vm_object.h revision 323537
1/*-
2 * Copyright (c) 1991, 1993
3 *	The Regents of the University of California.  All rights reserved.
4 *
5 * This code is derived from software contributed to Berkeley by
6 * The Mach Operating System project at Carnegie-Mellon University.
7 *
8 * Redistribution and use in source and binary forms, with or without
9 * modification, are permitted provided that the following conditions
10 * are met:
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 * 4. Neither the name of the University nor the names of its contributors
17 *    may be used to endorse or promote products derived from this software
18 *    without specific prior written permission.
19 *
20 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
21 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
22 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
23 * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
24 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
25 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
26 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
27 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
28 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
29 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
30 * SUCH DAMAGE.
31 *
32 *	from: @(#)vm_object.h	8.3 (Berkeley) 1/12/94
33 *
34 *
35 * Copyright (c) 1987, 1990 Carnegie-Mellon University.
36 * All rights reserved.
37 *
38 * Authors: Avadis Tevanian, Jr., Michael Wayne Young
39 *
40 * Permission to use, copy, modify and distribute this software and
41 * its documentation is hereby granted, provided that both the copyright
42 * notice and this permission notice appear in all copies of the
43 * software, derivative works or modified versions, and any portions
44 * thereof, and that both notices appear in supporting documentation.
45 *
46 * CARNEGIE MELLON ALLOWS FREE USE OF THIS SOFTWARE IN ITS "AS IS"
47 * CONDITION.  CARNEGIE MELLON DISCLAIMS ANY LIABILITY OF ANY KIND
48 * FOR ANY DAMAGES WHATSOEVER RESULTING FROM THE USE OF THIS SOFTWARE.
49 *
50 * Carnegie Mellon requests users of this software to return to
51 *
52 *  Software Distribution Coordinator  or  Software.Distribution@CS.CMU.EDU
53 *  School of Computer Science
54 *  Carnegie Mellon University
55 *  Pittsburgh PA 15213-3890
56 *
57 * any improvements or extensions that they make and grant Carnegie the
58 * rights to redistribute these changes.
59 *
60 * $FreeBSD: stable/11/sys/vm/vm_object.h 323537 2017-09-13 11:19:04Z kib $
61 */
62
63/*
64 *	Virtual memory object module definitions.
65 */
66
67#ifndef	_VM_OBJECT_
68#define	_VM_OBJECT_
69
70#include <sys/queue.h>
71#include <sys/_lock.h>
72#include <sys/_mutex.h>
73#include <sys/_pctrie.h>
74#include <sys/_rwlock.h>
75
76#include <vm/_vm_radix.h>
77
78/*
79 *	Types defined:
80 *
81 *	vm_object_t		Virtual memory object.
82 *
83 * List of locks
84 *	(c)	const until freed
85 *	(o)	per-object lock
86 *	(f)	free pages queue mutex
87 *
88 */
89
90struct vm_object {
91	struct rwlock lock;
92	TAILQ_ENTRY(vm_object) object_list; /* list of all objects */
93	LIST_HEAD(, vm_object) shadow_head; /* objects that this is a shadow for */
94	LIST_ENTRY(vm_object) shadow_list; /* chain of shadow objects */
95	TAILQ_HEAD(respgs, vm_page) memq; /* list of resident pages */
96	struct vm_radix rtree;		/* root of the resident page radix trie*/
97	vm_pindex_t size;		/* Object size */
98	int generation;			/* generation ID */
99	int ref_count;			/* How many refs?? */
100	int shadow_count;		/* how many objects that this is a shadow for */
101	vm_memattr_t memattr;		/* default memory attribute for pages */
102	objtype_t type;			/* type of pager */
103	u_short flags;			/* see below */
104	u_short pg_color;		/* (c) color of first page in obj */
105	u_int paging_in_progress;	/* Paging (in or out) so don't collapse or destroy */
106	int resident_page_count;	/* number of resident pages */
107	struct vm_object *backing_object; /* object that I'm a shadow of */
108	vm_ooffset_t backing_object_offset;/* Offset in backing object */
109	TAILQ_ENTRY(vm_object) pager_object_list; /* list of all objects of this pager type */
110	LIST_HEAD(, vm_reserv) rvq;	/* list of reservations */
111	void *handle;
112	union {
113		/*
114		 * VNode pager
115		 *
116		 *	vnp_size - current size of file
117		 */
118		struct {
119			off_t vnp_size;
120			vm_ooffset_t writemappings;
121		} vnp;
122
123		/*
124		 * Device pager
125		 *
126		 *	devp_pglist - list of allocated pages
127		 */
128		struct {
129			TAILQ_HEAD(, vm_page) devp_pglist;
130			struct cdev_pager_ops *ops;
131			struct cdev *dev;
132		} devp;
133
134		/*
135		 * SG pager
136		 *
137		 *	sgp_pglist - list of allocated pages
138		 */
139		struct {
140			TAILQ_HEAD(, vm_page) sgp_pglist;
141		} sgp;
142
143		/*
144		 * Swap pager
145		 *
146		 *	swp_tmpfs - back-pointer to the tmpfs vnode,
147		 *		     if any, which uses the vm object
148		 *		     as backing store.  The handle
149		 *		     cannot be reused for linking,
150		 *		     because the vnode can be
151		 *		     reclaimed and recreated, making
152		 *		     the handle changed and hash-chain
153		 *		     invalid.
154		 *
155		 *	swp_blks -   pc-trie of the allocated swap blocks.
156		 *
157		 */
158		struct {
159			void *swp_tmpfs;
160			struct pctrie swp_blks;
161		} swp;
162	} un_pager;
163	struct ucred *cred;
164	vm_ooffset_t charge;
165	void *umtx_data;
166};
167
168/*
169 * Flags
170 */
171#define	OBJ_FICTITIOUS	0x0001		/* (c) contains fictitious pages */
172#define	OBJ_UNMANAGED	0x0002		/* (c) contains unmanaged pages */
173#define	OBJ_POPULATE	0x0004		/* pager implements populate() */
174#define OBJ_DEAD	0x0008		/* dead objects (during rundown) */
175#define	OBJ_NOSPLIT	0x0010		/* dont split this object */
176#define	OBJ_UMTXDEAD	0x0020		/* umtx pshared was terminated */
177#define OBJ_PIPWNT	0x0040		/* paging in progress wanted */
178#define OBJ_MIGHTBEDIRTY 0x0100		/* object might be dirty, only for vnode */
179#define	OBJ_TMPFS_NODE	0x0200		/* object belongs to tmpfs VREG node */
180#define	OBJ_TMPFS_DIRTY	0x0400		/* dirty tmpfs obj */
181#define	OBJ_COLORED	0x1000		/* pg_color is defined */
182#define	OBJ_ONEMAPPING	0x2000		/* One USE (a single, non-forked) mapping flag */
183#define	OBJ_DISCONNECTWNT 0x4000	/* disconnect from vnode wanted */
184#define	OBJ_TMPFS	0x8000		/* has tmpfs vnode allocated */
185
186/*
187 * Helpers to perform conversion between vm_object page indexes and offsets.
188 * IDX_TO_OFF() converts an index into an offset.
189 * OFF_TO_IDX() converts an offset into an index.  Since offsets are signed
190 *   by default, the sign propagation in OFF_TO_IDX(), when applied to
191 *   negative offsets, is intentional and returns a vm_object page index
192 *   that cannot be created by a userspace mapping.
193 * UOFF_TO_IDX() treats the offset as an unsigned value and converts it
194 *   into an index accordingly.  Use it only when the full range of offset
195 *   values are allowed.  Currently, this only applies to device mappings.
196 * OBJ_MAX_SIZE specifies the maximum page index corresponding to the
197 *   maximum unsigned offset.
198 */
199#define	IDX_TO_OFF(idx) (((vm_ooffset_t)(idx)) << PAGE_SHIFT)
200#define	OFF_TO_IDX(off) ((vm_pindex_t)(((vm_ooffset_t)(off)) >> PAGE_SHIFT))
201#define	UOFF_TO_IDX(off) (((vm_pindex_t)(off)) >> PAGE_SHIFT)
202#define	OBJ_MAX_SIZE	(UOFF_TO_IDX(UINT64_MAX) + 1)
203
204#ifdef	_KERNEL
205
206#define OBJPC_SYNC	0x1			/* sync I/O */
207#define OBJPC_INVAL	0x2			/* invalidate */
208#define OBJPC_NOSYNC	0x4			/* skip if VPO_NOSYNC */
209
210/*
211 * The following options are supported by vm_object_page_remove().
212 */
213#define	OBJPR_CLEANONLY	0x1		/* Don't remove dirty pages. */
214#define	OBJPR_NOTMAPPED	0x2		/* Don't unmap pages. */
215
216TAILQ_HEAD(object_q, vm_object);
217
218extern struct object_q vm_object_list;	/* list of allocated objects */
219extern struct mtx vm_object_list_mtx;	/* lock for object list and count */
220
221extern struct vm_object kernel_object_store;
222extern struct vm_object kmem_object_store;
223
224#define	kernel_object	(&kernel_object_store)
225#define	kmem_object	(&kmem_object_store)
226
227#define	VM_OBJECT_ASSERT_LOCKED(object)					\
228	rw_assert(&(object)->lock, RA_LOCKED)
229#define	VM_OBJECT_ASSERT_RLOCKED(object)				\
230	rw_assert(&(object)->lock, RA_RLOCKED)
231#define	VM_OBJECT_ASSERT_WLOCKED(object)				\
232	rw_assert(&(object)->lock, RA_WLOCKED)
233#define	VM_OBJECT_ASSERT_UNLOCKED(object)				\
234	rw_assert(&(object)->lock, RA_UNLOCKED)
235#define	VM_OBJECT_LOCK_DOWNGRADE(object)				\
236	rw_downgrade(&(object)->lock)
237#define	VM_OBJECT_RLOCK(object)						\
238	rw_rlock(&(object)->lock)
239#define	VM_OBJECT_RUNLOCK(object)					\
240	rw_runlock(&(object)->lock)
241#define	VM_OBJECT_SLEEP(object, wchan, pri, wmesg, timo)		\
242	rw_sleep((wchan), &(object)->lock, (pri), (wmesg), (timo))
243#define	VM_OBJECT_TRYRLOCK(object)					\
244	rw_try_rlock(&(object)->lock)
245#define	VM_OBJECT_TRYWLOCK(object)					\
246	rw_try_wlock(&(object)->lock)
247#define	VM_OBJECT_TRYUPGRADE(object)					\
248	rw_try_upgrade(&(object)->lock)
249#define	VM_OBJECT_WLOCK(object)						\
250	rw_wlock(&(object)->lock)
251#define	VM_OBJECT_WOWNED(object)					\
252	rw_wowned(&(object)->lock)
253#define	VM_OBJECT_WUNLOCK(object)					\
254	rw_wunlock(&(object)->lock)
255
256/*
257 *	The object must be locked or thread private.
258 */
259static __inline void
260vm_object_set_flag(vm_object_t object, u_short bits)
261{
262
263	object->flags |= bits;
264}
265
266/*
267 *	Conditionally set the object's color, which (1) enables the allocation
268 *	of physical memory reservations for anonymous objects and larger-than-
269 *	superpage-sized named objects and (2) determines the first page offset
270 *	within the object at which a reservation may be allocated.  In other
271 *	words, the color determines the alignment of the object with respect
272 *	to the largest superpage boundary.  When mapping named objects, like
273 *	files or POSIX shared memory objects, the color should be set to zero
274 *	before a virtual address is selected for the mapping.  In contrast,
275 *	for anonymous objects, the color may be set after the virtual address
276 *	is selected.
277 *
278 *	The object must be locked.
279 */
280static __inline void
281vm_object_color(vm_object_t object, u_short color)
282{
283
284	if ((object->flags & OBJ_COLORED) == 0) {
285		object->pg_color = color;
286		object->flags |= OBJ_COLORED;
287	}
288}
289
290void vm_object_clear_flag(vm_object_t object, u_short bits);
291void vm_object_pip_add(vm_object_t object, short i);
292void vm_object_pip_subtract(vm_object_t object, short i);
293void vm_object_pip_wakeup(vm_object_t object);
294void vm_object_pip_wakeupn(vm_object_t object, short i);
295void vm_object_pip_wait(vm_object_t object, char *waitid);
296
297void umtx_shm_object_init(vm_object_t object);
298void umtx_shm_object_terminated(vm_object_t object);
299extern int umtx_shm_vnobj_persistent;
300
301vm_object_t vm_object_allocate (objtype_t, vm_pindex_t);
302boolean_t vm_object_coalesce(vm_object_t, vm_ooffset_t, vm_size_t, vm_size_t,
303   boolean_t);
304void vm_object_collapse (vm_object_t);
305void vm_object_deallocate (vm_object_t);
306void vm_object_destroy (vm_object_t);
307void vm_object_terminate (vm_object_t);
308void vm_object_set_writeable_dirty (vm_object_t);
309void vm_object_init (void);
310void vm_object_madvise(vm_object_t, vm_pindex_t, vm_pindex_t, int);
311boolean_t vm_object_page_clean(vm_object_t object, vm_ooffset_t start,
312    vm_ooffset_t end, int flags);
313void vm_object_page_noreuse(vm_object_t object, vm_pindex_t start,
314    vm_pindex_t end);
315void vm_object_page_remove(vm_object_t object, vm_pindex_t start,
316    vm_pindex_t end, int options);
317boolean_t vm_object_populate(vm_object_t, vm_pindex_t, vm_pindex_t);
318void vm_object_print(long addr, boolean_t have_addr, long count, char *modif);
319void vm_object_reference (vm_object_t);
320void vm_object_reference_locked(vm_object_t);
321int  vm_object_set_memattr(vm_object_t object, vm_memattr_t memattr);
322void vm_object_shadow (vm_object_t *, vm_ooffset_t *, vm_size_t);
323void vm_object_split(vm_map_entry_t);
324boolean_t vm_object_sync(vm_object_t, vm_ooffset_t, vm_size_t, boolean_t,
325    boolean_t);
326void vm_object_unwire(vm_object_t object, vm_ooffset_t offset,
327    vm_size_t length, uint8_t queue);
328struct vnode *vm_object_vnode(vm_object_t object);
329#endif				/* _KERNEL */
330
331#endif				/* _VM_OBJECT_ */
332