1 /* $OpenBSD: mke2fs.c,v 1.6 2013/12/27 19:17:28 deraadt Exp $ */ 2 /* $NetBSD: mke2fs.c,v 1.13 2009/10/19 18:41:08 bouyer Exp $ */ 3 4 /*- 5 * Copyright (c) 2007 Izumi Tsutsui. All rights reserved. 6 * 7 * Redistribution and use in source and binary forms, with or without 8 * modification, are permitted provided that the following conditions 9 * are met: 10 * 1. Redistributions of source code must retain the above copyright 11 * notice, this list of conditions and the following disclaimer. 12 * 2. Redistributions in binary form must reproduce the above copyright 13 * notice, this list of conditions and the following disclaimer in the 14 * documentation and/or other materials provided with the distribution. 15 * 16 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR 17 * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES 18 * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. 19 * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, 20 * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT 21 * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, 22 * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY 23 * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT 24 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF 25 * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. 26 */ 27 28 /* 29 * Copyright (c) 1980, 1989, 1993 30 * The Regents of the University of California. All rights reserved. 31 * 32 * Redistribution and use in source and binary forms, with or without 33 * modification, are permitted provided that the following conditions 34 * are met: 35 * 1. Redistributions of source code must retain the above copyright 36 * notice, this list of conditions and the following disclaimer. 37 * 2. Redistributions in binary form must reproduce the above copyright 38 * notice, this list of conditions and the following disclaimer in the 39 * documentation and/or other materials provided with the distribution. 40 * 3. Neither the name of the University nor the names of its contributors 41 * may be used to endorse or promote products derived from this software 42 * without specific prior written permission. 43 * 44 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 45 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 46 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 47 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 48 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 49 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 50 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 51 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 52 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 53 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 54 * SUCH DAMAGE. 55 */ 56 57 /* 58 * Copyright (c) 1997 Manuel Bouyer. 59 * 60 * Redistribution and use in source and binary forms, with or without 61 * modification, are permitted provided that the following conditions 62 * are met: 63 * 1. Redistributions of source code must retain the above copyright 64 * notice, this list of conditions and the following disclaimer. 65 * 2. Redistributions in binary form must reproduce the above copyright 66 * notice, this list of conditions and the following disclaimer in the 67 * documentation and/or other materials provided with the distribution. 68 * 69 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR 70 * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES 71 * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. 72 * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, 73 * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT 74 * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, 75 * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY 76 * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT 77 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF 78 * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. 79 */ 80 81 /* 82 * mke2fs.c: "re-invent (dumb but non-GPLed) wheel as a fun project" 83 * 84 * In spite of this name, there is no piece of code 85 * derived from GPLed e2fsprogs written for Linux. 86 * I referred them only to see how each structure 87 * member should be initialized. 88 * 89 * Reference: 90 * - All NetBSD sources under src/sys/ufs/ext2fs and src/sbin/fsck_ext2fs 91 * - Ext2fs Home Page 92 * http://e2fsprogs.sourceforge.net/ext2.html 93 * - Design and Implementation of the Second Extended Filesystem 94 * http://e2fsprogs.sourceforge.net/ext2intro.html 95 * - Linux Documentation "The Second Extended Filesystem" 96 * src/linux/Documentation/filesystems/ext2.txt 97 * in the Linux kernel distribution 98 */ 99 100 #include <sys/param.h> 101 #include <sys/mman.h> 102 #include <sys/time.h> 103 #include <ufs/ext2fs/ext2fs_dinode.h> 104 #include <ufs/ext2fs/ext2fs_dir.h> 105 #include <ufs/ext2fs/ext2fs.h> 106 #include <sys/ioctl.h> 107 108 #include <err.h> 109 #include <errno.h> 110 #include <inttypes.h> 111 #include <string.h> 112 #include <unistd.h> 113 #include <stdint.h> 114 #include <stdlib.h> 115 #include <stddef.h> 116 #include <stdio.h> 117 118 #include "extern.h" 119 120 static void initcg(uint); 121 static void zap_old_sblock(daddr32_t); 122 static uint cgoverhead(uint); 123 static int fsinit(const struct timeval *); 124 static int makedir(struct ext2fs_direct *, int); 125 static void copy_dir(struct ext2fs_direct *, struct ext2fs_direct *); 126 static void init_resizeino(const struct timeval *); 127 static uint32_t alloc(uint32_t, uint16_t); 128 static void iput(struct ext2fs_dinode *, ino_t); 129 static void rdfs(daddr32_t, int, void *); 130 static void wtfs(daddr32_t, int, void *); 131 static int ilog2(uint); 132 static int skpc(int, size_t, uint8_t *); 133 static void uuid_get(struct m_ext2fs *); 134 135 /* XXX: some of these macro should be into <ufs/ext2fs/ext2fs.h>? */ 136 #define EXT2_DEF_MAX_MNT_COUNT 20 137 #define EXT2_DEF_FSCKINTV (180 * 24 * 60 * 60) /* 180 days */ 138 #define EXT2_RESERVED_INODES (EXT2_FIRSTINO - 1) 139 #define EXT2_UMASK 0755 140 141 #define EXT2_INO_INDEX(ino) ((ino) - 1) /* no inode zero */ 142 143 #define EXT2_LOSTFOUNDSIZE 16384 144 #define EXT2_LOSTFOUNDINO EXT2_FIRSTINO /* XXX: not quite */ 145 #define EXT2_LOSTFOUNDUMASK 0700 146 147 #define EXT2_RESIZEINOUMASK 0600 148 149 #define NBLOCK_SUPERBLOCK 1 150 #define NBLOCK_BLOCK_BITMAP 1 151 #define NBLOCK_INODE_BITMAP 1 152 153 #define cgbase(fs, c) \ 154 ((fs)->e2fs.e2fs_first_dblock + (fs)->e2fs.e2fs_bpg * (c)) 155 156 #define rounddown(x,y) (((x)/(y))*(y)) 157 158 /* 159 * ext2fs super block and group descriptor structures 160 * 161 * We don't have to use or setup whole in-memory m_ext2fs structure, 162 * but prepare it to use several macro defined in kernel headers. 163 */ 164 union { 165 struct m_ext2fs m_ext2fs; 166 char pad[SBSIZE]; 167 } ext2fsun; 168 #define sblock ext2fsun.m_ext2fs 169 #define gd ext2fsun.m_ext2fs.e2fs_gd 170 171 static uint8_t *iobuf; /* for superblock and group descriptors */ 172 static int iobufsize; 173 174 static uint8_t buf[MAXBSIZE]; /* for initcg() and makedir() ops */ 175 176 static int fsi, fso; 177 178 void 179 mke2fs(const char *fsys, int fi, int fo) 180 { 181 struct timeval tv; 182 int64_t minfssize; 183 uint bcount, fbcount, ficount; 184 uint blocks_gd, blocks_per_cg, inodes_per_cg, iblocks_per_cg; 185 uint minblocks_per_cg, blocks_lastcg; 186 uint ncg, cylno, sboff; 187 int i, len, col, delta, fld_width, max_cols; 188 struct winsize winsize; 189 190 gettimeofday(&tv, NULL); 191 fsi = fi; 192 fso = fo; 193 194 /* 195 * collect and verify the block and fragment sizes 196 */ 197 if (!powerof2(bsize)) { 198 errx(EXIT_FAILURE, 199 "block size must be a power of 2, not %u\n", 200 bsize); 201 } 202 if (!powerof2(fsize)) { 203 errx(EXIT_FAILURE, 204 "fragment size must be a power of 2, not %u\n", 205 fsize); 206 } 207 if (fsize < sectorsize) { 208 errx(EXIT_FAILURE, 209 "fragment size %u is too small, minimum is %u\n", 210 fsize, sectorsize); 211 } 212 if (bsize < MINBSIZE) { 213 errx(EXIT_FAILURE, 214 "block size %u is too small, minimum is %u\n", 215 bsize, MINBSIZE); 216 } 217 if (bsize > EXT2_MAXBSIZE) { 218 errx(EXIT_FAILURE, 219 "block size %u is too large, maximum is %u\n", 220 bsize, MAXBSIZE); 221 } 222 if (bsize != fsize) { 223 /* 224 * There is no fragment support on current ext2fs (yet?), 225 * but some kernel code refers fsize or fpg as bsize or bpg 226 * and Linux seems to set the same values to them. 227 */ 228 errx(EXIT_FAILURE, 229 "block size (%u) can't be different from " 230 "fragment size (%u)\n", 231 bsize, fsize); 232 } 233 234 /* variable inodesize is REV1 feature */ 235 if (Oflag == 0 && inodesize != EXT2_REV0_DINODE_SIZE) { 236 errx(EXIT_FAILURE, "GOOD_OLD_REV file system format" 237 " doesn't support %d byte inode\n", inodesize); 238 } 239 240 sblock.e2fs.e2fs_log_bsize = ilog2(bsize) - LOG_MINBSIZE; 241 /* Umm, why not e2fs_log_fsize? */ 242 sblock.e2fs.e2fs_fsize = ilog2(fsize) - LOG_MINBSIZE; 243 244 sblock.e2fs_bsize = bsize; 245 sblock.e2fs_bshift = sblock.e2fs.e2fs_log_bsize + LOG_MINBSIZE; 246 sblock.e2fs_qbmask = sblock.e2fs_bsize - 1; 247 sblock.e2fs_bmask = ~sblock.e2fs_qbmask; 248 sblock.e2fs_fsbtodb = ilog2(sblock.e2fs_bsize) - ilog2(sectorsize); 249 sblock.e2fs_ipb = sblock.e2fs_bsize / inodesize; 250 251 /* 252 * Ext2fs preserves BBSIZE (1024 bytes) space at the top for 253 * bootloader (though it is not enough at all for our bootloader). 254 * If bsize == BBSIZE we have to preserve one block. 255 * If bsize > BBSIZE, the first block already contains BBSIZE space 256 * before superblock because superblock is allocated at SBOFF and 257 * bsize is a power of two (i.e. 2048 bytes or more). 258 */ 259 sblock.e2fs.e2fs_first_dblock = (sblock.e2fs_bsize > BBSIZE) ? 0 : 1; 260 minfssize = fsbtodb(&sblock, 261 sblock.e2fs.e2fs_first_dblock + 262 NBLOCK_SUPERBLOCK + 263 1 /* at least one group descriptor */ + 264 NBLOCK_BLOCK_BITMAP + 265 NBLOCK_INODE_BITMAP + 266 1 /* at least one inode table block */ + 267 1 /* at least one data block for rootdir */ + 268 1 /* at least one data block for data */ 269 ); /* XXX and more? */ 270 271 if (fssize < minfssize) 272 errx(EXIT_FAILURE, "Filesystem size %" PRId64 273 " < minimum size of %" PRId64 "\n", fssize, minfssize); 274 275 bcount = dbtofsb(&sblock, fssize); 276 277 /* 278 * While many people claim that ext2fs is a (bad) clone of ufs/ffs, 279 * it isn't actual ffs so maybe we should call it "block group" 280 * as their native name rather than ffs derived "cylinder group." 281 * But we'll use the latter here since other kernel sources use it. 282 * (I also agree "cylinder" based allocation is obsolete though) 283 */ 284 285 /* maybe "simple is the best" */ 286 blocks_per_cg = sblock.e2fs_bsize * NBBY; 287 288 ncg = howmany(bcount - sblock.e2fs.e2fs_first_dblock, blocks_per_cg); 289 blocks_gd = howmany(sizeof(struct ext2_gd) * ncg, bsize); 290 291 /* check range of inode number */ 292 if (num_inodes < EXT2_FIRSTINO) 293 num_inodes = EXT2_FIRSTINO; /* needs reserved inodes + 1 */ 294 if (num_inodes > UINT16_MAX * ncg) 295 num_inodes = UINT16_MAX * ncg; /* ext2bgd_nifree is uint16_t */ 296 297 inodes_per_cg = num_inodes / ncg; 298 iblocks_per_cg = howmany(inodesize * inodes_per_cg, bsize); 299 300 /* Check that the last cylinder group has enough space for inodes */ 301 minblocks_per_cg = 302 NBLOCK_BLOCK_BITMAP + 303 NBLOCK_INODE_BITMAP + 304 iblocks_per_cg + 305 1; /* at least one data block */ 306 if (Oflag == 0 || cg_has_sb(ncg - 1) != 0) 307 minblocks_per_cg += NBLOCK_SUPERBLOCK + blocks_gd; 308 309 blocks_lastcg = bcount - sblock.e2fs.e2fs_first_dblock - 310 blocks_per_cg * (ncg - 1); 311 if (blocks_lastcg < minblocks_per_cg) { 312 /* 313 * Since we make all the cylinder groups the same size, the 314 * last will only be small if there are more than one 315 * cylinder groups. If the last one is too small to store 316 * filesystem data, just kill it. 317 * 318 * XXX: Does fsck_ext2fs(8) properly handle this case? 319 */ 320 bcount -= blocks_lastcg; 321 ncg--; 322 blocks_lastcg = blocks_per_cg; 323 blocks_gd = howmany(sizeof(struct ext2_gd) * ncg, bsize); 324 inodes_per_cg = num_inodes / ncg; 325 } 326 /* roundup inodes_per_cg to make it use whole inode table blocks */ 327 inodes_per_cg = roundup(inodes_per_cg, sblock.e2fs_ipb); 328 num_inodes = inodes_per_cg * ncg; 329 iblocks_per_cg = inodes_per_cg / sblock.e2fs_ipb; 330 331 /* XXX: probably we should check these adjusted values again */ 332 333 sblock.e2fs.e2fs_bcount = bcount; 334 sblock.e2fs.e2fs_icount = num_inodes; 335 336 sblock.e2fs_ncg = ncg; 337 sblock.e2fs_ngdb = blocks_gd; 338 sblock.e2fs_itpg = iblocks_per_cg; 339 340 sblock.e2fs.e2fs_rbcount = sblock.e2fs.e2fs_bcount * minfree / 100; 341 /* e2fs_fbcount will be accounted later */ 342 /* e2fs_ficount will be accounted later */ 343 344 sblock.e2fs.e2fs_bpg = blocks_per_cg; 345 sblock.e2fs.e2fs_fpg = blocks_per_cg; 346 347 sblock.e2fs.e2fs_ipg = inodes_per_cg; 348 349 sblock.e2fs.e2fs_mtime = 0; 350 sblock.e2fs.e2fs_wtime = (u_int32_t)tv.tv_sec; 351 sblock.e2fs.e2fs_mnt_count = 0; 352 /* XXX: should add some entropy to avoid checking all fs at once? */ 353 sblock.e2fs.e2fs_max_mnt_count = EXT2_DEF_MAX_MNT_COUNT; 354 355 sblock.e2fs.e2fs_magic = E2FS_MAGIC; 356 sblock.e2fs.e2fs_state = E2FS_ISCLEAN; 357 sblock.e2fs.e2fs_beh = E2FS_BEH_DEFAULT; 358 sblock.e2fs.e2fs_minrev = 0; 359 sblock.e2fs.e2fs_lastfsck = (u_int32_t)tv.tv_sec; 360 sblock.e2fs.e2fs_fsckintv = EXT2_DEF_FSCKINTV; 361 362 /* 363 * Maybe we can use E2FS_OS_FREEBSD here and it would be more proper, 364 * but the purpose of this newfs_ext2fs(8) command is to provide 365 * a filesystem which can be recognized by firmware on some 366 * Linux based appliances that can load bootstrap files only from 367 * (their native) ext2fs, and anyway we will (and should) try to 368 * act like them as much as possible. 369 * 370 * Anyway, I hope that all newer such boxes will keep their support 371 * for the "GOOD_OLD_REV" ext2fs. 372 */ 373 sblock.e2fs.e2fs_creator = E2FS_OS_LINUX; 374 375 if (Oflag == 0) { 376 sblock.e2fs.e2fs_rev = E2FS_REV0; 377 sblock.e2fs.e2fs_features_compat = 0; 378 sblock.e2fs.e2fs_features_incompat = 0; 379 sblock.e2fs.e2fs_features_rocompat = 0; 380 } else { 381 sblock.e2fs.e2fs_rev = E2FS_REV1; 382 /* 383 * e2fsprogs say "REV1" is "dynamic" so 384 * it isn't quite a version and maybe it means 385 * "extended from REV0 so check compat features." 386 * 387 * XXX: We don't have any native tool to activate 388 * the EXT2F_COMPAT_RESIZE feature and 389 * fsck_ext2fs(8) might not fix structures for it. 390 */ 391 sblock.e2fs.e2fs_features_compat = EXT2F_COMPAT_RESIZE; 392 sblock.e2fs.e2fs_features_incompat = EXT2F_INCOMPAT_FTYPE; 393 sblock.e2fs.e2fs_features_rocompat = 394 EXT2F_ROCOMPAT_SPARSESUPER | EXT2F_ROCOMPAT_LARGEFILE; 395 } 396 397 sblock.e2fs.e2fs_ruid = geteuid(); 398 sblock.e2fs.e2fs_rgid = getegid(); 399 400 sblock.e2fs.e2fs_first_ino = EXT2_FIRSTINO; 401 sblock.e2fs.e2fs_inode_size = inodesize; 402 403 /* e2fs_block_group_nr is set on writing superblock to each group */ 404 405 uuid_get(&sblock); 406 if (volname != NULL) { 407 if (strlen(volname) > sizeof(sblock.e2fs.e2fs_vname)) 408 errx(EXIT_FAILURE, "Volume name is too long"); 409 strlcpy(sblock.e2fs.e2fs_vname, volname, 410 sizeof(sblock.e2fs.e2fs_vname)); 411 } 412 413 sblock.e2fs.e2fs_fsmnt[0] = '\0'; 414 sblock.e2fs_fsmnt[0] = '\0'; 415 416 sblock.e2fs.e2fs_algo = 0; /* XXX unsupported? */ 417 sblock.e2fs.e2fs_prealloc = 0; /* XXX unsupported? */ 418 sblock.e2fs.e2fs_dir_prealloc = 0; /* XXX unsupported? */ 419 420 /* calculate blocks for reserved group descriptors for resize */ 421 sblock.e2fs.e2fs_reserved_ngdb = 0; 422 if (sblock.e2fs.e2fs_rev > E2FS_REV0 && 423 (sblock.e2fs.e2fs_features_compat & EXT2F_COMPAT_RESIZE) != 0) { 424 uint64_t target_blocks; 425 uint target_ncg, target_ngdb, reserved_ngdb; 426 427 /* reserve descriptors for size as 1024 times as current */ 428 target_blocks = 429 (sblock.e2fs.e2fs_bcount - sblock.e2fs.e2fs_first_dblock) 430 * 1024ULL; 431 /* number of blocks must be in uint32_t */ 432 if (target_blocks > UINT32_MAX) 433 target_blocks = UINT32_MAX; 434 target_ncg = howmany(target_blocks, sblock.e2fs.e2fs_bpg); 435 target_ngdb = howmany(sizeof(struct ext2_gd) * target_ncg, 436 sblock.e2fs_bsize); 437 /* 438 * Reserved group descriptor blocks are preserved as 439 * the second level double indirect reference blocks in 440 * the EXT2_RESIZEINO inode, so the maximum number of 441 * the blocks is NINDIR(fs). 442 * (see also descriptions in init_resizeino() function) 443 * 444 * We check a number including current e2fs_ngdb here 445 * because they will be moved into reserved gdb on 446 * possible future size shrink, though e2fsprogs don't 447 * seem to care about it. 448 */ 449 if (target_ngdb > NINDIR(&sblock)) 450 target_ngdb = NINDIR(&sblock); 451 452 reserved_ngdb = target_ngdb - sblock.e2fs_ngdb; 453 454 /* make sure reserved_ngdb fits in the last cg */ 455 if (reserved_ngdb >= blocks_lastcg - cgoverhead(ncg - 1)) 456 reserved_ngdb = blocks_lastcg - cgoverhead(ncg - 1); 457 if (reserved_ngdb == 0) { 458 /* if no space for reserved gdb, disable the feature */ 459 sblock.e2fs.e2fs_features_compat &= 460 ~EXT2F_COMPAT_RESIZE; 461 } 462 sblock.e2fs.e2fs_reserved_ngdb = reserved_ngdb; 463 } 464 465 /* 466 * Initialize group descriptors 467 */ 468 gd = malloc(sblock.e2fs_ngdb * bsize); 469 if (gd == NULL) 470 errx(EXIT_FAILURE, "Can't allocate descriptors buffer"); 471 memset(gd, 0, sblock.e2fs_ngdb * bsize); 472 473 fbcount = 0; 474 ficount = 0; 475 for (cylno = 0; cylno < ncg; cylno++) { 476 uint boffset; 477 478 boffset = cgbase(&sblock, cylno); 479 if (sblock.e2fs.e2fs_rev == E2FS_REV0 || 480 (sblock.e2fs.e2fs_features_rocompat & 481 EXT2F_ROCOMPAT_SPARSESUPER) == 0 || 482 cg_has_sb(cylno)) { 483 boffset += NBLOCK_SUPERBLOCK + sblock.e2fs_ngdb; 484 if (sblock.e2fs.e2fs_rev > E2FS_REV0 && 485 (sblock.e2fs.e2fs_features_compat & 486 EXT2F_COMPAT_RESIZE) != 0) 487 boffset += sblock.e2fs.e2fs_reserved_ngdb; 488 } 489 gd[cylno].ext2bgd_b_bitmap = boffset; 490 boffset += NBLOCK_BLOCK_BITMAP; 491 gd[cylno].ext2bgd_i_bitmap = boffset; 492 boffset += NBLOCK_INODE_BITMAP; 493 gd[cylno].ext2bgd_i_tables = boffset; 494 if (cylno == (ncg - 1)) 495 gd[cylno].ext2bgd_nbfree = 496 blocks_lastcg - cgoverhead(cylno); 497 else 498 gd[cylno].ext2bgd_nbfree = 499 sblock.e2fs.e2fs_bpg - cgoverhead(cylno); 500 fbcount += gd[cylno].ext2bgd_nbfree; 501 gd[cylno].ext2bgd_nifree = sblock.e2fs.e2fs_ipg; 502 if (cylno == 0) { 503 /* take reserved inodes off nifree */ 504 gd[cylno].ext2bgd_nifree -= EXT2_RESERVED_INODES; 505 } 506 ficount += gd[cylno].ext2bgd_nifree; 507 gd[cylno].ext2bgd_ndirs = 0; 508 } 509 sblock.e2fs.e2fs_fbcount = fbcount; 510 sblock.e2fs.e2fs_ficount = ficount; 511 512 /* 513 * Dump out summary information about file system. 514 */ 515 if (verbosity > 0) { 516 printf("%s: %u.%1uMB (%" PRId64 " sectors) " 517 "block size %u, fragment size %u\n", 518 fsys, 519 (uint)(((uint64_t)bcount * bsize) / (1024 * 1024)), 520 (uint)((uint64_t)bcount * bsize - 521 rounddown((uint64_t)bcount * bsize, 1024 * 1024)) 522 / 1024 / 100, 523 fssize, bsize, fsize); 524 printf("\tusing %u block groups of %u.0MB, %u blks, " 525 "%u inodes.\n", 526 ncg, bsize * sblock.e2fs.e2fs_bpg / (1024 * 1024), 527 sblock.e2fs.e2fs_bpg, sblock.e2fs.e2fs_ipg); 528 } 529 530 /* 531 * allocate space for superblock and group descriptors 532 */ 533 iobufsize = (NBLOCK_SUPERBLOCK + sblock.e2fs_ngdb) * sblock.e2fs_bsize; 534 iobuf = mmap(0, iobufsize, PROT_READ|PROT_WRITE, 535 MAP_ANON|MAP_PRIVATE, -1, 0); 536 if (iobuf == NULL) 537 errx(EXIT_FAILURE, "Cannot allocate I/O buffer\n"); 538 memset(iobuf, 0, iobufsize); 539 540 /* 541 * We now start writing to the filesystem 542 */ 543 544 if (!Nflag) { 545 static const uint pbsize[] = { 1024, 2048, 4096, 0 }; 546 uint pblock, epblock; 547 /* 548 * Validate the given file system size. 549 * Verify that its last block can actually be accessed. 550 * Convert to file system fragment sized units. 551 */ 552 if (fssize <= 0) 553 errx(EXIT_FAILURE, "Preposterous size %" PRId64 "\n", 554 fssize); 555 wtfs(fssize - 1, sectorsize, iobuf); 556 557 /* 558 * Ensure there is nothing that looks like a filesystem 559 * superblock anywhere other than where ours will be. 560 * If fsck_ext2fs finds the wrong one all hell breaks loose! 561 * 562 * XXX: needs to check how fsck_ext2fs programs even 563 * on other OSes determine alternate superblocks 564 */ 565 for (i = 0; pbsize[i] != 0; i++) { 566 epblock = (uint64_t)bcount * bsize / pbsize[i]; 567 for (pblock = ((pbsize[i] == SBSIZE) ? 1 : 0); 568 pblock < epblock; 569 pblock += pbsize[i] * NBBY /* bpg */) 570 zap_old_sblock((daddr32_t)pblock * 571 pbsize[i] / sectorsize); 572 } 573 } 574 575 if (verbosity >= 3) 576 printf("super-block backups (for fsck_ext2fs -b #) at:\n"); 577 /* If we are printing more than one line of numbers, line up columns */ 578 fld_width = verbosity < 4 ? 1 : snprintf(NULL, 0, "%" PRIu64, 579 (uint64_t)cgbase(&sblock, ncg - 1)); 580 /* Get terminal width */ 581 if (ioctl(fileno(stdout), TIOCGWINSZ, &winsize) == 0) 582 max_cols = winsize.ws_col; 583 else 584 max_cols = 80; 585 if (Nflag && verbosity == 3) 586 /* Leave space to add " ..." after one row of numbers */ 587 max_cols -= 4; 588 #define BASE 0x10000 /* For some fixed-point maths */ 589 col = 0; 590 delta = verbosity > 2 ? 0 : max_cols * BASE / ncg; 591 for (cylno = 0; cylno < ncg; cylno++) { 592 fflush(stdout); 593 initcg(cylno); 594 if (verbosity < 2) 595 continue; 596 /* the first one is a master, not backup */ 597 if (cylno == 0) 598 continue; 599 /* skip if this cylinder doesn't have a backup */ 600 if (sblock.e2fs.e2fs_rev > E2FS_REV0 && 601 (sblock.e2fs.e2fs_features_rocompat & 602 EXT2F_ROCOMPAT_SPARSESUPER) != 0 && 603 cg_has_sb(cylno) == 0) 604 continue; 605 606 if (delta > 0) { 607 if (Nflag) 608 /* No point doing dots for -N */ 609 break; 610 /* Print dots scaled to end near RH margin */ 611 for (col += delta; col > BASE; col -= BASE) 612 printf("."); 613 continue; 614 } 615 /* Print superblock numbers */ 616 len = printf(" %*" PRIu64 "," + !col, fld_width, 617 (uint64_t)cgbase(&sblock, cylno)); 618 col += len; 619 if (col + len < max_cols) 620 /* Next number fits */ 621 continue; 622 /* Next number won't fit, need a newline */ 623 if (verbosity <= 3) { 624 /* Print dots for subsequent cylinder groups */ 625 delta = sblock.e2fs_ncg - cylno - 1; 626 if (delta != 0) { 627 if (Nflag) { 628 printf(" ..."); 629 break; 630 } 631 delta = max_cols * BASE / delta; 632 } 633 } 634 col = 0; 635 printf("\n"); 636 } 637 #undef BASE 638 if (col > 0) 639 printf("\n"); 640 if (Nflag) 641 return; 642 643 /* 644 * Now construct the initial file system, 645 */ 646 if (fsinit(&tv) == 0) 647 errx(EXIT_FAILURE, "Error making filesystem"); 648 /* 649 * Write out the superblock and group descriptors 650 */ 651 sblock.e2fs.e2fs_block_group_nr = 0; 652 sboff = 0; 653 if (cgbase(&sblock, 0) == 0) { 654 /* 655 * If the first block contains the boot block sectors, 656 * (i.e. in case of sblock.e2fs.e2fs_bsize > BBSIZE) 657 * we have to preserve data in it. 658 */ 659 sboff = SBOFF; 660 } 661 e2fs_sbsave(&sblock.e2fs, (struct ext2fs *)(iobuf + sboff)); 662 e2fs_cgsave(gd, (struct ext2_gd *)(iobuf + sblock.e2fs_bsize), 663 sizeof(struct ext2_gd) * sblock.e2fs_ncg); 664 wtfs(fsbtodb(&sblock, cgbase(&sblock, 0)) + sboff / sectorsize, 665 iobufsize - sboff, iobuf + sboff); 666 667 munmap(iobuf, iobufsize); 668 } 669 670 /* 671 * Initialize a cylinder (block) group. 672 */ 673 void 674 initcg(uint cylno) 675 { 676 uint nblcg, i, j, sboff; 677 struct ext2fs_dinode *dp; 678 679 /* 680 * Make a copy of the superblock and group descriptors. 681 */ 682 if (sblock.e2fs.e2fs_rev == E2FS_REV0 || 683 (sblock.e2fs.e2fs_features_rocompat & 684 EXT2F_ROCOMPAT_SPARSESUPER) == 0 || 685 cg_has_sb(cylno)) { 686 sblock.e2fs.e2fs_block_group_nr = cylno; 687 sboff = 0; 688 if (cgbase(&sblock, cylno) == 0) { 689 /* preserve data in bootblock in cg0 */ 690 sboff = SBOFF; 691 } 692 e2fs_sbsave(&sblock.e2fs, (struct ext2fs *)(iobuf + sboff)); 693 e2fs_cgsave(gd, (struct ext2_gd *)(iobuf + 694 sblock.e2fs_bsize * NBLOCK_SUPERBLOCK), 695 sizeof(struct ext2_gd) * sblock.e2fs_ncg); 696 /* write superblock and group descriptor backups */ 697 wtfs(fsbtodb(&sblock, cgbase(&sblock, cylno)) + 698 sboff / sectorsize, iobufsize - sboff, iobuf + sboff); 699 } 700 701 /* 702 * Initialize block bitmap. 703 */ 704 memset(buf, 0, sblock.e2fs_bsize); 705 if (cylno == (sblock.e2fs_ncg - 1)) { 706 /* The last group could have less blocks than e2fs_bpg. */ 707 nblcg = sblock.e2fs.e2fs_bcount - 708 cgbase(&sblock, sblock.e2fs_ncg - 1); 709 for (i = nblcg; i < roundup(nblcg, NBBY); i++) 710 setbit(buf, i); 711 memset(&buf[i / NBBY], ~0U, sblock.e2fs.e2fs_bpg - i); 712 } 713 /* set overhead (superblock, group descriptor etc.) blocks used */ 714 for (i = 0; i < cgoverhead(cylno) / NBBY; i++) 715 buf[i] = ~0; 716 i = i * NBBY; 717 for (; i < cgoverhead(cylno); i++) 718 setbit(buf, i); 719 wtfs(fsbtodb(&sblock, gd[cylno].ext2bgd_b_bitmap), sblock.e2fs_bsize, 720 buf); 721 722 /* 723 * Initialize inode bitmap. 724 * 725 * Assume e2fs_ipg is a multiple of NBBY since 726 * it's a multiple of e2fs_ipb (as we did above). 727 * Note even (possibly smaller) the last group has the same e2fs_ipg. 728 */ 729 i = sblock.e2fs.e2fs_ipg / NBBY; 730 memset(buf, 0, i); 731 memset(buf + i, ~0U, sblock.e2fs_bsize - i); 732 if (cylno == 0) { 733 /* mark reserved inodes */ 734 for (i = 1; i < EXT2_FIRSTINO; i++) 735 setbit(buf, EXT2_INO_INDEX(i)); 736 } 737 wtfs(fsbtodb(&sblock, gd[cylno].ext2bgd_i_bitmap), sblock.e2fs_bsize, 738 buf); 739 740 /* 741 * Initialize inode tables. 742 * 743 * Just initialize generation numbers for NFS security. 744 * XXX: sys/ufs/ext2fs/ext2fs_alloc.c:ext2fs_valloc() seems 745 * to override these generated numbers. 746 */ 747 memset(buf, 0, sblock.e2fs_bsize); 748 for (i = 0; i < sblock.e2fs_itpg; i++) { 749 for (j = 0; j < sblock.e2fs_ipb; j++) { 750 dp = (struct ext2fs_dinode *)(buf + inodesize * j); 751 /* h2fs32() just for consistency */ 752 dp->e2di_gen = h2fs32(arc4random()); 753 } 754 wtfs(fsbtodb(&sblock, gd[cylno].ext2bgd_i_tables + i), 755 sblock.e2fs_bsize, buf); 756 } 757 } 758 759 /* 760 * Zap possible lingering old superblock data 761 */ 762 static void 763 zap_old_sblock(daddr32_t sec) 764 { 765 static daddr32_t cg0_data; 766 uint32_t oldfs[SBSIZE / sizeof(uint32_t)]; 767 static const struct fsm { 768 uint32_t offset; 769 uint32_t magic; 770 uint32_t mask; 771 } fs_magics[] = { 772 {offsetof(struct ext2fs, e2fs_magic) / 4, E2FS_MAGIC, 0xffff}, 773 {offsetof(struct ext2fs, e2fs_magic) / 4, 774 E2FS_MAGIC << 16, 0xffff0000}, 775 {14, 0xef530000, 0xffff0000}, /* EXT2FS (big) */ 776 {0x55c / 4, 0x00011954, ~0U}, /* FS_UFS1_MAGIC */ 777 {0x55c / 4, 0x19540119, ~0U}, /* FS_UFS2_MAGIC */ 778 {0, 0x70162, ~0U}, /* LFS_MAGIC */ 779 {.offset = ~0U}, 780 }; 781 const struct fsm *fsm; 782 783 if (Nflag) 784 return; 785 786 /* don't override data before superblock */ 787 if (sec < SBOFF / sectorsize) 788 return; 789 790 if (cg0_data == 0) { 791 cg0_data = 792 ((daddr32_t)sblock.e2fs.e2fs_first_dblock + cgoverhead(0)) * 793 sblock.e2fs_bsize / sectorsize; 794 } 795 796 /* Ignore anything that is beyond our filesystem */ 797 if (sec >= fssize) 798 return; 799 /* Zero anything inside our filesystem... */ 800 if (sec >= sblock.e2fs.e2fs_first_dblock * bsize / sectorsize) { 801 /* ...unless we will write that area anyway */ 802 if (sec >= cg0_data) 803 /* assume iobuf is zero'ed here */ 804 wtfs(sec, roundup(SBSIZE, sectorsize), iobuf); 805 return; 806 } 807 808 /* 809 * The sector might contain boot code, so we must validate it 810 * 811 * XXX: ext2fs won't preserve data after SBOFF, 812 * but first_dblock could have a different value. 813 */ 814 rdfs(sec, sizeof(oldfs), &oldfs); 815 for (fsm = fs_magics;; fsm++) { 816 uint32_t v; 817 if (fsm->mask == 0) 818 return; 819 v = oldfs[fsm->offset]; 820 if ((v & fsm->mask) == fsm->magic || 821 (swap32(v) & fsm->mask) == fsm->magic) 822 break; 823 } 824 825 /* Just zap the magic number */ 826 oldfs[fsm->offset] = 0; 827 wtfs(sec, sizeof(oldfs), &oldfs); 828 } 829 830 /* 831 * uint cgoverhead(uint c) 832 * 833 * Return a number of reserved blocks on the specified group. 834 * XXX: should be shared with src/sbin/fsck_ext2fs/setup.c 835 */ 836 uint 837 cgoverhead(uint c) 838 { 839 uint overh; 840 841 overh = NBLOCK_BLOCK_BITMAP + NBLOCK_INODE_BITMAP + sblock.e2fs_itpg; 842 843 if (sblock.e2fs.e2fs_rev == E2FS_REV0 || 844 (sblock.e2fs.e2fs_features_rocompat & 845 EXT2F_ROCOMPAT_SPARSESUPER) == 0 || 846 cg_has_sb(c) != 0) { 847 overh += NBLOCK_SUPERBLOCK + sblock.e2fs_ngdb; 848 849 if (sblock.e2fs.e2fs_rev > E2FS_REV0 && 850 (sblock.e2fs.e2fs_features_compat & 851 EXT2F_COMPAT_RESIZE) != 0) 852 overh += sblock.e2fs.e2fs_reserved_ngdb; 853 } 854 855 return overh; 856 } 857 858 /* 859 * Initialize the file system 860 */ 861 862 #define LOSTDIR /* e2fsck complains if there is no lost+found */ 863 864 #define PREDEFDIR 2 865 866 #ifdef LOSTDIR 867 #define PREDEFROOTDIR (PREDEFDIR + 1) 868 #else 869 #define PREDEFROOTDIR PREDEFDIR 870 #endif 871 872 struct ext2fs_direct root_dir[] = { 873 { EXT2_ROOTINO, 0, 1, 0, "." }, 874 { EXT2_ROOTINO, 0, 2, 0, ".." }, 875 #ifdef LOSTDIR 876 { EXT2_LOSTFOUNDINO, 0, 10, 0, "lost+found" }, 877 #endif 878 }; 879 880 #ifdef LOSTDIR 881 struct ext2fs_direct lost_found_dir[] = { 882 { EXT2_LOSTFOUNDINO, 0, 1, 0, "." }, 883 { EXT2_ROOTINO, 0, 2, 0, ".." }, 884 }; 885 struct ext2fs_direct pad_dir = { 0, sizeof(struct ext2fs_direct), 0, 0, "" }; 886 #endif 887 888 int 889 fsinit(const struct timeval *tv) 890 { 891 struct ext2fs_dinode node; 892 #ifdef LOSTDIR 893 uint i, nblks_lostfound, blk; 894 #endif 895 896 /* 897 * Initialize the inode for the resizefs feature 898 */ 899 if (sblock.e2fs.e2fs_rev > E2FS_REV0 && 900 (sblock.e2fs.e2fs_features_compat & EXT2F_COMPAT_RESIZE) != 0) 901 init_resizeino(tv); 902 903 /* 904 * Initialize the node 905 */ 906 907 #ifdef LOSTDIR 908 /* 909 * Create the lost+found directory 910 */ 911 if (sblock.e2fs.e2fs_rev > E2FS_REV0 && 912 sblock.e2fs.e2fs_features_incompat & EXT2F_INCOMPAT_FTYPE) { 913 lost_found_dir[0].e2d_type = EXT2_FT_DIR; 914 lost_found_dir[1].e2d_type = EXT2_FT_DIR; 915 } 916 (void)makedir(lost_found_dir, nitems(lost_found_dir)); 917 918 /* prepare a bit large directory for preserved files */ 919 nblks_lostfound = EXT2_LOSTFOUNDSIZE / sblock.e2fs_bsize; 920 /* ...but only with direct blocks */ 921 if (nblks_lostfound > NDADDR) 922 nblks_lostfound = NDADDR; 923 924 memset(&node, 0, sizeof(node)); 925 node.e2di_mode = EXT2_IFDIR | EXT2_LOSTFOUNDUMASK; 926 node.e2di_uid_low = geteuid(); 927 node.e2di_size = sblock.e2fs_bsize * nblks_lostfound; 928 node.e2di_atime = (u_int32_t)tv->tv_sec; 929 node.e2di_ctime = (u_int32_t)tv->tv_sec; 930 node.e2di_mtime = (u_int32_t)tv->tv_sec; 931 node.e2di_gid_low = getegid(); 932 node.e2di_nlink = PREDEFDIR; 933 /* e2di_nblock is a number of disk blocks, not ext2fs blocks */ 934 node.e2di_nblock = fsbtodb(&sblock, nblks_lostfound); 935 node.e2di_blocks[0] = alloc(sblock.e2fs_bsize, node.e2di_mode); 936 if (node.e2di_blocks[0] == 0) { 937 printf("%s: can't allocate block for lost+found\n", __func__); 938 return 0; 939 } 940 for (i = 1; i < nblks_lostfound; i++) { 941 blk = alloc(sblock.e2fs_bsize, 0); 942 if (blk == 0) { 943 printf("%s: can't allocate blocks for lost+found\n", 944 __func__); 945 return 0; 946 } 947 node.e2di_blocks[i] = blk; 948 } 949 wtfs(fsbtodb(&sblock, node.e2di_blocks[0]), sblock.e2fs_bsize, buf); 950 pad_dir.e2d_reclen = sblock.e2fs_bsize; 951 for (i = 1; i < nblks_lostfound; i++) { 952 memset(buf, 0, sblock.e2fs_bsize); 953 copy_dir(&pad_dir, (struct ext2fs_direct *)buf); 954 wtfs(fsbtodb(&sblock, node.e2di_blocks[i]), sblock.e2fs_bsize, 955 buf); 956 } 957 iput(&node, EXT2_LOSTFOUNDINO); 958 #endif 959 /* 960 * create the root directory 961 */ 962 memset(&node, 0, sizeof(node)); 963 if (sblock.e2fs.e2fs_rev > E2FS_REV0 && 964 sblock.e2fs.e2fs_features_incompat & EXT2F_INCOMPAT_FTYPE) { 965 root_dir[0].e2d_type = EXT2_FT_DIR; 966 root_dir[1].e2d_type = EXT2_FT_DIR; 967 #ifdef LOSTDIR 968 root_dir[2].e2d_type = EXT2_FT_DIR; 969 #endif 970 } 971 node.e2di_mode = EXT2_IFDIR | EXT2_UMASK; 972 node.e2di_uid_low = geteuid(); 973 node.e2di_size = makedir(root_dir, nitems(root_dir)); 974 node.e2di_atime = (u_int32_t)tv->tv_sec; 975 node.e2di_ctime = (u_int32_t)tv->tv_sec; 976 node.e2di_mtime = (u_int32_t)tv->tv_sec; 977 node.e2di_gid_low = getegid(); 978 node.e2di_nlink = PREDEFROOTDIR; 979 /* e2di_nblock is a number of disk block, not ext2fs block */ 980 node.e2di_nblock = fsbtodb(&sblock, 1); 981 node.e2di_blocks[0] = alloc(node.e2di_size, node.e2di_mode); 982 if (node.e2di_blocks[0] == 0) { 983 printf("%s: can't allocate block for root dir\n", __func__); 984 return 0; 985 } 986 wtfs(fsbtodb(&sblock, node.e2di_blocks[0]), sblock.e2fs_bsize, buf); 987 iput(&node, EXT2_ROOTINO); 988 return 1; 989 } 990 991 /* 992 * Construct a set of directory entries in "buf". 993 * return size of directory. 994 */ 995 int 996 makedir(struct ext2fs_direct *protodir, int entries) 997 { 998 uint8_t *cp; 999 uint i, spcleft; 1000 uint dirblksiz; 1001 1002 dirblksiz = sblock.e2fs_bsize; 1003 memset(buf, 0, dirblksiz); 1004 spcleft = dirblksiz; 1005 for (cp = buf, i = 0; i < entries - 1; i++) { 1006 protodir[i].e2d_reclen = EXT2FS_DIRSIZ(protodir[i].e2d_namlen); 1007 copy_dir(&protodir[i], (struct ext2fs_direct *)cp); 1008 cp += protodir[i].e2d_reclen; 1009 spcleft -= protodir[i].e2d_reclen; 1010 } 1011 protodir[i].e2d_reclen = spcleft; 1012 copy_dir(&protodir[i], (struct ext2fs_direct *)cp); 1013 return dirblksiz; 1014 } 1015 1016 /* 1017 * Copy a direntry to a buffer, in fs byte order 1018 */ 1019 static void 1020 copy_dir(struct ext2fs_direct *dir, struct ext2fs_direct *dbuf) 1021 { 1022 1023 memcpy(dbuf, dir, EXT2FS_DIRSIZ(dir->e2d_namlen)); 1024 dbuf->e2d_ino = h2fs32(dir->e2d_ino); 1025 dbuf->e2d_reclen = h2fs16(dir->e2d_reclen); 1026 } 1027 1028 /* 1029 * void init_resizeino(const struct timeval *tv); 1030 * 1031 * Initialize the EXT2_RESEIZE_INO inode to preserve 1032 * reserved group descriptor blocks for future growth of this ext2fs. 1033 */ 1034 void 1035 init_resizeino(const struct timeval *tv) 1036 { 1037 struct ext2fs_dinode node; 1038 uint64_t isize; 1039 uint32_t *dindir_block, *reserved_gdb; 1040 uint nblock, i, cylno, n; 1041 1042 memset(&node, 0, sizeof(node)); 1043 1044 /* 1045 * Note this function only prepares required structures for 1046 * future resize. It's a quite different work to implement 1047 * a utility like resize_ext2fs(8) which handles actual 1048 * resize ops even on offline. 1049 * 1050 * Anyway, I'm not sure if there is any documentation about 1051 * this resize ext2fs feature and related data structures, 1052 * and I've written this function based on things what I see 1053 * on some existing implementation and real file system data 1054 * created by existing tools. To be honest, they are not 1055 * so easy to read, so I will try to implement it here without 1056 * any dumb optimization for people who would eventually 1057 * work on "yet another wheel" like resize_ext2fs(8). 1058 */ 1059 1060 /* 1061 * I'm not sure what type is appropriate for this inode. 1062 * The release notes of e2fsprogs says they changed e2fsck to allow 1063 * IFREG for RESIZEINO since a certain resize tool used it. Hmm. 1064 */ 1065 node.e2di_mode = EXT2_IFREG | EXT2_RESIZEINOUMASK; 1066 node.e2di_uid_low = geteuid(); 1067 node.e2di_atime = (u_int32_t)tv->tv_sec; 1068 node.e2di_ctime = (u_int32_t)tv->tv_sec; 1069 node.e2di_mtime = (u_int32_t)tv->tv_sec; 1070 node.e2di_gid_low = getegid(); 1071 node.e2di_nlink = 1; 1072 1073 /* 1074 * To preserve the reserved group descriptor blocks, 1075 * EXT2_RESIZEINO uses only double indirect reference 1076 * blocks in its inode entries. 1077 * 1078 * All entries for direct, single indirect and triple 1079 * indirect references are left zero'ed. Maybe it's safe 1080 * because no write operation will happen with this inode. 1081 * 1082 * We have to allocate a block for the first level double 1083 * indirect reference block. Indexes of inode entries in 1084 * this first level dindirect block are corresponding to 1085 * indexes of group descriptors including both used (e2fs_ngdb) 1086 * and reserved (e2fs_reserved_ngdb) group descriptor blocks. 1087 * 1088 * Inode entries of indexes for used (e2fs_ngdb) descriptors are 1089 * left zero'ed. Entries for reserved (e2fs_reserved_ngdb) ones 1090 * have block numbers of actual reserved group descriptors 1091 * allocated at block group zero. This means e2fs_reserved_ngdb 1092 * blocks are reserved as the second level dindirect reference 1093 * blocks, and they actually contain block numbers of indirect 1094 * references. It may be safe since they don't have to keep any 1095 * data yet. 1096 * 1097 * Each these second dindirect blocks (i.e. reserved group 1098 * descriptor blocks in the first block group) should have 1099 * block numbers of its backups in all other block groups. 1100 * I.e. reserved_ngdb[0] block in block group 0 contains block 1101 * numbers of resreved_ngdb[0] from group 1 through (e2fs_ncg - 1). 1102 * The number of backups can be determined by the 1103 * EXT2_ROCOMPAT_SPARSESUPER feature and cg_has_sb() macro 1104 * as done in the above initcg() function. 1105 */ 1106 1107 /* set e2di_size which occupies whole blocks through DINDIR blocks */ 1108 isize = (uint64_t)sblock.e2fs_bsize * NDADDR + 1109 (uint64_t)sblock.e2fs_bsize * NINDIR(&sblock) + 1110 (uint64_t)sblock.e2fs_bsize * NINDIR(&sblock) * NINDIR(&sblock); 1111 if (isize > UINT32_MAX && 1112 (sblock.e2fs.e2fs_features_rocompat & 1113 EXT2F_ROCOMPAT_LARGEFILE) == 0) { 1114 /* XXX should enable it here and update all backups? */ 1115 errx(EXIT_FAILURE, "%s: large_file rocompat feature is " 1116 "required to enable resize feature for this filesystem\n", 1117 __func__); 1118 } 1119 /* upper 32bit is stored into e2di_dacl on REV1 feature */ 1120 node.e2di_size = isize & UINT32_MAX; 1121 node.e2di_dacl = isize >> 32; 1122 1123 #define SINGLE 0 /* index of single indirect block */ 1124 #define DOUBLE 1 /* index of double indirect block */ 1125 #define TRIPLE 2 /* index of triple indirect block */ 1126 1127 /* zero out entries for direct references */ 1128 for (i = 0; i < NDADDR; i++) 1129 node.e2di_blocks[i] = 0; 1130 /* also zero out entries for single and triple indirect references */ 1131 node.e2di_blocks[NDADDR + SINGLE] = 0; 1132 node.e2di_blocks[NDADDR + TRIPLE] = 0; 1133 1134 /* allocate a block for the first level double indirect reference */ 1135 node.e2di_blocks[NDADDR + DOUBLE] = 1136 alloc(sblock.e2fs_bsize, node.e2di_mode); 1137 if (node.e2di_blocks[NDADDR + DOUBLE] == 0) 1138 errx(EXIT_FAILURE, "%s: Can't allocate a dindirect block", 1139 __func__); 1140 1141 /* account this first block */ 1142 nblock = fsbtodb(&sblock, 1); 1143 1144 /* allocate buffer to set data in the dindirect block */ 1145 dindir_block = malloc(sblock.e2fs_bsize); 1146 if (dindir_block == NULL) 1147 errx(EXIT_FAILURE, 1148 "%s: Can't allocate buffer for a dindirect block", 1149 __func__); 1150 1151 /* allocate buffer to set data in the group descriptor blocks */ 1152 reserved_gdb = malloc(sblock.e2fs_bsize); 1153 if (reserved_gdb == NULL) 1154 errx(EXIT_FAILURE, 1155 "%s: Can't allocate buffer for group descriptor blocks", 1156 __func__); 1157 1158 /* 1159 * Setup block entries in the first level dindirect blocks 1160 */ 1161 for (i = 0; i < sblock.e2fs_ngdb; i++) { 1162 /* no need to handle used group descriptor blocks */ 1163 dindir_block[i] = 0; 1164 } 1165 for (; i < sblock.e2fs_ngdb + sblock.e2fs.e2fs_reserved_ngdb; i++) { 1166 /* 1167 * point reserved group descriptor block in the first 1168 * (i.e. master) block group 1169 * 1170 * XXX: e2fsprogs seem to use "(i % NINDIR(&sblock))" here 1171 * to store maximum NINDIR(&sblock) reserved gdbs. 1172 * I'm not sure what will be done on future filesystem 1173 * shrink in that case on their way. 1174 */ 1175 if (i >= NINDIR(&sblock)) 1176 errx(EXIT_FAILURE, "%s: too many reserved " 1177 "group descriptors (%u) for resize inode", 1178 __func__, sblock.e2fs.e2fs_reserved_ngdb); 1179 dindir_block[i] = 1180 h2fs32(cgbase(&sblock, 0) + NBLOCK_SUPERBLOCK + i); 1181 1182 /* 1183 * Setup block entries in the second dindirect blocks 1184 * (which are primary reserved group descriptor blocks) 1185 * to point their backups. 1186 */ 1187 for (n = 0, cylno = 1; cylno < sblock.e2fs_ncg; cylno++) { 1188 /* skip block groups without backup */ 1189 if ((sblock.e2fs.e2fs_features_rocompat & 1190 EXT2F_ROCOMPAT_SPARSESUPER) != 0 && 1191 cg_has_sb(cylno) == 0) 1192 continue; 1193 1194 if (n >= NINDIR(&sblock)) 1195 errx(EXIT_FAILURE, "%s: too many block groups " 1196 "for the resize feature", __func__); 1197 /* 1198 * These blocks are already reserved in 1199 * initcg() so no need to use alloc() here. 1200 */ 1201 reserved_gdb[n++] = h2fs32(cgbase(&sblock, cylno) + 1202 NBLOCK_SUPERBLOCK + i); 1203 nblock += fsbtodb(&sblock, 1); 1204 } 1205 for (; n < NINDIR(&sblock); n++) 1206 reserved_gdb[n] = 0; 1207 1208 /* write group descriptor block as the second dindirect refs */ 1209 wtfs(fsbtodb(&sblock, fs2h32(dindir_block[i])), 1210 sblock.e2fs_bsize, reserved_gdb); 1211 nblock += fsbtodb(&sblock, 1); 1212 } 1213 for (; i < NINDIR(&sblock); i++) { 1214 /* leave trailing entries unallocated */ 1215 dindir_block[i] = 0; 1216 } 1217 free(reserved_gdb); 1218 1219 /* finally write the first level dindirect block */ 1220 wtfs(fsbtodb(&sblock, node.e2di_blocks[NDADDR + DOUBLE]), 1221 sblock.e2fs_bsize, dindir_block); 1222 free(dindir_block); 1223 1224 node.e2di_nblock = nblock; 1225 iput(&node, EXT2_RESIZEINO); 1226 } 1227 1228 /* 1229 * uint32_t alloc(uint32_t size, uint16_t mode) 1230 * 1231 * Allocate a block (from cylinder group 0) 1232 * Reference: src/sys/ufs/ext2fs/ext2fs_alloc.c:ext2fs_alloccg() 1233 */ 1234 uint32_t 1235 alloc(uint32_t size, uint16_t mode) 1236 { 1237 uint32_t loc, bno; 1238 uint8_t *bbp; 1239 uint len, map, i; 1240 1241 if (gd[0].ext2bgd_nbfree == 0) 1242 return 0; 1243 1244 if (size > sblock.e2fs_bsize) 1245 return 0; 1246 1247 bbp = malloc(sblock.e2fs_bsize); 1248 if (bbp == NULL) 1249 return 0; 1250 rdfs(fsbtodb(&sblock, gd[0].ext2bgd_b_bitmap), sblock.e2fs_bsize, bbp); 1251 1252 /* XXX: kernel uses e2fs_fpg here */ 1253 len = sblock.e2fs.e2fs_bpg / NBBY; 1254 1255 #if 0 /* no need block allocation for root or lost+found dir */ 1256 for (loc = 0; loc < len; loc++) { 1257 if (bbp[loc] == 0) { 1258 bno = loc * NBBY; 1259 goto gotit; 1260 } 1261 } 1262 #endif 1263 1264 loc = skpc(~0U, len, bbp); 1265 if (loc == 0) { 1266 free(bbp); 1267 return 0; 1268 } 1269 loc = len - loc; 1270 map = bbp[loc]; 1271 bno = loc * NBBY; 1272 for (i = 0; i < NBBY; i++, bno++) { 1273 if ((map & (1 << i)) == 0) 1274 goto gotit; 1275 } 1276 free(bbp); 1277 return 0; 1278 1279 gotit: 1280 if (isset(bbp, bno)) 1281 errx(EXIT_FAILURE, "%s: inconsistent bitmap\n", __func__); 1282 1283 setbit(bbp, bno); 1284 wtfs(fsbtodb(&sblock, gd[0].ext2bgd_b_bitmap), sblock.e2fs_bsize, bbp); 1285 free(bbp); 1286 /* XXX: modified group descriptors won't be written into backups */ 1287 gd[0].ext2bgd_nbfree--; 1288 if ((mode & EXT2_IFDIR) != 0) 1289 gd[0].ext2bgd_ndirs++; 1290 sblock.e2fs.e2fs_fbcount--; 1291 1292 return sblock.e2fs.e2fs_first_dblock + bno; 1293 } 1294 1295 /* 1296 * void iput(struct ext2fs_dinode *ip, ino_t ino) 1297 * 1298 * Put an inode entry into the corresponding table. 1299 */ 1300 static void 1301 iput(struct ext2fs_dinode *ip, ino_t ino) 1302 { 1303 daddr32_t d; 1304 uint c, i; 1305 struct ext2fs_dinode *dp; 1306 uint8_t *bp; 1307 1308 bp = malloc(sblock.e2fs_bsize); 1309 if (bp == NULL) 1310 errx(EXIT_FAILURE, "%s: can't allocate buffer for inode\n", 1311 __func__); 1312 1313 /* 1314 * Reserved inodes are allocated and accounted in initcg() 1315 * so skip checks of the bitmap and allocation for them. 1316 */ 1317 if (ino >= EXT2_FIRSTINO) { 1318 c = ino_to_cg(&sblock, ino); 1319 1320 /* sanity check */ 1321 if (gd[c].ext2bgd_nifree == 0) 1322 errx(EXIT_FAILURE, 1323 "%s: no free inode %" PRIu64 " in block group %u\n", 1324 __func__, (uint64_t)ino, c); 1325 1326 /* update inode bitmap */ 1327 rdfs(fsbtodb(&sblock, gd[0].ext2bgd_i_bitmap), 1328 sblock.e2fs_bsize, bp); 1329 1330 /* more sanity */ 1331 if (isset(bp, EXT2_INO_INDEX(ino))) 1332 errx(EXIT_FAILURE, "%s: inode %" PRIu64 1333 " already in use\n", __func__, (uint64_t)ino); 1334 setbit(bp, EXT2_INO_INDEX(ino)); 1335 wtfs(fsbtodb(&sblock, gd[0].ext2bgd_i_bitmap), 1336 sblock.e2fs_bsize, bp); 1337 gd[c].ext2bgd_nifree--; 1338 sblock.e2fs.e2fs_ficount--; 1339 } 1340 1341 if (ino >= sblock.e2fs.e2fs_ipg * sblock.e2fs_ncg) 1342 errx(EXIT_FAILURE, "%s: inode value out of range (%" PRIu64 1343 ").\n", __func__, (uint64_t)ino); 1344 1345 /* update an inode entry in the table */ 1346 d = fsbtodb(&sblock, ino_to_fsba(&sblock, ino)); 1347 rdfs(d, sblock.e2fs_bsize, bp); 1348 1349 dp = (struct ext2fs_dinode *)(bp + 1350 inodesize * ino_to_fsbo(&sblock, ino)); 1351 e2fs_isave(ip, dp); 1352 /* e2fs_i_bswap() doesn't swap e2di_blocks addrs */ 1353 if ((ip->e2di_mode & EXT2_IFMT) != EXT2_IFLNK) { 1354 for (i = 0; i < NDADDR + NIADDR; i++) 1355 dp->e2di_blocks[i] = h2fs32(ip->e2di_blocks[i]); 1356 } 1357 /* h2fs32() just for consistency */ 1358 dp->e2di_gen = h2fs32(arc4random()); 1359 1360 wtfs(d, sblock.e2fs_bsize, bp); 1361 free(bp); 1362 } 1363 1364 /* 1365 * Read a block from the file system 1366 */ 1367 void 1368 rdfs(daddr32_t bno, int size, void *bf) 1369 { 1370 int n; 1371 off_t offset; 1372 1373 offset = bno; 1374 n = pread(fsi, bf, size, offset * sectorsize); 1375 if (n != size) 1376 err(EXIT_FAILURE, "%s: read error for sector %" PRId64, 1377 __func__, (int64_t)bno); 1378 } 1379 1380 /* 1381 * Write a block to the file system 1382 */ 1383 void 1384 wtfs(daddr32_t bno, int size, void *bf) 1385 { 1386 int n; 1387 off_t offset; 1388 1389 if (Nflag) 1390 return; 1391 offset = bno; 1392 n = pwrite(fso, bf, size, offset * sectorsize); 1393 if (n != size) 1394 err(EXIT_FAILURE, "%s: write error for sector %" PRId64, 1395 __func__, (int64_t)bno); 1396 } 1397 1398 int 1399 ilog2(uint val) 1400 { 1401 1402 if (val == 0 || !powerof2(val)) 1403 errx(EXIT_FAILURE, "%s: %u is not a power of 2\n", 1404 __func__, val); 1405 1406 return ffs(val) - 1; 1407 } 1408 1409 /* 1410 * int skpc(int mask, size_t size, uint8_t *cp) 1411 * 1412 * Locate an unsigned character of value mask inside cp[]. 1413 * (from src/sys/lib/libkern/skpc.c) 1414 */ 1415 int 1416 skpc(int mask, size_t size, uint8_t *cp) 1417 { 1418 uint8_t *end; 1419 1420 end = &cp[size]; 1421 while (cp < end && *cp == (uint8_t)mask) 1422 cp++; 1423 1424 return end - cp; 1425 } 1426 1427 static void 1428 uuid_get(struct m_ext2fs *sb) 1429 { 1430 unsigned char buf[sizeof(sb->e2fs.e2fs_uuid)]; 1431 1432 arc4random_buf(buf, sizeof(buf)); 1433 /* UUID version 4: random */ 1434 buf[6] &= 0x0f; 1435 buf[6] |= 0x40; 1436 /* RFC4122 variant */ 1437 buf[8] &= 0x3f; 1438 buf[8] |= 0x80; 1439 memcpy(sb->e2fs.e2fs_uuid, buf, sizeof(sb->e2fs.e2fs_uuid)); 1440 } 1441