mm.h 43 KB

12345678910111213141516171819202122232425262728293031323334353637383940414243444546474849505152535455565758596061626364656667686970717273747576777879808182838485868788899091929394959697989910010110210310410510610710810911011111211311411511611711811912012112212312412512612712812913013113213313413513613713813914014114214314414514614714814915015115215315415515615715815916016116216316416516616716816917017117217317417517617717817918018118218318418518618718818919019119219319419519619719819920020120220320420520620720820921021121221321421521621721821922022122222322422522622722822923023123223323423523623723823924024124224324424524624724824925025125225325425525625725825926026126226326426526626726826927027127227327427527627727827928028128228328428528628728828929029129229329429529629729829930030130230330430530630730830931031131231331431531631731831932032132232332432532632732832933033133233333433533633733833934034134234334434534634734834935035135235335435535635735835936036136236336436536636736836937037137237337437537637737837938038138238338438538638738838939039139239339439539639739839940040140240340440540640740840941041141241341441541641741841942042142242342442542642742842943043143243343443543643743843944044144244344444544644744844945045145245345445545645745845946046146246346446546646746846947047147247347447547647747847948048148248348448548648748848949049149249349449549649749849950050150250350450550650750850951051151251351451551651751851952052152252352452552652752852953053153253353453553653753853954054154254354454554654754854955055155255355455555655755855956056156256356456556656756856957057157257357457557657757857958058158258358458558658758858959059159259359459559659759859960060160260360460560660760860961061161261361461561661761861962062162262362462562662762862963063163263363463563663763863964064164264364464564664764864965065165265365465565665765865966066166266366466566666766866967067167267367467567667767867968068168268368468568668768868969069169269369469569669769869970070170270370470570670770870971071171271371471571671771871972072172272372472572672772872973073173273373473573673773873974074174274374474574674774874975075175275375475575675775875976076176276376476576676776876977077177277377477577677777877978078178278378478578678778878979079179279379479579679779879980080180280380480580680780880981081181281381481581681781881982082182282382482582682782882983083183283383483583683783883984084184284384484584684784884985085185285385485585685785885986086186286386486586686786886987087187287387487587687787887988088188288388488588688788888989089189289389489589689789889990090190290390490590690790890991091191291391491591691791891992092192292392492592692792892993093193293393493593693793893994094194294394494594694794894995095195295395495595695795895996096196296396496596696796896997097197297397497597697797897998098198298398498598698798898999099199299399499599699799899910001001100210031004100510061007100810091010101110121013101410151016101710181019102010211022102310241025102610271028102910301031103210331034103510361037103810391040104110421043104410451046104710481049105010511052105310541055105610571058105910601061106210631064106510661067106810691070107110721073107410751076107710781079108010811082108310841085108610871088108910901091109210931094109510961097109810991100110111021103110411051106110711081109111011111112111311141115111611171118111911201121112211231124112511261127112811291130113111321133113411351136113711381139114011411142114311441145114611471148114911501151115211531154115511561157115811591160116111621163116411651166116711681169117011711172117311741175117611771178117911801181118211831184118511861187118811891190119111921193119411951196119711981199120012011202120312041205120612071208120912101211121212131214121512161217121812191220122112221223122412251226122712281229123012311232123312341235123612371238123912401241124212431244124512461247124812491250125112521253125412551256125712581259126012611262126312641265126612671268126912701271
  1. #ifndef _LINUX_MM_H
  2. #define _LINUX_MM_H
  3. #include <linux/errno.h>
  4. #ifdef __KERNEL__
  5. #include <linux/gfp.h>
  6. #include <linux/list.h>
  7. #include <linux/mmzone.h>
  8. #include <linux/rbtree.h>
  9. #include <linux/prio_tree.h>
  10. #include <linux/debug_locks.h>
  11. #include <linux/mm_types.h>
  12. struct mempolicy;
  13. struct anon_vma;
  14. struct file_ra_state;
  15. struct user_struct;
  16. struct writeback_control;
  17. #ifndef CONFIG_DISCONTIGMEM /* Don't use mapnrs, do it properly */
  18. extern unsigned long max_mapnr;
  19. #endif
  20. extern unsigned long num_physpages;
  21. extern void * high_memory;
  22. extern int page_cluster;
  23. #ifdef CONFIG_SYSCTL
  24. extern int sysctl_legacy_va_layout;
  25. #else
  26. #define sysctl_legacy_va_layout 0
  27. #endif
  28. extern unsigned long mmap_min_addr;
  29. #include <asm/page.h>
  30. #include <asm/pgtable.h>
  31. #include <asm/processor.h>
  32. #define nth_page(page,n) pfn_to_page(page_to_pfn((page)) + (n))
  33. /*
  34. * Linux kernel virtual memory manager primitives.
  35. * The idea being to have a "virtual" mm in the same way
  36. * we have a virtual fs - giving a cleaner interface to the
  37. * mm details, and allowing different kinds of memory mappings
  38. * (from shared memory to executable loading to arbitrary
  39. * mmap() functions).
  40. */
  41. extern struct kmem_cache *vm_area_cachep;
  42. /*
  43. * This struct defines the per-mm list of VMAs for uClinux. If CONFIG_MMU is
  44. * disabled, then there's a single shared list of VMAs maintained by the
  45. * system, and mm's subscribe to these individually
  46. */
  47. struct vm_list_struct {
  48. struct vm_list_struct *next;
  49. struct vm_area_struct *vma;
  50. };
  51. #ifndef CONFIG_MMU
  52. extern struct rb_root nommu_vma_tree;
  53. extern struct rw_semaphore nommu_vma_sem;
  54. extern unsigned int kobjsize(const void *objp);
  55. #endif
  56. /*
  57. * vm_flags..
  58. */
  59. #define VM_READ 0x00000001 /* currently active flags */
  60. #define VM_WRITE 0x00000002
  61. #define VM_EXEC 0x00000004
  62. #define VM_SHARED 0x00000008
  63. /* mprotect() hardcodes VM_MAYREAD >> 4 == VM_READ, and so for r/w/x bits. */
  64. #define VM_MAYREAD 0x00000010 /* limits for mprotect() etc */
  65. #define VM_MAYWRITE 0x00000020
  66. #define VM_MAYEXEC 0x00000040
  67. #define VM_MAYSHARE 0x00000080
  68. #define VM_GROWSDOWN 0x00000100 /* general info on the segment */
  69. #define VM_GROWSUP 0x00000200
  70. #define VM_PFNMAP 0x00000400 /* Page-ranges managed without "struct page", just pure PFN */
  71. #define VM_DENYWRITE 0x00000800 /* ETXTBSY on write attempts.. */
  72. #define VM_EXECUTABLE 0x00001000
  73. #define VM_LOCKED 0x00002000
  74. #define VM_IO 0x00004000 /* Memory mapped I/O or similar */
  75. /* Used by sys_madvise() */
  76. #define VM_SEQ_READ 0x00008000 /* App will access data sequentially */
  77. #define VM_RAND_READ 0x00010000 /* App will not benefit from clustered reads */
  78. #define VM_DONTCOPY 0x00020000 /* Do not copy this vma on fork */
  79. #define VM_DONTEXPAND 0x00040000 /* Cannot expand with mremap() */
  80. #define VM_RESERVED 0x00080000 /* Count as reserved_vm like IO */
  81. #define VM_ACCOUNT 0x00100000 /* Is a VM accounted object */
  82. #define VM_HUGETLB 0x00400000 /* Huge TLB Page VM */
  83. #define VM_NONLINEAR 0x00800000 /* Is non-linear (remap_file_pages) */
  84. #define VM_MAPPED_COPY 0x01000000 /* T if mapped copy of data (nommu mmap) */
  85. #define VM_INSERTPAGE 0x02000000 /* The vma has had "vm_insert_page()" done on it */
  86. #define VM_ALWAYSDUMP 0x04000000 /* Always include in core dumps */
  87. #define VM_CAN_NONLINEAR 0x08000000 /* Has ->fault & does nonlinear pages */
  88. #define VM_MIXEDMAP 0x10000000 /* Can contain "struct page" and pure PFN pages */
  89. #define VM_SAO 0x20000000 /* Strong Access Ordering (powerpc) */
  90. #ifndef VM_STACK_DEFAULT_FLAGS /* arch can override this */
  91. #define VM_STACK_DEFAULT_FLAGS VM_DATA_DEFAULT_FLAGS
  92. #endif
  93. #ifdef CONFIG_STACK_GROWSUP
  94. #define VM_STACK_FLAGS (VM_GROWSUP | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT)
  95. #else
  96. #define VM_STACK_FLAGS (VM_GROWSDOWN | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT)
  97. #endif
  98. #define VM_READHINTMASK (VM_SEQ_READ | VM_RAND_READ)
  99. #define VM_ClearReadHint(v) (v)->vm_flags &= ~VM_READHINTMASK
  100. #define VM_NormalReadHint(v) (!((v)->vm_flags & VM_READHINTMASK))
  101. #define VM_SequentialReadHint(v) ((v)->vm_flags & VM_SEQ_READ)
  102. #define VM_RandomReadHint(v) ((v)->vm_flags & VM_RAND_READ)
  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. /*
  111. * vm_fault is filled by the the pagefault handler and passed to the vma's
  112. * ->fault function. The vma's ->fault is responsible for returning a bitmask
  113. * of VM_FAULT_xxx flags that give details about how the fault was handled.
  114. *
  115. * pgoff should be used in favour of virtual_address, if possible. If pgoff
  116. * is used, one may set VM_CAN_NONLINEAR in the vma->vm_flags to get nonlinear
  117. * mapping support.
  118. */
  119. struct vm_fault {
  120. unsigned int flags; /* FAULT_FLAG_xxx flags */
  121. pgoff_t pgoff; /* Logical page offset based on vma */
  122. void __user *virtual_address; /* Faulting virtual address */
  123. struct page *page; /* ->fault handlers should return a
  124. * page here, unless VM_FAULT_NOPAGE
  125. * is set (which is also implied by
  126. * VM_FAULT_ERROR).
  127. */
  128. };
  129. /*
  130. * These are the virtual MM functions - opening of an area, closing and
  131. * unmapping it (needed to keep files on disk up-to-date etc), pointer
  132. * to the functions called when a no-page or a wp-page exception occurs.
  133. */
  134. struct vm_operations_struct {
  135. void (*open)(struct vm_area_struct * area);
  136. void (*close)(struct vm_area_struct * area);
  137. int (*fault)(struct vm_area_struct *vma, struct vm_fault *vmf);
  138. unsigned long (*nopfn)(struct vm_area_struct *area,
  139. unsigned long address);
  140. /* notification that a previously read-only page is about to become
  141. * writable, if an error is returned it will cause a SIGBUS */
  142. int (*page_mkwrite)(struct vm_area_struct *vma, struct page *page);
  143. #ifdef CONFIG_NUMA
  144. /*
  145. * set_policy() op must add a reference to any non-NULL @new mempolicy
  146. * to hold the policy upon return. Caller should pass NULL @new to
  147. * remove a policy and fall back to surrounding context--i.e. do not
  148. * install a MPOL_DEFAULT policy, nor the task or system default
  149. * mempolicy.
  150. */
  151. int (*set_policy)(struct vm_area_struct *vma, struct mempolicy *new);
  152. /*
  153. * get_policy() op must add reference [mpol_get()] to any policy at
  154. * (vma,addr) marked as MPOL_SHARED. The shared policy infrastructure
  155. * in mm/mempolicy.c will do this automatically.
  156. * get_policy() must NOT add a ref if the policy at (vma,addr) is not
  157. * marked as MPOL_SHARED. vma policies are protected by the mmap_sem.
  158. * If no [shared/vma] mempolicy exists at the addr, get_policy() op
  159. * must return NULL--i.e., do not "fallback" to task or system default
  160. * policy.
  161. */
  162. struct mempolicy *(*get_policy)(struct vm_area_struct *vma,
  163. unsigned long addr);
  164. int (*migrate)(struct vm_area_struct *vma, const nodemask_t *from,
  165. const nodemask_t *to, unsigned long flags);
  166. #endif
  167. };
  168. struct mmu_gather;
  169. struct inode;
  170. #define page_private(page) ((page)->private)
  171. #define set_page_private(page, v) ((page)->private = (v))
  172. /*
  173. * FIXME: take this include out, include page-flags.h in
  174. * files which need it (119 of them)
  175. */
  176. #include <linux/page-flags.h>
  177. #ifdef CONFIG_DEBUG_VM
  178. #define VM_BUG_ON(cond) BUG_ON(cond)
  179. #else
  180. #define VM_BUG_ON(condition) do { } while(0)
  181. #endif
  182. /*
  183. * Methods to modify the page usage count.
  184. *
  185. * What counts for a page usage:
  186. * - cache mapping (page->mapping)
  187. * - private data (page->private)
  188. * - page mapped in a task's page tables, each mapping
  189. * is counted separately
  190. *
  191. * Also, many kernel routines increase the page count before a critical
  192. * routine so they can be sure the page doesn't go away from under them.
  193. */
  194. /*
  195. * Drop a ref, return true if the refcount fell to zero (the page has no users)
  196. */
  197. static inline int put_page_testzero(struct page *page)
  198. {
  199. VM_BUG_ON(atomic_read(&page->_count) == 0);
  200. return atomic_dec_and_test(&page->_count);
  201. }
  202. /*
  203. * Try to grab a ref unless the page has a refcount of zero, return false if
  204. * that is the case.
  205. */
  206. static inline int get_page_unless_zero(struct page *page)
  207. {
  208. VM_BUG_ON(PageTail(page));
  209. return atomic_inc_not_zero(&page->_count);
  210. }
  211. /* Support for virtually mapped pages */
  212. struct page *vmalloc_to_page(const void *addr);
  213. unsigned long vmalloc_to_pfn(const void *addr);
  214. /*
  215. * Determine if an address is within the vmalloc range
  216. *
  217. * On nommu, vmalloc/vfree wrap through kmalloc/kfree directly, so there
  218. * is no special casing required.
  219. */
  220. static inline int is_vmalloc_addr(const void *x)
  221. {
  222. #ifdef CONFIG_MMU
  223. unsigned long addr = (unsigned long)x;
  224. return addr >= VMALLOC_START && addr < VMALLOC_END;
  225. #else
  226. return 0;
  227. #endif
  228. }
  229. static inline struct page *compound_head(struct page *page)
  230. {
  231. if (unlikely(PageTail(page)))
  232. return page->first_page;
  233. return page;
  234. }
  235. static inline int page_count(struct page *page)
  236. {
  237. return atomic_read(&compound_head(page)->_count);
  238. }
  239. static inline void get_page(struct page *page)
  240. {
  241. page = compound_head(page);
  242. VM_BUG_ON(atomic_read(&page->_count) == 0);
  243. atomic_inc(&page->_count);
  244. }
  245. static inline struct page *virt_to_head_page(const void *x)
  246. {
  247. struct page *page = virt_to_page(x);
  248. return compound_head(page);
  249. }
  250. /*
  251. * Setup the page count before being freed into the page allocator for
  252. * the first time (boot or memory hotplug)
  253. */
  254. static inline void init_page_count(struct page *page)
  255. {
  256. atomic_set(&page->_count, 1);
  257. }
  258. void put_page(struct page *page);
  259. void put_pages_list(struct list_head *pages);
  260. void split_page(struct page *page, unsigned int order);
  261. /*
  262. * Compound pages have a destructor function. Provide a
  263. * prototype for that function and accessor functions.
  264. * These are _only_ valid on the head of a PG_compound page.
  265. */
  266. typedef void compound_page_dtor(struct page *);
  267. static inline void set_compound_page_dtor(struct page *page,
  268. compound_page_dtor *dtor)
  269. {
  270. page[1].lru.next = (void *)dtor;
  271. }
  272. static inline compound_page_dtor *get_compound_page_dtor(struct page *page)
  273. {
  274. return (compound_page_dtor *)page[1].lru.next;
  275. }
  276. static inline int compound_order(struct page *page)
  277. {
  278. if (!PageHead(page))
  279. return 0;
  280. return (unsigned long)page[1].lru.prev;
  281. }
  282. static inline void set_compound_order(struct page *page, unsigned long order)
  283. {
  284. page[1].lru.prev = (void *)order;
  285. }
  286. /*
  287. * Multiple processes may "see" the same page. E.g. for untouched
  288. * mappings of /dev/null, all processes see the same page full of
  289. * zeroes, and text pages of executables and shared libraries have
  290. * only one copy in memory, at most, normally.
  291. *
  292. * For the non-reserved pages, page_count(page) denotes a reference count.
  293. * page_count() == 0 means the page is free. page->lru is then used for
  294. * freelist management in the buddy allocator.
  295. * page_count() > 0 means the page has been allocated.
  296. *
  297. * Pages are allocated by the slab allocator in order to provide memory
  298. * to kmalloc and kmem_cache_alloc. In this case, the management of the
  299. * page, and the fields in 'struct page' are the responsibility of mm/slab.c
  300. * unless a particular usage is carefully commented. (the responsibility of
  301. * freeing the kmalloc memory is the caller's, of course).
  302. *
  303. * A page may be used by anyone else who does a __get_free_page().
  304. * In this case, page_count still tracks the references, and should only
  305. * be used through the normal accessor functions. The top bits of page->flags
  306. * and page->virtual store page management information, but all other fields
  307. * are unused and could be used privately, carefully. The management of this
  308. * page is the responsibility of the one who allocated it, and those who have
  309. * subsequently been given references to it.
  310. *
  311. * The other pages (we may call them "pagecache pages") are completely
  312. * managed by the Linux memory manager: I/O, buffers, swapping etc.
  313. * The following discussion applies only to them.
  314. *
  315. * A pagecache page contains an opaque `private' member, which belongs to the
  316. * page's address_space. Usually, this is the address of a circular list of
  317. * the page's disk buffers. PG_private must be set to tell the VM to call
  318. * into the filesystem to release these pages.
  319. *
  320. * A page may belong to an inode's memory mapping. In this case, page->mapping
  321. * is the pointer to the inode, and page->index is the file offset of the page,
  322. * in units of PAGE_CACHE_SIZE.
  323. *
  324. * If pagecache pages are not associated with an inode, they are said to be
  325. * anonymous pages. These may become associated with the swapcache, and in that
  326. * case PG_swapcache is set, and page->private is an offset into the swapcache.
  327. *
  328. * In either case (swapcache or inode backed), the pagecache itself holds one
  329. * reference to the page. Setting PG_private should also increment the
  330. * refcount. The each user mapping also has a reference to the page.
  331. *
  332. * The pagecache pages are stored in a per-mapping radix tree, which is
  333. * rooted at mapping->page_tree, and indexed by offset.
  334. * Where 2.4 and early 2.6 kernels kept dirty/clean pages in per-address_space
  335. * lists, we instead now tag pages as dirty/writeback in the radix tree.
  336. *
  337. * All pagecache pages may be subject to I/O:
  338. * - inode pages may need to be read from disk,
  339. * - inode pages which have been modified and are MAP_SHARED may need
  340. * to be written back to the inode on disk,
  341. * - anonymous pages (including MAP_PRIVATE file mappings) which have been
  342. * modified may need to be swapped out to swap space and (later) to be read
  343. * back into memory.
  344. */
  345. /*
  346. * The zone field is never updated after free_area_init_core()
  347. * sets it, so none of the operations on it need to be atomic.
  348. */
  349. /*
  350. * page->flags layout:
  351. *
  352. * There are three possibilities for how page->flags get
  353. * laid out. The first is for the normal case, without
  354. * sparsemem. The second is for sparsemem when there is
  355. * plenty of space for node and section. The last is when
  356. * we have run out of space and have to fall back to an
  357. * alternate (slower) way of determining the node.
  358. *
  359. * No sparsemem or sparsemem vmemmap: | NODE | ZONE | ... | FLAGS |
  360. * classic sparse with space for node:| SECTION | NODE | ZONE | ... | FLAGS |
  361. * classic sparse no space for node: | SECTION | ZONE | ... | FLAGS |
  362. */
  363. #if defined(CONFIG_SPARSEMEM) && !defined(CONFIG_SPARSEMEM_VMEMMAP)
  364. #define SECTIONS_WIDTH SECTIONS_SHIFT
  365. #else
  366. #define SECTIONS_WIDTH 0
  367. #endif
  368. #define ZONES_WIDTH ZONES_SHIFT
  369. #if SECTIONS_WIDTH+ZONES_WIDTH+NODES_SHIFT <= BITS_PER_LONG - NR_PAGEFLAGS
  370. #define NODES_WIDTH NODES_SHIFT
  371. #else
  372. #ifdef CONFIG_SPARSEMEM_VMEMMAP
  373. #error "Vmemmap: No space for nodes field in page flags"
  374. #endif
  375. #define NODES_WIDTH 0
  376. #endif
  377. /* Page flags: | [SECTION] | [NODE] | ZONE | ... | FLAGS | */
  378. #define SECTIONS_PGOFF ((sizeof(unsigned long)*8) - SECTIONS_WIDTH)
  379. #define NODES_PGOFF (SECTIONS_PGOFF - NODES_WIDTH)
  380. #define ZONES_PGOFF (NODES_PGOFF - ZONES_WIDTH)
  381. /*
  382. * We are going to use the flags for the page to node mapping if its in
  383. * there. This includes the case where there is no node, so it is implicit.
  384. */
  385. #if !(NODES_WIDTH > 0 || NODES_SHIFT == 0)
  386. #define NODE_NOT_IN_PAGE_FLAGS
  387. #endif
  388. #ifndef PFN_SECTION_SHIFT
  389. #define PFN_SECTION_SHIFT 0
  390. #endif
  391. /*
  392. * Define the bit shifts to access each section. For non-existant
  393. * sections we define the shift as 0; that plus a 0 mask ensures
  394. * the compiler will optimise away reference to them.
  395. */
  396. #define SECTIONS_PGSHIFT (SECTIONS_PGOFF * (SECTIONS_WIDTH != 0))
  397. #define NODES_PGSHIFT (NODES_PGOFF * (NODES_WIDTH != 0))
  398. #define ZONES_PGSHIFT (ZONES_PGOFF * (ZONES_WIDTH != 0))
  399. /* NODE:ZONE or SECTION:ZONE is used to ID a zone for the buddy allcator */
  400. #ifdef NODE_NOT_IN_PAGEFLAGS
  401. #define ZONEID_SHIFT (SECTIONS_SHIFT + ZONES_SHIFT)
  402. #define ZONEID_PGOFF ((SECTIONS_PGOFF < ZONES_PGOFF)? \
  403. SECTIONS_PGOFF : ZONES_PGOFF)
  404. #else
  405. #define ZONEID_SHIFT (NODES_SHIFT + ZONES_SHIFT)
  406. #define ZONEID_PGOFF ((NODES_PGOFF < ZONES_PGOFF)? \
  407. NODES_PGOFF : ZONES_PGOFF)
  408. #endif
  409. #define ZONEID_PGSHIFT (ZONEID_PGOFF * (ZONEID_SHIFT != 0))
  410. #if SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > BITS_PER_LONG - NR_PAGEFLAGS
  411. #error SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > BITS_PER_LONG - NR_PAGEFLAGS
  412. #endif
  413. #define ZONES_MASK ((1UL << ZONES_WIDTH) - 1)
  414. #define NODES_MASK ((1UL << NODES_WIDTH) - 1)
  415. #define SECTIONS_MASK ((1UL << SECTIONS_WIDTH) - 1)
  416. #define ZONEID_MASK ((1UL << ZONEID_SHIFT) - 1)
  417. static inline enum zone_type page_zonenum(struct page *page)
  418. {
  419. return (page->flags >> ZONES_PGSHIFT) & ZONES_MASK;
  420. }
  421. /*
  422. * The identification function is only used by the buddy allocator for
  423. * determining if two pages could be buddies. We are not really
  424. * identifying a zone since we could be using a the section number
  425. * id if we have not node id available in page flags.
  426. * We guarantee only that it will return the same value for two
  427. * combinable pages in a zone.
  428. */
  429. static inline int page_zone_id(struct page *page)
  430. {
  431. return (page->flags >> ZONEID_PGSHIFT) & ZONEID_MASK;
  432. }
  433. static inline int zone_to_nid(struct zone *zone)
  434. {
  435. #ifdef CONFIG_NUMA
  436. return zone->node;
  437. #else
  438. return 0;
  439. #endif
  440. }
  441. #ifdef NODE_NOT_IN_PAGE_FLAGS
  442. extern int page_to_nid(struct page *page);
  443. #else
  444. static inline int page_to_nid(struct page *page)
  445. {
  446. return (page->flags >> NODES_PGSHIFT) & NODES_MASK;
  447. }
  448. #endif
  449. static inline struct zone *page_zone(struct page *page)
  450. {
  451. return &NODE_DATA(page_to_nid(page))->node_zones[page_zonenum(page)];
  452. }
  453. #if defined(CONFIG_SPARSEMEM) && !defined(CONFIG_SPARSEMEM_VMEMMAP)
  454. static inline unsigned long page_to_section(struct page *page)
  455. {
  456. return (page->flags >> SECTIONS_PGSHIFT) & SECTIONS_MASK;
  457. }
  458. #endif
  459. static inline void set_page_zone(struct page *page, enum zone_type zone)
  460. {
  461. page->flags &= ~(ZONES_MASK << ZONES_PGSHIFT);
  462. page->flags |= (zone & ZONES_MASK) << ZONES_PGSHIFT;
  463. }
  464. static inline void set_page_node(struct page *page, unsigned long node)
  465. {
  466. page->flags &= ~(NODES_MASK << NODES_PGSHIFT);
  467. page->flags |= (node & NODES_MASK) << NODES_PGSHIFT;
  468. }
  469. static inline void set_page_section(struct page *page, unsigned long section)
  470. {
  471. page->flags &= ~(SECTIONS_MASK << SECTIONS_PGSHIFT);
  472. page->flags |= (section & SECTIONS_MASK) << SECTIONS_PGSHIFT;
  473. }
  474. static inline void set_page_links(struct page *page, enum zone_type zone,
  475. unsigned long node, unsigned long pfn)
  476. {
  477. set_page_zone(page, zone);
  478. set_page_node(page, node);
  479. set_page_section(page, pfn_to_section_nr(pfn));
  480. }
  481. /*
  482. * If a hint addr is less than mmap_min_addr change hint to be as
  483. * low as possible but still greater than mmap_min_addr
  484. */
  485. static inline unsigned long round_hint_to_min(unsigned long hint)
  486. {
  487. #ifdef CONFIG_SECURITY
  488. hint &= PAGE_MASK;
  489. if (((void *)hint != NULL) &&
  490. (hint < mmap_min_addr))
  491. return PAGE_ALIGN(mmap_min_addr);
  492. #endif
  493. return hint;
  494. }
  495. /*
  496. * Some inline functions in vmstat.h depend on page_zone()
  497. */
  498. #include <linux/vmstat.h>
  499. static __always_inline void *lowmem_page_address(struct page *page)
  500. {
  501. return __va(page_to_pfn(page) << PAGE_SHIFT);
  502. }
  503. #if defined(CONFIG_HIGHMEM) && !defined(WANT_PAGE_VIRTUAL)
  504. #define HASHED_PAGE_VIRTUAL
  505. #endif
  506. #if defined(WANT_PAGE_VIRTUAL)
  507. #define page_address(page) ((page)->virtual)
  508. #define set_page_address(page, address) \
  509. do { \
  510. (page)->virtual = (address); \
  511. } while(0)
  512. #define page_address_init() do { } while(0)
  513. #endif
  514. #if defined(HASHED_PAGE_VIRTUAL)
  515. void *page_address(struct page *page);
  516. void set_page_address(struct page *page, void *virtual);
  517. void page_address_init(void);
  518. #endif
  519. #if !defined(HASHED_PAGE_VIRTUAL) && !defined(WANT_PAGE_VIRTUAL)
  520. #define page_address(page) lowmem_page_address(page)
  521. #define set_page_address(page, address) do { } while(0)
  522. #define page_address_init() do { } while(0)
  523. #endif
  524. /*
  525. * On an anonymous page mapped into a user virtual memory area,
  526. * page->mapping points to its anon_vma, not to a struct address_space;
  527. * with the PAGE_MAPPING_ANON bit set to distinguish it.
  528. *
  529. * Please note that, confusingly, "page_mapping" refers to the inode
  530. * address_space which maps the page from disk; whereas "page_mapped"
  531. * refers to user virtual address space into which the page is mapped.
  532. */
  533. #define PAGE_MAPPING_ANON 1
  534. extern struct address_space swapper_space;
  535. static inline struct address_space *page_mapping(struct page *page)
  536. {
  537. struct address_space *mapping = page->mapping;
  538. VM_BUG_ON(PageSlab(page));
  539. #ifdef CONFIG_SWAP
  540. if (unlikely(PageSwapCache(page)))
  541. mapping = &swapper_space;
  542. else
  543. #endif
  544. if (unlikely((unsigned long)mapping & PAGE_MAPPING_ANON))
  545. mapping = NULL;
  546. return mapping;
  547. }
  548. static inline int PageAnon(struct page *page)
  549. {
  550. return ((unsigned long)page->mapping & PAGE_MAPPING_ANON) != 0;
  551. }
  552. /*
  553. * Return the pagecache index of the passed page. Regular pagecache pages
  554. * use ->index whereas swapcache pages use ->private
  555. */
  556. static inline pgoff_t page_index(struct page *page)
  557. {
  558. if (unlikely(PageSwapCache(page)))
  559. return page_private(page);
  560. return page->index;
  561. }
  562. /*
  563. * The atomic page->_mapcount, like _count, starts from -1:
  564. * so that transitions both from it and to it can be tracked,
  565. * using atomic_inc_and_test and atomic_add_negative(-1).
  566. */
  567. static inline void reset_page_mapcount(struct page *page)
  568. {
  569. atomic_set(&(page)->_mapcount, -1);
  570. }
  571. static inline int page_mapcount(struct page *page)
  572. {
  573. return atomic_read(&(page)->_mapcount) + 1;
  574. }
  575. /*
  576. * Return true if this page is mapped into pagetables.
  577. */
  578. static inline int page_mapped(struct page *page)
  579. {
  580. return atomic_read(&(page)->_mapcount) >= 0;
  581. }
  582. /*
  583. * Error return values for the *_nopfn functions
  584. */
  585. #define NOPFN_SIGBUS ((unsigned long) -1)
  586. #define NOPFN_OOM ((unsigned long) -2)
  587. #define NOPFN_REFAULT ((unsigned long) -3)
  588. /*
  589. * Different kinds of faults, as returned by handle_mm_fault().
  590. * Used to decide whether a process gets delivered SIGBUS or
  591. * just gets major/minor fault counters bumped up.
  592. */
  593. #define VM_FAULT_MINOR 0 /* For backwards compat. Remove me quickly. */
  594. #define VM_FAULT_OOM 0x0001
  595. #define VM_FAULT_SIGBUS 0x0002
  596. #define VM_FAULT_MAJOR 0x0004
  597. #define VM_FAULT_WRITE 0x0008 /* Special case for get_user_pages */
  598. #define VM_FAULT_NOPAGE 0x0100 /* ->fault installed the pte, not return page */
  599. #define VM_FAULT_LOCKED 0x0200 /* ->fault locked the returned page */
  600. #define VM_FAULT_ERROR (VM_FAULT_OOM | VM_FAULT_SIGBUS)
  601. #define offset_in_page(p) ((unsigned long)(p) & ~PAGE_MASK)
  602. extern void show_free_areas(void);
  603. #ifdef CONFIG_SHMEM
  604. int shmem_lock(struct file *file, int lock, struct user_struct *user);
  605. #else
  606. static inline int shmem_lock(struct file *file, int lock,
  607. struct user_struct *user)
  608. {
  609. return 0;
  610. }
  611. #endif
  612. struct file *shmem_file_setup(char *name, loff_t size, unsigned long flags);
  613. int shmem_zero_setup(struct vm_area_struct *);
  614. #ifndef CONFIG_MMU
  615. extern unsigned long shmem_get_unmapped_area(struct file *file,
  616. unsigned long addr,
  617. unsigned long len,
  618. unsigned long pgoff,
  619. unsigned long flags);
  620. #endif
  621. extern int can_do_mlock(void);
  622. extern int user_shm_lock(size_t, struct user_struct *);
  623. extern void user_shm_unlock(size_t, struct user_struct *);
  624. /*
  625. * Parameter block passed down to zap_pte_range in exceptional cases.
  626. */
  627. struct zap_details {
  628. struct vm_area_struct *nonlinear_vma; /* Check page->index if set */
  629. struct address_space *check_mapping; /* Check page->mapping if set */
  630. pgoff_t first_index; /* Lowest page->index to unmap */
  631. pgoff_t last_index; /* Highest page->index to unmap */
  632. spinlock_t *i_mmap_lock; /* For unmap_mapping_range: */
  633. unsigned long truncate_count; /* Compare vm_truncate_count */
  634. };
  635. struct page *vm_normal_page(struct vm_area_struct *vma, unsigned long addr,
  636. pte_t pte);
  637. unsigned long zap_page_range(struct vm_area_struct *vma, unsigned long address,
  638. unsigned long size, struct zap_details *);
  639. unsigned long unmap_vmas(struct mmu_gather **tlb,
  640. struct vm_area_struct *start_vma, unsigned long start_addr,
  641. unsigned long end_addr, unsigned long *nr_accounted,
  642. struct zap_details *);
  643. /**
  644. * mm_walk - callbacks for walk_page_range
  645. * @pgd_entry: if set, called for each non-empty PGD (top-level) entry
  646. * @pud_entry: if set, called for each non-empty PUD (2nd-level) entry
  647. * @pmd_entry: if set, called for each non-empty PMD (3rd-level) entry
  648. * @pte_entry: if set, called for each non-empty PTE (4th-level) entry
  649. * @pte_hole: if set, called for each hole at all levels
  650. *
  651. * (see walk_page_range for more details)
  652. */
  653. struct mm_walk {
  654. int (*pgd_entry)(pgd_t *, unsigned long, unsigned long, struct mm_walk *);
  655. int (*pud_entry)(pud_t *, unsigned long, unsigned long, struct mm_walk *);
  656. int (*pmd_entry)(pmd_t *, unsigned long, unsigned long, struct mm_walk *);
  657. int (*pte_entry)(pte_t *, unsigned long, unsigned long, struct mm_walk *);
  658. int (*pte_hole)(unsigned long, unsigned long, struct mm_walk *);
  659. struct mm_struct *mm;
  660. void *private;
  661. };
  662. int walk_page_range(unsigned long addr, unsigned long end,
  663. struct mm_walk *walk);
  664. void free_pgd_range(struct mmu_gather **tlb, unsigned long addr,
  665. unsigned long end, unsigned long floor, unsigned long ceiling);
  666. void free_pgtables(struct mmu_gather **tlb, struct vm_area_struct *start_vma,
  667. unsigned long floor, unsigned long ceiling);
  668. int copy_page_range(struct mm_struct *dst, struct mm_struct *src,
  669. struct vm_area_struct *vma);
  670. void unmap_mapping_range(struct address_space *mapping,
  671. loff_t const holebegin, loff_t const holelen, int even_cows);
  672. static inline void unmap_shared_mapping_range(struct address_space *mapping,
  673. loff_t const holebegin, loff_t const holelen)
  674. {
  675. unmap_mapping_range(mapping, holebegin, holelen, 0);
  676. }
  677. extern int vmtruncate(struct inode * inode, loff_t offset);
  678. extern int vmtruncate_range(struct inode * inode, loff_t offset, loff_t end);
  679. #ifdef CONFIG_MMU
  680. extern int handle_mm_fault(struct mm_struct *mm, struct vm_area_struct *vma,
  681. unsigned long address, int write_access);
  682. #else
  683. static inline int handle_mm_fault(struct mm_struct *mm,
  684. struct vm_area_struct *vma, unsigned long address,
  685. int write_access)
  686. {
  687. /* should never happen if there's no MMU */
  688. BUG();
  689. return VM_FAULT_SIGBUS;
  690. }
  691. #endif
  692. extern int make_pages_present(unsigned long addr, unsigned long end);
  693. extern int access_process_vm(struct task_struct *tsk, unsigned long addr, void *buf, int len, int write);
  694. int get_user_pages(struct task_struct *tsk, struct mm_struct *mm, unsigned long start,
  695. int len, int write, int force, struct page **pages, struct vm_area_struct **vmas);
  696. void print_bad_pte(struct vm_area_struct *, pte_t, unsigned long);
  697. extern int try_to_release_page(struct page * page, gfp_t gfp_mask);
  698. extern void do_invalidatepage(struct page *page, unsigned long offset);
  699. int __set_page_dirty_nobuffers(struct page *page);
  700. int __set_page_dirty_no_writeback(struct page *page);
  701. int redirty_page_for_writepage(struct writeback_control *wbc,
  702. struct page *page);
  703. int set_page_dirty(struct page *page);
  704. int set_page_dirty_lock(struct page *page);
  705. int clear_page_dirty_for_io(struct page *page);
  706. extern unsigned long move_page_tables(struct vm_area_struct *vma,
  707. unsigned long old_addr, struct vm_area_struct *new_vma,
  708. unsigned long new_addr, unsigned long len);
  709. extern unsigned long do_mremap(unsigned long addr,
  710. unsigned long old_len, unsigned long new_len,
  711. unsigned long flags, unsigned long new_addr);
  712. extern int mprotect_fixup(struct vm_area_struct *vma,
  713. struct vm_area_struct **pprev, unsigned long start,
  714. unsigned long end, unsigned long newflags);
  715. /*
  716. * A callback you can register to apply pressure to ageable caches.
  717. *
  718. * 'shrink' is passed a count 'nr_to_scan' and a 'gfpmask'. It should
  719. * look through the least-recently-used 'nr_to_scan' entries and
  720. * attempt to free them up. It should return the number of objects
  721. * which remain in the cache. If it returns -1, it means it cannot do
  722. * any scanning at this time (eg. there is a risk of deadlock).
  723. *
  724. * The 'gfpmask' refers to the allocation we are currently trying to
  725. * fulfil.
  726. *
  727. * Note that 'shrink' will be passed nr_to_scan == 0 when the VM is
  728. * querying the cache size, so a fastpath for that case is appropriate.
  729. */
  730. struct shrinker {
  731. int (*shrink)(int nr_to_scan, gfp_t gfp_mask);
  732. int seeks; /* seeks to recreate an obj */
  733. /* These are for internal use */
  734. struct list_head list;
  735. long nr; /* objs pending delete */
  736. };
  737. #define DEFAULT_SEEKS 2 /* A good number if you don't know better. */
  738. extern void register_shrinker(struct shrinker *);
  739. extern void unregister_shrinker(struct shrinker *);
  740. int vma_wants_writenotify(struct vm_area_struct *vma);
  741. extern pte_t *get_locked_pte(struct mm_struct *mm, unsigned long addr, spinlock_t **ptl);
  742. #ifdef __PAGETABLE_PUD_FOLDED
  743. static inline int __pud_alloc(struct mm_struct *mm, pgd_t *pgd,
  744. unsigned long address)
  745. {
  746. return 0;
  747. }
  748. #else
  749. int __pud_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address);
  750. #endif
  751. #ifdef __PAGETABLE_PMD_FOLDED
  752. static inline int __pmd_alloc(struct mm_struct *mm, pud_t *pud,
  753. unsigned long address)
  754. {
  755. return 0;
  756. }
  757. #else
  758. int __pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address);
  759. #endif
  760. int __pte_alloc(struct mm_struct *mm, pmd_t *pmd, unsigned long address);
  761. int __pte_alloc_kernel(pmd_t *pmd, unsigned long address);
  762. /*
  763. * The following ifdef needed to get the 4level-fixup.h header to work.
  764. * Remove it when 4level-fixup.h has been removed.
  765. */
  766. #if defined(CONFIG_MMU) && !defined(__ARCH_HAS_4LEVEL_HACK)
  767. static inline pud_t *pud_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address)
  768. {
  769. return (unlikely(pgd_none(*pgd)) && __pud_alloc(mm, pgd, address))?
  770. NULL: pud_offset(pgd, address);
  771. }
  772. static inline pmd_t *pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address)
  773. {
  774. return (unlikely(pud_none(*pud)) && __pmd_alloc(mm, pud, address))?
  775. NULL: pmd_offset(pud, address);
  776. }
  777. #endif /* CONFIG_MMU && !__ARCH_HAS_4LEVEL_HACK */
  778. #if NR_CPUS >= CONFIG_SPLIT_PTLOCK_CPUS
  779. /*
  780. * We tuck a spinlock to guard each pagetable page into its struct page,
  781. * at page->private, with BUILD_BUG_ON to make sure that this will not
  782. * overflow into the next struct page (as it might with DEBUG_SPINLOCK).
  783. * When freeing, reset page->mapping so free_pages_check won't complain.
  784. */
  785. #define __pte_lockptr(page) &((page)->ptl)
  786. #define pte_lock_init(_page) do { \
  787. spin_lock_init(__pte_lockptr(_page)); \
  788. } while (0)
  789. #define pte_lock_deinit(page) ((page)->mapping = NULL)
  790. #define pte_lockptr(mm, pmd) ({(void)(mm); __pte_lockptr(pmd_page(*(pmd)));})
  791. #else
  792. /*
  793. * We use mm->page_table_lock to guard all pagetable pages of the mm.
  794. */
  795. #define pte_lock_init(page) do {} while (0)
  796. #define pte_lock_deinit(page) do {} while (0)
  797. #define pte_lockptr(mm, pmd) ({(void)(pmd); &(mm)->page_table_lock;})
  798. #endif /* NR_CPUS < CONFIG_SPLIT_PTLOCK_CPUS */
  799. static inline void pgtable_page_ctor(struct page *page)
  800. {
  801. pte_lock_init(page);
  802. inc_zone_page_state(page, NR_PAGETABLE);
  803. }
  804. static inline void pgtable_page_dtor(struct page *page)
  805. {
  806. pte_lock_deinit(page);
  807. dec_zone_page_state(page, NR_PAGETABLE);
  808. }
  809. #define pte_offset_map_lock(mm, pmd, address, ptlp) \
  810. ({ \
  811. spinlock_t *__ptl = pte_lockptr(mm, pmd); \
  812. pte_t *__pte = pte_offset_map(pmd, address); \
  813. *(ptlp) = __ptl; \
  814. spin_lock(__ptl); \
  815. __pte; \
  816. })
  817. #define pte_unmap_unlock(pte, ptl) do { \
  818. spin_unlock(ptl); \
  819. pte_unmap(pte); \
  820. } while (0)
  821. #define pte_alloc_map(mm, pmd, address) \
  822. ((unlikely(!pmd_present(*(pmd))) && __pte_alloc(mm, pmd, address))? \
  823. NULL: pte_offset_map(pmd, address))
  824. #define pte_alloc_map_lock(mm, pmd, address, ptlp) \
  825. ((unlikely(!pmd_present(*(pmd))) && __pte_alloc(mm, pmd, address))? \
  826. NULL: pte_offset_map_lock(mm, pmd, address, ptlp))
  827. #define pte_alloc_kernel(pmd, address) \
  828. ((unlikely(!pmd_present(*(pmd))) && __pte_alloc_kernel(pmd, address))? \
  829. NULL: pte_offset_kernel(pmd, address))
  830. extern void free_area_init(unsigned long * zones_size);
  831. extern void free_area_init_node(int nid, pg_data_t *pgdat,
  832. unsigned long * zones_size, unsigned long zone_start_pfn,
  833. unsigned long *zholes_size);
  834. #ifdef CONFIG_ARCH_POPULATES_NODE_MAP
  835. /*
  836. * With CONFIG_ARCH_POPULATES_NODE_MAP set, an architecture may initialise its
  837. * zones, allocate the backing mem_map and account for memory holes in a more
  838. * architecture independent manner. This is a substitute for creating the
  839. * zone_sizes[] and zholes_size[] arrays and passing them to
  840. * free_area_init_node()
  841. *
  842. * An architecture is expected to register range of page frames backed by
  843. * physical memory with add_active_range() before calling
  844. * free_area_init_nodes() passing in the PFN each zone ends at. At a basic
  845. * usage, an architecture is expected to do something like
  846. *
  847. * unsigned long max_zone_pfns[MAX_NR_ZONES] = {max_dma, max_normal_pfn,
  848. * max_highmem_pfn};
  849. * for_each_valid_physical_page_range()
  850. * add_active_range(node_id, start_pfn, end_pfn)
  851. * free_area_init_nodes(max_zone_pfns);
  852. *
  853. * If the architecture guarantees that there are no holes in the ranges
  854. * registered with add_active_range(), free_bootmem_active_regions()
  855. * will call free_bootmem_node() for each registered physical page range.
  856. * Similarly sparse_memory_present_with_active_regions() calls
  857. * memory_present() for each range when SPARSEMEM is enabled.
  858. *
  859. * See mm/page_alloc.c for more information on each function exposed by
  860. * CONFIG_ARCH_POPULATES_NODE_MAP
  861. */
  862. extern void free_area_init_nodes(unsigned long *max_zone_pfn);
  863. extern void add_active_range(unsigned int nid, unsigned long start_pfn,
  864. unsigned long end_pfn);
  865. extern void shrink_active_range(unsigned int nid, unsigned long old_end_pfn,
  866. unsigned long new_end_pfn);
  867. extern void push_node_boundaries(unsigned int nid, unsigned long start_pfn,
  868. unsigned long end_pfn);
  869. extern void remove_all_active_ranges(void);
  870. extern unsigned long absent_pages_in_range(unsigned long start_pfn,
  871. unsigned long end_pfn);
  872. extern void get_pfn_range_for_nid(unsigned int nid,
  873. unsigned long *start_pfn, unsigned long *end_pfn);
  874. extern unsigned long find_min_pfn_with_active_regions(void);
  875. extern unsigned long find_max_pfn_with_active_regions(void);
  876. extern void free_bootmem_with_active_regions(int nid,
  877. unsigned long max_low_pfn);
  878. extern void sparse_memory_present_with_active_regions(int nid);
  879. #ifndef CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID
  880. extern int early_pfn_to_nid(unsigned long pfn);
  881. #endif /* CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID */
  882. #endif /* CONFIG_ARCH_POPULATES_NODE_MAP */
  883. extern void set_dma_reserve(unsigned long new_dma_reserve);
  884. extern void memmap_init_zone(unsigned long, int, unsigned long,
  885. unsigned long, enum memmap_context);
  886. extern void setup_per_zone_pages_min(void);
  887. extern void mem_init(void);
  888. extern void show_mem(void);
  889. extern void si_meminfo(struct sysinfo * val);
  890. extern void si_meminfo_node(struct sysinfo *val, int nid);
  891. #ifdef CONFIG_NUMA
  892. extern void setup_per_cpu_pageset(void);
  893. #else
  894. static inline void setup_per_cpu_pageset(void) {}
  895. #endif
  896. /* prio_tree.c */
  897. void vma_prio_tree_add(struct vm_area_struct *, struct vm_area_struct *old);
  898. void vma_prio_tree_insert(struct vm_area_struct *, struct prio_tree_root *);
  899. void vma_prio_tree_remove(struct vm_area_struct *, struct prio_tree_root *);
  900. struct vm_area_struct *vma_prio_tree_next(struct vm_area_struct *vma,
  901. struct prio_tree_iter *iter);
  902. #define vma_prio_tree_foreach(vma, iter, root, begin, end) \
  903. for (prio_tree_iter_init(iter, root, begin, end), vma = NULL; \
  904. (vma = vma_prio_tree_next(vma, iter)); )
  905. static inline void vma_nonlinear_insert(struct vm_area_struct *vma,
  906. struct list_head *list)
  907. {
  908. vma->shared.vm_set.parent = NULL;
  909. list_add_tail(&vma->shared.vm_set.list, list);
  910. }
  911. /* mmap.c */
  912. extern int __vm_enough_memory(struct mm_struct *mm, long pages, int cap_sys_admin);
  913. extern void vma_adjust(struct vm_area_struct *vma, unsigned long start,
  914. unsigned long end, pgoff_t pgoff, struct vm_area_struct *insert);
  915. extern struct vm_area_struct *vma_merge(struct mm_struct *,
  916. struct vm_area_struct *prev, unsigned long addr, unsigned long end,
  917. unsigned long vm_flags, struct anon_vma *, struct file *, pgoff_t,
  918. struct mempolicy *);
  919. extern struct anon_vma *find_mergeable_anon_vma(struct vm_area_struct *);
  920. extern int split_vma(struct mm_struct *,
  921. struct vm_area_struct *, unsigned long addr, int new_below);
  922. extern int insert_vm_struct(struct mm_struct *, struct vm_area_struct *);
  923. extern void __vma_link_rb(struct mm_struct *, struct vm_area_struct *,
  924. struct rb_node **, struct rb_node *);
  925. extern void unlink_file_vma(struct vm_area_struct *);
  926. extern struct vm_area_struct *copy_vma(struct vm_area_struct **,
  927. unsigned long addr, unsigned long len, pgoff_t pgoff);
  928. extern void exit_mmap(struct mm_struct *);
  929. #ifdef CONFIG_PROC_FS
  930. /* From fs/proc/base.c. callers must _not_ hold the mm's exe_file_lock */
  931. extern void added_exe_file_vma(struct mm_struct *mm);
  932. extern void removed_exe_file_vma(struct mm_struct *mm);
  933. #else
  934. static inline void added_exe_file_vma(struct mm_struct *mm)
  935. {}
  936. static inline void removed_exe_file_vma(struct mm_struct *mm)
  937. {}
  938. #endif /* CONFIG_PROC_FS */
  939. extern int may_expand_vm(struct mm_struct *mm, unsigned long npages);
  940. extern int install_special_mapping(struct mm_struct *mm,
  941. unsigned long addr, unsigned long len,
  942. unsigned long flags, struct page **pages);
  943. extern unsigned long get_unmapped_area(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
  944. extern unsigned long do_mmap_pgoff(struct file *file, unsigned long addr,
  945. unsigned long len, unsigned long prot,
  946. unsigned long flag, unsigned long pgoff);
  947. extern unsigned long mmap_region(struct file *file, unsigned long addr,
  948. unsigned long len, unsigned long flags,
  949. unsigned int vm_flags, unsigned long pgoff,
  950. int accountable);
  951. static inline unsigned long do_mmap(struct file *file, unsigned long addr,
  952. unsigned long len, unsigned long prot,
  953. unsigned long flag, unsigned long offset)
  954. {
  955. unsigned long ret = -EINVAL;
  956. if ((offset + PAGE_ALIGN(len)) < offset)
  957. goto out;
  958. if (!(offset & ~PAGE_MASK))
  959. ret = do_mmap_pgoff(file, addr, len, prot, flag, offset >> PAGE_SHIFT);
  960. out:
  961. return ret;
  962. }
  963. extern int do_munmap(struct mm_struct *, unsigned long, size_t);
  964. extern unsigned long do_brk(unsigned long, unsigned long);
  965. /* filemap.c */
  966. extern unsigned long page_unuse(struct page *);
  967. extern void truncate_inode_pages(struct address_space *, loff_t);
  968. extern void truncate_inode_pages_range(struct address_space *,
  969. loff_t lstart, loff_t lend);
  970. /* generic vm_area_ops exported for stackable file systems */
  971. extern int filemap_fault(struct vm_area_struct *, struct vm_fault *);
  972. /* mm/page-writeback.c */
  973. int write_one_page(struct page *page, int wait);
  974. /* readahead.c */
  975. #define VM_MAX_READAHEAD 128 /* kbytes */
  976. #define VM_MIN_READAHEAD 16 /* kbytes (includes current page) */
  977. int do_page_cache_readahead(struct address_space *mapping, struct file *filp,
  978. pgoff_t offset, unsigned long nr_to_read);
  979. int force_page_cache_readahead(struct address_space *mapping, struct file *filp,
  980. pgoff_t offset, unsigned long nr_to_read);
  981. void page_cache_sync_readahead(struct address_space *mapping,
  982. struct file_ra_state *ra,
  983. struct file *filp,
  984. pgoff_t offset,
  985. unsigned long size);
  986. void page_cache_async_readahead(struct address_space *mapping,
  987. struct file_ra_state *ra,
  988. struct file *filp,
  989. struct page *pg,
  990. pgoff_t offset,
  991. unsigned long size);
  992. unsigned long max_sane_readahead(unsigned long nr);
  993. /* Do stack extension */
  994. extern int expand_stack(struct vm_area_struct *vma, unsigned long address);
  995. #ifdef CONFIG_IA64
  996. extern int expand_upwards(struct vm_area_struct *vma, unsigned long address);
  997. #endif
  998. extern int expand_stack_downwards(struct vm_area_struct *vma,
  999. unsigned long address);
  1000. /* Look up the first VMA which satisfies addr < vm_end, NULL if none. */
  1001. extern struct vm_area_struct * find_vma(struct mm_struct * mm, unsigned long addr);
  1002. extern struct vm_area_struct * find_vma_prev(struct mm_struct * mm, unsigned long addr,
  1003. struct vm_area_struct **pprev);
  1004. /* Look up the first VMA which intersects the interval start_addr..end_addr-1,
  1005. NULL if none. Assume start_addr < end_addr. */
  1006. static inline struct vm_area_struct * find_vma_intersection(struct mm_struct * mm, unsigned long start_addr, unsigned long end_addr)
  1007. {
  1008. struct vm_area_struct * vma = find_vma(mm,start_addr);
  1009. if (vma && end_addr <= vma->vm_start)
  1010. vma = NULL;
  1011. return vma;
  1012. }
  1013. static inline unsigned long vma_pages(struct vm_area_struct *vma)
  1014. {
  1015. return (vma->vm_end - vma->vm_start) >> PAGE_SHIFT;
  1016. }
  1017. pgprot_t vm_get_page_prot(unsigned long vm_flags);
  1018. struct vm_area_struct *find_extend_vma(struct mm_struct *, unsigned long addr);
  1019. int remap_pfn_range(struct vm_area_struct *, unsigned long addr,
  1020. unsigned long pfn, unsigned long size, pgprot_t);
  1021. int vm_insert_page(struct vm_area_struct *, unsigned long addr, struct page *);
  1022. int vm_insert_pfn(struct vm_area_struct *vma, unsigned long addr,
  1023. unsigned long pfn);
  1024. int vm_insert_mixed(struct vm_area_struct *vma, unsigned long addr,
  1025. unsigned long pfn);
  1026. struct page *follow_page(struct vm_area_struct *, unsigned long address,
  1027. unsigned int foll_flags);
  1028. #define FOLL_WRITE 0x01 /* check pte is writable */
  1029. #define FOLL_TOUCH 0x02 /* mark page accessed */
  1030. #define FOLL_GET 0x04 /* do get_page on page */
  1031. #define FOLL_ANON 0x08 /* give ZERO_PAGE if no pgtable */
  1032. typedef int (*pte_fn_t)(pte_t *pte, pgtable_t token, unsigned long addr,
  1033. void *data);
  1034. extern int apply_to_page_range(struct mm_struct *mm, unsigned long address,
  1035. unsigned long size, pte_fn_t fn, void *data);
  1036. #ifdef CONFIG_PROC_FS
  1037. void vm_stat_account(struct mm_struct *, unsigned long, struct file *, long);
  1038. #else
  1039. static inline void vm_stat_account(struct mm_struct *mm,
  1040. unsigned long flags, struct file *file, long pages)
  1041. {
  1042. }
  1043. #endif /* CONFIG_PROC_FS */
  1044. #ifdef CONFIG_DEBUG_PAGEALLOC
  1045. extern int debug_pagealloc_enabled;
  1046. extern void kernel_map_pages(struct page *page, int numpages, int enable);
  1047. static inline void enable_debug_pagealloc(void)
  1048. {
  1049. debug_pagealloc_enabled = 1;
  1050. }
  1051. #ifdef CONFIG_HIBERNATION
  1052. extern bool kernel_page_present(struct page *page);
  1053. #endif /* CONFIG_HIBERNATION */
  1054. #else
  1055. static inline void
  1056. kernel_map_pages(struct page *page, int numpages, int enable) {}
  1057. static inline void enable_debug_pagealloc(void)
  1058. {
  1059. }
  1060. #ifdef CONFIG_HIBERNATION
  1061. static inline bool kernel_page_present(struct page *page) { return true; }
  1062. #endif /* CONFIG_HIBERNATION */
  1063. #endif
  1064. extern struct vm_area_struct *get_gate_vma(struct task_struct *tsk);
  1065. #ifdef __HAVE_ARCH_GATE_AREA
  1066. int in_gate_area_no_task(unsigned long addr);
  1067. int in_gate_area(struct task_struct *task, unsigned long addr);
  1068. #else
  1069. int in_gate_area_no_task(unsigned long addr);
  1070. #define in_gate_area(task, addr) ({(void)task; in_gate_area_no_task(addr);})
  1071. #endif /* __HAVE_ARCH_GATE_AREA */
  1072. int drop_caches_sysctl_handler(struct ctl_table *, int, struct file *,
  1073. void __user *, size_t *, loff_t *);
  1074. unsigned long shrink_slab(unsigned long scanned, gfp_t gfp_mask,
  1075. unsigned long lru_pages);
  1076. #ifndef CONFIG_MMU
  1077. #define randomize_va_space 0
  1078. #else
  1079. extern int randomize_va_space;
  1080. #endif
  1081. const char * arch_vma_name(struct vm_area_struct *vma);
  1082. void print_vma_addr(char *prefix, unsigned long rip);
  1083. struct page *sparse_mem_map_populate(unsigned long pnum, int nid);
  1084. pgd_t *vmemmap_pgd_populate(unsigned long addr, int node);
  1085. pud_t *vmemmap_pud_populate(pgd_t *pgd, unsigned long addr, int node);
  1086. pmd_t *vmemmap_pmd_populate(pud_t *pud, unsigned long addr, int node);
  1087. pte_t *vmemmap_pte_populate(pmd_t *pmd, unsigned long addr, int node);
  1088. void *vmemmap_alloc_block(unsigned long size, int node);
  1089. void vmemmap_verify(pte_t *, int, unsigned long, unsigned long);
  1090. int vmemmap_populate_basepages(struct page *start_page,
  1091. unsigned long pages, int node);
  1092. int vmemmap_populate(struct page *start_page, unsigned long pages, int node);
  1093. void vmemmap_populate_print_last(void);
  1094. #endif /* __KERNEL__ */
  1095. #endif /* _LINUX_MM_H */