mm.h 43 KB

1234567891011121314151617181920212223242526272829303132333435363738394041424344454647484950515253545556575859606162636465666768697071727374757677787980818283848586878889909192939495969798991001011021031041051061071081091101111121131141151161171181191201211221231241251261271281291301311321331341351361371381391401411421431441451461471481491501511521531541551561571581591601611621631641651661671681691701711721731741751761771781791801811821831841851861871881891901911921931941951961971981992002012022032042052062072082092102112122132142152162172182192202212222232242252262272282292302312322332342352362372382392402412422432442452462472482492502512522532542552562572582592602612622632642652662672682692702712722732742752762772782792802812822832842852862872882892902912922932942952962972982993003013023033043053063073083093103113123133143153163173183193203213223233243253263273283293303313323333343353363373383393403413423433443453463473483493503513523533543553563573583593603613623633643653663673683693703713723733743753763773783793803813823833843853863873883893903913923933943953963973983994004014024034044054064074084094104114124134144154164174184194204214224234244254264274284294304314324334344354364374384394404414424434444454464474484494504514524534544554564574584594604614624634644654664674684694704714724734744754764774784794804814824834844854864874884894904914924934944954964974984995005015025035045055065075085095105115125135145155165175185195205215225235245255265275285295305315325335345355365375385395405415425435445455465475485495505515525535545555565575585595605615625635645655665675685695705715725735745755765775785795805815825835845855865875885895905915925935945955965975985996006016026036046056066076086096106116126136146156166176186196206216226236246256266276286296306316326336346356366376386396406416426436446456466476486496506516526536546556566576586596606616626636646656666676686696706716726736746756766776786796806816826836846856866876886896906916926936946956966976986997007017027037047057067077087097107117127137147157167177187197207217227237247257267277287297307317327337347357367377387397407417427437447457467477487497507517527537547557567577587597607617627637647657667677687697707717727737747757767777787797807817827837847857867877887897907917927937947957967977987998008018028038048058068078088098108118128138148158168178188198208218228238248258268278288298308318328338348358368378388398408418428438448458468478488498508518528538548558568578588598608618628638648658668678688698708718728738748758768778788798808818828838848858868878888898908918928938948958968978988999009019029039049059069079089099109119129139149159169179189199209219229239249259269279289299309319329339349359369379389399409419429439449459469479489499509519529539549559569579589599609619629639649659669679689699709719729739749759769779789799809819829839849859869879889899909919929939949959969979989991000100110021003100410051006100710081009101010111012101310141015101610171018101910201021102210231024102510261027102810291030103110321033103410351036103710381039104010411042104310441045104610471048104910501051105210531054105510561057105810591060106110621063106410651066106710681069107010711072107310741075107610771078107910801081108210831084108510861087108810891090109110921093109410951096109710981099110011011102110311041105110611071108110911101111111211131114111511161117111811191120112111221123112411251126112711281129113011311132113311341135113611371138113911401141114211431144114511461147114811491150115111521153115411551156115711581159116011611162116311641165116611671168116911701171117211731174117511761177117811791180118111821183118411851186118711881189119011911192119311941195119611971198119912001201120212031204120512061207120812091210121112121213121412151216121712181219122012211222122312241225122612271228122912301231123212331234123512361237123812391240124112421243124412451246124712481249125012511252125312541255125612571258125912601261126212631264126512661267126812691270
  1. #ifndef _LINUX_MM_H
  2. #define _LINUX_MM_H
  3. #include <linux/errno.h>
  4. #ifdef __KERNEL__
  5. #include <linux/gfp.h>
  6. #include <linux/list.h>
  7. #include <linux/mmzone.h>
  8. #include <linux/rbtree.h>
  9. #include <linux/prio_tree.h>
  10. #include <linux/debug_locks.h>
  11. #include <linux/mm_types.h>
  12. struct mempolicy;
  13. struct anon_vma;
  14. struct file_ra_state;
  15. struct user_struct;
  16. struct writeback_control;
  17. #ifndef CONFIG_DISCONTIGMEM /* Don't use mapnrs, do it properly */
  18. extern unsigned long max_mapnr;
  19. #endif
  20. extern unsigned long num_physpages;
  21. extern void * high_memory;
  22. extern int page_cluster;
  23. #ifdef CONFIG_SYSCTL
  24. extern int sysctl_legacy_va_layout;
  25. #else
  26. #define sysctl_legacy_va_layout 0
  27. #endif
  28. extern unsigned long mmap_min_addr;
  29. #include <asm/page.h>
  30. #include <asm/pgtable.h>
  31. #include <asm/processor.h>
  32. #define nth_page(page,n) pfn_to_page(page_to_pfn((page)) + (n))
  33. /*
  34. * Linux kernel virtual memory manager primitives.
  35. * The idea being to have a "virtual" mm in the same way
  36. * we have a virtual fs - giving a cleaner interface to the
  37. * mm details, and allowing different kinds of memory mappings
  38. * (from shared memory to executable loading to arbitrary
  39. * mmap() functions).
  40. */
  41. extern struct kmem_cache *vm_area_cachep;
  42. /*
  43. * This struct defines the per-mm list of VMAs for uClinux. If CONFIG_MMU is
  44. * disabled, then there's a single shared list of VMAs maintained by the
  45. * system, and mm's subscribe to these individually
  46. */
  47. struct vm_list_struct {
  48. struct vm_list_struct *next;
  49. struct vm_area_struct *vma;
  50. };
  51. #ifndef CONFIG_MMU
  52. extern struct rb_root nommu_vma_tree;
  53. extern struct rw_semaphore nommu_vma_sem;
  54. extern unsigned int kobjsize(const void *objp);
  55. #endif
  56. /*
  57. * vm_flags..
  58. */
  59. #define VM_READ 0x00000001 /* currently active flags */
  60. #define VM_WRITE 0x00000002
  61. #define VM_EXEC 0x00000004
  62. #define VM_SHARED 0x00000008
  63. /* mprotect() hardcodes VM_MAYREAD >> 4 == VM_READ, and so for r/w/x bits. */
  64. #define VM_MAYREAD 0x00000010 /* limits for mprotect() etc */
  65. #define VM_MAYWRITE 0x00000020
  66. #define VM_MAYEXEC 0x00000040
  67. #define VM_MAYSHARE 0x00000080
  68. #define VM_GROWSDOWN 0x00000100 /* general info on the segment */
  69. #define VM_GROWSUP 0x00000200
  70. #define VM_PFNMAP 0x00000400 /* Page-ranges managed without "struct page", just pure PFN */
  71. #define VM_DENYWRITE 0x00000800 /* ETXTBSY on write attempts.. */
  72. #define VM_EXECUTABLE 0x00001000
  73. #define VM_LOCKED 0x00002000
  74. #define VM_IO 0x00004000 /* Memory mapped I/O or similar */
  75. /* Used by sys_madvise() */
  76. #define VM_SEQ_READ 0x00008000 /* App will access data sequentially */
  77. #define VM_RAND_READ 0x00010000 /* App will not benefit from clustered reads */
  78. #define VM_DONTCOPY 0x00020000 /* Do not copy this vma on fork */
  79. #define VM_DONTEXPAND 0x00040000 /* Cannot expand with mremap() */
  80. #define VM_RESERVED 0x00080000 /* Count as reserved_vm like IO */
  81. #define VM_ACCOUNT 0x00100000 /* Is a VM accounted object */
  82. #define VM_HUGETLB 0x00400000 /* Huge TLB Page VM */
  83. #define VM_NONLINEAR 0x00800000 /* Is non-linear (remap_file_pages) */
  84. #define VM_MAPPED_COPY 0x01000000 /* T if mapped copy of data (nommu mmap) */
  85. #define VM_INSERTPAGE 0x02000000 /* The vma has had "vm_insert_page()" done on it */
  86. #define VM_ALWAYSDUMP 0x04000000 /* Always include in core dumps */
  87. #define VM_CAN_NONLINEAR 0x08000000 /* Has ->fault & does nonlinear pages */
  88. #define VM_MIXEDMAP 0x10000000 /* Can contain "struct page" and pure PFN pages */
  89. #define VM_SAO 0x20000000 /* Strong Access Ordering (powerpc) */
  90. #ifndef VM_STACK_DEFAULT_FLAGS /* arch can override this */
  91. #define VM_STACK_DEFAULT_FLAGS VM_DATA_DEFAULT_FLAGS
  92. #endif
  93. #ifdef CONFIG_STACK_GROWSUP
  94. #define VM_STACK_FLAGS (VM_GROWSUP | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT)
  95. #else
  96. #define VM_STACK_FLAGS (VM_GROWSDOWN | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT)
  97. #endif
  98. #define VM_READHINTMASK (VM_SEQ_READ | VM_RAND_READ)
  99. #define VM_ClearReadHint(v) (v)->vm_flags &= ~VM_READHINTMASK
  100. #define VM_NormalReadHint(v) (!((v)->vm_flags & VM_READHINTMASK))
  101. #define VM_SequentialReadHint(v) ((v)->vm_flags & VM_SEQ_READ)
  102. #define VM_RandomReadHint(v) ((v)->vm_flags & VM_RAND_READ)
  103. /*
  104. * mapping from the currently active vm_flags protection bits (the
  105. * low four bits) to a page protection mask..
  106. */
  107. extern pgprot_t protection_map[16];
  108. #define FAULT_FLAG_WRITE 0x01 /* Fault was a write access */
  109. #define FAULT_FLAG_NONLINEAR 0x02 /* Fault was via a nonlinear mapping */
  110. /*
  111. * vm_fault is filled by the the pagefault handler and passed to the vma's
  112. * ->fault function. The vma's ->fault is responsible for returning a bitmask
  113. * of VM_FAULT_xxx flags that give details about how the fault was handled.
  114. *
  115. * pgoff should be used in favour of virtual_address, if possible. If pgoff
  116. * is used, one may set VM_CAN_NONLINEAR in the vma->vm_flags to get nonlinear
  117. * mapping support.
  118. */
  119. struct vm_fault {
  120. unsigned int flags; /* FAULT_FLAG_xxx flags */
  121. pgoff_t pgoff; /* Logical page offset based on vma */
  122. void __user *virtual_address; /* Faulting virtual address */
  123. struct page *page; /* ->fault handlers should return a
  124. * page here, unless VM_FAULT_NOPAGE
  125. * is set (which is also implied by
  126. * VM_FAULT_ERROR).
  127. */
  128. };
  129. /*
  130. * These are the virtual MM functions - opening of an area, closing and
  131. * unmapping it (needed to keep files on disk up-to-date etc), pointer
  132. * to the functions called when a no-page or a wp-page exception occurs.
  133. */
  134. struct vm_operations_struct {
  135. void (*open)(struct vm_area_struct * area);
  136. void (*close)(struct vm_area_struct * area);
  137. int (*fault)(struct vm_area_struct *vma, struct vm_fault *vmf);
  138. /* notification that a previously read-only page is about to become
  139. * writable, if an error is returned it will cause a SIGBUS */
  140. int (*page_mkwrite)(struct vm_area_struct *vma, struct page *page);
  141. /* called by access_process_vm when get_user_pages() fails, typically
  142. * for use by special VMAs that can switch between memory and hardware
  143. */
  144. int (*access)(struct vm_area_struct *vma, unsigned long addr,
  145. void *buf, int len, int write);
  146. #ifdef CONFIG_NUMA
  147. /*
  148. * set_policy() op must add a reference to any non-NULL @new mempolicy
  149. * to hold the policy upon return. Caller should pass NULL @new to
  150. * remove a policy and fall back to surrounding context--i.e. do not
  151. * install a MPOL_DEFAULT policy, nor the task or system default
  152. * mempolicy.
  153. */
  154. int (*set_policy)(struct vm_area_struct *vma, struct mempolicy *new);
  155. /*
  156. * get_policy() op must add reference [mpol_get()] to any policy at
  157. * (vma,addr) marked as MPOL_SHARED. The shared policy infrastructure
  158. * in mm/mempolicy.c will do this automatically.
  159. * get_policy() must NOT add a ref if the policy at (vma,addr) is not
  160. * marked as MPOL_SHARED. vma policies are protected by the mmap_sem.
  161. * If no [shared/vma] mempolicy exists at the addr, get_policy() op
  162. * must return NULL--i.e., do not "fallback" to task or system default
  163. * policy.
  164. */
  165. struct mempolicy *(*get_policy)(struct vm_area_struct *vma,
  166. unsigned long addr);
  167. int (*migrate)(struct vm_area_struct *vma, const nodemask_t *from,
  168. const nodemask_t *to, unsigned long flags);
  169. #endif
  170. };
  171. struct mmu_gather;
  172. struct inode;
  173. #define page_private(page) ((page)->private)
  174. #define set_page_private(page, v) ((page)->private = (v))
  175. /*
  176. * FIXME: take this include out, include page-flags.h in
  177. * files which need it (119 of them)
  178. */
  179. #include <linux/page-flags.h>
  180. #ifdef CONFIG_DEBUG_VM
  181. #define VM_BUG_ON(cond) BUG_ON(cond)
  182. #else
  183. #define VM_BUG_ON(condition) do { } while(0)
  184. #endif
  185. /*
  186. * Methods to modify the page usage count.
  187. *
  188. * What counts for a page usage:
  189. * - cache mapping (page->mapping)
  190. * - private data (page->private)
  191. * - page mapped in a task's page tables, each mapping
  192. * is counted separately
  193. *
  194. * Also, many kernel routines increase the page count before a critical
  195. * routine so they can be sure the page doesn't go away from under them.
  196. */
  197. /*
  198. * Drop a ref, return true if the refcount fell to zero (the page has no users)
  199. */
  200. static inline int put_page_testzero(struct page *page)
  201. {
  202. VM_BUG_ON(atomic_read(&page->_count) == 0);
  203. return atomic_dec_and_test(&page->_count);
  204. }
  205. /*
  206. * Try to grab a ref unless the page has a refcount of zero, return false if
  207. * that is the case.
  208. */
  209. static inline int get_page_unless_zero(struct page *page)
  210. {
  211. VM_BUG_ON(PageTail(page));
  212. return atomic_inc_not_zero(&page->_count);
  213. }
  214. /* Support for virtually mapped pages */
  215. struct page *vmalloc_to_page(const void *addr);
  216. unsigned long vmalloc_to_pfn(const void *addr);
  217. /*
  218. * Determine if an address is within the vmalloc range
  219. *
  220. * On nommu, vmalloc/vfree wrap through kmalloc/kfree directly, so there
  221. * is no special casing required.
  222. */
  223. static inline int is_vmalloc_addr(const void *x)
  224. {
  225. #ifdef CONFIG_MMU
  226. unsigned long addr = (unsigned long)x;
  227. return addr >= VMALLOC_START && addr < VMALLOC_END;
  228. #else
  229. return 0;
  230. #endif
  231. }
  232. static inline struct page *compound_head(struct page *page)
  233. {
  234. if (unlikely(PageTail(page)))
  235. return page->first_page;
  236. return page;
  237. }
  238. static inline int page_count(struct page *page)
  239. {
  240. return atomic_read(&compound_head(page)->_count);
  241. }
  242. static inline void get_page(struct page *page)
  243. {
  244. page = compound_head(page);
  245. VM_BUG_ON(atomic_read(&page->_count) == 0);
  246. atomic_inc(&page->_count);
  247. }
  248. static inline struct page *virt_to_head_page(const void *x)
  249. {
  250. struct page *page = virt_to_page(x);
  251. return compound_head(page);
  252. }
  253. /*
  254. * Setup the page count before being freed into the page allocator for
  255. * the first time (boot or memory hotplug)
  256. */
  257. static inline void init_page_count(struct page *page)
  258. {
  259. atomic_set(&page->_count, 1);
  260. }
  261. void put_page(struct page *page);
  262. void put_pages_list(struct list_head *pages);
  263. void split_page(struct page *page, unsigned int order);
  264. /*
  265. * Compound pages have a destructor function. Provide a
  266. * prototype for that function and accessor functions.
  267. * These are _only_ valid on the head of a PG_compound page.
  268. */
  269. typedef void compound_page_dtor(struct page *);
  270. static inline void set_compound_page_dtor(struct page *page,
  271. compound_page_dtor *dtor)
  272. {
  273. page[1].lru.next = (void *)dtor;
  274. }
  275. static inline compound_page_dtor *get_compound_page_dtor(struct page *page)
  276. {
  277. return (compound_page_dtor *)page[1].lru.next;
  278. }
  279. static inline int compound_order(struct page *page)
  280. {
  281. if (!PageHead(page))
  282. return 0;
  283. return (unsigned long)page[1].lru.prev;
  284. }
  285. static inline void set_compound_order(struct page *page, unsigned long order)
  286. {
  287. page[1].lru.prev = (void *)order;
  288. }
  289. /*
  290. * Multiple processes may "see" the same page. E.g. for untouched
  291. * mappings of /dev/null, all processes see the same page full of
  292. * zeroes, and text pages of executables and shared libraries have
  293. * only one copy in memory, at most, normally.
  294. *
  295. * For the non-reserved pages, page_count(page) denotes a reference count.
  296. * page_count() == 0 means the page is free. page->lru is then used for
  297. * freelist management in the buddy allocator.
  298. * page_count() > 0 means the page has been allocated.
  299. *
  300. * Pages are allocated by the slab allocator in order to provide memory
  301. * to kmalloc and kmem_cache_alloc. In this case, the management of the
  302. * page, and the fields in 'struct page' are the responsibility of mm/slab.c
  303. * unless a particular usage is carefully commented. (the responsibility of
  304. * freeing the kmalloc memory is the caller's, of course).
  305. *
  306. * A page may be used by anyone else who does a __get_free_page().
  307. * In this case, page_count still tracks the references, and should only
  308. * be used through the normal accessor functions. The top bits of page->flags
  309. * and page->virtual store page management information, but all other fields
  310. * are unused and could be used privately, carefully. The management of this
  311. * page is the responsibility of the one who allocated it, and those who have
  312. * subsequently been given references to it.
  313. *
  314. * The other pages (we may call them "pagecache pages") are completely
  315. * managed by the Linux memory manager: I/O, buffers, swapping etc.
  316. * The following discussion applies only to them.
  317. *
  318. * A pagecache page contains an opaque `private' member, which belongs to the
  319. * page's address_space. Usually, this is the address of a circular list of
  320. * the page's disk buffers. PG_private must be set to tell the VM to call
  321. * into the filesystem to release these pages.
  322. *
  323. * A page may belong to an inode's memory mapping. In this case, page->mapping
  324. * is the pointer to the inode, and page->index is the file offset of the page,
  325. * in units of PAGE_CACHE_SIZE.
  326. *
  327. * If pagecache pages are not associated with an inode, they are said to be
  328. * anonymous pages. These may become associated with the swapcache, and in that
  329. * case PG_swapcache is set, and page->private is an offset into the swapcache.
  330. *
  331. * In either case (swapcache or inode backed), the pagecache itself holds one
  332. * reference to the page. Setting PG_private should also increment the
  333. * refcount. The each user mapping also has a reference to the page.
  334. *
  335. * The pagecache pages are stored in a per-mapping radix tree, which is
  336. * rooted at mapping->page_tree, and indexed by offset.
  337. * Where 2.4 and early 2.6 kernels kept dirty/clean pages in per-address_space
  338. * lists, we instead now tag pages as dirty/writeback in the radix tree.
  339. *
  340. * All pagecache pages may be subject to I/O:
  341. * - inode pages may need to be read from disk,
  342. * - inode pages which have been modified and are MAP_SHARED may need
  343. * to be written back to the inode on disk,
  344. * - anonymous pages (including MAP_PRIVATE file mappings) which have been
  345. * modified may need to be swapped out to swap space and (later) to be read
  346. * back into memory.
  347. */
  348. /*
  349. * The zone field is never updated after free_area_init_core()
  350. * sets it, so none of the operations on it need to be atomic.
  351. */
  352. /*
  353. * page->flags layout:
  354. *
  355. * There are three possibilities for how page->flags get
  356. * laid out. The first is for the normal case, without
  357. * sparsemem. The second is for sparsemem when there is
  358. * plenty of space for node and section. The last is when
  359. * we have run out of space and have to fall back to an
  360. * alternate (slower) way of determining the node.
  361. *
  362. * No sparsemem or sparsemem vmemmap: | NODE | ZONE | ... | FLAGS |
  363. * classic sparse with space for node:| SECTION | NODE | ZONE | ... | FLAGS |
  364. * classic sparse no space for node: | SECTION | ZONE | ... | FLAGS |
  365. */
  366. #if defined(CONFIG_SPARSEMEM) && !defined(CONFIG_SPARSEMEM_VMEMMAP)
  367. #define SECTIONS_WIDTH SECTIONS_SHIFT
  368. #else
  369. #define SECTIONS_WIDTH 0
  370. #endif
  371. #define ZONES_WIDTH ZONES_SHIFT
  372. #if SECTIONS_WIDTH+ZONES_WIDTH+NODES_SHIFT <= BITS_PER_LONG - NR_PAGEFLAGS
  373. #define NODES_WIDTH NODES_SHIFT
  374. #else
  375. #ifdef CONFIG_SPARSEMEM_VMEMMAP
  376. #error "Vmemmap: No space for nodes field in page flags"
  377. #endif
  378. #define NODES_WIDTH 0
  379. #endif
  380. /* Page flags: | [SECTION] | [NODE] | ZONE | ... | FLAGS | */
  381. #define SECTIONS_PGOFF ((sizeof(unsigned long)*8) - SECTIONS_WIDTH)
  382. #define NODES_PGOFF (SECTIONS_PGOFF - NODES_WIDTH)
  383. #define ZONES_PGOFF (NODES_PGOFF - ZONES_WIDTH)
  384. /*
  385. * We are going to use the flags for the page to node mapping if its in
  386. * there. This includes the case where there is no node, so it is implicit.
  387. */
  388. #if !(NODES_WIDTH > 0 || NODES_SHIFT == 0)
  389. #define NODE_NOT_IN_PAGE_FLAGS
  390. #endif
  391. #ifndef PFN_SECTION_SHIFT
  392. #define PFN_SECTION_SHIFT 0
  393. #endif
  394. /*
  395. * Define the bit shifts to access each section. For non-existant
  396. * sections we define the shift as 0; that plus a 0 mask ensures
  397. * the compiler will optimise away reference to them.
  398. */
  399. #define SECTIONS_PGSHIFT (SECTIONS_PGOFF * (SECTIONS_WIDTH != 0))
  400. #define NODES_PGSHIFT (NODES_PGOFF * (NODES_WIDTH != 0))
  401. #define ZONES_PGSHIFT (ZONES_PGOFF * (ZONES_WIDTH != 0))
  402. /* NODE:ZONE or SECTION:ZONE is used to ID a zone for the buddy allcator */
  403. #ifdef NODE_NOT_IN_PAGEFLAGS
  404. #define ZONEID_SHIFT (SECTIONS_SHIFT + ZONES_SHIFT)
  405. #define ZONEID_PGOFF ((SECTIONS_PGOFF < ZONES_PGOFF)? \
  406. SECTIONS_PGOFF : ZONES_PGOFF)
  407. #else
  408. #define ZONEID_SHIFT (NODES_SHIFT + ZONES_SHIFT)
  409. #define ZONEID_PGOFF ((NODES_PGOFF < ZONES_PGOFF)? \
  410. NODES_PGOFF : ZONES_PGOFF)
  411. #endif
  412. #define ZONEID_PGSHIFT (ZONEID_PGOFF * (ZONEID_SHIFT != 0))
  413. #if SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > BITS_PER_LONG - NR_PAGEFLAGS
  414. #error SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > BITS_PER_LONG - NR_PAGEFLAGS
  415. #endif
  416. #define ZONES_MASK ((1UL << ZONES_WIDTH) - 1)
  417. #define NODES_MASK ((1UL << NODES_WIDTH) - 1)
  418. #define SECTIONS_MASK ((1UL << SECTIONS_WIDTH) - 1)
  419. #define ZONEID_MASK ((1UL << ZONEID_SHIFT) - 1)
  420. static inline enum zone_type page_zonenum(struct page *page)
  421. {
  422. return (page->flags >> ZONES_PGSHIFT) & ZONES_MASK;
  423. }
  424. /*
  425. * The identification function is only used by the buddy allocator for
  426. * determining if two pages could be buddies. We are not really
  427. * identifying a zone since we could be using a the section number
  428. * id if we have not node id available in page flags.
  429. * We guarantee only that it will return the same value for two
  430. * combinable pages in a zone.
  431. */
  432. static inline int page_zone_id(struct page *page)
  433. {
  434. return (page->flags >> ZONEID_PGSHIFT) & ZONEID_MASK;
  435. }
  436. static inline int zone_to_nid(struct zone *zone)
  437. {
  438. #ifdef CONFIG_NUMA
  439. return zone->node;
  440. #else
  441. return 0;
  442. #endif
  443. }
  444. #ifdef NODE_NOT_IN_PAGE_FLAGS
  445. extern int page_to_nid(struct page *page);
  446. #else
  447. static inline int page_to_nid(struct page *page)
  448. {
  449. return (page->flags >> NODES_PGSHIFT) & NODES_MASK;
  450. }
  451. #endif
  452. static inline struct zone *page_zone(struct page *page)
  453. {
  454. return &NODE_DATA(page_to_nid(page))->node_zones[page_zonenum(page)];
  455. }
  456. #if defined(CONFIG_SPARSEMEM) && !defined(CONFIG_SPARSEMEM_VMEMMAP)
  457. static inline unsigned long page_to_section(struct page *page)
  458. {
  459. return (page->flags >> SECTIONS_PGSHIFT) & SECTIONS_MASK;
  460. }
  461. #endif
  462. static inline void set_page_zone(struct page *page, enum zone_type zone)
  463. {
  464. page->flags &= ~(ZONES_MASK << ZONES_PGSHIFT);
  465. page->flags |= (zone & ZONES_MASK) << ZONES_PGSHIFT;
  466. }
  467. static inline void set_page_node(struct page *page, unsigned long node)
  468. {
  469. page->flags &= ~(NODES_MASK << NODES_PGSHIFT);
  470. page->flags |= (node & NODES_MASK) << NODES_PGSHIFT;
  471. }
  472. static inline void set_page_section(struct page *page, unsigned long section)
  473. {
  474. page->flags &= ~(SECTIONS_MASK << SECTIONS_PGSHIFT);
  475. page->flags |= (section & SECTIONS_MASK) << SECTIONS_PGSHIFT;
  476. }
  477. static inline void set_page_links(struct page *page, enum zone_type zone,
  478. unsigned long node, unsigned long pfn)
  479. {
  480. set_page_zone(page, zone);
  481. set_page_node(page, node);
  482. set_page_section(page, pfn_to_section_nr(pfn));
  483. }
  484. /*
  485. * If a hint addr is less than mmap_min_addr change hint to be as
  486. * low as possible but still greater than mmap_min_addr
  487. */
  488. static inline unsigned long round_hint_to_min(unsigned long hint)
  489. {
  490. #ifdef CONFIG_SECURITY
  491. hint &= PAGE_MASK;
  492. if (((void *)hint != NULL) &&
  493. (hint < mmap_min_addr))
  494. return PAGE_ALIGN(mmap_min_addr);
  495. #endif
  496. return hint;
  497. }
  498. /*
  499. * Some inline functions in vmstat.h depend on page_zone()
  500. */
  501. #include <linux/vmstat.h>
  502. static __always_inline void *lowmem_page_address(struct page *page)
  503. {
  504. return __va(page_to_pfn(page) << PAGE_SHIFT);
  505. }
  506. #if defined(CONFIG_HIGHMEM) && !defined(WANT_PAGE_VIRTUAL)
  507. #define HASHED_PAGE_VIRTUAL
  508. #endif
  509. #if defined(WANT_PAGE_VIRTUAL)
  510. #define page_address(page) ((page)->virtual)
  511. #define set_page_address(page, address) \
  512. do { \
  513. (page)->virtual = (address); \
  514. } while(0)
  515. #define page_address_init() do { } while(0)
  516. #endif
  517. #if defined(HASHED_PAGE_VIRTUAL)
  518. void *page_address(struct page *page);
  519. void set_page_address(struct page *page, void *virtual);
  520. void page_address_init(void);
  521. #endif
  522. #if !defined(HASHED_PAGE_VIRTUAL) && !defined(WANT_PAGE_VIRTUAL)
  523. #define page_address(page) lowmem_page_address(page)
  524. #define set_page_address(page, address) do { } while(0)
  525. #define page_address_init() do { } while(0)
  526. #endif
  527. /*
  528. * On an anonymous page mapped into a user virtual memory area,
  529. * page->mapping points to its anon_vma, not to a struct address_space;
  530. * with the PAGE_MAPPING_ANON bit set to distinguish it.
  531. *
  532. * Please note that, confusingly, "page_mapping" refers to the inode
  533. * address_space which maps the page from disk; whereas "page_mapped"
  534. * refers to user virtual address space into which the page is mapped.
  535. */
  536. #define PAGE_MAPPING_ANON 1
  537. extern struct address_space swapper_space;
  538. static inline struct address_space *page_mapping(struct page *page)
  539. {
  540. struct address_space *mapping = page->mapping;
  541. VM_BUG_ON(PageSlab(page));
  542. #ifdef CONFIG_SWAP
  543. if (unlikely(PageSwapCache(page)))
  544. mapping = &swapper_space;
  545. else
  546. #endif
  547. if (unlikely((unsigned long)mapping & PAGE_MAPPING_ANON))
  548. mapping = NULL;
  549. return mapping;
  550. }
  551. static inline int PageAnon(struct page *page)
  552. {
  553. return ((unsigned long)page->mapping & PAGE_MAPPING_ANON) != 0;
  554. }
  555. /*
  556. * Return the pagecache index of the passed page. Regular pagecache pages
  557. * use ->index whereas swapcache pages use ->private
  558. */
  559. static inline pgoff_t page_index(struct page *page)
  560. {
  561. if (unlikely(PageSwapCache(page)))
  562. return page_private(page);
  563. return page->index;
  564. }
  565. /*
  566. * The atomic page->_mapcount, like _count, starts from -1:
  567. * so that transitions both from it and to it can be tracked,
  568. * using atomic_inc_and_test and atomic_add_negative(-1).
  569. */
  570. static inline void reset_page_mapcount(struct page *page)
  571. {
  572. atomic_set(&(page)->_mapcount, -1);
  573. }
  574. static inline int page_mapcount(struct page *page)
  575. {
  576. return atomic_read(&(page)->_mapcount) + 1;
  577. }
  578. /*
  579. * Return true if this page is mapped into pagetables.
  580. */
  581. static inline int page_mapped(struct page *page)
  582. {
  583. return atomic_read(&(page)->_mapcount) >= 0;
  584. }
  585. /*
  586. * Different kinds of faults, as returned by handle_mm_fault().
  587. * Used to decide whether a process gets delivered SIGBUS or
  588. * just gets major/minor fault counters bumped up.
  589. */
  590. #define VM_FAULT_MINOR 0 /* For backwards compat. Remove me quickly. */
  591. #define VM_FAULT_OOM 0x0001
  592. #define VM_FAULT_SIGBUS 0x0002
  593. #define VM_FAULT_MAJOR 0x0004
  594. #define VM_FAULT_WRITE 0x0008 /* Special case for get_user_pages */
  595. #define VM_FAULT_NOPAGE 0x0100 /* ->fault installed the pte, not return page */
  596. #define VM_FAULT_LOCKED 0x0200 /* ->fault locked the returned page */
  597. #define VM_FAULT_ERROR (VM_FAULT_OOM | VM_FAULT_SIGBUS)
  598. #define offset_in_page(p) ((unsigned long)(p) & ~PAGE_MASK)
  599. extern void show_free_areas(void);
  600. #ifdef CONFIG_SHMEM
  601. int shmem_lock(struct file *file, int lock, struct user_struct *user);
  602. #else
  603. static inline int shmem_lock(struct file *file, int lock,
  604. struct user_struct *user)
  605. {
  606. return 0;
  607. }
  608. #endif
  609. struct file *shmem_file_setup(char *name, loff_t size, unsigned long flags);
  610. int shmem_zero_setup(struct vm_area_struct *);
  611. #ifndef CONFIG_MMU
  612. extern unsigned long shmem_get_unmapped_area(struct file *file,
  613. unsigned long addr,
  614. unsigned long len,
  615. unsigned long pgoff,
  616. unsigned long flags);
  617. #endif
  618. extern int can_do_mlock(void);
  619. extern int user_shm_lock(size_t, struct user_struct *);
  620. extern void user_shm_unlock(size_t, struct user_struct *);
  621. /*
  622. * Parameter block passed down to zap_pte_range in exceptional cases.
  623. */
  624. struct zap_details {
  625. struct vm_area_struct *nonlinear_vma; /* Check page->index if set */
  626. struct address_space *check_mapping; /* Check page->mapping if set */
  627. pgoff_t first_index; /* Lowest page->index to unmap */
  628. pgoff_t last_index; /* Highest page->index to unmap */
  629. spinlock_t *i_mmap_lock; /* For unmap_mapping_range: */
  630. unsigned long truncate_count; /* Compare vm_truncate_count */
  631. };
  632. struct page *vm_normal_page(struct vm_area_struct *vma, unsigned long addr,
  633. pte_t pte);
  634. unsigned long zap_page_range(struct vm_area_struct *vma, unsigned long address,
  635. unsigned long size, struct zap_details *);
  636. unsigned long unmap_vmas(struct mmu_gather **tlb,
  637. struct vm_area_struct *start_vma, unsigned long start_addr,
  638. unsigned long end_addr, unsigned long *nr_accounted,
  639. struct zap_details *);
  640. /**
  641. * mm_walk - callbacks for walk_page_range
  642. * @pgd_entry: if set, called for each non-empty PGD (top-level) entry
  643. * @pud_entry: if set, called for each non-empty PUD (2nd-level) entry
  644. * @pmd_entry: if set, called for each non-empty PMD (3rd-level) entry
  645. * @pte_entry: if set, called for each non-empty PTE (4th-level) entry
  646. * @pte_hole: if set, called for each hole at all levels
  647. *
  648. * (see walk_page_range for more details)
  649. */
  650. struct mm_walk {
  651. int (*pgd_entry)(pgd_t *, unsigned long, unsigned long, struct mm_walk *);
  652. int (*pud_entry)(pud_t *, unsigned long, unsigned long, struct mm_walk *);
  653. int (*pmd_entry)(pmd_t *, unsigned long, unsigned long, struct mm_walk *);
  654. int (*pte_entry)(pte_t *, unsigned long, unsigned long, struct mm_walk *);
  655. int (*pte_hole)(unsigned long, unsigned long, struct mm_walk *);
  656. struct mm_struct *mm;
  657. void *private;
  658. };
  659. int walk_page_range(unsigned long addr, unsigned long end,
  660. struct mm_walk *walk);
  661. void free_pgd_range(struct mmu_gather *tlb, unsigned long addr,
  662. unsigned long end, unsigned long floor, unsigned long ceiling);
  663. int copy_page_range(struct mm_struct *dst, struct mm_struct *src,
  664. struct vm_area_struct *vma);
  665. void unmap_mapping_range(struct address_space *mapping,
  666. loff_t const holebegin, loff_t const holelen, int even_cows);
  667. int generic_access_phys(struct vm_area_struct *vma, unsigned long addr,
  668. void *buf, int len, int write);
  669. static inline void unmap_shared_mapping_range(struct address_space *mapping,
  670. loff_t const holebegin, loff_t const holelen)
  671. {
  672. unmap_mapping_range(mapping, holebegin, holelen, 0);
  673. }
  674. extern int vmtruncate(struct inode * inode, loff_t offset);
  675. extern int vmtruncate_range(struct inode * inode, loff_t offset, loff_t end);
  676. #ifdef CONFIG_MMU
  677. extern int handle_mm_fault(struct mm_struct *mm, struct vm_area_struct *vma,
  678. unsigned long address, int write_access);
  679. #else
  680. static inline int handle_mm_fault(struct mm_struct *mm,
  681. struct vm_area_struct *vma, unsigned long address,
  682. int write_access)
  683. {
  684. /* should never happen if there's no MMU */
  685. BUG();
  686. return VM_FAULT_SIGBUS;
  687. }
  688. #endif
  689. extern int make_pages_present(unsigned long addr, unsigned long end);
  690. extern int access_process_vm(struct task_struct *tsk, unsigned long addr, void *buf, int len, int write);
  691. int get_user_pages(struct task_struct *tsk, struct mm_struct *mm, unsigned long start,
  692. int len, int write, int force, struct page **pages, struct vm_area_struct **vmas);
  693. void print_bad_pte(struct vm_area_struct *, pte_t, unsigned long);
  694. extern int try_to_release_page(struct page * page, gfp_t gfp_mask);
  695. extern void do_invalidatepage(struct page *page, unsigned long offset);
  696. int __set_page_dirty_nobuffers(struct page *page);
  697. int __set_page_dirty_no_writeback(struct page *page);
  698. int redirty_page_for_writepage(struct writeback_control *wbc,
  699. struct page *page);
  700. int set_page_dirty(struct page *page);
  701. int set_page_dirty_lock(struct page *page);
  702. int clear_page_dirty_for_io(struct page *page);
  703. extern unsigned long move_page_tables(struct vm_area_struct *vma,
  704. unsigned long old_addr, struct vm_area_struct *new_vma,
  705. unsigned long new_addr, unsigned long len);
  706. extern unsigned long do_mremap(unsigned long addr,
  707. unsigned long old_len, unsigned long new_len,
  708. unsigned long flags, unsigned long new_addr);
  709. extern int mprotect_fixup(struct vm_area_struct *vma,
  710. struct vm_area_struct **pprev, unsigned long start,
  711. unsigned long end, unsigned long newflags);
  712. /*
  713. * A callback you can register to apply pressure to ageable caches.
  714. *
  715. * 'shrink' is passed a count 'nr_to_scan' and a 'gfpmask'. It should
  716. * look through the least-recently-used 'nr_to_scan' entries and
  717. * attempt to free them up. It should return the number of objects
  718. * which remain in the cache. If it returns -1, it means it cannot do
  719. * any scanning at this time (eg. there is a risk of deadlock).
  720. *
  721. * The 'gfpmask' refers to the allocation we are currently trying to
  722. * fulfil.
  723. *
  724. * Note that 'shrink' will be passed nr_to_scan == 0 when the VM is
  725. * querying the cache size, so a fastpath for that case is appropriate.
  726. */
  727. struct shrinker {
  728. int (*shrink)(int nr_to_scan, gfp_t gfp_mask);
  729. int seeks; /* seeks to recreate an obj */
  730. /* These are for internal use */
  731. struct list_head list;
  732. long nr; /* objs pending delete */
  733. };
  734. #define DEFAULT_SEEKS 2 /* A good number if you don't know better. */
  735. extern void register_shrinker(struct shrinker *);
  736. extern void unregister_shrinker(struct shrinker *);
  737. int vma_wants_writenotify(struct vm_area_struct *vma);
  738. extern pte_t *get_locked_pte(struct mm_struct *mm, unsigned long addr, spinlock_t **ptl);
  739. #ifdef __PAGETABLE_PUD_FOLDED
  740. static inline int __pud_alloc(struct mm_struct *mm, pgd_t *pgd,
  741. unsigned long address)
  742. {
  743. return 0;
  744. }
  745. #else
  746. int __pud_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address);
  747. #endif
  748. #ifdef __PAGETABLE_PMD_FOLDED
  749. static inline int __pmd_alloc(struct mm_struct *mm, pud_t *pud,
  750. unsigned long address)
  751. {
  752. return 0;
  753. }
  754. #else
  755. int __pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address);
  756. #endif
  757. int __pte_alloc(struct mm_struct *mm, pmd_t *pmd, unsigned long address);
  758. int __pte_alloc_kernel(pmd_t *pmd, unsigned long address);
  759. /*
  760. * The following ifdef needed to get the 4level-fixup.h header to work.
  761. * Remove it when 4level-fixup.h has been removed.
  762. */
  763. #if defined(CONFIG_MMU) && !defined(__ARCH_HAS_4LEVEL_HACK)
  764. static inline pud_t *pud_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address)
  765. {
  766. return (unlikely(pgd_none(*pgd)) && __pud_alloc(mm, pgd, address))?
  767. NULL: pud_offset(pgd, address);
  768. }
  769. static inline pmd_t *pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address)
  770. {
  771. return (unlikely(pud_none(*pud)) && __pmd_alloc(mm, pud, address))?
  772. NULL: pmd_offset(pud, address);
  773. }
  774. #endif /* CONFIG_MMU && !__ARCH_HAS_4LEVEL_HACK */
  775. #if NR_CPUS >= CONFIG_SPLIT_PTLOCK_CPUS
  776. /*
  777. * We tuck a spinlock to guard each pagetable page into its struct page,
  778. * at page->private, with BUILD_BUG_ON to make sure that this will not
  779. * overflow into the next struct page (as it might with DEBUG_SPINLOCK).
  780. * When freeing, reset page->mapping so free_pages_check won't complain.
  781. */
  782. #define __pte_lockptr(page) &((page)->ptl)
  783. #define pte_lock_init(_page) do { \
  784. spin_lock_init(__pte_lockptr(_page)); \
  785. } while (0)
  786. #define pte_lock_deinit(page) ((page)->mapping = NULL)
  787. #define pte_lockptr(mm, pmd) ({(void)(mm); __pte_lockptr(pmd_page(*(pmd)));})
  788. #else
  789. /*
  790. * We use mm->page_table_lock to guard all pagetable pages of the mm.
  791. */
  792. #define pte_lock_init(page) do {} while (0)
  793. #define pte_lock_deinit(page) do {} while (0)
  794. #define pte_lockptr(mm, pmd) ({(void)(pmd); &(mm)->page_table_lock;})
  795. #endif /* NR_CPUS < CONFIG_SPLIT_PTLOCK_CPUS */
  796. static inline void pgtable_page_ctor(struct page *page)
  797. {
  798. pte_lock_init(page);
  799. inc_zone_page_state(page, NR_PAGETABLE);
  800. }
  801. static inline void pgtable_page_dtor(struct page *page)
  802. {
  803. pte_lock_deinit(page);
  804. dec_zone_page_state(page, NR_PAGETABLE);
  805. }
  806. #define pte_offset_map_lock(mm, pmd, address, ptlp) \
  807. ({ \
  808. spinlock_t *__ptl = pte_lockptr(mm, pmd); \
  809. pte_t *__pte = pte_offset_map(pmd, address); \
  810. *(ptlp) = __ptl; \
  811. spin_lock(__ptl); \
  812. __pte; \
  813. })
  814. #define pte_unmap_unlock(pte, ptl) do { \
  815. spin_unlock(ptl); \
  816. pte_unmap(pte); \
  817. } while (0)
  818. #define pte_alloc_map(mm, pmd, address) \
  819. ((unlikely(!pmd_present(*(pmd))) && __pte_alloc(mm, pmd, address))? \
  820. NULL: pte_offset_map(pmd, address))
  821. #define pte_alloc_map_lock(mm, pmd, address, ptlp) \
  822. ((unlikely(!pmd_present(*(pmd))) && __pte_alloc(mm, pmd, address))? \
  823. NULL: pte_offset_map_lock(mm, pmd, address, ptlp))
  824. #define pte_alloc_kernel(pmd, address) \
  825. ((unlikely(!pmd_present(*(pmd))) && __pte_alloc_kernel(pmd, address))? \
  826. NULL: pte_offset_kernel(pmd, address))
  827. extern void free_area_init(unsigned long * zones_size);
  828. extern void free_area_init_node(int nid, unsigned long * zones_size,
  829. unsigned long zone_start_pfn, unsigned long *zholes_size);
  830. #ifdef CONFIG_ARCH_POPULATES_NODE_MAP
  831. /*
  832. * With CONFIG_ARCH_POPULATES_NODE_MAP set, an architecture may initialise its
  833. * zones, allocate the backing mem_map and account for memory holes in a more
  834. * architecture independent manner. This is a substitute for creating the
  835. * zone_sizes[] and zholes_size[] arrays and passing them to
  836. * free_area_init_node()
  837. *
  838. * An architecture is expected to register range of page frames backed by
  839. * physical memory with add_active_range() before calling
  840. * free_area_init_nodes() passing in the PFN each zone ends at. At a basic
  841. * usage, an architecture is expected to do something like
  842. *
  843. * unsigned long max_zone_pfns[MAX_NR_ZONES] = {max_dma, max_normal_pfn,
  844. * max_highmem_pfn};
  845. * for_each_valid_physical_page_range()
  846. * add_active_range(node_id, start_pfn, end_pfn)
  847. * free_area_init_nodes(max_zone_pfns);
  848. *
  849. * If the architecture guarantees that there are no holes in the ranges
  850. * registered with add_active_range(), free_bootmem_active_regions()
  851. * will call free_bootmem_node() for each registered physical page range.
  852. * Similarly sparse_memory_present_with_active_regions() calls
  853. * memory_present() for each range when SPARSEMEM is enabled.
  854. *
  855. * See mm/page_alloc.c for more information on each function exposed by
  856. * CONFIG_ARCH_POPULATES_NODE_MAP
  857. */
  858. extern void free_area_init_nodes(unsigned long *max_zone_pfn);
  859. extern void add_active_range(unsigned int nid, unsigned long start_pfn,
  860. unsigned long end_pfn);
  861. extern void remove_active_range(unsigned int nid, unsigned long start_pfn,
  862. unsigned long end_pfn);
  863. extern void push_node_boundaries(unsigned int nid, unsigned long start_pfn,
  864. unsigned long end_pfn);
  865. extern void remove_all_active_ranges(void);
  866. extern unsigned long absent_pages_in_range(unsigned long start_pfn,
  867. unsigned long end_pfn);
  868. extern void get_pfn_range_for_nid(unsigned int nid,
  869. unsigned long *start_pfn, unsigned long *end_pfn);
  870. extern unsigned long find_min_pfn_with_active_regions(void);
  871. extern unsigned long find_max_pfn_with_active_regions(void);
  872. extern void free_bootmem_with_active_regions(int nid,
  873. unsigned long max_low_pfn);
  874. typedef int (*work_fn_t)(unsigned long, unsigned long, void *);
  875. extern void work_with_active_regions(int nid, work_fn_t work_fn, void *data);
  876. extern void sparse_memory_present_with_active_regions(int nid);
  877. #ifndef CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID
  878. extern int early_pfn_to_nid(unsigned long pfn);
  879. #endif /* CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID */
  880. #endif /* CONFIG_ARCH_POPULATES_NODE_MAP */
  881. extern void set_dma_reserve(unsigned long new_dma_reserve);
  882. extern void memmap_init_zone(unsigned long, int, unsigned long,
  883. unsigned long, enum memmap_context);
  884. extern void setup_per_zone_pages_min(void);
  885. extern void mem_init(void);
  886. extern void show_mem(void);
  887. extern void si_meminfo(struct sysinfo * val);
  888. extern void si_meminfo_node(struct sysinfo *val, int nid);
  889. extern int after_bootmem;
  890. #ifdef CONFIG_NUMA
  891. extern void setup_per_cpu_pageset(void);
  892. #else
  893. static inline void setup_per_cpu_pageset(void) {}
  894. #endif
  895. /* prio_tree.c */
  896. void vma_prio_tree_add(struct vm_area_struct *, struct vm_area_struct *old);
  897. void vma_prio_tree_insert(struct vm_area_struct *, struct prio_tree_root *);
  898. void vma_prio_tree_remove(struct vm_area_struct *, struct prio_tree_root *);
  899. struct vm_area_struct *vma_prio_tree_next(struct vm_area_struct *vma,
  900. struct prio_tree_iter *iter);
  901. #define vma_prio_tree_foreach(vma, iter, root, begin, end) \
  902. for (prio_tree_iter_init(iter, root, begin, end), vma = NULL; \
  903. (vma = vma_prio_tree_next(vma, iter)); )
  904. static inline void vma_nonlinear_insert(struct vm_area_struct *vma,
  905. struct list_head *list)
  906. {
  907. vma->shared.vm_set.parent = NULL;
  908. list_add_tail(&vma->shared.vm_set.list, list);
  909. }
  910. /* mmap.c */
  911. extern int __vm_enough_memory(struct mm_struct *mm, long pages, int cap_sys_admin);
  912. extern void vma_adjust(struct vm_area_struct *vma, unsigned long start,
  913. unsigned long end, pgoff_t pgoff, struct vm_area_struct *insert);
  914. extern struct vm_area_struct *vma_merge(struct mm_struct *,
  915. struct vm_area_struct *prev, unsigned long addr, unsigned long end,
  916. unsigned long vm_flags, struct anon_vma *, struct file *, pgoff_t,
  917. struct mempolicy *);
  918. extern struct anon_vma *find_mergeable_anon_vma(struct vm_area_struct *);
  919. extern int split_vma(struct mm_struct *,
  920. struct vm_area_struct *, unsigned long addr, int new_below);
  921. extern int insert_vm_struct(struct mm_struct *, struct vm_area_struct *);
  922. extern void __vma_link_rb(struct mm_struct *, struct vm_area_struct *,
  923. struct rb_node **, struct rb_node *);
  924. extern void unlink_file_vma(struct vm_area_struct *);
  925. extern struct vm_area_struct *copy_vma(struct vm_area_struct **,
  926. unsigned long addr, unsigned long len, pgoff_t pgoff);
  927. extern void exit_mmap(struct mm_struct *);
  928. #ifdef CONFIG_PROC_FS
  929. /* From fs/proc/base.c. callers must _not_ hold the mm's exe_file_lock */
  930. extern void added_exe_file_vma(struct mm_struct *mm);
  931. extern void removed_exe_file_vma(struct mm_struct *mm);
  932. #else
  933. static inline void added_exe_file_vma(struct mm_struct *mm)
  934. {}
  935. static inline void removed_exe_file_vma(struct mm_struct *mm)
  936. {}
  937. #endif /* CONFIG_PROC_FS */
  938. extern int may_expand_vm(struct mm_struct *mm, unsigned long npages);
  939. extern int install_special_mapping(struct mm_struct *mm,
  940. unsigned long addr, unsigned long len,
  941. unsigned long flags, struct page **pages);
  942. extern unsigned long get_unmapped_area(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
  943. extern unsigned long do_mmap_pgoff(struct file *file, unsigned long addr,
  944. unsigned long len, unsigned long prot,
  945. unsigned long flag, unsigned long pgoff);
  946. extern unsigned long mmap_region(struct file *file, unsigned long addr,
  947. unsigned long len, unsigned long flags,
  948. unsigned int vm_flags, unsigned long pgoff,
  949. int accountable);
  950. static inline unsigned long do_mmap(struct file *file, unsigned long addr,
  951. unsigned long len, unsigned long prot,
  952. unsigned long flag, unsigned long offset)
  953. {
  954. unsigned long ret = -EINVAL;
  955. if ((offset + PAGE_ALIGN(len)) < offset)
  956. goto out;
  957. if (!(offset & ~PAGE_MASK))
  958. ret = do_mmap_pgoff(file, addr, len, prot, flag, offset >> PAGE_SHIFT);
  959. out:
  960. return ret;
  961. }
  962. extern int do_munmap(struct mm_struct *, unsigned long, size_t);
  963. extern unsigned long do_brk(unsigned long, unsigned long);
  964. /* filemap.c */
  965. extern unsigned long page_unuse(struct page *);
  966. extern void truncate_inode_pages(struct address_space *, loff_t);
  967. extern void truncate_inode_pages_range(struct address_space *,
  968. loff_t lstart, loff_t lend);
  969. /* generic vm_area_ops exported for stackable file systems */
  970. extern int filemap_fault(struct vm_area_struct *, struct vm_fault *);
  971. /* mm/page-writeback.c */
  972. int write_one_page(struct page *page, int wait);
  973. /* readahead.c */
  974. #define VM_MAX_READAHEAD 128 /* kbytes */
  975. #define VM_MIN_READAHEAD 16 /* kbytes (includes current page) */
  976. int do_page_cache_readahead(struct address_space *mapping, struct file *filp,
  977. pgoff_t offset, unsigned long nr_to_read);
  978. int force_page_cache_readahead(struct address_space *mapping, struct file *filp,
  979. pgoff_t offset, unsigned long nr_to_read);
  980. void page_cache_sync_readahead(struct address_space *mapping,
  981. struct file_ra_state *ra,
  982. struct file *filp,
  983. pgoff_t offset,
  984. unsigned long size);
  985. void page_cache_async_readahead(struct address_space *mapping,
  986. struct file_ra_state *ra,
  987. struct file *filp,
  988. struct page *pg,
  989. pgoff_t offset,
  990. unsigned long size);
  991. unsigned long max_sane_readahead(unsigned long nr);
  992. /* Do stack extension */
  993. extern int expand_stack(struct vm_area_struct *vma, unsigned long address);
  994. #ifdef CONFIG_IA64
  995. extern int expand_upwards(struct vm_area_struct *vma, unsigned long address);
  996. #endif
  997. extern int expand_stack_downwards(struct vm_area_struct *vma,
  998. unsigned long address);
  999. /* Look up the first VMA which satisfies addr < vm_end, NULL if none. */
  1000. extern struct vm_area_struct * find_vma(struct mm_struct * mm, unsigned long addr);
  1001. extern struct vm_area_struct * find_vma_prev(struct mm_struct * mm, unsigned long addr,
  1002. struct vm_area_struct **pprev);
  1003. /* Look up the first VMA which intersects the interval start_addr..end_addr-1,
  1004. NULL if none. Assume start_addr < end_addr. */
  1005. static inline struct vm_area_struct * find_vma_intersection(struct mm_struct * mm, unsigned long start_addr, unsigned long end_addr)
  1006. {
  1007. struct vm_area_struct * vma = find_vma(mm,start_addr);
  1008. if (vma && end_addr <= vma->vm_start)
  1009. vma = NULL;
  1010. return vma;
  1011. }
  1012. static inline unsigned long vma_pages(struct vm_area_struct *vma)
  1013. {
  1014. return (vma->vm_end - vma->vm_start) >> PAGE_SHIFT;
  1015. }
  1016. pgprot_t vm_get_page_prot(unsigned long vm_flags);
  1017. struct vm_area_struct *find_extend_vma(struct mm_struct *, unsigned long addr);
  1018. int remap_pfn_range(struct vm_area_struct *, unsigned long addr,
  1019. unsigned long pfn, unsigned long size, pgprot_t);
  1020. int vm_insert_page(struct vm_area_struct *, unsigned long addr, struct page *);
  1021. int vm_insert_pfn(struct vm_area_struct *vma, unsigned long addr,
  1022. unsigned long pfn);
  1023. int vm_insert_mixed(struct vm_area_struct *vma, unsigned long addr,
  1024. unsigned long pfn);
  1025. struct page *follow_page(struct vm_area_struct *, unsigned long address,
  1026. unsigned int foll_flags);
  1027. #define FOLL_WRITE 0x01 /* check pte is writable */
  1028. #define FOLL_TOUCH 0x02 /* mark page accessed */
  1029. #define FOLL_GET 0x04 /* do get_page on page */
  1030. #define FOLL_ANON 0x08 /* give ZERO_PAGE if no pgtable */
  1031. typedef int (*pte_fn_t)(pte_t *pte, pgtable_t token, unsigned long addr,
  1032. void *data);
  1033. extern int apply_to_page_range(struct mm_struct *mm, unsigned long address,
  1034. unsigned long size, pte_fn_t fn, void *data);
  1035. #ifdef CONFIG_PROC_FS
  1036. void vm_stat_account(struct mm_struct *, unsigned long, struct file *, long);
  1037. #else
  1038. static inline void vm_stat_account(struct mm_struct *mm,
  1039. unsigned long flags, struct file *file, long pages)
  1040. {
  1041. }
  1042. #endif /* CONFIG_PROC_FS */
  1043. #ifdef CONFIG_DEBUG_PAGEALLOC
  1044. extern int debug_pagealloc_enabled;
  1045. extern void kernel_map_pages(struct page *page, int numpages, int enable);
  1046. static inline void enable_debug_pagealloc(void)
  1047. {
  1048. debug_pagealloc_enabled = 1;
  1049. }
  1050. #ifdef CONFIG_HIBERNATION
  1051. extern bool kernel_page_present(struct page *page);
  1052. #endif /* CONFIG_HIBERNATION */
  1053. #else
  1054. static inline void
  1055. kernel_map_pages(struct page *page, int numpages, int enable) {}
  1056. static inline void enable_debug_pagealloc(void)
  1057. {
  1058. }
  1059. #ifdef CONFIG_HIBERNATION
  1060. static inline bool kernel_page_present(struct page *page) { return true; }
  1061. #endif /* CONFIG_HIBERNATION */
  1062. #endif
  1063. extern struct vm_area_struct *get_gate_vma(struct task_struct *tsk);
  1064. #ifdef __HAVE_ARCH_GATE_AREA
  1065. int in_gate_area_no_task(unsigned long addr);
  1066. int in_gate_area(struct task_struct *task, unsigned long addr);
  1067. #else
  1068. int in_gate_area_no_task(unsigned long addr);
  1069. #define in_gate_area(task, addr) ({(void)task; in_gate_area_no_task(addr);})
  1070. #endif /* __HAVE_ARCH_GATE_AREA */
  1071. int drop_caches_sysctl_handler(struct ctl_table *, int, struct file *,
  1072. void __user *, size_t *, loff_t *);
  1073. unsigned long shrink_slab(unsigned long scanned, gfp_t gfp_mask,
  1074. unsigned long lru_pages);
  1075. #ifndef CONFIG_MMU
  1076. #define randomize_va_space 0
  1077. #else
  1078. extern int randomize_va_space;
  1079. #endif
  1080. const char * arch_vma_name(struct vm_area_struct *vma);
  1081. void print_vma_addr(char *prefix, unsigned long rip);
  1082. struct page *sparse_mem_map_populate(unsigned long pnum, int nid);
  1083. pgd_t *vmemmap_pgd_populate(unsigned long addr, int node);
  1084. pud_t *vmemmap_pud_populate(pgd_t *pgd, unsigned long addr, int node);
  1085. pmd_t *vmemmap_pmd_populate(pud_t *pud, unsigned long addr, int node);
  1086. pte_t *vmemmap_pte_populate(pmd_t *pmd, unsigned long addr, int node);
  1087. void *vmemmap_alloc_block(unsigned long size, int node);
  1088. void vmemmap_verify(pte_t *, int, unsigned long, unsigned long);
  1089. int vmemmap_populate_basepages(struct page *start_page,
  1090. unsigned long pages, int node);
  1091. int vmemmap_populate(struct page *start_page, unsigned long pages, int node);
  1092. void vmemmap_populate_print_last(void);
  1093. #endif /* __KERNEL__ */
  1094. #endif /* _LINUX_MM_H */