blob: 2f6e2f886d4bf196bcefbaf97298681ab8949b30 [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
venkatesh.pallipadi@intel.com6bd9cd52008-12-19 13:47:26 -0800148/*
149 * This interface is used by x86 PAT code to identify a pfn mapping that is
150 * linear over entire vma. This is to optimize PAT code that deals with
151 * marking the physical region with a particular prot. This is not for generic
152 * mm use. Note also that this check will not work if the pfn mapping is
153 * linear for a vma starting at physical address 0. In which case PAT code
154 * falls back to slow path of reserving physical range page by page.
155 */
venkatesh.pallipadi@intel.com3c8bb732008-12-18 11:41:27 -0800156static inline int is_linear_pfn_mapping(struct vm_area_struct *vma)
157{
158 return ((vma->vm_flags & VM_PFNMAP) && vma->vm_pgoff);
159}
160
161static inline int is_pfn_mapping(struct vm_area_struct *vma)
162{
163 return (vma->vm_flags & VM_PFNMAP);
164}
Nick Piggind0217ac2007-07-19 01:47:03 -0700165
venkatesh.pallipadi@intel.com2ab64032008-12-18 11:41:29 -0800166extern int track_pfn_vma_new(struct vm_area_struct *vma, pgprot_t prot,
167 unsigned long pfn, unsigned long size);
168extern int track_pfn_vma_copy(struct vm_area_struct *vma);
169extern void untrack_pfn_vma(struct vm_area_struct *vma, unsigned long pfn,
170 unsigned long size);
171
Nick Piggin54cb8822007-07-19 01:46:59 -0700172/*
Nick Piggind0217ac2007-07-19 01:47:03 -0700173 * vm_fault is filled by the the pagefault handler and passed to the vma's
Nick Piggin83c54072007-07-19 01:47:05 -0700174 * ->fault function. The vma's ->fault is responsible for returning a bitmask
175 * of VM_FAULT_xxx flags that give details about how the fault was handled.
Nick Piggin54cb8822007-07-19 01:46:59 -0700176 *
Nick Piggind0217ac2007-07-19 01:47:03 -0700177 * pgoff should be used in favour of virtual_address, if possible. If pgoff
178 * is used, one may set VM_CAN_NONLINEAR in the vma->vm_flags to get nonlinear
179 * mapping support.
Nick Piggin54cb8822007-07-19 01:46:59 -0700180 */
Nick Piggind0217ac2007-07-19 01:47:03 -0700181struct vm_fault {
182 unsigned int flags; /* FAULT_FLAG_xxx flags */
183 pgoff_t pgoff; /* Logical page offset based on vma */
184 void __user *virtual_address; /* Faulting virtual address */
185
186 struct page *page; /* ->fault handlers should return a
Nick Piggin83c54072007-07-19 01:47:05 -0700187 * page here, unless VM_FAULT_NOPAGE
Nick Piggind0217ac2007-07-19 01:47:03 -0700188 * is set (which is also implied by
Nick Piggin83c54072007-07-19 01:47:05 -0700189 * VM_FAULT_ERROR).
Nick Piggind0217ac2007-07-19 01:47:03 -0700190 */
Nick Piggin54cb8822007-07-19 01:46:59 -0700191};
Linus Torvalds1da177e2005-04-16 15:20:36 -0700192
193/*
194 * These are the virtual MM functions - opening of an area, closing and
195 * unmapping it (needed to keep files on disk up-to-date etc), pointer
196 * to the functions called when a no-page or a wp-page exception occurs.
197 */
198struct vm_operations_struct {
199 void (*open)(struct vm_area_struct * area);
200 void (*close)(struct vm_area_struct * area);
Nick Piggind0217ac2007-07-19 01:47:03 -0700201 int (*fault)(struct vm_area_struct *vma, struct vm_fault *vmf);
David Howells9637a5e2006-06-23 02:03:43 -0700202
203 /* notification that a previously read-only page is about to become
204 * writable, if an error is returned it will cause a SIGBUS */
205 int (*page_mkwrite)(struct vm_area_struct *vma, struct page *page);
Rik van Riel28b2ee22008-07-23 21:27:05 -0700206
207 /* called by access_process_vm when get_user_pages() fails, typically
208 * for use by special VMAs that can switch between memory and hardware
209 */
210 int (*access)(struct vm_area_struct *vma, unsigned long addr,
211 void *buf, int len, int write);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700212#ifdef CONFIG_NUMA
Lee Schermerhorna6020ed2008-04-28 02:13:14 -0700213 /*
214 * set_policy() op must add a reference to any non-NULL @new mempolicy
215 * to hold the policy upon return. Caller should pass NULL @new to
216 * remove a policy and fall back to surrounding context--i.e. do not
217 * install a MPOL_DEFAULT policy, nor the task or system default
218 * mempolicy.
219 */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700220 int (*set_policy)(struct vm_area_struct *vma, struct mempolicy *new);
Lee Schermerhorna6020ed2008-04-28 02:13:14 -0700221
222 /*
223 * get_policy() op must add reference [mpol_get()] to any policy at
224 * (vma,addr) marked as MPOL_SHARED. The shared policy infrastructure
225 * in mm/mempolicy.c will do this automatically.
226 * get_policy() must NOT add a ref if the policy at (vma,addr) is not
227 * marked as MPOL_SHARED. vma policies are protected by the mmap_sem.
228 * If no [shared/vma] mempolicy exists at the addr, get_policy() op
229 * must return NULL--i.e., do not "fallback" to task or system default
230 * policy.
231 */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700232 struct mempolicy *(*get_policy)(struct vm_area_struct *vma,
233 unsigned long addr);
Christoph Lameter7b2259b2006-06-25 05:46:48 -0700234 int (*migrate)(struct vm_area_struct *vma, const nodemask_t *from,
235 const nodemask_t *to, unsigned long flags);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700236#endif
237};
238
239struct mmu_gather;
240struct inode;
241
Andrew Morton349aef02006-01-08 01:04:36 -0800242#define page_private(page) ((page)->private)
243#define set_page_private(page, v) ((page)->private = (v))
Hugh Dickins4c21e2f2005-10-29 18:16:40 -0700244
Linus Torvalds1da177e2005-04-16 15:20:36 -0700245/*
246 * FIXME: take this include out, include page-flags.h in
247 * files which need it (119 of them)
248 */
249#include <linux/page-flags.h>
250
251/*
252 * Methods to modify the page usage count.
253 *
254 * What counts for a page usage:
255 * - cache mapping (page->mapping)
256 * - private data (page->private)
257 * - page mapped in a task's page tables, each mapping
258 * is counted separately
259 *
260 * Also, many kernel routines increase the page count before a critical
261 * routine so they can be sure the page doesn't go away from under them.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700262 */
263
264/*
Nick Pigginda6052f2006-09-25 23:31:35 -0700265 * Drop a ref, return true if the refcount fell to zero (the page has no users)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700266 */
Nick Piggin7c8ee9a2006-03-22 00:08:03 -0800267static inline int put_page_testzero(struct page *page)
268{
Nick Piggin725d7042006-09-25 23:30:55 -0700269 VM_BUG_ON(atomic_read(&page->_count) == 0);
Nick Piggin8dc04ef2006-03-22 00:08:03 -0800270 return atomic_dec_and_test(&page->_count);
Nick Piggin7c8ee9a2006-03-22 00:08:03 -0800271}
Linus Torvalds1da177e2005-04-16 15:20:36 -0700272
273/*
Nick Piggin7c8ee9a2006-03-22 00:08:03 -0800274 * Try to grab a ref unless the page has a refcount of zero, return false if
275 * that is the case.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700276 */
Nick Piggin7c8ee9a2006-03-22 00:08:03 -0800277static inline int get_page_unless_zero(struct page *page)
278{
Christoph Lameteraec2c3e2008-02-04 22:28:35 -0800279 VM_BUG_ON(PageTail(page));
Nick Piggin8dc04ef2006-03-22 00:08:03 -0800280 return atomic_inc_not_zero(&page->_count);
Nick Piggin7c8ee9a2006-03-22 00:08:03 -0800281}
Linus Torvalds1da177e2005-04-16 15:20:36 -0700282
Christoph Lameter48667e72008-02-04 22:28:31 -0800283/* Support for virtually mapped pages */
Christoph Lameterb3bdda02008-02-04 22:28:32 -0800284struct page *vmalloc_to_page(const void *addr);
285unsigned long vmalloc_to_pfn(const void *addr);
Christoph Lameter48667e72008-02-04 22:28:31 -0800286
Paul Mundt0738c4b2008-03-12 16:51:31 +0900287/*
288 * Determine if an address is within the vmalloc range
289 *
290 * On nommu, vmalloc/vfree wrap through kmalloc/kfree directly, so there
291 * is no special casing required.
292 */
Christoph Lameter9e2779f2008-02-04 22:28:34 -0800293static inline int is_vmalloc_addr(const void *x)
294{
Paul Mundt0738c4b2008-03-12 16:51:31 +0900295#ifdef CONFIG_MMU
Christoph Lameter9e2779f2008-02-04 22:28:34 -0800296 unsigned long addr = (unsigned long)x;
297
298 return addr >= VMALLOC_START && addr < VMALLOC_END;
Paul Mundt0738c4b2008-03-12 16:51:31 +0900299#else
300 return 0;
David Howells8ca3ed82008-02-23 15:23:37 -0800301#endif
Paul Mundt0738c4b2008-03-12 16:51:31 +0900302}
Christoph Lameter9e2779f2008-02-04 22:28:34 -0800303
Christoph Lameterd85f3382007-05-06 14:49:39 -0700304static inline struct page *compound_head(struct page *page)
305{
Christoph Lameter6d777952007-05-06 14:49:40 -0700306 if (unlikely(PageTail(page)))
Christoph Lameterd85f3382007-05-06 14:49:39 -0700307 return page->first_page;
308 return page;
309}
310
Hugh Dickins4c21e2f2005-10-29 18:16:40 -0700311static inline int page_count(struct page *page)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700312{
Christoph Lameterd85f3382007-05-06 14:49:39 -0700313 return atomic_read(&compound_head(page)->_count);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700314}
315
316static inline void get_page(struct page *page)
317{
Christoph Lameterd85f3382007-05-06 14:49:39 -0700318 page = compound_head(page);
Nick Piggin725d7042006-09-25 23:30:55 -0700319 VM_BUG_ON(atomic_read(&page->_count) == 0);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700320 atomic_inc(&page->_count);
321}
322
Christoph Lameterb49af682007-05-06 14:49:41 -0700323static inline struct page *virt_to_head_page(const void *x)
324{
325 struct page *page = virt_to_page(x);
326 return compound_head(page);
327}
328
Nick Piggin7835e982006-03-22 00:08:40 -0800329/*
330 * Setup the page count before being freed into the page allocator for
331 * the first time (boot or memory hotplug)
332 */
333static inline void init_page_count(struct page *page)
334{
335 atomic_set(&page->_count, 1);
336}
337
Linus Torvalds1da177e2005-04-16 15:20:36 -0700338void put_page(struct page *page);
Alexander Zarochentsev1d7ea732006-08-13 23:24:27 -0700339void put_pages_list(struct list_head *pages);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700340
Nick Piggin8dfcc9b2006-03-22 00:08:05 -0800341void split_page(struct page *page, unsigned int order);
Nick Piggin8dfcc9b2006-03-22 00:08:05 -0800342
Linus Torvalds1da177e2005-04-16 15:20:36 -0700343/*
Andy Whitcroft33f2ef82006-12-06 20:33:32 -0800344 * Compound pages have a destructor function. Provide a
345 * prototype for that function and accessor functions.
346 * These are _only_ valid on the head of a PG_compound page.
347 */
348typedef void compound_page_dtor(struct page *);
349
350static inline void set_compound_page_dtor(struct page *page,
351 compound_page_dtor *dtor)
352{
353 page[1].lru.next = (void *)dtor;
354}
355
356static inline compound_page_dtor *get_compound_page_dtor(struct page *page)
357{
358 return (compound_page_dtor *)page[1].lru.next;
359}
360
Christoph Lameterd85f3382007-05-06 14:49:39 -0700361static inline int compound_order(struct page *page)
362{
Christoph Lameter6d777952007-05-06 14:49:40 -0700363 if (!PageHead(page))
Christoph Lameterd85f3382007-05-06 14:49:39 -0700364 return 0;
365 return (unsigned long)page[1].lru.prev;
366}
367
368static inline void set_compound_order(struct page *page, unsigned long order)
369{
370 page[1].lru.prev = (void *)order;
371}
372
Andy Whitcroft33f2ef82006-12-06 20:33:32 -0800373/*
Linus Torvalds1da177e2005-04-16 15:20:36 -0700374 * Multiple processes may "see" the same page. E.g. for untouched
375 * mappings of /dev/null, all processes see the same page full of
376 * zeroes, and text pages of executables and shared libraries have
377 * only one copy in memory, at most, normally.
378 *
379 * For the non-reserved pages, page_count(page) denotes a reference count.
Paolo 'Blaisorblade' Giarrusso7e871b62005-09-21 09:55:38 -0700380 * page_count() == 0 means the page is free. page->lru is then used for
381 * freelist management in the buddy allocator.
Nick Pigginda6052f2006-09-25 23:31:35 -0700382 * page_count() > 0 means the page has been allocated.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700383 *
Nick Pigginda6052f2006-09-25 23:31:35 -0700384 * Pages are allocated by the slab allocator in order to provide memory
385 * to kmalloc and kmem_cache_alloc. In this case, the management of the
386 * page, and the fields in 'struct page' are the responsibility of mm/slab.c
387 * unless a particular usage is carefully commented. (the responsibility of
388 * freeing the kmalloc memory is the caller's, of course).
Linus Torvalds1da177e2005-04-16 15:20:36 -0700389 *
Nick Pigginda6052f2006-09-25 23:31:35 -0700390 * A page may be used by anyone else who does a __get_free_page().
391 * In this case, page_count still tracks the references, and should only
392 * be used through the normal accessor functions. The top bits of page->flags
393 * and page->virtual store page management information, but all other fields
394 * are unused and could be used privately, carefully. The management of this
395 * page is the responsibility of the one who allocated it, and those who have
396 * subsequently been given references to it.
397 *
398 * The other pages (we may call them "pagecache pages") are completely
Linus Torvalds1da177e2005-04-16 15:20:36 -0700399 * managed by the Linux memory manager: I/O, buffers, swapping etc.
400 * The following discussion applies only to them.
401 *
Nick Pigginda6052f2006-09-25 23:31:35 -0700402 * A pagecache page contains an opaque `private' member, which belongs to the
403 * page's address_space. Usually, this is the address of a circular list of
404 * the page's disk buffers. PG_private must be set to tell the VM to call
405 * into the filesystem to release these pages.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700406 *
Nick Pigginda6052f2006-09-25 23:31:35 -0700407 * A page may belong to an inode's memory mapping. In this case, page->mapping
408 * is the pointer to the inode, and page->index is the file offset of the page,
409 * in units of PAGE_CACHE_SIZE.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700410 *
Nick Pigginda6052f2006-09-25 23:31:35 -0700411 * If pagecache pages are not associated with an inode, they are said to be
412 * anonymous pages. These may become associated with the swapcache, and in that
413 * case PG_swapcache is set, and page->private is an offset into the swapcache.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700414 *
Nick Pigginda6052f2006-09-25 23:31:35 -0700415 * In either case (swapcache or inode backed), the pagecache itself holds one
416 * reference to the page. Setting PG_private should also increment the
417 * refcount. The each user mapping also has a reference to the page.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700418 *
Nick Pigginda6052f2006-09-25 23:31:35 -0700419 * The pagecache pages are stored in a per-mapping radix tree, which is
420 * rooted at mapping->page_tree, and indexed by offset.
421 * Where 2.4 and early 2.6 kernels kept dirty/clean pages in per-address_space
422 * lists, we instead now tag pages as dirty/writeback in the radix tree.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700423 *
Nick Pigginda6052f2006-09-25 23:31:35 -0700424 * All pagecache pages may be subject to I/O:
Linus Torvalds1da177e2005-04-16 15:20:36 -0700425 * - inode pages may need to be read from disk,
426 * - inode pages which have been modified and are MAP_SHARED may need
Nick Pigginda6052f2006-09-25 23:31:35 -0700427 * to be written back to the inode on disk,
428 * - anonymous pages (including MAP_PRIVATE file mappings) which have been
429 * modified may need to be swapped out to swap space and (later) to be read
430 * back into memory.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700431 */
432
433/*
434 * The zone field is never updated after free_area_init_core()
435 * sets it, so none of the operations on it need to be atomic.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700436 */
Dave Hansen348f8b62005-06-23 00:07:40 -0700437
Andy Whitcroftd41dee32005-06-23 00:07:54 -0700438
439/*
440 * page->flags layout:
441 *
442 * There are three possibilities for how page->flags get
443 * laid out. The first is for the normal case, without
444 * sparsemem. The second is for sparsemem when there is
445 * plenty of space for node and section. The last is when
446 * we have run out of space and have to fall back to an
447 * alternate (slower) way of determining the node.
448 *
Christoph Lameter308c05e2008-04-28 02:12:43 -0700449 * No sparsemem or sparsemem vmemmap: | NODE | ZONE | ... | FLAGS |
450 * classic sparse with space for node:| SECTION | NODE | ZONE | ... | FLAGS |
451 * classic sparse no space for node: | SECTION | ZONE | ... | FLAGS |
Andy Whitcroftd41dee32005-06-23 00:07:54 -0700452 */
Christoph Lameter308c05e2008-04-28 02:12:43 -0700453#if defined(CONFIG_SPARSEMEM) && !defined(CONFIG_SPARSEMEM_VMEMMAP)
Andy Whitcroftd41dee32005-06-23 00:07:54 -0700454#define SECTIONS_WIDTH SECTIONS_SHIFT
455#else
456#define SECTIONS_WIDTH 0
457#endif
458
459#define ZONES_WIDTH ZONES_SHIFT
460
Christoph Lameter9223b412008-04-28 02:12:48 -0700461#if SECTIONS_WIDTH+ZONES_WIDTH+NODES_SHIFT <= BITS_PER_LONG - NR_PAGEFLAGS
Andy Whitcroftd41dee32005-06-23 00:07:54 -0700462#define NODES_WIDTH NODES_SHIFT
463#else
Christoph Lameter308c05e2008-04-28 02:12:43 -0700464#ifdef CONFIG_SPARSEMEM_VMEMMAP
465#error "Vmemmap: No space for nodes field in page flags"
466#endif
Andy Whitcroftd41dee32005-06-23 00:07:54 -0700467#define NODES_WIDTH 0
468#endif
469
470/* Page flags: | [SECTION] | [NODE] | ZONE | ... | FLAGS | */
Andi Kleen07808b72005-11-05 17:25:53 +0100471#define SECTIONS_PGOFF ((sizeof(unsigned long)*8) - SECTIONS_WIDTH)
Andy Whitcroftd41dee32005-06-23 00:07:54 -0700472#define NODES_PGOFF (SECTIONS_PGOFF - NODES_WIDTH)
473#define ZONES_PGOFF (NODES_PGOFF - ZONES_WIDTH)
474
475/*
476 * We are going to use the flags for the page to node mapping if its in
477 * there. This includes the case where there is no node, so it is implicit.
478 */
Christoph Lameter89689ae2006-12-06 20:31:45 -0800479#if !(NODES_WIDTH > 0 || NODES_SHIFT == 0)
480#define NODE_NOT_IN_PAGE_FLAGS
481#endif
Andy Whitcroftd41dee32005-06-23 00:07:54 -0700482
483#ifndef PFN_SECTION_SHIFT
484#define PFN_SECTION_SHIFT 0
485#endif
Dave Hansen348f8b62005-06-23 00:07:40 -0700486
487/*
488 * Define the bit shifts to access each section. For non-existant
489 * sections we define the shift as 0; that plus a 0 mask ensures
490 * the compiler will optimise away reference to them.
491 */
Andy Whitcroftd41dee32005-06-23 00:07:54 -0700492#define SECTIONS_PGSHIFT (SECTIONS_PGOFF * (SECTIONS_WIDTH != 0))
493#define NODES_PGSHIFT (NODES_PGOFF * (NODES_WIDTH != 0))
494#define ZONES_PGSHIFT (ZONES_PGOFF * (ZONES_WIDTH != 0))
Dave Hansen348f8b62005-06-23 00:07:40 -0700495
Christoph Lameter89689ae2006-12-06 20:31:45 -0800496/* NODE:ZONE or SECTION:ZONE is used to ID a zone for the buddy allcator */
497#ifdef NODE_NOT_IN_PAGEFLAGS
498#define ZONEID_SHIFT (SECTIONS_SHIFT + ZONES_SHIFT)
Andy Whitcroftbd8029b2007-02-10 01:43:14 -0800499#define ZONEID_PGOFF ((SECTIONS_PGOFF < ZONES_PGOFF)? \
500 SECTIONS_PGOFF : ZONES_PGOFF)
Andy Whitcroftd41dee32005-06-23 00:07:54 -0700501#else
Christoph Lameter89689ae2006-12-06 20:31:45 -0800502#define ZONEID_SHIFT (NODES_SHIFT + ZONES_SHIFT)
Andy Whitcroftbd8029b2007-02-10 01:43:14 -0800503#define ZONEID_PGOFF ((NODES_PGOFF < ZONES_PGOFF)? \
504 NODES_PGOFF : ZONES_PGOFF)
Andy Whitcroftd41dee32005-06-23 00:07:54 -0700505#endif
Christoph Lameter89689ae2006-12-06 20:31:45 -0800506
Andy Whitcroftbd8029b2007-02-10 01:43:14 -0800507#define ZONEID_PGSHIFT (ZONEID_PGOFF * (ZONEID_SHIFT != 0))
Dave Hansen348f8b62005-06-23 00:07:40 -0700508
Christoph Lameter9223b412008-04-28 02:12:48 -0700509#if SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > BITS_PER_LONG - NR_PAGEFLAGS
510#error SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > BITS_PER_LONG - NR_PAGEFLAGS
Dave Hansen348f8b62005-06-23 00:07:40 -0700511#endif
512
Andy Whitcroftd41dee32005-06-23 00:07:54 -0700513#define ZONES_MASK ((1UL << ZONES_WIDTH) - 1)
514#define NODES_MASK ((1UL << NODES_WIDTH) - 1)
515#define SECTIONS_MASK ((1UL << SECTIONS_WIDTH) - 1)
Christoph Lameter89689ae2006-12-06 20:31:45 -0800516#define ZONEID_MASK ((1UL << ZONEID_SHIFT) - 1)
Dave Hansen348f8b62005-06-23 00:07:40 -0700517
Christoph Lameter2f1b6242006-09-25 23:31:13 -0700518static inline enum zone_type page_zonenum(struct page *page)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700519{
Dave Hansen348f8b62005-06-23 00:07:40 -0700520 return (page->flags >> ZONES_PGSHIFT) & ZONES_MASK;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700521}
Linus Torvalds1da177e2005-04-16 15:20:36 -0700522
Christoph Lameter89689ae2006-12-06 20:31:45 -0800523/*
524 * The identification function is only used by the buddy allocator for
525 * determining if two pages could be buddies. We are not really
526 * identifying a zone since we could be using a the section number
527 * id if we have not node id available in page flags.
528 * We guarantee only that it will return the same value for two
529 * combinable pages in a zone.
530 */
Andy Whitcroftcb2b95e2006-06-23 02:03:01 -0700531static inline int page_zone_id(struct page *page)
532{
Christoph Lameter89689ae2006-12-06 20:31:45 -0800533 return (page->flags >> ZONEID_PGSHIFT) & ZONEID_MASK;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700534}
535
Andy Whitcroft25ba77c2006-12-06 20:33:03 -0800536static inline int zone_to_nid(struct zone *zone)
Christoph Lameter89fa3022006-09-25 23:31:55 -0700537{
Christoph Lameterd5f541e2006-09-27 01:50:08 -0700538#ifdef CONFIG_NUMA
539 return zone->node;
540#else
541 return 0;
542#endif
Christoph Lameter89fa3022006-09-25 23:31:55 -0700543}
544
Christoph Lameter89689ae2006-12-06 20:31:45 -0800545#ifdef NODE_NOT_IN_PAGE_FLAGS
Andy Whitcroft25ba77c2006-12-06 20:33:03 -0800546extern int page_to_nid(struct page *page);
Christoph Lameter89689ae2006-12-06 20:31:45 -0800547#else
Andy Whitcroft25ba77c2006-12-06 20:33:03 -0800548static inline int page_to_nid(struct page *page)
Andy Whitcroftd41dee32005-06-23 00:07:54 -0700549{
Christoph Lameter89689ae2006-12-06 20:31:45 -0800550 return (page->flags >> NODES_PGSHIFT) & NODES_MASK;
Andy Whitcroftd41dee32005-06-23 00:07:54 -0700551}
Christoph Lameter89689ae2006-12-06 20:31:45 -0800552#endif
553
554static inline struct zone *page_zone(struct page *page)
555{
556 return &NODE_DATA(page_to_nid(page))->node_zones[page_zonenum(page)];
557}
558
Christoph Lameter308c05e2008-04-28 02:12:43 -0700559#if defined(CONFIG_SPARSEMEM) && !defined(CONFIG_SPARSEMEM_VMEMMAP)
Andy Whitcroftd41dee32005-06-23 00:07:54 -0700560static inline unsigned long page_to_section(struct page *page)
561{
562 return (page->flags >> SECTIONS_PGSHIFT) & SECTIONS_MASK;
563}
Christoph Lameter308c05e2008-04-28 02:12:43 -0700564#endif
Andy Whitcroftd41dee32005-06-23 00:07:54 -0700565
Christoph Lameter2f1b6242006-09-25 23:31:13 -0700566static inline void set_page_zone(struct page *page, enum zone_type zone)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700567{
Dave Hansen348f8b62005-06-23 00:07:40 -0700568 page->flags &= ~(ZONES_MASK << ZONES_PGSHIFT);
569 page->flags |= (zone & ZONES_MASK) << ZONES_PGSHIFT;
570}
Christoph Lameter2f1b6242006-09-25 23:31:13 -0700571
Dave Hansen348f8b62005-06-23 00:07:40 -0700572static inline void set_page_node(struct page *page, unsigned long node)
573{
574 page->flags &= ~(NODES_MASK << NODES_PGSHIFT);
575 page->flags |= (node & NODES_MASK) << NODES_PGSHIFT;
576}
Christoph Lameter89689ae2006-12-06 20:31:45 -0800577
Andy Whitcroftd41dee32005-06-23 00:07:54 -0700578static inline void set_page_section(struct page *page, unsigned long section)
579{
580 page->flags &= ~(SECTIONS_MASK << SECTIONS_PGSHIFT);
581 page->flags |= (section & SECTIONS_MASK) << SECTIONS_PGSHIFT;
582}
Dave Hansen348f8b62005-06-23 00:07:40 -0700583
Christoph Lameter2f1b6242006-09-25 23:31:13 -0700584static inline void set_page_links(struct page *page, enum zone_type zone,
Andy Whitcroftd41dee32005-06-23 00:07:54 -0700585 unsigned long node, unsigned long pfn)
Dave Hansen348f8b62005-06-23 00:07:40 -0700586{
587 set_page_zone(page, zone);
588 set_page_node(page, node);
Andy Whitcroftd41dee32005-06-23 00:07:54 -0700589 set_page_section(page, pfn_to_section_nr(pfn));
Linus Torvalds1da177e2005-04-16 15:20:36 -0700590}
591
Christoph Lameterf6ac2352006-06-30 01:55:32 -0700592/*
Eric Paris7cd94142007-11-26 18:47:40 -0500593 * If a hint addr is less than mmap_min_addr change hint to be as
594 * low as possible but still greater than mmap_min_addr
595 */
596static inline unsigned long round_hint_to_min(unsigned long hint)
597{
598#ifdef CONFIG_SECURITY
599 hint &= PAGE_MASK;
600 if (((void *)hint != NULL) &&
601 (hint < mmap_min_addr))
602 return PAGE_ALIGN(mmap_min_addr);
603#endif
604 return hint;
605}
606
607/*
Christoph Lameterf6ac2352006-06-30 01:55:32 -0700608 * Some inline functions in vmstat.h depend on page_zone()
609 */
610#include <linux/vmstat.h>
611
Ingo Molnar652050a2006-01-14 13:21:30 -0800612static __always_inline void *lowmem_page_address(struct page *page)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700613{
614 return __va(page_to_pfn(page) << PAGE_SHIFT);
615}
616
617#if defined(CONFIG_HIGHMEM) && !defined(WANT_PAGE_VIRTUAL)
618#define HASHED_PAGE_VIRTUAL
619#endif
620
621#if defined(WANT_PAGE_VIRTUAL)
622#define page_address(page) ((page)->virtual)
623#define set_page_address(page, address) \
624 do { \
625 (page)->virtual = (address); \
626 } while(0)
627#define page_address_init() do { } while(0)
628#endif
629
630#if defined(HASHED_PAGE_VIRTUAL)
631void *page_address(struct page *page);
632void set_page_address(struct page *page, void *virtual);
633void page_address_init(void);
634#endif
635
636#if !defined(HASHED_PAGE_VIRTUAL) && !defined(WANT_PAGE_VIRTUAL)
637#define page_address(page) lowmem_page_address(page)
638#define set_page_address(page, address) do { } while(0)
639#define page_address_init() do { } while(0)
640#endif
641
642/*
643 * On an anonymous page mapped into a user virtual memory area,
644 * page->mapping points to its anon_vma, not to a struct address_space;
645 * with the PAGE_MAPPING_ANON bit set to distinguish it.
646 *
647 * Please note that, confusingly, "page_mapping" refers to the inode
648 * address_space which maps the page from disk; whereas "page_mapped"
649 * refers to user virtual address space into which the page is mapped.
650 */
651#define PAGE_MAPPING_ANON 1
652
653extern struct address_space swapper_space;
654static inline struct address_space *page_mapping(struct page *page)
655{
656 struct address_space *mapping = page->mapping;
657
Christoph Lameterb5fab142007-07-17 04:03:33 -0700658 VM_BUG_ON(PageSlab(page));
Andrew Morton726b8012008-04-28 02:12:44 -0700659#ifdef CONFIG_SWAP
Linus Torvalds1da177e2005-04-16 15:20:36 -0700660 if (unlikely(PageSwapCache(page)))
661 mapping = &swapper_space;
Andrew Morton726b8012008-04-28 02:12:44 -0700662 else
663#endif
664 if (unlikely((unsigned long)mapping & PAGE_MAPPING_ANON))
Linus Torvalds1da177e2005-04-16 15:20:36 -0700665 mapping = NULL;
666 return mapping;
667}
668
669static inline int PageAnon(struct page *page)
670{
671 return ((unsigned long)page->mapping & PAGE_MAPPING_ANON) != 0;
672}
673
674/*
675 * Return the pagecache index of the passed page. Regular pagecache pages
676 * use ->index whereas swapcache pages use ->private
677 */
678static inline pgoff_t page_index(struct page *page)
679{
680 if (unlikely(PageSwapCache(page)))
Hugh Dickins4c21e2f2005-10-29 18:16:40 -0700681 return page_private(page);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700682 return page->index;
683}
684
685/*
686 * The atomic page->_mapcount, like _count, starts from -1:
687 * so that transitions both from it and to it can be tracked,
688 * using atomic_inc_and_test and atomic_add_negative(-1).
689 */
690static inline void reset_page_mapcount(struct page *page)
691{
692 atomic_set(&(page)->_mapcount, -1);
693}
694
695static inline int page_mapcount(struct page *page)
696{
697 return atomic_read(&(page)->_mapcount) + 1;
698}
699
700/*
701 * Return true if this page is mapped into pagetables.
702 */
703static inline int page_mapped(struct page *page)
704{
705 return atomic_read(&(page)->_mapcount) >= 0;
706}
707
708/*
Linus Torvalds1da177e2005-04-16 15:20:36 -0700709 * Different kinds of faults, as returned by handle_mm_fault().
710 * Used to decide whether a process gets delivered SIGBUS or
711 * just gets major/minor fault counters bumped up.
712 */
Nick Piggind0217ac2007-07-19 01:47:03 -0700713
Nick Piggin83c54072007-07-19 01:47:05 -0700714#define VM_FAULT_MINOR 0 /* For backwards compat. Remove me quickly. */
Nick Piggind0217ac2007-07-19 01:47:03 -0700715
Nick Piggin83c54072007-07-19 01:47:05 -0700716#define VM_FAULT_OOM 0x0001
717#define VM_FAULT_SIGBUS 0x0002
718#define VM_FAULT_MAJOR 0x0004
719#define VM_FAULT_WRITE 0x0008 /* Special case for get_user_pages */
Nick Pigginf33ea7f2005-08-03 20:24:01 +1000720
Nick Piggin83c54072007-07-19 01:47:05 -0700721#define VM_FAULT_NOPAGE 0x0100 /* ->fault installed the pte, not return page */
722#define VM_FAULT_LOCKED 0x0200 /* ->fault locked the returned page */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700723
Nick Piggin83c54072007-07-19 01:47:05 -0700724#define VM_FAULT_ERROR (VM_FAULT_OOM | VM_FAULT_SIGBUS)
Nick Piggind0217ac2007-07-19 01:47:03 -0700725
Linus Torvalds1da177e2005-04-16 15:20:36 -0700726#define offset_in_page(p) ((unsigned long)(p) & ~PAGE_MASK)
727
728extern void show_free_areas(void);
729
730#ifdef CONFIG_SHMEM
Lee Schermerhorn89e004ea2008-10-18 20:26:43 -0700731extern int shmem_lock(struct file *file, int lock, struct user_struct *user);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700732#else
Russell King03b00eb2006-01-06 00:10:52 -0800733static inline int shmem_lock(struct file *file, int lock,
Lee Schermerhorn89e004ea2008-10-18 20:26:43 -0700734 struct user_struct *user)
Russell King03b00eb2006-01-06 00:10:52 -0800735{
736 return 0;
737}
Linus Torvalds1da177e2005-04-16 15:20:36 -0700738#endif
739struct file *shmem_file_setup(char *name, loff_t size, unsigned long flags);
740
741int shmem_zero_setup(struct vm_area_struct *);
742
David Howellsb0e15192006-01-06 00:11:42 -0800743#ifndef CONFIG_MMU
744extern unsigned long shmem_get_unmapped_area(struct file *file,
745 unsigned long addr,
746 unsigned long len,
747 unsigned long pgoff,
748 unsigned long flags);
749#endif
750
Alexey Dobriyane8edc6e2007-05-21 01:22:52 +0400751extern int can_do_mlock(void);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700752extern int user_shm_lock(size_t, struct user_struct *);
753extern void user_shm_unlock(size_t, struct user_struct *);
754
755/*
756 * Parameter block passed down to zap_pte_range in exceptional cases.
757 */
758struct zap_details {
759 struct vm_area_struct *nonlinear_vma; /* Check page->index if set */
760 struct address_space *check_mapping; /* Check page->mapping if set */
761 pgoff_t first_index; /* Lowest page->index to unmap */
762 pgoff_t last_index; /* Highest page->index to unmap */
763 spinlock_t *i_mmap_lock; /* For unmap_mapping_range: */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700764 unsigned long truncate_count; /* Compare vm_truncate_count */
765};
766
Nick Piggin7e675132008-04-28 02:13:00 -0700767struct page *vm_normal_page(struct vm_area_struct *vma, unsigned long addr,
768 pte_t pte);
769
Jack Steinerc627f9c2008-07-29 22:33:53 -0700770int zap_vma_ptes(struct vm_area_struct *vma, unsigned long address,
771 unsigned long size);
Hugh Dickinsee39b372005-04-19 13:29:15 -0700772unsigned long zap_page_range(struct vm_area_struct *vma, unsigned long address,
Linus Torvalds1da177e2005-04-16 15:20:36 -0700773 unsigned long size, struct zap_details *);
Hugh Dickins508034a2005-10-29 18:16:30 -0700774unsigned long unmap_vmas(struct mmu_gather **tlb,
Linus Torvalds1da177e2005-04-16 15:20:36 -0700775 struct vm_area_struct *start_vma, unsigned long start_addr,
776 unsigned long end_addr, unsigned long *nr_accounted,
777 struct zap_details *);
Matt Mackalle6473092008-02-04 22:29:01 -0800778
779/**
780 * mm_walk - callbacks for walk_page_range
781 * @pgd_entry: if set, called for each non-empty PGD (top-level) entry
782 * @pud_entry: if set, called for each non-empty PUD (2nd-level) entry
783 * @pmd_entry: if set, called for each non-empty PMD (3rd-level) entry
784 * @pte_entry: if set, called for each non-empty PTE (4th-level) entry
785 * @pte_hole: if set, called for each hole at all levels
786 *
787 * (see walk_page_range for more details)
788 */
789struct mm_walk {
Dave Hansen21650092008-06-12 15:21:47 -0700790 int (*pgd_entry)(pgd_t *, unsigned long, unsigned long, struct mm_walk *);
791 int (*pud_entry)(pud_t *, unsigned long, unsigned long, struct mm_walk *);
792 int (*pmd_entry)(pmd_t *, unsigned long, unsigned long, struct mm_walk *);
793 int (*pte_entry)(pte_t *, unsigned long, unsigned long, struct mm_walk *);
794 int (*pte_hole)(unsigned long, unsigned long, struct mm_walk *);
795 struct mm_struct *mm;
796 void *private;
Matt Mackalle6473092008-02-04 22:29:01 -0800797};
798
Dave Hansen21650092008-06-12 15:21:47 -0700799int walk_page_range(unsigned long addr, unsigned long end,
800 struct mm_walk *walk);
Jan Beulich42b77722008-07-23 21:27:10 -0700801void free_pgd_range(struct mmu_gather *tlb, unsigned long addr,
Hugh Dickins3bf5ee92005-04-19 13:29:16 -0700802 unsigned long end, unsigned long floor, unsigned long ceiling);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700803int copy_page_range(struct mm_struct *dst, struct mm_struct *src,
804 struct vm_area_struct *vma);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700805void unmap_mapping_range(struct address_space *mapping,
806 loff_t const holebegin, loff_t const holelen, int even_cows);
venkatesh.pallipadi@intel.comd87fe662008-12-19 13:47:27 -0800807int follow_phys(struct vm_area_struct *vma, unsigned long address,
808 unsigned int flags, unsigned long *prot, resource_size_t *phys);
Rik van Riel28b2ee22008-07-23 21:27:05 -0700809int generic_access_phys(struct vm_area_struct *vma, unsigned long addr,
810 void *buf, int len, int write);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700811
812static inline void unmap_shared_mapping_range(struct address_space *mapping,
813 loff_t const holebegin, loff_t const holelen)
814{
815 unmap_mapping_range(mapping, holebegin, holelen, 0);
816}
817
818extern int vmtruncate(struct inode * inode, loff_t offset);
Badari Pulavartyf6b3ec22006-01-06 00:10:38 -0800819extern int vmtruncate_range(struct inode * inode, loff_t offset, loff_t end);
Nick Pigginf33ea7f2005-08-03 20:24:01 +1000820
David Howells7ee1dd32006-01-06 00:11:44 -0800821#ifdef CONFIG_MMU
Nick Piggin83c54072007-07-19 01:47:05 -0700822extern int handle_mm_fault(struct mm_struct *mm, struct vm_area_struct *vma,
David Howells7ee1dd32006-01-06 00:11:44 -0800823 unsigned long address, int write_access);
David Howells7ee1dd32006-01-06 00:11:44 -0800824#else
825static inline int handle_mm_fault(struct mm_struct *mm,
826 struct vm_area_struct *vma, unsigned long address,
827 int write_access)
828{
829 /* should never happen if there's no MMU */
830 BUG();
831 return VM_FAULT_SIGBUS;
832}
833#endif
Nick Pigginf33ea7f2005-08-03 20:24:01 +1000834
Linus Torvalds1da177e2005-04-16 15:20:36 -0700835extern int make_pages_present(unsigned long addr, unsigned long end);
836extern int access_process_vm(struct task_struct *tsk, unsigned long addr, void *buf, int len, int write);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700837
838int get_user_pages(struct task_struct *tsk, struct mm_struct *mm, unsigned long start,
839 int len, int write, int force, struct page **pages, struct vm_area_struct **vmas);
840
David Howellscf9a2ae2006-08-29 19:05:54 +0100841extern int try_to_release_page(struct page * page, gfp_t gfp_mask);
842extern void do_invalidatepage(struct page *page, unsigned long offset);
843
Linus Torvalds1da177e2005-04-16 15:20:36 -0700844int __set_page_dirty_nobuffers(struct page *page);
Ken Chen76719322007-02-10 01:43:15 -0800845int __set_page_dirty_no_writeback(struct page *page);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700846int redirty_page_for_writepage(struct writeback_control *wbc,
847 struct page *page);
Harvey Harrisonb3c97522008-02-13 15:03:15 -0800848int set_page_dirty(struct page *page);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700849int set_page_dirty_lock(struct page *page);
850int clear_page_dirty_for_io(struct page *page);
851
Ollie Wildb6a2fea2007-07-19 01:48:16 -0700852extern unsigned long move_page_tables(struct vm_area_struct *vma,
853 unsigned long old_addr, struct vm_area_struct *new_vma,
854 unsigned long new_addr, unsigned long len);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700855extern unsigned long do_mremap(unsigned long addr,
856 unsigned long old_len, unsigned long new_len,
857 unsigned long flags, unsigned long new_addr);
Ollie Wildb6a2fea2007-07-19 01:48:16 -0700858extern int mprotect_fixup(struct vm_area_struct *vma,
859 struct vm_area_struct **pprev, unsigned long start,
860 unsigned long end, unsigned long newflags);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700861
Nick Piggin21cc1992008-07-25 19:45:22 -0700862/*
863 * get_user_pages_fast provides equivalent functionality to get_user_pages,
864 * operating on current and current->mm (force=0 and doesn't return any vmas).
865 *
866 * get_user_pages_fast may take mmap_sem and page tables, so no assumptions
867 * can be made about locking. get_user_pages_fast is to be implemented in a
868 * way that is advantageous (vs get_user_pages()) when the user memory area is
869 * already faulted in and present in ptes. However if the pages have to be
870 * faulted in, it may turn out to be slightly slower).
871 */
872int get_user_pages_fast(unsigned long start, int nr_pages, int write,
873 struct page **pages);
874
Linus Torvalds1da177e2005-04-16 15:20:36 -0700875/*
Rusty Russell8e1f9362007-07-17 04:03:17 -0700876 * A callback you can register to apply pressure to ageable caches.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700877 *
Rusty Russell8e1f9362007-07-17 04:03:17 -0700878 * 'shrink' is passed a count 'nr_to_scan' and a 'gfpmask'. It should
879 * look through the least-recently-used 'nr_to_scan' entries and
880 * attempt to free them up. It should return the number of objects
881 * which remain in the cache. If it returns -1, it means it cannot do
882 * any scanning at this time (eg. there is a risk of deadlock).
Linus Torvalds1da177e2005-04-16 15:20:36 -0700883 *
Rusty Russell8e1f9362007-07-17 04:03:17 -0700884 * The 'gfpmask' refers to the allocation we are currently trying to
885 * fulfil.
886 *
887 * Note that 'shrink' will be passed nr_to_scan == 0 when the VM is
888 * querying the cache size, so a fastpath for that case is appropriate.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700889 */
Rusty Russell8e1f9362007-07-17 04:03:17 -0700890struct shrinker {
891 int (*shrink)(int nr_to_scan, gfp_t gfp_mask);
892 int seeks; /* seeks to recreate an obj */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700893
Rusty Russell8e1f9362007-07-17 04:03:17 -0700894 /* These are for internal use */
895 struct list_head list;
896 long nr; /* objs pending delete */
897};
898#define DEFAULT_SEEKS 2 /* A good number if you don't know better. */
899extern void register_shrinker(struct shrinker *);
900extern void unregister_shrinker(struct shrinker *);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700901
Alexey Dobriyan4e950f62007-07-30 02:36:13 +0400902int vma_wants_writenotify(struct vm_area_struct *vma);
Peter Zijlstrad08b3852006-09-25 23:30:57 -0700903
Harvey Harrisonb3c97522008-02-13 15:03:15 -0800904extern pte_t *get_locked_pte(struct mm_struct *mm, unsigned long addr, spinlock_t **ptl);
Linus Torvaldsc9cfcdd2005-11-29 14:03:14 -0800905
Nick Piggin5f22df02007-05-06 14:49:02 -0700906#ifdef __PAGETABLE_PUD_FOLDED
907static inline int __pud_alloc(struct mm_struct *mm, pgd_t *pgd,
908 unsigned long address)
909{
910 return 0;
911}
912#else
Hugh Dickins1bb36302005-10-29 18:16:22 -0700913int __pud_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address);
Nick Piggin5f22df02007-05-06 14:49:02 -0700914#endif
915
916#ifdef __PAGETABLE_PMD_FOLDED
917static inline int __pmd_alloc(struct mm_struct *mm, pud_t *pud,
918 unsigned long address)
919{
920 return 0;
921}
922#else
Hugh Dickins1bb36302005-10-29 18:16:22 -0700923int __pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address);
Nick Piggin5f22df02007-05-06 14:49:02 -0700924#endif
925
Hugh Dickins1bb36302005-10-29 18:16:22 -0700926int __pte_alloc(struct mm_struct *mm, pmd_t *pmd, unsigned long address);
927int __pte_alloc_kernel(pmd_t *pmd, unsigned long address);
928
Linus Torvalds1da177e2005-04-16 15:20:36 -0700929/*
930 * The following ifdef needed to get the 4level-fixup.h header to work.
931 * Remove it when 4level-fixup.h has been removed.
932 */
Hugh Dickins1bb36302005-10-29 18:16:22 -0700933#if defined(CONFIG_MMU) && !defined(__ARCH_HAS_4LEVEL_HACK)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700934static inline pud_t *pud_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address)
935{
Hugh Dickins1bb36302005-10-29 18:16:22 -0700936 return (unlikely(pgd_none(*pgd)) && __pud_alloc(mm, pgd, address))?
937 NULL: pud_offset(pgd, address);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700938}
939
940static inline pmd_t *pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address)
941{
Hugh Dickins1bb36302005-10-29 18:16:22 -0700942 return (unlikely(pud_none(*pud)) && __pmd_alloc(mm, pud, address))?
943 NULL: pmd_offset(pud, address);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700944}
Hugh Dickins1bb36302005-10-29 18:16:22 -0700945#endif /* CONFIG_MMU && !__ARCH_HAS_4LEVEL_HACK */
946
Jeremy Fitzhardingef7d0b922008-09-09 15:43:22 -0700947#if USE_SPLIT_PTLOCKS
Hugh Dickins4c21e2f2005-10-29 18:16:40 -0700948/*
949 * We tuck a spinlock to guard each pagetable page into its struct page,
950 * at page->private, with BUILD_BUG_ON to make sure that this will not
951 * overflow into the next struct page (as it might with DEBUG_SPINLOCK).
952 * When freeing, reset page->mapping so free_pages_check won't complain.
953 */
Andrew Morton349aef02006-01-08 01:04:36 -0800954#define __pte_lockptr(page) &((page)->ptl)
Hugh Dickins4c21e2f2005-10-29 18:16:40 -0700955#define pte_lock_init(_page) do { \
956 spin_lock_init(__pte_lockptr(_page)); \
957} while (0)
958#define pte_lock_deinit(page) ((page)->mapping = NULL)
959#define pte_lockptr(mm, pmd) ({(void)(mm); __pte_lockptr(pmd_page(*(pmd)));})
Jeremy Fitzhardingef7d0b922008-09-09 15:43:22 -0700960#else /* !USE_SPLIT_PTLOCKS */
Hugh Dickins4c21e2f2005-10-29 18:16:40 -0700961/*
962 * We use mm->page_table_lock to guard all pagetable pages of the mm.
963 */
964#define pte_lock_init(page) do {} while (0)
965#define pte_lock_deinit(page) do {} while (0)
966#define pte_lockptr(mm, pmd) ({(void)(pmd); &(mm)->page_table_lock;})
Jeremy Fitzhardingef7d0b922008-09-09 15:43:22 -0700967#endif /* USE_SPLIT_PTLOCKS */
Hugh Dickins4c21e2f2005-10-29 18:16:40 -0700968
Martin Schwidefsky2f569af2008-02-08 04:22:04 -0800969static inline void pgtable_page_ctor(struct page *page)
970{
971 pte_lock_init(page);
972 inc_zone_page_state(page, NR_PAGETABLE);
973}
974
975static inline void pgtable_page_dtor(struct page *page)
976{
977 pte_lock_deinit(page);
978 dec_zone_page_state(page, NR_PAGETABLE);
979}
980
Hugh Dickinsc74df322005-10-29 18:16:23 -0700981#define pte_offset_map_lock(mm, pmd, address, ptlp) \
982({ \
Hugh Dickins4c21e2f2005-10-29 18:16:40 -0700983 spinlock_t *__ptl = pte_lockptr(mm, pmd); \
Hugh Dickinsc74df322005-10-29 18:16:23 -0700984 pte_t *__pte = pte_offset_map(pmd, address); \
985 *(ptlp) = __ptl; \
986 spin_lock(__ptl); \
987 __pte; \
988})
989
990#define pte_unmap_unlock(pte, ptl) do { \
991 spin_unlock(ptl); \
992 pte_unmap(pte); \
993} while (0)
994
Hugh Dickins1bb36302005-10-29 18:16:22 -0700995#define pte_alloc_map(mm, pmd, address) \
996 ((unlikely(!pmd_present(*(pmd))) && __pte_alloc(mm, pmd, address))? \
997 NULL: pte_offset_map(pmd, address))
998
Hugh Dickinsc74df322005-10-29 18:16:23 -0700999#define pte_alloc_map_lock(mm, pmd, address, ptlp) \
1000 ((unlikely(!pmd_present(*(pmd))) && __pte_alloc(mm, pmd, address))? \
1001 NULL: pte_offset_map_lock(mm, pmd, address, ptlp))
1002
Hugh Dickins1bb36302005-10-29 18:16:22 -07001003#define pte_alloc_kernel(pmd, address) \
1004 ((unlikely(!pmd_present(*(pmd))) && __pte_alloc_kernel(pmd, address))? \
1005 NULL: pte_offset_kernel(pmd, address))
Linus Torvalds1da177e2005-04-16 15:20:36 -07001006
1007extern void free_area_init(unsigned long * zones_size);
Johannes Weiner9109fb72008-07-23 21:27:20 -07001008extern void free_area_init_node(int nid, unsigned long * zones_size,
1009 unsigned long zone_start_pfn, unsigned long *zholes_size);
Mel Gormanc7132162006-09-27 01:49:43 -07001010#ifdef CONFIG_ARCH_POPULATES_NODE_MAP
1011/*
1012 * With CONFIG_ARCH_POPULATES_NODE_MAP set, an architecture may initialise its
1013 * zones, allocate the backing mem_map and account for memory holes in a more
1014 * architecture independent manner. This is a substitute for creating the
1015 * zone_sizes[] and zholes_size[] arrays and passing them to
1016 * free_area_init_node()
1017 *
1018 * An architecture is expected to register range of page frames backed by
1019 * physical memory with add_active_range() before calling
1020 * free_area_init_nodes() passing in the PFN each zone ends at. At a basic
1021 * usage, an architecture is expected to do something like
1022 *
1023 * unsigned long max_zone_pfns[MAX_NR_ZONES] = {max_dma, max_normal_pfn,
1024 * max_highmem_pfn};
1025 * for_each_valid_physical_page_range()
1026 * add_active_range(node_id, start_pfn, end_pfn)
1027 * free_area_init_nodes(max_zone_pfns);
1028 *
1029 * If the architecture guarantees that there are no holes in the ranges
1030 * registered with add_active_range(), free_bootmem_active_regions()
1031 * will call free_bootmem_node() for each registered physical page range.
1032 * Similarly sparse_memory_present_with_active_regions() calls
1033 * memory_present() for each range when SPARSEMEM is enabled.
1034 *
1035 * See mm/page_alloc.c for more information on each function exposed by
1036 * CONFIG_ARCH_POPULATES_NODE_MAP
1037 */
1038extern void free_area_init_nodes(unsigned long *max_zone_pfn);
1039extern void add_active_range(unsigned int nid, unsigned long start_pfn,
1040 unsigned long end_pfn);
Yinghai Lucc1050b2008-06-13 19:08:52 -07001041extern void remove_active_range(unsigned int nid, unsigned long start_pfn,
1042 unsigned long end_pfn);
Mel Gormanfb014392006-09-27 01:49:59 -07001043extern void push_node_boundaries(unsigned int nid, unsigned long start_pfn,
1044 unsigned long end_pfn);
Mel Gormanc7132162006-09-27 01:49:43 -07001045extern void remove_all_active_ranges(void);
1046extern unsigned long absent_pages_in_range(unsigned long start_pfn,
1047 unsigned long end_pfn);
1048extern void get_pfn_range_for_nid(unsigned int nid,
1049 unsigned long *start_pfn, unsigned long *end_pfn);
1050extern unsigned long find_min_pfn_with_active_regions(void);
Mel Gormanc7132162006-09-27 01:49:43 -07001051extern void free_bootmem_with_active_regions(int nid,
1052 unsigned long max_low_pfn);
Yinghai Lud52d53b2008-06-16 20:10:55 -07001053typedef int (*work_fn_t)(unsigned long, unsigned long, void *);
Yinghai Lub5bc6c02008-06-14 18:32:52 -07001054extern void work_with_active_regions(int nid, work_fn_t work_fn, void *data);
Mel Gormanc7132162006-09-27 01:49:43 -07001055extern void sparse_memory_present_with_active_regions(int nid);
1056#ifndef CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID
1057extern int early_pfn_to_nid(unsigned long pfn);
1058#endif /* CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID */
1059#endif /* CONFIG_ARCH_POPULATES_NODE_MAP */
Mel Gorman0e0b8642006-09-27 01:49:56 -07001060extern void set_dma_reserve(unsigned long new_dma_reserve);
Dave Hansena2f3aa02007-01-10 23:15:30 -08001061extern void memmap_init_zone(unsigned long, int, unsigned long,
1062 unsigned long, enum memmap_context);
Dave Hansen3947be12005-10-29 18:16:54 -07001063extern void setup_per_zone_pages_min(void);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001064extern void mem_init(void);
1065extern void show_mem(void);
1066extern void si_meminfo(struct sysinfo * val);
1067extern void si_meminfo_node(struct sysinfo *val, int nid);
Mike Travis3461b0a2008-05-12 21:21:13 +02001068extern int after_bootmem;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001069
Christoph Lametere7c8d5c2005-06-21 17:14:47 -07001070#ifdef CONFIG_NUMA
1071extern void setup_per_cpu_pageset(void);
1072#else
1073static inline void setup_per_cpu_pageset(void) {}
1074#endif
1075
Linus Torvalds1da177e2005-04-16 15:20:36 -07001076/* prio_tree.c */
1077void vma_prio_tree_add(struct vm_area_struct *, struct vm_area_struct *old);
1078void vma_prio_tree_insert(struct vm_area_struct *, struct prio_tree_root *);
1079void vma_prio_tree_remove(struct vm_area_struct *, struct prio_tree_root *);
1080struct vm_area_struct *vma_prio_tree_next(struct vm_area_struct *vma,
1081 struct prio_tree_iter *iter);
1082
1083#define vma_prio_tree_foreach(vma, iter, root, begin, end) \
1084 for (prio_tree_iter_init(iter, root, begin, end), vma = NULL; \
1085 (vma = vma_prio_tree_next(vma, iter)); )
1086
1087static inline void vma_nonlinear_insert(struct vm_area_struct *vma,
1088 struct list_head *list)
1089{
1090 vma->shared.vm_set.parent = NULL;
1091 list_add_tail(&vma->shared.vm_set.list, list);
1092}
1093
1094/* mmap.c */
Alan Cox34b4e4a2007-08-22 14:01:28 -07001095extern int __vm_enough_memory(struct mm_struct *mm, long pages, int cap_sys_admin);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001096extern void vma_adjust(struct vm_area_struct *vma, unsigned long start,
1097 unsigned long end, pgoff_t pgoff, struct vm_area_struct *insert);
1098extern struct vm_area_struct *vma_merge(struct mm_struct *,
1099 struct vm_area_struct *prev, unsigned long addr, unsigned long end,
1100 unsigned long vm_flags, struct anon_vma *, struct file *, pgoff_t,
1101 struct mempolicy *);
1102extern struct anon_vma *find_mergeable_anon_vma(struct vm_area_struct *);
1103extern int split_vma(struct mm_struct *,
1104 struct vm_area_struct *, unsigned long addr, int new_below);
1105extern int insert_vm_struct(struct mm_struct *, struct vm_area_struct *);
1106extern void __vma_link_rb(struct mm_struct *, struct vm_area_struct *,
1107 struct rb_node **, struct rb_node *);
Hugh Dickinsa8fb5612005-10-29 18:15:57 -07001108extern void unlink_file_vma(struct vm_area_struct *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001109extern struct vm_area_struct *copy_vma(struct vm_area_struct **,
1110 unsigned long addr, unsigned long len, pgoff_t pgoff);
1111extern void exit_mmap(struct mm_struct *);
Matt Helsley925d1c42008-04-29 01:01:36 -07001112
Andrea Arcangeli7906d002008-07-28 15:46:26 -07001113extern int mm_take_all_locks(struct mm_struct *mm);
1114extern void mm_drop_all_locks(struct mm_struct *mm);
1115
Matt Helsley925d1c42008-04-29 01:01:36 -07001116#ifdef CONFIG_PROC_FS
1117/* From fs/proc/base.c. callers must _not_ hold the mm's exe_file_lock */
1118extern void added_exe_file_vma(struct mm_struct *mm);
1119extern void removed_exe_file_vma(struct mm_struct *mm);
1120#else
1121static inline void added_exe_file_vma(struct mm_struct *mm)
1122{}
1123
1124static inline void removed_exe_file_vma(struct mm_struct *mm)
1125{}
1126#endif /* CONFIG_PROC_FS */
1127
akpm@osdl.org119f6572005-05-01 08:58:35 -07001128extern int may_expand_vm(struct mm_struct *mm, unsigned long npages);
Roland McGrathfa5dc222007-02-08 14:20:41 -08001129extern int install_special_mapping(struct mm_struct *mm,
1130 unsigned long addr, unsigned long len,
1131 unsigned long flags, struct page **pages);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001132
1133extern unsigned long get_unmapped_area(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
1134
1135extern unsigned long do_mmap_pgoff(struct file *file, unsigned long addr,
1136 unsigned long len, unsigned long prot,
1137 unsigned long flag, unsigned long pgoff);
Miklos Szeredi0165ab42007-07-15 23:38:26 -07001138extern unsigned long mmap_region(struct file *file, unsigned long addr,
1139 unsigned long len, unsigned long flags,
1140 unsigned int vm_flags, unsigned long pgoff,
1141 int accountable);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001142
1143static inline unsigned long do_mmap(struct file *file, unsigned long addr,
1144 unsigned long len, unsigned long prot,
1145 unsigned long flag, unsigned long offset)
1146{
1147 unsigned long ret = -EINVAL;
1148 if ((offset + PAGE_ALIGN(len)) < offset)
1149 goto out;
1150 if (!(offset & ~PAGE_MASK))
1151 ret = do_mmap_pgoff(file, addr, len, prot, flag, offset >> PAGE_SHIFT);
1152out:
1153 return ret;
1154}
1155
1156extern int do_munmap(struct mm_struct *, unsigned long, size_t);
1157
1158extern unsigned long do_brk(unsigned long, unsigned long);
1159
1160/* filemap.c */
1161extern unsigned long page_unuse(struct page *);
1162extern void truncate_inode_pages(struct address_space *, loff_t);
Hans Reiserd7339072006-01-06 00:10:36 -08001163extern void truncate_inode_pages_range(struct address_space *,
1164 loff_t lstart, loff_t lend);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001165
1166/* generic vm_area_ops exported for stackable file systems */
Nick Piggind0217ac2007-07-19 01:47:03 -07001167extern int filemap_fault(struct vm_area_struct *, struct vm_fault *);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001168
1169/* mm/page-writeback.c */
1170int write_one_page(struct page *page, int wait);
1171
1172/* readahead.c */
1173#define VM_MAX_READAHEAD 128 /* kbytes */
1174#define VM_MIN_READAHEAD 16 /* kbytes (includes current page) */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001175
1176int do_page_cache_readahead(struct address_space *mapping, struct file *filp,
Andrew Morton7361f4d2005-11-07 00:59:28 -08001177 pgoff_t offset, unsigned long nr_to_read);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001178int force_page_cache_readahead(struct address_space *mapping, struct file *filp,
Andrew Morton7361f4d2005-11-07 00:59:28 -08001179 pgoff_t offset, unsigned long nr_to_read);
Rusty Russellcf914a72007-07-19 01:48:08 -07001180
1181void page_cache_sync_readahead(struct address_space *mapping,
1182 struct file_ra_state *ra,
1183 struct file *filp,
1184 pgoff_t offset,
1185 unsigned long size);
1186
1187void page_cache_async_readahead(struct address_space *mapping,
1188 struct file_ra_state *ra,
1189 struct file *filp,
1190 struct page *pg,
1191 pgoff_t offset,
1192 unsigned long size);
1193
Linus Torvalds1da177e2005-04-16 15:20:36 -07001194unsigned long max_sane_readahead(unsigned long nr);
1195
1196/* Do stack extension */
Hugh Dickins46dea3d2005-10-29 18:16:20 -07001197extern int expand_stack(struct vm_area_struct *vma, unsigned long address);
Matthew Wilcox9ab88512005-11-18 16:16:42 -05001198#ifdef CONFIG_IA64
Hugh Dickins46dea3d2005-10-29 18:16:20 -07001199extern int expand_upwards(struct vm_area_struct *vma, unsigned long address);
Matthew Wilcox9ab88512005-11-18 16:16:42 -05001200#endif
Ollie Wildb6a2fea2007-07-19 01:48:16 -07001201extern int expand_stack_downwards(struct vm_area_struct *vma,
1202 unsigned long address);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001203
1204/* Look up the first VMA which satisfies addr < vm_end, NULL if none. */
1205extern struct vm_area_struct * find_vma(struct mm_struct * mm, unsigned long addr);
1206extern struct vm_area_struct * find_vma_prev(struct mm_struct * mm, unsigned long addr,
1207 struct vm_area_struct **pprev);
1208
1209/* Look up the first VMA which intersects the interval start_addr..end_addr-1,
1210 NULL if none. Assume start_addr < end_addr. */
1211static inline struct vm_area_struct * find_vma_intersection(struct mm_struct * mm, unsigned long start_addr, unsigned long end_addr)
1212{
1213 struct vm_area_struct * vma = find_vma(mm,start_addr);
1214
1215 if (vma && end_addr <= vma->vm_start)
1216 vma = NULL;
1217 return vma;
1218}
1219
1220static inline unsigned long vma_pages(struct vm_area_struct *vma)
1221{
1222 return (vma->vm_end - vma->vm_start) >> PAGE_SHIFT;
1223}
1224
Hugh Dickins804af2c2006-07-26 21:39:49 +01001225pgprot_t vm_get_page_prot(unsigned long vm_flags);
Hugh Dickinsdeceb6c2005-10-29 18:16:33 -07001226struct vm_area_struct *find_extend_vma(struct mm_struct *, unsigned long addr);
Hugh Dickinsdeceb6c2005-10-29 18:16:33 -07001227int remap_pfn_range(struct vm_area_struct *, unsigned long addr,
1228 unsigned long pfn, unsigned long size, pgprot_t);
Linus Torvaldsa145dd42005-11-30 09:35:19 -08001229int vm_insert_page(struct vm_area_struct *, unsigned long addr, struct page *);
Nick Piggine0dc0d82007-02-12 00:51:36 -08001230int vm_insert_pfn(struct vm_area_struct *vma, unsigned long addr,
1231 unsigned long pfn);
Nick Piggin423bad62008-04-28 02:13:01 -07001232int vm_insert_mixed(struct vm_area_struct *vma, unsigned long addr,
1233 unsigned long pfn);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001234
Linus Torvalds6aab3412005-11-28 14:34:23 -08001235struct page *follow_page(struct vm_area_struct *, unsigned long address,
Hugh Dickinsdeceb6c2005-10-29 18:16:33 -07001236 unsigned int foll_flags);
1237#define FOLL_WRITE 0x01 /* check pte is writable */
1238#define FOLL_TOUCH 0x02 /* mark page accessed */
1239#define FOLL_GET 0x04 /* do get_page on page */
1240#define FOLL_ANON 0x08 /* give ZERO_PAGE if no pgtable */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001241
venkatesh.pallipadi@intel.come121e412008-12-18 11:41:28 -08001242int follow_pfnmap_pte(struct vm_area_struct *vma,
1243 unsigned long address, pte_t *ret_ptep);
1244
Martin Schwidefsky2f569af2008-02-08 04:22:04 -08001245typedef int (*pte_fn_t)(pte_t *pte, pgtable_t token, unsigned long addr,
Jeremy Fitzhardingeaee16b32007-05-06 14:48:54 -07001246 void *data);
1247extern int apply_to_page_range(struct mm_struct *mm, unsigned long address,
1248 unsigned long size, pte_fn_t fn, void *data);
1249
Linus Torvalds1da177e2005-04-16 15:20:36 -07001250#ifdef CONFIG_PROC_FS
Hugh Dickinsab50b8e2005-10-29 18:15:56 -07001251void vm_stat_account(struct mm_struct *, unsigned long, struct file *, long);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001252#else
Hugh Dickinsab50b8e2005-10-29 18:15:56 -07001253static inline void vm_stat_account(struct mm_struct *mm,
Linus Torvalds1da177e2005-04-16 15:20:36 -07001254 unsigned long flags, struct file *file, long pages)
1255{
1256}
1257#endif /* CONFIG_PROC_FS */
1258
Ingo Molnar12d6f212008-01-30 13:33:58 +01001259#ifdef CONFIG_DEBUG_PAGEALLOC
1260extern int debug_pagealloc_enabled;
1261
1262extern void kernel_map_pages(struct page *page, int numpages, int enable);
1263
1264static inline void enable_debug_pagealloc(void)
1265{
1266 debug_pagealloc_enabled = 1;
1267}
Rafael J. Wysocki8a235ef2008-02-20 01:47:44 +01001268#ifdef CONFIG_HIBERNATION
1269extern bool kernel_page_present(struct page *page);
1270#endif /* CONFIG_HIBERNATION */
Ingo Molnar12d6f212008-01-30 13:33:58 +01001271#else
Linus Torvalds1da177e2005-04-16 15:20:36 -07001272static inline void
Nick Piggin9858db52006-10-11 01:21:30 -07001273kernel_map_pages(struct page *page, int numpages, int enable) {}
Ingo Molnar12d6f212008-01-30 13:33:58 +01001274static inline void enable_debug_pagealloc(void)
1275{
1276}
Rafael J. Wysocki8a235ef2008-02-20 01:47:44 +01001277#ifdef CONFIG_HIBERNATION
1278static inline bool kernel_page_present(struct page *page) { return true; }
1279#endif /* CONFIG_HIBERNATION */
Linus Torvalds1da177e2005-04-16 15:20:36 -07001280#endif
1281
1282extern struct vm_area_struct *get_gate_vma(struct task_struct *tsk);
1283#ifdef __HAVE_ARCH_GATE_AREA
1284int in_gate_area_no_task(unsigned long addr);
1285int in_gate_area(struct task_struct *task, unsigned long addr);
1286#else
1287int in_gate_area_no_task(unsigned long addr);
1288#define in_gate_area(task, addr) ({(void)task; in_gate_area_no_task(addr);})
1289#endif /* __HAVE_ARCH_GATE_AREA */
1290
Andrew Morton9d0243b2006-01-08 01:00:39 -08001291int drop_caches_sysctl_handler(struct ctl_table *, int, struct file *,
1292 void __user *, size_t *, loff_t *);
Andrew Morton69e05942006-03-22 00:08:19 -08001293unsigned long shrink_slab(unsigned long scanned, gfp_t gfp_mask,
Andrew Morton9d0243b2006-01-08 01:00:39 -08001294 unsigned long lru_pages);
Andrew Morton9d0243b2006-01-08 01:00:39 -08001295
Luke Yang7a9166e2006-02-20 18:28:07 -08001296#ifndef CONFIG_MMU
1297#define randomize_va_space 0
1298#else
Andi Kleena62eaf12006-02-16 23:41:58 +01001299extern int randomize_va_space;
Luke Yang7a9166e2006-02-20 18:28:07 -08001300#endif
Andi Kleena62eaf12006-02-16 23:41:58 +01001301
Sam Ravnborg045e72a2007-07-26 10:41:13 -07001302const char * arch_vma_name(struct vm_area_struct *vma);
Andi Kleen03252912008-01-30 13:33:18 +01001303void print_vma_addr(char *prefix, unsigned long rip);
Ingo Molnare6e54942006-06-27 02:53:50 -07001304
Yasunori Goto98f3cfc2007-10-16 01:26:14 -07001305struct page *sparse_mem_map_populate(unsigned long pnum, int nid);
Andy Whitcroft29c71112007-10-16 01:24:14 -07001306pgd_t *vmemmap_pgd_populate(unsigned long addr, int node);
1307pud_t *vmemmap_pud_populate(pgd_t *pgd, unsigned long addr, int node);
1308pmd_t *vmemmap_pmd_populate(pud_t *pud, unsigned long addr, int node);
1309pte_t *vmemmap_pte_populate(pmd_t *pmd, unsigned long addr, int node);
Christoph Lameter8f6aac42007-10-16 01:24:13 -07001310void *vmemmap_alloc_block(unsigned long size, int node);
1311void vmemmap_verify(pte_t *, int, unsigned long, unsigned long);
Andy Whitcroft29c71112007-10-16 01:24:14 -07001312int vmemmap_populate_basepages(struct page *start_page,
1313 unsigned long pages, int node);
1314int vmemmap_populate(struct page *start_page, unsigned long pages, int node);
Yinghai Luc2b91e22008-04-12 01:19:24 -07001315void vmemmap_populate_print_last(void);
Christoph Lameter8f6aac42007-10-16 01:24:13 -07001316
Linus Torvalds1da177e2005-04-16 15:20:36 -07001317#endif /* __KERNEL__ */
1318#endif /* _LINUX_MM_H */