2 * Copyright (c) 1989, 1990, 1993, 1994
3 * The Regents of the University of California. All rights reserved.
5 * Redistribution and use in source and binary forms, with or without
6 * modification, are permitted provided that the following conditions
8 * 1. Redistributions of source code must retain the above copyright
9 * notice, this list of conditions and the following disclaimer.
10 * 2. Redistributions in binary form must reproduce the above copyright
11 * notice, this list of conditions and the following disclaimer in the
12 * documentation and/or other materials provided with the distribution.
13 * 3. All advertising materials mentioning features or use of this software
14 * must display the following acknowledgement:
15 * This product includes software developed by the University of
16 * California, Berkeley and its contributors.
17 * 4. Neither the name of the University nor the names of its contributors
18 * may be used to endorse or promote products derived from this software
19 * without specific prior written permission.
21 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
22 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
23 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
24 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
25 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
26 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
27 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
28 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
29 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
30 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
33 * @(#)mfs_vfsops.c 8.11 (Berkeley) 6/19/95
34 * $FreeBSD: src/sys/ufs/mfs/mfs_vfsops.c,v 1.81.2.3 2001/07/04 17:35:21 tegge Exp $
35 * $DragonFly: src/sys/vfs/mfs/mfs_vfsops.c,v 1.40 2007/05/09 00:53:35 dillon Exp $
39 #include <sys/param.h>
40 #include <sys/systm.h>
42 #include <sys/device.h>
43 #include <sys/kernel.h>
46 #include <sys/mount.h>
47 #include <sys/signalvar.h>
48 #include <sys/signal2.h>
49 #include <sys/vnode.h>
50 #include <sys/malloc.h>
51 #include <sys/linker.h>
52 #include <sys/fcntl.h>
55 #include <vm/vm_object.h>
56 #include <vm/vm_page.h>
57 #include <vm/vm_pager.h>
58 #include <vm/vnode_pager.h>
61 #include <sys/thread2.h>
63 #include <vfs/ufs/quota.h>
64 #include <vfs/ufs/inode.h>
65 #include <vfs/ufs/ufsmount.h>
66 #include <vfs/ufs/ufs_extern.h>
67 #include <vfs/ufs/fs.h>
68 #include <vfs/ufs/ffs_extern.h>
71 #include "mfs_extern.h"
73 MALLOC_DEFINE(M_MFSNODE, "MFS node", "MFS vnode private part");
75 extern struct vop_ops *mfs_vnode_vops_p;
77 static int mfs_mount (struct mount *mp,
78 char *path, caddr_t data, struct ucred *td);
79 static int mfs_start (struct mount *mp, int flags);
80 static int mfs_statfs (struct mount *mp, struct statfs *sbp,
82 static int mfs_init (struct vfsconf *);
86 d_strategy_t mfsstrategy;
88 static struct dev_ops mfs_ops = {
89 { "MFS", -1, D_DISK },
94 .d_strategy = mfsstrategy,
100 static struct vfsops mfs_vfsops = {
101 .vfs_mount = mfs_mount,
102 .vfs_start = mfs_start,
103 .vfs_unmount = ffs_unmount,
104 .vfs_root = ufs_root,
105 .vfs_quotactl = ufs_quotactl,
106 .vfs_statfs = mfs_statfs,
107 .vfs_sync = ffs_sync,
108 .vfs_vget = ffs_vget,
109 .vfs_fhtovp = ffs_fhtovp,
110 .vfs_checkexp = ufs_check_export,
111 .vfs_vptofh = ffs_vptofh,
115 VFS_SET(mfs_vfsops, mfs, 0);
118 * We allow the underlying MFS block device to be opened and read.
121 mfsopen(struct dev_open_args *ap)
123 cdev_t dev = ap->a_head.a_dev;
125 if (ap->a_oflags & FWRITE)
133 mfsclose(struct dev_close_args *ap)
139 mfsstrategy(struct dev_strategy_args *ap)
141 cdev_t dev = ap->a_head.a_dev;
142 struct bio *bio = ap->a_bio;
143 struct buf *bp = bio->bio_buf;
144 off_t boff = bio->bio_offset;
145 off_t eoff = boff + bp->b_bcount;
146 struct mfsnode *mfsp;
148 if ((mfsp = dev->si_drv1) == NULL) {
154 if (eoff > mfsp->mfs_size) {
155 if (boff > mfsp->mfs_size || (bp->b_flags & B_BNOCLIP))
158 * Return EOF by completing the I/O with 0 bytes transfered.
159 * Set B_INVAL to indicate that any data in the buffer is not
162 if (boff == mfsp->mfs_size) {
163 bp->b_resid = bp->b_bcount;
164 bp->b_flags |= B_INVAL;
167 bp->b_bcount = mfsp->mfs_size - boff;
173 bioq_insert_tail(&mfsp->bio_queue, bio);
174 wakeup((caddr_t)mfsp);
178 * Failure conditions on bio
181 bp->b_error = EINVAL;
183 bp->b_flags |= B_ERROR | B_INVAL;
192 * Called when mounting local physical media
196 * mp mount point structure
197 * path NULL (flag for root mount!!!)
200 * p process (user credentials check [statfs])
203 * mp mount point structure
204 * path path to mount point
205 * data pointer to argument struct in user space
206 * ndp mount point namei() return (used for
207 * credentials on reload), reused to look
209 * p process (user credentials check)
212 * !0 error number (errno.h)
217 * mount point is locked
219 * mount point is locked
222 * A NULL path can be used for a flag since the mount
223 * system call will fail with EFAULT in copyinstr in
224 * namei() if it is a genuine NULL from the user.
228 mfs_mount(struct mount *mp, char *path, caddr_t data, struct ucred *cred)
231 struct mfs_args args;
232 struct ufsmount *ump;
234 struct mfsnode *mfsp;
241 * Use NULL path to flag a root mount
246 * Mounting root file system
251 panic("mfs_mount: mount MFS as root: not configured!");
256 * Mounting non-root file system or updating a file system
260 /* copy in user arguments*/
261 if ((err = copyin(data, (caddr_t)&args, sizeof (struct mfs_args))) != 0)
265 * If updating, check whether changing from read-only to
266 * read/write; if there is no device name, that's all we do.
268 if (mp->mnt_flag & MNT_UPDATE) {
276 if (fs->fs_ronly == 0 && (mp->mnt_flag & MNT_RDONLY)) {
278 if (mp->mnt_flag & MNT_FORCE)
280 err = ffs_flushfiles(mp, flags);
284 if (fs->fs_ronly && (mp->mnt_kern_flag & MNTK_WANTRDWR)) {
285 /* XXX reopen the device vnode read-write */
288 /* if not updating name...*/
289 if (args.fspec == 0) {
291 * Process export requests. Jumping to "success"
292 * will return the vfs_export() error code.
294 err = vfs_export(mp, &ump->um_export, &args.export);
298 /* XXX MFS does not support name updating*/
302 * Do the MALLOC before the getnewvnode since doing so afterward
303 * might cause a bogus v_data pointer to get dereferenced
304 * elsewhere if MALLOC should block.
306 MALLOC(mfsp, struct mfsnode *, sizeof *mfsp, M_MFSNODE, M_WAITOK);
308 err = getspecialvnode(VT_MFS, NULL, &mfs_vnode_vops_p, &devvp, 0, 0);
310 FREE(mfsp, M_MFSNODE);
314 minnum = (curproc->p_pid & 0xFF) |
315 ((curproc->p_pid & ~0xFF) << 8);
317 devvp->v_type = VCHR;
318 dev = make_dev(&mfs_ops, minnum, UID_ROOT, GID_WHEEL, 0600,
319 "MFS%d", minnum >> 16);
320 /* It is not clear that these will get initialized otherwise */
321 dev->si_bsize_phys = DEV_BSIZE;
322 dev->si_iosize_max = DFLTPHYS;
324 addaliasu(devvp, mfs_ops.head.maj, minnum);
325 devvp->v_data = mfsp;
326 mfsp->mfs_baseoff = args.base;
327 mfsp->mfs_size = args.size;
328 mfsp->mfs_vnode = devvp;
329 mfsp->mfs_dev = reference_dev(dev);
330 mfsp->mfs_td = curthread;
331 mfsp->mfs_active = 1;
332 bioq_init(&mfsp->bio_queue);
335 * Our 'block' device must be backed by a VM object. Theoretically
336 * we could use the anonymous memory VM object supplied by userland,
337 * but it would be somewhat of a complex task to deal with it
338 * that way since it would result in I/O requests which supply
339 * the VM pages from our own object.
341 * vnode_pager_alloc() is typically called when a VM object is
342 * being referenced externally. We have to undo the refs for
343 * the self reference between vnode and object.
345 vnode_pager_alloc(devvp, args.size, 0, 0);
347 --devvp->v_object->ref_count;
349 /* Save "mounted from" info for mount point (NULL pad)*/
350 copyinstr( args.fspec, /* device name*/
351 mp->mnt_stat.f_mntfromname, /* save area*/
352 MNAMELEN - 1, /* max size*/
353 &size); /* real size*/
354 bzero( mp->mnt_stat.f_mntfromname + size, MNAMELEN - size);
357 if ((err = ffs_mountfs(devvp, mp, M_MFSNODE)) != 0) {
358 mfsp->mfs_active = 0;
363 * Initialize FS stat information in mount struct; uses
364 * mp->mnt_stat.f_mntfromname.
366 * This code is common to root and non-root mounts
368 VFS_STATFS(mp, &mp->mnt_stat, cred);
372 error_2: /* error with devvp held*/
374 /* release devvp before failing*/
377 error_1: /* no state to back out*/
384 * Used to grab the process and keep it in the kernel to service
385 * memory filesystem I/O requests.
387 * Loop servicing I/O requests.
388 * Copy the requested data into or out of the memory filesystem
393 mfs_start(struct mount *mp, int flags)
395 struct vnode *vp = VFSTOUFS(mp)->um_devvp;
396 struct mfsnode *mfsp = VTOMFS(vp);
400 thread_t td = curthread;
403 * We must prevent the system from trying to swap
404 * out or kill ( when swap space is low, see vm/pageout.c ) the
405 * process. A deadlock can occur if the process is swapped out,
406 * and the system can loop trying to kill the unkillable ( while
407 * references exist ) MFS process when swap space is low.
414 while (mfsp->mfs_active) {
417 while ((bio = bioq_first(&mfsp->bio_queue)) != NULL) {
418 bioq_remove(&mfsp->bio_queue, bio);
429 * If a non-ignored signal is received, try to unmount.
430 * If that fails, clear the signal (it has been "processed"),
431 * otherwise we will loop here, as tsleep will always return
435 * Note that dounmount() may fail if work was queued after
436 * we slept. We have to jump hoops here to make sure that we
437 * process any buffers after the sleep, before we dounmount()
441 if (dounmount(mp, 0) != 0) {
442 KKASSERT(td->td_proc);
443 sig = CURSIG(td->td_lwp);
445 lwp_delsig(td->td_lwp, sig);
448 else if (tsleep((caddr_t)mfsp, PCATCH, "mfsidl", 0))
449 gotsig++; /* try to unmount in next pass */
452 v_release_rdev(vp); /* hack because we do not implement CLOSE */
453 /* XXX destroy/release devvp */
458 * Get file system statistics.
461 mfs_statfs(struct mount *mp, struct statfs *sbp, struct ucred *cred)
465 error = ffs_statfs(mp, sbp, cred);
466 sbp->f_type = mp->mnt_vfc->vfc_typenum;
471 * Memory based filesystem initialization.
474 mfs_init(struct vfsconf *vfsp)
476 dev_ops_add(&mfs_ops, 0, 0);