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 1.21 2006/03/24 18:35:34 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>
51 #include <sys/thread2.h>
54 #include "mfs_extern.h"
56 static int mfs_badop (struct vop_generic_args *);
57 static int mfs_bmap (struct vop_bmap_args *);
58 static int mfs_close (struct vop_close_args *);
59 static int mfs_fsync (struct vop_fsync_args *);
60 static int mfs_freeblks (struct vop_freeblks_args *);
61 static int mfs_inactive (struct vop_inactive_args *); /* XXX */
62 static int mfs_open (struct vop_open_args *);
63 static int mfs_reclaim (struct vop_reclaim_args *); /* XXX */
64 static int mfs_print (struct vop_print_args *); /* XXX */
65 static int mfs_strategy (struct vop_strategy_args *); /* XXX */
66 static int mfs_getpages (struct vop_getpages_args *); /* XXX */
68 * mfs vnode operations. Note: the vops here are used for the MFS block
69 * device, not for operations on files (MFS calls the ffs mount code for that)
71 struct vop_ops *mfs_vnode_vops;
72 static struct vnodeopv_entry_desc mfs_vnodeop_entries[] = {
73 { &vop_default_desc, (vnodeopv_entry_t) mfs_badop },
74 { &vop_bmap_desc, (vnodeopv_entry_t) mfs_bmap },
75 { &vop_bwrite_desc, vop_defaultop },
76 { &vop_close_desc, (vnodeopv_entry_t) mfs_close },
77 { &vop_createvobject_desc, (vnodeopv_entry_t) vop_stdcreatevobject },
78 { &vop_destroyvobject_desc, (vnodeopv_entry_t) vop_stddestroyvobject },
79 { &vop_freeblks_desc, (vnodeopv_entry_t) mfs_freeblks },
80 { &vop_fsync_desc, (vnodeopv_entry_t) mfs_fsync },
81 { &vop_getpages_desc, (vnodeopv_entry_t) mfs_getpages },
82 { &vop_getvobject_desc, (vnodeopv_entry_t) vop_stdgetvobject },
83 { &vop_inactive_desc, (vnodeopv_entry_t) mfs_inactive },
84 { &vop_ioctl_desc, vop_enotty },
85 { &vop_islocked_desc, vop_defaultop },
86 { &vop_lock_desc, vop_defaultop },
87 { &vop_open_desc, (vnodeopv_entry_t) mfs_open },
88 { &vop_print_desc, (vnodeopv_entry_t) mfs_print },
89 { &vop_reclaim_desc, (vnodeopv_entry_t) mfs_reclaim },
90 { &vop_strategy_desc, (vnodeopv_entry_t) mfs_strategy },
91 { &vop_unlock_desc, vop_defaultop },
94 static struct vnodeopv_desc mfs_vnodeop_opv_desc =
95 { &mfs_vnode_vops, mfs_vnodeop_entries, 0 };
97 VNODEOP_SET(mfs_vnodeop_opv_desc);
102 * Open called to allow memory filesystem to initialize and
103 * validate before actual IO. Record our process identifier
104 * so we can tell when we are doing I/O to ourself.
106 * NOTE: new device sequencing. mounts check the device reference count
107 * before calling open, so we must associate the device in open and
108 * disassociate it in close rather then faking it when we created the vnode.
110 * mfs_open(struct vnode *a_vp, int a_mode, struct ucred *a_cred,
111 * struct thread *a_td)
115 mfs_open(struct vop_open_args *ap)
117 struct vnode *vp = ap->a_vp;
119 if (vp->v_type != VCHR)
120 panic("mfs_open not VCHR");
121 v_associate_rdev(vp, udev2dev(vp->v_udev, 0));
126 mfs_fsync(struct vop_fsync_args *ap)
128 return (VOCALL(spec_vnode_vops, &ap->a_head));
132 * mfs_freeblks() - hook to allow us to free physical memory.
134 * We implement the B_FREEBUF strategy. We can't just madvise()
135 * here because we have to do it in the correct order vs other bio
136 * requests, so we queue it.
138 * Note: geteblk() sets B_INVAL. We leave it set to guarentee buffer
139 * throw-away on brelse()? XXX
141 * mfs_freeblks(struct vnode *a_vp, daddr_t a_addr, daddr_t a_length)
144 mfs_freeblks(struct vop_freeblks_args *ap)
147 struct vnode *vp = ap->a_vp;
149 bp = geteblk(ap->a_length);
150 bp->b_flags |= B_FREEBUF | B_ASYNC;
151 bp->b_bio1.bio_offset = ap->a_offset;
152 bp->b_bcount = ap->a_length;
154 vn_strategy(vp, &bp->b_bio1);
159 * Pass I/O requests to the memory filesystem process.
161 * mfs_strategy(struct vnode *a_vp, struct bio *a_bio)
164 mfs_strategy(struct vop_strategy_args *ap)
166 struct bio *bio = ap->a_bio;
167 struct buf *bp = bio->bio_buf;
168 struct mfsnode *mfsp;
169 struct thread *td = curthread; /* XXX */
171 mfsp = ap->a_vp->v_rdev->si_drv1;
174 bp->b_flags |= B_ERROR;
180 * splbio required for queueing/dequeueing, in case of forwarded
181 * BPs from bio interrupts (?). It may not be necessary.
186 if (mfsp->mfs_td == NULL) {
188 * mini-root. Note: B_FREEBUF not supported at the moment,
189 * I'm not sure what kind of dataspace b_data is in.
193 base = mfsp->mfs_baseoff + bio->bio_offset;
194 if (bp->b_flags & B_FREEBUF)
196 else if (bp->b_flags & B_READ)
197 bcopy(base, bp->b_data, bp->b_bcount);
199 bcopy(bp->b_data, base, bp->b_bcount);
201 } else if (mfsp->mfs_td == td) {
210 * VOP from some other process, queue to MFS process and
213 bioq_insert_tail(&mfsp->bio_queue, bio);
214 wakeup((caddr_t)mfsp);
221 * Memory file system I/O.
223 * Trivial on the HP since buffer has already been mapping into KVA space.
225 * Read and Write are handled with a simple copyin and copyout.
227 * We also partially support VOP_FREEBLKS() via B_FREEBUF. We can't implement
228 * completely -- for example, on fragments or inode metadata, but we can
229 * implement it for page-aligned requests.
232 mfs_doio(struct bio *bio, struct mfsnode *mfsp)
234 struct buf *bp = bio->bio_buf;
235 caddr_t base = mfsp->mfs_baseoff + bio->bio_offset;
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, off_t a_loffset, struct vnode **a_vpp,
285 * off_t *a_doffsetp, int *a_runp, int *a_runb)
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_doffsetp != NULL)
293 *ap->a_doffsetp = ap->a_loffset;
294 if (ap->a_runp != NULL)
296 if (ap->a_runb != NULL)
302 * Memory filesystem close routine
304 * mfs_close(struct vnode *a_vp, int a_fflag, struct ucred *a_cred,
305 * struct thread *a_td)
309 mfs_close(struct vop_close_args *ap)
311 struct vnode *vp = ap->a_vp;
312 struct mfsnode *mfsp = VTOMFS(vp);
317 * Finish any pending I/O requests.
319 while ((bio = bioq_first(&mfsp->bio_queue)) != NULL) {
320 bioq_remove(&mfsp->bio_queue, bio);
322 wakeup((caddr_t)bio->bio_buf);
325 * On last close of a memory filesystem
326 * we must invalidate any in core blocks, so that
327 * we can, free up its vnode.
329 if ((error = vinvalbuf(vp, 1, ap->a_td, 0, 0)) != 0)
332 * There should be no way to have any more uses of this
333 * vnode, so if we find any other uses, it is a panic.
335 if (vp->v_usecount > 1)
336 printf("mfs_close: ref count %d > 1\n", vp->v_usecount);
337 if (vp->v_usecount > 1 || (bioq_first(&mfsp->bio_queue) != NULL))
340 * Send a request to the filesystem server to exit.
342 mfsp->mfs_active = 0;
345 destroy_dev(mfsp->mfs_dev);
346 mfsp->mfs_dev = NULL;
348 wakeup((caddr_t)mfsp);
353 * Memory filesystem inactive routine
355 * mfs_inactive(struct vnode *a_vp, struct thread *a_td)
359 mfs_inactive(struct vop_inactive_args *ap)
361 struct vnode *vp = ap->a_vp;
362 struct mfsnode *mfsp = VTOMFS(vp);
364 if (bioq_first(&mfsp->bio_queue) != NULL)
365 panic("mfs_inactive: not inactive (next buffer %p)",
366 bioq_first(&mfsp->bio_queue));
371 * Reclaim a memory filesystem devvp so that it can be reused.
373 * mfs_reclaim(struct vnode *a_vp)
376 mfs_reclaim(struct vop_reclaim_args *ap)
378 struct vnode *vp = ap->a_vp;
380 FREE(vp->v_data, M_MFSNODE);
386 * Print out the contents of an mfsnode.
388 * mfs_print(struct vnode *a_vp)
391 mfs_print(struct vop_print_args *ap)
393 struct mfsnode *mfsp = VTOMFS(ap->a_vp);
395 printf("tag VT_MFS, td %p, base %p, size %ld\n",
396 mfsp->mfs_td, (void *)mfsp->mfs_baseoff, mfsp->mfs_size);
401 * Block device bad operation
404 mfs_badop(struct vop_generic_args *ap)
408 printf("mfs_badop[%s]\n", ap->a_desc->vdesc_name);
409 i = vop_defaultop(ap);
410 printf("mfs_badop[%s] = %d\n", ap->a_desc->vdesc_name, i);
415 mfs_getpages(struct vop_getpages_args *ap)
417 return (VOCALL(spec_vnode_vops, &ap->a_head));