mm.h 40 KB

12345678910111213141516171819202122232425262728293031323334353637383940414243444546474849505152535455565758596061626364656667686970717273747576777879808182838485868788899091929394959697989910010110210310410510610710810911011111211311411511611711811912012112212312412512612712812913013113213313413513613713813914014114214314414514614714814915015115215315415515615715815916016116216316416516616716816917017117217317417517617717817918018118218318418518618718818919019119219319419519619719819920020120220320420520620720820921021121221321421521621721821922022122222322422522622722822923023123223323423523623723823924024124224324424524624724824925025125225325425525625725825926026126226326426526626726826927027127227327427527627727827928028128228328428528628728828929029129229329429529629729829930030130230330430530630730830931031131231331431531631731831932032132232332432532632732832933033133233333433533633733833934034134234334434534634734834935035135235335435535635735835936036136236336436536636736836937037137237337437537637737837938038138238338438538638738838939039139239339439539639739839940040140240340440540640740840941041141241341441541641741841942042142242342442542642742842943043143243343443543643743843944044144244344444544644744844945045145245345445545645745845946046146246346446546646746846947047147247347447547647747847948048148248348448548648748848949049149249349449549649749849950050150250350450550650750850951051151251351451551651751851952052152252352452552652752852953053153253353453553653753853954054154254354454554654754854955055155255355455555655755855956056156256356456556656756856957057157257357457557657757857958058158258358458558658758858959059159259359459559659759859960060160260360460560660760860961061161261361461561661761861962062162262362462562662762862963063163263363463563663763863964064164264364464564664764864965065165265365465565665765865966066166266366466566666766866967067167267367467567667767867968068168268368468568668768868969069169269369469569669769869970070170270370470570670770870971071171271371471571671771871972072172272372472572672772872973073173273373473573673773873974074174274374474574674774874975075175275375475575675775875976076176276376476576676776876977077177277377477577677777877978078178278378478578678778878979079179279379479579679779879980080180280380480580680780880981081181281381481581681781881982082182282382482582682782882983083183283383483583683783883984084184284384484584684784884985085185285385485585685785885986086186286386486586686786886987087187287387487587687787887988088188288388488588688788888989089189289389489589689789889990090190290390490590690790890991091191291391491591691791891992092192292392492592692792892993093193293393493593693793893994094194294394494594694794894995095195295395495595695795895996096196296396496596696796896997097197297397497597697797897998098198298398498598698798898999099199299399499599699799899910001001100210031004100510061007100810091010101110121013101410151016101710181019102010211022102310241025102610271028102910301031103210331034103510361037103810391040104110421043104410451046104710481049105010511052105310541055105610571058105910601061106210631064106510661067106810691070107110721073107410751076107710781079108010811082108310841085108610871088108910901091109210931094109510961097109810991100110111021103110411051106110711081109111011111112111311141115111611171118111911201121112211231124112511261127112811291130113111321133113411351136113711381139114011411142114311441145114611471148114911501151115211531154115511561157115811591160116111621163116411651166116711681169117011711172
  1. #ifndef _LINUX_MM_H
  2. #define _LINUX_MM_H
  3. #include <linux/sched.h>
  4. #include <linux/errno.h>
  5. #include <linux/capability.h>
  6. #ifdef __KERNEL__
  7. #include <linux/gfp.h>
  8. #include <linux/list.h>
  9. #include <linux/mmzone.h>
  10. #include <linux/rbtree.h>
  11. #include <linux/prio_tree.h>
  12. #include <linux/fs.h>
  13. #include <linux/mutex.h>
  14. #include <linux/debug_locks.h>
  15. #include <linux/backing-dev.h>
  16. #include <linux/mm_types.h>
  17. struct mempolicy;
  18. struct anon_vma;
  19. #ifndef CONFIG_DISCONTIGMEM /* Don't use mapnrs, do it properly */
  20. extern unsigned long max_mapnr;
  21. #endif
  22. extern unsigned long num_physpages;
  23. extern void * high_memory;
  24. extern unsigned long vmalloc_earlyreserve;
  25. extern int page_cluster;
  26. #ifdef CONFIG_SYSCTL
  27. extern int sysctl_legacy_va_layout;
  28. #else
  29. #define sysctl_legacy_va_layout 0
  30. #endif
  31. #include <asm/page.h>
  32. #include <asm/pgtable.h>
  33. #include <asm/processor.h>
  34. #define nth_page(page,n) pfn_to_page(page_to_pfn((page)) + (n))
  35. /*
  36. * Linux kernel virtual memory manager primitives.
  37. * The idea being to have a "virtual" mm in the same way
  38. * we have a virtual fs - giving a cleaner interface to the
  39. * mm details, and allowing different kinds of memory mappings
  40. * (from shared memory to executable loading to arbitrary
  41. * mmap() functions).
  42. */
  43. /*
  44. * This struct defines a memory VMM memory area. There is one of these
  45. * per VM-area/task. A VM area is any part of the process virtual memory
  46. * space that has a special rule for the page-fault handlers (ie a shared
  47. * library, the executable area etc).
  48. */
  49. struct vm_area_struct {
  50. struct mm_struct * vm_mm; /* The address space we belong to. */
  51. unsigned long vm_start; /* Our start address within vm_mm. */
  52. unsigned long vm_end; /* The first byte after our end address
  53. within vm_mm. */
  54. /* linked list of VM areas per task, sorted by address */
  55. struct vm_area_struct *vm_next;
  56. pgprot_t vm_page_prot; /* Access permissions of this VMA. */
  57. unsigned long vm_flags; /* Flags, listed below. */
  58. struct rb_node vm_rb;
  59. /*
  60. * For areas with an address space and backing store,
  61. * linkage into the address_space->i_mmap prio tree, or
  62. * linkage to the list of like vmas hanging off its node, or
  63. * linkage of vma in the address_space->i_mmap_nonlinear list.
  64. */
  65. union {
  66. struct {
  67. struct list_head list;
  68. void *parent; /* aligns with prio_tree_node parent */
  69. struct vm_area_struct *head;
  70. } vm_set;
  71. struct raw_prio_tree_node prio_tree_node;
  72. } shared;
  73. /*
  74. * A file's MAP_PRIVATE vma can be in both i_mmap tree and anon_vma
  75. * list, after a COW of one of the file pages. A MAP_SHARED vma
  76. * can only be in the i_mmap tree. An anonymous MAP_PRIVATE, stack
  77. * or brk vma (with NULL file) can only be in an anon_vma list.
  78. */
  79. struct list_head anon_vma_node; /* Serialized by anon_vma->lock */
  80. struct anon_vma *anon_vma; /* Serialized by page_table_lock */
  81. /* Function pointers to deal with this struct. */
  82. struct vm_operations_struct * vm_ops;
  83. /* Information about our backing store: */
  84. unsigned long vm_pgoff; /* Offset (within vm_file) in PAGE_SIZE
  85. units, *not* PAGE_CACHE_SIZE */
  86. struct file * vm_file; /* File we map to (can be NULL). */
  87. void * vm_private_data; /* was vm_pte (shared mem) */
  88. unsigned long vm_truncate_count;/* truncate_count or restart_addr */
  89. #ifndef CONFIG_MMU
  90. atomic_t vm_usage; /* refcount (VMAs shared if !MMU) */
  91. #endif
  92. #ifdef CONFIG_NUMA
  93. struct mempolicy *vm_policy; /* NUMA policy for the VMA */
  94. #endif
  95. };
  96. extern struct kmem_cache *vm_area_cachep;
  97. /*
  98. * This struct defines the per-mm list of VMAs for uClinux. If CONFIG_MMU is
  99. * disabled, then there's a single shared list of VMAs maintained by the
  100. * system, and mm's subscribe to these individually
  101. */
  102. struct vm_list_struct {
  103. struct vm_list_struct *next;
  104. struct vm_area_struct *vma;
  105. };
  106. #ifndef CONFIG_MMU
  107. extern struct rb_root nommu_vma_tree;
  108. extern struct rw_semaphore nommu_vma_sem;
  109. extern unsigned int kobjsize(const void *objp);
  110. #endif
  111. /*
  112. * vm_flags..
  113. */
  114. #define VM_READ 0x00000001 /* currently active flags */
  115. #define VM_WRITE 0x00000002
  116. #define VM_EXEC 0x00000004
  117. #define VM_SHARED 0x00000008
  118. /* mprotect() hardcodes VM_MAYREAD >> 4 == VM_READ, and so for r/w/x bits. */
  119. #define VM_MAYREAD 0x00000010 /* limits for mprotect() etc */
  120. #define VM_MAYWRITE 0x00000020
  121. #define VM_MAYEXEC 0x00000040
  122. #define VM_MAYSHARE 0x00000080
  123. #define VM_GROWSDOWN 0x00000100 /* general info on the segment */
  124. #define VM_GROWSUP 0x00000200
  125. #define VM_PFNMAP 0x00000400 /* Page-ranges managed without "struct page", just pure PFN */
  126. #define VM_DENYWRITE 0x00000800 /* ETXTBSY on write attempts.. */
  127. #define VM_EXECUTABLE 0x00001000
  128. #define VM_LOCKED 0x00002000
  129. #define VM_IO 0x00004000 /* Memory mapped I/O or similar */
  130. /* Used by sys_madvise() */
  131. #define VM_SEQ_READ 0x00008000 /* App will access data sequentially */
  132. #define VM_RAND_READ 0x00010000 /* App will not benefit from clustered reads */
  133. #define VM_DONTCOPY 0x00020000 /* Do not copy this vma on fork */
  134. #define VM_DONTEXPAND 0x00040000 /* Cannot expand with mremap() */
  135. #define VM_RESERVED 0x00080000 /* Count as reserved_vm like IO */
  136. #define VM_ACCOUNT 0x00100000 /* Is a VM accounted object */
  137. #define VM_HUGETLB 0x00400000 /* Huge TLB Page VM */
  138. #define VM_NONLINEAR 0x00800000 /* Is non-linear (remap_file_pages) */
  139. #define VM_MAPPED_COPY 0x01000000 /* T if mapped copy of data (nommu mmap) */
  140. #define VM_INSERTPAGE 0x02000000 /* The vma has had "vm_insert_page()" done on it */
  141. #define VM_ALWAYSDUMP 0x04000000 /* Always include in core dumps */
  142. #ifndef VM_STACK_DEFAULT_FLAGS /* arch can override this */
  143. #define VM_STACK_DEFAULT_FLAGS VM_DATA_DEFAULT_FLAGS
  144. #endif
  145. #ifdef CONFIG_STACK_GROWSUP
  146. #define VM_STACK_FLAGS (VM_GROWSUP | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT)
  147. #else
  148. #define VM_STACK_FLAGS (VM_GROWSDOWN | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT)
  149. #endif
  150. #define VM_READHINTMASK (VM_SEQ_READ | VM_RAND_READ)
  151. #define VM_ClearReadHint(v) (v)->vm_flags &= ~VM_READHINTMASK
  152. #define VM_NormalReadHint(v) (!((v)->vm_flags & VM_READHINTMASK))
  153. #define VM_SequentialReadHint(v) ((v)->vm_flags & VM_SEQ_READ)
  154. #define VM_RandomReadHint(v) ((v)->vm_flags & VM_RAND_READ)
  155. /*
  156. * mapping from the currently active vm_flags protection bits (the
  157. * low four bits) to a page protection mask..
  158. */
  159. extern pgprot_t protection_map[16];
  160. /*
  161. * These are the virtual MM functions - opening of an area, closing and
  162. * unmapping it (needed to keep files on disk up-to-date etc), pointer
  163. * to the functions called when a no-page or a wp-page exception occurs.
  164. */
  165. struct vm_operations_struct {
  166. void (*open)(struct vm_area_struct * area);
  167. void (*close)(struct vm_area_struct * area);
  168. struct page * (*nopage)(struct vm_area_struct * area, unsigned long address, int *type);
  169. unsigned long (*nopfn)(struct vm_area_struct * area, unsigned long address);
  170. int (*populate)(struct vm_area_struct * area, unsigned long address, unsigned long len, pgprot_t prot, unsigned long pgoff, int nonblock);
  171. /* notification that a previously read-only page is about to become
  172. * writable, if an error is returned it will cause a SIGBUS */
  173. int (*page_mkwrite)(struct vm_area_struct *vma, struct page *page);
  174. #ifdef CONFIG_NUMA
  175. int (*set_policy)(struct vm_area_struct *vma, struct mempolicy *new);
  176. struct mempolicy *(*get_policy)(struct vm_area_struct *vma,
  177. unsigned long addr);
  178. int (*migrate)(struct vm_area_struct *vma, const nodemask_t *from,
  179. const nodemask_t *to, unsigned long flags);
  180. #endif
  181. };
  182. struct mmu_gather;
  183. struct inode;
  184. #define page_private(page) ((page)->private)
  185. #define set_page_private(page, v) ((page)->private = (v))
  186. /*
  187. * FIXME: take this include out, include page-flags.h in
  188. * files which need it (119 of them)
  189. */
  190. #include <linux/page-flags.h>
  191. #ifdef CONFIG_DEBUG_VM
  192. #define VM_BUG_ON(cond) BUG_ON(cond)
  193. #else
  194. #define VM_BUG_ON(condition) do { } while(0)
  195. #endif
  196. /*
  197. * Methods to modify the page usage count.
  198. *
  199. * What counts for a page usage:
  200. * - cache mapping (page->mapping)
  201. * - private data (page->private)
  202. * - page mapped in a task's page tables, each mapping
  203. * is counted separately
  204. *
  205. * Also, many kernel routines increase the page count before a critical
  206. * routine so they can be sure the page doesn't go away from under them.
  207. */
  208. /*
  209. * Drop a ref, return true if the refcount fell to zero (the page has no users)
  210. */
  211. static inline int put_page_testzero(struct page *page)
  212. {
  213. VM_BUG_ON(atomic_read(&page->_count) == 0);
  214. return atomic_dec_and_test(&page->_count);
  215. }
  216. /*
  217. * Try to grab a ref unless the page has a refcount of zero, return false if
  218. * that is the case.
  219. */
  220. static inline int get_page_unless_zero(struct page *page)
  221. {
  222. VM_BUG_ON(PageCompound(page));
  223. return atomic_inc_not_zero(&page->_count);
  224. }
  225. static inline int page_count(struct page *page)
  226. {
  227. if (unlikely(PageCompound(page)))
  228. page = (struct page *)page_private(page);
  229. return atomic_read(&page->_count);
  230. }
  231. static inline void get_page(struct page *page)
  232. {
  233. if (unlikely(PageCompound(page)))
  234. page = (struct page *)page_private(page);
  235. VM_BUG_ON(atomic_read(&page->_count) == 0);
  236. atomic_inc(&page->_count);
  237. }
  238. /*
  239. * Setup the page count before being freed into the page allocator for
  240. * the first time (boot or memory hotplug)
  241. */
  242. static inline void init_page_count(struct page *page)
  243. {
  244. atomic_set(&page->_count, 1);
  245. }
  246. void put_page(struct page *page);
  247. void put_pages_list(struct list_head *pages);
  248. void split_page(struct page *page, unsigned int order);
  249. /*
  250. * Compound pages have a destructor function. Provide a
  251. * prototype for that function and accessor functions.
  252. * These are _only_ valid on the head of a PG_compound page.
  253. */
  254. typedef void compound_page_dtor(struct page *);
  255. static inline void set_compound_page_dtor(struct page *page,
  256. compound_page_dtor *dtor)
  257. {
  258. page[1].lru.next = (void *)dtor;
  259. }
  260. static inline compound_page_dtor *get_compound_page_dtor(struct page *page)
  261. {
  262. return (compound_page_dtor *)page[1].lru.next;
  263. }
  264. /*
  265. * Multiple processes may "see" the same page. E.g. for untouched
  266. * mappings of /dev/null, all processes see the same page full of
  267. * zeroes, and text pages of executables and shared libraries have
  268. * only one copy in memory, at most, normally.
  269. *
  270. * For the non-reserved pages, page_count(page) denotes a reference count.
  271. * page_count() == 0 means the page is free. page->lru is then used for
  272. * freelist management in the buddy allocator.
  273. * page_count() > 0 means the page has been allocated.
  274. *
  275. * Pages are allocated by the slab allocator in order to provide memory
  276. * to kmalloc and kmem_cache_alloc. In this case, the management of the
  277. * page, and the fields in 'struct page' are the responsibility of mm/slab.c
  278. * unless a particular usage is carefully commented. (the responsibility of
  279. * freeing the kmalloc memory is the caller's, of course).
  280. *
  281. * A page may be used by anyone else who does a __get_free_page().
  282. * In this case, page_count still tracks the references, and should only
  283. * be used through the normal accessor functions. The top bits of page->flags
  284. * and page->virtual store page management information, but all other fields
  285. * are unused and could be used privately, carefully. The management of this
  286. * page is the responsibility of the one who allocated it, and those who have
  287. * subsequently been given references to it.
  288. *
  289. * The other pages (we may call them "pagecache pages") are completely
  290. * managed by the Linux memory manager: I/O, buffers, swapping etc.
  291. * The following discussion applies only to them.
  292. *
  293. * A pagecache page contains an opaque `private' member, which belongs to the
  294. * page's address_space. Usually, this is the address of a circular list of
  295. * the page's disk buffers. PG_private must be set to tell the VM to call
  296. * into the filesystem to release these pages.
  297. *
  298. * A page may belong to an inode's memory mapping. In this case, page->mapping
  299. * is the pointer to the inode, and page->index is the file offset of the page,
  300. * in units of PAGE_CACHE_SIZE.
  301. *
  302. * If pagecache pages are not associated with an inode, they are said to be
  303. * anonymous pages. These may become associated with the swapcache, and in that
  304. * case PG_swapcache is set, and page->private is an offset into the swapcache.
  305. *
  306. * In either case (swapcache or inode backed), the pagecache itself holds one
  307. * reference to the page. Setting PG_private should also increment the
  308. * refcount. The each user mapping also has a reference to the page.
  309. *
  310. * The pagecache pages are stored in a per-mapping radix tree, which is
  311. * rooted at mapping->page_tree, and indexed by offset.
  312. * Where 2.4 and early 2.6 kernels kept dirty/clean pages in per-address_space
  313. * lists, we instead now tag pages as dirty/writeback in the radix tree.
  314. *
  315. * All pagecache pages may be subject to I/O:
  316. * - inode pages may need to be read from disk,
  317. * - inode pages which have been modified and are MAP_SHARED may need
  318. * to be written back to the inode on disk,
  319. * - anonymous pages (including MAP_PRIVATE file mappings) which have been
  320. * modified may need to be swapped out to swap space and (later) to be read
  321. * back into memory.
  322. */
  323. /*
  324. * The zone field is never updated after free_area_init_core()
  325. * sets it, so none of the operations on it need to be atomic.
  326. */
  327. /*
  328. * page->flags layout:
  329. *
  330. * There are three possibilities for how page->flags get
  331. * laid out. The first is for the normal case, without
  332. * sparsemem. The second is for sparsemem when there is
  333. * plenty of space for node and section. The last is when
  334. * we have run out of space and have to fall back to an
  335. * alternate (slower) way of determining the node.
  336. *
  337. * No sparsemem: | NODE | ZONE | ... | FLAGS |
  338. * with space for node: | SECTION | NODE | ZONE | ... | FLAGS |
  339. * no space for node: | SECTION | ZONE | ... | FLAGS |
  340. */
  341. #ifdef CONFIG_SPARSEMEM
  342. #define SECTIONS_WIDTH SECTIONS_SHIFT
  343. #else
  344. #define SECTIONS_WIDTH 0
  345. #endif
  346. #define ZONES_WIDTH ZONES_SHIFT
  347. #if SECTIONS_WIDTH+ZONES_WIDTH+NODES_SHIFT <= FLAGS_RESERVED
  348. #define NODES_WIDTH NODES_SHIFT
  349. #else
  350. #define NODES_WIDTH 0
  351. #endif
  352. /* Page flags: | [SECTION] | [NODE] | ZONE | ... | FLAGS | */
  353. #define SECTIONS_PGOFF ((sizeof(unsigned long)*8) - SECTIONS_WIDTH)
  354. #define NODES_PGOFF (SECTIONS_PGOFF - NODES_WIDTH)
  355. #define ZONES_PGOFF (NODES_PGOFF - ZONES_WIDTH)
  356. /*
  357. * We are going to use the flags for the page to node mapping if its in
  358. * there. This includes the case where there is no node, so it is implicit.
  359. */
  360. #if !(NODES_WIDTH > 0 || NODES_SHIFT == 0)
  361. #define NODE_NOT_IN_PAGE_FLAGS
  362. #endif
  363. #ifndef PFN_SECTION_SHIFT
  364. #define PFN_SECTION_SHIFT 0
  365. #endif
  366. /*
  367. * Define the bit shifts to access each section. For non-existant
  368. * sections we define the shift as 0; that plus a 0 mask ensures
  369. * the compiler will optimise away reference to them.
  370. */
  371. #define SECTIONS_PGSHIFT (SECTIONS_PGOFF * (SECTIONS_WIDTH != 0))
  372. #define NODES_PGSHIFT (NODES_PGOFF * (NODES_WIDTH != 0))
  373. #define ZONES_PGSHIFT (ZONES_PGOFF * (ZONES_WIDTH != 0))
  374. /* NODE:ZONE or SECTION:ZONE is used to ID a zone for the buddy allcator */
  375. #ifdef NODE_NOT_IN_PAGEFLAGS
  376. #define ZONEID_SHIFT (SECTIONS_SHIFT + ZONES_SHIFT)
  377. #define ZONEID_PGOFF ((SECTIONS_PGOFF < ZONES_PGOFF)? \
  378. SECTIONS_PGOFF : ZONES_PGOFF)
  379. #else
  380. #define ZONEID_SHIFT (NODES_SHIFT + ZONES_SHIFT)
  381. #define ZONEID_PGOFF ((NODES_PGOFF < ZONES_PGOFF)? \
  382. NODES_PGOFF : ZONES_PGOFF)
  383. #endif
  384. #define ZONEID_PGSHIFT (ZONEID_PGOFF * (ZONEID_SHIFT != 0))
  385. #if SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > FLAGS_RESERVED
  386. #error SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > FLAGS_RESERVED
  387. #endif
  388. #define ZONES_MASK ((1UL << ZONES_WIDTH) - 1)
  389. #define NODES_MASK ((1UL << NODES_WIDTH) - 1)
  390. #define SECTIONS_MASK ((1UL << SECTIONS_WIDTH) - 1)
  391. #define ZONEID_MASK ((1UL << ZONEID_SHIFT) - 1)
  392. static inline enum zone_type page_zonenum(struct page *page)
  393. {
  394. return (page->flags >> ZONES_PGSHIFT) & ZONES_MASK;
  395. }
  396. /*
  397. * The identification function is only used by the buddy allocator for
  398. * determining if two pages could be buddies. We are not really
  399. * identifying a zone since we could be using a the section number
  400. * id if we have not node id available in page flags.
  401. * We guarantee only that it will return the same value for two
  402. * combinable pages in a zone.
  403. */
  404. static inline int page_zone_id(struct page *page)
  405. {
  406. return (page->flags >> ZONEID_PGSHIFT) & ZONEID_MASK;
  407. }
  408. static inline int zone_to_nid(struct zone *zone)
  409. {
  410. #ifdef CONFIG_NUMA
  411. return zone->node;
  412. #else
  413. return 0;
  414. #endif
  415. }
  416. #ifdef NODE_NOT_IN_PAGE_FLAGS
  417. extern int page_to_nid(struct page *page);
  418. #else
  419. static inline int page_to_nid(struct page *page)
  420. {
  421. return (page->flags >> NODES_PGSHIFT) & NODES_MASK;
  422. }
  423. #endif
  424. static inline struct zone *page_zone(struct page *page)
  425. {
  426. return &NODE_DATA(page_to_nid(page))->node_zones[page_zonenum(page)];
  427. }
  428. static inline unsigned long page_to_section(struct page *page)
  429. {
  430. return (page->flags >> SECTIONS_PGSHIFT) & SECTIONS_MASK;
  431. }
  432. static inline void set_page_zone(struct page *page, enum zone_type zone)
  433. {
  434. page->flags &= ~(ZONES_MASK << ZONES_PGSHIFT);
  435. page->flags |= (zone & ZONES_MASK) << ZONES_PGSHIFT;
  436. }
  437. static inline void set_page_node(struct page *page, unsigned long node)
  438. {
  439. page->flags &= ~(NODES_MASK << NODES_PGSHIFT);
  440. page->flags |= (node & NODES_MASK) << NODES_PGSHIFT;
  441. }
  442. static inline void set_page_section(struct page *page, unsigned long section)
  443. {
  444. page->flags &= ~(SECTIONS_MASK << SECTIONS_PGSHIFT);
  445. page->flags |= (section & SECTIONS_MASK) << SECTIONS_PGSHIFT;
  446. }
  447. static inline void set_page_links(struct page *page, enum zone_type zone,
  448. unsigned long node, unsigned long pfn)
  449. {
  450. set_page_zone(page, zone);
  451. set_page_node(page, node);
  452. set_page_section(page, pfn_to_section_nr(pfn));
  453. }
  454. /*
  455. * Some inline functions in vmstat.h depend on page_zone()
  456. */
  457. #include <linux/vmstat.h>
  458. static __always_inline void *lowmem_page_address(struct page *page)
  459. {
  460. return __va(page_to_pfn(page) << PAGE_SHIFT);
  461. }
  462. #if defined(CONFIG_HIGHMEM) && !defined(WANT_PAGE_VIRTUAL)
  463. #define HASHED_PAGE_VIRTUAL
  464. #endif
  465. #if defined(WANT_PAGE_VIRTUAL)
  466. #define page_address(page) ((page)->virtual)
  467. #define set_page_address(page, address) \
  468. do { \
  469. (page)->virtual = (address); \
  470. } while(0)
  471. #define page_address_init() do { } while(0)
  472. #endif
  473. #if defined(HASHED_PAGE_VIRTUAL)
  474. void *page_address(struct page *page);
  475. void set_page_address(struct page *page, void *virtual);
  476. void page_address_init(void);
  477. #endif
  478. #if !defined(HASHED_PAGE_VIRTUAL) && !defined(WANT_PAGE_VIRTUAL)
  479. #define page_address(page) lowmem_page_address(page)
  480. #define set_page_address(page, address) do { } while(0)
  481. #define page_address_init() do { } while(0)
  482. #endif
  483. /*
  484. * On an anonymous page mapped into a user virtual memory area,
  485. * page->mapping points to its anon_vma, not to a struct address_space;
  486. * with the PAGE_MAPPING_ANON bit set to distinguish it.
  487. *
  488. * Please note that, confusingly, "page_mapping" refers to the inode
  489. * address_space which maps the page from disk; whereas "page_mapped"
  490. * refers to user virtual address space into which the page is mapped.
  491. */
  492. #define PAGE_MAPPING_ANON 1
  493. extern struct address_space swapper_space;
  494. static inline struct address_space *page_mapping(struct page *page)
  495. {
  496. struct address_space *mapping = page->mapping;
  497. if (unlikely(PageSwapCache(page)))
  498. mapping = &swapper_space;
  499. else if (unlikely((unsigned long)mapping & PAGE_MAPPING_ANON))
  500. mapping = NULL;
  501. return mapping;
  502. }
  503. static inline int PageAnon(struct page *page)
  504. {
  505. return ((unsigned long)page->mapping & PAGE_MAPPING_ANON) != 0;
  506. }
  507. /*
  508. * Return the pagecache index of the passed page. Regular pagecache pages
  509. * use ->index whereas swapcache pages use ->private
  510. */
  511. static inline pgoff_t page_index(struct page *page)
  512. {
  513. if (unlikely(PageSwapCache(page)))
  514. return page_private(page);
  515. return page->index;
  516. }
  517. /*
  518. * The atomic page->_mapcount, like _count, starts from -1:
  519. * so that transitions both from it and to it can be tracked,
  520. * using atomic_inc_and_test and atomic_add_negative(-1).
  521. */
  522. static inline void reset_page_mapcount(struct page *page)
  523. {
  524. atomic_set(&(page)->_mapcount, -1);
  525. }
  526. static inline int page_mapcount(struct page *page)
  527. {
  528. return atomic_read(&(page)->_mapcount) + 1;
  529. }
  530. /*
  531. * Return true if this page is mapped into pagetables.
  532. */
  533. static inline int page_mapped(struct page *page)
  534. {
  535. return atomic_read(&(page)->_mapcount) >= 0;
  536. }
  537. /*
  538. * Error return values for the *_nopage functions
  539. */
  540. #define NOPAGE_SIGBUS (NULL)
  541. #define NOPAGE_OOM ((struct page *) (-1))
  542. #define NOPAGE_REFAULT ((struct page *) (-2)) /* Return to userspace, rerun */
  543. /*
  544. * Error return values for the *_nopfn functions
  545. */
  546. #define NOPFN_SIGBUS ((unsigned long) -1)
  547. #define NOPFN_OOM ((unsigned long) -2)
  548. #define NOPFN_REFAULT ((unsigned long) -3)
  549. /*
  550. * Different kinds of faults, as returned by handle_mm_fault().
  551. * Used to decide whether a process gets delivered SIGBUS or
  552. * just gets major/minor fault counters bumped up.
  553. */
  554. #define VM_FAULT_OOM 0x00
  555. #define VM_FAULT_SIGBUS 0x01
  556. #define VM_FAULT_MINOR 0x02
  557. #define VM_FAULT_MAJOR 0x03
  558. /*
  559. * Special case for get_user_pages.
  560. * Must be in a distinct bit from the above VM_FAULT_ flags.
  561. */
  562. #define VM_FAULT_WRITE 0x10
  563. #define offset_in_page(p) ((unsigned long)(p) & ~PAGE_MASK)
  564. extern void show_free_areas(void);
  565. #ifdef CONFIG_SHMEM
  566. int shmem_set_policy(struct vm_area_struct *vma, struct mempolicy *new);
  567. struct mempolicy *shmem_get_policy(struct vm_area_struct *vma,
  568. unsigned long addr);
  569. int shmem_lock(struct file *file, int lock, struct user_struct *user);
  570. #else
  571. static inline int shmem_lock(struct file *file, int lock,
  572. struct user_struct *user)
  573. {
  574. return 0;
  575. }
  576. static inline int shmem_set_policy(struct vm_area_struct *vma,
  577. struct mempolicy *new)
  578. {
  579. return 0;
  580. }
  581. static inline struct mempolicy *shmem_get_policy(struct vm_area_struct *vma,
  582. unsigned long addr)
  583. {
  584. return NULL;
  585. }
  586. #endif
  587. struct file *shmem_file_setup(char *name, loff_t size, unsigned long flags);
  588. int shmem_zero_setup(struct vm_area_struct *);
  589. #ifndef CONFIG_MMU
  590. extern unsigned long shmem_get_unmapped_area(struct file *file,
  591. unsigned long addr,
  592. unsigned long len,
  593. unsigned long pgoff,
  594. unsigned long flags);
  595. #endif
  596. static inline int can_do_mlock(void)
  597. {
  598. if (capable(CAP_IPC_LOCK))
  599. return 1;
  600. if (current->signal->rlim[RLIMIT_MEMLOCK].rlim_cur != 0)
  601. return 1;
  602. return 0;
  603. }
  604. extern int user_shm_lock(size_t, struct user_struct *);
  605. extern void user_shm_unlock(size_t, struct user_struct *);
  606. /*
  607. * Parameter block passed down to zap_pte_range in exceptional cases.
  608. */
  609. struct zap_details {
  610. struct vm_area_struct *nonlinear_vma; /* Check page->index if set */
  611. struct address_space *check_mapping; /* Check page->mapping if set */
  612. pgoff_t first_index; /* Lowest page->index to unmap */
  613. pgoff_t last_index; /* Highest page->index to unmap */
  614. spinlock_t *i_mmap_lock; /* For unmap_mapping_range: */
  615. unsigned long truncate_count; /* Compare vm_truncate_count */
  616. };
  617. struct page *vm_normal_page(struct vm_area_struct *, unsigned long, pte_t);
  618. unsigned long zap_page_range(struct vm_area_struct *vma, unsigned long address,
  619. unsigned long size, struct zap_details *);
  620. unsigned long unmap_vmas(struct mmu_gather **tlb,
  621. struct vm_area_struct *start_vma, unsigned long start_addr,
  622. unsigned long end_addr, unsigned long *nr_accounted,
  623. struct zap_details *);
  624. void free_pgd_range(struct mmu_gather **tlb, unsigned long addr,
  625. unsigned long end, unsigned long floor, unsigned long ceiling);
  626. void free_pgtables(struct mmu_gather **tlb, struct vm_area_struct *start_vma,
  627. unsigned long floor, unsigned long ceiling);
  628. int copy_page_range(struct mm_struct *dst, struct mm_struct *src,
  629. struct vm_area_struct *vma);
  630. int zeromap_page_range(struct vm_area_struct *vma, unsigned long from,
  631. unsigned long size, pgprot_t prot);
  632. void unmap_mapping_range(struct address_space *mapping,
  633. loff_t const holebegin, loff_t const holelen, int even_cows);
  634. static inline void unmap_shared_mapping_range(struct address_space *mapping,
  635. loff_t const holebegin, loff_t const holelen)
  636. {
  637. unmap_mapping_range(mapping, holebegin, holelen, 0);
  638. }
  639. extern int vmtruncate(struct inode * inode, loff_t offset);
  640. extern int vmtruncate_range(struct inode * inode, loff_t offset, loff_t end);
  641. extern int install_page(struct mm_struct *mm, struct vm_area_struct *vma, unsigned long addr, struct page *page, pgprot_t prot);
  642. extern int install_file_pte(struct mm_struct *mm, struct vm_area_struct *vma, unsigned long addr, unsigned long pgoff, pgprot_t prot);
  643. #ifdef CONFIG_MMU
  644. extern int __handle_mm_fault(struct mm_struct *mm,struct vm_area_struct *vma,
  645. unsigned long address, int write_access);
  646. static inline int handle_mm_fault(struct mm_struct *mm,
  647. struct vm_area_struct *vma, unsigned long address,
  648. int write_access)
  649. {
  650. return __handle_mm_fault(mm, vma, address, write_access) &
  651. (~VM_FAULT_WRITE);
  652. }
  653. #else
  654. static inline int handle_mm_fault(struct mm_struct *mm,
  655. struct vm_area_struct *vma, unsigned long address,
  656. int write_access)
  657. {
  658. /* should never happen if there's no MMU */
  659. BUG();
  660. return VM_FAULT_SIGBUS;
  661. }
  662. #endif
  663. extern int make_pages_present(unsigned long addr, unsigned long end);
  664. extern int access_process_vm(struct task_struct *tsk, unsigned long addr, void *buf, int len, int write);
  665. void install_arg_page(struct vm_area_struct *, struct page *, unsigned long);
  666. int get_user_pages(struct task_struct *tsk, struct mm_struct *mm, unsigned long start,
  667. int len, int write, int force, struct page **pages, struct vm_area_struct **vmas);
  668. void print_bad_pte(struct vm_area_struct *, pte_t, unsigned long);
  669. extern int try_to_release_page(struct page * page, gfp_t gfp_mask);
  670. extern void do_invalidatepage(struct page *page, unsigned long offset);
  671. int __set_page_dirty_nobuffers(struct page *page);
  672. int __set_page_dirty_no_writeback(struct page *page);
  673. int redirty_page_for_writepage(struct writeback_control *wbc,
  674. struct page *page);
  675. int FASTCALL(set_page_dirty(struct page *page));
  676. int set_page_dirty_lock(struct page *page);
  677. int clear_page_dirty_for_io(struct page *page);
  678. extern unsigned long do_mremap(unsigned long addr,
  679. unsigned long old_len, unsigned long new_len,
  680. unsigned long flags, unsigned long new_addr);
  681. /*
  682. * Prototype to add a shrinker callback for ageable caches.
  683. *
  684. * These functions are passed a count `nr_to_scan' and a gfpmask. They should
  685. * scan `nr_to_scan' objects, attempting to free them.
  686. *
  687. * The callback must return the number of objects which remain in the cache.
  688. *
  689. * The callback will be passed nr_to_scan == 0 when the VM is querying the
  690. * cache size, so a fastpath for that case is appropriate.
  691. */
  692. typedef int (*shrinker_t)(int nr_to_scan, gfp_t gfp_mask);
  693. /*
  694. * Add an aging callback. The int is the number of 'seeks' it takes
  695. * to recreate one of the objects that these functions age.
  696. */
  697. #define DEFAULT_SEEKS 2
  698. struct shrinker;
  699. extern struct shrinker *set_shrinker(int, shrinker_t);
  700. extern void remove_shrinker(struct shrinker *shrinker);
  701. /*
  702. * Some shared mappigns will want the pages marked read-only
  703. * to track write events. If so, we'll downgrade vm_page_prot
  704. * to the private version (using protection_map[] without the
  705. * VM_SHARED bit).
  706. */
  707. static inline int vma_wants_writenotify(struct vm_area_struct *vma)
  708. {
  709. unsigned int vm_flags = vma->vm_flags;
  710. /* If it was private or non-writable, the write bit is already clear */
  711. if ((vm_flags & (VM_WRITE|VM_SHARED)) != ((VM_WRITE|VM_SHARED)))
  712. return 0;
  713. /* The backer wishes to know when pages are first written to? */
  714. if (vma->vm_ops && vma->vm_ops->page_mkwrite)
  715. return 1;
  716. /* The open routine did something to the protections already? */
  717. if (pgprot_val(vma->vm_page_prot) !=
  718. pgprot_val(protection_map[vm_flags &
  719. (VM_READ|VM_WRITE|VM_EXEC|VM_SHARED)]))
  720. return 0;
  721. /* Specialty mapping? */
  722. if (vm_flags & (VM_PFNMAP|VM_INSERTPAGE))
  723. return 0;
  724. /* Can the mapping track the dirty pages? */
  725. return vma->vm_file && vma->vm_file->f_mapping &&
  726. mapping_cap_account_dirty(vma->vm_file->f_mapping);
  727. }
  728. extern pte_t *FASTCALL(get_locked_pte(struct mm_struct *mm, unsigned long addr, spinlock_t **ptl));
  729. int __pud_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address);
  730. int __pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address);
  731. int __pte_alloc(struct mm_struct *mm, pmd_t *pmd, unsigned long address);
  732. int __pte_alloc_kernel(pmd_t *pmd, unsigned long address);
  733. /*
  734. * The following ifdef needed to get the 4level-fixup.h header to work.
  735. * Remove it when 4level-fixup.h has been removed.
  736. */
  737. #if defined(CONFIG_MMU) && !defined(__ARCH_HAS_4LEVEL_HACK)
  738. static inline pud_t *pud_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address)
  739. {
  740. return (unlikely(pgd_none(*pgd)) && __pud_alloc(mm, pgd, address))?
  741. NULL: pud_offset(pgd, address);
  742. }
  743. static inline pmd_t *pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address)
  744. {
  745. return (unlikely(pud_none(*pud)) && __pmd_alloc(mm, pud, address))?
  746. NULL: pmd_offset(pud, address);
  747. }
  748. #endif /* CONFIG_MMU && !__ARCH_HAS_4LEVEL_HACK */
  749. #if NR_CPUS >= CONFIG_SPLIT_PTLOCK_CPUS
  750. /*
  751. * We tuck a spinlock to guard each pagetable page into its struct page,
  752. * at page->private, with BUILD_BUG_ON to make sure that this will not
  753. * overflow into the next struct page (as it might with DEBUG_SPINLOCK).
  754. * When freeing, reset page->mapping so free_pages_check won't complain.
  755. */
  756. #define __pte_lockptr(page) &((page)->ptl)
  757. #define pte_lock_init(_page) do { \
  758. spin_lock_init(__pte_lockptr(_page)); \
  759. } while (0)
  760. #define pte_lock_deinit(page) ((page)->mapping = NULL)
  761. #define pte_lockptr(mm, pmd) ({(void)(mm); __pte_lockptr(pmd_page(*(pmd)));})
  762. #else
  763. /*
  764. * We use mm->page_table_lock to guard all pagetable pages of the mm.
  765. */
  766. #define pte_lock_init(page) do {} while (0)
  767. #define pte_lock_deinit(page) do {} while (0)
  768. #define pte_lockptr(mm, pmd) ({(void)(pmd); &(mm)->page_table_lock;})
  769. #endif /* NR_CPUS < CONFIG_SPLIT_PTLOCK_CPUS */
  770. #define pte_offset_map_lock(mm, pmd, address, ptlp) \
  771. ({ \
  772. spinlock_t *__ptl = pte_lockptr(mm, pmd); \
  773. pte_t *__pte = pte_offset_map(pmd, address); \
  774. *(ptlp) = __ptl; \
  775. spin_lock(__ptl); \
  776. __pte; \
  777. })
  778. #define pte_unmap_unlock(pte, ptl) do { \
  779. spin_unlock(ptl); \
  780. pte_unmap(pte); \
  781. } while (0)
  782. #define pte_alloc_map(mm, pmd, address) \
  783. ((unlikely(!pmd_present(*(pmd))) && __pte_alloc(mm, pmd, address))? \
  784. NULL: pte_offset_map(pmd, address))
  785. #define pte_alloc_map_lock(mm, pmd, address, ptlp) \
  786. ((unlikely(!pmd_present(*(pmd))) && __pte_alloc(mm, pmd, address))? \
  787. NULL: pte_offset_map_lock(mm, pmd, address, ptlp))
  788. #define pte_alloc_kernel(pmd, address) \
  789. ((unlikely(!pmd_present(*(pmd))) && __pte_alloc_kernel(pmd, address))? \
  790. NULL: pte_offset_kernel(pmd, address))
  791. extern void free_area_init(unsigned long * zones_size);
  792. extern void free_area_init_node(int nid, pg_data_t *pgdat,
  793. unsigned long * zones_size, unsigned long zone_start_pfn,
  794. unsigned long *zholes_size);
  795. #ifdef CONFIG_ARCH_POPULATES_NODE_MAP
  796. /*
  797. * With CONFIG_ARCH_POPULATES_NODE_MAP set, an architecture may initialise its
  798. * zones, allocate the backing mem_map and account for memory holes in a more
  799. * architecture independent manner. This is a substitute for creating the
  800. * zone_sizes[] and zholes_size[] arrays and passing them to
  801. * free_area_init_node()
  802. *
  803. * An architecture is expected to register range of page frames backed by
  804. * physical memory with add_active_range() before calling
  805. * free_area_init_nodes() passing in the PFN each zone ends at. At a basic
  806. * usage, an architecture is expected to do something like
  807. *
  808. * unsigned long max_zone_pfns[MAX_NR_ZONES] = {max_dma, max_normal_pfn,
  809. * max_highmem_pfn};
  810. * for_each_valid_physical_page_range()
  811. * add_active_range(node_id, start_pfn, end_pfn)
  812. * free_area_init_nodes(max_zone_pfns);
  813. *
  814. * If the architecture guarantees that there are no holes in the ranges
  815. * registered with add_active_range(), free_bootmem_active_regions()
  816. * will call free_bootmem_node() for each registered physical page range.
  817. * Similarly sparse_memory_present_with_active_regions() calls
  818. * memory_present() for each range when SPARSEMEM is enabled.
  819. *
  820. * See mm/page_alloc.c for more information on each function exposed by
  821. * CONFIG_ARCH_POPULATES_NODE_MAP
  822. */
  823. extern void free_area_init_nodes(unsigned long *max_zone_pfn);
  824. extern void add_active_range(unsigned int nid, unsigned long start_pfn,
  825. unsigned long end_pfn);
  826. extern void shrink_active_range(unsigned int nid, unsigned long old_end_pfn,
  827. unsigned long new_end_pfn);
  828. extern void push_node_boundaries(unsigned int nid, unsigned long start_pfn,
  829. unsigned long end_pfn);
  830. extern void remove_all_active_ranges(void);
  831. extern unsigned long absent_pages_in_range(unsigned long start_pfn,
  832. unsigned long end_pfn);
  833. extern void get_pfn_range_for_nid(unsigned int nid,
  834. unsigned long *start_pfn, unsigned long *end_pfn);
  835. extern unsigned long find_min_pfn_with_active_regions(void);
  836. extern unsigned long find_max_pfn_with_active_regions(void);
  837. extern void free_bootmem_with_active_regions(int nid,
  838. unsigned long max_low_pfn);
  839. extern void sparse_memory_present_with_active_regions(int nid);
  840. #ifndef CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID
  841. extern int early_pfn_to_nid(unsigned long pfn);
  842. #endif /* CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID */
  843. #endif /* CONFIG_ARCH_POPULATES_NODE_MAP */
  844. extern void set_dma_reserve(unsigned long new_dma_reserve);
  845. extern void memmap_init_zone(unsigned long, int, unsigned long,
  846. unsigned long, enum memmap_context);
  847. extern void setup_per_zone_pages_min(void);
  848. extern void mem_init(void);
  849. extern void show_mem(void);
  850. extern void si_meminfo(struct sysinfo * val);
  851. extern void si_meminfo_node(struct sysinfo *val, int nid);
  852. #ifdef CONFIG_NUMA
  853. extern void setup_per_cpu_pageset(void);
  854. #else
  855. static inline void setup_per_cpu_pageset(void) {}
  856. #endif
  857. /* prio_tree.c */
  858. void vma_prio_tree_add(struct vm_area_struct *, struct vm_area_struct *old);
  859. void vma_prio_tree_insert(struct vm_area_struct *, struct prio_tree_root *);
  860. void vma_prio_tree_remove(struct vm_area_struct *, struct prio_tree_root *);
  861. struct vm_area_struct *vma_prio_tree_next(struct vm_area_struct *vma,
  862. struct prio_tree_iter *iter);
  863. #define vma_prio_tree_foreach(vma, iter, root, begin, end) \
  864. for (prio_tree_iter_init(iter, root, begin, end), vma = NULL; \
  865. (vma = vma_prio_tree_next(vma, iter)); )
  866. static inline void vma_nonlinear_insert(struct vm_area_struct *vma,
  867. struct list_head *list)
  868. {
  869. vma->shared.vm_set.parent = NULL;
  870. list_add_tail(&vma->shared.vm_set.list, list);
  871. }
  872. /* mmap.c */
  873. extern int __vm_enough_memory(long pages, int cap_sys_admin);
  874. extern void vma_adjust(struct vm_area_struct *vma, unsigned long start,
  875. unsigned long end, pgoff_t pgoff, struct vm_area_struct *insert);
  876. extern struct vm_area_struct *vma_merge(struct mm_struct *,
  877. struct vm_area_struct *prev, unsigned long addr, unsigned long end,
  878. unsigned long vm_flags, struct anon_vma *, struct file *, pgoff_t,
  879. struct mempolicy *);
  880. extern struct anon_vma *find_mergeable_anon_vma(struct vm_area_struct *);
  881. extern int split_vma(struct mm_struct *,
  882. struct vm_area_struct *, unsigned long addr, int new_below);
  883. extern int insert_vm_struct(struct mm_struct *, struct vm_area_struct *);
  884. extern void __vma_link_rb(struct mm_struct *, struct vm_area_struct *,
  885. struct rb_node **, struct rb_node *);
  886. extern void unlink_file_vma(struct vm_area_struct *);
  887. extern struct vm_area_struct *copy_vma(struct vm_area_struct **,
  888. unsigned long addr, unsigned long len, pgoff_t pgoff);
  889. extern void exit_mmap(struct mm_struct *);
  890. extern int may_expand_vm(struct mm_struct *mm, unsigned long npages);
  891. extern int install_special_mapping(struct mm_struct *mm,
  892. unsigned long addr, unsigned long len,
  893. unsigned long flags, struct page **pages);
  894. extern unsigned long get_unmapped_area(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
  895. extern unsigned long do_mmap_pgoff(struct file *file, unsigned long addr,
  896. unsigned long len, unsigned long prot,
  897. unsigned long flag, unsigned long pgoff);
  898. static inline unsigned long do_mmap(struct file *file, unsigned long addr,
  899. unsigned long len, unsigned long prot,
  900. unsigned long flag, unsigned long offset)
  901. {
  902. unsigned long ret = -EINVAL;
  903. if ((offset + PAGE_ALIGN(len)) < offset)
  904. goto out;
  905. if (!(offset & ~PAGE_MASK))
  906. ret = do_mmap_pgoff(file, addr, len, prot, flag, offset >> PAGE_SHIFT);
  907. out:
  908. return ret;
  909. }
  910. extern int do_munmap(struct mm_struct *, unsigned long, size_t);
  911. extern unsigned long do_brk(unsigned long, unsigned long);
  912. /* filemap.c */
  913. extern unsigned long page_unuse(struct page *);
  914. extern void truncate_inode_pages(struct address_space *, loff_t);
  915. extern void truncate_inode_pages_range(struct address_space *,
  916. loff_t lstart, loff_t lend);
  917. /* generic vm_area_ops exported for stackable file systems */
  918. extern struct page *filemap_nopage(struct vm_area_struct *, unsigned long, int *);
  919. extern int filemap_populate(struct vm_area_struct *, unsigned long,
  920. unsigned long, pgprot_t, unsigned long, int);
  921. /* mm/page-writeback.c */
  922. int write_one_page(struct page *page, int wait);
  923. /* readahead.c */
  924. #define VM_MAX_READAHEAD 128 /* kbytes */
  925. #define VM_MIN_READAHEAD 16 /* kbytes (includes current page) */
  926. #define VM_MAX_CACHE_HIT 256 /* max pages in a row in cache before
  927. * turning readahead off */
  928. int do_page_cache_readahead(struct address_space *mapping, struct file *filp,
  929. pgoff_t offset, unsigned long nr_to_read);
  930. int force_page_cache_readahead(struct address_space *mapping, struct file *filp,
  931. pgoff_t offset, unsigned long nr_to_read);
  932. unsigned long page_cache_readahead(struct address_space *mapping,
  933. struct file_ra_state *ra,
  934. struct file *filp,
  935. pgoff_t offset,
  936. unsigned long size);
  937. void handle_ra_miss(struct address_space *mapping,
  938. struct file_ra_state *ra, pgoff_t offset);
  939. unsigned long max_sane_readahead(unsigned long nr);
  940. /* Do stack extension */
  941. extern int expand_stack(struct vm_area_struct *vma, unsigned long address);
  942. #ifdef CONFIG_IA64
  943. extern int expand_upwards(struct vm_area_struct *vma, unsigned long address);
  944. #endif
  945. /* Look up the first VMA which satisfies addr < vm_end, NULL if none. */
  946. extern struct vm_area_struct * find_vma(struct mm_struct * mm, unsigned long addr);
  947. extern struct vm_area_struct * find_vma_prev(struct mm_struct * mm, unsigned long addr,
  948. struct vm_area_struct **pprev);
  949. /* Look up the first VMA which intersects the interval start_addr..end_addr-1,
  950. NULL if none. Assume start_addr < end_addr. */
  951. static inline struct vm_area_struct * find_vma_intersection(struct mm_struct * mm, unsigned long start_addr, unsigned long end_addr)
  952. {
  953. struct vm_area_struct * vma = find_vma(mm,start_addr);
  954. if (vma && end_addr <= vma->vm_start)
  955. vma = NULL;
  956. return vma;
  957. }
  958. static inline unsigned long vma_pages(struct vm_area_struct *vma)
  959. {
  960. return (vma->vm_end - vma->vm_start) >> PAGE_SHIFT;
  961. }
  962. pgprot_t vm_get_page_prot(unsigned long vm_flags);
  963. struct vm_area_struct *find_extend_vma(struct mm_struct *, unsigned long addr);
  964. struct page *vmalloc_to_page(void *addr);
  965. unsigned long vmalloc_to_pfn(void *addr);
  966. int remap_pfn_range(struct vm_area_struct *, unsigned long addr,
  967. unsigned long pfn, unsigned long size, pgprot_t);
  968. int vm_insert_page(struct vm_area_struct *, unsigned long addr, struct page *);
  969. int vm_insert_pfn(struct vm_area_struct *vma, unsigned long addr,
  970. unsigned long pfn);
  971. struct page *follow_page(struct vm_area_struct *, unsigned long address,
  972. unsigned int foll_flags);
  973. #define FOLL_WRITE 0x01 /* check pte is writable */
  974. #define FOLL_TOUCH 0x02 /* mark page accessed */
  975. #define FOLL_GET 0x04 /* do get_page on page */
  976. #define FOLL_ANON 0x08 /* give ZERO_PAGE if no pgtable */
  977. #ifdef CONFIG_PROC_FS
  978. void vm_stat_account(struct mm_struct *, unsigned long, struct file *, long);
  979. #else
  980. static inline void vm_stat_account(struct mm_struct *mm,
  981. unsigned long flags, struct file *file, long pages)
  982. {
  983. }
  984. #endif /* CONFIG_PROC_FS */
  985. #ifndef CONFIG_DEBUG_PAGEALLOC
  986. static inline void
  987. kernel_map_pages(struct page *page, int numpages, int enable) {}
  988. #endif
  989. extern struct vm_area_struct *get_gate_vma(struct task_struct *tsk);
  990. #ifdef __HAVE_ARCH_GATE_AREA
  991. int in_gate_area_no_task(unsigned long addr);
  992. int in_gate_area(struct task_struct *task, unsigned long addr);
  993. #else
  994. int in_gate_area_no_task(unsigned long addr);
  995. #define in_gate_area(task, addr) ({(void)task; in_gate_area_no_task(addr);})
  996. #endif /* __HAVE_ARCH_GATE_AREA */
  997. int drop_caches_sysctl_handler(struct ctl_table *, int, struct file *,
  998. void __user *, size_t *, loff_t *);
  999. unsigned long shrink_slab(unsigned long scanned, gfp_t gfp_mask,
  1000. unsigned long lru_pages);
  1001. void drop_pagecache(void);
  1002. void drop_slab(void);
  1003. #ifndef CONFIG_MMU
  1004. #define randomize_va_space 0
  1005. #else
  1006. extern int randomize_va_space;
  1007. #endif
  1008. __attribute__((weak)) const char *arch_vma_name(struct vm_area_struct *vma);
  1009. #endif /* __KERNEL__ */
  1010. #endif /* _LINUX_MM_H */