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 1.29 2006/05/04 18:32:23 dillon Exp $
41 #include <sys/param.h>
42 #include <sys/systm.h>
44 #include <sys/kernel.h>
47 #include <sys/mount.h>
48 #include <sys/signalvar.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");
76 extern struct vop_ops *mfs_vnode_vops;
78 static int mfs_mount (struct mount *mp,
79 char *path, caddr_t data, struct thread *td);
80 static int mfs_start (struct mount *mp, int flags, struct thread *td);
81 static int mfs_statfs (struct mount *mp, struct statfs *sbp,
83 static int mfs_init (struct vfsconf *);
87 d_strategy_t mfsstrategy;
89 #define MFS_CDEV_MAJOR 253
91 static struct cdevsw mfs_cdevsw = {
93 /* maj */ MFS_CDEV_MAJOR,
101 /* write */ physwrite,
105 /* strategy */ mfsstrategy,
111 * mfs vfs operations.
113 static struct vfsops mfs_vfsops = {
114 .vfs_mount = mfs_mount,
115 .vfs_start = mfs_start,
116 .vfs_unmount = ffs_unmount,
117 .vfs_root = ufs_root,
118 .vfs_quotactl = ufs_quotactl,
119 .vfs_statfs = mfs_statfs,
120 .vfs_sync = ffs_sync,
121 .vfs_vget = ffs_vget,
122 .vfs_fhtovp = ffs_fhtovp,
123 .vfs_checkexp = ufs_check_export,
124 .vfs_vptofh = ffs_vptofh,
128 VFS_SET(mfs_vfsops, mfs, 0);
131 * We allow the underlying MFS block device to be opened and read.
134 mfsopen(dev_t dev, int flags, int mode, struct thread *td)
144 mfsclose(dev_t dev, int flags, int mode, struct thread *td)
150 mfsstrategy(dev_t dev, struct bio *bio)
152 struct buf *bp = bio->bio_buf;
153 off_t boff = bio->bio_offset;
154 off_t eoff = boff + bp->b_bcount;
155 struct mfsnode *mfsp;
157 if ((mfsp = dev->si_drv1) == NULL) {
163 if (eoff > mfsp->mfs_size) {
164 if (boff > mfsp->mfs_size || (bp->b_flags & B_BNOCLIP))
167 * Return EOF by completing the I/O with 0 bytes transfered.
168 * Set B_INVAL to indicate that any data in the buffer is not
171 if (boff == mfsp->mfs_size) {
172 bp->b_resid = bp->b_bcount;
173 bp->b_flags |= B_INVAL;
176 bp->b_bcount = mfsp->mfs_size - boff;
182 bioq_insert_tail(&mfsp->bio_queue, bio);
183 wakeup((caddr_t)mfsp);
187 * Failure conditions on bio
190 bp->b_error = EINVAL;
192 bp->b_flags |= B_ERROR | B_INVAL;
200 * Called when mounting local physical media
204 * mp mount point structure
205 * path NULL (flag for root mount!!!)
208 * p process (user credentials check [statfs])
211 * mp mount point structure
212 * path path to mount point
213 * data pointer to argument struct in user space
214 * ndp mount point namei() return (used for
215 * credentials on reload), reused to look
217 * p process (user credentials check)
220 * !0 error number (errno.h)
225 * mount point is locked
227 * mount point is locked
230 * A NULL path can be used for a flag since the mount
231 * system call will fail with EFAULT in copyinstr in
232 * namei() if it is a genuine NULL from the user.
236 mfs_mount(struct mount *mp, char *path, caddr_t data, struct thread *td)
239 struct mfs_args args;
240 struct ufsmount *ump;
242 struct mfsnode *mfsp;
249 * Use NULL path to flag a root mount
254 * Mounting root file system
259 panic("mfs_mount: mount MFS as root: not configured!");
264 * Mounting non-root file system or updating a file system
268 /* copy in user arguments*/
269 if ((err = copyin(data, (caddr_t)&args, sizeof (struct mfs_args))) != 0)
273 * If updating, check whether changing from read-only to
274 * read/write; if there is no device name, that's all we do.
276 if (mp->mnt_flag & MNT_UPDATE) {
284 if (fs->fs_ronly == 0 && (mp->mnt_flag & MNT_RDONLY)) {
286 if (mp->mnt_flag & MNT_FORCE)
288 err = ffs_flushfiles(mp, flags, td);
292 if (fs->fs_ronly && (mp->mnt_kern_flag & MNTK_WANTRDWR)) {
293 /* XXX reopen the device vnode read-write */
296 /* if not updating name...*/
297 if (args.fspec == 0) {
299 * Process export requests. Jumping to "success"
300 * will return the vfs_export() error code.
302 err = vfs_export(mp, &ump->um_export, &args.export);
306 /* XXX MFS does not support name updating*/
310 * Do the MALLOC before the getnewvnode since doing so afterward
311 * might cause a bogus v_data pointer to get dereferenced
312 * elsewhere if MALLOC should block.
314 MALLOC(mfsp, struct mfsnode *, sizeof *mfsp, M_MFSNODE, M_WAITOK);
316 err = getspecialvnode(VT_MFS, NULL, &mfs_vnode_vops, &devvp, 0, 0);
318 FREE(mfsp, M_MFSNODE);
322 minnum = (curproc->p_pid & 0xFF) |
323 ((curproc->p_pid & ~0xFF) << 8);
325 devvp->v_type = VCHR;
326 dev = make_dev(&mfs_cdevsw, minnum, UID_ROOT, GID_WHEEL, 0600,
327 "MFS%d", minnum >> 16);
328 /* It is not clear that these will get initialized otherwise */
329 dev->si_bsize_phys = DEV_BSIZE;
330 dev->si_iosize_max = DFLTPHYS;
332 addaliasu(devvp, makeudev(MFS_CDEV_MAJOR, minnum));
333 devvp->v_data = mfsp;
334 mfsp->mfs_baseoff = args.base;
335 mfsp->mfs_size = args.size;
336 mfsp->mfs_vnode = devvp;
337 mfsp->mfs_dev = reference_dev(dev);
339 mfsp->mfs_active = 1;
340 bioq_init(&mfsp->bio_queue);
343 * Our 'block' device must be backed by a VM object. Theoretically
344 * we could use the anonymous memory VM object supplied by userland,
345 * but it would be somewhat of a complex task to deal with it
346 * that way since it would result in I/O requests which supply
347 * the VM pages from our own object.
349 * vnode_pager_alloc() is typically called when a VM object is
350 * being referenced externally. We have to undo the refs for
351 * the self reference between vnode and object.
353 vnode_pager_alloc(devvp, args.size, 0, 0);
355 --devvp->v_object->ref_count;
357 /* Save "mounted from" info for mount point (NULL pad)*/
358 copyinstr( args.fspec, /* device name*/
359 mp->mnt_stat.f_mntfromname, /* save area*/
360 MNAMELEN - 1, /* max size*/
361 &size); /* real size*/
362 bzero( mp->mnt_stat.f_mntfromname + size, MNAMELEN - size);
365 if ((err = ffs_mountfs(devvp, mp, td, M_MFSNODE)) != 0) {
366 mfsp->mfs_active = 0;
371 * Initialize FS stat information in mount struct; uses
372 * mp->mnt_stat.f_mntfromname.
374 * This code is common to root and non-root mounts
376 VFS_STATFS(mp, &mp->mnt_stat, td);
380 error_2: /* error with devvp held*/
382 /* release devvp before failing*/
385 error_1: /* no state to back out*/
392 * Used to grab the process and keep it in the kernel to service
393 * memory filesystem I/O requests.
395 * Loop servicing I/O requests.
396 * Copy the requested data into or out of the memory filesystem
401 mfs_start(struct mount *mp, int flags, struct thread *td)
403 struct vnode *vp = VFSTOUFS(mp)->um_devvp;
404 struct mfsnode *mfsp = VTOMFS(vp);
410 * We must prevent the system from trying to swap
411 * out or kill ( when swap space is low, see vm/pageout.c ) the
412 * process. A deadlock can occur if the process is swapped out,
413 * and the system can loop trying to kill the unkillable ( while
414 * references exist ) MFS process when swap space is low.
419 while (mfsp->mfs_active) {
422 while ((bio = bioq_first(&mfsp->bio_queue)) != NULL) {
423 bioq_remove(&mfsp->bio_queue, bio);
434 * If a non-ignored signal is received, try to unmount.
435 * If that fails, clear the signal (it has been "processed"),
436 * otherwise we will loop here, as tsleep will always return
440 * Note that dounmount() may fail if work was queued after
441 * we slept. We have to jump hoops here to make sure that we
442 * process any buffers after the sleep, before we dounmount()
446 if (dounmount(mp, 0, td) != 0) {
447 KKASSERT(td->td_proc);
448 sig = CURSIG(td->td_proc);
450 SIGDELSET(td->td_proc->p_siglist, sig);
453 else if (tsleep((caddr_t)mfsp, PCATCH, "mfsidl", 0))
454 gotsig++; /* try to unmount in next pass */
457 v_release_rdev(vp); /* hack because we do not implement CLOSE */
458 /* XXX destroy/release devvp */
463 * Get file system statistics.
466 mfs_statfs(struct mount *mp, struct statfs *sbp, struct thread *td)
470 error = ffs_statfs(mp, sbp, td);
471 sbp->f_type = mp->mnt_vfc->vfc_typenum;
476 * Memory based filesystem initialization.
479 mfs_init(struct vfsconf *vfsp)
481 cdevsw_add(&mfs_cdevsw, 0, 0);