internal.h 9.8 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344
  1. /* internal.h: mm/ internal definitions
  2. *
  3. * Copyright (C) 2004 Red Hat, Inc. All Rights Reserved.
  4. * Written by David Howells (dhowells@redhat.com)
  5. *
  6. * This program is free software; you can redistribute it and/or
  7. * modify it under the terms of the GNU General Public License
  8. * as published by the Free Software Foundation; either version
  9. * 2 of the License, or (at your option) any later version.
  10. */
  11. #ifndef __MM_INTERNAL_H
  12. #define __MM_INTERNAL_H
  13. #include <linux/mm.h>
  14. void free_pgtables(struct mmu_gather *tlb, struct vm_area_struct *start_vma,
  15. unsigned long floor, unsigned long ceiling);
  16. static inline void set_page_count(struct page *page, int v)
  17. {
  18. atomic_set(&page->_count, v);
  19. }
  20. /*
  21. * Turn a non-refcounted page (->_count == 0) into refcounted with
  22. * a count of one.
  23. */
  24. static inline void set_page_refcounted(struct page *page)
  25. {
  26. VM_BUG_ON(PageTail(page));
  27. VM_BUG_ON(atomic_read(&page->_count));
  28. set_page_count(page, 1);
  29. }
  30. static inline void __put_page(struct page *page)
  31. {
  32. atomic_dec(&page->_count);
  33. }
  34. static inline void __get_page_tail_foll(struct page *page,
  35. bool get_page_head)
  36. {
  37. /*
  38. * If we're getting a tail page, the elevated page->_count is
  39. * required only in the head page and we will elevate the head
  40. * page->_count and tail page->_mapcount.
  41. *
  42. * We elevate page_tail->_mapcount for tail pages to force
  43. * page_tail->_count to be zero at all times to avoid getting
  44. * false positives from get_page_unless_zero() with
  45. * speculative page access (like in
  46. * page_cache_get_speculative()) on tail pages.
  47. */
  48. VM_BUG_ON(atomic_read(&page->first_page->_count) <= 0);
  49. VM_BUG_ON(atomic_read(&page->_count) != 0);
  50. VM_BUG_ON(page_mapcount(page) < 0);
  51. if (get_page_head)
  52. atomic_inc(&page->first_page->_count);
  53. atomic_inc(&page->_mapcount);
  54. }
  55. /*
  56. * This is meant to be called as the FOLL_GET operation of
  57. * follow_page() and it must be called while holding the proper PT
  58. * lock while the pte (or pmd_trans_huge) is still mapping the page.
  59. */
  60. static inline void get_page_foll(struct page *page)
  61. {
  62. if (unlikely(PageTail(page)))
  63. /*
  64. * This is safe only because
  65. * __split_huge_page_refcount() can't run under
  66. * get_page_foll() because we hold the proper PT lock.
  67. */
  68. __get_page_tail_foll(page, true);
  69. else {
  70. /*
  71. * Getting a normal page or the head of a compound page
  72. * requires to already have an elevated page->_count.
  73. */
  74. VM_BUG_ON(atomic_read(&page->_count) <= 0);
  75. atomic_inc(&page->_count);
  76. }
  77. }
  78. extern unsigned long highest_memmap_pfn;
  79. /*
  80. * in mm/vmscan.c:
  81. */
  82. extern int isolate_lru_page(struct page *page);
  83. extern void putback_lru_page(struct page *page);
  84. /*
  85. * in mm/page_alloc.c
  86. */
  87. extern void __free_pages_bootmem(struct page *page, unsigned int order);
  88. extern void prep_compound_page(struct page *page, unsigned long order);
  89. #ifdef CONFIG_MEMORY_FAILURE
  90. extern bool is_free_buddy_page(struct page *page);
  91. #endif
  92. #if defined CONFIG_COMPACTION || defined CONFIG_CMA
  93. /*
  94. * in mm/compaction.c
  95. */
  96. /*
  97. * compact_control is used to track pages being migrated and the free pages
  98. * they are being migrated to during memory compaction. The free_pfn starts
  99. * at the end of a zone and migrate_pfn begins at the start. Movable pages
  100. * are moved to the end of a zone during a compaction run and the run
  101. * completes when free_pfn <= migrate_pfn
  102. */
  103. struct compact_control {
  104. struct list_head freepages; /* List of free pages to migrate to */
  105. struct list_head migratepages; /* List of pages being migrated */
  106. unsigned long nr_freepages; /* Number of isolated free pages */
  107. unsigned long nr_migratepages; /* Number of pages to migrate */
  108. unsigned long free_pfn; /* isolate_freepages search base */
  109. unsigned long migrate_pfn; /* isolate_migratepages search base */
  110. bool sync; /* Synchronous migration */
  111. int order; /* order a direct compactor needs */
  112. int migratetype; /* MOVABLE, RECLAIMABLE etc */
  113. struct zone *zone;
  114. };
  115. unsigned long
  116. isolate_freepages_range(unsigned long start_pfn, unsigned long end_pfn);
  117. unsigned long
  118. isolate_migratepages_range(struct zone *zone, struct compact_control *cc,
  119. unsigned long low_pfn, unsigned long end_pfn);
  120. #endif
  121. /*
  122. * function for dealing with page's order in buddy system.
  123. * zone->lock is already acquired when we use these.
  124. * So, we don't need atomic page->flags operations here.
  125. */
  126. static inline unsigned long page_order(struct page *page)
  127. {
  128. /* PageBuddy() must be checked by the caller */
  129. return page_private(page);
  130. }
  131. /* mm/util.c */
  132. void __vma_link_list(struct mm_struct *mm, struct vm_area_struct *vma,
  133. struct vm_area_struct *prev, struct rb_node *rb_parent);
  134. #ifdef CONFIG_MMU
  135. extern long mlock_vma_pages_range(struct vm_area_struct *vma,
  136. unsigned long start, unsigned long end);
  137. extern void munlock_vma_pages_range(struct vm_area_struct *vma,
  138. unsigned long start, unsigned long end);
  139. static inline void munlock_vma_pages_all(struct vm_area_struct *vma)
  140. {
  141. munlock_vma_pages_range(vma, vma->vm_start, vma->vm_end);
  142. }
  143. /*
  144. * Called only in fault path via page_evictable() for a new page
  145. * to determine if it's being mapped into a LOCKED vma.
  146. * If so, mark page as mlocked.
  147. */
  148. static inline int is_mlocked_vma(struct vm_area_struct *vma, struct page *page)
  149. {
  150. VM_BUG_ON(PageLRU(page));
  151. if (likely((vma->vm_flags & (VM_LOCKED | VM_SPECIAL)) != VM_LOCKED))
  152. return 0;
  153. if (!TestSetPageMlocked(page)) {
  154. inc_zone_page_state(page, NR_MLOCK);
  155. count_vm_event(UNEVICTABLE_PGMLOCKED);
  156. }
  157. return 1;
  158. }
  159. /*
  160. * must be called with vma's mmap_sem held for read or write, and page locked.
  161. */
  162. extern void mlock_vma_page(struct page *page);
  163. extern void munlock_vma_page(struct page *page);
  164. /*
  165. * Clear the page's PageMlocked(). This can be useful in a situation where
  166. * we want to unconditionally remove a page from the pagecache -- e.g.,
  167. * on truncation or freeing.
  168. *
  169. * It is legal to call this function for any page, mlocked or not.
  170. * If called for a page that is still mapped by mlocked vmas, all we do
  171. * is revert to lazy LRU behaviour -- semantics are not broken.
  172. */
  173. extern void __clear_page_mlock(struct page *page);
  174. static inline void clear_page_mlock(struct page *page)
  175. {
  176. if (unlikely(TestClearPageMlocked(page)))
  177. __clear_page_mlock(page);
  178. }
  179. /*
  180. * mlock_migrate_page - called only from migrate_page_copy() to
  181. * migrate the Mlocked page flag; update statistics.
  182. */
  183. static inline void mlock_migrate_page(struct page *newpage, struct page *page)
  184. {
  185. if (TestClearPageMlocked(page)) {
  186. unsigned long flags;
  187. local_irq_save(flags);
  188. __dec_zone_page_state(page, NR_MLOCK);
  189. SetPageMlocked(newpage);
  190. __inc_zone_page_state(newpage, NR_MLOCK);
  191. local_irq_restore(flags);
  192. }
  193. }
  194. #ifdef CONFIG_TRANSPARENT_HUGEPAGE
  195. extern unsigned long vma_address(struct page *page,
  196. struct vm_area_struct *vma);
  197. #endif
  198. #else /* !CONFIG_MMU */
  199. static inline int is_mlocked_vma(struct vm_area_struct *v, struct page *p)
  200. {
  201. return 0;
  202. }
  203. static inline void clear_page_mlock(struct page *page) { }
  204. static inline void mlock_vma_page(struct page *page) { }
  205. static inline void mlock_migrate_page(struct page *new, struct page *old) { }
  206. #endif /* !CONFIG_MMU */
  207. /*
  208. * Return the mem_map entry representing the 'offset' subpage within
  209. * the maximally aligned gigantic page 'base'. Handle any discontiguity
  210. * in the mem_map at MAX_ORDER_NR_PAGES boundaries.
  211. */
  212. static inline struct page *mem_map_offset(struct page *base, int offset)
  213. {
  214. if (unlikely(offset >= MAX_ORDER_NR_PAGES))
  215. return pfn_to_page(page_to_pfn(base) + offset);
  216. return base + offset;
  217. }
  218. /*
  219. * Iterator over all subpages within the maximally aligned gigantic
  220. * page 'base'. Handle any discontiguity in the mem_map.
  221. */
  222. static inline struct page *mem_map_next(struct page *iter,
  223. struct page *base, int offset)
  224. {
  225. if (unlikely((offset & (MAX_ORDER_NR_PAGES - 1)) == 0)) {
  226. unsigned long pfn = page_to_pfn(base) + offset;
  227. if (!pfn_valid(pfn))
  228. return NULL;
  229. return pfn_to_page(pfn);
  230. }
  231. return iter + 1;
  232. }
  233. /*
  234. * FLATMEM and DISCONTIGMEM configurations use alloc_bootmem_node,
  235. * so all functions starting at paging_init should be marked __init
  236. * in those cases. SPARSEMEM, however, allows for memory hotplug,
  237. * and alloc_bootmem_node is not used.
  238. */
  239. #ifdef CONFIG_SPARSEMEM
  240. #define __paginginit __meminit
  241. #else
  242. #define __paginginit __init
  243. #endif
  244. /* Memory initialisation debug and verification */
  245. enum mminit_level {
  246. MMINIT_WARNING,
  247. MMINIT_VERIFY,
  248. MMINIT_TRACE
  249. };
  250. #ifdef CONFIG_DEBUG_MEMORY_INIT
  251. extern int mminit_loglevel;
  252. #define mminit_dprintk(level, prefix, fmt, arg...) \
  253. do { \
  254. if (level < mminit_loglevel) { \
  255. printk(level <= MMINIT_WARNING ? KERN_WARNING : KERN_DEBUG); \
  256. printk(KERN_CONT "mminit::" prefix " " fmt, ##arg); \
  257. } \
  258. } while (0)
  259. extern void mminit_verify_pageflags_layout(void);
  260. extern void mminit_verify_page_links(struct page *page,
  261. enum zone_type zone, unsigned long nid, unsigned long pfn);
  262. extern void mminit_verify_zonelist(void);
  263. #else
  264. static inline void mminit_dprintk(enum mminit_level level,
  265. const char *prefix, const char *fmt, ...)
  266. {
  267. }
  268. static inline void mminit_verify_pageflags_layout(void)
  269. {
  270. }
  271. static inline void mminit_verify_page_links(struct page *page,
  272. enum zone_type zone, unsigned long nid, unsigned long pfn)
  273. {
  274. }
  275. static inline void mminit_verify_zonelist(void)
  276. {
  277. }
  278. #endif /* CONFIG_DEBUG_MEMORY_INIT */
  279. /* mminit_validate_memmodel_limits is independent of CONFIG_DEBUG_MEMORY_INIT */
  280. #if defined(CONFIG_SPARSEMEM)
  281. extern void mminit_validate_memmodel_limits(unsigned long *start_pfn,
  282. unsigned long *end_pfn);
  283. #else
  284. static inline void mminit_validate_memmodel_limits(unsigned long *start_pfn,
  285. unsigned long *end_pfn)
  286. {
  287. }
  288. #endif /* CONFIG_SPARSEMEM */
  289. #define ZONE_RECLAIM_NOSCAN -2
  290. #define ZONE_RECLAIM_FULL -1
  291. #define ZONE_RECLAIM_SOME 0
  292. #define ZONE_RECLAIM_SUCCESS 1
  293. #endif
  294. extern int hwpoison_filter(struct page *p);
  295. extern u32 hwpoison_filter_dev_major;
  296. extern u32 hwpoison_filter_dev_minor;
  297. extern u64 hwpoison_filter_flags_mask;
  298. extern u64 hwpoison_filter_flags_value;
  299. extern u64 hwpoison_filter_memcg;
  300. extern u32 hwpoison_filter_enable;