2 * Copyright (c) 2000-2004 Apple Computer, Inc. All rights reserved.
4 * @APPLE_OSREFERENCE_LICENSE_HEADER_START@
6 * This file contains Original Code and/or Modifications of Original Code
7 * as defined in and that are subject to the Apple Public Source License
8 * Version 2.0 (the 'License'). You may not use this file except in
9 * compliance with the License. The rights granted to you under the License
10 * may not be used to create, or enable the creation or redistribution of,
11 * unlawful or unlicensed copies of an Apple operating system, or to
12 * circumvent, violate, or enable the circumvention or violation of, any
13 * terms of an Apple operating system software license agreement.
15 * Please obtain a copy of the License at
16 * http://www.opensource.apple.com/apsl/ and read it before using this file.
18 * The Original Code and all software distributed under the License are
19 * distributed on an 'AS IS' basis, WITHOUT WARRANTY OF ANY KIND, EITHER
20 * EXPRESS OR IMPLIED, AND APPLE HEREBY DISCLAIMS ALL SUCH WARRANTIES,
21 * INCLUDING WITHOUT LIMITATION, ANY WARRANTIES OF MERCHANTABILITY,
22 * FITNESS FOR A PARTICULAR PURPOSE, QUIET ENJOYMENT OR NON-INFRINGEMENT.
23 * Please see the License for the specific language governing rights and
24 * limitations under the License.
26 * @APPLE_OSREFERENCE_LICENSE_HEADER_END@
30 * Copyright (c) 1988 University of Utah.
31 * Copyright (c) 1990, 1993
32 * The Regents of the University of California. All rights reserved.
34 * This code is derived from software contributed to Berkeley by
35 * the Systems Programming Group of the University of Utah Computer
38 * Redistribution and use in source and binary forms, with or without
39 * modification, are permitted provided that the following conditions
41 * 1. Redistributions of source code must retain the above copyright
42 * notice, this list of conditions and the following disclaimer.
43 * 2. Redistributions in binary form must reproduce the above copyright
44 * notice, this list of conditions and the following disclaimer in the
45 * documentation and/or other materials provided with the distribution.
46 * 3. All advertising materials mentioning features or use of this software
47 * must display the following acknowledgement:
48 * This product includes software developed by the University of
49 * California, Berkeley and its contributors.
50 * 4. Neither the name of the University nor the names of its contributors
51 * may be used to endorse or promote products derived from this software
52 * without specific prior written permission.
54 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
55 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
56 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
57 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
58 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
59 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
60 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
61 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
62 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
63 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
66 * from: Utah Hdr: vn.c 1.13 94/04/02
68 * from: @(#)vn.c 8.6 (Berkeley) 4/1/94
69 * $FreeBSD: src/sys/dev/vn/vn.c,v 1.105.2.4 2001/11/18 07:11:00 dillon Exp $
75 * Block/character interface to a vnode. Allows one to treat a file
76 * as a disk (e.g. build a filesystem in it, mount it, etc.).
78 * NOTE 1: This uses the vnop_blockmap/vnop_strategy interface to the vnode
79 * instead of a simple VOP_RDWR. We do this to avoid distorting the
82 * NOTE 2: There is a security issue involved with this driver.
83 * Once mounted all access to the contents of the "mapped" file via
84 * the special file is controlled by the permissions on the special
85 * file, the protection of the mapped file is ignored (effectively,
86 * by using root credentials in all transactions).
88 * NOTE 3: Doesn't interact with leases, should it?
95 #include <sys/param.h>
96 #include <sys/systm.h>
97 #include <sys/kernel.h>
98 #include <sys/mount.h>
99 #include <sys/namei.h>
100 #include <sys/proc.h>
101 #include <sys/kauth.h>
103 #include <sys/malloc.h>
104 #include <sys/vnode_internal.h>
105 #include <sys/fcntl.h>
106 #include <sys/conf.h>
107 #include <sys/disk.h>
108 #include <sys/stat.h>
109 #include <sys/conf.h>
110 #include <sys/uio_internal.h>
112 #include <sys/vnioctl.h>
116 #include <vm/vm_pager.h>
117 #include <mach/memory_object_types.h>
119 #include <miscfs/devfs/devfs.h>
124 static ioctl_fcn_t vnioctl_chr
;
125 static ioctl_fcn_t vnioctl_blk
;
126 static open_close_fcn_t vnopen
;
127 static open_close_fcn_t vnclose
;
128 static psize_fcn_t vnsize
;
129 static strategy_fcn_t vnstrategy
;
130 static read_write_fcn_t vnread
;
131 static read_write_fcn_t vnwrite
;
133 static int vndevice_bdev_major
;
134 static int vndevice_cdev_major
;
138 * D_DISK we want to look like a disk
139 * D_CANFREE We support B_FREEBUF
142 static struct bdevsw vn_bdevsw
= {
145 /* strategy */ vnstrategy
,
146 /* ioctl */ vnioctl_blk
,
152 static struct cdevsw vn_cdevsw
= {
157 /* ioctl */ vnioctl_chr
,
159 /* reset */ eno_reset
,
161 /* select */ eno_select
,
163 /* strategy */ eno_strat
,
170 u_int64_t sc_fsize
; /* file size in bytes */
171 u_int64_t sc_size
; /* size of vn, sc_secsize scale */
172 int sc_flags
; /* flags */
173 u_long sc_secsize
; /* sector size */
174 struct vnode
*sc_vp
; /* vnode if not NULL */
177 struct vnode
*sc_shadow_vp
; /* shadow vnode if not NULL */
178 uint32_t sc_shadow_vid
;
179 shadow_map_t
* sc_shadow_map
; /* shadow map if not NULL */
180 kauth_cred_t sc_cred
; /* credentials */
181 u_int32_t sc_options
; /* options */
184 } vn_table
[NVNDEVICE
];
186 #define ROOT_IMAGE_UNIT 0
189 #define VNF_INITED 0x01
190 #define VNF_READONLY 0x02
192 static u_int32_t vn_options
;
194 #define IFOPT(vn,opt) if (((vn)->sc_options|vn_options) & (opt))
195 #define TESTOPT(vn,opt) (((vn)->sc_options|vn_options) & (opt))
197 static int setcred(struct vnode
* vp
, struct proc
* p
,
199 static void vnclear (struct vn_softc
*vn
, struct proc
* p
);
200 static void vn_ioctl_to_64(struct vn_ioctl
*from
, struct user_vn_ioctl
*to
);
201 void vndevice_init(void);
202 int vndevice_root_image(char * path
, char devname
[], dev_t
* dev_p
);
205 vniocattach_file(struct vn_softc
*vn
,
206 struct user_vn_ioctl
*vniop
,
211 vniocattach_shadow(struct vn_softc
* vn
,
212 struct user_vn_ioctl
*vniop
,
216 static __inline__
int
223 vnclose(__unused dev_t dev
, __unused
int flags
,
224 __unused
int devtype
, __unused
struct proc
*p
)
230 vnopen(dev_t dev
, int flags
, __unused
int devtype
, __unused
struct proc
*p
)
236 if (vnunit(dev
) >= NVNDEVICE
) {
239 vn
= vn_table
+ unit
;
240 if ((flags
& FWRITE
) && (vn
->sc_flags
& VNF_READONLY
))
247 file_io(struct vnode
* vp
, struct vfs_context
* context_p
,
248 enum uio_rw op
, char * base
, off_t offset
, user_ssize_t count
,
249 user_ssize_t
* resid
)
253 char uio_buf
[UIO_SIZEOF(1)];
255 auio
= uio_createwithbuffer(1, offset
, UIO_SYSSPACE
, op
,
256 &uio_buf
[0], sizeof(uio_buf
));
257 uio_addiov(auio
, CAST_USER_ADDR_T(base
), count
);
259 error
= VNOP_READ(vp
, auio
, IO_SYNC
, context_p
);
261 error
= VNOP_WRITE(vp
, auio
, IO_SYNC
, context_p
);
264 *resid
= uio_resid(auio
);
269 static __inline__ off_t
270 block_round(off_t o
, int blocksize
)
272 return ((o
+ blocksize
- 1) / blocksize
);
275 static __inline__ off_t
276 block_truncate(off_t o
, int blocksize
)
278 return (o
/ blocksize
);
281 static __inline__
int
282 block_remainder(off_t o
, int blocksize
)
284 return (o
% blocksize
);
288 vnread_shadow(struct vn_softc
* vn
, struct uio
*uio
, int ioflag
,
289 struct vfs_context
* context_p
)
291 u_long blocksize
= vn
->sc_secsize
;
296 user_ssize_t orig_resid
;
298 orig_resid
= resid
= uio_resid(uio
);
299 orig_offset
= offset
= uio_offset(uio
);
303 u_long this_block_number
;
304 u_long this_block_count
;
306 user_ssize_t this_resid
;
309 /* figure out which blocks to read */
310 remainder
= block_remainder(offset
, blocksize
);
311 if (shadow_map_read(vn
->sc_shadow_map
,
312 block_truncate(offset
, blocksize
),
313 block_round(resid
+ remainder
, blocksize
),
314 &this_block_number
, &this_block_count
)) {
315 vp
= vn
->sc_shadow_vp
;
321 /* read the blocks (or parts thereof) */
322 this_offset
= (off_t
)this_block_number
* blocksize
+ remainder
;
323 uio_setoffset(uio
, this_offset
);
324 this_resid
= this_block_count
* blocksize
- remainder
;
325 if (this_resid
> resid
) {
328 uio_setresid(uio
, this_resid
);
329 error
= VNOP_READ(vp
, uio
, ioflag
, context_p
);
334 /* figure out how much we actually read */
335 this_resid
-= uio_resid(uio
);
336 if (this_resid
== 0) {
337 printf("vn device: vnread_shadow zero length read\n");
341 offset
+= this_resid
;
343 uio_setresid(uio
, resid
);
344 uio_setoffset(uio
, offset
);
349 vncopy_block_to_shadow(struct vn_softc
* vn
, struct vfs_context
* context_p
,
350 u_long file_block
, u_long shadow_block
)
355 tmpbuf
= _MALLOC(vn
->sc_secsize
, M_TEMP
, M_WAITOK
);
356 if (tmpbuf
== NULL
) {
359 /* read one block from file at file_block offset */
360 error
= file_io(vn
->sc_vp
, context_p
, UIO_READ
,
361 tmpbuf
, (off_t
)file_block
* vn
->sc_secsize
,
362 vn
->sc_secsize
, NULL
);
366 /* write one block to shadow file at shadow_block offset */
367 error
= file_io(vn
->sc_shadow_vp
, context_p
, UIO_WRITE
,
368 tmpbuf
, (off_t
)shadow_block
* vn
->sc_secsize
,
369 vn
->sc_secsize
, NULL
);
371 FREE(tmpbuf
, M_TEMP
);
376 FLAGS_FIRST_BLOCK_PARTIAL
= 0x1,
377 FLAGS_LAST_BLOCK_PARTIAL
= 0x2
381 vnwrite_shadow(struct vn_softc
* vn
, struct uio
*uio
, int ioflag
,
382 struct vfs_context
* context_p
)
384 u_long blocksize
= vn
->sc_secsize
;
389 resid
= uio_resid(uio
);
390 offset
= uio_offset(uio
);
394 u_long offset_block_number
;
396 u_long resid_block_count
;
397 u_long shadow_block_count
;
398 u_long shadow_block_number
;
399 user_ssize_t this_resid
;
401 /* figure out which blocks to write */
402 offset_block_number
= block_truncate(offset
, blocksize
);
403 remainder
= block_remainder(offset
, blocksize
);
404 resid_block_count
= block_round(resid
+ remainder
, blocksize
);
405 /* figure out if the first or last blocks are partial writes */
407 && !shadow_map_is_written(vn
->sc_shadow_map
,
408 offset_block_number
)) {
409 /* the first block is a partial write */
410 flags
|= FLAGS_FIRST_BLOCK_PARTIAL
;
412 if (resid_block_count
> 1
413 && !shadow_map_is_written(vn
->sc_shadow_map
,
415 + resid_block_count
- 1)
416 && block_remainder(offset
+ resid
, blocksize
) > 0) {
417 /* the last block is a partial write */
418 flags
|= FLAGS_LAST_BLOCK_PARTIAL
;
420 if (shadow_map_write(vn
->sc_shadow_map
,
421 offset_block_number
, resid_block_count
,
422 &shadow_block_number
,
423 &shadow_block_count
)) {
424 /* shadow file is growing */
426 /* truncate the file to its new length before write */
428 size
= (off_t
)shadow_map_shadow_size(vn
->sc_shadow_map
)
430 vnode_setsize(vn
->sc_shadow_vp
, size
, IO_SYNC
,
434 /* write the blocks (or parts thereof) */
435 uio_setoffset(uio
, (off_t
)
436 shadow_block_number
* blocksize
+ remainder
);
437 this_resid
= (off_t
)shadow_block_count
* blocksize
- remainder
;
438 if (this_resid
>= resid
) {
440 if ((flags
& FLAGS_LAST_BLOCK_PARTIAL
) != 0) {
441 /* copy the last block to the shadow */
445 s
= offset_block_number
446 + resid_block_count
- 1;
447 d
= shadow_block_number
448 + shadow_block_count
- 1;
449 error
= vncopy_block_to_shadow(vn
, context_p
,
452 printf("vnwrite_shadow: failed to copy"
453 " block %d to shadow block %d\n",
459 uio_setresid(uio
, this_resid
);
460 if ((flags
& FLAGS_FIRST_BLOCK_PARTIAL
) != 0) {
461 /* copy the first block to the shadow */
462 error
= vncopy_block_to_shadow(vn
, context_p
,
464 shadow_block_number
);
466 printf("vnwrite_shadow: failed to"
467 " copy block %d to shadow block %d\n",
469 shadow_block_number
);
473 error
= VNOP_WRITE(vn
->sc_shadow_vp
, uio
, ioflag
, context_p
);
477 /* figure out how much we actually wrote */
478 this_resid
-= uio_resid(uio
);
479 if (this_resid
== 0) {
480 printf("vn device: vnwrite_shadow zero length write\n");
484 offset
+= this_resid
;
486 uio_setresid(uio
, resid
);
487 uio_setoffset(uio
, offset
);
492 vnread(dev_t dev
, struct uio
*uio
, int ioflag
)
494 struct vfs_context context
;
496 boolean_t funnel_state
;
500 struct vn_softc
* vn
;
504 if (vnunit(dev
) >= NVNDEVICE
) {
508 funnel_state
= thread_funnel_set(kernel_flock
, TRUE
);
509 vn
= vn_table
+ unit
;
510 if ((vn
->sc_flags
& VNF_INITED
) == 0) {
514 error
= vnode_getwithvid(vn
->sc_vp
, vn
->sc_vid
);
516 /* the vnode is no longer available, abort */
522 resid
= uio_resid(uio
);
523 offset
= uio_offset(uio
);
526 * If out of bounds return an error. If at the EOF point,
529 if (offset
>= (off_t
)vn
->sc_fsize
) {
530 if (offset
> (off_t
)vn
->sc_fsize
) {
536 * If the request crosses EOF, truncate the request.
538 if ((offset
+ resid
) > (off_t
)vn
->sc_fsize
) {
539 resid
= vn
->sc_fsize
- offset
;
540 uio_setresid(uio
, resid
);
544 context
.vc_ucred
= vn
->sc_cred
;
545 if (vn
->sc_shadow_vp
!= NULL
) {
546 error
= vnode_getwithvid(vn
->sc_shadow_vp
,
549 /* the vnode is no longer available, abort */
551 vnode_put(vn
->sc_vp
);
555 error
= vnread_shadow(vn
, uio
, ioflag
, &context
);
556 vnode_put(vn
->sc_shadow_vp
);
558 error
= VNOP_READ(vn
->sc_vp
, uio
, ioflag
, &context
);
560 vnode_put(vn
->sc_vp
);
562 (void) thread_funnel_set(kernel_flock
, funnel_state
);
567 vnwrite(dev_t dev
, struct uio
*uio
, int ioflag
)
569 struct vfs_context context
;
571 boolean_t funnel_state
;
575 struct vn_softc
* vn
;
579 if (vnunit(dev
) >= NVNDEVICE
) {
583 funnel_state
= thread_funnel_set(kernel_flock
, TRUE
);
584 vn
= vn_table
+ unit
;
585 if ((vn
->sc_flags
& VNF_INITED
) == 0) {
589 if (vn
->sc_flags
& VNF_READONLY
) {
593 error
= vnode_getwithvid(vn
->sc_vp
, vn
->sc_vid
);
595 /* the vnode is no longer available, abort */
600 resid
= uio_resid(uio
);
601 offset
= uio_offset(uio
);
604 * If out of bounds return an error. If at the EOF point,
607 if (offset
>= (off_t
)vn
->sc_fsize
) {
608 if (offset
> (off_t
)vn
->sc_fsize
) {
614 * If the request crosses EOF, truncate the request.
616 if ((offset
+ resid
) > (off_t
)vn
->sc_fsize
) {
617 resid
= (off_t
)vn
->sc_fsize
- offset
;
618 uio_setresid(uio
, resid
);
622 context
.vc_ucred
= vn
->sc_cred
;
624 if (vn
->sc_shadow_vp
!= NULL
) {
625 error
= vnode_getwithvid(vn
->sc_shadow_vp
,
628 /* the vnode is no longer available, abort */
630 vnode_put(vn
->sc_vp
);
634 error
= vnwrite_shadow(vn
, uio
, ioflag
, &context
);
635 vnode_put(vn
->sc_shadow_vp
);
637 error
= VNOP_WRITE(vn
->sc_vp
, uio
, ioflag
, &context
);
639 vnode_put(vn
->sc_vp
);
641 (void) thread_funnel_set(kernel_flock
, funnel_state
);
646 shadow_read(struct vn_softc
* vn
, struct buf
* bp
, char * base
, struct proc
* p
)
648 u_long blocksize
= vn
->sc_secsize
;
649 struct vfs_context context
;
652 boolean_t read_shadow
;
657 context
.vc_ucred
= vn
->sc_cred
;
658 offset
= buf_blkno(bp
);
659 resid
= buf_resid(bp
) / blocksize
;
661 user_ssize_t temp_resid
;
666 read_shadow
= shadow_map_read(vn
->sc_shadow_map
,
668 &this_offset
, &this_resid
);
670 vp
= vn
->sc_shadow_vp
;
675 error
= file_io(vp
, &context
, UIO_READ
, base
+ start
,
676 (off_t
)this_offset
* blocksize
,
677 (user_ssize_t
)this_resid
* blocksize
,
682 this_resid
-= (temp_resid
/ blocksize
);
683 if (this_resid
== 0) {
684 printf("vn device: shadow_read zero length read\n");
688 offset
+= this_resid
;
689 start
+= this_resid
* blocksize
;
691 buf_setresid(bp
, resid
* blocksize
);
696 shadow_write(struct vn_softc
* vn
, struct buf
* bp
, char * base
,
699 u_long blocksize
= vn
->sc_secsize
;
700 struct vfs_context context
;
703 boolean_t shadow_grew
;
708 context
.vc_ucred
= vn
->sc_cred
;
709 offset
= buf_blkno(bp
);
710 resid
= buf_resid(bp
) / blocksize
;
712 user_ssize_t temp_resid
;
716 shadow_grew
= shadow_map_write(vn
->sc_shadow_map
,
718 &this_offset
, &this_resid
);
722 /* truncate the file to its new length before write */
723 size
= (off_t
)shadow_map_shadow_size(vn
->sc_shadow_map
)
725 vnode_setsize(vn
->sc_shadow_vp
, size
, IO_SYNC
,
729 error
= file_io(vn
->sc_shadow_vp
, &context
, UIO_WRITE
,
731 (off_t
)this_offset
* blocksize
,
732 (user_ssize_t
)this_resid
* blocksize
,
737 this_resid
-= (temp_resid
/ blocksize
);
738 if (this_resid
== 0) {
739 printf("vn device: shadow_write zero length write\n");
743 offset
+= this_resid
;
744 start
+= this_resid
* blocksize
;
746 buf_setresid(bp
, resid
* blocksize
);
751 vn_readwrite_io(struct vn_softc
* vn
, struct buf
* bp
, struct proc
* p
)
758 if (buf_map(bp
, &vaddr
))
759 panic("vn device: buf_map failed");
760 iov_base
= (char *)vaddr
;
762 if (vn
->sc_shadow_vp
== NULL
) {
763 struct vfs_context context
;
764 user_ssize_t temp_resid
;
767 context
.vc_ucred
= vn
->sc_cred
;
769 error
= file_io(vn
->sc_vp
, &context
,
770 buf_flags(bp
) & B_READ
? UIO_READ
: UIO_WRITE
,
772 (off_t
)buf_blkno(bp
) * vn
->sc_secsize
,
773 buf_resid(bp
), &temp_resid
);
774 buf_setresid(bp
, temp_resid
);
777 if (buf_flags(bp
) & B_READ
)
778 error
= shadow_read(vn
, bp
, iov_base
, p
);
780 error
= shadow_write(vn
, bp
, iov_base
, p
);
788 vnstrategy(struct buf
*bp
)
792 long sz
; /* in sc_secsize chunks */
794 boolean_t funnel_state
;
795 struct proc
* p
= current_proc();
796 struct vnode
* shadow_vp
= NULL
;
797 struct vnode
* vp
= NULL
;
799 funnel_state
= thread_funnel_set(kernel_flock
, TRUE
);
800 vn
= vn_table
+ vnunit(buf_device(bp
));
801 if ((vn
->sc_flags
& VNF_INITED
) == 0) {
806 buf_setresid(bp
, buf_count(bp
));
808 * Check for required alignment. Transfers must be a valid
809 * multiple of the sector size.
811 blk_num
= buf_blkno(bp
);
812 if (buf_count(bp
) % vn
->sc_secsize
!= 0) {
816 sz
= howmany(buf_count(bp
), vn
->sc_secsize
);
819 * If out of bounds return an error. If at the EOF point,
820 * simply read or write less.
822 if (blk_num
>= 0 && (u_int64_t
)blk_num
>= vn
->sc_size
) {
823 if (blk_num
> 0 && (u_int64_t
)blk_num
> vn
->sc_size
) {
829 * If the request crosses EOF, truncate the request.
831 if ((blk_num
+ sz
) > 0 && ((u_int64_t
)(blk_num
+ sz
)) > vn
->sc_size
) {
832 buf_setcount(bp
, (vn
->sc_size
- blk_num
) * vn
->sc_secsize
);
833 buf_setresid(bp
, buf_count(bp
));
840 error
= vnode_getwithvid(vp
, vn
->sc_vid
);
842 /* the vnode is no longer available, abort */
847 shadow_vp
= vn
->sc_shadow_vp
;
848 if (shadow_vp
!= NULL
) {
849 error
= vnode_getwithvid(shadow_vp
,
852 /* the vnode is no longer available, abort */
854 vnode_put(vn
->sc_vp
);
859 error
= vn_readwrite_io(vn
, bp
, p
);
861 if (shadow_vp
!= NULL
) {
862 vnode_put(shadow_vp
);
866 (void) thread_funnel_set(kernel_flock
, funnel_state
);
868 buf_seterror(bp
, error
);
876 vnioctl(dev_t dev
, u_long cmd
, caddr_t data
,
877 __unused
int flag
, struct proc
*p
,
881 struct user_vn_ioctl
*viop
;
886 struct vfsioattr ioattr
;
887 struct user_vn_ioctl user_vnio
;
888 boolean_t funnel_state
;
891 if (vnunit(dev
) >= NVNDEVICE
) {
895 funnel_state
= thread_funnel_set(kernel_flock
, TRUE
);
896 vn
= vn_table
+ unit
;
897 error
= proc_suser(p
);
902 viop
= (struct user_vn_ioctl
*)data
;
903 f
= (u_int32_t
*)data
;
904 o
= (u_int64_t
*)data
;
908 case DKIOCGETBLOCKSIZE
:
909 case DKIOCSETBLOCKSIZE
:
910 case DKIOCGETMAXBLOCKCOUNTREAD
:
911 case DKIOCGETMAXBLOCKCOUNTWRITE
:
912 case DKIOCGETMAXSEGMENTCOUNTREAD
:
913 case DKIOCGETMAXSEGMENTCOUNTWRITE
:
914 case DKIOCGETMAXSEGMENTBYTECOUNTREAD
:
915 case DKIOCGETMAXSEGMENTBYTECOUNTWRITE
:
916 case DKIOCGETBLOCKCOUNT
:
917 case DKIOCGETBLOCKCOUNT32
:
918 if ((vn
->sc_flags
& VNF_INITED
) == 0) {
927 if (vn
->sc_vp
!= NULL
)
928 vfs_ioattr(vnode_mount(vn
->sc_vp
), &ioattr
);
930 bzero(&ioattr
, sizeof(ioattr
));
936 case DKIOCGETMAXBLOCKCOUNTREAD
:
937 *o
= ioattr
.io_maxreadcnt
/ vn
->sc_secsize
;
939 case DKIOCGETMAXBLOCKCOUNTWRITE
:
940 *o
= ioattr
.io_maxwritecnt
/ vn
->sc_secsize
;
942 case DKIOCGETMAXBYTECOUNTREAD
:
943 *o
= ioattr
.io_maxreadcnt
;
945 case DKIOCGETMAXBYTECOUNTWRITE
:
946 *o
= ioattr
.io_maxwritecnt
;
948 case DKIOCGETMAXSEGMENTCOUNTREAD
:
949 *o
= ioattr
.io_segreadcnt
;
951 case DKIOCGETMAXSEGMENTCOUNTWRITE
:
952 *o
= ioattr
.io_segwritecnt
;
954 case DKIOCGETMAXSEGMENTBYTECOUNTREAD
:
955 *o
= ioattr
.io_maxsegreadsize
;
957 case DKIOCGETMAXSEGMENTBYTECOUNTWRITE
:
958 *o
= ioattr
.io_maxsegwritesize
;
960 case DKIOCGETBLOCKSIZE
:
963 case DKIOCSETBLOCKSIZE
:
965 /* can only set block size on block device */
969 if (*f
< DEV_BSIZE
) {
973 if (vn
->sc_shadow_vp
!= NULL
) {
974 if (*f
== (unsigned)vn
->sc_secsize
) {
977 /* can't change the block size if already shadowing */
982 /* recompute the size in terms of the new blocksize */
983 vn
->sc_size
= vn
->sc_fsize
/ vn
->sc_secsize
;
985 case DKIOCISWRITABLE
:
988 case DKIOCGETBLOCKCOUNT32
:
991 case DKIOCGETBLOCKCOUNT
:
996 if (vn
->sc_shadow_vp
!= NULL
) {
1000 if (vn
->sc_vp
== NULL
) {
1001 /* much be attached before we can shadow */
1005 if (!proc_is64bit(p
)) {
1006 /* downstream code expects LP64 version of vn_ioctl structure */
1007 vn_ioctl_to_64((struct vn_ioctl
*)viop
, &user_vnio
);
1010 if (viop
->vn_file
== USER_ADDR_NULL
) {
1014 error
= vniocattach_shadow(vn
, viop
, dev
, 0, p
);
1020 /* attach only on block device */
1024 if (vn
->sc_flags
& VNF_INITED
) {
1028 if (!proc_is64bit(p
)) {
1029 /* downstream code expects LP64 version of vn_ioctl structure */
1030 vn_ioctl_to_64((struct vn_ioctl
*)viop
, &user_vnio
);
1033 if (viop
->vn_file
== USER_ADDR_NULL
) {
1037 error
= vniocattach_file(vn
, viop
, dev
, 0, p
);
1043 /* detach only on block device */
1047 /* Note: spec_open won't open a mounted block device */
1050 * XXX handle i/o in progress. Return EBUSY, or wait, or
1052 * XXX handle multiple opens of the device. Return EBUSY,
1053 * or revoke the fd's.
1054 * How are these problems handled for removable and failing
1055 * hardware devices? (Hint: They are not)
1066 vn_options
&= ~(*f
);
1071 vn
->sc_options
|= *f
;
1072 *f
= vn
->sc_options
;
1076 vn
->sc_options
&= ~(*f
);
1077 *f
= vn
->sc_options
;
1085 (void) thread_funnel_set(kernel_flock
, funnel_state
);
1090 vnioctl_chr(dev_t dev
, u_long cmd
, caddr_t data
, int flag
, struct proc
*p
)
1092 return (vnioctl(dev
, cmd
, data
, flag
, p
, TRUE
));
1096 vnioctl_blk(dev_t dev
, u_long cmd
, caddr_t data
, int flag
, struct proc
*p
)
1098 return (vnioctl(dev
, cmd
, data
, flag
, p
, FALSE
));
1104 * Attach a file to a VN partition. Return the size in the vn_size
1109 vniocattach_file(struct vn_softc
*vn
,
1110 struct user_vn_ioctl
*vniop
,
1116 struct vfs_context context
;
1118 struct nameidata nd
;
1122 context
.vc_proc
= p
;
1123 context
.vc_ucred
= kauth_cred_proc_ref(p
);
1125 flags
= FREAD
|FWRITE
;
1127 NDINIT(&nd
, LOOKUP
, FOLLOW
, UIO_SYSSPACE32
, vniop
->vn_file
, &context
);
1130 NDINIT(&nd
, LOOKUP
, FOLLOW
,
1131 (IS_64BIT_PROCESS(p
) ? UIO_USERSPACE64
: UIO_USERSPACE32
),
1132 vniop
->vn_file
, &context
);
1134 /* vn_open gives both long- and short-term references */
1135 error
= vn_open(&nd
, flags
, 0);
1137 if (error
!= EACCES
&& error
!= EPERM
&& error
!= EROFS
) {
1138 kauth_cred_unref(&context
.vc_ucred
);
1143 NDINIT(&nd
, LOOKUP
, FOLLOW
, UIO_SYSSPACE32
,
1144 vniop
->vn_file
, &context
);
1147 NDINIT(&nd
, LOOKUP
, FOLLOW
,
1148 (IS_64BIT_PROCESS(p
) ? UIO_USERSPACE64
: UIO_USERSPACE32
),
1149 vniop
->vn_file
, &context
);
1151 error
= vn_open(&nd
, flags
, 0);
1153 kauth_cred_unref(&context
.vc_ucred
);
1157 if (nd
.ni_vp
->v_type
!= VREG
) {
1161 error
= vnode_size(nd
.ni_vp
, &file_size
, &context
);
1164 (void) vn_close(nd
.ni_vp
, flags
, context
.vc_ucred
, p
);
1165 vnode_put(nd
.ni_vp
);
1166 kauth_cred_unref(&context
.vc_ucred
);
1169 cred
= kauth_cred_proc_ref(p
);
1170 nd
.ni_vp
->v_flag
|= VNOCACHE_DATA
;
1171 error
= setcred(nd
.ni_vp
, p
, cred
);
1173 (void)vn_close(nd
.ni_vp
, flags
, context
.vc_ucred
, p
);
1174 vnode_put(nd
.ni_vp
);
1175 kauth_cred_unref(&cred
);
1176 kauth_cred_unref(&context
.vc_ucred
);
1179 vn
->sc_secsize
= DEV_BSIZE
;
1180 vn
->sc_fsize
= file_size
;
1181 vn
->sc_size
= file_size
/ vn
->sc_secsize
;
1182 vn
->sc_vp
= nd
.ni_vp
;
1183 vn
->sc_vid
= vnode_vid(nd
.ni_vp
);
1184 vn
->sc_open_flags
= flags
;
1186 cdev
= makedev(vndevice_cdev_major
, minor(dev
));
1187 vn
->sc_cdev
= devfs_make_node(cdev
, DEVFS_CHAR
,
1188 UID_ROOT
, GID_OPERATOR
,
1191 vn
->sc_flags
|= VNF_INITED
;
1193 vn
->sc_flags
|= VNF_READONLY
;
1194 /* lose the short-term reference */
1195 vnode_put(nd
.ni_vp
);
1196 kauth_cred_unref(&context
.vc_ucred
);
1201 vniocattach_shadow(struct vn_softc
*vn
, struct user_vn_ioctl
*vniop
,
1202 __unused
int dev
, int in_kernel
, struct proc
*p
)
1204 struct vfs_context context
;
1205 struct nameidata nd
;
1210 context
.vc_proc
= p
;
1211 context
.vc_ucred
= kauth_cred_proc_ref(p
);
1213 flags
= FREAD
|FWRITE
;
1215 NDINIT(&nd
, LOOKUP
, FOLLOW
, UIO_SYSSPACE32
, vniop
->vn_file
, &context
);
1218 NDINIT(&nd
, LOOKUP
, FOLLOW
,
1219 (IS_64BIT_PROCESS(p
) ? UIO_USERSPACE64
: UIO_USERSPACE32
),
1220 vniop
->vn_file
, &context
);
1222 /* vn_open gives both long- and short-term references */
1223 error
= vn_open(&nd
, flags
, 0);
1225 /* shadow MUST be writable! */
1226 kauth_cred_unref(&context
.vc_ucred
);
1229 if (nd
.ni_vp
->v_type
!= VREG
1230 || (error
= vnode_size(nd
.ni_vp
, &file_size
, &context
))) {
1231 (void)vn_close(nd
.ni_vp
, flags
, context
.vc_ucred
, p
);
1232 vnode_put(nd
.ni_vp
);
1233 kauth_cred_unref(&context
.vc_ucred
);
1234 return (error
? error
: EINVAL
);
1236 map
= shadow_map_create(vn
->sc_fsize
, file_size
,
1239 (void)vn_close(nd
.ni_vp
, flags
, context
.vc_ucred
, p
);
1240 vnode_put(nd
.ni_vp
);
1241 vn
->sc_shadow_vp
= NULL
;
1242 kauth_cred_unref(&context
.vc_ucred
);
1245 vn
->sc_shadow_vp
= nd
.ni_vp
;
1246 vn
->sc_shadow_vid
= vnode_vid(nd
.ni_vp
);
1247 vn
->sc_shadow_vp
->v_flag
|= VNOCACHE_DATA
;
1248 vn
->sc_shadow_map
= map
;
1249 vn
->sc_flags
&= ~VNF_READONLY
; /* we're now read/write */
1251 /* lose the short-term reference */
1252 vnode_put(nd
.ni_vp
);
1253 kauth_cred_unref(&context
.vc_ucred
);
1258 vndevice_root_image(char * path
, char devname
[], dev_t
* dev_p
)
1261 struct vn_softc
* vn
;
1262 struct user_vn_ioctl vnio
;
1264 vnio
.vn_file
= CAST_USER_ADDR_T(path
);
1267 vn
= vn_table
+ ROOT_IMAGE_UNIT
;
1268 *dev_p
= makedev(vndevice_bdev_major
,
1270 sprintf(devname
, "vn%d", ROOT_IMAGE_UNIT
);
1271 error
= vniocattach_file(vn
, &vnio
, *dev_p
, 1, current_proc());
1276 * Duplicate the current processes' credentials. Since we are called only
1277 * as the result of a SET ioctl and only root can do that, any future access
1278 * to this "disk" is essentially as root. Note that credentials may change
1279 * if some other uid can write directly to the mapped file (NFS).
1282 setcred(struct vnode
* vp
, struct proc
* p
, kauth_cred_t cred
)
1286 struct vfs_context context
;
1289 * Horrible kludge to establish credentials for NFS XXX.
1291 context
.vc_proc
= p
;
1292 context
.vc_ucred
= cred
;
1293 tmpbuf
= _MALLOC(DEV_BSIZE
, M_TEMP
, M_WAITOK
);
1294 error
= file_io(vp
, &context
, UIO_READ
, tmpbuf
, 0, DEV_BSIZE
, NULL
);
1295 FREE(tmpbuf
, M_TEMP
);
1300 vnclear(struct vn_softc
*vn
, struct proc
* p
)
1302 if (vn
->sc_vp
!= NULL
) {
1303 /* release long-term reference */
1304 (void)vn_close(vn
->sc_vp
, vn
->sc_open_flags
, vn
->sc_cred
, p
);
1307 if (vn
->sc_shadow_vp
!= NULL
) {
1308 /* release long-term reference */
1309 (void)vn_close(vn
->sc_shadow_vp
, FREAD
| FWRITE
,
1311 vn
->sc_shadow_vp
= NULL
;
1313 if (vn
->sc_shadow_map
!= NULL
) {
1314 shadow_map_free(vn
->sc_shadow_map
);
1315 vn
->sc_shadow_map
= NULL
;
1317 vn
->sc_flags
&= ~(VNF_INITED
| VNF_READONLY
);
1319 kauth_cred_unref(&vn
->sc_cred
);
1324 devfs_remove(vn
->sc_cdev
);
1333 struct vn_softc
*vn
;
1335 boolean_t funnel_state
;
1338 if (vnunit(dev
) >= NVNDEVICE
) {
1342 funnel_state
= thread_funnel_set(kernel_flock
, TRUE
);
1343 vn
= vn_table
+ unit
;
1344 if ((vn
->sc_flags
& VNF_INITED
) == 0)
1347 secsize
= vn
->sc_secsize
;
1348 (void) thread_funnel_set(kernel_flock
, funnel_state
);
1352 #define CDEV_MAJOR -1
1353 #define BDEV_MAJOR -1
1354 static int vndevice_inited
= 0;
1361 if (vndevice_inited
)
1363 vndevice_bdev_major
= bdevsw_add(BDEV_MAJOR
, &vn_bdevsw
);
1365 if (vndevice_bdev_major
< 0) {
1366 printf("vndevice_init: bdevsw_add() returned %d\n",
1367 vndevice_bdev_major
);
1370 vndevice_cdev_major
= cdevsw_add_with_bdev(CDEV_MAJOR
, &vn_cdevsw
,
1371 vndevice_bdev_major
);
1372 if (vndevice_cdev_major
< 0) {
1373 printf("vndevice_init: cdevsw_add() returned %d\n",
1374 vndevice_cdev_major
);
1377 for (i
= 0; i
< NVNDEVICE
; i
++) {
1378 dev_t dev
= makedev(vndevice_bdev_major
, i
);
1379 vn_table
[i
].sc_bdev
= devfs_make_node(dev
, DEVFS_BLOCK
,
1380 UID_ROOT
, GID_OPERATOR
,
1383 if (vn_table
[i
].sc_bdev
== NULL
)
1384 printf("vninit: devfs_make_node failed!\n");
1389 vn_ioctl_to_64(struct vn_ioctl
*from
, struct user_vn_ioctl
*to
)
1391 to
->vn_file
= CAST_USER_ADDR_T(from
->vn_file
);
1392 to
->vn_size
= from
->vn_size
;
1393 to
->vn_control
= from
->vn_control
;
1396 #endif /* NVNDEVICE */