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 *	@(#)vm_map.h	8.9 (Berkeley) 5/17/95
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 * $FreeBSD$
61 */
62
63/*
64 *	Virtual memory map module definitions.
65 */
66#ifndef	_VM_MAP_
67#define	_VM_MAP_
68
69#include <sys/lock.h>
70#include <sys/sx.h>
71#include <sys/_mutex.h>
72
73/*
74 *	Types defined:
75 *
76 *	vm_map_t		the high-level address map data structure.
77 *	vm_map_entry_t		an entry in an address map.
78 */
79
80typedef u_char vm_flags_t;
81typedef u_int vm_eflags_t;
82
83/*
84 *	Objects which live in maps may be either VM objects, or
85 *	another map (called a "sharing map") which denotes read-write
86 *	sharing with other maps.
87 */
88union vm_map_object {
89	struct vm_object *vm_object;	/* object object */
90	struct vm_map *sub_map;		/* belongs to another map */
91};
92
93/*
94 *	Address map entries consist of start and end addresses,
95 *	a VM object (or sharing map) and offset into that object,
96 *	and user-exported inheritance and protection information.
97 *	Also included is control information for virtual copy operations.
98 */
99struct vm_map_entry {
100	struct vm_map_entry *prev;	/* previous entry */
101	struct vm_map_entry *next;	/* next entry */
102	struct vm_map_entry *left;	/* left child in binary search tree */
103	struct vm_map_entry *right;	/* right child in binary search tree */
104	vm_offset_t start;		/* start address */
105	vm_offset_t end;		/* end address */
106	vm_offset_t avail_ssize;	/* amt can grow if this is a stack */
107	vm_size_t adj_free;		/* amount of adjacent free space */
108	vm_size_t max_free;		/* max free space in subtree */
109	union vm_map_object object;	/* object I point to */
110	vm_ooffset_t offset;		/* offset into object */
111	vm_eflags_t eflags;		/* map entry flags */
112	vm_prot_t protection;		/* protection code */
113	vm_prot_t max_protection;	/* maximum protection */
114	vm_inherit_t inheritance;	/* inheritance */
115	uint8_t read_ahead;		/* pages in the read-ahead window */
116	int wired_count;		/* can be paged if = 0 */
117	vm_pindex_t next_read;		/* index of the next sequential read */
118	struct ucred *cred;		/* tmp storage for creator ref */
119	struct thread *wiring_thread;
120};
121
122#define MAP_ENTRY_NOSYNC		0x0001
123#define MAP_ENTRY_IS_SUB_MAP		0x0002
124#define MAP_ENTRY_COW			0x0004
125#define MAP_ENTRY_NEEDS_COPY		0x0008
126#define MAP_ENTRY_NOFAULT		0x0010
127#define MAP_ENTRY_USER_WIRED		0x0020
128
129#define MAP_ENTRY_BEHAV_NORMAL		0x0000	/* default behavior */
130#define MAP_ENTRY_BEHAV_SEQUENTIAL	0x0040	/* expect sequential access */
131#define MAP_ENTRY_BEHAV_RANDOM		0x0080	/* expect random access */
132#define MAP_ENTRY_BEHAV_RESERVED	0x00C0	/* future use */
133
134#define MAP_ENTRY_BEHAV_MASK		0x00C0
135
136#define MAP_ENTRY_IN_TRANSITION		0x0100	/* entry being changed */
137#define MAP_ENTRY_NEEDS_WAKEUP		0x0200	/* waiters in transition */
138#define MAP_ENTRY_NOCOREDUMP		0x0400	/* don't include in a core */
139
140#define	MAP_ENTRY_GROWS_DOWN		0x1000	/* Top-down stacks */
141#define	MAP_ENTRY_GROWS_UP		0x2000	/* Bottom-up stacks */
142
143#define	MAP_ENTRY_WIRE_SKIPPED		0x4000
144#define	MAP_ENTRY_VN_WRITECNT		0x8000	/* writeable vnode mapping */
145
146#ifdef	_KERNEL
147static __inline u_char
148vm_map_entry_behavior(vm_map_entry_t entry)
149{
150	return (entry->eflags & MAP_ENTRY_BEHAV_MASK);
151}
152
153static __inline int
154vm_map_entry_user_wired_count(vm_map_entry_t entry)
155{
156	if (entry->eflags & MAP_ENTRY_USER_WIRED)
157		return (1);
158	return (0);
159}
160
161static __inline int
162vm_map_entry_system_wired_count(vm_map_entry_t entry)
163{
164	return (entry->wired_count - vm_map_entry_user_wired_count(entry));
165}
166#endif	/* _KERNEL */
167
168/*
169 *	A map is a set of map entries.  These map entries are
170 *	organized both as a binary search tree and as a doubly-linked
171 *	list.  Both structures are ordered based upon the start and
172 *	end addresses contained within each map entry.  Sleator and
173 *	Tarjan's top-down splay algorithm is employed to control
174 *	height imbalance in the binary search tree.
175 *
176 * List of locks
177 *	(c)	const until freed
178 */
179struct vm_map {
180	struct vm_map_entry header;	/* List of entries */
181	struct sx lock;			/* Lock for map data */
182	struct mtx system_mtx;
183	int nentries;			/* Number of entries */
184	vm_size_t size;			/* virtual size */
185	u_int timestamp;		/* Version number */
186	u_char needs_wakeup;
187	u_char system_map;		/* (c) Am I a system map? */
188	vm_flags_t flags;		/* flags for this vm_map */
189	vm_map_entry_t root;		/* Root of a binary search tree */
190	pmap_t pmap;			/* (c) Physical map */
191#define	min_offset	header.start	/* (c) */
192#define	max_offset	header.end	/* (c) */
193	int busy;
194};
195
196/*
197 * vm_flags_t values
198 */
199#define MAP_WIREFUTURE		0x01	/* wire all future pages */
200#define	MAP_BUSY_WAKEUP		0x02
201
202#ifdef	_KERNEL
203static __inline vm_offset_t
204vm_map_max(const struct vm_map *map)
205{
206	return (map->max_offset);
207}
208
209static __inline vm_offset_t
210vm_map_min(const struct vm_map *map)
211{
212	return (map->min_offset);
213}
214
215static __inline pmap_t
216vm_map_pmap(vm_map_t map)
217{
218	return (map->pmap);
219}
220
221static __inline void
222vm_map_modflags(vm_map_t map, vm_flags_t set, vm_flags_t clear)
223{
224	map->flags = (map->flags | set) & ~clear;
225}
226#endif	/* _KERNEL */
227
228/*
229 * Shareable process virtual address space.
230 *
231 * List of locks
232 *	(c)	const until freed
233 */
234struct vmspace {
235	struct vm_map vm_map;	/* VM address map */
236	struct shmmap_state *vm_shm;	/* SYS5 shared memory private data XXX */
237	segsz_t vm_swrss;	/* resident set size before last swap */
238	segsz_t vm_tsize;	/* text size (pages) XXX */
239	segsz_t vm_dsize;	/* data size (pages) XXX */
240	segsz_t vm_ssize;	/* stack size (pages) */
241	caddr_t vm_taddr;	/* (c) user virtual address of text */
242	caddr_t vm_daddr;	/* (c) user virtual address of data */
243	caddr_t vm_maxsaddr;	/* user VA at max stack growth */
244	volatile int vm_refcnt;	/* number of references */
245	/*
246	 * Keep the PMAP last, so that CPU-specific variations of that
247	 * structure on a single architecture don't result in offset
248	 * variations of the machine-independent fields in the vmspace.
249	 */
250	struct pmap vm_pmap;	/* private physical map */
251};
252
253#ifdef	_KERNEL
254static __inline pmap_t
255vmspace_pmap(struct vmspace *vmspace)
256{
257	return &vmspace->vm_pmap;
258}
259#endif	/* _KERNEL */
260
261#ifdef	_KERNEL
262/*
263 *	Macros:		vm_map_lock, etc.
264 *	Function:
265 *		Perform locking on the data portion of a map.  Note that
266 *		these macros mimic procedure calls returning void.  The
267 *		semicolon is supplied by the user of these macros, not
268 *		by the macros themselves.  The macros can safely be used
269 *		as unbraced elements in a higher level statement.
270 */
271
272void _vm_map_lock(vm_map_t map, const char *file, int line);
273void _vm_map_unlock(vm_map_t map, const char *file, int line);
274int _vm_map_unlock_and_wait(vm_map_t map, int timo, const char *file, int line);
275void _vm_map_lock_read(vm_map_t map, const char *file, int line);
276void _vm_map_unlock_read(vm_map_t map, const char *file, int line);
277int _vm_map_trylock(vm_map_t map, const char *file, int line);
278int _vm_map_trylock_read(vm_map_t map, const char *file, int line);
279int _vm_map_lock_upgrade(vm_map_t map, const char *file, int line);
280void _vm_map_lock_downgrade(vm_map_t map, const char *file, int line);
281int vm_map_locked(vm_map_t map);
282void vm_map_wakeup(vm_map_t map);
283void vm_map_busy(vm_map_t map);
284void vm_map_unbusy(vm_map_t map);
285void vm_map_wait_busy(vm_map_t map);
286
287#define	vm_map_lock(map)	_vm_map_lock(map, LOCK_FILE, LOCK_LINE)
288#define	vm_map_unlock(map)	_vm_map_unlock(map, LOCK_FILE, LOCK_LINE)
289#define	vm_map_unlock_and_wait(map, timo)	\
290			_vm_map_unlock_and_wait(map, timo, LOCK_FILE, LOCK_LINE)
291#define	vm_map_lock_read(map)	_vm_map_lock_read(map, LOCK_FILE, LOCK_LINE)
292#define	vm_map_unlock_read(map)	_vm_map_unlock_read(map, LOCK_FILE, LOCK_LINE)
293#define	vm_map_trylock(map)	_vm_map_trylock(map, LOCK_FILE, LOCK_LINE)
294#define	vm_map_trylock_read(map)	\
295			_vm_map_trylock_read(map, LOCK_FILE, LOCK_LINE)
296#define	vm_map_lock_upgrade(map)	\
297			_vm_map_lock_upgrade(map, LOCK_FILE, LOCK_LINE)
298#define	vm_map_lock_downgrade(map)	\
299			_vm_map_lock_downgrade(map, LOCK_FILE, LOCK_LINE)
300
301long vmspace_resident_count(struct vmspace *vmspace);
302#endif	/* _KERNEL */
303
304
305/* XXX: number of kernel maps and entries to statically allocate */
306#define MAX_KMAP	10
307#define	MAX_KMAPENT	128
308
309/*
310 * Copy-on-write flags for vm_map operations
311 */
312#define MAP_INHERIT_SHARE	0x0001
313#define MAP_COPY_ON_WRITE	0x0002
314#define MAP_NOFAULT		0x0004
315#define MAP_PREFAULT		0x0008
316#define MAP_PREFAULT_PARTIAL	0x0010
317#define MAP_DISABLE_SYNCER	0x0020
318#define MAP_DISABLE_COREDUMP	0x0100
319#define MAP_PREFAULT_MADVISE	0x0200	/* from (user) madvise request */
320#define	MAP_VN_WRITECOUNT	0x0400
321#define	MAP_STACK_GROWS_DOWN	0x1000
322#define	MAP_STACK_GROWS_UP	0x2000
323#define	MAP_ACC_CHARGED		0x4000
324#define	MAP_ACC_NO_CHARGE	0x8000
325
326/*
327 * vm_fault option flags
328 */
329#define VM_FAULT_NORMAL 0		/* Nothing special */
330#define VM_FAULT_CHANGE_WIRING 1	/* Change the wiring as appropriate */
331#define	VM_FAULT_DIRTY 2		/* Dirty the page; use w/VM_PROT_COPY */
332
333/*
334 * Initially, mappings are slightly sequential.  The maximum window size must
335 * account for the map entry's "read_ahead" field being defined as an uint8_t.
336 */
337#define	VM_FAULT_READ_AHEAD_MIN		7
338#define	VM_FAULT_READ_AHEAD_INIT	15
339#define	VM_FAULT_READ_AHEAD_MAX		min(atop(MAXPHYS) - 1, UINT8_MAX)
340
341/*
342 * The following "find_space" options are supported by vm_map_find().
343 *
344 * For VMFS_ALIGNED_SPACE, the desired alignment is specified to
345 * the macro argument as log base 2 of the desired alignment.
346 */
347#define	VMFS_NO_SPACE		0	/* don't find; use the given range */
348#define	VMFS_ANY_SPACE		1	/* find a range with any alignment */
349#define	VMFS_SUPER_SPACE	2	/* find a superpage-aligned range */
350#define	VMFS_ALIGNED_SPACE(x)	((x) << 8) /* find a range with fixed alignment */
351#if defined(__mips__)
352#define	VMFS_TLB_ALIGNED_SPACE	3	/* find a TLB entry aligned range */
353#endif
354#define	VMFS_OPTIMAL_SPACE	4	/* find a range with optimal alignment*/
355
356/*
357 * vm_map_wire and vm_map_unwire option flags
358 */
359#define VM_MAP_WIRE_SYSTEM	0	/* wiring in a kernel map */
360#define VM_MAP_WIRE_USER	1	/* wiring in a user map */
361
362#define VM_MAP_WIRE_NOHOLES	0	/* region must not have holes */
363#define VM_MAP_WIRE_HOLESOK	2	/* region may have holes */
364
365#define VM_MAP_WIRE_WRITE	4	/* Validate writable. */
366
367#ifdef _KERNEL
368boolean_t vm_map_check_protection (vm_map_t, vm_offset_t, vm_offset_t, vm_prot_t);
369vm_map_t vm_map_create(pmap_t, vm_offset_t, vm_offset_t);
370int vm_map_delete(vm_map_t, vm_offset_t, vm_offset_t);
371int vm_map_find(vm_map_t, vm_object_t, vm_ooffset_t, vm_offset_t *, vm_size_t,
372    int, vm_prot_t, vm_prot_t, int);
373int vm_map_fixed(vm_map_t, vm_object_t, vm_ooffset_t, vm_offset_t, vm_size_t,
374    vm_prot_t, vm_prot_t, int);
375int vm_map_findspace (vm_map_t, vm_offset_t, vm_size_t, vm_offset_t *);
376int vm_map_inherit (vm_map_t, vm_offset_t, vm_offset_t, vm_inherit_t);
377void vm_map_init(vm_map_t, pmap_t, vm_offset_t, vm_offset_t);
378int vm_map_insert (vm_map_t, vm_object_t, vm_ooffset_t, vm_offset_t, vm_offset_t, vm_prot_t, vm_prot_t, int);
379int vm_map_lookup (vm_map_t *, vm_offset_t, vm_prot_t, vm_map_entry_t *, vm_object_t *,
380    vm_pindex_t *, vm_prot_t *, boolean_t *);
381int vm_map_lookup_locked(vm_map_t *, vm_offset_t, vm_prot_t, vm_map_entry_t *, vm_object_t *,
382    vm_pindex_t *, vm_prot_t *, boolean_t *);
383void vm_map_lookup_done (vm_map_t, vm_map_entry_t);
384boolean_t vm_map_lookup_entry (vm_map_t, vm_offset_t, vm_map_entry_t *);
385void vm_map_pmap_enter(vm_map_t map, vm_offset_t addr, vm_prot_t prot,
386    vm_object_t object, vm_pindex_t pindex, vm_size_t size, int flags);
387int vm_map_protect (vm_map_t, vm_offset_t, vm_offset_t, vm_prot_t, boolean_t);
388int vm_map_remove (vm_map_t, vm_offset_t, vm_offset_t);
389void vm_map_startup (void);
390int vm_map_submap (vm_map_t, vm_offset_t, vm_offset_t, vm_map_t);
391int vm_map_sync(vm_map_t, vm_offset_t, vm_offset_t, boolean_t, boolean_t);
392int vm_map_madvise (vm_map_t, vm_offset_t, vm_offset_t, int);
393void vm_map_simplify_entry (vm_map_t, vm_map_entry_t);
394void vm_init2 (void);
395int vm_map_stack (vm_map_t, vm_offset_t, vm_size_t, vm_prot_t, vm_prot_t, int);
396int vm_map_growstack (struct proc *p, vm_offset_t addr);
397int vm_map_unwire(vm_map_t map, vm_offset_t start, vm_offset_t end,
398    int flags);
399int vm_map_wire(vm_map_t map, vm_offset_t start, vm_offset_t end,
400    int flags);
401long vmspace_swap_count(struct vmspace *vmspace);
402#endif				/* _KERNEL */
403#endif				/* _VM_MAP_ */
404