1 /* $NetBSD: ext2fs_vfsops.c,v 1.73 2004/07/05 07:28:45 pk Exp $ */ 2 3 /* 4 * Copyright (c) 1989, 1991, 1993, 1994 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_vfsops.c 8.14 (Berkeley) 11/28/94 32 * Modified for ext2fs by Manuel Bouyer. 33 */ 34 35 /* 36 * Copyright (c) 1997 Manuel Bouyer. 37 * 38 * Redistribution and use in source and binary forms, with or without 39 * modification, are permitted provided that the following conditions 40 * are met: 41 * 1. Redistributions of source code must retain the above copyright 42 * notice, this list of conditions and the following disclaimer. 43 * 2. Redistributions in binary form must reproduce the above copyright 44 * notice, this list of conditions and the following disclaimer in the 45 * documentation and/or other materials provided with the distribution. 46 * 3. All advertising materials mentioning features or use of this software 47 * must display the following acknowledgement: 48 * This product includes software developed by Manuel Bouyer. 49 * 4. The name of the author may not be used to endorse or promote products 50 * derived from this software without specific prior written permission. 51 * 52 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR 53 * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES 54 * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. 55 * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, 56 * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT 57 * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, 58 * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY 59 * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT 60 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF 61 * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. 62 * 63 * @(#)ffs_vfsops.c 8.14 (Berkeley) 11/28/94 64 * Modified for ext2fs by Manuel Bouyer. 65 */ 66 67 #include <sys/cdefs.h> 68 __KERNEL_RCSID(0, "$NetBSD: ext2fs_vfsops.c,v 1.73 2004/07/05 07:28:45 pk Exp $"); 69 70 #if defined(_KERNEL_OPT) 71 #include "opt_compat_netbsd.h" 72 #endif 73 74 #include <sys/param.h> 75 #include <sys/systm.h> 76 #include <sys/sysctl.h> 77 #include <sys/namei.h> 78 #include <sys/proc.h> 79 #include <sys/kernel.h> 80 #include <sys/vnode.h> 81 #include <sys/socket.h> 82 #include <sys/mount.h> 83 #include <sys/buf.h> 84 #include <sys/device.h> 85 #include <sys/mbuf.h> 86 #include <sys/file.h> 87 #include <sys/disklabel.h> 88 #include <sys/ioctl.h> 89 #include <sys/errno.h> 90 #include <sys/malloc.h> 91 #include <sys/pool.h> 92 #include <sys/lock.h> 93 #include <sys/conf.h> 94 95 #include <miscfs/specfs/specdev.h> 96 97 #include <ufs/ufs/quota.h> 98 #include <ufs/ufs/ufsmount.h> 99 #include <ufs/ufs/inode.h> 100 #include <ufs/ufs/dir.h> 101 #include <ufs/ufs/ufs_extern.h> 102 103 #include <ufs/ext2fs/ext2fs.h> 104 #include <ufs/ext2fs/ext2fs_extern.h> 105 106 extern struct lock ufs_hashlock; 107 108 int ext2fs_sbupdate __P((struct ufsmount *, int)); 109 static int ext2fs_checksb __P((struct ext2fs *, int)); 110 111 extern const struct vnodeopv_desc ext2fs_vnodeop_opv_desc; 112 extern const struct vnodeopv_desc ext2fs_specop_opv_desc; 113 extern const struct vnodeopv_desc ext2fs_fifoop_opv_desc; 114 115 const struct vnodeopv_desc * const ext2fs_vnodeopv_descs[] = { 116 &ext2fs_vnodeop_opv_desc, 117 &ext2fs_specop_opv_desc, 118 &ext2fs_fifoop_opv_desc, 119 NULL, 120 }; 121 122 struct vfsops ext2fs_vfsops = { 123 MOUNT_EXT2FS, 124 ext2fs_mount, 125 ufs_start, 126 ext2fs_unmount, 127 ufs_root, 128 ufs_quotactl, 129 ext2fs_statvfs, 130 ext2fs_sync, 131 ext2fs_vget, 132 ext2fs_fhtovp, 133 ext2fs_vptofh, 134 ext2fs_init, 135 ext2fs_reinit, 136 ext2fs_done, 137 NULL, 138 ext2fs_mountroot, 139 ufs_check_export, 140 (int (*)(struct mount *, struct vnode *, struct timespec *)) eopnotsupp, 141 ext2fs_vnodeopv_descs, 142 }; 143 144 struct genfs_ops ext2fs_genfsops = { 145 genfs_size, 146 ext2fs_gop_alloc, 147 genfs_gop_write, 148 }; 149 150 /* 151 * XXX Same structure as FFS inodes? Should we share a common pool? 152 */ 153 POOL_INIT(ext2fs_inode_pool, sizeof(struct inode), 0, 0, 0, "ext2fsinopl", 154 &pool_allocator_nointr); 155 POOL_INIT(ext2fs_dinode_pool, sizeof(struct ext2fs_dinode), 0, 0, 0, 156 "ext2dinopl", &pool_allocator_nointr); 157 158 extern u_long ext2gennumber; 159 160 void 161 ext2fs_init() 162 { 163 #ifdef _LKM 164 pool_init(&ext2fs_inode_pool, sizeof(struct inode), 0, 0, 0, 165 "ext2fsinopl", &pool_allocator_nointr); 166 pool_init(&ext2fs_dinode_pool, sizeof(struct ext2fs_dinode), 0, 0, 0, 167 "ext2dinopl", &pool_allocator_nointr); 168 #endif 169 ufs_init(); 170 } 171 172 void 173 ext2fs_reinit() 174 { 175 ufs_reinit(); 176 } 177 178 void 179 ext2fs_done() 180 { 181 ufs_done(); 182 #ifdef _LKM 183 pool_destroy(&ext2fs_inode_pool); 184 pool_destroy(&ext2fs_dinode_pool); 185 #endif 186 } 187 188 /* 189 * Called by main() when ext2fs is going to be mounted as root. 190 * 191 * Name is updated by mount(8) after booting. 192 */ 193 #define ROOTNAME "root_device" 194 195 int 196 ext2fs_mountroot() 197 { 198 extern struct vnode *rootvp; 199 struct m_ext2fs *fs; 200 struct mount *mp; 201 struct proc *p = curproc; /* XXX */ 202 struct ufsmount *ump; 203 int error; 204 205 if (root_device->dv_class != DV_DISK) 206 return (ENODEV); 207 208 /* 209 * Get vnodes for rootdev. 210 */ 211 if (bdevvp(rootdev, &rootvp)) 212 panic("ext2fs_mountroot: can't setup bdevvp's"); 213 214 if ((error = vfs_rootmountalloc(MOUNT_EXT2FS, "root_device", &mp))) { 215 vrele(rootvp); 216 return (error); 217 } 218 219 if ((error = ext2fs_mountfs(rootvp, mp, p)) != 0) { 220 mp->mnt_op->vfs_refcount--; 221 vfs_unbusy(mp); 222 free(mp, M_MOUNT); 223 vrele(rootvp); 224 return (error); 225 } 226 simple_lock(&mountlist_slock); 227 CIRCLEQ_INSERT_TAIL(&mountlist, mp, mnt_list); 228 simple_unlock(&mountlist_slock); 229 ump = VFSTOUFS(mp); 230 fs = ump->um_e2fs; 231 memset(fs->e2fs_fsmnt, 0, sizeof(fs->e2fs_fsmnt)); 232 (void) copystr(mp->mnt_stat.f_mntonname, fs->e2fs_fsmnt, 233 sizeof(fs->e2fs_fsmnt) - 1, 0); 234 if (fs->e2fs.e2fs_rev > E2FS_REV0) { 235 memset(fs->e2fs.e2fs_fsmnt, 0, sizeof(fs->e2fs.e2fs_fsmnt)); 236 (void) copystr(mp->mnt_stat.f_mntonname, fs->e2fs.e2fs_fsmnt, 237 sizeof(fs->e2fs.e2fs_fsmnt) - 1, 0); 238 } 239 (void)ext2fs_statvfs(mp, &mp->mnt_stat, p); 240 vfs_unbusy(mp); 241 setrootfstime((time_t)fs->e2fs.e2fs_wtime); 242 return (0); 243 } 244 245 /* 246 * VFS Operations. 247 * 248 * mount system call 249 */ 250 int 251 ext2fs_mount(mp, path, data, ndp, p) 252 struct mount *mp; 253 const char *path; 254 void * data; 255 struct nameidata *ndp; 256 struct proc *p; 257 { 258 struct vnode *devvp; 259 struct ufs_args args; 260 struct ufsmount *ump = NULL; 261 struct m_ext2fs *fs; 262 size_t size; 263 int error, flags; 264 mode_t accessmode; 265 266 if (mp->mnt_flag & MNT_GETARGS) { 267 ump = VFSTOUFS(mp); 268 if (ump == NULL) 269 return EIO; 270 args.fspec = NULL; 271 vfs_showexport(mp, &args.export, &ump->um_export); 272 return copyout(&args, data, sizeof(args)); 273 } 274 275 error = copyin(data, &args, sizeof (struct ufs_args)); 276 if (error) 277 return (error); 278 /* 279 * If updating, check whether changing from read-only to 280 * read/write; if there is no device name, that's all we do. 281 */ 282 if (mp->mnt_flag & MNT_UPDATE) { 283 ump = VFSTOUFS(mp); 284 fs = ump->um_e2fs; 285 if (fs->e2fs_ronly == 0 && (mp->mnt_flag & MNT_RDONLY)) { 286 flags = WRITECLOSE; 287 if (mp->mnt_flag & MNT_FORCE) 288 flags |= FORCECLOSE; 289 error = ext2fs_flushfiles(mp, flags, p); 290 if (error == 0 && 291 ext2fs_cgupdate(ump, MNT_WAIT) == 0 && 292 (fs->e2fs.e2fs_state & E2FS_ERRORS) == 0) { 293 fs->e2fs.e2fs_state = E2FS_ISCLEAN; 294 (void) ext2fs_sbupdate(ump, MNT_WAIT); 295 } 296 if (error) 297 return (error); 298 fs->e2fs_ronly = 1; 299 } 300 if (mp->mnt_flag & MNT_RELOAD) { 301 error = ext2fs_reload(mp, ndp->ni_cnd.cn_cred, p); 302 if (error) 303 return (error); 304 } 305 if (fs->e2fs_ronly && (mp->mnt_iflag & IMNT_WANTRDWR)) { 306 /* 307 * If upgrade to read-write by non-root, then verify 308 * that user has necessary permissions on the device. 309 */ 310 if (p->p_ucred->cr_uid != 0) { 311 devvp = ump->um_devvp; 312 vn_lock(devvp, LK_EXCLUSIVE | LK_RETRY); 313 error = VOP_ACCESS(devvp, VREAD | VWRITE, 314 p->p_ucred, p); 315 VOP_UNLOCK(devvp, 0); 316 if (error) 317 return (error); 318 } 319 fs->e2fs_ronly = 0; 320 if (fs->e2fs.e2fs_state == E2FS_ISCLEAN) 321 fs->e2fs.e2fs_state = 0; 322 else 323 fs->e2fs.e2fs_state = E2FS_ERRORS; 324 fs->e2fs_fmod = 1; 325 } 326 if (args.fspec == 0) { 327 /* 328 * Process export requests. 329 */ 330 return (vfs_export(mp, &ump->um_export, &args.export)); 331 } 332 } 333 /* 334 * Not an update, or updating the name: look up the name 335 * and verify that it refers to a sensible block device. 336 */ 337 NDINIT(ndp, LOOKUP, FOLLOW, UIO_USERSPACE, args.fspec, p); 338 if ((error = namei(ndp)) != 0) 339 return (error); 340 devvp = ndp->ni_vp; 341 342 if (devvp->v_type != VBLK) { 343 vrele(devvp); 344 return (ENOTBLK); 345 } 346 if (bdevsw_lookup(devvp->v_rdev) == NULL) { 347 vrele(devvp); 348 return (ENXIO); 349 } 350 /* 351 * If mount by non-root, then verify that user has necessary 352 * permissions on the device. 353 */ 354 if (p->p_ucred->cr_uid != 0) { 355 accessmode = VREAD; 356 if ((mp->mnt_flag & MNT_RDONLY) == 0) 357 accessmode |= VWRITE; 358 vn_lock(devvp, LK_EXCLUSIVE | LK_RETRY); 359 error = VOP_ACCESS(devvp, accessmode, p->p_ucred, p); 360 VOP_UNLOCK(devvp, 0); 361 if (error) { 362 vrele(devvp); 363 return (error); 364 } 365 } 366 if ((mp->mnt_flag & MNT_UPDATE) == 0) 367 error = ext2fs_mountfs(devvp, mp, p); 368 else { 369 if (devvp != ump->um_devvp) 370 error = EINVAL; /* needs translation */ 371 else 372 vrele(devvp); 373 } 374 if (error) { 375 vrele(devvp); 376 return (error); 377 } 378 ump = VFSTOUFS(mp); 379 fs = ump->um_e2fs; 380 error = set_statvfs_info(path, UIO_USERSPACE, args.fspec, 381 UIO_USERSPACE, mp, p); 382 (void) copystr(mp->mnt_stat.f_mntonname, fs->e2fs_fsmnt, 383 sizeof(fs->e2fs_fsmnt) - 1, &size); 384 memset(fs->e2fs_fsmnt + size, 0, sizeof(fs->e2fs_fsmnt) - size); 385 if (fs->e2fs.e2fs_rev > E2FS_REV0) { 386 (void) copystr(mp->mnt_stat.f_mntonname, fs->e2fs.e2fs_fsmnt, 387 sizeof(fs->e2fs.e2fs_fsmnt) - 1, &size); 388 memset(fs->e2fs.e2fs_fsmnt, 0, 389 sizeof(fs->e2fs.e2fs_fsmnt) - size); 390 } 391 if (fs->e2fs_fmod != 0) { /* XXX */ 392 fs->e2fs_fmod = 0; 393 if (fs->e2fs.e2fs_state == 0) 394 fs->e2fs.e2fs_wtime = time.tv_sec; 395 else 396 printf("%s: file system not clean; please fsck(8)\n", 397 mp->mnt_stat.f_mntfromname); 398 (void) ext2fs_cgupdate(ump, MNT_WAIT); 399 } 400 return error; 401 } 402 403 /* 404 * Reload all incore data for a filesystem (used after running fsck on 405 * the root filesystem and finding things to fix). The filesystem must 406 * be mounted read-only. 407 * 408 * Things to do to update the mount: 409 * 1) invalidate all cached meta-data. 410 * 2) re-read superblock from disk. 411 * 3) re-read summary information from disk. 412 * 4) invalidate all inactive vnodes. 413 * 5) invalidate all cached file data. 414 * 6) re-read inode data for all active vnodes. 415 */ 416 int 417 ext2fs_reload(mountp, cred, p) 418 struct mount *mountp; 419 struct ucred *cred; 420 struct proc *p; 421 { 422 struct vnode *vp, *nvp, *devvp; 423 struct inode *ip; 424 struct buf *bp; 425 struct m_ext2fs *fs; 426 struct ext2fs *newfs; 427 struct partinfo dpart; 428 int i, size, error; 429 caddr_t cp; 430 431 if ((mountp->mnt_flag & MNT_RDONLY) == 0) 432 return (EINVAL); 433 /* 434 * Step 1: invalidate all cached meta-data. 435 */ 436 devvp = VFSTOUFS(mountp)->um_devvp; 437 vn_lock(devvp, LK_EXCLUSIVE | LK_RETRY); 438 error = vinvalbuf(devvp, 0, cred, p, 0, 0); 439 VOP_UNLOCK(devvp, 0); 440 if (error) 441 panic("ext2fs_reload: dirty1"); 442 /* 443 * Step 2: re-read superblock from disk. 444 */ 445 if (VOP_IOCTL(devvp, DIOCGPART, &dpart, FREAD, NOCRED, p) != 0) 446 size = DEV_BSIZE; 447 else 448 size = dpart.disklab->d_secsize; 449 error = bread(devvp, (daddr_t)(SBOFF / size), SBSIZE, NOCRED, &bp); 450 if (error) { 451 brelse(bp); 452 return (error); 453 } 454 newfs = (struct ext2fs *)bp->b_data; 455 error = ext2fs_checksb(newfs, (mountp->mnt_flag & MNT_RDONLY) != 0); 456 if (error) { 457 brelse(bp); 458 return (error); 459 } 460 461 fs = VFSTOUFS(mountp)->um_e2fs; 462 /* 463 * copy in new superblock, and compute in-memory values 464 */ 465 e2fs_sbload(newfs, &fs->e2fs); 466 fs->e2fs_ncg = 467 howmany(fs->e2fs.e2fs_bcount - fs->e2fs.e2fs_first_dblock, 468 fs->e2fs.e2fs_bpg); 469 /* XXX assume hw bsize = 512 */ 470 fs->e2fs_fsbtodb = fs->e2fs.e2fs_log_bsize + 1; 471 fs->e2fs_bsize = 1024 << fs->e2fs.e2fs_log_bsize; 472 fs->e2fs_bshift = LOG_MINBSIZE + fs->e2fs.e2fs_log_bsize; 473 fs->e2fs_qbmask = fs->e2fs_bsize - 1; 474 fs->e2fs_bmask = ~fs->e2fs_qbmask; 475 fs->e2fs_ngdb = howmany(fs->e2fs_ncg, 476 fs->e2fs_bsize / sizeof(struct ext2_gd)); 477 fs->e2fs_ipb = fs->e2fs_bsize / EXT2_DINODE_SIZE; 478 fs->e2fs_itpg = fs->e2fs.e2fs_ipg/fs->e2fs_ipb; 479 480 /* 481 * Step 3: re-read summary information from disk. 482 */ 483 484 for (i=0; i < fs->e2fs_ngdb; i++) { 485 error = bread(devvp , 486 fsbtodb(fs, ((fs->e2fs_bsize>1024)? 0 : 1) + i + 1), 487 fs->e2fs_bsize, NOCRED, &bp); 488 if (error) { 489 brelse(bp); 490 return (error); 491 } 492 e2fs_cgload((struct ext2_gd*)bp->b_data, 493 &fs->e2fs_gd[i* fs->e2fs_bsize / sizeof(struct ext2_gd)], 494 fs->e2fs_bsize); 495 brelse(bp); 496 } 497 498 loop: 499 simple_lock(&mntvnode_slock); 500 for (vp = mountp->mnt_vnodelist.lh_first; vp != NULL; vp = nvp) { 501 if (vp->v_mount != mountp) { 502 simple_unlock(&mntvnode_slock); 503 goto loop; 504 } 505 nvp = vp->v_mntvnodes.le_next; 506 /* 507 * Step 4: invalidate all inactive vnodes. 508 */ 509 if (vrecycle(vp, &mntvnode_slock, p)) 510 goto loop; 511 /* 512 * Step 5: invalidate all cached file data. 513 */ 514 simple_lock(&vp->v_interlock); 515 simple_unlock(&mntvnode_slock); 516 if (vget(vp, LK_EXCLUSIVE | LK_INTERLOCK)) 517 goto loop; 518 if (vinvalbuf(vp, 0, cred, p, 0, 0)) 519 panic("ext2fs_reload: dirty2"); 520 /* 521 * Step 6: re-read inode data for all active vnodes. 522 */ 523 ip = VTOI(vp); 524 error = bread(devvp, fsbtodb(fs, ino_to_fsba(fs, ip->i_number)), 525 (int)fs->e2fs_bsize, NOCRED, &bp); 526 if (error) { 527 vput(vp); 528 return (error); 529 } 530 cp = (caddr_t)bp->b_data + 531 (ino_to_fsbo(fs, ip->i_number) * EXT2_DINODE_SIZE); 532 e2fs_iload((struct ext2fs_dinode *)cp, ip->i_din.e2fs_din); 533 brelse(bp); 534 vput(vp); 535 simple_lock(&mntvnode_slock); 536 } 537 simple_unlock(&mntvnode_slock); 538 return (0); 539 } 540 541 /* 542 * Common code for mount and mountroot 543 */ 544 int 545 ext2fs_mountfs(devvp, mp, p) 546 struct vnode *devvp; 547 struct mount *mp; 548 struct proc *p; 549 { 550 struct ufsmount *ump; 551 struct buf *bp; 552 struct ext2fs *fs; 553 struct m_ext2fs *m_fs; 554 dev_t dev; 555 struct partinfo dpart; 556 int error, i, size, ronly; 557 struct ucred *cred; 558 extern struct vnode *rootvp; 559 560 dev = devvp->v_rdev; 561 cred = p ? p->p_ucred : NOCRED; 562 /* 563 * Disallow multiple mounts of the same device. 564 * Disallow mounting of a device that is currently in use 565 * (except for root, which might share swap device for miniroot). 566 * Flush out any old buffers remaining from a previous use. 567 */ 568 if ((error = vfs_mountedon(devvp)) != 0) 569 return (error); 570 if (vcount(devvp) > 1 && devvp != rootvp) 571 return (EBUSY); 572 vn_lock(devvp, LK_EXCLUSIVE | LK_RETRY); 573 error = vinvalbuf(devvp, V_SAVE, cred, p, 0, 0); 574 VOP_UNLOCK(devvp, 0); 575 if (error) 576 return (error); 577 578 ronly = (mp->mnt_flag & MNT_RDONLY) != 0; 579 error = VOP_OPEN(devvp, ronly ? FREAD : FREAD|FWRITE, FSCRED, p); 580 if (error) 581 return (error); 582 if (VOP_IOCTL(devvp, DIOCGPART, &dpart, FREAD, cred, p) != 0) 583 size = DEV_BSIZE; 584 else 585 size = dpart.disklab->d_secsize; 586 587 bp = NULL; 588 ump = NULL; 589 590 #ifdef DEBUG_EXT2 591 printf("sb size: %d ino size %d\n", sizeof(struct ext2fs), 592 EXT2_DINODE_SIZE); 593 #endif 594 error = bread(devvp, (SBOFF / size), SBSIZE, cred, &bp); 595 if (error) 596 goto out; 597 fs = (struct ext2fs *)bp->b_data; 598 error = ext2fs_checksb(fs, ronly); 599 if (error) 600 goto out; 601 ump = malloc(sizeof *ump, M_UFSMNT, M_WAITOK); 602 memset(ump, 0, sizeof *ump); 603 ump->um_fstype = UFS1; 604 ump->um_e2fs = malloc(sizeof(struct m_ext2fs), M_UFSMNT, M_WAITOK); 605 memset(ump->um_e2fs, 0, sizeof(struct m_ext2fs)); 606 e2fs_sbload((struct ext2fs*)bp->b_data, &ump->um_e2fs->e2fs); 607 brelse(bp); 608 bp = NULL; 609 m_fs = ump->um_e2fs; 610 m_fs->e2fs_ronly = ronly; 611 if (ronly == 0) { 612 if (m_fs->e2fs.e2fs_state == E2FS_ISCLEAN) 613 m_fs->e2fs.e2fs_state = 0; 614 else 615 m_fs->e2fs.e2fs_state = E2FS_ERRORS; 616 m_fs->e2fs_fmod = 1; 617 } 618 619 /* compute dynamic sb infos */ 620 m_fs->e2fs_ncg = 621 howmany(m_fs->e2fs.e2fs_bcount - m_fs->e2fs.e2fs_first_dblock, 622 m_fs->e2fs.e2fs_bpg); 623 /* XXX assume hw bsize = 512 */ 624 m_fs->e2fs_fsbtodb = m_fs->e2fs.e2fs_log_bsize + 1; 625 m_fs->e2fs_bsize = 1024 << m_fs->e2fs.e2fs_log_bsize; 626 m_fs->e2fs_bshift = LOG_MINBSIZE + m_fs->e2fs.e2fs_log_bsize; 627 m_fs->e2fs_qbmask = m_fs->e2fs_bsize - 1; 628 m_fs->e2fs_bmask = ~m_fs->e2fs_qbmask; 629 m_fs->e2fs_ngdb = howmany(m_fs->e2fs_ncg, 630 m_fs->e2fs_bsize / sizeof(struct ext2_gd)); 631 m_fs->e2fs_ipb = m_fs->e2fs_bsize / EXT2_DINODE_SIZE; 632 m_fs->e2fs_itpg = m_fs->e2fs.e2fs_ipg/m_fs->e2fs_ipb; 633 634 m_fs->e2fs_gd = malloc(m_fs->e2fs_ngdb * m_fs->e2fs_bsize, 635 M_UFSMNT, M_WAITOK); 636 for (i=0; i < m_fs->e2fs_ngdb; i++) { 637 error = bread(devvp , 638 fsbtodb(m_fs, ((m_fs->e2fs_bsize>1024)? 0 : 1) + i + 1), 639 m_fs->e2fs_bsize, NOCRED, &bp); 640 if (error) { 641 free(m_fs->e2fs_gd, M_UFSMNT); 642 goto out; 643 } 644 e2fs_cgload((struct ext2_gd*)bp->b_data, 645 &m_fs->e2fs_gd[ 646 i * m_fs->e2fs_bsize / sizeof(struct ext2_gd)], 647 m_fs->e2fs_bsize); 648 brelse(bp); 649 bp = NULL; 650 } 651 652 mp->mnt_data = ump; 653 mp->mnt_stat.f_fsidx.__fsid_val[0] = (long)dev; 654 mp->mnt_stat.f_fsidx.__fsid_val[1] = makefstype(MOUNT_EXT2FS); 655 mp->mnt_stat.f_fsid = mp->mnt_stat.f_fsidx.__fsid_val[0]; 656 mp->mnt_stat.f_namemax = MAXNAMLEN; 657 mp->mnt_maxsymlinklen = EXT2_MAXSYMLINKLEN; 658 mp->mnt_flag |= MNT_LOCAL; 659 mp->mnt_dev_bshift = DEV_BSHIFT; /* XXX */ 660 mp->mnt_fs_bshift = m_fs->e2fs_bshift; 661 ump->um_flags = 0; 662 ump->um_mountp = mp; 663 ump->um_dev = dev; 664 ump->um_devvp = devvp; 665 ump->um_nindir = NINDIR(m_fs); 666 ump->um_lognindir = ffs(NINDIR(m_fs)) - 1; 667 ump->um_bptrtodb = m_fs->e2fs_fsbtodb; 668 ump->um_seqinc = 1; /* no frags */ 669 devvp->v_specmountpoint = mp; 670 return (0); 671 672 out: 673 if (bp) 674 brelse(bp); 675 vn_lock(devvp, LK_EXCLUSIVE | LK_RETRY); 676 (void)VOP_CLOSE(devvp, ronly ? FREAD : FREAD|FWRITE, cred, p); 677 VOP_UNLOCK(devvp, 0); 678 if (ump) { 679 free(ump->um_e2fs, M_UFSMNT); 680 free(ump, M_UFSMNT); 681 mp->mnt_data = NULL; 682 } 683 return (error); 684 } 685 686 /* 687 * unmount system call 688 */ 689 int 690 ext2fs_unmount(mp, mntflags, p) 691 struct mount *mp; 692 int mntflags; 693 struct proc *p; 694 { 695 struct ufsmount *ump; 696 struct m_ext2fs *fs; 697 int error, flags; 698 699 flags = 0; 700 if (mntflags & MNT_FORCE) 701 flags |= FORCECLOSE; 702 if ((error = ext2fs_flushfiles(mp, flags, p)) != 0) 703 return (error); 704 ump = VFSTOUFS(mp); 705 fs = ump->um_e2fs; 706 if (fs->e2fs_ronly == 0 && 707 ext2fs_cgupdate(ump, MNT_WAIT) == 0 && 708 (fs->e2fs.e2fs_state & E2FS_ERRORS) == 0) { 709 fs->e2fs.e2fs_state = E2FS_ISCLEAN; 710 (void) ext2fs_sbupdate(ump, MNT_WAIT); 711 } 712 if (ump->um_devvp->v_type != VBAD) 713 ump->um_devvp->v_specmountpoint = NULL; 714 vn_lock(ump->um_devvp, LK_EXCLUSIVE | LK_RETRY); 715 error = VOP_CLOSE(ump->um_devvp, fs->e2fs_ronly ? FREAD : FREAD|FWRITE, 716 NOCRED, p); 717 vput(ump->um_devvp); 718 free(fs->e2fs_gd, M_UFSMNT); 719 free(fs, M_UFSMNT); 720 free(ump, M_UFSMNT); 721 mp->mnt_data = NULL; 722 mp->mnt_flag &= ~MNT_LOCAL; 723 return (error); 724 } 725 726 /* 727 * Flush out all the files in a filesystem. 728 */ 729 int 730 ext2fs_flushfiles(mp, flags, p) 731 struct mount *mp; 732 int flags; 733 struct proc *p; 734 { 735 extern int doforce; 736 int error; 737 738 if (!doforce) 739 flags &= ~FORCECLOSE; 740 error = vflush(mp, NULLVP, flags); 741 return (error); 742 } 743 744 /* 745 * Get file system statistics. 746 */ 747 int 748 ext2fs_statvfs(mp, sbp, p) 749 struct mount *mp; 750 struct statvfs *sbp; 751 struct proc *p; 752 { 753 struct ufsmount *ump; 754 struct m_ext2fs *fs; 755 u_int32_t overhead, overhead_per_group; 756 int i, ngroups; 757 758 ump = VFSTOUFS(mp); 759 fs = ump->um_e2fs; 760 if (fs->e2fs.e2fs_magic != E2FS_MAGIC) 761 panic("ext2fs_statvfs"); 762 763 /* 764 * Compute the overhead (FS structures) 765 */ 766 overhead_per_group = 1 /* block bitmap */ + 767 1 /* inode bitmap */ + 768 fs->e2fs_itpg; 769 overhead = fs->e2fs.e2fs_first_dblock + 770 fs->e2fs_ncg * overhead_per_group; 771 if (fs->e2fs.e2fs_rev > E2FS_REV0 && 772 fs->e2fs.e2fs_features_rocompat & EXT2F_ROCOMPAT_SPARSESUPER) { 773 for (i = 0, ngroups = 0; i < fs->e2fs_ncg; i++) { 774 if (cg_has_sb(i)) 775 ngroups++; 776 } 777 } else { 778 ngroups = fs->e2fs_ncg; 779 } 780 overhead += ngroups * (1 + fs->e2fs_ngdb); 781 782 sbp->f_bsize = fs->e2fs_bsize; 783 sbp->f_frsize = fs->e2fs.e2fs_fsize; 784 sbp->f_iosize = fs->e2fs_bsize; 785 sbp->f_blocks = fs->e2fs.e2fs_bcount - overhead; 786 sbp->f_bfree = fs->e2fs.e2fs_fbcount; 787 sbp->f_bresvd = fs->e2fs.e2fs_rbcount; 788 if (sbp->f_bfree > sbp->f_bresvd) 789 sbp->f_bavail = sbp->f_bfree - sbp->f_bresvd; 790 else 791 sbp->f_bavail = 0; 792 sbp->f_files = fs->e2fs.e2fs_icount; 793 sbp->f_ffree = fs->e2fs.e2fs_ficount; 794 sbp->f_favail = fs->e2fs.e2fs_ficount; 795 sbp->f_fresvd = 0; 796 copy_statvfs_info(sbp, mp); 797 return (0); 798 } 799 800 /* 801 * Go through the disk queues to initiate sandbagged IO; 802 * go through the inodes to write those that have been modified; 803 * initiate the writing of the super block if it has been modified. 804 * 805 * Note: we are always called with the filesystem marked `MPBUSY'. 806 */ 807 int 808 ext2fs_sync(mp, waitfor, cred, p) 809 struct mount *mp; 810 int waitfor; 811 struct ucred *cred; 812 struct proc *p; 813 { 814 struct vnode *vp, *nvp; 815 struct inode *ip; 816 struct ufsmount *ump = VFSTOUFS(mp); 817 struct m_ext2fs *fs; 818 int error, allerror = 0; 819 820 fs = ump->um_e2fs; 821 if (fs->e2fs_fmod != 0 && fs->e2fs_ronly != 0) { /* XXX */ 822 printf("fs = %s\n", fs->e2fs_fsmnt); 823 panic("update: rofs mod"); 824 } 825 /* 826 * Write back each (modified) inode. 827 */ 828 simple_lock(&mntvnode_slock); 829 loop: 830 for (vp = LIST_FIRST(&mp->mnt_vnodelist); vp != NULL; vp = nvp) { 831 /* 832 * If the vnode that we are about to sync is no longer 833 * associated with this mount point, start over. 834 */ 835 if (vp->v_mount != mp) 836 goto loop; 837 simple_lock(&vp->v_interlock); 838 nvp = LIST_NEXT(vp, v_mntvnodes); 839 ip = VTOI(vp); 840 if (waitfor == MNT_LAZY || vp->v_type == VNON || 841 ((ip->i_flag & 842 (IN_ACCESS | IN_CHANGE | IN_UPDATE | IN_MODIFIED | IN_ACCESSED)) == 0 && 843 LIST_EMPTY(&vp->v_dirtyblkhd) && 844 vp->v_uobj.uo_npages == 0)) 845 { 846 simple_unlock(&vp->v_interlock); 847 continue; 848 } 849 simple_unlock(&mntvnode_slock); 850 error = vget(vp, LK_EXCLUSIVE | LK_NOWAIT | LK_INTERLOCK); 851 if (error) { 852 simple_lock(&mntvnode_slock); 853 if (error == ENOENT) 854 goto loop; 855 continue; 856 } 857 if ((error = VOP_FSYNC(vp, cred, 858 waitfor == MNT_WAIT ? FSYNC_WAIT : 0, 0, 0, p)) != 0) 859 allerror = error; 860 vput(vp); 861 simple_lock(&mntvnode_slock); 862 } 863 simple_unlock(&mntvnode_slock); 864 /* 865 * Force stale file system control information to be flushed. 866 */ 867 if (waitfor != MNT_LAZY) { 868 vn_lock(ump->um_devvp, LK_EXCLUSIVE | LK_RETRY); 869 if ((error = VOP_FSYNC(ump->um_devvp, cred, 870 waitfor == MNT_WAIT ? FSYNC_WAIT : 0, 0, 0, p)) != 0) 871 allerror = error; 872 VOP_UNLOCK(ump->um_devvp, 0); 873 } 874 /* 875 * Write back modified superblock. 876 */ 877 if (fs->e2fs_fmod != 0) { 878 fs->e2fs_fmod = 0; 879 fs->e2fs.e2fs_wtime = time.tv_sec; 880 if ((error = ext2fs_cgupdate(ump, waitfor))) 881 allerror = error; 882 } 883 return (allerror); 884 } 885 886 /* 887 * Look up a EXT2FS dinode number to find its incore vnode, otherwise read it 888 * in from disk. If it is in core, wait for the lock bit to clear, then 889 * return the inode locked. Detection and handling of mount points must be 890 * done by the calling routine. 891 */ 892 int 893 ext2fs_vget(mp, ino, vpp) 894 struct mount *mp; 895 ino_t ino; 896 struct vnode **vpp; 897 { 898 struct m_ext2fs *fs; 899 struct inode *ip; 900 struct ufsmount *ump; 901 struct buf *bp; 902 struct vnode *vp; 903 dev_t dev; 904 int error; 905 caddr_t cp; 906 907 ump = VFSTOUFS(mp); 908 dev = ump->um_dev; 909 910 if ((*vpp = ufs_ihashget(dev, ino, LK_EXCLUSIVE)) != NULL) 911 return (0); 912 913 /* Allocate a new vnode/inode. */ 914 if ((error = getnewvnode(VT_EXT2FS, mp, ext2fs_vnodeop_p, &vp)) != 0) { 915 *vpp = NULL; 916 return (error); 917 } 918 919 do { 920 if ((*vpp = ufs_ihashget(dev, ino, LK_EXCLUSIVE)) != NULL) { 921 ungetnewvnode(vp); 922 return (0); 923 } 924 } while (lockmgr(&ufs_hashlock, LK_EXCLUSIVE|LK_SLEEPFAIL, 0)); 925 926 ip = pool_get(&ext2fs_inode_pool, PR_WAITOK); 927 memset(ip, 0, sizeof(struct inode)); 928 vp->v_data = ip; 929 ip->i_vnode = vp; 930 ip->i_ump = ump; 931 ip->i_e2fs = fs = ump->um_e2fs; 932 ip->i_dev = dev; 933 ip->i_number = ino; 934 ip->i_e2fs_last_lblk = 0; 935 ip->i_e2fs_last_blk = 0; 936 937 /* 938 * Put it onto its hash chain and lock it so that other requests for 939 * this inode will block if they arrive while we are sleeping waiting 940 * for old data structures to be purged or for the contents of the 941 * disk portion of this inode to be read. 942 */ 943 944 ufs_ihashins(ip); 945 lockmgr(&ufs_hashlock, LK_RELEASE, 0); 946 947 /* Read in the disk contents for the inode, copy into the inode. */ 948 error = bread(ump->um_devvp, fsbtodb(fs, ino_to_fsba(fs, ino)), 949 (int)fs->e2fs_bsize, NOCRED, &bp); 950 if (error) { 951 952 /* 953 * The inode does not contain anything useful, so it would 954 * be misleading to leave it on its hash chain. With mode 955 * still zero, it will be unlinked and returned to the free 956 * list by vput(). 957 */ 958 959 vput(vp); 960 brelse(bp); 961 *vpp = NULL; 962 return (error); 963 } 964 cp = (caddr_t)bp->b_data + 965 (ino_to_fsbo(fs, ino) * EXT2_DINODE_SIZE); 966 ip->i_din.e2fs_din = pool_get(&ext2fs_dinode_pool, PR_WAITOK); 967 e2fs_iload((struct ext2fs_dinode *)cp, ip->i_din.e2fs_din); 968 brelse(bp); 969 970 /* If the inode was deleted, reset all fields */ 971 if (ip->i_e2fs_dtime != 0) { 972 ip->i_e2fs_mode = ip->i_e2fs_size = ip->i_e2fs_nblock = 0; 973 memset(ip->i_e2fs_blocks, 0, sizeof(ip->i_e2fs_blocks)); 974 } 975 976 /* 977 * Initialize the vnode from the inode, check for aliases. 978 * Note that the underlying vnode may have changed. 979 */ 980 981 error = ext2fs_vinit(mp, ext2fs_specop_p, ext2fs_fifoop_p, &vp); 982 if (error) { 983 vput(vp); 984 *vpp = NULL; 985 return (error); 986 } 987 /* 988 * Finish inode initialization now that aliasing has been resolved. 989 */ 990 991 genfs_node_init(vp, &ext2fs_genfsops); 992 ip->i_devvp = ump->um_devvp; 993 VREF(ip->i_devvp); 994 995 /* 996 * Set up a generation number for this inode if it does not 997 * already have one. This should only happen on old filesystems. 998 */ 999 1000 if (ip->i_e2fs_gen == 0) { 1001 if (++ext2gennumber < (u_long)time.tv_sec) 1002 ext2gennumber = time.tv_sec; 1003 ip->i_e2fs_gen = ext2gennumber; 1004 if ((vp->v_mount->mnt_flag & MNT_RDONLY) == 0) 1005 ip->i_flag |= IN_MODIFIED; 1006 } 1007 vp->v_size = ip->i_e2fs_size; 1008 *vpp = vp; 1009 return (0); 1010 } 1011 1012 /* 1013 * File handle to vnode 1014 * 1015 * Have to be really careful about stale file handles: 1016 * - check that the inode number is valid 1017 * - call ext2fs_vget() to get the locked inode 1018 * - check for an unallocated inode (i_mode == 0) 1019 */ 1020 int 1021 ext2fs_fhtovp(mp, fhp, vpp) 1022 struct mount *mp; 1023 struct fid *fhp; 1024 struct vnode **vpp; 1025 { 1026 struct inode *ip; 1027 struct vnode *nvp; 1028 int error; 1029 struct ufid *ufhp; 1030 struct m_ext2fs *fs; 1031 1032 ufhp = (struct ufid *)fhp; 1033 fs = VFSTOUFS(mp)->um_e2fs; 1034 if ((ufhp->ufid_ino < EXT2_FIRSTINO && ufhp->ufid_ino != EXT2_ROOTINO) || 1035 ufhp->ufid_ino >= fs->e2fs_ncg * fs->e2fs.e2fs_ipg) 1036 return (ESTALE); 1037 1038 if ((error = VFS_VGET(mp, ufhp->ufid_ino, &nvp)) != 0) { 1039 *vpp = NULLVP; 1040 return (error); 1041 } 1042 ip = VTOI(nvp); 1043 if (ip->i_e2fs_mode == 0 || ip->i_e2fs_dtime != 0 || 1044 ip->i_e2fs_gen != ufhp->ufid_gen) { 1045 vput(nvp); 1046 *vpp = NULLVP; 1047 return (ESTALE); 1048 } 1049 *vpp = nvp; 1050 return (0); 1051 } 1052 1053 /* 1054 * Vnode pointer to File handle 1055 */ 1056 /* ARGSUSED */ 1057 int 1058 ext2fs_vptofh(vp, fhp) 1059 struct vnode *vp; 1060 struct fid *fhp; 1061 { 1062 struct inode *ip; 1063 struct ufid *ufhp; 1064 1065 ip = VTOI(vp); 1066 ufhp = (struct ufid *)fhp; 1067 ufhp->ufid_len = sizeof(struct ufid); 1068 ufhp->ufid_ino = ip->i_number; 1069 ufhp->ufid_gen = ip->i_e2fs_gen; 1070 return (0); 1071 } 1072 1073 SYSCTL_SETUP(sysctl_vfs_ext2fs_setup, "sysctl vfs.ext2fs subtree setup") 1074 { 1075 1076 sysctl_createv(clog, 0, NULL, NULL, 1077 CTLFLAG_PERMANENT, 1078 CTLTYPE_NODE, "vfs", NULL, 1079 NULL, 0, NULL, 0, 1080 CTL_VFS, CTL_EOL); 1081 sysctl_createv(clog, 0, NULL, NULL, 1082 CTLFLAG_PERMANENT, 1083 CTLTYPE_NODE, "ext2fs", 1084 SYSCTL_DESCR("Linux EXT2FS file system"), 1085 NULL, 0, NULL, 0, 1086 CTL_VFS, 17, CTL_EOL); 1087 /* 1088 * XXX the "17" above could be dynamic, thereby eliminating 1089 * one more instance of the "number to vfs" mapping problem, 1090 * but "17" is the order as taken from sys/mount.h 1091 */ 1092 } 1093 1094 /* 1095 * Write a superblock and associated information back to disk. 1096 */ 1097 int 1098 ext2fs_sbupdate(mp, waitfor) 1099 struct ufsmount *mp; 1100 int waitfor; 1101 { 1102 struct m_ext2fs *fs = mp->um_e2fs; 1103 struct buf *bp; 1104 int error = 0; 1105 1106 bp = getblk(mp->um_devvp, SBLOCK, SBSIZE, 0, 0); 1107 e2fs_sbsave(&fs->e2fs, (struct ext2fs*)bp->b_data); 1108 if (waitfor == MNT_WAIT) 1109 error = bwrite(bp); 1110 else 1111 bawrite(bp); 1112 return (error); 1113 } 1114 1115 int 1116 ext2fs_cgupdate(mp, waitfor) 1117 struct ufsmount *mp; 1118 int waitfor; 1119 { 1120 struct m_ext2fs *fs = mp->um_e2fs; 1121 struct buf *bp; 1122 int i, error = 0, allerror = 0; 1123 1124 allerror = ext2fs_sbupdate(mp, waitfor); 1125 for (i = 0; i < fs->e2fs_ngdb; i++) { 1126 bp = getblk(mp->um_devvp, fsbtodb(fs, ((fs->e2fs_bsize>1024)?0:1)+i+1), 1127 fs->e2fs_bsize, 0, 0); 1128 e2fs_cgsave(&fs->e2fs_gd[i* fs->e2fs_bsize / sizeof(struct ext2_gd)], 1129 (struct ext2_gd*)bp->b_data, fs->e2fs_bsize); 1130 if (waitfor == MNT_WAIT) 1131 error = bwrite(bp); 1132 else 1133 bawrite(bp); 1134 } 1135 1136 if (!allerror && error) 1137 allerror = error; 1138 return (allerror); 1139 } 1140 1141 static int 1142 ext2fs_checksb(fs, ronly) 1143 struct ext2fs *fs; 1144 int ronly; 1145 { 1146 if (fs2h16(fs->e2fs_magic) != E2FS_MAGIC) { 1147 return (EIO); /* XXX needs translation */ 1148 } 1149 if (fs2h32(fs->e2fs_rev) > E2FS_REV1) { 1150 #ifdef DIAGNOSTIC 1151 printf("Ext2 fs: unsupported revision number: %x\n", 1152 fs2h32(fs->e2fs_rev)); 1153 #endif 1154 return (EIO); /* XXX needs translation */ 1155 } 1156 if (fs2h32(fs->e2fs_log_bsize) > 2) { /* block size = 1024|2048|4096 */ 1157 #ifdef DIAGNOSTIC 1158 printf("Ext2 fs: bad block size: %d (expected <=2 for ext2 fs)\n", 1159 fs2h32(fs->e2fs_log_bsize)); 1160 #endif 1161 return (EIO); /* XXX needs translation */ 1162 } 1163 if (fs2h32(fs->e2fs_rev) > E2FS_REV0) { 1164 if (fs2h32(fs->e2fs_first_ino) != EXT2_FIRSTINO || 1165 fs2h16(fs->e2fs_inode_size) != EXT2_DINODE_SIZE) { 1166 printf("Ext2 fs: unsupported inode size\n"); 1167 return (EINVAL); /* XXX needs translation */ 1168 } 1169 if (fs2h32(fs->e2fs_features_incompat) & 1170 ~EXT2F_INCOMPAT_SUPP) { 1171 printf("Ext2 fs: unsupported optional feature\n"); 1172 return (EINVAL); /* XXX needs translation */ 1173 } 1174 if (!ronly && fs2h32(fs->e2fs_features_rocompat) & 1175 ~EXT2F_ROCOMPAT_SUPP) { 1176 return (EROFS); /* XXX needs translation */ 1177 } 1178 } 1179 return (0); 1180 } 1181