mm.h 44 KB

12345678910111213141516171819202122232425262728293031323334353637383940414243444546474849505152535455565758596061626364656667686970717273747576777879808182838485868788899091929394959697989910010110210310410510610710810911011111211311411511611711811912012112212312412512612712812913013113213313413513613713813914014114214314414514614714814915015115215315415515615715815916016116216316416516616716816917017117217317417517617717817918018118218318418518618718818919019119219319419519619719819920020120220320420520620720820921021121221321421521621721821922022122222322422522622722822923023123223323423523623723823924024124224324424524624724824925025125225325425525625725825926026126226326426526626726826927027127227327427527627727827928028128228328428528628728828929029129229329429529629729829930030130230330430530630730830931031131231331431531631731831932032132232332432532632732832933033133233333433533633733833934034134234334434534634734834935035135235335435535635735835936036136236336436536636736836937037137237337437537637737837938038138238338438538638738838939039139239339439539639739839940040140240340440540640740840941041141241341441541641741841942042142242342442542642742842943043143243343443543643743843944044144244344444544644744844945045145245345445545645745845946046146246346446546646746846947047147247347447547647747847948048148248348448548648748848949049149249349449549649749849950050150250350450550650750850951051151251351451551651751851952052152252352452552652752852953053153253353453553653753853954054154254354454554654754854955055155255355455555655755855956056156256356456556656756856957057157257357457557657757857958058158258358458558658758858959059159259359459559659759859960060160260360460560660760860961061161261361461561661761861962062162262362462562662762862963063163263363463563663763863964064164264364464564664764864965065165265365465565665765865966066166266366466566666766866967067167267367467567667767867968068168268368468568668768868969069169269369469569669769869970070170270370470570670770870971071171271371471571671771871972072172272372472572672772872973073173273373473573673773873974074174274374474574674774874975075175275375475575675775875976076176276376476576676776876977077177277377477577677777877978078178278378478578678778878979079179279379479579679779879980080180280380480580680780880981081181281381481581681781881982082182282382482582682782882983083183283383483583683783883984084184284384484584684784884985085185285385485585685785885986086186286386486586686786886987087187287387487587687787887988088188288388488588688788888989089189289389489589689789889990090190290390490590690790890991091191291391491591691791891992092192292392492592692792892993093193293393493593693793893994094194294394494594694794894995095195295395495595695795895996096196296396496596696796896997097197297397497597697797897998098198298398498598698798898999099199299399499599699799899910001001100210031004100510061007100810091010101110121013101410151016101710181019102010211022102310241025102610271028102910301031103210331034103510361037103810391040104110421043104410451046104710481049105010511052105310541055105610571058105910601061106210631064106510661067106810691070107110721073107410751076107710781079108010811082108310841085108610871088108910901091109210931094109510961097109810991100110111021103110411051106110711081109111011111112111311141115111611171118111911201121112211231124112511261127112811291130113111321133113411351136113711381139114011411142114311441145114611471148114911501151115211531154115511561157115811591160116111621163116411651166116711681169117011711172117311741175117611771178117911801181118211831184118511861187118811891190119111921193119411951196119711981199120012011202120312041205120612071208120912101211121212131214121512161217121812191220122112221223122412251226122712281229123012311232123312341235123612371238123912401241124212431244124512461247124812491250125112521253125412551256125712581259126012611262126312641265126612671268126912701271127212731274127512761277127812791280128112821283128412851286128712881289129012911292
  1. #ifndef _LINUX_MM_H
  2. #define _LINUX_MM_H
  3. #include <linux/errno.h>
  4. #ifdef __KERNEL__
  5. #include <linux/gfp.h>
  6. #include <linux/list.h>
  7. #include <linux/mmdebug.h>
  8. #include <linux/mmzone.h>
  9. #include <linux/rbtree.h>
  10. #include <linux/prio_tree.h>
  11. #include <linux/debug_locks.h>
  12. #include <linux/mm_types.h>
  13. struct mempolicy;
  14. struct anon_vma;
  15. struct file_ra_state;
  16. struct user_struct;
  17. struct writeback_control;
  18. #ifndef CONFIG_DISCONTIGMEM /* Don't use mapnrs, do it properly */
  19. extern unsigned long max_mapnr;
  20. #endif
  21. extern unsigned long num_physpages;
  22. extern void * high_memory;
  23. extern int page_cluster;
  24. #ifdef CONFIG_SYSCTL
  25. extern int sysctl_legacy_va_layout;
  26. #else
  27. #define sysctl_legacy_va_layout 0
  28. #endif
  29. extern unsigned long mmap_min_addr;
  30. #include <asm/page.h>
  31. #include <asm/pgtable.h>
  32. #include <asm/processor.h>
  33. #define nth_page(page,n) pfn_to_page(page_to_pfn((page)) + (n))
  34. /* to align the pointer to the (next) page boundary */
  35. #define PAGE_ALIGN(addr) ALIGN(addr, PAGE_SIZE)
  36. /*
  37. * Linux kernel virtual memory manager primitives.
  38. * The idea being to have a "virtual" mm in the same way
  39. * we have a virtual fs - giving a cleaner interface to the
  40. * mm details, and allowing different kinds of memory mappings
  41. * (from shared memory to executable loading to arbitrary
  42. * mmap() functions).
  43. */
  44. extern struct kmem_cache *vm_area_cachep;
  45. /*
  46. * This struct defines the per-mm list of VMAs for uClinux. If CONFIG_MMU is
  47. * disabled, then there's a single shared list of VMAs maintained by the
  48. * system, and mm's subscribe to these individually
  49. */
  50. struct vm_list_struct {
  51. struct vm_list_struct *next;
  52. struct vm_area_struct *vma;
  53. };
  54. #ifndef CONFIG_MMU
  55. extern struct rb_root nommu_vma_tree;
  56. extern struct rw_semaphore nommu_vma_sem;
  57. extern unsigned int kobjsize(const void *objp);
  58. #endif
  59. /*
  60. * vm_flags in vm_area_struct, see mm_types.h.
  61. */
  62. #define VM_READ 0x00000001 /* currently active flags */
  63. #define VM_WRITE 0x00000002
  64. #define VM_EXEC 0x00000004
  65. #define VM_SHARED 0x00000008
  66. /* mprotect() hardcodes VM_MAYREAD >> 4 == VM_READ, and so for r/w/x bits. */
  67. #define VM_MAYREAD 0x00000010 /* limits for mprotect() etc */
  68. #define VM_MAYWRITE 0x00000020
  69. #define VM_MAYEXEC 0x00000040
  70. #define VM_MAYSHARE 0x00000080
  71. #define VM_GROWSDOWN 0x00000100 /* general info on the segment */
  72. #define VM_GROWSUP 0x00000200
  73. #define VM_PFNMAP 0x00000400 /* Page-ranges managed without "struct page", just pure PFN */
  74. #define VM_DENYWRITE 0x00000800 /* ETXTBSY on write attempts.. */
  75. #define VM_EXECUTABLE 0x00001000
  76. #define VM_LOCKED 0x00002000
  77. #define VM_IO 0x00004000 /* Memory mapped I/O or similar */
  78. /* Used by sys_madvise() */
  79. #define VM_SEQ_READ 0x00008000 /* App will access data sequentially */
  80. #define VM_RAND_READ 0x00010000 /* App will not benefit from clustered reads */
  81. #define VM_DONTCOPY 0x00020000 /* Do not copy this vma on fork */
  82. #define VM_DONTEXPAND 0x00040000 /* Cannot expand with mremap() */
  83. #define VM_RESERVED 0x00080000 /* Count as reserved_vm like IO */
  84. #define VM_ACCOUNT 0x00100000 /* Is a VM accounted object */
  85. #define VM_NORESERVE 0x00200000 /* should the VM suppress accounting */
  86. #define VM_HUGETLB 0x00400000 /* Huge TLB Page VM */
  87. #define VM_NONLINEAR 0x00800000 /* Is non-linear (remap_file_pages) */
  88. #define VM_MAPPED_COPY 0x01000000 /* T if mapped copy of data (nommu mmap) */
  89. #define VM_INSERTPAGE 0x02000000 /* The vma has had "vm_insert_page()" done on it */
  90. #define VM_ALWAYSDUMP 0x04000000 /* Always include in core dumps */
  91. #define VM_CAN_NONLINEAR 0x08000000 /* Has ->fault & does nonlinear pages */
  92. #define VM_MIXEDMAP 0x10000000 /* Can contain "struct page" and pure PFN pages */
  93. #define VM_SAO 0x20000000 /* Strong Access Ordering (powerpc) */
  94. #ifndef VM_STACK_DEFAULT_FLAGS /* arch can override this */
  95. #define VM_STACK_DEFAULT_FLAGS VM_DATA_DEFAULT_FLAGS
  96. #endif
  97. #ifdef CONFIG_STACK_GROWSUP
  98. #define VM_STACK_FLAGS (VM_GROWSUP | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT)
  99. #else
  100. #define VM_STACK_FLAGS (VM_GROWSDOWN | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT)
  101. #endif
  102. #define VM_READHINTMASK (VM_SEQ_READ | VM_RAND_READ)
  103. #define VM_ClearReadHint(v) (v)->vm_flags &= ~VM_READHINTMASK
  104. #define VM_NormalReadHint(v) (!((v)->vm_flags & VM_READHINTMASK))
  105. #define VM_SequentialReadHint(v) ((v)->vm_flags & VM_SEQ_READ)
  106. #define VM_RandomReadHint(v) ((v)->vm_flags & VM_RAND_READ)
  107. /*
  108. * special vmas that are non-mergable, non-mlock()able
  109. */
  110. #define VM_SPECIAL (VM_IO | VM_DONTEXPAND | VM_RESERVED | VM_PFNMAP)
  111. /*
  112. * mapping from the currently active vm_flags protection bits (the
  113. * low four bits) to a page protection mask..
  114. */
  115. extern pgprot_t protection_map[16];
  116. #define FAULT_FLAG_WRITE 0x01 /* Fault was a write access */
  117. #define FAULT_FLAG_NONLINEAR 0x02 /* Fault was via a nonlinear mapping */
  118. /*
  119. * vm_fault is filled by the the pagefault handler and passed to the vma's
  120. * ->fault function. The vma's ->fault is responsible for returning a bitmask
  121. * of VM_FAULT_xxx flags that give details about how the fault was handled.
  122. *
  123. * pgoff should be used in favour of virtual_address, if possible. If pgoff
  124. * is used, one may set VM_CAN_NONLINEAR in the vma->vm_flags to get nonlinear
  125. * mapping support.
  126. */
  127. struct vm_fault {
  128. unsigned int flags; /* FAULT_FLAG_xxx flags */
  129. pgoff_t pgoff; /* Logical page offset based on vma */
  130. void __user *virtual_address; /* Faulting virtual address */
  131. struct page *page; /* ->fault handlers should return a
  132. * page here, unless VM_FAULT_NOPAGE
  133. * is set (which is also implied by
  134. * VM_FAULT_ERROR).
  135. */
  136. };
  137. /*
  138. * These are the virtual MM functions - opening of an area, closing and
  139. * unmapping it (needed to keep files on disk up-to-date etc), pointer
  140. * to the functions called when a no-page or a wp-page exception occurs.
  141. */
  142. struct vm_operations_struct {
  143. void (*open)(struct vm_area_struct * area);
  144. void (*close)(struct vm_area_struct * area);
  145. int (*fault)(struct vm_area_struct *vma, struct vm_fault *vmf);
  146. /* notification that a previously read-only page is about to become
  147. * writable, if an error is returned it will cause a SIGBUS */
  148. int (*page_mkwrite)(struct vm_area_struct *vma, struct page *page);
  149. /* called by access_process_vm when get_user_pages() fails, typically
  150. * for use by special VMAs that can switch between memory and hardware
  151. */
  152. int (*access)(struct vm_area_struct *vma, unsigned long addr,
  153. void *buf, int len, int write);
  154. #ifdef CONFIG_NUMA
  155. /*
  156. * set_policy() op must add a reference to any non-NULL @new mempolicy
  157. * to hold the policy upon return. Caller should pass NULL @new to
  158. * remove a policy and fall back to surrounding context--i.e. do not
  159. * install a MPOL_DEFAULT policy, nor the task or system default
  160. * mempolicy.
  161. */
  162. int (*set_policy)(struct vm_area_struct *vma, struct mempolicy *new);
  163. /*
  164. * get_policy() op must add reference [mpol_get()] to any policy at
  165. * (vma,addr) marked as MPOL_SHARED. The shared policy infrastructure
  166. * in mm/mempolicy.c will do this automatically.
  167. * get_policy() must NOT add a ref if the policy at (vma,addr) is not
  168. * marked as MPOL_SHARED. vma policies are protected by the mmap_sem.
  169. * If no [shared/vma] mempolicy exists at the addr, get_policy() op
  170. * must return NULL--i.e., do not "fallback" to task or system default
  171. * policy.
  172. */
  173. struct mempolicy *(*get_policy)(struct vm_area_struct *vma,
  174. unsigned long addr);
  175. int (*migrate)(struct vm_area_struct *vma, const nodemask_t *from,
  176. const nodemask_t *to, unsigned long flags);
  177. #endif
  178. };
  179. struct mmu_gather;
  180. struct inode;
  181. #define page_private(page) ((page)->private)
  182. #define set_page_private(page, v) ((page)->private = (v))
  183. /*
  184. * FIXME: take this include out, include page-flags.h in
  185. * files which need it (119 of them)
  186. */
  187. #include <linux/page-flags.h>
  188. /*
  189. * Methods to modify the page usage count.
  190. *
  191. * What counts for a page usage:
  192. * - cache mapping (page->mapping)
  193. * - private data (page->private)
  194. * - page mapped in a task's page tables, each mapping
  195. * is counted separately
  196. *
  197. * Also, many kernel routines increase the page count before a critical
  198. * routine so they can be sure the page doesn't go away from under them.
  199. */
  200. /*
  201. * Drop a ref, return true if the refcount fell to zero (the page has no users)
  202. */
  203. static inline int put_page_testzero(struct page *page)
  204. {
  205. VM_BUG_ON(atomic_read(&page->_count) == 0);
  206. return atomic_dec_and_test(&page->_count);
  207. }
  208. /*
  209. * Try to grab a ref unless the page has a refcount of zero, return false if
  210. * that is the case.
  211. */
  212. static inline int get_page_unless_zero(struct page *page)
  213. {
  214. VM_BUG_ON(PageTail(page));
  215. return atomic_inc_not_zero(&page->_count);
  216. }
  217. /* Support for virtually mapped pages */
  218. struct page *vmalloc_to_page(const void *addr);
  219. unsigned long vmalloc_to_pfn(const void *addr);
  220. /*
  221. * Determine if an address is within the vmalloc range
  222. *
  223. * On nommu, vmalloc/vfree wrap through kmalloc/kfree directly, so there
  224. * is no special casing required.
  225. */
  226. static inline int is_vmalloc_addr(const void *x)
  227. {
  228. #ifdef CONFIG_MMU
  229. unsigned long addr = (unsigned long)x;
  230. return addr >= VMALLOC_START && addr < VMALLOC_END;
  231. #else
  232. return 0;
  233. #endif
  234. }
  235. static inline struct page *compound_head(struct page *page)
  236. {
  237. if (unlikely(PageTail(page)))
  238. return page->first_page;
  239. return page;
  240. }
  241. static inline int page_count(struct page *page)
  242. {
  243. return atomic_read(&compound_head(page)->_count);
  244. }
  245. static inline void get_page(struct page *page)
  246. {
  247. page = compound_head(page);
  248. VM_BUG_ON(atomic_read(&page->_count) == 0);
  249. atomic_inc(&page->_count);
  250. }
  251. static inline struct page *virt_to_head_page(const void *x)
  252. {
  253. struct page *page = virt_to_page(x);
  254. return compound_head(page);
  255. }
  256. /*
  257. * Setup the page count before being freed into the page allocator for
  258. * the first time (boot or memory hotplug)
  259. */
  260. static inline void init_page_count(struct page *page)
  261. {
  262. atomic_set(&page->_count, 1);
  263. }
  264. void put_page(struct page *page);
  265. void put_pages_list(struct list_head *pages);
  266. void split_page(struct page *page, unsigned int order);
  267. /*
  268. * Compound pages have a destructor function. Provide a
  269. * prototype for that function and accessor functions.
  270. * These are _only_ valid on the head of a PG_compound page.
  271. */
  272. typedef void compound_page_dtor(struct page *);
  273. static inline void set_compound_page_dtor(struct page *page,
  274. compound_page_dtor *dtor)
  275. {
  276. page[1].lru.next = (void *)dtor;
  277. }
  278. static inline compound_page_dtor *get_compound_page_dtor(struct page *page)
  279. {
  280. return (compound_page_dtor *)page[1].lru.next;
  281. }
  282. static inline int compound_order(struct page *page)
  283. {
  284. if (!PageHead(page))
  285. return 0;
  286. return (unsigned long)page[1].lru.prev;
  287. }
  288. static inline void set_compound_order(struct page *page, unsigned long order)
  289. {
  290. page[1].lru.prev = (void *)order;
  291. }
  292. /*
  293. * Multiple processes may "see" the same page. E.g. for untouched
  294. * mappings of /dev/null, all processes see the same page full of
  295. * zeroes, and text pages of executables and shared libraries have
  296. * only one copy in memory, at most, normally.
  297. *
  298. * For the non-reserved pages, page_count(page) denotes a reference count.
  299. * page_count() == 0 means the page is free. page->lru is then used for
  300. * freelist management in the buddy allocator.
  301. * page_count() > 0 means the page has been allocated.
  302. *
  303. * Pages are allocated by the slab allocator in order to provide memory
  304. * to kmalloc and kmem_cache_alloc. In this case, the management of the
  305. * page, and the fields in 'struct page' are the responsibility of mm/slab.c
  306. * unless a particular usage is carefully commented. (the responsibility of
  307. * freeing the kmalloc memory is the caller's, of course).
  308. *
  309. * A page may be used by anyone else who does a __get_free_page().
  310. * In this case, page_count still tracks the references, and should only
  311. * be used through the normal accessor functions. The top bits of page->flags
  312. * and page->virtual store page management information, but all other fields
  313. * are unused and could be used privately, carefully. The management of this
  314. * page is the responsibility of the one who allocated it, and those who have
  315. * subsequently been given references to it.
  316. *
  317. * The other pages (we may call them "pagecache pages") are completely
  318. * managed by the Linux memory manager: I/O, buffers, swapping etc.
  319. * The following discussion applies only to them.
  320. *
  321. * A pagecache page contains an opaque `private' member, which belongs to the
  322. * page's address_space. Usually, this is the address of a circular list of
  323. * the page's disk buffers. PG_private must be set to tell the VM to call
  324. * into the filesystem to release these pages.
  325. *
  326. * A page may belong to an inode's memory mapping. In this case, page->mapping
  327. * is the pointer to the inode, and page->index is the file offset of the page,
  328. * in units of PAGE_CACHE_SIZE.
  329. *
  330. * If pagecache pages are not associated with an inode, they are said to be
  331. * anonymous pages. These may become associated with the swapcache, and in that
  332. * case PG_swapcache is set, and page->private is an offset into the swapcache.
  333. *
  334. * In either case (swapcache or inode backed), the pagecache itself holds one
  335. * reference to the page. Setting PG_private should also increment the
  336. * refcount. The each user mapping also has a reference to the page.
  337. *
  338. * The pagecache pages are stored in a per-mapping radix tree, which is
  339. * rooted at mapping->page_tree, and indexed by offset.
  340. * Where 2.4 and early 2.6 kernels kept dirty/clean pages in per-address_space
  341. * lists, we instead now tag pages as dirty/writeback in the radix tree.
  342. *
  343. * All pagecache pages may be subject to I/O:
  344. * - inode pages may need to be read from disk,
  345. * - inode pages which have been modified and are MAP_SHARED may need
  346. * to be written back to the inode on disk,
  347. * - anonymous pages (including MAP_PRIVATE file mappings) which have been
  348. * modified may need to be swapped out to swap space and (later) to be read
  349. * back into memory.
  350. */
  351. /*
  352. * The zone field is never updated after free_area_init_core()
  353. * sets it, so none of the operations on it need to be atomic.
  354. */
  355. /*
  356. * page->flags layout:
  357. *
  358. * There are three possibilities for how page->flags get
  359. * laid out. The first is for the normal case, without
  360. * sparsemem. The second is for sparsemem when there is
  361. * plenty of space for node and section. The last is when
  362. * we have run out of space and have to fall back to an
  363. * alternate (slower) way of determining the node.
  364. *
  365. * No sparsemem or sparsemem vmemmap: | NODE | ZONE | ... | FLAGS |
  366. * classic sparse with space for node:| SECTION | NODE | ZONE | ... | FLAGS |
  367. * classic sparse no space for node: | SECTION | ZONE | ... | FLAGS |
  368. */
  369. #if defined(CONFIG_SPARSEMEM) && !defined(CONFIG_SPARSEMEM_VMEMMAP)
  370. #define SECTIONS_WIDTH SECTIONS_SHIFT
  371. #else
  372. #define SECTIONS_WIDTH 0
  373. #endif
  374. #define ZONES_WIDTH ZONES_SHIFT
  375. #if SECTIONS_WIDTH+ZONES_WIDTH+NODES_SHIFT <= BITS_PER_LONG - NR_PAGEFLAGS
  376. #define NODES_WIDTH NODES_SHIFT
  377. #else
  378. #ifdef CONFIG_SPARSEMEM_VMEMMAP
  379. #error "Vmemmap: No space for nodes field in page flags"
  380. #endif
  381. #define NODES_WIDTH 0
  382. #endif
  383. /* Page flags: | [SECTION] | [NODE] | ZONE | ... | FLAGS | */
  384. #define SECTIONS_PGOFF ((sizeof(unsigned long)*8) - SECTIONS_WIDTH)
  385. #define NODES_PGOFF (SECTIONS_PGOFF - NODES_WIDTH)
  386. #define ZONES_PGOFF (NODES_PGOFF - ZONES_WIDTH)
  387. /*
  388. * We are going to use the flags for the page to node mapping if its in
  389. * there. This includes the case where there is no node, so it is implicit.
  390. */
  391. #if !(NODES_WIDTH > 0 || NODES_SHIFT == 0)
  392. #define NODE_NOT_IN_PAGE_FLAGS
  393. #endif
  394. #ifndef PFN_SECTION_SHIFT
  395. #define PFN_SECTION_SHIFT 0
  396. #endif
  397. /*
  398. * Define the bit shifts to access each section. For non-existant
  399. * sections we define the shift as 0; that plus a 0 mask ensures
  400. * the compiler will optimise away reference to them.
  401. */
  402. #define SECTIONS_PGSHIFT (SECTIONS_PGOFF * (SECTIONS_WIDTH != 0))
  403. #define NODES_PGSHIFT (NODES_PGOFF * (NODES_WIDTH != 0))
  404. #define ZONES_PGSHIFT (ZONES_PGOFF * (ZONES_WIDTH != 0))
  405. /* NODE:ZONE or SECTION:ZONE is used to ID a zone for the buddy allcator */
  406. #ifdef NODE_NOT_IN_PAGEFLAGS
  407. #define ZONEID_SHIFT (SECTIONS_SHIFT + ZONES_SHIFT)
  408. #define ZONEID_PGOFF ((SECTIONS_PGOFF < ZONES_PGOFF)? \
  409. SECTIONS_PGOFF : ZONES_PGOFF)
  410. #else
  411. #define ZONEID_SHIFT (NODES_SHIFT + ZONES_SHIFT)
  412. #define ZONEID_PGOFF ((NODES_PGOFF < ZONES_PGOFF)? \
  413. NODES_PGOFF : ZONES_PGOFF)
  414. #endif
  415. #define ZONEID_PGSHIFT (ZONEID_PGOFF * (ZONEID_SHIFT != 0))
  416. #if SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > BITS_PER_LONG - NR_PAGEFLAGS
  417. #error SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > BITS_PER_LONG - NR_PAGEFLAGS
  418. #endif
  419. #define ZONES_MASK ((1UL << ZONES_WIDTH) - 1)
  420. #define NODES_MASK ((1UL << NODES_WIDTH) - 1)
  421. #define SECTIONS_MASK ((1UL << SECTIONS_WIDTH) - 1)
  422. #define ZONEID_MASK ((1UL << ZONEID_SHIFT) - 1)
  423. static inline enum zone_type page_zonenum(struct page *page)
  424. {
  425. return (page->flags >> ZONES_PGSHIFT) & ZONES_MASK;
  426. }
  427. /*
  428. * The identification function is only used by the buddy allocator for
  429. * determining if two pages could be buddies. We are not really
  430. * identifying a zone since we could be using a the section number
  431. * id if we have not node id available in page flags.
  432. * We guarantee only that it will return the same value for two
  433. * combinable pages in a zone.
  434. */
  435. static inline int page_zone_id(struct page *page)
  436. {
  437. return (page->flags >> ZONEID_PGSHIFT) & ZONEID_MASK;
  438. }
  439. static inline int zone_to_nid(struct zone *zone)
  440. {
  441. #ifdef CONFIG_NUMA
  442. return zone->node;
  443. #else
  444. return 0;
  445. #endif
  446. }
  447. #ifdef NODE_NOT_IN_PAGE_FLAGS
  448. extern int page_to_nid(struct page *page);
  449. #else
  450. static inline int page_to_nid(struct page *page)
  451. {
  452. return (page->flags >> NODES_PGSHIFT) & NODES_MASK;
  453. }
  454. #endif
  455. static inline struct zone *page_zone(struct page *page)
  456. {
  457. return &NODE_DATA(page_to_nid(page))->node_zones[page_zonenum(page)];
  458. }
  459. #if defined(CONFIG_SPARSEMEM) && !defined(CONFIG_SPARSEMEM_VMEMMAP)
  460. static inline unsigned long page_to_section(struct page *page)
  461. {
  462. return (page->flags >> SECTIONS_PGSHIFT) & SECTIONS_MASK;
  463. }
  464. #endif
  465. static inline void set_page_zone(struct page *page, enum zone_type zone)
  466. {
  467. page->flags &= ~(ZONES_MASK << ZONES_PGSHIFT);
  468. page->flags |= (zone & ZONES_MASK) << ZONES_PGSHIFT;
  469. }
  470. static inline void set_page_node(struct page *page, unsigned long node)
  471. {
  472. page->flags &= ~(NODES_MASK << NODES_PGSHIFT);
  473. page->flags |= (node & NODES_MASK) << NODES_PGSHIFT;
  474. }
  475. static inline void set_page_section(struct page *page, unsigned long section)
  476. {
  477. page->flags &= ~(SECTIONS_MASK << SECTIONS_PGSHIFT);
  478. page->flags |= (section & SECTIONS_MASK) << SECTIONS_PGSHIFT;
  479. }
  480. static inline void set_page_links(struct page *page, enum zone_type zone,
  481. unsigned long node, unsigned long pfn)
  482. {
  483. set_page_zone(page, zone);
  484. set_page_node(page, node);
  485. set_page_section(page, pfn_to_section_nr(pfn));
  486. }
  487. /*
  488. * If a hint addr is less than mmap_min_addr change hint to be as
  489. * low as possible but still greater than mmap_min_addr
  490. */
  491. static inline unsigned long round_hint_to_min(unsigned long hint)
  492. {
  493. #ifdef CONFIG_SECURITY
  494. hint &= PAGE_MASK;
  495. if (((void *)hint != NULL) &&
  496. (hint < mmap_min_addr))
  497. return PAGE_ALIGN(mmap_min_addr);
  498. #endif
  499. return hint;
  500. }
  501. /*
  502. * Some inline functions in vmstat.h depend on page_zone()
  503. */
  504. #include <linux/vmstat.h>
  505. static __always_inline void *lowmem_page_address(struct page *page)
  506. {
  507. return __va(page_to_pfn(page) << PAGE_SHIFT);
  508. }
  509. #if defined(CONFIG_HIGHMEM) && !defined(WANT_PAGE_VIRTUAL)
  510. #define HASHED_PAGE_VIRTUAL
  511. #endif
  512. #if defined(WANT_PAGE_VIRTUAL)
  513. #define page_address(page) ((page)->virtual)
  514. #define set_page_address(page, address) \
  515. do { \
  516. (page)->virtual = (address); \
  517. } while(0)
  518. #define page_address_init() do { } while(0)
  519. #endif
  520. #if defined(HASHED_PAGE_VIRTUAL)
  521. void *page_address(struct page *page);
  522. void set_page_address(struct page *page, void *virtual);
  523. void page_address_init(void);
  524. #endif
  525. #if !defined(HASHED_PAGE_VIRTUAL) && !defined(WANT_PAGE_VIRTUAL)
  526. #define page_address(page) lowmem_page_address(page)
  527. #define set_page_address(page, address) do { } while(0)
  528. #define page_address_init() do { } while(0)
  529. #endif
  530. /*
  531. * On an anonymous page mapped into a user virtual memory area,
  532. * page->mapping points to its anon_vma, not to a struct address_space;
  533. * with the PAGE_MAPPING_ANON bit set to distinguish it.
  534. *
  535. * Please note that, confusingly, "page_mapping" refers to the inode
  536. * address_space which maps the page from disk; whereas "page_mapped"
  537. * refers to user virtual address space into which the page is mapped.
  538. */
  539. #define PAGE_MAPPING_ANON 1
  540. extern struct address_space swapper_space;
  541. static inline struct address_space *page_mapping(struct page *page)
  542. {
  543. struct address_space *mapping = page->mapping;
  544. VM_BUG_ON(PageSlab(page));
  545. #ifdef CONFIG_SWAP
  546. if (unlikely(PageSwapCache(page)))
  547. mapping = &swapper_space;
  548. else
  549. #endif
  550. if (unlikely((unsigned long)mapping & PAGE_MAPPING_ANON))
  551. mapping = NULL;
  552. return mapping;
  553. }
  554. static inline int PageAnon(struct page *page)
  555. {
  556. return ((unsigned long)page->mapping & PAGE_MAPPING_ANON) != 0;
  557. }
  558. /*
  559. * Return the pagecache index of the passed page. Regular pagecache pages
  560. * use ->index whereas swapcache pages use ->private
  561. */
  562. static inline pgoff_t page_index(struct page *page)
  563. {
  564. if (unlikely(PageSwapCache(page)))
  565. return page_private(page);
  566. return page->index;
  567. }
  568. /*
  569. * The atomic page->_mapcount, like _count, starts from -1:
  570. * so that transitions both from it and to it can be tracked,
  571. * using atomic_inc_and_test and atomic_add_negative(-1).
  572. */
  573. static inline void reset_page_mapcount(struct page *page)
  574. {
  575. atomic_set(&(page)->_mapcount, -1);
  576. }
  577. static inline int page_mapcount(struct page *page)
  578. {
  579. return atomic_read(&(page)->_mapcount) + 1;
  580. }
  581. /*
  582. * Return true if this page is mapped into pagetables.
  583. */
  584. static inline int page_mapped(struct page *page)
  585. {
  586. return atomic_read(&(page)->_mapcount) >= 0;
  587. }
  588. /*
  589. * Different kinds of faults, as returned by handle_mm_fault().
  590. * Used to decide whether a process gets delivered SIGBUS or
  591. * just gets major/minor fault counters bumped up.
  592. */
  593. #define VM_FAULT_MINOR 0 /* For backwards compat. Remove me quickly. */
  594. #define VM_FAULT_OOM 0x0001
  595. #define VM_FAULT_SIGBUS 0x0002
  596. #define VM_FAULT_MAJOR 0x0004
  597. #define VM_FAULT_WRITE 0x0008 /* Special case for get_user_pages */
  598. #define VM_FAULT_NOPAGE 0x0100 /* ->fault installed the pte, not return page */
  599. #define VM_FAULT_LOCKED 0x0200 /* ->fault locked the returned page */
  600. #define VM_FAULT_ERROR (VM_FAULT_OOM | VM_FAULT_SIGBUS)
  601. #define offset_in_page(p) ((unsigned long)(p) & ~PAGE_MASK)
  602. extern void show_free_areas(void);
  603. #ifdef CONFIG_SHMEM
  604. extern int shmem_lock(struct file *file, int lock, struct user_struct *user);
  605. #else
  606. static inline int shmem_lock(struct file *file, int lock,
  607. struct user_struct *user)
  608. {
  609. return 0;
  610. }
  611. #endif
  612. struct file *shmem_file_setup(char *name, loff_t size, unsigned long flags);
  613. int shmem_zero_setup(struct vm_area_struct *);
  614. #ifndef CONFIG_MMU
  615. extern unsigned long shmem_get_unmapped_area(struct file *file,
  616. unsigned long addr,
  617. unsigned long len,
  618. unsigned long pgoff,
  619. unsigned long flags);
  620. #endif
  621. extern int can_do_mlock(void);
  622. extern int user_shm_lock(size_t, struct user_struct *);
  623. extern void user_shm_unlock(size_t, struct user_struct *);
  624. /*
  625. * Parameter block passed down to zap_pte_range in exceptional cases.
  626. */
  627. struct zap_details {
  628. struct vm_area_struct *nonlinear_vma; /* Check page->index if set */
  629. struct address_space *check_mapping; /* Check page->mapping if set */
  630. pgoff_t first_index; /* Lowest page->index to unmap */
  631. pgoff_t last_index; /* Highest page->index to unmap */
  632. spinlock_t *i_mmap_lock; /* For unmap_mapping_range: */
  633. unsigned long truncate_count; /* Compare vm_truncate_count */
  634. };
  635. struct page *vm_normal_page(struct vm_area_struct *vma, unsigned long addr,
  636. pte_t pte);
  637. int zap_vma_ptes(struct vm_area_struct *vma, unsigned long address,
  638. unsigned long size);
  639. unsigned long zap_page_range(struct vm_area_struct *vma, unsigned long address,
  640. unsigned long size, struct zap_details *);
  641. unsigned long unmap_vmas(struct mmu_gather **tlb,
  642. struct vm_area_struct *start_vma, unsigned long start_addr,
  643. unsigned long end_addr, unsigned long *nr_accounted,
  644. struct zap_details *);
  645. /**
  646. * mm_walk - callbacks for walk_page_range
  647. * @pgd_entry: if set, called for each non-empty PGD (top-level) entry
  648. * @pud_entry: if set, called for each non-empty PUD (2nd-level) entry
  649. * @pmd_entry: if set, called for each non-empty PMD (3rd-level) entry
  650. * @pte_entry: if set, called for each non-empty PTE (4th-level) entry
  651. * @pte_hole: if set, called for each hole at all levels
  652. *
  653. * (see walk_page_range for more details)
  654. */
  655. struct mm_walk {
  656. int (*pgd_entry)(pgd_t *, unsigned long, unsigned long, struct mm_walk *);
  657. int (*pud_entry)(pud_t *, unsigned long, unsigned long, struct mm_walk *);
  658. int (*pmd_entry)(pmd_t *, unsigned long, unsigned long, struct mm_walk *);
  659. int (*pte_entry)(pte_t *, unsigned long, unsigned long, struct mm_walk *);
  660. int (*pte_hole)(unsigned long, unsigned long, struct mm_walk *);
  661. struct mm_struct *mm;
  662. void *private;
  663. };
  664. int walk_page_range(unsigned long addr, unsigned long end,
  665. struct mm_walk *walk);
  666. void free_pgd_range(struct mmu_gather *tlb, unsigned long addr,
  667. unsigned long end, unsigned long floor, unsigned long ceiling);
  668. int copy_page_range(struct mm_struct *dst, struct mm_struct *src,
  669. struct vm_area_struct *vma);
  670. void unmap_mapping_range(struct address_space *mapping,
  671. loff_t const holebegin, loff_t const holelen, int even_cows);
  672. int generic_access_phys(struct vm_area_struct *vma, unsigned long addr,
  673. void *buf, int len, int write);
  674. static inline void unmap_shared_mapping_range(struct address_space *mapping,
  675. loff_t const holebegin, loff_t const holelen)
  676. {
  677. unmap_mapping_range(mapping, holebegin, holelen, 0);
  678. }
  679. extern int vmtruncate(struct inode * inode, loff_t offset);
  680. extern int vmtruncate_range(struct inode * inode, loff_t offset, loff_t end);
  681. #ifdef CONFIG_MMU
  682. extern int handle_mm_fault(struct mm_struct *mm, struct vm_area_struct *vma,
  683. unsigned long address, int write_access);
  684. #else
  685. static inline int handle_mm_fault(struct mm_struct *mm,
  686. struct vm_area_struct *vma, unsigned long address,
  687. int write_access)
  688. {
  689. /* should never happen if there's no MMU */
  690. BUG();
  691. return VM_FAULT_SIGBUS;
  692. }
  693. #endif
  694. extern int make_pages_present(unsigned long addr, unsigned long end);
  695. extern int access_process_vm(struct task_struct *tsk, unsigned long addr, void *buf, int len, int write);
  696. int get_user_pages(struct task_struct *tsk, struct mm_struct *mm, unsigned long start,
  697. int len, int write, int force, struct page **pages, struct vm_area_struct **vmas);
  698. extern int try_to_release_page(struct page * page, gfp_t gfp_mask);
  699. extern void do_invalidatepage(struct page *page, unsigned long offset);
  700. int __set_page_dirty_nobuffers(struct page *page);
  701. int __set_page_dirty_no_writeback(struct page *page);
  702. int redirty_page_for_writepage(struct writeback_control *wbc,
  703. struct page *page);
  704. int set_page_dirty(struct page *page);
  705. int set_page_dirty_lock(struct page *page);
  706. int clear_page_dirty_for_io(struct page *page);
  707. extern unsigned long move_page_tables(struct vm_area_struct *vma,
  708. unsigned long old_addr, struct vm_area_struct *new_vma,
  709. unsigned long new_addr, unsigned long len);
  710. extern unsigned long do_mremap(unsigned long addr,
  711. unsigned long old_len, unsigned long new_len,
  712. unsigned long flags, unsigned long new_addr);
  713. extern int mprotect_fixup(struct vm_area_struct *vma,
  714. struct vm_area_struct **pprev, unsigned long start,
  715. unsigned long end, unsigned long newflags);
  716. /*
  717. * get_user_pages_fast provides equivalent functionality to get_user_pages,
  718. * operating on current and current->mm (force=0 and doesn't return any vmas).
  719. *
  720. * get_user_pages_fast may take mmap_sem and page tables, so no assumptions
  721. * can be made about locking. get_user_pages_fast is to be implemented in a
  722. * way that is advantageous (vs get_user_pages()) when the user memory area is
  723. * already faulted in and present in ptes. However if the pages have to be
  724. * faulted in, it may turn out to be slightly slower).
  725. */
  726. int get_user_pages_fast(unsigned long start, int nr_pages, int write,
  727. struct page **pages);
  728. /*
  729. * A callback you can register to apply pressure to ageable caches.
  730. *
  731. * 'shrink' is passed a count 'nr_to_scan' and a 'gfpmask'. It should
  732. * look through the least-recently-used 'nr_to_scan' entries and
  733. * attempt to free them up. It should return the number of objects
  734. * which remain in the cache. If it returns -1, it means it cannot do
  735. * any scanning at this time (eg. there is a risk of deadlock).
  736. *
  737. * The 'gfpmask' refers to the allocation we are currently trying to
  738. * fulfil.
  739. *
  740. * Note that 'shrink' will be passed nr_to_scan == 0 when the VM is
  741. * querying the cache size, so a fastpath for that case is appropriate.
  742. */
  743. struct shrinker {
  744. int (*shrink)(int nr_to_scan, gfp_t gfp_mask);
  745. int seeks; /* seeks to recreate an obj */
  746. /* These are for internal use */
  747. struct list_head list;
  748. long nr; /* objs pending delete */
  749. };
  750. #define DEFAULT_SEEKS 2 /* A good number if you don't know better. */
  751. extern void register_shrinker(struct shrinker *);
  752. extern void unregister_shrinker(struct shrinker *);
  753. int vma_wants_writenotify(struct vm_area_struct *vma);
  754. extern pte_t *get_locked_pte(struct mm_struct *mm, unsigned long addr, spinlock_t **ptl);
  755. #ifdef __PAGETABLE_PUD_FOLDED
  756. static inline int __pud_alloc(struct mm_struct *mm, pgd_t *pgd,
  757. unsigned long address)
  758. {
  759. return 0;
  760. }
  761. #else
  762. int __pud_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address);
  763. #endif
  764. #ifdef __PAGETABLE_PMD_FOLDED
  765. static inline int __pmd_alloc(struct mm_struct *mm, pud_t *pud,
  766. unsigned long address)
  767. {
  768. return 0;
  769. }
  770. #else
  771. int __pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address);
  772. #endif
  773. int __pte_alloc(struct mm_struct *mm, pmd_t *pmd, unsigned long address);
  774. int __pte_alloc_kernel(pmd_t *pmd, unsigned long address);
  775. /*
  776. * The following ifdef needed to get the 4level-fixup.h header to work.
  777. * Remove it when 4level-fixup.h has been removed.
  778. */
  779. #if defined(CONFIG_MMU) && !defined(__ARCH_HAS_4LEVEL_HACK)
  780. static inline pud_t *pud_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address)
  781. {
  782. return (unlikely(pgd_none(*pgd)) && __pud_alloc(mm, pgd, address))?
  783. NULL: pud_offset(pgd, address);
  784. }
  785. static inline pmd_t *pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address)
  786. {
  787. return (unlikely(pud_none(*pud)) && __pmd_alloc(mm, pud, address))?
  788. NULL: pmd_offset(pud, address);
  789. }
  790. #endif /* CONFIG_MMU && !__ARCH_HAS_4LEVEL_HACK */
  791. #if USE_SPLIT_PTLOCKS
  792. /*
  793. * We tuck a spinlock to guard each pagetable page into its struct page,
  794. * at page->private, with BUILD_BUG_ON to make sure that this will not
  795. * overflow into the next struct page (as it might with DEBUG_SPINLOCK).
  796. * When freeing, reset page->mapping so free_pages_check won't complain.
  797. */
  798. #define __pte_lockptr(page) &((page)->ptl)
  799. #define pte_lock_init(_page) do { \
  800. spin_lock_init(__pte_lockptr(_page)); \
  801. } while (0)
  802. #define pte_lock_deinit(page) ((page)->mapping = NULL)
  803. #define pte_lockptr(mm, pmd) ({(void)(mm); __pte_lockptr(pmd_page(*(pmd)));})
  804. #else /* !USE_SPLIT_PTLOCKS */
  805. /*
  806. * We use mm->page_table_lock to guard all pagetable pages of the mm.
  807. */
  808. #define pte_lock_init(page) do {} while (0)
  809. #define pte_lock_deinit(page) do {} while (0)
  810. #define pte_lockptr(mm, pmd) ({(void)(pmd); &(mm)->page_table_lock;})
  811. #endif /* USE_SPLIT_PTLOCKS */
  812. static inline void pgtable_page_ctor(struct page *page)
  813. {
  814. pte_lock_init(page);
  815. inc_zone_page_state(page, NR_PAGETABLE);
  816. }
  817. static inline void pgtable_page_dtor(struct page *page)
  818. {
  819. pte_lock_deinit(page);
  820. dec_zone_page_state(page, NR_PAGETABLE);
  821. }
  822. #define pte_offset_map_lock(mm, pmd, address, ptlp) \
  823. ({ \
  824. spinlock_t *__ptl = pte_lockptr(mm, pmd); \
  825. pte_t *__pte = pte_offset_map(pmd, address); \
  826. *(ptlp) = __ptl; \
  827. spin_lock(__ptl); \
  828. __pte; \
  829. })
  830. #define pte_unmap_unlock(pte, ptl) do { \
  831. spin_unlock(ptl); \
  832. pte_unmap(pte); \
  833. } while (0)
  834. #define pte_alloc_map(mm, pmd, address) \
  835. ((unlikely(!pmd_present(*(pmd))) && __pte_alloc(mm, pmd, address))? \
  836. NULL: pte_offset_map(pmd, address))
  837. #define pte_alloc_map_lock(mm, pmd, address, ptlp) \
  838. ((unlikely(!pmd_present(*(pmd))) && __pte_alloc(mm, pmd, address))? \
  839. NULL: pte_offset_map_lock(mm, pmd, address, ptlp))
  840. #define pte_alloc_kernel(pmd, address) \
  841. ((unlikely(!pmd_present(*(pmd))) && __pte_alloc_kernel(pmd, address))? \
  842. NULL: pte_offset_kernel(pmd, address))
  843. extern void free_area_init(unsigned long * zones_size);
  844. extern void free_area_init_node(int nid, unsigned long * zones_size,
  845. unsigned long zone_start_pfn, unsigned long *zholes_size);
  846. #ifdef CONFIG_ARCH_POPULATES_NODE_MAP
  847. /*
  848. * With CONFIG_ARCH_POPULATES_NODE_MAP set, an architecture may initialise its
  849. * zones, allocate the backing mem_map and account for memory holes in a more
  850. * architecture independent manner. This is a substitute for creating the
  851. * zone_sizes[] and zholes_size[] arrays and passing them to
  852. * free_area_init_node()
  853. *
  854. * An architecture is expected to register range of page frames backed by
  855. * physical memory with add_active_range() before calling
  856. * free_area_init_nodes() passing in the PFN each zone ends at. At a basic
  857. * usage, an architecture is expected to do something like
  858. *
  859. * unsigned long max_zone_pfns[MAX_NR_ZONES] = {max_dma, max_normal_pfn,
  860. * max_highmem_pfn};
  861. * for_each_valid_physical_page_range()
  862. * add_active_range(node_id, start_pfn, end_pfn)
  863. * free_area_init_nodes(max_zone_pfns);
  864. *
  865. * If the architecture guarantees that there are no holes in the ranges
  866. * registered with add_active_range(), free_bootmem_active_regions()
  867. * will call free_bootmem_node() for each registered physical page range.
  868. * Similarly sparse_memory_present_with_active_regions() calls
  869. * memory_present() for each range when SPARSEMEM is enabled.
  870. *
  871. * See mm/page_alloc.c for more information on each function exposed by
  872. * CONFIG_ARCH_POPULATES_NODE_MAP
  873. */
  874. extern void free_area_init_nodes(unsigned long *max_zone_pfn);
  875. extern void add_active_range(unsigned int nid, unsigned long start_pfn,
  876. unsigned long end_pfn);
  877. extern void remove_active_range(unsigned int nid, unsigned long start_pfn,
  878. unsigned long end_pfn);
  879. extern void push_node_boundaries(unsigned int nid, unsigned long start_pfn,
  880. unsigned long end_pfn);
  881. extern void remove_all_active_ranges(void);
  882. extern unsigned long absent_pages_in_range(unsigned long start_pfn,
  883. unsigned long end_pfn);
  884. extern void get_pfn_range_for_nid(unsigned int nid,
  885. unsigned long *start_pfn, unsigned long *end_pfn);
  886. extern unsigned long find_min_pfn_with_active_regions(void);
  887. extern void free_bootmem_with_active_regions(int nid,
  888. unsigned long max_low_pfn);
  889. typedef int (*work_fn_t)(unsigned long, unsigned long, void *);
  890. extern void work_with_active_regions(int nid, work_fn_t work_fn, void *data);
  891. extern void sparse_memory_present_with_active_regions(int nid);
  892. #ifndef CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID
  893. extern int early_pfn_to_nid(unsigned long pfn);
  894. #endif /* CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID */
  895. #endif /* CONFIG_ARCH_POPULATES_NODE_MAP */
  896. extern void set_dma_reserve(unsigned long new_dma_reserve);
  897. extern void memmap_init_zone(unsigned long, int, unsigned long,
  898. unsigned long, enum memmap_context);
  899. extern void setup_per_zone_pages_min(void);
  900. extern void mem_init(void);
  901. extern void show_mem(void);
  902. extern void si_meminfo(struct sysinfo * val);
  903. extern void si_meminfo_node(struct sysinfo *val, int nid);
  904. extern int after_bootmem;
  905. #ifdef CONFIG_NUMA
  906. extern void setup_per_cpu_pageset(void);
  907. #else
  908. static inline void setup_per_cpu_pageset(void) {}
  909. #endif
  910. /* prio_tree.c */
  911. void vma_prio_tree_add(struct vm_area_struct *, struct vm_area_struct *old);
  912. void vma_prio_tree_insert(struct vm_area_struct *, struct prio_tree_root *);
  913. void vma_prio_tree_remove(struct vm_area_struct *, struct prio_tree_root *);
  914. struct vm_area_struct *vma_prio_tree_next(struct vm_area_struct *vma,
  915. struct prio_tree_iter *iter);
  916. #define vma_prio_tree_foreach(vma, iter, root, begin, end) \
  917. for (prio_tree_iter_init(iter, root, begin, end), vma = NULL; \
  918. (vma = vma_prio_tree_next(vma, iter)); )
  919. static inline void vma_nonlinear_insert(struct vm_area_struct *vma,
  920. struct list_head *list)
  921. {
  922. vma->shared.vm_set.parent = NULL;
  923. list_add_tail(&vma->shared.vm_set.list, list);
  924. }
  925. /* mmap.c */
  926. extern int __vm_enough_memory(struct mm_struct *mm, long pages, int cap_sys_admin);
  927. extern void vma_adjust(struct vm_area_struct *vma, unsigned long start,
  928. unsigned long end, pgoff_t pgoff, struct vm_area_struct *insert);
  929. extern struct vm_area_struct *vma_merge(struct mm_struct *,
  930. struct vm_area_struct *prev, unsigned long addr, unsigned long end,
  931. unsigned long vm_flags, struct anon_vma *, struct file *, pgoff_t,
  932. struct mempolicy *);
  933. extern struct anon_vma *find_mergeable_anon_vma(struct vm_area_struct *);
  934. extern int split_vma(struct mm_struct *,
  935. struct vm_area_struct *, unsigned long addr, int new_below);
  936. extern int insert_vm_struct(struct mm_struct *, struct vm_area_struct *);
  937. extern void __vma_link_rb(struct mm_struct *, struct vm_area_struct *,
  938. struct rb_node **, struct rb_node *);
  939. extern void unlink_file_vma(struct vm_area_struct *);
  940. extern struct vm_area_struct *copy_vma(struct vm_area_struct **,
  941. unsigned long addr, unsigned long len, pgoff_t pgoff);
  942. extern void exit_mmap(struct mm_struct *);
  943. extern int mm_take_all_locks(struct mm_struct *mm);
  944. extern void mm_drop_all_locks(struct mm_struct *mm);
  945. #ifdef CONFIG_PROC_FS
  946. /* From fs/proc/base.c. callers must _not_ hold the mm's exe_file_lock */
  947. extern void added_exe_file_vma(struct mm_struct *mm);
  948. extern void removed_exe_file_vma(struct mm_struct *mm);
  949. #else
  950. static inline void added_exe_file_vma(struct mm_struct *mm)
  951. {}
  952. static inline void removed_exe_file_vma(struct mm_struct *mm)
  953. {}
  954. #endif /* CONFIG_PROC_FS */
  955. extern int may_expand_vm(struct mm_struct *mm, unsigned long npages);
  956. extern int install_special_mapping(struct mm_struct *mm,
  957. unsigned long addr, unsigned long len,
  958. unsigned long flags, struct page **pages);
  959. extern unsigned long get_unmapped_area(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
  960. extern unsigned long do_mmap_pgoff(struct file *file, unsigned long addr,
  961. unsigned long len, unsigned long prot,
  962. unsigned long flag, unsigned long pgoff);
  963. extern unsigned long mmap_region(struct file *file, unsigned long addr,
  964. unsigned long len, unsigned long flags,
  965. unsigned int vm_flags, unsigned long pgoff,
  966. int accountable);
  967. static inline unsigned long do_mmap(struct file *file, unsigned long addr,
  968. unsigned long len, unsigned long prot,
  969. unsigned long flag, unsigned long offset)
  970. {
  971. unsigned long ret = -EINVAL;
  972. if ((offset + PAGE_ALIGN(len)) < offset)
  973. goto out;
  974. if (!(offset & ~PAGE_MASK))
  975. ret = do_mmap_pgoff(file, addr, len, prot, flag, offset >> PAGE_SHIFT);
  976. out:
  977. return ret;
  978. }
  979. extern int do_munmap(struct mm_struct *, unsigned long, size_t);
  980. extern unsigned long do_brk(unsigned long, unsigned long);
  981. /* filemap.c */
  982. extern unsigned long page_unuse(struct page *);
  983. extern void truncate_inode_pages(struct address_space *, loff_t);
  984. extern void truncate_inode_pages_range(struct address_space *,
  985. loff_t lstart, loff_t lend);
  986. /* generic vm_area_ops exported for stackable file systems */
  987. extern int filemap_fault(struct vm_area_struct *, struct vm_fault *);
  988. /* mm/page-writeback.c */
  989. int write_one_page(struct page *page, int wait);
  990. /* readahead.c */
  991. #define VM_MAX_READAHEAD 128 /* kbytes */
  992. #define VM_MIN_READAHEAD 16 /* kbytes (includes current page) */
  993. int do_page_cache_readahead(struct address_space *mapping, struct file *filp,
  994. pgoff_t offset, unsigned long nr_to_read);
  995. int force_page_cache_readahead(struct address_space *mapping, struct file *filp,
  996. pgoff_t offset, unsigned long nr_to_read);
  997. void page_cache_sync_readahead(struct address_space *mapping,
  998. struct file_ra_state *ra,
  999. struct file *filp,
  1000. pgoff_t offset,
  1001. unsigned long size);
  1002. void page_cache_async_readahead(struct address_space *mapping,
  1003. struct file_ra_state *ra,
  1004. struct file *filp,
  1005. struct page *pg,
  1006. pgoff_t offset,
  1007. unsigned long size);
  1008. unsigned long max_sane_readahead(unsigned long nr);
  1009. /* Do stack extension */
  1010. extern int expand_stack(struct vm_area_struct *vma, unsigned long address);
  1011. #ifdef CONFIG_IA64
  1012. extern int expand_upwards(struct vm_area_struct *vma, unsigned long address);
  1013. #endif
  1014. extern int expand_stack_downwards(struct vm_area_struct *vma,
  1015. unsigned long address);
  1016. /* Look up the first VMA which satisfies addr < vm_end, NULL if none. */
  1017. extern struct vm_area_struct * find_vma(struct mm_struct * mm, unsigned long addr);
  1018. extern struct vm_area_struct * find_vma_prev(struct mm_struct * mm, unsigned long addr,
  1019. struct vm_area_struct **pprev);
  1020. /* Look up the first VMA which intersects the interval start_addr..end_addr-1,
  1021. NULL if none. Assume start_addr < end_addr. */
  1022. static inline struct vm_area_struct * find_vma_intersection(struct mm_struct * mm, unsigned long start_addr, unsigned long end_addr)
  1023. {
  1024. struct vm_area_struct * vma = find_vma(mm,start_addr);
  1025. if (vma && end_addr <= vma->vm_start)
  1026. vma = NULL;
  1027. return vma;
  1028. }
  1029. static inline unsigned long vma_pages(struct vm_area_struct *vma)
  1030. {
  1031. return (vma->vm_end - vma->vm_start) >> PAGE_SHIFT;
  1032. }
  1033. pgprot_t vm_get_page_prot(unsigned long vm_flags);
  1034. struct vm_area_struct *find_extend_vma(struct mm_struct *, unsigned long addr);
  1035. int remap_pfn_range(struct vm_area_struct *, unsigned long addr,
  1036. unsigned long pfn, unsigned long size, pgprot_t);
  1037. int vm_insert_page(struct vm_area_struct *, unsigned long addr, struct page *);
  1038. int vm_insert_pfn(struct vm_area_struct *vma, unsigned long addr,
  1039. unsigned long pfn);
  1040. int vm_insert_mixed(struct vm_area_struct *vma, unsigned long addr,
  1041. unsigned long pfn);
  1042. struct page *follow_page(struct vm_area_struct *, unsigned long address,
  1043. unsigned int foll_flags);
  1044. #define FOLL_WRITE 0x01 /* check pte is writable */
  1045. #define FOLL_TOUCH 0x02 /* mark page accessed */
  1046. #define FOLL_GET 0x04 /* do get_page on page */
  1047. #define FOLL_ANON 0x08 /* give ZERO_PAGE if no pgtable */
  1048. typedef int (*pte_fn_t)(pte_t *pte, pgtable_t token, unsigned long addr,
  1049. void *data);
  1050. extern int apply_to_page_range(struct mm_struct *mm, unsigned long address,
  1051. unsigned long size, pte_fn_t fn, void *data);
  1052. #ifdef CONFIG_PROC_FS
  1053. void vm_stat_account(struct mm_struct *, unsigned long, struct file *, long);
  1054. #else
  1055. static inline void vm_stat_account(struct mm_struct *mm,
  1056. unsigned long flags, struct file *file, long pages)
  1057. {
  1058. }
  1059. #endif /* CONFIG_PROC_FS */
  1060. #ifdef CONFIG_DEBUG_PAGEALLOC
  1061. extern int debug_pagealloc_enabled;
  1062. extern void kernel_map_pages(struct page *page, int numpages, int enable);
  1063. static inline void enable_debug_pagealloc(void)
  1064. {
  1065. debug_pagealloc_enabled = 1;
  1066. }
  1067. #ifdef CONFIG_HIBERNATION
  1068. extern bool kernel_page_present(struct page *page);
  1069. #endif /* CONFIG_HIBERNATION */
  1070. #else
  1071. static inline void
  1072. kernel_map_pages(struct page *page, int numpages, int enable) {}
  1073. static inline void enable_debug_pagealloc(void)
  1074. {
  1075. }
  1076. #ifdef CONFIG_HIBERNATION
  1077. static inline bool kernel_page_present(struct page *page) { return true; }
  1078. #endif /* CONFIG_HIBERNATION */
  1079. #endif
  1080. extern struct vm_area_struct *get_gate_vma(struct task_struct *tsk);
  1081. #ifdef __HAVE_ARCH_GATE_AREA
  1082. int in_gate_area_no_task(unsigned long addr);
  1083. int in_gate_area(struct task_struct *task, unsigned long addr);
  1084. #else
  1085. int in_gate_area_no_task(unsigned long addr);
  1086. #define in_gate_area(task, addr) ({(void)task; in_gate_area_no_task(addr);})
  1087. #endif /* __HAVE_ARCH_GATE_AREA */
  1088. int drop_caches_sysctl_handler(struct ctl_table *, int, struct file *,
  1089. void __user *, size_t *, loff_t *);
  1090. unsigned long shrink_slab(unsigned long scanned, gfp_t gfp_mask,
  1091. unsigned long lru_pages);
  1092. #ifndef CONFIG_MMU
  1093. #define randomize_va_space 0
  1094. #else
  1095. extern int randomize_va_space;
  1096. #endif
  1097. const char * arch_vma_name(struct vm_area_struct *vma);
  1098. void print_vma_addr(char *prefix, unsigned long rip);
  1099. struct page *sparse_mem_map_populate(unsigned long pnum, int nid);
  1100. pgd_t *vmemmap_pgd_populate(unsigned long addr, int node);
  1101. pud_t *vmemmap_pud_populate(pgd_t *pgd, unsigned long addr, int node);
  1102. pmd_t *vmemmap_pmd_populate(pud_t *pud, unsigned long addr, int node);
  1103. pte_t *vmemmap_pte_populate(pmd_t *pmd, unsigned long addr, int node);
  1104. void *vmemmap_alloc_block(unsigned long size, int node);
  1105. void vmemmap_verify(pte_t *, int, unsigned long, unsigned long);
  1106. int vmemmap_populate_basepages(struct page *start_page,
  1107. unsigned long pages, int node);
  1108. int vmemmap_populate(struct page *start_page, unsigned long pages, int node);
  1109. void vmemmap_populate_print_last(void);
  1110. extern void *alloc_locked_buffer(size_t size);
  1111. extern void free_locked_buffer(void *buffer, size_t size);
  1112. #endif /* __KERNEL__ */
  1113. #endif /* _LINUX_MM_H */