mm.h 84 KB

1234567891011121314151617181920212223242526272829303132333435363738394041424344454647484950515253545556575859606162636465666768697071727374757677787980818283848586878889909192939495969798991001011021031041051061071081091101111121131141151161171181191201211221231241251261271281291301311321331341351361371381391401411421431441451461471481491501511521531541551561571581591601611621631641651661671681691701711721731741751761771781791801811821831841851861871881891901911921931941951961971981992002012022032042052062072082092102112122132142152162172182192202212222232242252262272282292302312322332342352362372382392402412422432442452462472482492502512522532542552562572582592602612622632642652662672682692702712722732742752762772782792802812822832842852862872882892902912922932942952962972982993003013023033043053063073083093103113123133143153163173183193203213223233243253263273283293303313323333343353363373383393403413423433443453463473483493503513523533543553563573583593603613623633643653663673683693703713723733743753763773783793803813823833843853863873883893903913923933943953963973983994004014024034044054064074084094104114124134144154164174184194204214224234244254264274284294304314324334344354364374384394404414424434444454464474484494504514524534544554564574584594604614624634644654664674684694704714724734744754764774784794804814824834844854864874884894904914924934944954964974984995005015025035045055065075085095105115125135145155165175185195205215225235245255265275285295305315325335345355365375385395405415425435445455465475485495505515525535545555565575585595605615625635645655665675685695705715725735745755765775785795805815825835845855865875885895905915925935945955965975985996006016026036046056066076086096106116126136146156166176186196206216226236246256266276286296306316326336346356366376386396406416426436446456466476486496506516526536546556566576586596606616626636646656666676686696706716726736746756766776786796806816826836846856866876886896906916926936946956966976986997007017027037047057067077087097107117127137147157167177187197207217227237247257267277287297307317327337347357367377387397407417427437447457467477487497507517527537547557567577587597607617627637647657667677687697707717727737747757767777787797807817827837847857867877887897907917927937947957967977987998008018028038048058068078088098108118128138148158168178188198208218228238248258268278288298308318328338348358368378388398408418428438448458468478488498508518528538548558568578588598608618628638648658668678688698708718728738748758768778788798808818828838848858868878888898908918928938948958968978988999009019029039049059069079089099109119129139149159169179189199209219229239249259269279289299309319329339349359369379389399409419429439449459469479489499509519529539549559569579589599609619629639649659669679689699709719729739749759769779789799809819829839849859869879889899909919929939949959969979989991000100110021003100410051006100710081009101010111012101310141015101610171018101910201021102210231024102510261027102810291030103110321033103410351036103710381039104010411042104310441045104610471048104910501051105210531054105510561057105810591060106110621063106410651066106710681069107010711072107310741075107610771078107910801081108210831084108510861087108810891090109110921093109410951096109710981099110011011102110311041105110611071108110911101111111211131114111511161117111811191120112111221123112411251126112711281129113011311132113311341135113611371138113911401141114211431144114511461147114811491150115111521153115411551156115711581159116011611162116311641165116611671168116911701171117211731174117511761177117811791180118111821183118411851186118711881189119011911192119311941195119611971198119912001201120212031204120512061207120812091210121112121213121412151216121712181219122012211222122312241225122612271228122912301231123212331234123512361237123812391240124112421243124412451246124712481249125012511252125312541255125612571258125912601261126212631264126512661267126812691270127112721273127412751276127712781279128012811282128312841285128612871288128912901291129212931294129512961297129812991300130113021303130413051306130713081309131013111312131313141315131613171318131913201321132213231324132513261327132813291330133113321333133413351336133713381339134013411342134313441345134613471348134913501351135213531354135513561357135813591360136113621363136413651366136713681369137013711372137313741375137613771378137913801381138213831384138513861387138813891390139113921393139413951396139713981399140014011402140314041405140614071408140914101411141214131414141514161417141814191420142114221423142414251426142714281429143014311432143314341435143614371438143914401441144214431444144514461447144814491450145114521453145414551456145714581459146014611462146314641465146614671468146914701471147214731474147514761477147814791480148114821483148414851486148714881489149014911492149314941495149614971498149915001501150215031504150515061507150815091510151115121513151415151516151715181519152015211522152315241525152615271528152915301531153215331534153515361537153815391540154115421543154415451546154715481549155015511552155315541555155615571558155915601561156215631564156515661567156815691570157115721573157415751576157715781579158015811582158315841585158615871588158915901591159215931594159515961597159815991600160116021603160416051606160716081609161016111612161316141615161616171618161916201621162216231624162516261627162816291630163116321633163416351636163716381639164016411642164316441645164616471648164916501651165216531654165516561657165816591660166116621663166416651666166716681669167016711672167316741675167616771678167916801681168216831684168516861687168816891690169116921693169416951696169716981699170017011702170317041705170617071708170917101711171217131714171517161717171817191720172117221723172417251726172717281729173017311732173317341735173617371738173917401741174217431744174517461747174817491750175117521753175417551756175717581759176017611762176317641765176617671768176917701771177217731774177517761777177817791780178117821783178417851786178717881789179017911792179317941795179617971798179918001801180218031804180518061807180818091810181118121813181418151816181718181819182018211822182318241825182618271828182918301831183218331834183518361837183818391840184118421843184418451846184718481849185018511852185318541855185618571858185918601861186218631864186518661867186818691870187118721873187418751876187718781879188018811882188318841885188618871888188918901891189218931894189518961897189818991900190119021903190419051906190719081909191019111912191319141915191619171918191919201921192219231924192519261927192819291930193119321933193419351936193719381939194019411942194319441945194619471948194919501951195219531954195519561957195819591960196119621963196419651966196719681969197019711972197319741975197619771978197919801981198219831984198519861987198819891990199119921993199419951996199719981999200020012002200320042005200620072008200920102011201220132014201520162017201820192020202120222023202420252026202720282029203020312032203320342035203620372038203920402041204220432044204520462047204820492050205120522053205420552056205720582059206020612062206320642065206620672068206920702071207220732074207520762077207820792080208120822083208420852086208720882089209020912092209320942095209620972098209921002101210221032104210521062107210821092110211121122113211421152116211721182119212021212122212321242125212621272128212921302131213221332134213521362137213821392140214121422143214421452146214721482149215021512152215321542155215621572158215921602161216221632164216521662167216821692170217121722173217421752176217721782179218021812182218321842185218621872188218921902191219221932194219521962197219821992200220122022203220422052206220722082209221022112212221322142215221622172218221922202221222222232224222522262227222822292230223122322233223422352236223722382239224022412242224322442245224622472248224922502251225222532254225522562257225822592260226122622263226422652266226722682269227022712272227322742275227622772278227922802281228222832284228522862287228822892290229122922293229422952296229722982299230023012302230323042305230623072308230923102311231223132314231523162317231823192320232123222323232423252326232723282329233023312332233323342335233623372338233923402341234223432344234523462347234823492350235123522353235423552356235723582359236023612362236323642365236623672368236923702371237223732374237523762377237823792380238123822383238423852386238723882389239023912392239323942395239623972398239924002401240224032404240524062407240824092410241124122413241424152416241724182419242024212422242324242425242624272428242924302431243224332434243524362437243824392440244124422443244424452446244724482449245024512452245324542455245624572458245924602461246224632464246524662467246824692470247124722473247424752476247724782479248024812482248324842485248624872488248924902491249224932494249524962497249824992500250125022503250425052506250725082509251025112512251325142515251625172518251925202521252225232524252525262527252825292530253125322533253425352536253725382539254025412542254325442545254625472548254925502551255225532554255525562557255825592560256125622563256425652566256725682569257025712572257325742575257625772578257925802581258225832584258525862587258825892590259125922593259425952596259725982599260026012602260326042605260626072608260926102611261226132614261526162617261826192620262126222623262426252626262726282629263026312632263326342635263626372638263926402641264226432644264526462647264826492650265126522653265426552656265726582659266026612662266326642665266626672668266926702671267226732674267526762677267826792680268126822683268426852686268726882689
  1. /* SPDX-License-Identifier: GPL-2.0 */
  2. #ifndef _LINUX_MM_H
  3. #define _LINUX_MM_H
  4. #include <linux/errno.h>
  5. #ifdef __KERNEL__
  6. #include <linux/mmdebug.h>
  7. #include <linux/gfp.h>
  8. #include <linux/bug.h>
  9. #include <linux/list.h>
  10. #include <linux/mmzone.h>
  11. #include <linux/rbtree.h>
  12. #include <linux/atomic.h>
  13. #include <linux/debug_locks.h>
  14. #include <linux/mm_types.h>
  15. #include <linux/range.h>
  16. #include <linux/pfn.h>
  17. #include <linux/percpu-refcount.h>
  18. #include <linux/bit_spinlock.h>
  19. #include <linux/shrinker.h>
  20. #include <linux/resource.h>
  21. #include <linux/page_ext.h>
  22. #include <linux/err.h>
  23. #include <linux/page_ref.h>
  24. #include <linux/memremap.h>
  25. struct mempolicy;
  26. struct anon_vma;
  27. struct anon_vma_chain;
  28. struct file_ra_state;
  29. struct user_struct;
  30. struct writeback_control;
  31. struct bdi_writeback;
  32. void init_mm_internals(void);
  33. #ifndef CONFIG_NEED_MULTIPLE_NODES /* Don't use mapnrs, do it properly */
  34. extern unsigned long max_mapnr;
  35. static inline void set_max_mapnr(unsigned long limit)
  36. {
  37. max_mapnr = limit;
  38. }
  39. #else
  40. static inline void set_max_mapnr(unsigned long limit) { }
  41. #endif
  42. extern unsigned long totalram_pages;
  43. extern void * high_memory;
  44. extern int page_cluster;
  45. #ifdef CONFIG_SYSCTL
  46. extern int sysctl_legacy_va_layout;
  47. #else
  48. #define sysctl_legacy_va_layout 0
  49. #endif
  50. #ifdef CONFIG_HAVE_ARCH_MMAP_RND_BITS
  51. extern const int mmap_rnd_bits_min;
  52. extern const int mmap_rnd_bits_max;
  53. extern int mmap_rnd_bits __read_mostly;
  54. #endif
  55. #ifdef CONFIG_HAVE_ARCH_MMAP_RND_COMPAT_BITS
  56. extern const int mmap_rnd_compat_bits_min;
  57. extern const int mmap_rnd_compat_bits_max;
  58. extern int mmap_rnd_compat_bits __read_mostly;
  59. #endif
  60. #include <asm/page.h>
  61. #include <asm/pgtable.h>
  62. #include <asm/processor.h>
  63. #ifndef __pa_symbol
  64. #define __pa_symbol(x) __pa(RELOC_HIDE((unsigned long)(x), 0))
  65. #endif
  66. #ifndef page_to_virt
  67. #define page_to_virt(x) __va(PFN_PHYS(page_to_pfn(x)))
  68. #endif
  69. #ifndef lm_alias
  70. #define lm_alias(x) __va(__pa_symbol(x))
  71. #endif
  72. /*
  73. * To prevent common memory management code establishing
  74. * a zero page mapping on a read fault.
  75. * This macro should be defined within <asm/pgtable.h>.
  76. * s390 does this to prevent multiplexing of hardware bits
  77. * related to the physical page in case of virtualization.
  78. */
  79. #ifndef mm_forbids_zeropage
  80. #define mm_forbids_zeropage(X) (0)
  81. #endif
  82. /*
  83. * On some architectures it is expensive to call memset() for small sizes.
  84. * Those architectures should provide their own implementation of "struct page"
  85. * zeroing by defining this macro in <asm/pgtable.h>.
  86. */
  87. #ifndef mm_zero_struct_page
  88. #define mm_zero_struct_page(pp) ((void)memset((pp), 0, sizeof(struct page)))
  89. #endif
  90. /*
  91. * Default maximum number of active map areas, this limits the number of vmas
  92. * per mm struct. Users can overwrite this number by sysctl but there is a
  93. * problem.
  94. *
  95. * When a program's coredump is generated as ELF format, a section is created
  96. * per a vma. In ELF, the number of sections is represented in unsigned short.
  97. * This means the number of sections should be smaller than 65535 at coredump.
  98. * Because the kernel adds some informative sections to a image of program at
  99. * generating coredump, we need some margin. The number of extra sections is
  100. * 1-3 now and depends on arch. We use "5" as safe margin, here.
  101. *
  102. * ELF extended numbering allows more than 65535 sections, so 16-bit bound is
  103. * not a hard limit any more. Although some userspace tools can be surprised by
  104. * that.
  105. */
  106. #define MAPCOUNT_ELF_CORE_MARGIN (5)
  107. #define DEFAULT_MAX_MAP_COUNT (USHRT_MAX - MAPCOUNT_ELF_CORE_MARGIN)
  108. extern int sysctl_max_map_count;
  109. extern unsigned long sysctl_user_reserve_kbytes;
  110. extern unsigned long sysctl_admin_reserve_kbytes;
  111. extern int sysctl_overcommit_memory;
  112. extern int sysctl_overcommit_ratio;
  113. extern unsigned long sysctl_overcommit_kbytes;
  114. extern int overcommit_ratio_handler(struct ctl_table *, int, void __user *,
  115. size_t *, loff_t *);
  116. extern int overcommit_kbytes_handler(struct ctl_table *, int, void __user *,
  117. size_t *, loff_t *);
  118. #define nth_page(page,n) pfn_to_page(page_to_pfn((page)) + (n))
  119. /* to align the pointer to the (next) page boundary */
  120. #define PAGE_ALIGN(addr) ALIGN(addr, PAGE_SIZE)
  121. /* test whether an address (unsigned long or pointer) is aligned to PAGE_SIZE */
  122. #define PAGE_ALIGNED(addr) IS_ALIGNED((unsigned long)(addr), PAGE_SIZE)
  123. /*
  124. * Linux kernel virtual memory manager primitives.
  125. * The idea being to have a "virtual" mm in the same way
  126. * we have a virtual fs - giving a cleaner interface to the
  127. * mm details, and allowing different kinds of memory mappings
  128. * (from shared memory to executable loading to arbitrary
  129. * mmap() functions).
  130. */
  131. extern struct kmem_cache *vm_area_cachep;
  132. #ifndef CONFIG_MMU
  133. extern struct rb_root nommu_region_tree;
  134. extern struct rw_semaphore nommu_region_sem;
  135. extern unsigned int kobjsize(const void *objp);
  136. #endif
  137. /*
  138. * vm_flags in vm_area_struct, see mm_types.h.
  139. * When changing, update also include/trace/events/mmflags.h
  140. */
  141. #define VM_NONE 0x00000000
  142. #define VM_READ 0x00000001 /* currently active flags */
  143. #define VM_WRITE 0x00000002
  144. #define VM_EXEC 0x00000004
  145. #define VM_SHARED 0x00000008
  146. /* mprotect() hardcodes VM_MAYREAD >> 4 == VM_READ, and so for r/w/x bits. */
  147. #define VM_MAYREAD 0x00000010 /* limits for mprotect() etc */
  148. #define VM_MAYWRITE 0x00000020
  149. #define VM_MAYEXEC 0x00000040
  150. #define VM_MAYSHARE 0x00000080
  151. #define VM_GROWSDOWN 0x00000100 /* general info on the segment */
  152. #define VM_UFFD_MISSING 0x00000200 /* missing pages tracking */
  153. #define VM_PFNMAP 0x00000400 /* Page-ranges managed without "struct page", just pure PFN */
  154. #define VM_DENYWRITE 0x00000800 /* ETXTBSY on write attempts.. */
  155. #define VM_UFFD_WP 0x00001000 /* wrprotect pages tracking */
  156. #define VM_LOCKED 0x00002000
  157. #define VM_IO 0x00004000 /* Memory mapped I/O or similar */
  158. /* Used by sys_madvise() */
  159. #define VM_SEQ_READ 0x00008000 /* App will access data sequentially */
  160. #define VM_RAND_READ 0x00010000 /* App will not benefit from clustered reads */
  161. #define VM_DONTCOPY 0x00020000 /* Do not copy this vma on fork */
  162. #define VM_DONTEXPAND 0x00040000 /* Cannot expand with mremap() */
  163. #define VM_LOCKONFAULT 0x00080000 /* Lock the pages covered when they are faulted in */
  164. #define VM_ACCOUNT 0x00100000 /* Is a VM accounted object */
  165. #define VM_NORESERVE 0x00200000 /* should the VM suppress accounting */
  166. #define VM_HUGETLB 0x00400000 /* Huge TLB Page VM */
  167. #define VM_SYNC 0x00800000 /* Synchronous page faults */
  168. #define VM_ARCH_1 0x01000000 /* Architecture-specific flag */
  169. #define VM_WIPEONFORK 0x02000000 /* Wipe VMA contents in child. */
  170. #define VM_DONTDUMP 0x04000000 /* Do not include in the core dump */
  171. #ifdef CONFIG_MEM_SOFT_DIRTY
  172. # define VM_SOFTDIRTY 0x08000000 /* Not soft dirty clean area */
  173. #else
  174. # define VM_SOFTDIRTY 0
  175. #endif
  176. #define VM_MIXEDMAP 0x10000000 /* Can contain "struct page" and pure PFN pages */
  177. #define VM_HUGEPAGE 0x20000000 /* MADV_HUGEPAGE marked this vma */
  178. #define VM_NOHUGEPAGE 0x40000000 /* MADV_NOHUGEPAGE marked this vma */
  179. #define VM_MERGEABLE 0x80000000 /* KSM may merge identical pages */
  180. #ifdef CONFIG_ARCH_USES_HIGH_VMA_FLAGS
  181. #define VM_HIGH_ARCH_BIT_0 32 /* bit only usable on 64-bit architectures */
  182. #define VM_HIGH_ARCH_BIT_1 33 /* bit only usable on 64-bit architectures */
  183. #define VM_HIGH_ARCH_BIT_2 34 /* bit only usable on 64-bit architectures */
  184. #define VM_HIGH_ARCH_BIT_3 35 /* bit only usable on 64-bit architectures */
  185. #define VM_HIGH_ARCH_BIT_4 36 /* bit only usable on 64-bit architectures */
  186. #define VM_HIGH_ARCH_0 BIT(VM_HIGH_ARCH_BIT_0)
  187. #define VM_HIGH_ARCH_1 BIT(VM_HIGH_ARCH_BIT_1)
  188. #define VM_HIGH_ARCH_2 BIT(VM_HIGH_ARCH_BIT_2)
  189. #define VM_HIGH_ARCH_3 BIT(VM_HIGH_ARCH_BIT_3)
  190. #define VM_HIGH_ARCH_4 BIT(VM_HIGH_ARCH_BIT_4)
  191. #endif /* CONFIG_ARCH_USES_HIGH_VMA_FLAGS */
  192. #if defined(CONFIG_X86)
  193. # define VM_PAT VM_ARCH_1 /* PAT reserves whole VMA at once (x86) */
  194. #if defined (CONFIG_X86_INTEL_MEMORY_PROTECTION_KEYS)
  195. # define VM_PKEY_SHIFT VM_HIGH_ARCH_BIT_0
  196. # define VM_PKEY_BIT0 VM_HIGH_ARCH_0 /* A protection key is a 4-bit value */
  197. # define VM_PKEY_BIT1 VM_HIGH_ARCH_1
  198. # define VM_PKEY_BIT2 VM_HIGH_ARCH_2
  199. # define VM_PKEY_BIT3 VM_HIGH_ARCH_3
  200. #endif
  201. #elif defined(CONFIG_PPC)
  202. # define VM_SAO VM_ARCH_1 /* Strong Access Ordering (powerpc) */
  203. #elif defined(CONFIG_PARISC)
  204. # define VM_GROWSUP VM_ARCH_1
  205. #elif defined(CONFIG_IA64)
  206. # define VM_GROWSUP VM_ARCH_1
  207. #elif defined(CONFIG_SPARC64)
  208. # define VM_SPARC_ADI VM_ARCH_1 /* Uses ADI tag for access control */
  209. # define VM_ARCH_CLEAR VM_SPARC_ADI
  210. #elif !defined(CONFIG_MMU)
  211. # define VM_MAPPED_COPY VM_ARCH_1 /* T if mapped copy of data (nommu mmap) */
  212. #endif
  213. #if defined(CONFIG_X86_INTEL_MPX)
  214. /* MPX specific bounds table or bounds directory */
  215. # define VM_MPX VM_HIGH_ARCH_4
  216. #else
  217. # define VM_MPX VM_NONE
  218. #endif
  219. #ifndef VM_GROWSUP
  220. # define VM_GROWSUP VM_NONE
  221. #endif
  222. /* Bits set in the VMA until the stack is in its final location */
  223. #define VM_STACK_INCOMPLETE_SETUP (VM_RAND_READ | VM_SEQ_READ)
  224. #ifndef VM_STACK_DEFAULT_FLAGS /* arch can override this */
  225. #define VM_STACK_DEFAULT_FLAGS VM_DATA_DEFAULT_FLAGS
  226. #endif
  227. #ifdef CONFIG_STACK_GROWSUP
  228. #define VM_STACK VM_GROWSUP
  229. #else
  230. #define VM_STACK VM_GROWSDOWN
  231. #endif
  232. #define VM_STACK_FLAGS (VM_STACK | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT)
  233. /*
  234. * Special vmas that are non-mergable, non-mlock()able.
  235. * Note: mm/huge_memory.c VM_NO_THP depends on this definition.
  236. */
  237. #define VM_SPECIAL (VM_IO | VM_DONTEXPAND | VM_PFNMAP | VM_MIXEDMAP)
  238. /* This mask defines which mm->def_flags a process can inherit its parent */
  239. #define VM_INIT_DEF_MASK VM_NOHUGEPAGE
  240. /* This mask is used to clear all the VMA flags used by mlock */
  241. #define VM_LOCKED_CLEAR_MASK (~(VM_LOCKED | VM_LOCKONFAULT))
  242. /* Arch-specific flags to clear when updating VM flags on protection change */
  243. #ifndef VM_ARCH_CLEAR
  244. # define VM_ARCH_CLEAR VM_NONE
  245. #endif
  246. #define VM_FLAGS_CLEAR (ARCH_VM_PKEY_FLAGS | VM_ARCH_CLEAR)
  247. /*
  248. * mapping from the currently active vm_flags protection bits (the
  249. * low four bits) to a page protection mask..
  250. */
  251. extern pgprot_t protection_map[16];
  252. #define FAULT_FLAG_WRITE 0x01 /* Fault was a write access */
  253. #define FAULT_FLAG_MKWRITE 0x02 /* Fault was mkwrite of existing pte */
  254. #define FAULT_FLAG_ALLOW_RETRY 0x04 /* Retry fault if blocking */
  255. #define FAULT_FLAG_RETRY_NOWAIT 0x08 /* Don't drop mmap_sem and wait when retrying */
  256. #define FAULT_FLAG_KILLABLE 0x10 /* The fault task is in SIGKILL killable region */
  257. #define FAULT_FLAG_TRIED 0x20 /* Second try */
  258. #define FAULT_FLAG_USER 0x40 /* The fault originated in userspace */
  259. #define FAULT_FLAG_REMOTE 0x80 /* faulting for non current tsk/mm */
  260. #define FAULT_FLAG_INSTRUCTION 0x100 /* The fault was during an instruction fetch */
  261. #define FAULT_FLAG_TRACE \
  262. { FAULT_FLAG_WRITE, "WRITE" }, \
  263. { FAULT_FLAG_MKWRITE, "MKWRITE" }, \
  264. { FAULT_FLAG_ALLOW_RETRY, "ALLOW_RETRY" }, \
  265. { FAULT_FLAG_RETRY_NOWAIT, "RETRY_NOWAIT" }, \
  266. { FAULT_FLAG_KILLABLE, "KILLABLE" }, \
  267. { FAULT_FLAG_TRIED, "TRIED" }, \
  268. { FAULT_FLAG_USER, "USER" }, \
  269. { FAULT_FLAG_REMOTE, "REMOTE" }, \
  270. { FAULT_FLAG_INSTRUCTION, "INSTRUCTION" }
  271. /*
  272. * vm_fault is filled by the the pagefault handler and passed to the vma's
  273. * ->fault function. The vma's ->fault is responsible for returning a bitmask
  274. * of VM_FAULT_xxx flags that give details about how the fault was handled.
  275. *
  276. * MM layer fills up gfp_mask for page allocations but fault handler might
  277. * alter it if its implementation requires a different allocation context.
  278. *
  279. * pgoff should be used in favour of virtual_address, if possible.
  280. */
  281. struct vm_fault {
  282. struct vm_area_struct *vma; /* Target VMA */
  283. unsigned int flags; /* FAULT_FLAG_xxx flags */
  284. gfp_t gfp_mask; /* gfp mask to be used for allocations */
  285. pgoff_t pgoff; /* Logical page offset based on vma */
  286. unsigned long address; /* Faulting virtual address */
  287. pmd_t *pmd; /* Pointer to pmd entry matching
  288. * the 'address' */
  289. pud_t *pud; /* Pointer to pud entry matching
  290. * the 'address'
  291. */
  292. pte_t orig_pte; /* Value of PTE at the time of fault */
  293. struct page *cow_page; /* Page handler may use for COW fault */
  294. struct mem_cgroup *memcg; /* Cgroup cow_page belongs to */
  295. struct page *page; /* ->fault handlers should return a
  296. * page here, unless VM_FAULT_NOPAGE
  297. * is set (which is also implied by
  298. * VM_FAULT_ERROR).
  299. */
  300. /* These three entries are valid only while holding ptl lock */
  301. pte_t *pte; /* Pointer to pte entry matching
  302. * the 'address'. NULL if the page
  303. * table hasn't been allocated.
  304. */
  305. spinlock_t *ptl; /* Page table lock.
  306. * Protects pte page table if 'pte'
  307. * is not NULL, otherwise pmd.
  308. */
  309. pgtable_t prealloc_pte; /* Pre-allocated pte page table.
  310. * vm_ops->map_pages() calls
  311. * alloc_set_pte() from atomic context.
  312. * do_fault_around() pre-allocates
  313. * page table to avoid allocation from
  314. * atomic context.
  315. */
  316. };
  317. /* page entry size for vm->huge_fault() */
  318. enum page_entry_size {
  319. PE_SIZE_PTE = 0,
  320. PE_SIZE_PMD,
  321. PE_SIZE_PUD,
  322. };
  323. /*
  324. * These are the virtual MM functions - opening of an area, closing and
  325. * unmapping it (needed to keep files on disk up-to-date etc), pointer
  326. * to the functions called when a no-page or a wp-page exception occurs.
  327. */
  328. struct vm_operations_struct {
  329. void (*open)(struct vm_area_struct * area);
  330. void (*close)(struct vm_area_struct * area);
  331. int (*split)(struct vm_area_struct * area, unsigned long addr);
  332. int (*mremap)(struct vm_area_struct * area);
  333. int (*fault)(struct vm_fault *vmf);
  334. int (*huge_fault)(struct vm_fault *vmf, enum page_entry_size pe_size);
  335. void (*map_pages)(struct vm_fault *vmf,
  336. pgoff_t start_pgoff, pgoff_t end_pgoff);
  337. unsigned long (*pagesize)(struct vm_area_struct * area);
  338. /* notification that a previously read-only page is about to become
  339. * writable, if an error is returned it will cause a SIGBUS */
  340. int (*page_mkwrite)(struct vm_fault *vmf);
  341. /* same as page_mkwrite when using VM_PFNMAP|VM_MIXEDMAP */
  342. int (*pfn_mkwrite)(struct vm_fault *vmf);
  343. /* called by access_process_vm when get_user_pages() fails, typically
  344. * for use by special VMAs that can switch between memory and hardware
  345. */
  346. int (*access)(struct vm_area_struct *vma, unsigned long addr,
  347. void *buf, int len, int write);
  348. /* Called by the /proc/PID/maps code to ask the vma whether it
  349. * has a special name. Returning non-NULL will also cause this
  350. * vma to be dumped unconditionally. */
  351. const char *(*name)(struct vm_area_struct *vma);
  352. #ifdef CONFIG_NUMA
  353. /*
  354. * set_policy() op must add a reference to any non-NULL @new mempolicy
  355. * to hold the policy upon return. Caller should pass NULL @new to
  356. * remove a policy and fall back to surrounding context--i.e. do not
  357. * install a MPOL_DEFAULT policy, nor the task or system default
  358. * mempolicy.
  359. */
  360. int (*set_policy)(struct vm_area_struct *vma, struct mempolicy *new);
  361. /*
  362. * get_policy() op must add reference [mpol_get()] to any policy at
  363. * (vma,addr) marked as MPOL_SHARED. The shared policy infrastructure
  364. * in mm/mempolicy.c will do this automatically.
  365. * get_policy() must NOT add a ref if the policy at (vma,addr) is not
  366. * marked as MPOL_SHARED. vma policies are protected by the mmap_sem.
  367. * If no [shared/vma] mempolicy exists at the addr, get_policy() op
  368. * must return NULL--i.e., do not "fallback" to task or system default
  369. * policy.
  370. */
  371. struct mempolicy *(*get_policy)(struct vm_area_struct *vma,
  372. unsigned long addr);
  373. #endif
  374. /*
  375. * Called by vm_normal_page() for special PTEs to find the
  376. * page for @addr. This is useful if the default behavior
  377. * (using pte_page()) would not find the correct page.
  378. */
  379. struct page *(*find_special_page)(struct vm_area_struct *vma,
  380. unsigned long addr);
  381. };
  382. struct mmu_gather;
  383. struct inode;
  384. #define page_private(page) ((page)->private)
  385. #define set_page_private(page, v) ((page)->private = (v))
  386. #if !defined(__HAVE_ARCH_PTE_DEVMAP) || !defined(CONFIG_TRANSPARENT_HUGEPAGE)
  387. static inline int pmd_devmap(pmd_t pmd)
  388. {
  389. return 0;
  390. }
  391. static inline int pud_devmap(pud_t pud)
  392. {
  393. return 0;
  394. }
  395. static inline int pgd_devmap(pgd_t pgd)
  396. {
  397. return 0;
  398. }
  399. #endif
  400. /*
  401. * FIXME: take this include out, include page-flags.h in
  402. * files which need it (119 of them)
  403. */
  404. #include <linux/page-flags.h>
  405. #include <linux/huge_mm.h>
  406. /*
  407. * Methods to modify the page usage count.
  408. *
  409. * What counts for a page usage:
  410. * - cache mapping (page->mapping)
  411. * - private data (page->private)
  412. * - page mapped in a task's page tables, each mapping
  413. * is counted separately
  414. *
  415. * Also, many kernel routines increase the page count before a critical
  416. * routine so they can be sure the page doesn't go away from under them.
  417. */
  418. /*
  419. * Drop a ref, return true if the refcount fell to zero (the page has no users)
  420. */
  421. static inline int put_page_testzero(struct page *page)
  422. {
  423. VM_BUG_ON_PAGE(page_ref_count(page) == 0, page);
  424. return page_ref_dec_and_test(page);
  425. }
  426. /*
  427. * Try to grab a ref unless the page has a refcount of zero, return false if
  428. * that is the case.
  429. * This can be called when MMU is off so it must not access
  430. * any of the virtual mappings.
  431. */
  432. static inline int get_page_unless_zero(struct page *page)
  433. {
  434. return page_ref_add_unless(page, 1, 0);
  435. }
  436. extern int page_is_ram(unsigned long pfn);
  437. enum {
  438. REGION_INTERSECTS,
  439. REGION_DISJOINT,
  440. REGION_MIXED,
  441. };
  442. int region_intersects(resource_size_t offset, size_t size, unsigned long flags,
  443. unsigned long desc);
  444. /* Support for virtually mapped pages */
  445. struct page *vmalloc_to_page(const void *addr);
  446. unsigned long vmalloc_to_pfn(const void *addr);
  447. /*
  448. * Determine if an address is within the vmalloc range
  449. *
  450. * On nommu, vmalloc/vfree wrap through kmalloc/kfree directly, so there
  451. * is no special casing required.
  452. */
  453. static inline bool is_vmalloc_addr(const void *x)
  454. {
  455. #ifdef CONFIG_MMU
  456. unsigned long addr = (unsigned long)x;
  457. return addr >= VMALLOC_START && addr < VMALLOC_END;
  458. #else
  459. return false;
  460. #endif
  461. }
  462. #ifdef CONFIG_MMU
  463. extern int is_vmalloc_or_module_addr(const void *x);
  464. #else
  465. static inline int is_vmalloc_or_module_addr(const void *x)
  466. {
  467. return 0;
  468. }
  469. #endif
  470. extern void *kvmalloc_node(size_t size, gfp_t flags, int node);
  471. static inline void *kvmalloc(size_t size, gfp_t flags)
  472. {
  473. return kvmalloc_node(size, flags, NUMA_NO_NODE);
  474. }
  475. static inline void *kvzalloc_node(size_t size, gfp_t flags, int node)
  476. {
  477. return kvmalloc_node(size, flags | __GFP_ZERO, node);
  478. }
  479. static inline void *kvzalloc(size_t size, gfp_t flags)
  480. {
  481. return kvmalloc(size, flags | __GFP_ZERO);
  482. }
  483. static inline void *kvmalloc_array(size_t n, size_t size, gfp_t flags)
  484. {
  485. if (size != 0 && n > SIZE_MAX / size)
  486. return NULL;
  487. return kvmalloc(n * size, flags);
  488. }
  489. extern void kvfree(const void *addr);
  490. static inline atomic_t *compound_mapcount_ptr(struct page *page)
  491. {
  492. return &page[1].compound_mapcount;
  493. }
  494. static inline int compound_mapcount(struct page *page)
  495. {
  496. VM_BUG_ON_PAGE(!PageCompound(page), page);
  497. page = compound_head(page);
  498. return atomic_read(compound_mapcount_ptr(page)) + 1;
  499. }
  500. /*
  501. * The atomic page->_mapcount, starts from -1: so that transitions
  502. * both from it and to it can be tracked, using atomic_inc_and_test
  503. * and atomic_add_negative(-1).
  504. */
  505. static inline void page_mapcount_reset(struct page *page)
  506. {
  507. atomic_set(&(page)->_mapcount, -1);
  508. }
  509. int __page_mapcount(struct page *page);
  510. static inline int page_mapcount(struct page *page)
  511. {
  512. VM_BUG_ON_PAGE(PageSlab(page), page);
  513. if (unlikely(PageCompound(page)))
  514. return __page_mapcount(page);
  515. return atomic_read(&page->_mapcount) + 1;
  516. }
  517. #ifdef CONFIG_TRANSPARENT_HUGEPAGE
  518. int total_mapcount(struct page *page);
  519. int page_trans_huge_mapcount(struct page *page, int *total_mapcount);
  520. #else
  521. static inline int total_mapcount(struct page *page)
  522. {
  523. return page_mapcount(page);
  524. }
  525. static inline int page_trans_huge_mapcount(struct page *page,
  526. int *total_mapcount)
  527. {
  528. int mapcount = page_mapcount(page);
  529. if (total_mapcount)
  530. *total_mapcount = mapcount;
  531. return mapcount;
  532. }
  533. #endif
  534. static inline struct page *virt_to_head_page(const void *x)
  535. {
  536. struct page *page = virt_to_page(x);
  537. return compound_head(page);
  538. }
  539. void __put_page(struct page *page);
  540. void put_pages_list(struct list_head *pages);
  541. void split_page(struct page *page, unsigned int order);
  542. /*
  543. * Compound pages have a destructor function. Provide a
  544. * prototype for that function and accessor functions.
  545. * These are _only_ valid on the head of a compound page.
  546. */
  547. typedef void compound_page_dtor(struct page *);
  548. /* Keep the enum in sync with compound_page_dtors array in mm/page_alloc.c */
  549. enum compound_dtor_id {
  550. NULL_COMPOUND_DTOR,
  551. COMPOUND_PAGE_DTOR,
  552. #ifdef CONFIG_HUGETLB_PAGE
  553. HUGETLB_PAGE_DTOR,
  554. #endif
  555. #ifdef CONFIG_TRANSPARENT_HUGEPAGE
  556. TRANSHUGE_PAGE_DTOR,
  557. #endif
  558. NR_COMPOUND_DTORS,
  559. };
  560. extern compound_page_dtor * const compound_page_dtors[];
  561. static inline void set_compound_page_dtor(struct page *page,
  562. enum compound_dtor_id compound_dtor)
  563. {
  564. VM_BUG_ON_PAGE(compound_dtor >= NR_COMPOUND_DTORS, page);
  565. page[1].compound_dtor = compound_dtor;
  566. }
  567. static inline compound_page_dtor *get_compound_page_dtor(struct page *page)
  568. {
  569. VM_BUG_ON_PAGE(page[1].compound_dtor >= NR_COMPOUND_DTORS, page);
  570. return compound_page_dtors[page[1].compound_dtor];
  571. }
  572. static inline unsigned int compound_order(struct page *page)
  573. {
  574. if (!PageHead(page))
  575. return 0;
  576. return page[1].compound_order;
  577. }
  578. static inline void set_compound_order(struct page *page, unsigned int order)
  579. {
  580. page[1].compound_order = order;
  581. }
  582. void free_compound_page(struct page *page);
  583. #ifdef CONFIG_MMU
  584. /*
  585. * Do pte_mkwrite, but only if the vma says VM_WRITE. We do this when
  586. * servicing faults for write access. In the normal case, do always want
  587. * pte_mkwrite. But get_user_pages can cause write faults for mappings
  588. * that do not have writing enabled, when used by access_process_vm.
  589. */
  590. static inline pte_t maybe_mkwrite(pte_t pte, struct vm_area_struct *vma)
  591. {
  592. if (likely(vma->vm_flags & VM_WRITE))
  593. pte = pte_mkwrite(pte);
  594. return pte;
  595. }
  596. int alloc_set_pte(struct vm_fault *vmf, struct mem_cgroup *memcg,
  597. struct page *page);
  598. int finish_fault(struct vm_fault *vmf);
  599. int finish_mkwrite_fault(struct vm_fault *vmf);
  600. #endif
  601. /*
  602. * Multiple processes may "see" the same page. E.g. for untouched
  603. * mappings of /dev/null, all processes see the same page full of
  604. * zeroes, and text pages of executables and shared libraries have
  605. * only one copy in memory, at most, normally.
  606. *
  607. * For the non-reserved pages, page_count(page) denotes a reference count.
  608. * page_count() == 0 means the page is free. page->lru is then used for
  609. * freelist management in the buddy allocator.
  610. * page_count() > 0 means the page has been allocated.
  611. *
  612. * Pages are allocated by the slab allocator in order to provide memory
  613. * to kmalloc and kmem_cache_alloc. In this case, the management of the
  614. * page, and the fields in 'struct page' are the responsibility of mm/slab.c
  615. * unless a particular usage is carefully commented. (the responsibility of
  616. * freeing the kmalloc memory is the caller's, of course).
  617. *
  618. * A page may be used by anyone else who does a __get_free_page().
  619. * In this case, page_count still tracks the references, and should only
  620. * be used through the normal accessor functions. The top bits of page->flags
  621. * and page->virtual store page management information, but all other fields
  622. * are unused and could be used privately, carefully. The management of this
  623. * page is the responsibility of the one who allocated it, and those who have
  624. * subsequently been given references to it.
  625. *
  626. * The other pages (we may call them "pagecache pages") are completely
  627. * managed by the Linux memory manager: I/O, buffers, swapping etc.
  628. * The following discussion applies only to them.
  629. *
  630. * A pagecache page contains an opaque `private' member, which belongs to the
  631. * page's address_space. Usually, this is the address of a circular list of
  632. * the page's disk buffers. PG_private must be set to tell the VM to call
  633. * into the filesystem to release these pages.
  634. *
  635. * A page may belong to an inode's memory mapping. In this case, page->mapping
  636. * is the pointer to the inode, and page->index is the file offset of the page,
  637. * in units of PAGE_SIZE.
  638. *
  639. * If pagecache pages are not associated with an inode, they are said to be
  640. * anonymous pages. These may become associated with the swapcache, and in that
  641. * case PG_swapcache is set, and page->private is an offset into the swapcache.
  642. *
  643. * In either case (swapcache or inode backed), the pagecache itself holds one
  644. * reference to the page. Setting PG_private should also increment the
  645. * refcount. The each user mapping also has a reference to the page.
  646. *
  647. * The pagecache pages are stored in a per-mapping radix tree, which is
  648. * rooted at mapping->page_tree, and indexed by offset.
  649. * Where 2.4 and early 2.6 kernels kept dirty/clean pages in per-address_space
  650. * lists, we instead now tag pages as dirty/writeback in the radix tree.
  651. *
  652. * All pagecache pages may be subject to I/O:
  653. * - inode pages may need to be read from disk,
  654. * - inode pages which have been modified and are MAP_SHARED may need
  655. * to be written back to the inode on disk,
  656. * - anonymous pages (including MAP_PRIVATE file mappings) which have been
  657. * modified may need to be swapped out to swap space and (later) to be read
  658. * back into memory.
  659. */
  660. /*
  661. * The zone field is never updated after free_area_init_core()
  662. * sets it, so none of the operations on it need to be atomic.
  663. */
  664. /* Page flags: | [SECTION] | [NODE] | ZONE | [LAST_CPUPID] | ... | FLAGS | */
  665. #define SECTIONS_PGOFF ((sizeof(unsigned long)*8) - SECTIONS_WIDTH)
  666. #define NODES_PGOFF (SECTIONS_PGOFF - NODES_WIDTH)
  667. #define ZONES_PGOFF (NODES_PGOFF - ZONES_WIDTH)
  668. #define LAST_CPUPID_PGOFF (ZONES_PGOFF - LAST_CPUPID_WIDTH)
  669. /*
  670. * Define the bit shifts to access each section. For non-existent
  671. * sections we define the shift as 0; that plus a 0 mask ensures
  672. * the compiler will optimise away reference to them.
  673. */
  674. #define SECTIONS_PGSHIFT (SECTIONS_PGOFF * (SECTIONS_WIDTH != 0))
  675. #define NODES_PGSHIFT (NODES_PGOFF * (NODES_WIDTH != 0))
  676. #define ZONES_PGSHIFT (ZONES_PGOFF * (ZONES_WIDTH != 0))
  677. #define LAST_CPUPID_PGSHIFT (LAST_CPUPID_PGOFF * (LAST_CPUPID_WIDTH != 0))
  678. /* NODE:ZONE or SECTION:ZONE is used to ID a zone for the buddy allocator */
  679. #ifdef NODE_NOT_IN_PAGE_FLAGS
  680. #define ZONEID_SHIFT (SECTIONS_SHIFT + ZONES_SHIFT)
  681. #define ZONEID_PGOFF ((SECTIONS_PGOFF < ZONES_PGOFF)? \
  682. SECTIONS_PGOFF : ZONES_PGOFF)
  683. #else
  684. #define ZONEID_SHIFT (NODES_SHIFT + ZONES_SHIFT)
  685. #define ZONEID_PGOFF ((NODES_PGOFF < ZONES_PGOFF)? \
  686. NODES_PGOFF : ZONES_PGOFF)
  687. #endif
  688. #define ZONEID_PGSHIFT (ZONEID_PGOFF * (ZONEID_SHIFT != 0))
  689. #if SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > BITS_PER_LONG - NR_PAGEFLAGS
  690. #error SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > BITS_PER_LONG - NR_PAGEFLAGS
  691. #endif
  692. #define ZONES_MASK ((1UL << ZONES_WIDTH) - 1)
  693. #define NODES_MASK ((1UL << NODES_WIDTH) - 1)
  694. #define SECTIONS_MASK ((1UL << SECTIONS_WIDTH) - 1)
  695. #define LAST_CPUPID_MASK ((1UL << LAST_CPUPID_SHIFT) - 1)
  696. #define ZONEID_MASK ((1UL << ZONEID_SHIFT) - 1)
  697. static inline enum zone_type page_zonenum(const struct page *page)
  698. {
  699. return (page->flags >> ZONES_PGSHIFT) & ZONES_MASK;
  700. }
  701. #ifdef CONFIG_ZONE_DEVICE
  702. static inline bool is_zone_device_page(const struct page *page)
  703. {
  704. return page_zonenum(page) == ZONE_DEVICE;
  705. }
  706. #else
  707. static inline bool is_zone_device_page(const struct page *page)
  708. {
  709. return false;
  710. }
  711. #endif
  712. #if defined(CONFIG_DEVICE_PRIVATE) || defined(CONFIG_DEVICE_PUBLIC)
  713. void put_zone_device_private_or_public_page(struct page *page);
  714. DECLARE_STATIC_KEY_FALSE(device_private_key);
  715. #define IS_HMM_ENABLED static_branch_unlikely(&device_private_key)
  716. static inline bool is_device_private_page(const struct page *page);
  717. static inline bool is_device_public_page(const struct page *page);
  718. #else /* CONFIG_DEVICE_PRIVATE || CONFIG_DEVICE_PUBLIC */
  719. static inline void put_zone_device_private_or_public_page(struct page *page)
  720. {
  721. }
  722. #define IS_HMM_ENABLED 0
  723. static inline bool is_device_private_page(const struct page *page)
  724. {
  725. return false;
  726. }
  727. static inline bool is_device_public_page(const struct page *page)
  728. {
  729. return false;
  730. }
  731. #endif /* CONFIG_DEVICE_PRIVATE || CONFIG_DEVICE_PUBLIC */
  732. static inline void get_page(struct page *page)
  733. {
  734. page = compound_head(page);
  735. /*
  736. * Getting a normal page or the head of a compound page
  737. * requires to already have an elevated page->_refcount.
  738. */
  739. VM_BUG_ON_PAGE(page_ref_count(page) <= 0, page);
  740. page_ref_inc(page);
  741. }
  742. static inline void put_page(struct page *page)
  743. {
  744. page = compound_head(page);
  745. /*
  746. * For private device pages we need to catch refcount transition from
  747. * 2 to 1, when refcount reach one it means the private device page is
  748. * free and we need to inform the device driver through callback. See
  749. * include/linux/memremap.h and HMM for details.
  750. */
  751. if (IS_HMM_ENABLED && unlikely(is_device_private_page(page) ||
  752. unlikely(is_device_public_page(page)))) {
  753. put_zone_device_private_or_public_page(page);
  754. return;
  755. }
  756. if (put_page_testzero(page))
  757. __put_page(page);
  758. }
  759. #if defined(CONFIG_SPARSEMEM) && !defined(CONFIG_SPARSEMEM_VMEMMAP)
  760. #define SECTION_IN_PAGE_FLAGS
  761. #endif
  762. /*
  763. * The identification function is mainly used by the buddy allocator for
  764. * determining if two pages could be buddies. We are not really identifying
  765. * the zone since we could be using the section number id if we do not have
  766. * node id available in page flags.
  767. * We only guarantee that it will return the same value for two combinable
  768. * pages in a zone.
  769. */
  770. static inline int page_zone_id(struct page *page)
  771. {
  772. return (page->flags >> ZONEID_PGSHIFT) & ZONEID_MASK;
  773. }
  774. static inline int zone_to_nid(struct zone *zone)
  775. {
  776. #ifdef CONFIG_NUMA
  777. return zone->node;
  778. #else
  779. return 0;
  780. #endif
  781. }
  782. #ifdef NODE_NOT_IN_PAGE_FLAGS
  783. extern int page_to_nid(const struct page *page);
  784. #else
  785. static inline int page_to_nid(const struct page *page)
  786. {
  787. struct page *p = (struct page *)page;
  788. return (PF_POISONED_CHECK(p)->flags >> NODES_PGSHIFT) & NODES_MASK;
  789. }
  790. #endif
  791. #ifdef CONFIG_NUMA_BALANCING
  792. static inline int cpu_pid_to_cpupid(int cpu, int pid)
  793. {
  794. return ((cpu & LAST__CPU_MASK) << LAST__PID_SHIFT) | (pid & LAST__PID_MASK);
  795. }
  796. static inline int cpupid_to_pid(int cpupid)
  797. {
  798. return cpupid & LAST__PID_MASK;
  799. }
  800. static inline int cpupid_to_cpu(int cpupid)
  801. {
  802. return (cpupid >> LAST__PID_SHIFT) & LAST__CPU_MASK;
  803. }
  804. static inline int cpupid_to_nid(int cpupid)
  805. {
  806. return cpu_to_node(cpupid_to_cpu(cpupid));
  807. }
  808. static inline bool cpupid_pid_unset(int cpupid)
  809. {
  810. return cpupid_to_pid(cpupid) == (-1 & LAST__PID_MASK);
  811. }
  812. static inline bool cpupid_cpu_unset(int cpupid)
  813. {
  814. return cpupid_to_cpu(cpupid) == (-1 & LAST__CPU_MASK);
  815. }
  816. static inline bool __cpupid_match_pid(pid_t task_pid, int cpupid)
  817. {
  818. return (task_pid & LAST__PID_MASK) == cpupid_to_pid(cpupid);
  819. }
  820. #define cpupid_match_pid(task, cpupid) __cpupid_match_pid(task->pid, cpupid)
  821. #ifdef LAST_CPUPID_NOT_IN_PAGE_FLAGS
  822. static inline int page_cpupid_xchg_last(struct page *page, int cpupid)
  823. {
  824. return xchg(&page->_last_cpupid, cpupid & LAST_CPUPID_MASK);
  825. }
  826. static inline int page_cpupid_last(struct page *page)
  827. {
  828. return page->_last_cpupid;
  829. }
  830. static inline void page_cpupid_reset_last(struct page *page)
  831. {
  832. page->_last_cpupid = -1 & LAST_CPUPID_MASK;
  833. }
  834. #else
  835. static inline int page_cpupid_last(struct page *page)
  836. {
  837. return (page->flags >> LAST_CPUPID_PGSHIFT) & LAST_CPUPID_MASK;
  838. }
  839. extern int page_cpupid_xchg_last(struct page *page, int cpupid);
  840. static inline void page_cpupid_reset_last(struct page *page)
  841. {
  842. page->flags |= LAST_CPUPID_MASK << LAST_CPUPID_PGSHIFT;
  843. }
  844. #endif /* LAST_CPUPID_NOT_IN_PAGE_FLAGS */
  845. #else /* !CONFIG_NUMA_BALANCING */
  846. static inline int page_cpupid_xchg_last(struct page *page, int cpupid)
  847. {
  848. return page_to_nid(page); /* XXX */
  849. }
  850. static inline int page_cpupid_last(struct page *page)
  851. {
  852. return page_to_nid(page); /* XXX */
  853. }
  854. static inline int cpupid_to_nid(int cpupid)
  855. {
  856. return -1;
  857. }
  858. static inline int cpupid_to_pid(int cpupid)
  859. {
  860. return -1;
  861. }
  862. static inline int cpupid_to_cpu(int cpupid)
  863. {
  864. return -1;
  865. }
  866. static inline int cpu_pid_to_cpupid(int nid, int pid)
  867. {
  868. return -1;
  869. }
  870. static inline bool cpupid_pid_unset(int cpupid)
  871. {
  872. return 1;
  873. }
  874. static inline void page_cpupid_reset_last(struct page *page)
  875. {
  876. }
  877. static inline bool cpupid_match_pid(struct task_struct *task, int cpupid)
  878. {
  879. return false;
  880. }
  881. #endif /* CONFIG_NUMA_BALANCING */
  882. static inline struct zone *page_zone(const struct page *page)
  883. {
  884. return &NODE_DATA(page_to_nid(page))->node_zones[page_zonenum(page)];
  885. }
  886. static inline pg_data_t *page_pgdat(const struct page *page)
  887. {
  888. return NODE_DATA(page_to_nid(page));
  889. }
  890. #ifdef SECTION_IN_PAGE_FLAGS
  891. static inline void set_page_section(struct page *page, unsigned long section)
  892. {
  893. page->flags &= ~(SECTIONS_MASK << SECTIONS_PGSHIFT);
  894. page->flags |= (section & SECTIONS_MASK) << SECTIONS_PGSHIFT;
  895. }
  896. static inline unsigned long page_to_section(const struct page *page)
  897. {
  898. return (page->flags >> SECTIONS_PGSHIFT) & SECTIONS_MASK;
  899. }
  900. #endif
  901. static inline void set_page_zone(struct page *page, enum zone_type zone)
  902. {
  903. page->flags &= ~(ZONES_MASK << ZONES_PGSHIFT);
  904. page->flags |= (zone & ZONES_MASK) << ZONES_PGSHIFT;
  905. }
  906. static inline void set_page_node(struct page *page, unsigned long node)
  907. {
  908. page->flags &= ~(NODES_MASK << NODES_PGSHIFT);
  909. page->flags |= (node & NODES_MASK) << NODES_PGSHIFT;
  910. }
  911. static inline void set_page_links(struct page *page, enum zone_type zone,
  912. unsigned long node, unsigned long pfn)
  913. {
  914. set_page_zone(page, zone);
  915. set_page_node(page, node);
  916. #ifdef SECTION_IN_PAGE_FLAGS
  917. set_page_section(page, pfn_to_section_nr(pfn));
  918. #endif
  919. }
  920. #ifdef CONFIG_MEMCG
  921. static inline struct mem_cgroup *page_memcg(struct page *page)
  922. {
  923. return page->mem_cgroup;
  924. }
  925. static inline struct mem_cgroup *page_memcg_rcu(struct page *page)
  926. {
  927. WARN_ON_ONCE(!rcu_read_lock_held());
  928. return READ_ONCE(page->mem_cgroup);
  929. }
  930. #else
  931. static inline struct mem_cgroup *page_memcg(struct page *page)
  932. {
  933. return NULL;
  934. }
  935. static inline struct mem_cgroup *page_memcg_rcu(struct page *page)
  936. {
  937. WARN_ON_ONCE(!rcu_read_lock_held());
  938. return NULL;
  939. }
  940. #endif
  941. /*
  942. * Some inline functions in vmstat.h depend on page_zone()
  943. */
  944. #include <linux/vmstat.h>
  945. static __always_inline void *lowmem_page_address(const struct page *page)
  946. {
  947. return page_to_virt(page);
  948. }
  949. #if defined(CONFIG_HIGHMEM) && !defined(WANT_PAGE_VIRTUAL)
  950. #define HASHED_PAGE_VIRTUAL
  951. #endif
  952. #if defined(WANT_PAGE_VIRTUAL)
  953. static inline void *page_address(const struct page *page)
  954. {
  955. return page->virtual;
  956. }
  957. static inline void set_page_address(struct page *page, void *address)
  958. {
  959. page->virtual = address;
  960. }
  961. #define page_address_init() do { } while(0)
  962. #endif
  963. #if defined(HASHED_PAGE_VIRTUAL)
  964. void *page_address(const struct page *page);
  965. void set_page_address(struct page *page, void *virtual);
  966. void page_address_init(void);
  967. #endif
  968. #if !defined(HASHED_PAGE_VIRTUAL) && !defined(WANT_PAGE_VIRTUAL)
  969. #define page_address(page) lowmem_page_address(page)
  970. #define set_page_address(page, address) do { } while(0)
  971. #define page_address_init() do { } while(0)
  972. #endif
  973. extern void *page_rmapping(struct page *page);
  974. extern struct anon_vma *page_anon_vma(struct page *page);
  975. extern struct address_space *page_mapping(struct page *page);
  976. extern struct address_space *__page_file_mapping(struct page *);
  977. static inline
  978. struct address_space *page_file_mapping(struct page *page)
  979. {
  980. if (unlikely(PageSwapCache(page)))
  981. return __page_file_mapping(page);
  982. return page->mapping;
  983. }
  984. extern pgoff_t __page_file_index(struct page *page);
  985. /*
  986. * Return the pagecache index of the passed page. Regular pagecache pages
  987. * use ->index whereas swapcache pages use swp_offset(->private)
  988. */
  989. static inline pgoff_t page_index(struct page *page)
  990. {
  991. if (unlikely(PageSwapCache(page)))
  992. return __page_file_index(page);
  993. return page->index;
  994. }
  995. bool page_mapped(struct page *page);
  996. struct address_space *page_mapping(struct page *page);
  997. struct address_space *page_mapping_file(struct page *page);
  998. /*
  999. * Return true only if the page has been allocated with
  1000. * ALLOC_NO_WATERMARKS and the low watermark was not
  1001. * met implying that the system is under some pressure.
  1002. */
  1003. static inline bool page_is_pfmemalloc(struct page *page)
  1004. {
  1005. /*
  1006. * Page index cannot be this large so this must be
  1007. * a pfmemalloc page.
  1008. */
  1009. return page->index == -1UL;
  1010. }
  1011. /*
  1012. * Only to be called by the page allocator on a freshly allocated
  1013. * page.
  1014. */
  1015. static inline void set_page_pfmemalloc(struct page *page)
  1016. {
  1017. page->index = -1UL;
  1018. }
  1019. static inline void clear_page_pfmemalloc(struct page *page)
  1020. {
  1021. page->index = 0;
  1022. }
  1023. /*
  1024. * Different kinds of faults, as returned by handle_mm_fault().
  1025. * Used to decide whether a process gets delivered SIGBUS or
  1026. * just gets major/minor fault counters bumped up.
  1027. */
  1028. #define VM_FAULT_OOM 0x0001
  1029. #define VM_FAULT_SIGBUS 0x0002
  1030. #define VM_FAULT_MAJOR 0x0004
  1031. #define VM_FAULT_WRITE 0x0008 /* Special case for get_user_pages */
  1032. #define VM_FAULT_HWPOISON 0x0010 /* Hit poisoned small page */
  1033. #define VM_FAULT_HWPOISON_LARGE 0x0020 /* Hit poisoned large page. Index encoded in upper bits */
  1034. #define VM_FAULT_SIGSEGV 0x0040
  1035. #define VM_FAULT_NOPAGE 0x0100 /* ->fault installed the pte, not return page */
  1036. #define VM_FAULT_LOCKED 0x0200 /* ->fault locked the returned page */
  1037. #define VM_FAULT_RETRY 0x0400 /* ->fault blocked, must retry */
  1038. #define VM_FAULT_FALLBACK 0x0800 /* huge page fault failed, fall back to small */
  1039. #define VM_FAULT_DONE_COW 0x1000 /* ->fault has fully handled COW */
  1040. #define VM_FAULT_NEEDDSYNC 0x2000 /* ->fault did not modify page tables
  1041. * and needs fsync() to complete (for
  1042. * synchronous page faults in DAX) */
  1043. #define VM_FAULT_ERROR (VM_FAULT_OOM | VM_FAULT_SIGBUS | VM_FAULT_SIGSEGV | \
  1044. VM_FAULT_HWPOISON | VM_FAULT_HWPOISON_LARGE | \
  1045. VM_FAULT_FALLBACK)
  1046. #define VM_FAULT_RESULT_TRACE \
  1047. { VM_FAULT_OOM, "OOM" }, \
  1048. { VM_FAULT_SIGBUS, "SIGBUS" }, \
  1049. { VM_FAULT_MAJOR, "MAJOR" }, \
  1050. { VM_FAULT_WRITE, "WRITE" }, \
  1051. { VM_FAULT_HWPOISON, "HWPOISON" }, \
  1052. { VM_FAULT_HWPOISON_LARGE, "HWPOISON_LARGE" }, \
  1053. { VM_FAULT_SIGSEGV, "SIGSEGV" }, \
  1054. { VM_FAULT_NOPAGE, "NOPAGE" }, \
  1055. { VM_FAULT_LOCKED, "LOCKED" }, \
  1056. { VM_FAULT_RETRY, "RETRY" }, \
  1057. { VM_FAULT_FALLBACK, "FALLBACK" }, \
  1058. { VM_FAULT_DONE_COW, "DONE_COW" }, \
  1059. { VM_FAULT_NEEDDSYNC, "NEEDDSYNC" }
  1060. /* Encode hstate index for a hwpoisoned large page */
  1061. #define VM_FAULT_SET_HINDEX(x) ((x) << 12)
  1062. #define VM_FAULT_GET_HINDEX(x) (((x) >> 12) & 0xf)
  1063. /*
  1064. * Can be called by the pagefault handler when it gets a VM_FAULT_OOM.
  1065. */
  1066. extern void pagefault_out_of_memory(void);
  1067. #define offset_in_page(p) ((unsigned long)(p) & ~PAGE_MASK)
  1068. /*
  1069. * Flags passed to show_mem() and show_free_areas() to suppress output in
  1070. * various contexts.
  1071. */
  1072. #define SHOW_MEM_FILTER_NODES (0x0001u) /* disallowed nodes */
  1073. extern void show_free_areas(unsigned int flags, nodemask_t *nodemask);
  1074. extern bool can_do_mlock(void);
  1075. extern int user_shm_lock(size_t, struct user_struct *);
  1076. extern void user_shm_unlock(size_t, struct user_struct *);
  1077. /*
  1078. * Parameter block passed down to zap_pte_range in exceptional cases.
  1079. */
  1080. struct zap_details {
  1081. struct address_space *check_mapping; /* Check page->mapping if set */
  1082. pgoff_t first_index; /* Lowest page->index to unmap */
  1083. pgoff_t last_index; /* Highest page->index to unmap */
  1084. };
  1085. struct page *_vm_normal_page(struct vm_area_struct *vma, unsigned long addr,
  1086. pte_t pte, bool with_public_device);
  1087. #define vm_normal_page(vma, addr, pte) _vm_normal_page(vma, addr, pte, false)
  1088. struct page *vm_normal_page_pmd(struct vm_area_struct *vma, unsigned long addr,
  1089. pmd_t pmd);
  1090. int zap_vma_ptes(struct vm_area_struct *vma, unsigned long address,
  1091. unsigned long size);
  1092. void zap_page_range(struct vm_area_struct *vma, unsigned long address,
  1093. unsigned long size);
  1094. void unmap_vmas(struct mmu_gather *tlb, struct vm_area_struct *start_vma,
  1095. unsigned long start, unsigned long end);
  1096. /**
  1097. * mm_walk - callbacks for walk_page_range
  1098. * @pud_entry: if set, called for each non-empty PUD (2nd-level) entry
  1099. * this handler should only handle pud_trans_huge() puds.
  1100. * the pmd_entry or pte_entry callbacks will be used for
  1101. * regular PUDs.
  1102. * @pmd_entry: if set, called for each non-empty PMD (3rd-level) entry
  1103. * this handler is required to be able to handle
  1104. * pmd_trans_huge() pmds. They may simply choose to
  1105. * split_huge_page() instead of handling it explicitly.
  1106. * @pte_entry: if set, called for each non-empty PTE (4th-level) entry
  1107. * @pte_hole: if set, called for each hole at all levels
  1108. * @hugetlb_entry: if set, called for each hugetlb entry
  1109. * @test_walk: caller specific callback function to determine whether
  1110. * we walk over the current vma or not. Returning 0
  1111. * value means "do page table walk over the current vma,"
  1112. * and a negative one means "abort current page table walk
  1113. * right now." 1 means "skip the current vma."
  1114. * @mm: mm_struct representing the target process of page table walk
  1115. * @vma: vma currently walked (NULL if walking outside vmas)
  1116. * @private: private data for callbacks' usage
  1117. *
  1118. * (see the comment on walk_page_range() for more details)
  1119. */
  1120. struct mm_walk {
  1121. int (*pud_entry)(pud_t *pud, unsigned long addr,
  1122. unsigned long next, struct mm_walk *walk);
  1123. int (*pmd_entry)(pmd_t *pmd, unsigned long addr,
  1124. unsigned long next, struct mm_walk *walk);
  1125. int (*pte_entry)(pte_t *pte, unsigned long addr,
  1126. unsigned long next, struct mm_walk *walk);
  1127. int (*pte_hole)(unsigned long addr, unsigned long next,
  1128. struct mm_walk *walk);
  1129. int (*hugetlb_entry)(pte_t *pte, unsigned long hmask,
  1130. unsigned long addr, unsigned long next,
  1131. struct mm_walk *walk);
  1132. int (*test_walk)(unsigned long addr, unsigned long next,
  1133. struct mm_walk *walk);
  1134. struct mm_struct *mm;
  1135. struct vm_area_struct *vma;
  1136. void *private;
  1137. };
  1138. int walk_page_range(unsigned long addr, unsigned long end,
  1139. struct mm_walk *walk);
  1140. int walk_page_vma(struct vm_area_struct *vma, struct mm_walk *walk);
  1141. void free_pgd_range(struct mmu_gather *tlb, unsigned long addr,
  1142. unsigned long end, unsigned long floor, unsigned long ceiling);
  1143. int copy_page_range(struct mm_struct *dst, struct mm_struct *src,
  1144. struct vm_area_struct *vma);
  1145. int follow_pte_pmd(struct mm_struct *mm, unsigned long address,
  1146. unsigned long *start, unsigned long *end,
  1147. pte_t **ptepp, pmd_t **pmdpp, spinlock_t **ptlp);
  1148. int follow_pfn(struct vm_area_struct *vma, unsigned long address,
  1149. unsigned long *pfn);
  1150. int follow_phys(struct vm_area_struct *vma, unsigned long address,
  1151. unsigned int flags, unsigned long *prot, resource_size_t *phys);
  1152. int generic_access_phys(struct vm_area_struct *vma, unsigned long addr,
  1153. void *buf, int len, int write);
  1154. extern void truncate_pagecache(struct inode *inode, loff_t new);
  1155. extern void truncate_setsize(struct inode *inode, loff_t newsize);
  1156. void pagecache_isize_extended(struct inode *inode, loff_t from, loff_t to);
  1157. void truncate_pagecache_range(struct inode *inode, loff_t offset, loff_t end);
  1158. int truncate_inode_page(struct address_space *mapping, struct page *page);
  1159. int generic_error_remove_page(struct address_space *mapping, struct page *page);
  1160. int invalidate_inode_page(struct page *page);
  1161. #ifdef CONFIG_MMU
  1162. extern int handle_mm_fault(struct vm_area_struct *vma, unsigned long address,
  1163. unsigned int flags);
  1164. extern int fixup_user_fault(struct task_struct *tsk, struct mm_struct *mm,
  1165. unsigned long address, unsigned int fault_flags,
  1166. bool *unlocked);
  1167. void unmap_mapping_pages(struct address_space *mapping,
  1168. pgoff_t start, pgoff_t nr, bool even_cows);
  1169. void unmap_mapping_range(struct address_space *mapping,
  1170. loff_t const holebegin, loff_t const holelen, int even_cows);
  1171. #else
  1172. static inline int handle_mm_fault(struct vm_area_struct *vma,
  1173. unsigned long address, unsigned int flags)
  1174. {
  1175. /* should never happen if there's no MMU */
  1176. BUG();
  1177. return VM_FAULT_SIGBUS;
  1178. }
  1179. static inline int fixup_user_fault(struct task_struct *tsk,
  1180. struct mm_struct *mm, unsigned long address,
  1181. unsigned int fault_flags, bool *unlocked)
  1182. {
  1183. /* should never happen if there's no MMU */
  1184. BUG();
  1185. return -EFAULT;
  1186. }
  1187. static inline void unmap_mapping_pages(struct address_space *mapping,
  1188. pgoff_t start, pgoff_t nr, bool even_cows) { }
  1189. static inline void unmap_mapping_range(struct address_space *mapping,
  1190. loff_t const holebegin, loff_t const holelen, int even_cows) { }
  1191. #endif
  1192. static inline void unmap_shared_mapping_range(struct address_space *mapping,
  1193. loff_t const holebegin, loff_t const holelen)
  1194. {
  1195. unmap_mapping_range(mapping, holebegin, holelen, 0);
  1196. }
  1197. extern int access_process_vm(struct task_struct *tsk, unsigned long addr,
  1198. void *buf, int len, unsigned int gup_flags);
  1199. extern int access_remote_vm(struct mm_struct *mm, unsigned long addr,
  1200. void *buf, int len, unsigned int gup_flags);
  1201. extern int __access_remote_vm(struct task_struct *tsk, struct mm_struct *mm,
  1202. unsigned long addr, void *buf, int len, unsigned int gup_flags);
  1203. long get_user_pages_remote(struct task_struct *tsk, struct mm_struct *mm,
  1204. unsigned long start, unsigned long nr_pages,
  1205. unsigned int gup_flags, struct page **pages,
  1206. struct vm_area_struct **vmas, int *locked);
  1207. long get_user_pages(unsigned long start, unsigned long nr_pages,
  1208. unsigned int gup_flags, struct page **pages,
  1209. struct vm_area_struct **vmas);
  1210. long get_user_pages_locked(unsigned long start, unsigned long nr_pages,
  1211. unsigned int gup_flags, struct page **pages, int *locked);
  1212. long get_user_pages_unlocked(unsigned long start, unsigned long nr_pages,
  1213. struct page **pages, unsigned int gup_flags);
  1214. #ifdef CONFIG_FS_DAX
  1215. long get_user_pages_longterm(unsigned long start, unsigned long nr_pages,
  1216. unsigned int gup_flags, struct page **pages,
  1217. struct vm_area_struct **vmas);
  1218. #else
  1219. static inline long get_user_pages_longterm(unsigned long start,
  1220. unsigned long nr_pages, unsigned int gup_flags,
  1221. struct page **pages, struct vm_area_struct **vmas)
  1222. {
  1223. return get_user_pages(start, nr_pages, gup_flags, pages, vmas);
  1224. }
  1225. #endif /* CONFIG_FS_DAX */
  1226. int get_user_pages_fast(unsigned long start, int nr_pages, int write,
  1227. struct page **pages);
  1228. /* Container for pinned pfns / pages */
  1229. struct frame_vector {
  1230. unsigned int nr_allocated; /* Number of frames we have space for */
  1231. unsigned int nr_frames; /* Number of frames stored in ptrs array */
  1232. bool got_ref; /* Did we pin pages by getting page ref? */
  1233. bool is_pfns; /* Does array contain pages or pfns? */
  1234. void *ptrs[0]; /* Array of pinned pfns / pages. Use
  1235. * pfns_vector_pages() or pfns_vector_pfns()
  1236. * for access */
  1237. };
  1238. struct frame_vector *frame_vector_create(unsigned int nr_frames);
  1239. void frame_vector_destroy(struct frame_vector *vec);
  1240. int get_vaddr_frames(unsigned long start, unsigned int nr_pfns,
  1241. unsigned int gup_flags, struct frame_vector *vec);
  1242. void put_vaddr_frames(struct frame_vector *vec);
  1243. int frame_vector_to_pages(struct frame_vector *vec);
  1244. void frame_vector_to_pfns(struct frame_vector *vec);
  1245. static inline unsigned int frame_vector_count(struct frame_vector *vec)
  1246. {
  1247. return vec->nr_frames;
  1248. }
  1249. static inline struct page **frame_vector_pages(struct frame_vector *vec)
  1250. {
  1251. if (vec->is_pfns) {
  1252. int err = frame_vector_to_pages(vec);
  1253. if (err)
  1254. return ERR_PTR(err);
  1255. }
  1256. return (struct page **)(vec->ptrs);
  1257. }
  1258. static inline unsigned long *frame_vector_pfns(struct frame_vector *vec)
  1259. {
  1260. if (!vec->is_pfns)
  1261. frame_vector_to_pfns(vec);
  1262. return (unsigned long *)(vec->ptrs);
  1263. }
  1264. struct kvec;
  1265. int get_kernel_pages(const struct kvec *iov, int nr_pages, int write,
  1266. struct page **pages);
  1267. int get_kernel_page(unsigned long start, int write, struct page **pages);
  1268. struct page *get_dump_page(unsigned long addr);
  1269. extern int try_to_release_page(struct page * page, gfp_t gfp_mask);
  1270. extern void do_invalidatepage(struct page *page, unsigned int offset,
  1271. unsigned int length);
  1272. int __set_page_dirty_nobuffers(struct page *page);
  1273. int __set_page_dirty_no_writeback(struct page *page);
  1274. int redirty_page_for_writepage(struct writeback_control *wbc,
  1275. struct page *page);
  1276. void account_page_dirtied(struct page *page, struct address_space *mapping);
  1277. void account_page_cleaned(struct page *page, struct address_space *mapping,
  1278. struct bdi_writeback *wb);
  1279. int set_page_dirty(struct page *page);
  1280. int set_page_dirty_lock(struct page *page);
  1281. void __cancel_dirty_page(struct page *page);
  1282. static inline void cancel_dirty_page(struct page *page)
  1283. {
  1284. /* Avoid atomic ops, locking, etc. when not actually needed. */
  1285. if (PageDirty(page))
  1286. __cancel_dirty_page(page);
  1287. }
  1288. int clear_page_dirty_for_io(struct page *page);
  1289. int get_cmdline(struct task_struct *task, char *buffer, int buflen);
  1290. static inline bool vma_is_anonymous(struct vm_area_struct *vma)
  1291. {
  1292. return !vma->vm_ops;
  1293. }
  1294. #ifdef CONFIG_SHMEM
  1295. /*
  1296. * The vma_is_shmem is not inline because it is used only by slow
  1297. * paths in userfault.
  1298. */
  1299. bool vma_is_shmem(struct vm_area_struct *vma);
  1300. #else
  1301. static inline bool vma_is_shmem(struct vm_area_struct *vma) { return false; }
  1302. #endif
  1303. int vma_is_stack_for_current(struct vm_area_struct *vma);
  1304. extern unsigned long move_page_tables(struct vm_area_struct *vma,
  1305. unsigned long old_addr, struct vm_area_struct *new_vma,
  1306. unsigned long new_addr, unsigned long len,
  1307. bool need_rmap_locks);
  1308. extern unsigned long change_protection(struct vm_area_struct *vma, unsigned long start,
  1309. unsigned long end, pgprot_t newprot,
  1310. int dirty_accountable, int prot_numa);
  1311. extern int mprotect_fixup(struct vm_area_struct *vma,
  1312. struct vm_area_struct **pprev, unsigned long start,
  1313. unsigned long end, unsigned long newflags);
  1314. /*
  1315. * doesn't attempt to fault and will return short.
  1316. */
  1317. int __get_user_pages_fast(unsigned long start, int nr_pages, int write,
  1318. struct page **pages);
  1319. /*
  1320. * per-process(per-mm_struct) statistics.
  1321. */
  1322. static inline unsigned long get_mm_counter(struct mm_struct *mm, int member)
  1323. {
  1324. long val = atomic_long_read(&mm->rss_stat.count[member]);
  1325. #ifdef SPLIT_RSS_COUNTING
  1326. /*
  1327. * counter is updated in asynchronous manner and may go to minus.
  1328. * But it's never be expected number for users.
  1329. */
  1330. if (val < 0)
  1331. val = 0;
  1332. #endif
  1333. return (unsigned long)val;
  1334. }
  1335. static inline void add_mm_counter(struct mm_struct *mm, int member, long value)
  1336. {
  1337. atomic_long_add(value, &mm->rss_stat.count[member]);
  1338. }
  1339. static inline void inc_mm_counter(struct mm_struct *mm, int member)
  1340. {
  1341. atomic_long_inc(&mm->rss_stat.count[member]);
  1342. }
  1343. static inline void dec_mm_counter(struct mm_struct *mm, int member)
  1344. {
  1345. atomic_long_dec(&mm->rss_stat.count[member]);
  1346. }
  1347. /* Optimized variant when page is already known not to be PageAnon */
  1348. static inline int mm_counter_file(struct page *page)
  1349. {
  1350. if (PageSwapBacked(page))
  1351. return MM_SHMEMPAGES;
  1352. return MM_FILEPAGES;
  1353. }
  1354. static inline int mm_counter(struct page *page)
  1355. {
  1356. if (PageAnon(page))
  1357. return MM_ANONPAGES;
  1358. return mm_counter_file(page);
  1359. }
  1360. static inline unsigned long get_mm_rss(struct mm_struct *mm)
  1361. {
  1362. return get_mm_counter(mm, MM_FILEPAGES) +
  1363. get_mm_counter(mm, MM_ANONPAGES) +
  1364. get_mm_counter(mm, MM_SHMEMPAGES);
  1365. }
  1366. static inline unsigned long get_mm_hiwater_rss(struct mm_struct *mm)
  1367. {
  1368. return max(mm->hiwater_rss, get_mm_rss(mm));
  1369. }
  1370. static inline unsigned long get_mm_hiwater_vm(struct mm_struct *mm)
  1371. {
  1372. return max(mm->hiwater_vm, mm->total_vm);
  1373. }
  1374. static inline void update_hiwater_rss(struct mm_struct *mm)
  1375. {
  1376. unsigned long _rss = get_mm_rss(mm);
  1377. if ((mm)->hiwater_rss < _rss)
  1378. (mm)->hiwater_rss = _rss;
  1379. }
  1380. static inline void update_hiwater_vm(struct mm_struct *mm)
  1381. {
  1382. if (mm->hiwater_vm < mm->total_vm)
  1383. mm->hiwater_vm = mm->total_vm;
  1384. }
  1385. static inline void reset_mm_hiwater_rss(struct mm_struct *mm)
  1386. {
  1387. mm->hiwater_rss = get_mm_rss(mm);
  1388. }
  1389. static inline void setmax_mm_hiwater_rss(unsigned long *maxrss,
  1390. struct mm_struct *mm)
  1391. {
  1392. unsigned long hiwater_rss = get_mm_hiwater_rss(mm);
  1393. if (*maxrss < hiwater_rss)
  1394. *maxrss = hiwater_rss;
  1395. }
  1396. #if defined(SPLIT_RSS_COUNTING)
  1397. void sync_mm_rss(struct mm_struct *mm);
  1398. #else
  1399. static inline void sync_mm_rss(struct mm_struct *mm)
  1400. {
  1401. }
  1402. #endif
  1403. #ifndef __HAVE_ARCH_PTE_DEVMAP
  1404. static inline int pte_devmap(pte_t pte)
  1405. {
  1406. return 0;
  1407. }
  1408. #endif
  1409. int vma_wants_writenotify(struct vm_area_struct *vma, pgprot_t vm_page_prot);
  1410. extern pte_t *__get_locked_pte(struct mm_struct *mm, unsigned long addr,
  1411. spinlock_t **ptl);
  1412. static inline pte_t *get_locked_pte(struct mm_struct *mm, unsigned long addr,
  1413. spinlock_t **ptl)
  1414. {
  1415. pte_t *ptep;
  1416. __cond_lock(*ptl, ptep = __get_locked_pte(mm, addr, ptl));
  1417. return ptep;
  1418. }
  1419. #ifdef __PAGETABLE_P4D_FOLDED
  1420. static inline int __p4d_alloc(struct mm_struct *mm, pgd_t *pgd,
  1421. unsigned long address)
  1422. {
  1423. return 0;
  1424. }
  1425. #else
  1426. int __p4d_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address);
  1427. #endif
  1428. #if defined(__PAGETABLE_PUD_FOLDED) || !defined(CONFIG_MMU)
  1429. static inline int __pud_alloc(struct mm_struct *mm, p4d_t *p4d,
  1430. unsigned long address)
  1431. {
  1432. return 0;
  1433. }
  1434. static inline void mm_inc_nr_puds(struct mm_struct *mm) {}
  1435. static inline void mm_dec_nr_puds(struct mm_struct *mm) {}
  1436. #else
  1437. int __pud_alloc(struct mm_struct *mm, p4d_t *p4d, unsigned long address);
  1438. static inline void mm_inc_nr_puds(struct mm_struct *mm)
  1439. {
  1440. atomic_long_add(PTRS_PER_PUD * sizeof(pud_t), &mm->pgtables_bytes);
  1441. }
  1442. static inline void mm_dec_nr_puds(struct mm_struct *mm)
  1443. {
  1444. atomic_long_sub(PTRS_PER_PUD * sizeof(pud_t), &mm->pgtables_bytes);
  1445. }
  1446. #endif
  1447. #if defined(__PAGETABLE_PMD_FOLDED) || !defined(CONFIG_MMU)
  1448. static inline int __pmd_alloc(struct mm_struct *mm, pud_t *pud,
  1449. unsigned long address)
  1450. {
  1451. return 0;
  1452. }
  1453. static inline void mm_inc_nr_pmds(struct mm_struct *mm) {}
  1454. static inline void mm_dec_nr_pmds(struct mm_struct *mm) {}
  1455. #else
  1456. int __pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address);
  1457. static inline void mm_inc_nr_pmds(struct mm_struct *mm)
  1458. {
  1459. atomic_long_add(PTRS_PER_PMD * sizeof(pmd_t), &mm->pgtables_bytes);
  1460. }
  1461. static inline void mm_dec_nr_pmds(struct mm_struct *mm)
  1462. {
  1463. atomic_long_sub(PTRS_PER_PMD * sizeof(pmd_t), &mm->pgtables_bytes);
  1464. }
  1465. #endif
  1466. #ifdef CONFIG_MMU
  1467. static inline void mm_pgtables_bytes_init(struct mm_struct *mm)
  1468. {
  1469. atomic_long_set(&mm->pgtables_bytes, 0);
  1470. }
  1471. static inline unsigned long mm_pgtables_bytes(const struct mm_struct *mm)
  1472. {
  1473. return atomic_long_read(&mm->pgtables_bytes);
  1474. }
  1475. static inline void mm_inc_nr_ptes(struct mm_struct *mm)
  1476. {
  1477. atomic_long_add(PTRS_PER_PTE * sizeof(pte_t), &mm->pgtables_bytes);
  1478. }
  1479. static inline void mm_dec_nr_ptes(struct mm_struct *mm)
  1480. {
  1481. atomic_long_sub(PTRS_PER_PTE * sizeof(pte_t), &mm->pgtables_bytes);
  1482. }
  1483. #else
  1484. static inline void mm_pgtables_bytes_init(struct mm_struct *mm) {}
  1485. static inline unsigned long mm_pgtables_bytes(const struct mm_struct *mm)
  1486. {
  1487. return 0;
  1488. }
  1489. static inline void mm_inc_nr_ptes(struct mm_struct *mm) {}
  1490. static inline void mm_dec_nr_ptes(struct mm_struct *mm) {}
  1491. #endif
  1492. int __pte_alloc(struct mm_struct *mm, pmd_t *pmd, unsigned long address);
  1493. int __pte_alloc_kernel(pmd_t *pmd, unsigned long address);
  1494. /*
  1495. * The following ifdef needed to get the 4level-fixup.h header to work.
  1496. * Remove it when 4level-fixup.h has been removed.
  1497. */
  1498. #if defined(CONFIG_MMU) && !defined(__ARCH_HAS_4LEVEL_HACK)
  1499. #ifndef __ARCH_HAS_5LEVEL_HACK
  1500. static inline p4d_t *p4d_alloc(struct mm_struct *mm, pgd_t *pgd,
  1501. unsigned long address)
  1502. {
  1503. return (unlikely(pgd_none(*pgd)) && __p4d_alloc(mm, pgd, address)) ?
  1504. NULL : p4d_offset(pgd, address);
  1505. }
  1506. static inline pud_t *pud_alloc(struct mm_struct *mm, p4d_t *p4d,
  1507. unsigned long address)
  1508. {
  1509. return (unlikely(p4d_none(*p4d)) && __pud_alloc(mm, p4d, address)) ?
  1510. NULL : pud_offset(p4d, address);
  1511. }
  1512. #endif /* !__ARCH_HAS_5LEVEL_HACK */
  1513. static inline pmd_t *pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address)
  1514. {
  1515. return (unlikely(pud_none(*pud)) && __pmd_alloc(mm, pud, address))?
  1516. NULL: pmd_offset(pud, address);
  1517. }
  1518. #endif /* CONFIG_MMU && !__ARCH_HAS_4LEVEL_HACK */
  1519. #if USE_SPLIT_PTE_PTLOCKS
  1520. #if ALLOC_SPLIT_PTLOCKS
  1521. void __init ptlock_cache_init(void);
  1522. extern bool ptlock_alloc(struct page *page);
  1523. extern void ptlock_free(struct page *page);
  1524. static inline spinlock_t *ptlock_ptr(struct page *page)
  1525. {
  1526. return page->ptl;
  1527. }
  1528. #else /* ALLOC_SPLIT_PTLOCKS */
  1529. static inline void ptlock_cache_init(void)
  1530. {
  1531. }
  1532. static inline bool ptlock_alloc(struct page *page)
  1533. {
  1534. return true;
  1535. }
  1536. static inline void ptlock_free(struct page *page)
  1537. {
  1538. }
  1539. static inline spinlock_t *ptlock_ptr(struct page *page)
  1540. {
  1541. return &page->ptl;
  1542. }
  1543. #endif /* ALLOC_SPLIT_PTLOCKS */
  1544. static inline spinlock_t *pte_lockptr(struct mm_struct *mm, pmd_t *pmd)
  1545. {
  1546. return ptlock_ptr(pmd_page(*pmd));
  1547. }
  1548. static inline bool ptlock_init(struct page *page)
  1549. {
  1550. /*
  1551. * prep_new_page() initialize page->private (and therefore page->ptl)
  1552. * with 0. Make sure nobody took it in use in between.
  1553. *
  1554. * It can happen if arch try to use slab for page table allocation:
  1555. * slab code uses page->slab_cache, which share storage with page->ptl.
  1556. */
  1557. VM_BUG_ON_PAGE(*(unsigned long *)&page->ptl, page);
  1558. if (!ptlock_alloc(page))
  1559. return false;
  1560. spin_lock_init(ptlock_ptr(page));
  1561. return true;
  1562. }
  1563. /* Reset page->mapping so free_pages_check won't complain. */
  1564. static inline void pte_lock_deinit(struct page *page)
  1565. {
  1566. page->mapping = NULL;
  1567. ptlock_free(page);
  1568. }
  1569. #else /* !USE_SPLIT_PTE_PTLOCKS */
  1570. /*
  1571. * We use mm->page_table_lock to guard all pagetable pages of the mm.
  1572. */
  1573. static inline spinlock_t *pte_lockptr(struct mm_struct *mm, pmd_t *pmd)
  1574. {
  1575. return &mm->page_table_lock;
  1576. }
  1577. static inline void ptlock_cache_init(void) {}
  1578. static inline bool ptlock_init(struct page *page) { return true; }
  1579. static inline void pte_lock_deinit(struct page *page) {}
  1580. #endif /* USE_SPLIT_PTE_PTLOCKS */
  1581. static inline void pgtable_init(void)
  1582. {
  1583. ptlock_cache_init();
  1584. pgtable_cache_init();
  1585. }
  1586. static inline bool pgtable_page_ctor(struct page *page)
  1587. {
  1588. if (!ptlock_init(page))
  1589. return false;
  1590. inc_zone_page_state(page, NR_PAGETABLE);
  1591. return true;
  1592. }
  1593. static inline void pgtable_page_dtor(struct page *page)
  1594. {
  1595. pte_lock_deinit(page);
  1596. dec_zone_page_state(page, NR_PAGETABLE);
  1597. }
  1598. #define pte_offset_map_lock(mm, pmd, address, ptlp) \
  1599. ({ \
  1600. spinlock_t *__ptl = pte_lockptr(mm, pmd); \
  1601. pte_t *__pte = pte_offset_map(pmd, address); \
  1602. *(ptlp) = __ptl; \
  1603. spin_lock(__ptl); \
  1604. __pte; \
  1605. })
  1606. #define pte_unmap_unlock(pte, ptl) do { \
  1607. spin_unlock(ptl); \
  1608. pte_unmap(pte); \
  1609. } while (0)
  1610. #define pte_alloc(mm, pmd, address) \
  1611. (unlikely(pmd_none(*(pmd))) && __pte_alloc(mm, pmd, address))
  1612. #define pte_alloc_map(mm, pmd, address) \
  1613. (pte_alloc(mm, pmd, address) ? NULL : pte_offset_map(pmd, address))
  1614. #define pte_alloc_map_lock(mm, pmd, address, ptlp) \
  1615. (pte_alloc(mm, pmd, address) ? \
  1616. NULL : pte_offset_map_lock(mm, pmd, address, ptlp))
  1617. #define pte_alloc_kernel(pmd, address) \
  1618. ((unlikely(pmd_none(*(pmd))) && __pte_alloc_kernel(pmd, address))? \
  1619. NULL: pte_offset_kernel(pmd, address))
  1620. #if USE_SPLIT_PMD_PTLOCKS
  1621. static struct page *pmd_to_page(pmd_t *pmd)
  1622. {
  1623. unsigned long mask = ~(PTRS_PER_PMD * sizeof(pmd_t) - 1);
  1624. return virt_to_page((void *)((unsigned long) pmd & mask));
  1625. }
  1626. static inline spinlock_t *pmd_lockptr(struct mm_struct *mm, pmd_t *pmd)
  1627. {
  1628. return ptlock_ptr(pmd_to_page(pmd));
  1629. }
  1630. static inline bool pgtable_pmd_page_ctor(struct page *page)
  1631. {
  1632. #ifdef CONFIG_TRANSPARENT_HUGEPAGE
  1633. page->pmd_huge_pte = NULL;
  1634. #endif
  1635. return ptlock_init(page);
  1636. }
  1637. static inline void pgtable_pmd_page_dtor(struct page *page)
  1638. {
  1639. #ifdef CONFIG_TRANSPARENT_HUGEPAGE
  1640. VM_BUG_ON_PAGE(page->pmd_huge_pte, page);
  1641. #endif
  1642. ptlock_free(page);
  1643. }
  1644. #define pmd_huge_pte(mm, pmd) (pmd_to_page(pmd)->pmd_huge_pte)
  1645. #else
  1646. static inline spinlock_t *pmd_lockptr(struct mm_struct *mm, pmd_t *pmd)
  1647. {
  1648. return &mm->page_table_lock;
  1649. }
  1650. static inline bool pgtable_pmd_page_ctor(struct page *page) { return true; }
  1651. static inline void pgtable_pmd_page_dtor(struct page *page) {}
  1652. #define pmd_huge_pte(mm, pmd) ((mm)->pmd_huge_pte)
  1653. #endif
  1654. static inline spinlock_t *pmd_lock(struct mm_struct *mm, pmd_t *pmd)
  1655. {
  1656. spinlock_t *ptl = pmd_lockptr(mm, pmd);
  1657. spin_lock(ptl);
  1658. return ptl;
  1659. }
  1660. /*
  1661. * No scalability reason to split PUD locks yet, but follow the same pattern
  1662. * as the PMD locks to make it easier if we decide to. The VM should not be
  1663. * considered ready to switch to split PUD locks yet; there may be places
  1664. * which need to be converted from page_table_lock.
  1665. */
  1666. static inline spinlock_t *pud_lockptr(struct mm_struct *mm, pud_t *pud)
  1667. {
  1668. return &mm->page_table_lock;
  1669. }
  1670. static inline spinlock_t *pud_lock(struct mm_struct *mm, pud_t *pud)
  1671. {
  1672. spinlock_t *ptl = pud_lockptr(mm, pud);
  1673. spin_lock(ptl);
  1674. return ptl;
  1675. }
  1676. extern void __init pagecache_init(void);
  1677. extern void free_area_init(unsigned long * zones_size);
  1678. extern void free_area_init_node(int nid, unsigned long * zones_size,
  1679. unsigned long zone_start_pfn, unsigned long *zholes_size);
  1680. extern void free_initmem(void);
  1681. /*
  1682. * Free reserved pages within range [PAGE_ALIGN(start), end & PAGE_MASK)
  1683. * into the buddy system. The freed pages will be poisoned with pattern
  1684. * "poison" if it's within range [0, UCHAR_MAX].
  1685. * Return pages freed into the buddy system.
  1686. */
  1687. extern unsigned long free_reserved_area(void *start, void *end,
  1688. int poison, char *s);
  1689. #ifdef CONFIG_HIGHMEM
  1690. /*
  1691. * Free a highmem page into the buddy system, adjusting totalhigh_pages
  1692. * and totalram_pages.
  1693. */
  1694. extern void free_highmem_page(struct page *page);
  1695. #endif
  1696. extern void adjust_managed_page_count(struct page *page, long count);
  1697. extern void mem_init_print_info(const char *str);
  1698. extern void reserve_bootmem_region(phys_addr_t start, phys_addr_t end);
  1699. /* Free the reserved page into the buddy system, so it gets managed. */
  1700. static inline void __free_reserved_page(struct page *page)
  1701. {
  1702. ClearPageReserved(page);
  1703. init_page_count(page);
  1704. __free_page(page);
  1705. }
  1706. static inline void free_reserved_page(struct page *page)
  1707. {
  1708. __free_reserved_page(page);
  1709. adjust_managed_page_count(page, 1);
  1710. }
  1711. static inline void mark_page_reserved(struct page *page)
  1712. {
  1713. SetPageReserved(page);
  1714. adjust_managed_page_count(page, -1);
  1715. }
  1716. /*
  1717. * Default method to free all the __init memory into the buddy system.
  1718. * The freed pages will be poisoned with pattern "poison" if it's within
  1719. * range [0, UCHAR_MAX].
  1720. * Return pages freed into the buddy system.
  1721. */
  1722. static inline unsigned long free_initmem_default(int poison)
  1723. {
  1724. extern char __init_begin[], __init_end[];
  1725. return free_reserved_area(&__init_begin, &__init_end,
  1726. poison, "unused kernel");
  1727. }
  1728. static inline unsigned long get_num_physpages(void)
  1729. {
  1730. int nid;
  1731. unsigned long phys_pages = 0;
  1732. for_each_online_node(nid)
  1733. phys_pages += node_present_pages(nid);
  1734. return phys_pages;
  1735. }
  1736. #ifdef CONFIG_HAVE_MEMBLOCK_NODE_MAP
  1737. /*
  1738. * With CONFIG_HAVE_MEMBLOCK_NODE_MAP set, an architecture may initialise its
  1739. * zones, allocate the backing mem_map and account for memory holes in a more
  1740. * architecture independent manner. This is a substitute for creating the
  1741. * zone_sizes[] and zholes_size[] arrays and passing them to
  1742. * free_area_init_node()
  1743. *
  1744. * An architecture is expected to register range of page frames backed by
  1745. * physical memory with memblock_add[_node]() before calling
  1746. * free_area_init_nodes() passing in the PFN each zone ends at. At a basic
  1747. * usage, an architecture is expected to do something like
  1748. *
  1749. * unsigned long max_zone_pfns[MAX_NR_ZONES] = {max_dma, max_normal_pfn,
  1750. * max_highmem_pfn};
  1751. * for_each_valid_physical_page_range()
  1752. * memblock_add_node(base, size, nid)
  1753. * free_area_init_nodes(max_zone_pfns);
  1754. *
  1755. * free_bootmem_with_active_regions() calls free_bootmem_node() for each
  1756. * registered physical page range. Similarly
  1757. * sparse_memory_present_with_active_regions() calls memory_present() for
  1758. * each range when SPARSEMEM is enabled.
  1759. *
  1760. * See mm/page_alloc.c for more information on each function exposed by
  1761. * CONFIG_HAVE_MEMBLOCK_NODE_MAP.
  1762. */
  1763. extern void free_area_init_nodes(unsigned long *max_zone_pfn);
  1764. unsigned long node_map_pfn_alignment(void);
  1765. unsigned long __absent_pages_in_range(int nid, unsigned long start_pfn,
  1766. unsigned long end_pfn);
  1767. extern unsigned long absent_pages_in_range(unsigned long start_pfn,
  1768. unsigned long end_pfn);
  1769. extern void get_pfn_range_for_nid(unsigned int nid,
  1770. unsigned long *start_pfn, unsigned long *end_pfn);
  1771. extern unsigned long find_min_pfn_with_active_regions(void);
  1772. extern void free_bootmem_with_active_regions(int nid,
  1773. unsigned long max_low_pfn);
  1774. extern void sparse_memory_present_with_active_regions(int nid);
  1775. #endif /* CONFIG_HAVE_MEMBLOCK_NODE_MAP */
  1776. #if !defined(CONFIG_HAVE_MEMBLOCK_NODE_MAP) && \
  1777. !defined(CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID)
  1778. static inline int __early_pfn_to_nid(unsigned long pfn,
  1779. struct mminit_pfnnid_cache *state)
  1780. {
  1781. return 0;
  1782. }
  1783. #else
  1784. /* please see mm/page_alloc.c */
  1785. extern int __meminit early_pfn_to_nid(unsigned long pfn);
  1786. /* there is a per-arch backend function. */
  1787. extern int __meminit __early_pfn_to_nid(unsigned long pfn,
  1788. struct mminit_pfnnid_cache *state);
  1789. #endif
  1790. #ifdef CONFIG_HAVE_MEMBLOCK
  1791. void zero_resv_unavail(void);
  1792. #else
  1793. static inline void zero_resv_unavail(void) {}
  1794. #endif
  1795. extern void set_dma_reserve(unsigned long new_dma_reserve);
  1796. extern void memmap_init_zone(unsigned long, int, unsigned long, unsigned long,
  1797. enum memmap_context, struct vmem_altmap *);
  1798. extern void setup_per_zone_wmarks(void);
  1799. extern int __meminit init_per_zone_wmark_min(void);
  1800. extern void mem_init(void);
  1801. extern void __init mmap_init(void);
  1802. extern void show_mem(unsigned int flags, nodemask_t *nodemask);
  1803. extern long si_mem_available(void);
  1804. extern void si_meminfo(struct sysinfo * val);
  1805. extern void si_meminfo_node(struct sysinfo *val, int nid);
  1806. #ifdef __HAVE_ARCH_RESERVED_KERNEL_PAGES
  1807. extern unsigned long arch_reserved_kernel_pages(void);
  1808. #endif
  1809. extern __printf(3, 4)
  1810. void warn_alloc(gfp_t gfp_mask, nodemask_t *nodemask, const char *fmt, ...);
  1811. extern void setup_per_cpu_pageset(void);
  1812. extern void zone_pcp_update(struct zone *zone);
  1813. extern void zone_pcp_reset(struct zone *zone);
  1814. /* page_alloc.c */
  1815. extern int min_free_kbytes;
  1816. extern int watermark_scale_factor;
  1817. /* nommu.c */
  1818. extern atomic_long_t mmap_pages_allocated;
  1819. extern int nommu_shrink_inode_mappings(struct inode *, size_t, size_t);
  1820. /* interval_tree.c */
  1821. void vma_interval_tree_insert(struct vm_area_struct *node,
  1822. struct rb_root_cached *root);
  1823. void vma_interval_tree_insert_after(struct vm_area_struct *node,
  1824. struct vm_area_struct *prev,
  1825. struct rb_root_cached *root);
  1826. void vma_interval_tree_remove(struct vm_area_struct *node,
  1827. struct rb_root_cached *root);
  1828. struct vm_area_struct *vma_interval_tree_iter_first(struct rb_root_cached *root,
  1829. unsigned long start, unsigned long last);
  1830. struct vm_area_struct *vma_interval_tree_iter_next(struct vm_area_struct *node,
  1831. unsigned long start, unsigned long last);
  1832. #define vma_interval_tree_foreach(vma, root, start, last) \
  1833. for (vma = vma_interval_tree_iter_first(root, start, last); \
  1834. vma; vma = vma_interval_tree_iter_next(vma, start, last))
  1835. void anon_vma_interval_tree_insert(struct anon_vma_chain *node,
  1836. struct rb_root_cached *root);
  1837. void anon_vma_interval_tree_remove(struct anon_vma_chain *node,
  1838. struct rb_root_cached *root);
  1839. struct anon_vma_chain *
  1840. anon_vma_interval_tree_iter_first(struct rb_root_cached *root,
  1841. unsigned long start, unsigned long last);
  1842. struct anon_vma_chain *anon_vma_interval_tree_iter_next(
  1843. struct anon_vma_chain *node, unsigned long start, unsigned long last);
  1844. #ifdef CONFIG_DEBUG_VM_RB
  1845. void anon_vma_interval_tree_verify(struct anon_vma_chain *node);
  1846. #endif
  1847. #define anon_vma_interval_tree_foreach(avc, root, start, last) \
  1848. for (avc = anon_vma_interval_tree_iter_first(root, start, last); \
  1849. avc; avc = anon_vma_interval_tree_iter_next(avc, start, last))
  1850. /* mmap.c */
  1851. extern int __vm_enough_memory(struct mm_struct *mm, long pages, int cap_sys_admin);
  1852. extern int __vma_adjust(struct vm_area_struct *vma, unsigned long start,
  1853. unsigned long end, pgoff_t pgoff, struct vm_area_struct *insert,
  1854. struct vm_area_struct *expand);
  1855. static inline int vma_adjust(struct vm_area_struct *vma, unsigned long start,
  1856. unsigned long end, pgoff_t pgoff, struct vm_area_struct *insert)
  1857. {
  1858. return __vma_adjust(vma, start, end, pgoff, insert, NULL);
  1859. }
  1860. extern struct vm_area_struct *vma_merge(struct mm_struct *,
  1861. struct vm_area_struct *prev, unsigned long addr, unsigned long end,
  1862. unsigned long vm_flags, struct anon_vma *, struct file *, pgoff_t,
  1863. struct mempolicy *, struct vm_userfaultfd_ctx);
  1864. extern struct anon_vma *find_mergeable_anon_vma(struct vm_area_struct *);
  1865. extern int __split_vma(struct mm_struct *, struct vm_area_struct *,
  1866. unsigned long addr, int new_below);
  1867. extern int split_vma(struct mm_struct *, struct vm_area_struct *,
  1868. unsigned long addr, int new_below);
  1869. extern int insert_vm_struct(struct mm_struct *, struct vm_area_struct *);
  1870. extern void __vma_link_rb(struct mm_struct *, struct vm_area_struct *,
  1871. struct rb_node **, struct rb_node *);
  1872. extern void unlink_file_vma(struct vm_area_struct *);
  1873. extern struct vm_area_struct *copy_vma(struct vm_area_struct **,
  1874. unsigned long addr, unsigned long len, pgoff_t pgoff,
  1875. bool *need_rmap_locks);
  1876. extern void exit_mmap(struct mm_struct *);
  1877. static inline int check_data_rlimit(unsigned long rlim,
  1878. unsigned long new,
  1879. unsigned long start,
  1880. unsigned long end_data,
  1881. unsigned long start_data)
  1882. {
  1883. if (rlim < RLIM_INFINITY) {
  1884. if (((new - start) + (end_data - start_data)) > rlim)
  1885. return -ENOSPC;
  1886. }
  1887. return 0;
  1888. }
  1889. extern int mm_take_all_locks(struct mm_struct *mm);
  1890. extern void mm_drop_all_locks(struct mm_struct *mm);
  1891. extern void set_mm_exe_file(struct mm_struct *mm, struct file *new_exe_file);
  1892. extern struct file *get_mm_exe_file(struct mm_struct *mm);
  1893. extern struct file *get_task_exe_file(struct task_struct *task);
  1894. extern bool may_expand_vm(struct mm_struct *, vm_flags_t, unsigned long npages);
  1895. extern void vm_stat_account(struct mm_struct *, vm_flags_t, long npages);
  1896. extern bool vma_is_special_mapping(const struct vm_area_struct *vma,
  1897. const struct vm_special_mapping *sm);
  1898. extern struct vm_area_struct *_install_special_mapping(struct mm_struct *mm,
  1899. unsigned long addr, unsigned long len,
  1900. unsigned long flags,
  1901. const struct vm_special_mapping *spec);
  1902. /* This is an obsolete alternative to _install_special_mapping. */
  1903. extern int install_special_mapping(struct mm_struct *mm,
  1904. unsigned long addr, unsigned long len,
  1905. unsigned long flags, struct page **pages);
  1906. extern unsigned long get_unmapped_area(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
  1907. extern unsigned long mmap_region(struct file *file, unsigned long addr,
  1908. unsigned long len, vm_flags_t vm_flags, unsigned long pgoff,
  1909. struct list_head *uf);
  1910. extern unsigned long do_mmap(struct file *file, unsigned long addr,
  1911. unsigned long len, unsigned long prot, unsigned long flags,
  1912. vm_flags_t vm_flags, unsigned long pgoff, unsigned long *populate,
  1913. struct list_head *uf);
  1914. extern int do_munmap(struct mm_struct *, unsigned long, size_t,
  1915. struct list_head *uf);
  1916. static inline unsigned long
  1917. do_mmap_pgoff(struct file *file, unsigned long addr,
  1918. unsigned long len, unsigned long prot, unsigned long flags,
  1919. unsigned long pgoff, unsigned long *populate,
  1920. struct list_head *uf)
  1921. {
  1922. return do_mmap(file, addr, len, prot, flags, 0, pgoff, populate, uf);
  1923. }
  1924. #ifdef CONFIG_MMU
  1925. extern int __mm_populate(unsigned long addr, unsigned long len,
  1926. int ignore_errors);
  1927. static inline void mm_populate(unsigned long addr, unsigned long len)
  1928. {
  1929. /* Ignore errors */
  1930. (void) __mm_populate(addr, len, 1);
  1931. }
  1932. #else
  1933. static inline void mm_populate(unsigned long addr, unsigned long len) {}
  1934. #endif
  1935. /* These take the mm semaphore themselves */
  1936. extern int __must_check vm_brk(unsigned long, unsigned long);
  1937. extern int __must_check vm_brk_flags(unsigned long, unsigned long, unsigned long);
  1938. extern int vm_munmap(unsigned long, size_t);
  1939. extern unsigned long __must_check vm_mmap(struct file *, unsigned long,
  1940. unsigned long, unsigned long,
  1941. unsigned long, unsigned long);
  1942. struct vm_unmapped_area_info {
  1943. #define VM_UNMAPPED_AREA_TOPDOWN 1
  1944. unsigned long flags;
  1945. unsigned long length;
  1946. unsigned long low_limit;
  1947. unsigned long high_limit;
  1948. unsigned long align_mask;
  1949. unsigned long align_offset;
  1950. };
  1951. extern unsigned long unmapped_area(struct vm_unmapped_area_info *info);
  1952. extern unsigned long unmapped_area_topdown(struct vm_unmapped_area_info *info);
  1953. /*
  1954. * Search for an unmapped address range.
  1955. *
  1956. * We are looking for a range that:
  1957. * - does not intersect with any VMA;
  1958. * - is contained within the [low_limit, high_limit) interval;
  1959. * - is at least the desired size.
  1960. * - satisfies (begin_addr & align_mask) == (align_offset & align_mask)
  1961. */
  1962. static inline unsigned long
  1963. vm_unmapped_area(struct vm_unmapped_area_info *info)
  1964. {
  1965. if (info->flags & VM_UNMAPPED_AREA_TOPDOWN)
  1966. return unmapped_area_topdown(info);
  1967. else
  1968. return unmapped_area(info);
  1969. }
  1970. /* truncate.c */
  1971. extern void truncate_inode_pages(struct address_space *, loff_t);
  1972. extern void truncate_inode_pages_range(struct address_space *,
  1973. loff_t lstart, loff_t lend);
  1974. extern void truncate_inode_pages_final(struct address_space *);
  1975. /* generic vm_area_ops exported for stackable file systems */
  1976. extern int filemap_fault(struct vm_fault *vmf);
  1977. extern void filemap_map_pages(struct vm_fault *vmf,
  1978. pgoff_t start_pgoff, pgoff_t end_pgoff);
  1979. extern int filemap_page_mkwrite(struct vm_fault *vmf);
  1980. /* mm/page-writeback.c */
  1981. int __must_check write_one_page(struct page *page);
  1982. void task_dirty_inc(struct task_struct *tsk);
  1983. /* readahead.c */
  1984. #define VM_MAX_READAHEAD 128 /* kbytes */
  1985. #define VM_MIN_READAHEAD 16 /* kbytes (includes current page) */
  1986. int force_page_cache_readahead(struct address_space *mapping, struct file *filp,
  1987. pgoff_t offset, unsigned long nr_to_read);
  1988. void page_cache_sync_readahead(struct address_space *mapping,
  1989. struct file_ra_state *ra,
  1990. struct file *filp,
  1991. pgoff_t offset,
  1992. unsigned long size);
  1993. void page_cache_async_readahead(struct address_space *mapping,
  1994. struct file_ra_state *ra,
  1995. struct file *filp,
  1996. struct page *pg,
  1997. pgoff_t offset,
  1998. unsigned long size);
  1999. extern unsigned long stack_guard_gap;
  2000. /* Generic expand stack which grows the stack according to GROWS{UP,DOWN} */
  2001. extern int expand_stack(struct vm_area_struct *vma, unsigned long address);
  2002. /* CONFIG_STACK_GROWSUP still needs to to grow downwards at some places */
  2003. extern int expand_downwards(struct vm_area_struct *vma,
  2004. unsigned long address);
  2005. #if VM_GROWSUP
  2006. extern int expand_upwards(struct vm_area_struct *vma, unsigned long address);
  2007. #else
  2008. #define expand_upwards(vma, address) (0)
  2009. #endif
  2010. /* Look up the first VMA which satisfies addr < vm_end, NULL if none. */
  2011. extern struct vm_area_struct * find_vma(struct mm_struct * mm, unsigned long addr);
  2012. extern struct vm_area_struct * find_vma_prev(struct mm_struct * mm, unsigned long addr,
  2013. struct vm_area_struct **pprev);
  2014. /* Look up the first VMA which intersects the interval start_addr..end_addr-1,
  2015. NULL if none. Assume start_addr < end_addr. */
  2016. static inline struct vm_area_struct * find_vma_intersection(struct mm_struct * mm, unsigned long start_addr, unsigned long end_addr)
  2017. {
  2018. struct vm_area_struct * vma = find_vma(mm,start_addr);
  2019. if (vma && end_addr <= vma->vm_start)
  2020. vma = NULL;
  2021. return vma;
  2022. }
  2023. static inline unsigned long vm_start_gap(struct vm_area_struct *vma)
  2024. {
  2025. unsigned long vm_start = vma->vm_start;
  2026. if (vma->vm_flags & VM_GROWSDOWN) {
  2027. vm_start -= stack_guard_gap;
  2028. if (vm_start > vma->vm_start)
  2029. vm_start = 0;
  2030. }
  2031. return vm_start;
  2032. }
  2033. static inline unsigned long vm_end_gap(struct vm_area_struct *vma)
  2034. {
  2035. unsigned long vm_end = vma->vm_end;
  2036. if (vma->vm_flags & VM_GROWSUP) {
  2037. vm_end += stack_guard_gap;
  2038. if (vm_end < vma->vm_end)
  2039. vm_end = -PAGE_SIZE;
  2040. }
  2041. return vm_end;
  2042. }
  2043. static inline unsigned long vma_pages(struct vm_area_struct *vma)
  2044. {
  2045. return (vma->vm_end - vma->vm_start) >> PAGE_SHIFT;
  2046. }
  2047. /* Look up the first VMA which exactly match the interval vm_start ... vm_end */
  2048. static inline struct vm_area_struct *find_exact_vma(struct mm_struct *mm,
  2049. unsigned long vm_start, unsigned long vm_end)
  2050. {
  2051. struct vm_area_struct *vma = find_vma(mm, vm_start);
  2052. if (vma && (vma->vm_start != vm_start || vma->vm_end != vm_end))
  2053. vma = NULL;
  2054. return vma;
  2055. }
  2056. #ifdef CONFIG_MMU
  2057. pgprot_t vm_get_page_prot(unsigned long vm_flags);
  2058. void vma_set_page_prot(struct vm_area_struct *vma);
  2059. #else
  2060. static inline pgprot_t vm_get_page_prot(unsigned long vm_flags)
  2061. {
  2062. return __pgprot(0);
  2063. }
  2064. static inline void vma_set_page_prot(struct vm_area_struct *vma)
  2065. {
  2066. vma->vm_page_prot = vm_get_page_prot(vma->vm_flags);
  2067. }
  2068. #endif
  2069. #ifdef CONFIG_NUMA_BALANCING
  2070. unsigned long change_prot_numa(struct vm_area_struct *vma,
  2071. unsigned long start, unsigned long end);
  2072. #endif
  2073. struct vm_area_struct *find_extend_vma(struct mm_struct *, unsigned long addr);
  2074. int remap_pfn_range(struct vm_area_struct *, unsigned long addr,
  2075. unsigned long pfn, unsigned long size, pgprot_t);
  2076. int vm_insert_page(struct vm_area_struct *, unsigned long addr, struct page *);
  2077. int vm_insert_pfn(struct vm_area_struct *vma, unsigned long addr,
  2078. unsigned long pfn);
  2079. int vm_insert_pfn_prot(struct vm_area_struct *vma, unsigned long addr,
  2080. unsigned long pfn, pgprot_t pgprot);
  2081. int vm_insert_mixed(struct vm_area_struct *vma, unsigned long addr,
  2082. pfn_t pfn);
  2083. int vm_insert_mixed_mkwrite(struct vm_area_struct *vma, unsigned long addr,
  2084. pfn_t pfn);
  2085. int vm_iomap_memory(struct vm_area_struct *vma, phys_addr_t start, unsigned long len);
  2086. struct page *follow_page_mask(struct vm_area_struct *vma,
  2087. unsigned long address, unsigned int foll_flags,
  2088. unsigned int *page_mask);
  2089. static inline struct page *follow_page(struct vm_area_struct *vma,
  2090. unsigned long address, unsigned int foll_flags)
  2091. {
  2092. unsigned int unused_page_mask;
  2093. return follow_page_mask(vma, address, foll_flags, &unused_page_mask);
  2094. }
  2095. #define FOLL_WRITE 0x01 /* check pte is writable */
  2096. #define FOLL_TOUCH 0x02 /* mark page accessed */
  2097. #define FOLL_GET 0x04 /* do get_page on page */
  2098. #define FOLL_DUMP 0x08 /* give error on hole if it would be zero */
  2099. #define FOLL_FORCE 0x10 /* get_user_pages read/write w/o permission */
  2100. #define FOLL_NOWAIT 0x20 /* if a disk transfer is needed, start the IO
  2101. * and return without waiting upon it */
  2102. #define FOLL_POPULATE 0x40 /* fault in page */
  2103. #define FOLL_SPLIT 0x80 /* don't return transhuge pages, split them */
  2104. #define FOLL_HWPOISON 0x100 /* check page is hwpoisoned */
  2105. #define FOLL_NUMA 0x200 /* force NUMA hinting page fault */
  2106. #define FOLL_MIGRATION 0x400 /* wait for page to replace migration entry */
  2107. #define FOLL_TRIED 0x800 /* a retry, previous pass started an IO */
  2108. #define FOLL_MLOCK 0x1000 /* lock present pages */
  2109. #define FOLL_REMOTE 0x2000 /* we are working on non-current tsk/mm */
  2110. #define FOLL_COW 0x4000 /* internal GUP flag */
  2111. static inline int vm_fault_to_errno(int vm_fault, int foll_flags)
  2112. {
  2113. if (vm_fault & VM_FAULT_OOM)
  2114. return -ENOMEM;
  2115. if (vm_fault & (VM_FAULT_HWPOISON | VM_FAULT_HWPOISON_LARGE))
  2116. return (foll_flags & FOLL_HWPOISON) ? -EHWPOISON : -EFAULT;
  2117. if (vm_fault & (VM_FAULT_SIGBUS | VM_FAULT_SIGSEGV))
  2118. return -EFAULT;
  2119. return 0;
  2120. }
  2121. typedef int (*pte_fn_t)(pte_t *pte, pgtable_t token, unsigned long addr,
  2122. void *data);
  2123. extern int apply_to_page_range(struct mm_struct *mm, unsigned long address,
  2124. unsigned long size, pte_fn_t fn, void *data);
  2125. #ifdef CONFIG_PAGE_POISONING
  2126. extern bool page_poisoning_enabled(void);
  2127. extern void kernel_poison_pages(struct page *page, int numpages, int enable);
  2128. extern bool page_is_poisoned(struct page *page);
  2129. #else
  2130. static inline bool page_poisoning_enabled(void) { return false; }
  2131. static inline void kernel_poison_pages(struct page *page, int numpages,
  2132. int enable) { }
  2133. static inline bool page_is_poisoned(struct page *page) { return false; }
  2134. #endif
  2135. #ifdef CONFIG_DEBUG_PAGEALLOC
  2136. extern bool _debug_pagealloc_enabled;
  2137. extern void __kernel_map_pages(struct page *page, int numpages, int enable);
  2138. static inline bool debug_pagealloc_enabled(void)
  2139. {
  2140. return _debug_pagealloc_enabled;
  2141. }
  2142. static inline void
  2143. kernel_map_pages(struct page *page, int numpages, int enable)
  2144. {
  2145. if (!debug_pagealloc_enabled())
  2146. return;
  2147. __kernel_map_pages(page, numpages, enable);
  2148. }
  2149. #ifdef CONFIG_HIBERNATION
  2150. extern bool kernel_page_present(struct page *page);
  2151. #endif /* CONFIG_HIBERNATION */
  2152. #else /* CONFIG_DEBUG_PAGEALLOC */
  2153. static inline void
  2154. kernel_map_pages(struct page *page, int numpages, int enable) {}
  2155. #ifdef CONFIG_HIBERNATION
  2156. static inline bool kernel_page_present(struct page *page) { return true; }
  2157. #endif /* CONFIG_HIBERNATION */
  2158. static inline bool debug_pagealloc_enabled(void)
  2159. {
  2160. return false;
  2161. }
  2162. #endif /* CONFIG_DEBUG_PAGEALLOC */
  2163. #ifdef __HAVE_ARCH_GATE_AREA
  2164. extern struct vm_area_struct *get_gate_vma(struct mm_struct *mm);
  2165. extern int in_gate_area_no_mm(unsigned long addr);
  2166. extern int in_gate_area(struct mm_struct *mm, unsigned long addr);
  2167. #else
  2168. static inline struct vm_area_struct *get_gate_vma(struct mm_struct *mm)
  2169. {
  2170. return NULL;
  2171. }
  2172. static inline int in_gate_area_no_mm(unsigned long addr) { return 0; }
  2173. static inline int in_gate_area(struct mm_struct *mm, unsigned long addr)
  2174. {
  2175. return 0;
  2176. }
  2177. #endif /* __HAVE_ARCH_GATE_AREA */
  2178. extern bool process_shares_mm(struct task_struct *p, struct mm_struct *mm);
  2179. #ifdef CONFIG_SYSCTL
  2180. extern int sysctl_drop_caches;
  2181. int drop_caches_sysctl_handler(struct ctl_table *, int,
  2182. void __user *, size_t *, loff_t *);
  2183. #endif
  2184. void drop_slab(void);
  2185. void drop_slab_node(int nid);
  2186. #ifndef CONFIG_MMU
  2187. #define randomize_va_space 0
  2188. #else
  2189. extern int randomize_va_space;
  2190. #endif
  2191. const char * arch_vma_name(struct vm_area_struct *vma);
  2192. void print_vma_addr(char *prefix, unsigned long rip);
  2193. void sparse_mem_maps_populate_node(struct page **map_map,
  2194. unsigned long pnum_begin,
  2195. unsigned long pnum_end,
  2196. unsigned long map_count,
  2197. int nodeid);
  2198. struct page *sparse_mem_map_populate(unsigned long pnum, int nid,
  2199. struct vmem_altmap *altmap);
  2200. pgd_t *vmemmap_pgd_populate(unsigned long addr, int node);
  2201. p4d_t *vmemmap_p4d_populate(pgd_t *pgd, unsigned long addr, int node);
  2202. pud_t *vmemmap_pud_populate(p4d_t *p4d, unsigned long addr, int node);
  2203. pmd_t *vmemmap_pmd_populate(pud_t *pud, unsigned long addr, int node);
  2204. pte_t *vmemmap_pte_populate(pmd_t *pmd, unsigned long addr, int node);
  2205. void *vmemmap_alloc_block(unsigned long size, int node);
  2206. struct vmem_altmap;
  2207. void *vmemmap_alloc_block_buf(unsigned long size, int node);
  2208. void *altmap_alloc_block_buf(unsigned long size, struct vmem_altmap *altmap);
  2209. void vmemmap_verify(pte_t *, int, unsigned long, unsigned long);
  2210. int vmemmap_populate_basepages(unsigned long start, unsigned long end,
  2211. int node);
  2212. int vmemmap_populate(unsigned long start, unsigned long end, int node,
  2213. struct vmem_altmap *altmap);
  2214. void vmemmap_populate_print_last(void);
  2215. #ifdef CONFIG_MEMORY_HOTPLUG
  2216. void vmemmap_free(unsigned long start, unsigned long end,
  2217. struct vmem_altmap *altmap);
  2218. #endif
  2219. void register_page_bootmem_memmap(unsigned long section_nr, struct page *map,
  2220. unsigned long nr_pages);
  2221. enum mf_flags {
  2222. MF_COUNT_INCREASED = 1 << 0,
  2223. MF_ACTION_REQUIRED = 1 << 1,
  2224. MF_MUST_KILL = 1 << 2,
  2225. MF_SOFT_OFFLINE = 1 << 3,
  2226. };
  2227. extern int memory_failure(unsigned long pfn, int flags);
  2228. extern void memory_failure_queue(unsigned long pfn, int flags);
  2229. extern int unpoison_memory(unsigned long pfn);
  2230. extern int get_hwpoison_page(struct page *page);
  2231. #define put_hwpoison_page(page) put_page(page)
  2232. extern int sysctl_memory_failure_early_kill;
  2233. extern int sysctl_memory_failure_recovery;
  2234. extern void shake_page(struct page *p, int access);
  2235. extern atomic_long_t num_poisoned_pages __read_mostly;
  2236. extern int soft_offline_page(struct page *page, int flags);
  2237. /*
  2238. * Error handlers for various types of pages.
  2239. */
  2240. enum mf_result {
  2241. MF_IGNORED, /* Error: cannot be handled */
  2242. MF_FAILED, /* Error: handling failed */
  2243. MF_DELAYED, /* Will be handled later */
  2244. MF_RECOVERED, /* Successfully recovered */
  2245. };
  2246. enum mf_action_page_type {
  2247. MF_MSG_KERNEL,
  2248. MF_MSG_KERNEL_HIGH_ORDER,
  2249. MF_MSG_SLAB,
  2250. MF_MSG_DIFFERENT_COMPOUND,
  2251. MF_MSG_POISONED_HUGE,
  2252. MF_MSG_HUGE,
  2253. MF_MSG_FREE_HUGE,
  2254. MF_MSG_NON_PMD_HUGE,
  2255. MF_MSG_UNMAP_FAILED,
  2256. MF_MSG_DIRTY_SWAPCACHE,
  2257. MF_MSG_CLEAN_SWAPCACHE,
  2258. MF_MSG_DIRTY_MLOCKED_LRU,
  2259. MF_MSG_CLEAN_MLOCKED_LRU,
  2260. MF_MSG_DIRTY_UNEVICTABLE_LRU,
  2261. MF_MSG_CLEAN_UNEVICTABLE_LRU,
  2262. MF_MSG_DIRTY_LRU,
  2263. MF_MSG_CLEAN_LRU,
  2264. MF_MSG_TRUNCATED_LRU,
  2265. MF_MSG_BUDDY,
  2266. MF_MSG_BUDDY_2ND,
  2267. MF_MSG_UNKNOWN,
  2268. };
  2269. #if defined(CONFIG_TRANSPARENT_HUGEPAGE) || defined(CONFIG_HUGETLBFS)
  2270. extern void clear_huge_page(struct page *page,
  2271. unsigned long addr_hint,
  2272. unsigned int pages_per_huge_page);
  2273. extern void copy_user_huge_page(struct page *dst, struct page *src,
  2274. unsigned long addr, struct vm_area_struct *vma,
  2275. unsigned int pages_per_huge_page);
  2276. extern long copy_huge_page_from_user(struct page *dst_page,
  2277. const void __user *usr_src,
  2278. unsigned int pages_per_huge_page,
  2279. bool allow_pagefault);
  2280. #endif /* CONFIG_TRANSPARENT_HUGEPAGE || CONFIG_HUGETLBFS */
  2281. extern struct page_ext_operations debug_guardpage_ops;
  2282. #ifdef CONFIG_DEBUG_PAGEALLOC
  2283. extern unsigned int _debug_guardpage_minorder;
  2284. extern bool _debug_guardpage_enabled;
  2285. static inline unsigned int debug_guardpage_minorder(void)
  2286. {
  2287. return _debug_guardpage_minorder;
  2288. }
  2289. static inline bool debug_guardpage_enabled(void)
  2290. {
  2291. return _debug_guardpage_enabled;
  2292. }
  2293. static inline bool page_is_guard(struct page *page)
  2294. {
  2295. struct page_ext *page_ext;
  2296. if (!debug_guardpage_enabled())
  2297. return false;
  2298. page_ext = lookup_page_ext(page);
  2299. if (unlikely(!page_ext))
  2300. return false;
  2301. return test_bit(PAGE_EXT_DEBUG_GUARD, &page_ext->flags);
  2302. }
  2303. #else
  2304. static inline unsigned int debug_guardpage_minorder(void) { return 0; }
  2305. static inline bool debug_guardpage_enabled(void) { return false; }
  2306. static inline bool page_is_guard(struct page *page) { return false; }
  2307. #endif /* CONFIG_DEBUG_PAGEALLOC */
  2308. #if MAX_NUMNODES > 1
  2309. void __init setup_nr_node_ids(void);
  2310. #else
  2311. static inline void setup_nr_node_ids(void) {}
  2312. #endif
  2313. #endif /* __KERNEL__ */
  2314. #endif /* _LINUX_MM_H */