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>
79 #ifndef _SYS_KERNEL_H_
80 #include <sys/kernel.h> /* ticks */
86 #ifndef _SYS_SYSREF_H_
87 #include <sys/sysref.h>
92 #ifndef _SYS_VKERNEL_H_
93 #include <sys/vkernel.h>
98 #ifndef _MACHINE_PMAP_H_
99 #include <machine/pmap.h>
101 #ifndef _VM_VM_OBJECT_H_
102 #include <vm/vm_object.h>
105 #include <sys/_null.h>
108 struct vm_map_rb_tree
;
109 RB_PROTOTYPE(vm_map_rb_tree
, vm_map_entry
, rb_entry
, rb_vm_map_compare
);
114 * vm_map_t the high-level address map data structure.
115 * vm_map_entry_t an entry in an address map.
118 typedef u_int vm_flags_t
;
119 typedef u_int vm_eflags_t
;
122 * Objects which live in maps may be either VM objects, or
123 * another map (called a "sharing map") which denotes read-write
124 * sharing with other maps.
126 union vm_map_object
{
127 struct vm_object
*vm_object
; /* object object */
128 struct vm_map
*sub_map
; /* belongs to another map */
132 vm_offset_t avail_ssize
; /* amt can grow if this is a stack */
133 vpte_t master_pde
; /* virtual page table root */
137 * Address map entries consist of start and end addresses,
138 * a VM object (or sharing map) and offset into that object,
139 * and user-exported inheritance and protection information.
140 * Also included is control information for virtual copy operations.
142 * When used with MAP_STACK, avail_ssize is used to determine the
143 * limits of stack growth.
145 * When used with VM_MAPTYPE_VPAGETABLE, avail_ssize stores the
146 * page directory index.
148 struct vm_map_entry
{
149 struct vm_map_entry
*prev
; /* previous entry */
150 struct vm_map_entry
*next
; /* next entry */
151 RB_ENTRY(vm_map_entry
) rb_entry
;
152 vm_offset_t start
; /* start address */
153 vm_offset_t end
; /* end address */
154 union vm_map_aux aux
; /* auxillary data */
155 union vm_map_object object
; /* object I point to */
156 vm_ooffset_t offset
; /* offset into object */
157 vm_eflags_t eflags
; /* map entry flags */
158 vm_maptype_t maptype
; /* type of VM mapping */
159 vm_prot_t protection
; /* protection code */
160 vm_prot_t max_protection
; /* maximum protection */
161 vm_inherit_t inheritance
; /* inheritance */
162 int wired_count
; /* can be paged if = 0 */
165 #define MAP_ENTRY_NOSYNC 0x0001
166 #define MAP_ENTRY_STACK 0x0002
167 #define MAP_ENTRY_COW 0x0004
168 #define MAP_ENTRY_NEEDS_COPY 0x0008
169 #define MAP_ENTRY_NOFAULT 0x0010
170 #define MAP_ENTRY_USER_WIRED 0x0020
172 #define MAP_ENTRY_BEHAV_NORMAL 0x0000 /* default behavior */
173 #define MAP_ENTRY_BEHAV_SEQUENTIAL 0x0040 /* expect sequential access */
174 #define MAP_ENTRY_BEHAV_RANDOM 0x0080 /* expect random access */
175 #define MAP_ENTRY_BEHAV_RESERVED 0x00C0 /* future use */
177 #define MAP_ENTRY_BEHAV_MASK 0x00C0
179 #define MAP_ENTRY_IN_TRANSITION 0x0100 /* entry being changed */
180 #define MAP_ENTRY_NEEDS_WAKEUP 0x0200 /* waiter's in transition */
181 #define MAP_ENTRY_NOCOREDUMP 0x0400 /* don't include in a core */
182 #define MAP_ENTRY_KSTACK 0x0800 /* guarded kernel stack */
185 * flags for vm_map_[un]clip_range()
187 #define MAP_CLIP_NO_HOLES 0x0001
190 * This reserve count for vm_map_entry_reserve() should cover all nominal
191 * single-insertion operations, including any necessary clipping.
193 #define MAP_RESERVE_COUNT 4
194 #define MAP_RESERVE_SLOP 32
196 static __inline u_char
197 vm_map_entry_behavior(struct vm_map_entry
*entry
)
199 return entry
->eflags
& MAP_ENTRY_BEHAV_MASK
;
203 vm_map_entry_set_behavior(struct vm_map_entry
*entry
, u_char behavior
)
205 entry
->eflags
= (entry
->eflags
& ~MAP_ENTRY_BEHAV_MASK
) |
206 (behavior
& MAP_ENTRY_BEHAV_MASK
);
210 * Maps are doubly-linked lists of map entries, kept sorted by address.
211 * A single hint is provided to start searches again from the last
212 * successful search, insertion, or removal.
214 * NOTE: The lock structure cannot be the first element of vm_map
215 * because this can result in a running lockup between two or more
216 * system processes trying to kmem_alloc_wait() due to kmem_alloc_wait()
217 * and free tsleep/waking up 'map' and the underlying lockmgr also
218 * sleeping and waking up on 'map'. The lockup occurs when the map fills
219 * up. The 'exec' map, for example.
221 * NOTE: The vm_map structure can be hard-locked with the lockmgr lock
222 * or soft-serialized with the token, or both.
225 struct vm_map_entry header
; /* List of entries */
226 RB_HEAD(vm_map_rb_tree
, vm_map_entry
) rb_root
;
227 struct lock lock
; /* Lock for map data */
228 int nentries
; /* Number of entries */
229 vm_size_t size
; /* virtual size */
230 u_char system_map
; /* Am I a system map? */
231 vm_map_entry_t hint
; /* hint for quick lookups */
232 unsigned int timestamp
; /* Version number */
233 vm_map_entry_t first_free
; /* First free space hint */
234 vm_flags_t flags
; /* flags for this vm_map */
235 struct pmap
*pmap
; /* Physical map */
236 u_int president_cache
; /* Remember president count */
237 u_int president_ticks
; /* Save ticks for cache */
238 struct lwkt_token token
; /* Soft serializer */
239 #define min_offset header.start
240 #define max_offset header.end
246 #define MAP_WIREFUTURE 0x01 /* wire all future pages */
254 struct vmupcall
*vu_next
;
255 void *vu_func
; /* user upcall function */
256 void *vu_data
; /* user data */
257 void *vu_ctx
; /* user context function */
258 struct lwp
*vu_lwp
; /* process that registered upcall */
259 int vu_id
; /* upcall identifier */
260 int vu_pending
; /* upcall request pending */
264 * Shareable process virtual address space.
266 * Refd pointers from vmresident, proc
269 struct vm_map vm_map
; /* VM address map */
270 struct pmap vm_pmap
; /* private physical map */
272 caddr_t vm_shm
; /* SYS5 shared memory private data XXX */
273 /* we copy from vm_startcopy to the end of the structure on fork */
274 #define vm_startcopy vm_rssize
275 segsz_t vm_rssize
; /* current resident set size in pages */
276 segsz_t vm_swrss
; /* resident set size before last swap */
277 segsz_t vm_tsize
; /* text size (pages) XXX */
278 segsz_t vm_dsize
; /* data size (pages) XXX */
279 segsz_t vm_ssize
; /* stack size (pages) */
280 caddr_t vm_taddr
; /* user virtual address of text XXX */
281 caddr_t vm_daddr
; /* user virtual address of data XXX */
282 caddr_t vm_maxsaddr
; /* user VA at max stack growth */
283 caddr_t vm_minsaddr
; /* user VA at max stack growth */
284 #define vm_endcopy vm_exitingcnt
285 int vm_exitingcnt
; /* several procsses zombied in exit1 */
286 int vm_upccount
; /* number of registered upcalls */
288 struct vmupcall
*vm_upcalls
; /* registered upcalls */
289 struct sysref vm_sysref
; /* sysref, refcnt, etc */
293 * Resident executable holding structure. A user program can take a snapshot
294 * of just its VM address space (typically done just after dynamic link
295 * libraries have completed loading) and register it as a resident
296 * executable associated with the program binary's vnode, which is also
297 * locked into memory. Future execs of the vnode will start with a copy
298 * of the resident vmspace instead of running the binary from scratch,
299 * avoiding both the kernel ELF loader *AND* all shared library mapping and
300 * relocation code, and will call a different entry point (the stack pointer
301 * is reset to the top of the stack) supplied when the vmspace was registered.
304 struct vnode
*vr_vnode
; /* associated vnode */
305 TAILQ_ENTRY(vmresident
) vr_link
; /* linked list of res sts */
306 struct vmspace
*vr_vmspace
; /* vmspace to fork */
307 intptr_t vr_entry_addr
; /* registered entry point */
308 struct sysentvec
*vr_sysent
; /* system call vects */
309 int vr_id
; /* registration id */
310 int vr_refs
; /* temporary refs */
315 * Macros: vm_map_lock, etc.
317 * Perform locking on the data portion of a map. Note that
318 * these macros mimic procedure calls returning void. The
319 * semicolon is supplied by the user of these macros, not
320 * by the macros themselves. The macros can safely be used
321 * as unbraced elements in a higher level statement.
324 #define ASSERT_VM_MAP_LOCKED(map) KKASSERT(lockowned(&(map)->lock))
327 /* #define MAP_LOCK_DIAGNOSTIC 1 */
328 #ifdef MAP_LOCK_DIAGNOSTIC
329 #define vm_map_lock(map) \
331 kprintf ("locking map LK_EXCLUSIVE: 0x%x\n", map); \
332 if (lockmgr(&(map)->lock, LK_EXCLUSIVE) != 0) { \
333 panic("vm_map_lock: failed to get lock"); \
335 (map)->timestamp++; \
338 #define vm_map_lock(map) \
340 if (lockmgr(&(map)->lock, LK_EXCLUSIVE) != 0) { \
341 panic("vm_map_lock: failed to get lock"); \
343 (map)->timestamp++; \
347 #define vm_map_lock(map) \
349 lockmgr(&(map)->lock, LK_EXCLUSIVE); \
350 (map)->timestamp++; \
352 #endif /* DIAGNOSTIC */
354 #if defined(MAP_LOCK_DIAGNOSTIC)
355 #define vm_map_unlock(map) \
357 kprintf ("locking map LK_RELEASE: 0x%x\n", map); \
358 lockmgr(&(map)->lock, LK_RELEASE); \
360 #define vm_map_lock_read(map) \
362 kprintf ("locking map LK_SHARED: 0x%x\n", map); \
363 lockmgr(&(map)->lock, LK_SHARED); \
365 #define vm_map_unlock_read(map) \
367 kprintf ("locking map LK_RELEASE: 0x%x\n", map); \
368 lockmgr(&(map)->lock, LK_RELEASE); \
371 #define vm_map_unlock(map) \
372 lockmgr(&(map)->lock, LK_RELEASE)
373 #define vm_map_lock_read(map) \
374 lockmgr(&(map)->lock, LK_SHARED)
375 #define vm_map_unlock_read(map) \
376 lockmgr(&(map)->lock, LK_RELEASE)
379 #define vm_map_lock_read_try(map) \
380 lockmgr(&(map)->lock, LK_SHARED | LK_NOWAIT)
382 static __inline__
int
383 vm_map_lock_read_to(vm_map_t map
)
387 #if defined(MAP_LOCK_DIAGNOSTIC)
388 kprintf ("locking map LK_SHARED: 0x%x\n", map
);
390 error
= lockmgr(&(map
)->lock
, LK_SHARED
| LK_TIMELOCK
);
394 static __inline__
int
395 vm_map_lock_upgrade(vm_map_t map
) {
397 #if defined(MAP_LOCK_DIAGNOSTIC)
398 kprintf("locking map LK_EXCLUPGRADE: 0x%x\n", map
);
400 error
= lockmgr(&map
->lock
, LK_EXCLUPGRADE
);
406 #if defined(MAP_LOCK_DIAGNOSTIC)
407 #define vm_map_lock_downgrade(map) \
409 kprintf ("locking map LK_DOWNGRADE: 0x%x\n", map); \
410 lockmgr(&(map)->lock, LK_DOWNGRADE); \
413 #define vm_map_lock_downgrade(map) \
414 lockmgr(&(map)->lock, LK_DOWNGRADE)
420 * Functions implemented as macros
422 #define vm_map_min(map) ((map)->min_offset)
423 #define vm_map_max(map) ((map)->max_offset)
424 #define vm_map_pmap(map) ((map)->pmap)
429 static __inline
struct pmap
*
430 vmspace_pmap(struct vmspace
*vmspace
)
432 return &vmspace
->vm_pmap
;
436 * Caller must hold the vmspace->vm_map.token
439 vmspace_resident_count(struct vmspace
*vmspace
)
441 return pmap_resident_count(vmspace_pmap(vmspace
));
445 * Calculates the proportional RSS and returning the
446 * accrued result. This is a loose value for statistics/display
447 * purposes only and will only be updated if we can acquire
448 * a non-blocking map lock.
450 * (used by userland or the kernel)
452 static __inline u_int
453 vmspace_president_count(struct vmspace
*vmspace
)
455 vm_map_t map
= &vmspace
->vm_map
;
462 if (map
->president_ticks
== ticks
/ hz
|| vm_map_lock_read_try(map
))
463 return(map
->president_cache
);
466 for (cur
= map
->header
.next
; cur
!= &map
->header
; cur
= cur
->next
) {
467 switch(cur
->maptype
) {
468 case VM_MAPTYPE_NORMAL
:
469 case VM_MAPTYPE_VPAGETABLE
:
470 if ((object
= cur
->object
.vm_object
) == NULL
)
472 if (object
->type
!= OBJT_DEFAULT
&&
473 object
->type
!= OBJT_SWAP
) {
477 * synchronize non-zero case, contents of field
478 * can change at any time due to pmap ops.
480 if ((n
= object
->agg_pv_list_count
) != 0) {
484 count
+= object
->resident_page_count
/ n
;
492 map
->president_cache
= count
;
493 map
->president_ticks
= ticks
/ hz
;
494 vm_map_unlock_read(map
);
501 * Number of kernel maps and entries to statically allocate, required
502 * during boot to bootstrap the VM system.
505 #define MAX_MAPENT 2048 /* required to support up to 64 cpus */
508 * Copy-on-write flags for vm_map operations
510 #define MAP_UNUSED_01 0x0001
511 #define MAP_COPY_ON_WRITE 0x0002
512 #define MAP_NOFAULT 0x0004
513 #define MAP_PREFAULT 0x0008
514 #define MAP_PREFAULT_PARTIAL 0x0010
515 #define MAP_DISABLE_SYNCER 0x0020
516 #define MAP_IS_STACK 0x0040
517 #define MAP_IS_KSTACK 0x0080
518 #define MAP_DISABLE_COREDUMP 0x0100
519 #define MAP_PREFAULT_MADVISE 0x0200 /* from (user) madvise request */
522 * vm_fault option flags
524 #define VM_FAULT_NORMAL 0x00 /* Nothing special */
525 #define VM_FAULT_CHANGE_WIRING 0x01 /* Change the wiring as appropriate */
526 #define VM_FAULT_USER_WIRE 0x02 /* Likewise, but for user purposes */
527 #define VM_FAULT_BURST 0x04 /* Burst fault can be done */
528 #define VM_FAULT_DIRTY 0x08 /* Dirty the page */
529 #define VM_FAULT_UNSWAP 0x10 /* Remove backing store from the page */
530 #define VM_FAULT_WIRE_MASK (VM_FAULT_CHANGE_WIRING|VM_FAULT_USER_WIRE)
534 extern struct sysref_class vmspace_sysref_class
;
536 boolean_t
vm_map_check_protection (vm_map_t
, vm_offset_t
, vm_offset_t
,
537 vm_prot_t
, boolean_t
);
540 void vm_map_entry_allocate_object(vm_map_entry_t
);
541 void vm_map_entry_reserve_cpu_init(struct globaldata
*gd
);
542 int vm_map_entry_reserve(int);
543 int vm_map_entry_kreserve(int);
544 void vm_map_entry_release(int);
545 void vm_map_entry_krelease(int);
546 vm_map_t
vm_map_create (vm_map_t
, struct pmap
*, vm_offset_t
, vm_offset_t
);
547 int vm_map_delete (vm_map_t
, vm_offset_t
, vm_offset_t
, int *);
548 int vm_map_find (vm_map_t
, vm_object_t
, vm_ooffset_t
,
549 vm_offset_t
*, vm_size_t
, vm_size_t
,
550 boolean_t
, vm_maptype_t
,
551 vm_prot_t
, vm_prot_t
,
553 int vm_map_findspace (vm_map_t
, vm_offset_t
, vm_size_t
, vm_size_t
,
555 vm_offset_t
vm_map_hint(struct proc
*, vm_offset_t
, vm_prot_t
);
556 int vm_map_inherit (vm_map_t
, vm_offset_t
, vm_offset_t
, vm_inherit_t
);
557 void vm_map_init (struct vm_map
*, vm_offset_t
, vm_offset_t
, pmap_t
);
558 int vm_map_insert (vm_map_t
, int *, vm_object_t
, vm_ooffset_t
,
559 vm_offset_t
, vm_offset_t
,
561 vm_prot_t
, vm_prot_t
,
563 int vm_map_lookup (vm_map_t
*, vm_offset_t
, vm_prot_t
, vm_map_entry_t
*, vm_object_t
*,
564 vm_pindex_t
*, vm_prot_t
*, boolean_t
*);
565 void vm_map_lookup_done (vm_map_t
, vm_map_entry_t
, int);
566 boolean_t
vm_map_lookup_entry (vm_map_t
, vm_offset_t
, vm_map_entry_t
*);
567 int vm_map_wire (vm_map_t
, vm_offset_t
, vm_offset_t
, int);
568 int vm_map_unwire (vm_map_t
, vm_offset_t
, vm_offset_t
, boolean_t
);
569 int vm_map_clean (vm_map_t
, vm_offset_t
, vm_offset_t
, boolean_t
, boolean_t
);
570 int vm_map_protect (vm_map_t
, vm_offset_t
, vm_offset_t
, vm_prot_t
, boolean_t
);
571 int vm_map_remove (vm_map_t
, vm_offset_t
, vm_offset_t
);
572 void vm_map_startup (void);
573 int vm_map_submap (vm_map_t
, vm_offset_t
, vm_offset_t
, vm_map_t
);
574 int vm_map_madvise (vm_map_t
, vm_offset_t
, vm_offset_t
, int, off_t
);
575 void vm_map_simplify_entry (vm_map_t
, vm_map_entry_t
, int *);
576 void vm_init2 (void);
577 int vm_uiomove (vm_map_t
, vm_object_t
, off_t
, int, vm_offset_t
, int *);
578 int vm_map_stack (vm_map_t
, vm_offset_t
, vm_size_t
, int,
579 vm_prot_t
, vm_prot_t
, int);
580 int vm_map_growstack (struct proc
*p
, vm_offset_t addr
);
581 int vmspace_swap_count (struct vmspace
*vmspace
);
582 int vmspace_anonymous_count (struct vmspace
*vmspace
);
583 void vm_map_set_wired_quick(vm_map_t map
, vm_offset_t addr
, vm_size_t size
, int *);
584 void vm_map_transition_wait(vm_map_t map
);
586 #if defined(__amd64__) && defined(_KERNEL_VIRTUAL)
587 int vkernel_module_memory_alloc(vm_offset_t
*, size_t);
588 void vkernel_module_memory_free(vm_offset_t
, size_t);
592 #endif /* _VM_VM_MAP_H_ */