2 * Copyright (c) 1991, 1993
3 * The Regents of the University of California. All rights reserved.
5 * This code is derived from software contributed to Berkeley by
6 * The Mach Operating System project at Carnegie-Mellon University.
8 * Redistribution and use in source and binary forms, with or without
9 * modification, are permitted provided that the following conditions
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 * 3. All advertising materials mentioning features or use of this software
17 * must display the following acknowledgement:
18 * This product includes software developed by the University of
19 * California, Berkeley and its contributors.
20 * 4. Neither the name of the University nor the names of its contributors
21 * may be used to endorse or promote products derived from this software
22 * without specific prior written permission.
24 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
25 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
26 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
27 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
28 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
29 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
30 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
31 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
32 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
33 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
36 * @(#)vm_map.h 8.9 (Berkeley) 5/17/95
39 * Copyright (c) 1987, 1990 Carnegie-Mellon University.
40 * All rights reserved.
42 * Authors: Avadis Tevanian, Jr., Michael Wayne Young
44 * Permission to use, copy, modify and distribute this software and
45 * its documentation is hereby granted, provided that both the copyright
46 * notice and this permission notice appear in all copies of the
47 * software, derivative works or modified versions, and any portions
48 * thereof, and that both notices appear in supporting documentation.
50 * CARNEGIE MELLON ALLOWS FREE USE OF THIS SOFTWARE IN ITS "AS IS"
51 * CONDITION. CARNEGIE MELLON DISCLAIMS ANY LIABILITY OF ANY KIND
52 * FOR ANY DAMAGES WHATSOEVER RESULTING FROM THE USE OF THIS SOFTWARE.
54 * Carnegie Mellon requests users of this software to return to
56 * Software Distribution Coordinator or Software.Distribution@CS.CMU.EDU
57 * School of Computer Science
58 * Carnegie Mellon University
59 * Pittsburgh PA 15213-3890
61 * any improvements or extensions that they make and grant Carnegie the
62 * rights to redistribute these changes.
64 * $FreeBSD: src/sys/vm/vm_map.h,v 1.54.2.5 2003/01/13 22:51:17 dillon Exp $
65 * $DragonFly: src/sys/vm/vm_map.h,v 1.30 2007/04/29 18:25:41 dillon Exp $
69 * Virtual memory map module definitions.
76 #include <sys/types.h>
81 #ifndef _SYS_SYSREF_H_
82 #include <sys/sysref.h>
87 #ifndef _SYS_VKERNEL_H_
88 #include <sys/vkernel.h>
93 #ifndef _MACHINE_PMAP_H_
94 #include <machine/pmap.h>
96 #ifndef _VM_VM_OBJECT_H_
97 #include <vm/vm_object.h>
100 #include <sys/_null.h>
103 struct vm_map_rb_tree;
104 RB_PROTOTYPE(vm_map_rb_tree, vm_map_entry, rb_entry, rb_vm_map_compare);
109 * vm_map_t the high-level address map data structure.
110 * vm_map_entry_t an entry in an address map.
113 typedef u_int vm_flags_t;
114 typedef u_int vm_eflags_t;
117 * Objects which live in maps may be either VM objects, or
118 * another map (called a "sharing map") which denotes read-write
119 * sharing with other maps.
121 union vm_map_object {
122 struct vm_object *vm_object; /* object object */
123 struct vm_map *sub_map; /* belongs to another map */
127 vm_offset_t avail_ssize; /* amt can grow if this is a stack */
128 vpte_t master_pde; /* virtual page table root */
132 * Address map entries consist of start and end addresses,
133 * a VM object (or sharing map) and offset into that object,
134 * and user-exported inheritance and protection information.
135 * Also included is control information for virtual copy operations.
137 * When used with MAP_STACK, avail_ssize is used to determine the
138 * limits of stack growth.
140 * When used with VM_MAPTYPE_VPAGETABLE, avail_ssize stores the
141 * page directory index.
143 struct vm_map_entry {
144 struct vm_map_entry *prev; /* previous entry */
145 struct vm_map_entry *next; /* next entry */
146 RB_ENTRY(vm_map_entry) rb_entry;
147 vm_offset_t start; /* start address */
148 vm_offset_t end; /* end address */
149 union vm_map_aux aux; /* auxillary data */
150 union vm_map_object object; /* object I point to */
151 vm_ooffset_t offset; /* offset into object */
152 vm_eflags_t eflags; /* map entry flags */
153 vm_maptype_t maptype; /* type of VM mapping */
154 vm_prot_t protection; /* protection code */
155 vm_prot_t max_protection; /* maximum protection */
156 vm_inherit_t inheritance; /* inheritance */
157 int wired_count; /* can be paged if = 0 */
160 #define MAP_ENTRY_NOSYNC 0x0001
161 #define MAP_ENTRY_STACK 0x0002
162 #define MAP_ENTRY_COW 0x0004
163 #define MAP_ENTRY_NEEDS_COPY 0x0008
164 #define MAP_ENTRY_NOFAULT 0x0010
165 #define MAP_ENTRY_USER_WIRED 0x0020
167 #define MAP_ENTRY_BEHAV_NORMAL 0x0000 /* default behavior */
168 #define MAP_ENTRY_BEHAV_SEQUENTIAL 0x0040 /* expect sequential access */
169 #define MAP_ENTRY_BEHAV_RANDOM 0x0080 /* expect random access */
170 #define MAP_ENTRY_BEHAV_RESERVED 0x00C0 /* future use */
172 #define MAP_ENTRY_BEHAV_MASK 0x00C0
174 #define MAP_ENTRY_IN_TRANSITION 0x0100 /* entry being changed */
175 #define MAP_ENTRY_NEEDS_WAKEUP 0x0200 /* waiter's in transition */
176 #define MAP_ENTRY_NOCOREDUMP 0x0400 /* don't include in a core */
177 #define MAP_ENTRY_KSTACK 0x0800 /* guarded kernel stack */
180 * flags for vm_map_[un]clip_range()
182 #define MAP_CLIP_NO_HOLES 0x0001
185 * This reserve count for vm_map_entry_reserve() should cover all nominal
186 * single-insertion operations, including any necessary clipping.
188 #define MAP_RESERVE_COUNT 4
189 #define MAP_RESERVE_SLOP 32
191 static __inline u_char
192 vm_map_entry_behavior(struct vm_map_entry *entry)
194 return entry->eflags & MAP_ENTRY_BEHAV_MASK;
198 vm_map_entry_set_behavior(struct vm_map_entry *entry, u_char behavior)
200 entry->eflags = (entry->eflags & ~MAP_ENTRY_BEHAV_MASK) |
201 (behavior & MAP_ENTRY_BEHAV_MASK);
205 * Maps are doubly-linked lists of map entries, kept sorted
206 * by address. A single hint is provided to start
207 * searches again from the last successful search,
208 * insertion, or removal.
210 * Note: the lock structure cannot be the first element of vm_map
211 * because this can result in a running lockup between two or more
212 * system processes trying to kmem_alloc_wait() due to kmem_alloc_wait()
213 * and free tsleep/waking up 'map' and the underlying lockmgr also
214 * sleeping and waking up on 'map'. The lockup occurs when the map fills
215 * up. The 'exec' map, for example.
218 struct vm_map_entry header; /* List of entries */
219 RB_HEAD(vm_map_rb_tree, vm_map_entry) rb_root;
220 struct lock lock; /* Lock for map data */
221 int nentries; /* Number of entries */
222 vm_size_t size; /* virtual size */
223 u_char system_map; /* Am I a system map? */
224 vm_map_entry_t hint; /* hint for quick lookups */
225 unsigned int timestamp; /* Version number */
226 vm_map_entry_t first_free; /* First free space hint */
227 vm_flags_t flags; /* flags for this vm_map */
228 struct pmap *pmap; /* Physical map */
229 #define min_offset header.start
230 #define max_offset header.end
236 #define MAP_WIREFUTURE 0x01 /* wire all future pages */
244 struct vmupcall *vu_next;
245 void *vu_func; /* user upcall function */
246 void *vu_data; /* user data */
247 void *vu_ctx; /* user context function */
248 struct lwp *vu_lwp; /* process that registered upcall */
249 int vu_id; /* upcall identifier */
250 int vu_pending; /* upcall request pending */
254 * Shareable process virtual address space.
256 * Refd pointers from vmresident, proc
259 struct vm_map vm_map; /* VM address map */
260 struct pmap vm_pmap; /* private physical map */
262 caddr_t vm_shm; /* SYS5 shared memory private data XXX */
263 /* we copy from vm_startcopy to the end of the structure on fork */
264 #define vm_startcopy vm_rssize
265 segsz_t vm_rssize; /* current resident set size in pages */
266 segsz_t vm_swrss; /* resident set size before last swap */
267 segsz_t vm_tsize; /* text size (pages) XXX */
268 segsz_t vm_dsize; /* data size (pages) XXX */
269 segsz_t vm_ssize; /* stack size (pages) */
270 caddr_t vm_taddr; /* user virtual address of text XXX */
271 caddr_t vm_daddr; /* user virtual address of data XXX */
272 caddr_t vm_maxsaddr; /* user VA at max stack growth */
273 caddr_t vm_minsaddr; /* user VA at max stack growth */
274 #define vm_endcopy vm_exitingcnt
275 int vm_exitingcnt; /* several procsses zombied in exit1 */
276 int vm_upccount; /* number of registered upcalls */
278 struct vmupcall *vm_upcalls; /* registered upcalls */
279 struct sysref vm_sysref; /* sysref, refcnt, etc */
283 * Resident executable holding structure. A user program can take a snapshot
284 * of just its VM address space (typically done just after dynamic link
285 * libraries have completed loading) and register it as a resident
286 * executable associated with the program binary's vnode, which is also
287 * locked into memory. Future execs of the vnode will start with a copy
288 * of the resident vmspace instead of running the binary from scratch,
289 * avoiding both the kernel ELF loader *AND* all shared library mapping and
290 * relocation code, and will call a different entry point (the stack pointer
291 * is reset to the top of the stack) supplied when the vmspace was registered.
294 struct vnode *vr_vnode; /* associated vnode */
295 TAILQ_ENTRY(vmresident) vr_link; /* linked list of res sts */
296 struct vmspace *vr_vmspace; /* vmspace to fork */
297 intptr_t vr_entry_addr; /* registered entry point */
298 struct sysentvec *vr_sysent; /* system call vects */
299 int vr_id; /* registration id */
300 int vr_refs; /* temporary refs */
305 * Macros: vm_map_lock, etc.
307 * Perform locking on the data portion of a map. Note that
308 * these macros mimic procedure calls returning void. The
309 * semicolon is supplied by the user of these macros, not
310 * by the macros themselves. The macros can safely be used
311 * as unbraced elements in a higher level statement.
314 #define ASSERT_VM_MAP_LOCKED(map) KKASSERT(lockowned(&(map)->lock))
317 /* #define MAP_LOCK_DIAGNOSTIC 1 */
318 #ifdef MAP_LOCK_DIAGNOSTIC
319 #define vm_map_lock(map) \
321 kprintf ("locking map LK_EXCLUSIVE: 0x%x\n", map); \
322 if (lockmgr(&(map)->lock, LK_EXCLUSIVE) != 0) { \
323 panic("vm_map_lock: failed to get lock"); \
325 (map)->timestamp++; \
328 #define vm_map_lock(map) \
330 if (lockmgr(&(map)->lock, LK_EXCLUSIVE) != 0) { \
331 panic("vm_map_lock: failed to get lock"); \
333 (map)->timestamp++; \
337 #define vm_map_lock(map) \
339 lockmgr(&(map)->lock, LK_EXCLUSIVE); \
340 (map)->timestamp++; \
342 #endif /* DIAGNOSTIC */
344 #if defined(MAP_LOCK_DIAGNOSTIC)
345 #define vm_map_unlock(map) \
347 kprintf ("locking map LK_RELEASE: 0x%x\n", map); \
348 lockmgr(&(map)->lock, LK_RELEASE); \
350 #define vm_map_lock_read(map) \
352 kprintf ("locking map LK_SHARED: 0x%x\n", map); \
353 lockmgr(&(map)->lock, LK_SHARED); \
355 #define vm_map_unlock_read(map) \
357 kprintf ("locking map LK_RELEASE: 0x%x\n", map); \
358 lockmgr(&(map)->lock, LK_RELEASE); \
361 #define vm_map_unlock(map) \
362 lockmgr(&(map)->lock, LK_RELEASE)
363 #define vm_map_lock_read(map) \
364 lockmgr(&(map)->lock, LK_SHARED)
365 #define vm_map_unlock_read(map) \
366 lockmgr(&(map)->lock, LK_RELEASE)
369 static __inline__ int
370 vm_map_lock_upgrade(vm_map_t map) {
372 #if defined(MAP_LOCK_DIAGNOSTIC)
373 kprintf("locking map LK_EXCLUPGRADE: 0x%x\n", map);
375 error = lockmgr(&map->lock, LK_EXCLUPGRADE);
381 #if defined(MAP_LOCK_DIAGNOSTIC)
382 #define vm_map_lock_downgrade(map) \
384 kprintf ("locking map LK_DOWNGRADE: 0x%x\n", map); \
385 lockmgr(&(map)->lock, LK_DOWNGRADE); \
388 #define vm_map_lock_downgrade(map) \
389 lockmgr(&(map)->lock, LK_DOWNGRADE)
395 * Functions implemented as macros
397 #define vm_map_min(map) ((map)->min_offset)
398 #define vm_map_max(map) ((map)->max_offset)
399 #define vm_map_pmap(map) ((map)->pmap)
404 static __inline struct pmap *
405 vmspace_pmap(struct vmspace *vmspace)
407 return &vmspace->vm_pmap;
411 vmspace_resident_count(struct vmspace *vmspace)
413 return pmap_resident_count(vmspace_pmap(vmspace));
416 /* Calculates the proportional RSS and returning the
419 static __inline u_int
420 vmspace_president_count(struct vmspace *vmspace)
422 vm_map_t map = &vmspace->vm_map;
427 for (cur = map->header.next; cur != &map->header; cur = cur->next) {
428 switch(cur->maptype) {
429 case VM_MAPTYPE_NORMAL:
430 case VM_MAPTYPE_VPAGETABLE:
431 if ((object = cur->object.vm_object) == NULL)
433 if (object->type != OBJT_DEFAULT &&
434 object->type != OBJT_SWAP) {
437 if(object->agg_pv_list_count != 0) {
438 count += (object->resident_page_count / object->agg_pv_list_count);
449 * Number of kernel maps and entries to statically allocate, required
450 * during boot to bootstrap the VM system.
453 #define MAX_MAPENT 2048 /* required to support up to 64 cpus */
456 * Copy-on-write flags for vm_map operations
458 #define MAP_UNUSED_01 0x0001
459 #define MAP_COPY_ON_WRITE 0x0002
460 #define MAP_NOFAULT 0x0004
461 #define MAP_PREFAULT 0x0008
462 #define MAP_PREFAULT_PARTIAL 0x0010
463 #define MAP_DISABLE_SYNCER 0x0020
464 #define MAP_IS_STACK 0x0040
465 #define MAP_IS_KSTACK 0x0080
466 #define MAP_DISABLE_COREDUMP 0x0100
467 #define MAP_PREFAULT_MADVISE 0x0200 /* from (user) madvise request */
470 * vm_fault option flags
472 #define VM_FAULT_NORMAL 0x00 /* Nothing special */
473 #define VM_FAULT_CHANGE_WIRING 0x01 /* Change the wiring as appropriate */
474 #define VM_FAULT_USER_WIRE 0x02 /* Likewise, but for user purposes */
475 #define VM_FAULT_BURST 0x04 /* Burst fault can be done */
476 #define VM_FAULT_DIRTY 0x08 /* Dirty the page */
477 #define VM_FAULT_UNSWAP 0x10 /* Remove backing store from the page */
478 #define VM_FAULT_WIRE_MASK (VM_FAULT_CHANGE_WIRING|VM_FAULT_USER_WIRE)
482 extern struct sysref_class vmspace_sysref_class;
484 boolean_t vm_map_check_protection (vm_map_t, vm_offset_t, vm_offset_t,
485 vm_prot_t, boolean_t);
488 void vm_map_entry_allocate_object(vm_map_entry_t);
489 void vm_map_entry_reserve_cpu_init(struct globaldata *gd);
490 int vm_map_entry_reserve(int);
491 int vm_map_entry_kreserve(int);
492 void vm_map_entry_release(int);
493 void vm_map_entry_krelease(int);
494 vm_map_t vm_map_create (vm_map_t, struct pmap *, vm_offset_t, vm_offset_t);
495 int vm_map_delete (vm_map_t, vm_offset_t, vm_offset_t, int *);
496 int vm_map_find (vm_map_t, vm_object_t, vm_ooffset_t,
497 vm_offset_t *, vm_size_t, vm_size_t,
498 boolean_t, vm_maptype_t,
499 vm_prot_t, vm_prot_t,
501 int vm_map_findspace (vm_map_t, vm_offset_t, vm_size_t, vm_size_t,
503 vm_offset_t vm_map_hint(struct proc *, vm_offset_t, vm_prot_t);
504 int vm_map_inherit (vm_map_t, vm_offset_t, vm_offset_t, vm_inherit_t);
505 void vm_map_init (struct vm_map *, vm_offset_t, vm_offset_t, pmap_t);
506 int vm_map_insert (vm_map_t, int *, vm_object_t, vm_ooffset_t,
507 vm_offset_t, vm_offset_t,
509 vm_prot_t, vm_prot_t,
511 int vm_map_lookup (vm_map_t *, vm_offset_t, vm_prot_t, vm_map_entry_t *, vm_object_t *,
512 vm_pindex_t *, vm_prot_t *, boolean_t *);
513 void vm_map_lookup_done (vm_map_t, vm_map_entry_t, int);
514 boolean_t vm_map_lookup_entry (vm_map_t, vm_offset_t, vm_map_entry_t *);
515 int vm_map_wire (vm_map_t, vm_offset_t, vm_offset_t, int);
516 int vm_map_unwire (vm_map_t, vm_offset_t, vm_offset_t, boolean_t);
517 int vm_map_clean (vm_map_t, vm_offset_t, vm_offset_t, boolean_t, boolean_t);
518 int vm_map_protect (vm_map_t, vm_offset_t, vm_offset_t, vm_prot_t, boolean_t);
519 int vm_map_remove (vm_map_t, vm_offset_t, vm_offset_t);
520 void vm_map_startup (void);
521 int vm_map_submap (vm_map_t, vm_offset_t, vm_offset_t, vm_map_t);
522 int vm_map_madvise (vm_map_t, vm_offset_t, vm_offset_t, int, off_t);
523 void vm_map_simplify_entry (vm_map_t, vm_map_entry_t, int *);
524 void vm_init2 (void);
525 int vm_uiomove (vm_map_t, vm_object_t, off_t, int, vm_offset_t, int *);
526 int vm_map_stack (vm_map_t, vm_offset_t, vm_size_t, int,
527 vm_prot_t, vm_prot_t, int);
528 int vm_map_growstack (struct proc *p, vm_offset_t addr);
529 int vmspace_swap_count (struct vmspace *vmspace);
530 int vmspace_anonymous_count (struct vmspace *vmspace);
531 void vm_map_set_wired_quick(vm_map_t map, vm_offset_t addr, vm_size_t size, int *);
532 void vm_map_transition_wait(vm_map_t map);
536 #endif /* _VM_VM_MAP_H_ */