blob: ffee2f74341856275ead62f71c6093c8de457199 [file] [log] [blame]
Linus Torvalds1da177e2005-04-16 15:20:36 -07001#ifndef _LINUX_MM_H
2#define _LINUX_MM_H
3
Linus Torvalds1da177e2005-04-16 15:20:36 -07004#include <linux/errno.h>
5
6#ifdef __KERNEL__
7
Linus Torvalds1da177e2005-04-16 15:20:36 -07008#include <linux/gfp.h>
9#include <linux/list.h>
Jiri Slaby59ea7462008-06-12 13:56:40 +020010#include <linux/mmdebug.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070011#include <linux/mmzone.h>
12#include <linux/rbtree.h>
13#include <linux/prio_tree.h>
Ingo Molnar9a11b49a2006-07-03 00:24:33 -070014#include <linux/debug_locks.h>
Heiko Carstens5b99cd02006-09-27 01:50:01 -070015#include <linux/mm_types.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070016
17struct mempolicy;
18struct anon_vma;
Alexey Dobriyan4e950f62007-07-30 02:36:13 +040019struct file_ra_state;
Alexey Dobriyane8edc6e2007-05-21 01:22:52 +040020struct user_struct;
Alexey Dobriyan4e950f62007-07-30 02:36:13 +040021struct writeback_control;
Linus Torvalds1da177e2005-04-16 15:20:36 -070022
23#ifndef CONFIG_DISCONTIGMEM /* Don't use mapnrs, do it properly */
24extern unsigned long max_mapnr;
25#endif
26
27extern unsigned long num_physpages;
28extern void * high_memory;
Linus Torvalds1da177e2005-04-16 15:20:36 -070029extern int page_cluster;
30
31#ifdef CONFIG_SYSCTL
32extern int sysctl_legacy_va_layout;
33#else
34#define sysctl_legacy_va_layout 0
35#endif
36
James Morris42d78962007-12-26 15:12:37 +110037extern unsigned long mmap_min_addr;
38
Linus Torvalds1da177e2005-04-16 15:20:36 -070039#include <asm/page.h>
40#include <asm/pgtable.h>
41#include <asm/processor.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070042
Linus Torvalds1da177e2005-04-16 15:20:36 -070043#define nth_page(page,n) pfn_to_page(page_to_pfn((page)) + (n))
44
Andrea Righi27ac7922008-07-23 21:28:13 -070045/* to align the pointer to the (next) page boundary */
46#define PAGE_ALIGN(addr) ALIGN(addr, PAGE_SIZE)
47
Linus Torvalds1da177e2005-04-16 15:20:36 -070048/*
49 * Linux kernel virtual memory manager primitives.
50 * The idea being to have a "virtual" mm in the same way
51 * we have a virtual fs - giving a cleaner interface to the
52 * mm details, and allowing different kinds of memory mappings
53 * (from shared memory to executable loading to arbitrary
54 * mmap() functions).
55 */
56
Christoph Lameterc43692e2006-12-06 20:32:48 -080057extern struct kmem_cache *vm_area_cachep;
58
Linus Torvalds1da177e2005-04-16 15:20:36 -070059/*
60 * This struct defines the per-mm list of VMAs for uClinux. If CONFIG_MMU is
61 * disabled, then there's a single shared list of VMAs maintained by the
62 * system, and mm's subscribe to these individually
63 */
64struct vm_list_struct {
65 struct vm_list_struct *next;
66 struct vm_area_struct *vma;
67};
68
69#ifndef CONFIG_MMU
70extern struct rb_root nommu_vma_tree;
71extern struct rw_semaphore nommu_vma_sem;
72
73extern unsigned int kobjsize(const void *objp);
74#endif
75
76/*
Hugh Dickins605d9282008-08-16 11:07:21 +010077 * vm_flags in vm_area_struct, see mm_types.h.
Linus Torvalds1da177e2005-04-16 15:20:36 -070078 */
79#define VM_READ 0x00000001 /* currently active flags */
80#define VM_WRITE 0x00000002
81#define VM_EXEC 0x00000004
82#define VM_SHARED 0x00000008
83
Paolo 'Blaisorblade' Giarrusso7e2cff42005-09-21 09:55:39 -070084/* mprotect() hardcodes VM_MAYREAD >> 4 == VM_READ, and so for r/w/x bits. */
Linus Torvalds1da177e2005-04-16 15:20:36 -070085#define VM_MAYREAD 0x00000010 /* limits for mprotect() etc */
86#define VM_MAYWRITE 0x00000020
87#define VM_MAYEXEC 0x00000040
88#define VM_MAYSHARE 0x00000080
89
90#define VM_GROWSDOWN 0x00000100 /* general info on the segment */
91#define VM_GROWSUP 0x00000200
Linus Torvalds6aab3412005-11-28 14:34:23 -080092#define VM_PFNMAP 0x00000400 /* Page-ranges managed without "struct page", just pure PFN */
Linus Torvalds1da177e2005-04-16 15:20:36 -070093#define VM_DENYWRITE 0x00000800 /* ETXTBSY on write attempts.. */
94
95#define VM_EXECUTABLE 0x00001000
96#define VM_LOCKED 0x00002000
97#define VM_IO 0x00004000 /* Memory mapped I/O or similar */
98
99 /* Used by sys_madvise() */
100#define VM_SEQ_READ 0x00008000 /* App will access data sequentially */
101#define VM_RAND_READ 0x00010000 /* App will not benefit from clustered reads */
102
103#define VM_DONTCOPY 0x00020000 /* Do not copy this vma on fork */
104#define VM_DONTEXPAND 0x00040000 /* Cannot expand with mremap() */
Hugh Dickins0b14c172005-11-21 21:32:15 -0800105#define VM_RESERVED 0x00080000 /* Count as reserved_vm like IO */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700106#define VM_ACCOUNT 0x00100000 /* Is a VM accounted object */
Andy Whitcroftcdfd4322008-07-23 21:27:28 -0700107#define VM_NORESERVE 0x00200000 /* should the VM suppress accounting */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700108#define VM_HUGETLB 0x00400000 /* Huge TLB Page VM */
109#define VM_NONLINEAR 0x00800000 /* Is non-linear (remap_file_pages) */
110#define VM_MAPPED_COPY 0x01000000 /* T if mapped copy of data (nommu mmap) */
Linus Torvalds4d7672b2005-12-16 10:21:23 -0800111#define VM_INSERTPAGE 0x02000000 /* The vma has had "vm_insert_page()" done on it */
Roland McGrathe5b97dd2007-01-26 00:56:48 -0800112#define VM_ALWAYSDUMP 0x04000000 /* Always include in core dumps */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700113
Nick Piggind0217ac2007-07-19 01:47:03 -0700114#define VM_CAN_NONLINEAR 0x08000000 /* Has ->fault & does nonlinear pages */
Jared Hulbertb379d792008-04-28 02:12:58 -0700115#define VM_MIXEDMAP 0x10000000 /* Can contain "struct page" and pure PFN pages */
Dave Kleikampaba46c52008-07-08 00:28:52 +1000116#define VM_SAO 0x20000000 /* Strong Access Ordering (powerpc) */
Nick Piggind00806b2007-07-19 01:46:57 -0700117
Linus Torvalds1da177e2005-04-16 15:20:36 -0700118#ifndef VM_STACK_DEFAULT_FLAGS /* arch can override this */
119#define VM_STACK_DEFAULT_FLAGS VM_DATA_DEFAULT_FLAGS
120#endif
121
122#ifdef CONFIG_STACK_GROWSUP
123#define VM_STACK_FLAGS (VM_GROWSUP | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT)
124#else
125#define VM_STACK_FLAGS (VM_GROWSDOWN | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT)
126#endif
127
128#define VM_READHINTMASK (VM_SEQ_READ | VM_RAND_READ)
129#define VM_ClearReadHint(v) (v)->vm_flags &= ~VM_READHINTMASK
130#define VM_NormalReadHint(v) (!((v)->vm_flags & VM_READHINTMASK))
131#define VM_SequentialReadHint(v) ((v)->vm_flags & VM_SEQ_READ)
132#define VM_RandomReadHint(v) ((v)->vm_flags & VM_RAND_READ)
133
134/*
Nick Pigginb291f002008-10-18 20:26:44 -0700135 * special vmas that are non-mergable, non-mlock()able
136 */
137#define VM_SPECIAL (VM_IO | VM_DONTEXPAND | VM_RESERVED | VM_PFNMAP)
138
139/*
Linus Torvalds1da177e2005-04-16 15:20:36 -0700140 * mapping from the currently active vm_flags protection bits (the
141 * low four bits) to a page protection mask..
142 */
143extern pgprot_t protection_map[16];
144
Nick Piggind0217ac2007-07-19 01:47:03 -0700145#define FAULT_FLAG_WRITE 0x01 /* Fault was a write access */
146#define FAULT_FLAG_NONLINEAR 0x02 /* Fault was via a nonlinear mapping */
147
148
Nick Piggin54cb8822007-07-19 01:46:59 -0700149/*
Nick Piggind0217ac2007-07-19 01:47:03 -0700150 * vm_fault is filled by the the pagefault handler and passed to the vma's
Nick Piggin83c54072007-07-19 01:47:05 -0700151 * ->fault function. The vma's ->fault is responsible for returning a bitmask
152 * of VM_FAULT_xxx flags that give details about how the fault was handled.
Nick Piggin54cb8822007-07-19 01:46:59 -0700153 *
Nick Piggind0217ac2007-07-19 01:47:03 -0700154 * pgoff should be used in favour of virtual_address, if possible. If pgoff
155 * is used, one may set VM_CAN_NONLINEAR in the vma->vm_flags to get nonlinear
156 * mapping support.
Nick Piggin54cb8822007-07-19 01:46:59 -0700157 */
Nick Piggind0217ac2007-07-19 01:47:03 -0700158struct vm_fault {
159 unsigned int flags; /* FAULT_FLAG_xxx flags */
160 pgoff_t pgoff; /* Logical page offset based on vma */
161 void __user *virtual_address; /* Faulting virtual address */
162
163 struct page *page; /* ->fault handlers should return a
Nick Piggin83c54072007-07-19 01:47:05 -0700164 * page here, unless VM_FAULT_NOPAGE
Nick Piggind0217ac2007-07-19 01:47:03 -0700165 * is set (which is also implied by
Nick Piggin83c54072007-07-19 01:47:05 -0700166 * VM_FAULT_ERROR).
Nick Piggind0217ac2007-07-19 01:47:03 -0700167 */
Nick Piggin54cb8822007-07-19 01:46:59 -0700168};
Linus Torvalds1da177e2005-04-16 15:20:36 -0700169
170/*
171 * These are the virtual MM functions - opening of an area, closing and
172 * unmapping it (needed to keep files on disk up-to-date etc), pointer
173 * to the functions called when a no-page or a wp-page exception occurs.
174 */
175struct vm_operations_struct {
176 void (*open)(struct vm_area_struct * area);
177 void (*close)(struct vm_area_struct * area);
Nick Piggind0217ac2007-07-19 01:47:03 -0700178 int (*fault)(struct vm_area_struct *vma, struct vm_fault *vmf);
David Howells9637a5e2006-06-23 02:03:43 -0700179
180 /* notification that a previously read-only page is about to become
181 * writable, if an error is returned it will cause a SIGBUS */
182 int (*page_mkwrite)(struct vm_area_struct *vma, struct page *page);
Rik van Riel28b2ee22008-07-23 21:27:05 -0700183
184 /* called by access_process_vm when get_user_pages() fails, typically
185 * for use by special VMAs that can switch between memory and hardware
186 */
187 int (*access)(struct vm_area_struct *vma, unsigned long addr,
188 void *buf, int len, int write);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700189#ifdef CONFIG_NUMA
Lee Schermerhorna6020ed2008-04-28 02:13:14 -0700190 /*
191 * set_policy() op must add a reference to any non-NULL @new mempolicy
192 * to hold the policy upon return. Caller should pass NULL @new to
193 * remove a policy and fall back to surrounding context--i.e. do not
194 * install a MPOL_DEFAULT policy, nor the task or system default
195 * mempolicy.
196 */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700197 int (*set_policy)(struct vm_area_struct *vma, struct mempolicy *new);
Lee Schermerhorna6020ed2008-04-28 02:13:14 -0700198
199 /*
200 * get_policy() op must add reference [mpol_get()] to any policy at
201 * (vma,addr) marked as MPOL_SHARED. The shared policy infrastructure
202 * in mm/mempolicy.c will do this automatically.
203 * get_policy() must NOT add a ref if the policy at (vma,addr) is not
204 * marked as MPOL_SHARED. vma policies are protected by the mmap_sem.
205 * If no [shared/vma] mempolicy exists at the addr, get_policy() op
206 * must return NULL--i.e., do not "fallback" to task or system default
207 * policy.
208 */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700209 struct mempolicy *(*get_policy)(struct vm_area_struct *vma,
210 unsigned long addr);
Christoph Lameter7b2259b2006-06-25 05:46:48 -0700211 int (*migrate)(struct vm_area_struct *vma, const nodemask_t *from,
212 const nodemask_t *to, unsigned long flags);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700213#endif
214};
215
216struct mmu_gather;
217struct inode;
218
Andrew Morton349aef02006-01-08 01:04:36 -0800219#define page_private(page) ((page)->private)
220#define set_page_private(page, v) ((page)->private = (v))
Hugh Dickins4c21e2f2005-10-29 18:16:40 -0700221
Linus Torvalds1da177e2005-04-16 15:20:36 -0700222/*
223 * FIXME: take this include out, include page-flags.h in
224 * files which need it (119 of them)
225 */
226#include <linux/page-flags.h>
227
228/*
229 * Methods to modify the page usage count.
230 *
231 * What counts for a page usage:
232 * - cache mapping (page->mapping)
233 * - private data (page->private)
234 * - page mapped in a task's page tables, each mapping
235 * is counted separately
236 *
237 * Also, many kernel routines increase the page count before a critical
238 * routine so they can be sure the page doesn't go away from under them.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700239 */
240
241/*
Nick Pigginda6052f2006-09-25 23:31:35 -0700242 * Drop a ref, return true if the refcount fell to zero (the page has no users)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700243 */
Nick Piggin7c8ee9a2006-03-22 00:08:03 -0800244static inline int put_page_testzero(struct page *page)
245{
Nick Piggin725d7042006-09-25 23:30:55 -0700246 VM_BUG_ON(atomic_read(&page->_count) == 0);
Nick Piggin8dc04ef2006-03-22 00:08:03 -0800247 return atomic_dec_and_test(&page->_count);
Nick Piggin7c8ee9a2006-03-22 00:08:03 -0800248}
Linus Torvalds1da177e2005-04-16 15:20:36 -0700249
250/*
Nick Piggin7c8ee9a2006-03-22 00:08:03 -0800251 * Try to grab a ref unless the page has a refcount of zero, return false if
252 * that is the case.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700253 */
Nick Piggin7c8ee9a2006-03-22 00:08:03 -0800254static inline int get_page_unless_zero(struct page *page)
255{
Christoph Lameteraec2c3e2008-02-04 22:28:35 -0800256 VM_BUG_ON(PageTail(page));
Nick Piggin8dc04ef2006-03-22 00:08:03 -0800257 return atomic_inc_not_zero(&page->_count);
Nick Piggin7c8ee9a2006-03-22 00:08:03 -0800258}
Linus Torvalds1da177e2005-04-16 15:20:36 -0700259
Christoph Lameter48667e72008-02-04 22:28:31 -0800260/* Support for virtually mapped pages */
Christoph Lameterb3bdda02008-02-04 22:28:32 -0800261struct page *vmalloc_to_page(const void *addr);
262unsigned long vmalloc_to_pfn(const void *addr);
Christoph Lameter48667e72008-02-04 22:28:31 -0800263
Paul Mundt0738c4b2008-03-12 16:51:31 +0900264/*
265 * Determine if an address is within the vmalloc range
266 *
267 * On nommu, vmalloc/vfree wrap through kmalloc/kfree directly, so there
268 * is no special casing required.
269 */
Christoph Lameter9e2779f2008-02-04 22:28:34 -0800270static inline int is_vmalloc_addr(const void *x)
271{
Paul Mundt0738c4b2008-03-12 16:51:31 +0900272#ifdef CONFIG_MMU
Christoph Lameter9e2779f2008-02-04 22:28:34 -0800273 unsigned long addr = (unsigned long)x;
274
275 return addr >= VMALLOC_START && addr < VMALLOC_END;
Paul Mundt0738c4b2008-03-12 16:51:31 +0900276#else
277 return 0;
David Howells8ca3ed82008-02-23 15:23:37 -0800278#endif
Paul Mundt0738c4b2008-03-12 16:51:31 +0900279}
Christoph Lameter9e2779f2008-02-04 22:28:34 -0800280
Christoph Lameterd85f3382007-05-06 14:49:39 -0700281static inline struct page *compound_head(struct page *page)
282{
Christoph Lameter6d777952007-05-06 14:49:40 -0700283 if (unlikely(PageTail(page)))
Christoph Lameterd85f3382007-05-06 14:49:39 -0700284 return page->first_page;
285 return page;
286}
287
Hugh Dickins4c21e2f2005-10-29 18:16:40 -0700288static inline int page_count(struct page *page)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700289{
Christoph Lameterd85f3382007-05-06 14:49:39 -0700290 return atomic_read(&compound_head(page)->_count);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700291}
292
293static inline void get_page(struct page *page)
294{
Christoph Lameterd85f3382007-05-06 14:49:39 -0700295 page = compound_head(page);
Nick Piggin725d7042006-09-25 23:30:55 -0700296 VM_BUG_ON(atomic_read(&page->_count) == 0);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700297 atomic_inc(&page->_count);
298}
299
Christoph Lameterb49af682007-05-06 14:49:41 -0700300static inline struct page *virt_to_head_page(const void *x)
301{
302 struct page *page = virt_to_page(x);
303 return compound_head(page);
304}
305
Nick Piggin7835e982006-03-22 00:08:40 -0800306/*
307 * Setup the page count before being freed into the page allocator for
308 * the first time (boot or memory hotplug)
309 */
310static inline void init_page_count(struct page *page)
311{
312 atomic_set(&page->_count, 1);
313}
314
Linus Torvalds1da177e2005-04-16 15:20:36 -0700315void put_page(struct page *page);
Alexander Zarochentsev1d7ea732006-08-13 23:24:27 -0700316void put_pages_list(struct list_head *pages);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700317
Nick Piggin8dfcc9b2006-03-22 00:08:05 -0800318void split_page(struct page *page, unsigned int order);
Nick Piggin8dfcc9b2006-03-22 00:08:05 -0800319
Linus Torvalds1da177e2005-04-16 15:20:36 -0700320/*
Andy Whitcroft33f2ef82006-12-06 20:33:32 -0800321 * Compound pages have a destructor function. Provide a
322 * prototype for that function and accessor functions.
323 * These are _only_ valid on the head of a PG_compound page.
324 */
325typedef void compound_page_dtor(struct page *);
326
327static inline void set_compound_page_dtor(struct page *page,
328 compound_page_dtor *dtor)
329{
330 page[1].lru.next = (void *)dtor;
331}
332
333static inline compound_page_dtor *get_compound_page_dtor(struct page *page)
334{
335 return (compound_page_dtor *)page[1].lru.next;
336}
337
Christoph Lameterd85f3382007-05-06 14:49:39 -0700338static inline int compound_order(struct page *page)
339{
Christoph Lameter6d777952007-05-06 14:49:40 -0700340 if (!PageHead(page))
Christoph Lameterd85f3382007-05-06 14:49:39 -0700341 return 0;
342 return (unsigned long)page[1].lru.prev;
343}
344
345static inline void set_compound_order(struct page *page, unsigned long order)
346{
347 page[1].lru.prev = (void *)order;
348}
349
Andy Whitcroft33f2ef82006-12-06 20:33:32 -0800350/*
Linus Torvalds1da177e2005-04-16 15:20:36 -0700351 * Multiple processes may "see" the same page. E.g. for untouched
352 * mappings of /dev/null, all processes see the same page full of
353 * zeroes, and text pages of executables and shared libraries have
354 * only one copy in memory, at most, normally.
355 *
356 * For the non-reserved pages, page_count(page) denotes a reference count.
Paolo 'Blaisorblade' Giarrusso7e871b62005-09-21 09:55:38 -0700357 * page_count() == 0 means the page is free. page->lru is then used for
358 * freelist management in the buddy allocator.
Nick Pigginda6052f2006-09-25 23:31:35 -0700359 * page_count() > 0 means the page has been allocated.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700360 *
Nick Pigginda6052f2006-09-25 23:31:35 -0700361 * Pages are allocated by the slab allocator in order to provide memory
362 * to kmalloc and kmem_cache_alloc. In this case, the management of the
363 * page, and the fields in 'struct page' are the responsibility of mm/slab.c
364 * unless a particular usage is carefully commented. (the responsibility of
365 * freeing the kmalloc memory is the caller's, of course).
Linus Torvalds1da177e2005-04-16 15:20:36 -0700366 *
Nick Pigginda6052f2006-09-25 23:31:35 -0700367 * A page may be used by anyone else who does a __get_free_page().
368 * In this case, page_count still tracks the references, and should only
369 * be used through the normal accessor functions. The top bits of page->flags
370 * and page->virtual store page management information, but all other fields
371 * are unused and could be used privately, carefully. The management of this
372 * page is the responsibility of the one who allocated it, and those who have
373 * subsequently been given references to it.
374 *
375 * The other pages (we may call them "pagecache pages") are completely
Linus Torvalds1da177e2005-04-16 15:20:36 -0700376 * managed by the Linux memory manager: I/O, buffers, swapping etc.
377 * The following discussion applies only to them.
378 *
Nick Pigginda6052f2006-09-25 23:31:35 -0700379 * A pagecache page contains an opaque `private' member, which belongs to the
380 * page's address_space. Usually, this is the address of a circular list of
381 * the page's disk buffers. PG_private must be set to tell the VM to call
382 * into the filesystem to release these pages.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700383 *
Nick Pigginda6052f2006-09-25 23:31:35 -0700384 * A page may belong to an inode's memory mapping. In this case, page->mapping
385 * is the pointer to the inode, and page->index is the file offset of the page,
386 * in units of PAGE_CACHE_SIZE.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700387 *
Nick Pigginda6052f2006-09-25 23:31:35 -0700388 * If pagecache pages are not associated with an inode, they are said to be
389 * anonymous pages. These may become associated with the swapcache, and in that
390 * case PG_swapcache is set, and page->private is an offset into the swapcache.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700391 *
Nick Pigginda6052f2006-09-25 23:31:35 -0700392 * In either case (swapcache or inode backed), the pagecache itself holds one
393 * reference to the page. Setting PG_private should also increment the
394 * refcount. The each user mapping also has a reference to the page.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700395 *
Nick Pigginda6052f2006-09-25 23:31:35 -0700396 * The pagecache pages are stored in a per-mapping radix tree, which is
397 * rooted at mapping->page_tree, and indexed by offset.
398 * Where 2.4 and early 2.6 kernels kept dirty/clean pages in per-address_space
399 * lists, we instead now tag pages as dirty/writeback in the radix tree.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700400 *
Nick Pigginda6052f2006-09-25 23:31:35 -0700401 * All pagecache pages may be subject to I/O:
Linus Torvalds1da177e2005-04-16 15:20:36 -0700402 * - inode pages may need to be read from disk,
403 * - inode pages which have been modified and are MAP_SHARED may need
Nick Pigginda6052f2006-09-25 23:31:35 -0700404 * to be written back to the inode on disk,
405 * - anonymous pages (including MAP_PRIVATE file mappings) which have been
406 * modified may need to be swapped out to swap space and (later) to be read
407 * back into memory.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700408 */
409
410/*
411 * The zone field is never updated after free_area_init_core()
412 * sets it, so none of the operations on it need to be atomic.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700413 */
Dave Hansen348f8b62005-06-23 00:07:40 -0700414
Andy Whitcroftd41dee32005-06-23 00:07:54 -0700415
416/*
417 * page->flags layout:
418 *
419 * There are three possibilities for how page->flags get
420 * laid out. The first is for the normal case, without
421 * sparsemem. The second is for sparsemem when there is
422 * plenty of space for node and section. The last is when
423 * we have run out of space and have to fall back to an
424 * alternate (slower) way of determining the node.
425 *
Christoph Lameter308c05e2008-04-28 02:12:43 -0700426 * No sparsemem or sparsemem vmemmap: | NODE | ZONE | ... | FLAGS |
427 * classic sparse with space for node:| SECTION | NODE | ZONE | ... | FLAGS |
428 * classic sparse no space for node: | SECTION | ZONE | ... | FLAGS |
Andy Whitcroftd41dee32005-06-23 00:07:54 -0700429 */
Christoph Lameter308c05e2008-04-28 02:12:43 -0700430#if defined(CONFIG_SPARSEMEM) && !defined(CONFIG_SPARSEMEM_VMEMMAP)
Andy Whitcroftd41dee32005-06-23 00:07:54 -0700431#define SECTIONS_WIDTH SECTIONS_SHIFT
432#else
433#define SECTIONS_WIDTH 0
434#endif
435
436#define ZONES_WIDTH ZONES_SHIFT
437
Christoph Lameter9223b412008-04-28 02:12:48 -0700438#if SECTIONS_WIDTH+ZONES_WIDTH+NODES_SHIFT <= BITS_PER_LONG - NR_PAGEFLAGS
Andy Whitcroftd41dee32005-06-23 00:07:54 -0700439#define NODES_WIDTH NODES_SHIFT
440#else
Christoph Lameter308c05e2008-04-28 02:12:43 -0700441#ifdef CONFIG_SPARSEMEM_VMEMMAP
442#error "Vmemmap: No space for nodes field in page flags"
443#endif
Andy Whitcroftd41dee32005-06-23 00:07:54 -0700444#define NODES_WIDTH 0
445#endif
446
447/* Page flags: | [SECTION] | [NODE] | ZONE | ... | FLAGS | */
Andi Kleen07808b72005-11-05 17:25:53 +0100448#define SECTIONS_PGOFF ((sizeof(unsigned long)*8) - SECTIONS_WIDTH)
Andy Whitcroftd41dee32005-06-23 00:07:54 -0700449#define NODES_PGOFF (SECTIONS_PGOFF - NODES_WIDTH)
450#define ZONES_PGOFF (NODES_PGOFF - ZONES_WIDTH)
451
452/*
453 * We are going to use the flags for the page to node mapping if its in
454 * there. This includes the case where there is no node, so it is implicit.
455 */
Christoph Lameter89689ae2006-12-06 20:31:45 -0800456#if !(NODES_WIDTH > 0 || NODES_SHIFT == 0)
457#define NODE_NOT_IN_PAGE_FLAGS
458#endif
Andy Whitcroftd41dee32005-06-23 00:07:54 -0700459
460#ifndef PFN_SECTION_SHIFT
461#define PFN_SECTION_SHIFT 0
462#endif
Dave Hansen348f8b62005-06-23 00:07:40 -0700463
464/*
465 * Define the bit shifts to access each section. For non-existant
466 * sections we define the shift as 0; that plus a 0 mask ensures
467 * the compiler will optimise away reference to them.
468 */
Andy Whitcroftd41dee32005-06-23 00:07:54 -0700469#define SECTIONS_PGSHIFT (SECTIONS_PGOFF * (SECTIONS_WIDTH != 0))
470#define NODES_PGSHIFT (NODES_PGOFF * (NODES_WIDTH != 0))
471#define ZONES_PGSHIFT (ZONES_PGOFF * (ZONES_WIDTH != 0))
Dave Hansen348f8b62005-06-23 00:07:40 -0700472
Christoph Lameter89689ae2006-12-06 20:31:45 -0800473/* NODE:ZONE or SECTION:ZONE is used to ID a zone for the buddy allcator */
474#ifdef NODE_NOT_IN_PAGEFLAGS
475#define ZONEID_SHIFT (SECTIONS_SHIFT + ZONES_SHIFT)
Andy Whitcroftbd8029b2007-02-10 01:43:14 -0800476#define ZONEID_PGOFF ((SECTIONS_PGOFF < ZONES_PGOFF)? \
477 SECTIONS_PGOFF : ZONES_PGOFF)
Andy Whitcroftd41dee32005-06-23 00:07:54 -0700478#else
Christoph Lameter89689ae2006-12-06 20:31:45 -0800479#define ZONEID_SHIFT (NODES_SHIFT + ZONES_SHIFT)
Andy Whitcroftbd8029b2007-02-10 01:43:14 -0800480#define ZONEID_PGOFF ((NODES_PGOFF < ZONES_PGOFF)? \
481 NODES_PGOFF : ZONES_PGOFF)
Andy Whitcroftd41dee32005-06-23 00:07:54 -0700482#endif
Christoph Lameter89689ae2006-12-06 20:31:45 -0800483
Andy Whitcroftbd8029b2007-02-10 01:43:14 -0800484#define ZONEID_PGSHIFT (ZONEID_PGOFF * (ZONEID_SHIFT != 0))
Dave Hansen348f8b62005-06-23 00:07:40 -0700485
Christoph Lameter9223b412008-04-28 02:12:48 -0700486#if SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > BITS_PER_LONG - NR_PAGEFLAGS
487#error SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > BITS_PER_LONG - NR_PAGEFLAGS
Dave Hansen348f8b62005-06-23 00:07:40 -0700488#endif
489
Andy Whitcroftd41dee32005-06-23 00:07:54 -0700490#define ZONES_MASK ((1UL << ZONES_WIDTH) - 1)
491#define NODES_MASK ((1UL << NODES_WIDTH) - 1)
492#define SECTIONS_MASK ((1UL << SECTIONS_WIDTH) - 1)
Christoph Lameter89689ae2006-12-06 20:31:45 -0800493#define ZONEID_MASK ((1UL << ZONEID_SHIFT) - 1)
Dave Hansen348f8b62005-06-23 00:07:40 -0700494
Christoph Lameter2f1b6242006-09-25 23:31:13 -0700495static inline enum zone_type page_zonenum(struct page *page)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700496{
Dave Hansen348f8b62005-06-23 00:07:40 -0700497 return (page->flags >> ZONES_PGSHIFT) & ZONES_MASK;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700498}
Linus Torvalds1da177e2005-04-16 15:20:36 -0700499
Christoph Lameter89689ae2006-12-06 20:31:45 -0800500/*
501 * The identification function is only used by the buddy allocator for
502 * determining if two pages could be buddies. We are not really
503 * identifying a zone since we could be using a the section number
504 * id if we have not node id available in page flags.
505 * We guarantee only that it will return the same value for two
506 * combinable pages in a zone.
507 */
Andy Whitcroftcb2b95e2006-06-23 02:03:01 -0700508static inline int page_zone_id(struct page *page)
509{
Christoph Lameter89689ae2006-12-06 20:31:45 -0800510 return (page->flags >> ZONEID_PGSHIFT) & ZONEID_MASK;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700511}
512
Andy Whitcroft25ba77c2006-12-06 20:33:03 -0800513static inline int zone_to_nid(struct zone *zone)
Christoph Lameter89fa3022006-09-25 23:31:55 -0700514{
Christoph Lameterd5f541e2006-09-27 01:50:08 -0700515#ifdef CONFIG_NUMA
516 return zone->node;
517#else
518 return 0;
519#endif
Christoph Lameter89fa3022006-09-25 23:31:55 -0700520}
521
Christoph Lameter89689ae2006-12-06 20:31:45 -0800522#ifdef NODE_NOT_IN_PAGE_FLAGS
Andy Whitcroft25ba77c2006-12-06 20:33:03 -0800523extern int page_to_nid(struct page *page);
Christoph Lameter89689ae2006-12-06 20:31:45 -0800524#else
Andy Whitcroft25ba77c2006-12-06 20:33:03 -0800525static inline int page_to_nid(struct page *page)
Andy Whitcroftd41dee32005-06-23 00:07:54 -0700526{
Christoph Lameter89689ae2006-12-06 20:31:45 -0800527 return (page->flags >> NODES_PGSHIFT) & NODES_MASK;
Andy Whitcroftd41dee32005-06-23 00:07:54 -0700528}
Christoph Lameter89689ae2006-12-06 20:31:45 -0800529#endif
530
531static inline struct zone *page_zone(struct page *page)
532{
533 return &NODE_DATA(page_to_nid(page))->node_zones[page_zonenum(page)];
534}
535
Christoph Lameter308c05e2008-04-28 02:12:43 -0700536#if defined(CONFIG_SPARSEMEM) && !defined(CONFIG_SPARSEMEM_VMEMMAP)
Andy Whitcroftd41dee32005-06-23 00:07:54 -0700537static inline unsigned long page_to_section(struct page *page)
538{
539 return (page->flags >> SECTIONS_PGSHIFT) & SECTIONS_MASK;
540}
Christoph Lameter308c05e2008-04-28 02:12:43 -0700541#endif
Andy Whitcroftd41dee32005-06-23 00:07:54 -0700542
Christoph Lameter2f1b6242006-09-25 23:31:13 -0700543static inline void set_page_zone(struct page *page, enum zone_type zone)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700544{
Dave Hansen348f8b62005-06-23 00:07:40 -0700545 page->flags &= ~(ZONES_MASK << ZONES_PGSHIFT);
546 page->flags |= (zone & ZONES_MASK) << ZONES_PGSHIFT;
547}
Christoph Lameter2f1b6242006-09-25 23:31:13 -0700548
Dave Hansen348f8b62005-06-23 00:07:40 -0700549static inline void set_page_node(struct page *page, unsigned long node)
550{
551 page->flags &= ~(NODES_MASK << NODES_PGSHIFT);
552 page->flags |= (node & NODES_MASK) << NODES_PGSHIFT;
553}
Christoph Lameter89689ae2006-12-06 20:31:45 -0800554
Andy Whitcroftd41dee32005-06-23 00:07:54 -0700555static inline void set_page_section(struct page *page, unsigned long section)
556{
557 page->flags &= ~(SECTIONS_MASK << SECTIONS_PGSHIFT);
558 page->flags |= (section & SECTIONS_MASK) << SECTIONS_PGSHIFT;
559}
Dave Hansen348f8b62005-06-23 00:07:40 -0700560
Christoph Lameter2f1b6242006-09-25 23:31:13 -0700561static inline void set_page_links(struct page *page, enum zone_type zone,
Andy Whitcroftd41dee32005-06-23 00:07:54 -0700562 unsigned long node, unsigned long pfn)
Dave Hansen348f8b62005-06-23 00:07:40 -0700563{
564 set_page_zone(page, zone);
565 set_page_node(page, node);
Andy Whitcroftd41dee32005-06-23 00:07:54 -0700566 set_page_section(page, pfn_to_section_nr(pfn));
Linus Torvalds1da177e2005-04-16 15:20:36 -0700567}
568
Christoph Lameterf6ac2352006-06-30 01:55:32 -0700569/*
Eric Paris7cd94142007-11-26 18:47:40 -0500570 * If a hint addr is less than mmap_min_addr change hint to be as
571 * low as possible but still greater than mmap_min_addr
572 */
573static inline unsigned long round_hint_to_min(unsigned long hint)
574{
575#ifdef CONFIG_SECURITY
576 hint &= PAGE_MASK;
577 if (((void *)hint != NULL) &&
578 (hint < mmap_min_addr))
579 return PAGE_ALIGN(mmap_min_addr);
580#endif
581 return hint;
582}
583
584/*
Christoph Lameterf6ac2352006-06-30 01:55:32 -0700585 * Some inline functions in vmstat.h depend on page_zone()
586 */
587#include <linux/vmstat.h>
588
Ingo Molnar652050a2006-01-14 13:21:30 -0800589static __always_inline void *lowmem_page_address(struct page *page)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700590{
591 return __va(page_to_pfn(page) << PAGE_SHIFT);
592}
593
594#if defined(CONFIG_HIGHMEM) && !defined(WANT_PAGE_VIRTUAL)
595#define HASHED_PAGE_VIRTUAL
596#endif
597
598#if defined(WANT_PAGE_VIRTUAL)
599#define page_address(page) ((page)->virtual)
600#define set_page_address(page, address) \
601 do { \
602 (page)->virtual = (address); \
603 } while(0)
604#define page_address_init() do { } while(0)
605#endif
606
607#if defined(HASHED_PAGE_VIRTUAL)
608void *page_address(struct page *page);
609void set_page_address(struct page *page, void *virtual);
610void page_address_init(void);
611#endif
612
613#if !defined(HASHED_PAGE_VIRTUAL) && !defined(WANT_PAGE_VIRTUAL)
614#define page_address(page) lowmem_page_address(page)
615#define set_page_address(page, address) do { } while(0)
616#define page_address_init() do { } while(0)
617#endif
618
619/*
620 * On an anonymous page mapped into a user virtual memory area,
621 * page->mapping points to its anon_vma, not to a struct address_space;
622 * with the PAGE_MAPPING_ANON bit set to distinguish it.
623 *
624 * Please note that, confusingly, "page_mapping" refers to the inode
625 * address_space which maps the page from disk; whereas "page_mapped"
626 * refers to user virtual address space into which the page is mapped.
627 */
628#define PAGE_MAPPING_ANON 1
629
630extern struct address_space swapper_space;
631static inline struct address_space *page_mapping(struct page *page)
632{
633 struct address_space *mapping = page->mapping;
634
Christoph Lameterb5fab142007-07-17 04:03:33 -0700635 VM_BUG_ON(PageSlab(page));
Andrew Morton726b8012008-04-28 02:12:44 -0700636#ifdef CONFIG_SWAP
Linus Torvalds1da177e2005-04-16 15:20:36 -0700637 if (unlikely(PageSwapCache(page)))
638 mapping = &swapper_space;
Andrew Morton726b8012008-04-28 02:12:44 -0700639 else
640#endif
641 if (unlikely((unsigned long)mapping & PAGE_MAPPING_ANON))
Linus Torvalds1da177e2005-04-16 15:20:36 -0700642 mapping = NULL;
643 return mapping;
644}
645
646static inline int PageAnon(struct page *page)
647{
648 return ((unsigned long)page->mapping & PAGE_MAPPING_ANON) != 0;
649}
650
651/*
652 * Return the pagecache index of the passed page. Regular pagecache pages
653 * use ->index whereas swapcache pages use ->private
654 */
655static inline pgoff_t page_index(struct page *page)
656{
657 if (unlikely(PageSwapCache(page)))
Hugh Dickins4c21e2f2005-10-29 18:16:40 -0700658 return page_private(page);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700659 return page->index;
660}
661
662/*
663 * The atomic page->_mapcount, like _count, starts from -1:
664 * so that transitions both from it and to it can be tracked,
665 * using atomic_inc_and_test and atomic_add_negative(-1).
666 */
667static inline void reset_page_mapcount(struct page *page)
668{
669 atomic_set(&(page)->_mapcount, -1);
670}
671
672static inline int page_mapcount(struct page *page)
673{
674 return atomic_read(&(page)->_mapcount) + 1;
675}
676
677/*
678 * Return true if this page is mapped into pagetables.
679 */
680static inline int page_mapped(struct page *page)
681{
682 return atomic_read(&(page)->_mapcount) >= 0;
683}
684
685/*
Linus Torvalds1da177e2005-04-16 15:20:36 -0700686 * Different kinds of faults, as returned by handle_mm_fault().
687 * Used to decide whether a process gets delivered SIGBUS or
688 * just gets major/minor fault counters bumped up.
689 */
Nick Piggind0217ac2007-07-19 01:47:03 -0700690
Nick Piggin83c54072007-07-19 01:47:05 -0700691#define VM_FAULT_MINOR 0 /* For backwards compat. Remove me quickly. */
Nick Piggind0217ac2007-07-19 01:47:03 -0700692
Nick Piggin83c54072007-07-19 01:47:05 -0700693#define VM_FAULT_OOM 0x0001
694#define VM_FAULT_SIGBUS 0x0002
695#define VM_FAULT_MAJOR 0x0004
696#define VM_FAULT_WRITE 0x0008 /* Special case for get_user_pages */
Nick Pigginf33ea7f2005-08-03 20:24:01 +1000697
Nick Piggin83c54072007-07-19 01:47:05 -0700698#define VM_FAULT_NOPAGE 0x0100 /* ->fault installed the pte, not return page */
699#define VM_FAULT_LOCKED 0x0200 /* ->fault locked the returned page */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700700
Nick Piggin83c54072007-07-19 01:47:05 -0700701#define VM_FAULT_ERROR (VM_FAULT_OOM | VM_FAULT_SIGBUS)
Nick Piggind0217ac2007-07-19 01:47:03 -0700702
Linus Torvalds1da177e2005-04-16 15:20:36 -0700703#define offset_in_page(p) ((unsigned long)(p) & ~PAGE_MASK)
704
705extern void show_free_areas(void);
706
707#ifdef CONFIG_SHMEM
Lee Schermerhorn89e004ea2008-10-18 20:26:43 -0700708extern int shmem_lock(struct file *file, int lock, struct user_struct *user);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700709#else
Russell King03b00eb2006-01-06 00:10:52 -0800710static inline int shmem_lock(struct file *file, int lock,
Lee Schermerhorn89e004ea2008-10-18 20:26:43 -0700711 struct user_struct *user)
Russell King03b00eb2006-01-06 00:10:52 -0800712{
713 return 0;
714}
Linus Torvalds1da177e2005-04-16 15:20:36 -0700715#endif
716struct file *shmem_file_setup(char *name, loff_t size, unsigned long flags);
717
718int shmem_zero_setup(struct vm_area_struct *);
719
David Howellsb0e15192006-01-06 00:11:42 -0800720#ifndef CONFIG_MMU
721extern unsigned long shmem_get_unmapped_area(struct file *file,
722 unsigned long addr,
723 unsigned long len,
724 unsigned long pgoff,
725 unsigned long flags);
726#endif
727
Alexey Dobriyane8edc6e2007-05-21 01:22:52 +0400728extern int can_do_mlock(void);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700729extern int user_shm_lock(size_t, struct user_struct *);
730extern void user_shm_unlock(size_t, struct user_struct *);
731
732/*
733 * Parameter block passed down to zap_pte_range in exceptional cases.
734 */
735struct zap_details {
736 struct vm_area_struct *nonlinear_vma; /* Check page->index if set */
737 struct address_space *check_mapping; /* Check page->mapping if set */
738 pgoff_t first_index; /* Lowest page->index to unmap */
739 pgoff_t last_index; /* Highest page->index to unmap */
740 spinlock_t *i_mmap_lock; /* For unmap_mapping_range: */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700741 unsigned long truncate_count; /* Compare vm_truncate_count */
742};
743
Nick Piggin7e675132008-04-28 02:13:00 -0700744struct page *vm_normal_page(struct vm_area_struct *vma, unsigned long addr,
745 pte_t pte);
746
Jack Steinerc627f9c2008-07-29 22:33:53 -0700747int zap_vma_ptes(struct vm_area_struct *vma, unsigned long address,
748 unsigned long size);
Hugh Dickinsee39b372005-04-19 13:29:15 -0700749unsigned long zap_page_range(struct vm_area_struct *vma, unsigned long address,
Linus Torvalds1da177e2005-04-16 15:20:36 -0700750 unsigned long size, struct zap_details *);
Hugh Dickins508034a2005-10-29 18:16:30 -0700751unsigned long unmap_vmas(struct mmu_gather **tlb,
Linus Torvalds1da177e2005-04-16 15:20:36 -0700752 struct vm_area_struct *start_vma, unsigned long start_addr,
753 unsigned long end_addr, unsigned long *nr_accounted,
754 struct zap_details *);
Matt Mackalle6473092008-02-04 22:29:01 -0800755
756/**
757 * mm_walk - callbacks for walk_page_range
758 * @pgd_entry: if set, called for each non-empty PGD (top-level) entry
759 * @pud_entry: if set, called for each non-empty PUD (2nd-level) entry
760 * @pmd_entry: if set, called for each non-empty PMD (3rd-level) entry
761 * @pte_entry: if set, called for each non-empty PTE (4th-level) entry
762 * @pte_hole: if set, called for each hole at all levels
763 *
764 * (see walk_page_range for more details)
765 */
766struct mm_walk {
Dave Hansen21650092008-06-12 15:21:47 -0700767 int (*pgd_entry)(pgd_t *, unsigned long, unsigned long, struct mm_walk *);
768 int (*pud_entry)(pud_t *, unsigned long, unsigned long, struct mm_walk *);
769 int (*pmd_entry)(pmd_t *, unsigned long, unsigned long, struct mm_walk *);
770 int (*pte_entry)(pte_t *, unsigned long, unsigned long, struct mm_walk *);
771 int (*pte_hole)(unsigned long, unsigned long, struct mm_walk *);
772 struct mm_struct *mm;
773 void *private;
Matt Mackalle6473092008-02-04 22:29:01 -0800774};
775
Dave Hansen21650092008-06-12 15:21:47 -0700776int walk_page_range(unsigned long addr, unsigned long end,
777 struct mm_walk *walk);
Jan Beulich42b77722008-07-23 21:27:10 -0700778void free_pgd_range(struct mmu_gather *tlb, unsigned long addr,
Hugh Dickins3bf5ee92005-04-19 13:29:16 -0700779 unsigned long end, unsigned long floor, unsigned long ceiling);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700780int copy_page_range(struct mm_struct *dst, struct mm_struct *src,
781 struct vm_area_struct *vma);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700782void unmap_mapping_range(struct address_space *mapping,
783 loff_t const holebegin, loff_t const holelen, int even_cows);
Rik van Riel28b2ee22008-07-23 21:27:05 -0700784int generic_access_phys(struct vm_area_struct *vma, unsigned long addr,
785 void *buf, int len, int write);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700786
787static inline void unmap_shared_mapping_range(struct address_space *mapping,
788 loff_t const holebegin, loff_t const holelen)
789{
790 unmap_mapping_range(mapping, holebegin, holelen, 0);
791}
792
793extern int vmtruncate(struct inode * inode, loff_t offset);
Badari Pulavartyf6b3ec22006-01-06 00:10:38 -0800794extern int vmtruncate_range(struct inode * inode, loff_t offset, loff_t end);
Nick Pigginf33ea7f2005-08-03 20:24:01 +1000795
David Howells7ee1dd32006-01-06 00:11:44 -0800796#ifdef CONFIG_MMU
Nick Piggin83c54072007-07-19 01:47:05 -0700797extern int handle_mm_fault(struct mm_struct *mm, struct vm_area_struct *vma,
David Howells7ee1dd32006-01-06 00:11:44 -0800798 unsigned long address, int write_access);
David Howells7ee1dd32006-01-06 00:11:44 -0800799#else
800static inline int handle_mm_fault(struct mm_struct *mm,
801 struct vm_area_struct *vma, unsigned long address,
802 int write_access)
803{
804 /* should never happen if there's no MMU */
805 BUG();
806 return VM_FAULT_SIGBUS;
807}
808#endif
Nick Pigginf33ea7f2005-08-03 20:24:01 +1000809
Linus Torvalds1da177e2005-04-16 15:20:36 -0700810extern int make_pages_present(unsigned long addr, unsigned long end);
811extern int access_process_vm(struct task_struct *tsk, unsigned long addr, void *buf, int len, int write);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700812
813int get_user_pages(struct task_struct *tsk, struct mm_struct *mm, unsigned long start,
814 int len, int write, int force, struct page **pages, struct vm_area_struct **vmas);
815
David Howellscf9a2ae2006-08-29 19:05:54 +0100816extern int try_to_release_page(struct page * page, gfp_t gfp_mask);
817extern void do_invalidatepage(struct page *page, unsigned long offset);
818
Linus Torvalds1da177e2005-04-16 15:20:36 -0700819int __set_page_dirty_nobuffers(struct page *page);
Ken Chen76719322007-02-10 01:43:15 -0800820int __set_page_dirty_no_writeback(struct page *page);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700821int redirty_page_for_writepage(struct writeback_control *wbc,
822 struct page *page);
Harvey Harrisonb3c97522008-02-13 15:03:15 -0800823int set_page_dirty(struct page *page);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700824int set_page_dirty_lock(struct page *page);
825int clear_page_dirty_for_io(struct page *page);
826
Ollie Wildb6a2fea2007-07-19 01:48:16 -0700827extern unsigned long move_page_tables(struct vm_area_struct *vma,
828 unsigned long old_addr, struct vm_area_struct *new_vma,
829 unsigned long new_addr, unsigned long len);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700830extern unsigned long do_mremap(unsigned long addr,
831 unsigned long old_len, unsigned long new_len,
832 unsigned long flags, unsigned long new_addr);
Ollie Wildb6a2fea2007-07-19 01:48:16 -0700833extern int mprotect_fixup(struct vm_area_struct *vma,
834 struct vm_area_struct **pprev, unsigned long start,
835 unsigned long end, unsigned long newflags);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700836
Nick Piggin21cc1992008-07-25 19:45:22 -0700837/*
838 * get_user_pages_fast provides equivalent functionality to get_user_pages,
839 * operating on current and current->mm (force=0 and doesn't return any vmas).
840 *
841 * get_user_pages_fast may take mmap_sem and page tables, so no assumptions
842 * can be made about locking. get_user_pages_fast is to be implemented in a
843 * way that is advantageous (vs get_user_pages()) when the user memory area is
844 * already faulted in and present in ptes. However if the pages have to be
845 * faulted in, it may turn out to be slightly slower).
846 */
847int get_user_pages_fast(unsigned long start, int nr_pages, int write,
848 struct page **pages);
849
Linus Torvalds1da177e2005-04-16 15:20:36 -0700850/*
Rusty Russell8e1f9362007-07-17 04:03:17 -0700851 * A callback you can register to apply pressure to ageable caches.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700852 *
Rusty Russell8e1f9362007-07-17 04:03:17 -0700853 * 'shrink' is passed a count 'nr_to_scan' and a 'gfpmask'. It should
854 * look through the least-recently-used 'nr_to_scan' entries and
855 * attempt to free them up. It should return the number of objects
856 * which remain in the cache. If it returns -1, it means it cannot do
857 * any scanning at this time (eg. there is a risk of deadlock).
Linus Torvalds1da177e2005-04-16 15:20:36 -0700858 *
Rusty Russell8e1f9362007-07-17 04:03:17 -0700859 * The 'gfpmask' refers to the allocation we are currently trying to
860 * fulfil.
861 *
862 * Note that 'shrink' will be passed nr_to_scan == 0 when the VM is
863 * querying the cache size, so a fastpath for that case is appropriate.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700864 */
Rusty Russell8e1f9362007-07-17 04:03:17 -0700865struct shrinker {
866 int (*shrink)(int nr_to_scan, gfp_t gfp_mask);
867 int seeks; /* seeks to recreate an obj */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700868
Rusty Russell8e1f9362007-07-17 04:03:17 -0700869 /* These are for internal use */
870 struct list_head list;
871 long nr; /* objs pending delete */
872};
873#define DEFAULT_SEEKS 2 /* A good number if you don't know better. */
874extern void register_shrinker(struct shrinker *);
875extern void unregister_shrinker(struct shrinker *);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700876
Alexey Dobriyan4e950f62007-07-30 02:36:13 +0400877int vma_wants_writenotify(struct vm_area_struct *vma);
Peter Zijlstrad08b3852006-09-25 23:30:57 -0700878
Harvey Harrisonb3c97522008-02-13 15:03:15 -0800879extern pte_t *get_locked_pte(struct mm_struct *mm, unsigned long addr, spinlock_t **ptl);
Linus Torvaldsc9cfcdd2005-11-29 14:03:14 -0800880
Nick Piggin5f22df02007-05-06 14:49:02 -0700881#ifdef __PAGETABLE_PUD_FOLDED
882static inline int __pud_alloc(struct mm_struct *mm, pgd_t *pgd,
883 unsigned long address)
884{
885 return 0;
886}
887#else
Hugh Dickins1bb36302005-10-29 18:16:22 -0700888int __pud_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address);
Nick Piggin5f22df02007-05-06 14:49:02 -0700889#endif
890
891#ifdef __PAGETABLE_PMD_FOLDED
892static inline int __pmd_alloc(struct mm_struct *mm, pud_t *pud,
893 unsigned long address)
894{
895 return 0;
896}
897#else
Hugh Dickins1bb36302005-10-29 18:16:22 -0700898int __pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address);
Nick Piggin5f22df02007-05-06 14:49:02 -0700899#endif
900
Hugh Dickins1bb36302005-10-29 18:16:22 -0700901int __pte_alloc(struct mm_struct *mm, pmd_t *pmd, unsigned long address);
902int __pte_alloc_kernel(pmd_t *pmd, unsigned long address);
903
Linus Torvalds1da177e2005-04-16 15:20:36 -0700904/*
905 * The following ifdef needed to get the 4level-fixup.h header to work.
906 * Remove it when 4level-fixup.h has been removed.
907 */
Hugh Dickins1bb36302005-10-29 18:16:22 -0700908#if defined(CONFIG_MMU) && !defined(__ARCH_HAS_4LEVEL_HACK)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700909static inline pud_t *pud_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address)
910{
Hugh Dickins1bb36302005-10-29 18:16:22 -0700911 return (unlikely(pgd_none(*pgd)) && __pud_alloc(mm, pgd, address))?
912 NULL: pud_offset(pgd, address);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700913}
914
915static inline pmd_t *pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address)
916{
Hugh Dickins1bb36302005-10-29 18:16:22 -0700917 return (unlikely(pud_none(*pud)) && __pmd_alloc(mm, pud, address))?
918 NULL: pmd_offset(pud, address);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700919}
Hugh Dickins1bb36302005-10-29 18:16:22 -0700920#endif /* CONFIG_MMU && !__ARCH_HAS_4LEVEL_HACK */
921
Jeremy Fitzhardingef7d0b922008-09-09 15:43:22 -0700922#if USE_SPLIT_PTLOCKS
Hugh Dickins4c21e2f2005-10-29 18:16:40 -0700923/*
924 * We tuck a spinlock to guard each pagetable page into its struct page,
925 * at page->private, with BUILD_BUG_ON to make sure that this will not
926 * overflow into the next struct page (as it might with DEBUG_SPINLOCK).
927 * When freeing, reset page->mapping so free_pages_check won't complain.
928 */
Andrew Morton349aef02006-01-08 01:04:36 -0800929#define __pte_lockptr(page) &((page)->ptl)
Hugh Dickins4c21e2f2005-10-29 18:16:40 -0700930#define pte_lock_init(_page) do { \
931 spin_lock_init(__pte_lockptr(_page)); \
932} while (0)
933#define pte_lock_deinit(page) ((page)->mapping = NULL)
934#define pte_lockptr(mm, pmd) ({(void)(mm); __pte_lockptr(pmd_page(*(pmd)));})
Jeremy Fitzhardingef7d0b922008-09-09 15:43:22 -0700935#else /* !USE_SPLIT_PTLOCKS */
Hugh Dickins4c21e2f2005-10-29 18:16:40 -0700936/*
937 * We use mm->page_table_lock to guard all pagetable pages of the mm.
938 */
939#define pte_lock_init(page) do {} while (0)
940#define pte_lock_deinit(page) do {} while (0)
941#define pte_lockptr(mm, pmd) ({(void)(pmd); &(mm)->page_table_lock;})
Jeremy Fitzhardingef7d0b922008-09-09 15:43:22 -0700942#endif /* USE_SPLIT_PTLOCKS */
Hugh Dickins4c21e2f2005-10-29 18:16:40 -0700943
Martin Schwidefsky2f569af2008-02-08 04:22:04 -0800944static inline void pgtable_page_ctor(struct page *page)
945{
946 pte_lock_init(page);
947 inc_zone_page_state(page, NR_PAGETABLE);
948}
949
950static inline void pgtable_page_dtor(struct page *page)
951{
952 pte_lock_deinit(page);
953 dec_zone_page_state(page, NR_PAGETABLE);
954}
955
Hugh Dickinsc74df322005-10-29 18:16:23 -0700956#define pte_offset_map_lock(mm, pmd, address, ptlp) \
957({ \
Hugh Dickins4c21e2f2005-10-29 18:16:40 -0700958 spinlock_t *__ptl = pte_lockptr(mm, pmd); \
Hugh Dickinsc74df322005-10-29 18:16:23 -0700959 pte_t *__pte = pte_offset_map(pmd, address); \
960 *(ptlp) = __ptl; \
961 spin_lock(__ptl); \
962 __pte; \
963})
964
965#define pte_unmap_unlock(pte, ptl) do { \
966 spin_unlock(ptl); \
967 pte_unmap(pte); \
968} while (0)
969
Hugh Dickins1bb36302005-10-29 18:16:22 -0700970#define pte_alloc_map(mm, pmd, address) \
971 ((unlikely(!pmd_present(*(pmd))) && __pte_alloc(mm, pmd, address))? \
972 NULL: pte_offset_map(pmd, address))
973
Hugh Dickinsc74df322005-10-29 18:16:23 -0700974#define pte_alloc_map_lock(mm, pmd, address, ptlp) \
975 ((unlikely(!pmd_present(*(pmd))) && __pte_alloc(mm, pmd, address))? \
976 NULL: pte_offset_map_lock(mm, pmd, address, ptlp))
977
Hugh Dickins1bb36302005-10-29 18:16:22 -0700978#define pte_alloc_kernel(pmd, address) \
979 ((unlikely(!pmd_present(*(pmd))) && __pte_alloc_kernel(pmd, address))? \
980 NULL: pte_offset_kernel(pmd, address))
Linus Torvalds1da177e2005-04-16 15:20:36 -0700981
982extern void free_area_init(unsigned long * zones_size);
Johannes Weiner9109fb72008-07-23 21:27:20 -0700983extern void free_area_init_node(int nid, unsigned long * zones_size,
984 unsigned long zone_start_pfn, unsigned long *zholes_size);
Mel Gormanc7132162006-09-27 01:49:43 -0700985#ifdef CONFIG_ARCH_POPULATES_NODE_MAP
986/*
987 * With CONFIG_ARCH_POPULATES_NODE_MAP set, an architecture may initialise its
988 * zones, allocate the backing mem_map and account for memory holes in a more
989 * architecture independent manner. This is a substitute for creating the
990 * zone_sizes[] and zholes_size[] arrays and passing them to
991 * free_area_init_node()
992 *
993 * An architecture is expected to register range of page frames backed by
994 * physical memory with add_active_range() before calling
995 * free_area_init_nodes() passing in the PFN each zone ends at. At a basic
996 * usage, an architecture is expected to do something like
997 *
998 * unsigned long max_zone_pfns[MAX_NR_ZONES] = {max_dma, max_normal_pfn,
999 * max_highmem_pfn};
1000 * for_each_valid_physical_page_range()
1001 * add_active_range(node_id, start_pfn, end_pfn)
1002 * free_area_init_nodes(max_zone_pfns);
1003 *
1004 * If the architecture guarantees that there are no holes in the ranges
1005 * registered with add_active_range(), free_bootmem_active_regions()
1006 * will call free_bootmem_node() for each registered physical page range.
1007 * Similarly sparse_memory_present_with_active_regions() calls
1008 * memory_present() for each range when SPARSEMEM is enabled.
1009 *
1010 * See mm/page_alloc.c for more information on each function exposed by
1011 * CONFIG_ARCH_POPULATES_NODE_MAP
1012 */
1013extern void free_area_init_nodes(unsigned long *max_zone_pfn);
1014extern void add_active_range(unsigned int nid, unsigned long start_pfn,
1015 unsigned long end_pfn);
Yinghai Lucc1050b2008-06-13 19:08:52 -07001016extern void remove_active_range(unsigned int nid, unsigned long start_pfn,
1017 unsigned long end_pfn);
Mel Gormanfb014392006-09-27 01:49:59 -07001018extern void push_node_boundaries(unsigned int nid, unsigned long start_pfn,
1019 unsigned long end_pfn);
Mel Gormanc7132162006-09-27 01:49:43 -07001020extern void remove_all_active_ranges(void);
1021extern unsigned long absent_pages_in_range(unsigned long start_pfn,
1022 unsigned long end_pfn);
1023extern void get_pfn_range_for_nid(unsigned int nid,
1024 unsigned long *start_pfn, unsigned long *end_pfn);
1025extern unsigned long find_min_pfn_with_active_regions(void);
Mel Gormanc7132162006-09-27 01:49:43 -07001026extern void free_bootmem_with_active_regions(int nid,
1027 unsigned long max_low_pfn);
Yinghai Lud52d53b2008-06-16 20:10:55 -07001028typedef int (*work_fn_t)(unsigned long, unsigned long, void *);
Yinghai Lub5bc6c02008-06-14 18:32:52 -07001029extern void work_with_active_regions(int nid, work_fn_t work_fn, void *data);
Mel Gormanc7132162006-09-27 01:49:43 -07001030extern void sparse_memory_present_with_active_regions(int nid);
1031#ifndef CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID
1032extern int early_pfn_to_nid(unsigned long pfn);
1033#endif /* CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID */
1034#endif /* CONFIG_ARCH_POPULATES_NODE_MAP */
Mel Gorman0e0b8642006-09-27 01:49:56 -07001035extern void set_dma_reserve(unsigned long new_dma_reserve);
Dave Hansena2f3aa02007-01-10 23:15:30 -08001036extern void memmap_init_zone(unsigned long, int, unsigned long,
1037 unsigned long, enum memmap_context);
Dave Hansen3947be12005-10-29 18:16:54 -07001038extern void setup_per_zone_pages_min(void);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001039extern void mem_init(void);
1040extern void show_mem(void);
1041extern void si_meminfo(struct sysinfo * val);
1042extern void si_meminfo_node(struct sysinfo *val, int nid);
Mike Travis3461b0a2008-05-12 21:21:13 +02001043extern int after_bootmem;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001044
Christoph Lametere7c8d5c2005-06-21 17:14:47 -07001045#ifdef CONFIG_NUMA
1046extern void setup_per_cpu_pageset(void);
1047#else
1048static inline void setup_per_cpu_pageset(void) {}
1049#endif
1050
Linus Torvalds1da177e2005-04-16 15:20:36 -07001051/* prio_tree.c */
1052void vma_prio_tree_add(struct vm_area_struct *, struct vm_area_struct *old);
1053void vma_prio_tree_insert(struct vm_area_struct *, struct prio_tree_root *);
1054void vma_prio_tree_remove(struct vm_area_struct *, struct prio_tree_root *);
1055struct vm_area_struct *vma_prio_tree_next(struct vm_area_struct *vma,
1056 struct prio_tree_iter *iter);
1057
1058#define vma_prio_tree_foreach(vma, iter, root, begin, end) \
1059 for (prio_tree_iter_init(iter, root, begin, end), vma = NULL; \
1060 (vma = vma_prio_tree_next(vma, iter)); )
1061
1062static inline void vma_nonlinear_insert(struct vm_area_struct *vma,
1063 struct list_head *list)
1064{
1065 vma->shared.vm_set.parent = NULL;
1066 list_add_tail(&vma->shared.vm_set.list, list);
1067}
1068
1069/* mmap.c */
Alan Cox34b4e4a2007-08-22 14:01:28 -07001070extern int __vm_enough_memory(struct mm_struct *mm, long pages, int cap_sys_admin);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001071extern void vma_adjust(struct vm_area_struct *vma, unsigned long start,
1072 unsigned long end, pgoff_t pgoff, struct vm_area_struct *insert);
1073extern struct vm_area_struct *vma_merge(struct mm_struct *,
1074 struct vm_area_struct *prev, unsigned long addr, unsigned long end,
1075 unsigned long vm_flags, struct anon_vma *, struct file *, pgoff_t,
1076 struct mempolicy *);
1077extern struct anon_vma *find_mergeable_anon_vma(struct vm_area_struct *);
1078extern int split_vma(struct mm_struct *,
1079 struct vm_area_struct *, unsigned long addr, int new_below);
1080extern int insert_vm_struct(struct mm_struct *, struct vm_area_struct *);
1081extern void __vma_link_rb(struct mm_struct *, struct vm_area_struct *,
1082 struct rb_node **, struct rb_node *);
Hugh Dickinsa8fb5612005-10-29 18:15:57 -07001083extern void unlink_file_vma(struct vm_area_struct *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001084extern struct vm_area_struct *copy_vma(struct vm_area_struct **,
1085 unsigned long addr, unsigned long len, pgoff_t pgoff);
1086extern void exit_mmap(struct mm_struct *);
Matt Helsley925d1c42008-04-29 01:01:36 -07001087
Andrea Arcangeli7906d002008-07-28 15:46:26 -07001088extern int mm_take_all_locks(struct mm_struct *mm);
1089extern void mm_drop_all_locks(struct mm_struct *mm);
1090
Matt Helsley925d1c42008-04-29 01:01:36 -07001091#ifdef CONFIG_PROC_FS
1092/* From fs/proc/base.c. callers must _not_ hold the mm's exe_file_lock */
1093extern void added_exe_file_vma(struct mm_struct *mm);
1094extern void removed_exe_file_vma(struct mm_struct *mm);
1095#else
1096static inline void added_exe_file_vma(struct mm_struct *mm)
1097{}
1098
1099static inline void removed_exe_file_vma(struct mm_struct *mm)
1100{}
1101#endif /* CONFIG_PROC_FS */
1102
akpm@osdl.org119f6572005-05-01 08:58:35 -07001103extern int may_expand_vm(struct mm_struct *mm, unsigned long npages);
Roland McGrathfa5dc222007-02-08 14:20:41 -08001104extern int install_special_mapping(struct mm_struct *mm,
1105 unsigned long addr, unsigned long len,
1106 unsigned long flags, struct page **pages);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001107
1108extern unsigned long get_unmapped_area(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
1109
1110extern unsigned long do_mmap_pgoff(struct file *file, unsigned long addr,
1111 unsigned long len, unsigned long prot,
1112 unsigned long flag, unsigned long pgoff);
Miklos Szeredi0165ab42007-07-15 23:38:26 -07001113extern unsigned long mmap_region(struct file *file, unsigned long addr,
1114 unsigned long len, unsigned long flags,
1115 unsigned int vm_flags, unsigned long pgoff,
1116 int accountable);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001117
1118static inline unsigned long do_mmap(struct file *file, unsigned long addr,
1119 unsigned long len, unsigned long prot,
1120 unsigned long flag, unsigned long offset)
1121{
1122 unsigned long ret = -EINVAL;
1123 if ((offset + PAGE_ALIGN(len)) < offset)
1124 goto out;
1125 if (!(offset & ~PAGE_MASK))
1126 ret = do_mmap_pgoff(file, addr, len, prot, flag, offset >> PAGE_SHIFT);
1127out:
1128 return ret;
1129}
1130
1131extern int do_munmap(struct mm_struct *, unsigned long, size_t);
1132
1133extern unsigned long do_brk(unsigned long, unsigned long);
1134
1135/* filemap.c */
1136extern unsigned long page_unuse(struct page *);
1137extern void truncate_inode_pages(struct address_space *, loff_t);
Hans Reiserd7339072006-01-06 00:10:36 -08001138extern void truncate_inode_pages_range(struct address_space *,
1139 loff_t lstart, loff_t lend);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001140
1141/* generic vm_area_ops exported for stackable file systems */
Nick Piggind0217ac2007-07-19 01:47:03 -07001142extern int filemap_fault(struct vm_area_struct *, struct vm_fault *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001143
1144/* mm/page-writeback.c */
1145int write_one_page(struct page *page, int wait);
1146
1147/* readahead.c */
1148#define VM_MAX_READAHEAD 128 /* kbytes */
1149#define VM_MIN_READAHEAD 16 /* kbytes (includes current page) */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001150
1151int do_page_cache_readahead(struct address_space *mapping, struct file *filp,
Andrew Morton7361f4d2005-11-07 00:59:28 -08001152 pgoff_t offset, unsigned long nr_to_read);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001153int force_page_cache_readahead(struct address_space *mapping, struct file *filp,
Andrew Morton7361f4d2005-11-07 00:59:28 -08001154 pgoff_t offset, unsigned long nr_to_read);
Rusty Russellcf914a72007-07-19 01:48:08 -07001155
1156void page_cache_sync_readahead(struct address_space *mapping,
1157 struct file_ra_state *ra,
1158 struct file *filp,
1159 pgoff_t offset,
1160 unsigned long size);
1161
1162void page_cache_async_readahead(struct address_space *mapping,
1163 struct file_ra_state *ra,
1164 struct file *filp,
1165 struct page *pg,
1166 pgoff_t offset,
1167 unsigned long size);
1168
Linus Torvalds1da177e2005-04-16 15:20:36 -07001169unsigned long max_sane_readahead(unsigned long nr);
1170
1171/* Do stack extension */
Hugh Dickins46dea3d2005-10-29 18:16:20 -07001172extern int expand_stack(struct vm_area_struct *vma, unsigned long address);
Matthew Wilcox9ab88512005-11-18 16:16:42 -05001173#ifdef CONFIG_IA64
Hugh Dickins46dea3d2005-10-29 18:16:20 -07001174extern int expand_upwards(struct vm_area_struct *vma, unsigned long address);
Matthew Wilcox9ab88512005-11-18 16:16:42 -05001175#endif
Ollie Wildb6a2fea2007-07-19 01:48:16 -07001176extern int expand_stack_downwards(struct vm_area_struct *vma,
1177 unsigned long address);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001178
1179/* Look up the first VMA which satisfies addr < vm_end, NULL if none. */
1180extern struct vm_area_struct * find_vma(struct mm_struct * mm, unsigned long addr);
1181extern struct vm_area_struct * find_vma_prev(struct mm_struct * mm, unsigned long addr,
1182 struct vm_area_struct **pprev);
1183
1184/* Look up the first VMA which intersects the interval start_addr..end_addr-1,
1185 NULL if none. Assume start_addr < end_addr. */
1186static inline struct vm_area_struct * find_vma_intersection(struct mm_struct * mm, unsigned long start_addr, unsigned long end_addr)
1187{
1188 struct vm_area_struct * vma = find_vma(mm,start_addr);
1189
1190 if (vma && end_addr <= vma->vm_start)
1191 vma = NULL;
1192 return vma;
1193}
1194
1195static inline unsigned long vma_pages(struct vm_area_struct *vma)
1196{
1197 return (vma->vm_end - vma->vm_start) >> PAGE_SHIFT;
1198}
1199
Hugh Dickins804af2c2006-07-26 21:39:49 +01001200pgprot_t vm_get_page_prot(unsigned long vm_flags);
Hugh Dickinsdeceb6c2005-10-29 18:16:33 -07001201struct vm_area_struct *find_extend_vma(struct mm_struct *, unsigned long addr);
Hugh Dickinsdeceb6c2005-10-29 18:16:33 -07001202int remap_pfn_range(struct vm_area_struct *, unsigned long addr,
1203 unsigned long pfn, unsigned long size, pgprot_t);
Linus Torvaldsa145dd42005-11-30 09:35:19 -08001204int vm_insert_page(struct vm_area_struct *, unsigned long addr, struct page *);
Nick Piggine0dc0d82007-02-12 00:51:36 -08001205int vm_insert_pfn(struct vm_area_struct *vma, unsigned long addr,
1206 unsigned long pfn);
Nick Piggin423bad62008-04-28 02:13:01 -07001207int vm_insert_mixed(struct vm_area_struct *vma, unsigned long addr,
1208 unsigned long pfn);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001209
Linus Torvalds6aab3412005-11-28 14:34:23 -08001210struct page *follow_page(struct vm_area_struct *, unsigned long address,
Hugh Dickinsdeceb6c2005-10-29 18:16:33 -07001211 unsigned int foll_flags);
1212#define FOLL_WRITE 0x01 /* check pte is writable */
1213#define FOLL_TOUCH 0x02 /* mark page accessed */
1214#define FOLL_GET 0x04 /* do get_page on page */
1215#define FOLL_ANON 0x08 /* give ZERO_PAGE if no pgtable */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001216
Martin Schwidefsky2f569af2008-02-08 04:22:04 -08001217typedef int (*pte_fn_t)(pte_t *pte, pgtable_t token, unsigned long addr,
Jeremy Fitzhardingeaee16b32007-05-06 14:48:54 -07001218 void *data);
1219extern int apply_to_page_range(struct mm_struct *mm, unsigned long address,
1220 unsigned long size, pte_fn_t fn, void *data);
1221
Linus Torvalds1da177e2005-04-16 15:20:36 -07001222#ifdef CONFIG_PROC_FS
Hugh Dickinsab50b8e2005-10-29 18:15:56 -07001223void vm_stat_account(struct mm_struct *, unsigned long, struct file *, long);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001224#else
Hugh Dickinsab50b8e2005-10-29 18:15:56 -07001225static inline void vm_stat_account(struct mm_struct *mm,
Linus Torvalds1da177e2005-04-16 15:20:36 -07001226 unsigned long flags, struct file *file, long pages)
1227{
1228}
1229#endif /* CONFIG_PROC_FS */
1230
Ingo Molnar12d6f212008-01-30 13:33:58 +01001231#ifdef CONFIG_DEBUG_PAGEALLOC
1232extern int debug_pagealloc_enabled;
1233
1234extern void kernel_map_pages(struct page *page, int numpages, int enable);
1235
1236static inline void enable_debug_pagealloc(void)
1237{
1238 debug_pagealloc_enabled = 1;
1239}
Rafael J. Wysocki8a235ef2008-02-20 01:47:44 +01001240#ifdef CONFIG_HIBERNATION
1241extern bool kernel_page_present(struct page *page);
1242#endif /* CONFIG_HIBERNATION */
Ingo Molnar12d6f212008-01-30 13:33:58 +01001243#else
Linus Torvalds1da177e2005-04-16 15:20:36 -07001244static inline void
Nick Piggin9858db52006-10-11 01:21:30 -07001245kernel_map_pages(struct page *page, int numpages, int enable) {}
Ingo Molnar12d6f212008-01-30 13:33:58 +01001246static inline void enable_debug_pagealloc(void)
1247{
1248}
Rafael J. Wysocki8a235ef2008-02-20 01:47:44 +01001249#ifdef CONFIG_HIBERNATION
1250static inline bool kernel_page_present(struct page *page) { return true; }
1251#endif /* CONFIG_HIBERNATION */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001252#endif
1253
1254extern struct vm_area_struct *get_gate_vma(struct task_struct *tsk);
1255#ifdef __HAVE_ARCH_GATE_AREA
1256int in_gate_area_no_task(unsigned long addr);
1257int in_gate_area(struct task_struct *task, unsigned long addr);
1258#else
1259int in_gate_area_no_task(unsigned long addr);
1260#define in_gate_area(task, addr) ({(void)task; in_gate_area_no_task(addr);})
1261#endif /* __HAVE_ARCH_GATE_AREA */
1262
Andrew Morton9d0243b2006-01-08 01:00:39 -08001263int drop_caches_sysctl_handler(struct ctl_table *, int, struct file *,
1264 void __user *, size_t *, loff_t *);
Andrew Morton69e05942006-03-22 00:08:19 -08001265unsigned long shrink_slab(unsigned long scanned, gfp_t gfp_mask,
Andrew Morton9d0243b2006-01-08 01:00:39 -08001266 unsigned long lru_pages);
Andrew Morton9d0243b2006-01-08 01:00:39 -08001267
Luke Yang7a9166e2006-02-20 18:28:07 -08001268#ifndef CONFIG_MMU
1269#define randomize_va_space 0
1270#else
Andi Kleena62eaf12006-02-16 23:41:58 +01001271extern int randomize_va_space;
Luke Yang7a9166e2006-02-20 18:28:07 -08001272#endif
Andi Kleena62eaf12006-02-16 23:41:58 +01001273
Sam Ravnborg045e72a2007-07-26 10:41:13 -07001274const char * arch_vma_name(struct vm_area_struct *vma);
Andi Kleen03252912008-01-30 13:33:18 +01001275void print_vma_addr(char *prefix, unsigned long rip);
Ingo Molnare6e54942006-06-27 02:53:50 -07001276
Yasunori Goto98f3cfc2007-10-16 01:26:14 -07001277struct page *sparse_mem_map_populate(unsigned long pnum, int nid);
Andy Whitcroft29c71112007-10-16 01:24:14 -07001278pgd_t *vmemmap_pgd_populate(unsigned long addr, int node);
1279pud_t *vmemmap_pud_populate(pgd_t *pgd, unsigned long addr, int node);
1280pmd_t *vmemmap_pmd_populate(pud_t *pud, unsigned long addr, int node);
1281pte_t *vmemmap_pte_populate(pmd_t *pmd, unsigned long addr, int node);
Christoph Lameter8f6aac42007-10-16 01:24:13 -07001282void *vmemmap_alloc_block(unsigned long size, int node);
1283void vmemmap_verify(pte_t *, int, unsigned long, unsigned long);
Andy Whitcroft29c71112007-10-16 01:24:14 -07001284int vmemmap_populate_basepages(struct page *start_page,
1285 unsigned long pages, int node);
1286int vmemmap_populate(struct page *start_page, unsigned long pages, int node);
Yinghai Luc2b91e22008-04-12 01:19:24 -07001287void vmemmap_populate_print_last(void);
Christoph Lameter8f6aac42007-10-16 01:24:13 -07001288
Linus Torvalds1da177e2005-04-16 15:20:36 -07001289#endif /* __KERNEL__ */
1290#endif /* _LINUX_MM_H */