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 * For stack gap map entries (MAP_ENTRY_GUARD | MAP_ENTRY_GROWS_DOWN
102 * or UP), the next_read member is reused as the stack_guard_page
103 * storage, and offset is the stack protection.
105 struct vm_map_entry {
106 struct vm_map_entry *left; /* left child or previous entry */
107 struct vm_map_entry *right; /* right child or next entry */
108 vm_offset_t start; /* start address */
109 vm_offset_t end; /* end address */
110 vm_offset_t next_read; /* vaddr of the next sequential read */
111 vm_size_t max_free; /* max free space in subtree */
112 union vm_map_object object; /* object I point to */
113 vm_ooffset_t offset; /* offset into object */
114 vm_eflags_t eflags; /* map entry flags */
115 vm_prot_t protection; /* protection code */
116 vm_prot_t max_protection; /* maximum protection */
117 vm_inherit_t inheritance; /* inheritance */
118 uint8_t read_ahead; /* pages in the read-ahead window */
119 int wired_count; /* can be paged if = 0 */
120 struct ucred *cred; /* tmp storage for creator ref */
121 struct thread *wiring_thread;
124 #define MAP_ENTRY_NOSYNC 0x00000001
125 #define MAP_ENTRY_IS_SUB_MAP 0x00000002
126 #define MAP_ENTRY_COW 0x00000004
127 #define MAP_ENTRY_NEEDS_COPY 0x00000008
128 #define MAP_ENTRY_NOFAULT 0x00000010
129 #define MAP_ENTRY_USER_WIRED 0x00000020
131 #define MAP_ENTRY_BEHAV_NORMAL 0x00000000 /* default behavior */
132 #define MAP_ENTRY_BEHAV_SEQUENTIAL 0x00000040 /* expect sequential
134 #define MAP_ENTRY_BEHAV_RANDOM 0x00000080 /* expect random
136 #define MAP_ENTRY_BEHAV_RESERVED 0x000000c0 /* future use */
137 #define MAP_ENTRY_BEHAV_MASK 0x000000c0
138 #define MAP_ENTRY_IN_TRANSITION 0x00000100 /* entry being
140 #define MAP_ENTRY_NEEDS_WAKEUP 0x00000200 /* waiters in
142 #define MAP_ENTRY_NOCOREDUMP 0x00000400 /* don't include in
144 #define MAP_ENTRY_VN_EXEC 0x00000800 /* text vnode mapping */
145 #define MAP_ENTRY_GROWS_DOWN 0x00001000 /* top-down stacks */
146 #define MAP_ENTRY_GROWS_UP 0x00002000 /* bottom-up stacks */
148 #define MAP_ENTRY_WIRE_SKIPPED 0x00004000
149 #define MAP_ENTRY_WRITECNT 0x00008000 /* tracked writeable
151 #define MAP_ENTRY_GUARD 0x00010000
152 #define MAP_ENTRY_STACK_GAP_DN 0x00020000
153 #define MAP_ENTRY_STACK_GAP_UP 0x00040000
154 #define MAP_ENTRY_HEADER 0x00080000
156 #define MAP_ENTRY_SPLIT_BOUNDARY_MASK 0x00300000
157 #define MAP_ENTRY_SPLIT_BOUNDARY_SHIFT 20
158 #define MAP_ENTRY_SPLIT_BOUNDARY_INDEX(entry) \
159 (((entry)->eflags & MAP_ENTRY_SPLIT_BOUNDARY_MASK) >> \
160 MAP_ENTRY_SPLIT_BOUNDARY_SHIFT)
163 static __inline u_char
164 vm_map_entry_behavior(vm_map_entry_t entry)
166 return (entry->eflags & MAP_ENTRY_BEHAV_MASK);
170 vm_map_entry_user_wired_count(vm_map_entry_t entry)
172 if (entry->eflags & MAP_ENTRY_USER_WIRED)
178 vm_map_entry_system_wired_count(vm_map_entry_t entry)
180 return (entry->wired_count - vm_map_entry_user_wired_count(entry));
185 * A map is a set of map entries. These map entries are
186 * organized as a threaded binary search tree. Both structures
187 * are ordered based upon the start and end addresses contained
188 * within each map entry. The largest gap between an entry in a
189 * subtree and one of its neighbors is saved in the max_free
190 * field, and that field is updated when the tree is
193 * Sleator and Tarjan's top-down splay algorithm is employed to
194 * control height imbalance in the binary search tree.
196 * The map's min offset value is stored in map->header.end, and
197 * its max offset value is stored in map->header.start. These
198 * values act as sentinels for any forward or backward address
199 * scan of the list. The right and left fields of the map
200 * header point to the first and list map entries. The map
201 * header has a special value for the eflags field,
202 * MAP_ENTRY_HEADER, that is set initially, is never changed,
203 * and prevents an eflags match of the header with any other map
207 * (c) const until freed
210 struct vm_map_entry header; /* List of entries */
211 struct sx lock; /* Lock for map data */
212 struct mtx system_mtx;
213 int nentries; /* Number of entries */
214 vm_size_t size; /* virtual size */
215 u_int timestamp; /* Version number */
217 u_char system_map; /* (c) Am I a system map? */
218 vm_flags_t flags; /* flags for this vm_map */
219 vm_map_entry_t root; /* Root of a binary search tree */
220 pmap_t pmap; /* (c) Physical map */
221 vm_offset_t anon_loc;
231 #define MAP_WIREFUTURE 0x01 /* wire all future pages */
232 #define MAP_BUSY_WAKEUP 0x02 /* thread(s) waiting on busy state */
233 #define MAP_IS_SUB_MAP 0x04 /* has parent */
234 #define MAP_ASLR 0x08 /* enabled ASLR */
235 #define MAP_ASLR_IGNSTART 0x10 /* ASLR ignores data segment */
236 #define MAP_REPLENISH 0x20 /* kmapent zone needs to be refilled */
237 #define MAP_WXORX 0x40 /* enforce W^X */
238 #define MAP_ASLR_STACK 0x80 /* stack location is randomized */
241 #if defined(KLD_MODULE) && !defined(KLD_TIED)
242 #define vm_map_max(map) vm_map_max_KBI((map))
243 #define vm_map_min(map) vm_map_min_KBI((map))
244 #define vm_map_pmap(map) vm_map_pmap_KBI((map))
245 #define vm_map_range_valid(map, start, end) \
246 vm_map_range_valid_KBI((map), (start), (end))
248 static __inline vm_offset_t
249 vm_map_max(const struct vm_map *map)
252 return (map->header.start);
255 static __inline vm_offset_t
256 vm_map_min(const struct vm_map *map)
259 return (map->header.end);
262 static __inline pmap_t
263 vm_map_pmap(vm_map_t map)
269 vm_map_modflags(vm_map_t map, vm_flags_t set, vm_flags_t clear)
271 map->flags = (map->flags | set) & ~clear;
275 vm_map_range_valid(vm_map_t map, vm_offset_t start, vm_offset_t end)
279 if (start < vm_map_min(map) || end > vm_map_max(map))
284 #endif /* KLD_MODULE */
288 * Shareable process virtual address space.
291 * (c) const until freed
294 struct vm_map vm_map; /* VM address map */
295 struct shmmap_state *vm_shm; /* SYS5 shared memory private data XXX */
296 segsz_t vm_swrss; /* resident set size before last swap */
297 segsz_t vm_tsize; /* text size (pages) XXX */
298 segsz_t vm_dsize; /* data size (pages) XXX */
299 segsz_t vm_ssize; /* stack size (pages) */
300 caddr_t vm_taddr; /* (c) user virtual address of text */
301 caddr_t vm_daddr; /* (c) user virtual address of data */
302 caddr_t vm_maxsaddr; /* user VA at max stack growth */
303 vm_offset_t vm_stacktop; /* top of the stack, may not be page-aligned */
304 vm_offset_t vm_shp_base; /* shared page address */
305 u_int vm_refcnt; /* number of references */
307 * Keep the PMAP last, so that CPU-specific variations of that
308 * structure on a single architecture don't result in offset
309 * variations of the machine-independent fields in the vmspace.
311 struct pmap vm_pmap; /* private physical map */
315 static __inline pmap_t
316 vmspace_pmap(struct vmspace *vmspace)
318 return &vmspace->vm_pmap;
324 * Macros: vm_map_lock, etc.
326 * Perform locking on the data portion of a map. Note that
327 * these macros mimic procedure calls returning void. The
328 * semicolon is supplied by the user of these macros, not
329 * by the macros themselves. The macros can safely be used
330 * as unbraced elements in a higher level statement.
333 void _vm_map_lock(vm_map_t map, const char *file, int line);
334 void _vm_map_unlock(vm_map_t map, const char *file, int line);
335 int _vm_map_unlock_and_wait(vm_map_t map, int timo, const char *file, int line);
336 void _vm_map_lock_read(vm_map_t map, const char *file, int line);
337 void _vm_map_unlock_read(vm_map_t map, const char *file, int line);
338 int _vm_map_trylock(vm_map_t map, const char *file, int line);
339 int _vm_map_trylock_read(vm_map_t map, const char *file, int line);
340 int _vm_map_lock_upgrade(vm_map_t map, const char *file, int line);
341 void _vm_map_lock_downgrade(vm_map_t map, const char *file, int line);
342 int vm_map_locked(vm_map_t map);
343 void vm_map_wakeup(vm_map_t map);
344 void vm_map_busy(vm_map_t map);
345 void vm_map_unbusy(vm_map_t map);
346 void vm_map_wait_busy(vm_map_t map);
347 vm_offset_t vm_map_max_KBI(const struct vm_map *map);
348 vm_offset_t vm_map_min_KBI(const struct vm_map *map);
349 pmap_t vm_map_pmap_KBI(vm_map_t map);
350 bool vm_map_range_valid_KBI(vm_map_t map, vm_offset_t start, vm_offset_t end);
352 #define vm_map_lock(map) _vm_map_lock(map, LOCK_FILE, LOCK_LINE)
353 #define vm_map_unlock(map) _vm_map_unlock(map, LOCK_FILE, LOCK_LINE)
354 #define vm_map_unlock_and_wait(map, timo) \
355 _vm_map_unlock_and_wait(map, timo, LOCK_FILE, LOCK_LINE)
356 #define vm_map_lock_read(map) _vm_map_lock_read(map, LOCK_FILE, LOCK_LINE)
357 #define vm_map_unlock_read(map) _vm_map_unlock_read(map, LOCK_FILE, LOCK_LINE)
358 #define vm_map_trylock(map) _vm_map_trylock(map, LOCK_FILE, LOCK_LINE)
359 #define vm_map_trylock_read(map) \
360 _vm_map_trylock_read(map, LOCK_FILE, LOCK_LINE)
361 #define vm_map_lock_upgrade(map) \
362 _vm_map_lock_upgrade(map, LOCK_FILE, LOCK_LINE)
363 #define vm_map_lock_downgrade(map) \
364 _vm_map_lock_downgrade(map, LOCK_FILE, LOCK_LINE)
366 long vmspace_resident_count(struct vmspace *vmspace);
370 * Copy-on-write flags for vm_map operations
372 #define MAP_INHERIT_SHARE 0x00000001
373 #define MAP_COPY_ON_WRITE 0x00000002
374 #define MAP_NOFAULT 0x00000004
375 #define MAP_PREFAULT 0x00000008
376 #define MAP_PREFAULT_PARTIAL 0x00000010
377 #define MAP_DISABLE_SYNCER 0x00000020
378 #define MAP_CHECK_EXCL 0x00000040
379 #define MAP_CREATE_GUARD 0x00000080
380 #define MAP_DISABLE_COREDUMP 0x00000100
381 #define MAP_PREFAULT_MADVISE 0x00000200 /* from (user) madvise request */
382 #define MAP_WRITECOUNT 0x00000400
383 #define MAP_REMAP 0x00000800
384 #define MAP_STACK_GROWS_DOWN 0x00001000
385 #define MAP_STACK_GROWS_UP 0x00002000
386 #define MAP_ACC_CHARGED 0x00004000
387 #define MAP_ACC_NO_CHARGE 0x00008000
388 #define MAP_CREATE_STACK_GAP_UP 0x00010000
389 #define MAP_CREATE_STACK_GAP_DN 0x00020000
390 #define MAP_VN_EXEC 0x00040000
391 #define MAP_SPLIT_BOUNDARY_MASK 0x00180000
392 #define MAP_NO_HINT 0x00200000
394 #define MAP_SPLIT_BOUNDARY_SHIFT 19
397 * vm_fault option flags
399 #define VM_FAULT_NORMAL 0x00 /* Nothing special */
400 #define VM_FAULT_WIRE 0x01 /* Wire the mapped page */
401 #define VM_FAULT_DIRTY 0x02 /* Dirty the page; use w/VM_PROT_COPY */
402 #define VM_FAULT_NOFILL 0x04 /* Fail if the pager doesn't have a copy */
405 * Initially, mappings are slightly sequential. The maximum window size must
406 * account for the map entry's "read_ahead" field being defined as an uint8_t.
408 #define VM_FAULT_READ_AHEAD_MIN 7
409 #define VM_FAULT_READ_AHEAD_INIT 15
410 #define VM_FAULT_READ_AHEAD_MAX min(atop(maxphys) - 1, UINT8_MAX)
413 * The following "find_space" options are supported by vm_map_find().
415 * For VMFS_ALIGNED_SPACE, the desired alignment is specified to
416 * the macro argument as log base 2 of the desired alignment.
418 #define VMFS_NO_SPACE 0 /* don't find; use the given range */
419 #define VMFS_ANY_SPACE 1 /* find a range with any alignment */
420 #define VMFS_OPTIMAL_SPACE 2 /* find a range with optimal alignment*/
421 #define VMFS_SUPER_SPACE 3 /* find a superpage-aligned range */
422 #define VMFS_ALIGNED_SPACE(x) ((x) << 8) /* find a range with fixed alignment */
425 * vm_map_wire and vm_map_unwire option flags
427 #define VM_MAP_WIRE_SYSTEM 0 /* wiring in a kernel map */
428 #define VM_MAP_WIRE_USER 1 /* wiring in a user map */
430 #define VM_MAP_WIRE_NOHOLES 0 /* region must not have holes */
431 #define VM_MAP_WIRE_HOLESOK 2 /* region may have holes */
433 #define VM_MAP_WIRE_WRITE 4 /* Validate writable. */
435 typedef int vm_map_entry_reader(void *token, vm_map_entry_t addr,
436 vm_map_entry_t dest);
440 * Find the successor of a map_entry, using a reader to dereference pointers.
441 * '*clone' is a copy of a vm_map entry. 'reader' is used to copy a map entry
442 * at some address into '*clone'. Change *clone to a copy of the next map
443 * entry, and return the address of that entry, or NULL if copying has failed.
445 * This function is made available to user-space code that needs to traverse
448 static inline vm_map_entry_t
449 vm_map_entry_read_succ(void *token, struct vm_map_entry *const clone,
450 vm_map_entry_reader reader)
452 vm_map_entry_t after, backup;
455 after = clone->right;
456 start = clone->start;
457 if (!reader(token, after, clone))
459 backup = clone->left;
460 if (!reader(token, backup, clone))
462 if (clone->start > start) {
465 backup = clone->left;
466 if (!reader(token, backup, clone))
468 } while (clone->start != start);
470 if (!reader(token, after, clone))
474 #endif /* ! _KERNEL */
477 boolean_t vm_map_check_protection (vm_map_t, vm_offset_t, vm_offset_t, vm_prot_t);
478 int vm_map_delete(vm_map_t, vm_offset_t, vm_offset_t);
479 int vm_map_find(vm_map_t, vm_object_t, vm_ooffset_t, vm_offset_t *, vm_size_t,
480 vm_offset_t, int, vm_prot_t, vm_prot_t, int);
481 int vm_map_find_min(vm_map_t, vm_object_t, vm_ooffset_t, vm_offset_t *,
482 vm_size_t, vm_offset_t, vm_offset_t, int, vm_prot_t, vm_prot_t, int);
483 int vm_map_find_aligned(vm_map_t map, vm_offset_t *addr, vm_size_t length,
484 vm_offset_t max_addr, vm_offset_t alignment);
485 int vm_map_fixed(vm_map_t, vm_object_t, vm_ooffset_t, vm_offset_t, vm_size_t,
486 vm_prot_t, vm_prot_t, int);
487 vm_offset_t vm_map_findspace(vm_map_t, vm_offset_t, vm_size_t);
488 int vm_map_inherit (vm_map_t, vm_offset_t, vm_offset_t, vm_inherit_t);
489 void vm_map_init(vm_map_t, pmap_t, vm_offset_t, vm_offset_t);
490 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);
491 int vm_map_lookup (vm_map_t *, vm_offset_t, vm_prot_t, vm_map_entry_t *, vm_object_t *,
492 vm_pindex_t *, vm_prot_t *, boolean_t *);
493 int vm_map_lookup_locked(vm_map_t *, vm_offset_t, vm_prot_t, vm_map_entry_t *, vm_object_t *,
494 vm_pindex_t *, vm_prot_t *, boolean_t *);
495 void vm_map_lookup_done (vm_map_t, vm_map_entry_t);
496 boolean_t vm_map_lookup_entry (vm_map_t, vm_offset_t, vm_map_entry_t *);
498 static inline vm_map_entry_t
499 vm_map_entry_first(vm_map_t map)
502 return (map->header.right);
505 static inline vm_map_entry_t
506 vm_map_entry_succ(vm_map_entry_t entry)
508 vm_map_entry_t after;
510 after = entry->right;
511 if (after->left->start > entry->start) {
514 while (after->left != entry);
519 #define VM_MAP_ENTRY_FOREACH(it, map) \
520 for ((it) = vm_map_entry_first(map); \
521 (it) != &(map)->header; \
522 (it) = vm_map_entry_succ(it))
524 #define VM_MAP_PROTECT_SET_PROT 0x0001
525 #define VM_MAP_PROTECT_SET_MAXPROT 0x0002
527 int vm_map_protect(vm_map_t map, vm_offset_t start, vm_offset_t end,
528 vm_prot_t new_prot, vm_prot_t new_maxprot, int flags);
529 int vm_map_remove (vm_map_t, vm_offset_t, vm_offset_t);
530 void vm_map_try_merge_entries(vm_map_t map, vm_map_entry_t prev,
531 vm_map_entry_t entry);
532 void vm_map_startup (void);
533 int vm_map_submap (vm_map_t, vm_offset_t, vm_offset_t, vm_map_t);
534 int vm_map_sync(vm_map_t, vm_offset_t, vm_offset_t, boolean_t, boolean_t);
535 int vm_map_madvise (vm_map_t, vm_offset_t, vm_offset_t, int);
536 int vm_map_stack (vm_map_t, vm_offset_t, vm_size_t, vm_prot_t, vm_prot_t, int);
537 int vm_map_unwire(vm_map_t map, vm_offset_t start, vm_offset_t end,
539 int vm_map_wire(vm_map_t map, vm_offset_t start, vm_offset_t end, int flags);
540 int vm_map_wire_locked(vm_map_t map, vm_offset_t start, vm_offset_t end,
542 long vmspace_swap_count(struct vmspace *vmspace);
543 void vm_map_entry_set_vnode_text(vm_map_entry_t entry, bool add);
545 #endif /* _VM_MAP_ */