kernel - simplify vm pager ops, add pre-faulting for zero-fill pages.
[dragonfly.git] / sys / vm / vm_map.h
CommitLineData
984263bc
MD
1/*
2 * Copyright (c) 1991, 1993
3 * The Regents of the University of California. All rights reserved.
4 *
5 * This code is derived from software contributed to Berkeley by
6 * The Mach Operating System project at Carnegie-Mellon University.
7 *
8 * Redistribution and use in source and binary forms, with or without
9 * modification, are permitted provided that the following conditions
10 * are met:
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.
23 *
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
34 * SUCH DAMAGE.
35 *
36 * @(#)vm_map.h 8.9 (Berkeley) 5/17/95
37 *
38 *
39 * Copyright (c) 1987, 1990 Carnegie-Mellon University.
40 * All rights reserved.
41 *
42 * Authors: Avadis Tevanian, Jr., Michael Wayne Young
43 *
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.
49 *
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.
53 *
54 * Carnegie Mellon requests users of this software to return to
55 *
56 * Software Distribution Coordinator or Software.Distribution@CS.CMU.EDU
57 * School of Computer Science
58 * Carnegie Mellon University
59 * Pittsburgh PA 15213-3890
60 *
61 * any improvements or extensions that they make and grant Carnegie the
62 * rights to redistribute these changes.
63 *
64 * $FreeBSD: src/sys/vm/vm_map.h,v 1.54.2.5 2003/01/13 22:51:17 dillon Exp $
e3161323 65 * $DragonFly: src/sys/vm/vm_map.h,v 1.30 2007/04/29 18:25:41 dillon Exp $
984263bc
MD
66 */
67
68/*
69 * Virtual memory map module definitions.
70 */
71
1bd40720
MD
72#ifndef _VM_VM_MAP_H_
73#define _VM_VM_MAP_H_
984263bc 74
1bd40720
MD
75#ifndef _SYS_TYPES_H_
76#include <sys/types.h>
77#endif
78#ifndef _SYS_TREE_H_
686dbf64 79#include <sys/tree.h>
1bd40720 80#endif
e3161323
MD
81#ifndef _SYS_SYSREF_H_
82#include <sys/sysref.h>
83#endif
1bd40720
MD
84#ifndef _SYS_LOCK_H_
85#include <sys/lock.h>
86#endif
afeabdca
MD
87#ifndef _SYS_VKERNEL_H_
88#include <sys/vkernel.h>
89#endif
1bd40720
MD
90#ifndef _VM_VM_H_
91#include <vm/vm.h>
92#endif
93#ifndef _MACHINE_PMAP_H_
94#include <machine/pmap.h>
95#endif
96
686dbf64
MD
97struct vm_map_rb_tree;
98RB_PROTOTYPE(vm_map_rb_tree, vm_map_entry, rb_entry, rb_vm_map_compare);
99
984263bc
MD
100/*
101 * Types defined:
102 *
103 * vm_map_t the high-level address map data structure.
104 * vm_map_entry_t an entry in an address map.
105 */
106
107typedef u_int vm_eflags_t;
108
109/*
110 * Objects which live in maps may be either VM objects, or
111 * another map (called a "sharing map") which denotes read-write
112 * sharing with other maps.
113 */
984263bc
MD
114union vm_map_object {
115 struct vm_object *vm_object; /* object object */
116 struct vm_map *sub_map; /* belongs to another map */
117};
118
afeabdca
MD
119union vm_map_aux {
120 vm_offset_t avail_ssize; /* amt can grow if this is a stack */
121 vpte_t master_pde; /* virtual page table root */
122};
123
984263bc
MD
124/*
125 * Address map entries consist of start and end addresses,
126 * a VM object (or sharing map) and offset into that object,
127 * and user-exported inheritance and protection information.
128 * Also included is control information for virtual copy operations.
568e6804
MD
129 *
130 * When used with MAP_STACK, avail_ssize is used to determine the
131 * limits of stack growth.
132 *
133 * When used with VM_MAPTYPE_VPAGETABLE, avail_ssize stores the
134 * page directory index.
984263bc
MD
135 */
136struct vm_map_entry {
137 struct vm_map_entry *prev; /* previous entry */
138 struct vm_map_entry *next; /* next entry */
686dbf64 139 RB_ENTRY(vm_map_entry) rb_entry;
984263bc
MD
140 vm_offset_t start; /* start address */
141 vm_offset_t end; /* end address */
afeabdca 142 union vm_map_aux aux; /* auxillary data */
984263bc
MD
143 union vm_map_object object; /* object I point to */
144 vm_ooffset_t offset; /* offset into object */
145 vm_eflags_t eflags; /* map entry flags */
1b874851 146 vm_maptype_t maptype; /* type of VM mapping */
984263bc
MD
147 vm_prot_t protection; /* protection code */
148 vm_prot_t max_protection; /* maximum protection */
149 vm_inherit_t inheritance; /* inheritance */
150 int wired_count; /* can be paged if = 0 */
984263bc
MD
151};
152
153#define MAP_ENTRY_NOSYNC 0x0001
c809941b 154#define MAP_ENTRY_STACK 0x0002
984263bc
MD
155#define MAP_ENTRY_COW 0x0004
156#define MAP_ENTRY_NEEDS_COPY 0x0008
157#define MAP_ENTRY_NOFAULT 0x0010
158#define MAP_ENTRY_USER_WIRED 0x0020
159
160#define MAP_ENTRY_BEHAV_NORMAL 0x0000 /* default behavior */
161#define MAP_ENTRY_BEHAV_SEQUENTIAL 0x0040 /* expect sequential access */
162#define MAP_ENTRY_BEHAV_RANDOM 0x0080 /* expect random access */
163#define MAP_ENTRY_BEHAV_RESERVED 0x00C0 /* future use */
164
165#define MAP_ENTRY_BEHAV_MASK 0x00C0
166
167#define MAP_ENTRY_IN_TRANSITION 0x0100 /* entry being changed */
168#define MAP_ENTRY_NEEDS_WAKEUP 0x0200 /* waiter's in transition */
169#define MAP_ENTRY_NOCOREDUMP 0x0400 /* don't include in a core */
170
171/*
172 * flags for vm_map_[un]clip_range()
173 */
174#define MAP_CLIP_NO_HOLES 0x0001
175
a108bf71
MD
176/*
177 * This reserve count for vm_map_entry_reserve() should cover all nominal
178 * single-insertion operations, including any necessary clipping.
179 */
180#define MAP_RESERVE_COUNT 4
181#define MAP_RESERVE_SLOP 32
182
984263bc
MD
183static __inline u_char
184vm_map_entry_behavior(struct vm_map_entry *entry)
185{
186 return entry->eflags & MAP_ENTRY_BEHAV_MASK;
187}
188
189static __inline void
190vm_map_entry_set_behavior(struct vm_map_entry *entry, u_char behavior)
191{
192 entry->eflags = (entry->eflags & ~MAP_ENTRY_BEHAV_MASK) |
193 (behavior & MAP_ENTRY_BEHAV_MASK);
194}
195
196/*
197 * Maps are doubly-linked lists of map entries, kept sorted
198 * by address. A single hint is provided to start
199 * searches again from the last successful search,
200 * insertion, or removal.
201 *
202 * Note: the lock structure cannot be the first element of vm_map
203 * because this can result in a running lockup between two or more
204 * system processes trying to kmem_alloc_wait() due to kmem_alloc_wait()
205 * and free tsleep/waking up 'map' and the underlying lockmgr also
206 * sleeping and waking up on 'map'. The lockup occurs when the map fills
207 * up. The 'exec' map, for example.
208 */
209struct vm_map {
210 struct vm_map_entry header; /* List of entries */
686dbf64 211 RB_HEAD(vm_map_rb_tree, vm_map_entry) rb_root;
984263bc
MD
212 struct lock lock; /* Lock for map data */
213 int nentries; /* Number of entries */
214 vm_size_t size; /* virtual size */
215 u_char system_map; /* Am I a system map? */
216 u_char infork; /* Am I in fork processing? */
217 vm_map_entry_t hint; /* hint for quick lookups */
218 unsigned int timestamp; /* Version number */
219 vm_map_entry_t first_free; /* First free space hint */
220 struct pmap *pmap; /* Physical map */
221#define min_offset header.start
222#define max_offset header.end
223};
224
a722be49
MD
225/*
226 * Registered upcall
227 */
228struct upcall;
229
230struct vmupcall {
231 struct vmupcall *vu_next;
232 void *vu_func; /* user upcall function */
233 void *vu_data; /* user data */
234 void *vu_ctx; /* user context function */
4170cc8b 235 struct lwp *vu_lwp; /* process that registered upcall */
a722be49
MD
236 int vu_id; /* upcall identifier */
237 int vu_pending; /* upcall request pending */
238};
239
984263bc
MD
240/*
241 * Shareable process virtual address space.
e3161323
MD
242 *
243 * Refd pointers from vmresident, proc
984263bc
MD
244 */
245struct vmspace {
246 struct vm_map vm_map; /* VM address map */
247 struct pmap vm_pmap; /* private physical map */
e3161323 248 int vm_unused01;
984263bc
MD
249 caddr_t vm_shm; /* SYS5 shared memory private data XXX */
250/* we copy from vm_startcopy to the end of the structure on fork */
251#define vm_startcopy vm_rssize
252 segsz_t vm_rssize; /* current resident set size in pages */
253 segsz_t vm_swrss; /* resident set size before last swap */
254 segsz_t vm_tsize; /* text size (pages) XXX */
255 segsz_t vm_dsize; /* data size (pages) XXX */
256 segsz_t vm_ssize; /* stack size (pages) */
257 caddr_t vm_taddr; /* user virtual address of text XXX */
258 caddr_t vm_daddr; /* user virtual address of data XXX */
259 caddr_t vm_maxsaddr; /* user VA at max stack growth */
260 caddr_t vm_minsaddr; /* user VA at max stack growth */
239b4df9 261#define vm_endcopy vm_exitingcnt
984263bc 262 int vm_exitingcnt; /* several procsses zombied in exit1 */
a722be49 263 int vm_upccount; /* number of registered upcalls */
46311ac2 264 int vm_pagesupply;
e3161323
MD
265 struct vmupcall *vm_upcalls; /* registered upcalls */
266 struct sysref vm_sysref; /* sysref, refcnt, etc */
984263bc
MD
267};
268
29802dbb
MD
269/*
270 * Resident executable holding structure. A user program can take a snapshot
271 * of just its VM address space (typically done just after dynamic link
272 * libraries have completed loading) and register it as a resident
273 * executable associated with the program binary's vnode, which is also
274 * locked into memory. Future execs of the vnode will start with a copy
275 * of the resident vmspace instead of running the binary from scratch,
276 * avoiding both the kernel ELF loader *AND* all shared library mapping and
277 * relocation code, and will call a different entry point (the stack pointer
278 * is reset to the top of the stack) supplied when the vmspace was registered.
279 */
280struct vmresident {
281 struct vnode *vr_vnode; /* associated vnode */
282 TAILQ_ENTRY(vmresident) vr_link; /* linked list of res sts */
283 struct vmspace *vr_vmspace; /* vmspace to fork */
284 intptr_t vr_entry_addr; /* registered entry point */
285 struct sysentvec *vr_sysent; /* system call vects */
286 int vr_id; /* registration id */
287};
288
d5fa53d4 289#ifdef _KERNEL
984263bc
MD
290/*
291 * Macros: vm_map_lock, etc.
292 * Function:
293 * Perform locking on the data portion of a map. Note that
294 * these macros mimic procedure calls returning void. The
295 * semicolon is supplied by the user of these macros, not
296 * by the macros themselves. The macros can safely be used
297 * as unbraced elements in a higher level statement.
298 */
299
984263bc
MD
300#ifdef DIAGNOSTIC
301/* #define MAP_LOCK_DIAGNOSTIC 1 */
302#ifdef MAP_LOCK_DIAGNOSTIC
303#define vm_map_lock(map) \
304 do { \
086c1d7e 305 kprintf ("locking map LK_EXCLUSIVE: 0x%x\n", map); \
df4f70a6 306 if (lockmgr(&(map)->lock, LK_EXCLUSIVE) != 0) { \
984263bc
MD
307 panic("vm_map_lock: failed to get lock"); \
308 } \
309 (map)->timestamp++; \
310 } while(0)
311#else
312#define vm_map_lock(map) \
313 do { \
df4f70a6 314 if (lockmgr(&(map)->lock, LK_EXCLUSIVE) != 0) { \
984263bc
MD
315 panic("vm_map_lock: failed to get lock"); \
316 } \
317 (map)->timestamp++; \
318 } while(0)
319#endif
320#else
321#define vm_map_lock(map) \
322 do { \
df4f70a6 323 lockmgr(&(map)->lock, LK_EXCLUSIVE); \
984263bc
MD
324 (map)->timestamp++; \
325 } while(0)
326#endif /* DIAGNOSTIC */
327
328#if defined(MAP_LOCK_DIAGNOSTIC)
329#define vm_map_unlock(map) \
330 do { \
086c1d7e 331 kprintf ("locking map LK_RELEASE: 0x%x\n", map); \
df4f70a6 332 lockmgr(&(map)->lock, LK_RELEASE); \
984263bc
MD
333 } while (0)
334#define vm_map_lock_read(map) \
335 do { \
086c1d7e 336 kprintf ("locking map LK_SHARED: 0x%x\n", map); \
df4f70a6 337 lockmgr(&(map)->lock, LK_SHARED); \
984263bc
MD
338 } while (0)
339#define vm_map_unlock_read(map) \
340 do { \
086c1d7e 341 kprintf ("locking map LK_RELEASE: 0x%x\n", map); \
df4f70a6 342 lockmgr(&(map)->lock, LK_RELEASE); \
984263bc
MD
343 } while (0)
344#else
345#define vm_map_unlock(map) \
df4f70a6 346 lockmgr(&(map)->lock, LK_RELEASE)
984263bc 347#define vm_map_lock_read(map) \
df4f70a6 348 lockmgr(&(map)->lock, LK_SHARED)
984263bc 349#define vm_map_unlock_read(map) \
df4f70a6 350 lockmgr(&(map)->lock, LK_RELEASE)
984263bc
MD
351#endif
352
353static __inline__ int
df4f70a6 354vm_map_lock_upgrade(vm_map_t map) {
984263bc
MD
355 int error;
356#if defined(MAP_LOCK_DIAGNOSTIC)
086c1d7e 357 kprintf("locking map LK_EXCLUPGRADE: 0x%x\n", map);
984263bc 358#endif
df4f70a6 359 error = lockmgr(&map->lock, LK_EXCLUPGRADE);
984263bc
MD
360 if (error == 0)
361 map->timestamp++;
362 return error;
363}
364
984263bc
MD
365#if defined(MAP_LOCK_DIAGNOSTIC)
366#define vm_map_lock_downgrade(map) \
367 do { \
086c1d7e 368 kprintf ("locking map LK_DOWNGRADE: 0x%x\n", map); \
df4f70a6 369 lockmgr(&(map)->lock, LK_DOWNGRADE); \
984263bc
MD
370 } while (0)
371#else
372#define vm_map_lock_downgrade(map) \
df4f70a6 373 lockmgr(&(map)->lock, LK_DOWNGRADE)
984263bc
MD
374#endif
375
d5fa53d4
HP
376#endif /* _KERNEL */
377
984263bc
MD
378/*
379 * Functions implemented as macros
380 */
381#define vm_map_min(map) ((map)->min_offset)
382#define vm_map_max(map) ((map)->max_offset)
383#define vm_map_pmap(map) ((map)->pmap)
384
385static __inline struct pmap *
386vmspace_pmap(struct vmspace *vmspace)
387{
388 return &vmspace->vm_pmap;
389}
390
391static __inline long
392vmspace_resident_count(struct vmspace *vmspace)
393{
394 return pmap_resident_count(vmspace_pmap(vmspace));
395}
396
a108bf71
MD
397/*
398 * Number of kernel maps and entries to statically allocate, required
399 * during boot to bootstrap the VM system.
400 */
984263bc 401#define MAX_KMAP 10
a108bf71 402#define MAX_MAPENT 256
984263bc
MD
403
404/*
405 * Copy-on-write flags for vm_map operations
406 */
407#define MAP_UNUSED_01 0x0001
408#define MAP_COPY_ON_WRITE 0x0002
409#define MAP_NOFAULT 0x0004
410#define MAP_PREFAULT 0x0008
411#define MAP_PREFAULT_PARTIAL 0x0010
412#define MAP_DISABLE_SYNCER 0x0020
c809941b 413#define MAP_IS_STACK 0x0040
984263bc
MD
414#define MAP_DISABLE_COREDUMP 0x0100
415#define MAP_PREFAULT_MADVISE 0x0200 /* from (user) madvise request */
416
417/*
418 * vm_fault option flags
419 */
0035dca9
MD
420#define VM_FAULT_NORMAL 0x00 /* Nothing special */
421#define VM_FAULT_CHANGE_WIRING 0x01 /* Change the wiring as appropriate */
422#define VM_FAULT_USER_WIRE 0x02 /* Likewise, but for user purposes */
1b9d3514 423#define VM_FAULT_BURST 0x04 /* Burst fault can be done */
0035dca9
MD
424#define VM_FAULT_DIRTY 0x08 /* Dirty the page */
425#define VM_FAULT_WIRE_MASK (VM_FAULT_CHANGE_WIRING|VM_FAULT_USER_WIRE)
984263bc
MD
426
427#ifdef _KERNEL
e3161323
MD
428
429extern struct sysref_class vmspace_sysref_class;
430
1388df65 431boolean_t vm_map_check_protection (vm_map_t, vm_offset_t, vm_offset_t, vm_prot_t);
984263bc 432struct pmap;
41a01a4d 433struct globaldata;
53025830 434void vm_map_entry_allocate_object(vm_map_entry_t);
41a01a4d 435void vm_map_entry_reserve_cpu_init(struct globaldata *gd);
a108bf71
MD
436int vm_map_entry_reserve(int);
437int vm_map_entry_kreserve(int);
438void vm_map_entry_release(int);
439void vm_map_entry_krelease(int);
e4846942 440vm_map_t vm_map_create (vm_map_t, struct pmap *, vm_offset_t, vm_offset_t);
a108bf71 441int vm_map_delete (vm_map_t, vm_offset_t, vm_offset_t, int *);
1b874851
MD
442int vm_map_find (vm_map_t, vm_object_t, vm_ooffset_t,
443 vm_offset_t *, vm_size_t,
444 boolean_t,
445 vm_maptype_t,
446 vm_prot_t, vm_prot_t,
447 int);
c809941b
MD
448int vm_map_findspace (vm_map_t, vm_offset_t, vm_size_t, vm_offset_t,
449 int, vm_offset_t *);
1388df65 450int vm_map_inherit (vm_map_t, vm_offset_t, vm_offset_t, vm_inherit_t);
e4846942 451void vm_map_init (struct vm_map *, vm_offset_t, vm_offset_t, pmap_t);
1b874851
MD
452int vm_map_insert (vm_map_t, int *, vm_object_t, vm_ooffset_t,
453 vm_offset_t, vm_offset_t,
454 vm_maptype_t,
455 vm_prot_t, vm_prot_t,
456 int);
1388df65
RG
457int vm_map_lookup (vm_map_t *, vm_offset_t, vm_prot_t, vm_map_entry_t *, vm_object_t *,
458 vm_pindex_t *, vm_prot_t *, boolean_t *);
a108bf71 459void vm_map_lookup_done (vm_map_t, vm_map_entry_t, int);
1388df65 460boolean_t vm_map_lookup_entry (vm_map_t, vm_offset_t, vm_map_entry_t *);
e1359933 461int vm_map_wire (vm_map_t, vm_offset_t, vm_offset_t, int);
cde87949 462int vm_map_unwire (vm_map_t, vm_offset_t, vm_offset_t, boolean_t);
1388df65
RG
463int vm_map_clean (vm_map_t, vm_offset_t, vm_offset_t, boolean_t, boolean_t);
464int vm_map_protect (vm_map_t, vm_offset_t, vm_offset_t, vm_prot_t, boolean_t);
465int vm_map_remove (vm_map_t, vm_offset_t, vm_offset_t);
466void vm_map_startup (void);
467int vm_map_submap (vm_map_t, vm_offset_t, vm_offset_t, vm_map_t);
afeabdca 468int vm_map_madvise (vm_map_t, vm_offset_t, vm_offset_t, int, off_t);
a108bf71 469void vm_map_simplify_entry (vm_map_t, vm_map_entry_t, int *);
1388df65
RG
470void vm_init2 (void);
471int vm_uiomove (vm_map_t, vm_object_t, off_t, int, vm_offset_t, int *);
c809941b 472int vm_map_stack (vm_map_t, vm_offset_t, vm_size_t, int,
85d25bcf 473 vm_prot_t, vm_prot_t, int);
1388df65
RG
474int vm_map_growstack (struct proc *p, vm_offset_t addr);
475int vmspace_swap_count (struct vmspace *vmspace);
20479584 476int vmspace_anonymous_count (struct vmspace *vmspace);
a108bf71 477void vm_map_set_wired_quick(vm_map_t map, vm_offset_t addr, vm_size_t size, int *);
984263bc
MD
478
479#endif
1bd40720 480#endif /* _VM_VM_MAP_H_ */