blob: 3f1fafc0245e7eebf0ee45334d3c7c3ff77f4a89 [file] [log] [blame]
Linus Torvalds1da177e2005-04-16 15:20:36 -07001#ifndef _LINUX_MM_H
2#define _LINUX_MM_H
3
4#include <linux/sched.h>
5#include <linux/errno.h>
6
7#ifdef __KERNEL__
8
9#include <linux/config.h>
10#include <linux/gfp.h>
11#include <linux/list.h>
12#include <linux/mmzone.h>
13#include <linux/rbtree.h>
14#include <linux/prio_tree.h>
15#include <linux/fs.h>
Ingo Molnarde5097c2006-01-09 15:59:21 -080016#include <linux/mutex.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070017
18struct mempolicy;
19struct anon_vma;
20
21#ifndef CONFIG_DISCONTIGMEM /* Don't use mapnrs, do it properly */
22extern unsigned long max_mapnr;
23#endif
24
25extern unsigned long num_physpages;
26extern void * high_memory;
27extern unsigned long vmalloc_earlyreserve;
28extern int page_cluster;
29
30#ifdef CONFIG_SYSCTL
31extern int sysctl_legacy_va_layout;
32#else
33#define sysctl_legacy_va_layout 0
34#endif
35
36#include <asm/page.h>
37#include <asm/pgtable.h>
38#include <asm/processor.h>
39#include <asm/atomic.h>
40
Linus Torvalds1da177e2005-04-16 15:20:36 -070041#define nth_page(page,n) pfn_to_page(page_to_pfn((page)) + (n))
42
43/*
44 * Linux kernel virtual memory manager primitives.
45 * The idea being to have a "virtual" mm in the same way
46 * we have a virtual fs - giving a cleaner interface to the
47 * mm details, and allowing different kinds of memory mappings
48 * (from shared memory to executable loading to arbitrary
49 * mmap() functions).
50 */
51
52/*
53 * This struct defines a memory VMM memory area. There is one of these
54 * per VM-area/task. A VM area is any part of the process virtual memory
55 * space that has a special rule for the page-fault handlers (ie a shared
56 * library, the executable area etc).
57 */
58struct vm_area_struct {
59 struct mm_struct * vm_mm; /* The address space we belong to. */
60 unsigned long vm_start; /* Our start address within vm_mm. */
61 unsigned long vm_end; /* The first byte after our end address
62 within vm_mm. */
63
64 /* linked list of VM areas per task, sorted by address */
65 struct vm_area_struct *vm_next;
66
67 pgprot_t vm_page_prot; /* Access permissions of this VMA. */
68 unsigned long vm_flags; /* Flags, listed below. */
69
70 struct rb_node vm_rb;
71
72 /*
73 * For areas with an address space and backing store,
74 * linkage into the address_space->i_mmap prio tree, or
75 * linkage to the list of like vmas hanging off its node, or
76 * linkage of vma in the address_space->i_mmap_nonlinear list.
77 */
78 union {
79 struct {
80 struct list_head list;
81 void *parent; /* aligns with prio_tree_node parent */
82 struct vm_area_struct *head;
83 } vm_set;
84
85 struct raw_prio_tree_node prio_tree_node;
86 } shared;
87
88 /*
89 * A file's MAP_PRIVATE vma can be in both i_mmap tree and anon_vma
90 * list, after a COW of one of the file pages. A MAP_SHARED vma
91 * can only be in the i_mmap tree. An anonymous MAP_PRIVATE, stack
92 * or brk vma (with NULL file) can only be in an anon_vma list.
93 */
94 struct list_head anon_vma_node; /* Serialized by anon_vma->lock */
95 struct anon_vma *anon_vma; /* Serialized by page_table_lock */
96
97 /* Function pointers to deal with this struct. */
98 struct vm_operations_struct * vm_ops;
99
100 /* Information about our backing store: */
101 unsigned long vm_pgoff; /* Offset (within vm_file) in PAGE_SIZE
102 units, *not* PAGE_CACHE_SIZE */
103 struct file * vm_file; /* File we map to (can be NULL). */
104 void * vm_private_data; /* was vm_pte (shared mem) */
105 unsigned long vm_truncate_count;/* truncate_count or restart_addr */
106
107#ifndef CONFIG_MMU
108 atomic_t vm_usage; /* refcount (VMAs shared if !MMU) */
109#endif
110#ifdef CONFIG_NUMA
111 struct mempolicy *vm_policy; /* NUMA policy for the VMA */
112#endif
113};
114
115/*
116 * This struct defines the per-mm list of VMAs for uClinux. If CONFIG_MMU is
117 * disabled, then there's a single shared list of VMAs maintained by the
118 * system, and mm's subscribe to these individually
119 */
120struct vm_list_struct {
121 struct vm_list_struct *next;
122 struct vm_area_struct *vma;
123};
124
125#ifndef CONFIG_MMU
126extern struct rb_root nommu_vma_tree;
127extern struct rw_semaphore nommu_vma_sem;
128
129extern unsigned int kobjsize(const void *objp);
130#endif
131
132/*
133 * vm_flags..
134 */
135#define VM_READ 0x00000001 /* currently active flags */
136#define VM_WRITE 0x00000002
137#define VM_EXEC 0x00000004
138#define VM_SHARED 0x00000008
139
Paolo 'Blaisorblade' Giarrusso7e2cff42005-09-21 09:55:39 -0700140/* mprotect() hardcodes VM_MAYREAD >> 4 == VM_READ, and so for r/w/x bits. */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700141#define VM_MAYREAD 0x00000010 /* limits for mprotect() etc */
142#define VM_MAYWRITE 0x00000020
143#define VM_MAYEXEC 0x00000040
144#define VM_MAYSHARE 0x00000080
145
146#define VM_GROWSDOWN 0x00000100 /* general info on the segment */
147#define VM_GROWSUP 0x00000200
Hugh Dickins0b14c172005-11-21 21:32:15 -0800148#define VM_SHM 0x00000000 /* Means nothing: delete it later */
Linus Torvalds6aab3412005-11-28 14:34:23 -0800149#define VM_PFNMAP 0x00000400 /* Page-ranges managed without "struct page", just pure PFN */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700150#define VM_DENYWRITE 0x00000800 /* ETXTBSY on write attempts.. */
151
152#define VM_EXECUTABLE 0x00001000
153#define VM_LOCKED 0x00002000
154#define VM_IO 0x00004000 /* Memory mapped I/O or similar */
155
156 /* Used by sys_madvise() */
157#define VM_SEQ_READ 0x00008000 /* App will access data sequentially */
158#define VM_RAND_READ 0x00010000 /* App will not benefit from clustered reads */
159
160#define VM_DONTCOPY 0x00020000 /* Do not copy this vma on fork */
161#define VM_DONTEXPAND 0x00040000 /* Cannot expand with mremap() */
Hugh Dickins0b14c172005-11-21 21:32:15 -0800162#define VM_RESERVED 0x00080000 /* Count as reserved_vm like IO */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700163#define VM_ACCOUNT 0x00100000 /* Is a VM accounted object */
164#define VM_HUGETLB 0x00400000 /* Huge TLB Page VM */
165#define VM_NONLINEAR 0x00800000 /* Is non-linear (remap_file_pages) */
166#define VM_MAPPED_COPY 0x01000000 /* T if mapped copy of data (nommu mmap) */
Linus Torvalds4d7672b2005-12-16 10:21:23 -0800167#define VM_INSERTPAGE 0x02000000 /* The vma has had "vm_insert_page()" done on it */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700168
169#ifndef VM_STACK_DEFAULT_FLAGS /* arch can override this */
170#define VM_STACK_DEFAULT_FLAGS VM_DATA_DEFAULT_FLAGS
171#endif
172
173#ifdef CONFIG_STACK_GROWSUP
174#define VM_STACK_FLAGS (VM_GROWSUP | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT)
175#else
176#define VM_STACK_FLAGS (VM_GROWSDOWN | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT)
177#endif
178
179#define VM_READHINTMASK (VM_SEQ_READ | VM_RAND_READ)
180#define VM_ClearReadHint(v) (v)->vm_flags &= ~VM_READHINTMASK
181#define VM_NormalReadHint(v) (!((v)->vm_flags & VM_READHINTMASK))
182#define VM_SequentialReadHint(v) ((v)->vm_flags & VM_SEQ_READ)
183#define VM_RandomReadHint(v) ((v)->vm_flags & VM_RAND_READ)
184
185/*
186 * mapping from the currently active vm_flags protection bits (the
187 * low four bits) to a page protection mask..
188 */
189extern pgprot_t protection_map[16];
190
191
192/*
193 * These are the virtual MM functions - opening of an area, closing and
194 * unmapping it (needed to keep files on disk up-to-date etc), pointer
195 * to the functions called when a no-page or a wp-page exception occurs.
196 */
197struct vm_operations_struct {
198 void (*open)(struct vm_area_struct * area);
199 void (*close)(struct vm_area_struct * area);
200 struct page * (*nopage)(struct vm_area_struct * area, unsigned long address, int *type);
201 int (*populate)(struct vm_area_struct * area, unsigned long address, unsigned long len, pgprot_t prot, unsigned long pgoff, int nonblock);
202#ifdef CONFIG_NUMA
203 int (*set_policy)(struct vm_area_struct *vma, struct mempolicy *new);
204 struct mempolicy *(*get_policy)(struct vm_area_struct *vma,
205 unsigned long addr);
206#endif
207};
208
209struct mmu_gather;
210struct inode;
211
Linus Torvalds1da177e2005-04-16 15:20:36 -0700212/*
213 * Each physical page in the system has a struct page associated with
214 * it to keep track of whatever it is we are using the page for at the
215 * moment. Note that we have no way to track which tasks are using
216 * a page.
217 */
218struct page {
Andi Kleen07808b72005-11-05 17:25:53 +0100219 unsigned long flags; /* Atomic flags, some possibly
Linus Torvalds1da177e2005-04-16 15:20:36 -0700220 * updated asynchronously */
221 atomic_t _count; /* Usage count, see below. */
222 atomic_t _mapcount; /* Count of ptes mapped in mms,
223 * to show when page is mapped
224 * & limit reverse map searches.
225 */
Hugh Dickins4c21e2f2005-10-29 18:16:40 -0700226 union {
Andrew Morton349aef02006-01-08 01:04:36 -0800227 struct {
228 unsigned long private; /* Mapping-private opaque data:
229 * usually used for buffer_heads
230 * if PagePrivate set; used for
231 * swp_entry_t if PageSwapCache.
232 * When page is free, this
233 * indicates order in the buddy
234 * system.
235 */
236 struct address_space *mapping; /* If low bit clear, points to
237 * inode address_space, or NULL.
238 * If page mapped as anonymous
239 * memory, low bit is set, and
240 * it points to anon_vma object:
241 * see PAGE_MAPPING_ANON below.
242 */
243 };
Hugh Dickins4c21e2f2005-10-29 18:16:40 -0700244#if NR_CPUS >= CONFIG_SPLIT_PTLOCK_CPUS
Andrew Morton349aef02006-01-08 01:04:36 -0800245 spinlock_t ptl;
Hugh Dickins4c21e2f2005-10-29 18:16:40 -0700246#endif
Andrew Morton349aef02006-01-08 01:04:36 -0800247 };
Linus Torvalds1da177e2005-04-16 15:20:36 -0700248 pgoff_t index; /* Our offset within mapping. */
249 struct list_head lru; /* Pageout list, eg. active_list
250 * protected by zone->lru_lock !
251 */
252 /*
253 * On machines where all RAM is mapped into kernel address space,
254 * we can simply calculate the virtual address. On machines with
255 * highmem some memory is mapped into kernel virtual memory
256 * dynamically, so we need a place to store that address.
257 * Note that this field could be 16 bits on x86 ... ;)
258 *
259 * Architectures with slow multiplication can define
260 * WANT_PAGE_VIRTUAL in asm/page.h
261 */
262#if defined(WANT_PAGE_VIRTUAL)
263 void *virtual; /* Kernel virtual address (NULL if
264 not kmapped, ie. highmem) */
265#endif /* WANT_PAGE_VIRTUAL */
266};
267
Andrew Morton349aef02006-01-08 01:04:36 -0800268#define page_private(page) ((page)->private)
269#define set_page_private(page, v) ((page)->private = (v))
Hugh Dickins4c21e2f2005-10-29 18:16:40 -0700270
Linus Torvalds1da177e2005-04-16 15:20:36 -0700271/*
272 * FIXME: take this include out, include page-flags.h in
273 * files which need it (119 of them)
274 */
275#include <linux/page-flags.h>
276
277/*
278 * Methods to modify the page usage count.
279 *
280 * What counts for a page usage:
281 * - cache mapping (page->mapping)
282 * - private data (page->private)
283 * - page mapped in a task's page tables, each mapping
284 * is counted separately
285 *
286 * Also, many kernel routines increase the page count before a critical
287 * routine so they can be sure the page doesn't go away from under them.
288 *
289 * Since 2.6.6 (approx), a free page has ->_count = -1. This is so that we
290 * can use atomic_add_negative(-1, page->_count) to detect when the page
291 * becomes free and so that we can also use atomic_inc_and_test to atomically
292 * detect when we just tried to grab a ref on a page which some other CPU has
293 * already deemed to be freeable.
294 *
295 * NO code should make assumptions about this internal detail! Use the provided
296 * macros which retain the old rules: page_count(page) == 0 is a free page.
297 */
298
299/*
300 * Drop a ref, return true if the logical refcount fell to zero (the page has
301 * no users)
302 */
303#define put_page_testzero(p) \
304 ({ \
305 BUG_ON(page_count(p) == 0); \
306 atomic_add_negative(-1, &(p)->_count); \
307 })
308
309/*
310 * Grab a ref, return true if the page previously had a logical refcount of
311 * zero. ie: returns true if we just grabbed an already-deemed-to-be-free page
312 */
313#define get_page_testone(p) atomic_inc_and_test(&(p)->_count)
314
Avishay Traeger152194a2006-01-08 01:00:58 -0800315#define set_page_count(p,v) atomic_set(&(p)->_count, (v) - 1)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700316#define __put_page(p) atomic_dec(&(p)->_count)
317
318extern void FASTCALL(__page_cache_release(struct page *));
319
Hugh Dickins4c21e2f2005-10-29 18:16:40 -0700320static inline int page_count(struct page *page)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700321{
Hugh Dickins4c21e2f2005-10-29 18:16:40 -0700322 if (PageCompound(page))
323 page = (struct page *)page_private(page);
324 return atomic_read(&page->_count) + 1;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700325}
326
327static inline void get_page(struct page *page)
328{
329 if (unlikely(PageCompound(page)))
Hugh Dickins4c21e2f2005-10-29 18:16:40 -0700330 page = (struct page *)page_private(page);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700331 atomic_inc(&page->_count);
332}
333
334void put_page(struct page *page);
335
Linus Torvalds1da177e2005-04-16 15:20:36 -0700336/*
337 * Multiple processes may "see" the same page. E.g. for untouched
338 * mappings of /dev/null, all processes see the same page full of
339 * zeroes, and text pages of executables and shared libraries have
340 * only one copy in memory, at most, normally.
341 *
342 * For the non-reserved pages, page_count(page) denotes a reference count.
Paolo 'Blaisorblade' Giarrusso7e871b62005-09-21 09:55:38 -0700343 * page_count() == 0 means the page is free. page->lru is then used for
344 * freelist management in the buddy allocator.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700345 * page_count() == 1 means the page is used for exactly one purpose
346 * (e.g. a private data page of one process).
347 *
348 * A page may be used for kmalloc() or anyone else who does a
349 * __get_free_page(). In this case the page_count() is at least 1, and
350 * all other fields are unused but should be 0 or NULL. The
351 * management of this page is the responsibility of the one who uses
352 * it.
353 *
354 * The other pages (we may call them "process pages") are completely
355 * managed by the Linux memory manager: I/O, buffers, swapping etc.
356 * The following discussion applies only to them.
357 *
358 * A page may belong to an inode's memory mapping. In this case,
359 * page->mapping is the pointer to the inode, and page->index is the
360 * file offset of the page, in units of PAGE_CACHE_SIZE.
361 *
362 * A page contains an opaque `private' member, which belongs to the
363 * page's address_space. Usually, this is the address of a circular
364 * list of the page's disk buffers.
365 *
366 * For pages belonging to inodes, the page_count() is the number of
367 * attaches, plus 1 if `private' contains something, plus one for
368 * the page cache itself.
369 *
Paolo 'Blaisorblade' Giarrusso7e871b62005-09-21 09:55:38 -0700370 * Instead of keeping dirty/clean pages in per address-space lists, we instead
371 * now tag pages as dirty/under writeback in the radix tree.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700372 *
373 * There is also a per-mapping radix tree mapping index to the page
374 * in memory if present. The tree is rooted at mapping->root.
375 *
376 * All process pages can do I/O:
377 * - inode pages may need to be read from disk,
378 * - inode pages which have been modified and are MAP_SHARED may need
379 * to be written to disk,
380 * - private pages which have been modified may need to be swapped out
381 * to swap space and (later) to be read back into memory.
382 */
383
384/*
385 * The zone field is never updated after free_area_init_core()
386 * sets it, so none of the operations on it need to be atomic.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700387 */
Dave Hansen348f8b62005-06-23 00:07:40 -0700388
Andy Whitcroftd41dee32005-06-23 00:07:54 -0700389
390/*
391 * page->flags layout:
392 *
393 * There are three possibilities for how page->flags get
394 * laid out. The first is for the normal case, without
395 * sparsemem. The second is for sparsemem when there is
396 * plenty of space for node and section. The last is when
397 * we have run out of space and have to fall back to an
398 * alternate (slower) way of determining the node.
399 *
400 * No sparsemem: | NODE | ZONE | ... | FLAGS |
401 * with space for node: | SECTION | NODE | ZONE | ... | FLAGS |
402 * no space for node: | SECTION | ZONE | ... | FLAGS |
403 */
404#ifdef CONFIG_SPARSEMEM
405#define SECTIONS_WIDTH SECTIONS_SHIFT
406#else
407#define SECTIONS_WIDTH 0
408#endif
409
410#define ZONES_WIDTH ZONES_SHIFT
411
412#if SECTIONS_WIDTH+ZONES_WIDTH+NODES_SHIFT <= FLAGS_RESERVED
413#define NODES_WIDTH NODES_SHIFT
414#else
415#define NODES_WIDTH 0
416#endif
417
418/* Page flags: | [SECTION] | [NODE] | ZONE | ... | FLAGS | */
Andi Kleen07808b72005-11-05 17:25:53 +0100419#define SECTIONS_PGOFF ((sizeof(unsigned long)*8) - SECTIONS_WIDTH)
Andy Whitcroftd41dee32005-06-23 00:07:54 -0700420#define NODES_PGOFF (SECTIONS_PGOFF - NODES_WIDTH)
421#define ZONES_PGOFF (NODES_PGOFF - ZONES_WIDTH)
422
423/*
424 * We are going to use the flags for the page to node mapping if its in
425 * there. This includes the case where there is no node, so it is implicit.
426 */
427#define FLAGS_HAS_NODE (NODES_WIDTH > 0 || NODES_SHIFT == 0)
428
429#ifndef PFN_SECTION_SHIFT
430#define PFN_SECTION_SHIFT 0
431#endif
Dave Hansen348f8b62005-06-23 00:07:40 -0700432
433/*
434 * Define the bit shifts to access each section. For non-existant
435 * sections we define the shift as 0; that plus a 0 mask ensures
436 * the compiler will optimise away reference to them.
437 */
Andy Whitcroftd41dee32005-06-23 00:07:54 -0700438#define SECTIONS_PGSHIFT (SECTIONS_PGOFF * (SECTIONS_WIDTH != 0))
439#define NODES_PGSHIFT (NODES_PGOFF * (NODES_WIDTH != 0))
440#define ZONES_PGSHIFT (ZONES_PGOFF * (ZONES_WIDTH != 0))
Dave Hansen348f8b62005-06-23 00:07:40 -0700441
Andy Whitcroftd41dee32005-06-23 00:07:54 -0700442/* NODE:ZONE or SECTION:ZONE is used to lookup the zone from a page. */
443#if FLAGS_HAS_NODE
Dave Hansen348f8b62005-06-23 00:07:40 -0700444#define ZONETABLE_SHIFT (NODES_SHIFT + ZONES_SHIFT)
Andy Whitcroftd41dee32005-06-23 00:07:54 -0700445#else
446#define ZONETABLE_SHIFT (SECTIONS_SHIFT + ZONES_SHIFT)
447#endif
Dave Hansen348f8b62005-06-23 00:07:40 -0700448#define ZONETABLE_PGSHIFT ZONES_PGSHIFT
449
Andy Whitcroftd41dee32005-06-23 00:07:54 -0700450#if SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > FLAGS_RESERVED
451#error SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > FLAGS_RESERVED
Dave Hansen348f8b62005-06-23 00:07:40 -0700452#endif
453
Andy Whitcroftd41dee32005-06-23 00:07:54 -0700454#define ZONES_MASK ((1UL << ZONES_WIDTH) - 1)
455#define NODES_MASK ((1UL << NODES_WIDTH) - 1)
456#define SECTIONS_MASK ((1UL << SECTIONS_WIDTH) - 1)
Dave Hansen348f8b62005-06-23 00:07:40 -0700457#define ZONETABLE_MASK ((1UL << ZONETABLE_SHIFT) - 1)
458
Linus Torvalds1da177e2005-04-16 15:20:36 -0700459static inline unsigned long page_zonenum(struct page *page)
460{
Dave Hansen348f8b62005-06-23 00:07:40 -0700461 return (page->flags >> ZONES_PGSHIFT) & ZONES_MASK;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700462}
Linus Torvalds1da177e2005-04-16 15:20:36 -0700463
464struct zone;
465extern struct zone *zone_table[];
466
467static inline struct zone *page_zone(struct page *page)
468{
Dave Hansen348f8b62005-06-23 00:07:40 -0700469 return zone_table[(page->flags >> ZONETABLE_PGSHIFT) &
470 ZONETABLE_MASK];
Linus Torvalds1da177e2005-04-16 15:20:36 -0700471}
472
Andy Whitcroftd41dee32005-06-23 00:07:54 -0700473static inline unsigned long page_to_nid(struct page *page)
474{
475 if (FLAGS_HAS_NODE)
476 return (page->flags >> NODES_PGSHIFT) & NODES_MASK;
477 else
478 return page_zone(page)->zone_pgdat->node_id;
479}
480static inline unsigned long page_to_section(struct page *page)
481{
482 return (page->flags >> SECTIONS_PGSHIFT) & SECTIONS_MASK;
483}
484
Dave Hansen348f8b62005-06-23 00:07:40 -0700485static inline void set_page_zone(struct page *page, unsigned long zone)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700486{
Dave Hansen348f8b62005-06-23 00:07:40 -0700487 page->flags &= ~(ZONES_MASK << ZONES_PGSHIFT);
488 page->flags |= (zone & ZONES_MASK) << ZONES_PGSHIFT;
489}
490static inline void set_page_node(struct page *page, unsigned long node)
491{
492 page->flags &= ~(NODES_MASK << NODES_PGSHIFT);
493 page->flags |= (node & NODES_MASK) << NODES_PGSHIFT;
494}
Andy Whitcroftd41dee32005-06-23 00:07:54 -0700495static inline void set_page_section(struct page *page, unsigned long section)
496{
497 page->flags &= ~(SECTIONS_MASK << SECTIONS_PGSHIFT);
498 page->flags |= (section & SECTIONS_MASK) << SECTIONS_PGSHIFT;
499}
Dave Hansen348f8b62005-06-23 00:07:40 -0700500
501static inline void set_page_links(struct page *page, unsigned long zone,
Andy Whitcroftd41dee32005-06-23 00:07:54 -0700502 unsigned long node, unsigned long pfn)
Dave Hansen348f8b62005-06-23 00:07:40 -0700503{
504 set_page_zone(page, zone);
505 set_page_node(page, node);
Andy Whitcroftd41dee32005-06-23 00:07:54 -0700506 set_page_section(page, pfn_to_section_nr(pfn));
Linus Torvalds1da177e2005-04-16 15:20:36 -0700507}
508
509#ifndef CONFIG_DISCONTIGMEM
510/* The array of struct pages - for discontigmem use pgdat->lmem_map */
511extern struct page *mem_map;
512#endif
513
514static inline void *lowmem_page_address(struct page *page)
515{
516 return __va(page_to_pfn(page) << PAGE_SHIFT);
517}
518
519#if defined(CONFIG_HIGHMEM) && !defined(WANT_PAGE_VIRTUAL)
520#define HASHED_PAGE_VIRTUAL
521#endif
522
523#if defined(WANT_PAGE_VIRTUAL)
524#define page_address(page) ((page)->virtual)
525#define set_page_address(page, address) \
526 do { \
527 (page)->virtual = (address); \
528 } while(0)
529#define page_address_init() do { } while(0)
530#endif
531
532#if defined(HASHED_PAGE_VIRTUAL)
533void *page_address(struct page *page);
534void set_page_address(struct page *page, void *virtual);
535void page_address_init(void);
536#endif
537
538#if !defined(HASHED_PAGE_VIRTUAL) && !defined(WANT_PAGE_VIRTUAL)
539#define page_address(page) lowmem_page_address(page)
540#define set_page_address(page, address) do { } while(0)
541#define page_address_init() do { } while(0)
542#endif
543
544/*
545 * On an anonymous page mapped into a user virtual memory area,
546 * page->mapping points to its anon_vma, not to a struct address_space;
547 * with the PAGE_MAPPING_ANON bit set to distinguish it.
548 *
549 * Please note that, confusingly, "page_mapping" refers to the inode
550 * address_space which maps the page from disk; whereas "page_mapped"
551 * refers to user virtual address space into which the page is mapped.
552 */
553#define PAGE_MAPPING_ANON 1
554
555extern struct address_space swapper_space;
556static inline struct address_space *page_mapping(struct page *page)
557{
558 struct address_space *mapping = page->mapping;
559
560 if (unlikely(PageSwapCache(page)))
561 mapping = &swapper_space;
562 else if (unlikely((unsigned long)mapping & PAGE_MAPPING_ANON))
563 mapping = NULL;
564 return mapping;
565}
566
567static inline int PageAnon(struct page *page)
568{
569 return ((unsigned long)page->mapping & PAGE_MAPPING_ANON) != 0;
570}
571
572/*
573 * Return the pagecache index of the passed page. Regular pagecache pages
574 * use ->index whereas swapcache pages use ->private
575 */
576static inline pgoff_t page_index(struct page *page)
577{
578 if (unlikely(PageSwapCache(page)))
Hugh Dickins4c21e2f2005-10-29 18:16:40 -0700579 return page_private(page);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700580 return page->index;
581}
582
583/*
584 * The atomic page->_mapcount, like _count, starts from -1:
585 * so that transitions both from it and to it can be tracked,
586 * using atomic_inc_and_test and atomic_add_negative(-1).
587 */
588static inline void reset_page_mapcount(struct page *page)
589{
590 atomic_set(&(page)->_mapcount, -1);
591}
592
593static inline int page_mapcount(struct page *page)
594{
595 return atomic_read(&(page)->_mapcount) + 1;
596}
597
598/*
599 * Return true if this page is mapped into pagetables.
600 */
601static inline int page_mapped(struct page *page)
602{
603 return atomic_read(&(page)->_mapcount) >= 0;
604}
605
606/*
607 * Error return values for the *_nopage functions
608 */
609#define NOPAGE_SIGBUS (NULL)
610#define NOPAGE_OOM ((struct page *) (-1))
611
612/*
613 * Different kinds of faults, as returned by handle_mm_fault().
614 * Used to decide whether a process gets delivered SIGBUS or
615 * just gets major/minor fault counters bumped up.
616 */
Nick Pigginf33ea7f2005-08-03 20:24:01 +1000617#define VM_FAULT_OOM 0x00
618#define VM_FAULT_SIGBUS 0x01
619#define VM_FAULT_MINOR 0x02
620#define VM_FAULT_MAJOR 0x03
621
622/*
623 * Special case for get_user_pages.
624 * Must be in a distinct bit from the above VM_FAULT_ flags.
625 */
626#define VM_FAULT_WRITE 0x10
Linus Torvalds1da177e2005-04-16 15:20:36 -0700627
628#define offset_in_page(p) ((unsigned long)(p) & ~PAGE_MASK)
629
630extern void show_free_areas(void);
631
632#ifdef CONFIG_SHMEM
633struct page *shmem_nopage(struct vm_area_struct *vma,
634 unsigned long address, int *type);
635int shmem_set_policy(struct vm_area_struct *vma, struct mempolicy *new);
636struct mempolicy *shmem_get_policy(struct vm_area_struct *vma,
637 unsigned long addr);
638int shmem_lock(struct file *file, int lock, struct user_struct *user);
639#else
640#define shmem_nopage filemap_nopage
Russell King03b00eb2006-01-06 00:10:52 -0800641
642static inline int shmem_lock(struct file *file, int lock,
643 struct user_struct *user)
644{
645 return 0;
646}
647
648static inline int shmem_set_policy(struct vm_area_struct *vma,
649 struct mempolicy *new)
650{
651 return 0;
652}
653
654static inline struct mempolicy *shmem_get_policy(struct vm_area_struct *vma,
655 unsigned long addr)
656{
657 return NULL;
658}
Linus Torvalds1da177e2005-04-16 15:20:36 -0700659#endif
660struct file *shmem_file_setup(char *name, loff_t size, unsigned long flags);
David Howellsb0e15192006-01-06 00:11:42 -0800661extern int shmem_mmap(struct file *file, struct vm_area_struct *vma);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700662
663int shmem_zero_setup(struct vm_area_struct *);
664
David Howellsb0e15192006-01-06 00:11:42 -0800665#ifndef CONFIG_MMU
666extern unsigned long shmem_get_unmapped_area(struct file *file,
667 unsigned long addr,
668 unsigned long len,
669 unsigned long pgoff,
670 unsigned long flags);
671#endif
672
Linus Torvalds1da177e2005-04-16 15:20:36 -0700673static inline int can_do_mlock(void)
674{
675 if (capable(CAP_IPC_LOCK))
676 return 1;
677 if (current->signal->rlim[RLIMIT_MEMLOCK].rlim_cur != 0)
678 return 1;
679 return 0;
680}
681extern int user_shm_lock(size_t, struct user_struct *);
682extern void user_shm_unlock(size_t, struct user_struct *);
683
684/*
685 * Parameter block passed down to zap_pte_range in exceptional cases.
686 */
687struct zap_details {
688 struct vm_area_struct *nonlinear_vma; /* Check page->index if set */
689 struct address_space *check_mapping; /* Check page->mapping if set */
690 pgoff_t first_index; /* Lowest page->index to unmap */
691 pgoff_t last_index; /* Highest page->index to unmap */
692 spinlock_t *i_mmap_lock; /* For unmap_mapping_range: */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700693 unsigned long truncate_count; /* Compare vm_truncate_count */
694};
695
Linus Torvalds6aab3412005-11-28 14:34:23 -0800696struct page *vm_normal_page(struct vm_area_struct *, unsigned long, pte_t);
Hugh Dickinsee39b372005-04-19 13:29:15 -0700697unsigned long zap_page_range(struct vm_area_struct *vma, unsigned long address,
Linus Torvalds1da177e2005-04-16 15:20:36 -0700698 unsigned long size, struct zap_details *);
Hugh Dickins508034a2005-10-29 18:16:30 -0700699unsigned long unmap_vmas(struct mmu_gather **tlb,
Linus Torvalds1da177e2005-04-16 15:20:36 -0700700 struct vm_area_struct *start_vma, unsigned long start_addr,
701 unsigned long end_addr, unsigned long *nr_accounted,
702 struct zap_details *);
Hugh Dickins3bf5ee92005-04-19 13:29:16 -0700703void free_pgd_range(struct mmu_gather **tlb, unsigned long addr,
704 unsigned long end, unsigned long floor, unsigned long ceiling);
705void free_pgtables(struct mmu_gather **tlb, struct vm_area_struct *start_vma,
Hugh Dickinse0da3822005-04-19 13:29:15 -0700706 unsigned long floor, unsigned long ceiling);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700707int copy_page_range(struct mm_struct *dst, struct mm_struct *src,
708 struct vm_area_struct *vma);
709int zeromap_page_range(struct vm_area_struct *vma, unsigned long from,
710 unsigned long size, pgprot_t prot);
711void unmap_mapping_range(struct address_space *mapping,
712 loff_t const holebegin, loff_t const holelen, int even_cows);
713
714static inline void unmap_shared_mapping_range(struct address_space *mapping,
715 loff_t const holebegin, loff_t const holelen)
716{
717 unmap_mapping_range(mapping, holebegin, holelen, 0);
718}
719
720extern int vmtruncate(struct inode * inode, loff_t offset);
Badari Pulavartyf6b3ec22006-01-06 00:10:38 -0800721extern int vmtruncate_range(struct inode * inode, loff_t offset, loff_t end);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700722extern int install_page(struct mm_struct *mm, struct vm_area_struct *vma, unsigned long addr, struct page *page, pgprot_t prot);
723extern int install_file_pte(struct mm_struct *mm, struct vm_area_struct *vma, unsigned long addr, unsigned long pgoff, pgprot_t prot);
Nick Pigginf33ea7f2005-08-03 20:24:01 +1000724
David Howells7ee1dd32006-01-06 00:11:44 -0800725#ifdef CONFIG_MMU
726extern int __handle_mm_fault(struct mm_struct *mm,struct vm_area_struct *vma,
727 unsigned long address, int write_access);
728
729static inline int handle_mm_fault(struct mm_struct *mm,
730 struct vm_area_struct *vma, unsigned long address,
731 int write_access)
Nick Pigginf33ea7f2005-08-03 20:24:01 +1000732{
David Howells7ee1dd32006-01-06 00:11:44 -0800733 return __handle_mm_fault(mm, vma, address, write_access) &
734 (~VM_FAULT_WRITE);
Nick Pigginf33ea7f2005-08-03 20:24:01 +1000735}
David Howells7ee1dd32006-01-06 00:11:44 -0800736#else
737static inline int handle_mm_fault(struct mm_struct *mm,
738 struct vm_area_struct *vma, unsigned long address,
739 int write_access)
740{
741 /* should never happen if there's no MMU */
742 BUG();
743 return VM_FAULT_SIGBUS;
744}
745#endif
Nick Pigginf33ea7f2005-08-03 20:24:01 +1000746
Linus Torvalds1da177e2005-04-16 15:20:36 -0700747extern int make_pages_present(unsigned long addr, unsigned long end);
748extern int access_process_vm(struct task_struct *tsk, unsigned long addr, void *buf, int len, int write);
749void install_arg_page(struct vm_area_struct *, struct page *, unsigned long);
750
751int get_user_pages(struct task_struct *tsk, struct mm_struct *mm, unsigned long start,
752 int len, int write, int force, struct page **pages, struct vm_area_struct **vmas);
Nick Pigginb5810032005-10-29 18:16:12 -0700753void print_bad_pte(struct vm_area_struct *, pte_t, unsigned long);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700754
755int __set_page_dirty_buffers(struct page *page);
756int __set_page_dirty_nobuffers(struct page *page);
757int redirty_page_for_writepage(struct writeback_control *wbc,
758 struct page *page);
759int FASTCALL(set_page_dirty(struct page *page));
760int set_page_dirty_lock(struct page *page);
761int clear_page_dirty_for_io(struct page *page);
762
763extern unsigned long do_mremap(unsigned long addr,
764 unsigned long old_len, unsigned long new_len,
765 unsigned long flags, unsigned long new_addr);
766
767/*
768 * Prototype to add a shrinker callback for ageable caches.
769 *
770 * These functions are passed a count `nr_to_scan' and a gfpmask. They should
771 * scan `nr_to_scan' objects, attempting to free them.
772 *
Domen Puncer845d3432005-05-05 16:16:14 -0700773 * The callback must return the number of objects which remain in the cache.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700774 *
Domen Puncer845d3432005-05-05 16:16:14 -0700775 * The callback will be passed nr_to_scan == 0 when the VM is querying the
Linus Torvalds1da177e2005-04-16 15:20:36 -0700776 * cache size, so a fastpath for that case is appropriate.
777 */
Al Viro6daa0e22005-10-21 03:18:50 -0400778typedef int (*shrinker_t)(int nr_to_scan, gfp_t gfp_mask);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700779
780/*
781 * Add an aging callback. The int is the number of 'seeks' it takes
782 * to recreate one of the objects that these functions age.
783 */
784
785#define DEFAULT_SEEKS 2
786struct shrinker;
787extern struct shrinker *set_shrinker(int, shrinker_t);
788extern void remove_shrinker(struct shrinker *shrinker);
789
Linus Torvaldsc9cfcdd2005-11-29 14:03:14 -0800790extern pte_t *FASTCALL(get_locked_pte(struct mm_struct *mm, unsigned long addr, spinlock_t **ptl));
791
Hugh Dickins1bb36302005-10-29 18:16:22 -0700792int __pud_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address);
793int __pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address);
794int __pte_alloc(struct mm_struct *mm, pmd_t *pmd, unsigned long address);
795int __pte_alloc_kernel(pmd_t *pmd, unsigned long address);
796
Linus Torvalds1da177e2005-04-16 15:20:36 -0700797/*
798 * The following ifdef needed to get the 4level-fixup.h header to work.
799 * Remove it when 4level-fixup.h has been removed.
800 */
Hugh Dickins1bb36302005-10-29 18:16:22 -0700801#if defined(CONFIG_MMU) && !defined(__ARCH_HAS_4LEVEL_HACK)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700802static inline pud_t *pud_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address)
803{
Hugh Dickins1bb36302005-10-29 18:16:22 -0700804 return (unlikely(pgd_none(*pgd)) && __pud_alloc(mm, pgd, address))?
805 NULL: pud_offset(pgd, address);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700806}
807
808static inline pmd_t *pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address)
809{
Hugh Dickins1bb36302005-10-29 18:16:22 -0700810 return (unlikely(pud_none(*pud)) && __pmd_alloc(mm, pud, address))?
811 NULL: pmd_offset(pud, address);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700812}
Hugh Dickins1bb36302005-10-29 18:16:22 -0700813#endif /* CONFIG_MMU && !__ARCH_HAS_4LEVEL_HACK */
814
Hugh Dickins4c21e2f2005-10-29 18:16:40 -0700815#if NR_CPUS >= CONFIG_SPLIT_PTLOCK_CPUS
816/*
817 * We tuck a spinlock to guard each pagetable page into its struct page,
818 * at page->private, with BUILD_BUG_ON to make sure that this will not
819 * overflow into the next struct page (as it might with DEBUG_SPINLOCK).
820 * When freeing, reset page->mapping so free_pages_check won't complain.
821 */
Andrew Morton349aef02006-01-08 01:04:36 -0800822#define __pte_lockptr(page) &((page)->ptl)
Hugh Dickins4c21e2f2005-10-29 18:16:40 -0700823#define pte_lock_init(_page) do { \
824 spin_lock_init(__pte_lockptr(_page)); \
825} while (0)
826#define pte_lock_deinit(page) ((page)->mapping = NULL)
827#define pte_lockptr(mm, pmd) ({(void)(mm); __pte_lockptr(pmd_page(*(pmd)));})
828#else
829/*
830 * We use mm->page_table_lock to guard all pagetable pages of the mm.
831 */
832#define pte_lock_init(page) do {} while (0)
833#define pte_lock_deinit(page) do {} while (0)
834#define pte_lockptr(mm, pmd) ({(void)(pmd); &(mm)->page_table_lock;})
835#endif /* NR_CPUS < CONFIG_SPLIT_PTLOCK_CPUS */
836
Hugh Dickinsc74df322005-10-29 18:16:23 -0700837#define pte_offset_map_lock(mm, pmd, address, ptlp) \
838({ \
Hugh Dickins4c21e2f2005-10-29 18:16:40 -0700839 spinlock_t *__ptl = pte_lockptr(mm, pmd); \
Hugh Dickinsc74df322005-10-29 18:16:23 -0700840 pte_t *__pte = pte_offset_map(pmd, address); \
841 *(ptlp) = __ptl; \
842 spin_lock(__ptl); \
843 __pte; \
844})
845
846#define pte_unmap_unlock(pte, ptl) do { \
847 spin_unlock(ptl); \
848 pte_unmap(pte); \
849} while (0)
850
Hugh Dickins1bb36302005-10-29 18:16:22 -0700851#define pte_alloc_map(mm, pmd, address) \
852 ((unlikely(!pmd_present(*(pmd))) && __pte_alloc(mm, pmd, address))? \
853 NULL: pte_offset_map(pmd, address))
854
Hugh Dickinsc74df322005-10-29 18:16:23 -0700855#define pte_alloc_map_lock(mm, pmd, address, ptlp) \
856 ((unlikely(!pmd_present(*(pmd))) && __pte_alloc(mm, pmd, address))? \
857 NULL: pte_offset_map_lock(mm, pmd, address, ptlp))
858
Hugh Dickins1bb36302005-10-29 18:16:22 -0700859#define pte_alloc_kernel(pmd, address) \
860 ((unlikely(!pmd_present(*(pmd))) && __pte_alloc_kernel(pmd, address))? \
861 NULL: pte_offset_kernel(pmd, address))
Linus Torvalds1da177e2005-04-16 15:20:36 -0700862
863extern void free_area_init(unsigned long * zones_size);
864extern void free_area_init_node(int nid, pg_data_t *pgdat,
865 unsigned long * zones_size, unsigned long zone_start_pfn,
866 unsigned long *zholes_size);
867extern void memmap_init_zone(unsigned long, int, unsigned long, unsigned long);
Dave Hansen3947be12005-10-29 18:16:54 -0700868extern void setup_per_zone_pages_min(void);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700869extern void mem_init(void);
870extern void show_mem(void);
871extern void si_meminfo(struct sysinfo * val);
872extern void si_meminfo_node(struct sysinfo *val, int nid);
873
Christoph Lametere7c8d5c2005-06-21 17:14:47 -0700874#ifdef CONFIG_NUMA
875extern void setup_per_cpu_pageset(void);
876#else
877static inline void setup_per_cpu_pageset(void) {}
878#endif
879
Linus Torvalds1da177e2005-04-16 15:20:36 -0700880/* prio_tree.c */
881void vma_prio_tree_add(struct vm_area_struct *, struct vm_area_struct *old);
882void vma_prio_tree_insert(struct vm_area_struct *, struct prio_tree_root *);
883void vma_prio_tree_remove(struct vm_area_struct *, struct prio_tree_root *);
884struct vm_area_struct *vma_prio_tree_next(struct vm_area_struct *vma,
885 struct prio_tree_iter *iter);
886
887#define vma_prio_tree_foreach(vma, iter, root, begin, end) \
888 for (prio_tree_iter_init(iter, root, begin, end), vma = NULL; \
889 (vma = vma_prio_tree_next(vma, iter)); )
890
891static inline void vma_nonlinear_insert(struct vm_area_struct *vma,
892 struct list_head *list)
893{
894 vma->shared.vm_set.parent = NULL;
895 list_add_tail(&vma->shared.vm_set.list, list);
896}
897
898/* mmap.c */
899extern int __vm_enough_memory(long pages, int cap_sys_admin);
900extern void vma_adjust(struct vm_area_struct *vma, unsigned long start,
901 unsigned long end, pgoff_t pgoff, struct vm_area_struct *insert);
902extern struct vm_area_struct *vma_merge(struct mm_struct *,
903 struct vm_area_struct *prev, unsigned long addr, unsigned long end,
904 unsigned long vm_flags, struct anon_vma *, struct file *, pgoff_t,
905 struct mempolicy *);
906extern struct anon_vma *find_mergeable_anon_vma(struct vm_area_struct *);
907extern int split_vma(struct mm_struct *,
908 struct vm_area_struct *, unsigned long addr, int new_below);
909extern int insert_vm_struct(struct mm_struct *, struct vm_area_struct *);
910extern void __vma_link_rb(struct mm_struct *, struct vm_area_struct *,
911 struct rb_node **, struct rb_node *);
Hugh Dickinsa8fb5612005-10-29 18:15:57 -0700912extern void unlink_file_vma(struct vm_area_struct *);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700913extern struct vm_area_struct *copy_vma(struct vm_area_struct **,
914 unsigned long addr, unsigned long len, pgoff_t pgoff);
915extern void exit_mmap(struct mm_struct *);
akpm@osdl.org119f6572005-05-01 08:58:35 -0700916extern int may_expand_vm(struct mm_struct *mm, unsigned long npages);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700917
918extern unsigned long get_unmapped_area(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
919
920extern unsigned long do_mmap_pgoff(struct file *file, unsigned long addr,
921 unsigned long len, unsigned long prot,
922 unsigned long flag, unsigned long pgoff);
923
924static inline unsigned long do_mmap(struct file *file, unsigned long addr,
925 unsigned long len, unsigned long prot,
926 unsigned long flag, unsigned long offset)
927{
928 unsigned long ret = -EINVAL;
929 if ((offset + PAGE_ALIGN(len)) < offset)
930 goto out;
931 if (!(offset & ~PAGE_MASK))
932 ret = do_mmap_pgoff(file, addr, len, prot, flag, offset >> PAGE_SHIFT);
933out:
934 return ret;
935}
936
937extern int do_munmap(struct mm_struct *, unsigned long, size_t);
938
939extern unsigned long do_brk(unsigned long, unsigned long);
940
941/* filemap.c */
942extern unsigned long page_unuse(struct page *);
943extern void truncate_inode_pages(struct address_space *, loff_t);
Hans Reiserd7339072006-01-06 00:10:36 -0800944extern void truncate_inode_pages_range(struct address_space *,
945 loff_t lstart, loff_t lend);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700946
947/* generic vm_area_ops exported for stackable file systems */
948extern struct page *filemap_nopage(struct vm_area_struct *, unsigned long, int *);
949extern int filemap_populate(struct vm_area_struct *, unsigned long,
950 unsigned long, pgprot_t, unsigned long, int);
951
952/* mm/page-writeback.c */
953int write_one_page(struct page *page, int wait);
954
955/* readahead.c */
956#define VM_MAX_READAHEAD 128 /* kbytes */
957#define VM_MIN_READAHEAD 16 /* kbytes (includes current page) */
958#define VM_MAX_CACHE_HIT 256 /* max pages in a row in cache before
959 * turning readahead off */
960
961int do_page_cache_readahead(struct address_space *mapping, struct file *filp,
Andrew Morton7361f4d2005-11-07 00:59:28 -0800962 pgoff_t offset, unsigned long nr_to_read);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700963int force_page_cache_readahead(struct address_space *mapping, struct file *filp,
Andrew Morton7361f4d2005-11-07 00:59:28 -0800964 pgoff_t offset, unsigned long nr_to_read);
965unsigned long page_cache_readahead(struct address_space *mapping,
Linus Torvalds1da177e2005-04-16 15:20:36 -0700966 struct file_ra_state *ra,
967 struct file *filp,
Andrew Morton7361f4d2005-11-07 00:59:28 -0800968 pgoff_t offset,
Linus Torvalds1da177e2005-04-16 15:20:36 -0700969 unsigned long size);
970void handle_ra_miss(struct address_space *mapping,
971 struct file_ra_state *ra, pgoff_t offset);
972unsigned long max_sane_readahead(unsigned long nr);
973
974/* Do stack extension */
Hugh Dickins46dea3d2005-10-29 18:16:20 -0700975extern int expand_stack(struct vm_area_struct *vma, unsigned long address);
Matthew Wilcox9ab88512005-11-18 16:16:42 -0500976#ifdef CONFIG_IA64
Hugh Dickins46dea3d2005-10-29 18:16:20 -0700977extern int expand_upwards(struct vm_area_struct *vma, unsigned long address);
Matthew Wilcox9ab88512005-11-18 16:16:42 -0500978#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -0700979
980/* Look up the first VMA which satisfies addr < vm_end, NULL if none. */
981extern struct vm_area_struct * find_vma(struct mm_struct * mm, unsigned long addr);
982extern struct vm_area_struct * find_vma_prev(struct mm_struct * mm, unsigned long addr,
983 struct vm_area_struct **pprev);
984
985/* Look up the first VMA which intersects the interval start_addr..end_addr-1,
986 NULL if none. Assume start_addr < end_addr. */
987static inline struct vm_area_struct * find_vma_intersection(struct mm_struct * mm, unsigned long start_addr, unsigned long end_addr)
988{
989 struct vm_area_struct * vma = find_vma(mm,start_addr);
990
991 if (vma && end_addr <= vma->vm_start)
992 vma = NULL;
993 return vma;
994}
995
996static inline unsigned long vma_pages(struct vm_area_struct *vma)
997{
998 return (vma->vm_end - vma->vm_start) >> PAGE_SHIFT;
999}
1000
Hugh Dickinsdeceb6c2005-10-29 18:16:33 -07001001struct vm_area_struct *find_extend_vma(struct mm_struct *, unsigned long addr);
1002struct page *vmalloc_to_page(void *addr);
1003unsigned long vmalloc_to_pfn(void *addr);
1004int remap_pfn_range(struct vm_area_struct *, unsigned long addr,
1005 unsigned long pfn, unsigned long size, pgprot_t);
Linus Torvaldsa145dd42005-11-30 09:35:19 -08001006int vm_insert_page(struct vm_area_struct *, unsigned long addr, struct page *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001007
Linus Torvalds6aab3412005-11-28 14:34:23 -08001008struct page *follow_page(struct vm_area_struct *, unsigned long address,
Hugh Dickinsdeceb6c2005-10-29 18:16:33 -07001009 unsigned int foll_flags);
1010#define FOLL_WRITE 0x01 /* check pte is writable */
1011#define FOLL_TOUCH 0x02 /* mark page accessed */
1012#define FOLL_GET 0x04 /* do get_page on page */
1013#define FOLL_ANON 0x08 /* give ZERO_PAGE if no pgtable */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001014
1015#ifdef CONFIG_PROC_FS
Hugh Dickinsab50b8e2005-10-29 18:15:56 -07001016void vm_stat_account(struct mm_struct *, unsigned long, struct file *, long);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001017#else
Hugh Dickinsab50b8e2005-10-29 18:15:56 -07001018static inline void vm_stat_account(struct mm_struct *mm,
Linus Torvalds1da177e2005-04-16 15:20:36 -07001019 unsigned long flags, struct file *file, long pages)
1020{
1021}
1022#endif /* CONFIG_PROC_FS */
1023
Linus Torvalds1da177e2005-04-16 15:20:36 -07001024#ifndef CONFIG_DEBUG_PAGEALLOC
1025static inline void
1026kernel_map_pages(struct page *page, int numpages, int enable)
1027{
Ingo Molnarde5097c2006-01-09 15:59:21 -08001028 if (!PageHighMem(page) && !enable)
1029 mutex_debug_check_no_locks_freed(page_address(page),
1030 page_address(page + numpages));
Linus Torvalds1da177e2005-04-16 15:20:36 -07001031}
1032#endif
1033
1034extern struct vm_area_struct *get_gate_vma(struct task_struct *tsk);
1035#ifdef __HAVE_ARCH_GATE_AREA
1036int in_gate_area_no_task(unsigned long addr);
1037int in_gate_area(struct task_struct *task, unsigned long addr);
1038#else
1039int in_gate_area_no_task(unsigned long addr);
1040#define in_gate_area(task, addr) ({(void)task; in_gate_area_no_task(addr);})
1041#endif /* __HAVE_ARCH_GATE_AREA */
1042
Andrea Arcangeli79befd02005-04-16 15:24:05 -07001043/* /proc/<pid>/oom_adj set to -17 protects from the oom-killer */
1044#define OOM_DISABLE -17
1045
Andrew Morton9d0243b2006-01-08 01:00:39 -08001046int drop_caches_sysctl_handler(struct ctl_table *, int, struct file *,
1047 void __user *, size_t *, loff_t *);
1048int shrink_slab(unsigned long scanned, gfp_t gfp_mask,
1049 unsigned long lru_pages);
1050void drop_pagecache(void);
1051void drop_slab(void);
1052
Linus Torvalds1da177e2005-04-16 15:20:36 -07001053#endif /* __KERNEL__ */
1054#endif /* _LINUX_MM_H */