mm.h 47 KB

12345678910111213141516171819202122232425262728293031323334353637383940414243444546474849505152535455565758596061626364656667686970717273747576777879808182838485868788899091929394959697989910010110210310410510610710810911011111211311411511611711811912012112212312412512612712812913013113213313413513613713813914014114214314414514614714814915015115215315415515615715815916016116216316416516616716816917017117217317417517617717817918018118218318418518618718818919019119219319419519619719819920020120220320420520620720820921021121221321421521621721821922022122222322422522622722822923023123223323423523623723823924024124224324424524624724824925025125225325425525625725825926026126226326426526626726826927027127227327427527627727827928028128228328428528628728828929029129229329429529629729829930030130230330430530630730830931031131231331431531631731831932032132232332432532632732832933033133233333433533633733833934034134234334434534634734834935035135235335435535635735835936036136236336436536636736836937037137237337437537637737837938038138238338438538638738838939039139239339439539639739839940040140240340440540640740840941041141241341441541641741841942042142242342442542642742842943043143243343443543643743843944044144244344444544644744844945045145245345445545645745845946046146246346446546646746846947047147247347447547647747847948048148248348448548648748848949049149249349449549649749849950050150250350450550650750850951051151251351451551651751851952052152252352452552652752852953053153253353453553653753853954054154254354454554654754854955055155255355455555655755855956056156256356456556656756856957057157257357457557657757857958058158258358458558658758858959059159259359459559659759859960060160260360460560660760860961061161261361461561661761861962062162262362462562662762862963063163263363463563663763863964064164264364464564664764864965065165265365465565665765865966066166266366466566666766866967067167267367467567667767867968068168268368468568668768868969069169269369469569669769869970070170270370470570670770870971071171271371471571671771871972072172272372472572672772872973073173273373473573673773873974074174274374474574674774874975075175275375475575675775875976076176276376476576676776876977077177277377477577677777877978078178278378478578678778878979079179279379479579679779879980080180280380480580680780880981081181281381481581681781881982082182282382482582682782882983083183283383483583683783883984084184284384484584684784884985085185285385485585685785885986086186286386486586686786886987087187287387487587687787887988088188288388488588688788888989089189289389489589689789889990090190290390490590690790890991091191291391491591691791891992092192292392492592692792892993093193293393493593693793893994094194294394494594694794894995095195295395495595695795895996096196296396496596696796896997097197297397497597697797897998098198298398498598698798898999099199299399499599699799899910001001100210031004100510061007100810091010101110121013101410151016101710181019102010211022102310241025102610271028102910301031103210331034103510361037103810391040104110421043104410451046104710481049105010511052105310541055105610571058105910601061106210631064106510661067106810691070107110721073107410751076107710781079108010811082108310841085108610871088108910901091109210931094109510961097109810991100110111021103110411051106110711081109111011111112111311141115111611171118111911201121112211231124112511261127112811291130113111321133113411351136113711381139114011411142114311441145114611471148114911501151115211531154115511561157115811591160116111621163116411651166116711681169117011711172117311741175117611771178117911801181118211831184118511861187118811891190119111921193119411951196119711981199120012011202120312041205120612071208120912101211121212131214121512161217121812191220122112221223122412251226122712281229123012311232123312341235123612371238123912401241124212431244124512461247124812491250125112521253125412551256125712581259126012611262126312641265126612671268126912701271127212731274127512761277127812791280128112821283128412851286128712881289129012911292129312941295129612971298129913001301130213031304130513061307130813091310131113121313131413151316131713181319132013211322132313241325132613271328132913301331133213331334133513361337133813391340134113421343134413451346134713481349
  1. #ifndef _LINUX_MM_H
  2. #define _LINUX_MM_H
  3. #include <linux/errno.h>
  4. #ifdef __KERNEL__
  5. #include <linux/gfp.h>
  6. #include <linux/list.h>
  7. #include <linux/mmzone.h>
  8. #include <linux/rbtree.h>
  9. #include <linux/prio_tree.h>
  10. #include <linux/debug_locks.h>
  11. #include <linux/mm_types.h>
  12. struct mempolicy;
  13. struct anon_vma;
  14. struct file_ra_state;
  15. struct user_struct;
  16. struct writeback_control;
  17. struct rlimit;
  18. #ifndef CONFIG_DISCONTIGMEM /* Don't use mapnrs, do it properly */
  19. extern unsigned long max_mapnr;
  20. #endif
  21. extern unsigned long num_physpages;
  22. extern unsigned long totalram_pages;
  23. extern void * high_memory;
  24. extern int page_cluster;
  25. #ifdef CONFIG_SYSCTL
  26. extern int sysctl_legacy_va_layout;
  27. #else
  28. #define sysctl_legacy_va_layout 0
  29. #endif
  30. #include <asm/page.h>
  31. #include <asm/pgtable.h>
  32. #include <asm/processor.h>
  33. #define nth_page(page,n) pfn_to_page(page_to_pfn((page)) + (n))
  34. /* to align the pointer to the (next) page boundary */
  35. #define PAGE_ALIGN(addr) ALIGN(addr, PAGE_SIZE)
  36. /*
  37. * Linux kernel virtual memory manager primitives.
  38. * The idea being to have a "virtual" mm in the same way
  39. * we have a virtual fs - giving a cleaner interface to the
  40. * mm details, and allowing different kinds of memory mappings
  41. * (from shared memory to executable loading to arbitrary
  42. * mmap() functions).
  43. */
  44. extern struct kmem_cache *vm_area_cachep;
  45. #ifndef CONFIG_MMU
  46. extern struct rb_root nommu_region_tree;
  47. extern struct rw_semaphore nommu_region_sem;
  48. extern unsigned int kobjsize(const void *objp);
  49. #endif
  50. /*
  51. * vm_flags in vm_area_struct, see mm_types.h.
  52. */
  53. #define VM_READ 0x00000001 /* currently active flags */
  54. #define VM_WRITE 0x00000002
  55. #define VM_EXEC 0x00000004
  56. #define VM_SHARED 0x00000008
  57. /* mprotect() hardcodes VM_MAYREAD >> 4 == VM_READ, and so for r/w/x bits. */
  58. #define VM_MAYREAD 0x00000010 /* limits for mprotect() etc */
  59. #define VM_MAYWRITE 0x00000020
  60. #define VM_MAYEXEC 0x00000040
  61. #define VM_MAYSHARE 0x00000080
  62. #define VM_GROWSDOWN 0x00000100 /* general info on the segment */
  63. #define VM_GROWSUP 0x00000200
  64. #define VM_PFNMAP 0x00000400 /* Page-ranges managed without "struct page", just pure PFN */
  65. #define VM_DENYWRITE 0x00000800 /* ETXTBSY on write attempts.. */
  66. #define VM_EXECUTABLE 0x00001000
  67. #define VM_LOCKED 0x00002000
  68. #define VM_IO 0x00004000 /* Memory mapped I/O or similar */
  69. /* Used by sys_madvise() */
  70. #define VM_SEQ_READ 0x00008000 /* App will access data sequentially */
  71. #define VM_RAND_READ 0x00010000 /* App will not benefit from clustered reads */
  72. #define VM_DONTCOPY 0x00020000 /* Do not copy this vma on fork */
  73. #define VM_DONTEXPAND 0x00040000 /* Cannot expand with mremap() */
  74. #define VM_RESERVED 0x00080000 /* Count as reserved_vm like IO */
  75. #define VM_ACCOUNT 0x00100000 /* Is a VM accounted object */
  76. #define VM_NORESERVE 0x00200000 /* should the VM suppress accounting */
  77. #define VM_HUGETLB 0x00400000 /* Huge TLB Page VM */
  78. #define VM_NONLINEAR 0x00800000 /* Is non-linear (remap_file_pages) */
  79. #define VM_MAPPED_COPY 0x01000000 /* T if mapped copy of data (nommu mmap) */
  80. #define VM_INSERTPAGE 0x02000000 /* The vma has had "vm_insert_page()" done on it */
  81. #define VM_ALWAYSDUMP 0x04000000 /* Always include in core dumps */
  82. #define VM_CAN_NONLINEAR 0x08000000 /* Has ->fault & does nonlinear pages */
  83. #define VM_MIXEDMAP 0x10000000 /* Can contain "struct page" and pure PFN pages */
  84. #define VM_SAO 0x20000000 /* Strong Access Ordering (powerpc) */
  85. #define VM_PFN_AT_MMAP 0x40000000 /* PFNMAP vma that is fully mapped at mmap time */
  86. #define VM_MERGEABLE 0x80000000 /* KSM may merge identical pages */
  87. #ifndef VM_STACK_DEFAULT_FLAGS /* arch can override this */
  88. #define VM_STACK_DEFAULT_FLAGS VM_DATA_DEFAULT_FLAGS
  89. #endif
  90. #ifdef CONFIG_STACK_GROWSUP
  91. #define VM_STACK_FLAGS (VM_GROWSUP | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT)
  92. #else
  93. #define VM_STACK_FLAGS (VM_GROWSDOWN | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT)
  94. #endif
  95. #define VM_READHINTMASK (VM_SEQ_READ | VM_RAND_READ)
  96. #define VM_ClearReadHint(v) (v)->vm_flags &= ~VM_READHINTMASK
  97. #define VM_NormalReadHint(v) (!((v)->vm_flags & VM_READHINTMASK))
  98. #define VM_SequentialReadHint(v) ((v)->vm_flags & VM_SEQ_READ)
  99. #define VM_RandomReadHint(v) ((v)->vm_flags & VM_RAND_READ)
  100. /*
  101. * special vmas that are non-mergable, non-mlock()able
  102. */
  103. #define VM_SPECIAL (VM_IO | VM_DONTEXPAND | VM_RESERVED | VM_PFNMAP)
  104. /*
  105. * mapping from the currently active vm_flags protection bits (the
  106. * low four bits) to a page protection mask..
  107. */
  108. extern pgprot_t protection_map[16];
  109. #define FAULT_FLAG_WRITE 0x01 /* Fault was a write access */
  110. #define FAULT_FLAG_NONLINEAR 0x02 /* Fault was via a nonlinear mapping */
  111. #define FAULT_FLAG_MKWRITE 0x04 /* Fault was mkwrite of existing pte */
  112. /*
  113. * This interface is used by x86 PAT code to identify a pfn mapping that is
  114. * linear over entire vma. This is to optimize PAT code that deals with
  115. * marking the physical region with a particular prot. This is not for generic
  116. * mm use. Note also that this check will not work if the pfn mapping is
  117. * linear for a vma starting at physical address 0. In which case PAT code
  118. * falls back to slow path of reserving physical range page by page.
  119. */
  120. static inline int is_linear_pfn_mapping(struct vm_area_struct *vma)
  121. {
  122. return (vma->vm_flags & VM_PFN_AT_MMAP);
  123. }
  124. static inline int is_pfn_mapping(struct vm_area_struct *vma)
  125. {
  126. return (vma->vm_flags & VM_PFNMAP);
  127. }
  128. /*
  129. * vm_fault is filled by the the pagefault handler and passed to the vma's
  130. * ->fault function. The vma's ->fault is responsible for returning a bitmask
  131. * of VM_FAULT_xxx flags that give details about how the fault was handled.
  132. *
  133. * pgoff should be used in favour of virtual_address, if possible. If pgoff
  134. * is used, one may set VM_CAN_NONLINEAR in the vma->vm_flags to get nonlinear
  135. * mapping support.
  136. */
  137. struct vm_fault {
  138. unsigned int flags; /* FAULT_FLAG_xxx flags */
  139. pgoff_t pgoff; /* Logical page offset based on vma */
  140. void __user *virtual_address; /* Faulting virtual address */
  141. struct page *page; /* ->fault handlers should return a
  142. * page here, unless VM_FAULT_NOPAGE
  143. * is set (which is also implied by
  144. * VM_FAULT_ERROR).
  145. */
  146. };
  147. /*
  148. * These are the virtual MM functions - opening of an area, closing and
  149. * unmapping it (needed to keep files on disk up-to-date etc), pointer
  150. * to the functions called when a no-page or a wp-page exception occurs.
  151. */
  152. struct vm_operations_struct {
  153. void (*open)(struct vm_area_struct * area);
  154. void (*close)(struct vm_area_struct * area);
  155. int (*fault)(struct vm_area_struct *vma, struct vm_fault *vmf);
  156. /* notification that a previously read-only page is about to become
  157. * writable, if an error is returned it will cause a SIGBUS */
  158. int (*page_mkwrite)(struct vm_area_struct *vma, struct vm_fault *vmf);
  159. /* called by access_process_vm when get_user_pages() fails, typically
  160. * for use by special VMAs that can switch between memory and hardware
  161. */
  162. int (*access)(struct vm_area_struct *vma, unsigned long addr,
  163. void *buf, int len, int write);
  164. #ifdef CONFIG_NUMA
  165. /*
  166. * set_policy() op must add a reference to any non-NULL @new mempolicy
  167. * to hold the policy upon return. Caller should pass NULL @new to
  168. * remove a policy and fall back to surrounding context--i.e. do not
  169. * install a MPOL_DEFAULT policy, nor the task or system default
  170. * mempolicy.
  171. */
  172. int (*set_policy)(struct vm_area_struct *vma, struct mempolicy *new);
  173. /*
  174. * get_policy() op must add reference [mpol_get()] to any policy at
  175. * (vma,addr) marked as MPOL_SHARED. The shared policy infrastructure
  176. * in mm/mempolicy.c will do this automatically.
  177. * get_policy() must NOT add a ref if the policy at (vma,addr) is not
  178. * marked as MPOL_SHARED. vma policies are protected by the mmap_sem.
  179. * If no [shared/vma] mempolicy exists at the addr, get_policy() op
  180. * must return NULL--i.e., do not "fallback" to task or system default
  181. * policy.
  182. */
  183. struct mempolicy *(*get_policy)(struct vm_area_struct *vma,
  184. unsigned long addr);
  185. int (*migrate)(struct vm_area_struct *vma, const nodemask_t *from,
  186. const nodemask_t *to, unsigned long flags);
  187. #endif
  188. };
  189. struct mmu_gather;
  190. struct inode;
  191. #define page_private(page) ((page)->private)
  192. #define set_page_private(page, v) ((page)->private = (v))
  193. /*
  194. * FIXME: take this include out, include page-flags.h in
  195. * files which need it (119 of them)
  196. */
  197. #include <linux/page-flags.h>
  198. /*
  199. * Methods to modify the page usage count.
  200. *
  201. * What counts for a page usage:
  202. * - cache mapping (page->mapping)
  203. * - private data (page->private)
  204. * - page mapped in a task's page tables, each mapping
  205. * is counted separately
  206. *
  207. * Also, many kernel routines increase the page count before a critical
  208. * routine so they can be sure the page doesn't go away from under them.
  209. */
  210. /*
  211. * Drop a ref, return true if the refcount fell to zero (the page has no users)
  212. */
  213. static inline int put_page_testzero(struct page *page)
  214. {
  215. VM_BUG_ON(atomic_read(&page->_count) == 0);
  216. return atomic_dec_and_test(&page->_count);
  217. }
  218. /*
  219. * Try to grab a ref unless the page has a refcount of zero, return false if
  220. * that is the case.
  221. */
  222. static inline int get_page_unless_zero(struct page *page)
  223. {
  224. return atomic_inc_not_zero(&page->_count);
  225. }
  226. extern int page_is_ram(unsigned long pfn);
  227. /* Support for virtually mapped pages */
  228. struct page *vmalloc_to_page(const void *addr);
  229. unsigned long vmalloc_to_pfn(const void *addr);
  230. /*
  231. * Determine if an address is within the vmalloc range
  232. *
  233. * On nommu, vmalloc/vfree wrap through kmalloc/kfree directly, so there
  234. * is no special casing required.
  235. */
  236. static inline int is_vmalloc_addr(const void *x)
  237. {
  238. #ifdef CONFIG_MMU
  239. unsigned long addr = (unsigned long)x;
  240. return addr >= VMALLOC_START && addr < VMALLOC_END;
  241. #else
  242. return 0;
  243. #endif
  244. }
  245. #ifdef CONFIG_MMU
  246. extern int is_vmalloc_or_module_addr(const void *x);
  247. #else
  248. static inline int is_vmalloc_or_module_addr(const void *x)
  249. {
  250. return 0;
  251. }
  252. #endif
  253. static inline struct page *compound_head(struct page *page)
  254. {
  255. if (unlikely(PageTail(page)))
  256. return page->first_page;
  257. return page;
  258. }
  259. static inline int page_count(struct page *page)
  260. {
  261. return atomic_read(&compound_head(page)->_count);
  262. }
  263. static inline void get_page(struct page *page)
  264. {
  265. page = compound_head(page);
  266. VM_BUG_ON(atomic_read(&page->_count) == 0);
  267. atomic_inc(&page->_count);
  268. }
  269. static inline struct page *virt_to_head_page(const void *x)
  270. {
  271. struct page *page = virt_to_page(x);
  272. return compound_head(page);
  273. }
  274. /*
  275. * Setup the page count before being freed into the page allocator for
  276. * the first time (boot or memory hotplug)
  277. */
  278. static inline void init_page_count(struct page *page)
  279. {
  280. atomic_set(&page->_count, 1);
  281. }
  282. void put_page(struct page *page);
  283. void put_pages_list(struct list_head *pages);
  284. void split_page(struct page *page, unsigned int order);
  285. /*
  286. * Compound pages have a destructor function. Provide a
  287. * prototype for that function and accessor functions.
  288. * These are _only_ valid on the head of a PG_compound page.
  289. */
  290. typedef void compound_page_dtor(struct page *);
  291. static inline void set_compound_page_dtor(struct page *page,
  292. compound_page_dtor *dtor)
  293. {
  294. page[1].lru.next = (void *)dtor;
  295. }
  296. static inline compound_page_dtor *get_compound_page_dtor(struct page *page)
  297. {
  298. return (compound_page_dtor *)page[1].lru.next;
  299. }
  300. static inline int compound_order(struct page *page)
  301. {
  302. if (!PageHead(page))
  303. return 0;
  304. return (unsigned long)page[1].lru.prev;
  305. }
  306. static inline void set_compound_order(struct page *page, unsigned long order)
  307. {
  308. page[1].lru.prev = (void *)order;
  309. }
  310. /*
  311. * Multiple processes may "see" the same page. E.g. for untouched
  312. * mappings of /dev/null, all processes see the same page full of
  313. * zeroes, and text pages of executables and shared libraries have
  314. * only one copy in memory, at most, normally.
  315. *
  316. * For the non-reserved pages, page_count(page) denotes a reference count.
  317. * page_count() == 0 means the page is free. page->lru is then used for
  318. * freelist management in the buddy allocator.
  319. * page_count() > 0 means the page has been allocated.
  320. *
  321. * Pages are allocated by the slab allocator in order to provide memory
  322. * to kmalloc and kmem_cache_alloc. In this case, the management of the
  323. * page, and the fields in 'struct page' are the responsibility of mm/slab.c
  324. * unless a particular usage is carefully commented. (the responsibility of
  325. * freeing the kmalloc memory is the caller's, of course).
  326. *
  327. * A page may be used by anyone else who does a __get_free_page().
  328. * In this case, page_count still tracks the references, and should only
  329. * be used through the normal accessor functions. The top bits of page->flags
  330. * and page->virtual store page management information, but all other fields
  331. * are unused and could be used privately, carefully. The management of this
  332. * page is the responsibility of the one who allocated it, and those who have
  333. * subsequently been given references to it.
  334. *
  335. * The other pages (we may call them "pagecache pages") are completely
  336. * managed by the Linux memory manager: I/O, buffers, swapping etc.
  337. * The following discussion applies only to them.
  338. *
  339. * A pagecache page contains an opaque `private' member, which belongs to the
  340. * page's address_space. Usually, this is the address of a circular list of
  341. * the page's disk buffers. PG_private must be set to tell the VM to call
  342. * into the filesystem to release these pages.
  343. *
  344. * A page may belong to an inode's memory mapping. In this case, page->mapping
  345. * is the pointer to the inode, and page->index is the file offset of the page,
  346. * in units of PAGE_CACHE_SIZE.
  347. *
  348. * If pagecache pages are not associated with an inode, they are said to be
  349. * anonymous pages. These may become associated with the swapcache, and in that
  350. * case PG_swapcache is set, and page->private is an offset into the swapcache.
  351. *
  352. * In either case (swapcache or inode backed), the pagecache itself holds one
  353. * reference to the page. Setting PG_private should also increment the
  354. * refcount. The each user mapping also has a reference to the page.
  355. *
  356. * The pagecache pages are stored in a per-mapping radix tree, which is
  357. * rooted at mapping->page_tree, and indexed by offset.
  358. * Where 2.4 and early 2.6 kernels kept dirty/clean pages in per-address_space
  359. * lists, we instead now tag pages as dirty/writeback in the radix tree.
  360. *
  361. * All pagecache pages may be subject to I/O:
  362. * - inode pages may need to be read from disk,
  363. * - inode pages which have been modified and are MAP_SHARED may need
  364. * to be written back to the inode on disk,
  365. * - anonymous pages (including MAP_PRIVATE file mappings) which have been
  366. * modified may need to be swapped out to swap space and (later) to be read
  367. * back into memory.
  368. */
  369. /*
  370. * The zone field is never updated after free_area_init_core()
  371. * sets it, so none of the operations on it need to be atomic.
  372. */
  373. /*
  374. * page->flags layout:
  375. *
  376. * There are three possibilities for how page->flags get
  377. * laid out. The first is for the normal case, without
  378. * sparsemem. The second is for sparsemem when there is
  379. * plenty of space for node and section. The last is when
  380. * we have run out of space and have to fall back to an
  381. * alternate (slower) way of determining the node.
  382. *
  383. * No sparsemem or sparsemem vmemmap: | NODE | ZONE | ... | FLAGS |
  384. * classic sparse with space for node:| SECTION | NODE | ZONE | ... | FLAGS |
  385. * classic sparse no space for node: | SECTION | ZONE | ... | FLAGS |
  386. */
  387. #if defined(CONFIG_SPARSEMEM) && !defined(CONFIG_SPARSEMEM_VMEMMAP)
  388. #define SECTIONS_WIDTH SECTIONS_SHIFT
  389. #else
  390. #define SECTIONS_WIDTH 0
  391. #endif
  392. #define ZONES_WIDTH ZONES_SHIFT
  393. #if SECTIONS_WIDTH+ZONES_WIDTH+NODES_SHIFT <= BITS_PER_LONG - NR_PAGEFLAGS
  394. #define NODES_WIDTH NODES_SHIFT
  395. #else
  396. #ifdef CONFIG_SPARSEMEM_VMEMMAP
  397. #error "Vmemmap: No space for nodes field in page flags"
  398. #endif
  399. #define NODES_WIDTH 0
  400. #endif
  401. /* Page flags: | [SECTION] | [NODE] | ZONE | ... | FLAGS | */
  402. #define SECTIONS_PGOFF ((sizeof(unsigned long)*8) - SECTIONS_WIDTH)
  403. #define NODES_PGOFF (SECTIONS_PGOFF - NODES_WIDTH)
  404. #define ZONES_PGOFF (NODES_PGOFF - ZONES_WIDTH)
  405. /*
  406. * We are going to use the flags for the page to node mapping if its in
  407. * there. This includes the case where there is no node, so it is implicit.
  408. */
  409. #if !(NODES_WIDTH > 0 || NODES_SHIFT == 0)
  410. #define NODE_NOT_IN_PAGE_FLAGS
  411. #endif
  412. #ifndef PFN_SECTION_SHIFT
  413. #define PFN_SECTION_SHIFT 0
  414. #endif
  415. /*
  416. * Define the bit shifts to access each section. For non-existant
  417. * sections we define the shift as 0; that plus a 0 mask ensures
  418. * the compiler will optimise away reference to them.
  419. */
  420. #define SECTIONS_PGSHIFT (SECTIONS_PGOFF * (SECTIONS_WIDTH != 0))
  421. #define NODES_PGSHIFT (NODES_PGOFF * (NODES_WIDTH != 0))
  422. #define ZONES_PGSHIFT (ZONES_PGOFF * (ZONES_WIDTH != 0))
  423. /* NODE:ZONE or SECTION:ZONE is used to ID a zone for the buddy allcator */
  424. #ifdef NODE_NOT_IN_PAGEFLAGS
  425. #define ZONEID_SHIFT (SECTIONS_SHIFT + ZONES_SHIFT)
  426. #define ZONEID_PGOFF ((SECTIONS_PGOFF < ZONES_PGOFF)? \
  427. SECTIONS_PGOFF : ZONES_PGOFF)
  428. #else
  429. #define ZONEID_SHIFT (NODES_SHIFT + ZONES_SHIFT)
  430. #define ZONEID_PGOFF ((NODES_PGOFF < ZONES_PGOFF)? \
  431. NODES_PGOFF : ZONES_PGOFF)
  432. #endif
  433. #define ZONEID_PGSHIFT (ZONEID_PGOFF * (ZONEID_SHIFT != 0))
  434. #if SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > BITS_PER_LONG - NR_PAGEFLAGS
  435. #error SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > BITS_PER_LONG - NR_PAGEFLAGS
  436. #endif
  437. #define ZONES_MASK ((1UL << ZONES_WIDTH) - 1)
  438. #define NODES_MASK ((1UL << NODES_WIDTH) - 1)
  439. #define SECTIONS_MASK ((1UL << SECTIONS_WIDTH) - 1)
  440. #define ZONEID_MASK ((1UL << ZONEID_SHIFT) - 1)
  441. static inline enum zone_type page_zonenum(struct page *page)
  442. {
  443. return (page->flags >> ZONES_PGSHIFT) & ZONES_MASK;
  444. }
  445. /*
  446. * The identification function is only used by the buddy allocator for
  447. * determining if two pages could be buddies. We are not really
  448. * identifying a zone since we could be using a the section number
  449. * id if we have not node id available in page flags.
  450. * We guarantee only that it will return the same value for two
  451. * combinable pages in a zone.
  452. */
  453. static inline int page_zone_id(struct page *page)
  454. {
  455. return (page->flags >> ZONEID_PGSHIFT) & ZONEID_MASK;
  456. }
  457. static inline int zone_to_nid(struct zone *zone)
  458. {
  459. #ifdef CONFIG_NUMA
  460. return zone->node;
  461. #else
  462. return 0;
  463. #endif
  464. }
  465. #ifdef NODE_NOT_IN_PAGE_FLAGS
  466. extern int page_to_nid(struct page *page);
  467. #else
  468. static inline int page_to_nid(struct page *page)
  469. {
  470. return (page->flags >> NODES_PGSHIFT) & NODES_MASK;
  471. }
  472. #endif
  473. static inline struct zone *page_zone(struct page *page)
  474. {
  475. return &NODE_DATA(page_to_nid(page))->node_zones[page_zonenum(page)];
  476. }
  477. #if defined(CONFIG_SPARSEMEM) && !defined(CONFIG_SPARSEMEM_VMEMMAP)
  478. static inline unsigned long page_to_section(struct page *page)
  479. {
  480. return (page->flags >> SECTIONS_PGSHIFT) & SECTIONS_MASK;
  481. }
  482. #endif
  483. static inline void set_page_zone(struct page *page, enum zone_type zone)
  484. {
  485. page->flags &= ~(ZONES_MASK << ZONES_PGSHIFT);
  486. page->flags |= (zone & ZONES_MASK) << ZONES_PGSHIFT;
  487. }
  488. static inline void set_page_node(struct page *page, unsigned long node)
  489. {
  490. page->flags &= ~(NODES_MASK << NODES_PGSHIFT);
  491. page->flags |= (node & NODES_MASK) << NODES_PGSHIFT;
  492. }
  493. static inline void set_page_section(struct page *page, unsigned long section)
  494. {
  495. page->flags &= ~(SECTIONS_MASK << SECTIONS_PGSHIFT);
  496. page->flags |= (section & SECTIONS_MASK) << SECTIONS_PGSHIFT;
  497. }
  498. static inline void set_page_links(struct page *page, enum zone_type zone,
  499. unsigned long node, unsigned long pfn)
  500. {
  501. set_page_zone(page, zone);
  502. set_page_node(page, node);
  503. set_page_section(page, pfn_to_section_nr(pfn));
  504. }
  505. /*
  506. * Some inline functions in vmstat.h depend on page_zone()
  507. */
  508. #include <linux/vmstat.h>
  509. static __always_inline void *lowmem_page_address(struct page *page)
  510. {
  511. return __va(page_to_pfn(page) << PAGE_SHIFT);
  512. }
  513. #if defined(CONFIG_HIGHMEM) && !defined(WANT_PAGE_VIRTUAL)
  514. #define HASHED_PAGE_VIRTUAL
  515. #endif
  516. #if defined(WANT_PAGE_VIRTUAL)
  517. #define page_address(page) ((page)->virtual)
  518. #define set_page_address(page, address) \
  519. do { \
  520. (page)->virtual = (address); \
  521. } while(0)
  522. #define page_address_init() do { } while(0)
  523. #endif
  524. #if defined(HASHED_PAGE_VIRTUAL)
  525. void *page_address(struct page *page);
  526. void set_page_address(struct page *page, void *virtual);
  527. void page_address_init(void);
  528. #endif
  529. #if !defined(HASHED_PAGE_VIRTUAL) && !defined(WANT_PAGE_VIRTUAL)
  530. #define page_address(page) lowmem_page_address(page)
  531. #define set_page_address(page, address) do { } while(0)
  532. #define page_address_init() do { } while(0)
  533. #endif
  534. /*
  535. * On an anonymous page mapped into a user virtual memory area,
  536. * page->mapping points to its anon_vma, not to a struct address_space;
  537. * with the PAGE_MAPPING_ANON bit set to distinguish it. See rmap.h.
  538. *
  539. * On an anonymous page in a VM_MERGEABLE area, if CONFIG_KSM is enabled,
  540. * the PAGE_MAPPING_KSM bit may be set along with the PAGE_MAPPING_ANON bit;
  541. * and then page->mapping points, not to an anon_vma, but to a private
  542. * structure which KSM associates with that merged page. See ksm.h.
  543. *
  544. * PAGE_MAPPING_KSM without PAGE_MAPPING_ANON is currently never used.
  545. *
  546. * Please note that, confusingly, "page_mapping" refers to the inode
  547. * address_space which maps the page from disk; whereas "page_mapped"
  548. * refers to user virtual address space into which the page is mapped.
  549. */
  550. #define PAGE_MAPPING_ANON 1
  551. #define PAGE_MAPPING_KSM 2
  552. #define PAGE_MAPPING_FLAGS (PAGE_MAPPING_ANON | PAGE_MAPPING_KSM)
  553. extern struct address_space swapper_space;
  554. static inline struct address_space *page_mapping(struct page *page)
  555. {
  556. struct address_space *mapping = page->mapping;
  557. VM_BUG_ON(PageSlab(page));
  558. if (unlikely(PageSwapCache(page)))
  559. mapping = &swapper_space;
  560. else if (unlikely((unsigned long)mapping & PAGE_MAPPING_ANON))
  561. mapping = NULL;
  562. return mapping;
  563. }
  564. /* Neutral page->mapping pointer to address_space or anon_vma or other */
  565. static inline void *page_rmapping(struct page *page)
  566. {
  567. return (void *)((unsigned long)page->mapping & ~PAGE_MAPPING_FLAGS);
  568. }
  569. static inline int PageAnon(struct page *page)
  570. {
  571. return ((unsigned long)page->mapping & PAGE_MAPPING_ANON) != 0;
  572. }
  573. /*
  574. * Return the pagecache index of the passed page. Regular pagecache pages
  575. * use ->index whereas swapcache pages use ->private
  576. */
  577. static inline pgoff_t page_index(struct page *page)
  578. {
  579. if (unlikely(PageSwapCache(page)))
  580. return page_private(page);
  581. return page->index;
  582. }
  583. /*
  584. * The atomic page->_mapcount, like _count, starts from -1:
  585. * so that transitions both from it and to it can be tracked,
  586. * using atomic_inc_and_test and atomic_add_negative(-1).
  587. */
  588. static inline void reset_page_mapcount(struct page *page)
  589. {
  590. atomic_set(&(page)->_mapcount, -1);
  591. }
  592. static inline int page_mapcount(struct page *page)
  593. {
  594. return atomic_read(&(page)->_mapcount) + 1;
  595. }
  596. /*
  597. * Return true if this page is mapped into pagetables.
  598. */
  599. static inline int page_mapped(struct page *page)
  600. {
  601. return atomic_read(&(page)->_mapcount) >= 0;
  602. }
  603. /*
  604. * Different kinds of faults, as returned by handle_mm_fault().
  605. * Used to decide whether a process gets delivered SIGBUS or
  606. * just gets major/minor fault counters bumped up.
  607. */
  608. #define VM_FAULT_MINOR 0 /* For backwards compat. Remove me quickly. */
  609. #define VM_FAULT_OOM 0x0001
  610. #define VM_FAULT_SIGBUS 0x0002
  611. #define VM_FAULT_MAJOR 0x0004
  612. #define VM_FAULT_WRITE 0x0008 /* Special case for get_user_pages */
  613. #define VM_FAULT_HWPOISON 0x0010 /* Hit poisoned page */
  614. #define VM_FAULT_NOPAGE 0x0100 /* ->fault installed the pte, not return page */
  615. #define VM_FAULT_LOCKED 0x0200 /* ->fault locked the returned page */
  616. #define VM_FAULT_ERROR (VM_FAULT_OOM | VM_FAULT_SIGBUS | VM_FAULT_HWPOISON)
  617. /*
  618. * Can be called by the pagefault handler when it gets a VM_FAULT_OOM.
  619. */
  620. extern void pagefault_out_of_memory(void);
  621. #define offset_in_page(p) ((unsigned long)(p) & ~PAGE_MASK)
  622. extern void show_free_areas(void);
  623. int shmem_lock(struct file *file, int lock, struct user_struct *user);
  624. struct file *shmem_file_setup(const char *name, loff_t size, unsigned long flags);
  625. int shmem_zero_setup(struct vm_area_struct *);
  626. #ifndef CONFIG_MMU
  627. extern unsigned long shmem_get_unmapped_area(struct file *file,
  628. unsigned long addr,
  629. unsigned long len,
  630. unsigned long pgoff,
  631. unsigned long flags);
  632. #endif
  633. extern int can_do_mlock(void);
  634. extern int user_shm_lock(size_t, struct user_struct *);
  635. extern void user_shm_unlock(size_t, struct user_struct *);
  636. /*
  637. * Parameter block passed down to zap_pte_range in exceptional cases.
  638. */
  639. struct zap_details {
  640. struct vm_area_struct *nonlinear_vma; /* Check page->index if set */
  641. struct address_space *check_mapping; /* Check page->mapping if set */
  642. pgoff_t first_index; /* Lowest page->index to unmap */
  643. pgoff_t last_index; /* Highest page->index to unmap */
  644. spinlock_t *i_mmap_lock; /* For unmap_mapping_range: */
  645. unsigned long truncate_count; /* Compare vm_truncate_count */
  646. };
  647. struct page *vm_normal_page(struct vm_area_struct *vma, unsigned long addr,
  648. pte_t pte);
  649. int zap_vma_ptes(struct vm_area_struct *vma, unsigned long address,
  650. unsigned long size);
  651. unsigned long zap_page_range(struct vm_area_struct *vma, unsigned long address,
  652. unsigned long size, struct zap_details *);
  653. unsigned long unmap_vmas(struct mmu_gather **tlb,
  654. struct vm_area_struct *start_vma, unsigned long start_addr,
  655. unsigned long end_addr, unsigned long *nr_accounted,
  656. struct zap_details *);
  657. /**
  658. * mm_walk - callbacks for walk_page_range
  659. * @pgd_entry: if set, called for each non-empty PGD (top-level) entry
  660. * @pud_entry: if set, called for each non-empty PUD (2nd-level) entry
  661. * @pmd_entry: if set, called for each non-empty PMD (3rd-level) entry
  662. * @pte_entry: if set, called for each non-empty PTE (4th-level) entry
  663. * @pte_hole: if set, called for each hole at all levels
  664. * @hugetlb_entry: if set, called for each hugetlb entry
  665. *
  666. * (see walk_page_range for more details)
  667. */
  668. struct mm_walk {
  669. int (*pgd_entry)(pgd_t *, unsigned long, unsigned long, struct mm_walk *);
  670. int (*pud_entry)(pud_t *, unsigned long, unsigned long, struct mm_walk *);
  671. int (*pmd_entry)(pmd_t *, unsigned long, unsigned long, struct mm_walk *);
  672. int (*pte_entry)(pte_t *, unsigned long, unsigned long, struct mm_walk *);
  673. int (*pte_hole)(unsigned long, unsigned long, struct mm_walk *);
  674. int (*hugetlb_entry)(pte_t *, unsigned long, unsigned long,
  675. struct mm_walk *);
  676. struct mm_struct *mm;
  677. void *private;
  678. };
  679. int walk_page_range(unsigned long addr, unsigned long end,
  680. struct mm_walk *walk);
  681. void free_pgd_range(struct mmu_gather *tlb, unsigned long addr,
  682. unsigned long end, unsigned long floor, unsigned long ceiling);
  683. int copy_page_range(struct mm_struct *dst, struct mm_struct *src,
  684. struct vm_area_struct *vma);
  685. void unmap_mapping_range(struct address_space *mapping,
  686. loff_t const holebegin, loff_t const holelen, int even_cows);
  687. int follow_pfn(struct vm_area_struct *vma, unsigned long address,
  688. unsigned long *pfn);
  689. int follow_phys(struct vm_area_struct *vma, unsigned long address,
  690. unsigned int flags, unsigned long *prot, resource_size_t *phys);
  691. int generic_access_phys(struct vm_area_struct *vma, unsigned long addr,
  692. void *buf, int len, int write);
  693. static inline void unmap_shared_mapping_range(struct address_space *mapping,
  694. loff_t const holebegin, loff_t const holelen)
  695. {
  696. unmap_mapping_range(mapping, holebegin, holelen, 0);
  697. }
  698. extern void truncate_pagecache(struct inode *inode, loff_t old, loff_t new);
  699. extern int vmtruncate(struct inode *inode, loff_t offset);
  700. extern int vmtruncate_range(struct inode *inode, loff_t offset, loff_t end);
  701. int truncate_inode_page(struct address_space *mapping, struct page *page);
  702. int generic_error_remove_page(struct address_space *mapping, struct page *page);
  703. int invalidate_inode_page(struct page *page);
  704. #ifdef CONFIG_MMU
  705. extern int handle_mm_fault(struct mm_struct *mm, struct vm_area_struct *vma,
  706. unsigned long address, unsigned int flags);
  707. #else
  708. static inline int handle_mm_fault(struct mm_struct *mm,
  709. struct vm_area_struct *vma, unsigned long address,
  710. unsigned int flags)
  711. {
  712. /* should never happen if there's no MMU */
  713. BUG();
  714. return VM_FAULT_SIGBUS;
  715. }
  716. #endif
  717. extern int make_pages_present(unsigned long addr, unsigned long end);
  718. extern int access_process_vm(struct task_struct *tsk, unsigned long addr, void *buf, int len, int write);
  719. int get_user_pages(struct task_struct *tsk, struct mm_struct *mm,
  720. unsigned long start, int nr_pages, int write, int force,
  721. struct page **pages, struct vm_area_struct **vmas);
  722. int get_user_pages_fast(unsigned long start, int nr_pages, int write,
  723. struct page **pages);
  724. struct page *get_dump_page(unsigned long addr);
  725. extern int try_to_release_page(struct page * page, gfp_t gfp_mask);
  726. extern void do_invalidatepage(struct page *page, unsigned long offset);
  727. int __set_page_dirty_nobuffers(struct page *page);
  728. int __set_page_dirty_no_writeback(struct page *page);
  729. int redirty_page_for_writepage(struct writeback_control *wbc,
  730. struct page *page);
  731. void account_page_dirtied(struct page *page, struct address_space *mapping);
  732. int set_page_dirty(struct page *page);
  733. int set_page_dirty_lock(struct page *page);
  734. int clear_page_dirty_for_io(struct page *page);
  735. extern unsigned long move_page_tables(struct vm_area_struct *vma,
  736. unsigned long old_addr, struct vm_area_struct *new_vma,
  737. unsigned long new_addr, unsigned long len);
  738. extern unsigned long do_mremap(unsigned long addr,
  739. unsigned long old_len, unsigned long new_len,
  740. unsigned long flags, unsigned long new_addr);
  741. extern int mprotect_fixup(struct vm_area_struct *vma,
  742. struct vm_area_struct **pprev, unsigned long start,
  743. unsigned long end, unsigned long newflags);
  744. /*
  745. * doesn't attempt to fault and will return short.
  746. */
  747. int __get_user_pages_fast(unsigned long start, int nr_pages, int write,
  748. struct page **pages);
  749. /*
  750. * A callback you can register to apply pressure to ageable caches.
  751. *
  752. * 'shrink' is passed a count 'nr_to_scan' and a 'gfpmask'. It should
  753. * look through the least-recently-used 'nr_to_scan' entries and
  754. * attempt to free them up. It should return the number of objects
  755. * which remain in the cache. If it returns -1, it means it cannot do
  756. * any scanning at this time (eg. there is a risk of deadlock).
  757. *
  758. * The 'gfpmask' refers to the allocation we are currently trying to
  759. * fulfil.
  760. *
  761. * Note that 'shrink' will be passed nr_to_scan == 0 when the VM is
  762. * querying the cache size, so a fastpath for that case is appropriate.
  763. */
  764. struct shrinker {
  765. int (*shrink)(int nr_to_scan, gfp_t gfp_mask);
  766. int seeks; /* seeks to recreate an obj */
  767. /* These are for internal use */
  768. struct list_head list;
  769. long nr; /* objs pending delete */
  770. };
  771. #define DEFAULT_SEEKS 2 /* A good number if you don't know better. */
  772. extern void register_shrinker(struct shrinker *);
  773. extern void unregister_shrinker(struct shrinker *);
  774. int vma_wants_writenotify(struct vm_area_struct *vma);
  775. extern pte_t *get_locked_pte(struct mm_struct *mm, unsigned long addr, spinlock_t **ptl);
  776. #ifdef __PAGETABLE_PUD_FOLDED
  777. static inline int __pud_alloc(struct mm_struct *mm, pgd_t *pgd,
  778. unsigned long address)
  779. {
  780. return 0;
  781. }
  782. #else
  783. int __pud_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address);
  784. #endif
  785. #ifdef __PAGETABLE_PMD_FOLDED
  786. static inline int __pmd_alloc(struct mm_struct *mm, pud_t *pud,
  787. unsigned long address)
  788. {
  789. return 0;
  790. }
  791. #else
  792. int __pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address);
  793. #endif
  794. int __pte_alloc(struct mm_struct *mm, pmd_t *pmd, unsigned long address);
  795. int __pte_alloc_kernel(pmd_t *pmd, unsigned long address);
  796. /*
  797. * The following ifdef needed to get the 4level-fixup.h header to work.
  798. * Remove it when 4level-fixup.h has been removed.
  799. */
  800. #if defined(CONFIG_MMU) && !defined(__ARCH_HAS_4LEVEL_HACK)
  801. static inline pud_t *pud_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address)
  802. {
  803. return (unlikely(pgd_none(*pgd)) && __pud_alloc(mm, pgd, address))?
  804. NULL: pud_offset(pgd, address);
  805. }
  806. static inline pmd_t *pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address)
  807. {
  808. return (unlikely(pud_none(*pud)) && __pmd_alloc(mm, pud, address))?
  809. NULL: pmd_offset(pud, address);
  810. }
  811. #endif /* CONFIG_MMU && !__ARCH_HAS_4LEVEL_HACK */
  812. #if USE_SPLIT_PTLOCKS
  813. /*
  814. * We tuck a spinlock to guard each pagetable page into its struct page,
  815. * at page->private, with BUILD_BUG_ON to make sure that this will not
  816. * overflow into the next struct page (as it might with DEBUG_SPINLOCK).
  817. * When freeing, reset page->mapping so free_pages_check won't complain.
  818. */
  819. #define __pte_lockptr(page) &((page)->ptl)
  820. #define pte_lock_init(_page) do { \
  821. spin_lock_init(__pte_lockptr(_page)); \
  822. } while (0)
  823. #define pte_lock_deinit(page) ((page)->mapping = NULL)
  824. #define pte_lockptr(mm, pmd) ({(void)(mm); __pte_lockptr(pmd_page(*(pmd)));})
  825. #else /* !USE_SPLIT_PTLOCKS */
  826. /*
  827. * We use mm->page_table_lock to guard all pagetable pages of the mm.
  828. */
  829. #define pte_lock_init(page) do {} while (0)
  830. #define pte_lock_deinit(page) do {} while (0)
  831. #define pte_lockptr(mm, pmd) ({(void)(pmd); &(mm)->page_table_lock;})
  832. #endif /* USE_SPLIT_PTLOCKS */
  833. static inline void pgtable_page_ctor(struct page *page)
  834. {
  835. pte_lock_init(page);
  836. inc_zone_page_state(page, NR_PAGETABLE);
  837. }
  838. static inline void pgtable_page_dtor(struct page *page)
  839. {
  840. pte_lock_deinit(page);
  841. dec_zone_page_state(page, NR_PAGETABLE);
  842. }
  843. #define pte_offset_map_lock(mm, pmd, address, ptlp) \
  844. ({ \
  845. spinlock_t *__ptl = pte_lockptr(mm, pmd); \
  846. pte_t *__pte = pte_offset_map(pmd, address); \
  847. *(ptlp) = __ptl; \
  848. spin_lock(__ptl); \
  849. __pte; \
  850. })
  851. #define pte_unmap_unlock(pte, ptl) do { \
  852. spin_unlock(ptl); \
  853. pte_unmap(pte); \
  854. } while (0)
  855. #define pte_alloc_map(mm, pmd, address) \
  856. ((unlikely(!pmd_present(*(pmd))) && __pte_alloc(mm, pmd, address))? \
  857. NULL: pte_offset_map(pmd, address))
  858. #define pte_alloc_map_lock(mm, pmd, address, ptlp) \
  859. ((unlikely(!pmd_present(*(pmd))) && __pte_alloc(mm, pmd, address))? \
  860. NULL: pte_offset_map_lock(mm, pmd, address, ptlp))
  861. #define pte_alloc_kernel(pmd, address) \
  862. ((unlikely(!pmd_present(*(pmd))) && __pte_alloc_kernel(pmd, address))? \
  863. NULL: pte_offset_kernel(pmd, address))
  864. extern void free_area_init(unsigned long * zones_size);
  865. extern void free_area_init_node(int nid, unsigned long * zones_size,
  866. unsigned long zone_start_pfn, unsigned long *zholes_size);
  867. #ifdef CONFIG_ARCH_POPULATES_NODE_MAP
  868. /*
  869. * With CONFIG_ARCH_POPULATES_NODE_MAP set, an architecture may initialise its
  870. * zones, allocate the backing mem_map and account for memory holes in a more
  871. * architecture independent manner. This is a substitute for creating the
  872. * zone_sizes[] and zholes_size[] arrays and passing them to
  873. * free_area_init_node()
  874. *
  875. * An architecture is expected to register range of page frames backed by
  876. * physical memory with add_active_range() before calling
  877. * free_area_init_nodes() passing in the PFN each zone ends at. At a basic
  878. * usage, an architecture is expected to do something like
  879. *
  880. * unsigned long max_zone_pfns[MAX_NR_ZONES] = {max_dma, max_normal_pfn,
  881. * max_highmem_pfn};
  882. * for_each_valid_physical_page_range()
  883. * add_active_range(node_id, start_pfn, end_pfn)
  884. * free_area_init_nodes(max_zone_pfns);
  885. *
  886. * If the architecture guarantees that there are no holes in the ranges
  887. * registered with add_active_range(), free_bootmem_active_regions()
  888. * will call free_bootmem_node() for each registered physical page range.
  889. * Similarly sparse_memory_present_with_active_regions() calls
  890. * memory_present() for each range when SPARSEMEM is enabled.
  891. *
  892. * See mm/page_alloc.c for more information on each function exposed by
  893. * CONFIG_ARCH_POPULATES_NODE_MAP
  894. */
  895. extern void free_area_init_nodes(unsigned long *max_zone_pfn);
  896. extern void add_active_range(unsigned int nid, unsigned long start_pfn,
  897. unsigned long end_pfn);
  898. extern void remove_active_range(unsigned int nid, unsigned long start_pfn,
  899. unsigned long end_pfn);
  900. extern void remove_all_active_ranges(void);
  901. void sort_node_map(void);
  902. unsigned long __absent_pages_in_range(int nid, unsigned long start_pfn,
  903. unsigned long end_pfn);
  904. extern unsigned long absent_pages_in_range(unsigned long start_pfn,
  905. unsigned long end_pfn);
  906. extern void get_pfn_range_for_nid(unsigned int nid,
  907. unsigned long *start_pfn, unsigned long *end_pfn);
  908. extern unsigned long find_min_pfn_with_active_regions(void);
  909. extern void free_bootmem_with_active_regions(int nid,
  910. unsigned long max_low_pfn);
  911. typedef int (*work_fn_t)(unsigned long, unsigned long, void *);
  912. extern void work_with_active_regions(int nid, work_fn_t work_fn, void *data);
  913. extern void sparse_memory_present_with_active_regions(int nid);
  914. #endif /* CONFIG_ARCH_POPULATES_NODE_MAP */
  915. #if !defined(CONFIG_ARCH_POPULATES_NODE_MAP) && \
  916. !defined(CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID)
  917. static inline int __early_pfn_to_nid(unsigned long pfn)
  918. {
  919. return 0;
  920. }
  921. #else
  922. /* please see mm/page_alloc.c */
  923. extern int __meminit early_pfn_to_nid(unsigned long pfn);
  924. #ifdef CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID
  925. /* there is a per-arch backend function. */
  926. extern int __meminit __early_pfn_to_nid(unsigned long pfn);
  927. #endif /* CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID */
  928. #endif
  929. extern void set_dma_reserve(unsigned long new_dma_reserve);
  930. extern void memmap_init_zone(unsigned long, int, unsigned long,
  931. unsigned long, enum memmap_context);
  932. extern void setup_per_zone_wmarks(void);
  933. extern void calculate_zone_inactive_ratio(struct zone *zone);
  934. extern void mem_init(void);
  935. extern void __init mmap_init(void);
  936. extern void show_mem(void);
  937. extern void si_meminfo(struct sysinfo * val);
  938. extern void si_meminfo_node(struct sysinfo *val, int nid);
  939. extern int after_bootmem;
  940. extern void setup_per_cpu_pageset(void);
  941. extern void zone_pcp_update(struct zone *zone);
  942. /* nommu.c */
  943. extern atomic_long_t mmap_pages_allocated;
  944. extern int nommu_shrink_inode_mappings(struct inode *, size_t, size_t);
  945. /* prio_tree.c */
  946. void vma_prio_tree_add(struct vm_area_struct *, struct vm_area_struct *old);
  947. void vma_prio_tree_insert(struct vm_area_struct *, struct prio_tree_root *);
  948. void vma_prio_tree_remove(struct vm_area_struct *, struct prio_tree_root *);
  949. struct vm_area_struct *vma_prio_tree_next(struct vm_area_struct *vma,
  950. struct prio_tree_iter *iter);
  951. #define vma_prio_tree_foreach(vma, iter, root, begin, end) \
  952. for (prio_tree_iter_init(iter, root, begin, end), vma = NULL; \
  953. (vma = vma_prio_tree_next(vma, iter)); )
  954. static inline void vma_nonlinear_insert(struct vm_area_struct *vma,
  955. struct list_head *list)
  956. {
  957. vma->shared.vm_set.parent = NULL;
  958. list_add_tail(&vma->shared.vm_set.list, list);
  959. }
  960. /* mmap.c */
  961. extern int __vm_enough_memory(struct mm_struct *mm, long pages, int cap_sys_admin);
  962. extern void vma_adjust(struct vm_area_struct *vma, unsigned long start,
  963. unsigned long end, pgoff_t pgoff, struct vm_area_struct *insert);
  964. extern struct vm_area_struct *vma_merge(struct mm_struct *,
  965. struct vm_area_struct *prev, unsigned long addr, unsigned long end,
  966. unsigned long vm_flags, struct anon_vma *, struct file *, pgoff_t,
  967. struct mempolicy *);
  968. extern struct anon_vma *find_mergeable_anon_vma(struct vm_area_struct *);
  969. extern int split_vma(struct mm_struct *,
  970. struct vm_area_struct *, unsigned long addr, int new_below);
  971. extern int insert_vm_struct(struct mm_struct *, struct vm_area_struct *);
  972. extern void __vma_link_rb(struct mm_struct *, struct vm_area_struct *,
  973. struct rb_node **, struct rb_node *);
  974. extern void unlink_file_vma(struct vm_area_struct *);
  975. extern struct vm_area_struct *copy_vma(struct vm_area_struct **,
  976. unsigned long addr, unsigned long len, pgoff_t pgoff);
  977. extern void exit_mmap(struct mm_struct *);
  978. extern int mm_take_all_locks(struct mm_struct *mm);
  979. extern void mm_drop_all_locks(struct mm_struct *mm);
  980. #ifdef CONFIG_PROC_FS
  981. /* From fs/proc/base.c. callers must _not_ hold the mm's exe_file_lock */
  982. extern void added_exe_file_vma(struct mm_struct *mm);
  983. extern void removed_exe_file_vma(struct mm_struct *mm);
  984. #else
  985. static inline void added_exe_file_vma(struct mm_struct *mm)
  986. {}
  987. static inline void removed_exe_file_vma(struct mm_struct *mm)
  988. {}
  989. #endif /* CONFIG_PROC_FS */
  990. extern int may_expand_vm(struct mm_struct *mm, unsigned long npages);
  991. extern int install_special_mapping(struct mm_struct *mm,
  992. unsigned long addr, unsigned long len,
  993. unsigned long flags, struct page **pages);
  994. extern unsigned long get_unmapped_area(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
  995. extern unsigned long do_mmap_pgoff(struct file *file, unsigned long addr,
  996. unsigned long len, unsigned long prot,
  997. unsigned long flag, unsigned long pgoff);
  998. extern unsigned long mmap_region(struct file *file, unsigned long addr,
  999. unsigned long len, unsigned long flags,
  1000. unsigned int vm_flags, unsigned long pgoff);
  1001. static inline unsigned long do_mmap(struct file *file, unsigned long addr,
  1002. unsigned long len, unsigned long prot,
  1003. unsigned long flag, unsigned long offset)
  1004. {
  1005. unsigned long ret = -EINVAL;
  1006. if ((offset + PAGE_ALIGN(len)) < offset)
  1007. goto out;
  1008. if (!(offset & ~PAGE_MASK))
  1009. ret = do_mmap_pgoff(file, addr, len, prot, flag, offset >> PAGE_SHIFT);
  1010. out:
  1011. return ret;
  1012. }
  1013. extern int do_munmap(struct mm_struct *, unsigned long, size_t);
  1014. extern unsigned long do_brk(unsigned long, unsigned long);
  1015. /* filemap.c */
  1016. extern unsigned long page_unuse(struct page *);
  1017. extern void truncate_inode_pages(struct address_space *, loff_t);
  1018. extern void truncate_inode_pages_range(struct address_space *,
  1019. loff_t lstart, loff_t lend);
  1020. /* generic vm_area_ops exported for stackable file systems */
  1021. extern int filemap_fault(struct vm_area_struct *, struct vm_fault *);
  1022. /* mm/page-writeback.c */
  1023. int write_one_page(struct page *page, int wait);
  1024. void task_dirty_inc(struct task_struct *tsk);
  1025. /* readahead.c */
  1026. #define VM_MAX_READAHEAD 128 /* kbytes */
  1027. #define VM_MIN_READAHEAD 16 /* kbytes (includes current page) */
  1028. int force_page_cache_readahead(struct address_space *mapping, struct file *filp,
  1029. pgoff_t offset, unsigned long nr_to_read);
  1030. void page_cache_sync_readahead(struct address_space *mapping,
  1031. struct file_ra_state *ra,
  1032. struct file *filp,
  1033. pgoff_t offset,
  1034. unsigned long size);
  1035. void page_cache_async_readahead(struct address_space *mapping,
  1036. struct file_ra_state *ra,
  1037. struct file *filp,
  1038. struct page *pg,
  1039. pgoff_t offset,
  1040. unsigned long size);
  1041. unsigned long max_sane_readahead(unsigned long nr);
  1042. unsigned long ra_submit(struct file_ra_state *ra,
  1043. struct address_space *mapping,
  1044. struct file *filp);
  1045. /* Do stack extension */
  1046. extern int expand_stack(struct vm_area_struct *vma, unsigned long address);
  1047. #ifdef CONFIG_IA64
  1048. extern int expand_upwards(struct vm_area_struct *vma, unsigned long address);
  1049. #endif
  1050. extern int expand_stack_downwards(struct vm_area_struct *vma,
  1051. unsigned long address);
  1052. /* Look up the first VMA which satisfies addr < vm_end, NULL if none. */
  1053. extern struct vm_area_struct * find_vma(struct mm_struct * mm, unsigned long addr);
  1054. extern struct vm_area_struct * find_vma_prev(struct mm_struct * mm, unsigned long addr,
  1055. struct vm_area_struct **pprev);
  1056. /* Look up the first VMA which intersects the interval start_addr..end_addr-1,
  1057. NULL if none. Assume start_addr < end_addr. */
  1058. static inline struct vm_area_struct * find_vma_intersection(struct mm_struct * mm, unsigned long start_addr, unsigned long end_addr)
  1059. {
  1060. struct vm_area_struct * vma = find_vma(mm,start_addr);
  1061. if (vma && end_addr <= vma->vm_start)
  1062. vma = NULL;
  1063. return vma;
  1064. }
  1065. static inline unsigned long vma_pages(struct vm_area_struct *vma)
  1066. {
  1067. return (vma->vm_end - vma->vm_start) >> PAGE_SHIFT;
  1068. }
  1069. pgprot_t vm_get_page_prot(unsigned long vm_flags);
  1070. struct vm_area_struct *find_extend_vma(struct mm_struct *, unsigned long addr);
  1071. int remap_pfn_range(struct vm_area_struct *, unsigned long addr,
  1072. unsigned long pfn, unsigned long size, pgprot_t);
  1073. int vm_insert_page(struct vm_area_struct *, unsigned long addr, struct page *);
  1074. int vm_insert_pfn(struct vm_area_struct *vma, unsigned long addr,
  1075. unsigned long pfn);
  1076. int vm_insert_mixed(struct vm_area_struct *vma, unsigned long addr,
  1077. unsigned long pfn);
  1078. struct page *follow_page(struct vm_area_struct *, unsigned long address,
  1079. unsigned int foll_flags);
  1080. #define FOLL_WRITE 0x01 /* check pte is writable */
  1081. #define FOLL_TOUCH 0x02 /* mark page accessed */
  1082. #define FOLL_GET 0x04 /* do get_page on page */
  1083. #define FOLL_DUMP 0x08 /* give error on hole if it would be zero */
  1084. #define FOLL_FORCE 0x10 /* get_user_pages read/write w/o permission */
  1085. typedef int (*pte_fn_t)(pte_t *pte, pgtable_t token, unsigned long addr,
  1086. void *data);
  1087. extern int apply_to_page_range(struct mm_struct *mm, unsigned long address,
  1088. unsigned long size, pte_fn_t fn, void *data);
  1089. #ifdef CONFIG_PROC_FS
  1090. void vm_stat_account(struct mm_struct *, unsigned long, struct file *, long);
  1091. #else
  1092. static inline void vm_stat_account(struct mm_struct *mm,
  1093. unsigned long flags, struct file *file, long pages)
  1094. {
  1095. }
  1096. #endif /* CONFIG_PROC_FS */
  1097. #ifdef CONFIG_DEBUG_PAGEALLOC
  1098. extern int debug_pagealloc_enabled;
  1099. extern void kernel_map_pages(struct page *page, int numpages, int enable);
  1100. static inline void enable_debug_pagealloc(void)
  1101. {
  1102. debug_pagealloc_enabled = 1;
  1103. }
  1104. #ifdef CONFIG_HIBERNATION
  1105. extern bool kernel_page_present(struct page *page);
  1106. #endif /* CONFIG_HIBERNATION */
  1107. #else
  1108. static inline void
  1109. kernel_map_pages(struct page *page, int numpages, int enable) {}
  1110. static inline void enable_debug_pagealloc(void)
  1111. {
  1112. }
  1113. #ifdef CONFIG_HIBERNATION
  1114. static inline bool kernel_page_present(struct page *page) { return true; }
  1115. #endif /* CONFIG_HIBERNATION */
  1116. #endif
  1117. extern struct vm_area_struct *get_gate_vma(struct task_struct *tsk);
  1118. #ifdef __HAVE_ARCH_GATE_AREA
  1119. int in_gate_area_no_task(unsigned long addr);
  1120. int in_gate_area(struct task_struct *task, unsigned long addr);
  1121. #else
  1122. int in_gate_area_no_task(unsigned long addr);
  1123. #define in_gate_area(task, addr) ({(void)task; in_gate_area_no_task(addr);})
  1124. #endif /* __HAVE_ARCH_GATE_AREA */
  1125. int drop_caches_sysctl_handler(struct ctl_table *, int,
  1126. void __user *, size_t *, loff_t *);
  1127. unsigned long shrink_slab(unsigned long scanned, gfp_t gfp_mask,
  1128. unsigned long lru_pages);
  1129. #ifndef CONFIG_MMU
  1130. #define randomize_va_space 0
  1131. #else
  1132. extern int randomize_va_space;
  1133. #endif
  1134. const char * arch_vma_name(struct vm_area_struct *vma);
  1135. void print_vma_addr(char *prefix, unsigned long rip);
  1136. struct page *sparse_mem_map_populate(unsigned long pnum, int nid);
  1137. pgd_t *vmemmap_pgd_populate(unsigned long addr, int node);
  1138. pud_t *vmemmap_pud_populate(pgd_t *pgd, unsigned long addr, int node);
  1139. pmd_t *vmemmap_pmd_populate(pud_t *pud, unsigned long addr, int node);
  1140. pte_t *vmemmap_pte_populate(pmd_t *pmd, unsigned long addr, int node);
  1141. void *vmemmap_alloc_block(unsigned long size, int node);
  1142. void vmemmap_verify(pte_t *, int, unsigned long, unsigned long);
  1143. int vmemmap_populate_basepages(struct page *start_page,
  1144. unsigned long pages, int node);
  1145. int vmemmap_populate(struct page *start_page, unsigned long pages, int node);
  1146. void vmemmap_populate_print_last(void);
  1147. extern int account_locked_memory(struct mm_struct *mm, struct rlimit *rlim,
  1148. size_t size);
  1149. extern void refund_locked_memory(struct mm_struct *mm, size_t size);
  1150. enum mf_flags {
  1151. MF_COUNT_INCREASED = 1 << 0,
  1152. };
  1153. extern void memory_failure(unsigned long pfn, int trapno);
  1154. extern int __memory_failure(unsigned long pfn, int trapno, int flags);
  1155. extern int unpoison_memory(unsigned long pfn);
  1156. extern int sysctl_memory_failure_early_kill;
  1157. extern int sysctl_memory_failure_recovery;
  1158. extern void shake_page(struct page *p, int access);
  1159. extern atomic_long_t mce_bad_pages;
  1160. extern int soft_offline_page(struct page *page, int flags);
  1161. #endif /* __KERNEL__ */
  1162. #endif /* _LINUX_MM_H */