mm.h 43 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662663664665666667668669670671672673674675676677678679680681682683684685686687688689690691692693694695696697698699700701702703704705706707708709710711712713714715716717718719720721722723724725726727728729730731732733734735736737738739740741742743744745746747748749750751752753754755756757758759760761762763764765766767768769770771772773774775776777778779780781782783784785786787788789790791792793794795796797798799800801802803804805806807808809810811812813814815816817818819820821822823824825826827828829830831832833834835836837838839840841842843844845846847848849850851852853854855856857858859860861862863864865866867868869870871872873874875876877878879880881882883884885886887888889890891892893894895896897898899900901902903904905906907908909910911912913914915916917918919920921922923924925926927928929930931932933934935936937938939940941942943944945946947948949950951952953954955956957958959960961962963964965966967968969970971972973974975976977978979980981982983984985986987988989990991992993994995996997998999100010011002100310041005100610071008100910101011101210131014101510161017101810191020102110221023102410251026102710281029103010311032103310341035103610371038103910401041104210431044104510461047104810491050105110521053105410551056105710581059106010611062106310641065106610671068106910701071107210731074107510761077107810791080108110821083108410851086108710881089109010911092109310941095109610971098109911001101110211031104110511061107110811091110111111121113111411151116111711181119112011211122112311241125112611271128112911301131113211331134113511361137113811391140114111421143114411451146114711481149115011511152115311541155115611571158115911601161116211631164116511661167116811691170117111721173117411751176117711781179118011811182118311841185118611871188118911901191119211931194119511961197119811991200120112021203120412051206120712081209121012111212121312141215121612171218121912201221122212231224122512261227122812291230123112321233123412351236123712381239124012411242124312441245124612471248124912501251125212531254125512561257125812591260126112621263126412651266126712681269127012711272
  1. #ifndef _LINUX_MM_H
  2. #define _LINUX_MM_H
  3. #include <linux/errno.h>
  4. #ifdef __KERNEL__
  5. #include <linux/gfp.h>
  6. #include <linux/list.h>
  7. #include <linux/mmzone.h>
  8. #include <linux/rbtree.h>
  9. #include <linux/prio_tree.h>
  10. #include <linux/debug_locks.h>
  11. #include <linux/mm_types.h>
  12. struct mempolicy;
  13. struct anon_vma;
  14. struct file_ra_state;
  15. struct user_struct;
  16. struct writeback_control;
  17. #ifndef CONFIG_DISCONTIGMEM /* Don't use mapnrs, do it properly */
  18. extern unsigned long max_mapnr;
  19. #endif
  20. extern unsigned long num_physpages;
  21. extern void * high_memory;
  22. extern int page_cluster;
  23. #ifdef CONFIG_SYSCTL
  24. extern int sysctl_legacy_va_layout;
  25. #else
  26. #define sysctl_legacy_va_layout 0
  27. #endif
  28. extern unsigned long mmap_min_addr;
  29. #include <asm/page.h>
  30. #include <asm/pgtable.h>
  31. #include <asm/processor.h>
  32. #define nth_page(page,n) pfn_to_page(page_to_pfn((page)) + (n))
  33. /*
  34. * Linux kernel virtual memory manager primitives.
  35. * The idea being to have a "virtual" mm in the same way
  36. * we have a virtual fs - giving a cleaner interface to the
  37. * mm details, and allowing different kinds of memory mappings
  38. * (from shared memory to executable loading to arbitrary
  39. * mmap() functions).
  40. */
  41. extern struct kmem_cache *vm_area_cachep;
  42. /*
  43. * This struct defines the per-mm list of VMAs for uClinux. If CONFIG_MMU is
  44. * disabled, then there's a single shared list of VMAs maintained by the
  45. * system, and mm's subscribe to these individually
  46. */
  47. struct vm_list_struct {
  48. struct vm_list_struct *next;
  49. struct vm_area_struct *vma;
  50. };
  51. #ifndef CONFIG_MMU
  52. extern struct rb_root nommu_vma_tree;
  53. extern struct rw_semaphore nommu_vma_sem;
  54. extern unsigned int kobjsize(const void *objp);
  55. #endif
  56. /*
  57. * vm_flags..
  58. */
  59. #define VM_READ 0x00000001 /* currently active flags */
  60. #define VM_WRITE 0x00000002
  61. #define VM_EXEC 0x00000004
  62. #define VM_SHARED 0x00000008
  63. /* mprotect() hardcodes VM_MAYREAD >> 4 == VM_READ, and so for r/w/x bits. */
  64. #define VM_MAYREAD 0x00000010 /* limits for mprotect() etc */
  65. #define VM_MAYWRITE 0x00000020
  66. #define VM_MAYEXEC 0x00000040
  67. #define VM_MAYSHARE 0x00000080
  68. #define VM_GROWSDOWN 0x00000100 /* general info on the segment */
  69. #define VM_GROWSUP 0x00000200
  70. #define VM_PFNMAP 0x00000400 /* Page-ranges managed without "struct page", just pure PFN */
  71. #define VM_DENYWRITE 0x00000800 /* ETXTBSY on write attempts.. */
  72. #define VM_EXECUTABLE 0x00001000
  73. #define VM_LOCKED 0x00002000
  74. #define VM_IO 0x00004000 /* Memory mapped I/O or similar */
  75. /* Used by sys_madvise() */
  76. #define VM_SEQ_READ 0x00008000 /* App will access data sequentially */
  77. #define VM_RAND_READ 0x00010000 /* App will not benefit from clustered reads */
  78. #define VM_DONTCOPY 0x00020000 /* Do not copy this vma on fork */
  79. #define VM_DONTEXPAND 0x00040000 /* Cannot expand with mremap() */
  80. #define VM_RESERVED 0x00080000 /* Count as reserved_vm like IO */
  81. #define VM_ACCOUNT 0x00100000 /* Is a VM accounted object */
  82. #define VM_HUGETLB 0x00400000 /* Huge TLB Page VM */
  83. #define VM_NONLINEAR 0x00800000 /* Is non-linear (remap_file_pages) */
  84. #define VM_MAPPED_COPY 0x01000000 /* T if mapped copy of data (nommu mmap) */
  85. #define VM_INSERTPAGE 0x02000000 /* The vma has had "vm_insert_page()" done on it */
  86. #define VM_ALWAYSDUMP 0x04000000 /* Always include in core dumps */
  87. #define VM_CAN_NONLINEAR 0x08000000 /* Has ->fault & does nonlinear pages */
  88. #define VM_MIXEDMAP 0x10000000 /* Can contain "struct page" and pure PFN pages */
  89. #ifndef VM_STACK_DEFAULT_FLAGS /* arch can override this */
  90. #define VM_STACK_DEFAULT_FLAGS VM_DATA_DEFAULT_FLAGS
  91. #endif
  92. #ifdef CONFIG_STACK_GROWSUP
  93. #define VM_STACK_FLAGS (VM_GROWSUP | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT)
  94. #else
  95. #define VM_STACK_FLAGS (VM_GROWSDOWN | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT)
  96. #endif
  97. #define VM_READHINTMASK (VM_SEQ_READ | VM_RAND_READ)
  98. #define VM_ClearReadHint(v) (v)->vm_flags &= ~VM_READHINTMASK
  99. #define VM_NormalReadHint(v) (!((v)->vm_flags & VM_READHINTMASK))
  100. #define VM_SequentialReadHint(v) ((v)->vm_flags & VM_SEQ_READ)
  101. #define VM_RandomReadHint(v) ((v)->vm_flags & VM_RAND_READ)
  102. /*
  103. * mapping from the currently active vm_flags protection bits (the
  104. * low four bits) to a page protection mask..
  105. */
  106. extern pgprot_t protection_map[16];
  107. #define FAULT_FLAG_WRITE 0x01 /* Fault was a write access */
  108. #define FAULT_FLAG_NONLINEAR 0x02 /* Fault was via a nonlinear mapping */
  109. /*
  110. * vm_fault is filled by the the pagefault handler and passed to the vma's
  111. * ->fault function. The vma's ->fault is responsible for returning a bitmask
  112. * of VM_FAULT_xxx flags that give details about how the fault was handled.
  113. *
  114. * pgoff should be used in favour of virtual_address, if possible. If pgoff
  115. * is used, one may set VM_CAN_NONLINEAR in the vma->vm_flags to get nonlinear
  116. * mapping support.
  117. */
  118. struct vm_fault {
  119. unsigned int flags; /* FAULT_FLAG_xxx flags */
  120. pgoff_t pgoff; /* Logical page offset based on vma */
  121. void __user *virtual_address; /* Faulting virtual address */
  122. struct page *page; /* ->fault handlers should return a
  123. * page here, unless VM_FAULT_NOPAGE
  124. * is set (which is also implied by
  125. * VM_FAULT_ERROR).
  126. */
  127. };
  128. /*
  129. * These are the virtual MM functions - opening of an area, closing and
  130. * unmapping it (needed to keep files on disk up-to-date etc), pointer
  131. * to the functions called when a no-page or a wp-page exception occurs.
  132. */
  133. struct vm_operations_struct {
  134. void (*open)(struct vm_area_struct * area);
  135. void (*close)(struct vm_area_struct * area);
  136. int (*fault)(struct vm_area_struct *vma, struct vm_fault *vmf);
  137. unsigned long (*nopfn)(struct vm_area_struct *area,
  138. unsigned long address);
  139. /* notification that a previously read-only page is about to become
  140. * writable, if an error is returned it will cause a SIGBUS */
  141. int (*page_mkwrite)(struct vm_area_struct *vma, struct page *page);
  142. #ifdef CONFIG_NUMA
  143. /*
  144. * set_policy() op must add a reference to any non-NULL @new mempolicy
  145. * to hold the policy upon return. Caller should pass NULL @new to
  146. * remove a policy and fall back to surrounding context--i.e. do not
  147. * install a MPOL_DEFAULT policy, nor the task or system default
  148. * mempolicy.
  149. */
  150. int (*set_policy)(struct vm_area_struct *vma, struct mempolicy *new);
  151. /*
  152. * get_policy() op must add reference [mpol_get()] to any policy at
  153. * (vma,addr) marked as MPOL_SHARED. The shared policy infrastructure
  154. * in mm/mempolicy.c will do this automatically.
  155. * get_policy() must NOT add a ref if the policy at (vma,addr) is not
  156. * marked as MPOL_SHARED. vma policies are protected by the mmap_sem.
  157. * If no [shared/vma] mempolicy exists at the addr, get_policy() op
  158. * must return NULL--i.e., do not "fallback" to task or system default
  159. * policy.
  160. */
  161. struct mempolicy *(*get_policy)(struct vm_area_struct *vma,
  162. unsigned long addr);
  163. int (*migrate)(struct vm_area_struct *vma, const nodemask_t *from,
  164. const nodemask_t *to, unsigned long flags);
  165. #endif
  166. };
  167. struct mmu_gather;
  168. struct inode;
  169. #define page_private(page) ((page)->private)
  170. #define set_page_private(page, v) ((page)->private = (v))
  171. /*
  172. * FIXME: take this include out, include page-flags.h in
  173. * files which need it (119 of them)
  174. */
  175. #include <linux/page-flags.h>
  176. #ifdef CONFIG_DEBUG_VM
  177. #define VM_BUG_ON(cond) BUG_ON(cond)
  178. #else
  179. #define VM_BUG_ON(condition) do { } while(0)
  180. #endif
  181. /*
  182. * Methods to modify the page usage count.
  183. *
  184. * What counts for a page usage:
  185. * - cache mapping (page->mapping)
  186. * - private data (page->private)
  187. * - page mapped in a task's page tables, each mapping
  188. * is counted separately
  189. *
  190. * Also, many kernel routines increase the page count before a critical
  191. * routine so they can be sure the page doesn't go away from under them.
  192. */
  193. /*
  194. * Drop a ref, return true if the refcount fell to zero (the page has no users)
  195. */
  196. static inline int put_page_testzero(struct page *page)
  197. {
  198. VM_BUG_ON(atomic_read(&page->_count) == 0);
  199. return atomic_dec_and_test(&page->_count);
  200. }
  201. /*
  202. * Try to grab a ref unless the page has a refcount of zero, return false if
  203. * that is the case.
  204. */
  205. static inline int get_page_unless_zero(struct page *page)
  206. {
  207. VM_BUG_ON(PageTail(page));
  208. return atomic_inc_not_zero(&page->_count);
  209. }
  210. /* Support for virtually mapped pages */
  211. struct page *vmalloc_to_page(const void *addr);
  212. unsigned long vmalloc_to_pfn(const void *addr);
  213. /*
  214. * Determine if an address is within the vmalloc range
  215. *
  216. * On nommu, vmalloc/vfree wrap through kmalloc/kfree directly, so there
  217. * is no special casing required.
  218. */
  219. static inline int is_vmalloc_addr(const void *x)
  220. {
  221. #ifdef CONFIG_MMU
  222. unsigned long addr = (unsigned long)x;
  223. return addr >= VMALLOC_START && addr < VMALLOC_END;
  224. #else
  225. return 0;
  226. #endif
  227. }
  228. static inline struct page *compound_head(struct page *page)
  229. {
  230. if (unlikely(PageTail(page)))
  231. return page->first_page;
  232. return page;
  233. }
  234. static inline int page_count(struct page *page)
  235. {
  236. return atomic_read(&compound_head(page)->_count);
  237. }
  238. static inline void get_page(struct page *page)
  239. {
  240. page = compound_head(page);
  241. VM_BUG_ON(atomic_read(&page->_count) == 0);
  242. atomic_inc(&page->_count);
  243. }
  244. static inline struct page *virt_to_head_page(const void *x)
  245. {
  246. struct page *page = virt_to_page(x);
  247. return compound_head(page);
  248. }
  249. /*
  250. * Setup the page count before being freed into the page allocator for
  251. * the first time (boot or memory hotplug)
  252. */
  253. static inline void init_page_count(struct page *page)
  254. {
  255. atomic_set(&page->_count, 1);
  256. }
  257. void put_page(struct page *page);
  258. void put_pages_list(struct list_head *pages);
  259. void split_page(struct page *page, unsigned int order);
  260. /*
  261. * Compound pages have a destructor function. Provide a
  262. * prototype for that function and accessor functions.
  263. * These are _only_ valid on the head of a PG_compound page.
  264. */
  265. typedef void compound_page_dtor(struct page *);
  266. static inline void set_compound_page_dtor(struct page *page,
  267. compound_page_dtor *dtor)
  268. {
  269. page[1].lru.next = (void *)dtor;
  270. }
  271. static inline compound_page_dtor *get_compound_page_dtor(struct page *page)
  272. {
  273. return (compound_page_dtor *)page[1].lru.next;
  274. }
  275. static inline int compound_order(struct page *page)
  276. {
  277. if (!PageHead(page))
  278. return 0;
  279. return (unsigned long)page[1].lru.prev;
  280. }
  281. static inline void set_compound_order(struct page *page, unsigned long order)
  282. {
  283. page[1].lru.prev = (void *)order;
  284. }
  285. /*
  286. * Multiple processes may "see" the same page. E.g. for untouched
  287. * mappings of /dev/null, all processes see the same page full of
  288. * zeroes, and text pages of executables and shared libraries have
  289. * only one copy in memory, at most, normally.
  290. *
  291. * For the non-reserved pages, page_count(page) denotes a reference count.
  292. * page_count() == 0 means the page is free. page->lru is then used for
  293. * freelist management in the buddy allocator.
  294. * page_count() > 0 means the page has been allocated.
  295. *
  296. * Pages are allocated by the slab allocator in order to provide memory
  297. * to kmalloc and kmem_cache_alloc. In this case, the management of the
  298. * page, and the fields in 'struct page' are the responsibility of mm/slab.c
  299. * unless a particular usage is carefully commented. (the responsibility of
  300. * freeing the kmalloc memory is the caller's, of course).
  301. *
  302. * A page may be used by anyone else who does a __get_free_page().
  303. * In this case, page_count still tracks the references, and should only
  304. * be used through the normal accessor functions. The top bits of page->flags
  305. * and page->virtual store page management information, but all other fields
  306. * are unused and could be used privately, carefully. The management of this
  307. * page is the responsibility of the one who allocated it, and those who have
  308. * subsequently been given references to it.
  309. *
  310. * The other pages (we may call them "pagecache pages") are completely
  311. * managed by the Linux memory manager: I/O, buffers, swapping etc.
  312. * The following discussion applies only to them.
  313. *
  314. * A pagecache page contains an opaque `private' member, which belongs to the
  315. * page's address_space. Usually, this is the address of a circular list of
  316. * the page's disk buffers. PG_private must be set to tell the VM to call
  317. * into the filesystem to release these pages.
  318. *
  319. * A page may belong to an inode's memory mapping. In this case, page->mapping
  320. * is the pointer to the inode, and page->index is the file offset of the page,
  321. * in units of PAGE_CACHE_SIZE.
  322. *
  323. * If pagecache pages are not associated with an inode, they are said to be
  324. * anonymous pages. These may become associated with the swapcache, and in that
  325. * case PG_swapcache is set, and page->private is an offset into the swapcache.
  326. *
  327. * In either case (swapcache or inode backed), the pagecache itself holds one
  328. * reference to the page. Setting PG_private should also increment the
  329. * refcount. The each user mapping also has a reference to the page.
  330. *
  331. * The pagecache pages are stored in a per-mapping radix tree, which is
  332. * rooted at mapping->page_tree, and indexed by offset.
  333. * Where 2.4 and early 2.6 kernels kept dirty/clean pages in per-address_space
  334. * lists, we instead now tag pages as dirty/writeback in the radix tree.
  335. *
  336. * All pagecache pages may be subject to I/O:
  337. * - inode pages may need to be read from disk,
  338. * - inode pages which have been modified and are MAP_SHARED may need
  339. * to be written back to the inode on disk,
  340. * - anonymous pages (including MAP_PRIVATE file mappings) which have been
  341. * modified may need to be swapped out to swap space and (later) to be read
  342. * back into memory.
  343. */
  344. /*
  345. * The zone field is never updated after free_area_init_core()
  346. * sets it, so none of the operations on it need to be atomic.
  347. */
  348. /*
  349. * page->flags layout:
  350. *
  351. * There are three possibilities for how page->flags get
  352. * laid out. The first is for the normal case, without
  353. * sparsemem. The second is for sparsemem when there is
  354. * plenty of space for node and section. The last is when
  355. * we have run out of space and have to fall back to an
  356. * alternate (slower) way of determining the node.
  357. *
  358. * No sparsemem or sparsemem vmemmap: | NODE | ZONE | ... | FLAGS |
  359. * classic sparse with space for node:| SECTION | NODE | ZONE | ... | FLAGS |
  360. * classic sparse no space for node: | SECTION | ZONE | ... | FLAGS |
  361. */
  362. #if defined(CONFIG_SPARSEMEM) && !defined(CONFIG_SPARSEMEM_VMEMMAP)
  363. #define SECTIONS_WIDTH SECTIONS_SHIFT
  364. #else
  365. #define SECTIONS_WIDTH 0
  366. #endif
  367. #define ZONES_WIDTH ZONES_SHIFT
  368. #if SECTIONS_WIDTH+ZONES_WIDTH+NODES_SHIFT <= BITS_PER_LONG - NR_PAGEFLAGS
  369. #define NODES_WIDTH NODES_SHIFT
  370. #else
  371. #ifdef CONFIG_SPARSEMEM_VMEMMAP
  372. #error "Vmemmap: No space for nodes field in page flags"
  373. #endif
  374. #define NODES_WIDTH 0
  375. #endif
  376. /* Page flags: | [SECTION] | [NODE] | ZONE | ... | FLAGS | */
  377. #define SECTIONS_PGOFF ((sizeof(unsigned long)*8) - SECTIONS_WIDTH)
  378. #define NODES_PGOFF (SECTIONS_PGOFF - NODES_WIDTH)
  379. #define ZONES_PGOFF (NODES_PGOFF - ZONES_WIDTH)
  380. /*
  381. * We are going to use the flags for the page to node mapping if its in
  382. * there. This includes the case where there is no node, so it is implicit.
  383. */
  384. #if !(NODES_WIDTH > 0 || NODES_SHIFT == 0)
  385. #define NODE_NOT_IN_PAGE_FLAGS
  386. #endif
  387. #ifndef PFN_SECTION_SHIFT
  388. #define PFN_SECTION_SHIFT 0
  389. #endif
  390. /*
  391. * Define the bit shifts to access each section. For non-existant
  392. * sections we define the shift as 0; that plus a 0 mask ensures
  393. * the compiler will optimise away reference to them.
  394. */
  395. #define SECTIONS_PGSHIFT (SECTIONS_PGOFF * (SECTIONS_WIDTH != 0))
  396. #define NODES_PGSHIFT (NODES_PGOFF * (NODES_WIDTH != 0))
  397. #define ZONES_PGSHIFT (ZONES_PGOFF * (ZONES_WIDTH != 0))
  398. /* NODE:ZONE or SECTION:ZONE is used to ID a zone for the buddy allcator */
  399. #ifdef NODE_NOT_IN_PAGEFLAGS
  400. #define ZONEID_SHIFT (SECTIONS_SHIFT + ZONES_SHIFT)
  401. #define ZONEID_PGOFF ((SECTIONS_PGOFF < ZONES_PGOFF)? \
  402. SECTIONS_PGOFF : ZONES_PGOFF)
  403. #else
  404. #define ZONEID_SHIFT (NODES_SHIFT + ZONES_SHIFT)
  405. #define ZONEID_PGOFF ((NODES_PGOFF < ZONES_PGOFF)? \
  406. NODES_PGOFF : ZONES_PGOFF)
  407. #endif
  408. #define ZONEID_PGSHIFT (ZONEID_PGOFF * (ZONEID_SHIFT != 0))
  409. #if SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > BITS_PER_LONG - NR_PAGEFLAGS
  410. #error SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > BITS_PER_LONG - NR_PAGEFLAGS
  411. #endif
  412. #define ZONES_MASK ((1UL << ZONES_WIDTH) - 1)
  413. #define NODES_MASK ((1UL << NODES_WIDTH) - 1)
  414. #define SECTIONS_MASK ((1UL << SECTIONS_WIDTH) - 1)
  415. #define ZONEID_MASK ((1UL << ZONEID_SHIFT) - 1)
  416. static inline enum zone_type page_zonenum(struct page *page)
  417. {
  418. return (page->flags >> ZONES_PGSHIFT) & ZONES_MASK;
  419. }
  420. /*
  421. * The identification function is only used by the buddy allocator for
  422. * determining if two pages could be buddies. We are not really
  423. * identifying a zone since we could be using a the section number
  424. * id if we have not node id available in page flags.
  425. * We guarantee only that it will return the same value for two
  426. * combinable pages in a zone.
  427. */
  428. static inline int page_zone_id(struct page *page)
  429. {
  430. return (page->flags >> ZONEID_PGSHIFT) & ZONEID_MASK;
  431. }
  432. static inline int zone_to_nid(struct zone *zone)
  433. {
  434. #ifdef CONFIG_NUMA
  435. return zone->node;
  436. #else
  437. return 0;
  438. #endif
  439. }
  440. #ifdef NODE_NOT_IN_PAGE_FLAGS
  441. extern int page_to_nid(struct page *page);
  442. #else
  443. static inline int page_to_nid(struct page *page)
  444. {
  445. return (page->flags >> NODES_PGSHIFT) & NODES_MASK;
  446. }
  447. #endif
  448. static inline struct zone *page_zone(struct page *page)
  449. {
  450. return &NODE_DATA(page_to_nid(page))->node_zones[page_zonenum(page)];
  451. }
  452. #if defined(CONFIG_SPARSEMEM) && !defined(CONFIG_SPARSEMEM_VMEMMAP)
  453. static inline unsigned long page_to_section(struct page *page)
  454. {
  455. return (page->flags >> SECTIONS_PGSHIFT) & SECTIONS_MASK;
  456. }
  457. #endif
  458. static inline void set_page_zone(struct page *page, enum zone_type zone)
  459. {
  460. page->flags &= ~(ZONES_MASK << ZONES_PGSHIFT);
  461. page->flags |= (zone & ZONES_MASK) << ZONES_PGSHIFT;
  462. }
  463. static inline void set_page_node(struct page *page, unsigned long node)
  464. {
  465. page->flags &= ~(NODES_MASK << NODES_PGSHIFT);
  466. page->flags |= (node & NODES_MASK) << NODES_PGSHIFT;
  467. }
  468. static inline void set_page_section(struct page *page, unsigned long section)
  469. {
  470. page->flags &= ~(SECTIONS_MASK << SECTIONS_PGSHIFT);
  471. page->flags |= (section & SECTIONS_MASK) << SECTIONS_PGSHIFT;
  472. }
  473. static inline void set_page_links(struct page *page, enum zone_type zone,
  474. unsigned long node, unsigned long pfn)
  475. {
  476. set_page_zone(page, zone);
  477. set_page_node(page, node);
  478. set_page_section(page, pfn_to_section_nr(pfn));
  479. }
  480. /*
  481. * If a hint addr is less than mmap_min_addr change hint to be as
  482. * low as possible but still greater than mmap_min_addr
  483. */
  484. static inline unsigned long round_hint_to_min(unsigned long hint)
  485. {
  486. #ifdef CONFIG_SECURITY
  487. hint &= PAGE_MASK;
  488. if (((void *)hint != NULL) &&
  489. (hint < mmap_min_addr))
  490. return PAGE_ALIGN(mmap_min_addr);
  491. #endif
  492. return hint;
  493. }
  494. /*
  495. * Some inline functions in vmstat.h depend on page_zone()
  496. */
  497. #include <linux/vmstat.h>
  498. static __always_inline void *lowmem_page_address(struct page *page)
  499. {
  500. return __va(page_to_pfn(page) << PAGE_SHIFT);
  501. }
  502. #if defined(CONFIG_HIGHMEM) && !defined(WANT_PAGE_VIRTUAL)
  503. #define HASHED_PAGE_VIRTUAL
  504. #endif
  505. #if defined(WANT_PAGE_VIRTUAL)
  506. #define page_address(page) ((page)->virtual)
  507. #define set_page_address(page, address) \
  508. do { \
  509. (page)->virtual = (address); \
  510. } while(0)
  511. #define page_address_init() do { } while(0)
  512. #endif
  513. #if defined(HASHED_PAGE_VIRTUAL)
  514. void *page_address(struct page *page);
  515. void set_page_address(struct page *page, void *virtual);
  516. void page_address_init(void);
  517. #endif
  518. #if !defined(HASHED_PAGE_VIRTUAL) && !defined(WANT_PAGE_VIRTUAL)
  519. #define page_address(page) lowmem_page_address(page)
  520. #define set_page_address(page, address) do { } while(0)
  521. #define page_address_init() do { } while(0)
  522. #endif
  523. /*
  524. * On an anonymous page mapped into a user virtual memory area,
  525. * page->mapping points to its anon_vma, not to a struct address_space;
  526. * with the PAGE_MAPPING_ANON bit set to distinguish it.
  527. *
  528. * Please note that, confusingly, "page_mapping" refers to the inode
  529. * address_space which maps the page from disk; whereas "page_mapped"
  530. * refers to user virtual address space into which the page is mapped.
  531. */
  532. #define PAGE_MAPPING_ANON 1
  533. extern struct address_space swapper_space;
  534. static inline struct address_space *page_mapping(struct page *page)
  535. {
  536. struct address_space *mapping = page->mapping;
  537. VM_BUG_ON(PageSlab(page));
  538. #ifdef CONFIG_SWAP
  539. if (unlikely(PageSwapCache(page)))
  540. mapping = &swapper_space;
  541. else
  542. #endif
  543. if (unlikely((unsigned long)mapping & PAGE_MAPPING_ANON))
  544. mapping = NULL;
  545. return mapping;
  546. }
  547. static inline int PageAnon(struct page *page)
  548. {
  549. return ((unsigned long)page->mapping & PAGE_MAPPING_ANON) != 0;
  550. }
  551. /*
  552. * Return the pagecache index of the passed page. Regular pagecache pages
  553. * use ->index whereas swapcache pages use ->private
  554. */
  555. static inline pgoff_t page_index(struct page *page)
  556. {
  557. if (unlikely(PageSwapCache(page)))
  558. return page_private(page);
  559. return page->index;
  560. }
  561. /*
  562. * The atomic page->_mapcount, like _count, starts from -1:
  563. * so that transitions both from it and to it can be tracked,
  564. * using atomic_inc_and_test and atomic_add_negative(-1).
  565. */
  566. static inline void reset_page_mapcount(struct page *page)
  567. {
  568. atomic_set(&(page)->_mapcount, -1);
  569. }
  570. static inline int page_mapcount(struct page *page)
  571. {
  572. return atomic_read(&(page)->_mapcount) + 1;
  573. }
  574. /*
  575. * Return true if this page is mapped into pagetables.
  576. */
  577. static inline int page_mapped(struct page *page)
  578. {
  579. return atomic_read(&(page)->_mapcount) >= 0;
  580. }
  581. /*
  582. * Error return values for the *_nopfn functions
  583. */
  584. #define NOPFN_SIGBUS ((unsigned long) -1)
  585. #define NOPFN_OOM ((unsigned long) -2)
  586. #define NOPFN_REFAULT ((unsigned long) -3)
  587. /*
  588. * Different kinds of faults, as returned by handle_mm_fault().
  589. * Used to decide whether a process gets delivered SIGBUS or
  590. * just gets major/minor fault counters bumped up.
  591. */
  592. #define VM_FAULT_MINOR 0 /* For backwards compat. Remove me quickly. */
  593. #define VM_FAULT_OOM 0x0001
  594. #define VM_FAULT_SIGBUS 0x0002
  595. #define VM_FAULT_MAJOR 0x0004
  596. #define VM_FAULT_WRITE 0x0008 /* Special case for get_user_pages */
  597. #define VM_FAULT_NOPAGE 0x0100 /* ->fault installed the pte, not return page */
  598. #define VM_FAULT_LOCKED 0x0200 /* ->fault locked the returned page */
  599. #define VM_FAULT_ERROR (VM_FAULT_OOM | VM_FAULT_SIGBUS)
  600. #define offset_in_page(p) ((unsigned long)(p) & ~PAGE_MASK)
  601. extern void show_free_areas(void);
  602. #ifdef CONFIG_SHMEM
  603. int shmem_lock(struct file *file, int lock, struct user_struct *user);
  604. #else
  605. static inline int shmem_lock(struct file *file, int lock,
  606. struct user_struct *user)
  607. {
  608. return 0;
  609. }
  610. #endif
  611. struct file *shmem_file_setup(char *name, loff_t size, unsigned long flags);
  612. int shmem_zero_setup(struct vm_area_struct *);
  613. #ifndef CONFIG_MMU
  614. extern unsigned long shmem_get_unmapped_area(struct file *file,
  615. unsigned long addr,
  616. unsigned long len,
  617. unsigned long pgoff,
  618. unsigned long flags);
  619. #endif
  620. extern int can_do_mlock(void);
  621. extern int user_shm_lock(size_t, struct user_struct *);
  622. extern void user_shm_unlock(size_t, struct user_struct *);
  623. /*
  624. * Parameter block passed down to zap_pte_range in exceptional cases.
  625. */
  626. struct zap_details {
  627. struct vm_area_struct *nonlinear_vma; /* Check page->index if set */
  628. struct address_space *check_mapping; /* Check page->mapping if set */
  629. pgoff_t first_index; /* Lowest page->index to unmap */
  630. pgoff_t last_index; /* Highest page->index to unmap */
  631. spinlock_t *i_mmap_lock; /* For unmap_mapping_range: */
  632. unsigned long truncate_count; /* Compare vm_truncate_count */
  633. };
  634. struct page *vm_normal_page(struct vm_area_struct *vma, unsigned long addr,
  635. pte_t pte);
  636. unsigned long zap_page_range(struct vm_area_struct *vma, unsigned long address,
  637. unsigned long size, struct zap_details *);
  638. unsigned long unmap_vmas(struct mmu_gather **tlb,
  639. struct vm_area_struct *start_vma, unsigned long start_addr,
  640. unsigned long end_addr, unsigned long *nr_accounted,
  641. struct zap_details *);
  642. /**
  643. * mm_walk - callbacks for walk_page_range
  644. * @pgd_entry: if set, called for each non-empty PGD (top-level) entry
  645. * @pud_entry: if set, called for each non-empty PUD (2nd-level) entry
  646. * @pmd_entry: if set, called for each non-empty PMD (3rd-level) entry
  647. * @pte_entry: if set, called for each non-empty PTE (4th-level) entry
  648. * @pte_hole: if set, called for each hole at all levels
  649. *
  650. * (see walk_page_range for more details)
  651. */
  652. struct mm_walk {
  653. int (*pgd_entry)(pgd_t *, unsigned long, unsigned long, struct mm_walk *);
  654. int (*pud_entry)(pud_t *, unsigned long, unsigned long, struct mm_walk *);
  655. int (*pmd_entry)(pmd_t *, unsigned long, unsigned long, struct mm_walk *);
  656. int (*pte_entry)(pte_t *, unsigned long, unsigned long, struct mm_walk *);
  657. int (*pte_hole)(unsigned long, unsigned long, struct mm_walk *);
  658. struct mm_struct *mm;
  659. void *private;
  660. };
  661. int walk_page_range(unsigned long addr, unsigned long end,
  662. struct mm_walk *walk);
  663. void free_pgd_range(struct mmu_gather **tlb, unsigned long addr,
  664. unsigned long end, unsigned long floor, unsigned long ceiling);
  665. void free_pgtables(struct mmu_gather **tlb, struct vm_area_struct *start_vma,
  666. unsigned long floor, unsigned long ceiling);
  667. int copy_page_range(struct mm_struct *dst, struct mm_struct *src,
  668. struct vm_area_struct *vma);
  669. void unmap_mapping_range(struct address_space *mapping,
  670. loff_t const holebegin, loff_t const holelen, int even_cows);
  671. static inline void unmap_shared_mapping_range(struct address_space *mapping,
  672. loff_t const holebegin, loff_t const holelen)
  673. {
  674. unmap_mapping_range(mapping, holebegin, holelen, 0);
  675. }
  676. extern int vmtruncate(struct inode * inode, loff_t offset);
  677. extern int vmtruncate_range(struct inode * inode, loff_t offset, loff_t end);
  678. #ifdef CONFIG_MMU
  679. extern int handle_mm_fault(struct mm_struct *mm, struct vm_area_struct *vma,
  680. unsigned long address, int write_access);
  681. #else
  682. static inline int handle_mm_fault(struct mm_struct *mm,
  683. struct vm_area_struct *vma, unsigned long address,
  684. int write_access)
  685. {
  686. /* should never happen if there's no MMU */
  687. BUG();
  688. return VM_FAULT_SIGBUS;
  689. }
  690. #endif
  691. extern int make_pages_present(unsigned long addr, unsigned long end);
  692. extern int access_process_vm(struct task_struct *tsk, unsigned long addr, void *buf, int len, int write);
  693. int get_user_pages(struct task_struct *tsk, struct mm_struct *mm, unsigned long start,
  694. int len, int write, int force, struct page **pages, struct vm_area_struct **vmas);
  695. void print_bad_pte(struct vm_area_struct *, pte_t, unsigned long);
  696. extern int try_to_release_page(struct page * page, gfp_t gfp_mask);
  697. extern void do_invalidatepage(struct page *page, unsigned long offset);
  698. int __set_page_dirty_nobuffers(struct page *page);
  699. int __set_page_dirty_no_writeback(struct page *page);
  700. int redirty_page_for_writepage(struct writeback_control *wbc,
  701. struct page *page);
  702. int set_page_dirty(struct page *page);
  703. int set_page_dirty_lock(struct page *page);
  704. int clear_page_dirty_for_io(struct page *page);
  705. extern unsigned long move_page_tables(struct vm_area_struct *vma,
  706. unsigned long old_addr, struct vm_area_struct *new_vma,
  707. unsigned long new_addr, unsigned long len);
  708. extern unsigned long do_mremap(unsigned long addr,
  709. unsigned long old_len, unsigned long new_len,
  710. unsigned long flags, unsigned long new_addr);
  711. extern int mprotect_fixup(struct vm_area_struct *vma,
  712. struct vm_area_struct **pprev, unsigned long start,
  713. unsigned long end, unsigned long newflags);
  714. /*
  715. * A callback you can register to apply pressure to ageable caches.
  716. *
  717. * 'shrink' is passed a count 'nr_to_scan' and a 'gfpmask'. It should
  718. * look through the least-recently-used 'nr_to_scan' entries and
  719. * attempt to free them up. It should return the number of objects
  720. * which remain in the cache. If it returns -1, it means it cannot do
  721. * any scanning at this time (eg. there is a risk of deadlock).
  722. *
  723. * The 'gfpmask' refers to the allocation we are currently trying to
  724. * fulfil.
  725. *
  726. * Note that 'shrink' will be passed nr_to_scan == 0 when the VM is
  727. * querying the cache size, so a fastpath for that case is appropriate.
  728. */
  729. struct shrinker {
  730. int (*shrink)(int nr_to_scan, gfp_t gfp_mask);
  731. int seeks; /* seeks to recreate an obj */
  732. /* These are for internal use */
  733. struct list_head list;
  734. long nr; /* objs pending delete */
  735. };
  736. #define DEFAULT_SEEKS 2 /* A good number if you don't know better. */
  737. extern void register_shrinker(struct shrinker *);
  738. extern void unregister_shrinker(struct shrinker *);
  739. int vma_wants_writenotify(struct vm_area_struct *vma);
  740. extern pte_t *get_locked_pte(struct mm_struct *mm, unsigned long addr, spinlock_t **ptl);
  741. #ifdef __PAGETABLE_PUD_FOLDED
  742. static inline int __pud_alloc(struct mm_struct *mm, pgd_t *pgd,
  743. unsigned long address)
  744. {
  745. return 0;
  746. }
  747. #else
  748. int __pud_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address);
  749. #endif
  750. #ifdef __PAGETABLE_PMD_FOLDED
  751. static inline int __pmd_alloc(struct mm_struct *mm, pud_t *pud,
  752. unsigned long address)
  753. {
  754. return 0;
  755. }
  756. #else
  757. int __pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address);
  758. #endif
  759. int __pte_alloc(struct mm_struct *mm, pmd_t *pmd, unsigned long address);
  760. int __pte_alloc_kernel(pmd_t *pmd, unsigned long address);
  761. /*
  762. * The following ifdef needed to get the 4level-fixup.h header to work.
  763. * Remove it when 4level-fixup.h has been removed.
  764. */
  765. #if defined(CONFIG_MMU) && !defined(__ARCH_HAS_4LEVEL_HACK)
  766. static inline pud_t *pud_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address)
  767. {
  768. return (unlikely(pgd_none(*pgd)) && __pud_alloc(mm, pgd, address))?
  769. NULL: pud_offset(pgd, address);
  770. }
  771. static inline pmd_t *pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address)
  772. {
  773. return (unlikely(pud_none(*pud)) && __pmd_alloc(mm, pud, address))?
  774. NULL: pmd_offset(pud, address);
  775. }
  776. #endif /* CONFIG_MMU && !__ARCH_HAS_4LEVEL_HACK */
  777. #if NR_CPUS >= CONFIG_SPLIT_PTLOCK_CPUS
  778. /*
  779. * We tuck a spinlock to guard each pagetable page into its struct page,
  780. * at page->private, with BUILD_BUG_ON to make sure that this will not
  781. * overflow into the next struct page (as it might with DEBUG_SPINLOCK).
  782. * When freeing, reset page->mapping so free_pages_check won't complain.
  783. */
  784. #define __pte_lockptr(page) &((page)->ptl)
  785. #define pte_lock_init(_page) do { \
  786. spin_lock_init(__pte_lockptr(_page)); \
  787. } while (0)
  788. #define pte_lock_deinit(page) ((page)->mapping = NULL)
  789. #define pte_lockptr(mm, pmd) ({(void)(mm); __pte_lockptr(pmd_page(*(pmd)));})
  790. #else
  791. /*
  792. * We use mm->page_table_lock to guard all pagetable pages of the mm.
  793. */
  794. #define pte_lock_init(page) do {} while (0)
  795. #define pte_lock_deinit(page) do {} while (0)
  796. #define pte_lockptr(mm, pmd) ({(void)(pmd); &(mm)->page_table_lock;})
  797. #endif /* NR_CPUS < CONFIG_SPLIT_PTLOCK_CPUS */
  798. static inline void pgtable_page_ctor(struct page *page)
  799. {
  800. pte_lock_init(page);
  801. inc_zone_page_state(page, NR_PAGETABLE);
  802. }
  803. static inline void pgtable_page_dtor(struct page *page)
  804. {
  805. pte_lock_deinit(page);
  806. dec_zone_page_state(page, NR_PAGETABLE);
  807. }
  808. #define pte_offset_map_lock(mm, pmd, address, ptlp) \
  809. ({ \
  810. spinlock_t *__ptl = pte_lockptr(mm, pmd); \
  811. pte_t *__pte = pte_offset_map(pmd, address); \
  812. *(ptlp) = __ptl; \
  813. spin_lock(__ptl); \
  814. __pte; \
  815. })
  816. #define pte_unmap_unlock(pte, ptl) do { \
  817. spin_unlock(ptl); \
  818. pte_unmap(pte); \
  819. } while (0)
  820. #define pte_alloc_map(mm, pmd, address) \
  821. ((unlikely(!pmd_present(*(pmd))) && __pte_alloc(mm, pmd, address))? \
  822. NULL: pte_offset_map(pmd, address))
  823. #define pte_alloc_map_lock(mm, pmd, address, ptlp) \
  824. ((unlikely(!pmd_present(*(pmd))) && __pte_alloc(mm, pmd, address))? \
  825. NULL: pte_offset_map_lock(mm, pmd, address, ptlp))
  826. #define pte_alloc_kernel(pmd, address) \
  827. ((unlikely(!pmd_present(*(pmd))) && __pte_alloc_kernel(pmd, address))? \
  828. NULL: pte_offset_kernel(pmd, address))
  829. extern void free_area_init(unsigned long * zones_size);
  830. extern void free_area_init_node(int nid, pg_data_t *pgdat,
  831. unsigned long * zones_size, unsigned long zone_start_pfn,
  832. unsigned long *zholes_size);
  833. #ifdef CONFIG_ARCH_POPULATES_NODE_MAP
  834. /*
  835. * With CONFIG_ARCH_POPULATES_NODE_MAP set, an architecture may initialise its
  836. * zones, allocate the backing mem_map and account for memory holes in a more
  837. * architecture independent manner. This is a substitute for creating the
  838. * zone_sizes[] and zholes_size[] arrays and passing them to
  839. * free_area_init_node()
  840. *
  841. * An architecture is expected to register range of page frames backed by
  842. * physical memory with add_active_range() before calling
  843. * free_area_init_nodes() passing in the PFN each zone ends at. At a basic
  844. * usage, an architecture is expected to do something like
  845. *
  846. * unsigned long max_zone_pfns[MAX_NR_ZONES] = {max_dma, max_normal_pfn,
  847. * max_highmem_pfn};
  848. * for_each_valid_physical_page_range()
  849. * add_active_range(node_id, start_pfn, end_pfn)
  850. * free_area_init_nodes(max_zone_pfns);
  851. *
  852. * If the architecture guarantees that there are no holes in the ranges
  853. * registered with add_active_range(), free_bootmem_active_regions()
  854. * will call free_bootmem_node() for each registered physical page range.
  855. * Similarly sparse_memory_present_with_active_regions() calls
  856. * memory_present() for each range when SPARSEMEM is enabled.
  857. *
  858. * See mm/page_alloc.c for more information on each function exposed by
  859. * CONFIG_ARCH_POPULATES_NODE_MAP
  860. */
  861. extern void free_area_init_nodes(unsigned long *max_zone_pfn);
  862. extern void add_active_range(unsigned int nid, unsigned long start_pfn,
  863. unsigned long end_pfn);
  864. extern void remove_active_range(unsigned int nid, unsigned long start_pfn,
  865. unsigned long end_pfn);
  866. extern void push_node_boundaries(unsigned int nid, unsigned long start_pfn,
  867. unsigned long end_pfn);
  868. extern void remove_all_active_ranges(void);
  869. extern unsigned long absent_pages_in_range(unsigned long start_pfn,
  870. unsigned long end_pfn);
  871. extern void get_pfn_range_for_nid(unsigned int nid,
  872. unsigned long *start_pfn, unsigned long *end_pfn);
  873. extern unsigned long find_min_pfn_with_active_regions(void);
  874. extern unsigned long find_max_pfn_with_active_regions(void);
  875. extern void free_bootmem_with_active_regions(int nid,
  876. unsigned long max_low_pfn);
  877. typedef void (*work_fn_t)(unsigned long, unsigned long, void *);
  878. extern void work_with_active_regions(int nid, work_fn_t work_fn, void *data);
  879. extern void sparse_memory_present_with_active_regions(int nid);
  880. #ifndef CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID
  881. extern int early_pfn_to_nid(unsigned long pfn);
  882. #endif /* CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID */
  883. #endif /* CONFIG_ARCH_POPULATES_NODE_MAP */
  884. extern void set_dma_reserve(unsigned long new_dma_reserve);
  885. extern void memmap_init_zone(unsigned long, int, unsigned long,
  886. unsigned long, enum memmap_context);
  887. extern void setup_per_zone_pages_min(void);
  888. extern void mem_init(void);
  889. extern void show_mem(void);
  890. extern void si_meminfo(struct sysinfo * val);
  891. extern void si_meminfo_node(struct sysinfo *val, int nid);
  892. #ifdef CONFIG_NUMA
  893. extern void setup_per_cpu_pageset(void);
  894. #else
  895. static inline void setup_per_cpu_pageset(void) {}
  896. #endif
  897. /* prio_tree.c */
  898. void vma_prio_tree_add(struct vm_area_struct *, struct vm_area_struct *old);
  899. void vma_prio_tree_insert(struct vm_area_struct *, struct prio_tree_root *);
  900. void vma_prio_tree_remove(struct vm_area_struct *, struct prio_tree_root *);
  901. struct vm_area_struct *vma_prio_tree_next(struct vm_area_struct *vma,
  902. struct prio_tree_iter *iter);
  903. #define vma_prio_tree_foreach(vma, iter, root, begin, end) \
  904. for (prio_tree_iter_init(iter, root, begin, end), vma = NULL; \
  905. (vma = vma_prio_tree_next(vma, iter)); )
  906. static inline void vma_nonlinear_insert(struct vm_area_struct *vma,
  907. struct list_head *list)
  908. {
  909. vma->shared.vm_set.parent = NULL;
  910. list_add_tail(&vma->shared.vm_set.list, list);
  911. }
  912. /* mmap.c */
  913. extern int __vm_enough_memory(struct mm_struct *mm, long pages, int cap_sys_admin);
  914. extern void vma_adjust(struct vm_area_struct *vma, unsigned long start,
  915. unsigned long end, pgoff_t pgoff, struct vm_area_struct *insert);
  916. extern struct vm_area_struct *vma_merge(struct mm_struct *,
  917. struct vm_area_struct *prev, unsigned long addr, unsigned long end,
  918. unsigned long vm_flags, struct anon_vma *, struct file *, pgoff_t,
  919. struct mempolicy *);
  920. extern struct anon_vma *find_mergeable_anon_vma(struct vm_area_struct *);
  921. extern int split_vma(struct mm_struct *,
  922. struct vm_area_struct *, unsigned long addr, int new_below);
  923. extern int insert_vm_struct(struct mm_struct *, struct vm_area_struct *);
  924. extern void __vma_link_rb(struct mm_struct *, struct vm_area_struct *,
  925. struct rb_node **, struct rb_node *);
  926. extern void unlink_file_vma(struct vm_area_struct *);
  927. extern struct vm_area_struct *copy_vma(struct vm_area_struct **,
  928. unsigned long addr, unsigned long len, pgoff_t pgoff);
  929. extern void exit_mmap(struct mm_struct *);
  930. #ifdef CONFIG_PROC_FS
  931. /* From fs/proc/base.c. callers must _not_ hold the mm's exe_file_lock */
  932. extern void added_exe_file_vma(struct mm_struct *mm);
  933. extern void removed_exe_file_vma(struct mm_struct *mm);
  934. #else
  935. static inline void added_exe_file_vma(struct mm_struct *mm)
  936. {}
  937. static inline void removed_exe_file_vma(struct mm_struct *mm)
  938. {}
  939. #endif /* CONFIG_PROC_FS */
  940. extern int may_expand_vm(struct mm_struct *mm, unsigned long npages);
  941. extern int install_special_mapping(struct mm_struct *mm,
  942. unsigned long addr, unsigned long len,
  943. unsigned long flags, struct page **pages);
  944. extern unsigned long get_unmapped_area(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
  945. extern unsigned long do_mmap_pgoff(struct file *file, unsigned long addr,
  946. unsigned long len, unsigned long prot,
  947. unsigned long flag, unsigned long pgoff);
  948. extern unsigned long mmap_region(struct file *file, unsigned long addr,
  949. unsigned long len, unsigned long flags,
  950. unsigned int vm_flags, unsigned long pgoff,
  951. int accountable);
  952. static inline unsigned long do_mmap(struct file *file, unsigned long addr,
  953. unsigned long len, unsigned long prot,
  954. unsigned long flag, unsigned long offset)
  955. {
  956. unsigned long ret = -EINVAL;
  957. if ((offset + PAGE_ALIGN(len)) < offset)
  958. goto out;
  959. if (!(offset & ~PAGE_MASK))
  960. ret = do_mmap_pgoff(file, addr, len, prot, flag, offset >> PAGE_SHIFT);
  961. out:
  962. return ret;
  963. }
  964. extern int do_munmap(struct mm_struct *, unsigned long, size_t);
  965. extern unsigned long do_brk(unsigned long, unsigned long);
  966. /* filemap.c */
  967. extern unsigned long page_unuse(struct page *);
  968. extern void truncate_inode_pages(struct address_space *, loff_t);
  969. extern void truncate_inode_pages_range(struct address_space *,
  970. loff_t lstart, loff_t lend);
  971. /* generic vm_area_ops exported for stackable file systems */
  972. extern int filemap_fault(struct vm_area_struct *, struct vm_fault *);
  973. /* mm/page-writeback.c */
  974. int write_one_page(struct page *page, int wait);
  975. /* readahead.c */
  976. #define VM_MAX_READAHEAD 128 /* kbytes */
  977. #define VM_MIN_READAHEAD 16 /* kbytes (includes current page) */
  978. int do_page_cache_readahead(struct address_space *mapping, struct file *filp,
  979. pgoff_t offset, unsigned long nr_to_read);
  980. int force_page_cache_readahead(struct address_space *mapping, struct file *filp,
  981. pgoff_t offset, unsigned long nr_to_read);
  982. void page_cache_sync_readahead(struct address_space *mapping,
  983. struct file_ra_state *ra,
  984. struct file *filp,
  985. pgoff_t offset,
  986. unsigned long size);
  987. void page_cache_async_readahead(struct address_space *mapping,
  988. struct file_ra_state *ra,
  989. struct file *filp,
  990. struct page *pg,
  991. pgoff_t offset,
  992. unsigned long size);
  993. unsigned long max_sane_readahead(unsigned long nr);
  994. /* Do stack extension */
  995. extern int expand_stack(struct vm_area_struct *vma, unsigned long address);
  996. #ifdef CONFIG_IA64
  997. extern int expand_upwards(struct vm_area_struct *vma, unsigned long address);
  998. #endif
  999. extern int expand_stack_downwards(struct vm_area_struct *vma,
  1000. unsigned long address);
  1001. /* Look up the first VMA which satisfies addr < vm_end, NULL if none. */
  1002. extern struct vm_area_struct * find_vma(struct mm_struct * mm, unsigned long addr);
  1003. extern struct vm_area_struct * find_vma_prev(struct mm_struct * mm, unsigned long addr,
  1004. struct vm_area_struct **pprev);
  1005. /* Look up the first VMA which intersects the interval start_addr..end_addr-1,
  1006. NULL if none. Assume start_addr < end_addr. */
  1007. static inline struct vm_area_struct * find_vma_intersection(struct mm_struct * mm, unsigned long start_addr, unsigned long end_addr)
  1008. {
  1009. struct vm_area_struct * vma = find_vma(mm,start_addr);
  1010. if (vma && end_addr <= vma->vm_start)
  1011. vma = NULL;
  1012. return vma;
  1013. }
  1014. static inline unsigned long vma_pages(struct vm_area_struct *vma)
  1015. {
  1016. return (vma->vm_end - vma->vm_start) >> PAGE_SHIFT;
  1017. }
  1018. pgprot_t vm_get_page_prot(unsigned long vm_flags);
  1019. struct vm_area_struct *find_extend_vma(struct mm_struct *, unsigned long addr);
  1020. int remap_pfn_range(struct vm_area_struct *, unsigned long addr,
  1021. unsigned long pfn, unsigned long size, pgprot_t);
  1022. int vm_insert_page(struct vm_area_struct *, unsigned long addr, struct page *);
  1023. int vm_insert_pfn(struct vm_area_struct *vma, unsigned long addr,
  1024. unsigned long pfn);
  1025. int vm_insert_mixed(struct vm_area_struct *vma, unsigned long addr,
  1026. unsigned long pfn);
  1027. struct page *follow_page(struct vm_area_struct *, unsigned long address,
  1028. unsigned int foll_flags);
  1029. #define FOLL_WRITE 0x01 /* check pte is writable */
  1030. #define FOLL_TOUCH 0x02 /* mark page accessed */
  1031. #define FOLL_GET 0x04 /* do get_page on page */
  1032. #define FOLL_ANON 0x08 /* give ZERO_PAGE if no pgtable */
  1033. typedef int (*pte_fn_t)(pte_t *pte, pgtable_t token, unsigned long addr,
  1034. void *data);
  1035. extern int apply_to_page_range(struct mm_struct *mm, unsigned long address,
  1036. unsigned long size, pte_fn_t fn, void *data);
  1037. #ifdef CONFIG_PROC_FS
  1038. void vm_stat_account(struct mm_struct *, unsigned long, struct file *, long);
  1039. #else
  1040. static inline void vm_stat_account(struct mm_struct *mm,
  1041. unsigned long flags, struct file *file, long pages)
  1042. {
  1043. }
  1044. #endif /* CONFIG_PROC_FS */
  1045. #ifdef CONFIG_DEBUG_PAGEALLOC
  1046. extern int debug_pagealloc_enabled;
  1047. extern void kernel_map_pages(struct page *page, int numpages, int enable);
  1048. static inline void enable_debug_pagealloc(void)
  1049. {
  1050. debug_pagealloc_enabled = 1;
  1051. }
  1052. #ifdef CONFIG_HIBERNATION
  1053. extern bool kernel_page_present(struct page *page);
  1054. #endif /* CONFIG_HIBERNATION */
  1055. #else
  1056. static inline void
  1057. kernel_map_pages(struct page *page, int numpages, int enable) {}
  1058. static inline void enable_debug_pagealloc(void)
  1059. {
  1060. }
  1061. #ifdef CONFIG_HIBERNATION
  1062. static inline bool kernel_page_present(struct page *page) { return true; }
  1063. #endif /* CONFIG_HIBERNATION */
  1064. #endif
  1065. extern struct vm_area_struct *get_gate_vma(struct task_struct *tsk);
  1066. #ifdef __HAVE_ARCH_GATE_AREA
  1067. int in_gate_area_no_task(unsigned long addr);
  1068. int in_gate_area(struct task_struct *task, unsigned long addr);
  1069. #else
  1070. int in_gate_area_no_task(unsigned long addr);
  1071. #define in_gate_area(task, addr) ({(void)task; in_gate_area_no_task(addr);})
  1072. #endif /* __HAVE_ARCH_GATE_AREA */
  1073. int drop_caches_sysctl_handler(struct ctl_table *, int, struct file *,
  1074. void __user *, size_t *, loff_t *);
  1075. unsigned long shrink_slab(unsigned long scanned, gfp_t gfp_mask,
  1076. unsigned long lru_pages);
  1077. #ifndef CONFIG_MMU
  1078. #define randomize_va_space 0
  1079. #else
  1080. extern int randomize_va_space;
  1081. #endif
  1082. const char * arch_vma_name(struct vm_area_struct *vma);
  1083. void print_vma_addr(char *prefix, unsigned long rip);
  1084. struct page *sparse_mem_map_populate(unsigned long pnum, int nid);
  1085. pgd_t *vmemmap_pgd_populate(unsigned long addr, int node);
  1086. pud_t *vmemmap_pud_populate(pgd_t *pgd, unsigned long addr, int node);
  1087. pmd_t *vmemmap_pmd_populate(pud_t *pud, unsigned long addr, int node);
  1088. pte_t *vmemmap_pte_populate(pmd_t *pmd, unsigned long addr, int node);
  1089. void *vmemmap_alloc_block(unsigned long size, int node);
  1090. void vmemmap_verify(pte_t *, int, unsigned long, unsigned long);
  1091. int vmemmap_populate_basepages(struct page *start_page,
  1092. unsigned long pages, int node);
  1093. int vmemmap_populate(struct page *start_page, unsigned long pages, int node);
  1094. void vmemmap_populate_print_last(void);
  1095. #endif /* __KERNEL__ */
  1096. #endif /* _LINUX_MM_H */