xref: /openbmc/linux/fs/ext4/extents.c (revision cc19db8b)
1 // SPDX-License-Identifier: GPL-2.0
2 /*
3  * Copyright (c) 2003-2006, Cluster File Systems, Inc, info@clusterfs.com
4  * Written by Alex Tomas <alex@clusterfs.com>
5  *
6  * Architecture independence:
7  *   Copyright (c) 2005, Bull S.A.
8  *   Written by Pierre Peiffer <pierre.peiffer@bull.net>
9  */
10 
11 /*
12  * Extents support for EXT4
13  *
14  * TODO:
15  *   - ext4*_error() should be used in some situations
16  *   - analyze all BUG()/BUG_ON(), use -EIO where appropriate
17  *   - smart tree reduction
18  */
19 
20 #include <linux/fs.h>
21 #include <linux/time.h>
22 #include <linux/jbd2.h>
23 #include <linux/highuid.h>
24 #include <linux/pagemap.h>
25 #include <linux/quotaops.h>
26 #include <linux/string.h>
27 #include <linux/slab.h>
28 #include <linux/uaccess.h>
29 #include <linux/fiemap.h>
30 #include <linux/iomap.h>
31 #include <linux/sched/mm.h>
32 #include "ext4_jbd2.h"
33 #include "ext4_extents.h"
34 #include "xattr.h"
35 
36 #include <trace/events/ext4.h>
37 
38 /*
39  * used by extent splitting.
40  */
41 #define EXT4_EXT_MAY_ZEROOUT	0x1  /* safe to zeroout if split fails \
42 					due to ENOSPC */
43 #define EXT4_EXT_MARK_UNWRIT1	0x2  /* mark first half unwritten */
44 #define EXT4_EXT_MARK_UNWRIT2	0x4  /* mark second half unwritten */
45 
46 #define EXT4_EXT_DATA_VALID1	0x8  /* first half contains valid data */
47 #define EXT4_EXT_DATA_VALID2	0x10 /* second half contains valid data */
48 
49 static __le32 ext4_extent_block_csum(struct inode *inode,
50 				     struct ext4_extent_header *eh)
51 {
52 	struct ext4_inode_info *ei = EXT4_I(inode);
53 	struct ext4_sb_info *sbi = EXT4_SB(inode->i_sb);
54 	__u32 csum;
55 
56 	csum = ext4_chksum(sbi, ei->i_csum_seed, (__u8 *)eh,
57 			   EXT4_EXTENT_TAIL_OFFSET(eh));
58 	return cpu_to_le32(csum);
59 }
60 
61 static int ext4_extent_block_csum_verify(struct inode *inode,
62 					 struct ext4_extent_header *eh)
63 {
64 	struct ext4_extent_tail *et;
65 
66 	if (!ext4_has_metadata_csum(inode->i_sb))
67 		return 1;
68 
69 	et = find_ext4_extent_tail(eh);
70 	if (et->et_checksum != ext4_extent_block_csum(inode, eh))
71 		return 0;
72 	return 1;
73 }
74 
75 static void ext4_extent_block_csum_set(struct inode *inode,
76 				       struct ext4_extent_header *eh)
77 {
78 	struct ext4_extent_tail *et;
79 
80 	if (!ext4_has_metadata_csum(inode->i_sb))
81 		return;
82 
83 	et = find_ext4_extent_tail(eh);
84 	et->et_checksum = ext4_extent_block_csum(inode, eh);
85 }
86 
87 static int ext4_split_extent_at(handle_t *handle,
88 			     struct inode *inode,
89 			     struct ext4_ext_path **ppath,
90 			     ext4_lblk_t split,
91 			     int split_flag,
92 			     int flags);
93 
94 static int ext4_ext_trunc_restart_fn(struct inode *inode, int *dropped)
95 {
96 	/*
97 	 * Drop i_data_sem to avoid deadlock with ext4_map_blocks.  At this
98 	 * moment, get_block can be called only for blocks inside i_size since
99 	 * page cache has been already dropped and writes are blocked by
100 	 * i_rwsem. So we can safely drop the i_data_sem here.
101 	 */
102 	BUG_ON(EXT4_JOURNAL(inode) == NULL);
103 	ext4_discard_preallocations(inode, 0);
104 	up_write(&EXT4_I(inode)->i_data_sem);
105 	*dropped = 1;
106 	return 0;
107 }
108 
109 /*
110  * Make sure 'handle' has at least 'check_cred' credits. If not, restart
111  * transaction with 'restart_cred' credits. The function drops i_data_sem
112  * when restarting transaction and gets it after transaction is restarted.
113  *
114  * The function returns 0 on success, 1 if transaction had to be restarted,
115  * and < 0 in case of fatal error.
116  */
117 int ext4_datasem_ensure_credits(handle_t *handle, struct inode *inode,
118 				int check_cred, int restart_cred,
119 				int revoke_cred)
120 {
121 	int ret;
122 	int dropped = 0;
123 
124 	ret = ext4_journal_ensure_credits_fn(handle, check_cred, restart_cred,
125 		revoke_cred, ext4_ext_trunc_restart_fn(inode, &dropped));
126 	if (dropped)
127 		down_write(&EXT4_I(inode)->i_data_sem);
128 	return ret;
129 }
130 
131 /*
132  * could return:
133  *  - EROFS
134  *  - ENOMEM
135  */
136 static int ext4_ext_get_access(handle_t *handle, struct inode *inode,
137 				struct ext4_ext_path *path)
138 {
139 	int err = 0;
140 
141 	if (path->p_bh) {
142 		/* path points to block */
143 		BUFFER_TRACE(path->p_bh, "get_write_access");
144 		err = ext4_journal_get_write_access(handle, inode->i_sb,
145 						    path->p_bh, EXT4_JTR_NONE);
146 		/*
147 		 * The extent buffer's verified bit will be set again in
148 		 * __ext4_ext_dirty(). We could leave an inconsistent
149 		 * buffer if the extents updating procudure break off du
150 		 * to some error happens, force to check it again.
151 		 */
152 		if (!err)
153 			clear_buffer_verified(path->p_bh);
154 	}
155 	/* path points to leaf/index in inode body */
156 	/* we use in-core data, no need to protect them */
157 	return err;
158 }
159 
160 /*
161  * could return:
162  *  - EROFS
163  *  - ENOMEM
164  *  - EIO
165  */
166 static int __ext4_ext_dirty(const char *where, unsigned int line,
167 			    handle_t *handle, struct inode *inode,
168 			    struct ext4_ext_path *path)
169 {
170 	int err;
171 
172 	WARN_ON(!rwsem_is_locked(&EXT4_I(inode)->i_data_sem));
173 	if (path->p_bh) {
174 		ext4_extent_block_csum_set(inode, ext_block_hdr(path->p_bh));
175 		/* path points to block */
176 		err = __ext4_handle_dirty_metadata(where, line, handle,
177 						   inode, path->p_bh);
178 		/* Extents updating done, re-set verified flag */
179 		if (!err)
180 			set_buffer_verified(path->p_bh);
181 	} else {
182 		/* path points to leaf/index in inode body */
183 		err = ext4_mark_inode_dirty(handle, inode);
184 	}
185 	return err;
186 }
187 
188 #define ext4_ext_dirty(handle, inode, path) \
189 		__ext4_ext_dirty(__func__, __LINE__, (handle), (inode), (path))
190 
191 static ext4_fsblk_t ext4_ext_find_goal(struct inode *inode,
192 			      struct ext4_ext_path *path,
193 			      ext4_lblk_t block)
194 {
195 	if (path) {
196 		int depth = path->p_depth;
197 		struct ext4_extent *ex;
198 
199 		/*
200 		 * Try to predict block placement assuming that we are
201 		 * filling in a file which will eventually be
202 		 * non-sparse --- i.e., in the case of libbfd writing
203 		 * an ELF object sections out-of-order but in a way
204 		 * the eventually results in a contiguous object or
205 		 * executable file, or some database extending a table
206 		 * space file.  However, this is actually somewhat
207 		 * non-ideal if we are writing a sparse file such as
208 		 * qemu or KVM writing a raw image file that is going
209 		 * to stay fairly sparse, since it will end up
210 		 * fragmenting the file system's free space.  Maybe we
211 		 * should have some hueristics or some way to allow
212 		 * userspace to pass a hint to file system,
213 		 * especially if the latter case turns out to be
214 		 * common.
215 		 */
216 		ex = path[depth].p_ext;
217 		if (ex) {
218 			ext4_fsblk_t ext_pblk = ext4_ext_pblock(ex);
219 			ext4_lblk_t ext_block = le32_to_cpu(ex->ee_block);
220 
221 			if (block > ext_block)
222 				return ext_pblk + (block - ext_block);
223 			else
224 				return ext_pblk - (ext_block - block);
225 		}
226 
227 		/* it looks like index is empty;
228 		 * try to find starting block from index itself */
229 		if (path[depth].p_bh)
230 			return path[depth].p_bh->b_blocknr;
231 	}
232 
233 	/* OK. use inode's group */
234 	return ext4_inode_to_goal_block(inode);
235 }
236 
237 /*
238  * Allocation for a meta data block
239  */
240 static ext4_fsblk_t
241 ext4_ext_new_meta_block(handle_t *handle, struct inode *inode,
242 			struct ext4_ext_path *path,
243 			struct ext4_extent *ex, int *err, unsigned int flags)
244 {
245 	ext4_fsblk_t goal, newblock;
246 
247 	goal = ext4_ext_find_goal(inode, path, le32_to_cpu(ex->ee_block));
248 	newblock = ext4_new_meta_blocks(handle, inode, goal, flags,
249 					NULL, err);
250 	return newblock;
251 }
252 
253 static inline int ext4_ext_space_block(struct inode *inode, int check)
254 {
255 	int size;
256 
257 	size = (inode->i_sb->s_blocksize - sizeof(struct ext4_extent_header))
258 			/ sizeof(struct ext4_extent);
259 #ifdef AGGRESSIVE_TEST
260 	if (!check && size > 6)
261 		size = 6;
262 #endif
263 	return size;
264 }
265 
266 static inline int ext4_ext_space_block_idx(struct inode *inode, int check)
267 {
268 	int size;
269 
270 	size = (inode->i_sb->s_blocksize - sizeof(struct ext4_extent_header))
271 			/ sizeof(struct ext4_extent_idx);
272 #ifdef AGGRESSIVE_TEST
273 	if (!check && size > 5)
274 		size = 5;
275 #endif
276 	return size;
277 }
278 
279 static inline int ext4_ext_space_root(struct inode *inode, int check)
280 {
281 	int size;
282 
283 	size = sizeof(EXT4_I(inode)->i_data);
284 	size -= sizeof(struct ext4_extent_header);
285 	size /= sizeof(struct ext4_extent);
286 #ifdef AGGRESSIVE_TEST
287 	if (!check && size > 3)
288 		size = 3;
289 #endif
290 	return size;
291 }
292 
293 static inline int ext4_ext_space_root_idx(struct inode *inode, int check)
294 {
295 	int size;
296 
297 	size = sizeof(EXT4_I(inode)->i_data);
298 	size -= sizeof(struct ext4_extent_header);
299 	size /= sizeof(struct ext4_extent_idx);
300 #ifdef AGGRESSIVE_TEST
301 	if (!check && size > 4)
302 		size = 4;
303 #endif
304 	return size;
305 }
306 
307 static inline int
308 ext4_force_split_extent_at(handle_t *handle, struct inode *inode,
309 			   struct ext4_ext_path **ppath, ext4_lblk_t lblk,
310 			   int nofail)
311 {
312 	struct ext4_ext_path *path = *ppath;
313 	int unwritten = ext4_ext_is_unwritten(path[path->p_depth].p_ext);
314 	int flags = EXT4_EX_NOCACHE | EXT4_GET_BLOCKS_PRE_IO;
315 
316 	if (nofail)
317 		flags |= EXT4_GET_BLOCKS_METADATA_NOFAIL | EXT4_EX_NOFAIL;
318 
319 	return ext4_split_extent_at(handle, inode, ppath, lblk, unwritten ?
320 			EXT4_EXT_MARK_UNWRIT1|EXT4_EXT_MARK_UNWRIT2 : 0,
321 			flags);
322 }
323 
324 static int
325 ext4_ext_max_entries(struct inode *inode, int depth)
326 {
327 	int max;
328 
329 	if (depth == ext_depth(inode)) {
330 		if (depth == 0)
331 			max = ext4_ext_space_root(inode, 1);
332 		else
333 			max = ext4_ext_space_root_idx(inode, 1);
334 	} else {
335 		if (depth == 0)
336 			max = ext4_ext_space_block(inode, 1);
337 		else
338 			max = ext4_ext_space_block_idx(inode, 1);
339 	}
340 
341 	return max;
342 }
343 
344 static int ext4_valid_extent(struct inode *inode, struct ext4_extent *ext)
345 {
346 	ext4_fsblk_t block = ext4_ext_pblock(ext);
347 	int len = ext4_ext_get_actual_len(ext);
348 	ext4_lblk_t lblock = le32_to_cpu(ext->ee_block);
349 
350 	/*
351 	 * We allow neither:
352 	 *  - zero length
353 	 *  - overflow/wrap-around
354 	 */
355 	if (lblock + len <= lblock)
356 		return 0;
357 	return ext4_inode_block_valid(inode, block, len);
358 }
359 
360 static int ext4_valid_extent_idx(struct inode *inode,
361 				struct ext4_extent_idx *ext_idx)
362 {
363 	ext4_fsblk_t block = ext4_idx_pblock(ext_idx);
364 
365 	return ext4_inode_block_valid(inode, block, 1);
366 }
367 
368 static int ext4_valid_extent_entries(struct inode *inode,
369 				     struct ext4_extent_header *eh,
370 				     ext4_lblk_t lblk, ext4_fsblk_t *pblk,
371 				     int depth)
372 {
373 	unsigned short entries;
374 	ext4_lblk_t lblock = 0;
375 	ext4_lblk_t prev = 0;
376 
377 	if (eh->eh_entries == 0)
378 		return 1;
379 
380 	entries = le16_to_cpu(eh->eh_entries);
381 
382 	if (depth == 0) {
383 		/* leaf entries */
384 		struct ext4_extent *ext = EXT_FIRST_EXTENT(eh);
385 
386 		/*
387 		 * The logical block in the first entry should equal to
388 		 * the number in the index block.
389 		 */
390 		if (depth != ext_depth(inode) &&
391 		    lblk != le32_to_cpu(ext->ee_block))
392 			return 0;
393 		while (entries) {
394 			if (!ext4_valid_extent(inode, ext))
395 				return 0;
396 
397 			/* Check for overlapping extents */
398 			lblock = le32_to_cpu(ext->ee_block);
399 			if ((lblock <= prev) && prev) {
400 				*pblk = ext4_ext_pblock(ext);
401 				return 0;
402 			}
403 			prev = lblock + ext4_ext_get_actual_len(ext) - 1;
404 			ext++;
405 			entries--;
406 		}
407 	} else {
408 		struct ext4_extent_idx *ext_idx = EXT_FIRST_INDEX(eh);
409 
410 		/*
411 		 * The logical block in the first entry should equal to
412 		 * the number in the parent index block.
413 		 */
414 		if (depth != ext_depth(inode) &&
415 		    lblk != le32_to_cpu(ext_idx->ei_block))
416 			return 0;
417 		while (entries) {
418 			if (!ext4_valid_extent_idx(inode, ext_idx))
419 				return 0;
420 
421 			/* Check for overlapping index extents */
422 			lblock = le32_to_cpu(ext_idx->ei_block);
423 			if ((lblock <= prev) && prev) {
424 				*pblk = ext4_idx_pblock(ext_idx);
425 				return 0;
426 			}
427 			ext_idx++;
428 			entries--;
429 			prev = lblock;
430 		}
431 	}
432 	return 1;
433 }
434 
435 static int __ext4_ext_check(const char *function, unsigned int line,
436 			    struct inode *inode, struct ext4_extent_header *eh,
437 			    int depth, ext4_fsblk_t pblk, ext4_lblk_t lblk)
438 {
439 	const char *error_msg;
440 	int max = 0, err = -EFSCORRUPTED;
441 
442 	if (unlikely(eh->eh_magic != EXT4_EXT_MAGIC)) {
443 		error_msg = "invalid magic";
444 		goto corrupted;
445 	}
446 	if (unlikely(le16_to_cpu(eh->eh_depth) != depth)) {
447 		error_msg = "unexpected eh_depth";
448 		goto corrupted;
449 	}
450 	if (unlikely(eh->eh_max == 0)) {
451 		error_msg = "invalid eh_max";
452 		goto corrupted;
453 	}
454 	max = ext4_ext_max_entries(inode, depth);
455 	if (unlikely(le16_to_cpu(eh->eh_max) > max)) {
456 		error_msg = "too large eh_max";
457 		goto corrupted;
458 	}
459 	if (unlikely(le16_to_cpu(eh->eh_entries) > le16_to_cpu(eh->eh_max))) {
460 		error_msg = "invalid eh_entries";
461 		goto corrupted;
462 	}
463 	if (!ext4_valid_extent_entries(inode, eh, lblk, &pblk, depth)) {
464 		error_msg = "invalid extent entries";
465 		goto corrupted;
466 	}
467 	if (unlikely(depth > 32)) {
468 		error_msg = "too large eh_depth";
469 		goto corrupted;
470 	}
471 	/* Verify checksum on non-root extent tree nodes */
472 	if (ext_depth(inode) != depth &&
473 	    !ext4_extent_block_csum_verify(inode, eh)) {
474 		error_msg = "extent tree corrupted";
475 		err = -EFSBADCRC;
476 		goto corrupted;
477 	}
478 	return 0;
479 
480 corrupted:
481 	ext4_error_inode_err(inode, function, line, 0, -err,
482 			     "pblk %llu bad header/extent: %s - magic %x, "
483 			     "entries %u, max %u(%u), depth %u(%u)",
484 			     (unsigned long long) pblk, error_msg,
485 			     le16_to_cpu(eh->eh_magic),
486 			     le16_to_cpu(eh->eh_entries),
487 			     le16_to_cpu(eh->eh_max),
488 			     max, le16_to_cpu(eh->eh_depth), depth);
489 	return err;
490 }
491 
492 #define ext4_ext_check(inode, eh, depth, pblk)			\
493 	__ext4_ext_check(__func__, __LINE__, (inode), (eh), (depth), (pblk), 0)
494 
495 int ext4_ext_check_inode(struct inode *inode)
496 {
497 	return ext4_ext_check(inode, ext_inode_hdr(inode), ext_depth(inode), 0);
498 }
499 
500 static void ext4_cache_extents(struct inode *inode,
501 			       struct ext4_extent_header *eh)
502 {
503 	struct ext4_extent *ex = EXT_FIRST_EXTENT(eh);
504 	ext4_lblk_t prev = 0;
505 	int i;
506 
507 	for (i = le16_to_cpu(eh->eh_entries); i > 0; i--, ex++) {
508 		unsigned int status = EXTENT_STATUS_WRITTEN;
509 		ext4_lblk_t lblk = le32_to_cpu(ex->ee_block);
510 		int len = ext4_ext_get_actual_len(ex);
511 
512 		if (prev && (prev != lblk))
513 			ext4_es_cache_extent(inode, prev, lblk - prev, ~0,
514 					     EXTENT_STATUS_HOLE);
515 
516 		if (ext4_ext_is_unwritten(ex))
517 			status = EXTENT_STATUS_UNWRITTEN;
518 		ext4_es_cache_extent(inode, lblk, len,
519 				     ext4_ext_pblock(ex), status);
520 		prev = lblk + len;
521 	}
522 }
523 
524 static struct buffer_head *
525 __read_extent_tree_block(const char *function, unsigned int line,
526 			 struct inode *inode, struct ext4_extent_idx *idx,
527 			 int depth, int flags)
528 {
529 	struct buffer_head		*bh;
530 	int				err;
531 	gfp_t				gfp_flags = __GFP_MOVABLE | GFP_NOFS;
532 	ext4_fsblk_t			pblk;
533 
534 	if (flags & EXT4_EX_NOFAIL)
535 		gfp_flags |= __GFP_NOFAIL;
536 
537 	pblk = ext4_idx_pblock(idx);
538 	bh = sb_getblk_gfp(inode->i_sb, pblk, gfp_flags);
539 	if (unlikely(!bh))
540 		return ERR_PTR(-ENOMEM);
541 
542 	if (!bh_uptodate_or_lock(bh)) {
543 		trace_ext4_ext_load_extent(inode, pblk, _RET_IP_);
544 		err = ext4_read_bh(bh, 0, NULL);
545 		if (err < 0)
546 			goto errout;
547 	}
548 	if (buffer_verified(bh) && !(flags & EXT4_EX_FORCE_CACHE))
549 		return bh;
550 	err = __ext4_ext_check(function, line, inode, ext_block_hdr(bh),
551 			       depth, pblk, le32_to_cpu(idx->ei_block));
552 	if (err)
553 		goto errout;
554 	set_buffer_verified(bh);
555 	/*
556 	 * If this is a leaf block, cache all of its entries
557 	 */
558 	if (!(flags & EXT4_EX_NOCACHE) && depth == 0) {
559 		struct ext4_extent_header *eh = ext_block_hdr(bh);
560 		ext4_cache_extents(inode, eh);
561 	}
562 	return bh;
563 errout:
564 	put_bh(bh);
565 	return ERR_PTR(err);
566 
567 }
568 
569 #define read_extent_tree_block(inode, idx, depth, flags)		\
570 	__read_extent_tree_block(__func__, __LINE__, (inode), (idx),	\
571 				 (depth), (flags))
572 
573 /*
574  * This function is called to cache a file's extent information in the
575  * extent status tree
576  */
577 int ext4_ext_precache(struct inode *inode)
578 {
579 	struct ext4_inode_info *ei = EXT4_I(inode);
580 	struct ext4_ext_path *path = NULL;
581 	struct buffer_head *bh;
582 	int i = 0, depth, ret = 0;
583 
584 	if (!ext4_test_inode_flag(inode, EXT4_INODE_EXTENTS))
585 		return 0;	/* not an extent-mapped inode */
586 
587 	down_read(&ei->i_data_sem);
588 	depth = ext_depth(inode);
589 
590 	/* Don't cache anything if there are no external extent blocks */
591 	if (!depth) {
592 		up_read(&ei->i_data_sem);
593 		return ret;
594 	}
595 
596 	path = kcalloc(depth + 1, sizeof(struct ext4_ext_path),
597 		       GFP_NOFS);
598 	if (path == NULL) {
599 		up_read(&ei->i_data_sem);
600 		return -ENOMEM;
601 	}
602 
603 	path[0].p_hdr = ext_inode_hdr(inode);
604 	ret = ext4_ext_check(inode, path[0].p_hdr, depth, 0);
605 	if (ret)
606 		goto out;
607 	path[0].p_idx = EXT_FIRST_INDEX(path[0].p_hdr);
608 	while (i >= 0) {
609 		/*
610 		 * If this is a leaf block or we've reached the end of
611 		 * the index block, go up
612 		 */
613 		if ((i == depth) ||
614 		    path[i].p_idx > EXT_LAST_INDEX(path[i].p_hdr)) {
615 			brelse(path[i].p_bh);
616 			path[i].p_bh = NULL;
617 			i--;
618 			continue;
619 		}
620 		bh = read_extent_tree_block(inode, path[i].p_idx++,
621 					    depth - i - 1,
622 					    EXT4_EX_FORCE_CACHE);
623 		if (IS_ERR(bh)) {
624 			ret = PTR_ERR(bh);
625 			break;
626 		}
627 		i++;
628 		path[i].p_bh = bh;
629 		path[i].p_hdr = ext_block_hdr(bh);
630 		path[i].p_idx = EXT_FIRST_INDEX(path[i].p_hdr);
631 	}
632 	ext4_set_inode_state(inode, EXT4_STATE_EXT_PRECACHED);
633 out:
634 	up_read(&ei->i_data_sem);
635 	ext4_ext_drop_refs(path);
636 	kfree(path);
637 	return ret;
638 }
639 
640 #ifdef EXT_DEBUG
641 static void ext4_ext_show_path(struct inode *inode, struct ext4_ext_path *path)
642 {
643 	int k, l = path->p_depth;
644 
645 	ext_debug(inode, "path:");
646 	for (k = 0; k <= l; k++, path++) {
647 		if (path->p_idx) {
648 			ext_debug(inode, "  %d->%llu",
649 				  le32_to_cpu(path->p_idx->ei_block),
650 				  ext4_idx_pblock(path->p_idx));
651 		} else if (path->p_ext) {
652 			ext_debug(inode, "  %d:[%d]%d:%llu ",
653 				  le32_to_cpu(path->p_ext->ee_block),
654 				  ext4_ext_is_unwritten(path->p_ext),
655 				  ext4_ext_get_actual_len(path->p_ext),
656 				  ext4_ext_pblock(path->p_ext));
657 		} else
658 			ext_debug(inode, "  []");
659 	}
660 	ext_debug(inode, "\n");
661 }
662 
663 static void ext4_ext_show_leaf(struct inode *inode, struct ext4_ext_path *path)
664 {
665 	int depth = ext_depth(inode);
666 	struct ext4_extent_header *eh;
667 	struct ext4_extent *ex;
668 	int i;
669 
670 	if (!path)
671 		return;
672 
673 	eh = path[depth].p_hdr;
674 	ex = EXT_FIRST_EXTENT(eh);
675 
676 	ext_debug(inode, "Displaying leaf extents\n");
677 
678 	for (i = 0; i < le16_to_cpu(eh->eh_entries); i++, ex++) {
679 		ext_debug(inode, "%d:[%d]%d:%llu ", le32_to_cpu(ex->ee_block),
680 			  ext4_ext_is_unwritten(ex),
681 			  ext4_ext_get_actual_len(ex), ext4_ext_pblock(ex));
682 	}
683 	ext_debug(inode, "\n");
684 }
685 
686 static void ext4_ext_show_move(struct inode *inode, struct ext4_ext_path *path,
687 			ext4_fsblk_t newblock, int level)
688 {
689 	int depth = ext_depth(inode);
690 	struct ext4_extent *ex;
691 
692 	if (depth != level) {
693 		struct ext4_extent_idx *idx;
694 		idx = path[level].p_idx;
695 		while (idx <= EXT_MAX_INDEX(path[level].p_hdr)) {
696 			ext_debug(inode, "%d: move %d:%llu in new index %llu\n",
697 				  level, le32_to_cpu(idx->ei_block),
698 				  ext4_idx_pblock(idx), newblock);
699 			idx++;
700 		}
701 
702 		return;
703 	}
704 
705 	ex = path[depth].p_ext;
706 	while (ex <= EXT_MAX_EXTENT(path[depth].p_hdr)) {
707 		ext_debug(inode, "move %d:%llu:[%d]%d in new leaf %llu\n",
708 				le32_to_cpu(ex->ee_block),
709 				ext4_ext_pblock(ex),
710 				ext4_ext_is_unwritten(ex),
711 				ext4_ext_get_actual_len(ex),
712 				newblock);
713 		ex++;
714 	}
715 }
716 
717 #else
718 #define ext4_ext_show_path(inode, path)
719 #define ext4_ext_show_leaf(inode, path)
720 #define ext4_ext_show_move(inode, path, newblock, level)
721 #endif
722 
723 void ext4_ext_drop_refs(struct ext4_ext_path *path)
724 {
725 	int depth, i;
726 
727 	if (!path)
728 		return;
729 	depth = path->p_depth;
730 	for (i = 0; i <= depth; i++, path++) {
731 		brelse(path->p_bh);
732 		path->p_bh = NULL;
733 	}
734 }
735 
736 /*
737  * ext4_ext_binsearch_idx:
738  * binary search for the closest index of the given block
739  * the header must be checked before calling this
740  */
741 static void
742 ext4_ext_binsearch_idx(struct inode *inode,
743 			struct ext4_ext_path *path, ext4_lblk_t block)
744 {
745 	struct ext4_extent_header *eh = path->p_hdr;
746 	struct ext4_extent_idx *r, *l, *m;
747 
748 
749 	ext_debug(inode, "binsearch for %u(idx):  ", block);
750 
751 	l = EXT_FIRST_INDEX(eh) + 1;
752 	r = EXT_LAST_INDEX(eh);
753 	while (l <= r) {
754 		m = l + (r - l) / 2;
755 		ext_debug(inode, "%p(%u):%p(%u):%p(%u) ", l,
756 			  le32_to_cpu(l->ei_block), m, le32_to_cpu(m->ei_block),
757 			  r, le32_to_cpu(r->ei_block));
758 
759 		if (block < le32_to_cpu(m->ei_block))
760 			r = m - 1;
761 		else
762 			l = m + 1;
763 	}
764 
765 	path->p_idx = l - 1;
766 	ext_debug(inode, "  -> %u->%lld ", le32_to_cpu(path->p_idx->ei_block),
767 		  ext4_idx_pblock(path->p_idx));
768 
769 #ifdef CHECK_BINSEARCH
770 	{
771 		struct ext4_extent_idx *chix, *ix;
772 		int k;
773 
774 		chix = ix = EXT_FIRST_INDEX(eh);
775 		for (k = 0; k < le16_to_cpu(eh->eh_entries); k++, ix++) {
776 			if (k != 0 && le32_to_cpu(ix->ei_block) <=
777 			    le32_to_cpu(ix[-1].ei_block)) {
778 				printk(KERN_DEBUG "k=%d, ix=0x%p, "
779 				       "first=0x%p\n", k,
780 				       ix, EXT_FIRST_INDEX(eh));
781 				printk(KERN_DEBUG "%u <= %u\n",
782 				       le32_to_cpu(ix->ei_block),
783 				       le32_to_cpu(ix[-1].ei_block));
784 			}
785 			BUG_ON(k && le32_to_cpu(ix->ei_block)
786 					   <= le32_to_cpu(ix[-1].ei_block));
787 			if (block < le32_to_cpu(ix->ei_block))
788 				break;
789 			chix = ix;
790 		}
791 		BUG_ON(chix != path->p_idx);
792 	}
793 #endif
794 
795 }
796 
797 /*
798  * ext4_ext_binsearch:
799  * binary search for closest extent of the given block
800  * the header must be checked before calling this
801  */
802 static void
803 ext4_ext_binsearch(struct inode *inode,
804 		struct ext4_ext_path *path, ext4_lblk_t block)
805 {
806 	struct ext4_extent_header *eh = path->p_hdr;
807 	struct ext4_extent *r, *l, *m;
808 
809 	if (eh->eh_entries == 0) {
810 		/*
811 		 * this leaf is empty:
812 		 * we get such a leaf in split/add case
813 		 */
814 		return;
815 	}
816 
817 	ext_debug(inode, "binsearch for %u:  ", block);
818 
819 	l = EXT_FIRST_EXTENT(eh) + 1;
820 	r = EXT_LAST_EXTENT(eh);
821 
822 	while (l <= r) {
823 		m = l + (r - l) / 2;
824 		ext_debug(inode, "%p(%u):%p(%u):%p(%u) ", l,
825 			  le32_to_cpu(l->ee_block), m, le32_to_cpu(m->ee_block),
826 			  r, le32_to_cpu(r->ee_block));
827 
828 		if (block < le32_to_cpu(m->ee_block))
829 			r = m - 1;
830 		else
831 			l = m + 1;
832 	}
833 
834 	path->p_ext = l - 1;
835 	ext_debug(inode, "  -> %d:%llu:[%d]%d ",
836 			le32_to_cpu(path->p_ext->ee_block),
837 			ext4_ext_pblock(path->p_ext),
838 			ext4_ext_is_unwritten(path->p_ext),
839 			ext4_ext_get_actual_len(path->p_ext));
840 
841 #ifdef CHECK_BINSEARCH
842 	{
843 		struct ext4_extent *chex, *ex;
844 		int k;
845 
846 		chex = ex = EXT_FIRST_EXTENT(eh);
847 		for (k = 0; k < le16_to_cpu(eh->eh_entries); k++, ex++) {
848 			BUG_ON(k && le32_to_cpu(ex->ee_block)
849 					  <= le32_to_cpu(ex[-1].ee_block));
850 			if (block < le32_to_cpu(ex->ee_block))
851 				break;
852 			chex = ex;
853 		}
854 		BUG_ON(chex != path->p_ext);
855 	}
856 #endif
857 
858 }
859 
860 void ext4_ext_tree_init(handle_t *handle, struct inode *inode)
861 {
862 	struct ext4_extent_header *eh;
863 
864 	eh = ext_inode_hdr(inode);
865 	eh->eh_depth = 0;
866 	eh->eh_entries = 0;
867 	eh->eh_magic = EXT4_EXT_MAGIC;
868 	eh->eh_max = cpu_to_le16(ext4_ext_space_root(inode, 0));
869 	eh->eh_generation = 0;
870 	ext4_mark_inode_dirty(handle, inode);
871 }
872 
873 struct ext4_ext_path *
874 ext4_find_extent(struct inode *inode, ext4_lblk_t block,
875 		 struct ext4_ext_path **orig_path, int flags)
876 {
877 	struct ext4_extent_header *eh;
878 	struct buffer_head *bh;
879 	struct ext4_ext_path *path = orig_path ? *orig_path : NULL;
880 	short int depth, i, ppos = 0;
881 	int ret;
882 	gfp_t gfp_flags = GFP_NOFS;
883 
884 	if (flags & EXT4_EX_NOFAIL)
885 		gfp_flags |= __GFP_NOFAIL;
886 
887 	eh = ext_inode_hdr(inode);
888 	depth = ext_depth(inode);
889 	if (depth < 0 || depth > EXT4_MAX_EXTENT_DEPTH) {
890 		EXT4_ERROR_INODE(inode, "inode has invalid extent depth: %d",
891 				 depth);
892 		ret = -EFSCORRUPTED;
893 		goto err;
894 	}
895 
896 	if (path) {
897 		ext4_ext_drop_refs(path);
898 		if (depth > path[0].p_maxdepth) {
899 			kfree(path);
900 			*orig_path = path = NULL;
901 		}
902 	}
903 	if (!path) {
904 		/* account possible depth increase */
905 		path = kcalloc(depth + 2, sizeof(struct ext4_ext_path),
906 				gfp_flags);
907 		if (unlikely(!path))
908 			return ERR_PTR(-ENOMEM);
909 		path[0].p_maxdepth = depth + 1;
910 	}
911 	path[0].p_hdr = eh;
912 	path[0].p_bh = NULL;
913 
914 	i = depth;
915 	if (!(flags & EXT4_EX_NOCACHE) && depth == 0)
916 		ext4_cache_extents(inode, eh);
917 	/* walk through the tree */
918 	while (i) {
919 		ext_debug(inode, "depth %d: num %d, max %d\n",
920 			  ppos, le16_to_cpu(eh->eh_entries), le16_to_cpu(eh->eh_max));
921 
922 		ext4_ext_binsearch_idx(inode, path + ppos, block);
923 		path[ppos].p_block = ext4_idx_pblock(path[ppos].p_idx);
924 		path[ppos].p_depth = i;
925 		path[ppos].p_ext = NULL;
926 
927 		bh = read_extent_tree_block(inode, path[ppos].p_idx, --i, flags);
928 		if (IS_ERR(bh)) {
929 			ret = PTR_ERR(bh);
930 			goto err;
931 		}
932 
933 		eh = ext_block_hdr(bh);
934 		ppos++;
935 		path[ppos].p_bh = bh;
936 		path[ppos].p_hdr = eh;
937 	}
938 
939 	path[ppos].p_depth = i;
940 	path[ppos].p_ext = NULL;
941 	path[ppos].p_idx = NULL;
942 
943 	/* find extent */
944 	ext4_ext_binsearch(inode, path + ppos, block);
945 	/* if not an empty leaf */
946 	if (path[ppos].p_ext)
947 		path[ppos].p_block = ext4_ext_pblock(path[ppos].p_ext);
948 
949 	ext4_ext_show_path(inode, path);
950 
951 	return path;
952 
953 err:
954 	ext4_ext_drop_refs(path);
955 	kfree(path);
956 	if (orig_path)
957 		*orig_path = NULL;
958 	return ERR_PTR(ret);
959 }
960 
961 /*
962  * ext4_ext_insert_index:
963  * insert new index [@logical;@ptr] into the block at @curp;
964  * check where to insert: before @curp or after @curp
965  */
966 static int ext4_ext_insert_index(handle_t *handle, struct inode *inode,
967 				 struct ext4_ext_path *curp,
968 				 int logical, ext4_fsblk_t ptr)
969 {
970 	struct ext4_extent_idx *ix;
971 	int len, err;
972 
973 	err = ext4_ext_get_access(handle, inode, curp);
974 	if (err)
975 		return err;
976 
977 	if (unlikely(logical == le32_to_cpu(curp->p_idx->ei_block))) {
978 		EXT4_ERROR_INODE(inode,
979 				 "logical %d == ei_block %d!",
980 				 logical, le32_to_cpu(curp->p_idx->ei_block));
981 		return -EFSCORRUPTED;
982 	}
983 
984 	if (unlikely(le16_to_cpu(curp->p_hdr->eh_entries)
985 			     >= le16_to_cpu(curp->p_hdr->eh_max))) {
986 		EXT4_ERROR_INODE(inode,
987 				 "eh_entries %d >= eh_max %d!",
988 				 le16_to_cpu(curp->p_hdr->eh_entries),
989 				 le16_to_cpu(curp->p_hdr->eh_max));
990 		return -EFSCORRUPTED;
991 	}
992 
993 	if (logical > le32_to_cpu(curp->p_idx->ei_block)) {
994 		/* insert after */
995 		ext_debug(inode, "insert new index %d after: %llu\n",
996 			  logical, ptr);
997 		ix = curp->p_idx + 1;
998 	} else {
999 		/* insert before */
1000 		ext_debug(inode, "insert new index %d before: %llu\n",
1001 			  logical, ptr);
1002 		ix = curp->p_idx;
1003 	}
1004 
1005 	len = EXT_LAST_INDEX(curp->p_hdr) - ix + 1;
1006 	BUG_ON(len < 0);
1007 	if (len > 0) {
1008 		ext_debug(inode, "insert new index %d: "
1009 				"move %d indices from 0x%p to 0x%p\n",
1010 				logical, len, ix, ix + 1);
1011 		memmove(ix + 1, ix, len * sizeof(struct ext4_extent_idx));
1012 	}
1013 
1014 	if (unlikely(ix > EXT_MAX_INDEX(curp->p_hdr))) {
1015 		EXT4_ERROR_INODE(inode, "ix > EXT_MAX_INDEX!");
1016 		return -EFSCORRUPTED;
1017 	}
1018 
1019 	ix->ei_block = cpu_to_le32(logical);
1020 	ext4_idx_store_pblock(ix, ptr);
1021 	le16_add_cpu(&curp->p_hdr->eh_entries, 1);
1022 
1023 	if (unlikely(ix > EXT_LAST_INDEX(curp->p_hdr))) {
1024 		EXT4_ERROR_INODE(inode, "ix > EXT_LAST_INDEX!");
1025 		return -EFSCORRUPTED;
1026 	}
1027 
1028 	err = ext4_ext_dirty(handle, inode, curp);
1029 	ext4_std_error(inode->i_sb, err);
1030 
1031 	return err;
1032 }
1033 
1034 /*
1035  * ext4_ext_split:
1036  * inserts new subtree into the path, using free index entry
1037  * at depth @at:
1038  * - allocates all needed blocks (new leaf and all intermediate index blocks)
1039  * - makes decision where to split
1040  * - moves remaining extents and index entries (right to the split point)
1041  *   into the newly allocated blocks
1042  * - initializes subtree
1043  */
1044 static int ext4_ext_split(handle_t *handle, struct inode *inode,
1045 			  unsigned int flags,
1046 			  struct ext4_ext_path *path,
1047 			  struct ext4_extent *newext, int at)
1048 {
1049 	struct buffer_head *bh = NULL;
1050 	int depth = ext_depth(inode);
1051 	struct ext4_extent_header *neh;
1052 	struct ext4_extent_idx *fidx;
1053 	int i = at, k, m, a;
1054 	ext4_fsblk_t newblock, oldblock;
1055 	__le32 border;
1056 	ext4_fsblk_t *ablocks = NULL; /* array of allocated blocks */
1057 	gfp_t gfp_flags = GFP_NOFS;
1058 	int err = 0;
1059 	size_t ext_size = 0;
1060 
1061 	if (flags & EXT4_EX_NOFAIL)
1062 		gfp_flags |= __GFP_NOFAIL;
1063 
1064 	/* make decision: where to split? */
1065 	/* FIXME: now decision is simplest: at current extent */
1066 
1067 	/* if current leaf will be split, then we should use
1068 	 * border from split point */
1069 	if (unlikely(path[depth].p_ext > EXT_MAX_EXTENT(path[depth].p_hdr))) {
1070 		EXT4_ERROR_INODE(inode, "p_ext > EXT_MAX_EXTENT!");
1071 		return -EFSCORRUPTED;
1072 	}
1073 	if (path[depth].p_ext != EXT_MAX_EXTENT(path[depth].p_hdr)) {
1074 		border = path[depth].p_ext[1].ee_block;
1075 		ext_debug(inode, "leaf will be split."
1076 				" next leaf starts at %d\n",
1077 				  le32_to_cpu(border));
1078 	} else {
1079 		border = newext->ee_block;
1080 		ext_debug(inode, "leaf will be added."
1081 				" next leaf starts at %d\n",
1082 				le32_to_cpu(border));
1083 	}
1084 
1085 	/*
1086 	 * If error occurs, then we break processing
1087 	 * and mark filesystem read-only. index won't
1088 	 * be inserted and tree will be in consistent
1089 	 * state. Next mount will repair buffers too.
1090 	 */
1091 
1092 	/*
1093 	 * Get array to track all allocated blocks.
1094 	 * We need this to handle errors and free blocks
1095 	 * upon them.
1096 	 */
1097 	ablocks = kcalloc(depth, sizeof(ext4_fsblk_t), gfp_flags);
1098 	if (!ablocks)
1099 		return -ENOMEM;
1100 
1101 	/* allocate all needed blocks */
1102 	ext_debug(inode, "allocate %d blocks for indexes/leaf\n", depth - at);
1103 	for (a = 0; a < depth - at; a++) {
1104 		newblock = ext4_ext_new_meta_block(handle, inode, path,
1105 						   newext, &err, flags);
1106 		if (newblock == 0)
1107 			goto cleanup;
1108 		ablocks[a] = newblock;
1109 	}
1110 
1111 	/* initialize new leaf */
1112 	newblock = ablocks[--a];
1113 	if (unlikely(newblock == 0)) {
1114 		EXT4_ERROR_INODE(inode, "newblock == 0!");
1115 		err = -EFSCORRUPTED;
1116 		goto cleanup;
1117 	}
1118 	bh = sb_getblk_gfp(inode->i_sb, newblock, __GFP_MOVABLE | GFP_NOFS);
1119 	if (unlikely(!bh)) {
1120 		err = -ENOMEM;
1121 		goto cleanup;
1122 	}
1123 	lock_buffer(bh);
1124 
1125 	err = ext4_journal_get_create_access(handle, inode->i_sb, bh,
1126 					     EXT4_JTR_NONE);
1127 	if (err)
1128 		goto cleanup;
1129 
1130 	neh = ext_block_hdr(bh);
1131 	neh->eh_entries = 0;
1132 	neh->eh_max = cpu_to_le16(ext4_ext_space_block(inode, 0));
1133 	neh->eh_magic = EXT4_EXT_MAGIC;
1134 	neh->eh_depth = 0;
1135 	neh->eh_generation = 0;
1136 
1137 	/* move remainder of path[depth] to the new leaf */
1138 	if (unlikely(path[depth].p_hdr->eh_entries !=
1139 		     path[depth].p_hdr->eh_max)) {
1140 		EXT4_ERROR_INODE(inode, "eh_entries %d != eh_max %d!",
1141 				 path[depth].p_hdr->eh_entries,
1142 				 path[depth].p_hdr->eh_max);
1143 		err = -EFSCORRUPTED;
1144 		goto cleanup;
1145 	}
1146 	/* start copy from next extent */
1147 	m = EXT_MAX_EXTENT(path[depth].p_hdr) - path[depth].p_ext++;
1148 	ext4_ext_show_move(inode, path, newblock, depth);
1149 	if (m) {
1150 		struct ext4_extent *ex;
1151 		ex = EXT_FIRST_EXTENT(neh);
1152 		memmove(ex, path[depth].p_ext, sizeof(struct ext4_extent) * m);
1153 		le16_add_cpu(&neh->eh_entries, m);
1154 	}
1155 
1156 	/* zero out unused area in the extent block */
1157 	ext_size = sizeof(struct ext4_extent_header) +
1158 		sizeof(struct ext4_extent) * le16_to_cpu(neh->eh_entries);
1159 	memset(bh->b_data + ext_size, 0, inode->i_sb->s_blocksize - ext_size);
1160 	ext4_extent_block_csum_set(inode, neh);
1161 	set_buffer_uptodate(bh);
1162 	unlock_buffer(bh);
1163 
1164 	err = ext4_handle_dirty_metadata(handle, inode, bh);
1165 	if (err)
1166 		goto cleanup;
1167 	brelse(bh);
1168 	bh = NULL;
1169 
1170 	/* correct old leaf */
1171 	if (m) {
1172 		err = ext4_ext_get_access(handle, inode, path + depth);
1173 		if (err)
1174 			goto cleanup;
1175 		le16_add_cpu(&path[depth].p_hdr->eh_entries, -m);
1176 		err = ext4_ext_dirty(handle, inode, path + depth);
1177 		if (err)
1178 			goto cleanup;
1179 
1180 	}
1181 
1182 	/* create intermediate indexes */
1183 	k = depth - at - 1;
1184 	if (unlikely(k < 0)) {
1185 		EXT4_ERROR_INODE(inode, "k %d < 0!", k);
1186 		err = -EFSCORRUPTED;
1187 		goto cleanup;
1188 	}
1189 	if (k)
1190 		ext_debug(inode, "create %d intermediate indices\n", k);
1191 	/* insert new index into current index block */
1192 	/* current depth stored in i var */
1193 	i = depth - 1;
1194 	while (k--) {
1195 		oldblock = newblock;
1196 		newblock = ablocks[--a];
1197 		bh = sb_getblk(inode->i_sb, newblock);
1198 		if (unlikely(!bh)) {
1199 			err = -ENOMEM;
1200 			goto cleanup;
1201 		}
1202 		lock_buffer(bh);
1203 
1204 		err = ext4_journal_get_create_access(handle, inode->i_sb, bh,
1205 						     EXT4_JTR_NONE);
1206 		if (err)
1207 			goto cleanup;
1208 
1209 		neh = ext_block_hdr(bh);
1210 		neh->eh_entries = cpu_to_le16(1);
1211 		neh->eh_magic = EXT4_EXT_MAGIC;
1212 		neh->eh_max = cpu_to_le16(ext4_ext_space_block_idx(inode, 0));
1213 		neh->eh_depth = cpu_to_le16(depth - i);
1214 		neh->eh_generation = 0;
1215 		fidx = EXT_FIRST_INDEX(neh);
1216 		fidx->ei_block = border;
1217 		ext4_idx_store_pblock(fidx, oldblock);
1218 
1219 		ext_debug(inode, "int.index at %d (block %llu): %u -> %llu\n",
1220 				i, newblock, le32_to_cpu(border), oldblock);
1221 
1222 		/* move remainder of path[i] to the new index block */
1223 		if (unlikely(EXT_MAX_INDEX(path[i].p_hdr) !=
1224 					EXT_LAST_INDEX(path[i].p_hdr))) {
1225 			EXT4_ERROR_INODE(inode,
1226 					 "EXT_MAX_INDEX != EXT_LAST_INDEX ee_block %d!",
1227 					 le32_to_cpu(path[i].p_ext->ee_block));
1228 			err = -EFSCORRUPTED;
1229 			goto cleanup;
1230 		}
1231 		/* start copy indexes */
1232 		m = EXT_MAX_INDEX(path[i].p_hdr) - path[i].p_idx++;
1233 		ext_debug(inode, "cur 0x%p, last 0x%p\n", path[i].p_idx,
1234 				EXT_MAX_INDEX(path[i].p_hdr));
1235 		ext4_ext_show_move(inode, path, newblock, i);
1236 		if (m) {
1237 			memmove(++fidx, path[i].p_idx,
1238 				sizeof(struct ext4_extent_idx) * m);
1239 			le16_add_cpu(&neh->eh_entries, m);
1240 		}
1241 		/* zero out unused area in the extent block */
1242 		ext_size = sizeof(struct ext4_extent_header) +
1243 		   (sizeof(struct ext4_extent) * le16_to_cpu(neh->eh_entries));
1244 		memset(bh->b_data + ext_size, 0,
1245 			inode->i_sb->s_blocksize - ext_size);
1246 		ext4_extent_block_csum_set(inode, neh);
1247 		set_buffer_uptodate(bh);
1248 		unlock_buffer(bh);
1249 
1250 		err = ext4_handle_dirty_metadata(handle, inode, bh);
1251 		if (err)
1252 			goto cleanup;
1253 		brelse(bh);
1254 		bh = NULL;
1255 
1256 		/* correct old index */
1257 		if (m) {
1258 			err = ext4_ext_get_access(handle, inode, path + i);
1259 			if (err)
1260 				goto cleanup;
1261 			le16_add_cpu(&path[i].p_hdr->eh_entries, -m);
1262 			err = ext4_ext_dirty(handle, inode, path + i);
1263 			if (err)
1264 				goto cleanup;
1265 		}
1266 
1267 		i--;
1268 	}
1269 
1270 	/* insert new index */
1271 	err = ext4_ext_insert_index(handle, inode, path + at,
1272 				    le32_to_cpu(border), newblock);
1273 
1274 cleanup:
1275 	if (bh) {
1276 		if (buffer_locked(bh))
1277 			unlock_buffer(bh);
1278 		brelse(bh);
1279 	}
1280 
1281 	if (err) {
1282 		/* free all allocated blocks in error case */
1283 		for (i = 0; i < depth; i++) {
1284 			if (!ablocks[i])
1285 				continue;
1286 			ext4_free_blocks(handle, inode, NULL, ablocks[i], 1,
1287 					 EXT4_FREE_BLOCKS_METADATA);
1288 		}
1289 	}
1290 	kfree(ablocks);
1291 
1292 	return err;
1293 }
1294 
1295 /*
1296  * ext4_ext_grow_indepth:
1297  * implements tree growing procedure:
1298  * - allocates new block
1299  * - moves top-level data (index block or leaf) into the new block
1300  * - initializes new top-level, creating index that points to the
1301  *   just created block
1302  */
1303 static int ext4_ext_grow_indepth(handle_t *handle, struct inode *inode,
1304 				 unsigned int flags)
1305 {
1306 	struct ext4_extent_header *neh;
1307 	struct buffer_head *bh;
1308 	ext4_fsblk_t newblock, goal = 0;
1309 	struct ext4_super_block *es = EXT4_SB(inode->i_sb)->s_es;
1310 	int err = 0;
1311 	size_t ext_size = 0;
1312 
1313 	/* Try to prepend new index to old one */
1314 	if (ext_depth(inode))
1315 		goal = ext4_idx_pblock(EXT_FIRST_INDEX(ext_inode_hdr(inode)));
1316 	if (goal > le32_to_cpu(es->s_first_data_block)) {
1317 		flags |= EXT4_MB_HINT_TRY_GOAL;
1318 		goal--;
1319 	} else
1320 		goal = ext4_inode_to_goal_block(inode);
1321 	newblock = ext4_new_meta_blocks(handle, inode, goal, flags,
1322 					NULL, &err);
1323 	if (newblock == 0)
1324 		return err;
1325 
1326 	bh = sb_getblk_gfp(inode->i_sb, newblock, __GFP_MOVABLE | GFP_NOFS);
1327 	if (unlikely(!bh))
1328 		return -ENOMEM;
1329 	lock_buffer(bh);
1330 
1331 	err = ext4_journal_get_create_access(handle, inode->i_sb, bh,
1332 					     EXT4_JTR_NONE);
1333 	if (err) {
1334 		unlock_buffer(bh);
1335 		goto out;
1336 	}
1337 
1338 	ext_size = sizeof(EXT4_I(inode)->i_data);
1339 	/* move top-level index/leaf into new block */
1340 	memmove(bh->b_data, EXT4_I(inode)->i_data, ext_size);
1341 	/* zero out unused area in the extent block */
1342 	memset(bh->b_data + ext_size, 0, inode->i_sb->s_blocksize - ext_size);
1343 
1344 	/* set size of new block */
1345 	neh = ext_block_hdr(bh);
1346 	/* old root could have indexes or leaves
1347 	 * so calculate e_max right way */
1348 	if (ext_depth(inode))
1349 		neh->eh_max = cpu_to_le16(ext4_ext_space_block_idx(inode, 0));
1350 	else
1351 		neh->eh_max = cpu_to_le16(ext4_ext_space_block(inode, 0));
1352 	neh->eh_magic = EXT4_EXT_MAGIC;
1353 	ext4_extent_block_csum_set(inode, neh);
1354 	set_buffer_uptodate(bh);
1355 	set_buffer_verified(bh);
1356 	unlock_buffer(bh);
1357 
1358 	err = ext4_handle_dirty_metadata(handle, inode, bh);
1359 	if (err)
1360 		goto out;
1361 
1362 	/* Update top-level index: num,max,pointer */
1363 	neh = ext_inode_hdr(inode);
1364 	neh->eh_entries = cpu_to_le16(1);
1365 	ext4_idx_store_pblock(EXT_FIRST_INDEX(neh), newblock);
1366 	if (neh->eh_depth == 0) {
1367 		/* Root extent block becomes index block */
1368 		neh->eh_max = cpu_to_le16(ext4_ext_space_root_idx(inode, 0));
1369 		EXT_FIRST_INDEX(neh)->ei_block =
1370 			EXT_FIRST_EXTENT(neh)->ee_block;
1371 	}
1372 	ext_debug(inode, "new root: num %d(%d), lblock %d, ptr %llu\n",
1373 		  le16_to_cpu(neh->eh_entries), le16_to_cpu(neh->eh_max),
1374 		  le32_to_cpu(EXT_FIRST_INDEX(neh)->ei_block),
1375 		  ext4_idx_pblock(EXT_FIRST_INDEX(neh)));
1376 
1377 	le16_add_cpu(&neh->eh_depth, 1);
1378 	err = ext4_mark_inode_dirty(handle, inode);
1379 out:
1380 	brelse(bh);
1381 
1382 	return err;
1383 }
1384 
1385 /*
1386  * ext4_ext_create_new_leaf:
1387  * finds empty index and adds new leaf.
1388  * if no free index is found, then it requests in-depth growing.
1389  */
1390 static int ext4_ext_create_new_leaf(handle_t *handle, struct inode *inode,
1391 				    unsigned int mb_flags,
1392 				    unsigned int gb_flags,
1393 				    struct ext4_ext_path **ppath,
1394 				    struct ext4_extent *newext)
1395 {
1396 	struct ext4_ext_path *path = *ppath;
1397 	struct ext4_ext_path *curp;
1398 	int depth, i, err = 0;
1399 
1400 repeat:
1401 	i = depth = ext_depth(inode);
1402 
1403 	/* walk up to the tree and look for free index entry */
1404 	curp = path + depth;
1405 	while (i > 0 && !EXT_HAS_FREE_INDEX(curp)) {
1406 		i--;
1407 		curp--;
1408 	}
1409 
1410 	/* we use already allocated block for index block,
1411 	 * so subsequent data blocks should be contiguous */
1412 	if (EXT_HAS_FREE_INDEX(curp)) {
1413 		/* if we found index with free entry, then use that
1414 		 * entry: create all needed subtree and add new leaf */
1415 		err = ext4_ext_split(handle, inode, mb_flags, path, newext, i);
1416 		if (err)
1417 			goto out;
1418 
1419 		/* refill path */
1420 		path = ext4_find_extent(inode,
1421 				    (ext4_lblk_t)le32_to_cpu(newext->ee_block),
1422 				    ppath, gb_flags);
1423 		if (IS_ERR(path))
1424 			err = PTR_ERR(path);
1425 	} else {
1426 		/* tree is full, time to grow in depth */
1427 		err = ext4_ext_grow_indepth(handle, inode, mb_flags);
1428 		if (err)
1429 			goto out;
1430 
1431 		/* refill path */
1432 		path = ext4_find_extent(inode,
1433 				   (ext4_lblk_t)le32_to_cpu(newext->ee_block),
1434 				    ppath, gb_flags);
1435 		if (IS_ERR(path)) {
1436 			err = PTR_ERR(path);
1437 			goto out;
1438 		}
1439 
1440 		/*
1441 		 * only first (depth 0 -> 1) produces free space;
1442 		 * in all other cases we have to split the grown tree
1443 		 */
1444 		depth = ext_depth(inode);
1445 		if (path[depth].p_hdr->eh_entries == path[depth].p_hdr->eh_max) {
1446 			/* now we need to split */
1447 			goto repeat;
1448 		}
1449 	}
1450 
1451 out:
1452 	return err;
1453 }
1454 
1455 /*
1456  * search the closest allocated block to the left for *logical
1457  * and returns it at @logical + it's physical address at @phys
1458  * if *logical is the smallest allocated block, the function
1459  * returns 0 at @phys
1460  * return value contains 0 (success) or error code
1461  */
1462 static int ext4_ext_search_left(struct inode *inode,
1463 				struct ext4_ext_path *path,
1464 				ext4_lblk_t *logical, ext4_fsblk_t *phys)
1465 {
1466 	struct ext4_extent_idx *ix;
1467 	struct ext4_extent *ex;
1468 	int depth, ee_len;
1469 
1470 	if (unlikely(path == NULL)) {
1471 		EXT4_ERROR_INODE(inode, "path == NULL *logical %d!", *logical);
1472 		return -EFSCORRUPTED;
1473 	}
1474 	depth = path->p_depth;
1475 	*phys = 0;
1476 
1477 	if (depth == 0 && path->p_ext == NULL)
1478 		return 0;
1479 
1480 	/* usually extent in the path covers blocks smaller
1481 	 * then *logical, but it can be that extent is the
1482 	 * first one in the file */
1483 
1484 	ex = path[depth].p_ext;
1485 	ee_len = ext4_ext_get_actual_len(ex);
1486 	if (*logical < le32_to_cpu(ex->ee_block)) {
1487 		if (unlikely(EXT_FIRST_EXTENT(path[depth].p_hdr) != ex)) {
1488 			EXT4_ERROR_INODE(inode,
1489 					 "EXT_FIRST_EXTENT != ex *logical %d ee_block %d!",
1490 					 *logical, le32_to_cpu(ex->ee_block));
1491 			return -EFSCORRUPTED;
1492 		}
1493 		while (--depth >= 0) {
1494 			ix = path[depth].p_idx;
1495 			if (unlikely(ix != EXT_FIRST_INDEX(path[depth].p_hdr))) {
1496 				EXT4_ERROR_INODE(inode,
1497 				  "ix (%d) != EXT_FIRST_INDEX (%d) (depth %d)!",
1498 				  ix != NULL ? le32_to_cpu(ix->ei_block) : 0,
1499 				  le32_to_cpu(EXT_FIRST_INDEX(path[depth].p_hdr)->ei_block),
1500 				  depth);
1501 				return -EFSCORRUPTED;
1502 			}
1503 		}
1504 		return 0;
1505 	}
1506 
1507 	if (unlikely(*logical < (le32_to_cpu(ex->ee_block) + ee_len))) {
1508 		EXT4_ERROR_INODE(inode,
1509 				 "logical %d < ee_block %d + ee_len %d!",
1510 				 *logical, le32_to_cpu(ex->ee_block), ee_len);
1511 		return -EFSCORRUPTED;
1512 	}
1513 
1514 	*logical = le32_to_cpu(ex->ee_block) + ee_len - 1;
1515 	*phys = ext4_ext_pblock(ex) + ee_len - 1;
1516 	return 0;
1517 }
1518 
1519 /*
1520  * Search the closest allocated block to the right for *logical
1521  * and returns it at @logical + it's physical address at @phys.
1522  * If not exists, return 0 and @phys is set to 0. We will return
1523  * 1 which means we found an allocated block and ret_ex is valid.
1524  * Or return a (< 0) error code.
1525  */
1526 static int ext4_ext_search_right(struct inode *inode,
1527 				 struct ext4_ext_path *path,
1528 				 ext4_lblk_t *logical, ext4_fsblk_t *phys,
1529 				 struct ext4_extent *ret_ex)
1530 {
1531 	struct buffer_head *bh = NULL;
1532 	struct ext4_extent_header *eh;
1533 	struct ext4_extent_idx *ix;
1534 	struct ext4_extent *ex;
1535 	int depth;	/* Note, NOT eh_depth; depth from top of tree */
1536 	int ee_len;
1537 
1538 	if (unlikely(path == NULL)) {
1539 		EXT4_ERROR_INODE(inode, "path == NULL *logical %d!", *logical);
1540 		return -EFSCORRUPTED;
1541 	}
1542 	depth = path->p_depth;
1543 	*phys = 0;
1544 
1545 	if (depth == 0 && path->p_ext == NULL)
1546 		return 0;
1547 
1548 	/* usually extent in the path covers blocks smaller
1549 	 * then *logical, but it can be that extent is the
1550 	 * first one in the file */
1551 
1552 	ex = path[depth].p_ext;
1553 	ee_len = ext4_ext_get_actual_len(ex);
1554 	if (*logical < le32_to_cpu(ex->ee_block)) {
1555 		if (unlikely(EXT_FIRST_EXTENT(path[depth].p_hdr) != ex)) {
1556 			EXT4_ERROR_INODE(inode,
1557 					 "first_extent(path[%d].p_hdr) != ex",
1558 					 depth);
1559 			return -EFSCORRUPTED;
1560 		}
1561 		while (--depth >= 0) {
1562 			ix = path[depth].p_idx;
1563 			if (unlikely(ix != EXT_FIRST_INDEX(path[depth].p_hdr))) {
1564 				EXT4_ERROR_INODE(inode,
1565 						 "ix != EXT_FIRST_INDEX *logical %d!",
1566 						 *logical);
1567 				return -EFSCORRUPTED;
1568 			}
1569 		}
1570 		goto found_extent;
1571 	}
1572 
1573 	if (unlikely(*logical < (le32_to_cpu(ex->ee_block) + ee_len))) {
1574 		EXT4_ERROR_INODE(inode,
1575 				 "logical %d < ee_block %d + ee_len %d!",
1576 				 *logical, le32_to_cpu(ex->ee_block), ee_len);
1577 		return -EFSCORRUPTED;
1578 	}
1579 
1580 	if (ex != EXT_LAST_EXTENT(path[depth].p_hdr)) {
1581 		/* next allocated block in this leaf */
1582 		ex++;
1583 		goto found_extent;
1584 	}
1585 
1586 	/* go up and search for index to the right */
1587 	while (--depth >= 0) {
1588 		ix = path[depth].p_idx;
1589 		if (ix != EXT_LAST_INDEX(path[depth].p_hdr))
1590 			goto got_index;
1591 	}
1592 
1593 	/* we've gone up to the root and found no index to the right */
1594 	return 0;
1595 
1596 got_index:
1597 	/* we've found index to the right, let's
1598 	 * follow it and find the closest allocated
1599 	 * block to the right */
1600 	ix++;
1601 	while (++depth < path->p_depth) {
1602 		/* subtract from p_depth to get proper eh_depth */
1603 		bh = read_extent_tree_block(inode, ix, path->p_depth - depth, 0);
1604 		if (IS_ERR(bh))
1605 			return PTR_ERR(bh);
1606 		eh = ext_block_hdr(bh);
1607 		ix = EXT_FIRST_INDEX(eh);
1608 		put_bh(bh);
1609 	}
1610 
1611 	bh = read_extent_tree_block(inode, ix, path->p_depth - depth, 0);
1612 	if (IS_ERR(bh))
1613 		return PTR_ERR(bh);
1614 	eh = ext_block_hdr(bh);
1615 	ex = EXT_FIRST_EXTENT(eh);
1616 found_extent:
1617 	*logical = le32_to_cpu(ex->ee_block);
1618 	*phys = ext4_ext_pblock(ex);
1619 	if (ret_ex)
1620 		*ret_ex = *ex;
1621 	if (bh)
1622 		put_bh(bh);
1623 	return 1;
1624 }
1625 
1626 /*
1627  * ext4_ext_next_allocated_block:
1628  * returns allocated block in subsequent extent or EXT_MAX_BLOCKS.
1629  * NOTE: it considers block number from index entry as
1630  * allocated block. Thus, index entries have to be consistent
1631  * with leaves.
1632  */
1633 ext4_lblk_t
1634 ext4_ext_next_allocated_block(struct ext4_ext_path *path)
1635 {
1636 	int depth;
1637 
1638 	BUG_ON(path == NULL);
1639 	depth = path->p_depth;
1640 
1641 	if (depth == 0 && path->p_ext == NULL)
1642 		return EXT_MAX_BLOCKS;
1643 
1644 	while (depth >= 0) {
1645 		struct ext4_ext_path *p = &path[depth];
1646 
1647 		if (depth == path->p_depth) {
1648 			/* leaf */
1649 			if (p->p_ext && p->p_ext != EXT_LAST_EXTENT(p->p_hdr))
1650 				return le32_to_cpu(p->p_ext[1].ee_block);
1651 		} else {
1652 			/* index */
1653 			if (p->p_idx != EXT_LAST_INDEX(p->p_hdr))
1654 				return le32_to_cpu(p->p_idx[1].ei_block);
1655 		}
1656 		depth--;
1657 	}
1658 
1659 	return EXT_MAX_BLOCKS;
1660 }
1661 
1662 /*
1663  * ext4_ext_next_leaf_block:
1664  * returns first allocated block from next leaf or EXT_MAX_BLOCKS
1665  */
1666 static ext4_lblk_t ext4_ext_next_leaf_block(struct ext4_ext_path *path)
1667 {
1668 	int depth;
1669 
1670 	BUG_ON(path == NULL);
1671 	depth = path->p_depth;
1672 
1673 	/* zero-tree has no leaf blocks at all */
1674 	if (depth == 0)
1675 		return EXT_MAX_BLOCKS;
1676 
1677 	/* go to index block */
1678 	depth--;
1679 
1680 	while (depth >= 0) {
1681 		if (path[depth].p_idx !=
1682 				EXT_LAST_INDEX(path[depth].p_hdr))
1683 			return (ext4_lblk_t)
1684 				le32_to_cpu(path[depth].p_idx[1].ei_block);
1685 		depth--;
1686 	}
1687 
1688 	return EXT_MAX_BLOCKS;
1689 }
1690 
1691 /*
1692  * ext4_ext_correct_indexes:
1693  * if leaf gets modified and modified extent is first in the leaf,
1694  * then we have to correct all indexes above.
1695  * TODO: do we need to correct tree in all cases?
1696  */
1697 static int ext4_ext_correct_indexes(handle_t *handle, struct inode *inode,
1698 				struct ext4_ext_path *path)
1699 {
1700 	struct ext4_extent_header *eh;
1701 	int depth = ext_depth(inode);
1702 	struct ext4_extent *ex;
1703 	__le32 border;
1704 	int k, err = 0;
1705 
1706 	eh = path[depth].p_hdr;
1707 	ex = path[depth].p_ext;
1708 
1709 	if (unlikely(ex == NULL || eh == NULL)) {
1710 		EXT4_ERROR_INODE(inode,
1711 				 "ex %p == NULL or eh %p == NULL", ex, eh);
1712 		return -EFSCORRUPTED;
1713 	}
1714 
1715 	if (depth == 0) {
1716 		/* there is no tree at all */
1717 		return 0;
1718 	}
1719 
1720 	if (ex != EXT_FIRST_EXTENT(eh)) {
1721 		/* we correct tree if first leaf got modified only */
1722 		return 0;
1723 	}
1724 
1725 	/*
1726 	 * TODO: we need correction if border is smaller than current one
1727 	 */
1728 	k = depth - 1;
1729 	border = path[depth].p_ext->ee_block;
1730 	err = ext4_ext_get_access(handle, inode, path + k);
1731 	if (err)
1732 		return err;
1733 	path[k].p_idx->ei_block = border;
1734 	err = ext4_ext_dirty(handle, inode, path + k);
1735 	if (err)
1736 		return err;
1737 
1738 	while (k--) {
1739 		/* change all left-side indexes */
1740 		if (path[k+1].p_idx != EXT_FIRST_INDEX(path[k+1].p_hdr))
1741 			break;
1742 		err = ext4_ext_get_access(handle, inode, path + k);
1743 		if (err)
1744 			break;
1745 		path[k].p_idx->ei_block = border;
1746 		err = ext4_ext_dirty(handle, inode, path + k);
1747 		if (err)
1748 			break;
1749 	}
1750 
1751 	return err;
1752 }
1753 
1754 static int ext4_can_extents_be_merged(struct inode *inode,
1755 				      struct ext4_extent *ex1,
1756 				      struct ext4_extent *ex2)
1757 {
1758 	unsigned short ext1_ee_len, ext2_ee_len;
1759 
1760 	if (ext4_ext_is_unwritten(ex1) != ext4_ext_is_unwritten(ex2))
1761 		return 0;
1762 
1763 	ext1_ee_len = ext4_ext_get_actual_len(ex1);
1764 	ext2_ee_len = ext4_ext_get_actual_len(ex2);
1765 
1766 	if (le32_to_cpu(ex1->ee_block) + ext1_ee_len !=
1767 			le32_to_cpu(ex2->ee_block))
1768 		return 0;
1769 
1770 	if (ext1_ee_len + ext2_ee_len > EXT_INIT_MAX_LEN)
1771 		return 0;
1772 
1773 	if (ext4_ext_is_unwritten(ex1) &&
1774 	    ext1_ee_len + ext2_ee_len > EXT_UNWRITTEN_MAX_LEN)
1775 		return 0;
1776 #ifdef AGGRESSIVE_TEST
1777 	if (ext1_ee_len >= 4)
1778 		return 0;
1779 #endif
1780 
1781 	if (ext4_ext_pblock(ex1) + ext1_ee_len == ext4_ext_pblock(ex2))
1782 		return 1;
1783 	return 0;
1784 }
1785 
1786 /*
1787  * This function tries to merge the "ex" extent to the next extent in the tree.
1788  * It always tries to merge towards right. If you want to merge towards
1789  * left, pass "ex - 1" as argument instead of "ex".
1790  * Returns 0 if the extents (ex and ex+1) were _not_ merged and returns
1791  * 1 if they got merged.
1792  */
1793 static int ext4_ext_try_to_merge_right(struct inode *inode,
1794 				 struct ext4_ext_path *path,
1795 				 struct ext4_extent *ex)
1796 {
1797 	struct ext4_extent_header *eh;
1798 	unsigned int depth, len;
1799 	int merge_done = 0, unwritten;
1800 
1801 	depth = ext_depth(inode);
1802 	BUG_ON(path[depth].p_hdr == NULL);
1803 	eh = path[depth].p_hdr;
1804 
1805 	while (ex < EXT_LAST_EXTENT(eh)) {
1806 		if (!ext4_can_extents_be_merged(inode, ex, ex + 1))
1807 			break;
1808 		/* merge with next extent! */
1809 		unwritten = ext4_ext_is_unwritten(ex);
1810 		ex->ee_len = cpu_to_le16(ext4_ext_get_actual_len(ex)
1811 				+ ext4_ext_get_actual_len(ex + 1));
1812 		if (unwritten)
1813 			ext4_ext_mark_unwritten(ex);
1814 
1815 		if (ex + 1 < EXT_LAST_EXTENT(eh)) {
1816 			len = (EXT_LAST_EXTENT(eh) - ex - 1)
1817 				* sizeof(struct ext4_extent);
1818 			memmove(ex + 1, ex + 2, len);
1819 		}
1820 		le16_add_cpu(&eh->eh_entries, -1);
1821 		merge_done = 1;
1822 		WARN_ON(eh->eh_entries == 0);
1823 		if (!eh->eh_entries)
1824 			EXT4_ERROR_INODE(inode, "eh->eh_entries = 0!");
1825 	}
1826 
1827 	return merge_done;
1828 }
1829 
1830 /*
1831  * This function does a very simple check to see if we can collapse
1832  * an extent tree with a single extent tree leaf block into the inode.
1833  */
1834 static void ext4_ext_try_to_merge_up(handle_t *handle,
1835 				     struct inode *inode,
1836 				     struct ext4_ext_path *path)
1837 {
1838 	size_t s;
1839 	unsigned max_root = ext4_ext_space_root(inode, 0);
1840 	ext4_fsblk_t blk;
1841 
1842 	if ((path[0].p_depth != 1) ||
1843 	    (le16_to_cpu(path[0].p_hdr->eh_entries) != 1) ||
1844 	    (le16_to_cpu(path[1].p_hdr->eh_entries) > max_root))
1845 		return;
1846 
1847 	/*
1848 	 * We need to modify the block allocation bitmap and the block
1849 	 * group descriptor to release the extent tree block.  If we
1850 	 * can't get the journal credits, give up.
1851 	 */
1852 	if (ext4_journal_extend(handle, 2,
1853 			ext4_free_metadata_revoke_credits(inode->i_sb, 1)))
1854 		return;
1855 
1856 	/*
1857 	 * Copy the extent data up to the inode
1858 	 */
1859 	blk = ext4_idx_pblock(path[0].p_idx);
1860 	s = le16_to_cpu(path[1].p_hdr->eh_entries) *
1861 		sizeof(struct ext4_extent_idx);
1862 	s += sizeof(struct ext4_extent_header);
1863 
1864 	path[1].p_maxdepth = path[0].p_maxdepth;
1865 	memcpy(path[0].p_hdr, path[1].p_hdr, s);
1866 	path[0].p_depth = 0;
1867 	path[0].p_ext = EXT_FIRST_EXTENT(path[0].p_hdr) +
1868 		(path[1].p_ext - EXT_FIRST_EXTENT(path[1].p_hdr));
1869 	path[0].p_hdr->eh_max = cpu_to_le16(max_root);
1870 
1871 	brelse(path[1].p_bh);
1872 	ext4_free_blocks(handle, inode, NULL, blk, 1,
1873 			 EXT4_FREE_BLOCKS_METADATA | EXT4_FREE_BLOCKS_FORGET);
1874 }
1875 
1876 /*
1877  * This function tries to merge the @ex extent to neighbours in the tree, then
1878  * tries to collapse the extent tree into the inode.
1879  */
1880 static void ext4_ext_try_to_merge(handle_t *handle,
1881 				  struct inode *inode,
1882 				  struct ext4_ext_path *path,
1883 				  struct ext4_extent *ex)
1884 {
1885 	struct ext4_extent_header *eh;
1886 	unsigned int depth;
1887 	int merge_done = 0;
1888 
1889 	depth = ext_depth(inode);
1890 	BUG_ON(path[depth].p_hdr == NULL);
1891 	eh = path[depth].p_hdr;
1892 
1893 	if (ex > EXT_FIRST_EXTENT(eh))
1894 		merge_done = ext4_ext_try_to_merge_right(inode, path, ex - 1);
1895 
1896 	if (!merge_done)
1897 		(void) ext4_ext_try_to_merge_right(inode, path, ex);
1898 
1899 	ext4_ext_try_to_merge_up(handle, inode, path);
1900 }
1901 
1902 /*
1903  * check if a portion of the "newext" extent overlaps with an
1904  * existing extent.
1905  *
1906  * If there is an overlap discovered, it updates the length of the newext
1907  * such that there will be no overlap, and then returns 1.
1908  * If there is no overlap found, it returns 0.
1909  */
1910 static unsigned int ext4_ext_check_overlap(struct ext4_sb_info *sbi,
1911 					   struct inode *inode,
1912 					   struct ext4_extent *newext,
1913 					   struct ext4_ext_path *path)
1914 {
1915 	ext4_lblk_t b1, b2;
1916 	unsigned int depth, len1;
1917 	unsigned int ret = 0;
1918 
1919 	b1 = le32_to_cpu(newext->ee_block);
1920 	len1 = ext4_ext_get_actual_len(newext);
1921 	depth = ext_depth(inode);
1922 	if (!path[depth].p_ext)
1923 		goto out;
1924 	b2 = EXT4_LBLK_CMASK(sbi, le32_to_cpu(path[depth].p_ext->ee_block));
1925 
1926 	/*
1927 	 * get the next allocated block if the extent in the path
1928 	 * is before the requested block(s)
1929 	 */
1930 	if (b2 < b1) {
1931 		b2 = ext4_ext_next_allocated_block(path);
1932 		if (b2 == EXT_MAX_BLOCKS)
1933 			goto out;
1934 		b2 = EXT4_LBLK_CMASK(sbi, b2);
1935 	}
1936 
1937 	/* check for wrap through zero on extent logical start block*/
1938 	if (b1 + len1 < b1) {
1939 		len1 = EXT_MAX_BLOCKS - b1;
1940 		newext->ee_len = cpu_to_le16(len1);
1941 		ret = 1;
1942 	}
1943 
1944 	/* check for overlap */
1945 	if (b1 + len1 > b2) {
1946 		newext->ee_len = cpu_to_le16(b2 - b1);
1947 		ret = 1;
1948 	}
1949 out:
1950 	return ret;
1951 }
1952 
1953 /*
1954  * ext4_ext_insert_extent:
1955  * tries to merge requested extent into the existing extent or
1956  * inserts requested extent as new one into the tree,
1957  * creating new leaf in the no-space case.
1958  */
1959 int ext4_ext_insert_extent(handle_t *handle, struct inode *inode,
1960 				struct ext4_ext_path **ppath,
1961 				struct ext4_extent *newext, int gb_flags)
1962 {
1963 	struct ext4_ext_path *path = *ppath;
1964 	struct ext4_extent_header *eh;
1965 	struct ext4_extent *ex, *fex;
1966 	struct ext4_extent *nearex; /* nearest extent */
1967 	struct ext4_ext_path *npath = NULL;
1968 	int depth, len, err;
1969 	ext4_lblk_t next;
1970 	int mb_flags = 0, unwritten;
1971 
1972 	if (gb_flags & EXT4_GET_BLOCKS_DELALLOC_RESERVE)
1973 		mb_flags |= EXT4_MB_DELALLOC_RESERVED;
1974 	if (unlikely(ext4_ext_get_actual_len(newext) == 0)) {
1975 		EXT4_ERROR_INODE(inode, "ext4_ext_get_actual_len(newext) == 0");
1976 		return -EFSCORRUPTED;
1977 	}
1978 	depth = ext_depth(inode);
1979 	ex = path[depth].p_ext;
1980 	eh = path[depth].p_hdr;
1981 	if (unlikely(path[depth].p_hdr == NULL)) {
1982 		EXT4_ERROR_INODE(inode, "path[%d].p_hdr == NULL", depth);
1983 		return -EFSCORRUPTED;
1984 	}
1985 
1986 	/* try to insert block into found extent and return */
1987 	if (ex && !(gb_flags & EXT4_GET_BLOCKS_PRE_IO)) {
1988 
1989 		/*
1990 		 * Try to see whether we should rather test the extent on
1991 		 * right from ex, or from the left of ex. This is because
1992 		 * ext4_find_extent() can return either extent on the
1993 		 * left, or on the right from the searched position. This
1994 		 * will make merging more effective.
1995 		 */
1996 		if (ex < EXT_LAST_EXTENT(eh) &&
1997 		    (le32_to_cpu(ex->ee_block) +
1998 		    ext4_ext_get_actual_len(ex) <
1999 		    le32_to_cpu(newext->ee_block))) {
2000 			ex += 1;
2001 			goto prepend;
2002 		} else if ((ex > EXT_FIRST_EXTENT(eh)) &&
2003 			   (le32_to_cpu(newext->ee_block) +
2004 			   ext4_ext_get_actual_len(newext) <
2005 			   le32_to_cpu(ex->ee_block)))
2006 			ex -= 1;
2007 
2008 		/* Try to append newex to the ex */
2009 		if (ext4_can_extents_be_merged(inode, ex, newext)) {
2010 			ext_debug(inode, "append [%d]%d block to %u:[%d]%d"
2011 				  "(from %llu)\n",
2012 				  ext4_ext_is_unwritten(newext),
2013 				  ext4_ext_get_actual_len(newext),
2014 				  le32_to_cpu(ex->ee_block),
2015 				  ext4_ext_is_unwritten(ex),
2016 				  ext4_ext_get_actual_len(ex),
2017 				  ext4_ext_pblock(ex));
2018 			err = ext4_ext_get_access(handle, inode,
2019 						  path + depth);
2020 			if (err)
2021 				return err;
2022 			unwritten = ext4_ext_is_unwritten(ex);
2023 			ex->ee_len = cpu_to_le16(ext4_ext_get_actual_len(ex)
2024 					+ ext4_ext_get_actual_len(newext));
2025 			if (unwritten)
2026 				ext4_ext_mark_unwritten(ex);
2027 			nearex = ex;
2028 			goto merge;
2029 		}
2030 
2031 prepend:
2032 		/* Try to prepend newex to the ex */
2033 		if (ext4_can_extents_be_merged(inode, newext, ex)) {
2034 			ext_debug(inode, "prepend %u[%d]%d block to %u:[%d]%d"
2035 				  "(from %llu)\n",
2036 				  le32_to_cpu(newext->ee_block),
2037 				  ext4_ext_is_unwritten(newext),
2038 				  ext4_ext_get_actual_len(newext),
2039 				  le32_to_cpu(ex->ee_block),
2040 				  ext4_ext_is_unwritten(ex),
2041 				  ext4_ext_get_actual_len(ex),
2042 				  ext4_ext_pblock(ex));
2043 			err = ext4_ext_get_access(handle, inode,
2044 						  path + depth);
2045 			if (err)
2046 				return err;
2047 
2048 			unwritten = ext4_ext_is_unwritten(ex);
2049 			ex->ee_block = newext->ee_block;
2050 			ext4_ext_store_pblock(ex, ext4_ext_pblock(newext));
2051 			ex->ee_len = cpu_to_le16(ext4_ext_get_actual_len(ex)
2052 					+ ext4_ext_get_actual_len(newext));
2053 			if (unwritten)
2054 				ext4_ext_mark_unwritten(ex);
2055 			nearex = ex;
2056 			goto merge;
2057 		}
2058 	}
2059 
2060 	depth = ext_depth(inode);
2061 	eh = path[depth].p_hdr;
2062 	if (le16_to_cpu(eh->eh_entries) < le16_to_cpu(eh->eh_max))
2063 		goto has_space;
2064 
2065 	/* probably next leaf has space for us? */
2066 	fex = EXT_LAST_EXTENT(eh);
2067 	next = EXT_MAX_BLOCKS;
2068 	if (le32_to_cpu(newext->ee_block) > le32_to_cpu(fex->ee_block))
2069 		next = ext4_ext_next_leaf_block(path);
2070 	if (next != EXT_MAX_BLOCKS) {
2071 		ext_debug(inode, "next leaf block - %u\n", next);
2072 		BUG_ON(npath != NULL);
2073 		npath = ext4_find_extent(inode, next, NULL, gb_flags);
2074 		if (IS_ERR(npath))
2075 			return PTR_ERR(npath);
2076 		BUG_ON(npath->p_depth != path->p_depth);
2077 		eh = npath[depth].p_hdr;
2078 		if (le16_to_cpu(eh->eh_entries) < le16_to_cpu(eh->eh_max)) {
2079 			ext_debug(inode, "next leaf isn't full(%d)\n",
2080 				  le16_to_cpu(eh->eh_entries));
2081 			path = npath;
2082 			goto has_space;
2083 		}
2084 		ext_debug(inode, "next leaf has no free space(%d,%d)\n",
2085 			  le16_to_cpu(eh->eh_entries), le16_to_cpu(eh->eh_max));
2086 	}
2087 
2088 	/*
2089 	 * There is no free space in the found leaf.
2090 	 * We're gonna add a new leaf in the tree.
2091 	 */
2092 	if (gb_flags & EXT4_GET_BLOCKS_METADATA_NOFAIL)
2093 		mb_flags |= EXT4_MB_USE_RESERVED;
2094 	err = ext4_ext_create_new_leaf(handle, inode, mb_flags, gb_flags,
2095 				       ppath, newext);
2096 	if (err)
2097 		goto cleanup;
2098 	depth = ext_depth(inode);
2099 	eh = path[depth].p_hdr;
2100 
2101 has_space:
2102 	nearex = path[depth].p_ext;
2103 
2104 	err = ext4_ext_get_access(handle, inode, path + depth);
2105 	if (err)
2106 		goto cleanup;
2107 
2108 	if (!nearex) {
2109 		/* there is no extent in this leaf, create first one */
2110 		ext_debug(inode, "first extent in the leaf: %u:%llu:[%d]%d\n",
2111 				le32_to_cpu(newext->ee_block),
2112 				ext4_ext_pblock(newext),
2113 				ext4_ext_is_unwritten(newext),
2114 				ext4_ext_get_actual_len(newext));
2115 		nearex = EXT_FIRST_EXTENT(eh);
2116 	} else {
2117 		if (le32_to_cpu(newext->ee_block)
2118 			   > le32_to_cpu(nearex->ee_block)) {
2119 			/* Insert after */
2120 			ext_debug(inode, "insert %u:%llu:[%d]%d before: "
2121 					"nearest %p\n",
2122 					le32_to_cpu(newext->ee_block),
2123 					ext4_ext_pblock(newext),
2124 					ext4_ext_is_unwritten(newext),
2125 					ext4_ext_get_actual_len(newext),
2126 					nearex);
2127 			nearex++;
2128 		} else {
2129 			/* Insert before */
2130 			BUG_ON(newext->ee_block == nearex->ee_block);
2131 			ext_debug(inode, "insert %u:%llu:[%d]%d after: "
2132 					"nearest %p\n",
2133 					le32_to_cpu(newext->ee_block),
2134 					ext4_ext_pblock(newext),
2135 					ext4_ext_is_unwritten(newext),
2136 					ext4_ext_get_actual_len(newext),
2137 					nearex);
2138 		}
2139 		len = EXT_LAST_EXTENT(eh) - nearex + 1;
2140 		if (len > 0) {
2141 			ext_debug(inode, "insert %u:%llu:[%d]%d: "
2142 					"move %d extents from 0x%p to 0x%p\n",
2143 					le32_to_cpu(newext->ee_block),
2144 					ext4_ext_pblock(newext),
2145 					ext4_ext_is_unwritten(newext),
2146 					ext4_ext_get_actual_len(newext),
2147 					len, nearex, nearex + 1);
2148 			memmove(nearex + 1, nearex,
2149 				len * sizeof(struct ext4_extent));
2150 		}
2151 	}
2152 
2153 	le16_add_cpu(&eh->eh_entries, 1);
2154 	path[depth].p_ext = nearex;
2155 	nearex->ee_block = newext->ee_block;
2156 	ext4_ext_store_pblock(nearex, ext4_ext_pblock(newext));
2157 	nearex->ee_len = newext->ee_len;
2158 
2159 merge:
2160 	/* try to merge extents */
2161 	if (!(gb_flags & EXT4_GET_BLOCKS_PRE_IO))
2162 		ext4_ext_try_to_merge(handle, inode, path, nearex);
2163 
2164 
2165 	/* time to correct all indexes above */
2166 	err = ext4_ext_correct_indexes(handle, inode, path);
2167 	if (err)
2168 		goto cleanup;
2169 
2170 	err = ext4_ext_dirty(handle, inode, path + path->p_depth);
2171 
2172 cleanup:
2173 	ext4_ext_drop_refs(npath);
2174 	kfree(npath);
2175 	return err;
2176 }
2177 
2178 static int ext4_fill_es_cache_info(struct inode *inode,
2179 				   ext4_lblk_t block, ext4_lblk_t num,
2180 				   struct fiemap_extent_info *fieinfo)
2181 {
2182 	ext4_lblk_t next, end = block + num - 1;
2183 	struct extent_status es;
2184 	unsigned char blksize_bits = inode->i_sb->s_blocksize_bits;
2185 	unsigned int flags;
2186 	int err;
2187 
2188 	while (block <= end) {
2189 		next = 0;
2190 		flags = 0;
2191 		if (!ext4_es_lookup_extent(inode, block, &next, &es))
2192 			break;
2193 		if (ext4_es_is_unwritten(&es))
2194 			flags |= FIEMAP_EXTENT_UNWRITTEN;
2195 		if (ext4_es_is_delayed(&es))
2196 			flags |= (FIEMAP_EXTENT_DELALLOC |
2197 				  FIEMAP_EXTENT_UNKNOWN);
2198 		if (ext4_es_is_hole(&es))
2199 			flags |= EXT4_FIEMAP_EXTENT_HOLE;
2200 		if (next == 0)
2201 			flags |= FIEMAP_EXTENT_LAST;
2202 		if (flags & (FIEMAP_EXTENT_DELALLOC|
2203 			     EXT4_FIEMAP_EXTENT_HOLE))
2204 			es.es_pblk = 0;
2205 		else
2206 			es.es_pblk = ext4_es_pblock(&es);
2207 		err = fiemap_fill_next_extent(fieinfo,
2208 				(__u64)es.es_lblk << blksize_bits,
2209 				(__u64)es.es_pblk << blksize_bits,
2210 				(__u64)es.es_len << blksize_bits,
2211 				flags);
2212 		if (next == 0)
2213 			break;
2214 		block = next;
2215 		if (err < 0)
2216 			return err;
2217 		if (err == 1)
2218 			return 0;
2219 	}
2220 	return 0;
2221 }
2222 
2223 
2224 /*
2225  * ext4_ext_determine_hole - determine hole around given block
2226  * @inode:	inode we lookup in
2227  * @path:	path in extent tree to @lblk
2228  * @lblk:	pointer to logical block around which we want to determine hole
2229  *
2230  * Determine hole length (and start if easily possible) around given logical
2231  * block. We don't try too hard to find the beginning of the hole but @path
2232  * actually points to extent before @lblk, we provide it.
2233  *
2234  * The function returns the length of a hole starting at @lblk. We update @lblk
2235  * to the beginning of the hole if we managed to find it.
2236  */
2237 static ext4_lblk_t ext4_ext_determine_hole(struct inode *inode,
2238 					   struct ext4_ext_path *path,
2239 					   ext4_lblk_t *lblk)
2240 {
2241 	int depth = ext_depth(inode);
2242 	struct ext4_extent *ex;
2243 	ext4_lblk_t len;
2244 
2245 	ex = path[depth].p_ext;
2246 	if (ex == NULL) {
2247 		/* there is no extent yet, so gap is [0;-] */
2248 		*lblk = 0;
2249 		len = EXT_MAX_BLOCKS;
2250 	} else if (*lblk < le32_to_cpu(ex->ee_block)) {
2251 		len = le32_to_cpu(ex->ee_block) - *lblk;
2252 	} else if (*lblk >= le32_to_cpu(ex->ee_block)
2253 			+ ext4_ext_get_actual_len(ex)) {
2254 		ext4_lblk_t next;
2255 
2256 		*lblk = le32_to_cpu(ex->ee_block) + ext4_ext_get_actual_len(ex);
2257 		next = ext4_ext_next_allocated_block(path);
2258 		BUG_ON(next == *lblk);
2259 		len = next - *lblk;
2260 	} else {
2261 		BUG();
2262 	}
2263 	return len;
2264 }
2265 
2266 /*
2267  * ext4_ext_put_gap_in_cache:
2268  * calculate boundaries of the gap that the requested block fits into
2269  * and cache this gap
2270  */
2271 static void
2272 ext4_ext_put_gap_in_cache(struct inode *inode, ext4_lblk_t hole_start,
2273 			  ext4_lblk_t hole_len)
2274 {
2275 	struct extent_status es;
2276 
2277 	ext4_es_find_extent_range(inode, &ext4_es_is_delayed, hole_start,
2278 				  hole_start + hole_len - 1, &es);
2279 	if (es.es_len) {
2280 		/* There's delayed extent containing lblock? */
2281 		if (es.es_lblk <= hole_start)
2282 			return;
2283 		hole_len = min(es.es_lblk - hole_start, hole_len);
2284 	}
2285 	ext_debug(inode, " -> %u:%u\n", hole_start, hole_len);
2286 	ext4_es_insert_extent(inode, hole_start, hole_len, ~0,
2287 			      EXTENT_STATUS_HOLE);
2288 }
2289 
2290 /*
2291  * ext4_ext_rm_idx:
2292  * removes index from the index block.
2293  */
2294 static int ext4_ext_rm_idx(handle_t *handle, struct inode *inode,
2295 			struct ext4_ext_path *path, int depth)
2296 {
2297 	int err;
2298 	ext4_fsblk_t leaf;
2299 
2300 	/* free index block */
2301 	depth--;
2302 	path = path + depth;
2303 	leaf = ext4_idx_pblock(path->p_idx);
2304 	if (unlikely(path->p_hdr->eh_entries == 0)) {
2305 		EXT4_ERROR_INODE(inode, "path->p_hdr->eh_entries == 0");
2306 		return -EFSCORRUPTED;
2307 	}
2308 	err = ext4_ext_get_access(handle, inode, path);
2309 	if (err)
2310 		return err;
2311 
2312 	if (path->p_idx != EXT_LAST_INDEX(path->p_hdr)) {
2313 		int len = EXT_LAST_INDEX(path->p_hdr) - path->p_idx;
2314 		len *= sizeof(struct ext4_extent_idx);
2315 		memmove(path->p_idx, path->p_idx + 1, len);
2316 	}
2317 
2318 	le16_add_cpu(&path->p_hdr->eh_entries, -1);
2319 	err = ext4_ext_dirty(handle, inode, path);
2320 	if (err)
2321 		return err;
2322 	ext_debug(inode, "index is empty, remove it, free block %llu\n", leaf);
2323 	trace_ext4_ext_rm_idx(inode, leaf);
2324 
2325 	ext4_free_blocks(handle, inode, NULL, leaf, 1,
2326 			 EXT4_FREE_BLOCKS_METADATA | EXT4_FREE_BLOCKS_FORGET);
2327 
2328 	while (--depth >= 0) {
2329 		if (path->p_idx != EXT_FIRST_INDEX(path->p_hdr))
2330 			break;
2331 		path--;
2332 		err = ext4_ext_get_access(handle, inode, path);
2333 		if (err)
2334 			break;
2335 		path->p_idx->ei_block = (path+1)->p_idx->ei_block;
2336 		err = ext4_ext_dirty(handle, inode, path);
2337 		if (err)
2338 			break;
2339 	}
2340 	return err;
2341 }
2342 
2343 /*
2344  * ext4_ext_calc_credits_for_single_extent:
2345  * This routine returns max. credits that needed to insert an extent
2346  * to the extent tree.
2347  * When pass the actual path, the caller should calculate credits
2348  * under i_data_sem.
2349  */
2350 int ext4_ext_calc_credits_for_single_extent(struct inode *inode, int nrblocks,
2351 						struct ext4_ext_path *path)
2352 {
2353 	if (path) {
2354 		int depth = ext_depth(inode);
2355 		int ret = 0;
2356 
2357 		/* probably there is space in leaf? */
2358 		if (le16_to_cpu(path[depth].p_hdr->eh_entries)
2359 				< le16_to_cpu(path[depth].p_hdr->eh_max)) {
2360 
2361 			/*
2362 			 *  There are some space in the leaf tree, no
2363 			 *  need to account for leaf block credit
2364 			 *
2365 			 *  bitmaps and block group descriptor blocks
2366 			 *  and other metadata blocks still need to be
2367 			 *  accounted.
2368 			 */
2369 			/* 1 bitmap, 1 block group descriptor */
2370 			ret = 2 + EXT4_META_TRANS_BLOCKS(inode->i_sb);
2371 			return ret;
2372 		}
2373 	}
2374 
2375 	return ext4_chunk_trans_blocks(inode, nrblocks);
2376 }
2377 
2378 /*
2379  * How many index/leaf blocks need to change/allocate to add @extents extents?
2380  *
2381  * If we add a single extent, then in the worse case, each tree level
2382  * index/leaf need to be changed in case of the tree split.
2383  *
2384  * If more extents are inserted, they could cause the whole tree split more
2385  * than once, but this is really rare.
2386  */
2387 int ext4_ext_index_trans_blocks(struct inode *inode, int extents)
2388 {
2389 	int index;
2390 	int depth;
2391 
2392 	/* If we are converting the inline data, only one is needed here. */
2393 	if (ext4_has_inline_data(inode))
2394 		return 1;
2395 
2396 	depth = ext_depth(inode);
2397 
2398 	if (extents <= 1)
2399 		index = depth * 2;
2400 	else
2401 		index = depth * 3;
2402 
2403 	return index;
2404 }
2405 
2406 static inline int get_default_free_blocks_flags(struct inode *inode)
2407 {
2408 	if (S_ISDIR(inode->i_mode) || S_ISLNK(inode->i_mode) ||
2409 	    ext4_test_inode_flag(inode, EXT4_INODE_EA_INODE))
2410 		return EXT4_FREE_BLOCKS_METADATA | EXT4_FREE_BLOCKS_FORGET;
2411 	else if (ext4_should_journal_data(inode))
2412 		return EXT4_FREE_BLOCKS_FORGET;
2413 	return 0;
2414 }
2415 
2416 /*
2417  * ext4_rereserve_cluster - increment the reserved cluster count when
2418  *                          freeing a cluster with a pending reservation
2419  *
2420  * @inode - file containing the cluster
2421  * @lblk - logical block in cluster to be reserved
2422  *
2423  * Increments the reserved cluster count and adjusts quota in a bigalloc
2424  * file system when freeing a partial cluster containing at least one
2425  * delayed and unwritten block.  A partial cluster meeting that
2426  * requirement will have a pending reservation.  If so, the
2427  * RERESERVE_CLUSTER flag is used when calling ext4_free_blocks() to
2428  * defer reserved and allocated space accounting to a subsequent call
2429  * to this function.
2430  */
2431 static void ext4_rereserve_cluster(struct inode *inode, ext4_lblk_t lblk)
2432 {
2433 	struct ext4_sb_info *sbi = EXT4_SB(inode->i_sb);
2434 	struct ext4_inode_info *ei = EXT4_I(inode);
2435 
2436 	dquot_reclaim_block(inode, EXT4_C2B(sbi, 1));
2437 
2438 	spin_lock(&ei->i_block_reservation_lock);
2439 	ei->i_reserved_data_blocks++;
2440 	percpu_counter_add(&sbi->s_dirtyclusters_counter, 1);
2441 	spin_unlock(&ei->i_block_reservation_lock);
2442 
2443 	percpu_counter_add(&sbi->s_freeclusters_counter, 1);
2444 	ext4_remove_pending(inode, lblk);
2445 }
2446 
2447 static int ext4_remove_blocks(handle_t *handle, struct inode *inode,
2448 			      struct ext4_extent *ex,
2449 			      struct partial_cluster *partial,
2450 			      ext4_lblk_t from, ext4_lblk_t to)
2451 {
2452 	struct ext4_sb_info *sbi = EXT4_SB(inode->i_sb);
2453 	unsigned short ee_len = ext4_ext_get_actual_len(ex);
2454 	ext4_fsblk_t last_pblk, pblk;
2455 	ext4_lblk_t num;
2456 	int flags;
2457 
2458 	/* only extent tail removal is allowed */
2459 	if (from < le32_to_cpu(ex->ee_block) ||
2460 	    to != le32_to_cpu(ex->ee_block) + ee_len - 1) {
2461 		ext4_error(sbi->s_sb,
2462 			   "strange request: removal(2) %u-%u from %u:%u",
2463 			   from, to, le32_to_cpu(ex->ee_block), ee_len);
2464 		return 0;
2465 	}
2466 
2467 #ifdef EXTENTS_STATS
2468 	spin_lock(&sbi->s_ext_stats_lock);
2469 	sbi->s_ext_blocks += ee_len;
2470 	sbi->s_ext_extents++;
2471 	if (ee_len < sbi->s_ext_min)
2472 		sbi->s_ext_min = ee_len;
2473 	if (ee_len > sbi->s_ext_max)
2474 		sbi->s_ext_max = ee_len;
2475 	if (ext_depth(inode) > sbi->s_depth_max)
2476 		sbi->s_depth_max = ext_depth(inode);
2477 	spin_unlock(&sbi->s_ext_stats_lock);
2478 #endif
2479 
2480 	trace_ext4_remove_blocks(inode, ex, from, to, partial);
2481 
2482 	/*
2483 	 * if we have a partial cluster, and it's different from the
2484 	 * cluster of the last block in the extent, we free it
2485 	 */
2486 	last_pblk = ext4_ext_pblock(ex) + ee_len - 1;
2487 
2488 	if (partial->state != initial &&
2489 	    partial->pclu != EXT4_B2C(sbi, last_pblk)) {
2490 		if (partial->state == tofree) {
2491 			flags = get_default_free_blocks_flags(inode);
2492 			if (ext4_is_pending(inode, partial->lblk))
2493 				flags |= EXT4_FREE_BLOCKS_RERESERVE_CLUSTER;
2494 			ext4_free_blocks(handle, inode, NULL,
2495 					 EXT4_C2B(sbi, partial->pclu),
2496 					 sbi->s_cluster_ratio, flags);
2497 			if (flags & EXT4_FREE_BLOCKS_RERESERVE_CLUSTER)
2498 				ext4_rereserve_cluster(inode, partial->lblk);
2499 		}
2500 		partial->state = initial;
2501 	}
2502 
2503 	num = le32_to_cpu(ex->ee_block) + ee_len - from;
2504 	pblk = ext4_ext_pblock(ex) + ee_len - num;
2505 
2506 	/*
2507 	 * We free the partial cluster at the end of the extent (if any),
2508 	 * unless the cluster is used by another extent (partial_cluster
2509 	 * state is nofree).  If a partial cluster exists here, it must be
2510 	 * shared with the last block in the extent.
2511 	 */
2512 	flags = get_default_free_blocks_flags(inode);
2513 
2514 	/* partial, left end cluster aligned, right end unaligned */
2515 	if ((EXT4_LBLK_COFF(sbi, to) != sbi->s_cluster_ratio - 1) &&
2516 	    (EXT4_LBLK_CMASK(sbi, to) >= from) &&
2517 	    (partial->state != nofree)) {
2518 		if (ext4_is_pending(inode, to))
2519 			flags |= EXT4_FREE_BLOCKS_RERESERVE_CLUSTER;
2520 		ext4_free_blocks(handle, inode, NULL,
2521 				 EXT4_PBLK_CMASK(sbi, last_pblk),
2522 				 sbi->s_cluster_ratio, flags);
2523 		if (flags & EXT4_FREE_BLOCKS_RERESERVE_CLUSTER)
2524 			ext4_rereserve_cluster(inode, to);
2525 		partial->state = initial;
2526 		flags = get_default_free_blocks_flags(inode);
2527 	}
2528 
2529 	flags |= EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER;
2530 
2531 	/*
2532 	 * For bigalloc file systems, we never free a partial cluster
2533 	 * at the beginning of the extent.  Instead, we check to see if we
2534 	 * need to free it on a subsequent call to ext4_remove_blocks,
2535 	 * or at the end of ext4_ext_rm_leaf or ext4_ext_remove_space.
2536 	 */
2537 	flags |= EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER;
2538 	ext4_free_blocks(handle, inode, NULL, pblk, num, flags);
2539 
2540 	/* reset the partial cluster if we've freed past it */
2541 	if (partial->state != initial && partial->pclu != EXT4_B2C(sbi, pblk))
2542 		partial->state = initial;
2543 
2544 	/*
2545 	 * If we've freed the entire extent but the beginning is not left
2546 	 * cluster aligned and is not marked as ineligible for freeing we
2547 	 * record the partial cluster at the beginning of the extent.  It
2548 	 * wasn't freed by the preceding ext4_free_blocks() call, and we
2549 	 * need to look farther to the left to determine if it's to be freed
2550 	 * (not shared with another extent). Else, reset the partial
2551 	 * cluster - we're either  done freeing or the beginning of the
2552 	 * extent is left cluster aligned.
2553 	 */
2554 	if (EXT4_LBLK_COFF(sbi, from) && num == ee_len) {
2555 		if (partial->state == initial) {
2556 			partial->pclu = EXT4_B2C(sbi, pblk);
2557 			partial->lblk = from;
2558 			partial->state = tofree;
2559 		}
2560 	} else {
2561 		partial->state = initial;
2562 	}
2563 
2564 	return 0;
2565 }
2566 
2567 /*
2568  * ext4_ext_rm_leaf() Removes the extents associated with the
2569  * blocks appearing between "start" and "end".  Both "start"
2570  * and "end" must appear in the same extent or EIO is returned.
2571  *
2572  * @handle: The journal handle
2573  * @inode:  The files inode
2574  * @path:   The path to the leaf
2575  * @partial_cluster: The cluster which we'll have to free if all extents
2576  *                   has been released from it.  However, if this value is
2577  *                   negative, it's a cluster just to the right of the
2578  *                   punched region and it must not be freed.
2579  * @start:  The first block to remove
2580  * @end:   The last block to remove
2581  */
2582 static int
2583 ext4_ext_rm_leaf(handle_t *handle, struct inode *inode,
2584 		 struct ext4_ext_path *path,
2585 		 struct partial_cluster *partial,
2586 		 ext4_lblk_t start, ext4_lblk_t end)
2587 {
2588 	struct ext4_sb_info *sbi = EXT4_SB(inode->i_sb);
2589 	int err = 0, correct_index = 0;
2590 	int depth = ext_depth(inode), credits, revoke_credits;
2591 	struct ext4_extent_header *eh;
2592 	ext4_lblk_t a, b;
2593 	unsigned num;
2594 	ext4_lblk_t ex_ee_block;
2595 	unsigned short ex_ee_len;
2596 	unsigned unwritten = 0;
2597 	struct ext4_extent *ex;
2598 	ext4_fsblk_t pblk;
2599 
2600 	/* the header must be checked already in ext4_ext_remove_space() */
2601 	ext_debug(inode, "truncate since %u in leaf to %u\n", start, end);
2602 	if (!path[depth].p_hdr)
2603 		path[depth].p_hdr = ext_block_hdr(path[depth].p_bh);
2604 	eh = path[depth].p_hdr;
2605 	if (unlikely(path[depth].p_hdr == NULL)) {
2606 		EXT4_ERROR_INODE(inode, "path[%d].p_hdr == NULL", depth);
2607 		return -EFSCORRUPTED;
2608 	}
2609 	/* find where to start removing */
2610 	ex = path[depth].p_ext;
2611 	if (!ex)
2612 		ex = EXT_LAST_EXTENT(eh);
2613 
2614 	ex_ee_block = le32_to_cpu(ex->ee_block);
2615 	ex_ee_len = ext4_ext_get_actual_len(ex);
2616 
2617 	trace_ext4_ext_rm_leaf(inode, start, ex, partial);
2618 
2619 	while (ex >= EXT_FIRST_EXTENT(eh) &&
2620 			ex_ee_block + ex_ee_len > start) {
2621 
2622 		if (ext4_ext_is_unwritten(ex))
2623 			unwritten = 1;
2624 		else
2625 			unwritten = 0;
2626 
2627 		ext_debug(inode, "remove ext %u:[%d]%d\n", ex_ee_block,
2628 			  unwritten, ex_ee_len);
2629 		path[depth].p_ext = ex;
2630 
2631 		a = ex_ee_block > start ? ex_ee_block : start;
2632 		b = ex_ee_block+ex_ee_len - 1 < end ?
2633 			ex_ee_block+ex_ee_len - 1 : end;
2634 
2635 		ext_debug(inode, "  border %u:%u\n", a, b);
2636 
2637 		/* If this extent is beyond the end of the hole, skip it */
2638 		if (end < ex_ee_block) {
2639 			/*
2640 			 * We're going to skip this extent and move to another,
2641 			 * so note that its first cluster is in use to avoid
2642 			 * freeing it when removing blocks.  Eventually, the
2643 			 * right edge of the truncated/punched region will
2644 			 * be just to the left.
2645 			 */
2646 			if (sbi->s_cluster_ratio > 1) {
2647 				pblk = ext4_ext_pblock(ex);
2648 				partial->pclu = EXT4_B2C(sbi, pblk);
2649 				partial->state = nofree;
2650 			}
2651 			ex--;
2652 			ex_ee_block = le32_to_cpu(ex->ee_block);
2653 			ex_ee_len = ext4_ext_get_actual_len(ex);
2654 			continue;
2655 		} else if (b != ex_ee_block + ex_ee_len - 1) {
2656 			EXT4_ERROR_INODE(inode,
2657 					 "can not handle truncate %u:%u "
2658 					 "on extent %u:%u",
2659 					 start, end, ex_ee_block,
2660 					 ex_ee_block + ex_ee_len - 1);
2661 			err = -EFSCORRUPTED;
2662 			goto out;
2663 		} else if (a != ex_ee_block) {
2664 			/* remove tail of the extent */
2665 			num = a - ex_ee_block;
2666 		} else {
2667 			/* remove whole extent: excellent! */
2668 			num = 0;
2669 		}
2670 		/*
2671 		 * 3 for leaf, sb, and inode plus 2 (bmap and group
2672 		 * descriptor) for each block group; assume two block
2673 		 * groups plus ex_ee_len/blocks_per_block_group for
2674 		 * the worst case
2675 		 */
2676 		credits = 7 + 2*(ex_ee_len/EXT4_BLOCKS_PER_GROUP(inode->i_sb));
2677 		if (ex == EXT_FIRST_EXTENT(eh)) {
2678 			correct_index = 1;
2679 			credits += (ext_depth(inode)) + 1;
2680 		}
2681 		credits += EXT4_MAXQUOTAS_TRANS_BLOCKS(inode->i_sb);
2682 		/*
2683 		 * We may end up freeing some index blocks and data from the
2684 		 * punched range. Note that partial clusters are accounted for
2685 		 * by ext4_free_data_revoke_credits().
2686 		 */
2687 		revoke_credits =
2688 			ext4_free_metadata_revoke_credits(inode->i_sb,
2689 							  ext_depth(inode)) +
2690 			ext4_free_data_revoke_credits(inode, b - a + 1);
2691 
2692 		err = ext4_datasem_ensure_credits(handle, inode, credits,
2693 						  credits, revoke_credits);
2694 		if (err) {
2695 			if (err > 0)
2696 				err = -EAGAIN;
2697 			goto out;
2698 		}
2699 
2700 		err = ext4_ext_get_access(handle, inode, path + depth);
2701 		if (err)
2702 			goto out;
2703 
2704 		err = ext4_remove_blocks(handle, inode, ex, partial, a, b);
2705 		if (err)
2706 			goto out;
2707 
2708 		if (num == 0)
2709 			/* this extent is removed; mark slot entirely unused */
2710 			ext4_ext_store_pblock(ex, 0);
2711 
2712 		ex->ee_len = cpu_to_le16(num);
2713 		/*
2714 		 * Do not mark unwritten if all the blocks in the
2715 		 * extent have been removed.
2716 		 */
2717 		if (unwritten && num)
2718 			ext4_ext_mark_unwritten(ex);
2719 		/*
2720 		 * If the extent was completely released,
2721 		 * we need to remove it from the leaf
2722 		 */
2723 		if (num == 0) {
2724 			if (end != EXT_MAX_BLOCKS - 1) {
2725 				/*
2726 				 * For hole punching, we need to scoot all the
2727 				 * extents up when an extent is removed so that
2728 				 * we dont have blank extents in the middle
2729 				 */
2730 				memmove(ex, ex+1, (EXT_LAST_EXTENT(eh) - ex) *
2731 					sizeof(struct ext4_extent));
2732 
2733 				/* Now get rid of the one at the end */
2734 				memset(EXT_LAST_EXTENT(eh), 0,
2735 					sizeof(struct ext4_extent));
2736 			}
2737 			le16_add_cpu(&eh->eh_entries, -1);
2738 		}
2739 
2740 		err = ext4_ext_dirty(handle, inode, path + depth);
2741 		if (err)
2742 			goto out;
2743 
2744 		ext_debug(inode, "new extent: %u:%u:%llu\n", ex_ee_block, num,
2745 				ext4_ext_pblock(ex));
2746 		ex--;
2747 		ex_ee_block = le32_to_cpu(ex->ee_block);
2748 		ex_ee_len = ext4_ext_get_actual_len(ex);
2749 	}
2750 
2751 	if (correct_index && eh->eh_entries)
2752 		err = ext4_ext_correct_indexes(handle, inode, path);
2753 
2754 	/*
2755 	 * If there's a partial cluster and at least one extent remains in
2756 	 * the leaf, free the partial cluster if it isn't shared with the
2757 	 * current extent.  If it is shared with the current extent
2758 	 * we reset the partial cluster because we've reached the start of the
2759 	 * truncated/punched region and we're done removing blocks.
2760 	 */
2761 	if (partial->state == tofree && ex >= EXT_FIRST_EXTENT(eh)) {
2762 		pblk = ext4_ext_pblock(ex) + ex_ee_len - 1;
2763 		if (partial->pclu != EXT4_B2C(sbi, pblk)) {
2764 			int flags = get_default_free_blocks_flags(inode);
2765 
2766 			if (ext4_is_pending(inode, partial->lblk))
2767 				flags |= EXT4_FREE_BLOCKS_RERESERVE_CLUSTER;
2768 			ext4_free_blocks(handle, inode, NULL,
2769 					 EXT4_C2B(sbi, partial->pclu),
2770 					 sbi->s_cluster_ratio, flags);
2771 			if (flags & EXT4_FREE_BLOCKS_RERESERVE_CLUSTER)
2772 				ext4_rereserve_cluster(inode, partial->lblk);
2773 		}
2774 		partial->state = initial;
2775 	}
2776 
2777 	/* if this leaf is free, then we should
2778 	 * remove it from index block above */
2779 	if (err == 0 && eh->eh_entries == 0 && path[depth].p_bh != NULL)
2780 		err = ext4_ext_rm_idx(handle, inode, path, depth);
2781 
2782 out:
2783 	return err;
2784 }
2785 
2786 /*
2787  * ext4_ext_more_to_rm:
2788  * returns 1 if current index has to be freed (even partial)
2789  */
2790 static int
2791 ext4_ext_more_to_rm(struct ext4_ext_path *path)
2792 {
2793 	BUG_ON(path->p_idx == NULL);
2794 
2795 	if (path->p_idx < EXT_FIRST_INDEX(path->p_hdr))
2796 		return 0;
2797 
2798 	/*
2799 	 * if truncate on deeper level happened, it wasn't partial,
2800 	 * so we have to consider current index for truncation
2801 	 */
2802 	if (le16_to_cpu(path->p_hdr->eh_entries) == path->p_block)
2803 		return 0;
2804 	return 1;
2805 }
2806 
2807 int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start,
2808 			  ext4_lblk_t end)
2809 {
2810 	struct ext4_sb_info *sbi = EXT4_SB(inode->i_sb);
2811 	int depth = ext_depth(inode);
2812 	struct ext4_ext_path *path = NULL;
2813 	struct partial_cluster partial;
2814 	handle_t *handle;
2815 	int i = 0, err = 0;
2816 
2817 	partial.pclu = 0;
2818 	partial.lblk = 0;
2819 	partial.state = initial;
2820 
2821 	ext_debug(inode, "truncate since %u to %u\n", start, end);
2822 
2823 	/* probably first extent we're gonna free will be last in block */
2824 	handle = ext4_journal_start_with_revoke(inode, EXT4_HT_TRUNCATE,
2825 			depth + 1,
2826 			ext4_free_metadata_revoke_credits(inode->i_sb, depth));
2827 	if (IS_ERR(handle))
2828 		return PTR_ERR(handle);
2829 
2830 again:
2831 	trace_ext4_ext_remove_space(inode, start, end, depth);
2832 
2833 	/*
2834 	 * Check if we are removing extents inside the extent tree. If that
2835 	 * is the case, we are going to punch a hole inside the extent tree
2836 	 * so we have to check whether we need to split the extent covering
2837 	 * the last block to remove so we can easily remove the part of it
2838 	 * in ext4_ext_rm_leaf().
2839 	 */
2840 	if (end < EXT_MAX_BLOCKS - 1) {
2841 		struct ext4_extent *ex;
2842 		ext4_lblk_t ee_block, ex_end, lblk;
2843 		ext4_fsblk_t pblk;
2844 
2845 		/* find extent for or closest extent to this block */
2846 		path = ext4_find_extent(inode, end, NULL,
2847 					EXT4_EX_NOCACHE | EXT4_EX_NOFAIL);
2848 		if (IS_ERR(path)) {
2849 			ext4_journal_stop(handle);
2850 			return PTR_ERR(path);
2851 		}
2852 		depth = ext_depth(inode);
2853 		/* Leaf not may not exist only if inode has no blocks at all */
2854 		ex = path[depth].p_ext;
2855 		if (!ex) {
2856 			if (depth) {
2857 				EXT4_ERROR_INODE(inode,
2858 						 "path[%d].p_hdr == NULL",
2859 						 depth);
2860 				err = -EFSCORRUPTED;
2861 			}
2862 			goto out;
2863 		}
2864 
2865 		ee_block = le32_to_cpu(ex->ee_block);
2866 		ex_end = ee_block + ext4_ext_get_actual_len(ex) - 1;
2867 
2868 		/*
2869 		 * See if the last block is inside the extent, if so split
2870 		 * the extent at 'end' block so we can easily remove the
2871 		 * tail of the first part of the split extent in
2872 		 * ext4_ext_rm_leaf().
2873 		 */
2874 		if (end >= ee_block && end < ex_end) {
2875 
2876 			/*
2877 			 * If we're going to split the extent, note that
2878 			 * the cluster containing the block after 'end' is
2879 			 * in use to avoid freeing it when removing blocks.
2880 			 */
2881 			if (sbi->s_cluster_ratio > 1) {
2882 				pblk = ext4_ext_pblock(ex) + end - ee_block + 1;
2883 				partial.pclu = EXT4_B2C(sbi, pblk);
2884 				partial.state = nofree;
2885 			}
2886 
2887 			/*
2888 			 * Split the extent in two so that 'end' is the last
2889 			 * block in the first new extent. Also we should not
2890 			 * fail removing space due to ENOSPC so try to use
2891 			 * reserved block if that happens.
2892 			 */
2893 			err = ext4_force_split_extent_at(handle, inode, &path,
2894 							 end + 1, 1);
2895 			if (err < 0)
2896 				goto out;
2897 
2898 		} else if (sbi->s_cluster_ratio > 1 && end >= ex_end &&
2899 			   partial.state == initial) {
2900 			/*
2901 			 * If we're punching, there's an extent to the right.
2902 			 * If the partial cluster hasn't been set, set it to
2903 			 * that extent's first cluster and its state to nofree
2904 			 * so it won't be freed should it contain blocks to be
2905 			 * removed. If it's already set (tofree/nofree), we're
2906 			 * retrying and keep the original partial cluster info
2907 			 * so a cluster marked tofree as a result of earlier
2908 			 * extent removal is not lost.
2909 			 */
2910 			lblk = ex_end + 1;
2911 			err = ext4_ext_search_right(inode, path, &lblk, &pblk,
2912 						    NULL);
2913 			if (err < 0)
2914 				goto out;
2915 			if (pblk) {
2916 				partial.pclu = EXT4_B2C(sbi, pblk);
2917 				partial.state = nofree;
2918 			}
2919 		}
2920 	}
2921 	/*
2922 	 * We start scanning from right side, freeing all the blocks
2923 	 * after i_size and walking into the tree depth-wise.
2924 	 */
2925 	depth = ext_depth(inode);
2926 	if (path) {
2927 		int k = i = depth;
2928 		while (--k > 0)
2929 			path[k].p_block =
2930 				le16_to_cpu(path[k].p_hdr->eh_entries)+1;
2931 	} else {
2932 		path = kcalloc(depth + 1, sizeof(struct ext4_ext_path),
2933 			       GFP_NOFS | __GFP_NOFAIL);
2934 		if (path == NULL) {
2935 			ext4_journal_stop(handle);
2936 			return -ENOMEM;
2937 		}
2938 		path[0].p_maxdepth = path[0].p_depth = depth;
2939 		path[0].p_hdr = ext_inode_hdr(inode);
2940 		i = 0;
2941 
2942 		if (ext4_ext_check(inode, path[0].p_hdr, depth, 0)) {
2943 			err = -EFSCORRUPTED;
2944 			goto out;
2945 		}
2946 	}
2947 	err = 0;
2948 
2949 	while (i >= 0 && err == 0) {
2950 		if (i == depth) {
2951 			/* this is leaf block */
2952 			err = ext4_ext_rm_leaf(handle, inode, path,
2953 					       &partial, start, end);
2954 			/* root level has p_bh == NULL, brelse() eats this */
2955 			brelse(path[i].p_bh);
2956 			path[i].p_bh = NULL;
2957 			i--;
2958 			continue;
2959 		}
2960 
2961 		/* this is index block */
2962 		if (!path[i].p_hdr) {
2963 			ext_debug(inode, "initialize header\n");
2964 			path[i].p_hdr = ext_block_hdr(path[i].p_bh);
2965 		}
2966 
2967 		if (!path[i].p_idx) {
2968 			/* this level hasn't been touched yet */
2969 			path[i].p_idx = EXT_LAST_INDEX(path[i].p_hdr);
2970 			path[i].p_block = le16_to_cpu(path[i].p_hdr->eh_entries)+1;
2971 			ext_debug(inode, "init index ptr: hdr 0x%p, num %d\n",
2972 				  path[i].p_hdr,
2973 				  le16_to_cpu(path[i].p_hdr->eh_entries));
2974 		} else {
2975 			/* we were already here, see at next index */
2976 			path[i].p_idx--;
2977 		}
2978 
2979 		ext_debug(inode, "level %d - index, first 0x%p, cur 0x%p\n",
2980 				i, EXT_FIRST_INDEX(path[i].p_hdr),
2981 				path[i].p_idx);
2982 		if (ext4_ext_more_to_rm(path + i)) {
2983 			struct buffer_head *bh;
2984 			/* go to the next level */
2985 			ext_debug(inode, "move to level %d (block %llu)\n",
2986 				  i + 1, ext4_idx_pblock(path[i].p_idx));
2987 			memset(path + i + 1, 0, sizeof(*path));
2988 			bh = read_extent_tree_block(inode, path[i].p_idx,
2989 						    depth - i - 1,
2990 						    EXT4_EX_NOCACHE);
2991 			if (IS_ERR(bh)) {
2992 				/* should we reset i_size? */
2993 				err = PTR_ERR(bh);
2994 				break;
2995 			}
2996 			/* Yield here to deal with large extent trees.
2997 			 * Should be a no-op if we did IO above. */
2998 			cond_resched();
2999 			if (WARN_ON(i + 1 > depth)) {
3000 				err = -EFSCORRUPTED;
3001 				break;
3002 			}
3003 			path[i + 1].p_bh = bh;
3004 
3005 			/* save actual number of indexes since this
3006 			 * number is changed at the next iteration */
3007 			path[i].p_block = le16_to_cpu(path[i].p_hdr->eh_entries);
3008 			i++;
3009 		} else {
3010 			/* we finished processing this index, go up */
3011 			if (path[i].p_hdr->eh_entries == 0 && i > 0) {
3012 				/* index is empty, remove it;
3013 				 * handle must be already prepared by the
3014 				 * truncatei_leaf() */
3015 				err = ext4_ext_rm_idx(handle, inode, path, i);
3016 			}
3017 			/* root level has p_bh == NULL, brelse() eats this */
3018 			brelse(path[i].p_bh);
3019 			path[i].p_bh = NULL;
3020 			i--;
3021 			ext_debug(inode, "return to level %d\n", i);
3022 		}
3023 	}
3024 
3025 	trace_ext4_ext_remove_space_done(inode, start, end, depth, &partial,
3026 					 path->p_hdr->eh_entries);
3027 
3028 	/*
3029 	 * if there's a partial cluster and we have removed the first extent
3030 	 * in the file, then we also free the partial cluster, if any
3031 	 */
3032 	if (partial.state == tofree && err == 0) {
3033 		int flags = get_default_free_blocks_flags(inode);
3034 
3035 		if (ext4_is_pending(inode, partial.lblk))
3036 			flags |= EXT4_FREE_BLOCKS_RERESERVE_CLUSTER;
3037 		ext4_free_blocks(handle, inode, NULL,
3038 				 EXT4_C2B(sbi, partial.pclu),
3039 				 sbi->s_cluster_ratio, flags);
3040 		if (flags & EXT4_FREE_BLOCKS_RERESERVE_CLUSTER)
3041 			ext4_rereserve_cluster(inode, partial.lblk);
3042 		partial.state = initial;
3043 	}
3044 
3045 	/* TODO: flexible tree reduction should be here */
3046 	if (path->p_hdr->eh_entries == 0) {
3047 		/*
3048 		 * truncate to zero freed all the tree,
3049 		 * so we need to correct eh_depth
3050 		 */
3051 		err = ext4_ext_get_access(handle, inode, path);
3052 		if (err == 0) {
3053 			ext_inode_hdr(inode)->eh_depth = 0;
3054 			ext_inode_hdr(inode)->eh_max =
3055 				cpu_to_le16(ext4_ext_space_root(inode, 0));
3056 			err = ext4_ext_dirty(handle, inode, path);
3057 		}
3058 	}
3059 out:
3060 	ext4_ext_drop_refs(path);
3061 	kfree(path);
3062 	path = NULL;
3063 	if (err == -EAGAIN)
3064 		goto again;
3065 	ext4_journal_stop(handle);
3066 
3067 	return err;
3068 }
3069 
3070 /*
3071  * called at mount time
3072  */
3073 void ext4_ext_init(struct super_block *sb)
3074 {
3075 	/*
3076 	 * possible initialization would be here
3077 	 */
3078 
3079 	if (ext4_has_feature_extents(sb)) {
3080 #if defined(AGGRESSIVE_TEST) || defined(CHECK_BINSEARCH) || defined(EXTENTS_STATS)
3081 		printk(KERN_INFO "EXT4-fs: file extents enabled"
3082 #ifdef AGGRESSIVE_TEST
3083 		       ", aggressive tests"
3084 #endif
3085 #ifdef CHECK_BINSEARCH
3086 		       ", check binsearch"
3087 #endif
3088 #ifdef EXTENTS_STATS
3089 		       ", stats"
3090 #endif
3091 		       "\n");
3092 #endif
3093 #ifdef EXTENTS_STATS
3094 		spin_lock_init(&EXT4_SB(sb)->s_ext_stats_lock);
3095 		EXT4_SB(sb)->s_ext_min = 1 << 30;
3096 		EXT4_SB(sb)->s_ext_max = 0;
3097 #endif
3098 	}
3099 }
3100 
3101 /*
3102  * called at umount time
3103  */
3104 void ext4_ext_release(struct super_block *sb)
3105 {
3106 	if (!ext4_has_feature_extents(sb))
3107 		return;
3108 
3109 #ifdef EXTENTS_STATS
3110 	if (EXT4_SB(sb)->s_ext_blocks && EXT4_SB(sb)->s_ext_extents) {
3111 		struct ext4_sb_info *sbi = EXT4_SB(sb);
3112 		printk(KERN_ERR "EXT4-fs: %lu blocks in %lu extents (%lu ave)\n",
3113 			sbi->s_ext_blocks, sbi->s_ext_extents,
3114 			sbi->s_ext_blocks / sbi->s_ext_extents);
3115 		printk(KERN_ERR "EXT4-fs: extents: %lu min, %lu max, max depth %lu\n",
3116 			sbi->s_ext_min, sbi->s_ext_max, sbi->s_depth_max);
3117 	}
3118 #endif
3119 }
3120 
3121 static int ext4_zeroout_es(struct inode *inode, struct ext4_extent *ex)
3122 {
3123 	ext4_lblk_t  ee_block;
3124 	ext4_fsblk_t ee_pblock;
3125 	unsigned int ee_len;
3126 
3127 	ee_block  = le32_to_cpu(ex->ee_block);
3128 	ee_len    = ext4_ext_get_actual_len(ex);
3129 	ee_pblock = ext4_ext_pblock(ex);
3130 
3131 	if (ee_len == 0)
3132 		return 0;
3133 
3134 	return ext4_es_insert_extent(inode, ee_block, ee_len, ee_pblock,
3135 				     EXTENT_STATUS_WRITTEN);
3136 }
3137 
3138 /* FIXME!! we need to try to merge to left or right after zero-out  */
3139 static int ext4_ext_zeroout(struct inode *inode, struct ext4_extent *ex)
3140 {
3141 	ext4_fsblk_t ee_pblock;
3142 	unsigned int ee_len;
3143 
3144 	ee_len    = ext4_ext_get_actual_len(ex);
3145 	ee_pblock = ext4_ext_pblock(ex);
3146 	return ext4_issue_zeroout(inode, le32_to_cpu(ex->ee_block), ee_pblock,
3147 				  ee_len);
3148 }
3149 
3150 /*
3151  * ext4_split_extent_at() splits an extent at given block.
3152  *
3153  * @handle: the journal handle
3154  * @inode: the file inode
3155  * @path: the path to the extent
3156  * @split: the logical block where the extent is splitted.
3157  * @split_flags: indicates if the extent could be zeroout if split fails, and
3158  *		 the states(init or unwritten) of new extents.
3159  * @flags: flags used to insert new extent to extent tree.
3160  *
3161  *
3162  * Splits extent [a, b] into two extents [a, @split) and [@split, b], states
3163  * of which are determined by split_flag.
3164  *
3165  * There are two cases:
3166  *  a> the extent are splitted into two extent.
3167  *  b> split is not needed, and just mark the extent.
3168  *
3169  * return 0 on success.
3170  */
3171 static int ext4_split_extent_at(handle_t *handle,
3172 			     struct inode *inode,
3173 			     struct ext4_ext_path **ppath,
3174 			     ext4_lblk_t split,
3175 			     int split_flag,
3176 			     int flags)
3177 {
3178 	struct ext4_ext_path *path = *ppath;
3179 	ext4_fsblk_t newblock;
3180 	ext4_lblk_t ee_block;
3181 	struct ext4_extent *ex, newex, orig_ex, zero_ex;
3182 	struct ext4_extent *ex2 = NULL;
3183 	unsigned int ee_len, depth;
3184 	int err = 0;
3185 
3186 	BUG_ON((split_flag & (EXT4_EXT_DATA_VALID1 | EXT4_EXT_DATA_VALID2)) ==
3187 	       (EXT4_EXT_DATA_VALID1 | EXT4_EXT_DATA_VALID2));
3188 
3189 	ext_debug(inode, "logical block %llu\n", (unsigned long long)split);
3190 
3191 	ext4_ext_show_leaf(inode, path);
3192 
3193 	depth = ext_depth(inode);
3194 	ex = path[depth].p_ext;
3195 	ee_block = le32_to_cpu(ex->ee_block);
3196 	ee_len = ext4_ext_get_actual_len(ex);
3197 	newblock = split - ee_block + ext4_ext_pblock(ex);
3198 
3199 	BUG_ON(split < ee_block || split >= (ee_block + ee_len));
3200 	BUG_ON(!ext4_ext_is_unwritten(ex) &&
3201 	       split_flag & (EXT4_EXT_MAY_ZEROOUT |
3202 			     EXT4_EXT_MARK_UNWRIT1 |
3203 			     EXT4_EXT_MARK_UNWRIT2));
3204 
3205 	err = ext4_ext_get_access(handle, inode, path + depth);
3206 	if (err)
3207 		goto out;
3208 
3209 	if (split == ee_block) {
3210 		/*
3211 		 * case b: block @split is the block that the extent begins with
3212 		 * then we just change the state of the extent, and splitting
3213 		 * is not needed.
3214 		 */
3215 		if (split_flag & EXT4_EXT_MARK_UNWRIT2)
3216 			ext4_ext_mark_unwritten(ex);
3217 		else
3218 			ext4_ext_mark_initialized(ex);
3219 
3220 		if (!(flags & EXT4_GET_BLOCKS_PRE_IO))
3221 			ext4_ext_try_to_merge(handle, inode, path, ex);
3222 
3223 		err = ext4_ext_dirty(handle, inode, path + path->p_depth);
3224 		goto out;
3225 	}
3226 
3227 	/* case a */
3228 	memcpy(&orig_ex, ex, sizeof(orig_ex));
3229 	ex->ee_len = cpu_to_le16(split - ee_block);
3230 	if (split_flag & EXT4_EXT_MARK_UNWRIT1)
3231 		ext4_ext_mark_unwritten(ex);
3232 
3233 	/*
3234 	 * path may lead to new leaf, not to original leaf any more
3235 	 * after ext4_ext_insert_extent() returns,
3236 	 */
3237 	err = ext4_ext_dirty(handle, inode, path + depth);
3238 	if (err)
3239 		goto fix_extent_len;
3240 
3241 	ex2 = &newex;
3242 	ex2->ee_block = cpu_to_le32(split);
3243 	ex2->ee_len   = cpu_to_le16(ee_len - (split - ee_block));
3244 	ext4_ext_store_pblock(ex2, newblock);
3245 	if (split_flag & EXT4_EXT_MARK_UNWRIT2)
3246 		ext4_ext_mark_unwritten(ex2);
3247 
3248 	err = ext4_ext_insert_extent(handle, inode, ppath, &newex, flags);
3249 	if (err != -ENOSPC && err != -EDQUOT)
3250 		goto out;
3251 
3252 	if (EXT4_EXT_MAY_ZEROOUT & split_flag) {
3253 		if (split_flag & (EXT4_EXT_DATA_VALID1|EXT4_EXT_DATA_VALID2)) {
3254 			if (split_flag & EXT4_EXT_DATA_VALID1) {
3255 				err = ext4_ext_zeroout(inode, ex2);
3256 				zero_ex.ee_block = ex2->ee_block;
3257 				zero_ex.ee_len = cpu_to_le16(
3258 						ext4_ext_get_actual_len(ex2));
3259 				ext4_ext_store_pblock(&zero_ex,
3260 						      ext4_ext_pblock(ex2));
3261 			} else {
3262 				err = ext4_ext_zeroout(inode, ex);
3263 				zero_ex.ee_block = ex->ee_block;
3264 				zero_ex.ee_len = cpu_to_le16(
3265 						ext4_ext_get_actual_len(ex));
3266 				ext4_ext_store_pblock(&zero_ex,
3267 						      ext4_ext_pblock(ex));
3268 			}
3269 		} else {
3270 			err = ext4_ext_zeroout(inode, &orig_ex);
3271 			zero_ex.ee_block = orig_ex.ee_block;
3272 			zero_ex.ee_len = cpu_to_le16(
3273 						ext4_ext_get_actual_len(&orig_ex));
3274 			ext4_ext_store_pblock(&zero_ex,
3275 					      ext4_ext_pblock(&orig_ex));
3276 		}
3277 
3278 		if (!err) {
3279 			/* update the extent length and mark as initialized */
3280 			ex->ee_len = cpu_to_le16(ee_len);
3281 			ext4_ext_try_to_merge(handle, inode, path, ex);
3282 			err = ext4_ext_dirty(handle, inode, path + path->p_depth);
3283 			if (!err)
3284 				/* update extent status tree */
3285 				err = ext4_zeroout_es(inode, &zero_ex);
3286 			/* If we failed at this point, we don't know in which
3287 			 * state the extent tree exactly is so don't try to fix
3288 			 * length of the original extent as it may do even more
3289 			 * damage.
3290 			 */
3291 			goto out;
3292 		}
3293 	}
3294 
3295 fix_extent_len:
3296 	ex->ee_len = orig_ex.ee_len;
3297 	/*
3298 	 * Ignore ext4_ext_dirty return value since we are already in error path
3299 	 * and err is a non-zero error code.
3300 	 */
3301 	ext4_ext_dirty(handle, inode, path + path->p_depth);
3302 	return err;
3303 out:
3304 	ext4_ext_show_leaf(inode, path);
3305 	return err;
3306 }
3307 
3308 /*
3309  * ext4_split_extents() splits an extent and mark extent which is covered
3310  * by @map as split_flags indicates
3311  *
3312  * It may result in splitting the extent into multiple extents (up to three)
3313  * There are three possibilities:
3314  *   a> There is no split required
3315  *   b> Splits in two extents: Split is happening at either end of the extent
3316  *   c> Splits in three extents: Somone is splitting in middle of the extent
3317  *
3318  */
3319 static int ext4_split_extent(handle_t *handle,
3320 			      struct inode *inode,
3321 			      struct ext4_ext_path **ppath,
3322 			      struct ext4_map_blocks *map,
3323 			      int split_flag,
3324 			      int flags)
3325 {
3326 	struct ext4_ext_path *path = *ppath;
3327 	ext4_lblk_t ee_block;
3328 	struct ext4_extent *ex;
3329 	unsigned int ee_len, depth;
3330 	int err = 0;
3331 	int unwritten;
3332 	int split_flag1, flags1;
3333 	int allocated = map->m_len;
3334 
3335 	depth = ext_depth(inode);
3336 	ex = path[depth].p_ext;
3337 	ee_block = le32_to_cpu(ex->ee_block);
3338 	ee_len = ext4_ext_get_actual_len(ex);
3339 	unwritten = ext4_ext_is_unwritten(ex);
3340 
3341 	if (map->m_lblk + map->m_len < ee_block + ee_len) {
3342 		split_flag1 = split_flag & EXT4_EXT_MAY_ZEROOUT;
3343 		flags1 = flags | EXT4_GET_BLOCKS_PRE_IO;
3344 		if (unwritten)
3345 			split_flag1 |= EXT4_EXT_MARK_UNWRIT1 |
3346 				       EXT4_EXT_MARK_UNWRIT2;
3347 		if (split_flag & EXT4_EXT_DATA_VALID2)
3348 			split_flag1 |= EXT4_EXT_DATA_VALID1;
3349 		err = ext4_split_extent_at(handle, inode, ppath,
3350 				map->m_lblk + map->m_len, split_flag1, flags1);
3351 		if (err)
3352 			goto out;
3353 	} else {
3354 		allocated = ee_len - (map->m_lblk - ee_block);
3355 	}
3356 	/*
3357 	 * Update path is required because previous ext4_split_extent_at() may
3358 	 * result in split of original leaf or extent zeroout.
3359 	 */
3360 	path = ext4_find_extent(inode, map->m_lblk, ppath, flags);
3361 	if (IS_ERR(path))
3362 		return PTR_ERR(path);
3363 	depth = ext_depth(inode);
3364 	ex = path[depth].p_ext;
3365 	if (!ex) {
3366 		EXT4_ERROR_INODE(inode, "unexpected hole at %lu",
3367 				 (unsigned long) map->m_lblk);
3368 		return -EFSCORRUPTED;
3369 	}
3370 	unwritten = ext4_ext_is_unwritten(ex);
3371 	split_flag1 = 0;
3372 
3373 	if (map->m_lblk >= ee_block) {
3374 		split_flag1 = split_flag & EXT4_EXT_DATA_VALID2;
3375 		if (unwritten) {
3376 			split_flag1 |= EXT4_EXT_MARK_UNWRIT1;
3377 			split_flag1 |= split_flag & (EXT4_EXT_MAY_ZEROOUT |
3378 						     EXT4_EXT_MARK_UNWRIT2);
3379 		}
3380 		err = ext4_split_extent_at(handle, inode, ppath,
3381 				map->m_lblk, split_flag1, flags);
3382 		if (err)
3383 			goto out;
3384 	}
3385 
3386 	ext4_ext_show_leaf(inode, path);
3387 out:
3388 	return err ? err : allocated;
3389 }
3390 
3391 /*
3392  * This function is called by ext4_ext_map_blocks() if someone tries to write
3393  * to an unwritten extent. It may result in splitting the unwritten
3394  * extent into multiple extents (up to three - one initialized and two
3395  * unwritten).
3396  * There are three possibilities:
3397  *   a> There is no split required: Entire extent should be initialized
3398  *   b> Splits in two extents: Write is happening at either end of the extent
3399  *   c> Splits in three extents: Somone is writing in middle of the extent
3400  *
3401  * Pre-conditions:
3402  *  - The extent pointed to by 'path' is unwritten.
3403  *  - The extent pointed to by 'path' contains a superset
3404  *    of the logical span [map->m_lblk, map->m_lblk + map->m_len).
3405  *
3406  * Post-conditions on success:
3407  *  - the returned value is the number of blocks beyond map->l_lblk
3408  *    that are allocated and initialized.
3409  *    It is guaranteed to be >= map->m_len.
3410  */
3411 static int ext4_ext_convert_to_initialized(handle_t *handle,
3412 					   struct inode *inode,
3413 					   struct ext4_map_blocks *map,
3414 					   struct ext4_ext_path **ppath,
3415 					   int flags)
3416 {
3417 	struct ext4_ext_path *path = *ppath;
3418 	struct ext4_sb_info *sbi;
3419 	struct ext4_extent_header *eh;
3420 	struct ext4_map_blocks split_map;
3421 	struct ext4_extent zero_ex1, zero_ex2;
3422 	struct ext4_extent *ex, *abut_ex;
3423 	ext4_lblk_t ee_block, eof_block;
3424 	unsigned int ee_len, depth, map_len = map->m_len;
3425 	int allocated = 0, max_zeroout = 0;
3426 	int err = 0;
3427 	int split_flag = EXT4_EXT_DATA_VALID2;
3428 
3429 	ext_debug(inode, "logical block %llu, max_blocks %u\n",
3430 		  (unsigned long long)map->m_lblk, map_len);
3431 
3432 	sbi = EXT4_SB(inode->i_sb);
3433 	eof_block = (EXT4_I(inode)->i_disksize + inode->i_sb->s_blocksize - 1)
3434 			>> inode->i_sb->s_blocksize_bits;
3435 	if (eof_block < map->m_lblk + map_len)
3436 		eof_block = map->m_lblk + map_len;
3437 
3438 	depth = ext_depth(inode);
3439 	eh = path[depth].p_hdr;
3440 	ex = path[depth].p_ext;
3441 	ee_block = le32_to_cpu(ex->ee_block);
3442 	ee_len = ext4_ext_get_actual_len(ex);
3443 	zero_ex1.ee_len = 0;
3444 	zero_ex2.ee_len = 0;
3445 
3446 	trace_ext4_ext_convert_to_initialized_enter(inode, map, ex);
3447 
3448 	/* Pre-conditions */
3449 	BUG_ON(!ext4_ext_is_unwritten(ex));
3450 	BUG_ON(!in_range(map->m_lblk, ee_block, ee_len));
3451 
3452 	/*
3453 	 * Attempt to transfer newly initialized blocks from the currently
3454 	 * unwritten extent to its neighbor. This is much cheaper
3455 	 * than an insertion followed by a merge as those involve costly
3456 	 * memmove() calls. Transferring to the left is the common case in
3457 	 * steady state for workloads doing fallocate(FALLOC_FL_KEEP_SIZE)
3458 	 * followed by append writes.
3459 	 *
3460 	 * Limitations of the current logic:
3461 	 *  - L1: we do not deal with writes covering the whole extent.
3462 	 *    This would require removing the extent if the transfer
3463 	 *    is possible.
3464 	 *  - L2: we only attempt to merge with an extent stored in the
3465 	 *    same extent tree node.
3466 	 */
3467 	if ((map->m_lblk == ee_block) &&
3468 		/* See if we can merge left */
3469 		(map_len < ee_len) &&		/*L1*/
3470 		(ex > EXT_FIRST_EXTENT(eh))) {	/*L2*/
3471 		ext4_lblk_t prev_lblk;
3472 		ext4_fsblk_t prev_pblk, ee_pblk;
3473 		unsigned int prev_len;
3474 
3475 		abut_ex = ex - 1;
3476 		prev_lblk = le32_to_cpu(abut_ex->ee_block);
3477 		prev_len = ext4_ext_get_actual_len(abut_ex);
3478 		prev_pblk = ext4_ext_pblock(abut_ex);
3479 		ee_pblk = ext4_ext_pblock(ex);
3480 
3481 		/*
3482 		 * A transfer of blocks from 'ex' to 'abut_ex' is allowed
3483 		 * upon those conditions:
3484 		 * - C1: abut_ex is initialized,
3485 		 * - C2: abut_ex is logically abutting ex,
3486 		 * - C3: abut_ex is physically abutting ex,
3487 		 * - C4: abut_ex can receive the additional blocks without
3488 		 *   overflowing the (initialized) length limit.
3489 		 */
3490 		if ((!ext4_ext_is_unwritten(abut_ex)) &&		/*C1*/
3491 			((prev_lblk + prev_len) == ee_block) &&		/*C2*/
3492 			((prev_pblk + prev_len) == ee_pblk) &&		/*C3*/
3493 			(prev_len < (EXT_INIT_MAX_LEN - map_len))) {	/*C4*/
3494 			err = ext4_ext_get_access(handle, inode, path + depth);
3495 			if (err)
3496 				goto out;
3497 
3498 			trace_ext4_ext_convert_to_initialized_fastpath(inode,
3499 				map, ex, abut_ex);
3500 
3501 			/* Shift the start of ex by 'map_len' blocks */
3502 			ex->ee_block = cpu_to_le32(ee_block + map_len);
3503 			ext4_ext_store_pblock(ex, ee_pblk + map_len);
3504 			ex->ee_len = cpu_to_le16(ee_len - map_len);
3505 			ext4_ext_mark_unwritten(ex); /* Restore the flag */
3506 
3507 			/* Extend abut_ex by 'map_len' blocks */
3508 			abut_ex->ee_len = cpu_to_le16(prev_len + map_len);
3509 
3510 			/* Result: number of initialized blocks past m_lblk */
3511 			allocated = map_len;
3512 		}
3513 	} else if (((map->m_lblk + map_len) == (ee_block + ee_len)) &&
3514 		   (map_len < ee_len) &&	/*L1*/
3515 		   ex < EXT_LAST_EXTENT(eh)) {	/*L2*/
3516 		/* See if we can merge right */
3517 		ext4_lblk_t next_lblk;
3518 		ext4_fsblk_t next_pblk, ee_pblk;
3519 		unsigned int next_len;
3520 
3521 		abut_ex = ex + 1;
3522 		next_lblk = le32_to_cpu(abut_ex->ee_block);
3523 		next_len = ext4_ext_get_actual_len(abut_ex);
3524 		next_pblk = ext4_ext_pblock(abut_ex);
3525 		ee_pblk = ext4_ext_pblock(ex);
3526 
3527 		/*
3528 		 * A transfer of blocks from 'ex' to 'abut_ex' is allowed
3529 		 * upon those conditions:
3530 		 * - C1: abut_ex is initialized,
3531 		 * - C2: abut_ex is logically abutting ex,
3532 		 * - C3: abut_ex is physically abutting ex,
3533 		 * - C4: abut_ex can receive the additional blocks without
3534 		 *   overflowing the (initialized) length limit.
3535 		 */
3536 		if ((!ext4_ext_is_unwritten(abut_ex)) &&		/*C1*/
3537 		    ((map->m_lblk + map_len) == next_lblk) &&		/*C2*/
3538 		    ((ee_pblk + ee_len) == next_pblk) &&		/*C3*/
3539 		    (next_len < (EXT_INIT_MAX_LEN - map_len))) {	/*C4*/
3540 			err = ext4_ext_get_access(handle, inode, path + depth);
3541 			if (err)
3542 				goto out;
3543 
3544 			trace_ext4_ext_convert_to_initialized_fastpath(inode,
3545 				map, ex, abut_ex);
3546 
3547 			/* Shift the start of abut_ex by 'map_len' blocks */
3548 			abut_ex->ee_block = cpu_to_le32(next_lblk - map_len);
3549 			ext4_ext_store_pblock(abut_ex, next_pblk - map_len);
3550 			ex->ee_len = cpu_to_le16(ee_len - map_len);
3551 			ext4_ext_mark_unwritten(ex); /* Restore the flag */
3552 
3553 			/* Extend abut_ex by 'map_len' blocks */
3554 			abut_ex->ee_len = cpu_to_le16(next_len + map_len);
3555 
3556 			/* Result: number of initialized blocks past m_lblk */
3557 			allocated = map_len;
3558 		}
3559 	}
3560 	if (allocated) {
3561 		/* Mark the block containing both extents as dirty */
3562 		err = ext4_ext_dirty(handle, inode, path + depth);
3563 
3564 		/* Update path to point to the right extent */
3565 		path[depth].p_ext = abut_ex;
3566 		goto out;
3567 	} else
3568 		allocated = ee_len - (map->m_lblk - ee_block);
3569 
3570 	WARN_ON(map->m_lblk < ee_block);
3571 	/*
3572 	 * It is safe to convert extent to initialized via explicit
3573 	 * zeroout only if extent is fully inside i_size or new_size.
3574 	 */
3575 	split_flag |= ee_block + ee_len <= eof_block ? EXT4_EXT_MAY_ZEROOUT : 0;
3576 
3577 	if (EXT4_EXT_MAY_ZEROOUT & split_flag)
3578 		max_zeroout = sbi->s_extent_max_zeroout_kb >>
3579 			(inode->i_sb->s_blocksize_bits - 10);
3580 
3581 	/*
3582 	 * five cases:
3583 	 * 1. split the extent into three extents.
3584 	 * 2. split the extent into two extents, zeroout the head of the first
3585 	 *    extent.
3586 	 * 3. split the extent into two extents, zeroout the tail of the second
3587 	 *    extent.
3588 	 * 4. split the extent into two extents with out zeroout.
3589 	 * 5. no splitting needed, just possibly zeroout the head and / or the
3590 	 *    tail of the extent.
3591 	 */
3592 	split_map.m_lblk = map->m_lblk;
3593 	split_map.m_len = map->m_len;
3594 
3595 	if (max_zeroout && (allocated > split_map.m_len)) {
3596 		if (allocated <= max_zeroout) {
3597 			/* case 3 or 5 */
3598 			zero_ex1.ee_block =
3599 				 cpu_to_le32(split_map.m_lblk +
3600 					     split_map.m_len);
3601 			zero_ex1.ee_len =
3602 				cpu_to_le16(allocated - split_map.m_len);
3603 			ext4_ext_store_pblock(&zero_ex1,
3604 				ext4_ext_pblock(ex) + split_map.m_lblk +
3605 				split_map.m_len - ee_block);
3606 			err = ext4_ext_zeroout(inode, &zero_ex1);
3607 			if (err)
3608 				goto fallback;
3609 			split_map.m_len = allocated;
3610 		}
3611 		if (split_map.m_lblk - ee_block + split_map.m_len <
3612 								max_zeroout) {
3613 			/* case 2 or 5 */
3614 			if (split_map.m_lblk != ee_block) {
3615 				zero_ex2.ee_block = ex->ee_block;
3616 				zero_ex2.ee_len = cpu_to_le16(split_map.m_lblk -
3617 							ee_block);
3618 				ext4_ext_store_pblock(&zero_ex2,
3619 						      ext4_ext_pblock(ex));
3620 				err = ext4_ext_zeroout(inode, &zero_ex2);
3621 				if (err)
3622 					goto fallback;
3623 			}
3624 
3625 			split_map.m_len += split_map.m_lblk - ee_block;
3626 			split_map.m_lblk = ee_block;
3627 			allocated = map->m_len;
3628 		}
3629 	}
3630 
3631 fallback:
3632 	err = ext4_split_extent(handle, inode, ppath, &split_map, split_flag,
3633 				flags);
3634 	if (err > 0)
3635 		err = 0;
3636 out:
3637 	/* If we have gotten a failure, don't zero out status tree */
3638 	if (!err) {
3639 		err = ext4_zeroout_es(inode, &zero_ex1);
3640 		if (!err)
3641 			err = ext4_zeroout_es(inode, &zero_ex2);
3642 	}
3643 	return err ? err : allocated;
3644 }
3645 
3646 /*
3647  * This function is called by ext4_ext_map_blocks() from
3648  * ext4_get_blocks_dio_write() when DIO to write
3649  * to an unwritten extent.
3650  *
3651  * Writing to an unwritten extent may result in splitting the unwritten
3652  * extent into multiple initialized/unwritten extents (up to three)
3653  * There are three possibilities:
3654  *   a> There is no split required: Entire extent should be unwritten
3655  *   b> Splits in two extents: Write is happening at either end of the extent
3656  *   c> Splits in three extents: Somone is writing in middle of the extent
3657  *
3658  * This works the same way in the case of initialized -> unwritten conversion.
3659  *
3660  * One of more index blocks maybe needed if the extent tree grow after
3661  * the unwritten extent split. To prevent ENOSPC occur at the IO
3662  * complete, we need to split the unwritten extent before DIO submit
3663  * the IO. The unwritten extent called at this time will be split
3664  * into three unwritten extent(at most). After IO complete, the part
3665  * being filled will be convert to initialized by the end_io callback function
3666  * via ext4_convert_unwritten_extents().
3667  *
3668  * Returns the size of unwritten extent to be written on success.
3669  */
3670 static int ext4_split_convert_extents(handle_t *handle,
3671 					struct inode *inode,
3672 					struct ext4_map_blocks *map,
3673 					struct ext4_ext_path **ppath,
3674 					int flags)
3675 {
3676 	struct ext4_ext_path *path = *ppath;
3677 	ext4_lblk_t eof_block;
3678 	ext4_lblk_t ee_block;
3679 	struct ext4_extent *ex;
3680 	unsigned int ee_len;
3681 	int split_flag = 0, depth;
3682 
3683 	ext_debug(inode, "logical block %llu, max_blocks %u\n",
3684 		  (unsigned long long)map->m_lblk, map->m_len);
3685 
3686 	eof_block = (EXT4_I(inode)->i_disksize + inode->i_sb->s_blocksize - 1)
3687 			>> inode->i_sb->s_blocksize_bits;
3688 	if (eof_block < map->m_lblk + map->m_len)
3689 		eof_block = map->m_lblk + map->m_len;
3690 	/*
3691 	 * It is safe to convert extent to initialized via explicit
3692 	 * zeroout only if extent is fully inside i_size or new_size.
3693 	 */
3694 	depth = ext_depth(inode);
3695 	ex = path[depth].p_ext;
3696 	ee_block = le32_to_cpu(ex->ee_block);
3697 	ee_len = ext4_ext_get_actual_len(ex);
3698 
3699 	/* Convert to unwritten */
3700 	if (flags & EXT4_GET_BLOCKS_CONVERT_UNWRITTEN) {
3701 		split_flag |= EXT4_EXT_DATA_VALID1;
3702 	/* Convert to initialized */
3703 	} else if (flags & EXT4_GET_BLOCKS_CONVERT) {
3704 		split_flag |= ee_block + ee_len <= eof_block ?
3705 			      EXT4_EXT_MAY_ZEROOUT : 0;
3706 		split_flag |= (EXT4_EXT_MARK_UNWRIT2 | EXT4_EXT_DATA_VALID2);
3707 	}
3708 	flags |= EXT4_GET_BLOCKS_PRE_IO;
3709 	return ext4_split_extent(handle, inode, ppath, map, split_flag, flags);
3710 }
3711 
3712 static int ext4_convert_unwritten_extents_endio(handle_t *handle,
3713 						struct inode *inode,
3714 						struct ext4_map_blocks *map,
3715 						struct ext4_ext_path **ppath)
3716 {
3717 	struct ext4_ext_path *path = *ppath;
3718 	struct ext4_extent *ex;
3719 	ext4_lblk_t ee_block;
3720 	unsigned int ee_len;
3721 	int depth;
3722 	int err = 0;
3723 
3724 	depth = ext_depth(inode);
3725 	ex = path[depth].p_ext;
3726 	ee_block = le32_to_cpu(ex->ee_block);
3727 	ee_len = ext4_ext_get_actual_len(ex);
3728 
3729 	ext_debug(inode, "logical block %llu, max_blocks %u\n",
3730 		  (unsigned long long)ee_block, ee_len);
3731 
3732 	/* If extent is larger than requested it is a clear sign that we still
3733 	 * have some extent state machine issues left. So extent_split is still
3734 	 * required.
3735 	 * TODO: Once all related issues will be fixed this situation should be
3736 	 * illegal.
3737 	 */
3738 	if (ee_block != map->m_lblk || ee_len > map->m_len) {
3739 #ifdef CONFIG_EXT4_DEBUG
3740 		ext4_warning(inode->i_sb, "Inode (%ld) finished: extent logical block %llu,"
3741 			     " len %u; IO logical block %llu, len %u",
3742 			     inode->i_ino, (unsigned long long)ee_block, ee_len,
3743 			     (unsigned long long)map->m_lblk, map->m_len);
3744 #endif
3745 		err = ext4_split_convert_extents(handle, inode, map, ppath,
3746 						 EXT4_GET_BLOCKS_CONVERT);
3747 		if (err < 0)
3748 			return err;
3749 		path = ext4_find_extent(inode, map->m_lblk, ppath, 0);
3750 		if (IS_ERR(path))
3751 			return PTR_ERR(path);
3752 		depth = ext_depth(inode);
3753 		ex = path[depth].p_ext;
3754 	}
3755 
3756 	err = ext4_ext_get_access(handle, inode, path + depth);
3757 	if (err)
3758 		goto out;
3759 	/* first mark the extent as initialized */
3760 	ext4_ext_mark_initialized(ex);
3761 
3762 	/* note: ext4_ext_correct_indexes() isn't needed here because
3763 	 * borders are not changed
3764 	 */
3765 	ext4_ext_try_to_merge(handle, inode, path, ex);
3766 
3767 	/* Mark modified extent as dirty */
3768 	err = ext4_ext_dirty(handle, inode, path + path->p_depth);
3769 out:
3770 	ext4_ext_show_leaf(inode, path);
3771 	return err;
3772 }
3773 
3774 static int
3775 convert_initialized_extent(handle_t *handle, struct inode *inode,
3776 			   struct ext4_map_blocks *map,
3777 			   struct ext4_ext_path **ppath,
3778 			   unsigned int *allocated)
3779 {
3780 	struct ext4_ext_path *path = *ppath;
3781 	struct ext4_extent *ex;
3782 	ext4_lblk_t ee_block;
3783 	unsigned int ee_len;
3784 	int depth;
3785 	int err = 0;
3786 
3787 	/*
3788 	 * Make sure that the extent is no bigger than we support with
3789 	 * unwritten extent
3790 	 */
3791 	if (map->m_len > EXT_UNWRITTEN_MAX_LEN)
3792 		map->m_len = EXT_UNWRITTEN_MAX_LEN / 2;
3793 
3794 	depth = ext_depth(inode);
3795 	ex = path[depth].p_ext;
3796 	ee_block = le32_to_cpu(ex->ee_block);
3797 	ee_len = ext4_ext_get_actual_len(ex);
3798 
3799 	ext_debug(inode, "logical block %llu, max_blocks %u\n",
3800 		  (unsigned long long)ee_block, ee_len);
3801 
3802 	if (ee_block != map->m_lblk || ee_len > map->m_len) {
3803 		err = ext4_split_convert_extents(handle, inode, map, ppath,
3804 				EXT4_GET_BLOCKS_CONVERT_UNWRITTEN);
3805 		if (err < 0)
3806 			return err;
3807 		path = ext4_find_extent(inode, map->m_lblk, ppath, 0);
3808 		if (IS_ERR(path))
3809 			return PTR_ERR(path);
3810 		depth = ext_depth(inode);
3811 		ex = path[depth].p_ext;
3812 		if (!ex) {
3813 			EXT4_ERROR_INODE(inode, "unexpected hole at %lu",
3814 					 (unsigned long) map->m_lblk);
3815 			return -EFSCORRUPTED;
3816 		}
3817 	}
3818 
3819 	err = ext4_ext_get_access(handle, inode, path + depth);
3820 	if (err)
3821 		return err;
3822 	/* first mark the extent as unwritten */
3823 	ext4_ext_mark_unwritten(ex);
3824 
3825 	/* note: ext4_ext_correct_indexes() isn't needed here because
3826 	 * borders are not changed
3827 	 */
3828 	ext4_ext_try_to_merge(handle, inode, path, ex);
3829 
3830 	/* Mark modified extent as dirty */
3831 	err = ext4_ext_dirty(handle, inode, path + path->p_depth);
3832 	if (err)
3833 		return err;
3834 	ext4_ext_show_leaf(inode, path);
3835 
3836 	ext4_update_inode_fsync_trans(handle, inode, 1);
3837 
3838 	map->m_flags |= EXT4_MAP_UNWRITTEN;
3839 	if (*allocated > map->m_len)
3840 		*allocated = map->m_len;
3841 	map->m_len = *allocated;
3842 	return 0;
3843 }
3844 
3845 static int
3846 ext4_ext_handle_unwritten_extents(handle_t *handle, struct inode *inode,
3847 			struct ext4_map_blocks *map,
3848 			struct ext4_ext_path **ppath, int flags,
3849 			unsigned int allocated, ext4_fsblk_t newblock)
3850 {
3851 	struct ext4_ext_path __maybe_unused *path = *ppath;
3852 	int ret = 0;
3853 	int err = 0;
3854 
3855 	ext_debug(inode, "logical block %llu, max_blocks %u, flags 0x%x, allocated %u\n",
3856 		  (unsigned long long)map->m_lblk, map->m_len, flags,
3857 		  allocated);
3858 	ext4_ext_show_leaf(inode, path);
3859 
3860 	/*
3861 	 * When writing into unwritten space, we should not fail to
3862 	 * allocate metadata blocks for the new extent block if needed.
3863 	 */
3864 	flags |= EXT4_GET_BLOCKS_METADATA_NOFAIL;
3865 
3866 	trace_ext4_ext_handle_unwritten_extents(inode, map, flags,
3867 						    allocated, newblock);
3868 
3869 	/* get_block() before submitting IO, split the extent */
3870 	if (flags & EXT4_GET_BLOCKS_PRE_IO) {
3871 		ret = ext4_split_convert_extents(handle, inode, map, ppath,
3872 					 flags | EXT4_GET_BLOCKS_CONVERT);
3873 		if (ret < 0) {
3874 			err = ret;
3875 			goto out2;
3876 		}
3877 		/*
3878 		 * shouldn't get a 0 return when splitting an extent unless
3879 		 * m_len is 0 (bug) or extent has been corrupted
3880 		 */
3881 		if (unlikely(ret == 0)) {
3882 			EXT4_ERROR_INODE(inode,
3883 					 "unexpected ret == 0, m_len = %u",
3884 					 map->m_len);
3885 			err = -EFSCORRUPTED;
3886 			goto out2;
3887 		}
3888 		map->m_flags |= EXT4_MAP_UNWRITTEN;
3889 		goto out;
3890 	}
3891 	/* IO end_io complete, convert the filled extent to written */
3892 	if (flags & EXT4_GET_BLOCKS_CONVERT) {
3893 		err = ext4_convert_unwritten_extents_endio(handle, inode, map,
3894 							   ppath);
3895 		if (err < 0)
3896 			goto out2;
3897 		ext4_update_inode_fsync_trans(handle, inode, 1);
3898 		goto map_out;
3899 	}
3900 	/* buffered IO cases */
3901 	/*
3902 	 * repeat fallocate creation request
3903 	 * we already have an unwritten extent
3904 	 */
3905 	if (flags & EXT4_GET_BLOCKS_UNWRIT_EXT) {
3906 		map->m_flags |= EXT4_MAP_UNWRITTEN;
3907 		goto map_out;
3908 	}
3909 
3910 	/* buffered READ or buffered write_begin() lookup */
3911 	if ((flags & EXT4_GET_BLOCKS_CREATE) == 0) {
3912 		/*
3913 		 * We have blocks reserved already.  We
3914 		 * return allocated blocks so that delalloc
3915 		 * won't do block reservation for us.  But
3916 		 * the buffer head will be unmapped so that
3917 		 * a read from the block returns 0s.
3918 		 */
3919 		map->m_flags |= EXT4_MAP_UNWRITTEN;
3920 		goto out1;
3921 	}
3922 
3923 	/*
3924 	 * Default case when (flags & EXT4_GET_BLOCKS_CREATE) == 1.
3925 	 * For buffered writes, at writepage time, etc.  Convert a
3926 	 * discovered unwritten extent to written.
3927 	 */
3928 	ret = ext4_ext_convert_to_initialized(handle, inode, map, ppath, flags);
3929 	if (ret < 0) {
3930 		err = ret;
3931 		goto out2;
3932 	}
3933 	ext4_update_inode_fsync_trans(handle, inode, 1);
3934 	/*
3935 	 * shouldn't get a 0 return when converting an unwritten extent
3936 	 * unless m_len is 0 (bug) or extent has been corrupted
3937 	 */
3938 	if (unlikely(ret == 0)) {
3939 		EXT4_ERROR_INODE(inode, "unexpected ret == 0, m_len = %u",
3940 				 map->m_len);
3941 		err = -EFSCORRUPTED;
3942 		goto out2;
3943 	}
3944 
3945 out:
3946 	allocated = ret;
3947 	map->m_flags |= EXT4_MAP_NEW;
3948 map_out:
3949 	map->m_flags |= EXT4_MAP_MAPPED;
3950 out1:
3951 	map->m_pblk = newblock;
3952 	if (allocated > map->m_len)
3953 		allocated = map->m_len;
3954 	map->m_len = allocated;
3955 	ext4_ext_show_leaf(inode, path);
3956 out2:
3957 	return err ? err : allocated;
3958 }
3959 
3960 /*
3961  * get_implied_cluster_alloc - check to see if the requested
3962  * allocation (in the map structure) overlaps with a cluster already
3963  * allocated in an extent.
3964  *	@sb	The filesystem superblock structure
3965  *	@map	The requested lblk->pblk mapping
3966  *	@ex	The extent structure which might contain an implied
3967  *			cluster allocation
3968  *
3969  * This function is called by ext4_ext_map_blocks() after we failed to
3970  * find blocks that were already in the inode's extent tree.  Hence,
3971  * we know that the beginning of the requested region cannot overlap
3972  * the extent from the inode's extent tree.  There are three cases we
3973  * want to catch.  The first is this case:
3974  *
3975  *		 |--- cluster # N--|
3976  *    |--- extent ---|	|---- requested region ---|
3977  *			|==========|
3978  *
3979  * The second case that we need to test for is this one:
3980  *
3981  *   |--------- cluster # N ----------------|
3982  *	   |--- requested region --|   |------- extent ----|
3983  *	   |=======================|
3984  *
3985  * The third case is when the requested region lies between two extents
3986  * within the same cluster:
3987  *          |------------- cluster # N-------------|
3988  * |----- ex -----|                  |---- ex_right ----|
3989  *                  |------ requested region ------|
3990  *                  |================|
3991  *
3992  * In each of the above cases, we need to set the map->m_pblk and
3993  * map->m_len so it corresponds to the return the extent labelled as
3994  * "|====|" from cluster #N, since it is already in use for data in
3995  * cluster EXT4_B2C(sbi, map->m_lblk).	We will then return 1 to
3996  * signal to ext4_ext_map_blocks() that map->m_pblk should be treated
3997  * as a new "allocated" block region.  Otherwise, we will return 0 and
3998  * ext4_ext_map_blocks() will then allocate one or more new clusters
3999  * by calling ext4_mb_new_blocks().
4000  */
4001 static int get_implied_cluster_alloc(struct super_block *sb,
4002 				     struct ext4_map_blocks *map,
4003 				     struct ext4_extent *ex,
4004 				     struct ext4_ext_path *path)
4005 {
4006 	struct ext4_sb_info *sbi = EXT4_SB(sb);
4007 	ext4_lblk_t c_offset = EXT4_LBLK_COFF(sbi, map->m_lblk);
4008 	ext4_lblk_t ex_cluster_start, ex_cluster_end;
4009 	ext4_lblk_t rr_cluster_start;
4010 	ext4_lblk_t ee_block = le32_to_cpu(ex->ee_block);
4011 	ext4_fsblk_t ee_start = ext4_ext_pblock(ex);
4012 	unsigned short ee_len = ext4_ext_get_actual_len(ex);
4013 
4014 	/* The extent passed in that we are trying to match */
4015 	ex_cluster_start = EXT4_B2C(sbi, ee_block);
4016 	ex_cluster_end = EXT4_B2C(sbi, ee_block + ee_len - 1);
4017 
4018 	/* The requested region passed into ext4_map_blocks() */
4019 	rr_cluster_start = EXT4_B2C(sbi, map->m_lblk);
4020 
4021 	if ((rr_cluster_start == ex_cluster_end) ||
4022 	    (rr_cluster_start == ex_cluster_start)) {
4023 		if (rr_cluster_start == ex_cluster_end)
4024 			ee_start += ee_len - 1;
4025 		map->m_pblk = EXT4_PBLK_CMASK(sbi, ee_start) + c_offset;
4026 		map->m_len = min(map->m_len,
4027 				 (unsigned) sbi->s_cluster_ratio - c_offset);
4028 		/*
4029 		 * Check for and handle this case:
4030 		 *
4031 		 *   |--------- cluster # N-------------|
4032 		 *		       |------- extent ----|
4033 		 *	   |--- requested region ---|
4034 		 *	   |===========|
4035 		 */
4036 
4037 		if (map->m_lblk < ee_block)
4038 			map->m_len = min(map->m_len, ee_block - map->m_lblk);
4039 
4040 		/*
4041 		 * Check for the case where there is already another allocated
4042 		 * block to the right of 'ex' but before the end of the cluster.
4043 		 *
4044 		 *          |------------- cluster # N-------------|
4045 		 * |----- ex -----|                  |---- ex_right ----|
4046 		 *                  |------ requested region ------|
4047 		 *                  |================|
4048 		 */
4049 		if (map->m_lblk > ee_block) {
4050 			ext4_lblk_t next = ext4_ext_next_allocated_block(path);
4051 			map->m_len = min(map->m_len, next - map->m_lblk);
4052 		}
4053 
4054 		trace_ext4_get_implied_cluster_alloc_exit(sb, map, 1);
4055 		return 1;
4056 	}
4057 
4058 	trace_ext4_get_implied_cluster_alloc_exit(sb, map, 0);
4059 	return 0;
4060 }
4061 
4062 
4063 /*
4064  * Block allocation/map/preallocation routine for extents based files
4065  *
4066  *
4067  * Need to be called with
4068  * down_read(&EXT4_I(inode)->i_data_sem) if not allocating file system block
4069  * (ie, create is zero). Otherwise down_write(&EXT4_I(inode)->i_data_sem)
4070  *
4071  * return > 0, number of blocks already mapped/allocated
4072  *          if create == 0 and these are pre-allocated blocks
4073  *          	buffer head is unmapped
4074  *          otherwise blocks are mapped
4075  *
4076  * return = 0, if plain look up failed (blocks have not been allocated)
4077  *          buffer head is unmapped
4078  *
4079  * return < 0, error case.
4080  */
4081 int ext4_ext_map_blocks(handle_t *handle, struct inode *inode,
4082 			struct ext4_map_blocks *map, int flags)
4083 {
4084 	struct ext4_ext_path *path = NULL;
4085 	struct ext4_extent newex, *ex, ex2;
4086 	struct ext4_sb_info *sbi = EXT4_SB(inode->i_sb);
4087 	ext4_fsblk_t newblock = 0, pblk;
4088 	int err = 0, depth, ret;
4089 	unsigned int allocated = 0, offset = 0;
4090 	unsigned int allocated_clusters = 0;
4091 	struct ext4_allocation_request ar;
4092 	ext4_lblk_t cluster_offset;
4093 
4094 	ext_debug(inode, "blocks %u/%u requested\n", map->m_lblk, map->m_len);
4095 	trace_ext4_ext_map_blocks_enter(inode, map->m_lblk, map->m_len, flags);
4096 
4097 	/* find extent for this block */
4098 	path = ext4_find_extent(inode, map->m_lblk, NULL, 0);
4099 	if (IS_ERR(path)) {
4100 		err = PTR_ERR(path);
4101 		path = NULL;
4102 		goto out;
4103 	}
4104 
4105 	depth = ext_depth(inode);
4106 
4107 	/*
4108 	 * consistent leaf must not be empty;
4109 	 * this situation is possible, though, _during_ tree modification;
4110 	 * this is why assert can't be put in ext4_find_extent()
4111 	 */
4112 	if (unlikely(path[depth].p_ext == NULL && depth != 0)) {
4113 		EXT4_ERROR_INODE(inode, "bad extent address "
4114 				 "lblock: %lu, depth: %d pblock %lld",
4115 				 (unsigned long) map->m_lblk, depth,
4116 				 path[depth].p_block);
4117 		err = -EFSCORRUPTED;
4118 		goto out;
4119 	}
4120 
4121 	ex = path[depth].p_ext;
4122 	if (ex) {
4123 		ext4_lblk_t ee_block = le32_to_cpu(ex->ee_block);
4124 		ext4_fsblk_t ee_start = ext4_ext_pblock(ex);
4125 		unsigned short ee_len;
4126 
4127 
4128 		/*
4129 		 * unwritten extents are treated as holes, except that
4130 		 * we split out initialized portions during a write.
4131 		 */
4132 		ee_len = ext4_ext_get_actual_len(ex);
4133 
4134 		trace_ext4_ext_show_extent(inode, ee_block, ee_start, ee_len);
4135 
4136 		/* if found extent covers block, simply return it */
4137 		if (in_range(map->m_lblk, ee_block, ee_len)) {
4138 			newblock = map->m_lblk - ee_block + ee_start;
4139 			/* number of remaining blocks in the extent */
4140 			allocated = ee_len - (map->m_lblk - ee_block);
4141 			ext_debug(inode, "%u fit into %u:%d -> %llu\n",
4142 				  map->m_lblk, ee_block, ee_len, newblock);
4143 
4144 			/*
4145 			 * If the extent is initialized check whether the
4146 			 * caller wants to convert it to unwritten.
4147 			 */
4148 			if ((!ext4_ext_is_unwritten(ex)) &&
4149 			    (flags & EXT4_GET_BLOCKS_CONVERT_UNWRITTEN)) {
4150 				err = convert_initialized_extent(handle,
4151 					inode, map, &path, &allocated);
4152 				goto out;
4153 			} else if (!ext4_ext_is_unwritten(ex)) {
4154 				map->m_flags |= EXT4_MAP_MAPPED;
4155 				map->m_pblk = newblock;
4156 				if (allocated > map->m_len)
4157 					allocated = map->m_len;
4158 				map->m_len = allocated;
4159 				ext4_ext_show_leaf(inode, path);
4160 				goto out;
4161 			}
4162 
4163 			ret = ext4_ext_handle_unwritten_extents(
4164 				handle, inode, map, &path, flags,
4165 				allocated, newblock);
4166 			if (ret < 0)
4167 				err = ret;
4168 			else
4169 				allocated = ret;
4170 			goto out;
4171 		}
4172 	}
4173 
4174 	/*
4175 	 * requested block isn't allocated yet;
4176 	 * we couldn't try to create block if create flag is zero
4177 	 */
4178 	if ((flags & EXT4_GET_BLOCKS_CREATE) == 0) {
4179 		ext4_lblk_t hole_start, hole_len;
4180 
4181 		hole_start = map->m_lblk;
4182 		hole_len = ext4_ext_determine_hole(inode, path, &hole_start);
4183 		/*
4184 		 * put just found gap into cache to speed up
4185 		 * subsequent requests
4186 		 */
4187 		ext4_ext_put_gap_in_cache(inode, hole_start, hole_len);
4188 
4189 		/* Update hole_len to reflect hole size after map->m_lblk */
4190 		if (hole_start != map->m_lblk)
4191 			hole_len -= map->m_lblk - hole_start;
4192 		map->m_pblk = 0;
4193 		map->m_len = min_t(unsigned int, map->m_len, hole_len);
4194 
4195 		goto out;
4196 	}
4197 
4198 	/*
4199 	 * Okay, we need to do block allocation.
4200 	 */
4201 	newex.ee_block = cpu_to_le32(map->m_lblk);
4202 	cluster_offset = EXT4_LBLK_COFF(sbi, map->m_lblk);
4203 
4204 	/*
4205 	 * If we are doing bigalloc, check to see if the extent returned
4206 	 * by ext4_find_extent() implies a cluster we can use.
4207 	 */
4208 	if (cluster_offset && ex &&
4209 	    get_implied_cluster_alloc(inode->i_sb, map, ex, path)) {
4210 		ar.len = allocated = map->m_len;
4211 		newblock = map->m_pblk;
4212 		goto got_allocated_blocks;
4213 	}
4214 
4215 	/* find neighbour allocated blocks */
4216 	ar.lleft = map->m_lblk;
4217 	err = ext4_ext_search_left(inode, path, &ar.lleft, &ar.pleft);
4218 	if (err)
4219 		goto out;
4220 	ar.lright = map->m_lblk;
4221 	err = ext4_ext_search_right(inode, path, &ar.lright, &ar.pright, &ex2);
4222 	if (err < 0)
4223 		goto out;
4224 
4225 	/* Check if the extent after searching to the right implies a
4226 	 * cluster we can use. */
4227 	if ((sbi->s_cluster_ratio > 1) && err &&
4228 	    get_implied_cluster_alloc(inode->i_sb, map, &ex2, path)) {
4229 		ar.len = allocated = map->m_len;
4230 		newblock = map->m_pblk;
4231 		goto got_allocated_blocks;
4232 	}
4233 
4234 	/*
4235 	 * See if request is beyond maximum number of blocks we can have in
4236 	 * a single extent. For an initialized extent this limit is
4237 	 * EXT_INIT_MAX_LEN and for an unwritten extent this limit is
4238 	 * EXT_UNWRITTEN_MAX_LEN.
4239 	 */
4240 	if (map->m_len > EXT_INIT_MAX_LEN &&
4241 	    !(flags & EXT4_GET_BLOCKS_UNWRIT_EXT))
4242 		map->m_len = EXT_INIT_MAX_LEN;
4243 	else if (map->m_len > EXT_UNWRITTEN_MAX_LEN &&
4244 		 (flags & EXT4_GET_BLOCKS_UNWRIT_EXT))
4245 		map->m_len = EXT_UNWRITTEN_MAX_LEN;
4246 
4247 	/* Check if we can really insert (m_lblk)::(m_lblk + m_len) extent */
4248 	newex.ee_len = cpu_to_le16(map->m_len);
4249 	err = ext4_ext_check_overlap(sbi, inode, &newex, path);
4250 	if (err)
4251 		allocated = ext4_ext_get_actual_len(&newex);
4252 	else
4253 		allocated = map->m_len;
4254 
4255 	/* allocate new block */
4256 	ar.inode = inode;
4257 	ar.goal = ext4_ext_find_goal(inode, path, map->m_lblk);
4258 	ar.logical = map->m_lblk;
4259 	/*
4260 	 * We calculate the offset from the beginning of the cluster
4261 	 * for the logical block number, since when we allocate a
4262 	 * physical cluster, the physical block should start at the
4263 	 * same offset from the beginning of the cluster.  This is
4264 	 * needed so that future calls to get_implied_cluster_alloc()
4265 	 * work correctly.
4266 	 */
4267 	offset = EXT4_LBLK_COFF(sbi, map->m_lblk);
4268 	ar.len = EXT4_NUM_B2C(sbi, offset+allocated);
4269 	ar.goal -= offset;
4270 	ar.logical -= offset;
4271 	if (S_ISREG(inode->i_mode))
4272 		ar.flags = EXT4_MB_HINT_DATA;
4273 	else
4274 		/* disable in-core preallocation for non-regular files */
4275 		ar.flags = 0;
4276 	if (flags & EXT4_GET_BLOCKS_NO_NORMALIZE)
4277 		ar.flags |= EXT4_MB_HINT_NOPREALLOC;
4278 	if (flags & EXT4_GET_BLOCKS_DELALLOC_RESERVE)
4279 		ar.flags |= EXT4_MB_DELALLOC_RESERVED;
4280 	if (flags & EXT4_GET_BLOCKS_METADATA_NOFAIL)
4281 		ar.flags |= EXT4_MB_USE_RESERVED;
4282 	newblock = ext4_mb_new_blocks(handle, &ar, &err);
4283 	if (!newblock)
4284 		goto out;
4285 	allocated_clusters = ar.len;
4286 	ar.len = EXT4_C2B(sbi, ar.len) - offset;
4287 	ext_debug(inode, "allocate new block: goal %llu, found %llu/%u, requested %u\n",
4288 		  ar.goal, newblock, ar.len, allocated);
4289 	if (ar.len > allocated)
4290 		ar.len = allocated;
4291 
4292 got_allocated_blocks:
4293 	/* try to insert new extent into found leaf and return */
4294 	pblk = newblock + offset;
4295 	ext4_ext_store_pblock(&newex, pblk);
4296 	newex.ee_len = cpu_to_le16(ar.len);
4297 	/* Mark unwritten */
4298 	if (flags & EXT4_GET_BLOCKS_UNWRIT_EXT) {
4299 		ext4_ext_mark_unwritten(&newex);
4300 		map->m_flags |= EXT4_MAP_UNWRITTEN;
4301 	}
4302 
4303 	err = ext4_ext_insert_extent(handle, inode, &path, &newex, flags);
4304 	if (err) {
4305 		if (allocated_clusters) {
4306 			int fb_flags = 0;
4307 
4308 			/*
4309 			 * free data blocks we just allocated.
4310 			 * not a good idea to call discard here directly,
4311 			 * but otherwise we'd need to call it every free().
4312 			 */
4313 			ext4_discard_preallocations(inode, 0);
4314 			if (flags & EXT4_GET_BLOCKS_DELALLOC_RESERVE)
4315 				fb_flags = EXT4_FREE_BLOCKS_NO_QUOT_UPDATE;
4316 			ext4_free_blocks(handle, inode, NULL, newblock,
4317 					 EXT4_C2B(sbi, allocated_clusters),
4318 					 fb_flags);
4319 		}
4320 		goto out;
4321 	}
4322 
4323 	/*
4324 	 * Reduce the reserved cluster count to reflect successful deferred
4325 	 * allocation of delayed allocated clusters or direct allocation of
4326 	 * clusters discovered to be delayed allocated.  Once allocated, a
4327 	 * cluster is not included in the reserved count.
4328 	 */
4329 	if (test_opt(inode->i_sb, DELALLOC) && allocated_clusters) {
4330 		if (flags & EXT4_GET_BLOCKS_DELALLOC_RESERVE) {
4331 			/*
4332 			 * When allocating delayed allocated clusters, simply
4333 			 * reduce the reserved cluster count and claim quota
4334 			 */
4335 			ext4_da_update_reserve_space(inode, allocated_clusters,
4336 							1);
4337 		} else {
4338 			ext4_lblk_t lblk, len;
4339 			unsigned int n;
4340 
4341 			/*
4342 			 * When allocating non-delayed allocated clusters
4343 			 * (from fallocate, filemap, DIO, or clusters
4344 			 * allocated when delalloc has been disabled by
4345 			 * ext4_nonda_switch), reduce the reserved cluster
4346 			 * count by the number of allocated clusters that
4347 			 * have previously been delayed allocated.  Quota
4348 			 * has been claimed by ext4_mb_new_blocks() above,
4349 			 * so release the quota reservations made for any
4350 			 * previously delayed allocated clusters.
4351 			 */
4352 			lblk = EXT4_LBLK_CMASK(sbi, map->m_lblk);
4353 			len = allocated_clusters << sbi->s_cluster_bits;
4354 			n = ext4_es_delayed_clu(inode, lblk, len);
4355 			if (n > 0)
4356 				ext4_da_update_reserve_space(inode, (int) n, 0);
4357 		}
4358 	}
4359 
4360 	/*
4361 	 * Cache the extent and update transaction to commit on fdatasync only
4362 	 * when it is _not_ an unwritten extent.
4363 	 */
4364 	if ((flags & EXT4_GET_BLOCKS_UNWRIT_EXT) == 0)
4365 		ext4_update_inode_fsync_trans(handle, inode, 1);
4366 	else
4367 		ext4_update_inode_fsync_trans(handle, inode, 0);
4368 
4369 	map->m_flags |= (EXT4_MAP_NEW | EXT4_MAP_MAPPED);
4370 	map->m_pblk = pblk;
4371 	map->m_len = ar.len;
4372 	allocated = map->m_len;
4373 	ext4_ext_show_leaf(inode, path);
4374 out:
4375 	ext4_ext_drop_refs(path);
4376 	kfree(path);
4377 
4378 	trace_ext4_ext_map_blocks_exit(inode, flags, map,
4379 				       err ? err : allocated);
4380 	return err ? err : allocated;
4381 }
4382 
4383 int ext4_ext_truncate(handle_t *handle, struct inode *inode)
4384 {
4385 	struct super_block *sb = inode->i_sb;
4386 	ext4_lblk_t last_block;
4387 	int err = 0;
4388 
4389 	/*
4390 	 * TODO: optimization is possible here.
4391 	 * Probably we need not scan at all,
4392 	 * because page truncation is enough.
4393 	 */
4394 
4395 	/* we have to know where to truncate from in crash case */
4396 	EXT4_I(inode)->i_disksize = inode->i_size;
4397 	err = ext4_mark_inode_dirty(handle, inode);
4398 	if (err)
4399 		return err;
4400 
4401 	last_block = (inode->i_size + sb->s_blocksize - 1)
4402 			>> EXT4_BLOCK_SIZE_BITS(sb);
4403 retry:
4404 	err = ext4_es_remove_extent(inode, last_block,
4405 				    EXT_MAX_BLOCKS - last_block);
4406 	if (err == -ENOMEM) {
4407 		memalloc_retry_wait(GFP_ATOMIC);
4408 		goto retry;
4409 	}
4410 	if (err)
4411 		return err;
4412 retry_remove_space:
4413 	err = ext4_ext_remove_space(inode, last_block, EXT_MAX_BLOCKS - 1);
4414 	if (err == -ENOMEM) {
4415 		memalloc_retry_wait(GFP_ATOMIC);
4416 		goto retry_remove_space;
4417 	}
4418 	return err;
4419 }
4420 
4421 static int ext4_alloc_file_blocks(struct file *file, ext4_lblk_t offset,
4422 				  ext4_lblk_t len, loff_t new_size,
4423 				  int flags)
4424 {
4425 	struct inode *inode = file_inode(file);
4426 	handle_t *handle;
4427 	int ret = 0, ret2 = 0, ret3 = 0;
4428 	int retries = 0;
4429 	int depth = 0;
4430 	struct ext4_map_blocks map;
4431 	unsigned int credits;
4432 	loff_t epos;
4433 
4434 	BUG_ON(!ext4_test_inode_flag(inode, EXT4_INODE_EXTENTS));
4435 	map.m_lblk = offset;
4436 	map.m_len = len;
4437 	/*
4438 	 * Don't normalize the request if it can fit in one extent so
4439 	 * that it doesn't get unnecessarily split into multiple
4440 	 * extents.
4441 	 */
4442 	if (len <= EXT_UNWRITTEN_MAX_LEN)
4443 		flags |= EXT4_GET_BLOCKS_NO_NORMALIZE;
4444 
4445 	/*
4446 	 * credits to insert 1 extent into extent tree
4447 	 */
4448 	credits = ext4_chunk_trans_blocks(inode, len);
4449 	depth = ext_depth(inode);
4450 
4451 retry:
4452 	while (len) {
4453 		/*
4454 		 * Recalculate credits when extent tree depth changes.
4455 		 */
4456 		if (depth != ext_depth(inode)) {
4457 			credits = ext4_chunk_trans_blocks(inode, len);
4458 			depth = ext_depth(inode);
4459 		}
4460 
4461 		handle = ext4_journal_start(inode, EXT4_HT_MAP_BLOCKS,
4462 					    credits);
4463 		if (IS_ERR(handle)) {
4464 			ret = PTR_ERR(handle);
4465 			break;
4466 		}
4467 		ret = ext4_map_blocks(handle, inode, &map, flags);
4468 		if (ret <= 0) {
4469 			ext4_debug("inode #%lu: block %u: len %u: "
4470 				   "ext4_ext_map_blocks returned %d",
4471 				   inode->i_ino, map.m_lblk,
4472 				   map.m_len, ret);
4473 			ext4_mark_inode_dirty(handle, inode);
4474 			ext4_journal_stop(handle);
4475 			break;
4476 		}
4477 		/*
4478 		 * allow a full retry cycle for any remaining allocations
4479 		 */
4480 		retries = 0;
4481 		map.m_lblk += ret;
4482 		map.m_len = len = len - ret;
4483 		epos = (loff_t)map.m_lblk << inode->i_blkbits;
4484 		inode->i_ctime = current_time(inode);
4485 		if (new_size) {
4486 			if (epos > new_size)
4487 				epos = new_size;
4488 			if (ext4_update_inode_size(inode, epos) & 0x1)
4489 				inode->i_mtime = inode->i_ctime;
4490 		}
4491 		ret2 = ext4_mark_inode_dirty(handle, inode);
4492 		ext4_update_inode_fsync_trans(handle, inode, 1);
4493 		ret3 = ext4_journal_stop(handle);
4494 		ret2 = ret3 ? ret3 : ret2;
4495 		if (unlikely(ret2))
4496 			break;
4497 	}
4498 	if (ret == -ENOSPC && ext4_should_retry_alloc(inode->i_sb, &retries))
4499 		goto retry;
4500 
4501 	return ret > 0 ? ret2 : ret;
4502 }
4503 
4504 static int ext4_collapse_range(struct inode *inode, loff_t offset, loff_t len);
4505 
4506 static int ext4_insert_range(struct inode *inode, loff_t offset, loff_t len);
4507 
4508 static long ext4_zero_range(struct file *file, loff_t offset,
4509 			    loff_t len, int mode)
4510 {
4511 	struct inode *inode = file_inode(file);
4512 	struct address_space *mapping = file->f_mapping;
4513 	handle_t *handle = NULL;
4514 	unsigned int max_blocks;
4515 	loff_t new_size = 0;
4516 	int ret = 0;
4517 	int flags;
4518 	int credits;
4519 	int partial_begin, partial_end;
4520 	loff_t start, end;
4521 	ext4_lblk_t lblk;
4522 	unsigned int blkbits = inode->i_blkbits;
4523 
4524 	trace_ext4_zero_range(inode, offset, len, mode);
4525 
4526 	/* Call ext4_force_commit to flush all data in case of data=journal. */
4527 	if (ext4_should_journal_data(inode)) {
4528 		ret = ext4_force_commit(inode->i_sb);
4529 		if (ret)
4530 			return ret;
4531 	}
4532 
4533 	/*
4534 	 * Round up offset. This is not fallocate, we need to zero out
4535 	 * blocks, so convert interior block aligned part of the range to
4536 	 * unwritten and possibly manually zero out unaligned parts of the
4537 	 * range.
4538 	 */
4539 	start = round_up(offset, 1 << blkbits);
4540 	end = round_down((offset + len), 1 << blkbits);
4541 
4542 	if (start < offset || end > offset + len)
4543 		return -EINVAL;
4544 	partial_begin = offset & ((1 << blkbits) - 1);
4545 	partial_end = (offset + len) & ((1 << blkbits) - 1);
4546 
4547 	lblk = start >> blkbits;
4548 	max_blocks = (end >> blkbits);
4549 	if (max_blocks < lblk)
4550 		max_blocks = 0;
4551 	else
4552 		max_blocks -= lblk;
4553 
4554 	inode_lock(inode);
4555 
4556 	/*
4557 	 * Indirect files do not support unwritten extents
4558 	 */
4559 	if (!(ext4_test_inode_flag(inode, EXT4_INODE_EXTENTS))) {
4560 		ret = -EOPNOTSUPP;
4561 		goto out_mutex;
4562 	}
4563 
4564 	if (!(mode & FALLOC_FL_KEEP_SIZE) &&
4565 	    (offset + len > inode->i_size ||
4566 	     offset + len > EXT4_I(inode)->i_disksize)) {
4567 		new_size = offset + len;
4568 		ret = inode_newsize_ok(inode, new_size);
4569 		if (ret)
4570 			goto out_mutex;
4571 	}
4572 
4573 	flags = EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT;
4574 
4575 	/* Wait all existing dio workers, newcomers will block on i_rwsem */
4576 	inode_dio_wait(inode);
4577 
4578 	/* Preallocate the range including the unaligned edges */
4579 	if (partial_begin || partial_end) {
4580 		ret = ext4_alloc_file_blocks(file,
4581 				round_down(offset, 1 << blkbits) >> blkbits,
4582 				(round_up((offset + len), 1 << blkbits) -
4583 				 round_down(offset, 1 << blkbits)) >> blkbits,
4584 				new_size, flags);
4585 		if (ret)
4586 			goto out_mutex;
4587 
4588 	}
4589 
4590 	/* Zero range excluding the unaligned edges */
4591 	if (max_blocks > 0) {
4592 		flags |= (EXT4_GET_BLOCKS_CONVERT_UNWRITTEN |
4593 			  EXT4_EX_NOCACHE);
4594 
4595 		/*
4596 		 * Prevent page faults from reinstantiating pages we have
4597 		 * released from page cache.
4598 		 */
4599 		filemap_invalidate_lock(mapping);
4600 
4601 		ret = ext4_break_layouts(inode);
4602 		if (ret) {
4603 			filemap_invalidate_unlock(mapping);
4604 			goto out_mutex;
4605 		}
4606 
4607 		ret = ext4_update_disksize_before_punch(inode, offset, len);
4608 		if (ret) {
4609 			filemap_invalidate_unlock(mapping);
4610 			goto out_mutex;
4611 		}
4612 		/* Now release the pages and zero block aligned part of pages */
4613 		truncate_pagecache_range(inode, start, end - 1);
4614 		inode->i_mtime = inode->i_ctime = current_time(inode);
4615 
4616 		ret = ext4_alloc_file_blocks(file, lblk, max_blocks, new_size,
4617 					     flags);
4618 		filemap_invalidate_unlock(mapping);
4619 		if (ret)
4620 			goto out_mutex;
4621 	}
4622 	if (!partial_begin && !partial_end)
4623 		goto out_mutex;
4624 
4625 	/*
4626 	 * In worst case we have to writeout two nonadjacent unwritten
4627 	 * blocks and update the inode
4628 	 */
4629 	credits = (2 * ext4_ext_index_trans_blocks(inode, 2)) + 1;
4630 	if (ext4_should_journal_data(inode))
4631 		credits += 2;
4632 	handle = ext4_journal_start(inode, EXT4_HT_MISC, credits);
4633 	if (IS_ERR(handle)) {
4634 		ret = PTR_ERR(handle);
4635 		ext4_std_error(inode->i_sb, ret);
4636 		goto out_mutex;
4637 	}
4638 
4639 	inode->i_mtime = inode->i_ctime = current_time(inode);
4640 	if (new_size)
4641 		ext4_update_inode_size(inode, new_size);
4642 	ret = ext4_mark_inode_dirty(handle, inode);
4643 	if (unlikely(ret))
4644 		goto out_handle;
4645 	/* Zero out partial block at the edges of the range */
4646 	ret = ext4_zero_partial_blocks(handle, inode, offset, len);
4647 	if (ret >= 0)
4648 		ext4_update_inode_fsync_trans(handle, inode, 1);
4649 
4650 	if (file->f_flags & O_SYNC)
4651 		ext4_handle_sync(handle);
4652 
4653 out_handle:
4654 	ext4_journal_stop(handle);
4655 out_mutex:
4656 	inode_unlock(inode);
4657 	return ret;
4658 }
4659 
4660 /*
4661  * preallocate space for a file. This implements ext4's fallocate file
4662  * operation, which gets called from sys_fallocate system call.
4663  * For block-mapped files, posix_fallocate should fall back to the method
4664  * of writing zeroes to the required new blocks (the same behavior which is
4665  * expected for file systems which do not support fallocate() system call).
4666  */
4667 long ext4_fallocate(struct file *file, int mode, loff_t offset, loff_t len)
4668 {
4669 	struct inode *inode = file_inode(file);
4670 	loff_t new_size = 0;
4671 	unsigned int max_blocks;
4672 	int ret = 0;
4673 	int flags;
4674 	ext4_lblk_t lblk;
4675 	unsigned int blkbits = inode->i_blkbits;
4676 
4677 	/*
4678 	 * Encrypted inodes can't handle collapse range or insert
4679 	 * range since we would need to re-encrypt blocks with a
4680 	 * different IV or XTS tweak (which are based on the logical
4681 	 * block number).
4682 	 */
4683 	if (IS_ENCRYPTED(inode) &&
4684 	    (mode & (FALLOC_FL_COLLAPSE_RANGE | FALLOC_FL_INSERT_RANGE)))
4685 		return -EOPNOTSUPP;
4686 
4687 	/* Return error if mode is not supported */
4688 	if (mode & ~(FALLOC_FL_KEEP_SIZE | FALLOC_FL_PUNCH_HOLE |
4689 		     FALLOC_FL_COLLAPSE_RANGE | FALLOC_FL_ZERO_RANGE |
4690 		     FALLOC_FL_INSERT_RANGE))
4691 		return -EOPNOTSUPP;
4692 
4693 	if (mode & FALLOC_FL_PUNCH_HOLE) {
4694 		ret = ext4_punch_hole(inode, offset, len);
4695 		goto exit;
4696 	}
4697 
4698 	ret = ext4_convert_inline_data(inode);
4699 	if (ret)
4700 		goto exit;
4701 
4702 	if (mode & FALLOC_FL_COLLAPSE_RANGE) {
4703 		ret = ext4_collapse_range(inode, offset, len);
4704 		goto exit;
4705 	}
4706 
4707 	if (mode & FALLOC_FL_INSERT_RANGE) {
4708 		ret = ext4_insert_range(inode, offset, len);
4709 		goto exit;
4710 	}
4711 
4712 	if (mode & FALLOC_FL_ZERO_RANGE) {
4713 		ret = ext4_zero_range(file, offset, len, mode);
4714 		goto exit;
4715 	}
4716 	trace_ext4_fallocate_enter(inode, offset, len, mode);
4717 	lblk = offset >> blkbits;
4718 
4719 	max_blocks = EXT4_MAX_BLOCKS(len, offset, blkbits);
4720 	flags = EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT;
4721 
4722 	inode_lock(inode);
4723 
4724 	/*
4725 	 * We only support preallocation for extent-based files only
4726 	 */
4727 	if (!(ext4_test_inode_flag(inode, EXT4_INODE_EXTENTS))) {
4728 		ret = -EOPNOTSUPP;
4729 		goto out;
4730 	}
4731 
4732 	if (!(mode & FALLOC_FL_KEEP_SIZE) &&
4733 	    (offset + len > inode->i_size ||
4734 	     offset + len > EXT4_I(inode)->i_disksize)) {
4735 		new_size = offset + len;
4736 		ret = inode_newsize_ok(inode, new_size);
4737 		if (ret)
4738 			goto out;
4739 	}
4740 
4741 	/* Wait all existing dio workers, newcomers will block on i_rwsem */
4742 	inode_dio_wait(inode);
4743 
4744 	ret = ext4_alloc_file_blocks(file, lblk, max_blocks, new_size, flags);
4745 	if (ret)
4746 		goto out;
4747 
4748 	if (file->f_flags & O_SYNC && EXT4_SB(inode->i_sb)->s_journal) {
4749 		ret = ext4_fc_commit(EXT4_SB(inode->i_sb)->s_journal,
4750 					EXT4_I(inode)->i_sync_tid);
4751 	}
4752 out:
4753 	inode_unlock(inode);
4754 	trace_ext4_fallocate_exit(inode, offset, max_blocks, ret);
4755 exit:
4756 	return ret;
4757 }
4758 
4759 /*
4760  * This function convert a range of blocks to written extents
4761  * The caller of this function will pass the start offset and the size.
4762  * all unwritten extents within this range will be converted to
4763  * written extents.
4764  *
4765  * This function is called from the direct IO end io call back
4766  * function, to convert the fallocated extents after IO is completed.
4767  * Returns 0 on success.
4768  */
4769 int ext4_convert_unwritten_extents(handle_t *handle, struct inode *inode,
4770 				   loff_t offset, ssize_t len)
4771 {
4772 	unsigned int max_blocks;
4773 	int ret = 0, ret2 = 0, ret3 = 0;
4774 	struct ext4_map_blocks map;
4775 	unsigned int blkbits = inode->i_blkbits;
4776 	unsigned int credits = 0;
4777 
4778 	map.m_lblk = offset >> blkbits;
4779 	max_blocks = EXT4_MAX_BLOCKS(len, offset, blkbits);
4780 
4781 	if (!handle) {
4782 		/*
4783 		 * credits to insert 1 extent into extent tree
4784 		 */
4785 		credits = ext4_chunk_trans_blocks(inode, max_blocks);
4786 	}
4787 	while (ret >= 0 && ret < max_blocks) {
4788 		map.m_lblk += ret;
4789 		map.m_len = (max_blocks -= ret);
4790 		if (credits) {
4791 			handle = ext4_journal_start(inode, EXT4_HT_MAP_BLOCKS,
4792 						    credits);
4793 			if (IS_ERR(handle)) {
4794 				ret = PTR_ERR(handle);
4795 				break;
4796 			}
4797 		}
4798 		ret = ext4_map_blocks(handle, inode, &map,
4799 				      EXT4_GET_BLOCKS_IO_CONVERT_EXT);
4800 		if (ret <= 0)
4801 			ext4_warning(inode->i_sb,
4802 				     "inode #%lu: block %u: len %u: "
4803 				     "ext4_ext_map_blocks returned %d",
4804 				     inode->i_ino, map.m_lblk,
4805 				     map.m_len, ret);
4806 		ret2 = ext4_mark_inode_dirty(handle, inode);
4807 		if (credits) {
4808 			ret3 = ext4_journal_stop(handle);
4809 			if (unlikely(ret3))
4810 				ret2 = ret3;
4811 		}
4812 
4813 		if (ret <= 0 || ret2)
4814 			break;
4815 	}
4816 	return ret > 0 ? ret2 : ret;
4817 }
4818 
4819 int ext4_convert_unwritten_io_end_vec(handle_t *handle, ext4_io_end_t *io_end)
4820 {
4821 	int ret = 0, err = 0;
4822 	struct ext4_io_end_vec *io_end_vec;
4823 
4824 	/*
4825 	 * This is somewhat ugly but the idea is clear: When transaction is
4826 	 * reserved, everything goes into it. Otherwise we rather start several
4827 	 * smaller transactions for conversion of each extent separately.
4828 	 */
4829 	if (handle) {
4830 		handle = ext4_journal_start_reserved(handle,
4831 						     EXT4_HT_EXT_CONVERT);
4832 		if (IS_ERR(handle))
4833 			return PTR_ERR(handle);
4834 	}
4835 
4836 	list_for_each_entry(io_end_vec, &io_end->list_vec, list) {
4837 		ret = ext4_convert_unwritten_extents(handle, io_end->inode,
4838 						     io_end_vec->offset,
4839 						     io_end_vec->size);
4840 		if (ret)
4841 			break;
4842 	}
4843 
4844 	if (handle)
4845 		err = ext4_journal_stop(handle);
4846 
4847 	return ret < 0 ? ret : err;
4848 }
4849 
4850 static int ext4_iomap_xattr_fiemap(struct inode *inode, struct iomap *iomap)
4851 {
4852 	__u64 physical = 0;
4853 	__u64 length = 0;
4854 	int blockbits = inode->i_sb->s_blocksize_bits;
4855 	int error = 0;
4856 	u16 iomap_type;
4857 
4858 	/* in-inode? */
4859 	if (ext4_test_inode_state(inode, EXT4_STATE_XATTR)) {
4860 		struct ext4_iloc iloc;
4861 		int offset;	/* offset of xattr in inode */
4862 
4863 		error = ext4_get_inode_loc(inode, &iloc);
4864 		if (error)
4865 			return error;
4866 		physical = (__u64)iloc.bh->b_blocknr << blockbits;
4867 		offset = EXT4_GOOD_OLD_INODE_SIZE +
4868 				EXT4_I(inode)->i_extra_isize;
4869 		physical += offset;
4870 		length = EXT4_SB(inode->i_sb)->s_inode_size - offset;
4871 		brelse(iloc.bh);
4872 		iomap_type = IOMAP_INLINE;
4873 	} else if (EXT4_I(inode)->i_file_acl) { /* external block */
4874 		physical = (__u64)EXT4_I(inode)->i_file_acl << blockbits;
4875 		length = inode->i_sb->s_blocksize;
4876 		iomap_type = IOMAP_MAPPED;
4877 	} else {
4878 		/* no in-inode or external block for xattr, so return -ENOENT */
4879 		error = -ENOENT;
4880 		goto out;
4881 	}
4882 
4883 	iomap->addr = physical;
4884 	iomap->offset = 0;
4885 	iomap->length = length;
4886 	iomap->type = iomap_type;
4887 	iomap->flags = 0;
4888 out:
4889 	return error;
4890 }
4891 
4892 static int ext4_iomap_xattr_begin(struct inode *inode, loff_t offset,
4893 				  loff_t length, unsigned flags,
4894 				  struct iomap *iomap, struct iomap *srcmap)
4895 {
4896 	int error;
4897 
4898 	error = ext4_iomap_xattr_fiemap(inode, iomap);
4899 	if (error == 0 && (offset >= iomap->length))
4900 		error = -ENOENT;
4901 	return error;
4902 }
4903 
4904 static const struct iomap_ops ext4_iomap_xattr_ops = {
4905 	.iomap_begin		= ext4_iomap_xattr_begin,
4906 };
4907 
4908 static int ext4_fiemap_check_ranges(struct inode *inode, u64 start, u64 *len)
4909 {
4910 	u64 maxbytes;
4911 
4912 	if (ext4_test_inode_flag(inode, EXT4_INODE_EXTENTS))
4913 		maxbytes = inode->i_sb->s_maxbytes;
4914 	else
4915 		maxbytes = EXT4_SB(inode->i_sb)->s_bitmap_maxbytes;
4916 
4917 	if (*len == 0)
4918 		return -EINVAL;
4919 	if (start > maxbytes)
4920 		return -EFBIG;
4921 
4922 	/*
4923 	 * Shrink request scope to what the fs can actually handle.
4924 	 */
4925 	if (*len > maxbytes || (maxbytes - *len) < start)
4926 		*len = maxbytes - start;
4927 	return 0;
4928 }
4929 
4930 int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
4931 		u64 start, u64 len)
4932 {
4933 	int error = 0;
4934 
4935 	if (fieinfo->fi_flags & FIEMAP_FLAG_CACHE) {
4936 		error = ext4_ext_precache(inode);
4937 		if (error)
4938 			return error;
4939 		fieinfo->fi_flags &= ~FIEMAP_FLAG_CACHE;
4940 	}
4941 
4942 	/*
4943 	 * For bitmap files the maximum size limit could be smaller than
4944 	 * s_maxbytes, so check len here manually instead of just relying on the
4945 	 * generic check.
4946 	 */
4947 	error = ext4_fiemap_check_ranges(inode, start, &len);
4948 	if (error)
4949 		return error;
4950 
4951 	if (fieinfo->fi_flags & FIEMAP_FLAG_XATTR) {
4952 		fieinfo->fi_flags &= ~FIEMAP_FLAG_XATTR;
4953 		return iomap_fiemap(inode, fieinfo, start, len,
4954 				    &ext4_iomap_xattr_ops);
4955 	}
4956 
4957 	return iomap_fiemap(inode, fieinfo, start, len, &ext4_iomap_report_ops);
4958 }
4959 
4960 int ext4_get_es_cache(struct inode *inode, struct fiemap_extent_info *fieinfo,
4961 		      __u64 start, __u64 len)
4962 {
4963 	ext4_lblk_t start_blk, len_blks;
4964 	__u64 last_blk;
4965 	int error = 0;
4966 
4967 	if (ext4_has_inline_data(inode)) {
4968 		int has_inline;
4969 
4970 		down_read(&EXT4_I(inode)->xattr_sem);
4971 		has_inline = ext4_has_inline_data(inode);
4972 		up_read(&EXT4_I(inode)->xattr_sem);
4973 		if (has_inline)
4974 			return 0;
4975 	}
4976 
4977 	if (fieinfo->fi_flags & FIEMAP_FLAG_CACHE) {
4978 		error = ext4_ext_precache(inode);
4979 		if (error)
4980 			return error;
4981 		fieinfo->fi_flags &= ~FIEMAP_FLAG_CACHE;
4982 	}
4983 
4984 	error = fiemap_prep(inode, fieinfo, start, &len, 0);
4985 	if (error)
4986 		return error;
4987 
4988 	error = ext4_fiemap_check_ranges(inode, start, &len);
4989 	if (error)
4990 		return error;
4991 
4992 	start_blk = start >> inode->i_sb->s_blocksize_bits;
4993 	last_blk = (start + len - 1) >> inode->i_sb->s_blocksize_bits;
4994 	if (last_blk >= EXT_MAX_BLOCKS)
4995 		last_blk = EXT_MAX_BLOCKS-1;
4996 	len_blks = ((ext4_lblk_t) last_blk) - start_blk + 1;
4997 
4998 	/*
4999 	 * Walk the extent tree gathering extent information
5000 	 * and pushing extents back to the user.
5001 	 */
5002 	return ext4_fill_es_cache_info(inode, start_blk, len_blks, fieinfo);
5003 }
5004 
5005 /*
5006  * ext4_ext_shift_path_extents:
5007  * Shift the extents of a path structure lying between path[depth].p_ext
5008  * and EXT_LAST_EXTENT(path[depth].p_hdr), by @shift blocks. @SHIFT tells
5009  * if it is right shift or left shift operation.
5010  */
5011 static int
5012 ext4_ext_shift_path_extents(struct ext4_ext_path *path, ext4_lblk_t shift,
5013 			    struct inode *inode, handle_t *handle,
5014 			    enum SHIFT_DIRECTION SHIFT)
5015 {
5016 	int depth, err = 0;
5017 	struct ext4_extent *ex_start, *ex_last;
5018 	bool update = false;
5019 	int credits, restart_credits;
5020 	depth = path->p_depth;
5021 
5022 	while (depth >= 0) {
5023 		if (depth == path->p_depth) {
5024 			ex_start = path[depth].p_ext;
5025 			if (!ex_start)
5026 				return -EFSCORRUPTED;
5027 
5028 			ex_last = EXT_LAST_EXTENT(path[depth].p_hdr);
5029 			/* leaf + sb + inode */
5030 			credits = 3;
5031 			if (ex_start == EXT_FIRST_EXTENT(path[depth].p_hdr)) {
5032 				update = true;
5033 				/* extent tree + sb + inode */
5034 				credits = depth + 2;
5035 			}
5036 
5037 			restart_credits = ext4_writepage_trans_blocks(inode);
5038 			err = ext4_datasem_ensure_credits(handle, inode, credits,
5039 					restart_credits, 0);
5040 			if (err) {
5041 				if (err > 0)
5042 					err = -EAGAIN;
5043 				goto out;
5044 			}
5045 
5046 			err = ext4_ext_get_access(handle, inode, path + depth);
5047 			if (err)
5048 				goto out;
5049 
5050 			while (ex_start <= ex_last) {
5051 				if (SHIFT == SHIFT_LEFT) {
5052 					le32_add_cpu(&ex_start->ee_block,
5053 						-shift);
5054 					/* Try to merge to the left. */
5055 					if ((ex_start >
5056 					    EXT_FIRST_EXTENT(path[depth].p_hdr))
5057 					    &&
5058 					    ext4_ext_try_to_merge_right(inode,
5059 					    path, ex_start - 1))
5060 						ex_last--;
5061 					else
5062 						ex_start++;
5063 				} else {
5064 					le32_add_cpu(&ex_last->ee_block, shift);
5065 					ext4_ext_try_to_merge_right(inode, path,
5066 						ex_last);
5067 					ex_last--;
5068 				}
5069 			}
5070 			err = ext4_ext_dirty(handle, inode, path + depth);
5071 			if (err)
5072 				goto out;
5073 
5074 			if (--depth < 0 || !update)
5075 				break;
5076 		}
5077 
5078 		/* Update index too */
5079 		err = ext4_ext_get_access(handle, inode, path + depth);
5080 		if (err)
5081 			goto out;
5082 
5083 		if (SHIFT == SHIFT_LEFT)
5084 			le32_add_cpu(&path[depth].p_idx->ei_block, -shift);
5085 		else
5086 			le32_add_cpu(&path[depth].p_idx->ei_block, shift);
5087 		err = ext4_ext_dirty(handle, inode, path + depth);
5088 		if (err)
5089 			goto out;
5090 
5091 		/* we are done if current index is not a starting index */
5092 		if (path[depth].p_idx != EXT_FIRST_INDEX(path[depth].p_hdr))
5093 			break;
5094 
5095 		depth--;
5096 	}
5097 
5098 out:
5099 	return err;
5100 }
5101 
5102 /*
5103  * ext4_ext_shift_extents:
5104  * All the extents which lies in the range from @start to the last allocated
5105  * block for the @inode are shifted either towards left or right (depending
5106  * upon @SHIFT) by @shift blocks.
5107  * On success, 0 is returned, error otherwise.
5108  */
5109 static int
5110 ext4_ext_shift_extents(struct inode *inode, handle_t *handle,
5111 		       ext4_lblk_t start, ext4_lblk_t shift,
5112 		       enum SHIFT_DIRECTION SHIFT)
5113 {
5114 	struct ext4_ext_path *path;
5115 	int ret = 0, depth;
5116 	struct ext4_extent *extent;
5117 	ext4_lblk_t stop, *iterator, ex_start, ex_end;
5118 	ext4_lblk_t tmp = EXT_MAX_BLOCKS;
5119 
5120 	/* Let path point to the last extent */
5121 	path = ext4_find_extent(inode, EXT_MAX_BLOCKS - 1, NULL,
5122 				EXT4_EX_NOCACHE);
5123 	if (IS_ERR(path))
5124 		return PTR_ERR(path);
5125 
5126 	depth = path->p_depth;
5127 	extent = path[depth].p_ext;
5128 	if (!extent)
5129 		goto out;
5130 
5131 	stop = le32_to_cpu(extent->ee_block);
5132 
5133        /*
5134 	* For left shifts, make sure the hole on the left is big enough to
5135 	* accommodate the shift.  For right shifts, make sure the last extent
5136 	* won't be shifted beyond EXT_MAX_BLOCKS.
5137 	*/
5138 	if (SHIFT == SHIFT_LEFT) {
5139 		path = ext4_find_extent(inode, start - 1, &path,
5140 					EXT4_EX_NOCACHE);
5141 		if (IS_ERR(path))
5142 			return PTR_ERR(path);
5143 		depth = path->p_depth;
5144 		extent =  path[depth].p_ext;
5145 		if (extent) {
5146 			ex_start = le32_to_cpu(extent->ee_block);
5147 			ex_end = le32_to_cpu(extent->ee_block) +
5148 				ext4_ext_get_actual_len(extent);
5149 		} else {
5150 			ex_start = 0;
5151 			ex_end = 0;
5152 		}
5153 
5154 		if ((start == ex_start && shift > ex_start) ||
5155 		    (shift > start - ex_end)) {
5156 			ret = -EINVAL;
5157 			goto out;
5158 		}
5159 	} else {
5160 		if (shift > EXT_MAX_BLOCKS -
5161 		    (stop + ext4_ext_get_actual_len(extent))) {
5162 			ret = -EINVAL;
5163 			goto out;
5164 		}
5165 	}
5166 
5167 	/*
5168 	 * In case of left shift, iterator points to start and it is increased
5169 	 * till we reach stop. In case of right shift, iterator points to stop
5170 	 * and it is decreased till we reach start.
5171 	 */
5172 again:
5173 	if (SHIFT == SHIFT_LEFT)
5174 		iterator = &start;
5175 	else
5176 		iterator = &stop;
5177 
5178 	if (tmp != EXT_MAX_BLOCKS)
5179 		*iterator = tmp;
5180 
5181 	/*
5182 	 * Its safe to start updating extents.  Start and stop are unsigned, so
5183 	 * in case of right shift if extent with 0 block is reached, iterator
5184 	 * becomes NULL to indicate the end of the loop.
5185 	 */
5186 	while (iterator && start <= stop) {
5187 		path = ext4_find_extent(inode, *iterator, &path,
5188 					EXT4_EX_NOCACHE);
5189 		if (IS_ERR(path))
5190 			return PTR_ERR(path);
5191 		depth = path->p_depth;
5192 		extent = path[depth].p_ext;
5193 		if (!extent) {
5194 			EXT4_ERROR_INODE(inode, "unexpected hole at %lu",
5195 					 (unsigned long) *iterator);
5196 			return -EFSCORRUPTED;
5197 		}
5198 		if (SHIFT == SHIFT_LEFT && *iterator >
5199 		    le32_to_cpu(extent->ee_block)) {
5200 			/* Hole, move to the next extent */
5201 			if (extent < EXT_LAST_EXTENT(path[depth].p_hdr)) {
5202 				path[depth].p_ext++;
5203 			} else {
5204 				*iterator = ext4_ext_next_allocated_block(path);
5205 				continue;
5206 			}
5207 		}
5208 
5209 		tmp = *iterator;
5210 		if (SHIFT == SHIFT_LEFT) {
5211 			extent = EXT_LAST_EXTENT(path[depth].p_hdr);
5212 			*iterator = le32_to_cpu(extent->ee_block) +
5213 					ext4_ext_get_actual_len(extent);
5214 		} else {
5215 			extent = EXT_FIRST_EXTENT(path[depth].p_hdr);
5216 			if (le32_to_cpu(extent->ee_block) > 0)
5217 				*iterator = le32_to_cpu(extent->ee_block) - 1;
5218 			else
5219 				/* Beginning is reached, end of the loop */
5220 				iterator = NULL;
5221 			/* Update path extent in case we need to stop */
5222 			while (le32_to_cpu(extent->ee_block) < start)
5223 				extent++;
5224 			path[depth].p_ext = extent;
5225 		}
5226 		ret = ext4_ext_shift_path_extents(path, shift, inode,
5227 				handle, SHIFT);
5228 		/* iterator can be NULL which means we should break */
5229 		if (ret == -EAGAIN)
5230 			goto again;
5231 		if (ret)
5232 			break;
5233 	}
5234 out:
5235 	ext4_ext_drop_refs(path);
5236 	kfree(path);
5237 	return ret;
5238 }
5239 
5240 /*
5241  * ext4_collapse_range:
5242  * This implements the fallocate's collapse range functionality for ext4
5243  * Returns: 0 and non-zero on error.
5244  */
5245 static int ext4_collapse_range(struct inode *inode, loff_t offset, loff_t len)
5246 {
5247 	struct super_block *sb = inode->i_sb;
5248 	struct address_space *mapping = inode->i_mapping;
5249 	ext4_lblk_t punch_start, punch_stop;
5250 	handle_t *handle;
5251 	unsigned int credits;
5252 	loff_t new_size, ioffset;
5253 	int ret;
5254 
5255 	/*
5256 	 * We need to test this early because xfstests assumes that a
5257 	 * collapse range of (0, 1) will return EOPNOTSUPP if the file
5258 	 * system does not support collapse range.
5259 	 */
5260 	if (!ext4_test_inode_flag(inode, EXT4_INODE_EXTENTS))
5261 		return -EOPNOTSUPP;
5262 
5263 	/* Collapse range works only on fs cluster size aligned regions. */
5264 	if (!IS_ALIGNED(offset | len, EXT4_CLUSTER_SIZE(sb)))
5265 		return -EINVAL;
5266 
5267 	trace_ext4_collapse_range(inode, offset, len);
5268 
5269 	punch_start = offset >> EXT4_BLOCK_SIZE_BITS(sb);
5270 	punch_stop = (offset + len) >> EXT4_BLOCK_SIZE_BITS(sb);
5271 
5272 	/* Call ext4_force_commit to flush all data in case of data=journal. */
5273 	if (ext4_should_journal_data(inode)) {
5274 		ret = ext4_force_commit(inode->i_sb);
5275 		if (ret)
5276 			return ret;
5277 	}
5278 
5279 	inode_lock(inode);
5280 	/*
5281 	 * There is no need to overlap collapse range with EOF, in which case
5282 	 * it is effectively a truncate operation
5283 	 */
5284 	if (offset + len >= inode->i_size) {
5285 		ret = -EINVAL;
5286 		goto out_mutex;
5287 	}
5288 
5289 	/* Currently just for extent based files */
5290 	if (!ext4_test_inode_flag(inode, EXT4_INODE_EXTENTS)) {
5291 		ret = -EOPNOTSUPP;
5292 		goto out_mutex;
5293 	}
5294 
5295 	/* Wait for existing dio to complete */
5296 	inode_dio_wait(inode);
5297 
5298 	/*
5299 	 * Prevent page faults from reinstantiating pages we have released from
5300 	 * page cache.
5301 	 */
5302 	filemap_invalidate_lock(mapping);
5303 
5304 	ret = ext4_break_layouts(inode);
5305 	if (ret)
5306 		goto out_mmap;
5307 
5308 	/*
5309 	 * Need to round down offset to be aligned with page size boundary
5310 	 * for page size > block size.
5311 	 */
5312 	ioffset = round_down(offset, PAGE_SIZE);
5313 	/*
5314 	 * Write tail of the last page before removed range since it will get
5315 	 * removed from the page cache below.
5316 	 */
5317 	ret = filemap_write_and_wait_range(mapping, ioffset, offset);
5318 	if (ret)
5319 		goto out_mmap;
5320 	/*
5321 	 * Write data that will be shifted to preserve them when discarding
5322 	 * page cache below. We are also protected from pages becoming dirty
5323 	 * by i_rwsem and invalidate_lock.
5324 	 */
5325 	ret = filemap_write_and_wait_range(mapping, offset + len,
5326 					   LLONG_MAX);
5327 	if (ret)
5328 		goto out_mmap;
5329 	truncate_pagecache(inode, ioffset);
5330 
5331 	credits = ext4_writepage_trans_blocks(inode);
5332 	handle = ext4_journal_start(inode, EXT4_HT_TRUNCATE, credits);
5333 	if (IS_ERR(handle)) {
5334 		ret = PTR_ERR(handle);
5335 		goto out_mmap;
5336 	}
5337 	ext4_fc_mark_ineligible(sb, EXT4_FC_REASON_FALLOC_RANGE, handle);
5338 
5339 	down_write(&EXT4_I(inode)->i_data_sem);
5340 	ext4_discard_preallocations(inode, 0);
5341 
5342 	ret = ext4_es_remove_extent(inode, punch_start,
5343 				    EXT_MAX_BLOCKS - punch_start);
5344 	if (ret) {
5345 		up_write(&EXT4_I(inode)->i_data_sem);
5346 		goto out_stop;
5347 	}
5348 
5349 	ret = ext4_ext_remove_space(inode, punch_start, punch_stop - 1);
5350 	if (ret) {
5351 		up_write(&EXT4_I(inode)->i_data_sem);
5352 		goto out_stop;
5353 	}
5354 	ext4_discard_preallocations(inode, 0);
5355 
5356 	ret = ext4_ext_shift_extents(inode, handle, punch_stop,
5357 				     punch_stop - punch_start, SHIFT_LEFT);
5358 	if (ret) {
5359 		up_write(&EXT4_I(inode)->i_data_sem);
5360 		goto out_stop;
5361 	}
5362 
5363 	new_size = inode->i_size - len;
5364 	i_size_write(inode, new_size);
5365 	EXT4_I(inode)->i_disksize = new_size;
5366 
5367 	up_write(&EXT4_I(inode)->i_data_sem);
5368 	if (IS_SYNC(inode))
5369 		ext4_handle_sync(handle);
5370 	inode->i_mtime = inode->i_ctime = current_time(inode);
5371 	ret = ext4_mark_inode_dirty(handle, inode);
5372 	ext4_update_inode_fsync_trans(handle, inode, 1);
5373 
5374 out_stop:
5375 	ext4_journal_stop(handle);
5376 out_mmap:
5377 	filemap_invalidate_unlock(mapping);
5378 out_mutex:
5379 	inode_unlock(inode);
5380 	return ret;
5381 }
5382 
5383 /*
5384  * ext4_insert_range:
5385  * This function implements the FALLOC_FL_INSERT_RANGE flag of fallocate.
5386  * The data blocks starting from @offset to the EOF are shifted by @len
5387  * towards right to create a hole in the @inode. Inode size is increased
5388  * by len bytes.
5389  * Returns 0 on success, error otherwise.
5390  */
5391 static int ext4_insert_range(struct inode *inode, loff_t offset, loff_t len)
5392 {
5393 	struct super_block *sb = inode->i_sb;
5394 	struct address_space *mapping = inode->i_mapping;
5395 	handle_t *handle;
5396 	struct ext4_ext_path *path;
5397 	struct ext4_extent *extent;
5398 	ext4_lblk_t offset_lblk, len_lblk, ee_start_lblk = 0;
5399 	unsigned int credits, ee_len;
5400 	int ret = 0, depth, split_flag = 0;
5401 	loff_t ioffset;
5402 
5403 	/*
5404 	 * We need to test this early because xfstests assumes that an
5405 	 * insert range of (0, 1) will return EOPNOTSUPP if the file
5406 	 * system does not support insert range.
5407 	 */
5408 	if (!ext4_test_inode_flag(inode, EXT4_INODE_EXTENTS))
5409 		return -EOPNOTSUPP;
5410 
5411 	/* Insert range works only on fs cluster size aligned regions. */
5412 	if (!IS_ALIGNED(offset | len, EXT4_CLUSTER_SIZE(sb)))
5413 		return -EINVAL;
5414 
5415 	trace_ext4_insert_range(inode, offset, len);
5416 
5417 	offset_lblk = offset >> EXT4_BLOCK_SIZE_BITS(sb);
5418 	len_lblk = len >> EXT4_BLOCK_SIZE_BITS(sb);
5419 
5420 	/* Call ext4_force_commit to flush all data in case of data=journal */
5421 	if (ext4_should_journal_data(inode)) {
5422 		ret = ext4_force_commit(inode->i_sb);
5423 		if (ret)
5424 			return ret;
5425 	}
5426 
5427 	inode_lock(inode);
5428 	/* Currently just for extent based files */
5429 	if (!ext4_test_inode_flag(inode, EXT4_INODE_EXTENTS)) {
5430 		ret = -EOPNOTSUPP;
5431 		goto out_mutex;
5432 	}
5433 
5434 	/* Check whether the maximum file size would be exceeded */
5435 	if (len > inode->i_sb->s_maxbytes - inode->i_size) {
5436 		ret = -EFBIG;
5437 		goto out_mutex;
5438 	}
5439 
5440 	/* Offset must be less than i_size */
5441 	if (offset >= inode->i_size) {
5442 		ret = -EINVAL;
5443 		goto out_mutex;
5444 	}
5445 
5446 	/* Wait for existing dio to complete */
5447 	inode_dio_wait(inode);
5448 
5449 	/*
5450 	 * Prevent page faults from reinstantiating pages we have released from
5451 	 * page cache.
5452 	 */
5453 	filemap_invalidate_lock(mapping);
5454 
5455 	ret = ext4_break_layouts(inode);
5456 	if (ret)
5457 		goto out_mmap;
5458 
5459 	/*
5460 	 * Need to round down to align start offset to page size boundary
5461 	 * for page size > block size.
5462 	 */
5463 	ioffset = round_down(offset, PAGE_SIZE);
5464 	/* Write out all dirty pages */
5465 	ret = filemap_write_and_wait_range(inode->i_mapping, ioffset,
5466 			LLONG_MAX);
5467 	if (ret)
5468 		goto out_mmap;
5469 	truncate_pagecache(inode, ioffset);
5470 
5471 	credits = ext4_writepage_trans_blocks(inode);
5472 	handle = ext4_journal_start(inode, EXT4_HT_TRUNCATE, credits);
5473 	if (IS_ERR(handle)) {
5474 		ret = PTR_ERR(handle);
5475 		goto out_mmap;
5476 	}
5477 	ext4_fc_mark_ineligible(sb, EXT4_FC_REASON_FALLOC_RANGE, handle);
5478 
5479 	/* Expand file to avoid data loss if there is error while shifting */
5480 	inode->i_size += len;
5481 	EXT4_I(inode)->i_disksize += len;
5482 	inode->i_mtime = inode->i_ctime = current_time(inode);
5483 	ret = ext4_mark_inode_dirty(handle, inode);
5484 	if (ret)
5485 		goto out_stop;
5486 
5487 	down_write(&EXT4_I(inode)->i_data_sem);
5488 	ext4_discard_preallocations(inode, 0);
5489 
5490 	path = ext4_find_extent(inode, offset_lblk, NULL, 0);
5491 	if (IS_ERR(path)) {
5492 		up_write(&EXT4_I(inode)->i_data_sem);
5493 		goto out_stop;
5494 	}
5495 
5496 	depth = ext_depth(inode);
5497 	extent = path[depth].p_ext;
5498 	if (extent) {
5499 		ee_start_lblk = le32_to_cpu(extent->ee_block);
5500 		ee_len = ext4_ext_get_actual_len(extent);
5501 
5502 		/*
5503 		 * If offset_lblk is not the starting block of extent, split
5504 		 * the extent @offset_lblk
5505 		 */
5506 		if ((offset_lblk > ee_start_lblk) &&
5507 				(offset_lblk < (ee_start_lblk + ee_len))) {
5508 			if (ext4_ext_is_unwritten(extent))
5509 				split_flag = EXT4_EXT_MARK_UNWRIT1 |
5510 					EXT4_EXT_MARK_UNWRIT2;
5511 			ret = ext4_split_extent_at(handle, inode, &path,
5512 					offset_lblk, split_flag,
5513 					EXT4_EX_NOCACHE |
5514 					EXT4_GET_BLOCKS_PRE_IO |
5515 					EXT4_GET_BLOCKS_METADATA_NOFAIL);
5516 		}
5517 
5518 		ext4_ext_drop_refs(path);
5519 		kfree(path);
5520 		if (ret < 0) {
5521 			up_write(&EXT4_I(inode)->i_data_sem);
5522 			goto out_stop;
5523 		}
5524 	} else {
5525 		ext4_ext_drop_refs(path);
5526 		kfree(path);
5527 	}
5528 
5529 	ret = ext4_es_remove_extent(inode, offset_lblk,
5530 			EXT_MAX_BLOCKS - offset_lblk);
5531 	if (ret) {
5532 		up_write(&EXT4_I(inode)->i_data_sem);
5533 		goto out_stop;
5534 	}
5535 
5536 	/*
5537 	 * if offset_lblk lies in a hole which is at start of file, use
5538 	 * ee_start_lblk to shift extents
5539 	 */
5540 	ret = ext4_ext_shift_extents(inode, handle,
5541 		ee_start_lblk > offset_lblk ? ee_start_lblk : offset_lblk,
5542 		len_lblk, SHIFT_RIGHT);
5543 
5544 	up_write(&EXT4_I(inode)->i_data_sem);
5545 	if (IS_SYNC(inode))
5546 		ext4_handle_sync(handle);
5547 	if (ret >= 0)
5548 		ext4_update_inode_fsync_trans(handle, inode, 1);
5549 
5550 out_stop:
5551 	ext4_journal_stop(handle);
5552 out_mmap:
5553 	filemap_invalidate_unlock(mapping);
5554 out_mutex:
5555 	inode_unlock(inode);
5556 	return ret;
5557 }
5558 
5559 /**
5560  * ext4_swap_extents() - Swap extents between two inodes
5561  * @handle: handle for this transaction
5562  * @inode1:	First inode
5563  * @inode2:	Second inode
5564  * @lblk1:	Start block for first inode
5565  * @lblk2:	Start block for second inode
5566  * @count:	Number of blocks to swap
5567  * @unwritten: Mark second inode's extents as unwritten after swap
5568  * @erp:	Pointer to save error value
5569  *
5570  * This helper routine does exactly what is promise "swap extents". All other
5571  * stuff such as page-cache locking consistency, bh mapping consistency or
5572  * extent's data copying must be performed by caller.
5573  * Locking:
5574  *		i_rwsem is held for both inodes
5575  * 		i_data_sem is locked for write for both inodes
5576  * Assumptions:
5577  *		All pages from requested range are locked for both inodes
5578  */
5579 int
5580 ext4_swap_extents(handle_t *handle, struct inode *inode1,
5581 		  struct inode *inode2, ext4_lblk_t lblk1, ext4_lblk_t lblk2,
5582 		  ext4_lblk_t count, int unwritten, int *erp)
5583 {
5584 	struct ext4_ext_path *path1 = NULL;
5585 	struct ext4_ext_path *path2 = NULL;
5586 	int replaced_count = 0;
5587 
5588 	BUG_ON(!rwsem_is_locked(&EXT4_I(inode1)->i_data_sem));
5589 	BUG_ON(!rwsem_is_locked(&EXT4_I(inode2)->i_data_sem));
5590 	BUG_ON(!inode_is_locked(inode1));
5591 	BUG_ON(!inode_is_locked(inode2));
5592 
5593 	*erp = ext4_es_remove_extent(inode1, lblk1, count);
5594 	if (unlikely(*erp))
5595 		return 0;
5596 	*erp = ext4_es_remove_extent(inode2, lblk2, count);
5597 	if (unlikely(*erp))
5598 		return 0;
5599 
5600 	while (count) {
5601 		struct ext4_extent *ex1, *ex2, tmp_ex;
5602 		ext4_lblk_t e1_blk, e2_blk;
5603 		int e1_len, e2_len, len;
5604 		int split = 0;
5605 
5606 		path1 = ext4_find_extent(inode1, lblk1, NULL, EXT4_EX_NOCACHE);
5607 		if (IS_ERR(path1)) {
5608 			*erp = PTR_ERR(path1);
5609 			path1 = NULL;
5610 		finish:
5611 			count = 0;
5612 			goto repeat;
5613 		}
5614 		path2 = ext4_find_extent(inode2, lblk2, NULL, EXT4_EX_NOCACHE);
5615 		if (IS_ERR(path2)) {
5616 			*erp = PTR_ERR(path2);
5617 			path2 = NULL;
5618 			goto finish;
5619 		}
5620 		ex1 = path1[path1->p_depth].p_ext;
5621 		ex2 = path2[path2->p_depth].p_ext;
5622 		/* Do we have something to swap ? */
5623 		if (unlikely(!ex2 || !ex1))
5624 			goto finish;
5625 
5626 		e1_blk = le32_to_cpu(ex1->ee_block);
5627 		e2_blk = le32_to_cpu(ex2->ee_block);
5628 		e1_len = ext4_ext_get_actual_len(ex1);
5629 		e2_len = ext4_ext_get_actual_len(ex2);
5630 
5631 		/* Hole handling */
5632 		if (!in_range(lblk1, e1_blk, e1_len) ||
5633 		    !in_range(lblk2, e2_blk, e2_len)) {
5634 			ext4_lblk_t next1, next2;
5635 
5636 			/* if hole after extent, then go to next extent */
5637 			next1 = ext4_ext_next_allocated_block(path1);
5638 			next2 = ext4_ext_next_allocated_block(path2);
5639 			/* If hole before extent, then shift to that extent */
5640 			if (e1_blk > lblk1)
5641 				next1 = e1_blk;
5642 			if (e2_blk > lblk2)
5643 				next2 = e2_blk;
5644 			/* Do we have something to swap */
5645 			if (next1 == EXT_MAX_BLOCKS || next2 == EXT_MAX_BLOCKS)
5646 				goto finish;
5647 			/* Move to the rightest boundary */
5648 			len = next1 - lblk1;
5649 			if (len < next2 - lblk2)
5650 				len = next2 - lblk2;
5651 			if (len > count)
5652 				len = count;
5653 			lblk1 += len;
5654 			lblk2 += len;
5655 			count -= len;
5656 			goto repeat;
5657 		}
5658 
5659 		/* Prepare left boundary */
5660 		if (e1_blk < lblk1) {
5661 			split = 1;
5662 			*erp = ext4_force_split_extent_at(handle, inode1,
5663 						&path1, lblk1, 0);
5664 			if (unlikely(*erp))
5665 				goto finish;
5666 		}
5667 		if (e2_blk < lblk2) {
5668 			split = 1;
5669 			*erp = ext4_force_split_extent_at(handle, inode2,
5670 						&path2,  lblk2, 0);
5671 			if (unlikely(*erp))
5672 				goto finish;
5673 		}
5674 		/* ext4_split_extent_at() may result in leaf extent split,
5675 		 * path must to be revalidated. */
5676 		if (split)
5677 			goto repeat;
5678 
5679 		/* Prepare right boundary */
5680 		len = count;
5681 		if (len > e1_blk + e1_len - lblk1)
5682 			len = e1_blk + e1_len - lblk1;
5683 		if (len > e2_blk + e2_len - lblk2)
5684 			len = e2_blk + e2_len - lblk2;
5685 
5686 		if (len != e1_len) {
5687 			split = 1;
5688 			*erp = ext4_force_split_extent_at(handle, inode1,
5689 						&path1, lblk1 + len, 0);
5690 			if (unlikely(*erp))
5691 				goto finish;
5692 		}
5693 		if (len != e2_len) {
5694 			split = 1;
5695 			*erp = ext4_force_split_extent_at(handle, inode2,
5696 						&path2, lblk2 + len, 0);
5697 			if (*erp)
5698 				goto finish;
5699 		}
5700 		/* ext4_split_extent_at() may result in leaf extent split,
5701 		 * path must to be revalidated. */
5702 		if (split)
5703 			goto repeat;
5704 
5705 		BUG_ON(e2_len != e1_len);
5706 		*erp = ext4_ext_get_access(handle, inode1, path1 + path1->p_depth);
5707 		if (unlikely(*erp))
5708 			goto finish;
5709 		*erp = ext4_ext_get_access(handle, inode2, path2 + path2->p_depth);
5710 		if (unlikely(*erp))
5711 			goto finish;
5712 
5713 		/* Both extents are fully inside boundaries. Swap it now */
5714 		tmp_ex = *ex1;
5715 		ext4_ext_store_pblock(ex1, ext4_ext_pblock(ex2));
5716 		ext4_ext_store_pblock(ex2, ext4_ext_pblock(&tmp_ex));
5717 		ex1->ee_len = cpu_to_le16(e2_len);
5718 		ex2->ee_len = cpu_to_le16(e1_len);
5719 		if (unwritten)
5720 			ext4_ext_mark_unwritten(ex2);
5721 		if (ext4_ext_is_unwritten(&tmp_ex))
5722 			ext4_ext_mark_unwritten(ex1);
5723 
5724 		ext4_ext_try_to_merge(handle, inode2, path2, ex2);
5725 		ext4_ext_try_to_merge(handle, inode1, path1, ex1);
5726 		*erp = ext4_ext_dirty(handle, inode2, path2 +
5727 				      path2->p_depth);
5728 		if (unlikely(*erp))
5729 			goto finish;
5730 		*erp = ext4_ext_dirty(handle, inode1, path1 +
5731 				      path1->p_depth);
5732 		/*
5733 		 * Looks scarry ah..? second inode already points to new blocks,
5734 		 * and it was successfully dirtied. But luckily error may happen
5735 		 * only due to journal error, so full transaction will be
5736 		 * aborted anyway.
5737 		 */
5738 		if (unlikely(*erp))
5739 			goto finish;
5740 		lblk1 += len;
5741 		lblk2 += len;
5742 		replaced_count += len;
5743 		count -= len;
5744 
5745 	repeat:
5746 		ext4_ext_drop_refs(path1);
5747 		kfree(path1);
5748 		ext4_ext_drop_refs(path2);
5749 		kfree(path2);
5750 		path1 = path2 = NULL;
5751 	}
5752 	return replaced_count;
5753 }
5754 
5755 /*
5756  * ext4_clu_mapped - determine whether any block in a logical cluster has
5757  *                   been mapped to a physical cluster
5758  *
5759  * @inode - file containing the logical cluster
5760  * @lclu - logical cluster of interest
5761  *
5762  * Returns 1 if any block in the logical cluster is mapped, signifying
5763  * that a physical cluster has been allocated for it.  Otherwise,
5764  * returns 0.  Can also return negative error codes.  Derived from
5765  * ext4_ext_map_blocks().
5766  */
5767 int ext4_clu_mapped(struct inode *inode, ext4_lblk_t lclu)
5768 {
5769 	struct ext4_sb_info *sbi = EXT4_SB(inode->i_sb);
5770 	struct ext4_ext_path *path;
5771 	int depth, mapped = 0, err = 0;
5772 	struct ext4_extent *extent;
5773 	ext4_lblk_t first_lblk, first_lclu, last_lclu;
5774 
5775 	/* search for the extent closest to the first block in the cluster */
5776 	path = ext4_find_extent(inode, EXT4_C2B(sbi, lclu), NULL, 0);
5777 	if (IS_ERR(path)) {
5778 		err = PTR_ERR(path);
5779 		path = NULL;
5780 		goto out;
5781 	}
5782 
5783 	depth = ext_depth(inode);
5784 
5785 	/*
5786 	 * A consistent leaf must not be empty.  This situation is possible,
5787 	 * though, _during_ tree modification, and it's why an assert can't
5788 	 * be put in ext4_find_extent().
5789 	 */
5790 	if (unlikely(path[depth].p_ext == NULL && depth != 0)) {
5791 		EXT4_ERROR_INODE(inode,
5792 		    "bad extent address - lblock: %lu, depth: %d, pblock: %lld",
5793 				 (unsigned long) EXT4_C2B(sbi, lclu),
5794 				 depth, path[depth].p_block);
5795 		err = -EFSCORRUPTED;
5796 		goto out;
5797 	}
5798 
5799 	extent = path[depth].p_ext;
5800 
5801 	/* can't be mapped if the extent tree is empty */
5802 	if (extent == NULL)
5803 		goto out;
5804 
5805 	first_lblk = le32_to_cpu(extent->ee_block);
5806 	first_lclu = EXT4_B2C(sbi, first_lblk);
5807 
5808 	/*
5809 	 * Three possible outcomes at this point - found extent spanning
5810 	 * the target cluster, to the left of the target cluster, or to the
5811 	 * right of the target cluster.  The first two cases are handled here.
5812 	 * The last case indicates the target cluster is not mapped.
5813 	 */
5814 	if (lclu >= first_lclu) {
5815 		last_lclu = EXT4_B2C(sbi, first_lblk +
5816 				     ext4_ext_get_actual_len(extent) - 1);
5817 		if (lclu <= last_lclu) {
5818 			mapped = 1;
5819 		} else {
5820 			first_lblk = ext4_ext_next_allocated_block(path);
5821 			first_lclu = EXT4_B2C(sbi, first_lblk);
5822 			if (lclu == first_lclu)
5823 				mapped = 1;
5824 		}
5825 	}
5826 
5827 out:
5828 	ext4_ext_drop_refs(path);
5829 	kfree(path);
5830 
5831 	return err ? err : mapped;
5832 }
5833 
5834 /*
5835  * Updates physical block address and unwritten status of extent
5836  * starting at lblk start and of len. If such an extent doesn't exist,
5837  * this function splits the extent tree appropriately to create an
5838  * extent like this.  This function is called in the fast commit
5839  * replay path.  Returns 0 on success and error on failure.
5840  */
5841 int ext4_ext_replay_update_ex(struct inode *inode, ext4_lblk_t start,
5842 			      int len, int unwritten, ext4_fsblk_t pblk)
5843 {
5844 	struct ext4_ext_path *path = NULL, *ppath;
5845 	struct ext4_extent *ex;
5846 	int ret;
5847 
5848 	path = ext4_find_extent(inode, start, NULL, 0);
5849 	if (IS_ERR(path))
5850 		return PTR_ERR(path);
5851 	ex = path[path->p_depth].p_ext;
5852 	if (!ex) {
5853 		ret = -EFSCORRUPTED;
5854 		goto out;
5855 	}
5856 
5857 	if (le32_to_cpu(ex->ee_block) != start ||
5858 		ext4_ext_get_actual_len(ex) != len) {
5859 		/* We need to split this extent to match our extent first */
5860 		ppath = path;
5861 		down_write(&EXT4_I(inode)->i_data_sem);
5862 		ret = ext4_force_split_extent_at(NULL, inode, &ppath, start, 1);
5863 		up_write(&EXT4_I(inode)->i_data_sem);
5864 		if (ret)
5865 			goto out;
5866 		kfree(path);
5867 		path = ext4_find_extent(inode, start, NULL, 0);
5868 		if (IS_ERR(path))
5869 			return -1;
5870 		ppath = path;
5871 		ex = path[path->p_depth].p_ext;
5872 		WARN_ON(le32_to_cpu(ex->ee_block) != start);
5873 		if (ext4_ext_get_actual_len(ex) != len) {
5874 			down_write(&EXT4_I(inode)->i_data_sem);
5875 			ret = ext4_force_split_extent_at(NULL, inode, &ppath,
5876 							 start + len, 1);
5877 			up_write(&EXT4_I(inode)->i_data_sem);
5878 			if (ret)
5879 				goto out;
5880 			kfree(path);
5881 			path = ext4_find_extent(inode, start, NULL, 0);
5882 			if (IS_ERR(path))
5883 				return -EINVAL;
5884 			ex = path[path->p_depth].p_ext;
5885 		}
5886 	}
5887 	if (unwritten)
5888 		ext4_ext_mark_unwritten(ex);
5889 	else
5890 		ext4_ext_mark_initialized(ex);
5891 	ext4_ext_store_pblock(ex, pblk);
5892 	down_write(&EXT4_I(inode)->i_data_sem);
5893 	ret = ext4_ext_dirty(NULL, inode, &path[path->p_depth]);
5894 	up_write(&EXT4_I(inode)->i_data_sem);
5895 out:
5896 	ext4_ext_drop_refs(path);
5897 	kfree(path);
5898 	ext4_mark_inode_dirty(NULL, inode);
5899 	return ret;
5900 }
5901 
5902 /* Try to shrink the extent tree */
5903 void ext4_ext_replay_shrink_inode(struct inode *inode, ext4_lblk_t end)
5904 {
5905 	struct ext4_ext_path *path = NULL;
5906 	struct ext4_extent *ex;
5907 	ext4_lblk_t old_cur, cur = 0;
5908 
5909 	while (cur < end) {
5910 		path = ext4_find_extent(inode, cur, NULL, 0);
5911 		if (IS_ERR(path))
5912 			return;
5913 		ex = path[path->p_depth].p_ext;
5914 		if (!ex) {
5915 			ext4_ext_drop_refs(path);
5916 			kfree(path);
5917 			ext4_mark_inode_dirty(NULL, inode);
5918 			return;
5919 		}
5920 		old_cur = cur;
5921 		cur = le32_to_cpu(ex->ee_block) + ext4_ext_get_actual_len(ex);
5922 		if (cur <= old_cur)
5923 			cur = old_cur + 1;
5924 		ext4_ext_try_to_merge(NULL, inode, path, ex);
5925 		down_write(&EXT4_I(inode)->i_data_sem);
5926 		ext4_ext_dirty(NULL, inode, &path[path->p_depth]);
5927 		up_write(&EXT4_I(inode)->i_data_sem);
5928 		ext4_mark_inode_dirty(NULL, inode);
5929 		ext4_ext_drop_refs(path);
5930 		kfree(path);
5931 	}
5932 }
5933 
5934 /* Check if *cur is a hole and if it is, skip it */
5935 static int skip_hole(struct inode *inode, ext4_lblk_t *cur)
5936 {
5937 	int ret;
5938 	struct ext4_map_blocks map;
5939 
5940 	map.m_lblk = *cur;
5941 	map.m_len = ((inode->i_size) >> inode->i_sb->s_blocksize_bits) - *cur;
5942 
5943 	ret = ext4_map_blocks(NULL, inode, &map, 0);
5944 	if (ret < 0)
5945 		return ret;
5946 	if (ret != 0)
5947 		return 0;
5948 	*cur = *cur + map.m_len;
5949 	return 0;
5950 }
5951 
5952 /* Count number of blocks used by this inode and update i_blocks */
5953 int ext4_ext_replay_set_iblocks(struct inode *inode)
5954 {
5955 	struct ext4_ext_path *path = NULL, *path2 = NULL;
5956 	struct ext4_extent *ex;
5957 	ext4_lblk_t cur = 0, end;
5958 	int numblks = 0, i, ret = 0;
5959 	ext4_fsblk_t cmp1, cmp2;
5960 	struct ext4_map_blocks map;
5961 
5962 	/* Determin the size of the file first */
5963 	path = ext4_find_extent(inode, EXT_MAX_BLOCKS - 1, NULL,
5964 					EXT4_EX_NOCACHE);
5965 	if (IS_ERR(path))
5966 		return PTR_ERR(path);
5967 	ex = path[path->p_depth].p_ext;
5968 	if (!ex) {
5969 		ext4_ext_drop_refs(path);
5970 		kfree(path);
5971 		goto out;
5972 	}
5973 	end = le32_to_cpu(ex->ee_block) + ext4_ext_get_actual_len(ex);
5974 	ext4_ext_drop_refs(path);
5975 	kfree(path);
5976 
5977 	/* Count the number of data blocks */
5978 	cur = 0;
5979 	while (cur < end) {
5980 		map.m_lblk = cur;
5981 		map.m_len = end - cur;
5982 		ret = ext4_map_blocks(NULL, inode, &map, 0);
5983 		if (ret < 0)
5984 			break;
5985 		if (ret > 0)
5986 			numblks += ret;
5987 		cur = cur + map.m_len;
5988 	}
5989 
5990 	/*
5991 	 * Count the number of extent tree blocks. We do it by looking up
5992 	 * two successive extents and determining the difference between
5993 	 * their paths. When path is different for 2 successive extents
5994 	 * we compare the blocks in the path at each level and increment
5995 	 * iblocks by total number of differences found.
5996 	 */
5997 	cur = 0;
5998 	ret = skip_hole(inode, &cur);
5999 	if (ret < 0)
6000 		goto out;
6001 	path = ext4_find_extent(inode, cur, NULL, 0);
6002 	if (IS_ERR(path))
6003 		goto out;
6004 	numblks += path->p_depth;
6005 	ext4_ext_drop_refs(path);
6006 	kfree(path);
6007 	while (cur < end) {
6008 		path = ext4_find_extent(inode, cur, NULL, 0);
6009 		if (IS_ERR(path))
6010 			break;
6011 		ex = path[path->p_depth].p_ext;
6012 		if (!ex) {
6013 			ext4_ext_drop_refs(path);
6014 			kfree(path);
6015 			return 0;
6016 		}
6017 		cur = max(cur + 1, le32_to_cpu(ex->ee_block) +
6018 					ext4_ext_get_actual_len(ex));
6019 		ret = skip_hole(inode, &cur);
6020 		if (ret < 0) {
6021 			ext4_ext_drop_refs(path);
6022 			kfree(path);
6023 			break;
6024 		}
6025 		path2 = ext4_find_extent(inode, cur, NULL, 0);
6026 		if (IS_ERR(path2)) {
6027 			ext4_ext_drop_refs(path);
6028 			kfree(path);
6029 			break;
6030 		}
6031 		for (i = 0; i <= max(path->p_depth, path2->p_depth); i++) {
6032 			cmp1 = cmp2 = 0;
6033 			if (i <= path->p_depth)
6034 				cmp1 = path[i].p_bh ?
6035 					path[i].p_bh->b_blocknr : 0;
6036 			if (i <= path2->p_depth)
6037 				cmp2 = path2[i].p_bh ?
6038 					path2[i].p_bh->b_blocknr : 0;
6039 			if (cmp1 != cmp2 && cmp2 != 0)
6040 				numblks++;
6041 		}
6042 		ext4_ext_drop_refs(path);
6043 		ext4_ext_drop_refs(path2);
6044 		kfree(path);
6045 		kfree(path2);
6046 	}
6047 
6048 out:
6049 	inode->i_blocks = numblks << (inode->i_sb->s_blocksize_bits - 9);
6050 	ext4_mark_inode_dirty(NULL, inode);
6051 	return 0;
6052 }
6053 
6054 int ext4_ext_clear_bb(struct inode *inode)
6055 {
6056 	struct ext4_ext_path *path = NULL;
6057 	struct ext4_extent *ex;
6058 	ext4_lblk_t cur = 0, end;
6059 	int j, ret = 0;
6060 	struct ext4_map_blocks map;
6061 
6062 	if (ext4_test_inode_flag(inode, EXT4_INODE_INLINE_DATA))
6063 		return 0;
6064 
6065 	/* Determin the size of the file first */
6066 	path = ext4_find_extent(inode, EXT_MAX_BLOCKS - 1, NULL,
6067 					EXT4_EX_NOCACHE);
6068 	if (IS_ERR(path))
6069 		return PTR_ERR(path);
6070 	ex = path[path->p_depth].p_ext;
6071 	if (!ex) {
6072 		ext4_ext_drop_refs(path);
6073 		kfree(path);
6074 		return 0;
6075 	}
6076 	end = le32_to_cpu(ex->ee_block) + ext4_ext_get_actual_len(ex);
6077 	ext4_ext_drop_refs(path);
6078 	kfree(path);
6079 
6080 	cur = 0;
6081 	while (cur < end) {
6082 		map.m_lblk = cur;
6083 		map.m_len = end - cur;
6084 		ret = ext4_map_blocks(NULL, inode, &map, 0);
6085 		if (ret < 0)
6086 			break;
6087 		if (ret > 0) {
6088 			path = ext4_find_extent(inode, map.m_lblk, NULL, 0);
6089 			if (!IS_ERR_OR_NULL(path)) {
6090 				for (j = 0; j < path->p_depth; j++) {
6091 
6092 					ext4_mb_mark_bb(inode->i_sb,
6093 							path[j].p_block, 1, 0);
6094 					ext4_fc_record_regions(inode->i_sb, inode->i_ino,
6095 							0, path[j].p_block, 1, 1);
6096 				}
6097 				ext4_ext_drop_refs(path);
6098 				kfree(path);
6099 			}
6100 			ext4_mb_mark_bb(inode->i_sb, map.m_pblk, map.m_len, 0);
6101 			ext4_fc_record_regions(inode->i_sb, inode->i_ino,
6102 					map.m_lblk, map.m_pblk, map.m_len, 1);
6103 		}
6104 		cur = cur + map.m_len;
6105 	}
6106 
6107 	return 0;
6108 }
6109