2 * Copyright (c) 1989, 1993
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_vnops.c 8.11 (Berkeley) 5/22/95
34 * $FreeBSD: src/sys/ufs/mfs/mfs_vnops.c,v 1.47.2.1 2001/05/22 02:06:43 bp Exp $
35 * $DragonFly: src/sys/vfs/mfs/mfs_vnops.c,v 1.2 2003/06/17 04:28:59 dillon Exp $
38 #include <sys/param.h>
39 #include <sys/systm.h>
40 #include <sys/kernel.h>
43 #include <sys/vnode.h>
44 #include <sys/malloc.h>
45 #include <sys/sysproto.h>
49 #include <ufs/mfs/mfsnode.h>
50 #include <ufs/mfs/mfs_extern.h>
52 static int mfs_badop __P((struct vop_generic_args *));
53 static int mfs_bmap __P((struct vop_bmap_args *));
54 static int mfs_close __P((struct vop_close_args *));
55 static int mfs_fsync __P((struct vop_fsync_args *));
56 static int mfs_freeblks __P((struct vop_freeblks_args *));
57 static int mfs_inactive __P((struct vop_inactive_args *)); /* XXX */
58 static int mfs_open __P((struct vop_open_args *));
59 static int mfs_reclaim __P((struct vop_reclaim_args *)); /* XXX */
60 static int mfs_print __P((struct vop_print_args *)); /* XXX */
61 static int mfs_strategy __P((struct vop_strategy_args *)); /* XXX */
62 static int mfs_getpages __P((struct vop_getpages_args *)); /* XXX */
64 * mfs vnode operations.
66 vop_t **mfs_vnodeop_p;
67 static struct vnodeopv_entry_desc mfs_vnodeop_entries[] = {
68 { &vop_default_desc, (vop_t *) mfs_badop },
69 { &vop_bmap_desc, (vop_t *) mfs_bmap },
70 { &vop_bwrite_desc, (vop_t *) vop_defaultop },
71 { &vop_close_desc, (vop_t *) mfs_close },
72 { &vop_createvobject_desc, (vop_t *) vop_stdcreatevobject },
73 { &vop_destroyvobject_desc, (vop_t *) vop_stddestroyvobject },
74 { &vop_freeblks_desc, (vop_t *) mfs_freeblks },
75 { &vop_fsync_desc, (vop_t *) mfs_fsync },
76 { &vop_getpages_desc, (vop_t *) mfs_getpages },
77 { &vop_getvobject_desc, (vop_t *) vop_stdgetvobject },
78 { &vop_inactive_desc, (vop_t *) mfs_inactive },
79 { &vop_ioctl_desc, (vop_t *) vop_enotty },
80 { &vop_islocked_desc, (vop_t *) vop_defaultop },
81 { &vop_lock_desc, (vop_t *) vop_defaultop },
82 { &vop_open_desc, (vop_t *) mfs_open },
83 { &vop_print_desc, (vop_t *) mfs_print },
84 { &vop_reclaim_desc, (vop_t *) mfs_reclaim },
85 { &vop_strategy_desc, (vop_t *) mfs_strategy },
86 { &vop_unlock_desc, (vop_t *) vop_defaultop },
89 static struct vnodeopv_desc mfs_vnodeop_opv_desc =
90 { &mfs_vnodeop_p, mfs_vnodeop_entries };
92 VNODEOP_SET(mfs_vnodeop_opv_desc);
97 * Open called to allow memory filesystem to initialize and
98 * validate before actual IO. Record our process identifier
99 * so we can tell when we are doing I/O to ourself.
104 struct vop_open_args /* {
107 struct ucred *a_cred;
112 if (ap->a_vp->v_type != VBLK) {
113 panic("mfs_open not VBLK");
121 struct vop_fsync_args *ap;
124 return (VOCALL(spec_vnodeop_p, VOFFSET(vop_fsync), ap));
128 * mfs_freeblks() - hook to allow us to free physical memory.
130 * We implement the B_FREEBUF strategy. We can't just madvise()
131 * here because we have to do it in the correct order vs other bio
132 * requests, so we queue it.
134 * Note: geteblk() sets B_INVAL. We leave it set to guarentee buffer
135 * throw-away on brelse()? XXX
140 struct vop_freeblks_args /* {
149 if (!vfinddev(ap->a_vp->v_rdev, VBLK, &vp) || vp->v_usecount == 0)
150 panic("mfs_freeblks: bad dev");
152 bp = geteblk(ap->a_length);
153 bp->b_flags |= B_FREEBUF | B_ASYNC;
154 bp->b_dev = ap->a_vp->v_rdev;
155 bp->b_blkno = ap->a_addr;
156 bp->b_offset = dbtob(ap->a_addr);
157 bp->b_bcount = ap->a_length;
159 VOP_STRATEGY(vp, bp);
164 * Pass I/O requests to the memory filesystem process.
168 struct vop_strategy_args /* {
173 register struct buf *bp = ap->a_bp;
174 register struct mfsnode *mfsp;
176 struct proc *p = curproc; /* XXX */
179 if (!vfinddev(bp->b_dev, VBLK, &vp) || vp->v_usecount == 0)
180 panic("mfs_strategy: bad dev");
184 * splbio required for queueing/dequeueing, in case of forwarded
185 * BPs from bio interrupts (?). It may not be necessary.
190 if (mfsp->mfs_pid == 0) {
192 * mini-root. Note: B_FREEBUF not supported at the moment,
193 * I'm not sure what kind of dataspace b_data is in.
197 base = mfsp->mfs_baseoff + (bp->b_blkno << DEV_BSHIFT);
198 if (bp->b_flags & B_FREEBUF)
200 if (bp->b_flags & B_READ)
201 bcopy(base, bp->b_data, bp->b_bcount);
203 bcopy(bp->b_data, base, bp->b_bcount);
205 } else if (mfsp->mfs_pid == p->p_pid) {
214 * VOP from some other process, queue to MFS process and
217 bufq_insert_tail(&mfsp->buf_queue, bp);
225 * Memory file system I/O.
227 * Trivial on the HP since buffer has already been mapping into KVA space.
229 * Read and Write are handled with a simple copyin and copyout.
231 * We also partially support VOP_FREEBLKS() via B_FREEBUF. We can't implement
232 * completely -- for example, on fragments or inode metadata, but we can
233 * implement it for page-aligned requests.
237 register struct buf *bp;
238 struct mfsnode *mfsp;
240 caddr_t base = mfsp->mfs_baseoff + (bp->b_blkno << DEV_BSHIFT);
242 if (bp->b_flags & B_FREEBUF) {
244 * Implement B_FREEBUF, which allows the filesystem to tell
245 * a block device when blocks are no longer needed (like when
246 * a file is deleted). We use the hook to MADV_FREE the VM.
247 * This makes an MFS filesystem work as well or better then
248 * a sun-style swap-mounted filesystem.
250 int bytes = bp->b_bcount;
252 if ((vm_offset_t)base & PAGE_MASK) {
253 int n = PAGE_SIZE - ((vm_offset_t)base & PAGE_MASK);
258 struct madvise_args uap;
262 bzero(&uap, sizeof(uap));
265 uap.behav = MADV_FREE;
266 madvise(curproc, &uap);
270 } else if (bp->b_flags & B_READ) {
272 * Read data from our 'memory' disk
274 bp->b_error = copyin(base, bp->b_data, bp->b_bcount);
277 * Write data to our 'memory' disk
279 bp->b_error = copyout(bp->b_data, base, bp->b_bcount);
282 bp->b_flags |= B_ERROR;
287 * This is a noop, simply returning what one has been given.
291 struct vop_bmap_args /* {
294 struct vnode **a_vpp;
300 if (ap->a_vpp != NULL)
301 *ap->a_vpp = ap->a_vp;
302 if (ap->a_bnp != NULL)
303 *ap->a_bnp = ap->a_bn;
304 if (ap->a_runp != NULL)
310 * Memory filesystem close routine
315 struct vop_close_args /* {
318 struct ucred *a_cred;
322 register struct vnode *vp = ap->a_vp;
323 register struct mfsnode *mfsp = VTOMFS(vp);
324 register struct buf *bp;
328 * Finish any pending I/O requests.
330 while ((bp = bufq_first(&mfsp->buf_queue)) != NULL) {
331 bufq_remove(&mfsp->buf_queue, bp);
336 * On last close of a memory filesystem
337 * we must invalidate any in core blocks, so that
338 * we can, free up its vnode.
340 if ((error = vinvalbuf(vp, 1, ap->a_cred, ap->a_p, 0, 0)) != 0)
343 * There should be no way to have any more uses of this
344 * vnode, so if we find any other uses, it is a panic.
346 if (vp->v_usecount > 1)
347 printf("mfs_close: ref count %d > 1\n", vp->v_usecount);
348 if (vp->v_usecount > 1 || (bufq_first(&mfsp->buf_queue) != NULL))
351 * Send a request to the filesystem server to exit.
353 mfsp->mfs_active = 0;
359 * Memory filesystem inactive routine
364 struct vop_inactive_args /* {
369 struct vnode *vp = ap->a_vp;
370 struct mfsnode *mfsp = VTOMFS(vp);
372 if (bufq_first(&mfsp->buf_queue) != NULL)
373 panic("mfs_inactive: not inactive (next buffer %p)",
374 bufq_first(&mfsp->buf_queue));
375 VOP_UNLOCK(vp, 0, ap->a_p);
380 * Reclaim a memory filesystem devvp so that it can be reused.
384 struct vop_reclaim_args /* {
388 register struct vnode *vp = ap->a_vp;
390 FREE(vp->v_data, M_MFSNODE);
396 * Print out the contents of an mfsnode.
400 struct vop_print_args /* {
404 register struct mfsnode *mfsp = VTOMFS(ap->a_vp);
406 printf("tag VT_MFS, pid %ld, base %p, size %ld\n",
407 (long)mfsp->mfs_pid, (void *)mfsp->mfs_baseoff, mfsp->mfs_size);
412 * Block device bad operation
415 mfs_badop(struct vop_generic_args *ap)
419 printf("mfs_badop[%s]\n", ap->a_desc->vdesc_name);
420 i = vop_defaultop(ap);
421 printf("mfs_badop[%s] = %d\n", ap->a_desc->vdesc_name,i);
428 struct vop_getpages_args *ap;
431 return (VOCALL(spec_vnodeop_p, VOFFSET(vop_getpages), ap));