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 1.17 2005/02/15 08:32:18 joerg 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>
52 #include "mfs_extern.h"
54 static int mfs_badop (struct vop_generic_args *);
55 static int mfs_bmap (struct vop_bmap_args *);
56 static int mfs_close (struct vop_close_args *);
57 static int mfs_fsync (struct vop_fsync_args *);
58 static int mfs_freeblks (struct vop_freeblks_args *);
59 static int mfs_inactive (struct vop_inactive_args *); /* XXX */
60 static int mfs_open (struct vop_open_args *);
61 static int mfs_reclaim (struct vop_reclaim_args *); /* XXX */
62 static int mfs_print (struct vop_print_args *); /* XXX */
63 static int mfs_strategy (struct vop_strategy_args *); /* XXX */
64 static int mfs_getpages (struct vop_getpages_args *); /* XXX */
66 * mfs vnode operations. Note: the vops here are used for the MFS block
67 * device, not for operations on files (MFS calls the ffs mount code for that)
69 struct vop_ops *mfs_vnode_vops;
70 static struct vnodeopv_entry_desc mfs_vnodeop_entries[] = {
71 { &vop_default_desc, (vnodeopv_entry_t) mfs_badop },
72 { &vop_bmap_desc, (vnodeopv_entry_t) mfs_bmap },
73 { &vop_bwrite_desc, vop_defaultop },
74 { &vop_close_desc, (vnodeopv_entry_t) mfs_close },
75 { &vop_createvobject_desc, (vnodeopv_entry_t) vop_stdcreatevobject },
76 { &vop_destroyvobject_desc, (vnodeopv_entry_t) vop_stddestroyvobject },
77 { &vop_freeblks_desc, (vnodeopv_entry_t) mfs_freeblks },
78 { &vop_fsync_desc, (vnodeopv_entry_t) mfs_fsync },
79 { &vop_getpages_desc, (vnodeopv_entry_t) mfs_getpages },
80 { &vop_getvobject_desc, (vnodeopv_entry_t) vop_stdgetvobject },
81 { &vop_inactive_desc, (vnodeopv_entry_t) mfs_inactive },
82 { &vop_ioctl_desc, vop_enotty },
83 { &vop_islocked_desc, vop_defaultop },
84 { &vop_lock_desc, vop_defaultop },
85 { &vop_open_desc, (vnodeopv_entry_t) mfs_open },
86 { &vop_print_desc, (vnodeopv_entry_t) mfs_print },
87 { &vop_reclaim_desc, (vnodeopv_entry_t) mfs_reclaim },
88 { &vop_strategy_desc, (vnodeopv_entry_t) mfs_strategy },
89 { &vop_unlock_desc, vop_defaultop },
92 static struct vnodeopv_desc mfs_vnodeop_opv_desc =
93 { &mfs_vnode_vops, mfs_vnodeop_entries };
95 VNODEOP_SET(mfs_vnodeop_opv_desc);
100 * Open called to allow memory filesystem to initialize and
101 * validate before actual IO. Record our process identifier
102 * so we can tell when we are doing I/O to ourself.
104 * NOTE: new device sequencing. mounts check the device reference count
105 * before calling open, so we must associate the device in open and
106 * disassociate it in close rather then faking it when we created the vnode.
108 * mfs_open(struct vnode *a_vp, int a_mode, struct ucred *a_cred,
109 * struct thread *a_td)
113 mfs_open(struct vop_open_args *ap)
115 struct vnode *vp = ap->a_vp;
117 if (vp->v_type != VCHR)
118 panic("mfs_open not VCHR");
119 v_associate_rdev(vp, udev2dev(vp->v_udev, 0));
124 mfs_fsync(struct vop_fsync_args *ap)
126 return (VOCALL(spec_vnode_vops, &ap->a_head));
130 * mfs_freeblks() - hook to allow us to free physical memory.
132 * We implement the B_FREEBUF strategy. We can't just madvise()
133 * here because we have to do it in the correct order vs other bio
134 * requests, so we queue it.
136 * Note: geteblk() sets B_INVAL. We leave it set to guarentee buffer
137 * throw-away on brelse()? XXX
139 * mfs_freeblks(struct vnode *a_vp, daddr_t a_addr, daddr_t a_length)
142 mfs_freeblks(struct vop_freeblks_args *ap)
145 struct vnode *vp = ap->a_vp;
147 bp = geteblk(ap->a_length);
148 bp->b_flags |= B_FREEBUF | B_ASYNC;
149 bp->b_dev = vp->v_rdev;
150 bp->b_blkno = ap->a_addr;
151 bp->b_offset = dbtob(ap->a_addr);
152 bp->b_bcount = ap->a_length;
154 VOP_STRATEGY(vp, bp);
159 * Pass I/O requests to the memory filesystem process.
161 * mfs_strategy(struct vnode *a_vp, struct buf *a_bp)
164 mfs_strategy(struct vop_strategy_args *ap)
166 struct buf *bp = ap->a_bp;
167 struct mfsnode *mfsp;
168 struct thread *td = curthread; /* XXX */
171 bp->b_dev = ap->a_vp->v_rdev;
172 mfsp = bp->b_dev->si_drv1;
175 bp->b_flags |= B_ERROR;
181 * splbio required for queueing/dequeueing, in case of forwarded
182 * BPs from bio interrupts (?). It may not be necessary.
187 if (mfsp->mfs_td == NULL) {
189 * mini-root. Note: B_FREEBUF not supported at the moment,
190 * I'm not sure what kind of dataspace b_data is in.
194 base = mfsp->mfs_baseoff + (bp->b_blkno << DEV_BSHIFT);
195 if (bp->b_flags & B_FREEBUF)
197 if (bp->b_flags & B_READ)
198 bcopy(base, bp->b_data, bp->b_bcount);
200 bcopy(bp->b_data, base, bp->b_bcount);
202 } else if (mfsp->mfs_td == td) {
211 * VOP from some other process, queue to MFS process and
214 bufq_insert_tail(&mfsp->buf_queue, bp);
215 wakeup((caddr_t)mfsp);
222 * Memory file system I/O.
224 * Trivial on the HP since buffer has already been mapping into KVA space.
226 * Read and Write are handled with a simple copyin and copyout.
228 * We also partially support VOP_FREEBLKS() via B_FREEBUF. We can't implement
229 * completely -- for example, on fragments or inode metadata, but we can
230 * implement it for page-aligned requests.
233 mfs_doio(struct buf *bp, struct mfsnode *mfsp)
235 caddr_t base = mfsp->mfs_baseoff + (bp->b_blkno << DEV_BSHIFT);
237 if (bp->b_flags & B_FREEBUF) {
239 * Implement B_FREEBUF, which allows the filesystem to tell
240 * a block device when blocks are no longer needed (like when
241 * a file is deleted). We use the hook to MADV_FREE the VM.
242 * This makes an MFS filesystem work as well or better then
243 * a sun-style swap-mounted filesystem.
245 int bytes = bp->b_bcount;
247 if ((vm_offset_t)base & PAGE_MASK) {
248 int n = PAGE_SIZE - ((vm_offset_t)base & PAGE_MASK);
253 struct madvise_args uap;
257 bzero(&uap, sizeof(uap));
260 uap.behav = MADV_FREE;
265 } else if (bp->b_flags & B_READ) {
267 * Read data from our 'memory' disk
269 bp->b_error = copyin(base, bp->b_data, bp->b_bcount);
272 * Write data to our 'memory' disk
274 bp->b_error = copyout(bp->b_data, base, bp->b_bcount);
277 bp->b_flags |= B_ERROR;
282 * This is a noop, simply returning what one has been given.
284 * mfs_bmap(struct vnode *a_vp, ufs_daddr_t a_bn, struct vnode **a_vpp,
285 * ufs_daddr_t *a_bnp, int *a_runp)
288 mfs_bmap(struct vop_bmap_args *ap)
290 if (ap->a_vpp != NULL)
291 *ap->a_vpp = ap->a_vp;
292 if (ap->a_bnp != NULL)
293 *ap->a_bnp = ap->a_bn;
294 if (ap->a_runp != NULL)
300 * Memory filesystem close routine
302 * mfs_close(struct vnode *a_vp, int a_fflag, struct ucred *a_cred,
303 * struct thread *a_td)
307 mfs_close(struct vop_close_args *ap)
309 struct vnode *vp = ap->a_vp;
310 struct mfsnode *mfsp = VTOMFS(vp);
315 * Finish any pending I/O requests.
317 while ((bp = bufq_first(&mfsp->buf_queue)) != NULL) {
318 bufq_remove(&mfsp->buf_queue, bp);
323 * On last close of a memory filesystem
324 * we must invalidate any in core blocks, so that
325 * we can, free up its vnode.
327 if ((error = vinvalbuf(vp, 1, ap->a_td, 0, 0)) != 0)
330 * There should be no way to have any more uses of this
331 * vnode, so if we find any other uses, it is a panic.
333 if (vp->v_usecount > 1)
334 printf("mfs_close: ref count %d > 1\n", vp->v_usecount);
335 if (vp->v_usecount > 1 || (bufq_first(&mfsp->buf_queue) != NULL))
338 * Send a request to the filesystem server to exit.
340 mfsp->mfs_active = 0;
343 destroy_dev(mfsp->mfs_dev);
344 mfsp->mfs_dev = NULL;
346 wakeup((caddr_t)mfsp);
351 * Memory filesystem inactive routine
353 * mfs_inactive(struct vnode *a_vp, struct thread *a_td)
357 mfs_inactive(struct vop_inactive_args *ap)
359 struct vnode *vp = ap->a_vp;
360 struct mfsnode *mfsp = VTOMFS(vp);
362 if (bufq_first(&mfsp->buf_queue) != NULL)
363 panic("mfs_inactive: not inactive (next buffer %p)",
364 bufq_first(&mfsp->buf_queue));
369 * Reclaim a memory filesystem devvp so that it can be reused.
371 * mfs_reclaim(struct vnode *a_vp)
374 mfs_reclaim(struct vop_reclaim_args *ap)
376 struct vnode *vp = ap->a_vp;
378 FREE(vp->v_data, M_MFSNODE);
384 * Print out the contents of an mfsnode.
386 * mfs_print(struct vnode *a_vp)
389 mfs_print(struct vop_print_args *ap)
391 struct mfsnode *mfsp = VTOMFS(ap->a_vp);
393 printf("tag VT_MFS, td %p, base %p, size %ld\n",
394 mfsp->mfs_td, (void *)mfsp->mfs_baseoff, mfsp->mfs_size);
399 * Block device bad operation
402 mfs_badop(struct vop_generic_args *ap)
406 printf("mfs_badop[%s]\n", ap->a_desc->vdesc_name);
407 i = vop_defaultop(ap);
408 printf("mfs_badop[%s] = %d\n", ap->a_desc->vdesc_name, i);
413 mfs_getpages(struct vop_getpages_args *ap)
415 return (VOCALL(spec_vnode_vops, &ap->a_head));