rightmost node
	 * at each level of the tree.
	 */
	for (i = depth-1; i >= 0; i--)
		if (path[i].p_idx != EXT_LAST_INDEX(path[i].p_hdr))
			return 0;
	ext4_clear_inode_flag(inode, EXT4_INODE_EOFBLOCKS);
	return ext4_mark_inode_dirty(handle, inode);
}

3177 3178 3179 3180 3181 3182 3183 3184 3185 3186 3187 3188 3189 3190 3191 3192 3193 3194 3195 3196 3197 3198 3199 3200 3201 3202 3203 3204 3205 3206 3207 3208 3209 3210 3211 3212 3213
/**
 * ext4_find_delalloc_range: find delayed allocated block in the given range.
 *
 * Goes through the buffer heads in the range [lblk_start, lblk_end] and returns
 * whether there are any buffers marked for delayed allocation. It returns '1'
 * on the first delalloc'ed buffer head found. If no buffer head in the given
 * range is marked for delalloc, it returns 0.
 * lblk_start should always be <= lblk_end.
 * search_hint_reverse is to indicate that searching in reverse from lblk_end to
 * lblk_start might be more efficient (i.e., we will likely hit the delalloc'ed
 * block sooner). This is useful when blocks are truncated sequentially from
 * lblk_start towards lblk_end.
 */
static int ext4_find_delalloc_range(struct inode *inode,
				    ext4_lblk_t lblk_start,
				    ext4_lblk_t lblk_end,
				    int search_hint_reverse)
{
	struct address_space *mapping = inode->i_mapping;
	struct buffer_head *head, *bh = NULL;
	struct page *page;
	ext4_lblk_t i, pg_lblk;
	pgoff_t index;

	/* reverse search wont work if fs block size is less than page size */
	if (inode->i_blkbits < PAGE_CACHE_SHIFT)
		search_hint_reverse = 0;

	if (search_hint_reverse)
		i = lblk_end;
	else
		i = lblk_start;

	index = i >> (PAGE_CACHE_SHIFT - inode->i_blkbits);

	while ((i >= lblk_start) && (i <= lblk_end)) {
		page = find_get_page(mapping, index);
3214
		if (!page)
3215 3216 3217 3218 3219 3220 3221 3222 3223 3224 3225 3226 3227 3228 3229 3230 3231 3232 3233 3234 3235 3236 3237 3238
			goto nextpage;

		if (!page_has_buffers(page))
			goto nextpage;

		head = page_buffers(page);
		if (!head)
			goto nextpage;

		bh = head;
		pg_lblk = index << (PAGE_CACHE_SHIFT -
						inode->i_blkbits);
		do {
			if (unlikely(pg_lblk < lblk_start)) {
				/*
				 * This is possible when fs block size is less
				 * than page size and our cluster starts/ends in
				 * middle of the page. So we need to skip the
				 * initial few blocks till we reach the 'lblk'
				 */
				pg_lblk++;
				continue;
			}

3239 3240 3241 3242 3243
			/* Check if the buffer is delayed allocated and that it
			 * is not yet mapped. (when da-buffers are mapped during
			 * their writeout, their da_mapped bit is set.)
			 */
			if (buffer_delay(bh) && !buffer_da_mapped(bh)) {
3244
				page_cache_release(page);
3245 3246 3247 3248
				trace_ext4_find_delalloc_range(inode,
						lblk_start, lblk_end,
						search_hint_reverse,
						1, i);
3249 3250 3251 3252 3253 3254 3255 3256 3257 3258 3259 3260 3261 3262 3263 3264 3265 3266 3267 3268 3269 3270
				return 1;
			}
			if (search_hint_reverse)
				i--;
			else
				i++;
		} while ((i >= lblk_start) && (i <= lblk_end) &&
				((bh = bh->b_this_page) != head));
nextpage:
		if (page)
			page_cache_release(page);
		/*
		 * Move to next page. 'i' will be the first lblk in the next
		 * page.
		 */
		if (search_hint_reverse)
			index--;
		else
			index++;
		i = index << (PAGE_CACHE_SHIFT - inode->i_blkbits);
	}

3271 3272
	trace_ext4_find_delalloc_range(inode, lblk_start, lblk_end,
					search_hint_reverse, 0, 0);
3273 3274 3275 3276 3277 3278 3279 3280 3281 3282 3283 3284 3285 3286 3287 3288 3289 3290 3291 3292 3293 3294 3295 3296 3297 3298 3299 3300 3301 3302 3303 3304 3305 3306 3307 3308 3309 3310 3311 3312 3313 3314 3315 3316 3317 3318 3319 3320 3321 3322 3323 3324 3325 3326 3327 3328 3329 3330 3331 3332 3333 3334 3335 3336 3337
	return 0;
}

int ext4_find_delalloc_cluster(struct inode *inode, ext4_lblk_t lblk,
			       int search_hint_reverse)
{
	struct ext4_sb_info *sbi = EXT4_SB(inode->i_sb);
	ext4_lblk_t lblk_start, lblk_end;
	lblk_start = lblk & (~(sbi->s_cluster_ratio - 1));
	lblk_end = lblk_start + sbi->s_cluster_ratio - 1;

	return ext4_find_delalloc_range(inode, lblk_start, lblk_end,
					search_hint_reverse);
}

/**
 * Determines how many complete clusters (out of those specified by the 'map')
 * are under delalloc and were reserved quota for.
 * This function is called when we are writing out the blocks that were
 * originally written with their allocation delayed, but then the space was
 * allocated using fallocate() before the delayed allocation could be resolved.
 * The cases to look for are:
 * ('=' indicated delayed allocated blocks
 *  '-' indicates non-delayed allocated blocks)
 * (a) partial clusters towards beginning and/or end outside of allocated range
 *     are not delalloc'ed.
 *	Ex:
 *	|----c---=|====c====|====c====|===-c----|
 *	         |++++++ allocated ++++++|
 *	==> 4 complete clusters in above example
 *
 * (b) partial cluster (outside of allocated range) towards either end is
 *     marked for delayed allocation. In this case, we will exclude that
 *     cluster.
 *	Ex:
 *	|----====c========|========c========|
 *	     |++++++ allocated ++++++|
 *	==> 1 complete clusters in above example
 *
 *	Ex:
 *	|================c================|
 *            |++++++ allocated ++++++|
 *	==> 0 complete clusters in above example
 *
 * The ext4_da_update_reserve_space will be called only if we
 * determine here that there were some "entire" clusters that span
 * this 'allocated' range.
 * In the non-bigalloc case, this function will just end up returning num_blks
 * without ever calling ext4_find_delalloc_range.
 */
static unsigned int
get_reserved_cluster_alloc(struct inode *inode, ext4_lblk_t lblk_start,
			   unsigned int num_blks)
{
	struct ext4_sb_info *sbi = EXT4_SB(inode->i_sb);
	ext4_lblk_t alloc_cluster_start, alloc_cluster_end;
	ext4_lblk_t lblk_from, lblk_to, c_offset;
	unsigned int allocated_clusters = 0;

	alloc_cluster_start = EXT4_B2C(sbi, lblk_start);
	alloc_cluster_end = EXT4_B2C(sbi, lblk_start + num_blks - 1);

	/* max possible clusters for this allocation */
	allocated_clusters = alloc_cluster_end - alloc_cluster_start + 1;

3338 3339
	trace_ext4_get_reserved_cluster_alloc(inode, lblk_start, num_blks);

3340 3341 3342 3343 3344 3345 3346 3347 3348 3349 3350 3351 3352 3353 3354 3355 3356 3357 3358 3359 3360 3361 3362
	/* Check towards left side */
	c_offset = lblk_start & (sbi->s_cluster_ratio - 1);
	if (c_offset) {
		lblk_from = lblk_start & (~(sbi->s_cluster_ratio - 1));
		lblk_to = lblk_from + c_offset - 1;

		if (ext4_find_delalloc_range(inode, lblk_from, lblk_to, 0))
			allocated_clusters--;
	}

	/* Now check towards right. */
	c_offset = (lblk_start + num_blks) & (sbi->s_cluster_ratio - 1);
	if (allocated_clusters && c_offset) {
		lblk_from = lblk_start + num_blks;
		lblk_to = lblk_from + (sbi->s_cluster_ratio - c_offset) - 1;

		if (ext4_find_delalloc_range(inode, lblk_from, lblk_to, 0))
			allocated_clusters--;
	}

	return allocated_clusters;
}

3363 3364
static int
ext4_ext_handle_uninitialized_extents(handle_t *handle, struct inode *inode,
3365
			struct ext4_map_blocks *map,
3366
			struct ext4_ext_path *path, int flags,
3367
			unsigned int allocated, ext4_fsblk_t newblock)
3368 3369 3370
{
	int ret = 0;
	int err = 0;
3371
	ext4_io_end_t *io = EXT4_I(inode)->cur_aio_dio;
3372 3373 3374

	ext_debug("ext4_ext_handle_uninitialized_extents: inode %lu, logical"
		  "block %llu, max_blocks %u, flags %d, allocated %u",
3375
		  inode->i_ino, (unsigned long long)map->m_lblk, map->m_len,
3376 3377 3378
		  flags, allocated);
	ext4_ext_show_leaf(inode, path);

3379 3380 3381
	trace_ext4_ext_handle_uninitialized_extents(inode, map, allocated,
						    newblock);

3382
	/* get_block() before submit the IO, split the extent */
3383
	if ((flags & EXT4_GET_BLOCKS_PRE_IO)) {
3384 3385
		ret = ext4_split_unwritten_extents(handle, inode, map,
						   path, flags);
3386 3387
		/*
		 * Flag the inode(non aio case) or end_io struct (aio case)
L
Lucas De Marchi 已提交
3388
		 * that this IO needs to conversion to written when IO is
3389 3390
		 * completed
		 */
3391
		if (io && !(io->flag & EXT4_IO_END_UNWRITTEN)) {
3392
			io->flag = EXT4_IO_END_UNWRITTEN;
3393 3394
			atomic_inc(&EXT4_I(inode)->i_aiodio_unwritten);
		} else
3395
			ext4_set_inode_state(inode, EXT4_STATE_DIO_UNWRITTEN);
3396
		if (ext4_should_dioread_nolock(inode))
3397
			map->m_flags |= EXT4_MAP_UNINIT;
3398 3399
		goto out;
	}
3400
	/* IO end_io complete, convert the filled extent to written */
3401
	if ((flags & EXT4_GET_BLOCKS_CONVERT)) {
3402
		ret = ext4_convert_unwritten_extents_endio(handle, inode,
3403
							path);
T
Theodore Ts'o 已提交
3404
		if (ret >= 0) {
3405
			ext4_update_inode_fsync_trans(handle, inode, 1);
3406 3407
			err = check_eofblocks_fl(handle, inode, map->m_lblk,
						 path, map->m_len);
T
Theodore Ts'o 已提交
3408 3409
		} else
			err = ret;
3410 3411 3412 3413 3414 3415 3416 3417 3418 3419 3420 3421 3422 3423 3424 3425 3426 3427 3428
		goto out2;
	}
	/* buffered IO case */
	/*
	 * repeat fallocate creation request
	 * we already have an unwritten extent
	 */
	if (flags & EXT4_GET_BLOCKS_UNINIT_EXT)
		goto map_out;

	/* buffered READ or buffered write_begin() lookup */
	if ((flags & EXT4_GET_BLOCKS_CREATE) == 0) {
		/*
		 * We have blocks reserved already.  We
		 * return allocated blocks so that delalloc
		 * won't do block reservation for us.  But
		 * the buffer head will be unmapped so that
		 * a read from the block returns 0s.
		 */
3429
		map->m_flags |= EXT4_MAP_UNWRITTEN;
3430 3431 3432 3433
		goto out1;
	}

	/* buffered write, writepage time, convert*/
3434
	ret = ext4_ext_convert_to_initialized(handle, inode, map, path);
T
Theodore Ts'o 已提交
3435
	if (ret >= 0) {
3436
		ext4_update_inode_fsync_trans(handle, inode, 1);
3437 3438
		err = check_eofblocks_fl(handle, inode, map->m_lblk, path,
					 map->m_len);
T
Theodore Ts'o 已提交
3439 3440 3441 3442
		if (err < 0)
			goto out2;
	}

3443 3444 3445 3446 3447 3448
out:
	if (ret <= 0) {
		err = ret;
		goto out2;
	} else
		allocated = ret;
3449
	map->m_flags |= EXT4_MAP_NEW;
3450 3451 3452 3453 3454 3455 3456
	/*
	 * if we allocated more blocks than requested
	 * we need to make sure we unmap the extra block
	 * allocated. The actual needed block will get
	 * unmapped later when we find the buffer_head marked
	 * new.
	 */
3457
	if (allocated > map->m_len) {
3458
		unmap_underlying_metadata_blocks(inode->i_sb->s_bdev,
3459 3460 3461
					newblock + map->m_len,
					allocated - map->m_len);
		allocated = map->m_len;
3462
	}
3463 3464 3465 3466 3467 3468 3469 3470

	/*
	 * If we have done fallocate with the offset that is already
	 * delayed allocated, we would have block reservation
	 * and quota reservation done in the delayed write path.
	 * But fallocate would have already updated quota and block
	 * count for this offset. So cancel these reservation
	 */
3471 3472 3473 3474 3475 3476 3477 3478 3479
	if (flags & EXT4_GET_BLOCKS_DELALLOC_RESERVE) {
		unsigned int reserved_clusters;
		reserved_clusters = get_reserved_cluster_alloc(inode,
				map->m_lblk, map->m_len);
		if (reserved_clusters)
			ext4_da_update_reserve_space(inode,
						     reserved_clusters,
						     0);
	}
3480

3481
map_out:
3482
	map->m_flags |= EXT4_MAP_MAPPED;
3483
out1:
3484 3485
	if (allocated > map->m_len)
		allocated = map->m_len;
3486
	ext4_ext_show_leaf(inode, path);
3487 3488
	map->m_pblk = newblock;
	map->m_len = allocated;
3489 3490 3491 3492 3493 3494 3495
out2:
	if (path) {
		ext4_ext_drop_refs(path);
		kfree(path);
	}
	return err ? err : allocated;
}
T
Theodore Ts'o 已提交
3496

3497 3498 3499 3500
/*
 * get_implied_cluster_alloc - check to see if the requested
 * allocation (in the map structure) overlaps with a cluster already
 * allocated in an extent.
3501
 *	@sb	The filesystem superblock structure
3502 3503 3504 3505 3506 3507 3508 3509 3510 3511 3512 3513 3514 3515 3516 3517 3518 3519 3520 3521 3522 3523 3524 3525 3526 3527 3528 3529 3530 3531 3532 3533 3534 3535 3536 3537
 *	@map	The requested lblk->pblk mapping
 *	@ex	The extent structure which might contain an implied
 *			cluster allocation
 *
 * This function is called by ext4_ext_map_blocks() after we failed to
 * find blocks that were already in the inode's extent tree.  Hence,
 * we know that the beginning of the requested region cannot overlap
 * the extent from the inode's extent tree.  There are three cases we
 * want to catch.  The first is this case:
 *
 *		 |--- cluster # N--|
 *    |--- extent ---|	|---- requested region ---|
 *			|==========|
 *
 * The second case that we need to test for is this one:
 *
 *   |--------- cluster # N ----------------|
 *	   |--- requested region --|   |------- extent ----|
 *	   |=======================|
 *
 * The third case is when the requested region lies between two extents
 * within the same cluster:
 *          |------------- cluster # N-------------|
 * |----- ex -----|                  |---- ex_right ----|
 *                  |------ requested region ------|
 *                  |================|
 *
 * In each of the above cases, we need to set the map->m_pblk and
 * map->m_len so it corresponds to the return the extent labelled as
 * "|====|" from cluster #N, since it is already in use for data in
 * cluster EXT4_B2C(sbi, map->m_lblk).	We will then return 1 to
 * signal to ext4_ext_map_blocks() that map->m_pblk should be treated
 * as a new "allocated" block region.  Otherwise, we will return 0 and
 * ext4_ext_map_blocks() will then allocate one or more new clusters
 * by calling ext4_mb_new_blocks().
 */
3538
static int get_implied_cluster_alloc(struct super_block *sb,
3539 3540 3541 3542
				     struct ext4_map_blocks *map,
				     struct ext4_extent *ex,
				     struct ext4_ext_path *path)
{
3543
	struct ext4_sb_info *sbi = EXT4_SB(sb);
3544 3545 3546 3547 3548 3549 3550 3551 3552 3553 3554 3555 3556 3557 3558 3559 3560 3561 3562 3563 3564 3565 3566 3567 3568 3569 3570 3571 3572 3573 3574 3575 3576 3577 3578 3579 3580 3581 3582 3583 3584 3585 3586 3587 3588 3589 3590 3591
	ext4_lblk_t c_offset = map->m_lblk & (sbi->s_cluster_ratio-1);
	ext4_lblk_t ex_cluster_start, ex_cluster_end;
	ext4_lblk_t rr_cluster_start, rr_cluster_end;
	ext4_lblk_t ee_block = le32_to_cpu(ex->ee_block);
	ext4_fsblk_t ee_start = ext4_ext_pblock(ex);
	unsigned short ee_len = ext4_ext_get_actual_len(ex);

	/* The extent passed in that we are trying to match */
	ex_cluster_start = EXT4_B2C(sbi, ee_block);
	ex_cluster_end = EXT4_B2C(sbi, ee_block + ee_len - 1);

	/* The requested region passed into ext4_map_blocks() */
	rr_cluster_start = EXT4_B2C(sbi, map->m_lblk);
	rr_cluster_end = EXT4_B2C(sbi, map->m_lblk + map->m_len - 1);

	if ((rr_cluster_start == ex_cluster_end) ||
	    (rr_cluster_start == ex_cluster_start)) {
		if (rr_cluster_start == ex_cluster_end)
			ee_start += ee_len - 1;
		map->m_pblk = (ee_start & ~(sbi->s_cluster_ratio - 1)) +
			c_offset;
		map->m_len = min(map->m_len,
				 (unsigned) sbi->s_cluster_ratio - c_offset);
		/*
		 * Check for and handle this case:
		 *
		 *   |--------- cluster # N-------------|
		 *		       |------- extent ----|
		 *	   |--- requested region ---|
		 *	   |===========|
		 */

		if (map->m_lblk < ee_block)
			map->m_len = min(map->m_len, ee_block - map->m_lblk);

		/*
		 * Check for the case where there is already another allocated
		 * block to the right of 'ex' but before the end of the cluster.
		 *
		 *          |------------- cluster # N-------------|
		 * |----- ex -----|                  |---- ex_right ----|
		 *                  |------ requested region ------|
		 *                  |================|
		 */
		if (map->m_lblk > ee_block) {
			ext4_lblk_t next = ext4_ext_next_allocated_block(path);
			map->m_len = min(map->m_len, next - map->m_lblk);
		}
3592 3593

		trace_ext4_get_implied_cluster_alloc_exit(sb, map, 1);
3594 3595
		return 1;
	}
3596 3597

	trace_ext4_get_implied_cluster_alloc_exit(sb, map, 0);
3598 3599 3600 3601
	return 0;
}


3602
/*
3603 3604 3605
 * Block allocation/map/preallocation routine for extents based files
 *
 *
3606
 * Need to be called with
3607 3608
 * 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)
3609 3610 3611 3612 3613 3614 3615 3616 3617 3618
 *
 * return > 0, number of of blocks already mapped/allocated
 *          if create == 0 and these are pre-allocated blocks
 *          	buffer head is unmapped
 *          otherwise blocks are mapped
 *
 * return = 0, if plain look up failed (blocks have not been allocated)
 *          buffer head is unmapped
 *
 * return < 0, error case.
3619
 */
3620 3621
int ext4_ext_map_blocks(handle_t *handle, struct inode *inode,
			struct ext4_map_blocks *map, int flags)
A
Alex Tomas 已提交
3622 3623
{
	struct ext4_ext_path *path = NULL;
3624 3625
	struct ext4_extent newex, *ex, *ex2;
	struct ext4_sb_info *sbi = EXT4_SB(inode->i_sb);
3626
	ext4_fsblk_t newblock = 0;
3627 3628
	int free_on_err = 0, err = 0, depth, ret;
	unsigned int allocated = 0, offset = 0;
3629
	unsigned int allocated_clusters = 0, reserved_clusters = 0;
3630 3631
	unsigned int punched_out = 0;
	unsigned int result = 0;
3632
	struct ext4_allocation_request ar;
3633
	ext4_io_end_t *io = EXT4_I(inode)->cur_aio_dio;
3634
	ext4_lblk_t cluster_offset;
3635
	struct ext4_map_blocks punch_map;
A
Alex Tomas 已提交
3636

3637
	ext_debug("blocks %u/%u requested for inode %lu\n",
3638
		  map->m_lblk, map->m_len, inode->i_ino);
3639
	trace_ext4_ext_map_blocks_enter(inode, map->m_lblk, map->m_len, flags);
A
Alex Tomas 已提交
3640 3641

	/* check in cache */
3642 3643
	if (!(flags & EXT4_GET_BLOCKS_PUNCH_OUT_EXT) &&
		ext4_ext_in_cache(inode, map->m_lblk, &newex)) {
3644
		if (!newex.ee_start_lo && !newex.ee_start_hi) {
3645 3646 3647 3648
			if ((sbi->s_cluster_ratio > 1) &&
			    ext4_find_delalloc_cluster(inode, map->m_lblk, 0))
				map->m_flags |= EXT4_MAP_FROM_CLUSTER;

3649
			if ((flags & EXT4_GET_BLOCKS_CREATE) == 0) {
3650 3651 3652 3653
				/*
				 * block isn't allocated yet and
				 * user doesn't want to allocate it
				 */
A
Alex Tomas 已提交
3654 3655 3656
				goto out2;
			}
			/* we should allocate requested block */
3657
		} else {
A
Alex Tomas 已提交
3658
			/* block is already allocated */
3659 3660
			if (sbi->s_cluster_ratio > 1)
				map->m_flags |= EXT4_MAP_FROM_CLUSTER;
3661
			newblock = map->m_lblk
D
Dave Kleikamp 已提交
3662
				   - le32_to_cpu(newex.ee_block)
3663
				   + ext4_ext_pblock(&newex);
3664
			/* number of remaining blocks in the extent */
3665
			allocated = ext4_ext_get_actual_len(&newex) -
3666
				(map->m_lblk - le32_to_cpu(newex.ee_block));
A
Alex Tomas 已提交
3667 3668 3669 3670 3671
			goto out;
		}
	}

	/* find extent for this block */
3672
	path = ext4_ext_find_extent(inode, map->m_lblk, NULL);
A
Alex Tomas 已提交
3673 3674 3675 3676 3677 3678 3679 3680 3681
	if (IS_ERR(path)) {
		err = PTR_ERR(path);
		path = NULL;
		goto out2;
	}

	depth = ext_depth(inode);

	/*
3682 3683
	 * consistent leaf must not be empty;
	 * this situation is possible, though, _during_ tree modification;
A
Alex Tomas 已提交
3684 3685
	 * this is why assert can't be put in ext4_ext_find_extent()
	 */
3686 3687
	if (unlikely(path[depth].p_ext == NULL && depth != 0)) {
		EXT4_ERROR_INODE(inode, "bad extent address "
3688 3689 3690
				 "lblock: %lu, depth: %d pblock %lld",
				 (unsigned long) map->m_lblk, depth,
				 path[depth].p_block);
3691 3692 3693
		err = -EIO;
		goto out2;
	}
A
Alex Tomas 已提交
3694

3695 3696
	ex = path[depth].p_ext;
	if (ex) {
A
Aneesh Kumar K.V 已提交
3697
		ext4_lblk_t ee_block = le32_to_cpu(ex->ee_block);
3698
		ext4_fsblk_t ee_start = ext4_ext_pblock(ex);
A
Amit Arora 已提交
3699
		unsigned short ee_len;
3700 3701 3702

		/*
		 * Uninitialized extents are treated as holes, except that
3703
		 * we split out initialized portions during a write.
3704
		 */
A
Amit Arora 已提交
3705
		ee_len = ext4_ext_get_actual_len(ex);
3706 3707 3708

		trace_ext4_ext_show_extent(inode, ee_block, ee_start, ee_len);

3709
		/* if found extent covers block, simply return it */
3710
		if (in_range(map->m_lblk, ee_block, ee_len)) {
3711 3712
			ext4_fsblk_t partial_cluster = 0;

3713
			newblock = map->m_lblk - ee_block + ee_start;
3714
			/* number of remaining blocks in the extent */
3715 3716 3717
			allocated = ee_len - (map->m_lblk - ee_block);
			ext_debug("%u fit into %u:%d -> %llu\n", map->m_lblk,
				  ee_block, ee_len, newblock);
3718

3719 3720 3721 3722 3723 3724 3725 3726 3727 3728 3729 3730 3731 3732
			if ((flags & EXT4_GET_BLOCKS_PUNCH_OUT_EXT) == 0) {
				/*
				 * Do not put uninitialized extent
				 * in the cache
				 */
				if (!ext4_ext_is_uninitialized(ex)) {
					ext4_ext_put_in_cache(inode, ee_block,
						ee_len, ee_start);
					goto out;
				}
				ret = ext4_ext_handle_uninitialized_extents(
					handle, inode, map, path, flags,
					allocated, newblock);
				return ret;
3733
			}
3734 3735 3736 3737 3738 3739 3740 3741 3742 3743 3744 3745 3746 3747 3748 3749 3750 3751 3752 3753 3754 3755 3756 3757 3758 3759 3760 3761 3762 3763 3764 3765 3766 3767 3768 3769 3770 3771 3772 3773 3774 3775 3776 3777 3778 3779 3780 3781 3782 3783 3784 3785 3786 3787 3788 3789 3790 3791 3792

			/*
			 * Punch out the map length, but only to the
			 * end of the extent
			 */
			punched_out = allocated < map->m_len ?
				allocated : map->m_len;

			/*
			 * Sense extents need to be converted to
			 * uninitialized, they must fit in an
			 * uninitialized extent
			 */
			if (punched_out > EXT_UNINIT_MAX_LEN)
				punched_out = EXT_UNINIT_MAX_LEN;

			punch_map.m_lblk = map->m_lblk;
			punch_map.m_pblk = newblock;
			punch_map.m_len = punched_out;
			punch_map.m_flags = 0;

			/* Check to see if the extent needs to be split */
			if (punch_map.m_len != ee_len ||
				punch_map.m_lblk != ee_block) {

				ret = ext4_split_extent(handle, inode,
				path, &punch_map, 0,
				EXT4_GET_BLOCKS_PUNCH_OUT_EXT |
				EXT4_GET_BLOCKS_PRE_IO);

				if (ret < 0) {
					err = ret;
					goto out2;
				}
				/*
				 * find extent for the block at
				 * the start of the hole
				 */
				ext4_ext_drop_refs(path);
				kfree(path);

				path = ext4_ext_find_extent(inode,
				map->m_lblk, NULL);
				if (IS_ERR(path)) {
					err = PTR_ERR(path);
					path = NULL;
					goto out2;
				}

				depth = ext_depth(inode);
				ex = path[depth].p_ext;
				ee_len = ext4_ext_get_actual_len(ex);
				ee_block = le32_to_cpu(ex->ee_block);
				ee_start = ext4_ext_pblock(ex);

			}

			ext4_ext_mark_uninitialized(ex);

3793 3794 3795
			ext4_ext_invalidate_cache(inode);

			err = ext4_ext_rm_leaf(handle, inode, path,
3796 3797
					       &partial_cluster, map->m_lblk,
					       map->m_lblk + punched_out);
3798 3799 3800 3801 3802 3803 3804 3805 3806 3807 3808 3809 3810 3811 3812 3813 3814

			if (!err && path->p_hdr->eh_entries == 0) {
				/*
				 * Punch hole freed all of this sub tree,
				 * so we need to correct eh_depth
				 */
				err = ext4_ext_get_access(handle, inode, path);
				if (err == 0) {
					ext_inode_hdr(inode)->eh_depth = 0;
					ext_inode_hdr(inode)->eh_max =
					cpu_to_le16(ext4_ext_space_root(
						inode, 0));

					err = ext4_ext_dirty(
						handle, inode, path);
				}
			}
3815 3816

			goto out2;
A
Alex Tomas 已提交
3817 3818 3819
		}
	}

3820 3821 3822 3823
	if ((sbi->s_cluster_ratio > 1) &&
	    ext4_find_delalloc_cluster(inode, map->m_lblk, 0))
		map->m_flags |= EXT4_MAP_FROM_CLUSTER;

A
Alex Tomas 已提交
3824
	/*
3825
	 * requested block isn't allocated yet;
A
Alex Tomas 已提交
3826 3827
	 * we couldn't try to create block if create flag is zero
	 */
3828
	if ((flags & EXT4_GET_BLOCKS_CREATE) == 0) {
3829 3830 3831 3832
		/*
		 * put just found gap into cache to speed up
		 * subsequent requests
		 */
3833
		ext4_ext_put_gap_in_cache(inode, path, map->m_lblk);
A
Alex Tomas 已提交
3834 3835
		goto out2;
	}
3836

A
Alex Tomas 已提交
3837
	/*
3838
	 * Okay, we need to do block allocation.
A
Andrew Morton 已提交
3839
	 */
3840
	map->m_flags &= ~EXT4_MAP_FROM_CLUSTER;
3841 3842 3843 3844 3845 3846 3847 3848
	newex.ee_block = cpu_to_le32(map->m_lblk);
	cluster_offset = map->m_lblk & (sbi->s_cluster_ratio-1);

	/*
	 * If we are doing bigalloc, check to see if the extent returned
	 * by ext4_ext_find_extent() implies a cluster we can use.
	 */
	if (cluster_offset && ex &&
3849
	    get_implied_cluster_alloc(inode->i_sb, map, ex, path)) {
3850 3851
		ar.len = allocated = map->m_len;
		newblock = map->m_pblk;
3852
		map->m_flags |= EXT4_MAP_FROM_CLUSTER;
3853 3854
		goto got_allocated_blocks;
	}
A
Alex Tomas 已提交
3855

3856
	/* find neighbour allocated blocks */
3857
	ar.lleft = map->m_lblk;
3858 3859 3860
	err = ext4_ext_search_left(inode, path, &ar.lleft, &ar.pleft);
	if (err)
		goto out2;
3861
	ar.lright = map->m_lblk;
3862 3863
	ex2 = NULL;
	err = ext4_ext_search_right(inode, path, &ar.lright, &ar.pright, &ex2);
3864 3865
	if (err)
		goto out2;
A
Amit Arora 已提交
3866

3867 3868 3869
	/* Check if the extent after searching to the right implies a
	 * cluster we can use. */
	if ((sbi->s_cluster_ratio > 1) && ex2 &&
3870
	    get_implied_cluster_alloc(inode->i_sb, map, ex2, path)) {
3871 3872
		ar.len = allocated = map->m_len;
		newblock = map->m_pblk;
3873
		map->m_flags |= EXT4_MAP_FROM_CLUSTER;
3874 3875 3876
		goto got_allocated_blocks;
	}

3877 3878 3879 3880 3881 3882
	/*
	 * See if request is beyond maximum number of blocks we can have in
	 * a single extent. For an initialized extent this limit is
	 * EXT_INIT_MAX_LEN and for an uninitialized extent this limit is
	 * EXT_UNINIT_MAX_LEN.
	 */
3883
	if (map->m_len > EXT_INIT_MAX_LEN &&
3884
	    !(flags & EXT4_GET_BLOCKS_UNINIT_EXT))
3885 3886
		map->m_len = EXT_INIT_MAX_LEN;
	else if (map->m_len > EXT_UNINIT_MAX_LEN &&
3887
		 (flags & EXT4_GET_BLOCKS_UNINIT_EXT))
3888
		map->m_len = EXT_UNINIT_MAX_LEN;
3889

3890 3891
	/* Check if we can really insert (m_lblk)::(m_lblk + m_len) extent */
	newex.ee_len = cpu_to_le16(map->m_len);
3892
	err = ext4_ext_check_overlap(sbi, inode, &newex, path);
A
Amit Arora 已提交
3893
	if (err)
3894
		allocated = ext4_ext_get_actual_len(&newex);
A
Amit Arora 已提交
3895
	else
3896
		allocated = map->m_len;
3897 3898 3899

	/* allocate new block */
	ar.inode = inode;
3900 3901
	ar.goal = ext4_ext_find_goal(inode, path, map->m_lblk);
	ar.logical = map->m_lblk;
3902 3903 3904 3905 3906 3907 3908 3909 3910 3911 3912 3913
	/*
	 * We calculate the offset from the beginning of the cluster
	 * for the logical block number, since when we allocate a
	 * physical cluster, the physical block should start at the
	 * same offset from the beginning of the cluster.  This is
	 * needed so that future calls to get_implied_cluster_alloc()
	 * work correctly.
	 */
	offset = map->m_lblk & (sbi->s_cluster_ratio - 1);
	ar.len = EXT4_NUM_B2C(sbi, offset+allocated);
	ar.goal -= offset;
	ar.logical -= offset;
3914 3915 3916 3917 3918
	if (S_ISREG(inode->i_mode))
		ar.flags = EXT4_MB_HINT_DATA;
	else
		/* disable in-core preallocation for non-regular files */
		ar.flags = 0;
3919 3920
	if (flags & EXT4_GET_BLOCKS_NO_NORMALIZE)
		ar.flags |= EXT4_MB_HINT_NOPREALLOC;
3921
	newblock = ext4_mb_new_blocks(handle, &ar, &err);
A
Alex Tomas 已提交
3922 3923
	if (!newblock)
		goto out2;
3924
	ext_debug("allocate new block: goal %llu, found %llu/%u\n",
3925
		  ar.goal, newblock, allocated);
3926
	free_on_err = 1;
3927
	allocated_clusters = ar.len;
3928 3929 3930
	ar.len = EXT4_C2B(sbi, ar.len) - offset;
	if (ar.len > allocated)
		ar.len = allocated;
A
Alex Tomas 已提交
3931

3932
got_allocated_blocks:
A
Alex Tomas 已提交
3933
	/* try to insert new extent into found leaf and return */
3934
	ext4_ext_store_pblock(&newex, newblock + offset);
3935
	newex.ee_len = cpu_to_le16(ar.len);
3936 3937
	/* Mark uninitialized */
	if (flags & EXT4_GET_BLOCKS_UNINIT_EXT){
A
Amit Arora 已提交
3938
		ext4_ext_mark_uninitialized(&newex);
3939
		/*
3940
		 * io_end structure was created for every IO write to an
L
Lucas De Marchi 已提交
3941
		 * uninitialized extent. To avoid unnecessary conversion,
3942
		 * here we flag the IO that really needs the conversion.
3943
		 * For non asycn direct IO case, flag the inode state
L
Lucas De Marchi 已提交
3944
		 * that we need to perform conversion when IO is done.
3945
		 */
3946
		if ((flags & EXT4_GET_BLOCKS_PRE_IO)) {
3947
			if (io && !(io->flag & EXT4_IO_END_UNWRITTEN)) {
3948
				io->flag = EXT4_IO_END_UNWRITTEN;
3949 3950
				atomic_inc(&EXT4_I(inode)->i_aiodio_unwritten);
			} else
3951 3952
				ext4_set_inode_state(inode,
						     EXT4_STATE_DIO_UNWRITTEN);
3953
		}
3954
		if (ext4_should_dioread_nolock(inode))
3955
			map->m_flags |= EXT4_MAP_UNINIT;
3956
	}
3957

3958
	err = check_eofblocks_fl(handle, inode, map->m_lblk, path, ar.len);
3959 3960 3961
	if (!err)
		err = ext4_ext_insert_extent(handle, inode, path,
					     &newex, flags);
3962
	if (err && free_on_err) {
3963 3964
		int fb_flags = flags & EXT4_GET_BLOCKS_DELALLOC_RESERVE ?
			EXT4_FREE_BLOCKS_NO_QUOT_UPDATE : 0;
3965
		/* free data blocks we just allocated */
3966 3967
		/* not a good idea to call discard here directly,
		 * but otherwise we'd need to call it every free() */
3968
		ext4_discard_preallocations(inode);
3969
		ext4_free_blocks(handle, inode, NULL, ext4_ext_pblock(&newex),
3970
				 ext4_ext_get_actual_len(&newex), fb_flags);
A
Alex Tomas 已提交
3971
		goto out2;
3972
	}
A
Alex Tomas 已提交
3973 3974

	/* previous routine could use block we allocated */
3975
	newblock = ext4_ext_pblock(&newex);
3976
	allocated = ext4_ext_get_actual_len(&newex);
3977 3978 3979
	if (allocated > map->m_len)
		allocated = map->m_len;
	map->m_flags |= EXT4_MAP_NEW;
A
Alex Tomas 已提交
3980

3981 3982 3983 3984
	/*
	 * Update reserved blocks/metadata blocks after successful
	 * block allocation which had been deferred till now.
	 */
3985 3986 3987 3988 3989 3990 3991 3992 3993 3994 3995 3996 3997 3998 3999 4000 4001 4002 4003 4004 4005 4006 4007 4008
	if (flags & EXT4_GET_BLOCKS_DELALLOC_RESERVE) {
		/*
		 * Check how many clusters we had reserved this allocted range.
		 */
		reserved_clusters = get_reserved_cluster_alloc(inode,
						map->m_lblk, allocated);
		if (map->m_flags & EXT4_MAP_FROM_CLUSTER) {
			if (reserved_clusters) {
				/*
				 * We have clusters reserved for this range.
				 * But since we are not doing actual allocation
				 * and are simply using blocks from previously
				 * allocated cluster, we should release the
				 * reservation and not claim quota.
				 */
				ext4_da_update_reserve_space(inode,
						reserved_clusters, 0);
			}
		} else {
			BUG_ON(allocated_clusters < reserved_clusters);
			/* We will claim quota for all newly allocated blocks.*/
			ext4_da_update_reserve_space(inode, allocated_clusters,
							1);
			if (reserved_clusters < allocated_clusters) {
4009
				struct ext4_inode_info *ei = EXT4_I(inode);
4010 4011 4012 4013 4014 4015 4016 4017 4018 4019 4020 4021 4022 4023 4024 4025 4026 4027 4028 4029 4030 4031 4032 4033 4034 4035 4036 4037 4038 4039 4040 4041 4042 4043 4044 4045 4046 4047 4048 4049 4050 4051
				int reservation = allocated_clusters -
						  reserved_clusters;
				/*
				 * It seems we claimed few clusters outside of
				 * the range of this allocation. We should give
				 * it back to the reservation pool. This can
				 * happen in the following case:
				 *
				 * * Suppose s_cluster_ratio is 4 (i.e., each
				 *   cluster has 4 blocks. Thus, the clusters
				 *   are [0-3],[4-7],[8-11]...
				 * * First comes delayed allocation write for
				 *   logical blocks 10 & 11. Since there were no
				 *   previous delayed allocated blocks in the
				 *   range [8-11], we would reserve 1 cluster
				 *   for this write.
				 * * Next comes write for logical blocks 3 to 8.
				 *   In this case, we will reserve 2 clusters
				 *   (for [0-3] and [4-7]; and not for [8-11] as
				 *   that range has a delayed allocated blocks.
				 *   Thus total reserved clusters now becomes 3.
				 * * Now, during the delayed allocation writeout
				 *   time, we will first write blocks [3-8] and
				 *   allocate 3 clusters for writing these
				 *   blocks. Also, we would claim all these
				 *   three clusters above.
				 * * Now when we come here to writeout the
				 *   blocks [10-11], we would expect to claim
				 *   the reservation of 1 cluster we had made
				 *   (and we would claim it since there are no
				 *   more delayed allocated blocks in the range
				 *   [8-11]. But our reserved cluster count had
				 *   already gone to 0.
				 *
				 *   Thus, at the step 4 above when we determine
				 *   that there are still some unwritten delayed
				 *   allocated blocks outside of our current
				 *   block range, we should increment the
				 *   reserved clusters count so that when the
				 *   remaining blocks finally gets written, we
				 *   could claim them.
				 */
4052 4053 4054 4055 4056
				dquot_reserve_block(inode,
						EXT4_C2B(sbi, reservation));
				spin_lock(&ei->i_block_reservation_lock);
				ei->i_reserved_data_blocks += reservation;
				spin_unlock(&ei->i_block_reservation_lock);
4057 4058 4059
			}
		}
	}
4060

4061 4062 4063 4064 4065
	/*
	 * Cache the extent and update transaction to commit on fdatasync only
	 * when it is _not_ an uninitialized extent.
	 */
	if ((flags & EXT4_GET_BLOCKS_UNINIT_EXT) == 0) {
4066
		ext4_ext_put_in_cache(inode, map->m_lblk, allocated, newblock);
4067 4068 4069
		ext4_update_inode_fsync_trans(handle, inode, 1);
	} else
		ext4_update_inode_fsync_trans(handle, inode, 0);
A
Alex Tomas 已提交
4070
out:
4071 4072
	if (allocated > map->m_len)
		allocated = map->m_len;
A
Alex Tomas 已提交
4073
	ext4_ext_show_leaf(inode, path);
4074 4075 4076
	map->m_flags |= EXT4_MAP_MAPPED;
	map->m_pblk = newblock;
	map->m_len = allocated;
A
Alex Tomas 已提交
4077 4078 4079 4080 4081
out2:
	if (path) {
		ext4_ext_drop_refs(path);
		kfree(path);
	}
4082 4083
	trace_ext4_ext_map_blocks_exit(inode, map->m_lblk,
		newblock, map->m_len, err ? err : allocated);
4084 4085 4086 4087 4088

	result = (flags & EXT4_GET_BLOCKS_PUNCH_OUT_EXT) ?
			punched_out : allocated;

	return err ? err : result;
A
Alex Tomas 已提交
4089 4090
}

4091
void ext4_ext_truncate(struct inode *inode)
A
Alex Tomas 已提交
4092 4093 4094
{
	struct address_space *mapping = inode->i_mapping;
	struct super_block *sb = inode->i_sb;
A
Aneesh Kumar K.V 已提交
4095
	ext4_lblk_t last_block;
A
Alex Tomas 已提交
4096
	handle_t *handle;
4097
	loff_t page_len;
A
Alex Tomas 已提交
4098 4099
	int err = 0;

4100 4101 4102 4103 4104 4105
	/*
	 * finish any pending end_io work so we won't run the risk of
	 * converting any truncated blocks to initialized later
	 */
	ext4_flush_completed_IO(inode);

A
Alex Tomas 已提交
4106 4107 4108
	/*
	 * probably first extent we're gonna free will be last in block
	 */
4109
	err = ext4_writepage_trans_blocks(inode);
A
Alex Tomas 已提交
4110
	handle = ext4_journal_start(inode, err);
4111
	if (IS_ERR(handle))
A
Alex Tomas 已提交
4112 4113
		return;

4114 4115 4116 4117 4118 4119 4120 4121 4122 4123
	if (inode->i_size % PAGE_CACHE_SIZE != 0) {
		page_len = PAGE_CACHE_SIZE -
			(inode->i_size & (PAGE_CACHE_SIZE - 1));

		err = ext4_discard_partial_page_buffers(handle,
			mapping, inode->i_size, page_len, 0);

		if (err)
			goto out_stop;
	}
A
Alex Tomas 已提交
4124

4125 4126 4127
	if (ext4_orphan_add(handle, inode))
		goto out_stop;

4128
	down_write(&EXT4_I(inode)->i_data_sem);
A
Alex Tomas 已提交
4129 4130
	ext4_ext_invalidate_cache(inode);

4131
	ext4_discard_preallocations(inode);
4132

A
Alex Tomas 已提交
4133
	/*
4134 4135 4136
	 * TODO: optimization is possible here.
	 * Probably we need not scan at all,
	 * because page truncation is enough.
A
Alex Tomas 已提交
4137 4138 4139 4140 4141 4142 4143 4144
	 */

	/* we have to know where to truncate from in crash case */
	EXT4_I(inode)->i_disksize = inode->i_size;
	ext4_mark_inode_dirty(handle, inode);

	last_block = (inode->i_size + sb->s_blocksize - 1)
			>> EXT4_BLOCK_SIZE_BITS(sb);
4145
	err = ext4_ext_remove_space(inode, last_block);
A
Alex Tomas 已提交
4146 4147

	/* In a multi-transaction truncate, we only make the final
4148 4149
	 * transaction synchronous.
	 */
A
Alex Tomas 已提交
4150
	if (IS_SYNC(inode))
4151
		ext4_handle_sync(handle);
A
Alex Tomas 已提交
4152

4153
	up_write(&EXT4_I(inode)->i_data_sem);
4154 4155

out_stop:
A
Alex Tomas 已提交
4156
	/*
4157
	 * If this was a simple ftruncate() and the file will remain alive,
A
Alex Tomas 已提交
4158 4159 4160 4161 4162 4163 4164 4165
	 * 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
	 * ext4_delete_inode(), and we allow that function to clean up the
	 * orphan info for us.
	 */
	if (inode->i_nlink)
		ext4_orphan_del(handle, inode);

4166 4167
	inode->i_mtime = inode->i_ctime = ext4_current_time(inode);
	ext4_mark_inode_dirty(handle, inode);
A
Alex Tomas 已提交
4168 4169 4170
	ext4_journal_stop(handle);
}

4171 4172 4173 4174 4175 4176 4177 4178 4179 4180 4181 4182 4183 4184
static void ext4_falloc_update_inode(struct inode *inode,
				int mode, loff_t new_size, int update_ctime)
{
	struct timespec now;

	if (update_ctime) {
		now = current_fs_time(inode->i_sb);
		if (!timespec_equal(&inode->i_ctime, &now))
			inode->i_ctime = now;
	}
	/*
	 * Update only when preallocation was requested beyond
	 * the file size.
	 */
4185 4186 4187 4188 4189
	if (!(mode & FALLOC_FL_KEEP_SIZE)) {
		if (new_size > i_size_read(inode))
			i_size_write(inode, new_size);
		if (new_size > EXT4_I(inode)->i_disksize)
			ext4_update_i_disksize(inode, new_size);
4190 4191 4192 4193 4194 4195
	} else {
		/*
		 * Mark that we allocate beyond EOF so the subsequent truncate
		 * can proceed even if the new size is the same as i_size.
		 */
		if (new_size > i_size_read(inode))
4196
			ext4_set_inode_flag(inode, EXT4_INODE_EOFBLOCKS);
4197 4198 4199 4200
	}

}

A
Amit Arora 已提交
4201
/*
4202
 * preallocate space for a file. This implements ext4's fallocate file
A
Amit Arora 已提交
4203 4204 4205 4206 4207
 * operation, which gets called from sys_fallocate system call.
 * For block-mapped files, posix_fallocate should fall back to the method
 * of writing zeroes to the required new blocks (the same behavior which is
 * expected for file systems which do not support fallocate() system call).
 */
4208
long ext4_fallocate(struct file *file, int mode, loff_t offset, loff_t len)
A
Amit Arora 已提交
4209
{
4210
	struct inode *inode = file->f_path.dentry->d_inode;
A
Amit Arora 已提交
4211
	handle_t *handle;
4212
	loff_t new_size;
4213
	unsigned int max_blocks;
A
Amit Arora 已提交
4214 4215 4216
	int ret = 0;
	int ret2 = 0;
	int retries = 0;
4217
	struct ext4_map_blocks map;
A
Amit Arora 已提交
4218 4219 4220 4221 4222 4223
	unsigned int credits, blkbits = inode->i_blkbits;

	/*
	 * currently supporting (pre)allocate mode for extent-based
	 * files _only_
	 */
4224
	if (!(ext4_test_inode_flag(inode, EXT4_INODE_EXTENTS)))
A
Amit Arora 已提交
4225 4226
		return -EOPNOTSUPP;

4227 4228 4229 4230 4231 4232 4233
	/* Return error if mode is not supported */
	if (mode & ~(FALLOC_FL_KEEP_SIZE | FALLOC_FL_PUNCH_HOLE))
		return -EOPNOTSUPP;

	if (mode & FALLOC_FL_PUNCH_HOLE)
		return ext4_punch_hole(file, offset, len);

4234
	trace_ext4_fallocate_enter(inode, offset, len, mode);
4235
	map.m_lblk = offset >> blkbits;
4236 4237 4238 4239
	/*
	 * We can't just convert len to max_blocks because
	 * If blocksize = 4096 offset = 3072 and len = 2048
	 */
A
Amit Arora 已提交
4240
	max_blocks = (EXT4_BLOCK_ALIGN(len + offset, blkbits) >> blkbits)
4241
		- map.m_lblk;
A
Amit Arora 已提交
4242
	/*
4243
	 * credits to insert 1 extent into extent tree
A
Amit Arora 已提交
4244
	 */
4245
	credits = ext4_chunk_trans_blocks(inode, max_blocks);
4246
	mutex_lock(&inode->i_mutex);
4247 4248 4249
	ret = inode_newsize_ok(inode, (len + offset));
	if (ret) {
		mutex_unlock(&inode->i_mutex);
4250
		trace_ext4_fallocate_exit(inode, offset, max_blocks, ret);
4251 4252
		return ret;
	}
A
Amit Arora 已提交
4253 4254
retry:
	while (ret >= 0 && ret < max_blocks) {
4255 4256
		map.m_lblk = map.m_lblk + ret;
		map.m_len = max_blocks = max_blocks - ret;
A
Amit Arora 已提交
4257 4258 4259 4260 4261
		handle = ext4_journal_start(inode, credits);
		if (IS_ERR(handle)) {
			ret = PTR_ERR(handle);
			break;
		}
4262
		ret = ext4_map_blocks(handle, inode, &map,
4263 4264
				      EXT4_GET_BLOCKS_CREATE_UNINIT_EXT |
				      EXT4_GET_BLOCKS_NO_NORMALIZE);
4265
		if (ret <= 0) {
4266 4267
#ifdef EXT4FS_DEBUG
			WARN_ON(ret <= 0);
4268
			printk(KERN_ERR "%s: ext4_ext_map_blocks "
4269
				    "returned error inode#%lu, block=%u, "
4270
				    "max_blocks=%u", __func__,
4271
				    inode->i_ino, map.m_lblk, max_blocks);
4272
#endif
A
Amit Arora 已提交
4273 4274 4275 4276
			ext4_mark_inode_dirty(handle, inode);
			ret2 = ext4_journal_stop(handle);
			break;
		}
4277
		if ((map.m_lblk + ret) >= (EXT4_BLOCK_ALIGN(offset + len,
4278 4279 4280
						blkbits) >> blkbits))
			new_size = offset + len;
		else
4281
			new_size = ((loff_t) map.m_lblk + ret) << blkbits;
A
Amit Arora 已提交
4282

4283
		ext4_falloc_update_inode(inode, mode, new_size,
4284
					 (map.m_flags & EXT4_MAP_NEW));
A
Amit Arora 已提交
4285 4286 4287 4288 4289
		ext4_mark_inode_dirty(handle, inode);
		ret2 = ext4_journal_stop(handle);
		if (ret2)
			break;
	}
4290 4291 4292
	if (ret == -ENOSPC &&
			ext4_should_retry_alloc(inode->i_sb, &retries)) {
		ret = 0;
A
Amit Arora 已提交
4293 4294
		goto retry;
	}
4295
	mutex_unlock(&inode->i_mutex);
4296 4297
	trace_ext4_fallocate_exit(inode, offset, max_blocks,
				ret > 0 ? ret2 : ret);
A
Amit Arora 已提交
4298 4299
	return ret > 0 ? ret2 : ret;
}
4300

4301 4302 4303 4304 4305 4306 4307 4308
/*
 * This function convert a range of blocks to written extents
 * The caller of this function will pass the start offset and the size.
 * all unwritten extents within this range will be converted to
 * written extents.
 *
 * This function is called from the direct IO end io call back
 * function, to convert the fallocated extents after IO is completed.
4309
 * Returns 0 on success.
4310 4311
 */
int ext4_convert_unwritten_extents(struct inode *inode, loff_t offset,
4312
				    ssize_t len)
4313 4314 4315 4316 4317
{
	handle_t *handle;
	unsigned int max_blocks;
	int ret = 0;
	int ret2 = 0;
4318
	struct ext4_map_blocks map;
4319 4320
	unsigned int credits, blkbits = inode->i_blkbits;

4321
	map.m_lblk = offset >> blkbits;
4322 4323 4324 4325
	/*
	 * We can't just convert len to max_blocks because
	 * If blocksize = 4096 offset = 3072 and len = 2048
	 */
4326 4327
	max_blocks = ((EXT4_BLOCK_ALIGN(len + offset, blkbits) >> blkbits) -
		      map.m_lblk);
4328 4329 4330 4331 4332
	/*
	 * credits to insert 1 extent into extent tree
	 */
	credits = ext4_chunk_trans_blocks(inode, max_blocks);
	while (ret >= 0 && ret < max_blocks) {
4333 4334
		map.m_lblk += ret;
		map.m_len = (max_blocks -= ret);
4335 4336 4337 4338 4339
		handle = ext4_journal_start(inode, credits);
		if (IS_ERR(handle)) {
			ret = PTR_ERR(handle);
			break;
		}
4340
		ret = ext4_map_blocks(handle, inode, &map,
4341
				      EXT4_GET_BLOCKS_IO_CONVERT_EXT);
4342 4343
		if (ret <= 0) {
			WARN_ON(ret <= 0);
4344
			printk(KERN_ERR "%s: ext4_ext_map_blocks "
4345 4346
				    "returned error inode#%lu, block=%u, "
				    "max_blocks=%u", __func__,
4347
				    inode->i_ino, map.m_lblk, map.m_len);
4348 4349 4350 4351 4352 4353 4354 4355
		}
		ext4_mark_inode_dirty(handle, inode);
		ret2 = ext4_journal_stop(handle);
		if (ret <= 0 || ret2 )
			break;
	}
	return ret > 0 ? ret2 : ret;
}
4356

4357 4358 4359
/*
 * Callback function called for each extent to gather FIEMAP information.
 */
4360
static int ext4_ext_fiemap_cb(struct inode *inode, ext4_lblk_t next,
4361 4362 4363 4364 4365 4366 4367
		       struct ext4_ext_cache *newex, struct ext4_extent *ex,
		       void *data)
{
	__u64	logical;
	__u64	physical;
	__u64	length;
	__u32	flags = 0;
4368 4369 4370
	int		ret = 0;
	struct fiemap_extent_info *fieinfo = data;
	unsigned char blksize_bits;
4371

4372 4373
	blksize_bits = inode->i_sb->s_blocksize_bits;
	logical = (__u64)newex->ec_block << blksize_bits;
4374

4375
	if (newex->ec_start == 0) {
4376 4377 4378 4379 4380 4381 4382 4383 4384 4385 4386 4387 4388 4389 4390 4391 4392 4393
		/*
		 * No extent in extent-tree contains block @newex->ec_start,
		 * then the block may stay in 1)a hole or 2)delayed-extent.
		 *
		 * Holes or delayed-extents are processed as follows.
		 * 1. lookup dirty pages with specified range in pagecache.
		 *    If no page is got, then there is no delayed-extent and
		 *    return with EXT_CONTINUE.
		 * 2. find the 1st mapped buffer,
		 * 3. check if the mapped buffer is both in the request range
		 *    and a delayed buffer. If not, there is no delayed-extent,
		 *    then return.
		 * 4. a delayed-extent is found, the extent will be collected.
		 */
		ext4_lblk_t	end = 0;
		pgoff_t		last_offset;
		pgoff_t		offset;
		pgoff_t		index;
4394
		pgoff_t		start_index = 0;
4395
		struct page	**pages = NULL;
4396
		struct buffer_head *bh = NULL;
4397 4398 4399 4400 4401 4402
		struct buffer_head *head = NULL;
		unsigned int nr_pages = PAGE_SIZE / sizeof(struct page *);

		pages = kmalloc(PAGE_SIZE, GFP_KERNEL);
		if (pages == NULL)
			return -ENOMEM;
4403 4404

		offset = logical >> PAGE_SHIFT;
4405 4406 4407 4408 4409 4410 4411 4412 4413 4414 4415 4416 4417 4418 4419 4420
repeat:
		last_offset = offset;
		head = NULL;
		ret = find_get_pages_tag(inode->i_mapping, &offset,
					PAGECACHE_TAG_DIRTY, nr_pages, pages);

		if (!(flags & FIEMAP_EXTENT_DELALLOC)) {
			/* First time, try to find a mapped buffer. */
			if (ret == 0) {
out:
				for (index = 0; index < ret; index++)
					page_cache_release(pages[index]);
				/* just a hole. */
				kfree(pages);
				return EXT_CONTINUE;
			}
4421
			index = 0;
4422

4423
next_page:
4424
			/* Try to find the 1st mapped buffer. */
4425
			end = ((__u64)pages[index]->index << PAGE_SHIFT) >>
4426
				  blksize_bits;
4427
			if (!page_has_buffers(pages[index]))
4428
				goto out;
4429
			head = page_buffers(pages[index]);
4430 4431
			if (!head)
				goto out;
4432

4433
			index++;
4434 4435
			bh = head;
			do {
4436 4437 4438 4439 4440 4441 4442 4443 4444 4445
				if (end >= newex->ec_block +
					newex->ec_len)
					/* The buffer is out of
					 * the request range.
					 */
					goto out;

				if (buffer_mapped(bh) &&
				    end >= newex->ec_block) {
					start_index = index - 1;
4446 4447 4448
					/* get the 1st mapped buffer. */
					goto found_mapped_buffer;
				}
4449

4450 4451 4452
				bh = bh->b_this_page;
				end++;
			} while (bh != head);
4453

4454 4455 4456 4457 4458 4459 4460 4461 4462 4463 4464
			/* No mapped buffer in the range found in this page,
			 * We need to look up next page.
			 */
			if (index >= ret) {
				/* There is no page left, but we need to limit
				 * newex->ec_len.
				 */
				newex->ec_len = end - newex->ec_block;
				goto out;
			}
			goto next_page;
4465
		} else {
4466 4467 4468 4469
			/*Find contiguous delayed buffers. */
			if (ret > 0 && pages[0]->index == last_offset)
				head = page_buffers(pages[0]);
			bh = head;
4470 4471
			index = 1;
			start_index = 0;
4472
		}
4473 4474 4475 4476 4477 4478 4479 4480 4481 4482 4483 4484 4485 4486 4487 4488 4489 4490 4491 4492 4493

found_mapped_buffer:
		if (bh != NULL && buffer_delay(bh)) {
			/* 1st or contiguous delayed buffer found. */
			if (!(flags & FIEMAP_EXTENT_DELALLOC)) {
				/*
				 * 1st delayed buffer found, record
				 * the start of extent.
				 */
				flags |= FIEMAP_EXTENT_DELALLOC;
				newex->ec_block = end;
				logical = (__u64)end << blksize_bits;
			}
			/* Find contiguous delayed buffers. */
			do {
				if (!buffer_delay(bh))
					goto found_delayed_extent;
				bh = bh->b_this_page;
				end++;
			} while (bh != head);

4494
			for (; index < ret; index++) {
4495 4496 4497 4498 4499 4500 4501 4502 4503
				if (!page_has_buffers(pages[index])) {
					bh = NULL;
					break;
				}
				head = page_buffers(pages[index]);
				if (!head) {
					bh = NULL;
					break;
				}
4504

4505
				if (pages[index]->index !=
4506 4507
				    pages[start_index]->index + index
				    - start_index) {
4508 4509 4510 4511 4512 4513 4514 4515 4516 4517 4518 4519 4520 4521 4522 4523 4524 4525 4526 4527 4528 4529 4530 4531 4532 4533 4534
					/* Blocks are not contiguous. */
					bh = NULL;
					break;
				}
				bh = head;
				do {
					if (!buffer_delay(bh))
						/* Delayed-extent ends. */
						goto found_delayed_extent;
					bh = bh->b_this_page;
					end++;
				} while (bh != head);
			}
		} else if (!(flags & FIEMAP_EXTENT_DELALLOC))
			/* a hole found. */
			goto out;

found_delayed_extent:
		newex->ec_len = min(end - newex->ec_block,
						(ext4_lblk_t)EXT_INIT_MAX_LEN);
		if (ret == nr_pages && bh != NULL &&
			newex->ec_len < EXT_INIT_MAX_LEN &&
			buffer_delay(bh)) {
			/* Have not collected an extent and continue. */
			for (index = 0; index < ret; index++)
				page_cache_release(pages[index]);
			goto repeat;
4535
		}
4536 4537 4538 4539

		for (index = 0; index < ret; index++)
			page_cache_release(pages[index]);
		kfree(pages);
4540 4541 4542 4543 4544 4545 4546 4547
	}

	physical = (__u64)newex->ec_start << blksize_bits;
	length =   (__u64)newex->ec_len << blksize_bits;

	if (ex && ext4_ext_is_uninitialized(ex))
		flags |= FIEMAP_EXTENT_UNWRITTEN;

4548
	if (next == EXT_MAX_BLOCKS)
4549 4550
		flags |= FIEMAP_EXTENT_LAST;

4551
	ret = fiemap_fill_next_extent(fieinfo, logical, physical,
4552
					length, flags);
4553 4554 4555
	if (ret < 0)
		return ret;
	if (ret == 1)
4556 4557 4558 4559 4560 4561
		return EXT_BREAK;
	return EXT_CONTINUE;
}
/* fiemap flags we can handle specified here */
#define EXT4_FIEMAP_FLAGS	(FIEMAP_FLAG_SYNC|FIEMAP_FLAG_XATTR)

A
Aneesh Kumar K.V 已提交
4562 4563
static int ext4_xattr_fiemap(struct inode *inode,
				struct fiemap_extent_info *fieinfo)
4564 4565 4566 4567 4568 4569 4570 4571
{
	__u64 physical = 0;
	__u64 length;
	__u32 flags = FIEMAP_EXTENT_LAST;
	int blockbits = inode->i_sb->s_blocksize_bits;
	int error = 0;

	/* in-inode? */
4572
	if (ext4_test_inode_state(inode, EXT4_STATE_XATTR)) {
4573 4574 4575 4576 4577 4578 4579 4580 4581 4582 4583 4584
		struct ext4_iloc iloc;
		int offset;	/* offset of xattr in inode */

		error = ext4_get_inode_loc(inode, &iloc);
		if (error)
			return error;
		physical = iloc.bh->b_blocknr << blockbits;
		offset = EXT4_GOOD_OLD_INODE_SIZE +
				EXT4_I(inode)->i_extra_isize;
		physical += offset;
		length = EXT4_SB(inode->i_sb)->s_inode_size - offset;
		flags |= FIEMAP_EXTENT_DATA_INLINE;
4585
		brelse(iloc.bh);
4586 4587 4588 4589 4590 4591 4592 4593 4594 4595 4596
	} else { /* external block */
		physical = EXT4_I(inode)->i_file_acl << blockbits;
		length = inode->i_sb->s_blocksize;
	}

	if (physical)
		error = fiemap_fill_next_extent(fieinfo, 0, physical,
						length, flags);
	return (error < 0 ? error : 0);
}

4597 4598 4599 4600 4601 4602 4603 4604 4605 4606 4607 4608 4609 4610 4611 4612 4613 4614 4615 4616 4617
/*
 * ext4_ext_punch_hole
 *
 * Punches a hole of "length" bytes in a file starting
 * at byte "offset"
 *
 * @inode:  The inode of the file to punch a hole in
 * @offset: The starting byte offset of the hole
 * @length: The length of the hole
 *
 * Returns the number of blocks removed or negative on err
 */
int ext4_ext_punch_hole(struct file *file, loff_t offset, loff_t length)
{
	struct inode *inode = file->f_path.dentry->d_inode;
	struct super_block *sb = inode->i_sb;
	struct ext4_ext_cache cache_ex;
	ext4_lblk_t first_block, last_block, num_blocks, iblock, max_blocks;
	struct address_space *mapping = inode->i_mapping;
	struct ext4_map_blocks map;
	handle_t *handle;
4618 4619
	loff_t first_page, last_page, page_len;
	loff_t first_page_offset, last_page_offset;
4620 4621
	int ret, credits, blocks_released, err = 0;

4622 4623 4624 4625 4626 4627 4628 4629 4630 4631 4632 4633 4634 4635
	/* No need to punch hole beyond i_size */
	if (offset >= inode->i_size)
		return 0;

	/*
	 * If the hole extends beyond i_size, set the hole
	 * to end after the page that contains i_size
	 */
	if (offset + length > inode->i_size) {
		length = inode->i_size +
		   PAGE_CACHE_SIZE - (inode->i_size & (PAGE_CACHE_SIZE - 1)) -
		   offset;
	}

4636 4637 4638 4639 4640 4641 4642 4643 4644 4645 4646 4647 4648 4649 4650 4651
	first_block = (offset + sb->s_blocksize - 1) >>
		EXT4_BLOCK_SIZE_BITS(sb);
	last_block = (offset + length) >> EXT4_BLOCK_SIZE_BITS(sb);

	first_page = (offset + PAGE_CACHE_SIZE - 1) >> PAGE_CACHE_SHIFT;
	last_page = (offset + length) >> PAGE_CACHE_SHIFT;

	first_page_offset = first_page << PAGE_CACHE_SHIFT;
	last_page_offset = last_page << PAGE_CACHE_SHIFT;

	/*
	 * Write out all dirty pages to avoid race conditions
	 * Then release them.
	 */
	if (mapping->nrpages && mapping_tagged(mapping, PAGECACHE_TAG_DIRTY)) {
		err = filemap_write_and_wait_range(mapping,
4652
			offset, offset + length - 1);
4653

4654 4655
		if (err)
			return err;
4656 4657 4658 4659 4660 4661 4662 4663 4664 4665 4666 4667 4668 4669 4670 4671 4672 4673 4674 4675 4676
	}

	/* Now release the pages */
	if (last_page_offset > first_page_offset) {
		truncate_inode_pages_range(mapping, first_page_offset,
					   last_page_offset-1);
	}

	/* finish any pending end_io work */
	ext4_flush_completed_IO(inode);

	credits = ext4_writepage_trans_blocks(inode);
	handle = ext4_journal_start(inode, credits);
	if (IS_ERR(handle))
		return PTR_ERR(handle);

	err = ext4_orphan_add(handle, inode);
	if (err)
		goto out;

	/*
4677 4678 4679 4680
	 * Now we need to zero out the non-page-aligned data in the
	 * pages at the start and tail of the hole, and unmap the buffer
	 * heads for the block aligned regions of the page that were
	 * completely zeroed.
4681
	 */
4682 4683 4684 4685 4686 4687 4688 4689 4690 4691 4692 4693 4694 4695 4696 4697 4698 4699 4700 4701 4702 4703 4704 4705 4706 4707 4708 4709 4710 4711 4712 4713 4714
	if (first_page > last_page) {
		/*
		 * If the file space being truncated is contained within a page
		 * just zero out and unmap the middle of that page
		 */
		err = ext4_discard_partial_page_buffers(handle,
			mapping, offset, length, 0);

		if (err)
			goto out;
	} else {
		/*
		 * zero out and unmap the partial page that contains
		 * the start of the hole
		 */
		page_len  = first_page_offset - offset;
		if (page_len > 0) {
			err = ext4_discard_partial_page_buffers(handle, mapping,
						   offset, page_len, 0);
			if (err)
				goto out;
		}

		/*
		 * zero out and unmap the partial page that contains
		 * the end of the hole
		 */
		page_len = offset + length - last_page_offset;
		if (page_len > 0) {
			err = ext4_discard_partial_page_buffers(handle, mapping,
					last_page_offset, page_len, 0);
			if (err)
				goto out;
4715 4716 4717
		}
	}

4718 4719 4720 4721 4722 4723 4724 4725 4726 4727 4728 4729 4730 4731 4732 4733 4734 4735 4736 4737

	/*
	 * If i_size is contained in the last page, we need to
	 * unmap and zero the partial page after i_size
	 */
	if (inode->i_size >> PAGE_CACHE_SHIFT == last_page &&
	   inode->i_size % PAGE_CACHE_SIZE != 0) {

		page_len = PAGE_CACHE_SIZE -
			(inode->i_size & (PAGE_CACHE_SIZE - 1));

		if (page_len > 0) {
			err = ext4_discard_partial_page_buffers(handle,
			  mapping, inode->i_size, page_len, 0);

			if (err)
				goto out;
		}
	}

4738 4739 4740 4741 4742 4743 4744 4745 4746 4747 4748 4749 4750 4751 4752 4753 4754 4755 4756 4757 4758 4759 4760 4761 4762 4763 4764 4765 4766 4767 4768 4769 4770 4771 4772 4773 4774 4775 4776 4777 4778 4779 4780 4781 4782 4783 4784 4785 4786 4787 4788 4789 4790 4791 4792 4793 4794 4795 4796 4797 4798 4799 4800 4801 4802 4803 4804 4805 4806 4807 4808 4809 4810 4811 4812 4813 4814 4815 4816 4817
	/* If there are no blocks to remove, return now */
	if (first_block >= last_block)
		goto out;

	down_write(&EXT4_I(inode)->i_data_sem);
	ext4_ext_invalidate_cache(inode);
	ext4_discard_preallocations(inode);

	/*
	 * Loop over all the blocks and identify blocks
	 * that need to be punched out
	 */
	iblock = first_block;
	blocks_released = 0;
	while (iblock < last_block) {
		max_blocks = last_block - iblock;
		num_blocks = 1;
		memset(&map, 0, sizeof(map));
		map.m_lblk = iblock;
		map.m_len = max_blocks;
		ret = ext4_ext_map_blocks(handle, inode, &map,
			EXT4_GET_BLOCKS_PUNCH_OUT_EXT);

		if (ret > 0) {
			blocks_released += ret;
			num_blocks = ret;
		} else if (ret == 0) {
			/*
			 * If map blocks could not find the block,
			 * then it is in a hole.  If the hole was
			 * not already cached, then map blocks should
			 * put it in the cache.  So we can get the hole
			 * out of the cache
			 */
			memset(&cache_ex, 0, sizeof(cache_ex));
			if ((ext4_ext_check_cache(inode, iblock, &cache_ex)) &&
				!cache_ex.ec_start) {

				/* The hole is cached */
				num_blocks = cache_ex.ec_block +
				cache_ex.ec_len - iblock;

			} else {
				/* The block could not be identified */
				err = -EIO;
				break;
			}
		} else {
			/* Map blocks error */
			err = ret;
			break;
		}

		if (num_blocks == 0) {
			/* This condition should never happen */
			ext_debug("Block lookup failed");
			err = -EIO;
			break;
		}

		iblock += num_blocks;
	}

	if (blocks_released > 0) {
		ext4_ext_invalidate_cache(inode);
		ext4_discard_preallocations(inode);
	}

	if (IS_SYNC(inode))
		ext4_handle_sync(handle);

	up_write(&EXT4_I(inode)->i_data_sem);

out:
	ext4_orphan_del(handle, inode);
	inode->i_mtime = inode->i_ctime = ext4_current_time(inode);
	ext4_mark_inode_dirty(handle, inode);
	ext4_journal_stop(handle);
	return err;
}
4818 4819 4820 4821 4822 4823 4824
int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
		__u64 start, __u64 len)
{
	ext4_lblk_t start_blk;
	int error = 0;

	/* fallback to generic here if not in extents fmt */
4825
	if (!(ext4_test_inode_flag(inode, EXT4_INODE_EXTENTS)))
4826 4827 4828 4829 4830 4831 4832 4833 4834
		return generic_block_fiemap(inode, fieinfo, start, len,
			ext4_get_block);

	if (fiemap_check_flags(fieinfo, EXT4_FIEMAP_FLAGS))
		return -EBADR;

	if (fieinfo->fi_flags & FIEMAP_FLAG_XATTR) {
		error = ext4_xattr_fiemap(inode, fieinfo);
	} else {
4835 4836 4837
		ext4_lblk_t len_blks;
		__u64 last_blk;

4838
		start_blk = start >> inode->i_sb->s_blocksize_bits;
4839
		last_blk = (start + len - 1) >> inode->i_sb->s_blocksize_bits;
4840 4841
		if (last_blk >= EXT_MAX_BLOCKS)
			last_blk = EXT_MAX_BLOCKS-1;
4842
		len_blks = ((ext4_lblk_t) last_blk) - start_blk + 1;
4843 4844 4845 4846 4847 4848 4849 4850 4851 4852 4853

		/*
		 * Walk the extent tree gathering extent information.
		 * ext4_ext_fiemap_cb will push extents back to user.
		 */
		error = ext4_ext_walk_space(inode, start_blk, len_blks,
					  ext4_ext_fiemap_cb, fieinfo);
	}

	return error;
}