mm.h 45 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662663664665666667668669670671672673674675676677678679680681682683684685686687688689690691692693694695696697698699700701702703704705706707708709710711712713714715716717718719720721722723724725726727728729730731732733734735736737738739740741742743744745746747748749750751752753754755756757758759760761762763764765766767768769770771772773774775776777778779780781782783784785786787788789790791792793794795796797798799800801802803804805806807808809810811812813814815816817818819820821822823824825826827828829830831832833834835836837838839840841842843844845846847848849850851852853854855856857858859860861862863864865866867868869870871872873874875876877878879880881882883884885886887888889890891892893894895896897898899900901902903904905906907908909910911912913914915916917918919920921922923924925926927928929930931932933934935936937938939940941942943944945946947948949950951952953954955956957958959960961962963964965966967968969970971972973974975976977978979980981982983984985986987988989990991992993994995996997998999100010011002100310041005100610071008100910101011101210131014101510161017101810191020102110221023102410251026102710281029103010311032103310341035103610371038103910401041104210431044104510461047104810491050105110521053105410551056105710581059106010611062106310641065106610671068106910701071107210731074107510761077107810791080108110821083108410851086108710881089109010911092109310941095109610971098109911001101110211031104110511061107110811091110111111121113111411151116111711181119112011211122112311241125112611271128112911301131113211331134113511361137113811391140114111421143114411451146114711481149115011511152115311541155115611571158115911601161116211631164116511661167116811691170117111721173117411751176117711781179118011811182118311841185118611871188118911901191119211931194119511961197119811991200120112021203120412051206120712081209121012111212121312141215121612171218121912201221122212231224122512261227122812291230123112321233123412351236123712381239124012411242124312441245124612471248124912501251125212531254125512561257125812591260126112621263126412651266126712681269127012711272127312741275127612771278127912801281128212831284128512861287128812891290129112921293129412951296129712981299130013011302130313041305130613071308130913101311131213131314131513161317131813191320132113221323132413251326
  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/mmdebug.h>
  8. #include <linux/mmzone.h>
  9. #include <linux/rbtree.h>
  10. #include <linux/prio_tree.h>
  11. #include <linux/debug_locks.h>
  12. #include <linux/mm_types.h>
  13. struct mempolicy;
  14. struct anon_vma;
  15. struct file_ra_state;
  16. struct user_struct;
  17. struct writeback_control;
  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 void * high_memory;
  23. extern int page_cluster;
  24. #ifdef CONFIG_SYSCTL
  25. extern int sysctl_legacy_va_layout;
  26. #else
  27. #define sysctl_legacy_va_layout 0
  28. #endif
  29. extern unsigned long mmap_min_addr;
  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. #ifndef VM_STACK_DEFAULT_FLAGS /* arch can override this */
  87. #define VM_STACK_DEFAULT_FLAGS VM_DATA_DEFAULT_FLAGS
  88. #endif
  89. #ifdef CONFIG_STACK_GROWSUP
  90. #define VM_STACK_FLAGS (VM_GROWSUP | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT)
  91. #else
  92. #define VM_STACK_FLAGS (VM_GROWSDOWN | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT)
  93. #endif
  94. #define VM_READHINTMASK (VM_SEQ_READ | VM_RAND_READ)
  95. #define VM_ClearReadHint(v) (v)->vm_flags &= ~VM_READHINTMASK
  96. #define VM_NormalReadHint(v) (!((v)->vm_flags & VM_READHINTMASK))
  97. #define VM_SequentialReadHint(v) ((v)->vm_flags & VM_SEQ_READ)
  98. #define VM_RandomReadHint(v) ((v)->vm_flags & VM_RAND_READ)
  99. /*
  100. * special vmas that are non-mergable, non-mlock()able
  101. */
  102. #define VM_SPECIAL (VM_IO | VM_DONTEXPAND | VM_RESERVED | VM_PFNMAP)
  103. /*
  104. * mapping from the currently active vm_flags protection bits (the
  105. * low four bits) to a page protection mask..
  106. */
  107. extern pgprot_t protection_map[16];
  108. #define FAULT_FLAG_WRITE 0x01 /* Fault was a write access */
  109. #define FAULT_FLAG_NONLINEAR 0x02 /* Fault was via a nonlinear mapping */
  110. #define FAULT_FLAG_MKWRITE 0x04 /* Fault was mkwrite of existing pte */
  111. /*
  112. * This interface is used by x86 PAT code to identify a pfn mapping that is
  113. * linear over entire vma. This is to optimize PAT code that deals with
  114. * marking the physical region with a particular prot. This is not for generic
  115. * mm use. Note also that this check will not work if the pfn mapping is
  116. * linear for a vma starting at physical address 0. In which case PAT code
  117. * falls back to slow path of reserving physical range page by page.
  118. */
  119. static inline int is_linear_pfn_mapping(struct vm_area_struct *vma)
  120. {
  121. return (vma->vm_flags & VM_PFN_AT_MMAP);
  122. }
  123. static inline int is_pfn_mapping(struct vm_area_struct *vma)
  124. {
  125. return (vma->vm_flags & VM_PFNMAP);
  126. }
  127. /*
  128. * vm_fault is filled by the the pagefault handler and passed to the vma's
  129. * ->fault function. The vma's ->fault is responsible for returning a bitmask
  130. * of VM_FAULT_xxx flags that give details about how the fault was handled.
  131. *
  132. * pgoff should be used in favour of virtual_address, if possible. If pgoff
  133. * is used, one may set VM_CAN_NONLINEAR in the vma->vm_flags to get nonlinear
  134. * mapping support.
  135. */
  136. struct vm_fault {
  137. unsigned int flags; /* FAULT_FLAG_xxx flags */
  138. pgoff_t pgoff; /* Logical page offset based on vma */
  139. void __user *virtual_address; /* Faulting virtual address */
  140. struct page *page; /* ->fault handlers should return a
  141. * page here, unless VM_FAULT_NOPAGE
  142. * is set (which is also implied by
  143. * VM_FAULT_ERROR).
  144. */
  145. };
  146. /*
  147. * These are the virtual MM functions - opening of an area, closing and
  148. * unmapping it (needed to keep files on disk up-to-date etc), pointer
  149. * to the functions called when a no-page or a wp-page exception occurs.
  150. */
  151. struct vm_operations_struct {
  152. void (*open)(struct vm_area_struct * area);
  153. void (*close)(struct vm_area_struct * area);
  154. int (*fault)(struct vm_area_struct *vma, struct vm_fault *vmf);
  155. /* notification that a previously read-only page is about to become
  156. * writable, if an error is returned it will cause a SIGBUS */
  157. int (*page_mkwrite)(struct vm_area_struct *vma, struct vm_fault *vmf);
  158. /* called by access_process_vm when get_user_pages() fails, typically
  159. * for use by special VMAs that can switch between memory and hardware
  160. */
  161. int (*access)(struct vm_area_struct *vma, unsigned long addr,
  162. void *buf, int len, int write);
  163. #ifdef CONFIG_NUMA
  164. /*
  165. * set_policy() op must add a reference to any non-NULL @new mempolicy
  166. * to hold the policy upon return. Caller should pass NULL @new to
  167. * remove a policy and fall back to surrounding context--i.e. do not
  168. * install a MPOL_DEFAULT policy, nor the task or system default
  169. * mempolicy.
  170. */
  171. int (*set_policy)(struct vm_area_struct *vma, struct mempolicy *new);
  172. /*
  173. * get_policy() op must add reference [mpol_get()] to any policy at
  174. * (vma,addr) marked as MPOL_SHARED. The shared policy infrastructure
  175. * in mm/mempolicy.c will do this automatically.
  176. * get_policy() must NOT add a ref if the policy at (vma,addr) is not
  177. * marked as MPOL_SHARED. vma policies are protected by the mmap_sem.
  178. * If no [shared/vma] mempolicy exists at the addr, get_policy() op
  179. * must return NULL--i.e., do not "fallback" to task or system default
  180. * policy.
  181. */
  182. struct mempolicy *(*get_policy)(struct vm_area_struct *vma,
  183. unsigned long addr);
  184. int (*migrate)(struct vm_area_struct *vma, const nodemask_t *from,
  185. const nodemask_t *to, unsigned long flags);
  186. #endif
  187. };
  188. struct mmu_gather;
  189. struct inode;
  190. #define page_private(page) ((page)->private)
  191. #define set_page_private(page, v) ((page)->private = (v))
  192. /*
  193. * FIXME: take this include out, include page-flags.h in
  194. * files which need it (119 of them)
  195. */
  196. #include <linux/page-flags.h>
  197. /*
  198. * Methods to modify the page usage count.
  199. *
  200. * What counts for a page usage:
  201. * - cache mapping (page->mapping)
  202. * - private data (page->private)
  203. * - page mapped in a task's page tables, each mapping
  204. * is counted separately
  205. *
  206. * Also, many kernel routines increase the page count before a critical
  207. * routine so they can be sure the page doesn't go away from under them.
  208. */
  209. /*
  210. * Drop a ref, return true if the refcount fell to zero (the page has no users)
  211. */
  212. static inline int put_page_testzero(struct page *page)
  213. {
  214. VM_BUG_ON(atomic_read(&page->_count) == 0);
  215. return atomic_dec_and_test(&page->_count);
  216. }
  217. /*
  218. * Try to grab a ref unless the page has a refcount of zero, return false if
  219. * that is the case.
  220. */
  221. static inline int get_page_unless_zero(struct page *page)
  222. {
  223. return atomic_inc_not_zero(&page->_count);
  224. }
  225. /* Support for virtually mapped pages */
  226. struct page *vmalloc_to_page(const void *addr);
  227. unsigned long vmalloc_to_pfn(const void *addr);
  228. /*
  229. * Determine if an address is within the vmalloc range
  230. *
  231. * On nommu, vmalloc/vfree wrap through kmalloc/kfree directly, so there
  232. * is no special casing required.
  233. */
  234. static inline int is_vmalloc_addr(const void *x)
  235. {
  236. #ifdef CONFIG_MMU
  237. unsigned long addr = (unsigned long)x;
  238. return addr >= VMALLOC_START && addr < VMALLOC_END;
  239. #else
  240. return 0;
  241. #endif
  242. }
  243. static inline struct page *compound_head(struct page *page)
  244. {
  245. if (unlikely(PageTail(page)))
  246. return page->first_page;
  247. return page;
  248. }
  249. static inline int page_count(struct page *page)
  250. {
  251. return atomic_read(&compound_head(page)->_count);
  252. }
  253. static inline void get_page(struct page *page)
  254. {
  255. page = compound_head(page);
  256. VM_BUG_ON(atomic_read(&page->_count) == 0);
  257. atomic_inc(&page->_count);
  258. }
  259. static inline struct page *virt_to_head_page(const void *x)
  260. {
  261. struct page *page = virt_to_page(x);
  262. return compound_head(page);
  263. }
  264. /*
  265. * Setup the page count before being freed into the page allocator for
  266. * the first time (boot or memory hotplug)
  267. */
  268. static inline void init_page_count(struct page *page)
  269. {
  270. atomic_set(&page->_count, 1);
  271. }
  272. void put_page(struct page *page);
  273. void put_pages_list(struct list_head *pages);
  274. void split_page(struct page *page, unsigned int order);
  275. /*
  276. * Compound pages have a destructor function. Provide a
  277. * prototype for that function and accessor functions.
  278. * These are _only_ valid on the head of a PG_compound page.
  279. */
  280. typedef void compound_page_dtor(struct page *);
  281. static inline void set_compound_page_dtor(struct page *page,
  282. compound_page_dtor *dtor)
  283. {
  284. page[1].lru.next = (void *)dtor;
  285. }
  286. static inline compound_page_dtor *get_compound_page_dtor(struct page *page)
  287. {
  288. return (compound_page_dtor *)page[1].lru.next;
  289. }
  290. static inline int compound_order(struct page *page)
  291. {
  292. if (!PageHead(page))
  293. return 0;
  294. return (unsigned long)page[1].lru.prev;
  295. }
  296. static inline void set_compound_order(struct page *page, unsigned long order)
  297. {
  298. page[1].lru.prev = (void *)order;
  299. }
  300. /*
  301. * Multiple processes may "see" the same page. E.g. for untouched
  302. * mappings of /dev/null, all processes see the same page full of
  303. * zeroes, and text pages of executables and shared libraries have
  304. * only one copy in memory, at most, normally.
  305. *
  306. * For the non-reserved pages, page_count(page) denotes a reference count.
  307. * page_count() == 0 means the page is free. page->lru is then used for
  308. * freelist management in the buddy allocator.
  309. * page_count() > 0 means the page has been allocated.
  310. *
  311. * Pages are allocated by the slab allocator in order to provide memory
  312. * to kmalloc and kmem_cache_alloc. In this case, the management of the
  313. * page, and the fields in 'struct page' are the responsibility of mm/slab.c
  314. * unless a particular usage is carefully commented. (the responsibility of
  315. * freeing the kmalloc memory is the caller's, of course).
  316. *
  317. * A page may be used by anyone else who does a __get_free_page().
  318. * In this case, page_count still tracks the references, and should only
  319. * be used through the normal accessor functions. The top bits of page->flags
  320. * and page->virtual store page management information, but all other fields
  321. * are unused and could be used privately, carefully. The management of this
  322. * page is the responsibility of the one who allocated it, and those who have
  323. * subsequently been given references to it.
  324. *
  325. * The other pages (we may call them "pagecache pages") are completely
  326. * managed by the Linux memory manager: I/O, buffers, swapping etc.
  327. * The following discussion applies only to them.
  328. *
  329. * A pagecache page contains an opaque `private' member, which belongs to the
  330. * page's address_space. Usually, this is the address of a circular list of
  331. * the page's disk buffers. PG_private must be set to tell the VM to call
  332. * into the filesystem to release these pages.
  333. *
  334. * A page may belong to an inode's memory mapping. In this case, page->mapping
  335. * is the pointer to the inode, and page->index is the file offset of the page,
  336. * in units of PAGE_CACHE_SIZE.
  337. *
  338. * If pagecache pages are not associated with an inode, they are said to be
  339. * anonymous pages. These may become associated with the swapcache, and in that
  340. * case PG_swapcache is set, and page->private is an offset into the swapcache.
  341. *
  342. * In either case (swapcache or inode backed), the pagecache itself holds one
  343. * reference to the page. Setting PG_private should also increment the
  344. * refcount. The each user mapping also has a reference to the page.
  345. *
  346. * The pagecache pages are stored in a per-mapping radix tree, which is
  347. * rooted at mapping->page_tree, and indexed by offset.
  348. * Where 2.4 and early 2.6 kernels kept dirty/clean pages in per-address_space
  349. * lists, we instead now tag pages as dirty/writeback in the radix tree.
  350. *
  351. * All pagecache pages may be subject to I/O:
  352. * - inode pages may need to be read from disk,
  353. * - inode pages which have been modified and are MAP_SHARED may need
  354. * to be written back to the inode on disk,
  355. * - anonymous pages (including MAP_PRIVATE file mappings) which have been
  356. * modified may need to be swapped out to swap space and (later) to be read
  357. * back into memory.
  358. */
  359. /*
  360. * The zone field is never updated after free_area_init_core()
  361. * sets it, so none of the operations on it need to be atomic.
  362. */
  363. /*
  364. * page->flags layout:
  365. *
  366. * There are three possibilities for how page->flags get
  367. * laid out. The first is for the normal case, without
  368. * sparsemem. The second is for sparsemem when there is
  369. * plenty of space for node and section. The last is when
  370. * we have run out of space and have to fall back to an
  371. * alternate (slower) way of determining the node.
  372. *
  373. * No sparsemem or sparsemem vmemmap: | NODE | ZONE | ... | FLAGS |
  374. * classic sparse with space for node:| SECTION | NODE | ZONE | ... | FLAGS |
  375. * classic sparse no space for node: | SECTION | ZONE | ... | FLAGS |
  376. */
  377. #if defined(CONFIG_SPARSEMEM) && !defined(CONFIG_SPARSEMEM_VMEMMAP)
  378. #define SECTIONS_WIDTH SECTIONS_SHIFT
  379. #else
  380. #define SECTIONS_WIDTH 0
  381. #endif
  382. #define ZONES_WIDTH ZONES_SHIFT
  383. #if SECTIONS_WIDTH+ZONES_WIDTH+NODES_SHIFT <= BITS_PER_LONG - NR_PAGEFLAGS
  384. #define NODES_WIDTH NODES_SHIFT
  385. #else
  386. #ifdef CONFIG_SPARSEMEM_VMEMMAP
  387. #error "Vmemmap: No space for nodes field in page flags"
  388. #endif
  389. #define NODES_WIDTH 0
  390. #endif
  391. /* Page flags: | [SECTION] | [NODE] | ZONE | ... | FLAGS | */
  392. #define SECTIONS_PGOFF ((sizeof(unsigned long)*8) - SECTIONS_WIDTH)
  393. #define NODES_PGOFF (SECTIONS_PGOFF - NODES_WIDTH)
  394. #define ZONES_PGOFF (NODES_PGOFF - ZONES_WIDTH)
  395. /*
  396. * We are going to use the flags for the page to node mapping if its in
  397. * there. This includes the case where there is no node, so it is implicit.
  398. */
  399. #if !(NODES_WIDTH > 0 || NODES_SHIFT == 0)
  400. #define NODE_NOT_IN_PAGE_FLAGS
  401. #endif
  402. #ifndef PFN_SECTION_SHIFT
  403. #define PFN_SECTION_SHIFT 0
  404. #endif
  405. /*
  406. * Define the bit shifts to access each section. For non-existant
  407. * sections we define the shift as 0; that plus a 0 mask ensures
  408. * the compiler will optimise away reference to them.
  409. */
  410. #define SECTIONS_PGSHIFT (SECTIONS_PGOFF * (SECTIONS_WIDTH != 0))
  411. #define NODES_PGSHIFT (NODES_PGOFF * (NODES_WIDTH != 0))
  412. #define ZONES_PGSHIFT (ZONES_PGOFF * (ZONES_WIDTH != 0))
  413. /* NODE:ZONE or SECTION:ZONE is used to ID a zone for the buddy allcator */
  414. #ifdef NODE_NOT_IN_PAGEFLAGS
  415. #define ZONEID_SHIFT (SECTIONS_SHIFT + ZONES_SHIFT)
  416. #define ZONEID_PGOFF ((SECTIONS_PGOFF < ZONES_PGOFF)? \
  417. SECTIONS_PGOFF : ZONES_PGOFF)
  418. #else
  419. #define ZONEID_SHIFT (NODES_SHIFT + ZONES_SHIFT)
  420. #define ZONEID_PGOFF ((NODES_PGOFF < ZONES_PGOFF)? \
  421. NODES_PGOFF : ZONES_PGOFF)
  422. #endif
  423. #define ZONEID_PGSHIFT (ZONEID_PGOFF * (ZONEID_SHIFT != 0))
  424. #if SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > BITS_PER_LONG - NR_PAGEFLAGS
  425. #error SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > BITS_PER_LONG - NR_PAGEFLAGS
  426. #endif
  427. #define ZONES_MASK ((1UL << ZONES_WIDTH) - 1)
  428. #define NODES_MASK ((1UL << NODES_WIDTH) - 1)
  429. #define SECTIONS_MASK ((1UL << SECTIONS_WIDTH) - 1)
  430. #define ZONEID_MASK ((1UL << ZONEID_SHIFT) - 1)
  431. static inline enum zone_type page_zonenum(struct page *page)
  432. {
  433. return (page->flags >> ZONES_PGSHIFT) & ZONES_MASK;
  434. }
  435. /*
  436. * The identification function is only used by the buddy allocator for
  437. * determining if two pages could be buddies. We are not really
  438. * identifying a zone since we could be using a the section number
  439. * id if we have not node id available in page flags.
  440. * We guarantee only that it will return the same value for two
  441. * combinable pages in a zone.
  442. */
  443. static inline int page_zone_id(struct page *page)
  444. {
  445. return (page->flags >> ZONEID_PGSHIFT) & ZONEID_MASK;
  446. }
  447. static inline int zone_to_nid(struct zone *zone)
  448. {
  449. #ifdef CONFIG_NUMA
  450. return zone->node;
  451. #else
  452. return 0;
  453. #endif
  454. }
  455. #ifdef NODE_NOT_IN_PAGE_FLAGS
  456. extern int page_to_nid(struct page *page);
  457. #else
  458. static inline int page_to_nid(struct page *page)
  459. {
  460. return (page->flags >> NODES_PGSHIFT) & NODES_MASK;
  461. }
  462. #endif
  463. static inline struct zone *page_zone(struct page *page)
  464. {
  465. return &NODE_DATA(page_to_nid(page))->node_zones[page_zonenum(page)];
  466. }
  467. #if defined(CONFIG_SPARSEMEM) && !defined(CONFIG_SPARSEMEM_VMEMMAP)
  468. static inline unsigned long page_to_section(struct page *page)
  469. {
  470. return (page->flags >> SECTIONS_PGSHIFT) & SECTIONS_MASK;
  471. }
  472. #endif
  473. static inline void set_page_zone(struct page *page, enum zone_type zone)
  474. {
  475. page->flags &= ~(ZONES_MASK << ZONES_PGSHIFT);
  476. page->flags |= (zone & ZONES_MASK) << ZONES_PGSHIFT;
  477. }
  478. static inline void set_page_node(struct page *page, unsigned long node)
  479. {
  480. page->flags &= ~(NODES_MASK << NODES_PGSHIFT);
  481. page->flags |= (node & NODES_MASK) << NODES_PGSHIFT;
  482. }
  483. static inline void set_page_section(struct page *page, unsigned long section)
  484. {
  485. page->flags &= ~(SECTIONS_MASK << SECTIONS_PGSHIFT);
  486. page->flags |= (section & SECTIONS_MASK) << SECTIONS_PGSHIFT;
  487. }
  488. static inline void set_page_links(struct page *page, enum zone_type zone,
  489. unsigned long node, unsigned long pfn)
  490. {
  491. set_page_zone(page, zone);
  492. set_page_node(page, node);
  493. set_page_section(page, pfn_to_section_nr(pfn));
  494. }
  495. /*
  496. * If a hint addr is less than mmap_min_addr change hint to be as
  497. * low as possible but still greater than mmap_min_addr
  498. */
  499. static inline unsigned long round_hint_to_min(unsigned long hint)
  500. {
  501. #ifdef CONFIG_SECURITY
  502. hint &= PAGE_MASK;
  503. if (((void *)hint != NULL) &&
  504. (hint < mmap_min_addr))
  505. return PAGE_ALIGN(mmap_min_addr);
  506. #endif
  507. return hint;
  508. }
  509. /*
  510. * Some inline functions in vmstat.h depend on page_zone()
  511. */
  512. #include <linux/vmstat.h>
  513. static __always_inline void *lowmem_page_address(struct page *page)
  514. {
  515. return __va(page_to_pfn(page) << PAGE_SHIFT);
  516. }
  517. #if defined(CONFIG_HIGHMEM) && !defined(WANT_PAGE_VIRTUAL)
  518. #define HASHED_PAGE_VIRTUAL
  519. #endif
  520. #if defined(WANT_PAGE_VIRTUAL)
  521. #define page_address(page) ((page)->virtual)
  522. #define set_page_address(page, address) \
  523. do { \
  524. (page)->virtual = (address); \
  525. } while(0)
  526. #define page_address_init() do { } while(0)
  527. #endif
  528. #if defined(HASHED_PAGE_VIRTUAL)
  529. void *page_address(struct page *page);
  530. void set_page_address(struct page *page, void *virtual);
  531. void page_address_init(void);
  532. #endif
  533. #if !defined(HASHED_PAGE_VIRTUAL) && !defined(WANT_PAGE_VIRTUAL)
  534. #define page_address(page) lowmem_page_address(page)
  535. #define set_page_address(page, address) do { } while(0)
  536. #define page_address_init() do { } while(0)
  537. #endif
  538. /*
  539. * On an anonymous page mapped into a user virtual memory area,
  540. * page->mapping points to its anon_vma, not to a struct address_space;
  541. * with the PAGE_MAPPING_ANON bit set to distinguish it.
  542. *
  543. * Please note that, confusingly, "page_mapping" refers to the inode
  544. * address_space which maps the page from disk; whereas "page_mapped"
  545. * refers to user virtual address space into which the page is mapped.
  546. */
  547. #define PAGE_MAPPING_ANON 1
  548. extern struct address_space swapper_space;
  549. static inline struct address_space *page_mapping(struct page *page)
  550. {
  551. struct address_space *mapping = page->mapping;
  552. VM_BUG_ON(PageSlab(page));
  553. #ifdef CONFIG_SWAP
  554. if (unlikely(PageSwapCache(page)))
  555. mapping = &swapper_space;
  556. else
  557. #endif
  558. if (unlikely((unsigned long)mapping & PAGE_MAPPING_ANON))
  559. mapping = NULL;
  560. return mapping;
  561. }
  562. static inline int PageAnon(struct page *page)
  563. {
  564. return ((unsigned long)page->mapping & PAGE_MAPPING_ANON) != 0;
  565. }
  566. /*
  567. * Return the pagecache index of the passed page. Regular pagecache pages
  568. * use ->index whereas swapcache pages use ->private
  569. */
  570. static inline pgoff_t page_index(struct page *page)
  571. {
  572. if (unlikely(PageSwapCache(page)))
  573. return page_private(page);
  574. return page->index;
  575. }
  576. /*
  577. * The atomic page->_mapcount, like _count, starts from -1:
  578. * so that transitions both from it and to it can be tracked,
  579. * using atomic_inc_and_test and atomic_add_negative(-1).
  580. */
  581. static inline void reset_page_mapcount(struct page *page)
  582. {
  583. atomic_set(&(page)->_mapcount, -1);
  584. }
  585. static inline int page_mapcount(struct page *page)
  586. {
  587. return atomic_read(&(page)->_mapcount) + 1;
  588. }
  589. /*
  590. * Return true if this page is mapped into pagetables.
  591. */
  592. static inline int page_mapped(struct page *page)
  593. {
  594. return atomic_read(&(page)->_mapcount) >= 0;
  595. }
  596. /*
  597. * Different kinds of faults, as returned by handle_mm_fault().
  598. * Used to decide whether a process gets delivered SIGBUS or
  599. * just gets major/minor fault counters bumped up.
  600. */
  601. #define VM_FAULT_MINOR 0 /* For backwards compat. Remove me quickly. */
  602. #define VM_FAULT_OOM 0x0001
  603. #define VM_FAULT_SIGBUS 0x0002
  604. #define VM_FAULT_MAJOR 0x0004
  605. #define VM_FAULT_WRITE 0x0008 /* Special case for get_user_pages */
  606. #define VM_FAULT_NOPAGE 0x0100 /* ->fault installed the pte, not return page */
  607. #define VM_FAULT_LOCKED 0x0200 /* ->fault locked the returned page */
  608. #define VM_FAULT_ERROR (VM_FAULT_OOM | VM_FAULT_SIGBUS)
  609. /*
  610. * Can be called by the pagefault handler when it gets a VM_FAULT_OOM.
  611. */
  612. extern void pagefault_out_of_memory(void);
  613. #define offset_in_page(p) ((unsigned long)(p) & ~PAGE_MASK)
  614. extern void show_free_areas(void);
  615. #ifdef CONFIG_SHMEM
  616. extern int shmem_lock(struct file *file, int lock, struct user_struct *user);
  617. #else
  618. static inline int shmem_lock(struct file *file, int lock,
  619. struct user_struct *user)
  620. {
  621. return 0;
  622. }
  623. #endif
  624. struct file *shmem_file_setup(char *name, loff_t size, unsigned long flags);
  625. int shmem_zero_setup(struct vm_area_struct *);
  626. #ifndef CONFIG_MMU
  627. extern unsigned long shmem_get_unmapped_area(struct file *file,
  628. unsigned long addr,
  629. unsigned long len,
  630. unsigned long pgoff,
  631. unsigned long flags);
  632. #endif
  633. extern int can_do_mlock(void);
  634. extern int user_shm_lock(size_t, struct user_struct *);
  635. extern void user_shm_unlock(size_t, struct user_struct *);
  636. /*
  637. * Parameter block passed down to zap_pte_range in exceptional cases.
  638. */
  639. struct zap_details {
  640. struct vm_area_struct *nonlinear_vma; /* Check page->index if set */
  641. struct address_space *check_mapping; /* Check page->mapping if set */
  642. pgoff_t first_index; /* Lowest page->index to unmap */
  643. pgoff_t last_index; /* Highest page->index to unmap */
  644. spinlock_t *i_mmap_lock; /* For unmap_mapping_range: */
  645. unsigned long truncate_count; /* Compare vm_truncate_count */
  646. };
  647. struct page *vm_normal_page(struct vm_area_struct *vma, unsigned long addr,
  648. pte_t pte);
  649. int zap_vma_ptes(struct vm_area_struct *vma, unsigned long address,
  650. unsigned long size);
  651. unsigned long zap_page_range(struct vm_area_struct *vma, unsigned long address,
  652. unsigned long size, struct zap_details *);
  653. unsigned long unmap_vmas(struct mmu_gather **tlb,
  654. struct vm_area_struct *start_vma, unsigned long start_addr,
  655. unsigned long end_addr, unsigned long *nr_accounted,
  656. struct zap_details *);
  657. /**
  658. * mm_walk - callbacks for walk_page_range
  659. * @pgd_entry: if set, called for each non-empty PGD (top-level) entry
  660. * @pud_entry: if set, called for each non-empty PUD (2nd-level) entry
  661. * @pmd_entry: if set, called for each non-empty PMD (3rd-level) entry
  662. * @pte_entry: if set, called for each non-empty PTE (4th-level) entry
  663. * @pte_hole: if set, called for each hole at all levels
  664. *
  665. * (see walk_page_range for more details)
  666. */
  667. struct mm_walk {
  668. int (*pgd_entry)(pgd_t *, unsigned long, unsigned long, struct mm_walk *);
  669. int (*pud_entry)(pud_t *, unsigned long, unsigned long, struct mm_walk *);
  670. int (*pmd_entry)(pmd_t *, unsigned long, unsigned long, struct mm_walk *);
  671. int (*pte_entry)(pte_t *, unsigned long, unsigned long, struct mm_walk *);
  672. int (*pte_hole)(unsigned long, unsigned long, struct mm_walk *);
  673. struct mm_struct *mm;
  674. void *private;
  675. };
  676. int walk_page_range(unsigned long addr, unsigned long end,
  677. struct mm_walk *walk);
  678. void free_pgd_range(struct mmu_gather *tlb, unsigned long addr,
  679. unsigned long end, unsigned long floor, unsigned long ceiling);
  680. int copy_page_range(struct mm_struct *dst, struct mm_struct *src,
  681. struct vm_area_struct *vma);
  682. void unmap_mapping_range(struct address_space *mapping,
  683. loff_t const holebegin, loff_t const holelen, int even_cows);
  684. int follow_phys(struct vm_area_struct *vma, unsigned long address,
  685. unsigned int flags, unsigned long *prot, resource_size_t *phys);
  686. int generic_access_phys(struct vm_area_struct *vma, unsigned long addr,
  687. void *buf, int len, int write);
  688. static inline void unmap_shared_mapping_range(struct address_space *mapping,
  689. loff_t const holebegin, loff_t const holelen)
  690. {
  691. unmap_mapping_range(mapping, holebegin, holelen, 0);
  692. }
  693. extern int vmtruncate(struct inode * inode, loff_t offset);
  694. extern int vmtruncate_range(struct inode * inode, loff_t offset, loff_t end);
  695. #ifdef CONFIG_MMU
  696. extern int handle_mm_fault(struct mm_struct *mm, struct vm_area_struct *vma,
  697. unsigned long address, int write_access);
  698. #else
  699. static inline int handle_mm_fault(struct mm_struct *mm,
  700. struct vm_area_struct *vma, unsigned long address,
  701. int write_access)
  702. {
  703. /* should never happen if there's no MMU */
  704. BUG();
  705. return VM_FAULT_SIGBUS;
  706. }
  707. #endif
  708. extern int make_pages_present(unsigned long addr, unsigned long end);
  709. extern int access_process_vm(struct task_struct *tsk, unsigned long addr, void *buf, int len, int write);
  710. int get_user_pages(struct task_struct *tsk, struct mm_struct *mm, unsigned long start,
  711. int len, int write, int force, struct page **pages, struct vm_area_struct **vmas);
  712. extern int try_to_release_page(struct page * page, gfp_t gfp_mask);
  713. extern void do_invalidatepage(struct page *page, unsigned long offset);
  714. int __set_page_dirty_nobuffers(struct page *page);
  715. int __set_page_dirty_no_writeback(struct page *page);
  716. int redirty_page_for_writepage(struct writeback_control *wbc,
  717. struct page *page);
  718. void account_page_dirtied(struct page *page, struct address_space *mapping);
  719. int set_page_dirty(struct page *page);
  720. int set_page_dirty_lock(struct page *page);
  721. int clear_page_dirty_for_io(struct page *page);
  722. extern unsigned long move_page_tables(struct vm_area_struct *vma,
  723. unsigned long old_addr, struct vm_area_struct *new_vma,
  724. unsigned long new_addr, unsigned long len);
  725. extern unsigned long do_mremap(unsigned long addr,
  726. unsigned long old_len, unsigned long new_len,
  727. unsigned long flags, unsigned long new_addr);
  728. extern int mprotect_fixup(struct vm_area_struct *vma,
  729. struct vm_area_struct **pprev, unsigned long start,
  730. unsigned long end, unsigned long newflags);
  731. /*
  732. * get_user_pages_fast provides equivalent functionality to get_user_pages,
  733. * operating on current and current->mm (force=0 and doesn't return any vmas).
  734. *
  735. * get_user_pages_fast may take mmap_sem and page tables, so no assumptions
  736. * can be made about locking. get_user_pages_fast is to be implemented in a
  737. * way that is advantageous (vs get_user_pages()) when the user memory area is
  738. * already faulted in and present in ptes. However if the pages have to be
  739. * faulted in, it may turn out to be slightly slower).
  740. */
  741. int get_user_pages_fast(unsigned long start, int nr_pages, int write,
  742. struct page **pages);
  743. /*
  744. * A callback you can register to apply pressure to ageable caches.
  745. *
  746. * 'shrink' is passed a count 'nr_to_scan' and a 'gfpmask'. It should
  747. * look through the least-recently-used 'nr_to_scan' entries and
  748. * attempt to free them up. It should return the number of objects
  749. * which remain in the cache. If it returns -1, it means it cannot do
  750. * any scanning at this time (eg. there is a risk of deadlock).
  751. *
  752. * The 'gfpmask' refers to the allocation we are currently trying to
  753. * fulfil.
  754. *
  755. * Note that 'shrink' will be passed nr_to_scan == 0 when the VM is
  756. * querying the cache size, so a fastpath for that case is appropriate.
  757. */
  758. struct shrinker {
  759. int (*shrink)(int nr_to_scan, gfp_t gfp_mask);
  760. int seeks; /* seeks to recreate an obj */
  761. /* These are for internal use */
  762. struct list_head list;
  763. long nr; /* objs pending delete */
  764. };
  765. #define DEFAULT_SEEKS 2 /* A good number if you don't know better. */
  766. extern void register_shrinker(struct shrinker *);
  767. extern void unregister_shrinker(struct shrinker *);
  768. int vma_wants_writenotify(struct vm_area_struct *vma);
  769. extern pte_t *get_locked_pte(struct mm_struct *mm, unsigned long addr, spinlock_t **ptl);
  770. #ifdef __PAGETABLE_PUD_FOLDED
  771. static inline int __pud_alloc(struct mm_struct *mm, pgd_t *pgd,
  772. unsigned long address)
  773. {
  774. return 0;
  775. }
  776. #else
  777. int __pud_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address);
  778. #endif
  779. #ifdef __PAGETABLE_PMD_FOLDED
  780. static inline int __pmd_alloc(struct mm_struct *mm, pud_t *pud,
  781. unsigned long address)
  782. {
  783. return 0;
  784. }
  785. #else
  786. int __pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address);
  787. #endif
  788. int __pte_alloc(struct mm_struct *mm, pmd_t *pmd, unsigned long address);
  789. int __pte_alloc_kernel(pmd_t *pmd, unsigned long address);
  790. /*
  791. * The following ifdef needed to get the 4level-fixup.h header to work.
  792. * Remove it when 4level-fixup.h has been removed.
  793. */
  794. #if defined(CONFIG_MMU) && !defined(__ARCH_HAS_4LEVEL_HACK)
  795. static inline pud_t *pud_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address)
  796. {
  797. return (unlikely(pgd_none(*pgd)) && __pud_alloc(mm, pgd, address))?
  798. NULL: pud_offset(pgd, address);
  799. }
  800. static inline pmd_t *pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address)
  801. {
  802. return (unlikely(pud_none(*pud)) && __pmd_alloc(mm, pud, address))?
  803. NULL: pmd_offset(pud, address);
  804. }
  805. #endif /* CONFIG_MMU && !__ARCH_HAS_4LEVEL_HACK */
  806. #if USE_SPLIT_PTLOCKS
  807. /*
  808. * We tuck a spinlock to guard each pagetable page into its struct page,
  809. * at page->private, with BUILD_BUG_ON to make sure that this will not
  810. * overflow into the next struct page (as it might with DEBUG_SPINLOCK).
  811. * When freeing, reset page->mapping so free_pages_check won't complain.
  812. */
  813. #define __pte_lockptr(page) &((page)->ptl)
  814. #define pte_lock_init(_page) do { \
  815. spin_lock_init(__pte_lockptr(_page)); \
  816. } while (0)
  817. #define pte_lock_deinit(page) ((page)->mapping = NULL)
  818. #define pte_lockptr(mm, pmd) ({(void)(mm); __pte_lockptr(pmd_page(*(pmd)));})
  819. #else /* !USE_SPLIT_PTLOCKS */
  820. /*
  821. * We use mm->page_table_lock to guard all pagetable pages of the mm.
  822. */
  823. #define pte_lock_init(page) do {} while (0)
  824. #define pte_lock_deinit(page) do {} while (0)
  825. #define pte_lockptr(mm, pmd) ({(void)(pmd); &(mm)->page_table_lock;})
  826. #endif /* USE_SPLIT_PTLOCKS */
  827. static inline void pgtable_page_ctor(struct page *page)
  828. {
  829. pte_lock_init(page);
  830. inc_zone_page_state(page, NR_PAGETABLE);
  831. }
  832. static inline void pgtable_page_dtor(struct page *page)
  833. {
  834. pte_lock_deinit(page);
  835. dec_zone_page_state(page, NR_PAGETABLE);
  836. }
  837. #define pte_offset_map_lock(mm, pmd, address, ptlp) \
  838. ({ \
  839. spinlock_t *__ptl = pte_lockptr(mm, pmd); \
  840. pte_t *__pte = pte_offset_map(pmd, address); \
  841. *(ptlp) = __ptl; \
  842. spin_lock(__ptl); \
  843. __pte; \
  844. })
  845. #define pte_unmap_unlock(pte, ptl) do { \
  846. spin_unlock(ptl); \
  847. pte_unmap(pte); \
  848. } while (0)
  849. #define pte_alloc_map(mm, pmd, address) \
  850. ((unlikely(!pmd_present(*(pmd))) && __pte_alloc(mm, pmd, address))? \
  851. NULL: pte_offset_map(pmd, address))
  852. #define pte_alloc_map_lock(mm, pmd, address, ptlp) \
  853. ((unlikely(!pmd_present(*(pmd))) && __pte_alloc(mm, pmd, address))? \
  854. NULL: pte_offset_map_lock(mm, pmd, address, ptlp))
  855. #define pte_alloc_kernel(pmd, address) \
  856. ((unlikely(!pmd_present(*(pmd))) && __pte_alloc_kernel(pmd, address))? \
  857. NULL: pte_offset_kernel(pmd, address))
  858. extern void free_area_init(unsigned long * zones_size);
  859. extern void free_area_init_node(int nid, unsigned long * zones_size,
  860. unsigned long zone_start_pfn, unsigned long *zholes_size);
  861. #ifdef CONFIG_ARCH_POPULATES_NODE_MAP
  862. /*
  863. * With CONFIG_ARCH_POPULATES_NODE_MAP set, an architecture may initialise its
  864. * zones, allocate the backing mem_map and account for memory holes in a more
  865. * architecture independent manner. This is a substitute for creating the
  866. * zone_sizes[] and zholes_size[] arrays and passing them to
  867. * free_area_init_node()
  868. *
  869. * An architecture is expected to register range of page frames backed by
  870. * physical memory with add_active_range() before calling
  871. * free_area_init_nodes() passing in the PFN each zone ends at. At a basic
  872. * usage, an architecture is expected to do something like
  873. *
  874. * unsigned long max_zone_pfns[MAX_NR_ZONES] = {max_dma, max_normal_pfn,
  875. * max_highmem_pfn};
  876. * for_each_valid_physical_page_range()
  877. * add_active_range(node_id, start_pfn, end_pfn)
  878. * free_area_init_nodes(max_zone_pfns);
  879. *
  880. * If the architecture guarantees that there are no holes in the ranges
  881. * registered with add_active_range(), free_bootmem_active_regions()
  882. * will call free_bootmem_node() for each registered physical page range.
  883. * Similarly sparse_memory_present_with_active_regions() calls
  884. * memory_present() for each range when SPARSEMEM is enabled.
  885. *
  886. * See mm/page_alloc.c for more information on each function exposed by
  887. * CONFIG_ARCH_POPULATES_NODE_MAP
  888. */
  889. extern void free_area_init_nodes(unsigned long *max_zone_pfn);
  890. extern void add_active_range(unsigned int nid, unsigned long start_pfn,
  891. unsigned long end_pfn);
  892. extern void remove_active_range(unsigned int nid, unsigned long start_pfn,
  893. unsigned long end_pfn);
  894. extern void push_node_boundaries(unsigned int nid, unsigned long start_pfn,
  895. unsigned long end_pfn);
  896. extern void remove_all_active_ranges(void);
  897. extern unsigned long absent_pages_in_range(unsigned long start_pfn,
  898. unsigned long end_pfn);
  899. extern void get_pfn_range_for_nid(unsigned int nid,
  900. unsigned long *start_pfn, unsigned long *end_pfn);
  901. extern unsigned long find_min_pfn_with_active_regions(void);
  902. extern void free_bootmem_with_active_regions(int nid,
  903. unsigned long max_low_pfn);
  904. typedef int (*work_fn_t)(unsigned long, unsigned long, void *);
  905. extern void work_with_active_regions(int nid, work_fn_t work_fn, void *data);
  906. extern void sparse_memory_present_with_active_regions(int nid);
  907. #endif /* CONFIG_ARCH_POPULATES_NODE_MAP */
  908. #if !defined(CONFIG_ARCH_POPULATES_NODE_MAP) && \
  909. !defined(CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID)
  910. static inline int __early_pfn_to_nid(unsigned long pfn)
  911. {
  912. return 0;
  913. }
  914. #else
  915. /* please see mm/page_alloc.c */
  916. extern int __meminit early_pfn_to_nid(unsigned long pfn);
  917. #ifdef CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID
  918. /* there is a per-arch backend function. */
  919. extern int __meminit __early_pfn_to_nid(unsigned long pfn);
  920. #endif /* CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID */
  921. #endif
  922. extern void set_dma_reserve(unsigned long new_dma_reserve);
  923. extern void memmap_init_zone(unsigned long, int, unsigned long,
  924. unsigned long, enum memmap_context);
  925. extern void setup_per_zone_pages_min(void);
  926. extern void mem_init(void);
  927. extern void __init mmap_init(void);
  928. extern void show_mem(void);
  929. extern void si_meminfo(struct sysinfo * val);
  930. extern void si_meminfo_node(struct sysinfo *val, int nid);
  931. extern int after_bootmem;
  932. #ifdef CONFIG_NUMA
  933. extern void setup_per_cpu_pageset(void);
  934. #else
  935. static inline void setup_per_cpu_pageset(void) {}
  936. #endif
  937. /* nommu.c */
  938. extern atomic_long_t mmap_pages_allocated;
  939. /* prio_tree.c */
  940. void vma_prio_tree_add(struct vm_area_struct *, struct vm_area_struct *old);
  941. void vma_prio_tree_insert(struct vm_area_struct *, struct prio_tree_root *);
  942. void vma_prio_tree_remove(struct vm_area_struct *, struct prio_tree_root *);
  943. struct vm_area_struct *vma_prio_tree_next(struct vm_area_struct *vma,
  944. struct prio_tree_iter *iter);
  945. #define vma_prio_tree_foreach(vma, iter, root, begin, end) \
  946. for (prio_tree_iter_init(iter, root, begin, end), vma = NULL; \
  947. (vma = vma_prio_tree_next(vma, iter)); )
  948. static inline void vma_nonlinear_insert(struct vm_area_struct *vma,
  949. struct list_head *list)
  950. {
  951. vma->shared.vm_set.parent = NULL;
  952. list_add_tail(&vma->shared.vm_set.list, list);
  953. }
  954. /* mmap.c */
  955. extern int __vm_enough_memory(struct mm_struct *mm, long pages, int cap_sys_admin);
  956. extern void vma_adjust(struct vm_area_struct *vma, unsigned long start,
  957. unsigned long end, pgoff_t pgoff, struct vm_area_struct *insert);
  958. extern struct vm_area_struct *vma_merge(struct mm_struct *,
  959. struct vm_area_struct *prev, unsigned long addr, unsigned long end,
  960. unsigned long vm_flags, struct anon_vma *, struct file *, pgoff_t,
  961. struct mempolicy *);
  962. extern struct anon_vma *find_mergeable_anon_vma(struct vm_area_struct *);
  963. extern int split_vma(struct mm_struct *,
  964. struct vm_area_struct *, unsigned long addr, int new_below);
  965. extern int insert_vm_struct(struct mm_struct *, struct vm_area_struct *);
  966. extern void __vma_link_rb(struct mm_struct *, struct vm_area_struct *,
  967. struct rb_node **, struct rb_node *);
  968. extern void unlink_file_vma(struct vm_area_struct *);
  969. extern struct vm_area_struct *copy_vma(struct vm_area_struct **,
  970. unsigned long addr, unsigned long len, pgoff_t pgoff);
  971. extern void exit_mmap(struct mm_struct *);
  972. extern int mm_take_all_locks(struct mm_struct *mm);
  973. extern void mm_drop_all_locks(struct mm_struct *mm);
  974. #ifdef CONFIG_PROC_FS
  975. /* From fs/proc/base.c. callers must _not_ hold the mm's exe_file_lock */
  976. extern void added_exe_file_vma(struct mm_struct *mm);
  977. extern void removed_exe_file_vma(struct mm_struct *mm);
  978. #else
  979. static inline void added_exe_file_vma(struct mm_struct *mm)
  980. {}
  981. static inline void removed_exe_file_vma(struct mm_struct *mm)
  982. {}
  983. #endif /* CONFIG_PROC_FS */
  984. extern int may_expand_vm(struct mm_struct *mm, unsigned long npages);
  985. extern int install_special_mapping(struct mm_struct *mm,
  986. unsigned long addr, unsigned long len,
  987. unsigned long flags, struct page **pages);
  988. extern unsigned long get_unmapped_area(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
  989. extern unsigned long do_mmap_pgoff(struct file *file, unsigned long addr,
  990. unsigned long len, unsigned long prot,
  991. unsigned long flag, unsigned long pgoff);
  992. extern unsigned long mmap_region(struct file *file, unsigned long addr,
  993. unsigned long len, unsigned long flags,
  994. unsigned int vm_flags, unsigned long pgoff);
  995. static inline unsigned long do_mmap(struct file *file, unsigned long addr,
  996. unsigned long len, unsigned long prot,
  997. unsigned long flag, unsigned long offset)
  998. {
  999. unsigned long ret = -EINVAL;
  1000. if ((offset + PAGE_ALIGN(len)) < offset)
  1001. goto out;
  1002. if (!(offset & ~PAGE_MASK))
  1003. ret = do_mmap_pgoff(file, addr, len, prot, flag, offset >> PAGE_SHIFT);
  1004. out:
  1005. return ret;
  1006. }
  1007. extern int do_munmap(struct mm_struct *, unsigned long, size_t);
  1008. extern unsigned long do_brk(unsigned long, unsigned long);
  1009. /* filemap.c */
  1010. extern unsigned long page_unuse(struct page *);
  1011. extern void truncate_inode_pages(struct address_space *, loff_t);
  1012. extern void truncate_inode_pages_range(struct address_space *,
  1013. loff_t lstart, loff_t lend);
  1014. /* generic vm_area_ops exported for stackable file systems */
  1015. extern int filemap_fault(struct vm_area_struct *, struct vm_fault *);
  1016. /* mm/page-writeback.c */
  1017. int write_one_page(struct page *page, int wait);
  1018. void task_dirty_inc(struct task_struct *tsk);
  1019. /* readahead.c */
  1020. #define VM_MAX_READAHEAD 128 /* kbytes */
  1021. #define VM_MIN_READAHEAD 16 /* kbytes (includes current page) */
  1022. int do_page_cache_readahead(struct address_space *mapping, struct file *filp,
  1023. pgoff_t offset, unsigned long nr_to_read);
  1024. int force_page_cache_readahead(struct address_space *mapping, struct file *filp,
  1025. pgoff_t offset, unsigned long nr_to_read);
  1026. void page_cache_sync_readahead(struct address_space *mapping,
  1027. struct file_ra_state *ra,
  1028. struct file *filp,
  1029. pgoff_t offset,
  1030. unsigned long size);
  1031. void page_cache_async_readahead(struct address_space *mapping,
  1032. struct file_ra_state *ra,
  1033. struct file *filp,
  1034. struct page *pg,
  1035. pgoff_t offset,
  1036. unsigned long size);
  1037. unsigned long max_sane_readahead(unsigned long nr);
  1038. /* Do stack extension */
  1039. extern int expand_stack(struct vm_area_struct *vma, unsigned long address);
  1040. #ifdef CONFIG_IA64
  1041. extern int expand_upwards(struct vm_area_struct *vma, unsigned long address);
  1042. #endif
  1043. extern int expand_stack_downwards(struct vm_area_struct *vma,
  1044. unsigned long address);
  1045. /* Look up the first VMA which satisfies addr < vm_end, NULL if none. */
  1046. extern struct vm_area_struct * find_vma(struct mm_struct * mm, unsigned long addr);
  1047. extern struct vm_area_struct * find_vma_prev(struct mm_struct * mm, unsigned long addr,
  1048. struct vm_area_struct **pprev);
  1049. /* Look up the first VMA which intersects the interval start_addr..end_addr-1,
  1050. NULL if none. Assume start_addr < end_addr. */
  1051. static inline struct vm_area_struct * find_vma_intersection(struct mm_struct * mm, unsigned long start_addr, unsigned long end_addr)
  1052. {
  1053. struct vm_area_struct * vma = find_vma(mm,start_addr);
  1054. if (vma && end_addr <= vma->vm_start)
  1055. vma = NULL;
  1056. return vma;
  1057. }
  1058. static inline unsigned long vma_pages(struct vm_area_struct *vma)
  1059. {
  1060. return (vma->vm_end - vma->vm_start) >> PAGE_SHIFT;
  1061. }
  1062. pgprot_t vm_get_page_prot(unsigned long vm_flags);
  1063. struct vm_area_struct *find_extend_vma(struct mm_struct *, unsigned long addr);
  1064. int remap_pfn_range(struct vm_area_struct *, unsigned long addr,
  1065. unsigned long pfn, unsigned long size, pgprot_t);
  1066. int vm_insert_page(struct vm_area_struct *, unsigned long addr, struct page *);
  1067. int vm_insert_pfn(struct vm_area_struct *vma, unsigned long addr,
  1068. unsigned long pfn);
  1069. int vm_insert_mixed(struct vm_area_struct *vma, unsigned long addr,
  1070. unsigned long pfn);
  1071. struct page *follow_page(struct vm_area_struct *, unsigned long address,
  1072. unsigned int foll_flags);
  1073. #define FOLL_WRITE 0x01 /* check pte is writable */
  1074. #define FOLL_TOUCH 0x02 /* mark page accessed */
  1075. #define FOLL_GET 0x04 /* do get_page on page */
  1076. #define FOLL_ANON 0x08 /* give ZERO_PAGE if no pgtable */
  1077. typedef int (*pte_fn_t)(pte_t *pte, pgtable_t token, unsigned long addr,
  1078. void *data);
  1079. extern int apply_to_page_range(struct mm_struct *mm, unsigned long address,
  1080. unsigned long size, pte_fn_t fn, void *data);
  1081. #ifdef CONFIG_PROC_FS
  1082. void vm_stat_account(struct mm_struct *, unsigned long, struct file *, long);
  1083. #else
  1084. static inline void vm_stat_account(struct mm_struct *mm,
  1085. unsigned long flags, struct file *file, long pages)
  1086. {
  1087. }
  1088. #endif /* CONFIG_PROC_FS */
  1089. #ifdef CONFIG_DEBUG_PAGEALLOC
  1090. extern int debug_pagealloc_enabled;
  1091. extern void kernel_map_pages(struct page *page, int numpages, int enable);
  1092. static inline void enable_debug_pagealloc(void)
  1093. {
  1094. debug_pagealloc_enabled = 1;
  1095. }
  1096. #ifdef CONFIG_HIBERNATION
  1097. extern bool kernel_page_present(struct page *page);
  1098. #endif /* CONFIG_HIBERNATION */
  1099. #else
  1100. static inline void
  1101. kernel_map_pages(struct page *page, int numpages, int enable) {}
  1102. static inline void enable_debug_pagealloc(void)
  1103. {
  1104. }
  1105. #ifdef CONFIG_HIBERNATION
  1106. static inline bool kernel_page_present(struct page *page) { return true; }
  1107. #endif /* CONFIG_HIBERNATION */
  1108. #endif
  1109. extern struct vm_area_struct *get_gate_vma(struct task_struct *tsk);
  1110. #ifdef __HAVE_ARCH_GATE_AREA
  1111. int in_gate_area_no_task(unsigned long addr);
  1112. int in_gate_area(struct task_struct *task, unsigned long addr);
  1113. #else
  1114. int in_gate_area_no_task(unsigned long addr);
  1115. #define in_gate_area(task, addr) ({(void)task; in_gate_area_no_task(addr);})
  1116. #endif /* __HAVE_ARCH_GATE_AREA */
  1117. int drop_caches_sysctl_handler(struct ctl_table *, int, struct file *,
  1118. void __user *, size_t *, loff_t *);
  1119. unsigned long shrink_slab(unsigned long scanned, gfp_t gfp_mask,
  1120. unsigned long lru_pages);
  1121. #ifndef CONFIG_MMU
  1122. #define randomize_va_space 0
  1123. #else
  1124. extern int randomize_va_space;
  1125. #endif
  1126. const char * arch_vma_name(struct vm_area_struct *vma);
  1127. void print_vma_addr(char *prefix, unsigned long rip);
  1128. struct page *sparse_mem_map_populate(unsigned long pnum, int nid);
  1129. pgd_t *vmemmap_pgd_populate(unsigned long addr, int node);
  1130. pud_t *vmemmap_pud_populate(pgd_t *pgd, unsigned long addr, int node);
  1131. pmd_t *vmemmap_pmd_populate(pud_t *pud, unsigned long addr, int node);
  1132. pte_t *vmemmap_pte_populate(pmd_t *pmd, unsigned long addr, int node);
  1133. void *vmemmap_alloc_block(unsigned long size, int node);
  1134. void vmemmap_verify(pte_t *, int, unsigned long, unsigned long);
  1135. int vmemmap_populate_basepages(struct page *start_page,
  1136. unsigned long pages, int node);
  1137. int vmemmap_populate(struct page *start_page, unsigned long pages, int node);
  1138. void vmemmap_populate_print_last(void);
  1139. extern void *alloc_locked_buffer(size_t size);
  1140. extern void free_locked_buffer(void *buffer, size_t size);
  1141. extern void refund_locked_buffer_memory(struct mm_struct *mm, size_t size);
  1142. #endif /* __KERNEL__ */
  1143. #endif /* _LINUX_MM_H */