| 1 | /* $NetBSD: ext2fs_htree.c,v 1.9 2016/08/23 06:23:26 christos Exp $ */ |
| 2 | |
| 3 | /*- |
| 4 | * Copyright (c) 2010, 2012 Zheng Liu <lz@freebsd.org> |
| 5 | * Copyright (c) 2012, Vyacheslav Matyushin |
| 6 | * All rights reserved. |
| 7 | * |
| 8 | * Redistribution and use in source and binary forms, with or without |
| 9 | * modification, are permitted provided that the following conditions |
| 10 | * are met: |
| 11 | * 1. Redistributions of source code must retain the above copyright |
| 12 | * notice, this list of conditions and the following disclaimer. |
| 13 | * 2. Redistributions in binary form must reproduce the above copyright |
| 14 | * notice, this list of conditions and the following disclaimer in the |
| 15 | * documentation and/or other materials provided with the distribution. |
| 16 | * |
| 17 | * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND |
| 18 | * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE |
| 19 | * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE |
| 20 | * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE |
| 21 | * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL |
| 22 | * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS |
| 23 | * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) |
| 24 | * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT |
| 25 | * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY |
| 26 | * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF |
| 27 | * SUCH DAMAGE. |
| 28 | * |
| 29 | * $FreeBSD: head/sys/fs/ext2fs/ext2fs_htree.c 294653 2016-01-24 02:41:49Z pfg $ |
| 30 | */ |
| 31 | #include <sys/cdefs.h> |
| 32 | __KERNEL_RCSID(0, "$NetBSD: ext2fs_htree.c,v 1.9 2016/08/23 06:23:26 christos Exp $" ); |
| 33 | |
| 34 | #include <sys/param.h> |
| 35 | #include <sys/systm.h> |
| 36 | #include <sys/resourcevar.h> |
| 37 | #include <sys/kernel.h> |
| 38 | #include <sys/file.h> |
| 39 | #include <sys/stat.h> |
| 40 | #include <sys/buf.h> |
| 41 | #include <sys/proc.h> |
| 42 | #include <sys/mount.h> |
| 43 | #include <sys/namei.h> |
| 44 | #include <sys/vnode.h> |
| 45 | #include <sys/lockf.h> |
| 46 | #include <sys/pool.h> |
| 47 | #include <sys/signalvar.h> |
| 48 | #include <sys/kauth.h> |
| 49 | #include <sys/malloc.h> |
| 50 | #include <ufs/ufs/dir.h> |
| 51 | |
| 52 | #include <ufs/ufs/inode.h> |
| 53 | #include <ufs/ufs/ufsmount.h> |
| 54 | #include <ufs/ext2fs/ext2fs.h> |
| 55 | |
| 56 | #include <ufs/ext2fs/ext2fs_extern.h> |
| 57 | #include <ufs/ext2fs/ext2fs_dinode.h> |
| 58 | #include <ufs/ext2fs/ext2fs_dir.h> |
| 59 | #include <ufs/ext2fs/ext2fs_htree.h> |
| 60 | #include <ufs/ext2fs/ext2fs_hash.h> |
| 61 | |
| 62 | static int ext2fs_htree_find_leaf(struct inode *, const char *, int , |
| 63 | uint32_t *, uint8_t *, struct ext2fs_htree_lookup_info *); |
| 64 | |
| 65 | int |
| 66 | ext2fs_htree_has_idx(struct inode *ip) |
| 67 | { |
| 68 | return EXT2F_HAS_COMPAT_FEATURE(ip->i_e2fs, EXT2F_COMPAT_DIRHASHINDEX) |
| 69 | && (ip->i_din.e2fs_din->e2di_flags & EXT2_INDEX); |
| 70 | } |
| 71 | |
| 72 | static off_t |
| 73 | ext2fs_htree_get_block(struct ext2fs_htree_entry *ep) |
| 74 | { |
| 75 | return ep->h_blk & 0x00FFFFFF; |
| 76 | } |
| 77 | |
| 78 | static void |
| 79 | ext2fs_htree_release(struct ext2fs_htree_lookup_info *info) |
| 80 | { |
| 81 | for (u_int i = 0; i < info->h_levels_num; i++) { |
| 82 | struct buf *bp = info->h_levels[i].h_bp; |
| 83 | if (bp != NULL) |
| 84 | brelse(bp, 0); |
| 85 | } |
| 86 | } |
| 87 | |
| 88 | static uint16_t |
| 89 | ext2fs_htree_get_limit(struct ext2fs_htree_entry *ep) |
| 90 | { |
| 91 | return ((struct ext2fs_htree_count *)(ep))->h_entries_max; |
| 92 | } |
| 93 | |
| 94 | static uint32_t |
| 95 | ext2fs_htree_root_limit(struct inode *ip, int len) |
| 96 | { |
| 97 | uint32_t space = ip->i_e2fs->e2fs_bsize - EXT2_DIR_REC_LEN(1) - |
| 98 | EXT2_DIR_REC_LEN(2) - len; |
| 99 | return space / sizeof(struct ext2fs_htree_entry); |
| 100 | } |
| 101 | |
| 102 | static uint16_t |
| 103 | ext2fs_htree_get_count(struct ext2fs_htree_entry *ep) |
| 104 | { |
| 105 | return ((struct ext2fs_htree_count *)(ep))->h_entries_num; |
| 106 | } |
| 107 | |
| 108 | static uint32_t |
| 109 | ext2fs_htree_get_hash(struct ext2fs_htree_entry *ep) |
| 110 | { |
| 111 | return ep->h_hash; |
| 112 | } |
| 113 | |
| 114 | |
| 115 | static void |
| 116 | ext2fs_htree_set_block(struct ext2fs_htree_entry *ep, uint32_t blk) |
| 117 | { |
| 118 | ep->h_blk = blk; |
| 119 | } |
| 120 | |
| 121 | static void |
| 122 | ext2fs_htree_set_count(struct ext2fs_htree_entry *ep, uint16_t cnt) |
| 123 | { |
| 124 | ((struct ext2fs_htree_count *)(ep))->h_entries_num = cnt; |
| 125 | } |
| 126 | |
| 127 | static void |
| 128 | ext2fs_htree_set_hash(struct ext2fs_htree_entry *ep, uint32_t hash) |
| 129 | { |
| 130 | ep->h_hash = hash; |
| 131 | } |
| 132 | |
| 133 | static void |
| 134 | ext2fs_htree_set_limit(struct ext2fs_htree_entry *ep, uint16_t limit) |
| 135 | { |
| 136 | ((struct ext2fs_htree_count *)(ep))->h_entries_max = limit; |
| 137 | } |
| 138 | |
| 139 | static uint32_t |
| 140 | ext2fs_htree_node_limit(struct inode *ip) |
| 141 | { |
| 142 | struct m_ext2fs *fs; |
| 143 | uint32_t space; |
| 144 | |
| 145 | fs = ip->i_e2fs; |
| 146 | space = fs->e2fs_bsize - EXT2_DIR_REC_LEN(0); |
| 147 | |
| 148 | return space / sizeof(struct ext2fs_htree_entry); |
| 149 | } |
| 150 | |
| 151 | static int |
| 152 | ext2fs_htree_append_block(struct vnode *vp, char *data, |
| 153 | struct componentname *cnp, uint32_t blksize) |
| 154 | { |
| 155 | struct iovec aiov; |
| 156 | struct uio auio; |
| 157 | struct inode *dp = VTOI(vp); |
| 158 | uint64_t cursize, newsize; |
| 159 | int error; |
| 160 | |
| 161 | cursize = roundup(dp->i_size, blksize); |
| 162 | newsize = cursize + blksize; |
| 163 | |
| 164 | auio.uio_offset = cursize; |
| 165 | auio.uio_resid = blksize; |
| 166 | aiov.iov_len = blksize; |
| 167 | aiov.iov_base = data; |
| 168 | auio.uio_iov = &aiov; |
| 169 | auio.uio_iovcnt = 1; |
| 170 | auio.uio_rw = UIO_WRITE; |
| 171 | auio.uio_vmspace = vmspace_kernel(); |
| 172 | error = VOP_WRITE(vp, &auio, IO_SYNC, cnp->cn_cred); |
| 173 | if (!error) |
| 174 | dp->i_size = newsize; |
| 175 | |
| 176 | return error; |
| 177 | } |
| 178 | |
| 179 | static int |
| 180 | ext2fs_htree_writebuf(struct ext2fs_htree_lookup_info *info) |
| 181 | { |
| 182 | int i, error; |
| 183 | |
| 184 | for (i = 0; i < info->h_levels_num; i++) { |
| 185 | struct buf *bp = info->h_levels[i].h_bp; |
| 186 | error = bwrite(bp); |
| 187 | if (error) |
| 188 | return error; |
| 189 | } |
| 190 | |
| 191 | return 0; |
| 192 | } |
| 193 | |
| 194 | static void |
| 195 | ext2fs_htree_insert_entry_to_level(struct ext2fs_htree_lookup_level *level, |
| 196 | uint32_t hash, uint32_t blk) |
| 197 | { |
| 198 | struct ext2fs_htree_entry *target; |
| 199 | int entries_num; |
| 200 | |
| 201 | target = level->h_entry + 1; |
| 202 | entries_num = ext2fs_htree_get_count(level->h_entries); |
| 203 | |
| 204 | memmove(target + 1, target, (char *)(level->h_entries + entries_num) - |
| 205 | (char *)target); |
| 206 | ext2fs_htree_set_block(target, blk); |
| 207 | ext2fs_htree_set_hash(target, hash); |
| 208 | ext2fs_htree_set_count(level->h_entries, entries_num + 1); |
| 209 | } |
| 210 | |
| 211 | /* |
| 212 | * Insert an index entry to the index node. |
| 213 | */ |
| 214 | static void |
| 215 | ext2fs_htree_insert_entry(struct ext2fs_htree_lookup_info *info, |
| 216 | uint32_t hash, uint32_t blk) |
| 217 | { |
| 218 | struct ext2fs_htree_lookup_level *level; |
| 219 | |
| 220 | level = &info->h_levels[info->h_levels_num - 1]; |
| 221 | ext2fs_htree_insert_entry_to_level(level, hash, blk); |
| 222 | } |
| 223 | |
| 224 | /* |
| 225 | * Compare two entry sort descriptors by name hash value. |
| 226 | * This is used together with qsort. |
| 227 | */ |
| 228 | static int |
| 229 | ext2fs_htree_cmp_sort_entry(const void *e1, const void *e2) |
| 230 | { |
| 231 | const struct ext2fs_htree_sort_entry *entry1, *entry2; |
| 232 | |
| 233 | entry1 = (const struct ext2fs_htree_sort_entry *)e1; |
| 234 | entry2 = (const struct ext2fs_htree_sort_entry *)e2; |
| 235 | |
| 236 | if (entry1->h_hash < entry2->h_hash) |
| 237 | return -1; |
| 238 | if (entry1->h_hash > entry2->h_hash) |
| 239 | return 1; |
| 240 | return 0; |
| 241 | } |
| 242 | |
| 243 | /* |
| 244 | * Append an entry to the end of the directory block. |
| 245 | */ |
| 246 | static void |
| 247 | ext2fs_append_entry(char *block, uint32_t blksize, |
| 248 | struct ext2fs_direct *last_entry, struct ext2fs_direct *new_entry) |
| 249 | { |
| 250 | uint16_t entry_len; |
| 251 | |
| 252 | entry_len = EXT2_DIR_REC_LEN(last_entry->e2d_namlen); |
| 253 | last_entry->e2d_reclen = entry_len; |
| 254 | last_entry = (struct ext2fs_direct *)((char *)last_entry + entry_len); |
| 255 | new_entry->e2d_reclen = block + blksize - (char *)last_entry; |
| 256 | memcpy(last_entry, new_entry, EXT2_DIR_REC_LEN(new_entry->e2d_namlen)); |
| 257 | } |
| 258 | |
| 259 | /* |
| 260 | * Move half of entries from the old directory block to the new one. |
| 261 | */ |
| 262 | static int |
| 263 | ext2fs_htree_split_dirblock(char *block1, char *block2, uint32_t blksize, |
| 264 | uint32_t *hash_seed, uint8_t hash_version, |
| 265 | uint32_t *split_hash, struct ext2fs_direct *entry) |
| 266 | { |
| 267 | int entry_cnt = 0; |
| 268 | int size = 0; |
| 269 | int i, k; |
| 270 | uint32_t offset; |
| 271 | uint16_t entry_len = 0; |
| 272 | uint32_t entry_hash; |
| 273 | struct ext2fs_direct *ep, *last; |
| 274 | char *dest; |
| 275 | struct ext2fs_htree_sort_entry *sort_info, dummy; |
| 276 | |
| 277 | ep = (struct ext2fs_direct *)block1; |
| 278 | dest = block2; |
| 279 | sort_info = (struct ext2fs_htree_sort_entry *) |
| 280 | ((char *)block2 + blksize); |
| 281 | |
| 282 | /* |
| 283 | * Calculate name hash value for the entry which is to be added. |
| 284 | */ |
| 285 | ext2fs_htree_hash(entry->e2d_name, entry->e2d_namlen, hash_seed, |
| 286 | hash_version, &entry_hash, NULL); |
| 287 | |
| 288 | /* |
| 289 | * Fill in directory entry sort descriptors. |
| 290 | */ |
| 291 | while ((char *)ep < block1 + blksize) { |
| 292 | if (ep->e2d_ino && ep->e2d_namlen) { |
| 293 | entry_cnt++; |
| 294 | sort_info--; |
| 295 | sort_info->h_size = ep->e2d_reclen; |
| 296 | sort_info->h_offset = (char *)ep - block1; |
| 297 | ext2fs_htree_hash(ep->e2d_name, ep->e2d_namlen, |
| 298 | hash_seed, hash_version, |
| 299 | &sort_info->h_hash, NULL); |
| 300 | } |
| 301 | ep = (struct ext2fs_direct *) |
| 302 | ((char *)ep + ep->e2d_reclen); |
| 303 | } |
| 304 | |
| 305 | /* |
| 306 | * Sort directory entry descriptors by name hash value. |
| 307 | */ |
| 308 | kheapsort(sort_info, entry_cnt, sizeof(struct ext2fs_htree_sort_entry), |
| 309 | ext2fs_htree_cmp_sort_entry, &dummy); |
| 310 | |
| 311 | /* |
| 312 | * Count the number of entries to move to directory block 2. |
| 313 | */ |
| 314 | for (i = entry_cnt - 1; i >= 0; i--) { |
| 315 | if (sort_info[i].h_size + size > blksize / 2) |
| 316 | break; |
| 317 | size += sort_info[i].h_size; |
| 318 | } |
| 319 | |
| 320 | *split_hash = sort_info[i + 1].h_hash; |
| 321 | |
| 322 | /* |
| 323 | * Set collision bit. |
| 324 | */ |
| 325 | if (*split_hash == sort_info[i].h_hash) |
| 326 | *split_hash += 1; |
| 327 | |
| 328 | /* |
| 329 | * Move half of directory entries from block 1 to block 2. |
| 330 | */ |
| 331 | for (k = i + 1; k < entry_cnt; k++) { |
| 332 | ep = (struct ext2fs_direct *)((char *)block1 + |
| 333 | sort_info[k].h_offset); |
| 334 | entry_len = EXT2_DIR_REC_LEN(ep->e2d_namlen); |
| 335 | memcpy(dest, ep, entry_len); |
| 336 | ((struct ext2fs_direct *)dest)->e2d_reclen = entry_len; |
| 337 | /* Mark directory entry as unused. */ |
| 338 | ep->e2d_ino = 0; |
| 339 | dest += entry_len; |
| 340 | } |
| 341 | dest -= entry_len; |
| 342 | |
| 343 | /* Shrink directory entries in block 1. */ |
| 344 | last = (struct ext2fs_direct *)block1; |
| 345 | entry_len = 0; |
| 346 | for (offset = 0; offset < blksize;) { |
| 347 | ep = (struct ext2fs_direct *)(block1 + offset); |
| 348 | offset += ep->e2d_reclen; |
| 349 | if (ep->e2d_ino) { |
| 350 | last = (struct ext2fs_direct *) |
| 351 | ((char *)last + entry_len); |
| 352 | entry_len = EXT2_DIR_REC_LEN(ep->e2d_namlen); |
| 353 | memcpy((void *)last, (void *)ep, entry_len); |
| 354 | last->e2d_reclen = entry_len; |
| 355 | } |
| 356 | } |
| 357 | |
| 358 | if (entry_hash >= *split_hash) { |
| 359 | /* Add entry to block 2. */ |
| 360 | ext2fs_append_entry(block2, blksize, |
| 361 | (struct ext2fs_direct *)dest, entry); |
| 362 | |
| 363 | /* Adjust length field of last entry of block 1. */ |
| 364 | last->e2d_reclen = block1 + blksize - (char *)last; |
| 365 | } else { |
| 366 | /* Add entry to block 1. */ |
| 367 | ext2fs_append_entry(block1, blksize, last, entry); |
| 368 | |
| 369 | /* Adjust length field of last entry of block 2. */ |
| 370 | ((struct ext2fs_direct *)dest)->e2d_reclen = |
| 371 | block2 + blksize - dest; |
| 372 | } |
| 373 | |
| 374 | return 0; |
| 375 | } |
| 376 | |
| 377 | /* |
| 378 | * Create an HTree index for a directory having entries which are no more |
| 379 | * accommodable in a single dir-block. |
| 380 | */ |
| 381 | int |
| 382 | ext2fs_htree_create_index(struct vnode *vp, struct componentname *cnp, |
| 383 | struct ext2fs_direct *new_entry) |
| 384 | { |
| 385 | struct buf *bp = NULL; |
| 386 | struct inode *dp; |
| 387 | struct ext2fs *fs; |
| 388 | struct m_ext2fs *m_fs; |
| 389 | struct ext2fs_direct *ep, *dotdot; |
| 390 | struct ext2fs_htree_root *root; |
| 391 | struct ext2fs_htree_lookup_info info; |
| 392 | uint32_t blksize, dirlen, split_hash; |
| 393 | uint8_t hash_version; |
| 394 | char *buf1 = NULL; |
| 395 | char *buf2 = NULL; |
| 396 | int error = 0; |
| 397 | |
| 398 | dp = VTOI(vp); |
| 399 | fs = &(dp->i_e2fs->e2fs); |
| 400 | m_fs = dp->i_e2fs; |
| 401 | blksize = m_fs->e2fs_bsize; |
| 402 | |
| 403 | buf1 = malloc(blksize, M_TEMP, M_WAITOK | M_ZERO); |
| 404 | buf2 = malloc(blksize, M_TEMP, M_WAITOK | M_ZERO); |
| 405 | |
| 406 | if ((error = ext2fs_blkatoff(vp, 0, NULL, &bp)) != 0) |
| 407 | goto out; |
| 408 | |
| 409 | root = (struct ext2fs_htree_root *)bp->b_data; |
| 410 | dotdot = (struct ext2fs_direct *)((char *)&(root->h_dotdot)); |
| 411 | ep = (struct ext2fs_direct *)((char *)dotdot + dotdot->e2d_reclen); |
| 412 | dirlen = (char *)root + blksize - (char *)ep; |
| 413 | memcpy(buf1, ep, dirlen); |
| 414 | ep = (struct ext2fs_direct *)buf1; |
| 415 | while ((char *)ep < buf1 + dirlen) |
| 416 | ep = (struct ext2fs_direct *)((char *)ep + ep->e2d_reclen); |
| 417 | ep->e2d_reclen = buf1 + blksize - (char *)ep; |
| 418 | /* XXX It should be made dp->i_flag |= IN_E3INDEX; */ |
| 419 | dp->i_din.e2fs_din->e2di_flags |= EXT2_INDEX; |
| 420 | |
| 421 | /* |
| 422 | * Initialize index root. |
| 423 | */ |
| 424 | dotdot->e2d_reclen = blksize - EXT2_DIR_REC_LEN(1); |
| 425 | memset(&root->h_info, 0, sizeof(root->h_info)); |
| 426 | root->h_info.h_hash_version = fs->e3fs_def_hash_version; |
| 427 | root->h_info.h_info_len = sizeof(root->h_info); |
| 428 | ext2fs_htree_set_block(root->h_entries, 1); |
| 429 | ext2fs_htree_set_count(root->h_entries, 1); |
| 430 | ext2fs_htree_set_limit(root->h_entries, |
| 431 | ext2fs_htree_root_limit(dp, sizeof(root->h_info))); |
| 432 | |
| 433 | memset(&info, 0, sizeof(info)); |
| 434 | info.h_levels_num = 1; |
| 435 | info.h_levels[0].h_entries = root->h_entries; |
| 436 | info.h_levels[0].h_entry = root->h_entries; |
| 437 | |
| 438 | hash_version = root->h_info.h_hash_version; |
| 439 | if (hash_version <= EXT2_HTREE_TEA) |
| 440 | hash_version += m_fs->e2fs_uhash; |
| 441 | ext2fs_htree_split_dirblock(buf1, buf2, blksize, fs->e3fs_hash_seed, |
| 442 | hash_version, &split_hash, new_entry); |
| 443 | ext2fs_htree_insert_entry(&info, split_hash, 2); |
| 444 | |
| 445 | /* |
| 446 | * Write directory block 0. |
| 447 | */ |
| 448 | if ( (vp)->v_mount->mnt_iflag & IO_SYNC) |
| 449 | (void)bwrite(bp); |
| 450 | else |
| 451 | bdwrite(bp); |
| 452 | |
| 453 | dp->i_flag |= IN_CHANGE | IN_UPDATE; |
| 454 | |
| 455 | /* |
| 456 | * Write directory block 1. |
| 457 | */ |
| 458 | error = ext2fs_htree_append_block(vp, buf1, cnp, blksize); |
| 459 | if (error) |
| 460 | goto out1; |
| 461 | |
| 462 | /* |
| 463 | * Write directory block 2. |
| 464 | */ |
| 465 | error = ext2fs_htree_append_block(vp, buf2, cnp, blksize); |
| 466 | goto out1; |
| 467 | out: |
| 468 | if (bp != NULL) |
| 469 | brelse(bp, 0); |
| 470 | out1: |
| 471 | free(buf1, M_TEMP); |
| 472 | free(buf2, M_TEMP); |
| 473 | return error; |
| 474 | } |
| 475 | |
| 476 | /* |
| 477 | * Add an entry to the directory using htree index. |
| 478 | */ |
| 479 | int |
| 480 | ext2fs_htree_add_entry(struct vnode *dvp, struct ext2fs_direct *entry, |
| 481 | struct componentname *cnp, size_t newentrysize) |
| 482 | { |
| 483 | struct ext2fs_htree_entry *entries, *leaf_node; |
| 484 | struct ext2fs_htree_lookup_info info; |
| 485 | struct buf *bp = NULL; |
| 486 | struct ext2fs *fs; |
| 487 | struct m_ext2fs *m_fs; |
| 488 | struct inode *ip; |
| 489 | uint16_t ent_num; |
| 490 | uint32_t dirhash, split_hash; |
| 491 | uint32_t blksize, blknum; |
| 492 | uint64_t cursize, dirsize; |
| 493 | uint8_t hash_version; |
| 494 | char *newdirblock = NULL; |
| 495 | char *newidxblock = NULL; |
| 496 | struct ext2fs_htree_node *dst_node; |
| 497 | struct ext2fs_htree_entry *dst_entries; |
| 498 | struct ext2fs_htree_entry *root_entires; |
| 499 | struct buf *dst_bp = NULL; |
| 500 | int error, write_bp = 0, write_dst_bp = 0, write_info = 0; |
| 501 | |
| 502 | ip = VTOI(dvp); |
| 503 | m_fs = ip->i_e2fs; |
| 504 | fs = &(m_fs->e2fs); |
| 505 | blksize = m_fs->e2fs_bsize; |
| 506 | |
| 507 | if (ip->i_crap.ulr_count != 0) |
| 508 | return ext2fs_add_entry(dvp, entry, &(ip->i_crap), newentrysize); |
| 509 | |
| 510 | /* Target directory block is full, split it */ |
| 511 | memset(&info, 0, sizeof(info)); |
| 512 | error = ext2fs_htree_find_leaf(ip, entry->e2d_name, entry->e2d_namlen, |
| 513 | &dirhash, &hash_version, &info); |
| 514 | if (error) |
| 515 | return error; |
| 516 | entries = info.h_levels[info.h_levels_num - 1].h_entries; |
| 517 | ent_num = ext2fs_htree_get_count(entries); |
| 518 | if (ent_num == ext2fs_htree_get_limit(entries)) { |
| 519 | /* Split the index node. */ |
| 520 | root_entires = info.h_levels[0].h_entries; |
| 521 | newidxblock = malloc(blksize, M_TEMP, M_WAITOK | M_ZERO); |
| 522 | dst_node = (struct ext2fs_htree_node *)newidxblock; |
| 523 | dst_entries = dst_node->h_entries; |
| 524 | memset(&dst_node->h_fake_dirent, 0, |
| 525 | sizeof(dst_node->h_fake_dirent)); |
| 526 | dst_node->h_fake_dirent.e2d_reclen = blksize; |
| 527 | |
| 528 | cursize = roundup(ip->i_size, blksize); |
| 529 | dirsize = cursize + blksize; |
| 530 | blknum = dirsize / blksize - 1; |
| 531 | |
| 532 | error = ext2fs_htree_append_block(dvp, newidxblock, |
| 533 | cnp, blksize); |
| 534 | if (error) |
| 535 | goto finish; |
| 536 | error = ext2fs_blkatoff(dvp, cursize, NULL, &dst_bp); |
| 537 | if (error) |
| 538 | goto finish; |
| 539 | |
| 540 | dst_node = (struct ext2fs_htree_node *)dst_bp->b_data; |
| 541 | dst_entries = dst_node->h_entries; |
| 542 | |
| 543 | if (info.h_levels_num == 2) { |
| 544 | uint16_t src_ent_num, dst_ent_num; |
| 545 | |
| 546 | if (ext2fs_htree_get_count(root_entires) == |
| 547 | ext2fs_htree_get_limit(root_entires)) { |
| 548 | /* Directory index is full */ |
| 549 | error = EIO; |
| 550 | goto finish; |
| 551 | } |
| 552 | |
| 553 | src_ent_num = ent_num / 2; |
| 554 | dst_ent_num = ent_num - src_ent_num; |
| 555 | split_hash = ext2fs_htree_get_hash(entries + src_ent_num); |
| 556 | |
| 557 | /* Move half of index entries to the new index node */ |
| 558 | memcpy(dst_entries, entries + src_ent_num, |
| 559 | dst_ent_num * sizeof(struct ext2fs_htree_entry)); |
| 560 | ext2fs_htree_set_count(entries, src_ent_num); |
| 561 | ext2fs_htree_set_count(dst_entries, dst_ent_num); |
| 562 | ext2fs_htree_set_limit(dst_entries, |
| 563 | ext2fs_htree_node_limit(ip)); |
| 564 | |
| 565 | if (info.h_levels[1].h_entry >= entries + src_ent_num) { |
| 566 | struct buf *tmp = info.h_levels[1].h_bp; |
| 567 | info.h_levels[1].h_bp = dst_bp; |
| 568 | dst_bp = tmp; |
| 569 | |
| 570 | info.h_levels[1].h_entry = |
| 571 | info.h_levels[1].h_entry - |
| 572 | (entries + src_ent_num) + |
| 573 | dst_entries; |
| 574 | info.h_levels[1].h_entries = dst_entries; |
| 575 | } |
| 576 | ext2fs_htree_insert_entry_to_level(&info.h_levels[0], |
| 577 | split_hash, blknum); |
| 578 | |
| 579 | /* Write new index node to disk */ |
| 580 | error = bwrite(dst_bp); |
| 581 | ip->i_flag |= IN_CHANGE | IN_UPDATE; |
| 582 | if (error) |
| 583 | goto finish; |
| 584 | |
| 585 | write_dst_bp = 1; |
| 586 | } else { |
| 587 | /* Create second level for htree index */ |
| 588 | |
| 589 | struct ext2fs_htree_root *idx_root; |
| 590 | |
| 591 | memcpy(dst_entries, entries, |
| 592 | ent_num * sizeof(struct ext2fs_htree_entry)); |
| 593 | ext2fs_htree_set_limit(dst_entries, |
| 594 | ext2fs_htree_node_limit(ip)); |
| 595 | |
| 596 | idx_root = (struct ext2fs_htree_root *) |
| 597 | info.h_levels[0].h_bp->b_data; |
| 598 | idx_root->h_info.h_ind_levels = 1; |
| 599 | |
| 600 | ext2fs_htree_set_count(entries, 1); |
| 601 | ext2fs_htree_set_block(entries, blknum); |
| 602 | |
| 603 | info.h_levels_num = 2; |
| 604 | info.h_levels[1].h_entries = dst_entries; |
| 605 | info.h_levels[1].h_entry = info.h_levels[0].h_entry - |
| 606 | info.h_levels[0].h_entries + dst_entries; |
| 607 | info.h_levels[1].h_bp = dst_bp; |
| 608 | dst_bp = NULL; |
| 609 | } |
| 610 | } |
| 611 | |
| 612 | leaf_node = info.h_levels[info.h_levels_num - 1].h_entry; |
| 613 | blknum = ext2fs_htree_get_block(leaf_node); |
| 614 | error = ext2fs_blkatoff(dvp, blknum * blksize, NULL, &bp); |
| 615 | if (error) |
| 616 | goto finish; |
| 617 | |
| 618 | /* Split target directory block */ |
| 619 | newdirblock = malloc(blksize, M_TEMP, M_WAITOK | M_ZERO); |
| 620 | ext2fs_htree_split_dirblock((char *)bp->b_data, newdirblock, blksize, |
| 621 | fs->e3fs_hash_seed, hash_version, &split_hash, entry); |
| 622 | cursize = roundup(ip->i_size, blksize); |
| 623 | dirsize = cursize + blksize; |
| 624 | blknum = dirsize / blksize - 1; |
| 625 | |
| 626 | /* Add index entry for the new directory block */ |
| 627 | ext2fs_htree_insert_entry(&info, split_hash, blknum); |
| 628 | |
| 629 | /* Write the new directory block to the end of the directory */ |
| 630 | error = ext2fs_htree_append_block(dvp, newdirblock, cnp, blksize); |
| 631 | |
| 632 | if (error) |
| 633 | goto finish; |
| 634 | |
| 635 | /* Write the target directory block */ |
| 636 | error = bwrite(bp); |
| 637 | ip->i_flag |= IN_CHANGE | IN_UPDATE; |
| 638 | |
| 639 | if (error) |
| 640 | goto finish; |
| 641 | write_bp = 1; |
| 642 | |
| 643 | /* Write the index block */ |
| 644 | error = ext2fs_htree_writebuf(&info); |
| 645 | if (!error) |
| 646 | write_info = 1; |
| 647 | |
| 648 | finish: |
| 649 | if (dst_bp != NULL && !write_dst_bp) |
| 650 | brelse(dst_bp, 0); |
| 651 | if (bp != NULL && !write_bp) |
| 652 | brelse(bp, 0); |
| 653 | if (newdirblock != NULL) |
| 654 | free(newdirblock, M_TEMP); |
| 655 | if (newidxblock != NULL) |
| 656 | free(newidxblock, M_TEMP); |
| 657 | if (!write_info) |
| 658 | ext2fs_htree_release(&info); |
| 659 | return error; |
| 660 | } |
| 661 | |
| 662 | static int |
| 663 | ext2fs_htree_check_next(struct inode *ip, uint32_t hash, const char *name, |
| 664 | struct ext2fs_htree_lookup_info *info) |
| 665 | { |
| 666 | struct vnode *vp = ITOV(ip); |
| 667 | struct ext2fs_htree_lookup_level *level; |
| 668 | struct buf *bp; |
| 669 | uint32_t next_hash; |
| 670 | int idx = info->h_levels_num - 1; |
| 671 | int levels = 0; |
| 672 | |
| 673 | for (;;) { |
| 674 | level = &info->h_levels[idx]; |
| 675 | level->h_entry++; |
| 676 | if (level->h_entry < level->h_entries + |
| 677 | ext2fs_htree_get_count(level->h_entries)) |
| 678 | break; |
| 679 | if (idx == 0) |
| 680 | return 0; |
| 681 | idx--; |
| 682 | levels++; |
| 683 | } |
| 684 | |
| 685 | next_hash = ext2fs_htree_get_hash(level->h_entry); |
| 686 | if ((hash & 1) == 0) { |
| 687 | if (hash != (next_hash & ~1)) |
| 688 | return 0; |
| 689 | } |
| 690 | |
| 691 | while (levels > 0) { |
| 692 | levels--; |
| 693 | if (ext2fs_blkatoff(vp, ext2fs_htree_get_block(level->h_entry) * |
| 694 | ip->i_e2fs->e2fs_bsize, NULL, &bp) != 0) |
| 695 | return 0; |
| 696 | level = &info->h_levels[idx + 1]; |
| 697 | brelse(level->h_bp, 0); |
| 698 | level->h_bp = bp; |
| 699 | level->h_entry = level->h_entries = |
| 700 | ((struct ext2fs_htree_node *)bp->b_data)->h_entries; |
| 701 | } |
| 702 | |
| 703 | return 1; |
| 704 | } |
| 705 | |
| 706 | static int |
| 707 | ext2fs_htree_find_leaf(struct inode *ip, const char *name, int namelen, |
| 708 | uint32_t *hash, uint8_t *hash_ver, |
| 709 | struct ext2fs_htree_lookup_info *info) |
| 710 | { |
| 711 | struct vnode *vp; |
| 712 | struct ext2fs *fs;/* F, G, and H are MD4 functions */ |
| 713 | struct m_ext2fs *m_fs; |
| 714 | struct buf *bp = NULL; |
| 715 | struct ext2fs_htree_root *rootp; |
| 716 | struct ext2fs_htree_entry *entp, *start, *end, *middle, *found; |
| 717 | struct ext2fs_htree_lookup_level *level_info; |
| 718 | uint32_t hash_major = 0, hash_minor = 0; |
| 719 | uint32_t levels, cnt; |
| 720 | uint8_t hash_version; |
| 721 | |
| 722 | if (name == NULL || info == NULL) |
| 723 | return -1; |
| 724 | |
| 725 | vp = ITOV(ip); |
| 726 | fs = &(ip->i_e2fs->e2fs); |
| 727 | m_fs = ip->i_e2fs; |
| 728 | |
| 729 | if (ext2fs_blkatoff(vp, 0, NULL, &bp) != 0) |
| 730 | return -1; |
| 731 | |
| 732 | info->h_levels_num = 1; |
| 733 | info->h_levels[0].h_bp = bp; |
| 734 | rootp = (struct ext2fs_htree_root *)bp->b_data; |
| 735 | if (rootp->h_info.h_hash_version != EXT2_HTREE_LEGACY && |
| 736 | rootp->h_info.h_hash_version != EXT2_HTREE_HALF_MD4 && |
| 737 | rootp->h_info.h_hash_version != EXT2_HTREE_TEA) |
| 738 | goto error; |
| 739 | |
| 740 | hash_version = rootp->h_info.h_hash_version; |
| 741 | if (hash_version <= EXT2_HTREE_TEA) |
| 742 | hash_version += m_fs->e2fs_uhash; |
| 743 | *hash_ver = hash_version; |
| 744 | |
| 745 | ext2fs_htree_hash(name, namelen, fs->e3fs_hash_seed, |
| 746 | hash_version, &hash_major, &hash_minor); |
| 747 | *hash = hash_major; |
| 748 | |
| 749 | if ((levels = rootp->h_info.h_ind_levels) > 1) |
| 750 | goto error; |
| 751 | |
| 752 | entp = (struct ext2fs_htree_entry *)(((char *)&rootp->h_info) + |
| 753 | rootp->h_info.h_info_len); |
| 754 | |
| 755 | if (ext2fs_htree_get_limit(entp) != |
| 756 | ext2fs_htree_root_limit(ip, rootp->h_info.h_info_len)) |
| 757 | goto error; |
| 758 | |
| 759 | for (;;) { |
| 760 | cnt = ext2fs_htree_get_count(entp); |
| 761 | if (cnt == 0 || cnt > ext2fs_htree_get_limit(entp)) |
| 762 | goto error; |
| 763 | |
| 764 | start = entp + 1; |
| 765 | end = entp + cnt - 1; |
| 766 | while (start <= end) { |
| 767 | middle = start + (end - start) / 2; |
| 768 | if (ext2fs_htree_get_hash(middle) > hash_major) |
| 769 | end = middle - 1; |
| 770 | else |
| 771 | start = middle + 1; |
| 772 | } |
| 773 | found = start - 1; |
| 774 | |
| 775 | level_info = &(info->h_levels[info->h_levels_num - 1]); |
| 776 | level_info->h_bp = bp; |
| 777 | level_info->h_entries = entp; |
| 778 | level_info->h_entry = found; |
| 779 | if (levels == 0) |
| 780 | return 0; |
| 781 | levels--; |
| 782 | if (ext2fs_blkatoff(vp, |
| 783 | ext2fs_htree_get_block(found) * m_fs->e2fs_bsize, |
| 784 | NULL, &bp) != 0) |
| 785 | goto error; |
| 786 | entp = ((struct ext2fs_htree_node *)bp->b_data)->h_entries; |
| 787 | info->h_levels_num++; |
| 788 | info->h_levels[info->h_levels_num - 1].h_bp = bp; |
| 789 | } |
| 790 | |
| 791 | error: |
| 792 | ext2fs_htree_release(info); |
| 793 | return -1; |
| 794 | } |
| 795 | |
| 796 | /* |
| 797 | * Try to lookup a directory entry in HTree index |
| 798 | */ |
| 799 | int |
| 800 | ext2fs_htree_lookup(struct inode *ip, const char *name, int namelen, |
| 801 | struct buf **bpp, int *entryoffp, doff_t *offp, |
| 802 | doff_t *prevoffp, doff_t *endusefulp, struct ext2fs_searchslot *ss) |
| 803 | { |
| 804 | struct vnode *vp; |
| 805 | struct ext2fs_htree_lookup_info info; |
| 806 | struct ext2fs_htree_entry *leaf_node; |
| 807 | struct m_ext2fs *m_fs; |
| 808 | struct buf *bp; |
| 809 | uint32_t blk; |
| 810 | uint32_t dirhash; |
| 811 | uint32_t bsize; |
| 812 | uint8_t hash_version; |
| 813 | int search_next; |
| 814 | |
| 815 | m_fs = ip->i_e2fs; |
| 816 | bsize = m_fs->e2fs_bsize; |
| 817 | vp = ITOV(ip); |
| 818 | |
| 819 | /* TODO: print error msg because we don't lookup '.' and '..' */ |
| 820 | |
| 821 | memset(&info, 0, sizeof(info)); |
| 822 | if (ext2fs_htree_find_leaf(ip, name, namelen, &dirhash, |
| 823 | &hash_version, &info)) { |
| 824 | return -1; |
| 825 | } |
| 826 | |
| 827 | do { |
| 828 | leaf_node = info.h_levels[info.h_levels_num - 1].h_entry; |
| 829 | blk = ext2fs_htree_get_block(leaf_node); |
| 830 | if (ext2fs_blkatoff(vp, blk * bsize, NULL, &bp) != 0) { |
| 831 | ext2fs_htree_release(&info); |
| 832 | return -1; |
| 833 | } |
| 834 | |
| 835 | *offp = blk * bsize; |
| 836 | *entryoffp = 0; |
| 837 | *prevoffp = blk * bsize; |
| 838 | *endusefulp = blk * bsize; |
| 839 | |
| 840 | if (ss->slotstatus == NONE) { |
| 841 | ss->slotoffset = -1; |
| 842 | ss->slotfreespace = 0; |
| 843 | } |
| 844 | |
| 845 | int found; |
| 846 | if (ext2fs_search_dirblock(ip, bp->b_data, &found, |
| 847 | name, namelen, entryoffp, offp, prevoffp, |
| 848 | endusefulp, ss) != 0) { |
| 849 | brelse(bp, 0); |
| 850 | ext2fs_htree_release(&info); |
| 851 | return -1; |
| 852 | } |
| 853 | |
| 854 | if (found) { |
| 855 | *bpp = bp; |
| 856 | ext2fs_htree_release(&info); |
| 857 | return 0; |
| 858 | } |
| 859 | |
| 860 | brelse(bp, 0); |
| 861 | search_next = ext2fs_htree_check_next(ip, dirhash, name, &info); |
| 862 | } while (search_next); |
| 863 | |
| 864 | ext2fs_htree_release(&info); |
| 865 | return ENOENT; |
| 866 | } |
| 867 | |