tools/llvm: Do not build with symbols
[minix3.git] / sys / ufs / ext2fs / ext2fs_vfsops.c
blob1a25831ff94deb6fd8bdc91e1ac1853a4693ace9
1 /* $NetBSD: ext2fs_vfsops.c,v 1.175 2013/11/23 13:35:36 christos Exp $ */
3 /*
4 * Copyright (c) 1989, 1991, 1993, 1994
5 * The Regents of the University of California. All rights reserved.
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.
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.
31 * @(#)ffs_vfsops.c 8.14 (Berkeley) 11/28/94
32 * Modified for ext2fs by Manuel Bouyer.
36 * Copyright (c) 1997 Manuel Bouyer.
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.
47 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR
48 * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
49 * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
50 * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT,
51 * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
52 * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
53 * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
54 * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
55 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF
56 * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
58 * @(#)ffs_vfsops.c 8.14 (Berkeley) 11/28/94
59 * Modified for ext2fs by Manuel Bouyer.
62 #include <sys/cdefs.h>
63 __KERNEL_RCSID(0, "$NetBSD: ext2fs_vfsops.c,v 1.175 2013/11/23 13:35:36 christos Exp $");
65 #if defined(_KERNEL_OPT)
66 #include "opt_compat_netbsd.h"
67 #endif
69 #include <sys/param.h>
70 #include <sys/systm.h>
71 #include <sys/sysctl.h>
72 #include <sys/namei.h>
73 #include <sys/proc.h>
74 #include <sys/kernel.h>
75 #include <sys/vnode.h>
76 #include <sys/socket.h>
77 #include <sys/mount.h>
78 #include <sys/buf.h>
79 #include <sys/device.h>
80 #include <sys/mbuf.h>
81 #include <sys/file.h>
82 #include <sys/disklabel.h>
83 #include <sys/ioctl.h>
84 #include <sys/errno.h>
85 #include <sys/malloc.h>
86 #include <sys/pool.h>
87 #include <sys/lock.h>
88 #include <sys/conf.h>
89 #include <sys/kauth.h>
90 #include <sys/module.h>
92 #include <miscfs/genfs/genfs.h>
93 #include <miscfs/specfs/specdev.h>
95 #include <ufs/ufs/quota.h>
96 #include <ufs/ufs/ufsmount.h>
97 #include <ufs/ufs/inode.h>
98 #include <ufs/ufs/dir.h>
99 #include <ufs/ufs/ufs_extern.h>
101 #include <ufs/ext2fs/ext2fs.h>
102 #include <ufs/ext2fs/ext2fs_dir.h>
103 #include <ufs/ext2fs/ext2fs_extern.h>
105 MODULE(MODULE_CLASS_VFS, ext2fs, "ffs");
107 int ext2fs_sbupdate(struct ufsmount *, int);
108 static int ext2fs_checksb(struct ext2fs *, int);
110 static struct sysctllog *ext2fs_sysctl_log;
112 extern const struct vnodeopv_desc ext2fs_vnodeop_opv_desc;
113 extern const struct vnodeopv_desc ext2fs_specop_opv_desc;
114 extern const struct vnodeopv_desc ext2fs_fifoop_opv_desc;
116 const struct vnodeopv_desc * const ext2fs_vnodeopv_descs[] = {
117 &ext2fs_vnodeop_opv_desc,
118 &ext2fs_specop_opv_desc,
119 &ext2fs_fifoop_opv_desc,
120 NULL,
123 struct vfsops ext2fs_vfsops = {
124 MOUNT_EXT2FS,
125 sizeof (struct ufs_args),
126 ext2fs_mount,
127 ufs_start,
128 ext2fs_unmount,
129 ufs_root,
130 ufs_quotactl,
131 ext2fs_statvfs,
132 ext2fs_sync,
133 ext2fs_vget,
134 ext2fs_fhtovp,
135 ext2fs_vptofh,
136 ext2fs_init,
137 ext2fs_reinit,
138 ext2fs_done,
139 ext2fs_mountroot,
140 (int (*)(struct mount *, struct vnode *, struct timespec *)) eopnotsupp,
141 vfs_stdextattrctl,
142 (void *)eopnotsupp, /* vfs_suspendctl */
143 genfs_renamelock_enter,
144 genfs_renamelock_exit,
145 (void *)eopnotsupp,
146 ext2fs_vnodeopv_descs,
148 { NULL, NULL },
151 static const struct genfs_ops ext2fs_genfsops = {
152 .gop_size = genfs_size,
153 .gop_alloc = ext2fs_gop_alloc,
154 .gop_write = genfs_gop_write,
155 .gop_markupdate = ufs_gop_markupdate,
158 static const struct ufs_ops ext2fs_ufsops = {
159 .uo_itimes = ext2fs_itimes,
160 .uo_update = ext2fs_update,
161 .uo_vfree = ext2fs_vfree,
164 /* Fill in the inode uid/gid from ext2 halves. */
165 void
166 ext2fs_set_inode_guid(struct inode *ip)
169 ip->i_gid = ip->i_e2fs_gid;
170 ip->i_uid = ip->i_e2fs_uid;
171 if (ip->i_e2fs->e2fs.e2fs_rev > E2FS_REV0) {
172 ip->i_gid |= ip->i_e2fs_gid_high << 16;
173 ip->i_uid |= ip->i_e2fs_uid_high << 16;
177 static int
178 ext2fs_modcmd(modcmd_t cmd, void *arg)
180 int error;
182 switch (cmd) {
183 case MODULE_CMD_INIT:
184 error = vfs_attach(&ext2fs_vfsops);
185 if (error != 0)
186 break;
187 sysctl_createv(&ext2fs_sysctl_log, 0, NULL, NULL,
188 CTLFLAG_PERMANENT,
189 CTLTYPE_NODE, "vfs", NULL,
190 NULL, 0, NULL, 0,
191 CTL_VFS, CTL_EOL);
192 sysctl_createv(&ext2fs_sysctl_log, 0, NULL, NULL,
193 CTLFLAG_PERMANENT,
194 CTLTYPE_NODE, "ext2fs",
195 SYSCTL_DESCR("Linux EXT2FS file system"),
196 NULL, 0, NULL, 0,
197 CTL_VFS, 17, CTL_EOL);
199 * XXX the "17" above could be dynamic, thereby eliminating
200 * one more instance of the "number to vfs" mapping problem,
201 * but "17" is the order as taken from sys/mount.h
203 break;
204 case MODULE_CMD_FINI:
205 error = vfs_detach(&ext2fs_vfsops);
206 if (error != 0)
207 break;
208 sysctl_teardown(&ext2fs_sysctl_log);
209 break;
210 default:
211 error = ENOTTY;
212 break;
215 return (error);
219 * XXX Same structure as FFS inodes? Should we share a common pool?
221 struct pool ext2fs_inode_pool;
222 struct pool ext2fs_dinode_pool;
224 extern u_long ext2gennumber;
226 void
227 ext2fs_init(void)
230 pool_init(&ext2fs_inode_pool, sizeof(struct inode), 0, 0, 0,
231 "ext2fsinopl", &pool_allocator_nointr, IPL_NONE);
232 pool_init(&ext2fs_dinode_pool, sizeof(struct ext2fs_dinode), 0, 0, 0,
233 "ext2dinopl", &pool_allocator_nointr, IPL_NONE);
234 ufs_init();
237 void
238 ext2fs_reinit(void)
240 ufs_reinit();
243 void
244 ext2fs_done(void)
247 ufs_done();
248 pool_destroy(&ext2fs_inode_pool);
249 pool_destroy(&ext2fs_dinode_pool);
253 * Called by main() when ext2fs is going to be mounted as root.
255 * Name is updated by mount(8) after booting.
257 #define ROOTNAME "root_device"
260 ext2fs_mountroot(void)
262 extern struct vnode *rootvp;
263 struct m_ext2fs *fs;
264 struct mount *mp;
265 struct ufsmount *ump;
266 int error;
268 if (device_class(root_device) != DV_DISK)
269 return (ENODEV);
271 if ((error = vfs_rootmountalloc(MOUNT_EXT2FS, "root_device", &mp))) {
272 vrele(rootvp);
273 return (error);
276 if ((error = ext2fs_mountfs(rootvp, mp)) != 0) {
277 vfs_unbusy(mp, false, NULL);
278 vfs_destroy(mp);
279 return (error);
281 mountlist_append(mp);
282 ump = VFSTOUFS(mp);
283 fs = ump->um_e2fs;
284 memset(fs->e2fs_fsmnt, 0, sizeof(fs->e2fs_fsmnt));
285 (void) copystr(mp->mnt_stat.f_mntonname, fs->e2fs_fsmnt,
286 sizeof(fs->e2fs_fsmnt) - 1, 0);
287 if (fs->e2fs.e2fs_rev > E2FS_REV0) {
288 memset(fs->e2fs.e2fs_fsmnt, 0, sizeof(fs->e2fs.e2fs_fsmnt));
289 (void) copystr(mp->mnt_stat.f_mntonname, fs->e2fs.e2fs_fsmnt,
290 sizeof(fs->e2fs.e2fs_fsmnt) - 1, 0);
292 (void)ext2fs_statvfs(mp, &mp->mnt_stat);
293 vfs_unbusy(mp, false, NULL);
294 setrootfstime((time_t)fs->e2fs.e2fs_wtime);
295 return (0);
299 * VFS Operations.
301 * mount system call
304 ext2fs_mount(struct mount *mp, const char *path, void *data, size_t *data_len)
306 struct lwp *l = curlwp;
307 struct vnode *devvp;
308 struct ufs_args *args = data;
309 struct ufsmount *ump = NULL;
310 struct m_ext2fs *fs;
311 size_t size;
312 int error = 0, flags, update;
313 mode_t accessmode;
315 if (*data_len < sizeof *args)
316 return EINVAL;
318 if (mp->mnt_flag & MNT_GETARGS) {
319 ump = VFSTOUFS(mp);
320 if (ump == NULL)
321 return EIO;
322 memset(args, 0, sizeof *args);
323 args->fspec = NULL;
324 *data_len = sizeof *args;
325 return 0;
328 update = mp->mnt_flag & MNT_UPDATE;
330 /* Check arguments */
331 if (args->fspec != NULL) {
333 * Look up the name and verify that it's sane.
335 error = namei_simple_user(args->fspec,
336 NSM_FOLLOW_NOEMULROOT, &devvp);
337 if (error != 0)
338 return (error);
340 if (!update) {
342 * Be sure this is a valid block device
344 if (devvp->v_type != VBLK)
345 error = ENOTBLK;
346 else if (bdevsw_lookup(devvp->v_rdev) == NULL)
347 error = ENXIO;
348 } else {
350 * Be sure we're still naming the same device
351 * used for our initial mount
353 ump = VFSTOUFS(mp);
354 if (devvp != ump->um_devvp) {
355 if (devvp->v_rdev != ump->um_devvp->v_rdev)
356 error = EINVAL;
357 else {
358 vrele(devvp);
359 devvp = ump->um_devvp;
360 vref(devvp);
364 } else {
365 if (!update) {
366 /* New mounts must have a filename for the device */
367 return (EINVAL);
368 } else {
369 ump = VFSTOUFS(mp);
370 devvp = ump->um_devvp;
371 vref(devvp);
376 * If mount by non-root, then verify that user has necessary
377 * permissions on the device.
379 * Permission to update a mount is checked higher, so here we presume
380 * updating the mount is okay (for example, as far as securelevel goes)
381 * which leaves us with the normal check.
383 if (error == 0) {
384 accessmode = VREAD;
385 if (update ?
386 (mp->mnt_iflag & IMNT_WANTRDWR) != 0 :
387 (mp->mnt_flag & MNT_RDONLY) == 0)
388 accessmode |= VWRITE;
389 vn_lock(devvp, LK_EXCLUSIVE | LK_RETRY);
390 error = kauth_authorize_system(l->l_cred, KAUTH_SYSTEM_MOUNT,
391 KAUTH_REQ_SYSTEM_MOUNT_DEVICE, mp, devvp,
392 KAUTH_ARG(accessmode));
393 VOP_UNLOCK(devvp);
396 if (error) {
397 vrele(devvp);
398 return (error);
401 if (!update) {
402 int xflags;
404 if (mp->mnt_flag & MNT_RDONLY)
405 xflags = FREAD;
406 else
407 xflags = FREAD|FWRITE;
408 vn_lock(devvp, LK_EXCLUSIVE | LK_RETRY);
409 error = VOP_OPEN(devvp, xflags, FSCRED);
410 VOP_UNLOCK(devvp);
411 if (error)
412 goto fail;
413 error = ext2fs_mountfs(devvp, mp);
414 if (error) {
415 vn_lock(devvp, LK_EXCLUSIVE | LK_RETRY);
416 (void)VOP_CLOSE(devvp, xflags, NOCRED);
417 VOP_UNLOCK(devvp);
418 goto fail;
421 ump = VFSTOUFS(mp);
422 fs = ump->um_e2fs;
423 } else {
425 * Update the mount.
429 * The initial mount got a reference on this
430 * device, so drop the one obtained via
431 * namei(), above.
433 vrele(devvp);
435 ump = VFSTOUFS(mp);
436 fs = ump->um_e2fs;
437 if (fs->e2fs_ronly == 0 && (mp->mnt_flag & MNT_RDONLY)) {
439 * Changing from r/w to r/o
441 flags = WRITECLOSE;
442 if (mp->mnt_flag & MNT_FORCE)
443 flags |= FORCECLOSE;
444 error = ext2fs_flushfiles(mp, flags);
445 if (error == 0 &&
446 ext2fs_cgupdate(ump, MNT_WAIT) == 0 &&
447 (fs->e2fs.e2fs_state & E2FS_ERRORS) == 0) {
448 fs->e2fs.e2fs_state = E2FS_ISCLEAN;
449 (void) ext2fs_sbupdate(ump, MNT_WAIT);
451 if (error)
452 return (error);
453 fs->e2fs_ronly = 1;
456 if (mp->mnt_flag & MNT_RELOAD) {
457 error = ext2fs_reload(mp, l->l_cred, l);
458 if (error)
459 return (error);
462 if (fs->e2fs_ronly && (mp->mnt_iflag & IMNT_WANTRDWR)) {
464 * Changing from read-only to read/write
466 fs->e2fs_ronly = 0;
467 if (fs->e2fs.e2fs_state == E2FS_ISCLEAN)
468 fs->e2fs.e2fs_state = 0;
469 else
470 fs->e2fs.e2fs_state = E2FS_ERRORS;
471 fs->e2fs_fmod = 1;
473 if (args->fspec == NULL)
474 return 0;
477 error = set_statvfs_info(path, UIO_USERSPACE, args->fspec,
478 UIO_USERSPACE, mp->mnt_op->vfs_name, mp, l);
479 (void) copystr(mp->mnt_stat.f_mntonname, fs->e2fs_fsmnt,
480 sizeof(fs->e2fs_fsmnt) - 1, &size);
481 memset(fs->e2fs_fsmnt + size, 0, sizeof(fs->e2fs_fsmnt) - size);
482 if (fs->e2fs.e2fs_rev > E2FS_REV0) {
483 (void) copystr(mp->mnt_stat.f_mntonname, fs->e2fs.e2fs_fsmnt,
484 sizeof(fs->e2fs.e2fs_fsmnt) - 1, &size);
485 memset(fs->e2fs.e2fs_fsmnt, 0,
486 sizeof(fs->e2fs.e2fs_fsmnt) - size);
488 if (fs->e2fs_fmod != 0) { /* XXX */
489 fs->e2fs_fmod = 0;
490 if (fs->e2fs.e2fs_state == 0)
491 fs->e2fs.e2fs_wtime = time_second;
492 else
493 printf("%s: file system not clean; please fsck(8)\n",
494 mp->mnt_stat.f_mntfromname);
495 (void) ext2fs_cgupdate(ump, MNT_WAIT);
497 return (error);
499 fail:
500 vrele(devvp);
501 return (error);
505 * Reload all incore data for a filesystem (used after running fsck on
506 * the root filesystem and finding things to fix). The filesystem must
507 * be mounted read-only.
509 * Things to do to update the mount:
510 * 1) invalidate all cached meta-data.
511 * 2) re-read superblock from disk.
512 * 3) re-read summary information from disk.
513 * 4) invalidate all inactive vnodes.
514 * 5) invalidate all cached file data.
515 * 6) re-read inode data for all active vnodes.
518 ext2fs_reload(struct mount *mp, kauth_cred_t cred, struct lwp *l)
520 struct vnode *vp, *mvp, *devvp;
521 struct inode *ip;
522 struct buf *bp;
523 struct m_ext2fs *fs;
524 struct ext2fs *newfs;
525 int i, error;
526 void *cp;
527 struct ufsmount *ump;
529 if ((mp->mnt_flag & MNT_RDONLY) == 0)
530 return (EINVAL);
532 ump = VFSTOUFS(mp);
534 * Step 1: invalidate all cached meta-data.
536 devvp = ump->um_devvp;
537 vn_lock(devvp, LK_EXCLUSIVE | LK_RETRY);
538 error = vinvalbuf(devvp, 0, cred, l, 0, 0);
539 VOP_UNLOCK(devvp);
540 if (error)
541 panic("ext2fs_reload: dirty1");
543 * Step 2: re-read superblock from disk.
545 error = bread(devvp, SBLOCK, SBSIZE, NOCRED, 0, &bp);
546 if (error) {
547 return (error);
549 newfs = (struct ext2fs *)bp->b_data;
550 error = ext2fs_checksb(newfs, (mp->mnt_flag & MNT_RDONLY) != 0);
551 if (error) {
552 brelse(bp, 0);
553 return (error);
556 fs = ump->um_e2fs;
558 * copy in new superblock, and compute in-memory values
560 e2fs_sbload(newfs, &fs->e2fs);
561 fs->e2fs_ncg =
562 howmany(fs->e2fs.e2fs_bcount - fs->e2fs.e2fs_first_dblock,
563 fs->e2fs.e2fs_bpg);
564 fs->e2fs_fsbtodb = fs->e2fs.e2fs_log_bsize + LOG_MINBSIZE - DEV_BSHIFT;
565 fs->e2fs_bsize = MINBSIZE << fs->e2fs.e2fs_log_bsize;
566 fs->e2fs_bshift = LOG_MINBSIZE + fs->e2fs.e2fs_log_bsize;
567 fs->e2fs_qbmask = fs->e2fs_bsize - 1;
568 fs->e2fs_bmask = ~fs->e2fs_qbmask;
569 fs->e2fs_ngdb =
570 howmany(fs->e2fs_ncg, fs->e2fs_bsize / sizeof(struct ext2_gd));
571 fs->e2fs_ipb = fs->e2fs_bsize / EXT2_DINODE_SIZE(fs);
572 fs->e2fs_itpg = fs->e2fs.e2fs_ipg / fs->e2fs_ipb;
573 brelse(bp, 0);
576 * Step 3: re-read summary information from disk.
579 for (i = 0; i < fs->e2fs_ngdb; i++) {
580 error = bread(devvp ,
581 EXT2_FSBTODB(fs, fs->e2fs.e2fs_first_dblock +
582 1 /* superblock */ + i),
583 fs->e2fs_bsize, NOCRED, 0, &bp);
584 if (error) {
585 return (error);
587 e2fs_cgload((struct ext2_gd *)bp->b_data,
588 &fs->e2fs_gd[i * fs->e2fs_bsize / sizeof(struct ext2_gd)],
589 fs->e2fs_bsize);
590 brelse(bp, 0);
593 /* Allocate a marker vnode. */
594 mvp = vnalloc(mp);
596 * NOTE: not using the TAILQ_FOREACH here since in this loop vgone()
597 * and vclean() can be called indirectly
599 mutex_enter(&mntvnode_lock);
600 loop:
601 for (vp = TAILQ_FIRST(&mp->mnt_vnodelist); vp; vp = vunmark(mvp)) {
602 vmark(mvp, vp);
603 if (vp->v_mount != mp || vismarker(vp))
604 continue;
606 * Step 4: invalidate all inactive vnodes.
608 if (vrecycle(vp, &mntvnode_lock)) {
609 mutex_enter(&mntvnode_lock);
610 (void)vunmark(mvp);
611 goto loop;
614 * Step 5: invalidate all cached file data.
616 mutex_enter(vp->v_interlock);
617 mutex_exit(&mntvnode_lock);
618 if (vget(vp, LK_EXCLUSIVE)) {
619 mutex_enter(&mntvnode_lock);
620 (void)vunmark(mvp);
621 goto loop;
623 if (vinvalbuf(vp, 0, cred, l, 0, 0))
624 panic("ext2fs_reload: dirty2");
626 * Step 6: re-read inode data for all active vnodes.
628 ip = VTOI(vp);
629 error = bread(devvp, EXT2_FSBTODB(fs, ino_to_fsba(fs, ip->i_number)),
630 (int)fs->e2fs_bsize, NOCRED, 0, &bp);
631 if (error) {
632 vput(vp);
633 mutex_enter(&mntvnode_lock);
634 (void)vunmark(mvp);
635 break;
637 cp = (char *)bp->b_data +
638 (ino_to_fsbo(fs, ip->i_number) * EXT2_DINODE_SIZE(fs));
639 e2fs_iload((struct ext2fs_dinode *)cp, ip->i_din.e2fs_din);
640 ext2fs_set_inode_guid(ip);
641 brelse(bp, 0);
642 vput(vp);
643 mutex_enter(&mntvnode_lock);
645 mutex_exit(&mntvnode_lock);
646 vnfree(mvp);
647 return (error);
651 * Common code for mount and mountroot
654 ext2fs_mountfs(struct vnode *devvp, struct mount *mp)
656 struct lwp *l = curlwp;
657 struct ufsmount *ump;
658 struct buf *bp;
659 struct ext2fs *fs;
660 struct m_ext2fs *m_fs;
661 dev_t dev;
662 int error, i, ronly;
663 kauth_cred_t cred;
665 dev = devvp->v_rdev;
666 cred = l ? l->l_cred : NOCRED;
668 /* Flush out any old buffers remaining from a previous use. */
669 vn_lock(devvp, LK_EXCLUSIVE | LK_RETRY);
670 error = vinvalbuf(devvp, V_SAVE, cred, l, 0, 0);
671 VOP_UNLOCK(devvp);
672 if (error)
673 return (error);
675 ronly = (mp->mnt_flag & MNT_RDONLY) != 0;
677 bp = NULL;
678 ump = NULL;
680 #ifdef DEBUG_EXT2
681 printf("ext2 sb size: %zu\n", sizeof(struct ext2fs));
682 #endif
683 error = bread(devvp, SBLOCK, SBSIZE, cred, 0, &bp);
684 if (error)
685 goto out;
686 fs = (struct ext2fs *)bp->b_data;
687 error = ext2fs_checksb(fs, ronly);
688 if (error)
689 goto out;
690 ump = kmem_zalloc(sizeof(*ump), KM_SLEEP);
691 ump->um_fstype = UFS1;
692 ump->um_ops = &ext2fs_ufsops;
693 ump->um_e2fs = kmem_zalloc(sizeof(struct m_ext2fs), KM_SLEEP);
694 e2fs_sbload((struct ext2fs *)bp->b_data, &ump->um_e2fs->e2fs);
695 brelse(bp, 0);
696 bp = NULL;
697 m_fs = ump->um_e2fs;
698 m_fs->e2fs_ronly = ronly;
700 #ifdef DEBUG_EXT2
701 printf("ext2 ino size %zu\n", EXT2_DINODE_SIZE(m_fs));
702 #endif
703 if (ronly == 0) {
704 if (m_fs->e2fs.e2fs_state == E2FS_ISCLEAN)
705 m_fs->e2fs.e2fs_state = 0;
706 else
707 m_fs->e2fs.e2fs_state = E2FS_ERRORS;
708 m_fs->e2fs_fmod = 1;
711 /* compute dynamic sb infos */
712 m_fs->e2fs_ncg =
713 howmany(m_fs->e2fs.e2fs_bcount - m_fs->e2fs.e2fs_first_dblock,
714 m_fs->e2fs.e2fs_bpg);
715 m_fs->e2fs_fsbtodb = m_fs->e2fs.e2fs_log_bsize + LOG_MINBSIZE - DEV_BSHIFT;
716 m_fs->e2fs_bsize = MINBSIZE << m_fs->e2fs.e2fs_log_bsize;
717 m_fs->e2fs_bshift = LOG_MINBSIZE + m_fs->e2fs.e2fs_log_bsize;
718 m_fs->e2fs_qbmask = m_fs->e2fs_bsize - 1;
719 m_fs->e2fs_bmask = ~m_fs->e2fs_qbmask;
720 m_fs->e2fs_ngdb =
721 howmany(m_fs->e2fs_ncg, m_fs->e2fs_bsize / sizeof(struct ext2_gd));
722 m_fs->e2fs_ipb = m_fs->e2fs_bsize / EXT2_DINODE_SIZE(m_fs);
723 m_fs->e2fs_itpg = m_fs->e2fs.e2fs_ipg / m_fs->e2fs_ipb;
725 m_fs->e2fs_gd = kmem_alloc(m_fs->e2fs_ngdb * m_fs->e2fs_bsize, KM_SLEEP);
726 for (i = 0; i < m_fs->e2fs_ngdb; i++) {
727 error = bread(devvp ,
728 EXT2_FSBTODB(m_fs, m_fs->e2fs.e2fs_first_dblock +
729 1 /* superblock */ + i),
730 m_fs->e2fs_bsize, NOCRED, 0, &bp);
731 if (error) {
732 kmem_free(m_fs->e2fs_gd,
733 m_fs->e2fs_ngdb * m_fs->e2fs_bsize);
734 goto out;
736 e2fs_cgload((struct ext2_gd *)bp->b_data,
737 &m_fs->e2fs_gd[
738 i * m_fs->e2fs_bsize / sizeof(struct ext2_gd)],
739 m_fs->e2fs_bsize);
740 brelse(bp, 0);
741 bp = NULL;
744 mp->mnt_data = ump;
745 mp->mnt_stat.f_fsidx.__fsid_val[0] = (long)dev;
746 mp->mnt_stat.f_fsidx.__fsid_val[1] = makefstype(MOUNT_EXT2FS);
747 mp->mnt_stat.f_fsid = mp->mnt_stat.f_fsidx.__fsid_val[0];
748 mp->mnt_stat.f_namemax = EXT2FS_MAXNAMLEN;
749 mp->mnt_flag |= MNT_LOCAL;
750 mp->mnt_dev_bshift = DEV_BSHIFT; /* XXX */
751 mp->mnt_fs_bshift = m_fs->e2fs_bshift;
752 mp->mnt_iflag |= IMNT_DTYPE;
753 ump->um_flags = 0;
754 ump->um_mountp = mp;
755 ump->um_dev = dev;
756 ump->um_devvp = devvp;
757 ump->um_nindir = EXT2_NINDIR(m_fs);
758 ump->um_lognindir = ffs(EXT2_NINDIR(m_fs)) - 1;
759 ump->um_bptrtodb = m_fs->e2fs_fsbtodb;
760 ump->um_seqinc = 1; /* no frags */
761 ump->um_maxsymlinklen = EXT2_MAXSYMLINKLEN;
762 ump->um_dirblksiz = m_fs->e2fs_bsize;
763 ump->um_maxfilesize = ((uint64_t)0x80000000 * m_fs->e2fs_bsize - 1);
764 spec_node_setmountedfs(devvp, mp);
765 return (0);
767 out:
768 if (bp != NULL)
769 brelse(bp, 0);
770 if (ump) {
771 kmem_free(ump->um_e2fs, sizeof(struct m_ext2fs));
772 kmem_free(ump, sizeof(*ump));
773 mp->mnt_data = NULL;
775 return (error);
779 * unmount system call
782 ext2fs_unmount(struct mount *mp, int mntflags)
784 struct ufsmount *ump;
785 struct m_ext2fs *fs;
786 int error, flags;
788 flags = 0;
789 if (mntflags & MNT_FORCE)
790 flags |= FORCECLOSE;
791 if ((error = ext2fs_flushfiles(mp, flags)) != 0)
792 return (error);
793 ump = VFSTOUFS(mp);
794 fs = ump->um_e2fs;
795 if (fs->e2fs_ronly == 0 &&
796 ext2fs_cgupdate(ump, MNT_WAIT) == 0 &&
797 (fs->e2fs.e2fs_state & E2FS_ERRORS) == 0) {
798 fs->e2fs.e2fs_state = E2FS_ISCLEAN;
799 (void) ext2fs_sbupdate(ump, MNT_WAIT);
801 if (ump->um_devvp->v_type != VBAD)
802 spec_node_setmountedfs(ump->um_devvp, NULL);
803 vn_lock(ump->um_devvp, LK_EXCLUSIVE | LK_RETRY);
804 error = VOP_CLOSE(ump->um_devvp, fs->e2fs_ronly ? FREAD : FREAD|FWRITE,
805 NOCRED);
806 vput(ump->um_devvp);
807 kmem_free(fs->e2fs_gd, fs->e2fs_ngdb * fs->e2fs_bsize);
808 kmem_free(fs, sizeof(*fs));
809 kmem_free(ump, sizeof(*ump));
810 mp->mnt_data = NULL;
811 mp->mnt_flag &= ~MNT_LOCAL;
812 return (error);
816 * Flush out all the files in a filesystem.
819 ext2fs_flushfiles(struct mount *mp, int flags)
821 extern int doforce;
822 int error;
824 if (!doforce)
825 flags &= ~FORCECLOSE;
826 error = vflush(mp, NULLVP, flags);
827 return (error);
831 * Get file system statistics.
834 ext2fs_statvfs(struct mount *mp, struct statvfs *sbp)
836 struct ufsmount *ump;
837 struct m_ext2fs *fs;
838 uint32_t overhead, overhead_per_group, ngdb;
839 int i, ngroups;
841 ump = VFSTOUFS(mp);
842 fs = ump->um_e2fs;
843 if (fs->e2fs.e2fs_magic != E2FS_MAGIC)
844 panic("ext2fs_statvfs");
847 * Compute the overhead (FS structures)
849 overhead_per_group =
850 1 /* block bitmap */ +
851 1 /* inode bitmap */ +
852 fs->e2fs_itpg;
853 overhead = fs->e2fs.e2fs_first_dblock +
854 fs->e2fs_ncg * overhead_per_group;
855 if (fs->e2fs.e2fs_rev > E2FS_REV0 &&
856 fs->e2fs.e2fs_features_rocompat & EXT2F_ROCOMPAT_SPARSESUPER) {
857 for (i = 0, ngroups = 0; i < fs->e2fs_ncg; i++) {
858 if (cg_has_sb(i))
859 ngroups++;
861 } else {
862 ngroups = fs->e2fs_ncg;
864 ngdb = fs->e2fs_ngdb;
865 if (fs->e2fs.e2fs_rev > E2FS_REV0 &&
866 fs->e2fs.e2fs_features_compat & EXT2F_COMPAT_RESIZE)
867 ngdb += fs->e2fs.e2fs_reserved_ngdb;
868 overhead += ngroups * (1 /* superblock */ + ngdb);
870 sbp->f_bsize = fs->e2fs_bsize;
871 sbp->f_frsize = MINBSIZE << fs->e2fs.e2fs_fsize;
872 sbp->f_iosize = fs->e2fs_bsize;
873 sbp->f_blocks = fs->e2fs.e2fs_bcount - overhead;
874 sbp->f_bfree = fs->e2fs.e2fs_fbcount;
875 sbp->f_bresvd = fs->e2fs.e2fs_rbcount;
876 if (sbp->f_bfree > sbp->f_bresvd)
877 sbp->f_bavail = sbp->f_bfree - sbp->f_bresvd;
878 else
879 sbp->f_bavail = 0;
880 sbp->f_files = fs->e2fs.e2fs_icount;
881 sbp->f_ffree = fs->e2fs.e2fs_ficount;
882 sbp->f_favail = fs->e2fs.e2fs_ficount;
883 sbp->f_fresvd = 0;
884 copy_statvfs_info(sbp, mp);
885 return (0);
889 * Go through the disk queues to initiate sandbagged IO;
890 * go through the inodes to write those that have been modified;
891 * initiate the writing of the super block if it has been modified.
893 * Note: we are always called with the filesystem marked `MPBUSY'.
896 ext2fs_sync(struct mount *mp, int waitfor, kauth_cred_t cred)
898 struct vnode *vp, *mvp;
899 struct inode *ip;
900 struct ufsmount *ump = VFSTOUFS(mp);
901 struct m_ext2fs *fs;
902 int error, allerror = 0;
904 fs = ump->um_e2fs;
905 if (fs->e2fs_fmod != 0 && fs->e2fs_ronly != 0) { /* XXX */
906 printf("fs = %s\n", fs->e2fs_fsmnt);
907 panic("update: rofs mod");
910 /* Allocate a marker vnode. */
911 mvp = vnalloc(mp);
914 * Write back each (modified) inode.
916 mutex_enter(&mntvnode_lock);
917 loop:
919 * NOTE: not using the TAILQ_FOREACH here since in this loop vgone()
920 * and vclean() can be called indirectly
922 for (vp = TAILQ_FIRST(&mp->mnt_vnodelist); vp; vp = vunmark(mvp)) {
923 vmark(mvp, vp);
924 if (vp->v_mount != mp || vismarker(vp))
925 continue;
926 mutex_enter(vp->v_interlock);
927 ip = VTOI(vp);
928 if (ip == NULL || (vp->v_iflag & (VI_XLOCK|VI_CLEAN)) != 0 ||
929 vp->v_type == VNON ||
930 ((ip->i_flag &
931 (IN_CHANGE | IN_UPDATE | IN_MODIFIED)) == 0 &&
932 LIST_EMPTY(&vp->v_dirtyblkhd) &&
933 UVM_OBJ_IS_CLEAN(&vp->v_uobj)))
935 mutex_exit(vp->v_interlock);
936 continue;
938 mutex_exit(&mntvnode_lock);
939 error = vget(vp, LK_EXCLUSIVE | LK_NOWAIT);
940 if (error) {
941 mutex_enter(&mntvnode_lock);
942 if (error == ENOENT) {
943 mutex_enter(&mntvnode_lock);
944 (void)vunmark(mvp);
945 goto loop;
947 continue;
949 if (vp->v_type == VREG && waitfor == MNT_LAZY)
950 error = ext2fs_update(vp, NULL, NULL, 0);
951 else
952 error = VOP_FSYNC(vp, cred,
953 waitfor == MNT_WAIT ? FSYNC_WAIT : 0, 0, 0);
954 if (error)
955 allerror = error;
956 vput(vp);
957 mutex_enter(&mntvnode_lock);
959 mutex_exit(&mntvnode_lock);
960 vnfree(mvp);
962 * Force stale file system control information to be flushed.
964 if (waitfor != MNT_LAZY) {
965 vn_lock(ump->um_devvp, LK_EXCLUSIVE | LK_RETRY);
966 if ((error = VOP_FSYNC(ump->um_devvp, cred,
967 waitfor == MNT_WAIT ? FSYNC_WAIT : 0, 0, 0)) != 0)
968 allerror = error;
969 VOP_UNLOCK(ump->um_devvp);
972 * Write back modified superblock.
974 if (fs->e2fs_fmod != 0) {
975 fs->e2fs_fmod = 0;
976 fs->e2fs.e2fs_wtime = time_second;
977 if ((error = ext2fs_cgupdate(ump, waitfor)))
978 allerror = error;
980 return (allerror);
984 * Look up a EXT2FS dinode number to find its incore vnode, otherwise read it
985 * in from disk. If it is in core, wait for the lock bit to clear, then
986 * return the inode locked. Detection and handling of mount points must be
987 * done by the calling routine.
990 ext2fs_vget(struct mount *mp, ino_t ino, struct vnode **vpp)
992 struct m_ext2fs *fs;
993 struct inode *ip;
994 struct ufsmount *ump;
995 struct buf *bp;
996 struct vnode *vp;
997 dev_t dev;
998 int error;
999 void *cp;
1001 ump = VFSTOUFS(mp);
1002 dev = ump->um_dev;
1003 retry:
1004 if ((*vpp = ufs_ihashget(dev, ino, LK_EXCLUSIVE)) != NULL)
1005 return (0);
1007 /* Allocate a new vnode/inode. */
1008 error = getnewvnode(VT_EXT2FS, mp, ext2fs_vnodeop_p, NULL, &vp);
1009 if (error) {
1010 *vpp = NULL;
1011 return (error);
1013 ip = pool_get(&ext2fs_inode_pool, PR_WAITOK);
1015 mutex_enter(&ufs_hashlock);
1016 if ((*vpp = ufs_ihashget(dev, ino, 0)) != NULL) {
1017 mutex_exit(&ufs_hashlock);
1018 ungetnewvnode(vp);
1019 pool_put(&ext2fs_inode_pool, ip);
1020 goto retry;
1023 vp->v_vflag |= VV_LOCKSWORK;
1025 memset(ip, 0, sizeof(struct inode));
1026 vp->v_data = ip;
1027 ip->i_vnode = vp;
1028 ip->i_ump = ump;
1029 ip->i_e2fs = fs = ump->um_e2fs;
1030 ip->i_dev = dev;
1031 ip->i_number = ino;
1032 ip->i_e2fs_last_lblk = 0;
1033 ip->i_e2fs_last_blk = 0;
1034 genfs_node_init(vp, &ext2fs_genfsops);
1037 * Put it onto its hash chain and lock it so that other requests for
1038 * this inode will block if they arrive while we are sleeping waiting
1039 * for old data structures to be purged or for the contents of the
1040 * disk portion of this inode to be read.
1043 ufs_ihashins(ip);
1044 mutex_exit(&ufs_hashlock);
1046 /* Read in the disk contents for the inode, copy into the inode. */
1047 error = bread(ump->um_devvp, EXT2_FSBTODB(fs, ino_to_fsba(fs, ino)),
1048 (int)fs->e2fs_bsize, NOCRED, 0, &bp);
1049 if (error) {
1052 * The inode does not contain anything useful, so it would
1053 * be misleading to leave it on its hash chain. With mode
1054 * still zero, it will be unlinked and returned to the free
1055 * list by vput().
1058 vput(vp);
1059 *vpp = NULL;
1060 return (error);
1062 cp = (char *)bp->b_data + (ino_to_fsbo(fs, ino) * EXT2_DINODE_SIZE(fs));
1063 ip->i_din.e2fs_din = pool_get(&ext2fs_dinode_pool, PR_WAITOK);
1064 e2fs_iload((struct ext2fs_dinode *)cp, ip->i_din.e2fs_din);
1065 ext2fs_set_inode_guid(ip);
1066 brelse(bp, 0);
1068 /* If the inode was deleted, reset all fields */
1069 if (ip->i_e2fs_dtime != 0) {
1070 ip->i_e2fs_mode = 0;
1071 (void)ext2fs_setsize(ip, 0);
1072 (void)ext2fs_setnblock(ip, 0);
1073 memset(ip->i_e2fs_blocks, 0, sizeof(ip->i_e2fs_blocks));
1077 * Initialize the vnode from the inode, check for aliases.
1080 error = ext2fs_vinit(mp, ext2fs_specop_p, ext2fs_fifoop_p, &vp);
1081 if (error) {
1082 vput(vp);
1083 *vpp = NULL;
1084 return (error);
1087 * Finish inode initialization now that aliasing has been resolved.
1090 ip->i_devvp = ump->um_devvp;
1091 vref(ip->i_devvp);
1094 * Set up a generation number for this inode if it does not
1095 * already have one. This should only happen on old filesystems.
1098 if (ip->i_e2fs_gen == 0) {
1099 if (++ext2gennumber < (u_long)time_second)
1100 ext2gennumber = time_second;
1101 ip->i_e2fs_gen = ext2gennumber;
1102 if ((vp->v_mount->mnt_flag & MNT_RDONLY) == 0)
1103 ip->i_flag |= IN_MODIFIED;
1105 uvm_vnp_setsize(vp, ext2fs_size(ip));
1106 *vpp = vp;
1107 return (0);
1111 * File handle to vnode
1113 * Have to be really careful about stale file handles:
1114 * - check that the inode number is valid
1115 * - call ext2fs_vget() to get the locked inode
1116 * - check for an unallocated inode (i_mode == 0)
1119 ext2fs_fhtovp(struct mount *mp, struct fid *fhp, struct vnode **vpp)
1121 struct inode *ip;
1122 struct vnode *nvp;
1123 int error;
1124 struct ufid ufh;
1125 struct m_ext2fs *fs;
1127 if (fhp->fid_len != sizeof(struct ufid))
1128 return EINVAL;
1130 memcpy(&ufh, fhp, sizeof(struct ufid));
1131 fs = VFSTOUFS(mp)->um_e2fs;
1132 if ((ufh.ufid_ino < EXT2_FIRSTINO && ufh.ufid_ino != EXT2_ROOTINO) ||
1133 ufh.ufid_ino >= fs->e2fs_ncg * fs->e2fs.e2fs_ipg)
1134 return (ESTALE);
1136 if ((error = VFS_VGET(mp, ufh.ufid_ino, &nvp)) != 0) {
1137 *vpp = NULLVP;
1138 return (error);
1140 ip = VTOI(nvp);
1141 if (ip->i_e2fs_mode == 0 || ip->i_e2fs_dtime != 0 ||
1142 ip->i_e2fs_gen != ufh.ufid_gen) {
1143 vput(nvp);
1144 *vpp = NULLVP;
1145 return (ESTALE);
1147 *vpp = nvp;
1148 return (0);
1152 * Vnode pointer to File handle
1154 /* ARGSUSED */
1156 ext2fs_vptofh(struct vnode *vp, struct fid *fhp, size_t *fh_size)
1158 struct inode *ip;
1159 struct ufid ufh;
1161 if (*fh_size < sizeof(struct ufid)) {
1162 *fh_size = sizeof(struct ufid);
1163 return E2BIG;
1165 *fh_size = sizeof(struct ufid);
1167 ip = VTOI(vp);
1168 memset(&ufh, 0, sizeof(ufh));
1169 ufh.ufid_len = sizeof(struct ufid);
1170 ufh.ufid_ino = ip->i_number;
1171 ufh.ufid_gen = ip->i_e2fs_gen;
1172 memcpy(fhp, &ufh, sizeof(ufh));
1173 return (0);
1177 * Write a superblock and associated information back to disk.
1180 ext2fs_sbupdate(struct ufsmount *mp, int waitfor)
1182 struct m_ext2fs *fs = mp->um_e2fs;
1183 struct buf *bp;
1184 int error = 0;
1186 bp = getblk(mp->um_devvp, SBLOCK, SBSIZE, 0, 0);
1187 e2fs_sbsave(&fs->e2fs, (struct ext2fs*)bp->b_data);
1188 if (waitfor == MNT_WAIT)
1189 error = bwrite(bp);
1190 else
1191 bawrite(bp);
1192 return (error);
1196 ext2fs_cgupdate(struct ufsmount *mp, int waitfor)
1198 struct m_ext2fs *fs = mp->um_e2fs;
1199 struct buf *bp;
1200 int i, error = 0, allerror = 0;
1202 allerror = ext2fs_sbupdate(mp, waitfor);
1203 for (i = 0; i < fs->e2fs_ngdb; i++) {
1204 bp = getblk(mp->um_devvp, EXT2_FSBTODB(fs,
1205 fs->e2fs.e2fs_first_dblock +
1206 1 /* superblock */ + i), fs->e2fs_bsize, 0, 0);
1207 e2fs_cgsave(&fs->e2fs_gd[
1208 i * fs->e2fs_bsize / sizeof(struct ext2_gd)],
1209 (struct ext2_gd *)bp->b_data, fs->e2fs_bsize);
1210 if (waitfor == MNT_WAIT)
1211 error = bwrite(bp);
1212 else
1213 bawrite(bp);
1216 if (!allerror && error)
1217 allerror = error;
1218 return (allerror);
1221 static int
1222 ext2fs_checksb(struct ext2fs *fs, int ronly)
1224 uint32_t u32;
1226 if (fs2h16(fs->e2fs_magic) != E2FS_MAGIC) {
1227 return (EINVAL); /* XXX needs translation */
1229 if (fs2h32(fs->e2fs_rev) > E2FS_REV1) {
1230 #ifdef DIAGNOSTIC
1231 printf("ext2fs: unsupported revision number: %x\n",
1232 fs2h32(fs->e2fs_rev));
1233 #endif
1234 return (EINVAL); /* XXX needs translation */
1236 if (fs2h32(fs->e2fs_log_bsize) > 2) { /* block size = 1024|2048|4096 */
1237 #ifdef DIAGNOSTIC
1238 printf("ext2fs: bad block size: %d "
1239 "(expected <= 2 for ext2 fs)\n",
1240 fs2h32(fs->e2fs_log_bsize));
1241 #endif
1242 return (EINVAL); /* XXX needs translation */
1244 if (fs2h32(fs->e2fs_rev) > E2FS_REV0) {
1245 char buf[256];
1246 if (fs2h32(fs->e2fs_first_ino) != EXT2_FIRSTINO) {
1247 printf("ext2fs: unsupported first inode position\n");
1248 return (EINVAL); /* XXX needs translation */
1250 u32 = fs2h32(fs->e2fs_features_incompat) & ~EXT2F_INCOMPAT_SUPP;
1251 if (u32) {
1252 snprintb(buf, sizeof(buf), EXT2F_INCOMPAT_BITS, u32);
1253 printf("ext2fs: unsupported incompat features: %s\n",
1254 buf);
1255 return EINVAL; /* XXX needs translation */
1257 u32 = fs2h32(fs->e2fs_features_rocompat) & ~EXT2F_ROCOMPAT_SUPP;
1258 if (!ronly && u32) {
1259 snprintb(buf, sizeof(buf), EXT2F_ROCOMPAT_BITS, u32);
1260 printf("ext2fs: unsupported ro-incompat features: %s\n",
1261 buf);
1262 return EROFS; /* XXX needs translation */
1265 return (0);