Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 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 | */ |
Nick Piggin | 0f8053a | 2006-03-22 00:08:33 -0800 | [diff] [blame] | 11 | #ifndef __MM_INTERNAL_H |
| 12 | #define __MM_INTERNAL_H |
| 13 | |
| 14 | #include <linux/mm.h> |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 15 | |
Jan Beulich | 42b7772 | 2008-07-23 21:27:10 -0700 | [diff] [blame] | 16 | void free_pgtables(struct mmu_gather *tlb, struct vm_area_struct *start_vma, |
| 17 | unsigned long floor, unsigned long ceiling); |
| 18 | |
Nick Piggin | 7835e98 | 2006-03-22 00:08:40 -0800 | [diff] [blame] | 19 | static inline void set_page_count(struct page *page, int v) |
Nick Piggin | 77a8a78 | 2006-01-06 00:10:57 -0800 | [diff] [blame] | 20 | { |
Nick Piggin | 7835e98 | 2006-03-22 00:08:40 -0800 | [diff] [blame] | 21 | atomic_set(&page->_count, v); |
| 22 | } |
| 23 | |
| 24 | /* |
| 25 | * Turn a non-refcounted page (->_count == 0) into refcounted with |
| 26 | * a count of one. |
| 27 | */ |
| 28 | static inline void set_page_refcounted(struct page *page) |
| 29 | { |
Qi Yong | ae1276b | 2008-02-04 22:29:27 -0800 | [diff] [blame] | 30 | VM_BUG_ON(PageTail(page)); |
Nick Piggin | 725d704 | 2006-09-25 23:30:55 -0700 | [diff] [blame] | 31 | VM_BUG_ON(atomic_read(&page->_count)); |
Nick Piggin | 77a8a78 | 2006-01-06 00:10:57 -0800 | [diff] [blame] | 32 | set_page_count(page, 1); |
Nick Piggin | 77a8a78 | 2006-01-06 00:10:57 -0800 | [diff] [blame] | 33 | } |
| 34 | |
Nick Piggin | 0f8053a | 2006-03-22 00:08:33 -0800 | [diff] [blame] | 35 | static inline void __put_page(struct page *page) |
| 36 | { |
| 37 | atomic_dec(&page->_count); |
| 38 | } |
| 39 | |
Hugh Dickins | 03f6462 | 2009-09-21 17:03:35 -0700 | [diff] [blame] | 40 | extern unsigned long highest_memmap_pfn; |
| 41 | |
Lee Schermerhorn | 894bc31 | 2008-10-18 20:26:39 -0700 | [diff] [blame] | 42 | /* |
| 43 | * in mm/vmscan.c: |
| 44 | */ |
Nick Piggin | 62695a8 | 2008-10-18 20:26:09 -0700 | [diff] [blame] | 45 | extern int isolate_lru_page(struct page *page); |
Lee Schermerhorn | 894bc31 | 2008-10-18 20:26:39 -0700 | [diff] [blame] | 46 | extern void putback_lru_page(struct page *page); |
Nick Piggin | 62695a8 | 2008-10-18 20:26:09 -0700 | [diff] [blame] | 47 | |
Lee Schermerhorn | 894bc31 | 2008-10-18 20:26:39 -0700 | [diff] [blame] | 48 | /* |
| 49 | * in mm/page_alloc.c |
| 50 | */ |
Yasunori Goto | 0c0a4a5 | 2008-04-28 02:13:34 -0700 | [diff] [blame] | 51 | extern void __free_pages_bootmem(struct page *page, unsigned int order); |
Wu Fengguang | 20a0307 | 2009-06-16 15:32:22 -0700 | [diff] [blame] | 52 | extern void prep_compound_page(struct page *page, unsigned long order); |
Wu Fengguang | 8d22ba1 | 2009-12-16 12:19:58 +0100 | [diff] [blame] | 53 | #ifdef CONFIG_MEMORY_FAILURE |
| 54 | extern bool is_free_buddy_page(struct page *page); |
| 55 | #endif |
Wu Fengguang | 20a0307 | 2009-06-16 15:32:22 -0700 | [diff] [blame] | 56 | |
Nick Piggin | 0f8053a | 2006-03-22 00:08:33 -0800 | [diff] [blame] | 57 | |
Mel Gorman | 48f13bf | 2007-10-16 01:26:10 -0700 | [diff] [blame] | 58 | /* |
| 59 | * function for dealing with page's order in buddy system. |
| 60 | * zone->lock is already acquired when we use these. |
| 61 | * So, we don't need atomic page->flags operations here. |
| 62 | */ |
| 63 | static inline unsigned long page_order(struct page *page) |
| 64 | { |
| 65 | VM_BUG_ON(!PageBuddy(page)); |
| 66 | return page_private(page); |
| 67 | } |
Alexander van Heukelum | b5a0e01 | 2008-02-23 15:24:06 -0800 | [diff] [blame] | 68 | |
Hugh Dickins | af8e335 | 2009-12-14 17:58:59 -0800 | [diff] [blame] | 69 | #ifdef CONFIG_MMU |
| 70 | extern long mlock_vma_pages_range(struct vm_area_struct *vma, |
| 71 | unsigned long start, unsigned long end); |
| 72 | extern void munlock_vma_pages_range(struct vm_area_struct *vma, |
| 73 | unsigned long start, unsigned long end); |
| 74 | static inline void munlock_vma_pages_all(struct vm_area_struct *vma) |
| 75 | { |
| 76 | munlock_vma_pages_range(vma, vma->vm_start, vma->vm_end); |
| 77 | } |
| 78 | |
Nick Piggin | b291f00 | 2008-10-18 20:26:44 -0700 | [diff] [blame] | 79 | /* |
| 80 | * Called only in fault path via page_evictable() for a new page |
| 81 | * to determine if it's being mapped into a LOCKED vma. |
| 82 | * If so, mark page as mlocked. |
| 83 | */ |
| 84 | static inline int is_mlocked_vma(struct vm_area_struct *vma, struct page *page) |
| 85 | { |
| 86 | VM_BUG_ON(PageLRU(page)); |
| 87 | |
| 88 | if (likely((vma->vm_flags & (VM_LOCKED | VM_SPECIAL)) != VM_LOCKED)) |
| 89 | return 0; |
| 90 | |
Nick Piggin | 5344b7e | 2008-10-18 20:26:51 -0700 | [diff] [blame] | 91 | if (!TestSetPageMlocked(page)) { |
| 92 | inc_zone_page_state(page, NR_MLOCK); |
| 93 | count_vm_event(UNEVICTABLE_PGMLOCKED); |
| 94 | } |
Nick Piggin | b291f00 | 2008-10-18 20:26:44 -0700 | [diff] [blame] | 95 | return 1; |
| 96 | } |
| 97 | |
| 98 | /* |
Hugh Dickins | 73848b4 | 2009-12-14 17:59:22 -0800 | [diff] [blame] | 99 | * must be called with vma's mmap_sem held for read or write, and page locked. |
Nick Piggin | b291f00 | 2008-10-18 20:26:44 -0700 | [diff] [blame] | 100 | */ |
| 101 | extern void mlock_vma_page(struct page *page); |
Hugh Dickins | 73848b4 | 2009-12-14 17:59:22 -0800 | [diff] [blame] | 102 | extern void munlock_vma_page(struct page *page); |
Nick Piggin | b291f00 | 2008-10-18 20:26:44 -0700 | [diff] [blame] | 103 | |
| 104 | /* |
| 105 | * Clear the page's PageMlocked(). This can be useful in a situation where |
| 106 | * we want to unconditionally remove a page from the pagecache -- e.g., |
| 107 | * on truncation or freeing. |
| 108 | * |
| 109 | * It is legal to call this function for any page, mlocked or not. |
| 110 | * If called for a page that is still mapped by mlocked vmas, all we do |
| 111 | * is revert to lazy LRU behaviour -- semantics are not broken. |
| 112 | */ |
| 113 | extern void __clear_page_mlock(struct page *page); |
| 114 | static inline void clear_page_mlock(struct page *page) |
| 115 | { |
| 116 | if (unlikely(TestClearPageMlocked(page))) |
| 117 | __clear_page_mlock(page); |
| 118 | } |
| 119 | |
| 120 | /* |
| 121 | * mlock_migrate_page - called only from migrate_page_copy() to |
Nick Piggin | 5344b7e | 2008-10-18 20:26:51 -0700 | [diff] [blame] | 122 | * migrate the Mlocked page flag; update statistics. |
Nick Piggin | b291f00 | 2008-10-18 20:26:44 -0700 | [diff] [blame] | 123 | */ |
| 124 | static inline void mlock_migrate_page(struct page *newpage, struct page *page) |
| 125 | { |
Nick Piggin | 5344b7e | 2008-10-18 20:26:51 -0700 | [diff] [blame] | 126 | if (TestClearPageMlocked(page)) { |
| 127 | unsigned long flags; |
| 128 | |
| 129 | local_irq_save(flags); |
| 130 | __dec_zone_page_state(page, NR_MLOCK); |
Nick Piggin | b291f00 | 2008-10-18 20:26:44 -0700 | [diff] [blame] | 131 | SetPageMlocked(newpage); |
Nick Piggin | 5344b7e | 2008-10-18 20:26:51 -0700 | [diff] [blame] | 132 | __inc_zone_page_state(newpage, NR_MLOCK); |
| 133 | local_irq_restore(flags); |
| 134 | } |
Nick Piggin | b291f00 | 2008-10-18 20:26:44 -0700 | [diff] [blame] | 135 | } |
| 136 | |
Hugh Dickins | af8e335 | 2009-12-14 17:58:59 -0800 | [diff] [blame] | 137 | #else /* !CONFIG_MMU */ |
Nick Piggin | b291f00 | 2008-10-18 20:26:44 -0700 | [diff] [blame] | 138 | static inline int is_mlocked_vma(struct vm_area_struct *v, struct page *p) |
| 139 | { |
| 140 | return 0; |
| 141 | } |
| 142 | static inline void clear_page_mlock(struct page *page) { } |
| 143 | static inline void mlock_vma_page(struct page *page) { } |
| 144 | static inline void mlock_migrate_page(struct page *new, struct page *old) { } |
| 145 | |
Hugh Dickins | af8e335 | 2009-12-14 17:58:59 -0800 | [diff] [blame] | 146 | #endif /* !CONFIG_MMU */ |
Lee Schermerhorn | 894bc31 | 2008-10-18 20:26:39 -0700 | [diff] [blame] | 147 | |
Alexander van Heukelum | b5a0e01 | 2008-02-23 15:24:06 -0800 | [diff] [blame] | 148 | /* |
Andy Whitcroft | 69d177c | 2008-11-06 12:53:26 -0800 | [diff] [blame] | 149 | * Return the mem_map entry representing the 'offset' subpage within |
| 150 | * the maximally aligned gigantic page 'base'. Handle any discontiguity |
| 151 | * in the mem_map at MAX_ORDER_NR_PAGES boundaries. |
| 152 | */ |
| 153 | static inline struct page *mem_map_offset(struct page *base, int offset) |
| 154 | { |
| 155 | if (unlikely(offset >= MAX_ORDER_NR_PAGES)) |
| 156 | return pfn_to_page(page_to_pfn(base) + offset); |
| 157 | return base + offset; |
| 158 | } |
| 159 | |
| 160 | /* |
| 161 | * Iterator over all subpages withing the maximally aligned gigantic |
| 162 | * page 'base'. Handle any discontiguity in the mem_map. |
| 163 | */ |
| 164 | static inline struct page *mem_map_next(struct page *iter, |
| 165 | struct page *base, int offset) |
| 166 | { |
| 167 | if (unlikely((offset & (MAX_ORDER_NR_PAGES - 1)) == 0)) { |
| 168 | unsigned long pfn = page_to_pfn(base) + offset; |
| 169 | if (!pfn_valid(pfn)) |
| 170 | return NULL; |
| 171 | return pfn_to_page(pfn); |
| 172 | } |
| 173 | return iter + 1; |
| 174 | } |
| 175 | |
| 176 | /* |
Alexander van Heukelum | b5a0e01 | 2008-02-23 15:24:06 -0800 | [diff] [blame] | 177 | * FLATMEM and DISCONTIGMEM configurations use alloc_bootmem_node, |
| 178 | * so all functions starting at paging_init should be marked __init |
| 179 | * in those cases. SPARSEMEM, however, allows for memory hotplug, |
| 180 | * and alloc_bootmem_node is not used. |
| 181 | */ |
| 182 | #ifdef CONFIG_SPARSEMEM |
| 183 | #define __paginginit __meminit |
| 184 | #else |
| 185 | #define __paginginit __init |
| 186 | #endif |
| 187 | |
Mel Gorman | 6b74ab9 | 2008-07-23 21:26:49 -0700 | [diff] [blame] | 188 | /* Memory initialisation debug and verification */ |
| 189 | enum mminit_level { |
| 190 | MMINIT_WARNING, |
| 191 | MMINIT_VERIFY, |
| 192 | MMINIT_TRACE |
| 193 | }; |
| 194 | |
| 195 | #ifdef CONFIG_DEBUG_MEMORY_INIT |
| 196 | |
| 197 | extern int mminit_loglevel; |
| 198 | |
| 199 | #define mminit_dprintk(level, prefix, fmt, arg...) \ |
| 200 | do { \ |
| 201 | if (level < mminit_loglevel) { \ |
| 202 | printk(level <= MMINIT_WARNING ? KERN_WARNING : KERN_DEBUG); \ |
| 203 | printk(KERN_CONT "mminit::" prefix " " fmt, ##arg); \ |
| 204 | } \ |
| 205 | } while (0) |
| 206 | |
Mel Gorman | 708614e | 2008-07-23 21:26:51 -0700 | [diff] [blame] | 207 | extern void mminit_verify_pageflags_layout(void); |
| 208 | extern void mminit_verify_page_links(struct page *page, |
| 209 | enum zone_type zone, unsigned long nid, unsigned long pfn); |
Mel Gorman | 68ad8df | 2008-07-23 21:26:52 -0700 | [diff] [blame] | 210 | extern void mminit_verify_zonelist(void); |
Mel Gorman | 708614e | 2008-07-23 21:26:51 -0700 | [diff] [blame] | 211 | |
Mel Gorman | 6b74ab9 | 2008-07-23 21:26:49 -0700 | [diff] [blame] | 212 | #else |
| 213 | |
| 214 | static inline void mminit_dprintk(enum mminit_level level, |
| 215 | const char *prefix, const char *fmt, ...) |
| 216 | { |
| 217 | } |
| 218 | |
Mel Gorman | 708614e | 2008-07-23 21:26:51 -0700 | [diff] [blame] | 219 | static inline void mminit_verify_pageflags_layout(void) |
| 220 | { |
| 221 | } |
| 222 | |
| 223 | static inline void mminit_verify_page_links(struct page *page, |
| 224 | enum zone_type zone, unsigned long nid, unsigned long pfn) |
| 225 | { |
| 226 | } |
Mel Gorman | 68ad8df | 2008-07-23 21:26:52 -0700 | [diff] [blame] | 227 | |
| 228 | static inline void mminit_verify_zonelist(void) |
| 229 | { |
| 230 | } |
Mel Gorman | 6b74ab9 | 2008-07-23 21:26:49 -0700 | [diff] [blame] | 231 | #endif /* CONFIG_DEBUG_MEMORY_INIT */ |
Mel Gorman | 2dbb51c | 2008-07-23 21:26:52 -0700 | [diff] [blame] | 232 | |
| 233 | /* mminit_validate_memmodel_limits is independent of CONFIG_DEBUG_MEMORY_INIT */ |
| 234 | #if defined(CONFIG_SPARSEMEM) |
| 235 | extern void mminit_validate_memmodel_limits(unsigned long *start_pfn, |
| 236 | unsigned long *end_pfn); |
| 237 | #else |
| 238 | static inline void mminit_validate_memmodel_limits(unsigned long *start_pfn, |
| 239 | unsigned long *end_pfn) |
| 240 | { |
| 241 | } |
| 242 | #endif /* CONFIG_SPARSEMEM */ |
| 243 | |
Nick Piggin | b291f00 | 2008-10-18 20:26:44 -0700 | [diff] [blame] | 244 | int __get_user_pages(struct task_struct *tsk, struct mm_struct *mm, |
Hugh Dickins | 58fa879 | 2009-09-21 17:03:31 -0700 | [diff] [blame] | 245 | unsigned long start, int len, unsigned int foll_flags, |
Nick Piggin | b291f00 | 2008-10-18 20:26:44 -0700 | [diff] [blame] | 246 | struct page **pages, struct vm_area_struct **vmas); |
| 247 | |
Mel Gorman | fa5e084 | 2009-06-16 15:33:22 -0700 | [diff] [blame] | 248 | #define ZONE_RECLAIM_NOSCAN -2 |
| 249 | #define ZONE_RECLAIM_FULL -1 |
| 250 | #define ZONE_RECLAIM_SOME 0 |
| 251 | #define ZONE_RECLAIM_SUCCESS 1 |
Nick Piggin | 0f8053a | 2006-03-22 00:08:33 -0800 | [diff] [blame] | 252 | #endif |
Wu Fengguang | 7c116f2 | 2009-12-16 12:19:59 +0100 | [diff] [blame] | 253 | |
Wu Fengguang | 31d3d34 | 2009-12-16 12:19:59 +0100 | [diff] [blame] | 254 | extern int hwpoison_filter(struct page *p); |
| 255 | |
Wu Fengguang | 7c116f2 | 2009-12-16 12:19:59 +0100 | [diff] [blame] | 256 | extern u32 hwpoison_filter_dev_major; |
| 257 | extern u32 hwpoison_filter_dev_minor; |
Wu Fengguang | 478c5ff | 2009-12-16 12:19:59 +0100 | [diff] [blame] | 258 | extern u64 hwpoison_filter_flags_mask; |
| 259 | extern u64 hwpoison_filter_flags_value; |
Andi Kleen | 4fd466e | 2009-12-16 12:19:59 +0100 | [diff] [blame] | 260 | extern u64 hwpoison_filter_memcg; |
Haicheng Li | 1bfe5fe | 2009-12-16 12:19:59 +0100 | [diff] [blame] | 261 | extern u32 hwpoison_filter_enable; |