mm.h 44 KB

1234567891011121314151617181920212223242526272829303132333435363738394041424344454647484950515253545556575859606162636465666768697071727374757677787980818283848586878889909192939495969798991001011021031041051061071081091101111121131141151161171181191201211221231241251261271281291301311321331341351361371381391401411421431441451461471481491501511521531541551561571581591601611621631641651661671681691701711721731741751761771781791801811821831841851861871881891901911921931941951961971981992002012022032042052062072082092102112122132142152162172182192202212222232242252262272282292302312322332342352362372382392402412422432442452462472482492502512522532542552562572582592602612622632642652662672682692702712722732742752762772782792802812822832842852862872882892902912922932942952962972982993003013023033043053063073083093103113123133143153163173183193203213223233243253263273283293303313323333343353363373383393403413423433443453463473483493503513523533543553563573583593603613623633643653663673683693703713723733743753763773783793803813823833843853863873883893903913923933943953963973983994004014024034044054064074084094104114124134144154164174184194204214224234244254264274284294304314324334344354364374384394404414424434444454464474484494504514524534544554564574584594604614624634644654664674684694704714724734744754764774784794804814824834844854864874884894904914924934944954964974984995005015025035045055065075085095105115125135145155165175185195205215225235245255265275285295305315325335345355365375385395405415425435445455465475485495505515525535545555565575585595605615625635645655665675685695705715725735745755765775785795805815825835845855865875885895905915925935945955965975985996006016026036046056066076086096106116126136146156166176186196206216226236246256266276286296306316326336346356366376386396406416426436446456466476486496506516526536546556566576586596606616626636646656666676686696706716726736746756766776786796806816826836846856866876886896906916926936946956966976986997007017027037047057067077087097107117127137147157167177187197207217227237247257267277287297307317327337347357367377387397407417427437447457467477487497507517527537547557567577587597607617627637647657667677687697707717727737747757767777787797807817827837847857867877887897907917927937947957967977987998008018028038048058068078088098108118128138148158168178188198208218228238248258268278288298308318328338348358368378388398408418428438448458468478488498508518528538548558568578588598608618628638648658668678688698708718728738748758768778788798808818828838848858868878888898908918928938948958968978988999009019029039049059069079089099109119129139149159169179189199209219229239249259269279289299309319329339349359369379389399409419429439449459469479489499509519529539549559569579589599609619629639649659669679689699709719729739749759769779789799809819829839849859869879889899909919929939949959969979989991000100110021003100410051006100710081009101010111012101310141015101610171018101910201021102210231024102510261027102810291030103110321033103410351036103710381039104010411042104310441045104610471048104910501051105210531054105510561057105810591060106110621063106410651066106710681069107010711072107310741075107610771078107910801081108210831084108510861087108810891090109110921093109410951096109710981099110011011102110311041105110611071108110911101111111211131114111511161117111811191120112111221123112411251126112711281129113011311132113311341135113611371138113911401141114211431144114511461147114811491150115111521153115411551156115711581159116011611162116311641165116611671168116911701171117211731174117511761177117811791180118111821183118411851186118711881189119011911192119311941195119611971198119912001201120212031204120512061207120812091210121112121213121412151216121712181219122012211222122312241225122612271228122912301231123212331234123512361237123812391240124112421243124412451246124712481249125012511252125312541255125612571258125912601261126212631264126512661267126812691270127112721273127412751276127712781279128012811282128312841285
  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. * mapping from the currently active vm_flags protection bits (the
  109. * low four bits) to a page protection mask..
  110. */
  111. extern pgprot_t protection_map[16];
  112. #define FAULT_FLAG_WRITE 0x01 /* Fault was a write access */
  113. #define FAULT_FLAG_NONLINEAR 0x02 /* Fault was via a nonlinear mapping */
  114. /*
  115. * vm_fault is filled by the the pagefault handler and passed to the vma's
  116. * ->fault function. The vma's ->fault is responsible for returning a bitmask
  117. * of VM_FAULT_xxx flags that give details about how the fault was handled.
  118. *
  119. * pgoff should be used in favour of virtual_address, if possible. If pgoff
  120. * is used, one may set VM_CAN_NONLINEAR in the vma->vm_flags to get nonlinear
  121. * mapping support.
  122. */
  123. struct vm_fault {
  124. unsigned int flags; /* FAULT_FLAG_xxx flags */
  125. pgoff_t pgoff; /* Logical page offset based on vma */
  126. void __user *virtual_address; /* Faulting virtual address */
  127. struct page *page; /* ->fault handlers should return a
  128. * page here, unless VM_FAULT_NOPAGE
  129. * is set (which is also implied by
  130. * VM_FAULT_ERROR).
  131. */
  132. };
  133. /*
  134. * These are the virtual MM functions - opening of an area, closing and
  135. * unmapping it (needed to keep files on disk up-to-date etc), pointer
  136. * to the functions called when a no-page or a wp-page exception occurs.
  137. */
  138. struct vm_operations_struct {
  139. void (*open)(struct vm_area_struct * area);
  140. void (*close)(struct vm_area_struct * area);
  141. int (*fault)(struct vm_area_struct *vma, struct vm_fault *vmf);
  142. /* notification that a previously read-only page is about to become
  143. * writable, if an error is returned it will cause a SIGBUS */
  144. int (*page_mkwrite)(struct vm_area_struct *vma, struct page *page);
  145. /* called by access_process_vm when get_user_pages() fails, typically
  146. * for use by special VMAs that can switch between memory and hardware
  147. */
  148. int (*access)(struct vm_area_struct *vma, unsigned long addr,
  149. void *buf, int len, int write);
  150. #ifdef CONFIG_NUMA
  151. /*
  152. * set_policy() op must add a reference to any non-NULL @new mempolicy
  153. * to hold the policy upon return. Caller should pass NULL @new to
  154. * remove a policy and fall back to surrounding context--i.e. do not
  155. * install a MPOL_DEFAULT policy, nor the task or system default
  156. * mempolicy.
  157. */
  158. int (*set_policy)(struct vm_area_struct *vma, struct mempolicy *new);
  159. /*
  160. * get_policy() op must add reference [mpol_get()] to any policy at
  161. * (vma,addr) marked as MPOL_SHARED. The shared policy infrastructure
  162. * in mm/mempolicy.c will do this automatically.
  163. * get_policy() must NOT add a ref if the policy at (vma,addr) is not
  164. * marked as MPOL_SHARED. vma policies are protected by the mmap_sem.
  165. * If no [shared/vma] mempolicy exists at the addr, get_policy() op
  166. * must return NULL--i.e., do not "fallback" to task or system default
  167. * policy.
  168. */
  169. struct mempolicy *(*get_policy)(struct vm_area_struct *vma,
  170. unsigned long addr);
  171. int (*migrate)(struct vm_area_struct *vma, const nodemask_t *from,
  172. const nodemask_t *to, unsigned long flags);
  173. #endif
  174. };
  175. struct mmu_gather;
  176. struct inode;
  177. #define page_private(page) ((page)->private)
  178. #define set_page_private(page, v) ((page)->private = (v))
  179. /*
  180. * FIXME: take this include out, include page-flags.h in
  181. * files which need it (119 of them)
  182. */
  183. #include <linux/page-flags.h>
  184. /*
  185. * Methods to modify the page usage count.
  186. *
  187. * What counts for a page usage:
  188. * - cache mapping (page->mapping)
  189. * - private data (page->private)
  190. * - page mapped in a task's page tables, each mapping
  191. * is counted separately
  192. *
  193. * Also, many kernel routines increase the page count before a critical
  194. * routine so they can be sure the page doesn't go away from under them.
  195. */
  196. /*
  197. * Drop a ref, return true if the refcount fell to zero (the page has no users)
  198. */
  199. static inline int put_page_testzero(struct page *page)
  200. {
  201. VM_BUG_ON(atomic_read(&page->_count) == 0);
  202. return atomic_dec_and_test(&page->_count);
  203. }
  204. /*
  205. * Try to grab a ref unless the page has a refcount of zero, return false if
  206. * that is the case.
  207. */
  208. static inline int get_page_unless_zero(struct page *page)
  209. {
  210. VM_BUG_ON(PageTail(page));
  211. return atomic_inc_not_zero(&page->_count);
  212. }
  213. /* Support for virtually mapped pages */
  214. struct page *vmalloc_to_page(const void *addr);
  215. unsigned long vmalloc_to_pfn(const void *addr);
  216. /*
  217. * Determine if an address is within the vmalloc range
  218. *
  219. * On nommu, vmalloc/vfree wrap through kmalloc/kfree directly, so there
  220. * is no special casing required.
  221. */
  222. static inline int is_vmalloc_addr(const void *x)
  223. {
  224. #ifdef CONFIG_MMU
  225. unsigned long addr = (unsigned long)x;
  226. return addr >= VMALLOC_START && addr < VMALLOC_END;
  227. #else
  228. return 0;
  229. #endif
  230. }
  231. static inline struct page *compound_head(struct page *page)
  232. {
  233. if (unlikely(PageTail(page)))
  234. return page->first_page;
  235. return page;
  236. }
  237. static inline int page_count(struct page *page)
  238. {
  239. return atomic_read(&compound_head(page)->_count);
  240. }
  241. static inline void get_page(struct page *page)
  242. {
  243. page = compound_head(page);
  244. VM_BUG_ON(atomic_read(&page->_count) == 0);
  245. atomic_inc(&page->_count);
  246. }
  247. static inline struct page *virt_to_head_page(const void *x)
  248. {
  249. struct page *page = virt_to_page(x);
  250. return compound_head(page);
  251. }
  252. /*
  253. * Setup the page count before being freed into the page allocator for
  254. * the first time (boot or memory hotplug)
  255. */
  256. static inline void init_page_count(struct page *page)
  257. {
  258. atomic_set(&page->_count, 1);
  259. }
  260. void put_page(struct page *page);
  261. void put_pages_list(struct list_head *pages);
  262. void split_page(struct page *page, unsigned int order);
  263. /*
  264. * Compound pages have a destructor function. Provide a
  265. * prototype for that function and accessor functions.
  266. * These are _only_ valid on the head of a PG_compound page.
  267. */
  268. typedef void compound_page_dtor(struct page *);
  269. static inline void set_compound_page_dtor(struct page *page,
  270. compound_page_dtor *dtor)
  271. {
  272. page[1].lru.next = (void *)dtor;
  273. }
  274. static inline compound_page_dtor *get_compound_page_dtor(struct page *page)
  275. {
  276. return (compound_page_dtor *)page[1].lru.next;
  277. }
  278. static inline int compound_order(struct page *page)
  279. {
  280. if (!PageHead(page))
  281. return 0;
  282. return (unsigned long)page[1].lru.prev;
  283. }
  284. static inline void set_compound_order(struct page *page, unsigned long order)
  285. {
  286. page[1].lru.prev = (void *)order;
  287. }
  288. /*
  289. * Multiple processes may "see" the same page. E.g. for untouched
  290. * mappings of /dev/null, all processes see the same page full of
  291. * zeroes, and text pages of executables and shared libraries have
  292. * only one copy in memory, at most, normally.
  293. *
  294. * For the non-reserved pages, page_count(page) denotes a reference count.
  295. * page_count() == 0 means the page is free. page->lru is then used for
  296. * freelist management in the buddy allocator.
  297. * page_count() > 0 means the page has been allocated.
  298. *
  299. * Pages are allocated by the slab allocator in order to provide memory
  300. * to kmalloc and kmem_cache_alloc. In this case, the management of the
  301. * page, and the fields in 'struct page' are the responsibility of mm/slab.c
  302. * unless a particular usage is carefully commented. (the responsibility of
  303. * freeing the kmalloc memory is the caller's, of course).
  304. *
  305. * A page may be used by anyone else who does a __get_free_page().
  306. * In this case, page_count still tracks the references, and should only
  307. * be used through the normal accessor functions. The top bits of page->flags
  308. * and page->virtual store page management information, but all other fields
  309. * are unused and could be used privately, carefully. The management of this
  310. * page is the responsibility of the one who allocated it, and those who have
  311. * subsequently been given references to it.
  312. *
  313. * The other pages (we may call them "pagecache pages") are completely
  314. * managed by the Linux memory manager: I/O, buffers, swapping etc.
  315. * The following discussion applies only to them.
  316. *
  317. * A pagecache page contains an opaque `private' member, which belongs to the
  318. * page's address_space. Usually, this is the address of a circular list of
  319. * the page's disk buffers. PG_private must be set to tell the VM to call
  320. * into the filesystem to release these pages.
  321. *
  322. * A page may belong to an inode's memory mapping. In this case, page->mapping
  323. * is the pointer to the inode, and page->index is the file offset of the page,
  324. * in units of PAGE_CACHE_SIZE.
  325. *
  326. * If pagecache pages are not associated with an inode, they are said to be
  327. * anonymous pages. These may become associated with the swapcache, and in that
  328. * case PG_swapcache is set, and page->private is an offset into the swapcache.
  329. *
  330. * In either case (swapcache or inode backed), the pagecache itself holds one
  331. * reference to the page. Setting PG_private should also increment the
  332. * refcount. The each user mapping also has a reference to the page.
  333. *
  334. * The pagecache pages are stored in a per-mapping radix tree, which is
  335. * rooted at mapping->page_tree, and indexed by offset.
  336. * Where 2.4 and early 2.6 kernels kept dirty/clean pages in per-address_space
  337. * lists, we instead now tag pages as dirty/writeback in the radix tree.
  338. *
  339. * All pagecache pages may be subject to I/O:
  340. * - inode pages may need to be read from disk,
  341. * - inode pages which have been modified and are MAP_SHARED may need
  342. * to be written back to the inode on disk,
  343. * - anonymous pages (including MAP_PRIVATE file mappings) which have been
  344. * modified may need to be swapped out to swap space and (later) to be read
  345. * back into memory.
  346. */
  347. /*
  348. * The zone field is never updated after free_area_init_core()
  349. * sets it, so none of the operations on it need to be atomic.
  350. */
  351. /*
  352. * page->flags layout:
  353. *
  354. * There are three possibilities for how page->flags get
  355. * laid out. The first is for the normal case, without
  356. * sparsemem. The second is for sparsemem when there is
  357. * plenty of space for node and section. The last is when
  358. * we have run out of space and have to fall back to an
  359. * alternate (slower) way of determining the node.
  360. *
  361. * No sparsemem or sparsemem vmemmap: | NODE | ZONE | ... | FLAGS |
  362. * classic sparse with space for node:| SECTION | NODE | ZONE | ... | FLAGS |
  363. * classic sparse no space for node: | SECTION | ZONE | ... | FLAGS |
  364. */
  365. #if defined(CONFIG_SPARSEMEM) && !defined(CONFIG_SPARSEMEM_VMEMMAP)
  366. #define SECTIONS_WIDTH SECTIONS_SHIFT
  367. #else
  368. #define SECTIONS_WIDTH 0
  369. #endif
  370. #define ZONES_WIDTH ZONES_SHIFT
  371. #if SECTIONS_WIDTH+ZONES_WIDTH+NODES_SHIFT <= BITS_PER_LONG - NR_PAGEFLAGS
  372. #define NODES_WIDTH NODES_SHIFT
  373. #else
  374. #ifdef CONFIG_SPARSEMEM_VMEMMAP
  375. #error "Vmemmap: No space for nodes field in page flags"
  376. #endif
  377. #define NODES_WIDTH 0
  378. #endif
  379. /* Page flags: | [SECTION] | [NODE] | ZONE | ... | FLAGS | */
  380. #define SECTIONS_PGOFF ((sizeof(unsigned long)*8) - SECTIONS_WIDTH)
  381. #define NODES_PGOFF (SECTIONS_PGOFF - NODES_WIDTH)
  382. #define ZONES_PGOFF (NODES_PGOFF - ZONES_WIDTH)
  383. /*
  384. * We are going to use the flags for the page to node mapping if its in
  385. * there. This includes the case where there is no node, so it is implicit.
  386. */
  387. #if !(NODES_WIDTH > 0 || NODES_SHIFT == 0)
  388. #define NODE_NOT_IN_PAGE_FLAGS
  389. #endif
  390. #ifndef PFN_SECTION_SHIFT
  391. #define PFN_SECTION_SHIFT 0
  392. #endif
  393. /*
  394. * Define the bit shifts to access each section. For non-existant
  395. * sections we define the shift as 0; that plus a 0 mask ensures
  396. * the compiler will optimise away reference to them.
  397. */
  398. #define SECTIONS_PGSHIFT (SECTIONS_PGOFF * (SECTIONS_WIDTH != 0))
  399. #define NODES_PGSHIFT (NODES_PGOFF * (NODES_WIDTH != 0))
  400. #define ZONES_PGSHIFT (ZONES_PGOFF * (ZONES_WIDTH != 0))
  401. /* NODE:ZONE or SECTION:ZONE is used to ID a zone for the buddy allcator */
  402. #ifdef NODE_NOT_IN_PAGEFLAGS
  403. #define ZONEID_SHIFT (SECTIONS_SHIFT + ZONES_SHIFT)
  404. #define ZONEID_PGOFF ((SECTIONS_PGOFF < ZONES_PGOFF)? \
  405. SECTIONS_PGOFF : ZONES_PGOFF)
  406. #else
  407. #define ZONEID_SHIFT (NODES_SHIFT + ZONES_SHIFT)
  408. #define ZONEID_PGOFF ((NODES_PGOFF < ZONES_PGOFF)? \
  409. NODES_PGOFF : ZONES_PGOFF)
  410. #endif
  411. #define ZONEID_PGSHIFT (ZONEID_PGOFF * (ZONEID_SHIFT != 0))
  412. #if SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > BITS_PER_LONG - NR_PAGEFLAGS
  413. #error SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > BITS_PER_LONG - NR_PAGEFLAGS
  414. #endif
  415. #define ZONES_MASK ((1UL << ZONES_WIDTH) - 1)
  416. #define NODES_MASK ((1UL << NODES_WIDTH) - 1)
  417. #define SECTIONS_MASK ((1UL << SECTIONS_WIDTH) - 1)
  418. #define ZONEID_MASK ((1UL << ZONEID_SHIFT) - 1)
  419. static inline enum zone_type page_zonenum(struct page *page)
  420. {
  421. return (page->flags >> ZONES_PGSHIFT) & ZONES_MASK;
  422. }
  423. /*
  424. * The identification function is only used by the buddy allocator for
  425. * determining if two pages could be buddies. We are not really
  426. * identifying a zone since we could be using a the section number
  427. * id if we have not node id available in page flags.
  428. * We guarantee only that it will return the same value for two
  429. * combinable pages in a zone.
  430. */
  431. static inline int page_zone_id(struct page *page)
  432. {
  433. return (page->flags >> ZONEID_PGSHIFT) & ZONEID_MASK;
  434. }
  435. static inline int zone_to_nid(struct zone *zone)
  436. {
  437. #ifdef CONFIG_NUMA
  438. return zone->node;
  439. #else
  440. return 0;
  441. #endif
  442. }
  443. #ifdef NODE_NOT_IN_PAGE_FLAGS
  444. extern int page_to_nid(struct page *page);
  445. #else
  446. static inline int page_to_nid(struct page *page)
  447. {
  448. return (page->flags >> NODES_PGSHIFT) & NODES_MASK;
  449. }
  450. #endif
  451. static inline struct zone *page_zone(struct page *page)
  452. {
  453. return &NODE_DATA(page_to_nid(page))->node_zones[page_zonenum(page)];
  454. }
  455. #if defined(CONFIG_SPARSEMEM) && !defined(CONFIG_SPARSEMEM_VMEMMAP)
  456. static inline unsigned long page_to_section(struct page *page)
  457. {
  458. return (page->flags >> SECTIONS_PGSHIFT) & SECTIONS_MASK;
  459. }
  460. #endif
  461. static inline void set_page_zone(struct page *page, enum zone_type zone)
  462. {
  463. page->flags &= ~(ZONES_MASK << ZONES_PGSHIFT);
  464. page->flags |= (zone & ZONES_MASK) << ZONES_PGSHIFT;
  465. }
  466. static inline void set_page_node(struct page *page, unsigned long node)
  467. {
  468. page->flags &= ~(NODES_MASK << NODES_PGSHIFT);
  469. page->flags |= (node & NODES_MASK) << NODES_PGSHIFT;
  470. }
  471. static inline void set_page_section(struct page *page, unsigned long section)
  472. {
  473. page->flags &= ~(SECTIONS_MASK << SECTIONS_PGSHIFT);
  474. page->flags |= (section & SECTIONS_MASK) << SECTIONS_PGSHIFT;
  475. }
  476. static inline void set_page_links(struct page *page, enum zone_type zone,
  477. unsigned long node, unsigned long pfn)
  478. {
  479. set_page_zone(page, zone);
  480. set_page_node(page, node);
  481. set_page_section(page, pfn_to_section_nr(pfn));
  482. }
  483. /*
  484. * If a hint addr is less than mmap_min_addr change hint to be as
  485. * low as possible but still greater than mmap_min_addr
  486. */
  487. static inline unsigned long round_hint_to_min(unsigned long hint)
  488. {
  489. #ifdef CONFIG_SECURITY
  490. hint &= PAGE_MASK;
  491. if (((void *)hint != NULL) &&
  492. (hint < mmap_min_addr))
  493. return PAGE_ALIGN(mmap_min_addr);
  494. #endif
  495. return hint;
  496. }
  497. /*
  498. * Some inline functions in vmstat.h depend on page_zone()
  499. */
  500. #include <linux/vmstat.h>
  501. static __always_inline void *lowmem_page_address(struct page *page)
  502. {
  503. return __va(page_to_pfn(page) << PAGE_SHIFT);
  504. }
  505. #if defined(CONFIG_HIGHMEM) && !defined(WANT_PAGE_VIRTUAL)
  506. #define HASHED_PAGE_VIRTUAL
  507. #endif
  508. #if defined(WANT_PAGE_VIRTUAL)
  509. #define page_address(page) ((page)->virtual)
  510. #define set_page_address(page, address) \
  511. do { \
  512. (page)->virtual = (address); \
  513. } while(0)
  514. #define page_address_init() do { } while(0)
  515. #endif
  516. #if defined(HASHED_PAGE_VIRTUAL)
  517. void *page_address(struct page *page);
  518. void set_page_address(struct page *page, void *virtual);
  519. void page_address_init(void);
  520. #endif
  521. #if !defined(HASHED_PAGE_VIRTUAL) && !defined(WANT_PAGE_VIRTUAL)
  522. #define page_address(page) lowmem_page_address(page)
  523. #define set_page_address(page, address) do { } while(0)
  524. #define page_address_init() do { } while(0)
  525. #endif
  526. /*
  527. * On an anonymous page mapped into a user virtual memory area,
  528. * page->mapping points to its anon_vma, not to a struct address_space;
  529. * with the PAGE_MAPPING_ANON bit set to distinguish it.
  530. *
  531. * Please note that, confusingly, "page_mapping" refers to the inode
  532. * address_space which maps the page from disk; whereas "page_mapped"
  533. * refers to user virtual address space into which the page is mapped.
  534. */
  535. #define PAGE_MAPPING_ANON 1
  536. extern struct address_space swapper_space;
  537. static inline struct address_space *page_mapping(struct page *page)
  538. {
  539. struct address_space *mapping = page->mapping;
  540. VM_BUG_ON(PageSlab(page));
  541. #ifdef CONFIG_SWAP
  542. if (unlikely(PageSwapCache(page)))
  543. mapping = &swapper_space;
  544. else
  545. #endif
  546. if (unlikely((unsigned long)mapping & PAGE_MAPPING_ANON))
  547. mapping = NULL;
  548. return mapping;
  549. }
  550. static inline int PageAnon(struct page *page)
  551. {
  552. return ((unsigned long)page->mapping & PAGE_MAPPING_ANON) != 0;
  553. }
  554. /*
  555. * Return the pagecache index of the passed page. Regular pagecache pages
  556. * use ->index whereas swapcache pages use ->private
  557. */
  558. static inline pgoff_t page_index(struct page *page)
  559. {
  560. if (unlikely(PageSwapCache(page)))
  561. return page_private(page);
  562. return page->index;
  563. }
  564. /*
  565. * The atomic page->_mapcount, like _count, starts from -1:
  566. * so that transitions both from it and to it can be tracked,
  567. * using atomic_inc_and_test and atomic_add_negative(-1).
  568. */
  569. static inline void reset_page_mapcount(struct page *page)
  570. {
  571. atomic_set(&(page)->_mapcount, -1);
  572. }
  573. static inline int page_mapcount(struct page *page)
  574. {
  575. return atomic_read(&(page)->_mapcount) + 1;
  576. }
  577. /*
  578. * Return true if this page is mapped into pagetables.
  579. */
  580. static inline int page_mapped(struct page *page)
  581. {
  582. return atomic_read(&(page)->_mapcount) >= 0;
  583. }
  584. /*
  585. * Different kinds of faults, as returned by handle_mm_fault().
  586. * Used to decide whether a process gets delivered SIGBUS or
  587. * just gets major/minor fault counters bumped up.
  588. */
  589. #define VM_FAULT_MINOR 0 /* For backwards compat. Remove me quickly. */
  590. #define VM_FAULT_OOM 0x0001
  591. #define VM_FAULT_SIGBUS 0x0002
  592. #define VM_FAULT_MAJOR 0x0004
  593. #define VM_FAULT_WRITE 0x0008 /* Special case for get_user_pages */
  594. #define VM_FAULT_NOPAGE 0x0100 /* ->fault installed the pte, not return page */
  595. #define VM_FAULT_LOCKED 0x0200 /* ->fault locked the returned page */
  596. #define VM_FAULT_ERROR (VM_FAULT_OOM | VM_FAULT_SIGBUS)
  597. #define offset_in_page(p) ((unsigned long)(p) & ~PAGE_MASK)
  598. extern void show_free_areas(void);
  599. #ifdef CONFIG_SHMEM
  600. int shmem_lock(struct file *file, int lock, struct user_struct *user);
  601. #else
  602. static inline int shmem_lock(struct file *file, int lock,
  603. struct user_struct *user)
  604. {
  605. return 0;
  606. }
  607. #endif
  608. struct file *shmem_file_setup(char *name, loff_t size, unsigned long flags);
  609. int shmem_zero_setup(struct vm_area_struct *);
  610. #ifndef CONFIG_MMU
  611. extern unsigned long shmem_get_unmapped_area(struct file *file,
  612. unsigned long addr,
  613. unsigned long len,
  614. unsigned long pgoff,
  615. unsigned long flags);
  616. #endif
  617. extern int can_do_mlock(void);
  618. extern int user_shm_lock(size_t, struct user_struct *);
  619. extern void user_shm_unlock(size_t, struct user_struct *);
  620. /*
  621. * Parameter block passed down to zap_pte_range in exceptional cases.
  622. */
  623. struct zap_details {
  624. struct vm_area_struct *nonlinear_vma; /* Check page->index if set */
  625. struct address_space *check_mapping; /* Check page->mapping if set */
  626. pgoff_t first_index; /* Lowest page->index to unmap */
  627. pgoff_t last_index; /* Highest page->index to unmap */
  628. spinlock_t *i_mmap_lock; /* For unmap_mapping_range: */
  629. unsigned long truncate_count; /* Compare vm_truncate_count */
  630. };
  631. struct page *vm_normal_page(struct vm_area_struct *vma, unsigned long addr,
  632. pte_t pte);
  633. int zap_vma_ptes(struct vm_area_struct *vma, unsigned long address,
  634. unsigned long size);
  635. unsigned long zap_page_range(struct vm_area_struct *vma, unsigned long address,
  636. unsigned long size, struct zap_details *);
  637. unsigned long unmap_vmas(struct mmu_gather **tlb,
  638. struct vm_area_struct *start_vma, unsigned long start_addr,
  639. unsigned long end_addr, unsigned long *nr_accounted,
  640. struct zap_details *);
  641. /**
  642. * mm_walk - callbacks for walk_page_range
  643. * @pgd_entry: if set, called for each non-empty PGD (top-level) entry
  644. * @pud_entry: if set, called for each non-empty PUD (2nd-level) entry
  645. * @pmd_entry: if set, called for each non-empty PMD (3rd-level) entry
  646. * @pte_entry: if set, called for each non-empty PTE (4th-level) entry
  647. * @pte_hole: if set, called for each hole at all levels
  648. *
  649. * (see walk_page_range for more details)
  650. */
  651. struct mm_walk {
  652. int (*pgd_entry)(pgd_t *, unsigned long, unsigned long, struct mm_walk *);
  653. int (*pud_entry)(pud_t *, unsigned long, unsigned long, struct mm_walk *);
  654. int (*pmd_entry)(pmd_t *, unsigned long, unsigned long, struct mm_walk *);
  655. int (*pte_entry)(pte_t *, unsigned long, unsigned long, struct mm_walk *);
  656. int (*pte_hole)(unsigned long, unsigned long, struct mm_walk *);
  657. struct mm_struct *mm;
  658. void *private;
  659. };
  660. int walk_page_range(unsigned long addr, unsigned long end,
  661. struct mm_walk *walk);
  662. void free_pgd_range(struct mmu_gather *tlb, unsigned long addr,
  663. unsigned long end, unsigned long floor, unsigned long ceiling);
  664. int copy_page_range(struct mm_struct *dst, struct mm_struct *src,
  665. struct vm_area_struct *vma);
  666. void unmap_mapping_range(struct address_space *mapping,
  667. loff_t const holebegin, loff_t const holelen, int even_cows);
  668. int generic_access_phys(struct vm_area_struct *vma, unsigned long addr,
  669. void *buf, int len, int write);
  670. static inline void unmap_shared_mapping_range(struct address_space *mapping,
  671. loff_t const holebegin, loff_t const holelen)
  672. {
  673. unmap_mapping_range(mapping, holebegin, holelen, 0);
  674. }
  675. extern int vmtruncate(struct inode * inode, loff_t offset);
  676. extern int vmtruncate_range(struct inode * inode, loff_t offset, loff_t end);
  677. #ifdef CONFIG_MMU
  678. extern int handle_mm_fault(struct mm_struct *mm, struct vm_area_struct *vma,
  679. unsigned long address, int write_access);
  680. #else
  681. static inline int handle_mm_fault(struct mm_struct *mm,
  682. struct vm_area_struct *vma, unsigned long address,
  683. int write_access)
  684. {
  685. /* should never happen if there's no MMU */
  686. BUG();
  687. return VM_FAULT_SIGBUS;
  688. }
  689. #endif
  690. extern int make_pages_present(unsigned long addr, unsigned long end);
  691. extern int access_process_vm(struct task_struct *tsk, unsigned long addr, void *buf, int len, int write);
  692. int get_user_pages(struct task_struct *tsk, struct mm_struct *mm, unsigned long start,
  693. int len, int write, int force, struct page **pages, struct vm_area_struct **vmas);
  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. * get_user_pages_fast provides equivalent functionality to get_user_pages,
  714. * operating on current and current->mm (force=0 and doesn't return any vmas).
  715. *
  716. * get_user_pages_fast may take mmap_sem and page tables, so no assumptions
  717. * can be made about locking. get_user_pages_fast is to be implemented in a
  718. * way that is advantageous (vs get_user_pages()) when the user memory area is
  719. * already faulted in and present in ptes. However if the pages have to be
  720. * faulted in, it may turn out to be slightly slower).
  721. */
  722. int get_user_pages_fast(unsigned long start, int nr_pages, int write,
  723. struct page **pages);
  724. /*
  725. * A callback you can register to apply pressure to ageable caches.
  726. *
  727. * 'shrink' is passed a count 'nr_to_scan' and a 'gfpmask'. It should
  728. * look through the least-recently-used 'nr_to_scan' entries and
  729. * attempt to free them up. It should return the number of objects
  730. * which remain in the cache. If it returns -1, it means it cannot do
  731. * any scanning at this time (eg. there is a risk of deadlock).
  732. *
  733. * The 'gfpmask' refers to the allocation we are currently trying to
  734. * fulfil.
  735. *
  736. * Note that 'shrink' will be passed nr_to_scan == 0 when the VM is
  737. * querying the cache size, so a fastpath for that case is appropriate.
  738. */
  739. struct shrinker {
  740. int (*shrink)(int nr_to_scan, gfp_t gfp_mask);
  741. int seeks; /* seeks to recreate an obj */
  742. /* These are for internal use */
  743. struct list_head list;
  744. long nr; /* objs pending delete */
  745. };
  746. #define DEFAULT_SEEKS 2 /* A good number if you don't know better. */
  747. extern void register_shrinker(struct shrinker *);
  748. extern void unregister_shrinker(struct shrinker *);
  749. int vma_wants_writenotify(struct vm_area_struct *vma);
  750. extern pte_t *get_locked_pte(struct mm_struct *mm, unsigned long addr, spinlock_t **ptl);
  751. #ifdef __PAGETABLE_PUD_FOLDED
  752. static inline int __pud_alloc(struct mm_struct *mm, pgd_t *pgd,
  753. unsigned long address)
  754. {
  755. return 0;
  756. }
  757. #else
  758. int __pud_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address);
  759. #endif
  760. #ifdef __PAGETABLE_PMD_FOLDED
  761. static inline int __pmd_alloc(struct mm_struct *mm, pud_t *pud,
  762. unsigned long address)
  763. {
  764. return 0;
  765. }
  766. #else
  767. int __pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address);
  768. #endif
  769. int __pte_alloc(struct mm_struct *mm, pmd_t *pmd, unsigned long address);
  770. int __pte_alloc_kernel(pmd_t *pmd, unsigned long address);
  771. /*
  772. * The following ifdef needed to get the 4level-fixup.h header to work.
  773. * Remove it when 4level-fixup.h has been removed.
  774. */
  775. #if defined(CONFIG_MMU) && !defined(__ARCH_HAS_4LEVEL_HACK)
  776. static inline pud_t *pud_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address)
  777. {
  778. return (unlikely(pgd_none(*pgd)) && __pud_alloc(mm, pgd, address))?
  779. NULL: pud_offset(pgd, address);
  780. }
  781. static inline pmd_t *pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address)
  782. {
  783. return (unlikely(pud_none(*pud)) && __pmd_alloc(mm, pud, address))?
  784. NULL: pmd_offset(pud, address);
  785. }
  786. #endif /* CONFIG_MMU && !__ARCH_HAS_4LEVEL_HACK */
  787. #if USE_SPLIT_PTLOCKS
  788. /*
  789. * We tuck a spinlock to guard each pagetable page into its struct page,
  790. * at page->private, with BUILD_BUG_ON to make sure that this will not
  791. * overflow into the next struct page (as it might with DEBUG_SPINLOCK).
  792. * When freeing, reset page->mapping so free_pages_check won't complain.
  793. */
  794. #define __pte_lockptr(page) &((page)->ptl)
  795. #define pte_lock_init(_page) do { \
  796. spin_lock_init(__pte_lockptr(_page)); \
  797. } while (0)
  798. #define pte_lock_deinit(page) ((page)->mapping = NULL)
  799. #define pte_lockptr(mm, pmd) ({(void)(mm); __pte_lockptr(pmd_page(*(pmd)));})
  800. #else /* !USE_SPLIT_PTLOCKS */
  801. /*
  802. * We use mm->page_table_lock to guard all pagetable pages of the mm.
  803. */
  804. #define pte_lock_init(page) do {} while (0)
  805. #define pte_lock_deinit(page) do {} while (0)
  806. #define pte_lockptr(mm, pmd) ({(void)(pmd); &(mm)->page_table_lock;})
  807. #endif /* USE_SPLIT_PTLOCKS */
  808. static inline void pgtable_page_ctor(struct page *page)
  809. {
  810. pte_lock_init(page);
  811. inc_zone_page_state(page, NR_PAGETABLE);
  812. }
  813. static inline void pgtable_page_dtor(struct page *page)
  814. {
  815. pte_lock_deinit(page);
  816. dec_zone_page_state(page, NR_PAGETABLE);
  817. }
  818. #define pte_offset_map_lock(mm, pmd, address, ptlp) \
  819. ({ \
  820. spinlock_t *__ptl = pte_lockptr(mm, pmd); \
  821. pte_t *__pte = pte_offset_map(pmd, address); \
  822. *(ptlp) = __ptl; \
  823. spin_lock(__ptl); \
  824. __pte; \
  825. })
  826. #define pte_unmap_unlock(pte, ptl) do { \
  827. spin_unlock(ptl); \
  828. pte_unmap(pte); \
  829. } while (0)
  830. #define pte_alloc_map(mm, pmd, address) \
  831. ((unlikely(!pmd_present(*(pmd))) && __pte_alloc(mm, pmd, address))? \
  832. NULL: pte_offset_map(pmd, address))
  833. #define pte_alloc_map_lock(mm, pmd, address, ptlp) \
  834. ((unlikely(!pmd_present(*(pmd))) && __pte_alloc(mm, pmd, address))? \
  835. NULL: pte_offset_map_lock(mm, pmd, address, ptlp))
  836. #define pte_alloc_kernel(pmd, address) \
  837. ((unlikely(!pmd_present(*(pmd))) && __pte_alloc_kernel(pmd, address))? \
  838. NULL: pte_offset_kernel(pmd, address))
  839. extern void free_area_init(unsigned long * zones_size);
  840. extern void free_area_init_node(int nid, unsigned long * zones_size,
  841. unsigned long zone_start_pfn, unsigned long *zholes_size);
  842. #ifdef CONFIG_ARCH_POPULATES_NODE_MAP
  843. /*
  844. * With CONFIG_ARCH_POPULATES_NODE_MAP set, an architecture may initialise its
  845. * zones, allocate the backing mem_map and account for memory holes in a more
  846. * architecture independent manner. This is a substitute for creating the
  847. * zone_sizes[] and zholes_size[] arrays and passing them to
  848. * free_area_init_node()
  849. *
  850. * An architecture is expected to register range of page frames backed by
  851. * physical memory with add_active_range() before calling
  852. * free_area_init_nodes() passing in the PFN each zone ends at. At a basic
  853. * usage, an architecture is expected to do something like
  854. *
  855. * unsigned long max_zone_pfns[MAX_NR_ZONES] = {max_dma, max_normal_pfn,
  856. * max_highmem_pfn};
  857. * for_each_valid_physical_page_range()
  858. * add_active_range(node_id, start_pfn, end_pfn)
  859. * free_area_init_nodes(max_zone_pfns);
  860. *
  861. * If the architecture guarantees that there are no holes in the ranges
  862. * registered with add_active_range(), free_bootmem_active_regions()
  863. * will call free_bootmem_node() for each registered physical page range.
  864. * Similarly sparse_memory_present_with_active_regions() calls
  865. * memory_present() for each range when SPARSEMEM is enabled.
  866. *
  867. * See mm/page_alloc.c for more information on each function exposed by
  868. * CONFIG_ARCH_POPULATES_NODE_MAP
  869. */
  870. extern void free_area_init_nodes(unsigned long *max_zone_pfn);
  871. extern void add_active_range(unsigned int nid, unsigned long start_pfn,
  872. unsigned long end_pfn);
  873. extern void remove_active_range(unsigned int nid, unsigned long start_pfn,
  874. unsigned long end_pfn);
  875. extern void push_node_boundaries(unsigned int nid, unsigned long start_pfn,
  876. unsigned long end_pfn);
  877. extern void remove_all_active_ranges(void);
  878. extern unsigned long absent_pages_in_range(unsigned long start_pfn,
  879. unsigned long end_pfn);
  880. extern void get_pfn_range_for_nid(unsigned int nid,
  881. unsigned long *start_pfn, unsigned long *end_pfn);
  882. extern unsigned long find_min_pfn_with_active_regions(void);
  883. extern void free_bootmem_with_active_regions(int nid,
  884. unsigned long max_low_pfn);
  885. typedef int (*work_fn_t)(unsigned long, unsigned long, void *);
  886. extern void work_with_active_regions(int nid, work_fn_t work_fn, void *data);
  887. extern void sparse_memory_present_with_active_regions(int nid);
  888. #ifndef CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID
  889. extern int early_pfn_to_nid(unsigned long pfn);
  890. #endif /* CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID */
  891. #endif /* CONFIG_ARCH_POPULATES_NODE_MAP */
  892. extern void set_dma_reserve(unsigned long new_dma_reserve);
  893. extern void memmap_init_zone(unsigned long, int, unsigned long,
  894. unsigned long, enum memmap_context);
  895. extern void setup_per_zone_pages_min(void);
  896. extern void mem_init(void);
  897. extern void show_mem(void);
  898. extern void si_meminfo(struct sysinfo * val);
  899. extern void si_meminfo_node(struct sysinfo *val, int nid);
  900. extern int after_bootmem;
  901. #ifdef CONFIG_NUMA
  902. extern void setup_per_cpu_pageset(void);
  903. #else
  904. static inline void setup_per_cpu_pageset(void) {}
  905. #endif
  906. /* prio_tree.c */
  907. void vma_prio_tree_add(struct vm_area_struct *, struct vm_area_struct *old);
  908. void vma_prio_tree_insert(struct vm_area_struct *, struct prio_tree_root *);
  909. void vma_prio_tree_remove(struct vm_area_struct *, struct prio_tree_root *);
  910. struct vm_area_struct *vma_prio_tree_next(struct vm_area_struct *vma,
  911. struct prio_tree_iter *iter);
  912. #define vma_prio_tree_foreach(vma, iter, root, begin, end) \
  913. for (prio_tree_iter_init(iter, root, begin, end), vma = NULL; \
  914. (vma = vma_prio_tree_next(vma, iter)); )
  915. static inline void vma_nonlinear_insert(struct vm_area_struct *vma,
  916. struct list_head *list)
  917. {
  918. vma->shared.vm_set.parent = NULL;
  919. list_add_tail(&vma->shared.vm_set.list, list);
  920. }
  921. /* mmap.c */
  922. extern int __vm_enough_memory(struct mm_struct *mm, long pages, int cap_sys_admin);
  923. extern void vma_adjust(struct vm_area_struct *vma, unsigned long start,
  924. unsigned long end, pgoff_t pgoff, struct vm_area_struct *insert);
  925. extern struct vm_area_struct *vma_merge(struct mm_struct *,
  926. struct vm_area_struct *prev, unsigned long addr, unsigned long end,
  927. unsigned long vm_flags, struct anon_vma *, struct file *, pgoff_t,
  928. struct mempolicy *);
  929. extern struct anon_vma *find_mergeable_anon_vma(struct vm_area_struct *);
  930. extern int split_vma(struct mm_struct *,
  931. struct vm_area_struct *, unsigned long addr, int new_below);
  932. extern int insert_vm_struct(struct mm_struct *, struct vm_area_struct *);
  933. extern void __vma_link_rb(struct mm_struct *, struct vm_area_struct *,
  934. struct rb_node **, struct rb_node *);
  935. extern void unlink_file_vma(struct vm_area_struct *);
  936. extern struct vm_area_struct *copy_vma(struct vm_area_struct **,
  937. unsigned long addr, unsigned long len, pgoff_t pgoff);
  938. extern void exit_mmap(struct mm_struct *);
  939. extern int mm_take_all_locks(struct mm_struct *mm);
  940. extern void mm_drop_all_locks(struct mm_struct *mm);
  941. #ifdef CONFIG_PROC_FS
  942. /* From fs/proc/base.c. callers must _not_ hold the mm's exe_file_lock */
  943. extern void added_exe_file_vma(struct mm_struct *mm);
  944. extern void removed_exe_file_vma(struct mm_struct *mm);
  945. #else
  946. static inline void added_exe_file_vma(struct mm_struct *mm)
  947. {}
  948. static inline void removed_exe_file_vma(struct mm_struct *mm)
  949. {}
  950. #endif /* CONFIG_PROC_FS */
  951. extern int may_expand_vm(struct mm_struct *mm, unsigned long npages);
  952. extern int install_special_mapping(struct mm_struct *mm,
  953. unsigned long addr, unsigned long len,
  954. unsigned long flags, struct page **pages);
  955. extern unsigned long get_unmapped_area(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
  956. extern unsigned long do_mmap_pgoff(struct file *file, unsigned long addr,
  957. unsigned long len, unsigned long prot,
  958. unsigned long flag, unsigned long pgoff);
  959. extern unsigned long mmap_region(struct file *file, unsigned long addr,
  960. unsigned long len, unsigned long flags,
  961. unsigned int vm_flags, unsigned long pgoff,
  962. int accountable);
  963. static inline unsigned long do_mmap(struct file *file, unsigned long addr,
  964. unsigned long len, unsigned long prot,
  965. unsigned long flag, unsigned long offset)
  966. {
  967. unsigned long ret = -EINVAL;
  968. if ((offset + PAGE_ALIGN(len)) < offset)
  969. goto out;
  970. if (!(offset & ~PAGE_MASK))
  971. ret = do_mmap_pgoff(file, addr, len, prot, flag, offset >> PAGE_SHIFT);
  972. out:
  973. return ret;
  974. }
  975. extern int do_munmap(struct mm_struct *, unsigned long, size_t);
  976. extern unsigned long do_brk(unsigned long, unsigned long);
  977. /* filemap.c */
  978. extern unsigned long page_unuse(struct page *);
  979. extern void truncate_inode_pages(struct address_space *, loff_t);
  980. extern void truncate_inode_pages_range(struct address_space *,
  981. loff_t lstart, loff_t lend);
  982. /* generic vm_area_ops exported for stackable file systems */
  983. extern int filemap_fault(struct vm_area_struct *, struct vm_fault *);
  984. /* mm/page-writeback.c */
  985. int write_one_page(struct page *page, int wait);
  986. /* readahead.c */
  987. #define VM_MAX_READAHEAD 128 /* kbytes */
  988. #define VM_MIN_READAHEAD 16 /* kbytes (includes current page) */
  989. int do_page_cache_readahead(struct address_space *mapping, struct file *filp,
  990. pgoff_t offset, unsigned long nr_to_read);
  991. int force_page_cache_readahead(struct address_space *mapping, struct file *filp,
  992. pgoff_t offset, unsigned long nr_to_read);
  993. void page_cache_sync_readahead(struct address_space *mapping,
  994. struct file_ra_state *ra,
  995. struct file *filp,
  996. pgoff_t offset,
  997. unsigned long size);
  998. void page_cache_async_readahead(struct address_space *mapping,
  999. struct file_ra_state *ra,
  1000. struct file *filp,
  1001. struct page *pg,
  1002. pgoff_t offset,
  1003. unsigned long size);
  1004. unsigned long max_sane_readahead(unsigned long nr);
  1005. /* Do stack extension */
  1006. extern int expand_stack(struct vm_area_struct *vma, unsigned long address);
  1007. #ifdef CONFIG_IA64
  1008. extern int expand_upwards(struct vm_area_struct *vma, unsigned long address);
  1009. #endif
  1010. extern int expand_stack_downwards(struct vm_area_struct *vma,
  1011. unsigned long address);
  1012. /* Look up the first VMA which satisfies addr < vm_end, NULL if none. */
  1013. extern struct vm_area_struct * find_vma(struct mm_struct * mm, unsigned long addr);
  1014. extern struct vm_area_struct * find_vma_prev(struct mm_struct * mm, unsigned long addr,
  1015. struct vm_area_struct **pprev);
  1016. /* Look up the first VMA which intersects the interval start_addr..end_addr-1,
  1017. NULL if none. Assume start_addr < end_addr. */
  1018. static inline struct vm_area_struct * find_vma_intersection(struct mm_struct * mm, unsigned long start_addr, unsigned long end_addr)
  1019. {
  1020. struct vm_area_struct * vma = find_vma(mm,start_addr);
  1021. if (vma && end_addr <= vma->vm_start)
  1022. vma = NULL;
  1023. return vma;
  1024. }
  1025. static inline unsigned long vma_pages(struct vm_area_struct *vma)
  1026. {
  1027. return (vma->vm_end - vma->vm_start) >> PAGE_SHIFT;
  1028. }
  1029. pgprot_t vm_get_page_prot(unsigned long vm_flags);
  1030. struct vm_area_struct *find_extend_vma(struct mm_struct *, unsigned long addr);
  1031. int remap_pfn_range(struct vm_area_struct *, unsigned long addr,
  1032. unsigned long pfn, unsigned long size, pgprot_t);
  1033. int vm_insert_page(struct vm_area_struct *, unsigned long addr, struct page *);
  1034. int vm_insert_pfn(struct vm_area_struct *vma, unsigned long addr,
  1035. unsigned long pfn);
  1036. int vm_insert_mixed(struct vm_area_struct *vma, unsigned long addr,
  1037. unsigned long pfn);
  1038. struct page *follow_page(struct vm_area_struct *, unsigned long address,
  1039. unsigned int foll_flags);
  1040. #define FOLL_WRITE 0x01 /* check pte is writable */
  1041. #define FOLL_TOUCH 0x02 /* mark page accessed */
  1042. #define FOLL_GET 0x04 /* do get_page on page */
  1043. #define FOLL_ANON 0x08 /* give ZERO_PAGE if no pgtable */
  1044. typedef int (*pte_fn_t)(pte_t *pte, pgtable_t token, unsigned long addr,
  1045. void *data);
  1046. extern int apply_to_page_range(struct mm_struct *mm, unsigned long address,
  1047. unsigned long size, pte_fn_t fn, void *data);
  1048. #ifdef CONFIG_PROC_FS
  1049. void vm_stat_account(struct mm_struct *, unsigned long, struct file *, long);
  1050. #else
  1051. static inline void vm_stat_account(struct mm_struct *mm,
  1052. unsigned long flags, struct file *file, long pages)
  1053. {
  1054. }
  1055. #endif /* CONFIG_PROC_FS */
  1056. #ifdef CONFIG_DEBUG_PAGEALLOC
  1057. extern int debug_pagealloc_enabled;
  1058. extern void kernel_map_pages(struct page *page, int numpages, int enable);
  1059. static inline void enable_debug_pagealloc(void)
  1060. {
  1061. debug_pagealloc_enabled = 1;
  1062. }
  1063. #ifdef CONFIG_HIBERNATION
  1064. extern bool kernel_page_present(struct page *page);
  1065. #endif /* CONFIG_HIBERNATION */
  1066. #else
  1067. static inline void
  1068. kernel_map_pages(struct page *page, int numpages, int enable) {}
  1069. static inline void enable_debug_pagealloc(void)
  1070. {
  1071. }
  1072. #ifdef CONFIG_HIBERNATION
  1073. static inline bool kernel_page_present(struct page *page) { return true; }
  1074. #endif /* CONFIG_HIBERNATION */
  1075. #endif
  1076. extern struct vm_area_struct *get_gate_vma(struct task_struct *tsk);
  1077. #ifdef __HAVE_ARCH_GATE_AREA
  1078. int in_gate_area_no_task(unsigned long addr);
  1079. int in_gate_area(struct task_struct *task, unsigned long addr);
  1080. #else
  1081. int in_gate_area_no_task(unsigned long addr);
  1082. #define in_gate_area(task, addr) ({(void)task; in_gate_area_no_task(addr);})
  1083. #endif /* __HAVE_ARCH_GATE_AREA */
  1084. int drop_caches_sysctl_handler(struct ctl_table *, int, struct file *,
  1085. void __user *, size_t *, loff_t *);
  1086. unsigned long shrink_slab(unsigned long scanned, gfp_t gfp_mask,
  1087. unsigned long lru_pages);
  1088. #ifndef CONFIG_MMU
  1089. #define randomize_va_space 0
  1090. #else
  1091. extern int randomize_va_space;
  1092. #endif
  1093. const char * arch_vma_name(struct vm_area_struct *vma);
  1094. void print_vma_addr(char *prefix, unsigned long rip);
  1095. struct page *sparse_mem_map_populate(unsigned long pnum, int nid);
  1096. pgd_t *vmemmap_pgd_populate(unsigned long addr, int node);
  1097. pud_t *vmemmap_pud_populate(pgd_t *pgd, unsigned long addr, int node);
  1098. pmd_t *vmemmap_pmd_populate(pud_t *pud, unsigned long addr, int node);
  1099. pte_t *vmemmap_pte_populate(pmd_t *pmd, unsigned long addr, int node);
  1100. void *vmemmap_alloc_block(unsigned long size, int node);
  1101. void vmemmap_verify(pte_t *, int, unsigned long, unsigned long);
  1102. int vmemmap_populate_basepages(struct page *start_page,
  1103. unsigned long pages, int node);
  1104. int vmemmap_populate(struct page *start_page, unsigned long pages, int node);
  1105. void vmemmap_populate_print_last(void);
  1106. #endif /* __KERNEL__ */
  1107. #endif /* _LINUX_MM_H */