mm: make register_page_bootmem_info_section() static
[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 102#define VM_ACCOUNT 0x00100000 /* Is a VM accounted object */
cdfd4325 103#define VM_NORESERVE 0x00200000 /* should the VM suppress accounting */
1da177e4
LT
104#define VM_HUGETLB 0x00400000 /* Huge TLB Page VM */
105#define VM_NONLINEAR 0x00800000 /* Is non-linear (remap_file_pages) */
106#define VM_MAPPED_COPY 0x01000000 /* T if mapped copy of data (nommu mmap) */
4d7672b4 107#define VM_INSERTPAGE 0x02000000 /* The vma has had "vm_insert_page()" done on it */
e5b97dde 108#define VM_ALWAYSDUMP 0x04000000 /* Always include in core dumps */
d00806b1 109
d0217ac0 110#define VM_CAN_NONLINEAR 0x08000000 /* Has ->fault & does nonlinear pages */
b379d790 111#define VM_MIXEDMAP 0x10000000 /* Can contain "struct page" and pure PFN pages */
aba46c50 112#define VM_SAO 0x20000000 /* Strong Access Ordering (powerpc) */
1da177e4
LT
113
114#ifndef VM_STACK_DEFAULT_FLAGS /* arch can override this */
115#define VM_STACK_DEFAULT_FLAGS VM_DATA_DEFAULT_FLAGS
116#endif
117
118#ifdef CONFIG_STACK_GROWSUP
119#define VM_STACK_FLAGS (VM_GROWSUP | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT)
120#else
121#define VM_STACK_FLAGS (VM_GROWSDOWN | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT)
122#endif
123
124#define VM_READHINTMASK (VM_SEQ_READ | VM_RAND_READ)
125#define VM_ClearReadHint(v) (v)->vm_flags &= ~VM_READHINTMASK
126#define VM_NormalReadHint(v) (!((v)->vm_flags & VM_READHINTMASK))
127#define VM_SequentialReadHint(v) ((v)->vm_flags & VM_SEQ_READ)
128#define VM_RandomReadHint(v) ((v)->vm_flags & VM_RAND_READ)
129
130/*
131 * mapping from the currently active vm_flags protection bits (the
132 * low four bits) to a page protection mask..
133 */
134extern pgprot_t protection_map[16];
135
d0217ac0
NP
136#define FAULT_FLAG_WRITE 0x01 /* Fault was a write access */
137#define FAULT_FLAG_NONLINEAR 0x02 /* Fault was via a nonlinear mapping */
138
139
54cb8821 140/*
d0217ac0 141 * vm_fault is filled by the the pagefault handler and passed to the vma's
83c54070
NP
142 * ->fault function. The vma's ->fault is responsible for returning a bitmask
143 * of VM_FAULT_xxx flags that give details about how the fault was handled.
54cb8821 144 *
d0217ac0
NP
145 * pgoff should be used in favour of virtual_address, if possible. If pgoff
146 * is used, one may set VM_CAN_NONLINEAR in the vma->vm_flags to get nonlinear
147 * mapping support.
54cb8821 148 */
d0217ac0
NP
149struct vm_fault {
150 unsigned int flags; /* FAULT_FLAG_xxx flags */
151 pgoff_t pgoff; /* Logical page offset based on vma */
152 void __user *virtual_address; /* Faulting virtual address */
153
154 struct page *page; /* ->fault handlers should return a
83c54070 155 * page here, unless VM_FAULT_NOPAGE
d0217ac0 156 * is set (which is also implied by
83c54070 157 * VM_FAULT_ERROR).
d0217ac0 158 */
54cb8821 159};
1da177e4
LT
160
161/*
162 * These are the virtual MM functions - opening of an area, closing and
163 * unmapping it (needed to keep files on disk up-to-date etc), pointer
164 * to the functions called when a no-page or a wp-page exception occurs.
165 */
166struct vm_operations_struct {
167 void (*open)(struct vm_area_struct * area);
168 void (*close)(struct vm_area_struct * area);
d0217ac0 169 int (*fault)(struct vm_area_struct *vma, struct vm_fault *vmf);
9637a5ef
DH
170
171 /* notification that a previously read-only page is about to become
172 * writable, if an error is returned it will cause a SIGBUS */
173 int (*page_mkwrite)(struct vm_area_struct *vma, struct page *page);
28b2ee20
RR
174
175 /* called by access_process_vm when get_user_pages() fails, typically
176 * for use by special VMAs that can switch between memory and hardware
177 */
178 int (*access)(struct vm_area_struct *vma, unsigned long addr,
179 void *buf, int len, int write);
1da177e4 180#ifdef CONFIG_NUMA
a6020ed7
LS
181 /*
182 * set_policy() op must add a reference to any non-NULL @new mempolicy
183 * to hold the policy upon return. Caller should pass NULL @new to
184 * remove a policy and fall back to surrounding context--i.e. do not
185 * install a MPOL_DEFAULT policy, nor the task or system default
186 * mempolicy.
187 */
1da177e4 188 int (*set_policy)(struct vm_area_struct *vma, struct mempolicy *new);
a6020ed7
LS
189
190 /*
191 * get_policy() op must add reference [mpol_get()] to any policy at
192 * (vma,addr) marked as MPOL_SHARED. The shared policy infrastructure
193 * in mm/mempolicy.c will do this automatically.
194 * get_policy() must NOT add a ref if the policy at (vma,addr) is not
195 * marked as MPOL_SHARED. vma policies are protected by the mmap_sem.
196 * If no [shared/vma] mempolicy exists at the addr, get_policy() op
197 * must return NULL--i.e., do not "fallback" to task or system default
198 * policy.
199 */
1da177e4
LT
200 struct mempolicy *(*get_policy)(struct vm_area_struct *vma,
201 unsigned long addr);
7b2259b3
CL
202 int (*migrate)(struct vm_area_struct *vma, const nodemask_t *from,
203 const nodemask_t *to, unsigned long flags);
1da177e4
LT
204#endif
205};
206
207struct mmu_gather;
208struct inode;
209
349aef0b
AM
210#define page_private(page) ((page)->private)
211#define set_page_private(page, v) ((page)->private = (v))
4c21e2f2 212
1da177e4
LT
213/*
214 * FIXME: take this include out, include page-flags.h in
215 * files which need it (119 of them)
216 */
217#include <linux/page-flags.h>
218
725d704e
NP
219#ifdef CONFIG_DEBUG_VM
220#define VM_BUG_ON(cond) BUG_ON(cond)
221#else
222#define VM_BUG_ON(condition) do { } while(0)
223#endif
224
1da177e4
LT
225/*
226 * Methods to modify the page usage count.
227 *
228 * What counts for a page usage:
229 * - cache mapping (page->mapping)
230 * - private data (page->private)
231 * - page mapped in a task's page tables, each mapping
232 * is counted separately
233 *
234 * Also, many kernel routines increase the page count before a critical
235 * routine so they can be sure the page doesn't go away from under them.
1da177e4
LT
236 */
237
238/*
da6052f7 239 * Drop a ref, return true if the refcount fell to zero (the page has no users)
1da177e4 240 */
7c8ee9a8
NP
241static inline int put_page_testzero(struct page *page)
242{
725d704e 243 VM_BUG_ON(atomic_read(&page->_count) == 0);
8dc04efb 244 return atomic_dec_and_test(&page->_count);
7c8ee9a8 245}
1da177e4
LT
246
247/*
7c8ee9a8
NP
248 * Try to grab a ref unless the page has a refcount of zero, return false if
249 * that is the case.
1da177e4 250 */
7c8ee9a8
NP
251static inline int get_page_unless_zero(struct page *page)
252{
aec2c3ed 253 VM_BUG_ON(PageTail(page));
8dc04efb 254 return atomic_inc_not_zero(&page->_count);
7c8ee9a8 255}
1da177e4 256
48667e7a 257/* Support for virtually mapped pages */
b3bdda02
CL
258struct page *vmalloc_to_page(const void *addr);
259unsigned long vmalloc_to_pfn(const void *addr);
48667e7a 260
0738c4bb
PM
261/*
262 * Determine if an address is within the vmalloc range
263 *
264 * On nommu, vmalloc/vfree wrap through kmalloc/kfree directly, so there
265 * is no special casing required.
266 */
9e2779fa
CL
267static inline int is_vmalloc_addr(const void *x)
268{
0738c4bb 269#ifdef CONFIG_MMU
9e2779fa
CL
270 unsigned long addr = (unsigned long)x;
271
272 return addr >= VMALLOC_START && addr < VMALLOC_END;
0738c4bb
PM
273#else
274 return 0;
8ca3ed87 275#endif
0738c4bb 276}
9e2779fa 277
d85f3385
CL
278static inline struct page *compound_head(struct page *page)
279{
6d777953 280 if (unlikely(PageTail(page)))
d85f3385
CL
281 return page->first_page;
282 return page;
283}
284
4c21e2f2 285static inline int page_count(struct page *page)
1da177e4 286{
d85f3385 287 return atomic_read(&compound_head(page)->_count);
1da177e4
LT
288}
289
290static inline void get_page(struct page *page)
291{
d85f3385 292 page = compound_head(page);
725d704e 293 VM_BUG_ON(atomic_read(&page->_count) == 0);
1da177e4
LT
294 atomic_inc(&page->_count);
295}
296
b49af68f
CL
297static inline struct page *virt_to_head_page(const void *x)
298{
299 struct page *page = virt_to_page(x);
300 return compound_head(page);
301}
302
7835e98b
NP
303/*
304 * Setup the page count before being freed into the page allocator for
305 * the first time (boot or memory hotplug)
306 */
307static inline void init_page_count(struct page *page)
308{
309 atomic_set(&page->_count, 1);
310}
311
1da177e4 312void put_page(struct page *page);
1d7ea732 313void put_pages_list(struct list_head *pages);
1da177e4 314
8dfcc9ba 315void split_page(struct page *page, unsigned int order);
8dfcc9ba 316
33f2ef89
AW
317/*
318 * Compound pages have a destructor function. Provide a
319 * prototype for that function and accessor functions.
320 * These are _only_ valid on the head of a PG_compound page.
321 */
322typedef void compound_page_dtor(struct page *);
323
324static inline void set_compound_page_dtor(struct page *page,
325 compound_page_dtor *dtor)
326{
327 page[1].lru.next = (void *)dtor;
328}
329
330static inline compound_page_dtor *get_compound_page_dtor(struct page *page)
331{
332 return (compound_page_dtor *)page[1].lru.next;
333}
334
d85f3385
CL
335static inline int compound_order(struct page *page)
336{
6d777953 337 if (!PageHead(page))
d85f3385
CL
338 return 0;
339 return (unsigned long)page[1].lru.prev;
340}
341
342static inline void set_compound_order(struct page *page, unsigned long order)
343{
344 page[1].lru.prev = (void *)order;
345}
346
1da177e4
LT
347/*
348 * Multiple processes may "see" the same page. E.g. for untouched
349 * mappings of /dev/null, all processes see the same page full of
350 * zeroes, and text pages of executables and shared libraries have
351 * only one copy in memory, at most, normally.
352 *
353 * For the non-reserved pages, page_count(page) denotes a reference count.
7e871b6c
PBG
354 * page_count() == 0 means the page is free. page->lru is then used for
355 * freelist management in the buddy allocator.
da6052f7 356 * page_count() > 0 means the page has been allocated.
1da177e4 357 *
da6052f7
NP
358 * Pages are allocated by the slab allocator in order to provide memory
359 * to kmalloc and kmem_cache_alloc. In this case, the management of the
360 * page, and the fields in 'struct page' are the responsibility of mm/slab.c
361 * unless a particular usage is carefully commented. (the responsibility of
362 * freeing the kmalloc memory is the caller's, of course).
1da177e4 363 *
da6052f7
NP
364 * A page may be used by anyone else who does a __get_free_page().
365 * In this case, page_count still tracks the references, and should only
366 * be used through the normal accessor functions. The top bits of page->flags
367 * and page->virtual store page management information, but all other fields
368 * are unused and could be used privately, carefully. The management of this
369 * page is the responsibility of the one who allocated it, and those who have
370 * subsequently been given references to it.
371 *
372 * The other pages (we may call them "pagecache pages") are completely
1da177e4
LT
373 * managed by the Linux memory manager: I/O, buffers, swapping etc.
374 * The following discussion applies only to them.
375 *
da6052f7
NP
376 * A pagecache page contains an opaque `private' member, which belongs to the
377 * page's address_space. Usually, this is the address of a circular list of
378 * the page's disk buffers. PG_private must be set to tell the VM to call
379 * into the filesystem to release these pages.
1da177e4 380 *
da6052f7
NP
381 * A page may belong to an inode's memory mapping. In this case, page->mapping
382 * is the pointer to the inode, and page->index is the file offset of the page,
383 * in units of PAGE_CACHE_SIZE.
1da177e4 384 *
da6052f7
NP
385 * If pagecache pages are not associated with an inode, they are said to be
386 * anonymous pages. These may become associated with the swapcache, and in that
387 * case PG_swapcache is set, and page->private is an offset into the swapcache.
1da177e4 388 *
da6052f7
NP
389 * In either case (swapcache or inode backed), the pagecache itself holds one
390 * reference to the page. Setting PG_private should also increment the
391 * refcount. The each user mapping also has a reference to the page.
1da177e4 392 *
da6052f7
NP
393 * The pagecache pages are stored in a per-mapping radix tree, which is
394 * rooted at mapping->page_tree, and indexed by offset.
395 * Where 2.4 and early 2.6 kernels kept dirty/clean pages in per-address_space
396 * lists, we instead now tag pages as dirty/writeback in the radix tree.
1da177e4 397 *
da6052f7 398 * All pagecache pages may be subject to I/O:
1da177e4
LT
399 * - inode pages may need to be read from disk,
400 * - inode pages which have been modified and are MAP_SHARED may need
da6052f7
NP
401 * to be written back to the inode on disk,
402 * - anonymous pages (including MAP_PRIVATE file mappings) which have been
403 * modified may need to be swapped out to swap space and (later) to be read
404 * back into memory.
1da177e4
LT
405 */
406
407/*
408 * The zone field is never updated after free_area_init_core()
409 * sets it, so none of the operations on it need to be atomic.
1da177e4 410 */
348f8b6c 411
d41dee36
AW
412
413/*
414 * page->flags layout:
415 *
416 * There are three possibilities for how page->flags get
417 * laid out. The first is for the normal case, without
418 * sparsemem. The second is for sparsemem when there is
419 * plenty of space for node and section. The last is when
420 * we have run out of space and have to fall back to an
421 * alternate (slower) way of determining the node.
422 *
308c05e3
CL
423 * No sparsemem or sparsemem vmemmap: | NODE | ZONE | ... | FLAGS |
424 * classic sparse with space for node:| SECTION | NODE | ZONE | ... | FLAGS |
425 * classic sparse no space for node: | SECTION | ZONE | ... | FLAGS |
d41dee36 426 */
308c05e3 427#if defined(CONFIG_SPARSEMEM) && !defined(CONFIG_SPARSEMEM_VMEMMAP)
d41dee36
AW
428#define SECTIONS_WIDTH SECTIONS_SHIFT
429#else
430#define SECTIONS_WIDTH 0
431#endif
432
433#define ZONES_WIDTH ZONES_SHIFT
434
9223b419 435#if SECTIONS_WIDTH+ZONES_WIDTH+NODES_SHIFT <= BITS_PER_LONG - NR_PAGEFLAGS
d41dee36
AW
436#define NODES_WIDTH NODES_SHIFT
437#else
308c05e3
CL
438#ifdef CONFIG_SPARSEMEM_VMEMMAP
439#error "Vmemmap: No space for nodes field in page flags"
440#endif
d41dee36
AW
441#define NODES_WIDTH 0
442#endif
443
444/* Page flags: | [SECTION] | [NODE] | ZONE | ... | FLAGS | */
07808b74 445#define SECTIONS_PGOFF ((sizeof(unsigned long)*8) - SECTIONS_WIDTH)
d41dee36
AW
446#define NODES_PGOFF (SECTIONS_PGOFF - NODES_WIDTH)
447#define ZONES_PGOFF (NODES_PGOFF - ZONES_WIDTH)
448
449/*
450 * We are going to use the flags for the page to node mapping if its in
451 * there. This includes the case where there is no node, so it is implicit.
452 */
89689ae7
CL
453#if !(NODES_WIDTH > 0 || NODES_SHIFT == 0)
454#define NODE_NOT_IN_PAGE_FLAGS
455#endif
d41dee36
AW
456
457#ifndef PFN_SECTION_SHIFT
458#define PFN_SECTION_SHIFT 0
459#endif
348f8b6c
DH
460
461/*
462 * Define the bit shifts to access each section. For non-existant
463 * sections we define the shift as 0; that plus a 0 mask ensures
464 * the compiler will optimise away reference to them.
465 */
d41dee36
AW
466#define SECTIONS_PGSHIFT (SECTIONS_PGOFF * (SECTIONS_WIDTH != 0))
467#define NODES_PGSHIFT (NODES_PGOFF * (NODES_WIDTH != 0))
468#define ZONES_PGSHIFT (ZONES_PGOFF * (ZONES_WIDTH != 0))
348f8b6c 469
89689ae7
CL
470/* NODE:ZONE or SECTION:ZONE is used to ID a zone for the buddy allcator */
471#ifdef NODE_NOT_IN_PAGEFLAGS
472#define ZONEID_SHIFT (SECTIONS_SHIFT + ZONES_SHIFT)
bd8029b6
AW
473#define ZONEID_PGOFF ((SECTIONS_PGOFF < ZONES_PGOFF)? \
474 SECTIONS_PGOFF : ZONES_PGOFF)
d41dee36 475#else
89689ae7 476#define ZONEID_SHIFT (NODES_SHIFT + ZONES_SHIFT)
bd8029b6
AW
477#define ZONEID_PGOFF ((NODES_PGOFF < ZONES_PGOFF)? \
478 NODES_PGOFF : ZONES_PGOFF)
89689ae7
CL
479#endif
480
bd8029b6 481#define ZONEID_PGSHIFT (ZONEID_PGOFF * (ZONEID_SHIFT != 0))
348f8b6c 482
9223b419
CL
483#if SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > BITS_PER_LONG - NR_PAGEFLAGS
484#error SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > BITS_PER_LONG - NR_PAGEFLAGS
348f8b6c
DH
485#endif
486
d41dee36
AW
487#define ZONES_MASK ((1UL << ZONES_WIDTH) - 1)
488#define NODES_MASK ((1UL << NODES_WIDTH) - 1)
489#define SECTIONS_MASK ((1UL << SECTIONS_WIDTH) - 1)
89689ae7 490#define ZONEID_MASK ((1UL << ZONEID_SHIFT) - 1)
348f8b6c 491
2f1b6248 492static inline enum zone_type page_zonenum(struct page *page)
1da177e4 493{
348f8b6c 494 return (page->flags >> ZONES_PGSHIFT) & ZONES_MASK;
1da177e4 495}
1da177e4 496
89689ae7
CL
497/*
498 * The identification function is only used by the buddy allocator for
499 * determining if two pages could be buddies. We are not really
500 * identifying a zone since we could be using a the section number
501 * id if we have not node id available in page flags.
502 * We guarantee only that it will return the same value for two
503 * combinable pages in a zone.
504 */
cb2b95e1
AW
505static inline int page_zone_id(struct page *page)
506{
89689ae7 507 return (page->flags >> ZONEID_PGSHIFT) & ZONEID_MASK;
348f8b6c
DH
508}
509
25ba77c1 510static inline int zone_to_nid(struct zone *zone)
89fa3024 511{
d5f541ed
CL
512#ifdef CONFIG_NUMA
513 return zone->node;
514#else
515 return 0;
516#endif
89fa3024
CL
517}
518
89689ae7 519#ifdef NODE_NOT_IN_PAGE_FLAGS
25ba77c1 520extern int page_to_nid(struct page *page);
89689ae7 521#else
25ba77c1 522static inline int page_to_nid(struct page *page)
d41dee36 523{
89689ae7 524 return (page->flags >> NODES_PGSHIFT) & NODES_MASK;
d41dee36 525}
89689ae7
CL
526#endif
527
528static inline struct zone *page_zone(struct page *page)
529{
530 return &NODE_DATA(page_to_nid(page))->node_zones[page_zonenum(page)];
531}
532
308c05e3 533#if defined(CONFIG_SPARSEMEM) && !defined(CONFIG_SPARSEMEM_VMEMMAP)
d41dee36
AW
534static inline unsigned long page_to_section(struct page *page)
535{
536 return (page->flags >> SECTIONS_PGSHIFT) & SECTIONS_MASK;
537}
308c05e3 538#endif
d41dee36 539
2f1b6248 540static inline void set_page_zone(struct page *page, enum zone_type zone)
348f8b6c
DH
541{
542 page->flags &= ~(ZONES_MASK << ZONES_PGSHIFT);
543 page->flags |= (zone & ZONES_MASK) << ZONES_PGSHIFT;
544}
2f1b6248 545
348f8b6c
DH
546static inline void set_page_node(struct page *page, unsigned long node)
547{
548 page->flags &= ~(NODES_MASK << NODES_PGSHIFT);
549 page->flags |= (node & NODES_MASK) << NODES_PGSHIFT;
1da177e4 550}
89689ae7 551
d41dee36
AW
552static inline void set_page_section(struct page *page, unsigned long section)
553{
554 page->flags &= ~(SECTIONS_MASK << SECTIONS_PGSHIFT);
555 page->flags |= (section & SECTIONS_MASK) << SECTIONS_PGSHIFT;
556}
1da177e4 557
2f1b6248 558static inline void set_page_links(struct page *page, enum zone_type zone,
d41dee36 559 unsigned long node, unsigned long pfn)
1da177e4 560{
348f8b6c
DH
561 set_page_zone(page, zone);
562 set_page_node(page, node);
d41dee36 563 set_page_section(page, pfn_to_section_nr(pfn));
1da177e4
LT
564}
565
7cd94146
EP
566/*
567 * If a hint addr is less than mmap_min_addr change hint to be as
568 * low as possible but still greater than mmap_min_addr
569 */
570static inline unsigned long round_hint_to_min(unsigned long hint)
571{
572#ifdef CONFIG_SECURITY
573 hint &= PAGE_MASK;
574 if (((void *)hint != NULL) &&
575 (hint < mmap_min_addr))
576 return PAGE_ALIGN(mmap_min_addr);
577#endif
578 return hint;
579}
580
f6ac2354
CL
581/*
582 * Some inline functions in vmstat.h depend on page_zone()
583 */
584#include <linux/vmstat.h>
585
652050ae 586static __always_inline void *lowmem_page_address(struct page *page)
1da177e4
LT
587{
588 return __va(page_to_pfn(page) << PAGE_SHIFT);
589}
590
591#if defined(CONFIG_HIGHMEM) && !defined(WANT_PAGE_VIRTUAL)
592#define HASHED_PAGE_VIRTUAL
593#endif
594
595#if defined(WANT_PAGE_VIRTUAL)
596#define page_address(page) ((page)->virtual)
597#define set_page_address(page, address) \
598 do { \
599 (page)->virtual = (address); \
600 } while(0)
601#define page_address_init() do { } while(0)
602#endif
603
604#if defined(HASHED_PAGE_VIRTUAL)
605void *page_address(struct page *page);
606void set_page_address(struct page *page, void *virtual);
607void page_address_init(void);
608#endif
609
610#if !defined(HASHED_PAGE_VIRTUAL) && !defined(WANT_PAGE_VIRTUAL)
611#define page_address(page) lowmem_page_address(page)
612#define set_page_address(page, address) do { } while(0)
613#define page_address_init() do { } while(0)
614#endif
615
616/*
617 * On an anonymous page mapped into a user virtual memory area,
618 * page->mapping points to its anon_vma, not to a struct address_space;
619 * with the PAGE_MAPPING_ANON bit set to distinguish it.
620 *
621 * Please note that, confusingly, "page_mapping" refers to the inode
622 * address_space which maps the page from disk; whereas "page_mapped"
623 * refers to user virtual address space into which the page is mapped.
624 */
625#define PAGE_MAPPING_ANON 1
626
627extern struct address_space swapper_space;
628static inline struct address_space *page_mapping(struct page *page)
629{
630 struct address_space *mapping = page->mapping;
631
b5fab14e 632 VM_BUG_ON(PageSlab(page));
726b8012 633#ifdef CONFIG_SWAP
1da177e4
LT
634 if (unlikely(PageSwapCache(page)))
635 mapping = &swapper_space;
726b8012
AM
636 else
637#endif
638 if (unlikely((unsigned long)mapping & PAGE_MAPPING_ANON))
1da177e4
LT
639 mapping = NULL;
640 return mapping;
641}
642
643static inline int PageAnon(struct page *page)
644{
645 return ((unsigned long)page->mapping & PAGE_MAPPING_ANON) != 0;
646}
647
648/*
649 * Return the pagecache index of the passed page. Regular pagecache pages
650 * use ->index whereas swapcache pages use ->private
651 */
652static inline pgoff_t page_index(struct page *page)
653{
654 if (unlikely(PageSwapCache(page)))
4c21e2f2 655 return page_private(page);
1da177e4
LT
656 return page->index;
657}
658
659/*
660 * The atomic page->_mapcount, like _count, starts from -1:
661 * so that transitions both from it and to it can be tracked,
662 * using atomic_inc_and_test and atomic_add_negative(-1).
663 */
664static inline void reset_page_mapcount(struct page *page)
665{
666 atomic_set(&(page)->_mapcount, -1);
667}
668
669static inline int page_mapcount(struct page *page)
670{
671 return atomic_read(&(page)->_mapcount) + 1;
672}
673
674/*
675 * Return true if this page is mapped into pagetables.
676 */
677static inline int page_mapped(struct page *page)
678{
679 return atomic_read(&(page)->_mapcount) >= 0;
680}
681
1da177e4
LT
682/*
683 * Different kinds of faults, as returned by handle_mm_fault().
684 * Used to decide whether a process gets delivered SIGBUS or
685 * just gets major/minor fault counters bumped up.
686 */
d0217ac0 687
83c54070 688#define VM_FAULT_MINOR 0 /* For backwards compat. Remove me quickly. */
d0217ac0 689
83c54070
NP
690#define VM_FAULT_OOM 0x0001
691#define VM_FAULT_SIGBUS 0x0002
692#define VM_FAULT_MAJOR 0x0004
693#define VM_FAULT_WRITE 0x0008 /* Special case for get_user_pages */
f33ea7f4 694
83c54070
NP
695#define VM_FAULT_NOPAGE 0x0100 /* ->fault installed the pte, not return page */
696#define VM_FAULT_LOCKED 0x0200 /* ->fault locked the returned page */
1da177e4 697
83c54070 698#define VM_FAULT_ERROR (VM_FAULT_OOM | VM_FAULT_SIGBUS)
d0217ac0 699
1da177e4
LT
700#define offset_in_page(p) ((unsigned long)(p) & ~PAGE_MASK)
701
702extern void show_free_areas(void);
703
704#ifdef CONFIG_SHMEM
1da177e4
LT
705int shmem_lock(struct file *file, int lock, struct user_struct *user);
706#else
03b00ebc
RK
707static inline int shmem_lock(struct file *file, int lock,
708 struct user_struct *user)
709{
710 return 0;
711}
1da177e4
LT
712#endif
713struct file *shmem_file_setup(char *name, loff_t size, unsigned long flags);
714
715int shmem_zero_setup(struct vm_area_struct *);
716
b0e15190
DH
717#ifndef CONFIG_MMU
718extern unsigned long shmem_get_unmapped_area(struct file *file,
719 unsigned long addr,
720 unsigned long len,
721 unsigned long pgoff,
722 unsigned long flags);
723#endif
724
e8edc6e0 725extern int can_do_mlock(void);
1da177e4
LT
726extern int user_shm_lock(size_t, struct user_struct *);
727extern void user_shm_unlock(size_t, struct user_struct *);
728
729/*
730 * Parameter block passed down to zap_pte_range in exceptional cases.
731 */
732struct zap_details {
733 struct vm_area_struct *nonlinear_vma; /* Check page->index if set */
734 struct address_space *check_mapping; /* Check page->mapping if set */
735 pgoff_t first_index; /* Lowest page->index to unmap */
736 pgoff_t last_index; /* Highest page->index to unmap */
737 spinlock_t *i_mmap_lock; /* For unmap_mapping_range: */
1da177e4
LT
738 unsigned long truncate_count; /* Compare vm_truncate_count */
739};
740
7e675137
NP
741struct page *vm_normal_page(struct vm_area_struct *vma, unsigned long addr,
742 pte_t pte);
743
ee39b37b 744unsigned long zap_page_range(struct vm_area_struct *vma, unsigned long address,
1da177e4 745 unsigned long size, struct zap_details *);
508034a3 746unsigned long unmap_vmas(struct mmu_gather **tlb,
1da177e4
LT
747 struct vm_area_struct *start_vma, unsigned long start_addr,
748 unsigned long end_addr, unsigned long *nr_accounted,
749 struct zap_details *);
e6473092
MM
750
751/**
752 * mm_walk - callbacks for walk_page_range
753 * @pgd_entry: if set, called for each non-empty PGD (top-level) entry
754 * @pud_entry: if set, called for each non-empty PUD (2nd-level) entry
755 * @pmd_entry: if set, called for each non-empty PMD (3rd-level) entry
756 * @pte_entry: if set, called for each non-empty PTE (4th-level) entry
757 * @pte_hole: if set, called for each hole at all levels
758 *
759 * (see walk_page_range for more details)
760 */
761struct mm_walk {
2165009b
DH
762 int (*pgd_entry)(pgd_t *, unsigned long, unsigned long, struct mm_walk *);
763 int (*pud_entry)(pud_t *, unsigned long, unsigned long, struct mm_walk *);
764 int (*pmd_entry)(pmd_t *, unsigned long, unsigned long, struct mm_walk *);
765 int (*pte_entry)(pte_t *, unsigned long, unsigned long, struct mm_walk *);
766 int (*pte_hole)(unsigned long, unsigned long, struct mm_walk *);
767 struct mm_struct *mm;
768 void *private;
e6473092
MM
769};
770
2165009b
DH
771int walk_page_range(unsigned long addr, unsigned long end,
772 struct mm_walk *walk);
42b77728 773void free_pgd_range(struct mmu_gather *tlb, unsigned long addr,
3bf5ee95 774 unsigned long end, unsigned long floor, unsigned long ceiling);
1da177e4
LT
775int copy_page_range(struct mm_struct *dst, struct mm_struct *src,
776 struct vm_area_struct *vma);
1da177e4
LT
777void unmap_mapping_range(struct address_space *mapping,
778 loff_t const holebegin, loff_t const holelen, int even_cows);
28b2ee20
RR
779int generic_access_phys(struct vm_area_struct *vma, unsigned long addr,
780 void *buf, int len, int write);
1da177e4
LT
781
782static inline void unmap_shared_mapping_range(struct address_space *mapping,
783 loff_t const holebegin, loff_t const holelen)
784{
785 unmap_mapping_range(mapping, holebegin, holelen, 0);
786}
787
788extern int vmtruncate(struct inode * inode, loff_t offset);
f6b3ec23 789extern int vmtruncate_range(struct inode * inode, loff_t offset, loff_t end);
f33ea7f4 790
7ee1dd3f 791#ifdef CONFIG_MMU
83c54070 792extern int handle_mm_fault(struct mm_struct *mm, struct vm_area_struct *vma,
7ee1dd3f 793 unsigned long address, int write_access);
7ee1dd3f
DH
794#else
795static inline int handle_mm_fault(struct mm_struct *mm,
796 struct vm_area_struct *vma, unsigned long address,
797 int write_access)
798{
799 /* should never happen if there's no MMU */
800 BUG();
801 return VM_FAULT_SIGBUS;
802}
803#endif
f33ea7f4 804
1da177e4
LT
805extern int make_pages_present(unsigned long addr, unsigned long end);
806extern int access_process_vm(struct task_struct *tsk, unsigned long addr, void *buf, int len, int write);
1da177e4
LT
807
808int get_user_pages(struct task_struct *tsk, struct mm_struct *mm, unsigned long start,
809 int len, int write, int force, struct page **pages, struct vm_area_struct **vmas);
b5810039 810void print_bad_pte(struct vm_area_struct *, pte_t, unsigned long);
1da177e4 811
cf9a2ae8
DH
812extern int try_to_release_page(struct page * page, gfp_t gfp_mask);
813extern void do_invalidatepage(struct page *page, unsigned long offset);
814
1da177e4 815int __set_page_dirty_nobuffers(struct page *page);
76719325 816int __set_page_dirty_no_writeback(struct page *page);
1da177e4
LT
817int redirty_page_for_writepage(struct writeback_control *wbc,
818 struct page *page);
b3c97528 819int set_page_dirty(struct page *page);
1da177e4
LT
820int set_page_dirty_lock(struct page *page);
821int clear_page_dirty_for_io(struct page *page);
822
b6a2fea3
OW
823extern unsigned long move_page_tables(struct vm_area_struct *vma,
824 unsigned long old_addr, struct vm_area_struct *new_vma,
825 unsigned long new_addr, unsigned long len);
1da177e4
LT
826extern unsigned long do_mremap(unsigned long addr,
827 unsigned long old_len, unsigned long new_len,
828 unsigned long flags, unsigned long new_addr);
b6a2fea3
OW
829extern int mprotect_fixup(struct vm_area_struct *vma,
830 struct vm_area_struct **pprev, unsigned long start,
831 unsigned long end, unsigned long newflags);
1da177e4
LT
832
833/*
8e1f936b 834 * A callback you can register to apply pressure to ageable caches.
1da177e4 835 *
8e1f936b
RR
836 * 'shrink' is passed a count 'nr_to_scan' and a 'gfpmask'. It should
837 * look through the least-recently-used 'nr_to_scan' entries and
838 * attempt to free them up. It should return the number of objects
839 * which remain in the cache. If it returns -1, it means it cannot do
840 * any scanning at this time (eg. there is a risk of deadlock).
1da177e4 841 *
8e1f936b
RR
842 * The 'gfpmask' refers to the allocation we are currently trying to
843 * fulfil.
844 *
845 * Note that 'shrink' will be passed nr_to_scan == 0 when the VM is
846 * querying the cache size, so a fastpath for that case is appropriate.
1da177e4 847 */
8e1f936b
RR
848struct shrinker {
849 int (*shrink)(int nr_to_scan, gfp_t gfp_mask);
850 int seeks; /* seeks to recreate an obj */
1da177e4 851
8e1f936b
RR
852 /* These are for internal use */
853 struct list_head list;
854 long nr; /* objs pending delete */
855};
856#define DEFAULT_SEEKS 2 /* A good number if you don't know better. */
857extern void register_shrinker(struct shrinker *);
858extern void unregister_shrinker(struct shrinker *);
1da177e4 859
4e950f6f 860int vma_wants_writenotify(struct vm_area_struct *vma);
d08b3851 861
b3c97528 862extern pte_t *get_locked_pte(struct mm_struct *mm, unsigned long addr, spinlock_t **ptl);
c9cfcddf 863
5f22df00
NP
864#ifdef __PAGETABLE_PUD_FOLDED
865static inline int __pud_alloc(struct mm_struct *mm, pgd_t *pgd,
866 unsigned long address)
867{
868 return 0;
869}
870#else
1bb3630e 871int __pud_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address);
5f22df00
NP
872#endif
873
874#ifdef __PAGETABLE_PMD_FOLDED
875static inline int __pmd_alloc(struct mm_struct *mm, pud_t *pud,
876 unsigned long address)
877{
878 return 0;
879}
880#else
1bb3630e 881int __pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address);
5f22df00
NP
882#endif
883
1bb3630e
HD
884int __pte_alloc(struct mm_struct *mm, pmd_t *pmd, unsigned long address);
885int __pte_alloc_kernel(pmd_t *pmd, unsigned long address);
886
1da177e4
LT
887/*
888 * The following ifdef needed to get the 4level-fixup.h header to work.
889 * Remove it when 4level-fixup.h has been removed.
890 */
1bb3630e 891#if defined(CONFIG_MMU) && !defined(__ARCH_HAS_4LEVEL_HACK)
1da177e4
LT
892static inline pud_t *pud_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address)
893{
1bb3630e
HD
894 return (unlikely(pgd_none(*pgd)) && __pud_alloc(mm, pgd, address))?
895 NULL: pud_offset(pgd, address);
1da177e4
LT
896}
897
898static inline pmd_t *pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address)
899{
1bb3630e
HD
900 return (unlikely(pud_none(*pud)) && __pmd_alloc(mm, pud, address))?
901 NULL: pmd_offset(pud, address);
1da177e4 902}
1bb3630e
HD
903#endif /* CONFIG_MMU && !__ARCH_HAS_4LEVEL_HACK */
904
4c21e2f2
HD
905#if NR_CPUS >= CONFIG_SPLIT_PTLOCK_CPUS
906/*
907 * We tuck a spinlock to guard each pagetable page into its struct page,
908 * at page->private, with BUILD_BUG_ON to make sure that this will not
909 * overflow into the next struct page (as it might with DEBUG_SPINLOCK).
910 * When freeing, reset page->mapping so free_pages_check won't complain.
911 */
349aef0b 912#define __pte_lockptr(page) &((page)->ptl)
4c21e2f2
HD
913#define pte_lock_init(_page) do { \
914 spin_lock_init(__pte_lockptr(_page)); \
915} while (0)
916#define pte_lock_deinit(page) ((page)->mapping = NULL)
917#define pte_lockptr(mm, pmd) ({(void)(mm); __pte_lockptr(pmd_page(*(pmd)));})
918#else
919/*
920 * We use mm->page_table_lock to guard all pagetable pages of the mm.
921 */
922#define pte_lock_init(page) do {} while (0)
923#define pte_lock_deinit(page) do {} while (0)
924#define pte_lockptr(mm, pmd) ({(void)(pmd); &(mm)->page_table_lock;})
925#endif /* NR_CPUS < CONFIG_SPLIT_PTLOCK_CPUS */
926
2f569afd
MS
927static inline void pgtable_page_ctor(struct page *page)
928{
929 pte_lock_init(page);
930 inc_zone_page_state(page, NR_PAGETABLE);
931}
932
933static inline void pgtable_page_dtor(struct page *page)
934{
935 pte_lock_deinit(page);
936 dec_zone_page_state(page, NR_PAGETABLE);
937}
938
c74df32c
HD
939#define pte_offset_map_lock(mm, pmd, address, ptlp) \
940({ \
4c21e2f2 941 spinlock_t *__ptl = pte_lockptr(mm, pmd); \
c74df32c
HD
942 pte_t *__pte = pte_offset_map(pmd, address); \
943 *(ptlp) = __ptl; \
944 spin_lock(__ptl); \
945 __pte; \
946})
947
948#define pte_unmap_unlock(pte, ptl) do { \
949 spin_unlock(ptl); \
950 pte_unmap(pte); \
951} while (0)
952
1bb3630e
HD
953#define pte_alloc_map(mm, pmd, address) \
954 ((unlikely(!pmd_present(*(pmd))) && __pte_alloc(mm, pmd, address))? \
955 NULL: pte_offset_map(pmd, address))
956
c74df32c
HD
957#define pte_alloc_map_lock(mm, pmd, address, ptlp) \
958 ((unlikely(!pmd_present(*(pmd))) && __pte_alloc(mm, pmd, address))? \
959 NULL: pte_offset_map_lock(mm, pmd, address, ptlp))
960
1bb3630e
HD
961#define pte_alloc_kernel(pmd, address) \
962 ((unlikely(!pmd_present(*(pmd))) && __pte_alloc_kernel(pmd, address))? \
963 NULL: pte_offset_kernel(pmd, address))
1da177e4
LT
964
965extern void free_area_init(unsigned long * zones_size);
9109fb7b
JW
966extern void free_area_init_node(int nid, unsigned long * zones_size,
967 unsigned long zone_start_pfn, unsigned long *zholes_size);
c713216d
MG
968#ifdef CONFIG_ARCH_POPULATES_NODE_MAP
969/*
970 * With CONFIG_ARCH_POPULATES_NODE_MAP set, an architecture may initialise its
971 * zones, allocate the backing mem_map and account for memory holes in a more
972 * architecture independent manner. This is a substitute for creating the
973 * zone_sizes[] and zholes_size[] arrays and passing them to
974 * free_area_init_node()
975 *
976 * An architecture is expected to register range of page frames backed by
977 * physical memory with add_active_range() before calling
978 * free_area_init_nodes() passing in the PFN each zone ends at. At a basic
979 * usage, an architecture is expected to do something like
980 *
981 * unsigned long max_zone_pfns[MAX_NR_ZONES] = {max_dma, max_normal_pfn,
982 * max_highmem_pfn};
983 * for_each_valid_physical_page_range()
984 * add_active_range(node_id, start_pfn, end_pfn)
985 * free_area_init_nodes(max_zone_pfns);
986 *
987 * If the architecture guarantees that there are no holes in the ranges
988 * registered with add_active_range(), free_bootmem_active_regions()
989 * will call free_bootmem_node() for each registered physical page range.
990 * Similarly sparse_memory_present_with_active_regions() calls
991 * memory_present() for each range when SPARSEMEM is enabled.
992 *
993 * See mm/page_alloc.c for more information on each function exposed by
994 * CONFIG_ARCH_POPULATES_NODE_MAP
995 */
996extern void free_area_init_nodes(unsigned long *max_zone_pfn);
997extern void add_active_range(unsigned int nid, unsigned long start_pfn,
998 unsigned long end_pfn);
cc1050ba
YL
999extern void remove_active_range(unsigned int nid, unsigned long start_pfn,
1000 unsigned long end_pfn);
fb01439c
MG
1001extern void push_node_boundaries(unsigned int nid, unsigned long start_pfn,
1002 unsigned long end_pfn);
c713216d
MG
1003extern void remove_all_active_ranges(void);
1004extern unsigned long absent_pages_in_range(unsigned long start_pfn,
1005 unsigned long end_pfn);
1006extern void get_pfn_range_for_nid(unsigned int nid,
1007 unsigned long *start_pfn, unsigned long *end_pfn);
1008extern unsigned long find_min_pfn_with_active_regions(void);
1009extern unsigned long find_max_pfn_with_active_regions(void);
1010extern void free_bootmem_with_active_regions(int nid,
1011 unsigned long max_low_pfn);
d52d53b8 1012typedef int (*work_fn_t)(unsigned long, unsigned long, void *);
b5bc6c0e 1013extern void work_with_active_regions(int nid, work_fn_t work_fn, void *data);
c713216d
MG
1014extern void sparse_memory_present_with_active_regions(int nid);
1015#ifndef CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID
1016extern int early_pfn_to_nid(unsigned long pfn);
1017#endif /* CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID */
1018#endif /* CONFIG_ARCH_POPULATES_NODE_MAP */
0e0b864e 1019extern void set_dma_reserve(unsigned long new_dma_reserve);
a2f3aa02
DH
1020extern void memmap_init_zone(unsigned long, int, unsigned long,
1021 unsigned long, enum memmap_context);
3947be19 1022extern void setup_per_zone_pages_min(void);
1da177e4
LT
1023extern void mem_init(void);
1024extern void show_mem(void);
1025extern void si_meminfo(struct sysinfo * val);
1026extern void si_meminfo_node(struct sysinfo *val, int nid);
3461b0af 1027extern int after_bootmem;
1da177e4 1028
e7c8d5c9
CL
1029#ifdef CONFIG_NUMA
1030extern void setup_per_cpu_pageset(void);
1031#else
1032static inline void setup_per_cpu_pageset(void) {}
1033#endif
1034
1da177e4
LT
1035/* prio_tree.c */
1036void vma_prio_tree_add(struct vm_area_struct *, struct vm_area_struct *old);
1037void vma_prio_tree_insert(struct vm_area_struct *, struct prio_tree_root *);
1038void vma_prio_tree_remove(struct vm_area_struct *, struct prio_tree_root *);
1039struct vm_area_struct *vma_prio_tree_next(struct vm_area_struct *vma,
1040 struct prio_tree_iter *iter);
1041
1042#define vma_prio_tree_foreach(vma, iter, root, begin, end) \
1043 for (prio_tree_iter_init(iter, root, begin, end), vma = NULL; \
1044 (vma = vma_prio_tree_next(vma, iter)); )
1045
1046static inline void vma_nonlinear_insert(struct vm_area_struct *vma,
1047 struct list_head *list)
1048{
1049 vma->shared.vm_set.parent = NULL;
1050 list_add_tail(&vma->shared.vm_set.list, list);
1051}
1052
1053/* mmap.c */
34b4e4aa 1054extern int __vm_enough_memory(struct mm_struct *mm, long pages, int cap_sys_admin);
1da177e4
LT
1055extern void vma_adjust(struct vm_area_struct *vma, unsigned long start,
1056 unsigned long end, pgoff_t pgoff, struct vm_area_struct *insert);
1057extern struct vm_area_struct *vma_merge(struct mm_struct *,
1058 struct vm_area_struct *prev, unsigned long addr, unsigned long end,
1059 unsigned long vm_flags, struct anon_vma *, struct file *, pgoff_t,
1060 struct mempolicy *);
1061extern struct anon_vma *find_mergeable_anon_vma(struct vm_area_struct *);
1062extern int split_vma(struct mm_struct *,
1063 struct vm_area_struct *, unsigned long addr, int new_below);
1064extern int insert_vm_struct(struct mm_struct *, struct vm_area_struct *);
1065extern void __vma_link_rb(struct mm_struct *, struct vm_area_struct *,
1066 struct rb_node **, struct rb_node *);
a8fb5618 1067extern void unlink_file_vma(struct vm_area_struct *);
1da177e4
LT
1068extern struct vm_area_struct *copy_vma(struct vm_area_struct **,
1069 unsigned long addr, unsigned long len, pgoff_t pgoff);
1070extern void exit_mmap(struct mm_struct *);
925d1c40
MH
1071
1072#ifdef CONFIG_PROC_FS
1073/* From fs/proc/base.c. callers must _not_ hold the mm's exe_file_lock */
1074extern void added_exe_file_vma(struct mm_struct *mm);
1075extern void removed_exe_file_vma(struct mm_struct *mm);
1076#else
1077static inline void added_exe_file_vma(struct mm_struct *mm)
1078{}
1079
1080static inline void removed_exe_file_vma(struct mm_struct *mm)
1081{}
1082#endif /* CONFIG_PROC_FS */
1083
119f657c 1084extern int may_expand_vm(struct mm_struct *mm, unsigned long npages);
fa5dc22f
RM
1085extern int install_special_mapping(struct mm_struct *mm,
1086 unsigned long addr, unsigned long len,
1087 unsigned long flags, struct page **pages);
1da177e4
LT
1088
1089extern unsigned long get_unmapped_area(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
1090
1091extern unsigned long do_mmap_pgoff(struct file *file, unsigned long addr,
1092 unsigned long len, unsigned long prot,
1093 unsigned long flag, unsigned long pgoff);
0165ab44
MS
1094extern unsigned long mmap_region(struct file *file, unsigned long addr,
1095 unsigned long len, unsigned long flags,
1096 unsigned int vm_flags, unsigned long pgoff,
1097 int accountable);
1da177e4
LT
1098
1099static inline unsigned long do_mmap(struct file *file, unsigned long addr,
1100 unsigned long len, unsigned long prot,
1101 unsigned long flag, unsigned long offset)
1102{
1103 unsigned long ret = -EINVAL;
1104 if ((offset + PAGE_ALIGN(len)) < offset)
1105 goto out;
1106 if (!(offset & ~PAGE_MASK))
1107 ret = do_mmap_pgoff(file, addr, len, prot, flag, offset >> PAGE_SHIFT);
1108out:
1109 return ret;
1110}
1111
1112extern int do_munmap(struct mm_struct *, unsigned long, size_t);
1113
1114extern unsigned long do_brk(unsigned long, unsigned long);
1115
1116/* filemap.c */
1117extern unsigned long page_unuse(struct page *);
1118extern void truncate_inode_pages(struct address_space *, loff_t);
d7339071
HR
1119extern void truncate_inode_pages_range(struct address_space *,
1120 loff_t lstart, loff_t lend);
1da177e4
LT
1121
1122/* generic vm_area_ops exported for stackable file systems */
d0217ac0 1123extern int filemap_fault(struct vm_area_struct *, struct vm_fault *);
1da177e4
LT
1124
1125/* mm/page-writeback.c */
1126int write_one_page(struct page *page, int wait);
1127
1128/* readahead.c */
1129#define VM_MAX_READAHEAD 128 /* kbytes */
1130#define VM_MIN_READAHEAD 16 /* kbytes (includes current page) */
1da177e4
LT
1131
1132int do_page_cache_readahead(struct address_space *mapping, struct file *filp,
7361f4d8 1133 pgoff_t offset, unsigned long nr_to_read);
1da177e4 1134int force_page_cache_readahead(struct address_space *mapping, struct file *filp,
7361f4d8 1135 pgoff_t offset, unsigned long nr_to_read);
cf914a7d
RR
1136
1137void page_cache_sync_readahead(struct address_space *mapping,
1138 struct file_ra_state *ra,
1139 struct file *filp,
1140 pgoff_t offset,
1141 unsigned long size);
1142
1143void page_cache_async_readahead(struct address_space *mapping,
1144 struct file_ra_state *ra,
1145 struct file *filp,
1146 struct page *pg,
1147 pgoff_t offset,
1148 unsigned long size);
1149
1da177e4
LT
1150unsigned long max_sane_readahead(unsigned long nr);
1151
1152/* Do stack extension */
46dea3d0 1153extern int expand_stack(struct vm_area_struct *vma, unsigned long address);
9ab88515 1154#ifdef CONFIG_IA64
46dea3d0 1155extern int expand_upwards(struct vm_area_struct *vma, unsigned long address);
9ab88515 1156#endif
b6a2fea3
OW
1157extern int expand_stack_downwards(struct vm_area_struct *vma,
1158 unsigned long address);
1da177e4
LT
1159
1160/* Look up the first VMA which satisfies addr < vm_end, NULL if none. */
1161extern struct vm_area_struct * find_vma(struct mm_struct * mm, unsigned long addr);
1162extern struct vm_area_struct * find_vma_prev(struct mm_struct * mm, unsigned long addr,
1163 struct vm_area_struct **pprev);
1164
1165/* Look up the first VMA which intersects the interval start_addr..end_addr-1,
1166 NULL if none. Assume start_addr < end_addr. */
1167static inline struct vm_area_struct * find_vma_intersection(struct mm_struct * mm, unsigned long start_addr, unsigned long end_addr)
1168{
1169 struct vm_area_struct * vma = find_vma(mm,start_addr);
1170
1171 if (vma && end_addr <= vma->vm_start)
1172 vma = NULL;
1173 return vma;
1174}
1175
1176static inline unsigned long vma_pages(struct vm_area_struct *vma)
1177{
1178 return (vma->vm_end - vma->vm_start) >> PAGE_SHIFT;
1179}
1180
804af2cf 1181pgprot_t vm_get_page_prot(unsigned long vm_flags);
deceb6cd 1182struct vm_area_struct *find_extend_vma(struct mm_struct *, unsigned long addr);
deceb6cd
HD
1183int remap_pfn_range(struct vm_area_struct *, unsigned long addr,
1184 unsigned long pfn, unsigned long size, pgprot_t);
a145dd41 1185int vm_insert_page(struct vm_area_struct *, unsigned long addr, struct page *);
e0dc0d8f
NP
1186int vm_insert_pfn(struct vm_area_struct *vma, unsigned long addr,
1187 unsigned long pfn);
423bad60
NP
1188int vm_insert_mixed(struct vm_area_struct *vma, unsigned long addr,
1189 unsigned long pfn);
deceb6cd 1190
6aab341e 1191struct page *follow_page(struct vm_area_struct *, unsigned long address,
deceb6cd
HD
1192 unsigned int foll_flags);
1193#define FOLL_WRITE 0x01 /* check pte is writable */
1194#define FOLL_TOUCH 0x02 /* mark page accessed */
1195#define FOLL_GET 0x04 /* do get_page on page */
1196#define FOLL_ANON 0x08 /* give ZERO_PAGE if no pgtable */
1da177e4 1197
2f569afd 1198typedef int (*pte_fn_t)(pte_t *pte, pgtable_t token, unsigned long addr,
aee16b3c
JF
1199 void *data);
1200extern int apply_to_page_range(struct mm_struct *mm, unsigned long address,
1201 unsigned long size, pte_fn_t fn, void *data);
1202
1da177e4 1203#ifdef CONFIG_PROC_FS
ab50b8ed 1204void vm_stat_account(struct mm_struct *, unsigned long, struct file *, long);
1da177e4 1205#else
ab50b8ed 1206static inline void vm_stat_account(struct mm_struct *mm,
1da177e4
LT
1207 unsigned long flags, struct file *file, long pages)
1208{
1209}
1210#endif /* CONFIG_PROC_FS */
1211
12d6f21e
IM
1212#ifdef CONFIG_DEBUG_PAGEALLOC
1213extern int debug_pagealloc_enabled;
1214
1215extern void kernel_map_pages(struct page *page, int numpages, int enable);
1216
1217static inline void enable_debug_pagealloc(void)
1218{
1219 debug_pagealloc_enabled = 1;
1220}
8a235efa
RW
1221#ifdef CONFIG_HIBERNATION
1222extern bool kernel_page_present(struct page *page);
1223#endif /* CONFIG_HIBERNATION */
12d6f21e 1224#else
1da177e4 1225static inline void
9858db50 1226kernel_map_pages(struct page *page, int numpages, int enable) {}
12d6f21e
IM
1227static inline void enable_debug_pagealloc(void)
1228{
1229}
8a235efa
RW
1230#ifdef CONFIG_HIBERNATION
1231static inline bool kernel_page_present(struct page *page) { return true; }
1232#endif /* CONFIG_HIBERNATION */
1da177e4
LT
1233#endif
1234
1235extern struct vm_area_struct *get_gate_vma(struct task_struct *tsk);
1236#ifdef __HAVE_ARCH_GATE_AREA
1237int in_gate_area_no_task(unsigned long addr);
1238int in_gate_area(struct task_struct *task, unsigned long addr);
1239#else
1240int in_gate_area_no_task(unsigned long addr);
1241#define in_gate_area(task, addr) ({(void)task; in_gate_area_no_task(addr);})
1242#endif /* __HAVE_ARCH_GATE_AREA */
1243
9d0243bc
AM
1244int drop_caches_sysctl_handler(struct ctl_table *, int, struct file *,
1245 void __user *, size_t *, loff_t *);
69e05944 1246unsigned long shrink_slab(unsigned long scanned, gfp_t gfp_mask,
9d0243bc 1247 unsigned long lru_pages);
9d0243bc 1248
7a9166e3
LY
1249#ifndef CONFIG_MMU
1250#define randomize_va_space 0
1251#else
a62eaf15 1252extern int randomize_va_space;
7a9166e3 1253#endif
a62eaf15 1254
045e72ac 1255const char * arch_vma_name(struct vm_area_struct *vma);
03252919 1256void print_vma_addr(char *prefix, unsigned long rip);
e6e5494c 1257
98f3cfc1 1258struct page *sparse_mem_map_populate(unsigned long pnum, int nid);
29c71111
AW
1259pgd_t *vmemmap_pgd_populate(unsigned long addr, int node);
1260pud_t *vmemmap_pud_populate(pgd_t *pgd, unsigned long addr, int node);
1261pmd_t *vmemmap_pmd_populate(pud_t *pud, unsigned long addr, int node);
1262pte_t *vmemmap_pte_populate(pmd_t *pmd, unsigned long addr, int node);
8f6aac41
CL
1263void *vmemmap_alloc_block(unsigned long size, int node);
1264void vmemmap_verify(pte_t *, int, unsigned long, unsigned long);
29c71111
AW
1265int vmemmap_populate_basepages(struct page *start_page,
1266 unsigned long pages, int node);
1267int vmemmap_populate(struct page *start_page, unsigned long pages, int node);
c2b91e2e 1268void vmemmap_populate_print_last(void);
8f6aac41 1269
1da177e4
LT
1270#endif /* __KERNEL__ */
1271#endif /* _LINUX_MM_H */