2 * SPDX-License-Identifier: (BSD-3-Clause AND MIT-CMU)
4 * Copyright (c) 1991, 1993
5 * The Regents of the University of California. All rights reserved.
7 * This code is derived from software contributed to Berkeley by
8 * The Mach Operating System project at Carnegie-Mellon University.
10 * Redistribution and use in source and binary forms, with or without
11 * modification, are permitted provided that the following conditions
13 * 1. Redistributions of source code must retain the above copyright
14 * notice, this list of conditions and the following disclaimer.
15 * 2. Redistributions in binary form must reproduce the above copyright
16 * notice, this list of conditions and the following disclaimer in the
17 * documentation and/or other materials provided with the distribution.
18 * 3. Neither the name of the University nor the names of its contributors
19 * may be used to endorse or promote products derived from this software
20 * without specific prior written permission.
22 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
23 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
24 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
25 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
26 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
27 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
28 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
29 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
30 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
31 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
34 * @(#)vm_map.h 8.9 (Berkeley) 5/17/95
37 * Copyright (c) 1987, 1990 Carnegie-Mellon University.
38 * All rights reserved.
40 * Authors: Avadis Tevanian, Jr., Michael Wayne Young
42 * Permission to use, copy, modify and distribute this software and
43 * its documentation is hereby granted, provided that both the copyright
44 * notice and this permission notice appear in all copies of the
45 * software, derivative works or modified versions, and any portions
46 * thereof, and that both notices appear in supporting documentation.
48 * CARNEGIE MELLON ALLOWS FREE USE OF THIS SOFTWARE IN ITS "AS IS"
49 * CONDITION. CARNEGIE MELLON DISCLAIMS ANY LIABILITY OF ANY KIND
50 * FOR ANY DAMAGES WHATSOEVER RESULTING FROM THE USE OF THIS SOFTWARE.
52 * Carnegie Mellon requests users of this software to return to
54 * Software Distribution Coordinator or Software.Distribution@CS.CMU.EDU
55 * School of Computer Science
56 * Carnegie Mellon University
57 * Pittsburgh PA 15213-3890
59 * any improvements or extensions that they make and grant Carnegie the
60 * rights to redistribute these changes.
66 * Virtual memory map module definitions.
73 #include <sys/_mutex.h>
78 * vm_map_t the high-level address map data structure.
79 * vm_map_entry_t an entry in an address map.
82 typedef u_char vm_flags_t;
83 typedef u_int vm_eflags_t;
86 * Objects which live in maps may be either VM objects, or
87 * another map (called a "sharing map") which denotes read-write
88 * sharing with other maps.
91 struct vm_object *vm_object; /* object object */
92 struct vm_map *sub_map; /* belongs to another map */
96 * Address map entries consist of start and end addresses,
97 * a VM object (or sharing map) and offset into that object,
98 * and user-exported inheritance and protection information.
99 * Also included is control information for virtual copy operations.
101 struct vm_map_entry {
102 struct vm_map_entry *left; /* left child or previous entry */
103 struct vm_map_entry *right; /* right child or next entry */
104 vm_offset_t start; /* start address */
105 vm_offset_t end; /* end address */
106 vm_offset_t next_read; /* vaddr of the next sequential read */
107 vm_size_t max_free; /* max free space in subtree */
108 union vm_map_object object; /* object I point to */
109 vm_ooffset_t offset; /* offset into object */
110 vm_eflags_t eflags; /* map entry flags */
111 vm_prot_t protection; /* protection code */
112 vm_prot_t max_protection; /* maximum protection */
113 vm_inherit_t inheritance; /* inheritance */
114 uint8_t read_ahead; /* pages in the read-ahead window */
115 int wired_count; /* can be paged if = 0 */
116 struct ucred *cred; /* tmp storage for creator ref */
117 struct thread *wiring_thread;
120 #define MAP_ENTRY_NOSYNC 0x00000001
121 #define MAP_ENTRY_IS_SUB_MAP 0x00000002
122 #define MAP_ENTRY_COW 0x00000004
123 #define MAP_ENTRY_NEEDS_COPY 0x00000008
124 #define MAP_ENTRY_NOFAULT 0x00000010
125 #define MAP_ENTRY_USER_WIRED 0x00000020
127 #define MAP_ENTRY_BEHAV_NORMAL 0x00000000 /* default behavior */
128 #define MAP_ENTRY_BEHAV_SEQUENTIAL 0x00000040 /* expect sequential
130 #define MAP_ENTRY_BEHAV_RANDOM 0x00000080 /* expect random
132 #define MAP_ENTRY_BEHAV_RESERVED 0x000000c0 /* future use */
133 #define MAP_ENTRY_BEHAV_MASK 0x000000c0
134 #define MAP_ENTRY_IN_TRANSITION 0x00000100 /* entry being
136 #define MAP_ENTRY_NEEDS_WAKEUP 0x00000200 /* waiters in
138 #define MAP_ENTRY_NOCOREDUMP 0x00000400 /* don't include in
140 #define MAP_ENTRY_VN_EXEC 0x00000800 /* text vnode mapping */
141 #define MAP_ENTRY_GROWS_DOWN 0x00001000 /* top-down stacks */
142 #define MAP_ENTRY_GROWS_UP 0x00002000 /* bottom-up stacks */
144 #define MAP_ENTRY_WIRE_SKIPPED 0x00004000
145 #define MAP_ENTRY_WRITECNT 0x00008000 /* tracked writeable
147 #define MAP_ENTRY_GUARD 0x00010000
148 #define MAP_ENTRY_STACK_GAP_DN 0x00020000
149 #define MAP_ENTRY_STACK_GAP_UP 0x00040000
150 #define MAP_ENTRY_HEADER 0x00080000
153 static __inline u_char
154 vm_map_entry_behavior(vm_map_entry_t entry)
156 return (entry->eflags & MAP_ENTRY_BEHAV_MASK);
160 vm_map_entry_user_wired_count(vm_map_entry_t entry)
162 if (entry->eflags & MAP_ENTRY_USER_WIRED)
168 vm_map_entry_system_wired_count(vm_map_entry_t entry)
170 return (entry->wired_count - vm_map_entry_user_wired_count(entry));
175 * A map is a set of map entries. These map entries are
176 * organized as a threaded binary search tree. Both structures
177 * are ordered based upon the start and end addresses contained
178 * within each map entry. The largest gap between an entry in a
179 * subtree and one of its neighbors is saved in the max_free
180 * field, and that field is updated when the tree is
183 * Sleator and Tarjan's top-down splay algorithm is employed to
184 * control height imbalance in the binary search tree.
186 * The map's min offset value is stored in map->header.end, and
187 * its max offset value is stored in map->header.start. These
188 * values act as sentinels for any forward or backward address
189 * scan of the list. The right and left fields of the map
190 * header point to the first and list map entries. The map
191 * header has a special value for the eflags field,
192 * MAP_ENTRY_HEADER, that is set initially, is never changed,
193 * and prevents an eflags match of the header with any other map
197 * (c) const until freed
200 struct vm_map_entry header; /* List of entries */
201 struct sx lock; /* Lock for map data */
202 struct mtx system_mtx;
203 int nentries; /* Number of entries */
204 vm_size_t size; /* virtual size */
205 u_int timestamp; /* Version number */
207 u_char system_map; /* (c) Am I a system map? */
208 vm_flags_t flags; /* flags for this vm_map */
209 vm_map_entry_t root; /* Root of a binary search tree */
210 pmap_t pmap; /* (c) Physical map */
211 vm_offset_t anon_loc;
221 #define MAP_WIREFUTURE 0x01 /* wire all future pages */
222 #define MAP_BUSY_WAKEUP 0x02
223 #define MAP_IS_SUB_MAP 0x04 /* has parent */
224 #define MAP_ASLR 0x08 /* enabled ASLR */
225 #define MAP_ASLR_IGNSTART 0x10
228 #if defined(KLD_MODULE) && !defined(KLD_TIED)
229 #define vm_map_max(map) vm_map_max_KBI((map))
230 #define vm_map_min(map) vm_map_min_KBI((map))
231 #define vm_map_pmap(map) vm_map_pmap_KBI((map))
233 static __inline vm_offset_t
234 vm_map_max(const struct vm_map *map)
237 return (map->header.start);
240 static __inline vm_offset_t
241 vm_map_min(const struct vm_map *map)
244 return (map->header.end);
247 static __inline pmap_t
248 vm_map_pmap(vm_map_t map)
254 vm_map_modflags(vm_map_t map, vm_flags_t set, vm_flags_t clear)
256 map->flags = (map->flags | set) & ~clear;
258 #endif /* KLD_MODULE */
262 * Shareable process virtual address space.
265 * (c) const until freed
268 struct vm_map vm_map; /* VM address map */
269 struct shmmap_state *vm_shm; /* SYS5 shared memory private data XXX */
270 segsz_t vm_swrss; /* resident set size before last swap */
271 segsz_t vm_tsize; /* text size (pages) XXX */
272 segsz_t vm_dsize; /* data size (pages) XXX */
273 segsz_t vm_ssize; /* stack size (pages) */
274 caddr_t vm_taddr; /* (c) user virtual address of text */
275 caddr_t vm_daddr; /* (c) user virtual address of data */
276 caddr_t vm_maxsaddr; /* user VA at max stack growth */
277 volatile int vm_refcnt; /* number of references */
279 * Keep the PMAP last, so that CPU-specific variations of that
280 * structure on a single architecture don't result in offset
281 * variations of the machine-independent fields in the vmspace.
283 struct pmap vm_pmap; /* private physical map */
287 static __inline pmap_t
288 vmspace_pmap(struct vmspace *vmspace)
290 return &vmspace->vm_pmap;
296 * Macros: vm_map_lock, etc.
298 * Perform locking on the data portion of a map. Note that
299 * these macros mimic procedure calls returning void. The
300 * semicolon is supplied by the user of these macros, not
301 * by the macros themselves. The macros can safely be used
302 * as unbraced elements in a higher level statement.
305 void _vm_map_lock(vm_map_t map, const char *file, int line);
306 void _vm_map_unlock(vm_map_t map, const char *file, int line);
307 int _vm_map_unlock_and_wait(vm_map_t map, int timo, const char *file, int line);
308 void _vm_map_lock_read(vm_map_t map, const char *file, int line);
309 void _vm_map_unlock_read(vm_map_t map, const char *file, int line);
310 int _vm_map_trylock(vm_map_t map, const char *file, int line);
311 int _vm_map_trylock_read(vm_map_t map, const char *file, int line);
312 int _vm_map_lock_upgrade(vm_map_t map, const char *file, int line);
313 void _vm_map_lock_downgrade(vm_map_t map, const char *file, int line);
314 int vm_map_locked(vm_map_t map);
315 void vm_map_wakeup(vm_map_t map);
316 void vm_map_busy(vm_map_t map);
317 void vm_map_unbusy(vm_map_t map);
318 void vm_map_wait_busy(vm_map_t map);
319 vm_offset_t vm_map_max_KBI(const struct vm_map *map);
320 vm_offset_t vm_map_min_KBI(const struct vm_map *map);
321 pmap_t vm_map_pmap_KBI(vm_map_t map);
323 #define vm_map_lock(map) _vm_map_lock(map, LOCK_FILE, LOCK_LINE)
324 #define vm_map_unlock(map) _vm_map_unlock(map, LOCK_FILE, LOCK_LINE)
325 #define vm_map_unlock_and_wait(map, timo) \
326 _vm_map_unlock_and_wait(map, timo, LOCK_FILE, LOCK_LINE)
327 #define vm_map_lock_read(map) _vm_map_lock_read(map, LOCK_FILE, LOCK_LINE)
328 #define vm_map_unlock_read(map) _vm_map_unlock_read(map, LOCK_FILE, LOCK_LINE)
329 #define vm_map_trylock(map) _vm_map_trylock(map, LOCK_FILE, LOCK_LINE)
330 #define vm_map_trylock_read(map) \
331 _vm_map_trylock_read(map, LOCK_FILE, LOCK_LINE)
332 #define vm_map_lock_upgrade(map) \
333 _vm_map_lock_upgrade(map, LOCK_FILE, LOCK_LINE)
334 #define vm_map_lock_downgrade(map) \
335 _vm_map_lock_downgrade(map, LOCK_FILE, LOCK_LINE)
337 long vmspace_resident_count(struct vmspace *vmspace);
341 /* XXX: number of kernel maps to statically allocate */
345 * Copy-on-write flags for vm_map operations
347 #define MAP_INHERIT_SHARE 0x00000001
348 #define MAP_COPY_ON_WRITE 0x00000002
349 #define MAP_NOFAULT 0x00000004
350 #define MAP_PREFAULT 0x00000008
351 #define MAP_PREFAULT_PARTIAL 0x00000010
352 #define MAP_DISABLE_SYNCER 0x00000020
353 #define MAP_CHECK_EXCL 0x00000040
354 #define MAP_CREATE_GUARD 0x00000080
355 #define MAP_DISABLE_COREDUMP 0x00000100
356 #define MAP_PREFAULT_MADVISE 0x00000200 /* from (user) madvise request */
357 #define MAP_WRITECOUNT 0x00000400
358 #define MAP_REMAP 0x00000800
359 #define MAP_STACK_GROWS_DOWN 0x00001000
360 #define MAP_STACK_GROWS_UP 0x00002000
361 #define MAP_ACC_CHARGED 0x00004000
362 #define MAP_ACC_NO_CHARGE 0x00008000
363 #define MAP_CREATE_STACK_GAP_UP 0x00010000
364 #define MAP_CREATE_STACK_GAP_DN 0x00020000
365 #define MAP_VN_EXEC 0x00040000
368 * vm_fault option flags
370 #define VM_FAULT_NORMAL 0 /* Nothing special */
371 #define VM_FAULT_WIRE 1 /* Wire the mapped page */
372 #define VM_FAULT_DIRTY 2 /* Dirty the page; use w/VM_PROT_COPY */
375 * Initially, mappings are slightly sequential. The maximum window size must
376 * account for the map entry's "read_ahead" field being defined as an uint8_t.
378 #define VM_FAULT_READ_AHEAD_MIN 7
379 #define VM_FAULT_READ_AHEAD_INIT 15
380 #define VM_FAULT_READ_AHEAD_MAX min(atop(MAXPHYS) - 1, UINT8_MAX)
383 * The following "find_space" options are supported by vm_map_find().
385 * For VMFS_ALIGNED_SPACE, the desired alignment is specified to
386 * the macro argument as log base 2 of the desired alignment.
388 #define VMFS_NO_SPACE 0 /* don't find; use the given range */
389 #define VMFS_ANY_SPACE 1 /* find a range with any alignment */
390 #define VMFS_OPTIMAL_SPACE 2 /* find a range with optimal alignment*/
391 #define VMFS_SUPER_SPACE 3 /* find a superpage-aligned range */
392 #define VMFS_ALIGNED_SPACE(x) ((x) << 8) /* find a range with fixed alignment */
395 * vm_map_wire and vm_map_unwire option flags
397 #define VM_MAP_WIRE_SYSTEM 0 /* wiring in a kernel map */
398 #define VM_MAP_WIRE_USER 1 /* wiring in a user map */
400 #define VM_MAP_WIRE_NOHOLES 0 /* region must not have holes */
401 #define VM_MAP_WIRE_HOLESOK 2 /* region may have holes */
403 #define VM_MAP_WIRE_WRITE 4 /* Validate writable. */
405 typedef int vm_map_entry_reader(void *token, vm_map_entry_t addr,
406 vm_map_entry_t dest);
410 * Find the successor of a map_entry, using a reader to dereference pointers.
411 * '*clone' is a copy of a vm_map entry. 'reader' is used to copy a map entry
412 * at some address into '*clone'. Change *clone to a copy of the next map
413 * entry, and return the address of that entry, or NULL if copying has failed.
415 * This function is made available to user-space code that needs to traverse
418 static inline vm_map_entry_t
419 vm_map_entry_read_succ(void *token, struct vm_map_entry *const clone,
420 vm_map_entry_reader reader)
422 vm_map_entry_t after, backup;
425 after = clone->right;
426 start = clone->start;
427 if (!reader(token, after, clone))
429 backup = clone->left;
430 if (!reader(token, backup, clone))
432 if (clone->start > start) {
435 backup = clone->left;
436 if (!reader(token, backup, clone))
438 } while (clone->start != start);
440 if (!reader(token, after, clone))
444 #endif /* ! _KERNEL */
447 boolean_t vm_map_check_protection (vm_map_t, vm_offset_t, vm_offset_t, vm_prot_t);
448 vm_map_t vm_map_create(pmap_t, vm_offset_t, vm_offset_t);
449 int vm_map_delete(vm_map_t, vm_offset_t, vm_offset_t);
450 int vm_map_find(vm_map_t, vm_object_t, vm_ooffset_t, vm_offset_t *, vm_size_t,
451 vm_offset_t, int, vm_prot_t, vm_prot_t, int);
452 int vm_map_find_min(vm_map_t, vm_object_t, vm_ooffset_t, vm_offset_t *,
453 vm_size_t, vm_offset_t, vm_offset_t, int, vm_prot_t, vm_prot_t, int);
454 int vm_map_fixed(vm_map_t, vm_object_t, vm_ooffset_t, vm_offset_t, vm_size_t,
455 vm_prot_t, vm_prot_t, int);
456 vm_offset_t vm_map_findspace(vm_map_t, vm_offset_t, vm_size_t);
457 int vm_map_inherit (vm_map_t, vm_offset_t, vm_offset_t, vm_inherit_t);
458 void vm_map_init(vm_map_t, pmap_t, vm_offset_t, vm_offset_t);
459 int 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);
460 int vm_map_lookup (vm_map_t *, vm_offset_t, vm_prot_t, vm_map_entry_t *, vm_object_t *,
461 vm_pindex_t *, vm_prot_t *, boolean_t *);
462 int vm_map_lookup_locked(vm_map_t *, vm_offset_t, vm_prot_t, vm_map_entry_t *, vm_object_t *,
463 vm_pindex_t *, vm_prot_t *, boolean_t *);
464 void vm_map_lookup_done (vm_map_t, vm_map_entry_t);
465 boolean_t vm_map_lookup_entry (vm_map_t, vm_offset_t, vm_map_entry_t *);
467 static inline vm_map_entry_t
468 vm_map_entry_first(vm_map_t map)
471 return (map->header.right);
474 static inline vm_map_entry_t
475 vm_map_entry_succ(vm_map_entry_t entry)
477 vm_map_entry_t after;
479 after = entry->right;
480 if (after->left->start > entry->start) {
483 while (after->left != entry);
488 #define VM_MAP_ENTRY_FOREACH(it, map) \
489 for ((it) = vm_map_entry_first(map); \
490 (it) != &(map)->header; \
491 (it) = vm_map_entry_succ(it))
492 int vm_map_protect (vm_map_t, vm_offset_t, vm_offset_t, vm_prot_t, boolean_t);
493 int vm_map_remove (vm_map_t, vm_offset_t, vm_offset_t);
494 void vm_map_try_merge_entries(vm_map_t map, vm_map_entry_t prev,
495 vm_map_entry_t entry);
496 void vm_map_startup (void);
497 int vm_map_submap (vm_map_t, vm_offset_t, vm_offset_t, vm_map_t);
498 int vm_map_sync(vm_map_t, vm_offset_t, vm_offset_t, boolean_t, boolean_t);
499 int vm_map_madvise (vm_map_t, vm_offset_t, vm_offset_t, int);
500 int vm_map_stack (vm_map_t, vm_offset_t, vm_size_t, vm_prot_t, vm_prot_t, int);
501 int vm_map_unwire(vm_map_t map, vm_offset_t start, vm_offset_t end,
503 int vm_map_wire(vm_map_t map, vm_offset_t start, vm_offset_t end, int flags);
504 int vm_map_wire_locked(vm_map_t map, vm_offset_t start, vm_offset_t end,
506 long vmspace_swap_count(struct vmspace *vmspace);
507 void vm_map_entry_set_vnode_text(vm_map_entry_t entry, bool add);
509 #endif /* _VM_MAP_ */