2 * Copyright (c) 1982, 1986, 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 * @(#)vm_swap.c 8.5 (Berkeley) 2/17/94
34 * $FreeBSD: src/sys/vm/vm_swap.c,v 1.96.2.2 2001/10/14 18:46:47 iedowse Exp $
35 * $DragonFly: src/sys/vm/vm_swap.c,v 1.35 2007/05/15 22:44:21 dillon Exp $
40 #include <sys/param.h>
41 #include <sys/systm.h>
42 #include <sys/sysproto.h>
45 #include <sys/nlookup.h>
46 #include <sys/dmap.h> /* XXX */
47 #include <sys/vnode.h>
48 #include <sys/fcntl.h>
49 #include <sys/blist.h>
50 #include <sys/kernel.h>
54 #include <sys/thread2.h>
56 #include <vm/vm_extern.h>
57 #include <vm/swap_pager.h>
58 #include <vm/vm_zone.h>
61 * Indirect driver for multi-controller paging.
67 static struct swdevt should_be_malloced
[NSWAPDEV
];
68 struct swdevt
*swdevt
= should_be_malloced
; /* exported to pstat/systat */
69 static int nswap
; /* first block after the interleaved devs */
70 int nswdev
= NSWAPDEV
; /* exported to pstat/systat */
73 static int swapdev_strategy (struct vop_strategy_args
*ap
);
74 struct vnode
*swapdev_vp
;
79 * vn_strategy() for swapdev_vp.
80 * Perform swap strategy interleave device selection.
82 * The bp is expected to be locked and on call.
84 * (struct vnode *a_vp, struct bio *b_bio)
88 swapdev_strategy(struct vop_strategy_args
*ap
)
90 struct bio
*bio
= ap
->a_bio
;
92 struct buf
*bp
= bio
->bio_buf
;
93 int sz
, off
, seg
, index
, blkno
, nblkno
;
98 sz
= howmany(bp
->b_bcount
, PAGE_SIZE
);
99 blkno
= (int)(bio
->bio_offset
>> PAGE_SHIFT
);
102 * Convert interleaved swap into per-device swap. Note that
103 * the block size is left in PAGE_SIZE'd chunks (for the newswap)
106 nbio
= push_bio(bio
);
109 if (off
+ sz
> dmmax
) {
110 bp
->b_error
= EINVAL
;
111 bp
->b_flags
|= B_ERROR
;
116 index
= seg
% nswdev
;
118 nbio
->bio_offset
= (off_t
)(seg
* dmmax
+ off
) << PAGE_SHIFT
;
121 nbio
->bio_offset
= bio
->bio_offset
;
123 nblkno
= (int)(nbio
->bio_offset
>> PAGE_SHIFT
);
125 if (nblkno
+ sz
> sp
->sw_nblks
) {
126 bp
->b_error
= EINVAL
;
127 bp
->b_flags
|= B_ERROR
;
128 /* I/O was never started on nbio, must biodone(bio) */
132 if (sp
->sw_vp
== NULL
) {
133 bp
->b_error
= ENODEV
;
134 bp
->b_flags
|= B_ERROR
;
135 /* I/O was never started on nbio, must biodone(bio) */
141 * Issue a strategy call on the appropriate swap vnode. Note that
142 * bp->b_vp is not modified. Strategy code is always supposed to
145 * XXX do a dev_dstrategy() call on sp->sw_device instead of on
148 vn_strategy(sp
->sw_vp
, nbio
);
153 * Create a special vnode op vector for swapdev_vp - we only use
154 * vn_strategy(), everything else returns an error.
156 static struct vop_ops swapdev_vnode_vops
= {
157 .vop_default
= vop_defaultop
,
158 .vop_strategy
= swapdev_strategy
160 static struct vop_ops
*swapdev_vnode_vops_p
= &swapdev_vnode_vops
;
162 VNODEOP_SET(swapdev_vnode_vops
);
165 * swapon_args(char *name)
167 * System call swapon(name) enables swapping on device name,
168 * which must be in the swdevsw. Return EBUSY
169 * if already swapping on this device.
173 sys_swapon(struct swapon_args
*uap
)
175 struct thread
*td
= curthread
;
178 struct nlookupdata nd
;
182 KKASSERT(td
->td_proc
);
183 cred
= td
->td_proc
->p_ucred
;
190 error
= nlookup_init(&nd
, uap
->name
, UIO_USERSPACE
, NLC_FOLLOW
);
192 error
= nlookup(&nd
);
194 error
= cache_vref(&nd
.nl_nch
, nd
.nl_cred
, &vp
);
199 if (vn_isdisk(vp
, &error
))
200 error
= swaponvp(td
, vp
, 0);
201 else if (vp
->v_type
== VREG
&& vp
->v_tag
== VT_NFS
&&
202 (error
= VOP_GETATTR(vp
, &attr
)) == 0) {
204 * Allow direct swapping to NFS regular files in the same
205 * way that nfs_mountroot() sets up diskless swapping.
207 error
= swaponvp(td
, vp
, attr
.va_size
/ DEV_BSIZE
);
217 * Swfree(index) frees the index'th portion of the swap map.
218 * Each of the nswdev devices provides 1/nswdev'th of the swap
219 * space, which is laid out with blocks of dmmax pages circularly
222 * The new swap code uses page-sized blocks. The old swap code used
223 * DEV_BSIZE'd chunks.
225 * XXX locking when multiple swapon's run in parallel
228 swaponvp(struct thread
*td
, struct vnode
*vp
, u_long nblks
)
230 u_long aligned_nblks
;
241 KKASSERT(td
->td_proc
);
242 cred
= td
->td_proc
->p_ucred
;
245 error
= getspecialvnode(VT_NON
, NULL
, &swapdev_vnode_vops_p
,
248 panic("Cannot get vnode for swapdev");
249 swapdev_vp
->v_type
= VNON
; /* Untyped */
250 vx_unlock(swapdev_vp
);
253 for (sp
= swdevt
, index
= 0 ; index
< nswdev
; index
++, sp
++) {
262 vn_lock(vp
, LK_EXCLUSIVE
| LK_RETRY
);
263 error
= VOP_OPEN(vp
, FREAD
| FWRITE
, cred
, NULL
);
269 * v_rdev is not valid until after the VOP_OPEN() call. dev_psize()
270 * must be supported if a character device has been specified.
272 if (vp
->v_type
== VCHR
)
277 if (nblks
== 0 && dev
!= NULL
) {
278 dpsize
= dev_dpsize(dev
);
280 VOP_CLOSE(vp
, FREAD
| FWRITE
);
283 if ((u_int64_t
)dpsize
< 0x100000000ULL
)
284 nblks
= (u_long
)dpsize
;
289 VOP_CLOSE(vp
, FREAD
| FWRITE
);
294 * If we go beyond this, we get overflows in the radix
297 if (nblks
> 0x40000000 / BLIST_META_RADIX
/ nswdev
) {
298 kprintf("exceeded maximum of %d blocks per swap unit\n",
299 0x40000000 / BLIST_META_RADIX
/ nswdev
);
300 VOP_CLOSE(vp
, FREAD
| FWRITE
);
304 * nblks is in DEV_BSIZE'd chunks, convert to PAGE_SIZE'd chunks.
305 * First chop nblks off to page-align it, then convert.
307 * sw->sw_nblks is in page-sized chunks now too.
309 nblks
&= ~(ctodb(1) - 1);
310 nblks
= dbtoc(nblks
);
313 sp
->sw_dev
= dev2udev(dev
);
315 sp
->sw_flags
|= SW_FREED
;
316 sp
->sw_nblks
= nblks
;
319 * nblks, nswap, and dmmax are PAGE_SIZE'd parameters now, not
320 * DEV_BSIZE'd. aligned_nblks is used to calculate the
321 * size of the swap bitmap, taking into account the stripe size.
323 aligned_nblks
= (nblks
+ (dmmax
- 1)) & ~(u_long
)(dmmax
- 1);
325 if (aligned_nblks
* nswdev
> nswap
)
326 nswap
= aligned_nblks
* nswdev
;
328 if (swapblist
== NULL
)
329 swapblist
= blist_create(nswap
);
331 blist_resize(&swapblist
, nswap
, 0);
333 for (dvbase
= dmmax
; dvbase
< nblks
; dvbase
+= dmmax
) {
334 blk
= min(nblks
- dvbase
, dmmax
);
335 vsbase
= index
* dmmax
+ dvbase
* nswdev
;
336 blist_free(swapblist
, vsbase
, blk
);