Theodore Ts'o | b7a0056 | 2002-07-20 00:28:07 -0400 | [diff] [blame] | 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 <errno.h> |
| 49 | #include "e2fsck.h" |
| 50 | #include "problem.h" |
| 51 | |
| 52 | struct fill_dir_struct { |
| 53 | char *buf; |
| 54 | struct ext2_inode *inode; |
| 55 | int err; |
| 56 | e2fsck_t ctx; |
| 57 | struct hash_entry *harray; |
| 58 | int max_array, num_array; |
| 59 | int dir_size; |
Theodore Ts'o | 850d05e | 2002-07-25 00:00:08 -0400 | [diff] [blame] | 60 | int compress; |
Theodore Ts'o | b7a0056 | 2002-07-20 00:28:07 -0400 | [diff] [blame] | 61 | ino_t parent; |
| 62 | }; |
| 63 | |
| 64 | struct hash_entry { |
| 65 | ext2_dirhash_t hash; |
| 66 | ext2_dirhash_t minor_hash; |
| 67 | struct ext2_dir_entry *dir; |
| 68 | }; |
| 69 | |
| 70 | struct out_dir { |
| 71 | int num; |
| 72 | int max; |
| 73 | char *buf; |
| 74 | ext2_dirhash_t *hashes; |
| 75 | }; |
| 76 | |
| 77 | static int fill_dir_block(ext2_filsys fs, |
| 78 | blk_t *block_nr, |
| 79 | e2_blkcnt_t blockcnt, |
| 80 | blk_t ref_block, |
| 81 | int ref_offset, |
| 82 | void *priv_data) |
| 83 | { |
| 84 | struct fill_dir_struct *fd = (struct fill_dir_struct *) priv_data; |
| 85 | struct hash_entry *new_array, *ent; |
| 86 | struct ext2_dir_entry *dirent; |
| 87 | char *dir; |
| 88 | int offset, dir_offset; |
| 89 | |
| 90 | if (blockcnt < 0) |
| 91 | return 0; |
| 92 | |
| 93 | offset = blockcnt * fs->blocksize; |
| 94 | if (offset + fs->blocksize > fd->inode->i_size) { |
| 95 | fd->err = EXT2_ET_DIR_CORRUPTED; |
| 96 | return BLOCK_ABORT; |
| 97 | } |
| 98 | dir = (fd->buf+offset); |
| 99 | if (HOLE_BLKADDR(*block_nr)) { |
| 100 | memset(dir, 0, fs->blocksize); |
| 101 | dirent = (struct ext2_dir_entry *) dir; |
| 102 | dirent->rec_len = fs->blocksize; |
| 103 | } else { |
| 104 | fd->err = ext2fs_read_dir_block(fs, *block_nr, dir); |
| 105 | if (fd->err) |
| 106 | return BLOCK_ABORT; |
| 107 | } |
| 108 | /* While the directory block is "hot", index it. */ |
| 109 | dir_offset = 0; |
| 110 | while (dir_offset < fs->blocksize) { |
| 111 | dirent = (struct ext2_dir_entry *) (dir + dir_offset); |
| 112 | if (((dir_offset + dirent->rec_len) > fs->blocksize) || |
| 113 | (dirent->rec_len < 8) || |
| 114 | ((dirent->rec_len % 4) != 0) || |
| 115 | (((dirent->name_len & 0xFF)+8) > dirent->rec_len)) { |
| 116 | fd->err = EXT2_ET_DIR_CORRUPTED; |
| 117 | return BLOCK_ABORT; |
| 118 | } |
Theodore Ts'o | 850d05e | 2002-07-25 00:00:08 -0400 | [diff] [blame] | 119 | dir_offset += dirent->rec_len; |
Theodore Ts'o | b7a0056 | 2002-07-20 00:28:07 -0400 | [diff] [blame] | 120 | if (dirent->inode == 0) |
Theodore Ts'o | 850d05e | 2002-07-25 00:00:08 -0400 | [diff] [blame] | 121 | continue; |
| 122 | if (!fd->compress && ((dirent->name_len&0xFF) == 1) && |
| 123 | (dirent->name[0] == '.')) |
| 124 | continue; |
| 125 | if (!fd->compress && ((dirent->name_len&0xFF) == 2) && |
Theodore Ts'o | b7a0056 | 2002-07-20 00:28:07 -0400 | [diff] [blame] | 126 | (dirent->name[0] == '.') && (dirent->name[1] == '.')) { |
| 127 | fd->parent = dirent->inode; |
Theodore Ts'o | 850d05e | 2002-07-25 00:00:08 -0400 | [diff] [blame] | 128 | continue; |
Theodore Ts'o | b7a0056 | 2002-07-20 00:28:07 -0400 | [diff] [blame] | 129 | } |
| 130 | if (fd->num_array >= fd->max_array) { |
| 131 | new_array = realloc(fd->harray, |
| 132 | sizeof(struct hash_entry) * (fd->max_array+500)); |
| 133 | if (!new_array) { |
| 134 | fd->err = ENOMEM; |
| 135 | return BLOCK_ABORT; |
| 136 | } |
| 137 | fd->harray = new_array; |
| 138 | fd->max_array += 500; |
| 139 | } |
Theodore Ts'o | 850d05e | 2002-07-25 00:00:08 -0400 | [diff] [blame] | 140 | ent = fd->harray + fd->num_array++; |
Theodore Ts'o | b7a0056 | 2002-07-20 00:28:07 -0400 | [diff] [blame] | 141 | ent->dir = dirent; |
Theodore Ts'o | b7a0056 | 2002-07-20 00:28:07 -0400 | [diff] [blame] | 142 | fd->dir_size += EXT2_DIR_REC_LEN(dirent->name_len & 0xFF); |
Theodore Ts'o | 850d05e | 2002-07-25 00:00:08 -0400 | [diff] [blame] | 143 | if (fd->compress) |
| 144 | ent->hash = ent->minor_hash = 0; |
| 145 | else { |
| 146 | fd->err = ext2fs_dirhash(fs->super->s_def_hash_version, |
| 147 | dirent->name, |
| 148 | dirent->name_len & 0xFF, |
| 149 | fs->super->s_hash_seed, |
| 150 | &ent->hash, &ent->minor_hash); |
| 151 | if (fd->err) |
| 152 | return BLOCK_ABORT; |
| 153 | } |
Theodore Ts'o | b7a0056 | 2002-07-20 00:28:07 -0400 | [diff] [blame] | 154 | } |
| 155 | |
| 156 | return 0; |
| 157 | } |
| 158 | |
| 159 | /* Used for sorting the hash entry */ |
Theodore Ts'o | 850d05e | 2002-07-25 00:00:08 -0400 | [diff] [blame] | 160 | static EXT2_QSORT_TYPE name_cmp(const void *a, const void *b) |
| 161 | { |
| 162 | const struct hash_entry *he_a = (const struct hash_entry *) a; |
| 163 | const struct hash_entry *he_b = (const struct hash_entry *) b; |
| 164 | int ret; |
| 165 | int min_len; |
| 166 | |
| 167 | min_len = he_a->dir->name_len; |
| 168 | if (min_len > he_b->dir->name_len) |
| 169 | min_len = he_b->dir->name_len; |
| 170 | |
| 171 | ret = strncmp(he_a->dir->name, he_b->dir->name, min_len); |
| 172 | if (ret == 0) { |
| 173 | if (he_a->dir->name_len > he_b->dir->name_len) |
| 174 | ret = 1; |
| 175 | else if (he_a->dir->name_len < he_b->dir->name_len) |
| 176 | ret = -1; |
| 177 | else |
| 178 | ret = 0; |
| 179 | } |
| 180 | return ret; |
| 181 | } |
| 182 | |
Theodore Ts'o | b0700a1 | 2003-03-14 01:43:56 -0500 | [diff] [blame] | 183 | /* Used for sorting the hash entry */ |
| 184 | static EXT2_QSORT_TYPE hash_cmp(const void *a, const void *b) |
| 185 | { |
| 186 | const struct hash_entry *he_a = (const struct hash_entry *) a; |
| 187 | const struct hash_entry *he_b = (const struct hash_entry *) b; |
| 188 | int ret; |
| 189 | |
| 190 | if (he_a->hash > he_b->hash) |
| 191 | ret = 1; |
| 192 | else if (he_a->hash < he_b->hash) |
| 193 | ret = -1; |
| 194 | else { |
| 195 | if (he_a->minor_hash > he_b->minor_hash) |
| 196 | ret = 1; |
| 197 | else if (he_a->minor_hash < he_b->minor_hash) |
| 198 | ret = -1; |
| 199 | else |
| 200 | ret = name_cmp(a, b); |
| 201 | } |
| 202 | return ret; |
| 203 | } |
| 204 | |
Theodore Ts'o | b7a0056 | 2002-07-20 00:28:07 -0400 | [diff] [blame] | 205 | static errcode_t alloc_size_dir(ext2_filsys fs, struct out_dir *outdir, |
| 206 | int blocks) |
| 207 | { |
| 208 | void *new_mem; |
| 209 | |
| 210 | if (outdir->max) { |
| 211 | new_mem = realloc(outdir->buf, blocks * fs->blocksize); |
| 212 | if (!new_mem) |
| 213 | return ENOMEM; |
| 214 | outdir->buf = new_mem; |
| 215 | new_mem = realloc(outdir->hashes, |
| 216 | blocks * sizeof(ext2_dirhash_t)); |
| 217 | if (!new_mem) |
| 218 | return ENOMEM; |
| 219 | outdir->hashes = new_mem; |
| 220 | } else { |
| 221 | outdir->buf = malloc(blocks * fs->blocksize); |
| 222 | outdir->hashes = malloc(blocks * sizeof(ext2_dirhash_t)); |
| 223 | outdir->num = 0; |
| 224 | } |
| 225 | outdir->max = blocks; |
| 226 | return 0; |
| 227 | } |
| 228 | |
| 229 | static void free_out_dir(struct out_dir *outdir) |
| 230 | { |
| 231 | free(outdir->buf); |
| 232 | free(outdir->hashes); |
| 233 | outdir->max = 0; |
| 234 | outdir->num =0; |
| 235 | } |
| 236 | |
Theodore Ts'o | 850d05e | 2002-07-25 00:00:08 -0400 | [diff] [blame] | 237 | static errcode_t get_next_block(ext2_filsys fs, struct out_dir *outdir, |
Theodore Ts'o | b7a0056 | 2002-07-20 00:28:07 -0400 | [diff] [blame] | 238 | char ** ret) |
| 239 | { |
| 240 | errcode_t retval; |
| 241 | |
| 242 | if (outdir->num >= outdir->max) { |
| 243 | retval = alloc_size_dir(fs, outdir, outdir->max + 50); |
| 244 | if (retval) |
| 245 | return retval; |
| 246 | } |
| 247 | *ret = outdir->buf + (outdir->num++ * fs->blocksize); |
Theodore Ts'o | 850d05e | 2002-07-25 00:00:08 -0400 | [diff] [blame] | 248 | memset(*ret, 0, fs->blocksize); |
Theodore Ts'o | b7a0056 | 2002-07-20 00:28:07 -0400 | [diff] [blame] | 249 | return 0; |
| 250 | } |
| 251 | |
Theodore Ts'o | b0700a1 | 2003-03-14 01:43:56 -0500 | [diff] [blame] | 252 | /* |
| 253 | * This function is used to make a unique filename. We do this by |
| 254 | * appending ~0, and then incrementing the number. However, we cannot |
| 255 | * expand the length of the filename beyond the padding available in |
| 256 | * the directory entry. |
| 257 | */ |
| 258 | static void mutate_name(char *str, __u16 *len) |
| 259 | { |
| 260 | int i; |
| 261 | __u16 l = *len & 0xFF, h = *len & 0xff00; |
| 262 | |
| 263 | /* |
| 264 | * First check to see if it looks the name has been mutated |
| 265 | * already |
| 266 | */ |
| 267 | for (i = l-1; i > 0; i--) { |
| 268 | if (!isdigit(str[i])) |
| 269 | break; |
| 270 | } |
| 271 | if ((i == l-1) || (str[i] != '~')) { |
| 272 | if (((l-1) & 3) < 2) |
| 273 | l += 2; |
| 274 | else |
| 275 | l = (l+3) & ~3; |
| 276 | str[l-2] = '~'; |
| 277 | str[l-1] = '0'; |
| 278 | *len = l | h; |
| 279 | return; |
| 280 | } |
| 281 | for (i = l-1; i >= 0; i--) { |
| 282 | if (isdigit(str[i])) { |
| 283 | if (str[i] == '9') |
| 284 | str[i] = '0'; |
| 285 | else { |
| 286 | str[i]++; |
| 287 | return; |
| 288 | } |
| 289 | continue; |
| 290 | } |
| 291 | if (i == 1) { |
| 292 | if (str[0] == 'z') |
| 293 | str[0] = 'A'; |
| 294 | else if (str[0] == 'Z') { |
| 295 | str[0] = '~'; |
| 296 | str[1] = '0'; |
| 297 | } else |
| 298 | str[0]++; |
| 299 | } else if (i > 0) { |
| 300 | str[i] = '1'; |
| 301 | str[i-1] = '~'; |
| 302 | } else { |
| 303 | if (str[0] == '~') |
| 304 | str[0] = 'a'; |
| 305 | else |
| 306 | str[0]++; |
| 307 | } |
| 308 | break; |
| 309 | } |
| 310 | } |
| 311 | |
| 312 | static int duplicate_search_and_fix(e2fsck_t ctx, ext2_filsys fs, |
| 313 | ext2_ino_t ino, |
| 314 | struct fill_dir_struct *fd) |
| 315 | { |
| 316 | struct problem_context pctx; |
| 317 | struct hash_entry *ent, *prev; |
| 318 | int i, j; |
| 319 | int fixed = 0; |
| 320 | char new_name[256]; |
| 321 | __u16 new_len; |
| 322 | |
| 323 | clear_problem_context(&pctx); |
| 324 | pctx.ino = ino; |
| 325 | |
| 326 | for (i=1; i < fd->num_array; i++) { |
| 327 | ent = fd->harray + i; |
| 328 | prev = ent - 1; |
| 329 | if (!ent->dir->inode || |
| 330 | ((ent->dir->name_len & 0xFF) != |
| 331 | (prev->dir->name_len & 0xFF)) || |
| 332 | (strncmp(ent->dir->name, prev->dir->name, |
| 333 | ent->dir->name_len & 0xFF))) |
| 334 | continue; |
| 335 | pctx.dirent = ent->dir; |
| 336 | if ((ent->dir->inode == prev->dir->inode) && |
| 337 | fix_problem(ctx, PR_2_DUPLICATE_DIRENT, &pctx)) { |
| 338 | e2fsck_adjust_inode_count(ctx, ent->dir->inode, -1); |
| 339 | ent->dir->inode = 0; |
| 340 | fixed++; |
| 341 | continue; |
| 342 | } |
| 343 | memcpy(new_name, ent->dir->name, ent->dir->name_len & 0xFF); |
| 344 | new_len = ent->dir->name_len; |
| 345 | mutate_name(new_name, &new_len); |
| 346 | for (j=0; j < fd->num_array; j++) { |
| 347 | if ((i==j) || |
| 348 | ((ent->dir->name_len & 0xFF) != |
| 349 | (fd->harray[j].dir->name_len & 0xFF)) || |
| 350 | (strncmp(new_name, fd->harray[j].dir->name, |
| 351 | new_len & 0xFF))) |
| 352 | continue; |
| 353 | mutate_name(new_name, &new_len); |
| 354 | |
| 355 | j = -1; |
| 356 | } |
| 357 | new_name[new_len & 0xFF] = 0; |
| 358 | pctx.str = new_name; |
| 359 | if (fix_problem(ctx, PR_2_NON_UNIQUE_FILE, &pctx)) { |
| 360 | memcpy(ent->dir->name, new_name, new_len & 0xFF); |
| 361 | ent->dir->name_len = new_len; |
| 362 | ext2fs_dirhash(fs->super->s_def_hash_version, |
| 363 | ent->dir->name, |
| 364 | ent->dir->name_len & 0xFF, |
| 365 | fs->super->s_hash_seed, |
| 366 | &ent->hash, &ent->minor_hash); |
| 367 | fixed++; |
| 368 | } |
| 369 | } |
| 370 | return fixed; |
| 371 | } |
| 372 | |
Theodore Ts'o | b7a0056 | 2002-07-20 00:28:07 -0400 | [diff] [blame] | 373 | |
Theodore Ts'o | 850d05e | 2002-07-25 00:00:08 -0400 | [diff] [blame] | 374 | static errcode_t copy_dir_entries(ext2_filsys fs, |
| 375 | struct fill_dir_struct *fd, |
| 376 | struct out_dir *outdir) |
| 377 | { |
| 378 | errcode_t retval; |
| 379 | char *block_start; |
| 380 | struct hash_entry *ent; |
| 381 | struct ext2_dir_entry *dirent; |
| 382 | int i, rec_len, left; |
| 383 | ext2_dirhash_t prev_hash; |
| 384 | int offset; |
| 385 | |
| 386 | outdir->max = 0; |
| 387 | retval = alloc_size_dir(fs, outdir, |
| 388 | (fd->dir_size / fs->blocksize) + 2); |
| 389 | if (retval) |
| 390 | return retval; |
| 391 | outdir->num = fd->compress ? 0 : 1; |
| 392 | offset = 0; |
| 393 | outdir->hashes[0] = 0; |
| 394 | prev_hash = 1; |
| 395 | if ((retval = get_next_block(fs, outdir, &block_start))) |
| 396 | return retval; |
| 397 | dirent = (struct ext2_dir_entry *) block_start; |
| 398 | left = fs->blocksize; |
| 399 | for (i=0; i < fd->num_array; i++) { |
| 400 | ent = fd->harray + i; |
Theodore Ts'o | b0700a1 | 2003-03-14 01:43:56 -0500 | [diff] [blame] | 401 | if (ent->dir->inode == 0) |
| 402 | continue; |
Theodore Ts'o | 850d05e | 2002-07-25 00:00:08 -0400 | [diff] [blame] | 403 | rec_len = EXT2_DIR_REC_LEN(ent->dir->name_len & 0xFF); |
Theodore Ts'o | 850d05e | 2002-07-25 00:00:08 -0400 | [diff] [blame] | 404 | if (rec_len > left) { |
Theodore Ts'o | fe5b72d | 2002-09-29 19:05:26 -0400 | [diff] [blame] | 405 | if (left) |
| 406 | dirent->rec_len += left; |
Theodore Ts'o | 850d05e | 2002-07-25 00:00:08 -0400 | [diff] [blame] | 407 | if ((retval = get_next_block(fs, outdir, |
| 408 | &block_start))) |
| 409 | return retval; |
Theodore Ts'o | fe5b72d | 2002-09-29 19:05:26 -0400 | [diff] [blame] | 410 | offset = 0; |
Theodore Ts'o | 850d05e | 2002-07-25 00:00:08 -0400 | [diff] [blame] | 411 | } |
Theodore Ts'o | fe5b72d | 2002-09-29 19:05:26 -0400 | [diff] [blame] | 412 | left = fs->blocksize - offset; |
| 413 | dirent = (struct ext2_dir_entry *) (block_start + offset); |
Theodore Ts'o | 850d05e | 2002-07-25 00:00:08 -0400 | [diff] [blame] | 414 | if (offset == 0) { |
| 415 | if (ent->hash == prev_hash) |
| 416 | outdir->hashes[outdir->num-1] = ent->hash | 1; |
| 417 | else |
| 418 | outdir->hashes[outdir->num-1] = ent->hash; |
| 419 | } |
| 420 | dirent->inode = ent->dir->inode; |
| 421 | dirent->name_len = ent->dir->name_len; |
| 422 | dirent->rec_len = rec_len; |
| 423 | memcpy(dirent->name, ent->dir->name, dirent->name_len & 0xFF); |
| 424 | offset += rec_len; |
| 425 | left -= rec_len; |
| 426 | if (left < 12) { |
| 427 | dirent->rec_len += left; |
| 428 | offset += left; |
Theodore Ts'o | cf3909e | 2002-09-06 10:14:12 -0400 | [diff] [blame] | 429 | left = 0; |
Theodore Ts'o | 850d05e | 2002-07-25 00:00:08 -0400 | [diff] [blame] | 430 | } |
| 431 | prev_hash = ent->hash; |
| 432 | } |
| 433 | if (left) |
| 434 | dirent->rec_len += left; |
| 435 | |
| 436 | return 0; |
| 437 | } |
| 438 | |
| 439 | |
| 440 | static struct ext2_dx_root_info *set_root_node(ext2_filsys fs, char *buf, |
Theodore Ts'o | b7a0056 | 2002-07-20 00:28:07 -0400 | [diff] [blame] | 441 | ext2_ino_t ino, ext2_ino_t parent) |
| 442 | { |
| 443 | struct ext2_dir_entry *dir; |
| 444 | struct ext2_dx_root_info *root; |
| 445 | struct ext2_dx_countlimit *limits; |
Theodore Ts'o | 850d05e | 2002-07-25 00:00:08 -0400 | [diff] [blame] | 446 | int filetype = 0; |
Theodore Ts'o | b7a0056 | 2002-07-20 00:28:07 -0400 | [diff] [blame] | 447 | |
| 448 | if (fs->super->s_feature_incompat & EXT2_FEATURE_INCOMPAT_FILETYPE) |
| 449 | filetype = EXT2_FT_DIR << 8; |
| 450 | |
| 451 | memset(buf, 0, fs->blocksize); |
| 452 | dir = (struct ext2_dir_entry *) buf; |
| 453 | dir->inode = ino; |
| 454 | dir->name[0] = '.'; |
| 455 | dir->name_len = 1 | filetype; |
| 456 | dir->rec_len = 12; |
| 457 | dir = (struct ext2_dir_entry *) (buf + 12); |
| 458 | dir->inode = parent; |
| 459 | dir->name[0] = '.'; |
| 460 | dir->name[1] = '.'; |
| 461 | dir->name_len = 2 | filetype; |
| 462 | dir->rec_len = fs->blocksize - 12; |
| 463 | |
| 464 | root = (struct ext2_dx_root_info *) (buf+24); |
| 465 | root->reserved_zero = 0; |
| 466 | root->hash_version = fs->super->s_def_hash_version; |
| 467 | root->info_length = 8; |
| 468 | root->indirect_levels = 0; |
| 469 | root->unused_flags = 0; |
| 470 | |
| 471 | limits = (struct ext2_dx_countlimit *) (buf+32); |
| 472 | limits->limit = (fs->blocksize - 32) / sizeof(struct ext2_dx_entry); |
| 473 | limits->count = 0; |
| 474 | |
| 475 | return root; |
| 476 | } |
| 477 | |
| 478 | |
Theodore Ts'o | 850d05e | 2002-07-25 00:00:08 -0400 | [diff] [blame] | 479 | static struct ext2_dx_entry *set_int_node(ext2_filsys fs, char *buf) |
Theodore Ts'o | b7a0056 | 2002-07-20 00:28:07 -0400 | [diff] [blame] | 480 | { |
| 481 | struct ext2_dir_entry *dir; |
| 482 | struct ext2_dx_countlimit *limits; |
| 483 | |
| 484 | memset(buf, 0, fs->blocksize); |
| 485 | dir = (struct ext2_dir_entry *) buf; |
| 486 | dir->inode = 0; |
| 487 | dir->rec_len = fs->blocksize; |
| 488 | |
| 489 | limits = (struct ext2_dx_countlimit *) (buf+8); |
| 490 | limits->limit = (fs->blocksize - 8) / sizeof(struct ext2_dx_entry); |
| 491 | limits->count = 0; |
| 492 | |
| 493 | return (struct ext2_dx_entry *) limits; |
| 494 | } |
| 495 | |
Theodore Ts'o | 850d05e | 2002-07-25 00:00:08 -0400 | [diff] [blame] | 496 | /* |
| 497 | * This function takes the leaf nodes which have been written in |
| 498 | * outdir, and populates the root node and any necessary interior nodes. |
| 499 | */ |
| 500 | static errcode_t calculate_tree(ext2_filsys fs, |
| 501 | struct out_dir *outdir, |
| 502 | ext2_ino_t ino, |
| 503 | ext2_ino_t parent) |
| 504 | { |
| 505 | struct ext2_dx_root_info *root_info; |
| 506 | struct ext2_dx_entry *root, *dx_ent = 0; |
| 507 | struct ext2_dx_countlimit *root_limit, *limit; |
| 508 | errcode_t retval; |
| 509 | char * block_start; |
| 510 | int i, c1, c2, nblks; |
| 511 | int limit_offset, root_offset; |
| 512 | |
| 513 | root_info = set_root_node(fs, outdir->buf, ino, parent); |
| 514 | root_offset = limit_offset = ((char *) root_info - outdir->buf) + |
| 515 | root_info->info_length; |
| 516 | root_limit = (struct ext2_dx_countlimit *) (outdir->buf + limit_offset); |
| 517 | c1 = root_limit->limit; |
| 518 | nblks = outdir->num; |
| 519 | |
| 520 | /* Write out the pointer blocks */ |
| 521 | if (nblks-1 <= c1) { |
| 522 | /* Just write out the root block, and we're done */ |
| 523 | root = (struct ext2_dx_entry *) (outdir->buf + root_offset); |
| 524 | for (i=1; i < nblks; i++) { |
| 525 | root->block = ext2fs_cpu_to_le32(i); |
| 526 | if (i != 1) |
| 527 | root->hash = |
| 528 | ext2fs_cpu_to_le32(outdir->hashes[i]); |
| 529 | root++; |
| 530 | c1--; |
| 531 | } |
| 532 | } else { |
| 533 | c2 = 0; |
| 534 | limit = 0; |
| 535 | root_info->indirect_levels = 1; |
| 536 | for (i=1; i < nblks; i++) { |
| 537 | if (c1 == 0) |
| 538 | return ENOSPC; |
| 539 | if (c2 == 0) { |
| 540 | if (limit) |
| 541 | limit->limit = limit->count = |
| 542 | ext2fs_cpu_to_le16(limit->limit); |
| 543 | root = (struct ext2_dx_entry *) |
| 544 | (outdir->buf + root_offset); |
| 545 | root->block = ext2fs_cpu_to_le32(outdir->num); |
| 546 | if (i != 1) |
| 547 | root->hash = |
| 548 | ext2fs_cpu_to_le32(outdir->hashes[i]); |
| 549 | if ((retval = get_next_block(fs, outdir, |
| 550 | &block_start))) |
| 551 | return retval; |
| 552 | dx_ent = set_int_node(fs, block_start); |
| 553 | limit = (struct ext2_dx_countlimit *) dx_ent; |
| 554 | c2 = limit->limit; |
| 555 | root_offset += sizeof(struct ext2_dx_entry); |
| 556 | c1--; |
| 557 | } |
| 558 | dx_ent->block = ext2fs_cpu_to_le32(i); |
| 559 | if (c2 != limit->limit) |
| 560 | dx_ent->hash = |
| 561 | ext2fs_cpu_to_le32(outdir->hashes[i]); |
| 562 | dx_ent++; |
| 563 | c2--; |
| 564 | } |
| 565 | limit->count = ext2fs_cpu_to_le16(limit->limit - c2); |
| 566 | limit->limit = ext2fs_cpu_to_le16(limit->limit); |
| 567 | } |
| 568 | root_limit = (struct ext2_dx_countlimit *) (outdir->buf + limit_offset); |
| 569 | root_limit->count = ext2fs_cpu_to_le16(root_limit->limit - c1); |
| 570 | root_limit->limit = ext2fs_cpu_to_le16(root_limit->limit); |
| 571 | |
| 572 | return 0; |
| 573 | } |
Theodore Ts'o | b7a0056 | 2002-07-20 00:28:07 -0400 | [diff] [blame] | 574 | |
| 575 | struct write_dir_struct { |
| 576 | struct out_dir *outdir; |
| 577 | errcode_t err; |
| 578 | e2fsck_t ctx; |
| 579 | int cleared; |
| 580 | }; |
| 581 | |
| 582 | /* |
Theodore Ts'o | 850d05e | 2002-07-25 00:00:08 -0400 | [diff] [blame] | 583 | * Helper function which writes out a directory block. |
Theodore Ts'o | b7a0056 | 2002-07-20 00:28:07 -0400 | [diff] [blame] | 584 | */ |
| 585 | static int write_dir_block(ext2_filsys fs, |
| 586 | blk_t *block_nr, |
| 587 | e2_blkcnt_t blockcnt, |
| 588 | blk_t ref_block, |
| 589 | int ref_offset, |
| 590 | void *priv_data) |
| 591 | { |
| 592 | struct write_dir_struct *wd = (struct write_dir_struct *) priv_data; |
Theodore Ts'o | 850d05e | 2002-07-25 00:00:08 -0400 | [diff] [blame] | 593 | blk_t blk; |
Theodore Ts'o | b7a0056 | 2002-07-20 00:28:07 -0400 | [diff] [blame] | 594 | char *dir; |
Theodore Ts'o | b7a0056 | 2002-07-20 00:28:07 -0400 | [diff] [blame] | 595 | |
| 596 | if (*block_nr == 0) |
| 597 | return 0; |
| 598 | if (blockcnt >= wd->outdir->num) { |
| 599 | e2fsck_read_bitmaps(wd->ctx); |
| 600 | blk = *block_nr; |
| 601 | ext2fs_unmark_block_bitmap(wd->ctx->block_found_map, blk); |
| 602 | ext2fs_block_alloc_stats(fs, blk, -1); |
| 603 | *block_nr = 0; |
| 604 | wd->cleared++; |
| 605 | return BLOCK_CHANGED; |
| 606 | } |
| 607 | if (blockcnt < 0) |
| 608 | return 0; |
| 609 | |
| 610 | dir = wd->outdir->buf + (blockcnt * fs->blocksize); |
| 611 | wd->err = ext2fs_write_dir_block(fs, *block_nr, dir); |
| 612 | if (wd->err) |
| 613 | return BLOCK_ABORT; |
| 614 | return 0; |
| 615 | } |
| 616 | |
Theodore Ts'o | b7a0056 | 2002-07-20 00:28:07 -0400 | [diff] [blame] | 617 | static errcode_t write_directory(e2fsck_t ctx, ext2_filsys fs, |
Theodore Ts'o | 850d05e | 2002-07-25 00:00:08 -0400 | [diff] [blame] | 618 | struct out_dir *outdir, |
| 619 | ext2_ino_t ino, int compress) |
Theodore Ts'o | b7a0056 | 2002-07-20 00:28:07 -0400 | [diff] [blame] | 620 | { |
| 621 | struct write_dir_struct wd; |
| 622 | errcode_t retval; |
| 623 | struct ext2_inode inode; |
| 624 | |
| 625 | retval = e2fsck_expand_directory(ctx, ino, -1, outdir->num); |
| 626 | if (retval) |
| 627 | return retval; |
| 628 | |
| 629 | wd.outdir = outdir; |
| 630 | wd.err = 0; |
| 631 | wd.ctx = ctx; |
| 632 | wd.cleared = 0; |
| 633 | |
| 634 | retval = ext2fs_block_iterate2(fs, ino, 0, 0, |
| 635 | write_dir_block, &wd); |
| 636 | if (retval) |
| 637 | return retval; |
| 638 | if (wd.err) |
| 639 | return wd.err; |
| 640 | |
| 641 | e2fsck_read_inode(ctx, ino, &inode, "rehash_dir"); |
Theodore Ts'o | e70ae99 | 2002-09-28 09:16:28 -0400 | [diff] [blame] | 642 | if (compress) |
| 643 | inode.i_flags &= ~EXT2_INDEX_FL; |
| 644 | else |
Theodore Ts'o | 850d05e | 2002-07-25 00:00:08 -0400 | [diff] [blame] | 645 | inode.i_flags |= EXT2_INDEX_FL; |
Theodore Ts'o | b7a0056 | 2002-07-20 00:28:07 -0400 | [diff] [blame] | 646 | inode.i_size = outdir->num * fs->blocksize; |
| 647 | inode.i_blocks -= (fs->blocksize / 512) * wd.cleared; |
| 648 | e2fsck_write_inode(ctx, ino, &inode, "rehash_dir"); |
| 649 | |
| 650 | return 0; |
| 651 | } |
| 652 | |
| 653 | errcode_t e2fsck_rehash_dir(e2fsck_t ctx, ext2_ino_t ino) |
| 654 | { |
| 655 | ext2_filsys fs = ctx->fs; |
| 656 | errcode_t retval; |
| 657 | struct ext2_inode inode; |
Theodore Ts'o | 850d05e | 2002-07-25 00:00:08 -0400 | [diff] [blame] | 658 | char *dir_buf = 0; |
Theodore Ts'o | b7a0056 | 2002-07-20 00:28:07 -0400 | [diff] [blame] | 659 | struct fill_dir_struct fd; |
Theodore Ts'o | b7a0056 | 2002-07-20 00:28:07 -0400 | [diff] [blame] | 660 | struct out_dir outdir; |
Theodore Ts'o | b7a0056 | 2002-07-20 00:28:07 -0400 | [diff] [blame] | 661 | |
| 662 | e2fsck_read_inode(ctx, ino, &inode, "rehash_dir"); |
Theodore Ts'o | b7a0056 | 2002-07-20 00:28:07 -0400 | [diff] [blame] | 663 | |
| 664 | retval = ENOMEM; |
| 665 | fd.harray = 0; |
| 666 | dir_buf = malloc(inode.i_size); |
| 667 | if (!dir_buf) |
| 668 | goto errout; |
| 669 | |
| 670 | fd.max_array = inode.i_size / 32; |
| 671 | fd.num_array = 0; |
| 672 | fd.harray = malloc(fd.max_array * sizeof(struct hash_entry)); |
| 673 | if (!fd.harray) |
| 674 | goto errout; |
| 675 | |
| 676 | fd.ctx = ctx; |
| 677 | fd.buf = dir_buf; |
| 678 | fd.inode = &inode; |
| 679 | fd.err = 0; |
| 680 | fd.dir_size = 0; |
Theodore Ts'o | 850d05e | 2002-07-25 00:00:08 -0400 | [diff] [blame] | 681 | fd.compress = 0; |
| 682 | if (!(fs->super->s_feature_compat & EXT2_FEATURE_COMPAT_DIR_INDEX) || |
Theodore Ts'o | e70ae99 | 2002-09-28 09:16:28 -0400 | [diff] [blame] | 683 | (inode.i_size / fs->blocksize) < 2) |
Theodore Ts'o | 850d05e | 2002-07-25 00:00:08 -0400 | [diff] [blame] | 684 | fd.compress = 1; |
Theodore Ts'o | b7a0056 | 2002-07-20 00:28:07 -0400 | [diff] [blame] | 685 | fd.parent = 0; |
| 686 | |
| 687 | /* Read in the entire directory into memory */ |
| 688 | retval = ext2fs_block_iterate2(fs, ino, 0, 0, |
| 689 | fill_dir_block, &fd); |
| 690 | if (fd.err) { |
| 691 | retval = fd.err; |
| 692 | goto errout; |
| 693 | } |
| 694 | |
| 695 | #if 0 |
| 696 | printf("%d entries (%d bytes) found in inode %d\n", |
| 697 | fd.num_array, fd.dir_size, ino); |
| 698 | #endif |
| 699 | |
Theodore Ts'o | 850d05e | 2002-07-25 00:00:08 -0400 | [diff] [blame] | 700 | /* Sort the list */ |
Theodore Ts'o | b0700a1 | 2003-03-14 01:43:56 -0500 | [diff] [blame] | 701 | resort: |
Theodore Ts'o | 850d05e | 2002-07-25 00:00:08 -0400 | [diff] [blame] | 702 | if (fd.compress) |
| 703 | qsort(fd.harray+2, fd.num_array-2, |
| 704 | sizeof(struct hash_entry), name_cmp); |
| 705 | else |
| 706 | qsort(fd.harray, fd.num_array, |
| 707 | sizeof(struct hash_entry), hash_cmp); |
Theodore Ts'o | b7a0056 | 2002-07-20 00:28:07 -0400 | [diff] [blame] | 708 | |
Theodore Ts'o | 850d05e | 2002-07-25 00:00:08 -0400 | [diff] [blame] | 709 | /* |
Theodore Ts'o | b0700a1 | 2003-03-14 01:43:56 -0500 | [diff] [blame] | 710 | * Look for duplicates |
| 711 | */ |
| 712 | if (duplicate_search_and_fix(ctx, fs, ino, &fd)) |
| 713 | goto resort; |
| 714 | |
| 715 | /* |
Theodore Ts'o | 850d05e | 2002-07-25 00:00:08 -0400 | [diff] [blame] | 716 | * Copy the directory entries. In a htree directory these |
| 717 | * will become the leaf nodes. |
| 718 | */ |
| 719 | retval = copy_dir_entries(fs, &fd, &outdir); |
Theodore Ts'o | b7a0056 | 2002-07-20 00:28:07 -0400 | [diff] [blame] | 720 | if (retval) |
| 721 | goto errout; |
Theodore Ts'o | 850d05e | 2002-07-25 00:00:08 -0400 | [diff] [blame] | 722 | |
Theodore Ts'o | b7a0056 | 2002-07-20 00:28:07 -0400 | [diff] [blame] | 723 | free(dir_buf); dir_buf = 0; |
| 724 | |
Theodore Ts'o | 850d05e | 2002-07-25 00:00:08 -0400 | [diff] [blame] | 725 | if (!fd.compress) { |
| 726 | /* Calculate the interior nodes */ |
| 727 | retval = calculate_tree(fs, &outdir, ino, fd.parent); |
| 728 | if (retval) |
| 729 | goto errout; |
Theodore Ts'o | b7a0056 | 2002-07-20 00:28:07 -0400 | [diff] [blame] | 730 | } |
Theodore Ts'o | 850d05e | 2002-07-25 00:00:08 -0400 | [diff] [blame] | 731 | |
| 732 | retval = write_directory(ctx, fs, &outdir, ino, fd.compress); |
Theodore Ts'o | b7a0056 | 2002-07-20 00:28:07 -0400 | [diff] [blame] | 733 | if (retval) |
| 734 | goto errout; |
| 735 | |
| 736 | errout: |
| 737 | if (dir_buf) |
| 738 | free(dir_buf); |
Theodore Ts'o | 850d05e | 2002-07-25 00:00:08 -0400 | [diff] [blame] | 739 | if (fd.harray) |
| 740 | free(fd.harray); |
| 741 | |
Theodore Ts'o | b7a0056 | 2002-07-20 00:28:07 -0400 | [diff] [blame] | 742 | free_out_dir(&outdir); |
| 743 | return retval; |
| 744 | } |
| 745 | |
| 746 | void e2fsck_rehash_directories(e2fsck_t ctx) |
| 747 | { |
Theodore Ts'o | b7a0056 | 2002-07-20 00:28:07 -0400 | [diff] [blame] | 748 | struct problem_context pctx; |
Theodore Ts'o | 850d05e | 2002-07-25 00:00:08 -0400 | [diff] [blame] | 749 | #ifdef RESOURCE_TRACK |
| 750 | struct resource_track rtrack; |
| 751 | #endif |
| 752 | struct dir_info *dir; |
| 753 | ext2_u32_iterate iter; |
| 754 | ext2_ino_t ino; |
| 755 | errcode_t retval; |
Theodore Ts'o | b0700a1 | 2003-03-14 01:43:56 -0500 | [diff] [blame] | 756 | int i, cur, max, all_dirs, dir_index, first = 1; |
Theodore Ts'o | b7a0056 | 2002-07-20 00:28:07 -0400 | [diff] [blame] | 757 | |
Theodore Ts'o | 850d05e | 2002-07-25 00:00:08 -0400 | [diff] [blame] | 758 | #ifdef RESOURCE_TRACK |
| 759 | init_resource_track(&rtrack); |
| 760 | #endif |
| 761 | |
| 762 | all_dirs = ctx->options & E2F_OPT_COMPRESS_DIRS; |
| 763 | |
| 764 | if (!ctx->dirs_to_hash && !all_dirs) |
Theodore Ts'o | b7a0056 | 2002-07-20 00:28:07 -0400 | [diff] [blame] | 765 | return; |
| 766 | |
Theodore Ts'o | 850d05e | 2002-07-25 00:00:08 -0400 | [diff] [blame] | 767 | e2fsck_get_lost_and_found(ctx, 0); |
Theodore Ts'o | b7a0056 | 2002-07-20 00:28:07 -0400 | [diff] [blame] | 768 | |
| 769 | clear_problem_context(&pctx); |
Theodore Ts'o | 850d05e | 2002-07-25 00:00:08 -0400 | [diff] [blame] | 770 | |
| 771 | dir_index = ctx->fs->super->s_feature_compat & EXT2_FEATURE_COMPAT_DIR_INDEX; |
Theodore Ts'o | b0700a1 | 2003-03-14 01:43:56 -0500 | [diff] [blame] | 772 | cur = 0; |
| 773 | if (all_dirs) { |
Theodore Ts'o | 850d05e | 2002-07-25 00:00:08 -0400 | [diff] [blame] | 774 | i = 0; |
Theodore Ts'o | b0700a1 | 2003-03-14 01:43:56 -0500 | [diff] [blame] | 775 | max = e2fsck_get_num_dirinfo(ctx); |
| 776 | } else { |
Theodore Ts'o | 850d05e | 2002-07-25 00:00:08 -0400 | [diff] [blame] | 777 | retval = ext2fs_u32_list_iterate_begin(ctx->dirs_to_hash, |
| 778 | &iter); |
| 779 | if (retval) { |
| 780 | pctx.errcode = retval; |
| 781 | fix_problem(ctx, PR_3A_OPTIMIZE_ITER, &pctx); |
| 782 | return; |
| 783 | } |
Theodore Ts'o | b0700a1 | 2003-03-14 01:43:56 -0500 | [diff] [blame] | 784 | max = ext2fs_u32_list_count(ctx->dirs_to_hash); |
Theodore Ts'o | b7a0056 | 2002-07-20 00:28:07 -0400 | [diff] [blame] | 785 | } |
Theodore Ts'o | 850d05e | 2002-07-25 00:00:08 -0400 | [diff] [blame] | 786 | while (1) { |
| 787 | if (all_dirs) { |
| 788 | if ((dir = e2fsck_dir_info_iter(ctx, &i)) == 0) |
| 789 | break; |
| 790 | ino = dir->ino; |
| 791 | } else { |
| 792 | if (!ext2fs_u32_list_iterate(iter, &ino)) |
| 793 | break; |
| 794 | } |
| 795 | if (ino == ctx->lost_and_found) |
Theodore Ts'o | b7a0056 | 2002-07-20 00:28:07 -0400 | [diff] [blame] | 796 | continue; |
| 797 | pctx.dir = ino; |
| 798 | if (first) { |
| 799 | fix_problem(ctx, PR_3A_PASS_HEADER, &pctx); |
| 800 | first = 0; |
| 801 | } |
Theodore Ts'o | b0700a1 | 2003-03-14 01:43:56 -0500 | [diff] [blame] | 802 | #if 0 |
Theodore Ts'o | 850d05e | 2002-07-25 00:00:08 -0400 | [diff] [blame] | 803 | fix_problem(ctx, PR_3A_OPTIMIZE_DIR, &pctx); |
Theodore Ts'o | b0700a1 | 2003-03-14 01:43:56 -0500 | [diff] [blame] | 804 | #endif |
Theodore Ts'o | b7a0056 | 2002-07-20 00:28:07 -0400 | [diff] [blame] | 805 | pctx.errcode = e2fsck_rehash_dir(ctx, ino); |
| 806 | if (pctx.errcode) { |
Theodore Ts'o | 850d05e | 2002-07-25 00:00:08 -0400 | [diff] [blame] | 807 | end_problem_latch(ctx, PR_LATCH_OPTIMIZE_DIR); |
| 808 | fix_problem(ctx, PR_3A_OPTIMIZE_DIR_ERR, &pctx); |
Theodore Ts'o | b7a0056 | 2002-07-20 00:28:07 -0400 | [diff] [blame] | 809 | } |
Theodore Ts'o | 52734dc | 2003-03-15 04:03:43 -0500 | [diff] [blame] | 810 | if (ctx->progress && !ctx->progress_fd) |
| 811 | e2fsck_simple_progress(ctx, "Rebuilding directory", |
| 812 | (float) (++cur) / (float) max, ino); |
Theodore Ts'o | b7a0056 | 2002-07-20 00:28:07 -0400 | [diff] [blame] | 813 | } |
Theodore Ts'o | 850d05e | 2002-07-25 00:00:08 -0400 | [diff] [blame] | 814 | end_problem_latch(ctx, PR_LATCH_OPTIMIZE_DIR); |
| 815 | if (!all_dirs) |
| 816 | ext2fs_u32_list_iterate_end(iter); |
Theodore Ts'o | b7a0056 | 2002-07-20 00:28:07 -0400 | [diff] [blame] | 817 | |
Theodore Ts'o | 850d05e | 2002-07-25 00:00:08 -0400 | [diff] [blame] | 818 | if (ctx->dirs_to_hash) |
| 819 | ext2fs_u32_list_free(ctx->dirs_to_hash); |
Theodore Ts'o | b7a0056 | 2002-07-20 00:28:07 -0400 | [diff] [blame] | 820 | ctx->dirs_to_hash = 0; |
Theodore Ts'o | 850d05e | 2002-07-25 00:00:08 -0400 | [diff] [blame] | 821 | |
| 822 | #ifdef RESOURCE_TRACK |
| 823 | if (ctx->options & E2F_OPT_TIME2) { |
| 824 | e2fsck_clear_progbar(ctx); |
| 825 | print_resource_track("Pass 3A", &rtrack); |
| 826 | } |
| 827 | #endif |
Theodore Ts'o | b7a0056 | 2002-07-20 00:28:07 -0400 | [diff] [blame] | 828 | } |