1/*- 2 * SPDX-License-Identifier: BSD-3-Clause 3 * 4 * Copyright (c) 1982, 1986, 1989, 1993 5 * The Regents of the University of California. 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 * 3. Neither the name of the University nor the names of its contributors 16 * may be used to endorse or promote products derived from this software 17 * without specific prior written permission. 18 * 19 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 20 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 21 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 22 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 23 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 24 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 25 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 26 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 27 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 28 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 29 * SUCH DAMAGE. 30 * 31 * @(#)ffs_subr.c 8.5 (Berkeley) 3/21/95 32 */ 33 34#include <sys/cdefs.h> 35__FBSDID("$FreeBSD$"); 36 37#include <sys/param.h> 38 39#ifndef _KERNEL 40#include <stdio.h> 41#include <string.h> 42#include <stdlib.h> 43#include <time.h> 44#include <sys/errno.h> 45#include <ufs/ufs/dinode.h> 46#include <ufs/ffs/fs.h> 47 48struct malloc_type; 49#define UFS_MALLOC(size, type, flags) malloc(size) 50#define UFS_FREE(ptr, type) free(ptr) 51#define UFS_TIME time(NULL) 52 53#else /* _KERNEL */ 54#include <sys/systm.h> 55#include <sys/lock.h> 56#include <sys/malloc.h> 57#include <sys/mount.h> 58#include <sys/vnode.h> 59#include <sys/bio.h> 60#include <sys/buf.h> 61#include <sys/ucred.h> 62 63#include <ufs/ufs/quota.h> 64#include <ufs/ufs/inode.h> 65#include <ufs/ufs/extattr.h> 66#include <ufs/ufs/ufsmount.h> 67#include <ufs/ufs/ufs_extern.h> 68#include <ufs/ffs/ffs_extern.h> 69#include <ufs/ffs/fs.h> 70 71#define UFS_MALLOC(size, type, flags) malloc(size, type, flags) 72#define UFS_FREE(ptr, type) free(ptr, type) 73#define UFS_TIME time_second 74 75/* 76 * Return buffer with the contents of block "offset" from the beginning of 77 * directory "ip". If "res" is non-zero, fill it in with a pointer to the 78 * remaining space in the directory. 79 */ 80int 81ffs_blkatoff(struct vnode *vp, off_t offset, char **res, struct buf **bpp) 82{ 83 struct inode *ip; 84 struct fs *fs; 85 struct buf *bp; 86 ufs_lbn_t lbn; 87 int bsize, error; 88 89 ip = VTOI(vp); 90 fs = ITOFS(ip); 91 lbn = lblkno(fs, offset); 92 bsize = blksize(fs, ip, lbn); 93 94 *bpp = NULL; 95 error = bread(vp, lbn, bsize, NOCRED, &bp); 96 if (error) { 97 brelse(bp); 98 return (error); 99 } 100 if (res) 101 *res = (char *)bp->b_data + blkoff(fs, offset); 102 *bpp = bp; 103 return (0); 104} 105 106/* 107 * Load up the contents of an inode and copy the appropriate pieces 108 * to the incore copy. 109 */ 110void 111ffs_load_inode(struct buf *bp, struct inode *ip, struct fs *fs, ino_t ino) 112{ 113 114 if (I_IS_UFS1(ip)) { 115 *ip->i_din1 = 116 *((struct ufs1_dinode *)bp->b_data + ino_to_fsbo(fs, ino)); 117 ip->i_mode = ip->i_din1->di_mode; 118 ip->i_nlink = ip->i_din1->di_nlink; 119 ip->i_size = ip->i_din1->di_size; 120 ip->i_flags = ip->i_din1->di_flags; 121 ip->i_gen = ip->i_din1->di_gen; 122 ip->i_uid = ip->i_din1->di_uid; 123 ip->i_gid = ip->i_din1->di_gid; 124 } else { 125 *ip->i_din2 = 126 *((struct ufs2_dinode *)bp->b_data + ino_to_fsbo(fs, ino)); 127 ip->i_mode = ip->i_din2->di_mode; 128 ip->i_nlink = ip->i_din2->di_nlink; 129 ip->i_size = ip->i_din2->di_size; 130 ip->i_flags = ip->i_din2->di_flags; 131 ip->i_gen = ip->i_din2->di_gen; 132 ip->i_uid = ip->i_din2->di_uid; 133 ip->i_gid = ip->i_din2->di_gid; 134 } 135} 136 137/* 138 * Verify that a filesystem block number is a valid data block. 139 * This routine is only called on untrusted filesystems. 140 */ 141int 142ffs_check_blkno(struct mount *mp, ino_t inum, ufs2_daddr_t daddr, int blksize) 143{ 144 struct fs *fs; 145 struct ufsmount *ump; 146 ufs2_daddr_t end_daddr; 147 int cg, havemtx; 148 149 KASSERT((mp->mnt_flag & MNT_UNTRUSTED) != 0, 150 ("ffs_check_blkno called on a trusted file system")); 151 ump = VFSTOUFS(mp); 152 fs = ump->um_fs; 153 cg = dtog(fs, daddr); 154 end_daddr = daddr + numfrags(fs, blksize); 155 /* 156 * Verify that the block number is a valid data block. Also check 157 * that it does not point to an inode block or a superblock. Accept 158 * blocks that are unalloacted (0) or part of snapshot metadata 159 * (BLK_NOCOPY or BLK_SNAP). 160 * 161 * Thus, the block must be in a valid range for the filesystem and 162 * either in the space before a backup superblock (except the first 163 * cylinder group where that space is used by the bootstrap code) or 164 * after the inode blocks and before the end of the cylinder group. 165 */ 166 if ((uint64_t)daddr <= BLK_SNAP || 167 ((uint64_t)end_daddr <= fs->fs_size && 168 ((cg > 0 && end_daddr <= cgsblock(fs, cg)) || 169 (daddr >= cgdmin(fs, cg) && 170 end_daddr <= cgbase(fs, cg) + fs->fs_fpg)))) 171 return (0); 172 if ((havemtx = mtx_owned(UFS_MTX(ump))) == 0) 173 UFS_LOCK(ump); 174 if (ppsratecheck(&ump->um_last_integritymsg, 175 &ump->um_secs_integritymsg, 1)) { 176 UFS_UNLOCK(ump); 177 uprintf("\n%s: inode %jd, out-of-range indirect block " 178 "number %jd\n", mp->mnt_stat.f_mntonname, inum, daddr); 179 if (havemtx) 180 UFS_LOCK(ump); 181 } else if (!havemtx) 182 UFS_UNLOCK(ump); 183 return (EIO); 184} 185#endif /* _KERNEL */ 186 187/* 188 * These are the low-level functions that actually read and write 189 * the superblock and its associated data. 190 */ 191static off_t sblock_try[] = SBLOCKSEARCH; 192static int readsuper(void *, struct fs **, off_t, int, 193 int (*)(void *, off_t, void **, int)); 194 195/* 196 * Read a superblock from the devfd device. 197 * 198 * If an alternate superblock is specified, it is read. Otherwise the 199 * set of locations given in the SBLOCKSEARCH list is searched for a 200 * superblock. Memory is allocated for the superblock by the readfunc and 201 * is returned. If filltype is non-NULL, additional memory is allocated 202 * of type filltype and filled in with the superblock summary information. 203 * All memory is freed when any error is returned. 204 * 205 * If a superblock is found, zero is returned. Otherwise one of the 206 * following error values is returned: 207 * EIO: non-existent or truncated superblock. 208 * EIO: error reading summary information. 209 * ENOENT: no usable known superblock found. 210 * ENOSPC: failed to allocate space for the superblock. 211 * EINVAL: The previous newfs operation on this volume did not complete. 212 * The administrator must complete newfs before using this volume. 213 */ 214int 215ffs_sbget(void *devfd, struct fs **fsp, off_t altsblock, 216 struct malloc_type *filltype, 217 int (*readfunc)(void *devfd, off_t loc, void **bufp, int size)) 218{ 219 struct fs *fs; 220 int i, error, size, blks; 221 uint8_t *space; 222 int32_t *lp; 223 char *buf; 224 225 fs = NULL; 226 *fsp = NULL; 227 if (altsblock != -1) { 228 if ((error = readsuper(devfd, &fs, altsblock, 1, 229 readfunc)) != 0) { 230 if (fs != NULL) 231 UFS_FREE(fs, filltype); 232 return (error); 233 } 234 } else { 235 for (i = 0; sblock_try[i] != -1; i++) { 236 if ((error = readsuper(devfd, &fs, sblock_try[i], 0, 237 readfunc)) == 0) 238 break; 239 if (fs != NULL) { 240 UFS_FREE(fs, filltype); 241 fs = NULL; 242 } 243 if (error == ENOENT) 244 continue; 245 return (error); 246 } 247 if (sblock_try[i] == -1) 248 return (ENOENT); 249 } 250 /* 251 * Read in the superblock summary information. 252 */ 253 size = fs->fs_cssize; 254 blks = howmany(size, fs->fs_fsize); 255 if (fs->fs_contigsumsize > 0) 256 size += fs->fs_ncg * sizeof(int32_t); 257 size += fs->fs_ncg * sizeof(u_int8_t); 258 /* When running in libufs or libsa, UFS_MALLOC may fail */ 259 if ((space = UFS_MALLOC(size, filltype, M_WAITOK)) == NULL) { 260 UFS_FREE(fs, filltype); 261 return (ENOSPC); 262 } 263 fs->fs_csp = (struct csum *)space; 264 for (i = 0; i < blks; i += fs->fs_frag) { 265 size = fs->fs_bsize; 266 if (i + fs->fs_frag > blks) 267 size = (blks - i) * fs->fs_fsize; 268 buf = NULL; 269 error = (*readfunc)(devfd, 270 dbtob(fsbtodb(fs, fs->fs_csaddr + i)), (void **)&buf, size); 271 if (error) { 272 if (buf != NULL) 273 UFS_FREE(buf, filltype); 274 UFS_FREE(fs->fs_csp, filltype); 275 UFS_FREE(fs, filltype); 276 return (error); 277 } 278 memcpy(space, buf, size); 279 UFS_FREE(buf, filltype); 280 space += size; 281 } 282 if (fs->fs_contigsumsize > 0) { 283 fs->fs_maxcluster = lp = (int32_t *)space; 284 for (i = 0; i < fs->fs_ncg; i++) 285 *lp++ = fs->fs_contigsumsize; 286 space = (uint8_t *)lp; 287 } 288 size = fs->fs_ncg * sizeof(u_int8_t); 289 fs->fs_contigdirs = (u_int8_t *)space; 290 bzero(fs->fs_contigdirs, size); 291 *fsp = fs; 292 return (0); 293} 294 295/* 296 * Try to read a superblock from the location specified by sblockloc. 297 * Return zero on success or an errno on failure. 298 */ 299static int 300readsuper(void *devfd, struct fs **fsp, off_t sblockloc, int isaltsblk, 301 int (*readfunc)(void *devfd, off_t loc, void **bufp, int size)) 302{ 303 struct fs *fs; 304 int error; 305 306 error = (*readfunc)(devfd, sblockloc, (void **)fsp, SBLOCKSIZE); 307 if (error != 0) 308 return (error); 309 fs = *fsp; 310 if (fs->fs_magic == FS_BAD_MAGIC) 311 return (EINVAL); 312 if (((fs->fs_magic == FS_UFS1_MAGIC && (isaltsblk || 313 sblockloc <= SBLOCK_UFS1)) || 314 (fs->fs_magic == FS_UFS2_MAGIC && (isaltsblk || 315 sblockloc == fs->fs_sblockloc))) && 316 fs->fs_ncg >= 1 && 317 fs->fs_bsize >= MINBSIZE && 318 fs->fs_bsize <= MAXBSIZE && 319 fs->fs_bsize >= roundup(sizeof(struct fs), DEV_BSIZE) && 320 fs->fs_sbsize <= SBLOCKSIZE) { 321 /* 322 * If the filesystem has been run on a kernel without 323 * metadata check hashes, disable them. 324 */ 325 if ((fs->fs_flags & FS_METACKHASH) == 0) 326 fs->fs_metackhash = 0; 327 /* 328 * Clear any check-hashes that are not maintained 329 * by this kernel. Also clear any unsupported flags. 330 */ 331 fs->fs_metackhash &= CK_SUPPORTED; 332 fs->fs_flags &= FS_SUPPORTED; 333 /* Have to set for old filesystems that predate this field */ 334 fs->fs_sblockactualloc = sblockloc; 335 /* Not yet any summary information */ 336 fs->fs_csp = NULL; 337 return (0); 338 } 339 return (ENOENT); 340} 341 342/* 343 * Write a superblock to the devfd device from the memory pointed to by fs. 344 * Write out the superblock summary information if it is present. 345 * 346 * If the write is successful, zero is returned. Otherwise one of the 347 * following error values is returned: 348 * EIO: failed to write superblock. 349 * EIO: failed to write superblock summary information. 350 */ 351int 352ffs_sbput(void *devfd, struct fs *fs, off_t loc, 353 int (*writefunc)(void *devfd, off_t loc, void *buf, int size)) 354{ 355 int i, error, blks, size; 356 uint8_t *space; 357 358 /* 359 * If there is summary information, write it first, so if there 360 * is an error, the superblock will not be marked as clean. 361 */ 362 if (fs->fs_csp != NULL) { 363 blks = howmany(fs->fs_cssize, fs->fs_fsize); 364 space = (uint8_t *)fs->fs_csp; 365 for (i = 0; i < blks; i += fs->fs_frag) { 366 size = fs->fs_bsize; 367 if (i + fs->fs_frag > blks) 368 size = (blks - i) * fs->fs_fsize; 369 if ((error = (*writefunc)(devfd, 370 dbtob(fsbtodb(fs, fs->fs_csaddr + i)), 371 space, size)) != 0) 372 return (error); 373 space += size; 374 } 375 } 376 fs->fs_fmod = 0; 377 fs->fs_time = UFS_TIME; 378 if ((error = (*writefunc)(devfd, loc, fs, fs->fs_sbsize)) != 0) 379 return (error); 380 return (0); 381} 382 383/* 384 * Update the frsum fields to reflect addition or deletion 385 * of some frags. 386 */ 387void 388ffs_fragacct(struct fs *fs, int fragmap, int32_t fraglist[], int cnt) 389{ 390 int inblk; 391 int field, subfield; 392 int siz, pos; 393 394 inblk = (int)(fragtbl[fs->fs_frag][fragmap]) << 1; 395 fragmap <<= 1; 396 for (siz = 1; siz < fs->fs_frag; siz++) { 397 if ((inblk & (1 << (siz + (fs->fs_frag % NBBY)))) == 0) 398 continue; 399 field = around[siz]; 400 subfield = inside[siz]; 401 for (pos = siz; pos <= fs->fs_frag; pos++) { 402 if ((fragmap & field) == subfield) { 403 fraglist[siz] += cnt; 404 pos += siz; 405 field <<= siz; 406 subfield <<= siz; 407 } 408 field <<= 1; 409 subfield <<= 1; 410 } 411 } 412} 413 414/* 415 * block operations 416 * 417 * check if a block is available 418 */ 419int 420ffs_isblock(struct fs *fs, unsigned char *cp, ufs1_daddr_t h) 421{ 422 unsigned char mask; 423 424 switch ((int)fs->fs_frag) { 425 case 8: 426 return (cp[h] == 0xff); 427 case 4: 428 mask = 0x0f << ((h & 0x1) << 2); 429 return ((cp[h >> 1] & mask) == mask); 430 case 2: 431 mask = 0x03 << ((h & 0x3) << 1); 432 return ((cp[h >> 2] & mask) == mask); 433 case 1: 434 mask = 0x01 << (h & 0x7); 435 return ((cp[h >> 3] & mask) == mask); 436 default: 437#ifdef _KERNEL 438 panic("ffs_isblock"); 439#endif 440 break; 441 } 442 return (0); 443} 444 445/* 446 * check if a block is free 447 */ 448int 449ffs_isfreeblock(struct fs *fs, u_char *cp, ufs1_daddr_t h) 450{ 451 452 switch ((int)fs->fs_frag) { 453 case 8: 454 return (cp[h] == 0); 455 case 4: 456 return ((cp[h >> 1] & (0x0f << ((h & 0x1) << 2))) == 0); 457 case 2: 458 return ((cp[h >> 2] & (0x03 << ((h & 0x3) << 1))) == 0); 459 case 1: 460 return ((cp[h >> 3] & (0x01 << (h & 0x7))) == 0); 461 default: 462#ifdef _KERNEL 463 panic("ffs_isfreeblock"); 464#endif 465 break; 466 } 467 return (0); 468} 469 470/* 471 * take a block out of the map 472 */ 473void 474ffs_clrblock(struct fs *fs, u_char *cp, ufs1_daddr_t h) 475{ 476 477 switch ((int)fs->fs_frag) { 478 case 8: 479 cp[h] = 0; 480 return; 481 case 4: 482 cp[h >> 1] &= ~(0x0f << ((h & 0x1) << 2)); 483 return; 484 case 2: 485 cp[h >> 2] &= ~(0x03 << ((h & 0x3) << 1)); 486 return; 487 case 1: 488 cp[h >> 3] &= ~(0x01 << (h & 0x7)); 489 return; 490 default: 491#ifdef _KERNEL 492 panic("ffs_clrblock"); 493#endif 494 break; 495 } 496} 497 498/* 499 * put a block into the map 500 */ 501void 502ffs_setblock(struct fs *fs, unsigned char *cp, ufs1_daddr_t h) 503{ 504 505 switch ((int)fs->fs_frag) { 506 507 case 8: 508 cp[h] = 0xff; 509 return; 510 case 4: 511 cp[h >> 1] |= (0x0f << ((h & 0x1) << 2)); 512 return; 513 case 2: 514 cp[h >> 2] |= (0x03 << ((h & 0x3) << 1)); 515 return; 516 case 1: 517 cp[h >> 3] |= (0x01 << (h & 0x7)); 518 return; 519 default: 520#ifdef _KERNEL 521 panic("ffs_setblock"); 522#endif 523 break; 524 } 525} 526 527/* 528 * Update the cluster map because of an allocation or free. 529 * 530 * Cnt == 1 means free; cnt == -1 means allocating. 531 */ 532void 533ffs_clusteracct(struct fs *fs, struct cg *cgp, ufs1_daddr_t blkno, int cnt) 534{ 535 int32_t *sump; 536 int32_t *lp; 537 u_char *freemapp, *mapp; 538 int i, start, end, forw, back, map; 539 u_int bit; 540 541 if (fs->fs_contigsumsize <= 0) 542 return; 543 freemapp = cg_clustersfree(cgp); 544 sump = cg_clustersum(cgp); 545 /* 546 * Allocate or clear the actual block. 547 */ 548 if (cnt > 0) 549 setbit(freemapp, blkno); 550 else 551 clrbit(freemapp, blkno); 552 /* 553 * Find the size of the cluster going forward. 554 */ 555 start = blkno + 1; 556 end = start + fs->fs_contigsumsize; 557 if (end >= cgp->cg_nclusterblks) 558 end = cgp->cg_nclusterblks; 559 mapp = &freemapp[start / NBBY]; 560 map = *mapp++; 561 bit = 1U << (start % NBBY); 562 for (i = start; i < end; i++) { 563 if ((map & bit) == 0) 564 break; 565 if ((i & (NBBY - 1)) != (NBBY - 1)) { 566 bit <<= 1; 567 } else { 568 map = *mapp++; 569 bit = 1; 570 } 571 } 572 forw = i - start; 573 /* 574 * Find the size of the cluster going backward. 575 */ 576 start = blkno - 1; 577 end = start - fs->fs_contigsumsize; 578 if (end < 0) 579 end = -1; 580 mapp = &freemapp[start / NBBY]; 581 map = *mapp--; 582 bit = 1U << (start % NBBY); 583 for (i = start; i > end; i--) { 584 if ((map & bit) == 0) 585 break; 586 if ((i & (NBBY - 1)) != 0) { 587 bit >>= 1; 588 } else { 589 map = *mapp--; 590 bit = 1U << (NBBY - 1); 591 } 592 } 593 back = start - i; 594 /* 595 * Account for old cluster and the possibly new forward and 596 * back clusters. 597 */ 598 i = back + forw + 1; 599 if (i > fs->fs_contigsumsize) 600 i = fs->fs_contigsumsize; 601 sump[i] += cnt; 602 if (back > 0) 603 sump[back] -= cnt; 604 if (forw > 0) 605 sump[forw] -= cnt; 606 /* 607 * Update cluster summary information. 608 */ 609 lp = &sump[fs->fs_contigsumsize]; 610 for (i = fs->fs_contigsumsize; i > 0; i--) 611 if (*lp-- > 0) 612 break; 613 fs->fs_maxcluster[cgp->cg_cgx] = i; 614} 615