lfs_vfsops.c revision 1.8 1 /* $NetBSD: lfs_vfsops.c,v 1.8 1995/03/09 12:06:10 mycroft 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. All advertising materials mentioning features or use of this software
16 * must display the following acknowledgement:
17 * This product includes software developed by the University of
18 * California, Berkeley and its contributors.
19 * 4. Neither the name of the University nor the names of its contributors
20 * may be used to endorse or promote products derived from this software
21 * without specific prior written permission.
22 *
23 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
24 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
25 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
26 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
27 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
28 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
29 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
30 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
31 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
32 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
33 * SUCH DAMAGE.
34 *
35 * @(#)lfs_vfsops.c 8.10 (Berkeley) 11/21/94
36 */
37
38 #include <sys/param.h>
39 #include <sys/systm.h>
40 #include <sys/namei.h>
41 #include <sys/proc.h>
42 #include <sys/kernel.h>
43 #include <sys/vnode.h>
44 #include <sys/mount.h>
45 #include <sys/buf.h>
46 #include <sys/mbuf.h>
47 #include <sys/file.h>
48 #include <sys/disklabel.h>
49 #include <sys/ioctl.h>
50 #include <sys/errno.h>
51 #include <sys/malloc.h>
52 #include <sys/socket.h>
53
54 #include <miscfs/specfs/specdev.h>
55
56 #include <ufs/ufs/quota.h>
57 #include <ufs/ufs/inode.h>
58 #include <ufs/ufs/ufsmount.h>
59 #include <ufs/ufs/ufs_extern.h>
60
61 #include <ufs/lfs/lfs.h>
62 #include <ufs/lfs/lfs_extern.h>
63
64 int lfs_mountfs __P((struct vnode *, struct mount *, struct proc *));
65
66 struct vfsops lfs_vfsops = {
67 MOUNT_LFS,
68 lfs_mount,
69 ufs_start,
70 lfs_unmount,
71 ufs_root,
72 ufs_quotactl,
73 lfs_statfs,
74 lfs_sync,
75 lfs_vget,
76 lfs_fhtovp,
77 lfs_vptofh,
78 lfs_init,
79 };
80
81 int
82 lfs_mountroot()
83 {
84 panic("lfs_mountroot"); /* XXX -- implement */
85 }
86
87 /*
88 * VFS Operations.
89 *
90 * mount system call
91 */
92 lfs_mount(mp, path, data, ndp, p)
93 register struct mount *mp;
94 char *path;
95 caddr_t data;
96 struct nameidata *ndp;
97 struct proc *p;
98 {
99 struct vnode *devvp;
100 struct ufs_args args;
101 struct ufsmount *ump;
102 register struct lfs *fs; /* LFS */
103 size_t size;
104 int error;
105 mode_t accessmode;
106
107 if (error = copyin(data, (caddr_t)&args, sizeof (struct ufs_args)))
108 return (error);
109
110 /* Until LFS can do NFS right. XXX */
111 if (args.export.ex_flags & MNT_EXPORTED)
112 return (EINVAL);
113
114 /*
115 * If updating, check whether changing from read-only to
116 * read/write; if there is no device name, that's all we do.
117 */
118 if (mp->mnt_flag & MNT_UPDATE) {
119 ump = VFSTOUFS(mp);
120 if (fs->lfs_ronly && (mp->mnt_flag & MNT_WANTRDWR)) {
121 /*
122 * If upgrade to read-write by non-root, then verify
123 * that user has necessary permissions on the device.
124 */
125 if (p->p_ucred->cr_uid != 0) {
126 VOP_LOCK(ump->um_devvp);
127 if (error = VOP_ACCESS(ump->um_devvp,
128 VREAD | VWRITE, p->p_ucred, p)) {
129 VOP_UNLOCK(ump->um_devvp);
130 return (error);
131 }
132 VOP_UNLOCK(ump->um_devvp);
133 }
134 fs->lfs_ronly = 0;
135 }
136 if (args.fspec == 0) {
137 /*
138 * Process export requests.
139 */
140 return (vfs_export(mp, &ump->um_export, &args.export));
141 }
142 }
143 /*
144 * Not an update, or updating the name: look up the name
145 * and verify that it refers to a sensible block device.
146 */
147 NDINIT(ndp, LOOKUP, FOLLOW, UIO_USERSPACE, args.fspec, p);
148 if (error = namei(ndp))
149 return (error);
150 devvp = ndp->ni_vp;
151 if (devvp->v_type != VBLK) {
152 vrele(devvp);
153 return (ENOTBLK);
154 }
155 if (major(devvp->v_rdev) >= nblkdev) {
156 vrele(devvp);
157 return (ENXIO);
158 }
159 /*
160 * If mount by non-root, then verify that user has necessary
161 * permissions on the device.
162 */
163 if (p->p_ucred->cr_uid != 0) {
164 accessmode = VREAD;
165 if ((mp->mnt_flag & MNT_RDONLY) == 0)
166 accessmode |= VWRITE;
167 VOP_LOCK(devvp);
168 if (error = VOP_ACCESS(devvp, accessmode, p->p_ucred, p)) {
169 vput(devvp);
170 return (error);
171 }
172 VOP_UNLOCK(devvp);
173 }
174 if ((mp->mnt_flag & MNT_UPDATE) == 0)
175 error = lfs_mountfs(devvp, mp, p); /* LFS */
176 else {
177 if (devvp != ump->um_devvp)
178 error = EINVAL; /* needs translation */
179 else
180 vrele(devvp);
181 }
182 if (error) {
183 vrele(devvp);
184 return (error);
185 }
186 ump = VFSTOUFS(mp);
187 fs = ump->um_lfs; /* LFS */
188 #ifdef NOTLFS /* LFS */
189 (void) copyinstr(path, fs->fs_fsmnt, sizeof(fs->fs_fsmnt) - 1, &size);
190 bzero(fs->fs_fsmnt + size, sizeof(fs->fs_fsmnt) - size);
191 bcopy(fs->fs_fsmnt, mp->mnt_stat.f_mntonname, MNAMELEN);
192 #else
193 (void)copyinstr(path, fs->lfs_fsmnt, sizeof(fs->lfs_fsmnt) - 1, &size);
194 bzero(fs->lfs_fsmnt + size, sizeof(fs->lfs_fsmnt) - size);
195 bcopy(fs->lfs_fsmnt, mp->mnt_stat.f_mntonname, MNAMELEN);
196 #endif
197 (void) copyinstr(args.fspec, mp->mnt_stat.f_mntfromname, MNAMELEN - 1,
198 &size);
199 bzero(mp->mnt_stat.f_mntfromname + size, MNAMELEN - size);
200 return (0);
201 }
202
203 /*
204 * Common code for mount and mountroot
205 * LFS specific
206 */
207 int
208 lfs_mountfs(devvp, mp, p)
209 register struct vnode *devvp;
210 struct mount *mp;
211 struct proc *p;
212 {
213 extern struct vnode *rootvp;
214 register struct lfs *fs;
215 register struct ufsmount *ump;
216 struct vnode *vp;
217 struct buf *bp;
218 struct partinfo dpart;
219 dev_t dev;
220 int error, i, ronly, size;
221 struct ucred *cred;
222
223 cred = p ? p->p_ucred : NOCRED;
224 /*
225 * Disallow multiple mounts of the same device.
226 * Disallow mounting of a device that is currently in use
227 * (except for root, which might share swap device for miniroot).
228 * Flush out any old buffers remaining from a previous use.
229 */
230 if (error = vfs_mountedon(devvp))
231 return (error);
232 if (vcount(devvp) > 1 && devvp != rootvp)
233 return (EBUSY);
234 if (error = vinvalbuf(devvp, V_SAVE, cred, p, 0, 0))
235 return (error);
236
237 ronly = (mp->mnt_flag & MNT_RDONLY) != 0;
238 if (error = VOP_OPEN(devvp, ronly ? FREAD : FREAD|FWRITE, FSCRED, p))
239 return (error);
240
241 if (VOP_IOCTL(devvp, DIOCGPART, (caddr_t)&dpart, FREAD, cred, p) != 0)
242 size = DEV_BSIZE;
243 else {
244 size = dpart.disklab->d_secsize;
245 #ifdef NEVER_USED
246 dpart.part->p_fstype = FS_LFS;
247 dpart.part->p_fsize = fs->lfs_fsize; /* frag size */
248 dpart.part->p_frag = fs->lfs_frag; /* frags per block */
249 dpart.part->p_cpg = fs->lfs_segshift; /* segment shift */
250 #endif
251 }
252
253 /* Don't free random space on error. */
254 bp = NULL;
255 ump = NULL;
256
257 /* Read in the superblock. */
258 if (error = bread(devvp, LFS_LABELPAD / size, LFS_SBPAD, cred, &bp))
259 goto out;
260 fs = (struct lfs *)bp->b_data;
261
262 /* Check the basics. */
263 if (fs->lfs_magic != LFS_MAGIC || fs->lfs_bsize > MAXBSIZE ||
264 fs->lfs_bsize < sizeof(struct lfs)) {
265 error = EINVAL; /* XXX needs translation */
266 goto out;
267 }
268
269 /* Allocate the mount structure, copy the superblock into it. */
270 ump = (struct ufsmount *)malloc(sizeof *ump, M_UFSMNT, M_WAITOK);
271 fs = ump->um_lfs = malloc(sizeof(struct lfs), M_UFSMNT, M_WAITOK);
272 bcopy(bp->b_data, fs, sizeof(struct lfs));
273 if (sizeof(struct lfs) < LFS_SBPAD) /* XXX why? */
274 bp->b_flags |= B_INVAL;
275 brelse(bp);
276 bp = NULL;
277
278 /* Set up the I/O information */
279 fs->lfs_iocount = 0;
280
281 /* Set up the ifile and lock aflags */
282 fs->lfs_doifile = 0;
283 fs->lfs_writer = 0;
284 fs->lfs_dirops = 0;
285 fs->lfs_seglock = 0;
286
287 /* Set the file system readonly/modify bits. */
288 fs->lfs_ronly = ronly;
289 if (ronly == 0)
290 fs->lfs_fmod = 1;
291
292 /* Initialize the mount structure. */
293 dev = devvp->v_rdev;
294 mp->mnt_data = (qaddr_t)ump;
295 mp->mnt_stat.f_fsid.val[0] = (long)dev;
296 mp->mnt_stat.f_fsid.val[1] = makefstype(MOUNT_LFS);
297 mp->mnt_maxsymlinklen = fs->lfs_maxsymlinklen;
298 mp->mnt_flag |= MNT_LOCAL;
299 ump->um_mountp = mp;
300 ump->um_dev = dev;
301 ump->um_devvp = devvp;
302 ump->um_bptrtodb = 0;
303 ump->um_seqinc = 1 << fs->lfs_fsbtodb;
304 ump->um_nindir = fs->lfs_nindir;
305 for (i = 0; i < MAXQUOTAS; i++)
306 ump->um_quotas[i] = NULLVP;
307 devvp->v_specflags |= SI_MOUNTEDON;
308
309 /*
310 * We use the ifile vnode for almost every operation. Instead of
311 * retrieving it from the hash table each time we retrieve it here,
312 * artificially increment the reference count and keep a pointer
313 * to it in the incore copy of the superblock.
314 */
315 if (error = VFS_VGET(mp, LFS_IFILE_INUM, &vp))
316 goto out;
317 fs->lfs_ivnode = vp;
318 VREF(vp);
319 vput(vp);
320
321 return (0);
322 out:
323 if (bp)
324 brelse(bp);
325 (void)VOP_CLOSE(devvp, ronly ? FREAD : FREAD|FWRITE, cred, p);
326 if (ump) {
327 free(ump->um_lfs, M_UFSMNT);
328 free(ump, M_UFSMNT);
329 mp->mnt_data = (qaddr_t)0;
330 }
331 return (error);
332 }
333
334 /*
335 * unmount system call
336 */
337 lfs_unmount(mp, mntflags, p)
338 struct mount *mp;
339 int mntflags;
340 struct proc *p;
341 {
342 extern int doforce;
343 register struct ufsmount *ump;
344 register struct lfs *fs;
345 int i, error, flags, ronly;
346
347 flags = 0;
348 if (mntflags & MNT_FORCE)
349 flags |= FORCECLOSE;
350
351 ump = VFSTOUFS(mp);
352 fs = ump->um_lfs;
353 #ifdef QUOTA
354 if (mp->mnt_flag & MNT_QUOTA) {
355 if (error = vflush(mp, fs->lfs_ivnode, SKIPSYSTEM|flags))
356 return (error);
357 for (i = 0; i < MAXQUOTAS; i++) {
358 if (ump->um_quotas[i] == NULLVP)
359 continue;
360 quotaoff(p, mp, i);
361 }
362 /*
363 * Here we fall through to vflush again to ensure
364 * that we have gotten rid of all the system vnodes.
365 */
366 }
367 #endif
368 if (error = vflush(mp, fs->lfs_ivnode, flags))
369 return (error);
370 fs->lfs_clean = 1;
371 if (error = VFS_SYNC(mp, 1, p->p_ucred, p))
372 return (error);
373 if (fs->lfs_ivnode->v_dirtyblkhd.lh_first)
374 panic("lfs_unmount: still dirty blocks on ifile vnode\n");
375 vrele(fs->lfs_ivnode);
376 vgone(fs->lfs_ivnode);
377
378 ronly = !fs->lfs_ronly;
379 ump->um_devvp->v_specflags &= ~SI_MOUNTEDON;
380 error = VOP_CLOSE(ump->um_devvp,
381 ronly ? FREAD : FREAD|FWRITE, NOCRED, p);
382 vrele(ump->um_devvp);
383 free(fs, M_UFSMNT);
384 free(ump, M_UFSMNT);
385 mp->mnt_data = (qaddr_t)0;
386 mp->mnt_flag &= ~MNT_LOCAL;
387 return (error);
388 }
389
390 /*
391 * Get file system statistics.
392 */
393 lfs_statfs(mp, sbp, p)
394 struct mount *mp;
395 register struct statfs *sbp;
396 struct proc *p;
397 {
398 register struct lfs *fs;
399 register struct ufsmount *ump;
400
401 ump = VFSTOUFS(mp);
402 fs = ump->um_lfs;
403 if (fs->lfs_magic != LFS_MAGIC)
404 panic("lfs_statfs: magic");
405 sbp->f_type = 0;
406 sbp->f_bsize = fs->lfs_bsize;
407 sbp->f_iosize = fs->lfs_bsize;
408 sbp->f_blocks = dbtofsb(fs,fs->lfs_dsize);
409 sbp->f_bfree = dbtofsb(fs, fs->lfs_bfree);
410 sbp->f_bavail = (fs->lfs_dsize * (100 - fs->lfs_minfree) / 100) -
411 (fs->lfs_dsize - fs->lfs_bfree);
412 sbp->f_bavail = dbtofsb(fs, sbp->f_bavail);
413 sbp->f_files = fs->lfs_nfiles;
414 sbp->f_ffree = sbp->f_bfree * INOPB(fs);
415 if (sbp != &mp->mnt_stat) {
416 bcopy(mp->mnt_stat.f_mntonname, sbp->f_mntonname, MNAMELEN);
417 bcopy(mp->mnt_stat.f_mntfromname, sbp->f_mntfromname, MNAMELEN);
418 }
419 strncpy(sbp->f_fstypename, mp->mnt_op->vfs_name, MFSNAMELEN);
420 sbp->f_fstypename[MFSNAMELEN] = '\0';
421 return (0);
422 }
423
424 /*
425 * Go through the disk queues to initiate sandbagged IO;
426 * go through the inodes to write those that have been modified;
427 * initiate the writing of the super block if it has been modified.
428 *
429 * Note: we are always called with the filesystem marked `MPBUSY'.
430 */
431 lfs_sync(mp, waitfor, cred, p)
432 struct mount *mp;
433 int waitfor;
434 struct ucred *cred;
435 struct proc *p;
436 {
437 int error;
438
439 /* All syncs must be checkpoints until roll-forward is implemented. */
440 error = lfs_segwrite(mp, SEGM_CKP | (waitfor ? SEGM_SYNC : 0));
441 #ifdef QUOTA
442 qsync(mp);
443 #endif
444 return (error);
445 }
446
447 /*
448 * Look up an LFS dinode number to find its incore vnode. If not already
449 * in core, read it in from the specified device. Return the inode locked.
450 * Detection and handling of mount points must be done by the calling routine.
451 */
452 int
453 lfs_vget(mp, ino, vpp)
454 struct mount *mp;
455 ino_t ino;
456 struct vnode **vpp;
457 {
458 register struct lfs *fs;
459 register struct inode *ip;
460 struct buf *bp;
461 struct ifile *ifp;
462 struct vnode *vp;
463 struct ufsmount *ump;
464 daddr_t daddr;
465 dev_t dev;
466 int error;
467
468 ump = VFSTOUFS(mp);
469 dev = ump->um_dev;
470 if ((*vpp = ufs_ihashget(dev, ino)) != NULL)
471 return (0);
472
473 /* Translate the inode number to a disk address. */
474 fs = ump->um_lfs;
475 if (ino == LFS_IFILE_INUM)
476 daddr = fs->lfs_idaddr;
477 else {
478 LFS_IENTRY(ifp, fs, ino, bp);
479 daddr = ifp->if_daddr;
480 brelse(bp);
481 if (daddr == LFS_UNUSED_DADDR)
482 return (ENOENT);
483 }
484
485 /* Allocate new vnode/inode. */
486 if (error = lfs_vcreate(mp, ino, &vp)) {
487 *vpp = NULL;
488 return (error);
489 }
490
491 /*
492 * Put it onto its hash chain and lock it so that other requests for
493 * this inode will block if they arrive while we are sleeping waiting
494 * for old data structures to be purged or for the contents of the
495 * disk portion of this inode to be read.
496 */
497 ip = VTOI(vp);
498 ufs_ihashins(ip);
499
500 /*
501 * XXX
502 * This may not need to be here, logically it should go down with
503 * the i_devvp initialization.
504 * Ask Kirk.
505 */
506 ip->i_lfs = ump->um_lfs;
507
508 /* Read in the disk contents for the inode, copy into the inode. */
509 if (error =
510 bread(ump->um_devvp, daddr, (int)fs->lfs_bsize, NOCRED, &bp)) {
511 /*
512 * The inode does not contain anything useful, so it would
513 * be misleading to leave it on its hash chain. With mode
514 * still zero, it will be unlinked and returned to the free
515 * list by vput().
516 */
517 vput(vp);
518 brelse(bp);
519 *vpp = NULL;
520 return (error);
521 }
522 ip->i_din = *lfs_ifind(fs, ino, (struct dinode *)bp->b_data);
523 brelse(bp);
524
525 /*
526 * Initialize the vnode from the inode, check for aliases. In all
527 * cases re-init ip, the underlying vnode/inode may have changed.
528 */
529 if (error = ufs_vinit(mp, lfs_specop_p, LFS_FIFOOPS, &vp)) {
530 vput(vp);
531 *vpp = NULL;
532 return (error);
533 }
534 /*
535 * Finish inode initialization now that aliasing has been resolved.
536 */
537 ip->i_devvp = ump->um_devvp;
538 VREF(ip->i_devvp);
539 *vpp = vp;
540 return (0);
541 }
542
543 /*
544 * File handle to vnode
545 *
546 * Have to be really careful about stale file handles:
547 * - check that the inode number is valid
548 * - call lfs_vget() to get the locked inode
549 * - check for an unallocated inode (i_mode == 0)
550 * - check that the given client host has export rights and return
551 * those rights via. exflagsp and credanonp
552 *
553 * XXX
554 * use ifile to see if inode is allocated instead of reading off disk
555 * what is the relationship between my generational number and the NFS
556 * generational number.
557 */
558 int
559 lfs_fhtovp(mp, fhp, nam, vpp, exflagsp, credanonp)
560 register struct mount *mp;
561 struct fid *fhp;
562 struct mbuf *nam;
563 struct vnode **vpp;
564 int *exflagsp;
565 struct ucred **credanonp;
566 {
567 register struct ufid *ufhp;
568
569 ufhp = (struct ufid *)fhp;
570 if (ufhp->ufid_ino < ROOTINO)
571 return (ESTALE);
572 return (ufs_check_export(mp, ufhp, nam, vpp, exflagsp, credanonp));
573 }
574
575 /*
576 * Vnode pointer to File handle
577 */
578 /* ARGSUSED */
579 lfs_vptofh(vp, fhp)
580 struct vnode *vp;
581 struct fid *fhp;
582 {
583 register struct inode *ip;
584 register struct ufid *ufhp;
585
586 ip = VTOI(vp);
587 ufhp = (struct ufid *)fhp;
588 ufhp->ufid_len = sizeof(struct ufid);
589 ufhp->ufid_ino = ip->i_number;
590 ufhp->ufid_gen = ip->i_gen;
591 return (0);
592 }
593