mm: remove nopfn
[GitHub/mt8127/android_kernel_alcatel_ttab.git] / include / linux / mm.h
CommitLineData
1da177e4
LT
1#ifndef _LINUX_MM_H
2#define _LINUX_MM_H
3
1da177e4
LT
4#include <linux/errno.h>
5
6#ifdef __KERNEL__
7
1da177e4
LT
8#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>
9a11b49a 13#include <linux/debug_locks.h>
5b99cd0e 14#include <linux/mm_types.h>
1da177e4
LT
15
16struct mempolicy;
17struct anon_vma;
4e950f6f 18struct file_ra_state;
e8edc6e0 19struct user_struct;
4e950f6f 20struct writeback_control;
1da177e4
LT
21
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;
1da177e4
LT
28extern int page_cluster;
29
30#ifdef CONFIG_SYSCTL
31extern int sysctl_legacy_va_layout;
32#else
33#define sysctl_legacy_va_layout 0
34#endif
35
42d7896e
JM
36extern unsigned long mmap_min_addr;
37
1da177e4
LT
38#include <asm/page.h>
39#include <asm/pgtable.h>
40#include <asm/processor.h>
1da177e4 41
1da177e4
LT
42#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
c43692e8
CL
53extern struct kmem_cache *vm_area_cachep;
54
1da177e4
LT
55/*
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
7e2cff42 80/* mprotect() hardcodes VM_MAYREAD >> 4 == VM_READ, and so for r/w/x bits. */
1da177e4
LT
81#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
6aab341e 88#define VM_PFNMAP 0x00000400 /* Page-ranges managed without "struct page", just pure PFN */
1da177e4
LT
89#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() */
0b14c179 101#define VM_RESERVED 0x00080000 /* Count as reserved_vm like IO */
1da177e4
LT
102#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) */
4d7672b4 106#define VM_INSERTPAGE 0x02000000 /* The vma has had "vm_insert_page()" done on it */
e5b97dde 107#define VM_ALWAYSDUMP 0x04000000 /* Always include in core dumps */
d00806b1 108
d0217ac0 109#define VM_CAN_NONLINEAR 0x08000000 /* Has ->fault & does nonlinear pages */
b379d790 110#define VM_MIXEDMAP 0x10000000 /* Can contain "struct page" and pure PFN pages */
aba46c50 111#define VM_SAO 0x20000000 /* Strong Access Ordering (powerpc) */
1da177e4
LT
112
113#ifndef VM_STACK_DEFAULT_FLAGS /* arch can override this */
114#define VM_STACK_DEFAULT_FLAGS VM_DATA_DEFAULT_FLAGS
115#endif
116
117#ifdef CONFIG_STACK_GROWSUP
118#define VM_STACK_FLAGS (VM_GROWSUP | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT)
119#else
120#define VM_STACK_FLAGS (VM_GROWSDOWN | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT)
121#endif
122
123#define VM_READHINTMASK (VM_SEQ_READ | VM_RAND_READ)
124#define VM_ClearReadHint(v) (v)->vm_flags &= ~VM_READHINTMASK
125#define VM_NormalReadHint(v) (!((v)->vm_flags & VM_READHINTMASK))
126#define VM_SequentialReadHint(v) ((v)->vm_flags & VM_SEQ_READ)
127#define VM_RandomReadHint(v) ((v)->vm_flags & VM_RAND_READ)
128
129/*
130 * mapping from the currently active vm_flags protection bits (the
131 * low four bits) to a page protection mask..
132 */
133extern pgprot_t protection_map[16];
134
d0217ac0
NP
135#define FAULT_FLAG_WRITE 0x01 /* Fault was a write access */
136#define FAULT_FLAG_NONLINEAR 0x02 /* Fault was via a nonlinear mapping */
137
138
54cb8821 139/*
d0217ac0 140 * vm_fault is filled by the the pagefault handler and passed to the vma's
83c54070
NP
141 * ->fault function. The vma's ->fault is responsible for returning a bitmask
142 * of VM_FAULT_xxx flags that give details about how the fault was handled.
54cb8821 143 *
d0217ac0
NP
144 * pgoff should be used in favour of virtual_address, if possible. If pgoff
145 * is used, one may set VM_CAN_NONLINEAR in the vma->vm_flags to get nonlinear
146 * mapping support.
54cb8821 147 */
d0217ac0
NP
148struct vm_fault {
149 unsigned int flags; /* FAULT_FLAG_xxx flags */
150 pgoff_t pgoff; /* Logical page offset based on vma */
151 void __user *virtual_address; /* Faulting virtual address */
152
153 struct page *page; /* ->fault handlers should return a
83c54070 154 * page here, unless VM_FAULT_NOPAGE
d0217ac0 155 * is set (which is also implied by
83c54070 156 * VM_FAULT_ERROR).
d0217ac0 157 */
54cb8821 158};
1da177e4
LT
159
160/*
161 * These are the virtual MM functions - opening of an area, closing and
162 * unmapping it (needed to keep files on disk up-to-date etc), pointer
163 * to the functions called when a no-page or a wp-page exception occurs.
164 */
165struct vm_operations_struct {
166 void (*open)(struct vm_area_struct * area);
167 void (*close)(struct vm_area_struct * area);
d0217ac0 168 int (*fault)(struct vm_area_struct *vma, struct vm_fault *vmf);
9637a5ef
DH
169
170 /* notification that a previously read-only page is about to become
171 * writable, if an error is returned it will cause a SIGBUS */
172 int (*page_mkwrite)(struct vm_area_struct *vma, struct page *page);
1da177e4 173#ifdef CONFIG_NUMA
a6020ed7
LS
174 /*
175 * set_policy() op must add a reference to any non-NULL @new mempolicy
176 * to hold the policy upon return. Caller should pass NULL @new to
177 * remove a policy and fall back to surrounding context--i.e. do not
178 * install a MPOL_DEFAULT policy, nor the task or system default
179 * mempolicy.
180 */
1da177e4 181 int (*set_policy)(struct vm_area_struct *vma, struct mempolicy *new);
a6020ed7
LS
182
183 /*
184 * get_policy() op must add reference [mpol_get()] to any policy at
185 * (vma,addr) marked as MPOL_SHARED. The shared policy infrastructure
186 * in mm/mempolicy.c will do this automatically.
187 * get_policy() must NOT add a ref if the policy at (vma,addr) is not
188 * marked as MPOL_SHARED. vma policies are protected by the mmap_sem.
189 * If no [shared/vma] mempolicy exists at the addr, get_policy() op
190 * must return NULL--i.e., do not "fallback" to task or system default
191 * policy.
192 */
1da177e4
LT
193 struct mempolicy *(*get_policy)(struct vm_area_struct *vma,
194 unsigned long addr);
7b2259b3
CL
195 int (*migrate)(struct vm_area_struct *vma, const nodemask_t *from,
196 const nodemask_t *to, unsigned long flags);
1da177e4
LT
197#endif
198};
199
200struct mmu_gather;
201struct inode;
202
349aef0b
AM
203#define page_private(page) ((page)->private)
204#define set_page_private(page, v) ((page)->private = (v))
4c21e2f2 205
1da177e4
LT
206/*
207 * FIXME: take this include out, include page-flags.h in
208 * files which need it (119 of them)
209 */
210#include <linux/page-flags.h>
211
725d704e
NP
212#ifdef CONFIG_DEBUG_VM
213#define VM_BUG_ON(cond) BUG_ON(cond)
214#else
215#define VM_BUG_ON(condition) do { } while(0)
216#endif
217
1da177e4
LT
218/*
219 * Methods to modify the page usage count.
220 *
221 * What counts for a page usage:
222 * - cache mapping (page->mapping)
223 * - private data (page->private)
224 * - page mapped in a task's page tables, each mapping
225 * is counted separately
226 *
227 * Also, many kernel routines increase the page count before a critical
228 * routine so they can be sure the page doesn't go away from under them.
1da177e4
LT
229 */
230
231/*
da6052f7 232 * Drop a ref, return true if the refcount fell to zero (the page has no users)
1da177e4 233 */
7c8ee9a8
NP
234static inline int put_page_testzero(struct page *page)
235{
725d704e 236 VM_BUG_ON(atomic_read(&page->_count) == 0);
8dc04efb 237 return atomic_dec_and_test(&page->_count);
7c8ee9a8 238}
1da177e4
LT
239
240/*
7c8ee9a8
NP
241 * Try to grab a ref unless the page has a refcount of zero, return false if
242 * that is the case.
1da177e4 243 */
7c8ee9a8
NP
244static inline int get_page_unless_zero(struct page *page)
245{
aec2c3ed 246 VM_BUG_ON(PageTail(page));
8dc04efb 247 return atomic_inc_not_zero(&page->_count);
7c8ee9a8 248}
1da177e4 249
48667e7a 250/* Support for virtually mapped pages */
b3bdda02
CL
251struct page *vmalloc_to_page(const void *addr);
252unsigned long vmalloc_to_pfn(const void *addr);
48667e7a 253
0738c4bb
PM
254/*
255 * Determine if an address is within the vmalloc range
256 *
257 * On nommu, vmalloc/vfree wrap through kmalloc/kfree directly, so there
258 * is no special casing required.
259 */
9e2779fa
CL
260static inline int is_vmalloc_addr(const void *x)
261{
0738c4bb 262#ifdef CONFIG_MMU
9e2779fa
CL
263 unsigned long addr = (unsigned long)x;
264
265 return addr >= VMALLOC_START && addr < VMALLOC_END;
0738c4bb
PM
266#else
267 return 0;
8ca3ed87 268#endif
0738c4bb 269}
9e2779fa 270
d85f3385
CL
271static inline struct page *compound_head(struct page *page)
272{
6d777953 273 if (unlikely(PageTail(page)))
d85f3385
CL
274 return page->first_page;
275 return page;
276}
277
4c21e2f2 278static inline int page_count(struct page *page)
1da177e4 279{
d85f3385 280 return atomic_read(&compound_head(page)->_count);
1da177e4
LT
281}
282
283static inline void get_page(struct page *page)
284{
d85f3385 285 page = compound_head(page);
725d704e 286 VM_BUG_ON(atomic_read(&page->_count) == 0);
1da177e4
LT
287 atomic_inc(&page->_count);
288}
289
b49af68f
CL
290static inline struct page *virt_to_head_page(const void *x)
291{
292 struct page *page = virt_to_page(x);
293 return compound_head(page);
294}
295
7835e98b
NP
296/*
297 * Setup the page count before being freed into the page allocator for
298 * the first time (boot or memory hotplug)
299 */
300static inline void init_page_count(struct page *page)
301{
302 atomic_set(&page->_count, 1);
303}
304
1da177e4 305void put_page(struct page *page);
1d7ea732 306void put_pages_list(struct list_head *pages);
1da177e4 307
8dfcc9ba 308void split_page(struct page *page, unsigned int order);
8dfcc9ba 309
33f2ef89
AW
310/*
311 * Compound pages have a destructor function. Provide a
312 * prototype for that function and accessor functions.
313 * These are _only_ valid on the head of a PG_compound page.
314 */
315typedef void compound_page_dtor(struct page *);
316
317static inline void set_compound_page_dtor(struct page *page,
318 compound_page_dtor *dtor)
319{
320 page[1].lru.next = (void *)dtor;
321}
322
323static inline compound_page_dtor *get_compound_page_dtor(struct page *page)
324{
325 return (compound_page_dtor *)page[1].lru.next;
326}
327
d85f3385
CL
328static inline int compound_order(struct page *page)
329{
6d777953 330 if (!PageHead(page))
d85f3385
CL
331 return 0;
332 return (unsigned long)page[1].lru.prev;
333}
334
335static inline void set_compound_order(struct page *page, unsigned long order)
336{
337 page[1].lru.prev = (void *)order;
338}
339
1da177e4
LT
340/*
341 * Multiple processes may "see" the same page. E.g. for untouched
342 * mappings of /dev/null, all processes see the same page full of
343 * zeroes, and text pages of executables and shared libraries have
344 * only one copy in memory, at most, normally.
345 *
346 * For the non-reserved pages, page_count(page) denotes a reference count.
7e871b6c
PBG
347 * page_count() == 0 means the page is free. page->lru is then used for
348 * freelist management in the buddy allocator.
da6052f7 349 * page_count() > 0 means the page has been allocated.
1da177e4 350 *
da6052f7
NP
351 * Pages are allocated by the slab allocator in order to provide memory
352 * to kmalloc and kmem_cache_alloc. In this case, the management of the
353 * page, and the fields in 'struct page' are the responsibility of mm/slab.c
354 * unless a particular usage is carefully commented. (the responsibility of
355 * freeing the kmalloc memory is the caller's, of course).
1da177e4 356 *
da6052f7
NP
357 * A page may be used by anyone else who does a __get_free_page().
358 * In this case, page_count still tracks the references, and should only
359 * be used through the normal accessor functions. The top bits of page->flags
360 * and page->virtual store page management information, but all other fields
361 * are unused and could be used privately, carefully. The management of this
362 * page is the responsibility of the one who allocated it, and those who have
363 * subsequently been given references to it.
364 *
365 * The other pages (we may call them "pagecache pages") are completely
1da177e4
LT
366 * managed by the Linux memory manager: I/O, buffers, swapping etc.
367 * The following discussion applies only to them.
368 *
da6052f7
NP
369 * A pagecache page contains an opaque `private' member, which belongs to the
370 * page's address_space. Usually, this is the address of a circular list of
371 * the page's disk buffers. PG_private must be set to tell the VM to call
372 * into the filesystem to release these pages.
1da177e4 373 *
da6052f7
NP
374 * A page may belong to an inode's memory mapping. In this case, page->mapping
375 * is the pointer to the inode, and page->index is the file offset of the page,
376 * in units of PAGE_CACHE_SIZE.
1da177e4 377 *
da6052f7
NP
378 * If pagecache pages are not associated with an inode, they are said to be
379 * anonymous pages. These may become associated with the swapcache, and in that
380 * case PG_swapcache is set, and page->private is an offset into the swapcache.
1da177e4 381 *
da6052f7
NP
382 * In either case (swapcache or inode backed), the pagecache itself holds one
383 * reference to the page. Setting PG_private should also increment the
384 * refcount. The each user mapping also has a reference to the page.
1da177e4 385 *
da6052f7
NP
386 * The pagecache pages are stored in a per-mapping radix tree, which is
387 * rooted at mapping->page_tree, and indexed by offset.
388 * Where 2.4 and early 2.6 kernels kept dirty/clean pages in per-address_space
389 * lists, we instead now tag pages as dirty/writeback in the radix tree.
1da177e4 390 *
da6052f7 391 * All pagecache pages may be subject to I/O:
1da177e4
LT
392 * - inode pages may need to be read from disk,
393 * - inode pages which have been modified and are MAP_SHARED may need
da6052f7
NP
394 * to be written back to the inode on disk,
395 * - anonymous pages (including MAP_PRIVATE file mappings) which have been
396 * modified may need to be swapped out to swap space and (later) to be read
397 * back into memory.
1da177e4
LT
398 */
399
400/*
401 * The zone field is never updated after free_area_init_core()
402 * sets it, so none of the operations on it need to be atomic.
1da177e4 403 */
348f8b6c 404
d41dee36
AW
405
406/*
407 * page->flags layout:
408 *
409 * There are three possibilities for how page->flags get
410 * laid out. The first is for the normal case, without
411 * sparsemem. The second is for sparsemem when there is
412 * plenty of space for node and section. The last is when
413 * we have run out of space and have to fall back to an
414 * alternate (slower) way of determining the node.
415 *
308c05e3
CL
416 * No sparsemem or sparsemem vmemmap: | NODE | ZONE | ... | FLAGS |
417 * classic sparse with space for node:| SECTION | NODE | ZONE | ... | FLAGS |
418 * classic sparse no space for node: | SECTION | ZONE | ... | FLAGS |
d41dee36 419 */
308c05e3 420#if defined(CONFIG_SPARSEMEM) && !defined(CONFIG_SPARSEMEM_VMEMMAP)
d41dee36
AW
421#define SECTIONS_WIDTH SECTIONS_SHIFT
422#else
423#define SECTIONS_WIDTH 0
424#endif
425
426#define ZONES_WIDTH ZONES_SHIFT
427
9223b419 428#if SECTIONS_WIDTH+ZONES_WIDTH+NODES_SHIFT <= BITS_PER_LONG - NR_PAGEFLAGS
d41dee36
AW
429#define NODES_WIDTH NODES_SHIFT
430#else
308c05e3
CL
431#ifdef CONFIG_SPARSEMEM_VMEMMAP
432#error "Vmemmap: No space for nodes field in page flags"
433#endif
d41dee36
AW
434#define NODES_WIDTH 0
435#endif
436
437/* Page flags: | [SECTION] | [NODE] | ZONE | ... | FLAGS | */
07808b74 438#define SECTIONS_PGOFF ((sizeof(unsigned long)*8) - SECTIONS_WIDTH)
d41dee36
AW
439#define NODES_PGOFF (SECTIONS_PGOFF - NODES_WIDTH)
440#define ZONES_PGOFF (NODES_PGOFF - ZONES_WIDTH)
441
442/*
443 * We are going to use the flags for the page to node mapping if its in
444 * there. This includes the case where there is no node, so it is implicit.
445 */
89689ae7
CL
446#if !(NODES_WIDTH > 0 || NODES_SHIFT == 0)
447#define NODE_NOT_IN_PAGE_FLAGS
448#endif
d41dee36
AW
449
450#ifndef PFN_SECTION_SHIFT
451#define PFN_SECTION_SHIFT 0
452#endif
348f8b6c
DH
453
454/*
455 * Define the bit shifts to access each section. For non-existant
456 * sections we define the shift as 0; that plus a 0 mask ensures
457 * the compiler will optimise away reference to them.
458 */
d41dee36
AW
459#define SECTIONS_PGSHIFT (SECTIONS_PGOFF * (SECTIONS_WIDTH != 0))
460#define NODES_PGSHIFT (NODES_PGOFF * (NODES_WIDTH != 0))
461#define ZONES_PGSHIFT (ZONES_PGOFF * (ZONES_WIDTH != 0))
348f8b6c 462
89689ae7
CL
463/* NODE:ZONE or SECTION:ZONE is used to ID a zone for the buddy allcator */
464#ifdef NODE_NOT_IN_PAGEFLAGS
465#define ZONEID_SHIFT (SECTIONS_SHIFT + ZONES_SHIFT)
bd8029b6
AW
466#define ZONEID_PGOFF ((SECTIONS_PGOFF < ZONES_PGOFF)? \
467 SECTIONS_PGOFF : ZONES_PGOFF)
d41dee36 468#else
89689ae7 469#define ZONEID_SHIFT (NODES_SHIFT + ZONES_SHIFT)
bd8029b6
AW
470#define ZONEID_PGOFF ((NODES_PGOFF < ZONES_PGOFF)? \
471 NODES_PGOFF : ZONES_PGOFF)
89689ae7
CL
472#endif
473
bd8029b6 474#define ZONEID_PGSHIFT (ZONEID_PGOFF * (ZONEID_SHIFT != 0))
348f8b6c 475
9223b419
CL
476#if SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > BITS_PER_LONG - NR_PAGEFLAGS
477#error SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > BITS_PER_LONG - NR_PAGEFLAGS
348f8b6c
DH
478#endif
479
d41dee36
AW
480#define ZONES_MASK ((1UL << ZONES_WIDTH) - 1)
481#define NODES_MASK ((1UL << NODES_WIDTH) - 1)
482#define SECTIONS_MASK ((1UL << SECTIONS_WIDTH) - 1)
89689ae7 483#define ZONEID_MASK ((1UL << ZONEID_SHIFT) - 1)
348f8b6c 484
2f1b6248 485static inline enum zone_type page_zonenum(struct page *page)
1da177e4 486{
348f8b6c 487 return (page->flags >> ZONES_PGSHIFT) & ZONES_MASK;
1da177e4 488}
1da177e4 489
89689ae7
CL
490/*
491 * The identification function is only used by the buddy allocator for
492 * determining if two pages could be buddies. We are not really
493 * identifying a zone since we could be using a the section number
494 * id if we have not node id available in page flags.
495 * We guarantee only that it will return the same value for two
496 * combinable pages in a zone.
497 */
cb2b95e1
AW
498static inline int page_zone_id(struct page *page)
499{
89689ae7 500 return (page->flags >> ZONEID_PGSHIFT) & ZONEID_MASK;
348f8b6c
DH
501}
502
25ba77c1 503static inline int zone_to_nid(struct zone *zone)
89fa3024 504{
d5f541ed
CL
505#ifdef CONFIG_NUMA
506 return zone->node;
507#else
508 return 0;
509#endif
89fa3024
CL
510}
511
89689ae7 512#ifdef NODE_NOT_IN_PAGE_FLAGS
25ba77c1 513extern int page_to_nid(struct page *page);
89689ae7 514#else
25ba77c1 515static inline int page_to_nid(struct page *page)
d41dee36 516{
89689ae7 517 return (page->flags >> NODES_PGSHIFT) & NODES_MASK;
d41dee36 518}
89689ae7
CL
519#endif
520
521static inline struct zone *page_zone(struct page *page)
522{
523 return &NODE_DATA(page_to_nid(page))->node_zones[page_zonenum(page)];
524}
525
308c05e3 526#if defined(CONFIG_SPARSEMEM) && !defined(CONFIG_SPARSEMEM_VMEMMAP)
d41dee36
AW
527static inline unsigned long page_to_section(struct page *page)
528{
529 return (page->flags >> SECTIONS_PGSHIFT) & SECTIONS_MASK;
530}
308c05e3 531#endif
d41dee36 532
2f1b6248 533static inline void set_page_zone(struct page *page, enum zone_type zone)
348f8b6c
DH
534{
535 page->flags &= ~(ZONES_MASK << ZONES_PGSHIFT);
536 page->flags |= (zone & ZONES_MASK) << ZONES_PGSHIFT;
537}
2f1b6248 538
348f8b6c
DH
539static inline void set_page_node(struct page *page, unsigned long node)
540{
541 page->flags &= ~(NODES_MASK << NODES_PGSHIFT);
542 page->flags |= (node & NODES_MASK) << NODES_PGSHIFT;
1da177e4 543}
89689ae7 544
d41dee36
AW
545static inline void set_page_section(struct page *page, unsigned long section)
546{
547 page->flags &= ~(SECTIONS_MASK << SECTIONS_PGSHIFT);
548 page->flags |= (section & SECTIONS_MASK) << SECTIONS_PGSHIFT;
549}
1da177e4 550
2f1b6248 551static inline void set_page_links(struct page *page, enum zone_type zone,
d41dee36 552 unsigned long node, unsigned long pfn)
1da177e4 553{
348f8b6c
DH
554 set_page_zone(page, zone);
555 set_page_node(page, node);
d41dee36 556 set_page_section(page, pfn_to_section_nr(pfn));
1da177e4
LT
557}
558
7cd94146
EP
559/*
560 * If a hint addr is less than mmap_min_addr change hint to be as
561 * low as possible but still greater than mmap_min_addr
562 */
563static inline unsigned long round_hint_to_min(unsigned long hint)
564{
565#ifdef CONFIG_SECURITY
566 hint &= PAGE_MASK;
567 if (((void *)hint != NULL) &&
568 (hint < mmap_min_addr))
569 return PAGE_ALIGN(mmap_min_addr);
570#endif
571 return hint;
572}
573
f6ac2354
CL
574/*
575 * Some inline functions in vmstat.h depend on page_zone()
576 */
577#include <linux/vmstat.h>
578
652050ae 579static __always_inline void *lowmem_page_address(struct page *page)
1da177e4
LT
580{
581 return __va(page_to_pfn(page) << PAGE_SHIFT);
582}
583
584#if defined(CONFIG_HIGHMEM) && !defined(WANT_PAGE_VIRTUAL)
585#define HASHED_PAGE_VIRTUAL
586#endif
587
588#if defined(WANT_PAGE_VIRTUAL)
589#define page_address(page) ((page)->virtual)
590#define set_page_address(page, address) \
591 do { \
592 (page)->virtual = (address); \
593 } while(0)
594#define page_address_init() do { } while(0)
595#endif
596
597#if defined(HASHED_PAGE_VIRTUAL)
598void *page_address(struct page *page);
599void set_page_address(struct page *page, void *virtual);
600void page_address_init(void);
601#endif
602
603#if !defined(HASHED_PAGE_VIRTUAL) && !defined(WANT_PAGE_VIRTUAL)
604#define page_address(page) lowmem_page_address(page)
605#define set_page_address(page, address) do { } while(0)
606#define page_address_init() do { } while(0)
607#endif
608
609/*
610 * On an anonymous page mapped into a user virtual memory area,
611 * page->mapping points to its anon_vma, not to a struct address_space;
612 * with the PAGE_MAPPING_ANON bit set to distinguish it.
613 *
614 * Please note that, confusingly, "page_mapping" refers to the inode
615 * address_space which maps the page from disk; whereas "page_mapped"
616 * refers to user virtual address space into which the page is mapped.
617 */
618#define PAGE_MAPPING_ANON 1
619
620extern struct address_space swapper_space;
621static inline struct address_space *page_mapping(struct page *page)
622{
623 struct address_space *mapping = page->mapping;
624
b5fab14e 625 VM_BUG_ON(PageSlab(page));
726b8012 626#ifdef CONFIG_SWAP
1da177e4
LT
627 if (unlikely(PageSwapCache(page)))
628 mapping = &swapper_space;
726b8012
AM
629 else
630#endif
631 if (unlikely((unsigned long)mapping & PAGE_MAPPING_ANON))
1da177e4
LT
632 mapping = NULL;
633 return mapping;
634}
635
636static inline int PageAnon(struct page *page)
637{
638 return ((unsigned long)page->mapping & PAGE_MAPPING_ANON) != 0;
639}
640
641/*
642 * Return the pagecache index of the passed page. Regular pagecache pages
643 * use ->index whereas swapcache pages use ->private
644 */
645static inline pgoff_t page_index(struct page *page)
646{
647 if (unlikely(PageSwapCache(page)))
4c21e2f2 648 return page_private(page);
1da177e4
LT
649 return page->index;
650}
651
652/*
653 * The atomic page->_mapcount, like _count, starts from -1:
654 * so that transitions both from it and to it can be tracked,
655 * using atomic_inc_and_test and atomic_add_negative(-1).
656 */
657static inline void reset_page_mapcount(struct page *page)
658{
659 atomic_set(&(page)->_mapcount, -1);
660}
661
662static inline int page_mapcount(struct page *page)
663{
664 return atomic_read(&(page)->_mapcount) + 1;
665}
666
667/*
668 * Return true if this page is mapped into pagetables.
669 */
670static inline int page_mapped(struct page *page)
671{
672 return atomic_read(&(page)->_mapcount) >= 0;
673}
674
1da177e4
LT
675/*
676 * Different kinds of faults, as returned by handle_mm_fault().
677 * Used to decide whether a process gets delivered SIGBUS or
678 * just gets major/minor fault counters bumped up.
679 */
d0217ac0 680
83c54070 681#define VM_FAULT_MINOR 0 /* For backwards compat. Remove me quickly. */
d0217ac0 682
83c54070
NP
683#define VM_FAULT_OOM 0x0001
684#define VM_FAULT_SIGBUS 0x0002
685#define VM_FAULT_MAJOR 0x0004
686#define VM_FAULT_WRITE 0x0008 /* Special case for get_user_pages */
f33ea7f4 687
83c54070
NP
688#define VM_FAULT_NOPAGE 0x0100 /* ->fault installed the pte, not return page */
689#define VM_FAULT_LOCKED 0x0200 /* ->fault locked the returned page */
1da177e4 690
83c54070 691#define VM_FAULT_ERROR (VM_FAULT_OOM | VM_FAULT_SIGBUS)
d0217ac0 692
1da177e4
LT
693#define offset_in_page(p) ((unsigned long)(p) & ~PAGE_MASK)
694
695extern void show_free_areas(void);
696
697#ifdef CONFIG_SHMEM
1da177e4
LT
698int shmem_lock(struct file *file, int lock, struct user_struct *user);
699#else
03b00ebc
RK
700static inline int shmem_lock(struct file *file, int lock,
701 struct user_struct *user)
702{
703 return 0;
704}
1da177e4
LT
705#endif
706struct file *shmem_file_setup(char *name, loff_t size, unsigned long flags);
707
708int shmem_zero_setup(struct vm_area_struct *);
709
b0e15190
DH
710#ifndef CONFIG_MMU
711extern unsigned long shmem_get_unmapped_area(struct file *file,
712 unsigned long addr,
713 unsigned long len,
714 unsigned long pgoff,
715 unsigned long flags);
716#endif
717
e8edc6e0 718extern int can_do_mlock(void);
1da177e4
LT
719extern int user_shm_lock(size_t, struct user_struct *);
720extern void user_shm_unlock(size_t, struct user_struct *);
721
722/*
723 * Parameter block passed down to zap_pte_range in exceptional cases.
724 */
725struct zap_details {
726 struct vm_area_struct *nonlinear_vma; /* Check page->index if set */
727 struct address_space *check_mapping; /* Check page->mapping if set */
728 pgoff_t first_index; /* Lowest page->index to unmap */
729 pgoff_t last_index; /* Highest page->index to unmap */
730 spinlock_t *i_mmap_lock; /* For unmap_mapping_range: */
1da177e4
LT
731 unsigned long truncate_count; /* Compare vm_truncate_count */
732};
733
7e675137
NP
734struct page *vm_normal_page(struct vm_area_struct *vma, unsigned long addr,
735 pte_t pte);
736
ee39b37b 737unsigned long zap_page_range(struct vm_area_struct *vma, unsigned long address,
1da177e4 738 unsigned long size, struct zap_details *);
508034a3 739unsigned long unmap_vmas(struct mmu_gather **tlb,
1da177e4
LT
740 struct vm_area_struct *start_vma, unsigned long start_addr,
741 unsigned long end_addr, unsigned long *nr_accounted,
742 struct zap_details *);
e6473092
MM
743
744/**
745 * mm_walk - callbacks for walk_page_range
746 * @pgd_entry: if set, called for each non-empty PGD (top-level) entry
747 * @pud_entry: if set, called for each non-empty PUD (2nd-level) entry
748 * @pmd_entry: if set, called for each non-empty PMD (3rd-level) entry
749 * @pte_entry: if set, called for each non-empty PTE (4th-level) entry
750 * @pte_hole: if set, called for each hole at all levels
751 *
752 * (see walk_page_range for more details)
753 */
754struct mm_walk {
2165009b
DH
755 int (*pgd_entry)(pgd_t *, unsigned long, unsigned long, struct mm_walk *);
756 int (*pud_entry)(pud_t *, unsigned long, unsigned long, struct mm_walk *);
757 int (*pmd_entry)(pmd_t *, unsigned long, unsigned long, struct mm_walk *);
758 int (*pte_entry)(pte_t *, unsigned long, unsigned long, struct mm_walk *);
759 int (*pte_hole)(unsigned long, unsigned long, struct mm_walk *);
760 struct mm_struct *mm;
761 void *private;
e6473092
MM
762};
763
2165009b
DH
764int walk_page_range(unsigned long addr, unsigned long end,
765 struct mm_walk *walk);
3bf5ee95
HD
766void free_pgd_range(struct mmu_gather **tlb, unsigned long addr,
767 unsigned long end, unsigned long floor, unsigned long ceiling);
768void free_pgtables(struct mmu_gather **tlb, struct vm_area_struct *start_vma,
e0da382c 769 unsigned long floor, unsigned long ceiling);
1da177e4
LT
770int copy_page_range(struct mm_struct *dst, struct mm_struct *src,
771 struct vm_area_struct *vma);
1da177e4
LT
772void unmap_mapping_range(struct address_space *mapping,
773 loff_t const holebegin, loff_t const holelen, int even_cows);
774
775static inline void unmap_shared_mapping_range(struct address_space *mapping,
776 loff_t const holebegin, loff_t const holelen)
777{
778 unmap_mapping_range(mapping, holebegin, holelen, 0);
779}
780
781extern int vmtruncate(struct inode * inode, loff_t offset);
f6b3ec23 782extern int vmtruncate_range(struct inode * inode, loff_t offset, loff_t end);
f33ea7f4 783
7ee1dd3f 784#ifdef CONFIG_MMU
83c54070 785extern int handle_mm_fault(struct mm_struct *mm, struct vm_area_struct *vma,
7ee1dd3f 786 unsigned long address, int write_access);
7ee1dd3f
DH
787#else
788static inline int handle_mm_fault(struct mm_struct *mm,
789 struct vm_area_struct *vma, unsigned long address,
790 int write_access)
791{
792 /* should never happen if there's no MMU */
793 BUG();
794 return VM_FAULT_SIGBUS;
795}
796#endif
f33ea7f4 797
1da177e4
LT
798extern int make_pages_present(unsigned long addr, unsigned long end);
799extern int access_process_vm(struct task_struct *tsk, unsigned long addr, void *buf, int len, int write);
1da177e4
LT
800
801int get_user_pages(struct task_struct *tsk, struct mm_struct *mm, unsigned long start,
802 int len, int write, int force, struct page **pages, struct vm_area_struct **vmas);
b5810039 803void print_bad_pte(struct vm_area_struct *, pte_t, unsigned long);
1da177e4 804
cf9a2ae8
DH
805extern int try_to_release_page(struct page * page, gfp_t gfp_mask);
806extern void do_invalidatepage(struct page *page, unsigned long offset);
807
1da177e4 808int __set_page_dirty_nobuffers(struct page *page);
76719325 809int __set_page_dirty_no_writeback(struct page *page);
1da177e4
LT
810int redirty_page_for_writepage(struct writeback_control *wbc,
811 struct page *page);
b3c97528 812int set_page_dirty(struct page *page);
1da177e4
LT
813int set_page_dirty_lock(struct page *page);
814int clear_page_dirty_for_io(struct page *page);
815
b6a2fea3
OW
816extern unsigned long move_page_tables(struct vm_area_struct *vma,
817 unsigned long old_addr, struct vm_area_struct *new_vma,
818 unsigned long new_addr, unsigned long len);
1da177e4
LT
819extern unsigned long do_mremap(unsigned long addr,
820 unsigned long old_len, unsigned long new_len,
821 unsigned long flags, unsigned long new_addr);
b6a2fea3
OW
822extern int mprotect_fixup(struct vm_area_struct *vma,
823 struct vm_area_struct **pprev, unsigned long start,
824 unsigned long end, unsigned long newflags);
1da177e4
LT
825
826/*
8e1f936b 827 * A callback you can register to apply pressure to ageable caches.
1da177e4 828 *
8e1f936b
RR
829 * 'shrink' is passed a count 'nr_to_scan' and a 'gfpmask'. It should
830 * look through the least-recently-used 'nr_to_scan' entries and
831 * attempt to free them up. It should return the number of objects
832 * which remain in the cache. If it returns -1, it means it cannot do
833 * any scanning at this time (eg. there is a risk of deadlock).
1da177e4 834 *
8e1f936b
RR
835 * The 'gfpmask' refers to the allocation we are currently trying to
836 * fulfil.
837 *
838 * Note that 'shrink' will be passed nr_to_scan == 0 when the VM is
839 * querying the cache size, so a fastpath for that case is appropriate.
1da177e4 840 */
8e1f936b
RR
841struct shrinker {
842 int (*shrink)(int nr_to_scan, gfp_t gfp_mask);
843 int seeks; /* seeks to recreate an obj */
1da177e4 844
8e1f936b
RR
845 /* These are for internal use */
846 struct list_head list;
847 long nr; /* objs pending delete */
848};
849#define DEFAULT_SEEKS 2 /* A good number if you don't know better. */
850extern void register_shrinker(struct shrinker *);
851extern void unregister_shrinker(struct shrinker *);
1da177e4 852
4e950f6f 853int vma_wants_writenotify(struct vm_area_struct *vma);
d08b3851 854
b3c97528 855extern pte_t *get_locked_pte(struct mm_struct *mm, unsigned long addr, spinlock_t **ptl);
c9cfcddf 856
5f22df00
NP
857#ifdef __PAGETABLE_PUD_FOLDED
858static inline int __pud_alloc(struct mm_struct *mm, pgd_t *pgd,
859 unsigned long address)
860{
861 return 0;
862}
863#else
1bb3630e 864int __pud_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address);
5f22df00
NP
865#endif
866
867#ifdef __PAGETABLE_PMD_FOLDED
868static inline int __pmd_alloc(struct mm_struct *mm, pud_t *pud,
869 unsigned long address)
870{
871 return 0;
872}
873#else
1bb3630e 874int __pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address);
5f22df00
NP
875#endif
876
1bb3630e
HD
877int __pte_alloc(struct mm_struct *mm, pmd_t *pmd, unsigned long address);
878int __pte_alloc_kernel(pmd_t *pmd, unsigned long address);
879
1da177e4
LT
880/*
881 * The following ifdef needed to get the 4level-fixup.h header to work.
882 * Remove it when 4level-fixup.h has been removed.
883 */
1bb3630e 884#if defined(CONFIG_MMU) && !defined(__ARCH_HAS_4LEVEL_HACK)
1da177e4
LT
885static inline pud_t *pud_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address)
886{
1bb3630e
HD
887 return (unlikely(pgd_none(*pgd)) && __pud_alloc(mm, pgd, address))?
888 NULL: pud_offset(pgd, address);
1da177e4
LT
889}
890
891static inline pmd_t *pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address)
892{
1bb3630e
HD
893 return (unlikely(pud_none(*pud)) && __pmd_alloc(mm, pud, address))?
894 NULL: pmd_offset(pud, address);
1da177e4 895}
1bb3630e
HD
896#endif /* CONFIG_MMU && !__ARCH_HAS_4LEVEL_HACK */
897
4c21e2f2
HD
898#if NR_CPUS >= CONFIG_SPLIT_PTLOCK_CPUS
899/*
900 * We tuck a spinlock to guard each pagetable page into its struct page,
901 * at page->private, with BUILD_BUG_ON to make sure that this will not
902 * overflow into the next struct page (as it might with DEBUG_SPINLOCK).
903 * When freeing, reset page->mapping so free_pages_check won't complain.
904 */
349aef0b 905#define __pte_lockptr(page) &((page)->ptl)
4c21e2f2
HD
906#define pte_lock_init(_page) do { \
907 spin_lock_init(__pte_lockptr(_page)); \
908} while (0)
909#define pte_lock_deinit(page) ((page)->mapping = NULL)
910#define pte_lockptr(mm, pmd) ({(void)(mm); __pte_lockptr(pmd_page(*(pmd)));})
911#else
912/*
913 * We use mm->page_table_lock to guard all pagetable pages of the mm.
914 */
915#define pte_lock_init(page) do {} while (0)
916#define pte_lock_deinit(page) do {} while (0)
917#define pte_lockptr(mm, pmd) ({(void)(pmd); &(mm)->page_table_lock;})
918#endif /* NR_CPUS < CONFIG_SPLIT_PTLOCK_CPUS */
919
2f569afd
MS
920static inline void pgtable_page_ctor(struct page *page)
921{
922 pte_lock_init(page);
923 inc_zone_page_state(page, NR_PAGETABLE);
924}
925
926static inline void pgtable_page_dtor(struct page *page)
927{
928 pte_lock_deinit(page);
929 dec_zone_page_state(page, NR_PAGETABLE);
930}
931
c74df32c
HD
932#define pte_offset_map_lock(mm, pmd, address, ptlp) \
933({ \
4c21e2f2 934 spinlock_t *__ptl = pte_lockptr(mm, pmd); \
c74df32c
HD
935 pte_t *__pte = pte_offset_map(pmd, address); \
936 *(ptlp) = __ptl; \
937 spin_lock(__ptl); \
938 __pte; \
939})
940
941#define pte_unmap_unlock(pte, ptl) do { \
942 spin_unlock(ptl); \
943 pte_unmap(pte); \
944} while (0)
945
1bb3630e
HD
946#define pte_alloc_map(mm, pmd, address) \
947 ((unlikely(!pmd_present(*(pmd))) && __pte_alloc(mm, pmd, address))? \
948 NULL: pte_offset_map(pmd, address))
949
c74df32c
HD
950#define pte_alloc_map_lock(mm, pmd, address, ptlp) \
951 ((unlikely(!pmd_present(*(pmd))) && __pte_alloc(mm, pmd, address))? \
952 NULL: pte_offset_map_lock(mm, pmd, address, ptlp))
953
1bb3630e
HD
954#define pte_alloc_kernel(pmd, address) \
955 ((unlikely(!pmd_present(*(pmd))) && __pte_alloc_kernel(pmd, address))? \
956 NULL: pte_offset_kernel(pmd, address))
1da177e4
LT
957
958extern void free_area_init(unsigned long * zones_size);
959extern void free_area_init_node(int nid, pg_data_t *pgdat,
960 unsigned long * zones_size, unsigned long zone_start_pfn,
961 unsigned long *zholes_size);
c713216d
MG
962#ifdef CONFIG_ARCH_POPULATES_NODE_MAP
963/*
964 * With CONFIG_ARCH_POPULATES_NODE_MAP set, an architecture may initialise its
965 * zones, allocate the backing mem_map and account for memory holes in a more
966 * architecture independent manner. This is a substitute for creating the
967 * zone_sizes[] and zholes_size[] arrays and passing them to
968 * free_area_init_node()
969 *
970 * An architecture is expected to register range of page frames backed by
971 * physical memory with add_active_range() before calling
972 * free_area_init_nodes() passing in the PFN each zone ends at. At a basic
973 * usage, an architecture is expected to do something like
974 *
975 * unsigned long max_zone_pfns[MAX_NR_ZONES] = {max_dma, max_normal_pfn,
976 * max_highmem_pfn};
977 * for_each_valid_physical_page_range()
978 * add_active_range(node_id, start_pfn, end_pfn)
979 * free_area_init_nodes(max_zone_pfns);
980 *
981 * If the architecture guarantees that there are no holes in the ranges
982 * registered with add_active_range(), free_bootmem_active_regions()
983 * will call free_bootmem_node() for each registered physical page range.
984 * Similarly sparse_memory_present_with_active_regions() calls
985 * memory_present() for each range when SPARSEMEM is enabled.
986 *
987 * See mm/page_alloc.c for more information on each function exposed by
988 * CONFIG_ARCH_POPULATES_NODE_MAP
989 */
990extern void free_area_init_nodes(unsigned long *max_zone_pfn);
991extern void add_active_range(unsigned int nid, unsigned long start_pfn,
992 unsigned long end_pfn);
cc1050ba
YL
993extern void remove_active_range(unsigned int nid, unsigned long start_pfn,
994 unsigned long end_pfn);
fb01439c
MG
995extern void push_node_boundaries(unsigned int nid, unsigned long start_pfn,
996 unsigned long end_pfn);
c713216d
MG
997extern void remove_all_active_ranges(void);
998extern unsigned long absent_pages_in_range(unsigned long start_pfn,
999 unsigned long end_pfn);
1000extern void get_pfn_range_for_nid(unsigned int nid,
1001 unsigned long *start_pfn, unsigned long *end_pfn);
1002extern unsigned long find_min_pfn_with_active_regions(void);
1003extern unsigned long find_max_pfn_with_active_regions(void);
1004extern void free_bootmem_with_active_regions(int nid,
1005 unsigned long max_low_pfn);
d52d53b8 1006typedef int (*work_fn_t)(unsigned long, unsigned long, void *);
b5bc6c0e 1007extern void work_with_active_regions(int nid, work_fn_t work_fn, void *data);
c713216d
MG
1008extern void sparse_memory_present_with_active_regions(int nid);
1009#ifndef CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID
1010extern int early_pfn_to_nid(unsigned long pfn);
1011#endif /* CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID */
1012#endif /* CONFIG_ARCH_POPULATES_NODE_MAP */
0e0b864e 1013extern void set_dma_reserve(unsigned long new_dma_reserve);
a2f3aa02
DH
1014extern void memmap_init_zone(unsigned long, int, unsigned long,
1015 unsigned long, enum memmap_context);
3947be19 1016extern void setup_per_zone_pages_min(void);
1da177e4
LT
1017extern void mem_init(void);
1018extern void show_mem(void);
1019extern void si_meminfo(struct sysinfo * val);
1020extern void si_meminfo_node(struct sysinfo *val, int nid);
3461b0af 1021extern int after_bootmem;
1da177e4 1022
e7c8d5c9
CL
1023#ifdef CONFIG_NUMA
1024extern void setup_per_cpu_pageset(void);
1025#else
1026static inline void setup_per_cpu_pageset(void) {}
1027#endif
1028
1da177e4
LT
1029/* prio_tree.c */
1030void vma_prio_tree_add(struct vm_area_struct *, struct vm_area_struct *old);
1031void vma_prio_tree_insert(struct vm_area_struct *, struct prio_tree_root *);
1032void vma_prio_tree_remove(struct vm_area_struct *, struct prio_tree_root *);
1033struct vm_area_struct *vma_prio_tree_next(struct vm_area_struct *vma,
1034 struct prio_tree_iter *iter);
1035
1036#define vma_prio_tree_foreach(vma, iter, root, begin, end) \
1037 for (prio_tree_iter_init(iter, root, begin, end), vma = NULL; \
1038 (vma = vma_prio_tree_next(vma, iter)); )
1039
1040static inline void vma_nonlinear_insert(struct vm_area_struct *vma,
1041 struct list_head *list)
1042{
1043 vma->shared.vm_set.parent = NULL;
1044 list_add_tail(&vma->shared.vm_set.list, list);
1045}
1046
1047/* mmap.c */
34b4e4aa 1048extern int __vm_enough_memory(struct mm_struct *mm, long pages, int cap_sys_admin);
1da177e4
LT
1049extern void vma_adjust(struct vm_area_struct *vma, unsigned long start,
1050 unsigned long end, pgoff_t pgoff, struct vm_area_struct *insert);
1051extern struct vm_area_struct *vma_merge(struct mm_struct *,
1052 struct vm_area_struct *prev, unsigned long addr, unsigned long end,
1053 unsigned long vm_flags, struct anon_vma *, struct file *, pgoff_t,
1054 struct mempolicy *);
1055extern struct anon_vma *find_mergeable_anon_vma(struct vm_area_struct *);
1056extern int split_vma(struct mm_struct *,
1057 struct vm_area_struct *, unsigned long addr, int new_below);
1058extern int insert_vm_struct(struct mm_struct *, struct vm_area_struct *);
1059extern void __vma_link_rb(struct mm_struct *, struct vm_area_struct *,
1060 struct rb_node **, struct rb_node *);
a8fb5618 1061extern void unlink_file_vma(struct vm_area_struct *);
1da177e4
LT
1062extern struct vm_area_struct *copy_vma(struct vm_area_struct **,
1063 unsigned long addr, unsigned long len, pgoff_t pgoff);
1064extern void exit_mmap(struct mm_struct *);
925d1c40
MH
1065
1066#ifdef CONFIG_PROC_FS
1067/* From fs/proc/base.c. callers must _not_ hold the mm's exe_file_lock */
1068extern void added_exe_file_vma(struct mm_struct *mm);
1069extern void removed_exe_file_vma(struct mm_struct *mm);
1070#else
1071static inline void added_exe_file_vma(struct mm_struct *mm)
1072{}
1073
1074static inline void removed_exe_file_vma(struct mm_struct *mm)
1075{}
1076#endif /* CONFIG_PROC_FS */
1077
119f657c 1078extern int may_expand_vm(struct mm_struct *mm, unsigned long npages);
fa5dc22f
RM
1079extern int install_special_mapping(struct mm_struct *mm,
1080 unsigned long addr, unsigned long len,
1081 unsigned long flags, struct page **pages);
1da177e4
LT
1082
1083extern unsigned long get_unmapped_area(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
1084
1085extern unsigned long do_mmap_pgoff(struct file *file, unsigned long addr,
1086 unsigned long len, unsigned long prot,
1087 unsigned long flag, unsigned long pgoff);
0165ab44
MS
1088extern unsigned long mmap_region(struct file *file, unsigned long addr,
1089 unsigned long len, unsigned long flags,
1090 unsigned int vm_flags, unsigned long pgoff,
1091 int accountable);
1da177e4
LT
1092
1093static inline unsigned long do_mmap(struct file *file, unsigned long addr,
1094 unsigned long len, unsigned long prot,
1095 unsigned long flag, unsigned long offset)
1096{
1097 unsigned long ret = -EINVAL;
1098 if ((offset + PAGE_ALIGN(len)) < offset)
1099 goto out;
1100 if (!(offset & ~PAGE_MASK))
1101 ret = do_mmap_pgoff(file, addr, len, prot, flag, offset >> PAGE_SHIFT);
1102out:
1103 return ret;
1104}
1105
1106extern int do_munmap(struct mm_struct *, unsigned long, size_t);
1107
1108extern unsigned long do_brk(unsigned long, unsigned long);
1109
1110/* filemap.c */
1111extern unsigned long page_unuse(struct page *);
1112extern void truncate_inode_pages(struct address_space *, loff_t);
d7339071
HR
1113extern void truncate_inode_pages_range(struct address_space *,
1114 loff_t lstart, loff_t lend);
1da177e4
LT
1115
1116/* generic vm_area_ops exported for stackable file systems */
d0217ac0 1117extern int filemap_fault(struct vm_area_struct *, struct vm_fault *);
1da177e4
LT
1118
1119/* mm/page-writeback.c */
1120int write_one_page(struct page *page, int wait);
1121
1122/* readahead.c */
1123#define VM_MAX_READAHEAD 128 /* kbytes */
1124#define VM_MIN_READAHEAD 16 /* kbytes (includes current page) */
1da177e4
LT
1125
1126int do_page_cache_readahead(struct address_space *mapping, struct file *filp,
7361f4d8 1127 pgoff_t offset, unsigned long nr_to_read);
1da177e4 1128int force_page_cache_readahead(struct address_space *mapping, struct file *filp,
7361f4d8 1129 pgoff_t offset, unsigned long nr_to_read);
cf914a7d
RR
1130
1131void page_cache_sync_readahead(struct address_space *mapping,
1132 struct file_ra_state *ra,
1133 struct file *filp,
1134 pgoff_t offset,
1135 unsigned long size);
1136
1137void page_cache_async_readahead(struct address_space *mapping,
1138 struct file_ra_state *ra,
1139 struct file *filp,
1140 struct page *pg,
1141 pgoff_t offset,
1142 unsigned long size);
1143
1da177e4
LT
1144unsigned long max_sane_readahead(unsigned long nr);
1145
1146/* Do stack extension */
46dea3d0 1147extern int expand_stack(struct vm_area_struct *vma, unsigned long address);
9ab88515 1148#ifdef CONFIG_IA64
46dea3d0 1149extern int expand_upwards(struct vm_area_struct *vma, unsigned long address);
9ab88515 1150#endif
b6a2fea3
OW
1151extern int expand_stack_downwards(struct vm_area_struct *vma,
1152 unsigned long address);
1da177e4
LT
1153
1154/* Look up the first VMA which satisfies addr < vm_end, NULL if none. */
1155extern struct vm_area_struct * find_vma(struct mm_struct * mm, unsigned long addr);
1156extern struct vm_area_struct * find_vma_prev(struct mm_struct * mm, unsigned long addr,
1157 struct vm_area_struct **pprev);
1158
1159/* Look up the first VMA which intersects the interval start_addr..end_addr-1,
1160 NULL if none. Assume start_addr < end_addr. */
1161static inline struct vm_area_struct * find_vma_intersection(struct mm_struct * mm, unsigned long start_addr, unsigned long end_addr)
1162{
1163 struct vm_area_struct * vma = find_vma(mm,start_addr);
1164
1165 if (vma && end_addr <= vma->vm_start)
1166 vma = NULL;
1167 return vma;
1168}
1169
1170static inline unsigned long vma_pages(struct vm_area_struct *vma)
1171{
1172 return (vma->vm_end - vma->vm_start) >> PAGE_SHIFT;
1173}
1174
804af2cf 1175pgprot_t vm_get_page_prot(unsigned long vm_flags);
deceb6cd 1176struct vm_area_struct *find_extend_vma(struct mm_struct *, unsigned long addr);
deceb6cd
HD
1177int remap_pfn_range(struct vm_area_struct *, unsigned long addr,
1178 unsigned long pfn, unsigned long size, pgprot_t);
a145dd41 1179int vm_insert_page(struct vm_area_struct *, unsigned long addr, struct page *);
e0dc0d8f
NP
1180int vm_insert_pfn(struct vm_area_struct *vma, unsigned long addr,
1181 unsigned long pfn);
423bad60
NP
1182int vm_insert_mixed(struct vm_area_struct *vma, unsigned long addr,
1183 unsigned long pfn);
deceb6cd 1184
6aab341e 1185struct page *follow_page(struct vm_area_struct *, unsigned long address,
deceb6cd
HD
1186 unsigned int foll_flags);
1187#define FOLL_WRITE 0x01 /* check pte is writable */
1188#define FOLL_TOUCH 0x02 /* mark page accessed */
1189#define FOLL_GET 0x04 /* do get_page on page */
1190#define FOLL_ANON 0x08 /* give ZERO_PAGE if no pgtable */
1da177e4 1191
2f569afd 1192typedef int (*pte_fn_t)(pte_t *pte, pgtable_t token, unsigned long addr,
aee16b3c
JF
1193 void *data);
1194extern int apply_to_page_range(struct mm_struct *mm, unsigned long address,
1195 unsigned long size, pte_fn_t fn, void *data);
1196
1da177e4 1197#ifdef CONFIG_PROC_FS
ab50b8ed 1198void vm_stat_account(struct mm_struct *, unsigned long, struct file *, long);
1da177e4 1199#else
ab50b8ed 1200static inline void vm_stat_account(struct mm_struct *mm,
1da177e4
LT
1201 unsigned long flags, struct file *file, long pages)
1202{
1203}
1204#endif /* CONFIG_PROC_FS */
1205
12d6f21e
IM
1206#ifdef CONFIG_DEBUG_PAGEALLOC
1207extern int debug_pagealloc_enabled;
1208
1209extern void kernel_map_pages(struct page *page, int numpages, int enable);
1210
1211static inline void enable_debug_pagealloc(void)
1212{
1213 debug_pagealloc_enabled = 1;
1214}
8a235efa
RW
1215#ifdef CONFIG_HIBERNATION
1216extern bool kernel_page_present(struct page *page);
1217#endif /* CONFIG_HIBERNATION */
12d6f21e 1218#else
1da177e4 1219static inline void
9858db50 1220kernel_map_pages(struct page *page, int numpages, int enable) {}
12d6f21e
IM
1221static inline void enable_debug_pagealloc(void)
1222{
1223}
8a235efa
RW
1224#ifdef CONFIG_HIBERNATION
1225static inline bool kernel_page_present(struct page *page) { return true; }
1226#endif /* CONFIG_HIBERNATION */
1da177e4
LT
1227#endif
1228
1229extern struct vm_area_struct *get_gate_vma(struct task_struct *tsk);
1230#ifdef __HAVE_ARCH_GATE_AREA
1231int in_gate_area_no_task(unsigned long addr);
1232int in_gate_area(struct task_struct *task, unsigned long addr);
1233#else
1234int in_gate_area_no_task(unsigned long addr);
1235#define in_gate_area(task, addr) ({(void)task; in_gate_area_no_task(addr);})
1236#endif /* __HAVE_ARCH_GATE_AREA */
1237
9d0243bc
AM
1238int drop_caches_sysctl_handler(struct ctl_table *, int, struct file *,
1239 void __user *, size_t *, loff_t *);
69e05944 1240unsigned long shrink_slab(unsigned long scanned, gfp_t gfp_mask,
9d0243bc 1241 unsigned long lru_pages);
9d0243bc 1242
7a9166e3
LY
1243#ifndef CONFIG_MMU
1244#define randomize_va_space 0
1245#else
a62eaf15 1246extern int randomize_va_space;
7a9166e3 1247#endif
a62eaf15 1248
045e72ac 1249const char * arch_vma_name(struct vm_area_struct *vma);
03252919 1250void print_vma_addr(char *prefix, unsigned long rip);
e6e5494c 1251
98f3cfc1 1252struct page *sparse_mem_map_populate(unsigned long pnum, int nid);
29c71111
AW
1253pgd_t *vmemmap_pgd_populate(unsigned long addr, int node);
1254pud_t *vmemmap_pud_populate(pgd_t *pgd, unsigned long addr, int node);
1255pmd_t *vmemmap_pmd_populate(pud_t *pud, unsigned long addr, int node);
1256pte_t *vmemmap_pte_populate(pmd_t *pmd, unsigned long addr, int node);
8f6aac41
CL
1257void *vmemmap_alloc_block(unsigned long size, int node);
1258void vmemmap_verify(pte_t *, int, unsigned long, unsigned long);
29c71111
AW
1259int vmemmap_populate_basepages(struct page *start_page,
1260 unsigned long pages, int node);
1261int vmemmap_populate(struct page *start_page, unsigned long pages, int node);
c2b91e2e 1262void vmemmap_populate_print_last(void);
8f6aac41 1263
1da177e4
LT
1264#endif /* __KERNEL__ */
1265#endif /* _LINUX_MM_H */