fs/super.c: add lockdep annotation to s_umount
[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>
59ea7463 10#include <linux/mmdebug.h>
1da177e4
LT
11#include <linux/mmzone.h>
12#include <linux/rbtree.h>
13#include <linux/prio_tree.h>
9a11b49a 14#include <linux/debug_locks.h>
5b99cd0e 15#include <linux/mm_types.h>
1da177e4
LT
16
17struct mempolicy;
18struct anon_vma;
4e950f6f 19struct file_ra_state;
e8edc6e0 20struct user_struct;
4e950f6f 21struct writeback_control;
1da177e4
LT
22
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;
1da177e4
LT
29extern 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
42d7896e
JM
37extern unsigned long mmap_min_addr;
38
1da177e4
LT
39#include <asm/page.h>
40#include <asm/pgtable.h>
41#include <asm/processor.h>
1da177e4 42
1da177e4
LT
43#define nth_page(page,n) pfn_to_page(page_to_pfn((page)) + (n))
44
27ac792c
AR
45/* to align the pointer to the (next) page boundary */
46#define PAGE_ALIGN(addr) ALIGN(addr, PAGE_SIZE)
47
1da177e4
LT
48/*
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
c43692e8
CL
57extern struct kmem_cache *vm_area_cachep;
58
1da177e4 59#ifndef CONFIG_MMU
8feae131
DH
60extern struct rb_root nommu_region_tree;
61extern struct rw_semaphore nommu_region_sem;
1da177e4
LT
62
63extern unsigned int kobjsize(const void *objp);
64#endif
65
66/*
605d9288 67 * vm_flags in vm_area_struct, see mm_types.h.
1da177e4
LT
68 */
69#define VM_READ 0x00000001 /* currently active flags */
70#define VM_WRITE 0x00000002
71#define VM_EXEC 0x00000004
72#define VM_SHARED 0x00000008
73
7e2cff42 74/* mprotect() hardcodes VM_MAYREAD >> 4 == VM_READ, and so for r/w/x bits. */
1da177e4
LT
75#define VM_MAYREAD 0x00000010 /* limits for mprotect() etc */
76#define VM_MAYWRITE 0x00000020
77#define VM_MAYEXEC 0x00000040
78#define VM_MAYSHARE 0x00000080
79
80#define VM_GROWSDOWN 0x00000100 /* general info on the segment */
81#define VM_GROWSUP 0x00000200
6aab341e 82#define VM_PFNMAP 0x00000400 /* Page-ranges managed without "struct page", just pure PFN */
1da177e4
LT
83#define VM_DENYWRITE 0x00000800 /* ETXTBSY on write attempts.. */
84
85#define VM_EXECUTABLE 0x00001000
86#define VM_LOCKED 0x00002000
87#define VM_IO 0x00004000 /* Memory mapped I/O or similar */
88
89 /* Used by sys_madvise() */
90#define VM_SEQ_READ 0x00008000 /* App will access data sequentially */
91#define VM_RAND_READ 0x00010000 /* App will not benefit from clustered reads */
92
93#define VM_DONTCOPY 0x00020000 /* Do not copy this vma on fork */
94#define VM_DONTEXPAND 0x00040000 /* Cannot expand with mremap() */
0b14c179 95#define VM_RESERVED 0x00080000 /* Count as reserved_vm like IO */
1da177e4 96#define VM_ACCOUNT 0x00100000 /* Is a VM accounted object */
cdfd4325 97#define VM_NORESERVE 0x00200000 /* should the VM suppress accounting */
1da177e4
LT
98#define VM_HUGETLB 0x00400000 /* Huge TLB Page VM */
99#define VM_NONLINEAR 0x00800000 /* Is non-linear (remap_file_pages) */
100#define VM_MAPPED_COPY 0x01000000 /* T if mapped copy of data (nommu mmap) */
4d7672b4 101#define VM_INSERTPAGE 0x02000000 /* The vma has had "vm_insert_page()" done on it */
e5b97dde 102#define VM_ALWAYSDUMP 0x04000000 /* Always include in core dumps */
d00806b1 103
d0217ac0 104#define VM_CAN_NONLINEAR 0x08000000 /* Has ->fault & does nonlinear pages */
b379d790 105#define VM_MIXEDMAP 0x10000000 /* Can contain "struct page" and pure PFN pages */
aba46c50 106#define VM_SAO 0x20000000 /* Strong Access Ordering (powerpc) */
1da177e4
LT
107
108#ifndef VM_STACK_DEFAULT_FLAGS /* arch can override this */
109#define VM_STACK_DEFAULT_FLAGS VM_DATA_DEFAULT_FLAGS
110#endif
111
112#ifdef CONFIG_STACK_GROWSUP
113#define VM_STACK_FLAGS (VM_GROWSUP | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT)
114#else
115#define VM_STACK_FLAGS (VM_GROWSDOWN | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT)
116#endif
117
118#define VM_READHINTMASK (VM_SEQ_READ | VM_RAND_READ)
119#define VM_ClearReadHint(v) (v)->vm_flags &= ~VM_READHINTMASK
120#define VM_NormalReadHint(v) (!((v)->vm_flags & VM_READHINTMASK))
121#define VM_SequentialReadHint(v) ((v)->vm_flags & VM_SEQ_READ)
122#define VM_RandomReadHint(v) ((v)->vm_flags & VM_RAND_READ)
123
b291f000
NP
124/*
125 * special vmas that are non-mergable, non-mlock()able
126 */
127#define VM_SPECIAL (VM_IO | VM_DONTEXPAND | VM_RESERVED | VM_PFNMAP)
128
1da177e4
LT
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
6bd9cd50 138/*
139 * This interface is used by x86 PAT code to identify a pfn mapping that is
140 * linear over entire vma. This is to optimize PAT code that deals with
141 * marking the physical region with a particular prot. This is not for generic
142 * mm use. Note also that this check will not work if the pfn mapping is
143 * linear for a vma starting at physical address 0. In which case PAT code
144 * falls back to slow path of reserving physical range page by page.
145 */
3c8bb73a 146static inline int is_linear_pfn_mapping(struct vm_area_struct *vma)
147{
148 return ((vma->vm_flags & VM_PFNMAP) && vma->vm_pgoff);
149}
150
151static inline int is_pfn_mapping(struct vm_area_struct *vma)
152{
153 return (vma->vm_flags & VM_PFNMAP);
154}
d0217ac0 155
54cb8821 156/*
d0217ac0 157 * vm_fault is filled by the the pagefault handler and passed to the vma's
83c54070
NP
158 * ->fault function. The vma's ->fault is responsible for returning a bitmask
159 * of VM_FAULT_xxx flags that give details about how the fault was handled.
54cb8821 160 *
d0217ac0
NP
161 * pgoff should be used in favour of virtual_address, if possible. If pgoff
162 * is used, one may set VM_CAN_NONLINEAR in the vma->vm_flags to get nonlinear
163 * mapping support.
54cb8821 164 */
d0217ac0
NP
165struct vm_fault {
166 unsigned int flags; /* FAULT_FLAG_xxx flags */
167 pgoff_t pgoff; /* Logical page offset based on vma */
168 void __user *virtual_address; /* Faulting virtual address */
169
170 struct page *page; /* ->fault handlers should return a
83c54070 171 * page here, unless VM_FAULT_NOPAGE
d0217ac0 172 * is set (which is also implied by
83c54070 173 * VM_FAULT_ERROR).
d0217ac0 174 */
54cb8821 175};
1da177e4
LT
176
177/*
178 * These are the virtual MM functions - opening of an area, closing and
179 * unmapping it (needed to keep files on disk up-to-date etc), pointer
180 * to the functions called when a no-page or a wp-page exception occurs.
181 */
182struct vm_operations_struct {
183 void (*open)(struct vm_area_struct * area);
184 void (*close)(struct vm_area_struct * area);
d0217ac0 185 int (*fault)(struct vm_area_struct *vma, struct vm_fault *vmf);
9637a5ef
DH
186
187 /* notification that a previously read-only page is about to become
188 * writable, if an error is returned it will cause a SIGBUS */
189 int (*page_mkwrite)(struct vm_area_struct *vma, struct page *page);
28b2ee20
RR
190
191 /* called by access_process_vm when get_user_pages() fails, typically
192 * for use by special VMAs that can switch between memory and hardware
193 */
194 int (*access)(struct vm_area_struct *vma, unsigned long addr,
195 void *buf, int len, int write);
1da177e4 196#ifdef CONFIG_NUMA
a6020ed7
LS
197 /*
198 * set_policy() op must add a reference to any non-NULL @new mempolicy
199 * to hold the policy upon return. Caller should pass NULL @new to
200 * remove a policy and fall back to surrounding context--i.e. do not
201 * install a MPOL_DEFAULT policy, nor the task or system default
202 * mempolicy.
203 */
1da177e4 204 int (*set_policy)(struct vm_area_struct *vma, struct mempolicy *new);
a6020ed7
LS
205
206 /*
207 * get_policy() op must add reference [mpol_get()] to any policy at
208 * (vma,addr) marked as MPOL_SHARED. The shared policy infrastructure
209 * in mm/mempolicy.c will do this automatically.
210 * get_policy() must NOT add a ref if the policy at (vma,addr) is not
211 * marked as MPOL_SHARED. vma policies are protected by the mmap_sem.
212 * If no [shared/vma] mempolicy exists at the addr, get_policy() op
213 * must return NULL--i.e., do not "fallback" to task or system default
214 * policy.
215 */
1da177e4
LT
216 struct mempolicy *(*get_policy)(struct vm_area_struct *vma,
217 unsigned long addr);
7b2259b3
CL
218 int (*migrate)(struct vm_area_struct *vma, const nodemask_t *from,
219 const nodemask_t *to, unsigned long flags);
1da177e4
LT
220#endif
221};
222
223struct mmu_gather;
224struct inode;
225
349aef0b
AM
226#define page_private(page) ((page)->private)
227#define set_page_private(page, v) ((page)->private = (v))
4c21e2f2 228
1da177e4
LT
229/*
230 * FIXME: take this include out, include page-flags.h in
231 * files which need it (119 of them)
232 */
233#include <linux/page-flags.h>
234
235/*
236 * Methods to modify the page usage count.
237 *
238 * What counts for a page usage:
239 * - cache mapping (page->mapping)
240 * - private data (page->private)
241 * - page mapped in a task's page tables, each mapping
242 * is counted separately
243 *
244 * Also, many kernel routines increase the page count before a critical
245 * routine so they can be sure the page doesn't go away from under them.
1da177e4
LT
246 */
247
248/*
da6052f7 249 * Drop a ref, return true if the refcount fell to zero (the page has no users)
1da177e4 250 */
7c8ee9a8
NP
251static inline int put_page_testzero(struct page *page)
252{
725d704e 253 VM_BUG_ON(atomic_read(&page->_count) == 0);
8dc04efb 254 return atomic_dec_and_test(&page->_count);
7c8ee9a8 255}
1da177e4
LT
256
257/*
7c8ee9a8
NP
258 * Try to grab a ref unless the page has a refcount of zero, return false if
259 * that is the case.
1da177e4 260 */
7c8ee9a8
NP
261static inline int get_page_unless_zero(struct page *page)
262{
8dc04efb 263 return atomic_inc_not_zero(&page->_count);
7c8ee9a8 264}
1da177e4 265
48667e7a 266/* Support for virtually mapped pages */
b3bdda02
CL
267struct page *vmalloc_to_page(const void *addr);
268unsigned long vmalloc_to_pfn(const void *addr);
48667e7a 269
0738c4bb
PM
270/*
271 * Determine if an address is within the vmalloc range
272 *
273 * On nommu, vmalloc/vfree wrap through kmalloc/kfree directly, so there
274 * is no special casing required.
275 */
9e2779fa
CL
276static inline int is_vmalloc_addr(const void *x)
277{
0738c4bb 278#ifdef CONFIG_MMU
9e2779fa
CL
279 unsigned long addr = (unsigned long)x;
280
281 return addr >= VMALLOC_START && addr < VMALLOC_END;
0738c4bb
PM
282#else
283 return 0;
8ca3ed87 284#endif
0738c4bb 285}
9e2779fa 286
d85f3385
CL
287static inline struct page *compound_head(struct page *page)
288{
6d777953 289 if (unlikely(PageTail(page)))
d85f3385
CL
290 return page->first_page;
291 return page;
292}
293
4c21e2f2 294static inline int page_count(struct page *page)
1da177e4 295{
d85f3385 296 return atomic_read(&compound_head(page)->_count);
1da177e4
LT
297}
298
299static inline void get_page(struct page *page)
300{
d85f3385 301 page = compound_head(page);
725d704e 302 VM_BUG_ON(atomic_read(&page->_count) == 0);
1da177e4
LT
303 atomic_inc(&page->_count);
304}
305
b49af68f
CL
306static inline struct page *virt_to_head_page(const void *x)
307{
308 struct page *page = virt_to_page(x);
309 return compound_head(page);
310}
311
7835e98b
NP
312/*
313 * Setup the page count before being freed into the page allocator for
314 * the first time (boot or memory hotplug)
315 */
316static inline void init_page_count(struct page *page)
317{
318 atomic_set(&page->_count, 1);
319}
320
1da177e4 321void put_page(struct page *page);
1d7ea732 322void put_pages_list(struct list_head *pages);
1da177e4 323
8dfcc9ba 324void split_page(struct page *page, unsigned int order);
8dfcc9ba 325
33f2ef89
AW
326/*
327 * Compound pages have a destructor function. Provide a
328 * prototype for that function and accessor functions.
329 * These are _only_ valid on the head of a PG_compound page.
330 */
331typedef void compound_page_dtor(struct page *);
332
333static inline void set_compound_page_dtor(struct page *page,
334 compound_page_dtor *dtor)
335{
336 page[1].lru.next = (void *)dtor;
337}
338
339static inline compound_page_dtor *get_compound_page_dtor(struct page *page)
340{
341 return (compound_page_dtor *)page[1].lru.next;
342}
343
d85f3385
CL
344static inline int compound_order(struct page *page)
345{
6d777953 346 if (!PageHead(page))
d85f3385
CL
347 return 0;
348 return (unsigned long)page[1].lru.prev;
349}
350
351static inline void set_compound_order(struct page *page, unsigned long order)
352{
353 page[1].lru.prev = (void *)order;
354}
355
1da177e4
LT
356/*
357 * Multiple processes may "see" the same page. E.g. for untouched
358 * mappings of /dev/null, all processes see the same page full of
359 * zeroes, and text pages of executables and shared libraries have
360 * only one copy in memory, at most, normally.
361 *
362 * For the non-reserved pages, page_count(page) denotes a reference count.
7e871b6c
PBG
363 * page_count() == 0 means the page is free. page->lru is then used for
364 * freelist management in the buddy allocator.
da6052f7 365 * page_count() > 0 means the page has been allocated.
1da177e4 366 *
da6052f7
NP
367 * Pages are allocated by the slab allocator in order to provide memory
368 * to kmalloc and kmem_cache_alloc. In this case, the management of the
369 * page, and the fields in 'struct page' are the responsibility of mm/slab.c
370 * unless a particular usage is carefully commented. (the responsibility of
371 * freeing the kmalloc memory is the caller's, of course).
1da177e4 372 *
da6052f7
NP
373 * A page may be used by anyone else who does a __get_free_page().
374 * In this case, page_count still tracks the references, and should only
375 * be used through the normal accessor functions. The top bits of page->flags
376 * and page->virtual store page management information, but all other fields
377 * are unused and could be used privately, carefully. The management of this
378 * page is the responsibility of the one who allocated it, and those who have
379 * subsequently been given references to it.
380 *
381 * The other pages (we may call them "pagecache pages") are completely
1da177e4
LT
382 * managed by the Linux memory manager: I/O, buffers, swapping etc.
383 * The following discussion applies only to them.
384 *
da6052f7
NP
385 * A pagecache page contains an opaque `private' member, which belongs to the
386 * page's address_space. Usually, this is the address of a circular list of
387 * the page's disk buffers. PG_private must be set to tell the VM to call
388 * into the filesystem to release these pages.
1da177e4 389 *
da6052f7
NP
390 * A page may belong to an inode's memory mapping. In this case, page->mapping
391 * is the pointer to the inode, and page->index is the file offset of the page,
392 * in units of PAGE_CACHE_SIZE.
1da177e4 393 *
da6052f7
NP
394 * If pagecache pages are not associated with an inode, they are said to be
395 * anonymous pages. These may become associated with the swapcache, and in that
396 * case PG_swapcache is set, and page->private is an offset into the swapcache.
1da177e4 397 *
da6052f7
NP
398 * In either case (swapcache or inode backed), the pagecache itself holds one
399 * reference to the page. Setting PG_private should also increment the
400 * refcount. The each user mapping also has a reference to the page.
1da177e4 401 *
da6052f7
NP
402 * The pagecache pages are stored in a per-mapping radix tree, which is
403 * rooted at mapping->page_tree, and indexed by offset.
404 * Where 2.4 and early 2.6 kernels kept dirty/clean pages in per-address_space
405 * lists, we instead now tag pages as dirty/writeback in the radix tree.
1da177e4 406 *
da6052f7 407 * All pagecache pages may be subject to I/O:
1da177e4
LT
408 * - inode pages may need to be read from disk,
409 * - inode pages which have been modified and are MAP_SHARED may need
da6052f7
NP
410 * to be written back to the inode on disk,
411 * - anonymous pages (including MAP_PRIVATE file mappings) which have been
412 * modified may need to be swapped out to swap space and (later) to be read
413 * back into memory.
1da177e4
LT
414 */
415
416/*
417 * The zone field is never updated after free_area_init_core()
418 * sets it, so none of the operations on it need to be atomic.
1da177e4 419 */
348f8b6c 420
d41dee36
AW
421
422/*
423 * page->flags layout:
424 *
425 * There are three possibilities for how page->flags get
426 * laid out. The first is for the normal case, without
427 * sparsemem. The second is for sparsemem when there is
428 * plenty of space for node and section. The last is when
429 * we have run out of space and have to fall back to an
430 * alternate (slower) way of determining the node.
431 *
308c05e3
CL
432 * No sparsemem or sparsemem vmemmap: | NODE | ZONE | ... | FLAGS |
433 * classic sparse with space for node:| SECTION | NODE | ZONE | ... | FLAGS |
434 * classic sparse no space for node: | SECTION | ZONE | ... | FLAGS |
d41dee36 435 */
308c05e3 436#if defined(CONFIG_SPARSEMEM) && !defined(CONFIG_SPARSEMEM_VMEMMAP)
d41dee36
AW
437#define SECTIONS_WIDTH SECTIONS_SHIFT
438#else
439#define SECTIONS_WIDTH 0
440#endif
441
442#define ZONES_WIDTH ZONES_SHIFT
443
9223b419 444#if SECTIONS_WIDTH+ZONES_WIDTH+NODES_SHIFT <= BITS_PER_LONG - NR_PAGEFLAGS
d41dee36
AW
445#define NODES_WIDTH NODES_SHIFT
446#else
308c05e3
CL
447#ifdef CONFIG_SPARSEMEM_VMEMMAP
448#error "Vmemmap: No space for nodes field in page flags"
449#endif
d41dee36
AW
450#define NODES_WIDTH 0
451#endif
452
453/* Page flags: | [SECTION] | [NODE] | ZONE | ... | FLAGS | */
07808b74 454#define SECTIONS_PGOFF ((sizeof(unsigned long)*8) - SECTIONS_WIDTH)
d41dee36
AW
455#define NODES_PGOFF (SECTIONS_PGOFF - NODES_WIDTH)
456#define ZONES_PGOFF (NODES_PGOFF - ZONES_WIDTH)
457
458/*
459 * We are going to use the flags for the page to node mapping if its in
460 * there. This includes the case where there is no node, so it is implicit.
461 */
89689ae7
CL
462#if !(NODES_WIDTH > 0 || NODES_SHIFT == 0)
463#define NODE_NOT_IN_PAGE_FLAGS
464#endif
d41dee36
AW
465
466#ifndef PFN_SECTION_SHIFT
467#define PFN_SECTION_SHIFT 0
468#endif
348f8b6c
DH
469
470/*
471 * Define the bit shifts to access each section. For non-existant
472 * sections we define the shift as 0; that plus a 0 mask ensures
473 * the compiler will optimise away reference to them.
474 */
d41dee36
AW
475#define SECTIONS_PGSHIFT (SECTIONS_PGOFF * (SECTIONS_WIDTH != 0))
476#define NODES_PGSHIFT (NODES_PGOFF * (NODES_WIDTH != 0))
477#define ZONES_PGSHIFT (ZONES_PGOFF * (ZONES_WIDTH != 0))
348f8b6c 478
89689ae7
CL
479/* NODE:ZONE or SECTION:ZONE is used to ID a zone for the buddy allcator */
480#ifdef NODE_NOT_IN_PAGEFLAGS
481#define ZONEID_SHIFT (SECTIONS_SHIFT + ZONES_SHIFT)
bd8029b6
AW
482#define ZONEID_PGOFF ((SECTIONS_PGOFF < ZONES_PGOFF)? \
483 SECTIONS_PGOFF : ZONES_PGOFF)
d41dee36 484#else
89689ae7 485#define ZONEID_SHIFT (NODES_SHIFT + ZONES_SHIFT)
bd8029b6
AW
486#define ZONEID_PGOFF ((NODES_PGOFF < ZONES_PGOFF)? \
487 NODES_PGOFF : ZONES_PGOFF)
89689ae7
CL
488#endif
489
bd8029b6 490#define ZONEID_PGSHIFT (ZONEID_PGOFF * (ZONEID_SHIFT != 0))
348f8b6c 491
9223b419
CL
492#if SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > BITS_PER_LONG - NR_PAGEFLAGS
493#error SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > BITS_PER_LONG - NR_PAGEFLAGS
348f8b6c
DH
494#endif
495
d41dee36
AW
496#define ZONES_MASK ((1UL << ZONES_WIDTH) - 1)
497#define NODES_MASK ((1UL << NODES_WIDTH) - 1)
498#define SECTIONS_MASK ((1UL << SECTIONS_WIDTH) - 1)
89689ae7 499#define ZONEID_MASK ((1UL << ZONEID_SHIFT) - 1)
348f8b6c 500
2f1b6248 501static inline enum zone_type page_zonenum(struct page *page)
1da177e4 502{
348f8b6c 503 return (page->flags >> ZONES_PGSHIFT) & ZONES_MASK;
1da177e4 504}
1da177e4 505
89689ae7
CL
506/*
507 * The identification function is only used by the buddy allocator for
508 * determining if two pages could be buddies. We are not really
509 * identifying a zone since we could be using a the section number
510 * id if we have not node id available in page flags.
511 * We guarantee only that it will return the same value for two
512 * combinable pages in a zone.
513 */
cb2b95e1
AW
514static inline int page_zone_id(struct page *page)
515{
89689ae7 516 return (page->flags >> ZONEID_PGSHIFT) & ZONEID_MASK;
348f8b6c
DH
517}
518
25ba77c1 519static inline int zone_to_nid(struct zone *zone)
89fa3024 520{
d5f541ed
CL
521#ifdef CONFIG_NUMA
522 return zone->node;
523#else
524 return 0;
525#endif
89fa3024
CL
526}
527
89689ae7 528#ifdef NODE_NOT_IN_PAGE_FLAGS
25ba77c1 529extern int page_to_nid(struct page *page);
89689ae7 530#else
25ba77c1 531static inline int page_to_nid(struct page *page)
d41dee36 532{
89689ae7 533 return (page->flags >> NODES_PGSHIFT) & NODES_MASK;
d41dee36 534}
89689ae7
CL
535#endif
536
537static inline struct zone *page_zone(struct page *page)
538{
539 return &NODE_DATA(page_to_nid(page))->node_zones[page_zonenum(page)];
540}
541
308c05e3 542#if defined(CONFIG_SPARSEMEM) && !defined(CONFIG_SPARSEMEM_VMEMMAP)
d41dee36
AW
543static inline unsigned long page_to_section(struct page *page)
544{
545 return (page->flags >> SECTIONS_PGSHIFT) & SECTIONS_MASK;
546}
308c05e3 547#endif
d41dee36 548
2f1b6248 549static inline void set_page_zone(struct page *page, enum zone_type zone)
348f8b6c
DH
550{
551 page->flags &= ~(ZONES_MASK << ZONES_PGSHIFT);
552 page->flags |= (zone & ZONES_MASK) << ZONES_PGSHIFT;
553}
2f1b6248 554
348f8b6c
DH
555static inline void set_page_node(struct page *page, unsigned long node)
556{
557 page->flags &= ~(NODES_MASK << NODES_PGSHIFT);
558 page->flags |= (node & NODES_MASK) << NODES_PGSHIFT;
1da177e4 559}
89689ae7 560
d41dee36
AW
561static inline void set_page_section(struct page *page, unsigned long section)
562{
563 page->flags &= ~(SECTIONS_MASK << SECTIONS_PGSHIFT);
564 page->flags |= (section & SECTIONS_MASK) << SECTIONS_PGSHIFT;
565}
1da177e4 566
2f1b6248 567static inline void set_page_links(struct page *page, enum zone_type zone,
d41dee36 568 unsigned long node, unsigned long pfn)
1da177e4 569{
348f8b6c
DH
570 set_page_zone(page, zone);
571 set_page_node(page, node);
d41dee36 572 set_page_section(page, pfn_to_section_nr(pfn));
1da177e4
LT
573}
574
7cd94146
EP
575/*
576 * If a hint addr is less than mmap_min_addr change hint to be as
577 * low as possible but still greater than mmap_min_addr
578 */
579static inline unsigned long round_hint_to_min(unsigned long hint)
580{
581#ifdef CONFIG_SECURITY
582 hint &= PAGE_MASK;
583 if (((void *)hint != NULL) &&
584 (hint < mmap_min_addr))
585 return PAGE_ALIGN(mmap_min_addr);
586#endif
587 return hint;
588}
589
f6ac2354
CL
590/*
591 * Some inline functions in vmstat.h depend on page_zone()
592 */
593#include <linux/vmstat.h>
594
652050ae 595static __always_inline void *lowmem_page_address(struct page *page)
1da177e4
LT
596{
597 return __va(page_to_pfn(page) << PAGE_SHIFT);
598}
599
600#if defined(CONFIG_HIGHMEM) && !defined(WANT_PAGE_VIRTUAL)
601#define HASHED_PAGE_VIRTUAL
602#endif
603
604#if defined(WANT_PAGE_VIRTUAL)
605#define page_address(page) ((page)->virtual)
606#define set_page_address(page, address) \
607 do { \
608 (page)->virtual = (address); \
609 } while(0)
610#define page_address_init() do { } while(0)
611#endif
612
613#if defined(HASHED_PAGE_VIRTUAL)
614void *page_address(struct page *page);
615void set_page_address(struct page *page, void *virtual);
616void page_address_init(void);
617#endif
618
619#if !defined(HASHED_PAGE_VIRTUAL) && !defined(WANT_PAGE_VIRTUAL)
620#define page_address(page) lowmem_page_address(page)
621#define set_page_address(page, address) do { } while(0)
622#define page_address_init() do { } while(0)
623#endif
624
625/*
626 * On an anonymous page mapped into a user virtual memory area,
627 * page->mapping points to its anon_vma, not to a struct address_space;
628 * with the PAGE_MAPPING_ANON bit set to distinguish it.
629 *
630 * Please note that, confusingly, "page_mapping" refers to the inode
631 * address_space which maps the page from disk; whereas "page_mapped"
632 * refers to user virtual address space into which the page is mapped.
633 */
634#define PAGE_MAPPING_ANON 1
635
636extern struct address_space swapper_space;
637static inline struct address_space *page_mapping(struct page *page)
638{
639 struct address_space *mapping = page->mapping;
640
b5fab14e 641 VM_BUG_ON(PageSlab(page));
726b8012 642#ifdef CONFIG_SWAP
1da177e4
LT
643 if (unlikely(PageSwapCache(page)))
644 mapping = &swapper_space;
726b8012
AM
645 else
646#endif
647 if (unlikely((unsigned long)mapping & PAGE_MAPPING_ANON))
1da177e4
LT
648 mapping = NULL;
649 return mapping;
650}
651
652static inline int PageAnon(struct page *page)
653{
654 return ((unsigned long)page->mapping & PAGE_MAPPING_ANON) != 0;
655}
656
657/*
658 * Return the pagecache index of the passed page. Regular pagecache pages
659 * use ->index whereas swapcache pages use ->private
660 */
661static inline pgoff_t page_index(struct page *page)
662{
663 if (unlikely(PageSwapCache(page)))
4c21e2f2 664 return page_private(page);
1da177e4
LT
665 return page->index;
666}
667
668/*
669 * The atomic page->_mapcount, like _count, starts from -1:
670 * so that transitions both from it and to it can be tracked,
671 * using atomic_inc_and_test and atomic_add_negative(-1).
672 */
673static inline void reset_page_mapcount(struct page *page)
674{
675 atomic_set(&(page)->_mapcount, -1);
676}
677
678static inline int page_mapcount(struct page *page)
679{
680 return atomic_read(&(page)->_mapcount) + 1;
681}
682
683/*
684 * Return true if this page is mapped into pagetables.
685 */
686static inline int page_mapped(struct page *page)
687{
688 return atomic_read(&(page)->_mapcount) >= 0;
689}
690
1da177e4
LT
691/*
692 * Different kinds of faults, as returned by handle_mm_fault().
693 * Used to decide whether a process gets delivered SIGBUS or
694 * just gets major/minor fault counters bumped up.
695 */
d0217ac0 696
83c54070 697#define VM_FAULT_MINOR 0 /* For backwards compat. Remove me quickly. */
d0217ac0 698
83c54070
NP
699#define VM_FAULT_OOM 0x0001
700#define VM_FAULT_SIGBUS 0x0002
701#define VM_FAULT_MAJOR 0x0004
702#define VM_FAULT_WRITE 0x0008 /* Special case for get_user_pages */
f33ea7f4 703
83c54070
NP
704#define VM_FAULT_NOPAGE 0x0100 /* ->fault installed the pte, not return page */
705#define VM_FAULT_LOCKED 0x0200 /* ->fault locked the returned page */
1da177e4 706
83c54070 707#define VM_FAULT_ERROR (VM_FAULT_OOM | VM_FAULT_SIGBUS)
d0217ac0 708
1c0fe6e3
NP
709/*
710 * Can be called by the pagefault handler when it gets a VM_FAULT_OOM.
711 */
712extern void pagefault_out_of_memory(void);
713
1da177e4
LT
714#define offset_in_page(p) ((unsigned long)(p) & ~PAGE_MASK)
715
716extern void show_free_areas(void);
717
718#ifdef CONFIG_SHMEM
89e004ea 719extern int shmem_lock(struct file *file, int lock, struct user_struct *user);
1da177e4 720#else
03b00ebc 721static inline int shmem_lock(struct file *file, int lock,
89e004ea 722 struct user_struct *user)
03b00ebc
RK
723{
724 return 0;
725}
1da177e4
LT
726#endif
727struct file *shmem_file_setup(char *name, loff_t size, unsigned long flags);
728
729int shmem_zero_setup(struct vm_area_struct *);
730
b0e15190
DH
731#ifndef CONFIG_MMU
732extern unsigned long shmem_get_unmapped_area(struct file *file,
733 unsigned long addr,
734 unsigned long len,
735 unsigned long pgoff,
736 unsigned long flags);
737#endif
738
e8edc6e0 739extern int can_do_mlock(void);
1da177e4
LT
740extern int user_shm_lock(size_t, struct user_struct *);
741extern void user_shm_unlock(size_t, struct user_struct *);
742
743/*
744 * Parameter block passed down to zap_pte_range in exceptional cases.
745 */
746struct zap_details {
747 struct vm_area_struct *nonlinear_vma; /* Check page->index if set */
748 struct address_space *check_mapping; /* Check page->mapping if set */
749 pgoff_t first_index; /* Lowest page->index to unmap */
750 pgoff_t last_index; /* Highest page->index to unmap */
751 spinlock_t *i_mmap_lock; /* For unmap_mapping_range: */
1da177e4
LT
752 unsigned long truncate_count; /* Compare vm_truncate_count */
753};
754
7e675137
NP
755struct page *vm_normal_page(struct vm_area_struct *vma, unsigned long addr,
756 pte_t pte);
757
c627f9cc
JS
758int zap_vma_ptes(struct vm_area_struct *vma, unsigned long address,
759 unsigned long size);
ee39b37b 760unsigned long zap_page_range(struct vm_area_struct *vma, unsigned long address,
1da177e4 761 unsigned long size, struct zap_details *);
508034a3 762unsigned long unmap_vmas(struct mmu_gather **tlb,
1da177e4
LT
763 struct vm_area_struct *start_vma, unsigned long start_addr,
764 unsigned long end_addr, unsigned long *nr_accounted,
765 struct zap_details *);
e6473092
MM
766
767/**
768 * mm_walk - callbacks for walk_page_range
769 * @pgd_entry: if set, called for each non-empty PGD (top-level) entry
770 * @pud_entry: if set, called for each non-empty PUD (2nd-level) entry
771 * @pmd_entry: if set, called for each non-empty PMD (3rd-level) entry
772 * @pte_entry: if set, called for each non-empty PTE (4th-level) entry
773 * @pte_hole: if set, called for each hole at all levels
774 *
775 * (see walk_page_range for more details)
776 */
777struct mm_walk {
2165009b
DH
778 int (*pgd_entry)(pgd_t *, unsigned long, unsigned long, struct mm_walk *);
779 int (*pud_entry)(pud_t *, unsigned long, unsigned long, struct mm_walk *);
780 int (*pmd_entry)(pmd_t *, unsigned long, unsigned long, struct mm_walk *);
781 int (*pte_entry)(pte_t *, unsigned long, unsigned long, struct mm_walk *);
782 int (*pte_hole)(unsigned long, unsigned long, struct mm_walk *);
783 struct mm_struct *mm;
784 void *private;
e6473092
MM
785};
786
2165009b
DH
787int walk_page_range(unsigned long addr, unsigned long end,
788 struct mm_walk *walk);
42b77728 789void free_pgd_range(struct mmu_gather *tlb, unsigned long addr,
3bf5ee95 790 unsigned long end, unsigned long floor, unsigned long ceiling);
1da177e4
LT
791int copy_page_range(struct mm_struct *dst, struct mm_struct *src,
792 struct vm_area_struct *vma);
1da177e4
LT
793void unmap_mapping_range(struct address_space *mapping,
794 loff_t const holebegin, loff_t const holelen, int even_cows);
d87fe660 795int follow_phys(struct vm_area_struct *vma, unsigned long address,
796 unsigned int flags, unsigned long *prot, resource_size_t *phys);
28b2ee20
RR
797int generic_access_phys(struct vm_area_struct *vma, unsigned long addr,
798 void *buf, int len, int write);
1da177e4
LT
799
800static inline void unmap_shared_mapping_range(struct address_space *mapping,
801 loff_t const holebegin, loff_t const holelen)
802{
803 unmap_mapping_range(mapping, holebegin, holelen, 0);
804}
805
806extern int vmtruncate(struct inode * inode, loff_t offset);
f6b3ec23 807extern int vmtruncate_range(struct inode * inode, loff_t offset, loff_t end);
f33ea7f4 808
7ee1dd3f 809#ifdef CONFIG_MMU
83c54070 810extern int handle_mm_fault(struct mm_struct *mm, struct vm_area_struct *vma,
7ee1dd3f 811 unsigned long address, int write_access);
7ee1dd3f
DH
812#else
813static inline int handle_mm_fault(struct mm_struct *mm,
814 struct vm_area_struct *vma, unsigned long address,
815 int write_access)
816{
817 /* should never happen if there's no MMU */
818 BUG();
819 return VM_FAULT_SIGBUS;
820}
821#endif
f33ea7f4 822
1da177e4
LT
823extern int make_pages_present(unsigned long addr, unsigned long end);
824extern int access_process_vm(struct task_struct *tsk, unsigned long addr, void *buf, int len, int write);
1da177e4
LT
825
826int get_user_pages(struct task_struct *tsk, struct mm_struct *mm, unsigned long start,
827 int len, int write, int force, struct page **pages, struct vm_area_struct **vmas);
828
cf9a2ae8
DH
829extern int try_to_release_page(struct page * page, gfp_t gfp_mask);
830extern void do_invalidatepage(struct page *page, unsigned long offset);
831
1da177e4 832int __set_page_dirty_nobuffers(struct page *page);
76719325 833int __set_page_dirty_no_writeback(struct page *page);
1da177e4
LT
834int redirty_page_for_writepage(struct writeback_control *wbc,
835 struct page *page);
b3c97528 836int set_page_dirty(struct page *page);
1da177e4
LT
837int set_page_dirty_lock(struct page *page);
838int clear_page_dirty_for_io(struct page *page);
839
b6a2fea3
OW
840extern unsigned long move_page_tables(struct vm_area_struct *vma,
841 unsigned long old_addr, struct vm_area_struct *new_vma,
842 unsigned long new_addr, unsigned long len);
1da177e4
LT
843extern unsigned long do_mremap(unsigned long addr,
844 unsigned long old_len, unsigned long new_len,
845 unsigned long flags, unsigned long new_addr);
b6a2fea3
OW
846extern int mprotect_fixup(struct vm_area_struct *vma,
847 struct vm_area_struct **pprev, unsigned long start,
848 unsigned long end, unsigned long newflags);
1da177e4 849
21cc199b
NP
850/*
851 * get_user_pages_fast provides equivalent functionality to get_user_pages,
852 * operating on current and current->mm (force=0 and doesn't return any vmas).
853 *
854 * get_user_pages_fast may take mmap_sem and page tables, so no assumptions
855 * can be made about locking. get_user_pages_fast is to be implemented in a
856 * way that is advantageous (vs get_user_pages()) when the user memory area is
857 * already faulted in and present in ptes. However if the pages have to be
858 * faulted in, it may turn out to be slightly slower).
859 */
860int get_user_pages_fast(unsigned long start, int nr_pages, int write,
861 struct page **pages);
862
1da177e4 863/*
8e1f936b 864 * A callback you can register to apply pressure to ageable caches.
1da177e4 865 *
8e1f936b
RR
866 * 'shrink' is passed a count 'nr_to_scan' and a 'gfpmask'. It should
867 * look through the least-recently-used 'nr_to_scan' entries and
868 * attempt to free them up. It should return the number of objects
869 * which remain in the cache. If it returns -1, it means it cannot do
870 * any scanning at this time (eg. there is a risk of deadlock).
1da177e4 871 *
8e1f936b
RR
872 * The 'gfpmask' refers to the allocation we are currently trying to
873 * fulfil.
874 *
875 * Note that 'shrink' will be passed nr_to_scan == 0 when the VM is
876 * querying the cache size, so a fastpath for that case is appropriate.
1da177e4 877 */
8e1f936b
RR
878struct shrinker {
879 int (*shrink)(int nr_to_scan, gfp_t gfp_mask);
880 int seeks; /* seeks to recreate an obj */
1da177e4 881
8e1f936b
RR
882 /* These are for internal use */
883 struct list_head list;
884 long nr; /* objs pending delete */
885};
886#define DEFAULT_SEEKS 2 /* A good number if you don't know better. */
887extern void register_shrinker(struct shrinker *);
888extern void unregister_shrinker(struct shrinker *);
1da177e4 889
4e950f6f 890int vma_wants_writenotify(struct vm_area_struct *vma);
d08b3851 891
b3c97528 892extern pte_t *get_locked_pte(struct mm_struct *mm, unsigned long addr, spinlock_t **ptl);
c9cfcddf 893
5f22df00
NP
894#ifdef __PAGETABLE_PUD_FOLDED
895static inline int __pud_alloc(struct mm_struct *mm, pgd_t *pgd,
896 unsigned long address)
897{
898 return 0;
899}
900#else
1bb3630e 901int __pud_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address);
5f22df00
NP
902#endif
903
904#ifdef __PAGETABLE_PMD_FOLDED
905static inline int __pmd_alloc(struct mm_struct *mm, pud_t *pud,
906 unsigned long address)
907{
908 return 0;
909}
910#else
1bb3630e 911int __pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address);
5f22df00
NP
912#endif
913
1bb3630e
HD
914int __pte_alloc(struct mm_struct *mm, pmd_t *pmd, unsigned long address);
915int __pte_alloc_kernel(pmd_t *pmd, unsigned long address);
916
1da177e4
LT
917/*
918 * The following ifdef needed to get the 4level-fixup.h header to work.
919 * Remove it when 4level-fixup.h has been removed.
920 */
1bb3630e 921#if defined(CONFIG_MMU) && !defined(__ARCH_HAS_4LEVEL_HACK)
1da177e4
LT
922static inline pud_t *pud_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address)
923{
1bb3630e
HD
924 return (unlikely(pgd_none(*pgd)) && __pud_alloc(mm, pgd, address))?
925 NULL: pud_offset(pgd, address);
1da177e4
LT
926}
927
928static inline pmd_t *pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address)
929{
1bb3630e
HD
930 return (unlikely(pud_none(*pud)) && __pmd_alloc(mm, pud, address))?
931 NULL: pmd_offset(pud, address);
1da177e4 932}
1bb3630e
HD
933#endif /* CONFIG_MMU && !__ARCH_HAS_4LEVEL_HACK */
934
f7d0b926 935#if USE_SPLIT_PTLOCKS
4c21e2f2
HD
936/*
937 * We tuck a spinlock to guard each pagetable page into its struct page,
938 * at page->private, with BUILD_BUG_ON to make sure that this will not
939 * overflow into the next struct page (as it might with DEBUG_SPINLOCK).
940 * When freeing, reset page->mapping so free_pages_check won't complain.
941 */
349aef0b 942#define __pte_lockptr(page) &((page)->ptl)
4c21e2f2
HD
943#define pte_lock_init(_page) do { \
944 spin_lock_init(__pte_lockptr(_page)); \
945} while (0)
946#define pte_lock_deinit(page) ((page)->mapping = NULL)
947#define pte_lockptr(mm, pmd) ({(void)(mm); __pte_lockptr(pmd_page(*(pmd)));})
f7d0b926 948#else /* !USE_SPLIT_PTLOCKS */
4c21e2f2
HD
949/*
950 * We use mm->page_table_lock to guard all pagetable pages of the mm.
951 */
952#define pte_lock_init(page) do {} while (0)
953#define pte_lock_deinit(page) do {} while (0)
954#define pte_lockptr(mm, pmd) ({(void)(pmd); &(mm)->page_table_lock;})
f7d0b926 955#endif /* USE_SPLIT_PTLOCKS */
4c21e2f2 956
2f569afd
MS
957static inline void pgtable_page_ctor(struct page *page)
958{
959 pte_lock_init(page);
960 inc_zone_page_state(page, NR_PAGETABLE);
961}
962
963static inline void pgtable_page_dtor(struct page *page)
964{
965 pte_lock_deinit(page);
966 dec_zone_page_state(page, NR_PAGETABLE);
967}
968
c74df32c
HD
969#define pte_offset_map_lock(mm, pmd, address, ptlp) \
970({ \
4c21e2f2 971 spinlock_t *__ptl = pte_lockptr(mm, pmd); \
c74df32c
HD
972 pte_t *__pte = pte_offset_map(pmd, address); \
973 *(ptlp) = __ptl; \
974 spin_lock(__ptl); \
975 __pte; \
976})
977
978#define pte_unmap_unlock(pte, ptl) do { \
979 spin_unlock(ptl); \
980 pte_unmap(pte); \
981} while (0)
982
1bb3630e
HD
983#define pte_alloc_map(mm, pmd, address) \
984 ((unlikely(!pmd_present(*(pmd))) && __pte_alloc(mm, pmd, address))? \
985 NULL: pte_offset_map(pmd, address))
986
c74df32c
HD
987#define pte_alloc_map_lock(mm, pmd, address, ptlp) \
988 ((unlikely(!pmd_present(*(pmd))) && __pte_alloc(mm, pmd, address))? \
989 NULL: pte_offset_map_lock(mm, pmd, address, ptlp))
990
1bb3630e
HD
991#define pte_alloc_kernel(pmd, address) \
992 ((unlikely(!pmd_present(*(pmd))) && __pte_alloc_kernel(pmd, address))? \
993 NULL: pte_offset_kernel(pmd, address))
1da177e4
LT
994
995extern void free_area_init(unsigned long * zones_size);
9109fb7b
JW
996extern void free_area_init_node(int nid, unsigned long * zones_size,
997 unsigned long zone_start_pfn, unsigned long *zholes_size);
c713216d
MG
998#ifdef CONFIG_ARCH_POPULATES_NODE_MAP
999/*
1000 * With CONFIG_ARCH_POPULATES_NODE_MAP set, an architecture may initialise its
1001 * zones, allocate the backing mem_map and account for memory holes in a more
1002 * architecture independent manner. This is a substitute for creating the
1003 * zone_sizes[] and zholes_size[] arrays and passing them to
1004 * free_area_init_node()
1005 *
1006 * An architecture is expected to register range of page frames backed by
1007 * physical memory with add_active_range() before calling
1008 * free_area_init_nodes() passing in the PFN each zone ends at. At a basic
1009 * usage, an architecture is expected to do something like
1010 *
1011 * unsigned long max_zone_pfns[MAX_NR_ZONES] = {max_dma, max_normal_pfn,
1012 * max_highmem_pfn};
1013 * for_each_valid_physical_page_range()
1014 * add_active_range(node_id, start_pfn, end_pfn)
1015 * free_area_init_nodes(max_zone_pfns);
1016 *
1017 * If the architecture guarantees that there are no holes in the ranges
1018 * registered with add_active_range(), free_bootmem_active_regions()
1019 * will call free_bootmem_node() for each registered physical page range.
1020 * Similarly sparse_memory_present_with_active_regions() calls
1021 * memory_present() for each range when SPARSEMEM is enabled.
1022 *
1023 * See mm/page_alloc.c for more information on each function exposed by
1024 * CONFIG_ARCH_POPULATES_NODE_MAP
1025 */
1026extern void free_area_init_nodes(unsigned long *max_zone_pfn);
1027extern void add_active_range(unsigned int nid, unsigned long start_pfn,
1028 unsigned long end_pfn);
cc1050ba
YL
1029extern void remove_active_range(unsigned int nid, unsigned long start_pfn,
1030 unsigned long end_pfn);
fb01439c
MG
1031extern void push_node_boundaries(unsigned int nid, unsigned long start_pfn,
1032 unsigned long end_pfn);
c713216d
MG
1033extern void remove_all_active_ranges(void);
1034extern unsigned long absent_pages_in_range(unsigned long start_pfn,
1035 unsigned long end_pfn);
1036extern void get_pfn_range_for_nid(unsigned int nid,
1037 unsigned long *start_pfn, unsigned long *end_pfn);
1038extern unsigned long find_min_pfn_with_active_regions(void);
c713216d
MG
1039extern void free_bootmem_with_active_regions(int nid,
1040 unsigned long max_low_pfn);
d52d53b8 1041typedef int (*work_fn_t)(unsigned long, unsigned long, void *);
b5bc6c0e 1042extern void work_with_active_regions(int nid, work_fn_t work_fn, void *data);
c713216d
MG
1043extern void sparse_memory_present_with_active_regions(int nid);
1044#ifndef CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID
1045extern int early_pfn_to_nid(unsigned long pfn);
1046#endif /* CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID */
1047#endif /* CONFIG_ARCH_POPULATES_NODE_MAP */
0e0b864e 1048extern void set_dma_reserve(unsigned long new_dma_reserve);
a2f3aa02
DH
1049extern void memmap_init_zone(unsigned long, int, unsigned long,
1050 unsigned long, enum memmap_context);
3947be19 1051extern void setup_per_zone_pages_min(void);
1da177e4 1052extern void mem_init(void);
8feae131 1053extern void __init mmap_init(void);
1da177e4
LT
1054extern void show_mem(void);
1055extern void si_meminfo(struct sysinfo * val);
1056extern void si_meminfo_node(struct sysinfo *val, int nid);
3461b0af 1057extern int after_bootmem;
1da177e4 1058
e7c8d5c9
CL
1059#ifdef CONFIG_NUMA
1060extern void setup_per_cpu_pageset(void);
1061#else
1062static inline void setup_per_cpu_pageset(void) {}
1063#endif
1064
8feae131
DH
1065/* nommu.c */
1066extern atomic_t mmap_pages_allocated;
1067
1da177e4
LT
1068/* prio_tree.c */
1069void vma_prio_tree_add(struct vm_area_struct *, struct vm_area_struct *old);
1070void vma_prio_tree_insert(struct vm_area_struct *, struct prio_tree_root *);
1071void vma_prio_tree_remove(struct vm_area_struct *, struct prio_tree_root *);
1072struct vm_area_struct *vma_prio_tree_next(struct vm_area_struct *vma,
1073 struct prio_tree_iter *iter);
1074
1075#define vma_prio_tree_foreach(vma, iter, root, begin, end) \
1076 for (prio_tree_iter_init(iter, root, begin, end), vma = NULL; \
1077 (vma = vma_prio_tree_next(vma, iter)); )
1078
1079static inline void vma_nonlinear_insert(struct vm_area_struct *vma,
1080 struct list_head *list)
1081{
1082 vma->shared.vm_set.parent = NULL;
1083 list_add_tail(&vma->shared.vm_set.list, list);
1084}
1085
1086/* mmap.c */
34b4e4aa 1087extern int __vm_enough_memory(struct mm_struct *mm, long pages, int cap_sys_admin);
1da177e4
LT
1088extern void vma_adjust(struct vm_area_struct *vma, unsigned long start,
1089 unsigned long end, pgoff_t pgoff, struct vm_area_struct *insert);
1090extern struct vm_area_struct *vma_merge(struct mm_struct *,
1091 struct vm_area_struct *prev, unsigned long addr, unsigned long end,
1092 unsigned long vm_flags, struct anon_vma *, struct file *, pgoff_t,
1093 struct mempolicy *);
1094extern struct anon_vma *find_mergeable_anon_vma(struct vm_area_struct *);
1095extern int split_vma(struct mm_struct *,
1096 struct vm_area_struct *, unsigned long addr, int new_below);
1097extern int insert_vm_struct(struct mm_struct *, struct vm_area_struct *);
1098extern void __vma_link_rb(struct mm_struct *, struct vm_area_struct *,
1099 struct rb_node **, struct rb_node *);
a8fb5618 1100extern void unlink_file_vma(struct vm_area_struct *);
1da177e4
LT
1101extern struct vm_area_struct *copy_vma(struct vm_area_struct **,
1102 unsigned long addr, unsigned long len, pgoff_t pgoff);
1103extern void exit_mmap(struct mm_struct *);
925d1c40 1104
7906d00c
AA
1105extern int mm_take_all_locks(struct mm_struct *mm);
1106extern void mm_drop_all_locks(struct mm_struct *mm);
1107
925d1c40
MH
1108#ifdef CONFIG_PROC_FS
1109/* From fs/proc/base.c. callers must _not_ hold the mm's exe_file_lock */
1110extern void added_exe_file_vma(struct mm_struct *mm);
1111extern void removed_exe_file_vma(struct mm_struct *mm);
1112#else
1113static inline void added_exe_file_vma(struct mm_struct *mm)
1114{}
1115
1116static inline void removed_exe_file_vma(struct mm_struct *mm)
1117{}
1118#endif /* CONFIG_PROC_FS */
1119
119f657c 1120extern int may_expand_vm(struct mm_struct *mm, unsigned long npages);
fa5dc22f
RM
1121extern int install_special_mapping(struct mm_struct *mm,
1122 unsigned long addr, unsigned long len,
1123 unsigned long flags, struct page **pages);
1da177e4
LT
1124
1125extern unsigned long get_unmapped_area(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
1126
1127extern unsigned long do_mmap_pgoff(struct file *file, unsigned long addr,
1128 unsigned long len, unsigned long prot,
1129 unsigned long flag, unsigned long pgoff);
0165ab44
MS
1130extern unsigned long mmap_region(struct file *file, unsigned long addr,
1131 unsigned long len, unsigned long flags,
5a6fe125 1132 unsigned int vm_flags, unsigned long pgoff);
1da177e4
LT
1133
1134static inline unsigned long do_mmap(struct file *file, unsigned long addr,
1135 unsigned long len, unsigned long prot,
1136 unsigned long flag, unsigned long offset)
1137{
1138 unsigned long ret = -EINVAL;
1139 if ((offset + PAGE_ALIGN(len)) < offset)
1140 goto out;
1141 if (!(offset & ~PAGE_MASK))
1142 ret = do_mmap_pgoff(file, addr, len, prot, flag, offset >> PAGE_SHIFT);
1143out:
1144 return ret;
1145}
1146
1147extern int do_munmap(struct mm_struct *, unsigned long, size_t);
1148
1149extern unsigned long do_brk(unsigned long, unsigned long);
1150
1151/* filemap.c */
1152extern unsigned long page_unuse(struct page *);
1153extern void truncate_inode_pages(struct address_space *, loff_t);
d7339071
HR
1154extern void truncate_inode_pages_range(struct address_space *,
1155 loff_t lstart, loff_t lend);
1da177e4
LT
1156
1157/* generic vm_area_ops exported for stackable file systems */
d0217ac0 1158extern int filemap_fault(struct vm_area_struct *, struct vm_fault *);
1da177e4
LT
1159
1160/* mm/page-writeback.c */
1161int write_one_page(struct page *page, int wait);
1cf6e7d8 1162void task_dirty_inc(struct task_struct *tsk);
1da177e4
LT
1163
1164/* readahead.c */
1165#define VM_MAX_READAHEAD 128 /* kbytes */
1166#define VM_MIN_READAHEAD 16 /* kbytes (includes current page) */
1da177e4
LT
1167
1168int do_page_cache_readahead(struct address_space *mapping, struct file *filp,
7361f4d8 1169 pgoff_t offset, unsigned long nr_to_read);
1da177e4 1170int force_page_cache_readahead(struct address_space *mapping, struct file *filp,
7361f4d8 1171 pgoff_t offset, unsigned long nr_to_read);
cf914a7d
RR
1172
1173void page_cache_sync_readahead(struct address_space *mapping,
1174 struct file_ra_state *ra,
1175 struct file *filp,
1176 pgoff_t offset,
1177 unsigned long size);
1178
1179void page_cache_async_readahead(struct address_space *mapping,
1180 struct file_ra_state *ra,
1181 struct file *filp,
1182 struct page *pg,
1183 pgoff_t offset,
1184 unsigned long size);
1185
1da177e4
LT
1186unsigned long max_sane_readahead(unsigned long nr);
1187
1188/* Do stack extension */
46dea3d0 1189extern int expand_stack(struct vm_area_struct *vma, unsigned long address);
9ab88515 1190#ifdef CONFIG_IA64
46dea3d0 1191extern int expand_upwards(struct vm_area_struct *vma, unsigned long address);
9ab88515 1192#endif
b6a2fea3
OW
1193extern int expand_stack_downwards(struct vm_area_struct *vma,
1194 unsigned long address);
1da177e4
LT
1195
1196/* Look up the first VMA which satisfies addr < vm_end, NULL if none. */
1197extern struct vm_area_struct * find_vma(struct mm_struct * mm, unsigned long addr);
1198extern struct vm_area_struct * find_vma_prev(struct mm_struct * mm, unsigned long addr,
1199 struct vm_area_struct **pprev);
1200
1201/* Look up the first VMA which intersects the interval start_addr..end_addr-1,
1202 NULL if none. Assume start_addr < end_addr. */
1203static inline struct vm_area_struct * find_vma_intersection(struct mm_struct * mm, unsigned long start_addr, unsigned long end_addr)
1204{
1205 struct vm_area_struct * vma = find_vma(mm,start_addr);
1206
1207 if (vma && end_addr <= vma->vm_start)
1208 vma = NULL;
1209 return vma;
1210}
1211
1212static inline unsigned long vma_pages(struct vm_area_struct *vma)
1213{
1214 return (vma->vm_end - vma->vm_start) >> PAGE_SHIFT;
1215}
1216
804af2cf 1217pgprot_t vm_get_page_prot(unsigned long vm_flags);
deceb6cd 1218struct vm_area_struct *find_extend_vma(struct mm_struct *, unsigned long addr);
deceb6cd
HD
1219int remap_pfn_range(struct vm_area_struct *, unsigned long addr,
1220 unsigned long pfn, unsigned long size, pgprot_t);
a145dd41 1221int vm_insert_page(struct vm_area_struct *, unsigned long addr, struct page *);
e0dc0d8f
NP
1222int vm_insert_pfn(struct vm_area_struct *vma, unsigned long addr,
1223 unsigned long pfn);
423bad60
NP
1224int vm_insert_mixed(struct vm_area_struct *vma, unsigned long addr,
1225 unsigned long pfn);
deceb6cd 1226
6aab341e 1227struct page *follow_page(struct vm_area_struct *, unsigned long address,
deceb6cd
HD
1228 unsigned int foll_flags);
1229#define FOLL_WRITE 0x01 /* check pte is writable */
1230#define FOLL_TOUCH 0x02 /* mark page accessed */
1231#define FOLL_GET 0x04 /* do get_page on page */
1232#define FOLL_ANON 0x08 /* give ZERO_PAGE if no pgtable */
1da177e4 1233
2f569afd 1234typedef int (*pte_fn_t)(pte_t *pte, pgtable_t token, unsigned long addr,
aee16b3c
JF
1235 void *data);
1236extern int apply_to_page_range(struct mm_struct *mm, unsigned long address,
1237 unsigned long size, pte_fn_t fn, void *data);
1238
1da177e4 1239#ifdef CONFIG_PROC_FS
ab50b8ed 1240void vm_stat_account(struct mm_struct *, unsigned long, struct file *, long);
1da177e4 1241#else
ab50b8ed 1242static inline void vm_stat_account(struct mm_struct *mm,
1da177e4
LT
1243 unsigned long flags, struct file *file, long pages)
1244{
1245}
1246#endif /* CONFIG_PROC_FS */
1247
12d6f21e
IM
1248#ifdef CONFIG_DEBUG_PAGEALLOC
1249extern int debug_pagealloc_enabled;
1250
1251extern void kernel_map_pages(struct page *page, int numpages, int enable);
1252
1253static inline void enable_debug_pagealloc(void)
1254{
1255 debug_pagealloc_enabled = 1;
1256}
8a235efa
RW
1257#ifdef CONFIG_HIBERNATION
1258extern bool kernel_page_present(struct page *page);
1259#endif /* CONFIG_HIBERNATION */
12d6f21e 1260#else
1da177e4 1261static inline void
9858db50 1262kernel_map_pages(struct page *page, int numpages, int enable) {}
12d6f21e
IM
1263static inline void enable_debug_pagealloc(void)
1264{
1265}
8a235efa
RW
1266#ifdef CONFIG_HIBERNATION
1267static inline bool kernel_page_present(struct page *page) { return true; }
1268#endif /* CONFIG_HIBERNATION */
1da177e4
LT
1269#endif
1270
1271extern struct vm_area_struct *get_gate_vma(struct task_struct *tsk);
1272#ifdef __HAVE_ARCH_GATE_AREA
1273int in_gate_area_no_task(unsigned long addr);
1274int in_gate_area(struct task_struct *task, unsigned long addr);
1275#else
1276int in_gate_area_no_task(unsigned long addr);
1277#define in_gate_area(task, addr) ({(void)task; in_gate_area_no_task(addr);})
1278#endif /* __HAVE_ARCH_GATE_AREA */
1279
9d0243bc
AM
1280int drop_caches_sysctl_handler(struct ctl_table *, int, struct file *,
1281 void __user *, size_t *, loff_t *);
69e05944 1282unsigned long shrink_slab(unsigned long scanned, gfp_t gfp_mask,
9d0243bc 1283 unsigned long lru_pages);
9d0243bc 1284
7a9166e3
LY
1285#ifndef CONFIG_MMU
1286#define randomize_va_space 0
1287#else
a62eaf15 1288extern int randomize_va_space;
7a9166e3 1289#endif
a62eaf15 1290
045e72ac 1291const char * arch_vma_name(struct vm_area_struct *vma);
03252919 1292void print_vma_addr(char *prefix, unsigned long rip);
e6e5494c 1293
98f3cfc1 1294struct page *sparse_mem_map_populate(unsigned long pnum, int nid);
29c71111
AW
1295pgd_t *vmemmap_pgd_populate(unsigned long addr, int node);
1296pud_t *vmemmap_pud_populate(pgd_t *pgd, unsigned long addr, int node);
1297pmd_t *vmemmap_pmd_populate(pud_t *pud, unsigned long addr, int node);
1298pte_t *vmemmap_pte_populate(pmd_t *pmd, unsigned long addr, int node);
8f6aac41
CL
1299void *vmemmap_alloc_block(unsigned long size, int node);
1300void vmemmap_verify(pte_t *, int, unsigned long, unsigned long);
29c71111
AW
1301int vmemmap_populate_basepages(struct page *start_page,
1302 unsigned long pages, int node);
1303int vmemmap_populate(struct page *start_page, unsigned long pages, int node);
c2b91e2e 1304void vmemmap_populate_print_last(void);
8f6aac41 1305
c5dee617
MM
1306extern void *alloc_locked_buffer(size_t size);
1307extern void free_locked_buffer(void *buffer, size_t size);
9f339e70 1308extern void release_locked_buffer(void *buffer, size_t size);
1da177e4
LT
1309#endif /* __KERNEL__ */
1310#endif /* _LINUX_MM_H */