1 /*
2 * rehash.c --- rebuild hash tree directories
3 *
4 * Copyright (C) 2002 Theodore Ts'o
5 *
6 * %Begin-Header%
7 * This file may be redistributed under the terms of the GNU Public
8 * License.
9 * %End-Header%
10 *
11 * This algorithm is designed for simplicity of implementation and to
12 * pack the directory as much as possible. It however requires twice
13 * as much memory as the size of the directory. The maximum size
14 * directory supported using a 4k blocksize is roughly a gigabyte, and
15 * so there may very well be problems with machines that don't have
16 * virtual memory, and obscenely large directories.
17 *
18 * An alternate algorithm which is much more disk intensive could be
19 * written, and probably will need to be written in the future. The
20 * design goals of such an algorithm are: (a) use (roughly) constant
21 * amounts of memory, no matter how large the directory, (b) the
22 * directory must be safe at all times, even if e2fsck is interrupted
23 * in the middle, (c) we must use minimal amounts of extra disk
24 * blocks. This pretty much requires an incremental approach, where
25 * we are reading from one part of the directory, and inserting into
26 * the front half. So the algorithm will have to keep track of a
27 * moving block boundary between the new tree and the old tree, and
28 * files will need to be moved from the old directory and inserted
29 * into the new tree. If the new directory requires space which isn't
30 * yet available, blocks from the beginning part of the old directory
31 * may need to be moved to the end of the directory to make room for
32 * the new tree:
33 *
34 * --------------------------------------------------------
35 * | new tree | | old tree |
36 * --------------------------------------------------------
37 * ^ ptr ^ptr
38 * tail new head old
39 *
40 * This is going to be a pain in the tuckus to implement, and will
41 * require a lot more disk accesses. So I'm going to skip it for now;
42 * it's only really going to be an issue for really, really big
43 * filesystems (when we reach the level of tens of millions of files
44 * in a single directory). It will probably be easier to simply
45 * require that e2fsck use VM first.
46 */
47
48 #include "config.h"
49 #include <string.h>
50 #include <ctype.h>
51 #include <errno.h>
52 #include "e2fsck.h"
53 #include "problem.h"
54 #include "support/sort_r.h"
55
56 /* Schedule a dir to be rebuilt during pass 3A. */
e2fsck_rehash_dir_later(e2fsck_t ctx,ext2_ino_t ino)57 void e2fsck_rehash_dir_later(e2fsck_t ctx, ext2_ino_t ino)
58 {
59 if (!ctx->dirs_to_hash)
60 ext2fs_u32_list_create(&ctx->dirs_to_hash, 50);
61 if (ctx->dirs_to_hash)
62 ext2fs_u32_list_add(ctx->dirs_to_hash, ino);
63 }
64
65 /* Ask if a dir will be rebuilt during pass 3A. */
e2fsck_dir_will_be_rehashed(e2fsck_t ctx,ext2_ino_t ino)66 int e2fsck_dir_will_be_rehashed(e2fsck_t ctx, ext2_ino_t ino)
67 {
68 if (ctx->options & E2F_OPT_COMPRESS_DIRS)
69 return 1;
70 if (!ctx->dirs_to_hash)
71 return 0;
72 return ext2fs_u32_list_test(ctx->dirs_to_hash, ino);
73 }
74
75 #undef REHASH_DEBUG
76
77 struct fill_dir_struct {
78 char *buf;
79 struct ext2_inode *inode;
80 ext2_ino_t ino;
81 errcode_t err;
82 e2fsck_t ctx;
83 struct hash_entry *harray;
84 blk_t max_array, num_array;
85 ext2_off64_t dir_size;
86 int compress;
87 ext2_ino_t parent;
88 ext2_ino_t dir;
89 };
90
91 struct hash_entry {
92 ext2_dirhash_t hash;
93 ext2_dirhash_t minor_hash;
94 ext2_ino_t ino;
95 struct ext2_dir_entry *dir;
96 };
97
98 struct out_dir {
99 blk_t num;
100 blk_t max;
101 char *buf;
102 ext2_dirhash_t *hashes;
103 };
104
105 #define DOTDOT_OFFSET 12
106
is_fake_entry(ext2_filsys fs,int lblk,unsigned int offset)107 static int is_fake_entry(ext2_filsys fs, int lblk, unsigned int offset)
108 {
109 /* Entries in the first block before this value refer to . or .. */
110 if (lblk == 0 && offset <= DOTDOT_OFFSET)
111 return 1;
112 /* Check if this is likely the csum entry */
113 if (ext2fs_has_feature_metadata_csum(fs->super) &&
114 (offset & (fs->blocksize - 1)) ==
115 fs->blocksize - sizeof(struct ext2_dir_entry_tail))
116 return 1;
117 return 0;
118 }
119
fill_dir_block(ext2_filsys fs,blk64_t * block_nr,e2_blkcnt_t blockcnt,blk64_t ref_block EXT2FS_ATTR ((unused)),int ref_offset EXT2FS_ATTR ((unused)),void * priv_data)120 static int fill_dir_block(ext2_filsys fs,
121 blk64_t *block_nr,
122 e2_blkcnt_t blockcnt,
123 blk64_t ref_block EXT2FS_ATTR((unused)),
124 int ref_offset EXT2FS_ATTR((unused)),
125 void *priv_data)
126 {
127 struct fill_dir_struct *fd = (struct fill_dir_struct *) priv_data;
128 struct hash_entry *ent;
129 struct ext2_dir_entry *dirent;
130 char *dir;
131 unsigned int offset, dir_offset, rec_len, name_len;
132 int hash_alg, hash_flags, hash_in_entry;
133
134 if (blockcnt < 0)
135 return 0;
136
137 offset = blockcnt * fs->blocksize;
138 if (offset + fs->blocksize > fd->inode->i_size) {
139 fd->err = EXT2_ET_DIR_CORRUPTED;
140 return BLOCK_ABORT;
141 }
142
143 dir = (fd->buf+offset);
144 if (*block_nr == 0) {
145 memset(dir, 0, fs->blocksize);
146 dirent = (struct ext2_dir_entry *) dir;
147 (void) ext2fs_set_rec_len(fs, fs->blocksize, dirent);
148 } else {
149 int flags = fs->flags;
150 fs->flags |= EXT2_FLAG_IGNORE_CSUM_ERRORS;
151 fd->err = ext2fs_read_dir_block4(fs, *block_nr, dir, 0,
152 fd->dir);
153 fs->flags = (flags & EXT2_FLAG_IGNORE_CSUM_ERRORS) |
154 (fs->flags & ~EXT2_FLAG_IGNORE_CSUM_ERRORS);
155 if (fd->err)
156 return BLOCK_ABORT;
157 }
158 hash_flags = fd->inode->i_flags & EXT4_CASEFOLD_FL;
159 hash_in_entry = ext4_hash_in_dirent(fd->inode);
160 hash_alg = fs->super->s_def_hash_version;
161 if ((hash_alg <= EXT2_HASH_TEA) &&
162 (fs->super->s_flags & EXT2_FLAGS_UNSIGNED_HASH))
163 hash_alg += 3;
164 /* While the directory block is "hot", index it. */
165 dir_offset = 0;
166 while (dir_offset < fs->blocksize) {
167 unsigned int min_rec = EXT2_DIR_ENTRY_HEADER_LEN;
168 int extended = hash_in_entry && !is_fake_entry(fs, blockcnt, dir_offset);
169
170 if (extended)
171 min_rec += EXT2_DIR_ENTRY_HASH_LEN;
172 dirent = (struct ext2_dir_entry *) (dir + dir_offset);
173 (void) ext2fs_get_rec_len(fs, dirent, &rec_len);
174 name_len = ext2fs_dirent_name_len(dirent);
175 if (((dir_offset + rec_len) > fs->blocksize) ||
176 (rec_len < min_rec) ||
177 ((rec_len % 4) != 0) ||
178 (name_len + min_rec > rec_len)) {
179 fd->err = EXT2_ET_DIR_CORRUPTED;
180 return BLOCK_ABORT;
181 }
182 dir_offset += rec_len;
183 if (dirent->inode == 0)
184 continue;
185 if ((name_len) == 0) {
186 fd->err = EXT2_ET_DIR_CORRUPTED;
187 return BLOCK_ABORT;
188 }
189 if (!fd->compress && (name_len == 1) &&
190 (dirent->name[0] == '.'))
191 continue;
192 if (!fd->compress && (name_len == 2) &&
193 (dirent->name[0] == '.') && (dirent->name[1] == '.')) {
194 fd->parent = dirent->inode;
195 continue;
196 }
197 if (fd->num_array >= fd->max_array) {
198 errcode_t retval;
199
200 retval = ext2fs_resize_array(sizeof(struct hash_entry),
201 fd->max_array,
202 fd->max_array + 500,
203 &fd->harray);
204 if (retval) {
205 fd->err = retval;
206 return BLOCK_ABORT;
207 }
208 fd->max_array += 500;
209 }
210 ent = fd->harray + fd->num_array++;
211 ent->dir = dirent;
212 fd->dir_size += ext2fs_dir_rec_len(name_len, extended);
213 ent->ino = dirent->inode;
214 if (extended) {
215 ent->hash = EXT2_DIRENT_HASH(dirent);
216 ent->minor_hash = EXT2_DIRENT_MINOR_HASH(dirent);
217 } else if (fd->compress) {
218 ent->hash = ent->minor_hash = 0;
219 } else {
220 fd->err = ext2fs_dirhash2(hash_alg,
221 dirent->name, name_len,
222 fs->encoding, hash_flags,
223 fs->super->s_hash_seed,
224 &ent->hash, &ent->minor_hash);
225 if (fd->err)
226 return BLOCK_ABORT;
227 }
228 }
229
230 return 0;
231 }
232
233 /* Used for sorting the hash entry */
ino_cmp(const void * a,const void * b)234 static EXT2_QSORT_TYPE ino_cmp(const void *a, const void *b)
235 {
236 const struct hash_entry *he_a = (const struct hash_entry *) a;
237 const struct hash_entry *he_b = (const struct hash_entry *) b;
238
239 return (he_a->ino - he_b->ino);
240 }
241
242 struct name_cmp_ctx
243 {
244 int casefold;
245 const struct ext2fs_nls_table *tbl;
246 };
247
same_name(const struct name_cmp_ctx * cmp_ctx,char * s1,int len1,char * s2,int len2)248 static int same_name(const struct name_cmp_ctx *cmp_ctx, char *s1,
249 int len1, char *s2, int len2)
250 {
251 if (!cmp_ctx->casefold)
252 return (len1 == len2 && !memcmp(s1, s2, len1));
253 else
254 return !ext2fs_casefold_cmp(cmp_ctx->tbl,
255 (unsigned char *) s1, len1,
256 (unsigned char *) s2, len2);
257 }
258
259 /* Used for sorting the hash entry */
name_cmp(const void * a,const void * b)260 static EXT2_QSORT_TYPE name_cmp(const void *a, const void *b)
261 {
262 const struct hash_entry *he_a = (const struct hash_entry *) a;
263 const struct hash_entry *he_b = (const struct hash_entry *) b;
264 unsigned int he_a_len, he_b_len, min_len;
265 int ret;
266
267 he_a_len = ext2fs_dirent_name_len(he_a->dir);
268 he_b_len = ext2fs_dirent_name_len(he_b->dir);
269 min_len = he_a_len;
270 if (min_len > he_b_len)
271 min_len = he_b_len;
272
273 ret = memcmp(he_a->dir->name, he_b->dir->name, min_len);
274 if (ret == 0) {
275 if (he_a_len > he_b_len)
276 ret = 1;
277 else if (he_a_len < he_b_len)
278 ret = -1;
279 else
280 ret = he_b->dir->inode - he_a->dir->inode;
281 }
282 return ret;
283 }
284
name_cf_cmp(const struct name_cmp_ctx * ctx,const void * a,const void * b)285 static EXT2_QSORT_TYPE name_cf_cmp(const struct name_cmp_ctx *ctx,
286 const void *a, const void *b)
287 {
288 const struct hash_entry *he_a = (const struct hash_entry *) a;
289 const struct hash_entry *he_b = (const struct hash_entry *) b;
290 unsigned int he_a_len, he_b_len;
291 int ret;
292
293 he_a_len = ext2fs_dirent_name_len(he_a->dir);
294 he_b_len = ext2fs_dirent_name_len(he_b->dir);
295
296 ret = ext2fs_casefold_cmp(ctx->tbl,
297 (unsigned char *) he_a->dir->name, he_a_len,
298 (unsigned char *) he_b->dir->name, he_b_len);
299 if (ret == 0) {
300 if (he_a_len > he_b_len)
301 ret = 1;
302 else if (he_a_len < he_b_len)
303 ret = -1;
304 else
305 ret = he_b->dir->inode - he_a->dir->inode;
306 }
307 return ret;
308 }
309
310 /* Used for sorting the hash entry */
hash_cmp(const void * a,const void * b,void * arg)311 static EXT2_QSORT_TYPE hash_cmp(const void *a, const void *b, void *arg)
312 {
313 const struct name_cmp_ctx *ctx = (struct name_cmp_ctx *) arg;
314 const struct hash_entry *he_a = (const struct hash_entry *) a;
315 const struct hash_entry *he_b = (const struct hash_entry *) b;
316 int ret;
317
318 if (he_a->hash > he_b->hash)
319 ret = 1;
320 else if (he_a->hash < he_b->hash)
321 ret = -1;
322 else {
323 if (he_a->minor_hash > he_b->minor_hash)
324 ret = 1;
325 else if (he_a->minor_hash < he_b->minor_hash)
326 ret = -1;
327 else {
328 if (ctx->casefold)
329 ret = name_cf_cmp(ctx, a, b);
330 else
331 ret = name_cmp(a, b);
332 }
333 }
334 return ret;
335 }
336
alloc_size_dir(ext2_filsys fs,struct out_dir * outdir,blk_t blocks)337 static errcode_t alloc_size_dir(ext2_filsys fs, struct out_dir *outdir,
338 blk_t blocks)
339 {
340 errcode_t retval;
341
342 if (outdir->max) {
343 retval = ext2fs_resize_array(fs->blocksize, outdir->max, blocks,
344 &outdir->buf);
345 if (retval)
346 return retval;
347 retval = ext2fs_resize_array(sizeof(ext2_dirhash_t),
348 outdir->max, blocks,
349 &outdir->hashes);
350 if (retval)
351 return retval;
352 } else {
353 retval = ext2fs_get_array(fs->blocksize, blocks, &outdir->buf);
354 if (retval)
355 return retval;
356 retval = ext2fs_get_array(sizeof(ext2_dirhash_t), blocks,
357 &outdir->hashes);
358 if (retval)
359 return retval;
360 outdir->num = 0;
361 }
362 outdir->max = blocks;
363 return 0;
364 }
365
free_out_dir(struct out_dir * outdir)366 static void free_out_dir(struct out_dir *outdir)
367 {
368 free(outdir->buf);
369 free(outdir->hashes);
370 outdir->max = 0;
371 outdir->num =0;
372 }
373
get_next_block(ext2_filsys fs,struct out_dir * outdir,char ** ret)374 static errcode_t get_next_block(ext2_filsys fs, struct out_dir *outdir,
375 char ** ret)
376 {
377 errcode_t retval;
378
379 if (outdir->num >= outdir->max) {
380 int increment = outdir->max / 10;
381
382 if (increment < 50)
383 increment = 50;
384 retval = alloc_size_dir(fs, outdir, outdir->max + increment);
385 if (retval)
386 return retval;
387 }
388 *ret = outdir->buf + (size_t)outdir->num++ * fs->blocksize;
389 memset(*ret, 0, fs->blocksize);
390 return 0;
391 }
392
393 /*
394 * This function is used to make a unique filename. We do this by
395 * appending ~0, and then incrementing the number. However, we cannot
396 * expand the length of the filename beyond the padding available in
397 * the directory entry.
398 */
mutate_name(char * str,unsigned int * len)399 static void mutate_name(char *str, unsigned int *len)
400 {
401 int i;
402 unsigned int l = *len;
403
404 /*
405 * First check to see if it looks the name has been mutated
406 * already
407 */
408 for (i = l-1; i > 0; i--) {
409 if (!isdigit(str[i]))
410 break;
411 }
412 if ((i == (int)l - 1) || (str[i] != '~')) {
413 if (((l-1) & 3) < 2)
414 l += 2;
415 else
416 l = (l+3) & ~3;
417 if (l > 255)
418 l = 255;
419 str[l-2] = '~';
420 str[l-1] = '0';
421 *len = l;
422 return;
423 }
424 for (i = l-1; i >= 0; i--) {
425 if (isdigit(str[i])) {
426 if (str[i] == '9')
427 str[i] = '0';
428 else {
429 str[i]++;
430 return;
431 }
432 continue;
433 }
434 if (i == 1) {
435 if (str[0] == 'z')
436 str[0] = 'A';
437 else if (str[0] == 'Z') {
438 str[0] = '~';
439 str[1] = '0';
440 } else
441 str[0]++;
442 } else if (i > 0) {
443 str[i] = '1';
444 str[i-1] = '~';
445 } else {
446 if (str[0] == '~')
447 str[0] = 'a';
448 else
449 str[0]++;
450 }
451 break;
452 }
453 }
454
duplicate_search_and_fix(e2fsck_t ctx,ext2_filsys fs,ext2_ino_t ino,struct fill_dir_struct * fd,const struct name_cmp_ctx * cmp_ctx)455 static int duplicate_search_and_fix(e2fsck_t ctx, ext2_filsys fs,
456 ext2_ino_t ino,
457 struct fill_dir_struct *fd,
458 const struct name_cmp_ctx *cmp_ctx)
459 {
460 struct problem_context pctx;
461 struct hash_entry *ent, *prev;
462 blk_t i, j;
463 int fixed = 0;
464 char new_name[256];
465 unsigned int new_len;
466 int hash_alg;
467 int hash_flags = fd->inode->i_flags & EXT4_CASEFOLD_FL;
468
469 clear_problem_context(&pctx);
470 pctx.ino = ino;
471
472 hash_alg = fs->super->s_def_hash_version;
473 if ((hash_alg <= EXT2_HASH_TEA) &&
474 (fs->super->s_flags & EXT2_FLAGS_UNSIGNED_HASH))
475 hash_alg += 3;
476
477 for (i=1; i < fd->num_array; i++) {
478 ent = fd->harray + i;
479 prev = ent - 1;
480 if (!ent->dir->inode ||
481 !same_name(cmp_ctx, ent->dir->name,
482 ext2fs_dirent_name_len(ent->dir),
483 prev->dir->name,
484 ext2fs_dirent_name_len(prev->dir)))
485 continue;
486 pctx.dirent = ent->dir;
487 if ((ent->dir->inode == prev->dir->inode) &&
488 fix_problem(ctx, PR_2_DUPLICATE_DIRENT, &pctx)) {
489 e2fsck_adjust_inode_count(ctx, ent->dir->inode, -1);
490 ent->dir->inode = 0;
491 fixed++;
492 continue;
493 }
494 /* Can't alter encrypted name without key, so just drop it */
495 if (fd->inode->i_flags & EXT4_ENCRYPT_FL) {
496 if (fix_problem(ctx, PR_2_NON_UNIQUE_FILE_NO_RENAME, &pctx)) {
497 e2fsck_adjust_inode_count(ctx, ent->dir->inode, -1);
498 ent->dir->inode = 0;
499 fixed++;
500 continue;
501 }
502 }
503 new_len = ext2fs_dirent_name_len(ent->dir);
504 if (new_len == 0) {
505 /* should never happen */
506 ext2fs_unmark_valid(fs);
507 continue;
508 }
509 memcpy(new_name, ent->dir->name, new_len);
510 mutate_name(new_name, &new_len);
511 for (j=0; j < fd->num_array; j++) {
512 if ((i==j) ||
513 !same_name(cmp_ctx, new_name, new_len,
514 fd->harray[j].dir->name,
515 ext2fs_dirent_name_len(fd->harray[j].dir))) {
516 continue;
517 }
518 mutate_name(new_name, &new_len);
519
520 j = -1;
521 }
522 new_name[new_len] = 0;
523 pctx.str = new_name;
524 if (fix_problem(ctx, PR_2_NON_UNIQUE_FILE, &pctx)) {
525 memcpy(ent->dir->name, new_name, new_len);
526 ext2fs_dirent_set_name_len(ent->dir, new_len);
527 ext2fs_dirhash2(hash_alg, new_name, new_len,
528 fs->encoding, hash_flags,
529 fs->super->s_hash_seed,
530 &ent->hash, &ent->minor_hash);
531 fixed++;
532 }
533 }
534 return fixed;
535 }
536
537
copy_dir_entries(e2fsck_t ctx,struct fill_dir_struct * fd,struct out_dir * outdir)538 static errcode_t copy_dir_entries(e2fsck_t ctx,
539 struct fill_dir_struct *fd,
540 struct out_dir *outdir)
541 {
542 ext2_filsys fs = ctx->fs;
543 errcode_t retval;
544 char *block_start;
545 struct hash_entry *ent;
546 struct ext2_dir_entry *dirent;
547 unsigned int rec_len, prev_rec_len, left, slack, offset;
548 blk_t i;
549 ext2_dirhash_t prev_hash;
550 int csum_size = 0;
551 struct ext2_dir_entry_tail *t;
552 int hash_in_entry = ext4_hash_in_dirent(fd->inode);
553 unsigned int min_rec_len = ext2fs_dir_rec_len(1, hash_in_entry);
554
555 if (ctx->htree_slack_percentage == 255) {
556 profile_get_uint(ctx->profile, "options",
557 "indexed_dir_slack_percentage",
558 0, 20,
559 &ctx->htree_slack_percentage);
560 if (ctx->htree_slack_percentage > 100)
561 ctx->htree_slack_percentage = 20;
562 }
563
564 if (ext2fs_has_feature_metadata_csum(fs->super))
565 csum_size = sizeof(struct ext2_dir_entry_tail);
566
567 outdir->max = 0;
568 retval = alloc_size_dir(fs, outdir,
569 (fd->dir_size / fs->blocksize) + 2);
570 if (retval)
571 return retval;
572 outdir->num = fd->compress ? 0 : 1;
573 offset = 0;
574 outdir->hashes[0] = 0;
575 prev_hash = 1;
576 if ((retval = get_next_block(fs, outdir, &block_start)))
577 return retval;
578 dirent = (struct ext2_dir_entry *) block_start;
579 prev_rec_len = 0;
580 rec_len = 0;
581 left = fs->blocksize - csum_size;
582 slack = fd->compress ? min_rec_len :
583 ((fs->blocksize - csum_size) * ctx->htree_slack_percentage)/100;
584 if (slack < min_rec_len)
585 slack = min_rec_len;
586 for (i = 0; i < fd->num_array; i++) {
587 ent = fd->harray + i;
588 if (ent->dir->inode == 0)
589 continue;
590 rec_len = ext2fs_dir_rec_len(ext2fs_dirent_name_len(ent->dir),
591 hash_in_entry);
592 if (rec_len > left) {
593 if (left) {
594 left += prev_rec_len;
595 retval = ext2fs_set_rec_len(fs, left, dirent);
596 if (retval)
597 return retval;
598 }
599 if (csum_size) {
600 t = EXT2_DIRENT_TAIL(block_start,
601 fs->blocksize);
602 ext2fs_initialize_dirent_tail(fs, t);
603 }
604 if ((retval = get_next_block(fs, outdir,
605 &block_start)))
606 return retval;
607 offset = 0;
608 }
609 left = (fs->blocksize - csum_size) - offset;
610 dirent = (struct ext2_dir_entry *) (block_start + offset);
611 if (offset == 0) {
612 if (ent->hash == prev_hash)
613 outdir->hashes[outdir->num-1] = ent->hash | 1;
614 else
615 outdir->hashes[outdir->num-1] = ent->hash;
616 }
617 dirent->inode = ent->dir->inode;
618 ext2fs_dirent_set_name_len(dirent,
619 ext2fs_dirent_name_len(ent->dir));
620 ext2fs_dirent_set_file_type(dirent,
621 ext2fs_dirent_file_type(ent->dir));
622 retval = ext2fs_set_rec_len(fs, rec_len, dirent);
623 if (retval)
624 return retval;
625 prev_rec_len = rec_len;
626 memcpy(dirent->name, ent->dir->name,
627 ext2fs_dirent_name_len(dirent));
628 if (hash_in_entry) {
629 EXT2_DIRENT_HASHES(dirent)->hash = ext2fs_cpu_to_le32(ent->hash);
630 EXT2_DIRENT_HASHES(dirent)->minor_hash =
631 ext2fs_cpu_to_le32(ent->minor_hash);
632 }
633 offset += rec_len;
634 left -= rec_len;
635 if (left < slack) {
636 prev_rec_len += left;
637 retval = ext2fs_set_rec_len(fs, prev_rec_len, dirent);
638 if (retval)
639 return retval;
640 offset += left;
641 left = 0;
642 }
643 prev_hash = ent->hash;
644 }
645 if (left)
646 retval = ext2fs_set_rec_len(fs, rec_len + left, dirent);
647 if (csum_size) {
648 t = EXT2_DIRENT_TAIL(block_start, fs->blocksize);
649 ext2fs_initialize_dirent_tail(fs, t);
650 }
651
652 return retval;
653 }
654
655
set_root_node(ext2_filsys fs,char * buf,ext2_ino_t ino,ext2_ino_t parent,struct ext2_inode * inode)656 static struct ext2_dx_root_info *set_root_node(ext2_filsys fs, char *buf,
657 ext2_ino_t ino, ext2_ino_t parent,
658 struct ext2_inode *inode)
659 {
660 struct ext2_dir_entry *dir;
661 struct ext2_dx_root_info *root;
662 struct ext2_dx_countlimit *limits;
663 int filetype = 0;
664 int csum_size = 0;
665
666 if (ext2fs_has_feature_filetype(fs->super))
667 filetype = EXT2_FT_DIR;
668
669 memset(buf, 0, fs->blocksize);
670 dir = (struct ext2_dir_entry *) buf;
671 dir->inode = ino;
672 dir->name[0] = '.';
673 ext2fs_dirent_set_name_len(dir, 1);
674 ext2fs_dirent_set_file_type(dir, filetype);
675 dir->rec_len = 12;
676 dir = (struct ext2_dir_entry *) (buf + 12);
677 dir->inode = parent;
678 dir->name[0] = '.';
679 dir->name[1] = '.';
680 ext2fs_dirent_set_name_len(dir, 2);
681 ext2fs_dirent_set_file_type(dir, filetype);
682 dir->rec_len = fs->blocksize - 12;
683
684 root = (struct ext2_dx_root_info *) (buf+24);
685 root->reserved_zero = 0;
686 if (ext4_hash_in_dirent(inode))
687 root->hash_version = EXT2_HASH_SIPHASH;
688 else
689 root->hash_version = fs->super->s_def_hash_version;
690 root->info_length = 8;
691 root->indirect_levels = 0;
692 root->unused_flags = 0;
693
694 if (ext2fs_has_feature_metadata_csum(fs->super))
695 csum_size = sizeof(struct ext2_dx_tail);
696
697 limits = (struct ext2_dx_countlimit *) (buf+32);
698 limits->limit = (fs->blocksize - (32 + csum_size)) /
699 sizeof(struct ext2_dx_entry);
700 limits->count = 0;
701
702 return root;
703 }
704
705
set_int_node(ext2_filsys fs,char * buf)706 static struct ext2_dx_entry *set_int_node(ext2_filsys fs, char *buf)
707 {
708 struct ext2_dir_entry *dir;
709 struct ext2_dx_countlimit *limits;
710 int csum_size = 0;
711
712 memset(buf, 0, fs->blocksize);
713 dir = (struct ext2_dir_entry *) buf;
714 dir->inode = 0;
715 (void) ext2fs_set_rec_len(fs, fs->blocksize, dir);
716
717 if (ext2fs_has_feature_metadata_csum(fs->super))
718 csum_size = sizeof(struct ext2_dx_tail);
719
720 limits = (struct ext2_dx_countlimit *) (buf+8);
721 limits->limit = (fs->blocksize - (8 + csum_size)) /
722 sizeof(struct ext2_dx_entry);
723 limits->count = 0;
724
725 return (struct ext2_dx_entry *) limits;
726 }
727
alloc_blocks(ext2_filsys fs,struct ext2_dx_countlimit ** limit,struct ext2_dx_entry ** prev_ent,struct ext2_dx_entry ** next_ent,int * prev_offset,int * next_offset,struct out_dir * outdir,int i,int * prev_count,int * next_count)728 static int alloc_blocks(ext2_filsys fs,
729 struct ext2_dx_countlimit **limit,
730 struct ext2_dx_entry **prev_ent,
731 struct ext2_dx_entry **next_ent,
732 int *prev_offset, int *next_offset,
733 struct out_dir *outdir, int i,
734 int *prev_count, int *next_count)
735 {
736 errcode_t retval;
737 char *block_start;
738
739 if (*limit)
740 (*limit)->limit = (*limit)->count =
741 ext2fs_cpu_to_le16((*limit)->limit);
742 *prev_ent = (struct ext2_dx_entry *) (outdir->buf + *prev_offset);
743 (*prev_ent)->block = ext2fs_cpu_to_le32(outdir->num);
744
745 if (i != 1)
746 (*prev_ent)->hash =
747 ext2fs_cpu_to_le32(outdir->hashes[i]);
748
749 retval = get_next_block(fs, outdir, &block_start);
750 if (retval)
751 return retval;
752
753 /* outdir->buf might be reallocated */
754 *prev_ent = (struct ext2_dx_entry *) (outdir->buf + *prev_offset);
755
756 *next_ent = set_int_node(fs, block_start);
757 *limit = (struct ext2_dx_countlimit *)(*next_ent);
758 if (next_offset)
759 *next_offset = ((char *) *next_ent - outdir->buf);
760
761 *next_count = (*limit)->limit;
762 (*prev_offset) += sizeof(struct ext2_dx_entry);
763 (*prev_count)--;
764
765 return 0;
766 }
767
768 /*
769 * This function takes the leaf nodes which have been written in
770 * outdir, and populates the root node and any necessary interior nodes.
771 */
calculate_tree(ext2_filsys fs,struct out_dir * outdir,ext2_ino_t ino,ext2_ino_t parent,struct ext2_inode * inode)772 static errcode_t calculate_tree(ext2_filsys fs,
773 struct out_dir *outdir,
774 ext2_ino_t ino,
775 ext2_ino_t parent,
776 struct ext2_inode *inode)
777 {
778 struct ext2_dx_root_info *root_info;
779 struct ext2_dx_entry *root, *int_ent, *dx_ent = 0;
780 struct ext2_dx_countlimit *root_limit, *int_limit, *limit;
781 errcode_t retval;
782 int i, c1, c2, c3, nblks;
783 int limit_offset, int_offset, root_offset;
784
785 root_info = set_root_node(fs, outdir->buf, ino, parent, inode);
786 root_offset = limit_offset = ((char *) root_info - outdir->buf) +
787 root_info->info_length;
788 root_limit = (struct ext2_dx_countlimit *) (outdir->buf + limit_offset);
789 c1 = root_limit->limit;
790 nblks = outdir->num;
791
792 /* Write out the pointer blocks */
793 if (nblks - 1 <= c1) {
794 /* Just write out the root block, and we're done */
795 root = (struct ext2_dx_entry *) (outdir->buf + root_offset);
796 for (i=1; i < nblks; i++) {
797 root->block = ext2fs_cpu_to_le32(i);
798 if (i != 1)
799 root->hash =
800 ext2fs_cpu_to_le32(outdir->hashes[i]);
801 root++;
802 c1--;
803 }
804 } else if (nblks - 1 <= ext2fs_htree_intnode_maxrecs(fs, c1)) {
805 c2 = 0;
806 limit = NULL;
807 root_info->indirect_levels = 1;
808 for (i=1; i < nblks; i++) {
809 if (c2 == 0 && c1 == 0)
810 return ENOSPC;
811 if (c2 == 0) {
812 retval = alloc_blocks(fs, &limit, &root,
813 &dx_ent, &root_offset,
814 NULL, outdir, i, &c1,
815 &c2);
816 if (retval)
817 return retval;
818 }
819 dx_ent->block = ext2fs_cpu_to_le32(i);
820 if (c2 != limit->limit)
821 dx_ent->hash =
822 ext2fs_cpu_to_le32(outdir->hashes[i]);
823 dx_ent++;
824 c2--;
825 }
826 limit->count = ext2fs_cpu_to_le16(limit->limit - c2);
827 limit->limit = ext2fs_cpu_to_le16(limit->limit);
828 } else {
829 c2 = 0;
830 c3 = 0;
831 limit = NULL;
832 int_limit = 0;
833 root_info->indirect_levels = 2;
834 for (i = 1; i < nblks; i++) {
835 if (c3 == 0 && c2 == 0 && c1 == 0)
836 return ENOSPC;
837 if (c3 == 0 && c2 == 0) {
838 retval = alloc_blocks(fs, &int_limit, &root,
839 &int_ent, &root_offset,
840 &int_offset, outdir, i,
841 &c1, &c2);
842 if (retval)
843 return retval;
844 }
845 if (c3 == 0) {
846 int delta1 = (char *)int_limit - outdir->buf;
847 int delta2 = (char *)root - outdir->buf;
848
849 retval = alloc_blocks(fs, &limit, &int_ent,
850 &dx_ent, &int_offset,
851 NULL, outdir, i, &c2,
852 &c3);
853 if (retval)
854 return retval;
855
856 /* outdir->buf might be reallocated */
857 int_limit = (struct ext2_dx_countlimit *)
858 (outdir->buf + delta1);
859 root = (struct ext2_dx_entry *)
860 (outdir->buf + delta2);
861 }
862 dx_ent->block = ext2fs_cpu_to_le32(i);
863 if (c3 != limit->limit)
864 dx_ent->hash =
865 ext2fs_cpu_to_le32(outdir->hashes[i]);
866 dx_ent++;
867 c3--;
868 }
869 int_limit->count = ext2fs_cpu_to_le16(limit->limit - c2);
870 int_limit->limit = ext2fs_cpu_to_le16(limit->limit);
871
872 limit->count = ext2fs_cpu_to_le16(limit->limit - c3);
873 limit->limit = ext2fs_cpu_to_le16(limit->limit);
874
875 }
876 root_limit = (struct ext2_dx_countlimit *) (outdir->buf + limit_offset);
877 root_limit->count = ext2fs_cpu_to_le16(root_limit->limit - c1);
878 root_limit->limit = ext2fs_cpu_to_le16(root_limit->limit);
879
880 return 0;
881 }
882
883 struct write_dir_struct {
884 struct out_dir *outdir;
885 errcode_t err;
886 ext2_ino_t ino;
887 e2fsck_t ctx;
888 ext2_ino_t dir;
889 };
890
891 /*
892 * Helper function which writes out a directory block.
893 */
write_dir_block(ext2_filsys fs,blk64_t * block_nr,e2_blkcnt_t blockcnt,blk64_t ref_block EXT2FS_ATTR ((unused)),int ref_offset EXT2FS_ATTR ((unused)),void * priv_data)894 static int write_dir_block(ext2_filsys fs,
895 blk64_t *block_nr,
896 e2_blkcnt_t blockcnt,
897 blk64_t ref_block EXT2FS_ATTR((unused)),
898 int ref_offset EXT2FS_ATTR((unused)),
899 void *priv_data)
900 {
901 struct write_dir_struct *wd = (struct write_dir_struct *) priv_data;
902 char *dir, *buf = 0;
903
904 #ifdef REHASH_DEBUG
905 printf("%u: write_dir_block %lld:%lld", wd->ino, blockcnt, *block_nr);
906 #endif
907 if ((*block_nr == 0) || (blockcnt < 0)) {
908 #ifdef REHASH_DEBUG
909 printf(" - skip\n");
910 #endif
911 return 0;
912 }
913 if (blockcnt < wd->outdir->num)
914 dir = wd->outdir->buf + (blockcnt * fs->blocksize);
915 else if (wd->ctx->lost_and_found == wd->dir) {
916 /* Don't release any extra directory blocks for lost+found */
917 wd->err = ext2fs_new_dir_block(fs, 0, 0, &buf);
918 if (wd->err)
919 return BLOCK_ABORT;
920 dir = buf;
921 wd->outdir->num++;
922 } else {
923 /* Don't free blocks at the end of the directory, they
924 * will be truncated by the caller. */
925 #ifdef REHASH_DEBUG
926 printf(" - not freed\n");
927 #endif
928 return 0;
929 }
930 wd->err = ext2fs_write_dir_block4(fs, *block_nr, dir, 0, wd->dir);
931 if (buf)
932 ext2fs_free_mem(&buf);
933
934 #ifdef REHASH_DEBUG
935 printf(" - write (%d)\n", wd->err);
936 #endif
937 if (wd->err)
938 return BLOCK_ABORT;
939 return 0;
940 }
941
write_directory(e2fsck_t ctx,ext2_filsys fs,struct out_dir * outdir,ext2_ino_t ino,struct ext2_inode * inode,int compress)942 static errcode_t write_directory(e2fsck_t ctx, ext2_filsys fs,
943 struct out_dir *outdir,
944 ext2_ino_t ino, struct ext2_inode *inode,
945 int compress)
946 {
947 struct write_dir_struct wd;
948 errcode_t retval;
949
950 retval = e2fsck_expand_directory(ctx, ino, -1, outdir->num);
951 if (retval)
952 return retval;
953
954 wd.outdir = outdir;
955 wd.err = 0;
956 wd.ino = ino;
957 wd.ctx = ctx;
958 wd.dir = ino;
959
960 retval = ext2fs_block_iterate3(fs, ino, 0, NULL,
961 write_dir_block, &wd);
962 if (retval)
963 return retval;
964 if (wd.err)
965 return wd.err;
966
967 e2fsck_read_inode(ctx, ino, inode, "rehash_dir");
968 if (compress)
969 inode->i_flags &= ~EXT2_INDEX_FL;
970 else
971 inode->i_flags |= EXT2_INDEX_FL;
972 #ifdef REHASH_DEBUG
973 printf("%u: set inode size to %u blocks = %u bytes\n",
974 ino, outdir->num, outdir->num * fs->blocksize);
975 #endif
976 retval = ext2fs_inode_size_set(fs, inode, (ext2_off64_t)outdir->num *
977 fs->blocksize);
978 if (retval)
979 return retval;
980
981 /* ext2fs_punch() calls ext2fs_write_inode() which writes the size */
982 return ext2fs_punch(fs, ino, inode, NULL, outdir->num, ~0ULL);
983 }
984
e2fsck_rehash_dir(e2fsck_t ctx,ext2_ino_t ino,struct problem_context * pctx)985 errcode_t e2fsck_rehash_dir(e2fsck_t ctx, ext2_ino_t ino,
986 struct problem_context *pctx)
987 {
988 ext2_filsys fs = ctx->fs;
989 errcode_t retval;
990 struct ext2_inode inode;
991 char *dir_buf = 0;
992 struct fill_dir_struct fd = { NULL, NULL, 0, 0, 0, NULL,
993 0, 0, 0, 0, 0, 0 };
994 struct out_dir outdir = { 0, 0, 0, 0 };
995 struct name_cmp_ctx name_cmp_ctx = {0, NULL};
996
997 e2fsck_read_inode(ctx, ino, &inode, "rehash_dir");
998
999 if (ext2fs_has_feature_inline_data(fs->super) &&
1000 (inode.i_flags & EXT4_INLINE_DATA_FL))
1001 return 0;
1002
1003 retval = ext2fs_get_mem(inode.i_size, &dir_buf);
1004 if (retval)
1005 goto errout;
1006
1007 fd.max_array = inode.i_size / 32;
1008 retval = ext2fs_get_array(sizeof(struct hash_entry),
1009 fd.max_array, &fd.harray);
1010 if (retval)
1011 goto errout;
1012
1013 fd.ino = ino;
1014 fd.ctx = ctx;
1015 fd.buf = dir_buf;
1016 fd.inode = &inode;
1017 fd.dir = ino;
1018 if (!ext2fs_has_feature_dir_index(fs->super) ||
1019 (inode.i_size / fs->blocksize) < 2)
1020 fd.compress = 1;
1021 fd.parent = 0;
1022
1023 if (fs->encoding && (inode.i_flags & EXT4_CASEFOLD_FL)) {
1024 name_cmp_ctx.casefold = 1;
1025 name_cmp_ctx.tbl = fs->encoding;
1026 }
1027
1028 retry_nohash:
1029 /* Read in the entire directory into memory */
1030 retval = ext2fs_block_iterate3(fs, ino, 0, 0,
1031 fill_dir_block, &fd);
1032 if (fd.err) {
1033 retval = fd.err;
1034 goto errout;
1035 }
1036
1037 /*
1038 * If the entries read are less than a block, then don't index
1039 * the directory
1040 */
1041 if (!fd.compress && (fd.dir_size < (fs->blocksize - 24))) {
1042 fd.compress = 1;
1043 fd.dir_size = 0;
1044 fd.num_array = 0;
1045 goto retry_nohash;
1046 }
1047
1048 #if 0
1049 printf("%d entries (%d bytes) found in inode %d\n",
1050 fd.num_array, fd.dir_size, ino);
1051 #endif
1052
1053 /* Sort the list */
1054 resort:
1055 if (fd.compress && fd.num_array > 1)
1056 sort_r(fd.harray+2, fd.num_array-2, sizeof(struct hash_entry),
1057 hash_cmp, &name_cmp_ctx);
1058 else
1059 sort_r(fd.harray, fd.num_array, sizeof(struct hash_entry),
1060 hash_cmp, &name_cmp_ctx);
1061
1062 /*
1063 * Look for duplicates
1064 */
1065 if (duplicate_search_and_fix(ctx, fs, ino, &fd, &name_cmp_ctx))
1066 goto resort;
1067
1068 if (ctx->options & E2F_OPT_NO) {
1069 retval = 0;
1070 goto errout;
1071 }
1072
1073 /* Sort non-hashed directories by inode number */
1074 if (fd.compress && fd.num_array > 1)
1075 qsort(fd.harray+2, fd.num_array-2,
1076 sizeof(struct hash_entry), ino_cmp);
1077
1078 /*
1079 * Copy the directory entries. In a htree directory these
1080 * will become the leaf nodes.
1081 */
1082 retval = copy_dir_entries(ctx, &fd, &outdir);
1083 if (retval)
1084 goto errout;
1085
1086 free(dir_buf); dir_buf = 0;
1087
1088 if (!fd.compress) {
1089 /* Calculate the interior nodes */
1090 retval = calculate_tree(fs, &outdir, ino, fd.parent, fd.inode);
1091 if (retval)
1092 goto errout;
1093 }
1094
1095 retval = write_directory(ctx, fs, &outdir, ino, &inode, fd.compress);
1096 if (retval)
1097 goto errout;
1098
1099 if (ctx->options & E2F_OPT_CONVERT_BMAP)
1100 retval = e2fsck_rebuild_extents_later(ctx, ino);
1101 else
1102 retval = e2fsck_check_rebuild_extents(ctx, ino, &inode, pctx);
1103 errout:
1104 ext2fs_free_mem(&dir_buf);
1105 ext2fs_free_mem(&fd.harray);
1106
1107 free_out_dir(&outdir);
1108 return retval;
1109 }
1110
e2fsck_rehash_directories(e2fsck_t ctx)1111 void e2fsck_rehash_directories(e2fsck_t ctx)
1112 {
1113 struct problem_context pctx;
1114 #ifdef RESOURCE_TRACK
1115 struct resource_track rtrack;
1116 #endif
1117 struct dir_info *dir;
1118 ext2_u32_iterate iter;
1119 struct dir_info_iter * dirinfo_iter = 0;
1120 ext2_ino_t ino;
1121 errcode_t retval;
1122 int cur, max, all_dirs, first = 1;
1123
1124 init_resource_track(&rtrack, ctx->fs->io);
1125 all_dirs = ctx->options & E2F_OPT_COMPRESS_DIRS;
1126
1127 if (!ctx->dirs_to_hash && !all_dirs)
1128 return;
1129
1130 (void) e2fsck_get_lost_and_found(ctx, 0);
1131
1132 clear_problem_context(&pctx);
1133
1134 cur = 0;
1135 if (all_dirs) {
1136 dirinfo_iter = e2fsck_dir_info_iter_begin(ctx);
1137 max = e2fsck_get_num_dirinfo(ctx);
1138 } else {
1139 retval = ext2fs_u32_list_iterate_begin(ctx->dirs_to_hash,
1140 &iter);
1141 if (retval) {
1142 pctx.errcode = retval;
1143 fix_problem(ctx, PR_3A_OPTIMIZE_ITER, &pctx);
1144 return;
1145 }
1146 max = ext2fs_u32_list_count(ctx->dirs_to_hash);
1147 }
1148 while (1) {
1149 if (all_dirs) {
1150 if ((dir = e2fsck_dir_info_iter(ctx,
1151 dirinfo_iter)) == 0)
1152 break;
1153 ino = dir->ino;
1154 } else {
1155 if (!ext2fs_u32_list_iterate(iter, &ino))
1156 break;
1157 }
1158 if (!ext2fs_test_inode_bitmap2(ctx->inode_dir_map, ino))
1159 continue;
1160
1161 pctx.dir = ino;
1162 if (first) {
1163 fix_problem(ctx, PR_3A_PASS_HEADER, &pctx);
1164 first = 0;
1165 }
1166 #if 0
1167 fix_problem(ctx, PR_3A_OPTIMIZE_DIR, &pctx);
1168 #endif
1169 pctx.errcode = e2fsck_rehash_dir(ctx, ino, &pctx);
1170 if (pctx.errcode) {
1171 end_problem_latch(ctx, PR_LATCH_OPTIMIZE_DIR);
1172 fix_problem(ctx, PR_3A_OPTIMIZE_DIR_ERR, &pctx);
1173 }
1174 if (ctx->progress && !ctx->progress_fd)
1175 e2fsck_simple_progress(ctx, "Rebuilding directory",
1176 100.0 * (float) (++cur) / (float) max, ino);
1177 }
1178 end_problem_latch(ctx, PR_LATCH_OPTIMIZE_DIR);
1179 if (all_dirs)
1180 e2fsck_dir_info_iter_end(ctx, dirinfo_iter);
1181 else
1182 ext2fs_u32_list_iterate_end(iter);
1183
1184 if (ctx->dirs_to_hash)
1185 ext2fs_u32_list_free(ctx->dirs_to_hash);
1186 ctx->dirs_to_hash = 0;
1187
1188 print_resource_track(ctx, "Pass 3A", &rtrack, ctx->fs->io);
1189 }
1190