2 * Copyright (c) 1991, 1993
3 * The Regents of the University of California. All rights reserved.
5 * This code is derived from software contributed to Berkeley by
6 * The Mach Operating System project at Carnegie-Mellon University.
8 * Redistribution and use in source and binary forms, with or without
9 * modification, are permitted provided that the following conditions
11 * 1. Redistributions of source code must retain the above copyright
12 * notice, this list of conditions and the following disclaimer.
13 * 2. Redistributions in binary form must reproduce the above copyright
14 * notice, this list of conditions and the following disclaimer in the
15 * documentation and/or other materials provided with the distribution.
16 * 3. All advertising materials mentioning features or use of this software
17 * must display the following acknowledgement:
18 * This product includes software developed by the University of
19 * California, Berkeley and its contributors.
20 * 4. Neither the name of the University nor the names of its contributors
21 * may be used to endorse or promote products derived from this software
22 * without specific prior written permission.
24 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
25 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
26 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
27 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
28 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
29 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
30 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
31 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
32 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
33 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
36 * from: @(#)vm_pager.c 8.6 (Berkeley) 1/12/94
39 * Copyright (c) 1987, 1990 Carnegie-Mellon University.
40 * All rights reserved.
42 * Authors: Avadis Tevanian, Jr., Michael Wayne Young
44 * Permission to use, copy, modify and distribute this software and
45 * its documentation is hereby granted, provided that both the copyright
46 * notice and this permission notice appear in all copies of the
47 * software, derivative works or modified versions, and any portions
48 * thereof, and that both notices appear in supporting documentation.
50 * CARNEGIE MELLON ALLOWS FREE USE OF THIS SOFTWARE IN ITS "AS IS"
51 * CONDITION. CARNEGIE MELLON DISCLAIMS ANY LIABILITY OF ANY KIND
52 * FOR ANY DAMAGES WHATSOEVER RESULTING FROM THE USE OF THIS SOFTWARE.
54 * Carnegie Mellon requests users of this software to return to
56 * Software Distribution Coordinator or Software.Distribution@CS.CMU.EDU
57 * School of Computer Science
58 * Carnegie Mellon University
59 * Pittsburgh PA 15213-3890
61 * any improvements or extensions that they make and grant Carnegie the
62 * rights to redistribute these changes.
64 * $FreeBSD: src/sys/vm/vm_pager.c,v 1.54.2.2 2001/11/18 07:11:00 dillon Exp $
65 * $DragonFly: src/sys/vm/vm_pager.c,v 1.24 2007/11/06 03:50:01 dillon Exp $
69 * Paging space routine stubs. Emulates a matchmaker-like interface
73 #include <sys/param.h>
74 #include <sys/systm.h>
75 #include <sys/kernel.h>
76 #include <sys/vnode.h>
78 #include <sys/ucred.h>
79 #include <sys/malloc.h>
81 #include <sys/thread2.h>
84 #include <vm/vm_param.h>
85 #include <vm/vm_object.h>
86 #include <vm/vm_page.h>
87 #include <vm/vm_pager.h>
88 #include <vm/vm_extern.h>
92 MALLOC_DEFINE(M_VMPGDATA
, "VM pgdata", "XXX: VM pager private data");
94 extern struct pagerops defaultpagerops
;
95 extern struct pagerops swappagerops
;
96 extern struct pagerops vnodepagerops
;
97 extern struct pagerops devicepagerops
;
98 extern struct pagerops physpagerops
;
100 int cluster_pbuf_freecnt
= -1; /* unlimited to begin with */
102 static int dead_pager_getpage (vm_object_t
, vm_page_t
*, int);
103 static vm_object_t
dead_pager_alloc (void *, off_t
, vm_prot_t
, off_t
);
104 static void dead_pager_putpages (vm_object_t
, vm_page_t
*, int, int, int *);
105 static boolean_t
dead_pager_haspage (vm_object_t
, vm_pindex_t
);
106 static void dead_pager_dealloc (vm_object_t
);
109 dead_pager_getpage(vm_object_t obj
, vm_page_t
*mpp
, int seqaccess
)
111 return VM_PAGER_FAIL
;
115 dead_pager_alloc(void *handle
, off_t size
, vm_prot_t prot
, off_t off
)
121 dead_pager_putpages(vm_object_t object
, vm_page_t
*m
, int count
, int flags
,
126 for (i
= 0; i
< count
; i
++) {
127 rtvals
[i
] = VM_PAGER_AGAIN
;
132 dead_pager_haspage(vm_object_t object
, vm_pindex_t pindex
)
138 dead_pager_dealloc(vm_object_t object
)
140 KKASSERT(object
->swblock_count
== 0);
144 static struct pagerops deadpagerops
= {
152 struct pagerops
*pagertab
[] = {
153 &defaultpagerops
, /* OBJT_DEFAULT */
154 &swappagerops
, /* OBJT_SWAP */
155 &vnodepagerops
, /* OBJT_VNODE */
156 &devicepagerops
, /* OBJT_DEVICE */
157 &physpagerops
, /* OBJT_PHYS */
158 &deadpagerops
/* OBJT_DEAD */
161 int npagers
= sizeof(pagertab
) / sizeof(pagertab
[0]);
164 * Kernel address space for mapping pages.
165 * Used by pagers where KVAs are needed for IO.
167 * XXX needs to be large enough to support the number of pending async
168 * cleaning requests (NPENDINGIO == 64) * the maximum swap cluster size
169 * (MAXPHYS == 64k) if you want to get the most efficiency.
171 #define PAGER_MAP_SIZE (8 * 1024 * 1024)
173 int pager_map_size
= PAGER_MAP_SIZE
;
174 struct vm_map pager_map
;
176 static int bswneeded
;
177 static vm_offset_t swapbkva
; /* swap buffers kva */
178 static TAILQ_HEAD(swqueue
, buf
) bswlist
;
179 static struct spinlock bswspin
= SPINLOCK_INITIALIZER(&bswspin
);
182 vm_pager_init(void *arg __unused
)
185 * Initialize the swap buffer list.
187 TAILQ_INIT(&bswlist
);
189 SYSINIT(vm_mem
, SI_BOOT1_VM
, SI_ORDER_SECOND
, vm_pager_init
, NULL
)
192 vm_pager_bufferinit(void)
198 * Reserve KVM space for pbuf data.
200 swapbkva
= kmem_alloc_pageable(&pager_map
, nswbuf
* MAXPHYS
);
202 panic("Not enough pager_map VM space for physical buffers");
205 * Initial pbuf setup.
208 for (i
= 0; i
< nswbuf
; ++i
, ++bp
) {
209 bp
->b_kvabase
= (caddr_t
)((intptr_t)i
* MAXPHYS
) + swapbkva
;
210 bp
->b_kvasize
= MAXPHYS
;
211 TAILQ_INSERT_HEAD(&bswlist
, bp
, b_freelist
);
217 * Allow the clustering code to use half of our pbufs.
219 cluster_pbuf_freecnt
= nswbuf
/ 2;
223 * Allocate an instance of a pager of the given type.
224 * Size, protection and offset parameters are passed in for pagers that
225 * need to perform page-level validation (e.g. the device pager).
228 vm_pager_allocate(objtype_t type
, void *handle
, vm_ooffset_t size
,
229 vm_prot_t prot
, off_t off
)
231 struct pagerops
*ops
;
233 ops
= pagertab
[type
];
235 return ((*ops
->pgo_alloc
) (handle
, size
, prot
, off
));
240 vm_pager_deallocate(vm_object_t object
)
242 (*pagertab
[object
->type
]->pgo_dealloc
) (object
);
246 * vm_pager_get_pages() - inline, see vm/vm_pager.h
247 * vm_pager_put_pages() - inline, see vm/vm_pager.h
248 * vm_pager_has_page() - inline, see vm/vm_pager.h
249 * vm_pager_page_inserted() - inline, see vm/vm_pager.h
250 * vm_pager_page_removed() - inline, see vm/vm_pager.h
257 * Called by pageout daemon before going back to sleep.
258 * Gives pagers a chance to clean up any completed async pageing
264 struct pagerops
**pgops
;
266 for (pgops
= pagertab
; pgops
< &pagertab
[npagers
]; pgops
++)
267 if (pgops
&& ((*pgops
)->pgo_sync
!= NULL
))
268 (*(*pgops
)->pgo_sync
) ();
274 * Initialize a physical buffer.
277 initpbuf(struct buf
*bp
)
279 bp
->b_qindex
= 0; /* BQUEUE_NONE */
280 bp
->b_data
= bp
->b_kvabase
;
281 bp
->b_flags
= B_PAGING
;
282 bp
->b_cmd
= BUF_CMD_DONE
;
285 bp
->b_bufsize
= MAXPHYS
;
287 xio_init(&bp
->b_xio
);
288 BUF_LOCK(bp
, LK_EXCLUSIVE
);
292 * Allocate a physical buffer
294 * There are a limited number (nswbuf) of physical buffers. We need
295 * to make sure that no single subsystem is able to hog all of them,
296 * so each subsystem implements a counter which is typically initialized
297 * to 1/2 nswbuf. getpbuf() decrements this counter in allocation and
298 * increments it on release, and blocks if the counter hits zero. A
299 * subsystem may initialize the counter to -1 to disable the feature,
300 * but it must still be sure to match up all uses of getpbuf() with
301 * relpbuf() using the same variable.
303 * NOTE: pfreecnt can be NULL, but this 'feature' will be removed
304 * relatively soon when the rest of the subsystems get smart about it. XXX
309 getpbuf(int *pfreecnt
)
313 spin_lock_wr(&bswspin
);
317 while (*pfreecnt
== 0)
318 ssleep(pfreecnt
, &bswspin
, 0, "wswbuf0", 0);
321 /* get a bp from the swap buffer header pool */
322 if ((bp
= TAILQ_FIRST(&bswlist
)) != NULL
)
325 ssleep(&bswneeded
, &bswspin
, 0, "wswbuf1", 0);
326 /* loop in case someone else grabbed one */
328 TAILQ_REMOVE(&bswlist
, bp
, b_freelist
);
332 spin_unlock_wr(&bswspin
);
339 * Allocate a physical buffer, if one is available.
341 * Note that there is no NULL hack here - all subsystems using this
342 * call understand how to use pfreecnt.
347 trypbuf(int *pfreecnt
)
351 spin_lock_wr(&bswspin
);
353 if (*pfreecnt
== 0 || (bp
= TAILQ_FIRST(&bswlist
)) == NULL
) {
354 spin_unlock_wr(&bswspin
);
357 TAILQ_REMOVE(&bswlist
, bp
, b_freelist
);
360 spin_unlock_wr(&bswspin
);
368 * Release a physical buffer
370 * NOTE: pfreecnt can be NULL, but this 'feature' will be removed
371 * relatively soon when the rest of the subsystems get smart about it. XXX
376 relpbuf(struct buf
*bp
, int *pfreecnt
)
379 int wake_freecnt
= 0;
381 KKASSERT(bp
->b_flags
& B_PAGING
);
383 spin_lock_wr(&bswspin
);
386 TAILQ_INSERT_HEAD(&bswlist
, bp
, b_freelist
);
392 if (++*pfreecnt
== 1)
396 spin_unlock_wr(&bswspin
);