mm.h 45 KB

1234567891011121314151617181920212223242526272829303132333435363738394041424344454647484950515253545556575859606162636465666768697071727374757677787980818283848586878889909192939495969798991001011021031041051061071081091101111121131141151161171181191201211221231241251261271281291301311321331341351361371381391401411421431441451461471481491501511521531541551561571581591601611621631641651661671681691701711721731741751761771781791801811821831841851861871881891901911921931941951961971981992002012022032042052062072082092102112122132142152162172182192202212222232242252262272282292302312322332342352362372382392402412422432442452462472482492502512522532542552562572582592602612622632642652662672682692702712722732742752762772782792802812822832842852862872882892902912922932942952962972982993003013023033043053063073083093103113123133143153163173183193203213223233243253263273283293303313323333343353363373383393403413423433443453463473483493503513523533543553563573583593603613623633643653663673683693703713723733743753763773783793803813823833843853863873883893903913923933943953963973983994004014024034044054064074084094104114124134144154164174184194204214224234244254264274284294304314324334344354364374384394404414424434444454464474484494504514524534544554564574584594604614624634644654664674684694704714724734744754764774784794804814824834844854864874884894904914924934944954964974984995005015025035045055065075085095105115125135145155165175185195205215225235245255265275285295305315325335345355365375385395405415425435445455465475485495505515525535545555565575585595605615625635645655665675685695705715725735745755765775785795805815825835845855865875885895905915925935945955965975985996006016026036046056066076086096106116126136146156166176186196206216226236246256266276286296306316326336346356366376386396406416426436446456466476486496506516526536546556566576586596606616626636646656666676686696706716726736746756766776786796806816826836846856866876886896906916926936946956966976986997007017027037047057067077087097107117127137147157167177187197207217227237247257267277287297307317327337347357367377387397407417427437447457467477487497507517527537547557567577587597607617627637647657667677687697707717727737747757767777787797807817827837847857867877887897907917927937947957967977987998008018028038048058068078088098108118128138148158168178188198208218228238248258268278288298308318328338348358368378388398408418428438448458468478488498508518528538548558568578588598608618628638648658668678688698708718728738748758768778788798808818828838848858868878888898908918928938948958968978988999009019029039049059069079089099109119129139149159169179189199209219229239249259269279289299309319329339349359369379389399409419429439449459469479489499509519529539549559569579589599609619629639649659669679689699709719729739749759769779789799809819829839849859869879889899909919929939949959969979989991000100110021003100410051006100710081009101010111012101310141015101610171018101910201021102210231024102510261027102810291030103110321033103410351036103710381039104010411042104310441045104610471048104910501051105210531054105510561057105810591060106110621063106410651066106710681069107010711072107310741075107610771078107910801081108210831084108510861087108810891090109110921093109410951096109710981099110011011102110311041105110611071108110911101111111211131114111511161117111811191120112111221123112411251126112711281129113011311132113311341135113611371138113911401141114211431144114511461147114811491150115111521153115411551156115711581159116011611162116311641165116611671168116911701171117211731174117511761177117811791180118111821183118411851186118711881189119011911192119311941195119611971198119912001201120212031204120512061207120812091210121112121213121412151216121712181219122012211222122312241225122612271228122912301231123212331234123512361237123812391240124112421243124412451246124712481249125012511252125312541255125612571258125912601261126212631264126512661267126812691270127112721273127412751276127712781279128012811282128312841285128612871288128912901291129212931294129512961297129812991300130113021303130413051306130713081309131013111312131313141315131613171318131913201321132213231324132513261327
  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. struct rlimit;
  19. #ifndef CONFIG_DISCONTIGMEM /* Don't use mapnrs, do it properly */
  20. extern unsigned long max_mapnr;
  21. #endif
  22. extern unsigned long num_physpages;
  23. extern void * high_memory;
  24. extern int page_cluster;
  25. #ifdef CONFIG_SYSCTL
  26. extern int sysctl_legacy_va_layout;
  27. #else
  28. #define sysctl_legacy_va_layout 0
  29. #endif
  30. extern unsigned long mmap_min_addr;
  31. #include <asm/page.h>
  32. #include <asm/pgtable.h>
  33. #include <asm/processor.h>
  34. #define nth_page(page,n) pfn_to_page(page_to_pfn((page)) + (n))
  35. /* to align the pointer to the (next) page boundary */
  36. #define PAGE_ALIGN(addr) ALIGN(addr, PAGE_SIZE)
  37. /*
  38. * Linux kernel virtual memory manager primitives.
  39. * The idea being to have a "virtual" mm in the same way
  40. * we have a virtual fs - giving a cleaner interface to the
  41. * mm details, and allowing different kinds of memory mappings
  42. * (from shared memory to executable loading to arbitrary
  43. * mmap() functions).
  44. */
  45. extern struct kmem_cache *vm_area_cachep;
  46. #ifndef CONFIG_MMU
  47. extern struct rb_root nommu_region_tree;
  48. extern struct rw_semaphore nommu_region_sem;
  49. extern unsigned int kobjsize(const void *objp);
  50. #endif
  51. /*
  52. * vm_flags in vm_area_struct, see mm_types.h.
  53. */
  54. #define VM_READ 0x00000001 /* currently active flags */
  55. #define VM_WRITE 0x00000002
  56. #define VM_EXEC 0x00000004
  57. #define VM_SHARED 0x00000008
  58. /* mprotect() hardcodes VM_MAYREAD >> 4 == VM_READ, and so for r/w/x bits. */
  59. #define VM_MAYREAD 0x00000010 /* limits for mprotect() etc */
  60. #define VM_MAYWRITE 0x00000020
  61. #define VM_MAYEXEC 0x00000040
  62. #define VM_MAYSHARE 0x00000080
  63. #define VM_GROWSDOWN 0x00000100 /* general info on the segment */
  64. #define VM_GROWSUP 0x00000200
  65. #define VM_PFNMAP 0x00000400 /* Page-ranges managed without "struct page", just pure PFN */
  66. #define VM_DENYWRITE 0x00000800 /* ETXTBSY on write attempts.. */
  67. #define VM_EXECUTABLE 0x00001000
  68. #define VM_LOCKED 0x00002000
  69. #define VM_IO 0x00004000 /* Memory mapped I/O or similar */
  70. /* Used by sys_madvise() */
  71. #define VM_SEQ_READ 0x00008000 /* App will access data sequentially */
  72. #define VM_RAND_READ 0x00010000 /* App will not benefit from clustered reads */
  73. #define VM_DONTCOPY 0x00020000 /* Do not copy this vma on fork */
  74. #define VM_DONTEXPAND 0x00040000 /* Cannot expand with mremap() */
  75. #define VM_RESERVED 0x00080000 /* Count as reserved_vm like IO */
  76. #define VM_ACCOUNT 0x00100000 /* Is a VM accounted object */
  77. #define VM_NORESERVE 0x00200000 /* should the VM suppress accounting */
  78. #define VM_HUGETLB 0x00400000 /* Huge TLB Page VM */
  79. #define VM_NONLINEAR 0x00800000 /* Is non-linear (remap_file_pages) */
  80. #define VM_MAPPED_COPY 0x01000000 /* T if mapped copy of data (nommu mmap) */
  81. #define VM_INSERTPAGE 0x02000000 /* The vma has had "vm_insert_page()" done on it */
  82. #define VM_ALWAYSDUMP 0x04000000 /* Always include in core dumps */
  83. #define VM_CAN_NONLINEAR 0x08000000 /* Has ->fault & does nonlinear pages */
  84. #define VM_MIXEDMAP 0x10000000 /* Can contain "struct page" and pure PFN pages */
  85. #define VM_SAO 0x20000000 /* Strong Access Ordering (powerpc) */
  86. #define VM_PFN_AT_MMAP 0x40000000 /* PFNMAP vma that is fully mapped at mmap time */
  87. #ifndef VM_STACK_DEFAULT_FLAGS /* arch can override this */
  88. #define VM_STACK_DEFAULT_FLAGS VM_DATA_DEFAULT_FLAGS
  89. #endif
  90. #ifdef CONFIG_STACK_GROWSUP
  91. #define VM_STACK_FLAGS (VM_GROWSUP | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT)
  92. #else
  93. #define VM_STACK_FLAGS (VM_GROWSDOWN | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT)
  94. #endif
  95. #define VM_READHINTMASK (VM_SEQ_READ | VM_RAND_READ)
  96. #define VM_ClearReadHint(v) (v)->vm_flags &= ~VM_READHINTMASK
  97. #define VM_NormalReadHint(v) (!((v)->vm_flags & VM_READHINTMASK))
  98. #define VM_SequentialReadHint(v) ((v)->vm_flags & VM_SEQ_READ)
  99. #define VM_RandomReadHint(v) ((v)->vm_flags & VM_RAND_READ)
  100. /*
  101. * special vmas that are non-mergable, non-mlock()able
  102. */
  103. #define VM_SPECIAL (VM_IO | VM_DONTEXPAND | VM_RESERVED | VM_PFNMAP)
  104. /*
  105. * mapping from the currently active vm_flags protection bits (the
  106. * low four bits) to a page protection mask..
  107. */
  108. extern pgprot_t protection_map[16];
  109. #define FAULT_FLAG_WRITE 0x01 /* Fault was a write access */
  110. #define FAULT_FLAG_NONLINEAR 0x02 /* Fault was via a nonlinear mapping */
  111. #define FAULT_FLAG_MKWRITE 0x04 /* Fault was mkwrite of existing pte */
  112. /*
  113. * This interface is used by x86 PAT code to identify a pfn mapping that is
  114. * linear over entire vma. This is to optimize PAT code that deals with
  115. * marking the physical region with a particular prot. This is not for generic
  116. * mm use. Note also that this check will not work if the pfn mapping is
  117. * linear for a vma starting at physical address 0. In which case PAT code
  118. * falls back to slow path of reserving physical range page by page.
  119. */
  120. static inline int is_linear_pfn_mapping(struct vm_area_struct *vma)
  121. {
  122. return (vma->vm_flags & VM_PFN_AT_MMAP);
  123. }
  124. static inline int is_pfn_mapping(struct vm_area_struct *vma)
  125. {
  126. return (vma->vm_flags & VM_PFNMAP);
  127. }
  128. /*
  129. * vm_fault is filled by the the pagefault handler and passed to the vma's
  130. * ->fault function. The vma's ->fault is responsible for returning a bitmask
  131. * of VM_FAULT_xxx flags that give details about how the fault was handled.
  132. *
  133. * pgoff should be used in favour of virtual_address, if possible. If pgoff
  134. * is used, one may set VM_CAN_NONLINEAR in the vma->vm_flags to get nonlinear
  135. * mapping support.
  136. */
  137. struct vm_fault {
  138. unsigned int flags; /* FAULT_FLAG_xxx flags */
  139. pgoff_t pgoff; /* Logical page offset based on vma */
  140. void __user *virtual_address; /* Faulting virtual address */
  141. struct page *page; /* ->fault handlers should return a
  142. * page here, unless VM_FAULT_NOPAGE
  143. * is set (which is also implied by
  144. * VM_FAULT_ERROR).
  145. */
  146. };
  147. /*
  148. * These are the virtual MM functions - opening of an area, closing and
  149. * unmapping it (needed to keep files on disk up-to-date etc), pointer
  150. * to the functions called when a no-page or a wp-page exception occurs.
  151. */
  152. struct vm_operations_struct {
  153. void (*open)(struct vm_area_struct * area);
  154. void (*close)(struct vm_area_struct * area);
  155. int (*fault)(struct vm_area_struct *vma, struct vm_fault *vmf);
  156. /* notification that a previously read-only page is about to become
  157. * writable, if an error is returned it will cause a SIGBUS */
  158. int (*page_mkwrite)(struct vm_area_struct *vma, struct vm_fault *vmf);
  159. /* called by access_process_vm when get_user_pages() fails, typically
  160. * for use by special VMAs that can switch between memory and hardware
  161. */
  162. int (*access)(struct vm_area_struct *vma, unsigned long addr,
  163. void *buf, int len, int write);
  164. #ifdef CONFIG_NUMA
  165. /*
  166. * set_policy() op must add a reference to any non-NULL @new mempolicy
  167. * to hold the policy upon return. Caller should pass NULL @new to
  168. * remove a policy and fall back to surrounding context--i.e. do not
  169. * install a MPOL_DEFAULT policy, nor the task or system default
  170. * mempolicy.
  171. */
  172. int (*set_policy)(struct vm_area_struct *vma, struct mempolicy *new);
  173. /*
  174. * get_policy() op must add reference [mpol_get()] to any policy at
  175. * (vma,addr) marked as MPOL_SHARED. The shared policy infrastructure
  176. * in mm/mempolicy.c will do this automatically.
  177. * get_policy() must NOT add a ref if the policy at (vma,addr) is not
  178. * marked as MPOL_SHARED. vma policies are protected by the mmap_sem.
  179. * If no [shared/vma] mempolicy exists at the addr, get_policy() op
  180. * must return NULL--i.e., do not "fallback" to task or system default
  181. * policy.
  182. */
  183. struct mempolicy *(*get_policy)(struct vm_area_struct *vma,
  184. unsigned long addr);
  185. int (*migrate)(struct vm_area_struct *vma, const nodemask_t *from,
  186. const nodemask_t *to, unsigned long flags);
  187. #endif
  188. };
  189. struct mmu_gather;
  190. struct inode;
  191. #define page_private(page) ((page)->private)
  192. #define set_page_private(page, v) ((page)->private = (v))
  193. /*
  194. * FIXME: take this include out, include page-flags.h in
  195. * files which need it (119 of them)
  196. */
  197. #include <linux/page-flags.h>
  198. /*
  199. * Methods to modify the page usage count.
  200. *
  201. * What counts for a page usage:
  202. * - cache mapping (page->mapping)
  203. * - private data (page->private)
  204. * - page mapped in a task's page tables, each mapping
  205. * is counted separately
  206. *
  207. * Also, many kernel routines increase the page count before a critical
  208. * routine so they can be sure the page doesn't go away from under them.
  209. */
  210. /*
  211. * Drop a ref, return true if the refcount fell to zero (the page has no users)
  212. */
  213. static inline int put_page_testzero(struct page *page)
  214. {
  215. VM_BUG_ON(atomic_read(&page->_count) == 0);
  216. return atomic_dec_and_test(&page->_count);
  217. }
  218. /*
  219. * Try to grab a ref unless the page has a refcount of zero, return false if
  220. * that is the case.
  221. */
  222. static inline int get_page_unless_zero(struct page *page)
  223. {
  224. return atomic_inc_not_zero(&page->_count);
  225. }
  226. /* Support for virtually mapped pages */
  227. struct page *vmalloc_to_page(const void *addr);
  228. unsigned long vmalloc_to_pfn(const void *addr);
  229. /*
  230. * Determine if an address is within the vmalloc range
  231. *
  232. * On nommu, vmalloc/vfree wrap through kmalloc/kfree directly, so there
  233. * is no special casing required.
  234. */
  235. static inline int is_vmalloc_addr(const void *x)
  236. {
  237. #ifdef CONFIG_MMU
  238. unsigned long addr = (unsigned long)x;
  239. return addr >= VMALLOC_START && addr < VMALLOC_END;
  240. #else
  241. return 0;
  242. #endif
  243. }
  244. static inline struct page *compound_head(struct page *page)
  245. {
  246. if (unlikely(PageTail(page)))
  247. return page->first_page;
  248. return page;
  249. }
  250. static inline int page_count(struct page *page)
  251. {
  252. return atomic_read(&compound_head(page)->_count);
  253. }
  254. static inline void get_page(struct page *page)
  255. {
  256. page = compound_head(page);
  257. VM_BUG_ON(atomic_read(&page->_count) == 0);
  258. atomic_inc(&page->_count);
  259. }
  260. static inline struct page *virt_to_head_page(const void *x)
  261. {
  262. struct page *page = virt_to_page(x);
  263. return compound_head(page);
  264. }
  265. /*
  266. * Setup the page count before being freed into the page allocator for
  267. * the first time (boot or memory hotplug)
  268. */
  269. static inline void init_page_count(struct page *page)
  270. {
  271. atomic_set(&page->_count, 1);
  272. }
  273. void put_page(struct page *page);
  274. void put_pages_list(struct list_head *pages);
  275. void split_page(struct page *page, unsigned int order);
  276. /*
  277. * Compound pages have a destructor function. Provide a
  278. * prototype for that function and accessor functions.
  279. * These are _only_ valid on the head of a PG_compound page.
  280. */
  281. typedef void compound_page_dtor(struct page *);
  282. static inline void set_compound_page_dtor(struct page *page,
  283. compound_page_dtor *dtor)
  284. {
  285. page[1].lru.next = (void *)dtor;
  286. }
  287. static inline compound_page_dtor *get_compound_page_dtor(struct page *page)
  288. {
  289. return (compound_page_dtor *)page[1].lru.next;
  290. }
  291. static inline int compound_order(struct page *page)
  292. {
  293. if (!PageHead(page))
  294. return 0;
  295. return (unsigned long)page[1].lru.prev;
  296. }
  297. static inline void set_compound_order(struct page *page, unsigned long order)
  298. {
  299. page[1].lru.prev = (void *)order;
  300. }
  301. /*
  302. * Multiple processes may "see" the same page. E.g. for untouched
  303. * mappings of /dev/null, all processes see the same page full of
  304. * zeroes, and text pages of executables and shared libraries have
  305. * only one copy in memory, at most, normally.
  306. *
  307. * For the non-reserved pages, page_count(page) denotes a reference count.
  308. * page_count() == 0 means the page is free. page->lru is then used for
  309. * freelist management in the buddy allocator.
  310. * page_count() > 0 means the page has been allocated.
  311. *
  312. * Pages are allocated by the slab allocator in order to provide memory
  313. * to kmalloc and kmem_cache_alloc. In this case, the management of the
  314. * page, and the fields in 'struct page' are the responsibility of mm/slab.c
  315. * unless a particular usage is carefully commented. (the responsibility of
  316. * freeing the kmalloc memory is the caller's, of course).
  317. *
  318. * A page may be used by anyone else who does a __get_free_page().
  319. * In this case, page_count still tracks the references, and should only
  320. * be used through the normal accessor functions. The top bits of page->flags
  321. * and page->virtual store page management information, but all other fields
  322. * are unused and could be used privately, carefully. The management of this
  323. * page is the responsibility of the one who allocated it, and those who have
  324. * subsequently been given references to it.
  325. *
  326. * The other pages (we may call them "pagecache pages") are completely
  327. * managed by the Linux memory manager: I/O, buffers, swapping etc.
  328. * The following discussion applies only to them.
  329. *
  330. * A pagecache page contains an opaque `private' member, which belongs to the
  331. * page's address_space. Usually, this is the address of a circular list of
  332. * the page's disk buffers. PG_private must be set to tell the VM to call
  333. * into the filesystem to release these pages.
  334. *
  335. * A page may belong to an inode's memory mapping. In this case, page->mapping
  336. * is the pointer to the inode, and page->index is the file offset of the page,
  337. * in units of PAGE_CACHE_SIZE.
  338. *
  339. * If pagecache pages are not associated with an inode, they are said to be
  340. * anonymous pages. These may become associated with the swapcache, and in that
  341. * case PG_swapcache is set, and page->private is an offset into the swapcache.
  342. *
  343. * In either case (swapcache or inode backed), the pagecache itself holds one
  344. * reference to the page. Setting PG_private should also increment the
  345. * refcount. The each user mapping also has a reference to the page.
  346. *
  347. * The pagecache pages are stored in a per-mapping radix tree, which is
  348. * rooted at mapping->page_tree, and indexed by offset.
  349. * Where 2.4 and early 2.6 kernels kept dirty/clean pages in per-address_space
  350. * lists, we instead now tag pages as dirty/writeback in the radix tree.
  351. *
  352. * All pagecache pages may be subject to I/O:
  353. * - inode pages may need to be read from disk,
  354. * - inode pages which have been modified and are MAP_SHARED may need
  355. * to be written back to the inode on disk,
  356. * - anonymous pages (including MAP_PRIVATE file mappings) which have been
  357. * modified may need to be swapped out to swap space and (later) to be read
  358. * back into memory.
  359. */
  360. /*
  361. * The zone field is never updated after free_area_init_core()
  362. * sets it, so none of the operations on it need to be atomic.
  363. */
  364. /*
  365. * page->flags layout:
  366. *
  367. * There are three possibilities for how page->flags get
  368. * laid out. The first is for the normal case, without
  369. * sparsemem. The second is for sparsemem when there is
  370. * plenty of space for node and section. The last is when
  371. * we have run out of space and have to fall back to an
  372. * alternate (slower) way of determining the node.
  373. *
  374. * No sparsemem or sparsemem vmemmap: | NODE | ZONE | ... | FLAGS |
  375. * classic sparse with space for node:| SECTION | NODE | ZONE | ... | FLAGS |
  376. * classic sparse no space for node: | SECTION | ZONE | ... | FLAGS |
  377. */
  378. #if defined(CONFIG_SPARSEMEM) && !defined(CONFIG_SPARSEMEM_VMEMMAP)
  379. #define SECTIONS_WIDTH SECTIONS_SHIFT
  380. #else
  381. #define SECTIONS_WIDTH 0
  382. #endif
  383. #define ZONES_WIDTH ZONES_SHIFT
  384. #if SECTIONS_WIDTH+ZONES_WIDTH+NODES_SHIFT <= BITS_PER_LONG - NR_PAGEFLAGS
  385. #define NODES_WIDTH NODES_SHIFT
  386. #else
  387. #ifdef CONFIG_SPARSEMEM_VMEMMAP
  388. #error "Vmemmap: No space for nodes field in page flags"
  389. #endif
  390. #define NODES_WIDTH 0
  391. #endif
  392. /* Page flags: | [SECTION] | [NODE] | ZONE | ... | FLAGS | */
  393. #define SECTIONS_PGOFF ((sizeof(unsigned long)*8) - SECTIONS_WIDTH)
  394. #define NODES_PGOFF (SECTIONS_PGOFF - NODES_WIDTH)
  395. #define ZONES_PGOFF (NODES_PGOFF - ZONES_WIDTH)
  396. /*
  397. * We are going to use the flags for the page to node mapping if its in
  398. * there. This includes the case where there is no node, so it is implicit.
  399. */
  400. #if !(NODES_WIDTH > 0 || NODES_SHIFT == 0)
  401. #define NODE_NOT_IN_PAGE_FLAGS
  402. #endif
  403. #ifndef PFN_SECTION_SHIFT
  404. #define PFN_SECTION_SHIFT 0
  405. #endif
  406. /*
  407. * Define the bit shifts to access each section. For non-existant
  408. * sections we define the shift as 0; that plus a 0 mask ensures
  409. * the compiler will optimise away reference to them.
  410. */
  411. #define SECTIONS_PGSHIFT (SECTIONS_PGOFF * (SECTIONS_WIDTH != 0))
  412. #define NODES_PGSHIFT (NODES_PGOFF * (NODES_WIDTH != 0))
  413. #define ZONES_PGSHIFT (ZONES_PGOFF * (ZONES_WIDTH != 0))
  414. /* NODE:ZONE or SECTION:ZONE is used to ID a zone for the buddy allcator */
  415. #ifdef NODE_NOT_IN_PAGEFLAGS
  416. #define ZONEID_SHIFT (SECTIONS_SHIFT + ZONES_SHIFT)
  417. #define ZONEID_PGOFF ((SECTIONS_PGOFF < ZONES_PGOFF)? \
  418. SECTIONS_PGOFF : ZONES_PGOFF)
  419. #else
  420. #define ZONEID_SHIFT (NODES_SHIFT + ZONES_SHIFT)
  421. #define ZONEID_PGOFF ((NODES_PGOFF < ZONES_PGOFF)? \
  422. NODES_PGOFF : ZONES_PGOFF)
  423. #endif
  424. #define ZONEID_PGSHIFT (ZONEID_PGOFF * (ZONEID_SHIFT != 0))
  425. #if SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > BITS_PER_LONG - NR_PAGEFLAGS
  426. #error SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > BITS_PER_LONG - NR_PAGEFLAGS
  427. #endif
  428. #define ZONES_MASK ((1UL << ZONES_WIDTH) - 1)
  429. #define NODES_MASK ((1UL << NODES_WIDTH) - 1)
  430. #define SECTIONS_MASK ((1UL << SECTIONS_WIDTH) - 1)
  431. #define ZONEID_MASK ((1UL << ZONEID_SHIFT) - 1)
  432. static inline enum zone_type page_zonenum(struct page *page)
  433. {
  434. return (page->flags >> ZONES_PGSHIFT) & ZONES_MASK;
  435. }
  436. /*
  437. * The identification function is only used by the buddy allocator for
  438. * determining if two pages could be buddies. We are not really
  439. * identifying a zone since we could be using a the section number
  440. * id if we have not node id available in page flags.
  441. * We guarantee only that it will return the same value for two
  442. * combinable pages in a zone.
  443. */
  444. static inline int page_zone_id(struct page *page)
  445. {
  446. return (page->flags >> ZONEID_PGSHIFT) & ZONEID_MASK;
  447. }
  448. static inline int zone_to_nid(struct zone *zone)
  449. {
  450. #ifdef CONFIG_NUMA
  451. return zone->node;
  452. #else
  453. return 0;
  454. #endif
  455. }
  456. #ifdef NODE_NOT_IN_PAGE_FLAGS
  457. extern int page_to_nid(struct page *page);
  458. #else
  459. static inline int page_to_nid(struct page *page)
  460. {
  461. return (page->flags >> NODES_PGSHIFT) & NODES_MASK;
  462. }
  463. #endif
  464. static inline struct zone *page_zone(struct page *page)
  465. {
  466. return &NODE_DATA(page_to_nid(page))->node_zones[page_zonenum(page)];
  467. }
  468. #if defined(CONFIG_SPARSEMEM) && !defined(CONFIG_SPARSEMEM_VMEMMAP)
  469. static inline unsigned long page_to_section(struct page *page)
  470. {
  471. return (page->flags >> SECTIONS_PGSHIFT) & SECTIONS_MASK;
  472. }
  473. #endif
  474. static inline void set_page_zone(struct page *page, enum zone_type zone)
  475. {
  476. page->flags &= ~(ZONES_MASK << ZONES_PGSHIFT);
  477. page->flags |= (zone & ZONES_MASK) << ZONES_PGSHIFT;
  478. }
  479. static inline void set_page_node(struct page *page, unsigned long node)
  480. {
  481. page->flags &= ~(NODES_MASK << NODES_PGSHIFT);
  482. page->flags |= (node & NODES_MASK) << NODES_PGSHIFT;
  483. }
  484. static inline void set_page_section(struct page *page, unsigned long section)
  485. {
  486. page->flags &= ~(SECTIONS_MASK << SECTIONS_PGSHIFT);
  487. page->flags |= (section & SECTIONS_MASK) << SECTIONS_PGSHIFT;
  488. }
  489. static inline void set_page_links(struct page *page, enum zone_type zone,
  490. unsigned long node, unsigned long pfn)
  491. {
  492. set_page_zone(page, zone);
  493. set_page_node(page, node);
  494. set_page_section(page, pfn_to_section_nr(pfn));
  495. }
  496. /*
  497. * If a hint addr is less than mmap_min_addr change hint to be as
  498. * low as possible but still greater than mmap_min_addr
  499. */
  500. static inline unsigned long round_hint_to_min(unsigned long hint)
  501. {
  502. #ifdef CONFIG_SECURITY
  503. hint &= PAGE_MASK;
  504. if (((void *)hint != NULL) &&
  505. (hint < mmap_min_addr))
  506. return PAGE_ALIGN(mmap_min_addr);
  507. #endif
  508. return hint;
  509. }
  510. /*
  511. * Some inline functions in vmstat.h depend on page_zone()
  512. */
  513. #include <linux/vmstat.h>
  514. static __always_inline void *lowmem_page_address(struct page *page)
  515. {
  516. return __va(page_to_pfn(page) << PAGE_SHIFT);
  517. }
  518. #if defined(CONFIG_HIGHMEM) && !defined(WANT_PAGE_VIRTUAL)
  519. #define HASHED_PAGE_VIRTUAL
  520. #endif
  521. #if defined(WANT_PAGE_VIRTUAL)
  522. #define page_address(page) ((page)->virtual)
  523. #define set_page_address(page, address) \
  524. do { \
  525. (page)->virtual = (address); \
  526. } while(0)
  527. #define page_address_init() do { } while(0)
  528. #endif
  529. #if defined(HASHED_PAGE_VIRTUAL)
  530. void *page_address(struct page *page);
  531. void set_page_address(struct page *page, void *virtual);
  532. void page_address_init(void);
  533. #endif
  534. #if !defined(HASHED_PAGE_VIRTUAL) && !defined(WANT_PAGE_VIRTUAL)
  535. #define page_address(page) lowmem_page_address(page)
  536. #define set_page_address(page, address) do { } while(0)
  537. #define page_address_init() do { } while(0)
  538. #endif
  539. /*
  540. * On an anonymous page mapped into a user virtual memory area,
  541. * page->mapping points to its anon_vma, not to a struct address_space;
  542. * with the PAGE_MAPPING_ANON bit set to distinguish it.
  543. *
  544. * Please note that, confusingly, "page_mapping" refers to the inode
  545. * address_space which maps the page from disk; whereas "page_mapped"
  546. * refers to user virtual address space into which the page is mapped.
  547. */
  548. #define PAGE_MAPPING_ANON 1
  549. extern struct address_space swapper_space;
  550. static inline struct address_space *page_mapping(struct page *page)
  551. {
  552. struct address_space *mapping = page->mapping;
  553. VM_BUG_ON(PageSlab(page));
  554. #ifdef CONFIG_SWAP
  555. if (unlikely(PageSwapCache(page)))
  556. mapping = &swapper_space;
  557. else
  558. #endif
  559. if (unlikely((unsigned long)mapping & PAGE_MAPPING_ANON))
  560. mapping = NULL;
  561. return mapping;
  562. }
  563. static inline int PageAnon(struct page *page)
  564. {
  565. return ((unsigned long)page->mapping & PAGE_MAPPING_ANON) != 0;
  566. }
  567. /*
  568. * Return the pagecache index of the passed page. Regular pagecache pages
  569. * use ->index whereas swapcache pages use ->private
  570. */
  571. static inline pgoff_t page_index(struct page *page)
  572. {
  573. if (unlikely(PageSwapCache(page)))
  574. return page_private(page);
  575. return page->index;
  576. }
  577. /*
  578. * The atomic page->_mapcount, like _count, starts from -1:
  579. * so that transitions both from it and to it can be tracked,
  580. * using atomic_inc_and_test and atomic_add_negative(-1).
  581. */
  582. static inline void reset_page_mapcount(struct page *page)
  583. {
  584. atomic_set(&(page)->_mapcount, -1);
  585. }
  586. static inline int page_mapcount(struct page *page)
  587. {
  588. return atomic_read(&(page)->_mapcount) + 1;
  589. }
  590. /*
  591. * Return true if this page is mapped into pagetables.
  592. */
  593. static inline int page_mapped(struct page *page)
  594. {
  595. return atomic_read(&(page)->_mapcount) >= 0;
  596. }
  597. /*
  598. * Different kinds of faults, as returned by handle_mm_fault().
  599. * Used to decide whether a process gets delivered SIGBUS or
  600. * just gets major/minor fault counters bumped up.
  601. */
  602. #define VM_FAULT_MINOR 0 /* For backwards compat. Remove me quickly. */
  603. #define VM_FAULT_OOM 0x0001
  604. #define VM_FAULT_SIGBUS 0x0002
  605. #define VM_FAULT_MAJOR 0x0004
  606. #define VM_FAULT_WRITE 0x0008 /* Special case for get_user_pages */
  607. #define VM_FAULT_NOPAGE 0x0100 /* ->fault installed the pte, not return page */
  608. #define VM_FAULT_LOCKED 0x0200 /* ->fault locked the returned page */
  609. #define VM_FAULT_ERROR (VM_FAULT_OOM | VM_FAULT_SIGBUS)
  610. /*
  611. * Can be called by the pagefault handler when it gets a VM_FAULT_OOM.
  612. */
  613. extern void pagefault_out_of_memory(void);
  614. #define offset_in_page(p) ((unsigned long)(p) & ~PAGE_MASK)
  615. extern void show_free_areas(void);
  616. #ifdef CONFIG_SHMEM
  617. extern int shmem_lock(struct file *file, int lock, struct user_struct *user);
  618. #else
  619. static inline int shmem_lock(struct file *file, int lock,
  620. struct user_struct *user)
  621. {
  622. return 0;
  623. }
  624. #endif
  625. struct file *shmem_file_setup(char *name, loff_t size, unsigned long flags);
  626. int shmem_zero_setup(struct vm_area_struct *);
  627. #ifndef CONFIG_MMU
  628. extern unsigned long shmem_get_unmapped_area(struct file *file,
  629. unsigned long addr,
  630. unsigned long len,
  631. unsigned long pgoff,
  632. unsigned long flags);
  633. #endif
  634. extern int can_do_mlock(void);
  635. extern int user_shm_lock(size_t, struct user_struct *);
  636. extern void user_shm_unlock(size_t, struct user_struct *);
  637. /*
  638. * Parameter block passed down to zap_pte_range in exceptional cases.
  639. */
  640. struct zap_details {
  641. struct vm_area_struct *nonlinear_vma; /* Check page->index if set */
  642. struct address_space *check_mapping; /* Check page->mapping if set */
  643. pgoff_t first_index; /* Lowest page->index to unmap */
  644. pgoff_t last_index; /* Highest page->index to unmap */
  645. spinlock_t *i_mmap_lock; /* For unmap_mapping_range: */
  646. unsigned long truncate_count; /* Compare vm_truncate_count */
  647. };
  648. struct page *vm_normal_page(struct vm_area_struct *vma, unsigned long addr,
  649. pte_t pte);
  650. int zap_vma_ptes(struct vm_area_struct *vma, unsigned long address,
  651. unsigned long size);
  652. unsigned long zap_page_range(struct vm_area_struct *vma, unsigned long address,
  653. unsigned long size, struct zap_details *);
  654. unsigned long unmap_vmas(struct mmu_gather **tlb,
  655. struct vm_area_struct *start_vma, unsigned long start_addr,
  656. unsigned long end_addr, unsigned long *nr_accounted,
  657. struct zap_details *);
  658. /**
  659. * mm_walk - callbacks for walk_page_range
  660. * @pgd_entry: if set, called for each non-empty PGD (top-level) entry
  661. * @pud_entry: if set, called for each non-empty PUD (2nd-level) entry
  662. * @pmd_entry: if set, called for each non-empty PMD (3rd-level) entry
  663. * @pte_entry: if set, called for each non-empty PTE (4th-level) entry
  664. * @pte_hole: if set, called for each hole at all levels
  665. *
  666. * (see walk_page_range for more details)
  667. */
  668. struct mm_walk {
  669. int (*pgd_entry)(pgd_t *, unsigned long, unsigned long, struct mm_walk *);
  670. int (*pud_entry)(pud_t *, unsigned long, unsigned long, struct mm_walk *);
  671. int (*pmd_entry)(pmd_t *, unsigned long, unsigned long, struct mm_walk *);
  672. int (*pte_entry)(pte_t *, unsigned long, unsigned long, struct mm_walk *);
  673. int (*pte_hole)(unsigned long, unsigned long, struct mm_walk *);
  674. struct mm_struct *mm;
  675. void *private;
  676. };
  677. int walk_page_range(unsigned long addr, unsigned long end,
  678. struct mm_walk *walk);
  679. void free_pgd_range(struct mmu_gather *tlb, unsigned long addr,
  680. unsigned long end, unsigned long floor, unsigned long ceiling);
  681. int copy_page_range(struct mm_struct *dst, struct mm_struct *src,
  682. struct vm_area_struct *vma);
  683. void unmap_mapping_range(struct address_space *mapping,
  684. loff_t const holebegin, loff_t const holelen, int even_cows);
  685. int follow_phys(struct vm_area_struct *vma, unsigned long address,
  686. unsigned int flags, unsigned long *prot, resource_size_t *phys);
  687. int generic_access_phys(struct vm_area_struct *vma, unsigned long addr,
  688. void *buf, int len, int write);
  689. static inline void unmap_shared_mapping_range(struct address_space *mapping,
  690. loff_t const holebegin, loff_t const holelen)
  691. {
  692. unmap_mapping_range(mapping, holebegin, holelen, 0);
  693. }
  694. extern int vmtruncate(struct inode * inode, loff_t offset);
  695. extern int vmtruncate_range(struct inode * inode, loff_t offset, loff_t end);
  696. #ifdef CONFIG_MMU
  697. extern int handle_mm_fault(struct mm_struct *mm, struct vm_area_struct *vma,
  698. unsigned long address, int write_access);
  699. #else
  700. static inline int handle_mm_fault(struct mm_struct *mm,
  701. struct vm_area_struct *vma, unsigned long address,
  702. int write_access)
  703. {
  704. /* should never happen if there's no MMU */
  705. BUG();
  706. return VM_FAULT_SIGBUS;
  707. }
  708. #endif
  709. extern int make_pages_present(unsigned long addr, unsigned long end);
  710. extern int access_process_vm(struct task_struct *tsk, unsigned long addr, void *buf, int len, int write);
  711. int get_user_pages(struct task_struct *tsk, struct mm_struct *mm, unsigned long start,
  712. int len, int write, int force, struct page **pages, struct vm_area_struct **vmas);
  713. extern int try_to_release_page(struct page * page, gfp_t gfp_mask);
  714. extern void do_invalidatepage(struct page *page, unsigned long offset);
  715. int __set_page_dirty_nobuffers(struct page *page);
  716. int __set_page_dirty_no_writeback(struct page *page);
  717. int redirty_page_for_writepage(struct writeback_control *wbc,
  718. struct page *page);
  719. void account_page_dirtied(struct page *page, struct address_space *mapping);
  720. int set_page_dirty(struct page *page);
  721. int set_page_dirty_lock(struct page *page);
  722. int clear_page_dirty_for_io(struct page *page);
  723. extern unsigned long move_page_tables(struct vm_area_struct *vma,
  724. unsigned long old_addr, struct vm_area_struct *new_vma,
  725. unsigned long new_addr, unsigned long len);
  726. extern unsigned long do_mremap(unsigned long addr,
  727. unsigned long old_len, unsigned long new_len,
  728. unsigned long flags, unsigned long new_addr);
  729. extern int mprotect_fixup(struct vm_area_struct *vma,
  730. struct vm_area_struct **pprev, unsigned long start,
  731. unsigned long end, unsigned long newflags);
  732. /*
  733. * get_user_pages_fast provides equivalent functionality to get_user_pages,
  734. * operating on current and current->mm (force=0 and doesn't return any vmas).
  735. *
  736. * get_user_pages_fast may take mmap_sem and page tables, so no assumptions
  737. * can be made about locking. get_user_pages_fast is to be implemented in a
  738. * way that is advantageous (vs get_user_pages()) when the user memory area is
  739. * already faulted in and present in ptes. However if the pages have to be
  740. * faulted in, it may turn out to be slightly slower).
  741. */
  742. int get_user_pages_fast(unsigned long start, int nr_pages, int write,
  743. struct page **pages);
  744. /*
  745. * A callback you can register to apply pressure to ageable caches.
  746. *
  747. * 'shrink' is passed a count 'nr_to_scan' and a 'gfpmask'. It should
  748. * look through the least-recently-used 'nr_to_scan' entries and
  749. * attempt to free them up. It should return the number of objects
  750. * which remain in the cache. If it returns -1, it means it cannot do
  751. * any scanning at this time (eg. there is a risk of deadlock).
  752. *
  753. * The 'gfpmask' refers to the allocation we are currently trying to
  754. * fulfil.
  755. *
  756. * Note that 'shrink' will be passed nr_to_scan == 0 when the VM is
  757. * querying the cache size, so a fastpath for that case is appropriate.
  758. */
  759. struct shrinker {
  760. int (*shrink)(int nr_to_scan, gfp_t gfp_mask);
  761. int seeks; /* seeks to recreate an obj */
  762. /* These are for internal use */
  763. struct list_head list;
  764. long nr; /* objs pending delete */
  765. };
  766. #define DEFAULT_SEEKS 2 /* A good number if you don't know better. */
  767. extern void register_shrinker(struct shrinker *);
  768. extern void unregister_shrinker(struct shrinker *);
  769. int vma_wants_writenotify(struct vm_area_struct *vma);
  770. extern pte_t *get_locked_pte(struct mm_struct *mm, unsigned long addr, spinlock_t **ptl);
  771. #ifdef __PAGETABLE_PUD_FOLDED
  772. static inline int __pud_alloc(struct mm_struct *mm, pgd_t *pgd,
  773. unsigned long address)
  774. {
  775. return 0;
  776. }
  777. #else
  778. int __pud_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address);
  779. #endif
  780. #ifdef __PAGETABLE_PMD_FOLDED
  781. static inline int __pmd_alloc(struct mm_struct *mm, pud_t *pud,
  782. unsigned long address)
  783. {
  784. return 0;
  785. }
  786. #else
  787. int __pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address);
  788. #endif
  789. int __pte_alloc(struct mm_struct *mm, pmd_t *pmd, unsigned long address);
  790. int __pte_alloc_kernel(pmd_t *pmd, unsigned long address);
  791. /*
  792. * The following ifdef needed to get the 4level-fixup.h header to work.
  793. * Remove it when 4level-fixup.h has been removed.
  794. */
  795. #if defined(CONFIG_MMU) && !defined(__ARCH_HAS_4LEVEL_HACK)
  796. static inline pud_t *pud_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address)
  797. {
  798. return (unlikely(pgd_none(*pgd)) && __pud_alloc(mm, pgd, address))?
  799. NULL: pud_offset(pgd, address);
  800. }
  801. static inline pmd_t *pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address)
  802. {
  803. return (unlikely(pud_none(*pud)) && __pmd_alloc(mm, pud, address))?
  804. NULL: pmd_offset(pud, address);
  805. }
  806. #endif /* CONFIG_MMU && !__ARCH_HAS_4LEVEL_HACK */
  807. #if USE_SPLIT_PTLOCKS
  808. /*
  809. * We tuck a spinlock to guard each pagetable page into its struct page,
  810. * at page->private, with BUILD_BUG_ON to make sure that this will not
  811. * overflow into the next struct page (as it might with DEBUG_SPINLOCK).
  812. * When freeing, reset page->mapping so free_pages_check won't complain.
  813. */
  814. #define __pte_lockptr(page) &((page)->ptl)
  815. #define pte_lock_init(_page) do { \
  816. spin_lock_init(__pte_lockptr(_page)); \
  817. } while (0)
  818. #define pte_lock_deinit(page) ((page)->mapping = NULL)
  819. #define pte_lockptr(mm, pmd) ({(void)(mm); __pte_lockptr(pmd_page(*(pmd)));})
  820. #else /* !USE_SPLIT_PTLOCKS */
  821. /*
  822. * We use mm->page_table_lock to guard all pagetable pages of the mm.
  823. */
  824. #define pte_lock_init(page) do {} while (0)
  825. #define pte_lock_deinit(page) do {} while (0)
  826. #define pte_lockptr(mm, pmd) ({(void)(pmd); &(mm)->page_table_lock;})
  827. #endif /* USE_SPLIT_PTLOCKS */
  828. static inline void pgtable_page_ctor(struct page *page)
  829. {
  830. pte_lock_init(page);
  831. inc_zone_page_state(page, NR_PAGETABLE);
  832. }
  833. static inline void pgtable_page_dtor(struct page *page)
  834. {
  835. pte_lock_deinit(page);
  836. dec_zone_page_state(page, NR_PAGETABLE);
  837. }
  838. #define pte_offset_map_lock(mm, pmd, address, ptlp) \
  839. ({ \
  840. spinlock_t *__ptl = pte_lockptr(mm, pmd); \
  841. pte_t *__pte = pte_offset_map(pmd, address); \
  842. *(ptlp) = __ptl; \
  843. spin_lock(__ptl); \
  844. __pte; \
  845. })
  846. #define pte_unmap_unlock(pte, ptl) do { \
  847. spin_unlock(ptl); \
  848. pte_unmap(pte); \
  849. } while (0)
  850. #define pte_alloc_map(mm, pmd, address) \
  851. ((unlikely(!pmd_present(*(pmd))) && __pte_alloc(mm, pmd, address))? \
  852. NULL: pte_offset_map(pmd, address))
  853. #define pte_alloc_map_lock(mm, pmd, address, ptlp) \
  854. ((unlikely(!pmd_present(*(pmd))) && __pte_alloc(mm, pmd, address))? \
  855. NULL: pte_offset_map_lock(mm, pmd, address, ptlp))
  856. #define pte_alloc_kernel(pmd, address) \
  857. ((unlikely(!pmd_present(*(pmd))) && __pte_alloc_kernel(pmd, address))? \
  858. NULL: pte_offset_kernel(pmd, address))
  859. extern void free_area_init(unsigned long * zones_size);
  860. extern void free_area_init_node(int nid, unsigned long * zones_size,
  861. unsigned long zone_start_pfn, unsigned long *zholes_size);
  862. #ifdef CONFIG_ARCH_POPULATES_NODE_MAP
  863. /*
  864. * With CONFIG_ARCH_POPULATES_NODE_MAP set, an architecture may initialise its
  865. * zones, allocate the backing mem_map and account for memory holes in a more
  866. * architecture independent manner. This is a substitute for creating the
  867. * zone_sizes[] and zholes_size[] arrays and passing them to
  868. * free_area_init_node()
  869. *
  870. * An architecture is expected to register range of page frames backed by
  871. * physical memory with add_active_range() before calling
  872. * free_area_init_nodes() passing in the PFN each zone ends at. At a basic
  873. * usage, an architecture is expected to do something like
  874. *
  875. * unsigned long max_zone_pfns[MAX_NR_ZONES] = {max_dma, max_normal_pfn,
  876. * max_highmem_pfn};
  877. * for_each_valid_physical_page_range()
  878. * add_active_range(node_id, start_pfn, end_pfn)
  879. * free_area_init_nodes(max_zone_pfns);
  880. *
  881. * If the architecture guarantees that there are no holes in the ranges
  882. * registered with add_active_range(), free_bootmem_active_regions()
  883. * will call free_bootmem_node() for each registered physical page range.
  884. * Similarly sparse_memory_present_with_active_regions() calls
  885. * memory_present() for each range when SPARSEMEM is enabled.
  886. *
  887. * See mm/page_alloc.c for more information on each function exposed by
  888. * CONFIG_ARCH_POPULATES_NODE_MAP
  889. */
  890. extern void free_area_init_nodes(unsigned long *max_zone_pfn);
  891. extern void add_active_range(unsigned int nid, unsigned long start_pfn,
  892. unsigned long end_pfn);
  893. extern void remove_active_range(unsigned int nid, unsigned long start_pfn,
  894. unsigned long end_pfn);
  895. extern void push_node_boundaries(unsigned int nid, unsigned long start_pfn,
  896. unsigned long end_pfn);
  897. extern void remove_all_active_ranges(void);
  898. extern unsigned long absent_pages_in_range(unsigned long start_pfn,
  899. unsigned long end_pfn);
  900. extern void get_pfn_range_for_nid(unsigned int nid,
  901. unsigned long *start_pfn, unsigned long *end_pfn);
  902. extern unsigned long find_min_pfn_with_active_regions(void);
  903. extern void free_bootmem_with_active_regions(int nid,
  904. unsigned long max_low_pfn);
  905. typedef int (*work_fn_t)(unsigned long, unsigned long, void *);
  906. extern void work_with_active_regions(int nid, work_fn_t work_fn, void *data);
  907. extern void sparse_memory_present_with_active_regions(int nid);
  908. #endif /* CONFIG_ARCH_POPULATES_NODE_MAP */
  909. #if !defined(CONFIG_ARCH_POPULATES_NODE_MAP) && \
  910. !defined(CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID)
  911. static inline int __early_pfn_to_nid(unsigned long pfn)
  912. {
  913. return 0;
  914. }
  915. #else
  916. /* please see mm/page_alloc.c */
  917. extern int __meminit early_pfn_to_nid(unsigned long pfn);
  918. #ifdef CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID
  919. /* there is a per-arch backend function. */
  920. extern int __meminit __early_pfn_to_nid(unsigned long pfn);
  921. #endif /* CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID */
  922. #endif
  923. extern void set_dma_reserve(unsigned long new_dma_reserve);
  924. extern void memmap_init_zone(unsigned long, int, unsigned long,
  925. unsigned long, enum memmap_context);
  926. extern void setup_per_zone_pages_min(void);
  927. extern void mem_init(void);
  928. extern void __init mmap_init(void);
  929. extern void show_mem(void);
  930. extern void si_meminfo(struct sysinfo * val);
  931. extern void si_meminfo_node(struct sysinfo *val, int nid);
  932. extern int after_bootmem;
  933. #ifdef CONFIG_NUMA
  934. extern void setup_per_cpu_pageset(void);
  935. #else
  936. static inline void setup_per_cpu_pageset(void) {}
  937. #endif
  938. /* nommu.c */
  939. extern atomic_long_t mmap_pages_allocated;
  940. /* prio_tree.c */
  941. void vma_prio_tree_add(struct vm_area_struct *, struct vm_area_struct *old);
  942. void vma_prio_tree_insert(struct vm_area_struct *, struct prio_tree_root *);
  943. void vma_prio_tree_remove(struct vm_area_struct *, struct prio_tree_root *);
  944. struct vm_area_struct *vma_prio_tree_next(struct vm_area_struct *vma,
  945. struct prio_tree_iter *iter);
  946. #define vma_prio_tree_foreach(vma, iter, root, begin, end) \
  947. for (prio_tree_iter_init(iter, root, begin, end), vma = NULL; \
  948. (vma = vma_prio_tree_next(vma, iter)); )
  949. static inline void vma_nonlinear_insert(struct vm_area_struct *vma,
  950. struct list_head *list)
  951. {
  952. vma->shared.vm_set.parent = NULL;
  953. list_add_tail(&vma->shared.vm_set.list, list);
  954. }
  955. /* mmap.c */
  956. extern int __vm_enough_memory(struct mm_struct *mm, long pages, int cap_sys_admin);
  957. extern void vma_adjust(struct vm_area_struct *vma, unsigned long start,
  958. unsigned long end, pgoff_t pgoff, struct vm_area_struct *insert);
  959. extern struct vm_area_struct *vma_merge(struct mm_struct *,
  960. struct vm_area_struct *prev, unsigned long addr, unsigned long end,
  961. unsigned long vm_flags, struct anon_vma *, struct file *, pgoff_t,
  962. struct mempolicy *);
  963. extern struct anon_vma *find_mergeable_anon_vma(struct vm_area_struct *);
  964. extern int split_vma(struct mm_struct *,
  965. struct vm_area_struct *, unsigned long addr, int new_below);
  966. extern int insert_vm_struct(struct mm_struct *, struct vm_area_struct *);
  967. extern void __vma_link_rb(struct mm_struct *, struct vm_area_struct *,
  968. struct rb_node **, struct rb_node *);
  969. extern void unlink_file_vma(struct vm_area_struct *);
  970. extern struct vm_area_struct *copy_vma(struct vm_area_struct **,
  971. unsigned long addr, unsigned long len, pgoff_t pgoff);
  972. extern void exit_mmap(struct mm_struct *);
  973. extern int mm_take_all_locks(struct mm_struct *mm);
  974. extern void mm_drop_all_locks(struct mm_struct *mm);
  975. #ifdef CONFIG_PROC_FS
  976. /* From fs/proc/base.c. callers must _not_ hold the mm's exe_file_lock */
  977. extern void added_exe_file_vma(struct mm_struct *mm);
  978. extern void removed_exe_file_vma(struct mm_struct *mm);
  979. #else
  980. static inline void added_exe_file_vma(struct mm_struct *mm)
  981. {}
  982. static inline void removed_exe_file_vma(struct mm_struct *mm)
  983. {}
  984. #endif /* CONFIG_PROC_FS */
  985. extern int may_expand_vm(struct mm_struct *mm, unsigned long npages);
  986. extern int install_special_mapping(struct mm_struct *mm,
  987. unsigned long addr, unsigned long len,
  988. unsigned long flags, struct page **pages);
  989. extern unsigned long get_unmapped_area(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
  990. extern unsigned long do_mmap_pgoff(struct file *file, unsigned long addr,
  991. unsigned long len, unsigned long prot,
  992. unsigned long flag, unsigned long pgoff);
  993. extern unsigned long mmap_region(struct file *file, unsigned long addr,
  994. unsigned long len, unsigned long flags,
  995. unsigned int vm_flags, unsigned long pgoff);
  996. static inline unsigned long do_mmap(struct file *file, unsigned long addr,
  997. unsigned long len, unsigned long prot,
  998. unsigned long flag, unsigned long offset)
  999. {
  1000. unsigned long ret = -EINVAL;
  1001. if ((offset + PAGE_ALIGN(len)) < offset)
  1002. goto out;
  1003. if (!(offset & ~PAGE_MASK))
  1004. ret = do_mmap_pgoff(file, addr, len, prot, flag, offset >> PAGE_SHIFT);
  1005. out:
  1006. return ret;
  1007. }
  1008. extern int do_munmap(struct mm_struct *, unsigned long, size_t);
  1009. extern unsigned long do_brk(unsigned long, unsigned long);
  1010. /* filemap.c */
  1011. extern unsigned long page_unuse(struct page *);
  1012. extern void truncate_inode_pages(struct address_space *, loff_t);
  1013. extern void truncate_inode_pages_range(struct address_space *,
  1014. loff_t lstart, loff_t lend);
  1015. /* generic vm_area_ops exported for stackable file systems */
  1016. extern int filemap_fault(struct vm_area_struct *, struct vm_fault *);
  1017. /* mm/page-writeback.c */
  1018. int write_one_page(struct page *page, int wait);
  1019. void task_dirty_inc(struct task_struct *tsk);
  1020. /* readahead.c */
  1021. #define VM_MAX_READAHEAD 128 /* kbytes */
  1022. #define VM_MIN_READAHEAD 16 /* kbytes (includes current page) */
  1023. int do_page_cache_readahead(struct address_space *mapping, struct file *filp,
  1024. pgoff_t offset, unsigned long nr_to_read);
  1025. int force_page_cache_readahead(struct address_space *mapping, struct file *filp,
  1026. pgoff_t offset, unsigned long nr_to_read);
  1027. void page_cache_sync_readahead(struct address_space *mapping,
  1028. struct file_ra_state *ra,
  1029. struct file *filp,
  1030. pgoff_t offset,
  1031. unsigned long size);
  1032. void page_cache_async_readahead(struct address_space *mapping,
  1033. struct file_ra_state *ra,
  1034. struct file *filp,
  1035. struct page *pg,
  1036. pgoff_t offset,
  1037. unsigned long size);
  1038. unsigned long max_sane_readahead(unsigned long nr);
  1039. /* Do stack extension */
  1040. extern int expand_stack(struct vm_area_struct *vma, unsigned long address);
  1041. #ifdef CONFIG_IA64
  1042. extern int expand_upwards(struct vm_area_struct *vma, unsigned long address);
  1043. #endif
  1044. extern int expand_stack_downwards(struct vm_area_struct *vma,
  1045. unsigned long address);
  1046. /* Look up the first VMA which satisfies addr < vm_end, NULL if none. */
  1047. extern struct vm_area_struct * find_vma(struct mm_struct * mm, unsigned long addr);
  1048. extern struct vm_area_struct * find_vma_prev(struct mm_struct * mm, unsigned long addr,
  1049. struct vm_area_struct **pprev);
  1050. /* Look up the first VMA which intersects the interval start_addr..end_addr-1,
  1051. NULL if none. Assume start_addr < end_addr. */
  1052. static inline struct vm_area_struct * find_vma_intersection(struct mm_struct * mm, unsigned long start_addr, unsigned long end_addr)
  1053. {
  1054. struct vm_area_struct * vma = find_vma(mm,start_addr);
  1055. if (vma && end_addr <= vma->vm_start)
  1056. vma = NULL;
  1057. return vma;
  1058. }
  1059. static inline unsigned long vma_pages(struct vm_area_struct *vma)
  1060. {
  1061. return (vma->vm_end - vma->vm_start) >> PAGE_SHIFT;
  1062. }
  1063. pgprot_t vm_get_page_prot(unsigned long vm_flags);
  1064. struct vm_area_struct *find_extend_vma(struct mm_struct *, unsigned long addr);
  1065. int remap_pfn_range(struct vm_area_struct *, unsigned long addr,
  1066. unsigned long pfn, unsigned long size, pgprot_t);
  1067. int vm_insert_page(struct vm_area_struct *, unsigned long addr, struct page *);
  1068. int vm_insert_pfn(struct vm_area_struct *vma, unsigned long addr,
  1069. unsigned long pfn);
  1070. int vm_insert_mixed(struct vm_area_struct *vma, unsigned long addr,
  1071. unsigned long pfn);
  1072. struct page *follow_page(struct vm_area_struct *, unsigned long address,
  1073. unsigned int foll_flags);
  1074. #define FOLL_WRITE 0x01 /* check pte is writable */
  1075. #define FOLL_TOUCH 0x02 /* mark page accessed */
  1076. #define FOLL_GET 0x04 /* do get_page on page */
  1077. #define FOLL_ANON 0x08 /* give ZERO_PAGE if no pgtable */
  1078. typedef int (*pte_fn_t)(pte_t *pte, pgtable_t token, unsigned long addr,
  1079. void *data);
  1080. extern int apply_to_page_range(struct mm_struct *mm, unsigned long address,
  1081. unsigned long size, pte_fn_t fn, void *data);
  1082. #ifdef CONFIG_PROC_FS
  1083. void vm_stat_account(struct mm_struct *, unsigned long, struct file *, long);
  1084. #else
  1085. static inline void vm_stat_account(struct mm_struct *mm,
  1086. unsigned long flags, struct file *file, long pages)
  1087. {
  1088. }
  1089. #endif /* CONFIG_PROC_FS */
  1090. #ifdef CONFIG_DEBUG_PAGEALLOC
  1091. extern int debug_pagealloc_enabled;
  1092. extern void kernel_map_pages(struct page *page, int numpages, int enable);
  1093. static inline void enable_debug_pagealloc(void)
  1094. {
  1095. debug_pagealloc_enabled = 1;
  1096. }
  1097. #ifdef CONFIG_HIBERNATION
  1098. extern bool kernel_page_present(struct page *page);
  1099. #endif /* CONFIG_HIBERNATION */
  1100. #else
  1101. static inline void
  1102. kernel_map_pages(struct page *page, int numpages, int enable) {}
  1103. static inline void enable_debug_pagealloc(void)
  1104. {
  1105. }
  1106. #ifdef CONFIG_HIBERNATION
  1107. static inline bool kernel_page_present(struct page *page) { return true; }
  1108. #endif /* CONFIG_HIBERNATION */
  1109. #endif
  1110. extern struct vm_area_struct *get_gate_vma(struct task_struct *tsk);
  1111. #ifdef __HAVE_ARCH_GATE_AREA
  1112. int in_gate_area_no_task(unsigned long addr);
  1113. int in_gate_area(struct task_struct *task, unsigned long addr);
  1114. #else
  1115. int in_gate_area_no_task(unsigned long addr);
  1116. #define in_gate_area(task, addr) ({(void)task; in_gate_area_no_task(addr);})
  1117. #endif /* __HAVE_ARCH_GATE_AREA */
  1118. int drop_caches_sysctl_handler(struct ctl_table *, int, struct file *,
  1119. void __user *, size_t *, loff_t *);
  1120. unsigned long shrink_slab(unsigned long scanned, gfp_t gfp_mask,
  1121. unsigned long lru_pages);
  1122. #ifndef CONFIG_MMU
  1123. #define randomize_va_space 0
  1124. #else
  1125. extern int randomize_va_space;
  1126. #endif
  1127. const char * arch_vma_name(struct vm_area_struct *vma);
  1128. void print_vma_addr(char *prefix, unsigned long rip);
  1129. struct page *sparse_mem_map_populate(unsigned long pnum, int nid);
  1130. pgd_t *vmemmap_pgd_populate(unsigned long addr, int node);
  1131. pud_t *vmemmap_pud_populate(pgd_t *pgd, unsigned long addr, int node);
  1132. pmd_t *vmemmap_pmd_populate(pud_t *pud, unsigned long addr, int node);
  1133. pte_t *vmemmap_pte_populate(pmd_t *pmd, unsigned long addr, int node);
  1134. void *vmemmap_alloc_block(unsigned long size, int node);
  1135. void vmemmap_verify(pte_t *, int, unsigned long, unsigned long);
  1136. int vmemmap_populate_basepages(struct page *start_page,
  1137. unsigned long pages, int node);
  1138. int vmemmap_populate(struct page *start_page, unsigned long pages, int node);
  1139. void vmemmap_populate_print_last(void);
  1140. extern int account_locked_memory(struct mm_struct *mm, struct rlimit *rlim,
  1141. size_t size);
  1142. extern void refund_locked_memory(struct mm_struct *mm, size_t size);
  1143. #endif /* __KERNEL__ */
  1144. #endif /* _LINUX_MM_H */