Merge branch 'for-linus' of git://git.kernel.dk/linux-block
[platform/adaptation/renesas_rcar/renesas_kernel.git] / fs / ext4 / extents.c
1 /*
2  * Copyright (c) 2003-2006, Cluster File Systems, Inc, info@clusterfs.com
3  * Written by Alex Tomas <alex@clusterfs.com>
4  *
5  * Architecture independence:
6  *   Copyright (c) 2005, Bull S.A.
7  *   Written by Pierre Peiffer <pierre.peiffer@bull.net>
8  *
9  * This program is free software; you can redistribute it and/or modify
10  * it under the terms of the GNU General Public License version 2 as
11  * published by the Free Software Foundation.
12  *
13  * This program is distributed in the hope that it will be useful,
14  * but WITHOUT ANY WARRANTY; without even the implied warranty of
15  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
16  * GNU General Public License for more details.
17  *
18  * You should have received a copy of the GNU General Public Licens
19  * along with this program; if not, write to the Free Software
20  * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-
21  */
22
23 /*
24  * Extents support for EXT4
25  *
26  * TODO:
27  *   - ext4*_error() should be used in some situations
28  *   - analyze all BUG()/BUG_ON(), use -EIO where appropriate
29  *   - smart tree reduction
30  */
31
32 #include <linux/fs.h>
33 #include <linux/time.h>
34 #include <linux/jbd2.h>
35 #include <linux/highuid.h>
36 #include <linux/pagemap.h>
37 #include <linux/quotaops.h>
38 #include <linux/string.h>
39 #include <linux/slab.h>
40 #include <linux/falloc.h>
41 #include <asm/uaccess.h>
42 #include <linux/fiemap.h>
43 #include "ext4_jbd2.h"
44 #include "ext4_extents.h"
45 #include "xattr.h"
46
47 #include <trace/events/ext4.h>
48
49 /*
50  * used by extent splitting.
51  */
52 #define EXT4_EXT_MAY_ZEROOUT    0x1  /* safe to zeroout if split fails \
53                                         due to ENOSPC */
54 #define EXT4_EXT_MARK_UNINIT1   0x2  /* mark first half uninitialized */
55 #define EXT4_EXT_MARK_UNINIT2   0x4  /* mark second half uninitialized */
56
57 #define EXT4_EXT_DATA_VALID1    0x8  /* first half contains valid data */
58 #define EXT4_EXT_DATA_VALID2    0x10 /* second half contains valid data */
59
60 static __le32 ext4_extent_block_csum(struct inode *inode,
61                                      struct ext4_extent_header *eh)
62 {
63         struct ext4_inode_info *ei = EXT4_I(inode);
64         struct ext4_sb_info *sbi = EXT4_SB(inode->i_sb);
65         __u32 csum;
66
67         csum = ext4_chksum(sbi, ei->i_csum_seed, (__u8 *)eh,
68                            EXT4_EXTENT_TAIL_OFFSET(eh));
69         return cpu_to_le32(csum);
70 }
71
72 static int ext4_extent_block_csum_verify(struct inode *inode,
73                                          struct ext4_extent_header *eh)
74 {
75         struct ext4_extent_tail *et;
76
77         if (!EXT4_HAS_RO_COMPAT_FEATURE(inode->i_sb,
78                 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM))
79                 return 1;
80
81         et = find_ext4_extent_tail(eh);
82         if (et->et_checksum != ext4_extent_block_csum(inode, eh))
83                 return 0;
84         return 1;
85 }
86
87 static void ext4_extent_block_csum_set(struct inode *inode,
88                                        struct ext4_extent_header *eh)
89 {
90         struct ext4_extent_tail *et;
91
92         if (!EXT4_HAS_RO_COMPAT_FEATURE(inode->i_sb,
93                 EXT4_FEATURE_RO_COMPAT_METADATA_CSUM))
94                 return;
95
96         et = find_ext4_extent_tail(eh);
97         et->et_checksum = ext4_extent_block_csum(inode, eh);
98 }
99
100 static int ext4_split_extent(handle_t *handle,
101                                 struct inode *inode,
102                                 struct ext4_ext_path *path,
103                                 struct ext4_map_blocks *map,
104                                 int split_flag,
105                                 int flags);
106
107 static int ext4_split_extent_at(handle_t *handle,
108                              struct inode *inode,
109                              struct ext4_ext_path *path,
110                              ext4_lblk_t split,
111                              int split_flag,
112                              int flags);
113
114 static int ext4_find_delayed_extent(struct inode *inode,
115                                     struct extent_status *newes);
116
117 static int ext4_ext_truncate_extend_restart(handle_t *handle,
118                                             struct inode *inode,
119                                             int needed)
120 {
121         int err;
122
123         if (!ext4_handle_valid(handle))
124                 return 0;
125         if (handle->h_buffer_credits > needed)
126                 return 0;
127         err = ext4_journal_extend(handle, needed);
128         if (err <= 0)
129                 return err;
130         err = ext4_truncate_restart_trans(handle, inode, needed);
131         if (err == 0)
132                 err = -EAGAIN;
133
134         return err;
135 }
136
137 /*
138  * could return:
139  *  - EROFS
140  *  - ENOMEM
141  */
142 static int ext4_ext_get_access(handle_t *handle, struct inode *inode,
143                                 struct ext4_ext_path *path)
144 {
145         if (path->p_bh) {
146                 /* path points to block */
147                 return ext4_journal_get_write_access(handle, path->p_bh);
148         }
149         /* path points to leaf/index in inode body */
150         /* we use in-core data, no need to protect them */
151         return 0;
152 }
153
154 /*
155  * could return:
156  *  - EROFS
157  *  - ENOMEM
158  *  - EIO
159  */
160 int __ext4_ext_dirty(const char *where, unsigned int line, handle_t *handle,
161                      struct inode *inode, struct ext4_ext_path *path)
162 {
163         int err;
164         if (path->p_bh) {
165                 ext4_extent_block_csum_set(inode, ext_block_hdr(path->p_bh));
166                 /* path points to block */
167                 err = __ext4_handle_dirty_metadata(where, line, handle,
168                                                    inode, path->p_bh);
169         } else {
170                 /* path points to leaf/index in inode body */
171                 err = ext4_mark_inode_dirty(handle, inode);
172         }
173         return err;
174 }
175
176 static ext4_fsblk_t ext4_ext_find_goal(struct inode *inode,
177                               struct ext4_ext_path *path,
178                               ext4_lblk_t block)
179 {
180         if (path) {
181                 int depth = path->p_depth;
182                 struct ext4_extent *ex;
183
184                 /*
185                  * Try to predict block placement assuming that we are
186                  * filling in a file which will eventually be
187                  * non-sparse --- i.e., in the case of libbfd writing
188                  * an ELF object sections out-of-order but in a way
189                  * the eventually results in a contiguous object or
190                  * executable file, or some database extending a table
191                  * space file.  However, this is actually somewhat
192                  * non-ideal if we are writing a sparse file such as
193                  * qemu or KVM writing a raw image file that is going
194                  * to stay fairly sparse, since it will end up
195                  * fragmenting the file system's free space.  Maybe we
196                  * should have some hueristics or some way to allow
197                  * userspace to pass a hint to file system,
198                  * especially if the latter case turns out to be
199                  * common.
200                  */
201                 ex = path[depth].p_ext;
202                 if (ex) {
203                         ext4_fsblk_t ext_pblk = ext4_ext_pblock(ex);
204                         ext4_lblk_t ext_block = le32_to_cpu(ex->ee_block);
205
206                         if (block > ext_block)
207                                 return ext_pblk + (block - ext_block);
208                         else
209                                 return ext_pblk - (ext_block - block);
210                 }
211
212                 /* it looks like index is empty;
213                  * try to find starting block from index itself */
214                 if (path[depth].p_bh)
215                         return path[depth].p_bh->b_blocknr;
216         }
217
218         /* OK. use inode's group */
219         return ext4_inode_to_goal_block(inode);
220 }
221
222 /*
223  * Allocation for a meta data block
224  */
225 static ext4_fsblk_t
226 ext4_ext_new_meta_block(handle_t *handle, struct inode *inode,
227                         struct ext4_ext_path *path,
228                         struct ext4_extent *ex, int *err, unsigned int flags)
229 {
230         ext4_fsblk_t goal, newblock;
231
232         goal = ext4_ext_find_goal(inode, path, le32_to_cpu(ex->ee_block));
233         newblock = ext4_new_meta_blocks(handle, inode, goal, flags,
234                                         NULL, err);
235         return newblock;
236 }
237
238 static inline int ext4_ext_space_block(struct inode *inode, int check)
239 {
240         int size;
241
242         size = (inode->i_sb->s_blocksize - sizeof(struct ext4_extent_header))
243                         / sizeof(struct ext4_extent);
244 #ifdef AGGRESSIVE_TEST
245         if (!check && size > 6)
246                 size = 6;
247 #endif
248         return size;
249 }
250
251 static inline int ext4_ext_space_block_idx(struct inode *inode, int check)
252 {
253         int size;
254
255         size = (inode->i_sb->s_blocksize - sizeof(struct ext4_extent_header))
256                         / sizeof(struct ext4_extent_idx);
257 #ifdef AGGRESSIVE_TEST
258         if (!check && size > 5)
259                 size = 5;
260 #endif
261         return size;
262 }
263
264 static inline int ext4_ext_space_root(struct inode *inode, int check)
265 {
266         int size;
267
268         size = sizeof(EXT4_I(inode)->i_data);
269         size -= sizeof(struct ext4_extent_header);
270         size /= sizeof(struct ext4_extent);
271 #ifdef AGGRESSIVE_TEST
272         if (!check && size > 3)
273                 size = 3;
274 #endif
275         return size;
276 }
277
278 static inline int ext4_ext_space_root_idx(struct inode *inode, int check)
279 {
280         int size;
281
282         size = sizeof(EXT4_I(inode)->i_data);
283         size -= sizeof(struct ext4_extent_header);
284         size /= sizeof(struct ext4_extent_idx);
285 #ifdef AGGRESSIVE_TEST
286         if (!check && size > 4)
287                 size = 4;
288 #endif
289         return size;
290 }
291
292 /*
293  * Calculate the number of metadata blocks needed
294  * to allocate @blocks
295  * Worse case is one block per extent
296  */
297 int ext4_ext_calc_metadata_amount(struct inode *inode, ext4_lblk_t lblock)
298 {
299         struct ext4_inode_info *ei = EXT4_I(inode);
300         int idxs;
301
302         idxs = ((inode->i_sb->s_blocksize - sizeof(struct ext4_extent_header))
303                 / sizeof(struct ext4_extent_idx));
304
305         /*
306          * If the new delayed allocation block is contiguous with the
307          * previous da block, it can share index blocks with the
308          * previous block, so we only need to allocate a new index
309          * block every idxs leaf blocks.  At ldxs**2 blocks, we need
310          * an additional index block, and at ldxs**3 blocks, yet
311          * another index blocks.
312          */
313         if (ei->i_da_metadata_calc_len &&
314             ei->i_da_metadata_calc_last_lblock+1 == lblock) {
315                 int num = 0;
316
317                 if ((ei->i_da_metadata_calc_len % idxs) == 0)
318                         num++;
319                 if ((ei->i_da_metadata_calc_len % (idxs*idxs)) == 0)
320                         num++;
321                 if ((ei->i_da_metadata_calc_len % (idxs*idxs*idxs)) == 0) {
322                         num++;
323                         ei->i_da_metadata_calc_len = 0;
324                 } else
325                         ei->i_da_metadata_calc_len++;
326                 ei->i_da_metadata_calc_last_lblock++;
327                 return num;
328         }
329
330         /*
331          * In the worst case we need a new set of index blocks at
332          * every level of the inode's extent tree.
333          */
334         ei->i_da_metadata_calc_len = 1;
335         ei->i_da_metadata_calc_last_lblock = lblock;
336         return ext_depth(inode) + 1;
337 }
338
339 static int
340 ext4_ext_max_entries(struct inode *inode, int depth)
341 {
342         int max;
343
344         if (depth == ext_depth(inode)) {
345                 if (depth == 0)
346                         max = ext4_ext_space_root(inode, 1);
347                 else
348                         max = ext4_ext_space_root_idx(inode, 1);
349         } else {
350                 if (depth == 0)
351                         max = ext4_ext_space_block(inode, 1);
352                 else
353                         max = ext4_ext_space_block_idx(inode, 1);
354         }
355
356         return max;
357 }
358
359 static int ext4_valid_extent(struct inode *inode, struct ext4_extent *ext)
360 {
361         ext4_fsblk_t block = ext4_ext_pblock(ext);
362         int len = ext4_ext_get_actual_len(ext);
363
364         if (len == 0)
365                 return 0;
366         return ext4_data_block_valid(EXT4_SB(inode->i_sb), block, len);
367 }
368
369 static int ext4_valid_extent_idx(struct inode *inode,
370                                 struct ext4_extent_idx *ext_idx)
371 {
372         ext4_fsblk_t block = ext4_idx_pblock(ext_idx);
373
374         return ext4_data_block_valid(EXT4_SB(inode->i_sb), block, 1);
375 }
376
377 static int ext4_valid_extent_entries(struct inode *inode,
378                                 struct ext4_extent_header *eh,
379                                 int depth)
380 {
381         unsigned short entries;
382         if (eh->eh_entries == 0)
383                 return 1;
384
385         entries = le16_to_cpu(eh->eh_entries);
386
387         if (depth == 0) {
388                 /* leaf entries */
389                 struct ext4_extent *ext = EXT_FIRST_EXTENT(eh);
390                 while (entries) {
391                         if (!ext4_valid_extent(inode, ext))
392                                 return 0;
393                         ext++;
394                         entries--;
395                 }
396         } else {
397                 struct ext4_extent_idx *ext_idx = EXT_FIRST_INDEX(eh);
398                 while (entries) {
399                         if (!ext4_valid_extent_idx(inode, ext_idx))
400                                 return 0;
401                         ext_idx++;
402                         entries--;
403                 }
404         }
405         return 1;
406 }
407
408 static int __ext4_ext_check(const char *function, unsigned int line,
409                             struct inode *inode, struct ext4_extent_header *eh,
410                             int depth, ext4_fsblk_t pblk)
411 {
412         const char *error_msg;
413         int max = 0;
414
415         if (unlikely(eh->eh_magic != EXT4_EXT_MAGIC)) {
416                 error_msg = "invalid magic";
417                 goto corrupted;
418         }
419         if (unlikely(le16_to_cpu(eh->eh_depth) != depth)) {
420                 error_msg = "unexpected eh_depth";
421                 goto corrupted;
422         }
423         if (unlikely(eh->eh_max == 0)) {
424                 error_msg = "invalid eh_max";
425                 goto corrupted;
426         }
427         max = ext4_ext_max_entries(inode, depth);
428         if (unlikely(le16_to_cpu(eh->eh_max) > max)) {
429                 error_msg = "too large eh_max";
430                 goto corrupted;
431         }
432         if (unlikely(le16_to_cpu(eh->eh_entries) > le16_to_cpu(eh->eh_max))) {
433                 error_msg = "invalid eh_entries";
434                 goto corrupted;
435         }
436         if (!ext4_valid_extent_entries(inode, eh, depth)) {
437                 error_msg = "invalid extent entries";
438                 goto corrupted;
439         }
440         /* Verify checksum on non-root extent tree nodes */
441         if (ext_depth(inode) != depth &&
442             !ext4_extent_block_csum_verify(inode, eh)) {
443                 error_msg = "extent tree corrupted";
444                 goto corrupted;
445         }
446         return 0;
447
448 corrupted:
449         ext4_error_inode(inode, function, line, 0,
450                          "pblk %llu bad header/extent: %s - magic %x, "
451                          "entries %u, max %u(%u), depth %u(%u)",
452                          (unsigned long long) pblk, error_msg,
453                          le16_to_cpu(eh->eh_magic),
454                          le16_to_cpu(eh->eh_entries), le16_to_cpu(eh->eh_max),
455                          max, le16_to_cpu(eh->eh_depth), depth);
456         return -EIO;
457 }
458
459 #define ext4_ext_check(inode, eh, depth, pblk)                  \
460         __ext4_ext_check(__func__, __LINE__, (inode), (eh), (depth), (pblk))
461
462 int ext4_ext_check_inode(struct inode *inode)
463 {
464         return ext4_ext_check(inode, ext_inode_hdr(inode), ext_depth(inode), 0);
465 }
466
467 static struct buffer_head *
468 __read_extent_tree_block(const char *function, unsigned int line,
469                          struct inode *inode, ext4_fsblk_t pblk, int depth,
470                          int flags)
471 {
472         struct buffer_head              *bh;
473         int                             err;
474
475         bh = sb_getblk(inode->i_sb, pblk);
476         if (unlikely(!bh))
477                 return ERR_PTR(-ENOMEM);
478
479         if (!bh_uptodate_or_lock(bh)) {
480                 trace_ext4_ext_load_extent(inode, pblk, _RET_IP_);
481                 err = bh_submit_read(bh);
482                 if (err < 0)
483                         goto errout;
484         }
485         if (buffer_verified(bh) && !(flags & EXT4_EX_FORCE_CACHE))
486                 return bh;
487         err = __ext4_ext_check(function, line, inode,
488                                ext_block_hdr(bh), depth, pblk);
489         if (err)
490                 goto errout;
491         set_buffer_verified(bh);
492         /*
493          * If this is a leaf block, cache all of its entries
494          */
495         if (!(flags & EXT4_EX_NOCACHE) && depth == 0) {
496                 struct ext4_extent_header *eh = ext_block_hdr(bh);
497                 struct ext4_extent *ex = EXT_FIRST_EXTENT(eh);
498                 ext4_lblk_t prev = 0;
499                 int i;
500
501                 for (i = le16_to_cpu(eh->eh_entries); i > 0; i--, ex++) {
502                         unsigned int status = EXTENT_STATUS_WRITTEN;
503                         ext4_lblk_t lblk = le32_to_cpu(ex->ee_block);
504                         int len = ext4_ext_get_actual_len(ex);
505
506                         if (prev && (prev != lblk))
507                                 ext4_es_cache_extent(inode, prev,
508                                                      lblk - prev, ~0,
509                                                      EXTENT_STATUS_HOLE);
510
511                         if (ext4_ext_is_uninitialized(ex))
512                                 status = EXTENT_STATUS_UNWRITTEN;
513                         ext4_es_cache_extent(inode, lblk, len,
514                                              ext4_ext_pblock(ex), status);
515                         prev = lblk + len;
516                 }
517         }
518         return bh;
519 errout:
520         put_bh(bh);
521         return ERR_PTR(err);
522
523 }
524
525 #define read_extent_tree_block(inode, pblk, depth, flags)               \
526         __read_extent_tree_block(__func__, __LINE__, (inode), (pblk),   \
527                                  (depth), (flags))
528
529 /*
530  * This function is called to cache a file's extent information in the
531  * extent status tree
532  */
533 int ext4_ext_precache(struct inode *inode)
534 {
535         struct ext4_inode_info *ei = EXT4_I(inode);
536         struct ext4_ext_path *path = NULL;
537         struct buffer_head *bh;
538         int i = 0, depth, ret = 0;
539
540         if (!ext4_test_inode_flag(inode, EXT4_INODE_EXTENTS))
541                 return 0;       /* not an extent-mapped inode */
542
543         down_read(&ei->i_data_sem);
544         depth = ext_depth(inode);
545
546         path = kzalloc(sizeof(struct ext4_ext_path) * (depth + 1),
547                        GFP_NOFS);
548         if (path == NULL) {
549                 up_read(&ei->i_data_sem);
550                 return -ENOMEM;
551         }
552
553         /* Don't cache anything if there are no external extent blocks */
554         if (depth == 0)
555                 goto out;
556         path[0].p_hdr = ext_inode_hdr(inode);
557         ret = ext4_ext_check(inode, path[0].p_hdr, depth, 0);
558         if (ret)
559                 goto out;
560         path[0].p_idx = EXT_FIRST_INDEX(path[0].p_hdr);
561         while (i >= 0) {
562                 /*
563                  * If this is a leaf block or we've reached the end of
564                  * the index block, go up
565                  */
566                 if ((i == depth) ||
567                     path[i].p_idx > EXT_LAST_INDEX(path[i].p_hdr)) {
568                         brelse(path[i].p_bh);
569                         path[i].p_bh = NULL;
570                         i--;
571                         continue;
572                 }
573                 bh = read_extent_tree_block(inode,
574                                             ext4_idx_pblock(path[i].p_idx++),
575                                             depth - i - 1,
576                                             EXT4_EX_FORCE_CACHE);
577                 if (IS_ERR(bh)) {
578                         ret = PTR_ERR(bh);
579                         break;
580                 }
581                 i++;
582                 path[i].p_bh = bh;
583                 path[i].p_hdr = ext_block_hdr(bh);
584                 path[i].p_idx = EXT_FIRST_INDEX(path[i].p_hdr);
585         }
586         ext4_set_inode_state(inode, EXT4_STATE_EXT_PRECACHED);
587 out:
588         up_read(&ei->i_data_sem);
589         ext4_ext_drop_refs(path);
590         kfree(path);
591         return ret;
592 }
593
594 #ifdef EXT_DEBUG
595 static void ext4_ext_show_path(struct inode *inode, struct ext4_ext_path *path)
596 {
597         int k, l = path->p_depth;
598
599         ext_debug("path:");
600         for (k = 0; k <= l; k++, path++) {
601                 if (path->p_idx) {
602                   ext_debug("  %d->%llu", le32_to_cpu(path->p_idx->ei_block),
603                             ext4_idx_pblock(path->p_idx));
604                 } else if (path->p_ext) {
605                         ext_debug("  %d:[%d]%d:%llu ",
606                                   le32_to_cpu(path->p_ext->ee_block),
607                                   ext4_ext_is_uninitialized(path->p_ext),
608                                   ext4_ext_get_actual_len(path->p_ext),
609                                   ext4_ext_pblock(path->p_ext));
610                 } else
611                         ext_debug("  []");
612         }
613         ext_debug("\n");
614 }
615
616 static void ext4_ext_show_leaf(struct inode *inode, struct ext4_ext_path *path)
617 {
618         int depth = ext_depth(inode);
619         struct ext4_extent_header *eh;
620         struct ext4_extent *ex;
621         int i;
622
623         if (!path)
624                 return;
625
626         eh = path[depth].p_hdr;
627         ex = EXT_FIRST_EXTENT(eh);
628
629         ext_debug("Displaying leaf extents for inode %lu\n", inode->i_ino);
630
631         for (i = 0; i < le16_to_cpu(eh->eh_entries); i++, ex++) {
632                 ext_debug("%d:[%d]%d:%llu ", le32_to_cpu(ex->ee_block),
633                           ext4_ext_is_uninitialized(ex),
634                           ext4_ext_get_actual_len(ex), ext4_ext_pblock(ex));
635         }
636         ext_debug("\n");
637 }
638
639 static void ext4_ext_show_move(struct inode *inode, struct ext4_ext_path *path,
640                         ext4_fsblk_t newblock, int level)
641 {
642         int depth = ext_depth(inode);
643         struct ext4_extent *ex;
644
645         if (depth != level) {
646                 struct ext4_extent_idx *idx;
647                 idx = path[level].p_idx;
648                 while (idx <= EXT_MAX_INDEX(path[level].p_hdr)) {
649                         ext_debug("%d: move %d:%llu in new index %llu\n", level,
650                                         le32_to_cpu(idx->ei_block),
651                                         ext4_idx_pblock(idx),
652                                         newblock);
653                         idx++;
654                 }
655
656                 return;
657         }
658
659         ex = path[depth].p_ext;
660         while (ex <= EXT_MAX_EXTENT(path[depth].p_hdr)) {
661                 ext_debug("move %d:%llu:[%d]%d in new leaf %llu\n",
662                                 le32_to_cpu(ex->ee_block),
663                                 ext4_ext_pblock(ex),
664                                 ext4_ext_is_uninitialized(ex),
665                                 ext4_ext_get_actual_len(ex),
666                                 newblock);
667                 ex++;
668         }
669 }
670
671 #else
672 #define ext4_ext_show_path(inode, path)
673 #define ext4_ext_show_leaf(inode, path)
674 #define ext4_ext_show_move(inode, path, newblock, level)
675 #endif
676
677 void ext4_ext_drop_refs(struct ext4_ext_path *path)
678 {
679         int depth = path->p_depth;
680         int i;
681
682         for (i = 0; i <= depth; i++, path++)
683                 if (path->p_bh) {
684                         brelse(path->p_bh);
685                         path->p_bh = NULL;
686                 }
687 }
688
689 /*
690  * ext4_ext_binsearch_idx:
691  * binary search for the closest index of the given block
692  * the header must be checked before calling this
693  */
694 static void
695 ext4_ext_binsearch_idx(struct inode *inode,
696                         struct ext4_ext_path *path, ext4_lblk_t block)
697 {
698         struct ext4_extent_header *eh = path->p_hdr;
699         struct ext4_extent_idx *r, *l, *m;
700
701
702         ext_debug("binsearch for %u(idx):  ", block);
703
704         l = EXT_FIRST_INDEX(eh) + 1;
705         r = EXT_LAST_INDEX(eh);
706         while (l <= r) {
707                 m = l + (r - l) / 2;
708                 if (block < le32_to_cpu(m->ei_block))
709                         r = m - 1;
710                 else
711                         l = m + 1;
712                 ext_debug("%p(%u):%p(%u):%p(%u) ", l, le32_to_cpu(l->ei_block),
713                                 m, le32_to_cpu(m->ei_block),
714                                 r, le32_to_cpu(r->ei_block));
715         }
716
717         path->p_idx = l - 1;
718         ext_debug("  -> %u->%lld ", le32_to_cpu(path->p_idx->ei_block),
719                   ext4_idx_pblock(path->p_idx));
720
721 #ifdef CHECK_BINSEARCH
722         {
723                 struct ext4_extent_idx *chix, *ix;
724                 int k;
725
726                 chix = ix = EXT_FIRST_INDEX(eh);
727                 for (k = 0; k < le16_to_cpu(eh->eh_entries); k++, ix++) {
728                   if (k != 0 &&
729                       le32_to_cpu(ix->ei_block) <= le32_to_cpu(ix[-1].ei_block)) {
730                                 printk(KERN_DEBUG "k=%d, ix=0x%p, "
731                                        "first=0x%p\n", k,
732                                        ix, EXT_FIRST_INDEX(eh));
733                                 printk(KERN_DEBUG "%u <= %u\n",
734                                        le32_to_cpu(ix->ei_block),
735                                        le32_to_cpu(ix[-1].ei_block));
736                         }
737                         BUG_ON(k && le32_to_cpu(ix->ei_block)
738                                            <= le32_to_cpu(ix[-1].ei_block));
739                         if (block < le32_to_cpu(ix->ei_block))
740                                 break;
741                         chix = ix;
742                 }
743                 BUG_ON(chix != path->p_idx);
744         }
745 #endif
746
747 }
748
749 /*
750  * ext4_ext_binsearch:
751  * binary search for closest extent of the given block
752  * the header must be checked before calling this
753  */
754 static void
755 ext4_ext_binsearch(struct inode *inode,
756                 struct ext4_ext_path *path, ext4_lblk_t block)
757 {
758         struct ext4_extent_header *eh = path->p_hdr;
759         struct ext4_extent *r, *l, *m;
760
761         if (eh->eh_entries == 0) {
762                 /*
763                  * this leaf is empty:
764                  * we get such a leaf in split/add case
765                  */
766                 return;
767         }
768
769         ext_debug("binsearch for %u:  ", block);
770
771         l = EXT_FIRST_EXTENT(eh) + 1;
772         r = EXT_LAST_EXTENT(eh);
773
774         while (l <= r) {
775                 m = l + (r - l) / 2;
776                 if (block < le32_to_cpu(m->ee_block))
777                         r = m - 1;
778                 else
779                         l = m + 1;
780                 ext_debug("%p(%u):%p(%u):%p(%u) ", l, le32_to_cpu(l->ee_block),
781                                 m, le32_to_cpu(m->ee_block),
782                                 r, le32_to_cpu(r->ee_block));
783         }
784
785         path->p_ext = l - 1;
786         ext_debug("  -> %d:%llu:[%d]%d ",
787                         le32_to_cpu(path->p_ext->ee_block),
788                         ext4_ext_pblock(path->p_ext),
789                         ext4_ext_is_uninitialized(path->p_ext),
790                         ext4_ext_get_actual_len(path->p_ext));
791
792 #ifdef CHECK_BINSEARCH
793         {
794                 struct ext4_extent *chex, *ex;
795                 int k;
796
797                 chex = ex = EXT_FIRST_EXTENT(eh);
798                 for (k = 0; k < le16_to_cpu(eh->eh_entries); k++, ex++) {
799                         BUG_ON(k && le32_to_cpu(ex->ee_block)
800                                           <= le32_to_cpu(ex[-1].ee_block));
801                         if (block < le32_to_cpu(ex->ee_block))
802                                 break;
803                         chex = ex;
804                 }
805                 BUG_ON(chex != path->p_ext);
806         }
807 #endif
808
809 }
810
811 int ext4_ext_tree_init(handle_t *handle, struct inode *inode)
812 {
813         struct ext4_extent_header *eh;
814
815         eh = ext_inode_hdr(inode);
816         eh->eh_depth = 0;
817         eh->eh_entries = 0;
818         eh->eh_magic = EXT4_EXT_MAGIC;
819         eh->eh_max = cpu_to_le16(ext4_ext_space_root(inode, 0));
820         ext4_mark_inode_dirty(handle, inode);
821         return 0;
822 }
823
824 struct ext4_ext_path *
825 ext4_ext_find_extent(struct inode *inode, ext4_lblk_t block,
826                      struct ext4_ext_path *path, int flags)
827 {
828         struct ext4_extent_header *eh;
829         struct buffer_head *bh;
830         short int depth, i, ppos = 0, alloc = 0;
831         int ret;
832
833         eh = ext_inode_hdr(inode);
834         depth = ext_depth(inode);
835
836         /* account possible depth increase */
837         if (!path) {
838                 path = kzalloc(sizeof(struct ext4_ext_path) * (depth + 2),
839                                 GFP_NOFS);
840                 if (!path)
841                         return ERR_PTR(-ENOMEM);
842                 alloc = 1;
843         }
844         path[0].p_hdr = eh;
845         path[0].p_bh = NULL;
846
847         i = depth;
848         /* walk through the tree */
849         while (i) {
850                 ext_debug("depth %d: num %d, max %d\n",
851                           ppos, le16_to_cpu(eh->eh_entries), le16_to_cpu(eh->eh_max));
852
853                 ext4_ext_binsearch_idx(inode, path + ppos, block);
854                 path[ppos].p_block = ext4_idx_pblock(path[ppos].p_idx);
855                 path[ppos].p_depth = i;
856                 path[ppos].p_ext = NULL;
857
858                 bh = read_extent_tree_block(inode, path[ppos].p_block, --i,
859                                             flags);
860                 if (IS_ERR(bh)) {
861                         ret = PTR_ERR(bh);
862                         goto err;
863                 }
864
865                 eh = ext_block_hdr(bh);
866                 ppos++;
867                 if (unlikely(ppos > depth)) {
868                         put_bh(bh);
869                         EXT4_ERROR_INODE(inode,
870                                          "ppos %d > depth %d", ppos, depth);
871                         ret = -EIO;
872                         goto err;
873                 }
874                 path[ppos].p_bh = bh;
875                 path[ppos].p_hdr = eh;
876         }
877
878         path[ppos].p_depth = i;
879         path[ppos].p_ext = NULL;
880         path[ppos].p_idx = NULL;
881
882         /* find extent */
883         ext4_ext_binsearch(inode, path + ppos, block);
884         /* if not an empty leaf */
885         if (path[ppos].p_ext)
886                 path[ppos].p_block = ext4_ext_pblock(path[ppos].p_ext);
887
888         ext4_ext_show_path(inode, path);
889
890         return path;
891
892 err:
893         ext4_ext_drop_refs(path);
894         if (alloc)
895                 kfree(path);
896         return ERR_PTR(ret);
897 }
898
899 /*
900  * ext4_ext_insert_index:
901  * insert new index [@logical;@ptr] into the block at @curp;
902  * check where to insert: before @curp or after @curp
903  */
904 static int ext4_ext_insert_index(handle_t *handle, struct inode *inode,
905                                  struct ext4_ext_path *curp,
906                                  int logical, ext4_fsblk_t ptr)
907 {
908         struct ext4_extent_idx *ix;
909         int len, err;
910
911         err = ext4_ext_get_access(handle, inode, curp);
912         if (err)
913                 return err;
914
915         if (unlikely(logical == le32_to_cpu(curp->p_idx->ei_block))) {
916                 EXT4_ERROR_INODE(inode,
917                                  "logical %d == ei_block %d!",
918                                  logical, le32_to_cpu(curp->p_idx->ei_block));
919                 return -EIO;
920         }
921
922         if (unlikely(le16_to_cpu(curp->p_hdr->eh_entries)
923                              >= le16_to_cpu(curp->p_hdr->eh_max))) {
924                 EXT4_ERROR_INODE(inode,
925                                  "eh_entries %d >= eh_max %d!",
926                                  le16_to_cpu(curp->p_hdr->eh_entries),
927                                  le16_to_cpu(curp->p_hdr->eh_max));
928                 return -EIO;
929         }
930
931         if (logical > le32_to_cpu(curp->p_idx->ei_block)) {
932                 /* insert after */
933                 ext_debug("insert new index %d after: %llu\n", logical, ptr);
934                 ix = curp->p_idx + 1;
935         } else {
936                 /* insert before */
937                 ext_debug("insert new index %d before: %llu\n", logical, ptr);
938                 ix = curp->p_idx;
939         }
940
941         len = EXT_LAST_INDEX(curp->p_hdr) - ix + 1;
942         BUG_ON(len < 0);
943         if (len > 0) {
944                 ext_debug("insert new index %d: "
945                                 "move %d indices from 0x%p to 0x%p\n",
946                                 logical, len, ix, ix + 1);
947                 memmove(ix + 1, ix, len * sizeof(struct ext4_extent_idx));
948         }
949
950         if (unlikely(ix > EXT_MAX_INDEX(curp->p_hdr))) {
951                 EXT4_ERROR_INODE(inode, "ix > EXT_MAX_INDEX!");
952                 return -EIO;
953         }
954
955         ix->ei_block = cpu_to_le32(logical);
956         ext4_idx_store_pblock(ix, ptr);
957         le16_add_cpu(&curp->p_hdr->eh_entries, 1);
958
959         if (unlikely(ix > EXT_LAST_INDEX(curp->p_hdr))) {
960                 EXT4_ERROR_INODE(inode, "ix > EXT_LAST_INDEX!");
961                 return -EIO;
962         }
963
964         err = ext4_ext_dirty(handle, inode, curp);
965         ext4_std_error(inode->i_sb, err);
966
967         return err;
968 }
969
970 /*
971  * ext4_ext_split:
972  * inserts new subtree into the path, using free index entry
973  * at depth @at:
974  * - allocates all needed blocks (new leaf and all intermediate index blocks)
975  * - makes decision where to split
976  * - moves remaining extents and index entries (right to the split point)
977  *   into the newly allocated blocks
978  * - initializes subtree
979  */
980 static int ext4_ext_split(handle_t *handle, struct inode *inode,
981                           unsigned int flags,
982                           struct ext4_ext_path *path,
983                           struct ext4_extent *newext, int at)
984 {
985         struct buffer_head *bh = NULL;
986         int depth = ext_depth(inode);
987         struct ext4_extent_header *neh;
988         struct ext4_extent_idx *fidx;
989         int i = at, k, m, a;
990         ext4_fsblk_t newblock, oldblock;
991         __le32 border;
992         ext4_fsblk_t *ablocks = NULL; /* array of allocated blocks */
993         int err = 0;
994
995         /* make decision: where to split? */
996         /* FIXME: now decision is simplest: at current extent */
997
998         /* if current leaf will be split, then we should use
999          * border from split point */
1000         if (unlikely(path[depth].p_ext > EXT_MAX_EXTENT(path[depth].p_hdr))) {
1001                 EXT4_ERROR_INODE(inode, "p_ext > EXT_MAX_EXTENT!");
1002                 return -EIO;
1003         }
1004         if (path[depth].p_ext != EXT_MAX_EXTENT(path[depth].p_hdr)) {
1005                 border = path[depth].p_ext[1].ee_block;
1006                 ext_debug("leaf will be split."
1007                                 " next leaf starts at %d\n",
1008                                   le32_to_cpu(border));
1009         } else {
1010                 border = newext->ee_block;
1011                 ext_debug("leaf will be added."
1012                                 " next leaf starts at %d\n",
1013                                 le32_to_cpu(border));
1014         }
1015
1016         /*
1017          * If error occurs, then we break processing
1018          * and mark filesystem read-only. index won't
1019          * be inserted and tree will be in consistent
1020          * state. Next mount will repair buffers too.
1021          */
1022
1023         /*
1024          * Get array to track all allocated blocks.
1025          * We need this to handle errors and free blocks
1026          * upon them.
1027          */
1028         ablocks = kzalloc(sizeof(ext4_fsblk_t) * depth, GFP_NOFS);
1029         if (!ablocks)
1030                 return -ENOMEM;
1031
1032         /* allocate all needed blocks */
1033         ext_debug("allocate %d blocks for indexes/leaf\n", depth - at);
1034         for (a = 0; a < depth - at; a++) {
1035                 newblock = ext4_ext_new_meta_block(handle, inode, path,
1036                                                    newext, &err, flags);
1037                 if (newblock == 0)
1038                         goto cleanup;
1039                 ablocks[a] = newblock;
1040         }
1041
1042         /* initialize new leaf */
1043         newblock = ablocks[--a];
1044         if (unlikely(newblock == 0)) {
1045                 EXT4_ERROR_INODE(inode, "newblock == 0!");
1046                 err = -EIO;
1047                 goto cleanup;
1048         }
1049         bh = sb_getblk(inode->i_sb, newblock);
1050         if (unlikely(!bh)) {
1051                 err = -ENOMEM;
1052                 goto cleanup;
1053         }
1054         lock_buffer(bh);
1055
1056         err = ext4_journal_get_create_access(handle, bh);
1057         if (err)
1058                 goto cleanup;
1059
1060         neh = ext_block_hdr(bh);
1061         neh->eh_entries = 0;
1062         neh->eh_max = cpu_to_le16(ext4_ext_space_block(inode, 0));
1063         neh->eh_magic = EXT4_EXT_MAGIC;
1064         neh->eh_depth = 0;
1065
1066         /* move remainder of path[depth] to the new leaf */
1067         if (unlikely(path[depth].p_hdr->eh_entries !=
1068                      path[depth].p_hdr->eh_max)) {
1069                 EXT4_ERROR_INODE(inode, "eh_entries %d != eh_max %d!",
1070                                  path[depth].p_hdr->eh_entries,
1071                                  path[depth].p_hdr->eh_max);
1072                 err = -EIO;
1073                 goto cleanup;
1074         }
1075         /* start copy from next extent */
1076         m = EXT_MAX_EXTENT(path[depth].p_hdr) - path[depth].p_ext++;
1077         ext4_ext_show_move(inode, path, newblock, depth);
1078         if (m) {
1079                 struct ext4_extent *ex;
1080                 ex = EXT_FIRST_EXTENT(neh);
1081                 memmove(ex, path[depth].p_ext, sizeof(struct ext4_extent) * m);
1082                 le16_add_cpu(&neh->eh_entries, m);
1083         }
1084
1085         ext4_extent_block_csum_set(inode, neh);
1086         set_buffer_uptodate(bh);
1087         unlock_buffer(bh);
1088
1089         err = ext4_handle_dirty_metadata(handle, inode, bh);
1090         if (err)
1091                 goto cleanup;
1092         brelse(bh);
1093         bh = NULL;
1094
1095         /* correct old leaf */
1096         if (m) {
1097                 err = ext4_ext_get_access(handle, inode, path + depth);
1098                 if (err)
1099                         goto cleanup;
1100                 le16_add_cpu(&path[depth].p_hdr->eh_entries, -m);
1101                 err = ext4_ext_dirty(handle, inode, path + depth);
1102                 if (err)
1103                         goto cleanup;
1104
1105         }
1106
1107         /* create intermediate indexes */
1108         k = depth - at - 1;
1109         if (unlikely(k < 0)) {
1110                 EXT4_ERROR_INODE(inode, "k %d < 0!", k);
1111                 err = -EIO;
1112                 goto cleanup;
1113         }
1114         if (k)
1115                 ext_debug("create %d intermediate indices\n", k);
1116         /* insert new index into current index block */
1117         /* current depth stored in i var */
1118         i = depth - 1;
1119         while (k--) {
1120                 oldblock = newblock;
1121                 newblock = ablocks[--a];
1122                 bh = sb_getblk(inode->i_sb, newblock);
1123                 if (unlikely(!bh)) {
1124                         err = -ENOMEM;
1125                         goto cleanup;
1126                 }
1127                 lock_buffer(bh);
1128
1129                 err = ext4_journal_get_create_access(handle, bh);
1130                 if (err)
1131                         goto cleanup;
1132
1133                 neh = ext_block_hdr(bh);
1134                 neh->eh_entries = cpu_to_le16(1);
1135                 neh->eh_magic = EXT4_EXT_MAGIC;
1136                 neh->eh_max = cpu_to_le16(ext4_ext_space_block_idx(inode, 0));
1137                 neh->eh_depth = cpu_to_le16(depth - i);
1138                 fidx = EXT_FIRST_INDEX(neh);
1139                 fidx->ei_block = border;
1140                 ext4_idx_store_pblock(fidx, oldblock);
1141
1142                 ext_debug("int.index at %d (block %llu): %u -> %llu\n",
1143                                 i, newblock, le32_to_cpu(border), oldblock);
1144
1145                 /* move remainder of path[i] to the new index block */
1146                 if (unlikely(EXT_MAX_INDEX(path[i].p_hdr) !=
1147                                         EXT_LAST_INDEX(path[i].p_hdr))) {
1148                         EXT4_ERROR_INODE(inode,
1149                                          "EXT_MAX_INDEX != EXT_LAST_INDEX ee_block %d!",
1150                                          le32_to_cpu(path[i].p_ext->ee_block));
1151                         err = -EIO;
1152                         goto cleanup;
1153                 }
1154                 /* start copy indexes */
1155                 m = EXT_MAX_INDEX(path[i].p_hdr) - path[i].p_idx++;
1156                 ext_debug("cur 0x%p, last 0x%p\n", path[i].p_idx,
1157                                 EXT_MAX_INDEX(path[i].p_hdr));
1158                 ext4_ext_show_move(inode, path, newblock, i);
1159                 if (m) {
1160                         memmove(++fidx, path[i].p_idx,
1161                                 sizeof(struct ext4_extent_idx) * m);
1162                         le16_add_cpu(&neh->eh_entries, m);
1163                 }
1164                 ext4_extent_block_csum_set(inode, neh);
1165                 set_buffer_uptodate(bh);
1166                 unlock_buffer(bh);
1167
1168                 err = ext4_handle_dirty_metadata(handle, inode, bh);
1169                 if (err)
1170                         goto cleanup;
1171                 brelse(bh);
1172                 bh = NULL;
1173
1174                 /* correct old index */
1175                 if (m) {
1176                         err = ext4_ext_get_access(handle, inode, path + i);
1177                         if (err)
1178                                 goto cleanup;
1179                         le16_add_cpu(&path[i].p_hdr->eh_entries, -m);
1180                         err = ext4_ext_dirty(handle, inode, path + i);
1181                         if (err)
1182                                 goto cleanup;
1183                 }
1184
1185                 i--;
1186         }
1187
1188         /* insert new index */
1189         err = ext4_ext_insert_index(handle, inode, path + at,
1190                                     le32_to_cpu(border), newblock);
1191
1192 cleanup:
1193         if (bh) {
1194                 if (buffer_locked(bh))
1195                         unlock_buffer(bh);
1196                 brelse(bh);
1197         }
1198
1199         if (err) {
1200                 /* free all allocated blocks in error case */
1201                 for (i = 0; i < depth; i++) {
1202                         if (!ablocks[i])
1203                                 continue;
1204                         ext4_free_blocks(handle, inode, NULL, ablocks[i], 1,
1205                                          EXT4_FREE_BLOCKS_METADATA);
1206                 }
1207         }
1208         kfree(ablocks);
1209
1210         return err;
1211 }
1212
1213 /*
1214  * ext4_ext_grow_indepth:
1215  * implements tree growing procedure:
1216  * - allocates new block
1217  * - moves top-level data (index block or leaf) into the new block
1218  * - initializes new top-level, creating index that points to the
1219  *   just created block
1220  */
1221 static int ext4_ext_grow_indepth(handle_t *handle, struct inode *inode,
1222                                  unsigned int flags,
1223                                  struct ext4_extent *newext)
1224 {
1225         struct ext4_extent_header *neh;
1226         struct buffer_head *bh;
1227         ext4_fsblk_t newblock;
1228         int err = 0;
1229
1230         newblock = ext4_ext_new_meta_block(handle, inode, NULL,
1231                 newext, &err, flags);
1232         if (newblock == 0)
1233                 return err;
1234
1235         bh = sb_getblk(inode->i_sb, newblock);
1236         if (unlikely(!bh))
1237                 return -ENOMEM;
1238         lock_buffer(bh);
1239
1240         err = ext4_journal_get_create_access(handle, bh);
1241         if (err) {
1242                 unlock_buffer(bh);
1243                 goto out;
1244         }
1245
1246         /* move top-level index/leaf into new block */
1247         memmove(bh->b_data, EXT4_I(inode)->i_data,
1248                 sizeof(EXT4_I(inode)->i_data));
1249
1250         /* set size of new block */
1251         neh = ext_block_hdr(bh);
1252         /* old root could have indexes or leaves
1253          * so calculate e_max right way */
1254         if (ext_depth(inode))
1255                 neh->eh_max = cpu_to_le16(ext4_ext_space_block_idx(inode, 0));
1256         else
1257                 neh->eh_max = cpu_to_le16(ext4_ext_space_block(inode, 0));
1258         neh->eh_magic = EXT4_EXT_MAGIC;
1259         ext4_extent_block_csum_set(inode, neh);
1260         set_buffer_uptodate(bh);
1261         unlock_buffer(bh);
1262
1263         err = ext4_handle_dirty_metadata(handle, inode, bh);
1264         if (err)
1265                 goto out;
1266
1267         /* Update top-level index: num,max,pointer */
1268         neh = ext_inode_hdr(inode);
1269         neh->eh_entries = cpu_to_le16(1);
1270         ext4_idx_store_pblock(EXT_FIRST_INDEX(neh), newblock);
1271         if (neh->eh_depth == 0) {
1272                 /* Root extent block becomes index block */
1273                 neh->eh_max = cpu_to_le16(ext4_ext_space_root_idx(inode, 0));
1274                 EXT_FIRST_INDEX(neh)->ei_block =
1275                         EXT_FIRST_EXTENT(neh)->ee_block;
1276         }
1277         ext_debug("new root: num %d(%d), lblock %d, ptr %llu\n",
1278                   le16_to_cpu(neh->eh_entries), le16_to_cpu(neh->eh_max),
1279                   le32_to_cpu(EXT_FIRST_INDEX(neh)->ei_block),
1280                   ext4_idx_pblock(EXT_FIRST_INDEX(neh)));
1281
1282         le16_add_cpu(&neh->eh_depth, 1);
1283         ext4_mark_inode_dirty(handle, inode);
1284 out:
1285         brelse(bh);
1286
1287         return err;
1288 }
1289
1290 /*
1291  * ext4_ext_create_new_leaf:
1292  * finds empty index and adds new leaf.
1293  * if no free index is found, then it requests in-depth growing.
1294  */
1295 static int ext4_ext_create_new_leaf(handle_t *handle, struct inode *inode,
1296                                     unsigned int mb_flags,
1297                                     unsigned int gb_flags,
1298                                     struct ext4_ext_path *path,
1299                                     struct ext4_extent *newext)
1300 {
1301         struct ext4_ext_path *curp;
1302         int depth, i, err = 0;
1303
1304 repeat:
1305         i = depth = ext_depth(inode);
1306
1307         /* walk up to the tree and look for free index entry */
1308         curp = path + depth;
1309         while (i > 0 && !EXT_HAS_FREE_INDEX(curp)) {
1310                 i--;
1311                 curp--;
1312         }
1313
1314         /* we use already allocated block for index block,
1315          * so subsequent data blocks should be contiguous */
1316         if (EXT_HAS_FREE_INDEX(curp)) {
1317                 /* if we found index with free entry, then use that
1318                  * entry: create all needed subtree and add new leaf */
1319                 err = ext4_ext_split(handle, inode, mb_flags, path, newext, i);
1320                 if (err)
1321                         goto out;
1322
1323                 /* refill path */
1324                 ext4_ext_drop_refs(path);
1325                 path = ext4_ext_find_extent(inode,
1326                                     (ext4_lblk_t)le32_to_cpu(newext->ee_block),
1327                                     path, gb_flags);
1328                 if (IS_ERR(path))
1329                         err = PTR_ERR(path);
1330         } else {
1331                 /* tree is full, time to grow in depth */
1332                 err = ext4_ext_grow_indepth(handle, inode, mb_flags, newext);
1333                 if (err)
1334                         goto out;
1335
1336                 /* refill path */
1337                 ext4_ext_drop_refs(path);
1338                 path = ext4_ext_find_extent(inode,
1339                                    (ext4_lblk_t)le32_to_cpu(newext->ee_block),
1340                                     path, gb_flags);
1341                 if (IS_ERR(path)) {
1342                         err = PTR_ERR(path);
1343                         goto out;
1344                 }
1345
1346                 /*
1347                  * only first (depth 0 -> 1) produces free space;
1348                  * in all other cases we have to split the grown tree
1349                  */
1350                 depth = ext_depth(inode);
1351                 if (path[depth].p_hdr->eh_entries == path[depth].p_hdr->eh_max) {
1352                         /* now we need to split */
1353                         goto repeat;
1354                 }
1355         }
1356
1357 out:
1358         return err;
1359 }
1360
1361 /*
1362  * search the closest allocated block to the left for *logical
1363  * and returns it at @logical + it's physical address at @phys
1364  * if *logical is the smallest allocated block, the function
1365  * returns 0 at @phys
1366  * return value contains 0 (success) or error code
1367  */
1368 static int ext4_ext_search_left(struct inode *inode,
1369                                 struct ext4_ext_path *path,
1370                                 ext4_lblk_t *logical, ext4_fsblk_t *phys)
1371 {
1372         struct ext4_extent_idx *ix;
1373         struct ext4_extent *ex;
1374         int depth, ee_len;
1375
1376         if (unlikely(path == NULL)) {
1377                 EXT4_ERROR_INODE(inode, "path == NULL *logical %d!", *logical);
1378                 return -EIO;
1379         }
1380         depth = path->p_depth;
1381         *phys = 0;
1382
1383         if (depth == 0 && path->p_ext == NULL)
1384                 return 0;
1385
1386         /* usually extent in the path covers blocks smaller
1387          * then *logical, but it can be that extent is the
1388          * first one in the file */
1389
1390         ex = path[depth].p_ext;
1391         ee_len = ext4_ext_get_actual_len(ex);
1392         if (*logical < le32_to_cpu(ex->ee_block)) {
1393                 if (unlikely(EXT_FIRST_EXTENT(path[depth].p_hdr) != ex)) {
1394                         EXT4_ERROR_INODE(inode,
1395                                          "EXT_FIRST_EXTENT != ex *logical %d ee_block %d!",
1396                                          *logical, le32_to_cpu(ex->ee_block));
1397                         return -EIO;
1398                 }
1399                 while (--depth >= 0) {
1400                         ix = path[depth].p_idx;
1401                         if (unlikely(ix != EXT_FIRST_INDEX(path[depth].p_hdr))) {
1402                                 EXT4_ERROR_INODE(inode,
1403                                   "ix (%d) != EXT_FIRST_INDEX (%d) (depth %d)!",
1404                                   ix != NULL ? le32_to_cpu(ix->ei_block) : 0,
1405                                   EXT_FIRST_INDEX(path[depth].p_hdr) != NULL ?
1406                 le32_to_cpu(EXT_FIRST_INDEX(path[depth].p_hdr)->ei_block) : 0,
1407                                   depth);
1408                                 return -EIO;
1409                         }
1410                 }
1411                 return 0;
1412         }
1413
1414         if (unlikely(*logical < (le32_to_cpu(ex->ee_block) + ee_len))) {
1415                 EXT4_ERROR_INODE(inode,
1416                                  "logical %d < ee_block %d + ee_len %d!",
1417                                  *logical, le32_to_cpu(ex->ee_block), ee_len);
1418                 return -EIO;
1419         }
1420
1421         *logical = le32_to_cpu(ex->ee_block) + ee_len - 1;
1422         *phys = ext4_ext_pblock(ex) + ee_len - 1;
1423         return 0;
1424 }
1425
1426 /*
1427  * search the closest allocated block to the right for *logical
1428  * and returns it at @logical + it's physical address at @phys
1429  * if *logical is the largest allocated block, the function
1430  * returns 0 at @phys
1431  * return value contains 0 (success) or error code
1432  */
1433 static int ext4_ext_search_right(struct inode *inode,
1434                                  struct ext4_ext_path *path,
1435                                  ext4_lblk_t *logical, ext4_fsblk_t *phys,
1436                                  struct ext4_extent **ret_ex)
1437 {
1438         struct buffer_head *bh = NULL;
1439         struct ext4_extent_header *eh;
1440         struct ext4_extent_idx *ix;
1441         struct ext4_extent *ex;
1442         ext4_fsblk_t block;
1443         int depth;      /* Note, NOT eh_depth; depth from top of tree */
1444         int ee_len;
1445
1446         if (unlikely(path == NULL)) {
1447                 EXT4_ERROR_INODE(inode, "path == NULL *logical %d!", *logical);
1448                 return -EIO;
1449         }
1450         depth = path->p_depth;
1451         *phys = 0;
1452
1453         if (depth == 0 && path->p_ext == NULL)
1454                 return 0;
1455
1456         /* usually extent in the path covers blocks smaller
1457          * then *logical, but it can be that extent is the
1458          * first one in the file */
1459
1460         ex = path[depth].p_ext;
1461         ee_len = ext4_ext_get_actual_len(ex);
1462         if (*logical < le32_to_cpu(ex->ee_block)) {
1463                 if (unlikely(EXT_FIRST_EXTENT(path[depth].p_hdr) != ex)) {
1464                         EXT4_ERROR_INODE(inode,
1465                                          "first_extent(path[%d].p_hdr) != ex",
1466                                          depth);
1467                         return -EIO;
1468                 }
1469                 while (--depth >= 0) {
1470                         ix = path[depth].p_idx;
1471                         if (unlikely(ix != EXT_FIRST_INDEX(path[depth].p_hdr))) {
1472                                 EXT4_ERROR_INODE(inode,
1473                                                  "ix != EXT_FIRST_INDEX *logical %d!",
1474                                                  *logical);
1475                                 return -EIO;
1476                         }
1477                 }
1478                 goto found_extent;
1479         }
1480
1481         if (unlikely(*logical < (le32_to_cpu(ex->ee_block) + ee_len))) {
1482                 EXT4_ERROR_INODE(inode,
1483                                  "logical %d < ee_block %d + ee_len %d!",
1484                                  *logical, le32_to_cpu(ex->ee_block), ee_len);
1485                 return -EIO;
1486         }
1487
1488         if (ex != EXT_LAST_EXTENT(path[depth].p_hdr)) {
1489                 /* next allocated block in this leaf */
1490                 ex++;
1491                 goto found_extent;
1492         }
1493
1494         /* go up and search for index to the right */
1495         while (--depth >= 0) {
1496                 ix = path[depth].p_idx;
1497                 if (ix != EXT_LAST_INDEX(path[depth].p_hdr))
1498                         goto got_index;
1499         }
1500
1501         /* we've gone up to the root and found no index to the right */
1502         return 0;
1503
1504 got_index:
1505         /* we've found index to the right, let's
1506          * follow it and find the closest allocated
1507          * block to the right */
1508         ix++;
1509         block = ext4_idx_pblock(ix);
1510         while (++depth < path->p_depth) {
1511                 /* subtract from p_depth to get proper eh_depth */
1512                 bh = read_extent_tree_block(inode, block,
1513                                             path->p_depth - depth, 0);
1514                 if (IS_ERR(bh))
1515                         return PTR_ERR(bh);
1516                 eh = ext_block_hdr(bh);
1517                 ix = EXT_FIRST_INDEX(eh);
1518                 block = ext4_idx_pblock(ix);
1519                 put_bh(bh);
1520         }
1521
1522         bh = read_extent_tree_block(inode, block, path->p_depth - depth, 0);
1523         if (IS_ERR(bh))
1524                 return PTR_ERR(bh);
1525         eh = ext_block_hdr(bh);
1526         ex = EXT_FIRST_EXTENT(eh);
1527 found_extent:
1528         *logical = le32_to_cpu(ex->ee_block);
1529         *phys = ext4_ext_pblock(ex);
1530         *ret_ex = ex;
1531         if (bh)
1532                 put_bh(bh);
1533         return 0;
1534 }
1535
1536 /*
1537  * ext4_ext_next_allocated_block:
1538  * returns allocated block in subsequent extent or EXT_MAX_BLOCKS.
1539  * NOTE: it considers block number from index entry as
1540  * allocated block. Thus, index entries have to be consistent
1541  * with leaves.
1542  */
1543 static ext4_lblk_t
1544 ext4_ext_next_allocated_block(struct ext4_ext_path *path)
1545 {
1546         int depth;
1547
1548         BUG_ON(path == NULL);
1549         depth = path->p_depth;
1550
1551         if (depth == 0 && path->p_ext == NULL)
1552                 return EXT_MAX_BLOCKS;
1553
1554         while (depth >= 0) {
1555                 if (depth == path->p_depth) {
1556                         /* leaf */
1557                         if (path[depth].p_ext &&
1558                                 path[depth].p_ext !=
1559                                         EXT_LAST_EXTENT(path[depth].p_hdr))
1560                           return le32_to_cpu(path[depth].p_ext[1].ee_block);
1561                 } else {
1562                         /* index */
1563                         if (path[depth].p_idx !=
1564                                         EXT_LAST_INDEX(path[depth].p_hdr))
1565                           return le32_to_cpu(path[depth].p_idx[1].ei_block);
1566                 }
1567                 depth--;
1568         }
1569
1570         return EXT_MAX_BLOCKS;
1571 }
1572
1573 /*
1574  * ext4_ext_next_leaf_block:
1575  * returns first allocated block from next leaf or EXT_MAX_BLOCKS
1576  */
1577 static ext4_lblk_t ext4_ext_next_leaf_block(struct ext4_ext_path *path)
1578 {
1579         int depth;
1580
1581         BUG_ON(path == NULL);
1582         depth = path->p_depth;
1583
1584         /* zero-tree has no leaf blocks at all */
1585         if (depth == 0)
1586                 return EXT_MAX_BLOCKS;
1587
1588         /* go to index block */
1589         depth--;
1590
1591         while (depth >= 0) {
1592                 if (path[depth].p_idx !=
1593                                 EXT_LAST_INDEX(path[depth].p_hdr))
1594                         return (ext4_lblk_t)
1595                                 le32_to_cpu(path[depth].p_idx[1].ei_block);
1596                 depth--;
1597         }
1598
1599         return EXT_MAX_BLOCKS;
1600 }
1601
1602 /*
1603  * ext4_ext_correct_indexes:
1604  * if leaf gets modified and modified extent is first in the leaf,
1605  * then we have to correct all indexes above.
1606  * TODO: do we need to correct tree in all cases?
1607  */
1608 static int ext4_ext_correct_indexes(handle_t *handle, struct inode *inode,
1609                                 struct ext4_ext_path *path)
1610 {
1611         struct ext4_extent_header *eh;
1612         int depth = ext_depth(inode);
1613         struct ext4_extent *ex;
1614         __le32 border;
1615         int k, err = 0;
1616
1617         eh = path[depth].p_hdr;
1618         ex = path[depth].p_ext;
1619
1620         if (unlikely(ex == NULL || eh == NULL)) {
1621                 EXT4_ERROR_INODE(inode,
1622                                  "ex %p == NULL or eh %p == NULL", ex, eh);
1623                 return -EIO;
1624         }
1625
1626         if (depth == 0) {
1627                 /* there is no tree at all */
1628                 return 0;
1629         }
1630
1631         if (ex != EXT_FIRST_EXTENT(eh)) {
1632                 /* we correct tree if first leaf got modified only */
1633                 return 0;
1634         }
1635
1636         /*
1637          * TODO: we need correction if border is smaller than current one
1638          */
1639         k = depth - 1;
1640         border = path[depth].p_ext->ee_block;
1641         err = ext4_ext_get_access(handle, inode, path + k);
1642         if (err)
1643                 return err;
1644         path[k].p_idx->ei_block = border;
1645         err = ext4_ext_dirty(handle, inode, path + k);
1646         if (err)
1647                 return err;
1648
1649         while (k--) {
1650                 /* change all left-side indexes */
1651                 if (path[k+1].p_idx != EXT_FIRST_INDEX(path[k+1].p_hdr))
1652                         break;
1653                 err = ext4_ext_get_access(handle, inode, path + k);
1654                 if (err)
1655                         break;
1656                 path[k].p_idx->ei_block = border;
1657                 err = ext4_ext_dirty(handle, inode, path + k);
1658                 if (err)
1659                         break;
1660         }
1661
1662         return err;
1663 }
1664
1665 int
1666 ext4_can_extents_be_merged(struct inode *inode, struct ext4_extent *ex1,
1667                                 struct ext4_extent *ex2)
1668 {
1669         unsigned short ext1_ee_len, ext2_ee_len;
1670
1671         /*
1672          * Make sure that both extents are initialized. We don't merge
1673          * uninitialized extents so that we can be sure that end_io code has
1674          * the extent that was written properly split out and conversion to
1675          * initialized is trivial.
1676          */
1677         if (ext4_ext_is_uninitialized(ex1) || ext4_ext_is_uninitialized(ex2))
1678                 return 0;
1679
1680         ext1_ee_len = ext4_ext_get_actual_len(ex1);
1681         ext2_ee_len = ext4_ext_get_actual_len(ex2);
1682
1683         if (le32_to_cpu(ex1->ee_block) + ext1_ee_len !=
1684                         le32_to_cpu(ex2->ee_block))
1685                 return 0;
1686
1687         /*
1688          * To allow future support for preallocated extents to be added
1689          * as an RO_COMPAT feature, refuse to merge to extents if
1690          * this can result in the top bit of ee_len being set.
1691          */
1692         if (ext1_ee_len + ext2_ee_len > EXT_INIT_MAX_LEN)
1693                 return 0;
1694 #ifdef AGGRESSIVE_TEST
1695         if (ext1_ee_len >= 4)
1696                 return 0;
1697 #endif
1698
1699         if (ext4_ext_pblock(ex1) + ext1_ee_len == ext4_ext_pblock(ex2))
1700                 return 1;
1701         return 0;
1702 }
1703
1704 /*
1705  * This function tries to merge the "ex" extent to the next extent in the tree.
1706  * It always tries to merge towards right. If you want to merge towards
1707  * left, pass "ex - 1" as argument instead of "ex".
1708  * Returns 0 if the extents (ex and ex+1) were _not_ merged and returns
1709  * 1 if they got merged.
1710  */
1711 static int ext4_ext_try_to_merge_right(struct inode *inode,
1712                                  struct ext4_ext_path *path,
1713                                  struct ext4_extent *ex)
1714 {
1715         struct ext4_extent_header *eh;
1716         unsigned int depth, len;
1717         int merge_done = 0;
1718
1719         depth = ext_depth(inode);
1720         BUG_ON(path[depth].p_hdr == NULL);
1721         eh = path[depth].p_hdr;
1722
1723         while (ex < EXT_LAST_EXTENT(eh)) {
1724                 if (!ext4_can_extents_be_merged(inode, ex, ex + 1))
1725                         break;
1726                 /* merge with next extent! */
1727                 ex->ee_len = cpu_to_le16(ext4_ext_get_actual_len(ex)
1728                                 + ext4_ext_get_actual_len(ex + 1));
1729
1730                 if (ex + 1 < EXT_LAST_EXTENT(eh)) {
1731                         len = (EXT_LAST_EXTENT(eh) - ex - 1)
1732                                 * sizeof(struct ext4_extent);
1733                         memmove(ex + 1, ex + 2, len);
1734                 }
1735                 le16_add_cpu(&eh->eh_entries, -1);
1736                 merge_done = 1;
1737                 WARN_ON(eh->eh_entries == 0);
1738                 if (!eh->eh_entries)
1739                         EXT4_ERROR_INODE(inode, "eh->eh_entries = 0!");
1740         }
1741
1742         return merge_done;
1743 }
1744
1745 /*
1746  * This function does a very simple check to see if we can collapse
1747  * an extent tree with a single extent tree leaf block into the inode.
1748  */
1749 static void ext4_ext_try_to_merge_up(handle_t *handle,
1750                                      struct inode *inode,
1751                                      struct ext4_ext_path *path)
1752 {
1753         size_t s;
1754         unsigned max_root = ext4_ext_space_root(inode, 0);
1755         ext4_fsblk_t blk;
1756
1757         if ((path[0].p_depth != 1) ||
1758             (le16_to_cpu(path[0].p_hdr->eh_entries) != 1) ||
1759             (le16_to_cpu(path[1].p_hdr->eh_entries) > max_root))
1760                 return;
1761
1762         /*
1763          * We need to modify the block allocation bitmap and the block
1764          * group descriptor to release the extent tree block.  If we
1765          * can't get the journal credits, give up.
1766          */
1767         if (ext4_journal_extend(handle, 2))
1768                 return;
1769
1770         /*
1771          * Copy the extent data up to the inode
1772          */
1773         blk = ext4_idx_pblock(path[0].p_idx);
1774         s = le16_to_cpu(path[1].p_hdr->eh_entries) *
1775                 sizeof(struct ext4_extent_idx);
1776         s += sizeof(struct ext4_extent_header);
1777
1778         memcpy(path[0].p_hdr, path[1].p_hdr, s);
1779         path[0].p_depth = 0;
1780         path[0].p_ext = EXT_FIRST_EXTENT(path[0].p_hdr) +
1781                 (path[1].p_ext - EXT_FIRST_EXTENT(path[1].p_hdr));
1782         path[0].p_hdr->eh_max = cpu_to_le16(max_root);
1783
1784         brelse(path[1].p_bh);
1785         ext4_free_blocks(handle, inode, NULL, blk, 1,
1786                          EXT4_FREE_BLOCKS_METADATA | EXT4_FREE_BLOCKS_FORGET |
1787                          EXT4_FREE_BLOCKS_RESERVE);
1788 }
1789
1790 /*
1791  * This function tries to merge the @ex extent to neighbours in the tree.
1792  * return 1 if merge left else 0.
1793  */
1794 static void ext4_ext_try_to_merge(handle_t *handle,
1795                                   struct inode *inode,
1796                                   struct ext4_ext_path *path,
1797                                   struct ext4_extent *ex) {
1798         struct ext4_extent_header *eh;
1799         unsigned int depth;
1800         int merge_done = 0;
1801
1802         depth = ext_depth(inode);
1803         BUG_ON(path[depth].p_hdr == NULL);
1804         eh = path[depth].p_hdr;
1805
1806         if (ex > EXT_FIRST_EXTENT(eh))
1807                 merge_done = ext4_ext_try_to_merge_right(inode, path, ex - 1);
1808
1809         if (!merge_done)
1810                 (void) ext4_ext_try_to_merge_right(inode, path, ex);
1811
1812         ext4_ext_try_to_merge_up(handle, inode, path);
1813 }
1814
1815 /*
1816  * check if a portion of the "newext" extent overlaps with an
1817  * existing extent.
1818  *
1819  * If there is an overlap discovered, it updates the length of the newext
1820  * such that there will be no overlap, and then returns 1.
1821  * If there is no overlap found, it returns 0.
1822  */
1823 static unsigned int ext4_ext_check_overlap(struct ext4_sb_info *sbi,
1824                                            struct inode *inode,
1825                                            struct ext4_extent *newext,
1826                                            struct ext4_ext_path *path)
1827 {
1828         ext4_lblk_t b1, b2;
1829         unsigned int depth, len1;
1830         unsigned int ret = 0;
1831
1832         b1 = le32_to_cpu(newext->ee_block);
1833         len1 = ext4_ext_get_actual_len(newext);
1834         depth = ext_depth(inode);
1835         if (!path[depth].p_ext)
1836                 goto out;
1837         b2 = le32_to_cpu(path[depth].p_ext->ee_block);
1838         b2 &= ~(sbi->s_cluster_ratio - 1);
1839
1840         /*
1841          * get the next allocated block if the extent in the path
1842          * is before the requested block(s)
1843          */
1844         if (b2 < b1) {
1845                 b2 = ext4_ext_next_allocated_block(path);
1846                 if (b2 == EXT_MAX_BLOCKS)
1847                         goto out;
1848                 b2 &= ~(sbi->s_cluster_ratio - 1);
1849         }
1850
1851         /* check for wrap through zero on extent logical start block*/
1852         if (b1 + len1 < b1) {
1853                 len1 = EXT_MAX_BLOCKS - b1;
1854                 newext->ee_len = cpu_to_le16(len1);
1855                 ret = 1;
1856         }
1857
1858         /* check for overlap */
1859         if (b1 + len1 > b2) {
1860                 newext->ee_len = cpu_to_le16(b2 - b1);
1861                 ret = 1;
1862         }
1863 out:
1864         return ret;
1865 }
1866
1867 /*
1868  * ext4_ext_insert_extent:
1869  * tries to merge requsted extent into the existing extent or
1870  * inserts requested extent as new one into the tree,
1871  * creating new leaf in the no-space case.
1872  */
1873 int ext4_ext_insert_extent(handle_t *handle, struct inode *inode,
1874                                 struct ext4_ext_path *path,
1875                                 struct ext4_extent *newext, int gb_flags)
1876 {
1877         struct ext4_extent_header *eh;
1878         struct ext4_extent *ex, *fex;
1879         struct ext4_extent *nearex; /* nearest extent */
1880         struct ext4_ext_path *npath = NULL;
1881         int depth, len, err;
1882         ext4_lblk_t next;
1883         int mb_flags = 0;
1884
1885         if (unlikely(ext4_ext_get_actual_len(newext) == 0)) {
1886                 EXT4_ERROR_INODE(inode, "ext4_ext_get_actual_len(newext) == 0");
1887                 return -EIO;
1888         }
1889         depth = ext_depth(inode);
1890         ex = path[depth].p_ext;
1891         eh = path[depth].p_hdr;
1892         if (unlikely(path[depth].p_hdr == NULL)) {
1893                 EXT4_ERROR_INODE(inode, "path[%d].p_hdr == NULL", depth);
1894                 return -EIO;
1895         }
1896
1897         /* try to insert block into found extent and return */
1898         if (ex && !(gb_flags & EXT4_GET_BLOCKS_PRE_IO)) {
1899
1900                 /*
1901                  * Try to see whether we should rather test the extent on
1902                  * right from ex, or from the left of ex. This is because
1903                  * ext4_ext_find_extent() can return either extent on the
1904                  * left, or on the right from the searched position. This
1905                  * will make merging more effective.
1906                  */
1907                 if (ex < EXT_LAST_EXTENT(eh) &&
1908                     (le32_to_cpu(ex->ee_block) +
1909                     ext4_ext_get_actual_len(ex) <
1910                     le32_to_cpu(newext->ee_block))) {
1911                         ex += 1;
1912                         goto prepend;
1913                 } else if ((ex > EXT_FIRST_EXTENT(eh)) &&
1914                            (le32_to_cpu(newext->ee_block) +
1915                            ext4_ext_get_actual_len(newext) <
1916                            le32_to_cpu(ex->ee_block)))
1917                         ex -= 1;
1918
1919                 /* Try to append newex to the ex */
1920                 if (ext4_can_extents_be_merged(inode, ex, newext)) {
1921                         ext_debug("append [%d]%d block to %u:[%d]%d"
1922                                   "(from %llu)\n",
1923                                   ext4_ext_is_uninitialized(newext),
1924                                   ext4_ext_get_actual_len(newext),
1925                                   le32_to_cpu(ex->ee_block),
1926                                   ext4_ext_is_uninitialized(ex),
1927                                   ext4_ext_get_actual_len(ex),
1928                                   ext4_ext_pblock(ex));
1929                         err = ext4_ext_get_access(handle, inode,
1930                                                   path + depth);
1931                         if (err)
1932                                 return err;
1933
1934                         ex->ee_len = cpu_to_le16(ext4_ext_get_actual_len(ex)
1935                                         + ext4_ext_get_actual_len(newext));
1936                         eh = path[depth].p_hdr;
1937                         nearex = ex;
1938                         goto merge;
1939                 }
1940
1941 prepend:
1942                 /* Try to prepend newex to the ex */
1943                 if (ext4_can_extents_be_merged(inode, newext, ex)) {
1944                         ext_debug("prepend %u[%d]%d block to %u:[%d]%d"
1945                                   "(from %llu)\n",
1946                                   le32_to_cpu(newext->ee_block),
1947                                   ext4_ext_is_uninitialized(newext),
1948                                   ext4_ext_get_actual_len(newext),
1949                                   le32_to_cpu(ex->ee_block),
1950                                   ext4_ext_is_uninitialized(ex),
1951                                   ext4_ext_get_actual_len(ex),
1952                                   ext4_ext_pblock(ex));
1953                         err = ext4_ext_get_access(handle, inode,
1954                                                   path + depth);
1955                         if (err)
1956                                 return err;
1957
1958                         ex->ee_block = newext->ee_block;
1959                         ext4_ext_store_pblock(ex, ext4_ext_pblock(newext));
1960                         ex->ee_len = cpu_to_le16(ext4_ext_get_actual_len(ex)
1961                                         + ext4_ext_get_actual_len(newext));
1962                         eh = path[depth].p_hdr;
1963                         nearex = ex;
1964                         goto merge;
1965                 }
1966         }
1967
1968         depth = ext_depth(inode);
1969         eh = path[depth].p_hdr;
1970         if (le16_to_cpu(eh->eh_entries) < le16_to_cpu(eh->eh_max))
1971                 goto has_space;
1972
1973         /* probably next leaf has space for us? */
1974         fex = EXT_LAST_EXTENT(eh);
1975         next = EXT_MAX_BLOCKS;
1976         if (le32_to_cpu(newext->ee_block) > le32_to_cpu(fex->ee_block))
1977                 next = ext4_ext_next_leaf_block(path);
1978         if (next != EXT_MAX_BLOCKS) {
1979                 ext_debug("next leaf block - %u\n", next);
1980                 BUG_ON(npath != NULL);
1981                 npath = ext4_ext_find_extent(inode, next, NULL, 0);
1982                 if (IS_ERR(npath))
1983                         return PTR_ERR(npath);
1984                 BUG_ON(npath->p_depth != path->p_depth);
1985                 eh = npath[depth].p_hdr;
1986                 if (le16_to_cpu(eh->eh_entries) < le16_to_cpu(eh->eh_max)) {
1987                         ext_debug("next leaf isn't full(%d)\n",
1988                                   le16_to_cpu(eh->eh_entries));
1989                         path = npath;
1990                         goto has_space;
1991                 }
1992                 ext_debug("next leaf has no free space(%d,%d)\n",
1993                           le16_to_cpu(eh->eh_entries), le16_to_cpu(eh->eh_max));
1994         }
1995
1996         /*
1997          * There is no free space in the found leaf.
1998          * We're gonna add a new leaf in the tree.
1999          */
2000         if (gb_flags & EXT4_GET_BLOCKS_METADATA_NOFAIL)
2001                 mb_flags = EXT4_MB_USE_RESERVED;
2002         err = ext4_ext_create_new_leaf(handle, inode, mb_flags, gb_flags,
2003                                        path, newext);
2004         if (err)
2005                 goto cleanup;
2006         depth = ext_depth(inode);
2007         eh = path[depth].p_hdr;
2008
2009 has_space:
2010         nearex = path[depth].p_ext;
2011
2012         err = ext4_ext_get_access(handle, inode, path + depth);
2013         if (err)
2014                 goto cleanup;
2015
2016         if (!nearex) {
2017                 /* there is no extent in this leaf, create first one */
2018                 ext_debug("first extent in the leaf: %u:%llu:[%d]%d\n",
2019                                 le32_to_cpu(newext->ee_block),
2020                                 ext4_ext_pblock(newext),
2021                                 ext4_ext_is_uninitialized(newext),
2022                                 ext4_ext_get_actual_len(newext));
2023                 nearex = EXT_FIRST_EXTENT(eh);
2024         } else {
2025                 if (le32_to_cpu(newext->ee_block)
2026                            > le32_to_cpu(nearex->ee_block)) {
2027                         /* Insert after */
2028                         ext_debug("insert %u:%llu:[%d]%d before: "
2029                                         "nearest %p\n",
2030                                         le32_to_cpu(newext->ee_block),
2031                                         ext4_ext_pblock(newext),
2032                                         ext4_ext_is_uninitialized(newext),
2033                                         ext4_ext_get_actual_len(newext),
2034                                         nearex);
2035                         nearex++;
2036                 } else {
2037                         /* Insert before */
2038                         BUG_ON(newext->ee_block == nearex->ee_block);
2039                         ext_debug("insert %u:%llu:[%d]%d after: "
2040                                         "nearest %p\n",
2041                                         le32_to_cpu(newext->ee_block),
2042                                         ext4_ext_pblock(newext),
2043                                         ext4_ext_is_uninitialized(newext),
2044                                         ext4_ext_get_actual_len(newext),
2045                                         nearex);
2046                 }
2047                 len = EXT_LAST_EXTENT(eh) - nearex + 1;
2048                 if (len > 0) {
2049                         ext_debug("insert %u:%llu:[%d]%d: "
2050                                         "move %d extents from 0x%p to 0x%p\n",
2051                                         le32_to_cpu(newext->ee_block),
2052                                         ext4_ext_pblock(newext),
2053                                         ext4_ext_is_uninitialized(newext),
2054                                         ext4_ext_get_actual_len(newext),
2055                                         len, nearex, nearex + 1);
2056                         memmove(nearex + 1, nearex,
2057                                 len * sizeof(struct ext4_extent));
2058                 }
2059         }
2060
2061         le16_add_cpu(&eh->eh_entries, 1);
2062         path[depth].p_ext = nearex;
2063         nearex->ee_block = newext->ee_block;
2064         ext4_ext_store_pblock(nearex, ext4_ext_pblock(newext));
2065         nearex->ee_len = newext->ee_len;
2066
2067 merge:
2068         /* try to merge extents */
2069         if (!(gb_flags & EXT4_GET_BLOCKS_PRE_IO))
2070                 ext4_ext_try_to_merge(handle, inode, path, nearex);
2071
2072
2073         /* time to correct all indexes above */
2074         err = ext4_ext_correct_indexes(handle, inode, path);
2075         if (err)
2076                 goto cleanup;
2077
2078         err = ext4_ext_dirty(handle, inode, path + path->p_depth);
2079
2080 cleanup:
2081         if (npath) {
2082                 ext4_ext_drop_refs(npath);
2083                 kfree(npath);
2084         }
2085         return err;
2086 }
2087
2088 static int ext4_fill_fiemap_extents(struct inode *inode,
2089                                     ext4_lblk_t block, ext4_lblk_t num,
2090                                     struct fiemap_extent_info *fieinfo)
2091 {
2092         struct ext4_ext_path *path = NULL;
2093         struct ext4_extent *ex;
2094         struct extent_status es;
2095         ext4_lblk_t next, next_del, start = 0, end = 0;
2096         ext4_lblk_t last = block + num;
2097         int exists, depth = 0, err = 0;
2098         unsigned int flags = 0;
2099         unsigned char blksize_bits = inode->i_sb->s_blocksize_bits;
2100
2101         while (block < last && block != EXT_MAX_BLOCKS) {
2102                 num = last - block;
2103                 /* find extent for this block */
2104                 down_read(&EXT4_I(inode)->i_data_sem);
2105
2106                 if (path && ext_depth(inode) != depth) {
2107                         /* depth was changed. we have to realloc path */
2108                         kfree(path);
2109                         path = NULL;
2110                 }
2111
2112                 path = ext4_ext_find_extent(inode, block, path, 0);
2113                 if (IS_ERR(path)) {
2114                         up_read(&EXT4_I(inode)->i_data_sem);
2115                         err = PTR_ERR(path);
2116                         path = NULL;
2117                         break;
2118                 }
2119
2120                 depth = ext_depth(inode);
2121                 if (unlikely(path[depth].p_hdr == NULL)) {
2122                         up_read(&EXT4_I(inode)->i_data_sem);
2123                         EXT4_ERROR_INODE(inode, "path[%d].p_hdr == NULL", depth);
2124                         err = -EIO;
2125                         break;
2126                 }
2127                 ex = path[depth].p_ext;
2128                 next = ext4_ext_next_allocated_block(path);
2129                 ext4_ext_drop_refs(path);
2130
2131                 flags = 0;
2132                 exists = 0;
2133                 if (!ex) {
2134                         /* there is no extent yet, so try to allocate
2135                          * all requested space */
2136                         start = block;
2137                         end = block + num;
2138                 } else if (le32_to_cpu(ex->ee_block) > block) {
2139                         /* need to allocate space before found extent */
2140                         start = block;
2141                         end = le32_to_cpu(ex->ee_block);
2142                         if (block + num < end)
2143                                 end = block + num;
2144                 } else if (block >= le32_to_cpu(ex->ee_block)
2145                                         + ext4_ext_get_actual_len(ex)) {
2146                         /* need to allocate space after found extent */
2147                         start = block;
2148                         end = block + num;
2149                         if (end >= next)
2150                                 end = next;
2151                 } else if (block >= le32_to_cpu(ex->ee_block)) {
2152                         /*
2153                          * some part of requested space is covered
2154                          * by found extent
2155                          */
2156                         start = block;
2157                         end = le32_to_cpu(ex->ee_block)
2158                                 + ext4_ext_get_actual_len(ex);
2159                         if (block + num < end)
2160                                 end = block + num;
2161                         exists = 1;
2162                 } else {
2163                         BUG();
2164                 }
2165                 BUG_ON(end <= start);
2166
2167                 if (!exists) {
2168                         es.es_lblk = start;
2169                         es.es_len = end - start;
2170                         es.es_pblk = 0;
2171                 } else {
2172                         es.es_lblk = le32_to_cpu(ex->ee_block);
2173                         es.es_len = ext4_ext_get_actual_len(ex);
2174                         es.es_pblk = ext4_ext_pblock(ex);
2175                         if (ext4_ext_is_uninitialized(ex))
2176                                 flags |= FIEMAP_EXTENT_UNWRITTEN;
2177                 }
2178
2179                 /*
2180                  * Find delayed extent and update es accordingly. We call
2181                  * it even in !exists case to find out whether es is the
2182                  * last existing extent or not.
2183                  */
2184                 next_del = ext4_find_delayed_extent(inode, &es);
2185                 if (!exists && next_del) {
2186                         exists = 1;
2187                         flags |= (FIEMAP_EXTENT_DELALLOC |
2188                                   FIEMAP_EXTENT_UNKNOWN);
2189                 }
2190                 up_read(&EXT4_I(inode)->i_data_sem);
2191
2192                 if (unlikely(es.es_len == 0)) {
2193                         EXT4_ERROR_INODE(inode, "es.es_len == 0");
2194                         err = -EIO;
2195                         break;
2196                 }
2197
2198                 /*
2199                  * This is possible iff next == next_del == EXT_MAX_BLOCKS.
2200                  * we need to check next == EXT_MAX_BLOCKS because it is
2201                  * possible that an extent is with unwritten and delayed
2202                  * status due to when an extent is delayed allocated and
2203                  * is allocated by fallocate status tree will track both of
2204                  * them in a extent.
2205                  *
2206                  * So we could return a unwritten and delayed extent, and
2207                  * its block is equal to 'next'.
2208                  */
2209                 if (next == next_del && next == EXT_MAX_BLOCKS) {
2210                         flags |= FIEMAP_EXTENT_LAST;
2211                         if (unlikely(next_del != EXT_MAX_BLOCKS ||
2212                                      next != EXT_MAX_BLOCKS)) {
2213                                 EXT4_ERROR_INODE(inode,
2214                                                  "next extent == %u, next "
2215                                                  "delalloc extent = %u",
2216                                                  next, next_del);
2217                                 err = -EIO;
2218                                 break;
2219                         }
2220                 }
2221
2222                 if (exists) {
2223                         err = fiemap_fill_next_extent(fieinfo,
2224                                 (__u64)es.es_lblk << blksize_bits,
2225                                 (__u64)es.es_pblk << blksize_bits,
2226                                 (__u64)es.es_len << blksize_bits,
2227                                 flags);
2228                         if (err < 0)
2229                                 break;
2230                         if (err == 1) {
2231                                 err = 0;
2232                                 break;
2233                         }
2234                 }
2235
2236                 block = es.es_lblk + es.es_len;
2237         }
2238
2239         if (path) {
2240                 ext4_ext_drop_refs(path);
2241                 kfree(path);
2242         }
2243
2244         return err;
2245 }
2246
2247 /*
2248  * ext4_ext_put_gap_in_cache:
2249  * calculate boundaries of the gap that the requested block fits into
2250  * and cache this gap
2251  */
2252 static void
2253 ext4_ext_put_gap_in_cache(struct inode *inode, struct ext4_ext_path *path,
2254                                 ext4_lblk_t block)
2255 {
2256         int depth = ext_depth(inode);
2257         unsigned long len = 0;
2258         ext4_lblk_t lblock = 0;
2259         struct ext4_extent *ex;
2260
2261         ex = path[depth].p_ext;
2262         if (ex == NULL) {
2263                 /*
2264                  * there is no extent yet, so gap is [0;-] and we
2265                  * don't cache it
2266                  */
2267                 ext_debug("cache gap(whole file):");
2268         } else if (block < le32_to_cpu(ex->ee_block)) {
2269                 lblock = block;
2270                 len = le32_to_cpu(ex->ee_block) - block;
2271                 ext_debug("cache gap(before): %u [%u:%u]",
2272                                 block,
2273                                 le32_to_cpu(ex->ee_block),
2274                                  ext4_ext_get_actual_len(ex));
2275                 if (!ext4_find_delalloc_range(inode, lblock, lblock + len - 1))
2276                         ext4_es_insert_extent(inode, lblock, len, ~0,
2277                                               EXTENT_STATUS_HOLE);
2278         } else if (block >= le32_to_cpu(ex->ee_block)
2279                         + ext4_ext_get_actual_len(ex)) {
2280                 ext4_lblk_t next;
2281                 lblock = le32_to_cpu(ex->ee_block)
2282                         + ext4_ext_get_actual_len(ex);
2283
2284                 next = ext4_ext_next_allocated_block(path);
2285                 ext_debug("cache gap(after): [%u:%u] %u",
2286                                 le32_to_cpu(ex->ee_block),
2287                                 ext4_ext_get_actual_len(ex),
2288                                 block);
2289                 BUG_ON(next == lblock);
2290                 len = next - lblock;
2291                 if (!ext4_find_delalloc_range(inode, lblock, lblock + len - 1))
2292                         ext4_es_insert_extent(inode, lblock, len, ~0,
2293                                               EXTENT_STATUS_HOLE);
2294         } else {
2295                 BUG();
2296         }
2297
2298         ext_debug(" -> %u:%lu\n", lblock, len);
2299 }
2300
2301 /*
2302  * ext4_ext_rm_idx:
2303  * removes index from the index block.
2304  */
2305 static int ext4_ext_rm_idx(handle_t *handle, struct inode *inode,
2306                         struct ext4_ext_path *path, int depth)
2307 {
2308         int err;
2309         ext4_fsblk_t leaf;
2310
2311         /* free index block */
2312         depth--;
2313         path = path + depth;
2314         leaf = ext4_idx_pblock(path->p_idx);
2315         if (unlikely(path->p_hdr->eh_entries == 0)) {
2316                 EXT4_ERROR_INODE(inode, "path->p_hdr->eh_entries == 0");
2317                 return -EIO;
2318         }
2319         err = ext4_ext_get_access(handle, inode, path);
2320         if (err)
2321                 return err;
2322
2323         if (path->p_idx != EXT_LAST_INDEX(path->p_hdr)) {
2324                 int len = EXT_LAST_INDEX(path->p_hdr) - path->p_idx;
2325                 len *= sizeof(struct ext4_extent_idx);
2326                 memmove(path->p_idx, path->p_idx + 1, len);
2327         }
2328
2329         le16_add_cpu(&path->p_hdr->eh_entries, -1);
2330         err = ext4_ext_dirty(handle, inode, path);
2331         if (err)
2332                 return err;
2333         ext_debug("index is empty, remove it, free block %llu\n", leaf);
2334         trace_ext4_ext_rm_idx(inode, leaf);
2335
2336         ext4_free_blocks(handle, inode, NULL, leaf, 1,
2337                          EXT4_FREE_BLOCKS_METADATA | EXT4_FREE_BLOCKS_FORGET);
2338
2339         while (--depth >= 0) {
2340                 if (path->p_idx != EXT_FIRST_INDEX(path->p_hdr))
2341                         break;
2342                 path--;
2343                 err = ext4_ext_get_access(handle, inode, path);
2344                 if (err)
2345                         break;
2346                 path->p_idx->ei_block = (path+1)->p_idx->ei_block;
2347                 err = ext4_ext_dirty(handle, inode, path);
2348                 if (err)
2349                         break;
2350         }
2351         return err;
2352 }
2353
2354 /*
2355  * ext4_ext_calc_credits_for_single_extent:
2356  * This routine returns max. credits that needed to insert an extent
2357  * to the extent tree.
2358  * When pass the actual path, the caller should calculate credits
2359  * under i_data_sem.
2360  */
2361 int ext4_ext_calc_credits_for_single_extent(struct inode *inode, int nrblocks,
2362                                                 struct ext4_ext_path *path)
2363 {
2364         if (path) {
2365                 int depth = ext_depth(inode);
2366                 int ret = 0;
2367
2368                 /* probably there is space in leaf? */
2369                 if (le16_to_cpu(path[depth].p_hdr->eh_entries)
2370                                 < le16_to_cpu(path[depth].p_hdr->eh_max)) {
2371
2372                         /*
2373                          *  There are some space in the leaf tree, no
2374                          *  need to account for leaf block credit
2375                          *
2376                          *  bitmaps and block group descriptor blocks
2377                          *  and other metadata blocks still need to be
2378                          *  accounted.
2379                          */
2380                         /* 1 bitmap, 1 block group descriptor */
2381                         ret = 2 + EXT4_META_TRANS_BLOCKS(inode->i_sb);
2382                         return ret;
2383                 }
2384         }
2385
2386         return ext4_chunk_trans_blocks(inode, nrblocks);
2387 }
2388
2389 /*
2390  * How many index/leaf blocks need to change/allocate to add @extents extents?
2391  *
2392  * If we add a single extent, then in the worse case, each tree level
2393  * index/leaf need to be changed in case of the tree split.
2394  *
2395  * If more extents are inserted, they could cause the whole tree split more
2396  * than once, but this is really rare.
2397  */
2398 int ext4_ext_index_trans_blocks(struct inode *inode, int extents)
2399 {
2400         int index;
2401         int depth;
2402
2403         /* If we are converting the inline data, only one is needed here. */
2404         if (ext4_has_inline_data(inode))
2405                 return 1;
2406
2407         depth = ext_depth(inode);
2408
2409         if (extents <= 1)
2410                 index = depth * 2;
2411         else
2412                 index = depth * 3;
2413
2414         return index;
2415 }
2416
2417 static inline int get_default_free_blocks_flags(struct inode *inode)
2418 {
2419         if (S_ISDIR(inode->i_mode) || S_ISLNK(inode->i_mode))
2420                 return EXT4_FREE_BLOCKS_METADATA | EXT4_FREE_BLOCKS_FORGET;
2421         else if (ext4_should_journal_data(inode))
2422                 return EXT4_FREE_BLOCKS_FORGET;
2423         return 0;
2424 }
2425
2426 static int ext4_remove_blocks(handle_t *handle, struct inode *inode,
2427                               struct ext4_extent *ex,
2428                               long long *partial_cluster,
2429                               ext4_lblk_t from, ext4_lblk_t to)
2430 {
2431         struct ext4_sb_info *sbi = EXT4_SB(inode->i_sb);
2432         unsigned short ee_len =  ext4_ext_get_actual_len(ex);
2433         ext4_fsblk_t pblk;
2434         int flags = get_default_free_blocks_flags(inode);
2435
2436         /*
2437          * For bigalloc file systems, we never free a partial cluster
2438          * at the beginning of the extent.  Instead, we make a note
2439          * that we tried freeing the cluster, and check to see if we
2440          * need to free it on a subsequent call to ext4_remove_blocks,
2441          * or at the end of the ext4_truncate() operation.
2442          */
2443         flags |= EXT4_FREE_BLOCKS_NOFREE_FIRST_CLUSTER;
2444
2445         trace_ext4_remove_blocks(inode, ex, from, to, *partial_cluster);
2446         /*
2447          * If we have a partial cluster, and it's different from the
2448          * cluster of the last block, we need to explicitly free the
2449          * partial cluster here.
2450          */
2451         pblk = ext4_ext_pblock(ex) + ee_len - 1;
2452         if ((*partial_cluster > 0) &&
2453             (EXT4_B2C(sbi, pblk) != *partial_cluster)) {
2454                 ext4_free_blocks(handle, inode, NULL,
2455                                  EXT4_C2B(sbi, *partial_cluster),
2456                                  sbi->s_cluster_ratio, flags);
2457                 *partial_cluster = 0;
2458         }
2459
2460 #ifdef EXTENTS_STATS
2461         {
2462                 struct ext4_sb_info *sbi = EXT4_SB(inode->i_sb);
2463                 spin_lock(&sbi->s_ext_stats_lock);
2464                 sbi->s_ext_blocks += ee_len;
2465                 sbi->s_ext_extents++;
2466                 if (ee_len < sbi->s_ext_min)
2467                         sbi->s_ext_min = ee_len;
2468                 if (ee_len > sbi->s_ext_max)
2469                         sbi->s_ext_max = ee_len;
2470                 if (ext_depth(inode) > sbi->s_depth_max)
2471                         sbi->s_depth_max = ext_depth(inode);
2472                 spin_unlock(&sbi->s_ext_stats_lock);
2473         }
2474 #endif
2475         if (from >= le32_to_cpu(ex->ee_block)
2476             && to == le32_to_cpu(ex->ee_block) + ee_len - 1) {
2477                 /* tail removal */
2478                 ext4_lblk_t num;
2479                 unsigned int unaligned;
2480
2481                 num = le32_to_cpu(ex->ee_block) + ee_len - from;
2482                 pblk = ext4_ext_pblock(ex) + ee_len - num;
2483                 /*
2484                  * Usually we want to free partial cluster at the end of the
2485                  * extent, except for the situation when the cluster is still
2486                  * used by any other extent (partial_cluster is negative).
2487                  */
2488                 if (*partial_cluster < 0 &&
2489                     -(*partial_cluster) == EXT4_B2C(sbi, pblk + num - 1))
2490                         flags |= EXT4_FREE_BLOCKS_NOFREE_LAST_CLUSTER;
2491
2492                 ext_debug("free last %u blocks starting %llu partial %lld\n",
2493                           num, pblk, *partial_cluster);
2494                 ext4_free_blocks(handle, inode, NULL, pblk, num, flags);
2495                 /*
2496                  * If the block range to be freed didn't start at the
2497                  * beginning of a cluster, and we removed the entire
2498                  * extent and the cluster is not used by any other extent,
2499                  * save the partial cluster here, since we might need to
2500                  * delete if we determine that the truncate operation has
2501                  * removed all of the blocks in the cluster.
2502                  *
2503                  * On the other hand, if we did not manage to free the whole
2504                  * extent, we have to mark the cluster as used (store negative
2505                  * cluster number in partial_cluster).
2506                  */
2507                 unaligned = pblk & (sbi->s_cluster_ratio - 1);
2508                 if (unaligned && (ee_len == num) &&
2509                     (*partial_cluster != -((long long)EXT4_B2C(sbi, pblk))))
2510                         *partial_cluster = EXT4_B2C(sbi, pblk);
2511                 else if (unaligned)
2512                         *partial_cluster = -((long long)EXT4_B2C(sbi, pblk));
2513                 else if (*partial_cluster > 0)
2514                         *partial_cluster = 0;
2515         } else
2516                 ext4_error(sbi->s_sb, "strange request: removal(2) "
2517                            "%u-%u from %u:%u\n",
2518                            from, to, le32_to_cpu(ex->ee_block), ee_len);
2519         return 0;
2520 }
2521
2522
2523 /*
2524  * ext4_ext_rm_leaf() Removes the extents associated with the
2525  * blocks appearing between "start" and "end", and splits the extents
2526  * if "start" and "end" appear in the same extent
2527  *
2528  * @handle: The journal handle
2529  * @inode:  The files inode
2530  * @path:   The path to the leaf
2531  * @partial_cluster: The cluster which we'll have to free if all extents
2532  *                   has been released from it. It gets negative in case
2533  *                   that the cluster is still used.
2534  * @start:  The first block to remove
2535  * @end:   The last block to remove
2536  */
2537 static int
2538 ext4_ext_rm_leaf(handle_t *handle, struct inode *inode,
2539                  struct ext4_ext_path *path,
2540                  long long *partial_cluster,
2541                  ext4_lblk_t start, ext4_lblk_t end)
2542 {
2543         struct ext4_sb_info *sbi = EXT4_SB(inode->i_sb);
2544         int err = 0, correct_index = 0;
2545         int depth = ext_depth(inode), credits;
2546         struct ext4_extent_header *eh;
2547         ext4_lblk_t a, b;
2548         unsigned num;
2549         ext4_lblk_t ex_ee_block;
2550         unsigned short ex_ee_len;
2551         unsigned uninitialized = 0;
2552         struct ext4_extent *ex;
2553         ext4_fsblk_t pblk;
2554
2555         /* the header must be checked already in ext4_ext_remove_space() */
2556         ext_debug("truncate since %u in leaf to %u\n", start, end);
2557         if (!path[depth].p_hdr)
2558                 path[depth].p_hdr = ext_block_hdr(path[depth].p_bh);
2559         eh = path[depth].p_hdr;
2560         if (unlikely(path[depth].p_hdr == NULL)) {
2561                 EXT4_ERROR_INODE(inode, "path[%d].p_hdr == NULL", depth);
2562                 return -EIO;
2563         }
2564         /* find where to start removing */
2565         ex = path[depth].p_ext;
2566         if (!ex)
2567                 ex = EXT_LAST_EXTENT(eh);
2568
2569         ex_ee_block = le32_to_cpu(ex->ee_block);
2570         ex_ee_len = ext4_ext_get_actual_len(ex);
2571
2572         trace_ext4_ext_rm_leaf(inode, start, ex, *partial_cluster);
2573
2574         while (ex >= EXT_FIRST_EXTENT(eh) &&
2575                         ex_ee_block + ex_ee_len > start) {
2576
2577                 if (ext4_ext_is_uninitialized(ex))
2578                         uninitialized = 1;
2579                 else
2580                         uninitialized = 0;
2581
2582                 ext_debug("remove ext %u:[%d]%d\n", ex_ee_block,
2583                          uninitialized, ex_ee_len);
2584                 path[depth].p_ext = ex;
2585
2586                 a = ex_ee_block > start ? ex_ee_block : start;
2587                 b = ex_ee_block+ex_ee_len - 1 < end ?
2588                         ex_ee_block+ex_ee_len - 1 : end;
2589
2590                 ext_debug("  border %u:%u\n", a, b);
2591
2592                 /* If this extent is beyond the end of the hole, skip it */
2593                 if (end < ex_ee_block) {
2594                         /*
2595                          * We're going to skip this extent and move to another,
2596                          * so if this extent is not cluster aligned we have
2597                          * to mark the current cluster as used to avoid
2598                          * accidentally freeing it later on
2599                          */
2600                         pblk = ext4_ext_pblock(ex);
2601                         if (pblk & (sbi->s_cluster_ratio - 1))
2602                                 *partial_cluster =
2603                                         -((long long)EXT4_B2C(sbi, pblk));
2604                         ex--;
2605                         ex_ee_block = le32_to_cpu(ex->ee_block);
2606                         ex_ee_len = ext4_ext_get_actual_len(ex);
2607                         continue;
2608                 } else if (b != ex_ee_block + ex_ee_len - 1) {
2609                         EXT4_ERROR_INODE(inode,
2610                                          "can not handle truncate %u:%u "
2611                                          "on extent %u:%u",
2612                                          start, end, ex_ee_block,
2613                                          ex_ee_block + ex_ee_len - 1);
2614                         err = -EIO;
2615                         goto out;
2616                 } else if (a != ex_ee_block) {
2617                         /* remove tail of the extent */
2618                         num = a - ex_ee_block;
2619                 } else {
2620                         /* remove whole extent: excellent! */
2621                         num = 0;
2622                 }
2623                 /*
2624                  * 3 for leaf, sb, and inode plus 2 (bmap and group
2625                  * descriptor) for each block group; assume two block
2626                  * groups plus ex_ee_len/blocks_per_block_group for
2627                  * the worst case
2628                  */
2629                 credits = 7 + 2*(ex_ee_len/EXT4_BLOCKS_PER_GROUP(inode->i_sb));
2630                 if (ex == EXT_FIRST_EXTENT(eh)) {
2631                         correct_index = 1;
2632                         credits += (ext_depth(inode)) + 1;
2633                 }
2634                 credits += EXT4_MAXQUOTAS_TRANS_BLOCKS(inode->i_sb);
2635
2636                 err = ext4_ext_truncate_extend_restart(handle, inode, credits);
2637                 if (err)
2638                         goto out;
2639
2640                 err = ext4_ext_get_access(handle, inode, path + depth);
2641                 if (err)
2642                         goto out;
2643
2644                 err = ext4_remove_blocks(handle, inode, ex, partial_cluster,
2645                                          a, b);
2646                 if (err)
2647                         goto out;
2648
2649                 if (num == 0)
2650                         /* this extent is removed; mark slot entirely unused */
2651                         ext4_ext_store_pblock(ex, 0);
2652
2653                 ex->ee_len = cpu_to_le16(num);
2654                 /*
2655                  * Do not mark uninitialized if all the blocks in the
2656                  * extent have been removed.
2657                  */
2658                 if (uninitialized && num)
2659                         ext4_ext_mark_uninitialized(ex);
2660                 /*
2661                  * If the extent was completely released,
2662                  * we need to remove it from the leaf
2663                  */
2664                 if (num == 0) {
2665                         if (end != EXT_MAX_BLOCKS - 1) {
2666                                 /*
2667                                  * For hole punching, we need to scoot all the
2668                                  * extents up when an extent is removed so that
2669                                  * we dont have blank extents in the middle
2670                                  */
2671                                 memmove(ex, ex+1, (EXT_LAST_EXTENT(eh) - ex) *
2672                                         sizeof(struct ext4_extent));
2673
2674                                 /* Now get rid of the one at the end */
2675                                 memset(EXT_LAST_EXTENT(eh), 0,
2676                                         sizeof(struct ext4_extent));
2677                         }
2678                         le16_add_cpu(&eh->eh_entries, -1);
2679                 } else if (*partial_cluster > 0)
2680                         *partial_cluster = 0;
2681
2682                 err = ext4_ext_dirty(handle, inode, path + depth);
2683                 if (err)
2684                         goto out;
2685
2686                 ext_debug("new extent: %u:%u:%llu\n", ex_ee_block, num,
2687                                 ext4_ext_pblock(ex));
2688                 ex--;
2689                 ex_ee_block = le32_to_cpu(ex->ee_block);
2690                 ex_ee_len = ext4_ext_get_actual_len(ex);
2691         }
2692
2693         if (correct_index && eh->eh_entries)
2694                 err = ext4_ext_correct_indexes(handle, inode, path);
2695
2696         /*
2697          * Free the partial cluster only if the current extent does not
2698          * reference it. Otherwise we might free used cluster.
2699          */
2700         if (*partial_cluster > 0 &&
2701             (EXT4_B2C(sbi, ext4_ext_pblock(ex) + ex_ee_len - 1) !=
2702              *partial_cluster)) {
2703                 int flags = get_default_free_blocks_flags(inode);
2704
2705                 ext4_free_blocks(handle, inode, NULL,
2706                                  EXT4_C2B(sbi, *partial_cluster),
2707                                  sbi->s_cluster_ratio, flags);
2708                 *partial_cluster = 0;
2709         }
2710
2711         /* if this leaf is free, then we should
2712          * remove it from index block above */
2713         if (err == 0 && eh->eh_entries == 0 && path[depth].p_bh != NULL)
2714                 err = ext4_ext_rm_idx(handle, inode, path, depth);
2715
2716 out:
2717         return err;
2718 }
2719
2720 /*
2721  * ext4_ext_more_to_rm:
2722  * returns 1 if current index has to be freed (even partial)
2723  */
2724 static int
2725 ext4_ext_more_to_rm(struct ext4_ext_path *path)
2726 {
2727         BUG_ON(path->p_idx == NULL);
2728
2729         if (path->p_idx < EXT_FIRST_INDEX(path->p_hdr))
2730                 return 0;
2731
2732         /*
2733          * if truncate on deeper level happened, it wasn't partial,
2734          * so we have to consider current index for truncation
2735          */
2736         if (le16_to_cpu(path->p_hdr->eh_entries) == path->p_block)
2737                 return 0;
2738         return 1;
2739 }
2740
2741 int ext4_ext_remove_space(struct inode *inode, ext4_lblk_t start,
2742                           ext4_lblk_t end)
2743 {
2744         struct super_block *sb = inode->i_sb;
2745         int depth = ext_depth(inode);
2746         struct ext4_ext_path *path = NULL;
2747         long long partial_cluster = 0;
2748         handle_t *handle;
2749         int i = 0, err = 0;
2750
2751         ext_debug("truncate since %u to %u\n", start, end);
2752
2753         /* probably first extent we're gonna free will be last in block */
2754         handle = ext4_journal_start(inode, EXT4_HT_TRUNCATE, depth + 1);
2755         if (IS_ERR(handle))
2756                 return PTR_ERR(handle);
2757
2758 again:
2759         trace_ext4_ext_remove_space(inode, start, end, depth);
2760
2761         /*
2762          * Check if we are removing extents inside the extent tree. If that
2763          * is the case, we are going to punch a hole inside the extent tree
2764          * so we have to check whether we need to split the extent covering
2765          * the last block to remove so we can easily remove the part of it
2766          * in ext4_ext_rm_leaf().
2767          */
2768         if (end < EXT_MAX_BLOCKS - 1) {
2769                 struct ext4_extent *ex;
2770                 ext4_lblk_t ee_block;
2771
2772                 /* find extent for this block */
2773                 path = ext4_ext_find_extent(inode, end, NULL, EXT4_EX_NOCACHE);
2774                 if (IS_ERR(path)) {
2775                         ext4_journal_stop(handle);
2776                         return PTR_ERR(path);
2777                 }
2778                 depth = ext_depth(inode);
2779                 /* Leaf not may not exist only if inode has no blocks at all */
2780                 ex = path[depth].p_ext;
2781                 if (!ex) {
2782                         if (depth) {
2783                                 EXT4_ERROR_INODE(inode,
2784                                                  "path[%d].p_hdr == NULL",
2785                                                  depth);
2786                                 err = -EIO;
2787                         }
2788                         goto out;
2789                 }
2790
2791                 ee_block = le32_to_cpu(ex->ee_block);
2792
2793                 /*
2794                  * See if the last block is inside the extent, if so split
2795                  * the extent at 'end' block so we can easily remove the
2796                  * tail of the first part of the split extent in
2797                  * ext4_ext_rm_leaf().
2798                  */
2799                 if (end >= ee_block &&
2800                     end < ee_block + ext4_ext_get_actual_len(ex) - 1) {
2801                         int split_flag = 0;
2802
2803                         if (ext4_ext_is_uninitialized(ex))
2804                                 split_flag = EXT4_EXT_MARK_UNINIT1 |
2805                                              EXT4_EXT_MARK_UNINIT2;
2806
2807                         /*
2808                          * Split the extent in two so that 'end' is the last
2809                          * block in the first new extent. Also we should not
2810                          * fail removing space due to ENOSPC so try to use
2811                          * reserved block if that happens.
2812                          */
2813                         err = ext4_split_extent_at(handle, inode, path,
2814                                         end + 1, split_flag,
2815                                         EXT4_EX_NOCACHE |
2816                                         EXT4_GET_BLOCKS_PRE_IO |
2817                                         EXT4_GET_BLOCKS_METADATA_NOFAIL);
2818
2819                         if (err < 0)
2820                                 goto out;
2821                 }
2822         }
2823         /*
2824          * We start scanning from right side, freeing all the blocks
2825          * after i_size and walking into the tree depth-wise.
2826          */
2827         depth = ext_depth(inode);
2828         if (path) {
2829                 int k = i = depth;
2830                 while (--k > 0)
2831                         path[k].p_block =
2832                                 le16_to_cpu(path[k].p_hdr->eh_entries)+1;
2833         } else {
2834                 path = kzalloc(sizeof(struct ext4_ext_path) * (depth + 1),
2835                                GFP_NOFS);
2836                 if (path == NULL) {
2837                         ext4_journal_stop(handle);
2838                         return -ENOMEM;
2839                 }
2840                 path[0].p_depth = depth;
2841                 path[0].p_hdr = ext_inode_hdr(inode);
2842                 i = 0;
2843
2844                 if (ext4_ext_check(inode, path[0].p_hdr, depth, 0)) {
2845                         err = -EIO;
2846                         goto out;
2847                 }
2848         }
2849         err = 0;
2850
2851         while (i >= 0 && err == 0) {
2852                 if (i == depth) {
2853                         /* this is leaf block */
2854                         err = ext4_ext_rm_leaf(handle, inode, path,
2855                                                &partial_cluster, start,
2856                                                end);
2857                         /* root level has p_bh == NULL, brelse() eats this */
2858                         brelse(path[i].p_bh);
2859                         path[i].p_bh = NULL;
2860                         i--;
2861                         continue;
2862                 }
2863
2864                 /* this is index block */
2865                 if (!path[i].p_hdr) {
2866                         ext_debug("initialize header\n");
2867                         path[i].p_hdr = ext_block_hdr(path[i].p_bh);
2868                 }
2869
2870                 if (!path[i].p_idx) {
2871                         /* this level hasn't been touched yet */
2872                         path[i].p_idx = EXT_LAST_INDEX(path[i].p_hdr);
2873                         path[i].p_block = le16_to_cpu(path[i].p_hdr->eh_entries)+1;
2874                         ext_debug("init index ptr: hdr 0x%p, num %d\n",
2875                                   path[i].p_hdr,
2876                                   le16_to_cpu(path[i].p_hdr->eh_entries));
2877                 } else {
2878                         /* we were already here, see at next index */
2879                         path[i].p_idx--;
2880                 }
2881
2882                 ext_debug("level %d - index, first 0x%p, cur 0x%p\n",
2883                                 i, EXT_FIRST_INDEX(path[i].p_hdr),
2884                                 path[i].p_idx);
2885                 if (ext4_ext_more_to_rm(path + i)) {
2886                         struct buffer_head *bh;
2887                         /* go to the next level */
2888                         ext_debug("move to level %d (block %llu)\n",
2889                                   i + 1, ext4_idx_pblock(path[i].p_idx));
2890                         memset(path + i + 1, 0, sizeof(*path));
2891                         bh = read_extent_tree_block(inode,
2892                                 ext4_idx_pblock(path[i].p_idx), depth - i - 1,
2893                                 EXT4_EX_NOCACHE);
2894                         if (IS_ERR(bh)) {
2895                                 /* should we reset i_size? */
2896                                 err = PTR_ERR(bh);
2897                                 break;
2898                         }
2899                         /* Yield here to deal with large extent trees.
2900                          * Should be a no-op if we did IO above. */
2901                         cond_resched();
2902                         if (WARN_ON(i + 1 > depth)) {
2903                                 err = -EIO;
2904                                 break;
2905                         }
2906                         path[i + 1].p_bh = bh;
2907
2908                         /* save actual number of indexes since this
2909                          * number is changed at the next iteration */
2910                         path[i].p_block = le16_to_cpu(path[i].p_hdr->eh_entries);
2911                         i++;
2912                 } else {
2913                         /* we finished processing this index, go up */
2914                         if (path[i].p_hdr->eh_entries == 0 && i > 0) {
2915                                 /* index is empty, remove it;
2916                                  * handle must be already prepared by the
2917                                  * truncatei_leaf() */
2918                                 err = ext4_ext_rm_idx(handle, inode, path, i);
2919                         }
2920                         /* root level has p_bh == NULL, brelse() eats this */
2921                         brelse(path[i].p_bh);
2922                         path[i].p_bh = NULL;
2923                         i--;
2924                         ext_debug("return to level %d\n", i);
2925                 }
2926         }
2927
2928         trace_ext4_ext_remove_space_done(inode, start, end, depth,
2929                         partial_cluster, path->p_hdr->eh_entries);
2930
2931         /* If we still have something in the partial cluster and we have removed
2932          * even the first extent, then we should free the blocks in the partial
2933          * cluster as well. */
2934         if (partial_cluster > 0 && path->p_hdr->eh_entries == 0) {
2935                 int flags = get_default_free_blocks_flags(inode);
2936
2937                 ext4_free_blocks(handle, inode, NULL,
2938                                  EXT4_C2B(EXT4_SB(sb), partial_cluster),
2939                                  EXT4_SB(sb)->s_cluster_ratio, flags);
2940                 partial_cluster = 0;
2941         }
2942
2943         /* TODO: flexible tree reduction should be here */
2944         if (path->p_hdr->eh_entries == 0) {
2945                 /*
2946                  * truncate to zero freed all the tree,
2947                  * so we need to correct eh_depth
2948                  */
2949                 err = ext4_ext_get_access(handle, inode, path);
2950                 if (err == 0) {
2951                         ext_inode_hdr(inode)->eh_depth = 0;
2952                         ext_inode_hdr(inode)->eh_max =
2953                                 cpu_to_le16(ext4_ext_space_root(inode, 0));
2954                         err = ext4_ext_dirty(handle, inode, path);
2955                 }
2956         }
2957 out:
2958         ext4_ext_drop_refs(path);
2959         kfree(path);
2960         if (err == -EAGAIN) {
2961                 path = NULL;
2962                 goto again;
2963         }
2964         ext4_journal_stop(handle);
2965
2966         return err;
2967 }
2968
2969 /*
2970  * called at mount time
2971  */
2972 void ext4_ext_init(struct super_block *sb)
2973 {
2974         /*
2975          * possible initialization would be here
2976          */
2977
2978         if (EXT4_HAS_INCOMPAT_FEATURE(sb, EXT4_FEATURE_INCOMPAT_EXTENTS)) {
2979 #if defined(AGGRESSIVE_TEST) || defined(CHECK_BINSEARCH) || defined(EXTENTS_STATS)
2980                 printk(KERN_INFO "EXT4-fs: file extents enabled"
2981 #ifdef AGGRESSIVE_TEST
2982                        ", aggressive tests"
2983 #endif
2984 #ifdef CHECK_BINSEARCH
2985                        ", check binsearch"
2986 #endif
2987 #ifdef EXTENTS_STATS
2988                        ", stats"
2989 #endif
2990                        "\n");
2991 #endif
2992 #ifdef EXTENTS_STATS
2993                 spin_lock_init(&EXT4_SB(sb)->s_ext_stats_lock);
2994                 EXT4_SB(sb)->s_ext_min = 1 << 30;
2995                 EXT4_SB(sb)->s_ext_max = 0;
2996 #endif
2997         }
2998 }
2999
3000 /*
3001  * called at umount time
3002  */
3003 void ext4_ext_release(struct super_block *sb)
3004 {
3005         if (!EXT4_HAS_INCOMPAT_FEATURE(sb, EXT4_FEATURE_INCOMPAT_EXTENTS))
3006                 return;
3007
3008 #ifdef EXTENTS_STATS
3009         if (EXT4_SB(sb)->s_ext_blocks && EXT4_SB(sb)->s_ext_extents) {
3010                 struct ext4_sb_info *sbi = EXT4_SB(sb);
3011                 printk(KERN_ERR "EXT4-fs: %lu blocks in %lu extents (%lu ave)\n",
3012                         sbi->s_ext_blocks, sbi->s_ext_extents,
3013                         sbi->s_ext_blocks / sbi->s_ext_extents);
3014                 printk(KERN_ERR "EXT4-fs: extents: %lu min, %lu max, max depth %lu\n",
3015                         sbi->s_ext_min, sbi->s_ext_max, sbi->s_depth_max);
3016         }
3017 #endif
3018 }
3019
3020 static int ext4_zeroout_es(struct inode *inode, struct ext4_extent *ex)
3021 {
3022         ext4_lblk_t  ee_block;
3023         ext4_fsblk_t ee_pblock;
3024         unsigned int ee_len;
3025
3026         ee_block  = le32_to_cpu(ex->ee_block);
3027         ee_len    = ext4_ext_get_actual_len(ex);
3028         ee_pblock = ext4_ext_pblock(ex);
3029
3030         if (ee_len == 0)
3031                 return 0;
3032
3033         return ext4_es_insert_extent(inode, ee_block, ee_len, ee_pblock,
3034                                      EXTENT_STATUS_WRITTEN);
3035 }
3036
3037 /* FIXME!! we need to try to merge to left or right after zero-out  */
3038 static int ext4_ext_zeroout(struct inode *inode, struct ext4_extent *ex)
3039 {
3040         ext4_fsblk_t ee_pblock;
3041         unsigned int ee_len;
3042         int ret;
3043
3044         ee_len    = ext4_ext_get_actual_len(ex);
3045         ee_pblock = ext4_ext_pblock(ex);
3046
3047         ret = sb_issue_zeroout(inode->i_sb, ee_pblock, ee_len, GFP_NOFS);
3048         if (ret > 0)
3049                 ret = 0;
3050
3051         return ret;
3052 }
3053
3054 /*
3055  * ext4_split_extent_at() splits an extent at given block.
3056  *
3057  * @handle: the journal handle
3058  * @inode: the file inode
3059  * @path: the path to the extent
3060  * @split: the logical block where the extent is splitted.
3061  * @split_flags: indicates if the extent could be zeroout if split fails, and
3062  *               the states(init or uninit) of new extents.
3063  * @flags: flags used to insert new extent to extent tree.
3064  *
3065  *
3066  * Splits extent [a, b] into two extents [a, @split) and [@split, b], states
3067  * of which are deterimined by split_flag.
3068  *
3069  * There are two cases:
3070  *  a> the extent are splitted into two extent.
3071  *  b> split is not needed, and just mark the extent.
3072  *
3073  * return 0 on success.
3074  */
3075 static int ext4_split_extent_at(handle_t *handle,
3076                              struct inode *inode,
3077                              struct ext4_ext_path *path,
3078                              ext4_lblk_t split,
3079                              int split_flag,
3080                              int flags)
3081 {
3082         ext4_fsblk_t newblock;
3083         ext4_lblk_t ee_block;
3084         struct ext4_extent *ex, newex, orig_ex, zero_ex;
3085         struct ext4_extent *ex2 = NULL;
3086         unsigned int ee_len, depth;
3087         int err = 0;
3088
3089         BUG_ON((split_flag & (EXT4_EXT_DATA_VALID1 | EXT4_EXT_DATA_VALID2)) ==
3090                (EXT4_EXT_DATA_VALID1 | EXT4_EXT_DATA_VALID2));
3091
3092         ext_debug("ext4_split_extents_at: inode %lu, logical"
3093                 "block %llu\n", inode->i_ino, (unsigned long long)split);
3094
3095         ext4_ext_show_leaf(inode, path);
3096
3097         depth = ext_depth(inode);
3098         ex = path[depth].p_ext;
3099         ee_block = le32_to_cpu(ex->ee_block);
3100         ee_len = ext4_ext_get_actual_len(ex);
3101         newblock = split - ee_block + ext4_ext_pblock(ex);
3102
3103         BUG_ON(split < ee_block || split >= (ee_block + ee_len));
3104         BUG_ON(!ext4_ext_is_uninitialized(ex) &&
3105                split_flag & (EXT4_EXT_MAY_ZEROOUT |
3106                              EXT4_EXT_MARK_UNINIT1 |
3107                              EXT4_EXT_MARK_UNINIT2));
3108
3109         err = ext4_ext_get_access(handle, inode, path + depth);
3110         if (err)
3111                 goto out;
3112
3113         if (split == ee_block) {
3114                 /*
3115                  * case b: block @split is the block that the extent begins with
3116                  * then we just change the state of the extent, and splitting
3117                  * is not needed.
3118                  */
3119                 if (split_flag & EXT4_EXT_MARK_UNINIT2)
3120                         ext4_ext_mark_uninitialized(ex);
3121                 else
3122                         ext4_ext_mark_initialized(ex);
3123
3124                 if (!(flags & EXT4_GET_BLOCKS_PRE_IO))
3125                         ext4_ext_try_to_merge(handle, inode, path, ex);
3126
3127                 err = ext4_ext_dirty(handle, inode, path + path->p_depth);
3128                 goto out;
3129         }
3130
3131         /* case a */
3132         memcpy(&orig_ex, ex, sizeof(orig_ex));
3133         ex->ee_len = cpu_to_le16(split - ee_block);
3134         if (split_flag & EXT4_EXT_MARK_UNINIT1)
3135                 ext4_ext_mark_uninitialized(ex);
3136
3137         /*
3138          * path may lead to new leaf, not to original leaf any more
3139          * after ext4_ext_insert_extent() returns,
3140          */
3141         err = ext4_ext_dirty(handle, inode, path + depth);
3142         if (err)
3143                 goto fix_extent_len;
3144
3145         ex2 = &newex;
3146         ex2->ee_block = cpu_to_le32(split);
3147         ex2->ee_len   = cpu_to_le16(ee_len - (split - ee_block));
3148         ext4_ext_store_pblock(ex2, newblock);
3149         if (split_flag & EXT4_EXT_MARK_UNINIT2)
3150                 ext4_ext_mark_uninitialized(ex2);
3151
3152         err = ext4_ext_insert_extent(handle, inode, path, &newex, flags);
3153         if (err == -ENOSPC && (EXT4_EXT_MAY_ZEROOUT & split_flag)) {
3154                 if (split_flag & (EXT4_EXT_DATA_VALID1|EXT4_EXT_DATA_VALID2)) {
3155                         if (split_flag & EXT4_EXT_DATA_VALID1) {
3156                                 err = ext4_ext_zeroout(inode, ex2);
3157                                 zero_ex.ee_block = ex2->ee_block;
3158                                 zero_ex.ee_len = cpu_to_le16(
3159                                                 ext4_ext_get_actual_len(ex2));
3160                                 ext4_ext_store_pblock(&zero_ex,
3161                                                       ext4_ext_pblock(ex2));
3162                         } else {
3163                                 err = ext4_ext_zeroout(inode, ex);
3164                                 zero_ex.ee_block = ex->ee_block;
3165                                 zero_ex.ee_len = cpu_to_le16(
3166                                                 ext4_ext_get_actual_len(ex));
3167                                 ext4_ext_store_pblock(&zero_ex,
3168                                                       ext4_ext_pblock(ex));
3169                         }
3170                 } else {
3171                         err = ext4_ext_zeroout(inode, &orig_ex);
3172                         zero_ex.ee_block = orig_ex.ee_block;
3173                         zero_ex.ee_len = cpu_to_le16(
3174                                                 ext4_ext_get_actual_len(&orig_ex));
3175                         ext4_ext_store_pblock(&zero_ex,
3176                                               ext4_ext_pblock(&orig_ex));
3177                 }
3178
3179                 if (err)
3180                         goto fix_extent_len;
3181                 /* update the extent length and mark as initialized */
3182                 ex->ee_len = cpu_to_le16(ee_len);
3183                 ext4_ext_try_to_merge(handle, inode, path, ex);
3184                 err = ext4_ext_dirty(handle, inode, path + path->p_depth);
3185                 if (err)
3186                         goto fix_extent_len;
3187
3188                 /* update extent status tree */
3189                 err = ext4_zeroout_es(inode, &zero_ex);
3190
3191                 goto out;
3192         } else if (err)
3193                 goto fix_extent_len;
3194
3195 out:
3196         ext4_ext_show_leaf(inode, path);
3197         return err;
3198
3199 fix_extent_len:
3200         ex->ee_len = orig_ex.ee_len;
3201         ext4_ext_dirty(handle, inode, path + depth);
3202         return err;
3203 }
3204
3205 /*
3206  * ext4_split_extents() splits an extent and mark extent which is covered
3207  * by @map as split_flags indicates
3208  *
3209  * It may result in splitting the extent into multiple extents (up to three)
3210  * There are three possibilities:
3211  *   a> There is no split required
3212  *   b> Splits in two extents: Split is happening at either end of the extent
3213  *   c> Splits in three extents: Somone is splitting in middle of the extent
3214  *
3215  */
3216 static int ext4_split_extent(handle_t *handle,
3217                               struct inode *inode,
3218                               struct ext4_ext_path *path,
3219                               struct ext4_map_blocks *map,
3220                               int split_flag,
3221                               int flags)
3222 {
3223         ext4_lblk_t ee_block;
3224         struct ext4_extent *ex;
3225         unsigned int ee_len, depth;
3226         int err = 0;
3227         int uninitialized;
3228         int split_flag1, flags1;
3229         int allocated = map->m_len;
3230
3231         depth = ext_depth(inode);
3232         ex = path[depth].p_ext;
3233         ee_block = le32_to_cpu(ex->ee_block);
3234         ee_len = ext4_ext_get_actual_len(ex);
3235         uninitialized = ext4_ext_is_uninitialized(ex);
3236
3237         if (map->m_lblk + map->m_len < ee_block + ee_len) {
3238                 split_flag1 = split_flag & EXT4_EXT_MAY_ZEROOUT;
3239                 flags1 = flags | EXT4_GET_BLOCKS_PRE_IO;
3240                 if (uninitialized)
3241                         split_flag1 |= EXT4_EXT_MARK_UNINIT1 |
3242                                        EXT4_EXT_MARK_UNINIT2;
3243                 if (split_flag & EXT4_EXT_DATA_VALID2)
3244                         split_flag1 |= EXT4_EXT_DATA_VALID1;
3245                 err = ext4_split_extent_at(handle, inode, path,
3246                                 map->m_lblk + map->m_len, split_flag1, flags1);
3247                 if (err)
3248                         goto out;
3249         } else {
3250                 allocated = ee_len - (map->m_lblk - ee_block);
3251         }
3252         /*
3253          * Update path is required because previous ext4_split_extent_at() may
3254          * result in split of original leaf or extent zeroout.
3255          */
3256         ext4_ext_drop_refs(path);
3257         path = ext4_ext_find_extent(inode, map->m_lblk, path, 0);
3258         if (IS_ERR(path))
3259                 return PTR_ERR(path);
3260         depth = ext_depth(inode);
3261         ex = path[depth].p_ext;
3262         uninitialized = ext4_ext_is_uninitialized(ex);
3263         split_flag1 = 0;
3264
3265         if (map->m_lblk >= ee_block) {
3266                 split_flag1 = split_flag & EXT4_EXT_DATA_VALID2;
3267                 if (uninitialized) {
3268                         split_flag1 |= EXT4_EXT_MARK_UNINIT1;
3269                         split_flag1 |= split_flag & (EXT4_EXT_MAY_ZEROOUT |
3270                                                      EXT4_EXT_MARK_UNINIT2);
3271                 }
3272                 err = ext4_split_extent_at(handle, inode, path,
3273                                 map->m_lblk, split_flag1, flags);
3274                 if (err)
3275                         goto out;
3276         }
3277
3278         ext4_ext_show_leaf(inode, path);
3279 out:
3280         return err ? err : allocated;
3281 }
3282
3283 /*
3284  * This function is called by ext4_ext_map_blocks() if someone tries to write
3285  * to an uninitialized extent. It may result in splitting the uninitialized
3286  * extent into multiple extents (up to three - one initialized and two
3287  * uninitialized).
3288  * There are three possibilities:
3289  *   a> There is no split required: Entire extent should be initialized
3290  *   b> Splits in two extents: Write is happening at either end of the extent
3291  *   c> Splits in three extents: Somone is writing in middle of the extent
3292  *
3293  * Pre-conditions:
3294  *  - The extent pointed to by 'path' is uninitialized.
3295  *  - The extent pointed to by 'path' contains a superset
3296  *    of the logical span [map->m_lblk, map->m_lblk + map->m_len).
3297  *
3298  * Post-conditions on success:
3299  *  - the returned value is the number of blocks beyond map->l_lblk
3300  *    that are allocated and initialized.
3301  *    It is guaranteed to be >= map->m_len.
3302  */
3303 static int ext4_ext_convert_to_initialized(handle_t *handle,
3304                                            struct inode *inode,
3305                                            struct ext4_map_blocks *map,
3306                                            struct ext4_ext_path *path,
3307                                            int flags)
3308 {
3309         struct ext4_sb_info *sbi;
3310         struct ext4_extent_header *eh;
3311         struct ext4_map_blocks split_map;
3312         struct ext4_extent zero_ex;
3313         struct ext4_extent *ex, *abut_ex;
3314         ext4_lblk_t ee_block, eof_block;
3315         unsigned int ee_len, depth, map_len = map->m_len;
3316         int allocated = 0, max_zeroout = 0;
3317         int err = 0;
3318         int split_flag = 0;
3319
3320         ext_debug("ext4_ext_convert_to_initialized: inode %lu, logical"
3321                 "block %llu, max_blocks %u\n", inode->i_ino,
3322                 (unsigned long long)map->m_lblk, map_len);
3323
3324         sbi = EXT4_SB(inode->i_sb);
3325         eof_block = (inode->i_size + inode->i_sb->s_blocksize - 1) >>
3326                 inode->i_sb->s_blocksize_bits;
3327         if (eof_block < map->m_lblk + map_len)
3328                 eof_block = map->m_lblk + map_len;
3329
3330         depth = ext_depth(inode);
3331         eh = path[depth].p_hdr;
3332         ex = path[depth].p_ext;
3333         ee_block = le32_to_cpu(ex->ee_block);
3334         ee_len = ext4_ext_get_actual_len(ex);
3335         zero_ex.ee_len = 0;
3336
3337         trace_ext4_ext_convert_to_initialized_enter(inode, map, ex);
3338
3339         /* Pre-conditions */
3340         BUG_ON(!ext4_ext_is_uninitialized(ex));
3341         BUG_ON(!in_range(map->m_lblk, ee_block, ee_len));
3342
3343         /*
3344          * Attempt to transfer newly initialized blocks from the currently
3345          * uninitialized extent to its neighbor. This is much cheaper
3346          * than an insertion followed by a merge as those involve costly
3347          * memmove() calls. Transferring to the left is the common case in
3348          * steady state for workloads doing fallocate(FALLOC_FL_KEEP_SIZE)
3349          * followed by append writes.
3350          *
3351          * Limitations of the current logic:
3352          *  - L1: we do not deal with writes covering the whole extent.
3353          *    This would require removing the extent if the transfer
3354          *    is possible.
3355          *  - L2: we only attempt to merge with an extent stored in the
3356          *    same extent tree node.
3357          */
3358         if ((map->m_lblk == ee_block) &&
3359                 /* See if we can merge left */
3360                 (map_len < ee_len) &&           /*L1*/
3361                 (ex > EXT_FIRST_EXTENT(eh))) {  /*L2*/
3362                 ext4_lblk_t prev_lblk;
3363                 ext4_fsblk_t prev_pblk, ee_pblk;
3364                 unsigned int prev_len;
3365
3366                 abut_ex = ex - 1;
3367                 prev_lblk = le32_to_cpu(abut_ex->ee_block);
3368                 prev_len = ext4_ext_get_actual_len(abut_ex);
3369                 prev_pblk = ext4_ext_pblock(abut_ex);
3370                 ee_pblk = ext4_ext_pblock(ex);
3371
3372                 /*
3373                  * A transfer of blocks from 'ex' to 'abut_ex' is allowed
3374                  * upon those conditions:
3375                  * - C1: abut_ex is initialized,
3376                  * - C2: abut_ex is logically abutting ex,
3377                  * - C3: abut_ex is physically abutting ex,
3378                  * - C4: abut_ex can receive the additional blocks without
3379                  *   overflowing the (initialized) length limit.
3380                  */
3381                 if ((!ext4_ext_is_uninitialized(abut_ex)) &&            /*C1*/
3382                         ((prev_lblk + prev_len) == ee_block) &&         /*C2*/
3383                         ((prev_pblk + prev_len) == ee_pblk) &&          /*C3*/
3384                         (prev_len < (EXT_INIT_MAX_LEN - map_len))) {    /*C4*/
3385                         err = ext4_ext_get_access(handle, inode, path + depth);
3386                         if (err)
3387                                 goto out;
3388
3389                         trace_ext4_ext_convert_to_initialized_fastpath(inode,
3390                                 map, ex, abut_ex);
3391
3392                         /* Shift the start of ex by 'map_len' blocks */
3393                         ex->ee_block = cpu_to_le32(ee_block + map_len);
3394                         ext4_ext_store_pblock(ex, ee_pblk + map_len);
3395                         ex->ee_len = cpu_to_le16(ee_len - map_len);
3396                         ext4_ext_mark_uninitialized(ex); /* Restore the flag */
3397
3398                         /* Extend abut_ex by 'map_len' blocks */
3399                         abut_ex->ee_len = cpu_to_le16(prev_len + map_len);
3400
3401                         /* Result: number of initialized blocks past m_lblk */
3402                         allocated = map_len;
3403                 }
3404         } else if (((map->m_lblk + map_len) == (ee_block + ee_len)) &&
3405                    (map_len < ee_len) &&        /*L1*/
3406                    ex < EXT_LAST_EXTENT(eh)) {  /*L2*/
3407                 /* See if we can merge right */
3408                 ext4_lblk_t next_lblk;
3409                 ext4_fsblk_t next_pblk, ee_pblk;
3410                 unsigned int next_len;
3411
3412                 abut_ex = ex + 1;
3413                 next_lblk = le32_to_cpu(abut_ex->ee_block);
3414                 next_len = ext4_ext_get_actual_len(abut_ex);
3415                 next_pblk = ext4_ext_pblock(abut_ex);
3416                 ee_pblk = ext4_ext_pblock(ex);
3417
3418                 /*
3419                  * A transfer of blocks from 'ex' to 'abut_ex' is allowed
3420                  * upon those conditions:
3421                  * - C1: abut_ex is initialized,
3422                  * - C2: abut_ex is logically abutting ex,
3423                  * - C3: abut_ex is physically abutting ex,
3424                  * - C4: abut_ex can receive the additional blocks without
3425                  *   overflowing the (initialized) length limit.
3426                  */
3427                 if ((!ext4_ext_is_uninitialized(abut_ex)) &&            /*C1*/
3428                     ((map->m_lblk + map_len) == next_lblk) &&           /*C2*/
3429                     ((ee_pblk + ee_len) == next_pblk) &&                /*C3*/
3430                     (next_len < (EXT_INIT_MAX_LEN - map_len))) {        /*C4*/
3431                         err = ext4_ext_get_access(handle, inode, path + depth);
3432                         if (err)
3433                                 goto out;
3434
3435                         trace_ext4_ext_convert_to_initialized_fastpath(inode,
3436                                 map, ex, abut_ex);
3437
3438                         /* Shift the start of abut_ex by 'map_len' blocks */
3439                         abut_ex->ee_block = cpu_to_le32(next_lblk - map_len);
3440                         ext4_ext_store_pblock(abut_ex, next_pblk - map_len);
3441                         ex->ee_len = cpu_to_le16(ee_len - map_len);
3442                         ext4_ext_mark_uninitialized(ex); /* Restore the flag */
3443
3444                         /* Extend abut_ex by 'map_len' blocks */
3445                         abut_ex->ee_len = cpu_to_le16(next_len + map_len);
3446
3447                         /* Result: number of initialized blocks past m_lblk */
3448                         allocated = map_len;
3449                 }
3450         }
3451         if (allocated) {
3452                 /* Mark the block containing both extents as dirty */
3453                 ext4_ext_dirty(handle, inode, path + depth);
3454
3455                 /* Update path to point to the right extent */
3456                 path[depth].p_ext = abut_ex;
3457                 goto out;
3458         } else
3459                 allocated = ee_len - (map->m_lblk - ee_block);
3460
3461         WARN_ON(map->m_lblk < ee_block);
3462         /*
3463          * It is safe to convert extent to initialized via explicit
3464          * zeroout only if extent is fully insde i_size or new_size.
3465          */
3466         split_flag |= ee_block + ee_len <= eof_block ? EXT4_EXT_MAY_ZEROOUT : 0;
3467
3468         if (EXT4_EXT_MAY_ZEROOUT & split_flag)
3469                 max_zeroout = sbi->s_extent_max_zeroout_kb >>
3470                         (inode->i_sb->s_blocksize_bits - 10);
3471
3472         /* If extent is less than s_max_zeroout_kb, zeroout directly */
3473         if (max_zeroout && (ee_len <= max_zeroout)) {
3474                 err = ext4_ext_zeroout(inode, ex);
3475                 if (err)
3476                         goto out;
3477                 zero_ex.ee_block = ex->ee_block;
3478                 zero_ex.ee_len = cpu_to_le16(ext4_ext_get_actual_len(ex));
3479                 ext4_ext_store_pblock(&zero_ex, ext4_ext_pblock(ex));
3480
3481                 err = ext4_ext_get_access(handle, inode, path + depth);
3482                 if (err)
3483                         goto out;
3484                 ext4_ext_mark_initialized(ex);
3485                 ext4_ext_try_to_merge(handle, inode, path, ex);
3486                 err = ext4_ext_dirty(handle, inode, path + path->p_depth);
3487                 goto out;
3488         }
3489
3490         /*
3491          * four cases:
3492          * 1. split the extent into three extents.
3493          * 2. split the extent into two extents, zeroout the first half.
3494          * 3. split the extent into two extents, zeroout the second half.
3495          * 4. split the extent into two extents with out zeroout.
3496          */
3497         split_map.m_lblk = map->m_lblk;
3498         split_map.m_len = map->m_len;
3499
3500         if (max_zeroout && (allocated > map->m_len)) {
3501                 if (allocated <= max_zeroout) {
3502                         /* case 3 */
3503                         zero_ex.ee_block =
3504                                          cpu_to_le32(map->m_lblk);
3505                         zero_ex.ee_len = cpu_to_le16(allocated);
3506                         ext4_ext_store_pblock(&zero_ex,
3507                                 ext4_ext_pblock(ex) + map->m_lblk - ee_block);
3508                         err = ext4_ext_zeroout(inode, &zero_ex);
3509                         if (err)
3510                                 goto out;
3511                         split_map.m_lblk = map->m_lblk;
3512                         split_map.m_len = allocated;
3513                 } else if (map->m_lblk - ee_block + map->m_len < max_zeroout) {
3514                         /* case 2 */
3515                         if (map->m_lblk != ee_block) {
3516                                 zero_ex.ee_block = ex->ee_block;
3517                                 zero_ex.ee_len = cpu_to_le16(map->m_lblk -
3518                                                         ee_block);
3519                                 ext4_ext_store_pblock(&zero_ex,
3520                                                       ext4_ext_pblock(ex));
3521                                 err = ext4_ext_zeroout(inode, &zero_ex);
3522                                 if (err)
3523                                         goto out;
3524                         }
3525
3526                         split_map.m_lblk = ee_block;
3527                         split_map.m_len = map->m_lblk - ee_block + map->m_len;
3528                         allocated = map->m_len;
3529                 }
3530         }
3531
3532         allocated = ext4_split_extent(handle, inode, path,
3533                                       &split_map, split_flag, flags);
3534         if (allocated < 0)
3535                 err = allocated;
3536
3537 out:
3538         /* If we have gotten a failure, don't zero out status tree */
3539         if (!err)
3540                 err = ext4_zeroout_es(inode, &zero_ex);
3541         return err ? err : allocated;
3542 }
3543
3544 /*
3545  * This function is called by ext4_ext_map_blocks() from
3546  * ext4_get_blocks_dio_write() when DIO to write
3547  * to an uninitialized extent.
3548  *
3549  * Writing to an uninitialized extent may result in splitting the uninitialized
3550  * extent into multiple initialized/uninitialized extents (up to three)
3551  * There are three possibilities:
3552  *   a> There is no split required: Entire extent should be uninitialized
3553  *   b> Splits in two extents: Write is happening at either end of the extent
3554  *   c> Splits in three extents: Somone is writing in middle of the extent
3555  *
3556  * One of more index blocks maybe needed if the extent tree grow after
3557  * the uninitialized extent split. To prevent ENOSPC occur at the IO
3558  * complete, we need to split the uninitialized extent before DIO submit
3559  * the IO. The uninitialized extent called at this time will be split
3560  * into three uninitialized extent(at most). After IO complete, the part
3561  * being filled will be convert to initialized by the end_io callback function
3562  * via ext4_convert_unwritten_extents().
3563  *
3564  * Returns the size of uninitialized extent to be written on success.
3565  */
3566 static int ext4_split_unwritten_extents(handle_t *handle,
3567                                         struct inode *inode,
3568                                         struct ext4_map_blocks *map,
3569                                         struct ext4_ext_path *path,
3570                                         int flags)
3571 {
3572         ext4_lblk_t eof_block;
3573         ext4_lblk_t ee_block;
3574         struct ext4_extent *ex;
3575         unsigned int ee_len;
3576         int split_flag = 0, depth;
3577
3578         ext_debug("ext4_split_unwritten_extents: inode %lu, logical"
3579                 "block %llu, max_blocks %u\n", inode->i_ino,
3580                 (unsigned long long)map->m_lblk, map->m_len);
3581
3582         eof_block = (inode->i_size + inode->i_sb->s_blocksize - 1) >>
3583                 inode->i_sb->s_blocksize_bits;
3584         if (eof_block < map->m_lblk + map->m_len)
3585                 eof_block = map->m_lblk + map->m_len;
3586         /*
3587          * It is safe to convert extent to initialized via explicit
3588          * zeroout only if extent is fully insde i_size or new_size.
3589          */
3590         depth = ext_depth(inode);
3591         ex = path[depth].p_ext;
3592         ee_block = le32_to_cpu(ex->ee_block);
3593         ee_len = ext4_ext_get_actual_len(ex);
3594
3595         split_flag |= ee_block + ee_len <= eof_block ? EXT4_EXT_MAY_ZEROOUT : 0;
3596         split_flag |= EXT4_EXT_MARK_UNINIT2;
3597         if (flags & EXT4_GET_BLOCKS_CONVERT)
3598                 split_flag |= EXT4_EXT_DATA_VALID2;
3599         flags |= EXT4_GET_BLOCKS_PRE_IO;
3600         return ext4_split_extent(handle, inode, path, map, split_flag, flags);
3601 }
3602
3603 static int ext4_convert_unwritten_extents_endio(handle_t *handle,
3604                                                 struct inode *inode,
3605                                                 struct ext4_map_blocks *map,
3606                                                 struct ext4_ext_path *path)
3607 {
3608         struct ext4_extent *ex;
3609         ext4_lblk_t ee_block;
3610         unsigned int ee_len;
3611         int depth;
3612         int err = 0;
3613
3614         depth = ext_depth(inode);
3615         ex = path[depth].p_ext;
3616         ee_block = le32_to_cpu(ex->ee_block);
3617         ee_len = ext4_ext_get_actual_len(ex);
3618
3619         ext_debug("ext4_convert_unwritten_extents_endio: inode %lu, logical"
3620                 "block %llu, max_blocks %u\n", inode->i_ino,
3621                   (unsigned long long)ee_block, ee_len);
3622
3623         /* If extent is larger than requested it is a clear sign that we still
3624          * have some extent state machine issues left. So extent_split is still
3625          * required.
3626          * TODO: Once all related issues will be fixed this situation should be
3627          * illegal.
3628          */
3629         if (ee_block != map->m_lblk || ee_len > map->m_len) {
3630 #ifdef EXT4_DEBUG
3631                 ext4_warning("Inode (%ld) finished: extent logical block %llu,"
3632                              " len %u; IO logical block %llu, len %u\n",
3633                              inode->i_ino, (unsigned long long)ee_block, ee_len,
3634                              (unsigned long long)map->m_lblk, map->m_len);
3635 #endif
3636                 err = ext4_split_unwritten_extents(handle, inode, map, path,
3637                                                    EXT4_GET_BLOCKS_CONVERT);
3638                 if (err < 0)
3639                         goto out;
3640                 ext4_ext_drop_refs(path);
3641                 path = ext4_ext_find_extent(inode, map->m_lblk, path, 0);
3642                 if (IS_ERR(path)) {
3643                         err = PTR_ERR(path);
3644                         goto out;
3645                 }
3646                 depth = ext_depth(inode);
3647                 ex = path[depth].p_ext;
3648         }
3649
3650         err = ext4_ext_get_access(handle, inode, path + depth);
3651         if (err)
3652                 goto out;
3653         /* first mark the extent as initialized */
3654         ext4_ext_mark_initialized(ex);
3655
3656         /* note: ext4_ext_correct_indexes() isn't needed here because
3657          * borders are not changed
3658          */
3659         ext4_ext_try_to_merge(handle, inode, path, ex);
3660
3661         /* Mark modified extent as dirty */
3662         err = ext4_ext_dirty(handle, inode, path + path->p_depth);
3663 out:
3664         ext4_ext_show_leaf(inode, path);
3665         return err;
3666 }
3667
3668 static void unmap_underlying_metadata_blocks(struct block_device *bdev,
3669                         sector_t block, int count)
3670 {
3671         int i;
3672         for (i = 0; i < count; i++)
3673                 unmap_underlying_metadata(bdev, block + i);
3674 }
3675
3676 /*
3677  * Handle EOFBLOCKS_FL flag, clearing it if necessary
3678  */
3679 static int check_eofblocks_fl(handle_t *handle, struct inode *inode,
3680                               ext4_lblk_t lblk,
3681                               struct ext4_ext_path *path,
3682                               unsigned int len)
3683 {
3684         int i, depth;
3685         struct ext4_extent_header *eh;
3686         struct ext4_extent *last_ex;
3687
3688         if (!ext4_test_inode_flag(inode, EXT4_INODE_EOFBLOCKS))
3689                 return 0;
3690
3691         depth = ext_depth(inode);
3692         eh = path[depth].p_hdr;
3693
3694         /*
3695          * We're going to remove EOFBLOCKS_FL entirely in future so we
3696          * do not care for this case anymore. Simply remove the flag
3697          * if there are no extents.
3698          */
3699         if (unlikely(!eh->eh_entries))
3700                 goto out;
3701         last_ex = EXT_LAST_EXTENT(eh);
3702         /*
3703          * We should clear the EOFBLOCKS_FL flag if we are writing the
3704          * last block in the last extent in the file.  We test this by
3705          * first checking to see if the caller to
3706          * ext4_ext_get_blocks() was interested in the last block (or
3707          * a block beyond the last block) in the current extent.  If
3708          * this turns out to be false, we can bail out from this
3709          * function immediately.
3710          */
3711         if (lblk + len < le32_to_cpu(last_ex->ee_block) +
3712             ext4_ext_get_actual_len(last_ex))
3713                 return 0;
3714         /*
3715          * If the caller does appear to be planning to write at or
3716          * beyond the end of the current extent, we then test to see
3717          * if the current extent is the last extent in the file, by
3718          * checking to make sure it was reached via the rightmost node
3719          * at each level of the tree.
3720          */
3721         for (i = depth-1; i >= 0; i--)
3722                 if (path[i].p_idx != EXT_LAST_INDEX(path[i].p_hdr))
3723                         return 0;
3724 out:
3725         ext4_clear_inode_flag(inode, EXT4_INODE_EOFBLOCKS);
3726         return ext4_mark_inode_dirty(handle, inode);
3727 }
3728
3729 /**
3730  * ext4_find_delalloc_range: find delayed allocated block in the given range.
3731  *
3732  * Return 1 if there is a delalloc block in the range, otherwise 0.
3733  */
3734 int ext4_find_delalloc_range(struct inode *inode,
3735                              ext4_lblk_t lblk_start,
3736                              ext4_lblk_t lblk_end)
3737 {
3738         struct extent_status es;
3739
3740         ext4_es_find_delayed_extent_range(inode, lblk_start, lblk_end, &es);
3741         if (es.es_len == 0)
3742                 return 0; /* there is no delay extent in this tree */
3743         else if (es.es_lblk <= lblk_start &&
3744                  lblk_start < es.es_lblk + es.es_len)
3745                 return 1;
3746         else if (lblk_start <= es.es_lblk && es.es_lblk <= lblk_end)
3747                 return 1;
3748         else
3749                 return 0;
3750 }
3751
3752 int ext4_find_delalloc_cluster(struct inode *inode, ext4_lblk_t lblk)
3753 {
3754         struct ext4_sb_info *sbi = EXT4_SB(inode->i_sb);
3755         ext4_lblk_t lblk_start, lblk_end;
3756         lblk_start = lblk & (~(sbi->s_cluster_ratio - 1));
3757         lblk_end = lblk_start + sbi->s_cluster_ratio - 1;
3758
3759         return ext4_find_delalloc_range(inode, lblk_start, lblk_end);
3760 }
3761
3762 /**
3763  * Determines how many complete clusters (out of those specified by the 'map')
3764  * are under delalloc and were reserved quota for.
3765  * This function is called when we are writing out the blocks that were
3766  * originally written with their allocation delayed, but then the space was
3767  * allocated using fallocate() before the delayed allocation could be resolved.
3768  * The cases to look for are:
3769  * ('=' indicated delayed allocated blocks
3770  *  '-' indicates non-delayed allocated blocks)
3771  * (a) partial clusters towards beginning and/or end outside of allocated range
3772  *     are not delalloc'ed.
3773  *      Ex:
3774  *      |----c---=|====c====|====c====|===-c----|
3775  *               |++++++ allocated ++++++|
3776  *      ==> 4 complete clusters in above example
3777  *
3778  * (b) partial cluster (outside of allocated range) towards either end is
3779  *     marked for delayed allocation. In this case, we will exclude that
3780  *     cluster.
3781  *      Ex:
3782  *      |----====c========|========c========|
3783  *           |++++++ allocated ++++++|
3784  *      ==> 1 complete clusters in above example
3785  *
3786  *      Ex:
3787  *      |================c================|
3788  *            |++++++ allocated ++++++|
3789  *      ==> 0 complete clusters in above example
3790  *
3791  * The ext4_da_update_reserve_space will be called only if we
3792  * determine here that there were some "entire" clusters that span
3793  * this 'allocated' range.
3794  * In the non-bigalloc case, this function will just end up returning num_blks
3795  * without ever calling ext4_find_delalloc_range.
3796  */
3797 static unsigned int
3798 get_reserved_cluster_alloc(struct inode *inode, ext4_lblk_t lblk_start,
3799                            unsigned int num_blks)
3800 {
3801         struct ext4_sb_info *sbi = EXT4_SB(inode->i_sb);
3802         ext4_lblk_t alloc_cluster_start, alloc_cluster_end;
3803         ext4_lblk_t lblk_from, lblk_to, c_offset;
3804         unsigned int allocated_clusters = 0;
3805
3806         alloc_cluster_start = EXT4_B2C(sbi, lblk_start);
3807         alloc_cluster_end = EXT4_B2C(sbi, lblk_start + num_blks - 1);
3808
3809         /* max possible clusters for this allocation */
3810         allocated_clusters = alloc_cluster_end - alloc_cluster_start + 1;
3811
3812         trace_ext4_get_reserved_cluster_alloc(inode, lblk_start, num_blks);
3813
3814         /* Check towards left side */
3815         c_offset = lblk_start & (sbi->s_cluster_ratio - 1);
3816         if (c_offset) {
3817                 lblk_from = lblk_start & (~(sbi->s_cluster_ratio - 1));
3818                 lblk_to = lblk_from + c_offset - 1;
3819
3820                 if (ext4_find_delalloc_range(inode, lblk_from, lblk_to))
3821                         allocated_clusters--;
3822         }
3823
3824         /* Now check towards right. */
3825         c_offset = (lblk_start + num_blks) & (sbi->s_cluster_ratio - 1);
3826         if (allocated_clusters && c_offset) {
3827                 lblk_from = lblk_start + num_blks;
3828                 lblk_to = lblk_from + (sbi->s_cluster_ratio - c_offset) - 1;
3829
3830                 if (ext4_find_delalloc_range(inode, lblk_from, lblk_to))
3831                         allocated_clusters--;
3832         }
3833
3834         return allocated_clusters;
3835 }
3836
3837 static int
3838 ext4_ext_handle_uninitialized_extents(handle_t *handle, struct inode *inode,
3839                         struct ext4_map_blocks *map,
3840                         struct ext4_ext_path *path, int flags,
3841                         unsigned int allocated, ext4_fsblk_t newblock)
3842 {
3843         int ret = 0;
3844         int err = 0;
3845         ext4_io_end_t *io = ext4_inode_aio(inode);
3846
3847         ext_debug("ext4_ext_handle_uninitialized_extents: inode %lu, logical "
3848                   "block %llu, max_blocks %u, flags %x, allocated %u\n",
3849                   inode->i_ino, (unsigned long long)map->m_lblk, map->m_len,
3850                   flags, allocated);
3851         ext4_ext_show_leaf(inode, path);
3852
3853         /*
3854          * When writing into uninitialized space, we should not fail to
3855          * allocate metadata blocks for the new extent block if needed.
3856          */
3857         flags |= EXT4_GET_BLOCKS_METADATA_NOFAIL;
3858
3859         trace_ext4_ext_handle_uninitialized_extents(inode, map, flags,
3860                                                     allocated, newblock);
3861
3862         /* get_block() before submit the IO, split the extent */
3863         if ((flags & EXT4_GET_BLOCKS_PRE_IO)) {
3864                 ret = ext4_split_unwritten_extents(handle, inode, map,
3865                                                    path, flags);
3866                 if (ret <= 0)
3867                         goto out;
3868                 /*
3869                  * Flag the inode(non aio case) or end_io struct (aio case)
3870                  * that this IO needs to conversion to written when IO is
3871                  * completed
3872                  */
3873                 if (io)
3874                         ext4_set_io_unwritten_flag(inode, io);
3875                 else
3876                         ext4_set_inode_state(inode, EXT4_STATE_DIO_UNWRITTEN);
3877                 map->m_flags |= EXT4_MAP_UNWRITTEN;
3878                 if (ext4_should_dioread_nolock(inode))
3879                         map->m_flags |= EXT4_MAP_UNINIT;
3880                 goto out;
3881         }
3882         /* IO end_io complete, convert the filled extent to written */
3883         if ((flags & EXT4_GET_BLOCKS_CONVERT)) {
3884                 ret = ext4_convert_unwritten_extents_endio(handle, inode, map,
3885                                                         path);
3886                 if (ret >= 0) {
3887                         ext4_update_inode_fsync_trans(handle, inode, 1);
3888                         err = check_eofblocks_fl(handle, inode, map->m_lblk,
3889                                                  path, map->m_len);
3890                 } else
3891                         err = ret;
3892                 map->m_flags |= EXT4_MAP_MAPPED;
3893                 if (allocated > map->m_len)
3894                         allocated = map->m_len;
3895                 map->m_len = allocated;
3896                 goto out2;
3897         }
3898         /* buffered IO case */
3899         /*
3900          * repeat fallocate creation request
3901          * we already have an unwritten extent
3902          */
3903         if (flags & EXT4_GET_BLOCKS_UNINIT_EXT) {
3904                 map->m_flags |= EXT4_MAP_UNWRITTEN;
3905                 goto map_out;
3906         }
3907
3908         /* buffered READ or buffered write_begin() lookup */
3909         if ((flags & EXT4_GET_BLOCKS_CREATE) == 0) {
3910                 /*
3911                  * We have blocks reserved already.  We
3912                  * return allocated blocks so that delalloc
3913                  * won't do block reservation for us.  But
3914                  * the buffer head will be unmapped so that
3915                  * a read from the block returns 0s.
3916                  */
3917                 map->m_flags |= EXT4_MAP_UNWRITTEN;
3918                 goto out1;
3919         }
3920
3921         /* buffered write, writepage time, convert*/
3922         ret = ext4_ext_convert_to_initialized(handle, inode, map, path, flags);
3923         if (ret >= 0)
3924                 ext4_update_inode_fsync_trans(handle, inode, 1);
3925 out:
3926         if (ret <= 0) {
3927                 err = ret;
3928                 goto out2;
3929         } else
3930                 allocated = ret;
3931         map->m_flags |= EXT4_MAP_NEW;
3932         /*
3933          * if we allocated more blocks than requested
3934          * we need to make sure we unmap the extra block
3935          * allocated. The actual needed block will get
3936          * unmapped later when we find the buffer_head marked
3937          * new.
3938          */
3939         if (allocated > map->m_len) {
3940                 unmap_underlying_metadata_blocks(inode->i_sb->s_bdev,
3941                                         newblock + map->m_len,
3942                                         allocated - map->m_len);
3943                 allocated = map->m_len;
3944         }
3945         map->m_len = allocated;
3946
3947         /*
3948          * If we have done fallocate with the offset that is already
3949          * delayed allocated, we would have block reservation
3950          * and quota reservation done in the delayed write path.
3951          * But fallocate would have already updated quota and block
3952          * count for this offset. So cancel these reservation
3953          */
3954         if (flags & EXT4_GET_BLOCKS_DELALLOC_RESERVE) {
3955                 unsigned int reserved_clusters;
3956                 reserved_clusters = get_reserved_cluster_alloc(inode,
3957                                 map->m_lblk, map->m_len);
3958                 if (reserved_clusters)
3959                         ext4_da_update_reserve_space(inode,
3960                                                      reserved_clusters,
3961                                                      0);
3962         }
3963
3964 map_out:
3965         map->m_flags |= EXT4_MAP_MAPPED;
3966         if ((flags & EXT4_GET_BLOCKS_KEEP_SIZE) == 0) {
3967                 err = check_eofblocks_fl(handle, inode, map->m_lblk, path,
3968                                          map->m_len);
3969                 if (err < 0)
3970                         goto out2;
3971         }
3972 out1:
3973         if (allocated > map->m_len)
3974                 allocated = map->m_len;
3975         ext4_ext_show_leaf(inode, path);
3976         map->m_pblk = newblock;
3977         map->m_len = allocated;
3978 out2:
3979         if (path) {
3980                 ext4_ext_drop_refs(path);
3981                 kfree(path);
3982         }
3983         return err ? err : allocated;
3984 }
3985
3986 /*
3987  * get_implied_cluster_alloc - check to see if the requested
3988  * allocation (in the map structure) overlaps with a cluster already
3989  * allocated in an extent.
3990  *      @sb     The filesystem superblock structure
3991  *      @map    The requested lblk->pblk mapping
3992  *      @ex     The extent structure which might contain an implied
3993  *                      cluster allocation
3994  *
3995  * This function is called by ext4_ext_map_blocks() after we failed to
3996  * find blocks that were already in the inode's extent tree.  Hence,
3997  * we know that the beginning of the requested region cannot overlap
3998  * the extent from the inode's extent tree.  There are three cases we
3999  * want to catch.  The first is this case:
4000  *
4001  *               |--- cluster # N--|
4002  *    |--- extent ---|  |---- requested region ---|
4003  *                      |==========|
4004  *
4005  * The second case that we need to test for is this one:
4006  *
4007  *   |--------- cluster # N ----------------|
4008  *         |--- requested region --|   |------- extent ----|
4009  *         |=======================|
4010  *
4011  * The third case is when the requested region lies between two extents
4012  * within the same cluster:
4013  *          |------------- cluster # N-------------|
4014  * |----- ex -----|                  |---- ex_right ----|
4015  *                  |------ requested region ------|
4016  *                  |================|
4017  *
4018  * In each of the above cases, we need to set the map->m_pblk and
4019  * map->m_len so it corresponds to the return the extent labelled as
4020  * "|====|" from cluster #N, since it is already in use for data in
4021  * cluster EXT4_B2C(sbi, map->m_lblk).  We will then return 1 to
4022  * signal to ext4_ext_map_blocks() that map->m_pblk should be treated
4023  * as a new "allocated" block region.  Otherwise, we will return 0 and
4024  * ext4_ext_map_blocks() will then allocate one or more new clusters
4025  * by calling ext4_mb_new_blocks().
4026  */
4027 static int get_implied_cluster_alloc(struct super_block *sb,
4028                                      struct ext4_map_blocks *map,
4029                                      struct ext4_extent *ex,
4030                                      struct ext4_ext_path *path)
4031 {
4032         struct ext4_sb_info *sbi = EXT4_SB(sb);
4033         ext4_lblk_t c_offset = map->m_lblk & (sbi->s_cluster_ratio-1);
4034         ext4_lblk_t ex_cluster_start, ex_cluster_end;
4035         ext4_lblk_t rr_cluster_start;
4036         ext4_lblk_t ee_block = le32_to_cpu(ex->ee_block);
4037         ext4_fsblk_t ee_start = ext4_ext_pblock(ex);
4038         unsigned short ee_len = ext4_ext_get_actual_len(ex);
4039
4040         /* The extent passed in that we are trying to match */
4041         ex_cluster_start = EXT4_B2C(sbi, ee_block);
4042         ex_cluster_end = EXT4_B2C(sbi, ee_block + ee_len - 1);
4043
4044         /* The requested region passed into ext4_map_blocks() */
4045         rr_cluster_start = EXT4_B2C(sbi, map->m_lblk);
4046
4047         if ((rr_cluster_start == ex_cluster_end) ||
4048             (rr_cluster_start == ex_cluster_start)) {
4049                 if (rr_cluster_start == ex_cluster_end)
4050                         ee_start += ee_len - 1;
4051                 map->m_pblk = (ee_start & ~(sbi->s_cluster_ratio - 1)) +
4052                         c_offset;
4053                 map->m_len = min(map->m_len,
4054                                  (unsigned) sbi->s_cluster_ratio - c_offset);
4055                 /*
4056                  * Check for and handle this case:
4057                  *
4058                  *   |--------- cluster # N-------------|
4059                  *                     |------- extent ----|
4060                  *         |--- requested region ---|
4061                  *         |===========|
4062                  */
4063
4064                 if (map->m_lblk < ee_block)
4065                         map->m_len = min(map->m_len, ee_block - map->m_lblk);
4066
4067                 /*
4068                  * Check for the case where there is already another allocated
4069                  * block to the right of 'ex' but before the end of the cluster.
4070                  *
4071                  *          |------------- cluster # N-------------|
4072                  * |----- ex -----|                  |---- ex_right ----|
4073                  *                  |------ requested region ------|
4074                  *                  |================|
4075                  */
4076                 if (map->m_lblk > ee_block) {
4077                         ext4_lblk_t next = ext4_ext_next_allocated_block(path);
4078                         map->m_len = min(map->m_len, next - map->m_lblk);
4079                 }
4080
4081                 trace_ext4_get_implied_cluster_alloc_exit(sb, map, 1);
4082                 return 1;
4083         }
4084
4085         trace_ext4_get_implied_cluster_alloc_exit(sb, map, 0);
4086         return 0;
4087 }
4088
4089
4090 /*
4091  * Block allocation/map/preallocation routine for extents based files
4092  *
4093  *
4094  * Need to be called with
4095  * down_read(&EXT4_I(inode)->i_data_sem) if not allocating file system block
4096  * (ie, create is zero). Otherwise down_write(&EXT4_I(inode)->i_data_sem)
4097  *
4098  * return > 0, number of of blocks already mapped/allocated
4099  *          if create == 0 and these are pre-allocated blocks
4100  *              buffer head is unmapped
4101  *          otherwise blocks are mapped
4102  *
4103  * return = 0, if plain look up failed (blocks have not been allocated)
4104  *          buffer head is unmapped
4105  *
4106  * return < 0, error case.
4107  */
4108 int ext4_ext_map_blocks(handle_t *handle, struct inode *inode,
4109                         struct ext4_map_blocks *map, int flags)
4110 {
4111         struct ext4_ext_path *path = NULL;
4112         struct ext4_extent newex, *ex, *ex2;
4113         struct ext4_sb_info *sbi = EXT4_SB(inode->i_sb);
4114         ext4_fsblk_t newblock = 0;
4115         int free_on_err = 0, err = 0, depth;
4116         unsigned int allocated = 0, offset = 0;
4117         unsigned int allocated_clusters = 0;
4118         struct ext4_allocation_request ar;
4119         ext4_io_end_t *io = ext4_inode_aio(inode);
4120         ext4_lblk_t cluster_offset;
4121         int set_unwritten = 0;
4122
4123         ext_debug("blocks %u/%u requested for inode %lu\n",
4124                   map->m_lblk, map->m_len, inode->i_ino);
4125         trace_ext4_ext_map_blocks_enter(inode, map->m_lblk, map->m_len, flags);
4126
4127         /* find extent for this block */
4128         path = ext4_ext_find_extent(inode, map->m_lblk, NULL, 0);
4129         if (IS_ERR(path)) {
4130                 err = PTR_ERR(path);
4131                 path = NULL;
4132                 goto out2;
4133         }
4134
4135         depth = ext_depth(inode);
4136
4137         /*
4138          * consistent leaf must not be empty;
4139          * this situation is possible, though, _during_ tree modification;
4140          * this is why assert can't be put in ext4_ext_find_extent()
4141          */
4142         if (unlikely(path[depth].p_ext == NULL && depth != 0)) {
4143                 EXT4_ERROR_INODE(inode, "bad extent address "
4144                                  "lblock: %lu, depth: %d pblock %lld",
4145                                  (unsigned long) map->m_lblk, depth,
4146                                  path[depth].p_block);
4147                 err = -EIO;
4148                 goto out2;
4149         }
4150
4151         ex = path[depth].p_ext;
4152         if (ex) {
4153                 ext4_lblk_t ee_block = le32_to_cpu(ex->ee_block);
4154                 ext4_fsblk_t ee_start = ext4_ext_pblock(ex);
4155                 unsigned short ee_len;
4156
4157                 /*
4158                  * Uninitialized extents are treated as holes, except that
4159                  * we split out initialized portions during a write.
4160                  */
4161                 ee_len = ext4_ext_get_actual_len(ex);
4162
4163                 trace_ext4_ext_show_extent(inode, ee_block, ee_start, ee_len);
4164
4165                 /* if found extent covers block, simply return it */
4166                 if (in_range(map->m_lblk, ee_block, ee_len)) {
4167                         newblock = map->m_lblk - ee_block + ee_start;
4168                         /* number of remaining blocks in the extent */
4169                         allocated = ee_len - (map->m_lblk - ee_block);
4170                         ext_debug("%u fit into %u:%d -> %llu\n", map->m_lblk,
4171                                   ee_block, ee_len, newblock);
4172
4173                         if (!ext4_ext_is_uninitialized(ex))
4174                                 goto out;
4175
4176                         allocated = ext4_ext_handle_uninitialized_extents(
4177                                 handle, inode, map, path, flags,
4178                                 allocated, newblock);
4179                         goto out3;
4180                 }
4181         }
4182
4183         if ((sbi->s_cluster_ratio > 1) &&
4184             ext4_find_delalloc_cluster(inode, map->m_lblk))
4185                 map->m_flags |= EXT4_MAP_FROM_CLUSTER;
4186
4187         /*
4188          * requested block isn't allocated yet;
4189          * we couldn't try to create block if create flag is zero
4190          */
4191         if ((flags & EXT4_GET_BLOCKS_CREATE) == 0) {
4192                 /*
4193                  * put just found gap into cache to speed up
4194                  * subsequent requests
4195                  */
4196                 if ((flags & EXT4_GET_BLOCKS_NO_PUT_HOLE) == 0)
4197                         ext4_ext_put_gap_in_cache(inode, path, map->m_lblk);
4198                 goto out2;
4199         }
4200
4201         /*
4202          * Okay, we need to do block allocation.
4203          */
4204         map->m_flags &= ~EXT4_MAP_FROM_CLUSTER;
4205         newex.ee_block = cpu_to_le32(map->m_lblk);
4206         cluster_offset = map->m_lblk & (sbi->s_cluster_ratio-1);
4207
4208         /*
4209          * If we are doing bigalloc, check to see if the extent returned
4210          * by ext4_ext_find_extent() implies a cluster we can use.
4211          */
4212         if (cluster_offset && ex &&
4213             get_implied_cluster_alloc(inode->i_sb, map, ex, path)) {
4214                 ar.len = allocated = map->m_len;
4215                 newblock = map->m_pblk;
4216                 map->m_flags |= EXT4_MAP_FROM_CLUSTER;
4217                 goto got_allocated_blocks;
4218         }
4219
4220         /* find neighbour allocated blocks */
4221         ar.lleft = map->m_lblk;
4222         err = ext4_ext_search_left(inode, path, &ar.lleft, &ar.pleft);
4223         if (err)
4224                 goto out2;
4225         ar.lright = map->m_lblk;
4226         ex2 = NULL;
4227         err = ext4_ext_search_right(inode, path, &ar.lright, &ar.pright, &ex2);
4228         if (err)
4229                 goto out2;
4230
4231         /* Check if the extent after searching to the right implies a
4232          * cluster we can use. */
4233         if ((sbi->s_cluster_ratio > 1) && ex2 &&
4234             get_implied_cluster_alloc(inode->i_sb, map, ex2, path)) {
4235                 ar.len = allocated = map->m_len;
4236                 newblock = map->m_pblk;
4237                 map->m_flags |= EXT4_MAP_FROM_CLUSTER;
4238                 goto got_allocated_blocks;
4239         }
4240
4241         /*
4242          * See if request is beyond maximum number of blocks we can have in
4243          * a single extent. For an initialized extent this limit is
4244          * EXT_INIT_MAX_LEN and for an uninitialized extent this limit is
4245          * EXT_UNINIT_MAX_LEN.
4246          */
4247         if (map->m_len > EXT_INIT_MAX_LEN &&
4248             !(flags & EXT4_GET_BLOCKS_UNINIT_EXT))
4249                 map->m_len = EXT_INIT_MAX_LEN;
4250         else if (map->m_len > EXT_UNINIT_MAX_LEN &&
4251                  (flags & EXT4_GET_BLOCKS_UNINIT_EXT))
4252                 map->m_len = EXT_UNINIT_MAX_LEN;
4253
4254         /* Check if we can really insert (m_lblk)::(m_lblk + m_len) extent */
4255         newex.ee_len = cpu_to_le16(map->m_len);
4256         err = ext4_ext_check_overlap(sbi, inode, &newex, path);
4257         if (err)
4258                 allocated = ext4_ext_get_actual_len(&newex);
4259         else
4260                 allocated = map->m_len;
4261
4262         /* allocate new block */
4263         ar.inode = inode;
4264         ar.goal = ext4_ext_find_goal(inode, path, map->m_lblk);
4265         ar.logical = map->m_lblk;
4266         /*
4267          * We calculate the offset from the beginning of the cluster
4268          * for the logical block number, since when we allocate a
4269          * physical cluster, the physical block should start at the
4270          * same offset from the beginning of the cluster.  This is
4271          * needed so that future calls to get_implied_cluster_alloc()
4272          * work correctly.
4273          */
4274         offset = map->m_lblk & (sbi->s_cluster_ratio - 1);
4275         ar.len = EXT4_NUM_B2C(sbi, offset+allocated);
4276         ar.goal -= offset;
4277         ar.logical -= offset;
4278         if (S_ISREG(inode->i_mode))
4279                 ar.flags = EXT4_MB_HINT_DATA;
4280         else
4281                 /* disable in-core preallocation for non-regular files */
4282                 ar.flags = 0;
4283         if (flags & EXT4_GET_BLOCKS_NO_NORMALIZE)
4284                 ar.flags |= EXT4_MB_HINT_NOPREALLOC;
4285         newblock = ext4_mb_new_blocks(handle, &ar, &err);
4286         if (!newblock)
4287                 goto out2;
4288         ext_debug("allocate new block: goal %llu, found %llu/%u\n",
4289                   ar.goal, newblock, allocated);
4290         free_on_err = 1;
4291         allocated_clusters = ar.len;
4292         ar.len = EXT4_C2B(sbi, ar.len) - offset;
4293         if (ar.len > allocated)
4294                 ar.len = allocated;
4295
4296 got_allocated_blocks:
4297         /* try to insert new extent into found leaf and return */
4298         ext4_ext_store_pblock(&newex, newblock + offset);
4299         newex.ee_len = cpu_to_le16(ar.len);
4300         /* Mark uninitialized */
4301         if (flags & EXT4_GET_BLOCKS_UNINIT_EXT){
4302                 ext4_ext_mark_uninitialized(&newex);
4303                 map->m_flags |= EXT4_MAP_UNWRITTEN;
4304                 /*
4305                  * io_end structure was created for every IO write to an
4306                  * uninitialized extent. To avoid unnecessary conversion,
4307                  * here we flag the IO that really needs the conversion.
4308                  * For non asycn direct IO case, flag the inode state
4309                  * that we need to perform conversion when IO is done.
4310                  */
4311                 if ((flags & EXT4_GET_BLOCKS_PRE_IO))
4312                         set_unwritten = 1;
4313                 if (ext4_should_dioread_nolock(inode))
4314                         map->m_flags |= EXT4_MAP_UNINIT;
4315         }
4316
4317         err = 0;
4318         if ((flags & EXT4_GET_BLOCKS_KEEP_SIZE) == 0)
4319                 err = check_eofblocks_fl(handle, inode, map->m_lblk,
4320                                          path, ar.len);
4321         if (!err)
4322                 err = ext4_ext_insert_extent(handle, inode, path,
4323                                              &newex, flags);
4324
4325         if (!err && set_unwritten) {
4326                 if (io)
4327                         ext4_set_io_unwritten_flag(inode, io);
4328                 else
4329                         ext4_set_inode_state(inode,
4330                                              EXT4_STATE_DIO_UNWRITTEN);
4331         }
4332
4333         if (err && free_on_err) {
4334                 int fb_flags = flags & EXT4_GET_BLOCKS_DELALLOC_RESERVE ?
4335                         EXT4_FREE_BLOCKS_NO_QUOT_UPDATE : 0;
4336                 /* free data blocks we just allocated */
4337                 /* not a good idea to call discard here directly,
4338                  * but otherwise we'd need to call it every free() */
4339                 ext4_discard_preallocations(inode);
4340                 ext4_free_blocks(handle, inode, NULL, newblock,
4341                                  EXT4_C2B(sbi, allocated_clusters), fb_flags);
4342                 goto out2;
4343         }
4344
4345         /* previous routine could use block we allocated */
4346         newblock = ext4_ext_pblock(&newex);
4347         allocated = ext4_ext_get_actual_len(&newex);
4348         if (allocated > map->m_len)
4349                 allocated = map->m_len;
4350         map->m_flags |= EXT4_MAP_NEW;
4351
4352         /*
4353          * Update reserved blocks/metadata blocks after successful
4354          * block allocation which had been deferred till now.
4355          */
4356         if (flags & EXT4_GET_BLOCKS_DELALLOC_RESERVE) {
4357                 unsigned int reserved_clusters;
4358                 /*
4359                  * Check how many clusters we had reserved this allocated range
4360                  */
4361                 reserved_clusters = get_reserved_cluster_alloc(inode,
4362                                                 map->m_lblk, allocated);
4363                 if (map->m_flags & EXT4_MAP_FROM_CLUSTER) {
4364                         if (reserved_clusters) {
4365                                 /*
4366                                  * We have clusters reserved for this range.
4367                                  * But since we are not doing actual allocation
4368                                  * and are simply using blocks from previously
4369                                  * allocated cluster, we should release the
4370                                  * reservation and not claim quota.
4371                                  */
4372                                 ext4_da_update_reserve_space(inode,
4373                                                 reserved_clusters, 0);
4374                         }
4375                 } else {
4376                         BUG_ON(allocated_clusters < reserved_clusters);
4377                         if (reserved_clusters < allocated_clusters) {
4378                                 struct ext4_inode_info *ei = EXT4_I(inode);
4379                                 int reservation = allocated_clusters -
4380                                                   reserved_clusters;
4381                                 /*
4382                                  * It seems we claimed few clusters outside of
4383                                  * the range of this allocation. We should give
4384                                  * it back to the reservation pool. This can
4385                                  * happen in the following case:
4386                                  *
4387                                  * * Suppose s_cluster_ratio is 4 (i.e., each
4388                                  *   cluster has 4 blocks. Thus, the clusters
4389                                  *   are [0-3],[4-7],[8-11]...
4390                                  * * First comes delayed allocation write for
4391                                  *   logical blocks 10 & 11. Since there were no
4392                                  *   previous delayed allocated blocks in the
4393                                  *   range [8-11], we would reserve 1 cluster
4394                                  *   for this write.
4395                                  * * Next comes write for logical blocks 3 to 8.
4396                                  *   In this case, we will reserve 2 clusters
4397                                  *   (for [0-3] and [4-7]; and not for [8-11] as
4398                                  *   that range has a delayed allocated blocks.
4399                                  *   Thus total reserved clusters now becomes 3.
4400                                  * * Now, during the delayed allocation writeout
4401                                  *   time, we will first write blocks [3-8] and
4402                                  *   allocate 3 clusters for writing these
4403                                  *   blocks. Also, we would claim all these
4404                                  *   three clusters above.
4405                                  * * Now when we come here to writeout the
4406                                  *   blocks [10-11], we would expect to claim
4407                                  *   the reservation of 1 cluster we had made
4408                                  *   (and we would claim it since there are no
4409                                  *   more delayed allocated blocks in the range
4410                                  *   [8-11]. But our reserved cluster count had
4411                                  *   already gone to 0.
4412                                  *
4413                                  *   Thus, at the step 4 above when we determine
4414                                  *   that there are still some unwritten delayed
4415                                  *   allocated blocks outside of our current
4416                                  *   block range, we should increment the
4417                                  *   reserved clusters count so that when the
4418                                  *   remaining blocks finally gets written, we
4419                                  *   could claim them.
4420                                  */
4421                                 dquot_reserve_block(inode,
4422                                                 EXT4_C2B(sbi, reservation));
4423                                 spin_lock(&ei->i_block_reservation_lock);
4424                                 ei->i_reserved_data_blocks += reservation;
4425                                 spin_unlock(&ei->i_block_reservation_lock);
4426                         }
4427                         /*
4428                          * We will claim quota for all newly allocated blocks.
4429                          * We're updating the reserved space *after* the
4430                          * correction above so we do not accidentally free
4431                          * all the metadata reservation because we might
4432                          * actually need it later on.
4433                          */
4434                         ext4_da_update_reserve_space(inode, allocated_clusters,
4435                                                         1);
4436                 }
4437         }
4438
4439         /*
4440          * Cache the extent and update transaction to commit on fdatasync only
4441          * when it is _not_ an uninitialized extent.
4442          */
4443         if ((flags & EXT4_GET_BLOCKS_UNINIT_EXT) == 0)
4444                 ext4_update_inode_fsync_trans(handle, inode, 1);
4445         else
4446                 ext4_update_inode_fsync_trans(handle, inode, 0);
4447 out:
4448         if (allocated > map->m_len)
4449                 allocated = map->m_len;
4450         ext4_ext_show_leaf(inode, path);
4451         map->m_flags |= EXT4_MAP_MAPPED;
4452         map->m_pblk = newblock;
4453         map->m_len = allocated;
4454 out2:
4455         if (path) {
4456                 ext4_ext_drop_refs(path);
4457                 kfree(path);
4458         }
4459
4460 out3:
4461         trace_ext4_ext_map_blocks_exit(inode, flags, map,
4462                                        err ? err : allocated);
4463         ext4_es_lru_add(inode);
4464         return err ? err : allocated;
4465 }
4466
4467 void ext4_ext_truncate(handle_t *handle, struct inode *inode)
4468 {
4469         struct super_block *sb = inode->i_sb;
4470         ext4_lblk_t last_block;
4471         int err = 0;
4472
4473         /*
4474          * TODO: optimization is possible here.
4475          * Probably we need not scan at all,
4476          * because page truncation is enough.
4477          */
4478
4479         /* we have to know where to truncate from in crash case */
4480         EXT4_I(inode)->i_disksize = inode->i_size;
4481         ext4_mark_inode_dirty(handle, inode);
4482
4483         last_block = (inode->i_size + sb->s_blocksize - 1)
4484                         >> EXT4_BLOCK_SIZE_BITS(sb);
4485 retry:
4486         err = ext4_es_remove_extent(inode, last_block,
4487                                     EXT_MAX_BLOCKS - last_block);
4488         if (err == -ENOMEM) {
4489                 cond_resched();
4490                 congestion_wait(BLK_RW_ASYNC, HZ/50);
4491                 goto retry;
4492         }
4493         if (err) {
4494                 ext4_std_error(inode->i_sb, err);
4495                 return;
4496         }
4497         err = ext4_ext_remove_space(inode, last_block, EXT_MAX_BLOCKS - 1);
4498         ext4_std_error(inode->i_sb, err);
4499 }
4500
4501 static void ext4_falloc_update_inode(struct inode *inode,
4502                                 int mode, loff_t new_size, int update_ctime)
4503 {
4504         struct timespec now;
4505
4506         if (update_ctime) {
4507                 now = current_fs_time(inode->i_sb);
4508                 if (!timespec_equal(&inode->i_ctime, &now))
4509                         inode->i_ctime = now;
4510         }
4511         /*
4512          * Update only when preallocation was requested beyond
4513          * the file size.
4514          */
4515         if (!(mode & FALLOC_FL_KEEP_SIZE)) {
4516                 if (new_size > i_size_read(inode))
4517                         i_size_write(inode, new_size);
4518                 if (new_size > EXT4_I(inode)->i_disksize)
4519                         ext4_update_i_disksize(inode, new_size);
4520         } else {
4521                 /*
4522                  * Mark that we allocate beyond EOF so the subsequent truncate
4523                  * can proceed even if the new size is the same as i_size.
4524                  */
4525                 if (new_size > i_size_read(inode))
4526                         ext4_set_inode_flag(inode, EXT4_INODE_EOFBLOCKS);
4527         }
4528
4529 }
4530
4531 /*
4532  * preallocate space for a file. This implements ext4's fallocate file
4533  * operation, which gets called from sys_fallocate system call.
4534  * For block-mapped files, posix_fallocate should fall back to the method
4535  * of writing zeroes to the required new blocks (the same behavior which is
4536  * expected for file systems which do not support fallocate() system call).
4537  */
4538 long ext4_fallocate(struct file *file, int mode, loff_t offset, loff_t len)
4539 {
4540         struct inode *inode = file_inode(file);
4541         handle_t *handle;
4542         loff_t new_size;
4543         unsigned int max_blocks;
4544         int ret = 0;
4545         int ret2 = 0;
4546         int retries = 0;
4547         int flags;
4548         struct ext4_map_blocks map;
4549         unsigned int credits, blkbits = inode->i_blkbits;
4550
4551         /* Return error if mode is not supported */
4552         if (mode & ~(FALLOC_FL_KEEP_SIZE | FALLOC_FL_PUNCH_HOLE))
4553                 return -EOPNOTSUPP;
4554
4555         if (mode & FALLOC_FL_PUNCH_HOLE)
4556                 return ext4_punch_hole(inode, offset, len);
4557
4558         ret = ext4_convert_inline_data(inode);
4559         if (ret)
4560                 return ret;
4561
4562         /*
4563          * currently supporting (pre)allocate mode for extent-based
4564          * files _only_
4565          */
4566         if (!(ext4_test_inode_flag(inode, EXT4_INODE_EXTENTS)))
4567                 return -EOPNOTSUPP;
4568
4569         trace_ext4_fallocate_enter(inode, offset, len, mode);
4570         map.m_lblk = offset >> blkbits;
4571         /*
4572          * We can't just convert len to max_blocks because
4573          * If blocksize = 4096 offset = 3072 and len = 2048
4574          */
4575         max_blocks = (EXT4_BLOCK_ALIGN(len + offset, blkbits) >> blkbits)
4576                 - map.m_lblk;
4577         /*
4578          * credits to insert 1 extent into extent tree
4579          */
4580         credits = ext4_chunk_trans_blocks(inode, max_blocks);
4581         mutex_lock(&inode->i_mutex);
4582         ret = inode_newsize_ok(inode, (len + offset));
4583         if (ret) {
4584                 mutex_unlock(&inode->i_mutex);
4585                 trace_ext4_fallocate_exit(inode, offset, max_blocks, ret);
4586                 return ret;
4587         }
4588         flags = EXT4_GET_BLOCKS_CREATE_UNINIT_EXT;
4589         if (mode & FALLOC_FL_KEEP_SIZE)
4590                 flags |= EXT4_GET_BLOCKS_KEEP_SIZE;
4591         /*
4592          * Don't normalize the request if it can fit in one extent so
4593          * that it doesn't get unnecessarily split into multiple
4594          * extents.
4595          */
4596         if (len <= EXT_UNINIT_MAX_LEN << blkbits)
4597                 flags |= EXT4_GET_BLOCKS_NO_NORMALIZE;
4598
4599 retry:
4600         while (ret >= 0 && ret < max_blocks) {
4601                 map.m_lblk = map.m_lblk + ret;
4602                 map.m_len = max_blocks = max_blocks - ret;
4603                 handle = ext4_journal_start(inode, EXT4_HT_MAP_BLOCKS,
4604                                             credits);
4605                 if (IS_ERR(handle)) {
4606                         ret = PTR_ERR(handle);
4607                         break;
4608                 }
4609                 ret = ext4_map_blocks(handle, inode, &map, flags);
4610                 if (ret <= 0) {
4611 #ifdef EXT4FS_DEBUG
4612                         ext4_warning(inode->i_sb,
4613                                      "inode #%lu: block %u: len %u: "
4614                                      "ext4_ext_map_blocks returned %d",
4615                                      inode->i_ino, map.m_lblk,
4616                                      map.m_len, ret);
4617 #endif
4618                         ext4_mark_inode_dirty(handle, inode);
4619                         ret2 = ext4_journal_stop(handle);
4620                         break;
4621                 }
4622                 if ((map.m_lblk + ret) >= (EXT4_BLOCK_ALIGN(offset + len,
4623                                                 blkbits) >> blkbits))
4624                         new_size = offset + len;
4625                 else
4626                         new_size = ((loff_t) map.m_lblk + ret) << blkbits;
4627
4628                 ext4_falloc_update_inode(inode, mode, new_size,
4629                                          (map.m_flags & EXT4_MAP_NEW));
4630                 ext4_mark_inode_dirty(handle, inode);
4631                 if ((file->f_flags & O_SYNC) && ret >= max_blocks)
4632                         ext4_handle_sync(handle);
4633                 ret2 = ext4_journal_stop(handle);
4634                 if (ret2)
4635                         break;
4636         }
4637         if (ret == -ENOSPC &&
4638                         ext4_should_retry_alloc(inode->i_sb, &retries)) {
4639                 ret = 0;
4640                 goto retry;
4641         }
4642         mutex_unlock(&inode->i_mutex);
4643         trace_ext4_fallocate_exit(inode, offset, max_blocks,
4644                                 ret > 0 ? ret2 : ret);
4645         return ret > 0 ? ret2 : ret;
4646 }
4647
4648 /*
4649  * This function convert a range of blocks to written extents
4650  * The caller of this function will pass the start offset and the size.
4651  * all unwritten extents within this range will be converted to
4652  * written extents.
4653  *
4654  * This function is called from the direct IO end io call back
4655  * function, to convert the fallocated extents after IO is completed.
4656  * Returns 0 on success.
4657  */
4658 int ext4_convert_unwritten_extents(handle_t *handle, struct inode *inode,
4659                                    loff_t offset, ssize_t len)
4660 {
4661         unsigned int max_blocks;
4662         int ret = 0;
4663         int ret2 = 0;
4664         struct ext4_map_blocks map;
4665         unsigned int credits, blkbits = inode->i_blkbits;
4666
4667         map.m_lblk = offset >> blkbits;
4668         /*
4669          * We can't just convert len to max_blocks because
4670          * If blocksize = 4096 offset = 3072 and len = 2048
4671          */
4672         max_blocks = ((EXT4_BLOCK_ALIGN(len + offset, blkbits) >> blkbits) -
4673                       map.m_lblk);
4674         /*
4675          * This is somewhat ugly but the idea is clear: When transaction is
4676          * reserved, everything goes into it. Otherwise we rather start several
4677          * smaller transactions for conversion of each extent separately.
4678          */
4679         if (handle) {
4680                 handle = ext4_journal_start_reserved(handle,
4681                                                      EXT4_HT_EXT_CONVERT);
4682                 if (IS_ERR(handle))
4683                         return PTR_ERR(handle);
4684                 credits = 0;
4685         } else {
4686                 /*
4687                  * credits to insert 1 extent into extent tree
4688                  */
4689                 credits = ext4_chunk_trans_blocks(inode, max_blocks);
4690         }
4691         while (ret >= 0 && ret < max_blocks) {
4692                 map.m_lblk += ret;
4693                 map.m_len = (max_blocks -= ret);
4694                 if (credits) {
4695                         handle = ext4_journal_start(inode, EXT4_HT_MAP_BLOCKS,
4696                                                     credits);
4697                         if (IS_ERR(handle)) {
4698                                 ret = PTR_ERR(handle);
4699                                 break;
4700                         }
4701                 }
4702                 ret = ext4_map_blocks(handle, inode, &map,
4703                                       EXT4_GET_BLOCKS_IO_CONVERT_EXT);
4704                 if (ret <= 0)
4705                         ext4_warning(inode->i_sb,
4706                                      "inode #%lu: block %u: len %u: "
4707                                      "ext4_ext_map_blocks returned %d",
4708                                      inode->i_ino, map.m_lblk,
4709                                      map.m_len, ret);
4710                 ext4_mark_inode_dirty(handle, inode);
4711                 if (credits)
4712                         ret2 = ext4_journal_stop(handle);
4713                 if (ret <= 0 || ret2)
4714                         break;
4715         }
4716         if (!credits)
4717                 ret2 = ext4_journal_stop(handle);
4718         return ret > 0 ? ret2 : ret;
4719 }
4720
4721 /*
4722  * If newes is not existing extent (newes->ec_pblk equals zero) find
4723  * delayed extent at start of newes and update newes accordingly and
4724  * return start of the next delayed extent.
4725  *
4726  * If newes is existing extent (newes->ec_pblk is not equal zero)
4727  * return start of next delayed extent or EXT_MAX_BLOCKS if no delayed
4728  * extent found. Leave newes unmodified.
4729  */
4730 static int ext4_find_delayed_extent(struct inode *inode,
4731                                     struct extent_status *newes)
4732 {
4733         struct extent_status es;
4734         ext4_lblk_t block, next_del;
4735
4736         if (newes->es_pblk == 0) {
4737                 ext4_es_find_delayed_extent_range(inode, newes->es_lblk,
4738                                 newes->es_lblk + newes->es_len - 1, &es);
4739
4740                 /*
4741                  * No extent in extent-tree contains block @newes->es_pblk,
4742                  * then the block may stay in 1)a hole or 2)delayed-extent.
4743                  */
4744                 if (es.es_len == 0)
4745                         /* A hole found. */
4746                         return 0;
4747
4748                 if (es.es_lblk > newes->es_lblk) {
4749                         /* A hole found. */
4750                         newes->es_len = min(es.es_lblk - newes->es_lblk,
4751                                             newes->es_len);
4752                         return 0;
4753                 }
4754
4755                 newes->es_len = es.es_lblk + es.es_len - newes->es_lblk;
4756         }
4757
4758         block = newes->es_lblk + newes->es_len;
4759         ext4_es_find_delayed_extent_range(inode, block, EXT_MAX_BLOCKS, &es);
4760         if (es.es_len == 0)
4761                 next_del = EXT_MAX_BLOCKS;
4762         else
4763                 next_del = es.es_lblk;
4764
4765         return next_del;
4766 }
4767 /* fiemap flags we can handle specified here */
4768 #define EXT4_FIEMAP_FLAGS       (FIEMAP_FLAG_SYNC|FIEMAP_FLAG_XATTR)
4769
4770 static int ext4_xattr_fiemap(struct inode *inode,
4771                                 struct fiemap_extent_info *fieinfo)
4772 {
4773         __u64 physical = 0;
4774         __u64 length;
4775         __u32 flags = FIEMAP_EXTENT_LAST;
4776         int blockbits = inode->i_sb->s_blocksize_bits;
4777         int error = 0;
4778
4779         /* in-inode? */
4780         if (ext4_test_inode_state(inode, EXT4_STATE_XATTR)) {
4781                 struct ext4_iloc iloc;
4782                 int offset;     /* offset of xattr in inode */
4783
4784                 error = ext4_get_inode_loc(inode, &iloc);
4785                 if (error)
4786                         return error;
4787                 physical = (__u64)iloc.bh->b_blocknr << blockbits;
4788                 offset = EXT4_GOOD_OLD_INODE_SIZE +
4789                                 EXT4_I(inode)->i_extra_isize;
4790                 physical += offset;
4791                 length = EXT4_SB(inode->i_sb)->s_inode_size - offset;
4792                 flags |= FIEMAP_EXTENT_DATA_INLINE;
4793                 brelse(iloc.bh);
4794         } else { /* external block */
4795                 physical = (__u64)EXT4_I(inode)->i_file_acl << blockbits;
4796                 length = inode->i_sb->s_blocksize;
4797         }
4798
4799         if (physical)
4800                 error = fiemap_fill_next_extent(fieinfo, 0, physical,
4801                                                 length, flags);
4802         return (error < 0 ? error : 0);
4803 }
4804
4805 int ext4_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
4806                 __u64 start, __u64 len)
4807 {
4808         ext4_lblk_t start_blk;
4809         int error = 0;
4810
4811         if (ext4_has_inline_data(inode)) {
4812                 int has_inline = 1;
4813
4814                 error = ext4_inline_data_fiemap(inode, fieinfo, &has_inline);
4815
4816                 if (has_inline)
4817                         return error;
4818         }
4819
4820         if (fieinfo->fi_flags & FIEMAP_FLAG_CACHE) {
4821                 error = ext4_ext_precache(inode);
4822                 if (error)
4823                         return error;
4824         }
4825
4826         /* fallback to generic here if not in extents fmt */
4827         if (!(ext4_test_inode_flag(inode, EXT4_INODE_EXTENTS)))
4828                 return generic_block_fiemap(inode, fieinfo, start, len,
4829                         ext4_get_block);
4830
4831         if (fiemap_check_flags(fieinfo, EXT4_FIEMAP_FLAGS))
4832                 return -EBADR;
4833
4834         if (fieinfo->fi_flags & FIEMAP_FLAG_XATTR) {
4835                 error = ext4_xattr_fiemap(inode, fieinfo);
4836         } else {
4837                 ext4_lblk_t len_blks;
4838                 __u64 last_blk;
4839
4840                 start_blk = start >> inode->i_sb->s_blocksize_bits;
4841                 last_blk = (start + len - 1) >> inode->i_sb->s_blocksize_bits;
4842                 if (last_blk >= EXT_MAX_BLOCKS)
4843                         last_blk = EXT_MAX_BLOCKS-1;
4844                 len_blks = ((ext4_lblk_t) last_blk) - start_blk + 1;
4845
4846                 /*
4847                  * Walk the extent tree gathering extent information
4848                  * and pushing extents back to the user.
4849                  */
4850                 error = ext4_fill_fiemap_extents(inode, start_blk,
4851                                                  len_blks, fieinfo);
4852         }
4853         ext4_es_lru_add(inode);
4854         return error;
4855 }