1 // SPDX-License-Identifier: GPL-2.0
3 * Copyright (c) 2003-2006, Cluster File Systems, Inc, info@clusterfs.com
4 * Written by Alex Tomas <alex@clusterfs.com>
6 * Architecture independence:
7 * Copyright (c) 2005, Bull S.A.
8 * Written by Pierre Peiffer <pierre.peiffer@bull.net>
12 * Extents support for EXT4
15 * - ext4*_error() should be used in some situations
16 * - analyze all BUG()/BUG_ON(), use -EIO where appropriate
17 * - smart tree reduction
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/backing-dev.h>
31 #include "ext4_jbd2.h"
32 #include "ext4_extents.h"
35 #include <trace/events/ext4.h>
38 * used by extent splitting.
40 #define EXT4_EXT_MAY_ZEROOUT 0x1 /* safe to zeroout if split fails \
42 #define EXT4_EXT_MARK_UNWRIT1 0x2 /* mark first half unwritten */
43 #define EXT4_EXT_MARK_UNWRIT2 0x4 /* mark second half unwritten */
45 #define EXT4_EXT_DATA_VALID1 0x8 /* first half contains valid data */
46 #define EXT4_EXT_DATA_VALID2 0x10 /* second half contains valid data */
48 static __le32 ext4_extent_block_csum(struct inode *inode,
49 struct ext4_extent_header *eh)
51 struct ext4_inode_info *ei = EXT4_I(inode);
52 struct ext4_sb_info *sbi = EXT4_SB(inode->i_sb);
55 csum = ext4_chksum(sbi, ei->i_csum_seed, (__u8 *)eh,
56 EXT4_EXTENT_TAIL_OFFSET(eh));
57 return cpu_to_le32(csum);
60 static int ext4_extent_block_csum_verify(struct inode *inode,
61 struct ext4_extent_header *eh)
63 struct ext4_extent_tail *et;
65 if (!ext4_has_metadata_csum(inode->i_sb))
68 et = find_ext4_extent_tail(eh);
69 if (et->et_checksum != ext4_extent_block_csum(inode, eh))
74 static void ext4_extent_block_csum_set(struct inode *inode,
75 struct ext4_extent_header *eh)
77 struct ext4_extent_tail *et;
79 if (!ext4_has_metadata_csum(inode->i_sb))
82 et = find_ext4_extent_tail(eh);
83 et->et_checksum = ext4_extent_block_csum(inode, eh);
86 static int ext4_split_extent(handle_t *handle,
88 struct ext4_ext_path **ppath,
89 struct ext4_map_blocks *map,
93 static int ext4_split_extent_at(handle_t *handle,
95 struct ext4_ext_path **ppath,
100 static int ext4_find_delayed_extent(struct inode *inode,
101 struct extent_status *newes);
103 static int ext4_ext_trunc_restart_fn(struct inode *inode, int *dropped)
106 * Drop i_data_sem to avoid deadlock with ext4_map_blocks. At this
107 * moment, get_block can be called only for blocks inside i_size since
108 * page cache has been already dropped and writes are blocked by
109 * i_mutex. So we can safely drop the i_data_sem here.
111 BUG_ON(EXT4_JOURNAL(inode) == NULL);
112 ext4_discard_preallocations(inode);
113 up_write(&EXT4_I(inode)->i_data_sem);
119 * Make sure 'handle' has at least 'check_cred' credits. If not, restart
120 * transaction with 'restart_cred' credits. The function drops i_data_sem
121 * when restarting transaction and gets it after transaction is restarted.
123 * The function returns 0 on success, 1 if transaction had to be restarted,
124 * and < 0 in case of fatal error.
126 int ext4_datasem_ensure_credits(handle_t *handle, struct inode *inode,
127 int check_cred, int restart_cred,
133 ret = ext4_journal_ensure_credits_fn(handle, check_cred, restart_cred,
134 revoke_cred, ext4_ext_trunc_restart_fn(inode, &dropped));
136 down_write(&EXT4_I(inode)->i_data_sem);
145 static int ext4_ext_get_access(handle_t *handle, struct inode *inode,
146 struct ext4_ext_path *path)
149 /* path points to block */
150 BUFFER_TRACE(path->p_bh, "get_write_access");
151 return ext4_journal_get_write_access(handle, path->p_bh);
153 /* path points to leaf/index in inode body */
154 /* we use in-core data, no need to protect them */
164 static int __ext4_ext_dirty(const char *where, unsigned int line,
165 handle_t *handle, struct inode *inode,
166 struct ext4_ext_path *path)
170 WARN_ON(!rwsem_is_locked(&EXT4_I(inode)->i_data_sem));
172 ext4_extent_block_csum_set(inode, ext_block_hdr(path->p_bh));
173 /* path points to block */
174 err = __ext4_handle_dirty_metadata(where, line, handle,
177 /* path points to leaf/index in inode body */
178 err = ext4_mark_inode_dirty(handle, inode);
183 #define ext4_ext_dirty(handle, inode, path) \
184 __ext4_ext_dirty(__func__, __LINE__, (handle), (inode), (path))
186 static ext4_fsblk_t ext4_ext_find_goal(struct inode *inode,
187 struct ext4_ext_path *path,
191 int depth = path->p_depth;
192 struct ext4_extent *ex;
195 * Try to predict block placement assuming that we are
196 * filling in a file which will eventually be
197 * non-sparse --- i.e., in the case of libbfd writing
198 * an ELF object sections out-of-order but in a way
199 * the eventually results in a contiguous object or
200 * executable file, or some database extending a table
201 * space file. However, this is actually somewhat
202 * non-ideal if we are writing a sparse file such as
203 * qemu or KVM writing a raw image file that is going
204 * to stay fairly sparse, since it will end up
205 * fragmenting the file system's free space. Maybe we
206 * should have some hueristics or some way to allow
207 * userspace to pass a hint to file system,
208 * especially if the latter case turns out to be
211 ex = path[depth].p_ext;
213 ext4_fsblk_t ext_pblk = ext4_ext_pblock(ex);
214 ext4_lblk_t ext_block = le32_to_cpu(ex->ee_block);
216 if (block > ext_block)
217 return ext_pblk + (block - ext_block);
219 return ext_pblk - (ext_block - block);
222 /* it looks like index is empty;
223 * try to find starting block from index itself */
224 if (path[depth].p_bh)
225 return path[depth].p_bh->b_blocknr;
228 /* OK. use inode's group */
229 return ext4_inode_to_goal_block(inode);
233 * Allocation for a meta data block
236 ext4_ext_new_meta_block(handle_t *handle, struct inode *inode,
237 struct ext4_ext_path *path,
238 struct ext4_extent *ex, int *err, unsigned int flags)
240 ext4_fsblk_t goal, newblock;
242 goal = ext4_ext_find_goal(inode, path, le32_to_cpu(ex->ee_block));
243 newblock = ext4_new_meta_blocks(handle, inode, goal, flags,
248 static inline int ext4_ext_space_block(struct inode *inode, int check)
252 size = (inode->i_sb->s_blocksize - sizeof(struct ext4_extent_header))
253 / sizeof(struct ext4_extent);
254 #ifdef AGGRESSIVE_TEST
255 if (!check && size > 6)
261 static inline int ext4_ext_space_block_idx(struct inode *inode, int check)
265 size = (inode->i_sb->s_blocksize - sizeof(struct ext4_extent_header))
266 / sizeof(struct ext4_extent_idx);
267 #ifdef AGGRESSIVE_TEST
268 if (!check && size > 5)
274 static inline int ext4_ext_space_root(struct inode *inode, int check)
278 size = sizeof(EXT4_I(inode)->i_data);
279 size -= sizeof(struct ext4_extent_header);
280 size /= sizeof(struct ext4_extent);
281 #ifdef AGGRESSIVE_TEST
282 if (!check && size > 3)
288 static inline int ext4_ext_space_root_idx(struct inode *inode, int check)
292 size = sizeof(EXT4_I(inode)->i_data);
293 size -= sizeof(struct ext4_extent_header);
294 size /= sizeof(struct ext4_extent_idx);
295 #ifdef AGGRESSIVE_TEST
296 if (!check && size > 4)
303 ext4_force_split_extent_at(handle_t *handle, struct inode *inode,
304 struct ext4_ext_path **ppath, ext4_lblk_t lblk,
307 struct ext4_ext_path *path = *ppath;
308 int unwritten = ext4_ext_is_unwritten(path[path->p_depth].p_ext);
310 return ext4_split_extent_at(handle, inode, ppath, lblk, unwritten ?
311 EXT4_EXT_MARK_UNWRIT1|EXT4_EXT_MARK_UNWRIT2 : 0,
312 EXT4_EX_NOCACHE | EXT4_GET_BLOCKS_PRE_IO |
313 (nofail ? EXT4_GET_BLOCKS_METADATA_NOFAIL:0));
317 ext4_ext_max_entries(struct inode *inode, int depth)
321 if (depth == ext_depth(inode)) {
323 max = ext4_ext_space_root(inode, 1);
325 max = ext4_ext_space_root_idx(inode, 1);
328 max = ext4_ext_space_block(inode, 1);
330 max = ext4_ext_space_block_idx(inode, 1);
336 static int ext4_valid_extent(struct inode *inode, struct ext4_extent *ext)
338 ext4_fsblk_t block = ext4_ext_pblock(ext);
339 int len = ext4_ext_get_actual_len(ext);
340 ext4_lblk_t lblock = le32_to_cpu(ext->ee_block);
345 * - overflow/wrap-around
347 if (lblock + len <= lblock)
349 return ext4_data_block_valid(EXT4_SB(inode->i_sb), block, len);
352 static int ext4_valid_extent_idx(struct inode *inode,
353 struct ext4_extent_idx *ext_idx)
355 ext4_fsblk_t block = ext4_idx_pblock(ext_idx);
357 return ext4_data_block_valid(EXT4_SB(inode->i_sb), block, 1);
360 static int ext4_valid_extent_entries(struct inode *inode,
361 struct ext4_extent_header *eh,
364 unsigned short entries;
365 if (eh->eh_entries == 0)
368 entries = le16_to_cpu(eh->eh_entries);
372 struct ext4_extent *ext = EXT_FIRST_EXTENT(eh);
373 struct ext4_super_block *es = EXT4_SB(inode->i_sb)->s_es;
374 ext4_fsblk_t pblock = 0;
375 ext4_lblk_t lblock = 0;
376 ext4_lblk_t prev = 0;
379 if (!ext4_valid_extent(inode, ext))
382 /* Check for overlapping extents */
383 lblock = le32_to_cpu(ext->ee_block);
384 len = ext4_ext_get_actual_len(ext);
385 if ((lblock <= prev) && prev) {
386 pblock = ext4_ext_pblock(ext);
387 es->s_last_error_block = cpu_to_le64(pblock);
392 prev = lblock + len - 1;
395 struct ext4_extent_idx *ext_idx = EXT_FIRST_INDEX(eh);
397 if (!ext4_valid_extent_idx(inode, ext_idx))
406 static int __ext4_ext_check(const char *function, unsigned int line,
407 struct inode *inode, struct ext4_extent_header *eh,
408 int depth, ext4_fsblk_t pblk)
410 const char *error_msg;
411 int max = 0, err = -EFSCORRUPTED;
413 if (unlikely(eh->eh_magic != EXT4_EXT_MAGIC)) {
414 error_msg = "invalid magic";
417 if (unlikely(le16_to_cpu(eh->eh_depth) != depth)) {
418 error_msg = "unexpected eh_depth";
421 if (unlikely(eh->eh_max == 0)) {
422 error_msg = "invalid eh_max";
425 max = ext4_ext_max_entries(inode, depth);
426 if (unlikely(le16_to_cpu(eh->eh_max) > max)) {
427 error_msg = "too large eh_max";
430 if (unlikely(le16_to_cpu(eh->eh_entries) > le16_to_cpu(eh->eh_max))) {
431 error_msg = "invalid eh_entries";
434 if (!ext4_valid_extent_entries(inode, eh, depth)) {
435 error_msg = "invalid extent entries";
438 if (unlikely(depth > 32)) {
439 error_msg = "too large eh_depth";
442 /* Verify checksum on non-root extent tree nodes */
443 if (ext_depth(inode) != depth &&
444 !ext4_extent_block_csum_verify(inode, eh)) {
445 error_msg = "extent tree corrupted";
452 ext4_set_errno(inode->i_sb, -err);
453 ext4_error_inode(inode, function, line, 0,
454 "pblk %llu bad header/extent: %s - magic %x, "
455 "entries %u, max %u(%u), depth %u(%u)",
456 (unsigned long long) pblk, error_msg,
457 le16_to_cpu(eh->eh_magic),
458 le16_to_cpu(eh->eh_entries), le16_to_cpu(eh->eh_max),
459 max, le16_to_cpu(eh->eh_depth), depth);
463 #define ext4_ext_check(inode, eh, depth, pblk) \
464 __ext4_ext_check(__func__, __LINE__, (inode), (eh), (depth), (pblk))
466 int ext4_ext_check_inode(struct inode *inode)
468 return ext4_ext_check(inode, ext_inode_hdr(inode), ext_depth(inode), 0);
471 static void ext4_cache_extents(struct inode *inode,
472 struct ext4_extent_header *eh)
474 struct ext4_extent *ex = EXT_FIRST_EXTENT(eh);
475 ext4_lblk_t prev = 0;
478 for (i = le16_to_cpu(eh->eh_entries); i > 0; i--, ex++) {
479 unsigned int status = EXTENT_STATUS_WRITTEN;
480 ext4_lblk_t lblk = le32_to_cpu(ex->ee_block);
481 int len = ext4_ext_get_actual_len(ex);
483 if (prev && (prev != lblk))
484 ext4_es_cache_extent(inode, prev, lblk - prev, ~0,
487 if (ext4_ext_is_unwritten(ex))
488 status = EXTENT_STATUS_UNWRITTEN;
489 ext4_es_cache_extent(inode, lblk, len,
490 ext4_ext_pblock(ex), status);
495 static struct buffer_head *
496 __read_extent_tree_block(const char *function, unsigned int line,
497 struct inode *inode, ext4_fsblk_t pblk, int depth,
500 struct buffer_head *bh;
503 bh = sb_getblk_gfp(inode->i_sb, pblk, __GFP_MOVABLE | GFP_NOFS);
505 return ERR_PTR(-ENOMEM);
507 if (!bh_uptodate_or_lock(bh)) {
508 trace_ext4_ext_load_extent(inode, pblk, _RET_IP_);
509 err = bh_submit_read(bh);
513 if (buffer_verified(bh) && !(flags & EXT4_EX_FORCE_CACHE))
515 if (!ext4_has_feature_journal(inode->i_sb) ||
517 le32_to_cpu(EXT4_SB(inode->i_sb)->s_es->s_journal_inum))) {
518 err = __ext4_ext_check(function, line, inode,
519 ext_block_hdr(bh), depth, pblk);
523 set_buffer_verified(bh);
525 * If this is a leaf block, cache all of its entries
527 if (!(flags & EXT4_EX_NOCACHE) && depth == 0) {
528 struct ext4_extent_header *eh = ext_block_hdr(bh);
529 ext4_cache_extents(inode, eh);
538 #define read_extent_tree_block(inode, pblk, depth, flags) \
539 __read_extent_tree_block(__func__, __LINE__, (inode), (pblk), \
543 * This function is called to cache a file's extent information in the
546 int ext4_ext_precache(struct inode *inode)
548 struct ext4_inode_info *ei = EXT4_I(inode);
549 struct ext4_ext_path *path = NULL;
550 struct buffer_head *bh;
551 int i = 0, depth, ret = 0;
553 if (!ext4_test_inode_flag(inode, EXT4_INODE_EXTENTS))
554 return 0; /* not an extent-mapped inode */
556 down_read(&ei->i_data_sem);
557 depth = ext_depth(inode);
559 path = kcalloc(depth + 1, sizeof(struct ext4_ext_path),
562 up_read(&ei->i_data_sem);
566 /* Don't cache anything if there are no external extent blocks */
569 path[0].p_hdr = ext_inode_hdr(inode);
570 ret = ext4_ext_check(inode, path[0].p_hdr, depth, 0);
573 path[0].p_idx = EXT_FIRST_INDEX(path[0].p_hdr);
576 * If this is a leaf block or we've reached the end of
577 * the index block, go up
580 path[i].p_idx > EXT_LAST_INDEX(path[i].p_hdr)) {
581 brelse(path[i].p_bh);
586 bh = read_extent_tree_block(inode,
587 ext4_idx_pblock(path[i].p_idx++),
589 EXT4_EX_FORCE_CACHE);
596 path[i].p_hdr = ext_block_hdr(bh);
597 path[i].p_idx = EXT_FIRST_INDEX(path[i].p_hdr);
599 ext4_set_inode_state(inode, EXT4_STATE_EXT_PRECACHED);
601 up_read(&ei->i_data_sem);
602 ext4_ext_drop_refs(path);
608 static void ext4_ext_show_path(struct inode *inode, struct ext4_ext_path *path)
610 int k, l = path->p_depth;
613 for (k = 0; k <= l; k++, path++) {
615 ext_debug(" %d->%llu",
616 le32_to_cpu(path->p_idx->ei_block),
617 ext4_idx_pblock(path->p_idx));
618 } else if (path->p_ext) {
619 ext_debug(" %d:[%d]%d:%llu ",
620 le32_to_cpu(path->p_ext->ee_block),
621 ext4_ext_is_unwritten(path->p_ext),
622 ext4_ext_get_actual_len(path->p_ext),
623 ext4_ext_pblock(path->p_ext));
630 static void ext4_ext_show_leaf(struct inode *inode, struct ext4_ext_path *path)
632 int depth = ext_depth(inode);
633 struct ext4_extent_header *eh;
634 struct ext4_extent *ex;
640 eh = path[depth].p_hdr;
641 ex = EXT_FIRST_EXTENT(eh);
643 ext_debug("Displaying leaf extents for inode %lu\n", inode->i_ino);
645 for (i = 0; i < le16_to_cpu(eh->eh_entries); i++, ex++) {
646 ext_debug("%d:[%d]%d:%llu ", le32_to_cpu(ex->ee_block),
647 ext4_ext_is_unwritten(ex),
648 ext4_ext_get_actual_len(ex), ext4_ext_pblock(ex));
653 static void ext4_ext_show_move(struct inode *inode, struct ext4_ext_path *path,
654 ext4_fsblk_t newblock, int level)
656 int depth = ext_depth(inode);
657 struct ext4_extent *ex;
659 if (depth != level) {
660 struct ext4_extent_idx *idx;
661 idx = path[level].p_idx;
662 while (idx <= EXT_MAX_INDEX(path[level].p_hdr)) {
663 ext_debug("%d: move %d:%llu in new index %llu\n", level,
664 le32_to_cpu(idx->ei_block),
665 ext4_idx_pblock(idx),
673 ex = path[depth].p_ext;
674 while (ex <= EXT_MAX_EXTENT(path[depth].p_hdr)) {
675 ext_debug("move %d:%llu:[%d]%d in new leaf %llu\n",
676 le32_to_cpu(ex->ee_block),
678 ext4_ext_is_unwritten(ex),
679 ext4_ext_get_actual_len(ex),
686 #define ext4_ext_show_path(inode, path)
687 #define ext4_ext_show_leaf(inode, path)
688 #define ext4_ext_show_move(inode, path, newblock, level)
691 void ext4_ext_drop_refs(struct ext4_ext_path *path)
697 depth = path->p_depth;
698 for (i = 0; i <= depth; i++, path++) {
707 * ext4_ext_binsearch_idx:
708 * binary search for the closest index of the given block
709 * the header must be checked before calling this
712 ext4_ext_binsearch_idx(struct inode *inode,
713 struct ext4_ext_path *path, ext4_lblk_t block)
715 struct ext4_extent_header *eh = path->p_hdr;
716 struct ext4_extent_idx *r, *l, *m;
719 ext_debug("binsearch for %u(idx): ", block);
721 l = EXT_FIRST_INDEX(eh) + 1;
722 r = EXT_LAST_INDEX(eh);
725 if (block < le32_to_cpu(m->ei_block))
729 ext_debug("%p(%u):%p(%u):%p(%u) ", l, le32_to_cpu(l->ei_block),
730 m, le32_to_cpu(m->ei_block),
731 r, le32_to_cpu(r->ei_block));
735 ext_debug(" -> %u->%lld ", le32_to_cpu(path->p_idx->ei_block),
736 ext4_idx_pblock(path->p_idx));
738 #ifdef CHECK_BINSEARCH
740 struct ext4_extent_idx *chix, *ix;
743 chix = ix = EXT_FIRST_INDEX(eh);
744 for (k = 0; k < le16_to_cpu(eh->eh_entries); k++, ix++) {
745 if (k != 0 && le32_to_cpu(ix->ei_block) <=
746 le32_to_cpu(ix[-1].ei_block)) {
747 printk(KERN_DEBUG "k=%d, ix=0x%p, "
749 ix, EXT_FIRST_INDEX(eh));
750 printk(KERN_DEBUG "%u <= %u\n",
751 le32_to_cpu(ix->ei_block),
752 le32_to_cpu(ix[-1].ei_block));
754 BUG_ON(k && le32_to_cpu(ix->ei_block)
755 <= le32_to_cpu(ix[-1].ei_block));
756 if (block < le32_to_cpu(ix->ei_block))
760 BUG_ON(chix != path->p_idx);
767 * ext4_ext_binsearch:
768 * binary search for closest extent of the given block
769 * the header must be checked before calling this
772 ext4_ext_binsearch(struct inode *inode,
773 struct ext4_ext_path *path, ext4_lblk_t block)
775 struct ext4_extent_header *eh = path->p_hdr;
776 struct ext4_extent *r, *l, *m;
778 if (eh->eh_entries == 0) {
780 * this leaf is empty:
781 * we get such a leaf in split/add case
786 ext_debug("binsearch for %u: ", block);
788 l = EXT_FIRST_EXTENT(eh) + 1;
789 r = EXT_LAST_EXTENT(eh);
793 if (block < le32_to_cpu(m->ee_block))
797 ext_debug("%p(%u):%p(%u):%p(%u) ", l, le32_to_cpu(l->ee_block),
798 m, le32_to_cpu(m->ee_block),
799 r, le32_to_cpu(r->ee_block));
803 ext_debug(" -> %d:%llu:[%d]%d ",
804 le32_to_cpu(path->p_ext->ee_block),
805 ext4_ext_pblock(path->p_ext),
806 ext4_ext_is_unwritten(path->p_ext),
807 ext4_ext_get_actual_len(path->p_ext));
809 #ifdef CHECK_BINSEARCH
811 struct ext4_extent *chex, *ex;
814 chex = ex = EXT_FIRST_EXTENT(eh);
815 for (k = 0; k < le16_to_cpu(eh->eh_entries); k++, ex++) {
816 BUG_ON(k && le32_to_cpu(ex->ee_block)
817 <= le32_to_cpu(ex[-1].ee_block));
818 if (block < le32_to_cpu(ex->ee_block))
822 BUG_ON(chex != path->p_ext);
828 int ext4_ext_tree_init(handle_t *handle, struct inode *inode)
830 struct ext4_extent_header *eh;
832 eh = ext_inode_hdr(inode);
835 eh->eh_magic = EXT4_EXT_MAGIC;
836 eh->eh_max = cpu_to_le16(ext4_ext_space_root(inode, 0));
837 ext4_mark_inode_dirty(handle, inode);
841 struct ext4_ext_path *
842 ext4_find_extent(struct inode *inode, ext4_lblk_t block,
843 struct ext4_ext_path **orig_path, int flags)
845 struct ext4_extent_header *eh;
846 struct buffer_head *bh;
847 struct ext4_ext_path *path = orig_path ? *orig_path : NULL;
848 short int depth, i, ppos = 0;
851 eh = ext_inode_hdr(inode);
852 depth = ext_depth(inode);
853 if (depth < 0 || depth > EXT4_MAX_EXTENT_DEPTH) {
854 EXT4_ERROR_INODE(inode, "inode has invalid extent depth: %d",
861 ext4_ext_drop_refs(path);
862 if (depth > path[0].p_maxdepth) {
864 *orig_path = path = NULL;
868 /* account possible depth increase */
869 path = kcalloc(depth + 2, sizeof(struct ext4_ext_path),
872 return ERR_PTR(-ENOMEM);
873 path[0].p_maxdepth = depth + 1;
879 if (!(flags & EXT4_EX_NOCACHE) && depth == 0)
880 ext4_cache_extents(inode, eh);
881 /* walk through the tree */
883 ext_debug("depth %d: num %d, max %d\n",
884 ppos, le16_to_cpu(eh->eh_entries), le16_to_cpu(eh->eh_max));
886 ext4_ext_binsearch_idx(inode, path + ppos, block);
887 path[ppos].p_block = ext4_idx_pblock(path[ppos].p_idx);
888 path[ppos].p_depth = i;
889 path[ppos].p_ext = NULL;
891 bh = read_extent_tree_block(inode, path[ppos].p_block, --i,
898 eh = ext_block_hdr(bh);
900 path[ppos].p_bh = bh;
901 path[ppos].p_hdr = eh;
904 path[ppos].p_depth = i;
905 path[ppos].p_ext = NULL;
906 path[ppos].p_idx = NULL;
909 ext4_ext_binsearch(inode, path + ppos, block);
910 /* if not an empty leaf */
911 if (path[ppos].p_ext)
912 path[ppos].p_block = ext4_ext_pblock(path[ppos].p_ext);
914 ext4_ext_show_path(inode, path);
919 ext4_ext_drop_refs(path);
927 * ext4_ext_insert_index:
928 * insert new index [@logical;@ptr] into the block at @curp;
929 * check where to insert: before @curp or after @curp
931 static int ext4_ext_insert_index(handle_t *handle, struct inode *inode,
932 struct ext4_ext_path *curp,
933 int logical, ext4_fsblk_t ptr)
935 struct ext4_extent_idx *ix;
938 err = ext4_ext_get_access(handle, inode, curp);
942 if (unlikely(logical == le32_to_cpu(curp->p_idx->ei_block))) {
943 EXT4_ERROR_INODE(inode,
944 "logical %d == ei_block %d!",
945 logical, le32_to_cpu(curp->p_idx->ei_block));
946 return -EFSCORRUPTED;
949 if (unlikely(le16_to_cpu(curp->p_hdr->eh_entries)
950 >= le16_to_cpu(curp->p_hdr->eh_max))) {
951 EXT4_ERROR_INODE(inode,
952 "eh_entries %d >= eh_max %d!",
953 le16_to_cpu(curp->p_hdr->eh_entries),
954 le16_to_cpu(curp->p_hdr->eh_max));
955 return -EFSCORRUPTED;
958 if (logical > le32_to_cpu(curp->p_idx->ei_block)) {
960 ext_debug("insert new index %d after: %llu\n", logical, ptr);
961 ix = curp->p_idx + 1;
964 ext_debug("insert new index %d before: %llu\n", logical, ptr);
968 len = EXT_LAST_INDEX(curp->p_hdr) - ix + 1;
971 ext_debug("insert new index %d: "
972 "move %d indices from 0x%p to 0x%p\n",
973 logical, len, ix, ix + 1);
974 memmove(ix + 1, ix, len * sizeof(struct ext4_extent_idx));
977 if (unlikely(ix > EXT_MAX_INDEX(curp->p_hdr))) {
978 EXT4_ERROR_INODE(inode, "ix > EXT_MAX_INDEX!");
979 return -EFSCORRUPTED;
982 ix->ei_block = cpu_to_le32(logical);
983 ext4_idx_store_pblock(ix, ptr);
984 le16_add_cpu(&curp->p_hdr->eh_entries, 1);
986 if (unlikely(ix > EXT_LAST_INDEX(curp->p_hdr))) {
987 EXT4_ERROR_INODE(inode, "ix > EXT_LAST_INDEX!");
988 return -EFSCORRUPTED;
991 err = ext4_ext_dirty(handle, inode, curp);
992 ext4_std_error(inode->i_sb, err);
999 * inserts new subtree into the path, using free index entry
1001 * - allocates all needed blocks (new leaf and all intermediate index blocks)
1002 * - makes decision where to split
1003 * - moves remaining extents and index entries (right to the split point)
1004 * into the newly allocated blocks
1005 * - initializes subtree
1007 static int ext4_ext_split(handle_t *handle, struct inode *inode,
1009 struct ext4_ext_path *path,
1010 struct ext4_extent *newext, int at)
1012 struct buffer_head *bh = NULL;
1013 int depth = ext_depth(inode);
1014 struct ext4_extent_header *neh;
1015 struct ext4_extent_idx *fidx;
1016 int i = at, k, m, a;
1017 ext4_fsblk_t newblock, oldblock;
1019 ext4_fsblk_t *ablocks = NULL; /* array of allocated blocks */
1021 size_t ext_size = 0;
1023 /* make decision: where to split? */
1024 /* FIXME: now decision is simplest: at current extent */
1026 /* if current leaf will be split, then we should use
1027 * border from split point */
1028 if (unlikely(path[depth].p_ext > EXT_MAX_EXTENT(path[depth].p_hdr))) {
1029 EXT4_ERROR_INODE(inode, "p_ext > EXT_MAX_EXTENT!");
1030 return -EFSCORRUPTED;
1032 if (path[depth].p_ext != EXT_MAX_EXTENT(path[depth].p_hdr)) {
1033 border = path[depth].p_ext[1].ee_block;
1034 ext_debug("leaf will be split."
1035 " next leaf starts at %d\n",
1036 le32_to_cpu(border));
1038 border = newext->ee_block;
1039 ext_debug("leaf will be added."
1040 " next leaf starts at %d\n",
1041 le32_to_cpu(border));
1045 * If error occurs, then we break processing
1046 * and mark filesystem read-only. index won't
1047 * be inserted and tree will be in consistent
1048 * state. Next mount will repair buffers too.
1052 * Get array to track all allocated blocks.
1053 * We need this to handle errors and free blocks
1056 ablocks = kcalloc(depth, sizeof(ext4_fsblk_t), GFP_NOFS);
1060 /* allocate all needed blocks */
1061 ext_debug("allocate %d blocks for indexes/leaf\n", depth - at);
1062 for (a = 0; a < depth - at; a++) {
1063 newblock = ext4_ext_new_meta_block(handle, inode, path,
1064 newext, &err, flags);
1067 ablocks[a] = newblock;
1070 /* initialize new leaf */
1071 newblock = ablocks[--a];
1072 if (unlikely(newblock == 0)) {
1073 EXT4_ERROR_INODE(inode, "newblock == 0!");
1074 err = -EFSCORRUPTED;
1077 bh = sb_getblk_gfp(inode->i_sb, newblock, __GFP_MOVABLE | GFP_NOFS);
1078 if (unlikely(!bh)) {
1084 err = ext4_journal_get_create_access(handle, bh);
1088 neh = ext_block_hdr(bh);
1089 neh->eh_entries = 0;
1090 neh->eh_max = cpu_to_le16(ext4_ext_space_block(inode, 0));
1091 neh->eh_magic = EXT4_EXT_MAGIC;
1094 /* move remainder of path[depth] to the new leaf */
1095 if (unlikely(path[depth].p_hdr->eh_entries !=
1096 path[depth].p_hdr->eh_max)) {
1097 EXT4_ERROR_INODE(inode, "eh_entries %d != eh_max %d!",
1098 path[depth].p_hdr->eh_entries,
1099 path[depth].p_hdr->eh_max);
1100 err = -EFSCORRUPTED;
1103 /* start copy from next extent */
1104 m = EXT_MAX_EXTENT(path[depth].p_hdr) - path[depth].p_ext++;
1105 ext4_ext_show_move(inode, path, newblock, depth);
1107 struct ext4_extent *ex;
1108 ex = EXT_FIRST_EXTENT(neh);
1109 memmove(ex, path[depth].p_ext, sizeof(struct ext4_extent) * m);
1110 le16_add_cpu(&neh->eh_entries, m);
1113 /* zero out unused area in the extent block */
1114 ext_size = sizeof(struct ext4_extent_header) +
1115 sizeof(struct ext4_extent) * le16_to_cpu(neh->eh_entries);
1116 memset(bh->b_data + ext_size, 0, inode->i_sb->s_blocksize - ext_size);
1117 ext4_extent_block_csum_set(inode, neh);
1118 set_buffer_uptodate(bh);
1121 err = ext4_handle_dirty_metadata(handle, inode, bh);
1127 /* correct old leaf */
1129 err = ext4_ext_get_access(handle, inode, path + depth);
1132 le16_add_cpu(&path[depth].p_hdr->eh_entries, -m);
1133 err = ext4_ext_dirty(handle, inode, path + depth);
1139 /* create intermediate indexes */
1141 if (unlikely(k < 0)) {
1142 EXT4_ERROR_INODE(inode, "k %d < 0!", k);
1143 err = -EFSCORRUPTED;
1147 ext_debug("create %d intermediate indices\n", k);
1148 /* insert new index into current index block */
1149 /* current depth stored in i var */
1152 oldblock = newblock;
1153 newblock = ablocks[--a];
1154 bh = sb_getblk(inode->i_sb, newblock);
1155 if (unlikely(!bh)) {
1161 err = ext4_journal_get_create_access(handle, bh);
1165 neh = ext_block_hdr(bh);
1166 neh->eh_entries = cpu_to_le16(1);
1167 neh->eh_magic = EXT4_EXT_MAGIC;
1168 neh->eh_max = cpu_to_le16(ext4_ext_space_block_idx(inode, 0));
1169 neh->eh_depth = cpu_to_le16(depth - i);
1170 fidx = EXT_FIRST_INDEX(neh);
1171 fidx->ei_block = border;
1172 ext4_idx_store_pblock(fidx, oldblock);
1174 ext_debug("int.index at %d (block %llu): %u -> %llu\n",
1175 i, newblock, le32_to_cpu(border), oldblock);
1177 /* move remainder of path[i] to the new index block */
1178 if (unlikely(EXT_MAX_INDEX(path[i].p_hdr) !=
1179 EXT_LAST_INDEX(path[i].p_hdr))) {
1180 EXT4_ERROR_INODE(inode,
1181 "EXT_MAX_INDEX != EXT_LAST_INDEX ee_block %d!",
1182 le32_to_cpu(path[i].p_ext->ee_block));
1183 err = -EFSCORRUPTED;
1186 /* start copy indexes */
1187 m = EXT_MAX_INDEX(path[i].p_hdr) - path[i].p_idx++;
1188 ext_debug("cur 0x%p, last 0x%p\n", path[i].p_idx,
1189 EXT_MAX_INDEX(path[i].p_hdr));
1190 ext4_ext_show_move(inode, path, newblock, i);
1192 memmove(++fidx, path[i].p_idx,
1193 sizeof(struct ext4_extent_idx) * m);
1194 le16_add_cpu(&neh->eh_entries, m);
1196 /* zero out unused area in the extent block */
1197 ext_size = sizeof(struct ext4_extent_header) +
1198 (sizeof(struct ext4_extent) * le16_to_cpu(neh->eh_entries));
1199 memset(bh->b_data + ext_size, 0,
1200 inode->i_sb->s_blocksize - ext_size);
1201 ext4_extent_block_csum_set(inode, neh);
1202 set_buffer_uptodate(bh);
1205 err = ext4_handle_dirty_metadata(handle, inode, bh);
1211 /* correct old index */
1213 err = ext4_ext_get_access(handle, inode, path + i);
1216 le16_add_cpu(&path[i].p_hdr->eh_entries, -m);
1217 err = ext4_ext_dirty(handle, inode, path + i);
1225 /* insert new index */
1226 err = ext4_ext_insert_index(handle, inode, path + at,
1227 le32_to_cpu(border), newblock);
1231 if (buffer_locked(bh))
1237 /* free all allocated blocks in error case */
1238 for (i = 0; i < depth; i++) {
1241 ext4_free_blocks(handle, inode, NULL, ablocks[i], 1,
1242 EXT4_FREE_BLOCKS_METADATA);
1251 * ext4_ext_grow_indepth:
1252 * implements tree growing procedure:
1253 * - allocates new block
1254 * - moves top-level data (index block or leaf) into the new block
1255 * - initializes new top-level, creating index that points to the
1256 * just created block
1258 static int ext4_ext_grow_indepth(handle_t *handle, struct inode *inode,
1261 struct ext4_extent_header *neh;
1262 struct buffer_head *bh;
1263 ext4_fsblk_t newblock, goal = 0;
1264 struct ext4_super_block *es = EXT4_SB(inode->i_sb)->s_es;
1266 size_t ext_size = 0;
1268 /* Try to prepend new index to old one */
1269 if (ext_depth(inode))
1270 goal = ext4_idx_pblock(EXT_FIRST_INDEX(ext_inode_hdr(inode)));
1271 if (goal > le32_to_cpu(es->s_first_data_block)) {
1272 flags |= EXT4_MB_HINT_TRY_GOAL;
1275 goal = ext4_inode_to_goal_block(inode);
1276 newblock = ext4_new_meta_blocks(handle, inode, goal, flags,
1281 bh = sb_getblk_gfp(inode->i_sb, newblock, __GFP_MOVABLE | GFP_NOFS);
1286 err = ext4_journal_get_create_access(handle, bh);
1292 ext_size = sizeof(EXT4_I(inode)->i_data);
1293 /* move top-level index/leaf into new block */
1294 memmove(bh->b_data, EXT4_I(inode)->i_data, ext_size);
1295 /* zero out unused area in the extent block */
1296 memset(bh->b_data + ext_size, 0, inode->i_sb->s_blocksize - ext_size);
1298 /* set size of new block */
1299 neh = ext_block_hdr(bh);
1300 /* old root could have indexes or leaves
1301 * so calculate e_max right way */
1302 if (ext_depth(inode))
1303 neh->eh_max = cpu_to_le16(ext4_ext_space_block_idx(inode, 0));
1305 neh->eh_max = cpu_to_le16(ext4_ext_space_block(inode, 0));
1306 neh->eh_magic = EXT4_EXT_MAGIC;
1307 ext4_extent_block_csum_set(inode, neh);
1308 set_buffer_uptodate(bh);
1311 err = ext4_handle_dirty_metadata(handle, inode, bh);
1315 /* Update top-level index: num,max,pointer */
1316 neh = ext_inode_hdr(inode);
1317 neh->eh_entries = cpu_to_le16(1);
1318 ext4_idx_store_pblock(EXT_FIRST_INDEX(neh), newblock);
1319 if (neh->eh_depth == 0) {
1320 /* Root extent block becomes index block */
1321 neh->eh_max = cpu_to_le16(ext4_ext_space_root_idx(inode, 0));
1322 EXT_FIRST_INDEX(neh)->ei_block =
1323 EXT_FIRST_EXTENT(neh)->ee_block;
1325 ext_debug("new root: num %d(%d), lblock %d, ptr %llu\n",
1326 le16_to_cpu(neh->eh_entries), le16_to_cpu(neh->eh_max),
1327 le32_to_cpu(EXT_FIRST_INDEX(neh)->ei_block),
1328 ext4_idx_pblock(EXT_FIRST_INDEX(neh)));
1330 le16_add_cpu(&neh->eh_depth, 1);
1331 ext4_mark_inode_dirty(handle, inode);
1339 * ext4_ext_create_new_leaf:
1340 * finds empty index and adds new leaf.
1341 * if no free index is found, then it requests in-depth growing.
1343 static int ext4_ext_create_new_leaf(handle_t *handle, struct inode *inode,
1344 unsigned int mb_flags,
1345 unsigned int gb_flags,
1346 struct ext4_ext_path **ppath,
1347 struct ext4_extent *newext)
1349 struct ext4_ext_path *path = *ppath;
1350 struct ext4_ext_path *curp;
1351 int depth, i, err = 0;
1354 i = depth = ext_depth(inode);
1356 /* walk up to the tree and look for free index entry */
1357 curp = path + depth;
1358 while (i > 0 && !EXT_HAS_FREE_INDEX(curp)) {
1363 /* we use already allocated block for index block,
1364 * so subsequent data blocks should be contiguous */
1365 if (EXT_HAS_FREE_INDEX(curp)) {
1366 /* if we found index with free entry, then use that
1367 * entry: create all needed subtree and add new leaf */
1368 err = ext4_ext_split(handle, inode, mb_flags, path, newext, i);
1373 path = ext4_find_extent(inode,
1374 (ext4_lblk_t)le32_to_cpu(newext->ee_block),
1377 err = PTR_ERR(path);
1379 /* tree is full, time to grow in depth */
1380 err = ext4_ext_grow_indepth(handle, inode, mb_flags);
1385 path = ext4_find_extent(inode,
1386 (ext4_lblk_t)le32_to_cpu(newext->ee_block),
1389 err = PTR_ERR(path);
1394 * only first (depth 0 -> 1) produces free space;
1395 * in all other cases we have to split the grown tree
1397 depth = ext_depth(inode);
1398 if (path[depth].p_hdr->eh_entries == path[depth].p_hdr->eh_max) {
1399 /* now we need to split */
1409 * search the closest allocated block to the left for *logical
1410 * and returns it at @logical + it's physical address at @phys
1411 * if *logical is the smallest allocated block, the function
1412 * returns 0 at @phys
1413 * return value contains 0 (success) or error code
1415 static int ext4_ext_search_left(struct inode *inode,
1416 struct ext4_ext_path *path,
1417 ext4_lblk_t *logical, ext4_fsblk_t *phys)
1419 struct ext4_extent_idx *ix;
1420 struct ext4_extent *ex;
1423 if (unlikely(path == NULL)) {
1424 EXT4_ERROR_INODE(inode, "path == NULL *logical %d!", *logical);
1425 return -EFSCORRUPTED;
1427 depth = path->p_depth;
1430 if (depth == 0 && path->p_ext == NULL)
1433 /* usually extent in the path covers blocks smaller
1434 * then *logical, but it can be that extent is the
1435 * first one in the file */
1437 ex = path[depth].p_ext;
1438 ee_len = ext4_ext_get_actual_len(ex);
1439 if (*logical < le32_to_cpu(ex->ee_block)) {
1440 if (unlikely(EXT_FIRST_EXTENT(path[depth].p_hdr) != ex)) {
1441 EXT4_ERROR_INODE(inode,
1442 "EXT_FIRST_EXTENT != ex *logical %d ee_block %d!",
1443 *logical, le32_to_cpu(ex->ee_block));
1444 return -EFSCORRUPTED;
1446 while (--depth >= 0) {
1447 ix = path[depth].p_idx;
1448 if (unlikely(ix != EXT_FIRST_INDEX(path[depth].p_hdr))) {
1449 EXT4_ERROR_INODE(inode,
1450 "ix (%d) != EXT_FIRST_INDEX (%d) (depth %d)!",
1451 ix != NULL ? le32_to_cpu(ix->ei_block) : 0,
1452 EXT_FIRST_INDEX(path[depth].p_hdr) != NULL ?
1453 le32_to_cpu(EXT_FIRST_INDEX(path[depth].p_hdr)->ei_block) : 0,
1455 return -EFSCORRUPTED;
1461 if (unlikely(*logical < (le32_to_cpu(ex->ee_block) + ee_len))) {
1462 EXT4_ERROR_INODE(inode,
1463 "logical %d < ee_block %d + ee_len %d!",
1464 *logical, le32_to_cpu(ex->ee_block), ee_len);
1465 return -EFSCORRUPTED;
1468 *logical = le32_to_cpu(ex->ee_block) + ee_len - 1;
1469 *phys = ext4_ext_pblock(ex) + ee_len - 1;
1474 * search the closest allocated block to the right for *logical
1475 * and returns it at @logical + it's physical address at @phys
1476 * if *logical is the largest allocated block, the function
1477 * returns 0 at @phys
1478 * return value contains 0 (success) or error code
1480 static int ext4_ext_search_right(struct inode *inode,
1481 struct ext4_ext_path *path,
1482 ext4_lblk_t *logical, ext4_fsblk_t *phys,
1483 struct ext4_extent **ret_ex)
1485 struct buffer_head *bh = NULL;
1486 struct ext4_extent_header *eh;
1487 struct ext4_extent_idx *ix;
1488 struct ext4_extent *ex;
1490 int depth; /* Note, NOT eh_depth; depth from top of tree */
1493 if (unlikely(path == NULL)) {
1494 EXT4_ERROR_INODE(inode, "path == NULL *logical %d!", *logical);
1495 return -EFSCORRUPTED;
1497 depth = path->p_depth;
1500 if (depth == 0 && path->p_ext == NULL)
1503 /* usually extent in the path covers blocks smaller
1504 * then *logical, but it can be that extent is the
1505 * first one in the file */
1507 ex = path[depth].p_ext;
1508 ee_len = ext4_ext_get_actual_len(ex);
1509 if (*logical < le32_to_cpu(ex->ee_block)) {
1510 if (unlikely(EXT_FIRST_EXTENT(path[depth].p_hdr) != ex)) {
1511 EXT4_ERROR_INODE(inode,
1512 "first_extent(path[%d].p_hdr) != ex",
1514 return -EFSCORRUPTED;
1516 while (--depth >= 0) {
1517 ix = path[depth].p_idx;
1518 if (unlikely(ix != EXT_FIRST_INDEX(path[depth].p_hdr))) {
1519 EXT4_ERROR_INODE(inode,
1520 "ix != EXT_FIRST_INDEX *logical %d!",
1522 return -EFSCORRUPTED;
1528 if (unlikely(*logical < (le32_to_cpu(ex->ee_block) + ee_len))) {
1529 EXT4_ERROR_INODE(inode,
1530 "logical %d < ee_block %d + ee_len %d!",
1531 *logical, le32_to_cpu(ex->ee_block), ee_len);
1532 return -EFSCORRUPTED;
1535 if (ex != EXT_LAST_EXTENT(path[depth].p_hdr)) {
1536 /* next allocated block in this leaf */
1541 /* go up and search for index to the right */
1542 while (--depth >= 0) {
1543 ix = path[depth].p_idx;
1544 if (ix != EXT_LAST_INDEX(path[depth].p_hdr))
1548 /* we've gone up to the root and found no index to the right */
1552 /* we've found index to the right, let's
1553 * follow it and find the closest allocated
1554 * block to the right */
1556 block = ext4_idx_pblock(ix);
1557 while (++depth < path->p_depth) {
1558 /* subtract from p_depth to get proper eh_depth */
1559 bh = read_extent_tree_block(inode, block,
1560 path->p_depth - depth, 0);
1563 eh = ext_block_hdr(bh);
1564 ix = EXT_FIRST_INDEX(eh);
1565 block = ext4_idx_pblock(ix);
1569 bh = read_extent_tree_block(inode, block, path->p_depth - depth, 0);
1572 eh = ext_block_hdr(bh);
1573 ex = EXT_FIRST_EXTENT(eh);
1575 *logical = le32_to_cpu(ex->ee_block);
1576 *phys = ext4_ext_pblock(ex);
1584 * ext4_ext_next_allocated_block:
1585 * returns allocated block in subsequent extent or EXT_MAX_BLOCKS.
1586 * NOTE: it considers block number from index entry as
1587 * allocated block. Thus, index entries have to be consistent
1591 ext4_ext_next_allocated_block(struct ext4_ext_path *path)
1595 BUG_ON(path == NULL);
1596 depth = path->p_depth;
1598 if (depth == 0 && path->p_ext == NULL)
1599 return EXT_MAX_BLOCKS;
1601 while (depth >= 0) {
1602 struct ext4_ext_path *p = &path[depth];
1604 if (depth == path->p_depth) {
1606 if (p->p_ext && p->p_ext != EXT_LAST_EXTENT(p->p_hdr))
1607 return le32_to_cpu(p->p_ext[1].ee_block);
1610 if (p->p_idx != EXT_LAST_INDEX(p->p_hdr))
1611 return le32_to_cpu(p->p_idx[1].ei_block);
1616 return EXT_MAX_BLOCKS;
1620 * ext4_ext_next_leaf_block:
1621 * returns first allocated block from next leaf or EXT_MAX_BLOCKS
1623 static ext4_lblk_t ext4_ext_next_leaf_block(struct ext4_ext_path *path)
1627 BUG_ON(path == NULL);
1628 depth = path->p_depth;
1630 /* zero-tree has no leaf blocks at all */
1632 return EXT_MAX_BLOCKS;
1634 /* go to index block */
1637 while (depth >= 0) {
1638 if (path[depth].p_idx !=
1639 EXT_LAST_INDEX(path[depth].p_hdr))
1640 return (ext4_lblk_t)
1641 le32_to_cpu(path[depth].p_idx[1].ei_block);
1645 return EXT_MAX_BLOCKS;
1649 * ext4_ext_correct_indexes:
1650 * if leaf gets modified and modified extent is first in the leaf,
1651 * then we have to correct all indexes above.
1652 * TODO: do we need to correct tree in all cases?
1654 static int ext4_ext_correct_indexes(handle_t *handle, struct inode *inode,
1655 struct ext4_ext_path *path)
1657 struct ext4_extent_header *eh;
1658 int depth = ext_depth(inode);
1659 struct ext4_extent *ex;
1663 eh = path[depth].p_hdr;
1664 ex = path[depth].p_ext;
1666 if (unlikely(ex == NULL || eh == NULL)) {
1667 EXT4_ERROR_INODE(inode,
1668 "ex %p == NULL or eh %p == NULL", ex, eh);
1669 return -EFSCORRUPTED;
1673 /* there is no tree at all */
1677 if (ex != EXT_FIRST_EXTENT(eh)) {
1678 /* we correct tree if first leaf got modified only */
1683 * TODO: we need correction if border is smaller than current one
1686 border = path[depth].p_ext->ee_block;
1687 err = ext4_ext_get_access(handle, inode, path + k);
1690 path[k].p_idx->ei_block = border;
1691 err = ext4_ext_dirty(handle, inode, path + k);
1696 /* change all left-side indexes */
1697 if (path[k+1].p_idx != EXT_FIRST_INDEX(path[k+1].p_hdr))
1699 err = ext4_ext_get_access(handle, inode, path + k);
1702 path[k].p_idx->ei_block = border;
1703 err = ext4_ext_dirty(handle, inode, path + k);
1711 static int ext4_can_extents_be_merged(struct inode *inode,
1712 struct ext4_extent *ex1,
1713 struct ext4_extent *ex2)
1715 unsigned short ext1_ee_len, ext2_ee_len;
1717 if (ext4_ext_is_unwritten(ex1) != ext4_ext_is_unwritten(ex2))
1720 ext1_ee_len = ext4_ext_get_actual_len(ex1);
1721 ext2_ee_len = ext4_ext_get_actual_len(ex2);
1723 if (le32_to_cpu(ex1->ee_block) + ext1_ee_len !=
1724 le32_to_cpu(ex2->ee_block))
1727 if (ext1_ee_len + ext2_ee_len > EXT_INIT_MAX_LEN)
1730 if (ext4_ext_is_unwritten(ex1) &&
1731 ext1_ee_len + ext2_ee_len > EXT_UNWRITTEN_MAX_LEN)
1733 #ifdef AGGRESSIVE_TEST
1734 if (ext1_ee_len >= 4)
1738 if (ext4_ext_pblock(ex1) + ext1_ee_len == ext4_ext_pblock(ex2))
1744 * This function tries to merge the "ex" extent to the next extent in the tree.
1745 * It always tries to merge towards right. If you want to merge towards
1746 * left, pass "ex - 1" as argument instead of "ex".
1747 * Returns 0 if the extents (ex and ex+1) were _not_ merged and returns
1748 * 1 if they got merged.
1750 static int ext4_ext_try_to_merge_right(struct inode *inode,
1751 struct ext4_ext_path *path,
1752 struct ext4_extent *ex)
1754 struct ext4_extent_header *eh;
1755 unsigned int depth, len;
1756 int merge_done = 0, unwritten;
1758 depth = ext_depth(inode);
1759 BUG_ON(path[depth].p_hdr == NULL);
1760 eh = path[depth].p_hdr;
1762 while (ex < EXT_LAST_EXTENT(eh)) {
1763 if (!ext4_can_extents_be_merged(inode, ex, ex + 1))
1765 /* merge with next extent! */
1766 unwritten = ext4_ext_is_unwritten(ex);
1767 ex->ee_len = cpu_to_le16(ext4_ext_get_actual_len(ex)
1768 + ext4_ext_get_actual_len(ex + 1));
1770 ext4_ext_mark_unwritten(ex);
1772 if (ex + 1 < EXT_LAST_EXTENT(eh)) {
1773 len = (EXT_LAST_EXTENT(eh) - ex - 1)
1774 * sizeof(struct ext4_extent);
1775 memmove(ex + 1, ex + 2, len);
1777 le16_add_cpu(&eh->eh_entries, -1);
1779 WARN_ON(eh->eh_entries == 0);
1780 if (!eh->eh_entries)
1781 EXT4_ERROR_INODE(inode, "eh->eh_entries = 0!");
1788 * This function does a very simple check to see if we can collapse
1789 * an extent tree with a single extent tree leaf block into the inode.
1791 static void ext4_ext_try_to_merge_up(handle_t *handle,
1792 struct inode *inode,
1793 struct ext4_ext_path *path)
1796 unsigned max_root = ext4_ext_space_root(inode, 0);
1799 if ((path[0].p_depth != 1) ||
1800 (le16_to_cpu(path[0].p_hdr->eh_entries) != 1) ||
1801 (le16_to_cpu(path[1].p_hdr->eh_entries) > max_root))
1805 * We need to modify the block allocation bitmap and the block
1806 * group descriptor to release the extent tree block. If we
1807 * can't get the journal credits, give up.
1809 if (ext4_journal_extend(handle, 2,
1810 ext4_free_metadata_revoke_credits(inode->i_sb, 1)))
1814 * Copy the extent data up to the inode
1816 blk = ext4_idx_pblock(path[0].p_idx);
1817 s = le16_to_cpu(path[1].p_hdr->eh_entries) *
1818 sizeof(struct ext4_extent_idx);
1819 s += sizeof(struct ext4_extent_header);
1821 path[1].p_maxdepth = path[0].p_maxdepth;
1822 memcpy(path[0].p_hdr, path[1].p_hdr, s);
1823 path[0].p_depth = 0;
1824 path[0].p_ext = EXT_FIRST_EXTENT(path[0].p_hdr) +
1825 (path[1].p_ext - EXT_FIRST_EXTENT(path[1].p_hdr));
1826 path[0].p_hdr->eh_max = cpu_to_le16(max_root);
1828 brelse(path[1].p_bh);
1829 ext4_free_blocks(handle, inode, NULL, blk, 1,
1830 EXT4_FREE_BLOCKS_METADATA | EXT4_FREE_BLOCKS_FORGET);
1834 * This function tries to merge the @ex extent to neighbours in the tree, then
1835 * tries to collapse the extent tree into the inode.
1837 static void ext4_ext_try_to_merge(handle_t *handle,
1838 struct inode *inode,
1839 struct ext4_ext_path *path,
1840 struct ext4_extent *ex)
1842 struct ext4_extent_header *eh;
1846 depth = ext_depth(inode);
1847 BUG_ON(path[depth].p_hdr == NULL);
1848 eh = path[depth].p_hdr;
1850 if (ex > EXT_FIRST_EXTENT(eh))
1851 merge_done = ext4_ext_try_to_merge_right(inode, path, ex - 1);
1854 (void) ext4_ext_try_to_merge_right(inode, path, ex);
1856 ext4_ext_try_to_merge_up(handle, inode, path);
1860 * check if a portion of the "newext" extent overlaps with an
1863 * If there is an overlap discovered, it updates the length of the newext
1864 * such that there will be no overlap, and then returns 1.
1865 * If there is no overlap found, it returns 0.
1867 static unsigned int ext4_ext_check_overlap(struct ext4_sb_info *sbi,
1868 struct inode *inode,
1869 struct ext4_extent *newext,
1870 struct ext4_ext_path *path)
1873 unsigned int depth, len1;
1874 unsigned int ret = 0;
1876 b1 = le32_to_cpu(newext->ee_block);
1877 len1 = ext4_ext_get_actual_len(newext);
1878 depth = ext_depth(inode);
1879 if (!path[depth].p_ext)
1881 b2 = EXT4_LBLK_CMASK(sbi, le32_to_cpu(path[depth].p_ext->ee_block));
1884 * get the next allocated block if the extent in the path
1885 * is before the requested block(s)
1888 b2 = ext4_ext_next_allocated_block(path);
1889 if (b2 == EXT_MAX_BLOCKS)
1891 b2 = EXT4_LBLK_CMASK(sbi, b2);
1894 /* check for wrap through zero on extent logical start block*/
1895 if (b1 + len1 < b1) {
1896 len1 = EXT_MAX_BLOCKS - b1;
1897 newext->ee_len = cpu_to_le16(len1);
1901 /* check for overlap */
1902 if (b1 + len1 > b2) {
1903 newext->ee_len = cpu_to_le16(b2 - b1);
1911 * ext4_ext_insert_extent:
1912 * tries to merge requsted extent into the existing extent or
1913 * inserts requested extent as new one into the tree,
1914 * creating new leaf in the no-space case.
1916 int ext4_ext_insert_extent(handle_t *handle, struct inode *inode,
1917 struct ext4_ext_path **ppath,
1918 struct ext4_extent *newext, int gb_flags)
1920 struct ext4_ext_path *path = *ppath;
1921 struct ext4_extent_header *eh;
1922 struct ext4_extent *ex, *fex;
1923 struct ext4_extent *nearex; /* nearest extent */
1924 struct ext4_ext_path *npath = NULL;
1925 int depth, len, err;
1927 int mb_flags = 0, unwritten;
1929 if (gb_flags & EXT4_GET_BLOCKS_DELALLOC_RESERVE)
1930 mb_flags |= EXT4_MB_DELALLOC_RESERVED;
1931 if (unlikely(ext4_ext_get_actual_len(newext) == 0)) {
1932 EXT4_ERROR_INODE(inode, "ext4_ext_get_actual_len(newext) == 0");
1933 return -EFSCORRUPTED;
1935 depth = ext_depth(inode);
1936 ex = path[depth].p_ext;
1937 eh = path[depth].p_hdr;
1938 if (unlikely(path[depth].p_hdr == NULL)) {
1939 EXT4_ERROR_INODE(inode, "path[%d].p_hdr == NULL", depth);
1940 return -EFSCORRUPTED;
1943 /* try to insert block into found extent and return */
1944 if (ex && !(gb_flags & EXT4_GET_BLOCKS_PRE_IO)) {
1947 * Try to see whether we should rather test the extent on
1948 * right from ex, or from the left of ex. This is because
1949 * ext4_find_extent() can return either extent on the
1950 * left, or on the right from the searched position. This
1951 * will make merging more effective.
1953 if (ex < EXT_LAST_EXTENT(eh) &&
1954 (le32_to_cpu(ex->ee_block) +
1955 ext4_ext_get_actual_len(ex) <
1956 le32_to_cpu(newext->ee_block))) {
1959 } else if ((ex > EXT_FIRST_EXTENT(eh)) &&
1960 (le32_to_cpu(newext->ee_block) +
1961 ext4_ext_get_actual_len(newext) <
1962 le32_to_cpu(ex->ee_block)))
1965 /* Try to append newex to the ex */
1966 if (ext4_can_extents_be_merged(inode, ex, newext)) {
1967 ext_debug("append [%d]%d block to %u:[%d]%d"
1969 ext4_ext_is_unwritten(newext),
1970 ext4_ext_get_actual_len(newext),
1971 le32_to_cpu(ex->ee_block),
1972 ext4_ext_is_unwritten(ex),
1973 ext4_ext_get_actual_len(ex),
1974 ext4_ext_pblock(ex));
1975 err = ext4_ext_get_access(handle, inode,
1979 unwritten = ext4_ext_is_unwritten(ex);
1980 ex->ee_len = cpu_to_le16(ext4_ext_get_actual_len(ex)
1981 + ext4_ext_get_actual_len(newext));
1983 ext4_ext_mark_unwritten(ex);
1984 eh = path[depth].p_hdr;
1990 /* Try to prepend newex to the ex */
1991 if (ext4_can_extents_be_merged(inode, newext, ex)) {
1992 ext_debug("prepend %u[%d]%d block to %u:[%d]%d"
1994 le32_to_cpu(newext->ee_block),
1995 ext4_ext_is_unwritten(newext),
1996 ext4_ext_get_actual_len(newext),
1997 le32_to_cpu(ex->ee_block),
1998 ext4_ext_is_unwritten(ex),
1999 ext4_ext_get_actual_len(ex),
2000 ext4_ext_pblock(ex));
2001 err = ext4_ext_get_access(handle, inode,
2006 unwritten = ext4_ext_is_unwritten(ex);
2007 ex->ee_block = newext->ee_block;
2008 ext4_ext_store_pblock(ex, ext4_ext_pblock(newext));
2009 ex->ee_len = cpu_to_le16(ext4_ext_get_actual_len(ex)
2010 + ext4_ext_get_actual_len(newext));
2012 ext4_ext_mark_unwritten(ex);
2013 eh = path[depth].p_hdr;
2019 depth = ext_depth(inode);
2020 eh = path[depth].p_hdr;
2021 if (le16_to_cpu(eh->eh_entries) < le16_to_cpu(eh->eh_max))
2024 /* probably next leaf has space for us? */
2025 fex = EXT_LAST_EXTENT(eh);
2026 next = EXT_MAX_BLOCKS;
2027 if (le32_to_cpu(newext->ee_block) > le32_to_cpu(fex->ee_block))
2028 next = ext4_ext_next_leaf_block(path);
2029 if (next != EXT_MAX_BLOCKS) {
2030 ext_debug("next leaf block - %u\n", next);
2031 BUG_ON(npath != NULL);
2032 npath = ext4_find_extent(inode, next, NULL, 0);
2034 return PTR_ERR(npath);
2035 BUG_ON(npath->p_depth != path->p_depth);
2036 eh = npath[depth].p_hdr;
2037 if (le16_to_cpu(eh->eh_entries) < le16_to_cpu(eh->eh_max)) {
2038 ext_debug("next leaf isn't full(%d)\n",
2039 le16_to_cpu(eh->eh_entries));
2043 ext_debug("next leaf has no free space(%d,%d)\n",
2044 le16_to_cpu(eh->eh_entries), le16_to_cpu(eh->eh_max));
2048 * There is no free space in the found leaf.
2049 * We're gonna add a new leaf in the tree.
2051 if (gb_flags & EXT4_GET_BLOCKS_METADATA_NOFAIL)
2052 mb_flags |= EXT4_MB_USE_RESERVED;
2053 err = ext4_ext_create_new_leaf(handle, inode, mb_flags, gb_flags,
2057 depth = ext_depth(inode);
2058 eh = path[depth].p_hdr;
2061 nearex = path[depth].p_ext;
2063 err = ext4_ext_get_access(handle, inode, path + depth);
2068 /* there is no extent in this leaf, create first one */
2069 ext_debug("first extent in the leaf: %u:%llu:[%d]%d\n",
2070 le32_to_cpu(newext->ee_block),
2071 ext4_ext_pblock(newext),
2072 ext4_ext_is_unwritten(newext),
2073 ext4_ext_get_actual_len(newext));
2074 nearex = EXT_FIRST_EXTENT(eh);
2076 if (le32_to_cpu(newext->ee_block)
2077 > le32_to_cpu(nearex->ee_block)) {
2079 ext_debug("insert %u:%llu:[%d]%d before: "
2081 le32_to_cpu(newext->ee_block),
2082 ext4_ext_pblock(newext),
2083 ext4_ext_is_unwritten(newext),
2084 ext4_ext_get_actual_len(newext),
2089 BUG_ON(newext->ee_block == nearex->ee_block);
2090 ext_debug("insert %u:%llu:[%d]%d after: "
2092 le32_to_cpu(newext->ee_block),
2093 ext4_ext_pblock(newext),
2094 ext4_ext_is_unwritten(newext),
2095 ext4_ext_get_actual_len(newext),
2098 len = EXT_LAST_EXTENT(eh) - nearex + 1;
2100 ext_debug("insert %u:%llu:[%d]%d: "
2101 "move %d extents from 0x%p to 0x%p\n",
2102 le32_to_cpu(newext->ee_block),
2103 ext4_ext_pblock(newext),
2104 ext4_ext_is_unwritten(newext),
2105 ext4_ext_get_actual_len(newext),
2106 len, nearex, nearex + 1);
2107 memmove(nearex + 1, nearex,
2108 len * sizeof(struct ext4_extent));
2112 le16_add_cpu(&eh->eh_entries, 1);
2113 path[depth].p_ext = nearex;
2114 nearex->ee_block = newext->ee_block;
2115 ext4_ext_store_pblock(nearex, ext4_ext_pblock(newext));
2116 nearex->ee_len = newext->ee_len;
2119 /* try to merge extents */
2120 if (!(gb_flags & EXT4_GET_BLOCKS_PRE_IO))
2121 ext4_ext_try_to_merge(handle, inode, path, nearex);
2124 /* time to correct all indexes above */
2125 err = ext4_ext_correct_indexes(handle, inode, path);
2129 err = ext4_ext_dirty(handle, inode, path + path->p_depth);
2132 ext4_ext_drop_refs(npath);
2137 static int ext4_fill_fiemap_extents(struct inode *inode,
2138 ext4_lblk_t block, ext4_lblk_t num,
2139 struct fiemap_extent_info *fieinfo)
2141 struct ext4_ext_path *path = NULL;
2142 struct ext4_extent *ex;
2143 struct extent_status es;
2144 ext4_lblk_t next, next_del, start = 0, end = 0;
2145 ext4_lblk_t last = block + num;
2146 int exists, depth = 0, err = 0;
2147 unsigned int flags = 0;
2148 unsigned char blksize_bits = inode->i_sb->s_blocksize_bits;
2150 while (block < last && block != EXT_MAX_BLOCKS) {
2152 /* find extent for this block */
2153 down_read(&EXT4_I(inode)->i_data_sem);
2155 path = ext4_find_extent(inode, block, &path, 0);
2157 up_read(&EXT4_I(inode)->i_data_sem);
2158 err = PTR_ERR(path);
2163 depth = ext_depth(inode);
2164 if (unlikely(path[depth].p_hdr == NULL)) {
2165 up_read(&EXT4_I(inode)->i_data_sem);
2166 EXT4_ERROR_INODE(inode, "path[%d].p_hdr == NULL", depth);
2167 err = -EFSCORRUPTED;
2170 ex = path[depth].p_ext;
2171 next = ext4_ext_next_allocated_block(path);
2176 /* there is no extent yet, so try to allocate
2177 * all requested space */
2180 } else if (le32_to_cpu(ex->ee_block) > block) {
2181 /* need to allocate space before found extent */
2183 end = le32_to_cpu(ex->ee_block);
2184 if (block + num < end)
2186 } else if (block >= le32_to_cpu(ex->ee_block)
2187 + ext4_ext_get_actual_len(ex)) {
2188 /* need to allocate space after found extent */
2193 } else if (block >= le32_to_cpu(ex->ee_block)) {
2195 * some part of requested space is covered
2199 end = le32_to_cpu(ex->ee_block)
2200 + ext4_ext_get_actual_len(ex);
2201 if (block + num < end)
2207 BUG_ON(end <= start);
2211 es.es_len = end - start;
2214 es.es_lblk = le32_to_cpu(ex->ee_block);
2215 es.es_len = ext4_ext_get_actual_len(ex);
2216 es.es_pblk = ext4_ext_pblock(ex);
2217 if (ext4_ext_is_unwritten(ex))
2218 flags |= FIEMAP_EXTENT_UNWRITTEN;
2222 * Find delayed extent and update es accordingly. We call
2223 * it even in !exists case to find out whether es is the
2224 * last existing extent or not.
2226 next_del = ext4_find_delayed_extent(inode, &es);
2227 if (!exists && next_del) {
2229 flags |= (FIEMAP_EXTENT_DELALLOC |
2230 FIEMAP_EXTENT_UNKNOWN);
2232 up_read(&EXT4_I(inode)->i_data_sem);
2234 if (unlikely(es.es_len == 0)) {
2235 EXT4_ERROR_INODE(inode, "es.es_len == 0");
2236 err = -EFSCORRUPTED;
2241 * This is possible iff next == next_del == EXT_MAX_BLOCKS.
2242 * we need to check next == EXT_MAX_BLOCKS because it is
2243 * possible that an extent is with unwritten and delayed
2244 * status due to when an extent is delayed allocated and
2245 * is allocated by fallocate status tree will track both of
2248 * So we could return a unwritten and delayed extent, and
2249 * its block is equal to 'next'.
2251 if (next == next_del && next == EXT_MAX_BLOCKS) {
2252 flags |= FIEMAP_EXTENT_LAST;
2253 if (unlikely(next_del != EXT_MAX_BLOCKS ||
2254 next != EXT_MAX_BLOCKS)) {
2255 EXT4_ERROR_INODE(inode,
2256 "next extent == %u, next "
2257 "delalloc extent = %u",
2259 err = -EFSCORRUPTED;
2265 err = fiemap_fill_next_extent(fieinfo,
2266 (__u64)es.es_lblk << blksize_bits,
2267 (__u64)es.es_pblk << blksize_bits,
2268 (__u64)es.es_len << blksize_bits,
2278 block = es.es_lblk + es.es_len;
2281 ext4_ext_drop_refs(path);
2286 static int ext4_fill_es_cache_info(struct inode *inode,
2287 ext4_lblk_t block, ext4_lblk_t num,
2288 struct fiemap_extent_info *fieinfo)
2290 ext4_lblk_t next, end = block + num - 1;
2291 struct extent_status es;
2292 unsigned char blksize_bits = inode->i_sb->s_blocksize_bits;
2296 while (block <= end) {
2299 if (!ext4_es_lookup_extent(inode, block, &next, &es))
2301 if (ext4_es_is_unwritten(&es))
2302 flags |= FIEMAP_EXTENT_UNWRITTEN;
2303 if (ext4_es_is_delayed(&es))
2304 flags |= (FIEMAP_EXTENT_DELALLOC |
2305 FIEMAP_EXTENT_UNKNOWN);
2306 if (ext4_es_is_hole(&es))
2307 flags |= EXT4_FIEMAP_EXTENT_HOLE;
2309 flags |= FIEMAP_EXTENT_LAST;
2310 if (flags & (FIEMAP_EXTENT_DELALLOC|
2311 EXT4_FIEMAP_EXTENT_HOLE))
2314 es.es_pblk = ext4_es_pblock(&es);
2315 err = fiemap_fill_next_extent(fieinfo,
2316 (__u64)es.es_lblk << blksize_bits,
2317 (__u64)es.es_pblk << blksize_bits,
2318 (__u64)es.es_len << blksize_bits,
2333 * ext4_ext_determine_hole - determine hole around given block
2334 * @inode: inode we lookup in
2335 * @path: path in extent tree to @lblk
2336 * @lblk: pointer to logical block around which we want to determine hole
2338 * Determine hole length (and start if easily possible) around given logical
2339 * block. We don't try too hard to find the beginning of the hole but @path
2340 * actually points to extent before @lblk, we provide it.
2342 * The function returns the length of a hole starting at @lblk. We update @lblk
2343 * to the beginning of the hole if we managed to find it.
2345 static ext4_lblk_t ext4_ext_determine_hole(struct inode *inode,
2346 struct ext4_ext_path *path,
2349 int depth = ext_depth(inode);
2350 struct ext4_extent *ex;
2353 ex = path[depth].p_ext;
2355 /* there is no extent yet, so gap is [0;-] */
2357 len = EXT_MAX_BLOCKS;
2358 } else if (*lblk < le32_to_cpu(ex->ee_block)) {
2359 len = le32_to_cpu(ex->ee_block) - *lblk;
2360 } else if (*lblk >= le32_to_cpu(ex->ee_block)
2361 + ext4_ext_get_actual_len(ex)) {
2364 *lblk = le32_to_cpu(ex->ee_block) + ext4_ext_get_actual_len(ex);
2365 next = ext4_ext_next_allocated_block(path);
2366 BUG_ON(next == *lblk);
2375 * ext4_ext_put_gap_in_cache:
2376 * calculate boundaries of the gap that the requested block fits into
2377 * and cache this gap
2380 ext4_ext_put_gap_in_cache(struct inode *inode, ext4_lblk_t hole_start,
2381 ext4_lblk_t hole_len)
2383 struct extent_status es;
2385 ext4_es_find_extent_range(inode, &ext4_es_is_delayed, hole_start,
2386 hole_start + hole_len - 1, &es);
2388 /* There's delayed extent containing lblock? */
2389 if (es.es_lblk <= hole_start)
2391 hole_len = min(es.es_lblk - hole_start, hole_len);
2393 ext_debug(" -> %u:%u\n", hole_start, hole_len);
2394 ext4_es_insert_extent(inode, hole_start, hole_len, ~0,
2395 EXTENT_STATUS_HOLE);
2400 * removes index from the index block.
2402 static int ext4_ext_rm_idx(handle_t *handle, struct inode *inode,
2403 struct ext4_ext_path *path, int depth)
2408 /* free index block */
2410 path = path + depth;
2411 leaf = ext4_idx_pblock(path->p_idx);
2412 if (unlikely(path->p_hdr->eh_entries == 0)) {
2413 EXT4_ERROR_INODE(inode, "path->p_hdr->eh_entries == 0");
2414 return -EFSCORRUPTED;
2416 err = ext4_ext_get_access(handle, inode, path);
2420 if (path->p_idx != EXT_LAST_INDEX(path->p_hdr)) {
2421 int len = EXT_LAST_INDEX(path->p_hdr) - path->p_idx;
2422 len *= sizeof(struct ext4_extent_idx);
2423 memmove(path->p_idx, path->p_idx + 1, len);
2426 le16_add_cpu(&path->p_hdr->eh_entries, -1);
2427 err = ext4_ext_dirty(handle, inode, path);
2430 ext_debug("index is empty, remove it, free block %llu\n", leaf);
2431 trace_ext4_ext_rm_idx(inode, leaf);
2433 ext4_free_blocks(handle, inode, NULL, leaf, 1,
2434 EXT4_FREE_BLOCKS_METADATA | EXT4_FREE_BLOCKS_FORGET);
2436 while (--depth >= 0) {
2437 if (path->p_idx != EXT_FIRST_INDEX(path->p_hdr))
2440 err = ext4_ext_get_access(handle, inode, path);
2443 path->p_idx->ei_block = (path+1)->p_idx->ei_block;
2444 err = ext4_ext_dirty(handle, inode, path);
2452 * ext4_ext_calc_credits_for_single_extent:
2453 * This routine returns max. credits that needed to insert an extent
2454 * to the extent tree.
2455 * When pass the actual path, the caller should calculate credits
2458 int ext4_ext_calc_credits_for_single_extent(struct inode *inode, int nrblocks,
2459 struct ext4_ext_path *path)
2462 int depth = ext_depth(inode);
2465 /* probably there is space in leaf? */
2466 if (le16_to_cpu(path[depth].p_hdr->eh_entries)
2467 < le16_to_cpu(path[depth].p_hdr->eh_max)) {
2470 * There are some space in the leaf tree, no
2471 * need to account for leaf block credit
2473 * bitmaps and block group descriptor blocks
2474 * and other metadata blocks still need to be
2477 /* 1 bitmap, 1 block group descriptor */
2478 ret = 2 + EXT4_META_TRANS_BLOCKS(inode->i_sb);
2483 return ext4_chunk_trans_blocks(inode, nrblocks);
2487 * How many index/leaf blocks need to change/allocate to add @extents extents?
2489 * If we add a single extent, then in the worse case, each tree level
2490 * index/leaf need to be changed in case of the tree split.
2492 * If more extents are inserted, they could cause the whole tree split more
2493 * than once, but this is really rare.
2495 int ext4_ext_index_trans_blocks(struct inode *inode, int extents)
2500 /* If we are converting the inline data, only one is needed here. */
2501 if (ext4_has_inline_data(inode))
2504 depth = ext_depth(inode);
2514 static inline int get_default_free_blocks_flags(struct inode *inode)
2516 if (S_ISDIR(inode->i_mode) || S_ISLNK(inode->i_mode) ||
2517 ext4_test_inode_flag(inode, EXT4_INODE_EA_INODE))
2518 return EXT4_FREE_BLOCKS_METADATA | EXT4_FREE_BLOCKS_FORGET;
2519 else if (ext4_should_journal_data(inode))
2520 return EXT4_FREE_BLOCKS_FORGET;
2525 * ext4_rereserve_cluster - increment the reserved cluster count when
2526 * freeing a cluster with a pending reservation
2528 * @inode - file containing the cluster
2529 * @lblk - logical block in cluster to be reserved
2531 * Increments the reserved cluster count and adjusts quota in a bigalloc
2532 * file system when freeing a partial cluster containing at least one
2533 * delayed and unwritten block. A partial cluster meeting that
2534 * requirement will have a pending reservation. If so, the
2535 * RERESERVE_CLUSTER flag is used when calling ext4_free_blocks() to
2536 * defer reserved and allocated space accounting to a subsequent call
2539 static void ext4_rereserve_cluster(struct inode *inode, ext4_lblk_t lblk)
2541 struct ext4_sb_info *sbi = EXT4_SB(inode->i_sb);
2542 struct ext4_inode_info *ei = EXT4_I(inode);
2544 dquot_reclaim_block(inode, EXT4_C2B(sbi, 1));
2546 spin_lock(&ei->i_block_reservation_lock);
2547 ei->i_reserved_data_blocks++;
2548 percpu_counter_add(&sbi->s_dirtyclusters_counter, 1);
2549 spin_unlock(&ei->i_block_reservation_lock);
2551 percpu_counter_add(&sbi->s_freeclusters_counter, 1);
2552 ext4_remove_pending(inode, lblk);
2555 static int ext4_remove_blocks(handle_t *handle, struct inode *inode,
2556 struct ext4_extent *ex,
2557 struct partial_cluster *partial,
2558 ext4_lblk_t from, ext4_lblk_t to)
2560 struct ext4_sb_info *sbi = EXT4_SB(inode->i_sb);
2561 unsigned short ee_len = ext4_ext_get_actual_len(ex);
2562 ext4_fsblk_t last_pblk, pblk;
2566 /* only extent tail removal is allowed */
2567 if (from < le32_to_cpu(ex->ee_block) ||
2568 to != le32_to_cpu(ex->ee_block) + ee_len - 1) {
2569 ext4_error(sbi->s_sb,
2570 "strange request: removal(2) %u-%u from %u:%u",
2571 from, to, le32_to_cpu(ex->ee_block), ee_len);
2575 #ifdef EXTENTS_STATS
2576 spin_lock(&sbi->s_ext_stats_lock);
2577 sbi->s_ext_blocks += ee_len;
2578 sbi->s_ext_extents++;
2579 if (ee_len < sbi->s_ext_min)
2580 sbi->s_ext_min = ee_len;
2581 if (ee_len > sbi->s_ext_max)
2582 sbi->s_ext_max = ee_len;
2583 if (ext_depth(inode) > sbi->s_depth_max)
2584 sbi->s_depth_max = ext_depth(inode);
2585 spin_unlock(&sbi->s_ext_stats_lock);
2588 trace_ext4_remove_blocks(inode, ex, from, to, partial);
2591 * if we have a partial cluster, and it's different from the
2592 * cluster of the last block in the extent, we free it
2594 last_pblk = ext4_ext_pblock(ex) + ee_len - 1;
2596 if (partial->state != initial &&
2597 partial->pclu != EXT4_B2C(sbi, last_pblk)) {
2598 if (partial->state == tofree) {
2599 flags = get_default_free_blocks_flags(inode);
2600 if (ext4_is_pending(inode, partial->lblk))
2601 flags |= EXT4_FREE_BLOCKS_RERESERVE_CLUSTER;
2602 ext4_free_blocks(handle, inode, NULL,
2603 EXT4_C2B(sbi, partial->pclu),
2604 sbi->s_cluster_ratio, flags);
2605 if (flags & EXT4_FREE_BLOCKS_RERESERVE_CLUSTER)
2606 ext4_rereserve_cluster(inode, partial->lblk);
2608 partial->state = initial;
2611 num = le32_to_cpu(ex->ee_block) + ee_len - from;
2612 pblk = ext4_ext_pblock(ex) + ee_len - num;
2615 * We free the partial cluster at the end of the extent (if any),
2616 * unless the cluster is used by another extent (partial_cluster
2617 * state is nofree). If a partial cluster exists here, it must be
2618 * shared with the last block in the extent.
2620 flags = get_default_free_blocks_flags(inode);
2622 /* partial, left end cluster aligned, right end unaligned */
2623 if ((EXT4_LBLK_COFF(sbi, to) != sbi->s_cluster_ratio - 1) &&
2624 (EXT4_LBLK_CMASK(sbi, to) >= from) &&
2625 (partial->state != nofree)) {
2626 if (ext4_is_pending(inode, to))
2627 flags |= EXT4_FREE_BLOCKS_RERESERVE_CLUSTER;
2628 ext4_free_blocks(handle, inode, NULL,
2629 EXT4_PBLK_CMASK(sbi, last_pblk),
2630 sbi->s_cluster_ratio, flags);
2631 if (flags & EXT4_FREE_BLOCKS_RERESERVE_CLUSTER)
2632 ext4_rereserve_cluster(inode, to);
2633 partial->state = initial;
2634 flags = get_default_free_blocks_flags(inode);
2637 flags |= EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER;
2640 * For bigalloc file systems, we never free a partial cluster
2641 * at the beginning of the extent. Instead, we check to see if we
2642 * need to free it on a subsequent call to ext4_remove_blocks,
2643 * or at the end of ext4_ext_rm_leaf or ext4_ext_remove_space.
2645 flags |= EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER;
2646 ext4_free_blocks(handle, inode, NULL, pblk, num, flags);
2648 /* reset the partial cluster if we've freed past it */
2649 if (partial->state != initial && partial->pclu != EXT4_B2C(sbi, pblk))
2650 partial->state = initial;
2653 * If we've freed the entire extent but the beginning is not left
2654 * cluster aligned and is not marked as ineligible for freeing we
2655 * record the partial cluster at the beginning of the extent. It
2656 * wasn't freed by the preceding ext4_free_blocks() call, and we
2657 * need to look farther to the left to determine if it's to be freed
2658 * (not shared with another extent). Else, reset the partial
2659 * cluster - we're either done freeing or the beginning of the
2660 * extent is left cluster aligned.
2662 if (EXT4_LBLK_COFF(sbi, from) && num == ee_len) {
2663 if (partial->state == initial) {
2664 partial->pclu = EXT4_B2C(sbi, pblk);
2665 partial->lblk = from;
2666 partial->state = tofree;
2669 partial->state = initial;
2676 * ext4_ext_rm_leaf() Removes the extents associated with the
2677 * blocks appearing between "start" and "end". Both "start"
2678 * and "end" must appear in the same extent or EIO is returned.
2680 * @handle: The journal handle
2681 * @inode: The files inode
2682 * @path: The path to the leaf
2683 * @partial_cluster: The cluster which we'll have to free if all extents
2684 * has been released from it. However, if this value is
2685 * negative, it's a cluster just to the right of the
2686 * punched region and it must not be freed.
2687 * @start: The first block to remove
2688 * @end: The last block to remove
2691 ext4_ext_rm_leaf(handle_t *handle, struct inode *inode,
2692 struct ext4_ext_path *path,
2693 struct partial_cluster *partial,
2694 ext4_lblk_t start, ext4_lblk_t end)
2696 struct ext4_sb_info *sbi = EXT4_SB(inode->i_sb);
2697 int err = 0, correct_index = 0;
2698 int depth = ext_depth(inode), credits, revoke_credits;
2699 struct ext4_extent_header *eh;
2702 ext4_lblk_t ex_ee_block;
2703 unsigned short ex_ee_len;
2704 unsigned unwritten = 0;
2705 struct ext4_extent *ex;
2708 /* the header must be checked already in ext4_ext_remove_space() */
2709 ext_debug("truncate since %u in leaf to %u\n", start, end);
2710 if (!path[depth].p_hdr)
2711 path[depth].p_hdr = ext_block_hdr(path[depth].p_bh);
2712 eh = path[depth].p_hdr;
2713 if (unlikely(path[depth].p_hdr == NULL)) {
2714 EXT4_ERROR_INODE(inode, "path[%d].p_hdr == NULL", depth);
2715 return -EFSCORRUPTED;
2717 /* find where to start removing */
2718 ex = path[depth].p_ext;
2720 ex = EXT_LAST_EXTENT(eh);
2722 ex_ee_block = le32_to_cpu(ex->ee_block);
2723 ex_ee_len = ext4_ext_get_actual_len(ex);
2725 trace_ext4_ext_rm_leaf(inode, start, ex, partial);
2727 while (ex >= EXT_FIRST_EXTENT(eh) &&
2728 ex_ee_block + ex_ee_len > start) {
2730 if (ext4_ext_is_unwritten(ex))
2735 ext_debug("remove ext %u:[%d]%d\n", ex_ee_block,
2736 unwritten, ex_ee_len);
2737 path[depth].p_ext = ex;
2739 a = ex_ee_block > start ? ex_ee_block : start;
2740 b = ex_ee_block+ex_ee_len - 1 < end ?
2741 ex_ee_block+ex_ee_len - 1 : end;
2743 ext_debug(" border %u:%u\n", a, b);
2745 /* If this extent is beyond the end of the hole, skip it */
2746 if (end < ex_ee_block) {
2748 * We're going to skip this extent and move to another,
2749 * so note that its first cluster is in use to avoid
2750 * freeing it when removing blocks. Eventually, the
2751 * right edge of the truncated/punched region will
2752 * be just to the left.
2754 if (sbi->s_cluster_ratio > 1) {
2755 pblk = ext4_ext_pblock(ex);
2756 partial->pclu = EXT4_B2C(sbi, pblk);
2757 partial->state = nofree;
2760 ex_ee_block = le32_to_cpu(ex->ee_block);
2761 ex_ee_len = ext4_ext_get_actual_len(ex);
2763 } else if (b != ex_ee_block + ex_ee_len - 1) {
2764 EXT4_ERROR_INODE(inode,
2765 "can not handle truncate %u:%u "
2767 start, end, ex_ee_block,
2768 ex_ee_block + ex_ee_len - 1);
2769 err = -EFSCORRUPTED;
2771 } else if (a != ex_ee_block) {
2772 /* remove tail of the extent */
2773 num = a - ex_ee_block;
2775 /* remove whole extent: excellent! */
2779 * 3 for leaf, sb, and inode plus 2 (bmap and group
2780 * descriptor) for each block group; assume two block
2781 * groups plus ex_ee_len/blocks_per_block_group for
2784 credits = 7 + 2*(ex_ee_len/EXT4_BLOCKS_PER_GROUP(inode->i_sb));
2785 if (ex == EXT_FIRST_EXTENT(eh)) {
2787 credits += (ext_depth(inode)) + 1;
2789 credits += EXT4_MAXQUOTAS_TRANS_BLOCKS(inode->i_sb);
2791 * We may end up freeing some index blocks and data from the
2792 * punched range. Note that partial clusters are accounted for
2793 * by ext4_free_data_revoke_credits().
2796 ext4_free_metadata_revoke_credits(inode->i_sb,
2798 ext4_free_data_revoke_credits(inode, b - a + 1);
2800 err = ext4_datasem_ensure_credits(handle, inode, credits,
2801 credits, revoke_credits);
2808 err = ext4_ext_get_access(handle, inode, path + depth);
2812 err = ext4_remove_blocks(handle, inode, ex, partial, a, b);
2817 /* this extent is removed; mark slot entirely unused */
2818 ext4_ext_store_pblock(ex, 0);
2820 ex->ee_len = cpu_to_le16(num);
2822 * Do not mark unwritten if all the blocks in the
2823 * extent have been removed.
2825 if (unwritten && num)
2826 ext4_ext_mark_unwritten(ex);
2828 * If the extent was completely released,
2829 * we need to remove it from the leaf
2832 if (end != EXT_MAX_BLOCKS - 1) {
2834 * For hole punching, we need to scoot all the
2835 * extents up when an extent is removed so that
2836 * we dont have blank extents in the middle
2838 memmove(ex, ex+1, (EXT_LAST_EXTENT(eh) - ex) *
2839 sizeof(struct ext4_extent));
2841 /* Now get rid of the one at the end */
2842 memset(EXT_LAST_EXTENT(eh), 0,
2843 sizeof(struct ext4_extent));
2845 le16_add_cpu(&eh->eh_entries, -1);
2848 err = ext4_ext_dirty(handle, inode, path + depth);
2852 ext_debug("new extent: %u:%u:%llu\n", ex_ee_block, num,
2853 ext4_ext_pblock(ex));
2855 ex_ee_block = le32_to_cpu(ex->ee_block);
2856 ex_ee_len = ext4_ext_get_actual_len(ex);
2859 if (correct_index && eh->eh_entries)
2860 err = ext4_ext_correct_indexes(handle, inode, path);
2863 * If there's a partial cluster and at least one extent remains in
2864 * the leaf, free the partial cluster if it isn't shared with the
2865 * current extent. If it is shared with the current extent
2866 * we reset the partial cluster because we've reached the start of the
2867 * truncated/punched region and we're done removing blocks.
2869 if (partial->state == tofree && ex >= EXT_FIRST_EXTENT(eh)) {
2870 pblk = ext4_ext_pblock(ex) + ex_ee_len - 1;
2871 if (partial->pclu != EXT4_B2C(sbi, pblk)) {
2872 int flags = get_default_free_blocks_flags(inode);
2874 if (ext4_is_pending(inode, partial->lblk))
2875 flags |= EXT4_FREE_BLOCKS_RERESERVE_CLUSTER;
2876 ext4_free_blocks(handle, inode, NULL,
2877 EXT4_C2B(sbi, partial->pclu),
2878 sbi->s_cluster_ratio, flags);
2879 if (flags & EXT4_FREE_BLOCKS_RERESERVE_CLUSTER)
2880 ext4_rereserve_cluster(inode, partial->lblk);
2882 partial->state = initial;
2885 /* if this leaf is free, then we should
2886 * remove it from index block above */
2887 if (err == 0 && eh->eh_entries == 0 && path[depth].p_bh != NULL)
2888 err = ext4_ext_rm_idx(handle, inode, path, depth);
2895 * ext4_ext_more_to_rm:
2896 * returns 1 if current index has to be freed (even partial)
2899 ext4_ext_more_to_rm(struct ext4_ext_path *path)
2901 BUG_ON(path->p_idx == NULL);
2903 if (path->p_idx < EXT_FIRST_INDEX(path->p_hdr))
2907 * if truncate on deeper level happened, it wasn't partial,
2908 * so we have to consider current index for truncation
2910 if (le16_to_cpu(path->p_hdr->eh_entries) == path->p_block)
2915 int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start,
2918 struct ext4_sb_info *sbi = EXT4_SB(inode->i_sb);
2919 int depth = ext_depth(inode);
2920 struct ext4_ext_path *path = NULL;
2921 struct partial_cluster partial;
2927 partial.state = initial;
2929 ext_debug("truncate since %u to %u\n", start, end);
2931 /* probably first extent we're gonna free will be last in block */
2932 handle = ext4_journal_start_with_revoke(inode, EXT4_HT_TRUNCATE,
2934 ext4_free_metadata_revoke_credits(inode->i_sb, depth));
2936 return PTR_ERR(handle);
2939 trace_ext4_ext_remove_space(inode, start, end, depth);
2942 * Check if we are removing extents inside the extent tree. If that
2943 * is the case, we are going to punch a hole inside the extent tree
2944 * so we have to check whether we need to split the extent covering
2945 * the last block to remove so we can easily remove the part of it
2946 * in ext4_ext_rm_leaf().
2948 if (end < EXT_MAX_BLOCKS - 1) {
2949 struct ext4_extent *ex;
2950 ext4_lblk_t ee_block, ex_end, lblk;
2953 /* find extent for or closest extent to this block */
2954 path = ext4_find_extent(inode, end, NULL, EXT4_EX_NOCACHE);
2956 ext4_journal_stop(handle);
2957 return PTR_ERR(path);
2959 depth = ext_depth(inode);
2960 /* Leaf not may not exist only if inode has no blocks at all */
2961 ex = path[depth].p_ext;
2964 EXT4_ERROR_INODE(inode,
2965 "path[%d].p_hdr == NULL",
2967 err = -EFSCORRUPTED;
2972 ee_block = le32_to_cpu(ex->ee_block);
2973 ex_end = ee_block + ext4_ext_get_actual_len(ex) - 1;
2976 * See if the last block is inside the extent, if so split
2977 * the extent at 'end' block so we can easily remove the
2978 * tail of the first part of the split extent in
2979 * ext4_ext_rm_leaf().
2981 if (end >= ee_block && end < ex_end) {
2984 * If we're going to split the extent, note that
2985 * the cluster containing the block after 'end' is
2986 * in use to avoid freeing it when removing blocks.
2988 if (sbi->s_cluster_ratio > 1) {
2989 pblk = ext4_ext_pblock(ex) + end - ee_block + 2;
2990 partial.pclu = EXT4_B2C(sbi, pblk);
2991 partial.state = nofree;
2995 * Split the extent in two so that 'end' is the last
2996 * block in the first new extent. Also we should not
2997 * fail removing space due to ENOSPC so try to use
2998 * reserved block if that happens.
3000 err = ext4_force_split_extent_at(handle, inode, &path,
3005 } else if (sbi->s_cluster_ratio > 1 && end >= ex_end &&
3006 partial.state == initial) {
3008 * If we're punching, there's an extent to the right.
3009 * If the partial cluster hasn't been set, set it to
3010 * that extent's first cluster and its state to nofree
3011 * so it won't be freed should it contain blocks to be
3012 * removed. If it's already set (tofree/nofree), we're
3013 * retrying and keep the original partial cluster info
3014 * so a cluster marked tofree as a result of earlier
3015 * extent removal is not lost.
3018 err = ext4_ext_search_right(inode, path, &lblk, &pblk,
3023 partial.pclu = EXT4_B2C(sbi, pblk);
3024 partial.state = nofree;
3029 * We start scanning from right side, freeing all the blocks
3030 * after i_size and walking into the tree depth-wise.
3032 depth = ext_depth(inode);
3037 le16_to_cpu(path[k].p_hdr->eh_entries)+1;
3039 path = kcalloc(depth + 1, sizeof(struct ext4_ext_path),
3042 ext4_journal_stop(handle);
3045 path[0].p_maxdepth = path[0].p_depth = depth;
3046 path[0].p_hdr = ext_inode_hdr(inode);
3049 if (ext4_ext_check(inode, path[0].p_hdr, depth, 0)) {
3050 err = -EFSCORRUPTED;
3056 while (i >= 0 && err == 0) {
3058 /* this is leaf block */
3059 err = ext4_ext_rm_leaf(handle, inode, path,
3060 &partial, start, end);
3061 /* root level has p_bh == NULL, brelse() eats this */
3062 brelse(path[i].p_bh);
3063 path[i].p_bh = NULL;
3068 /* this is index block */
3069 if (!path[i].p_hdr) {
3070 ext_debug("initialize header\n");
3071 path[i].p_hdr = ext_block_hdr(path[i].p_bh);
3074 if (!path[i].p_idx) {
3075 /* this level hasn't been touched yet */
3076 path[i].p_idx = EXT_LAST_INDEX(path[i].p_hdr);
3077 path[i].p_block = le16_to_cpu(path[i].p_hdr->eh_entries)+1;
3078 ext_debug("init index ptr: hdr 0x%p, num %d\n",
3080 le16_to_cpu(path[i].p_hdr->eh_entries));
3082 /* we were already here, see at next index */
3086 ext_debug("level %d - index, first 0x%p, cur 0x%p\n",
3087 i, EXT_FIRST_INDEX(path[i].p_hdr),
3089 if (ext4_ext_more_to_rm(path + i)) {
3090 struct buffer_head *bh;
3091 /* go to the next level */
3092 ext_debug("move to level %d (block %llu)\n",
3093 i + 1, ext4_idx_pblock(path[i].p_idx));
3094 memset(path + i + 1, 0, sizeof(*path));
3095 bh = read_extent_tree_block(inode,
3096 ext4_idx_pblock(path[i].p_idx), depth - i - 1,
3099 /* should we reset i_size? */
3103 /* Yield here to deal with large extent trees.
3104 * Should be a no-op if we did IO above. */
3106 if (WARN_ON(i + 1 > depth)) {
3107 err = -EFSCORRUPTED;
3110 path[i + 1].p_bh = bh;
3112 /* save actual number of indexes since this
3113 * number is changed at the next iteration */
3114 path[i].p_block = le16_to_cpu(path[i].p_hdr->eh_entries);
3117 /* we finished processing this index, go up */
3118 if (path[i].p_hdr->eh_entries == 0 && i > 0) {
3119 /* index is empty, remove it;
3120 * handle must be already prepared by the
3121 * truncatei_leaf() */
3122 err = ext4_ext_rm_idx(handle, inode, path, i);
3124 /* root level has p_bh == NULL, brelse() eats this */
3125 brelse(path[i].p_bh);
3126 path[i].p_bh = NULL;
3128 ext_debug("return to level %d\n", i);
3132 trace_ext4_ext_remove_space_done(inode, start, end, depth, &partial,
3133 path->p_hdr->eh_entries);
3136 * if there's a partial cluster and we have removed the first extent
3137 * in the file, then we also free the partial cluster, if any
3139 if (partial.state == tofree && err == 0) {
3140 int flags = get_default_free_blocks_flags(inode);
3142 if (ext4_is_pending(inode, partial.lblk))
3143 flags |= EXT4_FREE_BLOCKS_RERESERVE_CLUSTER;
3144 ext4_free_blocks(handle, inode, NULL,
3145 EXT4_C2B(sbi, partial.pclu),
3146 sbi->s_cluster_ratio, flags);
3147 if (flags & EXT4_FREE_BLOCKS_RERESERVE_CLUSTER)
3148 ext4_rereserve_cluster(inode, partial.lblk);
3149 partial.state = initial;
3152 /* TODO: flexible tree reduction should be here */
3153 if (path->p_hdr->eh_entries == 0) {
3155 * truncate to zero freed all the tree,
3156 * so we need to correct eh_depth
3158 err = ext4_ext_get_access(handle, inode, path);
3160 ext_inode_hdr(inode)->eh_depth = 0;
3161 ext_inode_hdr(inode)->eh_max =
3162 cpu_to_le16(ext4_ext_space_root(inode, 0));
3163 err = ext4_ext_dirty(handle, inode, path);
3167 ext4_ext_drop_refs(path);
3172 ext4_journal_stop(handle);
3178 * called at mount time
3180 void ext4_ext_init(struct super_block *sb)
3183 * possible initialization would be here
3186 if (ext4_has_feature_extents(sb)) {
3187 #if defined(AGGRESSIVE_TEST) || defined(CHECK_BINSEARCH) || defined(EXTENTS_STATS)
3188 printk(KERN_INFO "EXT4-fs: file extents enabled"
3189 #ifdef AGGRESSIVE_TEST
3190 ", aggressive tests"
3192 #ifdef CHECK_BINSEARCH
3195 #ifdef EXTENTS_STATS
3200 #ifdef EXTENTS_STATS
3201 spin_lock_init(&EXT4_SB(sb)->s_ext_stats_lock);
3202 EXT4_SB(sb)->s_ext_min = 1 << 30;
3203 EXT4_SB(sb)->s_ext_max = 0;
3209 * called at umount time
3211 void ext4_ext_release(struct super_block *sb)
3213 if (!ext4_has_feature_extents(sb))
3216 #ifdef EXTENTS_STATS
3217 if (EXT4_SB(sb)->s_ext_blocks && EXT4_SB(sb)->s_ext_extents) {
3218 struct ext4_sb_info *sbi = EXT4_SB(sb);
3219 printk(KERN_ERR "EXT4-fs: %lu blocks in %lu extents (%lu ave)\n",
3220 sbi->s_ext_blocks, sbi->s_ext_extents,
3221 sbi->s_ext_blocks / sbi->s_ext_extents);
3222 printk(KERN_ERR "EXT4-fs: extents: %lu min, %lu max, max depth %lu\n",
3223 sbi->s_ext_min, sbi->s_ext_max, sbi->s_depth_max);
3228 static int ext4_zeroout_es(struct inode *inode, struct ext4_extent *ex)
3230 ext4_lblk_t ee_block;
3231 ext4_fsblk_t ee_pblock;
3232 unsigned int ee_len;
3234 ee_block = le32_to_cpu(ex->ee_block);
3235 ee_len = ext4_ext_get_actual_len(ex);
3236 ee_pblock = ext4_ext_pblock(ex);
3241 return ext4_es_insert_extent(inode, ee_block, ee_len, ee_pblock,
3242 EXTENT_STATUS_WRITTEN);
3245 /* FIXME!! we need to try to merge to left or right after zero-out */
3246 static int ext4_ext_zeroout(struct inode *inode, struct ext4_extent *ex)
3248 ext4_fsblk_t ee_pblock;
3249 unsigned int ee_len;
3251 ee_len = ext4_ext_get_actual_len(ex);
3252 ee_pblock = ext4_ext_pblock(ex);
3253 return ext4_issue_zeroout(inode, le32_to_cpu(ex->ee_block), ee_pblock,
3258 * ext4_split_extent_at() splits an extent at given block.
3260 * @handle: the journal handle
3261 * @inode: the file inode
3262 * @path: the path to the extent
3263 * @split: the logical block where the extent is splitted.
3264 * @split_flags: indicates if the extent could be zeroout if split fails, and
3265 * the states(init or unwritten) of new extents.
3266 * @flags: flags used to insert new extent to extent tree.
3269 * Splits extent [a, b] into two extents [a, @split) and [@split, b], states
3270 * of which are deterimined by split_flag.
3272 * There are two cases:
3273 * a> the extent are splitted into two extent.
3274 * b> split is not needed, and just mark the extent.
3276 * return 0 on success.
3278 static int ext4_split_extent_at(handle_t *handle,
3279 struct inode *inode,
3280 struct ext4_ext_path **ppath,
3285 struct ext4_ext_path *path = *ppath;
3286 ext4_fsblk_t newblock;
3287 ext4_lblk_t ee_block;
3288 struct ext4_extent *ex, newex, orig_ex, zero_ex;
3289 struct ext4_extent *ex2 = NULL;
3290 unsigned int ee_len, depth;
3293 BUG_ON((split_flag & (EXT4_EXT_DATA_VALID1 | EXT4_EXT_DATA_VALID2)) ==
3294 (EXT4_EXT_DATA_VALID1 | EXT4_EXT_DATA_VALID2));
3296 ext_debug("ext4_split_extents_at: inode %lu, logical"
3297 "block %llu\n", inode->i_ino, (unsigned long long)split);
3299 ext4_ext_show_leaf(inode, path);
3301 depth = ext_depth(inode);
3302 ex = path[depth].p_ext;
3303 ee_block = le32_to_cpu(ex->ee_block);
3304 ee_len = ext4_ext_get_actual_len(ex);
3305 newblock = split - ee_block + ext4_ext_pblock(ex);
3307 BUG_ON(split < ee_block || split >= (ee_block + ee_len));
3308 BUG_ON(!ext4_ext_is_unwritten(ex) &&
3309 split_flag & (EXT4_EXT_MAY_ZEROOUT |
3310 EXT4_EXT_MARK_UNWRIT1 |
3311 EXT4_EXT_MARK_UNWRIT2));
3313 err = ext4_ext_get_access(handle, inode, path + depth);
3317 if (split == ee_block) {
3319 * case b: block @split is the block that the extent begins with
3320 * then we just change the state of the extent, and splitting
3323 if (split_flag & EXT4_EXT_MARK_UNWRIT2)
3324 ext4_ext_mark_unwritten(ex);
3326 ext4_ext_mark_initialized(ex);
3328 if (!(flags & EXT4_GET_BLOCKS_PRE_IO))
3329 ext4_ext_try_to_merge(handle, inode, path, ex);
3331 err = ext4_ext_dirty(handle, inode, path + path->p_depth);
3336 memcpy(&orig_ex, ex, sizeof(orig_ex));
3337 ex->ee_len = cpu_to_le16(split - ee_block);
3338 if (split_flag & EXT4_EXT_MARK_UNWRIT1)
3339 ext4_ext_mark_unwritten(ex);
3342 * path may lead to new leaf, not to original leaf any more
3343 * after ext4_ext_insert_extent() returns,
3345 err = ext4_ext_dirty(handle, inode, path + depth);
3347 goto fix_extent_len;
3350 ex2->ee_block = cpu_to_le32(split);
3351 ex2->ee_len = cpu_to_le16(ee_len - (split - ee_block));
3352 ext4_ext_store_pblock(ex2, newblock);
3353 if (split_flag & EXT4_EXT_MARK_UNWRIT2)
3354 ext4_ext_mark_unwritten(ex2);
3356 err = ext4_ext_insert_extent(handle, inode, ppath, &newex, flags);
3357 if (err == -ENOSPC && (EXT4_EXT_MAY_ZEROOUT & split_flag)) {
3358 if (split_flag & (EXT4_EXT_DATA_VALID1|EXT4_EXT_DATA_VALID2)) {
3359 if (split_flag & EXT4_EXT_DATA_VALID1) {
3360 err = ext4_ext_zeroout(inode, ex2);
3361 zero_ex.ee_block = ex2->ee_block;
3362 zero_ex.ee_len = cpu_to_le16(
3363 ext4_ext_get_actual_len(ex2));
3364 ext4_ext_store_pblock(&zero_ex,
3365 ext4_ext_pblock(ex2));
3367 err = ext4_ext_zeroout(inode, ex);
3368 zero_ex.ee_block = ex->ee_block;
3369 zero_ex.ee_len = cpu_to_le16(
3370 ext4_ext_get_actual_len(ex));
3371 ext4_ext_store_pblock(&zero_ex,
3372 ext4_ext_pblock(ex));
3375 err = ext4_ext_zeroout(inode, &orig_ex);
3376 zero_ex.ee_block = orig_ex.ee_block;
3377 zero_ex.ee_len = cpu_to_le16(
3378 ext4_ext_get_actual_len(&orig_ex));
3379 ext4_ext_store_pblock(&zero_ex,
3380 ext4_ext_pblock(&orig_ex));
3384 goto fix_extent_len;
3385 /* update the extent length and mark as initialized */
3386 ex->ee_len = cpu_to_le16(ee_len);
3387 ext4_ext_try_to_merge(handle, inode, path, ex);
3388 err = ext4_ext_dirty(handle, inode, path + path->p_depth);
3390 goto fix_extent_len;
3392 /* update extent status tree */
3393 err = ext4_zeroout_es(inode, &zero_ex);
3397 goto fix_extent_len;
3400 ext4_ext_show_leaf(inode, path);
3404 ex->ee_len = orig_ex.ee_len;
3405 ext4_ext_dirty(handle, inode, path + path->p_depth);
3410 * ext4_split_extents() splits an extent and mark extent which is covered
3411 * by @map as split_flags indicates
3413 * It may result in splitting the extent into multiple extents (up to three)
3414 * There are three possibilities:
3415 * a> There is no split required
3416 * b> Splits in two extents: Split is happening at either end of the extent
3417 * c> Splits in three extents: Somone is splitting in middle of the extent
3420 static int ext4_split_extent(handle_t *handle,
3421 struct inode *inode,
3422 struct ext4_ext_path **ppath,
3423 struct ext4_map_blocks *map,
3427 struct ext4_ext_path *path = *ppath;
3428 ext4_lblk_t ee_block;
3429 struct ext4_extent *ex;
3430 unsigned int ee_len, depth;
3433 int split_flag1, flags1;
3434 int allocated = map->m_len;
3436 depth = ext_depth(inode);
3437 ex = path[depth].p_ext;
3438 ee_block = le32_to_cpu(ex->ee_block);
3439 ee_len = ext4_ext_get_actual_len(ex);
3440 unwritten = ext4_ext_is_unwritten(ex);
3442 if (map->m_lblk + map->m_len < ee_block + ee_len) {
3443 split_flag1 = split_flag & EXT4_EXT_MAY_ZEROOUT;
3444 flags1 = flags | EXT4_GET_BLOCKS_PRE_IO;
3446 split_flag1 |= EXT4_EXT_MARK_UNWRIT1 |
3447 EXT4_EXT_MARK_UNWRIT2;
3448 if (split_flag & EXT4_EXT_DATA_VALID2)
3449 split_flag1 |= EXT4_EXT_DATA_VALID1;
3450 err = ext4_split_extent_at(handle, inode, ppath,
3451 map->m_lblk + map->m_len, split_flag1, flags1);
3455 allocated = ee_len - (map->m_lblk - ee_block);
3458 * Update path is required because previous ext4_split_extent_at() may
3459 * result in split of original leaf or extent zeroout.
3461 path = ext4_find_extent(inode, map->m_lblk, ppath, 0);
3463 return PTR_ERR(path);
3464 depth = ext_depth(inode);
3465 ex = path[depth].p_ext;
3467 EXT4_ERROR_INODE(inode, "unexpected hole at %lu",
3468 (unsigned long) map->m_lblk);
3469 return -EFSCORRUPTED;
3471 unwritten = ext4_ext_is_unwritten(ex);
3474 if (map->m_lblk >= ee_block) {
3475 split_flag1 = split_flag & EXT4_EXT_DATA_VALID2;
3477 split_flag1 |= EXT4_EXT_MARK_UNWRIT1;
3478 split_flag1 |= split_flag & (EXT4_EXT_MAY_ZEROOUT |
3479 EXT4_EXT_MARK_UNWRIT2);
3481 err = ext4_split_extent_at(handle, inode, ppath,
3482 map->m_lblk, split_flag1, flags);
3487 ext4_ext_show_leaf(inode, path);
3489 return err ? err : allocated;
3493 * This function is called by ext4_ext_map_blocks() if someone tries to write
3494 * to an unwritten extent. It may result in splitting the unwritten
3495 * extent into multiple extents (up to three - one initialized and two
3497 * There are three possibilities:
3498 * a> There is no split required: Entire extent should be initialized
3499 * b> Splits in two extents: Write is happening at either end of the extent
3500 * c> Splits in three extents: Somone is writing in middle of the extent
3503 * - The extent pointed to by 'path' is unwritten.
3504 * - The extent pointed to by 'path' contains a superset
3505 * of the logical span [map->m_lblk, map->m_lblk + map->m_len).
3507 * Post-conditions on success:
3508 * - the returned value is the number of blocks beyond map->l_lblk
3509 * that are allocated and initialized.
3510 * It is guaranteed to be >= map->m_len.
3512 static int ext4_ext_convert_to_initialized(handle_t *handle,
3513 struct inode *inode,
3514 struct ext4_map_blocks *map,
3515 struct ext4_ext_path **ppath,
3518 struct ext4_ext_path *path = *ppath;
3519 struct ext4_sb_info *sbi;
3520 struct ext4_extent_header *eh;
3521 struct ext4_map_blocks split_map;
3522 struct ext4_extent zero_ex1, zero_ex2;
3523 struct ext4_extent *ex, *abut_ex;
3524 ext4_lblk_t ee_block, eof_block;
3525 unsigned int ee_len, depth, map_len = map->m_len;
3526 int allocated = 0, max_zeroout = 0;
3528 int split_flag = EXT4_EXT_DATA_VALID2;
3530 ext_debug("ext4_ext_convert_to_initialized: inode %lu, logical"
3531 "block %llu, max_blocks %u\n", inode->i_ino,
3532 (unsigned long long)map->m_lblk, map_len);
3534 sbi = EXT4_SB(inode->i_sb);
3535 eof_block = (inode->i_size + inode->i_sb->s_blocksize - 1) >>
3536 inode->i_sb->s_blocksize_bits;
3537 if (eof_block < map->m_lblk + map_len)
3538 eof_block = map->m_lblk + map_len;
3540 depth = ext_depth(inode);
3541 eh = path[depth].p_hdr;
3542 ex = path[depth].p_ext;
3543 ee_block = le32_to_cpu(ex->ee_block);
3544 ee_len = ext4_ext_get_actual_len(ex);
3545 zero_ex1.ee_len = 0;
3546 zero_ex2.ee_len = 0;
3548 trace_ext4_ext_convert_to_initialized_enter(inode, map, ex);
3550 /* Pre-conditions */
3551 BUG_ON(!ext4_ext_is_unwritten(ex));
3552 BUG_ON(!in_range(map->m_lblk, ee_block, ee_len));
3555 * Attempt to transfer newly initialized blocks from the currently
3556 * unwritten extent to its neighbor. This is much cheaper
3557 * than an insertion followed by a merge as those involve costly
3558 * memmove() calls. Transferring to the left is the common case in
3559 * steady state for workloads doing fallocate(FALLOC_FL_KEEP_SIZE)
3560 * followed by append writes.
3562 * Limitations of the current logic:
3563 * - L1: we do not deal with writes covering the whole extent.
3564 * This would require removing the extent if the transfer
3566 * - L2: we only attempt to merge with an extent stored in the
3567 * same extent tree node.
3569 if ((map->m_lblk == ee_block) &&
3570 /* See if we can merge left */
3571 (map_len < ee_len) && /*L1*/
3572 (ex > EXT_FIRST_EXTENT(eh))) { /*L2*/
3573 ext4_lblk_t prev_lblk;
3574 ext4_fsblk_t prev_pblk, ee_pblk;
3575 unsigned int prev_len;
3578 prev_lblk = le32_to_cpu(abut_ex->ee_block);
3579 prev_len = ext4_ext_get_actual_len(abut_ex);
3580 prev_pblk = ext4_ext_pblock(abut_ex);
3581 ee_pblk = ext4_ext_pblock(ex);
3584 * A transfer of blocks from 'ex' to 'abut_ex' is allowed
3585 * upon those conditions:
3586 * - C1: abut_ex is initialized,
3587 * - C2: abut_ex is logically abutting ex,
3588 * - C3: abut_ex is physically abutting ex,
3589 * - C4: abut_ex can receive the additional blocks without
3590 * overflowing the (initialized) length limit.
3592 if ((!ext4_ext_is_unwritten(abut_ex)) && /*C1*/
3593 ((prev_lblk + prev_len) == ee_block) && /*C2*/
3594 ((prev_pblk + prev_len) == ee_pblk) && /*C3*/
3595 (prev_len < (EXT_INIT_MAX_LEN - map_len))) { /*C4*/
3596 err = ext4_ext_get_access(handle, inode, path + depth);
3600 trace_ext4_ext_convert_to_initialized_fastpath(inode,
3603 /* Shift the start of ex by 'map_len' blocks */
3604 ex->ee_block = cpu_to_le32(ee_block + map_len);
3605 ext4_ext_store_pblock(ex, ee_pblk + map_len);
3606 ex->ee_len = cpu_to_le16(ee_len - map_len);
3607 ext4_ext_mark_unwritten(ex); /* Restore the flag */
3609 /* Extend abut_ex by 'map_len' blocks */
3610 abut_ex->ee_len = cpu_to_le16(prev_len + map_len);
3612 /* Result: number of initialized blocks past m_lblk */
3613 allocated = map_len;
3615 } else if (((map->m_lblk + map_len) == (ee_block + ee_len)) &&
3616 (map_len < ee_len) && /*L1*/
3617 ex < EXT_LAST_EXTENT(eh)) { /*L2*/
3618 /* See if we can merge right */
3619 ext4_lblk_t next_lblk;
3620 ext4_fsblk_t next_pblk, ee_pblk;
3621 unsigned int next_len;
3624 next_lblk = le32_to_cpu(abut_ex->ee_block);
3625 next_len = ext4_ext_get_actual_len(abut_ex);
3626 next_pblk = ext4_ext_pblock(abut_ex);
3627 ee_pblk = ext4_ext_pblock(ex);
3630 * A transfer of blocks from 'ex' to 'abut_ex' is allowed
3631 * upon those conditions:
3632 * - C1: abut_ex is initialized,
3633 * - C2: abut_ex is logically abutting ex,
3634 * - C3: abut_ex is physically abutting ex,
3635 * - C4: abut_ex can receive the additional blocks without
3636 * overflowing the (initialized) length limit.
3638 if ((!ext4_ext_is_unwritten(abut_ex)) && /*C1*/
3639 ((map->m_lblk + map_len) == next_lblk) && /*C2*/
3640 ((ee_pblk + ee_len) == next_pblk) && /*C3*/
3641 (next_len < (EXT_INIT_MAX_LEN - map_len))) { /*C4*/
3642 err = ext4_ext_get_access(handle, inode, path + depth);
3646 trace_ext4_ext_convert_to_initialized_fastpath(inode,
3649 /* Shift the start of abut_ex by 'map_len' blocks */
3650 abut_ex->ee_block = cpu_to_le32(next_lblk - map_len);
3651 ext4_ext_store_pblock(abut_ex, next_pblk - map_len);
3652 ex->ee_len = cpu_to_le16(ee_len - map_len);
3653 ext4_ext_mark_unwritten(ex); /* Restore the flag */
3655 /* Extend abut_ex by 'map_len' blocks */
3656 abut_ex->ee_len = cpu_to_le16(next_len + map_len);
3658 /* Result: number of initialized blocks past m_lblk */
3659 allocated = map_len;
3663 /* Mark the block containing both extents as dirty */
3664 ext4_ext_dirty(handle, inode, path + depth);
3666 /* Update path to point to the right extent */
3667 path[depth].p_ext = abut_ex;
3670 allocated = ee_len - (map->m_lblk - ee_block);
3672 WARN_ON(map->m_lblk < ee_block);
3674 * It is safe to convert extent to initialized via explicit
3675 * zeroout only if extent is fully inside i_size or new_size.
3677 split_flag |= ee_block + ee_len <= eof_block ? EXT4_EXT_MAY_ZEROOUT : 0;
3679 if (EXT4_EXT_MAY_ZEROOUT & split_flag)
3680 max_zeroout = sbi->s_extent_max_zeroout_kb >>
3681 (inode->i_sb->s_blocksize_bits - 10);
3685 * 1. split the extent into three extents.
3686 * 2. split the extent into two extents, zeroout the head of the first
3688 * 3. split the extent into two extents, zeroout the tail of the second
3690 * 4. split the extent into two extents with out zeroout.
3691 * 5. no splitting needed, just possibly zeroout the head and / or the
3692 * tail of the extent.
3694 split_map.m_lblk = map->m_lblk;
3695 split_map.m_len = map->m_len;
3697 if (max_zeroout && (allocated > split_map.m_len)) {
3698 if (allocated <= max_zeroout) {
3701 cpu_to_le32(split_map.m_lblk +
3704 cpu_to_le16(allocated - split_map.m_len);
3705 ext4_ext_store_pblock(&zero_ex1,
3706 ext4_ext_pblock(ex) + split_map.m_lblk +
3707 split_map.m_len - ee_block);
3708 err = ext4_ext_zeroout(inode, &zero_ex1);
3711 split_map.m_len = allocated;
3713 if (split_map.m_lblk - ee_block + split_map.m_len <
3716 if (split_map.m_lblk != ee_block) {
3717 zero_ex2.ee_block = ex->ee_block;
3718 zero_ex2.ee_len = cpu_to_le16(split_map.m_lblk -
3720 ext4_ext_store_pblock(&zero_ex2,
3721 ext4_ext_pblock(ex));
3722 err = ext4_ext_zeroout(inode, &zero_ex2);
3727 split_map.m_len += split_map.m_lblk - ee_block;
3728 split_map.m_lblk = ee_block;
3729 allocated = map->m_len;
3733 err = ext4_split_extent(handle, inode, ppath, &split_map, split_flag,
3738 /* If we have gotten a failure, don't zero out status tree */
3740 err = ext4_zeroout_es(inode, &zero_ex1);
3742 err = ext4_zeroout_es(inode, &zero_ex2);
3744 return err ? err : allocated;
3748 * This function is called by ext4_ext_map_blocks() from
3749 * ext4_get_blocks_dio_write() when DIO to write
3750 * to an unwritten extent.
3752 * Writing to an unwritten extent may result in splitting the unwritten
3753 * extent into multiple initialized/unwritten extents (up to three)
3754 * There are three possibilities:
3755 * a> There is no split required: Entire extent should be unwritten
3756 * b> Splits in two extents: Write is happening at either end of the extent
3757 * c> Splits in three extents: Somone is writing in middle of the extent
3759 * This works the same way in the case of initialized -> unwritten conversion.
3761 * One of more index blocks maybe needed if the extent tree grow after
3762 * the unwritten extent split. To prevent ENOSPC occur at the IO
3763 * complete, we need to split the unwritten extent before DIO submit
3764 * the IO. The unwritten extent called at this time will be split
3765 * into three unwritten extent(at most). After IO complete, the part
3766 * being filled will be convert to initialized by the end_io callback function
3767 * via ext4_convert_unwritten_extents().
3769 * Returns the size of unwritten extent to be written on success.
3771 static int ext4_split_convert_extents(handle_t *handle,
3772 struct inode *inode,
3773 struct ext4_map_blocks *map,
3774 struct ext4_ext_path **ppath,
3777 struct ext4_ext_path *path = *ppath;
3778 ext4_lblk_t eof_block;
3779 ext4_lblk_t ee_block;
3780 struct ext4_extent *ex;
3781 unsigned int ee_len;
3782 int split_flag = 0, depth;
3784 ext_debug("%s: inode %lu, logical block %llu, max_blocks %u\n",
3785 __func__, inode->i_ino,
3786 (unsigned long long)map->m_lblk, map->m_len);
3788 eof_block = (inode->i_size + inode->i_sb->s_blocksize - 1) >>
3789 inode->i_sb->s_blocksize_bits;
3790 if (eof_block < map->m_lblk + map->m_len)
3791 eof_block = map->m_lblk + map->m_len;
3793 * It is safe to convert extent to initialized via explicit
3794 * zeroout only if extent is fully insde i_size or new_size.
3796 depth = ext_depth(inode);
3797 ex = path[depth].p_ext;
3798 ee_block = le32_to_cpu(ex->ee_block);
3799 ee_len = ext4_ext_get_actual_len(ex);
3801 /* Convert to unwritten */
3802 if (flags & EXT4_GET_BLOCKS_CONVERT_UNWRITTEN) {
3803 split_flag |= EXT4_EXT_DATA_VALID1;
3804 /* Convert to initialized */
3805 } else if (flags & EXT4_GET_BLOCKS_CONVERT) {
3806 split_flag |= ee_block + ee_len <= eof_block ?
3807 EXT4_EXT_MAY_ZEROOUT : 0;
3808 split_flag |= (EXT4_EXT_MARK_UNWRIT2 | EXT4_EXT_DATA_VALID2);
3810 flags |= EXT4_GET_BLOCKS_PRE_IO;
3811 return ext4_split_extent(handle, inode, ppath, map, split_flag, flags);
3814 static int ext4_convert_unwritten_extents_endio(handle_t *handle,
3815 struct inode *inode,
3816 struct ext4_map_blocks *map,
3817 struct ext4_ext_path **ppath)
3819 struct ext4_ext_path *path = *ppath;
3820 struct ext4_extent *ex;
3821 ext4_lblk_t ee_block;
3822 unsigned int ee_len;
3826 depth = ext_depth(inode);
3827 ex = path[depth].p_ext;
3828 ee_block = le32_to_cpu(ex->ee_block);
3829 ee_len = ext4_ext_get_actual_len(ex);
3831 ext_debug("ext4_convert_unwritten_extents_endio: inode %lu, logical"
3832 "block %llu, max_blocks %u\n", inode->i_ino,
3833 (unsigned long long)ee_block, ee_len);
3835 /* If extent is larger than requested it is a clear sign that we still
3836 * have some extent state machine issues left. So extent_split is still
3838 * TODO: Once all related issues will be fixed this situation should be
3841 if (ee_block != map->m_lblk || ee_len > map->m_len) {
3842 #ifdef CONFIG_EXT4_DEBUG
3843 ext4_warning(inode->i_sb, "Inode (%ld) finished: extent logical block %llu,"
3844 " len %u; IO logical block %llu, len %u",
3845 inode->i_ino, (unsigned long long)ee_block, ee_len,
3846 (unsigned long long)map->m_lblk, map->m_len);
3848 err = ext4_split_convert_extents(handle, inode, map, ppath,
3849 EXT4_GET_BLOCKS_CONVERT);
3852 path = ext4_find_extent(inode, map->m_lblk, ppath, 0);
3854 return PTR_ERR(path);
3855 depth = ext_depth(inode);
3856 ex = path[depth].p_ext;
3859 err = ext4_ext_get_access(handle, inode, path + depth);
3862 /* first mark the extent as initialized */
3863 ext4_ext_mark_initialized(ex);
3865 /* note: ext4_ext_correct_indexes() isn't needed here because
3866 * borders are not changed
3868 ext4_ext_try_to_merge(handle, inode, path, ex);
3870 /* Mark modified extent as dirty */
3871 err = ext4_ext_dirty(handle, inode, path + path->p_depth);
3873 ext4_ext_show_leaf(inode, path);
3878 * Handle EOFBLOCKS_FL flag, clearing it if necessary
3880 static int check_eofblocks_fl(handle_t *handle, struct inode *inode,
3882 struct ext4_ext_path *path,
3886 struct ext4_extent_header *eh;
3887 struct ext4_extent *last_ex;
3889 if (!ext4_test_inode_flag(inode, EXT4_INODE_EOFBLOCKS))
3892 depth = ext_depth(inode);
3893 eh = path[depth].p_hdr;
3896 * We're going to remove EOFBLOCKS_FL entirely in future so we
3897 * do not care for this case anymore. Simply remove the flag
3898 * if there are no extents.
3900 if (unlikely(!eh->eh_entries))
3902 last_ex = EXT_LAST_EXTENT(eh);
3904 * We should clear the EOFBLOCKS_FL flag if we are writing the
3905 * last block in the last extent in the file. We test this by
3906 * first checking to see if the caller to
3907 * ext4_ext_get_blocks() was interested in the last block (or
3908 * a block beyond the last block) in the current extent. If
3909 * this turns out to be false, we can bail out from this
3910 * function immediately.
3912 if (lblk + len < le32_to_cpu(last_ex->ee_block) +
3913 ext4_ext_get_actual_len(last_ex))
3916 * If the caller does appear to be planning to write at or
3917 * beyond the end of the current extent, we then test to see
3918 * if the current extent is the last extent in the file, by
3919 * checking to make sure it was reached via the rightmost node
3920 * at each level of the tree.
3922 for (i = depth-1; i >= 0; i--)
3923 if (path[i].p_idx != EXT_LAST_INDEX(path[i].p_hdr))
3926 ext4_clear_inode_flag(inode, EXT4_INODE_EOFBLOCKS);
3927 return ext4_mark_inode_dirty(handle, inode);
3931 convert_initialized_extent(handle_t *handle, struct inode *inode,
3932 struct ext4_map_blocks *map,
3933 struct ext4_ext_path **ppath,
3934 unsigned int allocated)
3936 struct ext4_ext_path *path = *ppath;
3937 struct ext4_extent *ex;
3938 ext4_lblk_t ee_block;
3939 unsigned int ee_len;
3944 * Make sure that the extent is no bigger than we support with
3947 if (map->m_len > EXT_UNWRITTEN_MAX_LEN)
3948 map->m_len = EXT_UNWRITTEN_MAX_LEN / 2;
3950 depth = ext_depth(inode);
3951 ex = path[depth].p_ext;
3952 ee_block = le32_to_cpu(ex->ee_block);
3953 ee_len = ext4_ext_get_actual_len(ex);
3955 ext_debug("%s: inode %lu, logical"
3956 "block %llu, max_blocks %u\n", __func__, inode->i_ino,
3957 (unsigned long long)ee_block, ee_len);
3959 if (ee_block != map->m_lblk || ee_len > map->m_len) {
3960 err = ext4_split_convert_extents(handle, inode, map, ppath,
3961 EXT4_GET_BLOCKS_CONVERT_UNWRITTEN);
3964 path = ext4_find_extent(inode, map->m_lblk, ppath, 0);
3966 return PTR_ERR(path);
3967 depth = ext_depth(inode);
3968 ex = path[depth].p_ext;
3970 EXT4_ERROR_INODE(inode, "unexpected hole at %lu",
3971 (unsigned long) map->m_lblk);
3972 return -EFSCORRUPTED;
3976 err = ext4_ext_get_access(handle, inode, path + depth);
3979 /* first mark the extent as unwritten */
3980 ext4_ext_mark_unwritten(ex);
3982 /* note: ext4_ext_correct_indexes() isn't needed here because
3983 * borders are not changed
3985 ext4_ext_try_to_merge(handle, inode, path, ex);
3987 /* Mark modified extent as dirty */
3988 err = ext4_ext_dirty(handle, inode, path + path->p_depth);
3991 ext4_ext_show_leaf(inode, path);
3993 ext4_update_inode_fsync_trans(handle, inode, 1);
3994 err = check_eofblocks_fl(handle, inode, map->m_lblk, path, map->m_len);
3997 map->m_flags |= EXT4_MAP_UNWRITTEN;
3998 if (allocated > map->m_len)
3999 allocated = map->m_len;
4000 map->m_len = allocated;
4005 ext4_ext_handle_unwritten_extents(handle_t *handle, struct inode *inode,
4006 struct ext4_map_blocks *map,
4007 struct ext4_ext_path **ppath, int flags,
4008 unsigned int allocated, ext4_fsblk_t newblock)
4010 struct ext4_ext_path *path = *ppath;
4014 ext_debug("ext4_ext_handle_unwritten_extents: inode %lu, logical "
4015 "block %llu, max_blocks %u, flags %x, allocated %u\n",
4016 inode->i_ino, (unsigned long long)map->m_lblk, map->m_len,
4018 ext4_ext_show_leaf(inode, path);
4021 * When writing into unwritten space, we should not fail to
4022 * allocate metadata blocks for the new extent block if needed.
4024 flags |= EXT4_GET_BLOCKS_METADATA_NOFAIL;
4026 trace_ext4_ext_handle_unwritten_extents(inode, map, flags,
4027 allocated, newblock);
4029 /* get_block() before submit the IO, split the extent */
4030 if (flags & EXT4_GET_BLOCKS_PRE_IO) {
4031 ret = ext4_split_convert_extents(handle, inode, map, ppath,
4032 flags | EXT4_GET_BLOCKS_CONVERT);
4035 map->m_flags |= EXT4_MAP_UNWRITTEN;
4038 /* IO end_io complete, convert the filled extent to written */
4039 if (flags & EXT4_GET_BLOCKS_CONVERT) {
4040 if (flags & EXT4_GET_BLOCKS_ZERO) {
4041 if (allocated > map->m_len)
4042 allocated = map->m_len;
4043 err = ext4_issue_zeroout(inode, map->m_lblk, newblock,
4048 ret = ext4_convert_unwritten_extents_endio(handle, inode, map,
4051 ext4_update_inode_fsync_trans(handle, inode, 1);
4052 err = check_eofblocks_fl(handle, inode, map->m_lblk,
4056 map->m_flags |= EXT4_MAP_MAPPED;
4057 map->m_pblk = newblock;
4058 if (allocated > map->m_len)
4059 allocated = map->m_len;
4060 map->m_len = allocated;
4063 /* buffered IO case */
4065 * repeat fallocate creation request
4066 * we already have an unwritten extent
4068 if (flags & EXT4_GET_BLOCKS_UNWRIT_EXT) {
4069 map->m_flags |= EXT4_MAP_UNWRITTEN;
4073 /* buffered READ or buffered write_begin() lookup */
4074 if ((flags & EXT4_GET_BLOCKS_CREATE) == 0) {
4076 * We have blocks reserved already. We
4077 * return allocated blocks so that delalloc
4078 * won't do block reservation for us. But
4079 * the buffer head will be unmapped so that
4080 * a read from the block returns 0s.
4082 map->m_flags |= EXT4_MAP_UNWRITTEN;
4086 /* buffered write, writepage time, convert*/
4087 ret = ext4_ext_convert_to_initialized(handle, inode, map, ppath, flags);
4089 ext4_update_inode_fsync_trans(handle, inode, 1);
4096 map->m_flags |= EXT4_MAP_NEW;
4097 if (allocated > map->m_len)
4098 allocated = map->m_len;
4099 map->m_len = allocated;
4102 map->m_flags |= EXT4_MAP_MAPPED;
4103 if ((flags & EXT4_GET_BLOCKS_KEEP_SIZE) == 0) {
4104 err = check_eofblocks_fl(handle, inode, map->m_lblk, path,
4110 if (allocated > map->m_len)
4111 allocated = map->m_len;
4112 ext4_ext_show_leaf(inode, path);
4113 map->m_pblk = newblock;
4114 map->m_len = allocated;
4116 return err ? err : allocated;
4120 * get_implied_cluster_alloc - check to see if the requested
4121 * allocation (in the map structure) overlaps with a cluster already
4122 * allocated in an extent.
4123 * @sb The filesystem superblock structure
4124 * @map The requested lblk->pblk mapping
4125 * @ex The extent structure which might contain an implied
4126 * cluster allocation
4128 * This function is called by ext4_ext_map_blocks() after we failed to
4129 * find blocks that were already in the inode's extent tree. Hence,
4130 * we know that the beginning of the requested region cannot overlap
4131 * the extent from the inode's extent tree. There are three cases we
4132 * want to catch. The first is this case:
4134 * |--- cluster # N--|
4135 * |--- extent ---| |---- requested region ---|
4138 * The second case that we need to test for is this one:
4140 * |--------- cluster # N ----------------|
4141 * |--- requested region --| |------- extent ----|
4142 * |=======================|
4144 * The third case is when the requested region lies between two extents
4145 * within the same cluster:
4146 * |------------- cluster # N-------------|
4147 * |----- ex -----| |---- ex_right ----|
4148 * |------ requested region ------|
4149 * |================|
4151 * In each of the above cases, we need to set the map->m_pblk and
4152 * map->m_len so it corresponds to the return the extent labelled as
4153 * "|====|" from cluster #N, since it is already in use for data in
4154 * cluster EXT4_B2C(sbi, map->m_lblk). We will then return 1 to
4155 * signal to ext4_ext_map_blocks() that map->m_pblk should be treated
4156 * as a new "allocated" block region. Otherwise, we will return 0 and
4157 * ext4_ext_map_blocks() will then allocate one or more new clusters
4158 * by calling ext4_mb_new_blocks().
4160 static int get_implied_cluster_alloc(struct super_block *sb,
4161 struct ext4_map_blocks *map,
4162 struct ext4_extent *ex,
4163 struct ext4_ext_path *path)
4165 struct ext4_sb_info *sbi = EXT4_SB(sb);
4166 ext4_lblk_t c_offset = EXT4_LBLK_COFF(sbi, map->m_lblk);
4167 ext4_lblk_t ex_cluster_start, ex_cluster_end;
4168 ext4_lblk_t rr_cluster_start;
4169 ext4_lblk_t ee_block = le32_to_cpu(ex->ee_block);
4170 ext4_fsblk_t ee_start = ext4_ext_pblock(ex);
4171 unsigned short ee_len = ext4_ext_get_actual_len(ex);
4173 /* The extent passed in that we are trying to match */
4174 ex_cluster_start = EXT4_B2C(sbi, ee_block);
4175 ex_cluster_end = EXT4_B2C(sbi, ee_block + ee_len - 1);
4177 /* The requested region passed into ext4_map_blocks() */
4178 rr_cluster_start = EXT4_B2C(sbi, map->m_lblk);
4180 if ((rr_cluster_start == ex_cluster_end) ||
4181 (rr_cluster_start == ex_cluster_start)) {
4182 if (rr_cluster_start == ex_cluster_end)
4183 ee_start += ee_len - 1;
4184 map->m_pblk = EXT4_PBLK_CMASK(sbi, ee_start) + c_offset;
4185 map->m_len = min(map->m_len,
4186 (unsigned) sbi->s_cluster_ratio - c_offset);
4188 * Check for and handle this case:
4190 * |--------- cluster # N-------------|
4191 * |------- extent ----|
4192 * |--- requested region ---|
4196 if (map->m_lblk < ee_block)
4197 map->m_len = min(map->m_len, ee_block - map->m_lblk);
4200 * Check for the case where there is already another allocated
4201 * block to the right of 'ex' but before the end of the cluster.
4203 * |------------- cluster # N-------------|
4204 * |----- ex -----| |---- ex_right ----|
4205 * |------ requested region ------|
4206 * |================|
4208 if (map->m_lblk > ee_block) {
4209 ext4_lblk_t next = ext4_ext_next_allocated_block(path);
4210 map->m_len = min(map->m_len, next - map->m_lblk);
4213 trace_ext4_get_implied_cluster_alloc_exit(sb, map, 1);
4217 trace_ext4_get_implied_cluster_alloc_exit(sb, map, 0);
4223 * Block allocation/map/preallocation routine for extents based files
4226 * Need to be called with
4227 * down_read(&EXT4_I(inode)->i_data_sem) if not allocating file system block
4228 * (ie, create is zero). Otherwise down_write(&EXT4_I(inode)->i_data_sem)
4230 * return > 0, number of of blocks already mapped/allocated
4231 * if create == 0 and these are pre-allocated blocks
4232 * buffer head is unmapped
4233 * otherwise blocks are mapped
4235 * return = 0, if plain look up failed (blocks have not been allocated)
4236 * buffer head is unmapped
4238 * return < 0, error case.
4240 int ext4_ext_map_blocks(handle_t *handle, struct inode *inode,
4241 struct ext4_map_blocks *map, int flags)
4243 struct ext4_ext_path *path = NULL;
4244 struct ext4_extent newex, *ex, *ex2;
4245 struct ext4_sb_info *sbi = EXT4_SB(inode->i_sb);
4246 ext4_fsblk_t newblock = 0;
4247 int free_on_err = 0, err = 0, depth, ret;
4248 unsigned int allocated = 0, offset = 0;
4249 unsigned int allocated_clusters = 0;
4250 struct ext4_allocation_request ar;
4251 ext4_lblk_t cluster_offset;
4252 bool map_from_cluster = false;
4254 ext_debug("blocks %u/%u requested for inode %lu\n",
4255 map->m_lblk, map->m_len, inode->i_ino);
4256 trace_ext4_ext_map_blocks_enter(inode, map->m_lblk, map->m_len, flags);
4258 /* find extent for this block */
4259 path = ext4_find_extent(inode, map->m_lblk, NULL, 0);
4261 err = PTR_ERR(path);
4266 depth = ext_depth(inode);
4269 * consistent leaf must not be empty;
4270 * this situation is possible, though, _during_ tree modification;
4271 * this is why assert can't be put in ext4_find_extent()
4273 if (unlikely(path[depth].p_ext == NULL && depth != 0)) {
4274 EXT4_ERROR_INODE(inode, "bad extent address "
4275 "lblock: %lu, depth: %d pblock %lld",
4276 (unsigned long) map->m_lblk, depth,
4277 path[depth].p_block);
4278 err = -EFSCORRUPTED;
4282 ex = path[depth].p_ext;
4284 ext4_lblk_t ee_block = le32_to_cpu(ex->ee_block);
4285 ext4_fsblk_t ee_start = ext4_ext_pblock(ex);
4286 unsigned short ee_len;
4290 * unwritten extents are treated as holes, except that
4291 * we split out initialized portions during a write.
4293 ee_len = ext4_ext_get_actual_len(ex);
4295 trace_ext4_ext_show_extent(inode, ee_block, ee_start, ee_len);
4297 /* if found extent covers block, simply return it */
4298 if (in_range(map->m_lblk, ee_block, ee_len)) {
4299 newblock = map->m_lblk - ee_block + ee_start;
4300 /* number of remaining blocks in the extent */
4301 allocated = ee_len - (map->m_lblk - ee_block);
4302 ext_debug("%u fit into %u:%d -> %llu\n", map->m_lblk,
4303 ee_block, ee_len, newblock);
4306 * If the extent is initialized check whether the
4307 * caller wants to convert it to unwritten.
4309 if ((!ext4_ext_is_unwritten(ex)) &&
4310 (flags & EXT4_GET_BLOCKS_CONVERT_UNWRITTEN)) {
4311 allocated = convert_initialized_extent(
4312 handle, inode, map, &path,
4315 } else if (!ext4_ext_is_unwritten(ex))
4318 ret = ext4_ext_handle_unwritten_extents(
4319 handle, inode, map, &path, flags,
4320 allocated, newblock);
4330 * requested block isn't allocated yet;
4331 * we couldn't try to create block if create flag is zero
4333 if ((flags & EXT4_GET_BLOCKS_CREATE) == 0) {
4334 ext4_lblk_t hole_start, hole_len;
4336 hole_start = map->m_lblk;
4337 hole_len = ext4_ext_determine_hole(inode, path, &hole_start);
4339 * put just found gap into cache to speed up
4340 * subsequent requests
4342 ext4_ext_put_gap_in_cache(inode, hole_start, hole_len);
4344 /* Update hole_len to reflect hole size after map->m_lblk */
4345 if (hole_start != map->m_lblk)
4346 hole_len -= map->m_lblk - hole_start;
4348 map->m_len = min_t(unsigned int, map->m_len, hole_len);
4354 * Okay, we need to do block allocation.
4356 newex.ee_block = cpu_to_le32(map->m_lblk);
4357 cluster_offset = EXT4_LBLK_COFF(sbi, map->m_lblk);
4360 * If we are doing bigalloc, check to see if the extent returned
4361 * by ext4_find_extent() implies a cluster we can use.
4363 if (cluster_offset && ex &&
4364 get_implied_cluster_alloc(inode->i_sb, map, ex, path)) {
4365 ar.len = allocated = map->m_len;
4366 newblock = map->m_pblk;
4367 map_from_cluster = true;
4368 goto got_allocated_blocks;
4371 /* find neighbour allocated blocks */
4372 ar.lleft = map->m_lblk;
4373 err = ext4_ext_search_left(inode, path, &ar.lleft, &ar.pleft);
4376 ar.lright = map->m_lblk;
4378 err = ext4_ext_search_right(inode, path, &ar.lright, &ar.pright, &ex2);
4382 /* Check if the extent after searching to the right implies a
4383 * cluster we can use. */
4384 if ((sbi->s_cluster_ratio > 1) && ex2 &&
4385 get_implied_cluster_alloc(inode->i_sb, map, ex2, path)) {
4386 ar.len = allocated = map->m_len;
4387 newblock = map->m_pblk;
4388 map_from_cluster = true;
4389 goto got_allocated_blocks;
4393 * See if request is beyond maximum number of blocks we can have in
4394 * a single extent. For an initialized extent this limit is
4395 * EXT_INIT_MAX_LEN and for an unwritten extent this limit is
4396 * EXT_UNWRITTEN_MAX_LEN.
4398 if (map->m_len > EXT_INIT_MAX_LEN &&
4399 !(flags & EXT4_GET_BLOCKS_UNWRIT_EXT))
4400 map->m_len = EXT_INIT_MAX_LEN;
4401 else if (map->m_len > EXT_UNWRITTEN_MAX_LEN &&
4402 (flags & EXT4_GET_BLOCKS_UNWRIT_EXT))
4403 map->m_len = EXT_UNWRITTEN_MAX_LEN;
4405 /* Check if we can really insert (m_lblk)::(m_lblk + m_len) extent */
4406 newex.ee_len = cpu_to_le16(map->m_len);
4407 err = ext4_ext_check_overlap(sbi, inode, &newex, path);
4409 allocated = ext4_ext_get_actual_len(&newex);
4411 allocated = map->m_len;
4413 /* allocate new block */
4415 ar.goal = ext4_ext_find_goal(inode, path, map->m_lblk);
4416 ar.logical = map->m_lblk;
4418 * We calculate the offset from the beginning of the cluster
4419 * for the logical block number, since when we allocate a
4420 * physical cluster, the physical block should start at the
4421 * same offset from the beginning of the cluster. This is
4422 * needed so that future calls to get_implied_cluster_alloc()
4425 offset = EXT4_LBLK_COFF(sbi, map->m_lblk);
4426 ar.len = EXT4_NUM_B2C(sbi, offset+allocated);
4428 ar.logical -= offset;
4429 if (S_ISREG(inode->i_mode))
4430 ar.flags = EXT4_MB_HINT_DATA;
4432 /* disable in-core preallocation for non-regular files */
4434 if (flags & EXT4_GET_BLOCKS_NO_NORMALIZE)
4435 ar.flags |= EXT4_MB_HINT_NOPREALLOC;
4436 if (flags & EXT4_GET_BLOCKS_DELALLOC_RESERVE)
4437 ar.flags |= EXT4_MB_DELALLOC_RESERVED;
4438 if (flags & EXT4_GET_BLOCKS_METADATA_NOFAIL)
4439 ar.flags |= EXT4_MB_USE_RESERVED;
4440 newblock = ext4_mb_new_blocks(handle, &ar, &err);
4443 ext_debug("allocate new block: goal %llu, found %llu/%u\n",
4444 ar.goal, newblock, allocated);
4446 allocated_clusters = ar.len;
4447 ar.len = EXT4_C2B(sbi, ar.len) - offset;
4448 if (ar.len > allocated)
4451 got_allocated_blocks:
4452 /* try to insert new extent into found leaf and return */
4453 ext4_ext_store_pblock(&newex, newblock + offset);
4454 newex.ee_len = cpu_to_le16(ar.len);
4455 /* Mark unwritten */
4456 if (flags & EXT4_GET_BLOCKS_UNWRIT_EXT){
4457 ext4_ext_mark_unwritten(&newex);
4458 map->m_flags |= EXT4_MAP_UNWRITTEN;
4462 if ((flags & EXT4_GET_BLOCKS_KEEP_SIZE) == 0)
4463 err = check_eofblocks_fl(handle, inode, map->m_lblk,
4466 err = ext4_ext_insert_extent(handle, inode, &path,
4469 if (err && free_on_err) {
4470 int fb_flags = flags & EXT4_GET_BLOCKS_DELALLOC_RESERVE ?
4471 EXT4_FREE_BLOCKS_NO_QUOT_UPDATE : 0;
4472 /* free data blocks we just allocated */
4473 /* not a good idea to call discard here directly,
4474 * but otherwise we'd need to call it every free() */
4475 ext4_discard_preallocations(inode);
4476 ext4_free_blocks(handle, inode, NULL, newblock,
4477 EXT4_C2B(sbi, allocated_clusters), fb_flags);
4481 /* previous routine could use block we allocated */
4482 newblock = ext4_ext_pblock(&newex);
4483 allocated = ext4_ext_get_actual_len(&newex);
4484 if (allocated > map->m_len)
4485 allocated = map->m_len;
4486 map->m_flags |= EXT4_MAP_NEW;
4489 * Reduce the reserved cluster count to reflect successful deferred
4490 * allocation of delayed allocated clusters or direct allocation of
4491 * clusters discovered to be delayed allocated. Once allocated, a
4492 * cluster is not included in the reserved count.
4494 if (test_opt(inode->i_sb, DELALLOC) && !map_from_cluster) {
4495 if (flags & EXT4_GET_BLOCKS_DELALLOC_RESERVE) {
4497 * When allocating delayed allocated clusters, simply
4498 * reduce the reserved cluster count and claim quota
4500 ext4_da_update_reserve_space(inode, allocated_clusters,
4503 ext4_lblk_t lblk, len;
4507 * When allocating non-delayed allocated clusters
4508 * (from fallocate, filemap, DIO, or clusters
4509 * allocated when delalloc has been disabled by
4510 * ext4_nonda_switch), reduce the reserved cluster
4511 * count by the number of allocated clusters that
4512 * have previously been delayed allocated. Quota
4513 * has been claimed by ext4_mb_new_blocks() above,
4514 * so release the quota reservations made for any
4515 * previously delayed allocated clusters.
4517 lblk = EXT4_LBLK_CMASK(sbi, map->m_lblk);
4518 len = allocated_clusters << sbi->s_cluster_bits;
4519 n = ext4_es_delayed_clu(inode, lblk, len);
4521 ext4_da_update_reserve_space(inode, (int) n, 0);
4526 * Cache the extent and update transaction to commit on fdatasync only
4527 * when it is _not_ an unwritten extent.
4529 if ((flags & EXT4_GET_BLOCKS_UNWRIT_EXT) == 0)
4530 ext4_update_inode_fsync_trans(handle, inode, 1);
4532 ext4_update_inode_fsync_trans(handle, inode, 0);
4534 if (allocated > map->m_len)
4535 allocated = map->m_len;
4536 ext4_ext_show_leaf(inode, path);
4537 map->m_flags |= EXT4_MAP_MAPPED;
4538 map->m_pblk = newblock;
4539 map->m_len = allocated;
4541 ext4_ext_drop_refs(path);
4544 trace_ext4_ext_map_blocks_exit(inode, flags, map,
4545 err ? err : allocated);
4546 return err ? err : allocated;
4549 int ext4_ext_truncate(handle_t *handle, struct inode *inode)
4551 struct super_block *sb = inode->i_sb;
4552 ext4_lblk_t last_block;
4556 * TODO: optimization is possible here.
4557 * Probably we need not scan at all,
4558 * because page truncation is enough.
4561 /* we have to know where to truncate from in crash case */
4562 EXT4_I(inode)->i_disksize = inode->i_size;
4563 err = ext4_mark_inode_dirty(handle, inode);
4567 last_block = (inode->i_size + sb->s_blocksize - 1)
4568 >> EXT4_BLOCK_SIZE_BITS(sb);
4570 err = ext4_es_remove_extent(inode, last_block,
4571 EXT_MAX_BLOCKS - last_block);
4572 if (err == -ENOMEM) {
4574 congestion_wait(BLK_RW_ASYNC, HZ/50);
4579 return ext4_ext_remove_space(inode, last_block, EXT_MAX_BLOCKS - 1);
4582 static int ext4_alloc_file_blocks(struct file *file, ext4_lblk_t offset,
4583 ext4_lblk_t len, loff_t new_size,
4586 struct inode *inode = file_inode(file);
4592 struct ext4_map_blocks map;
4593 unsigned int credits;
4596 BUG_ON(!ext4_test_inode_flag(inode, EXT4_INODE_EXTENTS));
4597 map.m_lblk = offset;
4600 * Don't normalize the request if it can fit in one extent so
4601 * that it doesn't get unnecessarily split into multiple
4604 if (len <= EXT_UNWRITTEN_MAX_LEN)
4605 flags |= EXT4_GET_BLOCKS_NO_NORMALIZE;
4608 * credits to insert 1 extent into extent tree
4610 credits = ext4_chunk_trans_blocks(inode, len);
4611 depth = ext_depth(inode);
4614 while (ret >= 0 && len) {
4616 * Recalculate credits when extent tree depth changes.
4618 if (depth != ext_depth(inode)) {
4619 credits = ext4_chunk_trans_blocks(inode, len);
4620 depth = ext_depth(inode);
4623 handle = ext4_journal_start(inode, EXT4_HT_MAP_BLOCKS,
4625 if (IS_ERR(handle)) {
4626 ret = PTR_ERR(handle);
4629 ret = ext4_map_blocks(handle, inode, &map, flags);
4631 ext4_debug("inode #%lu: block %u: len %u: "
4632 "ext4_ext_map_blocks returned %d",
4633 inode->i_ino, map.m_lblk,
4635 ext4_mark_inode_dirty(handle, inode);
4636 ret2 = ext4_journal_stop(handle);
4640 map.m_len = len = len - ret;
4641 epos = (loff_t)map.m_lblk << inode->i_blkbits;
4642 inode->i_ctime = current_time(inode);
4644 if (epos > new_size)
4646 if (ext4_update_inode_size(inode, epos) & 0x1)
4647 inode->i_mtime = inode->i_ctime;
4649 if (epos > inode->i_size)
4650 ext4_set_inode_flag(inode,
4651 EXT4_INODE_EOFBLOCKS);
4653 ext4_mark_inode_dirty(handle, inode);
4654 ext4_update_inode_fsync_trans(handle, inode, 1);
4655 ret2 = ext4_journal_stop(handle);
4659 if (ret == -ENOSPC &&
4660 ext4_should_retry_alloc(inode->i_sb, &retries)) {
4665 return ret > 0 ? ret2 : ret;
4668 static int ext4_collapse_range(struct inode *inode, loff_t offset, loff_t len);
4670 static int ext4_insert_range(struct inode *inode, loff_t offset, loff_t len);
4672 static long ext4_zero_range(struct file *file, loff_t offset,
4673 loff_t len, int mode)
4675 struct inode *inode = file_inode(file);
4676 handle_t *handle = NULL;
4677 unsigned int max_blocks;
4678 loff_t new_size = 0;
4682 int partial_begin, partial_end;
4685 unsigned int blkbits = inode->i_blkbits;
4687 trace_ext4_zero_range(inode, offset, len, mode);
4689 /* Call ext4_force_commit to flush all data in case of data=journal. */
4690 if (ext4_should_journal_data(inode)) {
4691 ret = ext4_force_commit(inode->i_sb);
4697 * Round up offset. This is not fallocate, we neet to zero out
4698 * blocks, so convert interior block aligned part of the range to
4699 * unwritten and possibly manually zero out unaligned parts of the
4702 start = round_up(offset, 1 << blkbits);
4703 end = round_down((offset + len), 1 << blkbits);
4705 if (start < offset || end > offset + len)
4707 partial_begin = offset & ((1 << blkbits) - 1);
4708 partial_end = (offset + len) & ((1 << blkbits) - 1);
4710 lblk = start >> blkbits;
4711 max_blocks = (end >> blkbits);
4712 if (max_blocks < lblk)
4720 * Indirect files do not support unwritten extnets
4722 if (!(ext4_test_inode_flag(inode, EXT4_INODE_EXTENTS))) {
4727 if (!(mode & FALLOC_FL_KEEP_SIZE) &&
4728 (offset + len > inode->i_size ||
4729 offset + len > EXT4_I(inode)->i_disksize)) {
4730 new_size = offset + len;
4731 ret = inode_newsize_ok(inode, new_size);
4736 flags = EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT;
4737 if (mode & FALLOC_FL_KEEP_SIZE)
4738 flags |= EXT4_GET_BLOCKS_KEEP_SIZE;
4740 /* Wait all existing dio workers, newcomers will block on i_mutex */
4741 inode_dio_wait(inode);
4743 /* Preallocate the range including the unaligned edges */
4744 if (partial_begin || partial_end) {
4745 ret = ext4_alloc_file_blocks(file,
4746 round_down(offset, 1 << blkbits) >> blkbits,
4747 (round_up((offset + len), 1 << blkbits) -
4748 round_down(offset, 1 << blkbits)) >> blkbits,
4755 /* Zero range excluding the unaligned edges */
4756 if (max_blocks > 0) {
4757 flags |= (EXT4_GET_BLOCKS_CONVERT_UNWRITTEN |
4761 * Prevent page faults from reinstantiating pages we have
4762 * released from page cache.
4764 down_write(&EXT4_I(inode)->i_mmap_sem);
4766 ret = ext4_break_layouts(inode);
4768 up_write(&EXT4_I(inode)->i_mmap_sem);
4772 ret = ext4_update_disksize_before_punch(inode, offset, len);
4774 up_write(&EXT4_I(inode)->i_mmap_sem);
4777 /* Now release the pages and zero block aligned part of pages */
4778 truncate_pagecache_range(inode, start, end - 1);
4779 inode->i_mtime = inode->i_ctime = current_time(inode);
4781 ret = ext4_alloc_file_blocks(file, lblk, max_blocks, new_size,
4783 up_write(&EXT4_I(inode)->i_mmap_sem);
4787 if (!partial_begin && !partial_end)
4791 * In worst case we have to writeout two nonadjacent unwritten
4792 * blocks and update the inode
4794 credits = (2 * ext4_ext_index_trans_blocks(inode, 2)) + 1;
4795 if (ext4_should_journal_data(inode))
4797 handle = ext4_journal_start(inode, EXT4_HT_MISC, credits);
4798 if (IS_ERR(handle)) {
4799 ret = PTR_ERR(handle);
4800 ext4_std_error(inode->i_sb, ret);
4804 inode->i_mtime = inode->i_ctime = current_time(inode);
4806 ext4_update_inode_size(inode, new_size);
4809 * Mark that we allocate beyond EOF so the subsequent truncate
4810 * can proceed even if the new size is the same as i_size.
4812 if (offset + len > inode->i_size)
4813 ext4_set_inode_flag(inode, EXT4_INODE_EOFBLOCKS);
4815 ext4_mark_inode_dirty(handle, inode);
4817 /* Zero out partial block at the edges of the range */
4818 ret = ext4_zero_partial_blocks(handle, inode, offset, len);
4820 ext4_update_inode_fsync_trans(handle, inode, 1);
4822 if (file->f_flags & O_SYNC)
4823 ext4_handle_sync(handle);
4825 ext4_journal_stop(handle);
4827 inode_unlock(inode);
4832 * preallocate space for a file. This implements ext4's fallocate file
4833 * operation, which gets called from sys_fallocate system call.
4834 * For block-mapped files, posix_fallocate should fall back to the method
4835 * of writing zeroes to the required new blocks (the same behavior which is
4836 * expected for file systems which do not support fallocate() system call).
4838 long ext4_fallocate(struct file *file, int mode, loff_t offset, loff_t len)
4840 struct inode *inode = file_inode(file);
4841 loff_t new_size = 0;
4842 unsigned int max_blocks;
4846 unsigned int blkbits = inode->i_blkbits;
4849 * Encrypted inodes can't handle collapse range or insert
4850 * range since we would need to re-encrypt blocks with a
4851 * different IV or XTS tweak (which are based on the logical
4854 if (IS_ENCRYPTED(inode) &&
4855 (mode & (FALLOC_FL_COLLAPSE_RANGE | FALLOC_FL_INSERT_RANGE)))
4858 /* Return error if mode is not supported */
4859 if (mode & ~(FALLOC_FL_KEEP_SIZE | FALLOC_FL_PUNCH_HOLE |
4860 FALLOC_FL_COLLAPSE_RANGE | FALLOC_FL_ZERO_RANGE |
4861 FALLOC_FL_INSERT_RANGE))
4864 if (mode & FALLOC_FL_PUNCH_HOLE)
4865 return ext4_punch_hole(inode, offset, len);
4867 ret = ext4_convert_inline_data(inode);
4871 if (mode & FALLOC_FL_COLLAPSE_RANGE)
4872 return ext4_collapse_range(inode, offset, len);
4874 if (mode & FALLOC_FL_INSERT_RANGE)
4875 return ext4_insert_range(inode, offset, len);
4877 if (mode & FALLOC_FL_ZERO_RANGE)
4878 return ext4_zero_range(file, offset, len, mode);
4880 trace_ext4_fallocate_enter(inode, offset, len, mode);
4881 lblk = offset >> blkbits;
4883 max_blocks = EXT4_MAX_BLOCKS(len, offset, blkbits);
4884 flags = EXT4_GET_BLOCKS_CREATE_UNWRIT_EXT;
4885 if (mode & FALLOC_FL_KEEP_SIZE)
4886 flags |= EXT4_GET_BLOCKS_KEEP_SIZE;
4891 * We only support preallocation for extent-based files only
4893 if (!(ext4_test_inode_flag(inode, EXT4_INODE_EXTENTS))) {
4898 if (!(mode & FALLOC_FL_KEEP_SIZE) &&
4899 (offset + len > inode->i_size ||
4900 offset + len > EXT4_I(inode)->i_disksize)) {
4901 new_size = offset + len;
4902 ret = inode_newsize_ok(inode, new_size);
4907 /* Wait all existing dio workers, newcomers will block on i_mutex */
4908 inode_dio_wait(inode);
4910 ret = ext4_alloc_file_blocks(file, lblk, max_blocks, new_size, flags);
4914 if (file->f_flags & O_SYNC && EXT4_SB(inode->i_sb)->s_journal) {
4915 ret = jbd2_complete_transaction(EXT4_SB(inode->i_sb)->s_journal,
4916 EXT4_I(inode)->i_sync_tid);
4919 inode_unlock(inode);
4920 trace_ext4_fallocate_exit(inode, offset, max_blocks, ret);
4925 * This function convert a range of blocks to written extents
4926 * The caller of this function will pass the start offset and the size.
4927 * all unwritten extents within this range will be converted to
4930 * This function is called from the direct IO end io call back
4931 * function, to convert the fallocated extents after IO is completed.
4932 * Returns 0 on success.
4934 int ext4_convert_unwritten_extents(handle_t *handle, struct inode *inode,
4935 loff_t offset, ssize_t len)
4937 unsigned int max_blocks;
4940 struct ext4_map_blocks map;
4941 unsigned int blkbits = inode->i_blkbits;
4942 unsigned int credits = 0;
4944 map.m_lblk = offset >> blkbits;
4945 max_blocks = EXT4_MAX_BLOCKS(len, offset, blkbits);
4949 * credits to insert 1 extent into extent tree
4951 credits = ext4_chunk_trans_blocks(inode, max_blocks);
4953 while (ret >= 0 && ret < max_blocks) {
4955 map.m_len = (max_blocks -= ret);
4957 handle = ext4_journal_start(inode, EXT4_HT_MAP_BLOCKS,
4959 if (IS_ERR(handle)) {
4960 ret = PTR_ERR(handle);
4964 ret = ext4_map_blocks(handle, inode, &map,
4965 EXT4_GET_BLOCKS_IO_CONVERT_EXT);
4967 ext4_warning(inode->i_sb,
4968 "inode #%lu: block %u: len %u: "
4969 "ext4_ext_map_blocks returned %d",
4970 inode->i_ino, map.m_lblk,
4972 ext4_mark_inode_dirty(handle, inode);
4974 ret2 = ext4_journal_stop(handle);
4975 if (ret <= 0 || ret2)
4978 return ret > 0 ? ret2 : ret;
4981 int ext4_convert_unwritten_io_end_vec(handle_t *handle, ext4_io_end_t *io_end)
4984 struct ext4_io_end_vec *io_end_vec;
4987 * This is somewhat ugly but the idea is clear: When transaction is
4988 * reserved, everything goes into it. Otherwise we rather start several
4989 * smaller transactions for conversion of each extent separately.
4992 handle = ext4_journal_start_reserved(handle,
4993 EXT4_HT_EXT_CONVERT);
4995 return PTR_ERR(handle);
4998 list_for_each_entry(io_end_vec, &io_end->list_vec, list) {
4999 ret = ext4_convert_unwritten_extents(handle, io_end->inode,
5007 err = ext4_journal_stop(handle);
5009 return ret < 0 ? ret : err;
5013 * If newes is not existing extent (newes->ec_pblk equals zero) find
5014 * delayed extent at start of newes and update newes accordingly and
5015 * return start of the next delayed extent.
5017 * If newes is existing extent (newes->ec_pblk is not equal zero)
5018 * return start of next delayed extent or EXT_MAX_BLOCKS if no delayed
5019 * extent found. Leave newes unmodified.
5021 static int ext4_find_delayed_extent(struct inode *inode,
5022 struct extent_status *newes)
5024 struct extent_status es;
5025 ext4_lblk_t block, next_del;
5027 if (newes->es_pblk == 0) {
5028 ext4_es_find_extent_range(inode, &ext4_es_is_delayed,
5030 newes->es_lblk + newes->es_len - 1,
5034 * No extent in extent-tree contains block @newes->es_pblk,
5035 * then the block may stay in 1)a hole or 2)delayed-extent.
5041 if (es.es_lblk > newes->es_lblk) {
5043 newes->es_len = min(es.es_lblk - newes->es_lblk,
5048 newes->es_len = es.es_lblk + es.es_len - newes->es_lblk;
5051 block = newes->es_lblk + newes->es_len;
5052 ext4_es_find_extent_range(inode, &ext4_es_is_delayed, block,
5053 EXT_MAX_BLOCKS, &es);
5055 next_del = EXT_MAX_BLOCKS;
5057 next_del = es.es_lblk;
5062 static int ext4_xattr_fiemap(struct inode *inode,
5063 struct fiemap_extent_info *fieinfo)
5067 __u32 flags = FIEMAP_EXTENT_LAST;
5068 int blockbits = inode->i_sb->s_blocksize_bits;
5072 if (ext4_test_inode_state(inode, EXT4_STATE_XATTR)) {
5073 struct ext4_iloc iloc;
5074 int offset; /* offset of xattr in inode */
5076 error = ext4_get_inode_loc(inode, &iloc);
5079 physical = (__u64)iloc.bh->b_blocknr << blockbits;
5080 offset = EXT4_GOOD_OLD_INODE_SIZE +
5081 EXT4_I(inode)->i_extra_isize;
5083 length = EXT4_SB(inode->i_sb)->s_inode_size - offset;
5084 flags |= FIEMAP_EXTENT_DATA_INLINE;
5086 } else { /* external block */
5087 physical = (__u64)EXT4_I(inode)->i_file_acl << blockbits;
5088 length = inode->i_sb->s_blocksize;
5092 error = fiemap_fill_next_extent(fieinfo, 0, physical,
5094 return (error < 0 ? error : 0);
5097 static int _ext4_fiemap(struct inode *inode,
5098 struct fiemap_extent_info *fieinfo,
5099 __u64 start, __u64 len,
5100 int (*fill)(struct inode *, ext4_lblk_t,
5102 struct fiemap_extent_info *))
5104 ext4_lblk_t start_blk;
5105 u32 ext4_fiemap_flags = FIEMAP_FLAG_SYNC|FIEMAP_FLAG_XATTR;
5109 if (ext4_has_inline_data(inode)) {
5112 error = ext4_inline_data_fiemap(inode, fieinfo, &has_inline,
5119 if (fieinfo->fi_flags & FIEMAP_FLAG_CACHE) {
5120 error = ext4_ext_precache(inode);
5123 fieinfo->fi_flags &= ~FIEMAP_FLAG_CACHE;
5126 /* fallback to generic here if not in extents fmt */
5127 if (!(ext4_test_inode_flag(inode, EXT4_INODE_EXTENTS)) &&
5128 fill == ext4_fill_fiemap_extents)
5129 return generic_block_fiemap(inode, fieinfo, start, len,
5132 if (fill == ext4_fill_es_cache_info)
5133 ext4_fiemap_flags &= FIEMAP_FLAG_XATTR;
5134 if (fiemap_check_flags(fieinfo, ext4_fiemap_flags))
5137 if (fieinfo->fi_flags & FIEMAP_FLAG_XATTR) {
5138 error = ext4_xattr_fiemap(inode, fieinfo);
5140 ext4_lblk_t len_blks;
5143 start_blk = start >> inode->i_sb->s_blocksize_bits;
5144 last_blk = (start + len - 1) >> inode->i_sb->s_blocksize_bits;
5145 if (last_blk >= EXT_MAX_BLOCKS)
5146 last_blk = EXT_MAX_BLOCKS-1;
5147 len_blks = ((ext4_lblk_t) last_blk) - start_blk + 1;
5150 * Walk the extent tree gathering extent information
5151 * and pushing extents back to the user.
5153 error = fill(inode, start_blk, len_blks, fieinfo);
5158 int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
5159 __u64 start, __u64 len)
5161 return _ext4_fiemap(inode, fieinfo, start, len,
5162 ext4_fill_fiemap_extents);
5165 int ext4_get_es_cache(struct inode *inode, struct fiemap_extent_info *fieinfo,
5166 __u64 start, __u64 len)
5168 if (ext4_has_inline_data(inode)) {
5171 down_read(&EXT4_I(inode)->xattr_sem);
5172 has_inline = ext4_has_inline_data(inode);
5173 up_read(&EXT4_I(inode)->xattr_sem);
5178 return _ext4_fiemap(inode, fieinfo, start, len,
5179 ext4_fill_es_cache_info);
5185 * Function to access the path buffer for marking it dirty.
5186 * It also checks if there are sufficient credits left in the journal handle
5190 ext4_access_path(handle_t *handle, struct inode *inode,
5191 struct ext4_ext_path *path)
5195 if (!ext4_handle_valid(handle))
5199 * Check if need to extend journal credits
5200 * 3 for leaf, sb, and inode plus 2 (bmap and group
5201 * descriptor) for each block group; assume two block
5204 credits = ext4_writepage_trans_blocks(inode);
5205 err = ext4_datasem_ensure_credits(handle, inode, 7, credits, 0);
5209 err = ext4_ext_get_access(handle, inode, path);
5214 * ext4_ext_shift_path_extents:
5215 * Shift the extents of a path structure lying between path[depth].p_ext
5216 * and EXT_LAST_EXTENT(path[depth].p_hdr), by @shift blocks. @SHIFT tells
5217 * if it is right shift or left shift operation.
5220 ext4_ext_shift_path_extents(struct ext4_ext_path *path, ext4_lblk_t shift,
5221 struct inode *inode, handle_t *handle,
5222 enum SHIFT_DIRECTION SHIFT)
5225 struct ext4_extent *ex_start, *ex_last;
5226 bool update = false;
5227 depth = path->p_depth;
5229 while (depth >= 0) {
5230 if (depth == path->p_depth) {
5231 ex_start = path[depth].p_ext;
5233 return -EFSCORRUPTED;
5235 ex_last = EXT_LAST_EXTENT(path[depth].p_hdr);
5237 err = ext4_access_path(handle, inode, path + depth);
5241 if (ex_start == EXT_FIRST_EXTENT(path[depth].p_hdr))
5244 while (ex_start <= ex_last) {
5245 if (SHIFT == SHIFT_LEFT) {
5246 le32_add_cpu(&ex_start->ee_block,
5248 /* Try to merge to the left. */
5250 EXT_FIRST_EXTENT(path[depth].p_hdr))
5252 ext4_ext_try_to_merge_right(inode,
5253 path, ex_start - 1))
5258 le32_add_cpu(&ex_last->ee_block, shift);
5259 ext4_ext_try_to_merge_right(inode, path,
5264 err = ext4_ext_dirty(handle, inode, path + depth);
5268 if (--depth < 0 || !update)
5272 /* Update index too */
5273 err = ext4_access_path(handle, inode, path + depth);
5277 if (SHIFT == SHIFT_LEFT)
5278 le32_add_cpu(&path[depth].p_idx->ei_block, -shift);
5280 le32_add_cpu(&path[depth].p_idx->ei_block, shift);
5281 err = ext4_ext_dirty(handle, inode, path + depth);
5285 /* we are done if current index is not a starting index */
5286 if (path[depth].p_idx != EXT_FIRST_INDEX(path[depth].p_hdr))
5297 * ext4_ext_shift_extents:
5298 * All the extents which lies in the range from @start to the last allocated
5299 * block for the @inode are shifted either towards left or right (depending
5300 * upon @SHIFT) by @shift blocks.
5301 * On success, 0 is returned, error otherwise.
5304 ext4_ext_shift_extents(struct inode *inode, handle_t *handle,
5305 ext4_lblk_t start, ext4_lblk_t shift,
5306 enum SHIFT_DIRECTION SHIFT)
5308 struct ext4_ext_path *path;
5310 struct ext4_extent *extent;
5311 ext4_lblk_t stop, *iterator, ex_start, ex_end;
5313 /* Let path point to the last extent */
5314 path = ext4_find_extent(inode, EXT_MAX_BLOCKS - 1, NULL,
5317 return PTR_ERR(path);
5319 depth = path->p_depth;
5320 extent = path[depth].p_ext;
5324 stop = le32_to_cpu(extent->ee_block);
5327 * For left shifts, make sure the hole on the left is big enough to
5328 * accommodate the shift. For right shifts, make sure the last extent
5329 * won't be shifted beyond EXT_MAX_BLOCKS.
5331 if (SHIFT == SHIFT_LEFT) {
5332 path = ext4_find_extent(inode, start - 1, &path,
5335 return PTR_ERR(path);
5336 depth = path->p_depth;
5337 extent = path[depth].p_ext;
5339 ex_start = le32_to_cpu(extent->ee_block);
5340 ex_end = le32_to_cpu(extent->ee_block) +
5341 ext4_ext_get_actual_len(extent);
5347 if ((start == ex_start && shift > ex_start) ||
5348 (shift > start - ex_end)) {
5353 if (shift > EXT_MAX_BLOCKS -
5354 (stop + ext4_ext_get_actual_len(extent))) {
5361 * In case of left shift, iterator points to start and it is increased
5362 * till we reach stop. In case of right shift, iterator points to stop
5363 * and it is decreased till we reach start.
5365 if (SHIFT == SHIFT_LEFT)
5371 * Its safe to start updating extents. Start and stop are unsigned, so
5372 * in case of right shift if extent with 0 block is reached, iterator
5373 * becomes NULL to indicate the end of the loop.
5375 while (iterator && start <= stop) {
5376 path = ext4_find_extent(inode, *iterator, &path,
5379 return PTR_ERR(path);
5380 depth = path->p_depth;
5381 extent = path[depth].p_ext;
5383 EXT4_ERROR_INODE(inode, "unexpected hole at %lu",
5384 (unsigned long) *iterator);
5385 return -EFSCORRUPTED;
5387 if (SHIFT == SHIFT_LEFT && *iterator >
5388 le32_to_cpu(extent->ee_block)) {
5389 /* Hole, move to the next extent */
5390 if (extent < EXT_LAST_EXTENT(path[depth].p_hdr)) {
5391 path[depth].p_ext++;
5393 *iterator = ext4_ext_next_allocated_block(path);
5398 if (SHIFT == SHIFT_LEFT) {
5399 extent = EXT_LAST_EXTENT(path[depth].p_hdr);
5400 *iterator = le32_to_cpu(extent->ee_block) +
5401 ext4_ext_get_actual_len(extent);
5403 extent = EXT_FIRST_EXTENT(path[depth].p_hdr);
5404 if (le32_to_cpu(extent->ee_block) > 0)
5405 *iterator = le32_to_cpu(extent->ee_block) - 1;
5407 /* Beginning is reached, end of the loop */
5409 /* Update path extent in case we need to stop */
5410 while (le32_to_cpu(extent->ee_block) < start)
5412 path[depth].p_ext = extent;
5414 ret = ext4_ext_shift_path_extents(path, shift, inode,
5420 ext4_ext_drop_refs(path);
5426 * ext4_collapse_range:
5427 * This implements the fallocate's collapse range functionality for ext4
5428 * Returns: 0 and non-zero on error.
5430 static int ext4_collapse_range(struct inode *inode, loff_t offset, loff_t len)
5432 struct super_block *sb = inode->i_sb;
5433 ext4_lblk_t punch_start, punch_stop;
5435 unsigned int credits;
5436 loff_t new_size, ioffset;
5440 * We need to test this early because xfstests assumes that a
5441 * collapse range of (0, 1) will return EOPNOTSUPP if the file
5442 * system does not support collapse range.
5444 if (!ext4_test_inode_flag(inode, EXT4_INODE_EXTENTS))
5447 /* Collapse range works only on fs cluster size aligned regions. */
5448 if (!IS_ALIGNED(offset | len, EXT4_CLUSTER_SIZE(sb)))
5451 trace_ext4_collapse_range(inode, offset, len);
5453 punch_start = offset >> EXT4_BLOCK_SIZE_BITS(sb);
5454 punch_stop = (offset + len) >> EXT4_BLOCK_SIZE_BITS(sb);
5456 /* Call ext4_force_commit to flush all data in case of data=journal. */
5457 if (ext4_should_journal_data(inode)) {
5458 ret = ext4_force_commit(inode->i_sb);
5465 * There is no need to overlap collapse range with EOF, in which case
5466 * it is effectively a truncate operation
5468 if (offset + len >= inode->i_size) {
5473 /* Currently just for extent based files */
5474 if (!ext4_test_inode_flag(inode, EXT4_INODE_EXTENTS)) {
5479 /* Wait for existing dio to complete */
5480 inode_dio_wait(inode);
5483 * Prevent page faults from reinstantiating pages we have released from
5486 down_write(&EXT4_I(inode)->i_mmap_sem);
5488 ret = ext4_break_layouts(inode);
5493 * Need to round down offset to be aligned with page size boundary
5494 * for page size > block size.
5496 ioffset = round_down(offset, PAGE_SIZE);
5498 * Write tail of the last page before removed range since it will get
5499 * removed from the page cache below.
5501 ret = filemap_write_and_wait_range(inode->i_mapping, ioffset, offset);
5505 * Write data that will be shifted to preserve them when discarding
5506 * page cache below. We are also protected from pages becoming dirty
5509 ret = filemap_write_and_wait_range(inode->i_mapping, offset + len,
5513 truncate_pagecache(inode, ioffset);
5515 credits = ext4_writepage_trans_blocks(inode);
5516 handle = ext4_journal_start(inode, EXT4_HT_TRUNCATE, credits);
5517 if (IS_ERR(handle)) {
5518 ret = PTR_ERR(handle);
5522 down_write(&EXT4_I(inode)->i_data_sem);
5523 ext4_discard_preallocations(inode);
5525 ret = ext4_es_remove_extent(inode, punch_start,
5526 EXT_MAX_BLOCKS - punch_start);
5528 up_write(&EXT4_I(inode)->i_data_sem);
5532 ret = ext4_ext_remove_space(inode, punch_start, punch_stop - 1);
5534 up_write(&EXT4_I(inode)->i_data_sem);
5537 ext4_discard_preallocations(inode);
5539 ret = ext4_ext_shift_extents(inode, handle, punch_stop,
5540 punch_stop - punch_start, SHIFT_LEFT);
5542 up_write(&EXT4_I(inode)->i_data_sem);
5546 new_size = inode->i_size - len;
5547 i_size_write(inode, new_size);
5548 EXT4_I(inode)->i_disksize = new_size;
5550 up_write(&EXT4_I(inode)->i_data_sem);
5552 ext4_handle_sync(handle);
5553 inode->i_mtime = inode->i_ctime = current_time(inode);
5554 ext4_mark_inode_dirty(handle, inode);
5555 ext4_update_inode_fsync_trans(handle, inode, 1);
5558 ext4_journal_stop(handle);
5560 up_write(&EXT4_I(inode)->i_mmap_sem);
5562 inode_unlock(inode);
5567 * ext4_insert_range:
5568 * This function implements the FALLOC_FL_INSERT_RANGE flag of fallocate.
5569 * The data blocks starting from @offset to the EOF are shifted by @len
5570 * towards right to create a hole in the @inode. Inode size is increased
5572 * Returns 0 on success, error otherwise.
5574 static int ext4_insert_range(struct inode *inode, loff_t offset, loff_t len)
5576 struct super_block *sb = inode->i_sb;
5578 struct ext4_ext_path *path;
5579 struct ext4_extent *extent;
5580 ext4_lblk_t offset_lblk, len_lblk, ee_start_lblk = 0;
5581 unsigned int credits, ee_len;
5582 int ret = 0, depth, split_flag = 0;
5586 * We need to test this early because xfstests assumes that an
5587 * insert range of (0, 1) will return EOPNOTSUPP if the file
5588 * system does not support insert range.
5590 if (!ext4_test_inode_flag(inode, EXT4_INODE_EXTENTS))
5593 /* Insert range works only on fs cluster size aligned regions. */
5594 if (!IS_ALIGNED(offset | len, EXT4_CLUSTER_SIZE(sb)))
5597 trace_ext4_insert_range(inode, offset, len);
5599 offset_lblk = offset >> EXT4_BLOCK_SIZE_BITS(sb);
5600 len_lblk = len >> EXT4_BLOCK_SIZE_BITS(sb);
5602 /* Call ext4_force_commit to flush all data in case of data=journal */
5603 if (ext4_should_journal_data(inode)) {
5604 ret = ext4_force_commit(inode->i_sb);
5610 /* Currently just for extent based files */
5611 if (!ext4_test_inode_flag(inode, EXT4_INODE_EXTENTS)) {
5616 /* Check whether the maximum file size would be exceeded */
5617 if (len > inode->i_sb->s_maxbytes - inode->i_size) {
5622 /* Offset must be less than i_size */
5623 if (offset >= inode->i_size) {
5628 /* Wait for existing dio to complete */
5629 inode_dio_wait(inode);
5632 * Prevent page faults from reinstantiating pages we have released from
5635 down_write(&EXT4_I(inode)->i_mmap_sem);
5637 ret = ext4_break_layouts(inode);
5642 * Need to round down to align start offset to page size boundary
5643 * for page size > block size.
5645 ioffset = round_down(offset, PAGE_SIZE);
5646 /* Write out all dirty pages */
5647 ret = filemap_write_and_wait_range(inode->i_mapping, ioffset,
5651 truncate_pagecache(inode, ioffset);
5653 credits = ext4_writepage_trans_blocks(inode);
5654 handle = ext4_journal_start(inode, EXT4_HT_TRUNCATE, credits);
5655 if (IS_ERR(handle)) {
5656 ret = PTR_ERR(handle);
5660 /* Expand file to avoid data loss if there is error while shifting */
5661 inode->i_size += len;
5662 EXT4_I(inode)->i_disksize += len;
5663 inode->i_mtime = inode->i_ctime = current_time(inode);
5664 ret = ext4_mark_inode_dirty(handle, inode);
5668 down_write(&EXT4_I(inode)->i_data_sem);
5669 ext4_discard_preallocations(inode);
5671 path = ext4_find_extent(inode, offset_lblk, NULL, 0);
5673 up_write(&EXT4_I(inode)->i_data_sem);
5677 depth = ext_depth(inode);
5678 extent = path[depth].p_ext;
5680 ee_start_lblk = le32_to_cpu(extent->ee_block);
5681 ee_len = ext4_ext_get_actual_len(extent);
5684 * If offset_lblk is not the starting block of extent, split
5685 * the extent @offset_lblk
5687 if ((offset_lblk > ee_start_lblk) &&
5688 (offset_lblk < (ee_start_lblk + ee_len))) {
5689 if (ext4_ext_is_unwritten(extent))
5690 split_flag = EXT4_EXT_MARK_UNWRIT1 |
5691 EXT4_EXT_MARK_UNWRIT2;
5692 ret = ext4_split_extent_at(handle, inode, &path,
5693 offset_lblk, split_flag,
5695 EXT4_GET_BLOCKS_PRE_IO |
5696 EXT4_GET_BLOCKS_METADATA_NOFAIL);
5699 ext4_ext_drop_refs(path);
5702 up_write(&EXT4_I(inode)->i_data_sem);
5706 ext4_ext_drop_refs(path);
5710 ret = ext4_es_remove_extent(inode, offset_lblk,
5711 EXT_MAX_BLOCKS - offset_lblk);
5713 up_write(&EXT4_I(inode)->i_data_sem);
5718 * if offset_lblk lies in a hole which is at start of file, use
5719 * ee_start_lblk to shift extents
5721 ret = ext4_ext_shift_extents(inode, handle,
5722 ee_start_lblk > offset_lblk ? ee_start_lblk : offset_lblk,
5723 len_lblk, SHIFT_RIGHT);
5725 up_write(&EXT4_I(inode)->i_data_sem);
5727 ext4_handle_sync(handle);
5729 ext4_update_inode_fsync_trans(handle, inode, 1);
5732 ext4_journal_stop(handle);
5734 up_write(&EXT4_I(inode)->i_mmap_sem);
5736 inode_unlock(inode);
5741 * ext4_swap_extents() - Swap extents between two inodes
5742 * @handle: handle for this transaction
5743 * @inode1: First inode
5744 * @inode2: Second inode
5745 * @lblk1: Start block for first inode
5746 * @lblk2: Start block for second inode
5747 * @count: Number of blocks to swap
5748 * @unwritten: Mark second inode's extents as unwritten after swap
5749 * @erp: Pointer to save error value
5751 * This helper routine does exactly what is promise "swap extents". All other
5752 * stuff such as page-cache locking consistency, bh mapping consistency or
5753 * extent's data copying must be performed by caller.
5755 * i_mutex is held for both inodes
5756 * i_data_sem is locked for write for both inodes
5758 * All pages from requested range are locked for both inodes
5761 ext4_swap_extents(handle_t *handle, struct inode *inode1,
5762 struct inode *inode2, ext4_lblk_t lblk1, ext4_lblk_t lblk2,
5763 ext4_lblk_t count, int unwritten, int *erp)
5765 struct ext4_ext_path *path1 = NULL;
5766 struct ext4_ext_path *path2 = NULL;
5767 int replaced_count = 0;
5769 BUG_ON(!rwsem_is_locked(&EXT4_I(inode1)->i_data_sem));
5770 BUG_ON(!rwsem_is_locked(&EXT4_I(inode2)->i_data_sem));
5771 BUG_ON(!inode_is_locked(inode1));
5772 BUG_ON(!inode_is_locked(inode2));
5774 *erp = ext4_es_remove_extent(inode1, lblk1, count);
5777 *erp = ext4_es_remove_extent(inode2, lblk2, count);
5782 struct ext4_extent *ex1, *ex2, tmp_ex;
5783 ext4_lblk_t e1_blk, e2_blk;
5784 int e1_len, e2_len, len;
5787 path1 = ext4_find_extent(inode1, lblk1, NULL, EXT4_EX_NOCACHE);
5788 if (IS_ERR(path1)) {
5789 *erp = PTR_ERR(path1);
5795 path2 = ext4_find_extent(inode2, lblk2, NULL, EXT4_EX_NOCACHE);
5796 if (IS_ERR(path2)) {
5797 *erp = PTR_ERR(path2);
5801 ex1 = path1[path1->p_depth].p_ext;
5802 ex2 = path2[path2->p_depth].p_ext;
5803 /* Do we have somthing to swap ? */
5804 if (unlikely(!ex2 || !ex1))
5807 e1_blk = le32_to_cpu(ex1->ee_block);
5808 e2_blk = le32_to_cpu(ex2->ee_block);
5809 e1_len = ext4_ext_get_actual_len(ex1);
5810 e2_len = ext4_ext_get_actual_len(ex2);
5813 if (!in_range(lblk1, e1_blk, e1_len) ||
5814 !in_range(lblk2, e2_blk, e2_len)) {
5815 ext4_lblk_t next1, next2;
5817 /* if hole after extent, then go to next extent */
5818 next1 = ext4_ext_next_allocated_block(path1);
5819 next2 = ext4_ext_next_allocated_block(path2);
5820 /* If hole before extent, then shift to that extent */
5825 /* Do we have something to swap */
5826 if (next1 == EXT_MAX_BLOCKS || next2 == EXT_MAX_BLOCKS)
5828 /* Move to the rightest boundary */
5829 len = next1 - lblk1;
5830 if (len < next2 - lblk2)
5831 len = next2 - lblk2;
5840 /* Prepare left boundary */
5841 if (e1_blk < lblk1) {
5843 *erp = ext4_force_split_extent_at(handle, inode1,
5848 if (e2_blk < lblk2) {
5850 *erp = ext4_force_split_extent_at(handle, inode2,
5855 /* ext4_split_extent_at() may result in leaf extent split,
5856 * path must to be revalidated. */
5860 /* Prepare right boundary */
5862 if (len > e1_blk + e1_len - lblk1)
5863 len = e1_blk + e1_len - lblk1;
5864 if (len > e2_blk + e2_len - lblk2)
5865 len = e2_blk + e2_len - lblk2;
5867 if (len != e1_len) {
5869 *erp = ext4_force_split_extent_at(handle, inode1,
5870 &path1, lblk1 + len, 0);
5874 if (len != e2_len) {
5876 *erp = ext4_force_split_extent_at(handle, inode2,
5877 &path2, lblk2 + len, 0);
5881 /* ext4_split_extent_at() may result in leaf extent split,
5882 * path must to be revalidated. */
5886 BUG_ON(e2_len != e1_len);
5887 *erp = ext4_ext_get_access(handle, inode1, path1 + path1->p_depth);
5890 *erp = ext4_ext_get_access(handle, inode2, path2 + path2->p_depth);
5894 /* Both extents are fully inside boundaries. Swap it now */
5896 ext4_ext_store_pblock(ex1, ext4_ext_pblock(ex2));
5897 ext4_ext_store_pblock(ex2, ext4_ext_pblock(&tmp_ex));
5898 ex1->ee_len = cpu_to_le16(e2_len);
5899 ex2->ee_len = cpu_to_le16(e1_len);
5901 ext4_ext_mark_unwritten(ex2);
5902 if (ext4_ext_is_unwritten(&tmp_ex))
5903 ext4_ext_mark_unwritten(ex1);
5905 ext4_ext_try_to_merge(handle, inode2, path2, ex2);
5906 ext4_ext_try_to_merge(handle, inode1, path1, ex1);
5907 *erp = ext4_ext_dirty(handle, inode2, path2 +
5911 *erp = ext4_ext_dirty(handle, inode1, path1 +
5914 * Looks scarry ah..? second inode already points to new blocks,
5915 * and it was successfully dirtied. But luckily error may happen
5916 * only due to journal error, so full transaction will be
5923 replaced_count += len;
5927 ext4_ext_drop_refs(path1);
5929 ext4_ext_drop_refs(path2);
5931 path1 = path2 = NULL;
5933 return replaced_count;
5937 * ext4_clu_mapped - determine whether any block in a logical cluster has
5938 * been mapped to a physical cluster
5940 * @inode - file containing the logical cluster
5941 * @lclu - logical cluster of interest
5943 * Returns 1 if any block in the logical cluster is mapped, signifying
5944 * that a physical cluster has been allocated for it. Otherwise,
5945 * returns 0. Can also return negative error codes. Derived from
5946 * ext4_ext_map_blocks().
5948 int ext4_clu_mapped(struct inode *inode, ext4_lblk_t lclu)
5950 struct ext4_sb_info *sbi = EXT4_SB(inode->i_sb);
5951 struct ext4_ext_path *path;
5952 int depth, mapped = 0, err = 0;
5953 struct ext4_extent *extent;
5954 ext4_lblk_t first_lblk, first_lclu, last_lclu;
5956 /* search for the extent closest to the first block in the cluster */
5957 path = ext4_find_extent(inode, EXT4_C2B(sbi, lclu), NULL, 0);
5959 err = PTR_ERR(path);
5964 depth = ext_depth(inode);
5967 * A consistent leaf must not be empty. This situation is possible,
5968 * though, _during_ tree modification, and it's why an assert can't
5969 * be put in ext4_find_extent().
5971 if (unlikely(path[depth].p_ext == NULL && depth != 0)) {
5972 EXT4_ERROR_INODE(inode,
5973 "bad extent address - lblock: %lu, depth: %d, pblock: %lld",
5974 (unsigned long) EXT4_C2B(sbi, lclu),
5975 depth, path[depth].p_block);
5976 err = -EFSCORRUPTED;
5980 extent = path[depth].p_ext;
5982 /* can't be mapped if the extent tree is empty */
5986 first_lblk = le32_to_cpu(extent->ee_block);
5987 first_lclu = EXT4_B2C(sbi, first_lblk);
5990 * Three possible outcomes at this point - found extent spanning
5991 * the target cluster, to the left of the target cluster, or to the
5992 * right of the target cluster. The first two cases are handled here.
5993 * The last case indicates the target cluster is not mapped.
5995 if (lclu >= first_lclu) {
5996 last_lclu = EXT4_B2C(sbi, first_lblk +
5997 ext4_ext_get_actual_len(extent) - 1);
5998 if (lclu <= last_lclu) {
6001 first_lblk = ext4_ext_next_allocated_block(path);
6002 first_lclu = EXT4_B2C(sbi, first_lblk);
6003 if (lclu == first_lclu)
6009 ext4_ext_drop_refs(path);
6012 return err ? err : mapped;