Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1 | #ifndef _LINUX_MM_H |
| 2 | #define _LINUX_MM_H |
| 3 | |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 4 | #include <linux/errno.h> |
| 5 | |
| 6 | #ifdef __KERNEL__ |
| 7 | |
Sasha Levin | 309381fea | 2014-01-23 15:52:54 -0800 | [diff] [blame] | 8 | #include <linux/mmdebug.h> |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 9 | #include <linux/gfp.h> |
Paul Gortmaker | 187f188 | 2011-11-23 20:12:59 -0500 | [diff] [blame] | 10 | #include <linux/bug.h> |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 11 | #include <linux/list.h> |
| 12 | #include <linux/mmzone.h> |
| 13 | #include <linux/rbtree.h> |
Konstantin Khlebnikov | 83aeead | 2011-12-08 14:33:54 -0800 | [diff] [blame] | 14 | #include <linux/atomic.h> |
Ingo Molnar | 9a11b49a | 2006-07-03 00:24:33 -0700 | [diff] [blame] | 15 | #include <linux/debug_locks.h> |
Heiko Carstens | 5b99cd0 | 2006-09-27 01:50:01 -0700 | [diff] [blame] | 16 | #include <linux/mm_types.h> |
Yinghai Lu | 0867721 | 2010-02-10 01:20:20 -0800 | [diff] [blame] | 17 | #include <linux/range.h> |
Chris Metcalf | c6f6b59 | 2010-05-24 14:32:53 -0700 | [diff] [blame] | 18 | #include <linux/pfn.h> |
Andrea Arcangeli | e9da73d | 2011-01-13 15:46:32 -0800 | [diff] [blame] | 19 | #include <linux/bit_spinlock.h> |
Dave Chinner | b0d40c9 | 2011-07-08 14:14:42 +1000 | [diff] [blame] | 20 | #include <linux/shrinker.h> |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 21 | |
| 22 | struct mempolicy; |
| 23 | struct anon_vma; |
Michel Lespinasse | bf181b9 | 2012-10-08 16:31:39 -0700 | [diff] [blame] | 24 | struct anon_vma_chain; |
Alexey Dobriyan | 4e950f6 | 2007-07-30 02:36:13 +0400 | [diff] [blame] | 25 | struct file_ra_state; |
Alexey Dobriyan | e8edc6e | 2007-05-21 01:22:52 +0400 | [diff] [blame] | 26 | struct user_struct; |
Alexey Dobriyan | 4e950f6 | 2007-07-30 02:36:13 +0400 | [diff] [blame] | 27 | struct writeback_control; |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 28 | |
Jiang Liu | fccc998 | 2013-07-03 15:04:23 -0700 | [diff] [blame] | 29 | #ifndef CONFIG_NEED_MULTIPLE_NODES /* Don't use mapnrs, do it properly */ |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 30 | extern unsigned long max_mapnr; |
Jiang Liu | fccc998 | 2013-07-03 15:04:23 -0700 | [diff] [blame] | 31 | |
| 32 | static inline void set_max_mapnr(unsigned long limit) |
| 33 | { |
| 34 | max_mapnr = limit; |
| 35 | } |
| 36 | #else |
| 37 | static inline void set_max_mapnr(unsigned long limit) { } |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 38 | #endif |
| 39 | |
Jan Beulich | 4481374 | 2009-09-21 17:03:05 -0700 | [diff] [blame] | 40 | extern unsigned long totalram_pages; |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 41 | extern void * high_memory; |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 42 | extern int page_cluster; |
| 43 | |
| 44 | #ifdef CONFIG_SYSCTL |
| 45 | extern int sysctl_legacy_va_layout; |
| 46 | #else |
| 47 | #define sysctl_legacy_va_layout 0 |
| 48 | #endif |
| 49 | |
| 50 | #include <asm/page.h> |
| 51 | #include <asm/pgtable.h> |
| 52 | #include <asm/processor.h> |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 53 | |
Tang Chen | 79442ed | 2013-11-12 15:07:59 -0800 | [diff] [blame] | 54 | #ifndef __pa_symbol |
| 55 | #define __pa_symbol(x) __pa(RELOC_HIDE((unsigned long)(x), 0)) |
| 56 | #endif |
| 57 | |
Andrew Shewmaker | c9b1d09 | 2013-04-29 15:08:10 -0700 | [diff] [blame] | 58 | extern unsigned long sysctl_user_reserve_kbytes; |
Andrew Shewmaker | 4eeab4f | 2013-04-29 15:08:11 -0700 | [diff] [blame] | 59 | extern unsigned long sysctl_admin_reserve_kbytes; |
Andrew Shewmaker | c9b1d09 | 2013-04-29 15:08:10 -0700 | [diff] [blame] | 60 | |
Jerome Marchand | 49f0ce5 | 2014-01-21 15:49:14 -0800 | [diff] [blame] | 61 | extern int sysctl_overcommit_memory; |
| 62 | extern int sysctl_overcommit_ratio; |
| 63 | extern unsigned long sysctl_overcommit_kbytes; |
| 64 | |
| 65 | extern int overcommit_ratio_handler(struct ctl_table *, int, void __user *, |
| 66 | size_t *, loff_t *); |
| 67 | extern int overcommit_kbytes_handler(struct ctl_table *, int, void __user *, |
| 68 | size_t *, loff_t *); |
| 69 | |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 70 | #define nth_page(page,n) pfn_to_page(page_to_pfn((page)) + (n)) |
| 71 | |
Andrea Righi | 27ac792 | 2008-07-23 21:28:13 -0700 | [diff] [blame] | 72 | /* to align the pointer to the (next) page boundary */ |
| 73 | #define PAGE_ALIGN(addr) ALIGN(addr, PAGE_SIZE) |
| 74 | |
Andrew Morton | 0fa73b8 | 2013-07-03 15:02:11 -0700 | [diff] [blame] | 75 | /* test whether an address (unsigned long or pointer) is aligned to PAGE_SIZE */ |
| 76 | #define PAGE_ALIGNED(addr) IS_ALIGNED((unsigned long)addr, PAGE_SIZE) |
| 77 | |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 78 | /* |
| 79 | * Linux kernel virtual memory manager primitives. |
| 80 | * The idea being to have a "virtual" mm in the same way |
| 81 | * we have a virtual fs - giving a cleaner interface to the |
| 82 | * mm details, and allowing different kinds of memory mappings |
| 83 | * (from shared memory to executable loading to arbitrary |
| 84 | * mmap() functions). |
| 85 | */ |
| 86 | |
Christoph Lameter | c43692e | 2006-12-06 20:32:48 -0800 | [diff] [blame] | 87 | extern struct kmem_cache *vm_area_cachep; |
| 88 | |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 89 | #ifndef CONFIG_MMU |
David Howells | 8feae13 | 2009-01-08 12:04:47 +0000 | [diff] [blame] | 90 | extern struct rb_root nommu_region_tree; |
| 91 | extern struct rw_semaphore nommu_region_sem; |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 92 | |
| 93 | extern unsigned int kobjsize(const void *objp); |
| 94 | #endif |
| 95 | |
| 96 | /* |
Hugh Dickins | 605d928 | 2008-08-16 11:07:21 +0100 | [diff] [blame] | 97 | * vm_flags in vm_area_struct, see mm_types.h. |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 98 | */ |
Konstantin Khlebnikov | cc2383e | 2012-10-08 16:28:37 -0700 | [diff] [blame] | 99 | #define VM_NONE 0x00000000 |
| 100 | |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 101 | #define VM_READ 0x00000001 /* currently active flags */ |
| 102 | #define VM_WRITE 0x00000002 |
| 103 | #define VM_EXEC 0x00000004 |
| 104 | #define VM_SHARED 0x00000008 |
| 105 | |
Paolo 'Blaisorblade' Giarrusso | 7e2cff4 | 2005-09-21 09:55:39 -0700 | [diff] [blame] | 106 | /* mprotect() hardcodes VM_MAYREAD >> 4 == VM_READ, and so for r/w/x bits. */ |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 107 | #define VM_MAYREAD 0x00000010 /* limits for mprotect() etc */ |
| 108 | #define VM_MAYWRITE 0x00000020 |
| 109 | #define VM_MAYEXEC 0x00000040 |
| 110 | #define VM_MAYSHARE 0x00000080 |
| 111 | |
| 112 | #define VM_GROWSDOWN 0x00000100 /* general info on the segment */ |
Linus Torvalds | 6aab341 | 2005-11-28 14:34:23 -0800 | [diff] [blame] | 113 | #define VM_PFNMAP 0x00000400 /* Page-ranges managed without "struct page", just pure PFN */ |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 114 | #define VM_DENYWRITE 0x00000800 /* ETXTBSY on write attempts.. */ |
| 115 | |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 116 | #define VM_LOCKED 0x00002000 |
| 117 | #define VM_IO 0x00004000 /* Memory mapped I/O or similar */ |
| 118 | |
| 119 | /* Used by sys_madvise() */ |
| 120 | #define VM_SEQ_READ 0x00008000 /* App will access data sequentially */ |
| 121 | #define VM_RAND_READ 0x00010000 /* App will not benefit from clustered reads */ |
| 122 | |
| 123 | #define VM_DONTCOPY 0x00020000 /* Do not copy this vma on fork */ |
| 124 | #define VM_DONTEXPAND 0x00040000 /* Cannot expand with mremap() */ |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 125 | #define VM_ACCOUNT 0x00100000 /* Is a VM accounted object */ |
Andy Whitcroft | cdfd432 | 2008-07-23 21:27:28 -0700 | [diff] [blame] | 126 | #define VM_NORESERVE 0x00200000 /* should the VM suppress accounting */ |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 127 | #define VM_HUGETLB 0x00400000 /* Huge TLB Page VM */ |
| 128 | #define VM_NONLINEAR 0x00800000 /* Is non-linear (remap_file_pages) */ |
Konstantin Khlebnikov | cc2383e | 2012-10-08 16:28:37 -0700 | [diff] [blame] | 129 | #define VM_ARCH_1 0x01000000 /* Architecture-specific flag */ |
Konstantin Khlebnikov | 0103bd1 | 2012-10-08 16:28:59 -0700 | [diff] [blame] | 130 | #define VM_DONTDUMP 0x04000000 /* Do not include in the core dump */ |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 131 | |
Cyrill Gorcunov | d9104d1 | 2013-09-11 14:22:24 -0700 | [diff] [blame] | 132 | #ifdef CONFIG_MEM_SOFT_DIRTY |
| 133 | # define VM_SOFTDIRTY 0x08000000 /* Not soft dirty clean area */ |
| 134 | #else |
| 135 | # define VM_SOFTDIRTY 0 |
| 136 | #endif |
| 137 | |
Jared Hulbert | b379d79 | 2008-04-28 02:12:58 -0700 | [diff] [blame] | 138 | #define VM_MIXEDMAP 0x10000000 /* Can contain "struct page" and pure PFN pages */ |
Konstantin Khlebnikov | cc2383e | 2012-10-08 16:28:37 -0700 | [diff] [blame] | 139 | #define VM_HUGEPAGE 0x20000000 /* MADV_HUGEPAGE marked this vma */ |
| 140 | #define VM_NOHUGEPAGE 0x40000000 /* MADV_NOHUGEPAGE marked this vma */ |
Hugh Dickins | f8af4da | 2009-09-21 17:01:57 -0700 | [diff] [blame] | 141 | #define VM_MERGEABLE 0x80000000 /* KSM may merge identical pages */ |
Nick Piggin | d00806b | 2007-07-19 01:46:57 -0700 | [diff] [blame] | 142 | |
Konstantin Khlebnikov | cc2383e | 2012-10-08 16:28:37 -0700 | [diff] [blame] | 143 | #if defined(CONFIG_X86) |
| 144 | # define VM_PAT VM_ARCH_1 /* PAT reserves whole VMA at once (x86) */ |
| 145 | #elif defined(CONFIG_PPC) |
| 146 | # define VM_SAO VM_ARCH_1 /* Strong Access Ordering (powerpc) */ |
| 147 | #elif defined(CONFIG_PARISC) |
| 148 | # define VM_GROWSUP VM_ARCH_1 |
James Hogan | 9ca52ed | 2012-10-16 10:16:14 +0100 | [diff] [blame] | 149 | #elif defined(CONFIG_METAG) |
| 150 | # define VM_GROWSUP VM_ARCH_1 |
Konstantin Khlebnikov | cc2383e | 2012-10-08 16:28:37 -0700 | [diff] [blame] | 151 | #elif defined(CONFIG_IA64) |
| 152 | # define VM_GROWSUP VM_ARCH_1 |
| 153 | #elif !defined(CONFIG_MMU) |
| 154 | # define VM_MAPPED_COPY VM_ARCH_1 /* T if mapped copy of data (nommu mmap) */ |
| 155 | #endif |
| 156 | |
| 157 | #ifndef VM_GROWSUP |
| 158 | # define VM_GROWSUP VM_NONE |
| 159 | #endif |
| 160 | |
Mel Gorman | a8bef8ff | 2010-05-24 14:32:24 -0700 | [diff] [blame] | 161 | /* Bits set in the VMA until the stack is in its final location */ |
| 162 | #define VM_STACK_INCOMPLETE_SETUP (VM_RAND_READ | VM_SEQ_READ) |
| 163 | |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 164 | #ifndef VM_STACK_DEFAULT_FLAGS /* arch can override this */ |
| 165 | #define VM_STACK_DEFAULT_FLAGS VM_DATA_DEFAULT_FLAGS |
| 166 | #endif |
| 167 | |
| 168 | #ifdef CONFIG_STACK_GROWSUP |
| 169 | #define VM_STACK_FLAGS (VM_GROWSUP | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT) |
| 170 | #else |
| 171 | #define VM_STACK_FLAGS (VM_GROWSDOWN | VM_STACK_DEFAULT_FLAGS | VM_ACCOUNT) |
| 172 | #endif |
| 173 | |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 174 | /* |
Andrea Arcangeli | 78f11a2 | 2011-04-27 15:26:45 -0700 | [diff] [blame] | 175 | * Special vmas that are non-mergable, non-mlock()able. |
| 176 | * Note: mm/huge_memory.c VM_NO_THP depends on this definition. |
Nick Piggin | b291f00 | 2008-10-18 20:26:44 -0700 | [diff] [blame] | 177 | */ |
Vlastimil Babka | 9050d7e | 2014-03-03 15:38:27 -0800 | [diff] [blame] | 178 | #define VM_SPECIAL (VM_IO | VM_DONTEXPAND | VM_PFNMAP | VM_MIXEDMAP) |
Nick Piggin | b291f00 | 2008-10-18 20:26:44 -0700 | [diff] [blame] | 179 | |
Alex Thorlton | a0715cc | 2014-04-07 15:37:10 -0700 | [diff] [blame] | 180 | /* This mask defines which mm->def_flags a process can inherit its parent */ |
| 181 | #define VM_INIT_DEF_MASK VM_NOHUGEPAGE |
| 182 | |
Nick Piggin | b291f00 | 2008-10-18 20:26:44 -0700 | [diff] [blame] | 183 | /* |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 184 | * mapping from the currently active vm_flags protection bits (the |
| 185 | * low four bits) to a page protection mask.. |
| 186 | */ |
| 187 | extern pgprot_t protection_map[16]; |
| 188 | |
Nick Piggin | d0217ac | 2007-07-19 01:47:03 -0700 | [diff] [blame] | 189 | #define FAULT_FLAG_WRITE 0x01 /* Fault was a write access */ |
| 190 | #define FAULT_FLAG_NONLINEAR 0x02 /* Fault was via a nonlinear mapping */ |
Nick Piggin | c2ec175 | 2009-03-31 15:23:21 -0700 | [diff] [blame] | 191 | #define FAULT_FLAG_MKWRITE 0x04 /* Fault was mkwrite of existing pte */ |
Michel Lespinasse | d065bd8 | 2010-10-26 14:21:57 -0700 | [diff] [blame] | 192 | #define FAULT_FLAG_ALLOW_RETRY 0x08 /* Retry fault if blocking */ |
Gleb Natapov | 318b275 | 2011-03-22 16:30:51 -0700 | [diff] [blame] | 193 | #define FAULT_FLAG_RETRY_NOWAIT 0x10 /* Don't drop mmap_sem and wait when retrying */ |
KOSAKI Motohiro | 37b23e0 | 2011-05-24 17:11:30 -0700 | [diff] [blame] | 194 | #define FAULT_FLAG_KILLABLE 0x20 /* The fault task is in SIGKILL killable region */ |
Shaohua Li | 45cac65 | 2012-10-08 16:32:19 -0700 | [diff] [blame] | 195 | #define FAULT_FLAG_TRIED 0x40 /* second try */ |
Johannes Weiner | 759496b | 2013-09-12 15:13:39 -0700 | [diff] [blame] | 196 | #define FAULT_FLAG_USER 0x80 /* The fault originated in userspace */ |
Nick Piggin | d0217ac | 2007-07-19 01:47:03 -0700 | [diff] [blame] | 197 | |
venkatesh.pallipadi@intel.com | 6bd9cd5 | 2008-12-19 13:47:26 -0800 | [diff] [blame] | 198 | /* |
Nick Piggin | d0217ac | 2007-07-19 01:47:03 -0700 | [diff] [blame] | 199 | * vm_fault is filled by the the pagefault handler and passed to the vma's |
Nick Piggin | 83c5407 | 2007-07-19 01:47:05 -0700 | [diff] [blame] | 200 | * ->fault function. The vma's ->fault is responsible for returning a bitmask |
| 201 | * of VM_FAULT_xxx flags that give details about how the fault was handled. |
Nick Piggin | 54cb882 | 2007-07-19 01:46:59 -0700 | [diff] [blame] | 202 | * |
Nick Piggin | d0217ac | 2007-07-19 01:47:03 -0700 | [diff] [blame] | 203 | * pgoff should be used in favour of virtual_address, if possible. If pgoff |
Konstantin Khlebnikov | 0b173bc | 2012-10-08 16:28:46 -0700 | [diff] [blame] | 204 | * is used, one may implement ->remap_pages to get nonlinear mapping support. |
Nick Piggin | 54cb882 | 2007-07-19 01:46:59 -0700 | [diff] [blame] | 205 | */ |
Nick Piggin | d0217ac | 2007-07-19 01:47:03 -0700 | [diff] [blame] | 206 | struct vm_fault { |
| 207 | unsigned int flags; /* FAULT_FLAG_xxx flags */ |
| 208 | pgoff_t pgoff; /* Logical page offset based on vma */ |
| 209 | void __user *virtual_address; /* Faulting virtual address */ |
| 210 | |
| 211 | struct page *page; /* ->fault handlers should return a |
Nick Piggin | 83c5407 | 2007-07-19 01:47:05 -0700 | [diff] [blame] | 212 | * page here, unless VM_FAULT_NOPAGE |
Nick Piggin | d0217ac | 2007-07-19 01:47:03 -0700 | [diff] [blame] | 213 | * is set (which is also implied by |
Nick Piggin | 83c5407 | 2007-07-19 01:47:05 -0700 | [diff] [blame] | 214 | * VM_FAULT_ERROR). |
Nick Piggin | d0217ac | 2007-07-19 01:47:03 -0700 | [diff] [blame] | 215 | */ |
Kirill A. Shutemov | 8c6e50b | 2014-04-07 15:37:18 -0700 | [diff] [blame] | 216 | /* for ->map_pages() only */ |
| 217 | pgoff_t max_pgoff; /* map pages for offset from pgoff till |
| 218 | * max_pgoff inclusive */ |
| 219 | pte_t *pte; /* pte entry associated with ->pgoff */ |
Nick Piggin | 54cb882 | 2007-07-19 01:46:59 -0700 | [diff] [blame] | 220 | }; |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 221 | |
| 222 | /* |
| 223 | * These are the virtual MM functions - opening of an area, closing and |
| 224 | * unmapping it (needed to keep files on disk up-to-date etc), pointer |
| 225 | * to the functions called when a no-page or a wp-page exception occurs. |
| 226 | */ |
| 227 | struct vm_operations_struct { |
| 228 | void (*open)(struct vm_area_struct * area); |
| 229 | void (*close)(struct vm_area_struct * area); |
Nick Piggin | d0217ac | 2007-07-19 01:47:03 -0700 | [diff] [blame] | 230 | int (*fault)(struct vm_area_struct *vma, struct vm_fault *vmf); |
Kirill A. Shutemov | 8c6e50b | 2014-04-07 15:37:18 -0700 | [diff] [blame] | 231 | void (*map_pages)(struct vm_area_struct *vma, struct vm_fault *vmf); |
David Howells | 9637a5e | 2006-06-23 02:03:43 -0700 | [diff] [blame] | 232 | |
| 233 | /* notification that a previously read-only page is about to become |
| 234 | * writable, if an error is returned it will cause a SIGBUS */ |
Nick Piggin | c2ec175 | 2009-03-31 15:23:21 -0700 | [diff] [blame] | 235 | int (*page_mkwrite)(struct vm_area_struct *vma, struct vm_fault *vmf); |
Rik van Riel | 28b2ee2 | 2008-07-23 21:27:05 -0700 | [diff] [blame] | 236 | |
| 237 | /* called by access_process_vm when get_user_pages() fails, typically |
| 238 | * for use by special VMAs that can switch between memory and hardware |
| 239 | */ |
| 240 | int (*access)(struct vm_area_struct *vma, unsigned long addr, |
| 241 | void *buf, int len, int write); |
Andy Lutomirski | 78d683e | 2014-05-19 15:58:32 -0700 | [diff] [blame] | 242 | |
| 243 | /* Called by the /proc/PID/maps code to ask the vma whether it |
| 244 | * has a special name. Returning non-NULL will also cause this |
| 245 | * vma to be dumped unconditionally. */ |
| 246 | const char *(*name)(struct vm_area_struct *vma); |
| 247 | |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 248 | #ifdef CONFIG_NUMA |
Lee Schermerhorn | a6020ed | 2008-04-28 02:13:14 -0700 | [diff] [blame] | 249 | /* |
| 250 | * set_policy() op must add a reference to any non-NULL @new mempolicy |
| 251 | * to hold the policy upon return. Caller should pass NULL @new to |
| 252 | * remove a policy and fall back to surrounding context--i.e. do not |
| 253 | * install a MPOL_DEFAULT policy, nor the task or system default |
| 254 | * mempolicy. |
| 255 | */ |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 256 | int (*set_policy)(struct vm_area_struct *vma, struct mempolicy *new); |
Lee Schermerhorn | a6020ed | 2008-04-28 02:13:14 -0700 | [diff] [blame] | 257 | |
| 258 | /* |
| 259 | * get_policy() op must add reference [mpol_get()] to any policy at |
| 260 | * (vma,addr) marked as MPOL_SHARED. The shared policy infrastructure |
| 261 | * in mm/mempolicy.c will do this automatically. |
| 262 | * get_policy() must NOT add a ref if the policy at (vma,addr) is not |
| 263 | * marked as MPOL_SHARED. vma policies are protected by the mmap_sem. |
| 264 | * If no [shared/vma] mempolicy exists at the addr, get_policy() op |
| 265 | * must return NULL--i.e., do not "fallback" to task or system default |
| 266 | * policy. |
| 267 | */ |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 268 | struct mempolicy *(*get_policy)(struct vm_area_struct *vma, |
| 269 | unsigned long addr); |
Christoph Lameter | 7b2259b | 2006-06-25 05:46:48 -0700 | [diff] [blame] | 270 | int (*migrate)(struct vm_area_struct *vma, const nodemask_t *from, |
| 271 | const nodemask_t *to, unsigned long flags); |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 272 | #endif |
Konstantin Khlebnikov | 0b173bc | 2012-10-08 16:28:46 -0700 | [diff] [blame] | 273 | /* called by sys_remap_file_pages() to populate non-linear mapping */ |
| 274 | int (*remap_pages)(struct vm_area_struct *vma, unsigned long addr, |
| 275 | unsigned long size, pgoff_t pgoff); |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 276 | }; |
| 277 | |
| 278 | struct mmu_gather; |
| 279 | struct inode; |
| 280 | |
Andrew Morton | 349aef0 | 2006-01-08 01:04:36 -0800 | [diff] [blame] | 281 | #define page_private(page) ((page)->private) |
| 282 | #define set_page_private(page, v) ((page)->private = (v)) |
Hugh Dickins | 4c21e2f | 2005-10-29 18:16:40 -0700 | [diff] [blame] | 283 | |
Minchan Kim | b12c4ad | 2012-10-08 16:32:08 -0700 | [diff] [blame] | 284 | /* It's valid only if the page is free path or free_list */ |
| 285 | static inline void set_freepage_migratetype(struct page *page, int migratetype) |
| 286 | { |
Minchan Kim | 95e3441 | 2012-10-08 16:32:11 -0700 | [diff] [blame] | 287 | page->index = migratetype; |
Minchan Kim | b12c4ad | 2012-10-08 16:32:08 -0700 | [diff] [blame] | 288 | } |
| 289 | |
| 290 | /* It's valid only if the page is free path or free_list */ |
| 291 | static inline int get_freepage_migratetype(struct page *page) |
| 292 | { |
Minchan Kim | 95e3441 | 2012-10-08 16:32:11 -0700 | [diff] [blame] | 293 | return page->index; |
Minchan Kim | b12c4ad | 2012-10-08 16:32:08 -0700 | [diff] [blame] | 294 | } |
| 295 | |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 296 | /* |
| 297 | * FIXME: take this include out, include page-flags.h in |
| 298 | * files which need it (119 of them) |
| 299 | */ |
| 300 | #include <linux/page-flags.h> |
Andrea Arcangeli | 71e3aac | 2011-01-13 15:46:52 -0800 | [diff] [blame] | 301 | #include <linux/huge_mm.h> |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 302 | |
| 303 | /* |
| 304 | * Methods to modify the page usage count. |
| 305 | * |
| 306 | * What counts for a page usage: |
| 307 | * - cache mapping (page->mapping) |
| 308 | * - private data (page->private) |
| 309 | * - page mapped in a task's page tables, each mapping |
| 310 | * is counted separately |
| 311 | * |
| 312 | * Also, many kernel routines increase the page count before a critical |
| 313 | * routine so they can be sure the page doesn't go away from under them. |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 314 | */ |
| 315 | |
| 316 | /* |
Nick Piggin | da6052f | 2006-09-25 23:31:35 -0700 | [diff] [blame] | 317 | * Drop a ref, return true if the refcount fell to zero (the page has no users) |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 318 | */ |
Nick Piggin | 7c8ee9a | 2006-03-22 00:08:03 -0800 | [diff] [blame] | 319 | static inline int put_page_testzero(struct page *page) |
| 320 | { |
Sasha Levin | 309381fea | 2014-01-23 15:52:54 -0800 | [diff] [blame] | 321 | VM_BUG_ON_PAGE(atomic_read(&page->_count) == 0, page); |
Nick Piggin | 8dc04ef | 2006-03-22 00:08:03 -0800 | [diff] [blame] | 322 | return atomic_dec_and_test(&page->_count); |
Nick Piggin | 7c8ee9a | 2006-03-22 00:08:03 -0800 | [diff] [blame] | 323 | } |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 324 | |
| 325 | /* |
Nick Piggin | 7c8ee9a | 2006-03-22 00:08:03 -0800 | [diff] [blame] | 326 | * Try to grab a ref unless the page has a refcount of zero, return false if |
| 327 | * that is the case. |
Alexey Kardashevskiy | 8e0861f | 2013-08-28 18:37:42 +1000 | [diff] [blame] | 328 | * This can be called when MMU is off so it must not access |
| 329 | * any of the virtual mappings. |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 330 | */ |
Nick Piggin | 7c8ee9a | 2006-03-22 00:08:03 -0800 | [diff] [blame] | 331 | static inline int get_page_unless_zero(struct page *page) |
| 332 | { |
Nick Piggin | 8dc04ef | 2006-03-22 00:08:03 -0800 | [diff] [blame] | 333 | return atomic_inc_not_zero(&page->_count); |
Nick Piggin | 7c8ee9a | 2006-03-22 00:08:03 -0800 | [diff] [blame] | 334 | } |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 335 | |
Alexey Kardashevskiy | 8e0861f | 2013-08-28 18:37:42 +1000 | [diff] [blame] | 336 | /* |
| 337 | * Try to drop a ref unless the page has a refcount of one, return false if |
| 338 | * that is the case. |
| 339 | * This is to make sure that the refcount won't become zero after this drop. |
| 340 | * This can be called when MMU is off so it must not access |
| 341 | * any of the virtual mappings. |
| 342 | */ |
| 343 | static inline int put_page_unless_one(struct page *page) |
| 344 | { |
| 345 | return atomic_add_unless(&page->_count, -1, 1); |
| 346 | } |
| 347 | |
Wu Fengguang | 53df8fd | 2010-01-27 11:06:39 +0800 | [diff] [blame] | 348 | extern int page_is_ram(unsigned long pfn); |
| 349 | |
Christoph Lameter | 48667e7 | 2008-02-04 22:28:31 -0800 | [diff] [blame] | 350 | /* Support for virtually mapped pages */ |
Christoph Lameter | b3bdda0 | 2008-02-04 22:28:32 -0800 | [diff] [blame] | 351 | struct page *vmalloc_to_page(const void *addr); |
| 352 | unsigned long vmalloc_to_pfn(const void *addr); |
Christoph Lameter | 48667e7 | 2008-02-04 22:28:31 -0800 | [diff] [blame] | 353 | |
Paul Mundt | 0738c4b | 2008-03-12 16:51:31 +0900 | [diff] [blame] | 354 | /* |
| 355 | * Determine if an address is within the vmalloc range |
| 356 | * |
| 357 | * On nommu, vmalloc/vfree wrap through kmalloc/kfree directly, so there |
| 358 | * is no special casing required. |
| 359 | */ |
Christoph Lameter | 9e2779f | 2008-02-04 22:28:34 -0800 | [diff] [blame] | 360 | static inline int is_vmalloc_addr(const void *x) |
| 361 | { |
Paul Mundt | 0738c4b | 2008-03-12 16:51:31 +0900 | [diff] [blame] | 362 | #ifdef CONFIG_MMU |
Christoph Lameter | 9e2779f | 2008-02-04 22:28:34 -0800 | [diff] [blame] | 363 | unsigned long addr = (unsigned long)x; |
| 364 | |
| 365 | return addr >= VMALLOC_START && addr < VMALLOC_END; |
Paul Mundt | 0738c4b | 2008-03-12 16:51:31 +0900 | [diff] [blame] | 366 | #else |
| 367 | return 0; |
David Howells | 8ca3ed8 | 2008-02-23 15:23:37 -0800 | [diff] [blame] | 368 | #endif |
Paul Mundt | 0738c4b | 2008-03-12 16:51:31 +0900 | [diff] [blame] | 369 | } |
KAMEZAWA Hiroyuki | 81ac3ad | 2009-09-22 16:45:49 -0700 | [diff] [blame] | 370 | #ifdef CONFIG_MMU |
| 371 | extern int is_vmalloc_or_module_addr(const void *x); |
| 372 | #else |
David Howells | 934831d | 2009-09-24 12:33:32 +0100 | [diff] [blame] | 373 | static inline int is_vmalloc_or_module_addr(const void *x) |
KAMEZAWA Hiroyuki | 81ac3ad | 2009-09-22 16:45:49 -0700 | [diff] [blame] | 374 | { |
| 375 | return 0; |
| 376 | } |
| 377 | #endif |
Christoph Lameter | 9e2779f | 2008-02-04 22:28:34 -0800 | [diff] [blame] | 378 | |
Al Viro | 39f1f78 | 2014-05-06 14:02:53 -0400 | [diff] [blame] | 379 | extern void kvfree(const void *addr); |
| 380 | |
Andrea Arcangeli | e9da73d | 2011-01-13 15:46:32 -0800 | [diff] [blame] | 381 | static inline void compound_lock(struct page *page) |
| 382 | { |
| 383 | #ifdef CONFIG_TRANSPARENT_HUGEPAGE |
Sasha Levin | 309381fea | 2014-01-23 15:52:54 -0800 | [diff] [blame] | 384 | VM_BUG_ON_PAGE(PageSlab(page), page); |
Andrea Arcangeli | e9da73d | 2011-01-13 15:46:32 -0800 | [diff] [blame] | 385 | bit_spin_lock(PG_compound_lock, &page->flags); |
| 386 | #endif |
| 387 | } |
| 388 | |
| 389 | static inline void compound_unlock(struct page *page) |
| 390 | { |
| 391 | #ifdef CONFIG_TRANSPARENT_HUGEPAGE |
Sasha Levin | 309381fea | 2014-01-23 15:52:54 -0800 | [diff] [blame] | 392 | VM_BUG_ON_PAGE(PageSlab(page), page); |
Andrea Arcangeli | e9da73d | 2011-01-13 15:46:32 -0800 | [diff] [blame] | 393 | bit_spin_unlock(PG_compound_lock, &page->flags); |
| 394 | #endif |
| 395 | } |
| 396 | |
| 397 | static inline unsigned long compound_lock_irqsave(struct page *page) |
| 398 | { |
| 399 | unsigned long uninitialized_var(flags); |
| 400 | #ifdef CONFIG_TRANSPARENT_HUGEPAGE |
| 401 | local_irq_save(flags); |
| 402 | compound_lock(page); |
| 403 | #endif |
| 404 | return flags; |
| 405 | } |
| 406 | |
| 407 | static inline void compound_unlock_irqrestore(struct page *page, |
| 408 | unsigned long flags) |
| 409 | { |
| 410 | #ifdef CONFIG_TRANSPARENT_HUGEPAGE |
| 411 | compound_unlock(page); |
| 412 | local_irq_restore(flags); |
| 413 | #endif |
| 414 | } |
| 415 | |
Jianyu Zhan | d2ee40e | 2014-06-04 16:08:02 -0700 | [diff] [blame] | 416 | static inline struct page *compound_head_by_tail(struct page *tail) |
| 417 | { |
| 418 | struct page *head = tail->first_page; |
| 419 | |
| 420 | /* |
| 421 | * page->first_page may be a dangling pointer to an old |
| 422 | * compound page, so recheck that it is still a tail |
| 423 | * page before returning. |
| 424 | */ |
| 425 | smp_rmb(); |
| 426 | if (likely(PageTail(tail))) |
| 427 | return head; |
| 428 | return tail; |
| 429 | } |
| 430 | |
Christoph Lameter | d85f338 | 2007-05-06 14:49:39 -0700 | [diff] [blame] | 431 | static inline struct page *compound_head(struct page *page) |
| 432 | { |
Jianyu Zhan | d2ee40e | 2014-06-04 16:08:02 -0700 | [diff] [blame] | 433 | if (unlikely(PageTail(page))) |
| 434 | return compound_head_by_tail(page); |
Christoph Lameter | d85f338 | 2007-05-06 14:49:39 -0700 | [diff] [blame] | 435 | return page; |
| 436 | } |
| 437 | |
Andrea Arcangeli | 70b50f9 | 2011-11-02 13:36:59 -0700 | [diff] [blame] | 438 | /* |
| 439 | * The atomic page->_mapcount, starts from -1: so that transitions |
| 440 | * both from it and to it can be tracked, using atomic_inc_and_test |
| 441 | * and atomic_add_negative(-1). |
| 442 | */ |
Mel Gorman | 22b751c | 2013-02-22 16:34:59 -0800 | [diff] [blame] | 443 | static inline void page_mapcount_reset(struct page *page) |
Andrea Arcangeli | 70b50f9 | 2011-11-02 13:36:59 -0700 | [diff] [blame] | 444 | { |
| 445 | atomic_set(&(page)->_mapcount, -1); |
| 446 | } |
| 447 | |
| 448 | static inline int page_mapcount(struct page *page) |
| 449 | { |
| 450 | return atomic_read(&(page)->_mapcount) + 1; |
| 451 | } |
| 452 | |
Hugh Dickins | 4c21e2f | 2005-10-29 18:16:40 -0700 | [diff] [blame] | 453 | static inline int page_count(struct page *page) |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 454 | { |
Christoph Lameter | d85f338 | 2007-05-06 14:49:39 -0700 | [diff] [blame] | 455 | return atomic_read(&compound_head(page)->_count); |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 456 | } |
| 457 | |
Andrea Arcangeli | 44518d2 | 2014-01-21 15:48:54 -0800 | [diff] [blame] | 458 | #ifdef CONFIG_HUGETLB_PAGE |
| 459 | extern int PageHeadHuge(struct page *page_head); |
| 460 | #else /* CONFIG_HUGETLB_PAGE */ |
| 461 | static inline int PageHeadHuge(struct page *page_head) |
| 462 | { |
| 463 | return 0; |
| 464 | } |
| 465 | #endif /* CONFIG_HUGETLB_PAGE */ |
| 466 | |
| 467 | static inline bool __compound_tail_refcounted(struct page *page) |
| 468 | { |
| 469 | return !PageSlab(page) && !PageHeadHuge(page); |
| 470 | } |
| 471 | |
| 472 | /* |
| 473 | * This takes a head page as parameter and tells if the |
| 474 | * tail page reference counting can be skipped. |
| 475 | * |
| 476 | * For this to be safe, PageSlab and PageHeadHuge must remain true on |
| 477 | * any given page where they return true here, until all tail pins |
| 478 | * have been released. |
| 479 | */ |
| 480 | static inline bool compound_tail_refcounted(struct page *page) |
| 481 | { |
Sasha Levin | 309381fea | 2014-01-23 15:52:54 -0800 | [diff] [blame] | 482 | VM_BUG_ON_PAGE(!PageHead(page), page); |
Andrea Arcangeli | 44518d2 | 2014-01-21 15:48:54 -0800 | [diff] [blame] | 483 | return __compound_tail_refcounted(page); |
| 484 | } |
| 485 | |
Andrea Arcangeli | b35a35b | 2011-11-02 13:37:36 -0700 | [diff] [blame] | 486 | static inline void get_huge_page_tail(struct page *page) |
| 487 | { |
| 488 | /* |
Oleg Nesterov | 5eaf1a9 | 2014-01-21 15:49:04 -0800 | [diff] [blame] | 489 | * __split_huge_page_refcount() cannot run from under us. |
Andrea Arcangeli | b35a35b | 2011-11-02 13:37:36 -0700 | [diff] [blame] | 490 | */ |
Sasha Levin | 309381fea | 2014-01-23 15:52:54 -0800 | [diff] [blame] | 491 | VM_BUG_ON_PAGE(!PageTail(page), page); |
| 492 | VM_BUG_ON_PAGE(page_mapcount(page) < 0, page); |
| 493 | VM_BUG_ON_PAGE(atomic_read(&page->_count) != 0, page); |
Oleg Nesterov | 5eaf1a9 | 2014-01-21 15:49:04 -0800 | [diff] [blame] | 494 | if (compound_tail_refcounted(page->first_page)) |
Andrea Arcangeli | 44518d2 | 2014-01-21 15:48:54 -0800 | [diff] [blame] | 495 | atomic_inc(&page->_mapcount); |
Andrea Arcangeli | b35a35b | 2011-11-02 13:37:36 -0700 | [diff] [blame] | 496 | } |
| 497 | |
Andrea Arcangeli | 70b50f9 | 2011-11-02 13:36:59 -0700 | [diff] [blame] | 498 | extern bool __get_page_tail(struct page *page); |
| 499 | |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 500 | static inline void get_page(struct page *page) |
| 501 | { |
Andrea Arcangeli | 70b50f9 | 2011-11-02 13:36:59 -0700 | [diff] [blame] | 502 | if (unlikely(PageTail(page))) |
| 503 | if (likely(__get_page_tail(page))) |
| 504 | return; |
Andrea Arcangeli | 9180706 | 2011-01-13 15:46:32 -0800 | [diff] [blame] | 505 | /* |
| 506 | * Getting a normal page or the head of a compound page |
Andrea Arcangeli | 70b50f9 | 2011-11-02 13:36:59 -0700 | [diff] [blame] | 507 | * requires to already have an elevated page->_count. |
Andrea Arcangeli | 9180706 | 2011-01-13 15:46:32 -0800 | [diff] [blame] | 508 | */ |
Sasha Levin | 309381fea | 2014-01-23 15:52:54 -0800 | [diff] [blame] | 509 | VM_BUG_ON_PAGE(atomic_read(&page->_count) <= 0, page); |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 510 | atomic_inc(&page->_count); |
| 511 | } |
| 512 | |
Christoph Lameter | b49af68 | 2007-05-06 14:49:41 -0700 | [diff] [blame] | 513 | static inline struct page *virt_to_head_page(const void *x) |
| 514 | { |
| 515 | struct page *page = virt_to_page(x); |
| 516 | return compound_head(page); |
| 517 | } |
| 518 | |
Nick Piggin | 7835e98 | 2006-03-22 00:08:40 -0800 | [diff] [blame] | 519 | /* |
| 520 | * Setup the page count before being freed into the page allocator for |
| 521 | * the first time (boot or memory hotplug) |
| 522 | */ |
| 523 | static inline void init_page_count(struct page *page) |
| 524 | { |
| 525 | atomic_set(&page->_count, 1); |
| 526 | } |
| 527 | |
Andrea Arcangeli | 5f24ce5 | 2011-01-13 15:47:00 -0800 | [diff] [blame] | 528 | /* |
| 529 | * PageBuddy() indicate that the page is free and in the buddy system |
| 530 | * (see mm/page_alloc.c). |
Andrea Arcangeli | ef2b4b9 | 2011-03-18 00:16:35 +0100 | [diff] [blame] | 531 | * |
| 532 | * PAGE_BUDDY_MAPCOUNT_VALUE must be <= -2 but better not too close to |
| 533 | * -2 so that an underflow of the page_mapcount() won't be mistaken |
| 534 | * for a genuine PAGE_BUDDY_MAPCOUNT_VALUE. -128 can be created very |
| 535 | * efficiently by most CPU architectures. |
Andrea Arcangeli | 5f24ce5 | 2011-01-13 15:47:00 -0800 | [diff] [blame] | 536 | */ |
Andrea Arcangeli | ef2b4b9 | 2011-03-18 00:16:35 +0100 | [diff] [blame] | 537 | #define PAGE_BUDDY_MAPCOUNT_VALUE (-128) |
| 538 | |
Andrea Arcangeli | 5f24ce5 | 2011-01-13 15:47:00 -0800 | [diff] [blame] | 539 | static inline int PageBuddy(struct page *page) |
| 540 | { |
Andrea Arcangeli | ef2b4b9 | 2011-03-18 00:16:35 +0100 | [diff] [blame] | 541 | return atomic_read(&page->_mapcount) == PAGE_BUDDY_MAPCOUNT_VALUE; |
Andrea Arcangeli | 5f24ce5 | 2011-01-13 15:47:00 -0800 | [diff] [blame] | 542 | } |
| 543 | |
| 544 | static inline void __SetPageBuddy(struct page *page) |
| 545 | { |
Sasha Levin | 309381fea | 2014-01-23 15:52:54 -0800 | [diff] [blame] | 546 | VM_BUG_ON_PAGE(atomic_read(&page->_mapcount) != -1, page); |
Andrea Arcangeli | ef2b4b9 | 2011-03-18 00:16:35 +0100 | [diff] [blame] | 547 | atomic_set(&page->_mapcount, PAGE_BUDDY_MAPCOUNT_VALUE); |
Andrea Arcangeli | 5f24ce5 | 2011-01-13 15:47:00 -0800 | [diff] [blame] | 548 | } |
| 549 | |
| 550 | static inline void __ClearPageBuddy(struct page *page) |
| 551 | { |
Sasha Levin | 309381fea | 2014-01-23 15:52:54 -0800 | [diff] [blame] | 552 | VM_BUG_ON_PAGE(!PageBuddy(page), page); |
Andrea Arcangeli | 5f24ce5 | 2011-01-13 15:47:00 -0800 | [diff] [blame] | 553 | atomic_set(&page->_mapcount, -1); |
| 554 | } |
| 555 | |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 556 | void put_page(struct page *page); |
Alexander Zarochentsev | 1d7ea73 | 2006-08-13 23:24:27 -0700 | [diff] [blame] | 557 | void put_pages_list(struct list_head *pages); |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 558 | |
Nick Piggin | 8dfcc9b | 2006-03-22 00:08:05 -0800 | [diff] [blame] | 559 | void split_page(struct page *page, unsigned int order); |
Mel Gorman | 748446b | 2010-05-24 14:32:27 -0700 | [diff] [blame] | 560 | int split_free_page(struct page *page); |
Nick Piggin | 8dfcc9b | 2006-03-22 00:08:05 -0800 | [diff] [blame] | 561 | |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 562 | /* |
Andy Whitcroft | 33f2ef8 | 2006-12-06 20:33:32 -0800 | [diff] [blame] | 563 | * Compound pages have a destructor function. Provide a |
| 564 | * prototype for that function and accessor functions. |
| 565 | * These are _only_ valid on the head of a PG_compound page. |
| 566 | */ |
| 567 | typedef void compound_page_dtor(struct page *); |
| 568 | |
| 569 | static inline void set_compound_page_dtor(struct page *page, |
| 570 | compound_page_dtor *dtor) |
| 571 | { |
| 572 | page[1].lru.next = (void *)dtor; |
| 573 | } |
| 574 | |
| 575 | static inline compound_page_dtor *get_compound_page_dtor(struct page *page) |
| 576 | { |
| 577 | return (compound_page_dtor *)page[1].lru.next; |
| 578 | } |
| 579 | |
Christoph Lameter | d85f338 | 2007-05-06 14:49:39 -0700 | [diff] [blame] | 580 | static inline int compound_order(struct page *page) |
| 581 | { |
Christoph Lameter | 6d77795 | 2007-05-06 14:49:40 -0700 | [diff] [blame] | 582 | if (!PageHead(page)) |
Christoph Lameter | d85f338 | 2007-05-06 14:49:39 -0700 | [diff] [blame] | 583 | return 0; |
| 584 | return (unsigned long)page[1].lru.prev; |
| 585 | } |
| 586 | |
| 587 | static inline void set_compound_order(struct page *page, unsigned long order) |
| 588 | { |
| 589 | page[1].lru.prev = (void *)order; |
| 590 | } |
| 591 | |
Michal Simek | 3dece37 | 2011-01-21 08:49:56 +0100 | [diff] [blame] | 592 | #ifdef CONFIG_MMU |
Andy Whitcroft | 33f2ef8 | 2006-12-06 20:33:32 -0800 | [diff] [blame] | 593 | /* |
Andrea Arcangeli | 14fd403 | 2011-01-13 15:46:37 -0800 | [diff] [blame] | 594 | * Do pte_mkwrite, but only if the vma says VM_WRITE. We do this when |
| 595 | * servicing faults for write access. In the normal case, do always want |
| 596 | * pte_mkwrite. But get_user_pages can cause write faults for mappings |
| 597 | * that do not have writing enabled, when used by access_process_vm. |
| 598 | */ |
| 599 | static inline pte_t maybe_mkwrite(pte_t pte, struct vm_area_struct *vma) |
| 600 | { |
| 601 | if (likely(vma->vm_flags & VM_WRITE)) |
| 602 | pte = pte_mkwrite(pte); |
| 603 | return pte; |
| 604 | } |
Kirill A. Shutemov | 8c6e50b | 2014-04-07 15:37:18 -0700 | [diff] [blame] | 605 | |
| 606 | void do_set_pte(struct vm_area_struct *vma, unsigned long address, |
| 607 | struct page *page, pte_t *pte, bool write, bool anon); |
Michal Simek | 3dece37 | 2011-01-21 08:49:56 +0100 | [diff] [blame] | 608 | #endif |
Andrea Arcangeli | 14fd403 | 2011-01-13 15:46:37 -0800 | [diff] [blame] | 609 | |
| 610 | /* |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 611 | * Multiple processes may "see" the same page. E.g. for untouched |
| 612 | * mappings of /dev/null, all processes see the same page full of |
| 613 | * zeroes, and text pages of executables and shared libraries have |
| 614 | * only one copy in memory, at most, normally. |
| 615 | * |
| 616 | * For the non-reserved pages, page_count(page) denotes a reference count. |
Paolo 'Blaisorblade' Giarrusso | 7e871b6 | 2005-09-21 09:55:38 -0700 | [diff] [blame] | 617 | * page_count() == 0 means the page is free. page->lru is then used for |
| 618 | * freelist management in the buddy allocator. |
Nick Piggin | da6052f | 2006-09-25 23:31:35 -0700 | [diff] [blame] | 619 | * page_count() > 0 means the page has been allocated. |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 620 | * |
Nick Piggin | da6052f | 2006-09-25 23:31:35 -0700 | [diff] [blame] | 621 | * Pages are allocated by the slab allocator in order to provide memory |
| 622 | * to kmalloc and kmem_cache_alloc. In this case, the management of the |
| 623 | * page, and the fields in 'struct page' are the responsibility of mm/slab.c |
| 624 | * unless a particular usage is carefully commented. (the responsibility of |
| 625 | * freeing the kmalloc memory is the caller's, of course). |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 626 | * |
Nick Piggin | da6052f | 2006-09-25 23:31:35 -0700 | [diff] [blame] | 627 | * A page may be used by anyone else who does a __get_free_page(). |
| 628 | * In this case, page_count still tracks the references, and should only |
| 629 | * be used through the normal accessor functions. The top bits of page->flags |
| 630 | * and page->virtual store page management information, but all other fields |
| 631 | * are unused and could be used privately, carefully. The management of this |
| 632 | * page is the responsibility of the one who allocated it, and those who have |
| 633 | * subsequently been given references to it. |
| 634 | * |
| 635 | * The other pages (we may call them "pagecache pages") are completely |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 636 | * managed by the Linux memory manager: I/O, buffers, swapping etc. |
| 637 | * The following discussion applies only to them. |
| 638 | * |
Nick Piggin | da6052f | 2006-09-25 23:31:35 -0700 | [diff] [blame] | 639 | * A pagecache page contains an opaque `private' member, which belongs to the |
| 640 | * page's address_space. Usually, this is the address of a circular list of |
| 641 | * the page's disk buffers. PG_private must be set to tell the VM to call |
| 642 | * into the filesystem to release these pages. |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 643 | * |
Nick Piggin | da6052f | 2006-09-25 23:31:35 -0700 | [diff] [blame] | 644 | * A page may belong to an inode's memory mapping. In this case, page->mapping |
| 645 | * is the pointer to the inode, and page->index is the file offset of the page, |
| 646 | * in units of PAGE_CACHE_SIZE. |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 647 | * |
Nick Piggin | da6052f | 2006-09-25 23:31:35 -0700 | [diff] [blame] | 648 | * If pagecache pages are not associated with an inode, they are said to be |
| 649 | * anonymous pages. These may become associated with the swapcache, and in that |
| 650 | * case PG_swapcache is set, and page->private is an offset into the swapcache. |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 651 | * |
Nick Piggin | da6052f | 2006-09-25 23:31:35 -0700 | [diff] [blame] | 652 | * In either case (swapcache or inode backed), the pagecache itself holds one |
| 653 | * reference to the page. Setting PG_private should also increment the |
| 654 | * refcount. The each user mapping also has a reference to the page. |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 655 | * |
Nick Piggin | da6052f | 2006-09-25 23:31:35 -0700 | [diff] [blame] | 656 | * The pagecache pages are stored in a per-mapping radix tree, which is |
| 657 | * rooted at mapping->page_tree, and indexed by offset. |
| 658 | * Where 2.4 and early 2.6 kernels kept dirty/clean pages in per-address_space |
| 659 | * lists, we instead now tag pages as dirty/writeback in the radix tree. |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 660 | * |
Nick Piggin | da6052f | 2006-09-25 23:31:35 -0700 | [diff] [blame] | 661 | * All pagecache pages may be subject to I/O: |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 662 | * - inode pages may need to be read from disk, |
| 663 | * - inode pages which have been modified and are MAP_SHARED may need |
Nick Piggin | da6052f | 2006-09-25 23:31:35 -0700 | [diff] [blame] | 664 | * to be written back to the inode on disk, |
| 665 | * - anonymous pages (including MAP_PRIVATE file mappings) which have been |
| 666 | * modified may need to be swapped out to swap space and (later) to be read |
| 667 | * back into memory. |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 668 | */ |
| 669 | |
| 670 | /* |
| 671 | * The zone field is never updated after free_area_init_core() |
| 672 | * sets it, so none of the operations on it need to be atomic. |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 673 | */ |
Dave Hansen | 348f8b6 | 2005-06-23 00:07:40 -0700 | [diff] [blame] | 674 | |
Peter Zijlstra | 9057289 | 2013-10-07 11:29:20 +0100 | [diff] [blame] | 675 | /* Page flags: | [SECTION] | [NODE] | ZONE | [LAST_CPUPID] | ... | FLAGS | */ |
Andi Kleen | 07808b7 | 2005-11-05 17:25:53 +0100 | [diff] [blame] | 676 | #define SECTIONS_PGOFF ((sizeof(unsigned long)*8) - SECTIONS_WIDTH) |
Andy Whitcroft | d41dee3 | 2005-06-23 00:07:54 -0700 | [diff] [blame] | 677 | #define NODES_PGOFF (SECTIONS_PGOFF - NODES_WIDTH) |
| 678 | #define ZONES_PGOFF (NODES_PGOFF - ZONES_WIDTH) |
Peter Zijlstra | 9057289 | 2013-10-07 11:29:20 +0100 | [diff] [blame] | 679 | #define LAST_CPUPID_PGOFF (ZONES_PGOFF - LAST_CPUPID_WIDTH) |
Andy Whitcroft | d41dee3 | 2005-06-23 00:07:54 -0700 | [diff] [blame] | 680 | |
| 681 | /* |
Lucas De Marchi | 25985ed | 2011-03-30 22:57:33 -0300 | [diff] [blame] | 682 | * Define the bit shifts to access each section. For non-existent |
Dave Hansen | 348f8b6 | 2005-06-23 00:07:40 -0700 | [diff] [blame] | 683 | * sections we define the shift as 0; that plus a 0 mask ensures |
| 684 | * the compiler will optimise away reference to them. |
| 685 | */ |
Andy Whitcroft | d41dee3 | 2005-06-23 00:07:54 -0700 | [diff] [blame] | 686 | #define SECTIONS_PGSHIFT (SECTIONS_PGOFF * (SECTIONS_WIDTH != 0)) |
| 687 | #define NODES_PGSHIFT (NODES_PGOFF * (NODES_WIDTH != 0)) |
| 688 | #define ZONES_PGSHIFT (ZONES_PGOFF * (ZONES_WIDTH != 0)) |
Peter Zijlstra | 9057289 | 2013-10-07 11:29:20 +0100 | [diff] [blame] | 689 | #define LAST_CPUPID_PGSHIFT (LAST_CPUPID_PGOFF * (LAST_CPUPID_WIDTH != 0)) |
Dave Hansen | 348f8b6 | 2005-06-23 00:07:40 -0700 | [diff] [blame] | 690 | |
Will Deacon | bce54bb | 2010-10-26 14:21:37 -0700 | [diff] [blame] | 691 | /* NODE:ZONE or SECTION:ZONE is used to ID a zone for the buddy allocator */ |
| 692 | #ifdef NODE_NOT_IN_PAGE_FLAGS |
Christoph Lameter | 89689ae | 2006-12-06 20:31:45 -0800 | [diff] [blame] | 693 | #define ZONEID_SHIFT (SECTIONS_SHIFT + ZONES_SHIFT) |
Andy Whitcroft | bd8029b | 2007-02-10 01:43:14 -0800 | [diff] [blame] | 694 | #define ZONEID_PGOFF ((SECTIONS_PGOFF < ZONES_PGOFF)? \ |
| 695 | SECTIONS_PGOFF : ZONES_PGOFF) |
Andy Whitcroft | d41dee3 | 2005-06-23 00:07:54 -0700 | [diff] [blame] | 696 | #else |
Christoph Lameter | 89689ae | 2006-12-06 20:31:45 -0800 | [diff] [blame] | 697 | #define ZONEID_SHIFT (NODES_SHIFT + ZONES_SHIFT) |
Andy Whitcroft | bd8029b | 2007-02-10 01:43:14 -0800 | [diff] [blame] | 698 | #define ZONEID_PGOFF ((NODES_PGOFF < ZONES_PGOFF)? \ |
| 699 | NODES_PGOFF : ZONES_PGOFF) |
Andy Whitcroft | d41dee3 | 2005-06-23 00:07:54 -0700 | [diff] [blame] | 700 | #endif |
Christoph Lameter | 89689ae | 2006-12-06 20:31:45 -0800 | [diff] [blame] | 701 | |
Andy Whitcroft | bd8029b | 2007-02-10 01:43:14 -0800 | [diff] [blame] | 702 | #define ZONEID_PGSHIFT (ZONEID_PGOFF * (ZONEID_SHIFT != 0)) |
Dave Hansen | 348f8b6 | 2005-06-23 00:07:40 -0700 | [diff] [blame] | 703 | |
Christoph Lameter | 9223b41 | 2008-04-28 02:12:48 -0700 | [diff] [blame] | 704 | #if SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > BITS_PER_LONG - NR_PAGEFLAGS |
| 705 | #error SECTIONS_WIDTH+NODES_WIDTH+ZONES_WIDTH > BITS_PER_LONG - NR_PAGEFLAGS |
Dave Hansen | 348f8b6 | 2005-06-23 00:07:40 -0700 | [diff] [blame] | 706 | #endif |
| 707 | |
Andy Whitcroft | d41dee3 | 2005-06-23 00:07:54 -0700 | [diff] [blame] | 708 | #define ZONES_MASK ((1UL << ZONES_WIDTH) - 1) |
| 709 | #define NODES_MASK ((1UL << NODES_WIDTH) - 1) |
| 710 | #define SECTIONS_MASK ((1UL << SECTIONS_WIDTH) - 1) |
Srikar Dronamraju | 834a964 | 2014-04-07 15:37:57 -0700 | [diff] [blame] | 711 | #define LAST_CPUPID_MASK ((1UL << LAST_CPUPID_SHIFT) - 1) |
Christoph Lameter | 89689ae | 2006-12-06 20:31:45 -0800 | [diff] [blame] | 712 | #define ZONEID_MASK ((1UL << ZONEID_SHIFT) - 1) |
Dave Hansen | 348f8b6 | 2005-06-23 00:07:40 -0700 | [diff] [blame] | 713 | |
Ian Campbell | 33dd4e0 | 2011-07-25 17:11:51 -0700 | [diff] [blame] | 714 | static inline enum zone_type page_zonenum(const struct page *page) |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 715 | { |
Dave Hansen | 348f8b6 | 2005-06-23 00:07:40 -0700 | [diff] [blame] | 716 | return (page->flags >> ZONES_PGSHIFT) & ZONES_MASK; |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 717 | } |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 718 | |
Cody P Schafer | 9127ab4 | 2013-02-22 16:35:21 -0800 | [diff] [blame] | 719 | #if defined(CONFIG_SPARSEMEM) && !defined(CONFIG_SPARSEMEM_VMEMMAP) |
| 720 | #define SECTION_IN_PAGE_FLAGS |
| 721 | #endif |
| 722 | |
Christoph Lameter | 89689ae | 2006-12-06 20:31:45 -0800 | [diff] [blame] | 723 | /* |
Vlastimil Babka | 7a8010c | 2013-09-11 14:22:35 -0700 | [diff] [blame] | 724 | * The identification function is mainly used by the buddy allocator for |
| 725 | * determining if two pages could be buddies. We are not really identifying |
| 726 | * the zone since we could be using the section number id if we do not have |
| 727 | * node id available in page flags. |
| 728 | * We only guarantee that it will return the same value for two combinable |
| 729 | * pages in a zone. |
Christoph Lameter | 89689ae | 2006-12-06 20:31:45 -0800 | [diff] [blame] | 730 | */ |
Andy Whitcroft | cb2b95e | 2006-06-23 02:03:01 -0700 | [diff] [blame] | 731 | static inline int page_zone_id(struct page *page) |
| 732 | { |
Christoph Lameter | 89689ae | 2006-12-06 20:31:45 -0800 | [diff] [blame] | 733 | return (page->flags >> ZONEID_PGSHIFT) & ZONEID_MASK; |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 734 | } |
| 735 | |
Andy Whitcroft | 25ba77c | 2006-12-06 20:33:03 -0800 | [diff] [blame] | 736 | static inline int zone_to_nid(struct zone *zone) |
Christoph Lameter | 89fa302 | 2006-09-25 23:31:55 -0700 | [diff] [blame] | 737 | { |
Christoph Lameter | d5f541e | 2006-09-27 01:50:08 -0700 | [diff] [blame] | 738 | #ifdef CONFIG_NUMA |
| 739 | return zone->node; |
| 740 | #else |
| 741 | return 0; |
| 742 | #endif |
Christoph Lameter | 89fa302 | 2006-09-25 23:31:55 -0700 | [diff] [blame] | 743 | } |
| 744 | |
Christoph Lameter | 89689ae | 2006-12-06 20:31:45 -0800 | [diff] [blame] | 745 | #ifdef NODE_NOT_IN_PAGE_FLAGS |
Ian Campbell | 33dd4e0 | 2011-07-25 17:11:51 -0700 | [diff] [blame] | 746 | extern int page_to_nid(const struct page *page); |
Christoph Lameter | 89689ae | 2006-12-06 20:31:45 -0800 | [diff] [blame] | 747 | #else |
Ian Campbell | 33dd4e0 | 2011-07-25 17:11:51 -0700 | [diff] [blame] | 748 | static inline int page_to_nid(const struct page *page) |
Andy Whitcroft | d41dee3 | 2005-06-23 00:07:54 -0700 | [diff] [blame] | 749 | { |
Christoph Lameter | 89689ae | 2006-12-06 20:31:45 -0800 | [diff] [blame] | 750 | return (page->flags >> NODES_PGSHIFT) & NODES_MASK; |
Andy Whitcroft | d41dee3 | 2005-06-23 00:07:54 -0700 | [diff] [blame] | 751 | } |
Christoph Lameter | 89689ae | 2006-12-06 20:31:45 -0800 | [diff] [blame] | 752 | #endif |
| 753 | |
Mel Gorman | 57e0a03 | 2012-11-12 09:06:20 +0000 | [diff] [blame] | 754 | #ifdef CONFIG_NUMA_BALANCING |
Peter Zijlstra | 9057289 | 2013-10-07 11:29:20 +0100 | [diff] [blame] | 755 | static inline int cpu_pid_to_cpupid(int cpu, int pid) |
Mel Gorman | 57e0a03 | 2012-11-12 09:06:20 +0000 | [diff] [blame] | 756 | { |
Peter Zijlstra | 9057289 | 2013-10-07 11:29:20 +0100 | [diff] [blame] | 757 | return ((cpu & LAST__CPU_MASK) << LAST__PID_SHIFT) | (pid & LAST__PID_MASK); |
Mel Gorman | 57e0a03 | 2012-11-12 09:06:20 +0000 | [diff] [blame] | 758 | } |
| 759 | |
Peter Zijlstra | 9057289 | 2013-10-07 11:29:20 +0100 | [diff] [blame] | 760 | static inline int cpupid_to_pid(int cpupid) |
Mel Gorman | 57e0a03 | 2012-11-12 09:06:20 +0000 | [diff] [blame] | 761 | { |
Peter Zijlstra | 9057289 | 2013-10-07 11:29:20 +0100 | [diff] [blame] | 762 | return cpupid & LAST__PID_MASK; |
Mel Gorman | 57e0a03 | 2012-11-12 09:06:20 +0000 | [diff] [blame] | 763 | } |
Mel Gorman | b795854 | 2013-10-07 11:29:07 +0100 | [diff] [blame] | 764 | |
Peter Zijlstra | 9057289 | 2013-10-07 11:29:20 +0100 | [diff] [blame] | 765 | static inline int cpupid_to_cpu(int cpupid) |
Mel Gorman | 57e0a03 | 2012-11-12 09:06:20 +0000 | [diff] [blame] | 766 | { |
Peter Zijlstra | 9057289 | 2013-10-07 11:29:20 +0100 | [diff] [blame] | 767 | return (cpupid >> LAST__PID_SHIFT) & LAST__CPU_MASK; |
Mel Gorman | b795854 | 2013-10-07 11:29:07 +0100 | [diff] [blame] | 768 | } |
| 769 | |
Peter Zijlstra | 9057289 | 2013-10-07 11:29:20 +0100 | [diff] [blame] | 770 | static inline int cpupid_to_nid(int cpupid) |
Mel Gorman | b795854 | 2013-10-07 11:29:07 +0100 | [diff] [blame] | 771 | { |
Peter Zijlstra | 9057289 | 2013-10-07 11:29:20 +0100 | [diff] [blame] | 772 | return cpu_to_node(cpupid_to_cpu(cpupid)); |
Mel Gorman | b795854 | 2013-10-07 11:29:07 +0100 | [diff] [blame] | 773 | } |
| 774 | |
Peter Zijlstra | 9057289 | 2013-10-07 11:29:20 +0100 | [diff] [blame] | 775 | static inline bool cpupid_pid_unset(int cpupid) |
Mel Gorman | b795854 | 2013-10-07 11:29:07 +0100 | [diff] [blame] | 776 | { |
Peter Zijlstra | 9057289 | 2013-10-07 11:29:20 +0100 | [diff] [blame] | 777 | return cpupid_to_pid(cpupid) == (-1 & LAST__PID_MASK); |
Mel Gorman | b795854 | 2013-10-07 11:29:07 +0100 | [diff] [blame] | 778 | } |
| 779 | |
Peter Zijlstra | 9057289 | 2013-10-07 11:29:20 +0100 | [diff] [blame] | 780 | static inline bool cpupid_cpu_unset(int cpupid) |
Mel Gorman | b795854 | 2013-10-07 11:29:07 +0100 | [diff] [blame] | 781 | { |
Peter Zijlstra | 9057289 | 2013-10-07 11:29:20 +0100 | [diff] [blame] | 782 | return cpupid_to_cpu(cpupid) == (-1 & LAST__CPU_MASK); |
Mel Gorman | b795854 | 2013-10-07 11:29:07 +0100 | [diff] [blame] | 783 | } |
| 784 | |
Peter Zijlstra | 8c8a743 | 2013-10-07 11:29:21 +0100 | [diff] [blame] | 785 | static inline bool __cpupid_match_pid(pid_t task_pid, int cpupid) |
| 786 | { |
| 787 | return (task_pid & LAST__PID_MASK) == cpupid_to_pid(cpupid); |
| 788 | } |
| 789 | |
| 790 | #define cpupid_match_pid(task, cpupid) __cpupid_match_pid(task->pid, cpupid) |
Peter Zijlstra | 9057289 | 2013-10-07 11:29:20 +0100 | [diff] [blame] | 791 | #ifdef LAST_CPUPID_NOT_IN_PAGE_FLAGS |
| 792 | static inline int page_cpupid_xchg_last(struct page *page, int cpupid) |
Mel Gorman | b795854 | 2013-10-07 11:29:07 +0100 | [diff] [blame] | 793 | { |
Liu Ping Fan | 1ae71d0 | 2014-03-03 15:38:39 -0800 | [diff] [blame] | 794 | return xchg(&page->_last_cpupid, cpupid & LAST_CPUPID_MASK); |
Mel Gorman | b795854 | 2013-10-07 11:29:07 +0100 | [diff] [blame] | 795 | } |
Peter Zijlstra | 9057289 | 2013-10-07 11:29:20 +0100 | [diff] [blame] | 796 | |
| 797 | static inline int page_cpupid_last(struct page *page) |
Mel Gorman | b795854 | 2013-10-07 11:29:07 +0100 | [diff] [blame] | 798 | { |
Peter Zijlstra | 9057289 | 2013-10-07 11:29:20 +0100 | [diff] [blame] | 799 | return page->_last_cpupid; |
| 800 | } |
| 801 | static inline void page_cpupid_reset_last(struct page *page) |
| 802 | { |
Liu Ping Fan | 1ae71d0 | 2014-03-03 15:38:39 -0800 | [diff] [blame] | 803 | page->_last_cpupid = -1 & LAST_CPUPID_MASK; |
Mel Gorman | 57e0a03 | 2012-11-12 09:06:20 +0000 | [diff] [blame] | 804 | } |
| 805 | #else |
Peter Zijlstra | 9057289 | 2013-10-07 11:29:20 +0100 | [diff] [blame] | 806 | static inline int page_cpupid_last(struct page *page) |
Peter Zijlstra | 75980e9 | 2013-02-22 16:34:32 -0800 | [diff] [blame] | 807 | { |
Peter Zijlstra | 9057289 | 2013-10-07 11:29:20 +0100 | [diff] [blame] | 808 | return (page->flags >> LAST_CPUPID_PGSHIFT) & LAST_CPUPID_MASK; |
Peter Zijlstra | 75980e9 | 2013-02-22 16:34:32 -0800 | [diff] [blame] | 809 | } |
| 810 | |
Peter Zijlstra | 9057289 | 2013-10-07 11:29:20 +0100 | [diff] [blame] | 811 | extern int page_cpupid_xchg_last(struct page *page, int cpupid); |
Peter Zijlstra | 75980e9 | 2013-02-22 16:34:32 -0800 | [diff] [blame] | 812 | |
Peter Zijlstra | 9057289 | 2013-10-07 11:29:20 +0100 | [diff] [blame] | 813 | static inline void page_cpupid_reset_last(struct page *page) |
Peter Zijlstra | 75980e9 | 2013-02-22 16:34:32 -0800 | [diff] [blame] | 814 | { |
Peter Zijlstra | 9057289 | 2013-10-07 11:29:20 +0100 | [diff] [blame] | 815 | int cpupid = (1 << LAST_CPUPID_SHIFT) - 1; |
Mel Gorman | 4468b8f | 2013-02-22 16:34:46 -0800 | [diff] [blame] | 816 | |
Peter Zijlstra | 9057289 | 2013-10-07 11:29:20 +0100 | [diff] [blame] | 817 | page->flags &= ~(LAST_CPUPID_MASK << LAST_CPUPID_PGSHIFT); |
| 818 | page->flags |= (cpupid & LAST_CPUPID_MASK) << LAST_CPUPID_PGSHIFT; |
Peter Zijlstra | 75980e9 | 2013-02-22 16:34:32 -0800 | [diff] [blame] | 819 | } |
Peter Zijlstra | 9057289 | 2013-10-07 11:29:20 +0100 | [diff] [blame] | 820 | #endif /* LAST_CPUPID_NOT_IN_PAGE_FLAGS */ |
| 821 | #else /* !CONFIG_NUMA_BALANCING */ |
| 822 | static inline int page_cpupid_xchg_last(struct page *page, int cpupid) |
Mel Gorman | 57e0a03 | 2012-11-12 09:06:20 +0000 | [diff] [blame] | 823 | { |
Peter Zijlstra | 9057289 | 2013-10-07 11:29:20 +0100 | [diff] [blame] | 824 | return page_to_nid(page); /* XXX */ |
Mel Gorman | 57e0a03 | 2012-11-12 09:06:20 +0000 | [diff] [blame] | 825 | } |
| 826 | |
Peter Zijlstra | 9057289 | 2013-10-07 11:29:20 +0100 | [diff] [blame] | 827 | static inline int page_cpupid_last(struct page *page) |
Mel Gorman | 57e0a03 | 2012-11-12 09:06:20 +0000 | [diff] [blame] | 828 | { |
Peter Zijlstra | 9057289 | 2013-10-07 11:29:20 +0100 | [diff] [blame] | 829 | return page_to_nid(page); /* XXX */ |
Mel Gorman | 57e0a03 | 2012-11-12 09:06:20 +0000 | [diff] [blame] | 830 | } |
| 831 | |
Peter Zijlstra | 9057289 | 2013-10-07 11:29:20 +0100 | [diff] [blame] | 832 | static inline int cpupid_to_nid(int cpupid) |
Mel Gorman | b795854 | 2013-10-07 11:29:07 +0100 | [diff] [blame] | 833 | { |
| 834 | return -1; |
| 835 | } |
| 836 | |
Peter Zijlstra | 9057289 | 2013-10-07 11:29:20 +0100 | [diff] [blame] | 837 | static inline int cpupid_to_pid(int cpupid) |
Mel Gorman | b795854 | 2013-10-07 11:29:07 +0100 | [diff] [blame] | 838 | { |
| 839 | return -1; |
| 840 | } |
| 841 | |
Peter Zijlstra | 9057289 | 2013-10-07 11:29:20 +0100 | [diff] [blame] | 842 | static inline int cpupid_to_cpu(int cpupid) |
Mel Gorman | b795854 | 2013-10-07 11:29:07 +0100 | [diff] [blame] | 843 | { |
| 844 | return -1; |
| 845 | } |
| 846 | |
Peter Zijlstra | 9057289 | 2013-10-07 11:29:20 +0100 | [diff] [blame] | 847 | static inline int cpu_pid_to_cpupid(int nid, int pid) |
| 848 | { |
| 849 | return -1; |
| 850 | } |
| 851 | |
| 852 | static inline bool cpupid_pid_unset(int cpupid) |
Mel Gorman | b795854 | 2013-10-07 11:29:07 +0100 | [diff] [blame] | 853 | { |
| 854 | return 1; |
| 855 | } |
| 856 | |
Peter Zijlstra | 9057289 | 2013-10-07 11:29:20 +0100 | [diff] [blame] | 857 | static inline void page_cpupid_reset_last(struct page *page) |
Mel Gorman | 57e0a03 | 2012-11-12 09:06:20 +0000 | [diff] [blame] | 858 | { |
| 859 | } |
Peter Zijlstra | 8c8a743 | 2013-10-07 11:29:21 +0100 | [diff] [blame] | 860 | |
| 861 | static inline bool cpupid_match_pid(struct task_struct *task, int cpupid) |
| 862 | { |
| 863 | return false; |
| 864 | } |
Peter Zijlstra | 9057289 | 2013-10-07 11:29:20 +0100 | [diff] [blame] | 865 | #endif /* CONFIG_NUMA_BALANCING */ |
Mel Gorman | 57e0a03 | 2012-11-12 09:06:20 +0000 | [diff] [blame] | 866 | |
Ian Campbell | 33dd4e0 | 2011-07-25 17:11:51 -0700 | [diff] [blame] | 867 | static inline struct zone *page_zone(const struct page *page) |
Christoph Lameter | 89689ae | 2006-12-06 20:31:45 -0800 | [diff] [blame] | 868 | { |
| 869 | return &NODE_DATA(page_to_nid(page))->node_zones[page_zonenum(page)]; |
| 870 | } |
| 871 | |
Cody P Schafer | 9127ab4 | 2013-02-22 16:35:21 -0800 | [diff] [blame] | 872 | #ifdef SECTION_IN_PAGE_FLAGS |
Daniel Kiper | bf4e890 | 2011-05-24 17:12:32 -0700 | [diff] [blame] | 873 | static inline void set_page_section(struct page *page, unsigned long section) |
| 874 | { |
| 875 | page->flags &= ~(SECTIONS_MASK << SECTIONS_PGSHIFT); |
| 876 | page->flags |= (section & SECTIONS_MASK) << SECTIONS_PGSHIFT; |
| 877 | } |
| 878 | |
Ian Campbell | aa462ab | 2011-08-17 17:40:33 +0100 | [diff] [blame] | 879 | static inline unsigned long page_to_section(const struct page *page) |
Andy Whitcroft | d41dee3 | 2005-06-23 00:07:54 -0700 | [diff] [blame] | 880 | { |
| 881 | return (page->flags >> SECTIONS_PGSHIFT) & SECTIONS_MASK; |
| 882 | } |
Christoph Lameter | 308c05e | 2008-04-28 02:12:43 -0700 | [diff] [blame] | 883 | #endif |
Andy Whitcroft | d41dee3 | 2005-06-23 00:07:54 -0700 | [diff] [blame] | 884 | |
Christoph Lameter | 2f1b624 | 2006-09-25 23:31:13 -0700 | [diff] [blame] | 885 | static inline void set_page_zone(struct page *page, enum zone_type zone) |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 886 | { |
Dave Hansen | 348f8b6 | 2005-06-23 00:07:40 -0700 | [diff] [blame] | 887 | page->flags &= ~(ZONES_MASK << ZONES_PGSHIFT); |
| 888 | page->flags |= (zone & ZONES_MASK) << ZONES_PGSHIFT; |
| 889 | } |
Christoph Lameter | 2f1b624 | 2006-09-25 23:31:13 -0700 | [diff] [blame] | 890 | |
Dave Hansen | 348f8b6 | 2005-06-23 00:07:40 -0700 | [diff] [blame] | 891 | static inline void set_page_node(struct page *page, unsigned long node) |
| 892 | { |
| 893 | page->flags &= ~(NODES_MASK << NODES_PGSHIFT); |
| 894 | page->flags |= (node & NODES_MASK) << NODES_PGSHIFT; |
| 895 | } |
Christoph Lameter | 89689ae | 2006-12-06 20:31:45 -0800 | [diff] [blame] | 896 | |
Christoph Lameter | 2f1b624 | 2006-09-25 23:31:13 -0700 | [diff] [blame] | 897 | static inline void set_page_links(struct page *page, enum zone_type zone, |
Andy Whitcroft | d41dee3 | 2005-06-23 00:07:54 -0700 | [diff] [blame] | 898 | unsigned long node, unsigned long pfn) |
Dave Hansen | 348f8b6 | 2005-06-23 00:07:40 -0700 | [diff] [blame] | 899 | { |
| 900 | set_page_zone(page, zone); |
| 901 | set_page_node(page, node); |
Cody P Schafer | 9127ab4 | 2013-02-22 16:35:21 -0800 | [diff] [blame] | 902 | #ifdef SECTION_IN_PAGE_FLAGS |
Andy Whitcroft | d41dee3 | 2005-06-23 00:07:54 -0700 | [diff] [blame] | 903 | set_page_section(page, pfn_to_section_nr(pfn)); |
Daniel Kiper | bf4e890 | 2011-05-24 17:12:32 -0700 | [diff] [blame] | 904 | #endif |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 905 | } |
| 906 | |
Christoph Lameter | f6ac235 | 2006-06-30 01:55:32 -0700 | [diff] [blame] | 907 | /* |
| 908 | * Some inline functions in vmstat.h depend on page_zone() |
| 909 | */ |
| 910 | #include <linux/vmstat.h> |
| 911 | |
Ian Campbell | 33dd4e0 | 2011-07-25 17:11:51 -0700 | [diff] [blame] | 912 | static __always_inline void *lowmem_page_address(const struct page *page) |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 913 | { |
Ian Campbell | aa462ab | 2011-08-17 17:40:33 +0100 | [diff] [blame] | 914 | return __va(PFN_PHYS(page_to_pfn(page))); |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 915 | } |
| 916 | |
| 917 | #if defined(CONFIG_HIGHMEM) && !defined(WANT_PAGE_VIRTUAL) |
| 918 | #define HASHED_PAGE_VIRTUAL |
| 919 | #endif |
| 920 | |
| 921 | #if defined(WANT_PAGE_VIRTUAL) |
Geert Uytterhoeven | f92f455 | 2014-01-21 15:48:47 -0800 | [diff] [blame] | 922 | static inline void *page_address(const struct page *page) |
| 923 | { |
| 924 | return page->virtual; |
| 925 | } |
| 926 | static inline void set_page_address(struct page *page, void *address) |
| 927 | { |
| 928 | page->virtual = address; |
| 929 | } |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 930 | #define page_address_init() do { } while(0) |
| 931 | #endif |
| 932 | |
| 933 | #if defined(HASHED_PAGE_VIRTUAL) |
Ian Campbell | f991879 | 2011-08-17 13:45:09 +0100 | [diff] [blame] | 934 | void *page_address(const struct page *page); |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 935 | void set_page_address(struct page *page, void *virtual); |
| 936 | void page_address_init(void); |
| 937 | #endif |
| 938 | |
| 939 | #if !defined(HASHED_PAGE_VIRTUAL) && !defined(WANT_PAGE_VIRTUAL) |
| 940 | #define page_address(page) lowmem_page_address(page) |
| 941 | #define set_page_address(page, address) do { } while(0) |
| 942 | #define page_address_init() do { } while(0) |
| 943 | #endif |
| 944 | |
| 945 | /* |
| 946 | * On an anonymous page mapped into a user virtual memory area, |
| 947 | * page->mapping points to its anon_vma, not to a struct address_space; |
Hugh Dickins | 3ca7b3c | 2009-12-14 17:58:57 -0800 | [diff] [blame] | 948 | * with the PAGE_MAPPING_ANON bit set to distinguish it. See rmap.h. |
| 949 | * |
| 950 | * On an anonymous page in a VM_MERGEABLE area, if CONFIG_KSM is enabled, |
| 951 | * the PAGE_MAPPING_KSM bit may be set along with the PAGE_MAPPING_ANON bit; |
| 952 | * and then page->mapping points, not to an anon_vma, but to a private |
| 953 | * structure which KSM associates with that merged page. See ksm.h. |
| 954 | * |
| 955 | * PAGE_MAPPING_KSM without PAGE_MAPPING_ANON is currently never used. |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 956 | * |
| 957 | * Please note that, confusingly, "page_mapping" refers to the inode |
| 958 | * address_space which maps the page from disk; whereas "page_mapped" |
| 959 | * refers to user virtual address space into which the page is mapped. |
| 960 | */ |
| 961 | #define PAGE_MAPPING_ANON 1 |
Hugh Dickins | 3ca7b3c | 2009-12-14 17:58:57 -0800 | [diff] [blame] | 962 | #define PAGE_MAPPING_KSM 2 |
| 963 | #define PAGE_MAPPING_FLAGS (PAGE_MAPPING_ANON | PAGE_MAPPING_KSM) |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 964 | |
Shaohua Li | 9800339 | 2013-02-22 16:34:35 -0800 | [diff] [blame] | 965 | extern struct address_space *page_mapping(struct page *page); |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 966 | |
Hugh Dickins | 3ca7b3c | 2009-12-14 17:58:57 -0800 | [diff] [blame] | 967 | /* Neutral page->mapping pointer to address_space or anon_vma or other */ |
| 968 | static inline void *page_rmapping(struct page *page) |
| 969 | { |
| 970 | return (void *)((unsigned long)page->mapping & ~PAGE_MAPPING_FLAGS); |
| 971 | } |
| 972 | |
Mel Gorman | f981c59 | 2012-07-31 16:44:47 -0700 | [diff] [blame] | 973 | extern struct address_space *__page_file_mapping(struct page *); |
| 974 | |
| 975 | static inline |
| 976 | struct address_space *page_file_mapping(struct page *page) |
| 977 | { |
| 978 | if (unlikely(PageSwapCache(page))) |
| 979 | return __page_file_mapping(page); |
| 980 | |
| 981 | return page->mapping; |
| 982 | } |
| 983 | |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 984 | static inline int PageAnon(struct page *page) |
| 985 | { |
| 986 | return ((unsigned long)page->mapping & PAGE_MAPPING_ANON) != 0; |
| 987 | } |
| 988 | |
| 989 | /* |
| 990 | * Return the pagecache index of the passed page. Regular pagecache pages |
| 991 | * use ->index whereas swapcache pages use ->private |
| 992 | */ |
| 993 | static inline pgoff_t page_index(struct page *page) |
| 994 | { |
| 995 | if (unlikely(PageSwapCache(page))) |
Hugh Dickins | 4c21e2f | 2005-10-29 18:16:40 -0700 | [diff] [blame] | 996 | return page_private(page); |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 997 | return page->index; |
| 998 | } |
| 999 | |
Mel Gorman | f981c59 | 2012-07-31 16:44:47 -0700 | [diff] [blame] | 1000 | extern pgoff_t __page_file_index(struct page *page); |
| 1001 | |
| 1002 | /* |
| 1003 | * Return the file index of the page. Regular pagecache pages use ->index |
| 1004 | * whereas swapcache pages use swp_offset(->private) |
| 1005 | */ |
| 1006 | static inline pgoff_t page_file_index(struct page *page) |
| 1007 | { |
| 1008 | if (unlikely(PageSwapCache(page))) |
| 1009 | return __page_file_index(page); |
| 1010 | |
| 1011 | return page->index; |
| 1012 | } |
| 1013 | |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1014 | /* |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1015 | * Return true if this page is mapped into pagetables. |
| 1016 | */ |
| 1017 | static inline int page_mapped(struct page *page) |
| 1018 | { |
| 1019 | return atomic_read(&(page)->_mapcount) >= 0; |
| 1020 | } |
| 1021 | |
| 1022 | /* |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1023 | * Different kinds of faults, as returned by handle_mm_fault(). |
| 1024 | * Used to decide whether a process gets delivered SIGBUS or |
| 1025 | * just gets major/minor fault counters bumped up. |
| 1026 | */ |
Nick Piggin | d0217ac | 2007-07-19 01:47:03 -0700 | [diff] [blame] | 1027 | |
Nick Piggin | 83c5407 | 2007-07-19 01:47:05 -0700 | [diff] [blame] | 1028 | #define VM_FAULT_MINOR 0 /* For backwards compat. Remove me quickly. */ |
Nick Piggin | d0217ac | 2007-07-19 01:47:03 -0700 | [diff] [blame] | 1029 | |
Nick Piggin | 83c5407 | 2007-07-19 01:47:05 -0700 | [diff] [blame] | 1030 | #define VM_FAULT_OOM 0x0001 |
| 1031 | #define VM_FAULT_SIGBUS 0x0002 |
| 1032 | #define VM_FAULT_MAJOR 0x0004 |
| 1033 | #define VM_FAULT_WRITE 0x0008 /* Special case for get_user_pages */ |
Andi Kleen | aa50d3a | 2010-10-06 21:45:00 +0200 | [diff] [blame] | 1034 | #define VM_FAULT_HWPOISON 0x0010 /* Hit poisoned small page */ |
| 1035 | #define VM_FAULT_HWPOISON_LARGE 0x0020 /* Hit poisoned large page. Index encoded in upper bits */ |
Nick Piggin | f33ea7f | 2005-08-03 20:24:01 +1000 | [diff] [blame] | 1036 | |
Nick Piggin | 83c5407 | 2007-07-19 01:47:05 -0700 | [diff] [blame] | 1037 | #define VM_FAULT_NOPAGE 0x0100 /* ->fault installed the pte, not return page */ |
| 1038 | #define VM_FAULT_LOCKED 0x0200 /* ->fault locked the returned page */ |
Michel Lespinasse | d065bd8 | 2010-10-26 14:21:57 -0700 | [diff] [blame] | 1039 | #define VM_FAULT_RETRY 0x0400 /* ->fault blocked, must retry */ |
Kirill A. Shutemov | c029255 | 2013-09-12 15:14:05 -0700 | [diff] [blame] | 1040 | #define VM_FAULT_FALLBACK 0x0800 /* huge page fault failed, fall back to small */ |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1041 | |
Andi Kleen | aa50d3a | 2010-10-06 21:45:00 +0200 | [diff] [blame] | 1042 | #define VM_FAULT_HWPOISON_LARGE_MASK 0xf000 /* encodes hpage index for large hwpoison */ |
| 1043 | |
| 1044 | #define VM_FAULT_ERROR (VM_FAULT_OOM | VM_FAULT_SIGBUS | VM_FAULT_HWPOISON | \ |
Kirill A. Shutemov | c029255 | 2013-09-12 15:14:05 -0700 | [diff] [blame] | 1045 | VM_FAULT_FALLBACK | VM_FAULT_HWPOISON_LARGE) |
Andi Kleen | aa50d3a | 2010-10-06 21:45:00 +0200 | [diff] [blame] | 1046 | |
| 1047 | /* Encode hstate index for a hwpoisoned large page */ |
| 1048 | #define VM_FAULT_SET_HINDEX(x) ((x) << 12) |
| 1049 | #define VM_FAULT_GET_HINDEX(x) (((x) >> 12) & 0xf) |
Nick Piggin | d0217ac | 2007-07-19 01:47:03 -0700 | [diff] [blame] | 1050 | |
Nick Piggin | 1c0fe6e | 2009-01-06 14:38:59 -0800 | [diff] [blame] | 1051 | /* |
| 1052 | * Can be called by the pagefault handler when it gets a VM_FAULT_OOM. |
| 1053 | */ |
| 1054 | extern void pagefault_out_of_memory(void); |
| 1055 | |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1056 | #define offset_in_page(p) ((unsigned long)(p) & ~PAGE_MASK) |
| 1057 | |
David Rientjes | ddd588b | 2011-03-22 16:30:46 -0700 | [diff] [blame] | 1058 | /* |
David Rientjes | 7bf02ea | 2011-05-24 17:11:16 -0700 | [diff] [blame] | 1059 | * Flags passed to show_mem() and show_free_areas() to suppress output in |
David Rientjes | ddd588b | 2011-03-22 16:30:46 -0700 | [diff] [blame] | 1060 | * various contexts. |
| 1061 | */ |
David Rientjes | 4b59e6c | 2013-04-29 15:06:11 -0700 | [diff] [blame] | 1062 | #define SHOW_MEM_FILTER_NODES (0x0001u) /* disallowed nodes */ |
David Rientjes | ddd588b | 2011-03-22 16:30:46 -0700 | [diff] [blame] | 1063 | |
David Rientjes | 7bf02ea | 2011-05-24 17:11:16 -0700 | [diff] [blame] | 1064 | extern void show_free_areas(unsigned int flags); |
| 1065 | extern bool skip_free_areas_node(unsigned int flags, int nid); |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1066 | |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1067 | int shmem_zero_setup(struct vm_area_struct *); |
Johannes Weiner | 0cd6144 | 2014-04-03 14:47:46 -0700 | [diff] [blame] | 1068 | #ifdef CONFIG_SHMEM |
| 1069 | bool shmem_mapping(struct address_space *mapping); |
| 1070 | #else |
| 1071 | static inline bool shmem_mapping(struct address_space *mapping) |
| 1072 | { |
| 1073 | return false; |
| 1074 | } |
| 1075 | #endif |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1076 | |
Alexey Dobriyan | e8edc6e | 2007-05-21 01:22:52 +0400 | [diff] [blame] | 1077 | extern int can_do_mlock(void); |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1078 | extern int user_shm_lock(size_t, struct user_struct *); |
| 1079 | extern void user_shm_unlock(size_t, struct user_struct *); |
| 1080 | |
| 1081 | /* |
| 1082 | * Parameter block passed down to zap_pte_range in exceptional cases. |
| 1083 | */ |
| 1084 | struct zap_details { |
| 1085 | struct vm_area_struct *nonlinear_vma; /* Check page->index if set */ |
| 1086 | struct address_space *check_mapping; /* Check page->mapping if set */ |
| 1087 | pgoff_t first_index; /* Lowest page->index to unmap */ |
| 1088 | pgoff_t last_index; /* Highest page->index to unmap */ |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1089 | }; |
| 1090 | |
Nick Piggin | 7e67513 | 2008-04-28 02:13:00 -0700 | [diff] [blame] | 1091 | struct page *vm_normal_page(struct vm_area_struct *vma, unsigned long addr, |
| 1092 | pte_t pte); |
| 1093 | |
Jack Steiner | c627f9c | 2008-07-29 22:33:53 -0700 | [diff] [blame] | 1094 | int zap_vma_ptes(struct vm_area_struct *vma, unsigned long address, |
| 1095 | unsigned long size); |
Al Viro | 14f5ff5 | 2012-03-05 13:38:09 -0500 | [diff] [blame] | 1096 | void zap_page_range(struct vm_area_struct *vma, unsigned long address, |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1097 | unsigned long size, struct zap_details *); |
Linus Torvalds | 4f74d2c | 2012-05-06 13:54:06 -0700 | [diff] [blame] | 1098 | void unmap_vmas(struct mmu_gather *tlb, struct vm_area_struct *start_vma, |
| 1099 | unsigned long start, unsigned long end); |
Matt Mackall | e647309 | 2008-02-04 22:29:01 -0800 | [diff] [blame] | 1100 | |
| 1101 | /** |
| 1102 | * mm_walk - callbacks for walk_page_range |
| 1103 | * @pgd_entry: if set, called for each non-empty PGD (top-level) entry |
| 1104 | * @pud_entry: if set, called for each non-empty PUD (2nd-level) entry |
| 1105 | * @pmd_entry: if set, called for each non-empty PMD (3rd-level) entry |
Dave Hansen | 0331932 | 2011-03-22 16:32:56 -0700 | [diff] [blame] | 1106 | * this handler is required to be able to handle |
| 1107 | * pmd_trans_huge() pmds. They may simply choose to |
| 1108 | * split_huge_page() instead of handling it explicitly. |
Matt Mackall | e647309 | 2008-02-04 22:29:01 -0800 | [diff] [blame] | 1109 | * @pte_entry: if set, called for each non-empty PTE (4th-level) entry |
| 1110 | * @pte_hole: if set, called for each hole at all levels |
Naoya Horiguchi | 5dc3764 | 2009-12-14 18:00:01 -0800 | [diff] [blame] | 1111 | * @hugetlb_entry: if set, called for each hugetlb entry |
KOSAKI Motohiro | c27fe4c8 | 2011-07-25 17:12:10 -0700 | [diff] [blame] | 1112 | * *Caution*: The caller must hold mmap_sem() if @hugetlb_entry |
| 1113 | * is used. |
Matt Mackall | e647309 | 2008-02-04 22:29:01 -0800 | [diff] [blame] | 1114 | * |
| 1115 | * (see walk_page_range for more details) |
| 1116 | */ |
| 1117 | struct mm_walk { |
Andrew Morton | 0f157a5 | 2013-05-07 16:18:10 -0700 | [diff] [blame] | 1118 | int (*pgd_entry)(pgd_t *pgd, unsigned long addr, |
| 1119 | unsigned long next, struct mm_walk *walk); |
| 1120 | int (*pud_entry)(pud_t *pud, unsigned long addr, |
| 1121 | unsigned long next, struct mm_walk *walk); |
| 1122 | int (*pmd_entry)(pmd_t *pmd, unsigned long addr, |
| 1123 | unsigned long next, struct mm_walk *walk); |
| 1124 | int (*pte_entry)(pte_t *pte, unsigned long addr, |
| 1125 | unsigned long next, struct mm_walk *walk); |
| 1126 | int (*pte_hole)(unsigned long addr, unsigned long next, |
| 1127 | struct mm_walk *walk); |
| 1128 | int (*hugetlb_entry)(pte_t *pte, unsigned long hmask, |
| 1129 | unsigned long addr, unsigned long next, |
| 1130 | struct mm_walk *walk); |
Dave Hansen | 2165009 | 2008-06-12 15:21:47 -0700 | [diff] [blame] | 1131 | struct mm_struct *mm; |
| 1132 | void *private; |
Matt Mackall | e647309 | 2008-02-04 22:29:01 -0800 | [diff] [blame] | 1133 | }; |
| 1134 | |
Dave Hansen | 2165009 | 2008-06-12 15:21:47 -0700 | [diff] [blame] | 1135 | int walk_page_range(unsigned long addr, unsigned long end, |
| 1136 | struct mm_walk *walk); |
Jan Beulich | 42b7772 | 2008-07-23 21:27:10 -0700 | [diff] [blame] | 1137 | void free_pgd_range(struct mmu_gather *tlb, unsigned long addr, |
Hugh Dickins | 3bf5ee9 | 2005-04-19 13:29:16 -0700 | [diff] [blame] | 1138 | unsigned long end, unsigned long floor, unsigned long ceiling); |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1139 | int copy_page_range(struct mm_struct *dst, struct mm_struct *src, |
| 1140 | struct vm_area_struct *vma); |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1141 | void unmap_mapping_range(struct address_space *mapping, |
| 1142 | loff_t const holebegin, loff_t const holelen, int even_cows); |
Johannes Weiner | 3b6748e | 2009-06-16 15:32:35 -0700 | [diff] [blame] | 1143 | int follow_pfn(struct vm_area_struct *vma, unsigned long address, |
| 1144 | unsigned long *pfn); |
venkatesh.pallipadi@intel.com | d87fe66 | 2008-12-19 13:47:27 -0800 | [diff] [blame] | 1145 | int follow_phys(struct vm_area_struct *vma, unsigned long address, |
| 1146 | unsigned int flags, unsigned long *prot, resource_size_t *phys); |
Rik van Riel | 28b2ee2 | 2008-07-23 21:27:05 -0700 | [diff] [blame] | 1147 | int generic_access_phys(struct vm_area_struct *vma, unsigned long addr, |
| 1148 | void *buf, int len, int write); |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1149 | |
| 1150 | static inline void unmap_shared_mapping_range(struct address_space *mapping, |
| 1151 | loff_t const holebegin, loff_t const holelen) |
| 1152 | { |
| 1153 | unmap_mapping_range(mapping, holebegin, holelen, 0); |
| 1154 | } |
| 1155 | |
Kirill A. Shutemov | 7caef26 | 2013-09-12 15:13:56 -0700 | [diff] [blame] | 1156 | extern void truncate_pagecache(struct inode *inode, loff_t new); |
Christoph Hellwig | 2c27c65 | 2010-06-04 11:30:04 +0200 | [diff] [blame] | 1157 | extern void truncate_setsize(struct inode *inode, loff_t newsize); |
Hugh Dickins | 623e3db | 2012-03-28 14:42:40 -0700 | [diff] [blame] | 1158 | void truncate_pagecache_range(struct inode *inode, loff_t offset, loff_t end); |
Nick Piggin | 750b498 | 2009-09-16 11:50:12 +0200 | [diff] [blame] | 1159 | int truncate_inode_page(struct address_space *mapping, struct page *page); |
Andi Kleen | 2571873 | 2009-09-16 11:50:13 +0200 | [diff] [blame] | 1160 | int generic_error_remove_page(struct address_space *mapping, struct page *page); |
Wu Fengguang | 83f7866 | 2009-09-16 11:50:13 +0200 | [diff] [blame] | 1161 | int invalidate_inode_page(struct page *page); |
| 1162 | |
David Howells | 7ee1dd3 | 2006-01-06 00:11:44 -0800 | [diff] [blame] | 1163 | #ifdef CONFIG_MMU |
Nick Piggin | 83c5407 | 2007-07-19 01:47:05 -0700 | [diff] [blame] | 1164 | extern int handle_mm_fault(struct mm_struct *mm, struct vm_area_struct *vma, |
Linus Torvalds | d06063c | 2009-04-10 09:01:23 -0700 | [diff] [blame] | 1165 | unsigned long address, unsigned int flags); |
Peter Zijlstra | 5c723ba | 2011-07-27 12:17:11 +0200 | [diff] [blame] | 1166 | extern int fixup_user_fault(struct task_struct *tsk, struct mm_struct *mm, |
| 1167 | unsigned long address, unsigned int fault_flags); |
David Howells | 7ee1dd3 | 2006-01-06 00:11:44 -0800 | [diff] [blame] | 1168 | #else |
| 1169 | static inline int handle_mm_fault(struct mm_struct *mm, |
| 1170 | struct vm_area_struct *vma, unsigned long address, |
Linus Torvalds | d06063c | 2009-04-10 09:01:23 -0700 | [diff] [blame] | 1171 | unsigned int flags) |
David Howells | 7ee1dd3 | 2006-01-06 00:11:44 -0800 | [diff] [blame] | 1172 | { |
| 1173 | /* should never happen if there's no MMU */ |
| 1174 | BUG(); |
| 1175 | return VM_FAULT_SIGBUS; |
| 1176 | } |
Peter Zijlstra | 5c723ba | 2011-07-27 12:17:11 +0200 | [diff] [blame] | 1177 | static inline int fixup_user_fault(struct task_struct *tsk, |
| 1178 | struct mm_struct *mm, unsigned long address, |
| 1179 | unsigned int fault_flags) |
| 1180 | { |
| 1181 | /* should never happen if there's no MMU */ |
| 1182 | BUG(); |
| 1183 | return -EFAULT; |
| 1184 | } |
David Howells | 7ee1dd3 | 2006-01-06 00:11:44 -0800 | [diff] [blame] | 1185 | #endif |
Nick Piggin | f33ea7f | 2005-08-03 20:24:01 +1000 | [diff] [blame] | 1186 | |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1187 | extern int access_process_vm(struct task_struct *tsk, unsigned long addr, void *buf, int len, int write); |
Stephen Wilson | 5ddd36b | 2011-03-13 15:49:20 -0400 | [diff] [blame] | 1188 | extern int access_remote_vm(struct mm_struct *mm, unsigned long addr, |
| 1189 | void *buf, int len, int write); |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1190 | |
Michel Lespinasse | 28a3571 | 2013-02-22 16:35:55 -0800 | [diff] [blame] | 1191 | long __get_user_pages(struct task_struct *tsk, struct mm_struct *mm, |
| 1192 | unsigned long start, unsigned long nr_pages, |
| 1193 | unsigned int foll_flags, struct page **pages, |
| 1194 | struct vm_area_struct **vmas, int *nonblocking); |
| 1195 | long get_user_pages(struct task_struct *tsk, struct mm_struct *mm, |
| 1196 | unsigned long start, unsigned long nr_pages, |
| 1197 | int write, int force, struct page **pages, |
| 1198 | struct vm_area_struct **vmas); |
Nick Piggin | d2bf6be | 2009-06-16 15:31:39 -0700 | [diff] [blame] | 1199 | int get_user_pages_fast(unsigned long start, int nr_pages, int write, |
| 1200 | struct page **pages); |
Mel Gorman | 18022c5 | 2012-07-31 16:44:51 -0700 | [diff] [blame] | 1201 | struct kvec; |
| 1202 | int get_kernel_pages(const struct kvec *iov, int nr_pages, int write, |
| 1203 | struct page **pages); |
| 1204 | int get_kernel_page(unsigned long start, int write, struct page **pages); |
Hugh Dickins | f3e8fcc | 2009-09-21 17:03:25 -0700 | [diff] [blame] | 1205 | struct page *get_dump_page(unsigned long addr); |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1206 | |
David Howells | cf9a2ae | 2006-08-29 19:05:54 +0100 | [diff] [blame] | 1207 | extern int try_to_release_page(struct page * page, gfp_t gfp_mask); |
Lukas Czerner | d47992f | 2013-05-21 23:17:23 -0400 | [diff] [blame] | 1208 | extern void do_invalidatepage(struct page *page, unsigned int offset, |
| 1209 | unsigned int length); |
David Howells | cf9a2ae | 2006-08-29 19:05:54 +0100 | [diff] [blame] | 1210 | |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1211 | int __set_page_dirty_nobuffers(struct page *page); |
Ken Chen | 7671932 | 2007-02-10 01:43:15 -0800 | [diff] [blame] | 1212 | int __set_page_dirty_no_writeback(struct page *page); |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1213 | int redirty_page_for_writepage(struct writeback_control *wbc, |
| 1214 | struct page *page); |
Edward Shishkin | e3a7cca | 2009-03-31 15:19:39 -0700 | [diff] [blame] | 1215 | void account_page_dirtied(struct page *page, struct address_space *mapping); |
Michael Rubin | f629d1c | 2010-10-26 14:21:33 -0700 | [diff] [blame] | 1216 | void account_page_writeback(struct page *page); |
Harvey Harrison | b3c9752 | 2008-02-13 15:03:15 -0800 | [diff] [blame] | 1217 | int set_page_dirty(struct page *page); |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1218 | int set_page_dirty_lock(struct page *page); |
| 1219 | int clear_page_dirty_for_io(struct page *page); |
William Roberts | a909025 | 2014-02-11 10:11:59 -0800 | [diff] [blame] | 1220 | int get_cmdline(struct task_struct *task, char *buffer, int buflen); |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1221 | |
Stefan Bader | 39aa3cb | 2010-08-31 15:52:27 +0200 | [diff] [blame] | 1222 | /* Is the vma a continuation of the stack vma above it? */ |
Mikulas Patocka | a09a79f | 2011-05-09 13:01:09 +0200 | [diff] [blame] | 1223 | static inline int vma_growsdown(struct vm_area_struct *vma, unsigned long addr) |
Stefan Bader | 39aa3cb | 2010-08-31 15:52:27 +0200 | [diff] [blame] | 1224 | { |
| 1225 | return vma && (vma->vm_end == addr) && (vma->vm_flags & VM_GROWSDOWN); |
| 1226 | } |
| 1227 | |
Mikulas Patocka | a09a79f | 2011-05-09 13:01:09 +0200 | [diff] [blame] | 1228 | static inline int stack_guard_page_start(struct vm_area_struct *vma, |
| 1229 | unsigned long addr) |
| 1230 | { |
| 1231 | return (vma->vm_flags & VM_GROWSDOWN) && |
| 1232 | (vma->vm_start == addr) && |
| 1233 | !vma_growsdown(vma->vm_prev, addr); |
| 1234 | } |
| 1235 | |
| 1236 | /* Is the vma a continuation of the stack vma below it? */ |
| 1237 | static inline int vma_growsup(struct vm_area_struct *vma, unsigned long addr) |
| 1238 | { |
| 1239 | return vma && (vma->vm_start == addr) && (vma->vm_flags & VM_GROWSUP); |
| 1240 | } |
| 1241 | |
| 1242 | static inline int stack_guard_page_end(struct vm_area_struct *vma, |
| 1243 | unsigned long addr) |
| 1244 | { |
| 1245 | return (vma->vm_flags & VM_GROWSUP) && |
| 1246 | (vma->vm_end == addr) && |
| 1247 | !vma_growsup(vma->vm_next, addr); |
| 1248 | } |
| 1249 | |
Siddhesh Poyarekar | b764375 | 2012-03-21 16:34:04 -0700 | [diff] [blame] | 1250 | extern pid_t |
| 1251 | vm_is_stack(struct task_struct *task, struct vm_area_struct *vma, int in_group); |
| 1252 | |
Ollie Wild | b6a2fea | 2007-07-19 01:48:16 -0700 | [diff] [blame] | 1253 | extern unsigned long move_page_tables(struct vm_area_struct *vma, |
| 1254 | unsigned long old_addr, struct vm_area_struct *new_vma, |
Michel Lespinasse | 38a7601 | 2012-10-08 16:31:50 -0700 | [diff] [blame] | 1255 | unsigned long new_addr, unsigned long len, |
| 1256 | bool need_rmap_locks); |
Peter Zijlstra | 7da4d64 | 2012-11-19 03:14:23 +0100 | [diff] [blame] | 1257 | extern unsigned long change_protection(struct vm_area_struct *vma, unsigned long start, |
| 1258 | unsigned long end, pgprot_t newprot, |
Mel Gorman | 4b10e7d | 2012-10-25 14:16:32 +0200 | [diff] [blame] | 1259 | int dirty_accountable, int prot_numa); |
Ollie Wild | b6a2fea | 2007-07-19 01:48:16 -0700 | [diff] [blame] | 1260 | extern int mprotect_fixup(struct vm_area_struct *vma, |
| 1261 | struct vm_area_struct **pprev, unsigned long start, |
| 1262 | unsigned long end, unsigned long newflags); |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1263 | |
Nick Piggin | 21cc199 | 2008-07-25 19:45:22 -0700 | [diff] [blame] | 1264 | /* |
Peter Zijlstra | 465a454 | 2009-06-15 12:31:37 +0200 | [diff] [blame] | 1265 | * doesn't attempt to fault and will return short. |
| 1266 | */ |
| 1267 | int __get_user_pages_fast(unsigned long start, int nr_pages, int write, |
| 1268 | struct page **pages); |
KAMEZAWA Hiroyuki | d559db0 | 2010-03-05 13:41:39 -0800 | [diff] [blame] | 1269 | /* |
| 1270 | * per-process(per-mm_struct) statistics. |
| 1271 | */ |
Matt Fleming | 172703b | 2011-05-24 17:12:36 -0700 | [diff] [blame] | 1272 | static inline unsigned long get_mm_counter(struct mm_struct *mm, int member) |
| 1273 | { |
Konstantin Khlebnikov | 69c9782 | 2012-03-21 16:33:49 -0700 | [diff] [blame] | 1274 | long val = atomic_long_read(&mm->rss_stat.count[member]); |
| 1275 | |
| 1276 | #ifdef SPLIT_RSS_COUNTING |
| 1277 | /* |
| 1278 | * counter is updated in asynchronous manner and may go to minus. |
| 1279 | * But it's never be expected number for users. |
| 1280 | */ |
| 1281 | if (val < 0) |
| 1282 | val = 0; |
Matt Fleming | 172703b | 2011-05-24 17:12:36 -0700 | [diff] [blame] | 1283 | #endif |
Konstantin Khlebnikov | 69c9782 | 2012-03-21 16:33:49 -0700 | [diff] [blame] | 1284 | return (unsigned long)val; |
| 1285 | } |
KAMEZAWA Hiroyuki | d559db0 | 2010-03-05 13:41:39 -0800 | [diff] [blame] | 1286 | |
| 1287 | static inline void add_mm_counter(struct mm_struct *mm, int member, long value) |
| 1288 | { |
| 1289 | atomic_long_add(value, &mm->rss_stat.count[member]); |
| 1290 | } |
| 1291 | |
| 1292 | static inline void inc_mm_counter(struct mm_struct *mm, int member) |
| 1293 | { |
| 1294 | atomic_long_inc(&mm->rss_stat.count[member]); |
| 1295 | } |
| 1296 | |
| 1297 | static inline void dec_mm_counter(struct mm_struct *mm, int member) |
| 1298 | { |
| 1299 | atomic_long_dec(&mm->rss_stat.count[member]); |
| 1300 | } |
| 1301 | |
KAMEZAWA Hiroyuki | d559db0 | 2010-03-05 13:41:39 -0800 | [diff] [blame] | 1302 | static inline unsigned long get_mm_rss(struct mm_struct *mm) |
| 1303 | { |
| 1304 | return get_mm_counter(mm, MM_FILEPAGES) + |
| 1305 | get_mm_counter(mm, MM_ANONPAGES); |
| 1306 | } |
| 1307 | |
| 1308 | static inline unsigned long get_mm_hiwater_rss(struct mm_struct *mm) |
| 1309 | { |
| 1310 | return max(mm->hiwater_rss, get_mm_rss(mm)); |
| 1311 | } |
| 1312 | |
| 1313 | static inline unsigned long get_mm_hiwater_vm(struct mm_struct *mm) |
| 1314 | { |
| 1315 | return max(mm->hiwater_vm, mm->total_vm); |
| 1316 | } |
| 1317 | |
| 1318 | static inline void update_hiwater_rss(struct mm_struct *mm) |
| 1319 | { |
| 1320 | unsigned long _rss = get_mm_rss(mm); |
| 1321 | |
| 1322 | if ((mm)->hiwater_rss < _rss) |
| 1323 | (mm)->hiwater_rss = _rss; |
| 1324 | } |
| 1325 | |
| 1326 | static inline void update_hiwater_vm(struct mm_struct *mm) |
| 1327 | { |
| 1328 | if (mm->hiwater_vm < mm->total_vm) |
| 1329 | mm->hiwater_vm = mm->total_vm; |
| 1330 | } |
| 1331 | |
| 1332 | static inline void setmax_mm_hiwater_rss(unsigned long *maxrss, |
| 1333 | struct mm_struct *mm) |
| 1334 | { |
| 1335 | unsigned long hiwater_rss = get_mm_hiwater_rss(mm); |
| 1336 | |
| 1337 | if (*maxrss < hiwater_rss) |
| 1338 | *maxrss = hiwater_rss; |
| 1339 | } |
| 1340 | |
KAMEZAWA Hiroyuki | 53bddb4 | 2010-03-10 15:20:38 -0800 | [diff] [blame] | 1341 | #if defined(SPLIT_RSS_COUNTING) |
David Rientjes | 05af2e1 | 2012-03-21 16:34:13 -0700 | [diff] [blame] | 1342 | void sync_mm_rss(struct mm_struct *mm); |
KAMEZAWA Hiroyuki | 53bddb4 | 2010-03-10 15:20:38 -0800 | [diff] [blame] | 1343 | #else |
David Rientjes | 05af2e1 | 2012-03-21 16:34:13 -0700 | [diff] [blame] | 1344 | static inline void sync_mm_rss(struct mm_struct *mm) |
KAMEZAWA Hiroyuki | 53bddb4 | 2010-03-10 15:20:38 -0800 | [diff] [blame] | 1345 | { |
| 1346 | } |
| 1347 | #endif |
Peter Zijlstra | 465a454 | 2009-06-15 12:31:37 +0200 | [diff] [blame] | 1348 | |
Alexey Dobriyan | 4e950f6 | 2007-07-30 02:36:13 +0400 | [diff] [blame] | 1349 | int vma_wants_writenotify(struct vm_area_struct *vma); |
Peter Zijlstra | d08b385 | 2006-09-25 23:30:57 -0700 | [diff] [blame] | 1350 | |
Namhyung Kim | 25ca1d6 | 2010-10-26 14:21:59 -0700 | [diff] [blame] | 1351 | extern pte_t *__get_locked_pte(struct mm_struct *mm, unsigned long addr, |
| 1352 | spinlock_t **ptl); |
| 1353 | static inline pte_t *get_locked_pte(struct mm_struct *mm, unsigned long addr, |
| 1354 | spinlock_t **ptl) |
| 1355 | { |
| 1356 | pte_t *ptep; |
| 1357 | __cond_lock(*ptl, ptep = __get_locked_pte(mm, addr, ptl)); |
| 1358 | return ptep; |
| 1359 | } |
Linus Torvalds | c9cfcdd | 2005-11-29 14:03:14 -0800 | [diff] [blame] | 1360 | |
Nick Piggin | 5f22df0 | 2007-05-06 14:49:02 -0700 | [diff] [blame] | 1361 | #ifdef __PAGETABLE_PUD_FOLDED |
| 1362 | static inline int __pud_alloc(struct mm_struct *mm, pgd_t *pgd, |
| 1363 | unsigned long address) |
| 1364 | { |
| 1365 | return 0; |
| 1366 | } |
| 1367 | #else |
Hugh Dickins | 1bb3630 | 2005-10-29 18:16:22 -0700 | [diff] [blame] | 1368 | int __pud_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address); |
Nick Piggin | 5f22df0 | 2007-05-06 14:49:02 -0700 | [diff] [blame] | 1369 | #endif |
| 1370 | |
| 1371 | #ifdef __PAGETABLE_PMD_FOLDED |
| 1372 | static inline int __pmd_alloc(struct mm_struct *mm, pud_t *pud, |
| 1373 | unsigned long address) |
| 1374 | { |
| 1375 | return 0; |
| 1376 | } |
| 1377 | #else |
Hugh Dickins | 1bb3630 | 2005-10-29 18:16:22 -0700 | [diff] [blame] | 1378 | int __pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address); |
Nick Piggin | 5f22df0 | 2007-05-06 14:49:02 -0700 | [diff] [blame] | 1379 | #endif |
| 1380 | |
Andrea Arcangeli | 8ac1f83 | 2011-01-13 15:46:43 -0800 | [diff] [blame] | 1381 | int __pte_alloc(struct mm_struct *mm, struct vm_area_struct *vma, |
| 1382 | pmd_t *pmd, unsigned long address); |
Hugh Dickins | 1bb3630 | 2005-10-29 18:16:22 -0700 | [diff] [blame] | 1383 | int __pte_alloc_kernel(pmd_t *pmd, unsigned long address); |
| 1384 | |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1385 | /* |
| 1386 | * The following ifdef needed to get the 4level-fixup.h header to work. |
| 1387 | * Remove it when 4level-fixup.h has been removed. |
| 1388 | */ |
Hugh Dickins | 1bb3630 | 2005-10-29 18:16:22 -0700 | [diff] [blame] | 1389 | #if defined(CONFIG_MMU) && !defined(__ARCH_HAS_4LEVEL_HACK) |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1390 | static inline pud_t *pud_alloc(struct mm_struct *mm, pgd_t *pgd, unsigned long address) |
| 1391 | { |
Hugh Dickins | 1bb3630 | 2005-10-29 18:16:22 -0700 | [diff] [blame] | 1392 | return (unlikely(pgd_none(*pgd)) && __pud_alloc(mm, pgd, address))? |
| 1393 | NULL: pud_offset(pgd, address); |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1394 | } |
| 1395 | |
| 1396 | static inline pmd_t *pmd_alloc(struct mm_struct *mm, pud_t *pud, unsigned long address) |
| 1397 | { |
Hugh Dickins | 1bb3630 | 2005-10-29 18:16:22 -0700 | [diff] [blame] | 1398 | return (unlikely(pud_none(*pud)) && __pmd_alloc(mm, pud, address))? |
| 1399 | NULL: pmd_offset(pud, address); |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1400 | } |
Hugh Dickins | 1bb3630 | 2005-10-29 18:16:22 -0700 | [diff] [blame] | 1401 | #endif /* CONFIG_MMU && !__ARCH_HAS_4LEVEL_HACK */ |
| 1402 | |
Kirill A. Shutemov | 57c1ffc | 2013-11-14 14:30:45 -0800 | [diff] [blame] | 1403 | #if USE_SPLIT_PTE_PTLOCKS |
Kirill A. Shutemov | 597d795 | 2013-12-20 13:35:58 +0200 | [diff] [blame] | 1404 | #if ALLOC_SPLIT_PTLOCKS |
Kirill A. Shutemov | b35f181 | 2014-01-21 15:49:07 -0800 | [diff] [blame] | 1405 | void __init ptlock_cache_init(void); |
Peter Zijlstra | 539edb5 | 2013-11-14 14:31:52 -0800 | [diff] [blame] | 1406 | extern bool ptlock_alloc(struct page *page); |
| 1407 | extern void ptlock_free(struct page *page); |
| 1408 | |
| 1409 | static inline spinlock_t *ptlock_ptr(struct page *page) |
| 1410 | { |
| 1411 | return page->ptl; |
| 1412 | } |
Kirill A. Shutemov | 597d795 | 2013-12-20 13:35:58 +0200 | [diff] [blame] | 1413 | #else /* ALLOC_SPLIT_PTLOCKS */ |
Kirill A. Shutemov | b35f181 | 2014-01-21 15:49:07 -0800 | [diff] [blame] | 1414 | static inline void ptlock_cache_init(void) |
| 1415 | { |
| 1416 | } |
| 1417 | |
Kirill A. Shutemov | 49076ec | 2013-11-14 14:31:51 -0800 | [diff] [blame] | 1418 | static inline bool ptlock_alloc(struct page *page) |
| 1419 | { |
Kirill A. Shutemov | 49076ec | 2013-11-14 14:31:51 -0800 | [diff] [blame] | 1420 | return true; |
| 1421 | } |
Peter Zijlstra | 539edb5 | 2013-11-14 14:31:52 -0800 | [diff] [blame] | 1422 | |
Kirill A. Shutemov | 49076ec | 2013-11-14 14:31:51 -0800 | [diff] [blame] | 1423 | static inline void ptlock_free(struct page *page) |
| 1424 | { |
Kirill A. Shutemov | 49076ec | 2013-11-14 14:31:51 -0800 | [diff] [blame] | 1425 | } |
| 1426 | |
| 1427 | static inline spinlock_t *ptlock_ptr(struct page *page) |
| 1428 | { |
Peter Zijlstra | 539edb5 | 2013-11-14 14:31:52 -0800 | [diff] [blame] | 1429 | return &page->ptl; |
Kirill A. Shutemov | 49076ec | 2013-11-14 14:31:51 -0800 | [diff] [blame] | 1430 | } |
Kirill A. Shutemov | 597d795 | 2013-12-20 13:35:58 +0200 | [diff] [blame] | 1431 | #endif /* ALLOC_SPLIT_PTLOCKS */ |
Kirill A. Shutemov | 49076ec | 2013-11-14 14:31:51 -0800 | [diff] [blame] | 1432 | |
| 1433 | static inline spinlock_t *pte_lockptr(struct mm_struct *mm, pmd_t *pmd) |
| 1434 | { |
| 1435 | return ptlock_ptr(pmd_page(*pmd)); |
| 1436 | } |
| 1437 | |
| 1438 | static inline bool ptlock_init(struct page *page) |
| 1439 | { |
| 1440 | /* |
| 1441 | * prep_new_page() initialize page->private (and therefore page->ptl) |
| 1442 | * with 0. Make sure nobody took it in use in between. |
| 1443 | * |
| 1444 | * It can happen if arch try to use slab for page table allocation: |
| 1445 | * slab code uses page->slab_cache and page->first_page (for tail |
| 1446 | * pages), which share storage with page->ptl. |
| 1447 | */ |
Sasha Levin | 309381fea | 2014-01-23 15:52:54 -0800 | [diff] [blame] | 1448 | VM_BUG_ON_PAGE(*(unsigned long *)&page->ptl, page); |
Kirill A. Shutemov | 49076ec | 2013-11-14 14:31:51 -0800 | [diff] [blame] | 1449 | if (!ptlock_alloc(page)) |
| 1450 | return false; |
| 1451 | spin_lock_init(ptlock_ptr(page)); |
| 1452 | return true; |
| 1453 | } |
| 1454 | |
| 1455 | /* Reset page->mapping so free_pages_check won't complain. */ |
| 1456 | static inline void pte_lock_deinit(struct page *page) |
| 1457 | { |
| 1458 | page->mapping = NULL; |
| 1459 | ptlock_free(page); |
| 1460 | } |
| 1461 | |
Kirill A. Shutemov | 57c1ffc | 2013-11-14 14:30:45 -0800 | [diff] [blame] | 1462 | #else /* !USE_SPLIT_PTE_PTLOCKS */ |
Hugh Dickins | 4c21e2f | 2005-10-29 18:16:40 -0700 | [diff] [blame] | 1463 | /* |
| 1464 | * We use mm->page_table_lock to guard all pagetable pages of the mm. |
| 1465 | */ |
Kirill A. Shutemov | 49076ec | 2013-11-14 14:31:51 -0800 | [diff] [blame] | 1466 | static inline spinlock_t *pte_lockptr(struct mm_struct *mm, pmd_t *pmd) |
| 1467 | { |
| 1468 | return &mm->page_table_lock; |
| 1469 | } |
Kirill A. Shutemov | b35f181 | 2014-01-21 15:49:07 -0800 | [diff] [blame] | 1470 | static inline void ptlock_cache_init(void) {} |
Kirill A. Shutemov | 49076ec | 2013-11-14 14:31:51 -0800 | [diff] [blame] | 1471 | static inline bool ptlock_init(struct page *page) { return true; } |
| 1472 | static inline void pte_lock_deinit(struct page *page) {} |
Kirill A. Shutemov | 57c1ffc | 2013-11-14 14:30:45 -0800 | [diff] [blame] | 1473 | #endif /* USE_SPLIT_PTE_PTLOCKS */ |
Hugh Dickins | 4c21e2f | 2005-10-29 18:16:40 -0700 | [diff] [blame] | 1474 | |
Kirill A. Shutemov | b35f181 | 2014-01-21 15:49:07 -0800 | [diff] [blame] | 1475 | static inline void pgtable_init(void) |
| 1476 | { |
| 1477 | ptlock_cache_init(); |
| 1478 | pgtable_cache_init(); |
| 1479 | } |
| 1480 | |
Kirill A. Shutemov | 390f44e | 2013-11-14 14:31:20 -0800 | [diff] [blame] | 1481 | static inline bool pgtable_page_ctor(struct page *page) |
Martin Schwidefsky | 2f569af | 2008-02-08 04:22:04 -0800 | [diff] [blame] | 1482 | { |
Martin Schwidefsky | 2f569af | 2008-02-08 04:22:04 -0800 | [diff] [blame] | 1483 | inc_zone_page_state(page, NR_PAGETABLE); |
Kirill A. Shutemov | 49076ec | 2013-11-14 14:31:51 -0800 | [diff] [blame] | 1484 | return ptlock_init(page); |
Martin Schwidefsky | 2f569af | 2008-02-08 04:22:04 -0800 | [diff] [blame] | 1485 | } |
| 1486 | |
| 1487 | static inline void pgtable_page_dtor(struct page *page) |
| 1488 | { |
| 1489 | pte_lock_deinit(page); |
| 1490 | dec_zone_page_state(page, NR_PAGETABLE); |
| 1491 | } |
| 1492 | |
Hugh Dickins | c74df32 | 2005-10-29 18:16:23 -0700 | [diff] [blame] | 1493 | #define pte_offset_map_lock(mm, pmd, address, ptlp) \ |
| 1494 | ({ \ |
Hugh Dickins | 4c21e2f | 2005-10-29 18:16:40 -0700 | [diff] [blame] | 1495 | spinlock_t *__ptl = pte_lockptr(mm, pmd); \ |
Hugh Dickins | c74df32 | 2005-10-29 18:16:23 -0700 | [diff] [blame] | 1496 | pte_t *__pte = pte_offset_map(pmd, address); \ |
| 1497 | *(ptlp) = __ptl; \ |
| 1498 | spin_lock(__ptl); \ |
| 1499 | __pte; \ |
| 1500 | }) |
| 1501 | |
| 1502 | #define pte_unmap_unlock(pte, ptl) do { \ |
| 1503 | spin_unlock(ptl); \ |
| 1504 | pte_unmap(pte); \ |
| 1505 | } while (0) |
| 1506 | |
Andrea Arcangeli | 8ac1f83 | 2011-01-13 15:46:43 -0800 | [diff] [blame] | 1507 | #define pte_alloc_map(mm, vma, pmd, address) \ |
| 1508 | ((unlikely(pmd_none(*(pmd))) && __pte_alloc(mm, vma, \ |
| 1509 | pmd, address))? \ |
| 1510 | NULL: pte_offset_map(pmd, address)) |
Hugh Dickins | 1bb3630 | 2005-10-29 18:16:22 -0700 | [diff] [blame] | 1511 | |
Hugh Dickins | c74df32 | 2005-10-29 18:16:23 -0700 | [diff] [blame] | 1512 | #define pte_alloc_map_lock(mm, pmd, address, ptlp) \ |
Andrea Arcangeli | 8ac1f83 | 2011-01-13 15:46:43 -0800 | [diff] [blame] | 1513 | ((unlikely(pmd_none(*(pmd))) && __pte_alloc(mm, NULL, \ |
| 1514 | pmd, address))? \ |
Hugh Dickins | c74df32 | 2005-10-29 18:16:23 -0700 | [diff] [blame] | 1515 | NULL: pte_offset_map_lock(mm, pmd, address, ptlp)) |
| 1516 | |
Hugh Dickins | 1bb3630 | 2005-10-29 18:16:22 -0700 | [diff] [blame] | 1517 | #define pte_alloc_kernel(pmd, address) \ |
Andrea Arcangeli | 8ac1f83 | 2011-01-13 15:46:43 -0800 | [diff] [blame] | 1518 | ((unlikely(pmd_none(*(pmd))) && __pte_alloc_kernel(pmd, address))? \ |
Hugh Dickins | 1bb3630 | 2005-10-29 18:16:22 -0700 | [diff] [blame] | 1519 | NULL: pte_offset_kernel(pmd, address)) |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1520 | |
Kirill A. Shutemov | e009bb3 | 2013-11-14 14:31:07 -0800 | [diff] [blame] | 1521 | #if USE_SPLIT_PMD_PTLOCKS |
| 1522 | |
Martin Schwidefsky | 634391a | 2014-02-13 13:53:33 +0100 | [diff] [blame] | 1523 | static struct page *pmd_to_page(pmd_t *pmd) |
| 1524 | { |
| 1525 | unsigned long mask = ~(PTRS_PER_PMD * sizeof(pmd_t) - 1); |
| 1526 | return virt_to_page((void *)((unsigned long) pmd & mask)); |
| 1527 | } |
| 1528 | |
Kirill A. Shutemov | e009bb3 | 2013-11-14 14:31:07 -0800 | [diff] [blame] | 1529 | static inline spinlock_t *pmd_lockptr(struct mm_struct *mm, pmd_t *pmd) |
| 1530 | { |
Martin Schwidefsky | 634391a | 2014-02-13 13:53:33 +0100 | [diff] [blame] | 1531 | return ptlock_ptr(pmd_to_page(pmd)); |
Kirill A. Shutemov | e009bb3 | 2013-11-14 14:31:07 -0800 | [diff] [blame] | 1532 | } |
| 1533 | |
| 1534 | static inline bool pgtable_pmd_page_ctor(struct page *page) |
| 1535 | { |
Kirill A. Shutemov | e009bb3 | 2013-11-14 14:31:07 -0800 | [diff] [blame] | 1536 | #ifdef CONFIG_TRANSPARENT_HUGEPAGE |
| 1537 | page->pmd_huge_pte = NULL; |
| 1538 | #endif |
Kirill A. Shutemov | 49076ec | 2013-11-14 14:31:51 -0800 | [diff] [blame] | 1539 | return ptlock_init(page); |
Kirill A. Shutemov | e009bb3 | 2013-11-14 14:31:07 -0800 | [diff] [blame] | 1540 | } |
| 1541 | |
| 1542 | static inline void pgtable_pmd_page_dtor(struct page *page) |
| 1543 | { |
| 1544 | #ifdef CONFIG_TRANSPARENT_HUGEPAGE |
Sasha Levin | 309381fea | 2014-01-23 15:52:54 -0800 | [diff] [blame] | 1545 | VM_BUG_ON_PAGE(page->pmd_huge_pte, page); |
Kirill A. Shutemov | e009bb3 | 2013-11-14 14:31:07 -0800 | [diff] [blame] | 1546 | #endif |
Kirill A. Shutemov | 49076ec | 2013-11-14 14:31:51 -0800 | [diff] [blame] | 1547 | ptlock_free(page); |
Kirill A. Shutemov | e009bb3 | 2013-11-14 14:31:07 -0800 | [diff] [blame] | 1548 | } |
| 1549 | |
Martin Schwidefsky | 634391a | 2014-02-13 13:53:33 +0100 | [diff] [blame] | 1550 | #define pmd_huge_pte(mm, pmd) (pmd_to_page(pmd)->pmd_huge_pte) |
Kirill A. Shutemov | e009bb3 | 2013-11-14 14:31:07 -0800 | [diff] [blame] | 1551 | |
| 1552 | #else |
| 1553 | |
Kirill A. Shutemov | 9a86cb7 | 2013-11-14 14:30:51 -0800 | [diff] [blame] | 1554 | static inline spinlock_t *pmd_lockptr(struct mm_struct *mm, pmd_t *pmd) |
| 1555 | { |
| 1556 | return &mm->page_table_lock; |
| 1557 | } |
| 1558 | |
Kirill A. Shutemov | e009bb3 | 2013-11-14 14:31:07 -0800 | [diff] [blame] | 1559 | static inline bool pgtable_pmd_page_ctor(struct page *page) { return true; } |
| 1560 | static inline void pgtable_pmd_page_dtor(struct page *page) {} |
| 1561 | |
Kirill A. Shutemov | c389a25 | 2013-11-14 14:30:59 -0800 | [diff] [blame] | 1562 | #define pmd_huge_pte(mm, pmd) ((mm)->pmd_huge_pte) |
Kirill A. Shutemov | 9a86cb7 | 2013-11-14 14:30:51 -0800 | [diff] [blame] | 1563 | |
Kirill A. Shutemov | e009bb3 | 2013-11-14 14:31:07 -0800 | [diff] [blame] | 1564 | #endif |
| 1565 | |
Kirill A. Shutemov | 9a86cb7 | 2013-11-14 14:30:51 -0800 | [diff] [blame] | 1566 | static inline spinlock_t *pmd_lock(struct mm_struct *mm, pmd_t *pmd) |
| 1567 | { |
| 1568 | spinlock_t *ptl = pmd_lockptr(mm, pmd); |
| 1569 | spin_lock(ptl); |
| 1570 | return ptl; |
| 1571 | } |
| 1572 | |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1573 | extern void free_area_init(unsigned long * zones_size); |
Johannes Weiner | 9109fb7 | 2008-07-23 21:27:20 -0700 | [diff] [blame] | 1574 | extern void free_area_init_node(int nid, unsigned long * zones_size, |
| 1575 | unsigned long zone_start_pfn, unsigned long *zholes_size); |
David Howells | 49a7f04 | 2012-03-28 18:30:03 +0100 | [diff] [blame] | 1576 | extern void free_initmem(void); |
| 1577 | |
Jiang Liu | 69afade | 2013-04-29 15:06:21 -0700 | [diff] [blame] | 1578 | /* |
| 1579 | * Free reserved pages within range [PAGE_ALIGN(start), end & PAGE_MASK) |
| 1580 | * into the buddy system. The freed pages will be poisoned with pattern |
Jiang Liu | dbe67df | 2013-07-03 15:02:51 -0700 | [diff] [blame] | 1581 | * "poison" if it's within range [0, UCHAR_MAX]. |
Jiang Liu | 69afade | 2013-04-29 15:06:21 -0700 | [diff] [blame] | 1582 | * Return pages freed into the buddy system. |
| 1583 | */ |
Jiang Liu | 1119969 | 2013-07-03 15:02:48 -0700 | [diff] [blame] | 1584 | extern unsigned long free_reserved_area(void *start, void *end, |
Jiang Liu | 69afade | 2013-04-29 15:06:21 -0700 | [diff] [blame] | 1585 | int poison, char *s); |
Jiang Liu | c3d5f5f | 2013-07-03 15:03:14 -0700 | [diff] [blame] | 1586 | |
Jiang Liu | cfa11e0 | 2013-04-29 15:07:00 -0700 | [diff] [blame] | 1587 | #ifdef CONFIG_HIGHMEM |
| 1588 | /* |
| 1589 | * Free a highmem page into the buddy system, adjusting totalhigh_pages |
| 1590 | * and totalram_pages. |
| 1591 | */ |
| 1592 | extern void free_highmem_page(struct page *page); |
| 1593 | #endif |
Jiang Liu | 69afade | 2013-04-29 15:06:21 -0700 | [diff] [blame] | 1594 | |
Jiang Liu | c3d5f5f | 2013-07-03 15:03:14 -0700 | [diff] [blame] | 1595 | extern void adjust_managed_page_count(struct page *page, long count); |
Jiang Liu | 7ee3d4e | 2013-07-03 15:03:41 -0700 | [diff] [blame] | 1596 | extern void mem_init_print_info(const char *str); |
Jiang Liu | 69afade | 2013-04-29 15:06:21 -0700 | [diff] [blame] | 1597 | |
| 1598 | /* Free the reserved page into the buddy system, so it gets managed. */ |
| 1599 | static inline void __free_reserved_page(struct page *page) |
| 1600 | { |
| 1601 | ClearPageReserved(page); |
| 1602 | init_page_count(page); |
| 1603 | __free_page(page); |
| 1604 | } |
| 1605 | |
| 1606 | static inline void free_reserved_page(struct page *page) |
| 1607 | { |
| 1608 | __free_reserved_page(page); |
| 1609 | adjust_managed_page_count(page, 1); |
| 1610 | } |
| 1611 | |
| 1612 | static inline void mark_page_reserved(struct page *page) |
| 1613 | { |
| 1614 | SetPageReserved(page); |
| 1615 | adjust_managed_page_count(page, -1); |
| 1616 | } |
| 1617 | |
| 1618 | /* |
| 1619 | * Default method to free all the __init memory into the buddy system. |
Jiang Liu | dbe67df | 2013-07-03 15:02:51 -0700 | [diff] [blame] | 1620 | * The freed pages will be poisoned with pattern "poison" if it's within |
| 1621 | * range [0, UCHAR_MAX]. |
| 1622 | * Return pages freed into the buddy system. |
Jiang Liu | 69afade | 2013-04-29 15:06:21 -0700 | [diff] [blame] | 1623 | */ |
| 1624 | static inline unsigned long free_initmem_default(int poison) |
| 1625 | { |
| 1626 | extern char __init_begin[], __init_end[]; |
| 1627 | |
Jiang Liu | 1119969 | 2013-07-03 15:02:48 -0700 | [diff] [blame] | 1628 | return free_reserved_area(&__init_begin, &__init_end, |
Jiang Liu | 69afade | 2013-04-29 15:06:21 -0700 | [diff] [blame] | 1629 | poison, "unused kernel"); |
| 1630 | } |
| 1631 | |
Jiang Liu | 7ee3d4e | 2013-07-03 15:03:41 -0700 | [diff] [blame] | 1632 | static inline unsigned long get_num_physpages(void) |
| 1633 | { |
| 1634 | int nid; |
| 1635 | unsigned long phys_pages = 0; |
| 1636 | |
| 1637 | for_each_online_node(nid) |
| 1638 | phys_pages += node_present_pages(nid); |
| 1639 | |
| 1640 | return phys_pages; |
| 1641 | } |
| 1642 | |
Tejun Heo | 0ee332c | 2011-12-08 10:22:09 -0800 | [diff] [blame] | 1643 | #ifdef CONFIG_HAVE_MEMBLOCK_NODE_MAP |
Mel Gorman | c713216 | 2006-09-27 01:49:43 -0700 | [diff] [blame] | 1644 | /* |
Tejun Heo | 0ee332c | 2011-12-08 10:22:09 -0800 | [diff] [blame] | 1645 | * With CONFIG_HAVE_MEMBLOCK_NODE_MAP set, an architecture may initialise its |
Mel Gorman | c713216 | 2006-09-27 01:49:43 -0700 | [diff] [blame] | 1646 | * zones, allocate the backing mem_map and account for memory holes in a more |
| 1647 | * architecture independent manner. This is a substitute for creating the |
| 1648 | * zone_sizes[] and zholes_size[] arrays and passing them to |
| 1649 | * free_area_init_node() |
| 1650 | * |
| 1651 | * An architecture is expected to register range of page frames backed by |
Tejun Heo | 0ee332c | 2011-12-08 10:22:09 -0800 | [diff] [blame] | 1652 | * physical memory with memblock_add[_node]() before calling |
Mel Gorman | c713216 | 2006-09-27 01:49:43 -0700 | [diff] [blame] | 1653 | * free_area_init_nodes() passing in the PFN each zone ends at. At a basic |
| 1654 | * usage, an architecture is expected to do something like |
| 1655 | * |
| 1656 | * unsigned long max_zone_pfns[MAX_NR_ZONES] = {max_dma, max_normal_pfn, |
| 1657 | * max_highmem_pfn}; |
| 1658 | * for_each_valid_physical_page_range() |
Tejun Heo | 0ee332c | 2011-12-08 10:22:09 -0800 | [diff] [blame] | 1659 | * memblock_add_node(base, size, nid) |
Mel Gorman | c713216 | 2006-09-27 01:49:43 -0700 | [diff] [blame] | 1660 | * free_area_init_nodes(max_zone_pfns); |
| 1661 | * |
Tejun Heo | 0ee332c | 2011-12-08 10:22:09 -0800 | [diff] [blame] | 1662 | * free_bootmem_with_active_regions() calls free_bootmem_node() for each |
| 1663 | * registered physical page range. Similarly |
| 1664 | * sparse_memory_present_with_active_regions() calls memory_present() for |
| 1665 | * each range when SPARSEMEM is enabled. |
Mel Gorman | c713216 | 2006-09-27 01:49:43 -0700 | [diff] [blame] | 1666 | * |
| 1667 | * See mm/page_alloc.c for more information on each function exposed by |
Tejun Heo | 0ee332c | 2011-12-08 10:22:09 -0800 | [diff] [blame] | 1668 | * CONFIG_HAVE_MEMBLOCK_NODE_MAP. |
Mel Gorman | c713216 | 2006-09-27 01:49:43 -0700 | [diff] [blame] | 1669 | */ |
| 1670 | extern void free_area_init_nodes(unsigned long *max_zone_pfn); |
Tejun Heo | 1e01979 | 2011-07-12 09:45:34 +0200 | [diff] [blame] | 1671 | unsigned long node_map_pfn_alignment(void); |
Yinghai Lu | 3299625 | 2009-12-15 17:59:02 -0800 | [diff] [blame] | 1672 | unsigned long __absent_pages_in_range(int nid, unsigned long start_pfn, |
| 1673 | unsigned long end_pfn); |
Mel Gorman | c713216 | 2006-09-27 01:49:43 -0700 | [diff] [blame] | 1674 | extern unsigned long absent_pages_in_range(unsigned long start_pfn, |
| 1675 | unsigned long end_pfn); |
| 1676 | extern void get_pfn_range_for_nid(unsigned int nid, |
| 1677 | unsigned long *start_pfn, unsigned long *end_pfn); |
| 1678 | extern unsigned long find_min_pfn_with_active_regions(void); |
Mel Gorman | c713216 | 2006-09-27 01:49:43 -0700 | [diff] [blame] | 1679 | extern void free_bootmem_with_active_regions(int nid, |
| 1680 | unsigned long max_low_pfn); |
| 1681 | extern void sparse_memory_present_with_active_regions(int nid); |
KAMEZAWA Hiroyuki | f2dbcfa | 2009-02-18 14:48:32 -0800 | [diff] [blame] | 1682 | |
Tejun Heo | 0ee332c | 2011-12-08 10:22:09 -0800 | [diff] [blame] | 1683 | #endif /* CONFIG_HAVE_MEMBLOCK_NODE_MAP */ |
Tejun Heo | 5dfe866 | 2011-07-14 09:46:10 +0200 | [diff] [blame] | 1684 | |
Tejun Heo | 0ee332c | 2011-12-08 10:22:09 -0800 | [diff] [blame] | 1685 | #if !defined(CONFIG_HAVE_MEMBLOCK_NODE_MAP) && \ |
KAMEZAWA Hiroyuki | f2dbcfa | 2009-02-18 14:48:32 -0800 | [diff] [blame] | 1686 | !defined(CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID) |
| 1687 | static inline int __early_pfn_to_nid(unsigned long pfn) |
| 1688 | { |
| 1689 | return 0; |
| 1690 | } |
| 1691 | #else |
| 1692 | /* please see mm/page_alloc.c */ |
| 1693 | extern int __meminit early_pfn_to_nid(unsigned long pfn); |
KAMEZAWA Hiroyuki | f2dbcfa | 2009-02-18 14:48:32 -0800 | [diff] [blame] | 1694 | /* there is a per-arch backend function. */ |
| 1695 | extern int __meminit __early_pfn_to_nid(unsigned long pfn); |
KAMEZAWA Hiroyuki | f2dbcfa | 2009-02-18 14:48:32 -0800 | [diff] [blame] | 1696 | #endif |
| 1697 | |
Mel Gorman | 0e0b864 | 2006-09-27 01:49:56 -0700 | [diff] [blame] | 1698 | extern void set_dma_reserve(unsigned long new_dma_reserve); |
Dave Hansen | a2f3aa02 | 2007-01-10 23:15:30 -0800 | [diff] [blame] | 1699 | extern void memmap_init_zone(unsigned long, int, unsigned long, |
| 1700 | unsigned long, enum memmap_context); |
Minchan Kim | bc75d33 | 2009-06-16 15:32:48 -0700 | [diff] [blame] | 1701 | extern void setup_per_zone_wmarks(void); |
KOSAKI Motohiro | 1b79acc | 2011-05-24 17:11:32 -0700 | [diff] [blame] | 1702 | extern int __meminit init_per_zone_wmark_min(void); |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1703 | extern void mem_init(void); |
David Howells | 8feae13 | 2009-01-08 12:04:47 +0000 | [diff] [blame] | 1704 | extern void __init mmap_init(void); |
David Rientjes | b2b755b | 2011-03-24 15:18:15 -0700 | [diff] [blame] | 1705 | extern void show_mem(unsigned int flags); |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1706 | extern void si_meminfo(struct sysinfo * val); |
| 1707 | extern void si_meminfo_node(struct sysinfo *val, int nid); |
| 1708 | |
Joe Perches | 3ee9a4f | 2011-10-31 17:08:35 -0700 | [diff] [blame] | 1709 | extern __printf(3, 4) |
| 1710 | void warn_alloc_failed(gfp_t gfp_mask, int order, const char *fmt, ...); |
Dave Hansen | a238ab5 | 2011-05-24 17:12:16 -0700 | [diff] [blame] | 1711 | |
Christoph Lameter | e7c8d5c | 2005-06-21 17:14:47 -0700 | [diff] [blame] | 1712 | extern void setup_per_cpu_pageset(void); |
Christoph Lameter | e7c8d5c | 2005-06-21 17:14:47 -0700 | [diff] [blame] | 1713 | |
Shaohua Li | 112067f | 2009-09-21 17:01:16 -0700 | [diff] [blame] | 1714 | extern void zone_pcp_update(struct zone *zone); |
Jiang Liu | 340175b | 2012-07-31 16:43:32 -0700 | [diff] [blame] | 1715 | extern void zone_pcp_reset(struct zone *zone); |
Shaohua Li | 112067f | 2009-09-21 17:01:16 -0700 | [diff] [blame] | 1716 | |
Paul Szabo | 75f7ad8 | 2013-02-22 16:34:42 -0800 | [diff] [blame] | 1717 | /* page_alloc.c */ |
| 1718 | extern int min_free_kbytes; |
| 1719 | |
David Howells | 8feae13 | 2009-01-08 12:04:47 +0000 | [diff] [blame] | 1720 | /* nommu.c */ |
David Howells | 33e5d769 | 2009-04-02 16:56:32 -0700 | [diff] [blame] | 1721 | extern atomic_long_t mmap_pages_allocated; |
David Howells | 7e66087 | 2010-01-15 17:01:39 -0800 | [diff] [blame] | 1722 | extern int nommu_shrink_inode_mappings(struct inode *, size_t, size_t); |
David Howells | 8feae13 | 2009-01-08 12:04:47 +0000 | [diff] [blame] | 1723 | |
Michel Lespinasse | 6b2dbba | 2012-10-08 16:31:25 -0700 | [diff] [blame] | 1724 | /* interval_tree.c */ |
Michel Lespinasse | 6b2dbba | 2012-10-08 16:31:25 -0700 | [diff] [blame] | 1725 | void vma_interval_tree_insert(struct vm_area_struct *node, |
| 1726 | struct rb_root *root); |
Michel Lespinasse | 9826a51 | 2012-10-08 16:31:35 -0700 | [diff] [blame] | 1727 | void vma_interval_tree_insert_after(struct vm_area_struct *node, |
| 1728 | struct vm_area_struct *prev, |
| 1729 | struct rb_root *root); |
Michel Lespinasse | 6b2dbba | 2012-10-08 16:31:25 -0700 | [diff] [blame] | 1730 | void vma_interval_tree_remove(struct vm_area_struct *node, |
| 1731 | struct rb_root *root); |
| 1732 | struct vm_area_struct *vma_interval_tree_iter_first(struct rb_root *root, |
| 1733 | unsigned long start, unsigned long last); |
| 1734 | struct vm_area_struct *vma_interval_tree_iter_next(struct vm_area_struct *node, |
| 1735 | unsigned long start, unsigned long last); |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1736 | |
Michel Lespinasse | 6b2dbba | 2012-10-08 16:31:25 -0700 | [diff] [blame] | 1737 | #define vma_interval_tree_foreach(vma, root, start, last) \ |
| 1738 | for (vma = vma_interval_tree_iter_first(root, start, last); \ |
| 1739 | vma; vma = vma_interval_tree_iter_next(vma, start, last)) |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1740 | |
| 1741 | static inline void vma_nonlinear_insert(struct vm_area_struct *vma, |
| 1742 | struct list_head *list) |
| 1743 | { |
Michel Lespinasse | 6b2dbba | 2012-10-08 16:31:25 -0700 | [diff] [blame] | 1744 | list_add_tail(&vma->shared.nonlinear, list); |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1745 | } |
| 1746 | |
Michel Lespinasse | bf181b9 | 2012-10-08 16:31:39 -0700 | [diff] [blame] | 1747 | void anon_vma_interval_tree_insert(struct anon_vma_chain *node, |
| 1748 | struct rb_root *root); |
| 1749 | void anon_vma_interval_tree_remove(struct anon_vma_chain *node, |
| 1750 | struct rb_root *root); |
| 1751 | struct anon_vma_chain *anon_vma_interval_tree_iter_first( |
| 1752 | struct rb_root *root, unsigned long start, unsigned long last); |
| 1753 | struct anon_vma_chain *anon_vma_interval_tree_iter_next( |
| 1754 | struct anon_vma_chain *node, unsigned long start, unsigned long last); |
Michel Lespinasse | ed8ea81 | 2012-10-08 16:31:45 -0700 | [diff] [blame] | 1755 | #ifdef CONFIG_DEBUG_VM_RB |
| 1756 | void anon_vma_interval_tree_verify(struct anon_vma_chain *node); |
| 1757 | #endif |
Michel Lespinasse | bf181b9 | 2012-10-08 16:31:39 -0700 | [diff] [blame] | 1758 | |
| 1759 | #define anon_vma_interval_tree_foreach(avc, root, start, last) \ |
| 1760 | for (avc = anon_vma_interval_tree_iter_first(root, start, last); \ |
| 1761 | avc; avc = anon_vma_interval_tree_iter_next(avc, start, last)) |
| 1762 | |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1763 | /* mmap.c */ |
Alan Cox | 34b4e4a | 2007-08-22 14:01:28 -0700 | [diff] [blame] | 1764 | extern int __vm_enough_memory(struct mm_struct *mm, long pages, int cap_sys_admin); |
Rik van Riel | 5beb493 | 2010-03-05 13:42:07 -0800 | [diff] [blame] | 1765 | extern int vma_adjust(struct vm_area_struct *vma, unsigned long start, |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1766 | unsigned long end, pgoff_t pgoff, struct vm_area_struct *insert); |
| 1767 | extern struct vm_area_struct *vma_merge(struct mm_struct *, |
| 1768 | struct vm_area_struct *prev, unsigned long addr, unsigned long end, |
| 1769 | unsigned long vm_flags, struct anon_vma *, struct file *, pgoff_t, |
| 1770 | struct mempolicy *); |
| 1771 | extern struct anon_vma *find_mergeable_anon_vma(struct vm_area_struct *); |
| 1772 | extern int split_vma(struct mm_struct *, |
| 1773 | struct vm_area_struct *, unsigned long addr, int new_below); |
| 1774 | extern int insert_vm_struct(struct mm_struct *, struct vm_area_struct *); |
| 1775 | extern void __vma_link_rb(struct mm_struct *, struct vm_area_struct *, |
| 1776 | struct rb_node **, struct rb_node *); |
Hugh Dickins | a8fb561 | 2005-10-29 18:15:57 -0700 | [diff] [blame] | 1777 | extern void unlink_file_vma(struct vm_area_struct *); |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1778 | extern struct vm_area_struct *copy_vma(struct vm_area_struct **, |
Michel Lespinasse | 38a7601 | 2012-10-08 16:31:50 -0700 | [diff] [blame] | 1779 | unsigned long addr, unsigned long len, pgoff_t pgoff, |
| 1780 | bool *need_rmap_locks); |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1781 | extern void exit_mmap(struct mm_struct *); |
Matt Helsley | 925d1c4 | 2008-04-29 01:01:36 -0700 | [diff] [blame] | 1782 | |
Andrea Arcangeli | 7906d00 | 2008-07-28 15:46:26 -0700 | [diff] [blame] | 1783 | extern int mm_take_all_locks(struct mm_struct *mm); |
| 1784 | extern void mm_drop_all_locks(struct mm_struct *mm); |
| 1785 | |
Jiri Slaby | 3864601 | 2011-05-26 16:25:46 -0700 | [diff] [blame] | 1786 | extern void set_mm_exe_file(struct mm_struct *mm, struct file *new_exe_file); |
| 1787 | extern struct file *get_mm_exe_file(struct mm_struct *mm); |
Matt Helsley | 925d1c4 | 2008-04-29 01:01:36 -0700 | [diff] [blame] | 1788 | |
akpm@osdl.org | 119f657 | 2005-05-01 08:58:35 -0700 | [diff] [blame] | 1789 | extern int may_expand_vm(struct mm_struct *mm, unsigned long npages); |
Stefani Seibold | 3935ed6 | 2014-03-17 23:22:02 +0100 | [diff] [blame] | 1790 | extern struct vm_area_struct *_install_special_mapping(struct mm_struct *mm, |
| 1791 | unsigned long addr, unsigned long len, |
Andy Lutomirski | a62c34b | 2014-05-19 15:58:33 -0700 | [diff] [blame] | 1792 | unsigned long flags, |
| 1793 | const struct vm_special_mapping *spec); |
| 1794 | /* This is an obsolete alternative to _install_special_mapping. */ |
Roland McGrath | fa5dc22 | 2007-02-08 14:20:41 -0800 | [diff] [blame] | 1795 | extern int install_special_mapping(struct mm_struct *mm, |
| 1796 | unsigned long addr, unsigned long len, |
| 1797 | unsigned long flags, struct page **pages); |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1798 | |
| 1799 | extern unsigned long get_unmapped_area(struct file *, unsigned long, unsigned long, unsigned long, unsigned long); |
| 1800 | |
Miklos Szeredi | 0165ab4 | 2007-07-15 23:38:26 -0700 | [diff] [blame] | 1801 | extern unsigned long mmap_region(struct file *file, unsigned long addr, |
Michel Lespinasse | c22c0d6 | 2013-02-22 16:32:43 -0800 | [diff] [blame] | 1802 | unsigned long len, vm_flags_t vm_flags, unsigned long pgoff); |
Michel Lespinasse | bebeb3d | 2013-02-22 16:32:37 -0800 | [diff] [blame] | 1803 | extern unsigned long do_mmap_pgoff(struct file *file, unsigned long addr, |
| 1804 | unsigned long len, unsigned long prot, unsigned long flags, |
Michel Lespinasse | 41badc1 | 2013-02-22 16:32:47 -0800 | [diff] [blame] | 1805 | unsigned long pgoff, unsigned long *populate); |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1806 | extern int do_munmap(struct mm_struct *, unsigned long, size_t); |
| 1807 | |
Michel Lespinasse | bebeb3d | 2013-02-22 16:32:37 -0800 | [diff] [blame] | 1808 | #ifdef CONFIG_MMU |
| 1809 | extern int __mm_populate(unsigned long addr, unsigned long len, |
| 1810 | int ignore_errors); |
| 1811 | static inline void mm_populate(unsigned long addr, unsigned long len) |
| 1812 | { |
| 1813 | /* Ignore errors */ |
| 1814 | (void) __mm_populate(addr, len, 1); |
| 1815 | } |
| 1816 | #else |
| 1817 | static inline void mm_populate(unsigned long addr, unsigned long len) {} |
| 1818 | #endif |
| 1819 | |
Linus Torvalds | e4eb1ff | 2012-04-20 15:35:40 -0700 | [diff] [blame] | 1820 | /* These take the mm semaphore themselves */ |
| 1821 | extern unsigned long vm_brk(unsigned long, unsigned long); |
Al Viro | bfce281 | 2012-04-20 21:57:04 -0400 | [diff] [blame] | 1822 | extern int vm_munmap(unsigned long, size_t); |
Linus Torvalds | 6be5ceb | 2012-04-20 17:13:58 -0700 | [diff] [blame] | 1823 | extern unsigned long vm_mmap(struct file *, unsigned long, |
| 1824 | unsigned long, unsigned long, |
| 1825 | unsigned long, unsigned long); |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1826 | |
Michel Lespinasse | db4fbfb | 2012-12-11 16:01:49 -0800 | [diff] [blame] | 1827 | struct vm_unmapped_area_info { |
| 1828 | #define VM_UNMAPPED_AREA_TOPDOWN 1 |
| 1829 | unsigned long flags; |
| 1830 | unsigned long length; |
| 1831 | unsigned long low_limit; |
| 1832 | unsigned long high_limit; |
| 1833 | unsigned long align_mask; |
| 1834 | unsigned long align_offset; |
| 1835 | }; |
| 1836 | |
| 1837 | extern unsigned long unmapped_area(struct vm_unmapped_area_info *info); |
| 1838 | extern unsigned long unmapped_area_topdown(struct vm_unmapped_area_info *info); |
| 1839 | |
| 1840 | /* |
| 1841 | * Search for an unmapped address range. |
| 1842 | * |
| 1843 | * We are looking for a range that: |
| 1844 | * - does not intersect with any VMA; |
| 1845 | * - is contained within the [low_limit, high_limit) interval; |
| 1846 | * - is at least the desired size. |
| 1847 | * - satisfies (begin_addr & align_mask) == (align_offset & align_mask) |
| 1848 | */ |
| 1849 | static inline unsigned long |
| 1850 | vm_unmapped_area(struct vm_unmapped_area_info *info) |
| 1851 | { |
| 1852 | if (!(info->flags & VM_UNMAPPED_AREA_TOPDOWN)) |
| 1853 | return unmapped_area(info); |
| 1854 | else |
| 1855 | return unmapped_area_topdown(info); |
| 1856 | } |
| 1857 | |
Hugh Dickins | 85821aa | 2011-07-25 17:12:23 -0700 | [diff] [blame] | 1858 | /* truncate.c */ |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1859 | extern void truncate_inode_pages(struct address_space *, loff_t); |
Hans Reiser | d733907 | 2006-01-06 00:10:36 -0800 | [diff] [blame] | 1860 | extern void truncate_inode_pages_range(struct address_space *, |
| 1861 | loff_t lstart, loff_t lend); |
Johannes Weiner | 91b0abe | 2014-04-03 14:47:49 -0700 | [diff] [blame] | 1862 | extern void truncate_inode_pages_final(struct address_space *); |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1863 | |
| 1864 | /* generic vm_area_ops exported for stackable file systems */ |
Nick Piggin | d0217ac | 2007-07-19 01:47:03 -0700 | [diff] [blame] | 1865 | extern int filemap_fault(struct vm_area_struct *, struct vm_fault *); |
Kirill A. Shutemov | f182036 | 2014-04-07 15:37:19 -0700 | [diff] [blame] | 1866 | extern void filemap_map_pages(struct vm_area_struct *vma, struct vm_fault *vmf); |
Jan Kara | 4fcf1c6 | 2012-06-12 16:20:29 +0200 | [diff] [blame] | 1867 | extern int filemap_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf); |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1868 | |
| 1869 | /* mm/page-writeback.c */ |
| 1870 | int write_one_page(struct page *page, int wait); |
Nick Piggin | 1cf6e7d | 2009-02-18 14:48:18 -0800 | [diff] [blame] | 1871 | void task_dirty_inc(struct task_struct *tsk); |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1872 | |
| 1873 | /* readahead.c */ |
| 1874 | #define VM_MAX_READAHEAD 128 /* kbytes */ |
| 1875 | #define VM_MIN_READAHEAD 16 /* kbytes (includes current page) */ |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1876 | |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1877 | int force_page_cache_readahead(struct address_space *mapping, struct file *filp, |
Andrew Morton | 7361f4d | 2005-11-07 00:59:28 -0800 | [diff] [blame] | 1878 | pgoff_t offset, unsigned long nr_to_read); |
Rusty Russell | cf914a7 | 2007-07-19 01:48:08 -0700 | [diff] [blame] | 1879 | |
| 1880 | void page_cache_sync_readahead(struct address_space *mapping, |
| 1881 | struct file_ra_state *ra, |
| 1882 | struct file *filp, |
| 1883 | pgoff_t offset, |
| 1884 | unsigned long size); |
| 1885 | |
| 1886 | void page_cache_async_readahead(struct address_space *mapping, |
| 1887 | struct file_ra_state *ra, |
| 1888 | struct file *filp, |
| 1889 | struct page *pg, |
| 1890 | pgoff_t offset, |
| 1891 | unsigned long size); |
| 1892 | |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1893 | unsigned long max_sane_readahead(unsigned long nr); |
| 1894 | |
Michal Hocko | d05f316 | 2011-05-24 17:11:44 -0700 | [diff] [blame] | 1895 | /* Generic expand stack which grows the stack according to GROWS{UP,DOWN} */ |
Hugh Dickins | 46dea3d | 2005-10-29 18:16:20 -0700 | [diff] [blame] | 1896 | extern int expand_stack(struct vm_area_struct *vma, unsigned long address); |
Michal Hocko | d05f316 | 2011-05-24 17:11:44 -0700 | [diff] [blame] | 1897 | |
| 1898 | /* CONFIG_STACK_GROWSUP still needs to to grow downwards at some places */ |
| 1899 | extern int expand_downwards(struct vm_area_struct *vma, |
| 1900 | unsigned long address); |
Luck, Tony | 8ca3eb0 | 2010-08-24 11:44:18 -0700 | [diff] [blame] | 1901 | #if VM_GROWSUP |
Hugh Dickins | 46dea3d | 2005-10-29 18:16:20 -0700 | [diff] [blame] | 1902 | extern int expand_upwards(struct vm_area_struct *vma, unsigned long address); |
Luck, Tony | 8ca3eb0 | 2010-08-24 11:44:18 -0700 | [diff] [blame] | 1903 | #else |
| 1904 | #define expand_upwards(vma, address) do { } while (0) |
Matthew Wilcox | 9ab8851 | 2005-11-18 16:16:42 -0500 | [diff] [blame] | 1905 | #endif |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1906 | |
| 1907 | /* Look up the first VMA which satisfies addr < vm_end, NULL if none. */ |
| 1908 | extern struct vm_area_struct * find_vma(struct mm_struct * mm, unsigned long addr); |
| 1909 | extern struct vm_area_struct * find_vma_prev(struct mm_struct * mm, unsigned long addr, |
| 1910 | struct vm_area_struct **pprev); |
| 1911 | |
| 1912 | /* Look up the first VMA which intersects the interval start_addr..end_addr-1, |
| 1913 | NULL if none. Assume start_addr < end_addr. */ |
| 1914 | static inline struct vm_area_struct * find_vma_intersection(struct mm_struct * mm, unsigned long start_addr, unsigned long end_addr) |
| 1915 | { |
| 1916 | struct vm_area_struct * vma = find_vma(mm,start_addr); |
| 1917 | |
| 1918 | if (vma && end_addr <= vma->vm_start) |
| 1919 | vma = NULL; |
| 1920 | return vma; |
| 1921 | } |
| 1922 | |
| 1923 | static inline unsigned long vma_pages(struct vm_area_struct *vma) |
| 1924 | { |
| 1925 | return (vma->vm_end - vma->vm_start) >> PAGE_SHIFT; |
| 1926 | } |
| 1927 | |
Pavel Emelyanov | 640708a | 2012-01-10 15:11:23 -0800 | [diff] [blame] | 1928 | /* Look up the first VMA which exactly match the interval vm_start ... vm_end */ |
| 1929 | static inline struct vm_area_struct *find_exact_vma(struct mm_struct *mm, |
| 1930 | unsigned long vm_start, unsigned long vm_end) |
| 1931 | { |
| 1932 | struct vm_area_struct *vma = find_vma(mm, vm_start); |
| 1933 | |
| 1934 | if (vma && (vma->vm_start != vm_start || vma->vm_end != vm_end)) |
| 1935 | vma = NULL; |
| 1936 | |
| 1937 | return vma; |
| 1938 | } |
| 1939 | |
David Howells | bad849b | 2010-08-26 16:00:34 +0100 | [diff] [blame] | 1940 | #ifdef CONFIG_MMU |
Hugh Dickins | 804af2c | 2006-07-26 21:39:49 +0100 | [diff] [blame] | 1941 | pgprot_t vm_get_page_prot(unsigned long vm_flags); |
David Howells | bad849b | 2010-08-26 16:00:34 +0100 | [diff] [blame] | 1942 | #else |
| 1943 | static inline pgprot_t vm_get_page_prot(unsigned long vm_flags) |
| 1944 | { |
| 1945 | return __pgprot(0); |
| 1946 | } |
| 1947 | #endif |
| 1948 | |
Aneesh Kumar K.V | 5877231 | 2013-12-06 00:08:22 +0530 | [diff] [blame] | 1949 | #ifdef CONFIG_NUMA_BALANCING |
Mel Gorman | 4b10e7d | 2012-10-25 14:16:32 +0200 | [diff] [blame] | 1950 | unsigned long change_prot_numa(struct vm_area_struct *vma, |
Lee Schermerhorn | b24f53a | 2012-10-25 14:16:32 +0200 | [diff] [blame] | 1951 | unsigned long start, unsigned long end); |
| 1952 | #endif |
| 1953 | |
Hugh Dickins | deceb6c | 2005-10-29 18:16:33 -0700 | [diff] [blame] | 1954 | struct vm_area_struct *find_extend_vma(struct mm_struct *, unsigned long addr); |
Hugh Dickins | deceb6c | 2005-10-29 18:16:33 -0700 | [diff] [blame] | 1955 | int remap_pfn_range(struct vm_area_struct *, unsigned long addr, |
| 1956 | unsigned long pfn, unsigned long size, pgprot_t); |
Linus Torvalds | a145dd4 | 2005-11-30 09:35:19 -0800 | [diff] [blame] | 1957 | int vm_insert_page(struct vm_area_struct *, unsigned long addr, struct page *); |
Nick Piggin | e0dc0d8 | 2007-02-12 00:51:36 -0800 | [diff] [blame] | 1958 | int vm_insert_pfn(struct vm_area_struct *vma, unsigned long addr, |
| 1959 | unsigned long pfn); |
Nick Piggin | 423bad6 | 2008-04-28 02:13:01 -0700 | [diff] [blame] | 1960 | int vm_insert_mixed(struct vm_area_struct *vma, unsigned long addr, |
| 1961 | unsigned long pfn); |
Linus Torvalds | b4cbb19 | 2013-04-16 13:45:37 -0700 | [diff] [blame] | 1962 | int vm_iomap_memory(struct vm_area_struct *vma, phys_addr_t start, unsigned long len); |
| 1963 | |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1964 | |
Michel Lespinasse | 240aade | 2013-02-22 16:35:56 -0800 | [diff] [blame] | 1965 | struct page *follow_page_mask(struct vm_area_struct *vma, |
| 1966 | unsigned long address, unsigned int foll_flags, |
| 1967 | unsigned int *page_mask); |
| 1968 | |
| 1969 | static inline struct page *follow_page(struct vm_area_struct *vma, |
| 1970 | unsigned long address, unsigned int foll_flags) |
| 1971 | { |
| 1972 | unsigned int unused_page_mask; |
| 1973 | return follow_page_mask(vma, address, foll_flags, &unused_page_mask); |
| 1974 | } |
| 1975 | |
Hugh Dickins | deceb6c | 2005-10-29 18:16:33 -0700 | [diff] [blame] | 1976 | #define FOLL_WRITE 0x01 /* check pte is writable */ |
| 1977 | #define FOLL_TOUCH 0x02 /* mark page accessed */ |
| 1978 | #define FOLL_GET 0x04 /* do get_page on page */ |
Hugh Dickins | 8e4b9a60 | 2009-09-21 17:03:26 -0700 | [diff] [blame] | 1979 | #define FOLL_DUMP 0x08 /* give error on hole if it would be zero */ |
Hugh Dickins | 58fa879 | 2009-09-21 17:03:31 -0700 | [diff] [blame] | 1980 | #define FOLL_FORCE 0x10 /* get_user_pages read/write w/o permission */ |
Gleb Natapov | 318b275 | 2011-03-22 16:30:51 -0700 | [diff] [blame] | 1981 | #define FOLL_NOWAIT 0x20 /* if a disk transfer is needed, start the IO |
| 1982 | * and return without waiting upon it */ |
Michel Lespinasse | 110d74a | 2011-01-13 15:46:11 -0800 | [diff] [blame] | 1983 | #define FOLL_MLOCK 0x40 /* mark page as mlocked */ |
Andrea Arcangeli | 500d65d | 2011-01-13 15:46:55 -0800 | [diff] [blame] | 1984 | #define FOLL_SPLIT 0x80 /* don't return transhuge pages, split them */ |
Huang Ying | 69ebb83 | 2011-01-30 11:15:48 +0800 | [diff] [blame] | 1985 | #define FOLL_HWPOISON 0x100 /* check page is hwpoisoned */ |
Andrea Arcangeli | 0b9d705 | 2012-10-05 21:36:27 +0200 | [diff] [blame] | 1986 | #define FOLL_NUMA 0x200 /* force NUMA hinting page fault */ |
Hugh Dickins | 5117b3b | 2013-02-22 16:36:07 -0800 | [diff] [blame] | 1987 | #define FOLL_MIGRATION 0x400 /* wait for page to replace migration entry */ |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1988 | |
Martin Schwidefsky | 2f569af | 2008-02-08 04:22:04 -0800 | [diff] [blame] | 1989 | typedef int (*pte_fn_t)(pte_t *pte, pgtable_t token, unsigned long addr, |
Jeremy Fitzhardinge | aee16b3 | 2007-05-06 14:48:54 -0700 | [diff] [blame] | 1990 | void *data); |
| 1991 | extern int apply_to_page_range(struct mm_struct *mm, unsigned long address, |
| 1992 | unsigned long size, pte_fn_t fn, void *data); |
| 1993 | |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1994 | #ifdef CONFIG_PROC_FS |
Hugh Dickins | ab50b8e | 2005-10-29 18:15:56 -0700 | [diff] [blame] | 1995 | void vm_stat_account(struct mm_struct *, unsigned long, struct file *, long); |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1996 | #else |
Hugh Dickins | ab50b8e | 2005-10-29 18:15:56 -0700 | [diff] [blame] | 1997 | static inline void vm_stat_account(struct mm_struct *mm, |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1998 | unsigned long flags, struct file *file, long pages) |
| 1999 | { |
Huang Shijie | 44de9d0 | 2012-07-31 16:41:49 -0700 | [diff] [blame] | 2000 | mm->total_vm += pages; |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 2001 | } |
| 2002 | #endif /* CONFIG_PROC_FS */ |
| 2003 | |
Ingo Molnar | 12d6f21 | 2008-01-30 13:33:58 +0100 | [diff] [blame] | 2004 | #ifdef CONFIG_DEBUG_PAGEALLOC |
Ingo Molnar | 12d6f21 | 2008-01-30 13:33:58 +0100 | [diff] [blame] | 2005 | extern void kernel_map_pages(struct page *page, int numpages, int enable); |
Rafael J. Wysocki | 8a235ef | 2008-02-20 01:47:44 +0100 | [diff] [blame] | 2006 | #ifdef CONFIG_HIBERNATION |
| 2007 | extern bool kernel_page_present(struct page *page); |
| 2008 | #endif /* CONFIG_HIBERNATION */ |
Ingo Molnar | 12d6f21 | 2008-01-30 13:33:58 +0100 | [diff] [blame] | 2009 | #else |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 2010 | static inline void |
Nick Piggin | 9858db5 | 2006-10-11 01:21:30 -0700 | [diff] [blame] | 2011 | kernel_map_pages(struct page *page, int numpages, int enable) {} |
Rafael J. Wysocki | 8a235ef | 2008-02-20 01:47:44 +0100 | [diff] [blame] | 2012 | #ifdef CONFIG_HIBERNATION |
| 2013 | static inline bool kernel_page_present(struct page *page) { return true; } |
| 2014 | #endif /* CONFIG_HIBERNATION */ |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 2015 | #endif |
| 2016 | |
Stephen Wilson | 31db58b | 2011-03-13 15:49:15 -0400 | [diff] [blame] | 2017 | extern struct vm_area_struct *get_gate_vma(struct mm_struct *mm); |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 2018 | #ifdef __HAVE_ARCH_GATE_AREA |
Stephen Wilson | cae5d39 | 2011-03-13 15:49:17 -0400 | [diff] [blame] | 2019 | int in_gate_area_no_mm(unsigned long addr); |
Stephen Wilson | 83b964b | 2011-03-13 15:49:16 -0400 | [diff] [blame] | 2020 | int in_gate_area(struct mm_struct *mm, unsigned long addr); |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 2021 | #else |
Stephen Wilson | cae5d39 | 2011-03-13 15:49:17 -0400 | [diff] [blame] | 2022 | int in_gate_area_no_mm(unsigned long addr); |
| 2023 | #define in_gate_area(mm, addr) ({(void)mm; in_gate_area_no_mm(addr);}) |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 2024 | #endif /* __HAVE_ARCH_GATE_AREA */ |
| 2025 | |
Josh Triplett | 146732c | 2013-04-29 15:07:22 -0700 | [diff] [blame] | 2026 | #ifdef CONFIG_SYSCTL |
| 2027 | extern int sysctl_drop_caches; |
Alexey Dobriyan | 8d65af7 | 2009-09-23 15:57:19 -0700 | [diff] [blame] | 2028 | int drop_caches_sysctl_handler(struct ctl_table *, int, |
Andrew Morton | 9d0243b | 2006-01-08 01:00:39 -0800 | [diff] [blame] | 2029 | void __user *, size_t *, loff_t *); |
Josh Triplett | 146732c | 2013-04-29 15:07:22 -0700 | [diff] [blame] | 2030 | #endif |
| 2031 | |
Ying Han | a09ed5e | 2011-05-24 17:12:26 -0700 | [diff] [blame] | 2032 | unsigned long shrink_slab(struct shrink_control *shrink, |
Ying Han | 1495f23 | 2011-05-24 17:12:27 -0700 | [diff] [blame] | 2033 | unsigned long nr_pages_scanned, |
| 2034 | unsigned long lru_pages); |
Andrew Morton | 9d0243b | 2006-01-08 01:00:39 -0800 | [diff] [blame] | 2035 | |
Luke Yang | 7a9166e | 2006-02-20 18:28:07 -0800 | [diff] [blame] | 2036 | #ifndef CONFIG_MMU |
| 2037 | #define randomize_va_space 0 |
| 2038 | #else |
Andi Kleen | a62eaf1 | 2006-02-16 23:41:58 +0100 | [diff] [blame] | 2039 | extern int randomize_va_space; |
Luke Yang | 7a9166e | 2006-02-20 18:28:07 -0800 | [diff] [blame] | 2040 | #endif |
Andi Kleen | a62eaf1 | 2006-02-16 23:41:58 +0100 | [diff] [blame] | 2041 | |
Sam Ravnborg | 045e72a | 2007-07-26 10:41:13 -0700 | [diff] [blame] | 2042 | const char * arch_vma_name(struct vm_area_struct *vma); |
Andi Kleen | 0325291 | 2008-01-30 13:33:18 +0100 | [diff] [blame] | 2043 | void print_vma_addr(char *prefix, unsigned long rip); |
Ingo Molnar | e6e5494 | 2006-06-27 02:53:50 -0700 | [diff] [blame] | 2044 | |
Yinghai Lu | 9bdac91 | 2010-02-10 01:20:22 -0800 | [diff] [blame] | 2045 | void sparse_mem_maps_populate_node(struct page **map_map, |
| 2046 | unsigned long pnum_begin, |
| 2047 | unsigned long pnum_end, |
| 2048 | unsigned long map_count, |
| 2049 | int nodeid); |
| 2050 | |
Yasunori Goto | 98f3cfc | 2007-10-16 01:26:14 -0700 | [diff] [blame] | 2051 | struct page *sparse_mem_map_populate(unsigned long pnum, int nid); |
Andy Whitcroft | 29c7111 | 2007-10-16 01:24:14 -0700 | [diff] [blame] | 2052 | pgd_t *vmemmap_pgd_populate(unsigned long addr, int node); |
| 2053 | pud_t *vmemmap_pud_populate(pgd_t *pgd, unsigned long addr, int node); |
| 2054 | pmd_t *vmemmap_pmd_populate(pud_t *pud, unsigned long addr, int node); |
| 2055 | pte_t *vmemmap_pte_populate(pmd_t *pmd, unsigned long addr, int node); |
Christoph Lameter | 8f6aac4 | 2007-10-16 01:24:13 -0700 | [diff] [blame] | 2056 | void *vmemmap_alloc_block(unsigned long size, int node); |
Yinghai Lu | 9bdac91 | 2010-02-10 01:20:22 -0800 | [diff] [blame] | 2057 | void *vmemmap_alloc_block_buf(unsigned long size, int node); |
Christoph Lameter | 8f6aac4 | 2007-10-16 01:24:13 -0700 | [diff] [blame] | 2058 | void vmemmap_verify(pte_t *, int, unsigned long, unsigned long); |
Johannes Weiner | 0aad818 | 2013-04-29 15:07:50 -0700 | [diff] [blame] | 2059 | int vmemmap_populate_basepages(unsigned long start, unsigned long end, |
| 2060 | int node); |
| 2061 | int vmemmap_populate(unsigned long start, unsigned long end, int node); |
Yinghai Lu | c2b91e2 | 2008-04-12 01:19:24 -0700 | [diff] [blame] | 2062 | void vmemmap_populate_print_last(void); |
Tang Chen | 0197518 | 2013-02-22 16:33:08 -0800 | [diff] [blame] | 2063 | #ifdef CONFIG_MEMORY_HOTPLUG |
Johannes Weiner | 0aad818 | 2013-04-29 15:07:50 -0700 | [diff] [blame] | 2064 | void vmemmap_free(unsigned long start, unsigned long end); |
Tang Chen | 0197518 | 2013-02-22 16:33:08 -0800 | [diff] [blame] | 2065 | #endif |
Yasuaki Ishimatsu | 46723bf | 2013-02-22 16:33:00 -0800 | [diff] [blame] | 2066 | void register_page_bootmem_memmap(unsigned long section_nr, struct page *map, |
| 2067 | unsigned long size); |
Andi Kleen | 6a46079 | 2009-09-16 11:50:15 +0200 | [diff] [blame] | 2068 | |
Andi Kleen | 82ba011 | 2009-12-16 12:19:57 +0100 | [diff] [blame] | 2069 | enum mf_flags { |
| 2070 | MF_COUNT_INCREASED = 1 << 0, |
Tony Luck | 7329bbe | 2011-12-13 09:27:58 -0800 | [diff] [blame] | 2071 | MF_ACTION_REQUIRED = 1 << 1, |
Tony Luck | 6751ed6 | 2012-07-11 10:20:47 -0700 | [diff] [blame] | 2072 | MF_MUST_KILL = 1 << 2, |
Naveen N. Rao | cf870c7 | 2013-07-10 14:57:01 +0530 | [diff] [blame] | 2073 | MF_SOFT_OFFLINE = 1 << 3, |
Andi Kleen | 82ba011 | 2009-12-16 12:19:57 +0100 | [diff] [blame] | 2074 | }; |
Tony Luck | cd42f4a | 2011-12-15 10:48:12 -0800 | [diff] [blame] | 2075 | extern int memory_failure(unsigned long pfn, int trapno, int flags); |
Huang Ying | ea8f5fb | 2011-07-13 13:14:27 +0800 | [diff] [blame] | 2076 | extern void memory_failure_queue(unsigned long pfn, int trapno, int flags); |
Wu Fengguang | 847ce40 | 2009-12-16 12:19:58 +0100 | [diff] [blame] | 2077 | extern int unpoison_memory(unsigned long pfn); |
Andi Kleen | 6a46079 | 2009-09-16 11:50:15 +0200 | [diff] [blame] | 2078 | extern int sysctl_memory_failure_early_kill; |
| 2079 | extern int sysctl_memory_failure_recovery; |
Andi Kleen | facb601 | 2009-12-16 12:20:00 +0100 | [diff] [blame] | 2080 | extern void shake_page(struct page *p, int access); |
Xishi Qiu | 293c07e | 2013-02-22 16:34:02 -0800 | [diff] [blame] | 2081 | extern atomic_long_t num_poisoned_pages; |
Andi Kleen | facb601 | 2009-12-16 12:20:00 +0100 | [diff] [blame] | 2082 | extern int soft_offline_page(struct page *page, int flags); |
Andi Kleen | 6a46079 | 2009-09-16 11:50:15 +0200 | [diff] [blame] | 2083 | |
Andrea Arcangeli | 47ad847 | 2011-01-13 15:46:47 -0800 | [diff] [blame] | 2084 | #if defined(CONFIG_TRANSPARENT_HUGEPAGE) || defined(CONFIG_HUGETLBFS) |
| 2085 | extern void clear_huge_page(struct page *page, |
| 2086 | unsigned long addr, |
| 2087 | unsigned int pages_per_huge_page); |
| 2088 | extern void copy_user_huge_page(struct page *dst, struct page *src, |
| 2089 | unsigned long addr, struct vm_area_struct *vma, |
| 2090 | unsigned int pages_per_huge_page); |
| 2091 | #endif /* CONFIG_TRANSPARENT_HUGEPAGE || CONFIG_HUGETLBFS */ |
| 2092 | |
Stanislaw Gruszka | c0a32fc5 | 2012-01-10 15:07:28 -0800 | [diff] [blame] | 2093 | #ifdef CONFIG_DEBUG_PAGEALLOC |
| 2094 | extern unsigned int _debug_guardpage_minorder; |
| 2095 | |
| 2096 | static inline unsigned int debug_guardpage_minorder(void) |
| 2097 | { |
| 2098 | return _debug_guardpage_minorder; |
| 2099 | } |
| 2100 | |
| 2101 | static inline bool page_is_guard(struct page *page) |
| 2102 | { |
| 2103 | return test_bit(PAGE_DEBUG_FLAG_GUARD, &page->debug_flags); |
| 2104 | } |
| 2105 | #else |
| 2106 | static inline unsigned int debug_guardpage_minorder(void) { return 0; } |
| 2107 | static inline bool page_is_guard(struct page *page) { return false; } |
| 2108 | #endif /* CONFIG_DEBUG_PAGEALLOC */ |
| 2109 | |
Cody P Schafer | f9872ca | 2013-04-29 15:08:01 -0700 | [diff] [blame] | 2110 | #if MAX_NUMNODES > 1 |
| 2111 | void __init setup_nr_node_ids(void); |
| 2112 | #else |
| 2113 | static inline void setup_nr_node_ids(void) {} |
| 2114 | #endif |
| 2115 | |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 2116 | #endif /* __KERNEL__ */ |
| 2117 | #endif /* _LINUX_MM_H */ |