blob: c31a9cd2a30e03a2672b8fa299c4d6a24a3fa186 [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>
10#include <linux/mmzone.h>
11#include <linux/rbtree.h>
12#include <linux/prio_tree.h>
Ingo Molnar9a11b49a2006-07-03 00:24:33 -070013#include <linux/debug_locks.h>
Heiko Carstens5b99cd02006-09-27 01:50:01 -070014#include <linux/mm_types.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070015
16struct mempolicy;
17struct anon_vma;
Alexey Dobriyan4e950f62007-07-30 02:36:13 +040018struct file_ra_state;
Alexey Dobriyane8edc6e2007-05-21 01:22:52 +040019struct user_struct;
Alexey Dobriyan4e950f62007-07-30 02:36:13 +040020struct writeback_control;
Linus Torvalds1da177e2005-04-16 15:20:36 -070021
22#ifndef CONFIG_DISCONTIGMEM /* Don't use mapnrs, do it properly */
23extern unsigned long max_mapnr;
24#endif
25
26extern unsigned long num_physpages;
27extern void * high_memory;
Linus Torvalds1da177e2005-04-16 15:20:36 -070028extern 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
James Morris42d78962007-12-26 15:12:37 +110036extern unsigned long mmap_min_addr;
37
Linus Torvalds1da177e2005-04-16 15:20:36 -070038#include <asm/page.h>
39#include <asm/pgtable.h>
40#include <asm/processor.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070041
Linus Torvalds1da177e2005-04-16 15:20:36 -070042#define nth_page(page,n) pfn_to_page(page_to_pfn((page)) + (n))
43
44/*
45 * Linux kernel virtual memory manager primitives.
46 * The idea being to have a "virtual" mm in the same way
47 * we have a virtual fs - giving a cleaner interface to the
48 * mm details, and allowing different kinds of memory mappings
49 * (from shared memory to executable loading to arbitrary
50 * mmap() functions).
51 */
52
Christoph Lameterc43692e2006-12-06 20:32:48 -080053extern struct kmem_cache *vm_area_cachep;
54
Linus Torvalds1da177e2005-04-16 15:20:36 -070055/*
56 * This struct defines the per-mm list of VMAs for uClinux. If CONFIG_MMU is
57 * disabled, then there's a single shared list of VMAs maintained by the
58 * system, and mm's subscribe to these individually
59 */
60struct vm_list_struct {
61 struct vm_list_struct *next;
62 struct vm_area_struct *vma;
63};
64
65#ifndef CONFIG_MMU
66extern struct rb_root nommu_vma_tree;
67extern struct rw_semaphore nommu_vma_sem;
68
69extern unsigned int kobjsize(const void *objp);
70#endif
71
72/*
73 * vm_flags..
74 */
75#define VM_READ 0x00000001 /* currently active flags */
76#define VM_WRITE 0x00000002
77#define VM_EXEC 0x00000004
78#define VM_SHARED 0x00000008
79
Paolo 'Blaisorblade' Giarrusso7e2cff42005-09-21 09:55:39 -070080/* mprotect() hardcodes VM_MAYREAD >> 4 == VM_READ, and so for r/w/x bits. */
Linus Torvalds1da177e2005-04-16 15:20:36 -070081#define VM_MAYREAD 0x00000010 /* limits for mprotect() etc */
82#define VM_MAYWRITE 0x00000020
83#define VM_MAYEXEC 0x00000040
84#define VM_MAYSHARE 0x00000080
85
86#define VM_GROWSDOWN 0x00000100 /* general info on the segment */
87#define VM_GROWSUP 0x00000200
Linus Torvalds6aab3412005-11-28 14:34:23 -080088#define VM_PFNMAP 0x00000400 /* Page-ranges managed without "struct page", just pure PFN */
Linus Torvalds1da177e2005-04-16 15:20:36 -070089#define VM_DENYWRITE 0x00000800 /* ETXTBSY on write attempts.. */
90
91#define VM_EXECUTABLE 0x00001000
92#define VM_LOCKED 0x00002000
93#define VM_IO 0x00004000 /* Memory mapped I/O or similar */
94
95 /* Used by sys_madvise() */
96#define VM_SEQ_READ 0x00008000 /* App will access data sequentially */
97#define VM_RAND_READ 0x00010000 /* App will not benefit from clustered reads */
98
99#define VM_DONTCOPY 0x00020000 /* Do not copy this vma on fork */
100#define VM_DONTEXPAND 0x00040000 /* Cannot expand with mremap() */
Hugh Dickins0b14c172005-11-21 21:32:15 -0800101#define VM_RESERVED 0x00080000 /* Count as reserved_vm like IO */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700102#define VM_ACCOUNT 0x00100000 /* Is a VM accounted object */
103#define VM_HUGETLB 0x00400000 /* Huge TLB Page VM */
104#define VM_NONLINEAR 0x00800000 /* Is non-linear (remap_file_pages) */
105#define VM_MAPPED_COPY 0x01000000 /* T if mapped copy of data (nommu mmap) */
Linus Torvalds4d7672b2005-12-16 10:21:23 -0800106#define VM_INSERTPAGE 0x02000000 /* The vma has had "vm_insert_page()" done on it */
Roland McGrathe5b97dd2007-01-26 00:56:48 -0800107#define VM_ALWAYSDUMP 0x04000000 /* Always include in core dumps */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700108
Nick Piggind0217ac2007-07-19 01:47:03 -0700109#define VM_CAN_NONLINEAR 0x08000000 /* Has ->fault & does nonlinear pages */
Jared Hulbertb379d792008-04-28 02:12:58 -0700110#define VM_MIXEDMAP 0x10000000 /* Can contain "struct page" and pure PFN pages */
Nick Piggind00806b2007-07-19 01:46:57 -0700111
Linus Torvalds1da177e2005-04-16 15:20:36 -0700112#ifndef VM_STACK_DEFAULT_FLAGS /* arch can override this */
113#define VM_STACK_DEFAULT_FLAGS VM_DATA_DEFAULT_FLAGS
114#endif
115
116#ifdef CONFIG_STACK_GROWSUP
117#define VM_STACK_FLAGS (VM_GROWSUP | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT)
118#else
119#define VM_STACK_FLAGS (VM_GROWSDOWN | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT)
120#endif
121
122#define VM_READHINTMASK (VM_SEQ_READ | VM_RAND_READ)
123#define VM_ClearReadHint(v) (v)->vm_flags &= ~VM_READHINTMASK
124#define VM_NormalReadHint(v) (!((v)->vm_flags & VM_READHINTMASK))
125#define VM_SequentialReadHint(v) ((v)->vm_flags & VM_SEQ_READ)
126#define VM_RandomReadHint(v) ((v)->vm_flags & VM_RAND_READ)
127
128/*
129 * mapping from the currently active vm_flags protection bits (the
130 * low four bits) to a page protection mask..
131 */
132extern pgprot_t protection_map[16];
133
Nick Piggind0217ac2007-07-19 01:47:03 -0700134#define FAULT_FLAG_WRITE 0x01 /* Fault was a write access */
135#define FAULT_FLAG_NONLINEAR 0x02 /* Fault was via a nonlinear mapping */
136
137
Nick Piggin54cb8822007-07-19 01:46:59 -0700138/*
Nick Piggind0217ac2007-07-19 01:47:03 -0700139 * vm_fault is filled by the the pagefault handler and passed to the vma's
Nick Piggin83c54072007-07-19 01:47:05 -0700140 * ->fault function. The vma's ->fault is responsible for returning a bitmask
141 * of VM_FAULT_xxx flags that give details about how the fault was handled.
Nick Piggin54cb8822007-07-19 01:46:59 -0700142 *
Nick Piggind0217ac2007-07-19 01:47:03 -0700143 * pgoff should be used in favour of virtual_address, if possible. If pgoff
144 * is used, one may set VM_CAN_NONLINEAR in the vma->vm_flags to get nonlinear
145 * mapping support.
Nick Piggin54cb8822007-07-19 01:46:59 -0700146 */
Nick Piggind0217ac2007-07-19 01:47:03 -0700147struct vm_fault {
148 unsigned int flags; /* FAULT_FLAG_xxx flags */
149 pgoff_t pgoff; /* Logical page offset based on vma */
150 void __user *virtual_address; /* Faulting virtual address */
151
152 struct page *page; /* ->fault handlers should return a
Nick Piggin83c54072007-07-19 01:47:05 -0700153 * page here, unless VM_FAULT_NOPAGE
Nick Piggind0217ac2007-07-19 01:47:03 -0700154 * is set (which is also implied by
Nick Piggin83c54072007-07-19 01:47:05 -0700155 * VM_FAULT_ERROR).
Nick Piggind0217ac2007-07-19 01:47:03 -0700156 */
Nick Piggin54cb8822007-07-19 01:46:59 -0700157};
Linus Torvalds1da177e2005-04-16 15:20:36 -0700158
159/*
160 * These are the virtual MM functions - opening of an area, closing and
161 * unmapping it (needed to keep files on disk up-to-date etc), pointer
162 * to the functions called when a no-page or a wp-page exception occurs.
163 */
164struct vm_operations_struct {
165 void (*open)(struct vm_area_struct * area);
166 void (*close)(struct vm_area_struct * area);
Nick Piggind0217ac2007-07-19 01:47:03 -0700167 int (*fault)(struct vm_area_struct *vma, struct vm_fault *vmf);
Nick Piggin54cb8822007-07-19 01:46:59 -0700168 unsigned long (*nopfn)(struct vm_area_struct *area,
169 unsigned long address);
David Howells9637a5e2006-06-23 02:03:43 -0700170
171 /* notification that a previously read-only page is about to become
172 * writable, if an error is returned it will cause a SIGBUS */
173 int (*page_mkwrite)(struct vm_area_struct *vma, struct page *page);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700174#ifdef CONFIG_NUMA
Lee Schermerhorna6020ed2008-04-28 02:13:14 -0700175 /*
176 * set_policy() op must add a reference to any non-NULL @new mempolicy
177 * to hold the policy upon return. Caller should pass NULL @new to
178 * remove a policy and fall back to surrounding context--i.e. do not
179 * install a MPOL_DEFAULT policy, nor the task or system default
180 * mempolicy.
181 */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700182 int (*set_policy)(struct vm_area_struct *vma, struct mempolicy *new);
Lee Schermerhorna6020ed2008-04-28 02:13:14 -0700183
184 /*
185 * get_policy() op must add reference [mpol_get()] to any policy at
186 * (vma,addr) marked as MPOL_SHARED. The shared policy infrastructure
187 * in mm/mempolicy.c will do this automatically.
188 * get_policy() must NOT add a ref if the policy at (vma,addr) is not
189 * marked as MPOL_SHARED. vma policies are protected by the mmap_sem.
190 * If no [shared/vma] mempolicy exists at the addr, get_policy() op
191 * must return NULL--i.e., do not "fallback" to task or system default
192 * policy.
193 */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700194 struct mempolicy *(*get_policy)(struct vm_area_struct *vma,
195 unsigned long addr);
Christoph Lameter7b2259b2006-06-25 05:46:48 -0700196 int (*migrate)(struct vm_area_struct *vma, const nodemask_t *from,
197 const nodemask_t *to, unsigned long flags);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700198#endif
199};
200
201struct mmu_gather;
202struct inode;
203
Andrew Morton349aef02006-01-08 01:04:36 -0800204#define page_private(page) ((page)->private)
205#define set_page_private(page, v) ((page)->private = (v))
Hugh Dickins4c21e2f2005-10-29 18:16:40 -0700206
Linus Torvalds1da177e2005-04-16 15:20:36 -0700207/*
208 * FIXME: take this include out, include page-flags.h in
209 * files which need it (119 of them)
210 */
211#include <linux/page-flags.h>
212
Nick Piggin725d7042006-09-25 23:30:55 -0700213#ifdef CONFIG_DEBUG_VM
214#define VM_BUG_ON(cond) BUG_ON(cond)
215#else
216#define VM_BUG_ON(condition) do { } while(0)
217#endif
218
Linus Torvalds1da177e2005-04-16 15:20:36 -0700219/*
220 * Methods to modify the page usage count.
221 *
222 * What counts for a page usage:
223 * - cache mapping (page->mapping)
224 * - private data (page->private)
225 * - page mapped in a task's page tables, each mapping
226 * is counted separately
227 *
228 * Also, many kernel routines increase the page count before a critical
229 * routine so they can be sure the page doesn't go away from under them.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700230 */
231
232/*
Nick Pigginda6052f2006-09-25 23:31:35 -0700233 * Drop a ref, return true if the refcount fell to zero (the page has no users)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700234 */
Nick Piggin7c8ee9a2006-03-22 00:08:03 -0800235static inline int put_page_testzero(struct page *page)
236{
Nick Piggin725d7042006-09-25 23:30:55 -0700237 VM_BUG_ON(atomic_read(&page->_count) == 0);
Nick Piggin8dc04ef2006-03-22 00:08:03 -0800238 return atomic_dec_and_test(&page->_count);
Nick Piggin7c8ee9a2006-03-22 00:08:03 -0800239}
Linus Torvalds1da177e2005-04-16 15:20:36 -0700240
241/*
Nick Piggin7c8ee9a2006-03-22 00:08:03 -0800242 * Try to grab a ref unless the page has a refcount of zero, return false if
243 * that is the case.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700244 */
Nick Piggin7c8ee9a2006-03-22 00:08:03 -0800245static inline int get_page_unless_zero(struct page *page)
246{
Christoph Lameteraec2c3e2008-02-04 22:28:35 -0800247 VM_BUG_ON(PageTail(page));
Nick Piggin8dc04ef2006-03-22 00:08:03 -0800248 return atomic_inc_not_zero(&page->_count);
Nick Piggin7c8ee9a2006-03-22 00:08:03 -0800249}
Linus Torvalds1da177e2005-04-16 15:20:36 -0700250
Christoph Lameter48667e72008-02-04 22:28:31 -0800251/* Support for virtually mapped pages */
Christoph Lameterb3bdda02008-02-04 22:28:32 -0800252struct page *vmalloc_to_page(const void *addr);
253unsigned long vmalloc_to_pfn(const void *addr);
Christoph Lameter48667e72008-02-04 22:28:31 -0800254
Paul Mundt0738c4b2008-03-12 16:51:31 +0900255/*
256 * Determine if an address is within the vmalloc range
257 *
258 * On nommu, vmalloc/vfree wrap through kmalloc/kfree directly, so there
259 * is no special casing required.
260 */
Christoph Lameter9e2779f2008-02-04 22:28:34 -0800261static inline int is_vmalloc_addr(const void *x)
262{
Paul Mundt0738c4b2008-03-12 16:51:31 +0900263#ifdef CONFIG_MMU
Christoph Lameter9e2779f2008-02-04 22:28:34 -0800264 unsigned long addr = (unsigned long)x;
265
266 return addr >= VMALLOC_START && addr < VMALLOC_END;
Paul Mundt0738c4b2008-03-12 16:51:31 +0900267#else
268 return 0;
David Howells8ca3ed82008-02-23 15:23:37 -0800269#endif
Paul Mundt0738c4b2008-03-12 16:51:31 +0900270}
Christoph Lameter9e2779f2008-02-04 22:28:34 -0800271
Christoph Lameterd85f3382007-05-06 14:49:39 -0700272static inline struct page *compound_head(struct page *page)
273{
Christoph Lameter6d777952007-05-06 14:49:40 -0700274 if (unlikely(PageTail(page)))
Christoph Lameterd85f3382007-05-06 14:49:39 -0700275 return page->first_page;
276 return page;
277}
278
Hugh Dickins4c21e2f2005-10-29 18:16:40 -0700279static inline int page_count(struct page *page)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700280{
Christoph Lameterd85f3382007-05-06 14:49:39 -0700281 return atomic_read(&compound_head(page)->_count);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700282}
283
284static inline void get_page(struct page *page)
285{
Christoph Lameterd85f3382007-05-06 14:49:39 -0700286 page = compound_head(page);
Nick Piggin725d7042006-09-25 23:30:55 -0700287 VM_BUG_ON(atomic_read(&page->_count) == 0);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700288 atomic_inc(&page->_count);
289}
290
Christoph Lameterb49af682007-05-06 14:49:41 -0700291static inline struct page *virt_to_head_page(const void *x)
292{
293 struct page *page = virt_to_page(x);
294 return compound_head(page);
295}
296
Nick Piggin7835e982006-03-22 00:08:40 -0800297/*
298 * Setup the page count before being freed into the page allocator for
299 * the first time (boot or memory hotplug)
300 */
301static inline void init_page_count(struct page *page)
302{
303 atomic_set(&page->_count, 1);
304}
305
Linus Torvalds1da177e2005-04-16 15:20:36 -0700306void put_page(struct page *page);
Alexander Zarochentsev1d7ea732006-08-13 23:24:27 -0700307void put_pages_list(struct list_head *pages);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700308
Nick Piggin8dfcc9b2006-03-22 00:08:05 -0800309void split_page(struct page *page, unsigned int order);
Nick Piggin8dfcc9b2006-03-22 00:08:05 -0800310
Linus Torvalds1da177e2005-04-16 15:20:36 -0700311/*
Andy Whitcroft33f2ef82006-12-06 20:33:32 -0800312 * Compound pages have a destructor function. Provide a
313 * prototype for that function and accessor functions.
314 * These are _only_ valid on the head of a PG_compound page.
315 */
316typedef void compound_page_dtor(struct page *);
317
318static inline void set_compound_page_dtor(struct page *page,
319 compound_page_dtor *dtor)
320{
321 page[1].lru.next = (void *)dtor;
322}
323
324static inline compound_page_dtor *get_compound_page_dtor(struct page *page)
325{
326 return (compound_page_dtor *)page[1].lru.next;
327}
328
Christoph Lameterd85f3382007-05-06 14:49:39 -0700329static inline int compound_order(struct page *page)
330{
Christoph Lameter6d777952007-05-06 14:49:40 -0700331 if (!PageHead(page))
Christoph Lameterd85f3382007-05-06 14:49:39 -0700332 return 0;
333 return (unsigned long)page[1].lru.prev;
334}
335
336static inline void set_compound_order(struct page *page, unsigned long order)
337{
338 page[1].lru.prev = (void *)order;
339}
340
Andy Whitcroft33f2ef82006-12-06 20:33:32 -0800341/*
Linus Torvalds1da177e2005-04-16 15:20:36 -0700342 * Multiple processes may "see" the same page. E.g. for untouched
343 * mappings of /dev/null, all processes see the same page full of
344 * zeroes, and text pages of executables and shared libraries have
345 * only one copy in memory, at most, normally.
346 *
347 * For the non-reserved pages, page_count(page) denotes a reference count.
Paolo 'Blaisorblade' Giarrusso7e871b62005-09-21 09:55:38 -0700348 * page_count() == 0 means the page is free. page->lru is then used for
349 * freelist management in the buddy allocator.
Nick Pigginda6052f2006-09-25 23:31:35 -0700350 * page_count() > 0 means the page has been allocated.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700351 *
Nick Pigginda6052f2006-09-25 23:31:35 -0700352 * Pages are allocated by the slab allocator in order to provide memory
353 * to kmalloc and kmem_cache_alloc. In this case, the management of the
354 * page, and the fields in 'struct page' are the responsibility of mm/slab.c
355 * unless a particular usage is carefully commented. (the responsibility of
356 * freeing the kmalloc memory is the caller's, of course).
Linus Torvalds1da177e2005-04-16 15:20:36 -0700357 *
Nick Pigginda6052f2006-09-25 23:31:35 -0700358 * A page may be used by anyone else who does a __get_free_page().
359 * In this case, page_count still tracks the references, and should only
360 * be used through the normal accessor functions. The top bits of page->flags
361 * and page->virtual store page management information, but all other fields
362 * are unused and could be used privately, carefully. The management of this
363 * page is the responsibility of the one who allocated it, and those who have
364 * subsequently been given references to it.
365 *
366 * The other pages (we may call them "pagecache pages") are completely
Linus Torvalds1da177e2005-04-16 15:20:36 -0700367 * managed by the Linux memory manager: I/O, buffers, swapping etc.
368 * The following discussion applies only to them.
369 *
Nick Pigginda6052f2006-09-25 23:31:35 -0700370 * A pagecache page contains an opaque `private' member, which belongs to the
371 * page's address_space. Usually, this is the address of a circular list of
372 * the page's disk buffers. PG_private must be set to tell the VM to call
373 * into the filesystem to release these pages.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700374 *
Nick Pigginda6052f2006-09-25 23:31:35 -0700375 * A page may belong to an inode's memory mapping. In this case, page->mapping
376 * is the pointer to the inode, and page->index is the file offset of the page,
377 * in units of PAGE_CACHE_SIZE.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700378 *
Nick Pigginda6052f2006-09-25 23:31:35 -0700379 * If pagecache pages are not associated with an inode, they are said to be
380 * anonymous pages. These may become associated with the swapcache, and in that
381 * case PG_swapcache is set, and page->private is an offset into the swapcache.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700382 *
Nick Pigginda6052f2006-09-25 23:31:35 -0700383 * In either case (swapcache or inode backed), the pagecache itself holds one
384 * reference to the page. Setting PG_private should also increment the
385 * refcount. The each user mapping also has a reference to the page.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700386 *
Nick Pigginda6052f2006-09-25 23:31:35 -0700387 * The pagecache pages are stored in a per-mapping radix tree, which is
388 * rooted at mapping->page_tree, and indexed by offset.
389 * Where 2.4 and early 2.6 kernels kept dirty/clean pages in per-address_space
390 * lists, we instead now tag pages as dirty/writeback in the radix tree.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700391 *
Nick Pigginda6052f2006-09-25 23:31:35 -0700392 * All pagecache pages may be subject to I/O:
Linus Torvalds1da177e2005-04-16 15:20:36 -0700393 * - inode pages may need to be read from disk,
394 * - inode pages which have been modified and are MAP_SHARED may need
Nick Pigginda6052f2006-09-25 23:31:35 -0700395 * to be written back to the inode on disk,
396 * - anonymous pages (including MAP_PRIVATE file mappings) which have been
397 * modified may need to be swapped out to swap space and (later) to be read
398 * back into memory.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700399 */
400
401/*
402 * The zone field is never updated after free_area_init_core()
403 * sets it, so none of the operations on it need to be atomic.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700404 */
Dave Hansen348f8b62005-06-23 00:07:40 -0700405
Andy Whitcroftd41dee32005-06-23 00:07:54 -0700406
407/*
408 * page->flags layout:
409 *
410 * There are three possibilities for how page->flags get
411 * laid out. The first is for the normal case, without
412 * sparsemem. The second is for sparsemem when there is
413 * plenty of space for node and section. The last is when
414 * we have run out of space and have to fall back to an
415 * alternate (slower) way of determining the node.
416 *
Christoph Lameter308c05e2008-04-28 02:12:43 -0700417 * No sparsemem or sparsemem vmemmap: | NODE | ZONE | ... | FLAGS |
418 * classic sparse with space for node:| SECTION | NODE | ZONE | ... | FLAGS |
419 * classic sparse no space for node: | SECTION | ZONE | ... | FLAGS |
Andy Whitcroftd41dee32005-06-23 00:07:54 -0700420 */
Christoph Lameter308c05e2008-04-28 02:12:43 -0700421#if defined(CONFIG_SPARSEMEM) && !defined(CONFIG_SPARSEMEM_VMEMMAP)
Andy Whitcroftd41dee32005-06-23 00:07:54 -0700422#define SECTIONS_WIDTH SECTIONS_SHIFT
423#else
424#define SECTIONS_WIDTH 0
425#endif
426
427#define ZONES_WIDTH ZONES_SHIFT
428
Christoph Lameter9223b412008-04-28 02:12:48 -0700429#if SECTIONS_WIDTH+ZONES_WIDTH+NODES_SHIFT <= BITS_PER_LONG - NR_PAGEFLAGS
Andy Whitcroftd41dee32005-06-23 00:07:54 -0700430#define NODES_WIDTH NODES_SHIFT
431#else
Christoph Lameter308c05e2008-04-28 02:12:43 -0700432#ifdef CONFIG_SPARSEMEM_VMEMMAP
433#error "Vmemmap: No space for nodes field in page flags"
434#endif
Andy Whitcroftd41dee32005-06-23 00:07:54 -0700435#define NODES_WIDTH 0
436#endif
437
438/* Page flags: | [SECTION] | [NODE] | ZONE | ... | FLAGS | */
Andi Kleen07808b72005-11-05 17:25:53 +0100439#define SECTIONS_PGOFF ((sizeof(unsigned long)*8) - SECTIONS_WIDTH)
Andy Whitcroftd41dee32005-06-23 00:07:54 -0700440#define NODES_PGOFF (SECTIONS_PGOFF - NODES_WIDTH)
441#define ZONES_PGOFF (NODES_PGOFF - ZONES_WIDTH)
442
443/*
444 * We are going to use the flags for the page to node mapping if its in
445 * there. This includes the case where there is no node, so it is implicit.
446 */
Christoph Lameter89689ae2006-12-06 20:31:45 -0800447#if !(NODES_WIDTH > 0 || NODES_SHIFT == 0)
448#define NODE_NOT_IN_PAGE_FLAGS
449#endif
Andy Whitcroftd41dee32005-06-23 00:07:54 -0700450
451#ifndef PFN_SECTION_SHIFT
452#define PFN_SECTION_SHIFT 0
453#endif
Dave Hansen348f8b62005-06-23 00:07:40 -0700454
455/*
456 * Define the bit shifts to access each section. For non-existant
457 * sections we define the shift as 0; that plus a 0 mask ensures
458 * the compiler will optimise away reference to them.
459 */
Andy Whitcroftd41dee32005-06-23 00:07:54 -0700460#define SECTIONS_PGSHIFT (SECTIONS_PGOFF * (SECTIONS_WIDTH != 0))
461#define NODES_PGSHIFT (NODES_PGOFF * (NODES_WIDTH != 0))
462#define ZONES_PGSHIFT (ZONES_PGOFF * (ZONES_WIDTH != 0))
Dave Hansen348f8b62005-06-23 00:07:40 -0700463
Christoph Lameter89689ae2006-12-06 20:31:45 -0800464/* NODE:ZONE or SECTION:ZONE is used to ID a zone for the buddy allcator */
465#ifdef NODE_NOT_IN_PAGEFLAGS
466#define ZONEID_SHIFT (SECTIONS_SHIFT + ZONES_SHIFT)
Andy Whitcroftbd8029b2007-02-10 01:43:14 -0800467#define ZONEID_PGOFF ((SECTIONS_PGOFF < ZONES_PGOFF)? \
468 SECTIONS_PGOFF : ZONES_PGOFF)
Andy Whitcroftd41dee32005-06-23 00:07:54 -0700469#else
Christoph Lameter89689ae2006-12-06 20:31:45 -0800470#define ZONEID_SHIFT (NODES_SHIFT + ZONES_SHIFT)
Andy Whitcroftbd8029b2007-02-10 01:43:14 -0800471#define ZONEID_PGOFF ((NODES_PGOFF < ZONES_PGOFF)? \
472 NODES_PGOFF : ZONES_PGOFF)
Andy Whitcroftd41dee32005-06-23 00:07:54 -0700473#endif
Christoph Lameter89689ae2006-12-06 20:31:45 -0800474
Andy Whitcroftbd8029b2007-02-10 01:43:14 -0800475#define ZONEID_PGSHIFT (ZONEID_PGOFF * (ZONEID_SHIFT != 0))
Dave Hansen348f8b62005-06-23 00:07:40 -0700476
Christoph Lameter9223b412008-04-28 02:12:48 -0700477#if SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > BITS_PER_LONG - NR_PAGEFLAGS
478#error SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > BITS_PER_LONG - NR_PAGEFLAGS
Dave Hansen348f8b62005-06-23 00:07:40 -0700479#endif
480
Andy Whitcroftd41dee32005-06-23 00:07:54 -0700481#define ZONES_MASK ((1UL << ZONES_WIDTH) - 1)
482#define NODES_MASK ((1UL << NODES_WIDTH) - 1)
483#define SECTIONS_MASK ((1UL << SECTIONS_WIDTH) - 1)
Christoph Lameter89689ae2006-12-06 20:31:45 -0800484#define ZONEID_MASK ((1UL << ZONEID_SHIFT) - 1)
Dave Hansen348f8b62005-06-23 00:07:40 -0700485
Christoph Lameter2f1b6242006-09-25 23:31:13 -0700486static inline enum zone_type page_zonenum(struct page *page)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700487{
Dave Hansen348f8b62005-06-23 00:07:40 -0700488 return (page->flags >> ZONES_PGSHIFT) & ZONES_MASK;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700489}
Linus Torvalds1da177e2005-04-16 15:20:36 -0700490
Christoph Lameter89689ae2006-12-06 20:31:45 -0800491/*
492 * The identification function is only used by the buddy allocator for
493 * determining if two pages could be buddies. We are not really
494 * identifying a zone since we could be using a the section number
495 * id if we have not node id available in page flags.
496 * We guarantee only that it will return the same value for two
497 * combinable pages in a zone.
498 */
Andy Whitcroftcb2b95e2006-06-23 02:03:01 -0700499static inline int page_zone_id(struct page *page)
500{
Christoph Lameter89689ae2006-12-06 20:31:45 -0800501 return (page->flags >> ZONEID_PGSHIFT) & ZONEID_MASK;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700502}
503
Andy Whitcroft25ba77c2006-12-06 20:33:03 -0800504static inline int zone_to_nid(struct zone *zone)
Christoph Lameter89fa3022006-09-25 23:31:55 -0700505{
Christoph Lameterd5f541e2006-09-27 01:50:08 -0700506#ifdef CONFIG_NUMA
507 return zone->node;
508#else
509 return 0;
510#endif
Christoph Lameter89fa3022006-09-25 23:31:55 -0700511}
512
Christoph Lameter89689ae2006-12-06 20:31:45 -0800513#ifdef NODE_NOT_IN_PAGE_FLAGS
Andy Whitcroft25ba77c2006-12-06 20:33:03 -0800514extern int page_to_nid(struct page *page);
Christoph Lameter89689ae2006-12-06 20:31:45 -0800515#else
Andy Whitcroft25ba77c2006-12-06 20:33:03 -0800516static inline int page_to_nid(struct page *page)
Andy Whitcroftd41dee32005-06-23 00:07:54 -0700517{
Christoph Lameter89689ae2006-12-06 20:31:45 -0800518 return (page->flags >> NODES_PGSHIFT) & NODES_MASK;
Andy Whitcroftd41dee32005-06-23 00:07:54 -0700519}
Christoph Lameter89689ae2006-12-06 20:31:45 -0800520#endif
521
522static inline struct zone *page_zone(struct page *page)
523{
524 return &NODE_DATA(page_to_nid(page))->node_zones[page_zonenum(page)];
525}
526
Christoph Lameter308c05e2008-04-28 02:12:43 -0700527#if defined(CONFIG_SPARSEMEM) && !defined(CONFIG_SPARSEMEM_VMEMMAP)
Andy Whitcroftd41dee32005-06-23 00:07:54 -0700528static inline unsigned long page_to_section(struct page *page)
529{
530 return (page->flags >> SECTIONS_PGSHIFT) & SECTIONS_MASK;
531}
Christoph Lameter308c05e2008-04-28 02:12:43 -0700532#endif
Andy Whitcroftd41dee32005-06-23 00:07:54 -0700533
Christoph Lameter2f1b6242006-09-25 23:31:13 -0700534static inline void set_page_zone(struct page *page, enum zone_type zone)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700535{
Dave Hansen348f8b62005-06-23 00:07:40 -0700536 page->flags &= ~(ZONES_MASK << ZONES_PGSHIFT);
537 page->flags |= (zone & ZONES_MASK) << ZONES_PGSHIFT;
538}
Christoph Lameter2f1b6242006-09-25 23:31:13 -0700539
Dave Hansen348f8b62005-06-23 00:07:40 -0700540static inline void set_page_node(struct page *page, unsigned long node)
541{
542 page->flags &= ~(NODES_MASK << NODES_PGSHIFT);
543 page->flags |= (node & NODES_MASK) << NODES_PGSHIFT;
544}
Christoph Lameter89689ae2006-12-06 20:31:45 -0800545
Andy Whitcroftd41dee32005-06-23 00:07:54 -0700546static inline void set_page_section(struct page *page, unsigned long section)
547{
548 page->flags &= ~(SECTIONS_MASK << SECTIONS_PGSHIFT);
549 page->flags |= (section & SECTIONS_MASK) << SECTIONS_PGSHIFT;
550}
Dave Hansen348f8b62005-06-23 00:07:40 -0700551
Christoph Lameter2f1b6242006-09-25 23:31:13 -0700552static inline void set_page_links(struct page *page, enum zone_type zone,
Andy Whitcroftd41dee32005-06-23 00:07:54 -0700553 unsigned long node, unsigned long pfn)
Dave Hansen348f8b62005-06-23 00:07:40 -0700554{
555 set_page_zone(page, zone);
556 set_page_node(page, node);
Andy Whitcroftd41dee32005-06-23 00:07:54 -0700557 set_page_section(page, pfn_to_section_nr(pfn));
Linus Torvalds1da177e2005-04-16 15:20:36 -0700558}
559
Christoph Lameterf6ac2352006-06-30 01:55:32 -0700560/*
Eric Paris7cd94142007-11-26 18:47:40 -0500561 * If a hint addr is less than mmap_min_addr change hint to be as
562 * low as possible but still greater than mmap_min_addr
563 */
564static inline unsigned long round_hint_to_min(unsigned long hint)
565{
566#ifdef CONFIG_SECURITY
567 hint &= PAGE_MASK;
568 if (((void *)hint != NULL) &&
569 (hint < mmap_min_addr))
570 return PAGE_ALIGN(mmap_min_addr);
571#endif
572 return hint;
573}
574
575/*
Christoph Lameterf6ac2352006-06-30 01:55:32 -0700576 * Some inline functions in vmstat.h depend on page_zone()
577 */
578#include <linux/vmstat.h>
579
Ingo Molnar652050a2006-01-14 13:21:30 -0800580static __always_inline void *lowmem_page_address(struct page *page)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700581{
582 return __va(page_to_pfn(page) << PAGE_SHIFT);
583}
584
585#if defined(CONFIG_HIGHMEM) && !defined(WANT_PAGE_VIRTUAL)
586#define HASHED_PAGE_VIRTUAL
587#endif
588
589#if defined(WANT_PAGE_VIRTUAL)
590#define page_address(page) ((page)->virtual)
591#define set_page_address(page, address) \
592 do { \
593 (page)->virtual = (address); \
594 } while(0)
595#define page_address_init() do { } while(0)
596#endif
597
598#if defined(HASHED_PAGE_VIRTUAL)
599void *page_address(struct page *page);
600void set_page_address(struct page *page, void *virtual);
601void page_address_init(void);
602#endif
603
604#if !defined(HASHED_PAGE_VIRTUAL) && !defined(WANT_PAGE_VIRTUAL)
605#define page_address(page) lowmem_page_address(page)
606#define set_page_address(page, address) do { } while(0)
607#define page_address_init() do { } while(0)
608#endif
609
610/*
611 * On an anonymous page mapped into a user virtual memory area,
612 * page->mapping points to its anon_vma, not to a struct address_space;
613 * with the PAGE_MAPPING_ANON bit set to distinguish it.
614 *
615 * Please note that, confusingly, "page_mapping" refers to the inode
616 * address_space which maps the page from disk; whereas "page_mapped"
617 * refers to user virtual address space into which the page is mapped.
618 */
619#define PAGE_MAPPING_ANON 1
620
621extern struct address_space swapper_space;
622static inline struct address_space *page_mapping(struct page *page)
623{
624 struct address_space *mapping = page->mapping;
625
Christoph Lameterb5fab14e2007-07-17 04:03:33 -0700626 VM_BUG_ON(PageSlab(page));
Andrew Morton726b8012008-04-28 02:12:44 -0700627#ifdef CONFIG_SWAP
Linus Torvalds1da177e2005-04-16 15:20:36 -0700628 if (unlikely(PageSwapCache(page)))
629 mapping = &swapper_space;
Andrew Morton726b8012008-04-28 02:12:44 -0700630 else
631#endif
632 if (unlikely((unsigned long)mapping & PAGE_MAPPING_ANON))
Linus Torvalds1da177e2005-04-16 15:20:36 -0700633 mapping = NULL;
634 return mapping;
635}
636
637static inline int PageAnon(struct page *page)
638{
639 return ((unsigned long)page->mapping & PAGE_MAPPING_ANON) != 0;
640}
641
642/*
643 * Return the pagecache index of the passed page. Regular pagecache pages
644 * use ->index whereas swapcache pages use ->private
645 */
646static inline pgoff_t page_index(struct page *page)
647{
648 if (unlikely(PageSwapCache(page)))
Hugh Dickins4c21e2f2005-10-29 18:16:40 -0700649 return page_private(page);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700650 return page->index;
651}
652
653/*
654 * The atomic page->_mapcount, like _count, starts from -1:
655 * so that transitions both from it and to it can be tracked,
656 * using atomic_inc_and_test and atomic_add_negative(-1).
657 */
658static inline void reset_page_mapcount(struct page *page)
659{
660 atomic_set(&(page)->_mapcount, -1);
661}
662
663static inline int page_mapcount(struct page *page)
664{
665 return atomic_read(&(page)->_mapcount) + 1;
666}
667
668/*
669 * Return true if this page is mapped into pagetables.
670 */
671static inline int page_mapped(struct page *page)
672{
673 return atomic_read(&(page)->_mapcount) >= 0;
674}
675
676/*
Jes Sorensenf4b81802006-09-27 01:50:10 -0700677 * Error return values for the *_nopfn functions
678 */
679#define NOPFN_SIGBUS ((unsigned long) -1)
680#define NOPFN_OOM ((unsigned long) -2)
Benjamin Herrenschmidt22cd25e2007-02-12 00:51:38 -0800681#define NOPFN_REFAULT ((unsigned long) -3)
Jes Sorensenf4b81802006-09-27 01:50:10 -0700682
683/*
Linus Torvalds1da177e2005-04-16 15:20:36 -0700684 * Different kinds of faults, as returned by handle_mm_fault().
685 * Used to decide whether a process gets delivered SIGBUS or
686 * just gets major/minor fault counters bumped up.
687 */
Nick Piggind0217ac2007-07-19 01:47:03 -0700688
Nick Piggin83c54072007-07-19 01:47:05 -0700689#define VM_FAULT_MINOR 0 /* For backwards compat. Remove me quickly. */
Nick Piggind0217ac2007-07-19 01:47:03 -0700690
Nick Piggin83c54072007-07-19 01:47:05 -0700691#define VM_FAULT_OOM 0x0001
692#define VM_FAULT_SIGBUS 0x0002
693#define VM_FAULT_MAJOR 0x0004
694#define VM_FAULT_WRITE 0x0008 /* Special case for get_user_pages */
Nick Pigginf33ea7f2005-08-03 20:24:01 +1000695
Nick Piggin83c54072007-07-19 01:47:05 -0700696#define VM_FAULT_NOPAGE 0x0100 /* ->fault installed the pte, not return page */
697#define VM_FAULT_LOCKED 0x0200 /* ->fault locked the returned page */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700698
Nick Piggin83c54072007-07-19 01:47:05 -0700699#define VM_FAULT_ERROR (VM_FAULT_OOM | VM_FAULT_SIGBUS)
Nick Piggind0217ac2007-07-19 01:47:03 -0700700
Linus Torvalds1da177e2005-04-16 15:20:36 -0700701#define offset_in_page(p) ((unsigned long)(p) & ~PAGE_MASK)
702
703extern void show_free_areas(void);
704
705#ifdef CONFIG_SHMEM
Linus Torvalds1da177e2005-04-16 15:20:36 -0700706int shmem_lock(struct file *file, int lock, struct user_struct *user);
707#else
Russell King03b00eb2006-01-06 00:10:52 -0800708static inline int shmem_lock(struct file *file, int lock,
709 struct user_struct *user)
710{
711 return 0;
712}
Linus Torvalds1da177e2005-04-16 15:20:36 -0700713#endif
714struct file *shmem_file_setup(char *name, loff_t size, unsigned long flags);
715
716int shmem_zero_setup(struct vm_area_struct *);
717
David Howellsb0e15192006-01-06 00:11:42 -0800718#ifndef CONFIG_MMU
719extern unsigned long shmem_get_unmapped_area(struct file *file,
720 unsigned long addr,
721 unsigned long len,
722 unsigned long pgoff,
723 unsigned long flags);
724#endif
725
Alexey Dobriyane8edc6e2007-05-21 01:22:52 +0400726extern int can_do_mlock(void);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700727extern int user_shm_lock(size_t, struct user_struct *);
728extern void user_shm_unlock(size_t, struct user_struct *);
729
730/*
731 * Parameter block passed down to zap_pte_range in exceptional cases.
732 */
733struct zap_details {
734 struct vm_area_struct *nonlinear_vma; /* Check page->index if set */
735 struct address_space *check_mapping; /* Check page->mapping if set */
736 pgoff_t first_index; /* Lowest page->index to unmap */
737 pgoff_t last_index; /* Highest page->index to unmap */
738 spinlock_t *i_mmap_lock; /* For unmap_mapping_range: */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700739 unsigned long truncate_count; /* Compare vm_truncate_count */
740};
741
Nick Piggin7e675132008-04-28 02:13:00 -0700742struct page *vm_normal_page(struct vm_area_struct *vma, unsigned long addr,
743 pte_t pte);
744
Hugh Dickinsee39b372005-04-19 13:29:15 -0700745unsigned long zap_page_range(struct vm_area_struct *vma, unsigned long address,
Linus Torvalds1da177e2005-04-16 15:20:36 -0700746 unsigned long size, struct zap_details *);
Hugh Dickins508034a2005-10-29 18:16:30 -0700747unsigned long unmap_vmas(struct mmu_gather **tlb,
Linus Torvalds1da177e2005-04-16 15:20:36 -0700748 struct vm_area_struct *start_vma, unsigned long start_addr,
749 unsigned long end_addr, unsigned long *nr_accounted,
750 struct zap_details *);
Matt Mackalle6473092008-02-04 22:29:01 -0800751
752/**
753 * mm_walk - callbacks for walk_page_range
754 * @pgd_entry: if set, called for each non-empty PGD (top-level) entry
755 * @pud_entry: if set, called for each non-empty PUD (2nd-level) entry
756 * @pmd_entry: if set, called for each non-empty PMD (3rd-level) entry
757 * @pte_entry: if set, called for each non-empty PTE (4th-level) entry
758 * @pte_hole: if set, called for each hole at all levels
759 *
760 * (see walk_page_range for more details)
761 */
762struct mm_walk {
763 int (*pgd_entry)(pgd_t *, unsigned long, unsigned long, void *);
764 int (*pud_entry)(pud_t *, unsigned long, unsigned long, void *);
765 int (*pmd_entry)(pmd_t *, unsigned long, unsigned long, void *);
766 int (*pte_entry)(pte_t *, unsigned long, unsigned long, void *);
767 int (*pte_hole)(unsigned long, unsigned long, void *);
768};
769
770int walk_page_range(const struct mm_struct *, unsigned long addr,
771 unsigned long end, const struct mm_walk *walk,
772 void *private);
Hugh Dickins3bf5ee92005-04-19 13:29:16 -0700773void free_pgd_range(struct mmu_gather **tlb, unsigned long addr,
774 unsigned long end, unsigned long floor, unsigned long ceiling);
775void free_pgtables(struct mmu_gather **tlb, struct vm_area_struct *start_vma,
Hugh Dickinse0da3822005-04-19 13:29:15 -0700776 unsigned long floor, unsigned long ceiling);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700777int copy_page_range(struct mm_struct *dst, struct mm_struct *src,
778 struct vm_area_struct *vma);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700779void unmap_mapping_range(struct address_space *mapping,
780 loff_t const holebegin, loff_t const holelen, int even_cows);
781
782static inline void unmap_shared_mapping_range(struct address_space *mapping,
783 loff_t const holebegin, loff_t const holelen)
784{
785 unmap_mapping_range(mapping, holebegin, holelen, 0);
786}
787
788extern int vmtruncate(struct inode * inode, loff_t offset);
Badari Pulavartyf6b3ec22006-01-06 00:10:38 -0800789extern int vmtruncate_range(struct inode * inode, loff_t offset, loff_t end);
Nick Pigginf33ea7f2005-08-03 20:24:01 +1000790
David Howells7ee1dd32006-01-06 00:11:44 -0800791#ifdef CONFIG_MMU
Nick Piggin83c54072007-07-19 01:47:05 -0700792extern int handle_mm_fault(struct mm_struct *mm, struct vm_area_struct *vma,
David Howells7ee1dd32006-01-06 00:11:44 -0800793 unsigned long address, int write_access);
David Howells7ee1dd32006-01-06 00:11:44 -0800794#else
795static inline int handle_mm_fault(struct mm_struct *mm,
796 struct vm_area_struct *vma, unsigned long address,
797 int write_access)
798{
799 /* should never happen if there's no MMU */
800 BUG();
801 return VM_FAULT_SIGBUS;
802}
803#endif
Nick Pigginf33ea7f2005-08-03 20:24:01 +1000804
Linus Torvalds1da177e2005-04-16 15:20:36 -0700805extern int make_pages_present(unsigned long addr, unsigned long end);
806extern int access_process_vm(struct task_struct *tsk, unsigned long addr, void *buf, int len, int write);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700807
808int get_user_pages(struct task_struct *tsk, struct mm_struct *mm, unsigned long start,
809 int len, int write, int force, struct page **pages, struct vm_area_struct **vmas);
Nick Pigginb5810032005-10-29 18:16:12 -0700810void print_bad_pte(struct vm_area_struct *, pte_t, unsigned long);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700811
David Howellscf9a2ae2006-08-29 19:05:54 +0100812extern int try_to_release_page(struct page * page, gfp_t gfp_mask);
813extern void do_invalidatepage(struct page *page, unsigned long offset);
814
Linus Torvalds1da177e2005-04-16 15:20:36 -0700815int __set_page_dirty_nobuffers(struct page *page);
Ken Chen76719322007-02-10 01:43:15 -0800816int __set_page_dirty_no_writeback(struct page *page);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700817int redirty_page_for_writepage(struct writeback_control *wbc,
818 struct page *page);
Harvey Harrisonb3c97522008-02-13 15:03:15 -0800819int set_page_dirty(struct page *page);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700820int set_page_dirty_lock(struct page *page);
821int clear_page_dirty_for_io(struct page *page);
822
Ollie Wildb6a2fea2007-07-19 01:48:16 -0700823extern unsigned long move_page_tables(struct vm_area_struct *vma,
824 unsigned long old_addr, struct vm_area_struct *new_vma,
825 unsigned long new_addr, unsigned long len);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700826extern unsigned long do_mremap(unsigned long addr,
827 unsigned long old_len, unsigned long new_len,
828 unsigned long flags, unsigned long new_addr);
Ollie Wildb6a2fea2007-07-19 01:48:16 -0700829extern int mprotect_fixup(struct vm_area_struct *vma,
830 struct vm_area_struct **pprev, unsigned long start,
831 unsigned long end, unsigned long newflags);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700832
833/*
Rusty Russell8e1f9362007-07-17 04:03:17 -0700834 * A callback you can register to apply pressure to ageable caches.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700835 *
Rusty Russell8e1f9362007-07-17 04:03:17 -0700836 * 'shrink' is passed a count 'nr_to_scan' and a 'gfpmask'. It should
837 * look through the least-recently-used 'nr_to_scan' entries and
838 * attempt to free them up. It should return the number of objects
839 * which remain in the cache. If it returns -1, it means it cannot do
840 * any scanning at this time (eg. there is a risk of deadlock).
Linus Torvalds1da177e2005-04-16 15:20:36 -0700841 *
Rusty Russell8e1f9362007-07-17 04:03:17 -0700842 * The 'gfpmask' refers to the allocation we are currently trying to
843 * fulfil.
844 *
845 * Note that 'shrink' will be passed nr_to_scan == 0 when the VM is
846 * querying the cache size, so a fastpath for that case is appropriate.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700847 */
Rusty Russell8e1f9362007-07-17 04:03:17 -0700848struct shrinker {
849 int (*shrink)(int nr_to_scan, gfp_t gfp_mask);
850 int seeks; /* seeks to recreate an obj */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700851
Rusty Russell8e1f9362007-07-17 04:03:17 -0700852 /* These are for internal use */
853 struct list_head list;
854 long nr; /* objs pending delete */
855};
856#define DEFAULT_SEEKS 2 /* A good number if you don't know better. */
857extern void register_shrinker(struct shrinker *);
858extern void unregister_shrinker(struct shrinker *);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700859
Alexey Dobriyan4e950f62007-07-30 02:36:13 +0400860int vma_wants_writenotify(struct vm_area_struct *vma);
Peter Zijlstrad08b3852006-09-25 23:30:57 -0700861
Harvey Harrisonb3c97522008-02-13 15:03:15 -0800862extern pte_t *get_locked_pte(struct mm_struct *mm, unsigned long addr, spinlock_t **ptl);
Linus Torvaldsc9cfcdd2005-11-29 14:03:14 -0800863
Nick Piggin5f22df02007-05-06 14:49:02 -0700864#ifdef __PAGETABLE_PUD_FOLDED
865static inline int __pud_alloc(struct mm_struct *mm, pgd_t *pgd,
866 unsigned long address)
867{
868 return 0;
869}
870#else
Hugh Dickins1bb36302005-10-29 18:16:22 -0700871int __pud_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address);
Nick Piggin5f22df02007-05-06 14:49:02 -0700872#endif
873
874#ifdef __PAGETABLE_PMD_FOLDED
875static inline int __pmd_alloc(struct mm_struct *mm, pud_t *pud,
876 unsigned long address)
877{
878 return 0;
879}
880#else
Hugh Dickins1bb36302005-10-29 18:16:22 -0700881int __pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address);
Nick Piggin5f22df02007-05-06 14:49:02 -0700882#endif
883
Hugh Dickins1bb36302005-10-29 18:16:22 -0700884int __pte_alloc(struct mm_struct *mm, pmd_t *pmd, unsigned long address);
885int __pte_alloc_kernel(pmd_t *pmd, unsigned long address);
886
Linus Torvalds1da177e2005-04-16 15:20:36 -0700887/*
888 * The following ifdef needed to get the 4level-fixup.h header to work.
889 * Remove it when 4level-fixup.h has been removed.
890 */
Hugh Dickins1bb36302005-10-29 18:16:22 -0700891#if defined(CONFIG_MMU) && !defined(__ARCH_HAS_4LEVEL_HACK)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700892static inline pud_t *pud_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address)
893{
Hugh Dickins1bb36302005-10-29 18:16:22 -0700894 return (unlikely(pgd_none(*pgd)) && __pud_alloc(mm, pgd, address))?
895 NULL: pud_offset(pgd, address);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700896}
897
898static inline pmd_t *pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address)
899{
Hugh Dickins1bb36302005-10-29 18:16:22 -0700900 return (unlikely(pud_none(*pud)) && __pmd_alloc(mm, pud, address))?
901 NULL: pmd_offset(pud, address);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700902}
Hugh Dickins1bb36302005-10-29 18:16:22 -0700903#endif /* CONFIG_MMU && !__ARCH_HAS_4LEVEL_HACK */
904
Hugh Dickins4c21e2f2005-10-29 18:16:40 -0700905#if NR_CPUS >= CONFIG_SPLIT_PTLOCK_CPUS
906/*
907 * We tuck a spinlock to guard each pagetable page into its struct page,
908 * at page->private, with BUILD_BUG_ON to make sure that this will not
909 * overflow into the next struct page (as it might with DEBUG_SPINLOCK).
910 * When freeing, reset page->mapping so free_pages_check won't complain.
911 */
Andrew Morton349aef02006-01-08 01:04:36 -0800912#define __pte_lockptr(page) &((page)->ptl)
Hugh Dickins4c21e2f2005-10-29 18:16:40 -0700913#define pte_lock_init(_page) do { \
914 spin_lock_init(__pte_lockptr(_page)); \
915} while (0)
916#define pte_lock_deinit(page) ((page)->mapping = NULL)
917#define pte_lockptr(mm, pmd) ({(void)(mm); __pte_lockptr(pmd_page(*(pmd)));})
918#else
919/*
920 * We use mm->page_table_lock to guard all pagetable pages of the mm.
921 */
922#define pte_lock_init(page) do {} while (0)
923#define pte_lock_deinit(page) do {} while (0)
924#define pte_lockptr(mm, pmd) ({(void)(pmd); &(mm)->page_table_lock;})
925#endif /* NR_CPUS < CONFIG_SPLIT_PTLOCK_CPUS */
926
Martin Schwidefsky2f569af2008-02-08 04:22:04 -0800927static inline void pgtable_page_ctor(struct page *page)
928{
929 pte_lock_init(page);
930 inc_zone_page_state(page, NR_PAGETABLE);
931}
932
933static inline void pgtable_page_dtor(struct page *page)
934{
935 pte_lock_deinit(page);
936 dec_zone_page_state(page, NR_PAGETABLE);
937}
938
Hugh Dickinsc74df322005-10-29 18:16:23 -0700939#define pte_offset_map_lock(mm, pmd, address, ptlp) \
940({ \
Hugh Dickins4c21e2f2005-10-29 18:16:40 -0700941 spinlock_t *__ptl = pte_lockptr(mm, pmd); \
Hugh Dickinsc74df322005-10-29 18:16:23 -0700942 pte_t *__pte = pte_offset_map(pmd, address); \
943 *(ptlp) = __ptl; \
944 spin_lock(__ptl); \
945 __pte; \
946})
947
948#define pte_unmap_unlock(pte, ptl) do { \
949 spin_unlock(ptl); \
950 pte_unmap(pte); \
951} while (0)
952
Hugh Dickins1bb36302005-10-29 18:16:22 -0700953#define pte_alloc_map(mm, pmd, address) \
954 ((unlikely(!pmd_present(*(pmd))) && __pte_alloc(mm, pmd, address))? \
955 NULL: pte_offset_map(pmd, address))
956
Hugh Dickinsc74df322005-10-29 18:16:23 -0700957#define pte_alloc_map_lock(mm, pmd, address, ptlp) \
958 ((unlikely(!pmd_present(*(pmd))) && __pte_alloc(mm, pmd, address))? \
959 NULL: pte_offset_map_lock(mm, pmd, address, ptlp))
960
Hugh Dickins1bb36302005-10-29 18:16:22 -0700961#define pte_alloc_kernel(pmd, address) \
962 ((unlikely(!pmd_present(*(pmd))) && __pte_alloc_kernel(pmd, address))? \
963 NULL: pte_offset_kernel(pmd, address))
Linus Torvalds1da177e2005-04-16 15:20:36 -0700964
965extern void free_area_init(unsigned long * zones_size);
966extern void free_area_init_node(int nid, pg_data_t *pgdat,
967 unsigned long * zones_size, unsigned long zone_start_pfn,
968 unsigned long *zholes_size);
Mel Gormanc7132162006-09-27 01:49:43 -0700969#ifdef CONFIG_ARCH_POPULATES_NODE_MAP
970/*
971 * With CONFIG_ARCH_POPULATES_NODE_MAP set, an architecture may initialise its
972 * zones, allocate the backing mem_map and account for memory holes in a more
973 * architecture independent manner. This is a substitute for creating the
974 * zone_sizes[] and zholes_size[] arrays and passing them to
975 * free_area_init_node()
976 *
977 * An architecture is expected to register range of page frames backed by
978 * physical memory with add_active_range() before calling
979 * free_area_init_nodes() passing in the PFN each zone ends at. At a basic
980 * usage, an architecture is expected to do something like
981 *
982 * unsigned long max_zone_pfns[MAX_NR_ZONES] = {max_dma, max_normal_pfn,
983 * max_highmem_pfn};
984 * for_each_valid_physical_page_range()
985 * add_active_range(node_id, start_pfn, end_pfn)
986 * free_area_init_nodes(max_zone_pfns);
987 *
988 * If the architecture guarantees that there are no holes in the ranges
989 * registered with add_active_range(), free_bootmem_active_regions()
990 * will call free_bootmem_node() for each registered physical page range.
991 * Similarly sparse_memory_present_with_active_regions() calls
992 * memory_present() for each range when SPARSEMEM is enabled.
993 *
994 * See mm/page_alloc.c for more information on each function exposed by
995 * CONFIG_ARCH_POPULATES_NODE_MAP
996 */
997extern void free_area_init_nodes(unsigned long *max_zone_pfn);
998extern void add_active_range(unsigned int nid, unsigned long start_pfn,
999 unsigned long end_pfn);
1000extern void shrink_active_range(unsigned int nid, unsigned long old_end_pfn,
1001 unsigned long new_end_pfn);
Mel Gormanfb014392006-09-27 01:49:59 -07001002extern void push_node_boundaries(unsigned int nid, unsigned long start_pfn,
1003 unsigned long end_pfn);
Mel Gormanc7132162006-09-27 01:49:43 -07001004extern void remove_all_active_ranges(void);
1005extern unsigned long absent_pages_in_range(unsigned long start_pfn,
1006 unsigned long end_pfn);
1007extern void get_pfn_range_for_nid(unsigned int nid,
1008 unsigned long *start_pfn, unsigned long *end_pfn);
1009extern unsigned long find_min_pfn_with_active_regions(void);
1010extern unsigned long find_max_pfn_with_active_regions(void);
1011extern void free_bootmem_with_active_regions(int nid,
1012 unsigned long max_low_pfn);
1013extern void sparse_memory_present_with_active_regions(int nid);
1014#ifndef CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID
1015extern int early_pfn_to_nid(unsigned long pfn);
1016#endif /* CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID */
1017#endif /* CONFIG_ARCH_POPULATES_NODE_MAP */
Mel Gorman0e0b8642006-09-27 01:49:56 -07001018extern void set_dma_reserve(unsigned long new_dma_reserve);
Dave Hansena2f3aa022007-01-10 23:15:30 -08001019extern void memmap_init_zone(unsigned long, int, unsigned long,
1020 unsigned long, enum memmap_context);
Dave Hansen3947be12005-10-29 18:16:54 -07001021extern void setup_per_zone_pages_min(void);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001022extern void mem_init(void);
1023extern void show_mem(void);
1024extern void si_meminfo(struct sysinfo * val);
1025extern void si_meminfo_node(struct sysinfo *val, int nid);
1026
Christoph Lametere7c8d5c2005-06-21 17:14:47 -07001027#ifdef CONFIG_NUMA
1028extern void setup_per_cpu_pageset(void);
1029#else
1030static inline void setup_per_cpu_pageset(void) {}
1031#endif
1032
Linus Torvalds1da177e2005-04-16 15:20:36 -07001033/* prio_tree.c */
1034void vma_prio_tree_add(struct vm_area_struct *, struct vm_area_struct *old);
1035void vma_prio_tree_insert(struct vm_area_struct *, struct prio_tree_root *);
1036void vma_prio_tree_remove(struct vm_area_struct *, struct prio_tree_root *);
1037struct vm_area_struct *vma_prio_tree_next(struct vm_area_struct *vma,
1038 struct prio_tree_iter *iter);
1039
1040#define vma_prio_tree_foreach(vma, iter, root, begin, end) \
1041 for (prio_tree_iter_init(iter, root, begin, end), vma = NULL; \
1042 (vma = vma_prio_tree_next(vma, iter)); )
1043
1044static inline void vma_nonlinear_insert(struct vm_area_struct *vma,
1045 struct list_head *list)
1046{
1047 vma->shared.vm_set.parent = NULL;
1048 list_add_tail(&vma->shared.vm_set.list, list);
1049}
1050
1051/* mmap.c */
Alan Cox34b4e4a2007-08-22 14:01:28 -07001052extern int __vm_enough_memory(struct mm_struct *mm, long pages, int cap_sys_admin);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001053extern void vma_adjust(struct vm_area_struct *vma, unsigned long start,
1054 unsigned long end, pgoff_t pgoff, struct vm_area_struct *insert);
1055extern struct vm_area_struct *vma_merge(struct mm_struct *,
1056 struct vm_area_struct *prev, unsigned long addr, unsigned long end,
1057 unsigned long vm_flags, struct anon_vma *, struct file *, pgoff_t,
1058 struct mempolicy *);
1059extern struct anon_vma *find_mergeable_anon_vma(struct vm_area_struct *);
1060extern int split_vma(struct mm_struct *,
1061 struct vm_area_struct *, unsigned long addr, int new_below);
1062extern int insert_vm_struct(struct mm_struct *, struct vm_area_struct *);
1063extern void __vma_link_rb(struct mm_struct *, struct vm_area_struct *,
1064 struct rb_node **, struct rb_node *);
Hugh Dickinsa8fb5612005-10-29 18:15:57 -07001065extern void unlink_file_vma(struct vm_area_struct *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001066extern struct vm_area_struct *copy_vma(struct vm_area_struct **,
1067 unsigned long addr, unsigned long len, pgoff_t pgoff);
1068extern void exit_mmap(struct mm_struct *);
Matt Helsley925d1c42008-04-29 01:01:36 -07001069
1070#ifdef CONFIG_PROC_FS
1071/* From fs/proc/base.c. callers must _not_ hold the mm's exe_file_lock */
1072extern void added_exe_file_vma(struct mm_struct *mm);
1073extern void removed_exe_file_vma(struct mm_struct *mm);
1074#else
1075static inline void added_exe_file_vma(struct mm_struct *mm)
1076{}
1077
1078static inline void removed_exe_file_vma(struct mm_struct *mm)
1079{}
1080#endif /* CONFIG_PROC_FS */
1081
akpm@osdl.org119f6572005-05-01 08:58:35 -07001082extern int may_expand_vm(struct mm_struct *mm, unsigned long npages);
Roland McGrathfa5dc222007-02-08 14:20:41 -08001083extern int install_special_mapping(struct mm_struct *mm,
1084 unsigned long addr, unsigned long len,
1085 unsigned long flags, struct page **pages);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001086
1087extern unsigned long get_unmapped_area(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
1088
1089extern unsigned long do_mmap_pgoff(struct file *file, unsigned long addr,
1090 unsigned long len, unsigned long prot,
1091 unsigned long flag, unsigned long pgoff);
Miklos Szeredi0165ab42007-07-15 23:38:26 -07001092extern unsigned long mmap_region(struct file *file, unsigned long addr,
1093 unsigned long len, unsigned long flags,
1094 unsigned int vm_flags, unsigned long pgoff,
1095 int accountable);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001096
1097static inline unsigned long do_mmap(struct file *file, unsigned long addr,
1098 unsigned long len, unsigned long prot,
1099 unsigned long flag, unsigned long offset)
1100{
1101 unsigned long ret = -EINVAL;
1102 if ((offset + PAGE_ALIGN(len)) < offset)
1103 goto out;
1104 if (!(offset & ~PAGE_MASK))
1105 ret = do_mmap_pgoff(file, addr, len, prot, flag, offset >> PAGE_SHIFT);
1106out:
1107 return ret;
1108}
1109
1110extern int do_munmap(struct mm_struct *, unsigned long, size_t);
1111
1112extern unsigned long do_brk(unsigned long, unsigned long);
1113
1114/* filemap.c */
1115extern unsigned long page_unuse(struct page *);
1116extern void truncate_inode_pages(struct address_space *, loff_t);
Hans Reiserd7339072006-01-06 00:10:36 -08001117extern void truncate_inode_pages_range(struct address_space *,
1118 loff_t lstart, loff_t lend);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001119
1120/* generic vm_area_ops exported for stackable file systems */
Nick Piggind0217ac2007-07-19 01:47:03 -07001121extern int filemap_fault(struct vm_area_struct *, struct vm_fault *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001122
1123/* mm/page-writeback.c */
1124int write_one_page(struct page *page, int wait);
1125
1126/* readahead.c */
1127#define VM_MAX_READAHEAD 128 /* kbytes */
1128#define VM_MIN_READAHEAD 16 /* kbytes (includes current page) */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001129
1130int do_page_cache_readahead(struct address_space *mapping, struct file *filp,
Andrew Morton7361f4d2005-11-07 00:59:28 -08001131 pgoff_t offset, unsigned long nr_to_read);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001132int force_page_cache_readahead(struct address_space *mapping, struct file *filp,
Andrew Morton7361f4d2005-11-07 00:59:28 -08001133 pgoff_t offset, unsigned long nr_to_read);
Rusty Russellcf914a72007-07-19 01:48:08 -07001134
1135void page_cache_sync_readahead(struct address_space *mapping,
1136 struct file_ra_state *ra,
1137 struct file *filp,
1138 pgoff_t offset,
1139 unsigned long size);
1140
1141void page_cache_async_readahead(struct address_space *mapping,
1142 struct file_ra_state *ra,
1143 struct file *filp,
1144 struct page *pg,
1145 pgoff_t offset,
1146 unsigned long size);
1147
Linus Torvalds1da177e2005-04-16 15:20:36 -07001148unsigned long max_sane_readahead(unsigned long nr);
1149
1150/* Do stack extension */
Hugh Dickins46dea3d2005-10-29 18:16:20 -07001151extern int expand_stack(struct vm_area_struct *vma, unsigned long address);
Matthew Wilcox9ab88512005-11-18 16:16:42 -05001152#ifdef CONFIG_IA64
Hugh Dickins46dea3d2005-10-29 18:16:20 -07001153extern int expand_upwards(struct vm_area_struct *vma, unsigned long address);
Matthew Wilcox9ab88512005-11-18 16:16:42 -05001154#endif
Ollie Wildb6a2fea2007-07-19 01:48:16 -07001155extern int expand_stack_downwards(struct vm_area_struct *vma,
1156 unsigned long address);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001157
1158/* Look up the first VMA which satisfies addr < vm_end, NULL if none. */
1159extern struct vm_area_struct * find_vma(struct mm_struct * mm, unsigned long addr);
1160extern struct vm_area_struct * find_vma_prev(struct mm_struct * mm, unsigned long addr,
1161 struct vm_area_struct **pprev);
1162
1163/* Look up the first VMA which intersects the interval start_addr..end_addr-1,
1164 NULL if none. Assume start_addr < end_addr. */
1165static inline struct vm_area_struct * find_vma_intersection(struct mm_struct * mm, unsigned long start_addr, unsigned long end_addr)
1166{
1167 struct vm_area_struct * vma = find_vma(mm,start_addr);
1168
1169 if (vma && end_addr <= vma->vm_start)
1170 vma = NULL;
1171 return vma;
1172}
1173
1174static inline unsigned long vma_pages(struct vm_area_struct *vma)
1175{
1176 return (vma->vm_end - vma->vm_start) >> PAGE_SHIFT;
1177}
1178
Hugh Dickins804af2c2006-07-26 21:39:49 +01001179pgprot_t vm_get_page_prot(unsigned long vm_flags);
Hugh Dickinsdeceb6c2005-10-29 18:16:33 -07001180struct vm_area_struct *find_extend_vma(struct mm_struct *, unsigned long addr);
Hugh Dickinsdeceb6c2005-10-29 18:16:33 -07001181int remap_pfn_range(struct vm_area_struct *, unsigned long addr,
1182 unsigned long pfn, unsigned long size, pgprot_t);
Linus Torvaldsa145dd42005-11-30 09:35:19 -08001183int vm_insert_page(struct vm_area_struct *, unsigned long addr, struct page *);
Nick Piggine0dc0d82007-02-12 00:51:36 -08001184int vm_insert_pfn(struct vm_area_struct *vma, unsigned long addr,
1185 unsigned long pfn);
Nick Piggin423bad62008-04-28 02:13:01 -07001186int vm_insert_mixed(struct vm_area_struct *vma, unsigned long addr,
1187 unsigned long pfn);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001188
Linus Torvalds6aab3412005-11-28 14:34:23 -08001189struct page *follow_page(struct vm_area_struct *, unsigned long address,
Hugh Dickinsdeceb6c2005-10-29 18:16:33 -07001190 unsigned int foll_flags);
1191#define FOLL_WRITE 0x01 /* check pte is writable */
1192#define FOLL_TOUCH 0x02 /* mark page accessed */
1193#define FOLL_GET 0x04 /* do get_page on page */
1194#define FOLL_ANON 0x08 /* give ZERO_PAGE if no pgtable */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001195
Martin Schwidefsky2f569af2008-02-08 04:22:04 -08001196typedef int (*pte_fn_t)(pte_t *pte, pgtable_t token, unsigned long addr,
Jeremy Fitzhardingeaee16b32007-05-06 14:48:54 -07001197 void *data);
1198extern int apply_to_page_range(struct mm_struct *mm, unsigned long address,
1199 unsigned long size, pte_fn_t fn, void *data);
1200
Linus Torvalds1da177e2005-04-16 15:20:36 -07001201#ifdef CONFIG_PROC_FS
Hugh Dickinsab50b8e2005-10-29 18:15:56 -07001202void vm_stat_account(struct mm_struct *, unsigned long, struct file *, long);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001203#else
Hugh Dickinsab50b8e2005-10-29 18:15:56 -07001204static inline void vm_stat_account(struct mm_struct *mm,
Linus Torvalds1da177e2005-04-16 15:20:36 -07001205 unsigned long flags, struct file *file, long pages)
1206{
1207}
1208#endif /* CONFIG_PROC_FS */
1209
Ingo Molnar12d6f212008-01-30 13:33:58 +01001210#ifdef CONFIG_DEBUG_PAGEALLOC
1211extern int debug_pagealloc_enabled;
1212
1213extern void kernel_map_pages(struct page *page, int numpages, int enable);
1214
1215static inline void enable_debug_pagealloc(void)
1216{
1217 debug_pagealloc_enabled = 1;
1218}
Rafael J. Wysocki8a235ef2008-02-20 01:47:44 +01001219#ifdef CONFIG_HIBERNATION
1220extern bool kernel_page_present(struct page *page);
1221#endif /* CONFIG_HIBERNATION */
Ingo Molnar12d6f212008-01-30 13:33:58 +01001222#else
Linus Torvalds1da177e2005-04-16 15:20:36 -07001223static inline void
Nick Piggin9858db52006-10-11 01:21:30 -07001224kernel_map_pages(struct page *page, int numpages, int enable) {}
Ingo Molnar12d6f212008-01-30 13:33:58 +01001225static inline void enable_debug_pagealloc(void)
1226{
1227}
Rafael J. Wysocki8a235ef2008-02-20 01:47:44 +01001228#ifdef CONFIG_HIBERNATION
1229static inline bool kernel_page_present(struct page *page) { return true; }
1230#endif /* CONFIG_HIBERNATION */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001231#endif
1232
1233extern struct vm_area_struct *get_gate_vma(struct task_struct *tsk);
1234#ifdef __HAVE_ARCH_GATE_AREA
1235int in_gate_area_no_task(unsigned long addr);
1236int in_gate_area(struct task_struct *task, unsigned long addr);
1237#else
1238int in_gate_area_no_task(unsigned long addr);
1239#define in_gate_area(task, addr) ({(void)task; in_gate_area_no_task(addr);})
1240#endif /* __HAVE_ARCH_GATE_AREA */
1241
Andrew Morton9d0243b2006-01-08 01:00:39 -08001242int drop_caches_sysctl_handler(struct ctl_table *, int, struct file *,
1243 void __user *, size_t *, loff_t *);
Andrew Morton69e05942006-03-22 00:08:19 -08001244unsigned long shrink_slab(unsigned long scanned, gfp_t gfp_mask,
Andrew Morton9d0243b2006-01-08 01:00:39 -08001245 unsigned long lru_pages);
Andrew Morton9d0243b2006-01-08 01:00:39 -08001246
Luke Yang7a9166e2006-02-20 18:28:07 -08001247#ifndef CONFIG_MMU
1248#define randomize_va_space 0
1249#else
Andi Kleena62eaf12006-02-16 23:41:58 +01001250extern int randomize_va_space;
Luke Yang7a9166e2006-02-20 18:28:07 -08001251#endif
Andi Kleena62eaf12006-02-16 23:41:58 +01001252
Sam Ravnborg045e72a2007-07-26 10:41:13 -07001253const char * arch_vma_name(struct vm_area_struct *vma);
Andi Kleen03252912008-01-30 13:33:18 +01001254void print_vma_addr(char *prefix, unsigned long rip);
Ingo Molnare6e54942006-06-27 02:53:50 -07001255
Yasunori Goto98f3cfc2007-10-16 01:26:14 -07001256struct page *sparse_mem_map_populate(unsigned long pnum, int nid);
Andy Whitcroft29c71112007-10-16 01:24:14 -07001257pgd_t *vmemmap_pgd_populate(unsigned long addr, int node);
1258pud_t *vmemmap_pud_populate(pgd_t *pgd, unsigned long addr, int node);
1259pmd_t *vmemmap_pmd_populate(pud_t *pud, unsigned long addr, int node);
1260pte_t *vmemmap_pte_populate(pmd_t *pmd, unsigned long addr, int node);
Christoph Lameter8f6aac42007-10-16 01:24:13 -07001261void *vmemmap_alloc_block(unsigned long size, int node);
1262void vmemmap_verify(pte_t *, int, unsigned long, unsigned long);
Andy Whitcroft29c71112007-10-16 01:24:14 -07001263int vmemmap_populate_basepages(struct page *start_page,
1264 unsigned long pages, int node);
1265int vmemmap_populate(struct page *start_page, unsigned long pages, int node);
Yinghai Luc2b91e22008-04-12 01:19:24 -07001266void vmemmap_populate_print_last(void);
Christoph Lameter8f6aac42007-10-16 01:24:13 -07001267
Linus Torvalds1da177e2005-04-16 15:20:36 -07001268#endif /* __KERNEL__ */
1269#endif /* _LINUX_MM_H */