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 1.24 2006/04/02 01:35:34 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>
50 #include <vm/vm_object.h>
51 #include <vm/vm_page.h>
52 #include <vm/vm_pager.h>
53 #include <vm/vnode_pager.h>
56 #include <sys/thread2.h>
59 #include "mfs_extern.h"
61 static int mfs_badop (struct vop_generic_args *);
62 static int mfs_bmap (struct vop_bmap_args *);
63 static int mfs_close (struct vop_close_args *);
64 static int mfs_fsync (struct vop_fsync_args *);
65 static int mfs_freeblks (struct vop_freeblks_args *);
66 static int mfs_inactive (struct vop_inactive_args *); /* XXX */
67 static int mfs_open (struct vop_open_args *);
68 static int mfs_reclaim (struct vop_reclaim_args *); /* XXX */
69 static int mfs_print (struct vop_print_args *); /* XXX */
70 static int mfs_strategy (struct vop_strategy_args *); /* XXX */
71 static int mfs_getpages (struct vop_getpages_args *); /* XXX */
73 * mfs vnode operations. Note: the vops here are used for the MFS block
74 * device, not for operations on files (MFS calls the ffs mount code for that)
76 struct vop_ops *mfs_vnode_vops;
77 static struct vnodeopv_entry_desc mfs_vnodeop_entries[] = {
78 { &vop_default_desc, (vnodeopv_entry_t) mfs_badop },
79 { &vop_bmap_desc, (vnodeopv_entry_t) mfs_bmap },
80 { &vop_bwrite_desc, vop_defaultop },
81 { &vop_close_desc, (vnodeopv_entry_t) mfs_close },
82 { &vop_freeblks_desc, (vnodeopv_entry_t) mfs_freeblks },
83 { &vop_fsync_desc, (vnodeopv_entry_t) mfs_fsync },
84 { &vop_getpages_desc, (vnodeopv_entry_t) mfs_getpages },
85 { &vop_inactive_desc, (vnodeopv_entry_t) mfs_inactive },
86 { &vop_ioctl_desc, vop_enotty },
87 { &vop_islocked_desc, vop_defaultop },
88 { &vop_lock_desc, vop_defaultop },
89 { &vop_open_desc, (vnodeopv_entry_t) mfs_open },
90 { &vop_print_desc, (vnodeopv_entry_t) mfs_print },
91 { &vop_reclaim_desc, (vnodeopv_entry_t) mfs_reclaim },
92 { &vop_strategy_desc, (vnodeopv_entry_t) mfs_strategy },
93 { &vop_unlock_desc, vop_defaultop },
96 static struct vnodeopv_desc mfs_vnodeop_opv_desc =
97 { &mfs_vnode_vops, mfs_vnodeop_entries, 0 };
99 VNODEOP_SET(mfs_vnodeop_opv_desc);
104 * Open called to allow memory filesystem to initialize and
105 * validate before actual IO. Record our process identifier
106 * so we can tell when we are doing I/O to ourself.
108 * NOTE: new device sequencing. mounts check the device reference count
109 * before calling open, so we must associate the device in open and
110 * disassociate it in close rather then faking it when we created the vnode.
112 * mfs_open(struct vnode *a_vp, int a_mode, struct ucred *a_cred,
113 * struct thread *a_td)
117 mfs_open(struct vop_open_args *ap)
119 struct vnode *vp = ap->a_vp;
121 if (vp->v_type != VCHR)
122 panic("mfs_open not VCHR");
123 v_associate_rdev(vp, udev2dev(vp->v_udev, 0));
124 return (vop_stdopen(ap));
128 mfs_fsync(struct vop_fsync_args *ap)
130 return (VOCALL(spec_vnode_vops, &ap->a_head));
134 * mfs_freeblks() - hook to allow us to free physical memory.
136 * We implement the B_FREEBUF strategy. We can't just madvise()
137 * here because we have to do it in the correct order vs other bio
138 * requests, so we queue it.
140 * Note: geteblk() sets B_INVAL. We leave it set to guarentee buffer
141 * throw-away on brelse()? XXX
143 * mfs_freeblks(struct vnode *a_vp, daddr_t a_addr, daddr_t a_length)
146 mfs_freeblks(struct vop_freeblks_args *ap)
149 struct vnode *vp = ap->a_vp;
151 bp = geteblk(ap->a_length);
152 bp->b_flags |= B_FREEBUF | B_ASYNC;
153 bp->b_bio1.bio_offset = ap->a_offset;
154 bp->b_bcount = ap->a_length;
156 vn_strategy(vp, &bp->b_bio1);
161 * Pass I/O requests to the memory filesystem process.
163 * mfs_strategy(struct vnode *a_vp, struct bio *a_bio)
166 mfs_strategy(struct vop_strategy_args *ap)
168 struct bio *bio = ap->a_bio;
169 struct buf *bp = bio->bio_buf;
170 struct mfsnode *mfsp;
171 struct thread *td = curthread; /* XXX */
173 mfsp = ap->a_vp->v_rdev->si_drv1;
176 bp->b_flags |= B_ERROR;
182 * splbio required for queueing/dequeueing, in case of forwarded
183 * BPs from bio interrupts (?). It may not be necessary.
188 if (mfsp->mfs_td == NULL) {
190 * mini-root. Note: B_FREEBUF not supported at the moment,
191 * I'm not sure what kind of dataspace b_data is in.
195 base = mfsp->mfs_baseoff + bio->bio_offset;
196 if (bp->b_flags & B_FREEBUF)
198 else if (bp->b_flags & B_READ)
199 bcopy(base, bp->b_data, bp->b_bcount);
201 bcopy(bp->b_data, base, bp->b_bcount);
203 } else if (mfsp->mfs_td == td) {
212 * VOP from some other process, queue to MFS process and
215 bioq_insert_tail(&mfsp->bio_queue, bio);
216 wakeup((caddr_t)mfsp);
223 * Memory file system I/O.
225 * Trivial on the HP since buffer has already been mapping into KVA space.
227 * Read and Write are handled with a simple copyin and copyout.
229 * We also partially support VOP_FREEBLKS() via B_FREEBUF. We can't implement
230 * completely -- for example, on fragments or inode metadata, but we can
231 * implement it for page-aligned requests.
234 mfs_doio(struct bio *bio, struct mfsnode *mfsp)
236 struct buf *bp = bio->bio_buf;
237 caddr_t base = mfsp->mfs_baseoff + bio->bio_offset;
239 if (bp->b_flags & B_FREEBUF) {
241 * Implement B_FREEBUF, which allows the filesystem to tell
242 * a block device when blocks are no longer needed (like when
243 * a file is deleted). We use the hook to MADV_FREE the VM.
244 * This makes an MFS filesystem work as well or better then
245 * a sun-style swap-mounted filesystem.
247 int bytes = bp->b_bcount;
249 if ((vm_offset_t)base & PAGE_MASK) {
250 int n = PAGE_SIZE - ((vm_offset_t)base & PAGE_MASK);
255 struct madvise_args uap;
259 bzero(&uap, sizeof(uap));
262 uap.behav = MADV_FREE;
267 } else if (bp->b_flags & B_READ) {
269 * Read data from our 'memory' disk
271 bp->b_error = copyin(base, bp->b_data, bp->b_bcount);
274 * Write data to our 'memory' disk
276 bp->b_error = copyout(bp->b_data, base, bp->b_bcount);
279 bp->b_flags |= B_ERROR;
284 * This is a noop, simply returning what one has been given.
286 * mfs_bmap(struct vnode *a_vp, off_t a_loffset, struct vnode **a_vpp,
287 * off_t *a_doffsetp, int *a_runp, int *a_runb)
290 mfs_bmap(struct vop_bmap_args *ap)
292 if (ap->a_vpp != NULL)
293 *ap->a_vpp = ap->a_vp;
294 if (ap->a_doffsetp != NULL)
295 *ap->a_doffsetp = ap->a_loffset;
296 if (ap->a_runp != NULL)
298 if (ap->a_runb != NULL)
304 * Memory filesystem close routine
306 * mfs_close(struct vnode *a_vp, int a_fflag, struct ucred *a_cred,
307 * struct thread *a_td)
311 mfs_close(struct vop_close_args *ap)
313 struct vnode *vp = ap->a_vp;
314 struct mfsnode *mfsp = VTOMFS(vp);
319 * Finish any pending I/O requests.
321 while ((bio = bioq_first(&mfsp->bio_queue)) != NULL) {
322 bioq_remove(&mfsp->bio_queue, bio);
324 wakeup((caddr_t)bio->bio_buf);
328 * We really only care about the last close
330 if (vp->v_opencount > 1)
334 * Synchronize any remaining buffers and then destroy them.
336 if ((error = vinvalbuf(vp, V_SAVE, ap->a_td, 0, 0)) != 0)
340 * Get rid of the pseudo-backing object. Since the object is
341 * not directly memory mapped, we don't have to worry about
345 vm_pager_deallocate(vp->v_object);
348 * There should be no way to have any more uses of this
349 * vnode, so if we find any other uses, it is a panic.
351 if (vp->v_usecount > 1)
352 printf("mfs_close: ref count %d > 1\n", vp->v_usecount);
353 if (vp->v_usecount > 1 || (bioq_first(&mfsp->bio_queue) != NULL))
356 * Send a request to the filesystem server to exit.
358 mfsp->mfs_active = 0;
361 destroy_dev(mfsp->mfs_dev);
362 mfsp->mfs_dev = NULL;
364 wakeup((caddr_t)mfsp);
371 * Memory filesystem inactive routine
373 * mfs_inactive(struct vnode *a_vp, struct thread *a_td)
377 mfs_inactive(struct vop_inactive_args *ap)
379 struct vnode *vp = ap->a_vp;
380 struct mfsnode *mfsp = VTOMFS(vp);
382 if (bioq_first(&mfsp->bio_queue) != NULL)
383 panic("mfs_inactive: not inactive (next buffer %p)",
384 bioq_first(&mfsp->bio_queue));
389 * Reclaim a memory filesystem devvp so that it can be reused.
391 * mfs_reclaim(struct vnode *a_vp)
394 mfs_reclaim(struct vop_reclaim_args *ap)
396 struct vnode *vp = ap->a_vp;
398 FREE(vp->v_data, M_MFSNODE);
404 * Print out the contents of an mfsnode.
406 * mfs_print(struct vnode *a_vp)
409 mfs_print(struct vop_print_args *ap)
411 struct mfsnode *mfsp = VTOMFS(ap->a_vp);
413 printf("tag VT_MFS, td %p, base %p, size %ld\n",
414 mfsp->mfs_td, (void *)mfsp->mfs_baseoff, mfsp->mfs_size);
419 * Block device bad operation
422 mfs_badop(struct vop_generic_args *ap)
426 printf("mfs_badop[%s]\n", ap->a_desc->vdesc_name);
427 i = vop_defaultop(ap);
428 printf("mfs_badop[%s] = %d\n", ap->a_desc->vdesc_name, i);
433 mfs_getpages(struct vop_getpages_args *ap)
435 return (VOCALL(spec_vnode_vops, &ap->a_head));