mm.h 47 KB

12345678910111213141516171819202122232425262728293031323334353637383940414243444546474849505152535455565758596061626364656667686970717273747576777879808182838485868788899091929394959697989910010110210310410510610710810911011111211311411511611711811912012112212312412512612712812913013113213313413513613713813914014114214314414514614714814915015115215315415515615715815916016116216316416516616716816917017117217317417517617717817918018118218318418518618718818919019119219319419519619719819920020120220320420520620720820921021121221321421521621721821922022122222322422522622722822923023123223323423523623723823924024124224324424524624724824925025125225325425525625725825926026126226326426526626726826927027127227327427527627727827928028128228328428528628728828929029129229329429529629729829930030130230330430530630730830931031131231331431531631731831932032132232332432532632732832933033133233333433533633733833934034134234334434534634734834935035135235335435535635735835936036136236336436536636736836937037137237337437537637737837938038138238338438538638738838939039139239339439539639739839940040140240340440540640740840941041141241341441541641741841942042142242342442542642742842943043143243343443543643743843944044144244344444544644744844945045145245345445545645745845946046146246346446546646746846947047147247347447547647747847948048148248348448548648748848949049149249349449549649749849950050150250350450550650750850951051151251351451551651751851952052152252352452552652752852953053153253353453553653753853954054154254354454554654754854955055155255355455555655755855956056156256356456556656756856957057157257357457557657757857958058158258358458558658758858959059159259359459559659759859960060160260360460560660760860961061161261361461561661761861962062162262362462562662762862963063163263363463563663763863964064164264364464564664764864965065165265365465565665765865966066166266366466566666766866967067167267367467567667767867968068168268368468568668768868969069169269369469569669769869970070170270370470570670770870971071171271371471571671771871972072172272372472572672772872973073173273373473573673773873974074174274374474574674774874975075175275375475575675775875976076176276376476576676776876977077177277377477577677777877978078178278378478578678778878979079179279379479579679779879980080180280380480580680780880981081181281381481581681781881982082182282382482582682782882983083183283383483583683783883984084184284384484584684784884985085185285385485585685785885986086186286386486586686786886987087187287387487587687787887988088188288388488588688788888989089189289389489589689789889990090190290390490590690790890991091191291391491591691791891992092192292392492592692792892993093193293393493593693793893994094194294394494594694794894995095195295395495595695795895996096196296396496596696796896997097197297397497597697797897998098198298398498598698798898999099199299399499599699799899910001001100210031004100510061007100810091010101110121013101410151016101710181019102010211022102310241025102610271028102910301031103210331034103510361037103810391040104110421043104410451046104710481049105010511052105310541055105610571058105910601061106210631064106510661067106810691070107110721073107410751076107710781079108010811082108310841085108610871088108910901091109210931094109510961097109810991100110111021103110411051106110711081109111011111112111311141115111611171118111911201121112211231124112511261127112811291130113111321133113411351136113711381139114011411142114311441145114611471148114911501151115211531154115511561157115811591160116111621163116411651166116711681169117011711172117311741175117611771178117911801181118211831184118511861187118811891190119111921193119411951196119711981199120012011202120312041205120612071208120912101211121212131214121512161217121812191220122112221223122412251226122712281229123012311232123312341235123612371238123912401241124212431244124512461247124812491250125112521253125412551256125712581259126012611262126312641265126612671268126912701271127212731274127512761277127812791280128112821283128412851286128712881289129012911292129312941295129612971298129913001301130213031304130513061307130813091310131113121313131413151316131713181319132013211322132313241325132613271328132913301331133213331334133513361337133813391340134113421343134413451346134713481349135013511352135313541355135613571358135913601361
  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. #include <linux/range.h>
  13. struct mempolicy;
  14. struct anon_vma;
  15. struct file_ra_state;
  16. struct user_struct;
  17. struct writeback_control;
  18. struct rlimit;
  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 unsigned long totalram_pages;
  24. extern void * high_memory;
  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. /* to align the pointer to the (next) page boundary */
  36. #define PAGE_ALIGN(addr) ALIGN(addr, PAGE_SIZE)
  37. /*
  38. * Linux kernel virtual memory manager primitives.
  39. * The idea being to have a "virtual" mm in the same way
  40. * we have a virtual fs - giving a cleaner interface to the
  41. * mm details, and allowing different kinds of memory mappings
  42. * (from shared memory to executable loading to arbitrary
  43. * mmap() functions).
  44. */
  45. extern struct kmem_cache *vm_area_cachep;
  46. #ifndef CONFIG_MMU
  47. extern struct rb_root nommu_region_tree;
  48. extern struct rw_semaphore nommu_region_sem;
  49. extern unsigned int kobjsize(const void *objp);
  50. #endif
  51. /*
  52. * vm_flags in vm_area_struct, see mm_types.h.
  53. */
  54. #define VM_READ 0x00000001 /* currently active flags */
  55. #define VM_WRITE 0x00000002
  56. #define VM_EXEC 0x00000004
  57. #define VM_SHARED 0x00000008
  58. /* mprotect() hardcodes VM_MAYREAD >> 4 == VM_READ, and so for r/w/x bits. */
  59. #define VM_MAYREAD 0x00000010 /* limits for mprotect() etc */
  60. #define VM_MAYWRITE 0x00000020
  61. #define VM_MAYEXEC 0x00000040
  62. #define VM_MAYSHARE 0x00000080
  63. #define VM_GROWSDOWN 0x00000100 /* general info on the segment */
  64. #define VM_GROWSUP 0x00000200
  65. #define VM_PFNMAP 0x00000400 /* Page-ranges managed without "struct page", just pure PFN */
  66. #define VM_DENYWRITE 0x00000800 /* ETXTBSY on write attempts.. */
  67. #define VM_EXECUTABLE 0x00001000
  68. #define VM_LOCKED 0x00002000
  69. #define VM_IO 0x00004000 /* Memory mapped I/O or similar */
  70. /* Used by sys_madvise() */
  71. #define VM_SEQ_READ 0x00008000 /* App will access data sequentially */
  72. #define VM_RAND_READ 0x00010000 /* App will not benefit from clustered reads */
  73. #define VM_DONTCOPY 0x00020000 /* Do not copy this vma on fork */
  74. #define VM_DONTEXPAND 0x00040000 /* Cannot expand with mremap() */
  75. #define VM_RESERVED 0x00080000 /* Count as reserved_vm like IO */
  76. #define VM_ACCOUNT 0x00100000 /* Is a VM accounted object */
  77. #define VM_NORESERVE 0x00200000 /* should the VM suppress accounting */
  78. #define VM_HUGETLB 0x00400000 /* Huge TLB Page VM */
  79. #define VM_NONLINEAR 0x00800000 /* Is non-linear (remap_file_pages) */
  80. #define VM_MAPPED_COPY 0x01000000 /* T if mapped copy of data (nommu mmap) */
  81. #define VM_INSERTPAGE 0x02000000 /* The vma has had "vm_insert_page()" done on it */
  82. #define VM_ALWAYSDUMP 0x04000000 /* Always include in core dumps */
  83. #define VM_CAN_NONLINEAR 0x08000000 /* Has ->fault & does nonlinear pages */
  84. #define VM_MIXEDMAP 0x10000000 /* Can contain "struct page" and pure PFN pages */
  85. #define VM_SAO 0x20000000 /* Strong Access Ordering (powerpc) */
  86. #define VM_PFN_AT_MMAP 0x40000000 /* PFNMAP vma that is fully mapped at mmap time */
  87. #define VM_MERGEABLE 0x80000000 /* KSM may merge identical pages */
  88. #ifndef VM_STACK_DEFAULT_FLAGS /* arch can override this */
  89. #define VM_STACK_DEFAULT_FLAGS VM_DATA_DEFAULT_FLAGS
  90. #endif
  91. #ifdef CONFIG_STACK_GROWSUP
  92. #define VM_STACK_FLAGS (VM_GROWSUP | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT)
  93. #else
  94. #define VM_STACK_FLAGS (VM_GROWSDOWN | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT)
  95. #endif
  96. #define VM_READHINTMASK (VM_SEQ_READ | VM_RAND_READ)
  97. #define VM_ClearReadHint(v) (v)->vm_flags &= ~VM_READHINTMASK
  98. #define VM_NormalReadHint(v) (!((v)->vm_flags & VM_READHINTMASK))
  99. #define VM_SequentialReadHint(v) ((v)->vm_flags & VM_SEQ_READ)
  100. #define VM_RandomReadHint(v) ((v)->vm_flags & VM_RAND_READ)
  101. /*
  102. * special vmas that are non-mergable, non-mlock()able
  103. */
  104. #define VM_SPECIAL (VM_IO | VM_DONTEXPAND | VM_RESERVED | VM_PFNMAP)
  105. /*
  106. * mapping from the currently active vm_flags protection bits (the
  107. * low four bits) to a page protection mask..
  108. */
  109. extern pgprot_t protection_map[16];
  110. #define FAULT_FLAG_WRITE 0x01 /* Fault was a write access */
  111. #define FAULT_FLAG_NONLINEAR 0x02 /* Fault was via a nonlinear mapping */
  112. #define FAULT_FLAG_MKWRITE 0x04 /* Fault was mkwrite of existing pte */
  113. /*
  114. * This interface is used by x86 PAT code to identify a pfn mapping that is
  115. * linear over entire vma. This is to optimize PAT code that deals with
  116. * marking the physical region with a particular prot. This is not for generic
  117. * mm use. Note also that this check will not work if the pfn mapping is
  118. * linear for a vma starting at physical address 0. In which case PAT code
  119. * falls back to slow path of reserving physical range page by page.
  120. */
  121. static inline int is_linear_pfn_mapping(struct vm_area_struct *vma)
  122. {
  123. return (vma->vm_flags & VM_PFN_AT_MMAP);
  124. }
  125. static inline int is_pfn_mapping(struct vm_area_struct *vma)
  126. {
  127. return (vma->vm_flags & VM_PFNMAP);
  128. }
  129. /*
  130. * vm_fault is filled by the the pagefault handler and passed to the vma's
  131. * ->fault function. The vma's ->fault is responsible for returning a bitmask
  132. * of VM_FAULT_xxx flags that give details about how the fault was handled.
  133. *
  134. * pgoff should be used in favour of virtual_address, if possible. If pgoff
  135. * is used, one may set VM_CAN_NONLINEAR in the vma->vm_flags to get nonlinear
  136. * mapping support.
  137. */
  138. struct vm_fault {
  139. unsigned int flags; /* FAULT_FLAG_xxx flags */
  140. pgoff_t pgoff; /* Logical page offset based on vma */
  141. void __user *virtual_address; /* Faulting virtual address */
  142. struct page *page; /* ->fault handlers should return a
  143. * page here, unless VM_FAULT_NOPAGE
  144. * is set (which is also implied by
  145. * VM_FAULT_ERROR).
  146. */
  147. };
  148. /*
  149. * These are the virtual MM functions - opening of an area, closing and
  150. * unmapping it (needed to keep files on disk up-to-date etc), pointer
  151. * to the functions called when a no-page or a wp-page exception occurs.
  152. */
  153. struct vm_operations_struct {
  154. void (*open)(struct vm_area_struct * area);
  155. void (*close)(struct vm_area_struct * area);
  156. int (*fault)(struct vm_area_struct *vma, struct vm_fault *vmf);
  157. /* notification that a previously read-only page is about to become
  158. * writable, if an error is returned it will cause a SIGBUS */
  159. int (*page_mkwrite)(struct vm_area_struct *vma, struct vm_fault *vmf);
  160. /* called by access_process_vm when get_user_pages() fails, typically
  161. * for use by special VMAs that can switch between memory and hardware
  162. */
  163. int (*access)(struct vm_area_struct *vma, unsigned long addr,
  164. void *buf, int len, int write);
  165. #ifdef CONFIG_NUMA
  166. /*
  167. * set_policy() op must add a reference to any non-NULL @new mempolicy
  168. * to hold the policy upon return. Caller should pass NULL @new to
  169. * remove a policy and fall back to surrounding context--i.e. do not
  170. * install a MPOL_DEFAULT policy, nor the task or system default
  171. * mempolicy.
  172. */
  173. int (*set_policy)(struct vm_area_struct *vma, struct mempolicy *new);
  174. /*
  175. * get_policy() op must add reference [mpol_get()] to any policy at
  176. * (vma,addr) marked as MPOL_SHARED. The shared policy infrastructure
  177. * in mm/mempolicy.c will do this automatically.
  178. * get_policy() must NOT add a ref if the policy at (vma,addr) is not
  179. * marked as MPOL_SHARED. vma policies are protected by the mmap_sem.
  180. * If no [shared/vma] mempolicy exists at the addr, get_policy() op
  181. * must return NULL--i.e., do not "fallback" to task or system default
  182. * policy.
  183. */
  184. struct mempolicy *(*get_policy)(struct vm_area_struct *vma,
  185. unsigned long addr);
  186. int (*migrate)(struct vm_area_struct *vma, const nodemask_t *from,
  187. const nodemask_t *to, unsigned long flags);
  188. #endif
  189. };
  190. struct mmu_gather;
  191. struct inode;
  192. #define page_private(page) ((page)->private)
  193. #define set_page_private(page, v) ((page)->private = (v))
  194. /*
  195. * FIXME: take this include out, include page-flags.h in
  196. * files which need it (119 of them)
  197. */
  198. #include <linux/page-flags.h>
  199. /*
  200. * Methods to modify the page usage count.
  201. *
  202. * What counts for a page usage:
  203. * - cache mapping (page->mapping)
  204. * - private data (page->private)
  205. * - page mapped in a task's page tables, each mapping
  206. * is counted separately
  207. *
  208. * Also, many kernel routines increase the page count before a critical
  209. * routine so they can be sure the page doesn't go away from under them.
  210. */
  211. /*
  212. * Drop a ref, return true if the refcount fell to zero (the page has no users)
  213. */
  214. static inline int put_page_testzero(struct page *page)
  215. {
  216. VM_BUG_ON(atomic_read(&page->_count) == 0);
  217. return atomic_dec_and_test(&page->_count);
  218. }
  219. /*
  220. * Try to grab a ref unless the page has a refcount of zero, return false if
  221. * that is the case.
  222. */
  223. static inline int get_page_unless_zero(struct page *page)
  224. {
  225. return atomic_inc_not_zero(&page->_count);
  226. }
  227. extern int page_is_ram(unsigned long pfn);
  228. /* Support for virtually mapped pages */
  229. struct page *vmalloc_to_page(const void *addr);
  230. unsigned long vmalloc_to_pfn(const void *addr);
  231. /*
  232. * Determine if an address is within the vmalloc range
  233. *
  234. * On nommu, vmalloc/vfree wrap through kmalloc/kfree directly, so there
  235. * is no special casing required.
  236. */
  237. static inline int is_vmalloc_addr(const void *x)
  238. {
  239. #ifdef CONFIG_MMU
  240. unsigned long addr = (unsigned long)x;
  241. return addr >= VMALLOC_START && addr < VMALLOC_END;
  242. #else
  243. return 0;
  244. #endif
  245. }
  246. #ifdef CONFIG_MMU
  247. extern int is_vmalloc_or_module_addr(const void *x);
  248. #else
  249. static inline int is_vmalloc_or_module_addr(const void *x)
  250. {
  251. return 0;
  252. }
  253. #endif
  254. static inline struct page *compound_head(struct page *page)
  255. {
  256. if (unlikely(PageTail(page)))
  257. return page->first_page;
  258. return page;
  259. }
  260. static inline int page_count(struct page *page)
  261. {
  262. return atomic_read(&compound_head(page)->_count);
  263. }
  264. static inline void get_page(struct page *page)
  265. {
  266. page = compound_head(page);
  267. VM_BUG_ON(atomic_read(&page->_count) == 0);
  268. atomic_inc(&page->_count);
  269. }
  270. static inline struct page *virt_to_head_page(const void *x)
  271. {
  272. struct page *page = virt_to_page(x);
  273. return compound_head(page);
  274. }
  275. /*
  276. * Setup the page count before being freed into the page allocator for
  277. * the first time (boot or memory hotplug)
  278. */
  279. static inline void init_page_count(struct page *page)
  280. {
  281. atomic_set(&page->_count, 1);
  282. }
  283. void put_page(struct page *page);
  284. void put_pages_list(struct list_head *pages);
  285. void split_page(struct page *page, unsigned int order);
  286. /*
  287. * Compound pages have a destructor function. Provide a
  288. * prototype for that function and accessor functions.
  289. * These are _only_ valid on the head of a PG_compound page.
  290. */
  291. typedef void compound_page_dtor(struct page *);
  292. static inline void set_compound_page_dtor(struct page *page,
  293. compound_page_dtor *dtor)
  294. {
  295. page[1].lru.next = (void *)dtor;
  296. }
  297. static inline compound_page_dtor *get_compound_page_dtor(struct page *page)
  298. {
  299. return (compound_page_dtor *)page[1].lru.next;
  300. }
  301. static inline int compound_order(struct page *page)
  302. {
  303. if (!PageHead(page))
  304. return 0;
  305. return (unsigned long)page[1].lru.prev;
  306. }
  307. static inline void set_compound_order(struct page *page, unsigned long order)
  308. {
  309. page[1].lru.prev = (void *)order;
  310. }
  311. /*
  312. * Multiple processes may "see" the same page. E.g. for untouched
  313. * mappings of /dev/null, all processes see the same page full of
  314. * zeroes, and text pages of executables and shared libraries have
  315. * only one copy in memory, at most, normally.
  316. *
  317. * For the non-reserved pages, page_count(page) denotes a reference count.
  318. * page_count() == 0 means the page is free. page->lru is then used for
  319. * freelist management in the buddy allocator.
  320. * page_count() > 0 means the page has been allocated.
  321. *
  322. * Pages are allocated by the slab allocator in order to provide memory
  323. * to kmalloc and kmem_cache_alloc. In this case, the management of the
  324. * page, and the fields in 'struct page' are the responsibility of mm/slab.c
  325. * unless a particular usage is carefully commented. (the responsibility of
  326. * freeing the kmalloc memory is the caller's, of course).
  327. *
  328. * A page may be used by anyone else who does a __get_free_page().
  329. * In this case, page_count still tracks the references, and should only
  330. * be used through the normal accessor functions. The top bits of page->flags
  331. * and page->virtual store page management information, but all other fields
  332. * are unused and could be used privately, carefully. The management of this
  333. * page is the responsibility of the one who allocated it, and those who have
  334. * subsequently been given references to it.
  335. *
  336. * The other pages (we may call them "pagecache pages") are completely
  337. * managed by the Linux memory manager: I/O, buffers, swapping etc.
  338. * The following discussion applies only to them.
  339. *
  340. * A pagecache page contains an opaque `private' member, which belongs to the
  341. * page's address_space. Usually, this is the address of a circular list of
  342. * the page's disk buffers. PG_private must be set to tell the VM to call
  343. * into the filesystem to release these pages.
  344. *
  345. * A page may belong to an inode's memory mapping. In this case, page->mapping
  346. * is the pointer to the inode, and page->index is the file offset of the page,
  347. * in units of PAGE_CACHE_SIZE.
  348. *
  349. * If pagecache pages are not associated with an inode, they are said to be
  350. * anonymous pages. These may become associated with the swapcache, and in that
  351. * case PG_swapcache is set, and page->private is an offset into the swapcache.
  352. *
  353. * In either case (swapcache or inode backed), the pagecache itself holds one
  354. * reference to the page. Setting PG_private should also increment the
  355. * refcount. The each user mapping also has a reference to the page.
  356. *
  357. * The pagecache pages are stored in a per-mapping radix tree, which is
  358. * rooted at mapping->page_tree, and indexed by offset.
  359. * Where 2.4 and early 2.6 kernels kept dirty/clean pages in per-address_space
  360. * lists, we instead now tag pages as dirty/writeback in the radix tree.
  361. *
  362. * All pagecache pages may be subject to I/O:
  363. * - inode pages may need to be read from disk,
  364. * - inode pages which have been modified and are MAP_SHARED may need
  365. * to be written back to the inode on disk,
  366. * - anonymous pages (including MAP_PRIVATE file mappings) which have been
  367. * modified may need to be swapped out to swap space and (later) to be read
  368. * back into memory.
  369. */
  370. /*
  371. * The zone field is never updated after free_area_init_core()
  372. * sets it, so none of the operations on it need to be atomic.
  373. */
  374. /*
  375. * page->flags layout:
  376. *
  377. * There are three possibilities for how page->flags get
  378. * laid out. The first is for the normal case, without
  379. * sparsemem. The second is for sparsemem when there is
  380. * plenty of space for node and section. The last is when
  381. * we have run out of space and have to fall back to an
  382. * alternate (slower) way of determining the node.
  383. *
  384. * No sparsemem or sparsemem vmemmap: | NODE | ZONE | ... | FLAGS |
  385. * classic sparse with space for node:| SECTION | NODE | ZONE | ... | FLAGS |
  386. * classic sparse no space for node: | SECTION | ZONE | ... | FLAGS |
  387. */
  388. #if defined(CONFIG_SPARSEMEM) && !defined(CONFIG_SPARSEMEM_VMEMMAP)
  389. #define SECTIONS_WIDTH SECTIONS_SHIFT
  390. #else
  391. #define SECTIONS_WIDTH 0
  392. #endif
  393. #define ZONES_WIDTH ZONES_SHIFT
  394. #if SECTIONS_WIDTH+ZONES_WIDTH+NODES_SHIFT <= BITS_PER_LONG - NR_PAGEFLAGS
  395. #define NODES_WIDTH NODES_SHIFT
  396. #else
  397. #ifdef CONFIG_SPARSEMEM_VMEMMAP
  398. #error "Vmemmap: No space for nodes field in page flags"
  399. #endif
  400. #define NODES_WIDTH 0
  401. #endif
  402. /* Page flags: | [SECTION] | [NODE] | ZONE | ... | FLAGS | */
  403. #define SECTIONS_PGOFF ((sizeof(unsigned long)*8) - SECTIONS_WIDTH)
  404. #define NODES_PGOFF (SECTIONS_PGOFF - NODES_WIDTH)
  405. #define ZONES_PGOFF (NODES_PGOFF - ZONES_WIDTH)
  406. /*
  407. * We are going to use the flags for the page to node mapping if its in
  408. * there. This includes the case where there is no node, so it is implicit.
  409. */
  410. #if !(NODES_WIDTH > 0 || NODES_SHIFT == 0)
  411. #define NODE_NOT_IN_PAGE_FLAGS
  412. #endif
  413. #ifndef PFN_SECTION_SHIFT
  414. #define PFN_SECTION_SHIFT 0
  415. #endif
  416. /*
  417. * Define the bit shifts to access each section. For non-existant
  418. * sections we define the shift as 0; that plus a 0 mask ensures
  419. * the compiler will optimise away reference to them.
  420. */
  421. #define SECTIONS_PGSHIFT (SECTIONS_PGOFF * (SECTIONS_WIDTH != 0))
  422. #define NODES_PGSHIFT (NODES_PGOFF * (NODES_WIDTH != 0))
  423. #define ZONES_PGSHIFT (ZONES_PGOFF * (ZONES_WIDTH != 0))
  424. /* NODE:ZONE or SECTION:ZONE is used to ID a zone for the buddy allcator */
  425. #ifdef NODE_NOT_IN_PAGEFLAGS
  426. #define ZONEID_SHIFT (SECTIONS_SHIFT + ZONES_SHIFT)
  427. #define ZONEID_PGOFF ((SECTIONS_PGOFF < ZONES_PGOFF)? \
  428. SECTIONS_PGOFF : ZONES_PGOFF)
  429. #else
  430. #define ZONEID_SHIFT (NODES_SHIFT + ZONES_SHIFT)
  431. #define ZONEID_PGOFF ((NODES_PGOFF < ZONES_PGOFF)? \
  432. NODES_PGOFF : ZONES_PGOFF)
  433. #endif
  434. #define ZONEID_PGSHIFT (ZONEID_PGOFF * (ZONEID_SHIFT != 0))
  435. #if SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > BITS_PER_LONG - NR_PAGEFLAGS
  436. #error SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > BITS_PER_LONG - NR_PAGEFLAGS
  437. #endif
  438. #define ZONES_MASK ((1UL << ZONES_WIDTH) - 1)
  439. #define NODES_MASK ((1UL << NODES_WIDTH) - 1)
  440. #define SECTIONS_MASK ((1UL << SECTIONS_WIDTH) - 1)
  441. #define ZONEID_MASK ((1UL << ZONEID_SHIFT) - 1)
  442. static inline enum zone_type page_zonenum(struct page *page)
  443. {
  444. return (page->flags >> ZONES_PGSHIFT) & ZONES_MASK;
  445. }
  446. /*
  447. * The identification function is only used by the buddy allocator for
  448. * determining if two pages could be buddies. We are not really
  449. * identifying a zone since we could be using a the section number
  450. * id if we have not node id available in page flags.
  451. * We guarantee only that it will return the same value for two
  452. * combinable pages in a zone.
  453. */
  454. static inline int page_zone_id(struct page *page)
  455. {
  456. return (page->flags >> ZONEID_PGSHIFT) & ZONEID_MASK;
  457. }
  458. static inline int zone_to_nid(struct zone *zone)
  459. {
  460. #ifdef CONFIG_NUMA
  461. return zone->node;
  462. #else
  463. return 0;
  464. #endif
  465. }
  466. #ifdef NODE_NOT_IN_PAGE_FLAGS
  467. extern int page_to_nid(struct page *page);
  468. #else
  469. static inline int page_to_nid(struct page *page)
  470. {
  471. return (page->flags >> NODES_PGSHIFT) & NODES_MASK;
  472. }
  473. #endif
  474. static inline struct zone *page_zone(struct page *page)
  475. {
  476. return &NODE_DATA(page_to_nid(page))->node_zones[page_zonenum(page)];
  477. }
  478. #if defined(CONFIG_SPARSEMEM) && !defined(CONFIG_SPARSEMEM_VMEMMAP)
  479. static inline unsigned long page_to_section(struct page *page)
  480. {
  481. return (page->flags >> SECTIONS_PGSHIFT) & SECTIONS_MASK;
  482. }
  483. #endif
  484. static inline void set_page_zone(struct page *page, enum zone_type zone)
  485. {
  486. page->flags &= ~(ZONES_MASK << ZONES_PGSHIFT);
  487. page->flags |= (zone & ZONES_MASK) << ZONES_PGSHIFT;
  488. }
  489. static inline void set_page_node(struct page *page, unsigned long node)
  490. {
  491. page->flags &= ~(NODES_MASK << NODES_PGSHIFT);
  492. page->flags |= (node & NODES_MASK) << NODES_PGSHIFT;
  493. }
  494. static inline void set_page_section(struct page *page, unsigned long section)
  495. {
  496. page->flags &= ~(SECTIONS_MASK << SECTIONS_PGSHIFT);
  497. page->flags |= (section & SECTIONS_MASK) << SECTIONS_PGSHIFT;
  498. }
  499. static inline void set_page_links(struct page *page, enum zone_type zone,
  500. unsigned long node, unsigned long pfn)
  501. {
  502. set_page_zone(page, zone);
  503. set_page_node(page, node);
  504. set_page_section(page, pfn_to_section_nr(pfn));
  505. }
  506. /*
  507. * Some inline functions in vmstat.h depend on page_zone()
  508. */
  509. #include <linux/vmstat.h>
  510. static __always_inline void *lowmem_page_address(struct page *page)
  511. {
  512. return __va(page_to_pfn(page) << PAGE_SHIFT);
  513. }
  514. #if defined(CONFIG_HIGHMEM) && !defined(WANT_PAGE_VIRTUAL)
  515. #define HASHED_PAGE_VIRTUAL
  516. #endif
  517. #if defined(WANT_PAGE_VIRTUAL)
  518. #define page_address(page) ((page)->virtual)
  519. #define set_page_address(page, address) \
  520. do { \
  521. (page)->virtual = (address); \
  522. } while(0)
  523. #define page_address_init() do { } while(0)
  524. #endif
  525. #if defined(HASHED_PAGE_VIRTUAL)
  526. void *page_address(struct page *page);
  527. void set_page_address(struct page *page, void *virtual);
  528. void page_address_init(void);
  529. #endif
  530. #if !defined(HASHED_PAGE_VIRTUAL) && !defined(WANT_PAGE_VIRTUAL)
  531. #define page_address(page) lowmem_page_address(page)
  532. #define set_page_address(page, address) do { } while(0)
  533. #define page_address_init() do { } while(0)
  534. #endif
  535. /*
  536. * On an anonymous page mapped into a user virtual memory area,
  537. * page->mapping points to its anon_vma, not to a struct address_space;
  538. * with the PAGE_MAPPING_ANON bit set to distinguish it. See rmap.h.
  539. *
  540. * On an anonymous page in a VM_MERGEABLE area, if CONFIG_KSM is enabled,
  541. * the PAGE_MAPPING_KSM bit may be set along with the PAGE_MAPPING_ANON bit;
  542. * and then page->mapping points, not to an anon_vma, but to a private
  543. * structure which KSM associates with that merged page. See ksm.h.
  544. *
  545. * PAGE_MAPPING_KSM without PAGE_MAPPING_ANON is currently never used.
  546. *
  547. * Please note that, confusingly, "page_mapping" refers to the inode
  548. * address_space which maps the page from disk; whereas "page_mapped"
  549. * refers to user virtual address space into which the page is mapped.
  550. */
  551. #define PAGE_MAPPING_ANON 1
  552. #define PAGE_MAPPING_KSM 2
  553. #define PAGE_MAPPING_FLAGS (PAGE_MAPPING_ANON | PAGE_MAPPING_KSM)
  554. extern struct address_space swapper_space;
  555. static inline struct address_space *page_mapping(struct page *page)
  556. {
  557. struct address_space *mapping = page->mapping;
  558. VM_BUG_ON(PageSlab(page));
  559. if (unlikely(PageSwapCache(page)))
  560. mapping = &swapper_space;
  561. else if (unlikely((unsigned long)mapping & PAGE_MAPPING_ANON))
  562. mapping = NULL;
  563. return mapping;
  564. }
  565. /* Neutral page->mapping pointer to address_space or anon_vma or other */
  566. static inline void *page_rmapping(struct page *page)
  567. {
  568. return (void *)((unsigned long)page->mapping & ~PAGE_MAPPING_FLAGS);
  569. }
  570. static inline int PageAnon(struct page *page)
  571. {
  572. return ((unsigned long)page->mapping & PAGE_MAPPING_ANON) != 0;
  573. }
  574. /*
  575. * Return the pagecache index of the passed page. Regular pagecache pages
  576. * use ->index whereas swapcache pages use ->private
  577. */
  578. static inline pgoff_t page_index(struct page *page)
  579. {
  580. if (unlikely(PageSwapCache(page)))
  581. return page_private(page);
  582. return page->index;
  583. }
  584. /*
  585. * The atomic page->_mapcount, like _count, starts from -1:
  586. * so that transitions both from it and to it can be tracked,
  587. * using atomic_inc_and_test and atomic_add_negative(-1).
  588. */
  589. static inline void reset_page_mapcount(struct page *page)
  590. {
  591. atomic_set(&(page)->_mapcount, -1);
  592. }
  593. static inline int page_mapcount(struct page *page)
  594. {
  595. return atomic_read(&(page)->_mapcount) + 1;
  596. }
  597. /*
  598. * Return true if this page is mapped into pagetables.
  599. */
  600. static inline int page_mapped(struct page *page)
  601. {
  602. return atomic_read(&(page)->_mapcount) >= 0;
  603. }
  604. /*
  605. * Different kinds of faults, as returned by handle_mm_fault().
  606. * Used to decide whether a process gets delivered SIGBUS or
  607. * just gets major/minor fault counters bumped up.
  608. */
  609. #define VM_FAULT_MINOR 0 /* For backwards compat. Remove me quickly. */
  610. #define VM_FAULT_OOM 0x0001
  611. #define VM_FAULT_SIGBUS 0x0002
  612. #define VM_FAULT_MAJOR 0x0004
  613. #define VM_FAULT_WRITE 0x0008 /* Special case for get_user_pages */
  614. #define VM_FAULT_HWPOISON 0x0010 /* Hit poisoned page */
  615. #define VM_FAULT_NOPAGE 0x0100 /* ->fault installed the pte, not return page */
  616. #define VM_FAULT_LOCKED 0x0200 /* ->fault locked the returned page */
  617. #define VM_FAULT_ERROR (VM_FAULT_OOM | VM_FAULT_SIGBUS | VM_FAULT_HWPOISON)
  618. /*
  619. * Can be called by the pagefault handler when it gets a VM_FAULT_OOM.
  620. */
  621. extern void pagefault_out_of_memory(void);
  622. #define offset_in_page(p) ((unsigned long)(p) & ~PAGE_MASK)
  623. extern void show_free_areas(void);
  624. int shmem_lock(struct file *file, int lock, struct user_struct *user);
  625. struct file *shmem_file_setup(const char *name, loff_t size, unsigned long flags);
  626. int shmem_zero_setup(struct vm_area_struct *);
  627. #ifndef CONFIG_MMU
  628. extern unsigned long shmem_get_unmapped_area(struct file *file,
  629. unsigned long addr,
  630. unsigned long len,
  631. unsigned long pgoff,
  632. unsigned long flags);
  633. #endif
  634. extern int can_do_mlock(void);
  635. extern int user_shm_lock(size_t, struct user_struct *);
  636. extern void user_shm_unlock(size_t, struct user_struct *);
  637. /*
  638. * Parameter block passed down to zap_pte_range in exceptional cases.
  639. */
  640. struct zap_details {
  641. struct vm_area_struct *nonlinear_vma; /* Check page->index if set */
  642. struct address_space *check_mapping; /* Check page->mapping if set */
  643. pgoff_t first_index; /* Lowest page->index to unmap */
  644. pgoff_t last_index; /* Highest page->index to unmap */
  645. spinlock_t *i_mmap_lock; /* For unmap_mapping_range: */
  646. unsigned long truncate_count; /* Compare vm_truncate_count */
  647. };
  648. struct page *vm_normal_page(struct vm_area_struct *vma, unsigned long addr,
  649. pte_t pte);
  650. int zap_vma_ptes(struct vm_area_struct *vma, unsigned long address,
  651. unsigned long size);
  652. unsigned long zap_page_range(struct vm_area_struct *vma, unsigned long address,
  653. unsigned long size, struct zap_details *);
  654. unsigned long unmap_vmas(struct mmu_gather **tlb,
  655. struct vm_area_struct *start_vma, unsigned long start_addr,
  656. unsigned long end_addr, unsigned long *nr_accounted,
  657. struct zap_details *);
  658. /**
  659. * mm_walk - callbacks for walk_page_range
  660. * @pgd_entry: if set, called for each non-empty PGD (top-level) entry
  661. * @pud_entry: if set, called for each non-empty PUD (2nd-level) entry
  662. * @pmd_entry: if set, called for each non-empty PMD (3rd-level) entry
  663. * @pte_entry: if set, called for each non-empty PTE (4th-level) entry
  664. * @pte_hole: if set, called for each hole at all levels
  665. * @hugetlb_entry: if set, called for each hugetlb entry
  666. *
  667. * (see walk_page_range for more details)
  668. */
  669. struct mm_walk {
  670. int (*pgd_entry)(pgd_t *, unsigned long, unsigned long, struct mm_walk *);
  671. int (*pud_entry)(pud_t *, unsigned long, unsigned long, struct mm_walk *);
  672. int (*pmd_entry)(pmd_t *, unsigned long, unsigned long, struct mm_walk *);
  673. int (*pte_entry)(pte_t *, unsigned long, unsigned long, struct mm_walk *);
  674. int (*pte_hole)(unsigned long, unsigned long, struct mm_walk *);
  675. int (*hugetlb_entry)(pte_t *, unsigned long, unsigned long,
  676. struct mm_walk *);
  677. struct mm_struct *mm;
  678. void *private;
  679. };
  680. int walk_page_range(unsigned long addr, unsigned long end,
  681. struct mm_walk *walk);
  682. void free_pgd_range(struct mmu_gather *tlb, unsigned long addr,
  683. unsigned long end, unsigned long floor, unsigned long ceiling);
  684. int copy_page_range(struct mm_struct *dst, struct mm_struct *src,
  685. struct vm_area_struct *vma);
  686. void unmap_mapping_range(struct address_space *mapping,
  687. loff_t const holebegin, loff_t const holelen, int even_cows);
  688. int follow_pfn(struct vm_area_struct *vma, unsigned long address,
  689. unsigned long *pfn);
  690. int follow_phys(struct vm_area_struct *vma, unsigned long address,
  691. unsigned int flags, unsigned long *prot, resource_size_t *phys);
  692. int generic_access_phys(struct vm_area_struct *vma, unsigned long addr,
  693. void *buf, int len, int write);
  694. static inline void unmap_shared_mapping_range(struct address_space *mapping,
  695. loff_t const holebegin, loff_t const holelen)
  696. {
  697. unmap_mapping_range(mapping, holebegin, holelen, 0);
  698. }
  699. extern void truncate_pagecache(struct inode *inode, loff_t old, loff_t new);
  700. extern int vmtruncate(struct inode *inode, loff_t offset);
  701. extern int vmtruncate_range(struct inode *inode, loff_t offset, loff_t end);
  702. int truncate_inode_page(struct address_space *mapping, struct page *page);
  703. int generic_error_remove_page(struct address_space *mapping, struct page *page);
  704. int invalidate_inode_page(struct page *page);
  705. #ifdef CONFIG_MMU
  706. extern int handle_mm_fault(struct mm_struct *mm, struct vm_area_struct *vma,
  707. unsigned long address, unsigned int flags);
  708. #else
  709. static inline int handle_mm_fault(struct mm_struct *mm,
  710. struct vm_area_struct *vma, unsigned long address,
  711. unsigned int flags)
  712. {
  713. /* should never happen if there's no MMU */
  714. BUG();
  715. return VM_FAULT_SIGBUS;
  716. }
  717. #endif
  718. extern int make_pages_present(unsigned long addr, unsigned long end);
  719. extern int access_process_vm(struct task_struct *tsk, unsigned long addr, void *buf, int len, int write);
  720. int get_user_pages(struct task_struct *tsk, struct mm_struct *mm,
  721. unsigned long start, int nr_pages, int write, int force,
  722. struct page **pages, struct vm_area_struct **vmas);
  723. int get_user_pages_fast(unsigned long start, int nr_pages, int write,
  724. struct page **pages);
  725. struct page *get_dump_page(unsigned long addr);
  726. extern int try_to_release_page(struct page * page, gfp_t gfp_mask);
  727. extern void do_invalidatepage(struct page *page, unsigned long offset);
  728. int __set_page_dirty_nobuffers(struct page *page);
  729. int __set_page_dirty_no_writeback(struct page *page);
  730. int redirty_page_for_writepage(struct writeback_control *wbc,
  731. struct page *page);
  732. void account_page_dirtied(struct page *page, struct address_space *mapping);
  733. int set_page_dirty(struct page *page);
  734. int set_page_dirty_lock(struct page *page);
  735. int clear_page_dirty_for_io(struct page *page);
  736. extern unsigned long move_page_tables(struct vm_area_struct *vma,
  737. unsigned long old_addr, struct vm_area_struct *new_vma,
  738. unsigned long new_addr, unsigned long len);
  739. extern unsigned long do_mremap(unsigned long addr,
  740. unsigned long old_len, unsigned long new_len,
  741. unsigned long flags, unsigned long new_addr);
  742. extern int mprotect_fixup(struct vm_area_struct *vma,
  743. struct vm_area_struct **pprev, unsigned long start,
  744. unsigned long end, unsigned long newflags);
  745. /*
  746. * doesn't attempt to fault and will return short.
  747. */
  748. int __get_user_pages_fast(unsigned long start, int nr_pages, int write,
  749. struct page **pages);
  750. /*
  751. * A callback you can register to apply pressure to ageable caches.
  752. *
  753. * 'shrink' is passed a count 'nr_to_scan' and a 'gfpmask'. It should
  754. * look through the least-recently-used 'nr_to_scan' entries and
  755. * attempt to free them up. It should return the number of objects
  756. * which remain in the cache. If it returns -1, it means it cannot do
  757. * any scanning at this time (eg. there is a risk of deadlock).
  758. *
  759. * The 'gfpmask' refers to the allocation we are currently trying to
  760. * fulfil.
  761. *
  762. * Note that 'shrink' will be passed nr_to_scan == 0 when the VM is
  763. * querying the cache size, so a fastpath for that case is appropriate.
  764. */
  765. struct shrinker {
  766. int (*shrink)(int nr_to_scan, gfp_t gfp_mask);
  767. int seeks; /* seeks to recreate an obj */
  768. /* These are for internal use */
  769. struct list_head list;
  770. long nr; /* objs pending delete */
  771. };
  772. #define DEFAULT_SEEKS 2 /* A good number if you don't know better. */
  773. extern void register_shrinker(struct shrinker *);
  774. extern void unregister_shrinker(struct shrinker *);
  775. int vma_wants_writenotify(struct vm_area_struct *vma);
  776. extern pte_t *get_locked_pte(struct mm_struct *mm, unsigned long addr, spinlock_t **ptl);
  777. #ifdef __PAGETABLE_PUD_FOLDED
  778. static inline int __pud_alloc(struct mm_struct *mm, pgd_t *pgd,
  779. unsigned long address)
  780. {
  781. return 0;
  782. }
  783. #else
  784. int __pud_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address);
  785. #endif
  786. #ifdef __PAGETABLE_PMD_FOLDED
  787. static inline int __pmd_alloc(struct mm_struct *mm, pud_t *pud,
  788. unsigned long address)
  789. {
  790. return 0;
  791. }
  792. #else
  793. int __pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address);
  794. #endif
  795. int __pte_alloc(struct mm_struct *mm, pmd_t *pmd, unsigned long address);
  796. int __pte_alloc_kernel(pmd_t *pmd, unsigned long address);
  797. /*
  798. * The following ifdef needed to get the 4level-fixup.h header to work.
  799. * Remove it when 4level-fixup.h has been removed.
  800. */
  801. #if defined(CONFIG_MMU) && !defined(__ARCH_HAS_4LEVEL_HACK)
  802. static inline pud_t *pud_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address)
  803. {
  804. return (unlikely(pgd_none(*pgd)) && __pud_alloc(mm, pgd, address))?
  805. NULL: pud_offset(pgd, address);
  806. }
  807. static inline pmd_t *pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address)
  808. {
  809. return (unlikely(pud_none(*pud)) && __pmd_alloc(mm, pud, address))?
  810. NULL: pmd_offset(pud, address);
  811. }
  812. #endif /* CONFIG_MMU && !__ARCH_HAS_4LEVEL_HACK */
  813. #if USE_SPLIT_PTLOCKS
  814. /*
  815. * We tuck a spinlock to guard each pagetable page into its struct page,
  816. * at page->private, with BUILD_BUG_ON to make sure that this will not
  817. * overflow into the next struct page (as it might with DEBUG_SPINLOCK).
  818. * When freeing, reset page->mapping so free_pages_check won't complain.
  819. */
  820. #define __pte_lockptr(page) &((page)->ptl)
  821. #define pte_lock_init(_page) do { \
  822. spin_lock_init(__pte_lockptr(_page)); \
  823. } while (0)
  824. #define pte_lock_deinit(page) ((page)->mapping = NULL)
  825. #define pte_lockptr(mm, pmd) ({(void)(mm); __pte_lockptr(pmd_page(*(pmd)));})
  826. #else /* !USE_SPLIT_PTLOCKS */
  827. /*
  828. * We use mm->page_table_lock to guard all pagetable pages of the mm.
  829. */
  830. #define pte_lock_init(page) do {} while (0)
  831. #define pte_lock_deinit(page) do {} while (0)
  832. #define pte_lockptr(mm, pmd) ({(void)(pmd); &(mm)->page_table_lock;})
  833. #endif /* USE_SPLIT_PTLOCKS */
  834. static inline void pgtable_page_ctor(struct page *page)
  835. {
  836. pte_lock_init(page);
  837. inc_zone_page_state(page, NR_PAGETABLE);
  838. }
  839. static inline void pgtable_page_dtor(struct page *page)
  840. {
  841. pte_lock_deinit(page);
  842. dec_zone_page_state(page, NR_PAGETABLE);
  843. }
  844. #define pte_offset_map_lock(mm, pmd, address, ptlp) \
  845. ({ \
  846. spinlock_t *__ptl = pte_lockptr(mm, pmd); \
  847. pte_t *__pte = pte_offset_map(pmd, address); \
  848. *(ptlp) = __ptl; \
  849. spin_lock(__ptl); \
  850. __pte; \
  851. })
  852. #define pte_unmap_unlock(pte, ptl) do { \
  853. spin_unlock(ptl); \
  854. pte_unmap(pte); \
  855. } while (0)
  856. #define pte_alloc_map(mm, pmd, address) \
  857. ((unlikely(!pmd_present(*(pmd))) && __pte_alloc(mm, pmd, address))? \
  858. NULL: pte_offset_map(pmd, address))
  859. #define pte_alloc_map_lock(mm, pmd, address, ptlp) \
  860. ((unlikely(!pmd_present(*(pmd))) && __pte_alloc(mm, pmd, address))? \
  861. NULL: pte_offset_map_lock(mm, pmd, address, ptlp))
  862. #define pte_alloc_kernel(pmd, address) \
  863. ((unlikely(!pmd_present(*(pmd))) && __pte_alloc_kernel(pmd, address))? \
  864. NULL: pte_offset_kernel(pmd, address))
  865. extern void free_area_init(unsigned long * zones_size);
  866. extern void free_area_init_node(int nid, unsigned long * zones_size,
  867. unsigned long zone_start_pfn, unsigned long *zholes_size);
  868. #ifdef CONFIG_ARCH_POPULATES_NODE_MAP
  869. /*
  870. * With CONFIG_ARCH_POPULATES_NODE_MAP set, an architecture may initialise its
  871. * zones, allocate the backing mem_map and account for memory holes in a more
  872. * architecture independent manner. This is a substitute for creating the
  873. * zone_sizes[] and zholes_size[] arrays and passing them to
  874. * free_area_init_node()
  875. *
  876. * An architecture is expected to register range of page frames backed by
  877. * physical memory with add_active_range() before calling
  878. * free_area_init_nodes() passing in the PFN each zone ends at. At a basic
  879. * usage, an architecture is expected to do something like
  880. *
  881. * unsigned long max_zone_pfns[MAX_NR_ZONES] = {max_dma, max_normal_pfn,
  882. * max_highmem_pfn};
  883. * for_each_valid_physical_page_range()
  884. * add_active_range(node_id, start_pfn, end_pfn)
  885. * free_area_init_nodes(max_zone_pfns);
  886. *
  887. * If the architecture guarantees that there are no holes in the ranges
  888. * registered with add_active_range(), free_bootmem_active_regions()
  889. * will call free_bootmem_node() for each registered physical page range.
  890. * Similarly sparse_memory_present_with_active_regions() calls
  891. * memory_present() for each range when SPARSEMEM is enabled.
  892. *
  893. * See mm/page_alloc.c for more information on each function exposed by
  894. * CONFIG_ARCH_POPULATES_NODE_MAP
  895. */
  896. extern void free_area_init_nodes(unsigned long *max_zone_pfn);
  897. extern void add_active_range(unsigned int nid, unsigned long start_pfn,
  898. unsigned long end_pfn);
  899. extern void remove_active_range(unsigned int nid, unsigned long start_pfn,
  900. unsigned long end_pfn);
  901. extern void remove_all_active_ranges(void);
  902. void sort_node_map(void);
  903. unsigned long __absent_pages_in_range(int nid, unsigned long start_pfn,
  904. unsigned long end_pfn);
  905. extern unsigned long absent_pages_in_range(unsigned long start_pfn,
  906. unsigned long end_pfn);
  907. extern void get_pfn_range_for_nid(unsigned int nid,
  908. unsigned long *start_pfn, unsigned long *end_pfn);
  909. extern unsigned long find_min_pfn_with_active_regions(void);
  910. extern void free_bootmem_with_active_regions(int nid,
  911. unsigned long max_low_pfn);
  912. int add_from_early_node_map(struct range *range, int az,
  913. int nr_range, int nid);
  914. void *__alloc_memory_core_early(int nodeid, u64 size, u64 align,
  915. u64 goal, u64 limit);
  916. typedef int (*work_fn_t)(unsigned long, unsigned long, void *);
  917. extern void work_with_active_regions(int nid, work_fn_t work_fn, void *data);
  918. extern void sparse_memory_present_with_active_regions(int nid);
  919. #endif /* CONFIG_ARCH_POPULATES_NODE_MAP */
  920. #if !defined(CONFIG_ARCH_POPULATES_NODE_MAP) && \
  921. !defined(CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID)
  922. static inline int __early_pfn_to_nid(unsigned long pfn)
  923. {
  924. return 0;
  925. }
  926. #else
  927. /* please see mm/page_alloc.c */
  928. extern int __meminit early_pfn_to_nid(unsigned long pfn);
  929. #ifdef CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID
  930. /* there is a per-arch backend function. */
  931. extern int __meminit __early_pfn_to_nid(unsigned long pfn);
  932. #endif /* CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID */
  933. #endif
  934. extern void set_dma_reserve(unsigned long new_dma_reserve);
  935. extern void memmap_init_zone(unsigned long, int, unsigned long,
  936. unsigned long, enum memmap_context);
  937. extern void setup_per_zone_wmarks(void);
  938. extern void calculate_zone_inactive_ratio(struct zone *zone);
  939. extern void mem_init(void);
  940. extern void __init mmap_init(void);
  941. extern void show_mem(void);
  942. extern void si_meminfo(struct sysinfo * val);
  943. extern void si_meminfo_node(struct sysinfo *val, int nid);
  944. extern int after_bootmem;
  945. extern void setup_per_cpu_pageset(void);
  946. extern void zone_pcp_update(struct zone *zone);
  947. /* nommu.c */
  948. extern atomic_long_t mmap_pages_allocated;
  949. extern int nommu_shrink_inode_mappings(struct inode *, size_t, size_t);
  950. /* prio_tree.c */
  951. void vma_prio_tree_add(struct vm_area_struct *, struct vm_area_struct *old);
  952. void vma_prio_tree_insert(struct vm_area_struct *, struct prio_tree_root *);
  953. void vma_prio_tree_remove(struct vm_area_struct *, struct prio_tree_root *);
  954. struct vm_area_struct *vma_prio_tree_next(struct vm_area_struct *vma,
  955. struct prio_tree_iter *iter);
  956. #define vma_prio_tree_foreach(vma, iter, root, begin, end) \
  957. for (prio_tree_iter_init(iter, root, begin, end), vma = NULL; \
  958. (vma = vma_prio_tree_next(vma, iter)); )
  959. static inline void vma_nonlinear_insert(struct vm_area_struct *vma,
  960. struct list_head *list)
  961. {
  962. vma->shared.vm_set.parent = NULL;
  963. list_add_tail(&vma->shared.vm_set.list, list);
  964. }
  965. /* mmap.c */
  966. extern int __vm_enough_memory(struct mm_struct *mm, long pages, int cap_sys_admin);
  967. extern void vma_adjust(struct vm_area_struct *vma, unsigned long start,
  968. unsigned long end, pgoff_t pgoff, struct vm_area_struct *insert);
  969. extern struct vm_area_struct *vma_merge(struct mm_struct *,
  970. struct vm_area_struct *prev, unsigned long addr, unsigned long end,
  971. unsigned long vm_flags, struct anon_vma *, struct file *, pgoff_t,
  972. struct mempolicy *);
  973. extern struct anon_vma *find_mergeable_anon_vma(struct vm_area_struct *);
  974. extern int split_vma(struct mm_struct *,
  975. struct vm_area_struct *, unsigned long addr, int new_below);
  976. extern int insert_vm_struct(struct mm_struct *, struct vm_area_struct *);
  977. extern void __vma_link_rb(struct mm_struct *, struct vm_area_struct *,
  978. struct rb_node **, struct rb_node *);
  979. extern void unlink_file_vma(struct vm_area_struct *);
  980. extern struct vm_area_struct *copy_vma(struct vm_area_struct **,
  981. unsigned long addr, unsigned long len, pgoff_t pgoff);
  982. extern void exit_mmap(struct mm_struct *);
  983. extern int mm_take_all_locks(struct mm_struct *mm);
  984. extern void mm_drop_all_locks(struct mm_struct *mm);
  985. #ifdef CONFIG_PROC_FS
  986. /* From fs/proc/base.c. callers must _not_ hold the mm's exe_file_lock */
  987. extern void added_exe_file_vma(struct mm_struct *mm);
  988. extern void removed_exe_file_vma(struct mm_struct *mm);
  989. #else
  990. static inline void added_exe_file_vma(struct mm_struct *mm)
  991. {}
  992. static inline void removed_exe_file_vma(struct mm_struct *mm)
  993. {}
  994. #endif /* CONFIG_PROC_FS */
  995. extern int may_expand_vm(struct mm_struct *mm, unsigned long npages);
  996. extern int install_special_mapping(struct mm_struct *mm,
  997. unsigned long addr, unsigned long len,
  998. unsigned long flags, struct page **pages);
  999. extern unsigned long get_unmapped_area(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
  1000. extern unsigned long do_mmap_pgoff(struct file *file, unsigned long addr,
  1001. unsigned long len, unsigned long prot,
  1002. unsigned long flag, unsigned long pgoff);
  1003. extern unsigned long mmap_region(struct file *file, unsigned long addr,
  1004. unsigned long len, unsigned long flags,
  1005. unsigned int vm_flags, unsigned long pgoff);
  1006. static inline unsigned long do_mmap(struct file *file, unsigned long addr,
  1007. unsigned long len, unsigned long prot,
  1008. unsigned long flag, unsigned long offset)
  1009. {
  1010. unsigned long ret = -EINVAL;
  1011. if ((offset + PAGE_ALIGN(len)) < offset)
  1012. goto out;
  1013. if (!(offset & ~PAGE_MASK))
  1014. ret = do_mmap_pgoff(file, addr, len, prot, flag, offset >> PAGE_SHIFT);
  1015. out:
  1016. return ret;
  1017. }
  1018. extern int do_munmap(struct mm_struct *, unsigned long, size_t);
  1019. extern unsigned long do_brk(unsigned long, unsigned long);
  1020. /* filemap.c */
  1021. extern unsigned long page_unuse(struct page *);
  1022. extern void truncate_inode_pages(struct address_space *, loff_t);
  1023. extern void truncate_inode_pages_range(struct address_space *,
  1024. loff_t lstart, loff_t lend);
  1025. /* generic vm_area_ops exported for stackable file systems */
  1026. extern int filemap_fault(struct vm_area_struct *, struct vm_fault *);
  1027. /* mm/page-writeback.c */
  1028. int write_one_page(struct page *page, int wait);
  1029. void task_dirty_inc(struct task_struct *tsk);
  1030. /* readahead.c */
  1031. #define VM_MAX_READAHEAD 128 /* kbytes */
  1032. #define VM_MIN_READAHEAD 16 /* kbytes (includes current page) */
  1033. int force_page_cache_readahead(struct address_space *mapping, struct file *filp,
  1034. pgoff_t offset, unsigned long nr_to_read);
  1035. void page_cache_sync_readahead(struct address_space *mapping,
  1036. struct file_ra_state *ra,
  1037. struct file *filp,
  1038. pgoff_t offset,
  1039. unsigned long size);
  1040. void page_cache_async_readahead(struct address_space *mapping,
  1041. struct file_ra_state *ra,
  1042. struct file *filp,
  1043. struct page *pg,
  1044. pgoff_t offset,
  1045. unsigned long size);
  1046. unsigned long max_sane_readahead(unsigned long nr);
  1047. unsigned long ra_submit(struct file_ra_state *ra,
  1048. struct address_space *mapping,
  1049. struct file *filp);
  1050. /* Do stack extension */
  1051. extern int expand_stack(struct vm_area_struct *vma, unsigned long address);
  1052. #ifdef CONFIG_IA64
  1053. extern int expand_upwards(struct vm_area_struct *vma, unsigned long address);
  1054. #endif
  1055. extern int expand_stack_downwards(struct vm_area_struct *vma,
  1056. unsigned long address);
  1057. /* Look up the first VMA which satisfies addr < vm_end, NULL if none. */
  1058. extern struct vm_area_struct * find_vma(struct mm_struct * mm, unsigned long addr);
  1059. extern struct vm_area_struct * find_vma_prev(struct mm_struct * mm, unsigned long addr,
  1060. struct vm_area_struct **pprev);
  1061. /* Look up the first VMA which intersects the interval start_addr..end_addr-1,
  1062. NULL if none. Assume start_addr < end_addr. */
  1063. static inline struct vm_area_struct * find_vma_intersection(struct mm_struct * mm, unsigned long start_addr, unsigned long end_addr)
  1064. {
  1065. struct vm_area_struct * vma = find_vma(mm,start_addr);
  1066. if (vma && end_addr <= vma->vm_start)
  1067. vma = NULL;
  1068. return vma;
  1069. }
  1070. static inline unsigned long vma_pages(struct vm_area_struct *vma)
  1071. {
  1072. return (vma->vm_end - vma->vm_start) >> PAGE_SHIFT;
  1073. }
  1074. pgprot_t vm_get_page_prot(unsigned long vm_flags);
  1075. struct vm_area_struct *find_extend_vma(struct mm_struct *, unsigned long addr);
  1076. int remap_pfn_range(struct vm_area_struct *, unsigned long addr,
  1077. unsigned long pfn, unsigned long size, pgprot_t);
  1078. int vm_insert_page(struct vm_area_struct *, unsigned long addr, struct page *);
  1079. int vm_insert_pfn(struct vm_area_struct *vma, unsigned long addr,
  1080. unsigned long pfn);
  1081. int vm_insert_mixed(struct vm_area_struct *vma, unsigned long addr,
  1082. unsigned long pfn);
  1083. struct page *follow_page(struct vm_area_struct *, unsigned long address,
  1084. unsigned int foll_flags);
  1085. #define FOLL_WRITE 0x01 /* check pte is writable */
  1086. #define FOLL_TOUCH 0x02 /* mark page accessed */
  1087. #define FOLL_GET 0x04 /* do get_page on page */
  1088. #define FOLL_DUMP 0x08 /* give error on hole if it would be zero */
  1089. #define FOLL_FORCE 0x10 /* get_user_pages read/write w/o permission */
  1090. typedef int (*pte_fn_t)(pte_t *pte, pgtable_t token, unsigned long addr,
  1091. void *data);
  1092. extern int apply_to_page_range(struct mm_struct *mm, unsigned long address,
  1093. unsigned long size, pte_fn_t fn, void *data);
  1094. #ifdef CONFIG_PROC_FS
  1095. void vm_stat_account(struct mm_struct *, unsigned long, struct file *, long);
  1096. #else
  1097. static inline void vm_stat_account(struct mm_struct *mm,
  1098. unsigned long flags, struct file *file, long pages)
  1099. {
  1100. }
  1101. #endif /* CONFIG_PROC_FS */
  1102. #ifdef CONFIG_DEBUG_PAGEALLOC
  1103. extern int debug_pagealloc_enabled;
  1104. extern void kernel_map_pages(struct page *page, int numpages, int enable);
  1105. static inline void enable_debug_pagealloc(void)
  1106. {
  1107. debug_pagealloc_enabled = 1;
  1108. }
  1109. #ifdef CONFIG_HIBERNATION
  1110. extern bool kernel_page_present(struct page *page);
  1111. #endif /* CONFIG_HIBERNATION */
  1112. #else
  1113. static inline void
  1114. kernel_map_pages(struct page *page, int numpages, int enable) {}
  1115. static inline void enable_debug_pagealloc(void)
  1116. {
  1117. }
  1118. #ifdef CONFIG_HIBERNATION
  1119. static inline bool kernel_page_present(struct page *page) { return true; }
  1120. #endif /* CONFIG_HIBERNATION */
  1121. #endif
  1122. extern struct vm_area_struct *get_gate_vma(struct task_struct *tsk);
  1123. #ifdef __HAVE_ARCH_GATE_AREA
  1124. int in_gate_area_no_task(unsigned long addr);
  1125. int in_gate_area(struct task_struct *task, unsigned long addr);
  1126. #else
  1127. int in_gate_area_no_task(unsigned long addr);
  1128. #define in_gate_area(task, addr) ({(void)task; in_gate_area_no_task(addr);})
  1129. #endif /* __HAVE_ARCH_GATE_AREA */
  1130. int drop_caches_sysctl_handler(struct ctl_table *, int,
  1131. void __user *, size_t *, loff_t *);
  1132. unsigned long shrink_slab(unsigned long scanned, gfp_t gfp_mask,
  1133. unsigned long lru_pages);
  1134. #ifndef CONFIG_MMU
  1135. #define randomize_va_space 0
  1136. #else
  1137. extern int randomize_va_space;
  1138. #endif
  1139. const char * arch_vma_name(struct vm_area_struct *vma);
  1140. void print_vma_addr(char *prefix, unsigned long rip);
  1141. void sparse_mem_maps_populate_node(struct page **map_map,
  1142. unsigned long pnum_begin,
  1143. unsigned long pnum_end,
  1144. unsigned long map_count,
  1145. int nodeid);
  1146. struct page *sparse_mem_map_populate(unsigned long pnum, int nid);
  1147. pgd_t *vmemmap_pgd_populate(unsigned long addr, int node);
  1148. pud_t *vmemmap_pud_populate(pgd_t *pgd, unsigned long addr, int node);
  1149. pmd_t *vmemmap_pmd_populate(pud_t *pud, unsigned long addr, int node);
  1150. pte_t *vmemmap_pte_populate(pmd_t *pmd, unsigned long addr, int node);
  1151. void *vmemmap_alloc_block(unsigned long size, int node);
  1152. void *vmemmap_alloc_block_buf(unsigned long size, int node);
  1153. void vmemmap_verify(pte_t *, int, unsigned long, unsigned long);
  1154. int vmemmap_populate_basepages(struct page *start_page,
  1155. unsigned long pages, int node);
  1156. int vmemmap_populate(struct page *start_page, unsigned long pages, int node);
  1157. void vmemmap_populate_print_last(void);
  1158. extern int account_locked_memory(struct mm_struct *mm, struct rlimit *rlim,
  1159. size_t size);
  1160. extern void refund_locked_memory(struct mm_struct *mm, size_t size);
  1161. enum mf_flags {
  1162. MF_COUNT_INCREASED = 1 << 0,
  1163. };
  1164. extern void memory_failure(unsigned long pfn, int trapno);
  1165. extern int __memory_failure(unsigned long pfn, int trapno, int flags);
  1166. extern int unpoison_memory(unsigned long pfn);
  1167. extern int sysctl_memory_failure_early_kill;
  1168. extern int sysctl_memory_failure_recovery;
  1169. extern void shake_page(struct page *p, int access);
  1170. extern atomic_long_t mce_bad_pages;
  1171. extern int soft_offline_page(struct page *page, int flags);
  1172. #endif /* __KERNEL__ */
  1173. #endif /* _LINUX_MM_H */