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
152 #define MAP_ENTRY_SPLIT_BOUNDARY_MASK 0x00300000
154 #define MAP_ENTRY_SPLIT_BOUNDARY_SHIFT 20
157 static __inline u_char
158 vm_map_entry_behavior(vm_map_entry_t entry)
160 return (entry->eflags & MAP_ENTRY_BEHAV_MASK);
164 vm_map_entry_user_wired_count(vm_map_entry_t entry)
166 if (entry->eflags & MAP_ENTRY_USER_WIRED)
172 vm_map_entry_system_wired_count(vm_map_entry_t entry)
174 return (entry->wired_count - vm_map_entry_user_wired_count(entry));
179 * A map is a set of map entries. These map entries are
180 * organized as a threaded binary search tree. Both structures
181 * are ordered based upon the start and end addresses contained
182 * within each map entry. The largest gap between an entry in a
183 * subtree and one of its neighbors is saved in the max_free
184 * field, and that field is updated when the tree is
187 * Sleator and Tarjan's top-down splay algorithm is employed to
188 * control height imbalance in the binary search tree.
190 * The map's min offset value is stored in map->header.end, and
191 * its max offset value is stored in map->header.start. These
192 * values act as sentinels for any forward or backward address
193 * scan of the list. The right and left fields of the map
194 * header point to the first and list map entries. The map
195 * header has a special value for the eflags field,
196 * MAP_ENTRY_HEADER, that is set initially, is never changed,
197 * and prevents an eflags match of the header with any other map
201 * (c) const until freed
204 struct vm_map_entry header; /* List of entries */
205 struct sx lock; /* Lock for map data */
206 struct mtx system_mtx;
207 int nentries; /* Number of entries */
208 vm_size_t size; /* virtual size */
209 u_int timestamp; /* Version number */
211 u_char system_map; /* (c) Am I a system map? */
212 vm_flags_t flags; /* flags for this vm_map */
213 vm_map_entry_t root; /* Root of a binary search tree */
214 pmap_t pmap; /* (c) Physical map */
215 vm_offset_t anon_loc;
225 #define MAP_WIREFUTURE 0x01 /* wire all future pages */
226 #define MAP_BUSY_WAKEUP 0x02
227 #define MAP_IS_SUB_MAP 0x04 /* has parent */
228 #define MAP_ASLR 0x08 /* enabled ASLR */
229 #define MAP_ASLR_IGNSTART 0x10
230 #define MAP_REPLENISH 0x20
233 #if defined(KLD_MODULE) && !defined(KLD_TIED)
234 #define vm_map_max(map) vm_map_max_KBI((map))
235 #define vm_map_min(map) vm_map_min_KBI((map))
236 #define vm_map_pmap(map) vm_map_pmap_KBI((map))
237 #define vm_map_range_valid(map, start, end) \
238 vm_map_range_valid_KBI((map), (start), (end))
240 static __inline vm_offset_t
241 vm_map_max(const struct vm_map *map)
244 return (map->header.start);
247 static __inline vm_offset_t
248 vm_map_min(const struct vm_map *map)
251 return (map->header.end);
254 static __inline pmap_t
255 vm_map_pmap(vm_map_t map)
261 vm_map_modflags(vm_map_t map, vm_flags_t set, vm_flags_t clear)
263 map->flags = (map->flags | set) & ~clear;
267 vm_map_range_valid(vm_map_t map, vm_offset_t start, vm_offset_t end)
271 if (start < vm_map_min(map) || end > vm_map_max(map))
276 #endif /* KLD_MODULE */
280 * Shareable process virtual address space.
283 * (c) const until freed
286 struct vm_map vm_map; /* VM address map */
287 struct shmmap_state *vm_shm; /* SYS5 shared memory private data XXX */
288 segsz_t vm_swrss; /* resident set size before last swap */
289 segsz_t vm_tsize; /* text size (pages) XXX */
290 segsz_t vm_dsize; /* data size (pages) XXX */
291 segsz_t vm_ssize; /* stack size (pages) */
292 caddr_t vm_taddr; /* (c) user virtual address of text */
293 caddr_t vm_daddr; /* (c) user virtual address of data */
294 caddr_t vm_maxsaddr; /* user VA at max stack growth */
295 u_int vm_refcnt; /* number of references */
297 * Keep the PMAP last, so that CPU-specific variations of that
298 * structure on a single architecture don't result in offset
299 * variations of the machine-independent fields in the vmspace.
301 struct pmap vm_pmap; /* private physical map */
305 static __inline pmap_t
306 vmspace_pmap(struct vmspace *vmspace)
308 return &vmspace->vm_pmap;
314 * Macros: vm_map_lock, etc.
316 * Perform locking on the data portion of a map. Note that
317 * these macros mimic procedure calls returning void. The
318 * semicolon is supplied by the user of these macros, not
319 * by the macros themselves. The macros can safely be used
320 * as unbraced elements in a higher level statement.
323 void _vm_map_lock(vm_map_t map, const char *file, int line);
324 void _vm_map_unlock(vm_map_t map, const char *file, int line);
325 int _vm_map_unlock_and_wait(vm_map_t map, int timo, const char *file, int line);
326 void _vm_map_lock_read(vm_map_t map, const char *file, int line);
327 void _vm_map_unlock_read(vm_map_t map, const char *file, int line);
328 int _vm_map_trylock(vm_map_t map, const char *file, int line);
329 int _vm_map_trylock_read(vm_map_t map, const char *file, int line);
330 int _vm_map_lock_upgrade(vm_map_t map, const char *file, int line);
331 void _vm_map_lock_downgrade(vm_map_t map, const char *file, int line);
332 int vm_map_locked(vm_map_t map);
333 void vm_map_wakeup(vm_map_t map);
334 void vm_map_busy(vm_map_t map);
335 void vm_map_unbusy(vm_map_t map);
336 void vm_map_wait_busy(vm_map_t map);
337 vm_offset_t vm_map_max_KBI(const struct vm_map *map);
338 vm_offset_t vm_map_min_KBI(const struct vm_map *map);
339 pmap_t vm_map_pmap_KBI(vm_map_t map);
340 bool vm_map_range_valid_KBI(vm_map_t map, vm_offset_t start, vm_offset_t end);
342 #define vm_map_lock(map) _vm_map_lock(map, LOCK_FILE, LOCK_LINE)
343 #define vm_map_unlock(map) _vm_map_unlock(map, LOCK_FILE, LOCK_LINE)
344 #define vm_map_unlock_and_wait(map, timo) \
345 _vm_map_unlock_and_wait(map, timo, LOCK_FILE, LOCK_LINE)
346 #define vm_map_lock_read(map) _vm_map_lock_read(map, LOCK_FILE, LOCK_LINE)
347 #define vm_map_unlock_read(map) _vm_map_unlock_read(map, LOCK_FILE, LOCK_LINE)
348 #define vm_map_trylock(map) _vm_map_trylock(map, LOCK_FILE, LOCK_LINE)
349 #define vm_map_trylock_read(map) \
350 _vm_map_trylock_read(map, LOCK_FILE, LOCK_LINE)
351 #define vm_map_lock_upgrade(map) \
352 _vm_map_lock_upgrade(map, LOCK_FILE, LOCK_LINE)
353 #define vm_map_lock_downgrade(map) \
354 _vm_map_lock_downgrade(map, LOCK_FILE, LOCK_LINE)
356 long vmspace_resident_count(struct vmspace *vmspace);
360 * Copy-on-write flags for vm_map operations
362 #define MAP_INHERIT_SHARE 0x00000001
363 #define MAP_COPY_ON_WRITE 0x00000002
364 #define MAP_NOFAULT 0x00000004
365 #define MAP_PREFAULT 0x00000008
366 #define MAP_PREFAULT_PARTIAL 0x00000010
367 #define MAP_DISABLE_SYNCER 0x00000020
368 #define MAP_CHECK_EXCL 0x00000040
369 #define MAP_CREATE_GUARD 0x00000080
370 #define MAP_DISABLE_COREDUMP 0x00000100
371 #define MAP_PREFAULT_MADVISE 0x00000200 /* from (user) madvise request */
372 #define MAP_WRITECOUNT 0x00000400
373 #define MAP_REMAP 0x00000800
374 #define MAP_STACK_GROWS_DOWN 0x00001000
375 #define MAP_STACK_GROWS_UP 0x00002000
376 #define MAP_ACC_CHARGED 0x00004000
377 #define MAP_ACC_NO_CHARGE 0x00008000
378 #define MAP_CREATE_STACK_GAP_UP 0x00010000
379 #define MAP_CREATE_STACK_GAP_DN 0x00020000
380 #define MAP_VN_EXEC 0x00040000
381 #define MAP_SPLIT_BOUNDARY_MASK 0x00180000
383 #define MAP_SPLIT_BOUNDARY_SHIFT 19
386 * vm_fault option flags
388 #define VM_FAULT_NORMAL 0x00 /* Nothing special */
389 #define VM_FAULT_WIRE 0x01 /* Wire the mapped page */
390 #define VM_FAULT_DIRTY 0x02 /* Dirty the page; use w/VM_PROT_COPY */
391 #define VM_FAULT_NOFILL 0x04 /* Fail if the pager doesn't have a copy */
394 * Initially, mappings are slightly sequential. The maximum window size must
395 * account for the map entry's "read_ahead" field being defined as an uint8_t.
397 #define VM_FAULT_READ_AHEAD_MIN 7
398 #define VM_FAULT_READ_AHEAD_INIT 15
399 #define VM_FAULT_READ_AHEAD_MAX min(atop(MAXPHYS) - 1, UINT8_MAX)
402 * The following "find_space" options are supported by vm_map_find().
404 * For VMFS_ALIGNED_SPACE, the desired alignment is specified to
405 * the macro argument as log base 2 of the desired alignment.
407 #define VMFS_NO_SPACE 0 /* don't find; use the given range */
408 #define VMFS_ANY_SPACE 1 /* find a range with any alignment */
409 #define VMFS_OPTIMAL_SPACE 2 /* find a range with optimal alignment*/
410 #define VMFS_SUPER_SPACE 3 /* find a superpage-aligned range */
411 #define VMFS_ALIGNED_SPACE(x) ((x) << 8) /* find a range with fixed alignment */
414 * vm_map_wire and vm_map_unwire option flags
416 #define VM_MAP_WIRE_SYSTEM 0 /* wiring in a kernel map */
417 #define VM_MAP_WIRE_USER 1 /* wiring in a user map */
419 #define VM_MAP_WIRE_NOHOLES 0 /* region must not have holes */
420 #define VM_MAP_WIRE_HOLESOK 2 /* region may have holes */
422 #define VM_MAP_WIRE_WRITE 4 /* Validate writable. */
424 typedef int vm_map_entry_reader(void *token, vm_map_entry_t addr,
425 vm_map_entry_t dest);
429 * Find the successor of a map_entry, using a reader to dereference pointers.
430 * '*clone' is a copy of a vm_map entry. 'reader' is used to copy a map entry
431 * at some address into '*clone'. Change *clone to a copy of the next map
432 * entry, and return the address of that entry, or NULL if copying has failed.
434 * This function is made available to user-space code that needs to traverse
437 static inline vm_map_entry_t
438 vm_map_entry_read_succ(void *token, struct vm_map_entry *const clone,
439 vm_map_entry_reader reader)
441 vm_map_entry_t after, backup;
444 after = clone->right;
445 start = clone->start;
446 if (!reader(token, after, clone))
448 backup = clone->left;
449 if (!reader(token, backup, clone))
451 if (clone->start > start) {
454 backup = clone->left;
455 if (!reader(token, backup, clone))
457 } while (clone->start != start);
459 if (!reader(token, after, clone))
463 #endif /* ! _KERNEL */
466 boolean_t vm_map_check_protection (vm_map_t, vm_offset_t, vm_offset_t, vm_prot_t);
467 int vm_map_delete(vm_map_t, vm_offset_t, vm_offset_t);
468 int vm_map_find(vm_map_t, vm_object_t, vm_ooffset_t, vm_offset_t *, vm_size_t,
469 vm_offset_t, int, vm_prot_t, vm_prot_t, int);
470 int vm_map_find_min(vm_map_t, vm_object_t, vm_ooffset_t, vm_offset_t *,
471 vm_size_t, vm_offset_t, vm_offset_t, int, vm_prot_t, vm_prot_t, int);
472 int vm_map_find_aligned(vm_map_t map, vm_offset_t *addr, vm_size_t length,
473 vm_offset_t max_addr, vm_offset_t alignment);
474 int vm_map_fixed(vm_map_t, vm_object_t, vm_ooffset_t, vm_offset_t, vm_size_t,
475 vm_prot_t, vm_prot_t, int);
476 vm_offset_t vm_map_findspace(vm_map_t, vm_offset_t, vm_size_t);
477 int vm_map_inherit (vm_map_t, vm_offset_t, vm_offset_t, vm_inherit_t);
478 void vm_map_init(vm_map_t, pmap_t, vm_offset_t, vm_offset_t);
479 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);
480 int vm_map_lookup (vm_map_t *, vm_offset_t, vm_prot_t, vm_map_entry_t *, vm_object_t *,
481 vm_pindex_t *, vm_prot_t *, boolean_t *);
482 int vm_map_lookup_locked(vm_map_t *, vm_offset_t, vm_prot_t, vm_map_entry_t *, vm_object_t *,
483 vm_pindex_t *, vm_prot_t *, boolean_t *);
484 void vm_map_lookup_done (vm_map_t, vm_map_entry_t);
485 boolean_t vm_map_lookup_entry (vm_map_t, vm_offset_t, vm_map_entry_t *);
487 static inline vm_map_entry_t
488 vm_map_entry_first(vm_map_t map)
491 return (map->header.right);
494 static inline vm_map_entry_t
495 vm_map_entry_succ(vm_map_entry_t entry)
497 vm_map_entry_t after;
499 after = entry->right;
500 if (after->left->start > entry->start) {
503 while (after->left != entry);
508 #define VM_MAP_ENTRY_FOREACH(it, map) \
509 for ((it) = vm_map_entry_first(map); \
510 (it) != &(map)->header; \
511 (it) = vm_map_entry_succ(it))
512 int vm_map_protect (vm_map_t, vm_offset_t, vm_offset_t, vm_prot_t, boolean_t);
513 int vm_map_remove (vm_map_t, vm_offset_t, vm_offset_t);
514 void vm_map_try_merge_entries(vm_map_t map, vm_map_entry_t prev,
515 vm_map_entry_t entry);
516 void vm_map_startup (void);
517 int vm_map_submap (vm_map_t, vm_offset_t, vm_offset_t, vm_map_t);
518 int vm_map_sync(vm_map_t, vm_offset_t, vm_offset_t, boolean_t, boolean_t);
519 int vm_map_madvise (vm_map_t, vm_offset_t, vm_offset_t, int);
520 int vm_map_stack (vm_map_t, vm_offset_t, vm_size_t, vm_prot_t, vm_prot_t, int);
521 int vm_map_unwire(vm_map_t map, vm_offset_t start, vm_offset_t end,
523 int vm_map_wire(vm_map_t map, vm_offset_t start, vm_offset_t end, int flags);
524 int vm_map_wire_locked(vm_map_t map, vm_offset_t start, vm_offset_t end,
526 long vmspace_swap_count(struct vmspace *vmspace);
527 void vm_map_entry_set_vnode_text(vm_map_entry_t entry, bool add);
529 #endif /* _VM_MAP_ */