mm.h 46 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662663664665666667668669670671672673674675676677678679680681682683684685686687688689690691692693694695696697698699700701702703704705706707708709710711712713714715716717718719720721722723724725726727728729730731732733734735736737738739740741742743744745746747748749750751752753754755756757758759760761762763764765766767768769770771772773774775776777778779780781782783784785786787788789790791792793794795796797798799800801802803804805806807808809810811812813814815816817818819820821822823824825826827828829830831832833834835836837838839840841842843844845846847848849850851852853854855856857858859860861862863864865866867868869870871872873874875876877878879880881882883884885886887888889890891892893894895896897898899900901902903904905906907908909910911912913914915916917918919920921922923924925926927928929930931932933934935936937938939940941942943944945946947948949950951952953954955956957958959960961962963964965966967968969970971972973974975976977978979980981982983984985986987988989990991992993994995996997998999100010011002100310041005100610071008100910101011101210131014101510161017101810191020102110221023102410251026102710281029103010311032103310341035103610371038103910401041104210431044104510461047104810491050105110521053105410551056105710581059106010611062106310641065106610671068106910701071107210731074107510761077107810791080108110821083108410851086108710881089109010911092109310941095109610971098109911001101110211031104110511061107110811091110111111121113111411151116111711181119112011211122112311241125112611271128112911301131113211331134113511361137113811391140114111421143114411451146114711481149115011511152115311541155115611571158115911601161116211631164116511661167116811691170117111721173117411751176117711781179118011811182118311841185118611871188118911901191119211931194119511961197119811991200120112021203120412051206120712081209121012111212121312141215121612171218121912201221122212231224122512261227122812291230123112321233123412351236123712381239124012411242124312441245124612471248124912501251125212531254125512561257125812591260126112621263126412651266126712681269127012711272127312741275127612771278127912801281128212831284128512861287128812891290129112921293129412951296129712981299130013011302130313041305130613071308130913101311131213131314131513161317131813191320132113221323132413251326132713281329133013311332133313341335133613371338133913401341
  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. struct rlimit;
  18. #ifndef CONFIG_DISCONTIGMEM /* Don't use mapnrs, do it properly */
  19. extern unsigned long max_mapnr;
  20. #endif
  21. extern unsigned long num_physpages;
  22. extern unsigned long totalram_pages;
  23. extern void * high_memory;
  24. extern int page_cluster;
  25. #ifdef CONFIG_SYSCTL
  26. extern int sysctl_legacy_va_layout;
  27. #else
  28. #define sysctl_legacy_va_layout 0
  29. #endif
  30. #include <asm/page.h>
  31. #include <asm/pgtable.h>
  32. #include <asm/processor.h>
  33. #define nth_page(page,n) pfn_to_page(page_to_pfn((page)) + (n))
  34. /* to align the pointer to the (next) page boundary */
  35. #define PAGE_ALIGN(addr) ALIGN(addr, PAGE_SIZE)
  36. /*
  37. * Linux kernel virtual memory manager primitives.
  38. * The idea being to have a "virtual" mm in the same way
  39. * we have a virtual fs - giving a cleaner interface to the
  40. * mm details, and allowing different kinds of memory mappings
  41. * (from shared memory to executable loading to arbitrary
  42. * mmap() functions).
  43. */
  44. extern struct kmem_cache *vm_area_cachep;
  45. #ifndef CONFIG_MMU
  46. extern struct rb_root nommu_region_tree;
  47. extern struct rw_semaphore nommu_region_sem;
  48. extern unsigned int kobjsize(const void *objp);
  49. #endif
  50. /*
  51. * vm_flags in vm_area_struct, see mm_types.h.
  52. */
  53. #define VM_READ 0x00000001 /* currently active flags */
  54. #define VM_WRITE 0x00000002
  55. #define VM_EXEC 0x00000004
  56. #define VM_SHARED 0x00000008
  57. /* mprotect() hardcodes VM_MAYREAD >> 4 == VM_READ, and so for r/w/x bits. */
  58. #define VM_MAYREAD 0x00000010 /* limits for mprotect() etc */
  59. #define VM_MAYWRITE 0x00000020
  60. #define VM_MAYEXEC 0x00000040
  61. #define VM_MAYSHARE 0x00000080
  62. #define VM_GROWSDOWN 0x00000100 /* general info on the segment */
  63. #define VM_GROWSUP 0x00000200
  64. #define VM_PFNMAP 0x00000400 /* Page-ranges managed without "struct page", just pure PFN */
  65. #define VM_DENYWRITE 0x00000800 /* ETXTBSY on write attempts.. */
  66. #define VM_EXECUTABLE 0x00001000
  67. #define VM_LOCKED 0x00002000
  68. #define VM_IO 0x00004000 /* Memory mapped I/O or similar */
  69. /* Used by sys_madvise() */
  70. #define VM_SEQ_READ 0x00008000 /* App will access data sequentially */
  71. #define VM_RAND_READ 0x00010000 /* App will not benefit from clustered reads */
  72. #define VM_DONTCOPY 0x00020000 /* Do not copy this vma on fork */
  73. #define VM_DONTEXPAND 0x00040000 /* Cannot expand with mremap() */
  74. #define VM_RESERVED 0x00080000 /* Count as reserved_vm like IO */
  75. #define VM_ACCOUNT 0x00100000 /* Is a VM accounted object */
  76. #define VM_NORESERVE 0x00200000 /* should the VM suppress accounting */
  77. #define VM_HUGETLB 0x00400000 /* Huge TLB Page VM */
  78. #define VM_NONLINEAR 0x00800000 /* Is non-linear (remap_file_pages) */
  79. #define VM_MAPPED_COPY 0x01000000 /* T if mapped copy of data (nommu mmap) */
  80. #define VM_INSERTPAGE 0x02000000 /* The vma has had "vm_insert_page()" done on it */
  81. #define VM_ALWAYSDUMP 0x04000000 /* Always include in core dumps */
  82. #define VM_CAN_NONLINEAR 0x08000000 /* Has ->fault & does nonlinear pages */
  83. #define VM_MIXEDMAP 0x10000000 /* Can contain "struct page" and pure PFN pages */
  84. #define VM_SAO 0x20000000 /* Strong Access Ordering (powerpc) */
  85. #define VM_PFN_AT_MMAP 0x40000000 /* PFNMAP vma that is fully mapped at mmap time */
  86. #define VM_MERGEABLE 0x80000000 /* KSM may merge identical pages */
  87. #ifndef VM_STACK_DEFAULT_FLAGS /* arch can override this */
  88. #define VM_STACK_DEFAULT_FLAGS VM_DATA_DEFAULT_FLAGS
  89. #endif
  90. #ifdef CONFIG_STACK_GROWSUP
  91. #define VM_STACK_FLAGS (VM_GROWSUP | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT)
  92. #else
  93. #define VM_STACK_FLAGS (VM_GROWSDOWN | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT)
  94. #endif
  95. #define VM_READHINTMASK (VM_SEQ_READ | VM_RAND_READ)
  96. #define VM_ClearReadHint(v) (v)->vm_flags &= ~VM_READHINTMASK
  97. #define VM_NormalReadHint(v) (!((v)->vm_flags & VM_READHINTMASK))
  98. #define VM_SequentialReadHint(v) ((v)->vm_flags & VM_SEQ_READ)
  99. #define VM_RandomReadHint(v) ((v)->vm_flags & VM_RAND_READ)
  100. /*
  101. * special vmas that are non-mergable, non-mlock()able
  102. */
  103. #define VM_SPECIAL (VM_IO | VM_DONTEXPAND | VM_RESERVED | VM_PFNMAP)
  104. /*
  105. * mapping from the currently active vm_flags protection bits (the
  106. * low four bits) to a page protection mask..
  107. */
  108. extern pgprot_t protection_map[16];
  109. #define FAULT_FLAG_WRITE 0x01 /* Fault was a write access */
  110. #define FAULT_FLAG_NONLINEAR 0x02 /* Fault was via a nonlinear mapping */
  111. #define FAULT_FLAG_MKWRITE 0x04 /* Fault was mkwrite of existing pte */
  112. /*
  113. * This interface is used by x86 PAT code to identify a pfn mapping that is
  114. * linear over entire vma. This is to optimize PAT code that deals with
  115. * marking the physical region with a particular prot. This is not for generic
  116. * mm use. Note also that this check will not work if the pfn mapping is
  117. * linear for a vma starting at physical address 0. In which case PAT code
  118. * falls back to slow path of reserving physical range page by page.
  119. */
  120. static inline int is_linear_pfn_mapping(struct vm_area_struct *vma)
  121. {
  122. return (vma->vm_flags & VM_PFN_AT_MMAP);
  123. }
  124. static inline int is_pfn_mapping(struct vm_area_struct *vma)
  125. {
  126. return (vma->vm_flags & VM_PFNMAP);
  127. }
  128. /*
  129. * vm_fault is filled by the the pagefault handler and passed to the vma's
  130. * ->fault function. The vma's ->fault is responsible for returning a bitmask
  131. * of VM_FAULT_xxx flags that give details about how the fault was handled.
  132. *
  133. * pgoff should be used in favour of virtual_address, if possible. If pgoff
  134. * is used, one may set VM_CAN_NONLINEAR in the vma->vm_flags to get nonlinear
  135. * mapping support.
  136. */
  137. struct vm_fault {
  138. unsigned int flags; /* FAULT_FLAG_xxx flags */
  139. pgoff_t pgoff; /* Logical page offset based on vma */
  140. void __user *virtual_address; /* Faulting virtual address */
  141. struct page *page; /* ->fault handlers should return a
  142. * page here, unless VM_FAULT_NOPAGE
  143. * is set (which is also implied by
  144. * VM_FAULT_ERROR).
  145. */
  146. };
  147. /*
  148. * These are the virtual MM functions - opening of an area, closing and
  149. * unmapping it (needed to keep files on disk up-to-date etc), pointer
  150. * to the functions called when a no-page or a wp-page exception occurs.
  151. */
  152. struct vm_operations_struct {
  153. void (*open)(struct vm_area_struct * area);
  154. void (*close)(struct vm_area_struct * area);
  155. int (*fault)(struct vm_area_struct *vma, struct vm_fault *vmf);
  156. /* notification that a previously read-only page is about to become
  157. * writable, if an error is returned it will cause a SIGBUS */
  158. int (*page_mkwrite)(struct vm_area_struct *vma, struct vm_fault *vmf);
  159. /* called by access_process_vm when get_user_pages() fails, typically
  160. * for use by special VMAs that can switch between memory and hardware
  161. */
  162. int (*access)(struct vm_area_struct *vma, unsigned long addr,
  163. void *buf, int len, int write);
  164. #ifdef CONFIG_NUMA
  165. /*
  166. * set_policy() op must add a reference to any non-NULL @new mempolicy
  167. * to hold the policy upon return. Caller should pass NULL @new to
  168. * remove a policy and fall back to surrounding context--i.e. do not
  169. * install a MPOL_DEFAULT policy, nor the task or system default
  170. * mempolicy.
  171. */
  172. int (*set_policy)(struct vm_area_struct *vma, struct mempolicy *new);
  173. /*
  174. * get_policy() op must add reference [mpol_get()] to any policy at
  175. * (vma,addr) marked as MPOL_SHARED. The shared policy infrastructure
  176. * in mm/mempolicy.c will do this automatically.
  177. * get_policy() must NOT add a ref if the policy at (vma,addr) is not
  178. * marked as MPOL_SHARED. vma policies are protected by the mmap_sem.
  179. * If no [shared/vma] mempolicy exists at the addr, get_policy() op
  180. * must return NULL--i.e., do not "fallback" to task or system default
  181. * policy.
  182. */
  183. struct mempolicy *(*get_policy)(struct vm_area_struct *vma,
  184. unsigned long addr);
  185. int (*migrate)(struct vm_area_struct *vma, const nodemask_t *from,
  186. const nodemask_t *to, unsigned long flags);
  187. #endif
  188. };
  189. struct mmu_gather;
  190. struct inode;
  191. #define page_private(page) ((page)->private)
  192. #define set_page_private(page, v) ((page)->private = (v))
  193. /*
  194. * FIXME: take this include out, include page-flags.h in
  195. * files which need it (119 of them)
  196. */
  197. #include <linux/page-flags.h>
  198. /*
  199. * Methods to modify the page usage count.
  200. *
  201. * What counts for a page usage:
  202. * - cache mapping (page->mapping)
  203. * - private data (page->private)
  204. * - page mapped in a task's page tables, each mapping
  205. * is counted separately
  206. *
  207. * Also, many kernel routines increase the page count before a critical
  208. * routine so they can be sure the page doesn't go away from under them.
  209. */
  210. /*
  211. * Drop a ref, return true if the refcount fell to zero (the page has no users)
  212. */
  213. static inline int put_page_testzero(struct page *page)
  214. {
  215. VM_BUG_ON(atomic_read(&page->_count) == 0);
  216. return atomic_dec_and_test(&page->_count);
  217. }
  218. /*
  219. * Try to grab a ref unless the page has a refcount of zero, return false if
  220. * that is the case.
  221. */
  222. static inline int get_page_unless_zero(struct page *page)
  223. {
  224. return atomic_inc_not_zero(&page->_count);
  225. }
  226. /* Support for virtually mapped pages */
  227. struct page *vmalloc_to_page(const void *addr);
  228. unsigned long vmalloc_to_pfn(const void *addr);
  229. /*
  230. * Determine if an address is within the vmalloc range
  231. *
  232. * On nommu, vmalloc/vfree wrap through kmalloc/kfree directly, so there
  233. * is no special casing required.
  234. */
  235. static inline int is_vmalloc_addr(const void *x)
  236. {
  237. #ifdef CONFIG_MMU
  238. unsigned long addr = (unsigned long)x;
  239. return addr >= VMALLOC_START && addr < VMALLOC_END;
  240. #else
  241. return 0;
  242. #endif
  243. }
  244. #ifdef CONFIG_MMU
  245. extern int is_vmalloc_or_module_addr(const void *x);
  246. #else
  247. static inline int is_vmalloc_or_module_addr(const void *x)
  248. {
  249. return 0;
  250. }
  251. #endif
  252. static inline struct page *compound_head(struct page *page)
  253. {
  254. if (unlikely(PageTail(page)))
  255. return page->first_page;
  256. return page;
  257. }
  258. static inline int page_count(struct page *page)
  259. {
  260. return atomic_read(&compound_head(page)->_count);
  261. }
  262. static inline void get_page(struct page *page)
  263. {
  264. page = compound_head(page);
  265. VM_BUG_ON(atomic_read(&page->_count) == 0);
  266. atomic_inc(&page->_count);
  267. }
  268. static inline struct page *virt_to_head_page(const void *x)
  269. {
  270. struct page *page = virt_to_page(x);
  271. return compound_head(page);
  272. }
  273. /*
  274. * Setup the page count before being freed into the page allocator for
  275. * the first time (boot or memory hotplug)
  276. */
  277. static inline void init_page_count(struct page *page)
  278. {
  279. atomic_set(&page->_count, 1);
  280. }
  281. void put_page(struct page *page);
  282. void put_pages_list(struct list_head *pages);
  283. void split_page(struct page *page, unsigned int order);
  284. /*
  285. * Compound pages have a destructor function. Provide a
  286. * prototype for that function and accessor functions.
  287. * These are _only_ valid on the head of a PG_compound page.
  288. */
  289. typedef void compound_page_dtor(struct page *);
  290. static inline void set_compound_page_dtor(struct page *page,
  291. compound_page_dtor *dtor)
  292. {
  293. page[1].lru.next = (void *)dtor;
  294. }
  295. static inline compound_page_dtor *get_compound_page_dtor(struct page *page)
  296. {
  297. return (compound_page_dtor *)page[1].lru.next;
  298. }
  299. static inline int compound_order(struct page *page)
  300. {
  301. if (!PageHead(page))
  302. return 0;
  303. return (unsigned long)page[1].lru.prev;
  304. }
  305. static inline void set_compound_order(struct page *page, unsigned long order)
  306. {
  307. page[1].lru.prev = (void *)order;
  308. }
  309. /*
  310. * Multiple processes may "see" the same page. E.g. for untouched
  311. * mappings of /dev/null, all processes see the same page full of
  312. * zeroes, and text pages of executables and shared libraries have
  313. * only one copy in memory, at most, normally.
  314. *
  315. * For the non-reserved pages, page_count(page) denotes a reference count.
  316. * page_count() == 0 means the page is free. page->lru is then used for
  317. * freelist management in the buddy allocator.
  318. * page_count() > 0 means the page has been allocated.
  319. *
  320. * Pages are allocated by the slab allocator in order to provide memory
  321. * to kmalloc and kmem_cache_alloc. In this case, the management of the
  322. * page, and the fields in 'struct page' are the responsibility of mm/slab.c
  323. * unless a particular usage is carefully commented. (the responsibility of
  324. * freeing the kmalloc memory is the caller's, of course).
  325. *
  326. * A page may be used by anyone else who does a __get_free_page().
  327. * In this case, page_count still tracks the references, and should only
  328. * be used through the normal accessor functions. The top bits of page->flags
  329. * and page->virtual store page management information, but all other fields
  330. * are unused and could be used privately, carefully. The management of this
  331. * page is the responsibility of the one who allocated it, and those who have
  332. * subsequently been given references to it.
  333. *
  334. * The other pages (we may call them "pagecache pages") are completely
  335. * managed by the Linux memory manager: I/O, buffers, swapping etc.
  336. * The following discussion applies only to them.
  337. *
  338. * A pagecache page contains an opaque `private' member, which belongs to the
  339. * page's address_space. Usually, this is the address of a circular list of
  340. * the page's disk buffers. PG_private must be set to tell the VM to call
  341. * into the filesystem to release these pages.
  342. *
  343. * A page may belong to an inode's memory mapping. In this case, page->mapping
  344. * is the pointer to the inode, and page->index is the file offset of the page,
  345. * in units of PAGE_CACHE_SIZE.
  346. *
  347. * If pagecache pages are not associated with an inode, they are said to be
  348. * anonymous pages. These may become associated with the swapcache, and in that
  349. * case PG_swapcache is set, and page->private is an offset into the swapcache.
  350. *
  351. * In either case (swapcache or inode backed), the pagecache itself holds one
  352. * reference to the page. Setting PG_private should also increment the
  353. * refcount. The each user mapping also has a reference to the page.
  354. *
  355. * The pagecache pages are stored in a per-mapping radix tree, which is
  356. * rooted at mapping->page_tree, and indexed by offset.
  357. * Where 2.4 and early 2.6 kernels kept dirty/clean pages in per-address_space
  358. * lists, we instead now tag pages as dirty/writeback in the radix tree.
  359. *
  360. * All pagecache pages may be subject to I/O:
  361. * - inode pages may need to be read from disk,
  362. * - inode pages which have been modified and are MAP_SHARED may need
  363. * to be written back to the inode on disk,
  364. * - anonymous pages (including MAP_PRIVATE file mappings) which have been
  365. * modified may need to be swapped out to swap space and (later) to be read
  366. * back into memory.
  367. */
  368. /*
  369. * The zone field is never updated after free_area_init_core()
  370. * sets it, so none of the operations on it need to be atomic.
  371. */
  372. /*
  373. * page->flags layout:
  374. *
  375. * There are three possibilities for how page->flags get
  376. * laid out. The first is for the normal case, without
  377. * sparsemem. The second is for sparsemem when there is
  378. * plenty of space for node and section. The last is when
  379. * we have run out of space and have to fall back to an
  380. * alternate (slower) way of determining the node.
  381. *
  382. * No sparsemem or sparsemem vmemmap: | NODE | ZONE | ... | FLAGS |
  383. * classic sparse with space for node:| SECTION | NODE | ZONE | ... | FLAGS |
  384. * classic sparse no space for node: | SECTION | ZONE | ... | FLAGS |
  385. */
  386. #if defined(CONFIG_SPARSEMEM) && !defined(CONFIG_SPARSEMEM_VMEMMAP)
  387. #define SECTIONS_WIDTH SECTIONS_SHIFT
  388. #else
  389. #define SECTIONS_WIDTH 0
  390. #endif
  391. #define ZONES_WIDTH ZONES_SHIFT
  392. #if SECTIONS_WIDTH+ZONES_WIDTH+NODES_SHIFT <= BITS_PER_LONG - NR_PAGEFLAGS
  393. #define NODES_WIDTH NODES_SHIFT
  394. #else
  395. #ifdef CONFIG_SPARSEMEM_VMEMMAP
  396. #error "Vmemmap: No space for nodes field in page flags"
  397. #endif
  398. #define NODES_WIDTH 0
  399. #endif
  400. /* Page flags: | [SECTION] | [NODE] | ZONE | ... | FLAGS | */
  401. #define SECTIONS_PGOFF ((sizeof(unsigned long)*8) - SECTIONS_WIDTH)
  402. #define NODES_PGOFF (SECTIONS_PGOFF - NODES_WIDTH)
  403. #define ZONES_PGOFF (NODES_PGOFF - ZONES_WIDTH)
  404. /*
  405. * We are going to use the flags for the page to node mapping if its in
  406. * there. This includes the case where there is no node, so it is implicit.
  407. */
  408. #if !(NODES_WIDTH > 0 || NODES_SHIFT == 0)
  409. #define NODE_NOT_IN_PAGE_FLAGS
  410. #endif
  411. #ifndef PFN_SECTION_SHIFT
  412. #define PFN_SECTION_SHIFT 0
  413. #endif
  414. /*
  415. * Define the bit shifts to access each section. For non-existant
  416. * sections we define the shift as 0; that plus a 0 mask ensures
  417. * the compiler will optimise away reference to them.
  418. */
  419. #define SECTIONS_PGSHIFT (SECTIONS_PGOFF * (SECTIONS_WIDTH != 0))
  420. #define NODES_PGSHIFT (NODES_PGOFF * (NODES_WIDTH != 0))
  421. #define ZONES_PGSHIFT (ZONES_PGOFF * (ZONES_WIDTH != 0))
  422. /* NODE:ZONE or SECTION:ZONE is used to ID a zone for the buddy allcator */
  423. #ifdef NODE_NOT_IN_PAGEFLAGS
  424. #define ZONEID_SHIFT (SECTIONS_SHIFT + ZONES_SHIFT)
  425. #define ZONEID_PGOFF ((SECTIONS_PGOFF < ZONES_PGOFF)? \
  426. SECTIONS_PGOFF : ZONES_PGOFF)
  427. #else
  428. #define ZONEID_SHIFT (NODES_SHIFT + ZONES_SHIFT)
  429. #define ZONEID_PGOFF ((NODES_PGOFF < ZONES_PGOFF)? \
  430. NODES_PGOFF : ZONES_PGOFF)
  431. #endif
  432. #define ZONEID_PGSHIFT (ZONEID_PGOFF * (ZONEID_SHIFT != 0))
  433. #if SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > BITS_PER_LONG - NR_PAGEFLAGS
  434. #error SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > BITS_PER_LONG - NR_PAGEFLAGS
  435. #endif
  436. #define ZONES_MASK ((1UL << ZONES_WIDTH) - 1)
  437. #define NODES_MASK ((1UL << NODES_WIDTH) - 1)
  438. #define SECTIONS_MASK ((1UL << SECTIONS_WIDTH) - 1)
  439. #define ZONEID_MASK ((1UL << ZONEID_SHIFT) - 1)
  440. static inline enum zone_type page_zonenum(struct page *page)
  441. {
  442. return (page->flags >> ZONES_PGSHIFT) & ZONES_MASK;
  443. }
  444. /*
  445. * The identification function is only used by the buddy allocator for
  446. * determining if two pages could be buddies. We are not really
  447. * identifying a zone since we could be using a the section number
  448. * id if we have not node id available in page flags.
  449. * We guarantee only that it will return the same value for two
  450. * combinable pages in a zone.
  451. */
  452. static inline int page_zone_id(struct page *page)
  453. {
  454. return (page->flags >> ZONEID_PGSHIFT) & ZONEID_MASK;
  455. }
  456. static inline int zone_to_nid(struct zone *zone)
  457. {
  458. #ifdef CONFIG_NUMA
  459. return zone->node;
  460. #else
  461. return 0;
  462. #endif
  463. }
  464. #ifdef NODE_NOT_IN_PAGE_FLAGS
  465. extern int page_to_nid(struct page *page);
  466. #else
  467. static inline int page_to_nid(struct page *page)
  468. {
  469. return (page->flags >> NODES_PGSHIFT) & NODES_MASK;
  470. }
  471. #endif
  472. static inline struct zone *page_zone(struct page *page)
  473. {
  474. return &NODE_DATA(page_to_nid(page))->node_zones[page_zonenum(page)];
  475. }
  476. #if defined(CONFIG_SPARSEMEM) && !defined(CONFIG_SPARSEMEM_VMEMMAP)
  477. static inline unsigned long page_to_section(struct page *page)
  478. {
  479. return (page->flags >> SECTIONS_PGSHIFT) & SECTIONS_MASK;
  480. }
  481. #endif
  482. static inline void set_page_zone(struct page *page, enum zone_type zone)
  483. {
  484. page->flags &= ~(ZONES_MASK << ZONES_PGSHIFT);
  485. page->flags |= (zone & ZONES_MASK) << ZONES_PGSHIFT;
  486. }
  487. static inline void set_page_node(struct page *page, unsigned long node)
  488. {
  489. page->flags &= ~(NODES_MASK << NODES_PGSHIFT);
  490. page->flags |= (node & NODES_MASK) << NODES_PGSHIFT;
  491. }
  492. static inline void set_page_section(struct page *page, unsigned long section)
  493. {
  494. page->flags &= ~(SECTIONS_MASK << SECTIONS_PGSHIFT);
  495. page->flags |= (section & SECTIONS_MASK) << SECTIONS_PGSHIFT;
  496. }
  497. static inline void set_page_links(struct page *page, enum zone_type zone,
  498. unsigned long node, unsigned long pfn)
  499. {
  500. set_page_zone(page, zone);
  501. set_page_node(page, node);
  502. set_page_section(page, pfn_to_section_nr(pfn));
  503. }
  504. /*
  505. * Some inline functions in vmstat.h depend on page_zone()
  506. */
  507. #include <linux/vmstat.h>
  508. static __always_inline void *lowmem_page_address(struct page *page)
  509. {
  510. return __va(page_to_pfn(page) << PAGE_SHIFT);
  511. }
  512. #if defined(CONFIG_HIGHMEM) && !defined(WANT_PAGE_VIRTUAL)
  513. #define HASHED_PAGE_VIRTUAL
  514. #endif
  515. #if defined(WANT_PAGE_VIRTUAL)
  516. #define page_address(page) ((page)->virtual)
  517. #define set_page_address(page, address) \
  518. do { \
  519. (page)->virtual = (address); \
  520. } while(0)
  521. #define page_address_init() do { } while(0)
  522. #endif
  523. #if defined(HASHED_PAGE_VIRTUAL)
  524. void *page_address(struct page *page);
  525. void set_page_address(struct page *page, void *virtual);
  526. void page_address_init(void);
  527. #endif
  528. #if !defined(HASHED_PAGE_VIRTUAL) && !defined(WANT_PAGE_VIRTUAL)
  529. #define page_address(page) lowmem_page_address(page)
  530. #define set_page_address(page, address) do { } while(0)
  531. #define page_address_init() do { } while(0)
  532. #endif
  533. /*
  534. * On an anonymous page mapped into a user virtual memory area,
  535. * page->mapping points to its anon_vma, not to a struct address_space;
  536. * with the PAGE_MAPPING_ANON bit set to distinguish it. See rmap.h.
  537. *
  538. * On an anonymous page in a VM_MERGEABLE area, if CONFIG_KSM is enabled,
  539. * the PAGE_MAPPING_KSM bit may be set along with the PAGE_MAPPING_ANON bit;
  540. * and then page->mapping points, not to an anon_vma, but to a private
  541. * structure which KSM associates with that merged page. See ksm.h.
  542. *
  543. * PAGE_MAPPING_KSM without PAGE_MAPPING_ANON is currently never used.
  544. *
  545. * Please note that, confusingly, "page_mapping" refers to the inode
  546. * address_space which maps the page from disk; whereas "page_mapped"
  547. * refers to user virtual address space into which the page is mapped.
  548. */
  549. #define PAGE_MAPPING_ANON 1
  550. #define PAGE_MAPPING_KSM 2
  551. #define PAGE_MAPPING_FLAGS (PAGE_MAPPING_ANON | PAGE_MAPPING_KSM)
  552. extern struct address_space swapper_space;
  553. static inline struct address_space *page_mapping(struct page *page)
  554. {
  555. struct address_space *mapping = page->mapping;
  556. VM_BUG_ON(PageSlab(page));
  557. #ifdef CONFIG_SWAP
  558. if (unlikely(PageSwapCache(page)))
  559. mapping = &swapper_space;
  560. else
  561. #endif
  562. if (unlikely((unsigned long)mapping & PAGE_MAPPING_ANON))
  563. mapping = NULL;
  564. return mapping;
  565. }
  566. /* Neutral page->mapping pointer to address_space or anon_vma or other */
  567. static inline void *page_rmapping(struct page *page)
  568. {
  569. return (void *)((unsigned long)page->mapping & ~PAGE_MAPPING_FLAGS);
  570. }
  571. static inline int PageAnon(struct page *page)
  572. {
  573. return ((unsigned long)page->mapping & PAGE_MAPPING_ANON) != 0;
  574. }
  575. /*
  576. * Return the pagecache index of the passed page. Regular pagecache pages
  577. * use ->index whereas swapcache pages use ->private
  578. */
  579. static inline pgoff_t page_index(struct page *page)
  580. {
  581. if (unlikely(PageSwapCache(page)))
  582. return page_private(page);
  583. return page->index;
  584. }
  585. /*
  586. * The atomic page->_mapcount, like _count, starts from -1:
  587. * so that transitions both from it and to it can be tracked,
  588. * using atomic_inc_and_test and atomic_add_negative(-1).
  589. */
  590. static inline void reset_page_mapcount(struct page *page)
  591. {
  592. atomic_set(&(page)->_mapcount, -1);
  593. }
  594. static inline int page_mapcount(struct page *page)
  595. {
  596. return atomic_read(&(page)->_mapcount) + 1;
  597. }
  598. /*
  599. * Return true if this page is mapped into pagetables.
  600. */
  601. static inline int page_mapped(struct page *page)
  602. {
  603. return atomic_read(&(page)->_mapcount) >= 0;
  604. }
  605. /*
  606. * Different kinds of faults, as returned by handle_mm_fault().
  607. * Used to decide whether a process gets delivered SIGBUS or
  608. * just gets major/minor fault counters bumped up.
  609. */
  610. #define VM_FAULT_MINOR 0 /* For backwards compat. Remove me quickly. */
  611. #define VM_FAULT_OOM 0x0001
  612. #define VM_FAULT_SIGBUS 0x0002
  613. #define VM_FAULT_MAJOR 0x0004
  614. #define VM_FAULT_WRITE 0x0008 /* Special case for get_user_pages */
  615. #define VM_FAULT_HWPOISON 0x0010 /* Hit poisoned page */
  616. #define VM_FAULT_NOPAGE 0x0100 /* ->fault installed the pte, not return page */
  617. #define VM_FAULT_LOCKED 0x0200 /* ->fault locked the returned page */
  618. #define VM_FAULT_ERROR (VM_FAULT_OOM | VM_FAULT_SIGBUS | VM_FAULT_HWPOISON)
  619. /*
  620. * Can be called by the pagefault handler when it gets a VM_FAULT_OOM.
  621. */
  622. extern void pagefault_out_of_memory(void);
  623. #define offset_in_page(p) ((unsigned long)(p) & ~PAGE_MASK)
  624. extern void show_free_areas(void);
  625. int shmem_lock(struct file *file, int lock, struct user_struct *user);
  626. struct file *shmem_file_setup(const char *name, loff_t size, unsigned long flags);
  627. int shmem_zero_setup(struct vm_area_struct *);
  628. #ifndef CONFIG_MMU
  629. extern unsigned long shmem_get_unmapped_area(struct file *file,
  630. unsigned long addr,
  631. unsigned long len,
  632. unsigned long pgoff,
  633. unsigned long flags);
  634. #endif
  635. extern int can_do_mlock(void);
  636. extern int user_shm_lock(size_t, struct user_struct *);
  637. extern void user_shm_unlock(size_t, struct user_struct *);
  638. /*
  639. * Parameter block passed down to zap_pte_range in exceptional cases.
  640. */
  641. struct zap_details {
  642. struct vm_area_struct *nonlinear_vma; /* Check page->index if set */
  643. struct address_space *check_mapping; /* Check page->mapping if set */
  644. pgoff_t first_index; /* Lowest page->index to unmap */
  645. pgoff_t last_index; /* Highest page->index to unmap */
  646. spinlock_t *i_mmap_lock; /* For unmap_mapping_range: */
  647. unsigned long truncate_count; /* Compare vm_truncate_count */
  648. };
  649. struct page *vm_normal_page(struct vm_area_struct *vma, unsigned long addr,
  650. pte_t pte);
  651. int zap_vma_ptes(struct vm_area_struct *vma, unsigned long address,
  652. unsigned long size);
  653. unsigned long zap_page_range(struct vm_area_struct *vma, unsigned long address,
  654. unsigned long size, struct zap_details *);
  655. unsigned long unmap_vmas(struct mmu_gather **tlb,
  656. struct vm_area_struct *start_vma, unsigned long start_addr,
  657. unsigned long end_addr, unsigned long *nr_accounted,
  658. struct zap_details *);
  659. /**
  660. * mm_walk - callbacks for walk_page_range
  661. * @pgd_entry: if set, called for each non-empty PGD (top-level) entry
  662. * @pud_entry: if set, called for each non-empty PUD (2nd-level) entry
  663. * @pmd_entry: if set, called for each non-empty PMD (3rd-level) entry
  664. * @pte_entry: if set, called for each non-empty PTE (4th-level) entry
  665. * @pte_hole: if set, called for each hole at all levels
  666. *
  667. * (see walk_page_range for more details)
  668. */
  669. struct mm_walk {
  670. int (*pgd_entry)(pgd_t *, unsigned long, unsigned long, struct mm_walk *);
  671. int (*pud_entry)(pud_t *, unsigned long, unsigned long, struct mm_walk *);
  672. int (*pmd_entry)(pmd_t *, unsigned long, unsigned long, struct mm_walk *);
  673. int (*pte_entry)(pte_t *, unsigned long, unsigned long, struct mm_walk *);
  674. int (*pte_hole)(unsigned long, unsigned long, struct mm_walk *);
  675. struct mm_struct *mm;
  676. void *private;
  677. };
  678. int walk_page_range(unsigned long addr, unsigned long end,
  679. struct mm_walk *walk);
  680. void free_pgd_range(struct mmu_gather *tlb, unsigned long addr,
  681. unsigned long end, unsigned long floor, unsigned long ceiling);
  682. int copy_page_range(struct mm_struct *dst, struct mm_struct *src,
  683. struct vm_area_struct *vma);
  684. void unmap_mapping_range(struct address_space *mapping,
  685. loff_t const holebegin, loff_t const holelen, int even_cows);
  686. int follow_pfn(struct vm_area_struct *vma, unsigned long address,
  687. unsigned long *pfn);
  688. int follow_phys(struct vm_area_struct *vma, unsigned long address,
  689. unsigned int flags, unsigned long *prot, resource_size_t *phys);
  690. int generic_access_phys(struct vm_area_struct *vma, unsigned long addr,
  691. void *buf, int len, int write);
  692. static inline void unmap_shared_mapping_range(struct address_space *mapping,
  693. loff_t const holebegin, loff_t const holelen)
  694. {
  695. unmap_mapping_range(mapping, holebegin, holelen, 0);
  696. }
  697. extern void truncate_pagecache(struct inode *inode, loff_t old, loff_t new);
  698. extern int vmtruncate(struct inode *inode, loff_t offset);
  699. extern int vmtruncate_range(struct inode *inode, loff_t offset, loff_t end);
  700. int truncate_inode_page(struct address_space *mapping, struct page *page);
  701. int generic_error_remove_page(struct address_space *mapping, struct page *page);
  702. int invalidate_inode_page(struct page *page);
  703. #ifdef CONFIG_MMU
  704. extern int handle_mm_fault(struct mm_struct *mm, struct vm_area_struct *vma,
  705. unsigned long address, unsigned int flags);
  706. #else
  707. static inline int handle_mm_fault(struct mm_struct *mm,
  708. struct vm_area_struct *vma, unsigned long address,
  709. unsigned int flags)
  710. {
  711. /* should never happen if there's no MMU */
  712. BUG();
  713. return VM_FAULT_SIGBUS;
  714. }
  715. #endif
  716. extern int make_pages_present(unsigned long addr, unsigned long end);
  717. extern int access_process_vm(struct task_struct *tsk, unsigned long addr, void *buf, int len, int write);
  718. int get_user_pages(struct task_struct *tsk, struct mm_struct *mm,
  719. unsigned long start, int nr_pages, int write, int force,
  720. struct page **pages, struct vm_area_struct **vmas);
  721. int get_user_pages_fast(unsigned long start, int nr_pages, int write,
  722. struct page **pages);
  723. struct page *get_dump_page(unsigned long addr);
  724. extern int try_to_release_page(struct page * page, gfp_t gfp_mask);
  725. extern void do_invalidatepage(struct page *page, unsigned long offset);
  726. int __set_page_dirty_nobuffers(struct page *page);
  727. int __set_page_dirty_no_writeback(struct page *page);
  728. int redirty_page_for_writepage(struct writeback_control *wbc,
  729. struct page *page);
  730. void account_page_dirtied(struct page *page, struct address_space *mapping);
  731. int set_page_dirty(struct page *page);
  732. int set_page_dirty_lock(struct page *page);
  733. int clear_page_dirty_for_io(struct page *page);
  734. extern unsigned long move_page_tables(struct vm_area_struct *vma,
  735. unsigned long old_addr, struct vm_area_struct *new_vma,
  736. unsigned long new_addr, unsigned long len);
  737. extern unsigned long do_mremap(unsigned long addr,
  738. unsigned long old_len, unsigned long new_len,
  739. unsigned long flags, unsigned long new_addr);
  740. extern int mprotect_fixup(struct vm_area_struct *vma,
  741. struct vm_area_struct **pprev, unsigned long start,
  742. unsigned long end, unsigned long newflags);
  743. /*
  744. * doesn't attempt to fault and will return short.
  745. */
  746. int __get_user_pages_fast(unsigned long start, int nr_pages, int write,
  747. struct page **pages);
  748. /*
  749. * A callback you can register to apply pressure to ageable caches.
  750. *
  751. * 'shrink' is passed a count 'nr_to_scan' and a 'gfpmask'. It should
  752. * look through the least-recently-used 'nr_to_scan' entries and
  753. * attempt to free them up. It should return the number of objects
  754. * which remain in the cache. If it returns -1, it means it cannot do
  755. * any scanning at this time (eg. there is a risk of deadlock).
  756. *
  757. * The 'gfpmask' refers to the allocation we are currently trying to
  758. * fulfil.
  759. *
  760. * Note that 'shrink' will be passed nr_to_scan == 0 when the VM is
  761. * querying the cache size, so a fastpath for that case is appropriate.
  762. */
  763. struct shrinker {
  764. int (*shrink)(int nr_to_scan, gfp_t gfp_mask);
  765. int seeks; /* seeks to recreate an obj */
  766. /* These are for internal use */
  767. struct list_head list;
  768. long nr; /* objs pending delete */
  769. };
  770. #define DEFAULT_SEEKS 2 /* A good number if you don't know better. */
  771. extern void register_shrinker(struct shrinker *);
  772. extern void unregister_shrinker(struct shrinker *);
  773. int vma_wants_writenotify(struct vm_area_struct *vma);
  774. extern pte_t *get_locked_pte(struct mm_struct *mm, unsigned long addr, spinlock_t **ptl);
  775. #ifdef __PAGETABLE_PUD_FOLDED
  776. static inline int __pud_alloc(struct mm_struct *mm, pgd_t *pgd,
  777. unsigned long address)
  778. {
  779. return 0;
  780. }
  781. #else
  782. int __pud_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address);
  783. #endif
  784. #ifdef __PAGETABLE_PMD_FOLDED
  785. static inline int __pmd_alloc(struct mm_struct *mm, pud_t *pud,
  786. unsigned long address)
  787. {
  788. return 0;
  789. }
  790. #else
  791. int __pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address);
  792. #endif
  793. int __pte_alloc(struct mm_struct *mm, pmd_t *pmd, unsigned long address);
  794. int __pte_alloc_kernel(pmd_t *pmd, unsigned long address);
  795. /*
  796. * The following ifdef needed to get the 4level-fixup.h header to work.
  797. * Remove it when 4level-fixup.h has been removed.
  798. */
  799. #if defined(CONFIG_MMU) && !defined(__ARCH_HAS_4LEVEL_HACK)
  800. static inline pud_t *pud_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address)
  801. {
  802. return (unlikely(pgd_none(*pgd)) && __pud_alloc(mm, pgd, address))?
  803. NULL: pud_offset(pgd, address);
  804. }
  805. static inline pmd_t *pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address)
  806. {
  807. return (unlikely(pud_none(*pud)) && __pmd_alloc(mm, pud, address))?
  808. NULL: pmd_offset(pud, address);
  809. }
  810. #endif /* CONFIG_MMU && !__ARCH_HAS_4LEVEL_HACK */
  811. #if USE_SPLIT_PTLOCKS
  812. /*
  813. * We tuck a spinlock to guard each pagetable page into its struct page,
  814. * at page->private, with BUILD_BUG_ON to make sure that this will not
  815. * overflow into the next struct page (as it might with DEBUG_SPINLOCK).
  816. * When freeing, reset page->mapping so free_pages_check won't complain.
  817. */
  818. #define __pte_lockptr(page) &((page)->ptl)
  819. #define pte_lock_init(_page) do { \
  820. spin_lock_init(__pte_lockptr(_page)); \
  821. } while (0)
  822. #define pte_lock_deinit(page) ((page)->mapping = NULL)
  823. #define pte_lockptr(mm, pmd) ({(void)(mm); __pte_lockptr(pmd_page(*(pmd)));})
  824. #else /* !USE_SPLIT_PTLOCKS */
  825. /*
  826. * We use mm->page_table_lock to guard all pagetable pages of the mm.
  827. */
  828. #define pte_lock_init(page) do {} while (0)
  829. #define pte_lock_deinit(page) do {} while (0)
  830. #define pte_lockptr(mm, pmd) ({(void)(pmd); &(mm)->page_table_lock;})
  831. #endif /* USE_SPLIT_PTLOCKS */
  832. static inline void pgtable_page_ctor(struct page *page)
  833. {
  834. pte_lock_init(page);
  835. inc_zone_page_state(page, NR_PAGETABLE);
  836. }
  837. static inline void pgtable_page_dtor(struct page *page)
  838. {
  839. pte_lock_deinit(page);
  840. dec_zone_page_state(page, NR_PAGETABLE);
  841. }
  842. #define pte_offset_map_lock(mm, pmd, address, ptlp) \
  843. ({ \
  844. spinlock_t *__ptl = pte_lockptr(mm, pmd); \
  845. pte_t *__pte = pte_offset_map(pmd, address); \
  846. *(ptlp) = __ptl; \
  847. spin_lock(__ptl); \
  848. __pte; \
  849. })
  850. #define pte_unmap_unlock(pte, ptl) do { \
  851. spin_unlock(ptl); \
  852. pte_unmap(pte); \
  853. } while (0)
  854. #define pte_alloc_map(mm, pmd, address) \
  855. ((unlikely(!pmd_present(*(pmd))) && __pte_alloc(mm, pmd, address))? \
  856. NULL: pte_offset_map(pmd, address))
  857. #define pte_alloc_map_lock(mm, pmd, address, ptlp) \
  858. ((unlikely(!pmd_present(*(pmd))) && __pte_alloc(mm, pmd, address))? \
  859. NULL: pte_offset_map_lock(mm, pmd, address, ptlp))
  860. #define pte_alloc_kernel(pmd, address) \
  861. ((unlikely(!pmd_present(*(pmd))) && __pte_alloc_kernel(pmd, address))? \
  862. NULL: pte_offset_kernel(pmd, address))
  863. extern void free_area_init(unsigned long * zones_size);
  864. extern void free_area_init_node(int nid, unsigned long * zones_size,
  865. unsigned long zone_start_pfn, unsigned long *zholes_size);
  866. #ifdef CONFIG_ARCH_POPULATES_NODE_MAP
  867. /*
  868. * With CONFIG_ARCH_POPULATES_NODE_MAP set, an architecture may initialise its
  869. * zones, allocate the backing mem_map and account for memory holes in a more
  870. * architecture independent manner. This is a substitute for creating the
  871. * zone_sizes[] and zholes_size[] arrays and passing them to
  872. * free_area_init_node()
  873. *
  874. * An architecture is expected to register range of page frames backed by
  875. * physical memory with add_active_range() before calling
  876. * free_area_init_nodes() passing in the PFN each zone ends at. At a basic
  877. * usage, an architecture is expected to do something like
  878. *
  879. * unsigned long max_zone_pfns[MAX_NR_ZONES] = {max_dma, max_normal_pfn,
  880. * max_highmem_pfn};
  881. * for_each_valid_physical_page_range()
  882. * add_active_range(node_id, start_pfn, end_pfn)
  883. * free_area_init_nodes(max_zone_pfns);
  884. *
  885. * If the architecture guarantees that there are no holes in the ranges
  886. * registered with add_active_range(), free_bootmem_active_regions()
  887. * will call free_bootmem_node() for each registered physical page range.
  888. * Similarly sparse_memory_present_with_active_regions() calls
  889. * memory_present() for each range when SPARSEMEM is enabled.
  890. *
  891. * See mm/page_alloc.c for more information on each function exposed by
  892. * CONFIG_ARCH_POPULATES_NODE_MAP
  893. */
  894. extern void free_area_init_nodes(unsigned long *max_zone_pfn);
  895. extern void add_active_range(unsigned int nid, unsigned long start_pfn,
  896. unsigned long end_pfn);
  897. extern void remove_active_range(unsigned int nid, unsigned long start_pfn,
  898. unsigned long end_pfn);
  899. extern void remove_all_active_ranges(void);
  900. extern unsigned long absent_pages_in_range(unsigned long start_pfn,
  901. unsigned long end_pfn);
  902. extern void get_pfn_range_for_nid(unsigned int nid,
  903. unsigned long *start_pfn, unsigned long *end_pfn);
  904. extern unsigned long find_min_pfn_with_active_regions(void);
  905. extern void free_bootmem_with_active_regions(int nid,
  906. unsigned long max_low_pfn);
  907. typedef int (*work_fn_t)(unsigned long, unsigned long, void *);
  908. extern void work_with_active_regions(int nid, work_fn_t work_fn, void *data);
  909. extern void sparse_memory_present_with_active_regions(int nid);
  910. #endif /* CONFIG_ARCH_POPULATES_NODE_MAP */
  911. #if !defined(CONFIG_ARCH_POPULATES_NODE_MAP) && \
  912. !defined(CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID)
  913. static inline int __early_pfn_to_nid(unsigned long pfn)
  914. {
  915. return 0;
  916. }
  917. #else
  918. /* please see mm/page_alloc.c */
  919. extern int __meminit early_pfn_to_nid(unsigned long pfn);
  920. #ifdef CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID
  921. /* there is a per-arch backend function. */
  922. extern int __meminit __early_pfn_to_nid(unsigned long pfn);
  923. #endif /* CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID */
  924. #endif
  925. extern void set_dma_reserve(unsigned long new_dma_reserve);
  926. extern void memmap_init_zone(unsigned long, int, unsigned long,
  927. unsigned long, enum memmap_context);
  928. extern void setup_per_zone_wmarks(void);
  929. extern void calculate_zone_inactive_ratio(struct zone *zone);
  930. extern void mem_init(void);
  931. extern void __init mmap_init(void);
  932. extern void show_mem(void);
  933. extern void si_meminfo(struct sysinfo * val);
  934. extern void si_meminfo_node(struct sysinfo *val, int nid);
  935. extern int after_bootmem;
  936. #ifdef CONFIG_NUMA
  937. extern void setup_per_cpu_pageset(void);
  938. #else
  939. static inline void setup_per_cpu_pageset(void) {}
  940. #endif
  941. extern void zone_pcp_update(struct zone *zone);
  942. /* nommu.c */
  943. extern atomic_long_t mmap_pages_allocated;
  944. /* prio_tree.c */
  945. void vma_prio_tree_add(struct vm_area_struct *, struct vm_area_struct *old);
  946. void vma_prio_tree_insert(struct vm_area_struct *, struct prio_tree_root *);
  947. void vma_prio_tree_remove(struct vm_area_struct *, struct prio_tree_root *);
  948. struct vm_area_struct *vma_prio_tree_next(struct vm_area_struct *vma,
  949. struct prio_tree_iter *iter);
  950. #define vma_prio_tree_foreach(vma, iter, root, begin, end) \
  951. for (prio_tree_iter_init(iter, root, begin, end), vma = NULL; \
  952. (vma = vma_prio_tree_next(vma, iter)); )
  953. static inline void vma_nonlinear_insert(struct vm_area_struct *vma,
  954. struct list_head *list)
  955. {
  956. vma->shared.vm_set.parent = NULL;
  957. list_add_tail(&vma->shared.vm_set.list, list);
  958. }
  959. /* mmap.c */
  960. extern int __vm_enough_memory(struct mm_struct *mm, long pages, int cap_sys_admin);
  961. extern void vma_adjust(struct vm_area_struct *vma, unsigned long start,
  962. unsigned long end, pgoff_t pgoff, struct vm_area_struct *insert);
  963. extern struct vm_area_struct *vma_merge(struct mm_struct *,
  964. struct vm_area_struct *prev, unsigned long addr, unsigned long end,
  965. unsigned long vm_flags, struct anon_vma *, struct file *, pgoff_t,
  966. struct mempolicy *);
  967. extern struct anon_vma *find_mergeable_anon_vma(struct vm_area_struct *);
  968. extern int split_vma(struct mm_struct *,
  969. struct vm_area_struct *, unsigned long addr, int new_below);
  970. extern int insert_vm_struct(struct mm_struct *, struct vm_area_struct *);
  971. extern void __vma_link_rb(struct mm_struct *, struct vm_area_struct *,
  972. struct rb_node **, struct rb_node *);
  973. extern void unlink_file_vma(struct vm_area_struct *);
  974. extern struct vm_area_struct *copy_vma(struct vm_area_struct **,
  975. unsigned long addr, unsigned long len, pgoff_t pgoff);
  976. extern void exit_mmap(struct mm_struct *);
  977. extern int mm_take_all_locks(struct mm_struct *mm);
  978. extern void mm_drop_all_locks(struct mm_struct *mm);
  979. #ifdef CONFIG_PROC_FS
  980. /* From fs/proc/base.c. callers must _not_ hold the mm's exe_file_lock */
  981. extern void added_exe_file_vma(struct mm_struct *mm);
  982. extern void removed_exe_file_vma(struct mm_struct *mm);
  983. #else
  984. static inline void added_exe_file_vma(struct mm_struct *mm)
  985. {}
  986. static inline void removed_exe_file_vma(struct mm_struct *mm)
  987. {}
  988. #endif /* CONFIG_PROC_FS */
  989. extern int may_expand_vm(struct mm_struct *mm, unsigned long npages);
  990. extern int install_special_mapping(struct mm_struct *mm,
  991. unsigned long addr, unsigned long len,
  992. unsigned long flags, struct page **pages);
  993. extern unsigned long get_unmapped_area(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
  994. extern unsigned long do_mmap_pgoff(struct file *file, unsigned long addr,
  995. unsigned long len, unsigned long prot,
  996. unsigned long flag, unsigned long pgoff);
  997. extern unsigned long mmap_region(struct file *file, unsigned long addr,
  998. unsigned long len, unsigned long flags,
  999. unsigned int vm_flags, unsigned long pgoff);
  1000. static inline unsigned long do_mmap(struct file *file, unsigned long addr,
  1001. unsigned long len, unsigned long prot,
  1002. unsigned long flag, unsigned long offset)
  1003. {
  1004. unsigned long ret = -EINVAL;
  1005. if ((offset + PAGE_ALIGN(len)) < offset)
  1006. goto out;
  1007. if (!(offset & ~PAGE_MASK))
  1008. ret = do_mmap_pgoff(file, addr, len, prot, flag, offset >> PAGE_SHIFT);
  1009. out:
  1010. return ret;
  1011. }
  1012. extern int do_munmap(struct mm_struct *, unsigned long, size_t);
  1013. extern unsigned long do_brk(unsigned long, unsigned long);
  1014. /* filemap.c */
  1015. extern unsigned long page_unuse(struct page *);
  1016. extern void truncate_inode_pages(struct address_space *, loff_t);
  1017. extern void truncate_inode_pages_range(struct address_space *,
  1018. loff_t lstart, loff_t lend);
  1019. /* generic vm_area_ops exported for stackable file systems */
  1020. extern int filemap_fault(struct vm_area_struct *, struct vm_fault *);
  1021. /* mm/page-writeback.c */
  1022. int write_one_page(struct page *page, int wait);
  1023. void task_dirty_inc(struct task_struct *tsk);
  1024. /* readahead.c */
  1025. #define VM_MAX_READAHEAD 128 /* kbytes */
  1026. #define VM_MIN_READAHEAD 16 /* kbytes (includes current page) */
  1027. int force_page_cache_readahead(struct address_space *mapping, struct file *filp,
  1028. pgoff_t offset, unsigned long nr_to_read);
  1029. void page_cache_sync_readahead(struct address_space *mapping,
  1030. struct file_ra_state *ra,
  1031. struct file *filp,
  1032. pgoff_t offset,
  1033. unsigned long size);
  1034. void page_cache_async_readahead(struct address_space *mapping,
  1035. struct file_ra_state *ra,
  1036. struct file *filp,
  1037. struct page *pg,
  1038. pgoff_t offset,
  1039. unsigned long size);
  1040. unsigned long max_sane_readahead(unsigned long nr);
  1041. unsigned long ra_submit(struct file_ra_state *ra,
  1042. struct address_space *mapping,
  1043. struct file *filp);
  1044. /* Do stack extension */
  1045. extern int expand_stack(struct vm_area_struct *vma, unsigned long address);
  1046. #ifdef CONFIG_IA64
  1047. extern int expand_upwards(struct vm_area_struct *vma, unsigned long address);
  1048. #endif
  1049. extern int expand_stack_downwards(struct vm_area_struct *vma,
  1050. unsigned long address);
  1051. /* Look up the first VMA which satisfies addr < vm_end, NULL if none. */
  1052. extern struct vm_area_struct * find_vma(struct mm_struct * mm, unsigned long addr);
  1053. extern struct vm_area_struct * find_vma_prev(struct mm_struct * mm, unsigned long addr,
  1054. struct vm_area_struct **pprev);
  1055. /* Look up the first VMA which intersects the interval start_addr..end_addr-1,
  1056. NULL if none. Assume start_addr < end_addr. */
  1057. static inline struct vm_area_struct * find_vma_intersection(struct mm_struct * mm, unsigned long start_addr, unsigned long end_addr)
  1058. {
  1059. struct vm_area_struct * vma = find_vma(mm,start_addr);
  1060. if (vma && end_addr <= vma->vm_start)
  1061. vma = NULL;
  1062. return vma;
  1063. }
  1064. static inline unsigned long vma_pages(struct vm_area_struct *vma)
  1065. {
  1066. return (vma->vm_end - vma->vm_start) >> PAGE_SHIFT;
  1067. }
  1068. pgprot_t vm_get_page_prot(unsigned long vm_flags);
  1069. struct vm_area_struct *find_extend_vma(struct mm_struct *, unsigned long addr);
  1070. int remap_pfn_range(struct vm_area_struct *, unsigned long addr,
  1071. unsigned long pfn, unsigned long size, pgprot_t);
  1072. int vm_insert_page(struct vm_area_struct *, unsigned long addr, struct page *);
  1073. int vm_insert_pfn(struct vm_area_struct *vma, unsigned long addr,
  1074. unsigned long pfn);
  1075. int vm_insert_mixed(struct vm_area_struct *vma, unsigned long addr,
  1076. unsigned long pfn);
  1077. struct page *follow_page(struct vm_area_struct *, unsigned long address,
  1078. unsigned int foll_flags);
  1079. #define FOLL_WRITE 0x01 /* check pte is writable */
  1080. #define FOLL_TOUCH 0x02 /* mark page accessed */
  1081. #define FOLL_GET 0x04 /* do get_page on page */
  1082. #define FOLL_DUMP 0x08 /* give error on hole if it would be zero */
  1083. #define FOLL_FORCE 0x10 /* get_user_pages read/write w/o permission */
  1084. typedef int (*pte_fn_t)(pte_t *pte, pgtable_t token, unsigned long addr,
  1085. void *data);
  1086. extern int apply_to_page_range(struct mm_struct *mm, unsigned long address,
  1087. unsigned long size, pte_fn_t fn, void *data);
  1088. #ifdef CONFIG_PROC_FS
  1089. void vm_stat_account(struct mm_struct *, unsigned long, struct file *, long);
  1090. #else
  1091. static inline void vm_stat_account(struct mm_struct *mm,
  1092. unsigned long flags, struct file *file, long pages)
  1093. {
  1094. }
  1095. #endif /* CONFIG_PROC_FS */
  1096. #ifdef CONFIG_DEBUG_PAGEALLOC
  1097. extern int debug_pagealloc_enabled;
  1098. extern void kernel_map_pages(struct page *page, int numpages, int enable);
  1099. static inline void enable_debug_pagealloc(void)
  1100. {
  1101. debug_pagealloc_enabled = 1;
  1102. }
  1103. #ifdef CONFIG_HIBERNATION
  1104. extern bool kernel_page_present(struct page *page);
  1105. #endif /* CONFIG_HIBERNATION */
  1106. #else
  1107. static inline void
  1108. kernel_map_pages(struct page *page, int numpages, int enable) {}
  1109. static inline void enable_debug_pagealloc(void)
  1110. {
  1111. }
  1112. #ifdef CONFIG_HIBERNATION
  1113. static inline bool kernel_page_present(struct page *page) { return true; }
  1114. #endif /* CONFIG_HIBERNATION */
  1115. #endif
  1116. extern struct vm_area_struct *get_gate_vma(struct task_struct *tsk);
  1117. #ifdef __HAVE_ARCH_GATE_AREA
  1118. int in_gate_area_no_task(unsigned long addr);
  1119. int in_gate_area(struct task_struct *task, unsigned long addr);
  1120. #else
  1121. int in_gate_area_no_task(unsigned long addr);
  1122. #define in_gate_area(task, addr) ({(void)task; in_gate_area_no_task(addr);})
  1123. #endif /* __HAVE_ARCH_GATE_AREA */
  1124. int drop_caches_sysctl_handler(struct ctl_table *, int,
  1125. void __user *, size_t *, loff_t *);
  1126. unsigned long shrink_slab(unsigned long scanned, gfp_t gfp_mask,
  1127. unsigned long lru_pages);
  1128. #ifndef CONFIG_MMU
  1129. #define randomize_va_space 0
  1130. #else
  1131. extern int randomize_va_space;
  1132. #endif
  1133. const char * arch_vma_name(struct vm_area_struct *vma);
  1134. void print_vma_addr(char *prefix, unsigned long rip);
  1135. struct page *sparse_mem_map_populate(unsigned long pnum, int nid);
  1136. pgd_t *vmemmap_pgd_populate(unsigned long addr, int node);
  1137. pud_t *vmemmap_pud_populate(pgd_t *pgd, unsigned long addr, int node);
  1138. pmd_t *vmemmap_pmd_populate(pud_t *pud, unsigned long addr, int node);
  1139. pte_t *vmemmap_pte_populate(pmd_t *pmd, unsigned long addr, int node);
  1140. void *vmemmap_alloc_block(unsigned long size, int node);
  1141. void vmemmap_verify(pte_t *, int, unsigned long, unsigned long);
  1142. int vmemmap_populate_basepages(struct page *start_page,
  1143. unsigned long pages, int node);
  1144. int vmemmap_populate(struct page *start_page, unsigned long pages, int node);
  1145. void vmemmap_populate_print_last(void);
  1146. extern int account_locked_memory(struct mm_struct *mm, struct rlimit *rlim,
  1147. size_t size);
  1148. extern void refund_locked_memory(struct mm_struct *mm, size_t size);
  1149. extern void memory_failure(unsigned long pfn, int trapno);
  1150. extern int __memory_failure(unsigned long pfn, int trapno, int ref);
  1151. extern int sysctl_memory_failure_early_kill;
  1152. extern int sysctl_memory_failure_recovery;
  1153. extern atomic_long_t mce_bad_pages;
  1154. #endif /* __KERNEL__ */
  1155. #endif /* _LINUX_MM_H */